diff --git a/docker-hub/FlashOCC/Flashocc/work_dirs/flashocc-r50/20260403_162421.log b/docker-hub/FlashOCC/Flashocc/work_dirs/flashocc-r50/20260403_162421.log deleted file mode 100644 index 75d29d8972f9f958c85f61554915da9f800beb49..0000000000000000000000000000000000000000 --- a/docker-hub/FlashOCC/Flashocc/work_dirs/flashocc-r50/20260403_162421.log +++ /dev/null @@ -1,1715 +0,0 @@ -2026-04-03 16:24:21,502 - mmdet - INFO - Environment info: ------------------------------------------------------------- -sys.platform: linux -Python: 3.10.12 (main, Aug 15 2025, 14:32:43) [GCC 11.4.0] -CUDA available: True -GPU 0,1,2,3,4,5,6,7: BW1000_H -CUDA_HOME: /opt/dtk -NVCC: Not Available -GCC: x86_64-linux-gnu-gcc (Ubuntu 11.4.0-1ubuntu1~22.04) 11.4.0 -PyTorch: 2.5.1 -PyTorch compiling details: PyTorch built with: - - GCC 10.3 - - C++ Version: 201703 - - Intel(R) Math Kernel Library Version 2020.0.4 Product Build 20200917 for Intel(R) 64 architecture applications - - OpenMP 201511 (a.k.a. OpenMP 4.5) - - LAPACK is enabled (usually provided by MKL) - - NNPACK is enabled - - CPU capability usage: AVX512 - - HIP Runtime 6.3.25521 - - MIOpen 2.18.0 - - Magma 2.8.0 - - Build settings: BLAS_INFO=mkl, BUILD_TYPE=Release, CXX_COMPILER=/opt/rh/gcc-toolset-10/root/usr/bin/c++, CXX_FLAGS= -D_GLIBCXX_USE_CXX11_ABI=1 -fvisibility-inlines-hidden -DUSE_PTHREADPOOL -DNDEBUG -DUSE_KINETO -DLIBKINETO_NOCUPTI -DLIBKINETO_NOXPUPTI=ON -DUSE_FBGEMM -DUSE_PYTORCH_QNNPACK -DUSE_XNNPACK -DSYMBOLICATE_MOBILE_DEBUG_HANDLE -O2 -fPIC -Wall -Wextra -Werror=return-type -Werror=non-virtual-dtor -Werror=bool-operation -Wnarrowing -Wno-missing-field-initializers -Wno-type-limits -Wno-array-bounds -Wno-unknown-pragmas -Wno-unused-parameter -Wno-strict-overflow -Wno-strict-aliasing -Wno-stringop-overflow -Wsuggest-override -Wno-psabi -Wno-error=old-style-cast -Wno-missing-braces -fdiagnostics-color=always -faligned-new -Wno-unused-but-set-variable -Wno-maybe-uninitialized -fno-math-errno -fno-trapping-math -Werror=format -Wno-stringop-overflow, FORCE_FALLBACK_CUDA_MPI=1, LAPACK_INFO=mkl, PERF_WITH_AVX=1, PERF_WITH_AVX2=1, TORCH_VERSION=2.5.1, USE_CUDA=0, USE_CUDNN=OFF, USE_CUSPARSELT=OFF, USE_EXCEPTION_PTR=1, USE_GFLAGS=1, USE_GLOG=1, USE_GLOO=1, USE_MKL=ON, USE_MKLDNN=0, USE_MPI=1, USE_NCCL=1, USE_NNPACK=ON, USE_OPENMP=1, USE_ROCM=ON, USE_ROCM_KERNEL_ASSERT=OFF, - -TorchVision: 0.20.1 -OpenCV: 4.12.0 -MMCV: 1.6.1 -MMCV Compiler: GCC 10.3 -MMCV CUDA Compiler: rocm not available -MMDetection: 2.25.1 -MMSegmentation: 0.25.0 -MMDetection3D: 1.0.0rc4+ -spconv2.0: False ------------------------------------------------------------- - -2026-04-03 16:24:22,070 - mmdet - INFO - Distributed training: True -2026-04-03 16:24:22,636 - mmdet - INFO - Config: -point_cloud_range = [-51.2, -51.2, -5.0, 51.2, 51.2, 3.0] -class_names = [ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier', - 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' -] -dataset_type = 'NuScenesDatasetOccpancy' -data_root = 'data/nuscenes/' -input_modality = dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False) -file_client_args = dict(backend='disk') -train_pipeline = [ - dict( - type='PrepareImageInputs', - is_train=True, - data_config=dict( - cams=[ - 'CAM_FRONT_LEFT', 'CAM_FRONT', 'CAM_FRONT_RIGHT', - 'CAM_BACK_LEFT', 'CAM_BACK', 'CAM_BACK_RIGHT' - ], - Ncams=6, - input_size=(256, 704), - src_size=(900, 1600), - resize=(-0.06, 0.11), - rot=(-5.4, 5.4), - flip=True, - crop_h=(0.0, 0.0), - resize_test=0.0), - sequential=False), - dict( - type='LoadAnnotationsBEVDepth', - bda_aug_conf=dict( - rot_lim=(-0.0, 0.0), - scale_lim=(1.0, 1.0), - flip_dx_ratio=0.5, - flip_dy_ratio=0.5), - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - is_train=True), - dict(type='LoadOccGTFromFile'), - dict( - type='LoadPointsFromFile', - coord_type='LIDAR', - load_dim=5, - use_dim=5, - file_client_args=dict(backend='disk')), - dict( - type='PointToMultiViewDepth', - downsample=1, - grid_config=dict( - x=[-40, 40, 0.4], - y=[-40, 40, 0.4], - z=[-1, 5.4, 6.4], - depth=[1.0, 45.0, 0.5])), - dict( - type='DefaultFormatBundle3D', - class_names=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ]), - dict( - type='Collect3D', - keys=[ - 'img_inputs', 'gt_depth', 'voxel_semantics', 'mask_lidar', - 'mask_camera' - ]) -] -test_pipeline = [ - dict( - type='PrepareImageInputs', - data_config=dict( - cams=[ - 'CAM_FRONT_LEFT', 'CAM_FRONT', 'CAM_FRONT_RIGHT', - 'CAM_BACK_LEFT', 'CAM_BACK', 'CAM_BACK_RIGHT' - ], - Ncams=6, - input_size=(256, 704), - src_size=(900, 1600), - resize=(-0.06, 0.11), - rot=(-5.4, 5.4), - flip=True, - crop_h=(0.0, 0.0), - resize_test=0.0), - sequential=False), - dict( - type='LoadAnnotationsBEVDepth', - bda_aug_conf=dict( - rot_lim=(-0.0, 0.0), - scale_lim=(1.0, 1.0), - flip_dx_ratio=0.5, - flip_dy_ratio=0.5), - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - is_train=False), - dict( - type='LoadPointsFromFile', - coord_type='LIDAR', - load_dim=5, - use_dim=5, - file_client_args=dict(backend='disk')), - dict( - type='MultiScaleFlipAug3D', - img_scale=(1333, 800), - pts_scale_ratio=1, - flip=False, - transforms=[ - dict( - type='DefaultFormatBundle3D', - class_names=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', - 'traffic_cone' - ], - with_label=False), - dict(type='Collect3D', keys=['points', 'img_inputs']) - ]) -] -eval_pipeline = [ - dict( - type='LoadPointsFromFile', - coord_type='LIDAR', - load_dim=5, - use_dim=5, - file_client_args=dict(backend='disk')), - dict( - type='LoadPointsFromMultiSweeps', - sweeps_num=10, - file_client_args=dict(backend='disk')), - dict( - type='DefaultFormatBundle3D', - class_names=[ - 'car', 'truck', 'trailer', 'bus', 'construction_vehicle', - 'bicycle', 'motorcycle', 'pedestrian', 'traffic_cone', 'barrier' - ], - with_label=False), - dict(type='Collect3D', keys=['points']) -] -data = dict( - samples_per_gpu=24, - workers_per_gpu=24, - train=dict( - type='NuScenesDatasetOccpancy', - data_root='data/nuscenes/', - ann_file='data/nuscenes/bevdetv2-nuscenes_infos_train.pkl', - pipeline=[ - dict( - type='PrepareImageInputs', - is_train=True, - data_config=dict( - cams=[ - 'CAM_FRONT_LEFT', 'CAM_FRONT', 'CAM_FRONT_RIGHT', - 'CAM_BACK_LEFT', 'CAM_BACK', 'CAM_BACK_RIGHT' - ], - Ncams=6, - input_size=(256, 704), - src_size=(900, 1600), - resize=(-0.06, 0.11), - rot=(-5.4, 5.4), - flip=True, - crop_h=(0.0, 0.0), - resize_test=0.0), - sequential=False), - dict( - type='LoadAnnotationsBEVDepth', - bda_aug_conf=dict( - rot_lim=(-0.0, 0.0), - scale_lim=(1.0, 1.0), - flip_dx_ratio=0.5, - flip_dy_ratio=0.5), - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', - 'traffic_cone' - ], - is_train=True), - dict(type='LoadOccGTFromFile'), - dict( - type='LoadPointsFromFile', - coord_type='LIDAR', - load_dim=5, - use_dim=5, - file_client_args=dict(backend='disk')), - dict( - type='PointToMultiViewDepth', - downsample=1, - grid_config=dict( - x=[-40, 40, 0.4], - y=[-40, 40, 0.4], - z=[-1, 5.4, 6.4], - depth=[1.0, 45.0, 0.5])), - dict( - type='DefaultFormatBundle3D', - class_names=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', - 'traffic_cone' - ]), - dict( - type='Collect3D', - keys=[ - 'img_inputs', 'gt_depth', 'voxel_semantics', 'mask_lidar', - 'mask_camera' - ]) - ], - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - test_mode=False, - box_type_3d='LiDAR', - use_valid_flag=True, - stereo=False, - filter_empty_gt=False, - img_info_prototype='bevdet'), - val=dict( - type='NuScenesDatasetOccpancy', - data_root='data/nuscenes/', - ann_file='data/nuscenes/bevdetv2-nuscenes_infos_val.pkl', - pipeline=[ - dict( - type='PrepareImageInputs', - data_config=dict( - cams=[ - 'CAM_FRONT_LEFT', 'CAM_FRONT', 'CAM_FRONT_RIGHT', - 'CAM_BACK_LEFT', 'CAM_BACK', 'CAM_BACK_RIGHT' - ], - Ncams=6, - input_size=(256, 704), - src_size=(900, 1600), - resize=(-0.06, 0.11), - rot=(-5.4, 5.4), - flip=True, - crop_h=(0.0, 0.0), - resize_test=0.0), - sequential=False), - dict( - type='LoadAnnotationsBEVDepth', - bda_aug_conf=dict( - rot_lim=(-0.0, 0.0), - scale_lim=(1.0, 1.0), - flip_dx_ratio=0.5, - flip_dy_ratio=0.5), - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', - 'traffic_cone' - ], - is_train=False), - dict( - type='LoadPointsFromFile', - coord_type='LIDAR', - load_dim=5, - use_dim=5, - file_client_args=dict(backend='disk')), - dict( - type='MultiScaleFlipAug3D', - img_scale=(1333, 800), - pts_scale_ratio=1, - flip=False, - transforms=[ - dict( - type='DefaultFormatBundle3D', - class_names=[ - 'car', 'truck', 'construction_vehicle', 'bus', - 'trailer', 'barrier', 'motorcycle', 'bicycle', - 'pedestrian', 'traffic_cone' - ], - with_label=False), - dict(type='Collect3D', keys=['points', 'img_inputs']) - ]) - ], - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - test_mode=True, - box_type_3d='LiDAR', - stereo=False, - filter_empty_gt=False, - img_info_prototype='bevdet'), - test=dict( - type='NuScenesDatasetOccpancy', - data_root='data/nuscenes/', - ann_file='data/nuscenes/bevdetv2-nuscenes_infos_val.pkl', - pipeline=[ - dict( - type='PrepareImageInputs', - data_config=dict( - cams=[ - 'CAM_FRONT_LEFT', 'CAM_FRONT', 'CAM_FRONT_RIGHT', - 'CAM_BACK_LEFT', 'CAM_BACK', 'CAM_BACK_RIGHT' - ], - Ncams=6, - input_size=(256, 704), - src_size=(900, 1600), - resize=(-0.06, 0.11), - rot=(-5.4, 5.4), - flip=True, - crop_h=(0.0, 0.0), - resize_test=0.0), - sequential=False), - dict( - type='LoadAnnotationsBEVDepth', - bda_aug_conf=dict( - rot_lim=(-0.0, 0.0), - scale_lim=(1.0, 1.0), - flip_dx_ratio=0.5, - flip_dy_ratio=0.5), - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', - 'traffic_cone' - ], - is_train=False), - dict( - type='LoadPointsFromFile', - coord_type='LIDAR', - load_dim=5, - use_dim=5, - file_client_args=dict(backend='disk')), - dict( - type='MultiScaleFlipAug3D', - img_scale=(1333, 800), - pts_scale_ratio=1, - flip=False, - transforms=[ - dict( - type='DefaultFormatBundle3D', - class_names=[ - 'car', 'truck', 'construction_vehicle', 'bus', - 'trailer', 'barrier', 'motorcycle', 'bicycle', - 'pedestrian', 'traffic_cone' - ], - with_label=False), - dict(type='Collect3D', keys=['points', 'img_inputs']) - ]) - ], - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - test_mode=True, - box_type_3d='LiDAR', - stereo=False, - filter_empty_gt=False, - img_info_prototype='bevdet')) -evaluation = dict( - interval=1, - pipeline=[ - dict( - type='PrepareImageInputs', - data_config=dict( - cams=[ - 'CAM_FRONT_LEFT', 'CAM_FRONT', 'CAM_FRONT_RIGHT', - 'CAM_BACK_LEFT', 'CAM_BACK', 'CAM_BACK_RIGHT' - ], - Ncams=6, - input_size=(256, 704), - src_size=(900, 1600), - resize=(-0.06, 0.11), - rot=(-5.4, 5.4), - flip=True, - crop_h=(0.0, 0.0), - resize_test=0.0), - sequential=False), - dict( - type='LoadAnnotationsBEVDepth', - bda_aug_conf=dict( - rot_lim=(-0.0, 0.0), - scale_lim=(1.0, 1.0), - flip_dx_ratio=0.5, - flip_dy_ratio=0.5), - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', - 'traffic_cone' - ], - is_train=False), - dict( - type='LoadPointsFromFile', - coord_type='LIDAR', - load_dim=5, - use_dim=5, - file_client_args=dict(backend='disk')), - dict( - type='MultiScaleFlipAug3D', - img_scale=(1333, 800), - pts_scale_ratio=1, - flip=False, - transforms=[ - dict( - type='DefaultFormatBundle3D', - class_names=[ - 'car', 'truck', 'construction_vehicle', 'bus', - 'trailer', 'barrier', 'motorcycle', 'bicycle', - 'pedestrian', 'traffic_cone' - ], - with_label=False), - dict(type='Collect3D', keys=['points', 'img_inputs']) - ]) - ], - start=20) -checkpoint_config = dict(interval=1, max_keep_ckpts=5) -log_config = dict( - interval=1, - hooks=[dict(type='TextLoggerHook'), - dict(type='TensorboardLoggerHook')]) -dist_params = dict(backend='nccl') -log_level = 'INFO' -work_dir = './work_dirs/flashocc-r50' -load_from = 'ckpts/bevdet-r50-cbgs.pth' -resume_from = None -workflow = [('train', 1)] -opencv_num_threads = 0 -mp_start_method = 'fork' -plugin = True -plugin_dir = 'projects/mmdet3d_plugin/' -data_config = dict( - cams=[ - 'CAM_FRONT_LEFT', 'CAM_FRONT', 'CAM_FRONT_RIGHT', 'CAM_BACK_LEFT', - 'CAM_BACK', 'CAM_BACK_RIGHT' - ], - Ncams=6, - input_size=(256, 704), - src_size=(900, 1600), - resize=(-0.06, 0.11), - rot=(-5.4, 5.4), - flip=True, - crop_h=(0.0, 0.0), - resize_test=0.0) -grid_config = dict( - x=[-40, 40, 0.4], - y=[-40, 40, 0.4], - z=[-1, 5.4, 6.4], - depth=[1.0, 45.0, 0.5]) -voxel_size = [0.1, 0.1, 0.2] -numC_Trans = 64 -model = dict( - type='BEVDetOCC', - img_backbone=dict( - type='ResNet', - depth=50, - num_stages=4, - out_indices=(2, 3), - frozen_stages=-1, - norm_cfg=dict(type='BN', requires_grad=True), - norm_eval=False, - with_cp=True, - style='pytorch'), - img_neck=dict( - type='CustomFPN', - in_channels=[1024, 2048], - out_channels=256, - num_outs=1, - start_level=0, - out_ids=[0]), - img_view_transformer=dict( - type='LSSViewTransformer', - grid_config=dict( - x=[-40, 40, 0.4], - y=[-40, 40, 0.4], - z=[-1, 5.4, 6.4], - depth=[1.0, 45.0, 0.5]), - input_size=(256, 704), - in_channels=256, - out_channels=64, - sid=False, - collapse_z=True, - downsample=16), - img_bev_encoder_backbone=dict( - type='CustomResNet', numC_input=64, num_channels=[128, 256, 512]), - img_bev_encoder_neck=dict( - type='FPN_LSS', in_channels=640, out_channels=256), - occ_head=dict( - type='BEVOCCHead2D', - in_dim=256, - out_dim=256, - Dz=16, - use_mask=True, - num_classes=18, - use_predicter=True, - class_balance=False, - loss_occ=dict( - type='CrossEntropyLoss', - use_sigmoid=False, - ignore_index=255, - loss_weight=1.0))) -bda_aug_conf = dict( - rot_lim=(-0.0, 0.0), - scale_lim=(1.0, 1.0), - flip_dx_ratio=0.5, - flip_dy_ratio=0.5) -share_data_config = dict( - type='NuScenesDatasetOccpancy', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier', - 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - stereo=False, - filter_empty_gt=False, - img_info_prototype='bevdet') -test_data_config = dict( - pipeline=[ - dict( - type='PrepareImageInputs', - data_config=dict( - cams=[ - 'CAM_FRONT_LEFT', 'CAM_FRONT', 'CAM_FRONT_RIGHT', - 'CAM_BACK_LEFT', 'CAM_BACK', 'CAM_BACK_RIGHT' - ], - Ncams=6, - input_size=(256, 704), - src_size=(900, 1600), - resize=(-0.06, 0.11), - rot=(-5.4, 5.4), - flip=True, - crop_h=(0.0, 0.0), - resize_test=0.0), - sequential=False), - dict( - type='LoadAnnotationsBEVDepth', - bda_aug_conf=dict( - rot_lim=(-0.0, 0.0), - scale_lim=(1.0, 1.0), - flip_dx_ratio=0.5, - flip_dy_ratio=0.5), - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', - 'traffic_cone' - ], - is_train=False), - dict( - type='LoadPointsFromFile', - coord_type='LIDAR', - load_dim=5, - use_dim=5, - file_client_args=dict(backend='disk')), - dict( - type='MultiScaleFlipAug3D', - img_scale=(1333, 800), - pts_scale_ratio=1, - flip=False, - transforms=[ - dict( - type='DefaultFormatBundle3D', - class_names=[ - 'car', 'truck', 'construction_vehicle', 'bus', - 'trailer', 'barrier', 'motorcycle', 'bicycle', - 'pedestrian', 'traffic_cone' - ], - with_label=False), - dict(type='Collect3D', keys=['points', 'img_inputs']) - ]) - ], - ann_file='data/nuscenes/bevdetv2-nuscenes_infos_val.pkl', - type='NuScenesDatasetOccpancy', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier', - 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - stereo=False, - filter_empty_gt=False, - img_info_prototype='bevdet') -key = 'test' -optimizer = dict(type='AdamW', lr=0.0001, weight_decay=0.01) -optimizer_config = dict(grad_clip=dict(max_norm=5, norm_type=2)) -lr_config = dict( - policy='step', - warmup='linear', - warmup_iters=200, - warmup_ratio=0.001, - step=[24]) -runner = dict(type='EpochBasedRunner', max_epochs=24) -custom_hooks = [ - dict(type='MEGVIIEMAHook', init_updates=10560, priority='NORMAL') -] -gpu_ids = range(0, 8) - -2026-04-03 16:24:22,636 - mmdet - INFO - Set random seed to 0, deterministic: False -2026-04-03 16:24:22,888 - mmdet - INFO - initialize ResNet with init_cfg [{'type': 'Kaiming', 'layer': 'Conv2d'}, {'type': 'Constant', 'val': 1, 'layer': ['_BatchNorm', 'GroupNorm']}] -2026-04-03 16:24:22,995 - mmdet - INFO - initialize Bottleneck with init_cfg {'type': 'Constant', 'val': 0, 'override': {'name': 'norm3'}} -2026-04-03 16:24:22,995 - mmdet - INFO - initialize Bottleneck with init_cfg {'type': 'Constant', 'val': 0, 'override': {'name': 'norm3'}} -2026-04-03 16:24:22,996 - mmdet - INFO - initialize Bottleneck with init_cfg {'type': 'Constant', 'val': 0, 'override': {'name': 'norm3'}} -2026-04-03 16:24:22,996 - mmdet - INFO - initialize Bottleneck with init_cfg {'type': 'Constant', 'val': 0, 'override': {'name': 'norm3'}} -2026-04-03 16:24:22,997 - mmdet - INFO - initialize Bottleneck with init_cfg {'type': 'Constant', 'val': 0, 'override': {'name': 'norm3'}} -2026-04-03 16:24:22,997 - mmdet - INFO - initialize Bottleneck with init_cfg {'type': 'Constant', 'val': 0, 'override': {'name': 'norm3'}} -2026-04-03 16:24:22,998 - mmdet - INFO - initialize Bottleneck with init_cfg {'type': 'Constant', 'val': 0, 'override': {'name': 'norm3'}} -2026-04-03 16:24:22,999 - mmdet - INFO - initialize Bottleneck with init_cfg {'type': 'Constant', 'val': 0, 'override': {'name': 'norm3'}} -2026-04-03 16:24:23,000 - mmdet - INFO - initialize Bottleneck with init_cfg {'type': 'Constant', 'val': 0, 'override': {'name': 'norm3'}} -2026-04-03 16:24:23,000 - mmdet - INFO - initialize Bottleneck with init_cfg {'type': 'Constant', 'val': 0, 'override': {'name': 'norm3'}} -2026-04-03 16:24:23,001 - mmdet - INFO - initialize Bottleneck with init_cfg {'type': 'Constant', 'val': 0, 'override': {'name': 'norm3'}} -2026-04-03 16:24:23,002 - mmdet - INFO - initialize Bottleneck with init_cfg {'type': 'Constant', 'val': 0, 'override': {'name': 'norm3'}} -2026-04-03 16:24:23,003 - mmdet - INFO - initialize Bottleneck with init_cfg {'type': 'Constant', 'val': 0, 'override': {'name': 'norm3'}} -2026-04-03 16:24:23,005 - mmdet - INFO - initialize Bottleneck with init_cfg {'type': 'Constant', 'val': 0, 'override': {'name': 'norm3'}} -2026-04-03 16:24:23,008 - mmdet - INFO - initialize Bottleneck with init_cfg {'type': 'Constant', 'val': 0, 'override': {'name': 'norm3'}} -2026-04-03 16:24:23,010 - mmdet - INFO - initialize Bottleneck with init_cfg {'type': 'Constant', 'val': 0, 'override': {'name': 'norm3'}} -2026-04-03 16:24:23,021 - mmdet - INFO - initialize CustomFPN with init_cfg {'type': 'Xavier', 'layer': 'Conv2d', 'distribution': 'uniform'} -Name of parameter - Initialization information - -img_backbone.conv1.weight - torch.Size([64, 3, 7, 7]): -KaimingInit: a=0, mode=fan_out, nonlinearity=relu, distribution =normal, bias=0 - -img_backbone.bn1.weight - torch.Size([64]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.bn1.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer1.0.conv1.weight - torch.Size([64, 64, 1, 1]): -KaimingInit: a=0, mode=fan_out, nonlinearity=relu, distribution =normal, bias=0 - -img_backbone.layer1.0.bn1.weight - torch.Size([64]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer1.0.bn1.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer1.0.conv2.weight - torch.Size([64, 64, 3, 3]): -KaimingInit: a=0, mode=fan_out, nonlinearity=relu, distribution =normal, bias=0 - -img_backbone.layer1.0.bn2.weight - torch.Size([64]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer1.0.bn2.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer1.0.conv3.weight - torch.Size([256, 64, 1, 1]): -KaimingInit: a=0, mode=fan_out, nonlinearity=relu, distribution =normal, bias=0 - -img_backbone.layer1.0.bn3.weight - torch.Size([256]): -ConstantInit: val=0, bias=0 - -img_backbone.layer1.0.bn3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer1.0.downsample.0.weight - torch.Size([256, 64, 1, 1]): -KaimingInit: a=0, mode=fan_out, nonlinearity=relu, distribution =normal, bias=0 - -img_backbone.layer1.0.downsample.1.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer1.0.downsample.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer1.1.conv1.weight - torch.Size([64, 256, 1, 1]): -KaimingInit: a=0, mode=fan_out, nonlinearity=relu, distribution =normal, bias=0 - -img_backbone.layer1.1.bn1.weight - torch.Size([64]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer1.1.bn1.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer1.1.conv2.weight - torch.Size([64, 64, 3, 3]): -KaimingInit: a=0, mode=fan_out, nonlinearity=relu, distribution =normal, bias=0 - -img_backbone.layer1.1.bn2.weight - torch.Size([64]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer1.1.bn2.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer1.1.conv3.weight - torch.Size([256, 64, 1, 1]): -KaimingInit: a=0, mode=fan_out, nonlinearity=relu, distribution =normal, bias=0 - -img_backbone.layer1.1.bn3.weight - torch.Size([256]): -ConstantInit: val=0, bias=0 - -img_backbone.layer1.1.bn3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer1.2.conv1.weight - torch.Size([64, 256, 1, 1]): -KaimingInit: a=0, mode=fan_out, nonlinearity=relu, distribution =normal, bias=0 - -img_backbone.layer1.2.bn1.weight - torch.Size([64]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer1.2.bn1.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer1.2.conv2.weight - torch.Size([64, 64, 3, 3]): -KaimingInit: a=0, mode=fan_out, nonlinearity=relu, distribution =normal, bias=0 - -img_backbone.layer1.2.bn2.weight - torch.Size([64]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer1.2.bn2.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer1.2.conv3.weight - torch.Size([256, 64, 1, 1]): -KaimingInit: a=0, mode=fan_out, nonlinearity=relu, distribution =normal, bias=0 - -img_backbone.layer1.2.bn3.weight - torch.Size([256]): -ConstantInit: val=0, bias=0 - -img_backbone.layer1.2.bn3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer2.0.conv1.weight - torch.Size([128, 256, 1, 1]): -KaimingInit: a=0, mode=fan_out, nonlinearity=relu, distribution =normal, bias=0 - -img_backbone.layer2.0.bn1.weight - torch.Size([128]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer2.0.bn1.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer2.0.conv2.weight - torch.Size([128, 128, 3, 3]): -KaimingInit: a=0, mode=fan_out, nonlinearity=relu, distribution =normal, bias=0 - -img_backbone.layer2.0.bn2.weight - torch.Size([128]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer2.0.bn2.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer2.0.conv3.weight - torch.Size([512, 128, 1, 1]): -KaimingInit: a=0, mode=fan_out, nonlinearity=relu, distribution =normal, bias=0 - -img_backbone.layer2.0.bn3.weight - torch.Size([512]): -ConstantInit: val=0, bias=0 - -img_backbone.layer2.0.bn3.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer2.0.downsample.0.weight - torch.Size([512, 256, 1, 1]): -KaimingInit: a=0, mode=fan_out, nonlinearity=relu, distribution =normal, bias=0 - -img_backbone.layer2.0.downsample.1.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer2.0.downsample.1.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer2.1.conv1.weight - torch.Size([128, 512, 1, 1]): -KaimingInit: a=0, mode=fan_out, nonlinearity=relu, distribution =normal, bias=0 - -img_backbone.layer2.1.bn1.weight - torch.Size([128]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer2.1.bn1.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer2.1.conv2.weight - torch.Size([128, 128, 3, 3]): -KaimingInit: a=0, mode=fan_out, nonlinearity=relu, distribution =normal, bias=0 - -img_backbone.layer2.1.bn2.weight - torch.Size([128]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer2.1.bn2.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer2.1.conv3.weight - torch.Size([512, 128, 1, 1]): -KaimingInit: a=0, mode=fan_out, nonlinearity=relu, distribution =normal, bias=0 - -img_backbone.layer2.1.bn3.weight - torch.Size([512]): -ConstantInit: val=0, bias=0 - -img_backbone.layer2.1.bn3.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer2.2.conv1.weight - torch.Size([128, 512, 1, 1]): -KaimingInit: a=0, mode=fan_out, nonlinearity=relu, distribution =normal, bias=0 - -img_backbone.layer2.2.bn1.weight - torch.Size([128]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer2.2.bn1.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer2.2.conv2.weight - torch.Size([128, 128, 3, 3]): -KaimingInit: a=0, mode=fan_out, nonlinearity=relu, distribution =normal, bias=0 - -img_backbone.layer2.2.bn2.weight - torch.Size([128]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer2.2.bn2.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer2.2.conv3.weight - torch.Size([512, 128, 1, 1]): -KaimingInit: a=0, mode=fan_out, nonlinearity=relu, distribution =normal, bias=0 - -img_backbone.layer2.2.bn3.weight - torch.Size([512]): -ConstantInit: val=0, bias=0 - -img_backbone.layer2.2.bn3.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer2.3.conv1.weight - torch.Size([128, 512, 1, 1]): -KaimingInit: a=0, mode=fan_out, nonlinearity=relu, distribution =normal, bias=0 - -img_backbone.layer2.3.bn1.weight - torch.Size([128]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer2.3.bn1.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer2.3.conv2.weight - torch.Size([128, 128, 3, 3]): -KaimingInit: a=0, mode=fan_out, nonlinearity=relu, distribution =normal, bias=0 - -img_backbone.layer2.3.bn2.weight - torch.Size([128]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer2.3.bn2.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer2.3.conv3.weight - torch.Size([512, 128, 1, 1]): -KaimingInit: a=0, mode=fan_out, nonlinearity=relu, distribution =normal, bias=0 - -img_backbone.layer2.3.bn3.weight - torch.Size([512]): -ConstantInit: val=0, bias=0 - -img_backbone.layer2.3.bn3.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer3.0.conv1.weight - torch.Size([256, 512, 1, 1]): -KaimingInit: a=0, mode=fan_out, nonlinearity=relu, distribution =normal, bias=0 - -img_backbone.layer3.0.bn1.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer3.0.bn1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer3.0.conv2.weight - torch.Size([256, 256, 3, 3]): -KaimingInit: a=0, mode=fan_out, nonlinearity=relu, distribution =normal, bias=0 - -img_backbone.layer3.0.bn2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer3.0.bn2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer3.0.conv3.weight - torch.Size([1024, 256, 1, 1]): -KaimingInit: a=0, mode=fan_out, nonlinearity=relu, distribution =normal, bias=0 - -img_backbone.layer3.0.bn3.weight - torch.Size([1024]): -ConstantInit: val=0, bias=0 - -img_backbone.layer3.0.bn3.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer3.0.downsample.0.weight - torch.Size([1024, 512, 1, 1]): -KaimingInit: a=0, mode=fan_out, nonlinearity=relu, distribution =normal, bias=0 - -img_backbone.layer3.0.downsample.1.weight - torch.Size([1024]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer3.0.downsample.1.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer3.1.conv1.weight - torch.Size([256, 1024, 1, 1]): -KaimingInit: a=0, mode=fan_out, nonlinearity=relu, distribution =normal, bias=0 - -img_backbone.layer3.1.bn1.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer3.1.bn1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer3.1.conv2.weight - torch.Size([256, 256, 3, 3]): -KaimingInit: a=0, mode=fan_out, nonlinearity=relu, distribution =normal, bias=0 - -img_backbone.layer3.1.bn2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer3.1.bn2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer3.1.conv3.weight - torch.Size([1024, 256, 1, 1]): -KaimingInit: a=0, mode=fan_out, nonlinearity=relu, distribution =normal, bias=0 - -img_backbone.layer3.1.bn3.weight - torch.Size([1024]): -ConstantInit: val=0, bias=0 - -img_backbone.layer3.1.bn3.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer3.2.conv1.weight - torch.Size([256, 1024, 1, 1]): -KaimingInit: a=0, mode=fan_out, nonlinearity=relu, distribution =normal, bias=0 - -img_backbone.layer3.2.bn1.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer3.2.bn1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer3.2.conv2.weight - torch.Size([256, 256, 3, 3]): -KaimingInit: a=0, mode=fan_out, nonlinearity=relu, distribution =normal, bias=0 - -img_backbone.layer3.2.bn2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer3.2.bn2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer3.2.conv3.weight - torch.Size([1024, 256, 1, 1]): -KaimingInit: a=0, mode=fan_out, nonlinearity=relu, distribution =normal, bias=0 - -img_backbone.layer3.2.bn3.weight - torch.Size([1024]): -ConstantInit: val=0, bias=0 - -img_backbone.layer3.2.bn3.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer3.3.conv1.weight - torch.Size([256, 1024, 1, 1]): -KaimingInit: a=0, mode=fan_out, nonlinearity=relu, distribution =normal, bias=0 - -img_backbone.layer3.3.bn1.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer3.3.bn1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer3.3.conv2.weight - torch.Size([256, 256, 3, 3]): -KaimingInit: a=0, mode=fan_out, nonlinearity=relu, distribution =normal, bias=0 - -img_backbone.layer3.3.bn2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer3.3.bn2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer3.3.conv3.weight - torch.Size([1024, 256, 1, 1]): -KaimingInit: a=0, mode=fan_out, nonlinearity=relu, distribution =normal, bias=0 - -img_backbone.layer3.3.bn3.weight - torch.Size([1024]): -ConstantInit: val=0, bias=0 - -img_backbone.layer3.3.bn3.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer3.4.conv1.weight - torch.Size([256, 1024, 1, 1]): -KaimingInit: a=0, mode=fan_out, nonlinearity=relu, distribution =normal, bias=0 - -img_backbone.layer3.4.bn1.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer3.4.bn1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer3.4.conv2.weight - torch.Size([256, 256, 3, 3]): -KaimingInit: a=0, mode=fan_out, nonlinearity=relu, distribution =normal, bias=0 - -img_backbone.layer3.4.bn2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer3.4.bn2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer3.4.conv3.weight - torch.Size([1024, 256, 1, 1]): -KaimingInit: a=0, mode=fan_out, nonlinearity=relu, distribution =normal, bias=0 - -img_backbone.layer3.4.bn3.weight - torch.Size([1024]): -ConstantInit: val=0, bias=0 - -img_backbone.layer3.4.bn3.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer3.5.conv1.weight - torch.Size([256, 1024, 1, 1]): -KaimingInit: a=0, mode=fan_out, nonlinearity=relu, distribution =normal, bias=0 - -img_backbone.layer3.5.bn1.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer3.5.bn1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer3.5.conv2.weight - torch.Size([256, 256, 3, 3]): -KaimingInit: a=0, mode=fan_out, nonlinearity=relu, distribution =normal, bias=0 - -img_backbone.layer3.5.bn2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer3.5.bn2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer3.5.conv3.weight - torch.Size([1024, 256, 1, 1]): -KaimingInit: a=0, mode=fan_out, nonlinearity=relu, distribution =normal, bias=0 - -img_backbone.layer3.5.bn3.weight - torch.Size([1024]): -ConstantInit: val=0, bias=0 - -img_backbone.layer3.5.bn3.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer4.0.conv1.weight - torch.Size([512, 1024, 1, 1]): -KaimingInit: a=0, mode=fan_out, nonlinearity=relu, distribution =normal, bias=0 - -img_backbone.layer4.0.bn1.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer4.0.bn1.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer4.0.conv2.weight - torch.Size([512, 512, 3, 3]): -KaimingInit: a=0, mode=fan_out, nonlinearity=relu, distribution =normal, bias=0 - -img_backbone.layer4.0.bn2.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer4.0.bn2.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer4.0.conv3.weight - torch.Size([2048, 512, 1, 1]): -KaimingInit: a=0, mode=fan_out, nonlinearity=relu, distribution =normal, bias=0 - -img_backbone.layer4.0.bn3.weight - torch.Size([2048]): -ConstantInit: val=0, bias=0 - -img_backbone.layer4.0.bn3.bias - torch.Size([2048]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer4.0.downsample.0.weight - torch.Size([2048, 1024, 1, 1]): -KaimingInit: a=0, mode=fan_out, nonlinearity=relu, distribution =normal, bias=0 - -img_backbone.layer4.0.downsample.1.weight - torch.Size([2048]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer4.0.downsample.1.bias - torch.Size([2048]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer4.1.conv1.weight - torch.Size([512, 2048, 1, 1]): -KaimingInit: a=0, mode=fan_out, nonlinearity=relu, distribution =normal, bias=0 - -img_backbone.layer4.1.bn1.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer4.1.bn1.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer4.1.conv2.weight - torch.Size([512, 512, 3, 3]): -KaimingInit: a=0, mode=fan_out, nonlinearity=relu, distribution =normal, bias=0 - -img_backbone.layer4.1.bn2.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer4.1.bn2.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer4.1.conv3.weight - torch.Size([2048, 512, 1, 1]): -KaimingInit: a=0, mode=fan_out, nonlinearity=relu, distribution =normal, bias=0 - -img_backbone.layer4.1.bn3.weight - torch.Size([2048]): -ConstantInit: val=0, bias=0 - -img_backbone.layer4.1.bn3.bias - torch.Size([2048]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer4.2.conv1.weight - torch.Size([512, 2048, 1, 1]): -KaimingInit: a=0, mode=fan_out, nonlinearity=relu, distribution =normal, bias=0 - -img_backbone.layer4.2.bn1.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer4.2.bn1.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer4.2.conv2.weight - torch.Size([512, 512, 3, 3]): -KaimingInit: a=0, mode=fan_out, nonlinearity=relu, distribution =normal, bias=0 - -img_backbone.layer4.2.bn2.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer4.2.bn2.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer4.2.conv3.weight - torch.Size([2048, 512, 1, 1]): -KaimingInit: a=0, mode=fan_out, nonlinearity=relu, distribution =normal, bias=0 - -img_backbone.layer4.2.bn3.weight - torch.Size([2048]): -ConstantInit: val=0, bias=0 - -img_backbone.layer4.2.bn3.bias - torch.Size([2048]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_neck.lateral_convs.0.conv.weight - torch.Size([256, 1024, 1, 1]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.lateral_convs.0.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_neck.lateral_convs.1.conv.weight - torch.Size([256, 2048, 1, 1]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.lateral_convs.1.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_neck.fpn_convs.0.conv.weight - torch.Size([256, 256, 3, 3]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.fpn_convs.0.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_view_transformer.depth_net.weight - torch.Size([152, 256, 1, 1]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_view_transformer.depth_net.bias - torch.Size([152]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_bev_encoder_backbone.layers.0.0.conv1.weight - torch.Size([128, 64, 3, 3]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_bev_encoder_backbone.layers.0.0.bn1.weight - torch.Size([128]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_bev_encoder_backbone.layers.0.0.bn1.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_bev_encoder_backbone.layers.0.0.conv2.weight - torch.Size([128, 128, 3, 3]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_bev_encoder_backbone.layers.0.0.bn2.weight - torch.Size([128]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_bev_encoder_backbone.layers.0.0.bn2.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_bev_encoder_backbone.layers.0.0.downsample.weight - torch.Size([128, 64, 3, 3]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_bev_encoder_backbone.layers.0.0.downsample.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_bev_encoder_backbone.layers.0.1.conv1.weight - torch.Size([128, 128, 3, 3]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_bev_encoder_backbone.layers.0.1.bn1.weight - torch.Size([128]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_bev_encoder_backbone.layers.0.1.bn1.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_bev_encoder_backbone.layers.0.1.conv2.weight - torch.Size([128, 128, 3, 3]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_bev_encoder_backbone.layers.0.1.bn2.weight - torch.Size([128]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_bev_encoder_backbone.layers.0.1.bn2.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_bev_encoder_backbone.layers.1.0.conv1.weight - torch.Size([256, 128, 3, 3]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_bev_encoder_backbone.layers.1.0.bn1.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_bev_encoder_backbone.layers.1.0.bn1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_bev_encoder_backbone.layers.1.0.conv2.weight - torch.Size([256, 256, 3, 3]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_bev_encoder_backbone.layers.1.0.bn2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_bev_encoder_backbone.layers.1.0.bn2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_bev_encoder_backbone.layers.1.0.downsample.weight - torch.Size([256, 128, 3, 3]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_bev_encoder_backbone.layers.1.0.downsample.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_bev_encoder_backbone.layers.1.1.conv1.weight - torch.Size([256, 256, 3, 3]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_bev_encoder_backbone.layers.1.1.bn1.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_bev_encoder_backbone.layers.1.1.bn1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_bev_encoder_backbone.layers.1.1.conv2.weight - torch.Size([256, 256, 3, 3]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_bev_encoder_backbone.layers.1.1.bn2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_bev_encoder_backbone.layers.1.1.bn2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_bev_encoder_backbone.layers.2.0.conv1.weight - torch.Size([512, 256, 3, 3]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_bev_encoder_backbone.layers.2.0.bn1.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_bev_encoder_backbone.layers.2.0.bn1.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_bev_encoder_backbone.layers.2.0.conv2.weight - torch.Size([512, 512, 3, 3]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_bev_encoder_backbone.layers.2.0.bn2.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_bev_encoder_backbone.layers.2.0.bn2.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_bev_encoder_backbone.layers.2.0.downsample.weight - torch.Size([512, 256, 3, 3]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_bev_encoder_backbone.layers.2.0.downsample.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_bev_encoder_backbone.layers.2.1.conv1.weight - torch.Size([512, 512, 3, 3]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_bev_encoder_backbone.layers.2.1.bn1.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_bev_encoder_backbone.layers.2.1.bn1.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_bev_encoder_backbone.layers.2.1.conv2.weight - torch.Size([512, 512, 3, 3]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_bev_encoder_backbone.layers.2.1.bn2.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_bev_encoder_backbone.layers.2.1.bn2.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_bev_encoder_neck.conv.0.weight - torch.Size([512, 640, 3, 3]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_bev_encoder_neck.conv.1.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_bev_encoder_neck.conv.1.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_bev_encoder_neck.conv.3.weight - torch.Size([512, 512, 3, 3]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_bev_encoder_neck.conv.4.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_bev_encoder_neck.conv.4.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_bev_encoder_neck.up2.1.weight - torch.Size([256, 512, 3, 3]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_bev_encoder_neck.up2.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_bev_encoder_neck.up2.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_bev_encoder_neck.up2.4.weight - torch.Size([256, 256, 1, 1]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_bev_encoder_neck.up2.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -occ_head.final_conv.conv.weight - torch.Size([256, 256, 3, 3]): -Initialized by user-defined `init_weights` in ConvModule - -occ_head.final_conv.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -occ_head.predicter.0.weight - torch.Size([512, 256]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -occ_head.predicter.0.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -occ_head.predicter.2.weight - torch.Size([288, 512]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -occ_head.predicter.2.bias - torch.Size([288]): -The value is the same before and after calling `init_weights` of BEVDetOCC -2026-04-03 16:24:23,032 - mmdet - INFO - Model: -BEVDetOCC( - (img_backbone): ResNet( - (conv1): Conv2d(3, 64, kernel_size=(7, 7), stride=(2, 2), padding=(3, 3), bias=False) - (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (maxpool): MaxPool2d(kernel_size=3, stride=2, padding=1, dilation=1, ceil_mode=False) - (layer1): ResLayer( - (0): Bottleneck( - (conv1): Conv2d(64, 64, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (downsample): Sequential( - (0): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - ) - ) - init_cfg={'type': 'Constant', 'val': 0, 'override': {'name': 'norm3'}} - (1): Bottleneck( - (conv1): Conv2d(256, 64, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - init_cfg={'type': 'Constant', 'val': 0, 'override': {'name': 'norm3'}} - (2): Bottleneck( - (conv1): Conv2d(256, 64, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - init_cfg={'type': 'Constant', 'val': 0, 'override': {'name': 'norm3'}} - ) - (layer2): ResLayer( - (0): Bottleneck( - (conv1): Conv2d(256, 128, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (downsample): Sequential( - (0): Conv2d(256, 512, kernel_size=(1, 1), stride=(2, 2), bias=False) - (1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - ) - ) - init_cfg={'type': 'Constant', 'val': 0, 'override': {'name': 'norm3'}} - (1): Bottleneck( - (conv1): Conv2d(512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - init_cfg={'type': 'Constant', 'val': 0, 'override': {'name': 'norm3'}} - (2): Bottleneck( - (conv1): Conv2d(512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - init_cfg={'type': 'Constant', 'val': 0, 'override': {'name': 'norm3'}} - (3): Bottleneck( - (conv1): Conv2d(512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - init_cfg={'type': 'Constant', 'val': 0, 'override': {'name': 'norm3'}} - ) - (layer3): ResLayer( - (0): Bottleneck( - (conv1): Conv2d(512, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (downsample): Sequential( - (0): Conv2d(512, 1024, kernel_size=(1, 1), stride=(2, 2), bias=False) - (1): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - ) - ) - init_cfg={'type': 'Constant', 'val': 0, 'override': {'name': 'norm3'}} - (1): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - init_cfg={'type': 'Constant', 'val': 0, 'override': {'name': 'norm3'}} - (2): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - init_cfg={'type': 'Constant', 'val': 0, 'override': {'name': 'norm3'}} - (3): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - init_cfg={'type': 'Constant', 'val': 0, 'override': {'name': 'norm3'}} - (4): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - init_cfg={'type': 'Constant', 'val': 0, 'override': {'name': 'norm3'}} - (5): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - init_cfg={'type': 'Constant', 'val': 0, 'override': {'name': 'norm3'}} - ) - (layer4): ResLayer( - (0): Bottleneck( - (conv1): Conv2d(1024, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(512, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (downsample): Sequential( - (0): Conv2d(1024, 2048, kernel_size=(1, 1), stride=(2, 2), bias=False) - (1): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - ) - ) - init_cfg={'type': 'Constant', 'val': 0, 'override': {'name': 'norm3'}} - (1): Bottleneck( - (conv1): Conv2d(2048, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(512, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - init_cfg={'type': 'Constant', 'val': 0, 'override': {'name': 'norm3'}} - (2): Bottleneck( - (conv1): Conv2d(2048, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(512, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - init_cfg={'type': 'Constant', 'val': 0, 'override': {'name': 'norm3'}} - ) - ) - init_cfg=[{'type': 'Kaiming', 'layer': 'Conv2d'}, {'type': 'Constant', 'val': 1, 'layer': ['_BatchNorm', 'GroupNorm']}] - (img_neck): CustomFPN( - (lateral_convs): ModuleList( - (0): ConvModule( - (conv): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1)) - ) - (1): ConvModule( - (conv): Conv2d(2048, 256, kernel_size=(1, 1), stride=(1, 1)) - ) - ) - (fpn_convs): ModuleList( - (0): ConvModule( - (conv): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1)) - ) - ) - ) - init_cfg={'type': 'Xavier', 'layer': 'Conv2d', 'distribution': 'uniform'} - (img_view_transformer): LSSViewTransformer( - (depth_net): Conv2d(256, 152, kernel_size=(1, 1), stride=(1, 1)) - ) - (img_bev_encoder_backbone): CustomResNet( - (layers): Sequential( - (0): Sequential( - (0): BasicBlock( - (conv1): Conv2d(64, 128, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False) - (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (downsample): Conv2d(64, 128, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1)) - ) - (1): BasicBlock( - (conv1): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - ) - (1): Sequential( - (0): BasicBlock( - (conv1): Conv2d(128, 256, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (downsample): Conv2d(128, 256, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1)) - ) - (1): BasicBlock( - (conv1): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - ) - (2): Sequential( - (0): BasicBlock( - (conv1): Conv2d(256, 512, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False) - (bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (downsample): Conv2d(256, 512, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1)) - ) - (1): BasicBlock( - (conv1): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - ) - ) - ) - (img_bev_encoder_neck): FPN_LSS( - (up): Upsample(scale_factor=4.0, mode='bilinear') - (conv): Sequential( - (0): Conv2d(640, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (2): ReLU(inplace=True) - (3): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (4): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (5): ReLU(inplace=True) - ) - (up2): Sequential( - (0): Upsample(scale_factor=2.0, mode='bilinear') - (1): Conv2d(512, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (3): ReLU(inplace=True) - (4): Conv2d(256, 256, kernel_size=(1, 1), stride=(1, 1)) - ) - ) - (occ_head): BEVOCCHead2D( - (final_conv): ConvModule( - (conv): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1)) - (activate): ReLU(inplace=True) - ) - (predicter): Sequential( - (0): Linear(in_features=256, out_features=512, bias=True) - (1): Softplus(beta=1.0, threshold=20.0) - (2): Linear(in_features=512, out_features=288, bias=True) - ) - (loss_occ): CrossEntropyLoss(avg_non_ignore=False) - ) -) -2026-04-03 16:24:32,890 - mmdet - INFO - load checkpoint from local path: ckpts/bevdet-r50-cbgs.pth -2026-04-03 16:24:32,998 - mmdet - WARNING - The model and loaded state dict do not match exactly - -size mismatch for img_view_transformer.depth_net.weight: copying a param with shape torch.Size([123, 256, 1, 1]) from checkpoint, the shape in current model is torch.Size([152, 256, 1, 1]). -size mismatch for img_view_transformer.depth_net.bias: copying a param with shape torch.Size([123]) from checkpoint, the shape in current model is torch.Size([152]). -unexpected key in source state_dict: pts_bbox_head.shared_conv.conv.weight, pts_bbox_head.shared_conv.bn.weight, pts_bbox_head.shared_conv.bn.bias, pts_bbox_head.shared_conv.bn.running_mean, pts_bbox_head.shared_conv.bn.running_var, pts_bbox_head.shared_conv.bn.num_batches_tracked, pts_bbox_head.task_heads.0.reg.0.conv.weight, pts_bbox_head.task_heads.0.reg.0.bn.weight, pts_bbox_head.task_heads.0.reg.0.bn.bias, pts_bbox_head.task_heads.0.reg.0.bn.running_mean, pts_bbox_head.task_heads.0.reg.0.bn.running_var, pts_bbox_head.task_heads.0.reg.0.bn.num_batches_tracked, pts_bbox_head.task_heads.0.reg.1.weight, pts_bbox_head.task_heads.0.reg.1.bias, pts_bbox_head.task_heads.0.height.0.conv.weight, pts_bbox_head.task_heads.0.height.0.bn.weight, pts_bbox_head.task_heads.0.height.0.bn.bias, pts_bbox_head.task_heads.0.height.0.bn.running_mean, pts_bbox_head.task_heads.0.height.0.bn.running_var, pts_bbox_head.task_heads.0.height.0.bn.num_batches_tracked, pts_bbox_head.task_heads.0.height.1.weight, pts_bbox_head.task_heads.0.height.1.bias, pts_bbox_head.task_heads.0.dim.0.conv.weight, pts_bbox_head.task_heads.0.dim.0.bn.weight, pts_bbox_head.task_heads.0.dim.0.bn.bias, pts_bbox_head.task_heads.0.dim.0.bn.running_mean, pts_bbox_head.task_heads.0.dim.0.bn.running_var, pts_bbox_head.task_heads.0.dim.0.bn.num_batches_tracked, pts_bbox_head.task_heads.0.dim.1.weight, pts_bbox_head.task_heads.0.dim.1.bias, pts_bbox_head.task_heads.0.rot.0.conv.weight, pts_bbox_head.task_heads.0.rot.0.bn.weight, pts_bbox_head.task_heads.0.rot.0.bn.bias, pts_bbox_head.task_heads.0.rot.0.bn.running_mean, pts_bbox_head.task_heads.0.rot.0.bn.running_var, pts_bbox_head.task_heads.0.rot.0.bn.num_batches_tracked, pts_bbox_head.task_heads.0.rot.1.weight, pts_bbox_head.task_heads.0.rot.1.bias, pts_bbox_head.task_heads.0.vel.0.conv.weight, pts_bbox_head.task_heads.0.vel.0.bn.weight, pts_bbox_head.task_heads.0.vel.0.bn.bias, pts_bbox_head.task_heads.0.vel.0.bn.running_mean, pts_bbox_head.task_heads.0.vel.0.bn.running_var, pts_bbox_head.task_heads.0.vel.0.bn.num_batches_tracked, pts_bbox_head.task_heads.0.vel.1.weight, pts_bbox_head.task_heads.0.vel.1.bias, pts_bbox_head.task_heads.0.heatmap.0.conv.weight, pts_bbox_head.task_heads.0.heatmap.0.bn.weight, pts_bbox_head.task_heads.0.heatmap.0.bn.bias, pts_bbox_head.task_heads.0.heatmap.0.bn.running_mean, pts_bbox_head.task_heads.0.heatmap.0.bn.running_var, pts_bbox_head.task_heads.0.heatmap.0.bn.num_batches_tracked, pts_bbox_head.task_heads.0.heatmap.1.weight, pts_bbox_head.task_heads.0.heatmap.1.bias - -missing keys in source state_dict: occ_head.final_conv.conv.weight, occ_head.final_conv.conv.bias, occ_head.predicter.0.weight, occ_head.predicter.0.bias, occ_head.predicter.2.weight, occ_head.predicter.2.bias - -2026-04-03 16:24:33,000 - mmdet - INFO - Start running, host: root@bw61, work_dir: /workspace/Flashocc/work_dirs/flashocc-r50 -2026-04-03 16:24:33,001 - mmdet - INFO - Hooks will be executed in the following order: -before_run: -(VERY_HIGH ) StepLrUpdaterHook -(NORMAL ) CheckpointHook -(NORMAL ) MEGVIIEMAHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -before_train_epoch: -(VERY_HIGH ) StepLrUpdaterHook -(NORMAL ) DistSamplerSeedHook -(LOW ) IterTimerHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -before_train_iter: -(VERY_HIGH ) StepLrUpdaterHook -(LOW ) IterTimerHook - -------------------- -after_train_iter: -(ABOVE_NORMAL) OptimizerHook -(NORMAL ) CheckpointHook -(NORMAL ) MEGVIIEMAHook -(LOW ) IterTimerHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -after_train_epoch: -(NORMAL ) CheckpointHook -(NORMAL ) MEGVIIEMAHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -before_val_epoch: -(NORMAL ) DistSamplerSeedHook -(LOW ) IterTimerHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -before_val_iter: -(LOW ) IterTimerHook - -------------------- -after_val_iter: -(LOW ) IterTimerHook - -------------------- -after_val_epoch: -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -after_run: -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -2026-04-03 16:24:33,001 - mmdet - INFO - workflow: [('train', 1)], max: 24 epochs -2026-04-03 16:24:33,001 - mmdet - INFO - Checkpoints will be saved to /workspace/Flashocc/work_dirs/flashocc-r50 by HardDiskBackend. diff --git a/docker-hub/FlashOCC/Flashocc/work_dirs/flashocc-r50/20260403_162421.log.json b/docker-hub/FlashOCC/Flashocc/work_dirs/flashocc-r50/20260403_162421.log.json deleted file mode 100644 index 1a9fc9805d6828418d3ae7dbcad96b5b9a6d5822..0000000000000000000000000000000000000000 --- a/docker-hub/FlashOCC/Flashocc/work_dirs/flashocc-r50/20260403_162421.log.json +++ /dev/null @@ -1 +0,0 @@ -{"env_info": "sys.platform: linux\nPython: 3.10.12 (main, Aug 15 2025, 14:32:43) [GCC 11.4.0]\nCUDA available: True\nGPU 0,1,2,3,4,5,6,7: BW1000_H\nCUDA_HOME: /opt/dtk\nNVCC: Not Available\nGCC: x86_64-linux-gnu-gcc (Ubuntu 11.4.0-1ubuntu1~22.04) 11.4.0\nPyTorch: 2.5.1\nPyTorch compiling details: PyTorch built with:\n - GCC 10.3\n - C++ Version: 201703\n - Intel(R) Math Kernel Library Version 2020.0.4 Product Build 20200917 for Intel(R) 64 architecture applications\n - OpenMP 201511 (a.k.a. OpenMP 4.5)\n - LAPACK is enabled (usually provided by MKL)\n - NNPACK is enabled\n - CPU capability usage: AVX512\n - HIP Runtime 6.3.25521\n - MIOpen 2.18.0\n - Magma 2.8.0\n - Build settings: BLAS_INFO=mkl, BUILD_TYPE=Release, CXX_COMPILER=/opt/rh/gcc-toolset-10/root/usr/bin/c++, CXX_FLAGS= -D_GLIBCXX_USE_CXX11_ABI=1 -fvisibility-inlines-hidden -DUSE_PTHREADPOOL -DNDEBUG -DUSE_KINETO -DLIBKINETO_NOCUPTI -DLIBKINETO_NOXPUPTI=ON -DUSE_FBGEMM -DUSE_PYTORCH_QNNPACK -DUSE_XNNPACK -DSYMBOLICATE_MOBILE_DEBUG_HANDLE -O2 -fPIC -Wall -Wextra -Werror=return-type -Werror=non-virtual-dtor -Werror=bool-operation -Wnarrowing -Wno-missing-field-initializers -Wno-type-limits -Wno-array-bounds -Wno-unknown-pragmas -Wno-unused-parameter -Wno-strict-overflow -Wno-strict-aliasing -Wno-stringop-overflow -Wsuggest-override -Wno-psabi -Wno-error=old-style-cast -Wno-missing-braces -fdiagnostics-color=always -faligned-new -Wno-unused-but-set-variable -Wno-maybe-uninitialized -fno-math-errno -fno-trapping-math -Werror=format -Wno-stringop-overflow, FORCE_FALLBACK_CUDA_MPI=1, LAPACK_INFO=mkl, PERF_WITH_AVX=1, PERF_WITH_AVX2=1, TORCH_VERSION=2.5.1, USE_CUDA=0, USE_CUDNN=OFF, USE_CUSPARSELT=OFF, USE_EXCEPTION_PTR=1, USE_GFLAGS=1, USE_GLOG=1, USE_GLOO=1, USE_MKL=ON, USE_MKLDNN=0, USE_MPI=1, USE_NCCL=1, USE_NNPACK=ON, USE_OPENMP=1, USE_ROCM=ON, USE_ROCM_KERNEL_ASSERT=OFF, \n\nTorchVision: 0.20.1\nOpenCV: 4.12.0\nMMCV: 1.6.1\nMMCV Compiler: GCC 10.3\nMMCV CUDA Compiler: rocm not available\nMMDetection: 2.25.1\nMMSegmentation: 0.25.0\nMMDetection3D: 1.0.0rc4+\nspconv2.0: False", "config": "point_cloud_range = [-51.2, -51.2, -5.0, 51.2, 51.2, 3.0]\nclass_names = [\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier',\n 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n]\ndataset_type = 'NuScenesDatasetOccpancy'\ndata_root = 'data/nuscenes/'\ninput_modality = dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False)\nfile_client_args = dict(backend='disk')\ntrain_pipeline = [\n dict(\n type='PrepareImageInputs',\n is_train=True,\n data_config=dict(\n cams=[\n 'CAM_FRONT_LEFT', 'CAM_FRONT', 'CAM_FRONT_RIGHT',\n 'CAM_BACK_LEFT', 'CAM_BACK', 'CAM_BACK_RIGHT'\n ],\n Ncams=6,\n input_size=(256, 704),\n src_size=(900, 1600),\n resize=(-0.06, 0.11),\n rot=(-5.4, 5.4),\n flip=True,\n crop_h=(0.0, 0.0),\n resize_test=0.0),\n sequential=False),\n dict(\n type='LoadAnnotationsBEVDepth',\n bda_aug_conf=dict(\n rot_lim=(-0.0, 0.0),\n scale_lim=(1.0, 1.0),\n flip_dx_ratio=0.5,\n flip_dy_ratio=0.5),\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ],\n is_train=True),\n dict(type='LoadOccGTFromFile'),\n dict(\n type='LoadPointsFromFile',\n coord_type='LIDAR',\n load_dim=5,\n use_dim=5,\n file_client_args=dict(backend='disk')),\n dict(\n type='PointToMultiViewDepth',\n downsample=1,\n grid_config=dict(\n x=[-40, 40, 0.4],\n y=[-40, 40, 0.4],\n z=[-1, 5.4, 6.4],\n depth=[1.0, 45.0, 0.5])),\n dict(\n type='DefaultFormatBundle3D',\n class_names=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ]),\n dict(\n type='Collect3D',\n keys=[\n 'img_inputs', 'gt_depth', 'voxel_semantics', 'mask_lidar',\n 'mask_camera'\n ])\n]\ntest_pipeline = [\n dict(\n type='PrepareImageInputs',\n data_config=dict(\n cams=[\n 'CAM_FRONT_LEFT', 'CAM_FRONT', 'CAM_FRONT_RIGHT',\n 'CAM_BACK_LEFT', 'CAM_BACK', 'CAM_BACK_RIGHT'\n ],\n Ncams=6,\n input_size=(256, 704),\n src_size=(900, 1600),\n resize=(-0.06, 0.11),\n rot=(-5.4, 5.4),\n flip=True,\n crop_h=(0.0, 0.0),\n resize_test=0.0),\n sequential=False),\n dict(\n type='LoadAnnotationsBEVDepth',\n bda_aug_conf=dict(\n rot_lim=(-0.0, 0.0),\n scale_lim=(1.0, 1.0),\n flip_dx_ratio=0.5,\n flip_dy_ratio=0.5),\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ],\n is_train=False),\n dict(\n type='LoadPointsFromFile',\n coord_type='LIDAR',\n load_dim=5,\n use_dim=5,\n file_client_args=dict(backend='disk')),\n dict(\n type='MultiScaleFlipAug3D',\n img_scale=(1333, 800),\n pts_scale_ratio=1,\n flip=False,\n transforms=[\n dict(\n type='DefaultFormatBundle3D',\n class_names=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian',\n 'traffic_cone'\n ],\n with_label=False),\n dict(type='Collect3D', keys=['points', 'img_inputs'])\n ])\n]\neval_pipeline = [\n dict(\n type='LoadPointsFromFile',\n coord_type='LIDAR',\n load_dim=5,\n use_dim=5,\n file_client_args=dict(backend='disk')),\n dict(\n type='LoadPointsFromMultiSweeps',\n sweeps_num=10,\n file_client_args=dict(backend='disk')),\n dict(\n type='DefaultFormatBundle3D',\n class_names=[\n 'car', 'truck', 'trailer', 'bus', 'construction_vehicle',\n 'bicycle', 'motorcycle', 'pedestrian', 'traffic_cone', 'barrier'\n ],\n with_label=False),\n dict(type='Collect3D', keys=['points'])\n]\ndata = dict(\n samples_per_gpu=24,\n workers_per_gpu=24,\n train=dict(\n type='NuScenesDatasetOccpancy',\n data_root='data/nuscenes/',\n ann_file='data/nuscenes/bevdetv2-nuscenes_infos_train.pkl',\n pipeline=[\n dict(\n type='PrepareImageInputs',\n is_train=True,\n data_config=dict(\n cams=[\n 'CAM_FRONT_LEFT', 'CAM_FRONT', 'CAM_FRONT_RIGHT',\n 'CAM_BACK_LEFT', 'CAM_BACK', 'CAM_BACK_RIGHT'\n ],\n Ncams=6,\n input_size=(256, 704),\n src_size=(900, 1600),\n resize=(-0.06, 0.11),\n rot=(-5.4, 5.4),\n flip=True,\n crop_h=(0.0, 0.0),\n resize_test=0.0),\n sequential=False),\n dict(\n type='LoadAnnotationsBEVDepth',\n bda_aug_conf=dict(\n rot_lim=(-0.0, 0.0),\n scale_lim=(1.0, 1.0),\n flip_dx_ratio=0.5,\n flip_dy_ratio=0.5),\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian',\n 'traffic_cone'\n ],\n is_train=True),\n dict(type='LoadOccGTFromFile'),\n dict(\n type='LoadPointsFromFile',\n coord_type='LIDAR',\n load_dim=5,\n use_dim=5,\n file_client_args=dict(backend='disk')),\n dict(\n type='PointToMultiViewDepth',\n downsample=1,\n grid_config=dict(\n x=[-40, 40, 0.4],\n y=[-40, 40, 0.4],\n z=[-1, 5.4, 6.4],\n depth=[1.0, 45.0, 0.5])),\n dict(\n type='DefaultFormatBundle3D',\n class_names=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian',\n 'traffic_cone'\n ]),\n dict(\n type='Collect3D',\n keys=[\n 'img_inputs', 'gt_depth', 'voxel_semantics', 'mask_lidar',\n 'mask_camera'\n ])\n ],\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ],\n modality=dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False),\n test_mode=False,\n box_type_3d='LiDAR',\n use_valid_flag=True,\n stereo=False,\n filter_empty_gt=False,\n img_info_prototype='bevdet'),\n val=dict(\n type='NuScenesDatasetOccpancy',\n data_root='data/nuscenes/',\n ann_file='data/nuscenes/bevdetv2-nuscenes_infos_val.pkl',\n pipeline=[\n dict(\n type='PrepareImageInputs',\n data_config=dict(\n cams=[\n 'CAM_FRONT_LEFT', 'CAM_FRONT', 'CAM_FRONT_RIGHT',\n 'CAM_BACK_LEFT', 'CAM_BACK', 'CAM_BACK_RIGHT'\n ],\n Ncams=6,\n input_size=(256, 704),\n src_size=(900, 1600),\n resize=(-0.06, 0.11),\n rot=(-5.4, 5.4),\n flip=True,\n crop_h=(0.0, 0.0),\n resize_test=0.0),\n sequential=False),\n dict(\n type='LoadAnnotationsBEVDepth',\n bda_aug_conf=dict(\n rot_lim=(-0.0, 0.0),\n scale_lim=(1.0, 1.0),\n flip_dx_ratio=0.5,\n flip_dy_ratio=0.5),\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian',\n 'traffic_cone'\n ],\n is_train=False),\n dict(\n type='LoadPointsFromFile',\n coord_type='LIDAR',\n load_dim=5,\n use_dim=5,\n file_client_args=dict(backend='disk')),\n dict(\n type='MultiScaleFlipAug3D',\n img_scale=(1333, 800),\n pts_scale_ratio=1,\n flip=False,\n transforms=[\n dict(\n type='DefaultFormatBundle3D',\n class_names=[\n 'car', 'truck', 'construction_vehicle', 'bus',\n 'trailer', 'barrier', 'motorcycle', 'bicycle',\n 'pedestrian', 'traffic_cone'\n ],\n with_label=False),\n dict(type='Collect3D', keys=['points', 'img_inputs'])\n ])\n ],\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ],\n modality=dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False),\n test_mode=True,\n box_type_3d='LiDAR',\n stereo=False,\n filter_empty_gt=False,\n img_info_prototype='bevdet'),\n test=dict(\n type='NuScenesDatasetOccpancy',\n data_root='data/nuscenes/',\n ann_file='data/nuscenes/bevdetv2-nuscenes_infos_val.pkl',\n pipeline=[\n dict(\n type='PrepareImageInputs',\n data_config=dict(\n cams=[\n 'CAM_FRONT_LEFT', 'CAM_FRONT', 'CAM_FRONT_RIGHT',\n 'CAM_BACK_LEFT', 'CAM_BACK', 'CAM_BACK_RIGHT'\n ],\n Ncams=6,\n input_size=(256, 704),\n src_size=(900, 1600),\n resize=(-0.06, 0.11),\n rot=(-5.4, 5.4),\n flip=True,\n crop_h=(0.0, 0.0),\n resize_test=0.0),\n sequential=False),\n dict(\n type='LoadAnnotationsBEVDepth',\n bda_aug_conf=dict(\n rot_lim=(-0.0, 0.0),\n scale_lim=(1.0, 1.0),\n flip_dx_ratio=0.5,\n flip_dy_ratio=0.5),\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian',\n 'traffic_cone'\n ],\n is_train=False),\n dict(\n type='LoadPointsFromFile',\n coord_type='LIDAR',\n load_dim=5,\n use_dim=5,\n file_client_args=dict(backend='disk')),\n dict(\n type='MultiScaleFlipAug3D',\n img_scale=(1333, 800),\n pts_scale_ratio=1,\n flip=False,\n transforms=[\n dict(\n type='DefaultFormatBundle3D',\n class_names=[\n 'car', 'truck', 'construction_vehicle', 'bus',\n 'trailer', 'barrier', 'motorcycle', 'bicycle',\n 'pedestrian', 'traffic_cone'\n ],\n with_label=False),\n dict(type='Collect3D', keys=['points', 'img_inputs'])\n ])\n ],\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ],\n modality=dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False),\n test_mode=True,\n box_type_3d='LiDAR',\n stereo=False,\n filter_empty_gt=False,\n img_info_prototype='bevdet'))\nevaluation = dict(\n interval=1,\n pipeline=[\n dict(\n type='PrepareImageInputs',\n data_config=dict(\n cams=[\n 'CAM_FRONT_LEFT', 'CAM_FRONT', 'CAM_FRONT_RIGHT',\n 'CAM_BACK_LEFT', 'CAM_BACK', 'CAM_BACK_RIGHT'\n ],\n Ncams=6,\n input_size=(256, 704),\n src_size=(900, 1600),\n resize=(-0.06, 0.11),\n rot=(-5.4, 5.4),\n flip=True,\n crop_h=(0.0, 0.0),\n resize_test=0.0),\n sequential=False),\n dict(\n type='LoadAnnotationsBEVDepth',\n bda_aug_conf=dict(\n rot_lim=(-0.0, 0.0),\n scale_lim=(1.0, 1.0),\n flip_dx_ratio=0.5,\n flip_dy_ratio=0.5),\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian',\n 'traffic_cone'\n ],\n is_train=False),\n dict(\n type='LoadPointsFromFile',\n coord_type='LIDAR',\n load_dim=5,\n use_dim=5,\n file_client_args=dict(backend='disk')),\n dict(\n type='MultiScaleFlipAug3D',\n img_scale=(1333, 800),\n pts_scale_ratio=1,\n flip=False,\n transforms=[\n dict(\n type='DefaultFormatBundle3D',\n class_names=[\n 'car', 'truck', 'construction_vehicle', 'bus',\n 'trailer', 'barrier', 'motorcycle', 'bicycle',\n 'pedestrian', 'traffic_cone'\n ],\n with_label=False),\n dict(type='Collect3D', keys=['points', 'img_inputs'])\n ])\n ],\n start=20)\ncheckpoint_config = dict(interval=1, max_keep_ckpts=5)\nlog_config = dict(\n interval=1,\n hooks=[dict(type='TextLoggerHook'),\n dict(type='TensorboardLoggerHook')])\ndist_params = dict(backend='nccl')\nlog_level = 'INFO'\nwork_dir = './work_dirs/flashocc-r50'\nload_from = 'ckpts/bevdet-r50-cbgs.pth'\nresume_from = None\nworkflow = [('train', 1)]\nopencv_num_threads = 0\nmp_start_method = 'fork'\nplugin = True\nplugin_dir = 'projects/mmdet3d_plugin/'\ndata_config = dict(\n cams=[\n 'CAM_FRONT_LEFT', 'CAM_FRONT', 'CAM_FRONT_RIGHT', 'CAM_BACK_LEFT',\n 'CAM_BACK', 'CAM_BACK_RIGHT'\n ],\n Ncams=6,\n input_size=(256, 704),\n src_size=(900, 1600),\n resize=(-0.06, 0.11),\n rot=(-5.4, 5.4),\n flip=True,\n crop_h=(0.0, 0.0),\n resize_test=0.0)\ngrid_config = dict(\n x=[-40, 40, 0.4],\n y=[-40, 40, 0.4],\n z=[-1, 5.4, 6.4],\n depth=[1.0, 45.0, 0.5])\nvoxel_size = [0.1, 0.1, 0.2]\nnumC_Trans = 64\nmodel = dict(\n type='BEVDetOCC',\n img_backbone=dict(\n type='ResNet',\n depth=50,\n num_stages=4,\n out_indices=(2, 3),\n frozen_stages=-1,\n norm_cfg=dict(type='BN', requires_grad=True),\n norm_eval=False,\n with_cp=True,\n style='pytorch'),\n img_neck=dict(\n type='CustomFPN',\n in_channels=[1024, 2048],\n out_channels=256,\n num_outs=1,\n start_level=0,\n out_ids=[0]),\n img_view_transformer=dict(\n type='LSSViewTransformer',\n grid_config=dict(\n x=[-40, 40, 0.4],\n y=[-40, 40, 0.4],\n z=[-1, 5.4, 6.4],\n depth=[1.0, 45.0, 0.5]),\n input_size=(256, 704),\n in_channels=256,\n out_channels=64,\n sid=False,\n collapse_z=True,\n downsample=16),\n img_bev_encoder_backbone=dict(\n type='CustomResNet', numC_input=64, num_channels=[128, 256, 512]),\n img_bev_encoder_neck=dict(\n type='FPN_LSS', in_channels=640, out_channels=256),\n occ_head=dict(\n type='BEVOCCHead2D',\n in_dim=256,\n out_dim=256,\n Dz=16,\n use_mask=True,\n num_classes=18,\n use_predicter=True,\n class_balance=False,\n loss_occ=dict(\n type='CrossEntropyLoss',\n use_sigmoid=False,\n ignore_index=255,\n loss_weight=1.0)))\nbda_aug_conf = dict(\n rot_lim=(-0.0, 0.0),\n scale_lim=(1.0, 1.0),\n flip_dx_ratio=0.5,\n flip_dy_ratio=0.5)\nshare_data_config = dict(\n type='NuScenesDatasetOccpancy',\n data_root='data/nuscenes/',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier',\n 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ],\n modality=dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False),\n stereo=False,\n filter_empty_gt=False,\n img_info_prototype='bevdet')\ntest_data_config = dict(\n pipeline=[\n dict(\n type='PrepareImageInputs',\n data_config=dict(\n cams=[\n 'CAM_FRONT_LEFT', 'CAM_FRONT', 'CAM_FRONT_RIGHT',\n 'CAM_BACK_LEFT', 'CAM_BACK', 'CAM_BACK_RIGHT'\n ],\n Ncams=6,\n input_size=(256, 704),\n src_size=(900, 1600),\n resize=(-0.06, 0.11),\n rot=(-5.4, 5.4),\n flip=True,\n crop_h=(0.0, 0.0),\n resize_test=0.0),\n sequential=False),\n dict(\n type='LoadAnnotationsBEVDepth',\n bda_aug_conf=dict(\n rot_lim=(-0.0, 0.0),\n scale_lim=(1.0, 1.0),\n flip_dx_ratio=0.5,\n flip_dy_ratio=0.5),\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian',\n 'traffic_cone'\n ],\n is_train=False),\n dict(\n type='LoadPointsFromFile',\n coord_type='LIDAR',\n load_dim=5,\n use_dim=5,\n file_client_args=dict(backend='disk')),\n dict(\n type='MultiScaleFlipAug3D',\n img_scale=(1333, 800),\n pts_scale_ratio=1,\n flip=False,\n transforms=[\n dict(\n type='DefaultFormatBundle3D',\n class_names=[\n 'car', 'truck', 'construction_vehicle', 'bus',\n 'trailer', 'barrier', 'motorcycle', 'bicycle',\n 'pedestrian', 'traffic_cone'\n ],\n with_label=False),\n dict(type='Collect3D', keys=['points', 'img_inputs'])\n ])\n ],\n ann_file='data/nuscenes/bevdetv2-nuscenes_infos_val.pkl',\n type='NuScenesDatasetOccpancy',\n data_root='data/nuscenes/',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier',\n 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ],\n modality=dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False),\n stereo=False,\n filter_empty_gt=False,\n img_info_prototype='bevdet')\nkey = 'test'\noptimizer = dict(type='AdamW', lr=0.0001, weight_decay=0.01)\noptimizer_config = dict(grad_clip=dict(max_norm=5, norm_type=2))\nlr_config = dict(\n policy='step',\n warmup='linear',\n warmup_iters=200,\n warmup_ratio=0.001,\n step=[24])\nrunner = dict(type='EpochBasedRunner', max_epochs=24)\ncustom_hooks = [\n dict(type='MEGVIIEMAHook', init_updates=10560, priority='NORMAL')\n]\ngpu_ids = range(0, 8)\n", "seed": 0, "exp_name": "flashocc-r50.py"} diff --git a/docker-hub/FlashOCC/Flashocc/work_dirs/flashocc-r50/20260403_162651.log b/docker-hub/FlashOCC/Flashocc/work_dirs/flashocc-r50/20260403_162651.log deleted file mode 100644 index e01032e949635352612860ebdadfe3373eddc6f6..0000000000000000000000000000000000000000 --- a/docker-hub/FlashOCC/Flashocc/work_dirs/flashocc-r50/20260403_162651.log +++ /dev/null @@ -1,1725 +0,0 @@ -2026-04-03 16:26:51,896 - mmdet - INFO - Environment info: ------------------------------------------------------------- -sys.platform: linux -Python: 3.10.12 (main, Aug 15 2025, 14:32:43) [GCC 11.4.0] -CUDA available: True -GPU 0,1,2,3,4,5,6,7: BW1000_H -CUDA_HOME: /opt/dtk -NVCC: Not Available -GCC: x86_64-linux-gnu-gcc (Ubuntu 11.4.0-1ubuntu1~22.04) 11.4.0 -PyTorch: 2.5.1 -PyTorch compiling details: PyTorch built with: - - GCC 10.3 - - C++ Version: 201703 - - Intel(R) Math Kernel Library Version 2020.0.4 Product Build 20200917 for Intel(R) 64 architecture applications - - OpenMP 201511 (a.k.a. OpenMP 4.5) - - LAPACK is enabled (usually provided by MKL) - - NNPACK is enabled - - CPU capability usage: AVX512 - - HIP Runtime 6.3.25521 - - MIOpen 2.18.0 - - Magma 2.8.0 - - Build settings: BLAS_INFO=mkl, BUILD_TYPE=Release, CXX_COMPILER=/opt/rh/gcc-toolset-10/root/usr/bin/c++, CXX_FLAGS= -D_GLIBCXX_USE_CXX11_ABI=1 -fvisibility-inlines-hidden -DUSE_PTHREADPOOL -DNDEBUG -DUSE_KINETO -DLIBKINETO_NOCUPTI -DLIBKINETO_NOXPUPTI=ON -DUSE_FBGEMM -DUSE_PYTORCH_QNNPACK -DUSE_XNNPACK -DSYMBOLICATE_MOBILE_DEBUG_HANDLE -O2 -fPIC -Wall -Wextra -Werror=return-type -Werror=non-virtual-dtor -Werror=bool-operation -Wnarrowing -Wno-missing-field-initializers -Wno-type-limits -Wno-array-bounds -Wno-unknown-pragmas -Wno-unused-parameter -Wno-strict-overflow -Wno-strict-aliasing -Wno-stringop-overflow -Wsuggest-override -Wno-psabi -Wno-error=old-style-cast -Wno-missing-braces -fdiagnostics-color=always -faligned-new -Wno-unused-but-set-variable -Wno-maybe-uninitialized -fno-math-errno -fno-trapping-math -Werror=format -Wno-stringop-overflow, FORCE_FALLBACK_CUDA_MPI=1, LAPACK_INFO=mkl, PERF_WITH_AVX=1, PERF_WITH_AVX2=1, TORCH_VERSION=2.5.1, USE_CUDA=0, USE_CUDNN=OFF, USE_CUSPARSELT=OFF, USE_EXCEPTION_PTR=1, USE_GFLAGS=1, USE_GLOG=1, USE_GLOO=1, USE_MKL=ON, USE_MKLDNN=0, USE_MPI=1, USE_NCCL=1, USE_NNPACK=ON, USE_OPENMP=1, USE_ROCM=ON, USE_ROCM_KERNEL_ASSERT=OFF, - -TorchVision: 0.20.1 -OpenCV: 4.12.0 -MMCV: 1.6.1 -MMCV Compiler: GCC 10.3 -MMCV CUDA Compiler: rocm not available -MMDetection: 2.25.1 -MMSegmentation: 0.25.0 -MMDetection3D: 1.0.0rc4+ -spconv2.0: False ------------------------------------------------------------- - -2026-04-03 16:26:52,455 - mmdet - INFO - Distributed training: True -2026-04-03 16:26:53,024 - mmdet - INFO - Config: -point_cloud_range = [-51.2, -51.2, -5.0, 51.2, 51.2, 3.0] -class_names = [ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier', - 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' -] -dataset_type = 'NuScenesDatasetOccpancy' -data_root = 'data/nuscenes/' -input_modality = dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False) -file_client_args = dict(backend='disk') -train_pipeline = [ - dict( - type='PrepareImageInputs', - is_train=True, - data_config=dict( - cams=[ - 'CAM_FRONT_LEFT', 'CAM_FRONT', 'CAM_FRONT_RIGHT', - 'CAM_BACK_LEFT', 'CAM_BACK', 'CAM_BACK_RIGHT' - ], - Ncams=6, - input_size=(256, 704), - src_size=(900, 1600), - resize=(-0.06, 0.11), - rot=(-5.4, 5.4), - flip=True, - crop_h=(0.0, 0.0), - resize_test=0.0), - sequential=False), - dict( - type='LoadAnnotationsBEVDepth', - bda_aug_conf=dict( - rot_lim=(-0.0, 0.0), - scale_lim=(1.0, 1.0), - flip_dx_ratio=0.5, - flip_dy_ratio=0.5), - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - is_train=True), - dict(type='LoadOccGTFromFile'), - dict( - type='LoadPointsFromFile', - coord_type='LIDAR', - load_dim=5, - use_dim=5, - file_client_args=dict(backend='disk')), - dict( - type='PointToMultiViewDepth', - downsample=1, - grid_config=dict( - x=[-40, 40, 0.4], - y=[-40, 40, 0.4], - z=[-1, 5.4, 6.4], - depth=[1.0, 45.0, 0.5])), - dict( - type='DefaultFormatBundle3D', - class_names=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ]), - dict( - type='Collect3D', - keys=[ - 'img_inputs', 'gt_depth', 'voxel_semantics', 'mask_lidar', - 'mask_camera' - ]) -] -test_pipeline = [ - dict( - type='PrepareImageInputs', - data_config=dict( - cams=[ - 'CAM_FRONT_LEFT', 'CAM_FRONT', 'CAM_FRONT_RIGHT', - 'CAM_BACK_LEFT', 'CAM_BACK', 'CAM_BACK_RIGHT' - ], - Ncams=6, - input_size=(256, 704), - src_size=(900, 1600), - resize=(-0.06, 0.11), - rot=(-5.4, 5.4), - flip=True, - crop_h=(0.0, 0.0), - resize_test=0.0), - sequential=False), - dict( - type='LoadAnnotationsBEVDepth', - bda_aug_conf=dict( - rot_lim=(-0.0, 0.0), - scale_lim=(1.0, 1.0), - flip_dx_ratio=0.5, - flip_dy_ratio=0.5), - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - is_train=False), - dict( - type='LoadPointsFromFile', - coord_type='LIDAR', - load_dim=5, - use_dim=5, - file_client_args=dict(backend='disk')), - dict( - type='MultiScaleFlipAug3D', - img_scale=(1333, 800), - pts_scale_ratio=1, - flip=False, - transforms=[ - dict( - type='DefaultFormatBundle3D', - class_names=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', - 'traffic_cone' - ], - with_label=False), - dict(type='Collect3D', keys=['points', 'img_inputs']) - ]) -] -eval_pipeline = [ - dict( - type='LoadPointsFromFile', - coord_type='LIDAR', - load_dim=5, - use_dim=5, - file_client_args=dict(backend='disk')), - dict( - type='LoadPointsFromMultiSweeps', - sweeps_num=10, - file_client_args=dict(backend='disk')), - dict( - type='DefaultFormatBundle3D', - class_names=[ - 'car', 'truck', 'trailer', 'bus', 'construction_vehicle', - 'bicycle', 'motorcycle', 'pedestrian', 'traffic_cone', 'barrier' - ], - with_label=False), - dict(type='Collect3D', keys=['points']) -] -data = dict( - samples_per_gpu=24, - workers_per_gpu=24, - train=dict( - type='NuScenesDatasetOccpancy', - data_root='data/nuscenes/', - ann_file='data/nuscenes/bevdetv2-nuscenes_infos_train.pkl', - pipeline=[ - dict( - type='PrepareImageInputs', - is_train=True, - data_config=dict( - cams=[ - 'CAM_FRONT_LEFT', 'CAM_FRONT', 'CAM_FRONT_RIGHT', - 'CAM_BACK_LEFT', 'CAM_BACK', 'CAM_BACK_RIGHT' - ], - Ncams=6, - input_size=(256, 704), - src_size=(900, 1600), - resize=(-0.06, 0.11), - rot=(-5.4, 5.4), - flip=True, - crop_h=(0.0, 0.0), - resize_test=0.0), - sequential=False), - dict( - type='LoadAnnotationsBEVDepth', - bda_aug_conf=dict( - rot_lim=(-0.0, 0.0), - scale_lim=(1.0, 1.0), - flip_dx_ratio=0.5, - flip_dy_ratio=0.5), - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', - 'traffic_cone' - ], - is_train=True), - dict(type='LoadOccGTFromFile'), - dict( - type='LoadPointsFromFile', - coord_type='LIDAR', - load_dim=5, - use_dim=5, - file_client_args=dict(backend='disk')), - dict( - type='PointToMultiViewDepth', - downsample=1, - grid_config=dict( - x=[-40, 40, 0.4], - y=[-40, 40, 0.4], - z=[-1, 5.4, 6.4], - depth=[1.0, 45.0, 0.5])), - dict( - type='DefaultFormatBundle3D', - class_names=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', - 'traffic_cone' - ]), - dict( - type='Collect3D', - keys=[ - 'img_inputs', 'gt_depth', 'voxel_semantics', 'mask_lidar', - 'mask_camera' - ]) - ], - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - test_mode=False, - box_type_3d='LiDAR', - use_valid_flag=True, - stereo=False, - filter_empty_gt=False, - img_info_prototype='bevdet'), - val=dict( - type='NuScenesDatasetOccpancy', - data_root='data/nuscenes/', - ann_file='data/nuscenes/bevdetv2-nuscenes_infos_val.pkl', - pipeline=[ - dict( - type='PrepareImageInputs', - data_config=dict( - cams=[ - 'CAM_FRONT_LEFT', 'CAM_FRONT', 'CAM_FRONT_RIGHT', - 'CAM_BACK_LEFT', 'CAM_BACK', 'CAM_BACK_RIGHT' - ], - Ncams=6, - input_size=(256, 704), - src_size=(900, 1600), - resize=(-0.06, 0.11), - rot=(-5.4, 5.4), - flip=True, - crop_h=(0.0, 0.0), - resize_test=0.0), - sequential=False), - dict( - type='LoadAnnotationsBEVDepth', - bda_aug_conf=dict( - rot_lim=(-0.0, 0.0), - scale_lim=(1.0, 1.0), - flip_dx_ratio=0.5, - flip_dy_ratio=0.5), - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', - 'traffic_cone' - ], - is_train=False), - dict( - type='LoadPointsFromFile', - coord_type='LIDAR', - load_dim=5, - use_dim=5, - file_client_args=dict(backend='disk')), - dict( - type='MultiScaleFlipAug3D', - img_scale=(1333, 800), - pts_scale_ratio=1, - flip=False, - transforms=[ - dict( - type='DefaultFormatBundle3D', - class_names=[ - 'car', 'truck', 'construction_vehicle', 'bus', - 'trailer', 'barrier', 'motorcycle', 'bicycle', - 'pedestrian', 'traffic_cone' - ], - with_label=False), - dict(type='Collect3D', keys=['points', 'img_inputs']) - ]) - ], - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - test_mode=True, - box_type_3d='LiDAR', - stereo=False, - filter_empty_gt=False, - img_info_prototype='bevdet'), - test=dict( - type='NuScenesDatasetOccpancy', - data_root='data/nuscenes/', - ann_file='data/nuscenes/bevdetv2-nuscenes_infos_val.pkl', - pipeline=[ - dict( - type='PrepareImageInputs', - data_config=dict( - cams=[ - 'CAM_FRONT_LEFT', 'CAM_FRONT', 'CAM_FRONT_RIGHT', - 'CAM_BACK_LEFT', 'CAM_BACK', 'CAM_BACK_RIGHT' - ], - Ncams=6, - input_size=(256, 704), - src_size=(900, 1600), - resize=(-0.06, 0.11), - rot=(-5.4, 5.4), - flip=True, - crop_h=(0.0, 0.0), - resize_test=0.0), - sequential=False), - dict( - type='LoadAnnotationsBEVDepth', - bda_aug_conf=dict( - rot_lim=(-0.0, 0.0), - scale_lim=(1.0, 1.0), - flip_dx_ratio=0.5, - flip_dy_ratio=0.5), - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', - 'traffic_cone' - ], - is_train=False), - dict( - type='LoadPointsFromFile', - coord_type='LIDAR', - load_dim=5, - use_dim=5, - file_client_args=dict(backend='disk')), - dict( - type='MultiScaleFlipAug3D', - img_scale=(1333, 800), - pts_scale_ratio=1, - flip=False, - transforms=[ - dict( - type='DefaultFormatBundle3D', - class_names=[ - 'car', 'truck', 'construction_vehicle', 'bus', - 'trailer', 'barrier', 'motorcycle', 'bicycle', - 'pedestrian', 'traffic_cone' - ], - with_label=False), - dict(type='Collect3D', keys=['points', 'img_inputs']) - ]) - ], - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - test_mode=True, - box_type_3d='LiDAR', - stereo=False, - filter_empty_gt=False, - img_info_prototype='bevdet')) -evaluation = dict( - interval=1, - pipeline=[ - dict( - type='PrepareImageInputs', - data_config=dict( - cams=[ - 'CAM_FRONT_LEFT', 'CAM_FRONT', 'CAM_FRONT_RIGHT', - 'CAM_BACK_LEFT', 'CAM_BACK', 'CAM_BACK_RIGHT' - ], - Ncams=6, - input_size=(256, 704), - src_size=(900, 1600), - resize=(-0.06, 0.11), - rot=(-5.4, 5.4), - flip=True, - crop_h=(0.0, 0.0), - resize_test=0.0), - sequential=False), - dict( - type='LoadAnnotationsBEVDepth', - bda_aug_conf=dict( - rot_lim=(-0.0, 0.0), - scale_lim=(1.0, 1.0), - flip_dx_ratio=0.5, - flip_dy_ratio=0.5), - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', - 'traffic_cone' - ], - is_train=False), - dict( - type='LoadPointsFromFile', - coord_type='LIDAR', - load_dim=5, - use_dim=5, - file_client_args=dict(backend='disk')), - dict( - type='MultiScaleFlipAug3D', - img_scale=(1333, 800), - pts_scale_ratio=1, - flip=False, - transforms=[ - dict( - type='DefaultFormatBundle3D', - class_names=[ - 'car', 'truck', 'construction_vehicle', 'bus', - 'trailer', 'barrier', 'motorcycle', 'bicycle', - 'pedestrian', 'traffic_cone' - ], - with_label=False), - dict(type='Collect3D', keys=['points', 'img_inputs']) - ]) - ], - start=20) -checkpoint_config = dict(interval=1, max_keep_ckpts=5) -log_config = dict( - interval=1, - hooks=[dict(type='TextLoggerHook'), - dict(type='TensorboardLoggerHook')]) -dist_params = dict(backend='nccl') -log_level = 'INFO' -work_dir = './work_dirs/flashocc-r50' -load_from = 'ckpts/bevdet-r50-cbgs.pth' -resume_from = None -workflow = [('train', 1)] -opencv_num_threads = 0 -mp_start_method = 'fork' -plugin = True -plugin_dir = 'projects/mmdet3d_plugin/' -data_config = dict( - cams=[ - 'CAM_FRONT_LEFT', 'CAM_FRONT', 'CAM_FRONT_RIGHT', 'CAM_BACK_LEFT', - 'CAM_BACK', 'CAM_BACK_RIGHT' - ], - Ncams=6, - input_size=(256, 704), - src_size=(900, 1600), - resize=(-0.06, 0.11), - rot=(-5.4, 5.4), - flip=True, - crop_h=(0.0, 0.0), - resize_test=0.0) -grid_config = dict( - x=[-40, 40, 0.4], - y=[-40, 40, 0.4], - z=[-1, 5.4, 6.4], - depth=[1.0, 45.0, 0.5]) -voxel_size = [0.1, 0.1, 0.2] -numC_Trans = 64 -model = dict( - type='BEVDetOCC', - img_backbone=dict( - type='ResNet', - depth=50, - num_stages=4, - out_indices=(2, 3), - frozen_stages=-1, - norm_cfg=dict(type='BN', requires_grad=True), - norm_eval=False, - with_cp=True, - style='pytorch'), - img_neck=dict( - type='CustomFPN', - in_channels=[1024, 2048], - out_channels=256, - num_outs=1, - start_level=0, - out_ids=[0]), - img_view_transformer=dict( - type='LSSViewTransformer', - grid_config=dict( - x=[-40, 40, 0.4], - y=[-40, 40, 0.4], - z=[-1, 5.4, 6.4], - depth=[1.0, 45.0, 0.5]), - input_size=(256, 704), - in_channels=256, - out_channels=64, - sid=False, - collapse_z=True, - downsample=16), - img_bev_encoder_backbone=dict( - type='CustomResNet', numC_input=64, num_channels=[128, 256, 512]), - img_bev_encoder_neck=dict( - type='FPN_LSS', in_channels=640, out_channels=256), - occ_head=dict( - type='BEVOCCHead2D', - in_dim=256, - out_dim=256, - Dz=16, - use_mask=True, - num_classes=18, - use_predicter=True, - class_balance=False, - loss_occ=dict( - type='CrossEntropyLoss', - use_sigmoid=False, - ignore_index=255, - loss_weight=1.0))) -bda_aug_conf = dict( - rot_lim=(-0.0, 0.0), - scale_lim=(1.0, 1.0), - flip_dx_ratio=0.5, - flip_dy_ratio=0.5) -share_data_config = dict( - type='NuScenesDatasetOccpancy', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier', - 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - stereo=False, - filter_empty_gt=False, - img_info_prototype='bevdet') -test_data_config = dict( - pipeline=[ - dict( - type='PrepareImageInputs', - data_config=dict( - cams=[ - 'CAM_FRONT_LEFT', 'CAM_FRONT', 'CAM_FRONT_RIGHT', - 'CAM_BACK_LEFT', 'CAM_BACK', 'CAM_BACK_RIGHT' - ], - Ncams=6, - input_size=(256, 704), - src_size=(900, 1600), - resize=(-0.06, 0.11), - rot=(-5.4, 5.4), - flip=True, - crop_h=(0.0, 0.0), - resize_test=0.0), - sequential=False), - dict( - type='LoadAnnotationsBEVDepth', - bda_aug_conf=dict( - rot_lim=(-0.0, 0.0), - scale_lim=(1.0, 1.0), - flip_dx_ratio=0.5, - flip_dy_ratio=0.5), - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', - 'traffic_cone' - ], - is_train=False), - dict( - type='LoadPointsFromFile', - coord_type='LIDAR', - load_dim=5, - use_dim=5, - file_client_args=dict(backend='disk')), - dict( - type='MultiScaleFlipAug3D', - img_scale=(1333, 800), - pts_scale_ratio=1, - flip=False, - transforms=[ - dict( - type='DefaultFormatBundle3D', - class_names=[ - 'car', 'truck', 'construction_vehicle', 'bus', - 'trailer', 'barrier', 'motorcycle', 'bicycle', - 'pedestrian', 'traffic_cone' - ], - with_label=False), - dict(type='Collect3D', keys=['points', 'img_inputs']) - ]) - ], - ann_file='data/nuscenes/bevdetv2-nuscenes_infos_val.pkl', - type='NuScenesDatasetOccpancy', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier', - 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - stereo=False, - filter_empty_gt=False, - img_info_prototype='bevdet') -key = 'test' -optimizer = dict(type='AdamW', lr=0.0001, weight_decay=0.01) -optimizer_config = dict(grad_clip=dict(max_norm=5, norm_type=2)) -lr_config = dict( - policy='step', - warmup='linear', - warmup_iters=200, - warmup_ratio=0.001, - step=[24]) -runner = dict(type='EpochBasedRunner', max_epochs=24) -custom_hooks = [ - dict(type='MEGVIIEMAHook', init_updates=10560, priority='NORMAL') -] -gpu_ids = range(0, 8) - -2026-04-03 16:26:53,024 - mmdet - INFO - Set random seed to 0, deterministic: False -2026-04-03 16:26:53,275 - mmdet - INFO - initialize ResNet with init_cfg [{'type': 'Kaiming', 'layer': 'Conv2d'}, {'type': 'Constant', 'val': 1, 'layer': ['_BatchNorm', 'GroupNorm']}] -2026-04-03 16:26:53,383 - mmdet - INFO - initialize Bottleneck with init_cfg {'type': 'Constant', 'val': 0, 'override': {'name': 'norm3'}} -2026-04-03 16:26:53,383 - mmdet - INFO - initialize Bottleneck with init_cfg {'type': 'Constant', 'val': 0, 'override': {'name': 'norm3'}} -2026-04-03 16:26:53,383 - mmdet - INFO - initialize Bottleneck with init_cfg {'type': 'Constant', 'val': 0, 'override': {'name': 'norm3'}} -2026-04-03 16:26:53,384 - mmdet - INFO - initialize Bottleneck with init_cfg {'type': 'Constant', 'val': 0, 'override': {'name': 'norm3'}} -2026-04-03 16:26:53,385 - mmdet - INFO - initialize Bottleneck with init_cfg {'type': 'Constant', 'val': 0, 'override': {'name': 'norm3'}} -2026-04-03 16:26:53,385 - mmdet - INFO - initialize Bottleneck with init_cfg {'type': 'Constant', 'val': 0, 'override': {'name': 'norm3'}} -2026-04-03 16:26:53,386 - mmdet - INFO - initialize Bottleneck with init_cfg {'type': 'Constant', 'val': 0, 'override': {'name': 'norm3'}} -2026-04-03 16:26:53,387 - mmdet - INFO - initialize Bottleneck with init_cfg {'type': 'Constant', 'val': 0, 'override': {'name': 'norm3'}} -2026-04-03 16:26:53,388 - mmdet - INFO - initialize Bottleneck with init_cfg {'type': 'Constant', 'val': 0, 'override': {'name': 'norm3'}} -2026-04-03 16:26:53,388 - mmdet - INFO - initialize Bottleneck with init_cfg {'type': 'Constant', 'val': 0, 'override': {'name': 'norm3'}} -2026-04-03 16:26:53,389 - mmdet - INFO - initialize Bottleneck with init_cfg {'type': 'Constant', 'val': 0, 'override': {'name': 'norm3'}} -2026-04-03 16:26:53,390 - mmdet - INFO - initialize Bottleneck with init_cfg {'type': 'Constant', 'val': 0, 'override': {'name': 'norm3'}} -2026-04-03 16:26:53,391 - mmdet - INFO - initialize Bottleneck with init_cfg {'type': 'Constant', 'val': 0, 'override': {'name': 'norm3'}} -2026-04-03 16:26:53,393 - mmdet - INFO - initialize Bottleneck with init_cfg {'type': 'Constant', 'val': 0, 'override': {'name': 'norm3'}} -2026-04-03 16:26:53,396 - mmdet - INFO - initialize Bottleneck with init_cfg {'type': 'Constant', 'val': 0, 'override': {'name': 'norm3'}} -2026-04-03 16:26:53,399 - mmdet - INFO - initialize Bottleneck with init_cfg {'type': 'Constant', 'val': 0, 'override': {'name': 'norm3'}} -2026-04-03 16:26:53,410 - mmdet - INFO - initialize CustomFPN with init_cfg {'type': 'Xavier', 'layer': 'Conv2d', 'distribution': 'uniform'} -Name of parameter - Initialization information - -img_backbone.conv1.weight - torch.Size([64, 3, 7, 7]): -KaimingInit: a=0, mode=fan_out, nonlinearity=relu, distribution =normal, bias=0 - -img_backbone.bn1.weight - torch.Size([64]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.bn1.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer1.0.conv1.weight - torch.Size([64, 64, 1, 1]): -KaimingInit: a=0, mode=fan_out, nonlinearity=relu, distribution =normal, bias=0 - -img_backbone.layer1.0.bn1.weight - torch.Size([64]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer1.0.bn1.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer1.0.conv2.weight - torch.Size([64, 64, 3, 3]): -KaimingInit: a=0, mode=fan_out, nonlinearity=relu, distribution =normal, bias=0 - -img_backbone.layer1.0.bn2.weight - torch.Size([64]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer1.0.bn2.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer1.0.conv3.weight - torch.Size([256, 64, 1, 1]): -KaimingInit: a=0, mode=fan_out, nonlinearity=relu, distribution =normal, bias=0 - -img_backbone.layer1.0.bn3.weight - torch.Size([256]): -ConstantInit: val=0, bias=0 - -img_backbone.layer1.0.bn3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer1.0.downsample.0.weight - torch.Size([256, 64, 1, 1]): -KaimingInit: a=0, mode=fan_out, nonlinearity=relu, distribution =normal, bias=0 - -img_backbone.layer1.0.downsample.1.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer1.0.downsample.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer1.1.conv1.weight - torch.Size([64, 256, 1, 1]): -KaimingInit: a=0, mode=fan_out, nonlinearity=relu, distribution =normal, bias=0 - -img_backbone.layer1.1.bn1.weight - torch.Size([64]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer1.1.bn1.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer1.1.conv2.weight - torch.Size([64, 64, 3, 3]): -KaimingInit: a=0, mode=fan_out, nonlinearity=relu, distribution =normal, bias=0 - -img_backbone.layer1.1.bn2.weight - torch.Size([64]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer1.1.bn2.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer1.1.conv3.weight - torch.Size([256, 64, 1, 1]): -KaimingInit: a=0, mode=fan_out, nonlinearity=relu, distribution =normal, bias=0 - -img_backbone.layer1.1.bn3.weight - torch.Size([256]): -ConstantInit: val=0, bias=0 - -img_backbone.layer1.1.bn3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer1.2.conv1.weight - torch.Size([64, 256, 1, 1]): -KaimingInit: a=0, mode=fan_out, nonlinearity=relu, distribution =normal, bias=0 - -img_backbone.layer1.2.bn1.weight - torch.Size([64]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer1.2.bn1.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer1.2.conv2.weight - torch.Size([64, 64, 3, 3]): -KaimingInit: a=0, mode=fan_out, nonlinearity=relu, distribution =normal, bias=0 - -img_backbone.layer1.2.bn2.weight - torch.Size([64]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer1.2.bn2.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer1.2.conv3.weight - torch.Size([256, 64, 1, 1]): -KaimingInit: a=0, mode=fan_out, nonlinearity=relu, distribution =normal, bias=0 - -img_backbone.layer1.2.bn3.weight - torch.Size([256]): -ConstantInit: val=0, bias=0 - -img_backbone.layer1.2.bn3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer2.0.conv1.weight - torch.Size([128, 256, 1, 1]): -KaimingInit: a=0, mode=fan_out, nonlinearity=relu, distribution =normal, bias=0 - -img_backbone.layer2.0.bn1.weight - torch.Size([128]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer2.0.bn1.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer2.0.conv2.weight - torch.Size([128, 128, 3, 3]): -KaimingInit: a=0, mode=fan_out, nonlinearity=relu, distribution =normal, bias=0 - -img_backbone.layer2.0.bn2.weight - torch.Size([128]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer2.0.bn2.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer2.0.conv3.weight - torch.Size([512, 128, 1, 1]): -KaimingInit: a=0, mode=fan_out, nonlinearity=relu, distribution =normal, bias=0 - -img_backbone.layer2.0.bn3.weight - torch.Size([512]): -ConstantInit: val=0, bias=0 - -img_backbone.layer2.0.bn3.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer2.0.downsample.0.weight - torch.Size([512, 256, 1, 1]): -KaimingInit: a=0, mode=fan_out, nonlinearity=relu, distribution =normal, bias=0 - -img_backbone.layer2.0.downsample.1.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer2.0.downsample.1.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer2.1.conv1.weight - torch.Size([128, 512, 1, 1]): -KaimingInit: a=0, mode=fan_out, nonlinearity=relu, distribution =normal, bias=0 - -img_backbone.layer2.1.bn1.weight - torch.Size([128]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer2.1.bn1.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer2.1.conv2.weight - torch.Size([128, 128, 3, 3]): -KaimingInit: a=0, mode=fan_out, nonlinearity=relu, distribution =normal, bias=0 - -img_backbone.layer2.1.bn2.weight - torch.Size([128]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer2.1.bn2.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer2.1.conv3.weight - torch.Size([512, 128, 1, 1]): -KaimingInit: a=0, mode=fan_out, nonlinearity=relu, distribution =normal, bias=0 - -img_backbone.layer2.1.bn3.weight - torch.Size([512]): -ConstantInit: val=0, bias=0 - -img_backbone.layer2.1.bn3.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer2.2.conv1.weight - torch.Size([128, 512, 1, 1]): -KaimingInit: a=0, mode=fan_out, nonlinearity=relu, distribution =normal, bias=0 - -img_backbone.layer2.2.bn1.weight - torch.Size([128]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer2.2.bn1.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer2.2.conv2.weight - torch.Size([128, 128, 3, 3]): -KaimingInit: a=0, mode=fan_out, nonlinearity=relu, distribution =normal, bias=0 - -img_backbone.layer2.2.bn2.weight - torch.Size([128]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer2.2.bn2.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer2.2.conv3.weight - torch.Size([512, 128, 1, 1]): -KaimingInit: a=0, mode=fan_out, nonlinearity=relu, distribution =normal, bias=0 - -img_backbone.layer2.2.bn3.weight - torch.Size([512]): -ConstantInit: val=0, bias=0 - -img_backbone.layer2.2.bn3.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer2.3.conv1.weight - torch.Size([128, 512, 1, 1]): -KaimingInit: a=0, mode=fan_out, nonlinearity=relu, distribution =normal, bias=0 - -img_backbone.layer2.3.bn1.weight - torch.Size([128]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer2.3.bn1.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer2.3.conv2.weight - torch.Size([128, 128, 3, 3]): -KaimingInit: a=0, mode=fan_out, nonlinearity=relu, distribution =normal, bias=0 - -img_backbone.layer2.3.bn2.weight - torch.Size([128]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer2.3.bn2.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer2.3.conv3.weight - torch.Size([512, 128, 1, 1]): -KaimingInit: a=0, mode=fan_out, nonlinearity=relu, distribution =normal, bias=0 - -img_backbone.layer2.3.bn3.weight - torch.Size([512]): -ConstantInit: val=0, bias=0 - -img_backbone.layer2.3.bn3.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer3.0.conv1.weight - torch.Size([256, 512, 1, 1]): -KaimingInit: a=0, mode=fan_out, nonlinearity=relu, distribution =normal, bias=0 - -img_backbone.layer3.0.bn1.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer3.0.bn1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer3.0.conv2.weight - torch.Size([256, 256, 3, 3]): -KaimingInit: a=0, mode=fan_out, nonlinearity=relu, distribution =normal, bias=0 - -img_backbone.layer3.0.bn2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer3.0.bn2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer3.0.conv3.weight - torch.Size([1024, 256, 1, 1]): -KaimingInit: a=0, mode=fan_out, nonlinearity=relu, distribution =normal, bias=0 - -img_backbone.layer3.0.bn3.weight - torch.Size([1024]): -ConstantInit: val=0, bias=0 - -img_backbone.layer3.0.bn3.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer3.0.downsample.0.weight - torch.Size([1024, 512, 1, 1]): -KaimingInit: a=0, mode=fan_out, nonlinearity=relu, distribution =normal, bias=0 - -img_backbone.layer3.0.downsample.1.weight - torch.Size([1024]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer3.0.downsample.1.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer3.1.conv1.weight - torch.Size([256, 1024, 1, 1]): -KaimingInit: a=0, mode=fan_out, nonlinearity=relu, distribution =normal, bias=0 - -img_backbone.layer3.1.bn1.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer3.1.bn1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer3.1.conv2.weight - torch.Size([256, 256, 3, 3]): -KaimingInit: a=0, mode=fan_out, nonlinearity=relu, distribution =normal, bias=0 - -img_backbone.layer3.1.bn2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer3.1.bn2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer3.1.conv3.weight - torch.Size([1024, 256, 1, 1]): -KaimingInit: a=0, mode=fan_out, nonlinearity=relu, distribution =normal, bias=0 - -img_backbone.layer3.1.bn3.weight - torch.Size([1024]): -ConstantInit: val=0, bias=0 - -img_backbone.layer3.1.bn3.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer3.2.conv1.weight - torch.Size([256, 1024, 1, 1]): -KaimingInit: a=0, mode=fan_out, nonlinearity=relu, distribution =normal, bias=0 - -img_backbone.layer3.2.bn1.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer3.2.bn1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer3.2.conv2.weight - torch.Size([256, 256, 3, 3]): -KaimingInit: a=0, mode=fan_out, nonlinearity=relu, distribution =normal, bias=0 - -img_backbone.layer3.2.bn2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer3.2.bn2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer3.2.conv3.weight - torch.Size([1024, 256, 1, 1]): -KaimingInit: a=0, mode=fan_out, nonlinearity=relu, distribution =normal, bias=0 - -img_backbone.layer3.2.bn3.weight - torch.Size([1024]): -ConstantInit: val=0, bias=0 - -img_backbone.layer3.2.bn3.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer3.3.conv1.weight - torch.Size([256, 1024, 1, 1]): -KaimingInit: a=0, mode=fan_out, nonlinearity=relu, distribution =normal, bias=0 - -img_backbone.layer3.3.bn1.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer3.3.bn1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer3.3.conv2.weight - torch.Size([256, 256, 3, 3]): -KaimingInit: a=0, mode=fan_out, nonlinearity=relu, distribution =normal, bias=0 - -img_backbone.layer3.3.bn2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer3.3.bn2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer3.3.conv3.weight - torch.Size([1024, 256, 1, 1]): -KaimingInit: a=0, mode=fan_out, nonlinearity=relu, distribution =normal, bias=0 - -img_backbone.layer3.3.bn3.weight - torch.Size([1024]): -ConstantInit: val=0, bias=0 - -img_backbone.layer3.3.bn3.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer3.4.conv1.weight - torch.Size([256, 1024, 1, 1]): -KaimingInit: a=0, mode=fan_out, nonlinearity=relu, distribution =normal, bias=0 - -img_backbone.layer3.4.bn1.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer3.4.bn1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer3.4.conv2.weight - torch.Size([256, 256, 3, 3]): -KaimingInit: a=0, mode=fan_out, nonlinearity=relu, distribution =normal, bias=0 - -img_backbone.layer3.4.bn2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer3.4.bn2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer3.4.conv3.weight - torch.Size([1024, 256, 1, 1]): -KaimingInit: a=0, mode=fan_out, nonlinearity=relu, distribution =normal, bias=0 - -img_backbone.layer3.4.bn3.weight - torch.Size([1024]): -ConstantInit: val=0, bias=0 - -img_backbone.layer3.4.bn3.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer3.5.conv1.weight - torch.Size([256, 1024, 1, 1]): -KaimingInit: a=0, mode=fan_out, nonlinearity=relu, distribution =normal, bias=0 - -img_backbone.layer3.5.bn1.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer3.5.bn1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer3.5.conv2.weight - torch.Size([256, 256, 3, 3]): -KaimingInit: a=0, mode=fan_out, nonlinearity=relu, distribution =normal, bias=0 - -img_backbone.layer3.5.bn2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer3.5.bn2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer3.5.conv3.weight - torch.Size([1024, 256, 1, 1]): -KaimingInit: a=0, mode=fan_out, nonlinearity=relu, distribution =normal, bias=0 - -img_backbone.layer3.5.bn3.weight - torch.Size([1024]): -ConstantInit: val=0, bias=0 - -img_backbone.layer3.5.bn3.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer4.0.conv1.weight - torch.Size([512, 1024, 1, 1]): -KaimingInit: a=0, mode=fan_out, nonlinearity=relu, distribution =normal, bias=0 - -img_backbone.layer4.0.bn1.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer4.0.bn1.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer4.0.conv2.weight - torch.Size([512, 512, 3, 3]): -KaimingInit: a=0, mode=fan_out, nonlinearity=relu, distribution =normal, bias=0 - -img_backbone.layer4.0.bn2.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer4.0.bn2.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer4.0.conv3.weight - torch.Size([2048, 512, 1, 1]): -KaimingInit: a=0, mode=fan_out, nonlinearity=relu, distribution =normal, bias=0 - -img_backbone.layer4.0.bn3.weight - torch.Size([2048]): -ConstantInit: val=0, bias=0 - -img_backbone.layer4.0.bn3.bias - torch.Size([2048]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer4.0.downsample.0.weight - torch.Size([2048, 1024, 1, 1]): -KaimingInit: a=0, mode=fan_out, nonlinearity=relu, distribution =normal, bias=0 - -img_backbone.layer4.0.downsample.1.weight - torch.Size([2048]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer4.0.downsample.1.bias - torch.Size([2048]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer4.1.conv1.weight - torch.Size([512, 2048, 1, 1]): -KaimingInit: a=0, mode=fan_out, nonlinearity=relu, distribution =normal, bias=0 - -img_backbone.layer4.1.bn1.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer4.1.bn1.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer4.1.conv2.weight - torch.Size([512, 512, 3, 3]): -KaimingInit: a=0, mode=fan_out, nonlinearity=relu, distribution =normal, bias=0 - -img_backbone.layer4.1.bn2.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer4.1.bn2.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer4.1.conv3.weight - torch.Size([2048, 512, 1, 1]): -KaimingInit: a=0, mode=fan_out, nonlinearity=relu, distribution =normal, bias=0 - -img_backbone.layer4.1.bn3.weight - torch.Size([2048]): -ConstantInit: val=0, bias=0 - -img_backbone.layer4.1.bn3.bias - torch.Size([2048]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer4.2.conv1.weight - torch.Size([512, 2048, 1, 1]): -KaimingInit: a=0, mode=fan_out, nonlinearity=relu, distribution =normal, bias=0 - -img_backbone.layer4.2.bn1.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer4.2.bn1.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer4.2.conv2.weight - torch.Size([512, 512, 3, 3]): -KaimingInit: a=0, mode=fan_out, nonlinearity=relu, distribution =normal, bias=0 - -img_backbone.layer4.2.bn2.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer4.2.bn2.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_backbone.layer4.2.conv3.weight - torch.Size([2048, 512, 1, 1]): -KaimingInit: a=0, mode=fan_out, nonlinearity=relu, distribution =normal, bias=0 - -img_backbone.layer4.2.bn3.weight - torch.Size([2048]): -ConstantInit: val=0, bias=0 - -img_backbone.layer4.2.bn3.bias - torch.Size([2048]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_neck.lateral_convs.0.conv.weight - torch.Size([256, 1024, 1, 1]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.lateral_convs.0.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_neck.lateral_convs.1.conv.weight - torch.Size([256, 2048, 1, 1]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.lateral_convs.1.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_neck.fpn_convs.0.conv.weight - torch.Size([256, 256, 3, 3]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.fpn_convs.0.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_view_transformer.depth_net.weight - torch.Size([152, 256, 1, 1]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_view_transformer.depth_net.bias - torch.Size([152]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_bev_encoder_backbone.layers.0.0.conv1.weight - torch.Size([128, 64, 3, 3]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_bev_encoder_backbone.layers.0.0.bn1.weight - torch.Size([128]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_bev_encoder_backbone.layers.0.0.bn1.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_bev_encoder_backbone.layers.0.0.conv2.weight - torch.Size([128, 128, 3, 3]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_bev_encoder_backbone.layers.0.0.bn2.weight - torch.Size([128]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_bev_encoder_backbone.layers.0.0.bn2.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_bev_encoder_backbone.layers.0.0.downsample.weight - torch.Size([128, 64, 3, 3]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_bev_encoder_backbone.layers.0.0.downsample.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_bev_encoder_backbone.layers.0.1.conv1.weight - torch.Size([128, 128, 3, 3]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_bev_encoder_backbone.layers.0.1.bn1.weight - torch.Size([128]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_bev_encoder_backbone.layers.0.1.bn1.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_bev_encoder_backbone.layers.0.1.conv2.weight - torch.Size([128, 128, 3, 3]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_bev_encoder_backbone.layers.0.1.bn2.weight - torch.Size([128]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_bev_encoder_backbone.layers.0.1.bn2.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_bev_encoder_backbone.layers.1.0.conv1.weight - torch.Size([256, 128, 3, 3]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_bev_encoder_backbone.layers.1.0.bn1.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_bev_encoder_backbone.layers.1.0.bn1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_bev_encoder_backbone.layers.1.0.conv2.weight - torch.Size([256, 256, 3, 3]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_bev_encoder_backbone.layers.1.0.bn2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_bev_encoder_backbone.layers.1.0.bn2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_bev_encoder_backbone.layers.1.0.downsample.weight - torch.Size([256, 128, 3, 3]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_bev_encoder_backbone.layers.1.0.downsample.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_bev_encoder_backbone.layers.1.1.conv1.weight - torch.Size([256, 256, 3, 3]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_bev_encoder_backbone.layers.1.1.bn1.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_bev_encoder_backbone.layers.1.1.bn1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_bev_encoder_backbone.layers.1.1.conv2.weight - torch.Size([256, 256, 3, 3]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_bev_encoder_backbone.layers.1.1.bn2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_bev_encoder_backbone.layers.1.1.bn2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_bev_encoder_backbone.layers.2.0.conv1.weight - torch.Size([512, 256, 3, 3]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_bev_encoder_backbone.layers.2.0.bn1.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_bev_encoder_backbone.layers.2.0.bn1.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_bev_encoder_backbone.layers.2.0.conv2.weight - torch.Size([512, 512, 3, 3]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_bev_encoder_backbone.layers.2.0.bn2.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_bev_encoder_backbone.layers.2.0.bn2.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_bev_encoder_backbone.layers.2.0.downsample.weight - torch.Size([512, 256, 3, 3]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_bev_encoder_backbone.layers.2.0.downsample.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_bev_encoder_backbone.layers.2.1.conv1.weight - torch.Size([512, 512, 3, 3]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_bev_encoder_backbone.layers.2.1.bn1.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_bev_encoder_backbone.layers.2.1.bn1.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_bev_encoder_backbone.layers.2.1.conv2.weight - torch.Size([512, 512, 3, 3]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_bev_encoder_backbone.layers.2.1.bn2.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_bev_encoder_backbone.layers.2.1.bn2.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_bev_encoder_neck.conv.0.weight - torch.Size([512, 640, 3, 3]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_bev_encoder_neck.conv.1.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_bev_encoder_neck.conv.1.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_bev_encoder_neck.conv.3.weight - torch.Size([512, 512, 3, 3]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_bev_encoder_neck.conv.4.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_bev_encoder_neck.conv.4.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_bev_encoder_neck.up2.1.weight - torch.Size([256, 512, 3, 3]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_bev_encoder_neck.up2.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_bev_encoder_neck.up2.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_bev_encoder_neck.up2.4.weight - torch.Size([256, 256, 1, 1]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -img_bev_encoder_neck.up2.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -occ_head.final_conv.conv.weight - torch.Size([256, 256, 3, 3]): -Initialized by user-defined `init_weights` in ConvModule - -occ_head.final_conv.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -occ_head.predicter.0.weight - torch.Size([512, 256]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -occ_head.predicter.0.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -occ_head.predicter.2.weight - torch.Size([288, 512]): -The value is the same before and after calling `init_weights` of BEVDetOCC - -occ_head.predicter.2.bias - torch.Size([288]): -The value is the same before and after calling `init_weights` of BEVDetOCC -2026-04-03 16:26:53,421 - mmdet - INFO - Model: -BEVDetOCC( - (img_backbone): ResNet( - (conv1): Conv2d(3, 64, kernel_size=(7, 7), stride=(2, 2), padding=(3, 3), bias=False) - (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (maxpool): MaxPool2d(kernel_size=3, stride=2, padding=1, dilation=1, ceil_mode=False) - (layer1): ResLayer( - (0): Bottleneck( - (conv1): Conv2d(64, 64, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (downsample): Sequential( - (0): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - ) - ) - init_cfg={'type': 'Constant', 'val': 0, 'override': {'name': 'norm3'}} - (1): Bottleneck( - (conv1): Conv2d(256, 64, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - init_cfg={'type': 'Constant', 'val': 0, 'override': {'name': 'norm3'}} - (2): Bottleneck( - (conv1): Conv2d(256, 64, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - init_cfg={'type': 'Constant', 'val': 0, 'override': {'name': 'norm3'}} - ) - (layer2): ResLayer( - (0): Bottleneck( - (conv1): Conv2d(256, 128, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (downsample): Sequential( - (0): Conv2d(256, 512, kernel_size=(1, 1), stride=(2, 2), bias=False) - (1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - ) - ) - init_cfg={'type': 'Constant', 'val': 0, 'override': {'name': 'norm3'}} - (1): Bottleneck( - (conv1): Conv2d(512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - init_cfg={'type': 'Constant', 'val': 0, 'override': {'name': 'norm3'}} - (2): Bottleneck( - (conv1): Conv2d(512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - init_cfg={'type': 'Constant', 'val': 0, 'override': {'name': 'norm3'}} - (3): Bottleneck( - (conv1): Conv2d(512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - init_cfg={'type': 'Constant', 'val': 0, 'override': {'name': 'norm3'}} - ) - (layer3): ResLayer( - (0): Bottleneck( - (conv1): Conv2d(512, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (downsample): Sequential( - (0): Conv2d(512, 1024, kernel_size=(1, 1), stride=(2, 2), bias=False) - (1): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - ) - ) - init_cfg={'type': 'Constant', 'val': 0, 'override': {'name': 'norm3'}} - (1): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - init_cfg={'type': 'Constant', 'val': 0, 'override': {'name': 'norm3'}} - (2): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - init_cfg={'type': 'Constant', 'val': 0, 'override': {'name': 'norm3'}} - (3): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - init_cfg={'type': 'Constant', 'val': 0, 'override': {'name': 'norm3'}} - (4): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - init_cfg={'type': 'Constant', 'val': 0, 'override': {'name': 'norm3'}} - (5): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - init_cfg={'type': 'Constant', 'val': 0, 'override': {'name': 'norm3'}} - ) - (layer4): ResLayer( - (0): Bottleneck( - (conv1): Conv2d(1024, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(512, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (downsample): Sequential( - (0): Conv2d(1024, 2048, kernel_size=(1, 1), stride=(2, 2), bias=False) - (1): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - ) - ) - init_cfg={'type': 'Constant', 'val': 0, 'override': {'name': 'norm3'}} - (1): Bottleneck( - (conv1): Conv2d(2048, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(512, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - init_cfg={'type': 'Constant', 'val': 0, 'override': {'name': 'norm3'}} - (2): Bottleneck( - (conv1): Conv2d(2048, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(512, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - init_cfg={'type': 'Constant', 'val': 0, 'override': {'name': 'norm3'}} - ) - ) - init_cfg=[{'type': 'Kaiming', 'layer': 'Conv2d'}, {'type': 'Constant', 'val': 1, 'layer': ['_BatchNorm', 'GroupNorm']}] - (img_neck): CustomFPN( - (lateral_convs): ModuleList( - (0): ConvModule( - (conv): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1)) - ) - (1): ConvModule( - (conv): Conv2d(2048, 256, kernel_size=(1, 1), stride=(1, 1)) - ) - ) - (fpn_convs): ModuleList( - (0): ConvModule( - (conv): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1)) - ) - ) - ) - init_cfg={'type': 'Xavier', 'layer': 'Conv2d', 'distribution': 'uniform'} - (img_view_transformer): LSSViewTransformer( - (depth_net): Conv2d(256, 152, kernel_size=(1, 1), stride=(1, 1)) - ) - (img_bev_encoder_backbone): CustomResNet( - (layers): Sequential( - (0): Sequential( - (0): BasicBlock( - (conv1): Conv2d(64, 128, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False) - (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (downsample): Conv2d(64, 128, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1)) - ) - (1): BasicBlock( - (conv1): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - ) - (1): Sequential( - (0): BasicBlock( - (conv1): Conv2d(128, 256, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (downsample): Conv2d(128, 256, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1)) - ) - (1): BasicBlock( - (conv1): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - ) - (2): Sequential( - (0): BasicBlock( - (conv1): Conv2d(256, 512, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False) - (bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (downsample): Conv2d(256, 512, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1)) - ) - (1): BasicBlock( - (conv1): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - ) - ) - ) - (img_bev_encoder_neck): FPN_LSS( - (up): Upsample(scale_factor=4.0, mode='bilinear') - (conv): Sequential( - (0): Conv2d(640, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (2): ReLU(inplace=True) - (3): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (4): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (5): ReLU(inplace=True) - ) - (up2): Sequential( - (0): Upsample(scale_factor=2.0, mode='bilinear') - (1): Conv2d(512, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (3): ReLU(inplace=True) - (4): Conv2d(256, 256, kernel_size=(1, 1), stride=(1, 1)) - ) - ) - (occ_head): BEVOCCHead2D( - (final_conv): ConvModule( - (conv): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1)) - (activate): ReLU(inplace=True) - ) - (predicter): Sequential( - (0): Linear(in_features=256, out_features=512, bias=True) - (1): Softplus(beta=1.0, threshold=20.0) - (2): Linear(in_features=512, out_features=288, bias=True) - ) - (loss_occ): CrossEntropyLoss(avg_non_ignore=False) - ) -) -2026-04-03 16:26:58,319 - mmdet - INFO - load checkpoint from local path: ckpts/bevdet-r50-cbgs.pth -2026-04-03 16:26:58,420 - mmdet - WARNING - The model and loaded state dict do not match exactly - -size mismatch for img_view_transformer.depth_net.weight: copying a param with shape torch.Size([123, 256, 1, 1]) from checkpoint, the shape in current model is torch.Size([152, 256, 1, 1]). -size mismatch for img_view_transformer.depth_net.bias: copying a param with shape torch.Size([123]) from checkpoint, the shape in current model is torch.Size([152]). -unexpected key in source state_dict: pts_bbox_head.shared_conv.conv.weight, pts_bbox_head.shared_conv.bn.weight, pts_bbox_head.shared_conv.bn.bias, pts_bbox_head.shared_conv.bn.running_mean, pts_bbox_head.shared_conv.bn.running_var, pts_bbox_head.shared_conv.bn.num_batches_tracked, pts_bbox_head.task_heads.0.reg.0.conv.weight, pts_bbox_head.task_heads.0.reg.0.bn.weight, pts_bbox_head.task_heads.0.reg.0.bn.bias, pts_bbox_head.task_heads.0.reg.0.bn.running_mean, pts_bbox_head.task_heads.0.reg.0.bn.running_var, pts_bbox_head.task_heads.0.reg.0.bn.num_batches_tracked, pts_bbox_head.task_heads.0.reg.1.weight, pts_bbox_head.task_heads.0.reg.1.bias, pts_bbox_head.task_heads.0.height.0.conv.weight, pts_bbox_head.task_heads.0.height.0.bn.weight, pts_bbox_head.task_heads.0.height.0.bn.bias, pts_bbox_head.task_heads.0.height.0.bn.running_mean, pts_bbox_head.task_heads.0.height.0.bn.running_var, pts_bbox_head.task_heads.0.height.0.bn.num_batches_tracked, pts_bbox_head.task_heads.0.height.1.weight, pts_bbox_head.task_heads.0.height.1.bias, pts_bbox_head.task_heads.0.dim.0.conv.weight, pts_bbox_head.task_heads.0.dim.0.bn.weight, pts_bbox_head.task_heads.0.dim.0.bn.bias, pts_bbox_head.task_heads.0.dim.0.bn.running_mean, pts_bbox_head.task_heads.0.dim.0.bn.running_var, pts_bbox_head.task_heads.0.dim.0.bn.num_batches_tracked, pts_bbox_head.task_heads.0.dim.1.weight, pts_bbox_head.task_heads.0.dim.1.bias, pts_bbox_head.task_heads.0.rot.0.conv.weight, pts_bbox_head.task_heads.0.rot.0.bn.weight, pts_bbox_head.task_heads.0.rot.0.bn.bias, pts_bbox_head.task_heads.0.rot.0.bn.running_mean, pts_bbox_head.task_heads.0.rot.0.bn.running_var, pts_bbox_head.task_heads.0.rot.0.bn.num_batches_tracked, pts_bbox_head.task_heads.0.rot.1.weight, pts_bbox_head.task_heads.0.rot.1.bias, pts_bbox_head.task_heads.0.vel.0.conv.weight, pts_bbox_head.task_heads.0.vel.0.bn.weight, pts_bbox_head.task_heads.0.vel.0.bn.bias, pts_bbox_head.task_heads.0.vel.0.bn.running_mean, pts_bbox_head.task_heads.0.vel.0.bn.running_var, pts_bbox_head.task_heads.0.vel.0.bn.num_batches_tracked, pts_bbox_head.task_heads.0.vel.1.weight, pts_bbox_head.task_heads.0.vel.1.bias, pts_bbox_head.task_heads.0.heatmap.0.conv.weight, pts_bbox_head.task_heads.0.heatmap.0.bn.weight, pts_bbox_head.task_heads.0.heatmap.0.bn.bias, pts_bbox_head.task_heads.0.heatmap.0.bn.running_mean, pts_bbox_head.task_heads.0.heatmap.0.bn.running_var, pts_bbox_head.task_heads.0.heatmap.0.bn.num_batches_tracked, pts_bbox_head.task_heads.0.heatmap.1.weight, pts_bbox_head.task_heads.0.heatmap.1.bias - -missing keys in source state_dict: occ_head.final_conv.conv.weight, occ_head.final_conv.conv.bias, occ_head.predicter.0.weight, occ_head.predicter.0.bias, occ_head.predicter.2.weight, occ_head.predicter.2.bias - -2026-04-03 16:26:58,422 - mmdet - INFO - Start running, host: root@bw61, work_dir: /workspace/Flashocc/work_dirs/flashocc-r50 -2026-04-03 16:26:58,422 - mmdet - INFO - Hooks will be executed in the following order: -before_run: -(VERY_HIGH ) StepLrUpdaterHook -(NORMAL ) CheckpointHook -(NORMAL ) MEGVIIEMAHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -before_train_epoch: -(VERY_HIGH ) StepLrUpdaterHook -(NORMAL ) DistSamplerSeedHook -(LOW ) IterTimerHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -before_train_iter: -(VERY_HIGH ) StepLrUpdaterHook -(LOW ) IterTimerHook - -------------------- -after_train_iter: -(ABOVE_NORMAL) OptimizerHook -(NORMAL ) CheckpointHook -(NORMAL ) MEGVIIEMAHook -(LOW ) IterTimerHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -after_train_epoch: -(NORMAL ) CheckpointHook -(NORMAL ) MEGVIIEMAHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -before_val_epoch: -(NORMAL ) DistSamplerSeedHook -(LOW ) IterTimerHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -before_val_iter: -(LOW ) IterTimerHook - -------------------- -after_val_iter: -(LOW ) IterTimerHook - -------------------- -after_val_epoch: -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -after_run: -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -2026-04-03 16:26:58,423 - mmdet - INFO - workflow: [('train', 1)], max: 24 epochs -2026-04-03 16:26:58,423 - mmdet - INFO - Checkpoints will be saved to /workspace/Flashocc/work_dirs/flashocc-r50 by HardDiskBackend. -2026-04-03 16:37:58,297 - mmdet - INFO - Epoch [1][1/147] lr: 1.000e-07, eta: 26 days, 22:22:17, time: 659.750, data_time: 15.432, memory: 32423, loss_occ: 3.0086, loss: 3.0086, grad_norm: 3.9001 -2026-04-03 16:38:03,001 - mmdet - INFO - Epoch [1][2/147] lr: 5.995e-07, eta: 13 days, 13:23:57, time: 4.707, data_time: 0.005, memory: 32769, loss_occ: 3.0129, loss: 3.0129, grad_norm: 3.9387 -2026-04-03 16:38:04,134 - mmdet - INFO - Epoch [1][3/147] lr: 1.099e-06, eta: 9 days, 1:14:27, time: 1.133, data_time: 0.003, memory: 32769, loss_occ: 3.0150, loss: 3.0150, grad_norm: 3.8985 -2026-04-03 16:38:05,261 - mmdet - INFO - Epoch [1][4/147] lr: 1.599e-06, eta: 6 days, 19:09:36, time: 1.126, data_time: 0.002, memory: 32769, loss_occ: 3.0085, loss: 3.0085, grad_norm: 3.8741 -2026-04-03 16:38:06,388 - mmdet - INFO - Epoch [1][5/147] lr: 2.098e-06, eta: 5 days, 10:42:41, time: 1.127, data_time: 0.003, memory: 32769, loss_occ: 3.0044, loss: 3.0044, grad_norm: 3.8639 -2026-04-03 16:38:07,513 - mmdet - INFO - Epoch [1][6/147] lr: 2.597e-06, eta: 4 days, 13:04:44, time: 1.126, data_time: 0.003, memory: 32769, loss_occ: 3.0085, loss: 3.0085, grad_norm: 3.8423 -2026-04-03 16:38:08,643 - mmdet - INFO - Epoch [1][7/147] lr: 3.097e-06, eta: 3 days, 21:37:37, time: 1.126, data_time: 0.003, memory: 32769, loss_occ: 3.0085, loss: 3.0085, grad_norm: 3.9473 -2026-04-03 16:38:09,771 - mmdet - INFO - Epoch [1][8/147] lr: 3.597e-06, eta: 3 days, 10:02:19, time: 1.130, data_time: 0.006, memory: 32769, loss_occ: 3.0038, loss: 3.0038, grad_norm: 3.9215 -2026-04-03 16:38:10,898 - mmdet - INFO - Epoch [1][9/147] lr: 4.096e-06, eta: 3 days, 1:01:30, time: 1.128, data_time: 0.004, memory: 32769, loss_occ: 2.9969, loss: 2.9969, grad_norm: 3.8659 -2026-04-03 16:38:12,026 - mmdet - INFO - Epoch [1][10/147] lr: 4.596e-06, eta: 2 days, 17:48:50, time: 1.127, data_time: 0.003, memory: 32769, loss_occ: 2.9957, loss: 2.9957, grad_norm: 3.8820 diff --git a/docker-hub/FlashOCC/Flashocc/work_dirs/flashocc-r50/20260403_162651.log.json b/docker-hub/FlashOCC/Flashocc/work_dirs/flashocc-r50/20260403_162651.log.json deleted file mode 100644 index 04c7364732d75b90a51b308c070bb7f6f0cb6682..0000000000000000000000000000000000000000 --- a/docker-hub/FlashOCC/Flashocc/work_dirs/flashocc-r50/20260403_162651.log.json +++ /dev/null @@ -1,11 +0,0 @@ -{"env_info": "sys.platform: linux\nPython: 3.10.12 (main, Aug 15 2025, 14:32:43) [GCC 11.4.0]\nCUDA available: True\nGPU 0,1,2,3,4,5,6,7: BW1000_H\nCUDA_HOME: /opt/dtk\nNVCC: Not Available\nGCC: x86_64-linux-gnu-gcc (Ubuntu 11.4.0-1ubuntu1~22.04) 11.4.0\nPyTorch: 2.5.1\nPyTorch compiling details: PyTorch built with:\n - GCC 10.3\n - C++ Version: 201703\n - Intel(R) Math Kernel Library Version 2020.0.4 Product Build 20200917 for Intel(R) 64 architecture applications\n - OpenMP 201511 (a.k.a. OpenMP 4.5)\n - LAPACK is enabled (usually provided by MKL)\n - NNPACK is enabled\n - CPU capability usage: AVX512\n - HIP Runtime 6.3.25521\n - MIOpen 2.18.0\n - Magma 2.8.0\n - Build settings: BLAS_INFO=mkl, BUILD_TYPE=Release, CXX_COMPILER=/opt/rh/gcc-toolset-10/root/usr/bin/c++, CXX_FLAGS= -D_GLIBCXX_USE_CXX11_ABI=1 -fvisibility-inlines-hidden -DUSE_PTHREADPOOL -DNDEBUG -DUSE_KINETO -DLIBKINETO_NOCUPTI -DLIBKINETO_NOXPUPTI=ON -DUSE_FBGEMM -DUSE_PYTORCH_QNNPACK -DUSE_XNNPACK -DSYMBOLICATE_MOBILE_DEBUG_HANDLE -O2 -fPIC -Wall -Wextra -Werror=return-type -Werror=non-virtual-dtor -Werror=bool-operation -Wnarrowing -Wno-missing-field-initializers -Wno-type-limits -Wno-array-bounds -Wno-unknown-pragmas -Wno-unused-parameter -Wno-strict-overflow -Wno-strict-aliasing -Wno-stringop-overflow -Wsuggest-override -Wno-psabi -Wno-error=old-style-cast -Wno-missing-braces -fdiagnostics-color=always -faligned-new -Wno-unused-but-set-variable -Wno-maybe-uninitialized -fno-math-errno -fno-trapping-math -Werror=format -Wno-stringop-overflow, FORCE_FALLBACK_CUDA_MPI=1, LAPACK_INFO=mkl, PERF_WITH_AVX=1, PERF_WITH_AVX2=1, TORCH_VERSION=2.5.1, USE_CUDA=0, USE_CUDNN=OFF, USE_CUSPARSELT=OFF, USE_EXCEPTION_PTR=1, USE_GFLAGS=1, USE_GLOG=1, USE_GLOO=1, USE_MKL=ON, USE_MKLDNN=0, USE_MPI=1, USE_NCCL=1, USE_NNPACK=ON, USE_OPENMP=1, USE_ROCM=ON, USE_ROCM_KERNEL_ASSERT=OFF, \n\nTorchVision: 0.20.1\nOpenCV: 4.12.0\nMMCV: 1.6.1\nMMCV Compiler: GCC 10.3\nMMCV CUDA Compiler: rocm not available\nMMDetection: 2.25.1\nMMSegmentation: 0.25.0\nMMDetection3D: 1.0.0rc4+\nspconv2.0: False", "config": "point_cloud_range = [-51.2, -51.2, -5.0, 51.2, 51.2, 3.0]\nclass_names = [\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier',\n 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n]\ndataset_type = 'NuScenesDatasetOccpancy'\ndata_root = 'data/nuscenes/'\ninput_modality = dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False)\nfile_client_args = dict(backend='disk')\ntrain_pipeline = [\n dict(\n type='PrepareImageInputs',\n is_train=True,\n data_config=dict(\n cams=[\n 'CAM_FRONT_LEFT', 'CAM_FRONT', 'CAM_FRONT_RIGHT',\n 'CAM_BACK_LEFT', 'CAM_BACK', 'CAM_BACK_RIGHT'\n ],\n Ncams=6,\n input_size=(256, 704),\n src_size=(900, 1600),\n resize=(-0.06, 0.11),\n rot=(-5.4, 5.4),\n flip=True,\n crop_h=(0.0, 0.0),\n resize_test=0.0),\n sequential=False),\n dict(\n type='LoadAnnotationsBEVDepth',\n bda_aug_conf=dict(\n rot_lim=(-0.0, 0.0),\n scale_lim=(1.0, 1.0),\n flip_dx_ratio=0.5,\n flip_dy_ratio=0.5),\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ],\n is_train=True),\n dict(type='LoadOccGTFromFile'),\n dict(\n type='LoadPointsFromFile',\n coord_type='LIDAR',\n load_dim=5,\n use_dim=5,\n file_client_args=dict(backend='disk')),\n dict(\n type='PointToMultiViewDepth',\n downsample=1,\n grid_config=dict(\n x=[-40, 40, 0.4],\n y=[-40, 40, 0.4],\n z=[-1, 5.4, 6.4],\n depth=[1.0, 45.0, 0.5])),\n dict(\n type='DefaultFormatBundle3D',\n class_names=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ]),\n dict(\n type='Collect3D',\n keys=[\n 'img_inputs', 'gt_depth', 'voxel_semantics', 'mask_lidar',\n 'mask_camera'\n ])\n]\ntest_pipeline = [\n dict(\n type='PrepareImageInputs',\n data_config=dict(\n cams=[\n 'CAM_FRONT_LEFT', 'CAM_FRONT', 'CAM_FRONT_RIGHT',\n 'CAM_BACK_LEFT', 'CAM_BACK', 'CAM_BACK_RIGHT'\n ],\n Ncams=6,\n input_size=(256, 704),\n src_size=(900, 1600),\n resize=(-0.06, 0.11),\n rot=(-5.4, 5.4),\n flip=True,\n crop_h=(0.0, 0.0),\n resize_test=0.0),\n sequential=False),\n dict(\n type='LoadAnnotationsBEVDepth',\n bda_aug_conf=dict(\n rot_lim=(-0.0, 0.0),\n scale_lim=(1.0, 1.0),\n flip_dx_ratio=0.5,\n flip_dy_ratio=0.5),\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ],\n is_train=False),\n dict(\n type='LoadPointsFromFile',\n coord_type='LIDAR',\n load_dim=5,\n use_dim=5,\n file_client_args=dict(backend='disk')),\n dict(\n type='MultiScaleFlipAug3D',\n img_scale=(1333, 800),\n pts_scale_ratio=1,\n flip=False,\n transforms=[\n dict(\n type='DefaultFormatBundle3D',\n class_names=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian',\n 'traffic_cone'\n ],\n with_label=False),\n dict(type='Collect3D', keys=['points', 'img_inputs'])\n ])\n]\neval_pipeline = [\n dict(\n type='LoadPointsFromFile',\n coord_type='LIDAR',\n load_dim=5,\n use_dim=5,\n file_client_args=dict(backend='disk')),\n dict(\n type='LoadPointsFromMultiSweeps',\n sweeps_num=10,\n file_client_args=dict(backend='disk')),\n dict(\n type='DefaultFormatBundle3D',\n class_names=[\n 'car', 'truck', 'trailer', 'bus', 'construction_vehicle',\n 'bicycle', 'motorcycle', 'pedestrian', 'traffic_cone', 'barrier'\n ],\n with_label=False),\n dict(type='Collect3D', keys=['points'])\n]\ndata = dict(\n samples_per_gpu=24,\n workers_per_gpu=24,\n train=dict(\n type='NuScenesDatasetOccpancy',\n data_root='data/nuscenes/',\n ann_file='data/nuscenes/bevdetv2-nuscenes_infos_train.pkl',\n pipeline=[\n dict(\n type='PrepareImageInputs',\n is_train=True,\n data_config=dict(\n cams=[\n 'CAM_FRONT_LEFT', 'CAM_FRONT', 'CAM_FRONT_RIGHT',\n 'CAM_BACK_LEFT', 'CAM_BACK', 'CAM_BACK_RIGHT'\n ],\n Ncams=6,\n input_size=(256, 704),\n src_size=(900, 1600),\n resize=(-0.06, 0.11),\n rot=(-5.4, 5.4),\n flip=True,\n crop_h=(0.0, 0.0),\n resize_test=0.0),\n sequential=False),\n dict(\n type='LoadAnnotationsBEVDepth',\n bda_aug_conf=dict(\n rot_lim=(-0.0, 0.0),\n scale_lim=(1.0, 1.0),\n flip_dx_ratio=0.5,\n flip_dy_ratio=0.5),\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian',\n 'traffic_cone'\n ],\n is_train=True),\n dict(type='LoadOccGTFromFile'),\n dict(\n type='LoadPointsFromFile',\n coord_type='LIDAR',\n load_dim=5,\n use_dim=5,\n file_client_args=dict(backend='disk')),\n dict(\n type='PointToMultiViewDepth',\n downsample=1,\n grid_config=dict(\n x=[-40, 40, 0.4],\n y=[-40, 40, 0.4],\n z=[-1, 5.4, 6.4],\n depth=[1.0, 45.0, 0.5])),\n dict(\n type='DefaultFormatBundle3D',\n class_names=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian',\n 'traffic_cone'\n ]),\n dict(\n type='Collect3D',\n keys=[\n 'img_inputs', 'gt_depth', 'voxel_semantics', 'mask_lidar',\n 'mask_camera'\n ])\n ],\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ],\n modality=dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False),\n test_mode=False,\n box_type_3d='LiDAR',\n use_valid_flag=True,\n stereo=False,\n filter_empty_gt=False,\n img_info_prototype='bevdet'),\n val=dict(\n type='NuScenesDatasetOccpancy',\n data_root='data/nuscenes/',\n ann_file='data/nuscenes/bevdetv2-nuscenes_infos_val.pkl',\n pipeline=[\n dict(\n type='PrepareImageInputs',\n data_config=dict(\n cams=[\n 'CAM_FRONT_LEFT', 'CAM_FRONT', 'CAM_FRONT_RIGHT',\n 'CAM_BACK_LEFT', 'CAM_BACK', 'CAM_BACK_RIGHT'\n ],\n Ncams=6,\n input_size=(256, 704),\n src_size=(900, 1600),\n resize=(-0.06, 0.11),\n rot=(-5.4, 5.4),\n flip=True,\n crop_h=(0.0, 0.0),\n resize_test=0.0),\n sequential=False),\n dict(\n type='LoadAnnotationsBEVDepth',\n bda_aug_conf=dict(\n rot_lim=(-0.0, 0.0),\n scale_lim=(1.0, 1.0),\n flip_dx_ratio=0.5,\n flip_dy_ratio=0.5),\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian',\n 'traffic_cone'\n ],\n is_train=False),\n dict(\n type='LoadPointsFromFile',\n coord_type='LIDAR',\n load_dim=5,\n use_dim=5,\n file_client_args=dict(backend='disk')),\n dict(\n type='MultiScaleFlipAug3D',\n img_scale=(1333, 800),\n pts_scale_ratio=1,\n flip=False,\n transforms=[\n dict(\n type='DefaultFormatBundle3D',\n class_names=[\n 'car', 'truck', 'construction_vehicle', 'bus',\n 'trailer', 'barrier', 'motorcycle', 'bicycle',\n 'pedestrian', 'traffic_cone'\n ],\n with_label=False),\n dict(type='Collect3D', keys=['points', 'img_inputs'])\n ])\n ],\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ],\n modality=dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False),\n test_mode=True,\n box_type_3d='LiDAR',\n stereo=False,\n filter_empty_gt=False,\n img_info_prototype='bevdet'),\n test=dict(\n type='NuScenesDatasetOccpancy',\n data_root='data/nuscenes/',\n ann_file='data/nuscenes/bevdetv2-nuscenes_infos_val.pkl',\n pipeline=[\n dict(\n type='PrepareImageInputs',\n data_config=dict(\n cams=[\n 'CAM_FRONT_LEFT', 'CAM_FRONT', 'CAM_FRONT_RIGHT',\n 'CAM_BACK_LEFT', 'CAM_BACK', 'CAM_BACK_RIGHT'\n ],\n Ncams=6,\n input_size=(256, 704),\n src_size=(900, 1600),\n resize=(-0.06, 0.11),\n rot=(-5.4, 5.4),\n flip=True,\n crop_h=(0.0, 0.0),\n resize_test=0.0),\n sequential=False),\n dict(\n type='LoadAnnotationsBEVDepth',\n bda_aug_conf=dict(\n rot_lim=(-0.0, 0.0),\n scale_lim=(1.0, 1.0),\n flip_dx_ratio=0.5,\n flip_dy_ratio=0.5),\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian',\n 'traffic_cone'\n ],\n is_train=False),\n dict(\n type='LoadPointsFromFile',\n coord_type='LIDAR',\n load_dim=5,\n use_dim=5,\n file_client_args=dict(backend='disk')),\n dict(\n type='MultiScaleFlipAug3D',\n img_scale=(1333, 800),\n pts_scale_ratio=1,\n flip=False,\n transforms=[\n dict(\n type='DefaultFormatBundle3D',\n class_names=[\n 'car', 'truck', 'construction_vehicle', 'bus',\n 'trailer', 'barrier', 'motorcycle', 'bicycle',\n 'pedestrian', 'traffic_cone'\n ],\n with_label=False),\n dict(type='Collect3D', keys=['points', 'img_inputs'])\n ])\n ],\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ],\n modality=dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False),\n test_mode=True,\n box_type_3d='LiDAR',\n stereo=False,\n filter_empty_gt=False,\n img_info_prototype='bevdet'))\nevaluation = dict(\n interval=1,\n pipeline=[\n dict(\n type='PrepareImageInputs',\n data_config=dict(\n cams=[\n 'CAM_FRONT_LEFT', 'CAM_FRONT', 'CAM_FRONT_RIGHT',\n 'CAM_BACK_LEFT', 'CAM_BACK', 'CAM_BACK_RIGHT'\n ],\n Ncams=6,\n input_size=(256, 704),\n src_size=(900, 1600),\n resize=(-0.06, 0.11),\n rot=(-5.4, 5.4),\n flip=True,\n crop_h=(0.0, 0.0),\n resize_test=0.0),\n sequential=False),\n dict(\n type='LoadAnnotationsBEVDepth',\n bda_aug_conf=dict(\n rot_lim=(-0.0, 0.0),\n scale_lim=(1.0, 1.0),\n flip_dx_ratio=0.5,\n flip_dy_ratio=0.5),\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian',\n 'traffic_cone'\n ],\n is_train=False),\n dict(\n type='LoadPointsFromFile',\n coord_type='LIDAR',\n load_dim=5,\n use_dim=5,\n file_client_args=dict(backend='disk')),\n dict(\n type='MultiScaleFlipAug3D',\n img_scale=(1333, 800),\n pts_scale_ratio=1,\n flip=False,\n transforms=[\n dict(\n type='DefaultFormatBundle3D',\n class_names=[\n 'car', 'truck', 'construction_vehicle', 'bus',\n 'trailer', 'barrier', 'motorcycle', 'bicycle',\n 'pedestrian', 'traffic_cone'\n ],\n with_label=False),\n dict(type='Collect3D', keys=['points', 'img_inputs'])\n ])\n ],\n start=20)\ncheckpoint_config = dict(interval=1, max_keep_ckpts=5)\nlog_config = dict(\n interval=1,\n hooks=[dict(type='TextLoggerHook'),\n dict(type='TensorboardLoggerHook')])\ndist_params = dict(backend='nccl')\nlog_level = 'INFO'\nwork_dir = './work_dirs/flashocc-r50'\nload_from = 'ckpts/bevdet-r50-cbgs.pth'\nresume_from = None\nworkflow = [('train', 1)]\nopencv_num_threads = 0\nmp_start_method = 'fork'\nplugin = True\nplugin_dir = 'projects/mmdet3d_plugin/'\ndata_config = dict(\n cams=[\n 'CAM_FRONT_LEFT', 'CAM_FRONT', 'CAM_FRONT_RIGHT', 'CAM_BACK_LEFT',\n 'CAM_BACK', 'CAM_BACK_RIGHT'\n ],\n Ncams=6,\n input_size=(256, 704),\n src_size=(900, 1600),\n resize=(-0.06, 0.11),\n rot=(-5.4, 5.4),\n flip=True,\n crop_h=(0.0, 0.0),\n resize_test=0.0)\ngrid_config = dict(\n x=[-40, 40, 0.4],\n y=[-40, 40, 0.4],\n z=[-1, 5.4, 6.4],\n depth=[1.0, 45.0, 0.5])\nvoxel_size = [0.1, 0.1, 0.2]\nnumC_Trans = 64\nmodel = dict(\n type='BEVDetOCC',\n img_backbone=dict(\n type='ResNet',\n depth=50,\n num_stages=4,\n out_indices=(2, 3),\n frozen_stages=-1,\n norm_cfg=dict(type='BN', requires_grad=True),\n norm_eval=False,\n with_cp=True,\n style='pytorch'),\n img_neck=dict(\n type='CustomFPN',\n in_channels=[1024, 2048],\n out_channels=256,\n num_outs=1,\n start_level=0,\n out_ids=[0]),\n img_view_transformer=dict(\n type='LSSViewTransformer',\n grid_config=dict(\n x=[-40, 40, 0.4],\n y=[-40, 40, 0.4],\n z=[-1, 5.4, 6.4],\n depth=[1.0, 45.0, 0.5]),\n input_size=(256, 704),\n in_channels=256,\n out_channels=64,\n sid=False,\n collapse_z=True,\n downsample=16),\n img_bev_encoder_backbone=dict(\n type='CustomResNet', numC_input=64, num_channels=[128, 256, 512]),\n img_bev_encoder_neck=dict(\n type='FPN_LSS', in_channels=640, out_channels=256),\n occ_head=dict(\n type='BEVOCCHead2D',\n in_dim=256,\n out_dim=256,\n Dz=16,\n use_mask=True,\n num_classes=18,\n use_predicter=True,\n class_balance=False,\n loss_occ=dict(\n type='CrossEntropyLoss',\n use_sigmoid=False,\n ignore_index=255,\n loss_weight=1.0)))\nbda_aug_conf = dict(\n rot_lim=(-0.0, 0.0),\n scale_lim=(1.0, 1.0),\n flip_dx_ratio=0.5,\n flip_dy_ratio=0.5)\nshare_data_config = dict(\n type='NuScenesDatasetOccpancy',\n data_root='data/nuscenes/',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier',\n 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ],\n modality=dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False),\n stereo=False,\n filter_empty_gt=False,\n img_info_prototype='bevdet')\ntest_data_config = dict(\n pipeline=[\n dict(\n type='PrepareImageInputs',\n data_config=dict(\n cams=[\n 'CAM_FRONT_LEFT', 'CAM_FRONT', 'CAM_FRONT_RIGHT',\n 'CAM_BACK_LEFT', 'CAM_BACK', 'CAM_BACK_RIGHT'\n ],\n Ncams=6,\n input_size=(256, 704),\n src_size=(900, 1600),\n resize=(-0.06, 0.11),\n rot=(-5.4, 5.4),\n flip=True,\n crop_h=(0.0, 0.0),\n resize_test=0.0),\n sequential=False),\n dict(\n type='LoadAnnotationsBEVDepth',\n bda_aug_conf=dict(\n rot_lim=(-0.0, 0.0),\n scale_lim=(1.0, 1.0),\n flip_dx_ratio=0.5,\n flip_dy_ratio=0.5),\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian',\n 'traffic_cone'\n ],\n is_train=False),\n dict(\n type='LoadPointsFromFile',\n coord_type='LIDAR',\n load_dim=5,\n use_dim=5,\n file_client_args=dict(backend='disk')),\n dict(\n type='MultiScaleFlipAug3D',\n img_scale=(1333, 800),\n pts_scale_ratio=1,\n flip=False,\n transforms=[\n dict(\n type='DefaultFormatBundle3D',\n class_names=[\n 'car', 'truck', 'construction_vehicle', 'bus',\n 'trailer', 'barrier', 'motorcycle', 'bicycle',\n 'pedestrian', 'traffic_cone'\n ],\n with_label=False),\n dict(type='Collect3D', keys=['points', 'img_inputs'])\n ])\n ],\n ann_file='data/nuscenes/bevdetv2-nuscenes_infos_val.pkl',\n type='NuScenesDatasetOccpancy',\n data_root='data/nuscenes/',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier',\n 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ],\n modality=dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False),\n stereo=False,\n filter_empty_gt=False,\n img_info_prototype='bevdet')\nkey = 'test'\noptimizer = dict(type='AdamW', lr=0.0001, weight_decay=0.01)\noptimizer_config = dict(grad_clip=dict(max_norm=5, norm_type=2))\nlr_config = dict(\n policy='step',\n warmup='linear',\n warmup_iters=200,\n warmup_ratio=0.001,\n step=[24])\nrunner = dict(type='EpochBasedRunner', max_epochs=24)\ncustom_hooks = [\n dict(type='MEGVIIEMAHook', init_updates=10560, priority='NORMAL')\n]\ngpu_ids = range(0, 8)\n", "seed": 0, "exp_name": "flashocc-r50.py"} -{"mode": "train", "epoch": 1, "iter": 1, "lr": 0.0, "memory": 32423, "data_time": 15.43241, "loss_occ": 3.00864, "loss": 3.00864, "grad_norm": 3.90007, "time": 659.74966} -{"mode": "train", "epoch": 1, "iter": 2, "lr": 0.0, "memory": 32769, "data_time": 0.00516, "loss_occ": 3.01287, "loss": 3.01287, "grad_norm": 3.9387, "time": 4.70719} -{"mode": "train", "epoch": 1, "iter": 3, "lr": 0.0, "memory": 32769, "data_time": 0.00255, "loss_occ": 3.015, "loss": 3.015, "grad_norm": 3.89853, "time": 1.13283} -{"mode": "train", "epoch": 1, "iter": 4, "lr": 0.0, "memory": 32769, "data_time": 0.00234, "loss_occ": 3.00854, "loss": 3.00854, "grad_norm": 3.87413, "time": 1.12622} -{"mode": "train", "epoch": 1, "iter": 5, "lr": 0.0, "memory": 32769, "data_time": 0.00279, "loss_occ": 3.0044, "loss": 3.0044, "grad_norm": 3.86394, "time": 1.12671} -{"mode": "train", "epoch": 1, "iter": 6, "lr": 0.0, "memory": 32769, "data_time": 0.00295, "loss_occ": 3.00848, "loss": 3.00848, "grad_norm": 3.84233, "time": 1.12617} -{"mode": "train", "epoch": 1, "iter": 7, "lr": 0.0, "memory": 32769, "data_time": 0.00263, "loss_occ": 3.0085, "loss": 3.0085, "grad_norm": 3.94733, "time": 1.12608} -{"mode": "train", "epoch": 1, "iter": 8, "lr": 0.0, "memory": 32769, "data_time": 0.00604, "loss_occ": 3.00383, "loss": 3.00383, "grad_norm": 3.9215, "time": 1.13024} -{"mode": "train", "epoch": 1, "iter": 9, "lr": 0.0, "memory": 32769, "data_time": 0.0036, "loss_occ": 2.99689, "loss": 2.99689, "grad_norm": 3.86593, "time": 1.1281} -{"mode": "train", "epoch": 1, "iter": 10, "lr": 0.0, "memory": 32769, "data_time": 0.00266, "loss_occ": 2.9957, "loss": 2.9957, "grad_norm": 3.88205, "time": 1.12727} diff --git a/docker-hub/FlashOCC/Flashocc/work_dirs/flashocc-r50/flashocc-r50.py b/docker-hub/FlashOCC/Flashocc/work_dirs/flashocc-r50/flashocc-r50.py deleted file mode 100644 index 3244e491153269fd4262b3ed3087b673905cd026..0000000000000000000000000000000000000000 --- a/docker-hub/FlashOCC/Flashocc/work_dirs/flashocc-r50/flashocc-r50.py +++ /dev/null @@ -1,617 +0,0 @@ -point_cloud_range = [-51.2, -51.2, -5.0, 51.2, 51.2, 3.0] -class_names = [ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier', - 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' -] -dataset_type = 'NuScenesDatasetOccpancy' -data_root = 'data/nuscenes/' -input_modality = dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False) -file_client_args = dict(backend='disk') -train_pipeline = [ - dict( - type='PrepareImageInputs', - is_train=True, - data_config=dict( - cams=[ - 'CAM_FRONT_LEFT', 'CAM_FRONT', 'CAM_FRONT_RIGHT', - 'CAM_BACK_LEFT', 'CAM_BACK', 'CAM_BACK_RIGHT' - ], - Ncams=6, - input_size=(256, 704), - src_size=(900, 1600), - resize=(-0.06, 0.11), - rot=(-5.4, 5.4), - flip=True, - crop_h=(0.0, 0.0), - resize_test=0.0), - sequential=False), - dict( - type='LoadAnnotationsBEVDepth', - bda_aug_conf=dict( - rot_lim=(-0.0, 0.0), - scale_lim=(1.0, 1.0), - flip_dx_ratio=0.5, - flip_dy_ratio=0.5), - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - is_train=True), - dict(type='LoadOccGTFromFile'), - dict( - type='LoadPointsFromFile', - coord_type='LIDAR', - load_dim=5, - use_dim=5, - file_client_args=dict(backend='disk')), - dict( - type='PointToMultiViewDepth', - downsample=1, - grid_config=dict( - x=[-40, 40, 0.4], - y=[-40, 40, 0.4], - z=[-1, 5.4, 6.4], - depth=[1.0, 45.0, 0.5])), - dict( - type='DefaultFormatBundle3D', - class_names=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ]), - dict( - type='Collect3D', - keys=[ - 'img_inputs', 'gt_depth', 'voxel_semantics', 'mask_lidar', - 'mask_camera' - ]) -] -test_pipeline = [ - dict( - type='PrepareImageInputs', - data_config=dict( - cams=[ - 'CAM_FRONT_LEFT', 'CAM_FRONT', 'CAM_FRONT_RIGHT', - 'CAM_BACK_LEFT', 'CAM_BACK', 'CAM_BACK_RIGHT' - ], - Ncams=6, - input_size=(256, 704), - src_size=(900, 1600), - resize=(-0.06, 0.11), - rot=(-5.4, 5.4), - flip=True, - crop_h=(0.0, 0.0), - resize_test=0.0), - sequential=False), - dict( - type='LoadAnnotationsBEVDepth', - bda_aug_conf=dict( - rot_lim=(-0.0, 0.0), - scale_lim=(1.0, 1.0), - flip_dx_ratio=0.5, - flip_dy_ratio=0.5), - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - is_train=False), - dict( - type='LoadPointsFromFile', - coord_type='LIDAR', - load_dim=5, - use_dim=5, - file_client_args=dict(backend='disk')), - dict( - type='MultiScaleFlipAug3D', - img_scale=(1333, 800), - pts_scale_ratio=1, - flip=False, - transforms=[ - dict( - type='DefaultFormatBundle3D', - class_names=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', - 'traffic_cone' - ], - with_label=False), - dict(type='Collect3D', keys=['points', 'img_inputs']) - ]) -] -eval_pipeline = [ - dict( - type='LoadPointsFromFile', - coord_type='LIDAR', - load_dim=5, - use_dim=5, - file_client_args=dict(backend='disk')), - dict( - type='LoadPointsFromMultiSweeps', - sweeps_num=10, - file_client_args=dict(backend='disk')), - dict( - type='DefaultFormatBundle3D', - class_names=[ - 'car', 'truck', 'trailer', 'bus', 'construction_vehicle', - 'bicycle', 'motorcycle', 'pedestrian', 'traffic_cone', 'barrier' - ], - with_label=False), - dict(type='Collect3D', keys=['points']) -] -data = dict( - samples_per_gpu=24, - workers_per_gpu=24, - train=dict( - type='NuScenesDatasetOccpancy', - data_root='data/nuscenes/', - ann_file='data/nuscenes/bevdetv2-nuscenes_infos_train.pkl', - pipeline=[ - dict( - type='PrepareImageInputs', - is_train=True, - data_config=dict( - cams=[ - 'CAM_FRONT_LEFT', 'CAM_FRONT', 'CAM_FRONT_RIGHT', - 'CAM_BACK_LEFT', 'CAM_BACK', 'CAM_BACK_RIGHT' - ], - Ncams=6, - input_size=(256, 704), - src_size=(900, 1600), - resize=(-0.06, 0.11), - rot=(-5.4, 5.4), - flip=True, - crop_h=(0.0, 0.0), - resize_test=0.0), - sequential=False), - dict( - type='LoadAnnotationsBEVDepth', - bda_aug_conf=dict( - rot_lim=(-0.0, 0.0), - scale_lim=(1.0, 1.0), - flip_dx_ratio=0.5, - flip_dy_ratio=0.5), - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', - 'traffic_cone' - ], - is_train=True), - dict(type='LoadOccGTFromFile'), - dict( - type='LoadPointsFromFile', - coord_type='LIDAR', - load_dim=5, - use_dim=5, - file_client_args=dict(backend='disk')), - dict( - type='PointToMultiViewDepth', - downsample=1, - grid_config=dict( - x=[-40, 40, 0.4], - y=[-40, 40, 0.4], - z=[-1, 5.4, 6.4], - depth=[1.0, 45.0, 0.5])), - dict( - type='DefaultFormatBundle3D', - class_names=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', - 'traffic_cone' - ]), - dict( - type='Collect3D', - keys=[ - 'img_inputs', 'gt_depth', 'voxel_semantics', 'mask_lidar', - 'mask_camera' - ]) - ], - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - test_mode=False, - box_type_3d='LiDAR', - use_valid_flag=True, - stereo=False, - filter_empty_gt=False, - img_info_prototype='bevdet'), - val=dict( - type='NuScenesDatasetOccpancy', - data_root='data/nuscenes/', - ann_file='data/nuscenes/bevdetv2-nuscenes_infos_val.pkl', - pipeline=[ - dict( - type='PrepareImageInputs', - data_config=dict( - cams=[ - 'CAM_FRONT_LEFT', 'CAM_FRONT', 'CAM_FRONT_RIGHT', - 'CAM_BACK_LEFT', 'CAM_BACK', 'CAM_BACK_RIGHT' - ], - Ncams=6, - input_size=(256, 704), - src_size=(900, 1600), - resize=(-0.06, 0.11), - rot=(-5.4, 5.4), - flip=True, - crop_h=(0.0, 0.0), - resize_test=0.0), - sequential=False), - dict( - type='LoadAnnotationsBEVDepth', - bda_aug_conf=dict( - rot_lim=(-0.0, 0.0), - scale_lim=(1.0, 1.0), - flip_dx_ratio=0.5, - flip_dy_ratio=0.5), - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', - 'traffic_cone' - ], - is_train=False), - dict( - type='LoadPointsFromFile', - coord_type='LIDAR', - load_dim=5, - use_dim=5, - file_client_args=dict(backend='disk')), - dict( - type='MultiScaleFlipAug3D', - img_scale=(1333, 800), - pts_scale_ratio=1, - flip=False, - transforms=[ - dict( - type='DefaultFormatBundle3D', - class_names=[ - 'car', 'truck', 'construction_vehicle', 'bus', - 'trailer', 'barrier', 'motorcycle', 'bicycle', - 'pedestrian', 'traffic_cone' - ], - with_label=False), - dict(type='Collect3D', keys=['points', 'img_inputs']) - ]) - ], - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - test_mode=True, - box_type_3d='LiDAR', - stereo=False, - filter_empty_gt=False, - img_info_prototype='bevdet'), - test=dict( - type='NuScenesDatasetOccpancy', - data_root='data/nuscenes/', - ann_file='data/nuscenes/bevdetv2-nuscenes_infos_val.pkl', - pipeline=[ - dict( - type='PrepareImageInputs', - data_config=dict( - cams=[ - 'CAM_FRONT_LEFT', 'CAM_FRONT', 'CAM_FRONT_RIGHT', - 'CAM_BACK_LEFT', 'CAM_BACK', 'CAM_BACK_RIGHT' - ], - Ncams=6, - input_size=(256, 704), - src_size=(900, 1600), - resize=(-0.06, 0.11), - rot=(-5.4, 5.4), - flip=True, - crop_h=(0.0, 0.0), - resize_test=0.0), - sequential=False), - dict( - type='LoadAnnotationsBEVDepth', - bda_aug_conf=dict( - rot_lim=(-0.0, 0.0), - scale_lim=(1.0, 1.0), - flip_dx_ratio=0.5, - flip_dy_ratio=0.5), - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', - 'traffic_cone' - ], - is_train=False), - dict( - type='LoadPointsFromFile', - coord_type='LIDAR', - load_dim=5, - use_dim=5, - file_client_args=dict(backend='disk')), - dict( - type='MultiScaleFlipAug3D', - img_scale=(1333, 800), - pts_scale_ratio=1, - flip=False, - transforms=[ - dict( - type='DefaultFormatBundle3D', - class_names=[ - 'car', 'truck', 'construction_vehicle', 'bus', - 'trailer', 'barrier', 'motorcycle', 'bicycle', - 'pedestrian', 'traffic_cone' - ], - with_label=False), - dict(type='Collect3D', keys=['points', 'img_inputs']) - ]) - ], - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - test_mode=True, - box_type_3d='LiDAR', - stereo=False, - filter_empty_gt=False, - img_info_prototype='bevdet')) -evaluation = dict( - interval=1, - pipeline=[ - dict( - type='PrepareImageInputs', - data_config=dict( - cams=[ - 'CAM_FRONT_LEFT', 'CAM_FRONT', 'CAM_FRONT_RIGHT', - 'CAM_BACK_LEFT', 'CAM_BACK', 'CAM_BACK_RIGHT' - ], - Ncams=6, - input_size=(256, 704), - src_size=(900, 1600), - resize=(-0.06, 0.11), - rot=(-5.4, 5.4), - flip=True, - crop_h=(0.0, 0.0), - resize_test=0.0), - sequential=False), - dict( - type='LoadAnnotationsBEVDepth', - bda_aug_conf=dict( - rot_lim=(-0.0, 0.0), - scale_lim=(1.0, 1.0), - flip_dx_ratio=0.5, - flip_dy_ratio=0.5), - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', - 'traffic_cone' - ], - is_train=False), - dict( - type='LoadPointsFromFile', - coord_type='LIDAR', - load_dim=5, - use_dim=5, - file_client_args=dict(backend='disk')), - dict( - type='MultiScaleFlipAug3D', - img_scale=(1333, 800), - pts_scale_ratio=1, - flip=False, - transforms=[ - dict( - type='DefaultFormatBundle3D', - class_names=[ - 'car', 'truck', 'construction_vehicle', 'bus', - 'trailer', 'barrier', 'motorcycle', 'bicycle', - 'pedestrian', 'traffic_cone' - ], - with_label=False), - dict(type='Collect3D', keys=['points', 'img_inputs']) - ]) - ], - start=20) -checkpoint_config = dict(interval=1, max_keep_ckpts=5) -log_config = dict( - interval=1, - hooks=[dict(type='TextLoggerHook'), - dict(type='TensorboardLoggerHook')]) -dist_params = dict(backend='nccl') -log_level = 'INFO' -work_dir = './work_dirs/flashocc-r50' -load_from = 'ckpts/bevdet-r50-cbgs.pth' -resume_from = None -workflow = [('train', 1)] -opencv_num_threads = 0 -mp_start_method = 'fork' -plugin = True -plugin_dir = 'projects/mmdet3d_plugin/' -data_config = dict( - cams=[ - 'CAM_FRONT_LEFT', 'CAM_FRONT', 'CAM_FRONT_RIGHT', 'CAM_BACK_LEFT', - 'CAM_BACK', 'CAM_BACK_RIGHT' - ], - Ncams=6, - input_size=(256, 704), - src_size=(900, 1600), - resize=(-0.06, 0.11), - rot=(-5.4, 5.4), - flip=True, - crop_h=(0.0, 0.0), - resize_test=0.0) -grid_config = dict( - x=[-40, 40, 0.4], - y=[-40, 40, 0.4], - z=[-1, 5.4, 6.4], - depth=[1.0, 45.0, 0.5]) -voxel_size = [0.1, 0.1, 0.2] -numC_Trans = 64 -model = dict( - type='BEVDetOCC', - img_backbone=dict( - type='ResNet', - depth=50, - num_stages=4, - out_indices=(2, 3), - frozen_stages=-1, - norm_cfg=dict(type='BN', requires_grad=True), - norm_eval=False, - with_cp=True, - style='pytorch'), - img_neck=dict( - type='CustomFPN', - in_channels=[1024, 2048], - out_channels=256, - num_outs=1, - start_level=0, - out_ids=[0]), - img_view_transformer=dict( - type='LSSViewTransformer', - grid_config=dict( - x=[-40, 40, 0.4], - y=[-40, 40, 0.4], - z=[-1, 5.4, 6.4], - depth=[1.0, 45.0, 0.5]), - input_size=(256, 704), - in_channels=256, - out_channels=64, - sid=False, - collapse_z=True, - downsample=16), - img_bev_encoder_backbone=dict( - type='CustomResNet', numC_input=64, num_channels=[128, 256, 512]), - img_bev_encoder_neck=dict( - type='FPN_LSS', in_channels=640, out_channels=256), - occ_head=dict( - type='BEVOCCHead2D', - in_dim=256, - out_dim=256, - Dz=16, - use_mask=True, - num_classes=18, - use_predicter=True, - class_balance=False, - loss_occ=dict( - type='CrossEntropyLoss', - use_sigmoid=False, - ignore_index=255, - loss_weight=1.0))) -bda_aug_conf = dict( - rot_lim=(-0.0, 0.0), - scale_lim=(1.0, 1.0), - flip_dx_ratio=0.5, - flip_dy_ratio=0.5) -share_data_config = dict( - type='NuScenesDatasetOccpancy', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier', - 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - stereo=False, - filter_empty_gt=False, - img_info_prototype='bevdet') -test_data_config = dict( - pipeline=[ - dict( - type='PrepareImageInputs', - data_config=dict( - cams=[ - 'CAM_FRONT_LEFT', 'CAM_FRONT', 'CAM_FRONT_RIGHT', - 'CAM_BACK_LEFT', 'CAM_BACK', 'CAM_BACK_RIGHT' - ], - Ncams=6, - input_size=(256, 704), - src_size=(900, 1600), - resize=(-0.06, 0.11), - rot=(-5.4, 5.4), - flip=True, - crop_h=(0.0, 0.0), - resize_test=0.0), - sequential=False), - dict( - type='LoadAnnotationsBEVDepth', - bda_aug_conf=dict( - rot_lim=(-0.0, 0.0), - scale_lim=(1.0, 1.0), - flip_dx_ratio=0.5, - flip_dy_ratio=0.5), - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', - 'traffic_cone' - ], - is_train=False), - dict( - type='LoadPointsFromFile', - coord_type='LIDAR', - load_dim=5, - use_dim=5, - file_client_args=dict(backend='disk')), - dict( - type='MultiScaleFlipAug3D', - img_scale=(1333, 800), - pts_scale_ratio=1, - flip=False, - transforms=[ - dict( - type='DefaultFormatBundle3D', - class_names=[ - 'car', 'truck', 'construction_vehicle', 'bus', - 'trailer', 'barrier', 'motorcycle', 'bicycle', - 'pedestrian', 'traffic_cone' - ], - with_label=False), - dict(type='Collect3D', keys=['points', 'img_inputs']) - ]) - ], - ann_file='data/nuscenes/bevdetv2-nuscenes_infos_val.pkl', - type='NuScenesDatasetOccpancy', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier', - 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - stereo=False, - filter_empty_gt=False, - img_info_prototype='bevdet') -key = 'test' -optimizer = dict(type='AdamW', lr=0.0001, weight_decay=0.01) -optimizer_config = dict(grad_clip=dict(max_norm=5, norm_type=2)) -lr_config = dict( - policy='step', - warmup='linear', - warmup_iters=200, - warmup_ratio=0.001, - step=[24]) -runner = dict(type='EpochBasedRunner', max_epochs=24) -custom_hooks = [ - dict(type='MEGVIIEMAHook', init_updates=10560, priority='NORMAL') -] -gpu_ids = range(0, 8) diff --git a/docker-hub/FlashOCC/Flashocc/work_dirs/flashocc-r50/tf_logs/events.out.tfevents.1775204673.bw61.849.0 b/docker-hub/FlashOCC/Flashocc/work_dirs/flashocc-r50/tf_logs/events.out.tfevents.1775204673.bw61.849.0 deleted file mode 100644 index 1bb971ede8cfafc60dec584ec4fb6fc996b0353d..0000000000000000000000000000000000000000 Binary files a/docker-hub/FlashOCC/Flashocc/work_dirs/flashocc-r50/tf_logs/events.out.tfevents.1775204673.bw61.849.0 and /dev/null differ diff --git a/docker-hub/FlashOCC/Flashocc/work_dirs/flashocc-r50/tf_logs/events.out.tfevents.1775204818.bw61.20636.0 b/docker-hub/FlashOCC/Flashocc/work_dirs/flashocc-r50/tf_logs/events.out.tfevents.1775204818.bw61.20636.0 deleted file mode 100644 index 6aa5d3638738e72579d46fe382f14b8ee7b56770..0000000000000000000000000000000000000000 Binary files a/docker-hub/FlashOCC/Flashocc/work_dirs/flashocc-r50/tf_logs/events.out.tfevents.1775204818.bw61.20636.0 and /dev/null differ diff --git a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20251112_142824.log b/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20251112_142824.log deleted file mode 100644 index af8341358d918bf6bdb5b43fb7bc39fd9adb4d71..0000000000000000000000000000000000000000 --- a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20251112_142824.log +++ /dev/null @@ -1,3647 +0,0 @@ -2025-11-12 14:28:24,757 - mmdet - INFO - Environment info: ------------------------------------------------------------- -sys.platform: linux -Python: 3.10.12 (main, May 27 2025, 17:12:29) [GCC 11.4.0] -CUDA available: True -GPU 0,1,2,3,4,5,6,7: BW200, UBB BW1000 -CUDA_HOME: /opt/dtk -NVCC: Not Available -GCC: x86_64-linux-gnu-gcc (Ubuntu 11.4.0-1ubuntu1~22.04) 11.4.0 -PyTorch: 2.4.1 -PyTorch compiling details: PyTorch built with: - - GCC 10.3 - - C++ Version: 201703 - - Intel(R) Math Kernel Library Version 2020.0.4 Product Build 20200917 for Intel(R) 64 architecture applications - - OpenMP 201511 (a.k.a. OpenMP 4.5) - - LAPACK is enabled (usually provided by MKL) - - NNPACK is enabled - - CPU capability usage: AVX512 - - HIP Runtime 6.3.25211 - - MIOpen 2.17.0 - - Magma 2.8.0 - - Build settings: BLAS_INFO=mkl, BUILD_TYPE=Release, CXX_COMPILER=/opt/rh/gcc-toolset-10/root/usr/bin/c++, CXX_FLAGS= -D_GLIBCXX_USE_CXX11_ABI=1 -fvisibility-inlines-hidden -DUSE_PTHREADPOOL -DNDEBUG -DUSE_KINETO -DLIBKINETO_NOCUPTI -DUSE_FBGEMM -DUSE_PYTORCH_QNNPACK -DUSE_XNNPACK -DSYMBOLICATE_MOBILE_DEBUG_HANDLE -O2 -fPIC -Wall -Wextra -Werror=return-type -Werror=non-virtual-dtor -Werror=bool-operation -Wnarrowing -Wno-missing-field-initializers -Wno-type-limits -Wno-array-bounds -Wno-unknown-pragmas -Wno-unused-parameter -Wno-unused-function -Wno-unused-result -Wno-strict-overflow -Wno-strict-aliasing -Wno-stringop-overflow -Wsuggest-override -Wno-psabi -Wno-error=pedantic -Wno-error=old-style-cast -Wno-missing-braces -fdiagnostics-color=always -faligned-new -Wno-unused-but-set-variable -Wno-maybe-uninitialized -fno-math-errno -fno-trapping-math -Werror=format -Wno-stringop-overflow, FORCE_FALLBACK_CUDA_MPI=1, LAPACK_INFO=mkl, PERF_WITH_AVX=1, PERF_WITH_AVX2=1, PERF_WITH_AVX512=1, TORCH_VERSION=2.4.1, USE_CUDA=0, USE_CUDNN=OFF, USE_CUSPARSELT=OFF, USE_EXCEPTION_PTR=1, USE_GFLAGS=1, USE_GLOG=1, USE_GLOO=1, USE_MKL=ON, USE_MKLDNN=0, USE_MPI=1, USE_NCCL=1, USE_NNPACK=ON, USE_OPENMP=1, USE_ROCM=ON, USE_ROCM_KERNEL_ASSERT=OFF, - -TorchVision: 0.19.1 -OpenCV: 4.11.0 -MMCV: 1.6.1 -MMCV Compiler: GCC 11.4 -MMCV CUDA Compiler: rocm not available -MMDetection: 2.26.0+c41df4b ------------------------------------------------------------- - -2025-11-12 14:28:25,494 - mmdet - INFO - Distributed training: True -2025-11-12 14:28:26,205 - mmdet - INFO - Config: -plugin = True -plugin_dir = 'projects/mmdet3d_plugin/' -dist_params = dict(backend='nccl') -log_level = 'INFO' -work_dir = './work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704' -total_batch_size = 160 -num_gpus = 8 -batch_size = 20 -num_iters_per_epoch = 175 -num_epochs = 100 -checkpoint_epoch_interval = 20 -checkpoint_config = dict(interval=3500) -log_config = dict( - interval=1, - hooks=[ - dict(type='TextLoggerHook', by_epoch=False), - dict(type='TensorboardLoggerHook') - ]) -load_from = None -resume_from = None -workflow = [('train', 1)] -fp16 = dict(loss_scale=32.0) -input_shape = (704, 256) -tracking_test = True -tracking_threshold = 0.2 -class_names = [ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier', - 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' -] -num_classes = 10 -embed_dims = 256 -num_groups = 8 -num_decoder = 6 -num_single_frame_decoder = 1 -use_deformable_func = True -strides = [4, 8, 16, 32] -num_levels = 4 -num_depth_layers = 3 -drop_out = 0.1 -temporal = True -decouple_attn = True -with_quality_estimation = True -model = dict( - type='Sparse4D', - use_grid_mask=True, - use_deformable_func=True, - img_backbone=dict( - type='ResNet', - depth=50, - num_stages=4, - frozen_stages=-1, - norm_eval=False, - style='pytorch', - with_cp=True, - out_indices=(0, 1, 2, 3), - norm_cfg=dict(type='BN', requires_grad=True), - pretrained='ckpt/resnet50-19c8e357.pth'), - img_neck=dict( - type='FPN', - num_outs=4, - start_level=0, - out_channels=256, - add_extra_convs='on_output', - relu_before_extra_convs=True, - in_channels=[256, 512, 1024, 2048]), - depth_branch=dict( - type='DenseDepthNet', - embed_dims=256, - num_depth_layers=3, - loss_weight=0.2), - head=dict( - type='Sparse4DHead', - cls_threshold_to_reg=0.05, - decouple_attn=True, - instance_bank=dict( - type='InstanceBank', - num_anchor=900, - embed_dims=256, - anchor='nuscenes_kmeans900.npy', - anchor_handler=dict(type='SparseBox3DKeyPointsGenerator'), - num_temp_instances=600, - confidence_decay=0.6, - feat_grad=False), - anchor_encoder=dict( - type='SparseBox3DEncoder', - vel_dims=3, - embed_dims=[128, 32, 32, 64], - mode='cat', - output_fc=False, - in_loops=1, - out_loops=4), - num_single_frame_decoder=1, - operation_order=[ - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine' - ], - temp_graph_model=dict( - type='MultiheadAttention', - embed_dims=512, - num_heads=8, - batch_first=True, - dropout=0.1), - graph_model=dict( - type='MultiheadAttention', - embed_dims=512, - num_heads=8, - batch_first=True, - dropout=0.1), - norm_layer=dict(type='LN', normalized_shape=256), - ffn=dict( - type='AsymmetricFFN', - in_channels=512, - pre_norm=dict(type='LN'), - embed_dims=256, - feedforward_channels=1024, - num_fcs=2, - ffn_drop=0.1, - act_cfg=dict(type='ReLU', inplace=True)), - deformable_model=dict( - type='DeformableFeatureAggregation', - embed_dims=256, - num_groups=8, - num_levels=4, - num_cams=6, - attn_drop=0.15, - use_deformable_func=True, - use_camera_embed=True, - residual_mode='cat', - kps_generator=dict( - type='SparseBox3DKeyPointsGenerator', - num_learnable_pts=6, - fix_scale=[[0, 0, 0], [0.45, 0, 0], [-0.45, 0, 0], - [0, 0.45, 0], [0, -0.45, 0], [0, 0, 0.45], - [0, 0, -0.45]])), - refine_layer=dict( - type='SparseBox3DRefinementModule', - embed_dims=256, - num_cls=10, - refine_yaw=True, - with_quality_estimation=True), - sampler=dict( - type='SparseBox3DTarget', - num_dn_groups=5, - num_temp_dn_groups=3, - dn_noise_scale=[2.0, 2.0, 2.0, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5], - max_dn_gt=32, - add_neg_dn=True, - cls_weight=2.0, - box_weight=0.25, - reg_weights=[2.0, 2.0, 2.0, 0.5, 0.5, 0.5, 0.0, 0.0, 0.0, 0.0], - cls_wise_reg_weights=dict( - {9: [2.0, 2.0, 2.0, 1.0, 1.0, 1.0, 0.0, 0.0, 1.0, 1.0]})), - loss_cls=dict( - type='FocalLoss', - use_sigmoid=True, - gamma=2.0, - alpha=0.25, - loss_weight=2.0), - loss_reg=dict( - type='SparseBox3DLoss', - loss_box=dict(type='L1Loss', loss_weight=0.25), - loss_centerness=dict(type='CrossEntropyLoss', use_sigmoid=True), - loss_yawness=dict(type='GaussianFocalLoss'), - cls_allow_reverse=[5]), - decoder=dict(type='SparseBox3DDecoder'), - reg_weights=[2.0, 2.0, 2.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0])) -dataset_type = 'NuScenes3DDetTrackDataset' -data_root = 'data/nuscenes/' -anno_root = 'data/nuscenes_anno_pkls/' -file_client_args = dict(backend='disk') -img_norm_cfg = dict( - mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375], to_rgb=True) -train_pipeline = [ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict( - type='LoadPointsFromFile', - coord_type='LIDAR', - load_dim=5, - use_dim=5, - file_client_args=dict(backend='disk')), - dict(type='ResizeCropFlipImage'), - dict(type='MultiScaleDepthMapGenerator', downsample=[4, 8, 16]), - dict(type='BBoxRotation'), - dict(type='PhotoMetricDistortionMultiViewImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict( - type='CircleObjectRangeFilter', - class_dist_thred=[55, 55, 55, 55, 55, 55, 55, 55, 55, 55]), - dict( - type='InstanceNameFilter', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ]), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=[ - 'img', 'timestamp', 'projection_mat', 'image_wh', 'gt_depth', - 'focal', 'gt_bboxes_3d', 'gt_labels_3d' - ], - meta_keys=['T_global', 'T_global_inv', 'timestamp', 'instance_id']) -] -test_pipeline = [ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='ResizeCropFlipImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=['img', 'timestamp', 'projection_mat', 'image_wh'], - meta_keys=['T_global', 'T_global_inv', 'timestamp']) -] -input_modality = dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False) -data_basic_config = dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier', - 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval') -data_aug_conf = dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]) -data = dict( - samples_per_gpu=20, - workers_per_gpu=20, - train=dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval', - ann_file='data/nuscenes_anno_pkls/nuscenes_infos_train.pkl', - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict( - type='LoadPointsFromFile', - coord_type='LIDAR', - load_dim=5, - use_dim=5, - file_client_args=dict(backend='disk')), - dict(type='ResizeCropFlipImage'), - dict(type='MultiScaleDepthMapGenerator', downsample=[4, 8, 16]), - dict(type='BBoxRotation'), - dict(type='PhotoMetricDistortionMultiViewImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict( - type='CircleObjectRangeFilter', - class_dist_thred=[55, 55, 55, 55, 55, 55, 55, 55, 55, 55]), - dict( - type='InstanceNameFilter', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', - 'traffic_cone' - ]), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=[ - 'img', 'timestamp', 'projection_mat', 'image_wh', - 'gt_depth', 'focal', 'gt_bboxes_3d', 'gt_labels_3d' - ], - meta_keys=[ - 'T_global', 'T_global_inv', 'timestamp', 'instance_id' - ]) - ], - test_mode=False, - data_aug_conf=dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]), - with_seq_flag=True, - sequences_split_num=2, - keep_consistent_seq_aug=True), - val=dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval', - ann_file='data/nuscenes_anno_pkls/nuscenes_infos_val.pkl', - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='ResizeCropFlipImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=['img', 'timestamp', 'projection_mat', 'image_wh'], - meta_keys=['T_global', 'T_global_inv', 'timestamp']) - ], - data_aug_conf=dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]), - test_mode=True, - tracking=True, - tracking_threshold=0.2), - test=dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval', - ann_file='data/nuscenes_anno_pkls/nuscenes_infos_val.pkl', - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='ResizeCropFlipImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=['img', 'timestamp', 'projection_mat', 'image_wh'], - meta_keys=['T_global', 'T_global_inv', 'timestamp']) - ], - data_aug_conf=dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]), - test_mode=True, - tracking=True, - tracking_threshold=0.2)) -optimizer = dict( - type='AdamW', - lr=0.0006, - weight_decay=0.001, - paramwise_cfg=dict(custom_keys=dict(img_backbone=dict(lr_mult=0.5)))) -optimizer_config = dict(grad_clip=dict(max_norm=25, norm_type=2)) -lr_config = dict( - policy='CosineAnnealing', - warmup='linear', - warmup_iters=500, - warmup_ratio=0.3333333333333333, - min_lr_ratio=0.001) -runner = dict(type='IterBasedRunner', max_iters=17500) -vis_pipeline = [ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='Collect', keys=['img'], meta_keys=['timestamp', 'lidar2img']) -] -evaluation = dict( - interval=3500, - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict( - type='Collect', keys=['img'], meta_keys=['timestamp', 'lidar2img']) - ]) -gpu_ids = range(0, 8) - -2025-11-12 14:28:26,205 - mmdet - INFO - Set random seed to 0, deterministic: False -2025-11-12 14:28:26,505 - mmdet - INFO - initialize ResNet with init_cfg {'type': 'Pretrained', 'checkpoint': 'ckpt/resnet50-19c8e357.pth'} -2025-11-12 14:28:26,875 - mmdet - INFO - initialize FPN with init_cfg {'type': 'Xavier', 'layer': 'Conv2d', 'distribution': 'uniform'} -Name of parameter - Initialization information - -img_backbone.conv1.weight - torch.Size([64, 3, 7, 7]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.bn1.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.bn1.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.conv1.weight - torch.Size([64, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn1.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn1.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.conv2.weight - torch.Size([64, 64, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn2.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn2.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.conv3.weight - torch.Size([256, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn3.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn3.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.downsample.0.weight - torch.Size([256, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.downsample.1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.downsample.1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.conv1.weight - torch.Size([64, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn1.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn1.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.conv2.weight - torch.Size([64, 64, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn2.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn2.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.conv3.weight - torch.Size([256, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn3.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn3.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.conv1.weight - torch.Size([64, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn1.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn1.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.conv2.weight - torch.Size([64, 64, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn2.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn2.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.conv3.weight - torch.Size([256, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn3.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn3.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.conv1.weight - torch.Size([128, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn1.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn1.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.conv2.weight - torch.Size([128, 128, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn2.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn2.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.conv3.weight - torch.Size([512, 128, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn3.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn3.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.downsample.0.weight - torch.Size([512, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.downsample.1.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.downsample.1.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.conv1.weight - torch.Size([128, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn1.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn1.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.conv2.weight - torch.Size([128, 128, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn2.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn2.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.conv3.weight - torch.Size([512, 128, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn3.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn3.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.conv1.weight - torch.Size([128, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn1.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn1.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.conv2.weight - torch.Size([128, 128, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn2.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn2.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.conv3.weight - torch.Size([512, 128, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn3.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn3.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.conv1.weight - torch.Size([128, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn1.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn1.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.conv2.weight - torch.Size([128, 128, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn2.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn2.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.conv3.weight - torch.Size([512, 128, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn3.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn3.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.conv1.weight - torch.Size([256, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.downsample.0.weight - torch.Size([1024, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.downsample.1.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.downsample.1.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.conv1.weight - torch.Size([512, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn1.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn1.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.conv2.weight - torch.Size([512, 512, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn2.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn2.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.conv3.weight - torch.Size([2048, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn3.weight - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn3.bias - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.downsample.0.weight - torch.Size([2048, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.downsample.1.weight - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.downsample.1.bias - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.conv1.weight - torch.Size([512, 2048, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn1.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn1.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.conv2.weight - torch.Size([512, 512, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn2.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn2.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.conv3.weight - torch.Size([2048, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn3.weight - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn3.bias - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.conv1.weight - torch.Size([512, 2048, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn1.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn1.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.conv2.weight - torch.Size([512, 512, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn2.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn2.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.conv3.weight - torch.Size([2048, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn3.weight - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn3.bias - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_neck.lateral_convs.0.conv.weight - torch.Size([256, 256, 1, 1]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.lateral_convs.0.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.lateral_convs.1.conv.weight - torch.Size([256, 512, 1, 1]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.lateral_convs.1.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.lateral_convs.2.conv.weight - torch.Size([256, 1024, 1, 1]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.lateral_convs.2.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.lateral_convs.3.conv.weight - torch.Size([256, 2048, 1, 1]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.lateral_convs.3.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.fpn_convs.0.conv.weight - torch.Size([256, 256, 3, 3]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.fpn_convs.0.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.fpn_convs.1.conv.weight - torch.Size([256, 256, 3, 3]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.fpn_convs.1.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.fpn_convs.2.conv.weight - torch.Size([256, 256, 3, 3]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.fpn_convs.2.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.fpn_convs.3.conv.weight - torch.Size([256, 256, 3, 3]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.fpn_convs.3.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.instance_bank.anchor - torch.Size([900, 11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.instance_bank.instance_feature - torch.Size([900, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.instance_bank.anchor_handler.fix_scale - torch.Size([1, 3]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.0.weight - torch.Size([128, 3]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.0.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.2.weight - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.2.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.3.weight - torch.Size([128, 128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.3.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.5.weight - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.5.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.6.weight - torch.Size([128, 128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.6.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.8.weight - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.8.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.9.weight - torch.Size([128, 128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.9.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.11.weight - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.11.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.0.weight - torch.Size([32, 3]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.0.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.2.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.2.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.3.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.3.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.5.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.5.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.6.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.6.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.8.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.8.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.9.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.9.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.11.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.11.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.0.weight - torch.Size([32, 2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.0.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.2.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.2.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.3.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.3.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.5.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.5.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.6.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.6.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.8.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.8.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.9.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.9.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.11.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.11.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.0.weight - torch.Size([64, 3]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.0.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.2.weight - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.2.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.3.weight - torch.Size([64, 64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.3.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.5.weight - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.5.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.6.weight - torch.Size([64, 64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.6.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.8.weight - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.8.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.9.weight - torch.Size([64, 64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.9.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.11.weight - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.11.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.1.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.1.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.1.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.1.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.1.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.1.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.1.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.1.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.3.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.4.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.4.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.4.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.4.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.5.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.5.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.5.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.5.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.6.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.6.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.8.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.8.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.8.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.8.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.8.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.8.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.8.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.8.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.10.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.11.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.11.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.11.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.11.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.12.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.12.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.12.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.12.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.13.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.13.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.15.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.15.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.15.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.15.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.15.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.15.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.15.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.15.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.16.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.16.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.17.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.18.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.18.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.18.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.18.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.19.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.19.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.19.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.19.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.20.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.20.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.22.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.22.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.22.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.22.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.22.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.22.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.22.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.22.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.23.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.23.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.24.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.25.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.25.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.25.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.25.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.26.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.26.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.26.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.26.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.27.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.27.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.29.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.29.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.29.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.29.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.29.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.29.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.29.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.29.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.30.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.30.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.31.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.32.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.32.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.32.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.32.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.33.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.33.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.33.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.33.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.34.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.34.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.36.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.36.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.36.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.36.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.36.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.36.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.36.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.36.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.37.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.37.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.38.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.fc_before.weight - torch.Size([512, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.fc_after.weight - torch.Size([256, 512]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.0.weight - torch.Size([1, 256, 1, 1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.0.bias - torch.Size([1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.1.weight - torch.Size([1, 256, 1, 1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.1.bias - torch.Size([1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.2.weight - torch.Size([1, 256, 1, 1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.2.bias - torch.Size([1]): -The value is the same before and after calling `init_weights` of Sparse4D -2025-11-12 14:28:26,965 - mmdet - INFO - Model: -Sparse4D( - (img_backbone): ResNet( - (conv1): Conv2d(3, 64, kernel_size=(7, 7), stride=(2, 2), padding=(3, 3), bias=False) - (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (maxpool): MaxPool2d(kernel_size=3, stride=2, padding=1, dilation=1, ceil_mode=False) - (layer1): ResLayer( - (0): Bottleneck( - (conv1): Conv2d(64, 64, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (downsample): Sequential( - (0): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - ) - ) - (1): Bottleneck( - (conv1): Conv2d(256, 64, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (2): Bottleneck( - (conv1): Conv2d(256, 64, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - ) - (layer2): ResLayer( - (0): Bottleneck( - (conv1): Conv2d(256, 128, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (downsample): Sequential( - (0): Conv2d(256, 512, kernel_size=(1, 1), stride=(2, 2), bias=False) - (1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - ) - ) - (1): Bottleneck( - (conv1): Conv2d(512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (2): Bottleneck( - (conv1): Conv2d(512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (3): Bottleneck( - (conv1): Conv2d(512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - ) - (layer3): ResLayer( - (0): Bottleneck( - (conv1): Conv2d(512, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (downsample): Sequential( - (0): Conv2d(512, 1024, kernel_size=(1, 1), stride=(2, 2), bias=False) - (1): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - ) - ) - (1): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (2): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (3): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (4): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (5): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - ) - (layer4): ResLayer( - (0): Bottleneck( - (conv1): Conv2d(1024, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(512, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (downsample): Sequential( - (0): Conv2d(1024, 2048, kernel_size=(1, 1), stride=(2, 2), bias=False) - (1): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - ) - ) - (1): Bottleneck( - (conv1): Conv2d(2048, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(512, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (2): Bottleneck( - (conv1): Conv2d(2048, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(512, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - ) - ) - init_cfg={'type': 'Pretrained', 'checkpoint': 'ckpt/resnet50-19c8e357.pth'} - (img_neck): FPN( - (lateral_convs): ModuleList( - (0): ConvModule( - (conv): Conv2d(256, 256, kernel_size=(1, 1), stride=(1, 1)) - ) - (1): ConvModule( - (conv): Conv2d(512, 256, kernel_size=(1, 1), stride=(1, 1)) - ) - (2): ConvModule( - (conv): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1)) - ) - (3): ConvModule( - (conv): Conv2d(2048, 256, kernel_size=(1, 1), stride=(1, 1)) - ) - ) - (fpn_convs): ModuleList( - (0-3): 4 x ConvModule( - (conv): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1)) - ) - ) - ) - init_cfg={'type': 'Xavier', 'layer': 'Conv2d', 'distribution': 'uniform'} - (head): Sparse4DHead( - (instance_bank): InstanceBank( - (anchor_handler): SparseBox3DKeyPointsGenerator() - ) - (anchor_encoder): SparseBox3DEncoder( - (pos_fc): Sequential( - (0): Linear(in_features=3, out_features=128, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((128,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=128, out_features=128, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((128,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=128, out_features=128, bias=True) - (7): ReLU(inplace=True) - (8): LayerNorm((128,), eps=1e-05, elementwise_affine=True) - (9): Linear(in_features=128, out_features=128, bias=True) - (10): ReLU(inplace=True) - (11): LayerNorm((128,), eps=1e-05, elementwise_affine=True) - ) - (size_fc): Sequential( - (0): Linear(in_features=3, out_features=32, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=32, out_features=32, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=32, out_features=32, bias=True) - (7): ReLU(inplace=True) - (8): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (9): Linear(in_features=32, out_features=32, bias=True) - (10): ReLU(inplace=True) - (11): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - ) - (yaw_fc): Sequential( - (0): Linear(in_features=2, out_features=32, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=32, out_features=32, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=32, out_features=32, bias=True) - (7): ReLU(inplace=True) - (8): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (9): Linear(in_features=32, out_features=32, bias=True) - (10): ReLU(inplace=True) - (11): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - ) - (vel_fc): Sequential( - (0): Linear(in_features=3, out_features=64, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((64,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=64, out_features=64, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((64,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=64, out_features=64, bias=True) - (7): ReLU(inplace=True) - (8): LayerNorm((64,), eps=1e-05, elementwise_affine=True) - (9): Linear(in_features=64, out_features=64, bias=True) - (10): ReLU(inplace=True) - (11): LayerNorm((64,), eps=1e-05, elementwise_affine=True) - ) - ) - (loss_cls): FocalLoss() - (loss_reg): SparseBox3DLoss( - (loss_box): L1Loss() - (loss_cns): CrossEntropyLoss(avg_non_ignore=False) - (loss_yns): GaussianFocalLoss() - ) - (layers): ModuleList( - (0): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (1): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (4-5): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (6): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (7): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (8): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (11-12): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (13): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (14): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (15): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (16): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (17): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (18-19): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (20): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (21): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (22): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (23): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (24): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (25-26): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (27): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (28): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (29): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (30): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (31): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (32-33): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (34): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (35): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (36): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (37): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (38): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - ) - (fc_before): Linear(in_features=256, out_features=512, bias=False) - (fc_after): Linear(in_features=512, out_features=256, bias=False) - ) - (depth_branch): DenseDepthNet( - (depth_layers): ModuleList( - (0-2): 3 x Conv2d(256, 1, kernel_size=(1, 1), stride=(1, 1)) - ) - ) - (grid_mask): GridMask() -) -2025-11-12 14:28:39,301 - mmdet - INFO - Start running, host: root@VM-120-96-tencentos, work_dir: /cfs/auto/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704 -2025-11-12 14:28:39,301 - mmdet - INFO - Hooks will be executed in the following order: -before_run: -(VERY_HIGH ) CosineAnnealingLrUpdaterHook -(ABOVE_NORMAL) Fp16OptimizerHook -(NORMAL ) CheckpointHook -(NORMAL ) CustomDistEvalHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -before_train_epoch: -(VERY_HIGH ) CosineAnnealingLrUpdaterHook -(NORMAL ) CustomDistEvalHook -(LOW ) IterTimerHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -before_train_iter: -(VERY_HIGH ) CosineAnnealingLrUpdaterHook -(NORMAL ) CustomDistEvalHook -(LOW ) IterTimerHook - -------------------- -after_train_iter: -(ABOVE_NORMAL) Fp16OptimizerHook -(NORMAL ) CheckpointHook -(NORMAL ) CustomDistEvalHook -(LOW ) IterTimerHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -after_train_epoch: -(NORMAL ) CheckpointHook -(NORMAL ) CustomDistEvalHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -before_val_epoch: -(LOW ) IterTimerHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -before_val_iter: -(LOW ) IterTimerHook - -------------------- -after_val_iter: -(LOW ) IterTimerHook - -------------------- -after_val_epoch: -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -after_run: -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -2025-11-12 14:28:39,302 - mmdet - INFO - workflow: [('train', 1)], max: 17500 iters -2025-11-12 14:28:39,304 - mmdet - INFO - Checkpoints will be saved to /cfs/auto/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704 by HardDiskBackend. -2025-11-12 14:30:37,253 - mmdet - INFO - Iter [1/17500] lr: 1.000e-04, eta: 23 days, 15:59:11, time: 116.850, data_time: 9.036, memory: 49164, loss_cls_0: 2.3612, loss_box_0: 0.0138, loss_cns_0: 0.0027, loss_yns_0: 0.0008, loss_cls_1: 2.1545, loss_box_1: 0.1081, loss_cns_1: 0.0245, loss_yns_1: 0.0067, loss_cls_2: 2.3121, loss_box_2: 0.0050, loss_cns_2: 0.0006, loss_yns_2: 0.0003, loss_cls_3: 2.3902, loss_box_3: 0.0295, loss_cns_3: 0.0050, loss_yns_3: 0.0014, loss_cls_4: 2.0281, loss_box_4: 0.4179, loss_cns_4: 0.0535, loss_yns_4: 0.0253, loss_cls_5: 2.4247, loss_box_5: 0.0180, loss_cns_5: 0.0022, loss_yns_5: 0.0016, loss_cls_dn_0: 1.1980, loss_box_dn_0: 1.4603, loss_cls_dn_1: 1.1102, loss_box_dn_1: 1.7318, loss_cls_dn_2: 1.1741, loss_box_dn_2: 1.9718, loss_cls_dn_3: 1.1721, loss_box_dn_3: 2.2418, loss_cls_dn_4: 1.0528, loss_box_dn_4: 2.4268, loss_cls_dn_5: 1.2387, loss_box_dn_5: 2.6773, loss_dense_depth: 1.8643, loss: 35.7080, grad_norm: 270.4737 -2025-11-12 14:30:39,270 - mmdet - INFO - Iter [2/17500] lr: 1.004e-04, eta: 12 days, 0:53:01, time: 2.019, data_time: 0.106, memory: 49164, loss_cls_0: 2.0434, loss_box_0: 0.0240, loss_cns_0: 0.0064, loss_yns_0: 0.0023, loss_cls_1: 2.0379, loss_box_1: 0.1256, loss_cns_1: 0.0259, loss_yns_1: 0.0058, loss_cls_2: 2.1118, loss_box_2: 0.2271, loss_cns_2: 0.0205, loss_yns_2: 0.0096, loss_cls_3: 1.9537, loss_box_3: 0.4594, loss_cns_3: 0.0593, loss_yns_3: 0.0202, loss_cls_4: 1.7976, loss_box_4: 1.5458, loss_cns_4: 0.1541, loss_yns_4: 0.0551, loss_cls_5: 2.0597, loss_box_5: 0.5620, loss_cns_5: 0.0615, loss_yns_5: 0.0194, loss_cls_dn_0: 1.0253, loss_box_dn_0: 1.2987, loss_cls_dn_1: 0.9587, loss_box_dn_1: 2.4141, loss_cls_dn_2: 0.9742, loss_box_dn_2: 2.5263, loss_cls_dn_3: 0.9109, loss_box_dn_3: 2.6100, loss_cls_dn_4: 0.8402, loss_box_dn_4: 2.8645, loss_cls_dn_5: 0.9866, loss_box_dn_5: 3.1079, loss_dense_depth: 1.7144, loss: 37.6199, grad_norm: 66.8553 -2025-11-12 14:30:40,806 - mmdet - INFO - Iter [3/17500] lr: 1.008e-04, eta: 8 days, 3:04:10, time: 1.538, data_time: 0.084, memory: 49164, loss_cls_0: 1.4452, loss_box_0: 2.5022, loss_cns_0: 0.6104, loss_yns_0: 0.2070, loss_cls_1: 1.7651, loss_box_1: 1.7029, loss_cns_1: 0.2667, loss_yns_1: 0.0982, loss_cls_2: 1.7911, loss_box_2: 3.7597, loss_cns_2: 0.3550, loss_yns_2: 0.1883, loss_cls_3: 1.6298, loss_box_3: 4.7163, loss_cns_3: 0.4146, loss_yns_3: 0.2065, loss_cls_4: 1.5890, loss_box_4: 3.8499, loss_cns_4: 0.3576, loss_yns_4: 0.1570, loss_cls_5: 1.6941, loss_box_5: 2.7867, loss_cns_5: 0.2157, loss_yns_5: 0.0916, loss_cls_dn_0: 0.6987, loss_box_dn_0: 1.1508, loss_cls_dn_1: 0.8290, loss_box_dn_1: 2.4190, loss_cls_dn_2: 0.8095, loss_box_dn_2: 2.6327, loss_cls_dn_3: 0.7144, loss_box_dn_3: 2.8313, loss_cls_dn_4: 0.7283, loss_box_dn_4: 3.1237, loss_cls_dn_5: 0.8077, loss_box_dn_5: 3.3585, loss_dense_depth: 1.6661, loss: 54.1706, grad_norm: 101.9969 -2025-11-12 14:30:42,400 - mmdet - INFO - Iter [4/17500] lr: 1.012e-04, eta: 6 days, 4:13:37, time: 1.591, data_time: 0.078, memory: 49164, loss_cls_0: 1.3616, loss_box_0: 2.5400, loss_cns_0: 0.5561, loss_yns_0: 0.1824, loss_cls_1: 1.6168, loss_box_1: 3.0691, loss_cns_1: 0.4582, loss_yns_1: 0.2034, loss_cls_2: 1.7151, loss_box_2: 3.6811, loss_cns_2: 0.4411, loss_yns_2: 0.1889, loss_cls_3: 1.5244, loss_box_3: 4.2430, loss_cns_3: 0.4458, loss_yns_3: 0.2175, loss_cls_4: 1.4702, loss_box_4: 4.8546, loss_cns_4: 0.3481, loss_yns_4: 0.1957, loss_cls_5: 1.5012, loss_box_5: 5.0785, loss_cns_5: 0.4416, loss_yns_5: 0.1878, loss_cls_dn_0: 0.5674, loss_box_dn_0: 1.1654, loss_cls_dn_1: 0.7236, loss_box_dn_1: 2.6004, loss_cls_dn_2: 0.6985, loss_box_dn_2: 2.6750, loss_cls_dn_3: 0.6139, loss_box_dn_3: 2.8883, loss_cls_dn_4: 0.5983, loss_box_dn_4: 3.1083, loss_cls_dn_5: 0.6716, loss_box_dn_5: 3.2954, loss_dense_depth: 1.5639, loss: 57.6921, grad_norm: 132.4867 -2025-11-12 14:30:43,988 - mmdet - INFO - Iter [5/17500] lr: 1.016e-04, eta: 5 days, 0:07:02, time: 1.587, data_time: 0.084, memory: 49164, loss_cls_0: 1.3311, loss_box_0: 2.8235, loss_cns_0: 0.4916, loss_yns_0: 0.2004, loss_cls_1: 1.5605, loss_box_1: 3.9524, loss_cns_1: 0.4015, loss_yns_1: 0.1983, loss_cls_2: 1.6093, loss_box_2: 4.0175, loss_cns_2: 0.3733, loss_yns_2: 0.2009, loss_cls_3: 1.4489, loss_box_3: 4.1501, loss_cns_3: 0.3961, loss_yns_3: 0.1975, loss_cls_4: 1.4170, loss_box_4: 4.2538, loss_cns_4: 0.3760, loss_yns_4: 0.1913, loss_cls_5: 1.3943, loss_box_5: 4.4721, loss_cns_5: 0.4127, loss_yns_5: 0.1965, loss_cls_dn_0: 0.5269, loss_box_dn_0: 1.2673, loss_cls_dn_1: 0.6597, loss_box_dn_1: 2.2531, loss_cls_dn_2: 0.6554, loss_box_dn_2: 2.3638, loss_cls_dn_3: 0.5644, loss_box_dn_3: 2.5044, loss_cls_dn_4: 0.5513, loss_box_dn_4: 2.6471, loss_cls_dn_5: 0.5837, loss_box_dn_5: 2.7150, loss_dense_depth: 1.4994, loss: 54.8581, grad_norm: 111.3347 -2025-11-12 14:30:45,603 - mmdet - INFO - Iter [6/17500] lr: 1.020e-04, eta: 4 days, 5:23:50, time: 1.611, data_time: 0.104, memory: 49164, loss_cls_0: 1.3035, loss_box_0: 2.5345, loss_cns_0: 0.5614, loss_yns_0: 0.1818, loss_cls_1: 1.4918, loss_box_1: 3.8888, loss_cns_1: 0.3700, loss_yns_1: 0.1932, loss_cls_2: 1.4853, loss_box_2: 4.0343, loss_cns_2: 0.3576, loss_yns_2: 0.1897, loss_cls_3: 1.3535, loss_box_3: 4.0761, loss_cns_3: 0.3479, loss_yns_3: 0.1933, loss_cls_4: 1.3177, loss_box_4: 4.3302, loss_cns_4: 0.3071, loss_yns_4: 0.1986, loss_cls_5: 1.3398, loss_box_5: 4.4594, loss_cns_5: 0.3050, loss_yns_5: 0.1995, loss_cls_dn_0: 0.5147, loss_box_dn_0: 1.1975, loss_cls_dn_1: 0.5956, loss_box_dn_1: 2.4218, loss_cls_dn_2: 0.5848, loss_box_dn_2: 2.4526, loss_cls_dn_3: 0.5214, loss_box_dn_3: 2.5115, loss_cls_dn_4: 0.4918, loss_box_dn_4: 2.7124, loss_cls_dn_5: 0.4868, loss_box_dn_5: 2.7429, loss_dense_depth: 1.4640, loss: 53.7175, grad_norm: 116.1794 -2025-11-12 14:30:47,180 - mmdet - INFO - Iter [7/17500] lr: 1.024e-04, eta: 3 days, 16:00:19, time: 1.582, data_time: 0.084, memory: 49164, loss_cls_0: 1.2624, loss_box_0: 2.3112, loss_cns_0: 0.6560, loss_yns_0: 0.1792, loss_cls_1: 1.3969, loss_box_1: 3.5179, loss_cns_1: 0.4499, loss_yns_1: 0.1993, loss_cls_2: 1.4259, loss_box_2: 3.5974, loss_cns_2: 0.4609, loss_yns_2: 0.1854, loss_cls_3: 1.3087, loss_box_3: 3.4763, loss_cns_3: 0.4754, loss_yns_3: 0.1892, loss_cls_4: 1.2836, loss_box_4: 3.7487, loss_cns_4: 0.4363, loss_yns_4: 0.1934, loss_cls_5: 1.3256, loss_box_5: 4.0202, loss_cns_5: 0.4049, loss_yns_5: 0.1841, loss_cls_dn_0: 0.5179, loss_box_dn_0: 1.1090, loss_cls_dn_1: 0.5331, loss_box_dn_1: 2.4554, loss_cls_dn_2: 0.5264, loss_box_dn_2: 2.4067, loss_cls_dn_3: 0.4763, loss_box_dn_3: 2.4161, loss_cls_dn_4: 0.4509, loss_box_dn_4: 2.6143, loss_cls_dn_5: 0.4365, loss_box_dn_5: 2.7031, loss_dense_depth: 1.4320, loss: 50.7665, grad_norm: 98.1441 -2025-11-12 14:30:48,770 - mmdet - INFO - Iter [8/17500] lr: 1.028e-04, eta: 3 days, 5:57:47, time: 1.585, data_time: 0.097, memory: 49164, loss_cls_0: 1.2264, loss_box_0: 2.2557, loss_cns_0: 0.6386, loss_yns_0: 0.1848, loss_cls_1: 1.3168, loss_box_1: 3.5499, loss_cns_1: 0.4836, loss_yns_1: 0.1846, loss_cls_2: 1.3852, loss_box_2: 3.7037, loss_cns_2: 0.4235, loss_yns_2: 0.1807, loss_cls_3: 1.3005, loss_box_3: 3.6425, loss_cns_3: 0.4281, loss_yns_3: 0.1833, loss_cls_4: 1.2872, loss_box_4: 3.7018, loss_cns_4: 0.4363, loss_yns_4: 0.1856, loss_cls_5: 1.3282, loss_box_5: 3.7832, loss_cns_5: 0.4639, loss_yns_5: 0.1894, loss_cls_dn_0: 0.5169, loss_box_dn_0: 1.0402, loss_cls_dn_1: 0.5345, loss_box_dn_1: 1.8415, loss_cls_dn_2: 0.5444, loss_box_dn_2: 1.7516, loss_cls_dn_3: 0.4928, loss_box_dn_3: 1.7501, loss_cls_dn_4: 0.4627, loss_box_dn_4: 1.8736, loss_cls_dn_5: 0.4383, loss_box_dn_5: 1.9130, loss_dense_depth: 1.3705, loss: 46.9935, grad_norm: 75.5766 -2025-11-12 14:30:50,348 - mmdet - INFO - Iter [9/17500] lr: 1.032e-04, eta: 2 days, 22:09:12, time: 1.587, data_time: 0.090, memory: 49164, loss_cls_0: 1.2225, loss_box_0: 2.2135, loss_cns_0: 0.6143, loss_yns_0: 0.1786, loss_cls_1: 1.2765, loss_box_1: 3.3283, loss_cns_1: 0.5177, loss_yns_1: 0.1840, loss_cls_2: 1.3696, loss_box_2: 3.4780, loss_cns_2: 0.4412, loss_yns_2: 0.1770, loss_cls_3: 1.2815, loss_box_3: 3.5966, loss_cns_3: 0.4594, loss_yns_3: 0.1970, loss_cls_4: 1.2698, loss_box_4: 3.4915, loss_cns_4: 0.4683, loss_yns_4: 0.1957, loss_cls_5: 1.3053, loss_box_5: 3.5567, loss_cns_5: 0.4835, loss_yns_5: 0.1833, loss_cls_dn_0: 0.5055, loss_box_dn_0: 1.0255, loss_cls_dn_1: 0.4872, loss_box_dn_1: 1.5167, loss_cls_dn_2: 0.5186, loss_box_dn_2: 1.5108, loss_cls_dn_3: 0.4713, loss_box_dn_3: 1.6062, loss_cls_dn_4: 0.4504, loss_box_dn_4: 1.5558, loss_cls_dn_5: 0.4312, loss_box_dn_5: 1.6611, loss_dense_depth: 1.3134, loss: 44.5435, grad_norm: 68.2433 -2025-11-12 14:30:51,914 - mmdet - INFO - Iter [10/17500] lr: 1.036e-04, eta: 2 days, 15:53:41, time: 1.565, data_time: 0.077, memory: 49164, loss_cls_0: 1.2105, loss_box_0: 2.2085, loss_cns_0: 0.6146, loss_yns_0: 0.1735, loss_cls_1: 1.2789, loss_box_1: 3.1561, loss_cns_1: 0.5455, loss_yns_1: 0.1847, loss_cls_2: 1.2829, loss_box_2: 3.2134, loss_cns_2: 0.5071, loss_yns_2: 0.1811, loss_cls_3: 1.2673, loss_box_3: 3.3737, loss_cns_3: 0.5460, loss_yns_3: 0.1958, loss_cls_4: 1.2479, loss_box_4: 3.2924, loss_cns_4: 0.5483, loss_yns_4: 0.1835, loss_cls_5: 1.2669, loss_box_5: 3.4219, loss_cns_5: 0.5438, loss_yns_5: 0.1794, loss_cls_dn_0: 0.4830, loss_box_dn_0: 1.0323, loss_cls_dn_1: 0.4430, loss_box_dn_1: 1.5932, loss_cls_dn_2: 0.4768, loss_box_dn_2: 1.6335, loss_cls_dn_3: 0.4349, loss_box_dn_3: 1.7453, loss_cls_dn_4: 0.4319, loss_box_dn_4: 1.7322, loss_cls_dn_5: 0.4321, loss_box_dn_5: 1.8770, loss_dense_depth: 1.3181, loss: 44.2566, grad_norm: 72.3677 -2025-11-12 14:30:53,488 - mmdet - INFO - Iter [11/17500] lr: 1.040e-04, eta: 2 days, 10:46:34, time: 1.570, data_time: 0.076, memory: 49164, loss_cls_0: 1.2253, loss_box_0: 2.2452, loss_cns_0: 0.6183, loss_yns_0: 0.1792, loss_cls_1: 1.2947, loss_box_1: 3.0589, loss_cns_1: 0.5349, loss_yns_1: 0.1785, loss_cls_2: 1.2778, loss_box_2: 3.0024, loss_cns_2: 0.5327, loss_yns_2: 0.1826, loss_cls_3: 1.2630, loss_box_3: 3.0951, loss_cns_3: 0.5705, loss_yns_3: 0.1809, loss_cls_4: 1.2455, loss_box_4: 3.0444, loss_cns_4: 0.5630, loss_yns_4: 0.1829, loss_cls_5: 1.2590, loss_box_5: 3.2201, loss_cns_5: 0.5397, loss_yns_5: 0.1822, loss_cls_dn_0: 0.4645, loss_box_dn_0: 1.0524, loss_cls_dn_1: 0.4121, loss_box_dn_1: 1.8054, loss_cls_dn_2: 0.4420, loss_box_dn_2: 1.8242, loss_cls_dn_3: 0.4108, loss_box_dn_3: 1.8965, loss_cls_dn_4: 0.4030, loss_box_dn_4: 1.9234, loss_cls_dn_5: 0.4208, loss_box_dn_5: 2.0693, loss_dense_depth: 1.3003, loss: 44.1014, grad_norm: 68.2458 -2025-11-12 14:30:55,055 - mmdet - INFO - Iter [12/17500] lr: 1.044e-04, eta: 2 days, 6:30:34, time: 1.567, data_time: 0.082, memory: 49164, loss_cls_0: 1.2307, loss_box_0: 2.2433, loss_cns_0: 0.6189, loss_yns_0: 0.1742, loss_cls_1: 1.2697, loss_box_1: 2.9500, loss_cns_1: 0.5018, loss_yns_1: 0.1759, loss_cls_2: 1.2872, loss_box_2: 2.9521, loss_cns_2: 0.5078, loss_yns_2: 0.1766, loss_cls_3: 1.2412, loss_box_3: 2.9297, loss_cns_3: 0.5312, loss_yns_3: 0.1787, loss_cls_4: 1.2480, loss_box_4: 2.9304, loss_cns_4: 0.5548, loss_yns_4: 0.1770, loss_cls_5: 1.2645, loss_box_5: 3.1073, loss_cns_5: 0.5327, loss_yns_5: 0.1812, loss_cls_dn_0: 0.4564, loss_box_dn_0: 1.0502, loss_cls_dn_1: 0.3867, loss_box_dn_1: 2.0114, loss_cls_dn_2: 0.4168, loss_box_dn_2: 2.0101, loss_cls_dn_3: 0.3965, loss_box_dn_3: 2.0305, loss_cls_dn_4: 0.3798, loss_box_dn_4: 2.0847, loss_cls_dn_5: 0.3970, loss_box_dn_5: 2.1722, loss_dense_depth: 1.3081, loss: 44.0655, grad_norm: 62.7498 -2025-11-12 14:30:56,623 - mmdet - INFO - Iter [13/17500] lr: 1.048e-04, eta: 2 days, 2:54:03, time: 1.572, data_time: 0.077, memory: 49164, loss_cls_0: 1.2034, loss_box_0: 2.2508, loss_cns_0: 0.5997, loss_yns_0: 0.1705, loss_cls_1: 1.2339, loss_box_1: 2.8291, loss_cns_1: 0.5192, loss_yns_1: 0.1804, loss_cls_2: 1.2702, loss_box_2: 2.8996, loss_cns_2: 0.5372, loss_yns_2: 0.1775, loss_cls_3: 1.2325, loss_box_3: 2.8655, loss_cns_3: 0.5366, loss_yns_3: 0.1773, loss_cls_4: 1.2646, loss_box_4: 2.9061, loss_cns_4: 0.5266, loss_yns_4: 0.1821, loss_cls_5: 1.2759, loss_box_5: 2.9121, loss_cns_5: 0.5331, loss_yns_5: 0.1796, loss_cls_dn_0: 0.4546, loss_box_dn_0: 1.0399, loss_cls_dn_1: 0.4288, loss_box_dn_1: 1.5131, loss_cls_dn_2: 0.4480, loss_box_dn_2: 1.5499, loss_cls_dn_3: 0.4489, loss_box_dn_3: 1.5717, loss_cls_dn_4: 0.4271, loss_box_dn_4: 1.7238, loss_cls_dn_5: 0.4423, loss_box_dn_5: 1.7069, loss_dense_depth: 1.1877, loss: 41.4061, grad_norm: 60.3168 -2025-11-12 14:30:58,202 - mmdet - INFO - Iter [14/17500] lr: 1.052e-04, eta: 1 day, 23:48:38, time: 1.580, data_time: 0.078, memory: 49164, loss_cls_0: 1.2000, loss_box_0: 2.2327, loss_cns_0: 0.5908, loss_yns_0: 0.1717, loss_cls_1: 1.2678, loss_box_1: 2.7237, loss_cns_1: 0.5487, loss_yns_1: 0.1746, loss_cls_2: 1.2819, loss_box_2: 2.8041, loss_cns_2: 0.5549, loss_yns_2: 0.1788, loss_cls_3: 1.2671, loss_box_3: 2.8389, loss_cns_3: 0.5563, loss_yns_3: 0.1782, loss_cls_4: 1.2952, loss_box_4: 2.8698, loss_cns_4: 0.5336, loss_yns_4: 0.1813, loss_cls_5: 1.2925, loss_box_5: 2.8829, loss_cns_5: 0.5571, loss_yns_5: 0.1791, loss_cls_dn_0: 0.4655, loss_box_dn_0: 1.0232, loss_cls_dn_1: 0.4497, loss_box_dn_1: 1.2845, loss_cls_dn_2: 0.4655, loss_box_dn_2: 1.2978, loss_cls_dn_3: 0.4661, loss_box_dn_3: 1.3828, loss_cls_dn_4: 0.4560, loss_box_dn_4: 1.5662, loss_cls_dn_5: 0.4607, loss_box_dn_5: 1.5289, loss_dense_depth: 1.2853, loss: 40.4936, grad_norm: 81.1896 -2025-11-12 14:30:59,763 - mmdet - INFO - Iter [15/17500] lr: 1.056e-04, eta: 1 day, 21:07:30, time: 1.558, data_time: 0.073, memory: 49164, loss_cls_0: 1.1910, loss_box_0: 2.2967, loss_cns_0: 0.5891, loss_yns_0: 0.1738, loss_cls_1: 1.2985, loss_box_1: 2.6433, loss_cns_1: 0.5750, loss_yns_1: 0.1736, loss_cls_2: 1.2826, loss_box_2: 2.6312, loss_cns_2: 0.5736, loss_yns_2: 0.1756, loss_cls_3: 1.2951, loss_box_3: 2.6945, loss_cns_3: 0.5775, loss_yns_3: 0.1789, loss_cls_4: 1.2886, loss_box_4: 2.7194, loss_cns_4: 0.5672, loss_yns_4: 0.1784, loss_cls_5: 1.2873, loss_box_5: 2.8238, loss_cns_5: 0.5909, loss_yns_5: 0.1789, loss_cls_dn_0: 0.4673, loss_box_dn_0: 1.0408, loss_cls_dn_1: 0.4499, loss_box_dn_1: 1.3781, loss_cls_dn_2: 0.4594, loss_box_dn_2: 1.3573, loss_cls_dn_3: 0.4508, loss_box_dn_3: 1.4397, loss_cls_dn_4: 0.4610, loss_box_dn_4: 1.5371, loss_cls_dn_5: 0.4584, loss_box_dn_5: 1.5663, loss_dense_depth: 1.1377, loss: 40.1886, grad_norm: 69.2365 -2025-11-12 14:31:01,336 - mmdet - INFO - Iter [16/17500] lr: 1.060e-04, eta: 1 day, 18:46:44, time: 1.570, data_time: 0.076, memory: 49164, loss_cls_0: 1.1813, loss_box_0: 2.3157, loss_cns_0: 0.5883, loss_yns_0: 0.1755, loss_cls_1: 1.2878, loss_box_1: 2.7088, loss_cns_1: 0.5694, loss_yns_1: 0.1744, loss_cls_2: 1.2947, loss_box_2: 2.7080, loss_cns_2: 0.5626, loss_yns_2: 0.1758, loss_cls_3: 1.3094, loss_box_3: 2.7068, loss_cns_3: 0.5702, loss_yns_3: 0.1753, loss_cls_4: 1.2605, loss_box_4: 2.7681, loss_cns_4: 0.5612, loss_yns_4: 0.1768, loss_cls_5: 1.2748, loss_box_5: 2.8488, loss_cns_5: 0.5819, loss_yns_5: 0.1726, loss_cls_dn_0: 0.4864, loss_box_dn_0: 1.0302, loss_cls_dn_1: 0.4666, loss_box_dn_1: 1.3610, loss_cls_dn_2: 0.4752, loss_box_dn_2: 1.3407, loss_cls_dn_3: 0.4433, loss_box_dn_3: 1.4193, loss_cls_dn_4: 0.4695, loss_box_dn_4: 1.4695, loss_cls_dn_5: 0.4736, loss_box_dn_5: 1.5521, loss_dense_depth: 1.2629, loss: 40.3992, grad_norm: 71.4048 -2025-11-12 14:31:02,919 - mmdet - INFO - Iter [17/17500] lr: 1.064e-04, eta: 1 day, 16:42:51, time: 1.589, data_time: 0.082, memory: 49164, loss_cls_0: 1.1741, loss_box_0: 2.2884, loss_cns_0: 0.5908, loss_yns_0: 0.1735, loss_cls_1: 1.2723, loss_box_1: 2.9321, loss_cns_1: 0.5397, loss_yns_1: 0.1736, loss_cls_2: 1.2882, loss_box_2: 2.9668, loss_cns_2: 0.5361, loss_yns_2: 0.1749, loss_cls_3: 1.2961, loss_box_3: 2.9227, loss_cns_3: 0.5420, loss_yns_3: 0.1750, loss_cls_4: 1.2572, loss_box_4: 2.9382, loss_cns_4: 0.5518, loss_yns_4: 0.1762, loss_cls_5: 1.2710, loss_box_5: 2.9331, loss_cns_5: 0.5470, loss_yns_5: 0.1730, loss_cls_dn_0: 0.4806, loss_box_dn_0: 1.0236, loss_cls_dn_1: 0.4559, loss_box_dn_1: 1.4996, loss_cls_dn_2: 0.4599, loss_box_dn_2: 1.4983, loss_cls_dn_3: 0.4214, loss_box_dn_3: 1.5645, loss_cls_dn_4: 0.4429, loss_box_dn_4: 1.5862, loss_cls_dn_5: 0.4610, loss_box_dn_5: 1.7016, loss_dense_depth: 1.1308, loss: 41.6199, grad_norm: 63.1654 -2025-11-12 14:31:04,494 - mmdet - INFO - Iter [18/17500] lr: 1.068e-04, eta: 1 day, 14:52:29, time: 1.574, data_time: 0.076, memory: 49164, loss_cls_0: 1.1787, loss_box_0: 2.2230, loss_cns_0: 0.6050, loss_yns_0: 0.1717, loss_cls_1: 1.2493, loss_box_1: 2.8918, loss_cns_1: 0.5351, loss_yns_1: 0.1774, loss_cls_2: 1.2605, loss_box_2: 2.9365, loss_cns_2: 0.5371, loss_yns_2: 0.1724, loss_cls_3: 1.2623, loss_box_3: 2.8936, loss_cns_3: 0.5414, loss_yns_3: 0.1752, loss_cls_4: 1.2501, loss_box_4: 2.9077, loss_cns_4: 0.5642, loss_yns_4: 0.1714, loss_cls_5: 1.2586, loss_box_5: 2.9516, loss_cns_5: 0.5361, loss_yns_5: 0.1737, loss_cls_dn_0: 0.4820, loss_box_dn_0: 1.0158, loss_cls_dn_1: 0.4470, loss_box_dn_1: 1.5412, loss_cls_dn_2: 0.4523, loss_box_dn_2: 1.5716, loss_cls_dn_3: 0.4182, loss_box_dn_3: 1.6432, loss_cls_dn_4: 0.4293, loss_box_dn_4: 1.6459, loss_cls_dn_5: 0.4540, loss_box_dn_5: 1.8061, loss_dense_depth: 1.0953, loss: 41.6264, grad_norm: 57.0068 -2025-11-12 14:31:06,073 - mmdet - INFO - Iter [19/17500] lr: 1.072e-04, eta: 1 day, 13:13:49, time: 1.580, data_time: 0.074, memory: 49164, loss_cls_0: 1.1754, loss_box_0: 2.2418, loss_cns_0: 0.5928, loss_yns_0: 0.1699, loss_cls_1: 1.2419, loss_box_1: 2.8203, loss_cns_1: 0.5418, loss_yns_1: 0.1802, loss_cls_2: 1.2613, loss_box_2: 2.8565, loss_cns_2: 0.5497, loss_yns_2: 0.1732, loss_cls_3: 1.2641, loss_box_3: 2.8662, loss_cns_3: 0.5390, loss_yns_3: 0.1735, loss_cls_4: 1.2636, loss_box_4: 2.8742, loss_cns_4: 0.5675, loss_yns_4: 0.1760, loss_cls_5: 1.2575, loss_box_5: 2.9768, loss_cns_5: 0.5324, loss_yns_5: 0.1742, loss_cls_dn_0: 0.4914, loss_box_dn_0: 1.0334, loss_cls_dn_1: 0.4546, loss_box_dn_1: 1.2878, loss_cls_dn_2: 0.4570, loss_box_dn_2: 1.3643, loss_cls_dn_3: 0.4345, loss_box_dn_3: 1.4638, loss_cls_dn_4: 0.4352, loss_box_dn_4: 1.4563, loss_cls_dn_5: 0.4599, loss_box_dn_5: 1.6381, loss_dense_depth: 1.1569, loss: 40.6030, grad_norm: 77.8069 -2025-11-12 14:31:07,639 - mmdet - INFO - Iter [20/17500] lr: 1.076e-04, eta: 1 day, 11:44:42, time: 1.558, data_time: 0.076, memory: 49164, loss_cls_0: 1.1651, loss_box_0: 2.2431, loss_cns_0: 0.5910, loss_yns_0: 0.1689, loss_cls_1: 1.2450, loss_box_1: 2.7939, loss_cns_1: 0.5623, loss_yns_1: 0.1778, loss_cls_2: 1.2775, loss_box_2: 2.7881, loss_cns_2: 0.5716, loss_yns_2: 0.1752, loss_cls_3: 1.2684, loss_box_3: 2.8347, loss_cns_3: 0.5602, loss_yns_3: 0.1769, loss_cls_4: 1.2689, loss_box_4: 2.7722, loss_cns_4: 0.5719, loss_yns_4: 0.1797, loss_cls_5: 1.2662, loss_box_5: 2.8435, loss_cns_5: 0.5712, loss_yns_5: 0.1831, loss_cls_dn_0: 0.4808, loss_box_dn_0: 1.0265, loss_cls_dn_1: 0.4148, loss_box_dn_1: 1.4226, loss_cls_dn_2: 0.4159, loss_box_dn_2: 1.4585, loss_cls_dn_3: 0.4071, loss_box_dn_3: 1.5184, loss_cls_dn_4: 0.4049, loss_box_dn_4: 1.4797, loss_cls_dn_5: 0.4168, loss_box_dn_5: 1.5647, loss_dense_depth: 1.0508, loss: 40.3179, grad_norm: 65.5594 -2025-11-12 14:31:09,309 - mmdet - INFO - Iter [21/17500] lr: 1.080e-04, eta: 1 day, 10:25:41, time: 1.673, data_time: 0.143, memory: 49164, loss_cls_0: 1.1541, loss_box_0: 2.1931, loss_cns_0: 0.5961, loss_yns_0: 0.1696, loss_cls_1: 1.2543, loss_box_1: 2.7743, loss_cns_1: 0.5867, loss_yns_1: 0.1747, loss_cls_2: 1.2838, loss_box_2: 2.7579, loss_cns_2: 0.5912, loss_yns_2: 0.1765, loss_cls_3: 1.2605, loss_box_3: 2.7851, loss_cns_3: 0.5917, loss_yns_3: 0.1735, loss_cls_4: 1.2505, loss_box_4: 2.7870, loss_cns_4: 0.5736, loss_yns_4: 0.1790, loss_cls_5: 1.2728, loss_box_5: 2.7887, loss_cns_5: 0.5941, loss_yns_5: 0.1824, loss_cls_dn_0: 0.4740, loss_box_dn_0: 1.0396, loss_cls_dn_1: 0.4286, loss_box_dn_1: 1.1820, loss_cls_dn_2: 0.4252, loss_box_dn_2: 1.2072, loss_cls_dn_3: 0.4359, loss_box_dn_3: 1.2611, loss_cls_dn_4: 0.4334, loss_box_dn_4: 1.2899, loss_cls_dn_5: 0.4346, loss_box_dn_5: 1.2881, loss_dense_depth: 1.0816, loss: 39.1323, grad_norm: 50.6531 -2025-11-12 14:31:10,928 - mmdet - INFO - Iter [22/17500] lr: 1.084e-04, eta: 1 day, 9:13:06, time: 1.618, data_time: 0.103, memory: 49164, loss_cls_0: 1.1789, loss_box_0: 2.1402, loss_cns_0: 0.6043, loss_yns_0: 0.1717, loss_cls_1: 1.2623, loss_box_1: 2.7971, loss_cns_1: 0.5712, loss_yns_1: 0.1689, loss_cls_2: 1.2729, loss_box_2: 2.8358, loss_cns_2: 0.5548, loss_yns_2: 0.1771, loss_cls_3: 1.2541, loss_box_3: 2.8733, loss_cns_3: 0.5637, loss_yns_3: 0.1732, loss_cls_4: 1.2444, loss_box_4: 2.9276, loss_cns_4: 0.5552, loss_yns_4: 0.1801, loss_cls_5: 1.2560, loss_box_5: 2.9536, loss_cns_5: 0.5637, loss_yns_5: 0.1797, loss_cls_dn_0: 0.4589, loss_box_dn_0: 1.0265, loss_cls_dn_1: 0.4420, loss_box_dn_1: 1.1337, loss_cls_dn_2: 0.4384, loss_box_dn_2: 1.1956, loss_cls_dn_3: 0.4577, loss_box_dn_3: 1.2869, loss_cls_dn_4: 0.4551, loss_box_dn_4: 1.4070, loss_cls_dn_5: 0.4481, loss_box_dn_5: 1.4040, loss_dense_depth: 1.0422, loss: 39.6562, grad_norm: 71.5034 -2025-11-12 14:31:12,518 - mmdet - INFO - Iter [23/17500] lr: 1.088e-04, eta: 1 day, 8:06:24, time: 1.585, data_time: 0.089, memory: 49164, loss_cls_0: 1.1622, loss_box_0: 2.1185, loss_cns_0: 0.6116, loss_yns_0: 0.1735, loss_cls_1: 1.2264, loss_box_1: 2.9637, loss_cns_1: 0.5292, loss_yns_1: 0.1699, loss_cls_2: 1.2502, loss_box_2: 2.9067, loss_cns_2: 0.5400, loss_yns_2: 0.1777, loss_cls_3: 1.2348, loss_box_3: 2.9381, loss_cns_3: 0.5482, loss_yns_3: 0.1758, loss_cls_4: 1.2270, loss_box_4: 2.9768, loss_cns_4: 0.5600, loss_yns_4: 0.1713, loss_cls_5: 1.2385, loss_box_5: 3.0048, loss_cns_5: 0.5601, loss_yns_5: 0.1734, loss_cls_dn_0: 0.4551, loss_box_dn_0: 0.9997, loss_cls_dn_1: 0.4449, loss_box_dn_1: 1.2294, loss_cls_dn_2: 0.4462, loss_box_dn_2: 1.2540, loss_cls_dn_3: 0.4625, loss_box_dn_3: 1.3587, loss_cls_dn_4: 0.4496, loss_box_dn_4: 1.5119, loss_cls_dn_5: 0.4527, loss_box_dn_5: 1.5146, loss_dense_depth: 1.0612, loss: 40.2788, grad_norm: 73.8528 -2025-11-12 14:31:14,115 - mmdet - INFO - Iter [24/17500] lr: 1.092e-04, eta: 1 day, 7:05:28, time: 1.601, data_time: 0.083, memory: 49164, loss_cls_0: 1.1745, loss_box_0: 2.1177, loss_cns_0: 0.6174, loss_yns_0: 0.1737, loss_cls_1: 1.2315, loss_box_1: 2.9614, loss_cns_1: 0.5446, loss_yns_1: 0.1733, loss_cls_2: 1.2438, loss_box_2: 2.8767, loss_cns_2: 0.5608, loss_yns_2: 0.1814, loss_cls_3: 1.2388, loss_box_3: 2.9017, loss_cns_3: 0.5576, loss_yns_3: 0.1785, loss_cls_4: 1.2351, loss_box_4: 2.9343, loss_cns_4: 0.5608, loss_yns_4: 0.1721, loss_cls_5: 1.2506, loss_box_5: 2.8953, loss_cns_5: 0.5629, loss_yns_5: 0.1759, loss_cls_dn_0: 0.4670, loss_box_dn_0: 1.0096, loss_cls_dn_1: 0.4323, loss_box_dn_1: 1.3606, loss_cls_dn_2: 0.4380, loss_box_dn_2: 1.3513, loss_cls_dn_3: 0.4456, loss_box_dn_3: 1.4414, loss_cls_dn_4: 0.4260, loss_box_dn_4: 1.5877, loss_cls_dn_5: 0.4412, loss_box_dn_5: 1.5640, loss_dense_depth: 1.0904, loss: 40.5757, grad_norm: 65.5368 -2025-11-12 14:31:15,687 - mmdet - INFO - Iter [25/17500] lr: 1.096e-04, eta: 1 day, 6:09:01, time: 1.569, data_time: 0.088, memory: 49164, loss_cls_0: 1.1645, loss_box_0: 2.1124, loss_cns_0: 0.6181, loss_yns_0: 0.1745, loss_cls_1: 1.2266, loss_box_1: 2.7581, loss_cns_1: 0.5517, loss_yns_1: 0.1723, loss_cls_2: 1.2325, loss_box_2: 2.7258, loss_cns_2: 0.5692, loss_yns_2: 0.1757, loss_cls_3: 1.2348, loss_box_3: 2.7391, loss_cns_3: 0.5679, loss_yns_3: 0.1713, loss_cls_4: 1.2415, loss_box_4: 2.7421, loss_cns_4: 0.5708, loss_yns_4: 0.1780, loss_cls_5: 1.2458, loss_box_5: 2.7407, loss_cns_5: 0.5818, loss_yns_5: 0.1718, loss_cls_dn_0: 0.4666, loss_box_dn_0: 0.9923, loss_cls_dn_1: 0.4189, loss_box_dn_1: 1.3339, loss_cls_dn_2: 0.4272, loss_box_dn_2: 1.3475, loss_cls_dn_3: 0.4262, loss_box_dn_3: 1.4171, loss_cls_dn_4: 0.4049, loss_box_dn_4: 1.5208, loss_cls_dn_5: 0.4286, loss_box_dn_5: 1.5080, loss_dense_depth: 1.0835, loss: 39.4424, grad_norm: 59.4028 -2025-11-12 14:31:17,277 - mmdet - INFO - Iter [26/17500] lr: 1.100e-04, eta: 1 day, 5:17:11, time: 1.592, data_time: 0.101, memory: 49164, loss_cls_0: 1.1500, loss_box_0: 2.0926, loss_cns_0: 0.6190, loss_yns_0: 0.1737, loss_cls_1: 1.2154, loss_box_1: 2.6694, loss_cns_1: 0.5669, loss_yns_1: 0.1708, loss_cls_2: 1.2307, loss_box_2: 2.6926, loss_cns_2: 0.5838, loss_yns_2: 0.1721, loss_cls_3: 1.2392, loss_box_3: 2.7234, loss_cns_3: 0.5832, loss_yns_3: 0.1729, loss_cls_4: 1.2458, loss_box_4: 2.7320, loss_cns_4: 0.5847, loss_yns_4: 0.1764, loss_cls_5: 1.2382, loss_box_5: 2.7449, loss_cns_5: 0.5992, loss_yns_5: 0.1746, loss_cls_dn_0: 0.4509, loss_box_dn_0: 0.9948, loss_cls_dn_1: 0.4336, loss_box_dn_1: 1.1535, loss_cls_dn_2: 0.4401, loss_box_dn_2: 1.2022, loss_cls_dn_3: 0.4357, loss_box_dn_3: 1.2563, loss_cls_dn_4: 0.4178, loss_box_dn_4: 1.3263, loss_cls_dn_5: 0.4409, loss_box_dn_5: 1.3278, loss_dense_depth: 1.0457, loss: 38.4774, grad_norm: 82.9434 -2025-11-12 14:31:18,851 - mmdet - INFO - Iter [27/17500] lr: 1.104e-04, eta: 1 day, 4:28:55, time: 1.568, data_time: 0.078, memory: 49164, loss_cls_0: 1.1410, loss_box_0: 2.1274, loss_cns_0: 0.6125, loss_yns_0: 0.1739, loss_cls_1: 1.2130, loss_box_1: 2.7238, loss_cns_1: 0.5684, loss_yns_1: 0.1728, loss_cls_2: 1.2301, loss_box_2: 2.7152, loss_cns_2: 0.5772, loss_yns_2: 0.1751, loss_cls_3: 1.2284, loss_box_3: 2.7174, loss_cns_3: 0.5797, loss_yns_3: 0.1751, loss_cls_4: 1.2274, loss_box_4: 2.7291, loss_cns_4: 0.5806, loss_yns_4: 0.1774, loss_cls_5: 1.2375, loss_box_5: 2.7436, loss_cns_5: 0.5749, loss_yns_5: 0.1811, loss_cls_dn_0: 0.4417, loss_box_dn_0: 0.9927, loss_cls_dn_1: 0.4203, loss_box_dn_1: 1.1776, loss_cls_dn_2: 0.4311, loss_box_dn_2: 1.1782, loss_cls_dn_3: 0.4255, loss_box_dn_3: 1.1901, loss_cls_dn_4: 0.4165, loss_box_dn_4: 1.2122, loss_cls_dn_5: 0.4253, loss_box_dn_5: 1.2547, loss_dense_depth: 1.0248, loss: 38.1732, grad_norm: 62.1459 -2025-11-12 14:31:20,433 - mmdet - INFO - Iter [28/17500] lr: 1.108e-04, eta: 1 day, 3:44:18, time: 1.589, data_time: 0.090, memory: 49164, loss_cls_0: 1.1298, loss_box_0: 2.1349, loss_cns_0: 0.6131, loss_yns_0: 0.1714, loss_cls_1: 1.2051, loss_box_1: 2.5381, loss_cns_1: 0.5803, loss_yns_1: 0.1724, loss_cls_2: 1.2160, loss_box_2: 2.5129, loss_cns_2: 0.5859, loss_yns_2: 0.1792, loss_cls_3: 1.2130, loss_box_3: 2.5358, loss_cns_3: 0.5848, loss_yns_3: 0.1770, loss_cls_4: 1.2067, loss_box_4: 2.5069, loss_cns_4: 0.5927, loss_yns_4: 0.1742, loss_cls_5: 1.2496, loss_box_5: 2.5478, loss_cns_5: 0.5932, loss_yns_5: 0.1754, loss_cls_dn_0: 0.4312, loss_box_dn_0: 0.9885, loss_cls_dn_1: 0.4066, loss_box_dn_1: 1.1752, loss_cls_dn_2: 0.4202, loss_box_dn_2: 1.1644, loss_cls_dn_3: 0.4097, loss_box_dn_3: 1.1935, loss_cls_dn_4: 0.4077, loss_box_dn_4: 1.2034, loss_cls_dn_5: 0.4029, loss_box_dn_5: 1.3157, loss_dense_depth: 1.0519, loss: 37.1672, grad_norm: 49.5942 -2025-11-12 14:31:22,016 - mmdet - INFO - Iter [29/17500] lr: 1.112e-04, eta: 1 day, 3:02:42, time: 1.581, data_time: 0.082, memory: 49164, loss_cls_0: 1.1023, loss_box_0: 2.1012, loss_cns_0: 0.6144, loss_yns_0: 0.1720, loss_cls_1: 1.1960, loss_box_1: 2.4345, loss_cns_1: 0.5907, loss_yns_1: 0.1725, loss_cls_2: 1.2023, loss_box_2: 2.4146, loss_cns_2: 0.6037, loss_yns_2: 0.1764, loss_cls_3: 1.2076, loss_box_3: 2.4433, loss_cns_3: 0.6032, loss_yns_3: 0.1772, loss_cls_4: 1.1960, loss_box_4: 2.4657, loss_cns_4: 0.6040, loss_yns_4: 0.1765, loss_cls_5: 1.2241, loss_box_5: 2.4968, loss_cns_5: 0.6087, loss_yns_5: 0.1693, loss_cls_dn_0: 0.4331, loss_box_dn_0: 0.9872, loss_cls_dn_1: 0.4083, loss_box_dn_1: 1.1129, loss_cls_dn_2: 0.4271, loss_box_dn_2: 1.1209, loss_cls_dn_3: 0.4103, loss_box_dn_3: 1.1948, loss_cls_dn_4: 0.4096, loss_box_dn_4: 1.2331, loss_cls_dn_5: 0.4089, loss_box_dn_5: 1.3773, loss_dense_depth: 0.9972, loss: 36.6735, grad_norm: 67.5593 -2025-11-12 14:31:23,595 - mmdet - INFO - Iter [30/17500] lr: 1.116e-04, eta: 1 day, 2:23:51, time: 1.581, data_time: 0.088, memory: 49164, loss_cls_0: 1.0773, loss_box_0: 2.0475, loss_cns_0: 0.6162, loss_yns_0: 0.1728, loss_cls_1: 1.1636, loss_box_1: 2.4398, loss_cns_1: 0.6009, loss_yns_1: 0.1726, loss_cls_2: 1.1784, loss_box_2: 2.4239, loss_cns_2: 0.6115, loss_yns_2: 0.1734, loss_cls_3: 1.1905, loss_box_3: 2.4213, loss_cns_3: 0.6112, loss_yns_3: 0.1750, loss_cls_4: 1.1761, loss_box_4: 2.4895, loss_cns_4: 0.6059, loss_yns_4: 0.1753, loss_cls_5: 1.1869, loss_box_5: 2.5157, loss_cns_5: 0.6054, loss_yns_5: 0.1709, loss_cls_dn_0: 0.4361, loss_box_dn_0: 0.9833, loss_cls_dn_1: 0.3971, loss_box_dn_1: 1.2233, loss_cls_dn_2: 0.4187, loss_box_dn_2: 1.2366, loss_cls_dn_3: 0.4007, loss_box_dn_3: 1.3064, loss_cls_dn_4: 0.4023, loss_box_dn_4: 1.3571, loss_cls_dn_5: 0.4053, loss_box_dn_5: 1.4804, loss_dense_depth: 1.0395, loss: 37.0884, grad_norm: 79.3057 -2025-11-12 14:31:25,181 - mmdet - INFO - Iter [31/17500] lr: 1.120e-04, eta: 1 day, 1:47:38, time: 1.592, data_time: 0.077, memory: 49164, loss_cls_0: 1.0662, loss_box_0: 2.0506, loss_cns_0: 0.6147, loss_yns_0: 0.1740, loss_cls_1: 1.1511, loss_box_1: 2.4285, loss_cns_1: 0.6092, loss_yns_1: 0.1748, loss_cls_2: 1.1681, loss_box_2: 2.3990, loss_cns_2: 0.6164, loss_yns_2: 0.1733, loss_cls_3: 1.1755, loss_box_3: 2.4133, loss_cns_3: 0.6124, loss_yns_3: 0.1732, loss_cls_4: 1.1636, loss_box_4: 2.4423, loss_cns_4: 0.6091, loss_yns_4: 0.1722, loss_cls_5: 1.1746, loss_box_5: 2.4837, loss_cns_5: 0.6033, loss_yns_5: 0.1710, loss_cls_dn_0: 0.4387, loss_box_dn_0: 0.9669, loss_cls_dn_1: 0.3932, loss_box_dn_1: 1.2726, loss_cls_dn_2: 0.4138, loss_box_dn_2: 1.2687, loss_cls_dn_3: 0.3965, loss_box_dn_3: 1.3113, loss_cls_dn_4: 0.3952, loss_box_dn_4: 1.3214, loss_cls_dn_5: 0.4092, loss_box_dn_5: 1.4157, loss_dense_depth: 0.9996, loss: 36.8230, grad_norm: 63.7240 -2025-11-12 14:31:26,755 - mmdet - INFO - Iter [32/17500] lr: 1.124e-04, eta: 1 day, 1:13:29, time: 1.573, data_time: 0.074, memory: 49164, loss_cls_0: 1.0500, loss_box_0: 2.0549, loss_cns_0: 0.6130, loss_yns_0: 0.1724, loss_cls_1: 1.1302, loss_box_1: 2.3605, loss_cns_1: 0.6069, loss_yns_1: 0.1771, loss_cls_2: 1.1538, loss_box_2: 2.3251, loss_cns_2: 0.6146, loss_yns_2: 0.1767, loss_cls_3: 1.1603, loss_box_3: 2.3545, loss_cns_3: 0.6148, loss_yns_3: 0.1715, loss_cls_4: 1.1508, loss_box_4: 2.4217, loss_cns_4: 0.6094, loss_yns_4: 0.1699, loss_cls_5: 1.1619, loss_box_5: 2.4589, loss_cns_5: 0.6061, loss_yns_5: 0.1733, loss_cls_dn_0: 0.4168, loss_box_dn_0: 0.9642, loss_cls_dn_1: 0.3817, loss_box_dn_1: 1.1420, loss_cls_dn_2: 0.3985, loss_box_dn_2: 1.1281, loss_cls_dn_3: 0.3845, loss_box_dn_3: 1.1547, loss_cls_dn_4: 0.3812, loss_box_dn_4: 1.1612, loss_cls_dn_5: 0.4071, loss_box_dn_5: 1.2270, loss_dense_depth: 0.9894, loss: 35.6251, grad_norm: 59.2249 -2025-11-12 14:31:28,325 - mmdet - INFO - Iter [33/17500] lr: 1.128e-04, eta: 1 day, 0:41:23, time: 1.570, data_time: 0.072, memory: 49164, loss_cls_0: 1.0844, loss_box_0: 2.0300, loss_cns_0: 0.6181, loss_yns_0: 0.1724, loss_cls_1: 1.1471, loss_box_1: 2.3464, loss_cns_1: 0.6074, loss_yns_1: 0.1807, loss_cls_2: 1.1579, loss_box_2: 2.3169, loss_cns_2: 0.6171, loss_yns_2: 0.1751, loss_cls_3: 1.1680, loss_box_3: 2.3281, loss_cns_3: 0.6216, loss_yns_3: 0.1729, loss_cls_4: 1.1511, loss_box_4: 2.3766, loss_cns_4: 0.6215, loss_yns_4: 0.1702, loss_cls_5: 1.1590, loss_box_5: 2.3955, loss_cns_5: 0.6211, loss_yns_5: 0.1750, loss_cls_dn_0: 0.4124, loss_box_dn_0: 0.9675, loss_cls_dn_1: 0.3495, loss_box_dn_1: 1.1355, loss_cls_dn_2: 0.3676, loss_box_dn_2: 1.1351, loss_cls_dn_3: 0.3550, loss_box_dn_3: 1.1553, loss_cls_dn_4: 0.3571, loss_box_dn_4: 1.1572, loss_cls_dn_5: 0.3760, loss_box_dn_5: 1.1914, loss_dense_depth: 1.0237, loss: 35.3971, grad_norm: 49.6332 -2025-11-12 14:31:29,890 - mmdet - INFO - Iter [34/17500] lr: 1.132e-04, eta: 1 day, 0:11:06, time: 1.561, data_time: 0.077, memory: 49164, loss_cls_0: 1.0692, loss_box_0: 1.9973, loss_cns_0: 0.6204, loss_yns_0: 0.1715, loss_cls_1: 1.1529, loss_box_1: 2.3931, loss_cns_1: 0.6039, loss_yns_1: 0.1729, loss_cls_2: 1.1571, loss_box_2: 2.4018, loss_cns_2: 0.6100, loss_yns_2: 0.1715, loss_cls_3: 1.1598, loss_box_3: 2.4293, loss_cns_3: 0.6176, loss_yns_3: 0.1711, loss_cls_4: 1.1491, loss_box_4: 2.4618, loss_cns_4: 0.6155, loss_yns_4: 0.1688, loss_cls_5: 1.1594, loss_box_5: 2.4716, loss_cns_5: 0.6181, loss_yns_5: 0.1729, loss_cls_dn_0: 0.4289, loss_box_dn_0: 0.9620, loss_cls_dn_1: 0.3317, loss_box_dn_1: 1.1534, loss_cls_dn_2: 0.3520, loss_box_dn_2: 1.1800, loss_cls_dn_3: 0.3527, loss_box_dn_3: 1.2094, loss_cls_dn_4: 0.3553, loss_box_dn_4: 1.2368, loss_cls_dn_5: 0.3638, loss_box_dn_5: 1.2399, loss_dense_depth: 1.0235, loss: 35.9062, grad_norm: 68.1854 -2025-11-12 14:31:31,474 - mmdet - INFO - Iter [35/17500] lr: 1.136e-04, eta: 23:42:42, time: 1.581, data_time: 0.081, memory: 49164, loss_cls_0: 1.0543, loss_box_0: 1.9561, loss_cns_0: 0.6232, loss_yns_0: 0.1714, loss_cls_1: 1.1672, loss_box_1: 2.3345, loss_cns_1: 0.6119, loss_yns_1: 0.1717, loss_cls_2: 1.1660, loss_box_2: 2.3664, loss_cns_2: 0.6117, loss_yns_2: 0.1731, loss_cls_3: 1.1517, loss_box_3: 2.3959, loss_cns_3: 0.6157, loss_yns_3: 0.1708, loss_cls_4: 1.1579, loss_box_4: 2.4723, loss_cns_4: 0.6161, loss_yns_4: 0.1709, loss_cls_5: 1.1643, loss_box_5: 2.4282, loss_cns_5: 0.6178, loss_yns_5: 0.1737, loss_cls_dn_0: 0.4342, loss_box_dn_0: 0.9450, loss_cls_dn_1: 0.3277, loss_box_dn_1: 1.1649, loss_cls_dn_2: 0.3475, loss_box_dn_2: 1.2115, loss_cls_dn_3: 0.3624, loss_box_dn_3: 1.2480, loss_cls_dn_4: 0.3630, loss_box_dn_4: 1.3281, loss_cls_dn_5: 0.3604, loss_box_dn_5: 1.3145, loss_dense_depth: 0.9842, loss: 35.9341, grad_norm: 87.5989 -2025-11-12 14:31:33,058 - mmdet - INFO - Iter [36/17500] lr: 1.140e-04, eta: 23:15:58, time: 1.590, data_time: 0.083, memory: 49164, loss_cls_0: 1.0418, loss_box_0: 1.9530, loss_cns_0: 0.6200, loss_yns_0: 0.1714, loss_cls_1: 1.1492, loss_box_1: 2.4014, loss_cns_1: 0.6044, loss_yns_1: 0.1735, loss_cls_2: 1.1536, loss_box_2: 2.4023, loss_cns_2: 0.6116, loss_yns_2: 0.1723, loss_cls_3: 1.1380, loss_box_3: 2.4028, loss_cns_3: 0.6121, loss_yns_3: 0.1692, loss_cls_4: 1.1300, loss_box_4: 2.4683, loss_cns_4: 0.6196, loss_yns_4: 0.1756, loss_cls_5: 1.1546, loss_box_5: 2.4734, loss_cns_5: 0.6122, loss_yns_5: 0.1757, loss_cls_dn_0: 0.4282, loss_box_dn_0: 0.9357, loss_cls_dn_1: 0.3552, loss_box_dn_1: 1.0840, loss_cls_dn_2: 0.3713, loss_box_dn_2: 1.1377, loss_cls_dn_3: 0.3928, loss_box_dn_3: 1.1703, loss_cls_dn_4: 0.3954, loss_box_dn_4: 1.3111, loss_cls_dn_5: 0.3842, loss_box_dn_5: 1.3154, loss_dense_depth: 0.9873, loss: 35.8545, grad_norm: 64.1945 -2025-11-12 14:31:34,618 - mmdet - INFO - Iter [37/17500] lr: 1.144e-04, eta: 22:50:27, time: 1.564, data_time: 0.078, memory: 49164, loss_cls_0: 1.0501, loss_box_0: 1.9578, loss_cns_0: 0.6207, loss_yns_0: 0.1719, loss_cls_1: 1.1208, loss_box_1: 2.5126, loss_cns_1: 0.5864, loss_yns_1: 0.1751, loss_cls_2: 1.1523, loss_box_2: 2.5058, loss_cns_2: 0.6004, loss_yns_2: 0.1749, loss_cls_3: 1.1435, loss_box_3: 2.5394, loss_cns_3: 0.5985, loss_yns_3: 0.1698, loss_cls_4: 1.1384, loss_box_4: 2.5607, loss_cns_4: 0.6133, loss_yns_4: 0.1796, loss_cls_5: 1.1724, loss_box_5: 2.5878, loss_cns_5: 0.6081, loss_yns_5: 0.1769, loss_cls_dn_0: 0.4158, loss_box_dn_0: 0.9346, loss_cls_dn_1: 0.3543, loss_box_dn_1: 1.2065, loss_cls_dn_2: 0.3643, loss_box_dn_2: 1.2279, loss_cls_dn_3: 0.3775, loss_box_dn_3: 1.2529, loss_cls_dn_4: 0.3809, loss_box_dn_4: 1.3592, loss_cls_dn_5: 0.3722, loss_box_dn_5: 1.3744, loss_dense_depth: 0.9627, loss: 36.7002, grad_norm: 76.1173 -2025-11-12 14:31:36,193 - mmdet - INFO - Iter [38/17500] lr: 1.148e-04, eta: 22:26:16, time: 1.562, data_time: 0.074, memory: 49164, loss_cls_0: 1.0570, loss_box_0: 1.9790, loss_cns_0: 0.6223, loss_yns_0: 0.1708, loss_cls_1: 1.0930, loss_box_1: 2.6041, loss_cns_1: 0.5770, loss_yns_1: 0.1720, loss_cls_2: 1.1387, loss_box_2: 2.5971, loss_cns_2: 0.5926, loss_yns_2: 0.1739, loss_cls_3: 1.1522, loss_box_3: 2.6252, loss_cns_3: 0.5963, loss_yns_3: 0.1700, loss_cls_4: 1.1638, loss_box_4: 2.6213, loss_cns_4: 0.6076, loss_yns_4: 0.1732, loss_cls_5: 1.1643, loss_box_5: 2.6523, loss_cns_5: 0.6093, loss_yns_5: 0.1744, loss_cls_dn_0: 0.3979, loss_box_dn_0: 0.9366, loss_cls_dn_1: 0.3611, loss_box_dn_1: 1.1882, loss_cls_dn_2: 0.3717, loss_box_dn_2: 1.1700, loss_cls_dn_3: 0.3763, loss_box_dn_3: 1.1792, loss_cls_dn_4: 0.3726, loss_box_dn_4: 1.2422, loss_cls_dn_5: 0.3791, loss_box_dn_5: 1.2579, loss_dense_depth: 1.0046, loss: 36.7247, grad_norm: 83.9319 -2025-11-12 14:31:37,838 - mmdet - INFO - Iter [39/17500] lr: 1.152e-04, eta: 22:03:54, time: 1.637, data_time: 0.084, memory: 49164, loss_cls_0: 1.0864, loss_box_0: 1.9627, loss_cns_0: 0.6221, loss_yns_0: 0.1686, loss_cls_1: 1.1060, loss_box_1: 2.4578, loss_cns_1: 0.5935, loss_yns_1: 0.1730, loss_cls_2: 1.1375, loss_box_2: 2.4679, loss_cns_2: 0.6037, loss_yns_2: 0.1715, loss_cls_3: 1.1506, loss_box_3: 2.4767, loss_cns_3: 0.6096, loss_yns_3: 0.1685, loss_cls_4: 1.1671, loss_box_4: 2.4597, loss_cns_4: 0.6118, loss_yns_4: 0.1709, loss_cls_5: 1.1443, loss_box_5: 2.4919, loss_cns_5: 0.6151, loss_yns_5: 0.1713, loss_cls_dn_0: 0.3941, loss_box_dn_0: 0.9217, loss_cls_dn_1: 0.3695, loss_box_dn_1: 1.0825, loss_cls_dn_2: 0.3833, loss_box_dn_2: 1.0532, loss_cls_dn_3: 0.3825, loss_box_dn_3: 1.0441, loss_cls_dn_4: 0.3662, loss_box_dn_4: 1.0740, loss_cls_dn_5: 0.3853, loss_box_dn_5: 1.0914, loss_dense_depth: 0.9523, loss: 35.2883, grad_norm: 68.2810 -2025-11-12 14:31:39,407 - mmdet - INFO - Iter [40/17500] lr: 1.156e-04, eta: 21:42:15, time: 1.585, data_time: 0.087, memory: 49164, loss_cls_0: 1.0597, loss_box_0: 1.9685, loss_cns_0: 0.6201, loss_yns_0: 0.1655, loss_cls_1: 1.0994, loss_box_1: 2.3271, loss_cns_1: 0.6025, loss_yns_1: 0.1718, loss_cls_2: 1.1381, loss_box_2: 2.3246, loss_cns_2: 0.6156, loss_yns_2: 0.1703, loss_cls_3: 1.1392, loss_box_3: 2.2987, loss_cns_3: 0.6198, loss_yns_3: 0.1687, loss_cls_4: 1.1252, loss_box_4: 2.2972, loss_cns_4: 0.6267, loss_yns_4: 0.1727, loss_cls_5: 1.1384, loss_box_5: 2.2913, loss_cns_5: 0.6337, loss_yns_5: 0.1669, loss_cls_dn_0: 0.4085, loss_box_dn_0: 0.9220, loss_cls_dn_1: 0.3746, loss_box_dn_1: 1.0402, loss_cls_dn_2: 0.3884, loss_box_dn_2: 1.0089, loss_cls_dn_3: 0.3829, loss_box_dn_3: 0.9870, loss_cls_dn_4: 0.3629, loss_box_dn_4: 0.9933, loss_cls_dn_5: 0.3866, loss_box_dn_5: 0.9946, loss_dense_depth: 1.0623, loss: 34.2536, grad_norm: 57.8070 -2025-11-12 14:31:41,043 - mmdet - INFO - Iter [41/17500] lr: 1.160e-04, eta: 21:22:02, time: 1.637, data_time: 0.101, memory: 49164, loss_cls_0: 1.0188, loss_box_0: 1.9484, loss_cns_0: 0.6220, loss_yns_0: 0.1631, loss_cls_1: 1.0835, loss_box_1: 2.2593, loss_cns_1: 0.6095, loss_yns_1: 0.1680, loss_cls_2: 1.1248, loss_box_2: 2.2395, loss_cns_2: 0.6295, loss_yns_2: 0.1679, loss_cls_3: 1.1390, loss_box_3: 2.2297, loss_cns_3: 0.6343, loss_yns_3: 0.1654, loss_cls_4: 1.1294, loss_box_4: 2.2351, loss_cns_4: 0.6521, loss_yns_4: 0.1678, loss_cls_5: 1.1545, loss_box_5: 2.2085, loss_cns_5: 0.6491, loss_yns_5: 0.1659, loss_cls_dn_0: 0.4226, loss_box_dn_0: 0.9255, loss_cls_dn_1: 0.3502, loss_box_dn_1: 1.0355, loss_cls_dn_2: 0.3635, loss_box_dn_2: 0.9835, loss_cls_dn_3: 0.3577, loss_box_dn_3: 0.9736, loss_cls_dn_4: 0.3540, loss_box_dn_4: 0.9751, loss_cls_dn_5: 0.3699, loss_box_dn_5: 0.9782, loss_dense_depth: 1.0973, loss: 33.7518, grad_norm: 58.6673 -2025-11-12 14:31:42,672 - mmdet - INFO - Iter [42/17500] lr: 1.164e-04, eta: 21:02:44, time: 1.630, data_time: 0.126, memory: 49164, loss_cls_0: 1.0122, loss_box_0: 1.9123, loss_cns_0: 0.6257, loss_yns_0: 0.1614, loss_cls_1: 1.0797, loss_box_1: 2.2087, loss_cns_1: 0.6145, loss_yns_1: 0.1662, loss_cls_2: 1.1229, loss_box_2: 2.1569, loss_cns_2: 0.6337, loss_yns_2: 0.1707, loss_cls_3: 1.1283, loss_box_3: 2.1846, loss_cns_3: 0.6388, loss_yns_3: 0.1692, loss_cls_4: 1.1193, loss_box_4: 2.1986, loss_cns_4: 0.6552, loss_yns_4: 0.1643, loss_cls_5: 1.1343, loss_box_5: 2.2184, loss_cns_5: 0.6368, loss_yns_5: 0.1674, loss_cls_dn_0: 0.4148, loss_box_dn_0: 0.9294, loss_cls_dn_1: 0.3473, loss_box_dn_1: 1.0420, loss_cls_dn_2: 0.3542, loss_box_dn_2: 0.9991, loss_cls_dn_3: 0.3540, loss_box_dn_3: 1.0269, loss_cls_dn_4: 0.3650, loss_box_dn_4: 1.0423, loss_cls_dn_5: 0.3769, loss_box_dn_5: 1.0944, loss_dense_depth: 0.9687, loss: 33.5950, grad_norm: 61.0783 -2025-11-12 14:31:44,287 - mmdet - INFO - Iter [43/17500] lr: 1.168e-04, eta: 20:44:14, time: 1.616, data_time: 0.084, memory: 49164, loss_cls_0: 1.0598, loss_box_0: 1.9437, loss_cns_0: 0.6234, loss_yns_0: 0.1618, loss_cls_1: 1.1243, loss_box_1: 2.2710, loss_cns_1: 0.6088, loss_yns_1: 0.1632, loss_cls_2: 1.1641, loss_box_2: 2.1940, loss_cns_2: 0.6253, loss_yns_2: 0.1666, loss_cls_3: 1.1634, loss_box_3: 2.2314, loss_cns_3: 0.6285, loss_yns_3: 0.1653, loss_cls_4: 1.1141, loss_box_4: 2.2294, loss_cns_4: 0.6448, loss_yns_4: 0.1645, loss_cls_5: 1.1353, loss_box_5: 2.2493, loss_cns_5: 0.6318, loss_yns_5: 0.1648, loss_cls_dn_0: 0.4139, loss_box_dn_0: 0.9241, loss_cls_dn_1: 0.3587, loss_box_dn_1: 1.0662, loss_cls_dn_2: 0.3600, loss_box_dn_2: 1.0489, loss_cls_dn_3: 0.3599, loss_box_dn_3: 1.1056, loss_cls_dn_4: 0.3812, loss_box_dn_4: 1.1287, loss_cls_dn_5: 0.3837, loss_box_dn_5: 1.2055, loss_dense_depth: 1.0747, loss: 34.4399, grad_norm: 62.0827 -2025-11-12 14:31:45,894 - mmdet - INFO - Iter [44/17500] lr: 1.172e-04, eta: 20:26:29, time: 1.603, data_time: 0.076, memory: 49164, loss_cls_0: 1.0738, loss_box_0: 2.0043, loss_cns_0: 0.6191, loss_yns_0: 0.1628, loss_cls_1: 1.1291, loss_box_1: 2.2745, loss_cns_1: 0.6050, loss_yns_1: 0.1614, loss_cls_2: 1.1679, loss_box_2: 2.2102, loss_cns_2: 0.6154, loss_yns_2: 0.1656, loss_cls_3: 1.1665, loss_box_3: 2.2411, loss_cns_3: 0.6153, loss_yns_3: 0.1634, loss_cls_4: 1.1255, loss_box_4: 2.2258, loss_cns_4: 0.6207, loss_yns_4: 0.1626, loss_cls_5: 1.1689, loss_box_5: 2.2376, loss_cns_5: 0.6190, loss_yns_5: 0.1612, loss_cls_dn_0: 0.4242, loss_box_dn_0: 0.9195, loss_cls_dn_1: 0.3687, loss_box_dn_1: 1.1374, loss_cls_dn_2: 0.3737, loss_box_dn_2: 1.1265, loss_cls_dn_3: 0.3740, loss_box_dn_3: 1.1794, loss_cls_dn_4: 0.3901, loss_box_dn_4: 1.1966, loss_cls_dn_5: 0.3918, loss_box_dn_5: 1.2686, loss_dense_depth: 1.0289, loss: 34.8758, grad_norm: 56.7187 -2025-11-12 14:31:47,505 - mmdet - INFO - Iter [45/17500] lr: 1.176e-04, eta: 20:09:35, time: 1.612, data_time: 0.089, memory: 49164, loss_cls_0: 1.0569, loss_box_0: 2.0143, loss_cns_0: 0.6120, loss_yns_0: 0.1639, loss_cls_1: 1.1151, loss_box_1: 2.3159, loss_cns_1: 0.6012, loss_yns_1: 0.1641, loss_cls_2: 1.1514, loss_box_2: 2.2522, loss_cns_2: 0.6162, loss_yns_2: 0.1674, loss_cls_3: 1.1419, loss_box_3: 2.2674, loss_cns_3: 0.6222, loss_yns_3: 0.1646, loss_cls_4: 1.1572, loss_box_4: 2.2751, loss_cns_4: 0.6203, loss_yns_4: 0.1650, loss_cls_5: 1.1794, loss_box_5: 2.2702, loss_cns_5: 0.6213, loss_yns_5: 0.1639, loss_cls_dn_0: 0.4256, loss_box_dn_0: 0.9249, loss_cls_dn_1: 0.3784, loss_box_dn_1: 1.1393, loss_cls_dn_2: 0.3852, loss_box_dn_2: 1.1252, loss_cls_dn_3: 0.3940, loss_box_dn_3: 1.1648, loss_cls_dn_4: 0.3888, loss_box_dn_4: 1.1818, loss_cls_dn_5: 0.4014, loss_box_dn_5: 1.2422, loss_dense_depth: 1.0519, loss: 35.0830, grad_norm: 56.5478 -2025-11-12 14:31:49,115 - mmdet - INFO - Iter [46/17500] lr: 1.180e-04, eta: 19:53:24, time: 1.609, data_time: 0.101, memory: 49164, loss_cls_0: 1.0433, loss_box_0: 1.9925, loss_cns_0: 0.6132, loss_yns_0: 0.1618, loss_cls_1: 1.0881, loss_box_1: 2.3423, loss_cns_1: 0.5982, loss_yns_1: 0.1664, loss_cls_2: 1.1078, loss_box_2: 2.2770, loss_cns_2: 0.6185, loss_yns_2: 0.1677, loss_cls_3: 1.1125, loss_box_3: 2.2718, loss_cns_3: 0.6275, loss_yns_3: 0.1653, loss_cls_4: 1.1180, loss_box_4: 2.2719, loss_cns_4: 0.6242, loss_yns_4: 0.1680, loss_cls_5: 1.1210, loss_box_5: 2.2817, loss_cns_5: 0.6272, loss_yns_5: 0.1648, loss_cls_dn_0: 0.4078, loss_box_dn_0: 0.9176, loss_cls_dn_1: 0.3681, loss_box_dn_1: 1.0939, loss_cls_dn_2: 0.3866, loss_box_dn_2: 1.0616, loss_cls_dn_3: 0.3959, loss_box_dn_3: 1.0726, loss_cls_dn_4: 0.3673, loss_box_dn_4: 1.0708, loss_cls_dn_5: 0.3897, loss_box_dn_5: 1.1143, loss_dense_depth: 1.0135, loss: 34.3905, grad_norm: 52.5360 -2025-11-12 14:31:50,704 - mmdet - INFO - Iter [47/17500] lr: 1.184e-04, eta: 19:37:45, time: 1.586, data_time: 0.081, memory: 49164, loss_cls_0: 1.0766, loss_box_0: 1.9654, loss_cns_0: 0.6221, loss_yns_0: 0.1616, loss_cls_1: 1.1348, loss_box_1: 2.3596, loss_cns_1: 0.6033, loss_yns_1: 0.1637, loss_cls_2: 1.1211, loss_box_2: 2.2898, loss_cns_2: 0.6253, loss_yns_2: 0.1669, loss_cls_3: 1.1391, loss_box_3: 2.2605, loss_cns_3: 0.6332, loss_yns_3: 0.1641, loss_cls_4: 1.1373, loss_box_4: 2.2659, loss_cns_4: 0.6331, loss_yns_4: 0.1675, loss_cls_5: 1.1412, loss_box_5: 2.2738, loss_cns_5: 0.6378, loss_yns_5: 0.1640, loss_cls_dn_0: 0.3967, loss_box_dn_0: 0.9100, loss_cls_dn_1: 0.3571, loss_box_dn_1: 1.0155, loss_cls_dn_2: 0.3867, loss_box_dn_2: 0.9835, loss_cls_dn_3: 0.3941, loss_box_dn_3: 0.9721, loss_cls_dn_4: 0.3561, loss_box_dn_4: 0.9706, loss_cls_dn_5: 0.3853, loss_box_dn_5: 0.9892, loss_dense_depth: 0.9453, loss: 33.9697, grad_norm: 44.9307 -2025-11-12 14:31:52,291 - mmdet - INFO - Iter [48/17500] lr: 1.188e-04, eta: 19:22:49, time: 1.595, data_time: 0.088, memory: 49164, loss_cls_0: 1.0412, loss_box_0: 1.9507, loss_cns_0: 0.6195, loss_yns_0: 0.1636, loss_cls_1: 1.1318, loss_box_1: 2.3283, loss_cns_1: 0.5997, loss_yns_1: 0.1637, loss_cls_2: 1.1249, loss_box_2: 2.2928, loss_cns_2: 0.6236, loss_yns_2: 0.1668, loss_cls_3: 1.1453, loss_box_3: 2.2562, loss_cns_3: 0.6299, loss_yns_3: 0.1699, loss_cls_4: 1.1526, loss_box_4: 2.2730, loss_cns_4: 0.6331, loss_yns_4: 0.1690, loss_cls_5: 1.1589, loss_box_5: 2.2567, loss_cns_5: 0.6364, loss_yns_5: 0.1669, loss_cls_dn_0: 0.3932, loss_box_dn_0: 0.9043, loss_cls_dn_1: 0.3277, loss_box_dn_1: 1.0359, loss_cls_dn_2: 0.3733, loss_box_dn_2: 0.9997, loss_cls_dn_3: 0.3683, loss_box_dn_3: 0.9799, loss_cls_dn_4: 0.3494, loss_box_dn_4: 0.9861, loss_cls_dn_5: 0.3736, loss_box_dn_5: 0.9813, loss_dense_depth: 1.0819, loss: 34.0090, grad_norm: 47.5284 -2025-11-12 14:31:53,899 - mmdet - INFO - Iter [49/17500] lr: 1.192e-04, eta: 19:08:33, time: 1.606, data_time: 0.083, memory: 49164, loss_cls_0: 1.0397, loss_box_0: 1.9554, loss_cns_0: 0.6161, loss_yns_0: 0.1645, loss_cls_1: 1.1152, loss_box_1: 2.2965, loss_cns_1: 0.6093, loss_yns_1: 0.1655, loss_cls_2: 1.1183, loss_box_2: 2.2785, loss_cns_2: 0.6269, loss_yns_2: 0.1669, loss_cls_3: 1.1358, loss_box_3: 2.2418, loss_cns_3: 0.6327, loss_yns_3: 0.1682, loss_cls_4: 1.1217, loss_box_4: 2.2746, loss_cns_4: 0.6360, loss_yns_4: 0.1705, loss_cls_5: 1.1332, loss_box_5: 2.2685, loss_cns_5: 0.6341, loss_yns_5: 0.1672, loss_cls_dn_0: 0.4016, loss_box_dn_0: 0.8973, loss_cls_dn_1: 0.3063, loss_box_dn_1: 1.0204, loss_cls_dn_2: 0.3500, loss_box_dn_2: 0.9901, loss_cls_dn_3: 0.3417, loss_box_dn_3: 0.9741, loss_cls_dn_4: 0.3524, loss_box_dn_4: 0.9999, loss_cls_dn_5: 0.3674, loss_box_dn_5: 1.0009, loss_dense_depth: 0.9791, loss: 33.7180, grad_norm: 43.6545 -2025-11-12 14:31:55,475 - mmdet - INFO - Iter [50/17500] lr: 1.196e-04, eta: 18:54:40, time: 1.574, data_time: 0.082, memory: 49164, loss_cls_0: 1.0227, loss_box_0: 1.9411, loss_cns_0: 0.6176, loss_yns_0: 0.1621, loss_cls_1: 1.1166, loss_box_1: 2.3079, loss_cns_1: 0.6203, loss_yns_1: 0.1644, loss_cls_2: 1.1154, loss_box_2: 2.2530, loss_cns_2: 0.6297, loss_yns_2: 0.1646, loss_cls_3: 1.1413, loss_box_3: 2.2848, loss_cns_3: 0.6321, loss_yns_3: 0.1654, loss_cls_4: 1.1069, loss_box_4: 2.2966, loss_cns_4: 0.6347, loss_yns_4: 0.1741, loss_cls_5: 1.1102, loss_box_5: 2.3084, loss_cns_5: 0.6303, loss_yns_5: 0.1661, loss_cls_dn_0: 0.3998, loss_box_dn_0: 0.9016, loss_cls_dn_1: 0.3127, loss_box_dn_1: 0.9985, loss_cls_dn_2: 0.3409, loss_box_dn_2: 0.9779, loss_cls_dn_3: 0.3388, loss_box_dn_3: 1.0060, loss_cls_dn_4: 0.3662, loss_box_dn_4: 1.0556, loss_cls_dn_5: 0.3768, loss_box_dn_5: 1.0799, loss_dense_depth: 1.0851, loss: 34.0059, grad_norm: 49.6365 -2025-11-12 14:31:57,059 - mmdet - INFO - Iter [51/17500] lr: 1.200e-04, eta: 18:41:22, time: 1.579, data_time: 0.077, memory: 49164, loss_cls_0: 1.0104, loss_box_0: 1.9222, loss_cns_0: 0.6194, loss_yns_0: 0.1617, loss_cls_1: 1.0980, loss_box_1: 2.4475, loss_cns_1: 0.6164, loss_yns_1: 0.1625, loss_cls_2: 1.1191, loss_box_2: 2.3999, loss_cns_2: 0.6255, loss_yns_2: 0.1647, loss_cls_3: 1.1527, loss_box_3: 2.4246, loss_cns_3: 0.6303, loss_yns_3: 0.1651, loss_cls_4: 1.1212, loss_box_4: 2.4284, loss_cns_4: 0.6310, loss_yns_4: 0.1739, loss_cls_5: 1.1256, loss_box_5: 2.4250, loss_cns_5: 0.6339, loss_yns_5: 0.1655, loss_cls_dn_0: 0.4086, loss_box_dn_0: 0.8882, loss_cls_dn_1: 0.3141, loss_box_dn_1: 1.0651, loss_cls_dn_2: 0.3371, loss_box_dn_2: 1.0446, loss_cls_dn_3: 0.3326, loss_box_dn_3: 1.0773, loss_cls_dn_4: 0.3460, loss_box_dn_4: 1.1257, loss_cls_dn_5: 0.3587, loss_box_dn_5: 1.1466, loss_dense_depth: 0.9208, loss: 34.7901, grad_norm: 53.7031 -2025-11-12 14:31:58,629 - mmdet - INFO - Iter [52/17500] lr: 1.204e-04, eta: 18:28:32, time: 1.571, data_time: 0.082, memory: 49164, loss_cls_0: 1.0133, loss_box_0: 1.9455, loss_cns_0: 0.6155, loss_yns_0: 0.1620, loss_cls_1: 1.1071, loss_box_1: 2.5120, loss_cns_1: 0.6101, loss_yns_1: 0.1625, loss_cls_2: 1.1169, loss_box_2: 2.4809, loss_cns_2: 0.6202, loss_yns_2: 0.1629, loss_cls_3: 1.1376, loss_box_3: 2.4596, loss_cns_3: 0.6244, loss_yns_3: 0.1644, loss_cls_4: 1.1558, loss_box_4: 2.4667, loss_cns_4: 0.6222, loss_yns_4: 0.1667, loss_cls_5: 1.1635, loss_box_5: 2.4691, loss_cns_5: 0.6257, loss_yns_5: 0.1623, loss_cls_dn_0: 0.4178, loss_box_dn_0: 0.8866, loss_cls_dn_1: 0.3279, loss_box_dn_1: 1.0783, loss_cls_dn_2: 0.3469, loss_box_dn_2: 1.0650, loss_cls_dn_3: 0.3476, loss_box_dn_3: 1.0785, loss_cls_dn_4: 0.3340, loss_box_dn_4: 1.1227, loss_cls_dn_5: 0.3512, loss_box_dn_5: 1.1390, loss_dense_depth: 1.0268, loss: 35.2490, grad_norm: 52.4064 -2025-11-12 14:32:00,191 - mmdet - INFO - Iter [53/17500] lr: 1.208e-04, eta: 18:16:08, time: 1.566, data_time: 0.080, memory: 49164, loss_cls_0: 1.0206, loss_box_0: 1.9730, loss_cns_0: 0.6173, loss_yns_0: 0.1634, loss_cls_1: 1.1003, loss_box_1: 2.3756, loss_cns_1: 0.6127, loss_yns_1: 0.1632, loss_cls_2: 1.1121, loss_box_2: 2.3598, loss_cns_2: 0.6231, loss_yns_2: 0.1653, loss_cls_3: 1.1135, loss_box_3: 2.3609, loss_cns_3: 0.6254, loss_yns_3: 0.1658, loss_cls_4: 1.1398, loss_box_4: 2.3800, loss_cns_4: 0.6223, loss_yns_4: 0.1632, loss_cls_5: 1.1568, loss_box_5: 2.3686, loss_cns_5: 0.6277, loss_yns_5: 0.1629, loss_cls_dn_0: 0.4165, loss_box_dn_0: 0.8781, loss_cls_dn_1: 0.3279, loss_box_dn_1: 0.9831, loss_cls_dn_2: 0.3543, loss_box_dn_2: 0.9656, loss_cls_dn_3: 0.3620, loss_box_dn_3: 0.9697, loss_cls_dn_4: 0.3294, loss_box_dn_4: 0.9980, loss_cls_dn_5: 0.3470, loss_box_dn_5: 1.0028, loss_dense_depth: 0.9758, loss: 34.0834, grad_norm: 37.8082 -2025-11-12 14:32:01,759 - mmdet - INFO - Iter [54/17500] lr: 1.212e-04, eta: 18:04:13, time: 1.567, data_time: 0.074, memory: 49164, loss_cls_0: 1.0234, loss_box_0: 1.9043, loss_cns_0: 0.6265, loss_yns_0: 0.1613, loss_cls_1: 1.0737, loss_box_1: 2.2532, loss_cns_1: 0.6178, loss_yns_1: 0.1634, loss_cls_2: 1.0974, loss_box_2: 2.2187, loss_cns_2: 0.6315, loss_yns_2: 0.1629, loss_cls_3: 1.0995, loss_box_3: 2.2382, loss_cns_3: 0.6388, loss_yns_3: 0.1615, loss_cls_4: 1.1126, loss_box_4: 2.2167, loss_cns_4: 0.6418, loss_yns_4: 0.1614, loss_cls_5: 1.1115, loss_box_5: 2.2098, loss_cns_5: 0.6429, loss_yns_5: 0.1612, loss_cls_dn_0: 0.3821, loss_box_dn_0: 0.8891, loss_cls_dn_1: 0.2988, loss_box_dn_1: 0.9758, loss_cls_dn_2: 0.3360, loss_box_dn_2: 0.9492, loss_cls_dn_3: 0.3544, loss_box_dn_3: 0.9585, loss_cls_dn_4: 0.3304, loss_box_dn_4: 0.9587, loss_cls_dn_5: 0.3424, loss_box_dn_5: 0.9597, loss_dense_depth: 0.9918, loss: 33.0569, grad_norm: 40.2988 -2025-11-12 14:32:03,339 - mmdet - INFO - Iter [55/17500] lr: 1.216e-04, eta: 17:52:48, time: 1.582, data_time: 0.079, memory: 49164, loss_cls_0: 1.0503, loss_box_0: 1.9078, loss_cns_0: 0.6213, loss_yns_0: 0.1633, loss_cls_1: 1.0978, loss_box_1: 2.2245, loss_cns_1: 0.6138, loss_yns_1: 0.1646, loss_cls_2: 1.1139, loss_box_2: 2.1667, loss_cns_2: 0.6309, loss_yns_2: 0.1662, loss_cls_3: 1.1075, loss_box_3: 2.1821, loss_cns_3: 0.6350, loss_yns_3: 0.1658, loss_cls_4: 1.1165, loss_box_4: 2.1543, loss_cns_4: 0.6386, loss_yns_4: 0.1711, loss_cls_5: 1.1169, loss_box_5: 2.1722, loss_cns_5: 0.6388, loss_yns_5: 0.1648, loss_cls_dn_0: 0.3714, loss_box_dn_0: 0.8793, loss_cls_dn_1: 0.2975, loss_box_dn_1: 0.8647, loss_cls_dn_2: 0.3434, loss_box_dn_2: 0.8358, loss_cls_dn_3: 0.3709, loss_box_dn_3: 0.8530, loss_cls_dn_4: 0.3724, loss_box_dn_4: 0.8465, loss_cls_dn_5: 0.3892, loss_box_dn_5: 0.8762, loss_dense_depth: 0.9682, loss: 32.4535, grad_norm: 43.0150 -2025-11-12 14:32:04,917 - mmdet - INFO - Iter [56/17500] lr: 1.220e-04, eta: 17:41:45, time: 1.571, data_time: 0.074, memory: 49164, loss_cls_0: 0.9987, loss_box_0: 1.8968, loss_cns_0: 0.6145, loss_yns_0: 0.1619, loss_cls_1: 1.1093, loss_box_1: 2.1004, loss_cns_1: 0.6179, loss_yns_1: 0.1625, loss_cls_2: 1.0982, loss_box_2: 2.0430, loss_cns_2: 0.6313, loss_yns_2: 0.1637, loss_cls_3: 1.0882, loss_box_3: 2.0470, loss_cns_3: 0.6344, loss_yns_3: 0.1657, loss_cls_4: 1.0839, loss_box_4: 2.0296, loss_cns_4: 0.6376, loss_yns_4: 0.1716, loss_cls_5: 1.0960, loss_box_5: 2.0386, loss_cns_5: 0.6367, loss_yns_5: 0.1633, loss_cls_dn_0: 0.3664, loss_box_dn_0: 0.8803, loss_cls_dn_1: 0.2862, loss_box_dn_1: 0.8831, loss_cls_dn_2: 0.3283, loss_box_dn_2: 0.8742, loss_cls_dn_3: 0.3447, loss_box_dn_3: 0.8965, loss_cls_dn_4: 0.3693, loss_box_dn_4: 0.9012, loss_cls_dn_5: 0.3917, loss_box_dn_5: 0.9478, loss_dense_depth: 0.9437, loss: 31.8043, grad_norm: 44.9614 -2025-11-12 14:32:06,486 - mmdet - INFO - Iter [57/17500] lr: 1.224e-04, eta: 17:31:05, time: 1.573, data_time: 0.083, memory: 49164, loss_cls_0: 0.9760, loss_box_0: 1.8706, loss_cns_0: 0.6178, loss_yns_0: 0.1601, loss_cls_1: 1.0571, loss_box_1: 2.0783, loss_cns_1: 0.6264, loss_yns_1: 0.1620, loss_cls_2: 1.1006, loss_box_2: 2.0471, loss_cns_2: 0.6376, loss_yns_2: 0.1631, loss_cls_3: 1.1273, loss_box_3: 2.0564, loss_cns_3: 0.6391, loss_yns_3: 0.1666, loss_cls_4: 1.0824, loss_box_4: 2.0711, loss_cns_4: 0.6382, loss_yns_4: 0.1661, loss_cls_5: 1.0836, loss_box_5: 2.0861, loss_cns_5: 0.6364, loss_yns_5: 0.1643, loss_cls_dn_0: 0.3818, loss_box_dn_0: 0.8690, loss_cls_dn_1: 0.2936, loss_box_dn_1: 0.9044, loss_cls_dn_2: 0.3245, loss_box_dn_2: 0.9097, loss_cls_dn_3: 0.3223, loss_box_dn_3: 0.9391, loss_cls_dn_4: 0.3402, loss_box_dn_4: 0.9522, loss_cls_dn_5: 0.3751, loss_box_dn_5: 1.0121, loss_dense_depth: 0.9771, loss: 32.0153, grad_norm: 61.7986 -2025-11-12 14:32:08,066 - mmdet - INFO - Iter [58/17500] lr: 1.228e-04, eta: 17:20:47, time: 1.573, data_time: 0.079, memory: 49164, loss_cls_0: 0.9698, loss_box_0: 1.8742, loss_cns_0: 0.6157, loss_yns_0: 0.1616, loss_cls_1: 1.0552, loss_box_1: 2.0881, loss_cns_1: 0.6296, loss_yns_1: 0.1643, loss_cls_2: 1.0857, loss_box_2: 2.0620, loss_cns_2: 0.6391, loss_yns_2: 0.1686, loss_cls_3: 1.1243, loss_box_3: 2.0701, loss_cns_3: 0.6419, loss_yns_3: 0.1663, loss_cls_4: 1.1252, loss_box_4: 2.0843, loss_cns_4: 0.6349, loss_yns_4: 0.1656, loss_cls_5: 1.0853, loss_box_5: 2.1143, loss_cns_5: 0.6391, loss_yns_5: 0.1658, loss_cls_dn_0: 0.3932, loss_box_dn_0: 0.8564, loss_cls_dn_1: 0.3093, loss_box_dn_1: 0.9365, loss_cls_dn_2: 0.3323, loss_box_dn_2: 0.9402, loss_cls_dn_3: 0.3164, loss_box_dn_3: 0.9697, loss_cls_dn_4: 0.3111, loss_box_dn_4: 0.9741, loss_cls_dn_5: 0.3510, loss_box_dn_5: 1.0350, loss_dense_depth: 0.9026, loss: 32.1590, grad_norm: 62.8748 -2025-11-12 14:32:09,625 - mmdet - INFO - Iter [59/17500] lr: 1.232e-04, eta: 17:10:49, time: 1.570, data_time: 0.082, memory: 49164, loss_cls_0: 0.9681, loss_box_0: 1.8917, loss_cns_0: 0.6138, loss_yns_0: 0.1615, loss_cls_1: 1.0753, loss_box_1: 2.1744, loss_cns_1: 0.6256, loss_yns_1: 0.1641, loss_cls_2: 1.0697, loss_box_2: 2.1210, loss_cns_2: 0.6386, loss_yns_2: 0.1696, loss_cls_3: 1.0750, loss_box_3: 2.1437, loss_cns_3: 0.6423, loss_yns_3: 0.1655, loss_cls_4: 1.1106, loss_box_4: 2.1003, loss_cns_4: 0.6438, loss_yns_4: 0.1645, loss_cls_5: 1.1091, loss_box_5: 2.1082, loss_cns_5: 0.6493, loss_yns_5: 0.1653, loss_cls_dn_0: 0.3929, loss_box_dn_0: 0.8444, loss_cls_dn_1: 0.3198, loss_box_dn_1: 0.9350, loss_cls_dn_2: 0.3391, loss_box_dn_2: 0.9214, loss_cls_dn_3: 0.3275, loss_box_dn_3: 0.9479, loss_cls_dn_4: 0.3038, loss_box_dn_4: 0.9297, loss_cls_dn_5: 0.3303, loss_box_dn_5: 0.9653, loss_dense_depth: 0.9092, loss: 32.2170, grad_norm: 45.4584 -2025-11-12 14:32:11,200 - mmdet - INFO - Iter [60/17500] lr: 1.236e-04, eta: 17:01:11, time: 1.570, data_time: 0.071, memory: 49164, loss_cls_0: 0.9801, loss_box_0: 1.8756, loss_cns_0: 0.6203, loss_yns_0: 0.1602, loss_cls_1: 1.0593, loss_box_1: 2.1058, loss_cns_1: 0.6306, loss_yns_1: 0.1613, loss_cls_2: 1.0765, loss_box_2: 2.0550, loss_cns_2: 0.6431, loss_yns_2: 0.1644, loss_cls_3: 1.0918, loss_box_3: 2.0910, loss_cns_3: 0.6449, loss_yns_3: 0.1628, loss_cls_4: 1.0817, loss_box_4: 2.0776, loss_cns_4: 0.6448, loss_yns_4: 0.1653, loss_cls_5: 1.1201, loss_box_5: 2.0838, loss_cns_5: 0.6469, loss_yns_5: 0.1639, loss_cls_dn_0: 0.3801, loss_box_dn_0: 0.8564, loss_cls_dn_1: 0.2973, loss_box_dn_1: 0.9315, loss_cls_dn_2: 0.3240, loss_box_dn_2: 0.9112, loss_cls_dn_3: 0.3416, loss_box_dn_3: 0.9317, loss_cls_dn_4: 0.3241, loss_box_dn_4: 0.9222, loss_cls_dn_5: 0.3246, loss_box_dn_5: 0.9389, loss_dense_depth: 0.8991, loss: 31.8892, grad_norm: 57.5588 -2025-11-12 14:32:12,819 - mmdet - INFO - Iter [61/17500] lr: 1.240e-04, eta: 16:52:08, time: 1.626, data_time: 0.105, memory: 49164, loss_cls_0: 0.9950, loss_box_0: 1.8907, loss_cns_0: 0.6164, loss_yns_0: 0.1585, loss_cls_1: 1.0469, loss_box_1: 2.1614, loss_cns_1: 0.6210, loss_yns_1: 0.1603, loss_cls_2: 1.1108, loss_box_2: 2.0878, loss_cns_2: 0.6391, loss_yns_2: 0.1612, loss_cls_3: 1.0992, loss_box_3: 2.0801, loss_cns_3: 0.6451, loss_yns_3: 0.1625, loss_cls_4: 1.0882, loss_box_4: 2.0919, loss_cns_4: 0.6431, loss_yns_4: 0.1661, loss_cls_5: 1.1061, loss_box_5: 2.0731, loss_cns_5: 0.6451, loss_yns_5: 0.1610, loss_cls_dn_0: 0.3707, loss_box_dn_0: 0.8555, loss_cls_dn_1: 0.2680, loss_box_dn_1: 0.9558, loss_cls_dn_2: 0.3008, loss_box_dn_2: 0.9073, loss_cls_dn_3: 0.3391, loss_box_dn_3: 0.8978, loss_cls_dn_4: 0.3396, loss_box_dn_4: 0.8991, loss_cls_dn_5: 0.3249, loss_box_dn_5: 0.8950, loss_dense_depth: 0.9303, loss: 31.8946, grad_norm: 56.0391 -2025-11-12 14:32:14,440 - mmdet - INFO - Iter [62/17500] lr: 1.244e-04, eta: 16:43:19, time: 1.614, data_time: 0.141, memory: 49164, loss_cls_0: 0.9948, loss_box_0: 1.8943, loss_cns_0: 0.6134, loss_yns_0: 0.1587, loss_cls_1: 1.1291, loss_box_1: 2.1927, loss_cns_1: 0.6223, loss_yns_1: 0.1620, loss_cls_2: 1.1381, loss_box_2: 2.1375, loss_cns_2: 0.6374, loss_yns_2: 0.1647, loss_cls_3: 1.0824, loss_box_3: 2.1118, loss_cns_3: 0.6423, loss_yns_3: 0.1631, loss_cls_4: 1.0804, loss_box_4: 2.1047, loss_cns_4: 0.6445, loss_yns_4: 0.1686, loss_cls_5: 1.0921, loss_box_5: 2.0955, loss_cns_5: 0.6432, loss_yns_5: 0.1639, loss_cls_dn_0: 0.3693, loss_box_dn_0: 0.8592, loss_cls_dn_1: 0.2570, loss_box_dn_1: 0.8876, loss_cls_dn_2: 0.2954, loss_box_dn_2: 0.8467, loss_cls_dn_3: 0.3461, loss_box_dn_3: 0.8398, loss_cls_dn_4: 0.3506, loss_box_dn_4: 0.8557, loss_cls_dn_5: 0.3527, loss_box_dn_5: 0.8692, loss_dense_depth: 0.8937, loss: 31.8605, grad_norm: 42.9728 -2025-11-12 14:32:16,038 - mmdet - INFO - Iter [63/17500] lr: 1.248e-04, eta: 16:34:41, time: 1.593, data_time: 0.080, memory: 49164, loss_cls_0: 0.9891, loss_box_0: 1.8708, loss_cns_0: 0.6168, loss_yns_0: 0.1614, loss_cls_1: 1.0872, loss_box_1: 2.1971, loss_cns_1: 0.6245, loss_yns_1: 0.1653, loss_cls_2: 1.0794, loss_box_2: 2.1425, loss_cns_2: 0.6420, loss_yns_2: 0.1674, loss_cls_3: 1.0955, loss_box_3: 2.1488, loss_cns_3: 0.6450, loss_yns_3: 0.1635, loss_cls_4: 1.0849, loss_box_4: 2.1658, loss_cns_4: 0.6466, loss_yns_4: 0.1639, loss_cls_5: 1.0975, loss_box_5: 2.1862, loss_cns_5: 0.6436, loss_yns_5: 0.1648, loss_cls_dn_0: 0.3731, loss_box_dn_0: 0.8515, loss_cls_dn_1: 0.2524, loss_box_dn_1: 0.8874, loss_cls_dn_2: 0.3072, loss_box_dn_2: 0.8693, loss_cls_dn_3: 0.3292, loss_box_dn_3: 0.8859, loss_cls_dn_4: 0.3205, loss_box_dn_4: 0.9240, loss_cls_dn_5: 0.3571, loss_box_dn_5: 0.9595, loss_dense_depth: 0.9307, loss: 32.1973, grad_norm: 58.3821 -2025-11-12 14:32:17,647 - mmdet - INFO - Iter [64/17500] lr: 1.252e-04, eta: 16:26:23, time: 1.610, data_time: 0.082, memory: 49164, loss_cls_0: 0.9869, loss_box_0: 1.8623, loss_cns_0: 0.6209, loss_yns_0: 0.1627, loss_cls_1: 1.0549, loss_box_1: 2.3005, loss_cns_1: 0.6241, loss_yns_1: 0.1678, loss_cls_2: 1.0787, loss_box_2: 2.2370, loss_cns_2: 0.6381, loss_yns_2: 0.1672, loss_cls_3: 1.1047, loss_box_3: 2.2233, loss_cns_3: 0.6416, loss_yns_3: 0.1631, loss_cls_4: 1.1222, loss_box_4: 2.2493, loss_cns_4: 0.6417, loss_yns_4: 0.1643, loss_cls_5: 1.0859, loss_box_5: 2.2365, loss_cns_5: 0.6406, loss_yns_5: 0.1640, loss_cls_dn_0: 0.3668, loss_box_dn_0: 0.8529, loss_cls_dn_1: 0.2513, loss_box_dn_1: 0.9525, loss_cls_dn_2: 0.3128, loss_box_dn_2: 0.9429, loss_cls_dn_3: 0.3010, loss_box_dn_3: 0.9560, loss_cls_dn_4: 0.2786, loss_box_dn_4: 1.0028, loss_cls_dn_5: 0.3398, loss_box_dn_5: 1.0273, loss_dense_depth: 0.8999, loss: 32.8229, grad_norm: 52.1381 -2025-11-12 14:32:19,253 - mmdet - INFO - Iter [65/17500] lr: 1.256e-04, eta: 16:18:22, time: 1.611, data_time: 0.085, memory: 49164, loss_cls_0: 0.9774, loss_box_0: 1.8626, loss_cns_0: 0.6197, loss_yns_0: 0.1628, loss_cls_1: 1.0743, loss_box_1: 2.1792, loss_cns_1: 0.6301, loss_yns_1: 0.1658, loss_cls_2: 1.0773, loss_box_2: 2.1264, loss_cns_2: 0.6417, loss_yns_2: 0.1643, loss_cls_3: 1.0973, loss_box_3: 2.1183, loss_cns_3: 0.6481, loss_yns_3: 0.1637, loss_cls_4: 1.1293, loss_box_4: 2.1408, loss_cns_4: 0.6459, loss_yns_4: 0.1679, loss_cls_5: 1.0960, loss_box_5: 2.1264, loss_cns_5: 0.6452, loss_yns_5: 0.1635, loss_cls_dn_0: 0.3703, loss_box_dn_0: 0.8615, loss_cls_dn_1: 0.2510, loss_box_dn_1: 1.0383, loss_cls_dn_2: 0.3035, loss_box_dn_2: 1.0211, loss_cls_dn_3: 0.2834, loss_box_dn_3: 1.0230, loss_cls_dn_4: 0.2623, loss_box_dn_4: 1.0568, loss_cls_dn_5: 0.3262, loss_box_dn_5: 1.0722, loss_dense_depth: 0.8665, loss: 32.5601, grad_norm: 56.1627 -2025-11-12 14:32:20,839 - mmdet - INFO - Iter [66/17500] lr: 1.260e-04, eta: 16:10:28, time: 1.588, data_time: 0.104, memory: 49164, loss_cls_0: 0.9578, loss_box_0: 1.8188, loss_cns_0: 0.6239, loss_yns_0: 0.1601, loss_cls_1: 1.0736, loss_box_1: 2.0674, loss_cns_1: 0.6327, loss_yns_1: 0.1674, loss_cls_2: 1.0748, loss_box_2: 2.0374, loss_cns_2: 0.6452, loss_yns_2: 0.1686, loss_cls_3: 1.0899, loss_box_3: 2.0758, loss_cns_3: 0.6486, loss_yns_3: 0.1631, loss_cls_4: 1.1015, loss_box_4: 2.0850, loss_cns_4: 0.6477, loss_yns_4: 0.1650, loss_cls_5: 1.1051, loss_box_5: 2.1061, loss_cns_5: 0.6454, loss_yns_5: 0.1628, loss_cls_dn_0: 0.3512, loss_box_dn_0: 0.8455, loss_cls_dn_1: 0.2421, loss_box_dn_1: 1.0682, loss_cls_dn_2: 0.2841, loss_box_dn_2: 1.0489, loss_cls_dn_3: 0.2813, loss_box_dn_3: 1.0551, loss_cls_dn_4: 0.2680, loss_box_dn_4: 1.0701, loss_cls_dn_5: 0.3110, loss_box_dn_5: 1.0901, loss_dense_depth: 0.8953, loss: 32.2349, grad_norm: 63.0118 -2025-11-12 14:32:22,390 - mmdet - INFO - Iter [67/17500] lr: 1.264e-04, eta: 16:02:40, time: 1.554, data_time: 0.073, memory: 49164, loss_cls_0: 0.9965, loss_box_0: 1.8496, loss_cns_0: 0.6228, loss_yns_0: 0.1626, loss_cls_1: 1.0229, loss_box_1: 2.0403, loss_cns_1: 0.6272, loss_yns_1: 0.1624, loss_cls_2: 1.0699, loss_box_2: 2.0146, loss_cns_2: 0.6419, loss_yns_2: 0.1627, loss_cls_3: 1.0656, loss_box_3: 2.0627, loss_cns_3: 0.6452, loss_yns_3: 0.1618, loss_cls_4: 1.0567, loss_box_4: 2.0240, loss_cns_4: 0.6454, loss_yns_4: 0.1618, loss_cls_5: 1.0733, loss_box_5: 2.0299, loss_cns_5: 0.6452, loss_yns_5: 0.1614, loss_cls_dn_0: 0.3490, loss_box_dn_0: 0.8502, loss_cls_dn_1: 0.2424, loss_box_dn_1: 0.9741, loss_cls_dn_2: 0.2831, loss_box_dn_2: 0.9508, loss_cls_dn_3: 0.3019, loss_box_dn_3: 0.9619, loss_cls_dn_4: 0.3036, loss_box_dn_4: 0.9542, loss_cls_dn_5: 0.3104, loss_box_dn_5: 0.9615, loss_dense_depth: 0.8671, loss: 31.4166, grad_norm: 46.7345 -2025-11-12 14:32:23,953 - mmdet - INFO - Iter [68/17500] lr: 1.268e-04, eta: 15:55:09, time: 1.565, data_time: 0.075, memory: 49164, loss_cls_0: 0.9911, loss_box_0: 1.8574, loss_cns_0: 0.6245, loss_yns_0: 0.1621, loss_cls_1: 1.0600, loss_box_1: 2.0186, loss_cns_1: 0.6296, loss_yns_1: 0.1636, loss_cls_2: 1.0989, loss_box_2: 1.9776, loss_cns_2: 0.6438, loss_yns_2: 0.1625, loss_cls_3: 1.0921, loss_box_3: 2.0220, loss_cns_3: 0.6488, loss_yns_3: 0.1625, loss_cls_4: 1.0921, loss_box_4: 2.0016, loss_cns_4: 0.6473, loss_yns_4: 0.1649, loss_cls_5: 1.1092, loss_box_5: 1.9675, loss_cns_5: 0.6460, loss_yns_5: 0.1636, loss_cls_dn_0: 0.3551, loss_box_dn_0: 0.8581, loss_cls_dn_1: 0.2436, loss_box_dn_1: 0.8753, loss_cls_dn_2: 0.2894, loss_box_dn_2: 0.8544, loss_cls_dn_3: 0.3229, loss_box_dn_3: 0.8730, loss_cls_dn_4: 0.3434, loss_box_dn_4: 0.8685, loss_cls_dn_5: 0.3194, loss_box_dn_5: 0.8586, loss_dense_depth: 0.9084, loss: 31.0775, grad_norm: 58.5612 -2025-11-12 14:32:25,561 - mmdet - INFO - Iter [69/17500] lr: 1.272e-04, eta: 15:48:01, time: 1.608, data_time: 0.075, memory: 49164, loss_cls_0: 0.9311, loss_box_0: 1.7958, loss_cns_0: 0.6241, loss_yns_0: 0.1593, loss_cls_1: 1.0595, loss_box_1: 2.0152, loss_cns_1: 0.6337, loss_yns_1: 0.1634, loss_cls_2: 1.0816, loss_box_2: 1.9920, loss_cns_2: 0.6488, loss_yns_2: 0.1620, loss_cls_3: 1.1049, loss_box_3: 2.0253, loss_cns_3: 0.6490, loss_yns_3: 0.1601, loss_cls_4: 1.1143, loss_box_4: 2.0169, loss_cns_4: 0.6472, loss_yns_4: 0.1611, loss_cls_5: 1.1280, loss_box_5: 1.9886, loss_cns_5: 0.6449, loss_yns_5: 0.1631, loss_cls_dn_0: 0.3632, loss_box_dn_0: 0.8359, loss_cls_dn_1: 0.2504, loss_box_dn_1: 0.8721, loss_cls_dn_2: 0.2992, loss_box_dn_2: 0.8625, loss_cls_dn_3: 0.3278, loss_box_dn_3: 0.8857, loss_cls_dn_4: 0.3487, loss_box_dn_4: 0.8850, loss_cls_dn_5: 0.3256, loss_box_dn_5: 0.8798, loss_dense_depth: 0.8462, loss: 31.0522, grad_norm: 65.1207 -2025-11-12 14:32:27,127 - mmdet - INFO - Iter [70/17500] lr: 1.276e-04, eta: 15:40:55, time: 1.563, data_time: 0.073, memory: 49164, loss_cls_0: 0.9467, loss_box_0: 1.7986, loss_cns_0: 0.6208, loss_yns_0: 0.1563, loss_cls_1: 0.9995, loss_box_1: 2.0315, loss_cns_1: 0.6251, loss_yns_1: 0.1636, loss_cls_2: 1.0463, loss_box_2: 1.9960, loss_cns_2: 0.6415, loss_yns_2: 0.1608, loss_cls_3: 1.0544, loss_box_3: 1.9852, loss_cns_3: 0.6453, loss_yns_3: 0.1599, loss_cls_4: 1.0760, loss_box_4: 1.9887, loss_cns_4: 0.6445, loss_yns_4: 0.1611, loss_cls_5: 1.0565, loss_box_5: 1.9763, loss_cns_5: 0.6425, loss_yns_5: 0.1630, loss_cls_dn_0: 0.3914, loss_box_dn_0: 0.8433, loss_cls_dn_1: 0.2608, loss_box_dn_1: 0.8359, loss_cls_dn_2: 0.3141, loss_box_dn_2: 0.8353, loss_cls_dn_3: 0.3337, loss_box_dn_3: 0.8564, loss_cls_dn_4: 0.3413, loss_box_dn_4: 0.8730, loss_cls_dn_5: 0.3417, loss_box_dn_5: 0.8966, loss_dense_depth: 0.8717, loss: 30.7351, grad_norm: 50.4857 -2025-11-12 14:32:28,714 - mmdet - INFO - Iter [71/17500] lr: 1.280e-04, eta: 15:34:06, time: 1.587, data_time: 0.071, memory: 49164, loss_cls_0: 0.9325, loss_box_0: 1.8429, loss_cns_0: 0.6218, loss_yns_0: 0.1566, loss_cls_1: 1.0019, loss_box_1: 2.0681, loss_cns_1: 0.6234, loss_yns_1: 0.1592, loss_cls_2: 1.0357, loss_box_2: 2.0138, loss_cns_2: 0.6444, loss_yns_2: 0.1583, loss_cls_3: 1.0486, loss_box_3: 2.0192, loss_cns_3: 0.6478, loss_yns_3: 0.1600, loss_cls_4: 1.0452, loss_box_4: 2.0189, loss_cns_4: 0.6455, loss_yns_4: 0.1665, loss_cls_5: 1.0459, loss_box_5: 1.9997, loss_cns_5: 0.6442, loss_yns_5: 0.1604, loss_cls_dn_0: 0.3750, loss_box_dn_0: 0.8644, loss_cls_dn_1: 0.2450, loss_box_dn_1: 0.8898, loss_cls_dn_2: 0.2913, loss_box_dn_2: 0.8811, loss_cls_dn_3: 0.3107, loss_box_dn_3: 0.9089, loss_cls_dn_4: 0.2966, loss_box_dn_4: 0.9246, loss_cls_dn_5: 0.3294, loss_box_dn_5: 0.9459, loss_dense_depth: 0.8548, loss: 30.9781, grad_norm: 62.9668 -2025-11-12 14:32:30,282 - mmdet - INFO - Iter [72/17500] lr: 1.284e-04, eta: 15:27:24, time: 1.568, data_time: 0.078, memory: 49164, loss_cls_0: 0.9424, loss_box_0: 1.8316, loss_cns_0: 0.6229, loss_yns_0: 0.1599, loss_cls_1: 1.0404, loss_box_1: 1.9998, loss_cns_1: 0.6303, loss_yns_1: 0.1594, loss_cls_2: 1.0772, loss_box_2: 1.9588, loss_cns_2: 0.6445, loss_yns_2: 0.1581, loss_cls_3: 1.1145, loss_box_3: 1.9746, loss_cns_3: 0.6488, loss_yns_3: 0.1585, loss_cls_4: 1.1290, loss_box_4: 1.9348, loss_cns_4: 0.6521, loss_yns_4: 0.1621, loss_cls_5: 1.0925, loss_box_5: 1.9310, loss_cns_5: 0.6496, loss_yns_5: 0.1651, loss_cls_dn_0: 0.3494, loss_box_dn_0: 0.8558, loss_cls_dn_1: 0.2415, loss_box_dn_1: 0.8663, loss_cls_dn_2: 0.2789, loss_box_dn_2: 0.8669, loss_cls_dn_3: 0.3051, loss_box_dn_3: 0.8966, loss_cls_dn_4: 0.2765, loss_box_dn_4: 0.8964, loss_cls_dn_5: 0.3275, loss_box_dn_5: 0.9138, loss_dense_depth: 0.8574, loss: 30.7697, grad_norm: 75.1198 -2025-11-12 14:32:31,860 - mmdet - INFO - Iter [73/17500] lr: 1.288e-04, eta: 15:20:54, time: 1.574, data_time: 0.073, memory: 49164, loss_cls_0: 0.9956, loss_box_0: 1.8535, loss_cns_0: 0.6181, loss_yns_0: 0.1606, loss_cls_1: 1.0302, loss_box_1: 2.0088, loss_cns_1: 0.6288, loss_yns_1: 0.1626, loss_cls_2: 1.0919, loss_box_2: 1.9757, loss_cns_2: 0.6419, loss_yns_2: 0.1609, loss_cls_3: 1.1023, loss_box_3: 1.9957, loss_cns_3: 0.6467, loss_yns_3: 0.1627, loss_cls_4: 1.1605, loss_box_4: 1.9575, loss_cns_4: 0.6504, loss_yns_4: 0.1636, loss_cls_5: 1.1012, loss_box_5: 1.9730, loss_cns_5: 0.6473, loss_yns_5: 0.1747, loss_cls_dn_0: 0.3310, loss_box_dn_0: 0.8522, loss_cls_dn_1: 0.2329, loss_box_dn_1: 0.8846, loss_cls_dn_2: 0.2652, loss_box_dn_2: 0.8869, loss_cls_dn_3: 0.2877, loss_box_dn_3: 0.9121, loss_cls_dn_4: 0.2621, loss_box_dn_4: 0.9096, loss_cls_dn_5: 0.3088, loss_box_dn_5: 0.9333, loss_dense_depth: 0.8835, loss: 31.0142, grad_norm: 63.2779 -2025-11-12 14:32:33,427 - mmdet - INFO - Iter [74/17500] lr: 1.292e-04, eta: 15:14:35, time: 1.574, data_time: 0.078, memory: 49164, loss_cls_0: 0.9956, loss_box_0: 1.8489, loss_cns_0: 0.6174, loss_yns_0: 0.1577, loss_cls_1: 1.0196, loss_box_1: 2.0443, loss_cns_1: 0.6337, loss_yns_1: 0.1626, loss_cls_2: 1.0607, loss_box_2: 1.9945, loss_cns_2: 0.6505, loss_yns_2: 0.1633, loss_cls_3: 1.0651, loss_box_3: 2.0189, loss_cns_3: 0.6521, loss_yns_3: 0.1661, loss_cls_4: 1.0598, loss_box_4: 2.0331, loss_cns_4: 0.6533, loss_yns_4: 0.1660, loss_cls_5: 1.0566, loss_box_5: 2.0423, loss_cns_5: 0.6502, loss_yns_5: 0.1726, loss_cls_dn_0: 0.3318, loss_box_dn_0: 0.8461, loss_cls_dn_1: 0.2267, loss_box_dn_1: 0.9310, loss_cls_dn_2: 0.2622, loss_box_dn_2: 0.9159, loss_cls_dn_3: 0.2892, loss_box_dn_3: 0.9339, loss_cls_dn_4: 0.2847, loss_box_dn_4: 0.9473, loss_cls_dn_5: 0.3218, loss_box_dn_5: 0.9661, loss_dense_depth: 0.8491, loss: 31.1905, grad_norm: 53.7441 -2025-11-12 14:32:35,012 - mmdet - INFO - Iter [75/17500] lr: 1.296e-04, eta: 15:08:28, time: 1.583, data_time: 0.074, memory: 49164, loss_cls_0: 0.9551, loss_box_0: 1.8710, loss_cns_0: 0.6136, loss_yns_0: 0.1601, loss_cls_1: 1.0427, loss_box_1: 2.1305, loss_cns_1: 0.6294, loss_yns_1: 0.1666, loss_cls_2: 1.0749, loss_box_2: 2.0797, loss_cns_2: 0.6483, loss_yns_2: 0.1647, loss_cls_3: 1.0785, loss_box_3: 2.0926, loss_cns_3: 0.6513, loss_yns_3: 0.1690, loss_cls_4: 1.1024, loss_box_4: 2.0966, loss_cns_4: 0.6473, loss_yns_4: 0.1689, loss_cls_5: 1.0745, loss_box_5: 2.1119, loss_cns_5: 0.6475, loss_yns_5: 0.1643, loss_cls_dn_0: 0.3445, loss_box_dn_0: 0.8343, loss_cls_dn_1: 0.2293, loss_box_dn_1: 0.9436, loss_cls_dn_2: 0.2707, loss_box_dn_2: 0.9180, loss_cls_dn_3: 0.3051, loss_box_dn_3: 0.9266, loss_cls_dn_4: 0.3253, loss_box_dn_4: 0.9370, loss_cls_dn_5: 0.3361, loss_box_dn_5: 0.9516, loss_dense_depth: 0.8494, loss: 31.7128, grad_norm: 65.3979 -2025-11-12 14:32:36,593 - mmdet - INFO - Iter [76/17500] lr: 1.300e-04, eta: 15:02:28, time: 1.572, data_time: 0.076, memory: 49164, loss_cls_0: 0.9475, loss_box_0: 1.8334, loss_cns_0: 0.6143, loss_yns_0: 0.1628, loss_cls_1: 1.0508, loss_box_1: 2.1637, loss_cns_1: 0.6302, loss_yns_1: 0.1648, loss_cls_2: 1.0872, loss_box_2: 2.0973, loss_cns_2: 0.6487, loss_yns_2: 0.1647, loss_cls_3: 1.0912, loss_box_3: 2.0824, loss_cns_3: 0.6560, loss_yns_3: 0.1658, loss_cls_4: 1.1174, loss_box_4: 2.0430, loss_cns_4: 0.6502, loss_yns_4: 0.1679, loss_cls_5: 1.0715, loss_box_5: 2.0836, loss_cns_5: 0.6530, loss_yns_5: 0.1644, loss_cls_dn_0: 0.3586, loss_box_dn_0: 0.8355, loss_cls_dn_1: 0.2376, loss_box_dn_1: 0.8769, loss_cls_dn_2: 0.2762, loss_box_dn_2: 0.8421, loss_cls_dn_3: 0.3048, loss_box_dn_3: 0.8337, loss_cls_dn_4: 0.3240, loss_box_dn_4: 0.8337, loss_cls_dn_5: 0.3199, loss_box_dn_5: 0.8500, loss_dense_depth: 0.8460, loss: 31.2505, grad_norm: 49.2882 -2025-11-12 14:32:38,154 - mmdet - INFO - Iter [77/17500] lr: 1.304e-04, eta: 14:56:37, time: 1.573, data_time: 0.078, memory: 49164, loss_cls_0: 0.9416, loss_box_0: 1.8512, loss_cns_0: 0.6144, loss_yns_0: 0.1617, loss_cls_1: 1.0337, loss_box_1: 2.0996, loss_cns_1: 0.6323, loss_yns_1: 0.1643, loss_cls_2: 1.0641, loss_box_2: 2.0675, loss_cns_2: 0.6510, loss_yns_2: 0.1622, loss_cls_3: 1.0817, loss_box_3: 2.0321, loss_cns_3: 0.6587, loss_yns_3: 0.1606, loss_cls_4: 1.0651, loss_box_4: 2.0453, loss_cns_4: 0.6557, loss_yns_4: 0.1637, loss_cls_5: 1.0753, loss_box_5: 2.0510, loss_cns_5: 0.6510, loss_yns_5: 0.1739, loss_cls_dn_0: 0.3576, loss_box_dn_0: 0.8257, loss_cls_dn_1: 0.2305, loss_box_dn_1: 0.8000, loss_cls_dn_2: 0.2657, loss_box_dn_2: 0.7849, loss_cls_dn_3: 0.2858, loss_box_dn_3: 0.7731, loss_cls_dn_4: 0.2856, loss_box_dn_4: 0.7903, loss_cls_dn_5: 0.2915, loss_box_dn_5: 0.8091, loss_dense_depth: 0.8684, loss: 30.6258, grad_norm: 42.3387 -2025-11-12 14:32:39,735 - mmdet - INFO - Iter [78/17500] lr: 1.308e-04, eta: 14:50:58, time: 1.581, data_time: 0.071, memory: 49164, loss_cls_0: 0.9590, loss_box_0: 1.8577, loss_cns_0: 0.6167, loss_yns_0: 0.1612, loss_cls_1: 1.0146, loss_box_1: 2.1155, loss_cns_1: 0.6274, loss_yns_1: 0.1647, loss_cls_2: 1.0698, loss_box_2: 2.0882, loss_cns_2: 0.6469, loss_yns_2: 0.1654, loss_cls_3: 1.0876, loss_box_3: 2.0852, loss_cns_3: 0.6532, loss_yns_3: 0.1672, loss_cls_4: 1.1509, loss_box_4: 2.1067, loss_cns_4: 0.6522, loss_yns_4: 0.1646, loss_cls_5: 1.1466, loss_box_5: 2.1073, loss_cns_5: 0.6472, loss_yns_5: 0.1769, loss_cls_dn_0: 0.3541, loss_box_dn_0: 0.8252, loss_cls_dn_1: 0.2106, loss_box_dn_1: 0.8214, loss_cls_dn_2: 0.2493, loss_box_dn_2: 0.8134, loss_cls_dn_3: 0.2657, loss_box_dn_3: 0.8194, loss_cls_dn_4: 0.2527, loss_box_dn_4: 0.8522, loss_cls_dn_5: 0.2783, loss_box_dn_5: 0.8738, loss_dense_depth: 0.8314, loss: 31.0802, grad_norm: 66.9393 -2025-11-12 14:32:41,315 - mmdet - INFO - Iter [79/17500] lr: 1.312e-04, eta: 14:45:27, time: 1.580, data_time: 0.072, memory: 49164, loss_cls_0: 0.9423, loss_box_0: 1.8481, loss_cns_0: 0.6233, loss_yns_0: 0.1608, loss_cls_1: 1.0331, loss_box_1: 2.0632, loss_cns_1: 0.6342, loss_yns_1: 0.1612, loss_cls_2: 1.0739, loss_box_2: 2.0518, loss_cns_2: 0.6465, loss_yns_2: 0.1670, loss_cls_3: 1.1038, loss_box_3: 2.0744, loss_cns_3: 0.6501, loss_yns_3: 0.1691, loss_cls_4: 1.1366, loss_box_4: 2.0844, loss_cns_4: 0.6506, loss_yns_4: 0.1640, loss_cls_5: 1.0826, loss_box_5: 2.0905, loss_cns_5: 0.6478, loss_yns_5: 0.1685, loss_cls_dn_0: 0.3321, loss_box_dn_0: 0.8276, loss_cls_dn_1: 0.2142, loss_box_dn_1: 0.8667, loss_cls_dn_2: 0.2503, loss_box_dn_2: 0.8648, loss_cls_dn_3: 0.2685, loss_box_dn_3: 0.8862, loss_cls_dn_4: 0.2514, loss_box_dn_4: 0.9196, loss_cls_dn_5: 0.2897, loss_box_dn_5: 0.9473, loss_dense_depth: 0.8397, loss: 31.1856, grad_norm: 69.4956 -2025-11-12 14:32:42,891 - mmdet - INFO - Iter [80/17500] lr: 1.316e-04, eta: 14:40:02, time: 1.571, data_time: 0.073, memory: 49164, loss_cls_0: 0.9493, loss_box_0: 1.8497, loss_cns_0: 0.6237, loss_yns_0: 0.1604, loss_cls_1: 1.0235, loss_box_1: 2.0998, loss_cns_1: 0.6309, loss_yns_1: 0.1615, loss_cls_2: 1.0388, loss_box_2: 2.0707, loss_cns_2: 0.6397, loss_yns_2: 0.1637, loss_cls_3: 1.0650, loss_box_3: 2.0806, loss_cns_3: 0.6483, loss_yns_3: 0.1643, loss_cls_4: 1.0463, loss_box_4: 2.0723, loss_cns_4: 0.6470, loss_yns_4: 0.1644, loss_cls_5: 1.0370, loss_box_5: 2.0877, loss_cns_5: 0.6460, loss_yns_5: 0.1644, loss_cls_dn_0: 0.3258, loss_box_dn_0: 0.8248, loss_cls_dn_1: 0.2214, loss_box_dn_1: 0.8963, loss_cls_dn_2: 0.2415, loss_box_dn_2: 0.8861, loss_cls_dn_3: 0.2688, loss_box_dn_3: 0.9020, loss_cls_dn_4: 0.2666, loss_box_dn_4: 0.9243, loss_cls_dn_5: 0.3029, loss_box_dn_5: 0.9510, loss_dense_depth: 0.8085, loss: 31.0551, grad_norm: 50.8392 -2025-11-12 14:32:44,569 - mmdet - INFO - Iter [81/17500] lr: 1.320e-04, eta: 14:35:08, time: 1.678, data_time: 0.107, memory: 49164, loss_cls_0: 0.9611, loss_box_0: 1.8854, loss_cns_0: 0.6197, loss_yns_0: 0.1610, loss_cls_1: 1.0086, loss_box_1: 2.0575, loss_cns_1: 0.6271, loss_yns_1: 0.1611, loss_cls_2: 1.0753, loss_box_2: 2.0095, loss_cns_2: 0.6435, loss_yns_2: 0.1642, loss_cls_3: 1.0455, loss_box_3: 2.0057, loss_cns_3: 0.6550, loss_yns_3: 0.1629, loss_cls_4: 1.0378, loss_box_4: 2.0125, loss_cns_4: 0.6517, loss_yns_4: 0.1647, loss_cls_5: 1.0654, loss_box_5: 2.0407, loss_cns_5: 0.6449, loss_yns_5: 0.1702, loss_cls_dn_0: 0.3313, loss_box_dn_0: 0.8289, loss_cls_dn_1: 0.2148, loss_box_dn_1: 0.9240, loss_cls_dn_2: 0.2394, loss_box_dn_2: 0.9068, loss_cls_dn_3: 0.2609, loss_box_dn_3: 0.9117, loss_cls_dn_4: 0.2695, loss_box_dn_4: 0.9299, loss_cls_dn_5: 0.2928, loss_box_dn_5: 0.9498, loss_dense_depth: 0.8460, loss: 30.9370, grad_norm: 42.2379 -2025-11-12 14:32:46,200 - mmdet - INFO - Iter [82/17500] lr: 1.324e-04, eta: 14:30:11, time: 1.631, data_time: 0.111, memory: 49164, loss_cls_0: 0.9546, loss_box_0: 1.8724, loss_cns_0: 0.6211, loss_yns_0: 0.1565, loss_cls_1: 1.0134, loss_box_1: 2.0742, loss_cns_1: 0.6250, loss_yns_1: 0.1592, loss_cls_2: 1.0556, loss_box_2: 2.0396, loss_cns_2: 0.6445, loss_yns_2: 0.1673, loss_cls_3: 1.0861, loss_box_3: 2.0536, loss_cns_3: 0.6520, loss_yns_3: 0.1648, loss_cls_4: 1.0723, loss_box_4: 2.0822, loss_cns_4: 0.6526, loss_yns_4: 0.1630, loss_cls_5: 1.1609, loss_box_5: 2.0396, loss_cns_5: 0.6474, loss_yns_5: 0.1714, loss_cls_dn_0: 0.3464, loss_box_dn_0: 0.8342, loss_cls_dn_1: 0.2045, loss_box_dn_1: 0.9076, loss_cls_dn_2: 0.2361, loss_box_dn_2: 0.8949, loss_cls_dn_3: 0.2536, loss_box_dn_3: 0.9067, loss_cls_dn_4: 0.2611, loss_box_dn_4: 0.9259, loss_cls_dn_5: 0.2760, loss_box_dn_5: 0.9188, loss_dense_depth: 0.8590, loss: 31.1539, grad_norm: 56.3532 -2025-11-12 14:32:47,796 - mmdet - INFO - Iter [83/17500] lr: 1.328e-04, eta: 14:25:13, time: 1.593, data_time: 0.085, memory: 49164, loss_cls_0: 0.9464, loss_box_0: 1.8265, loss_cns_0: 0.6254, loss_yns_0: 0.1591, loss_cls_1: 1.0120, loss_box_1: 1.9953, loss_cns_1: 0.6286, loss_yns_1: 0.1612, loss_cls_2: 1.0646, loss_box_2: 1.9482, loss_cns_2: 0.6439, loss_yns_2: 0.1663, loss_cls_3: 1.0799, loss_box_3: 1.9445, loss_cns_3: 0.6493, loss_yns_3: 0.1656, loss_cls_4: 1.0643, loss_box_4: 1.9691, loss_cns_4: 0.6511, loss_yns_4: 0.1631, loss_cls_5: 1.0818, loss_box_5: 1.9471, loss_cns_5: 0.6496, loss_yns_5: 0.1660, loss_cls_dn_0: 0.3507, loss_box_dn_0: 0.8333, loss_cls_dn_1: 0.2091, loss_box_dn_1: 0.8560, loss_cls_dn_2: 0.2385, loss_box_dn_2: 0.8354, loss_cls_dn_3: 0.2565, loss_box_dn_3: 0.8434, loss_cls_dn_4: 0.2639, loss_box_dn_4: 0.8605, loss_cls_dn_5: 0.2847, loss_box_dn_5: 0.8663, loss_dense_depth: 0.8272, loss: 30.2343, grad_norm: 49.5034 -2025-11-12 14:32:49,365 - mmdet - INFO - Iter [84/17500] lr: 1.332e-04, eta: 14:20:17, time: 1.567, data_time: 0.080, memory: 49164, loss_cls_0: 0.9021, loss_box_0: 1.8162, loss_cns_0: 0.6257, loss_yns_0: 0.1568, loss_cls_1: 0.9792, loss_box_1: 1.9846, loss_cns_1: 0.6365, loss_yns_1: 0.1586, loss_cls_2: 1.0751, loss_box_2: 1.9294, loss_cns_2: 0.6497, loss_yns_2: 0.1620, loss_cls_3: 1.0428, loss_box_3: 1.9425, loss_cns_3: 0.6519, loss_yns_3: 0.1656, loss_cls_4: 1.0656, loss_box_4: 1.9210, loss_cns_4: 0.6529, loss_yns_4: 0.1588, loss_cls_5: 1.0660, loss_box_5: 1.9511, loss_cns_5: 0.6512, loss_yns_5: 0.1571, loss_cls_dn_0: 0.3225, loss_box_dn_0: 0.8254, loss_cls_dn_1: 0.2134, loss_box_dn_1: 0.7934, loss_cls_dn_2: 0.2459, loss_box_dn_2: 0.7781, loss_cls_dn_3: 0.2623, loss_box_dn_3: 0.7816, loss_cls_dn_4: 0.2785, loss_box_dn_4: 0.7923, loss_cls_dn_5: 0.2918, loss_box_dn_5: 0.8258, loss_dense_depth: 0.8173, loss: 29.7307, grad_norm: 35.9579 -2025-11-12 14:32:50,966 - mmdet - INFO - Iter [85/17500] lr: 1.336e-04, eta: 14:15:35, time: 1.600, data_time: 0.089, memory: 49164, loss_cls_0: 0.9684, loss_box_0: 1.8443, loss_cns_0: 0.6235, loss_yns_0: 0.1577, loss_cls_1: 1.0133, loss_box_1: 1.9753, loss_cns_1: 0.6340, loss_yns_1: 0.1626, loss_cls_2: 1.0715, loss_box_2: 1.9320, loss_cns_2: 0.6484, loss_yns_2: 0.1601, loss_cls_3: 1.0708, loss_box_3: 1.9602, loss_cns_3: 0.6488, loss_yns_3: 0.1608, loss_cls_4: 1.0607, loss_box_4: 1.9523, loss_cns_4: 0.6503, loss_yns_4: 0.1598, loss_cls_5: 1.0711, loss_box_5: 1.9365, loss_cns_5: 0.6493, loss_yns_5: 0.1603, loss_cls_dn_0: 0.3150, loss_box_dn_0: 0.8475, loss_cls_dn_1: 0.2103, loss_box_dn_1: 0.8244, loss_cls_dn_2: 0.2447, loss_box_dn_2: 0.8193, loss_cls_dn_3: 0.2557, loss_box_dn_3: 0.8308, loss_cls_dn_4: 0.2566, loss_box_dn_4: 0.8448, loss_cls_dn_5: 0.2648, loss_box_dn_5: 0.8537, loss_dense_depth: 0.8396, loss: 30.0794, grad_norm: 46.2139 -2025-11-12 14:32:52,556 - mmdet - INFO - Iter [86/17500] lr: 1.340e-04, eta: 14:10:57, time: 1.591, data_time: 0.109, memory: 49164, loss_cls_0: 0.9847, loss_box_0: 1.8292, loss_cns_0: 0.6232, loss_yns_0: 0.1582, loss_cls_1: 1.0355, loss_box_1: 1.9195, loss_cns_1: 0.6347, loss_yns_1: 0.1629, loss_cls_2: 1.0408, loss_box_2: 1.8723, loss_cns_2: 0.6471, loss_yns_2: 0.1680, loss_cls_3: 1.0784, loss_box_3: 1.8812, loss_cns_3: 0.6503, loss_yns_3: 0.1634, loss_cls_4: 1.0883, loss_box_4: 1.8807, loss_cns_4: 0.6529, loss_yns_4: 0.1614, loss_cls_5: 1.1920, loss_box_5: 1.8709, loss_cns_5: 0.6499, loss_yns_5: 0.1691, loss_cls_dn_0: 0.2943, loss_box_dn_0: 0.8352, loss_cls_dn_1: 0.2031, loss_box_dn_1: 0.8211, loss_cls_dn_2: 0.2307, loss_box_dn_2: 0.8157, loss_cls_dn_3: 0.2392, loss_box_dn_3: 0.8268, loss_cls_dn_4: 0.2321, loss_box_dn_4: 0.8413, loss_cls_dn_5: 0.2477, loss_box_dn_5: 0.8505, loss_dense_depth: 0.8207, loss: 29.7735, grad_norm: 63.8130 -2025-11-12 14:32:54,114 - mmdet - INFO - Iter [87/17500] lr: 1.344e-04, eta: 14:06:20, time: 1.564, data_time: 0.080, memory: 49164, loss_cls_0: 0.9236, loss_box_0: 1.8218, loss_cns_0: 0.6216, loss_yns_0: 0.1586, loss_cls_1: 0.9887, loss_box_1: 1.9256, loss_cns_1: 0.6434, loss_yns_1: 0.1596, loss_cls_2: 1.0324, loss_box_2: 1.8988, loss_cns_2: 0.6524, loss_yns_2: 0.1675, loss_cls_3: 1.0395, loss_box_3: 1.9030, loss_cns_3: 0.6556, loss_yns_3: 0.1644, loss_cls_4: 1.0676, loss_box_4: 1.8733, loss_cns_4: 0.6576, loss_yns_4: 0.1612, loss_cls_5: 1.0411, loss_box_5: 1.9027, loss_cns_5: 0.6535, loss_yns_5: 0.1733, loss_cls_dn_0: 0.3033, loss_box_dn_0: 0.8237, loss_cls_dn_1: 0.2033, loss_box_dn_1: 0.8215, loss_cls_dn_2: 0.2225, loss_box_dn_2: 0.8111, loss_cls_dn_3: 0.2389, loss_box_dn_3: 0.8264, loss_cls_dn_4: 0.2281, loss_box_dn_4: 0.8247, loss_cls_dn_5: 0.2502, loss_box_dn_5: 0.8538, loss_dense_depth: 0.8282, loss: 29.5227, grad_norm: 45.1047 -2025-11-12 14:32:55,680 - mmdet - INFO - Iter [88/17500] lr: 1.348e-04, eta: 14:01:49, time: 1.558, data_time: 0.082, memory: 49164, loss_cls_0: 0.9210, loss_box_0: 1.7830, loss_cns_0: 0.6101, loss_yns_0: 0.1532, loss_cls_1: 1.0209, loss_box_1: 1.8988, loss_cns_1: 0.6405, loss_yns_1: 0.1590, loss_cls_2: 1.0297, loss_box_2: 1.8643, loss_cns_2: 0.6516, loss_yns_2: 0.1611, loss_cls_3: 1.1126, loss_box_3: 1.8371, loss_cns_3: 0.6474, loss_yns_3: 0.1613, loss_cls_4: 1.0365, loss_box_4: 1.8650, loss_cns_4: 0.6564, loss_yns_4: 0.1600, loss_cls_5: 1.0827, loss_box_5: 1.8536, loss_cns_5: 0.6535, loss_yns_5: 0.1684, loss_cls_dn_0: 0.3271, loss_box_dn_0: 0.8256, loss_cls_dn_1: 0.2068, loss_box_dn_1: 0.8229, loss_cls_dn_2: 0.2218, loss_box_dn_2: 0.8118, loss_cls_dn_3: 0.2564, loss_box_dn_3: 0.8199, loss_cls_dn_4: 0.2442, loss_box_dn_4: 0.8182, loss_cls_dn_5: 0.2827, loss_box_dn_5: 0.8325, loss_dense_depth: 0.8171, loss: 29.4149, grad_norm: 40.1224 -2025-11-12 14:32:57,276 - mmdet - INFO - Iter [89/17500] lr: 1.352e-04, eta: 13:57:31, time: 1.602, data_time: 0.088, memory: 49164, loss_cls_0: 0.9161, loss_box_0: 1.7823, loss_cns_0: 0.6149, loss_yns_0: 0.1531, loss_cls_1: 1.0097, loss_box_1: 1.9416, loss_cns_1: 0.6285, loss_yns_1: 0.1588, loss_cls_2: 1.0569, loss_box_2: 1.8735, loss_cns_2: 0.6482, loss_yns_2: 0.1600, loss_cls_3: 1.0403, loss_box_3: 1.8514, loss_cns_3: 0.6515, loss_yns_3: 0.1622, loss_cls_4: 1.0341, loss_box_4: 1.8821, loss_cns_4: 0.6542, loss_yns_4: 0.1604, loss_cls_5: 1.0523, loss_box_5: 1.8575, loss_cns_5: 0.6547, loss_yns_5: 0.1594, loss_cls_dn_0: 0.3147, loss_box_dn_0: 0.8151, loss_cls_dn_1: 0.2055, loss_box_dn_1: 0.8045, loss_cls_dn_2: 0.2247, loss_box_dn_2: 0.7877, loss_cls_dn_3: 0.2402, loss_box_dn_3: 0.7791, loss_cls_dn_4: 0.2425, loss_box_dn_4: 0.7928, loss_cls_dn_5: 0.2734, loss_box_dn_5: 0.7957, loss_dense_depth: 0.8044, loss: 29.1837, grad_norm: 42.8582 -2025-11-12 14:32:58,848 - mmdet - INFO - Iter [90/17500] lr: 1.356e-04, eta: 13:53:14, time: 1.569, data_time: 0.085, memory: 49164, loss_cls_0: 0.9465, loss_box_0: 1.8114, loss_cns_0: 0.6217, loss_yns_0: 0.1554, loss_cls_1: 0.9992, loss_box_1: 1.9962, loss_cns_1: 0.6350, loss_yns_1: 0.1605, loss_cls_2: 1.1018, loss_box_2: 1.9193, loss_cns_2: 0.6501, loss_yns_2: 0.1601, loss_cls_3: 1.1611, loss_box_3: 1.8935, loss_cns_3: 0.6557, loss_yns_3: 0.1609, loss_cls_4: 1.0815, loss_box_4: 1.9239, loss_cns_4: 0.6545, loss_yns_4: 0.1610, loss_cls_5: 1.0778, loss_box_5: 1.9065, loss_cns_5: 0.6525, loss_yns_5: 0.1604, loss_cls_dn_0: 0.3016, loss_box_dn_0: 0.8256, loss_cls_dn_1: 0.2032, loss_box_dn_1: 0.8156, loss_cls_dn_2: 0.2233, loss_box_dn_2: 0.7972, loss_cls_dn_3: 0.2268, loss_box_dn_3: 0.7899, loss_cls_dn_4: 0.2266, loss_box_dn_4: 0.8138, loss_cls_dn_5: 0.2503, loss_box_dn_5: 0.8146, loss_dense_depth: 0.8662, loss: 29.8012, grad_norm: 68.0871 -2025-11-12 14:33:00,459 - mmdet - INFO - Iter [91/17500] lr: 1.360e-04, eta: 13:49:11, time: 1.616, data_time: 0.087, memory: 49164, loss_cls_0: 0.9497, loss_box_0: 1.8101, loss_cns_0: 0.6221, loss_yns_0: 0.1537, loss_cls_1: 0.9878, loss_box_1: 1.9821, loss_cns_1: 0.6365, loss_yns_1: 0.1573, loss_cls_2: 1.0250, loss_box_2: 1.9194, loss_cns_2: 0.6490, loss_yns_2: 0.1632, loss_cls_3: 1.0765, loss_box_3: 1.9067, loss_cns_3: 0.6559, loss_yns_3: 0.1604, loss_cls_4: 1.0389, loss_box_4: 1.9234, loss_cns_4: 0.6532, loss_yns_4: 0.1631, loss_cls_5: 1.0859, loss_box_5: 1.9331, loss_cns_5: 0.6504, loss_yns_5: 0.1669, loss_cls_dn_0: 0.2869, loss_box_dn_0: 0.8103, loss_cls_dn_1: 0.2021, loss_box_dn_1: 0.8172, loss_cls_dn_2: 0.2184, loss_box_dn_2: 0.7955, loss_cls_dn_3: 0.2238, loss_box_dn_3: 0.7979, loss_cls_dn_4: 0.2229, loss_box_dn_4: 0.8209, loss_cls_dn_5: 0.2445, loss_box_dn_5: 0.8360, loss_dense_depth: 0.8151, loss: 29.5621, grad_norm: 52.4438 -2025-11-12 14:33:02,043 - mmdet - INFO - Iter [92/17500] lr: 1.364e-04, eta: 13:45:06, time: 1.579, data_time: 0.075, memory: 49164, loss_cls_0: 0.9139, loss_box_0: 1.8157, loss_cns_0: 0.6192, loss_yns_0: 0.1514, loss_cls_1: 0.9720, loss_box_1: 2.0282, loss_cns_1: 0.6300, loss_yns_1: 0.1572, loss_cls_2: 1.0514, loss_box_2: 1.9800, loss_cns_2: 0.6443, loss_yns_2: 0.1637, loss_cls_3: 1.0478, loss_box_3: 1.9654, loss_cns_3: 0.6504, loss_yns_3: 0.1594, loss_cls_4: 1.0642, loss_box_4: 1.9800, loss_cns_4: 0.6493, loss_yns_4: 0.1610, loss_cls_5: 1.0321, loss_box_5: 1.9908, loss_cns_5: 0.6476, loss_yns_5: 0.1666, loss_cls_dn_0: 0.2936, loss_box_dn_0: 0.8108, loss_cls_dn_1: 0.2005, loss_box_dn_1: 0.8120, loss_cls_dn_2: 0.2216, loss_box_dn_2: 0.7961, loss_cls_dn_3: 0.2361, loss_box_dn_3: 0.7978, loss_cls_dn_4: 0.2423, loss_box_dn_4: 0.8207, loss_cls_dn_5: 0.2547, loss_box_dn_5: 0.8451, loss_dense_depth: 0.8678, loss: 29.8410, grad_norm: 64.4417 -2025-11-12 14:33:03,633 - mmdet - INFO - Iter [93/17500] lr: 1.368e-04, eta: 13:41:08, time: 1.589, data_time: 0.082, memory: 49164, loss_cls_0: 0.9149, loss_box_0: 1.8055, loss_cns_0: 0.6199, loss_yns_0: 0.1493, loss_cls_1: 0.9832, loss_box_1: 2.0257, loss_cns_1: 0.6298, loss_yns_1: 0.1585, loss_cls_2: 1.1151, loss_box_2: 1.9605, loss_cns_2: 0.6426, loss_yns_2: 0.1598, loss_cls_3: 1.0798, loss_box_3: 1.9372, loss_cns_3: 0.6499, loss_yns_3: 0.1616, loss_cls_4: 1.1187, loss_box_4: 1.9309, loss_cns_4: 0.6521, loss_yns_4: 0.1584, loss_cls_5: 1.0382, loss_box_5: 1.9290, loss_cns_5: 0.6521, loss_yns_5: 0.1671, loss_cls_dn_0: 0.2939, loss_box_dn_0: 0.8138, loss_cls_dn_1: 0.2019, loss_box_dn_1: 0.8274, loss_cls_dn_2: 0.2306, loss_box_dn_2: 0.8140, loss_cls_dn_3: 0.2508, loss_box_dn_3: 0.8066, loss_cls_dn_4: 0.2684, loss_box_dn_4: 0.8293, loss_cls_dn_5: 0.2684, loss_box_dn_5: 0.8461, loss_dense_depth: 0.8539, loss: 29.9448, grad_norm: 55.2710 -2025-11-12 14:33:05,221 - mmdet - INFO - Iter [94/17500] lr: 1.372e-04, eta: 13:37:16, time: 1.592, data_time: 0.082, memory: 49164, loss_cls_0: 0.9232, loss_box_0: 1.8130, loss_cns_0: 0.6224, loss_yns_0: 0.1516, loss_cls_1: 0.9723, loss_box_1: 2.0308, loss_cns_1: 0.6287, loss_yns_1: 0.1569, loss_cls_2: 1.1230, loss_box_2: 1.9621, loss_cns_2: 0.6397, loss_yns_2: 0.1574, loss_cls_3: 1.0581, loss_box_3: 1.9603, loss_cns_3: 0.6515, loss_yns_3: 0.1656, loss_cls_4: 1.0866, loss_box_4: 1.9465, loss_cns_4: 0.6538, loss_yns_4: 0.1564, loss_cls_5: 1.0399, loss_box_5: 1.9360, loss_cns_5: 0.6571, loss_yns_5: 0.1636, loss_cls_dn_0: 0.2903, loss_box_dn_0: 0.8045, loss_cls_dn_1: 0.2036, loss_box_dn_1: 0.8247, loss_cls_dn_2: 0.2356, loss_box_dn_2: 0.8055, loss_cls_dn_3: 0.2537, loss_box_dn_3: 0.7994, loss_cls_dn_4: 0.2685, loss_box_dn_4: 0.8104, loss_cls_dn_5: 0.2726, loss_box_dn_5: 0.8266, loss_dense_depth: 0.8144, loss: 29.8662, grad_norm: 46.7874 -2025-11-12 14:33:06,790 - mmdet - INFO - Iter [95/17500] lr: 1.376e-04, eta: 13:33:24, time: 1.568, data_time: 0.076, memory: 49164, loss_cls_0: 0.9285, loss_box_0: 1.8169, loss_cns_0: 0.6207, loss_yns_0: 0.1523, loss_cls_1: 0.9778, loss_box_1: 2.0234, loss_cns_1: 0.6267, loss_yns_1: 0.1546, loss_cls_2: 1.0560, loss_box_2: 1.9932, loss_cns_2: 0.6430, loss_yns_2: 0.1565, loss_cls_3: 1.0341, loss_box_3: 2.0044, loss_cns_3: 0.6487, loss_yns_3: 0.1603, loss_cls_4: 1.0387, loss_box_4: 1.9726, loss_cns_4: 0.6538, loss_yns_4: 0.1564, loss_cls_5: 1.0690, loss_box_5: 1.9535, loss_cns_5: 0.6554, loss_yns_5: 0.1571, loss_cls_dn_0: 0.2752, loss_box_dn_0: 0.8098, loss_cls_dn_1: 0.1954, loss_box_dn_1: 0.8186, loss_cls_dn_2: 0.2196, loss_box_dn_2: 0.8019, loss_cls_dn_3: 0.2259, loss_box_dn_3: 0.8064, loss_cls_dn_4: 0.2307, loss_box_dn_4: 0.8083, loss_cls_dn_5: 0.2420, loss_box_dn_5: 0.8156, loss_dense_depth: 0.8652, loss: 29.7684, grad_norm: 59.2594 -2025-11-12 14:33:08,349 - mmdet - INFO - Iter [96/17500] lr: 1.380e-04, eta: 13:29:36, time: 1.562, data_time: 0.080, memory: 49164, loss_cls_0: 0.9593, loss_box_0: 1.8388, loss_cns_0: 0.6193, loss_yns_0: 0.1556, loss_cls_1: 1.0228, loss_box_1: 2.0130, loss_cns_1: 0.6197, loss_yns_1: 0.1554, loss_cls_2: 1.0319, loss_box_2: 1.9756, loss_cns_2: 0.6388, loss_yns_2: 0.1599, loss_cls_3: 1.1035, loss_box_3: 1.9891, loss_cns_3: 0.6443, loss_yns_3: 0.1574, loss_cls_4: 1.1404, loss_box_4: 1.9616, loss_cns_4: 0.6486, loss_yns_4: 0.1573, loss_cls_5: 1.1019, loss_box_5: 1.9184, loss_cns_5: 0.6491, loss_yns_5: 0.1586, loss_cls_dn_0: 0.2787, loss_box_dn_0: 0.8223, loss_cls_dn_1: 0.1946, loss_box_dn_1: 0.8099, loss_cls_dn_2: 0.2131, loss_box_dn_2: 0.7877, loss_cls_dn_3: 0.2189, loss_box_dn_3: 0.7918, loss_cls_dn_4: 0.2232, loss_box_dn_4: 0.7931, loss_cls_dn_5: 0.2351, loss_box_dn_5: 0.7909, loss_dense_depth: 0.8526, loss: 29.8322, grad_norm: 52.2382 -2025-11-12 14:33:09,940 - mmdet - INFO - Iter [97/17500] lr: 1.384e-04, eta: 13:25:57, time: 1.586, data_time: 0.077, memory: 49164, loss_cls_0: 0.9404, loss_box_0: 1.8039, loss_cns_0: 0.6200, loss_yns_0: 0.1544, loss_cls_1: 0.9890, loss_box_1: 1.9491, loss_cns_1: 0.6275, loss_yns_1: 0.1577, loss_cls_2: 1.0523, loss_box_2: 1.8707, loss_cns_2: 0.6457, loss_yns_2: 0.1645, loss_cls_3: 1.0672, loss_box_3: 1.8783, loss_cns_3: 0.6520, loss_yns_3: 0.1627, loss_cls_4: 1.0923, loss_box_4: 1.8955, loss_cns_4: 0.6509, loss_yns_4: 0.1576, loss_cls_5: 1.0637, loss_box_5: 1.8848, loss_cns_5: 0.6512, loss_yns_5: 0.1650, loss_cls_dn_0: 0.2902, loss_box_dn_0: 0.8181, loss_cls_dn_1: 0.1938, loss_box_dn_1: 0.7782, loss_cls_dn_2: 0.2088, loss_box_dn_2: 0.7488, loss_cls_dn_3: 0.2151, loss_box_dn_3: 0.7533, loss_cls_dn_4: 0.2206, loss_box_dn_4: 0.7681, loss_cls_dn_5: 0.2342, loss_box_dn_5: 0.7808, loss_dense_depth: 0.8393, loss: 29.1457, grad_norm: 42.7134 -2025-11-12 14:33:11,511 - mmdet - INFO - Iter [98/17500] lr: 1.388e-04, eta: 13:22:20, time: 1.569, data_time: 0.080, memory: 49164, loss_cls_0: 0.9312, loss_box_0: 1.7852, loss_cns_0: 0.6219, loss_yns_0: 0.1524, loss_cls_1: 0.9811, loss_box_1: 1.9133, loss_cns_1: 0.6248, loss_yns_1: 0.1551, loss_cls_2: 1.0250, loss_box_2: 1.8402, loss_cns_2: 0.6510, loss_yns_2: 0.1600, loss_cls_3: 1.0414, loss_box_3: 1.8233, loss_cns_3: 0.6583, loss_yns_3: 0.1677, loss_cls_4: 1.0442, loss_box_4: 1.8667, loss_cns_4: 0.6567, loss_yns_4: 0.1566, loss_cls_5: 1.0625, loss_box_5: 1.8603, loss_cns_5: 0.6548, loss_yns_5: 0.1680, loss_cls_dn_0: 0.2891, loss_box_dn_0: 0.8165, loss_cls_dn_1: 0.1961, loss_box_dn_1: 0.7867, loss_cls_dn_2: 0.2114, loss_box_dn_2: 0.7574, loss_cls_dn_3: 0.2182, loss_box_dn_3: 0.7664, loss_cls_dn_4: 0.2152, loss_box_dn_4: 0.7926, loss_cls_dn_5: 0.2321, loss_box_dn_5: 0.8141, loss_dense_depth: 0.8258, loss: 28.9231, grad_norm: 52.8419 -2025-11-12 14:33:13,084 - mmdet - INFO - Iter [99/17500] lr: 1.392e-04, eta: 13:18:47, time: 1.575, data_time: 0.083, memory: 49164, loss_cls_0: 0.9223, loss_box_0: 1.7849, loss_cns_0: 0.6214, loss_yns_0: 0.1532, loss_cls_1: 0.9892, loss_box_1: 1.8970, loss_cns_1: 0.6242, loss_yns_1: 0.1547, loss_cls_2: 1.0288, loss_box_2: 1.8424, loss_cns_2: 0.6510, loss_yns_2: 0.1562, loss_cls_3: 1.0545, loss_box_3: 1.8249, loss_cns_3: 0.6551, loss_yns_3: 0.1646, loss_cls_4: 1.0549, loss_box_4: 1.8460, loss_cns_4: 0.6543, loss_yns_4: 0.1583, loss_cls_5: 1.0731, loss_box_5: 1.8731, loss_cns_5: 0.6497, loss_yns_5: 0.1669, loss_cls_dn_0: 0.2762, loss_box_dn_0: 0.8090, loss_cls_dn_1: 0.1957, loss_box_dn_1: 0.8096, loss_cls_dn_2: 0.2092, loss_box_dn_2: 0.7954, loss_cls_dn_3: 0.2166, loss_box_dn_3: 0.8143, loss_cls_dn_4: 0.2133, loss_box_dn_4: 0.8322, loss_cls_dn_5: 0.2296, loss_box_dn_5: 0.8644, loss_dense_depth: 0.8277, loss: 29.0939, grad_norm: 53.6001 -2025-11-12 14:33:14,657 - mmdet - INFO - Iter [100/17500] lr: 1.396e-04, eta: 13:15:19, time: 1.575, data_time: 0.084, memory: 49164, loss_cls_0: 0.9669, loss_box_0: 1.8406, loss_cns_0: 0.6234, loss_yns_0: 0.1556, loss_cls_1: 0.9796, loss_box_1: 1.9625, loss_cns_1: 0.6175, loss_yns_1: 0.1589, loss_cls_2: 1.0328, loss_box_2: 1.9101, loss_cns_2: 0.6434, loss_yns_2: 0.1585, loss_cls_3: 1.0508, loss_box_3: 1.8863, loss_cns_3: 0.6471, loss_yns_3: 0.1576, loss_cls_4: 1.0672, loss_box_4: 1.8828, loss_cns_4: 0.6500, loss_yns_4: 0.1588, loss_cls_5: 1.0644, loss_box_5: 1.9006, loss_cns_5: 0.6450, loss_yns_5: 0.1587, loss_cls_dn_0: 0.2602, loss_box_dn_0: 0.8182, loss_cls_dn_1: 0.1922, loss_box_dn_1: 0.8164, loss_cls_dn_2: 0.2088, loss_box_dn_2: 0.8000, loss_cls_dn_3: 0.2117, loss_box_dn_3: 0.8135, loss_cls_dn_4: 0.2131, loss_box_dn_4: 0.8254, loss_cls_dn_5: 0.2225, loss_box_dn_5: 0.8467, loss_dense_depth: 0.8265, loss: 29.3741, grad_norm: 37.5787 -2025-11-12 14:33:16,284 - mmdet - INFO - Iter [101/17500] lr: 1.400e-04, eta: 13:12:05, time: 1.631, data_time: 0.122, memory: 49164, loss_cls_0: 0.9253, loss_box_0: 1.8019, loss_cns_0: 0.6196, loss_yns_0: 0.1546, loss_cls_1: 0.9742, loss_box_1: 1.9921, loss_cns_1: 0.6290, loss_yns_1: 0.1592, loss_cls_2: 1.0295, loss_box_2: 1.9544, loss_cns_2: 0.6472, loss_yns_2: 0.1665, loss_cls_3: 1.0431, loss_box_3: 1.9426, loss_cns_3: 0.6514, loss_yns_3: 0.1627, loss_cls_4: 1.0594, loss_box_4: 1.9406, loss_cns_4: 0.6538, loss_yns_4: 0.1598, loss_cls_5: 1.0506, loss_box_5: 1.9439, loss_cns_5: 0.6506, loss_yns_5: 0.1591, loss_cls_dn_0: 0.2670, loss_box_dn_0: 0.8079, loss_cls_dn_1: 0.1912, loss_box_dn_1: 0.8264, loss_cls_dn_2: 0.2082, loss_box_dn_2: 0.8125, loss_cls_dn_3: 0.2122, loss_box_dn_3: 0.8222, loss_cls_dn_4: 0.2162, loss_box_dn_4: 0.8308, loss_cls_dn_5: 0.2280, loss_box_dn_5: 0.8409, loss_dense_depth: 0.8173, loss: 29.5520, grad_norm: 52.5292 -2025-11-12 14:33:17,894 - mmdet - INFO - Iter [102/17500] lr: 1.404e-04, eta: 13:08:51, time: 1.610, data_time: 0.120, memory: 49164, loss_cls_0: 0.9468, loss_box_0: 1.7565, loss_cns_0: 0.6038, loss_yns_0: 0.1528, loss_cls_1: 0.9778, loss_box_1: 1.9626, loss_cns_1: 0.6339, loss_yns_1: 0.1576, loss_cls_2: 1.0148, loss_box_2: 1.9231, loss_cns_2: 0.6478, loss_yns_2: 0.1637, loss_cls_3: 1.0303, loss_box_3: 1.9201, loss_cns_3: 0.6533, loss_yns_3: 0.1660, loss_cls_4: 1.0396, loss_box_4: 1.9074, loss_cns_4: 0.6550, loss_yns_4: 0.1594, loss_cls_5: 1.0435, loss_box_5: 1.9369, loss_cns_5: 0.6521, loss_yns_5: 0.1649, loss_cls_dn_0: 0.2736, loss_box_dn_0: 0.8073, loss_cls_dn_1: 0.1924, loss_box_dn_1: 0.7981, loss_cls_dn_2: 0.2039, loss_box_dn_2: 0.7880, loss_cls_dn_3: 0.2081, loss_box_dn_3: 0.7977, loss_cls_dn_4: 0.2129, loss_box_dn_4: 0.7974, loss_cls_dn_5: 0.2244, loss_box_dn_5: 0.8164, loss_dense_depth: 0.8461, loss: 29.2362, grad_norm: 50.0923 -2025-11-12 14:33:19,461 - mmdet - INFO - Iter [103/17500] lr: 1.408e-04, eta: 13:05:34, time: 1.570, data_time: 0.080, memory: 49164, loss_cls_0: 0.9329, loss_box_0: 1.7449, loss_cns_0: 0.6047, loss_yns_0: 0.1530, loss_cls_1: 0.9919, loss_box_1: 1.9824, loss_cns_1: 0.6336, loss_yns_1: 0.1587, loss_cls_2: 1.0076, loss_box_2: 1.9092, loss_cns_2: 0.6488, loss_yns_2: 0.1600, loss_cls_3: 1.0403, loss_box_3: 1.8954, loss_cns_3: 0.6566, loss_yns_3: 0.1653, loss_cls_4: 1.0315, loss_box_4: 1.8816, loss_cns_4: 0.6549, loss_yns_4: 0.1591, loss_cls_5: 1.0503, loss_box_5: 1.9162, loss_cns_5: 0.6515, loss_yns_5: 0.1690, loss_cls_dn_0: 0.2707, loss_box_dn_0: 0.8026, loss_cls_dn_1: 0.1898, loss_box_dn_1: 0.8210, loss_cls_dn_2: 0.2001, loss_box_dn_2: 0.7966, loss_cls_dn_3: 0.1986, loss_box_dn_3: 0.7922, loss_cls_dn_4: 0.2013, loss_box_dn_4: 0.7894, loss_cls_dn_5: 0.2098, loss_box_dn_5: 0.8075, loss_dense_depth: 0.8170, loss: 29.0961, grad_norm: 36.8505 -2025-11-12 14:33:21,038 - mmdet - INFO - Iter [104/17500] lr: 1.412e-04, eta: 13:02:22, time: 1.577, data_time: 0.074, memory: 49164, loss_cls_0: 0.9209, loss_box_0: 1.8053, loss_cns_0: 0.6216, loss_yns_0: 0.1558, loss_cls_1: 0.9763, loss_box_1: 1.9685, loss_cns_1: 0.6420, loss_yns_1: 0.1585, loss_cls_2: 1.0396, loss_box_2: 1.9110, loss_cns_2: 0.6540, loss_yns_2: 0.1592, loss_cls_3: 1.0264, loss_box_3: 1.9304, loss_cns_3: 0.6602, loss_yns_3: 0.1577, loss_cls_4: 1.0242, loss_box_4: 1.9207, loss_cns_4: 0.6579, loss_yns_4: 0.1580, loss_cls_5: 1.0314, loss_box_5: 1.9100, loss_cns_5: 0.6557, loss_yns_5: 0.1654, loss_cls_dn_0: 0.2591, loss_box_dn_0: 0.8151, loss_cls_dn_1: 0.1859, loss_box_dn_1: 0.8057, loss_cls_dn_2: 0.1930, loss_box_dn_2: 0.7888, loss_cls_dn_3: 0.1917, loss_box_dn_3: 0.7923, loss_cls_dn_4: 0.1955, loss_box_dn_4: 0.8004, loss_cls_dn_5: 0.2025, loss_box_dn_5: 0.7962, loss_dense_depth: 0.8122, loss: 29.1491, grad_norm: 47.8251 -2025-11-12 14:33:22,637 - mmdet - INFO - Iter [105/17500] lr: 1.416e-04, eta: 12:59:17, time: 1.599, data_time: 0.080, memory: 49164, loss_cls_0: 0.9621, loss_box_0: 1.7833, loss_cns_0: 0.6251, loss_yns_0: 0.1560, loss_cls_1: 0.9679, loss_box_1: 1.9464, loss_cns_1: 0.6421, loss_yns_1: 0.1574, loss_cls_2: 1.0226, loss_box_2: 1.8902, loss_cns_2: 0.6553, loss_yns_2: 0.1584, loss_cls_3: 1.0195, loss_box_3: 1.9166, loss_cns_3: 0.6578, loss_yns_3: 0.1597, loss_cls_4: 1.0246, loss_box_4: 1.8969, loss_cns_4: 0.6593, loss_yns_4: 0.1570, loss_cls_5: 1.0263, loss_box_5: 1.8986, loss_cns_5: 0.6557, loss_yns_5: 0.1582, loss_cls_dn_0: 0.2472, loss_box_dn_0: 0.8174, loss_cls_dn_1: 0.1844, loss_box_dn_1: 0.8311, loss_cls_dn_2: 0.1915, loss_box_dn_2: 0.8146, loss_cls_dn_3: 0.1899, loss_box_dn_3: 0.8254, loss_cls_dn_4: 0.1960, loss_box_dn_4: 0.8289, loss_cls_dn_5: 0.2064, loss_box_dn_5: 0.8361, loss_dense_depth: 0.7886, loss: 29.1546, grad_norm: 53.0782 -2025-11-12 14:33:24,263 - mmdet - INFO - Iter [106/17500] lr: 1.420e-04, eta: 12:56:18, time: 1.615, data_time: 0.109, memory: 49164, loss_cls_0: 0.8888, loss_box_0: 1.7640, loss_cns_0: 0.6216, loss_yns_0: 0.1544, loss_cls_1: 0.9766, loss_box_1: 1.9251, loss_cns_1: 0.6413, loss_yns_1: 0.1593, loss_cls_2: 1.0055, loss_box_2: 1.8738, loss_cns_2: 0.6515, loss_yns_2: 0.1589, loss_cls_3: 1.0210, loss_box_3: 1.8747, loss_cns_3: 0.6574, loss_yns_3: 0.1614, loss_cls_4: 1.0274, loss_box_4: 1.8730, loss_cns_4: 0.6574, loss_yns_4: 0.1577, loss_cls_5: 1.0426, loss_box_5: 1.8598, loss_cns_5: 0.6549, loss_yns_5: 0.1599, loss_cls_dn_0: 0.2513, loss_box_dn_0: 0.7939, loss_cls_dn_1: 0.1824, loss_box_dn_1: 0.8215, loss_cls_dn_2: 0.1897, loss_box_dn_2: 0.8085, loss_cls_dn_3: 0.1922, loss_box_dn_3: 0.8242, loss_cls_dn_4: 0.2004, loss_box_dn_4: 0.8404, loss_cls_dn_5: 0.2216, loss_box_dn_5: 0.8611, loss_dense_depth: 0.7954, loss: 28.9507, grad_norm: 45.3279 -2025-11-12 14:33:25,848 - mmdet - INFO - Iter [107/17500] lr: 1.424e-04, eta: 12:53:20, time: 1.596, data_time: 0.087, memory: 49164, loss_cls_0: 0.9105, loss_box_0: 1.7258, loss_cns_0: 0.6109, loss_yns_0: 0.1527, loss_cls_1: 0.9632, loss_box_1: 1.9085, loss_cns_1: 0.6329, loss_yns_1: 0.1580, loss_cls_2: 0.9938, loss_box_2: 1.8552, loss_cns_2: 0.6455, loss_yns_2: 0.1576, loss_cls_3: 1.0152, loss_box_3: 1.8610, loss_cns_3: 0.6504, loss_yns_3: 0.1604, loss_cls_4: 1.0217, loss_box_4: 1.8626, loss_cns_4: 0.6488, loss_yns_4: 0.1583, loss_cls_5: 1.0411, loss_box_5: 1.8458, loss_cns_5: 0.6480, loss_yns_5: 0.1644, loss_cls_dn_0: 0.2592, loss_box_dn_0: 0.8115, loss_cls_dn_1: 0.1864, loss_box_dn_1: 0.8159, loss_cls_dn_2: 0.1933, loss_box_dn_2: 0.8035, loss_cls_dn_3: 0.2037, loss_box_dn_3: 0.8253, loss_cls_dn_4: 0.2099, loss_box_dn_4: 0.8499, loss_cls_dn_5: 0.2419, loss_box_dn_5: 0.8668, loss_dense_depth: 0.7906, loss: 28.8503, grad_norm: 48.8118 -2025-11-12 14:33:27,408 - mmdet - INFO - Iter [108/17500] lr: 1.428e-04, eta: 12:50:19, time: 1.559, data_time: 0.075, memory: 49164, loss_cls_0: 0.9021, loss_box_0: 1.7588, loss_cns_0: 0.6175, loss_yns_0: 0.1554, loss_cls_1: 0.9542, loss_box_1: 1.9175, loss_cns_1: 0.6306, loss_yns_1: 0.1602, loss_cls_2: 0.9978, loss_box_2: 1.8396, loss_cns_2: 0.6503, loss_yns_2: 0.1578, loss_cls_3: 1.0145, loss_box_3: 1.8584, loss_cns_3: 0.6570, loss_yns_3: 0.1604, loss_cls_4: 1.0227, loss_box_4: 1.8261, loss_cns_4: 0.6570, loss_yns_4: 0.1598, loss_cls_5: 1.0227, loss_box_5: 1.8344, loss_cns_5: 0.6512, loss_yns_5: 0.1658, loss_cls_dn_0: 0.2587, loss_box_dn_0: 0.8092, loss_cls_dn_1: 0.1846, loss_box_dn_1: 0.8065, loss_cls_dn_2: 0.1915, loss_box_dn_2: 0.7922, loss_cls_dn_3: 0.2034, loss_box_dn_3: 0.8154, loss_cls_dn_4: 0.2076, loss_box_dn_4: 0.8231, loss_cls_dn_5: 0.2297, loss_box_dn_5: 0.8398, loss_dense_depth: 0.7690, loss: 28.7027, grad_norm: 52.2358 -2025-11-12 14:33:29,002 - mmdet - INFO - Iter [109/17500] lr: 1.432e-04, eta: 12:47:26, time: 1.595, data_time: 0.082, memory: 49164, loss_cls_0: 0.9170, loss_box_0: 1.7874, loss_cns_0: 0.6158, loss_yns_0: 0.1564, loss_cls_1: 0.9588, loss_box_1: 1.9063, loss_cns_1: 0.6327, loss_yns_1: 0.1585, loss_cls_2: 1.0125, loss_box_2: 1.8641, loss_cns_2: 0.6480, loss_yns_2: 0.1576, loss_cls_3: 1.0072, loss_box_3: 1.8859, loss_cns_3: 0.6518, loss_yns_3: 0.1602, loss_cls_4: 1.0132, loss_box_4: 1.8313, loss_cns_4: 0.6531, loss_yns_4: 0.1586, loss_cls_5: 1.0131, loss_box_5: 1.8321, loss_cns_5: 0.6502, loss_yns_5: 0.1592, loss_cls_dn_0: 0.2526, loss_box_dn_0: 0.8090, loss_cls_dn_1: 0.1786, loss_box_dn_1: 0.7967, loss_cls_dn_2: 0.1855, loss_box_dn_2: 0.7895, loss_cls_dn_3: 0.1944, loss_box_dn_3: 0.7983, loss_cls_dn_4: 0.1986, loss_box_dn_4: 0.7887, loss_cls_dn_5: 0.2094, loss_box_dn_5: 0.7960, loss_dense_depth: 0.7868, loss: 28.6150, grad_norm: 43.4809 -2025-11-12 14:33:30,586 - mmdet - INFO - Iter [110/17500] lr: 1.436e-04, eta: 12:44:36, time: 1.584, data_time: 0.083, memory: 49164, loss_cls_0: 0.9363, loss_box_0: 1.8323, loss_cns_0: 0.6157, loss_yns_0: 0.1573, loss_cls_1: 0.9658, loss_box_1: 1.9348, loss_cns_1: 0.6279, loss_yns_1: 0.1582, loss_cls_2: 1.0335, loss_box_2: 1.8721, loss_cns_2: 0.6428, loss_yns_2: 0.1576, loss_cls_3: 1.0411, loss_box_3: 1.8861, loss_cns_3: 0.6458, loss_yns_3: 0.1600, loss_cls_4: 1.0198, loss_box_4: 1.8761, loss_cns_4: 0.6497, loss_yns_4: 0.1577, loss_cls_5: 1.0499, loss_box_5: 1.8754, loss_cns_5: 0.6465, loss_yns_5: 0.1573, loss_cls_dn_0: 0.2499, loss_box_dn_0: 0.8129, loss_cls_dn_1: 0.1812, loss_box_dn_1: 0.7700, loss_cls_dn_2: 0.1871, loss_box_dn_2: 0.7537, loss_cls_dn_3: 0.1893, loss_box_dn_3: 0.7562, loss_cls_dn_4: 0.1939, loss_box_dn_4: 0.7594, loss_cls_dn_5: 0.2013, loss_box_dn_5: 0.7695, loss_dense_depth: 0.7966, loss: 28.7204, grad_norm: 41.7115 -2025-11-12 14:33:32,166 - mmdet - INFO - Iter [111/17500] lr: 1.440e-04, eta: 12:41:46, time: 1.572, data_time: 0.072, memory: 49164, loss_cls_0: 0.8960, loss_box_0: 1.8094, loss_cns_0: 0.6172, loss_yns_0: 0.1549, loss_cls_1: 0.9659, loss_box_1: 1.9353, loss_cns_1: 0.6306, loss_yns_1: 0.1575, loss_cls_2: 1.0045, loss_box_2: 1.8796, loss_cns_2: 0.6416, loss_yns_2: 0.1574, loss_cls_3: 1.0159, loss_box_3: 1.8962, loss_cns_3: 0.6461, loss_yns_3: 0.1589, loss_cls_4: 1.0178, loss_box_4: 1.9046, loss_cns_4: 0.6482, loss_yns_4: 0.1571, loss_cls_5: 1.0435, loss_box_5: 1.8909, loss_cns_5: 0.6439, loss_yns_5: 0.1570, loss_cls_dn_0: 0.2540, loss_box_dn_0: 0.8104, loss_cls_dn_1: 0.1815, loss_box_dn_1: 0.7591, loss_cls_dn_2: 0.1888, loss_box_dn_2: 0.7445, loss_cls_dn_3: 0.1900, loss_box_dn_3: 0.7494, loss_cls_dn_4: 0.1928, loss_box_dn_4: 0.7603, loss_cls_dn_5: 0.2027, loss_box_dn_5: 0.7676, loss_dense_depth: 0.8287, loss: 28.6600, grad_norm: 41.5905 -2025-11-12 14:33:33,725 - mmdet - INFO - Iter [112/17500] lr: 1.444e-04, eta: 12:38:58, time: 1.563, data_time: 0.078, memory: 49164, loss_cls_0: 0.9407, loss_box_0: 1.7976, loss_cns_0: 0.6091, loss_yns_0: 0.1554, loss_cls_1: 0.9887, loss_box_1: 1.9836, loss_cns_1: 0.6219, loss_yns_1: 0.1572, loss_cls_2: 1.0699, loss_box_2: 1.8878, loss_cns_2: 0.6366, loss_yns_2: 0.1570, loss_cls_3: 1.0302, loss_box_3: 1.9057, loss_cns_3: 0.6435, loss_yns_3: 0.1589, loss_cls_4: 1.0193, loss_box_4: 1.9091, loss_cns_4: 0.6424, loss_yns_4: 0.1589, loss_cls_5: 1.0295, loss_box_5: 1.9113, loss_cns_5: 0.6381, loss_yns_5: 0.1580, loss_cls_dn_0: 0.2689, loss_box_dn_0: 0.7997, loss_cls_dn_1: 0.1799, loss_box_dn_1: 0.7688, loss_cls_dn_2: 0.1925, loss_box_dn_2: 0.7513, loss_cls_dn_3: 0.1898, loss_box_dn_3: 0.7594, loss_cls_dn_4: 0.1895, loss_box_dn_4: 0.7704, loss_cls_dn_5: 0.2024, loss_box_dn_5: 0.7801, loss_dense_depth: 0.8589, loss: 28.9217, grad_norm: 41.7423 -2025-11-12 14:33:35,311 - mmdet - INFO - Iter [113/17500] lr: 1.448e-04, eta: 12:36:16, time: 1.585, data_time: 0.073, memory: 49164, loss_cls_0: 0.8868, loss_box_0: 1.7783, loss_cns_0: 0.6190, loss_yns_0: 0.1528, loss_cls_1: 0.9297, loss_box_1: 1.9382, loss_cns_1: 0.6253, loss_yns_1: 0.1535, loss_cls_2: 1.0175, loss_box_2: 1.8453, loss_cns_2: 0.6424, loss_yns_2: 0.1543, loss_cls_3: 0.9905, loss_box_3: 1.8477, loss_cns_3: 0.6452, loss_yns_3: 0.1545, loss_cls_4: 0.9919, loss_box_4: 1.8570, loss_cns_4: 0.6455, loss_yns_4: 0.1544, loss_cls_5: 1.0015, loss_box_5: 1.8631, loss_cns_5: 0.6452, loss_yns_5: 0.1535, loss_cls_dn_0: 0.2501, loss_box_dn_0: 0.8012, loss_cls_dn_1: 0.1741, loss_box_dn_1: 0.7966, loss_cls_dn_2: 0.1865, loss_box_dn_2: 0.7740, loss_cls_dn_3: 0.1851, loss_box_dn_3: 0.7866, loss_cls_dn_4: 0.1853, loss_box_dn_4: 0.7986, loss_cls_dn_5: 0.2013, loss_box_dn_5: 0.8154, loss_dense_depth: 0.8100, loss: 28.4578, grad_norm: 40.0427 -2025-11-12 14:33:36,890 - mmdet - INFO - Iter [114/17500] lr: 1.452e-04, eta: 12:33:35, time: 1.572, data_time: 0.077, memory: 49164, loss_cls_0: 0.9086, loss_box_0: 1.7899, loss_cns_0: 0.6211, loss_yns_0: 0.1549, loss_cls_1: 0.9731, loss_box_1: 1.9022, loss_cns_1: 0.6363, loss_yns_1: 0.1533, loss_cls_2: 0.9856, loss_box_2: 1.8594, loss_cns_2: 0.6449, loss_yns_2: 0.1552, loss_cls_3: 1.0012, loss_box_3: 1.8664, loss_cns_3: 0.6461, loss_yns_3: 0.1551, loss_cls_4: 1.0093, loss_box_4: 1.8649, loss_cns_4: 0.6495, loss_yns_4: 0.1545, loss_cls_5: 1.0019, loss_box_5: 1.8636, loss_cns_5: 0.6473, loss_yns_5: 0.1538, loss_cls_dn_0: 0.2509, loss_box_dn_0: 0.7930, loss_cls_dn_1: 0.1739, loss_box_dn_1: 0.7837, loss_cls_dn_2: 0.1799, loss_box_dn_2: 0.7734, loss_cls_dn_3: 0.1858, loss_box_dn_3: 0.7862, loss_cls_dn_4: 0.1903, loss_box_dn_4: 0.7918, loss_cls_dn_5: 0.2027, loss_box_dn_5: 0.8067, loss_dense_depth: 0.8167, loss: 28.5334, grad_norm: 50.6369 -2025-11-12 14:33:38,469 - mmdet - INFO - Iter [115/17500] lr: 1.456e-04, eta: 12:30:59, time: 1.584, data_time: 0.079, memory: 49164, loss_cls_0: 0.8836, loss_box_0: 1.7635, loss_cns_0: 0.6228, loss_yns_0: 0.1527, loss_cls_1: 0.9737, loss_box_1: 1.8901, loss_cns_1: 0.6386, loss_yns_1: 0.1531, loss_cls_2: 1.0245, loss_box_2: 1.8480, loss_cns_2: 0.6489, loss_yns_2: 0.1519, loss_cls_3: 1.0173, loss_box_3: 1.8415, loss_cns_3: 0.6535, loss_yns_3: 0.1536, loss_cls_4: 1.0039, loss_box_4: 1.8503, loss_cns_4: 0.6558, loss_yns_4: 0.1540, loss_cls_5: 1.0135, loss_box_5: 1.8332, loss_cns_5: 0.6528, loss_yns_5: 0.1533, loss_cls_dn_0: 0.2422, loss_box_dn_0: 0.7922, loss_cls_dn_1: 0.1738, loss_box_dn_1: 0.7702, loss_cls_dn_2: 0.1803, loss_box_dn_2: 0.7567, loss_cls_dn_3: 0.1837, loss_box_dn_3: 0.7572, loss_cls_dn_4: 0.1863, loss_box_dn_4: 0.7621, loss_cls_dn_5: 0.1946, loss_box_dn_5: 0.7687, loss_dense_depth: 0.7712, loss: 28.2731, grad_norm: 38.9942 -2025-11-12 14:33:40,033 - mmdet - INFO - Iter [116/17500] lr: 1.460e-04, eta: 12:28:23, time: 1.566, data_time: 0.073, memory: 49164, loss_cls_0: 0.8928, loss_box_0: 1.7651, loss_cns_0: 0.6145, loss_yns_0: 0.1507, loss_cls_1: 0.9760, loss_box_1: 1.9119, loss_cns_1: 0.6375, loss_yns_1: 0.1535, loss_cls_2: 0.9953, loss_box_2: 1.8849, loss_cns_2: 0.6490, loss_yns_2: 0.1560, loss_cls_3: 1.0305, loss_box_3: 1.8870, loss_cns_3: 0.6512, loss_yns_3: 0.1557, loss_cls_4: 1.0490, loss_box_4: 1.9049, loss_cns_4: 0.6454, loss_yns_4: 0.1541, loss_cls_5: 1.0398, loss_box_5: 1.8861, loss_cns_5: 0.6462, loss_yns_5: 0.1553, loss_cls_dn_0: 0.2507, loss_box_dn_0: 0.7932, loss_cls_dn_1: 0.1720, loss_box_dn_1: 0.7468, loss_cls_dn_2: 0.1780, loss_box_dn_2: 0.7335, loss_cls_dn_3: 0.1858, loss_box_dn_3: 0.7364, loss_cls_dn_4: 0.1882, loss_box_dn_4: 0.7580, loss_cls_dn_5: 0.1968, loss_box_dn_5: 0.7499, loss_dense_depth: 0.8306, loss: 28.5122, grad_norm: 50.6243 -2025-11-12 14:33:41,601 - mmdet - INFO - Iter [117/17500] lr: 1.464e-04, eta: 12:25:49, time: 1.566, data_time: 0.073, memory: 49164, loss_cls_0: 0.8845, loss_box_0: 1.7109, loss_cns_0: 0.6143, loss_yns_0: 0.1482, loss_cls_1: 0.9710, loss_box_1: 1.8656, loss_cns_1: 0.6434, loss_yns_1: 0.1543, loss_cls_2: 0.9777, loss_box_2: 1.8388, loss_cns_2: 0.6519, loss_yns_2: 0.1553, loss_cls_3: 1.0159, loss_box_3: 1.8448, loss_cns_3: 0.6510, loss_yns_3: 0.1534, loss_cls_4: 1.0155, loss_box_4: 1.8345, loss_cns_4: 0.6499, loss_yns_4: 0.1521, loss_cls_5: 1.0127, loss_box_5: 1.8435, loss_cns_5: 0.6512, loss_yns_5: 0.1536, loss_cls_dn_0: 0.2461, loss_box_dn_0: 0.7916, loss_cls_dn_1: 0.1714, loss_box_dn_1: 0.7391, loss_cls_dn_2: 0.1772, loss_box_dn_2: 0.7271, loss_cls_dn_3: 0.1857, loss_box_dn_3: 0.7338, loss_cls_dn_4: 0.1861, loss_box_dn_4: 0.7446, loss_cls_dn_5: 0.1921, loss_box_dn_5: 0.7478, loss_dense_depth: 0.8462, loss: 28.0829, grad_norm: 41.8952 -2025-11-12 14:33:43,182 - mmdet - INFO - Iter [118/17500] lr: 1.468e-04, eta: 12:23:21, time: 1.583, data_time: 0.079, memory: 49164, loss_cls_0: 0.8933, loss_box_0: 1.7423, loss_cns_0: 0.6185, loss_yns_0: 0.1509, loss_cls_1: 0.9702, loss_box_1: 1.8480, loss_cns_1: 0.6421, loss_yns_1: 0.1534, loss_cls_2: 1.0041, loss_box_2: 1.7817, loss_cns_2: 0.6483, loss_yns_2: 0.1529, loss_cls_3: 1.0136, loss_box_3: 1.7822, loss_cns_3: 0.6531, loss_yns_3: 0.1526, loss_cls_4: 1.0185, loss_box_4: 1.7732, loss_cns_4: 0.6525, loss_yns_4: 0.1523, loss_cls_5: 0.9982, loss_box_5: 1.8331, loss_cns_5: 0.6468, loss_yns_5: 0.1531, loss_cls_dn_0: 0.2428, loss_box_dn_0: 0.7916, loss_cls_dn_1: 0.1709, loss_box_dn_1: 0.7529, loss_cls_dn_2: 0.1786, loss_box_dn_2: 0.7370, loss_cls_dn_3: 0.1847, loss_box_dn_3: 0.7460, loss_cls_dn_4: 0.1911, loss_box_dn_4: 0.7579, loss_cls_dn_5: 0.1928, loss_box_dn_5: 0.7894, loss_dense_depth: 0.9006, loss: 28.0712, grad_norm: 51.1715 -2025-11-12 14:33:44,757 - mmdet - INFO - Iter [119/17500] lr: 1.472e-04, eta: 12:20:53, time: 1.573, data_time: 0.073, memory: 49164, loss_cls_0: 0.8990, loss_box_0: 1.7559, loss_cns_0: 0.6174, loss_yns_0: 0.1504, loss_cls_1: 0.9876, loss_box_1: 1.8923, loss_cns_1: 0.6400, loss_yns_1: 0.1540, loss_cls_2: 1.0238, loss_box_2: 1.8153, loss_cns_2: 0.6475, loss_yns_2: 0.1551, loss_cls_3: 1.0515, loss_box_3: 1.8080, loss_cns_3: 0.6525, loss_yns_3: 0.1540, loss_cls_4: 1.0505, loss_box_4: 1.8107, loss_cns_4: 0.6519, loss_yns_4: 0.1549, loss_cls_5: 1.0263, loss_box_5: 1.8576, loss_cns_5: 0.6455, loss_yns_5: 0.1514, loss_cls_dn_0: 0.2369, loss_box_dn_0: 0.7825, loss_cls_dn_1: 0.1690, loss_box_dn_1: 0.7547, loss_cls_dn_2: 0.1730, loss_box_dn_2: 0.7354, loss_cls_dn_3: 0.1837, loss_box_dn_3: 0.7375, loss_cls_dn_4: 0.1838, loss_box_dn_4: 0.7474, loss_cls_dn_5: 0.1871, loss_box_dn_5: 0.7706, loss_dense_depth: 0.8221, loss: 28.2368, grad_norm: 50.2864 -2025-11-12 14:33:46,342 - mmdet - INFO - Iter [120/17500] lr: 1.476e-04, eta: 12:18:29, time: 1.580, data_time: 0.078, memory: 49164, loss_cls_0: 0.8793, loss_box_0: 1.7567, loss_cns_0: 0.6132, loss_yns_0: 0.1511, loss_cls_1: 0.9857, loss_box_1: 1.8212, loss_cns_1: 0.6462, loss_yns_1: 0.1566, loss_cls_2: 1.0124, loss_box_2: 1.8030, loss_cns_2: 0.6491, loss_yns_2: 0.1564, loss_cls_3: 1.0101, loss_box_3: 1.7918, loss_cns_3: 0.6545, loss_yns_3: 0.1558, loss_cls_4: 1.0135, loss_box_4: 1.7572, loss_cns_4: 0.6547, loss_yns_4: 0.1561, loss_cls_5: 1.0114, loss_box_5: 1.7557, loss_cns_5: 0.6526, loss_yns_5: 0.1556, loss_cls_dn_0: 0.2383, loss_box_dn_0: 0.7935, loss_cls_dn_1: 0.1744, loss_box_dn_1: 0.7465, loss_cls_dn_2: 0.1806, loss_box_dn_2: 0.7370, loss_cls_dn_3: 0.1859, loss_box_dn_3: 0.7394, loss_cls_dn_4: 0.1823, loss_box_dn_4: 0.7364, loss_cls_dn_5: 0.1863, loss_box_dn_5: 0.7378, loss_dense_depth: 0.7984, loss: 27.8368, grad_norm: 42.5032 -2025-11-12 14:33:47,986 - mmdet - INFO - Iter [121/17500] lr: 1.480e-04, eta: 12:16:17, time: 1.650, data_time: 0.108, memory: 49164, loss_cls_0: 0.9086, loss_box_0: 1.7500, loss_cns_0: 0.6149, loss_yns_0: 0.1531, loss_cls_1: 1.0150, loss_box_1: 1.8513, loss_cns_1: 0.6491, loss_yns_1: 0.1573, loss_cls_2: 1.0495, loss_box_2: 1.8208, loss_cns_2: 0.6533, loss_yns_2: 0.1537, loss_cls_3: 1.0360, loss_box_3: 1.8448, loss_cns_3: 0.6565, loss_yns_3: 0.1568, loss_cls_4: 1.0450, loss_box_4: 1.8085, loss_cns_4: 0.6565, loss_yns_4: 0.1570, loss_cls_5: 1.0248, loss_box_5: 1.8186, loss_cns_5: 0.6545, loss_yns_5: 0.1536, loss_cls_dn_0: 0.2435, loss_box_dn_0: 0.7966, loss_cls_dn_1: 0.1749, loss_box_dn_1: 0.7542, loss_cls_dn_2: 0.2001, loss_box_dn_2: 0.7475, loss_cls_dn_3: 0.2025, loss_box_dn_3: 0.7555, loss_cls_dn_4: 0.1965, loss_box_dn_4: 0.7514, loss_cls_dn_5: 0.1911, loss_box_dn_5: 0.7599, loss_dense_depth: 0.8235, loss: 28.3861, grad_norm: 53.2575 -2025-11-12 14:33:49,630 - mmdet - INFO - Iter [122/17500] lr: 1.484e-04, eta: 12:14:06, time: 1.639, data_time: 0.109, memory: 49164, loss_cls_0: 0.8948, loss_box_0: 1.7379, loss_cns_0: 0.6239, loss_yns_0: 0.1553, loss_cls_1: 0.9836, loss_box_1: 1.8430, loss_cns_1: 0.6436, loss_yns_1: 0.1571, loss_cls_2: 1.0373, loss_box_2: 1.8102, loss_cns_2: 0.6523, loss_yns_2: 0.1565, loss_cls_3: 1.0208, loss_box_3: 1.8233, loss_cns_3: 0.6516, loss_yns_3: 0.1573, loss_cls_4: 1.0290, loss_box_4: 1.8184, loss_cns_4: 0.6508, loss_yns_4: 0.1585, loss_cls_5: 1.0231, loss_box_5: 1.8114, loss_cns_5: 0.6522, loss_yns_5: 0.1563, loss_cls_dn_0: 0.2424, loss_box_dn_0: 0.7847, loss_cls_dn_1: 0.1700, loss_box_dn_1: 0.7490, loss_cls_dn_2: 0.1975, loss_box_dn_2: 0.7412, loss_cls_dn_3: 0.1902, loss_box_dn_3: 0.7435, loss_cls_dn_4: 0.1962, loss_box_dn_4: 0.7482, loss_cls_dn_5: 0.1970, loss_box_dn_5: 0.7549, loss_dense_depth: 0.7995, loss: 28.1626, grad_norm: 53.5410 -2025-11-12 14:33:51,240 - mmdet - INFO - Iter [123/17500] lr: 1.488e-04, eta: 12:11:54, time: 1.616, data_time: 0.094, memory: 49164, loss_cls_0: 0.9054, loss_box_0: 1.7365, loss_cns_0: 0.6250, loss_yns_0: 0.1552, loss_cls_1: 0.9573, loss_box_1: 1.8536, loss_cns_1: 0.6415, loss_yns_1: 0.1585, loss_cls_2: 1.0073, loss_box_2: 1.7943, loss_cns_2: 0.6550, loss_yns_2: 0.1552, loss_cls_3: 1.0212, loss_box_3: 1.7660, loss_cns_3: 0.6550, loss_yns_3: 0.1573, loss_cls_4: 1.0175, loss_box_4: 1.7643, loss_cns_4: 0.6559, loss_yns_4: 0.1573, loss_cls_5: 0.9986, loss_box_5: 1.7629, loss_cns_5: 0.6518, loss_yns_5: 0.1556, loss_cls_dn_0: 0.2295, loss_box_dn_0: 0.7826, loss_cls_dn_1: 0.1674, loss_box_dn_1: 0.7657, loss_cls_dn_2: 0.1872, loss_box_dn_2: 0.7537, loss_cls_dn_3: 0.1770, loss_box_dn_3: 0.7433, loss_cls_dn_4: 0.1849, loss_box_dn_4: 0.7515, loss_cls_dn_5: 0.1928, loss_box_dn_5: 0.7542, loss_dense_depth: 0.8232, loss: 27.9211, grad_norm: 45.8058 -2025-11-12 14:33:52,843 - mmdet - INFO - Iter [124/17500] lr: 1.492e-04, eta: 12:09:42, time: 1.604, data_time: 0.077, memory: 49164, loss_cls_0: 0.9021, loss_box_0: 1.7447, loss_cns_0: 0.6247, loss_yns_0: 0.1565, loss_cls_1: 1.0232, loss_box_1: 1.8296, loss_cns_1: 0.6443, loss_yns_1: 0.1579, loss_cls_2: 1.0174, loss_box_2: 1.7691, loss_cns_2: 0.6534, loss_yns_2: 0.1551, loss_cls_3: 1.0513, loss_box_3: 1.7586, loss_cns_3: 0.6553, loss_yns_3: 0.1567, loss_cls_4: 1.0486, loss_box_4: 1.7558, loss_cns_4: 0.6574, loss_yns_4: 0.1589, loss_cls_5: 1.0076, loss_box_5: 1.7730, loss_cns_5: 0.6521, loss_yns_5: 0.1574, loss_cls_dn_0: 0.2300, loss_box_dn_0: 0.7869, loss_cls_dn_1: 0.1629, loss_box_dn_1: 0.7516, loss_cls_dn_2: 0.1706, loss_box_dn_2: 0.7343, loss_cls_dn_3: 0.1767, loss_box_dn_3: 0.7321, loss_cls_dn_4: 0.1775, loss_box_dn_4: 0.7362, loss_cls_dn_5: 0.1857, loss_box_dn_5: 0.7461, loss_dense_depth: 0.7817, loss: 27.8828, grad_norm: 55.7699 -2025-11-12 14:33:54,466 - mmdet - INFO - Iter [125/17500] lr: 1.496e-04, eta: 12:07:34, time: 1.622, data_time: 0.076, memory: 49164, loss_cls_0: 0.9098, loss_box_0: 1.7369, loss_cns_0: 0.6143, loss_yns_0: 0.1562, loss_cls_1: 1.0039, loss_box_1: 1.8338, loss_cns_1: 0.6470, loss_yns_1: 0.1622, loss_cls_2: 1.0333, loss_box_2: 1.7920, loss_cns_2: 0.6591, loss_yns_2: 0.1576, loss_cls_3: 1.0518, loss_box_3: 1.8048, loss_cns_3: 0.6562, loss_yns_3: 0.1569, loss_cls_4: 1.0386, loss_box_4: 1.8162, loss_cns_4: 0.6582, loss_yns_4: 0.1580, loss_cls_5: 1.0251, loss_box_5: 1.8106, loss_cns_5: 0.6548, loss_yns_5: 0.1578, loss_cls_dn_0: 0.2398, loss_box_dn_0: 0.7808, loss_cls_dn_1: 0.1661, loss_box_dn_1: 0.7439, loss_cls_dn_2: 0.1710, loss_box_dn_2: 0.7290, loss_cls_dn_3: 0.1802, loss_box_dn_3: 0.7383, loss_cls_dn_4: 0.1802, loss_box_dn_4: 0.7438, loss_cls_dn_5: 0.1894, loss_box_dn_5: 0.7434, loss_dense_depth: 0.7911, loss: 28.0921, grad_norm: 55.9435 -2025-11-12 14:33:56,057 - mmdet - INFO - Iter [126/17500] lr: 1.500e-04, eta: 12:05:24, time: 1.589, data_time: 0.106, memory: 49164, loss_cls_0: 0.9334, loss_box_0: 1.6921, loss_cns_0: 0.6101, loss_yns_0: 0.1537, loss_cls_1: 0.9751, loss_box_1: 1.9254, loss_cns_1: 0.6387, loss_yns_1: 0.1570, loss_cls_2: 1.0259, loss_box_2: 1.8643, loss_cns_2: 0.6515, loss_yns_2: 0.1590, loss_cls_3: 1.0399, loss_box_3: 1.8548, loss_cns_3: 0.6497, loss_yns_3: 0.1570, loss_cls_4: 1.0225, loss_box_4: 1.8640, loss_cns_4: 0.6525, loss_yns_4: 0.1561, loss_cls_5: 1.0163, loss_box_5: 1.8498, loss_cns_5: 0.6487, loss_yns_5: 0.1560, loss_cls_dn_0: 0.2427, loss_box_dn_0: 0.7848, loss_cls_dn_1: 0.1710, loss_box_dn_1: 0.7490, loss_cls_dn_2: 0.1755, loss_box_dn_2: 0.7296, loss_cls_dn_3: 0.1815, loss_box_dn_3: 0.7311, loss_cls_dn_4: 0.1822, loss_box_dn_4: 0.7378, loss_cls_dn_5: 0.1897, loss_box_dn_5: 0.7357, loss_dense_depth: 0.7664, loss: 28.2303, grad_norm: 47.5023 -2025-11-12 14:33:57,644 - mmdet - INFO - Iter [127/17500] lr: 1.504e-04, eta: 12:03:16, time: 1.585, data_time: 0.076, memory: 49164, loss_cls_0: 0.9085, loss_box_0: 1.7186, loss_cns_0: 0.6136, loss_yns_0: 0.1529, loss_cls_1: 0.9817, loss_box_1: 1.8758, loss_cns_1: 0.6413, loss_yns_1: 0.1567, loss_cls_2: 1.0126, loss_box_2: 1.8061, loss_cns_2: 0.6490, loss_yns_2: 0.1541, loss_cls_3: 1.0286, loss_box_3: 1.7915, loss_cns_3: 0.6526, loss_yns_3: 0.1547, loss_cls_4: 1.0152, loss_box_4: 1.7784, loss_cns_4: 0.6544, loss_yns_4: 0.1573, loss_cls_5: 1.0024, loss_box_5: 1.7800, loss_cns_5: 0.6520, loss_yns_5: 0.1543, loss_cls_dn_0: 0.2396, loss_box_dn_0: 0.7814, loss_cls_dn_1: 0.1664, loss_box_dn_1: 0.7571, loss_cls_dn_2: 0.1726, loss_box_dn_2: 0.7398, loss_cls_dn_3: 0.1779, loss_box_dn_3: 0.7352, loss_cls_dn_4: 0.1814, loss_box_dn_4: 0.7389, loss_cls_dn_5: 0.1887, loss_box_dn_5: 0.7444, loss_dense_depth: 0.7716, loss: 27.8872, grad_norm: 39.4131 -2025-11-12 14:33:59,212 - mmdet - INFO - Iter [128/17500] lr: 1.508e-04, eta: 12:01:08, time: 1.571, data_time: 0.084, memory: 49164, loss_cls_0: 0.8952, loss_box_0: 1.7317, loss_cns_0: 0.6263, loss_yns_0: 0.1574, loss_cls_1: 0.9694, loss_box_1: 1.8661, loss_cns_1: 0.6371, loss_yns_1: 0.1585, loss_cls_2: 0.9983, loss_box_2: 1.7973, loss_cns_2: 0.6456, loss_yns_2: 0.1558, loss_cls_3: 1.0092, loss_box_3: 1.7951, loss_cns_3: 0.6489, loss_yns_3: 0.1556, loss_cls_4: 1.0125, loss_box_4: 1.7748, loss_cns_4: 0.6499, loss_yns_4: 0.1571, loss_cls_5: 1.0085, loss_box_5: 1.7704, loss_cns_5: 0.6506, loss_yns_5: 0.1567, loss_cls_dn_0: 0.2318, loss_box_dn_0: 0.7797, loss_cls_dn_1: 0.1627, loss_box_dn_1: 0.7508, loss_cls_dn_2: 0.1703, loss_box_dn_2: 0.7475, loss_cls_dn_3: 0.1775, loss_box_dn_3: 0.7470, loss_cls_dn_4: 0.1816, loss_box_dn_4: 0.7516, loss_cls_dn_5: 0.1899, loss_box_dn_5: 0.7557, loss_dense_depth: 0.8189, loss: 27.8928, grad_norm: 49.0904 -2025-11-12 14:34:00,788 - mmdet - INFO - Iter [129/17500] lr: 1.512e-04, eta: 11:59:02, time: 1.579, data_time: 0.084, memory: 49164, loss_cls_0: 0.9209, loss_box_0: 1.7299, loss_cns_0: 0.6238, loss_yns_0: 0.1596, loss_cls_1: 0.9545, loss_box_1: 1.8255, loss_cns_1: 0.6382, loss_yns_1: 0.1562, loss_cls_2: 0.9841, loss_box_2: 1.7603, loss_cns_2: 0.6495, loss_yns_2: 0.1578, loss_cls_3: 0.9989, loss_box_3: 1.7583, loss_cns_3: 0.6519, loss_yns_3: 0.1565, loss_cls_4: 1.0141, loss_box_4: 1.7336, loss_cns_4: 0.6546, loss_yns_4: 0.1577, loss_cls_5: 1.0224, loss_box_5: 1.7446, loss_cns_5: 0.6554, loss_yns_5: 0.1584, loss_cls_dn_0: 0.2249, loss_box_dn_0: 0.7749, loss_cls_dn_1: 0.1650, loss_box_dn_1: 0.7519, loss_cls_dn_2: 0.1716, loss_box_dn_2: 0.7420, loss_cls_dn_3: 0.1776, loss_box_dn_3: 0.7405, loss_cls_dn_4: 0.1779, loss_box_dn_4: 0.7387, loss_cls_dn_5: 0.1866, loss_box_dn_5: 0.7452, loss_dense_depth: 0.7677, loss: 27.6313, grad_norm: 45.1689 -2025-11-12 14:34:02,361 - mmdet - INFO - Iter [130/17500] lr: 1.516e-04, eta: 11:56:58, time: 1.571, data_time: 0.084, memory: 49164, loss_cls_0: 0.8916, loss_box_0: 1.7496, loss_cns_0: 0.6210, loss_yns_0: 0.1567, loss_cls_1: 0.9598, loss_box_1: 1.8103, loss_cns_1: 0.6418, loss_yns_1: 0.1551, loss_cls_2: 0.9906, loss_box_2: 1.7456, loss_cns_2: 0.6537, loss_yns_2: 0.1548, loss_cls_3: 1.0013, loss_box_3: 1.7486, loss_cns_3: 0.6556, loss_yns_3: 0.1543, loss_cls_4: 1.0152, loss_box_4: 1.7253, loss_cns_4: 0.6545, loss_yns_4: 0.1597, loss_cls_5: 1.0012, loss_box_5: 1.7341, loss_cns_5: 0.6575, loss_yns_5: 0.1572, loss_cls_dn_0: 0.2302, loss_box_dn_0: 0.7833, loss_cls_dn_1: 0.1626, loss_box_dn_1: 0.7438, loss_cls_dn_2: 0.1694, loss_box_dn_2: 0.7262, loss_cls_dn_3: 0.1732, loss_box_dn_3: 0.7250, loss_cls_dn_4: 0.1731, loss_box_dn_4: 0.7221, loss_cls_dn_5: 0.1802, loss_box_dn_5: 0.7263, loss_dense_depth: 0.8179, loss: 27.5286, grad_norm: 34.6191 -2025-11-12 14:34:03,988 - mmdet - INFO - Iter [131/17500] lr: 1.520e-04, eta: 11:55:02, time: 1.622, data_time: 0.078, memory: 49164, loss_cls_0: 0.9124, loss_box_0: 1.7210, loss_cns_0: 0.6176, loss_yns_0: 0.1505, loss_cls_1: 0.9615, loss_box_1: 1.8289, loss_cns_1: 0.6450, loss_yns_1: 0.1546, loss_cls_2: 1.0001, loss_box_2: 1.7527, loss_cns_2: 0.6560, loss_yns_2: 0.1546, loss_cls_3: 1.0090, loss_box_3: 1.7511, loss_cns_3: 0.6609, loss_yns_3: 0.1531, loss_cls_4: 1.0267, loss_box_4: 1.7504, loss_cns_4: 0.6589, loss_yns_4: 0.1553, loss_cls_5: 1.0154, loss_box_5: 1.7542, loss_cns_5: 0.6573, loss_yns_5: 0.1542, loss_cls_dn_0: 0.2362, loss_box_dn_0: 0.7853, loss_cls_dn_1: 0.1615, loss_box_dn_1: 0.7343, loss_cls_dn_2: 0.1691, loss_box_dn_2: 0.7143, loss_cls_dn_3: 0.1717, loss_box_dn_3: 0.7181, loss_cls_dn_4: 0.1800, loss_box_dn_4: 0.7259, loss_cls_dn_5: 0.1884, loss_box_dn_5: 0.7376, loss_dense_depth: 0.9143, loss: 27.7383, grad_norm: 41.9918 -2025-11-12 14:34:05,557 - mmdet - INFO - Iter [132/17500] lr: 1.524e-04, eta: 11:53:03, time: 1.579, data_time: 0.080, memory: 49164, loss_cls_0: 0.8958, loss_box_0: 1.7312, loss_cns_0: 0.6154, loss_yns_0: 0.1510, loss_cls_1: 0.9495, loss_box_1: 1.8427, loss_cns_1: 0.6403, loss_yns_1: 0.1521, loss_cls_2: 0.9864, loss_box_2: 1.7691, loss_cns_2: 0.6530, loss_yns_2: 0.1535, loss_cls_3: 0.9944, loss_box_3: 1.7760, loss_cns_3: 0.6517, loss_yns_3: 0.1516, loss_cls_4: 1.0003, loss_box_4: 1.7824, loss_cns_4: 0.6504, loss_yns_4: 0.1524, loss_cls_5: 1.0127, loss_box_5: 1.7829, loss_cns_5: 0.6502, loss_yns_5: 0.1514, loss_cls_dn_0: 0.2387, loss_box_dn_0: 0.7792, loss_cls_dn_1: 0.1693, loss_box_dn_1: 0.7486, loss_cls_dn_2: 0.1762, loss_box_dn_2: 0.7330, loss_cls_dn_3: 0.1781, loss_box_dn_3: 0.7484, loss_cls_dn_4: 0.1894, loss_box_dn_4: 0.7693, loss_cls_dn_5: 0.2000, loss_box_dn_5: 0.7904, loss_dense_depth: 0.8464, loss: 27.8634, grad_norm: 48.6091 -2025-11-12 14:34:07,138 - mmdet - INFO - Iter [133/17500] lr: 1.528e-04, eta: 11:51:05, time: 1.578, data_time: 0.076, memory: 49164, loss_cls_0: 0.8933, loss_box_0: 1.7359, loss_cns_0: 0.6155, loss_yns_0: 0.1530, loss_cls_1: 0.9597, loss_box_1: 1.9109, loss_cns_1: 0.6355, loss_yns_1: 0.1512, loss_cls_2: 0.9831, loss_box_2: 1.8387, loss_cns_2: 0.6467, loss_yns_2: 0.1547, loss_cls_3: 1.0059, loss_box_3: 1.8465, loss_cns_3: 0.6494, loss_yns_3: 0.1548, loss_cls_4: 1.0435, loss_box_4: 1.8390, loss_cns_4: 0.6482, loss_yns_4: 0.1564, loss_cls_5: 1.0295, loss_box_5: 1.8376, loss_cns_5: 0.6492, loss_yns_5: 0.1556, loss_cls_dn_0: 0.2362, loss_box_dn_0: 0.7898, loss_cls_dn_1: 0.1711, loss_box_dn_1: 0.7674, loss_cls_dn_2: 0.1742, loss_box_dn_2: 0.7549, loss_cls_dn_3: 0.1740, loss_box_dn_3: 0.7737, loss_cls_dn_4: 0.1838, loss_box_dn_4: 0.7962, loss_cls_dn_5: 0.1951, loss_box_dn_5: 0.8191, loss_dense_depth: 0.8635, loss: 28.3927, grad_norm: 47.3431 -2025-11-12 14:34:08,717 - mmdet - INFO - Iter [134/17500] lr: 1.532e-04, eta: 11:49:08, time: 1.573, data_time: 0.079, memory: 49164, loss_cls_0: 0.8578, loss_box_0: 1.7219, loss_cns_0: 0.6250, loss_yns_0: 0.1507, loss_cls_1: 0.9488, loss_box_1: 1.8653, loss_cns_1: 0.6414, loss_yns_1: 0.1533, loss_cls_2: 0.9759, loss_box_2: 1.8168, loss_cns_2: 0.6475, loss_yns_2: 0.1528, loss_cls_3: 1.0065, loss_box_3: 1.8233, loss_cns_3: 0.6520, loss_yns_3: 0.1561, loss_cls_4: 1.0280, loss_box_4: 1.8096, loss_cns_4: 0.6525, loss_yns_4: 0.1557, loss_cls_5: 0.9890, loss_box_5: 1.8214, loss_cns_5: 0.6503, loss_yns_5: 0.1548, loss_cls_dn_0: 0.2253, loss_box_dn_0: 0.7893, loss_cls_dn_1: 0.1677, loss_box_dn_1: 0.7831, loss_cls_dn_2: 0.1710, loss_box_dn_2: 0.7755, loss_cls_dn_3: 0.1722, loss_box_dn_3: 0.7874, loss_cls_dn_4: 0.1763, loss_box_dn_4: 0.8046, loss_cls_dn_5: 0.1849, loss_box_dn_5: 0.8276, loss_dense_depth: 0.8375, loss: 28.1583, grad_norm: 45.1178 -2025-11-12 14:34:10,299 - mmdet - INFO - Iter [135/17500] lr: 1.536e-04, eta: 11:47:14, time: 1.587, data_time: 0.082, memory: 49164, loss_cls_0: 0.8786, loss_box_0: 1.7619, loss_cns_0: 0.6226, loss_yns_0: 0.1538, loss_cls_1: 0.9481, loss_box_1: 1.8890, loss_cns_1: 0.6502, loss_yns_1: 0.1543, loss_cls_2: 0.9766, loss_box_2: 1.8414, loss_cns_2: 0.6551, loss_yns_2: 0.1538, loss_cls_3: 0.9933, loss_box_3: 1.8350, loss_cns_3: 0.6552, loss_yns_3: 0.1539, loss_cls_4: 0.9959, loss_box_4: 1.8420, loss_cns_4: 0.6566, loss_yns_4: 0.1574, loss_cls_5: 1.0072, loss_box_5: 1.8472, loss_cns_5: 0.6536, loss_yns_5: 0.1545, loss_cls_dn_0: 0.2375, loss_box_dn_0: 0.7900, loss_cls_dn_1: 0.1636, loss_box_dn_1: 0.7940, loss_cls_dn_2: 0.1681, loss_box_dn_2: 0.7815, loss_cls_dn_3: 0.1766, loss_box_dn_3: 0.7839, loss_cls_dn_4: 0.1784, loss_box_dn_4: 0.7989, loss_cls_dn_5: 0.1977, loss_box_dn_5: 0.8137, loss_dense_depth: 0.8028, loss: 28.3239, grad_norm: 49.6705 -2025-11-12 14:34:11,876 - mmdet - INFO - Iter [136/17500] lr: 1.540e-04, eta: 11:45:21, time: 1.575, data_time: 0.075, memory: 49164, loss_cls_0: 0.8421, loss_box_0: 1.7283, loss_cns_0: 0.6248, loss_yns_0: 0.1493, loss_cls_1: 0.9158, loss_box_1: 1.8454, loss_cns_1: 0.6484, loss_yns_1: 0.1493, loss_cls_2: 0.9497, loss_box_2: 1.7900, loss_cns_2: 0.6570, loss_yns_2: 0.1518, loss_cls_3: 0.9813, loss_box_3: 1.7779, loss_cns_3: 0.6570, loss_yns_3: 0.1490, loss_cls_4: 1.0017, loss_box_4: 1.7801, loss_cns_4: 0.6588, loss_yns_4: 0.1517, loss_cls_5: 0.9802, loss_box_5: 1.7839, loss_cns_5: 0.6570, loss_yns_5: 0.1509, loss_cls_dn_0: 0.2286, loss_box_dn_0: 0.7821, loss_cls_dn_1: 0.1599, loss_box_dn_1: 0.7674, loss_cls_dn_2: 0.1686, loss_box_dn_2: 0.7496, loss_cls_dn_3: 0.1752, loss_box_dn_3: 0.7487, loss_cls_dn_4: 0.1838, loss_box_dn_4: 0.7560, loss_cls_dn_5: 0.1908, loss_box_dn_5: 0.7621, loss_dense_depth: 0.8122, loss: 27.6666, grad_norm: 45.5392 -2025-11-12 14:34:13,447 - mmdet - INFO - Iter [137/17500] lr: 1.544e-04, eta: 11:43:29, time: 1.570, data_time: 0.076, memory: 49164, loss_cls_0: 0.8611, loss_box_0: 1.7293, loss_cns_0: 0.6257, loss_yns_0: 0.1515, loss_cls_1: 0.9249, loss_box_1: 1.8238, loss_cns_1: 0.6461, loss_yns_1: 0.1480, loss_cls_2: 0.9577, loss_box_2: 1.7763, loss_cns_2: 0.6489, loss_yns_2: 0.1498, loss_cls_3: 0.9883, loss_box_3: 1.7722, loss_cns_3: 0.6528, loss_yns_3: 0.1502, loss_cls_4: 1.0100, loss_box_4: 1.7606, loss_cns_4: 0.6525, loss_yns_4: 0.1502, loss_cls_5: 0.9652, loss_box_5: 1.7636, loss_cns_5: 0.6522, loss_yns_5: 0.1492, loss_cls_dn_0: 0.2366, loss_box_dn_0: 0.7836, loss_cls_dn_1: 0.1554, loss_box_dn_1: 0.7355, loss_cls_dn_2: 0.1704, loss_box_dn_2: 0.7205, loss_cls_dn_3: 0.1752, loss_box_dn_3: 0.7210, loss_cls_dn_4: 0.1891, loss_box_dn_4: 0.7207, loss_cls_dn_5: 0.1796, loss_box_dn_5: 0.7203, loss_dense_depth: 0.7877, loss: 27.4057, grad_norm: 36.3864 -2025-11-12 14:34:15,014 - mmdet - INFO - Iter [138/17500] lr: 1.548e-04, eta: 11:41:38, time: 1.573, data_time: 0.081, memory: 49164, loss_cls_0: 0.8592, loss_box_0: 1.7205, loss_cns_0: 0.6260, loss_yns_0: 0.1479, loss_cls_1: 0.9152, loss_box_1: 1.8399, loss_cns_1: 0.6464, loss_yns_1: 0.1507, loss_cls_2: 0.9440, loss_box_2: 1.7776, loss_cns_2: 0.6518, loss_yns_2: 0.1487, loss_cls_3: 0.9633, loss_box_3: 1.7632, loss_cns_3: 0.6554, loss_yns_3: 0.1509, loss_cls_4: 0.9776, loss_box_4: 1.7668, loss_cns_4: 0.6535, loss_yns_4: 0.1516, loss_cls_5: 0.9670, loss_box_5: 1.7818, loss_cns_5: 0.6517, loss_yns_5: 0.1502, loss_cls_dn_0: 0.2322, loss_box_dn_0: 0.7935, loss_cls_dn_1: 0.1493, loss_box_dn_1: 0.7348, loss_cls_dn_2: 0.1615, loss_box_dn_2: 0.7254, loss_cls_dn_3: 0.1610, loss_box_dn_3: 0.7309, loss_cls_dn_4: 0.1754, loss_box_dn_4: 0.7437, loss_cls_dn_5: 0.1686, loss_box_dn_5: 0.7605, loss_dense_depth: 0.8307, loss: 27.4285, grad_norm: 40.9929 -2025-11-12 14:34:16,585 - mmdet - INFO - Iter [139/17500] lr: 1.552e-04, eta: 11:39:48, time: 1.566, data_time: 0.076, memory: 49164, loss_cls_0: 0.8259, loss_box_0: 1.6798, loss_cns_0: 0.6251, loss_yns_0: 0.1452, loss_cls_1: 0.9049, loss_box_1: 1.8062, loss_cns_1: 0.6542, loss_yns_1: 0.1477, loss_cls_2: 0.9500, loss_box_2: 1.7577, loss_cns_2: 0.6539, loss_yns_2: 0.1460, loss_cls_3: 0.9611, loss_box_3: 1.7572, loss_cns_3: 0.6536, loss_yns_3: 0.1496, loss_cls_4: 0.9466, loss_box_4: 1.7655, loss_cns_4: 0.6538, loss_yns_4: 0.1471, loss_cls_5: 0.9669, loss_box_5: 1.7792, loss_cns_5: 0.6511, loss_yns_5: 0.1481, loss_cls_dn_0: 0.2251, loss_box_dn_0: 0.7845, loss_cls_dn_1: 0.1486, loss_box_dn_1: 0.7531, loss_cls_dn_2: 0.1537, loss_box_dn_2: 0.7551, loss_cls_dn_3: 0.1619, loss_box_dn_3: 0.7758, loss_cls_dn_4: 0.1691, loss_box_dn_4: 0.8089, loss_cls_dn_5: 0.1745, loss_box_dn_5: 0.8407, loss_dense_depth: 0.7907, loss: 27.4181, grad_norm: 52.6627 -2025-11-12 14:34:18,153 - mmdet - INFO - Iter [140/17500] lr: 1.556e-04, eta: 11:38:01, time: 1.570, data_time: 0.077, memory: 49164, loss_cls_0: 0.8323, loss_box_0: 1.6780, loss_cns_0: 0.6212, loss_yns_0: 0.1448, loss_cls_1: 0.8983, loss_box_1: 1.7918, loss_cns_1: 0.6519, loss_yns_1: 0.1481, loss_cls_2: 0.9342, loss_box_2: 1.7441, loss_cns_2: 0.6585, loss_yns_2: 0.1484, loss_cls_3: 0.9470, loss_box_3: 1.7453, loss_cns_3: 0.6594, loss_yns_3: 0.1486, loss_cls_4: 0.9516, loss_box_4: 1.7520, loss_cns_4: 0.6584, loss_yns_4: 0.1505, loss_cls_5: 0.9472, loss_box_5: 1.7698, loss_cns_5: 0.6581, loss_yns_5: 0.1494, loss_cls_dn_0: 0.2258, loss_box_dn_0: 0.7836, loss_cls_dn_1: 0.1455, loss_box_dn_1: 0.7873, loss_cls_dn_2: 0.1471, loss_box_dn_2: 0.7922, loss_cls_dn_3: 0.1580, loss_box_dn_3: 0.8119, loss_cls_dn_4: 0.1597, loss_box_dn_4: 0.8482, loss_cls_dn_5: 0.1708, loss_box_dn_5: 0.8763, loss_dense_depth: 0.7760, loss: 27.4712, grad_norm: 49.0103 -2025-11-12 14:34:19,826 - mmdet - INFO - Iter [141/17500] lr: 1.560e-04, eta: 11:36:27, time: 1.674, data_time: 0.106, memory: 49164, loss_cls_0: 0.8849, loss_box_0: 1.7359, loss_cns_0: 0.6212, loss_yns_0: 0.1489, loss_cls_1: 0.9191, loss_box_1: 1.8236, loss_cns_1: 0.6443, loss_yns_1: 0.1510, loss_cls_2: 0.9390, loss_box_2: 1.8087, loss_cns_2: 0.6518, loss_yns_2: 0.1507, loss_cls_3: 0.9541, loss_box_3: 1.8035, loss_cns_3: 0.6552, loss_yns_3: 0.1546, loss_cls_4: 0.9841, loss_box_4: 1.7802, loss_cns_4: 0.6549, loss_yns_4: 0.1540, loss_cls_5: 0.9603, loss_box_5: 1.7992, loss_cns_5: 0.6554, loss_yns_5: 0.1534, loss_cls_dn_0: 0.2407, loss_box_dn_0: 0.7841, loss_cls_dn_1: 0.1526, loss_box_dn_1: 0.8131, loss_cls_dn_2: 0.1569, loss_box_dn_2: 0.8121, loss_cls_dn_3: 0.1657, loss_box_dn_3: 0.8217, loss_cls_dn_4: 0.1723, loss_box_dn_4: 0.8379, loss_cls_dn_5: 0.1797, loss_box_dn_5: 0.8533, loss_dense_depth: 0.8142, loss: 27.9925, grad_norm: 47.4594 -2025-11-12 14:34:21,457 - mmdet - INFO - Iter [142/17500] lr: 1.564e-04, eta: 11:34:50, time: 1.628, data_time: 0.102, memory: 49164, loss_cls_0: 0.8870, loss_box_0: 1.7825, loss_cns_0: 0.6171, loss_yns_0: 0.1525, loss_cls_1: 0.9327, loss_box_1: 1.8034, loss_cns_1: 0.6425, loss_yns_1: 0.1554, loss_cls_2: 0.9574, loss_box_2: 1.8026, loss_cns_2: 0.6415, loss_yns_2: 0.1540, loss_cls_3: 0.9659, loss_box_3: 1.7703, loss_cns_3: 0.6514, loss_yns_3: 0.1575, loss_cls_4: 0.9876, loss_box_4: 1.7220, loss_cns_4: 0.6539, loss_yns_4: 0.1540, loss_cls_5: 0.9726, loss_box_5: 1.7230, loss_cns_5: 0.6535, loss_yns_5: 0.1529, loss_cls_dn_0: 0.2375, loss_box_dn_0: 0.7871, loss_cls_dn_1: 0.1545, loss_box_dn_1: 0.8128, loss_cls_dn_2: 0.1593, loss_box_dn_2: 0.8031, loss_cls_dn_3: 0.1668, loss_box_dn_3: 0.7965, loss_cls_dn_4: 0.1773, loss_box_dn_4: 0.7888, loss_cls_dn_5: 0.1768, loss_box_dn_5: 0.7905, loss_dense_depth: 0.8114, loss: 27.7556, grad_norm: 40.7067 -2025-11-12 14:34:23,040 - mmdet - INFO - Iter [143/17500] lr: 1.568e-04, eta: 11:33:08, time: 1.580, data_time: 0.082, memory: 49164, loss_cls_0: 0.8432, loss_box_0: 1.7614, loss_cns_0: 0.6251, loss_yns_0: 0.1492, loss_cls_1: 0.9213, loss_box_1: 1.7713, loss_cns_1: 0.6564, loss_yns_1: 0.1486, loss_cls_2: 0.9525, loss_box_2: 1.7149, loss_cns_2: 0.6590, loss_yns_2: 0.1507, loss_cls_3: 0.9533, loss_box_3: 1.7061, loss_cns_3: 0.6552, loss_yns_3: 0.1505, loss_cls_4: 0.9639, loss_box_4: 1.7098, loss_cns_4: 0.6549, loss_yns_4: 0.1484, loss_cls_5: 0.9681, loss_box_5: 1.7347, loss_cns_5: 0.6541, loss_yns_5: 0.1489, loss_cls_dn_0: 0.2221, loss_box_dn_0: 0.7817, loss_cls_dn_1: 0.1531, loss_box_dn_1: 0.7565, loss_cls_dn_2: 0.1628, loss_box_dn_2: 0.7342, loss_cls_dn_3: 0.1664, loss_box_dn_3: 0.7319, loss_cls_dn_4: 0.1738, loss_box_dn_4: 0.7359, loss_cls_dn_5: 0.1793, loss_box_dn_5: 0.7463, loss_dense_depth: 0.7868, loss: 27.1323, grad_norm: 37.9647 -2025-11-12 14:34:24,618 - mmdet - INFO - Iter [144/17500] lr: 1.572e-04, eta: 11:31:27, time: 1.583, data_time: 0.075, memory: 49164, loss_cls_0: 0.8745, loss_box_0: 1.7738, loss_cns_0: 0.6223, loss_yns_0: 0.1503, loss_cls_1: 0.9327, loss_box_1: 1.8143, loss_cns_1: 0.6475, loss_yns_1: 0.1485, loss_cls_2: 0.9686, loss_box_2: 1.7727, loss_cns_2: 0.6575, loss_yns_2: 0.1487, loss_cls_3: 0.9704, loss_box_3: 1.7466, loss_cns_3: 0.6557, loss_yns_3: 0.1499, loss_cls_4: 0.9797, loss_box_4: 1.7530, loss_cns_4: 0.6570, loss_yns_4: 0.1493, loss_cls_5: 0.9752, loss_box_5: 1.7597, loss_cns_5: 0.6551, loss_yns_5: 0.1503, loss_cls_dn_0: 0.2265, loss_box_dn_0: 0.7921, loss_cls_dn_1: 0.1586, loss_box_dn_1: 0.7604, loss_cls_dn_2: 0.1687, loss_box_dn_2: 0.7546, loss_cls_dn_3: 0.1731, loss_box_dn_3: 0.7645, loss_cls_dn_4: 0.1764, loss_box_dn_4: 0.7815, loss_cls_dn_5: 0.1860, loss_box_dn_5: 0.8019, loss_dense_depth: 0.8088, loss: 27.6661, grad_norm: 43.9773 -2025-11-12 14:34:26,222 - mmdet - INFO - Iter [145/17500] lr: 1.576e-04, eta: 11:29:50, time: 1.600, data_time: 0.082, memory: 49164, loss_cls_0: 0.8527, loss_box_0: 1.7734, loss_cns_0: 0.6231, loss_yns_0: 0.1519, loss_cls_1: 0.9218, loss_box_1: 1.8326, loss_cns_1: 0.6434, loss_yns_1: 0.1532, loss_cls_2: 0.9486, loss_box_2: 1.7951, loss_cns_2: 0.6523, loss_yns_2: 0.1503, loss_cls_3: 0.9589, loss_box_3: 1.7639, loss_cns_3: 0.6529, loss_yns_3: 0.1522, loss_cls_4: 0.9770, loss_box_4: 1.7617, loss_cns_4: 0.6525, loss_yns_4: 0.1496, loss_cls_5: 0.9591, loss_box_5: 1.7817, loss_cns_5: 0.6508, loss_yns_5: 0.1497, loss_cls_dn_0: 0.2217, loss_box_dn_0: 0.7822, loss_cls_dn_1: 0.1544, loss_box_dn_1: 0.7813, loss_cls_dn_2: 0.1611, loss_box_dn_2: 0.7824, loss_cls_dn_3: 0.1671, loss_box_dn_3: 0.7979, loss_cls_dn_4: 0.1722, loss_box_dn_4: 0.8203, loss_cls_dn_5: 0.1784, loss_box_dn_5: 0.8478, loss_dense_depth: 0.7994, loss: 27.7748, grad_norm: 44.8991 -2025-11-12 14:34:27,822 - mmdet - INFO - Iter [146/17500] lr: 1.580e-04, eta: 11:28:15, time: 1.602, data_time: 0.116, memory: 49164, loss_cls_0: 0.8362, loss_box_0: 1.7533, loss_cns_0: 0.6189, loss_yns_0: 0.1491, loss_cls_1: 0.9199, loss_box_1: 1.8237, loss_cns_1: 0.6446, loss_yns_1: 0.1492, loss_cls_2: 0.9482, loss_box_2: 1.7858, loss_cns_2: 0.6531, loss_yns_2: 0.1496, loss_cls_3: 0.9621, loss_box_3: 1.7540, loss_cns_3: 0.6513, loss_yns_3: 0.1603, loss_cls_4: 0.9659, loss_box_4: 1.7345, loss_cns_4: 0.6496, loss_yns_4: 0.1515, loss_cls_5: 0.9672, loss_box_5: 1.7584, loss_cns_5: 0.6502, loss_yns_5: 0.1499, loss_cls_dn_0: 0.2193, loss_box_dn_0: 0.7829, loss_cls_dn_1: 0.1480, loss_box_dn_1: 0.8068, loss_cls_dn_2: 0.1550, loss_box_dn_2: 0.8088, loss_cls_dn_3: 0.1581, loss_box_dn_3: 0.8257, loss_cls_dn_4: 0.1652, loss_box_dn_4: 0.8409, loss_cls_dn_5: 0.1704, loss_box_dn_5: 0.8690, loss_dense_depth: 0.7742, loss: 27.7110, grad_norm: 39.0619 -2025-11-12 14:34:29,386 - mmdet - INFO - Iter [147/17500] lr: 1.584e-04, eta: 11:26:36, time: 1.563, data_time: 0.074, memory: 49164, loss_cls_0: 0.8582, loss_box_0: 1.7382, loss_cns_0: 0.6177, loss_yns_0: 0.1481, loss_cls_1: 0.9428, loss_box_1: 1.8540, loss_cns_1: 0.6463, loss_yns_1: 0.1487, loss_cls_2: 0.9799, loss_box_2: 1.7625, loss_cns_2: 0.6564, loss_yns_2: 0.1507, loss_cls_3: 0.9760, loss_box_3: 1.7427, loss_cns_3: 0.6572, loss_yns_3: 0.1554, loss_cls_4: 0.9664, loss_box_4: 1.7476, loss_cns_4: 0.6536, loss_yns_4: 0.1522, loss_cls_5: 0.9989, loss_box_5: 1.7746, loss_cns_5: 0.6556, loss_yns_5: 0.1527, loss_cls_dn_0: 0.2239, loss_box_dn_0: 0.7792, loss_cls_dn_1: 0.1536, loss_box_dn_1: 0.8339, loss_cls_dn_2: 0.1655, loss_box_dn_2: 0.8140, loss_cls_dn_3: 0.1640, loss_box_dn_3: 0.8254, loss_cls_dn_4: 0.1681, loss_box_dn_4: 0.8395, loss_cls_dn_5: 0.1746, loss_box_dn_5: 0.8620, loss_dense_depth: 0.8075, loss: 27.9475, grad_norm: 41.7645 -2025-11-12 14:34:30,964 - mmdet - INFO - Iter [148/17500] lr: 1.588e-04, eta: 11:25:01, time: 1.581, data_time: 0.081, memory: 49164, loss_cls_0: 0.8505, loss_box_0: 1.7306, loss_cns_0: 0.6213, loss_yns_0: 0.1451, loss_cls_1: 0.9133, loss_box_1: 1.8447, loss_cns_1: 0.6411, loss_yns_1: 0.1465, loss_cls_2: 0.9395, loss_box_2: 1.7089, loss_cns_2: 0.6581, loss_yns_2: 0.1503, loss_cls_3: 0.9512, loss_box_3: 1.6847, loss_cns_3: 0.6599, loss_yns_3: 0.1525, loss_cls_4: 0.9580, loss_box_4: 1.6845, loss_cns_4: 0.6576, loss_yns_4: 0.1548, loss_cls_5: 0.9558, loss_box_5: 1.7009, loss_cns_5: 0.6612, loss_yns_5: 0.1552, loss_cls_dn_0: 0.2200, loss_box_dn_0: 0.7868, loss_cls_dn_1: 0.1551, loss_box_dn_1: 0.8105, loss_cls_dn_2: 0.1638, loss_box_dn_2: 0.7617, loss_cls_dn_3: 0.1597, loss_box_dn_3: 0.7595, loss_cls_dn_4: 0.1626, loss_box_dn_4: 0.7637, loss_cls_dn_5: 0.1666, loss_box_dn_5: 0.7725, loss_dense_depth: 0.7586, loss: 27.1671, grad_norm: 41.1640 -2025-11-12 14:34:32,569 - mmdet - INFO - Iter [149/17500] lr: 1.592e-04, eta: 11:23:30, time: 1.606, data_time: 0.081, memory: 49164, loss_cls_0: 0.8300, loss_box_0: 1.7275, loss_cns_0: 0.6209, loss_yns_0: 0.1479, loss_cls_1: 0.9087, loss_box_1: 1.7897, loss_cns_1: 0.6451, loss_yns_1: 0.1485, loss_cls_2: 0.9332, loss_box_2: 1.7400, loss_cns_2: 0.6525, loss_yns_2: 0.1486, loss_cls_3: 0.9505, loss_box_3: 1.7082, loss_cns_3: 0.6537, loss_yns_3: 0.1533, loss_cls_4: 0.9515, loss_box_4: 1.7253, loss_cns_4: 0.6506, loss_yns_4: 0.1522, loss_cls_5: 0.9555, loss_box_5: 1.7538, loss_cns_5: 0.6502, loss_yns_5: 0.1535, loss_cls_dn_0: 0.2158, loss_box_dn_0: 0.7806, loss_cls_dn_1: 0.1491, loss_box_dn_1: 0.7562, loss_cls_dn_2: 0.1518, loss_box_dn_2: 0.7465, loss_cls_dn_3: 0.1521, loss_box_dn_3: 0.7394, loss_cls_dn_4: 0.1571, loss_box_dn_4: 0.7513, loss_cls_dn_5: 0.1613, loss_box_dn_5: 0.7666, loss_dense_depth: 0.7645, loss: 27.0432, grad_norm: 42.2662 -2025-11-12 14:34:34,174 - mmdet - INFO - Iter [150/17500] lr: 1.596e-04, eta: 11:21:57, time: 1.582, data_time: 0.079, memory: 49164, loss_cls_0: 0.8547, loss_box_0: 1.7419, loss_cns_0: 0.6218, loss_yns_0: 0.1473, loss_cls_1: 0.9304, loss_box_1: 1.7591, loss_cns_1: 0.6535, loss_yns_1: 0.1469, loss_cls_2: 0.9496, loss_box_2: 1.7457, loss_cns_2: 0.6535, loss_yns_2: 0.1473, loss_cls_3: 0.9608, loss_box_3: 1.7205, loss_cns_3: 0.6603, loss_yns_3: 0.1566, loss_cls_4: 0.9545, loss_box_4: 1.7365, loss_cns_4: 0.6525, loss_yns_4: 0.1470, loss_cls_5: 0.9713, loss_box_5: 1.7679, loss_cns_5: 0.6529, loss_yns_5: 0.1473, loss_cls_dn_0: 0.2193, loss_box_dn_0: 0.7790, loss_cls_dn_1: 0.1451, loss_box_dn_1: 0.7534, loss_cls_dn_2: 0.1494, loss_box_dn_2: 0.7713, loss_cls_dn_3: 0.1538, loss_box_dn_3: 0.7805, loss_cls_dn_4: 0.1577, loss_box_dn_4: 0.8121, loss_cls_dn_5: 0.1665, loss_box_dn_5: 0.8456, loss_dense_depth: 0.7634, loss: 27.3768, grad_norm: 44.4350 -2025-11-12 14:34:35,743 - mmdet - INFO - Iter [151/17500] lr: 1.600e-04, eta: 11:20:26, time: 1.592, data_time: 0.085, memory: 49164, loss_cls_0: 0.8984, loss_box_0: 1.7414, loss_cns_0: 0.6274, loss_yns_0: 0.1460, loss_cls_1: 0.9286, loss_box_1: 1.8023, loss_cns_1: 0.6461, loss_yns_1: 0.1478, loss_cls_2: 0.9494, loss_box_2: 1.7861, loss_cns_2: 0.6492, loss_yns_2: 0.1480, loss_cls_3: 0.9597, loss_box_3: 1.7815, loss_cns_3: 0.6567, loss_yns_3: 0.1604, loss_cls_4: 0.9564, loss_box_4: 1.7701, loss_cns_4: 0.6501, loss_yns_4: 0.1498, loss_cls_5: 0.9640, loss_box_5: 1.7764, loss_cns_5: 0.6547, loss_yns_5: 0.1490, loss_cls_dn_0: 0.2177, loss_box_dn_0: 0.7797, loss_cls_dn_1: 0.1436, loss_box_dn_1: 0.7975, loss_cls_dn_2: 0.1492, loss_box_dn_2: 0.8168, loss_cls_dn_3: 0.1520, loss_box_dn_3: 0.8405, loss_cls_dn_4: 0.1569, loss_box_dn_4: 0.8656, loss_cls_dn_5: 0.1664, loss_box_dn_5: 0.8959, loss_dense_depth: 0.7787, loss: 27.8599, grad_norm: 49.2974 -2025-11-12 14:34:37,319 - mmdet - INFO - Iter [152/17500] lr: 1.604e-04, eta: 11:18:55, time: 1.576, data_time: 0.072, memory: 49164, loss_cls_0: 0.8804, loss_box_0: 1.7130, loss_cns_0: 0.6251, loss_yns_0: 0.1485, loss_cls_1: 0.9147, loss_box_1: 1.8467, loss_cns_1: 0.6333, loss_yns_1: 0.1526, loss_cls_2: 0.9464, loss_box_2: 1.7825, loss_cns_2: 0.6502, loss_yns_2: 0.1516, loss_cls_3: 0.9522, loss_box_3: 1.7681, loss_cns_3: 0.6537, loss_yns_3: 0.1574, loss_cls_4: 0.9612, loss_box_4: 1.7537, loss_cns_4: 0.6521, loss_yns_4: 0.1528, loss_cls_5: 0.9509, loss_box_5: 1.7583, loss_cns_5: 0.6542, loss_yns_5: 0.1531, loss_cls_dn_0: 0.2160, loss_box_dn_0: 0.7728, loss_cls_dn_1: 0.1445, loss_box_dn_1: 0.8297, loss_cls_dn_2: 0.1493, loss_box_dn_2: 0.8343, loss_cls_dn_3: 0.1529, loss_box_dn_3: 0.8495, loss_cls_dn_4: 0.1587, loss_box_dn_4: 0.8665, loss_cls_dn_5: 0.1641, loss_box_dn_5: 0.8942, loss_dense_depth: 0.7297, loss: 27.7751, grad_norm: 40.4926 -2025-11-12 14:34:38,899 - mmdet - INFO - Iter [153/17500] lr: 1.608e-04, eta: 11:17:26, time: 1.581, data_time: 0.071, memory: 49164, loss_cls_0: 0.8657, loss_box_0: 1.7162, loss_cns_0: 0.6162, loss_yns_0: 0.1496, loss_cls_1: 0.9174, loss_box_1: 1.9259, loss_cns_1: 0.6254, loss_yns_1: 0.1512, loss_cls_2: 0.9575, loss_box_2: 1.7766, loss_cns_2: 0.6485, loss_yns_2: 0.1506, loss_cls_3: 0.9654, loss_box_3: 1.7593, loss_cns_3: 0.6536, loss_yns_3: 0.1499, loss_cls_4: 0.9779, loss_box_4: 1.7465, loss_cns_4: 0.6515, loss_yns_4: 0.1504, loss_cls_5: 0.9686, loss_box_5: 1.7495, loss_cns_5: 0.6513, loss_yns_5: 0.1502, loss_cls_dn_0: 0.2285, loss_box_dn_0: 0.7789, loss_cls_dn_1: 0.1450, loss_box_dn_1: 0.8154, loss_cls_dn_2: 0.1496, loss_box_dn_2: 0.7772, loss_cls_dn_3: 0.1514, loss_box_dn_3: 0.7770, loss_cls_dn_4: 0.1583, loss_box_dn_4: 0.7837, loss_cls_dn_5: 0.1616, loss_box_dn_5: 0.7966, loss_dense_depth: 0.7667, loss: 27.5650, grad_norm: 43.5155 -2025-11-12 14:34:40,493 - mmdet - INFO - Iter [154/17500] lr: 1.612e-04, eta: 11:15:59, time: 1.591, data_time: 0.073, memory: 49164, loss_cls_0: 0.8699, loss_box_0: 1.6960, loss_cns_0: 0.6078, loss_yns_0: 0.1450, loss_cls_1: 0.9078, loss_box_1: 1.7765, loss_cns_1: 0.6417, loss_yns_1: 0.1517, loss_cls_2: 0.9419, loss_box_2: 1.7297, loss_cns_2: 0.6444, loss_yns_2: 0.1510, loss_cls_3: 0.9638, loss_box_3: 1.7153, loss_cns_3: 0.6559, loss_yns_3: 0.1563, loss_cls_4: 0.9610, loss_box_4: 1.7077, loss_cns_4: 0.6516, loss_yns_4: 0.1527, loss_cls_5: 0.9523, loss_box_5: 1.7273, loss_cns_5: 0.6525, loss_yns_5: 0.1513, loss_cls_dn_0: 0.2260, loss_box_dn_0: 0.7751, loss_cls_dn_1: 0.1444, loss_box_dn_1: 0.7624, loss_cls_dn_2: 0.1492, loss_box_dn_2: 0.7375, loss_cls_dn_3: 0.1520, loss_box_dn_3: 0.7348, loss_cls_dn_4: 0.1545, loss_box_dn_4: 0.7402, loss_cls_dn_5: 0.1574, loss_box_dn_5: 0.7416, loss_dense_depth: 0.7359, loss: 26.9222, grad_norm: 49.4688 -2025-11-12 14:34:42,078 - mmdet - INFO - Iter [155/17500] lr: 1.616e-04, eta: 11:14:32, time: 1.583, data_time: 0.079, memory: 49164, loss_cls_0: 0.8419, loss_box_0: 1.6890, loss_cns_0: 0.6154, loss_yns_0: 0.1476, loss_cls_1: 0.9033, loss_box_1: 1.7192, loss_cns_1: 0.6483, loss_yns_1: 0.1494, loss_cls_2: 0.9269, loss_box_2: 1.7008, loss_cns_2: 0.6513, loss_yns_2: 0.1498, loss_cls_3: 0.9448, loss_box_3: 1.6699, loss_cns_3: 0.6649, loss_yns_3: 0.1613, loss_cls_4: 0.9368, loss_box_4: 1.6679, loss_cns_4: 0.6574, loss_yns_4: 0.1519, loss_cls_5: 0.9360, loss_box_5: 1.6870, loss_cns_5: 0.6589, loss_yns_5: 0.1512, loss_cls_dn_0: 0.2232, loss_box_dn_0: 0.7780, loss_cls_dn_1: 0.1420, loss_box_dn_1: 0.7329, loss_cls_dn_2: 0.1443, loss_box_dn_2: 0.7211, loss_cls_dn_3: 0.1499, loss_box_dn_3: 0.7191, loss_cls_dn_4: 0.1478, loss_box_dn_4: 0.7295, loss_cls_dn_5: 0.1531, loss_box_dn_5: 0.7387, loss_dense_depth: 0.7223, loss: 26.5325, grad_norm: 41.4002 -2025-11-12 14:34:43,645 - mmdet - INFO - Iter [156/17500] lr: 1.620e-04, eta: 11:13:04, time: 1.563, data_time: 0.075, memory: 49164, loss_cls_0: 0.8293, loss_box_0: 1.7084, loss_cns_0: 0.6188, loss_yns_0: 0.1513, loss_cls_1: 0.9099, loss_box_1: 1.7391, loss_cns_1: 0.6422, loss_yns_1: 0.1504, loss_cls_2: 0.9341, loss_box_2: 1.6696, loss_cns_2: 0.6528, loss_yns_2: 0.1517, loss_cls_3: 0.9474, loss_box_3: 1.7190, loss_cns_3: 0.6513, loss_yns_3: 0.1571, loss_cls_4: 0.9518, loss_box_4: 1.7263, loss_cns_4: 0.6486, loss_yns_4: 0.1529, loss_cls_5: 0.9490, loss_box_5: 1.7189, loss_cns_5: 0.6503, loss_yns_5: 0.1540, loss_cls_dn_0: 0.2241, loss_box_dn_0: 0.7719, loss_cls_dn_1: 0.1433, loss_box_dn_1: 0.7757, loss_cls_dn_2: 0.1423, loss_box_dn_2: 0.7634, loss_cls_dn_3: 0.1441, loss_box_dn_3: 0.7922, loss_cls_dn_4: 0.1532, loss_box_dn_4: 0.8129, loss_cls_dn_5: 0.1590, loss_box_dn_5: 0.8298, loss_dense_depth: 0.7596, loss: 27.0558, grad_norm: 57.2870 -2025-11-12 14:34:45,207 - mmdet - INFO - Iter [157/17500] lr: 1.624e-04, eta: 11:11:38, time: 1.569, data_time: 0.081, memory: 49164, loss_cls_0: 0.8435, loss_box_0: 1.7296, loss_cns_0: 0.6224, loss_yns_0: 0.1542, loss_cls_1: 0.9079, loss_box_1: 1.7315, loss_cns_1: 0.6427, loss_yns_1: 0.1532, loss_cls_2: 0.9312, loss_box_2: 1.7077, loss_cns_2: 0.6491, loss_yns_2: 0.1529, loss_cls_3: 0.9610, loss_box_3: 1.7409, loss_cns_3: 0.6500, loss_yns_3: 0.1553, loss_cls_4: 0.9419, loss_box_4: 1.7359, loss_cns_4: 0.6479, loss_yns_4: 0.1570, loss_cls_5: 0.9480, loss_box_5: 1.7272, loss_cns_5: 0.6481, loss_yns_5: 0.1584, loss_cls_dn_0: 0.2168, loss_box_dn_0: 0.7710, loss_cls_dn_1: 0.1421, loss_box_dn_1: 0.7924, loss_cls_dn_2: 0.1422, loss_box_dn_2: 0.8056, loss_cls_dn_3: 0.1470, loss_box_dn_3: 0.8414, loss_cls_dn_4: 0.1529, loss_box_dn_4: 0.8698, loss_cls_dn_5: 0.1572, loss_box_dn_5: 0.8945, loss_dense_depth: 0.7269, loss: 27.3571, grad_norm: 48.2349 -2025-11-12 14:34:46,779 - mmdet - INFO - Iter [158/17500] lr: 1.628e-04, eta: 11:10:13, time: 1.574, data_time: 0.078, memory: 49164, loss_cls_0: 0.8164, loss_box_0: 1.7574, loss_cns_0: 0.6169, loss_yns_0: 0.1540, loss_cls_1: 0.9015, loss_box_1: 1.7922, loss_cns_1: 0.6355, loss_yns_1: 0.1518, loss_cls_2: 0.9250, loss_box_2: 1.7758, loss_cns_2: 0.6450, loss_yns_2: 0.1521, loss_cls_3: 0.9276, loss_box_3: 1.7103, loss_cns_3: 0.6596, loss_yns_3: 0.1537, loss_cls_4: 0.9256, loss_box_4: 1.7188, loss_cns_4: 0.6509, loss_yns_4: 0.1539, loss_cls_5: 0.9267, loss_box_5: 1.7451, loss_cns_5: 0.6498, loss_yns_5: 0.1537, loss_cls_dn_0: 0.2106, loss_box_dn_0: 0.7843, loss_cls_dn_1: 0.1428, loss_box_dn_1: 0.8354, loss_cls_dn_2: 0.1426, loss_box_dn_2: 0.8590, loss_cls_dn_3: 0.1465, loss_box_dn_3: 0.8711, loss_cls_dn_4: 0.1503, loss_box_dn_4: 0.9002, loss_cls_dn_5: 0.1546, loss_box_dn_5: 0.9341, loss_dense_depth: 0.7309, loss: 27.5615, grad_norm: 64.2047 -2025-11-12 14:34:48,354 - mmdet - INFO - Iter [159/17500] lr: 1.632e-04, eta: 11:08:49, time: 1.570, data_time: 0.074, memory: 49164, loss_cls_0: 0.8150, loss_box_0: 1.7410, loss_cns_0: 0.6221, loss_yns_0: 0.1532, loss_cls_1: 0.8998, loss_box_1: 1.8444, loss_cns_1: 0.6275, loss_yns_1: 0.1546, loss_cls_2: 0.9439, loss_box_2: 1.7651, loss_cns_2: 0.6491, loss_yns_2: 0.1549, loss_cls_3: 0.9304, loss_box_3: 1.6826, loss_cns_3: 0.6610, loss_yns_3: 0.1585, loss_cls_4: 0.9303, loss_box_4: 1.6836, loss_cns_4: 0.6539, loss_yns_4: 0.1558, loss_cls_5: 0.9310, loss_box_5: 1.7101, loss_cns_5: 0.6532, loss_yns_5: 0.1547, loss_cls_dn_0: 0.2115, loss_box_dn_0: 0.7697, loss_cls_dn_1: 0.1439, loss_box_dn_1: 0.8496, loss_cls_dn_2: 0.1449, loss_box_dn_2: 0.8426, loss_cls_dn_3: 0.1465, loss_box_dn_3: 0.8414, loss_cls_dn_4: 0.1496, loss_box_dn_4: 0.8567, loss_cls_dn_5: 0.1538, loss_box_dn_5: 0.8806, loss_dense_depth: 0.7427, loss: 27.4092, grad_norm: 53.7458 -2025-11-12 14:34:49,951 - mmdet - INFO - Iter [160/17500] lr: 1.636e-04, eta: 11:07:29, time: 1.591, data_time: 0.077, memory: 49164, loss_cls_0: 0.8132, loss_box_0: 1.7610, loss_cns_0: 0.6186, loss_yns_0: 0.1536, loss_cls_1: 0.8815, loss_box_1: 1.7519, loss_cns_1: 0.6430, loss_yns_1: 0.1528, loss_cls_2: 0.9132, loss_box_2: 1.7261, loss_cns_2: 0.6521, loss_yns_2: 0.1519, loss_cls_3: 0.9292, loss_box_3: 1.7151, loss_cns_3: 0.6554, loss_yns_3: 0.1591, loss_cls_4: 0.9272, loss_box_4: 1.6926, loss_cns_4: 0.6501, loss_yns_4: 0.1531, loss_cls_5: 0.9336, loss_box_5: 1.6790, loss_cns_5: 0.6514, loss_yns_5: 0.1519, loss_cls_dn_0: 0.2135, loss_box_dn_0: 0.7784, loss_cls_dn_1: 0.1431, loss_box_dn_1: 0.7876, loss_cls_dn_2: 0.1422, loss_box_dn_2: 0.7885, loss_cls_dn_3: 0.1455, loss_box_dn_3: 0.8014, loss_cls_dn_4: 0.1490, loss_box_dn_4: 0.7943, loss_cls_dn_5: 0.1546, loss_box_dn_5: 0.7969, loss_dense_depth: 0.7298, loss: 26.9415, grad_norm: 53.8988 -2025-11-12 14:34:51,579 - mmdet - INFO - Iter [161/17500] lr: 1.640e-04, eta: 11:06:14, time: 1.635, data_time: 0.111, memory: 49164, loss_cls_0: 0.8102, loss_box_0: 1.7900, loss_cns_0: 0.6133, loss_yns_0: 0.1541, loss_cls_1: 0.8807, loss_box_1: 1.7751, loss_cns_1: 0.6425, loss_yns_1: 0.1530, loss_cls_2: 0.9185, loss_box_2: 1.7867, loss_cns_2: 0.6424, loss_yns_2: 0.1556, loss_cls_3: 0.9306, loss_box_3: 1.7704, loss_cns_3: 0.6455, loss_yns_3: 0.1587, loss_cls_4: 0.9320, loss_box_4: 1.7207, loss_cns_4: 0.6474, loss_yns_4: 0.1548, loss_cls_5: 0.9301, loss_box_5: 1.7079, loss_cns_5: 0.6493, loss_yns_5: 0.1555, loss_cls_dn_0: 0.2111, loss_box_dn_0: 0.7946, loss_cls_dn_1: 0.1391, loss_box_dn_1: 0.7136, loss_cls_dn_2: 0.1391, loss_box_dn_2: 0.7379, loss_cls_dn_3: 0.1444, loss_box_dn_3: 0.7447, loss_cls_dn_4: 0.1482, loss_box_dn_4: 0.7280, loss_cls_dn_5: 0.1513, loss_box_dn_5: 0.7285, loss_dense_depth: 0.7301, loss: 26.8356, grad_norm: 59.0549 -2025-11-12 14:34:53,198 - mmdet - INFO - Iter [162/17500] lr: 1.644e-04, eta: 11:04:58, time: 1.617, data_time: 0.109, memory: 49164, loss_cls_0: 0.8176, loss_box_0: 1.7512, loss_cns_0: 0.6204, loss_yns_0: 0.1554, loss_cls_1: 0.8936, loss_box_1: 1.8057, loss_cns_1: 0.6365, loss_yns_1: 0.1551, loss_cls_2: 0.9320, loss_box_2: 1.6970, loss_cns_2: 0.6504, loss_yns_2: 0.1544, loss_cls_3: 0.9229, loss_box_3: 1.6995, loss_cns_3: 0.6556, loss_yns_3: 0.1558, loss_cls_4: 0.9257, loss_box_4: 1.6991, loss_cns_4: 0.6485, loss_yns_4: 0.1604, loss_cls_5: 0.9315, loss_box_5: 1.7150, loss_cns_5: 0.6546, loss_yns_5: 0.1564, loss_cls_dn_0: 0.2127, loss_box_dn_0: 0.7827, loss_cls_dn_1: 0.1350, loss_box_dn_1: 0.7345, loss_cls_dn_2: 0.1363, loss_box_dn_2: 0.7186, loss_cls_dn_3: 0.1383, loss_box_dn_3: 0.7347, loss_cls_dn_4: 0.1427, loss_box_dn_4: 0.7535, loss_cls_dn_5: 0.1534, loss_box_dn_5: 0.7752, loss_dense_depth: 0.7194, loss: 26.7309, grad_norm: 52.5920 -2025-11-12 14:34:54,784 - mmdet - INFO - Iter [163/17500] lr: 1.648e-04, eta: 11:03:39, time: 1.587, data_time: 0.087, memory: 49164, loss_cls_0: 0.8095, loss_box_0: 1.7555, loss_cns_0: 0.6192, loss_yns_0: 0.1562, loss_cls_1: 0.8840, loss_box_1: 1.8345, loss_cns_1: 0.6301, loss_yns_1: 0.1565, loss_cls_2: 0.9093, loss_box_2: 1.7098, loss_cns_2: 0.6517, loss_yns_2: 0.1550, loss_cls_3: 0.9205, loss_box_3: 1.7272, loss_cns_3: 0.6572, loss_yns_3: 0.1589, loss_cls_4: 0.9292, loss_box_4: 1.7656, loss_cns_4: 0.6454, loss_yns_4: 0.1617, loss_cls_5: 0.9315, loss_box_5: 1.7688, loss_cns_5: 0.6499, loss_yns_5: 0.1555, loss_cls_dn_0: 0.2121, loss_box_dn_0: 0.7668, loss_cls_dn_1: 0.1365, loss_box_dn_1: 0.7661, loss_cls_dn_2: 0.1409, loss_box_dn_2: 0.7454, loss_cls_dn_3: 0.1420, loss_box_dn_3: 0.7690, loss_cls_dn_4: 0.1447, loss_box_dn_4: 0.8099, loss_cls_dn_5: 0.1548, loss_box_dn_5: 0.8317, loss_dense_depth: 0.7154, loss: 27.0777, grad_norm: 56.0871 -2025-11-12 14:34:56,380 - mmdet - INFO - Iter [164/17500] lr: 1.652e-04, eta: 11:02:23, time: 1.601, data_time: 0.081, memory: 49164, loss_cls_0: 0.8312, loss_box_0: 1.7870, loss_cns_0: 0.6223, loss_yns_0: 0.1565, loss_cls_1: 0.8864, loss_box_1: 1.7442, loss_cns_1: 0.6540, loss_yns_1: 0.1580, loss_cls_2: 0.9117, loss_box_2: 1.6894, loss_cns_2: 0.6598, loss_yns_2: 0.1572, loss_cls_3: 0.9219, loss_box_3: 1.6778, loss_cns_3: 0.6734, loss_yns_3: 0.1614, loss_cls_4: 0.9149, loss_box_4: 1.7045, loss_cns_4: 0.6567, loss_yns_4: 0.1571, loss_cls_5: 0.9182, loss_box_5: 1.6871, loss_cns_5: 0.6582, loss_yns_5: 0.1563, loss_cls_dn_0: 0.2131, loss_box_dn_0: 0.7797, loss_cls_dn_1: 0.1353, loss_box_dn_1: 0.7933, loss_cls_dn_2: 0.1373, loss_box_dn_2: 0.7963, loss_cls_dn_3: 0.1408, loss_box_dn_3: 0.8049, loss_cls_dn_4: 0.1430, loss_box_dn_4: 0.8392, loss_cls_dn_5: 0.1511, loss_box_dn_5: 0.8492, loss_dense_depth: 0.7158, loss: 27.0444, grad_norm: 46.9662 -2025-11-12 14:34:57,970 - mmdet - INFO - Iter [165/17500] lr: 1.656e-04, eta: 11:01:07, time: 1.591, data_time: 0.078, memory: 49164, loss_cls_0: 0.8303, loss_box_0: 1.7955, loss_cns_0: 0.6169, loss_yns_0: 0.1577, loss_cls_1: 0.8992, loss_box_1: 1.7261, loss_cns_1: 0.6513, loss_yns_1: 0.1580, loss_cls_2: 0.9128, loss_box_2: 1.7301, loss_cns_2: 0.6488, loss_yns_2: 0.1573, loss_cls_3: 0.9157, loss_box_3: 1.7064, loss_cns_3: 0.6573, loss_yns_3: 0.1616, loss_cls_4: 0.9275, loss_box_4: 1.7271, loss_cns_4: 0.6495, loss_yns_4: 0.1595, loss_cls_5: 0.9662, loss_box_5: 1.7361, loss_cns_5: 0.6509, loss_yns_5: 0.1602, loss_cls_dn_0: 0.2153, loss_box_dn_0: 0.7754, loss_cls_dn_1: 0.1366, loss_box_dn_1: 0.7845, loss_cls_dn_2: 0.1365, loss_box_dn_2: 0.7959, loss_cls_dn_3: 0.1409, loss_box_dn_3: 0.7934, loss_cls_dn_4: 0.1431, loss_box_dn_4: 0.8184, loss_cls_dn_5: 0.1538, loss_box_dn_5: 0.8319, loss_dense_depth: 0.7310, loss: 27.1590, grad_norm: 58.2670 -2025-11-12 14:34:59,563 - mmdet - INFO - Iter [166/17500] lr: 1.660e-04, eta: 10:59:53, time: 1.593, data_time: 0.103, memory: 49164, loss_cls_0: 0.8022, loss_box_0: 1.7765, loss_cns_0: 0.6138, loss_yns_0: 0.1581, loss_cls_1: 0.9140, loss_box_1: 1.6874, loss_cns_1: 0.6536, loss_yns_1: 0.1617, loss_cls_2: 0.9094, loss_box_2: 1.6813, loss_cns_2: 0.6544, loss_yns_2: 0.1596, loss_cls_3: 0.9377, loss_box_3: 1.6456, loss_cns_3: 0.6587, loss_yns_3: 0.1607, loss_cls_4: 0.9280, loss_box_4: 1.6521, loss_cns_4: 0.6576, loss_yns_4: 0.1628, loss_cls_5: 0.9276, loss_box_5: 1.6608, loss_cns_5: 0.6649, loss_yns_5: 0.1599, loss_cls_dn_0: 0.2057, loss_box_dn_0: 0.7595, loss_cls_dn_1: 0.1364, loss_box_dn_1: 0.7836, loss_cls_dn_2: 0.1330, loss_box_dn_2: 0.7778, loss_cls_dn_3: 0.1366, loss_box_dn_3: 0.7637, loss_cls_dn_4: 0.1377, loss_box_dn_4: 0.7743, loss_cls_dn_5: 0.1424, loss_box_dn_5: 0.7835, loss_dense_depth: 0.7321, loss: 26.6546, grad_norm: 31.3539 -2025-11-12 14:35:06,361 - mmdet - INFO - Iter [167/17500] lr: 1.664e-04, eta: 11:07:39, time: 6.798, data_time: 0.074, memory: 49164, loss_cls_0: 0.8473, loss_box_0: 1.8034, loss_cns_0: 0.6101, loss_yns_0: 0.1607, loss_cls_1: 0.9346, loss_box_1: 1.6904, loss_cns_1: 0.6520, loss_yns_1: 0.1593, loss_cls_2: 0.9553, loss_box_2: 1.6896, loss_cns_2: 0.6518, loss_yns_2: 0.1597, loss_cls_3: 0.9607, loss_box_3: 1.6773, loss_cns_3: 0.6581, loss_yns_3: 0.1603, loss_cls_4: 0.9617, loss_box_4: 1.6574, loss_cns_4: 0.6531, loss_yns_4: 0.1609, loss_cls_5: 0.9711, loss_box_5: 1.6536, loss_cns_5: 0.6550, loss_yns_5: 0.1600, loss_cls_dn_0: 0.2184, loss_box_dn_0: 0.7855, loss_cls_dn_1: 0.1372, loss_box_dn_1: 0.7818, loss_cls_dn_2: 0.1379, loss_box_dn_2: 0.7822, loss_cls_dn_3: 0.1443, loss_box_dn_3: 0.7778, loss_cls_dn_4: 0.1496, loss_box_dn_4: 0.7805, loss_cls_dn_5: 0.1527, loss_box_dn_5: 0.7770, loss_dense_depth: 0.8153, loss: 27.0839, grad_norm: 50.1214 -2025-11-12 14:35:07,908 - mmdet - INFO - Iter [168/17500] lr: 1.668e-04, eta: 11:06:18, time: 1.547, data_time: 0.076, memory: 49164, loss_cls_0: 0.8337, loss_box_0: 1.7476, loss_cns_0: 0.6187, loss_yns_0: 0.1602, loss_cls_1: 0.9150, loss_box_1: 1.6537, loss_cns_1: 0.6515, loss_yns_1: 0.1604, loss_cls_2: 0.9451, loss_box_2: 1.6787, loss_cns_2: 0.6505, loss_yns_2: 0.1611, loss_cls_3: 0.9641, loss_box_3: 1.6590, loss_cns_3: 0.6671, loss_yns_3: 0.1621, loss_cls_4: 0.9629, loss_box_4: 1.6421, loss_cns_4: 0.6530, loss_yns_4: 0.1613, loss_cls_5: 0.9870, loss_box_5: 1.6418, loss_cns_5: 0.6519, loss_yns_5: 0.1608, loss_cls_dn_0: 0.2112, loss_box_dn_0: 0.7728, loss_cls_dn_1: 0.1361, loss_box_dn_1: 0.7274, loss_cls_dn_2: 0.1386, loss_box_dn_2: 0.7527, loss_cls_dn_3: 0.1498, loss_box_dn_3: 0.7521, loss_cls_dn_4: 0.1485, loss_box_dn_4: 0.7669, loss_cls_dn_5: 0.1538, loss_box_dn_5: 0.7711, loss_dense_depth: 0.7747, loss: 26.7447, grad_norm: 54.5127 -2025-11-12 14:35:09,507 - mmdet - INFO - Iter [169/17500] lr: 1.672e-04, eta: 11:05:02, time: 1.597, data_time: 0.077, memory: 49164, loss_cls_0: 0.8757, loss_box_0: 1.8004, loss_cns_0: 0.6106, loss_yns_0: 0.1599, loss_cls_1: 0.9441, loss_box_1: 1.7140, loss_cns_1: 0.6447, loss_yns_1: 0.1609, loss_cls_2: 0.9654, loss_box_2: 1.6660, loss_cns_2: 0.6502, loss_yns_2: 0.1604, loss_cls_3: 0.9841, loss_box_3: 1.6663, loss_cns_3: 0.6623, loss_yns_3: 0.1628, loss_cls_4: 0.9674, loss_box_4: 1.6678, loss_cns_4: 0.6493, loss_yns_4: 0.1608, loss_cls_5: 0.9998, loss_box_5: 1.6746, loss_cns_5: 0.6559, loss_yns_5: 0.1610, loss_cls_dn_0: 0.2198, loss_box_dn_0: 0.7845, loss_cls_dn_1: 0.1366, loss_box_dn_1: 0.7424, loss_cls_dn_2: 0.1407, loss_box_dn_2: 0.7430, loss_cls_dn_3: 0.1452, loss_box_dn_3: 0.7552, loss_cls_dn_4: 0.1445, loss_box_dn_4: 0.7789, loss_cls_dn_5: 0.1521, loss_box_dn_5: 0.7967, loss_dense_depth: 0.8192, loss: 27.1234, grad_norm: 33.4858 -2025-11-12 14:35:11,080 - mmdet - INFO - Iter [170/17500] lr: 1.676e-04, eta: 11:03:46, time: 1.574, data_time: 0.080, memory: 49164, loss_cls_0: 0.8560, loss_box_0: 1.7616, loss_cns_0: 0.6172, loss_yns_0: 0.1595, loss_cls_1: 0.9465, loss_box_1: 1.7298, loss_cns_1: 0.6480, loss_yns_1: 0.1594, loss_cls_2: 0.9813, loss_box_2: 1.6484, loss_cns_2: 0.6593, loss_yns_2: 0.1596, loss_cls_3: 0.9670, loss_box_3: 1.6315, loss_cns_3: 0.6589, loss_yns_3: 0.1660, loss_cls_4: 0.9998, loss_box_4: 1.6187, loss_cns_4: 0.6565, loss_yns_4: 0.1590, loss_cls_5: 0.9731, loss_box_5: 1.6244, loss_cns_5: 0.6587, loss_yns_5: 0.1590, loss_cls_dn_0: 0.2149, loss_box_dn_0: 0.7753, loss_cls_dn_1: 0.1365, loss_box_dn_1: 0.7824, loss_cls_dn_2: 0.1414, loss_box_dn_2: 0.7778, loss_cls_dn_3: 0.1420, loss_box_dn_3: 0.7927, loss_cls_dn_4: 0.1487, loss_box_dn_4: 0.8090, loss_cls_dn_5: 0.1575, loss_box_dn_5: 0.8327, loss_dense_depth: 0.8262, loss: 27.1362, grad_norm: 55.2972 -2025-11-12 14:35:12,676 - mmdet - INFO - Iter [171/17500] lr: 1.680e-04, eta: 11:02:32, time: 1.588, data_time: 0.072, memory: 49164, loss_cls_0: 0.8443, loss_box_0: 1.7091, loss_cns_0: 0.6153, loss_yns_0: 0.1572, loss_cls_1: 0.9245, loss_box_1: 1.7074, loss_cns_1: 0.6459, loss_yns_1: 0.1576, loss_cls_2: 0.9527, loss_box_2: 1.6621, loss_cns_2: 0.6561, loss_yns_2: 0.1583, loss_cls_3: 0.9662, loss_box_3: 1.6430, loss_cns_3: 0.6594, loss_yns_3: 0.1603, loss_cls_4: 0.9697, loss_box_4: 1.6303, loss_cns_4: 0.6583, loss_yns_4: 0.1598, loss_cls_5: 0.9669, loss_box_5: 1.6347, loss_cns_5: 0.6588, loss_yns_5: 0.1581, loss_cls_dn_0: 0.2142, loss_box_dn_0: 0.7619, loss_cls_dn_1: 0.1373, loss_box_dn_1: 0.7822, loss_cls_dn_2: 0.1397, loss_box_dn_2: 0.7835, loss_cls_dn_3: 0.1429, loss_box_dn_3: 0.7894, loss_cls_dn_4: 0.1458, loss_box_dn_4: 0.7977, loss_cls_dn_5: 0.1483, loss_box_dn_5: 0.8142, loss_dense_depth: 0.7468, loss: 26.8597, grad_norm: 37.3475 -2025-11-12 14:35:14,251 - mmdet - INFO - Iter [172/17500] lr: 1.684e-04, eta: 11:01:17, time: 1.576, data_time: 0.083, memory: 49164, loss_cls_0: 0.8736, loss_box_0: 1.7541, loss_cns_0: 0.6088, loss_yns_0: 0.1583, loss_cls_1: 0.9380, loss_box_1: 1.6700, loss_cns_1: 0.6479, loss_yns_1: 0.1580, loss_cls_2: 0.9576, loss_box_2: 1.6755, loss_cns_2: 0.6518, loss_yns_2: 0.1592, loss_cls_3: 0.9728, loss_box_3: 1.6446, loss_cns_3: 0.6657, loss_yns_3: 0.1596, loss_cls_4: 0.9914, loss_box_4: 1.6405, loss_cns_4: 0.6561, loss_yns_4: 0.1604, loss_cls_5: 0.9761, loss_box_5: 1.6525, loss_cns_5: 0.6575, loss_yns_5: 0.1591, loss_cls_dn_0: 0.2190, loss_box_dn_0: 0.7612, loss_cls_dn_1: 0.1355, loss_box_dn_1: 0.7547, loss_cls_dn_2: 0.1377, loss_box_dn_2: 0.7591, loss_cls_dn_3: 0.1416, loss_box_dn_3: 0.7481, loss_cls_dn_4: 0.1478, loss_box_dn_4: 0.7569, loss_cls_dn_5: 0.1477, loss_box_dn_5: 0.7642, loss_dense_depth: 0.8356, loss: 26.8979, grad_norm: 38.4480 -2025-11-12 14:35:15,812 - mmdet - INFO - Iter [173/17500] lr: 1.688e-04, eta: 11:00:02, time: 1.560, data_time: 0.081, memory: 49164, loss_cls_0: 0.8433, loss_box_0: 1.7450, loss_cns_0: 0.6180, loss_yns_0: 0.1583, loss_cls_1: 0.9147, loss_box_1: 1.6870, loss_cns_1: 0.6487, loss_yns_1: 0.1561, loss_cls_2: 0.9443, loss_box_2: 1.6562, loss_cns_2: 0.6548, loss_yns_2: 0.1573, loss_cls_3: 0.9548, loss_box_3: 1.6419, loss_cns_3: 0.6639, loss_yns_3: 0.1591, loss_cls_4: 0.9703, loss_box_4: 1.6325, loss_cns_4: 0.6579, loss_yns_4: 0.1593, loss_cls_5: 0.9630, loss_box_5: 1.6297, loss_cns_5: 0.6604, loss_yns_5: 0.1571, loss_cls_dn_0: 0.2156, loss_box_dn_0: 0.7621, loss_cls_dn_1: 0.1371, loss_box_dn_1: 0.7437, loss_cls_dn_2: 0.1374, loss_box_dn_2: 0.7383, loss_cls_dn_3: 0.1397, loss_box_dn_3: 0.7273, loss_cls_dn_4: 0.1447, loss_box_dn_4: 0.7321, loss_cls_dn_5: 0.1450, loss_box_dn_5: 0.7354, loss_dense_depth: 0.8001, loss: 26.5920, grad_norm: 37.2179 -2025-11-12 14:35:17,384 - mmdet - INFO - Iter [174/17500] lr: 1.692e-04, eta: 10:58:49, time: 1.580, data_time: 0.082, memory: 49164, loss_cls_0: 0.8519, loss_box_0: 1.7489, loss_cns_0: 0.6180, loss_yns_0: 0.1563, loss_cls_1: 0.9127, loss_box_1: 1.7426, loss_cns_1: 0.6445, loss_yns_1: 0.1549, loss_cls_2: 0.9421, loss_box_2: 1.6979, loss_cns_2: 0.6522, loss_yns_2: 0.1557, loss_cls_3: 0.9545, loss_box_3: 1.7166, loss_cns_3: 0.6514, loss_yns_3: 0.1587, loss_cls_4: 0.9562, loss_box_4: 1.7106, loss_cns_4: 0.6531, loss_yns_4: 0.1581, loss_cls_5: 0.9615, loss_box_5: 1.7067, loss_cns_5: 0.6489, loss_yns_5: 0.1561, loss_cls_dn_0: 0.2130, loss_box_dn_0: 0.7678, loss_cls_dn_1: 0.1378, loss_box_dn_1: 0.7465, loss_cls_dn_2: 0.1374, loss_box_dn_2: 0.7360, loss_cls_dn_3: 0.1410, loss_box_dn_3: 0.7471, loss_cls_dn_4: 0.1442, loss_box_dn_4: 0.7515, loss_cls_dn_5: 0.1455, loss_box_dn_5: 0.7633, loss_dense_depth: 0.8370, loss: 26.9779, grad_norm: 43.1406 -2025-11-12 14:35:18,955 - mmdet - INFO - Iter [175/17500] lr: 1.696e-04, eta: 10:57:36, time: 1.569, data_time: 0.073, memory: 49164, loss_cls_0: 0.8807, loss_box_0: 1.7312, loss_cns_0: 0.6233, loss_yns_0: 0.1571, loss_cls_1: 0.9173, loss_box_1: 1.7007, loss_cns_1: 0.6486, loss_yns_1: 0.1539, loss_cls_2: 0.9454, loss_box_2: 1.6680, loss_cns_2: 0.6539, loss_yns_2: 0.1540, loss_cls_3: 0.9553, loss_box_3: 1.6782, loss_cns_3: 0.6576, loss_yns_3: 0.1582, loss_cls_4: 0.9745, loss_box_4: 1.6743, loss_cns_4: 0.6594, loss_yns_4: 0.1554, loss_cls_5: 0.9595, loss_box_5: 1.6766, loss_cns_5: 0.6544, loss_yns_5: 0.1547, loss_cls_dn_0: 0.2168, loss_box_dn_0: 0.7597, loss_cls_dn_1: 0.1349, loss_box_dn_1: 0.7436, loss_cls_dn_2: 0.1325, loss_box_dn_2: 0.7446, loss_cls_dn_3: 0.1375, loss_box_dn_3: 0.7668, loss_cls_dn_4: 0.1428, loss_box_dn_4: 0.7844, loss_cls_dn_5: 0.1450, loss_box_dn_5: 0.8103, loss_dense_depth: 0.7970, loss: 26.9085, grad_norm: 40.4645 -2025-11-12 14:35:20,527 - mmdet - INFO - Iter [176/17500] lr: 1.700e-04, eta: 10:56:25, time: 1.573, data_time: 0.076, memory: 49164, loss_cls_0: 0.8267, loss_box_0: 1.7106, loss_cns_0: 0.6199, loss_yns_0: 0.1540, loss_cls_1: 0.9072, loss_box_1: 1.6966, loss_cns_1: 0.6479, loss_yns_1: 0.1527, loss_cls_2: 0.9455, loss_box_2: 1.6686, loss_cns_2: 0.6541, loss_yns_2: 0.1502, loss_cls_3: 0.9450, loss_box_3: 1.7182, loss_cns_3: 0.6541, loss_yns_3: 0.1535, loss_cls_4: 0.9560, loss_box_4: 1.6986, loss_cns_4: 0.6531, loss_yns_4: 0.1530, loss_cls_5: 0.9507, loss_box_5: 1.7096, loss_cns_5: 0.6496, loss_yns_5: 0.1531, loss_cls_dn_0: 0.2171, loss_box_dn_0: 0.7686, loss_cls_dn_1: 0.1377, loss_box_dn_1: 0.7648, loss_cls_dn_2: 0.1374, loss_box_dn_2: 0.7705, loss_cls_dn_3: 0.1418, loss_box_dn_3: 0.8043, loss_cls_dn_4: 0.1436, loss_box_dn_4: 0.8227, loss_cls_dn_5: 0.1469, loss_box_dn_5: 0.8510, loss_dense_depth: 0.8178, loss: 27.0525, grad_norm: 56.5117 -2025-11-12 14:35:22,092 - mmdet - INFO - Iter [177/17500] lr: 1.704e-04, eta: 10:55:13, time: 1.562, data_time: 0.078, memory: 49164, loss_cls_0: 0.8283, loss_box_0: 1.6885, loss_cns_0: 0.6154, loss_yns_0: 0.1504, loss_cls_1: 0.9111, loss_box_1: 1.6685, loss_cns_1: 0.6489, loss_yns_1: 0.1499, loss_cls_2: 0.9307, loss_box_2: 1.6322, loss_cns_2: 0.6551, loss_yns_2: 0.1495, loss_cls_3: 0.9306, loss_box_3: 1.6506, loss_cns_3: 0.6531, loss_yns_3: 0.1503, loss_cls_4: 0.9424, loss_box_4: 1.6503, loss_cns_4: 0.6552, loss_yns_4: 0.1506, loss_cls_5: 0.9412, loss_box_5: 1.6477, loss_cns_5: 0.6544, loss_yns_5: 0.1512, loss_cls_dn_0: 0.2180, loss_box_dn_0: 0.7591, loss_cls_dn_1: 0.1391, loss_box_dn_1: 0.8003, loss_cls_dn_2: 0.1372, loss_box_dn_2: 0.8002, loss_cls_dn_3: 0.1400, loss_box_dn_3: 0.8223, loss_cls_dn_4: 0.1445, loss_box_dn_4: 0.8429, loss_cls_dn_5: 0.1465, loss_box_dn_5: 0.8622, loss_dense_depth: 0.8530, loss: 26.8714, grad_norm: 41.5855 -2025-11-12 14:35:23,657 - mmdet - INFO - Iter [178/17500] lr: 1.708e-04, eta: 10:54:02, time: 1.564, data_time: 0.079, memory: 49164, loss_cls_0: 0.8378, loss_box_0: 1.6886, loss_cns_0: 0.6017, loss_yns_0: 0.1452, loss_cls_1: 0.9132, loss_box_1: 1.6967, loss_cns_1: 0.6489, loss_yns_1: 0.1490, loss_cls_2: 0.9285, loss_box_2: 1.6944, loss_cns_2: 0.6516, loss_yns_2: 0.1487, loss_cls_3: 0.9401, loss_box_3: 1.6652, loss_cns_3: 0.6543, loss_yns_3: 0.1482, loss_cls_4: 0.9394, loss_box_4: 1.6917, loss_cns_4: 0.6574, loss_yns_4: 0.1488, loss_cls_5: 0.9415, loss_box_5: 1.6993, loss_cns_5: 0.6556, loss_yns_5: 0.1503, loss_cls_dn_0: 0.2175, loss_box_dn_0: 0.7691, loss_cls_dn_1: 0.1394, loss_box_dn_1: 0.8087, loss_cls_dn_2: 0.1377, loss_box_dn_2: 0.8189, loss_cls_dn_3: 0.1417, loss_box_dn_3: 0.8208, loss_cls_dn_4: 0.1469, loss_box_dn_4: 0.8462, loss_cls_dn_5: 0.1512, loss_box_dn_5: 0.8581, loss_dense_depth: 0.7817, loss: 27.0340, grad_norm: 49.9817 -2025-11-12 14:35:25,215 - mmdet - INFO - Iter [179/17500] lr: 1.712e-04, eta: 10:52:51, time: 1.560, data_time: 0.075, memory: 49164, loss_cls_0: 0.8280, loss_box_0: 1.6977, loss_cns_0: 0.6074, loss_yns_0: 0.1462, loss_cls_1: 0.9126, loss_box_1: 1.7017, loss_cns_1: 0.6477, loss_yns_1: 0.1506, loss_cls_2: 0.9433, loss_box_2: 1.7004, loss_cns_2: 0.6491, loss_yns_2: 0.1504, loss_cls_3: 0.9498, loss_box_3: 1.6672, loss_cns_3: 0.6528, loss_yns_3: 0.1511, loss_cls_4: 0.9489, loss_box_4: 1.6659, loss_cns_4: 0.6543, loss_yns_4: 0.1491, loss_cls_5: 0.9494, loss_box_5: 1.6611, loss_cns_5: 0.6521, loss_yns_5: 0.1509, loss_cls_dn_0: 0.2171, loss_box_dn_0: 0.7790, loss_cls_dn_1: 0.1423, loss_box_dn_1: 0.8041, loss_cls_dn_2: 0.1390, loss_box_dn_2: 0.8087, loss_cls_dn_3: 0.1412, loss_box_dn_3: 0.7962, loss_cls_dn_4: 0.1447, loss_box_dn_4: 0.8052, loss_cls_dn_5: 0.1492, loss_box_dn_5: 0.8036, loss_dense_depth: 0.8048, loss: 26.9228, grad_norm: 47.7613 -2025-11-12 14:35:26,790 - mmdet - INFO - Iter [180/17500] lr: 1.716e-04, eta: 10:51:43, time: 1.578, data_time: 0.074, memory: 49164, loss_cls_0: 0.8089, loss_box_0: 1.6899, loss_cns_0: 0.6195, loss_yns_0: 0.1465, loss_cls_1: 0.9120, loss_box_1: 1.6759, loss_cns_1: 0.6511, loss_yns_1: 0.1505, loss_cls_2: 0.9324, loss_box_2: 1.6393, loss_cns_2: 0.6549, loss_yns_2: 0.1491, loss_cls_3: 0.9384, loss_box_3: 1.6362, loss_cns_3: 0.6566, loss_yns_3: 0.1491, loss_cls_4: 0.9462, loss_box_4: 1.6237, loss_cns_4: 0.6579, loss_yns_4: 0.1504, loss_cls_5: 0.9447, loss_box_5: 1.6277, loss_cns_5: 0.6559, loss_yns_5: 0.1479, loss_cls_dn_0: 0.2148, loss_box_dn_0: 0.7757, loss_cls_dn_1: 0.1378, loss_box_dn_1: 0.7854, loss_cls_dn_2: 0.1359, loss_box_dn_2: 0.7752, loss_cls_dn_3: 0.1370, loss_box_dn_3: 0.7664, loss_cls_dn_4: 0.1393, loss_box_dn_4: 0.7697, loss_cls_dn_5: 0.1428, loss_box_dn_5: 0.7701, loss_dense_depth: 0.8001, loss: 26.5151, grad_norm: 43.9413 -2025-11-12 14:35:28,434 - mmdet - INFO - Iter [181/17500] lr: 1.720e-04, eta: 10:50:42, time: 1.644, data_time: 0.101, memory: 49164, loss_cls_0: 0.8121, loss_box_0: 1.7240, loss_cns_0: 0.6188, loss_yns_0: 0.1493, loss_cls_1: 0.9001, loss_box_1: 1.6986, loss_cns_1: 0.6458, loss_yns_1: 0.1496, loss_cls_2: 0.9233, loss_box_2: 1.6365, loss_cns_2: 0.6500, loss_yns_2: 0.1520, loss_cls_3: 0.9261, loss_box_3: 1.6587, loss_cns_3: 0.6515, loss_yns_3: 0.1502, loss_cls_4: 0.9320, loss_box_4: 1.6512, loss_cns_4: 0.6551, loss_yns_4: 0.1519, loss_cls_5: 0.9347, loss_box_5: 1.6678, loss_cns_5: 0.6584, loss_yns_5: 0.1512, loss_cls_dn_0: 0.2155, loss_box_dn_0: 0.7717, loss_cls_dn_1: 0.1309, loss_box_dn_1: 0.7565, loss_cls_dn_2: 0.1328, loss_box_dn_2: 0.7467, loss_cls_dn_3: 0.1344, loss_box_dn_3: 0.7613, loss_cls_dn_4: 0.1399, loss_box_dn_4: 0.7758, loss_cls_dn_5: 0.1474, loss_box_dn_5: 0.7988, loss_dense_depth: 0.7896, loss: 26.5505, grad_norm: 43.3968 -2025-11-12 14:35:30,046 - mmdet - INFO - Iter [182/17500] lr: 1.724e-04, eta: 10:49:39, time: 1.614, data_time: 0.104, memory: 49164, loss_cls_0: 0.8026, loss_box_0: 1.6968, loss_cns_0: 0.6248, loss_yns_0: 0.1467, loss_cls_1: 0.8916, loss_box_1: 1.6710, loss_cns_1: 0.6529, loss_yns_1: 0.1467, loss_cls_2: 0.9100, loss_box_2: 1.6550, loss_cns_2: 0.6514, loss_yns_2: 0.1494, loss_cls_3: 0.9120, loss_box_3: 1.6767, loss_cns_3: 0.6519, loss_yns_3: 0.1489, loss_cls_4: 0.9287, loss_box_4: 1.6804, loss_cns_4: 0.6570, loss_yns_4: 0.1501, loss_cls_5: 0.9161, loss_box_5: 1.6835, loss_cns_5: 0.6529, loss_yns_5: 0.1484, loss_cls_dn_0: 0.2133, loss_box_dn_0: 0.7757, loss_cls_dn_1: 0.1323, loss_box_dn_1: 0.7825, loss_cls_dn_2: 0.1396, loss_box_dn_2: 0.7914, loss_cls_dn_3: 0.1429, loss_box_dn_3: 0.8229, loss_cls_dn_4: 0.1465, loss_box_dn_4: 0.8454, loss_cls_dn_5: 0.1576, loss_box_dn_5: 0.8766, loss_dense_depth: 0.7678, loss: 26.7997, grad_norm: 57.1573 -2025-11-12 14:35:31,622 - mmdet - INFO - Iter [183/17500] lr: 1.728e-04, eta: 10:48:32, time: 1.567, data_time: 0.078, memory: 49164, loss_cls_0: 0.7889, loss_box_0: 1.6876, loss_cns_0: 0.6282, loss_yns_0: 0.1466, loss_cls_1: 0.8900, loss_box_1: 1.6093, loss_cns_1: 0.6620, loss_yns_1: 0.1464, loss_cls_2: 0.9036, loss_box_2: 1.5895, loss_cns_2: 0.6571, loss_yns_2: 0.1454, loss_cls_3: 0.9169, loss_box_3: 1.6156, loss_cns_3: 0.6585, loss_yns_3: 0.1471, loss_cls_4: 0.9117, loss_box_4: 1.6237, loss_cns_4: 0.6566, loss_yns_4: 0.1470, loss_cls_5: 0.9115, loss_box_5: 1.6332, loss_cns_5: 0.6539, loss_yns_5: 0.1468, loss_cls_dn_0: 0.2136, loss_box_dn_0: 0.7627, loss_cls_dn_1: 0.1345, loss_box_dn_1: 0.7909, loss_cls_dn_2: 0.1304, loss_box_dn_2: 0.7987, loss_cls_dn_3: 0.1336, loss_box_dn_3: 0.8364, loss_cls_dn_4: 0.1368, loss_box_dn_4: 0.8588, loss_cls_dn_5: 0.1410, loss_box_dn_5: 0.8931, loss_dense_depth: 0.7827, loss: 26.4903, grad_norm: 56.6244 -2025-11-12 14:35:33,199 - mmdet - INFO - Iter [184/17500] lr: 1.732e-04, eta: 10:47:27, time: 1.585, data_time: 0.084, memory: 49164, loss_cls_0: 0.8103, loss_box_0: 1.7146, loss_cns_0: 0.6242, loss_yns_0: 0.1444, loss_cls_1: 0.9086, loss_box_1: 1.6193, loss_cns_1: 0.6601, loss_yns_1: 0.1471, loss_cls_2: 0.9133, loss_box_2: 1.6047, loss_cns_2: 0.6566, loss_yns_2: 0.1460, loss_cls_3: 0.9269, loss_box_3: 1.6120, loss_cns_3: 0.6569, loss_yns_3: 0.1450, loss_cls_4: 0.9193, loss_box_4: 1.6089, loss_cns_4: 0.6579, loss_yns_4: 0.1475, loss_cls_5: 0.9271, loss_box_5: 1.6295, loss_cns_5: 0.6589, loss_yns_5: 0.1461, loss_cls_dn_0: 0.2176, loss_box_dn_0: 0.7678, loss_cls_dn_1: 0.1348, loss_box_dn_1: 0.8074, loss_cls_dn_2: 0.1303, loss_box_dn_2: 0.8179, loss_cls_dn_3: 0.1387, loss_box_dn_3: 0.8523, loss_cls_dn_4: 0.1383, loss_box_dn_4: 0.8702, loss_cls_dn_5: 0.1418, loss_box_dn_5: 0.9036, loss_dense_depth: 0.7828, loss: 26.6887, grad_norm: 44.5095 -2025-11-12 14:35:34,799 - mmdet - INFO - Iter [185/17500] lr: 1.736e-04, eta: 10:46:25, time: 1.602, data_time: 0.079, memory: 49164, loss_cls_0: 0.8079, loss_box_0: 1.7026, loss_cns_0: 0.6233, loss_yns_0: 0.1456, loss_cls_1: 0.8949, loss_box_1: 1.6118, loss_cns_1: 0.6582, loss_yns_1: 0.1480, loss_cls_2: 0.9022, loss_box_2: 1.6135, loss_cns_2: 0.6575, loss_yns_2: 0.1486, loss_cls_3: 0.9164, loss_box_3: 1.6055, loss_cns_3: 0.6595, loss_yns_3: 0.1465, loss_cls_4: 0.9197, loss_box_4: 1.5941, loss_cns_4: 0.6609, loss_yns_4: 0.1509, loss_cls_5: 0.9233, loss_box_5: 1.5977, loss_cns_5: 0.6627, loss_yns_5: 0.1478, loss_cls_dn_0: 0.2179, loss_box_dn_0: 0.7664, loss_cls_dn_1: 0.1319, loss_box_dn_1: 0.8453, loss_cls_dn_2: 0.1328, loss_box_dn_2: 0.8541, loss_cls_dn_3: 0.1428, loss_box_dn_3: 0.8658, loss_cls_dn_4: 0.1455, loss_box_dn_4: 0.8729, loss_cls_dn_5: 0.1516, loss_box_dn_5: 0.8918, loss_dense_depth: 0.8024, loss: 26.7206, grad_norm: 59.5586 -2025-11-12 14:35:37,870 - mmdet - INFO - Iter [186/17500] lr: 1.740e-04, eta: 10:47:40, time: 3.071, data_time: 0.098, memory: 49164, loss_cls_0: 0.8024, loss_box_0: 1.6789, loss_cns_0: 0.6273, loss_yns_0: 0.1477, loss_cls_1: 0.8934, loss_box_1: 1.6105, loss_cns_1: 0.6603, loss_yns_1: 0.1507, loss_cls_2: 0.9044, loss_box_2: 1.5981, loss_cns_2: 0.6585, loss_yns_2: 0.1504, loss_cls_3: 0.9241, loss_box_3: 1.5746, loss_cns_3: 0.6630, loss_yns_3: 0.1474, loss_cls_4: 0.9193, loss_box_4: 1.5643, loss_cns_4: 0.6649, loss_yns_4: 0.1511, loss_cls_5: 0.9174, loss_box_5: 1.5733, loss_cns_5: 0.6658, loss_yns_5: 0.1505, loss_cls_dn_0: 0.2129, loss_box_dn_0: 0.7644, loss_cls_dn_1: 0.1349, loss_box_dn_1: 0.7878, loss_cls_dn_2: 0.1381, loss_box_dn_2: 0.7813, loss_cls_dn_3: 0.1408, loss_box_dn_3: 0.7726, loss_cls_dn_4: 0.1426, loss_box_dn_4: 0.7749, loss_cls_dn_5: 0.1476, loss_box_dn_5: 0.7870, loss_dense_depth: 0.7892, loss: 26.1724, grad_norm: 47.8700 -2025-11-12 14:35:39,391 - mmdet - INFO - Iter [187/17500] lr: 1.744e-04, eta: 10:46:31, time: 1.517, data_time: 0.044, memory: 49164, loss_cls_0: 0.8180, loss_box_0: 1.6909, loss_cns_0: 0.6228, loss_yns_0: 0.1481, loss_cls_1: 0.9142, loss_box_1: 1.6285, loss_cns_1: 0.6590, loss_yns_1: 0.1486, loss_cls_2: 0.9286, loss_box_2: 1.6017, loss_cns_2: 0.6598, loss_yns_2: 0.1495, loss_cls_3: 0.9383, loss_box_3: 1.6247, loss_cns_3: 0.6593, loss_yns_3: 0.1484, loss_cls_4: 0.9469, loss_box_4: 1.5963, loss_cns_4: 0.6632, loss_yns_4: 0.1494, loss_cls_5: 0.9424, loss_box_5: 1.5953, loss_cns_5: 0.6598, loss_yns_5: 0.1542, loss_cls_dn_0: 0.2167, loss_box_dn_0: 0.7733, loss_cls_dn_1: 0.1332, loss_box_dn_1: 0.7615, loss_cls_dn_2: 0.1360, loss_box_dn_2: 0.7495, loss_cls_dn_3: 0.1353, loss_box_dn_3: 0.7572, loss_cls_dn_4: 0.1403, loss_box_dn_4: 0.7510, loss_cls_dn_5: 0.1438, loss_box_dn_5: 0.7561, loss_dense_depth: 0.7732, loss: 26.2749, grad_norm: 46.5540 -2025-11-12 14:35:40,968 - mmdet - INFO - Iter [188/17500] lr: 1.748e-04, eta: 10:45:28, time: 1.581, data_time: 0.081, memory: 49164, loss_cls_0: 0.8114, loss_box_0: 1.6895, loss_cns_0: 0.6197, loss_yns_0: 0.1470, loss_cls_1: 0.9061, loss_box_1: 1.6452, loss_cns_1: 0.6532, loss_yns_1: 0.1470, loss_cls_2: 0.9239, loss_box_2: 1.6345, loss_cns_2: 0.6528, loss_yns_2: 0.1457, loss_cls_3: 0.9420, loss_box_3: 1.6500, loss_cns_3: 0.6529, loss_yns_3: 0.1481, loss_cls_4: 0.9485, loss_box_4: 1.6228, loss_cns_4: 0.6544, loss_yns_4: 0.1479, loss_cls_5: 0.9467, loss_box_5: 1.6173, loss_cns_5: 0.6529, loss_yns_5: 0.1475, loss_cls_dn_0: 0.2139, loss_box_dn_0: 0.7695, loss_cls_dn_1: 0.1339, loss_box_dn_1: 0.7245, loss_cls_dn_2: 0.1349, loss_box_dn_2: 0.7203, loss_cls_dn_3: 0.1392, loss_box_dn_3: 0.7464, loss_cls_dn_4: 0.1479, loss_box_dn_4: 0.7474, loss_cls_dn_5: 0.1543, loss_box_dn_5: 0.7642, loss_dense_depth: 0.7784, loss: 26.2820, grad_norm: 51.5995 -2025-11-12 14:35:42,569 - mmdet - INFO - Iter [189/17500] lr: 1.752e-04, eta: 10:44:27, time: 1.600, data_time: 0.074, memory: 49164, loss_cls_0: 0.8113, loss_box_0: 1.6819, loss_cns_0: 0.6248, loss_yns_0: 0.1483, loss_cls_1: 0.8991, loss_box_1: 1.6672, loss_cns_1: 0.6528, loss_yns_1: 0.1532, loss_cls_2: 0.9139, loss_box_2: 1.6739, loss_cns_2: 0.6502, loss_yns_2: 0.1516, loss_cls_3: 0.9402, loss_box_3: 1.6526, loss_cns_3: 0.6531, loss_yns_3: 0.1528, loss_cls_4: 0.9388, loss_box_4: 1.6664, loss_cns_4: 0.6527, loss_yns_4: 0.1547, loss_cls_5: 0.9374, loss_box_5: 1.6670, loss_cns_5: 0.6565, loss_yns_5: 0.1527, loss_cls_dn_0: 0.2168, loss_box_dn_0: 0.7676, loss_cls_dn_1: 0.1361, loss_box_dn_1: 0.7446, loss_cls_dn_2: 0.1360, loss_box_dn_2: 0.7612, loss_cls_dn_3: 0.1466, loss_box_dn_3: 0.7821, loss_cls_dn_4: 0.1507, loss_box_dn_4: 0.8097, loss_cls_dn_5: 0.1537, loss_box_dn_5: 0.8380, loss_dense_depth: 0.7862, loss: 26.6828, grad_norm: 57.1801 -2025-11-12 14:35:44,167 - mmdet - INFO - Iter [190/17500] lr: 1.755e-04, eta: 10:43:26, time: 1.593, data_time: 0.076, memory: 49164, loss_cls_0: 0.8316, loss_box_0: 1.7194, loss_cns_0: 0.6223, loss_yns_0: 0.1519, loss_cls_1: 0.9020, loss_box_1: 1.6910, loss_cns_1: 0.6495, loss_yns_1: 0.1562, loss_cls_2: 0.9199, loss_box_2: 1.6864, loss_cns_2: 0.6504, loss_yns_2: 0.1538, loss_cls_3: 0.9271, loss_box_3: 1.6548, loss_cns_3: 0.6529, loss_yns_3: 0.1517, loss_cls_4: 0.9317, loss_box_4: 1.6995, loss_cns_4: 0.6554, loss_yns_4: 0.1567, loss_cls_5: 0.9613, loss_box_5: 1.6949, loss_cns_5: 0.6556, loss_yns_5: 0.1581, loss_cls_dn_0: 0.2206, loss_box_dn_0: 0.7757, loss_cls_dn_1: 0.1379, loss_box_dn_1: 0.7860, loss_cls_dn_2: 0.1382, loss_box_dn_2: 0.8049, loss_cls_dn_3: 0.1439, loss_box_dn_3: 0.8194, loss_cls_dn_4: 0.1477, loss_box_dn_4: 0.8635, loss_cls_dn_5: 0.1551, loss_box_dn_5: 0.8898, loss_dense_depth: 0.8010, loss: 27.1178, grad_norm: 56.9233 -2025-11-12 14:35:45,759 - mmdet - INFO - Iter [191/17500] lr: 1.759e-04, eta: 10:42:26, time: 1.589, data_time: 0.080, memory: 49164, loss_cls_0: 0.8259, loss_box_0: 1.7282, loss_cns_0: 0.6219, loss_yns_0: 0.1508, loss_cls_1: 0.9091, loss_box_1: 1.6745, loss_cns_1: 0.6537, loss_yns_1: 0.1527, loss_cls_2: 0.9322, loss_box_2: 1.6575, loss_cns_2: 0.6530, loss_yns_2: 0.1535, loss_cls_3: 0.9273, loss_box_3: 1.6464, loss_cns_3: 0.6542, loss_yns_3: 0.1522, loss_cls_4: 0.9465, loss_box_4: 1.6692, loss_cns_4: 0.6596, loss_yns_4: 0.1541, loss_cls_5: 0.9563, loss_box_5: 1.6561, loss_cns_5: 0.6540, loss_yns_5: 0.1596, loss_cls_dn_0: 0.2198, loss_box_dn_0: 0.7662, loss_cls_dn_1: 0.1395, loss_box_dn_1: 0.8197, loss_cls_dn_2: 0.1379, loss_box_dn_2: 0.8282, loss_cls_dn_3: 0.1417, loss_box_dn_3: 0.8388, loss_cls_dn_4: 0.1457, loss_box_dn_4: 0.8703, loss_cls_dn_5: 0.1533, loss_box_dn_5: 0.8876, loss_dense_depth: 0.7595, loss: 27.0568, grad_norm: 53.4346 -2025-11-12 14:35:47,336 - mmdet - INFO - Iter [192/17500] lr: 1.763e-04, eta: 10:41:26, time: 1.581, data_time: 0.078, memory: 49164, loss_cls_0: 0.8083, loss_box_0: 1.7039, loss_cns_0: 0.6232, loss_yns_0: 0.1500, loss_cls_1: 0.8974, loss_box_1: 1.6757, loss_cns_1: 0.6526, loss_yns_1: 0.1510, loss_cls_2: 0.9183, loss_box_2: 1.6531, loss_cns_2: 0.6550, loss_yns_2: 0.1513, loss_cls_3: 0.9314, loss_box_3: 1.6416, loss_cns_3: 0.6545, loss_yns_3: 0.1535, loss_cls_4: 0.9350, loss_box_4: 1.6430, loss_cns_4: 0.6598, loss_yns_4: 0.1525, loss_cls_5: 0.9327, loss_box_5: 1.6519, loss_cns_5: 0.6552, loss_yns_5: 0.1544, loss_cls_dn_0: 0.2167, loss_box_dn_0: 0.7555, loss_cls_dn_1: 0.1388, loss_box_dn_1: 0.8347, loss_cls_dn_2: 0.1380, loss_box_dn_2: 0.8324, loss_cls_dn_3: 0.1466, loss_box_dn_3: 0.8327, loss_cls_dn_4: 0.1466, loss_box_dn_4: 0.8417, loss_cls_dn_5: 0.1508, loss_box_dn_5: 0.8589, loss_dense_depth: 0.7600, loss: 26.8591, grad_norm: 43.4390 -2025-11-12 14:35:48,936 - mmdet - INFO - Iter [193/17500] lr: 1.767e-04, eta: 10:40:26, time: 1.581, data_time: 0.076, memory: 49164, loss_cls_0: 0.7956, loss_box_0: 1.6835, loss_cns_0: 0.6275, loss_yns_0: 0.1487, loss_cls_1: 0.8804, loss_box_1: 1.6520, loss_cns_1: 0.6543, loss_yns_1: 0.1500, loss_cls_2: 0.9138, loss_box_2: 1.6234, loss_cns_2: 0.6572, loss_yns_2: 0.1518, loss_cls_3: 0.9150, loss_box_3: 1.6263, loss_cns_3: 0.6605, loss_yns_3: 0.1529, loss_cls_4: 0.9212, loss_box_4: 1.6183, loss_cns_4: 0.6596, loss_yns_4: 0.1573, loss_cls_5: 0.9399, loss_box_5: 1.6161, loss_cns_5: 0.6588, loss_yns_5: 0.1512, loss_cls_dn_0: 0.2149, loss_box_dn_0: 0.7654, loss_cls_dn_1: 0.1324, loss_box_dn_1: 0.7692, loss_cls_dn_2: 0.1432, loss_box_dn_2: 0.7595, loss_cls_dn_3: 0.1433, loss_box_dn_3: 0.7571, loss_cls_dn_4: 0.1412, loss_box_dn_4: 0.7557, loss_cls_dn_5: 0.1469, loss_box_dn_5: 0.7620, loss_dense_depth: 0.7456, loss: 26.2518, grad_norm: 50.2860 -2025-11-12 14:35:50,519 - mmdet - INFO - Iter [194/17500] lr: 1.771e-04, eta: 10:39:28, time: 1.599, data_time: 0.093, memory: 49164, loss_cls_0: 0.8207, loss_box_0: 1.6771, loss_cns_0: 0.6186, loss_yns_0: 0.1498, loss_cls_1: 0.9083, loss_box_1: 1.6844, loss_cns_1: 0.6505, loss_yns_1: 0.1547, loss_cls_2: 0.9380, loss_box_2: 1.6453, loss_cns_2: 0.6543, loss_yns_2: 0.1601, loss_cls_3: 0.9359, loss_box_3: 1.6586, loss_cns_3: 0.6579, loss_yns_3: 0.1564, loss_cls_4: 0.9413, loss_box_4: 1.6625, loss_cns_4: 0.6569, loss_yns_4: 0.1582, loss_cls_5: 0.9590, loss_box_5: 1.6475, loss_cns_5: 0.6510, loss_yns_5: 0.1543, loss_cls_dn_0: 0.2176, loss_box_dn_0: 0.7575, loss_cls_dn_1: 0.1304, loss_box_dn_1: 0.7479, loss_cls_dn_2: 0.1410, loss_box_dn_2: 0.7337, loss_cls_dn_3: 0.1363, loss_box_dn_3: 0.7375, loss_cls_dn_4: 0.1373, loss_box_dn_4: 0.7393, loss_cls_dn_5: 0.1429, loss_box_dn_5: 0.7401, loss_dense_depth: 0.7700, loss: 26.4329, grad_norm: 41.6307 -2025-11-12 14:35:52,166 - mmdet - INFO - Iter [195/17500] lr: 1.775e-04, eta: 10:38:35, time: 1.646, data_time: 0.080, memory: 49164, loss_cls_0: 0.7831, loss_box_0: 1.6589, loss_cns_0: 0.6252, loss_yns_0: 0.1442, loss_cls_1: 0.8812, loss_box_1: 1.6154, loss_cns_1: 0.6550, loss_yns_1: 0.1497, loss_cls_2: 0.9070, loss_box_2: 1.6172, loss_cns_2: 0.6541, loss_yns_2: 0.1465, loss_cls_3: 0.9048, loss_box_3: 1.6183, loss_cns_3: 0.6573, loss_yns_3: 0.1488, loss_cls_4: 0.9073, loss_box_4: 1.6288, loss_cns_4: 0.6564, loss_yns_4: 0.1478, loss_cls_5: 0.9322, loss_box_5: 1.6225, loss_cns_5: 0.6501, loss_yns_5: 0.1456, loss_cls_dn_0: 0.2040, loss_box_dn_0: 0.7525, loss_cls_dn_1: 0.1258, loss_box_dn_1: 0.7334, loss_cls_dn_2: 0.1256, loss_box_dn_2: 0.7239, loss_cls_dn_3: 0.1265, loss_box_dn_3: 0.7381, loss_cls_dn_4: 0.1320, loss_box_dn_4: 0.7538, loss_cls_dn_5: 0.1349, loss_box_dn_5: 0.7743, loss_dense_depth: 0.7389, loss: 25.9210, grad_norm: 50.0218 -2025-11-12 14:35:53,790 - mmdet - INFO - Iter [196/17500] lr: 1.779e-04, eta: 10:37:41, time: 1.629, data_time: 0.078, memory: 49164, loss_cls_0: 0.7677, loss_box_0: 1.6644, loss_cns_0: 0.6263, loss_yns_0: 0.1460, loss_cls_1: 0.8644, loss_box_1: 1.6271, loss_cns_1: 0.6524, loss_yns_1: 0.1458, loss_cls_2: 0.9019, loss_box_2: 1.6227, loss_cns_2: 0.6550, loss_yns_2: 0.1486, loss_cls_3: 0.8973, loss_box_3: 1.6105, loss_cns_3: 0.6571, loss_yns_3: 0.1483, loss_cls_4: 0.9026, loss_box_4: 1.5969, loss_cns_4: 0.6594, loss_yns_4: 0.1491, loss_cls_5: 0.9037, loss_box_5: 1.5992, loss_cns_5: 0.6551, loss_yns_5: 0.1463, loss_cls_dn_0: 0.2049, loss_box_dn_0: 0.7608, loss_cls_dn_1: 0.1252, loss_box_dn_1: 0.7624, loss_cls_dn_2: 0.1280, loss_box_dn_2: 0.7639, loss_cls_dn_3: 0.1332, loss_box_dn_3: 0.7839, loss_cls_dn_4: 0.1334, loss_box_dn_4: 0.7978, loss_cls_dn_5: 0.1413, loss_box_dn_5: 0.8220, loss_dense_depth: 0.7425, loss: 26.0469, grad_norm: 41.0392 -2025-11-12 14:35:55,388 - mmdet - INFO - Iter [197/17500] lr: 1.783e-04, eta: 10:36:45, time: 1.601, data_time: 0.075, memory: 49164, loss_cls_0: 0.8039, loss_box_0: 1.7266, loss_cns_0: 0.6236, loss_yns_0: 0.1493, loss_cls_1: 0.8751, loss_box_1: 1.6712, loss_cns_1: 0.6550, loss_yns_1: 0.1493, loss_cls_2: 0.8991, loss_box_2: 1.6502, loss_cns_2: 0.6599, loss_yns_2: 0.1522, loss_cls_3: 0.9074, loss_box_3: 1.6927, loss_cns_3: 0.6571, loss_yns_3: 0.1526, loss_cls_4: 0.9078, loss_box_4: 1.6818, loss_cns_4: 0.6595, loss_yns_4: 0.1510, loss_cls_5: 0.9163, loss_box_5: 1.7063, loss_cns_5: 0.6556, loss_yns_5: 0.1494, loss_cls_dn_0: 0.2146, loss_box_dn_0: 0.7541, loss_cls_dn_1: 0.1254, loss_box_dn_1: 0.7825, loss_cls_dn_2: 0.1270, loss_box_dn_2: 0.7914, loss_cls_dn_3: 0.1303, loss_box_dn_3: 0.8311, loss_cls_dn_4: 0.1358, loss_box_dn_4: 0.8531, loss_cls_dn_5: 0.1455, loss_box_dn_5: 0.8863, loss_dense_depth: 0.7817, loss: 26.8114, grad_norm: 52.2197 -2025-11-12 14:35:56,959 - mmdet - INFO - Iter [198/17500] lr: 1.787e-04, eta: 10:35:48, time: 1.573, data_time: 0.076, memory: 49164, loss_cls_0: 0.8176, loss_box_0: 1.7421, loss_cns_0: 0.6204, loss_yns_0: 0.1522, loss_cls_1: 0.8873, loss_box_1: 1.6629, loss_cns_1: 0.6514, loss_yns_1: 0.1525, loss_cls_2: 0.9204, loss_box_2: 1.6378, loss_cns_2: 0.6547, loss_yns_2: 0.1511, loss_cls_3: 0.9149, loss_box_3: 1.6698, loss_cns_3: 0.6530, loss_yns_3: 0.1540, loss_cls_4: 0.9115, loss_box_4: 1.6683, loss_cns_4: 0.6556, loss_yns_4: 0.1518, loss_cls_5: 0.9122, loss_box_5: 1.6826, loss_cns_5: 0.6553, loss_yns_5: 0.1519, loss_cls_dn_0: 0.2129, loss_box_dn_0: 0.7659, loss_cls_dn_1: 0.1311, loss_box_dn_1: 0.8591, loss_cls_dn_2: 0.1322, loss_box_dn_2: 0.8632, loss_cls_dn_3: 0.1342, loss_box_dn_3: 0.8939, loss_cls_dn_4: 0.1391, loss_box_dn_4: 0.9135, loss_cls_dn_5: 0.1450, loss_box_dn_5: 0.9355, loss_dense_depth: 0.7659, loss: 27.1227, grad_norm: 47.0641 -2025-11-12 14:35:58,536 - mmdet - INFO - Iter [199/17500] lr: 1.791e-04, eta: 10:34:50, time: 1.570, data_time: 0.077, memory: 49164, loss_cls_0: 0.7994, loss_box_0: 1.7269, loss_cns_0: 0.6194, loss_yns_0: 0.1543, loss_cls_1: 0.8756, loss_box_1: 1.6851, loss_cns_1: 0.6493, loss_yns_1: 0.1553, loss_cls_2: 0.9036, loss_box_2: 1.6532, loss_cns_2: 0.6529, loss_yns_2: 0.1534, loss_cls_3: 0.9082, loss_box_3: 1.6469, loss_cns_3: 0.6553, loss_yns_3: 0.1537, loss_cls_4: 0.9073, loss_box_4: 1.6566, loss_cns_4: 0.6552, loss_yns_4: 0.1539, loss_cls_5: 0.9185, loss_box_5: 1.6477, loss_cns_5: 0.6561, loss_yns_5: 0.1542, loss_cls_dn_0: 0.2085, loss_box_dn_0: 0.7551, loss_cls_dn_1: 0.1305, loss_box_dn_1: 0.8643, loss_cls_dn_2: 0.1315, loss_box_dn_2: 0.8569, loss_cls_dn_3: 0.1331, loss_box_dn_3: 0.8649, loss_cls_dn_4: 0.1358, loss_box_dn_4: 0.8738, loss_cls_dn_5: 0.1380, loss_box_dn_5: 0.8782, loss_dense_depth: 0.8040, loss: 26.9164, grad_norm: 33.4723 -2025-11-12 14:36:00,125 - mmdet - INFO - Iter [200/17500] lr: 1.795e-04, eta: 10:33:54, time: 1.577, data_time: 0.081, memory: 49164, loss_cls_0: 0.7939, loss_box_0: 1.7390, loss_cns_0: 0.6151, loss_yns_0: 0.1511, loss_cls_1: 0.8784, loss_box_1: 1.6876, loss_cns_1: 0.6498, loss_yns_1: 0.1540, loss_cls_2: 0.8958, loss_box_2: 1.6562, loss_cns_2: 0.6527, loss_yns_2: 0.1573, loss_cls_3: 0.9035, loss_box_3: 1.6465, loss_cns_3: 0.6552, loss_yns_3: 0.1547, loss_cls_4: 0.9053, loss_box_4: 1.6338, loss_cns_4: 0.6570, loss_yns_4: 0.1546, loss_cls_5: 0.9158, loss_box_5: 1.6435, loss_cns_5: 0.6553, loss_yns_5: 0.1536, loss_cls_dn_0: 0.2094, loss_box_dn_0: 0.7577, loss_cls_dn_1: 0.1274, loss_box_dn_1: 0.8459, loss_cls_dn_2: 0.1303, loss_box_dn_2: 0.8331, loss_cls_dn_3: 0.1354, loss_box_dn_3: 0.8300, loss_cls_dn_4: 0.1333, loss_box_dn_4: 0.8279, loss_cls_dn_5: 0.1352, loss_box_dn_5: 0.8323, loss_dense_depth: 0.8558, loss: 26.7632, grad_norm: 42.0511 -2025-11-12 14:36:01,758 - mmdet - INFO - Iter [201/17500] lr: 1.799e-04, eta: 10:33:05, time: 1.651, data_time: 0.117, memory: 49164, loss_cls_0: 0.7989, loss_box_0: 1.6931, loss_cns_0: 0.6148, loss_yns_0: 0.1489, loss_cls_1: 0.8681, loss_box_1: 1.6650, loss_cns_1: 0.6461, loss_yns_1: 0.1530, loss_cls_2: 0.8778, loss_box_2: 1.6623, loss_cns_2: 0.6481, loss_yns_2: 0.1543, loss_cls_3: 0.8949, loss_box_3: 1.6443, loss_cns_3: 0.6505, loss_yns_3: 0.1550, loss_cls_4: 0.8983, loss_box_4: 1.6387, loss_cns_4: 0.6560, loss_yns_4: 0.1537, loss_cls_5: 0.9133, loss_box_5: 1.6563, loss_cns_5: 0.6505, loss_yns_5: 0.1515, loss_cls_dn_0: 0.2046, loss_box_dn_0: 0.7516, loss_cls_dn_1: 0.1225, loss_box_dn_1: 0.8066, loss_cls_dn_2: 0.1245, loss_box_dn_2: 0.7939, loss_cls_dn_3: 0.1259, loss_box_dn_3: 0.7956, loss_cls_dn_4: 0.1276, loss_box_dn_4: 0.7980, loss_cls_dn_5: 0.1314, loss_box_dn_5: 0.8119, loss_dense_depth: 0.8127, loss: 26.4001, grad_norm: 36.5621 -2025-11-12 14:36:03,379 - mmdet - INFO - Iter [202/17500] lr: 1.803e-04, eta: 10:32:13, time: 1.618, data_time: 0.102, memory: 49164, loss_cls_0: 0.7868, loss_box_0: 1.6760, loss_cns_0: 0.6200, loss_yns_0: 0.1504, loss_cls_1: 0.8803, loss_box_1: 1.6427, loss_cns_1: 0.6495, loss_yns_1: 0.1494, loss_cls_2: 0.8897, loss_box_2: 1.6288, loss_cns_2: 0.6504, loss_yns_2: 0.1506, loss_cls_3: 0.8974, loss_box_3: 1.6214, loss_cns_3: 0.6524, loss_yns_3: 0.1530, loss_cls_4: 0.8995, loss_box_4: 1.6102, loss_cns_4: 0.6508, loss_yns_4: 0.1505, loss_cls_5: 0.8988, loss_box_5: 1.6170, loss_cns_5: 0.6487, loss_yns_5: 0.1521, loss_cls_dn_0: 0.2019, loss_box_dn_0: 0.7588, loss_cls_dn_1: 0.1246, loss_box_dn_1: 0.7673, loss_cls_dn_2: 0.1268, loss_box_dn_2: 0.7611, loss_cls_dn_3: 0.1256, loss_box_dn_3: 0.7797, loss_cls_dn_4: 0.1332, loss_box_dn_4: 0.7947, loss_cls_dn_5: 0.1361, loss_box_dn_5: 0.8223, loss_dense_depth: 0.7788, loss: 26.1373, grad_norm: 36.6998 -2025-11-12 14:36:04,963 - mmdet - INFO - Iter [203/17500] lr: 1.807e-04, eta: 10:31:19, time: 1.580, data_time: 0.083, memory: 49164, loss_cls_0: 0.7942, loss_box_0: 1.7074, loss_cns_0: 0.6223, loss_yns_0: 0.1516, loss_cls_1: 0.8797, loss_box_1: 1.7010, loss_cns_1: 0.6507, loss_yns_1: 0.1491, loss_cls_2: 0.8975, loss_box_2: 1.6925, loss_cns_2: 0.6521, loss_yns_2: 0.1507, loss_cls_3: 0.8986, loss_box_3: 1.6879, loss_cns_3: 0.6536, loss_yns_3: 0.1500, loss_cls_4: 0.8973, loss_box_4: 1.6776, loss_cns_4: 0.6529, loss_yns_4: 0.1495, loss_cls_5: 0.9194, loss_box_5: 1.6918, loss_cns_5: 0.6519, loss_yns_5: 0.1506, loss_cls_dn_0: 0.2038, loss_box_dn_0: 0.7606, loss_cls_dn_1: 0.1243, loss_box_dn_1: 0.7971, loss_cls_dn_2: 0.1235, loss_box_dn_2: 0.7969, loss_cls_dn_3: 0.1244, loss_box_dn_3: 0.8163, loss_cls_dn_4: 0.1318, loss_box_dn_4: 0.8337, loss_cls_dn_5: 0.1357, loss_box_dn_5: 0.8681, loss_dense_depth: 0.8063, loss: 26.7525, grad_norm: 41.6893 -2025-11-12 14:36:06,563 - mmdet - INFO - Iter [204/17500] lr: 1.811e-04, eta: 10:30:27, time: 1.606, data_time: 0.081, memory: 49164, loss_cls_0: 0.7967, loss_box_0: 1.7436, loss_cns_0: 0.6254, loss_yns_0: 0.1503, loss_cls_1: 0.8746, loss_box_1: 1.7119, loss_cns_1: 0.6563, loss_yns_1: 0.1499, loss_cls_2: 0.8894, loss_box_2: 1.7196, loss_cns_2: 0.6541, loss_yns_2: 0.1515, loss_cls_3: 0.8924, loss_box_3: 1.6877, loss_cns_3: 0.6568, loss_yns_3: 0.1494, loss_cls_4: 0.8890, loss_box_4: 1.6823, loss_cns_4: 0.6588, loss_yns_4: 0.1507, loss_cls_5: 0.9014, loss_box_5: 1.7048, loss_cns_5: 0.6552, loss_yns_5: 0.1485, loss_cls_dn_0: 0.2065, loss_box_dn_0: 0.7691, loss_cls_dn_1: 0.1226, loss_box_dn_1: 0.7746, loss_cls_dn_2: 0.1232, loss_box_dn_2: 0.7914, loss_cls_dn_3: 0.1274, loss_box_dn_3: 0.8020, loss_cls_dn_4: 0.1352, loss_box_dn_4: 0.8258, loss_cls_dn_5: 0.1398, loss_box_dn_5: 0.8669, loss_dense_depth: 0.7848, loss: 26.7698, grad_norm: 47.5823 -2025-11-12 14:36:08,162 - mmdet - INFO - Iter [205/17500] lr: 1.815e-04, eta: 10:29:35, time: 1.598, data_time: 0.080, memory: 49164, loss_cls_0: 0.7944, loss_box_0: 1.7530, loss_cns_0: 0.6249, loss_yns_0: 0.1512, loss_cls_1: 0.8749, loss_box_1: 1.6576, loss_cns_1: 0.6554, loss_yns_1: 0.1534, loss_cls_2: 0.8900, loss_box_2: 1.6482, loss_cns_2: 0.6565, loss_yns_2: 0.1515, loss_cls_3: 0.9068, loss_box_3: 1.6040, loss_cns_3: 0.6563, loss_yns_3: 0.1497, loss_cls_4: 0.8946, loss_box_4: 1.6093, loss_cns_4: 0.6625, loss_yns_4: 0.1521, loss_cls_5: 0.8982, loss_box_5: 1.6150, loss_cns_5: 0.6601, loss_yns_5: 0.1507, loss_cls_dn_0: 0.2119, loss_box_dn_0: 0.7727, loss_cls_dn_1: 0.1300, loss_box_dn_1: 0.7736, loss_cls_dn_2: 0.1316, loss_box_dn_2: 0.7790, loss_cls_dn_3: 0.1348, loss_box_dn_3: 0.7779, loss_cls_dn_4: 0.1379, loss_box_dn_4: 0.7942, loss_cls_dn_5: 0.1419, loss_box_dn_5: 0.8159, loss_dense_depth: 0.7694, loss: 26.3415, grad_norm: 34.0415 -2025-11-12 14:36:09,769 - mmdet - INFO - Iter [206/17500] lr: 1.819e-04, eta: 10:28:45, time: 1.608, data_time: 0.106, memory: 49164, loss_cls_0: 0.7922, loss_box_0: 1.7497, loss_cns_0: 0.6219, loss_yns_0: 0.1517, loss_cls_1: 0.8773, loss_box_1: 1.6726, loss_cns_1: 0.6573, loss_yns_1: 0.1508, loss_cls_2: 0.8942, loss_box_2: 1.6314, loss_cns_2: 0.6568, loss_yns_2: 0.1509, loss_cls_3: 0.8955, loss_box_3: 1.6357, loss_cns_3: 0.6563, loss_yns_3: 0.1500, loss_cls_4: 0.8972, loss_box_4: 1.6219, loss_cns_4: 0.6558, loss_yns_4: 0.1502, loss_cls_5: 0.9125, loss_box_5: 1.6269, loss_cns_5: 0.6569, loss_yns_5: 0.1498, loss_cls_dn_0: 0.2166, loss_box_dn_0: 0.7699, loss_cls_dn_1: 0.1275, loss_box_dn_1: 0.7452, loss_cls_dn_2: 0.1325, loss_box_dn_2: 0.7309, loss_cls_dn_3: 0.1359, loss_box_dn_3: 0.7342, loss_cls_dn_4: 0.1397, loss_box_dn_4: 0.7375, loss_cls_dn_5: 0.1456, loss_box_dn_5: 0.7465, loss_dense_depth: 0.7784, loss: 26.1559, grad_norm: 38.2289 -2025-11-12 14:36:11,358 - mmdet - INFO - Iter [207/17500] lr: 1.823e-04, eta: 10:27:52, time: 1.581, data_time: 0.079, memory: 49164, loss_cls_0: 0.7986, loss_box_0: 1.7164, loss_cns_0: 0.6216, loss_yns_0: 0.1537, loss_cls_1: 0.8727, loss_box_1: 1.6712, loss_cns_1: 0.6543, loss_yns_1: 0.1487, loss_cls_2: 0.8885, loss_box_2: 1.6332, loss_cns_2: 0.6558, loss_yns_2: 0.1518, loss_cls_3: 0.8918, loss_box_3: 1.6167, loss_cns_3: 0.6553, loss_yns_3: 0.1516, loss_cls_4: 0.8955, loss_box_4: 1.6168, loss_cns_4: 0.6538, loss_yns_4: 0.1523, loss_cls_5: 0.9021, loss_box_5: 1.6148, loss_cns_5: 0.6565, loss_yns_5: 0.1531, loss_cls_dn_0: 0.2156, loss_box_dn_0: 0.7561, loss_cls_dn_1: 0.1321, loss_box_dn_1: 0.7241, loss_cls_dn_2: 0.1350, loss_box_dn_2: 0.7116, loss_cls_dn_3: 0.1359, loss_box_dn_3: 0.7040, loss_cls_dn_4: 0.1392, loss_box_dn_4: 0.7086, loss_cls_dn_5: 0.1431, loss_box_dn_5: 0.7096, loss_dense_depth: 0.7541, loss: 25.8959, grad_norm: 29.0064 -2025-11-12 14:36:12,952 - mmdet - INFO - Iter [208/17500] lr: 1.827e-04, eta: 10:27:02, time: 1.596, data_time: 0.090, memory: 49164, loss_cls_0: 0.7997, loss_box_0: 1.6973, loss_cns_0: 0.6188, loss_yns_0: 0.1536, loss_cls_1: 0.8767, loss_box_1: 1.6662, loss_cns_1: 0.6531, loss_yns_1: 0.1529, loss_cls_2: 0.8896, loss_box_2: 1.6232, loss_cns_2: 0.6534, loss_yns_2: 0.1550, loss_cls_3: 0.9012, loss_box_3: 1.6158, loss_cns_3: 0.6556, loss_yns_3: 0.1545, loss_cls_4: 0.9065, loss_box_4: 1.6022, loss_cns_4: 0.6567, loss_yns_4: 0.1570, loss_cls_5: 0.9122, loss_box_5: 1.6084, loss_cns_5: 0.6554, loss_yns_5: 0.1562, loss_cls_dn_0: 0.2116, loss_box_dn_0: 0.7605, loss_cls_dn_1: 0.1278, loss_box_dn_1: 0.7078, loss_cls_dn_2: 0.1348, loss_box_dn_2: 0.6921, loss_cls_dn_3: 0.1324, loss_box_dn_3: 0.6913, loss_cls_dn_4: 0.1343, loss_box_dn_4: 0.6977, loss_cls_dn_5: 0.1383, loss_box_dn_5: 0.7123, loss_dense_depth: 0.7830, loss: 25.8452, grad_norm: 30.6328 -2025-11-12 14:36:14,569 - mmdet - INFO - Iter [209/17500] lr: 1.831e-04, eta: 10:26:13, time: 1.612, data_time: 0.092, memory: 49164, loss_cls_0: 0.8152, loss_box_0: 1.7438, loss_cns_0: 0.6193, loss_yns_0: 0.1518, loss_cls_1: 0.9059, loss_box_1: 1.6662, loss_cns_1: 0.6522, loss_yns_1: 0.1476, loss_cls_2: 0.9113, loss_box_2: 1.6604, loss_cns_2: 0.6546, loss_yns_2: 0.1490, loss_cls_3: 0.9108, loss_box_3: 1.6639, loss_cns_3: 0.6547, loss_yns_3: 0.1503, loss_cls_4: 0.9150, loss_box_4: 1.6626, loss_cns_4: 0.6573, loss_yns_4: 0.1504, loss_cls_5: 0.9229, loss_box_5: 1.6815, loss_cns_5: 0.6569, loss_yns_5: 0.1501, loss_cls_dn_0: 0.2132, loss_box_dn_0: 0.7684, loss_cls_dn_1: 0.1283, loss_box_dn_1: 0.7283, loss_cls_dn_2: 0.1298, loss_box_dn_2: 0.7229, loss_cls_dn_3: 0.1305, loss_box_dn_3: 0.7402, loss_cls_dn_4: 0.1329, loss_box_dn_4: 0.7522, loss_cls_dn_5: 0.1381, loss_box_dn_5: 0.7806, loss_dense_depth: 0.8175, loss: 26.4364, grad_norm: 40.3057 -2025-11-12 14:36:16,177 - mmdet - INFO - Iter [210/17500] lr: 1.835e-04, eta: 10:25:25, time: 1.615, data_time: 0.090, memory: 49164, loss_cls_0: 0.7884, loss_box_0: 1.7194, loss_cns_0: 0.6268, loss_yns_0: 0.1534, loss_cls_1: 0.8670, loss_box_1: 1.6601, loss_cns_1: 0.6497, loss_yns_1: 0.1489, loss_cls_2: 0.8858, loss_box_2: 1.6457, loss_cns_2: 0.6561, loss_yns_2: 0.1507, loss_cls_3: 0.8866, loss_box_3: 1.6338, loss_cns_3: 0.6558, loss_yns_3: 0.1498, loss_cls_4: 0.8944, loss_box_4: 1.6404, loss_cns_4: 0.6557, loss_yns_4: 0.1509, loss_cls_5: 0.8910, loss_box_5: 1.6572, loss_cns_5: 0.6566, loss_yns_5: 0.1510, loss_cls_dn_0: 0.2014, loss_box_dn_0: 0.7556, loss_cls_dn_1: 0.1209, loss_box_dn_1: 0.7659, loss_cls_dn_2: 0.1233, loss_box_dn_2: 0.7712, loss_cls_dn_3: 0.1255, loss_box_dn_3: 0.7857, loss_cls_dn_4: 0.1279, loss_box_dn_4: 0.8048, loss_cls_dn_5: 0.1324, loss_box_dn_5: 0.8343, loss_dense_depth: 0.7652, loss: 26.2893, grad_norm: 39.7005 -2025-11-12 14:36:17,790 - mmdet - INFO - Iter [211/17500] lr: 1.839e-04, eta: 10:24:37, time: 1.612, data_time: 0.078, memory: 49164, loss_cls_0: 0.7926, loss_box_0: 1.7014, loss_cns_0: 0.6255, loss_yns_0: 0.1522, loss_cls_1: 0.8647, loss_box_1: 1.6573, loss_cns_1: 0.6531, loss_yns_1: 0.1496, loss_cls_2: 0.8893, loss_box_2: 1.6342, loss_cns_2: 0.6567, loss_yns_2: 0.1498, loss_cls_3: 0.8950, loss_box_3: 1.6116, loss_cns_3: 0.6560, loss_yns_3: 0.1502, loss_cls_4: 0.9036, loss_box_4: 1.6167, loss_cns_4: 0.6572, loss_yns_4: 0.1502, loss_cls_5: 0.8895, loss_box_5: 1.6323, loss_cns_5: 0.6548, loss_yns_5: 0.1507, loss_cls_dn_0: 0.1996, loss_box_dn_0: 0.7551, loss_cls_dn_1: 0.1140, loss_box_dn_1: 0.7556, loss_cls_dn_2: 0.1153, loss_box_dn_2: 0.7556, loss_cls_dn_3: 0.1187, loss_box_dn_3: 0.7580, loss_cls_dn_4: 0.1195, loss_box_dn_4: 0.7679, loss_cls_dn_5: 0.1244, loss_box_dn_5: 0.7877, loss_dense_depth: 0.7577, loss: 26.0233, grad_norm: 45.2372 -2025-11-12 14:36:19,379 - mmdet - INFO - Iter [212/17500] lr: 1.843e-04, eta: 10:23:48, time: 1.591, data_time: 0.083, memory: 49164, loss_cls_0: 0.7773, loss_box_0: 1.6903, loss_cns_0: 0.6272, loss_yns_0: 0.1466, loss_cls_1: 0.8650, loss_box_1: 1.6827, loss_cns_1: 0.6534, loss_yns_1: 0.1467, loss_cls_2: 0.8783, loss_box_2: 1.6605, loss_cns_2: 0.6551, loss_yns_2: 0.1459, loss_cls_3: 0.8860, loss_box_3: 1.6369, loss_cns_3: 0.6571, loss_yns_3: 0.1482, loss_cls_4: 0.8900, loss_box_4: 1.6324, loss_cns_4: 0.6584, loss_yns_4: 0.1479, loss_cls_5: 0.8894, loss_box_5: 1.6271, loss_cns_5: 0.6583, loss_yns_5: 0.1477, loss_cls_dn_0: 0.2029, loss_box_dn_0: 0.7577, loss_cls_dn_1: 0.1214, loss_box_dn_1: 0.7528, loss_cls_dn_2: 0.1196, loss_box_dn_2: 0.7448, loss_cls_dn_3: 0.1239, loss_box_dn_3: 0.7367, loss_cls_dn_4: 0.1240, loss_box_dn_4: 0.7403, loss_cls_dn_5: 0.1290, loss_box_dn_5: 0.7404, loss_dense_depth: 0.7677, loss: 25.9695, grad_norm: 38.9872 -2025-11-12 14:36:20,964 - mmdet - INFO - Iter [213/17500] lr: 1.847e-04, eta: 10:22:58, time: 1.581, data_time: 0.081, memory: 49164, loss_cls_0: 0.7755, loss_box_0: 1.6905, loss_cns_0: 0.6250, loss_yns_0: 0.1496, loss_cls_1: 0.8580, loss_box_1: 1.6295, loss_cns_1: 0.6509, loss_yns_1: 0.1487, loss_cls_2: 0.8760, loss_box_2: 1.5871, loss_cns_2: 0.6547, loss_yns_2: 0.1519, loss_cls_3: 0.8983, loss_box_3: 1.6007, loss_cns_3: 0.6558, loss_yns_3: 0.1516, loss_cls_4: 0.8902, loss_box_4: 1.5828, loss_cns_4: 0.6562, loss_yns_4: 0.1515, loss_cls_5: 0.8895, loss_box_5: 1.5794, loss_cns_5: 0.6565, loss_yns_5: 0.1518, loss_cls_dn_0: 0.2036, loss_box_dn_0: 0.7557, loss_cls_dn_1: 0.1210, loss_box_dn_1: 0.6985, loss_cls_dn_2: 0.1187, loss_box_dn_2: 0.6812, loss_cls_dn_3: 0.1210, loss_box_dn_3: 0.6838, loss_cls_dn_4: 0.1242, loss_box_dn_4: 0.6847, loss_cls_dn_5: 0.1287, loss_box_dn_5: 0.6816, loss_dense_depth: 0.8155, loss: 25.4798, grad_norm: 34.1263 -2025-11-12 14:36:22,544 - mmdet - INFO - Iter [214/17500] lr: 1.851e-04, eta: 10:22:09, time: 1.583, data_time: 0.079, memory: 49164, loss_cls_0: 0.7928, loss_box_0: 1.6685, loss_cns_0: 0.6146, loss_yns_0: 0.1470, loss_cls_1: 0.8613, loss_box_1: 1.6148, loss_cns_1: 0.6494, loss_yns_1: 0.1520, loss_cls_2: 0.8826, loss_box_2: 1.5646, loss_cns_2: 0.6543, loss_yns_2: 0.1531, loss_cls_3: 0.9021, loss_box_3: 1.5743, loss_cns_3: 0.6523, loss_yns_3: 0.1516, loss_cls_4: 0.9022, loss_box_4: 1.5714, loss_cns_4: 0.6552, loss_yns_4: 0.1513, loss_cls_5: 0.9066, loss_box_5: 1.5885, loss_cns_5: 0.6558, loss_yns_5: 0.1515, loss_cls_dn_0: 0.2039, loss_box_dn_0: 0.7634, loss_cls_dn_1: 0.1162, loss_box_dn_1: 0.6910, loss_cls_dn_2: 0.1197, loss_box_dn_2: 0.6733, loss_cls_dn_3: 0.1199, loss_box_dn_3: 0.6771, loss_cls_dn_4: 0.1291, loss_box_dn_4: 0.6800, loss_cls_dn_5: 0.1266, loss_box_dn_5: 0.6914, loss_dense_depth: 0.7510, loss: 25.3605, grad_norm: 39.5373 -2025-11-12 14:36:24,120 - mmdet - INFO - Iter [215/17500] lr: 1.855e-04, eta: 10:21:20, time: 1.575, data_time: 0.074, memory: 49164, loss_cls_0: 0.8178, loss_box_0: 1.6906, loss_cns_0: 0.6192, loss_yns_0: 0.1490, loss_cls_1: 0.8771, loss_box_1: 1.6443, loss_cns_1: 0.6487, loss_yns_1: 0.1503, loss_cls_2: 0.8996, loss_box_2: 1.6139, loss_cns_2: 0.6527, loss_yns_2: 0.1503, loss_cls_3: 0.9102, loss_box_3: 1.5985, loss_cns_3: 0.6541, loss_yns_3: 0.1511, loss_cls_4: 0.9058, loss_box_4: 1.5933, loss_cns_4: 0.6547, loss_yns_4: 0.1507, loss_cls_5: 0.9056, loss_box_5: 1.6011, loss_cns_5: 0.6537, loss_yns_5: 0.1505, loss_cls_dn_0: 0.2132, loss_box_dn_0: 0.7576, loss_cls_dn_1: 0.1216, loss_box_dn_1: 0.6986, loss_cls_dn_2: 0.1291, loss_box_dn_2: 0.6982, loss_cls_dn_3: 0.1269, loss_box_dn_3: 0.6926, loss_cls_dn_4: 0.1358, loss_box_dn_4: 0.7013, loss_cls_dn_5: 0.1313, loss_box_dn_5: 0.7144, loss_dense_depth: 0.8083, loss: 25.7715, grad_norm: 36.1466 -2025-11-12 14:36:25,680 - mmdet - INFO - Iter [216/17500] lr: 1.859e-04, eta: 10:20:31, time: 1.565, data_time: 0.078, memory: 49164, loss_cls_0: 0.8181, loss_box_0: 1.7210, loss_cns_0: 0.6266, loss_yns_0: 0.1530, loss_cls_1: 0.8800, loss_box_1: 1.6235, loss_cns_1: 0.6540, loss_yns_1: 0.1522, loss_cls_2: 0.9008, loss_box_2: 1.6075, loss_cns_2: 0.6561, loss_yns_2: 0.1504, loss_cls_3: 0.9038, loss_box_3: 1.5991, loss_cns_3: 0.6570, loss_yns_3: 0.1523, loss_cls_4: 0.9081, loss_box_4: 1.5973, loss_cns_4: 0.6557, loss_yns_4: 0.1513, loss_cls_5: 0.9231, loss_box_5: 1.5916, loss_cns_5: 0.6562, loss_yns_5: 0.1521, loss_cls_dn_0: 0.2134, loss_box_dn_0: 0.7660, loss_cls_dn_1: 0.1257, loss_box_dn_1: 0.7039, loss_cls_dn_2: 0.1290, loss_box_dn_2: 0.7080, loss_cls_dn_3: 0.1271, loss_box_dn_3: 0.7052, loss_cls_dn_4: 0.1292, loss_box_dn_4: 0.7135, loss_cls_dn_5: 0.1372, loss_box_dn_5: 0.7196, loss_dense_depth: 0.7699, loss: 25.8386, grad_norm: 40.1532 -2025-11-12 14:36:27,245 - mmdet - INFO - Iter [217/17500] lr: 1.863e-04, eta: 10:19:42, time: 1.566, data_time: 0.073, memory: 49164, loss_cls_0: 0.7827, loss_box_0: 1.6936, loss_cns_0: 0.6265, loss_yns_0: 0.1562, loss_cls_1: 0.8579, loss_box_1: 1.6333, loss_cns_1: 0.6495, loss_yns_1: 0.1554, loss_cls_2: 0.8726, loss_box_2: 1.6211, loss_cns_2: 0.6526, loss_yns_2: 0.1556, loss_cls_3: 0.8822, loss_box_3: 1.6016, loss_cns_3: 0.6544, loss_yns_3: 0.1550, loss_cls_4: 0.8849, loss_box_4: 1.6086, loss_cns_4: 0.6554, loss_yns_4: 0.1559, loss_cls_5: 0.8897, loss_box_5: 1.6077, loss_cns_5: 0.6557, loss_yns_5: 0.1580, loss_cls_dn_0: 0.2058, loss_box_dn_0: 0.7494, loss_cls_dn_1: 0.1224, loss_box_dn_1: 0.7054, loss_cls_dn_2: 0.1244, loss_box_dn_2: 0.7054, loss_cls_dn_3: 0.1271, loss_box_dn_3: 0.6968, loss_cls_dn_4: 0.1332, loss_box_dn_4: 0.7111, loss_cls_dn_5: 0.1451, loss_box_dn_5: 0.7150, loss_dense_depth: 0.7198, loss: 25.6273, grad_norm: 47.0578 -2025-11-12 14:36:28,818 - mmdet - INFO - Iter [218/17500] lr: 1.867e-04, eta: 10:18:53, time: 1.572, data_time: 0.074, memory: 49164, loss_cls_0: 0.7842, loss_box_0: 1.6860, loss_cns_0: 0.6290, loss_yns_0: 0.1528, loss_cls_1: 0.8603, loss_box_1: 1.6430, loss_cns_1: 0.6516, loss_yns_1: 0.1539, loss_cls_2: 0.8785, loss_box_2: 1.6242, loss_cns_2: 0.6534, loss_yns_2: 0.1539, loss_cls_3: 0.8974, loss_box_3: 1.5803, loss_cns_3: 0.6564, loss_yns_3: 0.1544, loss_cls_4: 0.8882, loss_box_4: 1.5776, loss_cns_4: 0.6572, loss_yns_4: 0.1563, loss_cls_5: 0.8853, loss_box_5: 1.5756, loss_cns_5: 0.6535, loss_yns_5: 0.1550, loss_cls_dn_0: 0.2074, loss_box_dn_0: 0.7486, loss_cls_dn_1: 0.1330, loss_box_dn_1: 0.6891, loss_cls_dn_2: 0.1299, loss_box_dn_2: 0.6858, loss_cls_dn_3: 0.1305, loss_box_dn_3: 0.6661, loss_cls_dn_4: 0.1361, loss_box_dn_4: 0.6724, loss_cls_dn_5: 0.1402, loss_box_dn_5: 0.6756, loss_dense_depth: 0.7311, loss: 25.4538, grad_norm: 32.9936 -2025-11-12 14:36:30,405 - mmdet - INFO - Iter [219/17500] lr: 1.871e-04, eta: 10:18:07, time: 1.588, data_time: 0.075, memory: 49164, loss_cls_0: 0.7971, loss_box_0: 1.7075, loss_cns_0: 0.6225, loss_yns_0: 0.1528, loss_cls_1: 0.8683, loss_box_1: 1.6793, loss_cns_1: 0.6515, loss_yns_1: 0.1517, loss_cls_2: 0.8896, loss_box_2: 1.6454, loss_cns_2: 0.6542, loss_yns_2: 0.1516, loss_cls_3: 0.9083, loss_box_3: 1.6460, loss_cns_3: 0.6578, loss_yns_3: 0.1528, loss_cls_4: 0.9085, loss_box_4: 1.6180, loss_cns_4: 0.6574, loss_yns_4: 0.1515, loss_cls_5: 0.9131, loss_box_5: 1.6221, loss_cns_5: 0.6580, loss_yns_5: 0.1508, loss_cls_dn_0: 0.2128, loss_box_dn_0: 0.7516, loss_cls_dn_1: 0.1294, loss_box_dn_1: 0.6852, loss_cls_dn_2: 0.1323, loss_box_dn_2: 0.6751, loss_cls_dn_3: 0.1380, loss_box_dn_3: 0.6706, loss_cls_dn_4: 0.1456, loss_box_dn_4: 0.6682, loss_cls_dn_5: 0.1532, loss_box_dn_5: 0.6749, loss_dense_depth: 0.7331, loss: 25.7858, grad_norm: 43.3602 -2025-11-12 14:36:31,990 - mmdet - INFO - Iter [220/17500] lr: 1.875e-04, eta: 10:17:21, time: 1.585, data_time: 0.075, memory: 49164, loss_cls_0: 0.7723, loss_box_0: 1.6962, loss_cns_0: 0.6246, loss_yns_0: 0.1518, loss_cls_1: 0.8561, loss_box_1: 1.6398, loss_cns_1: 0.6562, loss_yns_1: 0.1519, loss_cls_2: 0.8590, loss_box_2: 1.6197, loss_cns_2: 0.6553, loss_yns_2: 0.1527, loss_cls_3: 0.8924, loss_box_3: 1.6099, loss_cns_3: 0.6626, loss_yns_3: 0.1541, loss_cls_4: 0.8814, loss_box_4: 1.5919, loss_cns_4: 0.6600, loss_yns_4: 0.1532, loss_cls_5: 0.8923, loss_box_5: 1.5869, loss_cns_5: 0.6601, loss_yns_5: 0.1516, loss_cls_dn_0: 0.2045, loss_box_dn_0: 0.7534, loss_cls_dn_1: 0.1302, loss_box_dn_1: 0.7018, loss_cls_dn_2: 0.1315, loss_box_dn_2: 0.6874, loss_cls_dn_3: 0.1395, loss_box_dn_3: 0.6893, loss_cls_dn_4: 0.1424, loss_box_dn_4: 0.6835, loss_cls_dn_5: 0.1481, loss_box_dn_5: 0.6870, loss_dense_depth: 0.7367, loss: 25.5672, grad_norm: 46.9971 -2025-11-12 14:36:35,494 - mmdet - INFO - Iter [221/17500] lr: 1.879e-04, eta: 10:19:05, time: 3.505, data_time: 0.104, memory: 49164, loss_cls_0: 0.7864, loss_box_0: 1.6812, loss_cns_0: 0.6190, loss_yns_0: 0.1517, loss_cls_1: 0.8693, loss_box_1: 1.6568, loss_cns_1: 0.6550, loss_yns_1: 0.1504, loss_cls_2: 0.8686, loss_box_2: 1.6240, loss_cns_2: 0.6551, loss_yns_2: 0.1520, loss_cls_3: 0.8959, loss_box_3: 1.6260, loss_cns_3: 0.6586, loss_yns_3: 0.1519, loss_cls_4: 0.8861, loss_box_4: 1.5954, loss_cns_4: 0.6578, loss_yns_4: 0.1525, loss_cls_5: 0.8860, loss_box_5: 1.5991, loss_cns_5: 0.6580, loss_yns_5: 0.1509, loss_cls_dn_0: 0.2093, loss_box_dn_0: 0.7557, loss_cls_dn_1: 0.1355, loss_box_dn_1: 0.6999, loss_cls_dn_2: 0.1332, loss_box_dn_2: 0.6796, loss_cls_dn_3: 0.1363, loss_box_dn_3: 0.6854, loss_cls_dn_4: 0.1380, loss_box_dn_4: 0.6741, loss_cls_dn_5: 0.1406, loss_box_dn_5: 0.6800, loss_dense_depth: 0.7473, loss: 25.6027, grad_norm: 29.5728 -2025-11-12 14:36:38,847 - mmdet - INFO - Iter [222/17500] lr: 1.883e-04, eta: 10:20:37, time: 3.352, data_time: 0.101, memory: 49164, loss_cls_0: 0.7751, loss_box_0: 1.6896, loss_cns_0: 0.6198, loss_yns_0: 0.1493, loss_cls_1: 0.8567, loss_box_1: 1.6366, loss_cns_1: 0.6495, loss_yns_1: 0.1478, loss_cls_2: 0.8682, loss_box_2: 1.5972, loss_cns_2: 0.6530, loss_yns_2: 0.1493, loss_cls_3: 0.8850, loss_box_3: 1.6222, loss_cns_3: 0.6569, loss_yns_3: 0.1506, loss_cls_4: 0.8863, loss_box_4: 1.5932, loss_cns_4: 0.6549, loss_yns_4: 0.1492, loss_cls_5: 0.8870, loss_box_5: 1.5926, loss_cns_5: 0.6557, loss_yns_5: 0.1482, loss_cls_dn_0: 0.2077, loss_box_dn_0: 0.7548, loss_cls_dn_1: 0.1284, loss_box_dn_1: 0.7012, loss_cls_dn_2: 0.1341, loss_box_dn_2: 0.6809, loss_cls_dn_3: 0.1327, loss_box_dn_3: 0.6870, loss_cls_dn_4: 0.1398, loss_box_dn_4: 0.6804, loss_cls_dn_5: 0.1377, loss_box_dn_5: 0.6823, loss_dense_depth: 0.7645, loss: 25.5056, grad_norm: 43.0175 -2025-11-12 14:36:42,169 - mmdet - INFO - Iter [223/17500] lr: 1.887e-04, eta: 10:22:05, time: 3.322, data_time: 0.081, memory: 49164, loss_cls_0: 0.8097, loss_box_0: 1.7328, loss_cns_0: 0.6239, loss_yns_0: 0.1522, loss_cls_1: 0.8602, loss_box_1: 1.6686, loss_cns_1: 0.6488, loss_yns_1: 0.1494, loss_cls_2: 0.8836, loss_box_2: 1.6321, loss_cns_2: 0.6523, loss_yns_2: 0.1501, loss_cls_3: 0.8951, loss_box_3: 1.6250, loss_cns_3: 0.6568, loss_yns_3: 0.1514, loss_cls_4: 0.9178, loss_box_4: 1.6169, loss_cns_4: 0.6541, loss_yns_4: 0.1505, loss_cls_5: 0.9127, loss_box_5: 1.5942, loss_cns_5: 0.6553, loss_yns_5: 0.1496, loss_cls_dn_0: 0.2152, loss_box_dn_0: 0.7564, loss_cls_dn_1: 0.1343, loss_box_dn_1: 0.7087, loss_cls_dn_2: 0.1413, loss_box_dn_2: 0.6926, loss_cls_dn_3: 0.1382, loss_box_dn_3: 0.6893, loss_cls_dn_4: 0.1411, loss_box_dn_4: 0.6893, loss_cls_dn_5: 0.1426, loss_box_dn_5: 0.6903, loss_dense_depth: 0.7735, loss: 25.8561, grad_norm: 36.0774 -2025-11-12 14:36:43,734 - mmdet - INFO - Iter [224/17500] lr: 1.891e-04, eta: 10:21:17, time: 1.565, data_time: 0.071, memory: 49164, loss_cls_0: 0.7786, loss_box_0: 1.7013, loss_cns_0: 0.6305, loss_yns_0: 0.1533, loss_cls_1: 0.8664, loss_box_1: 1.6341, loss_cns_1: 0.6538, loss_yns_1: 0.1511, loss_cls_2: 0.8751, loss_box_2: 1.5961, loss_cns_2: 0.6562, loss_yns_2: 0.1511, loss_cls_3: 0.8805, loss_box_3: 1.5996, loss_cns_3: 0.6600, loss_yns_3: 0.1499, loss_cls_4: 0.8815, loss_box_4: 1.5780, loss_cns_4: 0.6556, loss_yns_4: 0.1517, loss_cls_5: 0.9094, loss_box_5: 1.5758, loss_cns_5: 0.6556, loss_yns_5: 0.1508, loss_cls_dn_0: 0.2038, loss_box_dn_0: 0.7546, loss_cls_dn_1: 0.1254, loss_box_dn_1: 0.7033, loss_cls_dn_2: 0.1256, loss_box_dn_2: 0.6871, loss_cls_dn_3: 0.1245, loss_box_dn_3: 0.6876, loss_cls_dn_4: 0.1270, loss_box_dn_4: 0.6872, loss_cls_dn_5: 0.1319, loss_box_dn_5: 0.6915, loss_dense_depth: 0.7457, loss: 25.4914, grad_norm: 33.8712 -2025-11-12 14:36:45,307 - mmdet - INFO - Iter [225/17500] lr: 1.895e-04, eta: 10:20:30, time: 1.572, data_time: 0.080, memory: 49164, loss_cls_0: 0.7790, loss_box_0: 1.7101, loss_cns_0: 0.6274, loss_yns_0: 0.1527, loss_cls_1: 0.8576, loss_box_1: 1.6660, loss_cns_1: 0.6518, loss_yns_1: 0.1514, loss_cls_2: 0.8899, loss_box_2: 1.6241, loss_cns_2: 0.6535, loss_yns_2: 0.1521, loss_cls_3: 0.8918, loss_box_3: 1.6345, loss_cns_3: 0.6598, loss_yns_3: 0.1518, loss_cls_4: 0.8925, loss_box_4: 1.6121, loss_cns_4: 0.6553, loss_yns_4: 0.1514, loss_cls_5: 0.8900, loss_box_5: 1.6063, loss_cns_5: 0.6587, loss_yns_5: 0.1513, loss_cls_dn_0: 0.2045, loss_box_dn_0: 0.7449, loss_cls_dn_1: 0.1219, loss_box_dn_1: 0.6896, loss_cls_dn_2: 0.1245, loss_box_dn_2: 0.6744, loss_cls_dn_3: 0.1267, loss_box_dn_3: 0.6794, loss_cls_dn_4: 0.1327, loss_box_dn_4: 0.6806, loss_cls_dn_5: 0.1316, loss_box_dn_5: 0.6785, loss_dense_depth: 0.7783, loss: 25.6388, grad_norm: 30.6188 -2025-11-12 14:36:48,763 - mmdet - INFO - Iter [226/17500] lr: 1.899e-04, eta: 10:22:07, time: 3.456, data_time: 0.098, memory: 49164, loss_cls_0: 0.8037, loss_box_0: 1.6916, loss_cns_0: 0.6255, loss_yns_0: 0.1491, loss_cls_1: 0.8731, loss_box_1: 1.6739, loss_cns_1: 0.6536, loss_yns_1: 0.1510, loss_cls_2: 0.8991, loss_box_2: 1.6206, loss_cns_2: 0.6530, loss_yns_2: 0.1480, loss_cls_3: 0.9123, loss_box_3: 1.6121, loss_cns_3: 0.6554, loss_yns_3: 0.1480, loss_cls_4: 0.9167, loss_box_4: 1.6119, loss_cns_4: 0.6549, loss_yns_4: 0.1494, loss_cls_5: 0.9076, loss_box_5: 1.6181, loss_cns_5: 0.6556, loss_yns_5: 0.1491, loss_cls_dn_0: 0.2054, loss_box_dn_0: 0.7480, loss_cls_dn_1: 0.1252, loss_box_dn_1: 0.6829, loss_cls_dn_2: 0.1257, loss_box_dn_2: 0.6678, loss_cls_dn_3: 0.1296, loss_box_dn_3: 0.6674, loss_cls_dn_4: 0.1317, loss_box_dn_4: 0.6767, loss_cls_dn_5: 0.1329, loss_box_dn_5: 0.6747, loss_dense_depth: 0.7670, loss: 25.6683, grad_norm: 32.5378 -2025-11-12 14:36:50,327 - mmdet - INFO - Iter [227/17500] lr: 1.903e-04, eta: 10:21:19, time: 1.565, data_time: 0.069, memory: 49164, loss_cls_0: 0.7923, loss_box_0: 1.6986, loss_cns_0: 0.6170, loss_yns_0: 0.1467, loss_cls_1: 0.8758, loss_box_1: 1.6313, loss_cns_1: 0.6541, loss_yns_1: 0.1463, loss_cls_2: 0.8878, loss_box_2: 1.6011, loss_cns_2: 0.6547, loss_yns_2: 0.1474, loss_cls_3: 0.8925, loss_box_3: 1.5914, loss_cns_3: 0.6613, loss_yns_3: 0.1468, loss_cls_4: 0.8938, loss_box_4: 1.5956, loss_cns_4: 0.6590, loss_yns_4: 0.1522, loss_cls_5: 0.9076, loss_box_5: 1.5983, loss_cns_5: 0.6573, loss_yns_5: 0.1478, loss_cls_dn_0: 0.2064, loss_box_dn_0: 0.7472, loss_cls_dn_1: 0.1216, loss_box_dn_1: 0.6838, loss_cls_dn_2: 0.1230, loss_box_dn_2: 0.6710, loss_cls_dn_3: 0.1276, loss_box_dn_3: 0.6731, loss_cls_dn_4: 0.1249, loss_box_dn_4: 0.6777, loss_cls_dn_5: 0.1295, loss_box_dn_5: 0.6814, loss_dense_depth: 0.7630, loss: 25.4869, grad_norm: 33.4499 -2025-11-12 14:36:51,898 - mmdet - INFO - Iter [228/17500] lr: 1.907e-04, eta: 10:20:33, time: 1.571, data_time: 0.076, memory: 49164, loss_cls_0: 0.7846, loss_box_0: 1.7078, loss_cns_0: 0.6195, loss_yns_0: 0.1475, loss_cls_1: 0.8790, loss_box_1: 1.6211, loss_cns_1: 0.6568, loss_yns_1: 0.1470, loss_cls_2: 0.8849, loss_box_2: 1.6222, loss_cns_2: 0.6570, loss_yns_2: 0.1469, loss_cls_3: 0.8933, loss_box_3: 1.6010, loss_cns_3: 0.6648, loss_yns_3: 0.1478, loss_cls_4: 0.8959, loss_box_4: 1.6094, loss_cns_4: 0.6603, loss_yns_4: 0.1489, loss_cls_5: 0.8998, loss_box_5: 1.6055, loss_cns_5: 0.6613, loss_yns_5: 0.1476, loss_cls_dn_0: 0.2108, loss_box_dn_0: 0.7516, loss_cls_dn_1: 0.1192, loss_box_dn_1: 0.6843, loss_cls_dn_2: 0.1202, loss_box_dn_2: 0.6828, loss_cls_dn_3: 0.1250, loss_box_dn_3: 0.6822, loss_cls_dn_4: 0.1236, loss_box_dn_4: 0.6859, loss_cls_dn_5: 0.1314, loss_box_dn_5: 0.6958, loss_dense_depth: 0.7459, loss: 25.5685, grad_norm: 43.5274 -2025-11-12 14:36:53,497 - mmdet - INFO - Iter [229/17500] lr: 1.911e-04, eta: 10:19:47, time: 1.577, data_time: 0.083, memory: 49164, loss_cls_0: 0.8105, loss_box_0: 1.6999, loss_cns_0: 0.6211, loss_yns_0: 0.1460, loss_cls_1: 0.8701, loss_box_1: 1.6342, loss_cns_1: 0.6542, loss_yns_1: 0.1477, loss_cls_2: 0.8790, loss_box_2: 1.6331, loss_cns_2: 0.6579, loss_yns_2: 0.1435, loss_cls_3: 0.9219, loss_box_3: 1.6290, loss_cns_3: 0.6581, loss_yns_3: 0.1436, loss_cls_4: 0.8995, loss_box_4: 1.6087, loss_cns_4: 0.6582, loss_yns_4: 0.1445, loss_cls_5: 0.8948, loss_box_5: 1.6404, loss_cns_5: 0.6556, loss_yns_5: 0.1432, loss_cls_dn_0: 0.2036, loss_box_dn_0: 0.7404, loss_cls_dn_1: 0.1198, loss_box_dn_1: 0.6931, loss_cls_dn_2: 0.1203, loss_box_dn_2: 0.6920, loss_cls_dn_3: 0.1228, loss_box_dn_3: 0.6976, loss_cls_dn_4: 0.1224, loss_box_dn_4: 0.6978, loss_cls_dn_5: 0.1300, loss_box_dn_5: 0.7221, loss_dense_depth: 0.7787, loss: 25.7353, grad_norm: 42.5718 -2025-11-12 14:36:55,082 - mmdet - INFO - Iter [230/17500] lr: 1.915e-04, eta: 10:19:03, time: 1.599, data_time: 0.095, memory: 49164, loss_cls_0: 0.7987, loss_box_0: 1.7137, loss_cns_0: 0.6214, loss_yns_0: 0.1476, loss_cls_1: 0.8671, loss_box_1: 1.6609, loss_cns_1: 0.6418, loss_yns_1: 0.1463, loss_cls_2: 0.8808, loss_box_2: 1.6470, loss_cns_2: 0.6508, loss_yns_2: 0.1433, loss_cls_3: 0.8862, loss_box_3: 1.6432, loss_cns_3: 0.6531, loss_yns_3: 0.1447, loss_cls_4: 0.9023, loss_box_4: 1.6271, loss_cns_4: 0.6613, loss_yns_4: 0.1492, loss_cls_5: 0.9186, loss_box_5: 1.6225, loss_cns_5: 0.6521, loss_yns_5: 0.1463, loss_cls_dn_0: 0.2056, loss_box_dn_0: 0.7497, loss_cls_dn_1: 0.1204, loss_box_dn_1: 0.7069, loss_cls_dn_2: 0.1183, loss_box_dn_2: 0.6993, loss_cls_dn_3: 0.1190, loss_box_dn_3: 0.7042, loss_cls_dn_4: 0.1176, loss_box_dn_4: 0.7060, loss_cls_dn_5: 0.1229, loss_box_dn_5: 0.7211, loss_dense_depth: 0.7384, loss: 25.7554, grad_norm: 42.1623 -2025-11-12 14:36:58,419 - mmdet - INFO - Iter [231/17500] lr: 1.919e-04, eta: 10:20:30, time: 3.342, data_time: 0.079, memory: 49164, loss_cls_0: 0.7689, loss_box_0: 1.6959, loss_cns_0: 0.6271, loss_yns_0: 0.1470, loss_cls_1: 0.8432, loss_box_1: 1.5863, loss_cns_1: 0.6483, loss_yns_1: 0.1439, loss_cls_2: 0.8530, loss_box_2: 1.5812, loss_cns_2: 0.6550, loss_yns_2: 0.1444, loss_cls_3: 0.8634, loss_box_3: 1.5784, loss_cns_3: 0.6580, loss_yns_3: 0.1446, loss_cls_4: 0.8796, loss_box_4: 1.5871, loss_cns_4: 0.6662, loss_yns_4: 0.1480, loss_cls_5: 0.8791, loss_box_5: 1.5764, loss_cns_5: 0.6578, loss_yns_5: 0.1457, loss_cls_dn_0: 0.1971, loss_box_dn_0: 0.7457, loss_cls_dn_1: 0.1137, loss_box_dn_1: 0.6929, loss_cls_dn_2: 0.1145, loss_box_dn_2: 0.6941, loss_cls_dn_3: 0.1159, loss_box_dn_3: 0.7002, loss_cls_dn_4: 0.1197, loss_box_dn_4: 0.7099, loss_cls_dn_5: 0.1259, loss_box_dn_5: 0.7170, loss_dense_depth: 0.7294, loss: 25.2547, grad_norm: 51.7810 -2025-11-12 14:36:59,984 - mmdet - INFO - Iter [232/17500] lr: 1.923e-04, eta: 10:19:44, time: 1.562, data_time: 0.076, memory: 49164, loss_cls_0: 0.7773, loss_box_0: 1.6934, loss_cns_0: 0.6286, loss_yns_0: 0.1475, loss_cls_1: 0.8566, loss_box_1: 1.6230, loss_cns_1: 0.6569, loss_yns_1: 0.1445, loss_cls_2: 0.8817, loss_box_2: 1.6061, loss_cns_2: 0.6605, loss_yns_2: 0.1444, loss_cls_3: 0.8826, loss_box_3: 1.5989, loss_cns_3: 0.6616, loss_yns_3: 0.1457, loss_cls_4: 0.8891, loss_box_4: 1.5800, loss_cns_4: 0.6626, loss_yns_4: 0.1461, loss_cls_5: 0.8920, loss_box_5: 1.5844, loss_cns_5: 0.6625, loss_yns_5: 0.1447, loss_cls_dn_0: 0.1985, loss_box_dn_0: 0.7595, loss_cls_dn_1: 0.1159, loss_box_dn_1: 0.6936, loss_cls_dn_2: 0.1198, loss_box_dn_2: 0.6842, loss_cls_dn_3: 0.1203, loss_box_dn_3: 0.6873, loss_cls_dn_4: 0.1235, loss_box_dn_4: 0.6858, loss_cls_dn_5: 0.1292, loss_box_dn_5: 0.6916, loss_dense_depth: 0.7476, loss: 25.4275, grad_norm: 39.6415 -2025-11-12 14:37:01,550 - mmdet - INFO - Iter [233/17500] lr: 1.927e-04, eta: 10:18:58, time: 1.563, data_time: 0.077, memory: 49164, loss_cls_0: 0.7885, loss_box_0: 1.6776, loss_cns_0: 0.6238, loss_yns_0: 0.1472, loss_cls_1: 0.8919, loss_box_1: 1.5779, loss_cns_1: 0.6619, loss_yns_1: 0.1473, loss_cls_2: 0.9025, loss_box_2: 1.5473, loss_cns_2: 0.6624, loss_yns_2: 0.1465, loss_cls_3: 0.9110, loss_box_3: 1.5447, loss_cns_3: 0.6618, loss_yns_3: 0.1464, loss_cls_4: 0.9019, loss_box_4: 1.5402, loss_cns_4: 0.6643, loss_yns_4: 0.1465, loss_cls_5: 0.9089, loss_box_5: 1.5318, loss_cns_5: 0.6631, loss_yns_5: 0.1466, loss_cls_dn_0: 0.2010, loss_box_dn_0: 0.7623, loss_cls_dn_1: 0.1204, loss_box_dn_1: 0.6869, loss_cls_dn_2: 0.1215, loss_box_dn_2: 0.6717, loss_cls_dn_3: 0.1213, loss_box_dn_3: 0.6742, loss_cls_dn_4: 0.1227, loss_box_dn_4: 0.6731, loss_cls_dn_5: 0.1250, loss_box_dn_5: 0.6736, loss_dense_depth: 0.7440, loss: 25.2395, grad_norm: 41.5266 -2025-11-12 14:37:03,114 - mmdet - INFO - Iter [234/17500] lr: 1.931e-04, eta: 10:18:13, time: 1.568, data_time: 0.078, memory: 49164, loss_cls_0: 0.7780, loss_box_0: 1.6498, loss_cns_0: 0.6242, loss_yns_0: 0.1477, loss_cls_1: 0.8761, loss_box_1: 1.5634, loss_cns_1: 0.6634, loss_yns_1: 0.1493, loss_cls_2: 0.8880, loss_box_2: 1.5241, loss_cns_2: 0.6629, loss_yns_2: 0.1473, loss_cls_3: 0.9072, loss_box_3: 1.5224, loss_cns_3: 0.6616, loss_yns_3: 0.1472, loss_cls_4: 0.9041, loss_box_4: 1.5255, loss_cns_4: 0.6632, loss_yns_4: 0.1480, loss_cls_5: 0.8985, loss_box_5: 1.5171, loss_cns_5: 0.6615, loss_yns_5: 0.1485, loss_cls_dn_0: 0.1945, loss_box_dn_0: 0.7441, loss_cls_dn_1: 0.1123, loss_box_dn_1: 0.6878, loss_cls_dn_2: 0.1126, loss_box_dn_2: 0.6743, loss_cls_dn_3: 0.1143, loss_box_dn_3: 0.6811, loss_cls_dn_4: 0.1174, loss_box_dn_4: 0.6927, loss_cls_dn_5: 0.1197, loss_box_dn_5: 0.6919, loss_dense_depth: 0.7381, loss: 25.0598, grad_norm: 43.4040 -2025-11-12 14:37:04,693 - mmdet - INFO - Iter [235/17500] lr: 1.935e-04, eta: 10:17:29, time: 1.581, data_time: 0.078, memory: 49164, loss_cls_0: 0.7943, loss_box_0: 1.6697, loss_cns_0: 0.6216, loss_yns_0: 0.1480, loss_cls_1: 0.8753, loss_box_1: 1.6060, loss_cns_1: 0.6592, loss_yns_1: 0.1496, loss_cls_2: 0.8904, loss_box_2: 1.5644, loss_cns_2: 0.6582, loss_yns_2: 0.1485, loss_cls_3: 0.9033, loss_box_3: 1.5501, loss_cns_3: 0.6585, loss_yns_3: 0.1489, loss_cls_4: 0.9200, loss_box_4: 1.5315, loss_cns_4: 0.6556, loss_yns_4: 0.1477, loss_cls_5: 0.9178, loss_box_5: 1.5410, loss_cns_5: 0.6597, loss_yns_5: 0.1473, loss_cls_dn_0: 0.1983, loss_box_dn_0: 0.7417, loss_cls_dn_1: 0.1153, loss_box_dn_1: 0.6932, loss_cls_dn_2: 0.1148, loss_box_dn_2: 0.6814, loss_cls_dn_3: 0.1175, loss_box_dn_3: 0.6804, loss_cls_dn_4: 0.1218, loss_box_dn_4: 0.6856, loss_cls_dn_5: 0.1252, loss_box_dn_5: 0.6942, loss_dense_depth: 0.7392, loss: 25.2751, grad_norm: 41.3303 -2025-11-12 14:37:06,268 - mmdet - INFO - Iter [236/17500] lr: 1.939e-04, eta: 10:16:45, time: 1.571, data_time: 0.074, memory: 49164, loss_cls_0: 0.7624, loss_box_0: 1.6737, loss_cns_0: 0.6254, loss_yns_0: 0.1502, loss_cls_1: 0.8806, loss_box_1: 1.5907, loss_cns_1: 0.6579, loss_yns_1: 0.1508, loss_cls_2: 0.8845, loss_box_2: 1.5739, loss_cns_2: 0.6511, loss_yns_2: 0.1479, loss_cls_3: 0.8977, loss_box_3: 1.5762, loss_cns_3: 0.6533, loss_yns_3: 0.1498, loss_cls_4: 0.9188, loss_box_4: 1.5820, loss_cns_4: 0.6501, loss_yns_4: 0.1481, loss_cls_5: 0.9199, loss_box_5: 1.5910, loss_cns_5: 0.6576, loss_yns_5: 0.1509, loss_cls_dn_0: 0.1914, loss_box_dn_0: 0.7365, loss_cls_dn_1: 0.1171, loss_box_dn_1: 0.6804, loss_cls_dn_2: 0.1166, loss_box_dn_2: 0.6832, loss_cls_dn_3: 0.1214, loss_box_dn_3: 0.6847, loss_cls_dn_4: 0.1263, loss_box_dn_4: 0.7018, loss_cls_dn_5: 0.1345, loss_box_dn_5: 0.7182, loss_dense_depth: 0.7401, loss: 25.3969, grad_norm: 66.0761 -2025-11-12 14:37:07,850 - mmdet - INFO - Iter [237/17500] lr: 1.943e-04, eta: 10:16:01, time: 1.577, data_time: 0.077, memory: 49164, loss_cls_0: 0.7662, loss_box_0: 1.6509, loss_cns_0: 0.6274, loss_yns_0: 0.1493, loss_cls_1: 0.8619, loss_box_1: 1.6014, loss_cns_1: 0.6596, loss_yns_1: 0.1520, loss_cls_2: 0.8855, loss_box_2: 1.5747, loss_cns_2: 0.6530, loss_yns_2: 0.1489, loss_cls_3: 0.8993, loss_box_3: 1.5659, loss_cns_3: 0.6564, loss_yns_3: 0.1509, loss_cls_4: 0.9075, loss_box_4: 1.5933, loss_cns_4: 0.6581, loss_yns_4: 0.1506, loss_cls_5: 0.9164, loss_box_5: 1.5916, loss_cns_5: 0.6585, loss_yns_5: 0.1530, loss_cls_dn_0: 0.1949, loss_box_dn_0: 0.7447, loss_cls_dn_1: 0.1194, loss_box_dn_1: 0.7050, loss_cls_dn_2: 0.1246, loss_box_dn_2: 0.7158, loss_cls_dn_3: 0.1337, loss_box_dn_3: 0.7113, loss_cls_dn_4: 0.1346, loss_box_dn_4: 0.7336, loss_cls_dn_5: 0.1497, loss_box_dn_5: 0.7443, loss_dense_depth: 0.7218, loss: 25.5659, grad_norm: 77.8502 -2025-11-12 14:37:09,429 - mmdet - INFO - Iter [238/17500] lr: 1.947e-04, eta: 10:15:19, time: 1.589, data_time: 0.083, memory: 49164, loss_cls_0: 0.7712, loss_box_0: 1.6410, loss_cns_0: 0.6281, loss_yns_0: 0.1515, loss_cls_1: 0.8621, loss_box_1: 1.6034, loss_cns_1: 0.6583, loss_yns_1: 0.1534, loss_cls_2: 0.8868, loss_box_2: 1.5486, loss_cns_2: 0.6584, loss_yns_2: 0.1509, loss_cls_3: 0.8808, loss_box_3: 1.5558, loss_cns_3: 0.6599, loss_yns_3: 0.1532, loss_cls_4: 0.8791, loss_box_4: 1.5459, loss_cns_4: 0.6619, loss_yns_4: 0.1523, loss_cls_5: 0.8942, loss_box_5: 1.5498, loss_cns_5: 0.6616, loss_yns_5: 0.1547, loss_cls_dn_0: 0.1917, loss_box_dn_0: 0.7451, loss_cls_dn_1: 0.1213, loss_box_dn_1: 0.7220, loss_cls_dn_2: 0.1243, loss_box_dn_2: 0.7127, loss_cls_dn_3: 0.1269, loss_box_dn_3: 0.7090, loss_cls_dn_4: 0.1247, loss_box_dn_4: 0.7068, loss_cls_dn_5: 0.1330, loss_box_dn_5: 0.7102, loss_dense_depth: 0.7024, loss: 25.2928, grad_norm: 42.5807 -2025-11-12 14:37:11,010 - mmdet - INFO - Iter [239/17500] lr: 1.951e-04, eta: 10:14:37, time: 1.581, data_time: 0.074, memory: 49164, loss_cls_0: 0.7814, loss_box_0: 1.6146, loss_cns_0: 0.6302, loss_yns_0: 0.1506, loss_cls_1: 0.8638, loss_box_1: 1.5321, loss_cns_1: 0.6598, loss_yns_1: 0.1522, loss_cls_2: 0.8861, loss_box_2: 1.4770, loss_cns_2: 0.6637, loss_yns_2: 0.1511, loss_cls_3: 0.8957, loss_box_3: 1.5417, loss_cns_3: 0.6638, loss_yns_3: 0.1513, loss_cls_4: 0.9085, loss_box_4: 1.4828, loss_cns_4: 0.6645, loss_yns_4: 0.1523, loss_cls_5: 0.8953, loss_box_5: 1.4770, loss_cns_5: 0.6679, loss_yns_5: 0.1521, loss_cls_dn_0: 0.1943, loss_box_dn_0: 0.7355, loss_cls_dn_1: 0.1199, loss_box_dn_1: 0.7007, loss_cls_dn_2: 0.1250, loss_box_dn_2: 0.6817, loss_cls_dn_3: 0.1222, loss_box_dn_3: 0.7078, loss_cls_dn_4: 0.1257, loss_box_dn_4: 0.6891, loss_cls_dn_5: 0.1295, loss_box_dn_5: 0.6857, loss_dense_depth: 0.6816, loss: 24.9143, grad_norm: 49.0151 -2025-11-12 14:37:12,600 - mmdet - INFO - Iter [240/17500] lr: 1.955e-04, eta: 10:13:55, time: 1.591, data_time: 0.071, memory: 49164, loss_cls_0: 0.7679, loss_box_0: 1.6641, loss_cns_0: 0.6245, loss_yns_0: 0.1517, loss_cls_1: 0.8858, loss_box_1: 1.5159, loss_cns_1: 0.6644, loss_yns_1: 0.1509, loss_cls_2: 0.9018, loss_box_2: 1.4952, loss_cns_2: 0.6651, loss_yns_2: 0.1506, loss_cls_3: 0.9119, loss_box_3: 1.5237, loss_cns_3: 0.6693, loss_yns_3: 0.1516, loss_cls_4: 0.9168, loss_box_4: 1.4758, loss_cns_4: 0.6661, loss_yns_4: 0.1519, loss_cls_5: 0.9013, loss_box_5: 1.4808, loss_cns_5: 0.6712, loss_yns_5: 0.1525, loss_cls_dn_0: 0.1964, loss_box_dn_0: 0.7519, loss_cls_dn_1: 0.1201, loss_box_dn_1: 0.6817, loss_cls_dn_2: 0.1267, loss_box_dn_2: 0.6773, loss_cls_dn_3: 0.1250, loss_box_dn_3: 0.6966, loss_cls_dn_4: 0.1311, loss_box_dn_4: 0.6799, loss_cls_dn_5: 0.1331, loss_box_dn_5: 0.6818, loss_dense_depth: 0.6936, loss: 25.0058, grad_norm: 55.8324 -2025-11-12 14:37:14,251 - mmdet - INFO - Iter [241/17500] lr: 1.959e-04, eta: 10:13:19, time: 1.651, data_time: 0.099, memory: 49164, loss_cls_0: 0.7846, loss_box_0: 1.6615, loss_cns_0: 0.6239, loss_yns_0: 0.1549, loss_cls_1: 0.8920, loss_box_1: 1.5581, loss_cns_1: 0.6616, loss_yns_1: 0.1558, loss_cls_2: 0.9111, loss_box_2: 1.5375, loss_cns_2: 0.6637, loss_yns_2: 0.1557, loss_cls_3: 0.9057, loss_box_3: 1.5362, loss_cns_3: 0.6673, loss_yns_3: 0.1561, loss_cls_4: 0.9112, loss_box_4: 1.5209, loss_cns_4: 0.6641, loss_yns_4: 0.1560, loss_cls_5: 0.9081, loss_box_5: 1.5284, loss_cns_5: 0.6675, loss_yns_5: 0.1578, loss_cls_dn_0: 0.1945, loss_box_dn_0: 0.7461, loss_cls_dn_1: 0.1211, loss_box_dn_1: 0.6890, loss_cls_dn_2: 0.1269, loss_box_dn_2: 0.6811, loss_cls_dn_3: 0.1271, loss_box_dn_3: 0.6861, loss_cls_dn_4: 0.1284, loss_box_dn_4: 0.6775, loss_cls_dn_5: 0.1300, loss_box_dn_5: 0.6854, loss_dense_depth: 0.7227, loss: 25.2557, grad_norm: 44.8458 -2025-11-12 14:37:15,871 - mmdet - INFO - Iter [242/17500] lr: 1.963e-04, eta: 10:12:40, time: 1.620, data_time: 0.101, memory: 49164, loss_cls_0: 0.7925, loss_box_0: 1.6133, loss_cns_0: 0.6262, loss_yns_0: 0.1515, loss_cls_1: 0.8916, loss_box_1: 1.5020, loss_cns_1: 0.6705, loss_yns_1: 0.1546, loss_cls_2: 0.9037, loss_box_2: 1.4913, loss_cns_2: 0.6665, loss_yns_2: 0.1553, loss_cls_3: 0.8981, loss_box_3: 1.5148, loss_cns_3: 0.6653, loss_yns_3: 0.1547, loss_cls_4: 0.9344, loss_box_4: 1.5103, loss_cns_4: 0.6645, loss_yns_4: 0.1564, loss_cls_5: 0.9079, loss_box_5: 1.5048, loss_cns_5: 0.6619, loss_yns_5: 0.1572, loss_cls_dn_0: 0.1921, loss_box_dn_0: 0.7423, loss_cls_dn_1: 0.1203, loss_box_dn_1: 0.6790, loss_cls_dn_2: 0.1224, loss_box_dn_2: 0.6675, loss_cls_dn_3: 0.1247, loss_box_dn_3: 0.6817, loss_cls_dn_4: 0.1284, loss_box_dn_4: 0.6845, loss_cls_dn_5: 0.1310, loss_box_dn_5: 0.6883, loss_dense_depth: 0.6847, loss: 24.9962, grad_norm: 42.0736 -2025-11-12 14:37:17,479 - mmdet - INFO - Iter [243/17500] lr: 1.967e-04, eta: 10:12:00, time: 1.599, data_time: 0.079, memory: 49164, loss_cls_0: 0.8065, loss_box_0: 1.6235, loss_cns_0: 0.6224, loss_yns_0: 0.1519, loss_cls_1: 0.8921, loss_box_1: 1.5753, loss_cns_1: 0.6563, loss_yns_1: 0.1518, loss_cls_2: 0.9002, loss_box_2: 1.5615, loss_cns_2: 0.6597, loss_yns_2: 0.1515, loss_cls_3: 0.9009, loss_box_3: 1.5383, loss_cns_3: 0.6621, loss_yns_3: 0.1504, loss_cls_4: 0.9345, loss_box_4: 1.5318, loss_cns_4: 0.6600, loss_yns_4: 0.1519, loss_cls_5: 0.9023, loss_box_5: 1.5265, loss_cns_5: 0.6632, loss_yns_5: 0.1517, loss_cls_dn_0: 0.1985, loss_box_dn_0: 0.7444, loss_cls_dn_1: 0.1225, loss_box_dn_1: 0.6958, loss_cls_dn_2: 0.1243, loss_box_dn_2: 0.6833, loss_cls_dn_3: 0.1272, loss_box_dn_3: 0.6858, loss_cls_dn_4: 0.1363, loss_box_dn_4: 0.6917, loss_cls_dn_5: 0.1366, loss_box_dn_5: 0.6971, loss_dense_depth: 0.7024, loss: 25.2722, grad_norm: 43.2645 -2025-11-12 14:37:19,052 - mmdet - INFO - Iter [244/17500] lr: 1.971e-04, eta: 10:11:19, time: 1.576, data_time: 0.079, memory: 49164, loss_cls_0: 0.7896, loss_box_0: 1.6485, loss_cns_0: 0.6242, loss_yns_0: 0.1536, loss_cls_1: 0.8661, loss_box_1: 1.5998, loss_cns_1: 0.6549, loss_yns_1: 0.1527, loss_cls_2: 0.8877, loss_box_2: 1.5574, loss_cns_2: 0.6551, loss_yns_2: 0.1521, loss_cls_3: 0.8988, loss_box_3: 1.5649, loss_cns_3: 0.6565, loss_yns_3: 0.1526, loss_cls_4: 0.9207, loss_box_4: 1.5688, loss_cns_4: 0.6555, loss_yns_4: 0.1515, loss_cls_5: 0.9103, loss_box_5: 1.5754, loss_cns_5: 0.6614, loss_yns_5: 0.1517, loss_cls_dn_0: 0.1958, loss_box_dn_0: 0.7387, loss_cls_dn_1: 0.1243, loss_box_dn_1: 0.6994, loss_cls_dn_2: 0.1245, loss_box_dn_2: 0.6818, loss_cls_dn_3: 0.1243, loss_box_dn_3: 0.6882, loss_cls_dn_4: 0.1318, loss_box_dn_4: 0.6902, loss_cls_dn_5: 0.1301, loss_box_dn_5: 0.6991, loss_dense_depth: 0.7240, loss: 25.3622, grad_norm: 43.5876 -2025-11-12 14:37:20,646 - mmdet - INFO - Iter [245/17500] lr: 1.975e-04, eta: 10:10:40, time: 1.599, data_time: 0.085, memory: 49164, loss_cls_0: 0.7775, loss_box_0: 1.6686, loss_cns_0: 0.6240, loss_yns_0: 0.1516, loss_cls_1: 0.8362, loss_box_1: 1.5512, loss_cns_1: 0.6561, loss_yns_1: 0.1521, loss_cls_2: 0.8661, loss_box_2: 1.5230, loss_cns_2: 0.6563, loss_yns_2: 0.1533, loss_cls_3: 0.8699, loss_box_3: 1.5293, loss_cns_3: 0.6551, loss_yns_3: 0.1517, loss_cls_4: 0.8837, loss_box_4: 1.5092, loss_cns_4: 0.6602, loss_yns_4: 0.1516, loss_cls_5: 0.8806, loss_box_5: 1.5188, loss_cns_5: 0.6612, loss_yns_5: 0.1543, loss_cls_dn_0: 0.1989, loss_box_dn_0: 0.7330, loss_cls_dn_1: 0.1293, loss_box_dn_1: 0.6878, loss_cls_dn_2: 0.1288, loss_box_dn_2: 0.6746, loss_cls_dn_3: 0.1269, loss_box_dn_3: 0.6778, loss_cls_dn_4: 0.1290, loss_box_dn_4: 0.6675, loss_cls_dn_5: 0.1292, loss_box_dn_5: 0.6724, loss_dense_depth: 0.6959, loss: 24.8924, grad_norm: 36.5968 -2025-11-12 14:37:22,233 - mmdet - INFO - Iter [246/17500] lr: 1.979e-04, eta: 10:10:00, time: 1.588, data_time: 0.095, memory: 49164, loss_cls_0: 0.7703, loss_box_0: 1.6830, loss_cns_0: 0.6225, loss_yns_0: 0.1528, loss_cls_1: 0.8473, loss_box_1: 1.5766, loss_cns_1: 0.6570, loss_yns_1: 0.1520, loss_cls_2: 0.8673, loss_box_2: 1.5563, loss_cns_2: 0.6565, loss_yns_2: 0.1527, loss_cls_3: 0.8719, loss_box_3: 1.5642, loss_cns_3: 0.6577, loss_yns_3: 0.1523, loss_cls_4: 0.8856, loss_box_4: 1.5487, loss_cns_4: 0.6588, loss_yns_4: 0.1530, loss_cls_5: 0.8769, loss_box_5: 1.5468, loss_cns_5: 0.6572, loss_yns_5: 0.1547, loss_cls_dn_0: 0.1951, loss_box_dn_0: 0.7411, loss_cls_dn_1: 0.1220, loss_box_dn_1: 0.6833, loss_cls_dn_2: 0.1224, loss_box_dn_2: 0.6716, loss_cls_dn_3: 0.1242, loss_box_dn_3: 0.6722, loss_cls_dn_4: 0.1285, loss_box_dn_4: 0.6664, loss_cls_dn_5: 0.1294, loss_box_dn_5: 0.6666, loss_dense_depth: 0.7436, loss: 25.0887, grad_norm: 48.2492 -2025-11-12 14:37:23,822 - mmdet - INFO - Iter [247/17500] lr: 1.983e-04, eta: 10:09:21, time: 1.589, data_time: 0.073, memory: 49164, loss_cls_0: 0.8050, loss_box_0: 1.6791, loss_cns_0: 0.6273, loss_yns_0: 0.1570, loss_cls_1: 0.8760, loss_box_1: 1.5607, loss_cns_1: 0.6598, loss_yns_1: 0.1551, loss_cls_2: 0.8843, loss_box_2: 1.5556, loss_cns_2: 0.6608, loss_yns_2: 0.1545, loss_cls_3: 0.8862, loss_box_3: 1.5631, loss_cns_3: 0.6677, loss_yns_3: 0.1546, loss_cls_4: 0.9169, loss_box_4: 1.5694, loss_cns_4: 0.6635, loss_yns_4: 0.1551, loss_cls_5: 0.8990, loss_box_5: 1.5673, loss_cns_5: 0.6619, loss_yns_5: 0.1564, loss_cls_dn_0: 0.1940, loss_box_dn_0: 0.7405, loss_cls_dn_1: 0.1178, loss_box_dn_1: 0.6660, loss_cls_dn_2: 0.1185, loss_box_dn_2: 0.6608, loss_cls_dn_3: 0.1222, loss_box_dn_3: 0.6670, loss_cls_dn_4: 0.1261, loss_box_dn_4: 0.6750, loss_cls_dn_5: 0.1295, loss_box_dn_5: 0.6822, loss_dense_depth: 0.7111, loss: 25.2470, grad_norm: 37.3503 -2025-11-12 14:37:25,407 - mmdet - INFO - Iter [248/17500] lr: 1.987e-04, eta: 10:08:41, time: 1.584, data_time: 0.081, memory: 49164, loss_cls_0: 0.7888, loss_box_0: 1.7107, loss_cns_0: 0.6223, loss_yns_0: 0.1550, loss_cls_1: 0.8599, loss_box_1: 1.5881, loss_cns_1: 0.6559, loss_yns_1: 0.1540, loss_cls_2: 0.8795, loss_box_2: 1.5758, loss_cns_2: 0.6586, loss_yns_2: 0.1542, loss_cls_3: 0.8938, loss_box_3: 1.5697, loss_cns_3: 0.6689, loss_yns_3: 0.1552, loss_cls_4: 0.9269, loss_box_4: 1.5691, loss_cns_4: 0.6662, loss_yns_4: 0.1561, loss_cls_5: 0.8987, loss_box_5: 1.5604, loss_cns_5: 0.6710, loss_yns_5: 0.1562, loss_cls_dn_0: 0.1951, loss_box_dn_0: 0.7439, loss_cls_dn_1: 0.1177, loss_box_dn_1: 0.6766, loss_cls_dn_2: 0.1203, loss_box_dn_2: 0.6774, loss_cls_dn_3: 0.1222, loss_box_dn_3: 0.6869, loss_cls_dn_4: 0.1251, loss_box_dn_4: 0.7053, loss_cls_dn_5: 0.1261, loss_box_dn_5: 0.7170, loss_dense_depth: 0.7545, loss: 25.4631, grad_norm: 52.2863 -2025-11-12 14:37:27,002 - mmdet - INFO - Iter [249/17500] lr: 1.991e-04, eta: 10:08:03, time: 1.595, data_time: 0.083, memory: 49164, loss_cls_0: 0.8109, loss_box_0: 1.6930, loss_cns_0: 0.6281, loss_yns_0: 0.1552, loss_cls_1: 0.8774, loss_box_1: 1.6330, loss_cns_1: 0.6600, loss_yns_1: 0.1549, loss_cls_2: 0.8869, loss_box_2: 1.6010, loss_cns_2: 0.6630, loss_yns_2: 0.1550, loss_cls_3: 0.9025, loss_box_3: 1.5889, loss_cns_3: 0.6681, loss_yns_3: 0.1570, loss_cls_4: 0.9095, loss_box_4: 1.5833, loss_cns_4: 0.6621, loss_yns_4: 0.1563, loss_cls_5: 0.9001, loss_box_5: 1.5683, loss_cns_5: 0.6681, loss_yns_5: 0.1560, loss_cls_dn_0: 0.2067, loss_box_dn_0: 0.7436, loss_cls_dn_1: 0.1197, loss_box_dn_1: 0.6968, loss_cls_dn_2: 0.1177, loss_box_dn_2: 0.6876, loss_cls_dn_3: 0.1195, loss_box_dn_3: 0.6924, loss_cls_dn_4: 0.1232, loss_box_dn_4: 0.7051, loss_cls_dn_5: 0.1232, loss_box_dn_5: 0.7156, loss_dense_depth: 0.6922, loss: 25.5818, grad_norm: 47.3618 -2025-11-12 14:37:28,590 - mmdet - INFO - Iter [250/17500] lr: 1.995e-04, eta: 10:07:25, time: 1.587, data_time: 0.080, memory: 49164, loss_cls_0: 0.7713, loss_box_0: 1.6675, loss_cns_0: 0.6242, loss_yns_0: 0.1517, loss_cls_1: 0.8646, loss_box_1: 1.6008, loss_cns_1: 0.6608, loss_yns_1: 0.1515, loss_cls_2: 0.8678, loss_box_2: 1.5924, loss_cns_2: 0.6618, loss_yns_2: 0.1530, loss_cls_3: 0.8766, loss_box_3: 1.5751, loss_cns_3: 0.6602, loss_yns_3: 0.1542, loss_cls_4: 0.8976, loss_box_4: 1.5571, loss_cns_4: 0.6627, loss_yns_4: 0.1520, loss_cls_5: 0.9049, loss_box_5: 1.5549, loss_cns_5: 0.6573, loss_yns_5: 0.1548, loss_cls_dn_0: 0.1949, loss_box_dn_0: 0.7399, loss_cls_dn_1: 0.1213, loss_box_dn_1: 0.6958, loss_cls_dn_2: 0.1163, loss_box_dn_2: 0.6855, loss_cls_dn_3: 0.1198, loss_box_dn_3: 0.6864, loss_cls_dn_4: 0.1266, loss_box_dn_4: 0.6970, loss_cls_dn_5: 0.1311, loss_box_dn_5: 0.6972, loss_dense_depth: 0.7158, loss: 25.3023, grad_norm: 46.8600 -2025-11-12 14:37:30,205 - mmdet - INFO - Iter [251/17500] lr: 1.999e-04, eta: 10:06:48, time: 1.614, data_time: 0.075, memory: 49164, loss_cls_0: 0.7770, loss_box_0: 1.6270, loss_cns_0: 0.6221, loss_yns_0: 0.1496, loss_cls_1: 0.8533, loss_box_1: 1.5743, loss_cns_1: 0.6464, loss_yns_1: 0.1486, loss_cls_2: 0.8757, loss_box_2: 1.5791, loss_cns_2: 0.6590, loss_yns_2: 0.1506, loss_cls_3: 0.8896, loss_box_3: 1.5758, loss_cns_3: 0.6610, loss_yns_3: 0.1500, loss_cls_4: 0.9464, loss_box_4: 1.4900, loss_cns_4: 0.6538, loss_yns_4: 0.1461, loss_cls_5: 0.9201, loss_box_5: 1.5311, loss_cns_5: 0.6572, loss_yns_5: 0.1545, loss_cls_dn_0: 0.1913, loss_box_dn_0: 0.7457, loss_cls_dn_1: 0.1216, loss_box_dn_1: 0.6888, loss_cls_dn_2: 0.1203, loss_box_dn_2: 0.6773, loss_cls_dn_3: 0.1229, loss_box_dn_3: 0.6835, loss_cls_dn_4: 0.1276, loss_box_dn_4: 0.6773, loss_cls_dn_5: 0.1329, loss_box_dn_5: 0.6741, loss_dense_depth: 0.7538, loss: 25.1553, grad_norm: 48.7986 -2025-11-12 14:37:31,793 - mmdet - INFO - Iter [252/17500] lr: 2.003e-04, eta: 10:06:10, time: 1.589, data_time: 0.075, memory: 49164, loss_cls_0: 0.7550, loss_box_0: 1.6623, loss_cns_0: 0.6237, loss_yns_0: 0.1507, loss_cls_1: 0.8294, loss_box_1: 1.5866, loss_cns_1: 0.6533, loss_yns_1: 0.1526, loss_cls_2: 0.8758, loss_box_2: 1.5638, loss_cns_2: 0.6582, loss_yns_2: 0.1526, loss_cls_3: 0.8500, loss_box_3: 1.5903, loss_cns_3: 0.6666, loss_yns_3: 0.1506, loss_cls_4: 0.8704, loss_box_4: 1.5486, loss_cns_4: 0.6622, loss_yns_4: 0.1508, loss_cls_5: 0.8556, loss_box_5: 1.5805, loss_cns_5: 0.6636, loss_yns_5: 0.1530, loss_cls_dn_0: 0.1906, loss_box_dn_0: 0.7363, loss_cls_dn_1: 0.1177, loss_box_dn_1: 0.6666, loss_cls_dn_2: 0.1161, loss_box_dn_2: 0.6566, loss_cls_dn_3: 0.1161, loss_box_dn_3: 0.6673, loss_cls_dn_4: 0.1193, loss_box_dn_4: 0.6620, loss_cls_dn_5: 0.1204, loss_box_dn_5: 0.6729, loss_dense_depth: 0.7163, loss: 24.9644, grad_norm: 40.9789 -2025-11-12 14:37:33,377 - mmdet - INFO - Iter [253/17500] lr: 2.007e-04, eta: 10:05:33, time: 1.585, data_time: 0.074, memory: 49164, loss_cls_0: 0.7705, loss_box_0: 1.6786, loss_cns_0: 0.6281, loss_yns_0: 0.1466, loss_cls_1: 0.8380, loss_box_1: 1.5507, loss_cns_1: 0.6598, loss_yns_1: 0.1478, loss_cls_2: 0.8406, loss_box_2: 1.5330, loss_cns_2: 0.6596, loss_yns_2: 0.1477, loss_cls_3: 0.8650, loss_box_3: 1.5320, loss_cns_3: 0.6649, loss_yns_3: 0.1478, loss_cls_4: 0.8747, loss_box_4: 1.5350, loss_cns_4: 0.6612, loss_yns_4: 0.1480, loss_cls_5: 0.8817, loss_box_5: 1.5418, loss_cns_5: 0.6618, loss_yns_5: 0.1483, loss_cls_dn_0: 0.1977, loss_box_dn_0: 0.7443, loss_cls_dn_1: 0.1167, loss_box_dn_1: 0.6641, loss_cls_dn_2: 0.1173, loss_box_dn_2: 0.6602, loss_cls_dn_3: 0.1212, loss_box_dn_3: 0.6636, loss_cls_dn_4: 0.1234, loss_box_dn_4: 0.6748, loss_cls_dn_5: 0.1221, loss_box_dn_5: 0.6846, loss_dense_depth: 0.7271, loss: 24.8800, grad_norm: 40.7650 -2025-11-12 14:37:34,949 - mmdet - INFO - Iter [254/17500] lr: 2.011e-04, eta: 10:04:54, time: 1.570, data_time: 0.074, memory: 49164, loss_cls_0: 0.7739, loss_box_0: 1.6852, loss_cns_0: 0.6241, loss_yns_0: 0.1484, loss_cls_1: 0.8689, loss_box_1: 1.5646, loss_cns_1: 0.6617, loss_yns_1: 0.1469, loss_cls_2: 0.8667, loss_box_2: 1.5474, loss_cns_2: 0.6623, loss_yns_2: 0.1479, loss_cls_3: 0.8725, loss_box_3: 1.5468, loss_cns_3: 0.6638, loss_yns_3: 0.1463, loss_cls_4: 0.8980, loss_box_4: 1.5662, loss_cns_4: 0.6630, loss_yns_4: 0.1481, loss_cls_5: 0.8861, loss_box_5: 1.5682, loss_cns_5: 0.6628, loss_yns_5: 0.1481, loss_cls_dn_0: 0.1938, loss_box_dn_0: 0.7420, loss_cls_dn_1: 0.1201, loss_box_dn_1: 0.6815, loss_cls_dn_2: 0.1230, loss_box_dn_2: 0.6778, loss_cls_dn_3: 0.1274, loss_box_dn_3: 0.6858, loss_cls_dn_4: 0.1276, loss_box_dn_4: 0.7014, loss_cls_dn_5: 0.1270, loss_box_dn_5: 0.7094, loss_dense_depth: 0.7181, loss: 25.2029, grad_norm: 46.4580 -2025-11-12 14:37:36,535 - mmdet - INFO - Iter [255/17500] lr: 2.015e-04, eta: 10:04:17, time: 1.587, data_time: 0.077, memory: 49164, loss_cls_0: 0.7522, loss_box_0: 1.6502, loss_cns_0: 0.6288, loss_yns_0: 0.1497, loss_cls_1: 0.8636, loss_box_1: 1.5606, loss_cns_1: 0.6619, loss_yns_1: 0.1504, loss_cls_2: 0.8733, loss_box_2: 1.5199, loss_cns_2: 0.6644, loss_yns_2: 0.1497, loss_cls_3: 0.8571, loss_box_3: 1.5362, loss_cns_3: 0.6651, loss_yns_3: 0.1502, loss_cls_4: 0.8752, loss_box_4: 1.5425, loss_cns_4: 0.6656, loss_yns_4: 0.1506, loss_cls_5: 0.8659, loss_box_5: 1.5492, loss_cns_5: 0.6638, loss_yns_5: 0.1521, loss_cls_dn_0: 0.1855, loss_box_dn_0: 0.7382, loss_cls_dn_1: 0.1193, loss_box_dn_1: 0.7005, loss_cls_dn_2: 0.1193, loss_box_dn_2: 0.6855, loss_cls_dn_3: 0.1207, loss_box_dn_3: 0.6973, loss_cls_dn_4: 0.1213, loss_box_dn_4: 0.7051, loss_cls_dn_5: 0.1229, loss_box_dn_5: 0.7146, loss_dense_depth: 0.7203, loss: 25.0486, grad_norm: 45.1219 -2025-11-12 14:37:38,119 - mmdet - INFO - Iter [256/17500] lr: 2.019e-04, eta: 10:03:40, time: 1.583, data_time: 0.080, memory: 49164, loss_cls_0: 0.7627, loss_box_0: 1.6999, loss_cns_0: 0.6277, loss_yns_0: 0.1500, loss_cls_1: 0.8762, loss_box_1: 1.5365, loss_cns_1: 0.6641, loss_yns_1: 0.1480, loss_cls_2: 0.8928, loss_box_2: 1.5061, loss_cns_2: 0.6659, loss_yns_2: 0.1486, loss_cls_3: 0.8712, loss_box_3: 1.4864, loss_cns_3: 0.6657, loss_yns_3: 0.1471, loss_cls_4: 0.8833, loss_box_4: 1.4861, loss_cns_4: 0.6699, loss_yns_4: 0.1479, loss_cls_5: 0.8720, loss_box_5: 1.4857, loss_cns_5: 0.6694, loss_yns_5: 0.1503, loss_cls_dn_0: 0.1824, loss_box_dn_0: 0.7379, loss_cls_dn_1: 0.1117, loss_box_dn_1: 0.6720, loss_cls_dn_2: 0.1138, loss_box_dn_2: 0.6560, loss_cls_dn_3: 0.1136, loss_box_dn_3: 0.6490, loss_cls_dn_4: 0.1128, loss_box_dn_4: 0.6495, loss_cls_dn_5: 0.1163, loss_box_dn_5: 0.6563, loss_dense_depth: 0.7261, loss: 24.7110, grad_norm: 32.6382 -2025-11-12 14:37:39,690 - mmdet - INFO - Iter [257/17500] lr: 2.023e-04, eta: 10:03:02, time: 1.572, data_time: 0.073, memory: 49164, loss_cls_0: 0.7928, loss_box_0: 1.7255, loss_cns_0: 0.6280, loss_yns_0: 0.1496, loss_cls_1: 0.8663, loss_box_1: 1.5869, loss_cns_1: 0.6593, loss_yns_1: 0.1474, loss_cls_2: 0.8834, loss_box_2: 1.5480, loss_cns_2: 0.6598, loss_yns_2: 0.1471, loss_cls_3: 0.8758, loss_box_3: 1.5376, loss_cns_3: 0.6629, loss_yns_3: 0.1470, loss_cls_4: 0.8887, loss_box_4: 1.5604, loss_cns_4: 0.6616, loss_yns_4: 0.1478, loss_cls_5: 0.8800, loss_box_5: 1.5527, loss_cns_5: 0.6595, loss_yns_5: 0.1470, loss_cls_dn_0: 0.1888, loss_box_dn_0: 0.7342, loss_cls_dn_1: 0.1112, loss_box_dn_1: 0.6635, loss_cls_dn_2: 0.1140, loss_box_dn_2: 0.6455, loss_cls_dn_3: 0.1211, loss_box_dn_3: 0.6398, loss_cls_dn_4: 0.1170, loss_box_dn_4: 0.6520, loss_cls_dn_5: 0.1189, loss_box_dn_5: 0.6549, loss_dense_depth: 0.7431, loss: 25.0189, grad_norm: 37.4816 -2025-11-12 14:37:41,250 - mmdet - INFO - Iter [258/17500] lr: 2.027e-04, eta: 10:02:24, time: 1.560, data_time: 0.073, memory: 49164, loss_cls_0: 0.7897, loss_box_0: 1.7276, loss_cns_0: 0.6227, loss_yns_0: 0.1471, loss_cls_1: 0.8650, loss_box_1: 1.5680, loss_cns_1: 0.6561, loss_yns_1: 0.1448, loss_cls_2: 0.8834, loss_box_2: 1.5233, loss_cns_2: 0.6610, loss_yns_2: 0.1448, loss_cls_3: 0.8768, loss_box_3: 1.5255, loss_cns_3: 0.6607, loss_yns_3: 0.1445, loss_cls_4: 0.8951, loss_box_4: 1.5338, loss_cns_4: 0.6615, loss_yns_4: 0.1458, loss_cls_5: 0.8911, loss_box_5: 1.5212, loss_cns_5: 0.6592, loss_yns_5: 0.1436, loss_cls_dn_0: 0.1854, loss_box_dn_0: 0.7299, loss_cls_dn_1: 0.1185, loss_box_dn_1: 0.6565, loss_cls_dn_2: 0.1211, loss_box_dn_2: 0.6346, loss_cls_dn_3: 0.1220, loss_box_dn_3: 0.6378, loss_cls_dn_4: 0.1177, loss_box_dn_4: 0.6465, loss_cls_dn_5: 0.1193, loss_box_dn_5: 0.6474, loss_dense_depth: 0.7195, loss: 24.8487, grad_norm: 33.5882 -2025-11-12 14:37:42,816 - mmdet - INFO - Iter [259/17500] lr: 2.031e-04, eta: 10:01:47, time: 1.561, data_time: 0.074, memory: 49164, loss_cls_0: 0.7988, loss_box_0: 1.7136, loss_cns_0: 0.6185, loss_yns_0: 0.1481, loss_cls_1: 0.8663, loss_box_1: 1.6393, loss_cns_1: 0.6475, loss_yns_1: 0.1436, loss_cls_2: 0.8833, loss_box_2: 1.6198, loss_cns_2: 0.6566, loss_yns_2: 0.1455, loss_cls_3: 0.8914, loss_box_3: 1.5905, loss_cns_3: 0.6521, loss_yns_3: 0.1451, loss_cls_4: 0.9355, loss_box_4: 1.5642, loss_cns_4: 0.6544, loss_yns_4: 0.1450, loss_cls_5: 0.9027, loss_box_5: 1.5918, loss_cns_5: 0.6525, loss_yns_5: 0.1451, loss_cls_dn_0: 0.1941, loss_box_dn_0: 0.7413, loss_cls_dn_1: 0.1201, loss_box_dn_1: 0.6619, loss_cls_dn_2: 0.1230, loss_box_dn_2: 0.6546, loss_cls_dn_3: 0.1240, loss_box_dn_3: 0.6553, loss_cls_dn_4: 0.1230, loss_box_dn_4: 0.6551, loss_cls_dn_5: 0.1263, loss_box_dn_5: 0.6643, loss_dense_depth: 0.7619, loss: 25.3560, grad_norm: 34.8164 -2025-11-12 14:37:54,221 - mmdet - INFO - Iter [260/17500] lr: 2.035e-04, eta: 10:12:02, time: 11.410, data_time: 0.074, memory: 49164, loss_cls_0: 0.7850, loss_box_0: 1.6841, loss_cns_0: 0.6236, loss_yns_0: 0.1525, loss_cls_1: 0.8592, loss_box_1: 1.6565, loss_cns_1: 0.6490, loss_yns_1: 0.1515, loss_cls_2: 0.8885, loss_box_2: 1.6398, loss_cns_2: 0.6565, loss_yns_2: 0.1518, loss_cls_3: 0.8791, loss_box_3: 1.6112, loss_cns_3: 0.6549, loss_yns_3: 0.1511, loss_cls_4: 0.8988, loss_box_4: 1.5868, loss_cns_4: 0.6568, loss_yns_4: 0.1509, loss_cls_5: 0.9266, loss_box_5: 1.6155, loss_cns_5: 0.6557, loss_yns_5: 0.1507, loss_cls_dn_0: 0.1888, loss_box_dn_0: 0.7404, loss_cls_dn_1: 0.1219, loss_box_dn_1: 0.6788, loss_cls_dn_2: 0.1264, loss_box_dn_2: 0.6731, loss_cls_dn_3: 0.1297, loss_box_dn_3: 0.6703, loss_cls_dn_4: 0.1266, loss_box_dn_4: 0.6707, loss_cls_dn_5: 0.1405, loss_box_dn_5: 0.6812, loss_dense_depth: 0.7369, loss: 25.5214, grad_norm: 45.1137 -2025-11-12 14:37:55,852 - mmdet - INFO - Iter [261/17500] lr: 2.039e-04, eta: 10:11:27, time: 1.624, data_time: 0.093, memory: 49164, loss_cls_0: 0.7971, loss_box_0: 1.6843, loss_cns_0: 0.6221, loss_yns_0: 0.1476, loss_cls_1: 0.8799, loss_box_1: 1.5979, loss_cns_1: 0.6589, loss_yns_1: 0.1467, loss_cls_2: 0.8876, loss_box_2: 1.5714, loss_cns_2: 0.6611, loss_yns_2: 0.1470, loss_cls_3: 0.8915, loss_box_3: 1.5573, loss_cns_3: 0.6598, loss_yns_3: 0.1466, loss_cls_4: 0.9111, loss_box_4: 1.5650, loss_cns_4: 0.6604, loss_yns_4: 0.1473, loss_cls_5: 0.9060, loss_box_5: 1.5633, loss_cns_5: 0.6599, loss_yns_5: 0.1467, loss_cls_dn_0: 0.1876, loss_box_dn_0: 0.7373, loss_cls_dn_1: 0.1207, loss_box_dn_1: 0.6854, loss_cls_dn_2: 0.1219, loss_box_dn_2: 0.6763, loss_cls_dn_3: 0.1254, loss_box_dn_3: 0.6757, loss_cls_dn_4: 0.1247, loss_box_dn_4: 0.6812, loss_cls_dn_5: 0.1250, loss_box_dn_5: 0.6850, loss_dense_depth: 0.7227, loss: 25.2853, grad_norm: 29.7937 -2025-11-12 14:37:57,444 - mmdet - INFO - Iter [262/17500] lr: 2.043e-04, eta: 10:10:49, time: 1.594, data_time: 0.106, memory: 49164, loss_cls_0: 0.7931, loss_box_0: 1.6864, loss_cns_0: 0.6252, loss_yns_0: 0.1456, loss_cls_1: 0.8701, loss_box_1: 1.5801, loss_cns_1: 0.6592, loss_yns_1: 0.1465, loss_cls_2: 0.8888, loss_box_2: 1.5494, loss_cns_2: 0.6583, loss_yns_2: 0.1475, loss_cls_3: 0.8886, loss_box_3: 1.5484, loss_cns_3: 0.6622, loss_yns_3: 0.1434, loss_cls_4: 0.8994, loss_box_4: 1.5475, loss_cns_4: 0.6598, loss_yns_4: 0.1466, loss_cls_5: 0.9061, loss_box_5: 1.5263, loss_cns_5: 0.6578, loss_yns_5: 0.1449, loss_cls_dn_0: 0.1853, loss_box_dn_0: 0.7399, loss_cls_dn_1: 0.1143, loss_box_dn_1: 0.6773, loss_cls_dn_2: 0.1185, loss_box_dn_2: 0.6620, loss_cls_dn_3: 0.1229, loss_box_dn_3: 0.6648, loss_cls_dn_4: 0.1188, loss_box_dn_4: 0.6649, loss_cls_dn_5: 0.1234, loss_box_dn_5: 0.6633, loss_dense_depth: 0.7966, loss: 25.1330, grad_norm: 31.7405 -2025-11-12 14:37:59,042 - mmdet - INFO - Iter [263/17500] lr: 2.047e-04, eta: 10:10:13, time: 1.597, data_time: 0.082, memory: 49164, loss_cls_0: 0.7701, loss_box_0: 1.7158, loss_cns_0: 0.6217, loss_yns_0: 0.1472, loss_cls_1: 0.8595, loss_box_1: 1.6121, loss_cns_1: 0.6558, loss_yns_1: 0.1440, loss_cls_2: 0.8910, loss_box_2: 1.5601, loss_cns_2: 0.6543, loss_yns_2: 0.1433, loss_cls_3: 0.8866, loss_box_3: 1.5730, loss_cns_3: 0.6616, loss_yns_3: 0.1434, loss_cls_4: 0.8962, loss_box_4: 1.5593, loss_cns_4: 0.6588, loss_yns_4: 0.1449, loss_cls_5: 0.8961, loss_box_5: 1.5851, loss_cns_5: 0.6572, loss_yns_5: 0.1438, loss_cls_dn_0: 0.1864, loss_box_dn_0: 0.7619, loss_cls_dn_1: 0.1127, loss_box_dn_1: 0.6788, loss_cls_dn_2: 0.1169, loss_box_dn_2: 0.6581, loss_cls_dn_3: 0.1168, loss_box_dn_3: 0.6627, loss_cls_dn_4: 0.1175, loss_box_dn_4: 0.6551, loss_cls_dn_5: 0.1206, loss_box_dn_5: 0.6695, loss_dense_depth: 0.8062, loss: 25.2439, grad_norm: 37.9026 -2025-11-12 14:38:00,626 - mmdet - INFO - Iter [264/17500] lr: 2.051e-04, eta: 10:09:35, time: 1.582, data_time: 0.075, memory: 49164, loss_cls_0: 0.7644, loss_box_0: 1.7060, loss_cns_0: 0.6260, loss_yns_0: 0.1468, loss_cls_1: 0.8653, loss_box_1: 1.6042, loss_cns_1: 0.6617, loss_yns_1: 0.1431, loss_cls_2: 0.8839, loss_box_2: 1.5642, loss_cns_2: 0.6595, loss_yns_2: 0.1425, loss_cls_3: 0.9044, loss_box_3: 1.5617, loss_cns_3: 0.6616, loss_yns_3: 0.1432, loss_cls_4: 0.9223, loss_box_4: 1.5460, loss_cns_4: 0.6606, loss_yns_4: 0.1415, loss_cls_5: 0.8998, loss_box_5: 1.5520, loss_cns_5: 0.6606, loss_yns_5: 0.1420, loss_cls_dn_0: 0.1822, loss_box_dn_0: 0.7411, loss_cls_dn_1: 0.1148, loss_box_dn_1: 0.6676, loss_cls_dn_2: 0.1205, loss_box_dn_2: 0.6481, loss_cls_dn_3: 0.1192, loss_box_dn_3: 0.6512, loss_cls_dn_4: 0.1271, loss_box_dn_4: 0.6505, loss_cls_dn_5: 0.1217, loss_box_dn_5: 0.6644, loss_dense_depth: 0.7760, loss: 25.1478, grad_norm: 28.5703 -2025-11-12 14:38:02,206 - mmdet - INFO - Iter [265/17500] lr: 2.055e-04, eta: 10:08:57, time: 1.576, data_time: 0.087, memory: 49164, loss_cls_0: 0.7962, loss_box_0: 1.6946, loss_cns_0: 0.6252, loss_yns_0: 0.1426, loss_cls_1: 0.8873, loss_box_1: 1.5807, loss_cns_1: 0.6600, loss_yns_1: 0.1428, loss_cls_2: 0.8942, loss_box_2: 1.5699, loss_cns_2: 0.6588, loss_yns_2: 0.1431, loss_cls_3: 0.9045, loss_box_3: 1.5563, loss_cns_3: 0.6641, loss_yns_3: 0.1417, loss_cls_4: 0.9200, loss_box_4: 1.5726, loss_cns_4: 0.6594, loss_yns_4: 0.1446, loss_cls_5: 0.9090, loss_box_5: 1.5862, loss_cns_5: 0.6573, loss_yns_5: 0.1469, loss_cls_dn_0: 0.1936, loss_box_dn_0: 0.7513, loss_cls_dn_1: 0.1134, loss_box_dn_1: 0.6805, loss_cls_dn_2: 0.1156, loss_box_dn_2: 0.6779, loss_cls_dn_3: 0.1168, loss_box_dn_3: 0.6756, loss_cls_dn_4: 0.1172, loss_box_dn_4: 0.6923, loss_cls_dn_5: 0.1248, loss_box_dn_5: 0.7074, loss_dense_depth: 0.8771, loss: 25.5017, grad_norm: 50.5992 -2025-11-12 14:38:03,803 - mmdet - INFO - Iter [266/17500] lr: 2.059e-04, eta: 10:08:22, time: 1.603, data_time: 0.119, memory: 49164, loss_cls_0: 0.7851, loss_box_0: 1.6565, loss_cns_0: 0.6264, loss_yns_0: 0.1399, loss_cls_1: 0.8983, loss_box_1: 1.5490, loss_cns_1: 0.6597, loss_yns_1: 0.1435, loss_cls_2: 0.8934, loss_box_2: 1.5202, loss_cns_2: 0.6581, loss_yns_2: 0.1432, loss_cls_3: 0.9238, loss_box_3: 1.5273, loss_cns_3: 0.6708, loss_yns_3: 0.1406, loss_cls_4: 0.9320, loss_box_4: 1.5441, loss_cns_4: 0.6629, loss_yns_4: 0.1436, loss_cls_5: 0.9016, loss_box_5: 1.5645, loss_cns_5: 0.6638, loss_yns_5: 0.1459, loss_cls_dn_0: 0.1942, loss_box_dn_0: 0.7508, loss_cls_dn_1: 0.1200, loss_box_dn_1: 0.6917, loss_cls_dn_2: 0.1167, loss_box_dn_2: 0.6865, loss_cls_dn_3: 0.1212, loss_box_dn_3: 0.6918, loss_cls_dn_4: 0.1247, loss_box_dn_4: 0.7052, loss_cls_dn_5: 0.1345, loss_box_dn_5: 0.7202, loss_dense_depth: 0.8026, loss: 25.3543, grad_norm: 44.0204 -2025-11-12 14:38:05,376 - mmdet - INFO - Iter [267/17500] lr: 2.063e-04, eta: 10:07:45, time: 1.579, data_time: 0.079, memory: 49164, loss_cls_0: 0.7547, loss_box_0: 1.6633, loss_cns_0: 0.6280, loss_yns_0: 0.1463, loss_cls_1: 0.8636, loss_box_1: 1.5580, loss_cns_1: 0.6600, loss_yns_1: 0.1446, loss_cls_2: 0.8705, loss_box_2: 1.5238, loss_cns_2: 0.6588, loss_yns_2: 0.1450, loss_cls_3: 0.8986, loss_box_3: 1.5224, loss_cns_3: 0.6622, loss_yns_3: 0.1406, loss_cls_4: 0.9172, loss_box_4: 1.5210, loss_cns_4: 0.6613, loss_yns_4: 0.1418, loss_cls_5: 0.8857, loss_box_5: 1.5122, loss_cns_5: 0.6625, loss_yns_5: 0.1419, loss_cls_dn_0: 0.1897, loss_box_dn_0: 0.7380, loss_cls_dn_1: 0.1201, loss_box_dn_1: 0.7014, loss_cls_dn_2: 0.1171, loss_box_dn_2: 0.6908, loss_cls_dn_3: 0.1204, loss_box_dn_3: 0.6934, loss_cls_dn_4: 0.1238, loss_box_dn_4: 0.6992, loss_cls_dn_5: 0.1232, loss_box_dn_5: 0.6994, loss_dense_depth: 0.8344, loss: 25.1350, grad_norm: 45.6930 -2025-11-12 14:38:06,964 - mmdet - INFO - Iter [268/17500] lr: 2.067e-04, eta: 10:07:09, time: 1.588, data_time: 0.086, memory: 49164, loss_cls_0: 0.7961, loss_box_0: 1.6540, loss_cns_0: 0.6384, loss_yns_0: 0.1475, loss_cls_1: 0.8529, loss_box_1: 1.5392, loss_cns_1: 0.6642, loss_yns_1: 0.1440, loss_cls_2: 0.8826, loss_box_2: 1.4899, loss_cns_2: 0.6652, loss_yns_2: 0.1420, loss_cls_3: 0.8678, loss_box_3: 1.4751, loss_cns_3: 0.6660, loss_yns_3: 0.1413, loss_cls_4: 0.8865, loss_box_4: 1.4737, loss_cns_4: 0.6651, loss_yns_4: 0.1423, loss_cls_5: 0.8882, loss_box_5: 1.4755, loss_cns_5: 0.6664, loss_yns_5: 0.1431, loss_cls_dn_0: 0.1819, loss_box_dn_0: 0.7375, loss_cls_dn_1: 0.1171, loss_box_dn_1: 0.6933, loss_cls_dn_2: 0.1201, loss_box_dn_2: 0.6767, loss_cls_dn_3: 0.1213, loss_box_dn_3: 0.6721, loss_cls_dn_4: 0.1194, loss_box_dn_4: 0.6747, loss_cls_dn_5: 0.1317, loss_box_dn_5: 0.6780, loss_dense_depth: 0.7688, loss: 24.7997, grad_norm: 34.3213 -2025-11-12 14:38:08,541 - mmdet - INFO - Iter [269/17500] lr: 2.071e-04, eta: 10:06:32, time: 1.572, data_time: 0.082, memory: 49164, loss_cls_0: 0.7503, loss_box_0: 1.6398, loss_cns_0: 0.6318, loss_yns_0: 0.1445, loss_cls_1: 0.8503, loss_box_1: 1.5409, loss_cns_1: 0.6607, loss_yns_1: 0.1425, loss_cls_2: 0.8471, loss_box_2: 1.5191, loss_cns_2: 0.6610, loss_yns_2: 0.1432, loss_cls_3: 0.8675, loss_box_3: 1.5379, loss_cns_3: 0.6642, loss_yns_3: 0.1440, loss_cls_4: 0.8770, loss_box_4: 1.5338, loss_cns_4: 0.6624, loss_yns_4: 0.1474, loss_cls_5: 0.8694, loss_box_5: 1.5320, loss_cns_5: 0.6674, loss_yns_5: 0.1452, loss_cls_dn_0: 0.1733, loss_box_dn_0: 0.7409, loss_cls_dn_1: 0.1118, loss_box_dn_1: 0.6917, loss_cls_dn_2: 0.1106, loss_box_dn_2: 0.6785, loss_cls_dn_3: 0.1132, loss_box_dn_3: 0.6802, loss_cls_dn_4: 0.1152, loss_box_dn_4: 0.6849, loss_cls_dn_5: 0.1225, loss_box_dn_5: 0.6896, loss_dense_depth: 0.8043, loss: 24.8960, grad_norm: 54.8871 -2025-11-12 14:38:10,103 - mmdet - INFO - Iter [270/17500] lr: 2.075e-04, eta: 10:05:55, time: 1.565, data_time: 0.084, memory: 49164, loss_cls_0: 0.8005, loss_box_0: 1.6652, loss_cns_0: 0.6234, loss_yns_0: 0.1450, loss_cls_1: 0.9061, loss_box_1: 1.5778, loss_cns_1: 0.6482, loss_yns_1: 0.1434, loss_cls_2: 0.9054, loss_box_2: 1.5338, loss_cns_2: 0.6471, loss_yns_2: 0.1448, loss_cls_3: 0.9057, loss_box_3: 1.5614, loss_cns_3: 0.6581, loss_yns_3: 0.1459, loss_cls_4: 0.8918, loss_box_4: 1.5648, loss_cns_4: 0.6577, loss_yns_4: 0.1493, loss_cls_5: 0.9145, loss_box_5: 1.5479, loss_cns_5: 0.6605, loss_yns_5: 0.1478, loss_cls_dn_0: 0.1902, loss_box_dn_0: 0.7411, loss_cls_dn_1: 0.1117, loss_box_dn_1: 0.6989, loss_cls_dn_2: 0.1123, loss_box_dn_2: 0.6810, loss_cls_dn_3: 0.1157, loss_box_dn_3: 0.6866, loss_cls_dn_4: 0.1169, loss_box_dn_4: 0.6906, loss_cls_dn_5: 0.1194, loss_box_dn_5: 0.6975, loss_dense_depth: 0.8201, loss: 25.3284, grad_norm: 52.7024 -2025-11-12 14:38:11,668 - mmdet - INFO - Iter [271/17500] lr: 2.079e-04, eta: 10:05:18, time: 1.566, data_time: 0.075, memory: 49164, loss_cls_0: 0.7391, loss_box_0: 1.6546, loss_cns_0: 0.6343, loss_yns_0: 0.1480, loss_cls_1: 0.8541, loss_box_1: 1.5724, loss_cns_1: 0.6500, loss_yns_1: 0.1462, loss_cls_2: 0.8544, loss_box_2: 1.5392, loss_cns_2: 0.6545, loss_yns_2: 0.1456, loss_cls_3: 0.8713, loss_box_3: 1.5346, loss_cns_3: 0.6616, loss_yns_3: 0.1451, loss_cls_4: 0.8689, loss_box_4: 1.5434, loss_cns_4: 0.6612, loss_yns_4: 0.1457, loss_cls_5: 0.8760, loss_box_5: 1.5598, loss_cns_5: 0.6603, loss_yns_5: 0.1470, loss_cls_dn_0: 0.1871, loss_box_dn_0: 0.7497, loss_cls_dn_1: 0.1104, loss_box_dn_1: 0.6916, loss_cls_dn_2: 0.1094, loss_box_dn_2: 0.6790, loss_cls_dn_3: 0.1127, loss_box_dn_3: 0.6795, loss_cls_dn_4: 0.1143, loss_box_dn_4: 0.6855, loss_cls_dn_5: 0.1141, loss_box_dn_5: 0.7044, loss_dense_depth: 0.7438, loss: 24.9486, grad_norm: 39.9357 -2025-11-12 14:38:13,232 - mmdet - INFO - Iter [272/17500] lr: 2.083e-04, eta: 10:04:41, time: 1.558, data_time: 0.073, memory: 49164, loss_cls_0: 0.7438, loss_box_0: 1.6594, loss_cns_0: 0.6313, loss_yns_0: 0.1479, loss_cls_1: 0.8416, loss_box_1: 1.5562, loss_cns_1: 0.6593, loss_yns_1: 0.1486, loss_cls_2: 0.8470, loss_box_2: 1.5578, loss_cns_2: 0.6659, loss_yns_2: 0.1480, loss_cls_3: 0.9152, loss_box_3: 1.5278, loss_cns_3: 0.6656, loss_yns_3: 0.1487, loss_cls_4: 0.9183, loss_box_4: 1.5347, loss_cns_4: 0.6636, loss_yns_4: 0.1492, loss_cls_5: 0.8614, loss_box_5: 1.5327, loss_cns_5: 0.6631, loss_yns_5: 0.1498, loss_cls_dn_0: 0.1891, loss_box_dn_0: 0.7390, loss_cls_dn_1: 0.1068, loss_box_dn_1: 0.6897, loss_cls_dn_2: 0.1080, loss_box_dn_2: 0.6944, loss_cls_dn_3: 0.1128, loss_box_dn_3: 0.6894, loss_cls_dn_4: 0.1168, loss_box_dn_4: 0.7009, loss_cls_dn_5: 0.1144, loss_box_dn_5: 0.7127, loss_dense_depth: 0.8142, loss: 25.1249, grad_norm: 46.1240 -2025-11-12 14:38:14,794 - mmdet - INFO - Iter [273/17500] lr: 2.087e-04, eta: 10:04:05, time: 1.566, data_time: 0.079, memory: 49164, loss_cls_0: 0.7519, loss_box_0: 1.6432, loss_cns_0: 0.6270, loss_yns_0: 0.1470, loss_cls_1: 0.8307, loss_box_1: 1.5398, loss_cns_1: 0.6588, loss_yns_1: 0.1444, loss_cls_2: 0.8430, loss_box_2: 1.5437, loss_cns_2: 0.6623, loss_yns_2: 0.1440, loss_cls_3: 0.8561, loss_box_3: 1.5153, loss_cns_3: 0.6629, loss_yns_3: 0.1450, loss_cls_4: 0.8638, loss_box_4: 1.5262, loss_cns_4: 0.6616, loss_yns_4: 0.1474, loss_cls_5: 0.8542, loss_box_5: 1.5269, loss_cns_5: 0.6612, loss_yns_5: 0.1445, loss_cls_dn_0: 0.1828, loss_box_dn_0: 0.7390, loss_cls_dn_1: 0.1092, loss_box_dn_1: 0.6893, loss_cls_dn_2: 0.1090, loss_box_dn_2: 0.6963, loss_cls_dn_3: 0.1116, loss_box_dn_3: 0.6859, loss_cls_dn_4: 0.1144, loss_box_dn_4: 0.6982, loss_cls_dn_5: 0.1141, loss_box_dn_5: 0.7044, loss_dense_depth: 0.7258, loss: 24.7810, grad_norm: 47.5405 -2025-11-12 14:38:16,432 - mmdet - INFO - Iter [274/17500] lr: 2.091e-04, eta: 10:03:33, time: 1.635, data_time: 0.077, memory: 49164, loss_cls_0: 0.7558, loss_box_0: 1.6569, loss_cns_0: 0.6276, loss_yns_0: 0.1453, loss_cls_1: 0.8159, loss_box_1: 1.5247, loss_cns_1: 0.6591, loss_yns_1: 0.1423, loss_cls_2: 0.8358, loss_box_2: 1.4912, loss_cns_2: 0.6623, loss_yns_2: 0.1417, loss_cls_3: 0.8584, loss_box_3: 1.4608, loss_cns_3: 0.6609, loss_yns_3: 0.1423, loss_cls_4: 0.8659, loss_box_4: 1.4585, loss_cns_4: 0.6617, loss_yns_4: 0.1461, loss_cls_5: 0.8528, loss_box_5: 1.4663, loss_cns_5: 0.6704, loss_yns_5: 0.1439, loss_cls_dn_0: 0.1821, loss_box_dn_0: 0.7407, loss_cls_dn_1: 0.1076, loss_box_dn_1: 0.6648, loss_cls_dn_2: 0.1058, loss_box_dn_2: 0.6573, loss_cls_dn_3: 0.1101, loss_box_dn_3: 0.6460, loss_cls_dn_4: 0.1116, loss_box_dn_4: 0.6472, loss_cls_dn_5: 0.1120, loss_box_dn_5: 0.6512, loss_dense_depth: 0.7839, loss: 24.3670, grad_norm: 29.5870 -2025-11-12 14:38:18,005 - mmdet - INFO - Iter [275/17500] lr: 2.095e-04, eta: 10:02:58, time: 1.573, data_time: 0.076, memory: 49164, loss_cls_0: 0.7315, loss_box_0: 1.6479, loss_cns_0: 0.6348, loss_yns_0: 0.1427, loss_cls_1: 0.8017, loss_box_1: 1.5304, loss_cns_1: 0.6556, loss_yns_1: 0.1387, loss_cls_2: 0.8151, loss_box_2: 1.4853, loss_cns_2: 0.6598, loss_yns_2: 0.1388, loss_cls_3: 0.8669, loss_box_3: 1.4762, loss_cns_3: 0.6568, loss_yns_3: 0.1390, loss_cls_4: 0.8722, loss_box_4: 1.4609, loss_cns_4: 0.6586, loss_yns_4: 0.1401, loss_cls_5: 0.8393, loss_box_5: 1.4633, loss_cns_5: 0.6611, loss_yns_5: 0.1391, loss_cls_dn_0: 0.1749, loss_box_dn_0: 0.7447, loss_cls_dn_1: 0.1032, loss_box_dn_1: 0.6510, loss_cls_dn_2: 0.1028, loss_box_dn_2: 0.6392, loss_cls_dn_3: 0.1093, loss_box_dn_3: 0.6447, loss_cls_dn_4: 0.1094, loss_box_dn_4: 0.6410, loss_cls_dn_5: 0.1121, loss_box_dn_5: 0.6435, loss_dense_depth: 0.7096, loss: 24.1413, grad_norm: 54.7748 -2025-11-12 14:38:19,585 - mmdet - INFO - Iter [276/17500] lr: 2.099e-04, eta: 10:02:24, time: 1.579, data_time: 0.080, memory: 49164, loss_cls_0: 0.7521, loss_box_0: 1.6369, loss_cns_0: 0.6339, loss_yns_0: 0.1427, loss_cls_1: 0.8188, loss_box_1: 1.5128, loss_cns_1: 0.6592, loss_yns_1: 0.1388, loss_cls_2: 0.8337, loss_box_2: 1.4830, loss_cns_2: 0.6641, loss_yns_2: 0.1395, loss_cls_3: 0.8699, loss_box_3: 1.4631, loss_cns_3: 0.6572, loss_yns_3: 0.1385, loss_cls_4: 0.8605, loss_box_4: 1.4656, loss_cns_4: 0.6589, loss_yns_4: 0.1397, loss_cls_5: 0.8563, loss_box_5: 1.4842, loss_cns_5: 0.6591, loss_yns_5: 0.1393, loss_cls_dn_0: 0.1763, loss_box_dn_0: 0.7450, loss_cls_dn_1: 0.1062, loss_box_dn_1: 0.6524, loss_cls_dn_2: 0.1065, loss_box_dn_2: 0.6474, loss_cls_dn_3: 0.1135, loss_box_dn_3: 0.6514, loss_cls_dn_4: 0.1145, loss_box_dn_4: 0.6527, loss_cls_dn_5: 0.1169, loss_box_dn_5: 0.6638, loss_dense_depth: 0.7764, loss: 24.3309, grad_norm: 45.8595 -2025-11-12 14:38:21,163 - mmdet - INFO - Iter [277/17500] lr: 2.103e-04, eta: 10:01:49, time: 1.579, data_time: 0.076, memory: 49164, loss_cls_0: 0.7720, loss_box_0: 1.6631, loss_cns_0: 0.6325, loss_yns_0: 0.1451, loss_cls_1: 0.8299, loss_box_1: 1.5610, loss_cns_1: 0.6549, loss_yns_1: 0.1442, loss_cls_2: 0.8552, loss_box_2: 1.5519, loss_cns_2: 0.6621, loss_yns_2: 0.1391, loss_cls_3: 0.8446, loss_box_3: 1.5359, loss_cns_3: 0.6558, loss_yns_3: 0.1398, loss_cls_4: 0.8588, loss_box_4: 1.5487, loss_cns_4: 0.6630, loss_yns_4: 0.1413, loss_cls_5: 0.8814, loss_box_5: 1.5442, loss_cns_5: 0.6607, loss_yns_5: 0.1435, loss_cls_dn_0: 0.1697, loss_box_dn_0: 0.7427, loss_cls_dn_1: 0.1063, loss_box_dn_1: 0.6761, loss_cls_dn_2: 0.1072, loss_box_dn_2: 0.6806, loss_cls_dn_3: 0.1093, loss_box_dn_3: 0.6774, loss_cls_dn_4: 0.1137, loss_box_dn_4: 0.6882, loss_cls_dn_5: 0.1127, loss_box_dn_5: 0.6982, loss_dense_depth: 0.7488, loss: 24.8595, grad_norm: 52.3108 -2025-11-12 14:38:22,744 - mmdet - INFO - Iter [278/17500] lr: 2.107e-04, eta: 10:01:15, time: 1.585, data_time: 0.077, memory: 49164, loss_cls_0: 0.7596, loss_box_0: 1.6427, loss_cns_0: 0.6337, loss_yns_0: 0.1457, loss_cls_1: 0.8251, loss_box_1: 1.5634, loss_cns_1: 0.6555, loss_yns_1: 0.1431, loss_cls_2: 0.8568, loss_box_2: 1.5411, loss_cns_2: 0.6588, loss_yns_2: 0.1405, loss_cls_3: 0.8484, loss_box_3: 1.5176, loss_cns_3: 0.6584, loss_yns_3: 0.1404, loss_cls_4: 0.8615, loss_box_4: 1.5245, loss_cns_4: 0.6621, loss_yns_4: 0.1431, loss_cls_5: 0.8633, loss_box_5: 1.5117, loss_cns_5: 0.6627, loss_yns_5: 0.1431, loss_cls_dn_0: 0.1667, loss_box_dn_0: 0.7329, loss_cls_dn_1: 0.1040, loss_box_dn_1: 0.6854, loss_cls_dn_2: 0.1028, loss_box_dn_2: 0.6809, loss_cls_dn_3: 0.1072, loss_box_dn_3: 0.6732, loss_cls_dn_4: 0.1104, loss_box_dn_4: 0.6835, loss_cls_dn_5: 0.1069, loss_box_dn_5: 0.6890, loss_dense_depth: 0.7266, loss: 24.6724, grad_norm: 40.9874 -2025-11-12 14:38:24,312 - mmdet - INFO - Iter [279/17500] lr: 2.111e-04, eta: 10:00:41, time: 1.567, data_time: 0.076, memory: 49164, loss_cls_0: 0.7608, loss_box_0: 1.6296, loss_cns_0: 0.6287, loss_yns_0: 0.1437, loss_cls_1: 0.8576, loss_box_1: 1.4906, loss_cns_1: 0.6578, loss_yns_1: 0.1421, loss_cls_2: 0.8706, loss_box_2: 1.4997, loss_cns_2: 0.6596, loss_yns_2: 0.1423, loss_cls_3: 0.8608, loss_box_3: 1.4874, loss_cns_3: 0.6624, loss_yns_3: 0.1423, loss_cls_4: 0.8589, loss_box_4: 1.4985, loss_cns_4: 0.6622, loss_yns_4: 0.1454, loss_cls_5: 0.8600, loss_box_5: 1.4934, loss_cns_5: 0.6607, loss_yns_5: 0.1429, loss_cls_dn_0: 0.1748, loss_box_dn_0: 0.7324, loss_cls_dn_1: 0.1084, loss_box_dn_1: 0.6775, loss_cls_dn_2: 0.1082, loss_box_dn_2: 0.6783, loss_cls_dn_3: 0.1128, loss_box_dn_3: 0.6789, loss_cls_dn_4: 0.1153, loss_box_dn_4: 0.6871, loss_cls_dn_5: 0.1165, loss_box_dn_5: 0.6908, loss_dense_depth: 0.7409, loss: 24.5796, grad_norm: 49.6662 -2025-11-12 14:38:25,908 - mmdet - INFO - Iter [280/17500] lr: 2.115e-04, eta: 10:00:07, time: 1.579, data_time: 0.078, memory: 49164, loss_cls_0: 0.7706, loss_box_0: 1.6342, loss_cns_0: 0.6295, loss_yns_0: 0.1423, loss_cls_1: 0.8572, loss_box_1: 1.5333, loss_cns_1: 0.6592, loss_yns_1: 0.1428, loss_cls_2: 0.8681, loss_box_2: 1.5238, loss_cns_2: 0.6613, loss_yns_2: 0.1432, loss_cls_3: 0.8560, loss_box_3: 1.5245, loss_cns_3: 0.6627, loss_yns_3: 0.1433, loss_cls_4: 0.8646, loss_box_4: 1.5206, loss_cns_4: 0.6629, loss_yns_4: 0.1416, loss_cls_5: 0.8688, loss_box_5: 1.5071, loss_cns_5: 0.6580, loss_yns_5: 0.1418, loss_cls_dn_0: 0.1733, loss_box_dn_0: 0.7321, loss_cls_dn_1: 0.1113, loss_box_dn_1: 0.6833, loss_cls_dn_2: 0.1106, loss_box_dn_2: 0.6807, loss_cls_dn_3: 0.1101, loss_box_dn_3: 0.6818, loss_cls_dn_4: 0.1140, loss_box_dn_4: 0.6846, loss_cls_dn_5: 0.1153, loss_box_dn_5: 0.6816, loss_dense_depth: 0.7396, loss: 24.7357, grad_norm: 46.8185 -2025-11-12 14:38:27,574 - mmdet - INFO - Iter [281/17500] lr: 2.119e-04, eta: 9:59:40, time: 1.686, data_time: 0.114, memory: 49164, loss_cls_0: 0.7581, loss_box_0: 1.6384, loss_cns_0: 0.6330, loss_yns_0: 0.1418, loss_cls_1: 0.8351, loss_box_1: 1.4926, loss_cns_1: 0.6594, loss_yns_1: 0.1428, loss_cls_2: 0.8468, loss_box_2: 1.4730, loss_cns_2: 0.6627, loss_yns_2: 0.1417, loss_cls_3: 0.8531, loss_box_3: 1.4656, loss_cns_3: 0.6604, loss_yns_3: 0.1407, loss_cls_4: 0.8559, loss_box_4: 1.4559, loss_cns_4: 0.6612, loss_yns_4: 0.1401, loss_cls_5: 0.8689, loss_box_5: 1.4673, loss_cns_5: 0.6602, loss_yns_5: 0.1404, loss_cls_dn_0: 0.1751, loss_box_dn_0: 0.7378, loss_cls_dn_1: 0.1083, loss_box_dn_1: 0.6725, loss_cls_dn_2: 0.1066, loss_box_dn_2: 0.6605, loss_cls_dn_3: 0.1063, loss_box_dn_3: 0.6586, loss_cls_dn_4: 0.1076, loss_box_dn_4: 0.6560, loss_cls_dn_5: 0.1073, loss_box_dn_5: 0.6607, loss_dense_depth: 0.7285, loss: 24.2808, grad_norm: 32.2522 -2025-11-12 14:38:29,202 - mmdet - INFO - Iter [282/17500] lr: 2.123e-04, eta: 9:59:10, time: 1.627, data_time: 0.106, memory: 49164, loss_cls_0: 0.7556, loss_box_0: 1.6178, loss_cns_0: 0.6361, loss_yns_0: 0.1470, loss_cls_1: 0.8454, loss_box_1: 1.4750, loss_cns_1: 0.6607, loss_yns_1: 0.1456, loss_cls_2: 0.8615, loss_box_2: 1.4624, loss_cns_2: 0.6628, loss_yns_2: 0.1455, loss_cls_3: 0.8569, loss_box_3: 1.4495, loss_cns_3: 0.6620, loss_yns_3: 0.1442, loss_cls_4: 0.8724, loss_box_4: 1.4462, loss_cns_4: 0.6652, loss_yns_4: 0.1451, loss_cls_5: 0.8763, loss_box_5: 1.4470, loss_cns_5: 0.6641, loss_yns_5: 0.1440, loss_cls_dn_0: 0.1751, loss_box_dn_0: 0.7411, loss_cls_dn_1: 0.1062, loss_box_dn_1: 0.6626, loss_cls_dn_2: 0.1047, loss_box_dn_2: 0.6541, loss_cls_dn_3: 0.1082, loss_box_dn_3: 0.6519, loss_cls_dn_4: 0.1116, loss_box_dn_4: 0.6534, loss_cls_dn_5: 0.1101, loss_box_dn_5: 0.6617, loss_dense_depth: 0.7187, loss: 24.2479, grad_norm: 34.0768 -2025-11-12 14:38:30,808 - mmdet - INFO - Iter [283/17500] lr: 2.127e-04, eta: 9:58:38, time: 1.605, data_time: 0.082, memory: 49164, loss_cls_0: 0.7644, loss_box_0: 1.6441, loss_cns_0: 0.6316, loss_yns_0: 0.1456, loss_cls_1: 0.8640, loss_box_1: 1.4997, loss_cns_1: 0.6573, loss_yns_1: 0.1445, loss_cls_2: 0.8547, loss_box_2: 1.4667, loss_cns_2: 0.6586, loss_yns_2: 0.1457, loss_cls_3: 0.8620, loss_box_3: 1.4647, loss_cns_3: 0.6607, loss_yns_3: 0.1447, loss_cls_4: 0.8666, loss_box_4: 1.4805, loss_cns_4: 0.6607, loss_yns_4: 0.1446, loss_cls_5: 0.8768, loss_box_5: 1.4684, loss_cns_5: 0.6621, loss_yns_5: 0.1444, loss_cls_dn_0: 0.1835, loss_box_dn_0: 0.7379, loss_cls_dn_1: 0.1080, loss_box_dn_1: 0.6758, loss_cls_dn_2: 0.1070, loss_box_dn_2: 0.6627, loss_cls_dn_3: 0.1113, loss_box_dn_3: 0.6676, loss_cls_dn_4: 0.1148, loss_box_dn_4: 0.6803, loss_cls_dn_5: 0.1107, loss_box_dn_5: 0.6863, loss_dense_depth: 0.7244, loss: 24.4833, grad_norm: 38.6491 -2025-11-12 14:38:32,410 - mmdet - INFO - Iter [284/17500] lr: 2.131e-04, eta: 9:58:07, time: 1.602, data_time: 0.077, memory: 49164, loss_cls_0: 0.7520, loss_box_0: 1.6196, loss_cns_0: 0.6338, loss_yns_0: 0.1485, loss_cls_1: 0.8395, loss_box_1: 1.5223, loss_cns_1: 0.6582, loss_yns_1: 0.1471, loss_cls_2: 0.8550, loss_box_2: 1.4878, loss_cns_2: 0.6591, loss_yns_2: 0.1472, loss_cls_3: 0.8758, loss_box_3: 1.4625, loss_cns_3: 0.6582, loss_yns_3: 0.1470, loss_cls_4: 0.8682, loss_box_4: 1.4610, loss_cns_4: 0.6637, loss_yns_4: 0.1464, loss_cls_5: 0.8910, loss_box_5: 1.4605, loss_cns_5: 0.6595, loss_yns_5: 0.1459, loss_cls_dn_0: 0.1794, loss_box_dn_0: 0.7260, loss_cls_dn_1: 0.1064, loss_box_dn_1: 0.6890, loss_cls_dn_2: 0.1063, loss_box_dn_2: 0.6763, loss_cls_dn_3: 0.1091, loss_box_dn_3: 0.6765, loss_cls_dn_4: 0.1091, loss_box_dn_4: 0.6836, loss_cls_dn_5: 0.1095, loss_box_dn_5: 0.6899, loss_dense_depth: 0.7032, loss: 24.4738, grad_norm: 35.3757 -2025-11-12 14:38:33,991 - mmdet - INFO - Iter [285/17500] lr: 2.135e-04, eta: 9:57:34, time: 1.582, data_time: 0.080, memory: 49164, loss_cls_0: 0.7620, loss_box_0: 1.6122, loss_cns_0: 0.6273, loss_yns_0: 0.1493, loss_cls_1: 0.8483, loss_box_1: 1.4810, loss_cns_1: 0.6512, loss_yns_1: 0.1475, loss_cls_2: 0.8692, loss_box_2: 1.4859, loss_cns_2: 0.6503, loss_yns_2: 0.1460, loss_cls_3: 0.8981, loss_box_3: 1.4444, loss_cns_3: 0.6433, loss_yns_3: 0.1437, loss_cls_4: 0.8758, loss_box_4: 1.4690, loss_cns_4: 0.6579, loss_yns_4: 0.1488, loss_cls_5: 0.8955, loss_box_5: 1.4650, loss_cns_5: 0.6533, loss_yns_5: 0.1469, loss_cls_dn_0: 0.1762, loss_box_dn_0: 0.7389, loss_cls_dn_1: 0.1078, loss_box_dn_1: 0.6705, loss_cls_dn_2: 0.1073, loss_box_dn_2: 0.6705, loss_cls_dn_3: 0.1108, loss_box_dn_3: 0.6646, loss_cls_dn_4: 0.1113, loss_box_dn_4: 0.6669, loss_cls_dn_5: 0.1159, loss_box_dn_5: 0.6732, loss_dense_depth: 0.7194, loss: 24.4053, grad_norm: 44.5897 -2025-11-12 14:38:35,588 - mmdet - INFO - Iter [286/17500] lr: 2.139e-04, eta: 9:57:03, time: 1.592, data_time: 0.102, memory: 49164, loss_cls_0: 0.7471, loss_box_0: 1.5969, loss_cns_0: 0.6242, loss_yns_0: 0.1466, loss_cls_1: 0.8196, loss_box_1: 1.4800, loss_cns_1: 0.6471, loss_yns_1: 0.1473, loss_cls_2: 0.8350, loss_box_2: 1.4372, loss_cns_2: 0.6486, loss_yns_2: 0.1463, loss_cls_3: 0.8488, loss_box_3: 1.4223, loss_cns_3: 0.6490, loss_yns_3: 0.1446, loss_cls_4: 0.8469, loss_box_4: 1.4316, loss_cns_4: 0.6530, loss_yns_4: 0.1456, loss_cls_5: 0.8629, loss_box_5: 1.4339, loss_cns_5: 0.6528, loss_yns_5: 0.1473, loss_cls_dn_0: 0.1781, loss_box_dn_0: 0.7335, loss_cls_dn_1: 0.1084, loss_box_dn_1: 0.6673, loss_cls_dn_2: 0.1094, loss_box_dn_2: 0.6537, loss_cls_dn_3: 0.1105, loss_box_dn_3: 0.6506, loss_cls_dn_4: 0.1110, loss_box_dn_4: 0.6506, loss_cls_dn_5: 0.1141, loss_box_dn_5: 0.6575, loss_dense_depth: 0.7156, loss: 23.9748, grad_norm: 29.5253 -2025-11-12 14:38:37,159 - mmdet - INFO - Iter [287/17500] lr: 2.143e-04, eta: 9:56:30, time: 1.572, data_time: 0.077, memory: 49164, loss_cls_0: 0.7726, loss_box_0: 1.6104, loss_cns_0: 0.6274, loss_yns_0: 0.1496, loss_cls_1: 0.8393, loss_box_1: 1.5321, loss_cns_1: 0.6529, loss_yns_1: 0.1490, loss_cls_2: 0.8568, loss_box_2: 1.5524, loss_cns_2: 0.6550, loss_yns_2: 0.1497, loss_cls_3: 0.8779, loss_box_3: 1.5317, loss_cns_3: 0.6601, loss_yns_3: 0.1471, loss_cls_4: 0.8793, loss_box_4: 1.5060, loss_cns_4: 0.6604, loss_yns_4: 0.1475, loss_cls_5: 0.8801, loss_box_5: 1.5039, loss_cns_5: 0.6585, loss_yns_5: 0.1489, loss_cls_dn_0: 0.1797, loss_box_dn_0: 0.7474, loss_cls_dn_1: 0.1099, loss_box_dn_1: 0.6544, loss_cls_dn_2: 0.1102, loss_box_dn_2: 0.6512, loss_cls_dn_3: 0.1109, loss_box_dn_3: 0.6501, loss_cls_dn_4: 0.1121, loss_box_dn_4: 0.6427, loss_cls_dn_5: 0.1133, loss_box_dn_5: 0.6472, loss_dense_depth: 0.7434, loss: 24.6211, grad_norm: 42.3439 -2025-11-12 14:38:38,737 - mmdet - INFO - Iter [288/17500] lr: 2.147e-04, eta: 9:55:58, time: 1.582, data_time: 0.085, memory: 49164, loss_cls_0: 0.7645, loss_box_0: 1.6071, loss_cns_0: 0.6314, loss_yns_0: 0.1492, loss_cls_1: 0.8458, loss_box_1: 1.5247, loss_cns_1: 0.6622, loss_yns_1: 0.1492, loss_cls_2: 0.8627, loss_box_2: 1.5750, loss_cns_2: 0.6597, loss_yns_2: 0.1499, loss_cls_3: 0.8868, loss_box_3: 1.5423, loss_cns_3: 0.6600, loss_yns_3: 0.1497, loss_cls_4: 0.8858, loss_box_4: 1.5430, loss_cns_4: 0.6624, loss_yns_4: 0.1489, loss_cls_5: 0.8738, loss_box_5: 1.5284, loss_cns_5: 0.6624, loss_yns_5: 0.1488, loss_cls_dn_0: 0.1650, loss_box_dn_0: 0.7442, loss_cls_dn_1: 0.1048, loss_box_dn_1: 0.6537, loss_cls_dn_2: 0.1037, loss_box_dn_2: 0.6564, loss_cls_dn_3: 0.1076, loss_box_dn_3: 0.6534, loss_cls_dn_4: 0.1082, loss_box_dn_4: 0.6576, loss_cls_dn_5: 0.1115, loss_box_dn_5: 0.6612, loss_dense_depth: 0.7503, loss: 24.7513, grad_norm: 42.5648 -2025-11-12 14:38:40,352 - mmdet - INFO - Iter [289/17500] lr: 2.151e-04, eta: 9:55:28, time: 1.607, data_time: 0.079, memory: 49164, loss_cls_0: 0.7406, loss_box_0: 1.6316, loss_cns_0: 0.6257, loss_yns_0: 0.1476, loss_cls_1: 0.8390, loss_box_1: 1.5406, loss_cns_1: 0.6624, loss_yns_1: 0.1475, loss_cls_2: 0.8440, loss_box_2: 1.5022, loss_cns_2: 0.6685, loss_yns_2: 0.1488, loss_cls_3: 0.8557, loss_box_3: 1.5035, loss_cns_3: 0.6637, loss_yns_3: 0.1479, loss_cls_4: 0.8480, loss_box_4: 1.5186, loss_cns_4: 0.6601, loss_yns_4: 0.1470, loss_cls_5: 0.8557, loss_box_5: 1.5023, loss_cns_5: 0.6620, loss_yns_5: 0.1475, loss_cls_dn_0: 0.1625, loss_box_dn_0: 0.7431, loss_cls_dn_1: 0.1038, loss_box_dn_1: 0.6695, loss_cls_dn_2: 0.1031, loss_box_dn_2: 0.6592, loss_cls_dn_3: 0.1061, loss_box_dn_3: 0.6662, loss_cls_dn_4: 0.1068, loss_box_dn_4: 0.6804, loss_cls_dn_5: 0.1137, loss_box_dn_5: 0.6743, loss_dense_depth: 0.6885, loss: 24.4875, grad_norm: 43.2088 -2025-11-12 14:38:41,937 - mmdet - INFO - Iter [290/17500] lr: 2.155e-04, eta: 9:54:57, time: 1.587, data_time: 0.085, memory: 49164, loss_cls_0: 0.7929, loss_box_0: 1.6306, loss_cns_0: 0.6165, loss_yns_0: 0.1458, loss_cls_1: 0.8570, loss_box_1: 1.5452, loss_cns_1: 0.6557, loss_yns_1: 0.1473, loss_cls_2: 0.8665, loss_box_2: 1.4996, loss_cns_2: 0.6606, loss_yns_2: 0.1472, loss_cls_3: 0.8711, loss_box_3: 1.5091, loss_cns_3: 0.6634, loss_yns_3: 0.1467, loss_cls_4: 0.8753, loss_box_4: 1.4988, loss_cns_4: 0.6563, loss_yns_4: 0.1462, loss_cls_5: 0.8827, loss_box_5: 1.4945, loss_cns_5: 0.6581, loss_yns_5: 0.1464, loss_cls_dn_0: 0.1741, loss_box_dn_0: 0.7378, loss_cls_dn_1: 0.1083, loss_box_dn_1: 0.6773, loss_cls_dn_2: 0.1080, loss_box_dn_2: 0.6687, loss_cls_dn_3: 0.1091, loss_box_dn_3: 0.6760, loss_cls_dn_4: 0.1091, loss_box_dn_4: 0.6845, loss_cls_dn_5: 0.1151, loss_box_dn_5: 0.6823, loss_dense_depth: 0.7748, loss: 24.7387, grad_norm: 47.2117 -2025-11-12 14:38:43,536 - mmdet - INFO - Iter [291/17500] lr: 2.159e-04, eta: 9:54:27, time: 1.604, data_time: 0.082, memory: 49164, loss_cls_0: 0.8024, loss_box_0: 1.6323, loss_cns_0: 0.6255, loss_yns_0: 0.1507, loss_cls_1: 0.8599, loss_box_1: 1.4976, loss_cns_1: 0.6607, loss_yns_1: 0.1519, loss_cls_2: 0.8800, loss_box_2: 1.4939, loss_cns_2: 0.6589, loss_yns_2: 0.1518, loss_cls_3: 0.8799, loss_box_3: 1.4683, loss_cns_3: 0.6629, loss_yns_3: 0.1496, loss_cls_4: 0.8758, loss_box_4: 1.4638, loss_cns_4: 0.6578, loss_yns_4: 0.1479, loss_cls_5: 0.8800, loss_box_5: 1.4673, loss_cns_5: 0.6606, loss_yns_5: 0.1498, loss_cls_dn_0: 0.1650, loss_box_dn_0: 0.7411, loss_cls_dn_1: 0.1043, loss_box_dn_1: 0.6611, loss_cls_dn_2: 0.1053, loss_box_dn_2: 0.6521, loss_cls_dn_3: 0.1055, loss_box_dn_3: 0.6521, loss_cls_dn_4: 0.1066, loss_box_dn_4: 0.6551, loss_cls_dn_5: 0.1094, loss_box_dn_5: 0.6568, loss_dense_depth: 0.7697, loss: 24.5134, grad_norm: 28.9920 -2025-11-12 14:38:45,117 - mmdet - INFO - Iter [292/17500] lr: 2.163e-04, eta: 9:53:56, time: 1.580, data_time: 0.077, memory: 49164, loss_cls_0: 0.7442, loss_box_0: 1.6597, loss_cns_0: 0.6319, loss_yns_0: 0.1483, loss_cls_1: 0.8260, loss_box_1: 1.5238, loss_cns_1: 0.6606, loss_yns_1: 0.1481, loss_cls_2: 0.8362, loss_box_2: 1.5189, loss_cns_2: 0.6630, loss_yns_2: 0.1490, loss_cls_3: 0.8425, loss_box_3: 1.5035, loss_cns_3: 0.6599, loss_yns_3: 0.1475, loss_cls_4: 0.8409, loss_box_4: 1.4833, loss_cns_4: 0.6602, loss_yns_4: 0.1463, loss_cls_5: 0.8400, loss_box_5: 1.4751, loss_cns_5: 0.6609, loss_yns_5: 0.1468, loss_cls_dn_0: 0.1617, loss_box_dn_0: 0.7440, loss_cls_dn_1: 0.1078, loss_box_dn_1: 0.6526, loss_cls_dn_2: 0.1054, loss_box_dn_2: 0.6394, loss_cls_dn_3: 0.1076, loss_box_dn_3: 0.6417, loss_cls_dn_4: 0.1141, loss_box_dn_4: 0.6341, loss_cls_dn_5: 0.1128, loss_box_dn_5: 0.6330, loss_dense_depth: 0.7542, loss: 24.3251, grad_norm: 37.8853 -2025-11-12 14:38:46,695 - mmdet - INFO - Iter [293/17500] lr: 2.167e-04, eta: 9:53:25, time: 1.578, data_time: 0.076, memory: 49164, loss_cls_0: 0.7729, loss_box_0: 1.6781, loss_cns_0: 0.6327, loss_yns_0: 0.1465, loss_cls_1: 0.8482, loss_box_1: 1.4853, loss_cns_1: 0.6630, loss_yns_1: 0.1433, loss_cls_2: 0.8516, loss_box_2: 1.4647, loss_cns_2: 0.6653, loss_yns_2: 0.1436, loss_cls_3: 0.8572, loss_box_3: 1.4628, loss_cns_3: 0.6618, loss_yns_3: 0.1444, loss_cls_4: 0.8551, loss_box_4: 1.4599, loss_cns_4: 0.6623, loss_yns_4: 0.1439, loss_cls_5: 0.8619, loss_box_5: 1.4539, loss_cns_5: 0.6620, loss_yns_5: 0.1439, loss_cls_dn_0: 0.1796, loss_box_dn_0: 0.7347, loss_cls_dn_1: 0.1080, loss_box_dn_1: 0.6520, loss_cls_dn_2: 0.1058, loss_box_dn_2: 0.6429, loss_cls_dn_3: 0.1085, loss_box_dn_3: 0.6435, loss_cls_dn_4: 0.1116, loss_box_dn_4: 0.6430, loss_cls_dn_5: 0.1130, loss_box_dn_5: 0.6481, loss_dense_depth: 0.7402, loss: 24.2951, grad_norm: 28.8434 -2025-11-12 14:38:48,313 - mmdet - INFO - Iter [294/17500] lr: 2.170e-04, eta: 9:52:57, time: 1.621, data_time: 0.077, memory: 49164, loss_cls_0: 0.7653, loss_box_0: 1.6567, loss_cns_0: 0.6342, loss_yns_0: 0.1477, loss_cls_1: 0.8436, loss_box_1: 1.5065, loss_cns_1: 0.6609, loss_yns_1: 0.1424, loss_cls_2: 0.8495, loss_box_2: 1.4817, loss_cns_2: 0.6629, loss_yns_2: 0.1442, loss_cls_3: 0.8634, loss_box_3: 1.4568, loss_cns_3: 0.6619, loss_yns_3: 0.1436, loss_cls_4: 0.8743, loss_box_4: 1.4602, loss_cns_4: 0.6614, loss_yns_4: 0.1434, loss_cls_5: 0.8725, loss_box_5: 1.4591, loss_cns_5: 0.6622, loss_yns_5: 0.1432, loss_cls_dn_0: 0.1670, loss_box_dn_0: 0.7283, loss_cls_dn_1: 0.1064, loss_box_dn_1: 0.6548, loss_cls_dn_2: 0.1052, loss_box_dn_2: 0.6434, loss_cls_dn_3: 0.1071, loss_box_dn_3: 0.6385, loss_cls_dn_4: 0.1126, loss_box_dn_4: 0.6479, loss_cls_dn_5: 0.1122, loss_box_dn_5: 0.6530, loss_dense_depth: 0.7370, loss: 24.3108, grad_norm: 42.8229 -2025-11-12 14:38:49,887 - mmdet - INFO - Iter [295/17500] lr: 2.174e-04, eta: 9:52:26, time: 1.566, data_time: 0.072, memory: 49164, loss_cls_0: 0.7634, loss_box_0: 1.6637, loss_cns_0: 0.6352, loss_yns_0: 0.1495, loss_cls_1: 0.8680, loss_box_1: 1.5160, loss_cns_1: 0.6585, loss_yns_1: 0.1468, loss_cls_2: 0.8377, loss_box_2: 1.4884, loss_cns_2: 0.6605, loss_yns_2: 0.1493, loss_cls_3: 0.8472, loss_box_3: 1.4805, loss_cns_3: 0.6585, loss_yns_3: 0.1476, loss_cls_4: 0.8646, loss_box_4: 1.4830, loss_cns_4: 0.6577, loss_yns_4: 0.1463, loss_cls_5: 0.8615, loss_box_5: 1.4765, loss_cns_5: 0.6585, loss_yns_5: 0.1473, loss_cls_dn_0: 0.1617, loss_box_dn_0: 0.7319, loss_cls_dn_1: 0.1105, loss_box_dn_1: 0.6722, loss_cls_dn_2: 0.1078, loss_box_dn_2: 0.6538, loss_cls_dn_3: 0.1076, loss_box_dn_3: 0.6600, loss_cls_dn_4: 0.1206, loss_box_dn_4: 0.6709, loss_cls_dn_5: 0.1151, loss_box_dn_5: 0.6790, loss_dense_depth: 0.7336, loss: 24.4908, grad_norm: 37.4909 -2025-11-12 14:38:51,451 - mmdet - INFO - Iter [296/17500] lr: 2.178e-04, eta: 9:51:55, time: 1.572, data_time: 0.072, memory: 49164, loss_cls_0: 0.7651, loss_box_0: 1.6464, loss_cns_0: 0.6321, loss_yns_0: 0.1498, loss_cls_1: 0.8471, loss_box_1: 1.5203, loss_cns_1: 0.6560, loss_yns_1: 0.1484, loss_cls_2: 0.8341, loss_box_2: 1.5072, loss_cns_2: 0.6598, loss_yns_2: 0.1489, loss_cls_3: 0.8482, loss_box_3: 1.4962, loss_cns_3: 0.6609, loss_yns_3: 0.1482, loss_cls_4: 0.8536, loss_box_4: 1.4932, loss_cns_4: 0.6603, loss_yns_4: 0.1478, loss_cls_5: 0.8629, loss_box_5: 1.4846, loss_cns_5: 0.6598, loss_yns_5: 0.1483, loss_cls_dn_0: 0.1644, loss_box_dn_0: 0.7251, loss_cls_dn_1: 0.1064, loss_box_dn_1: 0.6915, loss_cls_dn_2: 0.1072, loss_box_dn_2: 0.6802, loss_cls_dn_3: 0.1056, loss_box_dn_3: 0.6841, loss_cls_dn_4: 0.1119, loss_box_dn_4: 0.6962, loss_cls_dn_5: 0.1099, loss_box_dn_5: 0.7022, loss_dense_depth: 0.7230, loss: 24.5869, grad_norm: 41.6543 -2025-11-12 14:38:53,034 - mmdet - INFO - Iter [297/17500] lr: 2.182e-04, eta: 9:51:25, time: 1.582, data_time: 0.069, memory: 49164, loss_cls_0: 0.7853, loss_box_0: 1.6843, loss_cns_0: 0.6290, loss_yns_0: 0.1495, loss_cls_1: 0.8624, loss_box_1: 1.5550, loss_cns_1: 0.6610, loss_yns_1: 0.1486, loss_cls_2: 0.8634, loss_box_2: 1.5562, loss_cns_2: 0.6644, loss_yns_2: 0.1497, loss_cls_3: 0.8943, loss_box_3: 1.5337, loss_cns_3: 0.6633, loss_yns_3: 0.1495, loss_cls_4: 0.8871, loss_box_4: 1.5323, loss_cns_4: 0.6632, loss_yns_4: 0.1504, loss_cls_5: 0.8871, loss_box_5: 1.5268, loss_cns_5: 0.6610, loss_yns_5: 0.1498, loss_cls_dn_0: 0.1857, loss_box_dn_0: 0.7363, loss_cls_dn_1: 0.1058, loss_box_dn_1: 0.6937, loss_cls_dn_2: 0.1076, loss_box_dn_2: 0.6873, loss_cls_dn_3: 0.1097, loss_box_dn_3: 0.6804, loss_cls_dn_4: 0.1139, loss_box_dn_4: 0.6897, loss_cls_dn_5: 0.1157, loss_box_dn_5: 0.6910, loss_dense_depth: 0.7436, loss: 25.0678, grad_norm: 44.8766 -2025-11-12 14:38:54,596 - mmdet - INFO - Iter [298/17500] lr: 2.186e-04, eta: 9:50:54, time: 1.564, data_time: 0.070, memory: 49164, loss_cls_0: 0.7714, loss_box_0: 1.6616, loss_cns_0: 0.6329, loss_yns_0: 0.1483, loss_cls_1: 0.8571, loss_box_1: 1.5384, loss_cns_1: 0.6619, loss_yns_1: 0.1498, loss_cls_2: 0.8504, loss_box_2: 1.5231, loss_cns_2: 0.6643, loss_yns_2: 0.1484, loss_cls_3: 0.8648, loss_box_3: 1.5143, loss_cns_3: 0.6611, loss_yns_3: 0.1485, loss_cls_4: 0.8776, loss_box_4: 1.4988, loss_cns_4: 0.6622, loss_yns_4: 0.1483, loss_cls_5: 0.8672, loss_box_5: 1.4955, loss_cns_5: 0.6617, loss_yns_5: 0.1480, loss_cls_dn_0: 0.1847, loss_box_dn_0: 0.7305, loss_cls_dn_1: 0.1084, loss_box_dn_1: 0.6724, loss_cls_dn_2: 0.1083, loss_box_dn_2: 0.6588, loss_cls_dn_3: 0.1097, loss_box_dn_3: 0.6580, loss_cls_dn_4: 0.1132, loss_box_dn_4: 0.6545, loss_cls_dn_5: 0.1153, loss_box_dn_5: 0.6558, loss_dense_depth: 0.7218, loss: 24.6467, grad_norm: 27.6917 -2025-11-12 14:38:56,153 - mmdet - INFO - Iter [299/17500] lr: 2.190e-04, eta: 9:50:23, time: 1.557, data_time: 0.075, memory: 49164, loss_cls_0: 0.7776, loss_box_0: 1.6573, loss_cns_0: 0.6267, loss_yns_0: 0.1505, loss_cls_1: 0.8632, loss_box_1: 1.5495, loss_cns_1: 0.6584, loss_yns_1: 0.1506, loss_cls_2: 0.8728, loss_box_2: 1.5126, loss_cns_2: 0.6546, loss_yns_2: 0.1490, loss_cls_3: 0.8880, loss_box_3: 1.5182, loss_cns_3: 0.6553, loss_yns_3: 0.1490, loss_cls_4: 0.8859, loss_box_4: 1.5206, loss_cns_4: 0.6609, loss_yns_4: 0.1509, loss_cls_5: 0.9005, loss_box_5: 1.5128, loss_cns_5: 0.6590, loss_yns_5: 0.1496, loss_cls_dn_0: 0.1581, loss_box_dn_0: 0.7350, loss_cls_dn_1: 0.1066, loss_box_dn_1: 0.6681, loss_cls_dn_2: 0.1036, loss_box_dn_2: 0.6527, loss_cls_dn_3: 0.1051, loss_box_dn_3: 0.6623, loss_cls_dn_4: 0.1063, loss_box_dn_4: 0.6566, loss_cls_dn_5: 0.1086, loss_box_dn_5: 0.6590, loss_dense_depth: 0.8050, loss: 24.8006, grad_norm: 39.3368 -2025-11-12 14:38:57,703 - mmdet - INFO - Iter [300/17500] lr: 2.194e-04, eta: 9:49:52, time: 1.550, data_time: 0.069, memory: 49164, loss_cls_0: 0.7745, loss_box_0: 1.6308, loss_cns_0: 0.6306, loss_yns_0: 0.1496, loss_cls_1: 0.8380, loss_box_1: 1.5625, loss_cns_1: 0.6543, loss_yns_1: 0.1473, loss_cls_2: 0.8579, loss_box_2: 1.5242, loss_cns_2: 0.6497, loss_yns_2: 0.1454, loss_cls_3: 0.8840, loss_box_3: 1.5244, loss_cns_3: 0.6582, loss_yns_3: 0.1464, loss_cls_4: 0.8779, loss_box_4: 1.5389, loss_cns_4: 0.6598, loss_yns_4: 0.1493, loss_cls_5: 0.9019, loss_box_5: 1.5257, loss_cns_5: 0.6556, loss_yns_5: 0.1474, loss_cls_dn_0: 0.1668, loss_box_dn_0: 0.7292, loss_cls_dn_1: 0.1062, loss_box_dn_1: 0.6623, loss_cls_dn_2: 0.1070, loss_box_dn_2: 0.6519, loss_cls_dn_3: 0.1100, loss_box_dn_3: 0.6570, loss_cls_dn_4: 0.1108, loss_box_dn_4: 0.6591, loss_cls_dn_5: 0.1151, loss_box_dn_5: 0.6604, loss_dense_depth: 0.7443, loss: 24.7142, grad_norm: 35.4407 -2025-11-12 14:38:59,341 - mmdet - INFO - Iter [301/17500] lr: 2.198e-04, eta: 9:49:26, time: 1.638, data_time: 0.097, memory: 49164, loss_cls_0: 0.8031, loss_box_0: 1.6559, loss_cns_0: 0.6346, loss_yns_0: 0.1517, loss_cls_1: 0.8364, loss_box_1: 1.5965, loss_cns_1: 0.6548, loss_yns_1: 0.1478, loss_cls_2: 0.8441, loss_box_2: 1.5577, loss_cns_2: 0.6615, loss_yns_2: 0.1468, loss_cls_3: 0.8668, loss_box_3: 1.5561, loss_cns_3: 0.6608, loss_yns_3: 0.1482, loss_cls_4: 0.8571, loss_box_4: 1.5908, loss_cns_4: 0.6592, loss_yns_4: 0.1484, loss_cls_5: 0.8672, loss_box_5: 1.5875, loss_cns_5: 0.6589, loss_yns_5: 0.1477, loss_cls_dn_0: 0.1791, loss_box_dn_0: 0.7340, loss_cls_dn_1: 0.1059, loss_box_dn_1: 0.6721, loss_cls_dn_2: 0.1065, loss_box_dn_2: 0.6625, loss_cls_dn_3: 0.1065, loss_box_dn_3: 0.6643, loss_cls_dn_4: 0.1102, loss_box_dn_4: 0.6753, loss_cls_dn_5: 0.1110, loss_box_dn_5: 0.6846, loss_dense_depth: 0.7902, loss: 25.0420, grad_norm: 41.0883 -2025-11-12 14:39:00,948 - mmdet - INFO - Iter [302/17500] lr: 2.202e-04, eta: 9:48:57, time: 1.599, data_time: 0.102, memory: 49164, loss_cls_0: 0.7864, loss_box_0: 1.6448, loss_cns_0: 0.6316, loss_yns_0: 0.1531, loss_cls_1: 0.8263, loss_box_1: 1.6146, loss_cns_1: 0.6534, loss_yns_1: 0.1502, loss_cls_2: 0.8422, loss_box_2: 1.5666, loss_cns_2: 0.6619, loss_yns_2: 0.1514, loss_cls_3: 0.8624, loss_box_3: 1.5599, loss_cns_3: 0.6573, loss_yns_3: 0.1508, loss_cls_4: 0.8640, loss_box_4: 1.5760, loss_cns_4: 0.6553, loss_yns_4: 0.1507, loss_cls_5: 0.8866, loss_box_5: 1.5687, loss_cns_5: 0.6566, loss_yns_5: 0.1529, loss_cls_dn_0: 0.1725, loss_box_dn_0: 0.7256, loss_cls_dn_1: 0.1059, loss_box_dn_1: 0.6826, loss_cls_dn_2: 0.1049, loss_box_dn_2: 0.6675, loss_cls_dn_3: 0.1042, loss_box_dn_3: 0.6728, loss_cls_dn_4: 0.1126, loss_box_dn_4: 0.6812, loss_cls_dn_5: 0.1108, loss_box_dn_5: 0.6893, loss_dense_depth: 0.7263, loss: 24.9802, grad_norm: 41.4251 -2025-11-12 14:39:02,546 - mmdet - INFO - Iter [303/17500] lr: 2.206e-04, eta: 9:48:30, time: 1.602, data_time: 0.086, memory: 49164, loss_cls_0: 0.7810, loss_box_0: 1.6656, loss_cns_0: 0.6257, loss_yns_0: 0.1513, loss_cls_1: 0.8370, loss_box_1: 1.5596, loss_cns_1: 0.6550, loss_yns_1: 0.1528, loss_cls_2: 0.8505, loss_box_2: 1.5351, loss_cns_2: 0.6585, loss_yns_2: 0.1531, loss_cls_3: 0.8674, loss_box_3: 1.5269, loss_cns_3: 0.6598, loss_yns_3: 0.1525, loss_cls_4: 0.8684, loss_box_4: 1.5361, loss_cns_4: 0.6621, loss_yns_4: 0.1531, loss_cls_5: 0.8984, loss_box_5: 1.5489, loss_cns_5: 0.6604, loss_yns_5: 0.1540, loss_cls_dn_0: 0.1590, loss_box_dn_0: 0.7347, loss_cls_dn_1: 0.1045, loss_box_dn_1: 0.6906, loss_cls_dn_2: 0.1024, loss_box_dn_2: 0.6851, loss_cls_dn_3: 0.1034, loss_box_dn_3: 0.6881, loss_cls_dn_4: 0.1057, loss_box_dn_4: 0.6958, loss_cls_dn_5: 0.1064, loss_box_dn_5: 0.7064, loss_dense_depth: 0.7606, loss: 24.9558, grad_norm: 36.8522 -2025-11-12 14:39:04,123 - mmdet - INFO - Iter [304/17500] lr: 2.210e-04, eta: 9:48:01, time: 1.575, data_time: 0.075, memory: 49164, loss_cls_0: 0.7927, loss_box_0: 1.6462, loss_cns_0: 0.6237, loss_yns_0: 0.1509, loss_cls_1: 0.8413, loss_box_1: 1.5177, loss_cns_1: 0.6603, loss_yns_1: 0.1516, loss_cls_2: 0.8463, loss_box_2: 1.5048, loss_cns_2: 0.6605, loss_yns_2: 0.1508, loss_cls_3: 0.8552, loss_box_3: 1.4972, loss_cns_3: 0.6687, loss_yns_3: 0.1503, loss_cls_4: 0.8533, loss_box_4: 1.4953, loss_cns_4: 0.6649, loss_yns_4: 0.1493, loss_cls_5: 0.8735, loss_box_5: 1.5067, loss_cns_5: 0.6650, loss_yns_5: 0.1493, loss_cls_dn_0: 0.1575, loss_box_dn_0: 0.7252, loss_cls_dn_1: 0.1056, loss_box_dn_1: 0.6708, loss_cls_dn_2: 0.1048, loss_box_dn_2: 0.6599, loss_cls_dn_3: 0.1045, loss_box_dn_3: 0.6578, loss_cls_dn_4: 0.1055, loss_box_dn_4: 0.6608, loss_cls_dn_5: 0.1067, loss_box_dn_5: 0.6686, loss_dense_depth: 0.7995, loss: 24.6031, grad_norm: 32.8787 -2025-11-12 14:39:05,703 - mmdet - INFO - Iter [305/17500] lr: 2.214e-04, eta: 9:47:32, time: 1.583, data_time: 0.083, memory: 49164, loss_cls_0: 0.7835, loss_box_0: 1.6401, loss_cns_0: 0.6250, loss_yns_0: 0.1497, loss_cls_1: 0.8247, loss_box_1: 1.5101, loss_cns_1: 0.6582, loss_yns_1: 0.1516, loss_cls_2: 0.8343, loss_box_2: 1.4928, loss_cns_2: 0.6621, loss_yns_2: 0.1517, loss_cls_3: 0.8485, loss_box_3: 1.4885, loss_cns_3: 0.6636, loss_yns_3: 0.1516, loss_cls_4: 0.8533, loss_box_4: 1.4746, loss_cns_4: 0.6581, loss_yns_4: 0.1523, loss_cls_5: 0.8683, loss_box_5: 1.4822, loss_cns_5: 0.6621, loss_yns_5: 0.1511, loss_cls_dn_0: 0.1664, loss_box_dn_0: 0.7364, loss_cls_dn_1: 0.1055, loss_box_dn_1: 0.6842, loss_cls_dn_2: 0.1071, loss_box_dn_2: 0.6664, loss_cls_dn_3: 0.1061, loss_box_dn_3: 0.6648, loss_cls_dn_4: 0.1085, loss_box_dn_4: 0.6637, loss_cls_dn_5: 0.1089, loss_box_dn_5: 0.6716, loss_dense_depth: 0.7514, loss: 24.4787, grad_norm: 36.2889 -2025-11-12 14:39:07,318 - mmdet - INFO - Iter [306/17500] lr: 2.218e-04, eta: 9:47:06, time: 1.618, data_time: 0.102, memory: 49164, loss_cls_0: 0.7747, loss_box_0: 1.6369, loss_cns_0: 0.6318, loss_yns_0: 0.1506, loss_cls_1: 0.8316, loss_box_1: 1.5102, loss_cns_1: 0.6597, loss_yns_1: 0.1491, loss_cls_2: 0.8415, loss_box_2: 1.5007, loss_cns_2: 0.6637, loss_yns_2: 0.1497, loss_cls_3: 0.8557, loss_box_3: 1.4778, loss_cns_3: 0.6660, loss_yns_3: 0.1494, loss_cls_4: 0.8545, loss_box_4: 1.4719, loss_cns_4: 0.6639, loss_yns_4: 0.1493, loss_cls_5: 0.8646, loss_box_5: 1.4866, loss_cns_5: 0.6645, loss_yns_5: 0.1482, loss_cls_dn_0: 0.1758, loss_box_dn_0: 0.7348, loss_cls_dn_1: 0.1065, loss_box_dn_1: 0.6736, loss_cls_dn_2: 0.1079, loss_box_dn_2: 0.6630, loss_cls_dn_3: 0.1097, loss_box_dn_3: 0.6652, loss_cls_dn_4: 0.1113, loss_box_dn_4: 0.6661, loss_cls_dn_5: 0.1119, loss_box_dn_5: 0.6790, loss_dense_depth: 0.7607, loss: 24.5179, grad_norm: 33.2419 -2025-11-12 14:39:08,908 - mmdet - INFO - Iter [307/17500] lr: 2.222e-04, eta: 9:46:38, time: 1.589, data_time: 0.069, memory: 49164, loss_cls_0: 0.7622, loss_box_0: 1.6399, loss_cns_0: 0.6346, loss_yns_0: 0.1503, loss_cls_1: 0.8329, loss_box_1: 1.5565, loss_cns_1: 0.6581, loss_yns_1: 0.1488, loss_cls_2: 0.8426, loss_box_2: 1.5326, loss_cns_2: 0.6612, loss_yns_2: 0.1504, loss_cls_3: 0.8517, loss_box_3: 1.5063, loss_cns_3: 0.6638, loss_yns_3: 0.1512, loss_cls_4: 0.8505, loss_box_4: 1.4927, loss_cns_4: 0.6666, loss_yns_4: 0.1504, loss_cls_5: 0.8607, loss_box_5: 1.4875, loss_cns_5: 0.6638, loss_yns_5: 0.1501, loss_cls_dn_0: 0.1614, loss_box_dn_0: 0.7247, loss_cls_dn_1: 0.1054, loss_box_dn_1: 0.6603, loss_cls_dn_2: 0.1037, loss_box_dn_2: 0.6553, loss_cls_dn_3: 0.1071, loss_box_dn_3: 0.6549, loss_cls_dn_4: 0.1071, loss_box_dn_4: 0.6591, loss_cls_dn_5: 0.1090, loss_box_dn_5: 0.6683, loss_dense_depth: 0.7987, loss: 24.5805, grad_norm: 34.8172 -2025-11-12 14:39:10,506 - mmdet - INFO - Iter [308/17500] lr: 2.226e-04, eta: 9:46:11, time: 1.596, data_time: 0.080, memory: 49164, loss_cls_0: 0.7788, loss_box_0: 1.6320, loss_cns_0: 0.6362, loss_yns_0: 0.1485, loss_cls_1: 0.8414, loss_box_1: 1.5362, loss_cns_1: 0.6574, loss_yns_1: 0.1482, loss_cls_2: 0.8407, loss_box_2: 1.5088, loss_cns_2: 0.6608, loss_yns_2: 0.1479, loss_cls_3: 0.8539, loss_box_3: 1.5132, loss_cns_3: 0.6653, loss_yns_3: 0.1479, loss_cls_4: 0.8571, loss_box_4: 1.5108, loss_cns_4: 0.6629, loss_yns_4: 0.1485, loss_cls_5: 0.8654, loss_box_5: 1.5118, loss_cns_5: 0.6617, loss_yns_5: 0.1471, loss_cls_dn_0: 0.1620, loss_box_dn_0: 0.7351, loss_cls_dn_1: 0.1087, loss_box_dn_1: 0.6680, loss_cls_dn_2: 0.1054, loss_box_dn_2: 0.6618, loss_cls_dn_3: 0.1075, loss_box_dn_3: 0.6690, loss_cls_dn_4: 0.1105, loss_box_dn_4: 0.6778, loss_cls_dn_5: 0.1120, loss_box_dn_5: 0.6903, loss_dense_depth: 0.6941, loss: 24.5844, grad_norm: 39.9060 -2025-11-12 14:39:12,085 - mmdet - INFO - Iter [309/17500] lr: 2.230e-04, eta: 9:45:43, time: 1.576, data_time: 0.082, memory: 49164, loss_cls_0: 0.7769, loss_box_0: 1.6448, loss_cns_0: 0.6328, loss_yns_0: 0.1509, loss_cls_1: 0.8344, loss_box_1: 1.5350, loss_cns_1: 0.6575, loss_yns_1: 0.1474, loss_cls_2: 0.8369, loss_box_2: 1.5223, loss_cns_2: 0.6573, loss_yns_2: 0.1462, loss_cls_3: 0.8526, loss_box_3: 1.5185, loss_cns_3: 0.6599, loss_yns_3: 0.1474, loss_cls_4: 0.8539, loss_box_4: 1.5092, loss_cns_4: 0.6567, loss_yns_4: 0.1473, loss_cls_5: 0.8582, loss_box_5: 1.5105, loss_cns_5: 0.6572, loss_yns_5: 0.1466, loss_cls_dn_0: 0.1711, loss_box_dn_0: 0.7288, loss_cls_dn_1: 0.1094, loss_box_dn_1: 0.6726, loss_cls_dn_2: 0.1075, loss_box_dn_2: 0.6653, loss_cls_dn_3: 0.1082, loss_box_dn_3: 0.6687, loss_cls_dn_4: 0.1119, loss_box_dn_4: 0.6701, loss_cls_dn_5: 0.1112, loss_box_dn_5: 0.6801, loss_dense_depth: 0.7671, loss: 24.6321, grad_norm: 40.3232 -2025-11-12 14:39:13,683 - mmdet - INFO - Iter [310/17500] lr: 2.234e-04, eta: 9:45:16, time: 1.595, data_time: 0.086, memory: 49164, loss_cls_0: 0.7565, loss_box_0: 1.6331, loss_cns_0: 0.6364, loss_yns_0: 0.1497, loss_cls_1: 0.8370, loss_box_1: 1.4931, loss_cns_1: 0.6541, loss_yns_1: 0.1450, loss_cls_2: 0.8438, loss_box_2: 1.4642, loss_cns_2: 0.6548, loss_yns_2: 0.1429, loss_cls_3: 0.8532, loss_box_3: 1.4432, loss_cns_3: 0.6511, loss_yns_3: 0.1428, loss_cls_4: 0.8500, loss_box_4: 1.4635, loss_cns_4: 0.6602, loss_yns_4: 0.1476, loss_cls_5: 0.8705, loss_box_5: 1.4754, loss_cns_5: 0.6582, loss_yns_5: 0.1460, loss_cls_dn_0: 0.1685, loss_box_dn_0: 0.7338, loss_cls_dn_1: 0.1065, loss_box_dn_1: 0.6578, loss_cls_dn_2: 0.1043, loss_box_dn_2: 0.6407, loss_cls_dn_3: 0.1050, loss_box_dn_3: 0.6369, loss_cls_dn_4: 0.1061, loss_box_dn_4: 0.6394, loss_cls_dn_5: 0.1107, loss_box_dn_5: 0.6482, loss_dense_depth: 0.7072, loss: 24.1375, grad_norm: 31.9717 -2025-11-12 14:39:15,299 - mmdet - INFO - Iter [311/17500] lr: 2.238e-04, eta: 9:44:50, time: 1.615, data_time: 0.082, memory: 49164, loss_cls_0: 0.7611, loss_box_0: 1.6213, loss_cns_0: 0.6314, loss_yns_0: 0.1480, loss_cls_1: 0.8346, loss_box_1: 1.4968, loss_cns_1: 0.6527, loss_yns_1: 0.1443, loss_cls_2: 0.8441, loss_box_2: 1.4813, loss_cns_2: 0.6553, loss_yns_2: 0.1438, loss_cls_3: 0.8514, loss_box_3: 1.4578, loss_cns_3: 0.6483, loss_yns_3: 0.1424, loss_cls_4: 0.8458, loss_box_4: 1.4806, loss_cns_4: 0.6561, loss_yns_4: 0.1464, loss_cls_5: 0.8732, loss_box_5: 1.4852, loss_cns_5: 0.6550, loss_yns_5: 0.1448, loss_cls_dn_0: 0.1615, loss_box_dn_0: 0.7311, loss_cls_dn_1: 0.1095, loss_box_dn_1: 0.6498, loss_cls_dn_2: 0.1087, loss_box_dn_2: 0.6408, loss_cls_dn_3: 0.1084, loss_box_dn_3: 0.6360, loss_cls_dn_4: 0.1091, loss_box_dn_4: 0.6391, loss_cls_dn_5: 0.1179, loss_box_dn_5: 0.6455, loss_dense_depth: 0.7953, loss: 24.2543, grad_norm: 35.5391 -2025-11-12 14:39:16,891 - mmdet - INFO - Iter [312/17500] lr: 2.242e-04, eta: 9:44:23, time: 1.593, data_time: 0.081, memory: 49164, loss_cls_0: 0.7775, loss_box_0: 1.6558, loss_cns_0: 0.6294, loss_yns_0: 0.1471, loss_cls_1: 0.8254, loss_box_1: 1.5225, loss_cns_1: 0.6527, loss_yns_1: 0.1451, loss_cls_2: 0.8428, loss_box_2: 1.5050, loss_cns_2: 0.6550, loss_yns_2: 0.1452, loss_cls_3: 0.8512, loss_box_3: 1.4829, loss_cns_3: 0.6537, loss_yns_3: 0.1452, loss_cls_4: 0.8472, loss_box_4: 1.4814, loss_cns_4: 0.6556, loss_yns_4: 0.1450, loss_cls_5: 0.8493, loss_box_5: 1.4945, loss_cns_5: 0.6564, loss_yns_5: 0.1455, loss_cls_dn_0: 0.1620, loss_box_dn_0: 0.7282, loss_cls_dn_1: 0.1134, loss_box_dn_1: 0.6480, loss_cls_dn_2: 0.1091, loss_box_dn_2: 0.6423, loss_cls_dn_3: 0.1107, loss_box_dn_3: 0.6420, loss_cls_dn_4: 0.1085, loss_box_dn_4: 0.6465, loss_cls_dn_5: 0.1103, loss_box_dn_5: 0.6589, loss_dense_depth: 0.7505, loss: 24.3419, grad_norm: 33.6367 -2025-11-12 14:39:18,474 - mmdet - INFO - Iter [313/17500] lr: 2.246e-04, eta: 9:43:56, time: 1.589, data_time: 0.079, memory: 49164, loss_cls_0: 0.7678, loss_box_0: 1.5980, loss_cns_0: 0.6347, loss_yns_0: 0.1489, loss_cls_1: 0.8278, loss_box_1: 1.4860, loss_cns_1: 0.6609, loss_yns_1: 0.1479, loss_cls_2: 0.8355, loss_box_2: 1.4640, loss_cns_2: 0.6616, loss_yns_2: 0.1471, loss_cls_3: 0.8251, loss_box_3: 1.4562, loss_cns_3: 0.6614, loss_yns_3: 0.1468, loss_cls_4: 0.8367, loss_box_4: 1.4523, loss_cns_4: 0.6616, loss_yns_4: 0.1468, loss_cls_5: 0.8426, loss_box_5: 1.4406, loss_cns_5: 0.6670, loss_yns_5: 0.1467, loss_cls_dn_0: 0.1545, loss_box_dn_0: 0.7260, loss_cls_dn_1: 0.1048, loss_box_dn_1: 0.6812, loss_cls_dn_2: 0.1042, loss_box_dn_2: 0.6734, loss_cls_dn_3: 0.1044, loss_box_dn_3: 0.6772, loss_cls_dn_4: 0.1107, loss_box_dn_4: 0.6823, loss_cls_dn_5: 0.1177, loss_box_dn_5: 0.6870, loss_dense_depth: 0.7871, loss: 24.2747, grad_norm: 40.3664 -2025-11-12 14:39:20,107 - mmdet - INFO - Iter [314/17500] lr: 2.250e-04, eta: 9:43:32, time: 1.630, data_time: 0.077, memory: 49164, loss_cls_0: 0.7798, loss_box_0: 1.6042, loss_cns_0: 0.6326, loss_yns_0: 0.1484, loss_cls_1: 0.8505, loss_box_1: 1.5175, loss_cns_1: 0.6590, loss_yns_1: 0.1464, loss_cls_2: 0.8534, loss_box_2: 1.5050, loss_cns_2: 0.6594, loss_yns_2: 0.1459, loss_cls_3: 0.8541, loss_box_3: 1.4860, loss_cns_3: 0.6622, loss_yns_3: 0.1468, loss_cls_4: 0.8656, loss_box_4: 1.4677, loss_cns_4: 0.6606, loss_yns_4: 0.1498, loss_cls_5: 0.8772, loss_box_5: 1.4730, loss_cns_5: 0.6637, loss_yns_5: 0.1477, loss_cls_dn_0: 0.1605, loss_box_dn_0: 0.7293, loss_cls_dn_1: 0.1137, loss_box_dn_1: 0.7054, loss_cls_dn_2: 0.1179, loss_box_dn_2: 0.7048, loss_cls_dn_3: 0.1161, loss_box_dn_3: 0.7044, loss_cls_dn_4: 0.1214, loss_box_dn_4: 0.7008, loss_cls_dn_5: 0.1315, loss_box_dn_5: 0.7067, loss_dense_depth: 0.7195, loss: 24.6883, grad_norm: 40.2051 -2025-11-12 14:39:21,692 - mmdet - INFO - Iter [315/17500] lr: 2.254e-04, eta: 9:43:05, time: 1.582, data_time: 0.074, memory: 49164, loss_cls_0: 0.7618, loss_box_0: 1.6111, loss_cns_0: 0.6303, loss_yns_0: 0.1468, loss_cls_1: 0.8237, loss_box_1: 1.5256, loss_cns_1: 0.6559, loss_yns_1: 0.1480, loss_cls_2: 0.8287, loss_box_2: 1.4833, loss_cns_2: 0.6564, loss_yns_2: 0.1456, loss_cls_3: 0.8386, loss_box_3: 1.4639, loss_cns_3: 0.6555, loss_yns_3: 0.1457, loss_cls_4: 0.8403, loss_box_4: 1.4627, loss_cns_4: 0.6619, loss_yns_4: 0.1467, loss_cls_5: 0.8586, loss_box_5: 1.4838, loss_cns_5: 0.6558, loss_yns_5: 0.1468, loss_cls_dn_0: 0.1539, loss_box_dn_0: 0.7309, loss_cls_dn_1: 0.1073, loss_box_dn_1: 0.6935, loss_cls_dn_2: 0.1066, loss_box_dn_2: 0.6826, loss_cls_dn_3: 0.1061, loss_box_dn_3: 0.6756, loss_cls_dn_4: 0.1062, loss_box_dn_4: 0.6744, loss_cls_dn_5: 0.1145, loss_box_dn_5: 0.6869, loss_dense_depth: 0.7587, loss: 24.3747, grad_norm: 31.6132 -2025-11-12 14:39:23,262 - mmdet - INFO - Iter [316/17500] lr: 2.258e-04, eta: 9:42:38, time: 1.571, data_time: 0.078, memory: 49164, loss_cls_0: 0.7619, loss_box_0: 1.5812, loss_cns_0: 0.6297, loss_yns_0: 0.1439, loss_cls_1: 0.8192, loss_box_1: 1.5021, loss_cns_1: 0.6583, loss_yns_1: 0.1437, loss_cls_2: 0.8326, loss_box_2: 1.4508, loss_cns_2: 0.6628, loss_yns_2: 0.1439, loss_cls_3: 0.8309, loss_box_3: 1.4368, loss_cns_3: 0.6596, loss_yns_3: 0.1426, loss_cls_4: 0.8404, loss_box_4: 1.4329, loss_cns_4: 0.6656, loss_yns_4: 0.1437, loss_cls_5: 0.8501, loss_box_5: 1.4424, loss_cns_5: 0.6608, loss_yns_5: 0.1433, loss_cls_dn_0: 0.1509, loss_box_dn_0: 0.7207, loss_cls_dn_1: 0.1075, loss_box_dn_1: 0.6750, loss_cls_dn_2: 0.1065, loss_box_dn_2: 0.6600, loss_cls_dn_3: 0.1078, loss_box_dn_3: 0.6525, loss_cls_dn_4: 0.1114, loss_box_dn_4: 0.6511, loss_cls_dn_5: 0.1099, loss_box_dn_5: 0.6573, loss_dense_depth: 0.6996, loss: 23.9894, grad_norm: 32.6366 -2025-11-12 14:39:24,857 - mmdet - INFO - Iter [317/17500] lr: 2.262e-04, eta: 9:42:12, time: 1.596, data_time: 0.080, memory: 49164, loss_cls_0: 0.7417, loss_box_0: 1.5659, loss_cns_0: 0.6286, loss_yns_0: 0.1422, loss_cls_1: 0.8144, loss_box_1: 1.4431, loss_cns_1: 0.6570, loss_yns_1: 0.1433, loss_cls_2: 0.8318, loss_box_2: 1.4280, loss_cns_2: 0.6597, loss_yns_2: 0.1445, loss_cls_3: 0.8213, loss_box_3: 1.4231, loss_cns_3: 0.6588, loss_yns_3: 0.1440, loss_cls_4: 0.8289, loss_box_4: 1.4108, loss_cns_4: 0.6597, loss_yns_4: 0.1445, loss_cls_5: 0.8478, loss_box_5: 1.4274, loss_cns_5: 0.6589, loss_yns_5: 0.1429, loss_cls_dn_0: 0.1499, loss_box_dn_0: 0.7277, loss_cls_dn_1: 0.1084, loss_box_dn_1: 0.6623, loss_cls_dn_2: 0.1072, loss_box_dn_2: 0.6516, loss_cls_dn_3: 0.1078, loss_box_dn_3: 0.6509, loss_cls_dn_4: 0.1098, loss_box_dn_4: 0.6496, loss_cls_dn_5: 0.1094, loss_box_dn_5: 0.6610, loss_dense_depth: 0.7230, loss: 23.7870, grad_norm: 29.1179 -2025-11-12 14:39:26,414 - mmdet - INFO - Iter [318/17500] lr: 2.266e-04, eta: 9:41:44, time: 1.558, data_time: 0.075, memory: 49164, loss_cls_0: 0.7767, loss_box_0: 1.5665, loss_cns_0: 0.6233, loss_yns_0: 0.1426, loss_cls_1: 0.8261, loss_box_1: 1.4728, loss_cns_1: 0.6594, loss_yns_1: 0.1444, loss_cls_2: 0.8394, loss_box_2: 1.4540, loss_cns_2: 0.6609, loss_yns_2: 0.1440, loss_cls_3: 0.8406, loss_box_3: 1.4518, loss_cns_3: 0.6624, loss_yns_3: 0.1452, loss_cls_4: 0.8433, loss_box_4: 1.4331, loss_cns_4: 0.6618, loss_yns_4: 0.1470, loss_cls_5: 0.8585, loss_box_5: 1.4430, loss_cns_5: 0.6588, loss_yns_5: 0.1443, loss_cls_dn_0: 0.1528, loss_box_dn_0: 0.7310, loss_cls_dn_1: 0.1015, loss_box_dn_1: 0.6528, loss_cls_dn_2: 0.0999, loss_box_dn_2: 0.6422, loss_cls_dn_3: 0.0997, loss_box_dn_3: 0.6491, loss_cls_dn_4: 0.1022, loss_box_dn_4: 0.6482, loss_cls_dn_5: 0.1057, loss_box_dn_5: 0.6580, loss_dense_depth: 0.7465, loss: 23.9896, grad_norm: 29.7883 -2025-11-12 14:39:27,990 - mmdet - INFO - Iter [319/17500] lr: 2.270e-04, eta: 9:41:18, time: 1.579, data_time: 0.080, memory: 49164, loss_cls_0: 0.7425, loss_box_0: 1.5642, loss_cns_0: 0.6347, loss_yns_0: 0.1458, loss_cls_1: 0.8019, loss_box_1: 1.4550, loss_cns_1: 0.6625, loss_yns_1: 0.1450, loss_cls_2: 0.8139, loss_box_2: 1.4220, loss_cns_2: 0.6676, loss_yns_2: 0.1450, loss_cls_3: 0.8264, loss_box_3: 1.4156, loss_cns_3: 0.6643, loss_yns_3: 0.1455, loss_cls_4: 0.8227, loss_box_4: 1.4160, loss_cns_4: 0.6684, loss_yns_4: 0.1441, loss_cls_5: 0.8291, loss_box_5: 1.4167, loss_cns_5: 0.6621, loss_yns_5: 0.1446, loss_cls_dn_0: 0.1423, loss_box_dn_0: 0.7341, loss_cls_dn_1: 0.0977, loss_box_dn_1: 0.6620, loss_cls_dn_2: 0.0968, loss_box_dn_2: 0.6528, loss_cls_dn_3: 0.0975, loss_box_dn_3: 0.6552, loss_cls_dn_4: 0.1003, loss_box_dn_4: 0.6614, loss_cls_dn_5: 0.1051, loss_box_dn_5: 0.6679, loss_dense_depth: 0.7044, loss: 23.7331, grad_norm: 28.9687 -2025-11-12 14:39:29,578 - mmdet - INFO - Iter [320/17500] lr: 2.274e-04, eta: 9:40:52, time: 1.588, data_time: 0.077, memory: 49164, loss_cls_0: 0.7486, loss_box_0: 1.5890, loss_cns_0: 0.6375, loss_yns_0: 0.1444, loss_cls_1: 0.8209, loss_box_1: 1.4548, loss_cns_1: 0.6625, loss_yns_1: 0.1432, loss_cls_2: 0.8457, loss_box_2: 1.4287, loss_cns_2: 0.6702, loss_yns_2: 0.1439, loss_cls_3: 0.8370, loss_box_3: 1.4175, loss_cns_3: 0.6658, loss_yns_3: 0.1437, loss_cls_4: 0.8336, loss_box_4: 1.4226, loss_cns_4: 0.6655, loss_yns_4: 0.1442, loss_cls_5: 0.8448, loss_box_5: 1.4136, loss_cns_5: 0.6633, loss_yns_5: 0.1439, loss_cls_dn_0: 0.1433, loss_box_dn_0: 0.7179, loss_cls_dn_1: 0.1023, loss_box_dn_1: 0.6595, loss_cls_dn_2: 0.1000, loss_box_dn_2: 0.6486, loss_cls_dn_3: 0.1003, loss_box_dn_3: 0.6463, loss_cls_dn_4: 0.1026, loss_box_dn_4: 0.6531, loss_cls_dn_5: 0.1041, loss_box_dn_5: 0.6575, loss_dense_depth: 0.7215, loss: 23.8421, grad_norm: 29.4316 -2025-11-12 14:39:31,251 - mmdet - INFO - Iter [321/17500] lr: 2.278e-04, eta: 9:40:31, time: 1.669, data_time: 0.113, memory: 49164, loss_cls_0: 0.7152, loss_box_0: 1.5708, loss_cns_0: 0.6419, loss_yns_0: 0.1473, loss_cls_1: 0.7991, loss_box_1: 1.4457, loss_cns_1: 0.6638, loss_yns_1: 0.1457, loss_cls_2: 0.8084, loss_box_2: 1.4296, loss_cns_2: 0.6672, loss_yns_2: 0.1453, loss_cls_3: 0.8145, loss_box_3: 1.4096, loss_cns_3: 0.6663, loss_yns_3: 0.1439, loss_cls_4: 0.8256, loss_box_4: 1.3950, loss_cns_4: 0.6621, loss_yns_4: 0.1444, loss_cls_5: 0.8270, loss_box_5: 1.3931, loss_cns_5: 0.6610, loss_yns_5: 0.1455, loss_cls_dn_0: 0.1425, loss_box_dn_0: 0.7216, loss_cls_dn_1: 0.1010, loss_box_dn_1: 0.6622, loss_cls_dn_2: 0.0977, loss_box_dn_2: 0.6489, loss_cls_dn_3: 0.1017, loss_box_dn_3: 0.6421, loss_cls_dn_4: 0.1042, loss_box_dn_4: 0.6382, loss_cls_dn_5: 0.1038, loss_box_dn_5: 0.6437, loss_dense_depth: 0.6764, loss: 23.5519, grad_norm: 31.5236 -2025-11-12 14:39:32,890 - mmdet - INFO - Iter [322/17500] lr: 2.282e-04, eta: 9:40:08, time: 1.644, data_time: 0.116, memory: 49164, loss_cls_0: 0.7298, loss_box_0: 1.5783, loss_cns_0: 0.6397, loss_yns_0: 0.1485, loss_cls_1: 0.8210, loss_box_1: 1.3932, loss_cns_1: 0.6620, loss_yns_1: 0.1444, loss_cls_2: 0.8215, loss_box_2: 1.3687, loss_cns_2: 0.6664, loss_yns_2: 0.1436, loss_cls_3: 0.8350, loss_box_3: 1.3442, loss_cns_3: 0.6600, loss_yns_3: 0.1428, loss_cls_4: 0.8297, loss_box_4: 1.3630, loss_cns_4: 0.6645, loss_yns_4: 0.1452, loss_cls_5: 0.8377, loss_box_5: 1.3587, loss_cns_5: 0.6633, loss_yns_5: 0.1455, loss_cls_dn_0: 0.1455, loss_box_dn_0: 0.7225, loss_cls_dn_1: 0.1007, loss_box_dn_1: 0.6611, loss_cls_dn_2: 0.0983, loss_box_dn_2: 0.6435, loss_cls_dn_3: 0.1015, loss_box_dn_3: 0.6413, loss_cls_dn_4: 0.1012, loss_box_dn_4: 0.6467, loss_cls_dn_5: 0.1014, loss_box_dn_5: 0.6507, loss_dense_depth: 0.6845, loss: 23.4058, grad_norm: 26.8259 -2025-11-12 14:39:34,489 - mmdet - INFO - Iter [323/17500] lr: 2.286e-04, eta: 9:39:44, time: 1.598, data_time: 0.082, memory: 49164, loss_cls_0: 0.7210, loss_box_0: 1.5813, loss_cns_0: 0.6409, loss_yns_0: 0.1444, loss_cls_1: 0.7906, loss_box_1: 1.4512, loss_cns_1: 0.6647, loss_yns_1: 0.1448, loss_cls_2: 0.8065, loss_box_2: 1.4261, loss_cns_2: 0.6660, loss_yns_2: 0.1456, loss_cls_3: 0.8187, loss_box_3: 1.4022, loss_cns_3: 0.6602, loss_yns_3: 0.1432, loss_cls_4: 0.8288, loss_box_4: 1.4238, loss_cns_4: 0.6694, loss_yns_4: 0.1463, loss_cls_5: 0.8263, loss_box_5: 1.4120, loss_cns_5: 0.6649, loss_yns_5: 0.1451, loss_cls_dn_0: 0.1439, loss_box_dn_0: 0.7257, loss_cls_dn_1: 0.0985, loss_box_dn_1: 0.6603, loss_cls_dn_2: 0.0976, loss_box_dn_2: 0.6464, loss_cls_dn_3: 0.0976, loss_box_dn_3: 0.6506, loss_cls_dn_4: 0.0975, loss_box_dn_4: 0.6602, loss_cls_dn_5: 0.0994, loss_box_dn_5: 0.6646, loss_dense_depth: 0.6588, loss: 23.6252, grad_norm: 33.9324 -2025-11-12 14:39:36,082 - mmdet - INFO - Iter [324/17500] lr: 2.290e-04, eta: 9:39:19, time: 1.592, data_time: 0.077, memory: 49164, loss_cls_0: 0.7347, loss_box_0: 1.6118, loss_cns_0: 0.6351, loss_yns_0: 0.1435, loss_cls_1: 0.8145, loss_box_1: 1.4710, loss_cns_1: 0.6652, loss_yns_1: 0.1448, loss_cls_2: 0.8138, loss_box_2: 1.4479, loss_cns_2: 0.6681, loss_yns_2: 0.1429, loss_cls_3: 0.8329, loss_box_3: 1.4128, loss_cns_3: 0.6589, loss_yns_3: 0.1405, loss_cls_4: 0.8318, loss_box_4: 1.4305, loss_cns_4: 0.6679, loss_yns_4: 0.1428, loss_cls_5: 0.8417, loss_box_5: 1.4314, loss_cns_5: 0.6678, loss_yns_5: 0.1424, loss_cls_dn_0: 0.1399, loss_box_dn_0: 0.7259, loss_cls_dn_1: 0.0980, loss_box_dn_1: 0.6578, loss_cls_dn_2: 0.0962, loss_box_dn_2: 0.6514, loss_cls_dn_3: 0.0969, loss_box_dn_3: 0.6475, loss_cls_dn_4: 0.0972, loss_box_dn_4: 0.6568, loss_cls_dn_5: 0.0996, loss_box_dn_5: 0.6649, loss_dense_depth: 0.6673, loss: 23.7944, grad_norm: 34.1145 -2025-11-12 14:39:37,673 - mmdet - INFO - Iter [325/17500] lr: 2.294e-04, eta: 9:38:53, time: 1.583, data_time: 0.082, memory: 49164, loss_cls_0: 0.7246, loss_box_0: 1.5743, loss_cns_0: 0.6376, loss_yns_0: 0.1440, loss_cls_1: 0.7919, loss_box_1: 1.4679, loss_cns_1: 0.6640, loss_yns_1: 0.1461, loss_cls_2: 0.7924, loss_box_2: 1.4509, loss_cns_2: 0.6642, loss_yns_2: 0.1451, loss_cls_3: 0.8032, loss_box_3: 1.4258, loss_cns_3: 0.6606, loss_yns_3: 0.1435, loss_cls_4: 0.8002, loss_box_4: 1.4381, loss_cns_4: 0.6651, loss_yns_4: 0.1451, loss_cls_5: 0.8097, loss_box_5: 1.4339, loss_cns_5: 0.6658, loss_yns_5: 0.1450, loss_cls_dn_0: 0.1463, loss_box_dn_0: 0.7175, loss_cls_dn_1: 0.0985, loss_box_dn_1: 0.6536, loss_cls_dn_2: 0.0974, loss_box_dn_2: 0.6502, loss_cls_dn_3: 0.0991, loss_box_dn_3: 0.6455, loss_cls_dn_4: 0.0991, loss_box_dn_4: 0.6463, loss_cls_dn_5: 0.1001, loss_box_dn_5: 0.6508, loss_dense_depth: 0.6868, loss: 23.6303, grad_norm: 29.8531 -2025-11-12 14:39:39,283 - mmdet - INFO - Iter [326/17500] lr: 2.298e-04, eta: 9:38:30, time: 1.614, data_time: 0.111, memory: 49164, loss_cls_0: 0.7125, loss_box_0: 1.6009, loss_cns_0: 0.6360, loss_yns_0: 0.1449, loss_cls_1: 0.7832, loss_box_1: 1.4613, loss_cns_1: 0.6632, loss_yns_1: 0.1433, loss_cls_2: 0.7863, loss_box_2: 1.4383, loss_cns_2: 0.6636, loss_yns_2: 0.1420, loss_cls_3: 0.7936, loss_box_3: 1.4240, loss_cns_3: 0.6628, loss_yns_3: 0.1420, loss_cls_4: 0.8037, loss_box_4: 1.4291, loss_cns_4: 0.6635, loss_yns_4: 0.1429, loss_cls_5: 0.8112, loss_box_5: 1.4240, loss_cns_5: 0.6638, loss_yns_5: 0.1405, loss_cls_dn_0: 0.1416, loss_box_dn_0: 0.7192, loss_cls_dn_1: 0.0968, loss_box_dn_1: 0.6311, loss_cls_dn_2: 0.0969, loss_box_dn_2: 0.6233, loss_cls_dn_3: 0.1004, loss_box_dn_3: 0.6177, loss_cls_dn_4: 0.0994, loss_box_dn_4: 0.6204, loss_cls_dn_5: 0.0987, loss_box_dn_5: 0.6232, loss_dense_depth: 0.6674, loss: 23.4126, grad_norm: 27.7309 -2025-11-12 14:39:40,860 - mmdet - INFO - Iter [327/17500] lr: 2.302e-04, eta: 9:38:04, time: 1.576, data_time: 0.078, memory: 49164, loss_cls_0: 0.7212, loss_box_0: 1.6080, loss_cns_0: 0.6394, loss_yns_0: 0.1395, loss_cls_1: 0.7762, loss_box_1: 1.4774, loss_cns_1: 0.6596, loss_yns_1: 0.1373, loss_cls_2: 0.7970, loss_box_2: 1.4419, loss_cns_2: 0.6615, loss_yns_2: 0.1391, loss_cls_3: 0.8072, loss_box_3: 1.4459, loss_cns_3: 0.6614, loss_yns_3: 0.1386, loss_cls_4: 0.8101, loss_box_4: 1.4493, loss_cns_4: 0.6610, loss_yns_4: 0.1377, loss_cls_5: 0.8168, loss_box_5: 1.4346, loss_cns_5: 0.6598, loss_yns_5: 0.1382, loss_cls_dn_0: 0.1397, loss_box_dn_0: 0.7144, loss_cls_dn_1: 0.0990, loss_box_dn_1: 0.6224, loss_cls_dn_2: 0.0995, loss_box_dn_2: 0.6091, loss_cls_dn_3: 0.1016, loss_box_dn_3: 0.6103, loss_cls_dn_4: 0.1014, loss_box_dn_4: 0.6141, loss_cls_dn_5: 0.1022, loss_box_dn_5: 0.6160, loss_dense_depth: 0.6857, loss: 23.4741, grad_norm: 30.7092 -2025-11-12 14:39:42,444 - mmdet - INFO - Iter [328/17500] lr: 2.306e-04, eta: 9:37:40, time: 1.587, data_time: 0.083, memory: 49164, loss_cls_0: 0.7161, loss_box_0: 1.6243, loss_cns_0: 0.6420, loss_yns_0: 0.1442, loss_cls_1: 0.7719, loss_box_1: 1.4739, loss_cns_1: 0.6618, loss_yns_1: 0.1399, loss_cls_2: 0.7983, loss_box_2: 1.4597, loss_cns_2: 0.6659, loss_yns_2: 0.1403, loss_cls_3: 0.8194, loss_box_3: 1.4358, loss_cns_3: 0.6687, loss_yns_3: 0.1398, loss_cls_4: 0.8074, loss_box_4: 1.4345, loss_cns_4: 0.6695, loss_yns_4: 0.1404, loss_cls_5: 0.8135, loss_box_5: 1.4333, loss_cns_5: 0.6691, loss_yns_5: 0.1389, loss_cls_dn_0: 0.1394, loss_box_dn_0: 0.7246, loss_cls_dn_1: 0.0994, loss_box_dn_1: 0.6359, loss_cls_dn_2: 0.0977, loss_box_dn_2: 0.6304, loss_cls_dn_3: 0.0968, loss_box_dn_3: 0.6255, loss_cls_dn_4: 0.1003, loss_box_dn_4: 0.6284, loss_cls_dn_5: 0.1041, loss_box_dn_5: 0.6379, loss_dense_depth: 0.7001, loss: 23.6293, grad_norm: 34.4178 -2025-11-12 14:39:44,049 - mmdet - INFO - Iter [329/17500] lr: 2.310e-04, eta: 9:37:16, time: 1.606, data_time: 0.089, memory: 49164, loss_cls_0: 0.7293, loss_box_0: 1.6241, loss_cns_0: 0.6356, loss_yns_0: 0.1426, loss_cls_1: 0.7932, loss_box_1: 1.4827, loss_cns_1: 0.6633, loss_yns_1: 0.1409, loss_cls_2: 0.8063, loss_box_2: 1.4684, loss_cns_2: 0.6670, loss_yns_2: 0.1432, loss_cls_3: 0.8269, loss_box_3: 1.4462, loss_cns_3: 0.6674, loss_yns_3: 0.1435, loss_cls_4: 0.8229, loss_box_4: 1.4523, loss_cns_4: 0.6698, loss_yns_4: 0.1485, loss_cls_5: 0.8196, loss_box_5: 1.4507, loss_cns_5: 0.6728, loss_yns_5: 0.1412, loss_cls_dn_0: 0.1463, loss_box_dn_0: 0.7238, loss_cls_dn_1: 0.1025, loss_box_dn_1: 0.6479, loss_cls_dn_2: 0.1038, loss_box_dn_2: 0.6425, loss_cls_dn_3: 0.1037, loss_box_dn_3: 0.6407, loss_cls_dn_4: 0.1042, loss_box_dn_4: 0.6503, loss_cls_dn_5: 0.1061, loss_box_dn_5: 0.6610, loss_dense_depth: 0.7479, loss: 23.9391, grad_norm: 29.1325 -2025-11-12 14:39:45,652 - mmdet - INFO - Iter [330/17500] lr: 2.314e-04, eta: 9:36:53, time: 1.601, data_time: 0.089, memory: 49164, loss_cls_0: 0.7098, loss_box_0: 1.6006, loss_cns_0: 0.6382, loss_yns_0: 0.1412, loss_cls_1: 0.7787, loss_box_1: 1.4635, loss_cns_1: 0.6657, loss_yns_1: 0.1389, loss_cls_2: 0.7887, loss_box_2: 1.4442, loss_cns_2: 0.6671, loss_yns_2: 0.1410, loss_cls_3: 0.8076, loss_box_3: 1.4522, loss_cns_3: 0.6667, loss_yns_3: 0.1408, loss_cls_4: 0.8039, loss_box_4: 1.4386, loss_cns_4: 0.6672, loss_yns_4: 0.1401, loss_cls_5: 0.8053, loss_box_5: 1.4303, loss_cns_5: 0.6720, loss_yns_5: 0.1393, loss_cls_dn_0: 0.1442, loss_box_dn_0: 0.7295, loss_cls_dn_1: 0.0998, loss_box_dn_1: 0.6587, loss_cls_dn_2: 0.1015, loss_box_dn_2: 0.6483, loss_cls_dn_3: 0.1009, loss_box_dn_3: 0.6529, loss_cls_dn_4: 0.1025, loss_box_dn_4: 0.6547, loss_cls_dn_5: 0.1007, loss_box_dn_5: 0.6570, loss_dense_depth: 0.7456, loss: 23.7380, grad_norm: 41.2340 -2025-11-12 14:39:47,257 - mmdet - INFO - Iter [331/17500] lr: 2.318e-04, eta: 9:36:29, time: 1.604, data_time: 0.079, memory: 49164, loss_cls_0: 0.7270, loss_box_0: 1.6161, loss_cns_0: 0.6328, loss_yns_0: 0.1397, loss_cls_1: 0.7862, loss_box_1: 1.4951, loss_cns_1: 0.6650, loss_yns_1: 0.1391, loss_cls_2: 0.7895, loss_box_2: 1.4669, loss_cns_2: 0.6636, loss_yns_2: 0.1376, loss_cls_3: 0.7982, loss_box_3: 1.4520, loss_cns_3: 0.6651, loss_yns_3: 0.1379, loss_cls_4: 0.7987, loss_box_4: 1.4397, loss_cns_4: 0.6624, loss_yns_4: 0.1382, loss_cls_5: 0.8082, loss_box_5: 1.4397, loss_cns_5: 0.6638, loss_yns_5: 0.1387, loss_cls_dn_0: 0.1453, loss_box_dn_0: 0.7131, loss_cls_dn_1: 0.0989, loss_box_dn_1: 0.6450, loss_cls_dn_2: 0.0979, loss_box_dn_2: 0.6318, loss_cls_dn_3: 0.0993, loss_box_dn_3: 0.6293, loss_cls_dn_4: 0.1003, loss_box_dn_4: 0.6288, loss_cls_dn_5: 0.1022, loss_box_dn_5: 0.6297, loss_dense_depth: 0.7106, loss: 23.6334, grad_norm: 27.1277 -2025-11-12 14:39:48,833 - mmdet - INFO - Iter [332/17500] lr: 2.322e-04, eta: 9:36:05, time: 1.579, data_time: 0.078, memory: 49164, loss_cls_0: 0.7097, loss_box_0: 1.6008, loss_cns_0: 0.6414, loss_yns_0: 0.1429, loss_cls_1: 0.7781, loss_box_1: 1.4553, loss_cns_1: 0.6673, loss_yns_1: 0.1413, loss_cls_2: 0.7887, loss_box_2: 1.4327, loss_cns_2: 0.6670, loss_yns_2: 0.1400, loss_cls_3: 0.7918, loss_box_3: 1.4312, loss_cns_3: 0.6667, loss_yns_3: 0.1398, loss_cls_4: 0.8046, loss_box_4: 1.4079, loss_cns_4: 0.6654, loss_yns_4: 0.1405, loss_cls_5: 0.8188, loss_box_5: 1.4141, loss_cns_5: 0.6686, loss_yns_5: 0.1418, loss_cls_dn_0: 0.1407, loss_box_dn_0: 0.7235, loss_cls_dn_1: 0.0996, loss_box_dn_1: 0.6423, loss_cls_dn_2: 0.0984, loss_box_dn_2: 0.6297, loss_cls_dn_3: 0.0994, loss_box_dn_3: 0.6298, loss_cls_dn_4: 0.0997, loss_box_dn_4: 0.6280, loss_cls_dn_5: 0.1023, loss_box_dn_5: 0.6295, loss_dense_depth: 0.6643, loss: 23.4434, grad_norm: 40.0010 -2025-11-12 14:39:50,419 - mmdet - INFO - Iter [333/17500] lr: 2.326e-04, eta: 9:35:40, time: 1.583, data_time: 0.077, memory: 49164, loss_cls_0: 0.7341, loss_box_0: 1.6146, loss_cns_0: 0.6386, loss_yns_0: 0.1429, loss_cls_1: 0.7772, loss_box_1: 1.4825, loss_cns_1: 0.6610, loss_yns_1: 0.1423, loss_cls_2: 0.7915, loss_box_2: 1.4540, loss_cns_2: 0.6551, loss_yns_2: 0.1390, loss_cls_3: 0.8019, loss_box_3: 1.4326, loss_cns_3: 0.6529, loss_yns_3: 0.1382, loss_cls_4: 0.8115, loss_box_4: 1.4289, loss_cns_4: 0.6597, loss_yns_4: 0.1412, loss_cls_5: 0.8200, loss_box_5: 1.4371, loss_cns_5: 0.6675, loss_yns_5: 0.1406, loss_cls_dn_0: 0.1481, loss_box_dn_0: 0.7257, loss_cls_dn_1: 0.1001, loss_box_dn_1: 0.6567, loss_cls_dn_2: 0.1008, loss_box_dn_2: 0.6447, loss_cls_dn_3: 0.1015, loss_box_dn_3: 0.6454, loss_cls_dn_4: 0.1004, loss_box_dn_4: 0.6438, loss_cls_dn_5: 0.1020, loss_box_dn_5: 0.6501, loss_dense_depth: 0.7268, loss: 23.7112, grad_norm: 41.5679 -2025-11-12 14:39:52,036 - mmdet - INFO - Iter [334/17500] lr: 2.330e-04, eta: 9:35:18, time: 1.623, data_time: 0.080, memory: 49164, loss_cls_0: 0.7593, loss_box_0: 1.6377, loss_cns_0: 0.6358, loss_yns_0: 0.1432, loss_cls_1: 0.7993, loss_box_1: 1.4926, loss_cns_1: 0.6567, loss_yns_1: 0.1404, loss_cls_2: 0.8023, loss_box_2: 1.4657, loss_cns_2: 0.6542, loss_yns_2: 0.1403, loss_cls_3: 0.8111, loss_box_3: 1.4649, loss_cns_3: 0.6520, loss_yns_3: 0.1378, loss_cls_4: 0.8167, loss_box_4: 1.4673, loss_cns_4: 0.6607, loss_yns_4: 0.1437, loss_cls_5: 0.8241, loss_box_5: 1.4606, loss_cns_5: 0.6654, loss_yns_5: 0.1395, loss_cls_dn_0: 0.1463, loss_box_dn_0: 0.7212, loss_cls_dn_1: 0.1024, loss_box_dn_1: 0.6567, loss_cls_dn_2: 0.1027, loss_box_dn_2: 0.6460, loss_cls_dn_3: 0.1016, loss_box_dn_3: 0.6507, loss_cls_dn_4: 0.1038, loss_box_dn_4: 0.6539, loss_cls_dn_5: 0.1062, loss_box_dn_5: 0.6588, loss_dense_depth: 0.7192, loss: 23.9409, grad_norm: 34.9920 -2025-11-12 14:39:53,607 - mmdet - INFO - Iter [335/17500] lr: 2.334e-04, eta: 9:34:54, time: 1.570, data_time: 0.074, memory: 49164, loss_cls_0: 0.7394, loss_box_0: 1.6287, loss_cns_0: 0.6359, loss_yns_0: 0.1439, loss_cls_1: 0.8006, loss_box_1: 1.4907, loss_cns_1: 0.6603, loss_yns_1: 0.1387, loss_cls_2: 0.8059, loss_box_2: 1.4687, loss_cns_2: 0.6611, loss_yns_2: 0.1404, loss_cls_3: 0.8074, loss_box_3: 1.4624, loss_cns_3: 0.6602, loss_yns_3: 0.1386, loss_cls_4: 0.8108, loss_box_4: 1.4618, loss_cns_4: 0.6619, loss_yns_4: 0.1413, loss_cls_5: 0.8252, loss_box_5: 1.4563, loss_cns_5: 0.6641, loss_yns_5: 0.1383, loss_cls_dn_0: 0.1459, loss_box_dn_0: 0.7221, loss_cls_dn_1: 0.1015, loss_box_dn_1: 0.6563, loss_cls_dn_2: 0.1004, loss_box_dn_2: 0.6495, loss_cls_dn_3: 0.1013, loss_box_dn_3: 0.6485, loss_cls_dn_4: 0.1047, loss_box_dn_4: 0.6532, loss_cls_dn_5: 0.1068, loss_box_dn_5: 0.6573, loss_dense_depth: 0.7027, loss: 23.8925, grad_norm: 35.4856 -2025-11-12 14:39:55,183 - mmdet - INFO - Iter [336/17500] lr: 2.338e-04, eta: 9:34:29, time: 1.573, data_time: 0.071, memory: 49164, loss_cls_0: 0.7285, loss_box_0: 1.6306, loss_cns_0: 0.6313, loss_yns_0: 0.1426, loss_cls_1: 0.8052, loss_box_1: 1.4986, loss_cns_1: 0.6619, loss_yns_1: 0.1409, loss_cls_2: 0.8106, loss_box_2: 1.4704, loss_cns_2: 0.6612, loss_yns_2: 0.1407, loss_cls_3: 0.8031, loss_box_3: 1.4656, loss_cns_3: 0.6624, loss_yns_3: 0.1407, loss_cls_4: 0.8128, loss_box_4: 1.4663, loss_cns_4: 0.6636, loss_yns_4: 0.1409, loss_cls_5: 0.8279, loss_box_5: 1.4791, loss_cns_5: 0.6628, loss_yns_5: 0.1399, loss_cls_dn_0: 0.1403, loss_box_dn_0: 0.7196, loss_cls_dn_1: 0.0996, loss_box_dn_1: 0.6543, loss_cls_dn_2: 0.0997, loss_box_dn_2: 0.6437, loss_cls_dn_3: 0.1005, loss_box_dn_3: 0.6438, loss_cls_dn_4: 0.1008, loss_box_dn_4: 0.6485, loss_cls_dn_5: 0.1026, loss_box_dn_5: 0.6546, loss_dense_depth: 0.7055, loss: 23.9009, grad_norm: 36.0420 -2025-11-12 14:39:56,748 - mmdet - INFO - Iter [337/17500] lr: 2.342e-04, eta: 9:34:05, time: 1.568, data_time: 0.073, memory: 49164, loss_cls_0: 0.7422, loss_box_0: 1.6218, loss_cns_0: 0.6329, loss_yns_0: 0.1410, loss_cls_1: 0.8175, loss_box_1: 1.5015, loss_cns_1: 0.6657, loss_yns_1: 0.1394, loss_cls_2: 0.8392, loss_box_2: 1.4763, loss_cns_2: 0.6639, loss_yns_2: 0.1370, loss_cls_3: 0.8327, loss_box_3: 1.4864, loss_cns_3: 0.6643, loss_yns_3: 0.1388, loss_cls_4: 0.8357, loss_box_4: 1.4653, loss_cns_4: 0.6653, loss_yns_4: 0.1394, loss_cls_5: 0.8292, loss_box_5: 1.4553, loss_cns_5: 0.6663, loss_yns_5: 0.1403, loss_cls_dn_0: 0.1458, loss_box_dn_0: 0.7142, loss_cls_dn_1: 0.1046, loss_box_dn_1: 0.6399, loss_cls_dn_2: 0.1060, loss_box_dn_2: 0.6285, loss_cls_dn_3: 0.1066, loss_box_dn_3: 0.6351, loss_cls_dn_4: 0.1084, loss_box_dn_4: 0.6288, loss_cls_dn_5: 0.1071, loss_box_dn_5: 0.6260, loss_dense_depth: 0.6919, loss: 23.9404, grad_norm: 42.5338 -2025-11-12 14:39:58,321 - mmdet - INFO - Iter [338/17500] lr: 2.346e-04, eta: 9:33:41, time: 1.574, data_time: 0.073, memory: 49164, loss_cls_0: 0.7342, loss_box_0: 1.5937, loss_cns_0: 0.6344, loss_yns_0: 0.1403, loss_cls_1: 0.8187, loss_box_1: 1.4718, loss_cns_1: 0.6655, loss_yns_1: 0.1365, loss_cls_2: 0.8337, loss_box_2: 1.4503, loss_cns_2: 0.6624, loss_yns_2: 0.1368, loss_cls_3: 0.8336, loss_box_3: 1.4446, loss_cns_3: 0.6617, loss_yns_3: 0.1373, loss_cls_4: 0.8361, loss_box_4: 1.4503, loss_cns_4: 0.6664, loss_yns_4: 0.1397, loss_cls_5: 0.8421, loss_box_5: 1.4384, loss_cns_5: 0.6670, loss_yns_5: 0.1389, loss_cls_dn_0: 0.1442, loss_box_dn_0: 0.7300, loss_cls_dn_1: 0.1061, loss_box_dn_1: 0.6307, loss_cls_dn_2: 0.1064, loss_box_dn_2: 0.6195, loss_cls_dn_3: 0.1058, loss_box_dn_3: 0.6190, loss_cls_dn_4: 0.1081, loss_box_dn_4: 0.6214, loss_cls_dn_5: 0.1098, loss_box_dn_5: 0.6218, loss_dense_depth: 0.6931, loss: 23.7502, grad_norm: 28.7069 -2025-11-12 14:39:59,886 - mmdet - INFO - Iter [339/17500] lr: 2.350e-04, eta: 9:33:17, time: 1.566, data_time: 0.068, memory: 49164, loss_cls_0: 0.7316, loss_box_0: 1.6218, loss_cns_0: 0.6335, loss_yns_0: 0.1423, loss_cls_1: 0.8143, loss_box_1: 1.5043, loss_cns_1: 0.6580, loss_yns_1: 0.1375, loss_cls_2: 0.8267, loss_box_2: 1.5009, loss_cns_2: 0.6649, loss_yns_2: 0.1376, loss_cls_3: 0.8267, loss_box_3: 1.5173, loss_cns_3: 0.6639, loss_yns_3: 0.1360, loss_cls_4: 0.8354, loss_box_4: 1.5110, loss_cns_4: 0.6566, loss_yns_4: 0.1367, loss_cls_5: 0.8426, loss_box_5: 1.4841, loss_cns_5: 0.6586, loss_yns_5: 0.1346, loss_cls_dn_0: 0.1451, loss_box_dn_0: 0.7267, loss_cls_dn_1: 0.1032, loss_box_dn_1: 0.6527, loss_cls_dn_2: 0.1024, loss_box_dn_2: 0.6515, loss_cls_dn_3: 0.1009, loss_box_dn_3: 0.6594, loss_cls_dn_4: 0.1060, loss_box_dn_4: 0.6660, loss_cls_dn_5: 0.1054, loss_box_dn_5: 0.6637, loss_dense_depth: 0.7529, loss: 24.2130, grad_norm: 62.0650 -2025-11-12 14:40:01,475 - mmdet - INFO - Iter [340/17500] lr: 2.354e-04, eta: 9:32:54, time: 1.588, data_time: 0.072, memory: 49164, loss_cls_0: 0.7786, loss_box_0: 1.6261, loss_cns_0: 0.6345, loss_yns_0: 0.1439, loss_cls_1: 0.8427, loss_box_1: 1.5450, loss_cns_1: 0.6601, loss_yns_1: 0.1425, loss_cls_2: 0.8592, loss_box_2: 1.4969, loss_cns_2: 0.6637, loss_yns_2: 0.1401, loss_cls_3: 0.8575, loss_box_3: 1.5170, loss_cns_3: 0.6664, loss_yns_3: 0.1393, loss_cls_4: 0.8664, loss_box_4: 1.4943, loss_cns_4: 0.6607, loss_yns_4: 0.1405, loss_cls_5: 0.8632, loss_box_5: 1.4772, loss_cns_5: 0.6595, loss_yns_5: 0.1421, loss_cls_dn_0: 0.1445, loss_box_dn_0: 0.7236, loss_cls_dn_1: 0.1055, loss_box_dn_1: 0.6655, loss_cls_dn_2: 0.1053, loss_box_dn_2: 0.6530, loss_cls_dn_3: 0.1049, loss_box_dn_3: 0.6594, loss_cls_dn_4: 0.1110, loss_box_dn_4: 0.6637, loss_cls_dn_5: 0.1098, loss_box_dn_5: 0.6676, loss_dense_depth: 0.7213, loss: 24.4525, grad_norm: 42.8875 -2025-11-12 14:40:03,139 - mmdet - INFO - Iter [341/17500] lr: 2.358e-04, eta: 9:32:35, time: 1.664, data_time: 0.107, memory: 49164, loss_cls_0: 0.7594, loss_box_0: 1.6344, loss_cns_0: 0.6299, loss_yns_0: 0.1415, loss_cls_1: 0.8160, loss_box_1: 1.5866, loss_cns_1: 0.6531, loss_yns_1: 0.1387, loss_cls_2: 0.8319, loss_box_2: 1.5582, loss_cns_2: 0.6496, loss_yns_2: 0.1376, loss_cls_3: 0.8357, loss_box_3: 1.5528, loss_cns_3: 0.6563, loss_yns_3: 0.1370, loss_cls_4: 0.8469, loss_box_4: 1.5731, loss_cns_4: 0.6542, loss_yns_4: 0.1361, loss_cls_5: 0.8508, loss_box_5: 1.6090, loss_cns_5: 0.6496, loss_yns_5: 0.1377, loss_cls_dn_0: 0.1469, loss_box_dn_0: 0.7274, loss_cls_dn_1: 0.1044, loss_box_dn_1: 0.6687, loss_cls_dn_2: 0.1033, loss_box_dn_2: 0.6666, loss_cls_dn_3: 0.1034, loss_box_dn_3: 0.6559, loss_cls_dn_4: 0.1070, loss_box_dn_4: 0.6717, loss_cls_dn_5: 0.1067, loss_box_dn_5: 0.6936, loss_dense_depth: 0.8033, loss: 24.7348, grad_norm: 60.9193 -2025-11-12 14:40:04,758 - mmdet - INFO - Iter [342/17500] lr: 2.362e-04, eta: 9:32:13, time: 1.617, data_time: 0.100, memory: 49164, loss_cls_0: 0.7571, loss_box_0: 1.5907, loss_cns_0: 0.6284, loss_yns_0: 0.1411, loss_cls_1: 0.7978, loss_box_1: 1.6063, loss_cns_1: 0.6473, loss_yns_1: 0.1412, loss_cls_2: 0.8228, loss_box_2: 1.5974, loss_cns_2: 0.6449, loss_yns_2: 0.1414, loss_cls_3: 0.8404, loss_box_3: 1.5695, loss_cns_3: 0.6529, loss_yns_3: 0.1421, loss_cls_4: 0.8345, loss_box_4: 1.5808, loss_cns_4: 0.6504, loss_yns_4: 0.1397, loss_cls_5: 0.8514, loss_box_5: 1.6115, loss_cns_5: 0.6470, loss_yns_5: 0.1404, loss_cls_dn_0: 0.1406, loss_box_dn_0: 0.7205, loss_cls_dn_1: 0.0997, loss_box_dn_1: 0.6624, loss_cls_dn_2: 0.0991, loss_box_dn_2: 0.6658, loss_cls_dn_3: 0.1000, loss_box_dn_3: 0.6468, loss_cls_dn_4: 0.1008, loss_box_dn_4: 0.6547, loss_cls_dn_5: 0.1023, loss_box_dn_5: 0.6729, loss_dense_depth: 0.7182, loss: 24.5608, grad_norm: 60.6281 -2025-11-12 14:40:06,348 - mmdet - INFO - Iter [343/17500] lr: 2.366e-04, eta: 9:31:51, time: 1.590, data_time: 0.084, memory: 49164, loss_cls_0: 0.7720, loss_box_0: 1.6300, loss_cns_0: 0.6253, loss_yns_0: 0.1419, loss_cls_1: 0.8189, loss_box_1: 1.6015, loss_cns_1: 0.6504, loss_yns_1: 0.1408, loss_cls_2: 0.8241, loss_box_2: 1.5877, loss_cns_2: 0.6588, loss_yns_2: 0.1425, loss_cls_3: 0.8332, loss_box_3: 1.5673, loss_cns_3: 0.6562, loss_yns_3: 0.1436, loss_cls_4: 0.8339, loss_box_4: 1.5611, loss_cns_4: 0.6542, loss_yns_4: 0.1411, loss_cls_5: 0.8478, loss_box_5: 1.5539, loss_cns_5: 0.6552, loss_yns_5: 0.1430, loss_cls_dn_0: 0.1398, loss_box_dn_0: 0.7454, loss_cls_dn_1: 0.1020, loss_box_dn_1: 0.6553, loss_cls_dn_2: 0.1005, loss_box_dn_2: 0.6502, loss_cls_dn_3: 0.0996, loss_box_dn_3: 0.6437, loss_cls_dn_4: 0.1023, loss_box_dn_4: 0.6448, loss_cls_dn_5: 0.1032, loss_box_dn_5: 0.6408, loss_dense_depth: 0.7542, loss: 24.5665, grad_norm: 35.7767 -2025-11-12 14:40:07,940 - mmdet - INFO - Iter [344/17500] lr: 2.370e-04, eta: 9:31:28, time: 1.587, data_time: 0.076, memory: 49164, loss_cls_0: 0.7725, loss_box_0: 1.6291, loss_cns_0: 0.6215, loss_yns_0: 0.1382, loss_cls_1: 0.8196, loss_box_1: 1.5536, loss_cns_1: 0.6513, loss_yns_1: 0.1363, loss_cls_2: 0.8369, loss_box_2: 1.5467, loss_cns_2: 0.6583, loss_yns_2: 0.1366, loss_cls_3: 0.8366, loss_box_3: 1.5471, loss_cns_3: 0.6539, loss_yns_3: 0.1361, loss_cls_4: 0.8467, loss_box_4: 1.5387, loss_cns_4: 0.6531, loss_yns_4: 0.1371, loss_cls_5: 0.8526, loss_box_5: 1.5174, loss_cns_5: 0.6529, loss_yns_5: 0.1359, loss_cls_dn_0: 0.1400, loss_box_dn_0: 0.7430, loss_cls_dn_1: 0.1008, loss_box_dn_1: 0.6481, loss_cls_dn_2: 0.1005, loss_box_dn_2: 0.6443, loss_cls_dn_3: 0.0987, loss_box_dn_3: 0.6493, loss_cls_dn_4: 0.0990, loss_box_dn_4: 0.6492, loss_cls_dn_5: 0.0997, loss_box_dn_5: 0.6441, loss_dense_depth: 0.7387, loss: 24.3641, grad_norm: 46.8027 -2025-11-12 14:40:09,527 - mmdet - INFO - Iter [345/17500] lr: 2.374e-04, eta: 9:31:06, time: 1.594, data_time: 0.088, memory: 49164, loss_cls_0: 0.7443, loss_box_0: 1.6273, loss_cns_0: 0.6300, loss_yns_0: 0.1393, loss_cls_1: 0.7887, loss_box_1: 1.5427, loss_cns_1: 0.6526, loss_yns_1: 0.1386, loss_cls_2: 0.8083, loss_box_2: 1.4950, loss_cns_2: 0.6593, loss_yns_2: 0.1376, loss_cls_3: 0.8135, loss_box_3: 1.4823, loss_cns_3: 0.6572, loss_yns_3: 0.1374, loss_cls_4: 0.8194, loss_box_4: 1.4807, loss_cns_4: 0.6565, loss_yns_4: 0.1382, loss_cls_5: 0.8288, loss_box_5: 1.4786, loss_cns_5: 0.6557, loss_yns_5: 0.1369, loss_cls_dn_0: 0.1448, loss_box_dn_0: 0.7201, loss_cls_dn_1: 0.0990, loss_box_dn_1: 0.6388, loss_cls_dn_2: 0.0971, loss_box_dn_2: 0.6224, loss_cls_dn_3: 0.0979, loss_box_dn_3: 0.6220, loss_cls_dn_4: 0.0979, loss_box_dn_4: 0.6249, loss_cls_dn_5: 0.0999, loss_box_dn_5: 0.6282, loss_dense_depth: 0.7211, loss: 23.8630, grad_norm: 30.6117 -2025-11-12 14:40:11,137 - mmdet - INFO - Iter [346/17500] lr: 2.378e-04, eta: 9:30:45, time: 1.604, data_time: 0.105, memory: 49164, loss_cls_0: 0.7425, loss_box_0: 1.6505, loss_cns_0: 0.6295, loss_yns_0: 0.1399, loss_cls_1: 0.7994, loss_box_1: 1.5288, loss_cns_1: 0.6568, loss_yns_1: 0.1388, loss_cls_2: 0.8141, loss_box_2: 1.4749, loss_cns_2: 0.6595, loss_yns_2: 0.1375, loss_cls_3: 0.8252, loss_box_3: 1.4551, loss_cns_3: 0.6594, loss_yns_3: 0.1372, loss_cls_4: 0.8252, loss_box_4: 1.4493, loss_cns_4: 0.6598, loss_yns_4: 0.1372, loss_cls_5: 0.8348, loss_box_5: 1.4351, loss_cns_5: 0.6598, loss_yns_5: 0.1366, loss_cls_dn_0: 0.1464, loss_box_dn_0: 0.7336, loss_cls_dn_1: 0.1018, loss_box_dn_1: 0.6590, loss_cls_dn_2: 0.0994, loss_box_dn_2: 0.6445, loss_cls_dn_3: 0.1032, loss_box_dn_3: 0.6375, loss_cls_dn_4: 0.1030, loss_box_dn_4: 0.6418, loss_cls_dn_5: 0.1073, loss_box_dn_5: 0.6417, loss_dense_depth: 0.7519, loss: 23.9580, grad_norm: 40.3009 -2025-11-12 14:40:12,717 - mmdet - INFO - Iter [347/17500] lr: 2.382e-04, eta: 9:30:22, time: 1.587, data_time: 0.079, memory: 49164, loss_cls_0: 0.7162, loss_box_0: 1.6037, loss_cns_0: 0.6374, loss_yns_0: 0.1398, loss_cls_1: 0.7762, loss_box_1: 1.4737, loss_cns_1: 0.6602, loss_yns_1: 0.1378, loss_cls_2: 0.7811, loss_box_2: 1.4471, loss_cns_2: 0.6614, loss_yns_2: 0.1371, loss_cls_3: 0.7968, loss_box_3: 1.4233, loss_cns_3: 0.6615, loss_yns_3: 0.1357, loss_cls_4: 0.8033, loss_box_4: 1.4152, loss_cns_4: 0.6603, loss_yns_4: 0.1356, loss_cls_5: 0.8098, loss_box_5: 1.4259, loss_cns_5: 0.6603, loss_yns_5: 0.1361, loss_cls_dn_0: 0.1355, loss_box_dn_0: 0.7124, loss_cls_dn_1: 0.0976, loss_box_dn_1: 0.6505, loss_cls_dn_2: 0.0945, loss_box_dn_2: 0.6384, loss_cls_dn_3: 0.0966, loss_box_dn_3: 0.6301, loss_cls_dn_4: 0.0953, loss_box_dn_4: 0.6311, loss_cls_dn_5: 0.0983, loss_box_dn_5: 0.6393, loss_dense_depth: 0.7228, loss: 23.4780, grad_norm: 26.0615 -2025-11-12 14:40:14,320 - mmdet - INFO - Iter [348/17500] lr: 2.386e-04, eta: 9:30:01, time: 1.597, data_time: 0.090, memory: 49164, loss_cls_0: 0.7376, loss_box_0: 1.6258, loss_cns_0: 0.6327, loss_yns_0: 0.1397, loss_cls_1: 0.7972, loss_box_1: 1.4634, loss_cns_1: 0.6632, loss_yns_1: 0.1366, loss_cls_2: 0.8049, loss_box_2: 1.4188, loss_cns_2: 0.6629, loss_yns_2: 0.1359, loss_cls_3: 0.8094, loss_box_3: 1.4143, loss_cns_3: 0.6612, loss_yns_3: 0.1353, loss_cls_4: 0.8131, loss_box_4: 1.4061, loss_cns_4: 0.6603, loss_yns_4: 0.1353, loss_cls_5: 0.8205, loss_box_5: 1.4248, loss_cns_5: 0.6594, loss_yns_5: 0.1361, loss_cls_dn_0: 0.1373, loss_box_dn_0: 0.7323, loss_cls_dn_1: 0.0988, loss_box_dn_1: 0.6545, loss_cls_dn_2: 0.0974, loss_box_dn_2: 0.6357, loss_cls_dn_3: 0.0982, loss_box_dn_3: 0.6364, loss_cls_dn_4: 0.0988, loss_box_dn_4: 0.6362, loss_cls_dn_5: 0.0989, loss_box_dn_5: 0.6440, loss_dense_depth: 0.8066, loss: 23.6697, grad_norm: 43.8782 -2025-11-12 14:40:15,886 - mmdet - INFO - Iter [349/17500] lr: 2.390e-04, eta: 9:29:38, time: 1.572, data_time: 0.083, memory: 49164, loss_cls_0: 0.7367, loss_box_0: 1.6308, loss_cns_0: 0.6345, loss_yns_0: 0.1423, loss_cls_1: 0.7990, loss_box_1: 1.4556, loss_cns_1: 0.6620, loss_yns_1: 0.1377, loss_cls_2: 0.8085, loss_box_2: 1.4021, loss_cns_2: 0.6633, loss_yns_2: 0.1368, loss_cls_3: 0.8094, loss_box_3: 1.3990, loss_cns_3: 0.6613, loss_yns_3: 0.1368, loss_cls_4: 0.8167, loss_box_4: 1.3919, loss_cns_4: 0.6609, loss_yns_4: 0.1365, loss_cls_5: 0.8149, loss_box_5: 1.3940, loss_cns_5: 0.6629, loss_yns_5: 0.1368, loss_cls_dn_0: 0.1407, loss_box_dn_0: 0.7245, loss_cls_dn_1: 0.1015, loss_box_dn_1: 0.6539, loss_cls_dn_2: 0.1017, loss_box_dn_2: 0.6305, loss_cls_dn_3: 0.1033, loss_box_dn_3: 0.6282, loss_cls_dn_4: 0.1030, loss_box_dn_4: 0.6265, loss_cls_dn_5: 0.1026, loss_box_dn_5: 0.6315, loss_dense_depth: 0.7842, loss: 23.5626, grad_norm: 33.3339 -2025-11-12 14:40:17,487 - mmdet - INFO - Iter [350/17500] lr: 2.394e-04, eta: 9:29:17, time: 1.598, data_time: 0.083, memory: 49164, loss_cls_0: 0.7099, loss_box_0: 1.5936, loss_cns_0: 0.6404, loss_yns_0: 0.1431, loss_cls_1: 0.7774, loss_box_1: 1.4211, loss_cns_1: 0.6655, loss_yns_1: 0.1386, loss_cls_2: 0.7870, loss_box_2: 1.3922, loss_cns_2: 0.6666, loss_yns_2: 0.1370, loss_cls_3: 0.8001, loss_box_3: 1.3959, loss_cns_3: 0.6663, loss_yns_3: 0.1361, loss_cls_4: 0.8008, loss_box_4: 1.4038, loss_cns_4: 0.6661, loss_yns_4: 0.1366, loss_cls_5: 0.8052, loss_box_5: 1.4027, loss_cns_5: 0.6646, loss_yns_5: 0.1362, loss_cls_dn_0: 0.1320, loss_box_dn_0: 0.7111, loss_cls_dn_1: 0.1000, loss_box_dn_1: 0.6296, loss_cls_dn_2: 0.0989, loss_box_dn_2: 0.6226, loss_cls_dn_3: 0.0994, loss_box_dn_3: 0.6235, loss_cls_dn_4: 0.0992, loss_box_dn_4: 0.6322, loss_cls_dn_5: 0.1000, loss_box_dn_5: 0.6368, loss_dense_depth: 0.7941, loss: 23.3661, grad_norm: 39.9991 -2025-11-12 14:40:19,093 - mmdet - INFO - Iter [351/17500] lr: 2.398e-04, eta: 9:28:56, time: 1.607, data_time: 0.079, memory: 49164, loss_cls_0: 0.7337, loss_box_0: 1.6123, loss_cns_0: 0.6351, loss_yns_0: 0.1447, loss_cls_1: 0.8171, loss_box_1: 1.4163, loss_cns_1: 0.6662, loss_yns_1: 0.1434, loss_cls_2: 0.8251, loss_box_2: 1.3979, loss_cns_2: 0.6660, loss_yns_2: 0.1410, loss_cls_3: 0.8479, loss_box_3: 1.3813, loss_cns_3: 0.6691, loss_yns_3: 0.1409, loss_cls_4: 0.8348, loss_box_4: 1.3813, loss_cns_4: 0.6662, loss_yns_4: 0.1416, loss_cls_5: 0.8498, loss_box_5: 1.3810, loss_cns_5: 0.6648, loss_yns_5: 0.1410, loss_cls_dn_0: 0.1369, loss_box_dn_0: 0.7171, loss_cls_dn_1: 0.0984, loss_box_dn_1: 0.6355, loss_cls_dn_2: 0.0979, loss_box_dn_2: 0.6337, loss_cls_dn_3: 0.1002, loss_box_dn_3: 0.6290, loss_cls_dn_4: 0.1017, loss_box_dn_4: 0.6356, loss_cls_dn_5: 0.1026, loss_box_dn_5: 0.6390, loss_dense_depth: 0.7363, loss: 23.5626, grad_norm: 40.6900 -2025-11-12 14:40:20,669 - mmdet - INFO - Iter [352/17500] lr: 2.402e-04, eta: 9:28:33, time: 1.570, data_time: 0.075, memory: 49164, loss_cls_0: 0.7367, loss_box_0: 1.5997, loss_cns_0: 0.6325, loss_yns_0: 0.1464, loss_cls_1: 0.8067, loss_box_1: 1.4333, loss_cns_1: 0.6626, loss_yns_1: 0.1441, loss_cls_2: 0.8164, loss_box_2: 1.4220, loss_cns_2: 0.6640, loss_yns_2: 0.1448, loss_cls_3: 0.8298, loss_box_3: 1.4008, loss_cns_3: 0.6642, loss_yns_3: 0.1437, loss_cls_4: 0.8290, loss_box_4: 1.3944, loss_cns_4: 0.6636, loss_yns_4: 0.1447, loss_cls_5: 0.8383, loss_box_5: 1.3868, loss_cns_5: 0.6644, loss_yns_5: 0.1453, loss_cls_dn_0: 0.1401, loss_box_dn_0: 0.7230, loss_cls_dn_1: 0.0978, loss_box_dn_1: 0.6453, loss_cls_dn_2: 0.0984, loss_box_dn_2: 0.6433, loss_cls_dn_3: 0.1001, loss_box_dn_3: 0.6363, loss_cls_dn_4: 0.1005, loss_box_dn_4: 0.6357, loss_cls_dn_5: 0.0998, loss_box_dn_5: 0.6369, loss_dense_depth: 0.7644, loss: 23.6358, grad_norm: 34.0635 -2025-11-12 14:40:22,247 - mmdet - INFO - Iter [353/17500] lr: 2.406e-04, eta: 9:28:11, time: 1.580, data_time: 0.081, memory: 49164, loss_cls_0: 0.7333, loss_box_0: 1.5745, loss_cns_0: 0.6348, loss_yns_0: 0.1458, loss_cls_1: 0.8043, loss_box_1: 1.3872, loss_cns_1: 0.6623, loss_yns_1: 0.1429, loss_cls_2: 0.8283, loss_box_2: 1.3552, loss_cns_2: 0.6613, loss_yns_2: 0.1433, loss_cls_3: 0.8244, loss_box_3: 1.3786, loss_cns_3: 0.6621, loss_yns_3: 0.1436, loss_cls_4: 0.8286, loss_box_4: 1.3704, loss_cns_4: 0.6628, loss_yns_4: 0.1443, loss_cls_5: 0.8296, loss_box_5: 1.3626, loss_cns_5: 0.6622, loss_yns_5: 0.1469, loss_cls_dn_0: 0.1344, loss_box_dn_0: 0.7151, loss_cls_dn_1: 0.0976, loss_box_dn_1: 0.6361, loss_cls_dn_2: 0.0979, loss_box_dn_2: 0.6240, loss_cls_dn_3: 0.0961, loss_box_dn_3: 0.6333, loss_cls_dn_4: 0.0978, loss_box_dn_4: 0.6329, loss_cls_dn_5: 0.0993, loss_box_dn_5: 0.6303, loss_dense_depth: 0.7352, loss: 23.3192, grad_norm: 39.5079 -2025-11-12 14:40:23,880 - mmdet - INFO - Iter [354/17500] lr: 2.410e-04, eta: 9:27:52, time: 1.630, data_time: 0.079, memory: 49164, loss_cls_0: 0.7316, loss_box_0: 1.5777, loss_cns_0: 0.6373, loss_yns_0: 0.1448, loss_cls_1: 0.8066, loss_box_1: 1.4075, loss_cns_1: 0.6615, loss_yns_1: 0.1430, loss_cls_2: 0.8304, loss_box_2: 1.3940, loss_cns_2: 0.6628, loss_yns_2: 0.1434, loss_cls_3: 0.8410, loss_box_3: 1.3903, loss_cns_3: 0.6631, loss_yns_3: 0.1447, loss_cls_4: 0.8334, loss_box_4: 1.3807, loss_cns_4: 0.6628, loss_yns_4: 0.1448, loss_cls_5: 0.8356, loss_box_5: 1.3732, loss_cns_5: 0.6617, loss_yns_5: 0.1452, loss_cls_dn_0: 0.1303, loss_box_dn_0: 0.7105, loss_cls_dn_1: 0.0958, loss_box_dn_1: 0.6318, loss_cls_dn_2: 0.0945, loss_box_dn_2: 0.6218, loss_cls_dn_3: 0.0930, loss_box_dn_3: 0.6223, loss_cls_dn_4: 0.0938, loss_box_dn_4: 0.6186, loss_cls_dn_5: 0.0966, loss_box_dn_5: 0.6138, loss_dense_depth: 0.7410, loss: 23.3810, grad_norm: 32.9427 -2025-11-12 14:40:25,465 - mmdet - INFO - Iter [355/17500] lr: 2.414e-04, eta: 9:27:31, time: 1.586, data_time: 0.082, memory: 49164, loss_cls_0: 0.7269, loss_box_0: 1.5751, loss_cns_0: 0.6333, loss_yns_0: 0.1441, loss_cls_1: 0.7992, loss_box_1: 1.4256, loss_cns_1: 0.6585, loss_yns_1: 0.1417, loss_cls_2: 0.8116, loss_box_2: 1.4155, loss_cns_2: 0.6587, loss_yns_2: 0.1425, loss_cls_3: 0.8252, loss_box_3: 1.4051, loss_cns_3: 0.6568, loss_yns_3: 0.1417, loss_cls_4: 0.8229, loss_box_4: 1.3983, loss_cns_4: 0.6577, loss_yns_4: 0.1420, loss_cls_5: 0.8315, loss_box_5: 1.4024, loss_cns_5: 0.6574, loss_yns_5: 0.1422, loss_cls_dn_0: 0.1384, loss_box_dn_0: 0.7248, loss_cls_dn_1: 0.0982, loss_box_dn_1: 0.6285, loss_cls_dn_2: 0.0959, loss_box_dn_2: 0.6201, loss_cls_dn_3: 0.0967, loss_box_dn_3: 0.6229, loss_cls_dn_4: 0.0969, loss_box_dn_4: 0.6240, loss_cls_dn_5: 0.0972, loss_box_dn_5: 0.6280, loss_dense_depth: 0.7092, loss: 23.3968, grad_norm: 29.7807 -2025-11-12 14:40:27,038 - mmdet - INFO - Iter [356/17500] lr: 2.418e-04, eta: 9:27:09, time: 1.580, data_time: 0.081, memory: 49164, loss_cls_0: 0.7595, loss_box_0: 1.5615, loss_cns_0: 0.6331, loss_yns_0: 0.1463, loss_cls_1: 0.8190, loss_box_1: 1.4290, loss_cns_1: 0.6592, loss_yns_1: 0.1447, loss_cls_2: 0.8318, loss_box_2: 1.4035, loss_cns_2: 0.6599, loss_yns_2: 0.1436, loss_cls_3: 0.8370, loss_box_3: 1.4209, loss_cns_3: 0.6576, loss_yns_3: 0.1437, loss_cls_4: 0.8496, loss_box_4: 1.3943, loss_cns_4: 0.6562, loss_yns_4: 0.1438, loss_cls_5: 0.8506, loss_box_5: 1.3960, loss_cns_5: 0.6585, loss_yns_5: 0.1468, loss_cls_dn_0: 0.1408, loss_box_dn_0: 0.7220, loss_cls_dn_1: 0.0993, loss_box_dn_1: 0.6222, loss_cls_dn_2: 0.0998, loss_box_dn_2: 0.6100, loss_cls_dn_3: 0.1006, loss_box_dn_3: 0.6223, loss_cls_dn_4: 0.1016, loss_box_dn_4: 0.6286, loss_cls_dn_5: 0.1019, loss_box_dn_5: 0.6372, loss_dense_depth: 0.7321, loss: 23.5646, grad_norm: 36.5598 -2025-11-12 14:40:28,616 - mmdet - INFO - Iter [357/17500] lr: 2.422e-04, eta: 9:26:47, time: 1.572, data_time: 0.075, memory: 49164, loss_cls_0: 0.7234, loss_box_0: 1.5605, loss_cns_0: 0.6325, loss_yns_0: 0.1424, loss_cls_1: 0.7896, loss_box_1: 1.4765, loss_cns_1: 0.6573, loss_yns_1: 0.1423, loss_cls_2: 0.8065, loss_box_2: 1.4353, loss_cns_2: 0.6573, loss_yns_2: 0.1418, loss_cls_3: 0.8079, loss_box_3: 1.4359, loss_cns_3: 0.6580, loss_yns_3: 0.1421, loss_cls_4: 0.8188, loss_box_4: 1.4278, loss_cns_4: 0.6600, loss_yns_4: 0.1413, loss_cls_5: 0.8245, loss_box_5: 1.4302, loss_cns_5: 0.6591, loss_yns_5: 0.1431, loss_cls_dn_0: 0.1354, loss_box_dn_0: 0.7101, loss_cls_dn_1: 0.0957, loss_box_dn_1: 0.6568, loss_cls_dn_2: 0.0954, loss_box_dn_2: 0.6380, loss_cls_dn_3: 0.0962, loss_box_dn_3: 0.6466, loss_cls_dn_4: 0.0969, loss_box_dn_4: 0.6559, loss_cls_dn_5: 0.0990, loss_box_dn_5: 0.6682, loss_dense_depth: 0.7119, loss: 23.6202, grad_norm: 32.3502 -2025-11-12 14:40:30,184 - mmdet - INFO - Iter [358/17500] lr: 2.426e-04, eta: 9:26:26, time: 1.572, data_time: 0.081, memory: 49164, loss_cls_0: 0.7267, loss_box_0: 1.5446, loss_cns_0: 0.6359, loss_yns_0: 0.1412, loss_cls_1: 0.7906, loss_box_1: 1.4413, loss_cns_1: 0.6596, loss_yns_1: 0.1414, loss_cls_2: 0.7959, loss_box_2: 1.4012, loss_cns_2: 0.6598, loss_yns_2: 0.1423, loss_cls_3: 0.8023, loss_box_3: 1.3926, loss_cns_3: 0.6620, loss_yns_3: 0.1432, loss_cls_4: 0.8087, loss_box_4: 1.3953, loss_cns_4: 0.6634, loss_yns_4: 0.1412, loss_cls_5: 0.8253, loss_box_5: 1.3944, loss_cns_5: 0.6613, loss_yns_5: 0.1424, loss_cls_dn_0: 0.1383, loss_box_dn_0: 0.7162, loss_cls_dn_1: 0.1004, loss_box_dn_1: 0.6641, loss_cls_dn_2: 0.1011, loss_box_dn_2: 0.6441, loss_cls_dn_3: 0.1019, loss_box_dn_3: 0.6440, loss_cls_dn_4: 0.1021, loss_box_dn_4: 0.6475, loss_cls_dn_5: 0.1029, loss_box_dn_5: 0.6530, loss_dense_depth: 0.7007, loss: 23.4290, grad_norm: 32.9683 -2025-11-12 14:40:31,767 - mmdet - INFO - Iter [359/17500] lr: 2.429e-04, eta: 9:26:04, time: 1.581, data_time: 0.081, memory: 49164, loss_cls_0: 0.7161, loss_box_0: 1.5199, loss_cns_0: 0.6395, loss_yns_0: 0.1367, loss_cls_1: 0.7802, loss_box_1: 1.3944, loss_cns_1: 0.6643, loss_yns_1: 0.1390, loss_cls_2: 0.7890, loss_box_2: 1.3805, loss_cns_2: 0.6634, loss_yns_2: 0.1385, loss_cls_3: 0.7962, loss_box_3: 1.3641, loss_cns_3: 0.6653, loss_yns_3: 0.1388, loss_cls_4: 0.7934, loss_box_4: 1.3561, loss_cns_4: 0.6638, loss_yns_4: 0.1388, loss_cls_5: 0.8109, loss_box_5: 1.3528, loss_cns_5: 0.6643, loss_yns_5: 0.1385, loss_cls_dn_0: 0.1399, loss_box_dn_0: 0.7093, loss_cls_dn_1: 0.1030, loss_box_dn_1: 0.6341, loss_cls_dn_2: 0.1035, loss_box_dn_2: 0.6241, loss_cls_dn_3: 0.1031, loss_box_dn_3: 0.6226, loss_cls_dn_4: 0.1027, loss_box_dn_4: 0.6164, loss_cls_dn_5: 0.1024, loss_box_dn_5: 0.6192, loss_dense_depth: 0.6818, loss: 23.0069, grad_norm: 24.9961 -2025-11-12 14:40:33,355 - mmdet - INFO - Iter [360/17500] lr: 2.433e-04, eta: 9:25:44, time: 1.591, data_time: 0.081, memory: 49164, loss_cls_0: 0.7230, loss_box_0: 1.5399, loss_cns_0: 0.6366, loss_yns_0: 0.1368, loss_cls_1: 0.7854, loss_box_1: 1.4120, loss_cns_1: 0.6620, loss_yns_1: 0.1380, loss_cls_2: 0.7958, loss_box_2: 1.4049, loss_cns_2: 0.6621, loss_yns_2: 0.1385, loss_cls_3: 0.7957, loss_box_3: 1.4010, loss_cns_3: 0.6647, loss_yns_3: 0.1361, loss_cls_4: 0.7979, loss_box_4: 1.4004, loss_cns_4: 0.6641, loss_yns_4: 0.1378, loss_cls_5: 0.8093, loss_box_5: 1.4115, loss_cns_5: 0.6657, loss_yns_5: 0.1398, loss_cls_dn_0: 0.1354, loss_box_dn_0: 0.7115, loss_cls_dn_1: 0.0985, loss_box_dn_1: 0.6201, loss_cls_dn_2: 0.0976, loss_box_dn_2: 0.6126, loss_cls_dn_3: 0.0970, loss_box_dn_3: 0.6109, loss_cls_dn_4: 0.0992, loss_box_dn_4: 0.6145, loss_cls_dn_5: 0.0996, loss_box_dn_5: 0.6203, loss_dense_depth: 0.7095, loss: 23.1860, grad_norm: 33.5784 -2025-11-12 14:40:35,015 - mmdet - INFO - Iter [361/17500] lr: 2.437e-04, eta: 9:25:27, time: 1.663, data_time: 0.106, memory: 49164, loss_cls_0: 0.7424, loss_box_0: 1.5453, loss_cns_0: 0.6381, loss_yns_0: 0.1430, loss_cls_1: 0.7935, loss_box_1: 1.4266, loss_cns_1: 0.6639, loss_yns_1: 0.1422, loss_cls_2: 0.8027, loss_box_2: 1.4169, loss_cns_2: 0.6644, loss_yns_2: 0.1416, loss_cls_3: 0.8039, loss_box_3: 1.4076, loss_cns_3: 0.6649, loss_yns_3: 0.1419, loss_cls_4: 0.8178, loss_box_4: 1.4111, loss_cns_4: 0.6670, loss_yns_4: 0.1412, loss_cls_5: 0.8231, loss_box_5: 1.4100, loss_cns_5: 0.6652, loss_yns_5: 0.1421, loss_cls_dn_0: 0.1339, loss_box_dn_0: 0.7144, loss_cls_dn_1: 0.0978, loss_box_dn_1: 0.6273, loss_cls_dn_2: 0.0977, loss_box_dn_2: 0.6236, loss_cls_dn_3: 0.0973, loss_box_dn_3: 0.6217, loss_cls_dn_4: 0.1002, loss_box_dn_4: 0.6312, loss_cls_dn_5: 0.1006, loss_box_dn_5: 0.6390, loss_dense_depth: 0.6963, loss: 23.3976, grad_norm: 35.8613 -2025-11-12 14:40:36,619 - mmdet - INFO - Iter [362/17500] lr: 2.441e-04, eta: 9:25:07, time: 1.596, data_time: 0.105, memory: 49164, loss_cls_0: 0.7441, loss_box_0: 1.5867, loss_cns_0: 0.6365, loss_yns_0: 0.1433, loss_cls_1: 0.8025, loss_box_1: 1.4406, loss_cns_1: 0.6642, loss_yns_1: 0.1405, loss_cls_2: 0.8146, loss_box_2: 1.4129, loss_cns_2: 0.6633, loss_yns_2: 0.1410, loss_cls_3: 0.8173, loss_box_3: 1.4085, loss_cns_3: 0.6615, loss_yns_3: 0.1414, loss_cls_4: 0.8159, loss_box_4: 1.4180, loss_cns_4: 0.6634, loss_yns_4: 0.1402, loss_cls_5: 0.8310, loss_box_5: 1.4065, loss_cns_5: 0.6665, loss_yns_5: 0.1413, loss_cls_dn_0: 0.1370, loss_box_dn_0: 0.7161, loss_cls_dn_1: 0.0979, loss_box_dn_1: 0.6359, loss_cls_dn_2: 0.0968, loss_box_dn_2: 0.6281, loss_cls_dn_3: 0.0966, loss_box_dn_3: 0.6334, loss_cls_dn_4: 0.0972, loss_box_dn_4: 0.6368, loss_cls_dn_5: 0.0986, loss_box_dn_5: 0.6429, loss_dense_depth: 0.7518, loss: 23.5709, grad_norm: 30.0778 -2025-11-12 14:40:38,211 - mmdet - INFO - Iter [363/17500] lr: 2.445e-04, eta: 9:24:47, time: 1.594, data_time: 0.085, memory: 49164, loss_cls_0: 0.7404, loss_box_0: 1.5585, loss_cns_0: 0.6334, loss_yns_0: 0.1357, loss_cls_1: 0.8058, loss_box_1: 1.4157, loss_cns_1: 0.6651, loss_yns_1: 0.1382, loss_cls_2: 0.8124, loss_box_2: 1.3949, loss_cns_2: 0.6657, loss_yns_2: 0.1365, loss_cls_3: 0.8189, loss_box_3: 1.3841, loss_cns_3: 0.6654, loss_yns_3: 0.1373, loss_cls_4: 0.8191, loss_box_4: 1.3960, loss_cns_4: 0.6659, loss_yns_4: 0.1372, loss_cls_5: 0.8283, loss_box_5: 1.3866, loss_cns_5: 0.6713, loss_yns_5: 0.1381, loss_cls_dn_0: 0.1300, loss_box_dn_0: 0.7256, loss_cls_dn_1: 0.0926, loss_box_dn_1: 0.6442, loss_cls_dn_2: 0.0907, loss_box_dn_2: 0.6316, loss_cls_dn_3: 0.0911, loss_box_dn_3: 0.6333, loss_cls_dn_4: 0.0921, loss_box_dn_4: 0.6338, loss_cls_dn_5: 0.0934, loss_box_dn_5: 0.6355, loss_dense_depth: 0.7498, loss: 23.3944, grad_norm: 33.3454 -2025-11-12 14:40:39,782 - mmdet - INFO - Iter [364/17500] lr: 2.449e-04, eta: 9:24:26, time: 1.573, data_time: 0.076, memory: 49164, loss_cls_0: 0.7391, loss_box_0: 1.5600, loss_cns_0: 0.6340, loss_yns_0: 0.1358, loss_cls_1: 0.8028, loss_box_1: 1.4600, loss_cns_1: 0.6635, loss_yns_1: 0.1393, loss_cls_2: 0.8066, loss_box_2: 1.4313, loss_cns_2: 0.6639, loss_yns_2: 0.1371, loss_cls_3: 0.8068, loss_box_3: 1.4275, loss_cns_3: 0.6632, loss_yns_3: 0.1370, loss_cls_4: 0.8151, loss_box_4: 1.4224, loss_cns_4: 0.6625, loss_yns_4: 0.1361, loss_cls_5: 0.8168, loss_box_5: 1.4087, loss_cns_5: 0.6613, loss_yns_5: 0.1359, loss_cls_dn_0: 0.1340, loss_box_dn_0: 0.7267, loss_cls_dn_1: 0.0947, loss_box_dn_1: 0.6344, loss_cls_dn_2: 0.0913, loss_box_dn_2: 0.6205, loss_cls_dn_3: 0.0907, loss_box_dn_3: 0.6193, loss_cls_dn_4: 0.0932, loss_box_dn_4: 0.6188, loss_cls_dn_5: 0.0934, loss_box_dn_5: 0.6180, loss_dense_depth: 0.7182, loss: 23.4200, grad_norm: 32.2425 -2025-11-12 14:40:41,355 - mmdet - INFO - Iter [365/17500] lr: 2.453e-04, eta: 9:24:05, time: 1.575, data_time: 0.085, memory: 49164, loss_cls_0: 0.7218, loss_box_0: 1.5791, loss_cns_0: 0.6378, loss_yns_0: 0.1357, loss_cls_1: 0.7937, loss_box_1: 1.4415, loss_cns_1: 0.6647, loss_yns_1: 0.1369, loss_cls_2: 0.7948, loss_box_2: 1.4083, loss_cns_2: 0.6650, loss_yns_2: 0.1352, loss_cls_3: 0.8020, loss_box_3: 1.4004, loss_cns_3: 0.6657, loss_yns_3: 0.1360, loss_cls_4: 0.8043, loss_box_4: 1.3916, loss_cns_4: 0.6653, loss_yns_4: 0.1353, loss_cls_5: 0.8090, loss_box_5: 1.4038, loss_cns_5: 0.6651, loss_yns_5: 0.1359, loss_cls_dn_0: 0.1300, loss_box_dn_0: 0.7225, loss_cls_dn_1: 0.0972, loss_box_dn_1: 0.6196, loss_cls_dn_2: 0.0958, loss_box_dn_2: 0.6042, loss_cls_dn_3: 0.0965, loss_box_dn_3: 0.5992, loss_cls_dn_4: 0.0966, loss_box_dn_4: 0.5977, loss_cls_dn_5: 0.0963, loss_box_dn_5: 0.6015, loss_dense_depth: 0.7058, loss: 23.1918, grad_norm: 26.0820 -2025-11-12 14:40:42,960 - mmdet - INFO - Iter [366/17500] lr: 2.457e-04, eta: 9:23:45, time: 1.602, data_time: 0.108, memory: 49164, loss_cls_0: 0.6956, loss_box_0: 1.5699, loss_cns_0: 0.6380, loss_yns_0: 0.1356, loss_cls_1: 0.7741, loss_box_1: 1.4385, loss_cns_1: 0.6621, loss_yns_1: 0.1342, loss_cls_2: 0.7859, loss_box_2: 1.4153, loss_cns_2: 0.6635, loss_yns_2: 0.1371, loss_cls_3: 0.7931, loss_box_3: 1.4190, loss_cns_3: 0.6627, loss_yns_3: 0.1353, loss_cls_4: 0.7977, loss_box_4: 1.4067, loss_cns_4: 0.6651, loss_yns_4: 0.1352, loss_cls_5: 0.7877, loss_box_5: 1.4011, loss_cns_5: 0.6635, loss_yns_5: 0.1358, loss_cls_dn_0: 0.1236, loss_box_dn_0: 0.7134, loss_cls_dn_1: 0.0925, loss_box_dn_1: 0.6152, loss_cls_dn_2: 0.0935, loss_box_dn_2: 0.6073, loss_cls_dn_3: 0.0959, loss_box_dn_3: 0.6119, loss_cls_dn_4: 0.0957, loss_box_dn_4: 0.6088, loss_cls_dn_5: 0.0941, loss_box_dn_5: 0.6081, loss_dense_depth: 0.7098, loss: 23.1222, grad_norm: 35.7093 -2025-11-12 14:40:44,538 - mmdet - INFO - Iter [367/17500] lr: 2.461e-04, eta: 9:23:25, time: 1.579, data_time: 0.079, memory: 49164, loss_cls_0: 0.7118, loss_box_0: 1.5931, loss_cns_0: 0.6402, loss_yns_0: 0.1361, loss_cls_1: 0.7792, loss_box_1: 1.4408, loss_cns_1: 0.6663, loss_yns_1: 0.1348, loss_cls_2: 0.8084, loss_box_2: 1.4014, loss_cns_2: 0.6648, loss_yns_2: 0.1348, loss_cls_3: 0.8022, loss_box_3: 1.4010, loss_cns_3: 0.6642, loss_yns_3: 0.1349, loss_cls_4: 0.8050, loss_box_4: 1.3937, loss_cns_4: 0.6671, loss_yns_4: 0.1337, loss_cls_5: 0.8033, loss_box_5: 1.3962, loss_cns_5: 0.6664, loss_yns_5: 0.1336, loss_cls_dn_0: 0.1275, loss_box_dn_0: 0.7174, loss_cls_dn_1: 0.0955, loss_box_dn_1: 0.6341, loss_cls_dn_2: 0.0973, loss_box_dn_2: 0.6212, loss_cls_dn_3: 0.0994, loss_box_dn_3: 0.6262, loss_cls_dn_4: 0.0995, loss_box_dn_4: 0.6250, loss_cls_dn_5: 0.0980, loss_box_dn_5: 0.6289, loss_dense_depth: 0.7287, loss: 23.3118, grad_norm: 35.9562 -2025-11-12 14:40:46,117 - mmdet - INFO - Iter [368/17500] lr: 2.465e-04, eta: 9:23:05, time: 1.583, data_time: 0.081, memory: 49164, loss_cls_0: 0.7234, loss_box_0: 1.6060, loss_cns_0: 0.6382, loss_yns_0: 0.1361, loss_cls_1: 0.7729, loss_box_1: 1.4394, loss_cns_1: 0.6651, loss_yns_1: 0.1346, loss_cls_2: 0.7916, loss_box_2: 1.4002, loss_cns_2: 0.6664, loss_yns_2: 0.1347, loss_cls_3: 0.7897, loss_box_3: 1.3776, loss_cns_3: 0.6654, loss_yns_3: 0.1336, loss_cls_4: 0.8020, loss_box_4: 1.3768, loss_cns_4: 0.6673, loss_yns_4: 0.1330, loss_cls_5: 0.7978, loss_box_5: 1.3843, loss_cns_5: 0.6666, loss_yns_5: 0.1312, loss_cls_dn_0: 0.1279, loss_box_dn_0: 0.7196, loss_cls_dn_1: 0.0963, loss_box_dn_1: 0.6438, loss_cls_dn_2: 0.0946, loss_box_dn_2: 0.6269, loss_cls_dn_3: 0.0959, loss_box_dn_3: 0.6243, loss_cls_dn_4: 0.0955, loss_box_dn_4: 0.6255, loss_cls_dn_5: 0.0962, loss_box_dn_5: 0.6323, loss_dense_depth: 0.7089, loss: 23.2215, grad_norm: 27.7761 -2025-11-12 14:40:47,708 - mmdet - INFO - Iter [369/17500] lr: 2.469e-04, eta: 9:22:45, time: 1.591, data_time: 0.084, memory: 49164, loss_cls_0: 0.6871, loss_box_0: 1.5781, loss_cns_0: 0.6395, loss_yns_0: 0.1355, loss_cls_1: 0.7505, loss_box_1: 1.4345, loss_cns_1: 0.6687, loss_yns_1: 0.1336, loss_cls_2: 0.7679, loss_box_2: 1.3764, loss_cns_2: 0.6687, loss_yns_2: 0.1311, loss_cls_3: 0.7792, loss_box_3: 1.3597, loss_cns_3: 0.6680, loss_yns_3: 0.1306, loss_cls_4: 0.7969, loss_box_4: 1.3567, loss_cns_4: 0.6690, loss_yns_4: 0.1312, loss_cls_5: 0.7879, loss_box_5: 1.3575, loss_cns_5: 0.6686, loss_yns_5: 0.1318, loss_cls_dn_0: 0.1192, loss_box_dn_0: 0.7188, loss_cls_dn_1: 0.0946, loss_box_dn_1: 0.6516, loss_cls_dn_2: 0.0949, loss_box_dn_2: 0.6243, loss_cls_dn_3: 0.0973, loss_box_dn_3: 0.6203, loss_cls_dn_4: 0.0969, loss_box_dn_4: 0.6185, loss_cls_dn_5: 0.0991, loss_box_dn_5: 0.6245, loss_dense_depth: 0.6897, loss: 22.9580, grad_norm: 32.9373 -2025-11-12 14:40:49,289 - mmdet - INFO - Iter [370/17500] lr: 2.473e-04, eta: 9:22:25, time: 1.581, data_time: 0.084, memory: 49164, loss_cls_0: 0.7146, loss_box_0: 1.5599, loss_cns_0: 0.6324, loss_yns_0: 0.1338, loss_cls_1: 0.7714, loss_box_1: 1.4323, loss_cns_1: 0.6636, loss_yns_1: 0.1306, loss_cls_2: 0.7883, loss_box_2: 1.3910, loss_cns_2: 0.6618, loss_yns_2: 0.1307, loss_cls_3: 0.8047, loss_box_3: 1.3840, loss_cns_3: 0.6657, loss_yns_3: 0.1302, loss_cls_4: 0.8050, loss_box_4: 1.3819, loss_cns_4: 0.6669, loss_yns_4: 0.1308, loss_cls_5: 0.8246, loss_box_5: 1.3692, loss_cns_5: 0.6664, loss_yns_5: 0.1311, loss_cls_dn_0: 0.1252, loss_box_dn_0: 0.7160, loss_cls_dn_1: 0.0974, loss_box_dn_1: 0.6348, loss_cls_dn_2: 0.1009, loss_box_dn_2: 0.6151, loss_cls_dn_3: 0.1064, loss_box_dn_3: 0.6134, loss_cls_dn_4: 0.1020, loss_box_dn_4: 0.6127, loss_cls_dn_5: 0.1039, loss_box_dn_5: 0.6115, loss_dense_depth: 0.7138, loss: 23.1241, grad_norm: 37.4024 -2025-11-12 14:40:50,910 - mmdet - INFO - Iter [371/17500] lr: 2.477e-04, eta: 9:22:07, time: 1.618, data_time: 0.077, memory: 49164, loss_cls_0: 0.7001, loss_box_0: 1.5339, loss_cns_0: 0.6348, loss_yns_0: 0.1344, loss_cls_1: 0.7566, loss_box_1: 1.4193, loss_cns_1: 0.6564, loss_yns_1: 0.1331, loss_cls_2: 0.7748, loss_box_2: 1.3847, loss_cns_2: 0.6561, loss_yns_2: 0.1337, loss_cls_3: 0.7841, loss_box_3: 1.3979, loss_cns_3: 0.6620, loss_yns_3: 0.1337, loss_cls_4: 0.7824, loss_box_4: 1.3973, loss_cns_4: 0.6621, loss_yns_4: 0.1331, loss_cls_5: 0.7905, loss_box_5: 1.3898, loss_cns_5: 0.6638, loss_yns_5: 0.1337, loss_cls_dn_0: 0.1185, loss_box_dn_0: 0.7163, loss_cls_dn_1: 0.0931, loss_box_dn_1: 0.6135, loss_cls_dn_2: 0.0936, loss_box_dn_2: 0.6019, loss_cls_dn_3: 0.0979, loss_box_dn_3: 0.6038, loss_cls_dn_4: 0.0950, loss_box_dn_4: 0.6045, loss_cls_dn_5: 0.0961, loss_box_dn_5: 0.6071, loss_dense_depth: 0.6808, loss: 22.8702, grad_norm: 31.3516 -2025-11-12 14:40:52,481 - mmdet - INFO - Iter [372/17500] lr: 2.481e-04, eta: 9:21:47, time: 1.570, data_time: 0.077, memory: 49164, loss_cls_0: 0.7234, loss_box_0: 1.5758, loss_cns_0: 0.6367, loss_yns_0: 0.1370, loss_cls_1: 0.7679, loss_box_1: 1.4344, loss_cns_1: 0.6590, loss_yns_1: 0.1366, loss_cls_2: 0.7822, loss_box_2: 1.4198, loss_cns_2: 0.6610, loss_yns_2: 0.1377, loss_cls_3: 0.7956, loss_box_3: 1.4212, loss_cns_3: 0.6610, loss_yns_3: 0.1368, loss_cls_4: 0.7890, loss_box_4: 1.4322, loss_cns_4: 0.6603, loss_yns_4: 0.1365, loss_cls_5: 0.7956, loss_box_5: 1.4305, loss_cns_5: 0.6628, loss_yns_5: 0.1390, loss_cls_dn_0: 0.1234, loss_box_dn_0: 0.7280, loss_cls_dn_1: 0.0927, loss_box_dn_1: 0.6222, loss_cls_dn_2: 0.0917, loss_box_dn_2: 0.6165, loss_cls_dn_3: 0.0937, loss_box_dn_3: 0.6197, loss_cls_dn_4: 0.0933, loss_box_dn_4: 0.6242, loss_cls_dn_5: 0.0944, loss_box_dn_5: 0.6309, loss_dense_depth: 0.7277, loss: 23.2904, grad_norm: 43.6423 -2025-11-12 14:40:54,068 - mmdet - INFO - Iter [373/17500] lr: 2.485e-04, eta: 9:21:27, time: 1.591, data_time: 0.076, memory: 49164, loss_cls_0: 0.7553, loss_box_0: 1.6191, loss_cns_0: 0.6391, loss_yns_0: 0.1398, loss_cls_1: 0.8022, loss_box_1: 1.4745, loss_cns_1: 0.6641, loss_yns_1: 0.1391, loss_cls_2: 0.8184, loss_box_2: 1.4393, loss_cns_2: 0.6676, loss_yns_2: 0.1398, loss_cls_3: 0.8339, loss_box_3: 1.4205, loss_cns_3: 0.6668, loss_yns_3: 0.1378, loss_cls_4: 0.8266, loss_box_4: 1.4242, loss_cns_4: 0.6701, loss_yns_4: 0.1383, loss_cls_5: 0.8297, loss_box_5: 1.4227, loss_cns_5: 0.6676, loss_yns_5: 0.1426, loss_cls_dn_0: 0.1236, loss_box_dn_0: 0.7119, loss_cls_dn_1: 0.0943, loss_box_dn_1: 0.6375, loss_cls_dn_2: 0.0925, loss_box_dn_2: 0.6217, loss_cls_dn_3: 0.0954, loss_box_dn_3: 0.6160, loss_cls_dn_4: 0.0936, loss_box_dn_4: 0.6176, loss_cls_dn_5: 0.0939, loss_box_dn_5: 0.6247, loss_dense_depth: 0.7023, loss: 23.6043, grad_norm: 27.1224 -2025-11-12 14:40:55,683 - mmdet - INFO - Iter [374/17500] lr: 2.489e-04, eta: 9:21:09, time: 1.612, data_time: 0.067, memory: 49164, loss_cls_0: 0.7278, loss_box_0: 1.5925, loss_cns_0: 0.6336, loss_yns_0: 0.1384, loss_cls_1: 0.7847, loss_box_1: 1.4736, loss_cns_1: 0.6610, loss_yns_1: 0.1387, loss_cls_2: 0.8058, loss_box_2: 1.4063, loss_cns_2: 0.6645, loss_yns_2: 0.1371, loss_cls_3: 0.8167, loss_box_3: 1.4093, loss_cns_3: 0.6642, loss_yns_3: 0.1370, loss_cls_4: 0.8083, loss_box_4: 1.3982, loss_cns_4: 0.6621, loss_yns_4: 0.1366, loss_cls_5: 0.8137, loss_box_5: 1.3915, loss_cns_5: 0.6616, loss_yns_5: 0.1387, loss_cls_dn_0: 0.1225, loss_box_dn_0: 0.7162, loss_cls_dn_1: 0.0932, loss_box_dn_1: 0.6328, loss_cls_dn_2: 0.0911, loss_box_dn_2: 0.6067, loss_cls_dn_3: 0.0938, loss_box_dn_3: 0.6111, loss_cls_dn_4: 0.0905, loss_box_dn_4: 0.6057, loss_cls_dn_5: 0.0948, loss_box_dn_5: 0.6077, loss_dense_depth: 0.7157, loss: 23.2839, grad_norm: 34.8968 -2025-11-12 14:40:57,258 - mmdet - INFO - Iter [375/17500] lr: 2.493e-04, eta: 9:20:49, time: 1.574, data_time: 0.071, memory: 49164, loss_cls_0: 0.7334, loss_box_0: 1.5656, loss_cns_0: 0.6316, loss_yns_0: 0.1377, loss_cls_1: 0.7986, loss_box_1: 1.4329, loss_cns_1: 0.6596, loss_yns_1: 0.1366, loss_cls_2: 0.8109, loss_box_2: 1.4056, loss_cns_2: 0.6619, loss_yns_2: 0.1380, loss_cls_3: 0.8208, loss_box_3: 1.4188, loss_cns_3: 0.6633, loss_yns_3: 0.1381, loss_cls_4: 0.8198, loss_box_4: 1.3859, loss_cns_4: 0.6630, loss_yns_4: 0.1344, loss_cls_5: 0.8193, loss_box_5: 1.3824, loss_cns_5: 0.6598, loss_yns_5: 0.1361, loss_cls_dn_0: 0.1219, loss_box_dn_0: 0.7128, loss_cls_dn_1: 0.0933, loss_box_dn_1: 0.6134, loss_cls_dn_2: 0.0932, loss_box_dn_2: 0.6060, loss_cls_dn_3: 0.0946, loss_box_dn_3: 0.6139, loss_cls_dn_4: 0.0948, loss_box_dn_4: 0.6054, loss_cls_dn_5: 0.0941, loss_box_dn_5: 0.6057, loss_dense_depth: 0.7135, loss: 23.2165, grad_norm: 41.7542 -2025-11-12 14:40:58,827 - mmdet - INFO - Iter [376/17500] lr: 2.497e-04, eta: 9:20:29, time: 1.573, data_time: 0.073, memory: 49164, loss_cls_0: 0.7613, loss_box_0: 1.5840, loss_cns_0: 0.6333, loss_yns_0: 0.1407, loss_cls_1: 0.8241, loss_box_1: 1.4531, loss_cns_1: 0.6611, loss_yns_1: 0.1377, loss_cls_2: 0.8409, loss_box_2: 1.4217, loss_cns_2: 0.6612, loss_yns_2: 0.1374, loss_cls_3: 0.8344, loss_box_3: 1.4233, loss_cns_3: 0.6610, loss_yns_3: 0.1379, loss_cls_4: 0.8334, loss_box_4: 1.4215, loss_cns_4: 0.6637, loss_yns_4: 0.1375, loss_cls_5: 0.8363, loss_box_5: 1.4377, loss_cns_5: 0.6620, loss_yns_5: 0.1389, loss_cls_dn_0: 0.1253, loss_box_dn_0: 0.7024, loss_cls_dn_1: 0.0954, loss_box_dn_1: 0.6460, loss_cls_dn_2: 0.0936, loss_box_dn_2: 0.6388, loss_cls_dn_3: 0.0932, loss_box_dn_3: 0.6445, loss_cls_dn_4: 0.0939, loss_box_dn_4: 0.6470, loss_cls_dn_5: 0.0928, loss_box_dn_5: 0.6567, loss_dense_depth: 0.7368, loss: 23.7106, grad_norm: 32.1698 -2025-11-12 14:41:00,400 - mmdet - INFO - Iter [377/17500] lr: 2.501e-04, eta: 9:20:10, time: 1.573, data_time: 0.071, memory: 49164, loss_cls_0: 0.7398, loss_box_0: 1.5519, loss_cns_0: 0.6375, loss_yns_0: 0.1420, loss_cls_1: 0.8114, loss_box_1: 1.4255, loss_cns_1: 0.6637, loss_yns_1: 0.1398, loss_cls_2: 0.8346, loss_box_2: 1.3894, loss_cns_2: 0.6613, loss_yns_2: 0.1383, loss_cls_3: 0.8249, loss_box_3: 1.3976, loss_cns_3: 0.6607, loss_yns_3: 0.1373, loss_cls_4: 0.8256, loss_box_4: 1.4029, loss_cns_4: 0.6644, loss_yns_4: 0.1379, loss_cls_5: 0.8256, loss_box_5: 1.3939, loss_cns_5: 0.6662, loss_yns_5: 0.1384, loss_cls_dn_0: 0.1180, loss_box_dn_0: 0.7117, loss_cls_dn_1: 0.0912, loss_box_dn_1: 0.6548, loss_cls_dn_2: 0.0912, loss_box_dn_2: 0.6419, loss_cls_dn_3: 0.0927, loss_box_dn_3: 0.6496, loss_cls_dn_4: 0.0922, loss_box_dn_4: 0.6537, loss_cls_dn_5: 0.0933, loss_box_dn_5: 0.6528, loss_dense_depth: 0.7385, loss: 23.4925, grad_norm: 36.9389 -2025-11-12 14:41:01,972 - mmdet - INFO - Iter [378/17500] lr: 2.505e-04, eta: 9:19:50, time: 1.563, data_time: 0.071, memory: 49164, loss_cls_0: 0.7349, loss_box_0: 1.5547, loss_cns_0: 0.6387, loss_yns_0: 0.1454, loss_cls_1: 0.8011, loss_box_1: 1.4533, loss_cns_1: 0.6638, loss_yns_1: 0.1427, loss_cls_2: 0.8151, loss_box_2: 1.4043, loss_cns_2: 0.6612, loss_yns_2: 0.1428, loss_cls_3: 0.8118, loss_box_3: 1.3868, loss_cns_3: 0.6591, loss_yns_3: 0.1428, loss_cls_4: 0.8149, loss_box_4: 1.3916, loss_cns_4: 0.6630, loss_yns_4: 0.1395, loss_cls_5: 0.8162, loss_box_5: 1.3871, loss_cns_5: 0.6621, loss_yns_5: 0.1398, loss_cls_dn_0: 0.1274, loss_box_dn_0: 0.7112, loss_cls_dn_1: 0.0924, loss_box_dn_1: 0.6512, loss_cls_dn_2: 0.0928, loss_box_dn_2: 0.6283, loss_cls_dn_3: 0.0954, loss_box_dn_3: 0.6273, loss_cls_dn_4: 0.0930, loss_box_dn_4: 0.6229, loss_cls_dn_5: 0.0961, loss_box_dn_5: 0.6246, loss_dense_depth: 0.7135, loss: 23.3488, grad_norm: 33.2095 -2025-11-12 14:41:03,560 - mmdet - INFO - Iter [379/17500] lr: 2.509e-04, eta: 9:19:31, time: 1.591, data_time: 0.076, memory: 49164, loss_cls_0: 0.7297, loss_box_0: 1.5722, loss_cns_0: 0.6371, loss_yns_0: 0.1426, loss_cls_1: 0.7991, loss_box_1: 1.4600, loss_cns_1: 0.6633, loss_yns_1: 0.1407, loss_cls_2: 0.8136, loss_box_2: 1.4174, loss_cns_2: 0.6615, loss_yns_2: 0.1394, loss_cls_3: 0.8222, loss_box_3: 1.3785, loss_cns_3: 0.6578, loss_yns_3: 0.1382, loss_cls_4: 0.8201, loss_box_4: 1.3850, loss_cns_4: 0.6615, loss_yns_4: 0.1394, loss_cls_5: 0.8245, loss_box_5: 1.3944, loss_cns_5: 0.6624, loss_yns_5: 0.1386, loss_cls_dn_0: 0.1239, loss_box_dn_0: 0.7040, loss_cls_dn_1: 0.0930, loss_box_dn_1: 0.6341, loss_cls_dn_2: 0.0918, loss_box_dn_2: 0.6197, loss_cls_dn_3: 0.0962, loss_box_dn_3: 0.6138, loss_cls_dn_4: 0.0937, loss_box_dn_4: 0.6139, loss_cls_dn_5: 0.0948, loss_box_dn_5: 0.6204, loss_dense_depth: 0.7096, loss: 23.3078, grad_norm: 32.1233 -2025-11-12 14:41:05,141 - mmdet - INFO - Iter [380/17500] lr: 2.513e-04, eta: 9:19:12, time: 1.584, data_time: 0.079, memory: 49164, loss_cls_0: 0.7387, loss_box_0: 1.5284, loss_cns_0: 0.6412, loss_yns_0: 0.1409, loss_cls_1: 0.7889, loss_box_1: 1.4147, loss_cns_1: 0.6605, loss_yns_1: 0.1369, loss_cls_2: 0.7980, loss_box_2: 1.3840, loss_cns_2: 0.6592, loss_yns_2: 0.1376, loss_cls_3: 0.8106, loss_box_3: 1.3736, loss_cns_3: 0.6571, loss_yns_3: 0.1367, loss_cls_4: 0.8143, loss_box_4: 1.3522, loss_cns_4: 0.6586, loss_yns_4: 0.1369, loss_cls_5: 0.8166, loss_box_5: 1.3498, loss_cns_5: 0.6605, loss_yns_5: 0.1378, loss_cls_dn_0: 0.1207, loss_box_dn_0: 0.7185, loss_cls_dn_1: 0.0932, loss_box_dn_1: 0.6403, loss_cls_dn_2: 0.0901, loss_box_dn_2: 0.6318, loss_cls_dn_3: 0.0928, loss_box_dn_3: 0.6361, loss_cls_dn_4: 0.0929, loss_box_dn_4: 0.6295, loss_cls_dn_5: 0.0932, loss_box_dn_5: 0.6280, loss_dense_depth: 0.7223, loss: 23.1230, grad_norm: 40.9484 -2025-11-12 14:41:06,798 - mmdet - INFO - Iter [381/17500] lr: 2.517e-04, eta: 9:18:56, time: 1.653, data_time: 0.105, memory: 49164, loss_cls_0: 0.7219, loss_box_0: 1.5056, loss_cns_0: 0.6385, loss_yns_0: 0.1360, loss_cls_1: 0.7795, loss_box_1: 1.3956, loss_cns_1: 0.6577, loss_yns_1: 0.1336, loss_cls_2: 0.7850, loss_box_2: 1.3723, loss_cns_2: 0.6590, loss_yns_2: 0.1357, loss_cls_3: 0.7888, loss_box_3: 1.3508, loss_cns_3: 0.6565, loss_yns_3: 0.1351, loss_cls_4: 0.7963, loss_box_4: 1.3258, loss_cns_4: 0.6569, loss_yns_4: 0.1319, loss_cls_5: 0.8080, loss_box_5: 1.3220, loss_cns_5: 0.6546, loss_yns_5: 0.1320, loss_cls_dn_0: 0.1211, loss_box_dn_0: 0.7110, loss_cls_dn_1: 0.0936, loss_box_dn_1: 0.6386, loss_cls_dn_2: 0.0916, loss_box_dn_2: 0.6215, loss_cls_dn_3: 0.0916, loss_box_dn_3: 0.6198, loss_cls_dn_4: 0.0918, loss_box_dn_4: 0.6139, loss_cls_dn_5: 0.0946, loss_box_dn_5: 0.6168, loss_dense_depth: 0.6942, loss: 22.7790, grad_norm: 30.4146 -2025-11-12 14:41:08,423 - mmdet - INFO - Iter [382/17500] lr: 2.521e-04, eta: 9:18:39, time: 1.623, data_time: 0.103, memory: 49164, loss_cls_0: 0.7106, loss_box_0: 1.5490, loss_cns_0: 0.6395, loss_yns_0: 0.1357, loss_cls_1: 0.7631, loss_box_1: 1.4121, loss_cns_1: 0.6622, loss_yns_1: 0.1341, loss_cls_2: 0.7643, loss_box_2: 1.3963, loss_cns_2: 0.6634, loss_yns_2: 0.1372, loss_cls_3: 0.7760, loss_box_3: 1.3798, loss_cns_3: 0.6619, loss_yns_3: 0.1330, loss_cls_4: 0.7791, loss_box_4: 1.3915, loss_cns_4: 0.6667, loss_yns_4: 0.1347, loss_cls_5: 0.7923, loss_box_5: 1.3740, loss_cns_5: 0.6638, loss_yns_5: 0.1323, loss_cls_dn_0: 0.1250, loss_box_dn_0: 0.7142, loss_cls_dn_1: 0.0930, loss_box_dn_1: 0.6235, loss_cls_dn_2: 0.0913, loss_box_dn_2: 0.6126, loss_cls_dn_3: 0.0936, loss_box_dn_3: 0.6092, loss_cls_dn_4: 0.0917, loss_box_dn_4: 0.6131, loss_cls_dn_5: 0.0935, loss_box_dn_5: 0.6093, loss_dense_depth: 0.7023, loss: 22.9246, grad_norm: 36.3261 -2025-11-12 14:41:10,018 - mmdet - INFO - Iter [383/17500] lr: 2.525e-04, eta: 9:18:21, time: 1.598, data_time: 0.098, memory: 49164, loss_cls_0: 0.7362, loss_box_0: 1.5323, loss_cns_0: 0.6379, loss_yns_0: 0.1356, loss_cls_1: 0.7774, loss_box_1: 1.4362, loss_cns_1: 0.6662, loss_yns_1: 0.1349, loss_cls_2: 0.7963, loss_box_2: 1.4007, loss_cns_2: 0.6674, loss_yns_2: 0.1354, loss_cls_3: 0.8069, loss_box_3: 1.3898, loss_cns_3: 0.6669, loss_yns_3: 0.1341, loss_cls_4: 0.8055, loss_box_4: 1.3824, loss_cns_4: 0.6709, loss_yns_4: 0.1360, loss_cls_5: 0.8140, loss_box_5: 1.3802, loss_cns_5: 0.6679, loss_yns_5: 0.1341, loss_cls_dn_0: 0.1187, loss_box_dn_0: 0.7073, loss_cls_dn_1: 0.0915, loss_box_dn_1: 0.6211, loss_cls_dn_2: 0.0928, loss_box_dn_2: 0.6103, loss_cls_dn_3: 0.0960, loss_box_dn_3: 0.6063, loss_cls_dn_4: 0.0947, loss_box_dn_4: 0.6014, loss_cls_dn_5: 0.0961, loss_box_dn_5: 0.6002, loss_dense_depth: 0.6945, loss: 23.0763, grad_norm: 36.9075 -2025-11-12 14:41:11,616 - mmdet - INFO - Iter [384/17500] lr: 2.529e-04, eta: 9:18:03, time: 1.605, data_time: 0.083, memory: 49164, loss_cls_0: 0.7516, loss_box_0: 1.5409, loss_cns_0: 0.6370, loss_yns_0: 0.1367, loss_cls_1: 0.7886, loss_box_1: 1.5185, loss_cns_1: 0.6584, loss_yns_1: 0.1355, loss_cls_2: 0.8097, loss_box_2: 1.4501, loss_cns_2: 0.6606, loss_yns_2: 0.1350, loss_cls_3: 0.8151, loss_box_3: 1.4555, loss_cns_3: 0.6619, loss_yns_3: 0.1338, loss_cls_4: 0.8105, loss_box_4: 1.4513, loss_cns_4: 0.6610, loss_yns_4: 0.1345, loss_cls_5: 0.8210, loss_box_5: 1.4487, loss_cns_5: 0.6625, loss_yns_5: 0.1349, loss_cls_dn_0: 0.1184, loss_box_dn_0: 0.7058, loss_cls_dn_1: 0.0912, loss_box_dn_1: 0.6317, loss_cls_dn_2: 0.0940, loss_box_dn_2: 0.6131, loss_cls_dn_3: 0.0961, loss_box_dn_3: 0.6200, loss_cls_dn_4: 0.0935, loss_box_dn_4: 0.6114, loss_cls_dn_5: 0.0943, loss_box_dn_5: 0.6097, loss_dense_depth: 0.7102, loss: 23.5028, grad_norm: 34.2469 -2025-11-12 14:41:13,217 - mmdet - INFO - Iter [385/17500] lr: 2.533e-04, eta: 9:17:46, time: 1.601, data_time: 0.083, memory: 49164, loss_cls_0: 0.7127, loss_box_0: 1.5272, loss_cns_0: 0.6415, loss_yns_0: 0.1380, loss_cls_1: 0.7671, loss_box_1: 1.4569, loss_cns_1: 0.6630, loss_yns_1: 0.1372, loss_cls_2: 0.7963, loss_box_2: 1.4177, loss_cns_2: 0.6634, loss_yns_2: 0.1367, loss_cls_3: 0.7919, loss_box_3: 1.4297, loss_cns_3: 0.6631, loss_yns_3: 0.1361, loss_cls_4: 0.7874, loss_box_4: 1.3988, loss_cns_4: 0.6630, loss_yns_4: 0.1353, loss_cls_5: 0.7911, loss_box_5: 1.3783, loss_cns_5: 0.6639, loss_yns_5: 0.1352, loss_cls_dn_0: 0.1191, loss_box_dn_0: 0.7107, loss_cls_dn_1: 0.0880, loss_box_dn_1: 0.6261, loss_cls_dn_2: 0.0886, loss_box_dn_2: 0.6258, loss_cls_dn_3: 0.0876, loss_box_dn_3: 0.6402, loss_cls_dn_4: 0.0886, loss_box_dn_4: 0.6237, loss_cls_dn_5: 0.0877, loss_box_dn_5: 0.6102, loss_dense_depth: 0.6775, loss: 23.1053, grad_norm: 38.6686 -2025-11-12 14:41:14,834 - mmdet - INFO - Iter [386/17500] lr: 2.537e-04, eta: 9:17:28, time: 1.612, data_time: 0.107, memory: 49164, loss_cls_0: 0.7317, loss_box_0: 1.5525, loss_cns_0: 0.6391, loss_yns_0: 0.1360, loss_cls_1: 0.7792, loss_box_1: 1.4714, loss_cns_1: 0.6613, loss_yns_1: 0.1345, loss_cls_2: 0.8038, loss_box_2: 1.4556, loss_cns_2: 0.6634, loss_yns_2: 0.1344, loss_cls_3: 0.8009, loss_box_3: 1.4452, loss_cns_3: 0.6615, loss_yns_3: 0.1343, loss_cls_4: 0.8098, loss_box_4: 1.4195, loss_cns_4: 0.6631, loss_yns_4: 0.1332, loss_cls_5: 0.8069, loss_box_5: 1.4200, loss_cns_5: 0.6617, loss_yns_5: 0.1331, loss_cls_dn_0: 0.1202, loss_box_dn_0: 0.7168, loss_cls_dn_1: 0.0892, loss_box_dn_1: 0.6210, loss_cls_dn_2: 0.0897, loss_box_dn_2: 0.6197, loss_cls_dn_3: 0.0882, loss_box_dn_3: 0.6198, loss_cls_dn_4: 0.0913, loss_box_dn_4: 0.6066, loss_cls_dn_5: 0.0896, loss_box_dn_5: 0.6055, loss_dense_depth: 0.7037, loss: 23.3135, grad_norm: 32.6865 -2025-11-12 14:41:16,413 - mmdet - INFO - Iter [387/17500] lr: 2.541e-04, eta: 9:17:10, time: 1.579, data_time: 0.081, memory: 49164, loss_cls_0: 0.6952, loss_box_0: 1.5660, loss_cns_0: 0.6389, loss_yns_0: 0.1398, loss_cls_1: 0.7577, loss_box_1: 1.4385, loss_cns_1: 0.6607, loss_yns_1: 0.1383, loss_cls_2: 0.7812, loss_box_2: 1.4042, loss_cns_2: 0.6610, loss_yns_2: 0.1354, loss_cls_3: 0.7854, loss_box_3: 1.3954, loss_cns_3: 0.6610, loss_yns_3: 0.1357, loss_cls_4: 0.7872, loss_box_4: 1.3765, loss_cns_4: 0.6660, loss_yns_4: 0.1337, loss_cls_5: 0.7923, loss_box_5: 1.3929, loss_cns_5: 0.6600, loss_yns_5: 0.1337, loss_cls_dn_0: 0.1152, loss_box_dn_0: 0.7121, loss_cls_dn_1: 0.0875, loss_box_dn_1: 0.6378, loss_cls_dn_2: 0.0888, loss_box_dn_2: 0.6196, loss_cls_dn_3: 0.0867, loss_box_dn_3: 0.6163, loss_cls_dn_4: 0.0878, loss_box_dn_4: 0.6152, loss_cls_dn_5: 0.0882, loss_box_dn_5: 0.6199, loss_dense_depth: 0.7003, loss: 23.0120, grad_norm: 39.1965 -2025-11-12 14:41:18,001 - mmdet - INFO - Iter [388/17500] lr: 2.545e-04, eta: 9:16:52, time: 1.584, data_time: 0.089, memory: 49164, loss_cls_0: 0.6985, loss_box_0: 1.5319, loss_cns_0: 0.6410, loss_yns_0: 0.1380, loss_cls_1: 0.7492, loss_box_1: 1.4104, loss_cns_1: 0.6625, loss_yns_1: 0.1358, loss_cls_2: 0.7757, loss_box_2: 1.3658, loss_cns_2: 0.6642, loss_yns_2: 0.1341, loss_cls_3: 0.7699, loss_box_3: 1.3596, loss_cns_3: 0.6607, loss_yns_3: 0.1349, loss_cls_4: 0.7732, loss_box_4: 1.3577, loss_cns_4: 0.6662, loss_yns_4: 0.1352, loss_cls_5: 0.7828, loss_box_5: 1.3714, loss_cns_5: 0.6619, loss_yns_5: 0.1356, loss_cls_dn_0: 0.1133, loss_box_dn_0: 0.7119, loss_cls_dn_1: 0.0889, loss_box_dn_1: 0.6372, loss_cls_dn_2: 0.0899, loss_box_dn_2: 0.6165, loss_cls_dn_3: 0.0879, loss_box_dn_3: 0.6121, loss_cls_dn_4: 0.0901, loss_box_dn_4: 0.6115, loss_cls_dn_5: 0.0913, loss_box_dn_5: 0.6200, loss_dense_depth: 0.6953, loss: 22.7821, grad_norm: 24.2955 -2025-11-12 14:41:29,546 - mmdet - INFO - Iter [389/17500] lr: 2.549e-04, eta: 9:23:52, time: 11.551, data_time: 0.088, memory: 49164, loss_cls_0: 0.6930, loss_box_0: 1.5344, loss_cns_0: 0.6414, loss_yns_0: 0.1382, loss_cls_1: 0.7539, loss_box_1: 1.4189, loss_cns_1: 0.6660, loss_yns_1: 0.1366, loss_cls_2: 0.7898, loss_box_2: 1.3739, loss_cns_2: 0.6643, loss_yns_2: 0.1358, loss_cls_3: 0.7795, loss_box_3: 1.3585, loss_cns_3: 0.6613, loss_yns_3: 0.1346, loss_cls_4: 0.7920, loss_box_4: 1.3479, loss_cns_4: 0.6623, loss_yns_4: 0.1372, loss_cls_5: 0.7884, loss_box_5: 1.3539, loss_cns_5: 0.6604, loss_yns_5: 0.1345, loss_cls_dn_0: 0.1101, loss_box_dn_0: 0.7027, loss_cls_dn_1: 0.0890, loss_box_dn_1: 0.6411, loss_cls_dn_2: 0.0889, loss_box_dn_2: 0.6248, loss_cls_dn_3: 0.0870, loss_box_dn_3: 0.6179, loss_cls_dn_4: 0.0887, loss_box_dn_4: 0.6113, loss_cls_dn_5: 0.0897, loss_box_dn_5: 0.6141, loss_dense_depth: 0.7239, loss: 22.8455, grad_norm: 34.7568 -2025-11-12 14:41:31,106 - mmdet - INFO - Iter [390/17500] lr: 2.553e-04, eta: 9:23:32, time: 1.561, data_time: 0.083, memory: 49164, loss_cls_0: 0.6957, loss_box_0: 1.5256, loss_cns_0: 0.6404, loss_yns_0: 0.1427, loss_cls_1: 0.7660, loss_box_1: 1.4045, loss_cns_1: 0.6635, loss_yns_1: 0.1380, loss_cls_2: 0.7814, loss_box_2: 1.3722, loss_cns_2: 0.6617, loss_yns_2: 0.1363, loss_cls_3: 0.7824, loss_box_3: 1.3608, loss_cns_3: 0.6640, loss_yns_3: 0.1391, loss_cls_4: 0.7887, loss_box_4: 1.3560, loss_cns_4: 0.6677, loss_yns_4: 0.1388, loss_cls_5: 0.7860, loss_box_5: 1.3643, loss_cns_5: 0.6644, loss_yns_5: 0.1370, loss_cls_dn_0: 0.1119, loss_box_dn_0: 0.6983, loss_cls_dn_1: 0.0911, loss_box_dn_1: 0.6217, loss_cls_dn_2: 0.0913, loss_box_dn_2: 0.6107, loss_cls_dn_3: 0.0920, loss_box_dn_3: 0.6079, loss_cls_dn_4: 0.0922, loss_box_dn_4: 0.6084, loss_cls_dn_5: 0.0911, loss_box_dn_5: 0.6134, loss_dense_depth: 0.7206, loss: 22.8281, grad_norm: 32.4642 -2025-11-12 14:41:32,720 - mmdet - INFO - Iter [391/17500] lr: 2.557e-04, eta: 9:23:14, time: 1.612, data_time: 0.073, memory: 49164, loss_cls_0: 0.6916, loss_box_0: 1.5280, loss_cns_0: 0.6404, loss_yns_0: 0.1387, loss_cls_1: 0.7649, loss_box_1: 1.3928, loss_cns_1: 0.6635, loss_yns_1: 0.1355, loss_cls_2: 0.7837, loss_box_2: 1.3642, loss_cns_2: 0.6688, loss_yns_2: 0.1376, loss_cls_3: 0.7776, loss_box_3: 1.3546, loss_cns_3: 0.6657, loss_yns_3: 0.1382, loss_cls_4: 0.7881, loss_box_4: 1.3409, loss_cns_4: 0.6712, loss_yns_4: 0.1346, loss_cls_5: 0.7914, loss_box_5: 1.3577, loss_cns_5: 0.6639, loss_yns_5: 0.1338, loss_cls_dn_0: 0.1144, loss_box_dn_0: 0.7136, loss_cls_dn_1: 0.0951, loss_box_dn_1: 0.6282, loss_cls_dn_2: 0.0916, loss_box_dn_2: 0.6167, loss_cls_dn_3: 0.0921, loss_box_dn_3: 0.6137, loss_cls_dn_4: 0.0937, loss_box_dn_4: 0.6082, loss_cls_dn_5: 0.0942, loss_box_dn_5: 0.6156, loss_dense_depth: 0.7432, loss: 22.8474, grad_norm: 28.9269 -2025-11-12 14:41:34,301 - mmdet - INFO - Iter [392/17500] lr: 2.561e-04, eta: 9:22:55, time: 1.581, data_time: 0.077, memory: 49164, loss_cls_0: 0.7096, loss_box_0: 1.5458, loss_cns_0: 0.6362, loss_yns_0: 0.1388, loss_cls_1: 0.7614, loss_box_1: 1.4222, loss_cns_1: 0.6635, loss_yns_1: 0.1375, loss_cls_2: 0.7852, loss_box_2: 1.3962, loss_cns_2: 0.6670, loss_yns_2: 0.1379, loss_cls_3: 0.7901, loss_box_3: 1.4046, loss_cns_3: 0.6623, loss_yns_3: 0.1370, loss_cls_4: 0.8039, loss_box_4: 1.4021, loss_cns_4: 0.6647, loss_yns_4: 0.1396, loss_cls_5: 0.8055, loss_box_5: 1.3984, loss_cns_5: 0.6605, loss_yns_5: 0.1384, loss_cls_dn_0: 0.1185, loss_box_dn_0: 0.7115, loss_cls_dn_1: 0.0926, loss_box_dn_1: 0.6299, loss_cls_dn_2: 0.0934, loss_box_dn_2: 0.6170, loss_cls_dn_3: 0.0950, loss_box_dn_3: 0.6189, loss_cls_dn_4: 0.0982, loss_box_dn_4: 0.6188, loss_cls_dn_5: 0.0988, loss_box_dn_5: 0.6168, loss_dense_depth: 0.7077, loss: 23.1257, grad_norm: 36.1466 -2025-11-12 14:41:35,886 - mmdet - INFO - Iter [393/17500] lr: 2.565e-04, eta: 9:22:36, time: 1.584, data_time: 0.077, memory: 49164, loss_cls_0: 0.7225, loss_box_0: 1.5792, loss_cns_0: 0.6371, loss_yns_0: 0.1451, loss_cls_1: 0.7762, loss_box_1: 1.4296, loss_cns_1: 0.6633, loss_yns_1: 0.1381, loss_cls_2: 0.7884, loss_box_2: 1.3851, loss_cns_2: 0.6637, loss_yns_2: 0.1367, loss_cls_3: 0.7923, loss_box_3: 1.3855, loss_cns_3: 0.6630, loss_yns_3: 0.1373, loss_cls_4: 0.7946, loss_box_4: 1.3804, loss_cns_4: 0.6658, loss_yns_4: 0.1409, loss_cls_5: 0.8051, loss_box_5: 1.3803, loss_cns_5: 0.6637, loss_yns_5: 0.1389, loss_cls_dn_0: 0.1174, loss_box_dn_0: 0.7132, loss_cls_dn_1: 0.0929, loss_box_dn_1: 0.6348, loss_cls_dn_2: 0.0947, loss_box_dn_2: 0.6176, loss_cls_dn_3: 0.0955, loss_box_dn_3: 0.6176, loss_cls_dn_4: 0.0940, loss_box_dn_4: 0.6204, loss_cls_dn_5: 0.0945, loss_box_dn_5: 0.6256, loss_dense_depth: 0.7087, loss: 23.1394, grad_norm: 27.6648 -2025-11-12 14:41:37,499 - mmdet - INFO - Iter [394/17500] lr: 2.569e-04, eta: 9:22:18, time: 1.616, data_time: 0.082, memory: 49164, loss_cls_0: 0.6971, loss_box_0: 1.5253, loss_cns_0: 0.6383, loss_yns_0: 0.1444, loss_cls_1: 0.7549, loss_box_1: 1.4130, loss_cns_1: 0.6629, loss_yns_1: 0.1391, loss_cls_2: 0.7725, loss_box_2: 1.3591, loss_cns_2: 0.6617, loss_yns_2: 0.1384, loss_cls_3: 0.7757, loss_box_3: 1.3427, loss_cns_3: 0.6652, loss_yns_3: 0.1382, loss_cls_4: 0.7801, loss_box_4: 1.3483, loss_cns_4: 0.6625, loss_yns_4: 0.1399, loss_cls_5: 0.7874, loss_box_5: 1.3620, loss_cns_5: 0.6632, loss_yns_5: 0.1367, loss_cls_dn_0: 0.1185, loss_box_dn_0: 0.7054, loss_cls_dn_1: 0.0933, loss_box_dn_1: 0.6287, loss_cls_dn_2: 0.0928, loss_box_dn_2: 0.6101, loss_cls_dn_3: 0.0913, loss_box_dn_3: 0.6032, loss_cls_dn_4: 0.0924, loss_box_dn_4: 0.6096, loss_cls_dn_5: 0.0939, loss_box_dn_5: 0.6165, loss_dense_depth: 0.7063, loss: 22.7707, grad_norm: 40.2198 -2025-11-12 14:41:39,082 - mmdet - INFO - Iter [395/17500] lr: 2.573e-04, eta: 9:21:59, time: 1.586, data_time: 0.075, memory: 49164, loss_cls_0: 0.7031, loss_box_0: 1.5311, loss_cns_0: 0.6375, loss_yns_0: 0.1446, loss_cls_1: 0.7644, loss_box_1: 1.3752, loss_cns_1: 0.6638, loss_yns_1: 0.1406, loss_cls_2: 0.7765, loss_box_2: 1.3437, loss_cns_2: 0.6638, loss_yns_2: 0.1410, loss_cls_3: 0.7743, loss_box_3: 1.3327, loss_cns_3: 0.6648, loss_yns_3: 0.1420, loss_cls_4: 0.7743, loss_box_4: 1.3411, loss_cns_4: 0.6653, loss_yns_4: 0.1400, loss_cls_5: 0.7764, loss_box_5: 1.3258, loss_cns_5: 0.6629, loss_yns_5: 0.1400, loss_cls_dn_0: 0.1161, loss_box_dn_0: 0.7122, loss_cls_dn_1: 0.0912, loss_box_dn_1: 0.6194, loss_cls_dn_2: 0.0876, loss_box_dn_2: 0.6097, loss_cls_dn_3: 0.0895, loss_box_dn_3: 0.6065, loss_cls_dn_4: 0.0925, loss_box_dn_4: 0.6116, loss_cls_dn_5: 0.0962, loss_box_dn_5: 0.6058, loss_dense_depth: 0.7156, loss: 22.6787, grad_norm: 35.8014 -2025-11-12 14:41:40,672 - mmdet - INFO - Iter [396/17500] lr: 2.577e-04, eta: 9:21:41, time: 1.590, data_time: 0.075, memory: 49164, loss_cls_0: 0.7035, loss_box_0: 1.5002, loss_cns_0: 0.6412, loss_yns_0: 0.1434, loss_cls_1: 0.7731, loss_box_1: 1.3550, loss_cns_1: 0.6671, loss_yns_1: 0.1418, loss_cls_2: 0.7929, loss_box_2: 1.3130, loss_cns_2: 0.6658, loss_yns_2: 0.1417, loss_cls_3: 0.7938, loss_box_3: 1.3075, loss_cns_3: 0.6657, loss_yns_3: 0.1412, loss_cls_4: 0.7887, loss_box_4: 1.3125, loss_cns_4: 0.6711, loss_yns_4: 0.1417, loss_cls_5: 0.7987, loss_box_5: 1.3015, loss_cns_5: 0.6649, loss_yns_5: 0.1406, loss_cls_dn_0: 0.1158, loss_box_dn_0: 0.7133, loss_cls_dn_1: 0.0904, loss_box_dn_1: 0.6125, loss_cls_dn_2: 0.0878, loss_box_dn_2: 0.5969, loss_cls_dn_3: 0.0873, loss_box_dn_3: 0.5944, loss_cls_dn_4: 0.0873, loss_box_dn_4: 0.5961, loss_cls_dn_5: 0.0907, loss_box_dn_5: 0.5927, loss_dense_depth: 0.7315, loss: 22.5637, grad_norm: 27.3052 -2025-11-12 14:41:42,260 - mmdet - INFO - Iter [397/17500] lr: 2.581e-04, eta: 9:21:22, time: 1.587, data_time: 0.072, memory: 49164, loss_cls_0: 0.7214, loss_box_0: 1.5360, loss_cns_0: 0.6351, loss_yns_0: 0.1434, loss_cls_1: 0.7891, loss_box_1: 1.3648, loss_cns_1: 0.6668, loss_yns_1: 0.1403, loss_cls_2: 0.7980, loss_box_2: 1.3463, loss_cns_2: 0.6661, loss_yns_2: 0.1393, loss_cls_3: 0.8027, loss_box_3: 1.3443, loss_cns_3: 0.6663, loss_yns_3: 0.1396, loss_cls_4: 0.8068, loss_box_4: 1.3502, loss_cns_4: 0.6718, loss_yns_4: 0.1423, loss_cls_5: 0.8177, loss_box_5: 1.3484, loss_cns_5: 0.6651, loss_yns_5: 0.1414, loss_cls_dn_0: 0.1182, loss_box_dn_0: 0.7132, loss_cls_dn_1: 0.0922, loss_box_dn_1: 0.6114, loss_cls_dn_2: 0.0922, loss_box_dn_2: 0.6019, loss_cls_dn_3: 0.0911, loss_box_dn_3: 0.5973, loss_cls_dn_4: 0.0893, loss_box_dn_4: 0.5983, loss_cls_dn_5: 0.0909, loss_box_dn_5: 0.6025, loss_dense_depth: 0.7265, loss: 22.8684, grad_norm: 39.1414 -2025-11-12 14:41:43,843 - mmdet - INFO - Iter [398/17500] lr: 2.585e-04, eta: 9:21:04, time: 1.584, data_time: 0.076, memory: 49164, loss_cls_0: 0.7520, loss_box_0: 1.5309, loss_cns_0: 0.6300, loss_yns_0: 0.1432, loss_cls_1: 0.7982, loss_box_1: 1.3961, loss_cns_1: 0.6630, loss_yns_1: 0.1395, loss_cls_2: 0.8035, loss_box_2: 1.3767, loss_cns_2: 0.6614, loss_yns_2: 0.1392, loss_cls_3: 0.7996, loss_box_3: 1.3579, loss_cns_3: 0.6629, loss_yns_3: 0.1383, loss_cls_4: 0.8089, loss_box_4: 1.3642, loss_cns_4: 0.6617, loss_yns_4: 0.1408, loss_cls_5: 0.8126, loss_box_5: 1.3658, loss_cns_5: 0.6610, loss_yns_5: 0.1395, loss_cls_dn_0: 0.1173, loss_box_dn_0: 0.7083, loss_cls_dn_1: 0.0887, loss_box_dn_1: 0.6069, loss_cls_dn_2: 0.0881, loss_box_dn_2: 0.5984, loss_cls_dn_3: 0.0899, loss_box_dn_3: 0.5889, loss_cls_dn_4: 0.0890, loss_box_dn_4: 0.5935, loss_cls_dn_5: 0.0887, loss_box_dn_5: 0.5958, loss_dense_depth: 0.7558, loss: 22.9562, grad_norm: 29.1547 -2025-11-12 14:41:45,411 - mmdet - INFO - Iter [399/17500] lr: 2.589e-04, eta: 9:20:45, time: 1.568, data_time: 0.074, memory: 49164, loss_cls_0: 0.7175, loss_box_0: 1.5187, loss_cns_0: 0.6359, loss_yns_0: 0.1400, loss_cls_1: 0.7796, loss_box_1: 1.4035, loss_cns_1: 0.6654, loss_yns_1: 0.1399, loss_cls_2: 0.8003, loss_box_2: 1.3690, loss_cns_2: 0.6626, loss_yns_2: 0.1411, loss_cls_3: 0.7874, loss_box_3: 1.3572, loss_cns_3: 0.6635, loss_yns_3: 0.1389, loss_cls_4: 0.7952, loss_box_4: 1.3535, loss_cns_4: 0.6607, loss_yns_4: 0.1387, loss_cls_5: 0.8113, loss_box_5: 1.3635, loss_cns_5: 0.6625, loss_yns_5: 0.1390, loss_cls_dn_0: 0.1186, loss_box_dn_0: 0.7070, loss_cls_dn_1: 0.0898, loss_box_dn_1: 0.6107, loss_cls_dn_2: 0.0893, loss_box_dn_2: 0.5987, loss_cls_dn_3: 0.0897, loss_box_dn_3: 0.5942, loss_cls_dn_4: 0.0915, loss_box_dn_4: 0.5922, loss_cls_dn_5: 0.0941, loss_box_dn_5: 0.5963, loss_dense_depth: 0.7554, loss: 22.8724, grad_norm: 37.9899 -2025-11-12 14:41:46,988 - mmdet - INFO - Iter [400/17500] lr: 2.593e-04, eta: 9:20:26, time: 1.577, data_time: 0.074, memory: 49164, loss_cls_0: 0.7364, loss_box_0: 1.5273, loss_cns_0: 0.6417, loss_yns_0: 0.1424, loss_cls_1: 0.7942, loss_box_1: 1.4094, loss_cns_1: 0.6679, loss_yns_1: 0.1406, loss_cls_2: 0.8068, loss_box_2: 1.3623, loss_cns_2: 0.6650, loss_yns_2: 0.1419, loss_cls_3: 0.7987, loss_box_3: 1.3267, loss_cns_3: 0.6618, loss_yns_3: 0.1395, loss_cls_4: 0.8052, loss_box_4: 1.3199, loss_cns_4: 0.6673, loss_yns_4: 0.1393, loss_cls_5: 0.8180, loss_box_5: 1.3382, loss_cns_5: 0.6623, loss_yns_5: 0.1414, loss_cls_dn_0: 0.1184, loss_box_dn_0: 0.7191, loss_cls_dn_1: 0.0899, loss_box_dn_1: 0.6370, loss_cls_dn_2: 0.0902, loss_box_dn_2: 0.6178, loss_cls_dn_3: 0.0898, loss_box_dn_3: 0.6091, loss_cls_dn_4: 0.0887, loss_box_dn_4: 0.6123, loss_cls_dn_5: 0.0913, loss_box_dn_5: 0.6191, loss_dense_depth: 0.7170, loss: 22.9539, grad_norm: 35.3114 -2025-11-12 14:41:48,648 - mmdet - INFO - Iter [401/17500] lr: 2.597e-04, eta: 9:20:11, time: 1.660, data_time: 0.104, memory: 49164, loss_cls_0: 0.7422, loss_box_0: 1.5453, loss_cns_0: 0.6426, loss_yns_0: 0.1417, loss_cls_1: 0.8021, loss_box_1: 1.4259, loss_cns_1: 0.6594, loss_yns_1: 0.1415, loss_cls_2: 0.8119, loss_box_2: 1.3994, loss_cns_2: 0.6592, loss_yns_2: 0.1415, loss_cls_3: 0.8188, loss_box_3: 1.3725, loss_cns_3: 0.6604, loss_yns_3: 0.1410, loss_cls_4: 0.8258, loss_box_4: 1.3733, loss_cns_4: 0.6685, loss_yns_4: 0.1396, loss_cls_5: 0.8404, loss_box_5: 1.3762, loss_cns_5: 0.6591, loss_yns_5: 0.1414, loss_cls_dn_0: 0.1173, loss_box_dn_0: 0.7147, loss_cls_dn_1: 0.0919, loss_box_dn_1: 0.6209, loss_cls_dn_2: 0.0936, loss_box_dn_2: 0.6066, loss_cls_dn_3: 0.0920, loss_box_dn_3: 0.5990, loss_cls_dn_4: 0.0928, loss_box_dn_4: 0.6033, loss_cls_dn_5: 0.0942, loss_box_dn_5: 0.6065, loss_dense_depth: 0.7813, loss: 23.2437, grad_norm: 42.2604 -2025-11-12 14:41:50,291 - mmdet - INFO - Iter [402/17500] lr: 2.601e-04, eta: 9:19:55, time: 1.642, data_time: 0.099, memory: 49164, loss_cls_0: 0.7152, loss_box_0: 1.5368, loss_cns_0: 0.6434, loss_yns_0: 0.1408, loss_cls_1: 0.7857, loss_box_1: 1.4005, loss_cns_1: 0.6595, loss_yns_1: 0.1363, loss_cls_2: 0.7981, loss_box_2: 1.3712, loss_cns_2: 0.6574, loss_yns_2: 0.1358, loss_cls_3: 0.7879, loss_box_3: 1.3821, loss_cns_3: 0.6624, loss_yns_3: 0.1381, loss_cls_4: 0.8004, loss_box_4: 1.3676, loss_cns_4: 0.6636, loss_yns_4: 0.1375, loss_cls_5: 0.8098, loss_box_5: 1.3624, loss_cns_5: 0.6592, loss_yns_5: 0.1366, loss_cls_dn_0: 0.1130, loss_box_dn_0: 0.7078, loss_cls_dn_1: 0.0904, loss_box_dn_1: 0.6126, loss_cls_dn_2: 0.0905, loss_box_dn_2: 0.5996, loss_cls_dn_3: 0.0881, loss_box_dn_3: 0.5976, loss_cls_dn_4: 0.0904, loss_box_dn_4: 0.5968, loss_cls_dn_5: 0.0919, loss_box_dn_5: 0.5966, loss_dense_depth: 0.7424, loss: 22.9062, grad_norm: 28.9180 -2025-11-12 14:41:51,864 - mmdet - INFO - Iter [403/17500] lr: 2.605e-04, eta: 9:19:37, time: 1.574, data_time: 0.086, memory: 49164, loss_cls_0: 0.7160, loss_box_0: 1.5589, loss_cns_0: 0.6402, loss_yns_0: 0.1392, loss_cls_1: 0.7620, loss_box_1: 1.4230, loss_cns_1: 0.6606, loss_yns_1: 0.1347, loss_cls_2: 0.7788, loss_box_2: 1.3991, loss_cns_2: 0.6633, loss_yns_2: 0.1342, loss_cls_3: 0.7907, loss_box_3: 1.3984, loss_cns_3: 0.6616, loss_yns_3: 0.1345, loss_cls_4: 0.7890, loss_box_4: 1.3895, loss_cns_4: 0.6611, loss_yns_4: 0.1342, loss_cls_5: 0.7989, loss_box_5: 1.3850, loss_cns_5: 0.6617, loss_yns_5: 0.1344, loss_cls_dn_0: 0.1179, loss_box_dn_0: 0.7202, loss_cls_dn_1: 0.0900, loss_box_dn_1: 0.6176, loss_cls_dn_2: 0.0895, loss_box_dn_2: 0.6079, loss_cls_dn_3: 0.0921, loss_box_dn_3: 0.6037, loss_cls_dn_4: 0.0898, loss_box_dn_4: 0.6021, loss_cls_dn_5: 0.0912, loss_box_dn_5: 0.6040, loss_dense_depth: 0.7919, loss: 23.0670, grad_norm: 39.0816 -2025-11-12 14:41:53,440 - mmdet - INFO - Iter [404/17500] lr: 2.609e-04, eta: 9:19:18, time: 1.568, data_time: 0.073, memory: 49164, loss_cls_0: 0.7128, loss_box_0: 1.5463, loss_cns_0: 0.6403, loss_yns_0: 0.1373, loss_cls_1: 0.7631, loss_box_1: 1.4034, loss_cns_1: 0.6634, loss_yns_1: 0.1363, loss_cls_2: 0.7963, loss_box_2: 1.3709, loss_cns_2: 0.6661, loss_yns_2: 0.1349, loss_cls_3: 0.7990, loss_box_3: 1.3680, loss_cns_3: 0.6637, loss_yns_3: 0.1326, loss_cls_4: 0.7870, loss_box_4: 1.3724, loss_cns_4: 0.6655, loss_yns_4: 0.1336, loss_cls_5: 0.7962, loss_box_5: 1.3543, loss_cns_5: 0.6659, loss_yns_5: 0.1356, loss_cls_dn_0: 0.1178, loss_box_dn_0: 0.7121, loss_cls_dn_1: 0.0907, loss_box_dn_1: 0.6080, loss_cls_dn_2: 0.0912, loss_box_dn_2: 0.5954, loss_cls_dn_3: 0.0919, loss_box_dn_3: 0.5948, loss_cls_dn_4: 0.0880, loss_box_dn_4: 0.5975, loss_cls_dn_5: 0.0904, loss_box_dn_5: 0.5959, loss_dense_depth: 0.7401, loss: 22.8588, grad_norm: 39.6760 -2025-11-12 14:41:55,037 - mmdet - INFO - Iter [405/17500] lr: 2.613e-04, eta: 9:19:01, time: 1.597, data_time: 0.088, memory: 49164, loss_cls_0: 0.7399, loss_box_0: 1.5492, loss_cns_0: 0.6350, loss_yns_0: 0.1363, loss_cls_1: 0.7710, loss_box_1: 1.4223, loss_cns_1: 0.6609, loss_yns_1: 0.1357, loss_cls_2: 0.7849, loss_box_2: 1.3995, loss_cns_2: 0.6612, loss_yns_2: 0.1361, loss_cls_3: 0.8026, loss_box_3: 1.3872, loss_cns_3: 0.6616, loss_yns_3: 0.1344, loss_cls_4: 0.7991, loss_box_4: 1.4029, loss_cns_4: 0.6631, loss_yns_4: 0.1351, loss_cls_5: 0.8156, loss_box_5: 1.3809, loss_cns_5: 0.6616, loss_yns_5: 0.1346, loss_cls_dn_0: 0.1219, loss_box_dn_0: 0.7195, loss_cls_dn_1: 0.0899, loss_box_dn_1: 0.6121, loss_cls_dn_2: 0.0892, loss_box_dn_2: 0.5986, loss_cls_dn_3: 0.0900, loss_box_dn_3: 0.5995, loss_cls_dn_4: 0.0896, loss_box_dn_4: 0.6040, loss_cls_dn_5: 0.0919, loss_box_dn_5: 0.6014, loss_dense_depth: 0.7496, loss: 23.0676, grad_norm: 31.9995 -2025-11-12 14:41:56,663 - mmdet - INFO - Iter [406/17500] lr: 2.617e-04, eta: 9:18:44, time: 1.623, data_time: 0.117, memory: 49164, loss_cls_0: 0.7284, loss_box_0: 1.5254, loss_cns_0: 0.6353, loss_yns_0: 0.1341, loss_cls_1: 0.7663, loss_box_1: 1.4061, loss_cns_1: 0.6625, loss_yns_1: 0.1345, loss_cls_2: 0.7793, loss_box_2: 1.3754, loss_cns_2: 0.6625, loss_yns_2: 0.1347, loss_cls_3: 0.8018, loss_box_3: 1.3676, loss_cns_3: 0.6603, loss_yns_3: 0.1344, loss_cls_4: 0.7842, loss_box_4: 1.3738, loss_cns_4: 0.6636, loss_yns_4: 0.1351, loss_cls_5: 0.7962, loss_box_5: 1.3630, loss_cns_5: 0.6628, loss_yns_5: 0.1340, loss_cls_dn_0: 0.1165, loss_box_dn_0: 0.7118, loss_cls_dn_1: 0.0899, loss_box_dn_1: 0.6096, loss_cls_dn_2: 0.0905, loss_box_dn_2: 0.5974, loss_cls_dn_3: 0.0933, loss_box_dn_3: 0.5993, loss_cls_dn_4: 0.0909, loss_box_dn_4: 0.5991, loss_cls_dn_5: 0.0928, loss_box_dn_5: 0.5948, loss_dense_depth: 0.7291, loss: 22.8362, grad_norm: 36.0338 -2025-11-12 14:41:58,257 - mmdet - INFO - Iter [407/17500] lr: 2.621e-04, eta: 9:18:27, time: 1.604, data_time: 0.082, memory: 49164, loss_cls_0: 0.7479, loss_box_0: 1.5601, loss_cns_0: 0.6394, loss_yns_0: 0.1358, loss_cls_1: 0.7844, loss_box_1: 1.3949, loss_cns_1: 0.6657, loss_yns_1: 0.1327, loss_cls_2: 0.8001, loss_box_2: 1.3534, loss_cns_2: 0.6644, loss_yns_2: 0.1339, loss_cls_3: 0.8147, loss_box_3: 1.3468, loss_cns_3: 0.6650, loss_yns_3: 0.1338, loss_cls_4: 0.7981, loss_box_4: 1.3674, loss_cns_4: 0.6713, loss_yns_4: 0.1342, loss_cls_5: 0.8035, loss_box_5: 1.3601, loss_cns_5: 0.6696, loss_yns_5: 0.1335, loss_cls_dn_0: 0.1143, loss_box_dn_0: 0.7123, loss_cls_dn_1: 0.0917, loss_box_dn_1: 0.6225, loss_cls_dn_2: 0.0925, loss_box_dn_2: 0.6052, loss_cls_dn_3: 0.0938, loss_box_dn_3: 0.6059, loss_cls_dn_4: 0.0894, loss_box_dn_4: 0.6157, loss_cls_dn_5: 0.0914, loss_box_dn_5: 0.6156, loss_dense_depth: 0.7174, loss: 22.9782, grad_norm: 27.7005 -2025-11-12 14:41:59,834 - mmdet - INFO - Iter [408/17500] lr: 2.624e-04, eta: 9:18:09, time: 1.570, data_time: 0.074, memory: 49164, loss_cls_0: 0.7245, loss_box_0: 1.5734, loss_cns_0: 0.6381, loss_yns_0: 0.1380, loss_cls_1: 0.7696, loss_box_1: 1.3966, loss_cns_1: 0.6688, loss_yns_1: 0.1337, loss_cls_2: 0.7846, loss_box_2: 1.3631, loss_cns_2: 0.6691, loss_yns_2: 0.1345, loss_cls_3: 0.8014, loss_box_3: 1.3549, loss_cns_3: 0.6689, loss_yns_3: 0.1345, loss_cls_4: 0.7847, loss_box_4: 1.3731, loss_cns_4: 0.6725, loss_yns_4: 0.1351, loss_cls_5: 0.7905, loss_box_5: 1.3631, loss_cns_5: 0.6701, loss_yns_5: 0.1351, loss_cls_dn_0: 0.1139, loss_box_dn_0: 0.7160, loss_cls_dn_1: 0.0915, loss_box_dn_1: 0.6353, loss_cls_dn_2: 0.0909, loss_box_dn_2: 0.6185, loss_cls_dn_3: 0.0910, loss_box_dn_3: 0.6160, loss_cls_dn_4: 0.0885, loss_box_dn_4: 0.6262, loss_cls_dn_5: 0.0903, loss_box_dn_5: 0.6230, loss_dense_depth: 0.7142, loss: 22.9935, grad_norm: 39.1010 -2025-11-12 14:42:01,428 - mmdet - INFO - Iter [409/17500] lr: 2.628e-04, eta: 9:17:52, time: 1.594, data_time: 0.087, memory: 49164, loss_cls_0: 0.7099, loss_box_0: 1.5448, loss_cns_0: 0.6441, loss_yns_0: 0.1395, loss_cls_1: 0.7783, loss_box_1: 1.4043, loss_cns_1: 0.6682, loss_yns_1: 0.1383, loss_cls_2: 0.7820, loss_box_2: 1.3727, loss_cns_2: 0.6711, loss_yns_2: 0.1360, loss_cls_3: 0.7870, loss_box_3: 1.3694, loss_cns_3: 0.6682, loss_yns_3: 0.1359, loss_cls_4: 0.7773, loss_box_4: 1.3761, loss_cns_4: 0.6685, loss_yns_4: 0.1351, loss_cls_5: 0.7806, loss_box_5: 1.3775, loss_cns_5: 0.6695, loss_yns_5: 0.1365, loss_cls_dn_0: 0.1163, loss_box_dn_0: 0.7084, loss_cls_dn_1: 0.0956, loss_box_dn_1: 0.6200, loss_cls_dn_2: 0.0943, loss_box_dn_2: 0.6021, loss_cls_dn_3: 0.0909, loss_box_dn_3: 0.5998, loss_cls_dn_4: 0.0927, loss_box_dn_4: 0.6011, loss_cls_dn_5: 0.0931, loss_box_dn_5: 0.6037, loss_dense_depth: 0.6972, loss: 22.8859, grad_norm: 23.1848 -2025-11-12 14:42:02,999 - mmdet - INFO - Iter [410/17500] lr: 2.632e-04, eta: 9:17:34, time: 1.572, data_time: 0.088, memory: 49164, loss_cls_0: 0.7061, loss_box_0: 1.5337, loss_cns_0: 0.6443, loss_yns_0: 0.1386, loss_cls_1: 0.7717, loss_box_1: 1.3820, loss_cns_1: 0.6644, loss_yns_1: 0.1367, loss_cls_2: 0.7802, loss_box_2: 1.3587, loss_cns_2: 0.6676, loss_yns_2: 0.1392, loss_cls_3: 0.7850, loss_box_3: 1.3420, loss_cns_3: 0.6647, loss_yns_3: 0.1351, loss_cls_4: 0.7765, loss_box_4: 1.3524, loss_cns_4: 0.6705, loss_yns_4: 0.1371, loss_cls_5: 0.7842, loss_box_5: 1.3506, loss_cns_5: 0.6660, loss_yns_5: 0.1377, loss_cls_dn_0: 0.1110, loss_box_dn_0: 0.7088, loss_cls_dn_1: 0.0919, loss_box_dn_1: 0.5996, loss_cls_dn_2: 0.0921, loss_box_dn_2: 0.5877, loss_cls_dn_3: 0.0896, loss_box_dn_3: 0.5817, loss_cls_dn_4: 0.0909, loss_box_dn_4: 0.5831, loss_cls_dn_5: 0.0903, loss_box_dn_5: 0.5838, loss_dense_depth: 0.7120, loss: 22.6474, grad_norm: 32.1731 -2025-11-12 14:42:04,603 - mmdet - INFO - Iter [411/17500] lr: 2.636e-04, eta: 9:17:17, time: 1.603, data_time: 0.079, memory: 49164, loss_cls_0: 0.7245, loss_box_0: 1.5653, loss_cns_0: 0.6416, loss_yns_0: 0.1371, loss_cls_1: 0.7734, loss_box_1: 1.3842, loss_cns_1: 0.6613, loss_yns_1: 0.1347, loss_cls_2: 0.7840, loss_box_2: 1.3670, loss_cns_2: 0.6619, loss_yns_2: 0.1366, loss_cls_3: 0.7960, loss_box_3: 1.3430, loss_cns_3: 0.6584, loss_yns_3: 0.1339, loss_cls_4: 0.8172, loss_box_4: 1.3385, loss_cns_4: 0.6566, loss_yns_4: 0.1339, loss_cls_5: 0.8209, loss_box_5: 1.3380, loss_cns_5: 0.6575, loss_yns_5: 0.1345, loss_cls_dn_0: 0.1137, loss_box_dn_0: 0.7136, loss_cls_dn_1: 0.0885, loss_box_dn_1: 0.6221, loss_cls_dn_2: 0.0881, loss_box_dn_2: 0.6099, loss_cls_dn_3: 0.0879, loss_box_dn_3: 0.6095, loss_cls_dn_4: 0.0872, loss_box_dn_4: 0.6149, loss_cls_dn_5: 0.0894, loss_box_dn_5: 0.6121, loss_dense_depth: 0.7328, loss: 22.8696, grad_norm: 31.1719 -2025-11-12 14:42:06,176 - mmdet - INFO - Iter [412/17500] lr: 2.640e-04, eta: 9:16:59, time: 1.574, data_time: 0.080, memory: 49164, loss_cls_0: 0.7099, loss_box_0: 1.5627, loss_cns_0: 0.6466, loss_yns_0: 0.1386, loss_cls_1: 0.7591, loss_box_1: 1.3880, loss_cns_1: 0.6637, loss_yns_1: 0.1325, loss_cls_2: 0.7715, loss_box_2: 1.3559, loss_cns_2: 0.6656, loss_yns_2: 0.1317, loss_cls_3: 0.7766, loss_box_3: 1.3433, loss_cns_3: 0.6628, loss_yns_3: 0.1321, loss_cls_4: 0.7771, loss_box_4: 1.3384, loss_cns_4: 0.6621, loss_yns_4: 0.1330, loss_cls_5: 0.7799, loss_box_5: 1.3377, loss_cns_5: 0.6650, loss_yns_5: 0.1325, loss_cls_dn_0: 0.1150, loss_box_dn_0: 0.7219, loss_cls_dn_1: 0.0890, loss_box_dn_1: 0.6246, loss_cls_dn_2: 0.0885, loss_box_dn_2: 0.6080, loss_cls_dn_3: 0.0868, loss_box_dn_3: 0.6052, loss_cls_dn_4: 0.0872, loss_box_dn_4: 0.6031, loss_cls_dn_5: 0.0868, loss_box_dn_5: 0.6043, loss_dense_depth: 0.6813, loss: 22.6680, grad_norm: 27.7342 -2025-11-12 14:42:07,744 - mmdet - INFO - Iter [413/17500] lr: 2.644e-04, eta: 9:16:41, time: 1.571, data_time: 0.078, memory: 49164, loss_cls_0: 0.7091, loss_box_0: 1.5517, loss_cns_0: 0.6435, loss_yns_0: 0.1395, loss_cls_1: 0.7687, loss_box_1: 1.3859, loss_cns_1: 0.6646, loss_yns_1: 0.1355, loss_cls_2: 0.7800, loss_box_2: 1.3727, loss_cns_2: 0.6689, loss_yns_2: 0.1359, loss_cls_3: 0.7859, loss_box_3: 1.3587, loss_cns_3: 0.6653, loss_yns_3: 0.1354, loss_cls_4: 0.7923, loss_box_4: 1.3678, loss_cns_4: 0.6655, loss_yns_4: 0.1352, loss_cls_5: 0.8065, loss_box_5: 1.3612, loss_cns_5: 0.6679, loss_yns_5: 0.1341, loss_cls_dn_0: 0.1140, loss_box_dn_0: 0.7137, loss_cls_dn_1: 0.0887, loss_box_dn_1: 0.6155, loss_cls_dn_2: 0.0870, loss_box_dn_2: 0.6012, loss_cls_dn_3: 0.0851, loss_box_dn_3: 0.5989, loss_cls_dn_4: 0.0868, loss_box_dn_4: 0.5995, loss_cls_dn_5: 0.0877, loss_box_dn_5: 0.6015, loss_dense_depth: 0.6915, loss: 22.8028, grad_norm: 39.3093 -2025-11-12 14:42:09,357 - mmdet - INFO - Iter [414/17500] lr: 2.648e-04, eta: 9:16:25, time: 1.609, data_time: 0.076, memory: 49164, loss_cls_0: 0.7227, loss_box_0: 1.5812, loss_cns_0: 0.6435, loss_yns_0: 0.1419, loss_cls_1: 0.7766, loss_box_1: 1.3981, loss_cns_1: 0.6668, loss_yns_1: 0.1366, loss_cls_2: 0.7875, loss_box_2: 1.3562, loss_cns_2: 0.6629, loss_yns_2: 0.1351, loss_cls_3: 0.7890, loss_box_3: 1.3541, loss_cns_3: 0.6638, loss_yns_3: 0.1360, loss_cls_4: 0.7962, loss_box_4: 1.3563, loss_cns_4: 0.6663, loss_yns_4: 0.1371, loss_cls_5: 0.8068, loss_box_5: 1.3505, loss_cns_5: 0.6659, loss_yns_5: 0.1363, loss_cls_dn_0: 0.1196, loss_box_dn_0: 0.7221, loss_cls_dn_1: 0.0935, loss_box_dn_1: 0.6264, loss_cls_dn_2: 0.0919, loss_box_dn_2: 0.6079, loss_cls_dn_3: 0.0888, loss_box_dn_3: 0.6055, loss_cls_dn_4: 0.0892, loss_box_dn_4: 0.6056, loss_cls_dn_5: 0.0912, loss_box_dn_5: 0.6064, loss_dense_depth: 0.6872, loss: 22.9025, grad_norm: 27.9227 -2025-11-12 14:42:10,922 - mmdet - INFO - Iter [415/17500] lr: 2.652e-04, eta: 9:16:08, time: 1.573, data_time: 0.075, memory: 49164, loss_cls_0: 0.7151, loss_box_0: 1.5060, loss_cns_0: 0.6446, loss_yns_0: 0.1374, loss_cls_1: 0.7634, loss_box_1: 1.3646, loss_cns_1: 0.6652, loss_yns_1: 0.1382, loss_cls_2: 0.7774, loss_box_2: 1.3455, loss_cns_2: 0.6641, loss_yns_2: 0.1368, loss_cls_3: 0.7800, loss_box_3: 1.3208, loss_cns_3: 0.6652, loss_yns_3: 0.1386, loss_cls_4: 0.7881, loss_box_4: 1.3375, loss_cns_4: 0.6665, loss_yns_4: 0.1399, loss_cls_5: 0.7960, loss_box_5: 1.3222, loss_cns_5: 0.6660, loss_yns_5: 0.1410, loss_cls_dn_0: 0.1169, loss_box_dn_0: 0.7118, loss_cls_dn_1: 0.0917, loss_box_dn_1: 0.6209, loss_cls_dn_2: 0.0892, loss_box_dn_2: 0.6085, loss_cls_dn_3: 0.0883, loss_box_dn_3: 0.5996, loss_cls_dn_4: 0.0875, loss_box_dn_4: 0.6049, loss_cls_dn_5: 0.0882, loss_box_dn_5: 0.6021, loss_dense_depth: 0.6731, loss: 22.6028, grad_norm: 32.0566 -2025-11-12 14:42:12,483 - mmdet - INFO - Iter [416/17500] lr: 2.656e-04, eta: 9:15:49, time: 1.560, data_time: 0.072, memory: 49164, loss_cls_0: 0.7035, loss_box_0: 1.5377, loss_cns_0: 0.6432, loss_yns_0: 0.1389, loss_cls_1: 0.7461, loss_box_1: 1.3888, loss_cns_1: 0.6673, loss_yns_1: 0.1380, loss_cls_2: 0.7540, loss_box_2: 1.3533, loss_cns_2: 0.6661, loss_yns_2: 0.1368, loss_cls_3: 0.7632, loss_box_3: 1.3336, loss_cns_3: 0.6648, loss_yns_3: 0.1378, loss_cls_4: 0.7682, loss_box_4: 1.3521, loss_cns_4: 0.6649, loss_yns_4: 0.1392, loss_cls_5: 0.7774, loss_box_5: 1.3320, loss_cns_5: 0.6653, loss_yns_5: 0.1370, loss_cls_dn_0: 0.1167, loss_box_dn_0: 0.7157, loss_cls_dn_1: 0.0912, loss_box_dn_1: 0.6197, loss_cls_dn_2: 0.0900, loss_box_dn_2: 0.6022, loss_cls_dn_3: 0.0926, loss_box_dn_3: 0.5966, loss_cls_dn_4: 0.0916, loss_box_dn_4: 0.5977, loss_cls_dn_5: 0.0956, loss_box_dn_5: 0.5944, loss_dense_depth: 0.6767, loss: 22.5898, grad_norm: 28.9323 -2025-11-12 14:42:14,060 - mmdet - INFO - Iter [417/17500] lr: 2.660e-04, eta: 9:15:32, time: 1.573, data_time: 0.071, memory: 49164, loss_cls_0: 0.7324, loss_box_0: 1.5408, loss_cns_0: 0.6428, loss_yns_0: 0.1407, loss_cls_1: 0.7666, loss_box_1: 1.3897, loss_cns_1: 0.6644, loss_yns_1: 0.1382, loss_cls_2: 0.7722, loss_box_2: 1.3692, loss_cns_2: 0.6656, loss_yns_2: 0.1368, loss_cls_3: 0.7762, loss_box_3: 1.3406, loss_cns_3: 0.6599, loss_yns_3: 0.1364, loss_cls_4: 0.7716, loss_box_4: 1.3680, loss_cns_4: 0.6625, loss_yns_4: 0.1382, loss_cls_5: 0.7843, loss_box_5: 1.3582, loss_cns_5: 0.6654, loss_yns_5: 0.1378, loss_cls_dn_0: 0.1182, loss_box_dn_0: 0.7162, loss_cls_dn_1: 0.0964, loss_box_dn_1: 0.6180, loss_cls_dn_2: 0.0950, loss_box_dn_2: 0.6049, loss_cls_dn_3: 0.0971, loss_box_dn_3: 0.6006, loss_cls_dn_4: 0.0947, loss_box_dn_4: 0.6021, loss_cls_dn_5: 0.0999, loss_box_dn_5: 0.6021, loss_dense_depth: 0.6919, loss: 22.7956, grad_norm: 30.8965 -2025-11-12 14:42:15,639 - mmdet - INFO - Iter [418/17500] lr: 2.664e-04, eta: 9:15:15, time: 1.584, data_time: 0.076, memory: 49164, loss_cls_0: 0.6967, loss_box_0: 1.5435, loss_cns_0: 0.6419, loss_yns_0: 0.1400, loss_cls_1: 0.7568, loss_box_1: 1.3769, loss_cns_1: 0.6660, loss_yns_1: 0.1380, loss_cls_2: 0.7591, loss_box_2: 1.3562, loss_cns_2: 0.6640, loss_yns_2: 0.1389, loss_cls_3: 0.7737, loss_box_3: 1.3448, loss_cns_3: 0.6607, loss_yns_3: 0.1383, loss_cls_4: 0.7778, loss_box_4: 1.3383, loss_cns_4: 0.6619, loss_yns_4: 0.1389, loss_cls_5: 0.7856, loss_box_5: 1.3307, loss_cns_5: 0.6624, loss_yns_5: 0.1383, loss_cls_dn_0: 0.1097, loss_box_dn_0: 0.6989, loss_cls_dn_1: 0.0926, loss_box_dn_1: 0.6120, loss_cls_dn_2: 0.0912, loss_box_dn_2: 0.5996, loss_cls_dn_3: 0.0915, loss_box_dn_3: 0.5960, loss_cls_dn_4: 0.0927, loss_box_dn_4: 0.5945, loss_cls_dn_5: 0.0934, loss_box_dn_5: 0.5969, loss_dense_depth: 0.6803, loss: 22.5788, grad_norm: 28.0982 -2025-11-12 14:42:17,231 - mmdet - INFO - Iter [419/17500] lr: 2.668e-04, eta: 9:14:58, time: 1.586, data_time: 0.074, memory: 49164, loss_cls_0: 0.7514, loss_box_0: 1.5892, loss_cns_0: 0.6322, loss_yns_0: 0.1386, loss_cls_1: 0.7812, loss_box_1: 1.4162, loss_cns_1: 0.6641, loss_yns_1: 0.1367, loss_cls_2: 0.7879, loss_box_2: 1.3930, loss_cns_2: 0.6624, loss_yns_2: 0.1365, loss_cls_3: 0.7972, loss_box_3: 1.3804, loss_cns_3: 0.6603, loss_yns_3: 0.1367, loss_cls_4: 0.8100, loss_box_4: 1.3840, loss_cns_4: 0.6609, loss_yns_4: 0.1359, loss_cls_5: 0.8138, loss_box_5: 1.3750, loss_cns_5: 0.6584, loss_yns_5: 0.1350, loss_cls_dn_0: 0.1187, loss_box_dn_0: 0.7042, loss_cls_dn_1: 0.0921, loss_box_dn_1: 0.6218, loss_cls_dn_2: 0.0911, loss_box_dn_2: 0.6093, loss_cls_dn_3: 0.0908, loss_box_dn_3: 0.6071, loss_cls_dn_4: 0.0914, loss_box_dn_4: 0.6110, loss_cls_dn_5: 0.0918, loss_box_dn_5: 0.6102, loss_dense_depth: 0.7442, loss: 23.1204, grad_norm: 32.3790 -2025-11-12 14:42:18,815 - mmdet - INFO - Iter [420/17500] lr: 2.672e-04, eta: 9:14:42, time: 1.588, data_time: 0.078, memory: 49164, loss_cls_0: 0.7257, loss_box_0: 1.5869, loss_cns_0: 0.6344, loss_yns_0: 0.1374, loss_cls_1: 0.7777, loss_box_1: 1.3993, loss_cns_1: 0.6648, loss_yns_1: 0.1342, loss_cls_2: 0.7842, loss_box_2: 1.3690, loss_cns_2: 0.6653, loss_yns_2: 0.1347, loss_cls_3: 0.7911, loss_box_3: 1.3529, loss_cns_3: 0.6648, loss_yns_3: 0.1341, loss_cls_4: 0.7959, loss_box_4: 1.3681, loss_cns_4: 0.6645, loss_yns_4: 0.1354, loss_cls_5: 0.8038, loss_box_5: 1.3535, loss_cns_5: 0.6634, loss_yns_5: 0.1339, loss_cls_dn_0: 0.1176, loss_box_dn_0: 0.7176, loss_cls_dn_1: 0.0936, loss_box_dn_1: 0.6330, loss_cls_dn_2: 0.0926, loss_box_dn_2: 0.6181, loss_cls_dn_3: 0.0933, loss_box_dn_3: 0.6146, loss_cls_dn_4: 0.0933, loss_box_dn_4: 0.6208, loss_cls_dn_5: 0.0952, loss_box_dn_5: 0.6191, loss_dense_depth: 0.7335, loss: 23.0170, grad_norm: 38.6310 -2025-11-12 14:42:20,461 - mmdet - INFO - Iter [421/17500] lr: 2.676e-04, eta: 9:14:27, time: 1.642, data_time: 0.103, memory: 49164, loss_cls_0: 0.7409, loss_box_0: 1.6072, loss_cns_0: 0.6394, loss_yns_0: 0.1386, loss_cls_1: 0.7847, loss_box_1: 1.4344, loss_cns_1: 0.6658, loss_yns_1: 0.1362, loss_cls_2: 0.7902, loss_box_2: 1.3959, loss_cns_2: 0.6683, loss_yns_2: 0.1359, loss_cls_3: 0.7965, loss_box_3: 1.3928, loss_cns_3: 0.6662, loss_yns_3: 0.1353, loss_cls_4: 0.8018, loss_box_4: 1.3902, loss_cns_4: 0.6643, loss_yns_4: 0.1373, loss_cls_5: 0.8066, loss_box_5: 1.3887, loss_cns_5: 0.6662, loss_yns_5: 0.1345, loss_cls_dn_0: 0.1183, loss_box_dn_0: 0.7096, loss_cls_dn_1: 0.0929, loss_box_dn_1: 0.6507, loss_cls_dn_2: 0.0927, loss_box_dn_2: 0.6294, loss_cls_dn_3: 0.0920, loss_box_dn_3: 0.6266, loss_cls_dn_4: 0.0901, loss_box_dn_4: 0.6219, loss_cls_dn_5: 0.0933, loss_box_dn_5: 0.6270, loss_dense_depth: 0.6870, loss: 23.2493, grad_norm: 32.8775 -2025-11-12 14:42:22,088 - mmdet - INFO - Iter [422/17500] lr: 2.680e-04, eta: 9:14:13, time: 1.634, data_time: 0.099, memory: 49164, loss_cls_0: 0.7057, loss_box_0: 1.6115, loss_cns_0: 0.6360, loss_yns_0: 0.1364, loss_cls_1: 0.7662, loss_box_1: 1.3852, loss_cns_1: 0.6612, loss_yns_1: 0.1333, loss_cls_2: 0.7780, loss_box_2: 1.3784, loss_cns_2: 0.6620, loss_yns_2: 0.1316, loss_cls_3: 0.7818, loss_box_3: 1.3835, loss_cns_3: 0.6632, loss_yns_3: 0.1310, loss_cls_4: 0.7743, loss_box_4: 1.3840, loss_cns_4: 0.6642, loss_yns_4: 0.1334, loss_cls_5: 0.7864, loss_box_5: 1.3827, loss_cns_5: 0.6686, loss_yns_5: 0.1327, loss_cls_dn_0: 0.1131, loss_box_dn_0: 0.7139, loss_cls_dn_1: 0.0924, loss_box_dn_1: 0.6216, loss_cls_dn_2: 0.0901, loss_box_dn_2: 0.6141, loss_cls_dn_3: 0.0920, loss_box_dn_3: 0.6131, loss_cls_dn_4: 0.0883, loss_box_dn_4: 0.6122, loss_cls_dn_5: 0.0924, loss_box_dn_5: 0.6129, loss_dense_depth: 0.7136, loss: 22.9410, grad_norm: 45.4513 -2025-11-12 14:42:23,682 - mmdet - INFO - Iter [423/17500] lr: 2.684e-04, eta: 9:13:56, time: 1.588, data_time: 0.085, memory: 49164, loss_cls_0: 0.7078, loss_box_0: 1.5752, loss_cns_0: 0.6418, loss_yns_0: 0.1407, loss_cls_1: 0.7619, loss_box_1: 1.3610, loss_cns_1: 0.6642, loss_yns_1: 0.1343, loss_cls_2: 0.7824, loss_box_2: 1.3384, loss_cns_2: 0.6653, loss_yns_2: 0.1332, loss_cls_3: 0.7808, loss_box_3: 1.3523, loss_cns_3: 0.6679, loss_yns_3: 0.1330, loss_cls_4: 0.7726, loss_box_4: 1.3572, loss_cns_4: 0.6689, loss_yns_4: 0.1335, loss_cls_5: 0.7808, loss_box_5: 1.3652, loss_cns_5: 0.6703, loss_yns_5: 0.1338, loss_cls_dn_0: 0.1131, loss_box_dn_0: 0.7089, loss_cls_dn_1: 0.0941, loss_box_dn_1: 0.6155, loss_cls_dn_2: 0.0912, loss_box_dn_2: 0.6054, loss_cls_dn_3: 0.0939, loss_box_dn_3: 0.6046, loss_cls_dn_4: 0.0915, loss_box_dn_4: 0.6068, loss_cls_dn_5: 0.0924, loss_box_dn_5: 0.6122, loss_dense_depth: 0.6785, loss: 22.7311, grad_norm: 39.7717 -2025-11-12 14:42:25,287 - mmdet - INFO - Iter [424/17500] lr: 2.688e-04, eta: 9:13:41, time: 1.609, data_time: 0.082, memory: 49164, loss_cls_0: 0.7554, loss_box_0: 1.6186, loss_cns_0: 0.6405, loss_yns_0: 0.1400, loss_cls_1: 0.7778, loss_box_1: 1.4302, loss_cns_1: 0.6644, loss_yns_1: 0.1328, loss_cls_2: 0.7937, loss_box_2: 1.3974, loss_cns_2: 0.6676, loss_yns_2: 0.1318, loss_cls_3: 0.7937, loss_box_3: 1.4138, loss_cns_3: 0.6686, loss_yns_3: 0.1324, loss_cls_4: 0.8020, loss_box_4: 1.4139, loss_cns_4: 0.6707, loss_yns_4: 0.1333, loss_cls_5: 0.8154, loss_box_5: 1.4094, loss_cns_5: 0.6706, loss_yns_5: 0.1318, loss_cls_dn_0: 0.1258, loss_box_dn_0: 0.7172, loss_cls_dn_1: 0.0949, loss_box_dn_1: 0.6462, loss_cls_dn_2: 0.0949, loss_box_dn_2: 0.6372, loss_cls_dn_3: 0.0932, loss_box_dn_3: 0.6357, loss_cls_dn_4: 0.0946, loss_box_dn_4: 0.6320, loss_cls_dn_5: 0.0952, loss_box_dn_5: 0.6350, loss_dense_depth: 0.7113, loss: 23.4186, grad_norm: 43.8958 -2025-11-12 14:42:26,851 - mmdet - INFO - Iter [425/17500] lr: 2.692e-04, eta: 9:13:23, time: 1.565, data_time: 0.079, memory: 49164, loss_cls_0: 0.7286, loss_box_0: 1.5938, loss_cns_0: 0.6471, loss_yns_0: 0.1377, loss_cls_1: 0.7831, loss_box_1: 1.4264, loss_cns_1: 0.6690, loss_yns_1: 0.1305, loss_cls_2: 0.7839, loss_box_2: 1.3983, loss_cns_2: 0.6697, loss_yns_2: 0.1306, loss_cls_3: 0.7764, loss_box_3: 1.3923, loss_cns_3: 0.6688, loss_yns_3: 0.1309, loss_cls_4: 0.7815, loss_box_4: 1.3808, loss_cns_4: 0.6682, loss_yns_4: 0.1317, loss_cls_5: 0.7909, loss_box_5: 1.3716, loss_cns_5: 0.6706, loss_yns_5: 0.1297, loss_cls_dn_0: 0.1179, loss_box_dn_0: 0.7163, loss_cls_dn_1: 0.0980, loss_box_dn_1: 0.6450, loss_cls_dn_2: 0.1001, loss_box_dn_2: 0.6354, loss_cls_dn_3: 0.1004, loss_box_dn_3: 0.6300, loss_cls_dn_4: 0.1013, loss_box_dn_4: 0.6208, loss_cls_dn_5: 0.1020, loss_box_dn_5: 0.6204, loss_dense_depth: 0.6870, loss: 23.1668, grad_norm: 35.1808 -2025-11-12 14:42:28,430 - mmdet - INFO - Iter [426/17500] lr: 2.696e-04, eta: 9:13:07, time: 1.576, data_time: 0.101, memory: 49164, loss_cls_0: 0.7185, loss_box_0: 1.5479, loss_cns_0: 0.6448, loss_yns_0: 0.1381, loss_cls_1: 0.7739, loss_box_1: 1.4158, loss_cns_1: 0.6665, loss_yns_1: 0.1342, loss_cls_2: 0.7892, loss_box_2: 1.3771, loss_cns_2: 0.6645, loss_yns_2: 0.1327, loss_cls_3: 0.7897, loss_box_3: 1.3893, loss_cns_3: 0.6631, loss_yns_3: 0.1348, loss_cls_4: 0.7923, loss_box_4: 1.3748, loss_cns_4: 0.6642, loss_yns_4: 0.1326, loss_cls_5: 0.8105, loss_box_5: 1.3806, loss_cns_5: 0.6623, loss_yns_5: 0.1316, loss_cls_dn_0: 0.1089, loss_box_dn_0: 0.7103, loss_cls_dn_1: 0.0963, loss_box_dn_1: 0.6355, loss_cls_dn_2: 0.0997, loss_box_dn_2: 0.6181, loss_cls_dn_3: 0.1004, loss_box_dn_3: 0.6178, loss_cls_dn_4: 0.0980, loss_box_dn_4: 0.6154, loss_cls_dn_5: 0.1009, loss_box_dn_5: 0.6232, loss_dense_depth: 0.6526, loss: 23.0062, grad_norm: 42.5162 -2025-11-12 14:42:30,008 - mmdet - INFO - Iter [427/17500] lr: 2.700e-04, eta: 9:12:50, time: 1.579, data_time: 0.072, memory: 49164, loss_cls_0: 0.7203, loss_box_0: 1.5620, loss_cns_0: 0.6383, loss_yns_0: 0.1383, loss_cls_1: 0.7675, loss_box_1: 1.4021, loss_cns_1: 0.6637, loss_yns_1: 0.1357, loss_cls_2: 0.7844, loss_box_2: 1.3707, loss_cns_2: 0.6616, loss_yns_2: 0.1356, loss_cls_3: 0.7844, loss_box_3: 1.3757, loss_cns_3: 0.6631, loss_yns_3: 0.1359, loss_cls_4: 0.7945, loss_box_4: 1.3610, loss_cns_4: 0.6629, loss_yns_4: 0.1341, loss_cls_5: 0.8087, loss_box_5: 1.3511, loss_cns_5: 0.6581, loss_yns_5: 0.1343, loss_cls_dn_0: 0.1143, loss_box_dn_0: 0.7070, loss_cls_dn_1: 0.0960, loss_box_dn_1: 0.6274, loss_cls_dn_2: 0.0980, loss_box_dn_2: 0.6121, loss_cls_dn_3: 0.0980, loss_box_dn_3: 0.6183, loss_cls_dn_4: 0.0936, loss_box_dn_4: 0.6214, loss_cls_dn_5: 0.0963, loss_box_dn_5: 0.6258, loss_dense_depth: 0.6808, loss: 22.9330, grad_norm: 36.8594 diff --git a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20251112_142824.log.json b/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20251112_142824.log.json deleted file mode 100644 index 338a9051b9b234e816a9250848f96a1dc3fa3d92..0000000000000000000000000000000000000000 --- a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20251112_142824.log.json +++ /dev/null @@ -1,428 +0,0 @@ -{"env_info": "sys.platform: linux\nPython: 3.10.12 (main, May 27 2025, 17:12:29) [GCC 11.4.0]\nCUDA available: True\nGPU 0,1,2,3,4,5,6,7: BW200, UBB BW1000\nCUDA_HOME: /opt/dtk\nNVCC: Not Available\nGCC: x86_64-linux-gnu-gcc (Ubuntu 11.4.0-1ubuntu1~22.04) 11.4.0\nPyTorch: 2.4.1\nPyTorch compiling details: PyTorch built with:\n - GCC 10.3\n - C++ Version: 201703\n - Intel(R) Math Kernel Library Version 2020.0.4 Product Build 20200917 for Intel(R) 64 architecture applications\n - OpenMP 201511 (a.k.a. OpenMP 4.5)\n - LAPACK is enabled (usually provided by MKL)\n - NNPACK is enabled\n - CPU capability usage: AVX512\n - HIP Runtime 6.3.25211\n - MIOpen 2.17.0\n - Magma 2.8.0\n - Build settings: BLAS_INFO=mkl, BUILD_TYPE=Release, CXX_COMPILER=/opt/rh/gcc-toolset-10/root/usr/bin/c++, CXX_FLAGS= -D_GLIBCXX_USE_CXX11_ABI=1 -fvisibility-inlines-hidden -DUSE_PTHREADPOOL -DNDEBUG -DUSE_KINETO -DLIBKINETO_NOCUPTI -DUSE_FBGEMM -DUSE_PYTORCH_QNNPACK -DUSE_XNNPACK -DSYMBOLICATE_MOBILE_DEBUG_HANDLE -O2 -fPIC -Wall -Wextra -Werror=return-type -Werror=non-virtual-dtor -Werror=bool-operation -Wnarrowing -Wno-missing-field-initializers -Wno-type-limits -Wno-array-bounds -Wno-unknown-pragmas -Wno-unused-parameter -Wno-unused-function -Wno-unused-result -Wno-strict-overflow -Wno-strict-aliasing -Wno-stringop-overflow -Wsuggest-override -Wno-psabi -Wno-error=pedantic -Wno-error=old-style-cast -Wno-missing-braces -fdiagnostics-color=always -faligned-new -Wno-unused-but-set-variable -Wno-maybe-uninitialized -fno-math-errno -fno-trapping-math -Werror=format -Wno-stringop-overflow, FORCE_FALLBACK_CUDA_MPI=1, LAPACK_INFO=mkl, PERF_WITH_AVX=1, PERF_WITH_AVX2=1, PERF_WITH_AVX512=1, TORCH_VERSION=2.4.1, USE_CUDA=0, USE_CUDNN=OFF, USE_CUSPARSELT=OFF, USE_EXCEPTION_PTR=1, USE_GFLAGS=1, USE_GLOG=1, USE_GLOO=1, USE_MKL=ON, USE_MKLDNN=0, USE_MPI=1, USE_NCCL=1, USE_NNPACK=ON, USE_OPENMP=1, USE_ROCM=ON, USE_ROCM_KERNEL_ASSERT=OFF, \n\nTorchVision: 0.19.1\nOpenCV: 4.11.0\nMMCV: 1.6.1\nMMCV Compiler: GCC 11.4\nMMCV CUDA Compiler: rocm not available\nMMDetection: 2.26.0+c41df4b", "config": "plugin = True\nplugin_dir = 'projects/mmdet3d_plugin/'\ndist_params = dict(backend='nccl')\nlog_level = 'INFO'\nwork_dir = './work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704'\ntotal_batch_size = 160\nnum_gpus = 8\nbatch_size = 20\nnum_iters_per_epoch = 175\nnum_epochs = 100\ncheckpoint_epoch_interval = 20\ncheckpoint_config = dict(interval=3500)\nlog_config = dict(\n interval=1,\n hooks=[\n dict(type='TextLoggerHook', by_epoch=False),\n dict(type='TensorboardLoggerHook')\n ])\nload_from = None\nresume_from = None\nworkflow = [('train', 1)]\nfp16 = dict(loss_scale=32.0)\ninput_shape = (704, 256)\ntracking_test = True\ntracking_threshold = 0.2\nclass_names = [\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier',\n 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n]\nnum_classes = 10\nembed_dims = 256\nnum_groups = 8\nnum_decoder = 6\nnum_single_frame_decoder = 1\nuse_deformable_func = True\nstrides = [4, 8, 16, 32]\nnum_levels = 4\nnum_depth_layers = 3\ndrop_out = 0.1\ntemporal = True\ndecouple_attn = True\nwith_quality_estimation = True\nmodel = dict(\n type='Sparse4D',\n use_grid_mask=True,\n use_deformable_func=True,\n img_backbone=dict(\n type='ResNet',\n depth=50,\n num_stages=4,\n frozen_stages=-1,\n norm_eval=False,\n style='pytorch',\n with_cp=True,\n out_indices=(0, 1, 2, 3),\n norm_cfg=dict(type='BN', requires_grad=True),\n pretrained='ckpt/resnet50-19c8e357.pth'),\n img_neck=dict(\n type='FPN',\n num_outs=4,\n start_level=0,\n out_channels=256,\n add_extra_convs='on_output',\n relu_before_extra_convs=True,\n in_channels=[256, 512, 1024, 2048]),\n depth_branch=dict(\n type='DenseDepthNet',\n embed_dims=256,\n num_depth_layers=3,\n loss_weight=0.2),\n head=dict(\n type='Sparse4DHead',\n cls_threshold_to_reg=0.05,\n decouple_attn=True,\n instance_bank=dict(\n type='InstanceBank',\n num_anchor=900,\n embed_dims=256,\n anchor='nuscenes_kmeans900.npy',\n anchor_handler=dict(type='SparseBox3DKeyPointsGenerator'),\n num_temp_instances=600,\n confidence_decay=0.6,\n feat_grad=False),\n anchor_encoder=dict(\n type='SparseBox3DEncoder',\n vel_dims=3,\n embed_dims=[128, 32, 32, 64],\n mode='cat',\n output_fc=False,\n in_loops=1,\n out_loops=4),\n num_single_frame_decoder=1,\n operation_order=[\n 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm',\n 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm',\n 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm',\n 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm',\n 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm',\n 'deformable', 'ffn', 'norm', 'refine'\n ],\n temp_graph_model=dict(\n type='MultiheadAttention',\n embed_dims=512,\n num_heads=8,\n batch_first=True,\n dropout=0.1),\n graph_model=dict(\n type='MultiheadAttention',\n embed_dims=512,\n num_heads=8,\n batch_first=True,\n dropout=0.1),\n norm_layer=dict(type='LN', normalized_shape=256),\n ffn=dict(\n type='AsymmetricFFN',\n in_channels=512,\n pre_norm=dict(type='LN'),\n embed_dims=256,\n feedforward_channels=1024,\n num_fcs=2,\n ffn_drop=0.1,\n act_cfg=dict(type='ReLU', inplace=True)),\n deformable_model=dict(\n type='DeformableFeatureAggregation',\n embed_dims=256,\n num_groups=8,\n num_levels=4,\n num_cams=6,\n attn_drop=0.15,\n use_deformable_func=True,\n use_camera_embed=True,\n residual_mode='cat',\n kps_generator=dict(\n type='SparseBox3DKeyPointsGenerator',\n num_learnable_pts=6,\n fix_scale=[[0, 0, 0], [0.45, 0, 0], [-0.45, 0, 0],\n [0, 0.45, 0], [0, -0.45, 0], [0, 0, 0.45],\n [0, 0, -0.45]])),\n refine_layer=dict(\n type='SparseBox3DRefinementModule',\n embed_dims=256,\n num_cls=10,\n refine_yaw=True,\n with_quality_estimation=True),\n sampler=dict(\n type='SparseBox3DTarget',\n num_dn_groups=5,\n num_temp_dn_groups=3,\n dn_noise_scale=[2.0, 2.0, 2.0, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5],\n max_dn_gt=32,\n add_neg_dn=True,\n cls_weight=2.0,\n box_weight=0.25,\n reg_weights=[2.0, 2.0, 2.0, 0.5, 0.5, 0.5, 0.0, 0.0, 0.0, 0.0],\n cls_wise_reg_weights=dict(\n {9: [2.0, 2.0, 2.0, 1.0, 1.0, 1.0, 0.0, 0.0, 1.0, 1.0]})),\n loss_cls=dict(\n type='FocalLoss',\n use_sigmoid=True,\n gamma=2.0,\n alpha=0.25,\n loss_weight=2.0),\n loss_reg=dict(\n type='SparseBox3DLoss',\n loss_box=dict(type='L1Loss', loss_weight=0.25),\n loss_centerness=dict(type='CrossEntropyLoss', use_sigmoid=True),\n loss_yawness=dict(type='GaussianFocalLoss'),\n cls_allow_reverse=[5]),\n decoder=dict(type='SparseBox3DDecoder'),\n reg_weights=[2.0, 2.0, 2.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0]))\ndataset_type = 'NuScenes3DDetTrackDataset'\ndata_root = 'data/nuscenes/'\nanno_root = 'data/nuscenes_anno_pkls/'\nfile_client_args = dict(backend='disk')\nimg_norm_cfg = dict(\n mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375], to_rgb=True)\ntrain_pipeline = [\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(\n type='LoadPointsFromFile',\n coord_type='LIDAR',\n load_dim=5,\n use_dim=5,\n file_client_args=dict(backend='disk')),\n dict(type='ResizeCropFlipImage'),\n dict(type='MultiScaleDepthMapGenerator', downsample=[4, 8, 16]),\n dict(type='BBoxRotation'),\n dict(type='PhotoMetricDistortionMultiViewImage'),\n dict(\n type='NormalizeMultiviewImage',\n mean=[123.675, 116.28, 103.53],\n std=[58.395, 57.12, 57.375],\n to_rgb=True),\n dict(\n type='CircleObjectRangeFilter',\n class_dist_thred=[55, 55, 55, 55, 55, 55, 55, 55, 55, 55]),\n dict(\n type='InstanceNameFilter',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ]),\n dict(type='NuScenesSparse4DAdaptor'),\n dict(\n type='Collect',\n keys=[\n 'img', 'timestamp', 'projection_mat', 'image_wh', 'gt_depth',\n 'focal', 'gt_bboxes_3d', 'gt_labels_3d'\n ],\n meta_keys=['T_global', 'T_global_inv', 'timestamp', 'instance_id'])\n]\ntest_pipeline = [\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(type='ResizeCropFlipImage'),\n dict(\n type='NormalizeMultiviewImage',\n mean=[123.675, 116.28, 103.53],\n std=[58.395, 57.12, 57.375],\n to_rgb=True),\n dict(type='NuScenesSparse4DAdaptor'),\n dict(\n type='Collect',\n keys=['img', 'timestamp', 'projection_mat', 'image_wh'],\n meta_keys=['T_global', 'T_global_inv', 'timestamp'])\n]\ninput_modality = dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False)\ndata_basic_config = dict(\n type='NuScenes3DDetTrackDataset',\n data_root='data/nuscenes/',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier',\n 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ],\n modality=dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False),\n version='v1.0-trainval')\ndata_aug_conf = dict(\n resize_lim=(0.4, 0.47),\n final_dim=(256, 704),\n bot_pct_lim=(0.0, 0.0),\n rot_lim=(-5.4, 5.4),\n H=900,\n W=1600,\n rand_flip=True,\n rot3d_range=[-0.3925, 0.3925])\ndata = dict(\n samples_per_gpu=20,\n workers_per_gpu=20,\n train=dict(\n type='NuScenes3DDetTrackDataset',\n data_root='data/nuscenes/',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ],\n modality=dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False),\n version='v1.0-trainval',\n ann_file='data/nuscenes_anno_pkls/nuscenes_infos_train.pkl',\n pipeline=[\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(\n type='LoadPointsFromFile',\n coord_type='LIDAR',\n load_dim=5,\n use_dim=5,\n file_client_args=dict(backend='disk')),\n dict(type='ResizeCropFlipImage'),\n dict(type='MultiScaleDepthMapGenerator', downsample=[4, 8, 16]),\n dict(type='BBoxRotation'),\n dict(type='PhotoMetricDistortionMultiViewImage'),\n dict(\n type='NormalizeMultiviewImage',\n mean=[123.675, 116.28, 103.53],\n std=[58.395, 57.12, 57.375],\n to_rgb=True),\n dict(\n type='CircleObjectRangeFilter',\n class_dist_thred=[55, 55, 55, 55, 55, 55, 55, 55, 55, 55]),\n dict(\n type='InstanceNameFilter',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian',\n 'traffic_cone'\n ]),\n dict(type='NuScenesSparse4DAdaptor'),\n dict(\n type='Collect',\n keys=[\n 'img', 'timestamp', 'projection_mat', 'image_wh',\n 'gt_depth', 'focal', 'gt_bboxes_3d', 'gt_labels_3d'\n ],\n meta_keys=[\n 'T_global', 'T_global_inv', 'timestamp', 'instance_id'\n ])\n ],\n test_mode=False,\n data_aug_conf=dict(\n resize_lim=(0.4, 0.47),\n final_dim=(256, 704),\n bot_pct_lim=(0.0, 0.0),\n rot_lim=(-5.4, 5.4),\n H=900,\n W=1600,\n rand_flip=True,\n rot3d_range=[-0.3925, 0.3925]),\n with_seq_flag=True,\n sequences_split_num=2,\n keep_consistent_seq_aug=True),\n val=dict(\n type='NuScenes3DDetTrackDataset',\n data_root='data/nuscenes/',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ],\n modality=dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False),\n version='v1.0-trainval',\n ann_file='data/nuscenes_anno_pkls/nuscenes_infos_val.pkl',\n pipeline=[\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(type='ResizeCropFlipImage'),\n dict(\n type='NormalizeMultiviewImage',\n mean=[123.675, 116.28, 103.53],\n std=[58.395, 57.12, 57.375],\n to_rgb=True),\n dict(type='NuScenesSparse4DAdaptor'),\n dict(\n type='Collect',\n keys=['img', 'timestamp', 'projection_mat', 'image_wh'],\n meta_keys=['T_global', 'T_global_inv', 'timestamp'])\n ],\n data_aug_conf=dict(\n resize_lim=(0.4, 0.47),\n final_dim=(256, 704),\n bot_pct_lim=(0.0, 0.0),\n rot_lim=(-5.4, 5.4),\n H=900,\n W=1600,\n rand_flip=True,\n rot3d_range=[-0.3925, 0.3925]),\n test_mode=True,\n tracking=True,\n tracking_threshold=0.2),\n test=dict(\n type='NuScenes3DDetTrackDataset',\n data_root='data/nuscenes/',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ],\n modality=dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False),\n version='v1.0-trainval',\n ann_file='data/nuscenes_anno_pkls/nuscenes_infos_val.pkl',\n pipeline=[\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(type='ResizeCropFlipImage'),\n dict(\n type='NormalizeMultiviewImage',\n mean=[123.675, 116.28, 103.53],\n std=[58.395, 57.12, 57.375],\n to_rgb=True),\n dict(type='NuScenesSparse4DAdaptor'),\n dict(\n type='Collect',\n keys=['img', 'timestamp', 'projection_mat', 'image_wh'],\n meta_keys=['T_global', 'T_global_inv', 'timestamp'])\n ],\n data_aug_conf=dict(\n resize_lim=(0.4, 0.47),\n final_dim=(256, 704),\n bot_pct_lim=(0.0, 0.0),\n rot_lim=(-5.4, 5.4),\n H=900,\n W=1600,\n rand_flip=True,\n rot3d_range=[-0.3925, 0.3925]),\n test_mode=True,\n tracking=True,\n tracking_threshold=0.2))\noptimizer = dict(\n type='AdamW',\n lr=0.0006,\n weight_decay=0.001,\n paramwise_cfg=dict(custom_keys=dict(img_backbone=dict(lr_mult=0.5))))\noptimizer_config = dict(grad_clip=dict(max_norm=25, norm_type=2))\nlr_config = dict(\n policy='CosineAnnealing',\n warmup='linear',\n warmup_iters=500,\n warmup_ratio=0.3333333333333333,\n min_lr_ratio=0.001)\nrunner = dict(type='IterBasedRunner', max_iters=17500)\nvis_pipeline = [\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(type='Collect', keys=['img'], meta_keys=['timestamp', 'lidar2img'])\n]\nevaluation = dict(\n interval=3500,\n pipeline=[\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(\n type='Collect', keys=['img'], meta_keys=['timestamp', 'lidar2img'])\n ])\ngpu_ids = range(0, 8)\n", "seed": 0, "exp_name": "sparse4dv3_temporal_r50_1x8_bs6_256x704.py"} -{"mode": "train", "epoch": 1, "iter": 1, "lr": 0.0001, "memory": 49164, "data_time": 9.03575, "loss_cls_0": 2.36122, "loss_box_0": 0.01384, "loss_cns_0": 0.0027, "loss_yns_0": 0.00079, "loss_cls_1": 2.15453, "loss_box_1": 0.10814, "loss_cns_1": 0.02455, "loss_yns_1": 0.00669, "loss_cls_2": 2.31212, "loss_box_2": 0.00504, "loss_cns_2": 0.00059, "loss_yns_2": 0.00029, "loss_cls_3": 2.3902, "loss_box_3": 0.02945, "loss_cns_3": 0.00504, "loss_yns_3": 0.00144, "loss_cls_4": 2.02811, "loss_box_4": 0.41793, "loss_cns_4": 0.05352, "loss_yns_4": 0.02533, "loss_cls_5": 2.42472, "loss_box_5": 0.01801, "loss_cns_5": 0.00218, "loss_yns_5": 0.0016, "loss_cls_dn_0": 1.19803, "loss_box_dn_0": 1.46028, "loss_cls_dn_1": 1.11018, "loss_box_dn_1": 1.73177, "loss_cls_dn_2": 1.17413, "loss_box_dn_2": 1.97185, "loss_cls_dn_3": 1.17206, "loss_box_dn_3": 2.24179, "loss_cls_dn_4": 1.05278, "loss_box_dn_4": 2.42682, "loss_cls_dn_5": 1.23867, "loss_box_dn_5": 2.67727, "loss_dense_depth": 1.86433, "loss": 35.70798, "grad_norm": 270.47369, "time": 116.84963} -{"mode": "train", "epoch": 1, "iter": 2, "lr": 0.0001, "memory": 49164, "data_time": 0.10606, "loss_cls_0": 2.0434, "loss_box_0": 0.024, "loss_cns_0": 0.0064, "loss_yns_0": 0.0023, "loss_cls_1": 2.03785, "loss_box_1": 0.12565, "loss_cns_1": 0.02589, "loss_yns_1": 0.00582, "loss_cls_2": 2.11184, "loss_box_2": 0.22715, "loss_cns_2": 0.02047, "loss_yns_2": 0.00961, "loss_cls_3": 1.95373, "loss_box_3": 0.45943, "loss_cns_3": 0.05932, "loss_yns_3": 0.02025, "loss_cls_4": 1.79759, "loss_box_4": 1.54584, "loss_cns_4": 0.1541, "loss_yns_4": 0.05507, "loss_cls_5": 2.05966, "loss_box_5": 0.56203, "loss_cns_5": 0.06146, "loss_yns_5": 0.01936, "loss_cls_dn_0": 1.02529, "loss_box_dn_0": 1.29865, "loss_cls_dn_1": 0.95867, "loss_box_dn_1": 2.41407, "loss_cls_dn_2": 0.97421, "loss_box_dn_2": 2.52632, "loss_cls_dn_3": 0.9109, "loss_box_dn_3": 2.60999, "loss_cls_dn_4": 0.84021, "loss_box_dn_4": 2.86447, "loss_cls_dn_5": 0.98659, "loss_box_dn_5": 3.10787, "loss_dense_depth": 1.7144, "loss": 37.61988, "grad_norm": 66.85527, "time": 2.019} -{"mode": "train", "epoch": 1, "iter": 3, "lr": 0.0001, "memory": 49164, "data_time": 0.08353, "loss_cls_0": 1.44521, "loss_box_0": 2.50225, "loss_cns_0": 0.61041, "loss_yns_0": 0.20695, "loss_cls_1": 1.76515, "loss_box_1": 1.7029, "loss_cns_1": 0.26675, "loss_yns_1": 0.09823, "loss_cls_2": 1.79105, "loss_box_2": 3.75972, "loss_cns_2": 0.35498, "loss_yns_2": 0.18832, "loss_cls_3": 1.62981, "loss_box_3": 4.71635, "loss_cns_3": 0.41462, "loss_yns_3": 0.20652, "loss_cls_4": 1.58897, "loss_box_4": 3.84992, "loss_cns_4": 0.35758, "loss_yns_4": 0.15703, "loss_cls_5": 1.69412, "loss_box_5": 2.78673, "loss_cns_5": 0.21571, "loss_yns_5": 0.09164, "loss_cls_dn_0": 0.69867, "loss_box_dn_0": 1.1508, "loss_cls_dn_1": 0.82898, "loss_box_dn_1": 2.41905, "loss_cls_dn_2": 0.80952, "loss_box_dn_2": 2.63267, "loss_cls_dn_3": 0.71442, "loss_box_dn_3": 2.83128, "loss_cls_dn_4": 0.72826, "loss_box_dn_4": 3.12366, "loss_cls_dn_5": 0.80773, "loss_box_dn_5": 3.35848, "loss_dense_depth": 1.66614, "loss": 54.17055, "grad_norm": 101.99692, "time": 1.53776} -{"mode": "train", "epoch": 1, "iter": 4, "lr": 0.0001, "memory": 49164, "data_time": 0.07793, "loss_cls_0": 1.36163, "loss_box_0": 2.54003, "loss_cns_0": 0.55612, "loss_yns_0": 0.18236, "loss_cls_1": 1.61676, "loss_box_1": 3.06912, "loss_cns_1": 0.45817, "loss_yns_1": 0.20341, "loss_cls_2": 1.71514, "loss_box_2": 3.68113, "loss_cns_2": 0.44109, "loss_yns_2": 0.18888, "loss_cls_3": 1.52443, "loss_box_3": 4.24297, "loss_cns_3": 0.44581, "loss_yns_3": 0.2175, "loss_cls_4": 1.47017, "loss_box_4": 4.85458, "loss_cns_4": 0.34806, "loss_yns_4": 0.19572, "loss_cls_5": 1.50124, "loss_box_5": 5.07846, "loss_cns_5": 0.44159, "loss_yns_5": 0.18784, "loss_cls_dn_0": 0.56738, "loss_box_dn_0": 1.16538, "loss_cls_dn_1": 0.72361, "loss_box_dn_1": 2.60043, "loss_cls_dn_2": 0.69849, "loss_box_dn_2": 2.67496, "loss_cls_dn_3": 0.61393, "loss_box_dn_3": 2.88829, "loss_cls_dn_4": 0.59831, "loss_box_dn_4": 3.10826, "loss_cls_dn_5": 0.6716, "loss_box_dn_5": 3.29535, "loss_dense_depth": 1.56391, "loss": 57.6921, "grad_norm": 132.48674, "time": 1.59113} -{"mode": "train", "epoch": 1, "iter": 5, "lr": 0.0001, "memory": 49164, "data_time": 0.08431, "loss_cls_0": 1.33114, "loss_box_0": 2.82355, "loss_cns_0": 0.49157, "loss_yns_0": 0.20038, "loss_cls_1": 1.56048, "loss_box_1": 3.95244, "loss_cns_1": 0.40147, "loss_yns_1": 0.19831, "loss_cls_2": 1.60932, "loss_box_2": 4.01748, "loss_cns_2": 0.37328, "loss_yns_2": 0.20085, "loss_cls_3": 1.44894, "loss_box_3": 4.15014, "loss_cns_3": 0.39607, "loss_yns_3": 0.19747, "loss_cls_4": 1.41701, "loss_box_4": 4.25385, "loss_cns_4": 0.376, "loss_yns_4": 0.19125, "loss_cls_5": 1.39427, "loss_box_5": 4.47206, "loss_cns_5": 0.41266, "loss_yns_5": 0.19648, "loss_cls_dn_0": 0.5269, "loss_box_dn_0": 1.26733, "loss_cls_dn_1": 0.65973, "loss_box_dn_1": 2.25312, "loss_cls_dn_2": 0.6554, "loss_box_dn_2": 2.36381, "loss_cls_dn_3": 0.56436, "loss_box_dn_3": 2.5044, "loss_cls_dn_4": 0.55129, "loss_box_dn_4": 2.64708, "loss_cls_dn_5": 0.58374, "loss_box_dn_5": 2.71501, "loss_dense_depth": 1.49944, "loss": 54.85807, "grad_norm": 111.33468, "time": 1.5872} -{"mode": "train", "epoch": 1, "iter": 6, "lr": 0.0001, "memory": 49164, "data_time": 0.10366, "loss_cls_0": 1.30345, "loss_box_0": 2.53449, "loss_cns_0": 0.56138, "loss_yns_0": 0.18177, "loss_cls_1": 1.49181, "loss_box_1": 3.88879, "loss_cns_1": 0.37003, "loss_yns_1": 0.19318, "loss_cls_2": 1.48527, "loss_box_2": 4.03429, "loss_cns_2": 0.35756, "loss_yns_2": 0.18971, "loss_cls_3": 1.35351, "loss_box_3": 4.07609, "loss_cns_3": 0.34793, "loss_yns_3": 0.19325, "loss_cls_4": 1.31769, "loss_box_4": 4.33018, "loss_cns_4": 0.30705, "loss_yns_4": 0.19858, "loss_cls_5": 1.33975, "loss_box_5": 4.45943, "loss_cns_5": 0.30502, "loss_yns_5": 0.19948, "loss_cls_dn_0": 0.51468, "loss_box_dn_0": 1.19751, "loss_cls_dn_1": 0.59562, "loss_box_dn_1": 2.42181, "loss_cls_dn_2": 0.58484, "loss_box_dn_2": 2.45258, "loss_cls_dn_3": 0.52141, "loss_box_dn_3": 2.51148, "loss_cls_dn_4": 0.49184, "loss_box_dn_4": 2.71241, "loss_cls_dn_5": 0.48676, "loss_box_dn_5": 2.7429, "loss_dense_depth": 1.46399, "loss": 53.71753, "grad_norm": 116.1794, "time": 1.61136} -{"mode": "train", "epoch": 1, "iter": 7, "lr": 0.0001, "memory": 49164, "data_time": 0.08354, "loss_cls_0": 1.26242, "loss_box_0": 2.31121, "loss_cns_0": 0.65601, "loss_yns_0": 0.17919, "loss_cls_1": 1.39685, "loss_box_1": 3.51791, "loss_cns_1": 0.44991, "loss_yns_1": 0.19927, "loss_cls_2": 1.42588, "loss_box_2": 3.59741, "loss_cns_2": 0.46093, "loss_yns_2": 0.18543, "loss_cls_3": 1.30868, "loss_box_3": 3.47633, "loss_cns_3": 0.47541, "loss_yns_3": 0.18925, "loss_cls_4": 1.28364, "loss_box_4": 3.74865, "loss_cns_4": 0.43634, "loss_yns_4": 0.19344, "loss_cls_5": 1.32556, "loss_box_5": 4.02016, "loss_cns_5": 0.40494, "loss_yns_5": 0.18409, "loss_cls_dn_0": 0.51793, "loss_box_dn_0": 1.10896, "loss_cls_dn_1": 0.53308, "loss_box_dn_1": 2.45544, "loss_cls_dn_2": 0.52636, "loss_box_dn_2": 2.40669, "loss_cls_dn_3": 0.47631, "loss_box_dn_3": 2.41607, "loss_cls_dn_4": 0.45092, "loss_box_dn_4": 2.61426, "loss_cls_dn_5": 0.4365, "loss_box_dn_5": 2.70311, "loss_dense_depth": 1.432, "loss": 50.76654, "grad_norm": 98.14408, "time": 1.58239} -{"mode": "train", "epoch": 1, "iter": 8, "lr": 0.0001, "memory": 49164, "data_time": 0.09665, "loss_cls_0": 1.22642, "loss_box_0": 2.25574, "loss_cns_0": 0.6386, "loss_yns_0": 0.18478, "loss_cls_1": 1.31678, "loss_box_1": 3.54987, "loss_cns_1": 0.48355, "loss_yns_1": 0.18459, "loss_cls_2": 1.38521, "loss_box_2": 3.70367, "loss_cns_2": 0.42355, "loss_yns_2": 0.18067, "loss_cls_3": 1.30047, "loss_box_3": 3.64252, "loss_cns_3": 0.42812, "loss_yns_3": 0.18329, "loss_cls_4": 1.28716, "loss_box_4": 3.70182, "loss_cns_4": 0.43635, "loss_yns_4": 0.18561, "loss_cls_5": 1.32823, "loss_box_5": 3.78321, "loss_cns_5": 0.46385, "loss_yns_5": 0.18941, "loss_cls_dn_0": 0.51687, "loss_box_dn_0": 1.04015, "loss_cls_dn_1": 0.53454, "loss_box_dn_1": 1.84152, "loss_cls_dn_2": 0.54439, "loss_box_dn_2": 1.7516, "loss_cls_dn_3": 0.49281, "loss_box_dn_3": 1.7501, "loss_cls_dn_4": 0.46266, "loss_box_dn_4": 1.87362, "loss_cls_dn_5": 0.43831, "loss_box_dn_5": 1.91299, "loss_dense_depth": 1.37049, "loss": 46.99349, "grad_norm": 75.57657, "time": 1.58518} -{"mode": "train", "epoch": 1, "iter": 9, "lr": 0.0001, "memory": 49164, "data_time": 0.09008, "loss_cls_0": 1.22251, "loss_box_0": 2.21347, "loss_cns_0": 0.61434, "loss_yns_0": 0.17865, "loss_cls_1": 1.27648, "loss_box_1": 3.3283, "loss_cns_1": 0.51771, "loss_yns_1": 0.18399, "loss_cls_2": 1.3696, "loss_box_2": 3.478, "loss_cns_2": 0.44124, "loss_yns_2": 0.17699, "loss_cls_3": 1.28147, "loss_box_3": 3.5966, "loss_cns_3": 0.4594, "loss_yns_3": 0.19697, "loss_cls_4": 1.26982, "loss_box_4": 3.49151, "loss_cns_4": 0.46825, "loss_yns_4": 0.19573, "loss_cls_5": 1.30533, "loss_box_5": 3.55673, "loss_cns_5": 0.48354, "loss_yns_5": 0.18335, "loss_cls_dn_0": 0.50554, "loss_box_dn_0": 1.02545, "loss_cls_dn_1": 0.48718, "loss_box_dn_1": 1.51672, "loss_cls_dn_2": 0.51859, "loss_box_dn_2": 1.5108, "loss_cls_dn_3": 0.47131, "loss_box_dn_3": 1.60619, "loss_cls_dn_4": 0.45036, "loss_box_dn_4": 1.55581, "loss_cls_dn_5": 0.43118, "loss_box_dn_5": 1.66109, "loss_dense_depth": 1.31335, "loss": 44.54353, "grad_norm": 68.24327, "time": 1.58709} -{"mode": "train", "epoch": 1, "iter": 10, "lr": 0.0001, "memory": 49164, "data_time": 0.07746, "loss_cls_0": 1.21046, "loss_box_0": 2.20847, "loss_cns_0": 0.61455, "loss_yns_0": 0.17346, "loss_cls_1": 1.27892, "loss_box_1": 3.15613, "loss_cns_1": 0.54554, "loss_yns_1": 0.18475, "loss_cls_2": 1.28288, "loss_box_2": 3.21337, "loss_cns_2": 0.5071, "loss_yns_2": 0.18107, "loss_cls_3": 1.2673, "loss_box_3": 3.37372, "loss_cns_3": 0.54602, "loss_yns_3": 0.19579, "loss_cls_4": 1.24793, "loss_box_4": 3.29238, "loss_cns_4": 0.54825, "loss_yns_4": 0.18347, "loss_cls_5": 1.26685, "loss_box_5": 3.42186, "loss_cns_5": 0.54378, "loss_yns_5": 0.17942, "loss_cls_dn_0": 0.48296, "loss_box_dn_0": 1.03228, "loss_cls_dn_1": 0.44299, "loss_box_dn_1": 1.59319, "loss_cls_dn_2": 0.47677, "loss_box_dn_2": 1.63352, "loss_cls_dn_3": 0.43488, "loss_box_dn_3": 1.74532, "loss_cls_dn_4": 0.43185, "loss_box_dn_4": 1.73216, "loss_cls_dn_5": 0.43208, "loss_box_dn_5": 1.87703, "loss_dense_depth": 1.31812, "loss": 44.2566, "grad_norm": 72.36771, "time": 1.56539} -{"mode": "train", "epoch": 1, "iter": 11, "lr": 0.0001, "memory": 49164, "data_time": 0.0762, "loss_cls_0": 1.22529, "loss_box_0": 2.24518, "loss_cns_0": 0.61829, "loss_yns_0": 0.1792, "loss_cls_1": 1.29466, "loss_box_1": 3.05893, "loss_cns_1": 0.53495, "loss_yns_1": 0.17848, "loss_cls_2": 1.27776, "loss_box_2": 3.00236, "loss_cns_2": 0.53267, "loss_yns_2": 0.18257, "loss_cls_3": 1.263, "loss_box_3": 3.09508, "loss_cns_3": 0.57049, "loss_yns_3": 0.18093, "loss_cls_4": 1.24554, "loss_box_4": 3.04435, "loss_cns_4": 0.56305, "loss_yns_4": 0.1829, "loss_cls_5": 1.25902, "loss_box_5": 3.22012, "loss_cns_5": 0.5397, "loss_yns_5": 0.18222, "loss_cls_dn_0": 0.46448, "loss_box_dn_0": 1.0524, "loss_cls_dn_1": 0.41214, "loss_box_dn_1": 1.80539, "loss_cls_dn_2": 0.442, "loss_box_dn_2": 1.82416, "loss_cls_dn_3": 0.41076, "loss_box_dn_3": 1.89655, "loss_cls_dn_4": 0.40297, "loss_box_dn_4": 1.92342, "loss_cls_dn_5": 0.42083, "loss_box_dn_5": 2.06931, "loss_dense_depth": 1.30026, "loss": 44.10141, "grad_norm": 68.24578, "time": 1.56971} -{"mode": "train", "epoch": 1, "iter": 12, "lr": 0.0001, "memory": 49164, "data_time": 0.08161, "loss_cls_0": 1.23069, "loss_box_0": 2.24328, "loss_cns_0": 0.6189, "loss_yns_0": 0.17424, "loss_cls_1": 1.26969, "loss_box_1": 2.95003, "loss_cns_1": 0.50177, "loss_yns_1": 0.17594, "loss_cls_2": 1.28721, "loss_box_2": 2.95213, "loss_cns_2": 0.50783, "loss_yns_2": 0.17656, "loss_cls_3": 1.2412, "loss_box_3": 2.92973, "loss_cns_3": 0.53124, "loss_yns_3": 0.1787, "loss_cls_4": 1.24798, "loss_box_4": 2.93042, "loss_cns_4": 0.55475, "loss_yns_4": 0.17701, "loss_cls_5": 1.26451, "loss_box_5": 3.10731, "loss_cns_5": 0.5327, "loss_yns_5": 0.18125, "loss_cls_dn_0": 0.45641, "loss_box_dn_0": 1.05019, "loss_cls_dn_1": 0.38665, "loss_box_dn_1": 2.01144, "loss_cls_dn_2": 0.41681, "loss_box_dn_2": 2.01013, "loss_cls_dn_3": 0.39648, "loss_box_dn_3": 2.03053, "loss_cls_dn_4": 0.3798, "loss_box_dn_4": 2.08467, "loss_cls_dn_5": 0.39704, "loss_box_dn_5": 2.17217, "loss_dense_depth": 1.30809, "loss": 44.06547, "grad_norm": 62.74979, "time": 1.56706} -{"mode": "train", "epoch": 1, "iter": 13, "lr": 0.0001, "memory": 49164, "data_time": 0.07686, "loss_cls_0": 1.20341, "loss_box_0": 2.25077, "loss_cns_0": 0.59975, "loss_yns_0": 0.17047, "loss_cls_1": 1.23388, "loss_box_1": 2.82912, "loss_cns_1": 0.51916, "loss_yns_1": 0.1804, "loss_cls_2": 1.2702, "loss_box_2": 2.8996, "loss_cns_2": 0.53724, "loss_yns_2": 0.17745, "loss_cls_3": 1.23246, "loss_box_3": 2.8655, "loss_cns_3": 0.5366, "loss_yns_3": 0.17727, "loss_cls_4": 1.26462, "loss_box_4": 2.90614, "loss_cns_4": 0.52663, "loss_yns_4": 0.18213, "loss_cls_5": 1.27591, "loss_box_5": 2.91214, "loss_cns_5": 0.53305, "loss_yns_5": 0.17958, "loss_cls_dn_0": 0.45457, "loss_box_dn_0": 1.03986, "loss_cls_dn_1": 0.42881, "loss_box_dn_1": 1.5131, "loss_cls_dn_2": 0.44799, "loss_box_dn_2": 1.54993, "loss_cls_dn_3": 0.44887, "loss_box_dn_3": 1.57173, "loss_cls_dn_4": 0.42706, "loss_box_dn_4": 1.72376, "loss_cls_dn_5": 0.4423, "loss_box_dn_5": 1.70691, "loss_dense_depth": 1.18772, "loss": 41.40609, "grad_norm": 60.31679, "time": 1.57237} -{"mode": "train", "epoch": 1, "iter": 14, "lr": 0.00011, "memory": 49164, "data_time": 0.07757, "loss_cls_0": 1.19999, "loss_box_0": 2.23269, "loss_cns_0": 0.59081, "loss_yns_0": 0.17165, "loss_cls_1": 1.26784, "loss_box_1": 2.72366, "loss_cns_1": 0.54865, "loss_yns_1": 0.17458, "loss_cls_2": 1.28193, "loss_box_2": 2.80409, "loss_cns_2": 0.55488, "loss_yns_2": 0.1788, "loss_cls_3": 1.26707, "loss_box_3": 2.83894, "loss_cns_3": 0.55625, "loss_yns_3": 0.17822, "loss_cls_4": 1.29521, "loss_box_4": 2.86976, "loss_cns_4": 0.5336, "loss_yns_4": 0.18131, "loss_cls_5": 1.29245, "loss_box_5": 2.8829, "loss_cns_5": 0.55711, "loss_yns_5": 0.17909, "loss_cls_dn_0": 0.46552, "loss_box_dn_0": 1.02319, "loss_cls_dn_1": 0.44968, "loss_box_dn_1": 1.28453, "loss_cls_dn_2": 0.46545, "loss_box_dn_2": 1.29777, "loss_cls_dn_3": 0.4661, "loss_box_dn_3": 1.38281, "loss_cls_dn_4": 0.45599, "loss_box_dn_4": 1.56617, "loss_cls_dn_5": 0.46069, "loss_box_dn_5": 1.52889, "loss_dense_depth": 1.28532, "loss": 40.49361, "grad_norm": 81.18964, "time": 1.57959} -{"mode": "train", "epoch": 1, "iter": 15, "lr": 0.00011, "memory": 49164, "data_time": 0.07305, "loss_cls_0": 1.19101, "loss_box_0": 2.29674, "loss_cns_0": 0.58913, "loss_yns_0": 0.17384, "loss_cls_1": 1.29854, "loss_box_1": 2.64328, "loss_cns_1": 0.57502, "loss_yns_1": 0.17356, "loss_cls_2": 1.2826, "loss_box_2": 2.63119, "loss_cns_2": 0.57365, "loss_yns_2": 0.17556, "loss_cls_3": 1.29508, "loss_box_3": 2.69451, "loss_cns_3": 0.57745, "loss_yns_3": 0.1789, "loss_cls_4": 1.28861, "loss_box_4": 2.71944, "loss_cns_4": 0.56723, "loss_yns_4": 0.17839, "loss_cls_5": 1.28728, "loss_box_5": 2.82377, "loss_cns_5": 0.59094, "loss_yns_5": 0.17892, "loss_cls_dn_0": 0.46732, "loss_box_dn_0": 1.04079, "loss_cls_dn_1": 0.44993, "loss_box_dn_1": 1.37813, "loss_cls_dn_2": 0.45942, "loss_box_dn_2": 1.35726, "loss_cls_dn_3": 0.45079, "loss_box_dn_3": 1.43968, "loss_cls_dn_4": 0.46104, "loss_box_dn_4": 1.53709, "loss_cls_dn_5": 0.45838, "loss_box_dn_5": 1.56635, "loss_dense_depth": 1.13774, "loss": 40.18859, "grad_norm": 69.23654, "time": 1.5583} -{"mode": "train", "epoch": 1, "iter": 16, "lr": 0.00011, "memory": 49164, "data_time": 0.07587, "loss_cls_0": 1.18129, "loss_box_0": 2.31569, "loss_cns_0": 0.58831, "loss_yns_0": 0.1755, "loss_cls_1": 1.28785, "loss_box_1": 2.70876, "loss_cns_1": 0.56938, "loss_yns_1": 0.17445, "loss_cls_2": 1.29474, "loss_box_2": 2.70804, "loss_cns_2": 0.56256, "loss_yns_2": 0.17578, "loss_cls_3": 1.30941, "loss_box_3": 2.70683, "loss_cns_3": 0.57023, "loss_yns_3": 0.17534, "loss_cls_4": 1.26051, "loss_box_4": 2.76813, "loss_cns_4": 0.56115, "loss_yns_4": 0.17681, "loss_cls_5": 1.27476, "loss_box_5": 2.84876, "loss_cns_5": 0.58187, "loss_yns_5": 0.17263, "loss_cls_dn_0": 0.48635, "loss_box_dn_0": 1.03024, "loss_cls_dn_1": 0.46657, "loss_box_dn_1": 1.36104, "loss_cls_dn_2": 0.47522, "loss_box_dn_2": 1.34068, "loss_cls_dn_3": 0.44334, "loss_box_dn_3": 1.41933, "loss_cls_dn_4": 0.46954, "loss_box_dn_4": 1.46954, "loss_cls_dn_5": 0.47358, "loss_box_dn_5": 1.55211, "loss_dense_depth": 1.26291, "loss": 40.39923, "grad_norm": 71.40475, "time": 1.57} -{"mode": "train", "epoch": 1, "iter": 17, "lr": 0.00011, "memory": 49164, "data_time": 0.08242, "loss_cls_0": 1.17415, "loss_box_0": 2.28839, "loss_cns_0": 0.59075, "loss_yns_0": 0.17347, "loss_cls_1": 1.27233, "loss_box_1": 2.93206, "loss_cns_1": 0.53971, "loss_yns_1": 0.17355, "loss_cls_2": 1.28818, "loss_box_2": 2.96682, "loss_cns_2": 0.53605, "loss_yns_2": 0.17493, "loss_cls_3": 1.29609, "loss_box_3": 2.92273, "loss_cns_3": 0.542, "loss_yns_3": 0.17503, "loss_cls_4": 1.25715, "loss_box_4": 2.93815, "loss_cns_4": 0.55178, "loss_yns_4": 0.17624, "loss_cls_5": 1.27099, "loss_box_5": 2.93308, "loss_cns_5": 0.54705, "loss_yns_5": 0.173, "loss_cls_dn_0": 0.48057, "loss_box_dn_0": 1.02364, "loss_cls_dn_1": 0.45586, "loss_box_dn_1": 1.49955, "loss_cls_dn_2": 0.45989, "loss_box_dn_2": 1.49834, "loss_cls_dn_3": 0.42139, "loss_box_dn_3": 1.56448, "loss_cls_dn_4": 0.44293, "loss_box_dn_4": 1.58618, "loss_cls_dn_5": 0.46101, "loss_box_dn_5": 1.70156, "loss_dense_depth": 1.13081, "loss": 41.6199, "grad_norm": 63.16544, "time": 1.58897} -{"mode": "train", "epoch": 1, "iter": 18, "lr": 0.00011, "memory": 49164, "data_time": 0.07644, "loss_cls_0": 1.1787, "loss_box_0": 2.22302, "loss_cns_0": 0.605, "loss_yns_0": 0.1717, "loss_cls_1": 1.24932, "loss_box_1": 2.89183, "loss_cns_1": 0.53512, "loss_yns_1": 0.17739, "loss_cls_2": 1.26051, "loss_box_2": 2.9365, "loss_cns_2": 0.53713, "loss_yns_2": 0.17236, "loss_cls_3": 1.26227, "loss_box_3": 2.89359, "loss_cns_3": 0.54144, "loss_yns_3": 0.17524, "loss_cls_4": 1.25008, "loss_box_4": 2.90769, "loss_cns_4": 0.56424, "loss_yns_4": 0.1714, "loss_cls_5": 1.25862, "loss_box_5": 2.95157, "loss_cns_5": 0.53608, "loss_yns_5": 0.1737, "loss_cls_dn_0": 0.48197, "loss_box_dn_0": 1.01582, "loss_cls_dn_1": 0.44699, "loss_box_dn_1": 1.54123, "loss_cls_dn_2": 0.45234, "loss_box_dn_2": 1.57155, "loss_cls_dn_3": 0.41822, "loss_box_dn_3": 1.64321, "loss_cls_dn_4": 0.42931, "loss_box_dn_4": 1.64592, "loss_cls_dn_5": 0.45397, "loss_box_dn_5": 1.80611, "loss_dense_depth": 1.09527, "loss": 41.62641, "grad_norm": 57.00684, "time": 1.574} -{"mode": "train", "epoch": 1, "iter": 19, "lr": 0.00011, "memory": 49164, "data_time": 0.07398, "loss_cls_0": 1.1754, "loss_box_0": 2.24182, "loss_cns_0": 0.5928, "loss_yns_0": 0.16987, "loss_cls_1": 1.24193, "loss_box_1": 2.82035, "loss_cns_1": 0.54178, "loss_yns_1": 0.18018, "loss_cls_2": 1.26128, "loss_box_2": 2.85646, "loss_cns_2": 0.54967, "loss_yns_2": 0.17321, "loss_cls_3": 1.26414, "loss_box_3": 2.86617, "loss_cns_3": 0.53902, "loss_yns_3": 0.1735, "loss_cls_4": 1.26365, "loss_box_4": 2.87415, "loss_cns_4": 0.56753, "loss_yns_4": 0.176, "loss_cls_5": 1.25748, "loss_box_5": 2.9768, "loss_cns_5": 0.53235, "loss_yns_5": 0.17417, "loss_cls_dn_0": 0.49143, "loss_box_dn_0": 1.03342, "loss_cls_dn_1": 0.45457, "loss_box_dn_1": 1.28784, "loss_cls_dn_2": 0.45698, "loss_box_dn_2": 1.36426, "loss_cls_dn_3": 0.43451, "loss_box_dn_3": 1.4638, "loss_cls_dn_4": 0.43522, "loss_box_dn_4": 1.45633, "loss_cls_dn_5": 0.45986, "loss_box_dn_5": 1.63814, "loss_dense_depth": 1.15695, "loss": 40.60299, "grad_norm": 77.80691, "time": 1.58026} -{"mode": "train", "epoch": 1, "iter": 20, "lr": 0.00011, "memory": 49164, "data_time": 0.07554, "loss_cls_0": 1.16508, "loss_box_0": 2.24309, "loss_cns_0": 0.59097, "loss_yns_0": 0.16895, "loss_cls_1": 1.24505, "loss_box_1": 2.79395, "loss_cns_1": 0.5623, "loss_yns_1": 0.17781, "loss_cls_2": 1.27752, "loss_box_2": 2.78806, "loss_cns_2": 0.57158, "loss_yns_2": 0.17525, "loss_cls_3": 1.26838, "loss_box_3": 2.83473, "loss_cns_3": 0.56019, "loss_yns_3": 0.17687, "loss_cls_4": 1.26886, "loss_box_4": 2.77221, "loss_cns_4": 0.57186, "loss_yns_4": 0.17969, "loss_cls_5": 1.26616, "loss_box_5": 2.84351, "loss_cns_5": 0.57124, "loss_yns_5": 0.18307, "loss_cls_dn_0": 0.4808, "loss_box_dn_0": 1.02646, "loss_cls_dn_1": 0.4148, "loss_box_dn_1": 1.42256, "loss_cls_dn_2": 0.41593, "loss_box_dn_2": 1.45853, "loss_cls_dn_3": 0.40711, "loss_box_dn_3": 1.51843, "loss_cls_dn_4": 0.40492, "loss_box_dn_4": 1.47967, "loss_cls_dn_5": 0.41683, "loss_box_dn_5": 1.56474, "loss_dense_depth": 1.05075, "loss": 40.3179, "grad_norm": 65.55939, "time": 1.55812} -{"mode": "train", "epoch": 1, "iter": 21, "lr": 0.00011, "memory": 49164, "data_time": 0.14265, "loss_cls_0": 1.15412, "loss_box_0": 2.19314, "loss_cns_0": 0.59612, "loss_yns_0": 0.16958, "loss_cls_1": 1.25431, "loss_box_1": 2.77427, "loss_cns_1": 0.58672, "loss_yns_1": 0.1747, "loss_cls_2": 1.28377, "loss_box_2": 2.75786, "loss_cns_2": 0.5912, "loss_yns_2": 0.17646, "loss_cls_3": 1.26055, "loss_box_3": 2.78509, "loss_cns_3": 0.59175, "loss_yns_3": 0.1735, "loss_cls_4": 1.25045, "loss_box_4": 2.78699, "loss_cns_4": 0.57365, "loss_yns_4": 0.17902, "loss_cls_5": 1.27275, "loss_box_5": 2.78867, "loss_cns_5": 0.59407, "loss_yns_5": 0.18242, "loss_cls_dn_0": 0.47398, "loss_box_dn_0": 1.03963, "loss_cls_dn_1": 0.4286, "loss_box_dn_1": 1.18198, "loss_cls_dn_2": 0.42519, "loss_box_dn_2": 1.20724, "loss_cls_dn_3": 0.43585, "loss_box_dn_3": 1.26105, "loss_cls_dn_4": 0.43337, "loss_box_dn_4": 1.28994, "loss_cls_dn_5": 0.43458, "loss_box_dn_5": 1.28811, "loss_dense_depth": 1.08161, "loss": 39.13231, "grad_norm": 50.65307, "time": 1.67347} -{"mode": "train", "epoch": 1, "iter": 22, "lr": 0.00011, "memory": 49164, "data_time": 0.10329, "loss_cls_0": 1.17895, "loss_box_0": 2.1402, "loss_cns_0": 0.60428, "loss_yns_0": 0.17171, "loss_cls_1": 1.26233, "loss_box_1": 2.79715, "loss_cns_1": 0.57124, "loss_yns_1": 0.16894, "loss_cls_2": 1.27295, "loss_box_2": 2.83576, "loss_cns_2": 0.55477, "loss_yns_2": 0.17707, "loss_cls_3": 1.25412, "loss_box_3": 2.8733, "loss_cns_3": 0.56372, "loss_yns_3": 0.1732, "loss_cls_4": 1.24435, "loss_box_4": 2.92765, "loss_cns_4": 0.55524, "loss_yns_4": 0.18007, "loss_cls_5": 1.25598, "loss_box_5": 2.95362, "loss_cns_5": 0.56373, "loss_yns_5": 0.17965, "loss_cls_dn_0": 0.45894, "loss_box_dn_0": 1.02652, "loss_cls_dn_1": 0.442, "loss_box_dn_1": 1.13372, "loss_cls_dn_2": 0.4384, "loss_box_dn_2": 1.19564, "loss_cls_dn_3": 0.45772, "loss_box_dn_3": 1.28695, "loss_cls_dn_4": 0.45511, "loss_box_dn_4": 1.40697, "loss_cls_dn_5": 0.44814, "loss_box_dn_5": 1.404, "loss_dense_depth": 1.04216, "loss": 39.65622, "grad_norm": 71.50336, "time": 1.6182} -{"mode": "train", "epoch": 1, "iter": 23, "lr": 0.00011, "memory": 49164, "data_time": 0.08862, "loss_cls_0": 1.16216, "loss_box_0": 2.11854, "loss_cns_0": 0.61156, "loss_yns_0": 0.17348, "loss_cls_1": 1.22641, "loss_box_1": 2.96367, "loss_cns_1": 0.52924, "loss_yns_1": 0.1699, "loss_cls_2": 1.25015, "loss_box_2": 2.90666, "loss_cns_2": 0.53996, "loss_yns_2": 0.17772, "loss_cls_3": 1.23476, "loss_box_3": 2.93808, "loss_cns_3": 0.54815, "loss_yns_3": 0.17576, "loss_cls_4": 1.22698, "loss_box_4": 2.9768, "loss_cns_4": 0.56003, "loss_yns_4": 0.17133, "loss_cls_5": 1.23852, "loss_box_5": 3.00482, "loss_cns_5": 0.56008, "loss_yns_5": 0.17342, "loss_cls_dn_0": 0.45508, "loss_box_dn_0": 0.9997, "loss_cls_dn_1": 0.44494, "loss_box_dn_1": 1.22944, "loss_cls_dn_2": 0.44623, "loss_box_dn_2": 1.25405, "loss_cls_dn_3": 0.46254, "loss_box_dn_3": 1.35866, "loss_cls_dn_4": 0.4496, "loss_box_dn_4": 1.51186, "loss_cls_dn_5": 0.45275, "loss_box_dn_5": 1.51462, "loss_dense_depth": 1.06118, "loss": 40.27881, "grad_norm": 73.85281, "time": 1.585} -{"mode": "train", "epoch": 1, "iter": 24, "lr": 0.00011, "memory": 49164, "data_time": 0.08336, "loss_cls_0": 1.17447, "loss_box_0": 2.11775, "loss_cns_0": 0.61744, "loss_yns_0": 0.1737, "loss_cls_1": 1.2315, "loss_box_1": 2.96144, "loss_cns_1": 0.54463, "loss_yns_1": 0.17325, "loss_cls_2": 1.2438, "loss_box_2": 2.87671, "loss_cns_2": 0.56078, "loss_yns_2": 0.18139, "loss_cls_3": 1.23883, "loss_box_3": 2.90171, "loss_cns_3": 0.55763, "loss_yns_3": 0.17853, "loss_cls_4": 1.23512, "loss_box_4": 2.93428, "loss_cns_4": 0.56078, "loss_yns_4": 0.17208, "loss_cls_5": 1.25055, "loss_box_5": 2.89534, "loss_cns_5": 0.56291, "loss_yns_5": 0.17592, "loss_cls_dn_0": 0.46697, "loss_box_dn_0": 1.00958, "loss_cls_dn_1": 0.43226, "loss_box_dn_1": 1.36062, "loss_cls_dn_2": 0.43799, "loss_box_dn_2": 1.35131, "loss_cls_dn_3": 0.44559, "loss_box_dn_3": 1.44145, "loss_cls_dn_4": 0.42598, "loss_box_dn_4": 1.58769, "loss_cls_dn_5": 0.4412, "loss_box_dn_5": 1.56405, "loss_dense_depth": 1.09043, "loss": 40.5757, "grad_norm": 65.5368, "time": 1.60097} -{"mode": "train", "epoch": 1, "iter": 25, "lr": 0.00011, "memory": 49164, "data_time": 0.08822, "loss_cls_0": 1.16447, "loss_box_0": 2.11237, "loss_cns_0": 0.61805, "loss_yns_0": 0.17454, "loss_cls_1": 1.22661, "loss_box_1": 2.7581, "loss_cns_1": 0.55169, "loss_yns_1": 0.17226, "loss_cls_2": 1.23253, "loss_box_2": 2.72577, "loss_cns_2": 0.56917, "loss_yns_2": 0.17574, "loss_cls_3": 1.23483, "loss_box_3": 2.73908, "loss_cns_3": 0.5679, "loss_yns_3": 0.17135, "loss_cls_4": 1.24151, "loss_box_4": 2.74213, "loss_cns_4": 0.57077, "loss_yns_4": 0.17797, "loss_cls_5": 1.24579, "loss_box_5": 2.74069, "loss_cns_5": 0.58185, "loss_yns_5": 0.17178, "loss_cls_dn_0": 0.46662, "loss_box_dn_0": 0.99229, "loss_cls_dn_1": 0.41886, "loss_box_dn_1": 1.33389, "loss_cls_dn_2": 0.42722, "loss_box_dn_2": 1.34751, "loss_cls_dn_3": 0.42619, "loss_box_dn_3": 1.4171, "loss_cls_dn_4": 0.40487, "loss_box_dn_4": 1.52081, "loss_cls_dn_5": 0.42857, "loss_box_dn_5": 1.50797, "loss_dense_depth": 1.08355, "loss": 39.44241, "grad_norm": 59.40277, "time": 1.56906} -{"mode": "train", "epoch": 1, "iter": 26, "lr": 0.00011, "memory": 49164, "data_time": 0.10057, "loss_cls_0": 1.14998, "loss_box_0": 2.0926, "loss_cns_0": 0.61904, "loss_yns_0": 0.17368, "loss_cls_1": 1.21537, "loss_box_1": 2.66944, "loss_cns_1": 0.56691, "loss_yns_1": 0.1708, "loss_cls_2": 1.23075, "loss_box_2": 2.69257, "loss_cns_2": 0.58377, "loss_yns_2": 0.17212, "loss_cls_3": 1.23924, "loss_box_3": 2.72341, "loss_cns_3": 0.58321, "loss_yns_3": 0.17293, "loss_cls_4": 1.24579, "loss_box_4": 2.73205, "loss_cns_4": 0.58467, "loss_yns_4": 0.17642, "loss_cls_5": 1.2382, "loss_box_5": 2.74491, "loss_cns_5": 0.59919, "loss_yns_5": 0.17457, "loss_cls_dn_0": 0.45091, "loss_box_dn_0": 0.99482, "loss_cls_dn_1": 0.43364, "loss_box_dn_1": 1.15353, "loss_cls_dn_2": 0.44011, "loss_box_dn_2": 1.20219, "loss_cls_dn_3": 0.43569, "loss_box_dn_3": 1.25634, "loss_cls_dn_4": 0.41783, "loss_box_dn_4": 1.32627, "loss_cls_dn_5": 0.4409, "loss_box_dn_5": 1.3278, "loss_dense_depth": 1.04574, "loss": 38.47739, "grad_norm": 82.94336, "time": 1.59224} -{"mode": "train", "epoch": 1, "iter": 27, "lr": 0.00011, "memory": 49164, "data_time": 0.07817, "loss_cls_0": 1.14104, "loss_box_0": 2.12738, "loss_cns_0": 0.61252, "loss_yns_0": 0.17392, "loss_cls_1": 1.21301, "loss_box_1": 2.72383, "loss_cns_1": 0.56844, "loss_yns_1": 0.17283, "loss_cls_2": 1.23008, "loss_box_2": 2.71516, "loss_cns_2": 0.5772, "loss_yns_2": 0.17515, "loss_cls_3": 1.22836, "loss_box_3": 2.71743, "loss_cns_3": 0.57965, "loss_yns_3": 0.17507, "loss_cls_4": 1.22739, "loss_box_4": 2.72911, "loss_cns_4": 0.58064, "loss_yns_4": 0.17743, "loss_cls_5": 1.23748, "loss_box_5": 2.74358, "loss_cns_5": 0.57485, "loss_yns_5": 0.18108, "loss_cls_dn_0": 0.44168, "loss_box_dn_0": 0.99274, "loss_cls_dn_1": 0.42028, "loss_box_dn_1": 1.17759, "loss_cls_dn_2": 0.43105, "loss_box_dn_2": 1.17822, "loss_cls_dn_3": 0.42548, "loss_box_dn_3": 1.19006, "loss_cls_dn_4": 0.4165, "loss_box_dn_4": 1.21217, "loss_cls_dn_5": 0.42526, "loss_box_dn_5": 1.25466, "loss_dense_depth": 1.02481, "loss": 38.17316, "grad_norm": 62.14588, "time": 1.56789} -{"mode": "train", "epoch": 1, "iter": 28, "lr": 0.00011, "memory": 49164, "data_time": 0.08978, "loss_cls_0": 1.12982, "loss_box_0": 2.13493, "loss_cns_0": 0.61312, "loss_yns_0": 0.17145, "loss_cls_1": 1.20514, "loss_box_1": 2.53809, "loss_cns_1": 0.58035, "loss_yns_1": 0.17237, "loss_cls_2": 1.21596, "loss_box_2": 2.51287, "loss_cns_2": 0.58587, "loss_yns_2": 0.17919, "loss_cls_3": 1.21304, "loss_box_3": 2.53582, "loss_cns_3": 0.58483, "loss_yns_3": 0.17698, "loss_cls_4": 1.20669, "loss_box_4": 2.50695, "loss_cns_4": 0.59274, "loss_yns_4": 0.17418, "loss_cls_5": 1.24959, "loss_box_5": 2.54785, "loss_cns_5": 0.59319, "loss_yns_5": 0.17538, "loss_cls_dn_0": 0.43123, "loss_box_dn_0": 0.9885, "loss_cls_dn_1": 0.40663, "loss_box_dn_1": 1.17517, "loss_cls_dn_2": 0.42018, "loss_box_dn_2": 1.16436, "loss_cls_dn_3": 0.40966, "loss_box_dn_3": 1.19354, "loss_cls_dn_4": 0.40765, "loss_box_dn_4": 1.20337, "loss_cls_dn_5": 0.40295, "loss_box_dn_5": 1.31569, "loss_dense_depth": 1.05191, "loss": 37.16722, "grad_norm": 49.59421, "time": 1.58869} -{"mode": "train", "epoch": 1, "iter": 29, "lr": 0.00011, "memory": 49164, "data_time": 0.08189, "loss_cls_0": 1.10232, "loss_box_0": 2.10123, "loss_cns_0": 0.61437, "loss_yns_0": 0.17195, "loss_cls_1": 1.19596, "loss_box_1": 2.43447, "loss_cns_1": 0.59069, "loss_yns_1": 0.17251, "loss_cls_2": 1.20227, "loss_box_2": 2.41458, "loss_cns_2": 0.6037, "loss_yns_2": 0.1764, "loss_cls_3": 1.20762, "loss_box_3": 2.44331, "loss_cns_3": 0.60321, "loss_yns_3": 0.17722, "loss_cls_4": 1.19596, "loss_box_4": 2.46566, "loss_cns_4": 0.60404, "loss_yns_4": 0.17648, "loss_cls_5": 1.22407, "loss_box_5": 2.49675, "loss_cns_5": 0.60867, "loss_yns_5": 0.16932, "loss_cls_dn_0": 0.43309, "loss_box_dn_0": 0.98721, "loss_cls_dn_1": 0.40833, "loss_box_dn_1": 1.11287, "loss_cls_dn_2": 0.42709, "loss_box_dn_2": 1.1209, "loss_cls_dn_3": 0.41032, "loss_box_dn_3": 1.19478, "loss_cls_dn_4": 0.40959, "loss_box_dn_4": 1.23315, "loss_cls_dn_5": 0.40892, "loss_box_dn_5": 1.37728, "loss_dense_depth": 0.99721, "loss": 36.67351, "grad_norm": 67.55933, "time": 1.58056} -{"mode": "train", "epoch": 1, "iter": 30, "lr": 0.00011, "memory": 49164, "data_time": 0.08783, "loss_cls_0": 1.07735, "loss_box_0": 2.04748, "loss_cns_0": 0.61622, "loss_yns_0": 0.17275, "loss_cls_1": 1.16362, "loss_box_1": 2.43977, "loss_cns_1": 0.60087, "loss_yns_1": 0.17255, "loss_cls_2": 1.17835, "loss_box_2": 2.42394, "loss_cns_2": 0.61147, "loss_yns_2": 0.17344, "loss_cls_3": 1.19045, "loss_box_3": 2.4213, "loss_cns_3": 0.61117, "loss_yns_3": 0.17505, "loss_cls_4": 1.17612, "loss_box_4": 2.48952, "loss_cns_4": 0.60585, "loss_yns_4": 0.17527, "loss_cls_5": 1.18692, "loss_box_5": 2.51571, "loss_cns_5": 0.60537, "loss_yns_5": 0.17087, "loss_cls_dn_0": 0.43606, "loss_box_dn_0": 0.98334, "loss_cls_dn_1": 0.39705, "loss_box_dn_1": 1.22335, "loss_cls_dn_2": 0.41871, "loss_box_dn_2": 1.23665, "loss_cls_dn_3": 0.40071, "loss_box_dn_3": 1.30643, "loss_cls_dn_4": 0.40233, "loss_box_dn_4": 1.35708, "loss_cls_dn_5": 0.4053, "loss_box_dn_5": 1.48042, "loss_dense_depth": 1.03953, "loss": 37.08841, "grad_norm": 79.30573, "time": 1.5809} -{"mode": "train", "epoch": 1, "iter": 31, "lr": 0.00011, "memory": 49164, "data_time": 0.07687, "loss_cls_0": 1.06615, "loss_box_0": 2.05063, "loss_cns_0": 0.61475, "loss_yns_0": 0.17405, "loss_cls_1": 1.15112, "loss_box_1": 2.4285, "loss_cns_1": 0.60919, "loss_yns_1": 0.1748, "loss_cls_2": 1.16815, "loss_box_2": 2.39895, "loss_cns_2": 0.61638, "loss_yns_2": 0.1733, "loss_cls_3": 1.17549, "loss_box_3": 2.41333, "loss_cns_3": 0.61237, "loss_yns_3": 0.17317, "loss_cls_4": 1.16358, "loss_box_4": 2.44226, "loss_cns_4": 0.60909, "loss_yns_4": 0.17224, "loss_cls_5": 1.17464, "loss_box_5": 2.48374, "loss_cns_5": 0.60332, "loss_yns_5": 0.17099, "loss_cls_dn_0": 0.43872, "loss_box_dn_0": 0.96686, "loss_cls_dn_1": 0.39318, "loss_box_dn_1": 1.27256, "loss_cls_dn_2": 0.41376, "loss_box_dn_2": 1.26871, "loss_cls_dn_3": 0.39652, "loss_box_dn_3": 1.31133, "loss_cls_dn_4": 0.39518, "loss_box_dn_4": 1.32138, "loss_cls_dn_5": 0.40924, "loss_box_dn_5": 1.4157, "loss_dense_depth": 0.99962, "loss": 36.82296, "grad_norm": 63.72398, "time": 1.59177} -{"mode": "train", "epoch": 1, "iter": 32, "lr": 0.00011, "memory": 49164, "data_time": 0.07376, "loss_cls_0": 1.05, "loss_box_0": 2.05491, "loss_cns_0": 0.61297, "loss_yns_0": 0.17242, "loss_cls_1": 1.13018, "loss_box_1": 2.36053, "loss_cns_1": 0.60691, "loss_yns_1": 0.17713, "loss_cls_2": 1.15381, "loss_box_2": 2.32508, "loss_cns_2": 0.61461, "loss_yns_2": 0.17675, "loss_cls_3": 1.16032, "loss_box_3": 2.35454, "loss_cns_3": 0.61484, "loss_yns_3": 0.17154, "loss_cls_4": 1.15077, "loss_box_4": 2.4217, "loss_cns_4": 0.60936, "loss_yns_4": 0.16993, "loss_cls_5": 1.16187, "loss_box_5": 2.45893, "loss_cns_5": 0.60607, "loss_yns_5": 0.17329, "loss_cls_dn_0": 0.41682, "loss_box_dn_0": 0.9642, "loss_cls_dn_1": 0.38175, "loss_box_dn_1": 1.14204, "loss_cls_dn_2": 0.39851, "loss_box_dn_2": 1.12811, "loss_cls_dn_3": 0.38451, "loss_box_dn_3": 1.15472, "loss_cls_dn_4": 0.38115, "loss_box_dn_4": 1.16118, "loss_cls_dn_5": 0.40714, "loss_box_dn_5": 1.22704, "loss_dense_depth": 0.98943, "loss": 35.62506, "grad_norm": 59.22495, "time": 1.57256} -{"mode": "train", "epoch": 1, "iter": 33, "lr": 0.00011, "memory": 49164, "data_time": 0.07233, "loss_cls_0": 1.08443, "loss_box_0": 2.03, "loss_cns_0": 0.61805, "loss_yns_0": 0.17244, "loss_cls_1": 1.14713, "loss_box_1": 2.34639, "loss_cns_1": 0.60736, "loss_yns_1": 0.18065, "loss_cls_2": 1.15789, "loss_box_2": 2.31687, "loss_cns_2": 0.61711, "loss_yns_2": 0.1751, "loss_cls_3": 1.16799, "loss_box_3": 2.32812, "loss_cns_3": 0.62156, "loss_yns_3": 0.17287, "loss_cls_4": 1.15107, "loss_box_4": 2.37658, "loss_cns_4": 0.62151, "loss_yns_4": 0.17017, "loss_cls_5": 1.15902, "loss_box_5": 2.39546, "loss_cns_5": 0.62106, "loss_yns_5": 0.17499, "loss_cls_dn_0": 0.41242, "loss_box_dn_0": 0.96747, "loss_cls_dn_1": 0.34949, "loss_box_dn_1": 1.13548, "loss_cls_dn_2": 0.36764, "loss_box_dn_2": 1.13506, "loss_cls_dn_3": 0.35499, "loss_box_dn_3": 1.15529, "loss_cls_dn_4": 0.35714, "loss_box_dn_4": 1.15724, "loss_cls_dn_5": 0.37598, "loss_box_dn_5": 1.19143, "loss_dense_depth": 1.02366, "loss": 35.39712, "grad_norm": 49.63317, "time": 1.56954} -{"mode": "train", "epoch": 1, "iter": 34, "lr": 0.00011, "memory": 49164, "data_time": 0.07686, "loss_cls_0": 1.06919, "loss_box_0": 1.99731, "loss_cns_0": 0.62044, "loss_yns_0": 0.1715, "loss_cls_1": 1.15294, "loss_box_1": 2.39311, "loss_cns_1": 0.60394, "loss_yns_1": 0.17291, "loss_cls_2": 1.15711, "loss_box_2": 2.4018, "loss_cns_2": 0.60998, "loss_yns_2": 0.17153, "loss_cls_3": 1.15979, "loss_box_3": 2.42934, "loss_cns_3": 0.61758, "loss_yns_3": 0.17106, "loss_cls_4": 1.14912, "loss_box_4": 2.46178, "loss_cns_4": 0.61547, "loss_yns_4": 0.16882, "loss_cls_5": 1.15937, "loss_box_5": 2.4716, "loss_cns_5": 0.61808, "loss_yns_5": 0.17291, "loss_cls_dn_0": 0.42893, "loss_box_dn_0": 0.96204, "loss_cls_dn_1": 0.33168, "loss_box_dn_1": 1.15342, "loss_cls_dn_2": 0.35205, "loss_box_dn_2": 1.17999, "loss_cls_dn_3": 0.35267, "loss_box_dn_3": 1.20944, "loss_cls_dn_4": 0.35527, "loss_box_dn_4": 1.23678, "loss_cls_dn_5": 0.36385, "loss_box_dn_5": 1.23991, "loss_dense_depth": 1.02355, "loss": 35.90622, "grad_norm": 68.18538, "time": 1.56081} -{"mode": "train", "epoch": 1, "iter": 35, "lr": 0.00011, "memory": 49164, "data_time": 0.08099, "loss_cls_0": 1.05425, "loss_box_0": 1.95615, "loss_cns_0": 0.62324, "loss_yns_0": 0.17138, "loss_cls_1": 1.1672, "loss_box_1": 2.33447, "loss_cns_1": 0.61194, "loss_yns_1": 0.17169, "loss_cls_2": 1.16597, "loss_box_2": 2.3664, "loss_cns_2": 0.61169, "loss_yns_2": 0.17307, "loss_cls_3": 1.15175, "loss_box_3": 2.39592, "loss_cns_3": 0.61569, "loss_yns_3": 0.17081, "loss_cls_4": 1.15789, "loss_box_4": 2.47227, "loss_cns_4": 0.61605, "loss_yns_4": 0.1709, "loss_cls_5": 1.16434, "loss_box_5": 2.42825, "loss_cns_5": 0.61779, "loss_yns_5": 0.17369, "loss_cls_dn_0": 0.43415, "loss_box_dn_0": 0.94495, "loss_cls_dn_1": 0.32768, "loss_box_dn_1": 1.16488, "loss_cls_dn_2": 0.3475, "loss_box_dn_2": 1.21148, "loss_cls_dn_3": 0.3624, "loss_box_dn_3": 1.24803, "loss_cls_dn_4": 0.363, "loss_box_dn_4": 1.32812, "loss_cls_dn_5": 0.36036, "loss_box_dn_5": 1.31454, "loss_dense_depth": 0.98419, "loss": 35.93407, "grad_norm": 87.59892, "time": 1.58137} -{"mode": "train", "epoch": 1, "iter": 36, "lr": 0.00011, "memory": 49164, "data_time": 0.08324, "loss_cls_0": 1.04176, "loss_box_0": 1.95299, "loss_cns_0": 0.62002, "loss_yns_0": 0.17141, "loss_cls_1": 1.14917, "loss_box_1": 2.40136, "loss_cns_1": 0.60444, "loss_yns_1": 0.17352, "loss_cls_2": 1.15362, "loss_box_2": 2.40226, "loss_cns_2": 0.61161, "loss_yns_2": 0.17231, "loss_cls_3": 1.138, "loss_box_3": 2.40278, "loss_cns_3": 0.61214, "loss_yns_3": 0.16916, "loss_cls_4": 1.12999, "loss_box_4": 2.46833, "loss_cns_4": 0.61962, "loss_yns_4": 0.1756, "loss_cls_5": 1.1546, "loss_box_5": 2.47341, "loss_cns_5": 0.61216, "loss_yns_5": 0.17571, "loss_cls_dn_0": 0.42821, "loss_box_dn_0": 0.93568, "loss_cls_dn_1": 0.35525, "loss_box_dn_1": 1.08397, "loss_cls_dn_2": 0.37132, "loss_box_dn_2": 1.13775, "loss_cls_dn_3": 0.39276, "loss_box_dn_3": 1.17027, "loss_cls_dn_4": 0.39538, "loss_box_dn_4": 1.31107, "loss_cls_dn_5": 0.3842, "loss_box_dn_5": 1.31541, "loss_dense_depth": 0.98726, "loss": 35.85451, "grad_norm": 64.19453, "time": 1.58996} -{"mode": "train", "epoch": 1, "iter": 37, "lr": 0.00011, "memory": 49164, "data_time": 0.07759, "loss_cls_0": 1.05014, "loss_box_0": 1.95779, "loss_cns_0": 0.62068, "loss_yns_0": 0.1719, "loss_cls_1": 1.1208, "loss_box_1": 2.51262, "loss_cns_1": 0.58644, "loss_yns_1": 0.17508, "loss_cls_2": 1.15231, "loss_box_2": 2.50575, "loss_cns_2": 0.6004, "loss_yns_2": 0.17485, "loss_cls_3": 1.14355, "loss_box_3": 2.5394, "loss_cns_3": 0.59848, "loss_yns_3": 0.1698, "loss_cls_4": 1.13835, "loss_box_4": 2.56073, "loss_cns_4": 0.61327, "loss_yns_4": 0.17961, "loss_cls_5": 1.17235, "loss_box_5": 2.58783, "loss_cns_5": 0.6081, "loss_yns_5": 0.17686, "loss_cls_dn_0": 0.41577, "loss_box_dn_0": 0.93458, "loss_cls_dn_1": 0.35429, "loss_box_dn_1": 1.2065, "loss_cls_dn_2": 0.36427, "loss_box_dn_2": 1.22787, "loss_cls_dn_3": 0.37752, "loss_box_dn_3": 1.25289, "loss_cls_dn_4": 0.38094, "loss_box_dn_4": 1.35917, "loss_cls_dn_5": 0.37216, "loss_box_dn_5": 1.37441, "loss_dense_depth": 0.9627, "loss": 36.70023, "grad_norm": 76.11732, "time": 1.56352} -{"mode": "train", "epoch": 1, "iter": 38, "lr": 0.00011, "memory": 49164, "data_time": 0.07394, "loss_cls_0": 1.05697, "loss_box_0": 1.97899, "loss_cns_0": 0.62227, "loss_yns_0": 0.17077, "loss_cls_1": 1.09304, "loss_box_1": 2.60411, "loss_cns_1": 0.57697, "loss_yns_1": 0.17197, "loss_cls_2": 1.13872, "loss_box_2": 2.59713, "loss_cns_2": 0.59263, "loss_yns_2": 0.17385, "loss_cls_3": 1.15225, "loss_box_3": 2.62523, "loss_cns_3": 0.59631, "loss_yns_3": 0.17002, "loss_cls_4": 1.16379, "loss_box_4": 2.62132, "loss_cns_4": 0.6076, "loss_yns_4": 0.1732, "loss_cls_5": 1.16431, "loss_box_5": 2.65227, "loss_cns_5": 0.60931, "loss_yns_5": 0.17444, "loss_cls_dn_0": 0.39792, "loss_box_dn_0": 0.9366, "loss_cls_dn_1": 0.3611, "loss_box_dn_1": 1.18817, "loss_cls_dn_2": 0.37171, "loss_box_dn_2": 1.17002, "loss_cls_dn_3": 0.37625, "loss_box_dn_3": 1.17917, "loss_cls_dn_4": 0.37256, "loss_box_dn_4": 1.24216, "loss_cls_dn_5": 0.37913, "loss_box_dn_5": 1.25787, "loss_dense_depth": 1.00459, "loss": 36.72469, "grad_norm": 83.93192, "time": 1.56218} -{"mode": "train", "epoch": 1, "iter": 39, "lr": 0.00012, "memory": 49164, "data_time": 0.08408, "loss_cls_0": 1.08641, "loss_box_0": 1.96272, "loss_cns_0": 0.62207, "loss_yns_0": 0.16864, "loss_cls_1": 1.10603, "loss_box_1": 2.45784, "loss_cns_1": 0.5935, "loss_yns_1": 0.17296, "loss_cls_2": 1.13748, "loss_box_2": 2.4679, "loss_cns_2": 0.60373, "loss_yns_2": 0.17148, "loss_cls_3": 1.1506, "loss_box_3": 2.47668, "loss_cns_3": 0.60955, "loss_yns_3": 0.16853, "loss_cls_4": 1.16714, "loss_box_4": 2.45967, "loss_cns_4": 0.61179, "loss_yns_4": 0.17085, "loss_cls_5": 1.14434, "loss_box_5": 2.49189, "loss_cns_5": 0.61512, "loss_yns_5": 0.17126, "loss_cls_dn_0": 0.39408, "loss_box_dn_0": 0.92172, "loss_cls_dn_1": 0.36948, "loss_box_dn_1": 1.08247, "loss_cls_dn_2": 0.38327, "loss_box_dn_2": 1.05323, "loss_cls_dn_3": 0.38246, "loss_box_dn_3": 1.04412, "loss_cls_dn_4": 0.3662, "loss_box_dn_4": 1.07405, "loss_cls_dn_5": 0.38532, "loss_box_dn_5": 1.09144, "loss_dense_depth": 0.9523, "loss": 35.28834, "grad_norm": 68.28097, "time": 1.63742} -{"mode": "train", "epoch": 1, "iter": 40, "lr": 0.00012, "memory": 49164, "data_time": 0.08657, "loss_cls_0": 1.05969, "loss_box_0": 1.96846, "loss_cns_0": 0.62013, "loss_yns_0": 0.16547, "loss_cls_1": 1.09942, "loss_box_1": 2.3271, "loss_cns_1": 0.60246, "loss_yns_1": 0.17178, "loss_cls_2": 1.13807, "loss_box_2": 2.32457, "loss_cns_2": 0.61556, "loss_yns_2": 0.17026, "loss_cls_3": 1.13925, "loss_box_3": 2.29871, "loss_cns_3": 0.61975, "loss_yns_3": 0.16867, "loss_cls_4": 1.1252, "loss_box_4": 2.29715, "loss_cns_4": 0.62669, "loss_yns_4": 0.17269, "loss_cls_5": 1.13836, "loss_box_5": 2.29132, "loss_cns_5": 0.63368, "loss_yns_5": 0.16693, "loss_cls_dn_0": 0.40846, "loss_box_dn_0": 0.92197, "loss_cls_dn_1": 0.37465, "loss_box_dn_1": 1.04025, "loss_cls_dn_2": 0.38845, "loss_box_dn_2": 1.00889, "loss_cls_dn_3": 0.38288, "loss_box_dn_3": 0.98703, "loss_cls_dn_4": 0.36292, "loss_box_dn_4": 0.99329, "loss_cls_dn_5": 0.38657, "loss_box_dn_5": 0.99463, "loss_dense_depth": 1.06226, "loss": 34.25364, "grad_norm": 57.80703, "time": 1.58507} -{"mode": "train", "epoch": 1, "iter": 41, "lr": 0.00012, "memory": 49164, "data_time": 0.10068, "loss_cls_0": 1.01877, "loss_box_0": 1.94843, "loss_cns_0": 0.62201, "loss_yns_0": 0.16305, "loss_cls_1": 1.08346, "loss_box_1": 2.2593, "loss_cns_1": 0.60955, "loss_yns_1": 0.16797, "loss_cls_2": 1.12476, "loss_box_2": 2.23955, "loss_cns_2": 0.62953, "loss_yns_2": 0.16794, "loss_cls_3": 1.13895, "loss_box_3": 2.22968, "loss_cns_3": 0.63434, "loss_yns_3": 0.16541, "loss_cls_4": 1.12936, "loss_box_4": 2.23513, "loss_cns_4": 0.65214, "loss_yns_4": 0.16781, "loss_cls_5": 1.15445, "loss_box_5": 2.20854, "loss_cns_5": 0.64906, "loss_yns_5": 0.16594, "loss_cls_dn_0": 0.42256, "loss_box_dn_0": 0.92553, "loss_cls_dn_1": 0.35025, "loss_box_dn_1": 1.03546, "loss_cls_dn_2": 0.36348, "loss_box_dn_2": 0.98351, "loss_cls_dn_3": 0.35772, "loss_box_dn_3": 0.97356, "loss_cls_dn_4": 0.35401, "loss_box_dn_4": 0.9751, "loss_cls_dn_5": 0.36992, "loss_box_dn_5": 0.97825, "loss_dense_depth": 1.09728, "loss": 33.75176, "grad_norm": 58.6673, "time": 1.63685} -{"mode": "train", "epoch": 1, "iter": 42, "lr": 0.00012, "memory": 49164, "data_time": 0.1265, "loss_cls_0": 1.01223, "loss_box_0": 1.9123, "loss_cns_0": 0.62574, "loss_yns_0": 0.16145, "loss_cls_1": 1.07968, "loss_box_1": 2.20868, "loss_cns_1": 0.61453, "loss_yns_1": 0.16625, "loss_cls_2": 1.12289, "loss_box_2": 2.1569, "loss_cns_2": 0.6337, "loss_yns_2": 0.17072, "loss_cls_3": 1.12827, "loss_box_3": 2.18464, "loss_cns_3": 0.63882, "loss_yns_3": 0.16916, "loss_cls_4": 1.11929, "loss_box_4": 2.19855, "loss_cns_4": 0.65517, "loss_yns_4": 0.16431, "loss_cls_5": 1.13432, "loss_box_5": 2.21839, "loss_cns_5": 0.63683, "loss_yns_5": 0.16738, "loss_cls_dn_0": 0.41485, "loss_box_dn_0": 0.92935, "loss_cls_dn_1": 0.34726, "loss_box_dn_1": 1.04195, "loss_cls_dn_2": 0.35416, "loss_box_dn_2": 0.99908, "loss_cls_dn_3": 0.354, "loss_box_dn_3": 1.02687, "loss_cls_dn_4": 0.36497, "loss_box_dn_4": 1.04232, "loss_cls_dn_5": 0.37688, "loss_box_dn_5": 1.09442, "loss_dense_depth": 0.96865, "loss": 33.59497, "grad_norm": 61.07832, "time": 1.62991} -{"mode": "train", "epoch": 1, "iter": 43, "lr": 0.00012, "memory": 49164, "data_time": 0.08429, "loss_cls_0": 1.05984, "loss_box_0": 1.94372, "loss_cns_0": 0.62338, "loss_yns_0": 0.16181, "loss_cls_1": 1.1243, "loss_box_1": 2.271, "loss_cns_1": 0.6088, "loss_yns_1": 0.16324, "loss_cls_2": 1.16409, "loss_box_2": 2.19405, "loss_cns_2": 0.6253, "loss_yns_2": 0.16657, "loss_cls_3": 1.16344, "loss_box_3": 2.23139, "loss_cns_3": 0.62845, "loss_yns_3": 0.16526, "loss_cls_4": 1.11412, "loss_box_4": 2.22943, "loss_cns_4": 0.64481, "loss_yns_4": 0.16455, "loss_cls_5": 1.13531, "loss_box_5": 2.2493, "loss_cns_5": 0.6318, "loss_yns_5": 0.16483, "loss_cls_dn_0": 0.41391, "loss_box_dn_0": 0.92413, "loss_cls_dn_1": 0.35865, "loss_box_dn_1": 1.0662, "loss_cls_dn_2": 0.36003, "loss_box_dn_2": 1.04886, "loss_cls_dn_3": 0.3599, "loss_box_dn_3": 1.10565, "loss_cls_dn_4": 0.38121, "loss_box_dn_4": 1.12869, "loss_cls_dn_5": 0.38375, "loss_box_dn_5": 1.20549, "loss_dense_depth": 1.07469, "loss": 34.43993, "grad_norm": 62.08274, "time": 1.61569} -{"mode": "train", "epoch": 1, "iter": 44, "lr": 0.00012, "memory": 49164, "data_time": 0.07568, "loss_cls_0": 1.07383, "loss_box_0": 2.00426, "loss_cns_0": 0.61915, "loss_yns_0": 0.16285, "loss_cls_1": 1.12912, "loss_box_1": 2.27449, "loss_cns_1": 0.60498, "loss_yns_1": 0.16138, "loss_cls_2": 1.16791, "loss_box_2": 2.21015, "loss_cns_2": 0.61538, "loss_yns_2": 0.16556, "loss_cls_3": 1.16647, "loss_box_3": 2.24107, "loss_cns_3": 0.61532, "loss_yns_3": 0.1634, "loss_cls_4": 1.12547, "loss_box_4": 2.22578, "loss_cns_4": 0.62066, "loss_yns_4": 0.16264, "loss_cls_5": 1.16888, "loss_box_5": 2.23758, "loss_cns_5": 0.61898, "loss_yns_5": 0.16119, "loss_cls_dn_0": 0.4242, "loss_box_dn_0": 0.91951, "loss_cls_dn_1": 0.36867, "loss_box_dn_1": 1.13741, "loss_cls_dn_2": 0.37372, "loss_box_dn_2": 1.12646, "loss_cls_dn_3": 0.37402, "loss_box_dn_3": 1.17943, "loss_cls_dn_4": 0.39007, "loss_box_dn_4": 1.1966, "loss_cls_dn_5": 0.39178, "loss_box_dn_5": 1.26859, "loss_dense_depth": 1.02888, "loss": 34.87583, "grad_norm": 56.71872, "time": 1.60313} -{"mode": "train", "epoch": 1, "iter": 45, "lr": 0.00012, "memory": 49164, "data_time": 0.08925, "loss_cls_0": 1.05687, "loss_box_0": 2.01434, "loss_cns_0": 0.61204, "loss_yns_0": 0.16388, "loss_cls_1": 1.11509, "loss_box_1": 2.3159, "loss_cns_1": 0.60123, "loss_yns_1": 0.16409, "loss_cls_2": 1.1514, "loss_box_2": 2.25224, "loss_cns_2": 0.61624, "loss_yns_2": 0.16742, "loss_cls_3": 1.14195, "loss_box_3": 2.26742, "loss_cns_3": 0.62218, "loss_yns_3": 0.16458, "loss_cls_4": 1.15724, "loss_box_4": 2.27505, "loss_cns_4": 0.62034, "loss_yns_4": 0.16499, "loss_cls_5": 1.17939, "loss_box_5": 2.27017, "loss_cns_5": 0.62135, "loss_yns_5": 0.16395, "loss_cls_dn_0": 0.42565, "loss_box_dn_0": 0.92492, "loss_cls_dn_1": 0.37838, "loss_box_dn_1": 1.13934, "loss_cls_dn_2": 0.38524, "loss_box_dn_2": 1.12521, "loss_cls_dn_3": 0.394, "loss_box_dn_3": 1.16478, "loss_cls_dn_4": 0.38881, "loss_box_dn_4": 1.18181, "loss_cls_dn_5": 0.40137, "loss_box_dn_5": 1.24221, "loss_dense_depth": 1.05192, "loss": 35.08301, "grad_norm": 56.54783, "time": 1.61247} -{"mode": "train", "epoch": 1, "iter": 46, "lr": 0.00012, "memory": 49164, "data_time": 0.10122, "loss_cls_0": 1.04332, "loss_box_0": 1.99252, "loss_cns_0": 0.61318, "loss_yns_0": 0.16182, "loss_cls_1": 1.08813, "loss_box_1": 2.34229, "loss_cns_1": 0.59815, "loss_yns_1": 0.16644, "loss_cls_2": 1.10784, "loss_box_2": 2.27695, "loss_cns_2": 0.61846, "loss_yns_2": 0.16771, "loss_cls_3": 1.11245, "loss_box_3": 2.27185, "loss_cns_3": 0.62752, "loss_yns_3": 0.16531, "loss_cls_4": 1.11797, "loss_box_4": 2.27192, "loss_cns_4": 0.62417, "loss_yns_4": 0.16805, "loss_cls_5": 1.121, "loss_box_5": 2.28167, "loss_cns_5": 0.6272, "loss_yns_5": 0.16481, "loss_cls_dn_0": 0.40777, "loss_box_dn_0": 0.91764, "loss_cls_dn_1": 0.36812, "loss_box_dn_1": 1.09388, "loss_cls_dn_2": 0.38662, "loss_box_dn_2": 1.06164, "loss_cls_dn_3": 0.39586, "loss_box_dn_3": 1.0726, "loss_cls_dn_4": 0.36735, "loss_box_dn_4": 1.0708, "loss_cls_dn_5": 0.38973, "loss_box_dn_5": 1.11429, "loss_dense_depth": 1.01352, "loss": 34.39054, "grad_norm": 52.53603, "time": 1.60871} -{"mode": "train", "epoch": 1, "iter": 47, "lr": 0.00012, "memory": 49164, "data_time": 0.08146, "loss_cls_0": 1.0766, "loss_box_0": 1.96539, "loss_cns_0": 0.62212, "loss_yns_0": 0.16158, "loss_cls_1": 1.13484, "loss_box_1": 2.35961, "loss_cns_1": 0.60329, "loss_yns_1": 0.16367, "loss_cls_2": 1.12109, "loss_box_2": 2.28985, "loss_cns_2": 0.62528, "loss_yns_2": 0.16685, "loss_cls_3": 1.13906, "loss_box_3": 2.26052, "loss_cns_3": 0.63316, "loss_yns_3": 0.16405, "loss_cls_4": 1.13731, "loss_box_4": 2.2659, "loss_cns_4": 0.63312, "loss_yns_4": 0.16746, "loss_cls_5": 1.14124, "loss_box_5": 2.27384, "loss_cns_5": 0.63775, "loss_yns_5": 0.16395, "loss_cls_dn_0": 0.39672, "loss_box_dn_0": 0.90996, "loss_cls_dn_1": 0.35708, "loss_box_dn_1": 1.01548, "loss_cls_dn_2": 0.38667, "loss_box_dn_2": 0.98348, "loss_cls_dn_3": 0.39413, "loss_box_dn_3": 0.97215, "loss_cls_dn_4": 0.35612, "loss_box_dn_4": 0.97064, "loss_cls_dn_5": 0.38529, "loss_box_dn_5": 0.98922, "loss_dense_depth": 0.94525, "loss": 33.96971, "grad_norm": 44.93067, "time": 1.58622} -{"mode": "train", "epoch": 1, "iter": 48, "lr": 0.00012, "memory": 49164, "data_time": 0.08849, "loss_cls_0": 1.04115, "loss_box_0": 1.95067, "loss_cns_0": 0.61949, "loss_yns_0": 0.16356, "loss_cls_1": 1.13179, "loss_box_1": 2.32833, "loss_cns_1": 0.59974, "loss_yns_1": 0.16374, "loss_cls_2": 1.12485, "loss_box_2": 2.29283, "loss_cns_2": 0.62361, "loss_yns_2": 0.16676, "loss_cls_3": 1.14534, "loss_box_3": 2.25621, "loss_cns_3": 0.62987, "loss_yns_3": 0.16988, "loss_cls_4": 1.15255, "loss_box_4": 2.27297, "loss_cns_4": 0.63311, "loss_yns_4": 0.16895, "loss_cls_5": 1.15888, "loss_box_5": 2.2567, "loss_cns_5": 0.63645, "loss_yns_5": 0.16688, "loss_cls_dn_0": 0.39316, "loss_box_dn_0": 0.90429, "loss_cls_dn_1": 0.32773, "loss_box_dn_1": 1.03586, "loss_cls_dn_2": 0.37331, "loss_box_dn_2": 0.99971, "loss_cls_dn_3": 0.36832, "loss_box_dn_3": 0.97993, "loss_cls_dn_4": 0.34942, "loss_box_dn_4": 0.98614, "loss_cls_dn_5": 0.3736, "loss_box_dn_5": 0.98131, "loss_dense_depth": 1.08185, "loss": 34.00896, "grad_norm": 47.52842, "time": 1.59522} -{"mode": "train", "epoch": 1, "iter": 49, "lr": 0.00012, "memory": 49164, "data_time": 0.08337, "loss_cls_0": 1.03975, "loss_box_0": 1.95538, "loss_cns_0": 0.61612, "loss_yns_0": 0.16455, "loss_cls_1": 1.11518, "loss_box_1": 2.29647, "loss_cns_1": 0.60926, "loss_yns_1": 0.16555, "loss_cls_2": 1.11828, "loss_box_2": 2.27846, "loss_cns_2": 0.62687, "loss_yns_2": 0.16686, "loss_cls_3": 1.13577, "loss_box_3": 2.24176, "loss_cns_3": 0.63269, "loss_yns_3": 0.16817, "loss_cls_4": 1.12169, "loss_box_4": 2.27458, "loss_cns_4": 0.63601, "loss_yns_4": 0.17049, "loss_cls_5": 1.1332, "loss_box_5": 2.26849, "loss_cns_5": 0.63408, "loss_yns_5": 0.16722, "loss_cls_dn_0": 0.40157, "loss_box_dn_0": 0.89729, "loss_cls_dn_1": 0.30634, "loss_box_dn_1": 1.02039, "loss_cls_dn_2": 0.35002, "loss_box_dn_2": 0.99012, "loss_cls_dn_3": 0.34166, "loss_box_dn_3": 0.97405, "loss_cls_dn_4": 0.35238, "loss_box_dn_4": 0.99989, "loss_cls_dn_5": 0.36744, "loss_box_dn_5": 1.00089, "loss_dense_depth": 0.97906, "loss": 33.71797, "grad_norm": 43.65452, "time": 1.60631} -{"mode": "train", "epoch": 1, "iter": 50, "lr": 0.00012, "memory": 49164, "data_time": 0.0821, "loss_cls_0": 1.02266, "loss_box_0": 1.94107, "loss_cns_0": 0.61764, "loss_yns_0": 0.16209, "loss_cls_1": 1.11665, "loss_box_1": 2.30793, "loss_cns_1": 0.62025, "loss_yns_1": 0.16436, "loss_cls_2": 1.1154, "loss_box_2": 2.25296, "loss_cns_2": 0.62972, "loss_yns_2": 0.1646, "loss_cls_3": 1.14128, "loss_box_3": 2.28481, "loss_cns_3": 0.63213, "loss_yns_3": 0.16537, "loss_cls_4": 1.10687, "loss_box_4": 2.29657, "loss_cns_4": 0.63468, "loss_yns_4": 0.17408, "loss_cls_5": 1.1102, "loss_box_5": 2.30835, "loss_cns_5": 0.63028, "loss_yns_5": 0.16606, "loss_cls_dn_0": 0.39985, "loss_box_dn_0": 0.9016, "loss_cls_dn_1": 0.31267, "loss_box_dn_1": 0.99848, "loss_cls_dn_2": 0.34091, "loss_box_dn_2": 0.97794, "loss_cls_dn_3": 0.33882, "loss_box_dn_3": 1.00604, "loss_cls_dn_4": 0.36617, "loss_box_dn_4": 1.05564, "loss_cls_dn_5": 0.37679, "loss_box_dn_5": 1.07989, "loss_dense_depth": 1.08508, "loss": 34.00588, "grad_norm": 49.63653, "time": 1.57413} -{"mode": "train", "epoch": 1, "iter": 51, "lr": 0.00012, "memory": 49164, "data_time": 0.07724, "loss_cls_0": 1.01044, "loss_box_0": 1.92224, "loss_cns_0": 0.61943, "loss_yns_0": 0.16172, "loss_cls_1": 1.09804, "loss_box_1": 2.44749, "loss_cns_1": 0.61639, "loss_yns_1": 0.16247, "loss_cls_2": 1.11911, "loss_box_2": 2.39992, "loss_cns_2": 0.62555, "loss_yns_2": 0.16471, "loss_cls_3": 1.15273, "loss_box_3": 2.42459, "loss_cns_3": 0.63034, "loss_yns_3": 0.16515, "loss_cls_4": 1.12118, "loss_box_4": 2.4284, "loss_cns_4": 0.63098, "loss_yns_4": 0.17388, "loss_cls_5": 1.12556, "loss_box_5": 2.42501, "loss_cns_5": 0.63387, "loss_yns_5": 0.1655, "loss_cls_dn_0": 0.40856, "loss_box_dn_0": 0.88819, "loss_cls_dn_1": 0.31414, "loss_box_dn_1": 1.06506, "loss_cls_dn_2": 0.33713, "loss_box_dn_2": 1.04461, "loss_cls_dn_3": 0.33258, "loss_box_dn_3": 1.07735, "loss_cls_dn_4": 0.34597, "loss_box_dn_4": 1.12573, "loss_cls_dn_5": 0.35867, "loss_box_dn_5": 1.14655, "loss_dense_depth": 0.92082, "loss": 34.79008, "grad_norm": 53.70309, "time": 1.57938} -{"mode": "train", "epoch": 1, "iter": 52, "lr": 0.00012, "memory": 49164, "data_time": 0.08199, "loss_cls_0": 1.01331, "loss_box_0": 1.94548, "loss_cns_0": 0.61553, "loss_yns_0": 0.162, "loss_cls_1": 1.10713, "loss_box_1": 2.51204, "loss_cns_1": 0.61009, "loss_yns_1": 0.16247, "loss_cls_2": 1.11685, "loss_box_2": 2.4809, "loss_cns_2": 0.62018, "loss_yns_2": 0.1629, "loss_cls_3": 1.13759, "loss_box_3": 2.45963, "loss_cns_3": 0.62444, "loss_yns_3": 0.16436, "loss_cls_4": 1.15583, "loss_box_4": 2.46665, "loss_cns_4": 0.62216, "loss_yns_4": 0.16671, "loss_cls_5": 1.16354, "loss_box_5": 2.46908, "loss_cns_5": 0.62573, "loss_yns_5": 0.16233, "loss_cls_dn_0": 0.41776, "loss_box_dn_0": 0.88664, "loss_cls_dn_1": 0.32785, "loss_box_dn_1": 1.07832, "loss_cls_dn_2": 0.34686, "loss_box_dn_2": 1.06497, "loss_cls_dn_3": 0.34756, "loss_box_dn_3": 1.07847, "loss_cls_dn_4": 0.33397, "loss_box_dn_4": 1.12269, "loss_cls_dn_5": 0.3512, "loss_box_dn_5": 1.13897, "loss_dense_depth": 1.02679, "loss": 35.24901, "grad_norm": 52.40639, "time": 1.57107} -{"mode": "train", "epoch": 1, "iter": 53, "lr": 0.00012, "memory": 49164, "data_time": 0.07999, "loss_cls_0": 1.02057, "loss_box_0": 1.97297, "loss_cns_0": 0.61727, "loss_yns_0": 0.16339, "loss_cls_1": 1.10031, "loss_box_1": 2.37558, "loss_cns_1": 0.61271, "loss_yns_1": 0.16317, "loss_cls_2": 1.11208, "loss_box_2": 2.35979, "loss_cns_2": 0.62307, "loss_yns_2": 0.16529, "loss_cls_3": 1.11352, "loss_box_3": 2.36093, "loss_cns_3": 0.62539, "loss_yns_3": 0.1658, "loss_cls_4": 1.1398, "loss_box_4": 2.37999, "loss_cns_4": 0.62229, "loss_yns_4": 0.16322, "loss_cls_5": 1.15685, "loss_box_5": 2.3686, "loss_cns_5": 0.62767, "loss_yns_5": 0.16295, "loss_cls_dn_0": 0.41654, "loss_box_dn_0": 0.87809, "loss_cls_dn_1": 0.32792, "loss_box_dn_1": 0.98306, "loss_cls_dn_2": 0.35427, "loss_box_dn_2": 0.9656, "loss_cls_dn_3": 0.362, "loss_box_dn_3": 0.96974, "loss_cls_dn_4": 0.32936, "loss_box_dn_4": 0.99797, "loss_cls_dn_5": 0.34705, "loss_box_dn_5": 1.00277, "loss_dense_depth": 0.97581, "loss": 34.08337, "grad_norm": 37.80824, "time": 1.56613} -{"mode": "train", "epoch": 1, "iter": 54, "lr": 0.00012, "memory": 49164, "data_time": 0.07429, "loss_cls_0": 1.02336, "loss_box_0": 1.90433, "loss_cns_0": 0.62651, "loss_yns_0": 0.16129, "loss_cls_1": 1.07369, "loss_box_1": 2.25324, "loss_cns_1": 0.61776, "loss_yns_1": 0.16342, "loss_cls_2": 1.0974, "loss_box_2": 2.21872, "loss_cns_2": 0.63145, "loss_yns_2": 0.1629, "loss_cls_3": 1.09955, "loss_box_3": 2.23819, "loss_cns_3": 0.63877, "loss_yns_3": 0.16148, "loss_cls_4": 1.11264, "loss_box_4": 2.2167, "loss_cns_4": 0.64176, "loss_yns_4": 0.16138, "loss_cls_5": 1.11147, "loss_box_5": 2.20977, "loss_cns_5": 0.64286, "loss_yns_5": 0.16123, "loss_cls_dn_0": 0.3821, "loss_box_dn_0": 0.88915, "loss_cls_dn_1": 0.2988, "loss_box_dn_1": 0.97578, "loss_cls_dn_2": 0.33604, "loss_box_dn_2": 0.94921, "loss_cls_dn_3": 0.35438, "loss_box_dn_3": 0.95848, "loss_cls_dn_4": 0.33041, "loss_box_dn_4": 0.95868, "loss_cls_dn_5": 0.3424, "loss_box_dn_5": 0.95973, "loss_dense_depth": 0.99184, "loss": 33.05686, "grad_norm": 40.29876, "time": 1.56715} -{"mode": "train", "epoch": 1, "iter": 55, "lr": 0.00012, "memory": 49164, "data_time": 0.07922, "loss_cls_0": 1.05034, "loss_box_0": 1.9078, "loss_cns_0": 0.6213, "loss_yns_0": 0.16334, "loss_cls_1": 1.09785, "loss_box_1": 2.22449, "loss_cns_1": 0.61375, "loss_yns_1": 0.16461, "loss_cls_2": 1.11394, "loss_box_2": 2.16674, "loss_cns_2": 0.6309, "loss_yns_2": 0.16623, "loss_cls_3": 1.10749, "loss_box_3": 2.18209, "loss_cns_3": 0.63496, "loss_yns_3": 0.16581, "loss_cls_4": 1.11651, "loss_box_4": 2.15432, "loss_cns_4": 0.63859, "loss_yns_4": 0.17109, "loss_cls_5": 1.11693, "loss_box_5": 2.17217, "loss_cns_5": 0.63881, "loss_yns_5": 0.16484, "loss_cls_dn_0": 0.37144, "loss_box_dn_0": 0.87933, "loss_cls_dn_1": 0.29754, "loss_box_dn_1": 0.86473, "loss_cls_dn_2": 0.34338, "loss_box_dn_2": 0.83582, "loss_cls_dn_3": 0.37091, "loss_box_dn_3": 0.85297, "loss_cls_dn_4": 0.37244, "loss_box_dn_4": 0.8465, "loss_cls_dn_5": 0.38919, "loss_box_dn_5": 0.8762, "loss_dense_depth": 0.96816, "loss": 32.45351, "grad_norm": 43.015, "time": 1.58182} -{"mode": "train", "epoch": 1, "iter": 56, "lr": 0.00012, "memory": 49164, "data_time": 0.07396, "loss_cls_0": 0.99871, "loss_box_0": 1.89675, "loss_cns_0": 0.61451, "loss_yns_0": 0.16189, "loss_cls_1": 1.10929, "loss_box_1": 2.10042, "loss_cns_1": 0.61794, "loss_yns_1": 0.16249, "loss_cls_2": 1.09823, "loss_box_2": 2.04297, "loss_cns_2": 0.63133, "loss_yns_2": 0.16374, "loss_cls_3": 1.08815, "loss_box_3": 2.04698, "loss_cns_3": 0.63444, "loss_yns_3": 0.16575, "loss_cls_4": 1.08387, "loss_box_4": 2.0296, "loss_cns_4": 0.63765, "loss_yns_4": 0.17162, "loss_cls_5": 1.09598, "loss_box_5": 2.03858, "loss_cns_5": 0.63675, "loss_yns_5": 0.1633, "loss_cls_dn_0": 0.36643, "loss_box_dn_0": 0.88029, "loss_cls_dn_1": 0.28618, "loss_box_dn_1": 0.88314, "loss_cls_dn_2": 0.32827, "loss_box_dn_2": 0.87418, "loss_cls_dn_3": 0.34468, "loss_box_dn_3": 0.89649, "loss_cls_dn_4": 0.36932, "loss_box_dn_4": 0.90118, "loss_cls_dn_5": 0.39173, "loss_box_dn_5": 0.94779, "loss_dense_depth": 0.94371, "loss": 31.80433, "grad_norm": 44.9614, "time": 1.57074} -{"mode": "train", "epoch": 1, "iter": 57, "lr": 0.00012, "memory": 49164, "data_time": 0.08314, "loss_cls_0": 0.976, "loss_box_0": 1.87058, "loss_cns_0": 0.61777, "loss_yns_0": 0.16008, "loss_cls_1": 1.05705, "loss_box_1": 2.07832, "loss_cns_1": 0.62645, "loss_yns_1": 0.16196, "loss_cls_2": 1.10059, "loss_box_2": 2.0471, "loss_cns_2": 0.63755, "loss_yns_2": 0.16307, "loss_cls_3": 1.12733, "loss_box_3": 2.0564, "loss_cns_3": 0.63914, "loss_yns_3": 0.16663, "loss_cls_4": 1.08242, "loss_box_4": 2.07107, "loss_cns_4": 0.63816, "loss_yns_4": 0.1661, "loss_cls_5": 1.08355, "loss_box_5": 2.08612, "loss_cns_5": 0.63645, "loss_yns_5": 0.16429, "loss_cls_dn_0": 0.38182, "loss_box_dn_0": 0.86896, "loss_cls_dn_1": 0.29362, "loss_box_dn_1": 0.90438, "loss_cls_dn_2": 0.32448, "loss_box_dn_2": 0.90968, "loss_cls_dn_3": 0.3223, "loss_box_dn_3": 0.93913, "loss_cls_dn_4": 0.3402, "loss_box_dn_4": 0.95222, "loss_cls_dn_5": 0.3751, "loss_box_dn_5": 1.0121, "loss_dense_depth": 0.97715, "loss": 32.01532, "grad_norm": 61.79861, "time": 1.57266} -{"mode": "train", "epoch": 1, "iter": 58, "lr": 0.00012, "memory": 49164, "data_time": 0.07904, "loss_cls_0": 0.9698, "loss_box_0": 1.87416, "loss_cns_0": 0.61569, "loss_yns_0": 0.16161, "loss_cls_1": 1.05525, "loss_box_1": 2.08809, "loss_cns_1": 0.62957, "loss_yns_1": 0.16432, "loss_cls_2": 1.08567, "loss_box_2": 2.06202, "loss_cns_2": 0.63908, "loss_yns_2": 0.16863, "loss_cls_3": 1.12434, "loss_box_3": 2.07014, "loss_cns_3": 0.64187, "loss_yns_3": 0.16635, "loss_cls_4": 1.12517, "loss_box_4": 2.08428, "loss_cns_4": 0.63495, "loss_yns_4": 0.16564, "loss_cls_5": 1.08529, "loss_box_5": 2.11434, "loss_cns_5": 0.63906, "loss_yns_5": 0.1658, "loss_cls_dn_0": 0.39316, "loss_box_dn_0": 0.85644, "loss_cls_dn_1": 0.30934, "loss_box_dn_1": 0.93653, "loss_cls_dn_2": 0.33227, "loss_box_dn_2": 0.9402, "loss_cls_dn_3": 0.31644, "loss_box_dn_3": 0.96969, "loss_cls_dn_4": 0.31106, "loss_box_dn_4": 0.97411, "loss_cls_dn_5": 0.35097, "loss_box_dn_5": 1.03503, "loss_dense_depth": 0.90264, "loss": 32.15901, "grad_norm": 62.87479, "time": 1.57305} -{"mode": "train", "epoch": 1, "iter": 59, "lr": 0.00012, "memory": 49164, "data_time": 0.08196, "loss_cls_0": 0.96808, "loss_box_0": 1.89166, "loss_cns_0": 0.61381, "loss_yns_0": 0.16147, "loss_cls_1": 1.07534, "loss_box_1": 2.17439, "loss_cns_1": 0.62561, "loss_yns_1": 0.16411, "loss_cls_2": 1.06973, "loss_box_2": 2.12101, "loss_cns_2": 0.63856, "loss_yns_2": 0.16964, "loss_cls_3": 1.07504, "loss_box_3": 2.14366, "loss_cns_3": 0.64227, "loss_yns_3": 0.16555, "loss_cls_4": 1.11058, "loss_box_4": 2.10028, "loss_cns_4": 0.64379, "loss_yns_4": 0.16452, "loss_cls_5": 1.10909, "loss_box_5": 2.10817, "loss_cns_5": 0.64933, "loss_yns_5": 0.16527, "loss_cls_dn_0": 0.39286, "loss_box_dn_0": 0.84444, "loss_cls_dn_1": 0.31977, "loss_box_dn_1": 0.93498, "loss_cls_dn_2": 0.33905, "loss_box_dn_2": 0.92136, "loss_cls_dn_3": 0.32747, "loss_box_dn_3": 0.94791, "loss_cls_dn_4": 0.30375, "loss_box_dn_4": 0.92968, "loss_cls_dn_5": 0.33029, "loss_box_dn_5": 0.96526, "loss_dense_depth": 0.9092, "loss": 32.21697, "grad_norm": 45.4584, "time": 1.57004} -{"mode": "train", "epoch": 1, "iter": 60, "lr": 0.00012, "memory": 49164, "data_time": 0.07056, "loss_cls_0": 0.98006, "loss_box_0": 1.87564, "loss_cns_0": 0.62027, "loss_yns_0": 0.16021, "loss_cls_1": 1.05926, "loss_box_1": 2.10575, "loss_cns_1": 0.63061, "loss_yns_1": 0.1613, "loss_cls_2": 1.0765, "loss_box_2": 2.05499, "loss_cns_2": 0.64306, "loss_yns_2": 0.16437, "loss_cls_3": 1.09184, "loss_box_3": 2.09096, "loss_cns_3": 0.64488, "loss_yns_3": 0.16276, "loss_cls_4": 1.08172, "loss_box_4": 2.07759, "loss_cns_4": 0.64476, "loss_yns_4": 0.16526, "loss_cls_5": 1.1201, "loss_box_5": 2.08375, "loss_cns_5": 0.64688, "loss_yns_5": 0.16393, "loss_cls_dn_0": 0.38011, "loss_box_dn_0": 0.85638, "loss_cls_dn_1": 0.29734, "loss_box_dn_1": 0.93148, "loss_cls_dn_2": 0.32402, "loss_box_dn_2": 0.91125, "loss_cls_dn_3": 0.34159, "loss_box_dn_3": 0.93167, "loss_cls_dn_4": 0.32414, "loss_box_dn_4": 0.92217, "loss_cls_dn_5": 0.32462, "loss_box_dn_5": 0.93892, "loss_dense_depth": 0.8991, "loss": 31.88925, "grad_norm": 57.55884, "time": 1.56999} -{"mode": "train", "epoch": 1, "iter": 61, "lr": 0.00012, "memory": 49164, "data_time": 0.10511, "loss_cls_0": 0.995, "loss_box_0": 1.89075, "loss_cns_0": 0.6164, "loss_yns_0": 0.15851, "loss_cls_1": 1.04692, "loss_box_1": 2.16142, "loss_cns_1": 0.62105, "loss_yns_1": 0.1603, "loss_cls_2": 1.11084, "loss_box_2": 2.08777, "loss_cns_2": 0.63906, "loss_yns_2": 0.16122, "loss_cls_3": 1.0992, "loss_box_3": 2.08011, "loss_cns_3": 0.64507, "loss_yns_3": 0.16252, "loss_cls_4": 1.08822, "loss_box_4": 2.09189, "loss_cns_4": 0.64313, "loss_yns_4": 0.16611, "loss_cls_5": 1.10612, "loss_box_5": 2.07308, "loss_cns_5": 0.64514, "loss_yns_5": 0.16096, "loss_cls_dn_0": 0.37065, "loss_box_dn_0": 0.85545, "loss_cls_dn_1": 0.26797, "loss_box_dn_1": 0.95579, "loss_cls_dn_2": 0.30082, "loss_box_dn_2": 0.90726, "loss_cls_dn_3": 0.33915, "loss_box_dn_3": 0.89777, "loss_cls_dn_4": 0.33956, "loss_box_dn_4": 0.89914, "loss_cls_dn_5": 0.32495, "loss_box_dn_5": 0.89496, "loss_dense_depth": 0.93034, "loss": 31.89459, "grad_norm": 56.03912, "time": 1.62577} -{"mode": "train", "epoch": 1, "iter": 62, "lr": 0.00012, "memory": 49164, "data_time": 0.14086, "loss_cls_0": 0.99479, "loss_box_0": 1.89425, "loss_cns_0": 0.61342, "loss_yns_0": 0.15872, "loss_cls_1": 1.12911, "loss_box_1": 2.19272, "loss_cns_1": 0.62231, "loss_yns_1": 0.16202, "loss_cls_2": 1.13808, "loss_box_2": 2.13751, "loss_cns_2": 0.63742, "loss_yns_2": 0.16471, "loss_cls_3": 1.0824, "loss_box_3": 2.11183, "loss_cns_3": 0.64232, "loss_yns_3": 0.16306, "loss_cls_4": 1.08039, "loss_box_4": 2.10471, "loss_cns_4": 0.64455, "loss_yns_4": 0.16856, "loss_cls_5": 1.09211, "loss_box_5": 2.09552, "loss_cns_5": 0.64319, "loss_yns_5": 0.16391, "loss_cls_dn_0": 0.36927, "loss_box_dn_0": 0.85915, "loss_cls_dn_1": 0.25701, "loss_box_dn_1": 0.88761, "loss_cls_dn_2": 0.29544, "loss_box_dn_2": 0.84672, "loss_cls_dn_3": 0.34609, "loss_box_dn_3": 0.83976, "loss_cls_dn_4": 0.3506, "loss_box_dn_4": 0.85567, "loss_cls_dn_5": 0.3527, "loss_box_dn_5": 0.86917, "loss_dense_depth": 0.89371, "loss": 31.86049, "grad_norm": 42.97277, "time": 1.61417} -{"mode": "train", "epoch": 1, "iter": 63, "lr": 0.00012, "memory": 49164, "data_time": 0.07981, "loss_cls_0": 0.98909, "loss_box_0": 1.87076, "loss_cns_0": 0.61685, "loss_yns_0": 0.16135, "loss_cls_1": 1.08716, "loss_box_1": 2.19715, "loss_cns_1": 0.62452, "loss_yns_1": 0.16527, "loss_cls_2": 1.07938, "loss_box_2": 2.14252, "loss_cns_2": 0.64199, "loss_yns_2": 0.1674, "loss_cls_3": 1.0955, "loss_box_3": 2.14881, "loss_cns_3": 0.64498, "loss_yns_3": 0.16352, "loss_cls_4": 1.08487, "loss_box_4": 2.16584, "loss_cns_4": 0.64657, "loss_yns_4": 0.16391, "loss_cls_5": 1.09749, "loss_box_5": 2.18618, "loss_cns_5": 0.64357, "loss_yns_5": 0.1648, "loss_cls_dn_0": 0.37308, "loss_box_dn_0": 0.85152, "loss_cls_dn_1": 0.25244, "loss_box_dn_1": 0.88743, "loss_cls_dn_2": 0.30717, "loss_box_dn_2": 0.86927, "loss_cls_dn_3": 0.32925, "loss_box_dn_3": 0.88592, "loss_cls_dn_4": 0.32053, "loss_box_dn_4": 0.92396, "loss_cls_dn_5": 0.35712, "loss_box_dn_5": 0.95947, "loss_dense_depth": 0.93069, "loss": 32.19732, "grad_norm": 58.38211, "time": 1.59272} -{"mode": "train", "epoch": 1, "iter": 64, "lr": 0.00013, "memory": 49164, "data_time": 0.08237, "loss_cls_0": 0.98695, "loss_box_0": 1.86231, "loss_cns_0": 0.62085, "loss_yns_0": 0.16274, "loss_cls_1": 1.05493, "loss_box_1": 2.30054, "loss_cns_1": 0.62409, "loss_yns_1": 0.16783, "loss_cls_2": 1.0787, "loss_box_2": 2.23705, "loss_cns_2": 0.63806, "loss_yns_2": 0.16722, "loss_cls_3": 1.1047, "loss_box_3": 2.22326, "loss_cns_3": 0.64162, "loss_yns_3": 0.16314, "loss_cls_4": 1.12217, "loss_box_4": 2.24927, "loss_cns_4": 0.6417, "loss_yns_4": 0.16429, "loss_cls_5": 1.08589, "loss_box_5": 2.23645, "loss_cns_5": 0.64062, "loss_yns_5": 0.16395, "loss_cls_dn_0": 0.36682, "loss_box_dn_0": 0.85286, "loss_cls_dn_1": 0.25126, "loss_box_dn_1": 0.9525, "loss_cls_dn_2": 0.31279, "loss_box_dn_2": 0.94293, "loss_cls_dn_3": 0.30101, "loss_box_dn_3": 0.95597, "loss_cls_dn_4": 0.27862, "loss_box_dn_4": 1.00282, "loss_cls_dn_5": 0.33982, "loss_box_dn_5": 1.02725, "loss_dense_depth": 0.89994, "loss": 32.82291, "grad_norm": 52.13807, "time": 1.60974} -{"mode": "train", "epoch": 1, "iter": 65, "lr": 0.00013, "memory": 49164, "data_time": 0.08482, "loss_cls_0": 0.97738, "loss_box_0": 1.86257, "loss_cns_0": 0.61972, "loss_yns_0": 0.1628, "loss_cls_1": 1.07429, "loss_box_1": 2.17923, "loss_cns_1": 0.63008, "loss_yns_1": 0.16577, "loss_cls_2": 1.07729, "loss_box_2": 2.12635, "loss_cns_2": 0.64175, "loss_yns_2": 0.16431, "loss_cls_3": 1.09732, "loss_box_3": 2.11834, "loss_cns_3": 0.64814, "loss_yns_3": 0.16365, "loss_cls_4": 1.12925, "loss_box_4": 2.14082, "loss_cns_4": 0.64591, "loss_yns_4": 0.1679, "loss_cls_5": 1.09604, "loss_box_5": 2.12644, "loss_cns_5": 0.64524, "loss_yns_5": 0.16349, "loss_cls_dn_0": 0.37031, "loss_box_dn_0": 0.86147, "loss_cls_dn_1": 0.25096, "loss_box_dn_1": 1.03827, "loss_cls_dn_2": 0.3035, "loss_box_dn_2": 1.02105, "loss_cls_dn_3": 0.28344, "loss_box_dn_3": 1.023, "loss_cls_dn_4": 0.26234, "loss_box_dn_4": 1.05684, "loss_cls_dn_5": 0.32618, "loss_box_dn_5": 1.07221, "loss_dense_depth": 0.86648, "loss": 32.56013, "grad_norm": 56.16266, "time": 1.61115} -{"mode": "train", "epoch": 1, "iter": 66, "lr": 0.00013, "memory": 49164, "data_time": 0.10417, "loss_cls_0": 0.95778, "loss_box_0": 1.81885, "loss_cns_0": 0.62392, "loss_yns_0": 0.1601, "loss_cls_1": 1.0736, "loss_box_1": 2.06745, "loss_cns_1": 0.63273, "loss_yns_1": 0.16739, "loss_cls_2": 1.07482, "loss_box_2": 2.03744, "loss_cns_2": 0.64523, "loss_yns_2": 0.16859, "loss_cls_3": 1.08989, "loss_box_3": 2.07585, "loss_cns_3": 0.6486, "loss_yns_3": 0.16309, "loss_cls_4": 1.10153, "loss_box_4": 2.08495, "loss_cns_4": 0.64765, "loss_yns_4": 0.16496, "loss_cls_5": 1.10514, "loss_box_5": 2.10615, "loss_cns_5": 0.64536, "loss_yns_5": 0.16277, "loss_cls_dn_0": 0.35124, "loss_box_dn_0": 0.84552, "loss_cls_dn_1": 0.24209, "loss_box_dn_1": 1.06819, "loss_cls_dn_2": 0.28414, "loss_box_dn_2": 1.04892, "loss_cls_dn_3": 0.28131, "loss_box_dn_3": 1.05506, "loss_cls_dn_4": 0.268, "loss_box_dn_4": 1.07014, "loss_cls_dn_5": 0.31101, "loss_box_dn_5": 1.09011, "loss_dense_depth": 0.89533, "loss": 32.2349, "grad_norm": 63.01182, "time": 1.58779} -{"mode": "train", "epoch": 1, "iter": 67, "lr": 0.00013, "memory": 49164, "data_time": 0.07254, "loss_cls_0": 0.99648, "loss_box_0": 1.84965, "loss_cns_0": 0.62275, "loss_yns_0": 0.16256, "loss_cls_1": 1.02292, "loss_box_1": 2.0403, "loss_cns_1": 0.6272, "loss_yns_1": 0.16239, "loss_cls_2": 1.06993, "loss_box_2": 2.0146, "loss_cns_2": 0.64188, "loss_yns_2": 0.16269, "loss_cls_3": 1.06564, "loss_box_3": 2.06267, "loss_cns_3": 0.64518, "loss_yns_3": 0.16176, "loss_cls_4": 1.05674, "loss_box_4": 2.02398, "loss_cns_4": 0.64544, "loss_yns_4": 0.16177, "loss_cls_5": 1.07335, "loss_box_5": 2.02994, "loss_cns_5": 0.64522, "loss_yns_5": 0.16144, "loss_cls_dn_0": 0.349, "loss_box_dn_0": 0.85019, "loss_cls_dn_1": 0.24239, "loss_box_dn_1": 0.97407, "loss_cls_dn_2": 0.28308, "loss_box_dn_2": 0.95082, "loss_cls_dn_3": 0.30193, "loss_box_dn_3": 0.96192, "loss_cls_dn_4": 0.30359, "loss_box_dn_4": 0.95415, "loss_cls_dn_5": 0.31038, "loss_box_dn_5": 0.96149, "loss_dense_depth": 0.86707, "loss": 31.41658, "grad_norm": 46.73452, "time": 1.55356} -{"mode": "train", "epoch": 1, "iter": 68, "lr": 0.00013, "memory": 49164, "data_time": 0.07477, "loss_cls_0": 0.99111, "loss_box_0": 1.85735, "loss_cns_0": 0.62454, "loss_yns_0": 0.16206, "loss_cls_1": 1.06001, "loss_box_1": 2.01864, "loss_cns_1": 0.62963, "loss_yns_1": 0.16363, "loss_cls_2": 1.09887, "loss_box_2": 1.97755, "loss_cns_2": 0.6438, "loss_yns_2": 0.16254, "loss_cls_3": 1.09205, "loss_box_3": 2.02195, "loss_cns_3": 0.64879, "loss_yns_3": 0.16249, "loss_cls_4": 1.0921, "loss_box_4": 2.00163, "loss_cns_4": 0.64726, "loss_yns_4": 0.16489, "loss_cls_5": 1.10922, "loss_box_5": 1.96752, "loss_cns_5": 0.64602, "loss_yns_5": 0.16363, "loss_cls_dn_0": 0.35511, "loss_box_dn_0": 0.8581, "loss_cls_dn_1": 0.24364, "loss_box_dn_1": 0.87531, "loss_cls_dn_2": 0.28942, "loss_box_dn_2": 0.85441, "loss_cls_dn_3": 0.3229, "loss_box_dn_3": 0.87302, "loss_cls_dn_4": 0.34339, "loss_box_dn_4": 0.86855, "loss_cls_dn_5": 0.31936, "loss_box_dn_5": 0.8586, "loss_dense_depth": 0.9084, "loss": 31.07751, "grad_norm": 58.56116, "time": 1.56492} -{"mode": "train", "epoch": 1, "iter": 69, "lr": 0.00013, "memory": 49164, "data_time": 0.07504, "loss_cls_0": 0.93109, "loss_box_0": 1.79581, "loss_cns_0": 0.62409, "loss_yns_0": 0.15933, "loss_cls_1": 1.05949, "loss_box_1": 2.01518, "loss_cns_1": 0.63369, "loss_yns_1": 0.16339, "loss_cls_2": 1.08161, "loss_box_2": 1.99196, "loss_cns_2": 0.64883, "loss_yns_2": 0.16204, "loss_cls_3": 1.10494, "loss_box_3": 2.02527, "loss_cns_3": 0.64897, "loss_yns_3": 0.16014, "loss_cls_4": 1.11426, "loss_box_4": 2.01692, "loss_cns_4": 0.64723, "loss_yns_4": 0.1611, "loss_cls_5": 1.128, "loss_box_5": 1.98859, "loss_cns_5": 0.6449, "loss_yns_5": 0.16311, "loss_cls_dn_0": 0.36319, "loss_box_dn_0": 0.83595, "loss_cls_dn_1": 0.25039, "loss_box_dn_1": 0.87214, "loss_cls_dn_2": 0.29924, "loss_box_dn_2": 0.86248, "loss_cls_dn_3": 0.3278, "loss_box_dn_3": 0.88569, "loss_cls_dn_4": 0.34872, "loss_box_dn_4": 0.885, "loss_cls_dn_5": 0.3256, "loss_box_dn_5": 0.87983, "loss_dense_depth": 0.84622, "loss": 31.05217, "grad_norm": 65.12071, "time": 1.60769} -{"mode": "train", "epoch": 1, "iter": 70, "lr": 0.00013, "memory": 49164, "data_time": 0.07331, "loss_cls_0": 0.94672, "loss_box_0": 1.79859, "loss_cns_0": 0.62076, "loss_yns_0": 0.15632, "loss_cls_1": 0.99947, "loss_box_1": 2.03145, "loss_cns_1": 0.62513, "loss_yns_1": 0.16365, "loss_cls_2": 1.04631, "loss_box_2": 1.99601, "loss_cns_2": 0.64153, "loss_yns_2": 0.16077, "loss_cls_3": 1.05444, "loss_box_3": 1.98519, "loss_cns_3": 0.64531, "loss_yns_3": 0.15986, "loss_cls_4": 1.07602, "loss_box_4": 1.98867, "loss_cns_4": 0.64449, "loss_yns_4": 0.1611, "loss_cls_5": 1.05646, "loss_box_5": 1.97627, "loss_cns_5": 0.64247, "loss_yns_5": 0.16298, "loss_cls_dn_0": 0.39137, "loss_box_dn_0": 0.84326, "loss_cls_dn_1": 0.26078, "loss_box_dn_1": 0.83594, "loss_cls_dn_2": 0.31409, "loss_box_dn_2": 0.8353, "loss_cls_dn_3": 0.33373, "loss_box_dn_3": 0.8564, "loss_cls_dn_4": 0.3413, "loss_box_dn_4": 0.87299, "loss_cls_dn_5": 0.34168, "loss_box_dn_5": 0.89664, "loss_dense_depth": 0.87171, "loss": 30.73514, "grad_norm": 50.48566, "time": 1.56317} -{"mode": "train", "epoch": 1, "iter": 71, "lr": 0.00013, "memory": 49164, "data_time": 0.07124, "loss_cls_0": 0.93251, "loss_box_0": 1.84285, "loss_cns_0": 0.6218, "loss_yns_0": 0.15664, "loss_cls_1": 1.00187, "loss_box_1": 2.06812, "loss_cns_1": 0.62338, "loss_yns_1": 0.15924, "loss_cls_2": 1.0357, "loss_box_2": 2.01381, "loss_cns_2": 0.64444, "loss_yns_2": 0.15835, "loss_cls_3": 1.04864, "loss_box_3": 2.01921, "loss_cns_3": 0.64783, "loss_yns_3": 0.15996, "loss_cls_4": 1.04515, "loss_box_4": 2.01892, "loss_cns_4": 0.64551, "loss_yns_4": 0.16654, "loss_cls_5": 1.04589, "loss_box_5": 1.99974, "loss_cns_5": 0.64417, "loss_yns_5": 0.16044, "loss_cls_dn_0": 0.37496, "loss_box_dn_0": 0.86442, "loss_cls_dn_1": 0.24497, "loss_box_dn_1": 0.88976, "loss_cls_dn_2": 0.29128, "loss_box_dn_2": 0.88106, "loss_cls_dn_3": 0.31073, "loss_box_dn_3": 0.90887, "loss_cls_dn_4": 0.29656, "loss_box_dn_4": 0.92464, "loss_cls_dn_5": 0.32945, "loss_box_dn_5": 0.9459, "loss_dense_depth": 0.85483, "loss": 30.97814, "grad_norm": 62.96682, "time": 1.58712} -{"mode": "train", "epoch": 1, "iter": 72, "lr": 0.00013, "memory": 49164, "data_time": 0.07769, "loss_cls_0": 0.94238, "loss_box_0": 1.83158, "loss_cns_0": 0.62286, "loss_yns_0": 0.15995, "loss_cls_1": 1.04037, "loss_box_1": 1.99976, "loss_cns_1": 0.63025, "loss_yns_1": 0.1594, "loss_cls_2": 1.07721, "loss_box_2": 1.95879, "loss_cns_2": 0.64449, "loss_yns_2": 0.15809, "loss_cls_3": 1.11448, "loss_box_3": 1.97459, "loss_cns_3": 0.64882, "loss_yns_3": 0.15851, "loss_cls_4": 1.12901, "loss_box_4": 1.93482, "loss_cns_4": 0.65213, "loss_yns_4": 0.16208, "loss_cls_5": 1.09254, "loss_box_5": 1.93096, "loss_cns_5": 0.64961, "loss_yns_5": 0.16508, "loss_cls_dn_0": 0.34941, "loss_box_dn_0": 0.8558, "loss_cls_dn_1": 0.24152, "loss_box_dn_1": 0.86625, "loss_cls_dn_2": 0.27891, "loss_box_dn_2": 0.86687, "loss_cls_dn_3": 0.30509, "loss_box_dn_3": 0.89657, "loss_cls_dn_4": 0.27646, "loss_box_dn_4": 0.89636, "loss_cls_dn_5": 0.32746, "loss_box_dn_5": 0.91383, "loss_dense_depth": 0.85742, "loss": 30.76967, "grad_norm": 75.11981, "time": 1.56752} -{"mode": "train", "epoch": 1, "iter": 73, "lr": 0.00013, "memory": 49164, "data_time": 0.07328, "loss_cls_0": 0.99565, "loss_box_0": 1.85355, "loss_cns_0": 0.6181, "loss_yns_0": 0.16065, "loss_cls_1": 1.03018, "loss_box_1": 2.00884, "loss_cns_1": 0.62884, "loss_yns_1": 0.16261, "loss_cls_2": 1.09194, "loss_box_2": 1.97567, "loss_cns_2": 0.64189, "loss_yns_2": 0.16089, "loss_cls_3": 1.10229, "loss_box_3": 1.99572, "loss_cns_3": 0.64674, "loss_yns_3": 0.16267, "loss_cls_4": 1.16047, "loss_box_4": 1.95753, "loss_cns_4": 0.65042, "loss_yns_4": 0.16359, "loss_cls_5": 1.10118, "loss_box_5": 1.97296, "loss_cns_5": 0.64728, "loss_yns_5": 0.17469, "loss_cls_dn_0": 0.33098, "loss_box_dn_0": 0.85224, "loss_cls_dn_1": 0.23289, "loss_box_dn_1": 0.88458, "loss_cls_dn_2": 0.26525, "loss_box_dn_2": 0.88685, "loss_cls_dn_3": 0.28765, "loss_box_dn_3": 0.91206, "loss_cls_dn_4": 0.2621, "loss_box_dn_4": 0.90964, "loss_cls_dn_5": 0.30881, "loss_box_dn_5": 0.93334, "loss_dense_depth": 0.88349, "loss": 31.0142, "grad_norm": 63.27786, "time": 1.57357} -{"mode": "train", "epoch": 1, "iter": 74, "lr": 0.00013, "memory": 49164, "data_time": 0.07827, "loss_cls_0": 0.99557, "loss_box_0": 1.84893, "loss_cns_0": 0.6174, "loss_yns_0": 0.15773, "loss_cls_1": 1.01963, "loss_box_1": 2.04427, "loss_cns_1": 0.63366, "loss_yns_1": 0.16257, "loss_cls_2": 1.0607, "loss_box_2": 1.99446, "loss_cns_2": 0.65048, "loss_yns_2": 0.16327, "loss_cls_3": 1.06515, "loss_box_3": 2.01886, "loss_cns_3": 0.65214, "loss_yns_3": 0.16612, "loss_cls_4": 1.05985, "loss_box_4": 2.03309, "loss_cns_4": 0.65331, "loss_yns_4": 0.16597, "loss_cls_5": 1.05662, "loss_box_5": 2.0423, "loss_cns_5": 0.65021, "loss_yns_5": 0.17256, "loss_cls_dn_0": 0.3318, "loss_box_dn_0": 0.84615, "loss_cls_dn_1": 0.2267, "loss_box_dn_1": 0.93101, "loss_cls_dn_2": 0.2622, "loss_box_dn_2": 0.91585, "loss_cls_dn_3": 0.28917, "loss_box_dn_3": 0.93395, "loss_cls_dn_4": 0.28467, "loss_box_dn_4": 0.94727, "loss_cls_dn_5": 0.32176, "loss_box_dn_5": 0.96608, "loss_dense_depth": 0.84908, "loss": 31.19051, "grad_norm": 53.74409, "time": 1.5742} -{"mode": "train", "epoch": 1, "iter": 75, "lr": 0.00013, "memory": 49164, "data_time": 0.07363, "loss_cls_0": 0.95507, "loss_box_0": 1.87103, "loss_cns_0": 0.61356, "loss_yns_0": 0.16013, "loss_cls_1": 1.04267, "loss_box_1": 2.13051, "loss_cns_1": 0.62944, "loss_yns_1": 0.16664, "loss_cls_2": 1.0749, "loss_box_2": 2.07965, "loss_cns_2": 0.6483, "loss_yns_2": 0.16473, "loss_cls_3": 1.07852, "loss_box_3": 2.09264, "loss_cns_3": 0.65126, "loss_yns_3": 0.16898, "loss_cls_4": 1.10235, "loss_box_4": 2.09665, "loss_cns_4": 0.64726, "loss_yns_4": 0.16888, "loss_cls_5": 1.07448, "loss_box_5": 2.11192, "loss_cns_5": 0.64747, "loss_yns_5": 0.16431, "loss_cls_dn_0": 0.34451, "loss_box_dn_0": 0.83426, "loss_cls_dn_1": 0.22931, "loss_box_dn_1": 0.94357, "loss_cls_dn_2": 0.27071, "loss_box_dn_2": 0.91801, "loss_cls_dn_3": 0.30506, "loss_box_dn_3": 0.92657, "loss_cls_dn_4": 0.3253, "loss_box_dn_4": 0.93701, "loss_cls_dn_5": 0.33615, "loss_box_dn_5": 0.95163, "loss_dense_depth": 0.84937, "loss": 31.71281, "grad_norm": 65.39788, "time": 1.58268} -{"mode": "train", "epoch": 1, "iter": 76, "lr": 0.00013, "memory": 49164, "data_time": 0.07645, "loss_cls_0": 0.94752, "loss_box_0": 1.83344, "loss_cns_0": 0.61432, "loss_yns_0": 0.16278, "loss_cls_1": 1.0508, "loss_box_1": 2.16366, "loss_cns_1": 0.63016, "loss_yns_1": 0.16476, "loss_cls_2": 1.08715, "loss_box_2": 2.09731, "loss_cns_2": 0.64871, "loss_yns_2": 0.16468, "loss_cls_3": 1.09122, "loss_box_3": 2.0824, "loss_cns_3": 0.65602, "loss_yns_3": 0.16585, "loss_cls_4": 1.11742, "loss_box_4": 2.04299, "loss_cns_4": 0.65017, "loss_yns_4": 0.16785, "loss_cls_5": 1.07152, "loss_box_5": 2.08356, "loss_cns_5": 0.65299, "loss_yns_5": 0.16444, "loss_cls_dn_0": 0.35864, "loss_box_dn_0": 0.83548, "loss_cls_dn_1": 0.23758, "loss_box_dn_1": 0.87685, "loss_cls_dn_2": 0.27615, "loss_box_dn_2": 0.84208, "loss_cls_dn_3": 0.30476, "loss_box_dn_3": 0.83372, "loss_cls_dn_4": 0.32396, "loss_box_dn_4": 0.83371, "loss_cls_dn_5": 0.31988, "loss_box_dn_5": 0.84996, "loss_dense_depth": 0.84599, "loss": 31.25051, "grad_norm": 49.28823, "time": 1.57172} -{"mode": "train", "epoch": 1, "iter": 77, "lr": 0.00013, "memory": 49164, "data_time": 0.07801, "loss_cls_0": 0.94159, "loss_box_0": 1.85124, "loss_cns_0": 0.61442, "loss_yns_0": 0.16171, "loss_cls_1": 1.03367, "loss_box_1": 2.09963, "loss_cns_1": 0.63232, "loss_yns_1": 0.16428, "loss_cls_2": 1.06407, "loss_box_2": 2.06746, "loss_cns_2": 0.65099, "loss_yns_2": 0.16223, "loss_cls_3": 1.0817, "loss_box_3": 2.03207, "loss_cns_3": 0.65869, "loss_yns_3": 0.16058, "loss_cls_4": 1.0651, "loss_box_4": 2.04531, "loss_cns_4": 0.65568, "loss_yns_4": 0.16367, "loss_cls_5": 1.07527, "loss_box_5": 2.05101, "loss_cns_5": 0.65095, "loss_yns_5": 0.1739, "loss_cls_dn_0": 0.35756, "loss_box_dn_0": 0.82567, "loss_cls_dn_1": 0.23047, "loss_box_dn_1": 0.80004, "loss_cls_dn_2": 0.26567, "loss_box_dn_2": 0.78492, "loss_cls_dn_3": 0.2858, "loss_box_dn_3": 0.77314, "loss_cls_dn_4": 0.28564, "loss_box_dn_4": 0.79035, "loss_cls_dn_5": 0.29151, "loss_box_dn_5": 0.8091, "loss_dense_depth": 0.86839, "loss": 30.62579, "grad_norm": 42.33875, "time": 1.57285} -{"mode": "train", "epoch": 1, "iter": 78, "lr": 0.00013, "memory": 49164, "data_time": 0.07126, "loss_cls_0": 0.95901, "loss_box_0": 1.85769, "loss_cns_0": 0.6167, "loss_yns_0": 0.16117, "loss_cls_1": 1.01457, "loss_box_1": 2.11552, "loss_cns_1": 0.62744, "loss_yns_1": 0.16473, "loss_cls_2": 1.06981, "loss_box_2": 2.08824, "loss_cns_2": 0.64689, "loss_yns_2": 0.1654, "loss_cls_3": 1.0876, "loss_box_3": 2.08525, "loss_cns_3": 0.65316, "loss_yns_3": 0.16721, "loss_cls_4": 1.15087, "loss_box_4": 2.10667, "loss_cns_4": 0.65218, "loss_yns_4": 0.16462, "loss_cls_5": 1.14662, "loss_box_5": 2.10729, "loss_cns_5": 0.64716, "loss_yns_5": 0.17687, "loss_cls_dn_0": 0.35405, "loss_box_dn_0": 0.82523, "loss_cls_dn_1": 0.21063, "loss_box_dn_1": 0.82138, "loss_cls_dn_2": 0.24927, "loss_box_dn_2": 0.81344, "loss_cls_dn_3": 0.26572, "loss_box_dn_3": 0.81936, "loss_cls_dn_4": 0.2527, "loss_box_dn_4": 0.85219, "loss_cls_dn_5": 0.2783, "loss_box_dn_5": 0.87383, "loss_dense_depth": 0.8314, "loss": 31.08016, "grad_norm": 66.93925, "time": 1.58104} -{"mode": "train", "epoch": 1, "iter": 79, "lr": 0.00013, "memory": 49164, "data_time": 0.07219, "loss_cls_0": 0.94228, "loss_box_0": 1.84805, "loss_cns_0": 0.62331, "loss_yns_0": 0.16079, "loss_cls_1": 1.03309, "loss_box_1": 2.06321, "loss_cns_1": 0.63416, "loss_yns_1": 0.1612, "loss_cls_2": 1.07385, "loss_box_2": 2.05176, "loss_cns_2": 0.64651, "loss_yns_2": 0.16702, "loss_cls_3": 1.10381, "loss_box_3": 2.07437, "loss_cns_3": 0.65008, "loss_yns_3": 0.16913, "loss_cls_4": 1.13662, "loss_box_4": 2.08435, "loss_cns_4": 0.65055, "loss_yns_4": 0.16404, "loss_cls_5": 1.08263, "loss_box_5": 2.09048, "loss_cns_5": 0.64777, "loss_yns_5": 0.16852, "loss_cls_dn_0": 0.33212, "loss_box_dn_0": 0.82757, "loss_cls_dn_1": 0.21418, "loss_box_dn_1": 0.86666, "loss_cls_dn_2": 0.25027, "loss_box_dn_2": 0.86485, "loss_cls_dn_3": 0.26849, "loss_box_dn_3": 0.88617, "loss_cls_dn_4": 0.25142, "loss_box_dn_4": 0.91958, "loss_cls_dn_5": 0.28965, "loss_box_dn_5": 0.94727, "loss_dense_depth": 0.83974, "loss": 31.18558, "grad_norm": 69.49559, "time": 1.58044} -{"mode": "train", "epoch": 1, "iter": 80, "lr": 0.00013, "memory": 49164, "data_time": 0.07301, "loss_cls_0": 0.94927, "loss_box_0": 1.8497, "loss_cns_0": 0.62371, "loss_yns_0": 0.16044, "loss_cls_1": 1.02352, "loss_box_1": 2.09979, "loss_cns_1": 0.63087, "loss_yns_1": 0.16154, "loss_cls_2": 1.03883, "loss_box_2": 2.07068, "loss_cns_2": 0.63973, "loss_yns_2": 0.1637, "loss_cls_3": 1.06499, "loss_box_3": 2.08061, "loss_cns_3": 0.64833, "loss_yns_3": 0.1643, "loss_cls_4": 1.04633, "loss_box_4": 2.0723, "loss_cns_4": 0.64697, "loss_yns_4": 0.16443, "loss_cls_5": 1.03704, "loss_box_5": 2.0877, "loss_cns_5": 0.64595, "loss_yns_5": 0.16443, "loss_cls_dn_0": 0.32577, "loss_box_dn_0": 0.82479, "loss_cls_dn_1": 0.22136, "loss_box_dn_1": 0.89634, "loss_cls_dn_2": 0.24149, "loss_box_dn_2": 0.88607, "loss_cls_dn_3": 0.26883, "loss_box_dn_3": 0.90199, "loss_cls_dn_4": 0.26658, "loss_box_dn_4": 0.92432, "loss_cls_dn_5": 0.30291, "loss_box_dn_5": 0.95103, "loss_dense_depth": 0.80846, "loss": 31.05512, "grad_norm": 50.8392, "time": 1.57115} -{"mode": "train", "epoch": 1, "iter": 81, "lr": 0.00013, "memory": 49164, "data_time": 0.10678, "loss_cls_0": 0.96106, "loss_box_0": 1.8854, "loss_cns_0": 0.61968, "loss_yns_0": 0.16098, "loss_cls_1": 1.00863, "loss_box_1": 2.05754, "loss_cns_1": 0.62708, "loss_yns_1": 0.16111, "loss_cls_2": 1.07533, "loss_box_2": 2.00946, "loss_cns_2": 0.64349, "loss_yns_2": 0.16418, "loss_cls_3": 1.04552, "loss_box_3": 2.0057, "loss_cns_3": 0.65501, "loss_yns_3": 0.1629, "loss_cls_4": 1.03784, "loss_box_4": 2.01255, "loss_cns_4": 0.65169, "loss_yns_4": 0.16466, "loss_cls_5": 1.06543, "loss_box_5": 2.04074, "loss_cns_5": 0.6449, "loss_yns_5": 0.17024, "loss_cls_dn_0": 0.33128, "loss_box_dn_0": 0.82888, "loss_cls_dn_1": 0.21481, "loss_box_dn_1": 0.92398, "loss_cls_dn_2": 0.23941, "loss_box_dn_2": 0.90681, "loss_cls_dn_3": 0.26095, "loss_box_dn_3": 0.91167, "loss_cls_dn_4": 0.26951, "loss_box_dn_4": 0.92992, "loss_cls_dn_5": 0.2928, "loss_box_dn_5": 0.94984, "loss_dense_depth": 0.84598, "loss": 30.93697, "grad_norm": 42.23786, "time": 1.67827} -{"mode": "train", "epoch": 1, "iter": 82, "lr": 0.00013, "memory": 49164, "data_time": 0.1111, "loss_cls_0": 0.9546, "loss_box_0": 1.87241, "loss_cns_0": 0.6211, "loss_yns_0": 0.15647, "loss_cls_1": 1.0134, "loss_box_1": 2.07415, "loss_cns_1": 0.625, "loss_yns_1": 0.15925, "loss_cls_2": 1.05556, "loss_box_2": 2.03955, "loss_cns_2": 0.64446, "loss_yns_2": 0.16728, "loss_cls_3": 1.08611, "loss_box_3": 2.05358, "loss_cns_3": 0.65198, "loss_yns_3": 0.16477, "loss_cls_4": 1.07228, "loss_box_4": 2.0822, "loss_cns_4": 0.65257, "loss_yns_4": 0.16304, "loss_cls_5": 1.16092, "loss_box_5": 2.03965, "loss_cns_5": 0.64743, "loss_yns_5": 0.17137, "loss_cls_dn_0": 0.34637, "loss_box_dn_0": 0.83418, "loss_cls_dn_1": 0.20447, "loss_box_dn_1": 0.90763, "loss_cls_dn_2": 0.23609, "loss_box_dn_2": 0.89491, "loss_cls_dn_3": 0.25361, "loss_box_dn_3": 0.9067, "loss_cls_dn_4": 0.2611, "loss_box_dn_4": 0.92589, "loss_cls_dn_5": 0.27602, "loss_box_dn_5": 0.91876, "loss_dense_depth": 0.85903, "loss": 31.15388, "grad_norm": 56.35323, "time": 1.63072} -{"mode": "train", "epoch": 1, "iter": 83, "lr": 0.00013, "memory": 49164, "data_time": 0.08523, "loss_cls_0": 0.94638, "loss_box_0": 1.82654, "loss_cns_0": 0.6254, "loss_yns_0": 0.15909, "loss_cls_1": 1.01197, "loss_box_1": 1.99531, "loss_cns_1": 0.62865, "loss_yns_1": 0.16119, "loss_cls_2": 1.06458, "loss_box_2": 1.94824, "loss_cns_2": 0.64385, "loss_yns_2": 0.16626, "loss_cls_3": 1.07989, "loss_box_3": 1.94455, "loss_cns_3": 0.64929, "loss_yns_3": 0.16561, "loss_cls_4": 1.06426, "loss_box_4": 1.96907, "loss_cns_4": 0.65114, "loss_yns_4": 0.16314, "loss_cls_5": 1.08182, "loss_box_5": 1.94709, "loss_cns_5": 0.6496, "loss_yns_5": 0.16602, "loss_cls_dn_0": 0.35071, "loss_box_dn_0": 0.83327, "loss_cls_dn_1": 0.20913, "loss_box_dn_1": 0.85595, "loss_cls_dn_2": 0.23854, "loss_box_dn_2": 0.83538, "loss_cls_dn_3": 0.25653, "loss_box_dn_3": 0.84339, "loss_cls_dn_4": 0.26389, "loss_box_dn_4": 0.86053, "loss_cls_dn_5": 0.28467, "loss_box_dn_5": 0.86627, "loss_dense_depth": 0.82716, "loss": 30.23434, "grad_norm": 49.50338, "time": 1.59331} -{"mode": "train", "epoch": 1, "iter": 84, "lr": 0.00013, "memory": 49164, "data_time": 0.07966, "loss_cls_0": 0.90214, "loss_box_0": 1.81619, "loss_cns_0": 0.62566, "loss_yns_0": 0.15683, "loss_cls_1": 0.97917, "loss_box_1": 1.98459, "loss_cns_1": 0.63655, "loss_yns_1": 0.15863, "loss_cls_2": 1.07509, "loss_box_2": 1.92938, "loss_cns_2": 0.64969, "loss_yns_2": 0.16196, "loss_cls_3": 1.04281, "loss_box_3": 1.94247, "loss_cns_3": 0.65188, "loss_yns_3": 0.16564, "loss_cls_4": 1.06562, "loss_box_4": 1.92102, "loss_cns_4": 0.65293, "loss_yns_4": 0.15883, "loss_cls_5": 1.06596, "loss_box_5": 1.95112, "loss_cns_5": 0.65116, "loss_yns_5": 0.15715, "loss_cls_dn_0": 0.32252, "loss_box_dn_0": 0.82543, "loss_cls_dn_1": 0.21337, "loss_box_dn_1": 0.79343, "loss_cls_dn_2": 0.24593, "loss_box_dn_2": 0.77811, "loss_cls_dn_3": 0.26234, "loss_box_dn_3": 0.78158, "loss_cls_dn_4": 0.27846, "loss_box_dn_4": 0.79227, "loss_cls_dn_5": 0.29177, "loss_box_dn_5": 0.82579, "loss_dense_depth": 0.81728, "loss": 29.73072, "grad_norm": 35.95794, "time": 1.56749} -{"mode": "train", "epoch": 1, "iter": 85, "lr": 0.00013, "memory": 49164, "data_time": 0.08893, "loss_cls_0": 0.96835, "loss_box_0": 1.84429, "loss_cns_0": 0.62348, "loss_yns_0": 0.15775, "loss_cls_1": 1.01332, "loss_box_1": 1.97535, "loss_cns_1": 0.63395, "loss_yns_1": 0.16262, "loss_cls_2": 1.07152, "loss_box_2": 1.93198, "loss_cns_2": 0.6484, "loss_yns_2": 0.16015, "loss_cls_3": 1.07077, "loss_box_3": 1.96021, "loss_cns_3": 0.64877, "loss_yns_3": 0.1608, "loss_cls_4": 1.06072, "loss_box_4": 1.95235, "loss_cns_4": 0.65031, "loss_yns_4": 0.15979, "loss_cls_5": 1.07113, "loss_box_5": 1.93651, "loss_cns_5": 0.64927, "loss_yns_5": 0.16032, "loss_cls_dn_0": 0.31502, "loss_box_dn_0": 0.84751, "loss_cls_dn_1": 0.21031, "loss_box_dn_1": 0.82439, "loss_cls_dn_2": 0.2447, "loss_box_dn_2": 0.81928, "loss_cls_dn_3": 0.25569, "loss_box_dn_3": 0.83084, "loss_cls_dn_4": 0.25664, "loss_box_dn_4": 0.84482, "loss_cls_dn_5": 0.26485, "loss_box_dn_5": 0.85366, "loss_dense_depth": 0.8396, "loss": 30.07941, "grad_norm": 46.21389, "time": 1.59984} -{"mode": "train", "epoch": 1, "iter": 86, "lr": 0.00013, "memory": 49164, "data_time": 0.10855, "loss_cls_0": 0.98474, "loss_box_0": 1.82917, "loss_cns_0": 0.62325, "loss_yns_0": 0.15823, "loss_cls_1": 1.03546, "loss_box_1": 1.91955, "loss_cns_1": 0.63469, "loss_yns_1": 0.16291, "loss_cls_2": 1.04077, "loss_box_2": 1.87226, "loss_cns_2": 0.64712, "loss_yns_2": 0.16801, "loss_cls_3": 1.07842, "loss_box_3": 1.88117, "loss_cns_3": 0.65032, "loss_yns_3": 0.16345, "loss_cls_4": 1.08834, "loss_box_4": 1.88074, "loss_cns_4": 0.65292, "loss_yns_4": 0.16143, "loss_cls_5": 1.19197, "loss_box_5": 1.87093, "loss_cns_5": 0.64993, "loss_yns_5": 0.16909, "loss_cls_dn_0": 0.29426, "loss_box_dn_0": 0.8352, "loss_cls_dn_1": 0.2031, "loss_box_dn_1": 0.82114, "loss_cls_dn_2": 0.23074, "loss_box_dn_2": 0.81573, "loss_cls_dn_3": 0.23925, "loss_box_dn_3": 0.82683, "loss_cls_dn_4": 0.23209, "loss_box_dn_4": 0.84132, "loss_cls_dn_5": 0.24775, "loss_box_dn_5": 0.85052, "loss_dense_depth": 0.8207, "loss": 29.77346, "grad_norm": 63.81302, "time": 1.59113} -{"mode": "train", "epoch": 1, "iter": 87, "lr": 0.00013, "memory": 49164, "data_time": 0.08027, "loss_cls_0": 0.92356, "loss_box_0": 1.82175, "loss_cns_0": 0.62163, "loss_yns_0": 0.15865, "loss_cls_1": 0.98872, "loss_box_1": 1.92563, "loss_cns_1": 0.64335, "loss_yns_1": 0.15963, "loss_cls_2": 1.03242, "loss_box_2": 1.89883, "loss_cns_2": 0.65242, "loss_yns_2": 0.16748, "loss_cls_3": 1.03954, "loss_box_3": 1.90303, "loss_cns_3": 0.65562, "loss_yns_3": 0.16442, "loss_cls_4": 1.06759, "loss_box_4": 1.87334, "loss_cns_4": 0.65755, "loss_yns_4": 0.16122, "loss_cls_5": 1.04107, "loss_box_5": 1.90266, "loss_cns_5": 0.65346, "loss_yns_5": 0.17334, "loss_cls_dn_0": 0.30328, "loss_box_dn_0": 0.8237, "loss_cls_dn_1": 0.20327, "loss_box_dn_1": 0.82151, "loss_cls_dn_2": 0.22253, "loss_box_dn_2": 0.81109, "loss_cls_dn_3": 0.23894, "loss_box_dn_3": 0.82644, "loss_cls_dn_4": 0.22815, "loss_box_dn_4": 0.82466, "loss_cls_dn_5": 0.25017, "loss_box_dn_5": 0.85385, "loss_dense_depth": 0.82825, "loss": 29.5227, "grad_norm": 45.10469, "time": 1.56382} -{"mode": "train", "epoch": 1, "iter": 88, "lr": 0.00013, "memory": 49164, "data_time": 0.08212, "loss_cls_0": 0.92101, "loss_box_0": 1.78299, "loss_cns_0": 0.61007, "loss_yns_0": 0.15318, "loss_cls_1": 1.02095, "loss_box_1": 1.8988, "loss_cns_1": 0.64047, "loss_yns_1": 0.15901, "loss_cls_2": 1.0297, "loss_box_2": 1.86429, "loss_cns_2": 0.65165, "loss_yns_2": 0.16107, "loss_cls_3": 1.11259, "loss_box_3": 1.83712, "loss_cns_3": 0.64739, "loss_yns_3": 0.16134, "loss_cls_4": 1.03655, "loss_box_4": 1.86504, "loss_cns_4": 0.65641, "loss_yns_4": 0.15997, "loss_cls_5": 1.08272, "loss_box_5": 1.85355, "loss_cns_5": 0.65348, "loss_yns_5": 0.16843, "loss_cls_dn_0": 0.32715, "loss_box_dn_0": 0.82562, "loss_cls_dn_1": 0.20679, "loss_box_dn_1": 0.82294, "loss_cls_dn_2": 0.22179, "loss_box_dn_2": 0.8118, "loss_cls_dn_3": 0.25641, "loss_box_dn_3": 0.81989, "loss_cls_dn_4": 0.24418, "loss_box_dn_4": 0.81817, "loss_cls_dn_5": 0.2827, "loss_box_dn_5": 0.83251, "loss_dense_depth": 0.81713, "loss": 29.41489, "grad_norm": 40.12243, "time": 1.55807} -{"mode": "train", "epoch": 1, "iter": 89, "lr": 0.00014, "memory": 49164, "data_time": 0.08842, "loss_cls_0": 0.91613, "loss_box_0": 1.78231, "loss_cns_0": 0.61487, "loss_yns_0": 0.15309, "loss_cls_1": 1.00967, "loss_box_1": 1.94157, "loss_cns_1": 0.62846, "loss_yns_1": 0.1588, "loss_cls_2": 1.05694, "loss_box_2": 1.8735, "loss_cns_2": 0.64815, "loss_yns_2": 0.16, "loss_cls_3": 1.04028, "loss_box_3": 1.85141, "loss_cns_3": 0.65145, "loss_yns_3": 0.1622, "loss_cls_4": 1.03406, "loss_box_4": 1.88209, "loss_cns_4": 0.65424, "loss_yns_4": 0.16041, "loss_cls_5": 1.05226, "loss_box_5": 1.85752, "loss_cns_5": 0.65469, "loss_yns_5": 0.15943, "loss_cls_dn_0": 0.31467, "loss_box_dn_0": 0.81507, "loss_cls_dn_1": 0.20548, "loss_box_dn_1": 0.80445, "loss_cls_dn_2": 0.22469, "loss_box_dn_2": 0.78767, "loss_cls_dn_3": 0.24022, "loss_box_dn_3": 0.7791, "loss_cls_dn_4": 0.24249, "loss_box_dn_4": 0.7928, "loss_cls_dn_5": 0.27343, "loss_box_dn_5": 0.79568, "loss_dense_depth": 0.80438, "loss": 29.18367, "grad_norm": 42.85815, "time": 1.60155} -{"mode": "train", "epoch": 1, "iter": 90, "lr": 0.00014, "memory": 49164, "data_time": 0.0851, "loss_cls_0": 0.94652, "loss_box_0": 1.81136, "loss_cns_0": 0.62171, "loss_yns_0": 0.15539, "loss_cls_1": 0.99918, "loss_box_1": 1.99619, "loss_cns_1": 0.63504, "loss_yns_1": 0.16047, "loss_cls_2": 1.10178, "loss_box_2": 1.91926, "loss_cns_2": 0.65014, "loss_yns_2": 0.16014, "loss_cls_3": 1.16107, "loss_box_3": 1.89354, "loss_cns_3": 0.65565, "loss_yns_3": 0.16088, "loss_cls_4": 1.08154, "loss_box_4": 1.92388, "loss_cns_4": 0.65455, "loss_yns_4": 0.16102, "loss_cls_5": 1.07776, "loss_box_5": 1.90652, "loss_cns_5": 0.65252, "loss_yns_5": 0.16039, "loss_cls_dn_0": 0.30163, "loss_box_dn_0": 0.82556, "loss_cls_dn_1": 0.20324, "loss_box_dn_1": 0.81563, "loss_cls_dn_2": 0.22328, "loss_box_dn_2": 0.79718, "loss_cls_dn_3": 0.22679, "loss_box_dn_3": 0.78986, "loss_cls_dn_4": 0.22659, "loss_box_dn_4": 0.81384, "loss_cls_dn_5": 0.25028, "loss_box_dn_5": 0.81461, "loss_dense_depth": 0.86617, "loss": 29.8012, "grad_norm": 68.08707, "time": 1.56866} -{"mode": "train", "epoch": 1, "iter": 91, "lr": 0.00014, "memory": 49164, "data_time": 0.08739, "loss_cls_0": 0.94975, "loss_box_0": 1.81015, "loss_cns_0": 0.62212, "loss_yns_0": 0.1537, "loss_cls_1": 0.98779, "loss_box_1": 1.98205, "loss_cns_1": 0.63652, "loss_yns_1": 0.15729, "loss_cls_2": 1.02501, "loss_box_2": 1.9194, "loss_cns_2": 0.64896, "loss_yns_2": 0.16325, "loss_cls_3": 1.07652, "loss_box_3": 1.90673, "loss_cns_3": 0.65586, "loss_yns_3": 0.16045, "loss_cls_4": 1.03893, "loss_box_4": 1.92344, "loss_cns_4": 0.65325, "loss_yns_4": 0.16313, "loss_cls_5": 1.08594, "loss_box_5": 1.93309, "loss_cns_5": 0.65039, "loss_yns_5": 0.16693, "loss_cls_dn_0": 0.2869, "loss_box_dn_0": 0.8103, "loss_cls_dn_1": 0.20208, "loss_box_dn_1": 0.81725, "loss_cls_dn_2": 0.21836, "loss_box_dn_2": 0.79546, "loss_cls_dn_3": 0.22383, "loss_box_dn_3": 0.79785, "loss_cls_dn_4": 0.22292, "loss_box_dn_4": 0.82092, "loss_cls_dn_5": 0.24448, "loss_box_dn_5": 0.83599, "loss_dense_depth": 0.81515, "loss": 29.56211, "grad_norm": 52.44379, "time": 1.61561} -{"mode": "train", "epoch": 1, "iter": 92, "lr": 0.00014, "memory": 49164, "data_time": 0.07524, "loss_cls_0": 0.91394, "loss_box_0": 1.8157, "loss_cns_0": 0.61917, "loss_yns_0": 0.15135, "loss_cls_1": 0.97202, "loss_box_1": 2.02817, "loss_cns_1": 0.62999, "loss_yns_1": 0.15722, "loss_cls_2": 1.05142, "loss_box_2": 1.98004, "loss_cns_2": 0.64427, "loss_yns_2": 0.16371, "loss_cls_3": 1.04783, "loss_box_3": 1.96541, "loss_cns_3": 0.65041, "loss_yns_3": 0.15937, "loss_cls_4": 1.0642, "loss_box_4": 1.97998, "loss_cns_4": 0.64931, "loss_yns_4": 0.161, "loss_cls_5": 1.03212, "loss_box_5": 1.99081, "loss_cns_5": 0.64763, "loss_yns_5": 0.1666, "loss_cls_dn_0": 0.29364, "loss_box_dn_0": 0.81082, "loss_cls_dn_1": 0.20052, "loss_box_dn_1": 0.81201, "loss_cls_dn_2": 0.22162, "loss_box_dn_2": 0.79611, "loss_cls_dn_3": 0.23613, "loss_box_dn_3": 0.79784, "loss_cls_dn_4": 0.24229, "loss_box_dn_4": 0.82071, "loss_cls_dn_5": 0.25469, "loss_box_dn_5": 0.84512, "loss_dense_depth": 0.86784, "loss": 29.84102, "grad_norm": 64.44172, "time": 1.57907} -{"mode": "train", "epoch": 1, "iter": 93, "lr": 0.00014, "memory": 49164, "data_time": 0.08188, "loss_cls_0": 0.91489, "loss_box_0": 1.80555, "loss_cns_0": 0.61985, "loss_yns_0": 0.14931, "loss_cls_1": 0.98315, "loss_box_1": 2.02573, "loss_cns_1": 0.62979, "loss_yns_1": 0.15846, "loss_cls_2": 1.11509, "loss_box_2": 1.96047, "loss_cns_2": 0.64257, "loss_yns_2": 0.15985, "loss_cls_3": 1.07977, "loss_box_3": 1.93722, "loss_cns_3": 0.64991, "loss_yns_3": 0.16156, "loss_cls_4": 1.11865, "loss_box_4": 1.9309, "loss_cns_4": 0.65207, "loss_yns_4": 0.15843, "loss_cls_5": 1.03824, "loss_box_5": 1.92904, "loss_cns_5": 0.65205, "loss_yns_5": 0.16711, "loss_cls_dn_0": 0.29394, "loss_box_dn_0": 0.81378, "loss_cls_dn_1": 0.20186, "loss_box_dn_1": 0.82739, "loss_cls_dn_2": 0.23059, "loss_box_dn_2": 0.81401, "loss_cls_dn_3": 0.25076, "loss_box_dn_3": 0.80664, "loss_cls_dn_4": 0.26836, "loss_box_dn_4": 0.82932, "loss_cls_dn_5": 0.26843, "loss_box_dn_5": 0.84612, "loss_dense_depth": 0.85391, "loss": 29.94478, "grad_norm": 55.27099, "time": 1.58883} -{"mode": "train", "epoch": 1, "iter": 94, "lr": 0.00014, "memory": 49164, "data_time": 0.08249, "loss_cls_0": 0.92324, "loss_box_0": 1.81304, "loss_cns_0": 0.62244, "loss_yns_0": 0.15159, "loss_cls_1": 0.97232, "loss_box_1": 2.03075, "loss_cns_1": 0.62865, "loss_yns_1": 0.15688, "loss_cls_2": 1.12296, "loss_box_2": 1.9621, "loss_cns_2": 0.63968, "loss_yns_2": 0.15744, "loss_cls_3": 1.05812, "loss_box_3": 1.96032, "loss_cns_3": 0.6515, "loss_yns_3": 0.16556, "loss_cls_4": 1.08656, "loss_box_4": 1.9465, "loss_cns_4": 0.65377, "loss_yns_4": 0.1564, "loss_cls_5": 1.03986, "loss_box_5": 1.93599, "loss_cns_5": 0.65713, "loss_yns_5": 0.16361, "loss_cls_dn_0": 0.29027, "loss_box_dn_0": 0.80451, "loss_cls_dn_1": 0.20362, "loss_box_dn_1": 0.82472, "loss_cls_dn_2": 0.23564, "loss_box_dn_2": 0.8055, "loss_cls_dn_3": 0.25372, "loss_box_dn_3": 0.79938, "loss_cls_dn_4": 0.26849, "loss_box_dn_4": 0.81044, "loss_cls_dn_5": 0.27257, "loss_box_dn_5": 0.82657, "loss_dense_depth": 0.81437, "loss": 29.86622, "grad_norm": 46.78736, "time": 1.59233} -{"mode": "train", "epoch": 1, "iter": 95, "lr": 0.00014, "memory": 49164, "data_time": 0.07604, "loss_cls_0": 0.92849, "loss_box_0": 1.81692, "loss_cns_0": 0.62074, "loss_yns_0": 0.15234, "loss_cls_1": 0.97777, "loss_box_1": 2.02341, "loss_cns_1": 0.6267, "loss_yns_1": 0.15455, "loss_cls_2": 1.05603, "loss_box_2": 1.99321, "loss_cns_2": 0.64304, "loss_yns_2": 0.15651, "loss_cls_3": 1.03412, "loss_box_3": 2.00439, "loss_cns_3": 0.64874, "loss_yns_3": 0.16033, "loss_cls_4": 1.03871, "loss_box_4": 1.97258, "loss_cns_4": 0.65378, "loss_yns_4": 0.1564, "loss_cls_5": 1.06899, "loss_box_5": 1.9535, "loss_cns_5": 0.65541, "loss_yns_5": 0.15714, "loss_cls_dn_0": 0.27519, "loss_box_dn_0": 0.80983, "loss_cls_dn_1": 0.19537, "loss_box_dn_1": 0.81858, "loss_cls_dn_2": 0.2196, "loss_box_dn_2": 0.80193, "loss_cls_dn_3": 0.22592, "loss_box_dn_3": 0.80639, "loss_cls_dn_4": 0.23069, "loss_box_dn_4": 0.80832, "loss_cls_dn_5": 0.24197, "loss_box_dn_5": 0.81559, "loss_dense_depth": 0.86519, "loss": 29.76837, "grad_norm": 59.25943, "time": 1.56823} -{"mode": "train", "epoch": 1, "iter": 96, "lr": 0.00014, "memory": 49164, "data_time": 0.07983, "loss_cls_0": 0.95933, "loss_box_0": 1.8388, "loss_cns_0": 0.61929, "loss_yns_0": 0.15556, "loss_cls_1": 1.02276, "loss_box_1": 2.01302, "loss_cns_1": 0.61971, "loss_yns_1": 0.15542, "loss_cls_2": 1.03192, "loss_box_2": 1.97559, "loss_cns_2": 0.63878, "loss_yns_2": 0.15988, "loss_cls_3": 1.10352, "loss_box_3": 1.98911, "loss_cns_3": 0.64435, "loss_yns_3": 0.15743, "loss_cls_4": 1.14044, "loss_box_4": 1.96159, "loss_cns_4": 0.64857, "loss_yns_4": 0.15729, "loss_cls_5": 1.10187, "loss_box_5": 1.91844, "loss_cns_5": 0.64906, "loss_yns_5": 0.15858, "loss_cls_dn_0": 0.27868, "loss_box_dn_0": 0.82226, "loss_cls_dn_1": 0.19459, "loss_box_dn_1": 0.80985, "loss_cls_dn_2": 0.21314, "loss_box_dn_2": 0.78773, "loss_cls_dn_3": 0.21895, "loss_box_dn_3": 0.7918, "loss_cls_dn_4": 0.22323, "loss_box_dn_4": 0.79307, "loss_cls_dn_5": 0.23513, "loss_box_dn_5": 0.79088, "loss_dense_depth": 0.85264, "loss": 29.83224, "grad_norm": 52.23816, "time": 1.56193} -{"mode": "train", "epoch": 1, "iter": 97, "lr": 0.00014, "memory": 49164, "data_time": 0.07745, "loss_cls_0": 0.94036, "loss_box_0": 1.80386, "loss_cns_0": 0.62, "loss_yns_0": 0.15442, "loss_cls_1": 0.98904, "loss_box_1": 1.9491, "loss_cns_1": 0.62754, "loss_yns_1": 0.15769, "loss_cls_2": 1.0523, "loss_box_2": 1.87075, "loss_cns_2": 0.64567, "loss_yns_2": 0.16447, "loss_cls_3": 1.06724, "loss_box_3": 1.87827, "loss_cns_3": 0.65203, "loss_yns_3": 0.16274, "loss_cls_4": 1.09233, "loss_box_4": 1.89545, "loss_cns_4": 0.65088, "loss_yns_4": 0.15757, "loss_cls_5": 1.06365, "loss_box_5": 1.88484, "loss_cns_5": 0.65115, "loss_yns_5": 0.16496, "loss_cls_dn_0": 0.29024, "loss_box_dn_0": 0.81809, "loss_cls_dn_1": 0.19378, "loss_box_dn_1": 0.77816, "loss_cls_dn_2": 0.20882, "loss_box_dn_2": 0.74882, "loss_cls_dn_3": 0.21512, "loss_box_dn_3": 0.75335, "loss_cls_dn_4": 0.22059, "loss_box_dn_4": 0.76808, "loss_cls_dn_5": 0.23418, "loss_box_dn_5": 0.78084, "loss_dense_depth": 0.83934, "loss": 29.14574, "grad_norm": 42.71344, "time": 1.58581} -{"mode": "train", "epoch": 1, "iter": 98, "lr": 0.00014, "memory": 49164, "data_time": 0.07958, "loss_cls_0": 0.93117, "loss_box_0": 1.78518, "loss_cns_0": 0.62186, "loss_yns_0": 0.15243, "loss_cls_1": 0.98113, "loss_box_1": 1.91332, "loss_cns_1": 0.62479, "loss_yns_1": 0.15506, "loss_cls_2": 1.02502, "loss_box_2": 1.84017, "loss_cns_2": 0.65098, "loss_yns_2": 0.15997, "loss_cls_3": 1.04137, "loss_box_3": 1.82335, "loss_cns_3": 0.65828, "loss_yns_3": 0.16768, "loss_cls_4": 1.04422, "loss_box_4": 1.86668, "loss_cns_4": 0.65672, "loss_yns_4": 0.15656, "loss_cls_5": 1.06246, "loss_box_5": 1.86025, "loss_cns_5": 0.65484, "loss_yns_5": 0.16805, "loss_cls_dn_0": 0.28911, "loss_box_dn_0": 0.8165, "loss_cls_dn_1": 0.1961, "loss_box_dn_1": 0.7867, "loss_cls_dn_2": 0.21141, "loss_box_dn_2": 0.75737, "loss_cls_dn_3": 0.2182, "loss_box_dn_3": 0.76635, "loss_cls_dn_4": 0.21523, "loss_box_dn_4": 0.79255, "loss_cls_dn_5": 0.23214, "loss_box_dn_5": 0.81407, "loss_dense_depth": 0.82583, "loss": 28.92309, "grad_norm": 52.84194, "time": 1.56929} -{"mode": "train", "epoch": 1, "iter": 99, "lr": 0.00014, "memory": 49164, "data_time": 0.08289, "loss_cls_0": 0.92225, "loss_box_0": 1.78487, "loss_cns_0": 0.62143, "loss_yns_0": 0.15321, "loss_cls_1": 0.98917, "loss_box_1": 1.897, "loss_cns_1": 0.62421, "loss_yns_1": 0.15471, "loss_cls_2": 1.0288, "loss_box_2": 1.84239, "loss_cns_2": 0.65103, "loss_yns_2": 0.15625, "loss_cls_3": 1.05448, "loss_box_3": 1.82489, "loss_cns_3": 0.65506, "loss_yns_3": 0.16457, "loss_cls_4": 1.05494, "loss_box_4": 1.84604, "loss_cns_4": 0.65434, "loss_yns_4": 0.15829, "loss_cls_5": 1.07309, "loss_box_5": 1.87309, "loss_cns_5": 0.64966, "loss_yns_5": 0.16688, "loss_cls_dn_0": 0.27618, "loss_box_dn_0": 0.80902, "loss_cls_dn_1": 0.19572, "loss_box_dn_1": 0.80963, "loss_cls_dn_2": 0.20916, "loss_box_dn_2": 0.79541, "loss_cls_dn_3": 0.21657, "loss_box_dn_3": 0.81432, "loss_cls_dn_4": 0.21327, "loss_box_dn_4": 0.83223, "loss_cls_dn_5": 0.22963, "loss_box_dn_5": 0.8644, "loss_dense_depth": 0.82771, "loss": 29.09388, "grad_norm": 53.60011, "time": 1.57451} -{"mode": "train", "epoch": 1, "iter": 100, "lr": 0.00014, "memory": 49164, "data_time": 0.0844, "loss_cls_0": 0.96689, "loss_box_0": 1.84061, "loss_cns_0": 0.62342, "loss_yns_0": 0.15558, "loss_cls_1": 0.97956, "loss_box_1": 1.96246, "loss_cns_1": 0.61745, "loss_yns_1": 0.15894, "loss_cls_2": 1.03276, "loss_box_2": 1.91007, "loss_cns_2": 0.64344, "loss_yns_2": 0.15852, "loss_cls_3": 1.05084, "loss_box_3": 1.88625, "loss_cns_3": 0.64714, "loss_yns_3": 0.15762, "loss_cls_4": 1.06722, "loss_box_4": 1.88277, "loss_cns_4": 0.65002, "loss_yns_4": 0.15876, "loss_cls_5": 1.06435, "loss_box_5": 1.90059, "loss_cns_5": 0.645, "loss_yns_5": 0.15869, "loss_cls_dn_0": 0.26018, "loss_box_dn_0": 0.81817, "loss_cls_dn_1": 0.19223, "loss_box_dn_1": 0.81642, "loss_cls_dn_2": 0.20878, "loss_box_dn_2": 0.80002, "loss_cls_dn_3": 0.21173, "loss_box_dn_3": 0.81348, "loss_cls_dn_4": 0.21309, "loss_box_dn_4": 0.82539, "loss_cls_dn_5": 0.22254, "loss_box_dn_5": 0.84666, "loss_dense_depth": 0.82645, "loss": 29.37408, "grad_norm": 37.5787, "time": 1.57497} -{"mode": "train", "epoch": 1, "iter": 101, "lr": 0.00014, "memory": 49164, "data_time": 0.12244, "loss_cls_0": 0.92529, "loss_box_0": 1.80188, "loss_cns_0": 0.61957, "loss_yns_0": 0.15459, "loss_cls_1": 0.97424, "loss_box_1": 1.99214, "loss_cns_1": 0.62905, "loss_yns_1": 0.15916, "loss_cls_2": 1.0295, "loss_box_2": 1.95443, "loss_cns_2": 0.64725, "loss_yns_2": 0.16652, "loss_cls_3": 1.04307, "loss_box_3": 1.94259, "loss_cns_3": 0.65136, "loss_yns_3": 0.16266, "loss_cls_4": 1.05938, "loss_box_4": 1.94056, "loss_cns_4": 0.65376, "loss_yns_4": 0.15983, "loss_cls_5": 1.05064, "loss_box_5": 1.94395, "loss_cns_5": 0.65056, "loss_yns_5": 0.1591, "loss_cls_dn_0": 0.26695, "loss_box_dn_0": 0.80787, "loss_cls_dn_1": 0.1912, "loss_box_dn_1": 0.82643, "loss_cls_dn_2": 0.20825, "loss_box_dn_2": 0.81246, "loss_cls_dn_3": 0.21223, "loss_box_dn_3": 0.82223, "loss_cls_dn_4": 0.21621, "loss_box_dn_4": 0.83077, "loss_cls_dn_5": 0.22802, "loss_box_dn_5": 0.84095, "loss_dense_depth": 0.81733, "loss": 29.55199, "grad_norm": 52.52924, "time": 1.6314} -{"mode": "train", "epoch": 1, "iter": 102, "lr": 0.00014, "memory": 49164, "data_time": 0.11993, "loss_cls_0": 0.94684, "loss_box_0": 1.75652, "loss_cns_0": 0.60377, "loss_yns_0": 0.15282, "loss_cls_1": 0.97781, "loss_box_1": 1.96263, "loss_cns_1": 0.63387, "loss_yns_1": 0.15757, "loss_cls_2": 1.01481, "loss_box_2": 1.92313, "loss_cns_2": 0.64783, "loss_yns_2": 0.16367, "loss_cls_3": 1.03028, "loss_box_3": 1.92009, "loss_cns_3": 0.65334, "loss_yns_3": 0.16598, "loss_cls_4": 1.03964, "loss_box_4": 1.90744, "loss_cns_4": 0.655, "loss_yns_4": 0.15943, "loss_cls_5": 1.04348, "loss_box_5": 1.93687, "loss_cns_5": 0.65208, "loss_yns_5": 0.16486, "loss_cls_dn_0": 0.27359, "loss_box_dn_0": 0.80734, "loss_cls_dn_1": 0.19237, "loss_box_dn_1": 0.79815, "loss_cls_dn_2": 0.20393, "loss_box_dn_2": 0.78803, "loss_cls_dn_3": 0.20809, "loss_box_dn_3": 0.79767, "loss_cls_dn_4": 0.21289, "loss_box_dn_4": 0.79741, "loss_cls_dn_5": 0.22445, "loss_box_dn_5": 0.81643, "loss_dense_depth": 0.84612, "loss": 29.23621, "grad_norm": 50.09231, "time": 1.60988} -{"mode": "train", "epoch": 1, "iter": 103, "lr": 0.00014, "memory": 49164, "data_time": 0.08034, "loss_cls_0": 0.93293, "loss_box_0": 1.74494, "loss_cns_0": 0.60473, "loss_yns_0": 0.15296, "loss_cls_1": 0.99195, "loss_box_1": 1.98237, "loss_cns_1": 0.63357, "loss_yns_1": 0.15875, "loss_cls_2": 1.00761, "loss_box_2": 1.90923, "loss_cns_2": 0.64879, "loss_yns_2": 0.15999, "loss_cls_3": 1.04028, "loss_box_3": 1.89539, "loss_cns_3": 0.65656, "loss_yns_3": 0.16531, "loss_cls_4": 1.03146, "loss_box_4": 1.88158, "loss_cns_4": 0.65493, "loss_yns_4": 0.15907, "loss_cls_5": 1.05034, "loss_box_5": 1.9162, "loss_cns_5": 0.65148, "loss_yns_5": 0.16899, "loss_cls_dn_0": 0.27072, "loss_box_dn_0": 0.80265, "loss_cls_dn_1": 0.18983, "loss_box_dn_1": 0.82101, "loss_cls_dn_2": 0.20008, "loss_box_dn_2": 0.79663, "loss_cls_dn_3": 0.19855, "loss_box_dn_3": 0.79223, "loss_cls_dn_4": 0.20131, "loss_box_dn_4": 0.78944, "loss_cls_dn_5": 0.20979, "loss_box_dn_5": 0.80745, "loss_dense_depth": 0.81699, "loss": 29.09608, "grad_norm": 36.85048, "time": 1.57007} -{"mode": "train", "epoch": 1, "iter": 104, "lr": 0.00014, "memory": 49164, "data_time": 0.07402, "loss_cls_0": 0.92095, "loss_box_0": 1.80529, "loss_cns_0": 0.62161, "loss_yns_0": 0.1558, "loss_cls_1": 0.97634, "loss_box_1": 1.96848, "loss_cns_1": 0.64197, "loss_yns_1": 0.15846, "loss_cls_2": 1.03956, "loss_box_2": 1.91098, "loss_cns_2": 0.65401, "loss_yns_2": 0.15918, "loss_cls_3": 1.02635, "loss_box_3": 1.93042, "loss_cns_3": 0.6602, "loss_yns_3": 0.15768, "loss_cls_4": 1.02423, "loss_box_4": 1.92065, "loss_cns_4": 0.65786, "loss_yns_4": 0.15796, "loss_cls_5": 1.03138, "loss_box_5": 1.91003, "loss_cns_5": 0.65567, "loss_yns_5": 0.16544, "loss_cls_dn_0": 0.25913, "loss_box_dn_0": 0.8151, "loss_cls_dn_1": 0.18588, "loss_box_dn_1": 0.80575, "loss_cls_dn_2": 0.19298, "loss_box_dn_2": 0.78878, "loss_cls_dn_3": 0.19172, "loss_box_dn_3": 0.79235, "loss_cls_dn_4": 0.19553, "loss_box_dn_4": 0.80039, "loss_cls_dn_5": 0.20255, "loss_box_dn_5": 0.79616, "loss_dense_depth": 0.81221, "loss": 29.14906, "grad_norm": 47.82512, "time": 1.577} -{"mode": "train", "epoch": 1, "iter": 105, "lr": 0.00014, "memory": 49164, "data_time": 0.08012, "loss_cls_0": 0.96207, "loss_box_0": 1.78328, "loss_cns_0": 0.62512, "loss_yns_0": 0.15602, "loss_cls_1": 0.96791, "loss_box_1": 1.94643, "loss_cns_1": 0.6421, "loss_yns_1": 0.15745, "loss_cls_2": 1.02259, "loss_box_2": 1.89018, "loss_cns_2": 0.65535, "loss_yns_2": 0.15841, "loss_cls_3": 1.01949, "loss_box_3": 1.91657, "loss_cns_3": 0.65782, "loss_yns_3": 0.15969, "loss_cls_4": 1.02459, "loss_box_4": 1.89693, "loss_cns_4": 0.65926, "loss_yns_4": 0.15698, "loss_cls_5": 1.02627, "loss_box_5": 1.89855, "loss_cns_5": 0.65574, "loss_yns_5": 0.15818, "loss_cls_dn_0": 0.24724, "loss_box_dn_0": 0.81743, "loss_cls_dn_1": 0.18443, "loss_box_dn_1": 0.83109, "loss_cls_dn_2": 0.19147, "loss_box_dn_2": 0.81458, "loss_cls_dn_3": 0.18994, "loss_box_dn_3": 0.82542, "loss_cls_dn_4": 0.19603, "loss_box_dn_4": 0.82893, "loss_cls_dn_5": 0.20635, "loss_box_dn_5": 0.83613, "loss_dense_depth": 0.78855, "loss": 29.15456, "grad_norm": 53.07818, "time": 1.59885} -{"mode": "train", "epoch": 1, "iter": 106, "lr": 0.00014, "memory": 49164, "data_time": 0.10881, "loss_cls_0": 0.88879, "loss_box_0": 1.76396, "loss_cns_0": 0.62164, "loss_yns_0": 0.15437, "loss_cls_1": 0.9766, "loss_box_1": 1.9251, "loss_cns_1": 0.64134, "loss_yns_1": 0.15926, "loss_cls_2": 1.00549, "loss_box_2": 1.87378, "loss_cns_2": 0.6515, "loss_yns_2": 0.15894, "loss_cls_3": 1.02104, "loss_box_3": 1.87475, "loss_cns_3": 0.65736, "loss_yns_3": 0.16137, "loss_cls_4": 1.02744, "loss_box_4": 1.87302, "loss_cns_4": 0.6574, "loss_yns_4": 0.15773, "loss_cls_5": 1.04258, "loss_box_5": 1.8598, "loss_cns_5": 0.6549, "loss_yns_5": 0.15991, "loss_cls_dn_0": 0.25126, "loss_box_dn_0": 0.79392, "loss_cls_dn_1": 0.18244, "loss_box_dn_1": 0.82152, "loss_cls_dn_2": 0.1897, "loss_box_dn_2": 0.80847, "loss_cls_dn_3": 0.19224, "loss_box_dn_3": 0.82418, "loss_cls_dn_4": 0.20036, "loss_box_dn_4": 0.84043, "loss_cls_dn_5": 0.22162, "loss_box_dn_5": 0.86111, "loss_dense_depth": 0.79542, "loss": 28.95073, "grad_norm": 45.32787, "time": 1.61503} -{"mode": "train", "epoch": 1, "iter": 107, "lr": 0.00014, "memory": 49164, "data_time": 0.08748, "loss_cls_0": 0.9105, "loss_box_0": 1.72584, "loss_cns_0": 0.61086, "loss_yns_0": 0.15267, "loss_cls_1": 0.96324, "loss_box_1": 1.90851, "loss_cns_1": 0.63286, "loss_yns_1": 0.15805, "loss_cls_2": 0.99382, "loss_box_2": 1.85519, "loss_cns_2": 0.6455, "loss_yns_2": 0.1576, "loss_cls_3": 1.01522, "loss_box_3": 1.86104, "loss_cns_3": 0.65041, "loss_yns_3": 0.16043, "loss_cls_4": 1.0217, "loss_box_4": 1.86262, "loss_cns_4": 0.64876, "loss_yns_4": 0.1583, "loss_cls_5": 1.04108, "loss_box_5": 1.84583, "loss_cns_5": 0.64803, "loss_yns_5": 0.16438, "loss_cls_dn_0": 0.2592, "loss_box_dn_0": 0.81146, "loss_cls_dn_1": 0.18636, "loss_box_dn_1": 0.81588, "loss_cls_dn_2": 0.19327, "loss_box_dn_2": 0.80351, "loss_cls_dn_3": 0.20367, "loss_box_dn_3": 0.82531, "loss_cls_dn_4": 0.20992, "loss_box_dn_4": 0.84986, "loss_cls_dn_5": 0.24194, "loss_box_dn_5": 0.86681, "loss_dense_depth": 0.79063, "loss": 28.85026, "grad_norm": 48.81184, "time": 1.59551} -{"mode": "train", "epoch": 1, "iter": 108, "lr": 0.00014, "memory": 49164, "data_time": 0.07527, "loss_cls_0": 0.90212, "loss_box_0": 1.75876, "loss_cns_0": 0.61753, "loss_yns_0": 0.15536, "loss_cls_1": 0.95419, "loss_box_1": 1.91749, "loss_cns_1": 0.63065, "loss_yns_1": 0.16017, "loss_cls_2": 0.99777, "loss_box_2": 1.83957, "loss_cns_2": 0.65033, "loss_yns_2": 0.15785, "loss_cls_3": 1.01453, "loss_box_3": 1.85844, "loss_cns_3": 0.65704, "loss_yns_3": 0.16038, "loss_cls_4": 1.02268, "loss_box_4": 1.82608, "loss_cns_4": 0.65701, "loss_yns_4": 0.15983, "loss_cls_5": 1.02275, "loss_box_5": 1.83438, "loss_cns_5": 0.65118, "loss_yns_5": 0.1658, "loss_cls_dn_0": 0.25872, "loss_box_dn_0": 0.80919, "loss_cls_dn_1": 0.1846, "loss_box_dn_1": 0.80652, "loss_cls_dn_2": 0.19151, "loss_box_dn_2": 0.79223, "loss_cls_dn_3": 0.20344, "loss_box_dn_3": 0.81537, "loss_cls_dn_4": 0.20764, "loss_box_dn_4": 0.82309, "loss_cls_dn_5": 0.22966, "loss_box_dn_5": 0.8398, "loss_dense_depth": 0.76903, "loss": 28.70271, "grad_norm": 52.23577, "time": 1.5587} -{"mode": "train", "epoch": 1, "iter": 109, "lr": 0.00014, "memory": 49164, "data_time": 0.08173, "loss_cls_0": 0.91697, "loss_box_0": 1.78735, "loss_cns_0": 0.61579, "loss_yns_0": 0.15641, "loss_cls_1": 0.95882, "loss_box_1": 1.90632, "loss_cns_1": 0.63273, "loss_yns_1": 0.15846, "loss_cls_2": 1.01252, "loss_box_2": 1.86409, "loss_cns_2": 0.64798, "loss_yns_2": 0.15759, "loss_cls_3": 1.00717, "loss_box_3": 1.88594, "loss_cns_3": 0.65184, "loss_yns_3": 0.1602, "loss_cls_4": 1.01318, "loss_box_4": 1.8313, "loss_cns_4": 0.65307, "loss_yns_4": 0.15858, "loss_cls_5": 1.01308, "loss_box_5": 1.83208, "loss_cns_5": 0.65019, "loss_yns_5": 0.15918, "loss_cls_dn_0": 0.25258, "loss_box_dn_0": 0.80904, "loss_cls_dn_1": 0.17862, "loss_box_dn_1": 0.79669, "loss_cls_dn_2": 0.18549, "loss_box_dn_2": 0.78954, "loss_cls_dn_3": 0.19438, "loss_box_dn_3": 0.7983, "loss_cls_dn_4": 0.19862, "loss_box_dn_4": 0.78868, "loss_cls_dn_5": 0.20945, "loss_box_dn_5": 0.79596, "loss_dense_depth": 0.78676, "loss": 28.61496, "grad_norm": 43.48086, "time": 1.59534} -{"mode": "train", "epoch": 1, "iter": 110, "lr": 0.00014, "memory": 49164, "data_time": 0.08267, "loss_cls_0": 0.93628, "loss_box_0": 1.83227, "loss_cns_0": 0.61569, "loss_yns_0": 0.15729, "loss_cls_1": 0.96577, "loss_box_1": 1.93477, "loss_cns_1": 0.62793, "loss_yns_1": 0.15818, "loss_cls_2": 1.03351, "loss_box_2": 1.87208, "loss_cns_2": 0.6428, "loss_yns_2": 0.15756, "loss_cls_3": 1.0411, "loss_box_3": 1.88609, "loss_cns_3": 0.64582, "loss_yns_3": 0.15995, "loss_cls_4": 1.01982, "loss_box_4": 1.8761, "loss_cns_4": 0.64966, "loss_yns_4": 0.1577, "loss_cls_5": 1.04992, "loss_box_5": 1.87539, "loss_cns_5": 0.64651, "loss_yns_5": 0.15732, "loss_cls_dn_0": 0.24985, "loss_box_dn_0": 0.81295, "loss_cls_dn_1": 0.18125, "loss_box_dn_1": 0.76996, "loss_cls_dn_2": 0.18708, "loss_box_dn_2": 0.75367, "loss_cls_dn_3": 0.18926, "loss_box_dn_3": 0.75619, "loss_cls_dn_4": 0.19394, "loss_box_dn_4": 0.75941, "loss_cls_dn_5": 0.20132, "loss_box_dn_5": 0.76945, "loss_dense_depth": 0.79658, "loss": 28.72042, "grad_norm": 41.71151, "time": 1.58363} -{"mode": "train", "epoch": 1, "iter": 111, "lr": 0.00014, "memory": 49164, "data_time": 0.07189, "loss_cls_0": 0.89605, "loss_box_0": 1.80937, "loss_cns_0": 0.61721, "loss_yns_0": 0.15495, "loss_cls_1": 0.9659, "loss_box_1": 1.93532, "loss_cns_1": 0.63059, "loss_yns_1": 0.15751, "loss_cls_2": 1.00451, "loss_box_2": 1.87957, "loss_cns_2": 0.64163, "loss_yns_2": 0.15739, "loss_cls_3": 1.01594, "loss_box_3": 1.89617, "loss_cns_3": 0.64611, "loss_yns_3": 0.15887, "loss_cls_4": 1.01783, "loss_box_4": 1.90457, "loss_cns_4": 0.64823, "loss_yns_4": 0.15713, "loss_cls_5": 1.04346, "loss_box_5": 1.89091, "loss_cns_5": 0.64393, "loss_yns_5": 0.15698, "loss_cls_dn_0": 0.254, "loss_box_dn_0": 0.81037, "loss_cls_dn_1": 0.18154, "loss_box_dn_1": 0.75915, "loss_cls_dn_2": 0.18882, "loss_box_dn_2": 0.74453, "loss_cls_dn_3": 0.18997, "loss_box_dn_3": 0.74939, "loss_cls_dn_4": 0.19284, "loss_box_dn_4": 0.76033, "loss_cls_dn_5": 0.20266, "loss_box_dn_5": 0.76762, "loss_dense_depth": 0.82869, "loss": 28.66002, "grad_norm": 41.59046, "time": 1.57237} -{"mode": "train", "epoch": 1, "iter": 112, "lr": 0.00014, "memory": 49164, "data_time": 0.07783, "loss_cls_0": 0.9407, "loss_box_0": 1.79759, "loss_cns_0": 0.60906, "loss_yns_0": 0.15544, "loss_cls_1": 0.98872, "loss_box_1": 1.98356, "loss_cns_1": 0.62189, "loss_yns_1": 0.15717, "loss_cls_2": 1.06989, "loss_box_2": 1.88779, "loss_cns_2": 0.63665, "loss_yns_2": 0.15704, "loss_cls_3": 1.03019, "loss_box_3": 1.90565, "loss_cns_3": 0.64347, "loss_yns_3": 0.15885, "loss_cls_4": 1.01927, "loss_box_4": 1.90906, "loss_cns_4": 0.64244, "loss_yns_4": 0.15891, "loss_cls_5": 1.02946, "loss_box_5": 1.91126, "loss_cns_5": 0.63808, "loss_yns_5": 0.15805, "loss_cls_dn_0": 0.26886, "loss_box_dn_0": 0.79966, "loss_cls_dn_1": 0.17994, "loss_box_dn_1": 0.7688, "loss_cls_dn_2": 0.19248, "loss_box_dn_2": 0.75135, "loss_cls_dn_3": 0.18977, "loss_box_dn_3": 0.75937, "loss_cls_dn_4": 0.18952, "loss_box_dn_4": 0.77035, "loss_cls_dn_5": 0.20241, "loss_box_dn_5": 0.78007, "loss_dense_depth": 0.85893, "loss": 28.9217, "grad_norm": 41.74226, "time": 1.56346} -{"mode": "train", "epoch": 1, "iter": 113, "lr": 0.00014, "memory": 49164, "data_time": 0.07267, "loss_cls_0": 0.88683, "loss_box_0": 1.77827, "loss_cns_0": 0.61902, "loss_yns_0": 0.15281, "loss_cls_1": 0.92972, "loss_box_1": 1.93816, "loss_cns_1": 0.62529, "loss_yns_1": 0.15354, "loss_cls_2": 1.01748, "loss_box_2": 1.84529, "loss_cns_2": 0.64235, "loss_yns_2": 0.15431, "loss_cls_3": 0.99051, "loss_box_3": 1.84769, "loss_cns_3": 0.64522, "loss_yns_3": 0.15445, "loss_cls_4": 0.9919, "loss_box_4": 1.857, "loss_cns_4": 0.64547, "loss_yns_4": 0.15442, "loss_cls_5": 1.0015, "loss_box_5": 1.86311, "loss_cns_5": 0.64519, "loss_yns_5": 0.15348, "loss_cls_dn_0": 0.25007, "loss_box_dn_0": 0.80122, "loss_cls_dn_1": 0.17407, "loss_box_dn_1": 0.7966, "loss_cls_dn_2": 0.18648, "loss_box_dn_2": 0.774, "loss_cls_dn_3": 0.1851, "loss_box_dn_3": 0.78661, "loss_cls_dn_4": 0.18527, "loss_box_dn_4": 0.79862, "loss_cls_dn_5": 0.20131, "loss_box_dn_5": 0.81543, "loss_dense_depth": 0.80997, "loss": 28.45777, "grad_norm": 40.04265, "time": 1.58512} -{"mode": "train", "epoch": 1, "iter": 114, "lr": 0.00015, "memory": 49164, "data_time": 0.0774, "loss_cls_0": 0.9086, "loss_box_0": 1.78992, "loss_cns_0": 0.62106, "loss_yns_0": 0.15493, "loss_cls_1": 0.97306, "loss_box_1": 1.90218, "loss_cns_1": 0.63631, "loss_yns_1": 0.15332, "loss_cls_2": 0.98561, "loss_box_2": 1.85945, "loss_cns_2": 0.64494, "loss_yns_2": 0.15518, "loss_cls_3": 1.00122, "loss_box_3": 1.86641, "loss_cns_3": 0.64613, "loss_yns_3": 0.15514, "loss_cls_4": 1.00933, "loss_box_4": 1.86486, "loss_cns_4": 0.64947, "loss_yns_4": 0.15451, "loss_cls_5": 1.00191, "loss_box_5": 1.86359, "loss_cns_5": 0.64733, "loss_yns_5": 0.1538, "loss_cls_dn_0": 0.25091, "loss_box_dn_0": 0.79299, "loss_cls_dn_1": 0.17394, "loss_box_dn_1": 0.78367, "loss_cls_dn_2": 0.17993, "loss_box_dn_2": 0.77339, "loss_cls_dn_3": 0.18582, "loss_box_dn_3": 0.78623, "loss_cls_dn_4": 0.1903, "loss_box_dn_4": 0.79183, "loss_cls_dn_5": 0.20272, "loss_box_dn_5": 0.80665, "loss_dense_depth": 0.81675, "loss": 28.53339, "grad_norm": 50.63693, "time": 1.57193} -{"mode": "train", "epoch": 1, "iter": 115, "lr": 0.00015, "memory": 49164, "data_time": 0.07906, "loss_cls_0": 0.8836, "loss_box_0": 1.76346, "loss_cns_0": 0.62283, "loss_yns_0": 0.15273, "loss_cls_1": 0.97369, "loss_box_1": 1.89012, "loss_cns_1": 0.63864, "loss_yns_1": 0.1531, "loss_cls_2": 1.02448, "loss_box_2": 1.848, "loss_cns_2": 0.64886, "loss_yns_2": 0.1519, "loss_cls_3": 1.01732, "loss_box_3": 1.84146, "loss_cns_3": 0.65352, "loss_yns_3": 0.1536, "loss_cls_4": 1.00391, "loss_box_4": 1.85031, "loss_cns_4": 0.65577, "loss_yns_4": 0.15397, "loss_cls_5": 1.01345, "loss_box_5": 1.83322, "loss_cns_5": 0.65284, "loss_yns_5": 0.15326, "loss_cls_dn_0": 0.24224, "loss_box_dn_0": 0.7922, "loss_cls_dn_1": 0.17384, "loss_box_dn_1": 0.7702, "loss_cls_dn_2": 0.18027, "loss_box_dn_2": 0.7567, "loss_cls_dn_3": 0.18369, "loss_box_dn_3": 0.75719, "loss_cls_dn_4": 0.18626, "loss_box_dn_4": 0.76207, "loss_cls_dn_5": 0.19457, "loss_box_dn_5": 0.76869, "loss_dense_depth": 0.77117, "loss": 28.27313, "grad_norm": 38.99418, "time": 1.5841} -{"mode": "train", "epoch": 1, "iter": 116, "lr": 0.00015, "memory": 49164, "data_time": 0.07339, "loss_cls_0": 0.89277, "loss_box_0": 1.7651, "loss_cns_0": 0.61447, "loss_yns_0": 0.15074, "loss_cls_1": 0.97597, "loss_box_1": 1.91195, "loss_cns_1": 0.63755, "loss_yns_1": 0.15349, "loss_cls_2": 0.99529, "loss_box_2": 1.88487, "loss_cns_2": 0.649, "loss_yns_2": 0.15596, "loss_cls_3": 1.03055, "loss_box_3": 1.88704, "loss_cns_3": 0.65121, "loss_yns_3": 0.15568, "loss_cls_4": 1.049, "loss_box_4": 1.90486, "loss_cns_4": 0.64539, "loss_yns_4": 0.15413, "loss_cls_5": 1.03979, "loss_box_5": 1.88609, "loss_cns_5": 0.64623, "loss_yns_5": 0.15531, "loss_cls_dn_0": 0.25067, "loss_box_dn_0": 0.79317, "loss_cls_dn_1": 0.17196, "loss_box_dn_1": 0.74679, "loss_cls_dn_2": 0.17804, "loss_box_dn_2": 0.73348, "loss_cls_dn_3": 0.1858, "loss_box_dn_3": 0.73642, "loss_cls_dn_4": 0.1882, "loss_box_dn_4": 0.75801, "loss_cls_dn_5": 0.19682, "loss_box_dn_5": 0.74989, "loss_dense_depth": 0.83056, "loss": 28.51224, "grad_norm": 50.62425, "time": 1.56631} -{"mode": "train", "epoch": 1, "iter": 117, "lr": 0.00015, "memory": 49164, "data_time": 0.07315, "loss_cls_0": 0.88446, "loss_box_0": 1.71089, "loss_cns_0": 0.61435, "loss_yns_0": 0.14822, "loss_cls_1": 0.97103, "loss_box_1": 1.86565, "loss_cns_1": 0.64339, "loss_yns_1": 0.15432, "loss_cls_2": 0.97768, "loss_box_2": 1.83879, "loss_cns_2": 0.65192, "loss_yns_2": 0.15533, "loss_cls_3": 1.01592, "loss_box_3": 1.84479, "loss_cns_3": 0.65099, "loss_yns_3": 0.15337, "loss_cls_4": 1.01554, "loss_box_4": 1.83447, "loss_cns_4": 0.64991, "loss_yns_4": 0.15208, "loss_cls_5": 1.01274, "loss_box_5": 1.84347, "loss_cns_5": 0.65124, "loss_yns_5": 0.15363, "loss_cls_dn_0": 0.2461, "loss_box_dn_0": 0.79157, "loss_cls_dn_1": 0.17144, "loss_box_dn_1": 0.73913, "loss_cls_dn_2": 0.1772, "loss_box_dn_2": 0.72706, "loss_cls_dn_3": 0.18569, "loss_box_dn_3": 0.73376, "loss_cls_dn_4": 0.1861, "loss_box_dn_4": 0.74461, "loss_cls_dn_5": 0.19211, "loss_box_dn_5": 0.74779, "loss_dense_depth": 0.84617, "loss": 28.08294, "grad_norm": 41.89524, "time": 1.56584} -{"mode": "train", "epoch": 1, "iter": 118, "lr": 0.00015, "memory": 49164, "data_time": 0.07926, "loss_cls_0": 0.89327, "loss_box_0": 1.74233, "loss_cns_0": 0.61848, "loss_yns_0": 0.15089, "loss_cls_1": 0.97025, "loss_box_1": 1.84798, "loss_cns_1": 0.6421, "loss_yns_1": 0.15341, "loss_cls_2": 1.00405, "loss_box_2": 1.78172, "loss_cns_2": 0.64834, "loss_yns_2": 0.15287, "loss_cls_3": 1.01365, "loss_box_3": 1.78222, "loss_cns_3": 0.65313, "loss_yns_3": 0.15263, "loss_cls_4": 1.01845, "loss_box_4": 1.77318, "loss_cns_4": 0.65254, "loss_yns_4": 0.15227, "loss_cls_5": 0.99815, "loss_box_5": 1.83312, "loss_cns_5": 0.64685, "loss_yns_5": 0.15311, "loss_cls_dn_0": 0.24283, "loss_box_dn_0": 0.79162, "loss_cls_dn_1": 0.17087, "loss_box_dn_1": 0.75294, "loss_cls_dn_2": 0.17861, "loss_box_dn_2": 0.73695, "loss_cls_dn_3": 0.18469, "loss_box_dn_3": 0.74596, "loss_cls_dn_4": 0.19107, "loss_box_dn_4": 0.75789, "loss_cls_dn_5": 0.19279, "loss_box_dn_5": 0.7894, "loss_dense_depth": 0.90061, "loss": 28.07121, "grad_norm": 51.1715, "time": 1.58288} -{"mode": "train", "epoch": 1, "iter": 119, "lr": 0.00015, "memory": 49164, "data_time": 0.07321, "loss_cls_0": 0.89897, "loss_box_0": 1.75588, "loss_cns_0": 0.61736, "loss_yns_0": 0.15039, "loss_cls_1": 0.98758, "loss_box_1": 1.89226, "loss_cns_1": 0.64001, "loss_yns_1": 0.15399, "loss_cls_2": 1.02378, "loss_box_2": 1.81528, "loss_cns_2": 0.64748, "loss_yns_2": 0.15513, "loss_cls_3": 1.05155, "loss_box_3": 1.80802, "loss_cns_3": 0.65255, "loss_yns_3": 0.15399, "loss_cls_4": 1.05054, "loss_box_4": 1.81072, "loss_cns_4": 0.65193, "loss_yns_4": 0.15487, "loss_cls_5": 1.02634, "loss_box_5": 1.85757, "loss_cns_5": 0.64549, "loss_yns_5": 0.1514, "loss_cls_dn_0": 0.23689, "loss_box_dn_0": 0.78255, "loss_cls_dn_1": 0.169, "loss_box_dn_1": 0.75466, "loss_cls_dn_2": 0.17296, "loss_box_dn_2": 0.73538, "loss_cls_dn_3": 0.18375, "loss_box_dn_3": 0.73751, "loss_cls_dn_4": 0.18384, "loss_box_dn_4": 0.74744, "loss_cls_dn_5": 0.1871, "loss_box_dn_5": 0.7706, "loss_dense_depth": 0.82211, "loss": 28.23683, "grad_norm": 50.28642, "time": 1.57334} -{"mode": "train", "epoch": 1, "iter": 120, "lr": 0.00015, "memory": 49164, "data_time": 0.07751, "loss_cls_0": 0.8793, "loss_box_0": 1.75671, "loss_cns_0": 0.61316, "loss_yns_0": 0.15113, "loss_cls_1": 0.98572, "loss_box_1": 1.82118, "loss_cns_1": 0.64623, "loss_yns_1": 0.15661, "loss_cls_2": 1.0124, "loss_box_2": 1.80298, "loss_cns_2": 0.64913, "loss_yns_2": 0.15637, "loss_cls_3": 1.01006, "loss_box_3": 1.79178, "loss_cns_3": 0.65455, "loss_yns_3": 0.15577, "loss_cls_4": 1.01347, "loss_box_4": 1.75723, "loss_cns_4": 0.65474, "loss_yns_4": 0.15612, "loss_cls_5": 1.01137, "loss_box_5": 1.75574, "loss_cns_5": 0.65265, "loss_yns_5": 0.15557, "loss_cls_dn_0": 0.23832, "loss_box_dn_0": 0.79346, "loss_cls_dn_1": 0.17443, "loss_box_dn_1": 0.74649, "loss_cls_dn_2": 0.18062, "loss_box_dn_2": 0.73704, "loss_cls_dn_3": 0.18594, "loss_box_dn_3": 0.73941, "loss_cls_dn_4": 0.18231, "loss_box_dn_4": 0.73635, "loss_cls_dn_5": 0.18627, "loss_box_dn_5": 0.73782, "loss_dense_depth": 0.79841, "loss": 27.83683, "grad_norm": 42.50321, "time": 1.57972} -{"mode": "train", "epoch": 1, "iter": 121, "lr": 0.00015, "memory": 49164, "data_time": 0.10757, "loss_cls_0": 0.90856, "loss_box_0": 1.74995, "loss_cns_0": 0.61488, "loss_yns_0": 0.15306, "loss_cls_1": 1.015, "loss_box_1": 1.8513, "loss_cns_1": 0.64913, "loss_yns_1": 0.15731, "loss_cls_2": 1.04949, "loss_box_2": 1.82078, "loss_cns_2": 0.65334, "loss_yns_2": 0.15368, "loss_cls_3": 1.03599, "loss_box_3": 1.84482, "loss_cns_3": 0.65648, "loss_yns_3": 0.15676, "loss_cls_4": 1.04495, "loss_box_4": 1.80849, "loss_cns_4": 0.65654, "loss_yns_4": 0.15696, "loss_cls_5": 1.02479, "loss_box_5": 1.8186, "loss_cns_5": 0.65449, "loss_yns_5": 0.15359, "loss_cls_dn_0": 0.24354, "loss_box_dn_0": 0.79661, "loss_cls_dn_1": 0.17487, "loss_box_dn_1": 0.75421, "loss_cls_dn_2": 0.20008, "loss_box_dn_2": 0.74748, "loss_cls_dn_3": 0.20249, "loss_box_dn_3": 0.75549, "loss_cls_dn_4": 0.19649, "loss_box_dn_4": 0.75135, "loss_cls_dn_5": 0.19112, "loss_box_dn_5": 0.75993, "loss_dense_depth": 0.82348, "loss": 28.38608, "grad_norm": 53.25749, "time": 1.65036} -{"mode": "train", "epoch": 1, "iter": 122, "lr": 0.00015, "memory": 49164, "data_time": 0.10883, "loss_cls_0": 0.89476, "loss_box_0": 1.73788, "loss_cns_0": 0.62389, "loss_yns_0": 0.15528, "loss_cls_1": 0.98359, "loss_box_1": 1.843, "loss_cns_1": 0.6436, "loss_yns_1": 0.1571, "loss_cls_2": 1.03734, "loss_box_2": 1.8102, "loss_cns_2": 0.65233, "loss_yns_2": 0.15652, "loss_cls_3": 1.02078, "loss_box_3": 1.82335, "loss_cns_3": 0.65161, "loss_yns_3": 0.15733, "loss_cls_4": 1.02905, "loss_box_4": 1.81844, "loss_cns_4": 0.65078, "loss_yns_4": 0.15854, "loss_cls_5": 1.02308, "loss_box_5": 1.81139, "loss_cns_5": 0.65216, "loss_yns_5": 0.15626, "loss_cls_dn_0": 0.24244, "loss_box_dn_0": 0.78468, "loss_cls_dn_1": 0.17001, "loss_box_dn_1": 0.74896, "loss_cls_dn_2": 0.19754, "loss_box_dn_2": 0.74121, "loss_cls_dn_3": 0.19019, "loss_box_dn_3": 0.74351, "loss_cls_dn_4": 0.1962, "loss_box_dn_4": 0.74825, "loss_cls_dn_5": 0.19696, "loss_box_dn_5": 0.75487, "loss_dense_depth": 0.79954, "loss": 28.1626, "grad_norm": 53.54099, "time": 1.63896} -{"mode": "train", "epoch": 1, "iter": 123, "lr": 0.00015, "memory": 49164, "data_time": 0.09361, "loss_cls_0": 0.90537, "loss_box_0": 1.73647, "loss_cns_0": 0.625, "loss_yns_0": 0.15517, "loss_cls_1": 0.95728, "loss_box_1": 1.8536, "loss_cns_1": 0.64146, "loss_yns_1": 0.15848, "loss_cls_2": 1.00735, "loss_box_2": 1.79431, "loss_cns_2": 0.65495, "loss_yns_2": 0.15519, "loss_cls_3": 1.02123, "loss_box_3": 1.76596, "loss_cns_3": 0.655, "loss_yns_3": 0.15728, "loss_cls_4": 1.01748, "loss_box_4": 1.76433, "loss_cns_4": 0.65588, "loss_yns_4": 0.15734, "loss_cls_5": 0.99861, "loss_box_5": 1.76287, "loss_cns_5": 0.65183, "loss_yns_5": 0.15555, "loss_cls_dn_0": 0.22952, "loss_box_dn_0": 0.78263, "loss_cls_dn_1": 0.16736, "loss_box_dn_1": 0.76567, "loss_cls_dn_2": 0.18717, "loss_box_dn_2": 0.7537, "loss_cls_dn_3": 0.17699, "loss_box_dn_3": 0.74334, "loss_cls_dn_4": 0.1849, "loss_box_dn_4": 0.75155, "loss_cls_dn_5": 0.19284, "loss_box_dn_5": 0.75425, "loss_dense_depth": 0.82322, "loss": 27.92112, "grad_norm": 45.80585, "time": 1.61554} -{"mode": "train", "epoch": 1, "iter": 124, "lr": 0.00015, "memory": 49164, "data_time": 0.07678, "loss_cls_0": 0.90214, "loss_box_0": 1.74465, "loss_cns_0": 0.62468, "loss_yns_0": 0.1565, "loss_cls_1": 1.02319, "loss_box_1": 1.82965, "loss_cns_1": 0.64427, "loss_yns_1": 0.15786, "loss_cls_2": 1.01741, "loss_box_2": 1.7691, "loss_cns_2": 0.65337, "loss_yns_2": 0.15508, "loss_cls_3": 1.05131, "loss_box_3": 1.75857, "loss_cns_3": 0.65526, "loss_yns_3": 0.15673, "loss_cls_4": 1.04859, "loss_box_4": 1.75582, "loss_cns_4": 0.65739, "loss_yns_4": 0.15886, "loss_cls_5": 1.00756, "loss_box_5": 1.773, "loss_cns_5": 0.65211, "loss_yns_5": 0.15741, "loss_cls_dn_0": 0.23002, "loss_box_dn_0": 0.78692, "loss_cls_dn_1": 0.16286, "loss_box_dn_1": 0.75161, "loss_cls_dn_2": 0.17063, "loss_box_dn_2": 0.73425, "loss_cls_dn_3": 0.17669, "loss_box_dn_3": 0.73211, "loss_cls_dn_4": 0.17746, "loss_box_dn_4": 0.73624, "loss_cls_dn_5": 0.18568, "loss_box_dn_5": 0.7461, "loss_dense_depth": 0.7817, "loss": 27.8828, "grad_norm": 55.76987, "time": 1.60384} -{"mode": "train", "epoch": 1, "iter": 125, "lr": 0.00015, "memory": 49164, "data_time": 0.0758, "loss_cls_0": 0.90977, "loss_box_0": 1.73691, "loss_cns_0": 0.61434, "loss_yns_0": 0.15617, "loss_cls_1": 1.00392, "loss_box_1": 1.83382, "loss_cns_1": 0.647, "loss_yns_1": 0.1622, "loss_cls_2": 1.0333, "loss_box_2": 1.79199, "loss_cns_2": 0.65909, "loss_yns_2": 0.15764, "loss_cls_3": 1.05181, "loss_box_3": 1.80476, "loss_cns_3": 0.6562, "loss_yns_3": 0.1569, "loss_cls_4": 1.03864, "loss_box_4": 1.81616, "loss_cns_4": 0.65823, "loss_yns_4": 0.15799, "loss_cls_5": 1.0251, "loss_box_5": 1.81063, "loss_cns_5": 0.6548, "loss_yns_5": 0.15785, "loss_cls_dn_0": 0.23984, "loss_box_dn_0": 0.78081, "loss_cls_dn_1": 0.16614, "loss_box_dn_1": 0.74389, "loss_cls_dn_2": 0.17096, "loss_box_dn_2": 0.72897, "loss_cls_dn_3": 0.1802, "loss_box_dn_3": 0.73825, "loss_cls_dn_4": 0.18017, "loss_box_dn_4": 0.74382, "loss_cls_dn_5": 0.18942, "loss_box_dn_5": 0.7434, "loss_dense_depth": 0.79106, "loss": 28.09214, "grad_norm": 55.9435, "time": 1.62151} -{"mode": "train", "epoch": 1, "iter": 126, "lr": 0.00015, "memory": 49164, "data_time": 0.10605, "loss_cls_0": 0.93343, "loss_box_0": 1.69207, "loss_cns_0": 0.6101, "loss_yns_0": 0.15366, "loss_cls_1": 0.97509, "loss_box_1": 1.9254, "loss_cns_1": 0.63865, "loss_yns_1": 0.15698, "loss_cls_2": 1.02588, "loss_box_2": 1.8643, "loss_cns_2": 0.6515, "loss_yns_2": 0.15898, "loss_cls_3": 1.03993, "loss_box_3": 1.85477, "loss_cns_3": 0.64973, "loss_yns_3": 0.15696, "loss_cls_4": 1.02251, "loss_box_4": 1.86397, "loss_cns_4": 0.65248, "loss_yns_4": 0.15606, "loss_cls_5": 1.01632, "loss_box_5": 1.84975, "loss_cns_5": 0.64867, "loss_yns_5": 0.15601, "loss_cls_dn_0": 0.24272, "loss_box_dn_0": 0.78481, "loss_cls_dn_1": 0.171, "loss_box_dn_1": 0.74898, "loss_cls_dn_2": 0.17545, "loss_box_dn_2": 0.72959, "loss_cls_dn_3": 0.1815, "loss_box_dn_3": 0.73112, "loss_cls_dn_4": 0.18217, "loss_box_dn_4": 0.73785, "loss_cls_dn_5": 0.18975, "loss_box_dn_5": 0.73572, "loss_dense_depth": 0.76642, "loss": 28.23028, "grad_norm": 47.50231, "time": 1.58878} -{"mode": "train", "epoch": 1, "iter": 127, "lr": 0.00015, "memory": 49164, "data_time": 0.07647, "loss_cls_0": 0.90848, "loss_box_0": 1.71855, "loss_cns_0": 0.61356, "loss_yns_0": 0.15292, "loss_cls_1": 0.98174, "loss_box_1": 1.87584, "loss_cns_1": 0.64126, "loss_yns_1": 0.1567, "loss_cls_2": 1.01264, "loss_box_2": 1.80609, "loss_cns_2": 0.64896, "loss_yns_2": 0.15406, "loss_cls_3": 1.02858, "loss_box_3": 1.79151, "loss_cns_3": 0.65263, "loss_yns_3": 0.15474, "loss_cls_4": 1.0152, "loss_box_4": 1.77836, "loss_cns_4": 0.65437, "loss_yns_4": 0.15727, "loss_cls_5": 1.00244, "loss_box_5": 1.77995, "loss_cns_5": 0.65198, "loss_yns_5": 0.15429, "loss_cls_dn_0": 0.23964, "loss_box_dn_0": 0.78136, "loss_cls_dn_1": 0.16642, "loss_box_dn_1": 0.75712, "loss_cls_dn_2": 0.17258, "loss_box_dn_2": 0.7398, "loss_cls_dn_3": 0.17791, "loss_box_dn_3": 0.73515, "loss_cls_dn_4": 0.18143, "loss_box_dn_4": 0.73891, "loss_cls_dn_5": 0.18866, "loss_box_dn_5": 0.74442, "loss_dense_depth": 0.77162, "loss": 27.88716, "grad_norm": 39.41309, "time": 1.58496} -{"mode": "train", "epoch": 1, "iter": 128, "lr": 0.00015, "memory": 49164, "data_time": 0.08436, "loss_cls_0": 0.89515, "loss_box_0": 1.73173, "loss_cns_0": 0.62634, "loss_yns_0": 0.15739, "loss_cls_1": 0.96936, "loss_box_1": 1.86607, "loss_cns_1": 0.63707, "loss_yns_1": 0.15848, "loss_cls_2": 0.99829, "loss_box_2": 1.79727, "loss_cns_2": 0.64565, "loss_yns_2": 0.15575, "loss_cls_3": 1.00921, "loss_box_3": 1.79515, "loss_cns_3": 0.64886, "loss_yns_3": 0.15561, "loss_cls_4": 1.0125, "loss_box_4": 1.77485, "loss_cns_4": 0.64986, "loss_yns_4": 0.15706, "loss_cls_5": 1.00848, "loss_box_5": 1.77043, "loss_cns_5": 0.65059, "loss_yns_5": 0.1567, "loss_cls_dn_0": 0.23182, "loss_box_dn_0": 0.7797, "loss_cls_dn_1": 0.16273, "loss_box_dn_1": 0.75083, "loss_cls_dn_2": 0.17027, "loss_box_dn_2": 0.74745, "loss_cls_dn_3": 0.1775, "loss_box_dn_3": 0.74698, "loss_cls_dn_4": 0.18156, "loss_box_dn_4": 0.75164, "loss_cls_dn_5": 0.18987, "loss_box_dn_5": 0.75572, "loss_dense_depth": 0.81887, "loss": 27.89278, "grad_norm": 49.09036, "time": 1.57075} -{"mode": "train", "epoch": 1, "iter": 129, "lr": 0.00015, "memory": 49164, "data_time": 0.08416, "loss_cls_0": 0.92094, "loss_box_0": 1.72992, "loss_cns_0": 0.6238, "loss_yns_0": 0.15958, "loss_cls_1": 0.95449, "loss_box_1": 1.82554, "loss_cns_1": 0.63823, "loss_yns_1": 0.15621, "loss_cls_2": 0.98408, "loss_box_2": 1.7603, "loss_cns_2": 0.64947, "loss_yns_2": 0.15783, "loss_cls_3": 0.99887, "loss_box_3": 1.75835, "loss_cns_3": 0.65195, "loss_yns_3": 0.15654, "loss_cls_4": 1.01405, "loss_box_4": 1.73361, "loss_cns_4": 0.65463, "loss_yns_4": 0.15775, "loss_cls_5": 1.02241, "loss_box_5": 1.74459, "loss_cns_5": 0.65537, "loss_yns_5": 0.15838, "loss_cls_dn_0": 0.22487, "loss_box_dn_0": 0.77489, "loss_cls_dn_1": 0.16497, "loss_box_dn_1": 0.7519, "loss_cls_dn_2": 0.17158, "loss_box_dn_2": 0.74202, "loss_cls_dn_3": 0.17761, "loss_box_dn_3": 0.74049, "loss_cls_dn_4": 0.17791, "loss_box_dn_4": 0.7387, "loss_cls_dn_5": 0.18659, "loss_box_dn_5": 0.74524, "loss_dense_depth": 0.7677, "loss": 27.63134, "grad_norm": 45.16889, "time": 1.57882} -{"mode": "train", "epoch": 1, "iter": 130, "lr": 0.00015, "memory": 49164, "data_time": 0.08428, "loss_cls_0": 0.89155, "loss_box_0": 1.74956, "loss_cns_0": 0.62101, "loss_yns_0": 0.15668, "loss_cls_1": 0.95984, "loss_box_1": 1.81034, "loss_cns_1": 0.64182, "loss_yns_1": 0.15511, "loss_cls_2": 0.99063, "loss_box_2": 1.74558, "loss_cns_2": 0.6537, "loss_yns_2": 0.1548, "loss_cls_3": 1.00134, "loss_box_3": 1.74857, "loss_cns_3": 0.65563, "loss_yns_3": 0.15427, "loss_cls_4": 1.01522, "loss_box_4": 1.72528, "loss_cns_4": 0.6545, "loss_yns_4": 0.15972, "loss_cls_5": 1.0012, "loss_box_5": 1.73409, "loss_cns_5": 0.65747, "loss_yns_5": 0.15724, "loss_cls_dn_0": 0.23022, "loss_box_dn_0": 0.78334, "loss_cls_dn_1": 0.16261, "loss_box_dn_1": 0.74375, "loss_cls_dn_2": 0.16944, "loss_box_dn_2": 0.72617, "loss_cls_dn_3": 0.17324, "loss_box_dn_3": 0.72503, "loss_cls_dn_4": 0.17311, "loss_box_dn_4": 0.72209, "loss_cls_dn_5": 0.18024, "loss_box_dn_5": 0.72631, "loss_dense_depth": 0.81789, "loss": 27.52862, "grad_norm": 34.61907, "time": 1.57144} -{"mode": "train", "epoch": 1, "iter": 131, "lr": 0.00015, "memory": 49164, "data_time": 0.07778, "loss_cls_0": 0.91237, "loss_box_0": 1.72099, "loss_cns_0": 0.61763, "loss_yns_0": 0.1505, "loss_cls_1": 0.96154, "loss_box_1": 1.82886, "loss_cns_1": 0.64499, "loss_yns_1": 0.15464, "loss_cls_2": 1.00011, "loss_box_2": 1.75269, "loss_cns_2": 0.65604, "loss_yns_2": 0.15461, "loss_cls_3": 1.009, "loss_box_3": 1.75111, "loss_cns_3": 0.66093, "loss_yns_3": 0.15314, "loss_cls_4": 1.02669, "loss_box_4": 1.75041, "loss_cns_4": 0.65891, "loss_yns_4": 0.1553, "loss_cls_5": 1.01544, "loss_box_5": 1.75424, "loss_cns_5": 0.65726, "loss_yns_5": 0.15418, "loss_cls_dn_0": 0.23619, "loss_box_dn_0": 0.78533, "loss_cls_dn_1": 0.16148, "loss_box_dn_1": 0.73429, "loss_cls_dn_2": 0.16905, "loss_box_dn_2": 0.71431, "loss_cls_dn_3": 0.17172, "loss_box_dn_3": 0.71808, "loss_cls_dn_4": 0.18003, "loss_box_dn_4": 0.72589, "loss_cls_dn_5": 0.1884, "loss_box_dn_5": 0.73759, "loss_dense_depth": 0.91434, "loss": 27.73832, "grad_norm": 41.99177, "time": 1.62182} -{"mode": "train", "epoch": 1, "iter": 132, "lr": 0.00015, "memory": 49164, "data_time": 0.08012, "loss_cls_0": 0.89582, "loss_box_0": 1.7312, "loss_cns_0": 0.61541, "loss_yns_0": 0.15098, "loss_cls_1": 0.94949, "loss_box_1": 1.84267, "loss_cns_1": 0.64031, "loss_yns_1": 0.15206, "loss_cls_2": 0.98645, "loss_box_2": 1.76911, "loss_cns_2": 0.653, "loss_yns_2": 0.15347, "loss_cls_3": 0.99438, "loss_box_3": 1.77596, "loss_cns_3": 0.65172, "loss_yns_3": 0.1516, "loss_cls_4": 1.00028, "loss_box_4": 1.7824, "loss_cns_4": 0.65036, "loss_yns_4": 0.1524, "loss_cls_5": 1.01269, "loss_box_5": 1.78291, "loss_cns_5": 0.65017, "loss_yns_5": 0.15138, "loss_cls_dn_0": 0.23868, "loss_box_dn_0": 0.77922, "loss_cls_dn_1": 0.16934, "loss_box_dn_1": 0.74864, "loss_cls_dn_2": 0.17623, "loss_box_dn_2": 0.73304, "loss_cls_dn_3": 0.17815, "loss_box_dn_3": 0.7484, "loss_cls_dn_4": 0.1894, "loss_box_dn_4": 0.76929, "loss_cls_dn_5": 0.2, "loss_box_dn_5": 0.79042, "loss_dense_depth": 0.84637, "loss": 27.8634, "grad_norm": 48.60912, "time": 1.57939} -{"mode": "train", "epoch": 1, "iter": 133, "lr": 0.00015, "memory": 49164, "data_time": 0.07583, "loss_cls_0": 0.89328, "loss_box_0": 1.73595, "loss_cns_0": 0.61552, "loss_yns_0": 0.15304, "loss_cls_1": 0.9597, "loss_box_1": 1.91088, "loss_cns_1": 0.63548, "loss_yns_1": 0.15118, "loss_cls_2": 0.98314, "loss_box_2": 1.83873, "loss_cns_2": 0.64671, "loss_yns_2": 0.15467, "loss_cls_3": 1.00586, "loss_box_3": 1.84653, "loss_cns_3": 0.6494, "loss_yns_3": 0.15477, "loss_cls_4": 1.0435, "loss_box_4": 1.83903, "loss_cns_4": 0.64816, "loss_yns_4": 0.15641, "loss_cls_5": 1.02946, "loss_box_5": 1.83757, "loss_cns_5": 0.64922, "loss_yns_5": 0.15557, "loss_cls_dn_0": 0.23624, "loss_box_dn_0": 0.78976, "loss_cls_dn_1": 0.17111, "loss_box_dn_1": 0.76741, "loss_cls_dn_2": 0.17422, "loss_box_dn_2": 0.75487, "loss_cls_dn_3": 0.17395, "loss_box_dn_3": 0.77367, "loss_cls_dn_4": 0.18381, "loss_box_dn_4": 0.79619, "loss_cls_dn_5": 0.19512, "loss_box_dn_5": 0.81908, "loss_dense_depth": 0.86352, "loss": 28.39271, "grad_norm": 47.34305, "time": 1.57804} -{"mode": "train", "epoch": 1, "iter": 134, "lr": 0.00015, "memory": 49164, "data_time": 0.0794, "loss_cls_0": 0.85776, "loss_box_0": 1.72192, "loss_cns_0": 0.62499, "loss_yns_0": 0.15071, "loss_cls_1": 0.94876, "loss_box_1": 1.86528, "loss_cns_1": 0.64139, "loss_yns_1": 0.15329, "loss_cls_2": 0.97586, "loss_box_2": 1.81677, "loss_cns_2": 0.6475, "loss_yns_2": 0.15277, "loss_cls_3": 1.0065, "loss_box_3": 1.82334, "loss_cns_3": 0.65197, "loss_yns_3": 0.1561, "loss_cls_4": 1.02798, "loss_box_4": 1.80962, "loss_cns_4": 0.65254, "loss_yns_4": 0.15573, "loss_cls_5": 0.98897, "loss_box_5": 1.82137, "loss_cns_5": 0.65027, "loss_yns_5": 0.15478, "loss_cls_dn_0": 0.22529, "loss_box_dn_0": 0.78925, "loss_cls_dn_1": 0.16771, "loss_box_dn_1": 0.78312, "loss_cls_dn_2": 0.17095, "loss_box_dn_2": 0.77548, "loss_cls_dn_3": 0.17218, "loss_box_dn_3": 0.78736, "loss_cls_dn_4": 0.17626, "loss_box_dn_4": 0.80456, "loss_cls_dn_5": 0.18491, "loss_box_dn_5": 0.82758, "loss_dense_depth": 0.83747, "loss": 28.1583, "grad_norm": 45.1178, "time": 1.57326} -{"mode": "train", "epoch": 1, "iter": 135, "lr": 0.00015, "memory": 49164, "data_time": 0.08199, "loss_cls_0": 0.87863, "loss_box_0": 1.76185, "loss_cns_0": 0.62259, "loss_yns_0": 0.15384, "loss_cls_1": 0.94812, "loss_box_1": 1.88898, "loss_cns_1": 0.65023, "loss_yns_1": 0.1543, "loss_cls_2": 0.97663, "loss_box_2": 1.84142, "loss_cns_2": 0.65507, "loss_yns_2": 0.15381, "loss_cls_3": 0.99329, "loss_box_3": 1.83505, "loss_cns_3": 0.65519, "loss_yns_3": 0.15392, "loss_cls_4": 0.99585, "loss_box_4": 1.84199, "loss_cns_4": 0.65661, "loss_yns_4": 0.15743, "loss_cls_5": 1.00721, "loss_box_5": 1.84722, "loss_cns_5": 0.65364, "loss_yns_5": 0.15446, "loss_cls_dn_0": 0.23751, "loss_box_dn_0": 0.78999, "loss_cls_dn_1": 0.16357, "loss_box_dn_1": 0.79403, "loss_cls_dn_2": 0.16809, "loss_box_dn_2": 0.7815, "loss_cls_dn_3": 0.17656, "loss_box_dn_3": 0.78392, "loss_cls_dn_4": 0.17839, "loss_box_dn_4": 0.79891, "loss_cls_dn_5": 0.19765, "loss_box_dn_5": 0.81365, "loss_dense_depth": 0.80283, "loss": 28.32393, "grad_norm": 49.67049, "time": 1.58717} -{"mode": "train", "epoch": 1, "iter": 136, "lr": 0.00015, "memory": 49164, "data_time": 0.07454, "loss_cls_0": 0.84209, "loss_box_0": 1.7283, "loss_cns_0": 0.62484, "loss_yns_0": 0.14934, "loss_cls_1": 0.91581, "loss_box_1": 1.84537, "loss_cns_1": 0.64845, "loss_yns_1": 0.14931, "loss_cls_2": 0.94973, "loss_box_2": 1.79001, "loss_cns_2": 0.65701, "loss_yns_2": 0.15175, "loss_cls_3": 0.98131, "loss_box_3": 1.77791, "loss_cns_3": 0.65704, "loss_yns_3": 0.14898, "loss_cls_4": 1.00175, "loss_box_4": 1.78008, "loss_cns_4": 0.65875, "loss_yns_4": 0.15168, "loss_cls_5": 0.98019, "loss_box_5": 1.78388, "loss_cns_5": 0.65701, "loss_yns_5": 0.15093, "loss_cls_dn_0": 0.22863, "loss_box_dn_0": 0.78206, "loss_cls_dn_1": 0.15986, "loss_box_dn_1": 0.76744, "loss_cls_dn_2": 0.1686, "loss_box_dn_2": 0.74957, "loss_cls_dn_3": 0.17524, "loss_box_dn_3": 0.7487, "loss_cls_dn_4": 0.1838, "loss_box_dn_4": 0.75603, "loss_cls_dn_5": 0.19083, "loss_box_dn_5": 0.7621, "loss_dense_depth": 0.81224, "loss": 27.66662, "grad_norm": 45.53922, "time": 1.5753} -{"mode": "train", "epoch": 1, "iter": 137, "lr": 0.00015, "memory": 49164, "data_time": 0.07554, "loss_cls_0": 0.86108, "loss_box_0": 1.72929, "loss_cns_0": 0.6257, "loss_yns_0": 0.15154, "loss_cls_1": 0.92487, "loss_box_1": 1.82384, "loss_cns_1": 0.6461, "loss_yns_1": 0.14803, "loss_cls_2": 0.95774, "loss_box_2": 1.77632, "loss_cns_2": 0.64886, "loss_yns_2": 0.1498, "loss_cls_3": 0.98832, "loss_box_3": 1.77224, "loss_cns_3": 0.65283, "loss_yns_3": 0.15018, "loss_cls_4": 1.00998, "loss_box_4": 1.76056, "loss_cns_4": 0.65248, "loss_yns_4": 0.15019, "loss_cls_5": 0.96518, "loss_box_5": 1.76357, "loss_cns_5": 0.65223, "loss_yns_5": 0.14922, "loss_cls_dn_0": 0.23659, "loss_box_dn_0": 0.78362, "loss_cls_dn_1": 0.15536, "loss_box_dn_1": 0.73552, "loss_cls_dn_2": 0.17037, "loss_box_dn_2": 0.72045, "loss_cls_dn_3": 0.17524, "loss_box_dn_3": 0.72095, "loss_cls_dn_4": 0.18913, "loss_box_dn_4": 0.72071, "loss_cls_dn_5": 0.17956, "loss_box_dn_5": 0.72033, "loss_dense_depth": 0.78769, "loss": 27.4057, "grad_norm": 36.38641, "time": 1.57039} -{"mode": "train", "epoch": 1, "iter": 138, "lr": 0.00015, "memory": 49164, "data_time": 0.08084, "loss_cls_0": 0.85923, "loss_box_0": 1.72046, "loss_cns_0": 0.62602, "loss_yns_0": 0.14793, "loss_cls_1": 0.91516, "loss_box_1": 1.83994, "loss_cns_1": 0.64637, "loss_yns_1": 0.15069, "loss_cls_2": 0.94401, "loss_box_2": 1.77757, "loss_cns_2": 0.65177, "loss_yns_2": 0.14865, "loss_cls_3": 0.96329, "loss_box_3": 1.76322, "loss_cns_3": 0.65538, "loss_yns_3": 0.15089, "loss_cls_4": 0.97764, "loss_box_4": 1.76681, "loss_cns_4": 0.65354, "loss_yns_4": 0.15164, "loss_cls_5": 0.967, "loss_box_5": 1.78184, "loss_cns_5": 0.65174, "loss_yns_5": 0.1502, "loss_cls_dn_0": 0.23222, "loss_box_dn_0": 0.7935, "loss_cls_dn_1": 0.14925, "loss_box_dn_1": 0.7348, "loss_cls_dn_2": 0.16152, "loss_box_dn_2": 0.72537, "loss_cls_dn_3": 0.16104, "loss_box_dn_3": 0.73087, "loss_cls_dn_4": 0.17544, "loss_box_dn_4": 0.74367, "loss_cls_dn_5": 0.16865, "loss_box_dn_5": 0.76053, "loss_dense_depth": 0.83066, "loss": 27.42852, "grad_norm": 40.99288, "time": 1.57252} -{"mode": "train", "epoch": 1, "iter": 139, "lr": 0.00016, "memory": 49164, "data_time": 0.07649, "loss_cls_0": 0.82588, "loss_box_0": 1.6798, "loss_cns_0": 0.62509, "loss_yns_0": 0.14519, "loss_cls_1": 0.90489, "loss_box_1": 1.80615, "loss_cns_1": 0.65425, "loss_yns_1": 0.14768, "loss_cls_2": 0.94999, "loss_box_2": 1.7577, "loss_cns_2": 0.65392, "loss_yns_2": 0.146, "loss_cls_3": 0.96109, "loss_box_3": 1.75717, "loss_cns_3": 0.65364, "loss_yns_3": 0.14963, "loss_cls_4": 0.94661, "loss_box_4": 1.76546, "loss_cns_4": 0.6538, "loss_yns_4": 0.14709, "loss_cls_5": 0.96693, "loss_box_5": 1.77917, "loss_cns_5": 0.65111, "loss_yns_5": 0.14815, "loss_cls_dn_0": 0.22508, "loss_box_dn_0": 0.78449, "loss_cls_dn_1": 0.14859, "loss_box_dn_1": 0.75314, "loss_cls_dn_2": 0.15373, "loss_box_dn_2": 0.7551, "loss_cls_dn_3": 0.16187, "loss_box_dn_3": 0.77582, "loss_cls_dn_4": 0.16908, "loss_box_dn_4": 0.80893, "loss_cls_dn_5": 0.17453, "loss_box_dn_5": 0.84068, "loss_dense_depth": 0.79072, "loss": 27.41813, "grad_norm": 52.66274, "time": 1.56583} -{"mode": "train", "epoch": 1, "iter": 140, "lr": 0.00016, "memory": 49164, "data_time": 0.07708, "loss_cls_0": 0.83234, "loss_box_0": 1.67799, "loss_cns_0": 0.62116, "loss_yns_0": 0.14476, "loss_cls_1": 0.89829, "loss_box_1": 1.79184, "loss_cns_1": 0.65187, "loss_yns_1": 0.14806, "loss_cls_2": 0.93425, "loss_box_2": 1.74409, "loss_cns_2": 0.6585, "loss_yns_2": 0.14836, "loss_cls_3": 0.94695, "loss_box_3": 1.74531, "loss_cns_3": 0.65939, "loss_yns_3": 0.14862, "loss_cls_4": 0.9516, "loss_box_4": 1.75197, "loss_cns_4": 0.65839, "loss_yns_4": 0.15053, "loss_cls_5": 0.94724, "loss_box_5": 1.76981, "loss_cns_5": 0.65807, "loss_yns_5": 0.14942, "loss_cls_dn_0": 0.22576, "loss_box_dn_0": 0.78355, "loss_cls_dn_1": 0.1455, "loss_box_dn_1": 0.78725, "loss_cls_dn_2": 0.14712, "loss_box_dn_2": 0.79218, "loss_cls_dn_3": 0.158, "loss_box_dn_3": 0.81193, "loss_cls_dn_4": 0.15973, "loss_box_dn_4": 0.84825, "loss_cls_dn_5": 0.17076, "loss_box_dn_5": 0.87631, "loss_dense_depth": 0.77601, "loss": 27.47116, "grad_norm": 49.01032, "time": 1.56997} -{"mode": "train", "epoch": 1, "iter": 141, "lr": 0.00016, "memory": 49164, "data_time": 0.10573, "loss_cls_0": 0.88491, "loss_box_0": 1.73587, "loss_cns_0": 0.62118, "loss_yns_0": 0.14887, "loss_cls_1": 0.91914, "loss_box_1": 1.82358, "loss_cns_1": 0.64433, "loss_yns_1": 0.151, "loss_cls_2": 0.93896, "loss_box_2": 1.80868, "loss_cns_2": 0.65184, "loss_yns_2": 0.15066, "loss_cls_3": 0.95407, "loss_box_3": 1.8035, "loss_cns_3": 0.65524, "loss_yns_3": 0.1546, "loss_cls_4": 0.98405, "loss_box_4": 1.78018, "loss_cns_4": 0.65493, "loss_yns_4": 0.15403, "loss_cls_5": 0.96031, "loss_box_5": 1.79918, "loss_cns_5": 0.6554, "loss_yns_5": 0.15339, "loss_cls_dn_0": 0.24071, "loss_box_dn_0": 0.78412, "loss_cls_dn_1": 0.15265, "loss_box_dn_1": 0.81314, "loss_cls_dn_2": 0.15693, "loss_box_dn_2": 0.81214, "loss_cls_dn_3": 0.16575, "loss_box_dn_3": 0.82175, "loss_cls_dn_4": 0.17229, "loss_box_dn_4": 0.83791, "loss_cls_dn_5": 0.17966, "loss_box_dn_5": 0.85335, "loss_dense_depth": 0.81418, "loss": 27.99246, "grad_norm": 47.45939, "time": 1.67413} -{"mode": "train", "epoch": 1, "iter": 142, "lr": 0.00016, "memory": 49164, "data_time": 0.10189, "loss_cls_0": 0.88695, "loss_box_0": 1.78254, "loss_cns_0": 0.61705, "loss_yns_0": 0.1525, "loss_cls_1": 0.93274, "loss_box_1": 1.80337, "loss_cns_1": 0.64249, "loss_yns_1": 0.15538, "loss_cls_2": 0.95745, "loss_box_2": 1.80258, "loss_cns_2": 0.64154, "loss_yns_2": 0.15397, "loss_cls_3": 0.96591, "loss_box_3": 1.77028, "loss_cns_3": 0.65136, "loss_yns_3": 0.15747, "loss_cls_4": 0.98755, "loss_box_4": 1.72199, "loss_cns_4": 0.65394, "loss_yns_4": 0.15398, "loss_cls_5": 0.97256, "loss_box_5": 1.72301, "loss_cns_5": 0.65354, "loss_yns_5": 0.15291, "loss_cls_dn_0": 0.2375, "loss_box_dn_0": 0.78713, "loss_cls_dn_1": 0.15451, "loss_box_dn_1": 0.81285, "loss_cls_dn_2": 0.15935, "loss_box_dn_2": 0.80312, "loss_cls_dn_3": 0.16681, "loss_box_dn_3": 0.79649, "loss_cls_dn_4": 0.17728, "loss_box_dn_4": 0.78884, "loss_cls_dn_5": 0.17678, "loss_box_dn_5": 0.79054, "loss_dense_depth": 0.81136, "loss": 27.75563, "grad_norm": 40.70667, "time": 1.62828} -{"mode": "train", "epoch": 1, "iter": 143, "lr": 0.00016, "memory": 49164, "data_time": 0.08242, "loss_cls_0": 0.84321, "loss_box_0": 1.7614, "loss_cns_0": 0.62509, "loss_yns_0": 0.14918, "loss_cls_1": 0.92127, "loss_box_1": 1.77125, "loss_cns_1": 0.65642, "loss_yns_1": 0.14859, "loss_cls_2": 0.95254, "loss_box_2": 1.71486, "loss_cns_2": 0.65897, "loss_yns_2": 0.15073, "loss_cls_3": 0.95327, "loss_box_3": 1.70608, "loss_cns_3": 0.65519, "loss_yns_3": 0.15053, "loss_cls_4": 0.96392, "loss_box_4": 1.70984, "loss_cns_4": 0.65489, "loss_yns_4": 0.14843, "loss_cls_5": 0.96812, "loss_box_5": 1.73467, "loss_cns_5": 0.65408, "loss_yns_5": 0.1489, "loss_cls_dn_0": 0.22211, "loss_box_dn_0": 0.78173, "loss_cls_dn_1": 0.15308, "loss_box_dn_1": 0.75652, "loss_cls_dn_2": 0.1628, "loss_box_dn_2": 0.73418, "loss_cls_dn_3": 0.16639, "loss_box_dn_3": 0.73194, "loss_cls_dn_4": 0.17376, "loss_box_dn_4": 0.73593, "loss_cls_dn_5": 0.17933, "loss_box_dn_5": 0.74633, "loss_dense_depth": 0.7868, "loss": 27.13235, "grad_norm": 37.96473, "time": 1.58044} -{"mode": "train", "epoch": 1, "iter": 144, "lr": 0.00016, "memory": 49164, "data_time": 0.07523, "loss_cls_0": 0.87446, "loss_box_0": 1.77376, "loss_cns_0": 0.62229, "loss_yns_0": 0.15033, "loss_cls_1": 0.93265, "loss_box_1": 1.81433, "loss_cns_1": 0.64748, "loss_yns_1": 0.14848, "loss_cls_2": 0.96862, "loss_box_2": 1.77272, "loss_cns_2": 0.65746, "loss_yns_2": 0.14867, "loss_cls_3": 0.97042, "loss_box_3": 1.74657, "loss_cns_3": 0.65566, "loss_yns_3": 0.14991, "loss_cls_4": 0.97969, "loss_box_4": 1.75302, "loss_cns_4": 0.65697, "loss_yns_4": 0.14927, "loss_cls_5": 0.97521, "loss_box_5": 1.75971, "loss_cns_5": 0.65505, "loss_yns_5": 0.15034, "loss_cls_dn_0": 0.22651, "loss_box_dn_0": 0.79212, "loss_cls_dn_1": 0.15856, "loss_box_dn_1": 0.76035, "loss_cls_dn_2": 0.1687, "loss_box_dn_2": 0.75465, "loss_cls_dn_3": 0.17313, "loss_box_dn_3": 0.76446, "loss_cls_dn_4": 0.17637, "loss_box_dn_4": 0.7815, "loss_cls_dn_5": 0.18596, "loss_box_dn_5": 0.80191, "loss_dense_depth": 0.80881, "loss": 27.6661, "grad_norm": 43.97732, "time": 1.58319} -{"mode": "train", "epoch": 1, "iter": 145, "lr": 0.00016, "memory": 49164, "data_time": 0.08233, "loss_cls_0": 0.85268, "loss_box_0": 1.77343, "loss_cns_0": 0.62315, "loss_yns_0": 0.15188, "loss_cls_1": 0.92178, "loss_box_1": 1.83258, "loss_cns_1": 0.64339, "loss_yns_1": 0.15319, "loss_cls_2": 0.94861, "loss_box_2": 1.79508, "loss_cns_2": 0.65231, "loss_yns_2": 0.15034, "loss_cls_3": 0.95895, "loss_box_3": 1.7639, "loss_cns_3": 0.65292, "loss_yns_3": 0.15224, "loss_cls_4": 0.97703, "loss_box_4": 1.76169, "loss_cns_4": 0.65254, "loss_yns_4": 0.14955, "loss_cls_5": 0.95913, "loss_box_5": 1.78167, "loss_cns_5": 0.65082, "loss_yns_5": 0.14973, "loss_cls_dn_0": 0.22174, "loss_box_dn_0": 0.78224, "loss_cls_dn_1": 0.15435, "loss_box_dn_1": 0.78134, "loss_cls_dn_2": 0.16109, "loss_box_dn_2": 0.78241, "loss_cls_dn_3": 0.16707, "loss_box_dn_3": 0.79792, "loss_cls_dn_4": 0.17218, "loss_box_dn_4": 0.8203, "loss_cls_dn_5": 0.17838, "loss_box_dn_5": 0.84778, "loss_dense_depth": 0.79936, "loss": 27.77476, "grad_norm": 44.89906, "time": 1.59999} -{"mode": "train", "epoch": 1, "iter": 146, "lr": 0.00016, "memory": 49164, "data_time": 0.11554, "loss_cls_0": 0.83624, "loss_box_0": 1.7533, "loss_cns_0": 0.61888, "loss_yns_0": 0.14914, "loss_cls_1": 0.91994, "loss_box_1": 1.82366, "loss_cns_1": 0.64461, "loss_yns_1": 0.14917, "loss_cls_2": 0.94817, "loss_box_2": 1.7858, "loss_cns_2": 0.65315, "loss_yns_2": 0.14965, "loss_cls_3": 0.96213, "loss_box_3": 1.75404, "loss_cns_3": 0.6513, "loss_yns_3": 0.16026, "loss_cls_4": 0.96588, "loss_box_4": 1.73454, "loss_cns_4": 0.64961, "loss_yns_4": 0.15151, "loss_cls_5": 0.96725, "loss_box_5": 1.75844, "loss_cns_5": 0.65019, "loss_yns_5": 0.14994, "loss_cls_dn_0": 0.21929, "loss_box_dn_0": 0.78288, "loss_cls_dn_1": 0.14799, "loss_box_dn_1": 0.80678, "loss_cls_dn_2": 0.15496, "loss_box_dn_2": 0.80883, "loss_cls_dn_3": 0.15806, "loss_box_dn_3": 0.82571, "loss_cls_dn_4": 0.16519, "loss_box_dn_4": 0.84094, "loss_cls_dn_5": 0.17038, "loss_box_dn_5": 0.86904, "loss_dense_depth": 0.7742, "loss": 27.71104, "grad_norm": 39.06192, "time": 1.60228} -{"mode": "train", "epoch": 1, "iter": 147, "lr": 0.00016, "memory": 49164, "data_time": 0.07427, "loss_cls_0": 0.85816, "loss_box_0": 1.73822, "loss_cns_0": 0.61766, "loss_yns_0": 0.14811, "loss_cls_1": 0.94277, "loss_box_1": 1.85403, "loss_cns_1": 0.64631, "loss_yns_1": 0.14868, "loss_cls_2": 0.97992, "loss_box_2": 1.76252, "loss_cns_2": 0.65639, "loss_yns_2": 0.15066, "loss_cls_3": 0.97596, "loss_box_3": 1.74274, "loss_cns_3": 0.65724, "loss_yns_3": 0.15539, "loss_cls_4": 0.96638, "loss_box_4": 1.74759, "loss_cns_4": 0.65357, "loss_yns_4": 0.15221, "loss_cls_5": 0.99894, "loss_box_5": 1.77461, "loss_cns_5": 0.65555, "loss_yns_5": 0.15271, "loss_cls_dn_0": 0.22394, "loss_box_dn_0": 0.77919, "loss_cls_dn_1": 0.15359, "loss_box_dn_1": 0.83386, "loss_cls_dn_2": 0.16553, "loss_box_dn_2": 0.81401, "loss_cls_dn_3": 0.16397, "loss_box_dn_3": 0.8254, "loss_cls_dn_4": 0.1681, "loss_box_dn_4": 0.83947, "loss_cls_dn_5": 0.17461, "loss_box_dn_5": 0.86202, "loss_dense_depth": 0.80749, "loss": 27.94751, "grad_norm": 41.76451, "time": 1.56339} -{"mode": "train", "epoch": 1, "iter": 148, "lr": 0.00016, "memory": 49164, "data_time": 0.08083, "loss_cls_0": 0.8505, "loss_box_0": 1.73062, "loss_cns_0": 0.62129, "loss_yns_0": 0.1451, "loss_cls_1": 0.91331, "loss_box_1": 1.84472, "loss_cns_1": 0.64109, "loss_yns_1": 0.14647, "loss_cls_2": 0.9395, "loss_box_2": 1.7089, "loss_cns_2": 0.65813, "loss_yns_2": 0.15034, "loss_cls_3": 0.95121, "loss_box_3": 1.68466, "loss_cns_3": 0.65989, "loss_yns_3": 0.15246, "loss_cls_4": 0.95802, "loss_box_4": 1.68447, "loss_cns_4": 0.65763, "loss_yns_4": 0.15477, "loss_cls_5": 0.95578, "loss_box_5": 1.7009, "loss_cns_5": 0.66117, "loss_yns_5": 0.15518, "loss_cls_dn_0": 0.21998, "loss_box_dn_0": 0.78683, "loss_cls_dn_1": 0.15509, "loss_box_dn_1": 0.81045, "loss_cls_dn_2": 0.16381, "loss_box_dn_2": 0.7617, "loss_cls_dn_3": 0.15968, "loss_box_dn_3": 0.75945, "loss_cls_dn_4": 0.16264, "loss_box_dn_4": 0.76367, "loss_cls_dn_5": 0.16663, "loss_box_dn_5": 0.77247, "loss_dense_depth": 0.75861, "loss": 27.16714, "grad_norm": 41.16401, "time": 1.5814} -{"mode": "train", "epoch": 1, "iter": 149, "lr": 0.00016, "memory": 49164, "data_time": 0.08104, "loss_cls_0": 0.83004, "loss_box_0": 1.72755, "loss_cns_0": 0.62094, "loss_yns_0": 0.14794, "loss_cls_1": 0.90871, "loss_box_1": 1.78966, "loss_cns_1": 0.64513, "loss_yns_1": 0.14846, "loss_cls_2": 0.93319, "loss_box_2": 1.73997, "loss_cns_2": 0.65249, "loss_yns_2": 0.14861, "loss_cls_3": 0.95053, "loss_box_3": 1.70819, "loss_cns_3": 0.65366, "loss_yns_3": 0.15325, "loss_cls_4": 0.9515, "loss_box_4": 1.7253, "loss_cns_4": 0.65062, "loss_yns_4": 0.15219, "loss_cls_5": 0.95547, "loss_box_5": 1.7538, "loss_cns_5": 0.65021, "loss_yns_5": 0.15346, "loss_cls_dn_0": 0.21584, "loss_box_dn_0": 0.78058, "loss_cls_dn_1": 0.14906, "loss_box_dn_1": 0.75623, "loss_cls_dn_2": 0.15177, "loss_box_dn_2": 0.74645, "loss_cls_dn_3": 0.15208, "loss_box_dn_3": 0.73944, "loss_cls_dn_4": 0.15714, "loss_box_dn_4": 0.75133, "loss_cls_dn_5": 0.16132, "loss_box_dn_5": 0.76656, "loss_dense_depth": 0.76448, "loss": 27.04317, "grad_norm": 42.26622, "time": 1.60552} -{"mode": "train", "epoch": 1, "iter": 150, "lr": 0.00016, "memory": 49164, "data_time": 0.07925, "loss_cls_0": 0.85467, "loss_box_0": 1.74192, "loss_cns_0": 0.62181, "loss_yns_0": 0.14727, "loss_cls_1": 0.93036, "loss_box_1": 1.75907, "loss_cns_1": 0.65347, "loss_yns_1": 0.14688, "loss_cls_2": 0.94961, "loss_box_2": 1.74571, "loss_cns_2": 0.65348, "loss_yns_2": 0.14728, "loss_cls_3": 0.96085, "loss_box_3": 1.7205, "loss_cns_3": 0.66032, "loss_yns_3": 0.15664, "loss_cls_4": 0.95447, "loss_box_4": 1.73648, "loss_cns_4": 0.65247, "loss_yns_4": 0.14698, "loss_cls_5": 0.97127, "loss_box_5": 1.76786, "loss_cns_5": 0.6529, "loss_yns_5": 0.14733, "loss_cls_dn_0": 0.21934, "loss_box_dn_0": 0.77901, "loss_cls_dn_1": 0.14515, "loss_box_dn_1": 0.75343, "loss_cls_dn_2": 0.14944, "loss_box_dn_2": 0.77127, "loss_cls_dn_3": 0.15383, "loss_box_dn_3": 0.78047, "loss_cls_dn_4": 0.15769, "loss_box_dn_4": 0.81208, "loss_cls_dn_5": 0.16648, "loss_box_dn_5": 0.84565, "loss_dense_depth": 0.76337, "loss": 27.37678, "grad_norm": 44.43503, "time": 1.58164} -{"mode": "train", "epoch": 1, "iter": 151, "lr": 0.00016, "memory": 49164, "data_time": 0.08488, "loss_cls_0": 0.89836, "loss_box_0": 1.74136, "loss_cns_0": 0.62738, "loss_yns_0": 0.14598, "loss_cls_1": 0.92863, "loss_box_1": 1.8023, "loss_cns_1": 0.64611, "loss_yns_1": 0.14777, "loss_cls_2": 0.94938, "loss_box_2": 1.78614, "loss_cns_2": 0.64921, "loss_yns_2": 0.14803, "loss_cls_3": 0.95968, "loss_box_3": 1.78147, "loss_cns_3": 0.65671, "loss_yns_3": 0.1604, "loss_cls_4": 0.95642, "loss_box_4": 1.77006, "loss_cns_4": 0.65011, "loss_yns_4": 0.14982, "loss_cls_5": 0.96398, "loss_box_5": 1.77636, "loss_cns_5": 0.65465, "loss_yns_5": 0.14905, "loss_cls_dn_0": 0.21775, "loss_box_dn_0": 0.77973, "loss_cls_dn_1": 0.14355, "loss_box_dn_1": 0.79747, "loss_cls_dn_2": 0.14922, "loss_box_dn_2": 0.8168, "loss_cls_dn_3": 0.152, "loss_box_dn_3": 0.84047, "loss_cls_dn_4": 0.15686, "loss_box_dn_4": 0.86563, "loss_cls_dn_5": 0.16642, "loss_box_dn_5": 0.89593, "loss_dense_depth": 0.77872, "loss": 27.85992, "grad_norm": 49.29742, "time": 1.59172} -{"mode": "train", "epoch": 1, "iter": 152, "lr": 0.00016, "memory": 49164, "data_time": 0.07176, "loss_cls_0": 0.88041, "loss_box_0": 1.71304, "loss_cns_0": 0.6251, "loss_yns_0": 0.14853, "loss_cls_1": 0.91467, "loss_box_1": 1.8467, "loss_cns_1": 0.63331, "loss_yns_1": 0.15264, "loss_cls_2": 0.94638, "loss_box_2": 1.78251, "loss_cns_2": 0.65016, "loss_yns_2": 0.15156, "loss_cls_3": 0.95225, "loss_box_3": 1.76807, "loss_cns_3": 0.65375, "loss_yns_3": 0.15745, "loss_cls_4": 0.96124, "loss_box_4": 1.75368, "loss_cns_4": 0.65213, "loss_yns_4": 0.15283, "loss_cls_5": 0.95087, "loss_box_5": 1.75826, "loss_cns_5": 0.65416, "loss_yns_5": 0.1531, "loss_cls_dn_0": 0.21602, "loss_box_dn_0": 0.77281, "loss_cls_dn_1": 0.1445, "loss_box_dn_1": 0.82969, "loss_cls_dn_2": 0.14934, "loss_box_dn_2": 0.83426, "loss_cls_dn_3": 0.1529, "loss_box_dn_3": 0.84947, "loss_cls_dn_4": 0.15873, "loss_box_dn_4": 0.86653, "loss_cls_dn_5": 0.16414, "loss_box_dn_5": 0.89421, "loss_dense_depth": 0.72973, "loss": 27.77512, "grad_norm": 40.49258, "time": 1.57602} -{"mode": "train", "epoch": 1, "iter": 153, "lr": 0.00016, "memory": 49164, "data_time": 0.07124, "loss_cls_0": 0.86571, "loss_box_0": 1.71622, "loss_cns_0": 0.61623, "loss_yns_0": 0.14959, "loss_cls_1": 0.91744, "loss_box_1": 1.92595, "loss_cns_1": 0.62538, "loss_yns_1": 0.15125, "loss_cls_2": 0.95752, "loss_box_2": 1.77657, "loss_cns_2": 0.64853, "loss_yns_2": 0.15063, "loss_cls_3": 0.96541, "loss_box_3": 1.75928, "loss_cns_3": 0.65357, "loss_yns_3": 0.14993, "loss_cls_4": 0.9779, "loss_box_4": 1.74653, "loss_cns_4": 0.65153, "loss_yns_4": 0.15042, "loss_cls_5": 0.96859, "loss_box_5": 1.74949, "loss_cns_5": 0.65128, "loss_yns_5": 0.15016, "loss_cls_dn_0": 0.22849, "loss_box_dn_0": 0.77892, "loss_cls_dn_1": 0.145, "loss_box_dn_1": 0.8154, "loss_cls_dn_2": 0.14961, "loss_box_dn_2": 0.77717, "loss_cls_dn_3": 0.15138, "loss_box_dn_3": 0.77704, "loss_cls_dn_4": 0.15832, "loss_box_dn_4": 0.78369, "loss_cls_dn_5": 0.16156, "loss_box_dn_5": 0.7966, "loss_dense_depth": 0.76666, "loss": 27.56495, "grad_norm": 43.51552, "time": 1.58094} -{"mode": "train", "epoch": 1, "iter": 154, "lr": 0.00016, "memory": 49164, "data_time": 0.07326, "loss_cls_0": 0.86989, "loss_box_0": 1.69603, "loss_cns_0": 0.60778, "loss_yns_0": 0.14499, "loss_cls_1": 0.90783, "loss_box_1": 1.77651, "loss_cns_1": 0.64169, "loss_yns_1": 0.15175, "loss_cls_2": 0.94194, "loss_box_2": 1.72967, "loss_cns_2": 0.64444, "loss_yns_2": 0.15101, "loss_cls_3": 0.96379, "loss_box_3": 1.71533, "loss_cns_3": 0.65589, "loss_yns_3": 0.15634, "loss_cls_4": 0.96096, "loss_box_4": 1.70766, "loss_cns_4": 0.65162, "loss_yns_4": 0.15271, "loss_cls_5": 0.95225, "loss_box_5": 1.72732, "loss_cns_5": 0.65246, "loss_yns_5": 0.15126, "loss_cls_dn_0": 0.22601, "loss_box_dn_0": 0.77513, "loss_cls_dn_1": 0.14441, "loss_box_dn_1": 0.76236, "loss_cls_dn_2": 0.14919, "loss_box_dn_2": 0.73753, "loss_cls_dn_3": 0.15201, "loss_box_dn_3": 0.73481, "loss_cls_dn_4": 0.15453, "loss_box_dn_4": 0.74021, "loss_cls_dn_5": 0.15744, "loss_box_dn_5": 0.74162, "loss_dense_depth": 0.73585, "loss": 26.92224, "grad_norm": 49.46885, "time": 1.59112} -{"mode": "train", "epoch": 1, "iter": 155, "lr": 0.00016, "memory": 49164, "data_time": 0.07902, "loss_cls_0": 0.84191, "loss_box_0": 1.68901, "loss_cns_0": 0.6154, "loss_yns_0": 0.14761, "loss_cls_1": 0.90325, "loss_box_1": 1.71917, "loss_cns_1": 0.64833, "loss_yns_1": 0.1494, "loss_cls_2": 0.92692, "loss_box_2": 1.70075, "loss_cns_2": 0.65135, "loss_yns_2": 0.14984, "loss_cls_3": 0.94477, "loss_box_3": 1.66991, "loss_cns_3": 0.66492, "loss_yns_3": 0.1613, "loss_cls_4": 0.93678, "loss_box_4": 1.66795, "loss_cns_4": 0.65737, "loss_yns_4": 0.15191, "loss_cls_5": 0.93595, "loss_box_5": 1.68695, "loss_cns_5": 0.65893, "loss_yns_5": 0.15117, "loss_cls_dn_0": 0.22316, "loss_box_dn_0": 0.77797, "loss_cls_dn_1": 0.14197, "loss_box_dn_1": 0.73289, "loss_cls_dn_2": 0.14427, "loss_box_dn_2": 0.72106, "loss_cls_dn_3": 0.14992, "loss_box_dn_3": 0.71909, "loss_cls_dn_4": 0.14781, "loss_box_dn_4": 0.72947, "loss_cls_dn_5": 0.15307, "loss_box_dn_5": 0.73867, "loss_dense_depth": 0.72226, "loss": 26.53247, "grad_norm": 41.40023, "time": 1.58274} -{"mode": "train", "epoch": 1, "iter": 156, "lr": 0.00016, "memory": 49164, "data_time": 0.07529, "loss_cls_0": 0.82932, "loss_box_0": 1.70843, "loss_cns_0": 0.61881, "loss_yns_0": 0.15129, "loss_cls_1": 0.90991, "loss_box_1": 1.73907, "loss_cns_1": 0.64216, "loss_yns_1": 0.15035, "loss_cls_2": 0.93415, "loss_box_2": 1.66962, "loss_cns_2": 0.65285, "loss_yns_2": 0.15166, "loss_cls_3": 0.94737, "loss_box_3": 1.71904, "loss_cns_3": 0.65129, "loss_yns_3": 0.15712, "loss_cls_4": 0.95184, "loss_box_4": 1.72629, "loss_cns_4": 0.64862, "loss_yns_4": 0.15287, "loss_cls_5": 0.94897, "loss_box_5": 1.71893, "loss_cns_5": 0.65032, "loss_yns_5": 0.15399, "loss_cls_dn_0": 0.22405, "loss_box_dn_0": 0.77194, "loss_cls_dn_1": 0.14333, "loss_box_dn_1": 0.77573, "loss_cls_dn_2": 0.14228, "loss_box_dn_2": 0.76337, "loss_cls_dn_3": 0.14413, "loss_box_dn_3": 0.79218, "loss_cls_dn_4": 0.15319, "loss_box_dn_4": 0.81294, "loss_cls_dn_5": 0.15898, "loss_box_dn_5": 0.82976, "loss_dense_depth": 0.75965, "loss": 27.0558, "grad_norm": 57.28703, "time": 1.56285} -{"mode": "train", "epoch": 1, "iter": 157, "lr": 0.00016, "memory": 49164, "data_time": 0.0812, "loss_cls_0": 0.84352, "loss_box_0": 1.72959, "loss_cns_0": 0.62237, "loss_yns_0": 0.15417, "loss_cls_1": 0.90794, "loss_box_1": 1.73154, "loss_cns_1": 0.6427, "loss_yns_1": 0.15316, "loss_cls_2": 0.93116, "loss_box_2": 1.70765, "loss_cns_2": 0.64908, "loss_yns_2": 0.15287, "loss_cls_3": 0.961, "loss_box_3": 1.74093, "loss_cns_3": 0.64997, "loss_yns_3": 0.15533, "loss_cls_4": 0.94189, "loss_box_4": 1.73595, "loss_cns_4": 0.64786, "loss_yns_4": 0.15695, "loss_cls_5": 0.94802, "loss_box_5": 1.72718, "loss_cns_5": 0.64815, "loss_yns_5": 0.1584, "loss_cls_dn_0": 0.2168, "loss_box_dn_0": 0.77096, "loss_cls_dn_1": 0.14213, "loss_box_dn_1": 0.79236, "loss_cls_dn_2": 0.14224, "loss_box_dn_2": 0.80559, "loss_cls_dn_3": 0.14701, "loss_box_dn_3": 0.84137, "loss_cls_dn_4": 0.15291, "loss_box_dn_4": 0.86977, "loss_cls_dn_5": 0.15724, "loss_box_dn_5": 0.89446, "loss_dense_depth": 0.72687, "loss": 27.35708, "grad_norm": 48.23488, "time": 1.569} -{"mode": "train", "epoch": 1, "iter": 158, "lr": 0.00016, "memory": 49164, "data_time": 0.07821, "loss_cls_0": 0.81638, "loss_box_0": 1.75742, "loss_cns_0": 0.61689, "loss_yns_0": 0.15403, "loss_cls_1": 0.90146, "loss_box_1": 1.79218, "loss_cns_1": 0.63551, "loss_yns_1": 0.15183, "loss_cls_2": 0.92499, "loss_box_2": 1.77577, "loss_cns_2": 0.64497, "loss_yns_2": 0.15206, "loss_cls_3": 0.92756, "loss_box_3": 1.71026, "loss_cns_3": 0.65955, "loss_yns_3": 0.15366, "loss_cls_4": 0.92558, "loss_box_4": 1.71881, "loss_cns_4": 0.6509, "loss_yns_4": 0.15393, "loss_cls_5": 0.92671, "loss_box_5": 1.74507, "loss_cns_5": 0.64979, "loss_yns_5": 0.15368, "loss_cls_dn_0": 0.21063, "loss_box_dn_0": 0.78429, "loss_cls_dn_1": 0.1428, "loss_box_dn_1": 0.83536, "loss_cls_dn_2": 0.1426, "loss_box_dn_2": 0.859, "loss_cls_dn_3": 0.14654, "loss_box_dn_3": 0.8711, "loss_cls_dn_4": 0.15029, "loss_box_dn_4": 0.90016, "loss_cls_dn_5": 0.15465, "loss_box_dn_5": 0.93412, "loss_dense_depth": 0.73092, "loss": 27.56146, "grad_norm": 64.20471, "time": 1.57358} -{"mode": "train", "epoch": 1, "iter": 159, "lr": 0.00016, "memory": 49164, "data_time": 0.07352, "loss_cls_0": 0.81497, "loss_box_0": 1.74097, "loss_cns_0": 0.62214, "loss_yns_0": 0.15323, "loss_cls_1": 0.89976, "loss_box_1": 1.84443, "loss_cns_1": 0.62753, "loss_yns_1": 0.1546, "loss_cls_2": 0.94391, "loss_box_2": 1.76511, "loss_cns_2": 0.64912, "loss_yns_2": 0.15489, "loss_cls_3": 0.93042, "loss_box_3": 1.68259, "loss_cns_3": 0.66102, "loss_yns_3": 0.15853, "loss_cls_4": 0.93034, "loss_box_4": 1.68355, "loss_cns_4": 0.65386, "loss_yns_4": 0.15577, "loss_cls_5": 0.931, "loss_box_5": 1.71007, "loss_cns_5": 0.65324, "loss_yns_5": 0.1547, "loss_cls_dn_0": 0.21146, "loss_box_dn_0": 0.7697, "loss_cls_dn_1": 0.14393, "loss_box_dn_1": 0.84958, "loss_cls_dn_2": 0.14489, "loss_box_dn_2": 0.84255, "loss_cls_dn_3": 0.14653, "loss_box_dn_3": 0.84144, "loss_cls_dn_4": 0.14959, "loss_box_dn_4": 0.85668, "loss_cls_dn_5": 0.15381, "loss_box_dn_5": 0.88056, "loss_dense_depth": 0.7427, "loss": 27.40916, "grad_norm": 53.74581, "time": 1.57036} -{"mode": "train", "epoch": 1, "iter": 160, "lr": 0.00016, "memory": 49164, "data_time": 0.077, "loss_cls_0": 0.81316, "loss_box_0": 1.76105, "loss_cns_0": 0.61858, "loss_yns_0": 0.15358, "loss_cls_1": 0.88148, "loss_box_1": 1.75188, "loss_cns_1": 0.64297, "loss_yns_1": 0.15281, "loss_cls_2": 0.91324, "loss_box_2": 1.72611, "loss_cns_2": 0.65213, "loss_yns_2": 0.1519, "loss_cls_3": 0.92922, "loss_box_3": 1.71508, "loss_cns_3": 0.6554, "loss_yns_3": 0.15912, "loss_cls_4": 0.92722, "loss_box_4": 1.6926, "loss_cns_4": 0.65015, "loss_yns_4": 0.15313, "loss_cls_5": 0.93356, "loss_box_5": 1.67904, "loss_cns_5": 0.65142, "loss_yns_5": 0.15186, "loss_cls_dn_0": 0.21349, "loss_box_dn_0": 0.77841, "loss_cls_dn_1": 0.14313, "loss_box_dn_1": 0.7876, "loss_cls_dn_2": 0.14219, "loss_box_dn_2": 0.78849, "loss_cls_dn_3": 0.14546, "loss_box_dn_3": 0.80143, "loss_cls_dn_4": 0.14901, "loss_box_dn_4": 0.79433, "loss_cls_dn_5": 0.15455, "loss_box_dn_5": 0.79685, "loss_dense_depth": 0.72985, "loss": 26.94149, "grad_norm": 53.89881, "time": 1.59073} -{"mode": "train", "epoch": 1, "iter": 161, "lr": 0.00016, "memory": 49164, "data_time": 0.11089, "loss_cls_0": 0.81023, "loss_box_0": 1.78998, "loss_cns_0": 0.6133, "loss_yns_0": 0.15414, "loss_cls_1": 0.88071, "loss_box_1": 1.77506, "loss_cns_1": 0.64246, "loss_yns_1": 0.15299, "loss_cls_2": 0.91852, "loss_box_2": 1.78671, "loss_cns_2": 0.64238, "loss_yns_2": 0.15564, "loss_cls_3": 0.93056, "loss_box_3": 1.77035, "loss_cns_3": 0.64549, "loss_yns_3": 0.15865, "loss_cls_4": 0.93202, "loss_box_4": 1.72074, "loss_cns_4": 0.64742, "loss_yns_4": 0.15476, "loss_cls_5": 0.93011, "loss_box_5": 1.70793, "loss_cns_5": 0.64926, "loss_yns_5": 0.1555, "loss_cls_dn_0": 0.21112, "loss_box_dn_0": 0.79463, "loss_cls_dn_1": 0.13914, "loss_box_dn_1": 0.71356, "loss_cls_dn_2": 0.13913, "loss_box_dn_2": 0.7379, "loss_cls_dn_3": 0.14442, "loss_box_dn_3": 0.74468, "loss_cls_dn_4": 0.1482, "loss_box_dn_4": 0.72801, "loss_cls_dn_5": 0.15133, "loss_box_dn_5": 0.72845, "loss_dense_depth": 0.73012, "loss": 26.83563, "grad_norm": 59.05489, "time": 1.63506} -{"mode": "train", "epoch": 1, "iter": 162, "lr": 0.00016, "memory": 49164, "data_time": 0.10923, "loss_cls_0": 0.81756, "loss_box_0": 1.75118, "loss_cns_0": 0.62041, "loss_yns_0": 0.1554, "loss_cls_1": 0.89359, "loss_box_1": 1.80571, "loss_cns_1": 0.6365, "loss_yns_1": 0.1551, "loss_cls_2": 0.93204, "loss_box_2": 1.69698, "loss_cns_2": 0.65038, "loss_yns_2": 0.15435, "loss_cls_3": 0.92288, "loss_box_3": 1.69946, "loss_cns_3": 0.65556, "loss_yns_3": 0.15577, "loss_cls_4": 0.92566, "loss_box_4": 1.69909, "loss_cns_4": 0.64852, "loss_yns_4": 0.16045, "loss_cls_5": 0.93146, "loss_box_5": 1.71496, "loss_cns_5": 0.6546, "loss_yns_5": 0.15638, "loss_cls_dn_0": 0.21267, "loss_box_dn_0": 0.78267, "loss_cls_dn_1": 0.13496, "loss_box_dn_1": 0.73447, "loss_cls_dn_2": 0.13629, "loss_box_dn_2": 0.71863, "loss_cls_dn_3": 0.1383, "loss_box_dn_3": 0.73471, "loss_cls_dn_4": 0.14273, "loss_box_dn_4": 0.75347, "loss_cls_dn_5": 0.1534, "loss_box_dn_5": 0.77518, "loss_dense_depth": 0.71943, "loss": 26.73094, "grad_norm": 52.59198, "time": 1.61681} -{"mode": "train", "epoch": 1, "iter": 163, "lr": 0.00016, "memory": 49164, "data_time": 0.08716, "loss_cls_0": 0.80949, "loss_box_0": 1.7555, "loss_cns_0": 0.61924, "loss_yns_0": 0.15624, "loss_cls_1": 0.88399, "loss_box_1": 1.83447, "loss_cns_1": 0.63007, "loss_yns_1": 0.15649, "loss_cls_2": 0.90929, "loss_box_2": 1.70984, "loss_cns_2": 0.65166, "loss_yns_2": 0.15497, "loss_cls_3": 0.92053, "loss_box_3": 1.72716, "loss_cns_3": 0.65723, "loss_yns_3": 0.15892, "loss_cls_4": 0.92916, "loss_box_4": 1.76558, "loss_cns_4": 0.64538, "loss_yns_4": 0.16172, "loss_cls_5": 0.9315, "loss_box_5": 1.76879, "loss_cns_5": 0.64989, "loss_yns_5": 0.15545, "loss_cls_dn_0": 0.21215, "loss_box_dn_0": 0.76677, "loss_cls_dn_1": 0.13645, "loss_box_dn_1": 0.76609, "loss_cls_dn_2": 0.14089, "loss_box_dn_2": 0.74535, "loss_cls_dn_3": 0.14196, "loss_box_dn_3": 0.769, "loss_cls_dn_4": 0.14466, "loss_box_dn_4": 0.80991, "loss_cls_dn_5": 0.15479, "loss_box_dn_5": 0.83169, "loss_dense_depth": 0.71543, "loss": 27.0777, "grad_norm": 56.0871, "time": 1.58685} -{"mode": "train", "epoch": 1, "iter": 164, "lr": 0.00017, "memory": 49164, "data_time": 0.08117, "loss_cls_0": 0.83118, "loss_box_0": 1.78696, "loss_cns_0": 0.62231, "loss_yns_0": 0.15648, "loss_cls_1": 0.88637, "loss_box_1": 1.74423, "loss_cns_1": 0.65401, "loss_yns_1": 0.15796, "loss_cls_2": 0.91168, "loss_box_2": 1.68944, "loss_cns_2": 0.65984, "loss_yns_2": 0.15721, "loss_cls_3": 0.9219, "loss_box_3": 1.67783, "loss_cns_3": 0.67344, "loss_yns_3": 0.16138, "loss_cls_4": 0.91488, "loss_box_4": 1.70455, "loss_cns_4": 0.65667, "loss_yns_4": 0.15714, "loss_cls_5": 0.91824, "loss_box_5": 1.68713, "loss_cns_5": 0.65823, "loss_yns_5": 0.15633, "loss_cls_dn_0": 0.21313, "loss_box_dn_0": 0.77969, "loss_cls_dn_1": 0.13529, "loss_box_dn_1": 0.79326, "loss_cls_dn_2": 0.13726, "loss_box_dn_2": 0.7963, "loss_cls_dn_3": 0.14082, "loss_box_dn_3": 0.80488, "loss_cls_dn_4": 0.14303, "loss_box_dn_4": 0.83919, "loss_cls_dn_5": 0.15112, "loss_box_dn_5": 0.84924, "loss_dense_depth": 0.71578, "loss": 27.04438, "grad_norm": 46.96624, "time": 1.60122} -{"mode": "train", "epoch": 1, "iter": 165, "lr": 0.00017, "memory": 49164, "data_time": 0.07837, "loss_cls_0": 0.83031, "loss_box_0": 1.79546, "loss_cns_0": 0.61692, "loss_yns_0": 0.15773, "loss_cls_1": 0.89918, "loss_box_1": 1.72615, "loss_cns_1": 0.65129, "loss_yns_1": 0.15798, "loss_cls_2": 0.91281, "loss_box_2": 1.73006, "loss_cns_2": 0.64881, "loss_yns_2": 0.15733, "loss_cls_3": 0.91573, "loss_box_3": 1.70643, "loss_cns_3": 0.65733, "loss_yns_3": 0.16165, "loss_cls_4": 0.92748, "loss_box_4": 1.72713, "loss_cns_4": 0.64952, "loss_yns_4": 0.15947, "loss_cls_5": 0.96618, "loss_box_5": 1.73611, "loss_cns_5": 0.65094, "loss_yns_5": 0.16017, "loss_cls_dn_0": 0.2153, "loss_box_dn_0": 0.77537, "loss_cls_dn_1": 0.13665, "loss_box_dn_1": 0.7845, "loss_cls_dn_2": 0.13649, "loss_box_dn_2": 0.79589, "loss_cls_dn_3": 0.14089, "loss_box_dn_3": 0.7934, "loss_cls_dn_4": 0.14308, "loss_box_dn_4": 0.81845, "loss_cls_dn_5": 0.15384, "loss_box_dn_5": 0.83191, "loss_dense_depth": 0.73104, "loss": 27.15897, "grad_norm": 58.26701, "time": 1.59098} -{"mode": "train", "epoch": 1, "iter": 166, "lr": 0.00017, "memory": 49164, "data_time": 0.10329, "loss_cls_0": 0.80221, "loss_box_0": 1.77653, "loss_cns_0": 0.61384, "loss_yns_0": 0.15805, "loss_cls_1": 0.91398, "loss_box_1": 1.68742, "loss_cns_1": 0.65362, "loss_yns_1": 0.16174, "loss_cls_2": 0.90941, "loss_box_2": 1.68126, "loss_cns_2": 0.6544, "loss_yns_2": 0.15964, "loss_cls_3": 0.93771, "loss_box_3": 1.64558, "loss_cns_3": 0.65865, "loss_yns_3": 0.16065, "loss_cls_4": 0.92798, "loss_box_4": 1.65207, "loss_cns_4": 0.65756, "loss_yns_4": 0.16279, "loss_cls_5": 0.92758, "loss_box_5": 1.66078, "loss_cns_5": 0.66488, "loss_yns_5": 0.1599, "loss_cls_dn_0": 0.20573, "loss_box_dn_0": 0.75955, "loss_cls_dn_1": 0.13641, "loss_box_dn_1": 0.78358, "loss_cls_dn_2": 0.133, "loss_box_dn_2": 0.77783, "loss_cls_dn_3": 0.13658, "loss_box_dn_3": 0.76368, "loss_cls_dn_4": 0.13766, "loss_box_dn_4": 0.77431, "loss_cls_dn_5": 0.14243, "loss_box_dn_5": 0.78346, "loss_dense_depth": 0.7321, "loss": 26.65457, "grad_norm": 31.3539, "time": 1.59323} -{"mode": "train", "epoch": 1, "iter": 167, "lr": 0.00017, "memory": 49164, "data_time": 0.07415, "loss_cls_0": 0.84728, "loss_box_0": 1.80342, "loss_cns_0": 0.61009, "loss_yns_0": 0.16071, "loss_cls_1": 0.93457, "loss_box_1": 1.69041, "loss_cns_1": 0.65205, "loss_yns_1": 0.15928, "loss_cls_2": 0.95531, "loss_box_2": 1.68959, "loss_cns_2": 0.65184, "loss_yns_2": 0.15971, "loss_cls_3": 0.96071, "loss_box_3": 1.67734, "loss_cns_3": 0.65809, "loss_yns_3": 0.16031, "loss_cls_4": 0.96168, "loss_box_4": 1.65744, "loss_cns_4": 0.65312, "loss_yns_4": 0.16095, "loss_cls_5": 0.97113, "loss_box_5": 1.65361, "loss_cns_5": 0.65498, "loss_yns_5": 0.16001, "loss_cls_dn_0": 0.21838, "loss_box_dn_0": 0.7855, "loss_cls_dn_1": 0.13716, "loss_box_dn_1": 0.78181, "loss_cls_dn_2": 0.13789, "loss_box_dn_2": 0.78225, "loss_cls_dn_3": 0.14434, "loss_box_dn_3": 0.77776, "loss_cls_dn_4": 0.14964, "loss_box_dn_4": 0.78049, "loss_cls_dn_5": 0.15267, "loss_box_dn_5": 0.77704, "loss_dense_depth": 0.81533, "loss": 27.08391, "grad_norm": 50.12139, "time": 6.79762} -{"mode": "train", "epoch": 1, "iter": 168, "lr": 0.00017, "memory": 49164, "data_time": 0.07594, "loss_cls_0": 0.83374, "loss_box_0": 1.74759, "loss_cns_0": 0.61867, "loss_yns_0": 0.16018, "loss_cls_1": 0.91501, "loss_box_1": 1.65366, "loss_cns_1": 0.65146, "loss_yns_1": 0.16045, "loss_cls_2": 0.9451, "loss_box_2": 1.67867, "loss_cns_2": 0.65048, "loss_yns_2": 0.16112, "loss_cls_3": 0.96411, "loss_box_3": 1.65895, "loss_cns_3": 0.66713, "loss_yns_3": 0.16209, "loss_cls_4": 0.96291, "loss_box_4": 1.64211, "loss_cns_4": 0.65301, "loss_yns_4": 0.16126, "loss_cls_5": 0.98699, "loss_box_5": 1.64182, "loss_cns_5": 0.6519, "loss_yns_5": 0.16082, "loss_cls_dn_0": 0.21116, "loss_box_dn_0": 0.77276, "loss_cls_dn_1": 0.1361, "loss_box_dn_1": 0.72737, "loss_cls_dn_2": 0.13858, "loss_box_dn_2": 0.75267, "loss_cls_dn_3": 0.14977, "loss_box_dn_3": 0.75211, "loss_cls_dn_4": 0.14846, "loss_box_dn_4": 0.76686, "loss_cls_dn_5": 0.15382, "loss_box_dn_5": 0.77109, "loss_dense_depth": 0.77471, "loss": 26.74466, "grad_norm": 54.51274, "time": 1.54684} -{"mode": "train", "epoch": 1, "iter": 169, "lr": 0.00017, "memory": 49164, "data_time": 0.07687, "loss_cls_0": 0.87567, "loss_box_0": 1.80045, "loss_cns_0": 0.61057, "loss_yns_0": 0.15991, "loss_cls_1": 0.94413, "loss_box_1": 1.71403, "loss_cns_1": 0.64465, "loss_yns_1": 0.16092, "loss_cls_2": 0.96545, "loss_box_2": 1.66603, "loss_cns_2": 0.65016, "loss_yns_2": 0.16042, "loss_cls_3": 0.98413, "loss_box_3": 1.66627, "loss_cns_3": 0.66231, "loss_yns_3": 0.16276, "loss_cls_4": 0.96742, "loss_box_4": 1.66776, "loss_cns_4": 0.6493, "loss_yns_4": 0.16079, "loss_cls_5": 0.99985, "loss_box_5": 1.67455, "loss_cns_5": 0.6559, "loss_yns_5": 0.16102, "loss_cls_dn_0": 0.21983, "loss_box_dn_0": 0.78446, "loss_cls_dn_1": 0.13663, "loss_box_dn_1": 0.74244, "loss_cls_dn_2": 0.14075, "loss_box_dn_2": 0.74301, "loss_cls_dn_3": 0.14519, "loss_box_dn_3": 0.75522, "loss_cls_dn_4": 0.14446, "loss_box_dn_4": 0.77891, "loss_cls_dn_5": 0.15211, "loss_box_dn_5": 0.79673, "loss_dense_depth": 0.81924, "loss": 27.12343, "grad_norm": 33.48582, "time": 1.59677} -{"mode": "train", "epoch": 1, "iter": 170, "lr": 0.00017, "memory": 49164, "data_time": 0.08041, "loss_cls_0": 0.85596, "loss_box_0": 1.76157, "loss_cns_0": 0.61718, "loss_yns_0": 0.15951, "loss_cls_1": 0.94649, "loss_box_1": 1.72981, "loss_cns_1": 0.64797, "loss_yns_1": 0.15943, "loss_cls_2": 0.98128, "loss_box_2": 1.64845, "loss_cns_2": 0.65927, "loss_yns_2": 0.15961, "loss_cls_3": 0.96697, "loss_box_3": 1.63146, "loss_cns_3": 0.65894, "loss_yns_3": 0.16604, "loss_cls_4": 0.99978, "loss_box_4": 1.61868, "loss_cns_4": 0.65647, "loss_yns_4": 0.15904, "loss_cls_5": 0.97315, "loss_box_5": 1.62439, "loss_cns_5": 0.65867, "loss_yns_5": 0.15899, "loss_cls_dn_0": 0.21492, "loss_box_dn_0": 0.77534, "loss_cls_dn_1": 0.13651, "loss_box_dn_1": 0.78237, "loss_cls_dn_2": 0.1414, "loss_box_dn_2": 0.77784, "loss_cls_dn_3": 0.14203, "loss_box_dn_3": 0.79266, "loss_cls_dn_4": 0.14874, "loss_box_dn_4": 0.809, "loss_cls_dn_5": 0.15745, "loss_box_dn_5": 0.83265, "loss_dense_depth": 0.82617, "loss": 27.13619, "grad_norm": 55.29721, "time": 1.5742} -{"mode": "train", "epoch": 1, "iter": 171, "lr": 0.00017, "memory": 49164, "data_time": 0.07194, "loss_cls_0": 0.84428, "loss_box_0": 1.70906, "loss_cns_0": 0.61528, "loss_yns_0": 0.15716, "loss_cls_1": 0.92454, "loss_box_1": 1.70736, "loss_cns_1": 0.64587, "loss_yns_1": 0.15758, "loss_cls_2": 0.95272, "loss_box_2": 1.66211, "loss_cns_2": 0.65614, "loss_yns_2": 0.15835, "loss_cls_3": 0.96619, "loss_box_3": 1.64299, "loss_cns_3": 0.65943, "loss_yns_3": 0.16033, "loss_cls_4": 0.96973, "loss_box_4": 1.63026, "loss_cns_4": 0.6583, "loss_yns_4": 0.15979, "loss_cls_5": 0.96692, "loss_box_5": 1.63467, "loss_cns_5": 0.65884, "loss_yns_5": 0.15805, "loss_cls_dn_0": 0.21416, "loss_box_dn_0": 0.76188, "loss_cls_dn_1": 0.13732, "loss_box_dn_1": 0.78218, "loss_cls_dn_2": 0.13969, "loss_box_dn_2": 0.78349, "loss_cls_dn_3": 0.1429, "loss_box_dn_3": 0.78937, "loss_cls_dn_4": 0.14576, "loss_box_dn_4": 0.79771, "loss_cls_dn_5": 0.14826, "loss_box_dn_5": 0.81418, "loss_dense_depth": 0.74681, "loss": 26.85969, "grad_norm": 37.34747, "time": 1.58776} -{"mode": "train", "epoch": 1, "iter": 172, "lr": 0.00017, "memory": 49164, "data_time": 0.08272, "loss_cls_0": 0.87355, "loss_box_0": 1.75408, "loss_cns_0": 0.60882, "loss_yns_0": 0.15826, "loss_cls_1": 0.938, "loss_box_1": 1.67, "loss_cns_1": 0.64788, "loss_yns_1": 0.15798, "loss_cls_2": 0.95757, "loss_box_2": 1.67549, "loss_cns_2": 0.65176, "loss_yns_2": 0.15917, "loss_cls_3": 0.97276, "loss_box_3": 1.64462, "loss_cns_3": 0.66566, "loss_yns_3": 0.15957, "loss_cls_4": 0.99136, "loss_box_4": 1.64047, "loss_cns_4": 0.6561, "loss_yns_4": 0.16037, "loss_cls_5": 0.97612, "loss_box_5": 1.65254, "loss_cns_5": 0.65753, "loss_yns_5": 0.15914, "loss_cls_dn_0": 0.21902, "loss_box_dn_0": 0.76122, "loss_cls_dn_1": 0.13548, "loss_box_dn_1": 0.75467, "loss_cls_dn_2": 0.1377, "loss_box_dn_2": 0.75915, "loss_cls_dn_3": 0.14161, "loss_box_dn_3": 0.74811, "loss_cls_dn_4": 0.14778, "loss_box_dn_4": 0.75694, "loss_cls_dn_5": 0.14766, "loss_box_dn_5": 0.76416, "loss_dense_depth": 0.83562, "loss": 26.89793, "grad_norm": 38.44796, "time": 1.57568} -{"mode": "train", "epoch": 1, "iter": 173, "lr": 0.00017, "memory": 49164, "data_time": 0.08065, "loss_cls_0": 0.84334, "loss_box_0": 1.74498, "loss_cns_0": 0.61802, "loss_yns_0": 0.15828, "loss_cls_1": 0.91474, "loss_box_1": 1.68701, "loss_cns_1": 0.64867, "loss_yns_1": 0.15613, "loss_cls_2": 0.94428, "loss_box_2": 1.6562, "loss_cns_2": 0.65484, "loss_yns_2": 0.15728, "loss_cls_3": 0.95482, "loss_box_3": 1.64191, "loss_cns_3": 0.66389, "loss_yns_3": 0.15906, "loss_cls_4": 0.97031, "loss_box_4": 1.63248, "loss_cns_4": 0.65794, "loss_yns_4": 0.15929, "loss_cls_5": 0.96305, "loss_box_5": 1.62971, "loss_cns_5": 0.66044, "loss_yns_5": 0.1571, "loss_cls_dn_0": 0.2156, "loss_box_dn_0": 0.76208, "loss_cls_dn_1": 0.13708, "loss_box_dn_1": 0.7437, "loss_cls_dn_2": 0.13736, "loss_box_dn_2": 0.73833, "loss_cls_dn_3": 0.13969, "loss_box_dn_3": 0.72728, "loss_cls_dn_4": 0.14468, "loss_box_dn_4": 0.73206, "loss_cls_dn_5": 0.14501, "loss_box_dn_5": 0.73536, "loss_dense_depth": 0.80006, "loss": 26.59202, "grad_norm": 37.21786, "time": 1.56006} -{"mode": "train", "epoch": 1, "iter": 174, "lr": 0.00017, "memory": 49164, "data_time": 0.08163, "loss_cls_0": 0.85187, "loss_box_0": 1.7489, "loss_cns_0": 0.61795, "loss_yns_0": 0.15631, "loss_cls_1": 0.91274, "loss_box_1": 1.74261, "loss_cns_1": 0.64447, "loss_yns_1": 0.1549, "loss_cls_2": 0.9421, "loss_box_2": 1.6979, "loss_cns_2": 0.65216, "loss_yns_2": 0.15567, "loss_cls_3": 0.95454, "loss_box_3": 1.71663, "loss_cns_3": 0.65141, "loss_yns_3": 0.1587, "loss_cls_4": 0.95616, "loss_box_4": 1.71056, "loss_cns_4": 0.6531, "loss_yns_4": 0.15806, "loss_cls_5": 0.96148, "loss_box_5": 1.70673, "loss_cns_5": 0.64888, "loss_yns_5": 0.15607, "loss_cls_dn_0": 0.21297, "loss_box_dn_0": 0.76781, "loss_cls_dn_1": 0.13779, "loss_box_dn_1": 0.74649, "loss_cls_dn_2": 0.13736, "loss_box_dn_2": 0.73599, "loss_cls_dn_3": 0.141, "loss_box_dn_3": 0.74711, "loss_cls_dn_4": 0.14421, "loss_box_dn_4": 0.7515, "loss_cls_dn_5": 0.14549, "loss_box_dn_5": 0.76328, "loss_dense_depth": 0.83698, "loss": 26.97786, "grad_norm": 43.14055, "time": 1.57964} -{"mode": "train", "epoch": 1, "iter": 175, "lr": 0.00017, "memory": 49164, "data_time": 0.07333, "loss_cls_0": 0.8807, "loss_box_0": 1.73116, "loss_cns_0": 0.62331, "loss_yns_0": 0.15713, "loss_cls_1": 0.91726, "loss_box_1": 1.70074, "loss_cns_1": 0.64856, "loss_yns_1": 0.1539, "loss_cls_2": 0.94543, "loss_box_2": 1.66804, "loss_cns_2": 0.6539, "loss_yns_2": 0.15398, "loss_cls_3": 0.9553, "loss_box_3": 1.67824, "loss_cns_3": 0.65765, "loss_yns_3": 0.15822, "loss_cls_4": 0.97453, "loss_box_4": 1.67431, "loss_cns_4": 0.6594, "loss_yns_4": 0.15539, "loss_cls_5": 0.95947, "loss_box_5": 1.67662, "loss_cns_5": 0.65443, "loss_yns_5": 0.15475, "loss_cls_dn_0": 0.21679, "loss_box_dn_0": 0.7597, "loss_cls_dn_1": 0.13489, "loss_box_dn_1": 0.74365, "loss_cls_dn_2": 0.13251, "loss_box_dn_2": 0.7446, "loss_cls_dn_3": 0.13754, "loss_box_dn_3": 0.76685, "loss_cls_dn_4": 0.14284, "loss_box_dn_4": 0.78442, "loss_cls_dn_5": 0.14498, "loss_box_dn_5": 0.81032, "loss_dense_depth": 0.79698, "loss": 26.90845, "grad_norm": 40.46447, "time": 1.56902} -{"mode": "train", "epoch": 1, "iter": 176, "lr": 0.00017, "memory": 49164, "data_time": 0.07554, "loss_cls_0": 0.82667, "loss_box_0": 1.7106, "loss_cns_0": 0.61989, "loss_yns_0": 0.15396, "loss_cls_1": 0.90723, "loss_box_1": 1.69659, "loss_cns_1": 0.64794, "loss_yns_1": 0.15265, "loss_cls_2": 0.94545, "loss_box_2": 1.66863, "loss_cns_2": 0.65414, "loss_yns_2": 0.15021, "loss_cls_3": 0.94502, "loss_box_3": 1.71817, "loss_cns_3": 0.65408, "loss_yns_3": 0.1535, "loss_cls_4": 0.95601, "loss_box_4": 1.69862, "loss_cns_4": 0.6531, "loss_yns_4": 0.15298, "loss_cls_5": 0.95069, "loss_box_5": 1.70959, "loss_cns_5": 0.64962, "loss_yns_5": 0.15306, "loss_cls_dn_0": 0.21705, "loss_box_dn_0": 0.76864, "loss_cls_dn_1": 0.13775, "loss_box_dn_1": 0.76475, "loss_cls_dn_2": 0.13736, "loss_box_dn_2": 0.77055, "loss_cls_dn_3": 0.14176, "loss_box_dn_3": 0.80427, "loss_cls_dn_4": 0.14359, "loss_box_dn_4": 0.82273, "loss_cls_dn_5": 0.14687, "loss_box_dn_5": 0.85099, "loss_dense_depth": 0.81778, "loss": 27.0525, "grad_norm": 56.5117, "time": 1.57272} -{"mode": "train", "epoch": 1, "iter": 177, "lr": 0.00017, "memory": 49164, "data_time": 0.07768, "loss_cls_0": 0.82834, "loss_box_0": 1.68853, "loss_cns_0": 0.6154, "loss_yns_0": 0.15041, "loss_cls_1": 0.91106, "loss_box_1": 1.66847, "loss_cns_1": 0.64893, "loss_yns_1": 0.14987, "loss_cls_2": 0.93073, "loss_box_2": 1.63215, "loss_cns_2": 0.65512, "loss_yns_2": 0.14952, "loss_cls_3": 0.93063, "loss_box_3": 1.65057, "loss_cns_3": 0.65309, "loss_yns_3": 0.15026, "loss_cls_4": 0.94239, "loss_box_4": 1.6503, "loss_cns_4": 0.65524, "loss_yns_4": 0.15065, "loss_cls_5": 0.94116, "loss_box_5": 1.64768, "loss_cns_5": 0.65444, "loss_yns_5": 0.15124, "loss_cls_dn_0": 0.21797, "loss_box_dn_0": 0.75908, "loss_cls_dn_1": 0.13912, "loss_box_dn_1": 0.80035, "loss_cls_dn_2": 0.13724, "loss_box_dn_2": 0.80017, "loss_cls_dn_3": 0.14003, "loss_box_dn_3": 0.8223, "loss_cls_dn_4": 0.14447, "loss_box_dn_4": 0.84286, "loss_cls_dn_5": 0.14647, "loss_box_dn_5": 0.86221, "loss_dense_depth": 0.85298, "loss": 26.87144, "grad_norm": 41.58554, "time": 1.56235} -{"mode": "train", "epoch": 1, "iter": 178, "lr": 0.00017, "memory": 49164, "data_time": 0.07901, "loss_cls_0": 0.83779, "loss_box_0": 1.6886, "loss_cns_0": 0.60166, "loss_yns_0": 0.14522, "loss_cls_1": 0.91323, "loss_box_1": 1.6967, "loss_cns_1": 0.64895, "loss_yns_1": 0.14897, "loss_cls_2": 0.92853, "loss_box_2": 1.69439, "loss_cns_2": 0.65158, "loss_yns_2": 0.14874, "loss_cls_3": 0.94006, "loss_box_3": 1.66518, "loss_cns_3": 0.65435, "loss_yns_3": 0.14817, "loss_cls_4": 0.93944, "loss_box_4": 1.69165, "loss_cns_4": 0.6574, "loss_yns_4": 0.14881, "loss_cls_5": 0.94155, "loss_box_5": 1.69925, "loss_cns_5": 0.65557, "loss_yns_5": 0.15032, "loss_cls_dn_0": 0.21747, "loss_box_dn_0": 0.76908, "loss_cls_dn_1": 0.13935, "loss_box_dn_1": 0.80874, "loss_cls_dn_2": 0.13766, "loss_box_dn_2": 0.8189, "loss_cls_dn_3": 0.14174, "loss_box_dn_3": 0.82082, "loss_cls_dn_4": 0.1469, "loss_box_dn_4": 0.84621, "loss_cls_dn_5": 0.15121, "loss_box_dn_5": 0.85811, "loss_dense_depth": 0.78172, "loss": 27.03402, "grad_norm": 49.98172, "time": 1.56369} -{"mode": "train", "epoch": 1, "iter": 179, "lr": 0.00017, "memory": 49164, "data_time": 0.07543, "loss_cls_0": 0.82799, "loss_box_0": 1.69774, "loss_cns_0": 0.60735, "loss_yns_0": 0.14617, "loss_cls_1": 0.91262, "loss_box_1": 1.70172, "loss_cns_1": 0.64773, "loss_yns_1": 0.15063, "loss_cls_2": 0.94329, "loss_box_2": 1.70039, "loss_cns_2": 0.64911, "loss_yns_2": 0.15038, "loss_cls_3": 0.94981, "loss_box_3": 1.6672, "loss_cns_3": 0.65285, "loss_yns_3": 0.15109, "loss_cls_4": 0.94894, "loss_box_4": 1.66592, "loss_cns_4": 0.65429, "loss_yns_4": 0.1491, "loss_cls_5": 0.94943, "loss_box_5": 1.66106, "loss_cns_5": 0.6521, "loss_yns_5": 0.15094, "loss_cls_dn_0": 0.21707, "loss_box_dn_0": 0.77898, "loss_cls_dn_1": 0.14226, "loss_box_dn_1": 0.80409, "loss_cls_dn_2": 0.13896, "loss_box_dn_2": 0.80872, "loss_cls_dn_3": 0.14118, "loss_box_dn_3": 0.79624, "loss_cls_dn_4": 0.14467, "loss_box_dn_4": 0.80523, "loss_cls_dn_5": 0.14917, "loss_box_dn_5": 0.80359, "loss_dense_depth": 0.80481, "loss": 26.92282, "grad_norm": 47.76125, "time": 1.55956} -{"mode": "train", "epoch": 1, "iter": 180, "lr": 0.00017, "memory": 49164, "data_time": 0.07372, "loss_cls_0": 0.80893, "loss_box_0": 1.68986, "loss_cns_0": 0.61952, "loss_yns_0": 0.1465, "loss_cls_1": 0.91202, "loss_box_1": 1.67594, "loss_cns_1": 0.65108, "loss_yns_1": 0.15053, "loss_cls_2": 0.93236, "loss_box_2": 1.63925, "loss_cns_2": 0.65492, "loss_yns_2": 0.14914, "loss_cls_3": 0.93844, "loss_box_3": 1.63619, "loss_cns_3": 0.65664, "loss_yns_3": 0.1491, "loss_cls_4": 0.94618, "loss_box_4": 1.62373, "loss_cns_4": 0.65787, "loss_yns_4": 0.15042, "loss_cls_5": 0.94469, "loss_box_5": 1.62771, "loss_cns_5": 0.65594, "loss_yns_5": 0.14792, "loss_cls_dn_0": 0.21481, "loss_box_dn_0": 0.77569, "loss_cls_dn_1": 0.13778, "loss_box_dn_1": 0.78537, "loss_cls_dn_2": 0.13586, "loss_box_dn_2": 0.77522, "loss_cls_dn_3": 0.13703, "loss_box_dn_3": 0.76639, "loss_cls_dn_4": 0.13935, "loss_box_dn_4": 0.76972, "loss_cls_dn_5": 0.14283, "loss_box_dn_5": 0.7701, "loss_dense_depth": 0.8001, "loss": 26.51513, "grad_norm": 43.94125, "time": 1.57796} -{"mode": "train", "epoch": 1, "iter": 181, "lr": 0.00017, "memory": 49164, "data_time": 0.1011, "loss_cls_0": 0.81212, "loss_box_0": 1.72402, "loss_cns_0": 0.61881, "loss_yns_0": 0.14935, "loss_cls_1": 0.90011, "loss_box_1": 1.69856, "loss_cns_1": 0.64579, "loss_yns_1": 0.14965, "loss_cls_2": 0.92325, "loss_box_2": 1.6365, "loss_cns_2": 0.65004, "loss_yns_2": 0.15203, "loss_cls_3": 0.92607, "loss_box_3": 1.65867, "loss_cns_3": 0.65155, "loss_yns_3": 0.1502, "loss_cls_4": 0.93203, "loss_box_4": 1.65125, "loss_cns_4": 0.65514, "loss_yns_4": 0.15191, "loss_cls_5": 0.93474, "loss_box_5": 1.66776, "loss_cns_5": 0.6584, "loss_yns_5": 0.15124, "loss_cls_dn_0": 0.21549, "loss_box_dn_0": 0.77168, "loss_cls_dn_1": 0.13089, "loss_box_dn_1": 0.75651, "loss_cls_dn_2": 0.13283, "loss_box_dn_2": 0.7467, "loss_cls_dn_3": 0.13444, "loss_box_dn_3": 0.76128, "loss_cls_dn_4": 0.13995, "loss_box_dn_4": 0.77576, "loss_cls_dn_5": 0.14735, "loss_box_dn_5": 0.79884, "loss_dense_depth": 0.78965, "loss": 26.55055, "grad_norm": 43.39682, "time": 1.64358} -{"mode": "train", "epoch": 1, "iter": 182, "lr": 0.00017, "memory": 49164, "data_time": 0.10395, "loss_cls_0": 0.80257, "loss_box_0": 1.69679, "loss_cns_0": 0.62481, "loss_yns_0": 0.14669, "loss_cls_1": 0.89159, "loss_box_1": 1.671, "loss_cns_1": 0.65287, "loss_yns_1": 0.14666, "loss_cls_2": 0.91003, "loss_box_2": 1.65496, "loss_cns_2": 0.65135, "loss_yns_2": 0.14937, "loss_cls_3": 0.91197, "loss_box_3": 1.67672, "loss_cns_3": 0.65192, "loss_yns_3": 0.14894, "loss_cls_4": 0.92867, "loss_box_4": 1.68042, "loss_cns_4": 0.65699, "loss_yns_4": 0.15006, "loss_cls_5": 0.91609, "loss_box_5": 1.68348, "loss_cns_5": 0.65291, "loss_yns_5": 0.14841, "loss_cls_dn_0": 0.2133, "loss_box_dn_0": 0.77572, "loss_cls_dn_1": 0.13228, "loss_box_dn_1": 0.78246, "loss_cls_dn_2": 0.13962, "loss_box_dn_2": 0.79138, "loss_cls_dn_3": 0.1429, "loss_box_dn_3": 0.82295, "loss_cls_dn_4": 0.14649, "loss_box_dn_4": 0.8454, "loss_cls_dn_5": 0.15757, "loss_box_dn_5": 0.87657, "loss_dense_depth": 0.76777, "loss": 26.79972, "grad_norm": 57.15733, "time": 1.61367} -{"mode": "train", "epoch": 1, "iter": 183, "lr": 0.00017, "memory": 49164, "data_time": 0.07806, "loss_cls_0": 0.78894, "loss_box_0": 1.68763, "loss_cns_0": 0.62819, "loss_yns_0": 0.14659, "loss_cls_1": 0.88996, "loss_box_1": 1.60926, "loss_cns_1": 0.66204, "loss_yns_1": 0.14642, "loss_cls_2": 0.90357, "loss_box_2": 1.58948, "loss_cns_2": 0.65711, "loss_yns_2": 0.14538, "loss_cls_3": 0.91693, "loss_box_3": 1.6156, "loss_cns_3": 0.65852, "loss_yns_3": 0.1471, "loss_cls_4": 0.91172, "loss_box_4": 1.62366, "loss_cns_4": 0.65661, "loss_yns_4": 0.14698, "loss_cls_5": 0.91153, "loss_box_5": 1.6332, "loss_cns_5": 0.65392, "loss_yns_5": 0.14683, "loss_cls_dn_0": 0.21362, "loss_box_dn_0": 0.76269, "loss_cls_dn_1": 0.13447, "loss_box_dn_1": 0.79092, "loss_cls_dn_2": 0.13036, "loss_box_dn_2": 0.79871, "loss_cls_dn_3": 0.13359, "loss_box_dn_3": 0.83639, "loss_cls_dn_4": 0.13682, "loss_box_dn_4": 0.85878, "loss_cls_dn_5": 0.14102, "loss_box_dn_5": 0.89309, "loss_dense_depth": 0.78266, "loss": 26.49029, "grad_norm": 56.62438, "time": 1.56675} -{"mode": "train", "epoch": 1, "iter": 184, "lr": 0.00017, "memory": 49164, "data_time": 0.08395, "loss_cls_0": 0.81026, "loss_box_0": 1.71462, "loss_cns_0": 0.62421, "loss_yns_0": 0.14445, "loss_cls_1": 0.90857, "loss_box_1": 1.61926, "loss_cns_1": 0.66006, "loss_yns_1": 0.14712, "loss_cls_2": 0.91331, "loss_box_2": 1.60473, "loss_cns_2": 0.65659, "loss_yns_2": 0.14598, "loss_cls_3": 0.92686, "loss_box_3": 1.61195, "loss_cns_3": 0.65695, "loss_yns_3": 0.14499, "loss_cls_4": 0.91926, "loss_box_4": 1.60891, "loss_cns_4": 0.6579, "loss_yns_4": 0.14751, "loss_cls_5": 0.92707, "loss_box_5": 1.62952, "loss_cns_5": 0.6589, "loss_yns_5": 0.1461, "loss_cls_dn_0": 0.21763, "loss_box_dn_0": 0.76776, "loss_cls_dn_1": 0.13481, "loss_box_dn_1": 0.80744, "loss_cls_dn_2": 0.13033, "loss_box_dn_2": 0.81787, "loss_cls_dn_3": 0.13867, "loss_box_dn_3": 0.85234, "loss_cls_dn_4": 0.1383, "loss_box_dn_4": 0.87018, "loss_cls_dn_5": 0.14179, "loss_box_dn_5": 0.90361, "loss_dense_depth": 0.78285, "loss": 26.68867, "grad_norm": 44.50948, "time": 1.58528} -{"mode": "train", "epoch": 1, "iter": 185, "lr": 0.00017, "memory": 49164, "data_time": 0.07912, "loss_cls_0": 0.80794, "loss_box_0": 1.70261, "loss_cns_0": 0.6233, "loss_yns_0": 0.1456, "loss_cls_1": 0.89495, "loss_box_1": 1.61177, "loss_cns_1": 0.65816, "loss_yns_1": 0.14805, "loss_cls_2": 0.90221, "loss_box_2": 1.61353, "loss_cns_2": 0.65752, "loss_yns_2": 0.1486, "loss_cls_3": 0.91642, "loss_box_3": 1.60552, "loss_cns_3": 0.65954, "loss_yns_3": 0.1465, "loss_cls_4": 0.9197, "loss_box_4": 1.59412, "loss_cns_4": 0.66089, "loss_yns_4": 0.15093, "loss_cls_5": 0.92329, "loss_box_5": 1.5977, "loss_cns_5": 0.66274, "loss_yns_5": 0.14778, "loss_cls_dn_0": 0.21795, "loss_box_dn_0": 0.76638, "loss_cls_dn_1": 0.13185, "loss_box_dn_1": 0.8453, "loss_cls_dn_2": 0.13284, "loss_box_dn_2": 0.85411, "loss_cls_dn_3": 0.14275, "loss_box_dn_3": 0.86577, "loss_cls_dn_4": 0.14554, "loss_box_dn_4": 0.87288, "loss_cls_dn_5": 0.15161, "loss_box_dn_5": 0.89179, "loss_dense_depth": 0.80244, "loss": 26.7206, "grad_norm": 59.55862, "time": 1.60224} -{"mode": "train", "epoch": 1, "iter": 186, "lr": 0.00017, "memory": 49164, "data_time": 0.09835, "loss_cls_0": 0.80241, "loss_box_0": 1.67891, "loss_cns_0": 0.62735, "loss_yns_0": 0.14766, "loss_cls_1": 0.89341, "loss_box_1": 1.61051, "loss_cns_1": 0.66032, "loss_yns_1": 0.15065, "loss_cls_2": 0.90441, "loss_box_2": 1.59809, "loss_cns_2": 0.65846, "loss_yns_2": 0.15039, "loss_cls_3": 0.92411, "loss_box_3": 1.57459, "loss_cns_3": 0.663, "loss_yns_3": 0.14738, "loss_cls_4": 0.91929, "loss_box_4": 1.56431, "loss_cns_4": 0.66494, "loss_yns_4": 0.15114, "loss_cls_5": 0.91738, "loss_box_5": 1.57332, "loss_cns_5": 0.6658, "loss_yns_5": 0.15055, "loss_cls_dn_0": 0.21292, "loss_box_dn_0": 0.76436, "loss_cls_dn_1": 0.13491, "loss_box_dn_1": 0.78776, "loss_cls_dn_2": 0.13809, "loss_box_dn_2": 0.78133, "loss_cls_dn_3": 0.14082, "loss_box_dn_3": 0.77262, "loss_cls_dn_4": 0.14259, "loss_box_dn_4": 0.77495, "loss_cls_dn_5": 0.14758, "loss_box_dn_5": 0.78701, "loss_dense_depth": 0.78915, "loss": 26.17245, "grad_norm": 47.86999, "time": 3.07081} -{"mode": "train", "epoch": 1, "iter": 187, "lr": 0.00017, "memory": 49164, "data_time": 0.04435, "loss_cls_0": 0.81796, "loss_box_0": 1.69089, "loss_cns_0": 0.62278, "loss_yns_0": 0.14811, "loss_cls_1": 0.91415, "loss_box_1": 1.62847, "loss_cns_1": 0.65898, "loss_yns_1": 0.14858, "loss_cls_2": 0.92863, "loss_box_2": 1.60166, "loss_cns_2": 0.65979, "loss_yns_2": 0.14955, "loss_cls_3": 0.93826, "loss_box_3": 1.62471, "loss_cns_3": 0.65932, "loss_yns_3": 0.14842, "loss_cls_4": 0.94688, "loss_box_4": 1.59632, "loss_cns_4": 0.66323, "loss_yns_4": 0.14937, "loss_cls_5": 0.94235, "loss_box_5": 1.5953, "loss_cns_5": 0.65982, "loss_yns_5": 0.15416, "loss_cls_dn_0": 0.2167, "loss_box_dn_0": 0.77333, "loss_cls_dn_1": 0.13318, "loss_box_dn_1": 0.76151, "loss_cls_dn_2": 0.136, "loss_box_dn_2": 0.74954, "loss_cls_dn_3": 0.13528, "loss_box_dn_3": 0.75724, "loss_cls_dn_4": 0.14029, "loss_box_dn_4": 0.75098, "loss_cls_dn_5": 0.14382, "loss_box_dn_5": 0.75607, "loss_dense_depth": 0.77324, "loss": 26.2749, "grad_norm": 46.55399, "time": 1.51708} -{"mode": "train", "epoch": 1, "iter": 188, "lr": 0.00017, "memory": 49164, "data_time": 0.08061, "loss_cls_0": 0.8114, "loss_box_0": 1.68949, "loss_cns_0": 0.61974, "loss_yns_0": 0.14699, "loss_cls_1": 0.9061, "loss_box_1": 1.64523, "loss_cns_1": 0.65319, "loss_yns_1": 0.14703, "loss_cls_2": 0.92389, "loss_box_2": 1.63449, "loss_cns_2": 0.65281, "loss_yns_2": 0.1457, "loss_cls_3": 0.94199, "loss_box_3": 1.65, "loss_cns_3": 0.65285, "loss_yns_3": 0.14811, "loss_cls_4": 0.94845, "loss_box_4": 1.62283, "loss_cns_4": 0.65445, "loss_yns_4": 0.14792, "loss_cls_5": 0.94669, "loss_box_5": 1.6173, "loss_cns_5": 0.65292, "loss_yns_5": 0.1475, "loss_cls_dn_0": 0.21387, "loss_box_dn_0": 0.76955, "loss_cls_dn_1": 0.13392, "loss_box_dn_1": 0.72454, "loss_cls_dn_2": 0.1349, "loss_box_dn_2": 0.72034, "loss_cls_dn_3": 0.13919, "loss_box_dn_3": 0.7464, "loss_cls_dn_4": 0.14794, "loss_box_dn_4": 0.7474, "loss_cls_dn_5": 0.1543, "loss_box_dn_5": 0.76417, "loss_dense_depth": 0.77842, "loss": 26.282, "grad_norm": 51.59954, "time": 1.58058} -{"mode": "train", "epoch": 1, "iter": 189, "lr": 0.00018, "memory": 49164, "data_time": 0.0738, "loss_cls_0": 0.81127, "loss_box_0": 1.68195, "loss_cns_0": 0.6248, "loss_yns_0": 0.14826, "loss_cls_1": 0.89915, "loss_box_1": 1.66718, "loss_cns_1": 0.65282, "loss_yns_1": 0.15325, "loss_cls_2": 0.9139, "loss_box_2": 1.6739, "loss_cns_2": 0.65024, "loss_yns_2": 0.15163, "loss_cls_3": 0.94017, "loss_box_3": 1.65262, "loss_cns_3": 0.65309, "loss_yns_3": 0.15282, "loss_cls_4": 0.93885, "loss_box_4": 1.66644, "loss_cns_4": 0.65274, "loss_yns_4": 0.15468, "loss_cls_5": 0.93741, "loss_box_5": 1.66703, "loss_cns_5": 0.65655, "loss_yns_5": 0.15268, "loss_cls_dn_0": 0.21677, "loss_box_dn_0": 0.76762, "loss_cls_dn_1": 0.13607, "loss_box_dn_1": 0.74463, "loss_cls_dn_2": 0.13602, "loss_box_dn_2": 0.76124, "loss_cls_dn_3": 0.14656, "loss_box_dn_3": 0.78212, "loss_cls_dn_4": 0.15071, "loss_box_dn_4": 0.8097, "loss_cls_dn_5": 0.15373, "loss_box_dn_5": 0.83799, "loss_dense_depth": 0.78621, "loss": 26.68279, "grad_norm": 57.18006, "time": 1.59977} -{"mode": "train", "epoch": 1, "iter": 190, "lr": 0.00018, "memory": 49164, "data_time": 0.07599, "loss_cls_0": 0.83161, "loss_box_0": 1.71938, "loss_cns_0": 0.62234, "loss_yns_0": 0.1519, "loss_cls_1": 0.90199, "loss_box_1": 1.69096, "loss_cns_1": 0.64955, "loss_yns_1": 0.1562, "loss_cls_2": 0.91994, "loss_box_2": 1.68637, "loss_cns_2": 0.65041, "loss_yns_2": 0.15381, "loss_cls_3": 0.92709, "loss_box_3": 1.65476, "loss_cns_3": 0.65286, "loss_yns_3": 0.15171, "loss_cls_4": 0.93173, "loss_box_4": 1.69955, "loss_cns_4": 0.65544, "loss_yns_4": 0.15675, "loss_cls_5": 0.96133, "loss_box_5": 1.69494, "loss_cns_5": 0.65555, "loss_yns_5": 0.15806, "loss_cls_dn_0": 0.2206, "loss_box_dn_0": 0.7757, "loss_cls_dn_1": 0.13789, "loss_box_dn_1": 0.78599, "loss_cls_dn_2": 0.13822, "loss_box_dn_2": 0.80491, "loss_cls_dn_3": 0.14385, "loss_box_dn_3": 0.81941, "loss_cls_dn_4": 0.14767, "loss_box_dn_4": 0.86353, "loss_cls_dn_5": 0.15506, "loss_box_dn_5": 0.88981, "loss_dense_depth": 0.80096, "loss": 27.11781, "grad_norm": 56.92331, "time": 1.59324} -{"mode": "train", "epoch": 1, "iter": 191, "lr": 0.00018, "memory": 49164, "data_time": 0.08, "loss_cls_0": 0.82591, "loss_box_0": 1.72823, "loss_cns_0": 0.62193, "loss_yns_0": 0.15077, "loss_cls_1": 0.90913, "loss_box_1": 1.67446, "loss_cns_1": 0.65374, "loss_yns_1": 0.15271, "loss_cls_2": 0.93222, "loss_box_2": 1.65747, "loss_cns_2": 0.65299, "loss_yns_2": 0.15347, "loss_cls_3": 0.9273, "loss_box_3": 1.64643, "loss_cns_3": 0.65423, "loss_yns_3": 0.15222, "loss_cls_4": 0.94646, "loss_box_4": 1.66921, "loss_cns_4": 0.65955, "loss_yns_4": 0.15413, "loss_cls_5": 0.95633, "loss_box_5": 1.65607, "loss_cns_5": 0.65402, "loss_yns_5": 0.15956, "loss_cls_dn_0": 0.2198, "loss_box_dn_0": 0.76623, "loss_cls_dn_1": 0.13951, "loss_box_dn_1": 0.81968, "loss_cls_dn_2": 0.13791, "loss_box_dn_2": 0.82824, "loss_cls_dn_3": 0.14168, "loss_box_dn_3": 0.83875, "loss_cls_dn_4": 0.14573, "loss_box_dn_4": 0.87032, "loss_cls_dn_5": 0.15331, "loss_box_dn_5": 0.88761, "loss_dense_depth": 0.75946, "loss": 27.0568, "grad_norm": 53.43457, "time": 1.58892} -{"mode": "train", "epoch": 1, "iter": 192, "lr": 0.00018, "memory": 49164, "data_time": 0.07756, "loss_cls_0": 0.80832, "loss_box_0": 1.70385, "loss_cns_0": 0.62324, "loss_yns_0": 0.15003, "loss_cls_1": 0.89741, "loss_box_1": 1.67566, "loss_cns_1": 0.65261, "loss_yns_1": 0.15099, "loss_cls_2": 0.91829, "loss_box_2": 1.65315, "loss_cns_2": 0.65496, "loss_yns_2": 0.15126, "loss_cls_3": 0.93139, "loss_box_3": 1.64161, "loss_cns_3": 0.65452, "loss_yns_3": 0.1535, "loss_cls_4": 0.93505, "loss_box_4": 1.64304, "loss_cns_4": 0.65981, "loss_yns_4": 0.1525, "loss_cls_5": 0.93272, "loss_box_5": 1.65191, "loss_cns_5": 0.65523, "loss_yns_5": 0.15444, "loss_cls_dn_0": 0.21674, "loss_box_dn_0": 0.75547, "loss_cls_dn_1": 0.13884, "loss_box_dn_1": 0.83469, "loss_cls_dn_2": 0.13804, "loss_box_dn_2": 0.83243, "loss_cls_dn_3": 0.1466, "loss_box_dn_3": 0.83275, "loss_cls_dn_4": 0.14664, "loss_box_dn_4": 0.84166, "loss_cls_dn_5": 0.15084, "loss_box_dn_5": 0.85894, "loss_dense_depth": 0.76002, "loss": 26.85913, "grad_norm": 43.43899, "time": 1.58107} -{"mode": "train", "epoch": 1, "iter": 193, "lr": 0.00018, "memory": 49164, "data_time": 0.07558, "loss_cls_0": 0.79564, "loss_box_0": 1.68352, "loss_cns_0": 0.62753, "loss_yns_0": 0.14868, "loss_cls_1": 0.88037, "loss_box_1": 1.65204, "loss_cns_1": 0.65427, "loss_yns_1": 0.15003, "loss_cls_2": 0.91385, "loss_box_2": 1.6234, "loss_cns_2": 0.65716, "loss_yns_2": 0.15178, "loss_cls_3": 0.91495, "loss_box_3": 1.62628, "loss_cns_3": 0.66046, "loss_yns_3": 0.15287, "loss_cls_4": 0.92121, "loss_box_4": 1.61834, "loss_cns_4": 0.65961, "loss_yns_4": 0.15733, "loss_cls_5": 0.9399, "loss_box_5": 1.61613, "loss_cns_5": 0.65881, "loss_yns_5": 0.15123, "loss_cls_dn_0": 0.21493, "loss_box_dn_0": 0.76539, "loss_cls_dn_1": 0.13237, "loss_box_dn_1": 0.76916, "loss_cls_dn_2": 0.14322, "loss_box_dn_2": 0.75954, "loss_cls_dn_3": 0.14326, "loss_box_dn_3": 0.75707, "loss_cls_dn_4": 0.14123, "loss_box_dn_4": 0.75573, "loss_cls_dn_5": 0.1469, "loss_box_dn_5": 0.76202, "loss_dense_depth": 0.7456, "loss": 26.2518, "grad_norm": 50.28603, "time": 1.58077} -{"mode": "train", "epoch": 1, "iter": 194, "lr": 0.00018, "memory": 49164, "data_time": 0.0926, "loss_cls_0": 0.82074, "loss_box_0": 1.67709, "loss_cns_0": 0.61859, "loss_yns_0": 0.14978, "loss_cls_1": 0.90827, "loss_box_1": 1.68445, "loss_cns_1": 0.65046, "loss_yns_1": 0.15468, "loss_cls_2": 0.93796, "loss_box_2": 1.64532, "loss_cns_2": 0.65432, "loss_yns_2": 0.1601, "loss_cls_3": 0.9359, "loss_box_3": 1.65858, "loss_cns_3": 0.6579, "loss_yns_3": 0.15644, "loss_cls_4": 0.94132, "loss_box_4": 1.66249, "loss_cns_4": 0.65691, "loss_yns_4": 0.1582, "loss_cls_5": 0.95905, "loss_box_5": 1.64752, "loss_cns_5": 0.65102, "loss_yns_5": 0.15431, "loss_cls_dn_0": 0.21757, "loss_box_dn_0": 0.75747, "loss_cls_dn_1": 0.13045, "loss_box_dn_1": 0.74794, "loss_cls_dn_2": 0.14098, "loss_box_dn_2": 0.73368, "loss_cls_dn_3": 0.13632, "loss_box_dn_3": 0.73748, "loss_cls_dn_4": 0.1373, "loss_box_dn_4": 0.7393, "loss_cls_dn_5": 0.14288, "loss_box_dn_5": 0.74013, "loss_dense_depth": 0.77001, "loss": 26.4329, "grad_norm": 41.63073, "time": 1.59913} -{"mode": "train", "epoch": 1, "iter": 195, "lr": 0.00018, "memory": 49164, "data_time": 0.07975, "loss_cls_0": 0.78315, "loss_box_0": 1.65887, "loss_cns_0": 0.6252, "loss_yns_0": 0.14416, "loss_cls_1": 0.88122, "loss_box_1": 1.6154, "loss_cns_1": 0.65501, "loss_yns_1": 0.14966, "loss_cls_2": 0.90695, "loss_box_2": 1.61719, "loss_cns_2": 0.65407, "loss_yns_2": 0.14645, "loss_cls_3": 0.90484, "loss_box_3": 1.61827, "loss_cns_3": 0.65731, "loss_yns_3": 0.14885, "loss_cls_4": 0.90725, "loss_box_4": 1.6288, "loss_cns_4": 0.65645, "loss_yns_4": 0.14781, "loss_cls_5": 0.93223, "loss_box_5": 1.62249, "loss_cns_5": 0.65009, "loss_yns_5": 0.14556, "loss_cls_dn_0": 0.204, "loss_box_dn_0": 0.75247, "loss_cls_dn_1": 0.12583, "loss_box_dn_1": 0.73336, "loss_cls_dn_2": 0.12563, "loss_box_dn_2": 0.7239, "loss_cls_dn_3": 0.12648, "loss_box_dn_3": 0.73813, "loss_cls_dn_4": 0.13202, "loss_box_dn_4": 0.75383, "loss_cls_dn_5": 0.13494, "loss_box_dn_5": 0.7743, "loss_dense_depth": 0.73888, "loss": 25.92105, "grad_norm": 50.02182, "time": 1.64555} -{"mode": "train", "epoch": 1, "iter": 196, "lr": 0.00018, "memory": 49164, "data_time": 0.0776, "loss_cls_0": 0.76774, "loss_box_0": 1.66441, "loss_cns_0": 0.62627, "loss_yns_0": 0.146, "loss_cls_1": 0.86442, "loss_box_1": 1.62714, "loss_cns_1": 0.65243, "loss_yns_1": 0.14576, "loss_cls_2": 0.90188, "loss_box_2": 1.62266, "loss_cns_2": 0.65504, "loss_yns_2": 0.14862, "loss_cls_3": 0.89726, "loss_box_3": 1.61046, "loss_cns_3": 0.6571, "loss_yns_3": 0.14827, "loss_cls_4": 0.90257, "loss_box_4": 1.59688, "loss_cns_4": 0.65938, "loss_yns_4": 0.14905, "loss_cls_5": 0.90374, "loss_box_5": 1.59918, "loss_cns_5": 0.65512, "loss_yns_5": 0.1463, "loss_cls_dn_0": 0.20489, "loss_box_dn_0": 0.76083, "loss_cls_dn_1": 0.12518, "loss_box_dn_1": 0.76242, "loss_cls_dn_2": 0.12796, "loss_box_dn_2": 0.76389, "loss_cls_dn_3": 0.13321, "loss_box_dn_3": 0.78392, "loss_cls_dn_4": 0.13336, "loss_box_dn_4": 0.79775, "loss_cls_dn_5": 0.14133, "loss_box_dn_5": 0.82201, "loss_dense_depth": 0.74247, "loss": 26.04692, "grad_norm": 41.03923, "time": 1.6286} -{"mode": "train", "epoch": 1, "iter": 197, "lr": 0.00018, "memory": 49164, "data_time": 0.07549, "loss_cls_0": 0.80391, "loss_box_0": 1.72657, "loss_cns_0": 0.62359, "loss_yns_0": 0.14935, "loss_cls_1": 0.87507, "loss_box_1": 1.67119, "loss_cns_1": 0.65499, "loss_yns_1": 0.14934, "loss_cls_2": 0.8991, "loss_box_2": 1.65021, "loss_cns_2": 0.65988, "loss_yns_2": 0.15215, "loss_cls_3": 0.9074, "loss_box_3": 1.69265, "loss_cns_3": 0.65714, "loss_yns_3": 0.15258, "loss_cls_4": 0.90784, "loss_box_4": 1.68177, "loss_cns_4": 0.65946, "loss_yns_4": 0.15103, "loss_cls_5": 0.91628, "loss_box_5": 1.70633, "loss_cns_5": 0.6556, "loss_yns_5": 0.14941, "loss_cls_dn_0": 0.21462, "loss_box_dn_0": 0.75405, "loss_cls_dn_1": 0.1254, "loss_box_dn_1": 0.78246, "loss_cls_dn_2": 0.12696, "loss_box_dn_2": 0.79137, "loss_cls_dn_3": 0.13028, "loss_box_dn_3": 0.83108, "loss_cls_dn_4": 0.13579, "loss_box_dn_4": 0.85308, "loss_cls_dn_5": 0.14546, "loss_box_dn_5": 0.88626, "loss_dense_depth": 0.78173, "loss": 26.81139, "grad_norm": 52.21975, "time": 1.60084} -{"mode": "train", "epoch": 1, "iter": 198, "lr": 0.00018, "memory": 49164, "data_time": 0.07557, "loss_cls_0": 0.81758, "loss_box_0": 1.74213, "loss_cns_0": 0.62044, "loss_yns_0": 0.15223, "loss_cls_1": 0.88727, "loss_box_1": 1.66292, "loss_cns_1": 0.65136, "loss_yns_1": 0.15246, "loss_cls_2": 0.92044, "loss_box_2": 1.63782, "loss_cns_2": 0.65474, "loss_yns_2": 0.15108, "loss_cls_3": 0.91486, "loss_box_3": 1.66977, "loss_cns_3": 0.65305, "loss_yns_3": 0.15401, "loss_cls_4": 0.91149, "loss_box_4": 1.66833, "loss_cns_4": 0.65557, "loss_yns_4": 0.1518, "loss_cls_5": 0.91216, "loss_box_5": 1.68257, "loss_cns_5": 0.65528, "loss_yns_5": 0.15195, "loss_cls_dn_0": 0.21286, "loss_box_dn_0": 0.76593, "loss_cls_dn_1": 0.13109, "loss_box_dn_1": 0.85905, "loss_cls_dn_2": 0.13216, "loss_box_dn_2": 0.86323, "loss_cls_dn_3": 0.13424, "loss_box_dn_3": 0.89388, "loss_cls_dn_4": 0.13906, "loss_box_dn_4": 0.9135, "loss_cls_dn_5": 0.14502, "loss_box_dn_5": 0.9355, "loss_dense_depth": 0.76593, "loss": 27.12273, "grad_norm": 47.06406, "time": 1.57293} -{"mode": "train", "epoch": 1, "iter": 199, "lr": 0.00018, "memory": 49164, "data_time": 0.07717, "loss_cls_0": 0.79935, "loss_box_0": 1.72688, "loss_cns_0": 0.61941, "loss_yns_0": 0.15429, "loss_cls_1": 0.87563, "loss_box_1": 1.68508, "loss_cns_1": 0.64929, "loss_yns_1": 0.15527, "loss_cls_2": 0.90363, "loss_box_2": 1.65318, "loss_cns_2": 0.65289, "loss_yns_2": 0.15339, "loss_cls_3": 0.90816, "loss_box_3": 1.6469, "loss_cns_3": 0.65527, "loss_yns_3": 0.15368, "loss_cls_4": 0.90726, "loss_box_4": 1.65655, "loss_cns_4": 0.65522, "loss_yns_4": 0.15391, "loss_cls_5": 0.91846, "loss_box_5": 1.64766, "loss_cns_5": 0.65611, "loss_yns_5": 0.15421, "loss_cls_dn_0": 0.20848, "loss_box_dn_0": 0.75506, "loss_cls_dn_1": 0.13053, "loss_box_dn_1": 0.86435, "loss_cls_dn_2": 0.1315, "loss_box_dn_2": 0.85687, "loss_cls_dn_3": 0.13315, "loss_box_dn_3": 0.8649, "loss_cls_dn_4": 0.13582, "loss_box_dn_4": 0.8738, "loss_cls_dn_5": 0.13804, "loss_box_dn_5": 0.87821, "loss_dense_depth": 0.80401, "loss": 26.91639, "grad_norm": 33.47229, "time": 1.57049} -{"mode": "train", "epoch": 1, "iter": 200, "lr": 0.00018, "memory": 49164, "data_time": 0.0811, "loss_cls_0": 0.79387, "loss_box_0": 1.73901, "loss_cns_0": 0.61509, "loss_yns_0": 0.15107, "loss_cls_1": 0.8784, "loss_box_1": 1.68758, "loss_cns_1": 0.64985, "loss_yns_1": 0.15402, "loss_cls_2": 0.8958, "loss_box_2": 1.65619, "loss_cns_2": 0.65269, "loss_yns_2": 0.15728, "loss_cls_3": 0.90346, "loss_box_3": 1.64645, "loss_cns_3": 0.65517, "loss_yns_3": 0.15474, "loss_cls_4": 0.90535, "loss_box_4": 1.63377, "loss_cns_4": 0.65696, "loss_yns_4": 0.15458, "loss_cls_5": 0.91585, "loss_box_5": 1.64346, "loss_cns_5": 0.65531, "loss_yns_5": 0.15363, "loss_cls_dn_0": 0.20943, "loss_box_dn_0": 0.75766, "loss_cls_dn_1": 0.12739, "loss_box_dn_1": 0.84591, "loss_cls_dn_2": 0.13031, "loss_box_dn_2": 0.83315, "loss_cls_dn_3": 0.1354, "loss_box_dn_3": 0.82997, "loss_cls_dn_4": 0.13325, "loss_box_dn_4": 0.82786, "loss_cls_dn_5": 0.13521, "loss_box_dn_5": 0.83234, "loss_dense_depth": 0.85579, "loss": 26.76324, "grad_norm": 42.05111, "time": 1.5766} -{"mode": "train", "epoch": 1, "iter": 201, "lr": 0.00018, "memory": 49164, "data_time": 0.11691, "loss_cls_0": 0.79888, "loss_box_0": 1.69312, "loss_cns_0": 0.6148, "loss_yns_0": 0.14886, "loss_cls_1": 0.86812, "loss_box_1": 1.66503, "loss_cns_1": 0.64607, "loss_yns_1": 0.15299, "loss_cls_2": 0.87777, "loss_box_2": 1.66233, "loss_cns_2": 0.64806, "loss_yns_2": 0.15429, "loss_cls_3": 0.89495, "loss_box_3": 1.64434, "loss_cns_3": 0.65046, "loss_yns_3": 0.15498, "loss_cls_4": 0.89829, "loss_box_4": 1.63869, "loss_cns_4": 0.65601, "loss_yns_4": 0.15368, "loss_cls_5": 0.91326, "loss_box_5": 1.65632, "loss_cns_5": 0.65047, "loss_yns_5": 0.15151, "loss_cls_dn_0": 0.20458, "loss_box_dn_0": 0.75157, "loss_cls_dn_1": 0.12253, "loss_box_dn_1": 0.80661, "loss_cls_dn_2": 0.12454, "loss_box_dn_2": 0.79393, "loss_cls_dn_3": 0.12591, "loss_box_dn_3": 0.79558, "loss_cls_dn_4": 0.12755, "loss_box_dn_4": 0.79801, "loss_cls_dn_5": 0.13136, "loss_box_dn_5": 0.81186, "loss_dense_depth": 0.81272, "loss": 26.40005, "grad_norm": 36.56207, "time": 1.65141} -{"mode": "train", "epoch": 1, "iter": 202, "lr": 0.00018, "memory": 49164, "data_time": 0.10202, "loss_cls_0": 0.78684, "loss_box_0": 1.67599, "loss_cns_0": 0.62, "loss_yns_0": 0.15036, "loss_cls_1": 0.88027, "loss_box_1": 1.64265, "loss_cns_1": 0.64947, "loss_yns_1": 0.14943, "loss_cls_2": 0.88967, "loss_box_2": 1.62876, "loss_cns_2": 0.65035, "loss_yns_2": 0.15065, "loss_cls_3": 0.89742, "loss_box_3": 1.62143, "loss_cns_3": 0.65244, "loss_yns_3": 0.15302, "loss_cls_4": 0.89954, "loss_box_4": 1.61023, "loss_cns_4": 0.65076, "loss_yns_4": 0.15048, "loss_cls_5": 0.89883, "loss_box_5": 1.61696, "loss_cns_5": 0.64872, "loss_yns_5": 0.15207, "loss_cls_dn_0": 0.2019, "loss_box_dn_0": 0.75884, "loss_cls_dn_1": 0.12457, "loss_box_dn_1": 0.76734, "loss_cls_dn_2": 0.12685, "loss_box_dn_2": 0.76106, "loss_cls_dn_3": 0.12564, "loss_box_dn_3": 0.77972, "loss_cls_dn_4": 0.1332, "loss_box_dn_4": 0.79469, "loss_cls_dn_5": 0.13609, "loss_box_dn_5": 0.82228, "loss_dense_depth": 0.77881, "loss": 26.13733, "grad_norm": 36.6998, "time": 1.61785} -{"mode": "train", "epoch": 1, "iter": 203, "lr": 0.00018, "memory": 49164, "data_time": 0.08295, "loss_cls_0": 0.79421, "loss_box_0": 1.70745, "loss_cns_0": 0.62227, "loss_yns_0": 0.15157, "loss_cls_1": 0.87968, "loss_box_1": 1.70095, "loss_cns_1": 0.65074, "loss_yns_1": 0.14907, "loss_cls_2": 0.89748, "loss_box_2": 1.69255, "loss_cns_2": 0.65214, "loss_yns_2": 0.15071, "loss_cls_3": 0.89858, "loss_box_3": 1.68791, "loss_cns_3": 0.65364, "loss_yns_3": 0.14997, "loss_cls_4": 0.89728, "loss_box_4": 1.67759, "loss_cns_4": 0.65289, "loss_yns_4": 0.14955, "loss_cls_5": 0.91939, "loss_box_5": 1.69184, "loss_cns_5": 0.6519, "loss_yns_5": 0.15061, "loss_cls_dn_0": 0.20384, "loss_box_dn_0": 0.76058, "loss_cls_dn_1": 0.1243, "loss_box_dn_1": 0.7971, "loss_cls_dn_2": 0.12355, "loss_box_dn_2": 0.79686, "loss_cls_dn_3": 0.12438, "loss_box_dn_3": 0.81631, "loss_cls_dn_4": 0.13181, "loss_box_dn_4": 0.83367, "loss_cls_dn_5": 0.13566, "loss_box_dn_5": 0.86814, "loss_dense_depth": 0.8063, "loss": 26.75247, "grad_norm": 41.68928, "time": 1.57992} -{"mode": "train", "epoch": 1, "iter": 204, "lr": 0.00018, "memory": 49164, "data_time": 0.08091, "loss_cls_0": 0.79672, "loss_box_0": 1.74357, "loss_cns_0": 0.62541, "loss_yns_0": 0.15025, "loss_cls_1": 0.87465, "loss_box_1": 1.71194, "loss_cns_1": 0.65626, "loss_yns_1": 0.14992, "loss_cls_2": 0.88937, "loss_box_2": 1.71955, "loss_cns_2": 0.65411, "loss_yns_2": 0.15152, "loss_cls_3": 0.89244, "loss_box_3": 1.68767, "loss_cns_3": 0.65683, "loss_yns_3": 0.14937, "loss_cls_4": 0.88901, "loss_box_4": 1.68229, "loss_cns_4": 0.65878, "loss_yns_4": 0.15075, "loss_cls_5": 0.90143, "loss_box_5": 1.70483, "loss_cns_5": 0.65524, "loss_yns_5": 0.14852, "loss_cls_dn_0": 0.20651, "loss_box_dn_0": 0.7691, "loss_cls_dn_1": 0.12259, "loss_box_dn_1": 0.77458, "loss_cls_dn_2": 0.12318, "loss_box_dn_2": 0.7914, "loss_cls_dn_3": 0.12738, "loss_box_dn_3": 0.80201, "loss_cls_dn_4": 0.1352, "loss_box_dn_4": 0.82582, "loss_cls_dn_5": 0.13978, "loss_box_dn_5": 0.86692, "loss_dense_depth": 0.78484, "loss": 26.76977, "grad_norm": 47.58227, "time": 1.60615} -{"mode": "train", "epoch": 1, "iter": 205, "lr": 0.00018, "memory": 49164, "data_time": 0.08041, "loss_cls_0": 0.79441, "loss_box_0": 1.75302, "loss_cns_0": 0.62495, "loss_yns_0": 0.15121, "loss_cls_1": 0.87492, "loss_box_1": 1.65759, "loss_cns_1": 0.65545, "loss_yns_1": 0.15341, "loss_cls_2": 0.88998, "loss_box_2": 1.64822, "loss_cns_2": 0.65655, "loss_yns_2": 0.15154, "loss_cls_3": 0.90685, "loss_box_3": 1.60399, "loss_cns_3": 0.65631, "loss_yns_3": 0.14972, "loss_cls_4": 0.8946, "loss_box_4": 1.60932, "loss_cns_4": 0.66249, "loss_yns_4": 0.15212, "loss_cls_5": 0.89822, "loss_box_5": 1.61501, "loss_cns_5": 0.66007, "loss_yns_5": 0.15074, "loss_cls_dn_0": 0.21195, "loss_box_dn_0": 0.77265, "loss_cls_dn_1": 0.13, "loss_box_dn_1": 0.77357, "loss_cls_dn_2": 0.13157, "loss_box_dn_2": 0.77903, "loss_cls_dn_3": 0.13479, "loss_box_dn_3": 0.7779, "loss_cls_dn_4": 0.13793, "loss_box_dn_4": 0.79424, "loss_cls_dn_5": 0.14186, "loss_box_dn_5": 0.81594, "loss_dense_depth": 0.76944, "loss": 26.34155, "grad_norm": 34.04148, "time": 1.598} -{"mode": "train", "epoch": 1, "iter": 206, "lr": 0.00018, "memory": 49164, "data_time": 0.10625, "loss_cls_0": 0.79222, "loss_box_0": 1.74971, "loss_cns_0": 0.62186, "loss_yns_0": 0.15172, "loss_cls_1": 0.87732, "loss_box_1": 1.67258, "loss_cns_1": 0.65727, "loss_yns_1": 0.15081, "loss_cls_2": 0.89423, "loss_box_2": 1.63145, "loss_cns_2": 0.65681, "loss_yns_2": 0.15086, "loss_cls_3": 0.89548, "loss_box_3": 1.63569, "loss_cns_3": 0.65627, "loss_yns_3": 0.14998, "loss_cls_4": 0.89718, "loss_box_4": 1.62192, "loss_cns_4": 0.65579, "loss_yns_4": 0.15017, "loss_cls_5": 0.91245, "loss_box_5": 1.62685, "loss_cns_5": 0.6569, "loss_yns_5": 0.14983, "loss_cls_dn_0": 0.21657, "loss_box_dn_0": 0.76995, "loss_cls_dn_1": 0.12753, "loss_box_dn_1": 0.74522, "loss_cls_dn_2": 0.13248, "loss_box_dn_2": 0.73095, "loss_cls_dn_3": 0.13593, "loss_box_dn_3": 0.73417, "loss_cls_dn_4": 0.1397, "loss_box_dn_4": 0.73746, "loss_cls_dn_5": 0.1456, "loss_box_dn_5": 0.74654, "loss_dense_depth": 0.77843, "loss": 26.15588, "grad_norm": 38.22887, "time": 1.60822} -{"mode": "train", "epoch": 1, "iter": 207, "lr": 0.00018, "memory": 49164, "data_time": 0.07861, "loss_cls_0": 0.79864, "loss_box_0": 1.71639, "loss_cns_0": 0.6216, "loss_yns_0": 0.15372, "loss_cls_1": 0.87274, "loss_box_1": 1.67123, "loss_cns_1": 0.65435, "loss_yns_1": 0.14868, "loss_cls_2": 0.88846, "loss_box_2": 1.63317, "loss_cns_2": 0.65577, "loss_yns_2": 0.15185, "loss_cls_3": 0.89179, "loss_box_3": 1.61669, "loss_cns_3": 0.65526, "loss_yns_3": 0.15159, "loss_cls_4": 0.89552, "loss_box_4": 1.6168, "loss_cns_4": 0.65381, "loss_yns_4": 0.15229, "loss_cls_5": 0.90213, "loss_box_5": 1.61483, "loss_cns_5": 0.65649, "loss_yns_5": 0.1531, "loss_cls_dn_0": 0.21558, "loss_box_dn_0": 0.75609, "loss_cls_dn_1": 0.13206, "loss_box_dn_1": 0.72415, "loss_cls_dn_2": 0.13501, "loss_box_dn_2": 0.71156, "loss_cls_dn_3": 0.13587, "loss_box_dn_3": 0.70399, "loss_cls_dn_4": 0.13924, "loss_box_dn_4": 0.70859, "loss_cls_dn_5": 0.14308, "loss_box_dn_5": 0.70964, "loss_dense_depth": 0.75409, "loss": 25.89586, "grad_norm": 29.00636, "time": 1.58144} -{"mode": "train", "epoch": 1, "iter": 208, "lr": 0.00018, "memory": 49164, "data_time": 0.09004, "loss_cls_0": 0.79971, "loss_box_0": 1.69733, "loss_cns_0": 0.61881, "loss_yns_0": 0.15359, "loss_cls_1": 0.87667, "loss_box_1": 1.66617, "loss_cns_1": 0.65308, "loss_yns_1": 0.15286, "loss_cls_2": 0.8896, "loss_box_2": 1.62323, "loss_cns_2": 0.65342, "loss_yns_2": 0.15505, "loss_cls_3": 0.90117, "loss_box_3": 1.61581, "loss_cns_3": 0.65564, "loss_yns_3": 0.15452, "loss_cls_4": 0.90648, "loss_box_4": 1.60224, "loss_cns_4": 0.65675, "loss_yns_4": 0.15703, "loss_cls_5": 0.91219, "loss_box_5": 1.60835, "loss_cns_5": 0.65539, "loss_yns_5": 0.15619, "loss_cls_dn_0": 0.21164, "loss_box_dn_0": 0.76047, "loss_cls_dn_1": 0.12782, "loss_box_dn_1": 0.70783, "loss_cls_dn_2": 0.1348, "loss_box_dn_2": 0.6921, "loss_cls_dn_3": 0.13235, "loss_box_dn_3": 0.69128, "loss_cls_dn_4": 0.13427, "loss_box_dn_4": 0.69772, "loss_cls_dn_5": 0.13832, "loss_box_dn_5": 0.71232, "loss_dense_depth": 0.78304, "loss": 25.84524, "grad_norm": 30.63281, "time": 1.59636} -{"mode": "train", "epoch": 1, "iter": 209, "lr": 0.00018, "memory": 49164, "data_time": 0.09212, "loss_cls_0": 0.81522, "loss_box_0": 1.74375, "loss_cns_0": 0.61932, "loss_yns_0": 0.1518, "loss_cls_1": 0.90588, "loss_box_1": 1.66616, "loss_cns_1": 0.65215, "loss_yns_1": 0.14759, "loss_cls_2": 0.91126, "loss_box_2": 1.66038, "loss_cns_2": 0.65458, "loss_yns_2": 0.14895, "loss_cls_3": 0.9108, "loss_box_3": 1.66386, "loss_cns_3": 0.65475, "loss_yns_3": 0.15026, "loss_cls_4": 0.91504, "loss_box_4": 1.66255, "loss_cns_4": 0.65728, "loss_yns_4": 0.15037, "loss_cls_5": 0.92295, "loss_box_5": 1.68147, "loss_cns_5": 0.65694, "loss_yns_5": 0.15011, "loss_cls_dn_0": 0.21321, "loss_box_dn_0": 0.76835, "loss_cls_dn_1": 0.12832, "loss_box_dn_1": 0.72832, "loss_cls_dn_2": 0.12979, "loss_box_dn_2": 0.72291, "loss_cls_dn_3": 0.13052, "loss_box_dn_3": 0.74021, "loss_cls_dn_4": 0.13288, "loss_box_dn_4": 0.75221, "loss_cls_dn_5": 0.13815, "loss_box_dn_5": 0.78056, "loss_dense_depth": 0.81755, "loss": 26.43637, "grad_norm": 40.30566, "time": 1.61176} -{"mode": "train", "epoch": 1, "iter": 210, "lr": 0.00018, "memory": 49164, "data_time": 0.09033, "loss_cls_0": 0.7884, "loss_box_0": 1.71935, "loss_cns_0": 0.62676, "loss_yns_0": 0.1534, "loss_cls_1": 0.86703, "loss_box_1": 1.66013, "loss_cns_1": 0.64967, "loss_yns_1": 0.14888, "loss_cls_2": 0.88578, "loss_box_2": 1.64574, "loss_cns_2": 0.65609, "loss_yns_2": 0.15066, "loss_cls_3": 0.88664, "loss_box_3": 1.63381, "loss_cns_3": 0.65581, "loss_yns_3": 0.14984, "loss_cls_4": 0.89437, "loss_box_4": 1.64041, "loss_cns_4": 0.65573, "loss_yns_4": 0.15087, "loss_cls_5": 0.89097, "loss_box_5": 1.6572, "loss_cns_5": 0.65656, "loss_yns_5": 0.151, "loss_cls_dn_0": 0.20145, "loss_box_dn_0": 0.75564, "loss_cls_dn_1": 0.12093, "loss_box_dn_1": 0.76593, "loss_cls_dn_2": 0.12333, "loss_box_dn_2": 0.77119, "loss_cls_dn_3": 0.12553, "loss_box_dn_3": 0.78568, "loss_cls_dn_4": 0.12788, "loss_box_dn_4": 0.80483, "loss_cls_dn_5": 0.1324, "loss_box_dn_5": 0.83425, "loss_dense_depth": 0.76516, "loss": 26.2893, "grad_norm": 39.70048, "time": 1.61501} -{"mode": "train", "epoch": 1, "iter": 211, "lr": 0.00018, "memory": 49164, "data_time": 0.07845, "loss_cls_0": 0.79262, "loss_box_0": 1.70138, "loss_cns_0": 0.62552, "loss_yns_0": 0.15219, "loss_cls_1": 0.86469, "loss_box_1": 1.65731, "loss_cns_1": 0.65308, "loss_yns_1": 0.14956, "loss_cls_2": 0.88935, "loss_box_2": 1.63423, "loss_cns_2": 0.65667, "loss_yns_2": 0.14983, "loss_cls_3": 0.89497, "loss_box_3": 1.61161, "loss_cns_3": 0.65599, "loss_yns_3": 0.15016, "loss_cls_4": 0.90363, "loss_box_4": 1.61674, "loss_cns_4": 0.65721, "loss_yns_4": 0.15016, "loss_cls_5": 0.8895, "loss_box_5": 1.63231, "loss_cns_5": 0.65479, "loss_yns_5": 0.15066, "loss_cls_dn_0": 0.19957, "loss_box_dn_0": 0.75513, "loss_cls_dn_1": 0.11397, "loss_box_dn_1": 0.75557, "loss_cls_dn_2": 0.11525, "loss_box_dn_2": 0.75561, "loss_cls_dn_3": 0.11867, "loss_box_dn_3": 0.75803, "loss_cls_dn_4": 0.11952, "loss_box_dn_4": 0.76793, "loss_cls_dn_5": 0.12444, "loss_box_dn_5": 0.78772, "loss_dense_depth": 0.75773, "loss": 26.02332, "grad_norm": 45.23724, "time": 1.61201} -{"mode": "train", "epoch": 1, "iter": 212, "lr": 0.00018, "memory": 49164, "data_time": 0.08296, "loss_cls_0": 0.77734, "loss_box_0": 1.69031, "loss_cns_0": 0.62717, "loss_yns_0": 0.14664, "loss_cls_1": 0.86496, "loss_box_1": 1.68265, "loss_cns_1": 0.65342, "loss_yns_1": 0.14669, "loss_cls_2": 0.87835, "loss_box_2": 1.66049, "loss_cns_2": 0.65506, "loss_yns_2": 0.14593, "loss_cls_3": 0.88596, "loss_box_3": 1.63689, "loss_cns_3": 0.6571, "loss_yns_3": 0.14823, "loss_cls_4": 0.89001, "loss_box_4": 1.63244, "loss_cns_4": 0.65838, "loss_yns_4": 0.14785, "loss_cls_5": 0.8894, "loss_box_5": 1.62706, "loss_cns_5": 0.65835, "loss_yns_5": 0.14772, "loss_cls_dn_0": 0.20287, "loss_box_dn_0": 0.75767, "loss_cls_dn_1": 0.12141, "loss_box_dn_1": 0.75275, "loss_cls_dn_2": 0.11956, "loss_box_dn_2": 0.74483, "loss_cls_dn_3": 0.12386, "loss_box_dn_3": 0.73675, "loss_cls_dn_4": 0.12404, "loss_box_dn_4": 0.74031, "loss_cls_dn_5": 0.129, "loss_box_dn_5": 0.74036, "loss_dense_depth": 0.76773, "loss": 25.96955, "grad_norm": 38.98722, "time": 1.59076} -{"mode": "train", "epoch": 1, "iter": 213, "lr": 0.00018, "memory": 49164, "data_time": 0.08132, "loss_cls_0": 0.77548, "loss_box_0": 1.69053, "loss_cns_0": 0.62499, "loss_yns_0": 0.14961, "loss_cls_1": 0.85796, "loss_box_1": 1.62946, "loss_cns_1": 0.65092, "loss_yns_1": 0.14871, "loss_cls_2": 0.87605, "loss_box_2": 1.58712, "loss_cns_2": 0.65468, "loss_yns_2": 0.15187, "loss_cls_3": 0.89834, "loss_box_3": 1.60067, "loss_cns_3": 0.65578, "loss_yns_3": 0.15158, "loss_cls_4": 0.89019, "loss_box_4": 1.58277, "loss_cns_4": 0.65623, "loss_yns_4": 0.15152, "loss_cls_5": 0.88951, "loss_box_5": 1.57936, "loss_cns_5": 0.65649, "loss_yns_5": 0.1518, "loss_cls_dn_0": 0.20362, "loss_box_dn_0": 0.75572, "loss_cls_dn_1": 0.12096, "loss_box_dn_1": 0.69846, "loss_cls_dn_2": 0.11871, "loss_box_dn_2": 0.68123, "loss_cls_dn_3": 0.12105, "loss_box_dn_3": 0.68382, "loss_cls_dn_4": 0.12422, "loss_box_dn_4": 0.68465, "loss_cls_dn_5": 0.12872, "loss_box_dn_5": 0.68155, "loss_dense_depth": 0.81551, "loss": 25.47984, "grad_norm": 34.1263, "time": 1.58098} -{"mode": "train", "epoch": 1, "iter": 214, "lr": 0.00019, "memory": 49164, "data_time": 0.07857, "loss_cls_0": 0.7928, "loss_box_0": 1.66847, "loss_cns_0": 0.6146, "loss_yns_0": 0.14702, "loss_cls_1": 0.86128, "loss_box_1": 1.61475, "loss_cns_1": 0.64936, "loss_yns_1": 0.15201, "loss_cls_2": 0.88262, "loss_box_2": 1.56455, "loss_cns_2": 0.65431, "loss_yns_2": 0.15315, "loss_cls_3": 0.90212, "loss_box_3": 1.57427, "loss_cns_3": 0.65234, "loss_yns_3": 0.15159, "loss_cls_4": 0.90216, "loss_box_4": 1.57143, "loss_cns_4": 0.65522, "loss_yns_4": 0.15129, "loss_cls_5": 0.90663, "loss_box_5": 1.58853, "loss_cns_5": 0.65581, "loss_yns_5": 0.15153, "loss_cls_dn_0": 0.20392, "loss_box_dn_0": 0.76337, "loss_cls_dn_1": 0.1162, "loss_box_dn_1": 0.691, "loss_cls_dn_2": 0.11968, "loss_box_dn_2": 0.67332, "loss_cls_dn_3": 0.11994, "loss_box_dn_3": 0.67714, "loss_cls_dn_4": 0.12913, "loss_box_dn_4": 0.67997, "loss_cls_dn_5": 0.12659, "loss_box_dn_5": 0.69145, "loss_dense_depth": 0.75096, "loss": 25.3605, "grad_norm": 39.53734, "time": 1.58321} -{"mode": "train", "epoch": 1, "iter": 215, "lr": 0.00019, "memory": 49164, "data_time": 0.07446, "loss_cls_0": 0.81779, "loss_box_0": 1.69058, "loss_cns_0": 0.61917, "loss_yns_0": 0.14896, "loss_cls_1": 0.87708, "loss_box_1": 1.64432, "loss_cns_1": 0.64874, "loss_yns_1": 0.15029, "loss_cls_2": 0.89961, "loss_box_2": 1.61391, "loss_cns_2": 0.6527, "loss_yns_2": 0.15026, "loss_cls_3": 0.91015, "loss_box_3": 1.59847, "loss_cns_3": 0.65412, "loss_yns_3": 0.15108, "loss_cls_4": 0.90579, "loss_box_4": 1.59327, "loss_cns_4": 0.65472, "loss_yns_4": 0.15067, "loss_cls_5": 0.90561, "loss_box_5": 1.60113, "loss_cns_5": 0.65375, "loss_yns_5": 0.15054, "loss_cls_dn_0": 0.21318, "loss_box_dn_0": 0.75757, "loss_cls_dn_1": 0.12164, "loss_box_dn_1": 0.69856, "loss_cls_dn_2": 0.12909, "loss_box_dn_2": 0.69824, "loss_cls_dn_3": 0.12692, "loss_box_dn_3": 0.69255, "loss_cls_dn_4": 0.13578, "loss_box_dn_4": 0.70129, "loss_cls_dn_5": 0.13132, "loss_box_dn_5": 0.71437, "loss_dense_depth": 0.8083, "loss": 25.77152, "grad_norm": 36.14664, "time": 1.57525} -{"mode": "train", "epoch": 1, "iter": 216, "lr": 0.00019, "memory": 49164, "data_time": 0.07766, "loss_cls_0": 0.81814, "loss_box_0": 1.72104, "loss_cns_0": 0.62659, "loss_yns_0": 0.15298, "loss_cls_1": 0.87996, "loss_box_1": 1.6235, "loss_cns_1": 0.65404, "loss_yns_1": 0.15221, "loss_cls_2": 0.90077, "loss_box_2": 1.60748, "loss_cns_2": 0.65613, "loss_yns_2": 0.15042, "loss_cls_3": 0.90377, "loss_box_3": 1.59911, "loss_cns_3": 0.65705, "loss_yns_3": 0.15233, "loss_cls_4": 0.90807, "loss_box_4": 1.59732, "loss_cns_4": 0.65568, "loss_yns_4": 0.15127, "loss_cls_5": 0.92312, "loss_box_5": 1.59159, "loss_cns_5": 0.6562, "loss_yns_5": 0.15205, "loss_cls_dn_0": 0.21341, "loss_box_dn_0": 0.76602, "loss_cls_dn_1": 0.12575, "loss_box_dn_1": 0.70394, "loss_cls_dn_2": 0.12903, "loss_box_dn_2": 0.70795, "loss_cls_dn_3": 0.12709, "loss_box_dn_3": 0.70517, "loss_cls_dn_4": 0.12918, "loss_box_dn_4": 0.71355, "loss_cls_dn_5": 0.13723, "loss_box_dn_5": 0.7196, "loss_dense_depth": 0.76989, "loss": 25.83862, "grad_norm": 40.15321, "time": 1.56534} -{"mode": "train", "epoch": 1, "iter": 217, "lr": 0.00019, "memory": 49164, "data_time": 0.07293, "loss_cls_0": 0.78269, "loss_box_0": 1.69364, "loss_cns_0": 0.62652, "loss_yns_0": 0.15621, "loss_cls_1": 0.85788, "loss_box_1": 1.63331, "loss_cns_1": 0.64951, "loss_yns_1": 0.15541, "loss_cls_2": 0.87265, "loss_box_2": 1.62113, "loss_cns_2": 0.65259, "loss_yns_2": 0.15563, "loss_cls_3": 0.88222, "loss_box_3": 1.60156, "loss_cns_3": 0.65442, "loss_yns_3": 0.15501, "loss_cls_4": 0.88491, "loss_box_4": 1.6086, "loss_cns_4": 0.65538, "loss_yns_4": 0.15593, "loss_cls_5": 0.88975, "loss_box_5": 1.6077, "loss_cns_5": 0.65569, "loss_yns_5": 0.15798, "loss_cls_dn_0": 0.20581, "loss_box_dn_0": 0.74943, "loss_cls_dn_1": 0.12238, "loss_box_dn_1": 0.70541, "loss_cls_dn_2": 0.12445, "loss_box_dn_2": 0.70536, "loss_cls_dn_3": 0.12706, "loss_box_dn_3": 0.69681, "loss_cls_dn_4": 0.13323, "loss_box_dn_4": 0.71107, "loss_cls_dn_5": 0.14512, "loss_box_dn_5": 0.71504, "loss_dense_depth": 0.71977, "loss": 25.62726, "grad_norm": 47.0578, "time": 1.56562} -{"mode": "train", "epoch": 1, "iter": 218, "lr": 0.00019, "memory": 49164, "data_time": 0.07441, "loss_cls_0": 0.78425, "loss_box_0": 1.68604, "loss_cns_0": 0.62897, "loss_yns_0": 0.15281, "loss_cls_1": 0.86033, "loss_box_1": 1.643, "loss_cns_1": 0.6516, "loss_yns_1": 0.15395, "loss_cls_2": 0.87846, "loss_box_2": 1.62417, "loss_cns_2": 0.65339, "loss_yns_2": 0.15387, "loss_cls_3": 0.89743, "loss_box_3": 1.58028, "loss_cns_3": 0.65639, "loss_yns_3": 0.15442, "loss_cls_4": 0.88815, "loss_box_4": 1.57755, "loss_cns_4": 0.65721, "loss_yns_4": 0.15627, "loss_cls_5": 0.88526, "loss_box_5": 1.57559, "loss_cns_5": 0.65353, "loss_yns_5": 0.15504, "loss_cls_dn_0": 0.20742, "loss_box_dn_0": 0.74858, "loss_cls_dn_1": 0.133, "loss_box_dn_1": 0.68914, "loss_cls_dn_2": 0.12986, "loss_box_dn_2": 0.68583, "loss_cls_dn_3": 0.1305, "loss_box_dn_3": 0.6661, "loss_cls_dn_4": 0.13613, "loss_box_dn_4": 0.67239, "loss_cls_dn_5": 0.14016, "loss_box_dn_5": 0.67558, "loss_dense_depth": 0.73112, "loss": 25.45377, "grad_norm": 32.99363, "time": 1.57229} -{"mode": "train", "epoch": 1, "iter": 219, "lr": 0.00019, "memory": 49164, "data_time": 0.07479, "loss_cls_0": 0.7971, "loss_box_0": 1.70746, "loss_cns_0": 0.62246, "loss_yns_0": 0.15281, "loss_cls_1": 0.86827, "loss_box_1": 1.67931, "loss_cns_1": 0.65148, "loss_yns_1": 0.15171, "loss_cls_2": 0.88961, "loss_box_2": 1.64544, "loss_cns_2": 0.65425, "loss_yns_2": 0.15159, "loss_cls_3": 0.90825, "loss_box_3": 1.64596, "loss_cns_3": 0.65777, "loss_yns_3": 0.15283, "loss_cls_4": 0.90851, "loss_box_4": 1.61803, "loss_cns_4": 0.65742, "loss_yns_4": 0.15154, "loss_cls_5": 0.91306, "loss_box_5": 1.62209, "loss_cns_5": 0.65797, "loss_yns_5": 0.15082, "loss_cls_dn_0": 0.21279, "loss_box_dn_0": 0.75163, "loss_cls_dn_1": 0.12943, "loss_box_dn_1": 0.68516, "loss_cls_dn_2": 0.13227, "loss_box_dn_2": 0.67506, "loss_cls_dn_3": 0.13802, "loss_box_dn_3": 0.67064, "loss_cls_dn_4": 0.14556, "loss_box_dn_4": 0.66824, "loss_cls_dn_5": 0.15325, "loss_box_dn_5": 0.67493, "loss_dense_depth": 0.73306, "loss": 25.78578, "grad_norm": 43.3602, "time": 1.58775} -{"mode": "train", "epoch": 1, "iter": 220, "lr": 0.00019, "memory": 49164, "data_time": 0.07508, "loss_cls_0": 0.77229, "loss_box_0": 1.69616, "loss_cns_0": 0.62462, "loss_yns_0": 0.15184, "loss_cls_1": 0.85614, "loss_box_1": 1.63979, "loss_cns_1": 0.65623, "loss_yns_1": 0.15189, "loss_cls_2": 0.85905, "loss_box_2": 1.61969, "loss_cns_2": 0.65528, "loss_yns_2": 0.15274, "loss_cls_3": 0.89245, "loss_box_3": 1.60987, "loss_cns_3": 0.66259, "loss_yns_3": 0.1541, "loss_cls_4": 0.88135, "loss_box_4": 1.59188, "loss_cns_4": 0.65999, "loss_yns_4": 0.15315, "loss_cls_5": 0.89226, "loss_box_5": 1.58689, "loss_cns_5": 0.66007, "loss_yns_5": 0.15162, "loss_cls_dn_0": 0.2045, "loss_box_dn_0": 0.75343, "loss_cls_dn_1": 0.13019, "loss_box_dn_1": 0.70184, "loss_cls_dn_2": 0.13151, "loss_box_dn_2": 0.68736, "loss_cls_dn_3": 0.1395, "loss_box_dn_3": 0.68933, "loss_cls_dn_4": 0.14236, "loss_box_dn_4": 0.68346, "loss_cls_dn_5": 0.14808, "loss_box_dn_5": 0.68703, "loss_dense_depth": 0.73666, "loss": 25.56717, "grad_norm": 46.99715, "time": 1.5846} -{"mode": "train", "epoch": 1, "iter": 221, "lr": 0.00019, "memory": 49164, "data_time": 0.10414, "loss_cls_0": 0.7864, "loss_box_0": 1.68121, "loss_cns_0": 0.61899, "loss_yns_0": 0.15171, "loss_cls_1": 0.86927, "loss_box_1": 1.65681, "loss_cns_1": 0.65496, "loss_yns_1": 0.15043, "loss_cls_2": 0.86863, "loss_box_2": 1.62401, "loss_cns_2": 0.65506, "loss_yns_2": 0.152, "loss_cls_3": 0.8959, "loss_box_3": 1.62596, "loss_cns_3": 0.65861, "loss_yns_3": 0.15195, "loss_cls_4": 0.88609, "loss_box_4": 1.59542, "loss_cns_4": 0.65783, "loss_yns_4": 0.15252, "loss_cls_5": 0.88596, "loss_box_5": 1.59911, "loss_cns_5": 0.65797, "loss_yns_5": 0.1509, "loss_cls_dn_0": 0.20935, "loss_box_dn_0": 0.75568, "loss_cls_dn_1": 0.13554, "loss_box_dn_1": 0.69985, "loss_cls_dn_2": 0.13316, "loss_box_dn_2": 0.67965, "loss_cls_dn_3": 0.13633, "loss_box_dn_3": 0.68541, "loss_cls_dn_4": 0.13798, "loss_box_dn_4": 0.67412, "loss_cls_dn_5": 0.14065, "loss_box_dn_5": 0.67997, "loss_dense_depth": 0.7473, "loss": 25.60272, "grad_norm": 29.57277, "time": 3.5047} -{"mode": "train", "epoch": 1, "iter": 222, "lr": 0.00019, "memory": 49164, "data_time": 0.10068, "loss_cls_0": 0.77509, "loss_box_0": 1.68963, "loss_cns_0": 0.61976, "loss_yns_0": 0.14931, "loss_cls_1": 0.85668, "loss_box_1": 1.63659, "loss_cns_1": 0.64953, "loss_yns_1": 0.14783, "loss_cls_2": 0.86821, "loss_box_2": 1.59723, "loss_cns_2": 0.65303, "loss_yns_2": 0.14934, "loss_cls_3": 0.88502, "loss_box_3": 1.62224, "loss_cns_3": 0.65689, "loss_yns_3": 0.1506, "loss_cls_4": 0.88627, "loss_box_4": 1.59318, "loss_cns_4": 0.65487, "loss_yns_4": 0.14917, "loss_cls_5": 0.88695, "loss_box_5": 1.59262, "loss_cns_5": 0.65572, "loss_yns_5": 0.14825, "loss_cls_dn_0": 0.20768, "loss_box_dn_0": 0.75481, "loss_cls_dn_1": 0.12841, "loss_box_dn_1": 0.70116, "loss_cls_dn_2": 0.1341, "loss_box_dn_2": 0.68095, "loss_cls_dn_3": 0.13269, "loss_box_dn_3": 0.68701, "loss_cls_dn_4": 0.1398, "loss_box_dn_4": 0.6804, "loss_cls_dn_5": 0.13775, "loss_box_dn_5": 0.68234, "loss_dense_depth": 0.76453, "loss": 25.50559, "grad_norm": 43.01746, "time": 3.35248} -{"mode": "train", "epoch": 1, "iter": 223, "lr": 0.00019, "memory": 49164, "data_time": 0.08081, "loss_cls_0": 0.80968, "loss_box_0": 1.73278, "loss_cns_0": 0.62394, "loss_yns_0": 0.15224, "loss_cls_1": 0.86018, "loss_box_1": 1.66859, "loss_cns_1": 0.64881, "loss_yns_1": 0.14936, "loss_cls_2": 0.88361, "loss_box_2": 1.6321, "loss_cns_2": 0.65229, "loss_yns_2": 0.15007, "loss_cls_3": 0.89514, "loss_box_3": 1.62503, "loss_cns_3": 0.65681, "loss_yns_3": 0.15135, "loss_cls_4": 0.91778, "loss_box_4": 1.61688, "loss_cns_4": 0.65414, "loss_yns_4": 0.15048, "loss_cls_5": 0.91274, "loss_box_5": 1.59421, "loss_cns_5": 0.6553, "loss_yns_5": 0.14956, "loss_cls_dn_0": 0.21524, "loss_box_dn_0": 0.75636, "loss_cls_dn_1": 0.13433, "loss_box_dn_1": 0.7087, "loss_cls_dn_2": 0.14131, "loss_box_dn_2": 0.69263, "loss_cls_dn_3": 0.1382, "loss_box_dn_3": 0.68932, "loss_cls_dn_4": 0.14113, "loss_box_dn_4": 0.68928, "loss_cls_dn_5": 0.14265, "loss_box_dn_5": 0.69032, "loss_dense_depth": 0.77353, "loss": 25.85605, "grad_norm": 36.07742, "time": 3.32215} -{"mode": "train", "epoch": 1, "iter": 224, "lr": 0.00019, "memory": 49164, "data_time": 0.07058, "loss_cls_0": 0.77858, "loss_box_0": 1.70128, "loss_cns_0": 0.63045, "loss_yns_0": 0.15326, "loss_cls_1": 0.8664, "loss_box_1": 1.63415, "loss_cns_1": 0.65383, "loss_yns_1": 0.15113, "loss_cls_2": 0.87512, "loss_box_2": 1.59608, "loss_cns_2": 0.65621, "loss_yns_2": 0.15112, "loss_cls_3": 0.88054, "loss_box_3": 1.59956, "loss_cns_3": 0.66002, "loss_yns_3": 0.14988, "loss_cls_4": 0.88145, "loss_box_4": 1.57801, "loss_cns_4": 0.65562, "loss_yns_4": 0.15167, "loss_cls_5": 0.90935, "loss_box_5": 1.57584, "loss_cns_5": 0.6556, "loss_yns_5": 0.15085, "loss_cls_dn_0": 0.20385, "loss_box_dn_0": 0.75462, "loss_cls_dn_1": 0.1254, "loss_box_dn_1": 0.70334, "loss_cls_dn_2": 0.12561, "loss_box_dn_2": 0.68705, "loss_cls_dn_3": 0.12452, "loss_box_dn_3": 0.68762, "loss_cls_dn_4": 0.12697, "loss_box_dn_4": 0.68722, "loss_cls_dn_5": 0.13195, "loss_box_dn_5": 0.69151, "loss_dense_depth": 0.74571, "loss": 25.4914, "grad_norm": 33.87122, "time": 1.56499} -{"mode": "train", "epoch": 1, "iter": 225, "lr": 0.00019, "memory": 49164, "data_time": 0.07955, "loss_cls_0": 0.77904, "loss_box_0": 1.71006, "loss_cns_0": 0.62742, "loss_yns_0": 0.15272, "loss_cls_1": 0.85758, "loss_box_1": 1.66596, "loss_cns_1": 0.65176, "loss_yns_1": 0.15142, "loss_cls_2": 0.88995, "loss_box_2": 1.62406, "loss_cns_2": 0.65355, "loss_yns_2": 0.15208, "loss_cls_3": 0.89181, "loss_box_3": 1.6345, "loss_cns_3": 0.65985, "loss_yns_3": 0.15181, "loss_cls_4": 0.89251, "loss_box_4": 1.61211, "loss_cns_4": 0.65526, "loss_yns_4": 0.15137, "loss_cls_5": 0.89004, "loss_box_5": 1.60634, "loss_cns_5": 0.65868, "loss_yns_5": 0.15133, "loss_cls_dn_0": 0.20446, "loss_box_dn_0": 0.7449, "loss_cls_dn_1": 0.12186, "loss_box_dn_1": 0.68962, "loss_cls_dn_2": 0.12452, "loss_box_dn_2": 0.67442, "loss_cls_dn_3": 0.12671, "loss_box_dn_3": 0.67937, "loss_cls_dn_4": 0.13274, "loss_box_dn_4": 0.6806, "loss_cls_dn_5": 0.13157, "loss_box_dn_5": 0.67853, "loss_dense_depth": 0.77827, "loss": 25.63876, "grad_norm": 30.61879, "time": 1.57215} -{"mode": "train", "epoch": 1, "iter": 226, "lr": 0.00019, "memory": 49164, "data_time": 0.09767, "loss_cls_0": 0.80367, "loss_box_0": 1.69156, "loss_cns_0": 0.62549, "loss_yns_0": 0.14914, "loss_cls_1": 0.87307, "loss_box_1": 1.67389, "loss_cns_1": 0.65363, "loss_yns_1": 0.15095, "loss_cls_2": 0.89909, "loss_box_2": 1.62064, "loss_cns_2": 0.65296, "loss_yns_2": 0.14798, "loss_cls_3": 0.91232, "loss_box_3": 1.61214, "loss_cns_3": 0.65541, "loss_yns_3": 0.14796, "loss_cls_4": 0.91675, "loss_box_4": 1.61188, "loss_cns_4": 0.65494, "loss_yns_4": 0.14939, "loss_cls_5": 0.90756, "loss_box_5": 1.61812, "loss_cns_5": 0.65556, "loss_yns_5": 0.14906, "loss_cls_dn_0": 0.20539, "loss_box_dn_0": 0.748, "loss_cls_dn_1": 0.1252, "loss_box_dn_1": 0.68294, "loss_cls_dn_2": 0.12574, "loss_box_dn_2": 0.66785, "loss_cls_dn_3": 0.12964, "loss_box_dn_3": 0.66743, "loss_cls_dn_4": 0.13171, "loss_box_dn_4": 0.67672, "loss_cls_dn_5": 0.13291, "loss_box_dn_5": 0.67467, "loss_dense_depth": 0.76698, "loss": 25.66835, "grad_norm": 32.53785, "time": 3.45563} -{"mode": "train", "epoch": 1, "iter": 227, "lr": 0.00019, "memory": 49164, "data_time": 0.06941, "loss_cls_0": 0.79227, "loss_box_0": 1.69856, "loss_cns_0": 0.617, "loss_yns_0": 0.1467, "loss_cls_1": 0.87585, "loss_box_1": 1.63127, "loss_cns_1": 0.6541, "loss_yns_1": 0.14629, "loss_cls_2": 0.88775, "loss_box_2": 1.60112, "loss_cns_2": 0.65465, "loss_yns_2": 0.14743, "loss_cls_3": 0.89248, "loss_box_3": 1.59145, "loss_cns_3": 0.66134, "loss_yns_3": 0.14683, "loss_cls_4": 0.89384, "loss_box_4": 1.59563, "loss_cns_4": 0.65903, "loss_yns_4": 0.15221, "loss_cls_5": 0.90761, "loss_box_5": 1.59828, "loss_cns_5": 0.65727, "loss_yns_5": 0.14778, "loss_cls_dn_0": 0.2064, "loss_box_dn_0": 0.74724, "loss_cls_dn_1": 0.12163, "loss_box_dn_1": 0.6838, "loss_cls_dn_2": 0.12303, "loss_box_dn_2": 0.67097, "loss_cls_dn_3": 0.12756, "loss_box_dn_3": 0.67312, "loss_cls_dn_4": 0.12485, "loss_box_dn_4": 0.67771, "loss_cls_dn_5": 0.12946, "loss_box_dn_5": 0.68141, "loss_dense_depth": 0.76298, "loss": 25.48693, "grad_norm": 33.44994, "time": 1.56529} -{"mode": "train", "epoch": 1, "iter": 228, "lr": 0.00019, "memory": 49164, "data_time": 0.07618, "loss_cls_0": 0.78464, "loss_box_0": 1.70777, "loss_cns_0": 0.6195, "loss_yns_0": 0.14746, "loss_cls_1": 0.87896, "loss_box_1": 1.62111, "loss_cns_1": 0.65683, "loss_yns_1": 0.14702, "loss_cls_2": 0.88493, "loss_box_2": 1.6222, "loss_cns_2": 0.65704, "loss_yns_2": 0.14688, "loss_cls_3": 0.89331, "loss_box_3": 1.60096, "loss_cns_3": 0.6648, "loss_yns_3": 0.14776, "loss_cls_4": 0.89588, "loss_box_4": 1.60942, "loss_cns_4": 0.66029, "loss_yns_4": 0.14893, "loss_cls_5": 0.89983, "loss_box_5": 1.60554, "loss_cns_5": 0.66128, "loss_yns_5": 0.1476, "loss_cls_dn_0": 0.21081, "loss_box_dn_0": 0.75156, "loss_cls_dn_1": 0.11919, "loss_box_dn_1": 0.6843, "loss_cls_dn_2": 0.12022, "loss_box_dn_2": 0.68276, "loss_cls_dn_3": 0.12499, "loss_box_dn_3": 0.68217, "loss_cls_dn_4": 0.12357, "loss_box_dn_4": 0.68589, "loss_cls_dn_5": 0.13143, "loss_box_dn_5": 0.69575, "loss_dense_depth": 0.74595, "loss": 25.56852, "grad_norm": 43.52741, "time": 1.57109} -{"mode": "train", "epoch": 1, "iter": 229, "lr": 0.00019, "memory": 49164, "data_time": 0.08271, "loss_cls_0": 0.81048, "loss_box_0": 1.69992, "loss_cns_0": 0.62107, "loss_yns_0": 0.14599, "loss_cls_1": 0.8701, "loss_box_1": 1.63423, "loss_cns_1": 0.65425, "loss_yns_1": 0.14766, "loss_cls_2": 0.87896, "loss_box_2": 1.6331, "loss_cns_2": 0.65786, "loss_yns_2": 0.14352, "loss_cls_3": 0.92191, "loss_box_3": 1.62903, "loss_cns_3": 0.65808, "loss_yns_3": 0.14356, "loss_cls_4": 0.89947, "loss_box_4": 1.60874, "loss_cns_4": 0.65818, "loss_yns_4": 0.14452, "loss_cls_5": 0.89477, "loss_box_5": 1.64044, "loss_cns_5": 0.6556, "loss_yns_5": 0.14316, "loss_cls_dn_0": 0.20357, "loss_box_dn_0": 0.74044, "loss_cls_dn_1": 0.11979, "loss_box_dn_1": 0.69311, "loss_cls_dn_2": 0.12031, "loss_box_dn_2": 0.69197, "loss_cls_dn_3": 0.1228, "loss_box_dn_3": 0.69759, "loss_cls_dn_4": 0.12244, "loss_box_dn_4": 0.69777, "loss_cls_dn_5": 0.13003, "loss_box_dn_5": 0.72214, "loss_dense_depth": 0.7787, "loss": 25.73527, "grad_norm": 42.57179, "time": 1.57688} -{"mode": "train", "epoch": 1, "iter": 230, "lr": 0.00019, "memory": 49164, "data_time": 0.09541, "loss_cls_0": 0.79871, "loss_box_0": 1.71367, "loss_cns_0": 0.62145, "loss_yns_0": 0.14756, "loss_cls_1": 0.86712, "loss_box_1": 1.66087, "loss_cns_1": 0.64183, "loss_yns_1": 0.14628, "loss_cls_2": 0.88078, "loss_box_2": 1.647, "loss_cns_2": 0.65082, "loss_yns_2": 0.14334, "loss_cls_3": 0.88621, "loss_box_3": 1.6432, "loss_cns_3": 0.65305, "loss_yns_3": 0.1447, "loss_cls_4": 0.90226, "loss_box_4": 1.62707, "loss_cns_4": 0.66134, "loss_yns_4": 0.14915, "loss_cls_5": 0.91862, "loss_box_5": 1.62249, "loss_cns_5": 0.6521, "loss_yns_5": 0.14627, "loss_cls_dn_0": 0.20564, "loss_box_dn_0": 0.7497, "loss_cls_dn_1": 0.12039, "loss_box_dn_1": 0.70693, "loss_cls_dn_2": 0.11827, "loss_box_dn_2": 0.69927, "loss_cls_dn_3": 0.11904, "loss_box_dn_3": 0.70419, "loss_cls_dn_4": 0.11757, "loss_box_dn_4": 0.70601, "loss_cls_dn_5": 0.12293, "loss_box_dn_5": 0.7211, "loss_dense_depth": 0.73842, "loss": 25.75536, "grad_norm": 42.16234, "time": 1.59919} -{"mode": "train", "epoch": 1, "iter": 231, "lr": 0.00019, "memory": 49164, "data_time": 0.07927, "loss_cls_0": 0.76893, "loss_box_0": 1.69595, "loss_cns_0": 0.62705, "loss_yns_0": 0.147, "loss_cls_1": 0.84323, "loss_box_1": 1.5863, "loss_cns_1": 0.64828, "loss_yns_1": 0.14391, "loss_cls_2": 0.85298, "loss_box_2": 1.58124, "loss_cns_2": 0.65499, "loss_yns_2": 0.14437, "loss_cls_3": 0.86339, "loss_box_3": 1.5784, "loss_cns_3": 0.65802, "loss_yns_3": 0.14462, "loss_cls_4": 0.87955, "loss_box_4": 1.58714, "loss_cns_4": 0.66623, "loss_yns_4": 0.148, "loss_cls_5": 0.87911, "loss_box_5": 1.57644, "loss_cns_5": 0.65776, "loss_yns_5": 0.14566, "loss_cls_dn_0": 0.19706, "loss_box_dn_0": 0.74575, "loss_cls_dn_1": 0.11373, "loss_box_dn_1": 0.69293, "loss_cls_dn_2": 0.11446, "loss_box_dn_2": 0.6941, "loss_cls_dn_3": 0.11586, "loss_box_dn_3": 0.70024, "loss_cls_dn_4": 0.11974, "loss_box_dn_4": 0.7099, "loss_cls_dn_5": 0.12594, "loss_box_dn_5": 0.71705, "loss_dense_depth": 0.72942, "loss": 25.25474, "grad_norm": 51.78098, "time": 3.34165} -{"mode": "train", "epoch": 1, "iter": 232, "lr": 0.00019, "memory": 49164, "data_time": 0.07568, "loss_cls_0": 0.7773, "loss_box_0": 1.69344, "loss_cns_0": 0.62858, "loss_yns_0": 0.14748, "loss_cls_1": 0.85664, "loss_box_1": 1.62302, "loss_cns_1": 0.65691, "loss_yns_1": 0.14449, "loss_cls_2": 0.88166, "loss_box_2": 1.60613, "loss_cns_2": 0.66049, "loss_yns_2": 0.14443, "loss_cls_3": 0.88265, "loss_box_3": 1.59891, "loss_cns_3": 0.66161, "loss_yns_3": 0.14574, "loss_cls_4": 0.88914, "loss_box_4": 1.58, "loss_cns_4": 0.66255, "loss_yns_4": 0.14609, "loss_cls_5": 0.89197, "loss_box_5": 1.58437, "loss_cns_5": 0.66253, "loss_yns_5": 0.14468, "loss_cls_dn_0": 0.19847, "loss_box_dn_0": 0.75949, "loss_cls_dn_1": 0.11593, "loss_box_dn_1": 0.6936, "loss_cls_dn_2": 0.11978, "loss_box_dn_2": 0.68417, "loss_cls_dn_3": 0.12025, "loss_box_dn_3": 0.68726, "loss_cls_dn_4": 0.12354, "loss_box_dn_4": 0.68579, "loss_cls_dn_5": 0.12915, "loss_box_dn_5": 0.69163, "loss_dense_depth": 0.74759, "loss": 25.42746, "grad_norm": 39.64152, "time": 1.56214} -{"mode": "train", "epoch": 1, "iter": 233, "lr": 0.00019, "memory": 49164, "data_time": 0.07745, "loss_cls_0": 0.7885, "loss_box_0": 1.67757, "loss_cns_0": 0.62377, "loss_yns_0": 0.14719, "loss_cls_1": 0.8919, "loss_box_1": 1.57785, "loss_cns_1": 0.66193, "loss_yns_1": 0.14734, "loss_cls_2": 0.90251, "loss_box_2": 1.54734, "loss_cns_2": 0.66243, "loss_yns_2": 0.14648, "loss_cls_3": 0.91098, "loss_box_3": 1.54466, "loss_cns_3": 0.66185, "loss_yns_3": 0.14636, "loss_cls_4": 0.90192, "loss_box_4": 1.54016, "loss_cns_4": 0.66426, "loss_yns_4": 0.14649, "loss_cls_5": 0.90891, "loss_box_5": 1.53178, "loss_cns_5": 0.66307, "loss_yns_5": 0.14656, "loss_cls_dn_0": 0.20097, "loss_box_dn_0": 0.76232, "loss_cls_dn_1": 0.12042, "loss_box_dn_1": 0.68691, "loss_cls_dn_2": 0.12147, "loss_box_dn_2": 0.67165, "loss_cls_dn_3": 0.12133, "loss_box_dn_3": 0.67422, "loss_cls_dn_4": 0.12273, "loss_box_dn_4": 0.67305, "loss_cls_dn_5": 0.12498, "loss_box_dn_5": 0.67355, "loss_dense_depth": 0.74403, "loss": 25.23946, "grad_norm": 41.52663, "time": 1.56349} -{"mode": "train", "epoch": 1, "iter": 234, "lr": 0.00019, "memory": 49164, "data_time": 0.0784, "loss_cls_0": 0.77801, "loss_box_0": 1.64977, "loss_cns_0": 0.62419, "loss_yns_0": 0.14766, "loss_cls_1": 0.87607, "loss_box_1": 1.56339, "loss_cns_1": 0.66339, "loss_yns_1": 0.14933, "loss_cls_2": 0.88795, "loss_box_2": 1.52415, "loss_cns_2": 0.66286, "loss_yns_2": 0.14732, "loss_cls_3": 0.9072, "loss_box_3": 1.52237, "loss_cns_3": 0.66162, "loss_yns_3": 0.14724, "loss_cls_4": 0.90408, "loss_box_4": 1.52548, "loss_cns_4": 0.66317, "loss_yns_4": 0.14799, "loss_cls_5": 0.89847, "loss_box_5": 1.51707, "loss_cns_5": 0.66146, "loss_yns_5": 0.14846, "loss_cls_dn_0": 0.19455, "loss_box_dn_0": 0.74405, "loss_cls_dn_1": 0.11233, "loss_box_dn_1": 0.68778, "loss_cls_dn_2": 0.11262, "loss_box_dn_2": 0.67428, "loss_cls_dn_3": 0.11435, "loss_box_dn_3": 0.68114, "loss_cls_dn_4": 0.11744, "loss_box_dn_4": 0.69273, "loss_cls_dn_5": 0.11973, "loss_box_dn_5": 0.69192, "loss_dense_depth": 0.73813, "loss": 25.05976, "grad_norm": 43.40402, "time": 1.56814} -{"mode": "train", "epoch": 1, "iter": 235, "lr": 0.00019, "memory": 49164, "data_time": 0.0782, "loss_cls_0": 0.7943, "loss_box_0": 1.66968, "loss_cns_0": 0.62157, "loss_yns_0": 0.14797, "loss_cls_1": 0.87526, "loss_box_1": 1.60599, "loss_cns_1": 0.6592, "loss_yns_1": 0.14955, "loss_cls_2": 0.89044, "loss_box_2": 1.56443, "loss_cns_2": 0.65824, "loss_yns_2": 0.14847, "loss_cls_3": 0.9033, "loss_box_3": 1.55014, "loss_cns_3": 0.65849, "loss_yns_3": 0.14886, "loss_cls_4": 0.91999, "loss_box_4": 1.53151, "loss_cns_4": 0.65559, "loss_yns_4": 0.14769, "loss_cls_5": 0.91776, "loss_box_5": 1.54102, "loss_cns_5": 0.65973, "loss_yns_5": 0.14734, "loss_cls_dn_0": 0.19835, "loss_box_dn_0": 0.74171, "loss_cls_dn_1": 0.1153, "loss_box_dn_1": 0.69316, "loss_cls_dn_2": 0.11481, "loss_box_dn_2": 0.68142, "loss_cls_dn_3": 0.11748, "loss_box_dn_3": 0.68041, "loss_cls_dn_4": 0.12176, "loss_box_dn_4": 0.68564, "loss_cls_dn_5": 0.12516, "loss_box_dn_5": 0.69418, "loss_dense_depth": 0.73916, "loss": 25.27506, "grad_norm": 41.33026, "time": 1.58132} -{"mode": "train", "epoch": 1, "iter": 236, "lr": 0.00019, "memory": 49164, "data_time": 0.07406, "loss_cls_0": 0.76236, "loss_box_0": 1.67371, "loss_cns_0": 0.62543, "loss_yns_0": 0.15023, "loss_cls_1": 0.88063, "loss_box_1": 1.59071, "loss_cns_1": 0.65794, "loss_yns_1": 0.15083, "loss_cls_2": 0.88449, "loss_box_2": 1.57387, "loss_cns_2": 0.6511, "loss_yns_2": 0.14792, "loss_cls_3": 0.89775, "loss_box_3": 1.57621, "loss_cns_3": 0.65331, "loss_yns_3": 0.14976, "loss_cls_4": 0.91883, "loss_box_4": 1.582, "loss_cns_4": 0.65008, "loss_yns_4": 0.14809, "loss_cls_5": 0.91989, "loss_box_5": 1.59099, "loss_cns_5": 0.65761, "loss_yns_5": 0.1509, "loss_cls_dn_0": 0.19137, "loss_box_dn_0": 0.7365, "loss_cls_dn_1": 0.1171, "loss_box_dn_1": 0.68043, "loss_cls_dn_2": 0.11663, "loss_box_dn_2": 0.68318, "loss_cls_dn_3": 0.12144, "loss_box_dn_3": 0.68469, "loss_cls_dn_4": 0.12628, "loss_box_dn_4": 0.7018, "loss_cls_dn_5": 0.13454, "loss_box_dn_5": 0.71824, "loss_dense_depth": 0.74006, "loss": 25.39692, "grad_norm": 66.07609, "time": 1.57088} -{"mode": "train", "epoch": 1, "iter": 237, "lr": 0.00019, "memory": 49164, "data_time": 0.0768, "loss_cls_0": 0.76616, "loss_box_0": 1.65092, "loss_cns_0": 0.62744, "loss_yns_0": 0.14927, "loss_cls_1": 0.86186, "loss_box_1": 1.60143, "loss_cns_1": 0.6596, "loss_yns_1": 0.15204, "loss_cls_2": 0.88548, "loss_box_2": 1.5747, "loss_cns_2": 0.65303, "loss_yns_2": 0.1489, "loss_cls_3": 0.89933, "loss_box_3": 1.56594, "loss_cns_3": 0.65637, "loss_yns_3": 0.15085, "loss_cls_4": 0.90753, "loss_box_4": 1.59333, "loss_cns_4": 0.65807, "loss_yns_4": 0.15065, "loss_cls_5": 0.91641, "loss_box_5": 1.59161, "loss_cns_5": 0.65852, "loss_yns_5": 0.15299, "loss_cls_dn_0": 0.19489, "loss_box_dn_0": 0.74469, "loss_cls_dn_1": 0.11941, "loss_box_dn_1": 0.70502, "loss_cls_dn_2": 0.12464, "loss_box_dn_2": 0.71579, "loss_cls_dn_3": 0.13374, "loss_box_dn_3": 0.71133, "loss_cls_dn_4": 0.1346, "loss_box_dn_4": 0.73364, "loss_cls_dn_5": 0.1497, "loss_box_dn_5": 0.74429, "loss_dense_depth": 0.72179, "loss": 25.56595, "grad_norm": 77.85015, "time": 1.57683} -{"mode": "train", "epoch": 1, "iter": 238, "lr": 0.00019, "memory": 49164, "data_time": 0.08335, "loss_cls_0": 0.77117, "loss_box_0": 1.64097, "loss_cns_0": 0.62808, "loss_yns_0": 0.15149, "loss_cls_1": 0.86208, "loss_box_1": 1.60342, "loss_cns_1": 0.65835, "loss_yns_1": 0.15338, "loss_cls_2": 0.88677, "loss_box_2": 1.54858, "loss_cns_2": 0.65841, "loss_yns_2": 0.15094, "loss_cls_3": 0.88079, "loss_box_3": 1.55579, "loss_cns_3": 0.6599, "loss_yns_3": 0.15321, "loss_cls_4": 0.87911, "loss_box_4": 1.5459, "loss_cns_4": 0.66185, "loss_yns_4": 0.15232, "loss_cls_5": 0.89419, "loss_box_5": 1.54979, "loss_cns_5": 0.66157, "loss_yns_5": 0.15472, "loss_cls_dn_0": 0.19169, "loss_box_dn_0": 0.74509, "loss_cls_dn_1": 0.12135, "loss_box_dn_1": 0.72197, "loss_cls_dn_2": 0.12428, "loss_box_dn_2": 0.71272, "loss_cls_dn_3": 0.12691, "loss_box_dn_3": 0.70897, "loss_cls_dn_4": 0.12469, "loss_box_dn_4": 0.70676, "loss_cls_dn_5": 0.13295, "loss_box_dn_5": 0.7102, "loss_dense_depth": 0.70242, "loss": 25.29279, "grad_norm": 42.5807, "time": 1.58946} -{"mode": "train", "epoch": 1, "iter": 239, "lr": 0.0002, "memory": 49164, "data_time": 0.07367, "loss_cls_0": 0.78145, "loss_box_0": 1.61457, "loss_cns_0": 0.63018, "loss_yns_0": 0.15062, "loss_cls_1": 0.86376, "loss_box_1": 1.53205, "loss_cns_1": 0.65982, "loss_yns_1": 0.15224, "loss_cls_2": 0.88607, "loss_box_2": 1.47696, "loss_cns_2": 0.66373, "loss_yns_2": 0.15113, "loss_cls_3": 0.89575, "loss_box_3": 1.54165, "loss_cns_3": 0.6638, "loss_yns_3": 0.15126, "loss_cls_4": 0.90848, "loss_box_4": 1.48284, "loss_cns_4": 0.66454, "loss_yns_4": 0.15233, "loss_cls_5": 0.89531, "loss_box_5": 1.47702, "loss_cns_5": 0.66792, "loss_yns_5": 0.15214, "loss_cls_dn_0": 0.19434, "loss_box_dn_0": 0.7355, "loss_cls_dn_1": 0.11991, "loss_box_dn_1": 0.70066, "loss_cls_dn_2": 0.12503, "loss_box_dn_2": 0.68175, "loss_cls_dn_3": 0.12224, "loss_box_dn_3": 0.70778, "loss_cls_dn_4": 0.12571, "loss_box_dn_4": 0.68905, "loss_cls_dn_5": 0.12947, "loss_box_dn_5": 0.68569, "loss_dense_depth": 0.68159, "loss": 24.91435, "grad_norm": 49.01508, "time": 1.58115} -{"mode": "train", "epoch": 1, "iter": 240, "lr": 0.0002, "memory": 49164, "data_time": 0.07081, "loss_cls_0": 0.76793, "loss_box_0": 1.66405, "loss_cns_0": 0.62446, "loss_yns_0": 0.15175, "loss_cls_1": 0.88584, "loss_box_1": 1.51588, "loss_cns_1": 0.66442, "loss_yns_1": 0.15085, "loss_cls_2": 0.9018, "loss_box_2": 1.49523, "loss_cns_2": 0.66506, "loss_yns_2": 0.15062, "loss_cls_3": 0.91185, "loss_box_3": 1.52369, "loss_cns_3": 0.66935, "loss_yns_3": 0.15155, "loss_cls_4": 0.91682, "loss_box_4": 1.47577, "loss_cns_4": 0.6661, "loss_yns_4": 0.15185, "loss_cls_5": 0.90129, "loss_box_5": 1.48078, "loss_cns_5": 0.67115, "loss_yns_5": 0.15255, "loss_cls_dn_0": 0.19635, "loss_box_dn_0": 0.75187, "loss_cls_dn_1": 0.1201, "loss_box_dn_1": 0.68174, "loss_cls_dn_2": 0.12674, "loss_box_dn_2": 0.67729, "loss_cls_dn_3": 0.12495, "loss_box_dn_3": 0.69656, "loss_cls_dn_4": 0.13109, "loss_box_dn_4": 0.67994, "loss_cls_dn_5": 0.13312, "loss_box_dn_5": 0.68182, "loss_dense_depth": 0.69361, "loss": 25.00583, "grad_norm": 55.83244, "time": 1.59074} -{"mode": "train", "epoch": 1, "iter": 241, "lr": 0.0002, "memory": 49164, "data_time": 0.09939, "loss_cls_0": 0.78464, "loss_box_0": 1.66152, "loss_cns_0": 0.62392, "loss_yns_0": 0.15488, "loss_cls_1": 0.89197, "loss_box_1": 1.5581, "loss_cns_1": 0.66164, "loss_yns_1": 0.15579, "loss_cls_2": 0.91115, "loss_box_2": 1.53746, "loss_cns_2": 0.66368, "loss_yns_2": 0.15566, "loss_cls_3": 0.90569, "loss_box_3": 1.53624, "loss_cns_3": 0.66733, "loss_yns_3": 0.15609, "loss_cls_4": 0.91121, "loss_box_4": 1.52095, "loss_cns_4": 0.66409, "loss_yns_4": 0.156, "loss_cls_5": 0.9081, "loss_box_5": 1.52836, "loss_cns_5": 0.66751, "loss_yns_5": 0.15778, "loss_cls_dn_0": 0.1945, "loss_box_dn_0": 0.74608, "loss_cls_dn_1": 0.12115, "loss_box_dn_1": 0.68896, "loss_cls_dn_2": 0.1269, "loss_box_dn_2": 0.68106, "loss_cls_dn_3": 0.12709, "loss_box_dn_3": 0.68607, "loss_cls_dn_4": 0.1284, "loss_box_dn_4": 0.67751, "loss_cls_dn_5": 0.13004, "loss_box_dn_5": 0.68541, "loss_dense_depth": 0.72273, "loss": 25.25567, "grad_norm": 44.84576, "time": 1.65079} -{"mode": "train", "epoch": 1, "iter": 242, "lr": 0.0002, "memory": 49164, "data_time": 0.10111, "loss_cls_0": 0.79246, "loss_box_0": 1.61327, "loss_cns_0": 0.62622, "loss_yns_0": 0.15149, "loss_cls_1": 0.89161, "loss_box_1": 1.50195, "loss_cns_1": 0.67053, "loss_yns_1": 0.15455, "loss_cls_2": 0.90375, "loss_box_2": 1.49128, "loss_cns_2": 0.66647, "loss_yns_2": 0.15526, "loss_cls_3": 0.89807, "loss_box_3": 1.51482, "loss_cns_3": 0.66533, "loss_yns_3": 0.15473, "loss_cls_4": 0.93445, "loss_box_4": 1.51029, "loss_cns_4": 0.66453, "loss_yns_4": 0.15637, "loss_cls_5": 0.90793, "loss_box_5": 1.50477, "loss_cns_5": 0.66193, "loss_yns_5": 0.15719, "loss_cls_dn_0": 0.19213, "loss_box_dn_0": 0.74227, "loss_cls_dn_1": 0.12033, "loss_box_dn_1": 0.67903, "loss_cls_dn_2": 0.12241, "loss_box_dn_2": 0.66754, "loss_cls_dn_3": 0.12471, "loss_box_dn_3": 0.68168, "loss_cls_dn_4": 0.12841, "loss_box_dn_4": 0.68448, "loss_cls_dn_5": 0.13099, "loss_box_dn_5": 0.68826, "loss_dense_depth": 0.68469, "loss": 24.99617, "grad_norm": 42.07364, "time": 1.62005} -{"mode": "train", "epoch": 1, "iter": 243, "lr": 0.0002, "memory": 49164, "data_time": 0.07871, "loss_cls_0": 0.8065, "loss_box_0": 1.62347, "loss_cns_0": 0.62239, "loss_yns_0": 0.15186, "loss_cls_1": 0.89208, "loss_box_1": 1.57528, "loss_cns_1": 0.65632, "loss_yns_1": 0.15178, "loss_cls_2": 0.90025, "loss_box_2": 1.56154, "loss_cns_2": 0.65965, "loss_yns_2": 0.15148, "loss_cls_3": 0.90092, "loss_box_3": 1.5383, "loss_cns_3": 0.66214, "loss_yns_3": 0.15043, "loss_cls_4": 0.93448, "loss_box_4": 1.53182, "loss_cns_4": 0.65997, "loss_yns_4": 0.15187, "loss_cls_5": 0.90227, "loss_box_5": 1.52653, "loss_cns_5": 0.66318, "loss_yns_5": 0.15174, "loss_cls_dn_0": 0.19849, "loss_box_dn_0": 0.74439, "loss_cls_dn_1": 0.12246, "loss_box_dn_1": 0.69584, "loss_cls_dn_2": 0.12432, "loss_box_dn_2": 0.68333, "loss_cls_dn_3": 0.12723, "loss_box_dn_3": 0.68581, "loss_cls_dn_4": 0.13628, "loss_box_dn_4": 0.69174, "loss_cls_dn_5": 0.13662, "loss_box_dn_5": 0.69706, "loss_dense_depth": 0.70241, "loss": 25.27222, "grad_norm": 43.2645, "time": 1.59905} -{"mode": "train", "epoch": 1, "iter": 244, "lr": 0.0002, "memory": 49164, "data_time": 0.07877, "loss_cls_0": 0.78963, "loss_box_0": 1.64847, "loss_cns_0": 0.62418, "loss_yns_0": 0.15365, "loss_cls_1": 0.86612, "loss_box_1": 1.59979, "loss_cns_1": 0.65489, "loss_yns_1": 0.15272, "loss_cls_2": 0.88772, "loss_box_2": 1.55743, "loss_cns_2": 0.65513, "loss_yns_2": 0.15214, "loss_cls_3": 0.8988, "loss_box_3": 1.56491, "loss_cns_3": 0.65652, "loss_yns_3": 0.1526, "loss_cls_4": 0.92071, "loss_box_4": 1.56875, "loss_cns_4": 0.65554, "loss_yns_4": 0.15148, "loss_cls_5": 0.91025, "loss_box_5": 1.57541, "loss_cns_5": 0.66141, "loss_yns_5": 0.15166, "loss_cls_dn_0": 0.19581, "loss_box_dn_0": 0.73871, "loss_cls_dn_1": 0.12431, "loss_box_dn_1": 0.69935, "loss_cls_dn_2": 0.12454, "loss_box_dn_2": 0.68183, "loss_cls_dn_3": 0.1243, "loss_box_dn_3": 0.68824, "loss_cls_dn_4": 0.13177, "loss_box_dn_4": 0.69018, "loss_cls_dn_5": 0.13011, "loss_box_dn_5": 0.69913, "loss_dense_depth": 0.72402, "loss": 25.3622, "grad_norm": 43.58756, "time": 1.57612} -{"mode": "train", "epoch": 1, "iter": 245, "lr": 0.0002, "memory": 49164, "data_time": 0.08499, "loss_cls_0": 0.7775, "loss_box_0": 1.66865, "loss_cns_0": 0.62396, "loss_yns_0": 0.15156, "loss_cls_1": 0.83624, "loss_box_1": 1.55117, "loss_cns_1": 0.65613, "loss_yns_1": 0.15208, "loss_cls_2": 0.86607, "loss_box_2": 1.52301, "loss_cns_2": 0.6563, "loss_yns_2": 0.15332, "loss_cls_3": 0.86992, "loss_box_3": 1.52929, "loss_cns_3": 0.65507, "loss_yns_3": 0.15165, "loss_cls_4": 0.88365, "loss_box_4": 1.50925, "loss_cns_4": 0.66016, "loss_yns_4": 0.15163, "loss_cls_5": 0.8806, "loss_box_5": 1.51878, "loss_cns_5": 0.66116, "loss_yns_5": 0.15429, "loss_cls_dn_0": 0.19888, "loss_box_dn_0": 0.73302, "loss_cls_dn_1": 0.12926, "loss_box_dn_1": 0.68783, "loss_cls_dn_2": 0.12878, "loss_box_dn_2": 0.67455, "loss_cls_dn_3": 0.12692, "loss_box_dn_3": 0.6778, "loss_cls_dn_4": 0.12896, "loss_box_dn_4": 0.66746, "loss_cls_dn_5": 0.12917, "loss_box_dn_5": 0.67245, "loss_dense_depth": 0.69588, "loss": 24.89238, "grad_norm": 36.59685, "time": 1.59944} -{"mode": "train", "epoch": 1, "iter": 246, "lr": 0.0002, "memory": 49164, "data_time": 0.09528, "loss_cls_0": 0.77035, "loss_box_0": 1.68296, "loss_cns_0": 0.62253, "loss_yns_0": 0.15284, "loss_cls_1": 0.84735, "loss_box_1": 1.57664, "loss_cns_1": 0.65704, "loss_yns_1": 0.15196, "loss_cls_2": 0.86727, "loss_box_2": 1.55634, "loss_cns_2": 0.65648, "loss_yns_2": 0.15272, "loss_cls_3": 0.87186, "loss_box_3": 1.56425, "loss_cns_3": 0.65767, "loss_yns_3": 0.1523, "loss_cls_4": 0.88559, "loss_box_4": 1.54869, "loss_cns_4": 0.65882, "loss_yns_4": 0.15296, "loss_cls_5": 0.87687, "loss_box_5": 1.54677, "loss_cns_5": 0.65721, "loss_yns_5": 0.15466, "loss_cls_dn_0": 0.1951, "loss_box_dn_0": 0.74115, "loss_cls_dn_1": 0.12202, "loss_box_dn_1": 0.68332, "loss_cls_dn_2": 0.12238, "loss_box_dn_2": 0.67158, "loss_cls_dn_3": 0.12425, "loss_box_dn_3": 0.6722, "loss_cls_dn_4": 0.12852, "loss_box_dn_4": 0.66642, "loss_cls_dn_5": 0.12939, "loss_box_dn_5": 0.6666, "loss_dense_depth": 0.7436, "loss": 25.08867, "grad_norm": 48.24923, "time": 1.58778} -{"mode": "train", "epoch": 1, "iter": 247, "lr": 0.0002, "memory": 49164, "data_time": 0.07301, "loss_cls_0": 0.80497, "loss_box_0": 1.67909, "loss_cns_0": 0.62726, "loss_yns_0": 0.15698, "loss_cls_1": 0.87598, "loss_box_1": 1.56073, "loss_cns_1": 0.6598, "loss_yns_1": 0.15515, "loss_cls_2": 0.88427, "loss_box_2": 1.55556, "loss_cns_2": 0.66083, "loss_yns_2": 0.15453, "loss_cls_3": 0.88616, "loss_box_3": 1.56306, "loss_cns_3": 0.66766, "loss_yns_3": 0.15459, "loss_cls_4": 0.91694, "loss_box_4": 1.56938, "loss_cns_4": 0.66351, "loss_yns_4": 0.15511, "loss_cls_5": 0.89902, "loss_box_5": 1.56726, "loss_cns_5": 0.66189, "loss_yns_5": 0.15642, "loss_cls_dn_0": 0.19402, "loss_box_dn_0": 0.74054, "loss_cls_dn_1": 0.11781, "loss_box_dn_1": 0.66603, "loss_cls_dn_2": 0.11851, "loss_box_dn_2": 0.66084, "loss_cls_dn_3": 0.12225, "loss_box_dn_3": 0.66697, "loss_cls_dn_4": 0.12611, "loss_box_dn_4": 0.675, "loss_cls_dn_5": 0.12949, "loss_box_dn_5": 0.68224, "loss_dense_depth": 0.71108, "loss": 25.24704, "grad_norm": 37.35027, "time": 1.58885} -{"mode": "train", "epoch": 1, "iter": 248, "lr": 0.0002, "memory": 49164, "data_time": 0.08062, "loss_cls_0": 0.78885, "loss_box_0": 1.71072, "loss_cns_0": 0.62228, "loss_yns_0": 0.15501, "loss_cls_1": 0.85993, "loss_box_1": 1.58812, "loss_cns_1": 0.65588, "loss_yns_1": 0.15398, "loss_cls_2": 0.87948, "loss_box_2": 1.57575, "loss_cns_2": 0.65859, "loss_yns_2": 0.15419, "loss_cls_3": 0.89377, "loss_box_3": 1.56974, "loss_cns_3": 0.66892, "loss_yns_3": 0.15522, "loss_cls_4": 0.92694, "loss_box_4": 1.56907, "loss_cns_4": 0.66615, "loss_yns_4": 0.1561, "loss_cls_5": 0.89871, "loss_box_5": 1.56045, "loss_cns_5": 0.67097, "loss_yns_5": 0.15618, "loss_cls_dn_0": 0.19515, "loss_box_dn_0": 0.74387, "loss_cls_dn_1": 0.11769, "loss_box_dn_1": 0.67664, "loss_cls_dn_2": 0.12031, "loss_box_dn_2": 0.6774, "loss_cls_dn_3": 0.12216, "loss_box_dn_3": 0.68692, "loss_cls_dn_4": 0.12506, "loss_box_dn_4": 0.70531, "loss_cls_dn_5": 0.12611, "loss_box_dn_5": 0.71704, "loss_dense_depth": 0.75451, "loss": 25.46314, "grad_norm": 52.28632, "time": 1.58438} -{"mode": "train", "epoch": 1, "iter": 249, "lr": 0.0002, "memory": 49164, "data_time": 0.08256, "loss_cls_0": 0.81086, "loss_box_0": 1.69302, "loss_cns_0": 0.62809, "loss_yns_0": 0.15519, "loss_cls_1": 0.87737, "loss_box_1": 1.63299, "loss_cns_1": 0.66, "loss_yns_1": 0.15487, "loss_cls_2": 0.88685, "loss_box_2": 1.60103, "loss_cns_2": 0.66302, "loss_yns_2": 0.15503, "loss_cls_3": 0.90245, "loss_box_3": 1.58891, "loss_cns_3": 0.66807, "loss_yns_3": 0.15703, "loss_cls_4": 0.90954, "loss_box_4": 1.58325, "loss_cns_4": 0.66215, "loss_yns_4": 0.1563, "loss_cls_5": 0.90009, "loss_box_5": 1.56827, "loss_cns_5": 0.66807, "loss_yns_5": 0.15597, "loss_cls_dn_0": 0.20673, "loss_box_dn_0": 0.74358, "loss_cls_dn_1": 0.11974, "loss_box_dn_1": 0.69684, "loss_cls_dn_2": 0.11768, "loss_box_dn_2": 0.68761, "loss_cls_dn_3": 0.11951, "loss_box_dn_3": 0.6924, "loss_cls_dn_4": 0.12322, "loss_box_dn_4": 0.70511, "loss_cls_dn_5": 0.12323, "loss_box_dn_5": 0.71558, "loss_dense_depth": 0.69216, "loss": 25.5818, "grad_norm": 47.36179, "time": 1.59529} -{"mode": "train", "epoch": 1, "iter": 250, "lr": 0.0002, "memory": 49164, "data_time": 0.07995, "loss_cls_0": 0.77127, "loss_box_0": 1.66748, "loss_cns_0": 0.62423, "loss_yns_0": 0.15166, "loss_cls_1": 0.86458, "loss_box_1": 1.60084, "loss_cns_1": 0.66078, "loss_yns_1": 0.15154, "loss_cls_2": 0.8678, "loss_box_2": 1.59239, "loss_cns_2": 0.66184, "loss_yns_2": 0.15303, "loss_cls_3": 0.87659, "loss_box_3": 1.57509, "loss_cns_3": 0.66023, "loss_yns_3": 0.15422, "loss_cls_4": 0.89762, "loss_box_4": 1.55708, "loss_cns_4": 0.66271, "loss_yns_4": 0.15204, "loss_cls_5": 0.90488, "loss_box_5": 1.5549, "loss_cns_5": 0.65729, "loss_yns_5": 0.15479, "loss_cls_dn_0": 0.19492, "loss_box_dn_0": 0.73989, "loss_cls_dn_1": 0.12126, "loss_box_dn_1": 0.6958, "loss_cls_dn_2": 0.11628, "loss_box_dn_2": 0.68551, "loss_cls_dn_3": 0.11977, "loss_box_dn_3": 0.68637, "loss_cls_dn_4": 0.12664, "loss_box_dn_4": 0.69701, "loss_cls_dn_5": 0.13105, "loss_box_dn_5": 0.69719, "loss_dense_depth": 0.71577, "loss": 25.30235, "grad_norm": 46.86003, "time": 1.58744} -{"mode": "train", "epoch": 1, "iter": 251, "lr": 0.0002, "memory": 49164, "data_time": 0.07512, "loss_cls_0": 0.77698, "loss_box_0": 1.62704, "loss_cns_0": 0.62214, "loss_yns_0": 0.1496, "loss_cls_1": 0.85326, "loss_box_1": 1.5743, "loss_cns_1": 0.64639, "loss_yns_1": 0.14856, "loss_cls_2": 0.87566, "loss_box_2": 1.57906, "loss_cns_2": 0.65902, "loss_yns_2": 0.1506, "loss_cls_3": 0.88959, "loss_box_3": 1.57582, "loss_cns_3": 0.66102, "loss_yns_3": 0.15002, "loss_cls_4": 0.9464, "loss_box_4": 1.48995, "loss_cns_4": 0.65377, "loss_yns_4": 0.14606, "loss_cls_5": 0.92006, "loss_box_5": 1.53108, "loss_cns_5": 0.6572, "loss_yns_5": 0.15448, "loss_cls_dn_0": 0.19127, "loss_box_dn_0": 0.74569, "loss_cls_dn_1": 0.12164, "loss_box_dn_1": 0.68884, "loss_cls_dn_2": 0.12031, "loss_box_dn_2": 0.67729, "loss_cls_dn_3": 0.12294, "loss_box_dn_3": 0.6835, "loss_cls_dn_4": 0.1276, "loss_box_dn_4": 0.67734, "loss_cls_dn_5": 0.1329, "loss_box_dn_5": 0.67409, "loss_dense_depth": 0.75376, "loss": 25.15525, "grad_norm": 48.79861, "time": 1.61445} -{"mode": "train", "epoch": 1, "iter": 252, "lr": 0.0002, "memory": 49164, "data_time": 0.07509, "loss_cls_0": 0.75498, "loss_box_0": 1.66229, "loss_cns_0": 0.62367, "loss_yns_0": 0.15074, "loss_cls_1": 0.82945, "loss_box_1": 1.58664, "loss_cns_1": 0.65334, "loss_yns_1": 0.15255, "loss_cls_2": 0.87583, "loss_box_2": 1.56376, "loss_cns_2": 0.65824, "loss_yns_2": 0.15256, "loss_cls_3": 0.84998, "loss_box_3": 1.59029, "loss_cns_3": 0.66664, "loss_yns_3": 0.15064, "loss_cls_4": 0.87043, "loss_box_4": 1.54857, "loss_cns_4": 0.66219, "loss_yns_4": 0.15084, "loss_cls_5": 0.85564, "loss_box_5": 1.58052, "loss_cns_5": 0.66356, "loss_yns_5": 0.15296, "loss_cls_dn_0": 0.19059, "loss_box_dn_0": 0.73626, "loss_cls_dn_1": 0.1177, "loss_box_dn_1": 0.66659, "loss_cls_dn_2": 0.11605, "loss_box_dn_2": 0.65659, "loss_cls_dn_3": 0.11605, "loss_box_dn_3": 0.66731, "loss_cls_dn_4": 0.1193, "loss_box_dn_4": 0.66203, "loss_cls_dn_5": 0.12044, "loss_box_dn_5": 0.67293, "loss_dense_depth": 0.71625, "loss": 24.96441, "grad_norm": 40.97886, "time": 1.58884} -{"mode": "train", "epoch": 1, "iter": 253, "lr": 0.0002, "memory": 49164, "data_time": 0.07402, "loss_cls_0": 0.77049, "loss_box_0": 1.67861, "loss_cns_0": 0.62811, "loss_yns_0": 0.14658, "loss_cls_1": 0.83795, "loss_box_1": 1.55066, "loss_cns_1": 0.65975, "loss_yns_1": 0.14776, "loss_cls_2": 0.84057, "loss_box_2": 1.53298, "loss_cns_2": 0.65955, "loss_yns_2": 0.14773, "loss_cls_3": 0.86499, "loss_box_3": 1.53196, "loss_cns_3": 0.66489, "loss_yns_3": 0.14781, "loss_cls_4": 0.87468, "loss_box_4": 1.53495, "loss_cns_4": 0.66119, "loss_yns_4": 0.14801, "loss_cls_5": 0.8817, "loss_box_5": 1.5418, "loss_cns_5": 0.66182, "loss_yns_5": 0.14835, "loss_cls_dn_0": 0.19775, "loss_box_dn_0": 0.74426, "loss_cls_dn_1": 0.11668, "loss_box_dn_1": 0.6641, "loss_cls_dn_2": 0.11728, "loss_box_dn_2": 0.66024, "loss_cls_dn_3": 0.12124, "loss_box_dn_3": 0.66356, "loss_cls_dn_4": 0.12337, "loss_box_dn_4": 0.67481, "loss_cls_dn_5": 0.12208, "loss_box_dn_5": 0.68457, "loss_dense_depth": 0.72714, "loss": 24.87999, "grad_norm": 40.76501, "time": 1.58455} -{"mode": "train", "epoch": 1, "iter": 254, "lr": 0.0002, "memory": 49164, "data_time": 0.07392, "loss_cls_0": 0.77395, "loss_box_0": 1.68518, "loss_cns_0": 0.6241, "loss_yns_0": 0.14844, "loss_cls_1": 0.86888, "loss_box_1": 1.5646, "loss_cns_1": 0.66174, "loss_yns_1": 0.14691, "loss_cls_2": 0.8667, "loss_box_2": 1.54736, "loss_cns_2": 0.66235, "loss_yns_2": 0.14787, "loss_cls_3": 0.87249, "loss_box_3": 1.54677, "loss_cns_3": 0.66378, "loss_yns_3": 0.14633, "loss_cls_4": 0.89798, "loss_box_4": 1.56619, "loss_cns_4": 0.66302, "loss_yns_4": 0.14812, "loss_cls_5": 0.88608, "loss_box_5": 1.56819, "loss_cns_5": 0.66276, "loss_yns_5": 0.14811, "loss_cls_dn_0": 0.1938, "loss_box_dn_0": 0.742, "loss_cls_dn_1": 0.1201, "loss_box_dn_1": 0.68149, "loss_cls_dn_2": 0.12299, "loss_box_dn_2": 0.67781, "loss_cls_dn_3": 0.12745, "loss_box_dn_3": 0.6858, "loss_cls_dn_4": 0.12762, "loss_box_dn_4": 0.70143, "loss_cls_dn_5": 0.12705, "loss_box_dn_5": 0.70935, "loss_dense_depth": 0.71811, "loss": 25.20289, "grad_norm": 46.45803, "time": 1.57045} -{"mode": "train", "epoch": 1, "iter": 255, "lr": 0.0002, "memory": 49164, "data_time": 0.07749, "loss_cls_0": 0.75225, "loss_box_0": 1.65021, "loss_cns_0": 0.6288, "loss_yns_0": 0.14966, "loss_cls_1": 0.86363, "loss_box_1": 1.56059, "loss_cns_1": 0.66187, "loss_yns_1": 0.15039, "loss_cls_2": 0.87333, "loss_box_2": 1.51985, "loss_cns_2": 0.66436, "loss_yns_2": 0.14971, "loss_cls_3": 0.8571, "loss_box_3": 1.53619, "loss_cns_3": 0.66506, "loss_yns_3": 0.15019, "loss_cls_4": 0.87521, "loss_box_4": 1.54252, "loss_cns_4": 0.66559, "loss_yns_4": 0.15061, "loss_cls_5": 0.86592, "loss_box_5": 1.54915, "loss_cns_5": 0.66381, "loss_yns_5": 0.1521, "loss_cls_dn_0": 0.18547, "loss_box_dn_0": 0.73818, "loss_cls_dn_1": 0.11933, "loss_box_dn_1": 0.70047, "loss_cls_dn_2": 0.11927, "loss_box_dn_2": 0.68553, "loss_cls_dn_3": 0.12066, "loss_box_dn_3": 0.6973, "loss_cls_dn_4": 0.12133, "loss_box_dn_4": 0.70513, "loss_cls_dn_5": 0.12291, "loss_box_dn_5": 0.71459, "loss_dense_depth": 0.7203, "loss": 25.04859, "grad_norm": 45.12188, "time": 1.58672} -{"mode": "train", "epoch": 1, "iter": 256, "lr": 0.0002, "memory": 49164, "data_time": 0.08002, "loss_cls_0": 0.76274, "loss_box_0": 1.69992, "loss_cns_0": 0.62773, "loss_yns_0": 0.14998, "loss_cls_1": 0.87617, "loss_box_1": 1.53654, "loss_cns_1": 0.66408, "loss_yns_1": 0.14798, "loss_cls_2": 0.89277, "loss_box_2": 1.50615, "loss_cns_2": 0.66587, "loss_yns_2": 0.14858, "loss_cls_3": 0.87119, "loss_box_3": 1.48637, "loss_cns_3": 0.66568, "loss_yns_3": 0.14709, "loss_cls_4": 0.88333, "loss_box_4": 1.48611, "loss_cns_4": 0.66986, "loss_yns_4": 0.14793, "loss_cls_5": 0.87205, "loss_box_5": 1.48573, "loss_cns_5": 0.66939, "loss_yns_5": 0.15029, "loss_cls_dn_0": 0.18244, "loss_box_dn_0": 0.73786, "loss_cls_dn_1": 0.11169, "loss_box_dn_1": 0.67204, "loss_cls_dn_2": 0.11383, "loss_box_dn_2": 0.65604, "loss_cls_dn_3": 0.11361, "loss_box_dn_3": 0.64899, "loss_cls_dn_4": 0.11275, "loss_box_dn_4": 0.6495, "loss_cls_dn_5": 0.11628, "loss_box_dn_5": 0.65625, "loss_dense_depth": 0.72613, "loss": 24.71095, "grad_norm": 32.63818, "time": 1.58272} -{"mode": "train", "epoch": 1, "iter": 257, "lr": 0.0002, "memory": 49164, "data_time": 0.07285, "loss_cls_0": 0.79278, "loss_box_0": 1.72546, "loss_cns_0": 0.62802, "loss_yns_0": 0.14962, "loss_cls_1": 0.86625, "loss_box_1": 1.58687, "loss_cns_1": 0.65927, "loss_yns_1": 0.14745, "loss_cls_2": 0.88341, "loss_box_2": 1.54801, "loss_cns_2": 0.65981, "loss_yns_2": 0.14708, "loss_cls_3": 0.87584, "loss_box_3": 1.53759, "loss_cns_3": 0.66294, "loss_yns_3": 0.14703, "loss_cls_4": 0.88873, "loss_box_4": 1.56035, "loss_cns_4": 0.66155, "loss_yns_4": 0.14776, "loss_cls_5": 0.87996, "loss_box_5": 1.5527, "loss_cns_5": 0.65951, "loss_yns_5": 0.14705, "loss_cls_dn_0": 0.18883, "loss_box_dn_0": 0.73423, "loss_cls_dn_1": 0.11119, "loss_box_dn_1": 0.6635, "loss_cls_dn_2": 0.11395, "loss_box_dn_2": 0.64551, "loss_cls_dn_3": 0.1211, "loss_box_dn_3": 0.63976, "loss_cls_dn_4": 0.117, "loss_box_dn_4": 0.65196, "loss_cls_dn_5": 0.11889, "loss_box_dn_5": 0.65488, "loss_dense_depth": 0.74307, "loss": 25.01891, "grad_norm": 37.48157, "time": 1.57213} -{"mode": "train", "epoch": 1, "iter": 258, "lr": 0.0002, "memory": 49164, "data_time": 0.07276, "loss_cls_0": 0.78971, "loss_box_0": 1.72758, "loss_cns_0": 0.62273, "loss_yns_0": 0.14706, "loss_cls_1": 0.86495, "loss_box_1": 1.56795, "loss_cns_1": 0.65612, "loss_yns_1": 0.14484, "loss_cls_2": 0.88344, "loss_box_2": 1.52329, "loss_cns_2": 0.66099, "loss_yns_2": 0.14484, "loss_cls_3": 0.8768, "loss_box_3": 1.5255, "loss_cns_3": 0.66066, "loss_yns_3": 0.14451, "loss_cls_4": 0.89512, "loss_box_4": 1.53383, "loss_cns_4": 0.66145, "loss_yns_4": 0.1458, "loss_cls_5": 0.89113, "loss_box_5": 1.52117, "loss_cns_5": 0.65925, "loss_yns_5": 0.14364, "loss_cls_dn_0": 0.18537, "loss_box_dn_0": 0.72987, "loss_cls_dn_1": 0.11847, "loss_box_dn_1": 0.65648, "loss_cls_dn_2": 0.12114, "loss_box_dn_2": 0.63463, "loss_cls_dn_3": 0.122, "loss_box_dn_3": 0.63781, "loss_cls_dn_4": 0.11775, "loss_box_dn_4": 0.64654, "loss_cls_dn_5": 0.11931, "loss_box_dn_5": 0.64745, "loss_dense_depth": 0.7195, "loss": 24.8487, "grad_norm": 33.58824, "time": 1.55998} -{"mode": "train", "epoch": 1, "iter": 259, "lr": 0.0002, "memory": 49164, "data_time": 0.0737, "loss_cls_0": 0.79881, "loss_box_0": 1.71355, "loss_cns_0": 0.61848, "loss_yns_0": 0.14807, "loss_cls_1": 0.86631, "loss_box_1": 1.63926, "loss_cns_1": 0.6475, "loss_yns_1": 0.14357, "loss_cls_2": 0.8833, "loss_box_2": 1.6198, "loss_cns_2": 0.65665, "loss_yns_2": 0.14547, "loss_cls_3": 0.89143, "loss_box_3": 1.59053, "loss_cns_3": 0.65207, "loss_yns_3": 0.14507, "loss_cls_4": 0.93553, "loss_box_4": 1.56422, "loss_cns_4": 0.65442, "loss_yns_4": 0.14503, "loss_cls_5": 0.90265, "loss_box_5": 1.59175, "loss_cns_5": 0.65246, "loss_yns_5": 0.14509, "loss_cls_dn_0": 0.19411, "loss_box_dn_0": 0.74126, "loss_cls_dn_1": 0.12013, "loss_box_dn_1": 0.66191, "loss_cls_dn_2": 0.12302, "loss_box_dn_2": 0.65458, "loss_cls_dn_3": 0.12404, "loss_box_dn_3": 0.65525, "loss_cls_dn_4": 0.12302, "loss_box_dn_4": 0.65514, "loss_cls_dn_5": 0.12633, "loss_box_dn_5": 0.66432, "loss_dense_depth": 0.76188, "loss": 25.35603, "grad_norm": 34.81636, "time": 1.56127} -{"mode": "train", "epoch": 1, "iter": 260, "lr": 0.0002, "memory": 49164, "data_time": 0.07432, "loss_cls_0": 0.78498, "loss_box_0": 1.68414, "loss_cns_0": 0.62363, "loss_yns_0": 0.1525, "loss_cls_1": 0.85916, "loss_box_1": 1.6565, "loss_cns_1": 0.64904, "loss_yns_1": 0.15148, "loss_cls_2": 0.88849, "loss_box_2": 1.63981, "loss_cns_2": 0.65653, "loss_yns_2": 0.15182, "loss_cls_3": 0.87909, "loss_box_3": 1.6112, "loss_cns_3": 0.65491, "loss_yns_3": 0.15111, "loss_cls_4": 0.89877, "loss_box_4": 1.58676, "loss_cns_4": 0.65683, "loss_yns_4": 0.15086, "loss_cls_5": 0.92662, "loss_box_5": 1.61552, "loss_cns_5": 0.6557, "loss_yns_5": 0.15074, "loss_cls_dn_0": 0.18881, "loss_box_dn_0": 0.74042, "loss_cls_dn_1": 0.12191, "loss_box_dn_1": 0.67879, "loss_cls_dn_2": 0.12635, "loss_box_dn_2": 0.67313, "loss_cls_dn_3": 0.12969, "loss_box_dn_3": 0.67029, "loss_cls_dn_4": 0.12658, "loss_box_dn_4": 0.67074, "loss_cls_dn_5": 0.14045, "loss_box_dn_5": 0.68123, "loss_dense_depth": 0.73688, "loss": 25.52143, "grad_norm": 45.11368, "time": 11.41018} -{"mode": "train", "epoch": 1, "iter": 261, "lr": 0.0002, "memory": 49164, "data_time": 0.0931, "loss_cls_0": 0.79712, "loss_box_0": 1.68434, "loss_cns_0": 0.62206, "loss_yns_0": 0.14757, "loss_cls_1": 0.87987, "loss_box_1": 1.59791, "loss_cns_1": 0.65891, "loss_yns_1": 0.14666, "loss_cls_2": 0.88761, "loss_box_2": 1.57144, "loss_cns_2": 0.66111, "loss_yns_2": 0.147, "loss_cls_3": 0.89155, "loss_box_3": 1.55725, "loss_cns_3": 0.65979, "loss_yns_3": 0.14662, "loss_cls_4": 0.91109, "loss_box_4": 1.56499, "loss_cns_4": 0.66042, "loss_yns_4": 0.14735, "loss_cls_5": 0.90603, "loss_box_5": 1.56326, "loss_cns_5": 0.65994, "loss_yns_5": 0.14674, "loss_cls_dn_0": 0.18757, "loss_box_dn_0": 0.73733, "loss_cls_dn_1": 0.12067, "loss_box_dn_1": 0.68538, "loss_cls_dn_2": 0.12188, "loss_box_dn_2": 0.67632, "loss_cls_dn_3": 0.12536, "loss_box_dn_3": 0.67567, "loss_cls_dn_4": 0.12469, "loss_box_dn_4": 0.68118, "loss_cls_dn_5": 0.12498, "loss_box_dn_5": 0.68496, "loss_dense_depth": 0.72268, "loss": 25.28528, "grad_norm": 29.79366, "time": 1.62408} -{"mode": "train", "epoch": 1, "iter": 262, "lr": 0.0002, "memory": 49164, "data_time": 0.10649, "loss_cls_0": 0.79306, "loss_box_0": 1.6864, "loss_cns_0": 0.62521, "loss_yns_0": 0.14557, "loss_cls_1": 0.87008, "loss_box_1": 1.58013, "loss_cns_1": 0.65919, "loss_yns_1": 0.14646, "loss_cls_2": 0.8888, "loss_box_2": 1.54944, "loss_cns_2": 0.6583, "loss_yns_2": 0.14749, "loss_cls_3": 0.88859, "loss_box_3": 1.54838, "loss_cns_3": 0.66218, "loss_yns_3": 0.14342, "loss_cls_4": 0.89936, "loss_box_4": 1.54753, "loss_cns_4": 0.65982, "loss_yns_4": 0.1466, "loss_cls_5": 0.90613, "loss_box_5": 1.52634, "loss_cns_5": 0.65778, "loss_yns_5": 0.14489, "loss_cls_dn_0": 0.18529, "loss_box_dn_0": 0.73991, "loss_cls_dn_1": 0.11425, "loss_box_dn_1": 0.6773, "loss_cls_dn_2": 0.11851, "loss_box_dn_2": 0.66196, "loss_cls_dn_3": 0.12285, "loss_box_dn_3": 0.66476, "loss_cls_dn_4": 0.11882, "loss_box_dn_4": 0.66492, "loss_cls_dn_5": 0.12336, "loss_box_dn_5": 0.66327, "loss_dense_depth": 0.79664, "loss": 25.13299, "grad_norm": 31.74053, "time": 1.59373} -{"mode": "train", "epoch": 1, "iter": 263, "lr": 0.0002, "memory": 49164, "data_time": 0.08227, "loss_cls_0": 0.77011, "loss_box_0": 1.71583, "loss_cns_0": 0.62174, "loss_yns_0": 0.14715, "loss_cls_1": 0.85954, "loss_box_1": 1.6121, "loss_cns_1": 0.65576, "loss_yns_1": 0.14397, "loss_cls_2": 0.89103, "loss_box_2": 1.56014, "loss_cns_2": 0.65426, "loss_yns_2": 0.14332, "loss_cls_3": 0.88664, "loss_box_3": 1.573, "loss_cns_3": 0.66156, "loss_yns_3": 0.14337, "loss_cls_4": 0.89618, "loss_box_4": 1.55927, "loss_cns_4": 0.65879, "loss_yns_4": 0.14486, "loss_cls_5": 0.89606, "loss_box_5": 1.58508, "loss_cns_5": 0.65715, "loss_yns_5": 0.14377, "loss_cls_dn_0": 0.18639, "loss_box_dn_0": 0.7619, "loss_cls_dn_1": 0.1127, "loss_box_dn_1": 0.67881, "loss_cls_dn_2": 0.11689, "loss_box_dn_2": 0.65807, "loss_cls_dn_3": 0.11679, "loss_box_dn_3": 0.66266, "loss_cls_dn_4": 0.11754, "loss_box_dn_4": 0.65512, "loss_cls_dn_5": 0.12064, "loss_box_dn_5": 0.66951, "loss_dense_depth": 0.80619, "loss": 25.2439, "grad_norm": 37.90257, "time": 1.59704} -{"mode": "train", "epoch": 1, "iter": 264, "lr": 0.00021, "memory": 49164, "data_time": 0.07459, "loss_cls_0": 0.76443, "loss_box_0": 1.70601, "loss_cns_0": 0.62599, "loss_yns_0": 0.14678, "loss_cls_1": 0.86527, "loss_box_1": 1.60418, "loss_cns_1": 0.66168, "loss_yns_1": 0.14312, "loss_cls_2": 0.88393, "loss_box_2": 1.56416, "loss_cns_2": 0.65955, "loss_yns_2": 0.14251, "loss_cls_3": 0.90442, "loss_box_3": 1.56168, "loss_cns_3": 0.66156, "loss_yns_3": 0.14317, "loss_cls_4": 0.92229, "loss_box_4": 1.54602, "loss_cns_4": 0.66063, "loss_yns_4": 0.14151, "loss_cls_5": 0.89983, "loss_box_5": 1.552, "loss_cns_5": 0.6606, "loss_yns_5": 0.14202, "loss_cls_dn_0": 0.18218, "loss_box_dn_0": 0.74109, "loss_cls_dn_1": 0.11485, "loss_box_dn_1": 0.66764, "loss_cls_dn_2": 0.12052, "loss_box_dn_2": 0.64815, "loss_cls_dn_3": 0.11916, "loss_box_dn_3": 0.65121, "loss_cls_dn_4": 0.12707, "loss_box_dn_4": 0.65047, "loss_cls_dn_5": 0.12168, "loss_box_dn_5": 0.66443, "loss_dense_depth": 0.77597, "loss": 25.14776, "grad_norm": 28.57029, "time": 1.58191} -{"mode": "train", "epoch": 1, "iter": 265, "lr": 0.00021, "memory": 49164, "data_time": 0.08712, "loss_cls_0": 0.79617, "loss_box_0": 1.69464, "loss_cns_0": 0.62519, "loss_yns_0": 0.14263, "loss_cls_1": 0.88732, "loss_box_1": 1.58073, "loss_cns_1": 0.66001, "loss_yns_1": 0.14277, "loss_cls_2": 0.8942, "loss_box_2": 1.56987, "loss_cns_2": 0.65878, "loss_yns_2": 0.14311, "loss_cls_3": 0.90453, "loss_box_3": 1.55632, "loss_cns_3": 0.66415, "loss_yns_3": 0.14171, "loss_cls_4": 0.91997, "loss_box_4": 1.57255, "loss_cns_4": 0.65945, "loss_yns_4": 0.14462, "loss_cls_5": 0.90905, "loss_box_5": 1.58622, "loss_cns_5": 0.65729, "loss_yns_5": 0.14688, "loss_cls_dn_0": 0.19356, "loss_box_dn_0": 0.75133, "loss_cls_dn_1": 0.11342, "loss_box_dn_1": 0.68047, "loss_cls_dn_2": 0.11564, "loss_box_dn_2": 0.67793, "loss_cls_dn_3": 0.11683, "loss_box_dn_3": 0.67561, "loss_cls_dn_4": 0.11721, "loss_box_dn_4": 0.69233, "loss_cls_dn_5": 0.1248, "loss_box_dn_5": 0.70739, "loss_dense_depth": 0.87705, "loss": 25.50169, "grad_norm": 50.59924, "time": 1.57568} -{"mode": "train", "epoch": 1, "iter": 266, "lr": 0.00021, "memory": 49164, "data_time": 0.11911, "loss_cls_0": 0.78512, "loss_box_0": 1.65646, "loss_cns_0": 0.62637, "loss_yns_0": 0.13988, "loss_cls_1": 0.89835, "loss_box_1": 1.54903, "loss_cns_1": 0.65971, "loss_yns_1": 0.14347, "loss_cls_2": 0.89342, "loss_box_2": 1.52017, "loss_cns_2": 0.6581, "loss_yns_2": 0.14317, "loss_cls_3": 0.92376, "loss_box_3": 1.5273, "loss_cns_3": 0.67084, "loss_yns_3": 0.14058, "loss_cls_4": 0.93204, "loss_box_4": 1.54411, "loss_cns_4": 0.66295, "loss_yns_4": 0.14357, "loss_cls_5": 0.90164, "loss_box_5": 1.56451, "loss_cns_5": 0.66382, "loss_yns_5": 0.14588, "loss_cls_dn_0": 0.19423, "loss_box_dn_0": 0.75084, "loss_cls_dn_1": 0.12, "loss_box_dn_1": 0.69169, "loss_cls_dn_2": 0.11673, "loss_box_dn_2": 0.68649, "loss_cls_dn_3": 0.12124, "loss_box_dn_3": 0.69175, "loss_cls_dn_4": 0.12474, "loss_box_dn_4": 0.70515, "loss_cls_dn_5": 0.13447, "loss_box_dn_5": 0.72016, "loss_dense_depth": 0.80258, "loss": 25.35433, "grad_norm": 44.02044, "time": 1.60295} -{"mode": "train", "epoch": 1, "iter": 267, "lr": 0.00021, "memory": 49164, "data_time": 0.07909, "loss_cls_0": 0.75468, "loss_box_0": 1.66331, "loss_cns_0": 0.62803, "loss_yns_0": 0.14628, "loss_cls_1": 0.86363, "loss_box_1": 1.55796, "loss_cns_1": 0.66003, "loss_yns_1": 0.14464, "loss_cls_2": 0.87046, "loss_box_2": 1.52382, "loss_cns_2": 0.65883, "loss_yns_2": 0.14501, "loss_cls_3": 0.89864, "loss_box_3": 1.52236, "loss_cns_3": 0.66218, "loss_yns_3": 0.14058, "loss_cls_4": 0.91719, "loss_box_4": 1.52103, "loss_cns_4": 0.66126, "loss_yns_4": 0.14182, "loss_cls_5": 0.88573, "loss_box_5": 1.51219, "loss_cns_5": 0.66255, "loss_yns_5": 0.14194, "loss_cls_dn_0": 0.18971, "loss_box_dn_0": 0.73801, "loss_cls_dn_1": 0.12011, "loss_box_dn_1": 0.70135, "loss_cls_dn_2": 0.11711, "loss_box_dn_2": 0.69076, "loss_cls_dn_3": 0.1204, "loss_box_dn_3": 0.69342, "loss_cls_dn_4": 0.12381, "loss_box_dn_4": 0.6992, "loss_cls_dn_5": 0.12324, "loss_box_dn_5": 0.69938, "loss_dense_depth": 0.83438, "loss": 25.13502, "grad_norm": 45.693, "time": 1.57906} -{"mode": "train", "epoch": 1, "iter": 268, "lr": 0.00021, "memory": 49164, "data_time": 0.08582, "loss_cls_0": 0.7961, "loss_box_0": 1.65401, "loss_cns_0": 0.63836, "loss_yns_0": 0.14748, "loss_cls_1": 0.85293, "loss_box_1": 1.53918, "loss_cns_1": 0.66423, "loss_yns_1": 0.14403, "loss_cls_2": 0.88262, "loss_box_2": 1.48985, "loss_cns_2": 0.66521, "loss_yns_2": 0.14201, "loss_cls_3": 0.86783, "loss_box_3": 1.47511, "loss_cns_3": 0.666, "loss_yns_3": 0.14127, "loss_cls_4": 0.8865, "loss_box_4": 1.47375, "loss_cns_4": 0.66509, "loss_yns_4": 0.14232, "loss_cls_5": 0.88822, "loss_box_5": 1.47546, "loss_cns_5": 0.66641, "loss_yns_5": 0.14306, "loss_cls_dn_0": 0.18187, "loss_box_dn_0": 0.73754, "loss_cls_dn_1": 0.11709, "loss_box_dn_1": 0.69332, "loss_cls_dn_2": 0.12006, "loss_box_dn_2": 0.67675, "loss_cls_dn_3": 0.12128, "loss_box_dn_3": 0.67209, "loss_cls_dn_4": 0.11944, "loss_box_dn_4": 0.67469, "loss_cls_dn_5": 0.13171, "loss_box_dn_5": 0.67803, "loss_dense_depth": 0.76878, "loss": 24.79968, "grad_norm": 34.32131, "time": 1.58768} -{"mode": "train", "epoch": 1, "iter": 269, "lr": 0.00021, "memory": 49164, "data_time": 0.08213, "loss_cls_0": 0.75032, "loss_box_0": 1.63977, "loss_cns_0": 0.63184, "loss_yns_0": 0.14454, "loss_cls_1": 0.85028, "loss_box_1": 1.54095, "loss_cns_1": 0.66074, "loss_yns_1": 0.14249, "loss_cls_2": 0.84713, "loss_box_2": 1.5191, "loss_cns_2": 0.66097, "loss_yns_2": 0.14322, "loss_cls_3": 0.86747, "loss_box_3": 1.53786, "loss_cns_3": 0.66417, "loss_yns_3": 0.14396, "loss_cls_4": 0.87695, "loss_box_4": 1.53378, "loss_cns_4": 0.66245, "loss_yns_4": 0.1474, "loss_cls_5": 0.86944, "loss_box_5": 1.53204, "loss_cns_5": 0.66735, "loss_yns_5": 0.14519, "loss_cls_dn_0": 0.17326, "loss_box_dn_0": 0.74085, "loss_cls_dn_1": 0.11183, "loss_box_dn_1": 0.69168, "loss_cls_dn_2": 0.11061, "loss_box_dn_2": 0.67846, "loss_cls_dn_3": 0.11317, "loss_box_dn_3": 0.68016, "loss_cls_dn_4": 0.11524, "loss_box_dn_4": 0.68491, "loss_cls_dn_5": 0.12249, "loss_box_dn_5": 0.68964, "loss_dense_depth": 0.80432, "loss": 24.89603, "grad_norm": 54.88711, "time": 1.57155} -{"mode": "train", "epoch": 1, "iter": 270, "lr": 0.00021, "memory": 49164, "data_time": 0.08392, "loss_cls_0": 0.80054, "loss_box_0": 1.6652, "loss_cns_0": 0.62345, "loss_yns_0": 0.14505, "loss_cls_1": 0.90611, "loss_box_1": 1.57783, "loss_cns_1": 0.64821, "loss_yns_1": 0.14341, "loss_cls_2": 0.90544, "loss_box_2": 1.53378, "loss_cns_2": 0.64713, "loss_yns_2": 0.14482, "loss_cls_3": 0.90574, "loss_box_3": 1.56141, "loss_cns_3": 0.65808, "loss_yns_3": 0.14594, "loss_cls_4": 0.89176, "loss_box_4": 1.56479, "loss_cns_4": 0.65771, "loss_yns_4": 0.14935, "loss_cls_5": 0.9145, "loss_box_5": 1.54787, "loss_cns_5": 0.66046, "loss_yns_5": 0.14777, "loss_cls_dn_0": 0.19019, "loss_box_dn_0": 0.74107, "loss_cls_dn_1": 0.11169, "loss_box_dn_1": 0.69889, "loss_cls_dn_2": 0.11233, "loss_box_dn_2": 0.681, "loss_cls_dn_3": 0.11569, "loss_box_dn_3": 0.68663, "loss_cls_dn_4": 0.11688, "loss_box_dn_4": 0.6906, "loss_cls_dn_5": 0.11944, "loss_box_dn_5": 0.69748, "loss_dense_depth": 0.82013, "loss": 25.32839, "grad_norm": 52.70241, "time": 1.56503} -{"mode": "train", "epoch": 1, "iter": 271, "lr": 0.00021, "memory": 49164, "data_time": 0.07485, "loss_cls_0": 0.73908, "loss_box_0": 1.65463, "loss_cns_0": 0.63427, "loss_yns_0": 0.14796, "loss_cls_1": 0.85406, "loss_box_1": 1.57245, "loss_cns_1": 0.64999, "loss_yns_1": 0.14623, "loss_cls_2": 0.85443, "loss_box_2": 1.5392, "loss_cns_2": 0.65455, "loss_yns_2": 0.1456, "loss_cls_3": 0.87127, "loss_box_3": 1.53455, "loss_cns_3": 0.66156, "loss_yns_3": 0.1451, "loss_cls_4": 0.86886, "loss_box_4": 1.54341, "loss_cns_4": 0.66119, "loss_yns_4": 0.14569, "loss_cls_5": 0.87599, "loss_box_5": 1.55975, "loss_cns_5": 0.6603, "loss_yns_5": 0.14699, "loss_cls_dn_0": 0.18711, "loss_box_dn_0": 0.74971, "loss_cls_dn_1": 0.11044, "loss_box_dn_1": 0.69156, "loss_cls_dn_2": 0.10938, "loss_box_dn_2": 0.67898, "loss_cls_dn_3": 0.11271, "loss_box_dn_3": 0.67954, "loss_cls_dn_4": 0.11433, "loss_box_dn_4": 0.68549, "loss_cls_dn_5": 0.11409, "loss_box_dn_5": 0.70438, "loss_dense_depth": 0.7438, "loss": 24.94863, "grad_norm": 39.93568, "time": 1.56622} -{"mode": "train", "epoch": 1, "iter": 272, "lr": 0.00021, "memory": 49164, "data_time": 0.0728, "loss_cls_0": 0.74378, "loss_box_0": 1.65936, "loss_cns_0": 0.63126, "loss_yns_0": 0.14792, "loss_cls_1": 0.84161, "loss_box_1": 1.55617, "loss_cns_1": 0.65928, "loss_yns_1": 0.14863, "loss_cls_2": 0.84701, "loss_box_2": 1.55783, "loss_cns_2": 0.66587, "loss_yns_2": 0.14795, "loss_cls_3": 0.91516, "loss_box_3": 1.52784, "loss_cns_3": 0.66558, "loss_yns_3": 0.14873, "loss_cls_4": 0.91835, "loss_box_4": 1.53468, "loss_cns_4": 0.66364, "loss_yns_4": 0.14921, "loss_cls_5": 0.86136, "loss_box_5": 1.53266, "loss_cns_5": 0.66307, "loss_yns_5": 0.14983, "loss_cls_dn_0": 0.18907, "loss_box_dn_0": 0.739, "loss_cls_dn_1": 0.10685, "loss_box_dn_1": 0.68973, "loss_cls_dn_2": 0.10798, "loss_box_dn_2": 0.69437, "loss_cls_dn_3": 0.11284, "loss_box_dn_3": 0.68938, "loss_cls_dn_4": 0.11676, "loss_box_dn_4": 0.70088, "loss_cls_dn_5": 0.11444, "loss_box_dn_5": 0.71266, "loss_dense_depth": 0.81415, "loss": 25.12489, "grad_norm": 46.12395, "time": 1.5585} -{"mode": "train", "epoch": 1, "iter": 273, "lr": 0.00021, "memory": 49164, "data_time": 0.0789, "loss_cls_0": 0.75187, "loss_box_0": 1.64323, "loss_cns_0": 0.62704, "loss_yns_0": 0.14704, "loss_cls_1": 0.83072, "loss_box_1": 1.53985, "loss_cns_1": 0.6588, "loss_yns_1": 0.14444, "loss_cls_2": 0.843, "loss_box_2": 1.5437, "loss_cns_2": 0.66232, "loss_yns_2": 0.14401, "loss_cls_3": 0.85613, "loss_box_3": 1.51535, "loss_cns_3": 0.66292, "loss_yns_3": 0.14499, "loss_cls_4": 0.86385, "loss_box_4": 1.52619, "loss_cns_4": 0.66156, "loss_yns_4": 0.14741, "loss_cls_5": 0.8542, "loss_box_5": 1.5269, "loss_cns_5": 0.66122, "loss_yns_5": 0.14451, "loss_cls_dn_0": 0.18281, "loss_box_dn_0": 0.73896, "loss_cls_dn_1": 0.10923, "loss_box_dn_1": 0.68934, "loss_cls_dn_2": 0.10896, "loss_box_dn_2": 0.69628, "loss_cls_dn_3": 0.11158, "loss_box_dn_3": 0.68588, "loss_cls_dn_4": 0.11438, "loss_box_dn_4": 0.69816, "loss_cls_dn_5": 0.1141, "loss_box_dn_5": 0.70436, "loss_dense_depth": 0.72576, "loss": 24.78104, "grad_norm": 47.54048, "time": 1.56643} -{"mode": "train", "epoch": 1, "iter": 274, "lr": 0.00021, "memory": 49164, "data_time": 0.07737, "loss_cls_0": 0.75576, "loss_box_0": 1.65691, "loss_cns_0": 0.62764, "loss_yns_0": 0.14527, "loss_cls_1": 0.8159, "loss_box_1": 1.52472, "loss_cns_1": 0.65914, "loss_yns_1": 0.14227, "loss_cls_2": 0.83581, "loss_box_2": 1.49121, "loss_cns_2": 0.66231, "loss_yns_2": 0.14168, "loss_cls_3": 0.85841, "loss_box_3": 1.46082, "loss_cns_3": 0.66094, "loss_yns_3": 0.1423, "loss_cls_4": 0.86593, "loss_box_4": 1.45853, "loss_cns_4": 0.66166, "loss_yns_4": 0.14606, "loss_cls_5": 0.85285, "loss_box_5": 1.46632, "loss_cns_5": 0.67044, "loss_yns_5": 0.14389, "loss_cls_dn_0": 0.18209, "loss_box_dn_0": 0.74065, "loss_cls_dn_1": 0.10757, "loss_box_dn_1": 0.66481, "loss_cls_dn_2": 0.10582, "loss_box_dn_2": 0.65731, "loss_cls_dn_3": 0.11006, "loss_box_dn_3": 0.64605, "loss_cls_dn_4": 0.11163, "loss_box_dn_4": 0.6472, "loss_cls_dn_5": 0.11199, "loss_box_dn_5": 0.65119, "loss_dense_depth": 0.78389, "loss": 24.36701, "grad_norm": 29.58696, "time": 1.63486} -{"mode": "train", "epoch": 1, "iter": 275, "lr": 0.00021, "memory": 49164, "data_time": 0.07645, "loss_cls_0": 0.73153, "loss_box_0": 1.64791, "loss_cns_0": 0.63484, "loss_yns_0": 0.14265, "loss_cls_1": 0.80166, "loss_box_1": 1.53044, "loss_cns_1": 0.6556, "loss_yns_1": 0.13869, "loss_cls_2": 0.8151, "loss_box_2": 1.48535, "loss_cns_2": 0.65981, "loss_yns_2": 0.13879, "loss_cls_3": 0.86691, "loss_box_3": 1.47617, "loss_cns_3": 0.65676, "loss_yns_3": 0.13905, "loss_cls_4": 0.87221, "loss_box_4": 1.46093, "loss_cns_4": 0.65861, "loss_yns_4": 0.14012, "loss_cls_5": 0.83925, "loss_box_5": 1.46325, "loss_cns_5": 0.66112, "loss_yns_5": 0.13911, "loss_cls_dn_0": 0.17489, "loss_box_dn_0": 0.74468, "loss_cls_dn_1": 0.10318, "loss_box_dn_1": 0.651, "loss_cls_dn_2": 0.10283, "loss_box_dn_2": 0.63921, "loss_cls_dn_3": 0.10926, "loss_box_dn_3": 0.6447, "loss_cls_dn_4": 0.10939, "loss_box_dn_4": 0.64105, "loss_cls_dn_5": 0.11213, "loss_box_dn_5": 0.64354, "loss_dense_depth": 0.70962, "loss": 24.14133, "grad_norm": 54.77482, "time": 1.57276} -{"mode": "train", "epoch": 1, "iter": 276, "lr": 0.00021, "memory": 49164, "data_time": 0.07976, "loss_cls_0": 0.75208, "loss_box_0": 1.6369, "loss_cns_0": 0.63389, "loss_yns_0": 0.14266, "loss_cls_1": 0.81881, "loss_box_1": 1.51282, "loss_cns_1": 0.65922, "loss_yns_1": 0.1388, "loss_cls_2": 0.83368, "loss_box_2": 1.48296, "loss_cns_2": 0.66408, "loss_yns_2": 0.1395, "loss_cls_3": 0.86988, "loss_box_3": 1.46309, "loss_cns_3": 0.65723, "loss_yns_3": 0.1385, "loss_cls_4": 0.86055, "loss_box_4": 1.46562, "loss_cns_4": 0.65892, "loss_yns_4": 0.13969, "loss_cls_5": 0.85634, "loss_box_5": 1.48419, "loss_cns_5": 0.65914, "loss_yns_5": 0.13926, "loss_cls_dn_0": 0.17632, "loss_box_dn_0": 0.74504, "loss_cls_dn_1": 0.10617, "loss_box_dn_1": 0.65238, "loss_cls_dn_2": 0.1065, "loss_box_dn_2": 0.64739, "loss_cls_dn_3": 0.1135, "loss_box_dn_3": 0.6514, "loss_cls_dn_4": 0.11449, "loss_box_dn_4": 0.65272, "loss_cls_dn_5": 0.11694, "loss_box_dn_5": 0.66382, "loss_dense_depth": 0.77638, "loss": 24.33086, "grad_norm": 45.85955, "time": 1.57911} -{"mode": "train", "epoch": 1, "iter": 277, "lr": 0.00021, "memory": 49164, "data_time": 0.07577, "loss_cls_0": 0.772, "loss_box_0": 1.66306, "loss_cns_0": 0.63248, "loss_yns_0": 0.14515, "loss_cls_1": 0.82988, "loss_box_1": 1.56104, "loss_cns_1": 0.65491, "loss_yns_1": 0.14417, "loss_cls_2": 0.85516, "loss_box_2": 1.55192, "loss_cns_2": 0.66208, "loss_yns_2": 0.13906, "loss_cls_3": 0.84463, "loss_box_3": 1.53586, "loss_cns_3": 0.65585, "loss_yns_3": 0.13976, "loss_cls_4": 0.85877, "loss_box_4": 1.54872, "loss_cns_4": 0.66305, "loss_yns_4": 0.1413, "loss_cls_5": 0.88137, "loss_box_5": 1.5442, "loss_cns_5": 0.66068, "loss_yns_5": 0.14354, "loss_cls_dn_0": 0.16972, "loss_box_dn_0": 0.74266, "loss_cls_dn_1": 0.1063, "loss_box_dn_1": 0.67613, "loss_cls_dn_2": 0.10717, "loss_box_dn_2": 0.68058, "loss_cls_dn_3": 0.10927, "loss_box_dn_3": 0.67743, "loss_cls_dn_4": 0.11374, "loss_box_dn_4": 0.68824, "loss_cls_dn_5": 0.11266, "loss_box_dn_5": 0.69817, "loss_dense_depth": 0.74882, "loss": 24.85954, "grad_norm": 52.31081, "time": 1.57932} -{"mode": "train", "epoch": 1, "iter": 278, "lr": 0.00021, "memory": 49164, "data_time": 0.07653, "loss_cls_0": 0.75959, "loss_box_0": 1.64266, "loss_cns_0": 0.63374, "loss_yns_0": 0.14568, "loss_cls_1": 0.82514, "loss_box_1": 1.56342, "loss_cns_1": 0.65554, "loss_yns_1": 0.14315, "loss_cls_2": 0.85681, "loss_box_2": 1.54112, "loss_cns_2": 0.65878, "loss_yns_2": 0.14051, "loss_cls_3": 0.8484, "loss_box_3": 1.5176, "loss_cns_3": 0.65836, "loss_yns_3": 0.14037, "loss_cls_4": 0.8615, "loss_box_4": 1.52453, "loss_cns_4": 0.6621, "loss_yns_4": 0.14307, "loss_cls_5": 0.86334, "loss_box_5": 1.5117, "loss_cns_5": 0.66274, "loss_yns_5": 0.14314, "loss_cls_dn_0": 0.16668, "loss_box_dn_0": 0.73294, "loss_cls_dn_1": 0.10399, "loss_box_dn_1": 0.68537, "loss_cls_dn_2": 0.10278, "loss_box_dn_2": 0.68086, "loss_cls_dn_3": 0.10722, "loss_box_dn_3": 0.67321, "loss_cls_dn_4": 0.11037, "loss_box_dn_4": 0.68352, "loss_cls_dn_5": 0.10688, "loss_box_dn_5": 0.689, "loss_dense_depth": 0.72663, "loss": 24.67243, "grad_norm": 40.98744, "time": 1.58547} -{"mode": "train", "epoch": 1, "iter": 279, "lr": 0.00021, "memory": 49164, "data_time": 0.07575, "loss_cls_0": 0.76077, "loss_box_0": 1.62964, "loss_cns_0": 0.62868, "loss_yns_0": 0.14374, "loss_cls_1": 0.85757, "loss_box_1": 1.49058, "loss_cns_1": 0.65777, "loss_yns_1": 0.14209, "loss_cls_2": 0.8706, "loss_box_2": 1.49973, "loss_cns_2": 0.65961, "loss_yns_2": 0.1423, "loss_cls_3": 0.86081, "loss_box_3": 1.48737, "loss_cns_3": 0.66239, "loss_yns_3": 0.14231, "loss_cls_4": 0.85891, "loss_box_4": 1.49847, "loss_cns_4": 0.66217, "loss_yns_4": 0.14537, "loss_cls_5": 0.85995, "loss_box_5": 1.49336, "loss_cns_5": 0.66065, "loss_yns_5": 0.14291, "loss_cls_dn_0": 0.17481, "loss_box_dn_0": 0.73242, "loss_cls_dn_1": 0.10845, "loss_box_dn_1": 0.6775, "loss_cls_dn_2": 0.10815, "loss_box_dn_2": 0.67835, "loss_cls_dn_3": 0.11281, "loss_box_dn_3": 0.6789, "loss_cls_dn_4": 0.11527, "loss_box_dn_4": 0.68709, "loss_cls_dn_5": 0.11646, "loss_box_dn_5": 0.69083, "loss_dense_depth": 0.74086, "loss": 24.57964, "grad_norm": 49.66616, "time": 1.56695} -{"mode": "train", "epoch": 1, "iter": 280, "lr": 0.00021, "memory": 49164, "data_time": 0.07771, "loss_cls_0": 0.77062, "loss_box_0": 1.63416, "loss_cns_0": 0.62953, "loss_yns_0": 0.14225, "loss_cls_1": 0.85722, "loss_box_1": 1.53326, "loss_cns_1": 0.65918, "loss_yns_1": 0.14279, "loss_cls_2": 0.86807, "loss_box_2": 1.52377, "loss_cns_2": 0.66134, "loss_yns_2": 0.14322, "loss_cls_3": 0.85601, "loss_box_3": 1.5245, "loss_cns_3": 0.66275, "loss_yns_3": 0.14331, "loss_cls_4": 0.86463, "loss_box_4": 1.52059, "loss_cns_4": 0.66291, "loss_yns_4": 0.14155, "loss_cls_5": 0.8688, "loss_box_5": 1.50708, "loss_cns_5": 0.65797, "loss_yns_5": 0.14181, "loss_cls_dn_0": 0.17326, "loss_box_dn_0": 0.73207, "loss_cls_dn_1": 0.11133, "loss_box_dn_1": 0.68326, "loss_cls_dn_2": 0.11063, "loss_box_dn_2": 0.68075, "loss_cls_dn_3": 0.11009, "loss_box_dn_3": 0.68181, "loss_cls_dn_4": 0.11401, "loss_box_dn_4": 0.68461, "loss_cls_dn_5": 0.1153, "loss_box_dn_5": 0.68159, "loss_dense_depth": 0.73964, "loss": 24.73568, "grad_norm": 46.81848, "time": 1.57856} -{"mode": "train", "epoch": 1, "iter": 281, "lr": 0.00021, "memory": 49164, "data_time": 0.11368, "loss_cls_0": 0.7581, "loss_box_0": 1.63835, "loss_cns_0": 0.63301, "loss_yns_0": 0.14178, "loss_cls_1": 0.83507, "loss_box_1": 1.49263, "loss_cns_1": 0.65938, "loss_yns_1": 0.14275, "loss_cls_2": 0.84684, "loss_box_2": 1.47305, "loss_cns_2": 0.66266, "loss_yns_2": 0.14166, "loss_cls_3": 0.85311, "loss_box_3": 1.46557, "loss_cns_3": 0.66042, "loss_yns_3": 0.14074, "loss_cls_4": 0.85593, "loss_box_4": 1.45595, "loss_cns_4": 0.66122, "loss_yns_4": 0.14007, "loss_cls_5": 0.8689, "loss_box_5": 1.46725, "loss_cns_5": 0.66019, "loss_yns_5": 0.14044, "loss_cls_dn_0": 0.17508, "loss_box_dn_0": 0.73775, "loss_cls_dn_1": 0.10834, "loss_box_dn_1": 0.67246, "loss_cls_dn_2": 0.10663, "loss_box_dn_2": 0.6605, "loss_cls_dn_3": 0.10632, "loss_box_dn_3": 0.65861, "loss_cls_dn_4": 0.10761, "loss_box_dn_4": 0.65598, "loss_cls_dn_5": 0.10729, "loss_box_dn_5": 0.66066, "loss_dense_depth": 0.72849, "loss": 24.28079, "grad_norm": 32.25221, "time": 1.68573} -{"mode": "train", "epoch": 1, "iter": 282, "lr": 0.00021, "memory": 49164, "data_time": 0.10556, "loss_cls_0": 0.75559, "loss_box_0": 1.6178, "loss_cns_0": 0.63608, "loss_yns_0": 0.14704, "loss_cls_1": 0.84544, "loss_box_1": 1.47503, "loss_cns_1": 0.66065, "loss_yns_1": 0.1456, "loss_cls_2": 0.86152, "loss_box_2": 1.46237, "loss_cns_2": 0.66283, "loss_yns_2": 0.1455, "loss_cls_3": 0.85692, "loss_box_3": 1.44953, "loss_cns_3": 0.66199, "loss_yns_3": 0.14424, "loss_cls_4": 0.87244, "loss_box_4": 1.44624, "loss_cns_4": 0.66519, "loss_yns_4": 0.14511, "loss_cls_5": 0.87626, "loss_box_5": 1.447, "loss_cns_5": 0.66408, "loss_yns_5": 0.14404, "loss_cls_dn_0": 0.1751, "loss_box_dn_0": 0.74113, "loss_cls_dn_1": 0.10616, "loss_box_dn_1": 0.66256, "loss_cls_dn_2": 0.10472, "loss_box_dn_2": 0.6541, "loss_cls_dn_3": 0.10821, "loss_box_dn_3": 0.65195, "loss_cls_dn_4": 0.11159, "loss_box_dn_4": 0.65339, "loss_cls_dn_5": 0.11005, "loss_box_dn_5": 0.66168, "loss_dense_depth": 0.71873, "loss": 24.24786, "grad_norm": 34.07682, "time": 1.62696} -{"mode": "train", "epoch": 1, "iter": 283, "lr": 0.00021, "memory": 49164, "data_time": 0.08239, "loss_cls_0": 0.76438, "loss_box_0": 1.64409, "loss_cns_0": 0.63165, "loss_yns_0": 0.14559, "loss_cls_1": 0.86399, "loss_box_1": 1.49966, "loss_cns_1": 0.65733, "loss_yns_1": 0.14452, "loss_cls_2": 0.85469, "loss_box_2": 1.46667, "loss_cns_2": 0.65861, "loss_yns_2": 0.14569, "loss_cls_3": 0.86196, "loss_box_3": 1.46471, "loss_cns_3": 0.66067, "loss_yns_3": 0.14474, "loss_cls_4": 0.86657, "loss_box_4": 1.48049, "loss_cns_4": 0.66075, "loss_yns_4": 0.14457, "loss_cls_5": 0.87682, "loss_box_5": 1.46844, "loss_cns_5": 0.66206, "loss_yns_5": 0.14439, "loss_cls_dn_0": 0.18345, "loss_box_dn_0": 0.7379, "loss_cls_dn_1": 0.10802, "loss_box_dn_1": 0.67578, "loss_cls_dn_2": 0.10699, "loss_box_dn_2": 0.66268, "loss_cls_dn_3": 0.1113, "loss_box_dn_3": 0.66762, "loss_cls_dn_4": 0.11482, "loss_box_dn_4": 0.68029, "loss_cls_dn_5": 0.11071, "loss_box_dn_5": 0.68627, "loss_dense_depth": 0.7244, "loss": 24.48329, "grad_norm": 38.64912, "time": 1.60455} -{"mode": "train", "epoch": 1, "iter": 284, "lr": 0.00021, "memory": 49164, "data_time": 0.0772, "loss_cls_0": 0.75202, "loss_box_0": 1.61955, "loss_cns_0": 0.63379, "loss_yns_0": 0.14847, "loss_cls_1": 0.8395, "loss_box_1": 1.52228, "loss_cns_1": 0.6582, "loss_yns_1": 0.14708, "loss_cls_2": 0.85502, "loss_box_2": 1.48779, "loss_cns_2": 0.65911, "loss_yns_2": 0.14719, "loss_cls_3": 0.87577, "loss_box_3": 1.46247, "loss_cns_3": 0.65816, "loss_yns_3": 0.14703, "loss_cls_4": 0.86816, "loss_box_4": 1.461, "loss_cns_4": 0.6637, "loss_yns_4": 0.14636, "loss_cls_5": 0.89096, "loss_box_5": 1.46053, "loss_cns_5": 0.65953, "loss_yns_5": 0.14594, "loss_cls_dn_0": 0.17936, "loss_box_dn_0": 0.72604, "loss_cls_dn_1": 0.10636, "loss_box_dn_1": 0.68897, "loss_cls_dn_2": 0.10627, "loss_box_dn_2": 0.67632, "loss_cls_dn_3": 0.10907, "loss_box_dn_3": 0.6765, "loss_cls_dn_4": 0.10909, "loss_box_dn_4": 0.68358, "loss_cls_dn_5": 0.10953, "loss_box_dn_5": 0.68992, "loss_dense_depth": 0.7032, "loss": 24.47383, "grad_norm": 35.37569, "time": 1.60237} -{"mode": "train", "epoch": 1, "iter": 285, "lr": 0.00021, "memory": 49164, "data_time": 0.08, "loss_cls_0": 0.76204, "loss_box_0": 1.6122, "loss_cns_0": 0.62733, "loss_yns_0": 0.14927, "loss_cls_1": 0.84829, "loss_box_1": 1.48099, "loss_cns_1": 0.65122, "loss_yns_1": 0.14748, "loss_cls_2": 0.86924, "loss_box_2": 1.48592, "loss_cns_2": 0.65025, "loss_yns_2": 0.14603, "loss_cls_3": 0.89808, "loss_box_3": 1.44442, "loss_cns_3": 0.64334, "loss_yns_3": 0.14365, "loss_cls_4": 0.87582, "loss_box_4": 1.46903, "loss_cns_4": 0.65793, "loss_yns_4": 0.14883, "loss_cls_5": 0.89551, "loss_box_5": 1.46498, "loss_cns_5": 0.65328, "loss_yns_5": 0.14686, "loss_cls_dn_0": 0.1762, "loss_box_dn_0": 0.73892, "loss_cls_dn_1": 0.10784, "loss_box_dn_1": 0.67049, "loss_cls_dn_2": 0.1073, "loss_box_dn_2": 0.67051, "loss_cls_dn_3": 0.11083, "loss_box_dn_3": 0.66459, "loss_cls_dn_4": 0.11127, "loss_box_dn_4": 0.66685, "loss_cls_dn_5": 0.11587, "loss_box_dn_5": 0.67319, "loss_dense_depth": 0.71944, "loss": 24.40531, "grad_norm": 44.58969, "time": 1.58209} -{"mode": "train", "epoch": 1, "iter": 286, "lr": 0.00021, "memory": 49164, "data_time": 0.10195, "loss_cls_0": 0.74712, "loss_box_0": 1.59688, "loss_cns_0": 0.62421, "loss_yns_0": 0.14658, "loss_cls_1": 0.81957, "loss_box_1": 1.47999, "loss_cns_1": 0.64712, "loss_yns_1": 0.14725, "loss_cls_2": 0.83497, "loss_box_2": 1.43717, "loss_cns_2": 0.64859, "loss_yns_2": 0.14632, "loss_cls_3": 0.84883, "loss_box_3": 1.42233, "loss_cns_3": 0.64895, "loss_yns_3": 0.14456, "loss_cls_4": 0.84686, "loss_box_4": 1.43165, "loss_cns_4": 0.65297, "loss_yns_4": 0.14565, "loss_cls_5": 0.8629, "loss_box_5": 1.43389, "loss_cns_5": 0.65284, "loss_yns_5": 0.14731, "loss_cls_dn_0": 0.1781, "loss_box_dn_0": 0.73347, "loss_cls_dn_1": 0.10843, "loss_box_dn_1": 0.66735, "loss_cls_dn_2": 0.10943, "loss_box_dn_2": 0.65372, "loss_cls_dn_3": 0.1105, "loss_box_dn_3": 0.65063, "loss_cls_dn_4": 0.11102, "loss_box_dn_4": 0.65061, "loss_cls_dn_5": 0.11407, "loss_box_dn_5": 0.65745, "loss_dense_depth": 0.71555, "loss": 23.97485, "grad_norm": 29.52532, "time": 1.59235} -{"mode": "train", "epoch": 1, "iter": 287, "lr": 0.00021, "memory": 49164, "data_time": 0.07674, "loss_cls_0": 0.77263, "loss_box_0": 1.61041, "loss_cns_0": 0.62743, "loss_yns_0": 0.14964, "loss_cls_1": 0.83927, "loss_box_1": 1.53209, "loss_cns_1": 0.65285, "loss_yns_1": 0.149, "loss_cls_2": 0.85681, "loss_box_2": 1.55237, "loss_cns_2": 0.65495, "loss_yns_2": 0.14971, "loss_cls_3": 0.87792, "loss_box_3": 1.53168, "loss_cns_3": 0.66013, "loss_yns_3": 0.14713, "loss_cls_4": 0.87927, "loss_box_4": 1.50597, "loss_cns_4": 0.66043, "loss_yns_4": 0.14753, "loss_cls_5": 0.88008, "loss_box_5": 1.50386, "loss_cns_5": 0.65853, "loss_yns_5": 0.14889, "loss_cls_dn_0": 0.17967, "loss_box_dn_0": 0.74739, "loss_cls_dn_1": 0.10993, "loss_box_dn_1": 0.65435, "loss_cls_dn_2": 0.11018, "loss_box_dn_2": 0.65122, "loss_cls_dn_3": 0.11091, "loss_box_dn_3": 0.65007, "loss_cls_dn_4": 0.11208, "loss_box_dn_4": 0.64275, "loss_cls_dn_5": 0.11334, "loss_box_dn_5": 0.6472, "loss_dense_depth": 0.74341, "loss": 24.6211, "grad_norm": 42.34392, "time": 1.57209} -{"mode": "train", "epoch": 1, "iter": 288, "lr": 0.00021, "memory": 49164, "data_time": 0.08494, "loss_cls_0": 0.76449, "loss_box_0": 1.60709, "loss_cns_0": 0.63144, "loss_yns_0": 0.14917, "loss_cls_1": 0.84577, "loss_box_1": 1.52473, "loss_cns_1": 0.66216, "loss_yns_1": 0.14923, "loss_cls_2": 0.86271, "loss_box_2": 1.57501, "loss_cns_2": 0.65973, "loss_yns_2": 0.14993, "loss_cls_3": 0.88681, "loss_box_3": 1.54233, "loss_cns_3": 0.66001, "loss_yns_3": 0.1497, "loss_cls_4": 0.88576, "loss_box_4": 1.54297, "loss_cns_4": 0.6624, "loss_yns_4": 0.14894, "loss_cls_5": 0.87383, "loss_box_5": 1.52836, "loss_cns_5": 0.66244, "loss_yns_5": 0.14876, "loss_cls_dn_0": 0.16495, "loss_box_dn_0": 0.74425, "loss_cls_dn_1": 0.10482, "loss_box_dn_1": 0.65367, "loss_cls_dn_2": 0.10368, "loss_box_dn_2": 0.6564, "loss_cls_dn_3": 0.10762, "loss_box_dn_3": 0.65342, "loss_cls_dn_4": 0.10817, "loss_box_dn_4": 0.65761, "loss_cls_dn_5": 0.11147, "loss_box_dn_5": 0.66118, "loss_dense_depth": 0.75028, "loss": 24.7513, "grad_norm": 42.56479, "time": 1.58207} -{"mode": "train", "epoch": 1, "iter": 289, "lr": 0.00022, "memory": 49164, "data_time": 0.07892, "loss_cls_0": 0.7406, "loss_box_0": 1.63155, "loss_cns_0": 0.62567, "loss_yns_0": 0.14762, "loss_cls_1": 0.83904, "loss_box_1": 1.54062, "loss_cns_1": 0.66239, "loss_yns_1": 0.14747, "loss_cls_2": 0.84404, "loss_box_2": 1.50216, "loss_cns_2": 0.6685, "loss_yns_2": 0.14875, "loss_cls_3": 0.85575, "loss_box_3": 1.50353, "loss_cns_3": 0.66366, "loss_yns_3": 0.14789, "loss_cls_4": 0.84804, "loss_box_4": 1.51865, "loss_cns_4": 0.66005, "loss_yns_4": 0.14697, "loss_cls_5": 0.85566, "loss_box_5": 1.50234, "loss_cns_5": 0.662, "loss_yns_5": 0.14745, "loss_cls_dn_0": 0.16249, "loss_box_dn_0": 0.74312, "loss_cls_dn_1": 0.10375, "loss_box_dn_1": 0.66948, "loss_cls_dn_2": 0.10311, "loss_box_dn_2": 0.65916, "loss_cls_dn_3": 0.10614, "loss_box_dn_3": 0.66625, "loss_cls_dn_4": 0.10682, "loss_box_dn_4": 0.68041, "loss_cls_dn_5": 0.11366, "loss_box_dn_5": 0.67426, "loss_dense_depth": 0.68845, "loss": 24.48751, "grad_norm": 43.20884, "time": 1.60715} -{"mode": "train", "epoch": 1, "iter": 290, "lr": 0.00022, "memory": 49164, "data_time": 0.08473, "loss_cls_0": 0.79293, "loss_box_0": 1.63061, "loss_cns_0": 0.61647, "loss_yns_0": 0.14578, "loss_cls_1": 0.85699, "loss_box_1": 1.54517, "loss_cns_1": 0.65572, "loss_yns_1": 0.1473, "loss_cls_2": 0.86649, "loss_box_2": 1.49964, "loss_cns_2": 0.66064, "loss_yns_2": 0.1472, "loss_cls_3": 0.87107, "loss_box_3": 1.50906, "loss_cns_3": 0.66344, "loss_yns_3": 0.14666, "loss_cls_4": 0.87535, "loss_box_4": 1.49875, "loss_cns_4": 0.65628, "loss_yns_4": 0.14624, "loss_cls_5": 0.88266, "loss_box_5": 1.49448, "loss_cns_5": 0.65814, "loss_yns_5": 0.14639, "loss_cls_dn_0": 0.17411, "loss_box_dn_0": 0.73783, "loss_cls_dn_1": 0.10833, "loss_box_dn_1": 0.67726, "loss_cls_dn_2": 0.10796, "loss_box_dn_2": 0.66871, "loss_cls_dn_3": 0.10912, "loss_box_dn_3": 0.67603, "loss_cls_dn_4": 0.10914, "loss_box_dn_4": 0.6845, "loss_cls_dn_5": 0.11512, "loss_box_dn_5": 0.68232, "loss_dense_depth": 0.77483, "loss": 24.7387, "grad_norm": 47.21167, "time": 1.58653} -{"mode": "train", "epoch": 1, "iter": 291, "lr": 0.00022, "memory": 49164, "data_time": 0.08243, "loss_cls_0": 0.80244, "loss_box_0": 1.63231, "loss_cns_0": 0.6255, "loss_yns_0": 0.15071, "loss_cls_1": 0.85991, "loss_box_1": 1.49761, "loss_cns_1": 0.66067, "loss_yns_1": 0.1519, "loss_cls_2": 0.87999, "loss_box_2": 1.49391, "loss_cns_2": 0.65892, "loss_yns_2": 0.15182, "loss_cls_3": 0.87991, "loss_box_3": 1.46834, "loss_cns_3": 0.66291, "loss_yns_3": 0.14962, "loss_cls_4": 0.87583, "loss_box_4": 1.46383, "loss_cns_4": 0.65775, "loss_yns_4": 0.14786, "loss_cls_5": 0.87995, "loss_box_5": 1.46727, "loss_cns_5": 0.66065, "loss_yns_5": 0.14979, "loss_cls_dn_0": 0.16496, "loss_box_dn_0": 0.74107, "loss_cls_dn_1": 0.10428, "loss_box_dn_1": 0.66108, "loss_cls_dn_2": 0.10526, "loss_box_dn_2": 0.65215, "loss_cls_dn_3": 0.10549, "loss_box_dn_3": 0.65208, "loss_cls_dn_4": 0.10658, "loss_box_dn_4": 0.6551, "loss_cls_dn_5": 0.10938, "loss_box_dn_5": 0.65679, "loss_dense_depth": 0.76974, "loss": 24.51337, "grad_norm": 28.99197, "time": 1.60369} -{"mode": "train", "epoch": 1, "iter": 292, "lr": 0.00022, "memory": 49164, "data_time": 0.07675, "loss_cls_0": 0.74424, "loss_box_0": 1.65975, "loss_cns_0": 0.63187, "loss_yns_0": 0.14833, "loss_cls_1": 0.82598, "loss_box_1": 1.52376, "loss_cns_1": 0.66059, "loss_yns_1": 0.14811, "loss_cls_2": 0.83623, "loss_box_2": 1.51889, "loss_cns_2": 0.66297, "loss_yns_2": 0.14898, "loss_cls_3": 0.84253, "loss_box_3": 1.50347, "loss_cns_3": 0.65986, "loss_yns_3": 0.14747, "loss_cls_4": 0.8409, "loss_box_4": 1.48332, "loss_cns_4": 0.6602, "loss_yns_4": 0.14633, "loss_cls_5": 0.84004, "loss_box_5": 1.47506, "loss_cns_5": 0.66091, "loss_yns_5": 0.14684, "loss_cls_dn_0": 0.16174, "loss_box_dn_0": 0.74404, "loss_cls_dn_1": 0.10778, "loss_box_dn_1": 0.65257, "loss_cls_dn_2": 0.10544, "loss_box_dn_2": 0.63935, "loss_cls_dn_3": 0.10762, "loss_box_dn_3": 0.64169, "loss_cls_dn_4": 0.11406, "loss_box_dn_4": 0.63412, "loss_cls_dn_5": 0.11279, "loss_box_dn_5": 0.63304, "loss_dense_depth": 0.75419, "loss": 24.32505, "grad_norm": 37.88528, "time": 1.5797} -{"mode": "train", "epoch": 1, "iter": 293, "lr": 0.00022, "memory": 49164, "data_time": 0.07604, "loss_cls_0": 0.77295, "loss_box_0": 1.6781, "loss_cns_0": 0.63274, "loss_yns_0": 0.14648, "loss_cls_1": 0.84818, "loss_box_1": 1.48534, "loss_cns_1": 0.66297, "loss_yns_1": 0.14331, "loss_cls_2": 0.85157, "loss_box_2": 1.46466, "loss_cns_2": 0.66533, "loss_yns_2": 0.1436, "loss_cls_3": 0.85723, "loss_box_3": 1.46277, "loss_cns_3": 0.6618, "loss_yns_3": 0.14438, "loss_cls_4": 0.85508, "loss_box_4": 1.45993, "loss_cns_4": 0.66227, "loss_yns_4": 0.14393, "loss_cls_5": 0.86187, "loss_box_5": 1.45395, "loss_cns_5": 0.66196, "loss_yns_5": 0.14387, "loss_cls_dn_0": 0.17963, "loss_box_dn_0": 0.73467, "loss_cls_dn_1": 0.10804, "loss_box_dn_1": 0.65197, "loss_cls_dn_2": 0.10581, "loss_box_dn_2": 0.64287, "loss_cls_dn_3": 0.10847, "loss_box_dn_3": 0.64354, "loss_cls_dn_4": 0.11161, "loss_box_dn_4": 0.64297, "loss_cls_dn_5": 0.11296, "loss_box_dn_5": 0.64807, "loss_dense_depth": 0.74024, "loss": 24.29514, "grad_norm": 28.84343, "time": 1.57757} -{"mode": "train", "epoch": 1, "iter": 294, "lr": 0.00022, "memory": 49164, "data_time": 0.07723, "loss_cls_0": 0.76525, "loss_box_0": 1.65668, "loss_cns_0": 0.63417, "loss_yns_0": 0.14766, "loss_cls_1": 0.84356, "loss_box_1": 1.50648, "loss_cns_1": 0.66092, "loss_yns_1": 0.1424, "loss_cls_2": 0.84945, "loss_box_2": 1.48168, "loss_cns_2": 0.66288, "loss_yns_2": 0.14416, "loss_cls_3": 0.8634, "loss_box_3": 1.4568, "loss_cns_3": 0.66188, "loss_yns_3": 0.14363, "loss_cls_4": 0.87433, "loss_box_4": 1.4602, "loss_cns_4": 0.66142, "loss_yns_4": 0.14337, "loss_cls_5": 0.87254, "loss_box_5": 1.4591, "loss_cns_5": 0.66225, "loss_yns_5": 0.14323, "loss_cls_dn_0": 0.16695, "loss_box_dn_0": 0.72826, "loss_cls_dn_1": 0.10639, "loss_box_dn_1": 0.65483, "loss_cls_dn_2": 0.10518, "loss_box_dn_2": 0.64341, "loss_cls_dn_3": 0.10709, "loss_box_dn_3": 0.63854, "loss_cls_dn_4": 0.11261, "loss_box_dn_4": 0.64791, "loss_cls_dn_5": 0.11217, "loss_box_dn_5": 0.65301, "loss_dense_depth": 0.73697, "loss": 24.3108, "grad_norm": 42.8229, "time": 1.62097} -{"mode": "train", "epoch": 1, "iter": 295, "lr": 0.00022, "memory": 49164, "data_time": 0.07229, "loss_cls_0": 0.76341, "loss_box_0": 1.66374, "loss_cns_0": 0.6352, "loss_yns_0": 0.14949, "loss_cls_1": 0.86801, "loss_box_1": 1.51599, "loss_cns_1": 0.65851, "loss_yns_1": 0.14684, "loss_cls_2": 0.83767, "loss_box_2": 1.48844, "loss_cns_2": 0.66047, "loss_yns_2": 0.14933, "loss_cls_3": 0.84715, "loss_box_3": 1.48047, "loss_cns_3": 0.65849, "loss_yns_3": 0.14756, "loss_cls_4": 0.86456, "loss_box_4": 1.48303, "loss_cns_4": 0.65771, "loss_yns_4": 0.1463, "loss_cls_5": 0.86146, "loss_box_5": 1.47654, "loss_cns_5": 0.6585, "loss_yns_5": 0.14735, "loss_cls_dn_0": 0.16168, "loss_box_dn_0": 0.73193, "loss_cls_dn_1": 0.11047, "loss_box_dn_1": 0.67222, "loss_cls_dn_2": 0.10777, "loss_box_dn_2": 0.65384, "loss_cls_dn_3": 0.10757, "loss_box_dn_3": 0.65996, "loss_cls_dn_4": 0.12055, "loss_box_dn_4": 0.67091, "loss_cls_dn_5": 0.1151, "loss_box_dn_5": 0.67902, "loss_dense_depth": 0.73359, "loss": 24.49084, "grad_norm": 37.49094, "time": 1.56564} -{"mode": "train", "epoch": 1, "iter": 296, "lr": 0.00022, "memory": 49164, "data_time": 0.07214, "loss_cls_0": 0.76507, "loss_box_0": 1.6464, "loss_cns_0": 0.63208, "loss_yns_0": 0.14978, "loss_cls_1": 0.84709, "loss_box_1": 1.52034, "loss_cns_1": 0.65599, "loss_yns_1": 0.14839, "loss_cls_2": 0.83414, "loss_box_2": 1.50717, "loss_cns_2": 0.65979, "loss_yns_2": 0.14891, "loss_cls_3": 0.84819, "loss_box_3": 1.49622, "loss_cns_3": 0.66092, "loss_yns_3": 0.14824, "loss_cls_4": 0.85357, "loss_box_4": 1.49317, "loss_cns_4": 0.66027, "loss_yns_4": 0.1478, "loss_cls_5": 0.86293, "loss_box_5": 1.48459, "loss_cns_5": 0.65985, "loss_yns_5": 0.14833, "loss_cls_dn_0": 0.16438, "loss_box_dn_0": 0.72513, "loss_cls_dn_1": 0.1064, "loss_box_dn_1": 0.6915, "loss_cls_dn_2": 0.10716, "loss_box_dn_2": 0.68025, "loss_cls_dn_3": 0.10558, "loss_box_dn_3": 0.68413, "loss_cls_dn_4": 0.1119, "loss_box_dn_4": 0.69615, "loss_cls_dn_5": 0.10986, "loss_box_dn_5": 0.70218, "loss_dense_depth": 0.72302, "loss": 24.58688, "grad_norm": 41.65434, "time": 1.57214} -{"mode": "train", "epoch": 1, "iter": 297, "lr": 0.00022, "memory": 49164, "data_time": 0.06869, "loss_cls_0": 0.78528, "loss_box_0": 1.68433, "loss_cns_0": 0.62896, "loss_yns_0": 0.14951, "loss_cls_1": 0.86244, "loss_box_1": 1.55503, "loss_cns_1": 0.66104, "loss_yns_1": 0.14856, "loss_cls_2": 0.86335, "loss_box_2": 1.55616, "loss_cns_2": 0.6644, "loss_yns_2": 0.14973, "loss_cls_3": 0.89429, "loss_box_3": 1.53369, "loss_cns_3": 0.66335, "loss_yns_3": 0.14953, "loss_cls_4": 0.88706, "loss_box_4": 1.53234, "loss_cns_4": 0.66318, "loss_yns_4": 0.15037, "loss_cls_5": 0.88705, "loss_box_5": 1.5268, "loss_cns_5": 0.66102, "loss_yns_5": 0.14984, "loss_cls_dn_0": 0.18574, "loss_box_dn_0": 0.73634, "loss_cls_dn_1": 0.10583, "loss_box_dn_1": 0.69374, "loss_cls_dn_2": 0.10758, "loss_box_dn_2": 0.68727, "loss_cls_dn_3": 0.10973, "loss_box_dn_3": 0.6804, "loss_cls_dn_4": 0.11394, "loss_box_dn_4": 0.6897, "loss_cls_dn_5": 0.11565, "loss_box_dn_5": 0.69096, "loss_dense_depth": 0.74358, "loss": 25.06776, "grad_norm": 44.87664, "time": 1.58157} -{"mode": "train", "epoch": 1, "iter": 298, "lr": 0.00022, "memory": 49164, "data_time": 0.07018, "loss_cls_0": 0.77138, "loss_box_0": 1.66158, "loss_cns_0": 0.63287, "loss_yns_0": 0.14833, "loss_cls_1": 0.85708, "loss_box_1": 1.53844, "loss_cns_1": 0.66189, "loss_yns_1": 0.14976, "loss_cls_2": 0.85042, "loss_box_2": 1.52308, "loss_cns_2": 0.66427, "loss_yns_2": 0.1484, "loss_cls_3": 0.86481, "loss_box_3": 1.51428, "loss_cns_3": 0.66114, "loss_yns_3": 0.14846, "loss_cls_4": 0.87758, "loss_box_4": 1.4988, "loss_cns_4": 0.66217, "loss_yns_4": 0.14833, "loss_cls_5": 0.8672, "loss_box_5": 1.49545, "loss_cns_5": 0.6617, "loss_yns_5": 0.14799, "loss_cls_dn_0": 0.18468, "loss_box_dn_0": 0.73052, "loss_cls_dn_1": 0.10842, "loss_box_dn_1": 0.6724, "loss_cls_dn_2": 0.10827, "loss_box_dn_2": 0.65876, "loss_cls_dn_3": 0.10975, "loss_box_dn_3": 0.65803, "loss_cls_dn_4": 0.11317, "loss_box_dn_4": 0.6545, "loss_cls_dn_5": 0.11532, "loss_box_dn_5": 0.65576, "loss_dense_depth": 0.72177, "loss": 24.64675, "grad_norm": 27.69175, "time": 1.56394} -{"mode": "train", "epoch": 1, "iter": 299, "lr": 0.00022, "memory": 49164, "data_time": 0.07482, "loss_cls_0": 0.77762, "loss_box_0": 1.65731, "loss_cns_0": 0.62673, "loss_yns_0": 0.15055, "loss_cls_1": 0.86319, "loss_box_1": 1.5495, "loss_cns_1": 0.65842, "loss_yns_1": 0.15063, "loss_cls_2": 0.8728, "loss_box_2": 1.51256, "loss_cns_2": 0.65456, "loss_yns_2": 0.149, "loss_cls_3": 0.88804, "loss_box_3": 1.51816, "loss_cns_3": 0.65531, "loss_yns_3": 0.14902, "loss_cls_4": 0.88589, "loss_box_4": 1.52056, "loss_cns_4": 0.66094, "loss_yns_4": 0.15091, "loss_cls_5": 0.90047, "loss_box_5": 1.51285, "loss_cns_5": 0.65901, "loss_yns_5": 0.1496, "loss_cls_dn_0": 0.15811, "loss_box_dn_0": 0.73503, "loss_cls_dn_1": 0.10656, "loss_box_dn_1": 0.66805, "loss_cls_dn_2": 0.10358, "loss_box_dn_2": 0.65275, "loss_cls_dn_3": 0.1051, "loss_box_dn_3": 0.66228, "loss_cls_dn_4": 0.10627, "loss_box_dn_4": 0.65665, "loss_cls_dn_5": 0.10864, "loss_box_dn_5": 0.65898, "loss_dense_depth": 0.80495, "loss": 24.80059, "grad_norm": 39.33683, "time": 1.55749} -{"mode": "train", "epoch": 1, "iter": 300, "lr": 0.00022, "memory": 49164, "data_time": 0.06928, "loss_cls_0": 0.77447, "loss_box_0": 1.63076, "loss_cns_0": 0.63063, "loss_yns_0": 0.14959, "loss_cls_1": 0.83801, "loss_box_1": 1.56245, "loss_cns_1": 0.65434, "loss_yns_1": 0.14734, "loss_cls_2": 0.85794, "loss_box_2": 1.52417, "loss_cns_2": 0.64971, "loss_yns_2": 0.14543, "loss_cls_3": 0.88399, "loss_box_3": 1.52437, "loss_cns_3": 0.65821, "loss_yns_3": 0.14645, "loss_cls_4": 0.87791, "loss_box_4": 1.53889, "loss_cns_4": 0.65981, "loss_yns_4": 0.14927, "loss_cls_5": 0.90186, "loss_box_5": 1.52574, "loss_cns_5": 0.65558, "loss_yns_5": 0.14736, "loss_cls_dn_0": 0.16676, "loss_box_dn_0": 0.72915, "loss_cls_dn_1": 0.10618, "loss_box_dn_1": 0.66227, "loss_cls_dn_2": 0.10701, "loss_box_dn_2": 0.65194, "loss_cls_dn_3": 0.10999, "loss_box_dn_3": 0.65696, "loss_cls_dn_4": 0.11076, "loss_box_dn_4": 0.65913, "loss_cls_dn_5": 0.11511, "loss_box_dn_5": 0.66035, "loss_dense_depth": 0.74432, "loss": 24.7142, "grad_norm": 35.44067, "time": 1.55029} -{"mode": "train", "epoch": 1, "iter": 301, "lr": 0.00022, "memory": 49164, "data_time": 0.09661, "loss_cls_0": 0.8031, "loss_box_0": 1.65589, "loss_cns_0": 0.63464, "loss_yns_0": 0.15171, "loss_cls_1": 0.83645, "loss_box_1": 1.59646, "loss_cns_1": 0.6548, "loss_yns_1": 0.14781, "loss_cls_2": 0.84406, "loss_box_2": 1.55772, "loss_cns_2": 0.66148, "loss_yns_2": 0.14683, "loss_cls_3": 0.86681, "loss_box_3": 1.55607, "loss_cns_3": 0.66081, "loss_yns_3": 0.14823, "loss_cls_4": 0.85711, "loss_box_4": 1.59081, "loss_cns_4": 0.65919, "loss_yns_4": 0.14836, "loss_cls_5": 0.86716, "loss_box_5": 1.58751, "loss_cns_5": 0.65895, "loss_yns_5": 0.14769, "loss_cls_dn_0": 0.17915, "loss_box_dn_0": 0.73405, "loss_cls_dn_1": 0.1059, "loss_box_dn_1": 0.67208, "loss_cls_dn_2": 0.10648, "loss_box_dn_2": 0.66253, "loss_cls_dn_3": 0.1065, "loss_box_dn_3": 0.6643, "loss_cls_dn_4": 0.11018, "loss_box_dn_4": 0.67533, "loss_cls_dn_5": 0.111, "loss_box_dn_5": 0.68462, "loss_dense_depth": 0.79021, "loss": 25.04198, "grad_norm": 41.08827, "time": 1.63773} -{"mode": "train", "epoch": 1, "iter": 302, "lr": 0.00022, "memory": 49164, "data_time": 0.10201, "loss_cls_0": 0.78636, "loss_box_0": 1.64481, "loss_cns_0": 0.63163, "loss_yns_0": 0.15306, "loss_cls_1": 0.82632, "loss_box_1": 1.61462, "loss_cns_1": 0.65344, "loss_yns_1": 0.15023, "loss_cls_2": 0.84218, "loss_box_2": 1.56663, "loss_cns_2": 0.66192, "loss_yns_2": 0.15143, "loss_cls_3": 0.86239, "loss_box_3": 1.55994, "loss_cns_3": 0.65727, "loss_yns_3": 0.15084, "loss_cls_4": 0.86403, "loss_box_4": 1.57603, "loss_cns_4": 0.65533, "loss_yns_4": 0.15071, "loss_cls_5": 0.88662, "loss_box_5": 1.56874, "loss_cns_5": 0.65661, "loss_yns_5": 0.15288, "loss_cls_dn_0": 0.1725, "loss_box_dn_0": 0.72563, "loss_cls_dn_1": 0.1059, "loss_box_dn_1": 0.68257, "loss_cls_dn_2": 0.10489, "loss_box_dn_2": 0.66751, "loss_cls_dn_3": 0.10421, "loss_box_dn_3": 0.67276, "loss_cls_dn_4": 0.1126, "loss_box_dn_4": 0.68121, "loss_cls_dn_5": 0.11077, "loss_box_dn_5": 0.68928, "loss_dense_depth": 0.72634, "loss": 24.98019, "grad_norm": 41.42509, "time": 1.59875} -{"mode": "train", "epoch": 1, "iter": 303, "lr": 0.00022, "memory": 49164, "data_time": 0.0864, "loss_cls_0": 0.78105, "loss_box_0": 1.6656, "loss_cns_0": 0.62566, "loss_yns_0": 0.1513, "loss_cls_1": 0.83699, "loss_box_1": 1.55963, "loss_cns_1": 0.65498, "loss_yns_1": 0.15278, "loss_cls_2": 0.85049, "loss_box_2": 1.53506, "loss_cns_2": 0.65848, "loss_yns_2": 0.15315, "loss_cls_3": 0.86737, "loss_box_3": 1.52693, "loss_cns_3": 0.65984, "loss_yns_3": 0.15247, "loss_cls_4": 0.86845, "loss_box_4": 1.53614, "loss_cns_4": 0.66208, "loss_yns_4": 0.15307, "loss_cls_5": 0.89836, "loss_box_5": 1.54894, "loss_cns_5": 0.66038, "loss_yns_5": 0.15402, "loss_cls_dn_0": 0.15903, "loss_box_dn_0": 0.73466, "loss_cls_dn_1": 0.10454, "loss_box_dn_1": 0.69061, "loss_cls_dn_2": 0.1024, "loss_box_dn_2": 0.68515, "loss_cls_dn_3": 0.1034, "loss_box_dn_3": 0.68808, "loss_cls_dn_4": 0.10566, "loss_box_dn_4": 0.69575, "loss_cls_dn_5": 0.10637, "loss_box_dn_5": 0.70636, "loss_dense_depth": 0.76061, "loss": 24.95582, "grad_norm": 36.85221, "time": 1.60166} -{"mode": "train", "epoch": 1, "iter": 304, "lr": 0.00022, "memory": 49164, "data_time": 0.07494, "loss_cls_0": 0.79269, "loss_box_0": 1.64624, "loss_cns_0": 0.62373, "loss_yns_0": 0.15095, "loss_cls_1": 0.84125, "loss_box_1": 1.51774, "loss_cns_1": 0.66032, "loss_yns_1": 0.15156, "loss_cls_2": 0.84626, "loss_box_2": 1.50481, "loss_cns_2": 0.66046, "loss_yns_2": 0.15084, "loss_cls_3": 0.85518, "loss_box_3": 1.4972, "loss_cns_3": 0.66867, "loss_yns_3": 0.15031, "loss_cls_4": 0.85331, "loss_box_4": 1.49534, "loss_cns_4": 0.66492, "loss_yns_4": 0.14933, "loss_cls_5": 0.87354, "loss_box_5": 1.50675, "loss_cns_5": 0.66504, "loss_yns_5": 0.14934, "loss_cls_dn_0": 0.15753, "loss_box_dn_0": 0.72524, "loss_cls_dn_1": 0.10561, "loss_box_dn_1": 0.67075, "loss_cls_dn_2": 0.10483, "loss_box_dn_2": 0.65994, "loss_cls_dn_3": 0.10445, "loss_box_dn_3": 0.65783, "loss_cls_dn_4": 0.10554, "loss_box_dn_4": 0.66077, "loss_cls_dn_5": 0.10673, "loss_box_dn_5": 0.66861, "loss_dense_depth": 0.7995, "loss": 24.60311, "grad_norm": 32.87874, "time": 1.57496} -{"mode": "train", "epoch": 1, "iter": 305, "lr": 0.00022, "memory": 49164, "data_time": 0.08268, "loss_cls_0": 0.7835, "loss_box_0": 1.64006, "loss_cns_0": 0.625, "loss_yns_0": 0.14966, "loss_cls_1": 0.82469, "loss_box_1": 1.51008, "loss_cns_1": 0.65824, "loss_yns_1": 0.15159, "loss_cls_2": 0.83427, "loss_box_2": 1.4928, "loss_cns_2": 0.66212, "loss_yns_2": 0.15171, "loss_cls_3": 0.84846, "loss_box_3": 1.4885, "loss_cns_3": 0.66363, "loss_yns_3": 0.1516, "loss_cls_4": 0.85334, "loss_box_4": 1.47456, "loss_cns_4": 0.65811, "loss_yns_4": 0.15233, "loss_cls_5": 0.86825, "loss_box_5": 1.48218, "loss_cns_5": 0.66205, "loss_yns_5": 0.15108, "loss_cls_dn_0": 0.16638, "loss_box_dn_0": 0.73637, "loss_cls_dn_1": 0.10555, "loss_box_dn_1": 0.68422, "loss_cls_dn_2": 0.10708, "loss_box_dn_2": 0.66636, "loss_cls_dn_3": 0.10607, "loss_box_dn_3": 0.66482, "loss_cls_dn_4": 0.1085, "loss_box_dn_4": 0.6637, "loss_cls_dn_5": 0.10888, "loss_box_dn_5": 0.67157, "loss_dense_depth": 0.75142, "loss": 24.47873, "grad_norm": 36.28895, "time": 1.58337} -{"mode": "train", "epoch": 1, "iter": 306, "lr": 0.00022, "memory": 49164, "data_time": 0.10189, "loss_cls_0": 0.77475, "loss_box_0": 1.63687, "loss_cns_0": 0.63182, "loss_yns_0": 0.15062, "loss_cls_1": 0.83157, "loss_box_1": 1.51023, "loss_cns_1": 0.65972, "loss_yns_1": 0.14909, "loss_cls_2": 0.84145, "loss_box_2": 1.5007, "loss_cns_2": 0.66367, "loss_yns_2": 0.14968, "loss_cls_3": 0.85568, "loss_box_3": 1.47784, "loss_cns_3": 0.666, "loss_yns_3": 0.1494, "loss_cls_4": 0.85449, "loss_box_4": 1.47193, "loss_cns_4": 0.66388, "loss_yns_4": 0.14926, "loss_cls_5": 0.8646, "loss_box_5": 1.48656, "loss_cns_5": 0.66453, "loss_yns_5": 0.14821, "loss_cls_dn_0": 0.17576, "loss_box_dn_0": 0.7348, "loss_cls_dn_1": 0.10648, "loss_box_dn_1": 0.67364, "loss_cls_dn_2": 0.10791, "loss_box_dn_2": 0.66301, "loss_cls_dn_3": 0.10968, "loss_box_dn_3": 0.66516, "loss_cls_dn_4": 0.1113, "loss_box_dn_4": 0.66607, "loss_cls_dn_5": 0.11186, "loss_box_dn_5": 0.67902, "loss_dense_depth": 0.76071, "loss": 24.51794, "grad_norm": 33.24194, "time": 1.61847} -{"mode": "train", "epoch": 1, "iter": 307, "lr": 0.00022, "memory": 49164, "data_time": 0.06908, "loss_cls_0": 0.76221, "loss_box_0": 1.63993, "loss_cns_0": 0.63459, "loss_yns_0": 0.1503, "loss_cls_1": 0.83294, "loss_box_1": 1.55652, "loss_cns_1": 0.65808, "loss_yns_1": 0.14884, "loss_cls_2": 0.84257, "loss_box_2": 1.5326, "loss_cns_2": 0.66121, "loss_yns_2": 0.15038, "loss_cls_3": 0.85168, "loss_box_3": 1.50628, "loss_cns_3": 0.66378, "loss_yns_3": 0.15119, "loss_cls_4": 0.8505, "loss_box_4": 1.49274, "loss_cns_4": 0.66663, "loss_yns_4": 0.15044, "loss_cls_5": 0.86071, "loss_box_5": 1.48745, "loss_cns_5": 0.66384, "loss_yns_5": 0.15005, "loss_cls_dn_0": 0.16143, "loss_box_dn_0": 0.72468, "loss_cls_dn_1": 0.10536, "loss_box_dn_1": 0.66033, "loss_cls_dn_2": 0.10368, "loss_box_dn_2": 0.65533, "loss_cls_dn_3": 0.10708, "loss_box_dn_3": 0.65494, "loss_cls_dn_4": 0.1071, "loss_box_dn_4": 0.65914, "loss_cls_dn_5": 0.10899, "loss_box_dn_5": 0.6683, "loss_dense_depth": 0.79865, "loss": 24.58046, "grad_norm": 34.81723, "time": 1.58875} -{"mode": "train", "epoch": 1, "iter": 308, "lr": 0.00022, "memory": 49164, "data_time": 0.08036, "loss_cls_0": 0.77882, "loss_box_0": 1.63195, "loss_cns_0": 0.63615, "loss_yns_0": 0.1485, "loss_cls_1": 0.84143, "loss_box_1": 1.53617, "loss_cns_1": 0.65738, "loss_yns_1": 0.14818, "loss_cls_2": 0.84072, "loss_box_2": 1.50881, "loss_cns_2": 0.66078, "loss_yns_2": 0.14795, "loss_cls_3": 0.85385, "loss_box_3": 1.5132, "loss_cns_3": 0.66533, "loss_yns_3": 0.14788, "loss_cls_4": 0.85705, "loss_box_4": 1.5108, "loss_cns_4": 0.66291, "loss_yns_4": 0.14849, "loss_cls_5": 0.86537, "loss_box_5": 1.51178, "loss_cns_5": 0.66169, "loss_yns_5": 0.14713, "loss_cls_dn_0": 0.16204, "loss_box_dn_0": 0.73509, "loss_cls_dn_1": 0.10869, "loss_box_dn_1": 0.66795, "loss_cls_dn_2": 0.10538, "loss_box_dn_2": 0.66176, "loss_cls_dn_3": 0.10745, "loss_box_dn_3": 0.66897, "loss_cls_dn_4": 0.11048, "loss_box_dn_4": 0.67784, "loss_cls_dn_5": 0.11197, "loss_box_dn_5": 0.69035, "loss_dense_depth": 0.69405, "loss": 24.58437, "grad_norm": 39.90596, "time": 1.59599} -{"mode": "train", "epoch": 1, "iter": 309, "lr": 0.00022, "memory": 49164, "data_time": 0.08233, "loss_cls_0": 0.77692, "loss_box_0": 1.64479, "loss_cns_0": 0.63281, "loss_yns_0": 0.15094, "loss_cls_1": 0.83437, "loss_box_1": 1.53496, "loss_cns_1": 0.65754, "loss_yns_1": 0.14738, "loss_cls_2": 0.83686, "loss_box_2": 1.52235, "loss_cns_2": 0.6573, "loss_yns_2": 0.14619, "loss_cls_3": 0.85261, "loss_box_3": 1.51845, "loss_cns_3": 0.65986, "loss_yns_3": 0.14737, "loss_cls_4": 0.85391, "loss_box_4": 1.50915, "loss_cns_4": 0.65675, "loss_yns_4": 0.14727, "loss_cls_5": 0.85824, "loss_box_5": 1.51046, "loss_cns_5": 0.65715, "loss_yns_5": 0.14659, "loss_cls_dn_0": 0.17113, "loss_box_dn_0": 0.72881, "loss_cls_dn_1": 0.10936, "loss_box_dn_1": 0.6726, "loss_cls_dn_2": 0.10748, "loss_box_dn_2": 0.66525, "loss_cls_dn_3": 0.10824, "loss_box_dn_3": 0.66868, "loss_cls_dn_4": 0.11194, "loss_box_dn_4": 0.67007, "loss_cls_dn_5": 0.11121, "loss_box_dn_5": 0.68006, "loss_dense_depth": 0.76706, "loss": 24.6321, "grad_norm": 40.32322, "time": 1.57602} -{"mode": "train", "epoch": 1, "iter": 310, "lr": 0.00022, "memory": 49164, "data_time": 0.08602, "loss_cls_0": 0.75652, "loss_box_0": 1.63312, "loss_cns_0": 0.63644, "loss_yns_0": 0.1497, "loss_cls_1": 0.837, "loss_box_1": 1.49308, "loss_cns_1": 0.65415, "loss_yns_1": 0.14498, "loss_cls_2": 0.84382, "loss_box_2": 1.4642, "loss_cns_2": 0.6548, "loss_yns_2": 0.14286, "loss_cls_3": 0.85322, "loss_box_3": 1.44324, "loss_cns_3": 0.65109, "loss_yns_3": 0.14284, "loss_cls_4": 0.84996, "loss_box_4": 1.46352, "loss_cns_4": 0.66018, "loss_yns_4": 0.14761, "loss_cls_5": 0.87047, "loss_box_5": 1.47541, "loss_cns_5": 0.65823, "loss_yns_5": 0.14599, "loss_cls_dn_0": 0.1685, "loss_box_dn_0": 0.73383, "loss_cls_dn_1": 0.10649, "loss_box_dn_1": 0.6578, "loss_cls_dn_2": 0.10434, "loss_box_dn_2": 0.64069, "loss_cls_dn_3": 0.10499, "loss_box_dn_3": 0.63694, "loss_cls_dn_4": 0.1061, "loss_box_dn_4": 0.63939, "loss_cls_dn_5": 0.11067, "loss_box_dn_5": 0.64822, "loss_dense_depth": 0.70715, "loss": 24.13753, "grad_norm": 31.97168, "time": 1.59512} -{"mode": "train", "epoch": 1, "iter": 311, "lr": 0.00022, "memory": 49164, "data_time": 0.08188, "loss_cls_0": 0.76107, "loss_box_0": 1.6213, "loss_cns_0": 0.6314, "loss_yns_0": 0.14797, "loss_cls_1": 0.83462, "loss_box_1": 1.49677, "loss_cns_1": 0.65266, "loss_yns_1": 0.14433, "loss_cls_2": 0.84408, "loss_box_2": 1.48128, "loss_cns_2": 0.65526, "loss_yns_2": 0.14381, "loss_cls_3": 0.85137, "loss_box_3": 1.45784, "loss_cns_3": 0.64827, "loss_yns_3": 0.14245, "loss_cls_4": 0.84582, "loss_box_4": 1.48062, "loss_cns_4": 0.65606, "loss_yns_4": 0.14644, "loss_cls_5": 0.87324, "loss_box_5": 1.48521, "loss_cns_5": 0.65502, "loss_yns_5": 0.14484, "loss_cls_dn_0": 0.16151, "loss_box_dn_0": 0.73114, "loss_cls_dn_1": 0.10947, "loss_box_dn_1": 0.64979, "loss_cls_dn_2": 0.10866, "loss_box_dn_2": 0.64076, "loss_cls_dn_3": 0.1084, "loss_box_dn_3": 0.63599, "loss_cls_dn_4": 0.10906, "loss_box_dn_4": 0.63912, "loss_cls_dn_5": 0.11788, "loss_box_dn_5": 0.64548, "loss_dense_depth": 0.79532, "loss": 24.25431, "grad_norm": 35.53914, "time": 1.61549} -{"mode": "train", "epoch": 1, "iter": 312, "lr": 0.00022, "memory": 49164, "data_time": 0.08142, "loss_cls_0": 0.77749, "loss_box_0": 1.65579, "loss_cns_0": 0.62937, "loss_yns_0": 0.14714, "loss_cls_1": 0.82542, "loss_box_1": 1.52251, "loss_cns_1": 0.65268, "loss_yns_1": 0.14513, "loss_cls_2": 0.8428, "loss_box_2": 1.50498, "loss_cns_2": 0.655, "loss_yns_2": 0.14519, "loss_cls_3": 0.85118, "loss_box_3": 1.48288, "loss_cns_3": 0.65368, "loss_yns_3": 0.1452, "loss_cls_4": 0.84724, "loss_box_4": 1.48136, "loss_cns_4": 0.65559, "loss_yns_4": 0.14496, "loss_cls_5": 0.84934, "loss_box_5": 1.49454, "loss_cns_5": 0.65644, "loss_yns_5": 0.14554, "loss_cls_dn_0": 0.16198, "loss_box_dn_0": 0.72821, "loss_cls_dn_1": 0.11336, "loss_box_dn_1": 0.64804, "loss_cls_dn_2": 0.10914, "loss_box_dn_2": 0.64228, "loss_cls_dn_3": 0.11072, "loss_box_dn_3": 0.642, "loss_cls_dn_4": 0.10853, "loss_box_dn_4": 0.64648, "loss_cls_dn_5": 0.11035, "loss_box_dn_5": 0.65888, "loss_dense_depth": 0.75047, "loss": 24.34185, "grad_norm": 33.63667, "time": 1.59258} -{"mode": "train", "epoch": 1, "iter": 313, "lr": 0.00022, "memory": 49164, "data_time": 0.07911, "loss_cls_0": 0.76784, "loss_box_0": 1.59798, "loss_cns_0": 0.63471, "loss_yns_0": 0.14889, "loss_cls_1": 0.82781, "loss_box_1": 1.48602, "loss_cns_1": 0.66088, "loss_yns_1": 0.14791, "loss_cls_2": 0.83551, "loss_box_2": 1.46397, "loss_cns_2": 0.66156, "loss_yns_2": 0.14714, "loss_cls_3": 0.82513, "loss_box_3": 1.45623, "loss_cns_3": 0.66143, "loss_yns_3": 0.14685, "loss_cls_4": 0.83672, "loss_box_4": 1.4523, "loss_cns_4": 0.66162, "loss_yns_4": 0.14684, "loss_cls_5": 0.84258, "loss_box_5": 1.44063, "loss_cns_5": 0.66704, "loss_yns_5": 0.14672, "loss_cls_dn_0": 0.1545, "loss_box_dn_0": 0.72597, "loss_cls_dn_1": 0.10485, "loss_box_dn_1": 0.68123, "loss_cls_dn_2": 0.10423, "loss_box_dn_2": 0.67341, "loss_cls_dn_3": 0.10435, "loss_box_dn_3": 0.67722, "loss_cls_dn_4": 0.11069, "loss_box_dn_4": 0.6823, "loss_cls_dn_5": 0.11768, "loss_box_dn_5": 0.68695, "loss_dense_depth": 0.78706, "loss": 24.27474, "grad_norm": 40.36639, "time": 1.58948} -{"mode": "train", "epoch": 1, "iter": 314, "lr": 0.00023, "memory": 49164, "data_time": 0.07669, "loss_cls_0": 0.77983, "loss_box_0": 1.60421, "loss_cns_0": 0.63265, "loss_yns_0": 0.14843, "loss_cls_1": 0.85048, "loss_box_1": 1.51751, "loss_cns_1": 0.65903, "loss_yns_1": 0.14637, "loss_cls_2": 0.85343, "loss_box_2": 1.50496, "loss_cns_2": 0.65937, "loss_yns_2": 0.14585, "loss_cls_3": 0.85408, "loss_box_3": 1.486, "loss_cns_3": 0.66218, "loss_yns_3": 0.14677, "loss_cls_4": 0.86555, "loss_box_4": 1.46767, "loss_cns_4": 0.66063, "loss_yns_4": 0.14978, "loss_cls_5": 0.87723, "loss_box_5": 1.47299, "loss_cns_5": 0.66367, "loss_yns_5": 0.14766, "loss_cls_dn_0": 0.16055, "loss_box_dn_0": 0.72927, "loss_cls_dn_1": 0.11375, "loss_box_dn_1": 0.70543, "loss_cls_dn_2": 0.11787, "loss_box_dn_2": 0.7048, "loss_cls_dn_3": 0.11606, "loss_box_dn_3": 0.70438, "loss_cls_dn_4": 0.1214, "loss_box_dn_4": 0.70076, "loss_cls_dn_5": 0.13151, "loss_box_dn_5": 0.70665, "loss_dense_depth": 0.71953, "loss": 24.68828, "grad_norm": 40.20514, "time": 1.63015} -{"mode": "train", "epoch": 1, "iter": 315, "lr": 0.00023, "memory": 49164, "data_time": 0.07391, "loss_cls_0": 0.76177, "loss_box_0": 1.61112, "loss_cns_0": 0.6303, "loss_yns_0": 0.14678, "loss_cls_1": 0.82366, "loss_box_1": 1.52557, "loss_cns_1": 0.65591, "loss_yns_1": 0.14799, "loss_cls_2": 0.82874, "loss_box_2": 1.4833, "loss_cns_2": 0.6564, "loss_yns_2": 0.1456, "loss_cls_3": 0.8386, "loss_box_3": 1.46389, "loss_cns_3": 0.65547, "loss_yns_3": 0.14574, "loss_cls_4": 0.84032, "loss_box_4": 1.46272, "loss_cns_4": 0.66193, "loss_yns_4": 0.1467, "loss_cls_5": 0.85857, "loss_box_5": 1.48381, "loss_cns_5": 0.65576, "loss_yns_5": 0.14681, "loss_cls_dn_0": 0.15386, "loss_box_dn_0": 0.73095, "loss_cls_dn_1": 0.10733, "loss_box_dn_1": 0.69346, "loss_cls_dn_2": 0.10664, "loss_box_dn_2": 0.68255, "loss_cls_dn_3": 0.10606, "loss_box_dn_3": 0.67565, "loss_cls_dn_4": 0.10623, "loss_box_dn_4": 0.67437, "loss_cls_dn_5": 0.11452, "loss_box_dn_5": 0.68686, "loss_dense_depth": 0.75873, "loss": 24.37466, "grad_norm": 31.61317, "time": 1.58159} -{"mode": "train", "epoch": 1, "iter": 316, "lr": 0.00023, "memory": 49164, "data_time": 0.07815, "loss_cls_0": 0.76189, "loss_box_0": 1.58119, "loss_cns_0": 0.62967, "loss_yns_0": 0.14386, "loss_cls_1": 0.81921, "loss_box_1": 1.50209, "loss_cns_1": 0.65833, "loss_yns_1": 0.14373, "loss_cls_2": 0.83265, "loss_box_2": 1.45076, "loss_cns_2": 0.6628, "loss_yns_2": 0.1439, "loss_cls_3": 0.83094, "loss_box_3": 1.43682, "loss_cns_3": 0.65959, "loss_yns_3": 0.14257, "loss_cls_4": 0.84042, "loss_box_4": 1.43292, "loss_cns_4": 0.66559, "loss_yns_4": 0.1437, "loss_cls_5": 0.85007, "loss_box_5": 1.44241, "loss_cns_5": 0.66076, "loss_yns_5": 0.14334, "loss_cls_dn_0": 0.15086, "loss_box_dn_0": 0.72072, "loss_cls_dn_1": 0.10754, "loss_box_dn_1": 0.67504, "loss_cls_dn_2": 0.10646, "loss_box_dn_2": 0.65999, "loss_cls_dn_3": 0.10775, "loss_box_dn_3": 0.65248, "loss_cls_dn_4": 0.11135, "loss_box_dn_4": 0.65113, "loss_cls_dn_5": 0.10995, "loss_box_dn_5": 0.65732, "loss_dense_depth": 0.69958, "loss": 23.98936, "grad_norm": 32.63664, "time": 1.57091} -{"mode": "train", "epoch": 1, "iter": 317, "lr": 0.00023, "memory": 49164, "data_time": 0.08003, "loss_cls_0": 0.74172, "loss_box_0": 1.56591, "loss_cns_0": 0.62857, "loss_yns_0": 0.14223, "loss_cls_1": 0.81436, "loss_box_1": 1.44306, "loss_cns_1": 0.65698, "loss_yns_1": 0.14333, "loss_cls_2": 0.83176, "loss_box_2": 1.42801, "loss_cns_2": 0.65968, "loss_yns_2": 0.14449, "loss_cls_3": 0.82132, "loss_box_3": 1.42314, "loss_cns_3": 0.65877, "loss_yns_3": 0.14398, "loss_cls_4": 0.82888, "loss_box_4": 1.41085, "loss_cns_4": 0.65973, "loss_yns_4": 0.14452, "loss_cls_5": 0.8478, "loss_box_5": 1.42739, "loss_cns_5": 0.65894, "loss_yns_5": 0.14287, "loss_cls_dn_0": 0.14992, "loss_box_dn_0": 0.7277, "loss_cls_dn_1": 0.1084, "loss_box_dn_1": 0.66232, "loss_cls_dn_2": 0.10718, "loss_box_dn_2": 0.65164, "loss_cls_dn_3": 0.10778, "loss_box_dn_3": 0.65094, "loss_cls_dn_4": 0.1098, "loss_box_dn_4": 0.64962, "loss_cls_dn_5": 0.10936, "loss_box_dn_5": 0.66101, "loss_dense_depth": 0.72296, "loss": 23.78695, "grad_norm": 29.11787, "time": 1.5958} -{"mode": "train", "epoch": 1, "iter": 318, "lr": 0.00023, "memory": 49164, "data_time": 0.07504, "loss_cls_0": 0.77667, "loss_box_0": 1.5665, "loss_cns_0": 0.62333, "loss_yns_0": 0.1426, "loss_cls_1": 0.82611, "loss_box_1": 1.47279, "loss_cns_1": 0.65938, "loss_yns_1": 0.14436, "loss_cls_2": 0.83941, "loss_box_2": 1.45403, "loss_cns_2": 0.6609, "loss_yns_2": 0.14403, "loss_cls_3": 0.84056, "loss_box_3": 1.45177, "loss_cns_3": 0.66236, "loss_yns_3": 0.14524, "loss_cls_4": 0.84328, "loss_box_4": 1.43314, "loss_cns_4": 0.66179, "loss_yns_4": 0.14704, "loss_cls_5": 0.85846, "loss_box_5": 1.44297, "loss_cns_5": 0.65884, "loss_yns_5": 0.14429, "loss_cls_dn_0": 0.15279, "loss_box_dn_0": 0.73104, "loss_cls_dn_1": 0.10152, "loss_box_dn_1": 0.65277, "loss_cls_dn_2": 0.09995, "loss_box_dn_2": 0.6422, "loss_cls_dn_3": 0.09975, "loss_box_dn_3": 0.64913, "loss_cls_dn_4": 0.10222, "loss_box_dn_4": 0.64818, "loss_cls_dn_5": 0.10575, "loss_box_dn_5": 0.65801, "loss_dense_depth": 0.74645, "loss": 23.98963, "grad_norm": 29.78831, "time": 1.55837} -{"mode": "train", "epoch": 1, "iter": 319, "lr": 0.00023, "memory": 49164, "data_time": 0.08039, "loss_cls_0": 0.74251, "loss_box_0": 1.56425, "loss_cns_0": 0.63468, "loss_yns_0": 0.14578, "loss_cls_1": 0.80192, "loss_box_1": 1.45497, "loss_cns_1": 0.66247, "loss_yns_1": 0.14503, "loss_cls_2": 0.81387, "loss_box_2": 1.422, "loss_cns_2": 0.66762, "loss_yns_2": 0.14502, "loss_cls_3": 0.82645, "loss_box_3": 1.41559, "loss_cns_3": 0.66432, "loss_yns_3": 0.14555, "loss_cls_4": 0.82268, "loss_box_4": 1.41599, "loss_cns_4": 0.66841, "loss_yns_4": 0.14412, "loss_cls_5": 0.82905, "loss_box_5": 1.41672, "loss_cns_5": 0.66206, "loss_yns_5": 0.1446, "loss_cls_dn_0": 0.14227, "loss_box_dn_0": 0.73409, "loss_cls_dn_1": 0.09768, "loss_box_dn_1": 0.66202, "loss_cls_dn_2": 0.09677, "loss_box_dn_2": 0.65277, "loss_cls_dn_3": 0.09749, "loss_box_dn_3": 0.65522, "loss_cls_dn_4": 0.10034, "loss_box_dn_4": 0.66137, "loss_cls_dn_5": 0.10514, "loss_box_dn_5": 0.6679, "loss_dense_depth": 0.7044, "loss": 23.73311, "grad_norm": 28.9687, "time": 1.57939} -{"mode": "train", "epoch": 1, "iter": 320, "lr": 0.00023, "memory": 49164, "data_time": 0.07662, "loss_cls_0": 0.74863, "loss_box_0": 1.58899, "loss_cns_0": 0.63746, "loss_yns_0": 0.14445, "loss_cls_1": 0.82095, "loss_box_1": 1.45483, "loss_cns_1": 0.66245, "loss_yns_1": 0.14325, "loss_cls_2": 0.84573, "loss_box_2": 1.42872, "loss_cns_2": 0.67023, "loss_yns_2": 0.1439, "loss_cls_3": 0.83697, "loss_box_3": 1.41755, "loss_cns_3": 0.66584, "loss_yns_3": 0.14366, "loss_cls_4": 0.8336, "loss_box_4": 1.4226, "loss_cns_4": 0.66555, "loss_yns_4": 0.14421, "loss_cls_5": 0.84476, "loss_box_5": 1.4136, "loss_cns_5": 0.66329, "loss_yns_5": 0.1439, "loss_cls_dn_0": 0.1433, "loss_box_dn_0": 0.71794, "loss_cls_dn_1": 0.1023, "loss_box_dn_1": 0.65951, "loss_cls_dn_2": 0.09998, "loss_box_dn_2": 0.64858, "loss_cls_dn_3": 0.1003, "loss_box_dn_3": 0.64628, "loss_cls_dn_4": 0.10265, "loss_box_dn_4": 0.65311, "loss_cls_dn_5": 0.10412, "loss_box_dn_5": 0.65746, "loss_dense_depth": 0.72145, "loss": 23.84211, "grad_norm": 29.43163, "time": 1.58813} -{"mode": "train", "epoch": 1, "iter": 321, "lr": 0.00023, "memory": 49164, "data_time": 0.11258, "loss_cls_0": 0.71522, "loss_box_0": 1.57077, "loss_cns_0": 0.64189, "loss_yns_0": 0.14735, "loss_cls_1": 0.79906, "loss_box_1": 1.44571, "loss_cns_1": 0.66375, "loss_yns_1": 0.14572, "loss_cls_2": 0.80837, "loss_box_2": 1.42963, "loss_cns_2": 0.66723, "loss_yns_2": 0.14531, "loss_cls_3": 0.81454, "loss_box_3": 1.40964, "loss_cns_3": 0.66629, "loss_yns_3": 0.14394, "loss_cls_4": 0.8256, "loss_box_4": 1.39499, "loss_cns_4": 0.66211, "loss_yns_4": 0.14444, "loss_cls_5": 0.82698, "loss_box_5": 1.39306, "loss_cns_5": 0.66097, "loss_yns_5": 0.14548, "loss_cls_dn_0": 0.14248, "loss_box_dn_0": 0.72157, "loss_cls_dn_1": 0.10101, "loss_box_dn_1": 0.66216, "loss_cls_dn_2": 0.09767, "loss_box_dn_2": 0.64887, "loss_cls_dn_3": 0.10166, "loss_box_dn_3": 0.64214, "loss_cls_dn_4": 0.10417, "loss_box_dn_4": 0.63824, "loss_cls_dn_5": 0.10378, "loss_box_dn_5": 0.64372, "loss_dense_depth": 0.67639, "loss": 23.55189, "grad_norm": 31.52357, "time": 1.66936} -{"mode": "train", "epoch": 1, "iter": 322, "lr": 0.00023, "memory": 49164, "data_time": 0.11612, "loss_cls_0": 0.72985, "loss_box_0": 1.57826, "loss_cns_0": 0.63974, "loss_yns_0": 0.14848, "loss_cls_1": 0.82104, "loss_box_1": 1.3932, "loss_cns_1": 0.66201, "loss_yns_1": 0.14437, "loss_cls_2": 0.82149, "loss_box_2": 1.36868, "loss_cns_2": 0.66643, "loss_yns_2": 0.14365, "loss_cls_3": 0.835, "loss_box_3": 1.34416, "loss_cns_3": 0.66001, "loss_yns_3": 0.14282, "loss_cls_4": 0.82968, "loss_box_4": 1.36299, "loss_cns_4": 0.66454, "loss_yns_4": 0.14518, "loss_cls_5": 0.83773, "loss_box_5": 1.35873, "loss_cns_5": 0.66327, "loss_yns_5": 0.14549, "loss_cls_dn_0": 0.14549, "loss_box_dn_0": 0.72247, "loss_cls_dn_1": 0.10068, "loss_box_dn_1": 0.66115, "loss_cls_dn_2": 0.0983, "loss_box_dn_2": 0.64354, "loss_cls_dn_3": 0.10155, "loss_box_dn_3": 0.64133, "loss_cls_dn_4": 0.10117, "loss_box_dn_4": 0.64672, "loss_cls_dn_5": 0.10139, "loss_box_dn_5": 0.65072, "loss_dense_depth": 0.68446, "loss": 23.40576, "grad_norm": 26.82585, "time": 1.64446} -{"mode": "train", "epoch": 1, "iter": 323, "lr": 0.00023, "memory": 49164, "data_time": 0.08171, "loss_cls_0": 0.72095, "loss_box_0": 1.58126, "loss_cns_0": 0.64088, "loss_yns_0": 0.14437, "loss_cls_1": 0.79062, "loss_box_1": 1.45122, "loss_cns_1": 0.66468, "loss_yns_1": 0.14482, "loss_cls_2": 0.8065, "loss_box_2": 1.42607, "loss_cns_2": 0.66603, "loss_yns_2": 0.1456, "loss_cls_3": 0.81872, "loss_box_3": 1.4022, "loss_cns_3": 0.66024, "loss_yns_3": 0.14317, "loss_cls_4": 0.82877, "loss_box_4": 1.42381, "loss_cns_4": 0.66945, "loss_yns_4": 0.14628, "loss_cls_5": 0.82635, "loss_box_5": 1.41205, "loss_cns_5": 0.6649, "loss_yns_5": 0.14512, "loss_cls_dn_0": 0.14392, "loss_box_dn_0": 0.72574, "loss_cls_dn_1": 0.09852, "loss_box_dn_1": 0.66027, "loss_cls_dn_2": 0.09763, "loss_box_dn_2": 0.64637, "loss_cls_dn_3": 0.09764, "loss_box_dn_3": 0.65061, "loss_cls_dn_4": 0.09746, "loss_box_dn_4": 0.66019, "loss_cls_dn_5": 0.09942, "loss_box_dn_5": 0.66458, "loss_dense_depth": 0.65877, "loss": 23.62517, "grad_norm": 33.93236, "time": 1.59839} -{"mode": "train", "epoch": 1, "iter": 324, "lr": 0.00023, "memory": 49164, "data_time": 0.07694, "loss_cls_0": 0.73468, "loss_box_0": 1.61182, "loss_cns_0": 0.63514, "loss_yns_0": 0.14353, "loss_cls_1": 0.81455, "loss_box_1": 1.47096, "loss_cns_1": 0.66519, "loss_yns_1": 0.1448, "loss_cls_2": 0.81384, "loss_box_2": 1.44785, "loss_cns_2": 0.6681, "loss_yns_2": 0.14291, "loss_cls_3": 0.83295, "loss_box_3": 1.41282, "loss_cns_3": 0.65889, "loss_yns_3": 0.14047, "loss_cls_4": 0.83184, "loss_box_4": 1.43052, "loss_cns_4": 0.66791, "loss_yns_4": 0.14279, "loss_cls_5": 0.84174, "loss_box_5": 1.43142, "loss_cns_5": 0.6678, "loss_yns_5": 0.14244, "loss_cls_dn_0": 0.13993, "loss_box_dn_0": 0.72595, "loss_cls_dn_1": 0.09798, "loss_box_dn_1": 0.65781, "loss_cls_dn_2": 0.09623, "loss_box_dn_2": 0.65141, "loss_cls_dn_3": 0.09695, "loss_box_dn_3": 0.64748, "loss_cls_dn_4": 0.09716, "loss_box_dn_4": 0.65683, "loss_cls_dn_5": 0.09963, "loss_box_dn_5": 0.66487, "loss_dense_depth": 0.66727, "loss": 23.79445, "grad_norm": 34.11447, "time": 1.59198} -{"mode": "train", "epoch": 1, "iter": 325, "lr": 0.00023, "memory": 49164, "data_time": 0.08155, "loss_cls_0": 0.72457, "loss_box_0": 1.57432, "loss_cns_0": 0.63758, "loss_yns_0": 0.144, "loss_cls_1": 0.79187, "loss_box_1": 1.46794, "loss_cns_1": 0.66396, "loss_yns_1": 0.14605, "loss_cls_2": 0.79236, "loss_box_2": 1.45087, "loss_cns_2": 0.66418, "loss_yns_2": 0.14514, "loss_cls_3": 0.80325, "loss_box_3": 1.42584, "loss_cns_3": 0.66062, "loss_yns_3": 0.14348, "loss_cls_4": 0.80021, "loss_box_4": 1.43812, "loss_cns_4": 0.66505, "loss_yns_4": 0.14505, "loss_cls_5": 0.80968, "loss_box_5": 1.43392, "loss_cns_5": 0.66583, "loss_yns_5": 0.14502, "loss_cls_dn_0": 0.14631, "loss_box_dn_0": 0.7175, "loss_cls_dn_1": 0.09852, "loss_box_dn_1": 0.65364, "loss_cls_dn_2": 0.09744, "loss_box_dn_2": 0.65025, "loss_cls_dn_3": 0.09914, "loss_box_dn_3": 0.64548, "loss_cls_dn_4": 0.09906, "loss_box_dn_4": 0.64633, "loss_cls_dn_5": 0.10008, "loss_box_dn_5": 0.65081, "loss_dense_depth": 0.68683, "loss": 23.63033, "grad_norm": 29.85307, "time": 1.58292} -{"mode": "train", "epoch": 1, "iter": 326, "lr": 0.00023, "memory": 49164, "data_time": 0.11088, "loss_cls_0": 0.71252, "loss_box_0": 1.60093, "loss_cns_0": 0.63597, "loss_yns_0": 0.14486, "loss_cls_1": 0.78322, "loss_box_1": 1.46131, "loss_cns_1": 0.66319, "loss_yns_1": 0.14332, "loss_cls_2": 0.78627, "loss_box_2": 1.43827, "loss_cns_2": 0.66356, "loss_yns_2": 0.14196, "loss_cls_3": 0.79363, "loss_box_3": 1.42395, "loss_cns_3": 0.66279, "loss_yns_3": 0.14204, "loss_cls_4": 0.80367, "loss_box_4": 1.4291, "loss_cns_4": 0.66348, "loss_yns_4": 0.14293, "loss_cls_5": 0.81123, "loss_box_5": 1.424, "loss_cns_5": 0.66381, "loss_yns_5": 0.14048, "loss_cls_dn_0": 0.14157, "loss_box_dn_0": 0.71923, "loss_cls_dn_1": 0.09677, "loss_box_dn_1": 0.63113, "loss_cls_dn_2": 0.09689, "loss_box_dn_2": 0.62327, "loss_cls_dn_3": 0.10044, "loss_box_dn_3": 0.61771, "loss_cls_dn_4": 0.09938, "loss_box_dn_4": 0.62039, "loss_cls_dn_5": 0.09874, "loss_box_dn_5": 0.62316, "loss_dense_depth": 0.66737, "loss": 23.41256, "grad_norm": 27.73094, "time": 1.61413} -{"mode": "train", "epoch": 1, "iter": 327, "lr": 0.00023, "memory": 49164, "data_time": 0.07831, "loss_cls_0": 0.72123, "loss_box_0": 1.608, "loss_cns_0": 0.63945, "loss_yns_0": 0.13948, "loss_cls_1": 0.77615, "loss_box_1": 1.47741, "loss_cns_1": 0.65961, "loss_yns_1": 0.13732, "loss_cls_2": 0.797, "loss_box_2": 1.44194, "loss_cns_2": 0.66147, "loss_yns_2": 0.13909, "loss_cls_3": 0.80721, "loss_box_3": 1.44586, "loss_cns_3": 0.66137, "loss_yns_3": 0.13863, "loss_cls_4": 0.8101, "loss_box_4": 1.44934, "loss_cns_4": 0.66104, "loss_yns_4": 0.13766, "loss_cls_5": 0.81676, "loss_box_5": 1.43462, "loss_cns_5": 0.65975, "loss_yns_5": 0.13822, "loss_cls_dn_0": 0.13971, "loss_box_dn_0": 0.71438, "loss_cls_dn_1": 0.09895, "loss_box_dn_1": 0.62244, "loss_cls_dn_2": 0.09946, "loss_box_dn_2": 0.60913, "loss_cls_dn_3": 0.1016, "loss_box_dn_3": 0.61034, "loss_cls_dn_4": 0.10145, "loss_box_dn_4": 0.61405, "loss_cls_dn_5": 0.10222, "loss_box_dn_5": 0.61597, "loss_dense_depth": 0.68573, "loss": 23.47414, "grad_norm": 30.70921, "time": 1.57567} -{"mode": "train", "epoch": 1, "iter": 328, "lr": 0.00023, "memory": 49164, "data_time": 0.08311, "loss_cls_0": 0.71614, "loss_box_0": 1.62426, "loss_cns_0": 0.64199, "loss_yns_0": 0.14423, "loss_cls_1": 0.77193, "loss_box_1": 1.47395, "loss_cns_1": 0.66178, "loss_yns_1": 0.13992, "loss_cls_2": 0.79834, "loss_box_2": 1.45969, "loss_cns_2": 0.66594, "loss_yns_2": 0.1403, "loss_cls_3": 0.81937, "loss_box_3": 1.43582, "loss_cns_3": 0.6687, "loss_yns_3": 0.13976, "loss_cls_4": 0.80738, "loss_box_4": 1.43445, "loss_cns_4": 0.66953, "loss_yns_4": 0.14044, "loss_cls_5": 0.81351, "loss_box_5": 1.43334, "loss_cns_5": 0.66915, "loss_yns_5": 0.13889, "loss_cls_dn_0": 0.13937, "loss_box_dn_0": 0.72459, "loss_cls_dn_1": 0.09943, "loss_box_dn_1": 0.63586, "loss_cls_dn_2": 0.09772, "loss_box_dn_2": 0.63038, "loss_cls_dn_3": 0.09677, "loss_box_dn_3": 0.62551, "loss_cls_dn_4": 0.10027, "loss_box_dn_4": 0.62841, "loss_cls_dn_5": 0.10409, "loss_box_dn_5": 0.6379, "loss_dense_depth": 0.70014, "loss": 23.62928, "grad_norm": 34.41777, "time": 1.58654} -{"mode": "train", "epoch": 1, "iter": 329, "lr": 0.00023, "memory": 49164, "data_time": 0.08948, "loss_cls_0": 0.7293, "loss_box_0": 1.62411, "loss_cns_0": 0.63557, "loss_yns_0": 0.14256, "loss_cls_1": 0.79322, "loss_box_1": 1.48269, "loss_cns_1": 0.66327, "loss_yns_1": 0.14092, "loss_cls_2": 0.80627, "loss_box_2": 1.46837, "loss_cns_2": 0.66701, "loss_yns_2": 0.14324, "loss_cls_3": 0.82687, "loss_box_3": 1.44617, "loss_cns_3": 0.66736, "loss_yns_3": 0.1435, "loss_cls_4": 0.82286, "loss_box_4": 1.45233, "loss_cns_4": 0.66979, "loss_yns_4": 0.14854, "loss_cls_5": 0.81958, "loss_box_5": 1.45066, "loss_cns_5": 0.67282, "loss_yns_5": 0.14125, "loss_cls_dn_0": 0.14628, "loss_box_dn_0": 0.72383, "loss_cls_dn_1": 0.10255, "loss_box_dn_1": 0.64787, "loss_cls_dn_2": 0.10385, "loss_box_dn_2": 0.64246, "loss_cls_dn_3": 0.10375, "loss_box_dn_3": 0.64067, "loss_cls_dn_4": 0.10423, "loss_box_dn_4": 0.65035, "loss_cls_dn_5": 0.10611, "loss_box_dn_5": 0.66102, "loss_dense_depth": 0.74789, "loss": 23.93912, "grad_norm": 29.13249, "time": 1.60599} -{"mode": "train", "epoch": 1, "iter": 330, "lr": 0.00023, "memory": 49164, "data_time": 0.0893, "loss_cls_0": 0.70978, "loss_box_0": 1.60064, "loss_cns_0": 0.63815, "loss_yns_0": 0.14122, "loss_cls_1": 0.77873, "loss_box_1": 1.46347, "loss_cns_1": 0.66571, "loss_yns_1": 0.13893, "loss_cls_2": 0.78866, "loss_box_2": 1.44423, "loss_cns_2": 0.66708, "loss_yns_2": 0.141, "loss_cls_3": 0.8076, "loss_box_3": 1.45223, "loss_cns_3": 0.66665, "loss_yns_3": 0.14078, "loss_cls_4": 0.80394, "loss_box_4": 1.43861, "loss_cns_4": 0.66721, "loss_yns_4": 0.14013, "loss_cls_5": 0.80532, "loss_box_5": 1.43029, "loss_cns_5": 0.67197, "loss_yns_5": 0.13929, "loss_cls_dn_0": 0.14424, "loss_box_dn_0": 0.72953, "loss_cls_dn_1": 0.09984, "loss_box_dn_1": 0.65869, "loss_cls_dn_2": 0.10147, "loss_box_dn_2": 0.64826, "loss_cls_dn_3": 0.10087, "loss_box_dn_3": 0.65291, "loss_cls_dn_4": 0.10251, "loss_box_dn_4": 0.65475, "loss_cls_dn_5": 0.10068, "loss_box_dn_5": 0.65704, "loss_dense_depth": 0.7456, "loss": 23.73801, "grad_norm": 41.23397, "time": 1.6014} -{"mode": "train", "epoch": 1, "iter": 331, "lr": 0.00023, "memory": 49164, "data_time": 0.07922, "loss_cls_0": 0.72704, "loss_box_0": 1.61611, "loss_cns_0": 0.63282, "loss_yns_0": 0.13974, "loss_cls_1": 0.7862, "loss_box_1": 1.49512, "loss_cns_1": 0.66497, "loss_yns_1": 0.13914, "loss_cls_2": 0.78955, "loss_box_2": 1.46694, "loss_cns_2": 0.66362, "loss_yns_2": 0.13756, "loss_cls_3": 0.79817, "loss_box_3": 1.45198, "loss_cns_3": 0.66512, "loss_yns_3": 0.13786, "loss_cls_4": 0.79875, "loss_box_4": 1.43965, "loss_cns_4": 0.66239, "loss_yns_4": 0.13816, "loss_cls_5": 0.80816, "loss_box_5": 1.43973, "loss_cns_5": 0.66385, "loss_yns_5": 0.13869, "loss_cls_dn_0": 0.14528, "loss_box_dn_0": 0.7131, "loss_cls_dn_1": 0.09893, "loss_box_dn_1": 0.64497, "loss_cls_dn_2": 0.09793, "loss_box_dn_2": 0.63182, "loss_cls_dn_3": 0.09927, "loss_box_dn_3": 0.62926, "loss_cls_dn_4": 0.10028, "loss_box_dn_4": 0.62876, "loss_cls_dn_5": 0.10224, "loss_box_dn_5": 0.6297, "loss_dense_depth": 0.71058, "loss": 23.63341, "grad_norm": 27.12765, "time": 1.60387} -{"mode": "train", "epoch": 1, "iter": 332, "lr": 0.00023, "memory": 49164, "data_time": 0.07842, "loss_cls_0": 0.70971, "loss_box_0": 1.60083, "loss_cns_0": 0.64142, "loss_yns_0": 0.14291, "loss_cls_1": 0.7781, "loss_box_1": 1.45527, "loss_cns_1": 0.66729, "loss_yns_1": 0.14127, "loss_cls_2": 0.78868, "loss_box_2": 1.43272, "loss_cns_2": 0.66703, "loss_yns_2": 0.13996, "loss_cls_3": 0.7918, "loss_box_3": 1.43118, "loss_cns_3": 0.66667, "loss_yns_3": 0.13979, "loss_cls_4": 0.80459, "loss_box_4": 1.40788, "loss_cns_4": 0.66539, "loss_yns_4": 0.14052, "loss_cls_5": 0.81878, "loss_box_5": 1.41409, "loss_cns_5": 0.66864, "loss_yns_5": 0.14179, "loss_cls_dn_0": 0.14073, "loss_box_dn_0": 0.72348, "loss_cls_dn_1": 0.09959, "loss_box_dn_1": 0.64228, "loss_cls_dn_2": 0.09837, "loss_box_dn_2": 0.6297, "loss_cls_dn_3": 0.09937, "loss_box_dn_3": 0.62979, "loss_cls_dn_4": 0.09967, "loss_box_dn_4": 0.62805, "loss_cls_dn_5": 0.10231, "loss_box_dn_5": 0.62948, "loss_dense_depth": 0.66431, "loss": 23.44343, "grad_norm": 40.00101, "time": 1.57854} -{"mode": "train", "epoch": 1, "iter": 333, "lr": 0.00023, "memory": 49164, "data_time": 0.07734, "loss_cls_0": 0.73411, "loss_box_0": 1.61456, "loss_cns_0": 0.63861, "loss_yns_0": 0.14292, "loss_cls_1": 0.77725, "loss_box_1": 1.4825, "loss_cns_1": 0.66097, "loss_yns_1": 0.14235, "loss_cls_2": 0.79146, "loss_box_2": 1.454, "loss_cns_2": 0.65505, "loss_yns_2": 0.13897, "loss_cls_3": 0.80188, "loss_box_3": 1.43264, "loss_cns_3": 0.65295, "loss_yns_3": 0.13816, "loss_cls_4": 0.81146, "loss_box_4": 1.4289, "loss_cns_4": 0.65973, "loss_yns_4": 0.14124, "loss_cls_5": 0.81997, "loss_box_5": 1.43709, "loss_cns_5": 0.66754, "loss_yns_5": 0.1406, "loss_cls_dn_0": 0.14813, "loss_box_dn_0": 0.72574, "loss_cls_dn_1": 0.10011, "loss_box_dn_1": 0.65674, "loss_cls_dn_2": 0.10083, "loss_box_dn_2": 0.64469, "loss_cls_dn_3": 0.10155, "loss_box_dn_3": 0.64539, "loss_cls_dn_4": 0.10035, "loss_box_dn_4": 0.64383, "loss_cls_dn_5": 0.10202, "loss_box_dn_5": 0.65011, "loss_dense_depth": 0.72676, "loss": 23.71116, "grad_norm": 41.56788, "time": 1.58279} -{"mode": "train", "epoch": 1, "iter": 334, "lr": 0.00023, "memory": 49164, "data_time": 0.08007, "loss_cls_0": 0.7593, "loss_box_0": 1.63775, "loss_cns_0": 0.63585, "loss_yns_0": 0.14317, "loss_cls_1": 0.79933, "loss_box_1": 1.49256, "loss_cns_1": 0.65669, "loss_yns_1": 0.14036, "loss_cls_2": 0.8023, "loss_box_2": 1.46572, "loss_cns_2": 0.65419, "loss_yns_2": 0.14029, "loss_cls_3": 0.81109, "loss_box_3": 1.4649, "loss_cns_3": 0.65203, "loss_yns_3": 0.13777, "loss_cls_4": 0.81673, "loss_box_4": 1.46727, "loss_cns_4": 0.66068, "loss_yns_4": 0.14368, "loss_cls_5": 0.82412, "loss_box_5": 1.46056, "loss_cns_5": 0.66538, "loss_yns_5": 0.13951, "loss_cls_dn_0": 0.14635, "loss_box_dn_0": 0.72123, "loss_cls_dn_1": 0.10242, "loss_box_dn_1": 0.65669, "loss_cls_dn_2": 0.10275, "loss_box_dn_2": 0.64603, "loss_cls_dn_3": 0.10159, "loss_box_dn_3": 0.65067, "loss_cls_dn_4": 0.10381, "loss_box_dn_4": 0.65389, "loss_cls_dn_5": 0.10622, "loss_box_dn_5": 0.65875, "loss_dense_depth": 0.71923, "loss": 23.94086, "grad_norm": 34.99201, "time": 1.6226} -{"mode": "train", "epoch": 1, "iter": 335, "lr": 0.00023, "memory": 49164, "data_time": 0.07435, "loss_cls_0": 0.7394, "loss_box_0": 1.62873, "loss_cns_0": 0.63594, "loss_yns_0": 0.1439, "loss_cls_1": 0.80055, "loss_box_1": 1.49073, "loss_cns_1": 0.66025, "loss_yns_1": 0.13866, "loss_cls_2": 0.80588, "loss_box_2": 1.46865, "loss_cns_2": 0.66108, "loss_yns_2": 0.14044, "loss_cls_3": 0.80742, "loss_box_3": 1.46244, "loss_cns_3": 0.66016, "loss_yns_3": 0.13863, "loss_cls_4": 0.8108, "loss_box_4": 1.46176, "loss_cns_4": 0.66193, "loss_yns_4": 0.14126, "loss_cls_5": 0.82516, "loss_box_5": 1.45626, "loss_cns_5": 0.66415, "loss_yns_5": 0.13832, "loss_cls_dn_0": 0.14586, "loss_box_dn_0": 0.72206, "loss_cls_dn_1": 0.10151, "loss_box_dn_1": 0.65627, "loss_cls_dn_2": 0.10037, "loss_box_dn_2": 0.64946, "loss_cls_dn_3": 0.10126, "loss_box_dn_3": 0.64852, "loss_cls_dn_4": 0.10469, "loss_box_dn_4": 0.65325, "loss_cls_dn_5": 0.10684, "loss_box_dn_5": 0.65727, "loss_dense_depth": 0.70269, "loss": 23.89253, "grad_norm": 35.48561, "time": 1.57032} -{"mode": "train", "epoch": 1, "iter": 336, "lr": 0.00023, "memory": 49164, "data_time": 0.07103, "loss_cls_0": 0.72855, "loss_box_0": 1.63063, "loss_cns_0": 0.63131, "loss_yns_0": 0.14257, "loss_cls_1": 0.80515, "loss_box_1": 1.49862, "loss_cns_1": 0.66186, "loss_yns_1": 0.14091, "loss_cls_2": 0.8106, "loss_box_2": 1.47043, "loss_cns_2": 0.66124, "loss_yns_2": 0.14069, "loss_cls_3": 0.80313, "loss_box_3": 1.46562, "loss_cns_3": 0.66235, "loss_yns_3": 0.14068, "loss_cls_4": 0.81283, "loss_box_4": 1.46628, "loss_cns_4": 0.66358, "loss_yns_4": 0.14086, "loss_cls_5": 0.82785, "loss_box_5": 1.4791, "loss_cns_5": 0.6628, "loss_yns_5": 0.1399, "loss_cls_dn_0": 0.14029, "loss_box_dn_0": 0.71955, "loss_cls_dn_1": 0.09964, "loss_box_dn_1": 0.65427, "loss_cls_dn_2": 0.09966, "loss_box_dn_2": 0.64375, "loss_cls_dn_3": 0.1005, "loss_box_dn_3": 0.6438, "loss_cls_dn_4": 0.10078, "loss_box_dn_4": 0.64851, "loss_cls_dn_5": 0.10256, "loss_box_dn_5": 0.65462, "loss_dense_depth": 0.70545, "loss": 23.90091, "grad_norm": 36.04195, "time": 1.57322} -{"mode": "train", "epoch": 1, "iter": 337, "lr": 0.00023, "memory": 49164, "data_time": 0.07267, "loss_cls_0": 0.74219, "loss_box_0": 1.62181, "loss_cns_0": 0.63287, "loss_yns_0": 0.14102, "loss_cls_1": 0.81747, "loss_box_1": 1.50151, "loss_cns_1": 0.66569, "loss_yns_1": 0.13942, "loss_cls_2": 0.83923, "loss_box_2": 1.47627, "loss_cns_2": 0.6639, "loss_yns_2": 0.13702, "loss_cls_3": 0.83273, "loss_box_3": 1.48645, "loss_cns_3": 0.6643, "loss_yns_3": 0.13875, "loss_cls_4": 0.83567, "loss_box_4": 1.46529, "loss_cns_4": 0.66534, "loss_yns_4": 0.13944, "loss_cls_5": 0.82916, "loss_box_5": 1.45529, "loss_cns_5": 0.66627, "loss_yns_5": 0.14029, "loss_cls_dn_0": 0.1458, "loss_box_dn_0": 0.71424, "loss_cls_dn_1": 0.10461, "loss_box_dn_1": 0.63991, "loss_cls_dn_2": 0.10598, "loss_box_dn_2": 0.62852, "loss_cls_dn_3": 0.10662, "loss_box_dn_3": 0.63515, "loss_cls_dn_4": 0.10844, "loss_box_dn_4": 0.62876, "loss_cls_dn_5": 0.10712, "loss_box_dn_5": 0.62599, "loss_dense_depth": 0.69188, "loss": 23.94037, "grad_norm": 42.5338, "time": 1.56756} -{"mode": "train", "epoch": 1, "iter": 338, "lr": 0.00023, "memory": 49164, "data_time": 0.07343, "loss_cls_0": 0.73421, "loss_box_0": 1.59374, "loss_cns_0": 0.63443, "loss_yns_0": 0.14034, "loss_cls_1": 0.81869, "loss_box_1": 1.47177, "loss_cns_1": 0.66555, "loss_yns_1": 0.13651, "loss_cls_2": 0.8337, "loss_box_2": 1.4503, "loss_cns_2": 0.66239, "loss_yns_2": 0.13679, "loss_cls_3": 0.83358, "loss_box_3": 1.44455, "loss_cns_3": 0.66168, "loss_yns_3": 0.13732, "loss_cls_4": 0.83606, "loss_box_4": 1.45034, "loss_cns_4": 0.6664, "loss_yns_4": 0.13965, "loss_cls_5": 0.84211, "loss_box_5": 1.4384, "loss_cns_5": 0.66696, "loss_yns_5": 0.13889, "loss_cls_dn_0": 0.14417, "loss_box_dn_0": 0.72998, "loss_cls_dn_1": 0.10612, "loss_box_dn_1": 0.63075, "loss_cls_dn_2": 0.10639, "loss_box_dn_2": 0.61949, "loss_cls_dn_3": 0.10583, "loss_box_dn_3": 0.619, "loss_cls_dn_4": 0.10807, "loss_box_dn_4": 0.62143, "loss_cls_dn_5": 0.10977, "loss_box_dn_5": 0.6218, "loss_dense_depth": 0.69311, "loss": 23.75025, "grad_norm": 28.70694, "time": 1.57433} -{"mode": "train", "epoch": 1, "iter": 339, "lr": 0.00023, "memory": 49164, "data_time": 0.06838, "loss_cls_0": 0.73156, "loss_box_0": 1.62181, "loss_cns_0": 0.63351, "loss_yns_0": 0.1423, "loss_cls_1": 0.81432, "loss_box_1": 1.50433, "loss_cns_1": 0.658, "loss_yns_1": 0.13752, "loss_cls_2": 0.82674, "loss_box_2": 1.50087, "loss_cns_2": 0.66486, "loss_yns_2": 0.13763, "loss_cls_3": 0.82675, "loss_box_3": 1.51727, "loss_cns_3": 0.66392, "loss_yns_3": 0.13599, "loss_cls_4": 0.83541, "loss_box_4": 1.51103, "loss_cns_4": 0.65655, "loss_yns_4": 0.13668, "loss_cls_5": 0.84264, "loss_box_5": 1.48412, "loss_cns_5": 0.65864, "loss_yns_5": 0.13462, "loss_cls_dn_0": 0.14507, "loss_box_dn_0": 0.72673, "loss_cls_dn_1": 0.10317, "loss_box_dn_1": 0.65269, "loss_cls_dn_2": 0.1024, "loss_box_dn_2": 0.65147, "loss_cls_dn_3": 0.10091, "loss_box_dn_3": 0.65945, "loss_cls_dn_4": 0.10604, "loss_box_dn_4": 0.66602, "loss_cls_dn_5": 0.10539, "loss_box_dn_5": 0.66366, "loss_dense_depth": 0.75287, "loss": 24.21298, "grad_norm": 62.06498, "time": 1.56562} -{"mode": "train", "epoch": 1, "iter": 340, "lr": 0.00024, "memory": 49164, "data_time": 0.07202, "loss_cls_0": 0.7786, "loss_box_0": 1.62609, "loss_cns_0": 0.63453, "loss_yns_0": 0.14391, "loss_cls_1": 0.84275, "loss_box_1": 1.54502, "loss_cns_1": 0.66008, "loss_yns_1": 0.14254, "loss_cls_2": 0.85916, "loss_box_2": 1.49688, "loss_cns_2": 0.66366, "loss_yns_2": 0.14011, "loss_cls_3": 0.85749, "loss_box_3": 1.51702, "loss_cns_3": 0.66635, "loss_yns_3": 0.13925, "loss_cls_4": 0.86645, "loss_box_4": 1.49428, "loss_cns_4": 0.66069, "loss_yns_4": 0.14053, "loss_cls_5": 0.86323, "loss_box_5": 1.47716, "loss_cns_5": 0.65951, "loss_yns_5": 0.14215, "loss_cls_dn_0": 0.14451, "loss_box_dn_0": 0.72358, "loss_cls_dn_1": 0.10552, "loss_box_dn_1": 0.66549, "loss_cls_dn_2": 0.10532, "loss_box_dn_2": 0.65296, "loss_cls_dn_3": 0.10488, "loss_box_dn_3": 0.65942, "loss_cls_dn_4": 0.11096, "loss_box_dn_4": 0.66368, "loss_cls_dn_5": 0.10975, "loss_box_dn_5": 0.66763, "loss_dense_depth": 0.72133, "loss": 24.45248, "grad_norm": 42.88754, "time": 1.58824} -{"mode": "train", "epoch": 1, "iter": 341, "lr": 0.00024, "memory": 49164, "data_time": 0.10688, "loss_cls_0": 0.75937, "loss_box_0": 1.63441, "loss_cns_0": 0.62989, "loss_yns_0": 0.14147, "loss_cls_1": 0.81597, "loss_box_1": 1.5866, "loss_cns_1": 0.65308, "loss_yns_1": 0.13873, "loss_cls_2": 0.83187, "loss_box_2": 1.55816, "loss_cns_2": 0.64962, "loss_yns_2": 0.13758, "loss_cls_3": 0.83572, "loss_box_3": 1.55281, "loss_cns_3": 0.65634, "loss_yns_3": 0.13703, "loss_cls_4": 0.84693, "loss_box_4": 1.57309, "loss_cns_4": 0.65416, "loss_yns_4": 0.13605, "loss_cls_5": 0.85076, "loss_box_5": 1.60898, "loss_cns_5": 0.64957, "loss_yns_5": 0.13774, "loss_cls_dn_0": 0.14687, "loss_box_dn_0": 0.72741, "loss_cls_dn_1": 0.10435, "loss_box_dn_1": 0.66868, "loss_cls_dn_2": 0.10326, "loss_box_dn_2": 0.66657, "loss_cls_dn_3": 0.10343, "loss_box_dn_3": 0.65588, "loss_cls_dn_4": 0.10702, "loss_box_dn_4": 0.67173, "loss_cls_dn_5": 0.10672, "loss_box_dn_5": 0.69363, "loss_dense_depth": 0.80332, "loss": 24.73483, "grad_norm": 60.91932, "time": 1.66403} -{"mode": "train", "epoch": 1, "iter": 342, "lr": 0.00024, "memory": 49164, "data_time": 0.10026, "loss_cls_0": 0.75712, "loss_box_0": 1.59071, "loss_cns_0": 0.62839, "loss_yns_0": 0.1411, "loss_cls_1": 0.79782, "loss_box_1": 1.60631, "loss_cns_1": 0.64726, "loss_yns_1": 0.14121, "loss_cls_2": 0.8228, "loss_box_2": 1.59735, "loss_cns_2": 0.64494, "loss_yns_2": 0.14135, "loss_cls_3": 0.84042, "loss_box_3": 1.56949, "loss_cns_3": 0.65286, "loss_yns_3": 0.14211, "loss_cls_4": 0.83449, "loss_box_4": 1.5808, "loss_cns_4": 0.65043, "loss_yns_4": 0.13971, "loss_cls_5": 0.85143, "loss_box_5": 1.61155, "loss_cns_5": 0.64704, "loss_yns_5": 0.14041, "loss_cls_dn_0": 0.14061, "loss_box_dn_0": 0.72055, "loss_cls_dn_1": 0.09968, "loss_box_dn_1": 0.66238, "loss_cls_dn_2": 0.09905, "loss_box_dn_2": 0.66582, "loss_cls_dn_3": 0.1, "loss_box_dn_3": 0.64676, "loss_cls_dn_4": 0.10079, "loss_box_dn_4": 0.6547, "loss_cls_dn_5": 0.10229, "loss_box_dn_5": 0.67286, "loss_dense_depth": 0.71816, "loss": 24.56076, "grad_norm": 60.6281, "time": 1.61665} -{"mode": "train", "epoch": 1, "iter": 343, "lr": 0.00024, "memory": 49164, "data_time": 0.08437, "loss_cls_0": 0.772, "loss_box_0": 1.63004, "loss_cns_0": 0.62533, "loss_yns_0": 0.14194, "loss_cls_1": 0.81892, "loss_box_1": 1.60154, "loss_cns_1": 0.65043, "loss_yns_1": 0.14076, "loss_cls_2": 0.82415, "loss_box_2": 1.58775, "loss_cns_2": 0.65878, "loss_yns_2": 0.14246, "loss_cls_3": 0.83324, "loss_box_3": 1.56732, "loss_cns_3": 0.6562, "loss_yns_3": 0.1436, "loss_cls_4": 0.83391, "loss_box_4": 1.56112, "loss_cns_4": 0.65425, "loss_yns_4": 0.14114, "loss_cls_5": 0.84777, "loss_box_5": 1.55389, "loss_cns_5": 0.65515, "loss_yns_5": 0.14302, "loss_cls_dn_0": 0.13978, "loss_box_dn_0": 0.74543, "loss_cls_dn_1": 0.10197, "loss_box_dn_1": 0.65528, "loss_cls_dn_2": 0.10055, "loss_box_dn_2": 0.65021, "loss_cls_dn_3": 0.09955, "loss_box_dn_3": 0.6437, "loss_cls_dn_4": 0.10226, "loss_box_dn_4": 0.64483, "loss_cls_dn_5": 0.10319, "loss_box_dn_5": 0.64082, "loss_dense_depth": 0.7542, "loss": 24.56649, "grad_norm": 35.77671, "time": 1.58991} -{"mode": "train", "epoch": 1, "iter": 344, "lr": 0.00024, "memory": 49164, "data_time": 0.07563, "loss_cls_0": 0.77246, "loss_box_0": 1.62911, "loss_cns_0": 0.62152, "loss_yns_0": 0.13821, "loss_cls_1": 0.81963, "loss_box_1": 1.55359, "loss_cns_1": 0.65133, "loss_yns_1": 0.13625, "loss_cls_2": 0.83686, "loss_box_2": 1.5467, "loss_cns_2": 0.65828, "loss_yns_2": 0.13661, "loss_cls_3": 0.83665, "loss_box_3": 1.54709, "loss_cns_3": 0.65385, "loss_yns_3": 0.13606, "loss_cls_4": 0.84674, "loss_box_4": 1.53866, "loss_cns_4": 0.65311, "loss_yns_4": 0.13707, "loss_cls_5": 0.85263, "loss_box_5": 1.51741, "loss_cns_5": 0.65285, "loss_yns_5": 0.13593, "loss_cls_dn_0": 0.14004, "loss_box_dn_0": 0.74301, "loss_cls_dn_1": 0.10081, "loss_box_dn_1": 0.64815, "loss_cls_dn_2": 0.10054, "loss_box_dn_2": 0.64427, "loss_cls_dn_3": 0.09871, "loss_box_dn_3": 0.64932, "loss_cls_dn_4": 0.09895, "loss_box_dn_4": 0.64917, "loss_cls_dn_5": 0.0997, "loss_box_dn_5": 0.64406, "loss_dense_depth": 0.73873, "loss": 24.36407, "grad_norm": 46.80268, "time": 1.58718} -{"mode": "train", "epoch": 1, "iter": 345, "lr": 0.00024, "memory": 49164, "data_time": 0.08791, "loss_cls_0": 0.74434, "loss_box_0": 1.62731, "loss_cns_0": 0.62996, "loss_yns_0": 0.13927, "loss_cls_1": 0.78873, "loss_box_1": 1.54266, "loss_cns_1": 0.65257, "loss_yns_1": 0.13859, "loss_cls_2": 0.80825, "loss_box_2": 1.49497, "loss_cns_2": 0.65931, "loss_yns_2": 0.13762, "loss_cls_3": 0.81353, "loss_box_3": 1.48232, "loss_cns_3": 0.65717, "loss_yns_3": 0.13744, "loss_cls_4": 0.81936, "loss_box_4": 1.48068, "loss_cns_4": 0.65651, "loss_yns_4": 0.13824, "loss_cls_5": 0.82876, "loss_box_5": 1.47859, "loss_cns_5": 0.65575, "loss_yns_5": 0.13687, "loss_cls_dn_0": 0.14477, "loss_box_dn_0": 0.72007, "loss_cls_dn_1": 0.09902, "loss_box_dn_1": 0.63879, "loss_cls_dn_2": 0.09715, "loss_box_dn_2": 0.62239, "loss_cls_dn_3": 0.09793, "loss_box_dn_3": 0.62197, "loss_cls_dn_4": 0.09794, "loss_box_dn_4": 0.62486, "loss_cls_dn_5": 0.09995, "loss_box_dn_5": 0.62822, "loss_dense_depth": 0.72108, "loss": 23.86297, "grad_norm": 30.61167, "time": 1.59405} -{"mode": "train", "epoch": 1, "iter": 346, "lr": 0.00024, "memory": 49164, "data_time": 0.10513, "loss_cls_0": 0.74251, "loss_box_0": 1.65046, "loss_cns_0": 0.62953, "loss_yns_0": 0.13992, "loss_cls_1": 0.79941, "loss_box_1": 1.5288, "loss_cns_1": 0.65677, "loss_yns_1": 0.13884, "loss_cls_2": 0.81412, "loss_box_2": 1.47489, "loss_cns_2": 0.65947, "loss_yns_2": 0.13755, "loss_cls_3": 0.82517, "loss_box_3": 1.45512, "loss_cns_3": 0.65942, "loss_yns_3": 0.1372, "loss_cls_4": 0.82521, "loss_box_4": 1.44926, "loss_cns_4": 0.65977, "loss_yns_4": 0.1372, "loss_cls_5": 0.83479, "loss_box_5": 1.43513, "loss_cns_5": 0.65983, "loss_yns_5": 0.13656, "loss_cls_dn_0": 0.14644, "loss_box_dn_0": 0.73357, "loss_cls_dn_1": 0.10178, "loss_box_dn_1": 0.65902, "loss_cls_dn_2": 0.09941, "loss_box_dn_2": 0.6445, "loss_cls_dn_3": 0.10321, "loss_box_dn_3": 0.63748, "loss_cls_dn_4": 0.10297, "loss_box_dn_4": 0.64177, "loss_cls_dn_5": 0.10733, "loss_box_dn_5": 0.64171, "loss_dense_depth": 0.75193, "loss": 23.95805, "grad_norm": 40.30091, "time": 1.60365} -{"mode": "train", "epoch": 1, "iter": 347, "lr": 0.00024, "memory": 49164, "data_time": 0.07924, "loss_cls_0": 0.71616, "loss_box_0": 1.6037, "loss_cns_0": 0.63744, "loss_yns_0": 0.13984, "loss_cls_1": 0.77617, "loss_box_1": 1.47366, "loss_cns_1": 0.66024, "loss_yns_1": 0.13778, "loss_cls_2": 0.78106, "loss_box_2": 1.4471, "loss_cns_2": 0.66144, "loss_yns_2": 0.13714, "loss_cls_3": 0.79685, "loss_box_3": 1.42334, "loss_cns_3": 0.66153, "loss_yns_3": 0.13575, "loss_cls_4": 0.80333, "loss_box_4": 1.41517, "loss_cns_4": 0.66026, "loss_yns_4": 0.13557, "loss_cls_5": 0.80983, "loss_box_5": 1.42591, "loss_cns_5": 0.66028, "loss_yns_5": 0.13609, "loss_cls_dn_0": 0.13555, "loss_box_dn_0": 0.71238, "loss_cls_dn_1": 0.09759, "loss_box_dn_1": 0.65045, "loss_cls_dn_2": 0.0945, "loss_box_dn_2": 0.63843, "loss_cls_dn_3": 0.09662, "loss_box_dn_3": 0.63012, "loss_cls_dn_4": 0.09533, "loss_box_dn_4": 0.63107, "loss_cls_dn_5": 0.09826, "loss_box_dn_5": 0.63926, "loss_dense_depth": 0.7228, "loss": 23.47799, "grad_norm": 26.06146, "time": 1.58721} -{"mode": "train", "epoch": 1, "iter": 348, "lr": 0.00024, "memory": 49164, "data_time": 0.09011, "loss_cls_0": 0.73759, "loss_box_0": 1.62583, "loss_cns_0": 0.63269, "loss_yns_0": 0.13966, "loss_cls_1": 0.79724, "loss_box_1": 1.46338, "loss_cns_1": 0.66322, "loss_yns_1": 0.13658, "loss_cls_2": 0.80493, "loss_box_2": 1.41876, "loss_cns_2": 0.66291, "loss_yns_2": 0.13589, "loss_cls_3": 0.80938, "loss_box_3": 1.4143, "loss_cns_3": 0.66124, "loss_yns_3": 0.13531, "loss_cls_4": 0.81315, "loss_box_4": 1.40614, "loss_cns_4": 0.66033, "loss_yns_4": 0.1353, "loss_cls_5": 0.82049, "loss_box_5": 1.42475, "loss_cns_5": 0.65943, "loss_yns_5": 0.13606, "loss_cls_dn_0": 0.1373, "loss_box_dn_0": 0.7323, "loss_cls_dn_1": 0.09882, "loss_box_dn_1": 0.65452, "loss_cls_dn_2": 0.09737, "loss_box_dn_2": 0.63569, "loss_cls_dn_3": 0.09822, "loss_box_dn_3": 0.63643, "loss_cls_dn_4": 0.09881, "loss_box_dn_4": 0.63624, "loss_cls_dn_5": 0.09893, "loss_box_dn_5": 0.64399, "loss_dense_depth": 0.80659, "loss": 23.66974, "grad_norm": 43.87815, "time": 1.59683} -{"mode": "train", "epoch": 1, "iter": 349, "lr": 0.00024, "memory": 49164, "data_time": 0.08304, "loss_cls_0": 0.73669, "loss_box_0": 1.63082, "loss_cns_0": 0.63445, "loss_yns_0": 0.14233, "loss_cls_1": 0.79905, "loss_box_1": 1.45563, "loss_cns_1": 0.66201, "loss_yns_1": 0.13773, "loss_cls_2": 0.80854, "loss_box_2": 1.4021, "loss_cns_2": 0.66329, "loss_yns_2": 0.13679, "loss_cls_3": 0.80944, "loss_box_3": 1.39902, "loss_cns_3": 0.66126, "loss_yns_3": 0.13677, "loss_cls_4": 0.81671, "loss_box_4": 1.3919, "loss_cns_4": 0.6609, "loss_yns_4": 0.13652, "loss_cls_5": 0.81486, "loss_box_5": 1.39402, "loss_cns_5": 0.66292, "loss_yns_5": 0.13677, "loss_cls_dn_0": 0.14068, "loss_box_dn_0": 0.72455, "loss_cls_dn_1": 0.10147, "loss_box_dn_1": 0.65389, "loss_cls_dn_2": 0.10169, "loss_box_dn_2": 0.63055, "loss_cls_dn_3": 0.10326, "loss_box_dn_3": 0.62821, "loss_cls_dn_4": 0.10297, "loss_box_dn_4": 0.6265, "loss_cls_dn_5": 0.10258, "loss_box_dn_5": 0.6315, "loss_dense_depth": 0.78423, "loss": 23.56258, "grad_norm": 33.33389, "time": 1.57241} -{"mode": "train", "epoch": 1, "iter": 350, "lr": 0.00024, "memory": 49164, "data_time": 0.08339, "loss_cls_0": 0.70985, "loss_box_0": 1.59356, "loss_cns_0": 0.64035, "loss_yns_0": 0.14313, "loss_cls_1": 0.77737, "loss_box_1": 1.42106, "loss_cns_1": 0.6655, "loss_yns_1": 0.13858, "loss_cls_2": 0.78702, "loss_box_2": 1.39221, "loss_cns_2": 0.66665, "loss_yns_2": 0.13695, "loss_cls_3": 0.80007, "loss_box_3": 1.39588, "loss_cns_3": 0.6663, "loss_yns_3": 0.1361, "loss_cls_4": 0.80078, "loss_box_4": 1.40378, "loss_cns_4": 0.66614, "loss_yns_4": 0.13664, "loss_cls_5": 0.8052, "loss_box_5": 1.40272, "loss_cns_5": 0.66458, "loss_yns_5": 0.13616, "loss_cls_dn_0": 0.132, "loss_box_dn_0": 0.71113, "loss_cls_dn_1": 0.10003, "loss_box_dn_1": 0.62962, "loss_cls_dn_2": 0.09891, "loss_box_dn_2": 0.62262, "loss_cls_dn_3": 0.09942, "loss_box_dn_3": 0.62347, "loss_cls_dn_4": 0.0992, "loss_box_dn_4": 0.63215, "loss_cls_dn_5": 0.10004, "loss_box_dn_5": 0.63683, "loss_dense_depth": 0.79414, "loss": 23.36614, "grad_norm": 39.99905, "time": 1.59759} -{"mode": "train", "epoch": 1, "iter": 351, "lr": 0.00024, "memory": 49164, "data_time": 0.07892, "loss_cls_0": 0.73375, "loss_box_0": 1.61229, "loss_cns_0": 0.63515, "loss_yns_0": 0.14469, "loss_cls_1": 0.81706, "loss_box_1": 1.4163, "loss_cns_1": 0.66624, "loss_yns_1": 0.14341, "loss_cls_2": 0.82512, "loss_box_2": 1.39789, "loss_cns_2": 0.66605, "loss_yns_2": 0.14103, "loss_cls_3": 0.84787, "loss_box_3": 1.38129, "loss_cns_3": 0.66909, "loss_yns_3": 0.14089, "loss_cls_4": 0.83483, "loss_box_4": 1.38129, "loss_cns_4": 0.66616, "loss_yns_4": 0.14164, "loss_cls_5": 0.84984, "loss_box_5": 1.38102, "loss_cns_5": 0.66484, "loss_yns_5": 0.141, "loss_cls_dn_0": 0.13688, "loss_box_dn_0": 0.71712, "loss_cls_dn_1": 0.09841, "loss_box_dn_1": 0.6355, "loss_cls_dn_2": 0.09791, "loss_box_dn_2": 0.63368, "loss_cls_dn_3": 0.10019, "loss_box_dn_3": 0.62897, "loss_cls_dn_4": 0.10169, "loss_box_dn_4": 0.63559, "loss_cls_dn_5": 0.10258, "loss_box_dn_5": 0.63904, "loss_dense_depth": 0.73634, "loss": 23.56263, "grad_norm": 40.68996, "time": 1.60685} -{"mode": "train", "epoch": 1, "iter": 352, "lr": 0.00024, "memory": 49164, "data_time": 0.07468, "loss_cls_0": 0.7367, "loss_box_0": 1.59966, "loss_cns_0": 0.63247, "loss_yns_0": 0.14637, "loss_cls_1": 0.8067, "loss_box_1": 1.43325, "loss_cns_1": 0.6626, "loss_yns_1": 0.14411, "loss_cls_2": 0.81641, "loss_box_2": 1.42199, "loss_cns_2": 0.66402, "loss_yns_2": 0.14484, "loss_cls_3": 0.82979, "loss_box_3": 1.40079, "loss_cns_3": 0.6642, "loss_yns_3": 0.14369, "loss_cls_4": 0.82902, "loss_box_4": 1.39443, "loss_cns_4": 0.66357, "loss_yns_4": 0.14471, "loss_cls_5": 0.83832, "loss_box_5": 1.38679, "loss_cns_5": 0.6644, "loss_yns_5": 0.14527, "loss_cls_dn_0": 0.14006, "loss_box_dn_0": 0.72304, "loss_cls_dn_1": 0.09783, "loss_box_dn_1": 0.64533, "loss_cls_dn_2": 0.09841, "loss_box_dn_2": 0.64331, "loss_cls_dn_3": 0.10005, "loss_box_dn_3": 0.63634, "loss_cls_dn_4": 0.10055, "loss_box_dn_4": 0.63571, "loss_cls_dn_5": 0.09981, "loss_box_dn_5": 0.63693, "loss_dense_depth": 0.76435, "loss": 23.63582, "grad_norm": 34.06349, "time": 1.56957} -{"mode": "train", "epoch": 1, "iter": 353, "lr": 0.00024, "memory": 49164, "data_time": 0.0812, "loss_cls_0": 0.73334, "loss_box_0": 1.57448, "loss_cns_0": 0.63484, "loss_yns_0": 0.14578, "loss_cls_1": 0.80433, "loss_box_1": 1.38724, "loss_cns_1": 0.66228, "loss_yns_1": 0.14293, "loss_cls_2": 0.82835, "loss_box_2": 1.35517, "loss_cns_2": 0.66132, "loss_yns_2": 0.14327, "loss_cls_3": 0.82435, "loss_box_3": 1.37863, "loss_cns_3": 0.66209, "loss_yns_3": 0.14357, "loss_cls_4": 0.82863, "loss_box_4": 1.37044, "loss_cns_4": 0.66277, "loss_yns_4": 0.14429, "loss_cls_5": 0.82957, "loss_box_5": 1.36258, "loss_cns_5": 0.66219, "loss_yns_5": 0.14692, "loss_cls_dn_0": 0.13436, "loss_box_dn_0": 0.71507, "loss_cls_dn_1": 0.09763, "loss_box_dn_1": 0.63607, "loss_cls_dn_2": 0.09794, "loss_box_dn_2": 0.624, "loss_cls_dn_3": 0.09609, "loss_box_dn_3": 0.63325, "loss_cls_dn_4": 0.09777, "loss_box_dn_4": 0.63288, "loss_cls_dn_5": 0.09926, "loss_box_dn_5": 0.63033, "loss_dense_depth": 0.7352, "loss": 23.31921, "grad_norm": 39.50788, "time": 1.57981} -{"mode": "train", "epoch": 1, "iter": 354, "lr": 0.00024, "memory": 49164, "data_time": 0.07901, "loss_cls_0": 0.7316, "loss_box_0": 1.5777, "loss_cns_0": 0.63734, "loss_yns_0": 0.14478, "loss_cls_1": 0.80658, "loss_box_1": 1.4075, "loss_cns_1": 0.66147, "loss_yns_1": 0.14303, "loss_cls_2": 0.8304, "loss_box_2": 1.39402, "loss_cns_2": 0.66277, "loss_yns_2": 0.14336, "loss_cls_3": 0.84103, "loss_box_3": 1.39029, "loss_cns_3": 0.66311, "loss_yns_3": 0.14473, "loss_cls_4": 0.83338, "loss_box_4": 1.38067, "loss_cns_4": 0.6628, "loss_yns_4": 0.14476, "loss_cls_5": 0.8356, "loss_box_5": 1.37318, "loss_cns_5": 0.66169, "loss_yns_5": 0.14524, "loss_cls_dn_0": 0.13035, "loss_box_dn_0": 0.71046, "loss_cls_dn_1": 0.09584, "loss_box_dn_1": 0.63176, "loss_cls_dn_2": 0.09454, "loss_box_dn_2": 0.62182, "loss_cls_dn_3": 0.09301, "loss_box_dn_3": 0.62228, "loss_cls_dn_4": 0.09384, "loss_box_dn_4": 0.61863, "loss_cls_dn_5": 0.09664, "loss_box_dn_5": 0.61382, "loss_dense_depth": 0.74099, "loss": 23.381, "grad_norm": 32.94272, "time": 1.63026} -{"mode": "train", "epoch": 1, "iter": 355, "lr": 0.00024, "memory": 49164, "data_time": 0.08187, "loss_cls_0": 0.7269, "loss_box_0": 1.57506, "loss_cns_0": 0.63329, "loss_yns_0": 0.14414, "loss_cls_1": 0.79922, "loss_box_1": 1.42557, "loss_cns_1": 0.65853, "loss_yns_1": 0.14165, "loss_cls_2": 0.81162, "loss_box_2": 1.41547, "loss_cns_2": 0.65875, "loss_yns_2": 0.14249, "loss_cls_3": 0.82522, "loss_box_3": 1.40513, "loss_cns_3": 0.65682, "loss_yns_3": 0.14165, "loss_cls_4": 0.82291, "loss_box_4": 1.39832, "loss_cns_4": 0.65773, "loss_yns_4": 0.14201, "loss_cls_5": 0.83148, "loss_box_5": 1.40243, "loss_cns_5": 0.65742, "loss_yns_5": 0.14219, "loss_cls_dn_0": 0.13839, "loss_box_dn_0": 0.72481, "loss_cls_dn_1": 0.0982, "loss_box_dn_1": 0.62849, "loss_cls_dn_2": 0.09588, "loss_box_dn_2": 0.62012, "loss_cls_dn_3": 0.09667, "loss_box_dn_3": 0.62291, "loss_cls_dn_4": 0.09695, "loss_box_dn_4": 0.62396, "loss_cls_dn_5": 0.09721, "loss_box_dn_5": 0.62797, "loss_dense_depth": 0.70922, "loss": 23.39679, "grad_norm": 29.78074, "time": 1.58586} -{"mode": "train", "epoch": 1, "iter": 356, "lr": 0.00024, "memory": 49164, "data_time": 0.08123, "loss_cls_0": 0.75948, "loss_box_0": 1.56151, "loss_cns_0": 0.63308, "loss_yns_0": 0.1463, "loss_cls_1": 0.81898, "loss_box_1": 1.42901, "loss_cns_1": 0.65923, "loss_yns_1": 0.14471, "loss_cls_2": 0.83179, "loss_box_2": 1.4035, "loss_cns_2": 0.65992, "loss_yns_2": 0.14358, "loss_cls_3": 0.83705, "loss_box_3": 1.42091, "loss_cns_3": 0.6576, "loss_yns_3": 0.14369, "loss_cls_4": 0.84964, "loss_box_4": 1.39433, "loss_cns_4": 0.65624, "loss_yns_4": 0.14383, "loss_cls_5": 0.85055, "loss_box_5": 1.39598, "loss_cns_5": 0.6585, "loss_yns_5": 0.14682, "loss_cls_dn_0": 0.14082, "loss_box_dn_0": 0.72202, "loss_cls_dn_1": 0.0993, "loss_box_dn_1": 0.62224, "loss_cls_dn_2": 0.09983, "loss_box_dn_2": 0.61001, "loss_cls_dn_3": 0.1006, "loss_box_dn_3": 0.62227, "loss_cls_dn_4": 0.10164, "loss_box_dn_4": 0.62855, "loss_cls_dn_5": 0.10188, "loss_box_dn_5": 0.63718, "loss_dense_depth": 0.73206, "loss": 23.56463, "grad_norm": 36.55982, "time": 1.57962} -{"mode": "train", "epoch": 1, "iter": 357, "lr": 0.00024, "memory": 49164, "data_time": 0.07517, "loss_cls_0": 0.72337, "loss_box_0": 1.56054, "loss_cns_0": 0.63252, "loss_yns_0": 0.14244, "loss_cls_1": 0.78957, "loss_box_1": 1.47653, "loss_cns_1": 0.65733, "loss_yns_1": 0.14225, "loss_cls_2": 0.80648, "loss_box_2": 1.4353, "loss_cns_2": 0.65731, "loss_yns_2": 0.14176, "loss_cls_3": 0.80785, "loss_box_3": 1.43586, "loss_cns_3": 0.65803, "loss_yns_3": 0.14208, "loss_cls_4": 0.81876, "loss_box_4": 1.42784, "loss_cns_4": 0.66003, "loss_yns_4": 0.14135, "loss_cls_5": 0.82449, "loss_box_5": 1.43015, "loss_cns_5": 0.65909, "loss_yns_5": 0.14308, "loss_cls_dn_0": 0.13541, "loss_box_dn_0": 0.71015, "loss_cls_dn_1": 0.09568, "loss_box_dn_1": 0.65677, "loss_cls_dn_2": 0.09542, "loss_box_dn_2": 0.63796, "loss_cls_dn_3": 0.09623, "loss_box_dn_3": 0.64663, "loss_cls_dn_4": 0.09695, "loss_box_dn_4": 0.65591, "loss_cls_dn_5": 0.09898, "loss_box_dn_5": 0.66816, "loss_dense_depth": 0.71194, "loss": 23.62022, "grad_norm": 32.35017, "time": 1.5717} -{"mode": "train", "epoch": 1, "iter": 358, "lr": 0.00024, "memory": 49164, "data_time": 0.08085, "loss_cls_0": 0.72669, "loss_box_0": 1.54465, "loss_cns_0": 0.6359, "loss_yns_0": 0.14124, "loss_cls_1": 0.79059, "loss_box_1": 1.44131, "loss_cns_1": 0.6596, "loss_yns_1": 0.1414, "loss_cls_2": 0.79592, "loss_box_2": 1.40121, "loss_cns_2": 0.65985, "loss_yns_2": 0.14234, "loss_cls_3": 0.80231, "loss_box_3": 1.39258, "loss_cns_3": 0.66205, "loss_yns_3": 0.14321, "loss_cls_4": 0.80875, "loss_box_4": 1.39531, "loss_cns_4": 0.66345, "loss_yns_4": 0.14122, "loss_cls_5": 0.8253, "loss_box_5": 1.39436, "loss_cns_5": 0.66128, "loss_yns_5": 0.14243, "loss_cls_dn_0": 0.13829, "loss_box_dn_0": 0.71619, "loss_cls_dn_1": 0.10045, "loss_box_dn_1": 0.66409, "loss_cls_dn_2": 0.10107, "loss_box_dn_2": 0.64409, "loss_cls_dn_3": 0.10185, "loss_box_dn_3": 0.64399, "loss_cls_dn_4": 0.10206, "loss_box_dn_4": 0.64745, "loss_cls_dn_5": 0.10288, "loss_box_dn_5": 0.65297, "loss_dense_depth": 0.70069, "loss": 23.429, "grad_norm": 32.9683, "time": 1.5717} -{"mode": "train", "epoch": 1, "iter": 359, "lr": 0.00024, "memory": 49164, "data_time": 0.08113, "loss_cls_0": 0.71607, "loss_box_0": 1.51995, "loss_cns_0": 0.63954, "loss_yns_0": 0.13665, "loss_cls_1": 0.78025, "loss_box_1": 1.3944, "loss_cns_1": 0.66431, "loss_yns_1": 0.13904, "loss_cls_2": 0.78904, "loss_box_2": 1.38051, "loss_cns_2": 0.66345, "loss_yns_2": 0.13846, "loss_cls_3": 0.79619, "loss_box_3": 1.36407, "loss_cns_3": 0.6653, "loss_yns_3": 0.1388, "loss_cls_4": 0.79335, "loss_box_4": 1.35607, "loss_cns_4": 0.66381, "loss_yns_4": 0.13877, "loss_cls_5": 0.81094, "loss_box_5": 1.35281, "loss_cns_5": 0.66434, "loss_yns_5": 0.13852, "loss_cls_dn_0": 0.13991, "loss_box_dn_0": 0.70928, "loss_cls_dn_1": 0.103, "loss_box_dn_1": 0.63412, "loss_cls_dn_2": 0.10349, "loss_box_dn_2": 0.62414, "loss_cls_dn_3": 0.10312, "loss_box_dn_3": 0.62264, "loss_cls_dn_4": 0.10266, "loss_box_dn_4": 0.61643, "loss_cls_dn_5": 0.10239, "loss_box_dn_5": 0.61925, "loss_dense_depth": 0.68184, "loss": 23.00691, "grad_norm": 24.99608, "time": 1.58093} -{"mode": "train", "epoch": 1, "iter": 360, "lr": 0.00024, "memory": 49164, "data_time": 0.08122, "loss_cls_0": 0.72301, "loss_box_0": 1.53986, "loss_cns_0": 0.63663, "loss_yns_0": 0.1368, "loss_cls_1": 0.78545, "loss_box_1": 1.41199, "loss_cns_1": 0.66202, "loss_yns_1": 0.13803, "loss_cls_2": 0.79581, "loss_box_2": 1.40494, "loss_cns_2": 0.66211, "loss_yns_2": 0.13849, "loss_cls_3": 0.7957, "loss_box_3": 1.40102, "loss_cns_3": 0.66474, "loss_yns_3": 0.13613, "loss_cls_4": 0.79791, "loss_box_4": 1.4004, "loss_cns_4": 0.66408, "loss_yns_4": 0.13781, "loss_cls_5": 0.80925, "loss_box_5": 1.41151, "loss_cns_5": 0.66569, "loss_yns_5": 0.13985, "loss_cls_dn_0": 0.13541, "loss_box_dn_0": 0.71153, "loss_cls_dn_1": 0.09846, "loss_box_dn_1": 0.62011, "loss_cls_dn_2": 0.09755, "loss_box_dn_2": 0.61264, "loss_cls_dn_3": 0.09699, "loss_box_dn_3": 0.61089, "loss_cls_dn_4": 0.09919, "loss_box_dn_4": 0.61449, "loss_cls_dn_5": 0.09962, "loss_box_dn_5": 0.62033, "loss_dense_depth": 0.70953, "loss": 23.18597, "grad_norm": 33.57843, "time": 1.59097} -{"mode": "train", "epoch": 1, "iter": 361, "lr": 0.00024, "memory": 49164, "data_time": 0.10626, "loss_cls_0": 0.7424, "loss_box_0": 1.54528, "loss_cns_0": 0.63807, "loss_yns_0": 0.14303, "loss_cls_1": 0.79353, "loss_box_1": 1.42658, "loss_cns_1": 0.66392, "loss_yns_1": 0.14222, "loss_cls_2": 0.80269, "loss_box_2": 1.41694, "loss_cns_2": 0.66441, "loss_yns_2": 0.14157, "loss_cls_3": 0.80388, "loss_box_3": 1.40756, "loss_cns_3": 0.66487, "loss_yns_3": 0.14191, "loss_cls_4": 0.81785, "loss_box_4": 1.41106, "loss_cns_4": 0.66705, "loss_yns_4": 0.14123, "loss_cls_5": 0.82312, "loss_box_5": 1.40998, "loss_cns_5": 0.6652, "loss_yns_5": 0.1421, "loss_cls_dn_0": 0.13391, "loss_box_dn_0": 0.71437, "loss_cls_dn_1": 0.09782, "loss_box_dn_1": 0.62734, "loss_cls_dn_2": 0.09775, "loss_box_dn_2": 0.62364, "loss_cls_dn_3": 0.0973, "loss_box_dn_3": 0.62172, "loss_cls_dn_4": 0.10025, "loss_box_dn_4": 0.63118, "loss_cls_dn_5": 0.10055, "loss_box_dn_5": 0.63902, "loss_dense_depth": 0.69631, "loss": 23.39763, "grad_norm": 35.86131, "time": 1.6635} -{"mode": "train", "epoch": 1, "iter": 362, "lr": 0.00024, "memory": 49164, "data_time": 0.10497, "loss_cls_0": 0.74412, "loss_box_0": 1.5867, "loss_cns_0": 0.63649, "loss_yns_0": 0.14328, "loss_cls_1": 0.80248, "loss_box_1": 1.44056, "loss_cns_1": 0.66423, "loss_yns_1": 0.14052, "loss_cls_2": 0.81462, "loss_box_2": 1.41285, "loss_cns_2": 0.66333, "loss_yns_2": 0.14099, "loss_cls_3": 0.81732, "loss_box_3": 1.40849, "loss_cns_3": 0.66153, "loss_yns_3": 0.14144, "loss_cls_4": 0.81586, "loss_box_4": 1.41798, "loss_cns_4": 0.66344, "loss_yns_4": 0.14019, "loss_cls_5": 0.83098, "loss_box_5": 1.40652, "loss_cns_5": 0.6665, "loss_yns_5": 0.14126, "loss_cls_dn_0": 0.13705, "loss_box_dn_0": 0.71615, "loss_cls_dn_1": 0.09788, "loss_box_dn_1": 0.63591, "loss_cls_dn_2": 0.09681, "loss_box_dn_2": 0.62812, "loss_cls_dn_3": 0.09664, "loss_box_dn_3": 0.63336, "loss_cls_dn_4": 0.09722, "loss_box_dn_4": 0.63676, "loss_cls_dn_5": 0.09864, "loss_box_dn_5": 0.64291, "loss_dense_depth": 0.7518, "loss": 23.57094, "grad_norm": 30.07775, "time": 1.59626} -{"mode": "train", "epoch": 1, "iter": 363, "lr": 0.00024, "memory": 49164, "data_time": 0.08521, "loss_cls_0": 0.74042, "loss_box_0": 1.55846, "loss_cns_0": 0.6334, "loss_yns_0": 0.13574, "loss_cls_1": 0.80583, "loss_box_1": 1.41571, "loss_cns_1": 0.66513, "loss_yns_1": 0.13824, "loss_cls_2": 0.81235, "loss_box_2": 1.39492, "loss_cns_2": 0.6657, "loss_yns_2": 0.13655, "loss_cls_3": 0.81891, "loss_box_3": 1.38407, "loss_cns_3": 0.66543, "loss_yns_3": 0.1373, "loss_cls_4": 0.81912, "loss_box_4": 1.39603, "loss_cns_4": 0.66589, "loss_yns_4": 0.13724, "loss_cls_5": 0.82829, "loss_box_5": 1.38662, "loss_cns_5": 0.67129, "loss_yns_5": 0.13805, "loss_cls_dn_0": 0.13003, "loss_box_dn_0": 0.72561, "loss_cls_dn_1": 0.0926, "loss_box_dn_1": 0.64425, "loss_cls_dn_2": 0.09067, "loss_box_dn_2": 0.63159, "loss_cls_dn_3": 0.09105, "loss_box_dn_3": 0.63333, "loss_cls_dn_4": 0.09206, "loss_box_dn_4": 0.63376, "loss_cls_dn_5": 0.09342, "loss_box_dn_5": 0.63555, "loss_dense_depth": 0.7498, "loss": 23.39441, "grad_norm": 33.34536, "time": 1.59406} -{"mode": "train", "epoch": 1, "iter": 364, "lr": 0.00024, "memory": 49164, "data_time": 0.07556, "loss_cls_0": 0.73913, "loss_box_0": 1.55999, "loss_cns_0": 0.63397, "loss_yns_0": 0.13576, "loss_cls_1": 0.80281, "loss_box_1": 1.45995, "loss_cns_1": 0.66347, "loss_yns_1": 0.13933, "loss_cls_2": 0.80663, "loss_box_2": 1.43128, "loss_cns_2": 0.66393, "loss_yns_2": 0.13706, "loss_cls_3": 0.80683, "loss_box_3": 1.42749, "loss_cns_3": 0.66315, "loss_yns_3": 0.13702, "loss_cls_4": 0.81507, "loss_box_4": 1.4224, "loss_cns_4": 0.66253, "loss_yns_4": 0.13613, "loss_cls_5": 0.81684, "loss_box_5": 1.40873, "loss_cns_5": 0.6613, "loss_yns_5": 0.13591, "loss_cls_dn_0": 0.134, "loss_box_dn_0": 0.72666, "loss_cls_dn_1": 0.09466, "loss_box_dn_1": 0.63444, "loss_cls_dn_2": 0.09127, "loss_box_dn_2": 0.62049, "loss_cls_dn_3": 0.09072, "loss_box_dn_3": 0.61929, "loss_cls_dn_4": 0.09324, "loss_box_dn_4": 0.61885, "loss_cls_dn_5": 0.09343, "loss_box_dn_5": 0.61803, "loss_dense_depth": 0.7182, "loss": 23.41997, "grad_norm": 32.24253, "time": 1.5732} -{"mode": "train", "epoch": 1, "iter": 365, "lr": 0.00025, "memory": 49164, "data_time": 0.08515, "loss_cls_0": 0.72181, "loss_box_0": 1.57915, "loss_cns_0": 0.63779, "loss_yns_0": 0.13572, "loss_cls_1": 0.7937, "loss_box_1": 1.44148, "loss_cns_1": 0.6647, "loss_yns_1": 0.13686, "loss_cls_2": 0.79484, "loss_box_2": 1.40829, "loss_cns_2": 0.66504, "loss_yns_2": 0.13518, "loss_cls_3": 0.80203, "loss_box_3": 1.40037, "loss_cns_3": 0.66569, "loss_yns_3": 0.13601, "loss_cls_4": 0.80429, "loss_box_4": 1.39156, "loss_cns_4": 0.66527, "loss_yns_4": 0.13529, "loss_cls_5": 0.80905, "loss_box_5": 1.40382, "loss_cns_5": 0.66509, "loss_yns_5": 0.13591, "loss_cls_dn_0": 0.12999, "loss_box_dn_0": 0.72251, "loss_cls_dn_1": 0.09716, "loss_box_dn_1": 0.6196, "loss_cls_dn_2": 0.09583, "loss_box_dn_2": 0.60422, "loss_cls_dn_3": 0.09648, "loss_box_dn_3": 0.59922, "loss_cls_dn_4": 0.09655, "loss_box_dn_4": 0.59767, "loss_cls_dn_5": 0.09627, "loss_box_dn_5": 0.60155, "loss_dense_depth": 0.70577, "loss": 23.19175, "grad_norm": 26.08198, "time": 1.57463} -{"mode": "train", "epoch": 1, "iter": 366, "lr": 0.00025, "memory": 49164, "data_time": 0.10799, "loss_cls_0": 0.69561, "loss_box_0": 1.56988, "loss_cns_0": 0.63799, "loss_yns_0": 0.13564, "loss_cls_1": 0.77409, "loss_box_1": 1.43848, "loss_cns_1": 0.66213, "loss_yns_1": 0.13421, "loss_cls_2": 0.78591, "loss_box_2": 1.41532, "loss_cns_2": 0.66349, "loss_yns_2": 0.13707, "loss_cls_3": 0.79311, "loss_box_3": 1.41897, "loss_cns_3": 0.66269, "loss_yns_3": 0.13533, "loss_cls_4": 0.79768, "loss_box_4": 1.40667, "loss_cns_4": 0.66513, "loss_yns_4": 0.13517, "loss_cls_5": 0.78766, "loss_box_5": 1.40112, "loss_cns_5": 0.66345, "loss_yns_5": 0.13575, "loss_cls_dn_0": 0.12363, "loss_box_dn_0": 0.71339, "loss_cls_dn_1": 0.09253, "loss_box_dn_1": 0.61516, "loss_cls_dn_2": 0.09351, "loss_box_dn_2": 0.60727, "loss_cls_dn_3": 0.0959, "loss_box_dn_3": 0.61191, "loss_cls_dn_4": 0.09566, "loss_box_dn_4": 0.60875, "loss_cls_dn_5": 0.09406, "loss_box_dn_5": 0.60813, "loss_dense_depth": 0.70978, "loss": 23.12222, "grad_norm": 35.70926, "time": 1.6019} -{"mode": "train", "epoch": 1, "iter": 367, "lr": 0.00025, "memory": 49164, "data_time": 0.07876, "loss_cls_0": 0.71183, "loss_box_0": 1.5931, "loss_cns_0": 0.64021, "loss_yns_0": 0.13609, "loss_cls_1": 0.77921, "loss_box_1": 1.44083, "loss_cns_1": 0.66627, "loss_yns_1": 0.13479, "loss_cls_2": 0.80836, "loss_box_2": 1.40139, "loss_cns_2": 0.66478, "loss_yns_2": 0.13481, "loss_cls_3": 0.80225, "loss_box_3": 1.40102, "loss_cns_3": 0.66424, "loss_yns_3": 0.13494, "loss_cls_4": 0.80497, "loss_box_4": 1.39365, "loss_cns_4": 0.66711, "loss_yns_4": 0.13371, "loss_cls_5": 0.80331, "loss_box_5": 1.39616, "loss_cns_5": 0.66637, "loss_yns_5": 0.1336, "loss_cls_dn_0": 0.12747, "loss_box_dn_0": 0.71745, "loss_cls_dn_1": 0.09549, "loss_box_dn_1": 0.63413, "loss_cls_dn_2": 0.09734, "loss_box_dn_2": 0.62119, "loss_cls_dn_3": 0.09944, "loss_box_dn_3": 0.62619, "loss_cls_dn_4": 0.09952, "loss_box_dn_4": 0.625, "loss_cls_dn_5": 0.09797, "loss_box_dn_5": 0.6289, "loss_dense_depth": 0.7287, "loss": 23.31177, "grad_norm": 35.95616, "time": 1.57875} -{"mode": "train", "epoch": 1, "iter": 368, "lr": 0.00025, "memory": 49164, "data_time": 0.08119, "loss_cls_0": 0.72339, "loss_box_0": 1.60596, "loss_cns_0": 0.63824, "loss_yns_0": 0.13611, "loss_cls_1": 0.77292, "loss_box_1": 1.43944, "loss_cns_1": 0.66514, "loss_yns_1": 0.13459, "loss_cls_2": 0.79157, "loss_box_2": 1.40017, "loss_cns_2": 0.66638, "loss_yns_2": 0.13471, "loss_cls_3": 0.78967, "loss_box_3": 1.37762, "loss_cns_3": 0.66536, "loss_yns_3": 0.13355, "loss_cls_4": 0.80204, "loss_box_4": 1.37676, "loss_cns_4": 0.6673, "loss_yns_4": 0.13301, "loss_cls_5": 0.79782, "loss_box_5": 1.3843, "loss_cns_5": 0.66656, "loss_yns_5": 0.13116, "loss_cls_dn_0": 0.1279, "loss_box_dn_0": 0.71959, "loss_cls_dn_1": 0.09632, "loss_box_dn_1": 0.64382, "loss_cls_dn_2": 0.09455, "loss_box_dn_2": 0.62689, "loss_cls_dn_3": 0.09594, "loss_box_dn_3": 0.62434, "loss_cls_dn_4": 0.09547, "loss_box_dn_4": 0.62547, "loss_cls_dn_5": 0.09615, "loss_box_dn_5": 0.63234, "loss_dense_depth": 0.70892, "loss": 23.22151, "grad_norm": 27.77613, "time": 1.5826} -{"mode": "train", "epoch": 1, "iter": 369, "lr": 0.00025, "memory": 49164, "data_time": 0.08425, "loss_cls_0": 0.68707, "loss_box_0": 1.57813, "loss_cns_0": 0.63947, "loss_yns_0": 0.13547, "loss_cls_1": 0.75051, "loss_box_1": 1.43448, "loss_cns_1": 0.66866, "loss_yns_1": 0.13357, "loss_cls_2": 0.76794, "loss_box_2": 1.37636, "loss_cns_2": 0.66867, "loss_yns_2": 0.13109, "loss_cls_3": 0.77918, "loss_box_3": 1.3597, "loss_cns_3": 0.66804, "loss_yns_3": 0.13058, "loss_cls_4": 0.79691, "loss_box_4": 1.35666, "loss_cns_4": 0.66905, "loss_yns_4": 0.13119, "loss_cls_5": 0.7879, "loss_box_5": 1.35748, "loss_cns_5": 0.66856, "loss_yns_5": 0.13175, "loss_cls_dn_0": 0.11918, "loss_box_dn_0": 0.71877, "loss_cls_dn_1": 0.09458, "loss_box_dn_1": 0.65157, "loss_cls_dn_2": 0.09486, "loss_box_dn_2": 0.62432, "loss_cls_dn_3": 0.0973, "loss_box_dn_3": 0.62031, "loss_cls_dn_4": 0.09685, "loss_box_dn_4": 0.61855, "loss_cls_dn_5": 0.09906, "loss_box_dn_5": 0.62449, "loss_dense_depth": 0.68972, "loss": 22.95796, "grad_norm": 32.93731, "time": 1.5912} -{"mode": "train", "epoch": 1, "iter": 370, "lr": 0.00025, "memory": 49164, "data_time": 0.08401, "loss_cls_0": 0.71458, "loss_box_0": 1.55992, "loss_cns_0": 0.63243, "loss_yns_0": 0.13376, "loss_cls_1": 0.77144, "loss_box_1": 1.4323, "loss_cns_1": 0.66359, "loss_yns_1": 0.13065, "loss_cls_2": 0.7883, "loss_box_2": 1.39104, "loss_cns_2": 0.66183, "loss_yns_2": 0.1307, "loss_cls_3": 0.80471, "loss_box_3": 1.38396, "loss_cns_3": 0.66572, "loss_yns_3": 0.13015, "loss_cls_4": 0.80503, "loss_box_4": 1.38187, "loss_cns_4": 0.66689, "loss_yns_4": 0.13082, "loss_cls_5": 0.8246, "loss_box_5": 1.36919, "loss_cns_5": 0.66642, "loss_yns_5": 0.13114, "loss_cls_dn_0": 0.12521, "loss_box_dn_0": 0.71598, "loss_cls_dn_1": 0.09744, "loss_box_dn_1": 0.63481, "loss_cls_dn_2": 0.10089, "loss_box_dn_2": 0.61506, "loss_cls_dn_3": 0.1064, "loss_box_dn_3": 0.61345, "loss_cls_dn_4": 0.10196, "loss_box_dn_4": 0.6127, "loss_cls_dn_5": 0.10388, "loss_box_dn_5": 0.61148, "loss_dense_depth": 0.71382, "loss": 23.12412, "grad_norm": 37.40245, "time": 1.58119} -{"mode": "train", "epoch": 1, "iter": 371, "lr": 0.00025, "memory": 49164, "data_time": 0.0773, "loss_cls_0": 0.70007, "loss_box_0": 1.53392, "loss_cns_0": 0.63477, "loss_yns_0": 0.13443, "loss_cls_1": 0.75657, "loss_box_1": 1.41925, "loss_cns_1": 0.65636, "loss_yns_1": 0.13312, "loss_cls_2": 0.77483, "loss_box_2": 1.38472, "loss_cns_2": 0.6561, "loss_yns_2": 0.13365, "loss_cls_3": 0.78408, "loss_box_3": 1.39789, "loss_cns_3": 0.66198, "loss_yns_3": 0.13369, "loss_cls_4": 0.78237, "loss_box_4": 1.39729, "loss_cns_4": 0.66212, "loss_yns_4": 0.13312, "loss_cls_5": 0.79046, "loss_box_5": 1.38978, "loss_cns_5": 0.66375, "loss_yns_5": 0.13367, "loss_cls_dn_0": 0.1185, "loss_box_dn_0": 0.71632, "loss_cls_dn_1": 0.09313, "loss_box_dn_1": 0.61352, "loss_cls_dn_2": 0.09364, "loss_box_dn_2": 0.60193, "loss_cls_dn_3": 0.0979, "loss_box_dn_3": 0.60382, "loss_cls_dn_4": 0.095, "loss_box_dn_4": 0.60448, "loss_cls_dn_5": 0.09611, "loss_box_dn_5": 0.60712, "loss_dense_depth": 0.68077, "loss": 22.87023, "grad_norm": 31.35162, "time": 1.61846} -{"mode": "train", "epoch": 1, "iter": 372, "lr": 0.00025, "memory": 49164, "data_time": 0.07721, "loss_cls_0": 0.72336, "loss_box_0": 1.57585, "loss_cns_0": 0.63673, "loss_yns_0": 0.13701, "loss_cls_1": 0.76789, "loss_box_1": 1.4344, "loss_cns_1": 0.65898, "loss_yns_1": 0.13661, "loss_cls_2": 0.78221, "loss_box_2": 1.41982, "loss_cns_2": 0.66098, "loss_yns_2": 0.13775, "loss_cls_3": 0.79559, "loss_box_3": 1.42116, "loss_cns_3": 0.66101, "loss_yns_3": 0.13683, "loss_cls_4": 0.78904, "loss_box_4": 1.43221, "loss_cns_4": 0.66035, "loss_yns_4": 0.13647, "loss_cls_5": 0.79555, "loss_box_5": 1.4305, "loss_cns_5": 0.66277, "loss_yns_5": 0.13903, "loss_cls_dn_0": 0.12337, "loss_box_dn_0": 0.72797, "loss_cls_dn_1": 0.09266, "loss_box_dn_1": 0.62223, "loss_cls_dn_2": 0.09173, "loss_box_dn_2": 0.61649, "loss_cls_dn_3": 0.09372, "loss_box_dn_3": 0.61966, "loss_cls_dn_4": 0.09329, "loss_box_dn_4": 0.62415, "loss_cls_dn_5": 0.09438, "loss_box_dn_5": 0.63093, "loss_dense_depth": 0.72773, "loss": 23.29042, "grad_norm": 43.64226, "time": 1.56998} -{"mode": "train", "epoch": 1, "iter": 373, "lr": 0.00025, "memory": 49164, "data_time": 0.07587, "loss_cls_0": 0.75532, "loss_box_0": 1.61912, "loss_cns_0": 0.63909, "loss_yns_0": 0.13978, "loss_cls_1": 0.80224, "loss_box_1": 1.47453, "loss_cns_1": 0.66407, "loss_yns_1": 0.13909, "loss_cls_2": 0.8184, "loss_box_2": 1.43927, "loss_cns_2": 0.66762, "loss_yns_2": 0.13983, "loss_cls_3": 0.83394, "loss_box_3": 1.4205, "loss_cns_3": 0.66678, "loss_yns_3": 0.13783, "loss_cls_4": 0.82659, "loss_box_4": 1.42417, "loss_cns_4": 0.67006, "loss_yns_4": 0.13834, "loss_cls_5": 0.82969, "loss_box_5": 1.42266, "loss_cns_5": 0.66759, "loss_yns_5": 0.14258, "loss_cls_dn_0": 0.12363, "loss_box_dn_0": 0.71192, "loss_cls_dn_1": 0.09434, "loss_box_dn_1": 0.63752, "loss_cls_dn_2": 0.09247, "loss_box_dn_2": 0.62173, "loss_cls_dn_3": 0.09541, "loss_box_dn_3": 0.61605, "loss_cls_dn_4": 0.09357, "loss_box_dn_4": 0.61763, "loss_cls_dn_5": 0.09388, "loss_box_dn_5": 0.62474, "loss_dense_depth": 0.70234, "loss": 23.60431, "grad_norm": 27.12244, "time": 1.59107} -{"mode": "train", "epoch": 1, "iter": 374, "lr": 0.00025, "memory": 49164, "data_time": 0.06672, "loss_cls_0": 0.72784, "loss_box_0": 1.5925, "loss_cns_0": 0.63362, "loss_yns_0": 0.1384, "loss_cls_1": 0.78472, "loss_box_1": 1.47361, "loss_cns_1": 0.661, "loss_yns_1": 0.13868, "loss_cls_2": 0.80575, "loss_box_2": 1.40631, "loss_cns_2": 0.66454, "loss_yns_2": 0.13706, "loss_cls_3": 0.81674, "loss_box_3": 1.40932, "loss_cns_3": 0.66424, "loss_yns_3": 0.13699, "loss_cls_4": 0.80829, "loss_box_4": 1.39822, "loss_cns_4": 0.66209, "loss_yns_4": 0.13657, "loss_cls_5": 0.81374, "loss_box_5": 1.39153, "loss_cns_5": 0.66162, "loss_yns_5": 0.13874, "loss_cls_dn_0": 0.12253, "loss_box_dn_0": 0.71624, "loss_cls_dn_1": 0.09324, "loss_box_dn_1": 0.6328, "loss_cls_dn_2": 0.09114, "loss_box_dn_2": 0.60667, "loss_cls_dn_3": 0.09375, "loss_box_dn_3": 0.6111, "loss_cls_dn_4": 0.09046, "loss_box_dn_4": 0.60574, "loss_cls_dn_5": 0.09476, "loss_box_dn_5": 0.60771, "loss_dense_depth": 0.71569, "loss": 23.28395, "grad_norm": 34.89682, "time": 1.61175} -{"mode": "train", "epoch": 1, "iter": 375, "lr": 0.00025, "memory": 49164, "data_time": 0.07113, "loss_cls_0": 0.73335, "loss_box_0": 1.56558, "loss_cns_0": 0.6316, "loss_yns_0": 0.13767, "loss_cls_1": 0.79856, "loss_box_1": 1.43287, "loss_cns_1": 0.65956, "loss_yns_1": 0.13657, "loss_cls_2": 0.81091, "loss_box_2": 1.40556, "loss_cns_2": 0.66189, "loss_yns_2": 0.13804, "loss_cls_3": 0.8208, "loss_box_3": 1.41882, "loss_cns_3": 0.66327, "loss_yns_3": 0.13806, "loss_cls_4": 0.81976, "loss_box_4": 1.38592, "loss_cns_4": 0.66298, "loss_yns_4": 0.13444, "loss_cls_5": 0.81931, "loss_box_5": 1.38239, "loss_cns_5": 0.6598, "loss_yns_5": 0.13611, "loss_cls_dn_0": 0.12192, "loss_box_dn_0": 0.71285, "loss_cls_dn_1": 0.0933, "loss_box_dn_1": 0.61342, "loss_cls_dn_2": 0.09325, "loss_box_dn_2": 0.60604, "loss_cls_dn_3": 0.09462, "loss_box_dn_3": 0.61388, "loss_cls_dn_4": 0.09478, "loss_box_dn_4": 0.60538, "loss_cls_dn_5": 0.09409, "loss_box_dn_5": 0.60567, "loss_dense_depth": 0.71347, "loss": 23.21648, "grad_norm": 41.75422, "time": 1.574} -{"mode": "train", "epoch": 1, "iter": 376, "lr": 0.00025, "memory": 49164, "data_time": 0.07285, "loss_cls_0": 0.76129, "loss_box_0": 1.58398, "loss_cns_0": 0.63327, "loss_yns_0": 0.14066, "loss_cls_1": 0.82409, "loss_box_1": 1.45314, "loss_cns_1": 0.66109, "loss_yns_1": 0.13767, "loss_cls_2": 0.8409, "loss_box_2": 1.42165, "loss_cns_2": 0.66123, "loss_yns_2": 0.13743, "loss_cls_3": 0.83443, "loss_box_3": 1.42333, "loss_cns_3": 0.66101, "loss_yns_3": 0.13793, "loss_cls_4": 0.83337, "loss_box_4": 1.42152, "loss_cns_4": 0.66373, "loss_yns_4": 0.1375, "loss_cls_5": 0.83625, "loss_box_5": 1.43774, "loss_cns_5": 0.66204, "loss_yns_5": 0.13886, "loss_cls_dn_0": 0.12532, "loss_box_dn_0": 0.70243, "loss_cls_dn_1": 0.09544, "loss_box_dn_1": 0.64604, "loss_cls_dn_2": 0.09358, "loss_box_dn_2": 0.63883, "loss_cls_dn_3": 0.09316, "loss_box_dn_3": 0.64452, "loss_cls_dn_4": 0.09387, "loss_box_dn_4": 0.64696, "loss_cls_dn_5": 0.09284, "loss_box_dn_5": 0.6567, "loss_dense_depth": 0.73682, "loss": 23.71062, "grad_norm": 32.16979, "time": 1.57277} -{"mode": "train", "epoch": 1, "iter": 377, "lr": 0.00025, "memory": 49164, "data_time": 0.0711, "loss_cls_0": 0.73977, "loss_box_0": 1.55194, "loss_cns_0": 0.63753, "loss_yns_0": 0.14201, "loss_cls_1": 0.81142, "loss_box_1": 1.42546, "loss_cns_1": 0.66374, "loss_yns_1": 0.13982, "loss_cls_2": 0.83463, "loss_box_2": 1.38939, "loss_cns_2": 0.66128, "loss_yns_2": 0.13833, "loss_cls_3": 0.82486, "loss_box_3": 1.39764, "loss_cns_3": 0.66071, "loss_yns_3": 0.13729, "loss_cls_4": 0.82561, "loss_box_4": 1.40291, "loss_cns_4": 0.66442, "loss_yns_4": 0.13787, "loss_cls_5": 0.82563, "loss_box_5": 1.39386, "loss_cns_5": 0.66622, "loss_yns_5": 0.13845, "loss_cls_dn_0": 0.11798, "loss_box_dn_0": 0.71167, "loss_cls_dn_1": 0.09124, "loss_box_dn_1": 0.65481, "loss_cls_dn_2": 0.09121, "loss_box_dn_2": 0.64193, "loss_cls_dn_3": 0.09272, "loss_box_dn_3": 0.64961, "loss_cls_dn_4": 0.09222, "loss_box_dn_4": 0.65373, "loss_cls_dn_5": 0.09332, "loss_box_dn_5": 0.65276, "loss_dense_depth": 0.73849, "loss": 23.49249, "grad_norm": 36.93893, "time": 1.57291} -{"mode": "train", "epoch": 1, "iter": 378, "lr": 0.00025, "memory": 49164, "data_time": 0.07097, "loss_cls_0": 0.73495, "loss_box_0": 1.55469, "loss_cns_0": 0.63872, "loss_yns_0": 0.1454, "loss_cls_1": 0.80108, "loss_box_1": 1.45326, "loss_cns_1": 0.6638, "loss_yns_1": 0.14271, "loss_cls_2": 0.81513, "loss_box_2": 1.40433, "loss_cns_2": 0.6612, "loss_yns_2": 0.14277, "loss_cls_3": 0.81178, "loss_box_3": 1.38678, "loss_cns_3": 0.65905, "loss_yns_3": 0.14279, "loss_cls_4": 0.81493, "loss_box_4": 1.39162, "loss_cns_4": 0.66298, "loss_yns_4": 0.13954, "loss_cls_5": 0.81616, "loss_box_5": 1.3871, "loss_cns_5": 0.6621, "loss_yns_5": 0.13977, "loss_cls_dn_0": 0.12741, "loss_box_dn_0": 0.71116, "loss_cls_dn_1": 0.09245, "loss_box_dn_1": 0.65123, "loss_cls_dn_2": 0.09279, "loss_box_dn_2": 0.62834, "loss_cls_dn_3": 0.0954, "loss_box_dn_3": 0.62726, "loss_cls_dn_4": 0.09295, "loss_box_dn_4": 0.62294, "loss_cls_dn_5": 0.09613, "loss_box_dn_5": 0.6246, "loss_dense_depth": 0.71348, "loss": 23.34876, "grad_norm": 33.2095, "time": 1.56273} -{"mode": "train", "epoch": 1, "iter": 379, "lr": 0.00025, "memory": 49164, "data_time": 0.07638, "loss_cls_0": 0.72967, "loss_box_0": 1.57222, "loss_cns_0": 0.63709, "loss_yns_0": 0.1426, "loss_cls_1": 0.79912, "loss_box_1": 1.46, "loss_cns_1": 0.66332, "loss_yns_1": 0.14065, "loss_cls_2": 0.8136, "loss_box_2": 1.41745, "loss_cns_2": 0.66147, "loss_yns_2": 0.13936, "loss_cls_3": 0.82222, "loss_box_3": 1.37846, "loss_cns_3": 0.65776, "loss_yns_3": 0.13816, "loss_cls_4": 0.82012, "loss_box_4": 1.38495, "loss_cns_4": 0.66151, "loss_yns_4": 0.13943, "loss_cls_5": 0.82446, "loss_box_5": 1.39438, "loss_cns_5": 0.66243, "loss_yns_5": 0.13859, "loss_cls_dn_0": 0.12389, "loss_box_dn_0": 0.70404, "loss_cls_dn_1": 0.09301, "loss_box_dn_1": 0.63406, "loss_cls_dn_2": 0.09178, "loss_box_dn_2": 0.61967, "loss_cls_dn_3": 0.09625, "loss_box_dn_3": 0.61378, "loss_cls_dn_4": 0.09371, "loss_box_dn_4": 0.61385, "loss_cls_dn_5": 0.09483, "loss_box_dn_5": 0.62039, "loss_dense_depth": 0.70956, "loss": 23.30783, "grad_norm": 32.12333, "time": 1.59051} -{"mode": "train", "epoch": 1, "iter": 380, "lr": 0.00025, "memory": 49164, "data_time": 0.07893, "loss_cls_0": 0.73871, "loss_box_0": 1.52843, "loss_cns_0": 0.64123, "loss_yns_0": 0.14092, "loss_cls_1": 0.78893, "loss_box_1": 1.4147, "loss_cns_1": 0.66048, "loss_yns_1": 0.13692, "loss_cls_2": 0.798, "loss_box_2": 1.38404, "loss_cns_2": 0.65924, "loss_yns_2": 0.13758, "loss_cls_3": 0.81059, "loss_box_3": 1.37361, "loss_cns_3": 0.65708, "loss_yns_3": 0.13671, "loss_cls_4": 0.81428, "loss_box_4": 1.35218, "loss_cns_4": 0.65862, "loss_yns_4": 0.13693, "loss_cls_5": 0.81661, "loss_box_5": 1.34983, "loss_cns_5": 0.66046, "loss_yns_5": 0.13775, "loss_cls_dn_0": 0.12066, "loss_box_dn_0": 0.71847, "loss_cls_dn_1": 0.09318, "loss_box_dn_1": 0.6403, "loss_cls_dn_2": 0.09008, "loss_box_dn_2": 0.63179, "loss_cls_dn_3": 0.09276, "loss_box_dn_3": 0.63615, "loss_cls_dn_4": 0.09288, "loss_box_dn_4": 0.62946, "loss_cls_dn_5": 0.0932, "loss_box_dn_5": 0.62801, "loss_dense_depth": 0.7223, "loss": 23.12305, "grad_norm": 40.94841, "time": 1.58358} -{"mode": "train", "epoch": 1, "iter": 381, "lr": 0.00025, "memory": 49164, "data_time": 0.10494, "loss_cls_0": 0.7219, "loss_box_0": 1.5056, "loss_cns_0": 0.63846, "loss_yns_0": 0.13597, "loss_cls_1": 0.77945, "loss_box_1": 1.39555, "loss_cns_1": 0.65771, "loss_yns_1": 0.13357, "loss_cls_2": 0.785, "loss_box_2": 1.37231, "loss_cns_2": 0.65896, "loss_yns_2": 0.13569, "loss_cls_3": 0.78882, "loss_box_3": 1.35081, "loss_cns_3": 0.65648, "loss_yns_3": 0.13511, "loss_cls_4": 0.79629, "loss_box_4": 1.32584, "loss_cns_4": 0.65694, "loss_yns_4": 0.13193, "loss_cls_5": 0.80801, "loss_box_5": 1.32203, "loss_cns_5": 0.65456, "loss_yns_5": 0.13203, "loss_cls_dn_0": 0.12112, "loss_box_dn_0": 0.71096, "loss_cls_dn_1": 0.09358, "loss_box_dn_1": 0.63859, "loss_cls_dn_2": 0.09163, "loss_box_dn_2": 0.6215, "loss_cls_dn_3": 0.0916, "loss_box_dn_3": 0.61975, "loss_cls_dn_4": 0.09179, "loss_box_dn_4": 0.61386, "loss_cls_dn_5": 0.09461, "loss_box_dn_5": 0.61678, "loss_dense_depth": 0.69424, "loss": 22.77903, "grad_norm": 30.41458, "time": 1.65324} -{"mode": "train", "epoch": 1, "iter": 382, "lr": 0.00025, "memory": 49164, "data_time": 0.10258, "loss_cls_0": 0.71061, "loss_box_0": 1.54897, "loss_cns_0": 0.63946, "loss_yns_0": 0.13573, "loss_cls_1": 0.76306, "loss_box_1": 1.41208, "loss_cns_1": 0.66219, "loss_yns_1": 0.13407, "loss_cls_2": 0.76433, "loss_box_2": 1.39629, "loss_cns_2": 0.6634, "loss_yns_2": 0.13724, "loss_cls_3": 0.77598, "loss_box_3": 1.37977, "loss_cns_3": 0.66186, "loss_yns_3": 0.13297, "loss_cls_4": 0.77911, "loss_box_4": 1.39147, "loss_cns_4": 0.66672, "loss_yns_4": 0.13472, "loss_cls_5": 0.79233, "loss_box_5": 1.37403, "loss_cns_5": 0.66382, "loss_yns_5": 0.13226, "loss_cls_dn_0": 0.125, "loss_box_dn_0": 0.71416, "loss_cls_dn_1": 0.09295, "loss_box_dn_1": 0.62353, "loss_cls_dn_2": 0.0913, "loss_box_dn_2": 0.61257, "loss_cls_dn_3": 0.09355, "loss_box_dn_3": 0.60921, "loss_cls_dn_4": 0.09173, "loss_box_dn_4": 0.61306, "loss_cls_dn_5": 0.09349, "loss_box_dn_5": 0.60929, "loss_dense_depth": 0.70225, "loss": 22.92457, "grad_norm": 36.32612, "time": 1.62291} -{"mode": "train", "epoch": 1, "iter": 383, "lr": 0.00025, "memory": 49164, "data_time": 0.09805, "loss_cls_0": 0.73618, "loss_box_0": 1.53233, "loss_cns_0": 0.63792, "loss_yns_0": 0.13556, "loss_cls_1": 0.77741, "loss_box_1": 1.43619, "loss_cns_1": 0.66623, "loss_yns_1": 0.13494, "loss_cls_2": 0.79634, "loss_box_2": 1.40073, "loss_cns_2": 0.66738, "loss_yns_2": 0.1354, "loss_cls_3": 0.80692, "loss_box_3": 1.38979, "loss_cns_3": 0.6669, "loss_yns_3": 0.13415, "loss_cls_4": 0.8055, "loss_box_4": 1.38244, "loss_cns_4": 0.67091, "loss_yns_4": 0.13603, "loss_cls_5": 0.81399, "loss_box_5": 1.38018, "loss_cns_5": 0.66786, "loss_yns_5": 0.13413, "loss_cls_dn_0": 0.11874, "loss_box_dn_0": 0.70729, "loss_cls_dn_1": 0.09145, "loss_box_dn_1": 0.62113, "loss_cls_dn_2": 0.09276, "loss_box_dn_2": 0.61029, "loss_cls_dn_3": 0.09596, "loss_box_dn_3": 0.60632, "loss_cls_dn_4": 0.09471, "loss_box_dn_4": 0.60135, "loss_cls_dn_5": 0.09606, "loss_box_dn_5": 0.60025, "loss_dense_depth": 0.69454, "loss": 23.07627, "grad_norm": 36.90752, "time": 1.59778} -{"mode": "train", "epoch": 1, "iter": 384, "lr": 0.00025, "memory": 49164, "data_time": 0.08303, "loss_cls_0": 0.75159, "loss_box_0": 1.5409, "loss_cns_0": 0.63704, "loss_yns_0": 0.13675, "loss_cls_1": 0.7886, "loss_box_1": 1.5185, "loss_cns_1": 0.65839, "loss_yns_1": 0.13548, "loss_cls_2": 0.80972, "loss_box_2": 1.45014, "loss_cns_2": 0.66064, "loss_yns_2": 0.135, "loss_cls_3": 0.8151, "loss_box_3": 1.45548, "loss_cns_3": 0.66187, "loss_yns_3": 0.13384, "loss_cls_4": 0.81046, "loss_box_4": 1.45132, "loss_cns_4": 0.661, "loss_yns_4": 0.13454, "loss_cls_5": 0.82095, "loss_box_5": 1.44875, "loss_cns_5": 0.66245, "loss_yns_5": 0.13495, "loss_cls_dn_0": 0.11837, "loss_box_dn_0": 0.70576, "loss_cls_dn_1": 0.0912, "loss_box_dn_1": 0.63166, "loss_cls_dn_2": 0.09403, "loss_box_dn_2": 0.61313, "loss_cls_dn_3": 0.0961, "loss_box_dn_3": 0.62002, "loss_cls_dn_4": 0.0935, "loss_box_dn_4": 0.61144, "loss_cls_dn_5": 0.09425, "loss_box_dn_5": 0.60971, "loss_dense_depth": 0.71016, "loss": 23.50277, "grad_norm": 34.24694, "time": 1.60461} -{"mode": "train", "epoch": 1, "iter": 385, "lr": 0.00025, "memory": 49164, "data_time": 0.08278, "loss_cls_0": 0.71266, "loss_box_0": 1.52722, "loss_cns_0": 0.64154, "loss_yns_0": 0.13801, "loss_cls_1": 0.76707, "loss_box_1": 1.45687, "loss_cns_1": 0.66298, "loss_yns_1": 0.13723, "loss_cls_2": 0.79625, "loss_box_2": 1.41768, "loss_cns_2": 0.66336, "loss_yns_2": 0.13667, "loss_cls_3": 0.79186, "loss_box_3": 1.42966, "loss_cns_3": 0.66306, "loss_yns_3": 0.13613, "loss_cls_4": 0.78744, "loss_box_4": 1.39882, "loss_cns_4": 0.66301, "loss_yns_4": 0.13531, "loss_cls_5": 0.79107, "loss_box_5": 1.3783, "loss_cns_5": 0.66385, "loss_yns_5": 0.13524, "loss_cls_dn_0": 0.11913, "loss_box_dn_0": 0.71075, "loss_cls_dn_1": 0.08799, "loss_box_dn_1": 0.62613, "loss_cls_dn_2": 0.08855, "loss_box_dn_2": 0.62584, "loss_cls_dn_3": 0.08763, "loss_box_dn_3": 0.64024, "loss_cls_dn_4": 0.08863, "loss_box_dn_4": 0.62374, "loss_cls_dn_5": 0.08766, "loss_box_dn_5": 0.61023, "loss_dense_depth": 0.67749, "loss": 23.1053, "grad_norm": 38.66861, "time": 1.6014} -{"mode": "train", "epoch": 1, "iter": 386, "lr": 0.00025, "memory": 49164, "data_time": 0.10741, "loss_cls_0": 0.73172, "loss_box_0": 1.55254, "loss_cns_0": 0.63908, "loss_yns_0": 0.13597, "loss_cls_1": 0.77921, "loss_box_1": 1.47135, "loss_cns_1": 0.6613, "loss_yns_1": 0.13448, "loss_cls_2": 0.80378, "loss_box_2": 1.4556, "loss_cns_2": 0.66343, "loss_yns_2": 0.13442, "loss_cls_3": 0.8009, "loss_box_3": 1.44522, "loss_cns_3": 0.66152, "loss_yns_3": 0.13433, "loss_cls_4": 0.80982, "loss_box_4": 1.41953, "loss_cns_4": 0.66308, "loss_yns_4": 0.13318, "loss_cls_5": 0.80692, "loss_box_5": 1.42003, "loss_cns_5": 0.6617, "loss_yns_5": 0.13312, "loss_cls_dn_0": 0.12021, "loss_box_dn_0": 0.71679, "loss_cls_dn_1": 0.08923, "loss_box_dn_1": 0.62102, "loss_cls_dn_2": 0.08968, "loss_box_dn_2": 0.61966, "loss_cls_dn_3": 0.08819, "loss_box_dn_3": 0.61984, "loss_cls_dn_4": 0.09134, "loss_box_dn_4": 0.60662, "loss_cls_dn_5": 0.08955, "loss_box_dn_5": 0.60548, "loss_dense_depth": 0.70372, "loss": 23.31353, "grad_norm": 32.6865, "time": 1.61171} -{"mode": "train", "epoch": 1, "iter": 387, "lr": 0.00025, "memory": 49164, "data_time": 0.08141, "loss_cls_0": 0.69524, "loss_box_0": 1.56597, "loss_cns_0": 0.6389, "loss_yns_0": 0.13977, "loss_cls_1": 0.7577, "loss_box_1": 1.43848, "loss_cns_1": 0.66069, "loss_yns_1": 0.13825, "loss_cls_2": 0.78122, "loss_box_2": 1.40418, "loss_cns_2": 0.66099, "loss_yns_2": 0.13545, "loss_cls_3": 0.78544, "loss_box_3": 1.39543, "loss_cns_3": 0.66095, "loss_yns_3": 0.13572, "loss_cls_4": 0.78722, "loss_box_4": 1.37651, "loss_cns_4": 0.666, "loss_yns_4": 0.13373, "loss_cls_5": 0.79227, "loss_box_5": 1.39287, "loss_cns_5": 0.66, "loss_yns_5": 0.13368, "loss_cls_dn_0": 0.11516, "loss_box_dn_0": 0.71209, "loss_cls_dn_1": 0.08746, "loss_box_dn_1": 0.63781, "loss_cls_dn_2": 0.08885, "loss_box_dn_2": 0.61958, "loss_cls_dn_3": 0.0867, "loss_box_dn_3": 0.61632, "loss_cls_dn_4": 0.08783, "loss_box_dn_4": 0.61522, "loss_cls_dn_5": 0.08821, "loss_box_dn_5": 0.61988, "loss_dense_depth": 0.70027, "loss": 23.01203, "grad_norm": 39.19648, "time": 1.57946} -{"mode": "train", "epoch": 1, "iter": 388, "lr": 0.00025, "memory": 49164, "data_time": 0.08939, "loss_cls_0": 0.69851, "loss_box_0": 1.53192, "loss_cns_0": 0.64099, "loss_yns_0": 0.13803, "loss_cls_1": 0.74923, "loss_box_1": 1.41042, "loss_cns_1": 0.66253, "loss_yns_1": 0.13581, "loss_cls_2": 0.77571, "loss_box_2": 1.3658, "loss_cns_2": 0.66418, "loss_yns_2": 0.13406, "loss_cls_3": 0.76987, "loss_box_3": 1.35959, "loss_cns_3": 0.66067, "loss_yns_3": 0.13485, "loss_cls_4": 0.77324, "loss_box_4": 1.35768, "loss_cns_4": 0.66617, "loss_yns_4": 0.1352, "loss_cls_5": 0.78282, "loss_box_5": 1.3714, "loss_cns_5": 0.66189, "loss_yns_5": 0.13562, "loss_cls_dn_0": 0.11335, "loss_box_dn_0": 0.71193, "loss_cls_dn_1": 0.08885, "loss_box_dn_1": 0.63722, "loss_cls_dn_2": 0.08989, "loss_box_dn_2": 0.61645, "loss_cls_dn_3": 0.08786, "loss_box_dn_3": 0.61211, "loss_cls_dn_4": 0.09014, "loss_box_dn_4": 0.61151, "loss_cls_dn_5": 0.09125, "loss_box_dn_5": 0.62002, "loss_dense_depth": 0.69533, "loss": 22.78212, "grad_norm": 24.29554, "time": 1.58434} -{"mode": "train", "epoch": 1, "iter": 389, "lr": 0.00025, "memory": 49164, "data_time": 0.08833, "loss_cls_0": 0.69298, "loss_box_0": 1.53436, "loss_cns_0": 0.64144, "loss_yns_0": 0.13816, "loss_cls_1": 0.75386, "loss_box_1": 1.4189, "loss_cns_1": 0.66596, "loss_yns_1": 0.13661, "loss_cls_2": 0.78978, "loss_box_2": 1.37394, "loss_cns_2": 0.66425, "loss_yns_2": 0.13575, "loss_cls_3": 0.77946, "loss_box_3": 1.35852, "loss_cns_3": 0.66126, "loss_yns_3": 0.13456, "loss_cls_4": 0.79198, "loss_box_4": 1.34791, "loss_cns_4": 0.66227, "loss_yns_4": 0.1372, "loss_cls_5": 0.78837, "loss_box_5": 1.35387, "loss_cns_5": 0.66039, "loss_yns_5": 0.13446, "loss_cls_dn_0": 0.11013, "loss_box_dn_0": 0.70268, "loss_cls_dn_1": 0.08899, "loss_box_dn_1": 0.64111, "loss_cls_dn_2": 0.08887, "loss_box_dn_2": 0.62484, "loss_cls_dn_3": 0.08704, "loss_box_dn_3": 0.61788, "loss_cls_dn_4": 0.08866, "loss_box_dn_4": 0.61134, "loss_cls_dn_5": 0.08969, "loss_box_dn_5": 0.61413, "loss_dense_depth": 0.72391, "loss": 22.84553, "grad_norm": 34.75682, "time": 11.55058} -{"mode": "train", "epoch": 1, "iter": 390, "lr": 0.00026, "memory": 49164, "data_time": 0.08289, "loss_cls_0": 0.69569, "loss_box_0": 1.52559, "loss_cns_0": 0.64036, "loss_yns_0": 0.1427, "loss_cls_1": 0.76604, "loss_box_1": 1.40445, "loss_cns_1": 0.66354, "loss_yns_1": 0.13802, "loss_cls_2": 0.78142, "loss_box_2": 1.37223, "loss_cns_2": 0.66173, "loss_yns_2": 0.13632, "loss_cls_3": 0.78244, "loss_box_3": 1.36082, "loss_cns_3": 0.66404, "loss_yns_3": 0.13911, "loss_cls_4": 0.78874, "loss_box_4": 1.356, "loss_cns_4": 0.66774, "loss_yns_4": 0.13883, "loss_cls_5": 0.78597, "loss_box_5": 1.36434, "loss_cns_5": 0.66441, "loss_yns_5": 0.13698, "loss_cls_dn_0": 0.11189, "loss_box_dn_0": 0.69831, "loss_cls_dn_1": 0.09111, "loss_box_dn_1": 0.62165, "loss_cls_dn_2": 0.09126, "loss_box_dn_2": 0.6107, "loss_cls_dn_3": 0.09202, "loss_box_dn_3": 0.6079, "loss_cls_dn_4": 0.0922, "loss_box_dn_4": 0.60841, "loss_cls_dn_5": 0.09112, "loss_box_dn_5": 0.61345, "loss_dense_depth": 0.72058, "loss": 22.82808, "grad_norm": 32.46416, "time": 1.5607} -{"mode": "train", "epoch": 1, "iter": 391, "lr": 0.00026, "memory": 49164, "data_time": 0.07302, "loss_cls_0": 0.69157, "loss_box_0": 1.52798, "loss_cns_0": 0.64036, "loss_yns_0": 0.13868, "loss_cls_1": 0.76486, "loss_box_1": 1.3928, "loss_cns_1": 0.66348, "loss_yns_1": 0.13552, "loss_cls_2": 0.78367, "loss_box_2": 1.36424, "loss_cns_2": 0.66877, "loss_yns_2": 0.13758, "loss_cls_3": 0.77764, "loss_box_3": 1.35462, "loss_cns_3": 0.66569, "loss_yns_3": 0.13817, "loss_cls_4": 0.78815, "loss_box_4": 1.34085, "loss_cns_4": 0.67116, "loss_yns_4": 0.13456, "loss_cls_5": 0.79141, "loss_box_5": 1.35768, "loss_cns_5": 0.66388, "loss_yns_5": 0.13376, "loss_cls_dn_0": 0.11435, "loss_box_dn_0": 0.71361, "loss_cls_dn_1": 0.09506, "loss_box_dn_1": 0.62824, "loss_cls_dn_2": 0.09157, "loss_box_dn_2": 0.61672, "loss_cls_dn_3": 0.09212, "loss_box_dn_3": 0.61368, "loss_cls_dn_4": 0.09366, "loss_box_dn_4": 0.60821, "loss_cls_dn_5": 0.09419, "loss_box_dn_5": 0.61564, "loss_dense_depth": 0.74322, "loss": 22.84736, "grad_norm": 28.9269, "time": 1.61154} -{"mode": "train", "epoch": 1, "iter": 392, "lr": 0.00026, "memory": 49164, "data_time": 0.07736, "loss_cls_0": 0.70962, "loss_box_0": 1.54583, "loss_cns_0": 0.63625, "loss_yns_0": 0.13883, "loss_cls_1": 0.7614, "loss_box_1": 1.4222, "loss_cns_1": 0.6635, "loss_yns_1": 0.13752, "loss_cls_2": 0.78517, "loss_box_2": 1.39618, "loss_cns_2": 0.66699, "loss_yns_2": 0.13789, "loss_cls_3": 0.79014, "loss_box_3": 1.40458, "loss_cns_3": 0.66226, "loss_yns_3": 0.13696, "loss_cls_4": 0.80389, "loss_box_4": 1.40215, "loss_cns_4": 0.66468, "loss_yns_4": 0.13962, "loss_cls_5": 0.80551, "loss_box_5": 1.39841, "loss_cns_5": 0.66048, "loss_yns_5": 0.13836, "loss_cls_dn_0": 0.11848, "loss_box_dn_0": 0.71155, "loss_cls_dn_1": 0.09264, "loss_box_dn_1": 0.62991, "loss_cls_dn_2": 0.09343, "loss_box_dn_2": 0.61705, "loss_cls_dn_3": 0.09497, "loss_box_dn_3": 0.61895, "loss_cls_dn_4": 0.0982, "loss_box_dn_4": 0.61884, "loss_cls_dn_5": 0.0988, "loss_box_dn_5": 0.61682, "loss_dense_depth": 0.70766, "loss": 23.12572, "grad_norm": 36.14661, "time": 1.58139} -{"mode": "train", "epoch": 1, "iter": 393, "lr": 0.00026, "memory": 49164, "data_time": 0.07671, "loss_cls_0": 0.72249, "loss_box_0": 1.57919, "loss_cns_0": 0.63707, "loss_yns_0": 0.14506, "loss_cls_1": 0.77623, "loss_box_1": 1.42955, "loss_cns_1": 0.66332, "loss_yns_1": 0.13813, "loss_cls_2": 0.78836, "loss_box_2": 1.38508, "loss_cns_2": 0.66369, "loss_yns_2": 0.13673, "loss_cls_3": 0.79233, "loss_box_3": 1.38546, "loss_cns_3": 0.66304, "loss_yns_3": 0.13732, "loss_cls_4": 0.79461, "loss_box_4": 1.38039, "loss_cns_4": 0.66576, "loss_yns_4": 0.14087, "loss_cls_5": 0.80512, "loss_box_5": 1.38028, "loss_cns_5": 0.66367, "loss_yns_5": 0.13889, "loss_cls_dn_0": 0.11742, "loss_box_dn_0": 0.71317, "loss_cls_dn_1": 0.09288, "loss_box_dn_1": 0.63481, "loss_cls_dn_2": 0.09467, "loss_box_dn_2": 0.61758, "loss_cls_dn_3": 0.09552, "loss_box_dn_3": 0.61757, "loss_cls_dn_4": 0.09395, "loss_box_dn_4": 0.62043, "loss_cls_dn_5": 0.09447, "loss_box_dn_5": 0.62563, "loss_dense_depth": 0.70867, "loss": 23.13943, "grad_norm": 27.66484, "time": 1.58367} -{"mode": "train", "epoch": 1, "iter": 394, "lr": 0.00026, "memory": 49164, "data_time": 0.08196, "loss_cls_0": 0.69712, "loss_box_0": 1.52532, "loss_cns_0": 0.63832, "loss_yns_0": 0.14435, "loss_cls_1": 0.75491, "loss_box_1": 1.41299, "loss_cns_1": 0.66293, "loss_yns_1": 0.13907, "loss_cls_2": 0.77253, "loss_box_2": 1.35907, "loss_cns_2": 0.66167, "loss_yns_2": 0.13843, "loss_cls_3": 0.77575, "loss_box_3": 1.34273, "loss_cns_3": 0.66521, "loss_yns_3": 0.13824, "loss_cls_4": 0.78014, "loss_box_4": 1.34832, "loss_cns_4": 0.66253, "loss_yns_4": 0.13994, "loss_cls_5": 0.78736, "loss_box_5": 1.36204, "loss_cns_5": 0.66325, "loss_yns_5": 0.13666, "loss_cls_dn_0": 0.11847, "loss_box_dn_0": 0.70541, "loss_cls_dn_1": 0.09333, "loss_box_dn_1": 0.62865, "loss_cls_dn_2": 0.09277, "loss_box_dn_2": 0.61011, "loss_cls_dn_3": 0.09128, "loss_box_dn_3": 0.60319, "loss_cls_dn_4": 0.09235, "loss_box_dn_4": 0.60957, "loss_cls_dn_5": 0.09389, "loss_box_dn_5": 0.61649, "loss_dense_depth": 0.70633, "loss": 22.77073, "grad_norm": 40.21982, "time": 1.61595} -{"mode": "train", "epoch": 1, "iter": 395, "lr": 0.00026, "memory": 49164, "data_time": 0.07485, "loss_cls_0": 0.70313, "loss_box_0": 1.53112, "loss_cns_0": 0.63754, "loss_yns_0": 0.14459, "loss_cls_1": 0.76439, "loss_box_1": 1.37519, "loss_cns_1": 0.66383, "loss_yns_1": 0.14056, "loss_cls_2": 0.77652, "loss_box_2": 1.3437, "loss_cns_2": 0.66379, "loss_yns_2": 0.14096, "loss_cls_3": 0.77432, "loss_box_3": 1.3327, "loss_cns_3": 0.66479, "loss_yns_3": 0.14195, "loss_cls_4": 0.77429, "loss_box_4": 1.34108, "loss_cns_4": 0.66526, "loss_yns_4": 0.14004, "loss_cls_5": 0.77635, "loss_box_5": 1.32583, "loss_cns_5": 0.66292, "loss_yns_5": 0.13995, "loss_cls_dn_0": 0.11614, "loss_box_dn_0": 0.71217, "loss_cls_dn_1": 0.09124, "loss_box_dn_1": 0.61942, "loss_cls_dn_2": 0.08757, "loss_box_dn_2": 0.60967, "loss_cls_dn_3": 0.08948, "loss_box_dn_3": 0.60648, "loss_cls_dn_4": 0.09247, "loss_box_dn_4": 0.61158, "loss_cls_dn_5": 0.09623, "loss_box_dn_5": 0.60579, "loss_dense_depth": 0.71563, "loss": 22.6787, "grad_norm": 35.80139, "time": 1.58579} -{"mode": "train", "epoch": 1, "iter": 396, "lr": 0.00026, "memory": 49164, "data_time": 0.07533, "loss_cls_0": 0.7035, "loss_box_0": 1.50017, "loss_cns_0": 0.64123, "loss_yns_0": 0.14345, "loss_cls_1": 0.77306, "loss_box_1": 1.35505, "loss_cns_1": 0.66714, "loss_yns_1": 0.14183, "loss_cls_2": 0.79288, "loss_box_2": 1.31304, "loss_cns_2": 0.66584, "loss_yns_2": 0.14174, "loss_cls_3": 0.7938, "loss_box_3": 1.30747, "loss_cns_3": 0.66573, "loss_yns_3": 0.14122, "loss_cls_4": 0.78875, "loss_box_4": 1.31254, "loss_cns_4": 0.6711, "loss_yns_4": 0.14169, "loss_cls_5": 0.79871, "loss_box_5": 1.30149, "loss_cns_5": 0.66492, "loss_yns_5": 0.14056, "loss_cls_dn_0": 0.11577, "loss_box_dn_0": 0.71331, "loss_cls_dn_1": 0.09043, "loss_box_dn_1": 0.61248, "loss_cls_dn_2": 0.08782, "loss_box_dn_2": 0.59691, "loss_cls_dn_3": 0.08728, "loss_box_dn_3": 0.59442, "loss_cls_dn_4": 0.08733, "loss_box_dn_4": 0.59611, "loss_cls_dn_5": 0.09072, "loss_box_dn_5": 0.5927, "loss_dense_depth": 0.73154, "loss": 22.56372, "grad_norm": 27.30521, "time": 1.58973} -{"mode": "train", "epoch": 1, "iter": 397, "lr": 0.00026, "memory": 49164, "data_time": 0.07155, "loss_cls_0": 0.72136, "loss_box_0": 1.53605, "loss_cns_0": 0.63512, "loss_yns_0": 0.14336, "loss_cls_1": 0.7891, "loss_box_1": 1.36485, "loss_cns_1": 0.66681, "loss_yns_1": 0.14032, "loss_cls_2": 0.79802, "loss_box_2": 1.34629, "loss_cns_2": 0.66608, "loss_yns_2": 0.13927, "loss_cls_3": 0.80269, "loss_box_3": 1.34432, "loss_cns_3": 0.66629, "loss_yns_3": 0.13963, "loss_cls_4": 0.80682, "loss_box_4": 1.35024, "loss_cns_4": 0.67181, "loss_yns_4": 0.14231, "loss_cls_5": 0.81774, "loss_box_5": 1.34843, "loss_cns_5": 0.66511, "loss_yns_5": 0.14136, "loss_cls_dn_0": 0.11818, "loss_box_dn_0": 0.71316, "loss_cls_dn_1": 0.09218, "loss_box_dn_1": 0.61139, "loss_cls_dn_2": 0.09222, "loss_box_dn_2": 0.60187, "loss_cls_dn_3": 0.09112, "loss_box_dn_3": 0.59732, "loss_cls_dn_4": 0.08934, "loss_box_dn_4": 0.59835, "loss_cls_dn_5": 0.09088, "loss_box_dn_5": 0.60246, "loss_dense_depth": 0.72651, "loss": 22.86836, "grad_norm": 39.14138, "time": 1.58719} -{"mode": "train", "epoch": 1, "iter": 398, "lr": 0.00026, "memory": 49164, "data_time": 0.07599, "loss_cls_0": 0.75197, "loss_box_0": 1.53086, "loss_cns_0": 0.62997, "loss_yns_0": 0.14316, "loss_cls_1": 0.79823, "loss_box_1": 1.39613, "loss_cns_1": 0.663, "loss_yns_1": 0.13948, "loss_cls_2": 0.80346, "loss_box_2": 1.3767, "loss_cns_2": 0.66138, "loss_yns_2": 0.13916, "loss_cls_3": 0.79963, "loss_box_3": 1.35787, "loss_cns_3": 0.66294, "loss_yns_3": 0.13826, "loss_cls_4": 0.80891, "loss_box_4": 1.36418, "loss_cns_4": 0.66175, "loss_yns_4": 0.14085, "loss_cls_5": 0.8126, "loss_box_5": 1.36577, "loss_cns_5": 0.66099, "loss_yns_5": 0.13948, "loss_cls_dn_0": 0.11735, "loss_box_dn_0": 0.7083, "loss_cls_dn_1": 0.08868, "loss_box_dn_1": 0.60689, "loss_cls_dn_2": 0.08809, "loss_box_dn_2": 0.5984, "loss_cls_dn_3": 0.08995, "loss_box_dn_3": 0.5889, "loss_cls_dn_4": 0.08904, "loss_box_dn_4": 0.59349, "loss_cls_dn_5": 0.08872, "loss_box_dn_5": 0.59584, "loss_dense_depth": 0.75581, "loss": 22.9562, "grad_norm": 29.15473, "time": 1.58418} -{"mode": "train", "epoch": 1, "iter": 399, "lr": 0.00026, "memory": 49164, "data_time": 0.07372, "loss_cls_0": 0.71752, "loss_box_0": 1.51869, "loss_cns_0": 0.63586, "loss_yns_0": 0.13999, "loss_cls_1": 0.77957, "loss_box_1": 1.40348, "loss_cns_1": 0.66541, "loss_yns_1": 0.13989, "loss_cls_2": 0.80027, "loss_box_2": 1.36897, "loss_cns_2": 0.66259, "loss_yns_2": 0.14106, "loss_cls_3": 0.78739, "loss_box_3": 1.35716, "loss_cns_3": 0.66352, "loss_yns_3": 0.13892, "loss_cls_4": 0.79524, "loss_box_4": 1.35353, "loss_cns_4": 0.66075, "loss_yns_4": 0.13872, "loss_cls_5": 0.81126, "loss_box_5": 1.36353, "loss_cns_5": 0.66252, "loss_yns_5": 0.139, "loss_cls_dn_0": 0.11857, "loss_box_dn_0": 0.70701, "loss_cls_dn_1": 0.08985, "loss_box_dn_1": 0.6107, "loss_cls_dn_2": 0.08929, "loss_box_dn_2": 0.5987, "loss_cls_dn_3": 0.08969, "loss_box_dn_3": 0.59422, "loss_cls_dn_4": 0.09148, "loss_box_dn_4": 0.59221, "loss_cls_dn_5": 0.09407, "loss_box_dn_5": 0.59633, "loss_dense_depth": 0.75539, "loss": 22.87238, "grad_norm": 37.98988, "time": 1.56786} -{"mode": "train", "epoch": 1, "iter": 400, "lr": 0.00026, "memory": 49164, "data_time": 0.07404, "loss_cls_0": 0.73637, "loss_box_0": 1.52733, "loss_cns_0": 0.64173, "loss_yns_0": 0.14245, "loss_cls_1": 0.79422, "loss_box_1": 1.40944, "loss_cns_1": 0.6679, "loss_yns_1": 0.14056, "loss_cls_2": 0.80678, "loss_box_2": 1.36231, "loss_cns_2": 0.66503, "loss_yns_2": 0.14188, "loss_cls_3": 0.79872, "loss_box_3": 1.32666, "loss_cns_3": 0.66185, "loss_yns_3": 0.13951, "loss_cls_4": 0.80516, "loss_box_4": 1.31992, "loss_cns_4": 0.66727, "loss_yns_4": 0.13932, "loss_cls_5": 0.81802, "loss_box_5": 1.33822, "loss_cns_5": 0.66227, "loss_yns_5": 0.14136, "loss_cls_dn_0": 0.11837, "loss_box_dn_0": 0.7191, "loss_cls_dn_1": 0.08994, "loss_box_dn_1": 0.63701, "loss_cls_dn_2": 0.09018, "loss_box_dn_2": 0.6178, "loss_cls_dn_3": 0.08976, "loss_box_dn_3": 0.60906, "loss_cls_dn_4": 0.08875, "loss_box_dn_4": 0.61232, "loss_cls_dn_5": 0.09126, "loss_box_dn_5": 0.61906, "loss_dense_depth": 0.71701, "loss": 22.95389, "grad_norm": 35.31143, "time": 1.57653} -{"mode": "train", "epoch": 1, "iter": 401, "lr": 0.00026, "memory": 49164, "data_time": 0.10355, "loss_cls_0": 0.74224, "loss_box_0": 1.54526, "loss_cns_0": 0.64259, "loss_yns_0": 0.14172, "loss_cls_1": 0.80211, "loss_box_1": 1.42586, "loss_cns_1": 0.65938, "loss_yns_1": 0.14152, "loss_cls_2": 0.81187, "loss_box_2": 1.3994, "loss_cns_2": 0.65924, "loss_yns_2": 0.14149, "loss_cls_3": 0.81882, "loss_box_3": 1.37247, "loss_cns_3": 0.6604, "loss_yns_3": 0.14103, "loss_cls_4": 0.82576, "loss_box_4": 1.37334, "loss_cns_4": 0.66851, "loss_yns_4": 0.13956, "loss_cls_5": 0.84036, "loss_box_5": 1.37615, "loss_cns_5": 0.65914, "loss_yns_5": 0.14137, "loss_cls_dn_0": 0.11734, "loss_box_dn_0": 0.71469, "loss_cls_dn_1": 0.09195, "loss_box_dn_1": 0.62092, "loss_cls_dn_2": 0.09362, "loss_box_dn_2": 0.60662, "loss_cls_dn_3": 0.09199, "loss_box_dn_3": 0.59896, "loss_cls_dn_4": 0.09279, "loss_box_dn_4": 0.60325, "loss_cls_dn_5": 0.09423, "loss_box_dn_5": 0.60647, "loss_dense_depth": 0.78126, "loss": 23.24368, "grad_norm": 42.2604, "time": 1.65983} -{"mode": "train", "epoch": 1, "iter": 402, "lr": 0.00026, "memory": 49164, "data_time": 0.09926, "loss_cls_0": 0.71523, "loss_box_0": 1.53684, "loss_cns_0": 0.64338, "loss_yns_0": 0.14076, "loss_cls_1": 0.78575, "loss_box_1": 1.40054, "loss_cns_1": 0.65952, "loss_yns_1": 0.1363, "loss_cls_2": 0.79813, "loss_box_2": 1.37118, "loss_cns_2": 0.6574, "loss_yns_2": 0.13582, "loss_cls_3": 0.78786, "loss_box_3": 1.38212, "loss_cns_3": 0.66242, "loss_yns_3": 0.13812, "loss_cls_4": 0.80044, "loss_box_4": 1.36757, "loss_cns_4": 0.66365, "loss_yns_4": 0.13754, "loss_cls_5": 0.80979, "loss_box_5": 1.36245, "loss_cns_5": 0.65916, "loss_yns_5": 0.13656, "loss_cls_dn_0": 0.11302, "loss_box_dn_0": 0.70779, "loss_cls_dn_1": 0.09044, "loss_box_dn_1": 0.61258, "loss_cls_dn_2": 0.09051, "loss_box_dn_2": 0.59961, "loss_cls_dn_3": 0.0881, "loss_box_dn_3": 0.59758, "loss_cls_dn_4": 0.0904, "loss_box_dn_4": 0.59679, "loss_cls_dn_5": 0.0919, "loss_box_dn_5": 0.59659, "loss_dense_depth": 0.74238, "loss": 22.90622, "grad_norm": 28.91803, "time": 1.64242} -{"mode": "train", "epoch": 1, "iter": 403, "lr": 0.00026, "memory": 49164, "data_time": 0.08573, "loss_cls_0": 0.71602, "loss_box_0": 1.55893, "loss_cns_0": 0.64025, "loss_yns_0": 0.13917, "loss_cls_1": 0.76203, "loss_box_1": 1.42299, "loss_cns_1": 0.66058, "loss_yns_1": 0.13468, "loss_cls_2": 0.77877, "loss_box_2": 1.39909, "loss_cns_2": 0.66334, "loss_yns_2": 0.13417, "loss_cls_3": 0.79075, "loss_box_3": 1.39837, "loss_cns_3": 0.66159, "loss_yns_3": 0.13454, "loss_cls_4": 0.78896, "loss_box_4": 1.38947, "loss_cns_4": 0.6611, "loss_yns_4": 0.13417, "loss_cls_5": 0.79889, "loss_box_5": 1.38498, "loss_cns_5": 0.66168, "loss_yns_5": 0.13445, "loss_cls_dn_0": 0.11794, "loss_box_dn_0": 0.72018, "loss_cls_dn_1": 0.09002, "loss_box_dn_1": 0.61761, "loss_cls_dn_2": 0.08951, "loss_box_dn_2": 0.60791, "loss_cls_dn_3": 0.09214, "loss_box_dn_3": 0.60375, "loss_cls_dn_4": 0.08975, "loss_box_dn_4": 0.60205, "loss_cls_dn_5": 0.09124, "loss_box_dn_5": 0.60402, "loss_dense_depth": 0.79188, "loss": 23.06699, "grad_norm": 39.08163, "time": 1.57407} -{"mode": "train", "epoch": 1, "iter": 404, "lr": 0.00026, "memory": 49164, "data_time": 0.07342, "loss_cls_0": 0.71284, "loss_box_0": 1.54631, "loss_cns_0": 0.64026, "loss_yns_0": 0.13734, "loss_cls_1": 0.7631, "loss_box_1": 1.40342, "loss_cns_1": 0.66342, "loss_yns_1": 0.13629, "loss_cls_2": 0.79627, "loss_box_2": 1.3709, "loss_cns_2": 0.66608, "loss_yns_2": 0.13494, "loss_cls_3": 0.79902, "loss_box_3": 1.368, "loss_cns_3": 0.66369, "loss_yns_3": 0.13263, "loss_cls_4": 0.78701, "loss_box_4": 1.37236, "loss_cns_4": 0.66546, "loss_yns_4": 0.13358, "loss_cls_5": 0.79622, "loss_box_5": 1.35433, "loss_cns_5": 0.66594, "loss_yns_5": 0.13557, "loss_cls_dn_0": 0.11781, "loss_box_dn_0": 0.71205, "loss_cls_dn_1": 0.09074, "loss_box_dn_1": 0.60801, "loss_cls_dn_2": 0.09121, "loss_box_dn_2": 0.59537, "loss_cls_dn_3": 0.09187, "loss_box_dn_3": 0.59483, "loss_cls_dn_4": 0.088, "loss_box_dn_4": 0.59746, "loss_cls_dn_5": 0.09045, "loss_box_dn_5": 0.59594, "loss_dense_depth": 0.7401, "loss": 22.85883, "grad_norm": 39.67599, "time": 1.56831} -{"mode": "train", "epoch": 1, "iter": 405, "lr": 0.00026, "memory": 49164, "data_time": 0.08769, "loss_cls_0": 0.7399, "loss_box_0": 1.54922, "loss_cns_0": 0.63503, "loss_yns_0": 0.13627, "loss_cls_1": 0.77096, "loss_box_1": 1.42225, "loss_cns_1": 0.66089, "loss_yns_1": 0.13571, "loss_cls_2": 0.78489, "loss_box_2": 1.39946, "loss_cns_2": 0.66117, "loss_yns_2": 0.13607, "loss_cls_3": 0.80262, "loss_box_3": 1.38717, "loss_cns_3": 0.66162, "loss_yns_3": 0.13441, "loss_cls_4": 0.79914, "loss_box_4": 1.40289, "loss_cns_4": 0.66312, "loss_yns_4": 0.13506, "loss_cls_5": 0.81561, "loss_box_5": 1.38091, "loss_cns_5": 0.66164, "loss_yns_5": 0.13461, "loss_cls_dn_0": 0.12189, "loss_box_dn_0": 0.71948, "loss_cls_dn_1": 0.08985, "loss_box_dn_1": 0.61213, "loss_cls_dn_2": 0.08919, "loss_box_dn_2": 0.59858, "loss_cls_dn_3": 0.08995, "loss_box_dn_3": 0.59948, "loss_cls_dn_4": 0.08964, "loss_box_dn_4": 0.60399, "loss_cls_dn_5": 0.09187, "loss_box_dn_5": 0.6014, "loss_dense_depth": 0.74956, "loss": 23.06764, "grad_norm": 31.99951, "time": 1.59672} -{"mode": "train", "epoch": 1, "iter": 406, "lr": 0.00026, "memory": 49164, "data_time": 0.11718, "loss_cls_0": 0.72841, "loss_box_0": 1.52542, "loss_cns_0": 0.63535, "loss_yns_0": 0.13411, "loss_cls_1": 0.76626, "loss_box_1": 1.40614, "loss_cns_1": 0.66246, "loss_yns_1": 0.13449, "loss_cls_2": 0.77929, "loss_box_2": 1.37541, "loss_cns_2": 0.66252, "loss_yns_2": 0.13466, "loss_cls_3": 0.80177, "loss_box_3": 1.36755, "loss_cns_3": 0.66035, "loss_yns_3": 0.13441, "loss_cls_4": 0.78424, "loss_box_4": 1.37378, "loss_cns_4": 0.66359, "loss_yns_4": 0.13513, "loss_cls_5": 0.79616, "loss_box_5": 1.36302, "loss_cns_5": 0.66283, "loss_yns_5": 0.13395, "loss_cls_dn_0": 0.11653, "loss_box_dn_0": 0.71178, "loss_cls_dn_1": 0.08988, "loss_box_dn_1": 0.60955, "loss_cls_dn_2": 0.09051, "loss_box_dn_2": 0.59743, "loss_cls_dn_3": 0.09335, "loss_box_dn_3": 0.59932, "loss_cls_dn_4": 0.09087, "loss_box_dn_4": 0.59909, "loss_cls_dn_5": 0.09278, "loss_box_dn_5": 0.59477, "loss_dense_depth": 0.72906, "loss": 22.83622, "grad_norm": 36.03376, "time": 1.62289} -{"mode": "train", "epoch": 1, "iter": 407, "lr": 0.00026, "memory": 49164, "data_time": 0.08235, "loss_cls_0": 0.74791, "loss_box_0": 1.56011, "loss_cns_0": 0.63942, "loss_yns_0": 0.13578, "loss_cls_1": 0.7844, "loss_box_1": 1.39488, "loss_cns_1": 0.66569, "loss_yns_1": 0.13268, "loss_cls_2": 0.80012, "loss_box_2": 1.35335, "loss_cns_2": 0.66442, "loss_yns_2": 0.13387, "loss_cls_3": 0.81468, "loss_box_3": 1.34684, "loss_cns_3": 0.66495, "loss_yns_3": 0.13377, "loss_cls_4": 0.79814, "loss_box_4": 1.36739, "loss_cns_4": 0.67126, "loss_yns_4": 0.1342, "loss_cls_5": 0.80353, "loss_box_5": 1.36008, "loss_cns_5": 0.66959, "loss_yns_5": 0.13347, "loss_cls_dn_0": 0.11434, "loss_box_dn_0": 0.71231, "loss_cls_dn_1": 0.09167, "loss_box_dn_1": 0.62253, "loss_cls_dn_2": 0.0925, "loss_box_dn_2": 0.60519, "loss_cls_dn_3": 0.0938, "loss_box_dn_3": 0.60589, "loss_cls_dn_4": 0.08943, "loss_box_dn_4": 0.61574, "loss_cls_dn_5": 0.09138, "loss_box_dn_5": 0.61555, "loss_dense_depth": 0.71737, "loss": 22.97824, "grad_norm": 27.70052, "time": 1.60442} -{"mode": "train", "epoch": 1, "iter": 408, "lr": 0.00026, "memory": 49164, "data_time": 0.07448, "loss_cls_0": 0.72447, "loss_box_0": 1.5734, "loss_cns_0": 0.63809, "loss_yns_0": 0.13801, "loss_cls_1": 0.76957, "loss_box_1": 1.39662, "loss_cns_1": 0.66885, "loss_yns_1": 0.1337, "loss_cls_2": 0.78458, "loss_box_2": 1.36315, "loss_cns_2": 0.66912, "loss_yns_2": 0.13449, "loss_cls_3": 0.80139, "loss_box_3": 1.35494, "loss_cns_3": 0.66886, "loss_yns_3": 0.13454, "loss_cls_4": 0.78471, "loss_box_4": 1.37306, "loss_cns_4": 0.67245, "loss_yns_4": 0.13513, "loss_cls_5": 0.79051, "loss_box_5": 1.36315, "loss_cns_5": 0.67013, "loss_yns_5": 0.13511, "loss_cls_dn_0": 0.11389, "loss_box_dn_0": 0.71605, "loss_cls_dn_1": 0.09146, "loss_box_dn_1": 0.63534, "loss_cls_dn_2": 0.0909, "loss_box_dn_2": 0.6185, "loss_cls_dn_3": 0.09104, "loss_box_dn_3": 0.61598, "loss_cls_dn_4": 0.0885, "loss_box_dn_4": 0.62623, "loss_cls_dn_5": 0.09033, "loss_box_dn_5": 0.62302, "loss_dense_depth": 0.71424, "loss": 22.99351, "grad_norm": 39.10096, "time": 1.56965} -{"mode": "train", "epoch": 1, "iter": 409, "lr": 0.00026, "memory": 49164, "data_time": 0.08705, "loss_cls_0": 0.70989, "loss_box_0": 1.54478, "loss_cns_0": 0.6441, "loss_yns_0": 0.13945, "loss_cls_1": 0.77831, "loss_box_1": 1.4043, "loss_cns_1": 0.6682, "loss_yns_1": 0.13829, "loss_cls_2": 0.782, "loss_box_2": 1.37266, "loss_cns_2": 0.67108, "loss_yns_2": 0.13601, "loss_cls_3": 0.78696, "loss_box_3": 1.36942, "loss_cns_3": 0.66824, "loss_yns_3": 0.13587, "loss_cls_4": 0.7773, "loss_box_4": 1.37607, "loss_cns_4": 0.66847, "loss_yns_4": 0.13514, "loss_cls_5": 0.78056, "loss_box_5": 1.37749, "loss_cns_5": 0.66948, "loss_yns_5": 0.13654, "loss_cls_dn_0": 0.11631, "loss_box_dn_0": 0.70843, "loss_cls_dn_1": 0.09556, "loss_box_dn_1": 0.62, "loss_cls_dn_2": 0.09434, "loss_box_dn_2": 0.60212, "loss_cls_dn_3": 0.09089, "loss_box_dn_3": 0.59984, "loss_cls_dn_4": 0.09268, "loss_box_dn_4": 0.60113, "loss_cls_dn_5": 0.0931, "loss_box_dn_5": 0.60374, "loss_dense_depth": 0.69716, "loss": 22.8859, "grad_norm": 23.18479, "time": 1.59404} -{"mode": "train", "epoch": 1, "iter": 410, "lr": 0.00026, "memory": 49164, "data_time": 0.08827, "loss_cls_0": 0.70605, "loss_box_0": 1.53372, "loss_cns_0": 0.64431, "loss_yns_0": 0.1386, "loss_cls_1": 0.77167, "loss_box_1": 1.38203, "loss_cns_1": 0.66444, "loss_yns_1": 0.13666, "loss_cls_2": 0.7802, "loss_box_2": 1.35869, "loss_cns_2": 0.66758, "loss_yns_2": 0.13917, "loss_cls_3": 0.78502, "loss_box_3": 1.34203, "loss_cns_3": 0.66472, "loss_yns_3": 0.13508, "loss_cls_4": 0.77646, "loss_box_4": 1.35243, "loss_cns_4": 0.67051, "loss_yns_4": 0.13706, "loss_cls_5": 0.78421, "loss_box_5": 1.3506, "loss_cns_5": 0.66599, "loss_yns_5": 0.13768, "loss_cls_dn_0": 0.11097, "loss_box_dn_0": 0.70879, "loss_cls_dn_1": 0.09194, "loss_box_dn_1": 0.5996, "loss_cls_dn_2": 0.09205, "loss_box_dn_2": 0.58774, "loss_cls_dn_3": 0.08958, "loss_box_dn_3": 0.58167, "loss_cls_dn_4": 0.09094, "loss_box_dn_4": 0.58309, "loss_cls_dn_5": 0.09029, "loss_box_dn_5": 0.58378, "loss_dense_depth": 0.71204, "loss": 22.64739, "grad_norm": 32.17308, "time": 1.57235} -{"mode": "train", "epoch": 1, "iter": 411, "lr": 0.00026, "memory": 49164, "data_time": 0.07933, "loss_cls_0": 0.72452, "loss_box_0": 1.56527, "loss_cns_0": 0.64156, "loss_yns_0": 0.13706, "loss_cls_1": 0.77335, "loss_box_1": 1.38416, "loss_cns_1": 0.66135, "loss_yns_1": 0.13466, "loss_cls_2": 0.78404, "loss_box_2": 1.36702, "loss_cns_2": 0.66193, "loss_yns_2": 0.13663, "loss_cls_3": 0.79598, "loss_box_3": 1.34302, "loss_cns_3": 0.65837, "loss_yns_3": 0.13388, "loss_cls_4": 0.81719, "loss_box_4": 1.33854, "loss_cns_4": 0.65655, "loss_yns_4": 0.13394, "loss_cls_5": 0.82091, "loss_box_5": 1.33801, "loss_cns_5": 0.65746, "loss_yns_5": 0.13446, "loss_cls_dn_0": 0.11373, "loss_box_dn_0": 0.71365, "loss_cls_dn_1": 0.08854, "loss_box_dn_1": 0.6221, "loss_cls_dn_2": 0.08805, "loss_box_dn_2": 0.60993, "loss_cls_dn_3": 0.08788, "loss_box_dn_3": 0.6095, "loss_cls_dn_4": 0.08724, "loss_box_dn_4": 0.6149, "loss_cls_dn_5": 0.08938, "loss_box_dn_5": 0.61206, "loss_dense_depth": 0.73279, "loss": 22.8696, "grad_norm": 31.17189, "time": 1.60333} -{"mode": "train", "epoch": 1, "iter": 412, "lr": 0.00026, "memory": 49164, "data_time": 0.07952, "loss_cls_0": 0.70987, "loss_box_0": 1.56271, "loss_cns_0": 0.64665, "loss_yns_0": 0.13858, "loss_cls_1": 0.75909, "loss_box_1": 1.38802, "loss_cns_1": 0.6637, "loss_yns_1": 0.13255, "loss_cls_2": 0.77148, "loss_box_2": 1.35587, "loss_cns_2": 0.66556, "loss_yns_2": 0.13175, "loss_cls_3": 0.77656, "loss_box_3": 1.34328, "loss_cns_3": 0.66279, "loss_yns_3": 0.13206, "loss_cls_4": 0.77705, "loss_box_4": 1.33843, "loss_cns_4": 0.66213, "loss_yns_4": 0.133, "loss_cls_5": 0.77989, "loss_box_5": 1.33771, "loss_cns_5": 0.665, "loss_yns_5": 0.13251, "loss_cls_dn_0": 0.11498, "loss_box_dn_0": 0.72192, "loss_cls_dn_1": 0.08898, "loss_box_dn_1": 0.62455, "loss_cls_dn_2": 0.08846, "loss_box_dn_2": 0.608, "loss_cls_dn_3": 0.08684, "loss_box_dn_3": 0.60517, "loss_cls_dn_4": 0.08725, "loss_box_dn_4": 0.60313, "loss_cls_dn_5": 0.08684, "loss_box_dn_5": 0.60433, "loss_dense_depth": 0.68134, "loss": 22.668, "grad_norm": 27.73416, "time": 1.57445} -{"mode": "train", "epoch": 1, "iter": 413, "lr": 0.00026, "memory": 49164, "data_time": 0.07775, "loss_cls_0": 0.70907, "loss_box_0": 1.55167, "loss_cns_0": 0.64345, "loss_yns_0": 0.13949, "loss_cls_1": 0.76867, "loss_box_1": 1.38588, "loss_cns_1": 0.66457, "loss_yns_1": 0.13553, "loss_cls_2": 0.78, "loss_box_2": 1.37269, "loss_cns_2": 0.66894, "loss_yns_2": 0.13588, "loss_cls_3": 0.78594, "loss_box_3": 1.35867, "loss_cns_3": 0.66534, "loss_yns_3": 0.13536, "loss_cls_4": 0.7923, "loss_box_4": 1.36782, "loss_cns_4": 0.66551, "loss_yns_4": 0.13523, "loss_cls_5": 0.80646, "loss_box_5": 1.36122, "loss_cns_5": 0.66789, "loss_yns_5": 0.13414, "loss_cls_dn_0": 0.114, "loss_box_dn_0": 0.71368, "loss_cls_dn_1": 0.08874, "loss_box_dn_1": 0.61547, "loss_cls_dn_2": 0.08699, "loss_box_dn_2": 0.60123, "loss_cls_dn_3": 0.0851, "loss_box_dn_3": 0.59886, "loss_cls_dn_4": 0.08681, "loss_box_dn_4": 0.5995, "loss_cls_dn_5": 0.08766, "loss_box_dn_5": 0.60155, "loss_dense_depth": 0.69149, "loss": 22.8028, "grad_norm": 39.30933, "time": 1.57098} -{"mode": "train", "epoch": 1, "iter": 414, "lr": 0.00026, "memory": 49164, "data_time": 0.07565, "loss_cls_0": 0.7227, "loss_box_0": 1.58118, "loss_cns_0": 0.6435, "loss_yns_0": 0.14188, "loss_cls_1": 0.77665, "loss_box_1": 1.39809, "loss_cns_1": 0.66684, "loss_yns_1": 0.13656, "loss_cls_2": 0.78746, "loss_box_2": 1.35618, "loss_cns_2": 0.66291, "loss_yns_2": 0.13509, "loss_cls_3": 0.789, "loss_box_3": 1.35415, "loss_cns_3": 0.66376, "loss_yns_3": 0.13599, "loss_cls_4": 0.79625, "loss_box_4": 1.35631, "loss_cns_4": 0.6663, "loss_yns_4": 0.13711, "loss_cls_5": 0.80677, "loss_box_5": 1.35047, "loss_cns_5": 0.6659, "loss_yns_5": 0.13627, "loss_cls_dn_0": 0.11961, "loss_box_dn_0": 0.72209, "loss_cls_dn_1": 0.09349, "loss_box_dn_1": 0.62635, "loss_cls_dn_2": 0.09186, "loss_box_dn_2": 0.60795, "loss_cls_dn_3": 0.08876, "loss_box_dn_3": 0.60548, "loss_cls_dn_4": 0.08922, "loss_box_dn_4": 0.60557, "loss_cls_dn_5": 0.09124, "loss_box_dn_5": 0.60639, "loss_dense_depth": 0.68716, "loss": 22.90247, "grad_norm": 27.92273, "time": 1.60863} -{"mode": "train", "epoch": 1, "iter": 415, "lr": 0.00027, "memory": 49164, "data_time": 0.07458, "loss_cls_0": 0.71507, "loss_box_0": 1.50604, "loss_cns_0": 0.64457, "loss_yns_0": 0.13744, "loss_cls_1": 0.76342, "loss_box_1": 1.36457, "loss_cns_1": 0.66515, "loss_yns_1": 0.13816, "loss_cls_2": 0.77743, "loss_box_2": 1.34548, "loss_cns_2": 0.66408, "loss_yns_2": 0.13679, "loss_cls_3": 0.77998, "loss_box_3": 1.32078, "loss_cns_3": 0.66519, "loss_yns_3": 0.13862, "loss_cls_4": 0.78807, "loss_box_4": 1.33751, "loss_cns_4": 0.66654, "loss_yns_4": 0.13994, "loss_cls_5": 0.79597, "loss_box_5": 1.32216, "loss_cns_5": 0.66601, "loss_yns_5": 0.14099, "loss_cls_dn_0": 0.11689, "loss_box_dn_0": 0.71175, "loss_cls_dn_1": 0.09174, "loss_box_dn_1": 0.62093, "loss_cls_dn_2": 0.08924, "loss_box_dn_2": 0.60853, "loss_cls_dn_3": 0.08832, "loss_box_dn_3": 0.59961, "loss_cls_dn_4": 0.08752, "loss_box_dn_4": 0.60489, "loss_cls_dn_5": 0.08822, "loss_box_dn_5": 0.60209, "loss_dense_depth": 0.67314, "loss": 22.60284, "grad_norm": 32.05662, "time": 1.57262} -{"mode": "train", "epoch": 1, "iter": 416, "lr": 0.00027, "memory": 49164, "data_time": 0.07224, "loss_cls_0": 0.70348, "loss_box_0": 1.53775, "loss_cns_0": 0.64319, "loss_yns_0": 0.1389, "loss_cls_1": 0.74614, "loss_box_1": 1.38878, "loss_cns_1": 0.66727, "loss_yns_1": 0.13799, "loss_cls_2": 0.75396, "loss_box_2": 1.35326, "loss_cns_2": 0.66614, "loss_yns_2": 0.13676, "loss_cls_3": 0.76319, "loss_box_3": 1.3336, "loss_cns_3": 0.66476, "loss_yns_3": 0.13783, "loss_cls_4": 0.7682, "loss_box_4": 1.35207, "loss_cns_4": 0.66492, "loss_yns_4": 0.13923, "loss_cls_5": 0.77744, "loss_box_5": 1.33199, "loss_cns_5": 0.66528, "loss_yns_5": 0.13705, "loss_cls_dn_0": 0.11669, "loss_box_dn_0": 0.71567, "loss_cls_dn_1": 0.09116, "loss_box_dn_1": 0.6197, "loss_cls_dn_2": 0.09005, "loss_box_dn_2": 0.60223, "loss_cls_dn_3": 0.09257, "loss_box_dn_3": 0.59661, "loss_cls_dn_4": 0.09159, "loss_box_dn_4": 0.59766, "loss_cls_dn_5": 0.09562, "loss_box_dn_5": 0.59437, "loss_dense_depth": 0.67669, "loss": 22.58979, "grad_norm": 28.93233, "time": 1.55959} -{"mode": "train", "epoch": 1, "iter": 417, "lr": 0.00027, "memory": 49164, "data_time": 0.07116, "loss_cls_0": 0.73238, "loss_box_0": 1.54078, "loss_cns_0": 0.64277, "loss_yns_0": 0.14072, "loss_cls_1": 0.76661, "loss_box_1": 1.38965, "loss_cns_1": 0.66436, "loss_yns_1": 0.13818, "loss_cls_2": 0.77223, "loss_box_2": 1.36924, "loss_cns_2": 0.66565, "loss_yns_2": 0.13676, "loss_cls_3": 0.77623, "loss_box_3": 1.34064, "loss_cns_3": 0.65986, "loss_yns_3": 0.13642, "loss_cls_4": 0.77156, "loss_box_4": 1.36799, "loss_cns_4": 0.66253, "loss_yns_4": 0.13822, "loss_cls_5": 0.78429, "loss_box_5": 1.35821, "loss_cns_5": 0.66542, "loss_yns_5": 0.13778, "loss_cls_dn_0": 0.1182, "loss_box_dn_0": 0.71616, "loss_cls_dn_1": 0.09642, "loss_box_dn_1": 0.61803, "loss_cls_dn_2": 0.09504, "loss_box_dn_2": 0.60492, "loss_cls_dn_3": 0.09709, "loss_box_dn_3": 0.6006, "loss_cls_dn_4": 0.09466, "loss_box_dn_4": 0.60211, "loss_cls_dn_5": 0.0999, "loss_box_dn_5": 0.60211, "loss_dense_depth": 0.69187, "loss": 22.79556, "grad_norm": 30.89649, "time": 1.57279} -{"mode": "train", "epoch": 1, "iter": 418, "lr": 0.00027, "memory": 49164, "data_time": 0.07624, "loss_cls_0": 0.69672, "loss_box_0": 1.54349, "loss_cns_0": 0.64186, "loss_yns_0": 0.13998, "loss_cls_1": 0.75679, "loss_box_1": 1.37693, "loss_cns_1": 0.66597, "loss_yns_1": 0.13804, "loss_cls_2": 0.75913, "loss_box_2": 1.35623, "loss_cns_2": 0.66404, "loss_yns_2": 0.13886, "loss_cls_3": 0.77365, "loss_box_3": 1.34485, "loss_cns_3": 0.6607, "loss_yns_3": 0.13826, "loss_cls_4": 0.7778, "loss_box_4": 1.33827, "loss_cns_4": 0.66191, "loss_yns_4": 0.13886, "loss_cls_5": 0.78557, "loss_box_5": 1.33071, "loss_cns_5": 0.66236, "loss_yns_5": 0.13833, "loss_cls_dn_0": 0.10975, "loss_box_dn_0": 0.6989, "loss_cls_dn_1": 0.0926, "loss_box_dn_1": 0.61201, "loss_cls_dn_2": 0.09121, "loss_box_dn_2": 0.59965, "loss_cls_dn_3": 0.09151, "loss_box_dn_3": 0.59604, "loss_cls_dn_4": 0.09269, "loss_box_dn_4": 0.59449, "loss_cls_dn_5": 0.09337, "loss_box_dn_5": 0.59695, "loss_dense_depth": 0.68031, "loss": 22.57879, "grad_norm": 28.09821, "time": 1.58366} -{"mode": "train", "epoch": 1, "iter": 419, "lr": 0.00027, "memory": 49164, "data_time": 0.07362, "loss_cls_0": 0.75137, "loss_box_0": 1.58916, "loss_cns_0": 0.63218, "loss_yns_0": 0.13858, "loss_cls_1": 0.78119, "loss_box_1": 1.41615, "loss_cns_1": 0.66411, "loss_yns_1": 0.13669, "loss_cls_2": 0.78789, "loss_box_2": 1.39303, "loss_cns_2": 0.6624, "loss_yns_2": 0.13649, "loss_cls_3": 0.79718, "loss_box_3": 1.38036, "loss_cns_3": 0.66032, "loss_yns_3": 0.1367, "loss_cls_4": 0.80996, "loss_box_4": 1.38403, "loss_cns_4": 0.66095, "loss_yns_4": 0.13591, "loss_cls_5": 0.81375, "loss_box_5": 1.37495, "loss_cns_5": 0.65838, "loss_yns_5": 0.13502, "loss_cls_dn_0": 0.11869, "loss_box_dn_0": 0.70421, "loss_cls_dn_1": 0.09209, "loss_box_dn_1": 0.62184, "loss_cls_dn_2": 0.09109, "loss_box_dn_2": 0.6093, "loss_cls_dn_3": 0.09078, "loss_box_dn_3": 0.60707, "loss_cls_dn_4": 0.09139, "loss_box_dn_4": 0.611, "loss_cls_dn_5": 0.09179, "loss_box_dn_5": 0.61021, "loss_dense_depth": 0.74421, "loss": 23.1204, "grad_norm": 32.37903, "time": 1.58598} -{"mode": "train", "epoch": 1, "iter": 420, "lr": 0.00027, "memory": 49164, "data_time": 0.07817, "loss_cls_0": 0.72567, "loss_box_0": 1.58695, "loss_cns_0": 0.63442, "loss_yns_0": 0.13738, "loss_cls_1": 0.77766, "loss_box_1": 1.39928, "loss_cns_1": 0.66478, "loss_yns_1": 0.13418, "loss_cls_2": 0.78421, "loss_box_2": 1.36899, "loss_cns_2": 0.66534, "loss_yns_2": 0.13465, "loss_cls_3": 0.79107, "loss_box_3": 1.35294, "loss_cns_3": 0.66477, "loss_yns_3": 0.13409, "loss_cls_4": 0.79589, "loss_box_4": 1.3681, "loss_cns_4": 0.66449, "loss_yns_4": 0.13543, "loss_cls_5": 0.80378, "loss_box_5": 1.35345, "loss_cns_5": 0.66336, "loss_yns_5": 0.13387, "loss_cls_dn_0": 0.1176, "loss_box_dn_0": 0.71761, "loss_cls_dn_1": 0.09356, "loss_box_dn_1": 0.63302, "loss_cls_dn_2": 0.09262, "loss_box_dn_2": 0.61814, "loss_cls_dn_3": 0.09327, "loss_box_dn_3": 0.61462, "loss_cls_dn_4": 0.0933, "loss_box_dn_4": 0.62079, "loss_cls_dn_5": 0.09518, "loss_box_dn_5": 0.61906, "loss_dense_depth": 0.73347, "loss": 23.01696, "grad_norm": 38.63104, "time": 1.58831} -{"mode": "train", "epoch": 1, "iter": 421, "lr": 0.00027, "memory": 49164, "data_time": 0.10268, "loss_cls_0": 0.74089, "loss_box_0": 1.6072, "loss_cns_0": 0.63942, "loss_yns_0": 0.13861, "loss_cls_1": 0.78472, "loss_box_1": 1.4344, "loss_cns_1": 0.66584, "loss_yns_1": 0.13623, "loss_cls_2": 0.79016, "loss_box_2": 1.39586, "loss_cns_2": 0.66833, "loss_yns_2": 0.13592, "loss_cls_3": 0.7965, "loss_box_3": 1.39279, "loss_cns_3": 0.6662, "loss_yns_3": 0.13529, "loss_cls_4": 0.80184, "loss_box_4": 1.39016, "loss_cns_4": 0.66434, "loss_yns_4": 0.13726, "loss_cls_5": 0.80661, "loss_box_5": 1.38866, "loss_cns_5": 0.66616, "loss_yns_5": 0.13451, "loss_cls_dn_0": 0.11828, "loss_box_dn_0": 0.70956, "loss_cls_dn_1": 0.09285, "loss_box_dn_1": 0.65075, "loss_cls_dn_2": 0.09267, "loss_box_dn_2": 0.62941, "loss_cls_dn_3": 0.09198, "loss_box_dn_3": 0.62664, "loss_cls_dn_4": 0.09006, "loss_box_dn_4": 0.62187, "loss_cls_dn_5": 0.09332, "loss_box_dn_5": 0.62695, "loss_dense_depth": 0.68704, "loss": 23.24927, "grad_norm": 32.8775, "time": 1.64153} -{"mode": "train", "epoch": 1, "iter": 422, "lr": 0.00027, "memory": 49164, "data_time": 0.09924, "loss_cls_0": 0.70569, "loss_box_0": 1.61149, "loss_cns_0": 0.63601, "loss_yns_0": 0.13637, "loss_cls_1": 0.76622, "loss_box_1": 1.3852, "loss_cns_1": 0.66118, "loss_yns_1": 0.13328, "loss_cls_2": 0.77801, "loss_box_2": 1.3784, "loss_cns_2": 0.66196, "loss_yns_2": 0.13163, "loss_cls_3": 0.78184, "loss_box_3": 1.38351, "loss_cns_3": 0.66325, "loss_yns_3": 0.13105, "loss_cls_4": 0.77428, "loss_box_4": 1.38397, "loss_cns_4": 0.66417, "loss_yns_4": 0.13337, "loss_cls_5": 0.78641, "loss_box_5": 1.38272, "loss_cns_5": 0.66859, "loss_yns_5": 0.13267, "loss_cls_dn_0": 0.11314, "loss_box_dn_0": 0.7139, "loss_cls_dn_1": 0.09241, "loss_box_dn_1": 0.62163, "loss_cls_dn_2": 0.09008, "loss_box_dn_2": 0.61411, "loss_cls_dn_3": 0.09203, "loss_box_dn_3": 0.61305, "loss_cls_dn_4": 0.08831, "loss_box_dn_4": 0.61224, "loss_cls_dn_5": 0.09239, "loss_box_dn_5": 0.61291, "loss_dense_depth": 0.71355, "loss": 22.941, "grad_norm": 45.45131, "time": 1.63359} -{"mode": "train", "epoch": 1, "iter": 423, "lr": 0.00027, "memory": 49164, "data_time": 0.0853, "loss_cls_0": 0.70781, "loss_box_0": 1.57524, "loss_cns_0": 0.64182, "loss_yns_0": 0.14067, "loss_cls_1": 0.76189, "loss_box_1": 1.36105, "loss_cns_1": 0.66423, "loss_yns_1": 0.13431, "loss_cls_2": 0.7824, "loss_box_2": 1.33839, "loss_cns_2": 0.66534, "loss_yns_2": 0.13322, "loss_cls_3": 0.78081, "loss_box_3": 1.3523, "loss_cns_3": 0.66791, "loss_yns_3": 0.13302, "loss_cls_4": 0.77265, "loss_box_4": 1.35725, "loss_cns_4": 0.66892, "loss_yns_4": 0.13346, "loss_cls_5": 0.78085, "loss_box_5": 1.3652, "loss_cns_5": 0.6703, "loss_yns_5": 0.13379, "loss_cls_dn_0": 0.11313, "loss_box_dn_0": 0.70891, "loss_cls_dn_1": 0.09413, "loss_box_dn_1": 0.61554, "loss_cls_dn_2": 0.09123, "loss_box_dn_2": 0.60536, "loss_cls_dn_3": 0.09395, "loss_box_dn_3": 0.60459, "loss_cls_dn_4": 0.09149, "loss_box_dn_4": 0.6068, "loss_cls_dn_5": 0.09244, "loss_box_dn_5": 0.61224, "loss_dense_depth": 0.6785, "loss": 22.7311, "grad_norm": 39.77167, "time": 1.58806} -{"mode": "train", "epoch": 1, "iter": 424, "lr": 0.00027, "memory": 49164, "data_time": 0.0816, "loss_cls_0": 0.75535, "loss_box_0": 1.61862, "loss_cns_0": 0.64047, "loss_yns_0": 0.14, "loss_cls_1": 0.77779, "loss_box_1": 1.43019, "loss_cns_1": 0.66438, "loss_yns_1": 0.1328, "loss_cls_2": 0.79373, "loss_box_2": 1.39741, "loss_cns_2": 0.66758, "loss_yns_2": 0.13181, "loss_cls_3": 0.79365, "loss_box_3": 1.41382, "loss_cns_3": 0.66858, "loss_yns_3": 0.1324, "loss_cls_4": 0.802, "loss_box_4": 1.41386, "loss_cns_4": 0.67067, "loss_yns_4": 0.13328, "loss_cls_5": 0.81535, "loss_box_5": 1.40937, "loss_cns_5": 0.67056, "loss_yns_5": 0.13175, "loss_cls_dn_0": 0.12578, "loss_box_dn_0": 0.71718, "loss_cls_dn_1": 0.09486, "loss_box_dn_1": 0.64619, "loss_cls_dn_2": 0.09494, "loss_box_dn_2": 0.63724, "loss_cls_dn_3": 0.09325, "loss_box_dn_3": 0.63574, "loss_cls_dn_4": 0.09458, "loss_box_dn_4": 0.63199, "loss_cls_dn_5": 0.09517, "loss_box_dn_5": 0.635, "loss_dense_depth": 0.71125, "loss": 23.4186, "grad_norm": 43.89584, "time": 1.60941} -{"mode": "train", "epoch": 1, "iter": 425, "lr": 0.00027, "memory": 49164, "data_time": 0.07862, "loss_cls_0": 0.72861, "loss_box_0": 1.59378, "loss_cns_0": 0.64713, "loss_yns_0": 0.13767, "loss_cls_1": 0.78314, "loss_box_1": 1.42644, "loss_cns_1": 0.66896, "loss_yns_1": 0.13047, "loss_cls_2": 0.78392, "loss_box_2": 1.39833, "loss_cns_2": 0.66973, "loss_yns_2": 0.13057, "loss_cls_3": 0.77641, "loss_box_3": 1.39229, "loss_cns_3": 0.66877, "loss_yns_3": 0.13088, "loss_cls_4": 0.78146, "loss_box_4": 1.38085, "loss_cns_4": 0.66817, "loss_yns_4": 0.13172, "loss_cls_5": 0.7909, "loss_box_5": 1.37161, "loss_cns_5": 0.67055, "loss_yns_5": 0.12969, "loss_cls_dn_0": 0.11785, "loss_box_dn_0": 0.71629, "loss_cls_dn_1": 0.09804, "loss_box_dn_1": 0.645, "loss_cls_dn_2": 0.10007, "loss_box_dn_2": 0.63544, "loss_cls_dn_3": 0.10041, "loss_box_dn_3": 0.62997, "loss_cls_dn_4": 0.10133, "loss_box_dn_4": 0.62084, "loss_cls_dn_5": 0.10202, "loss_box_dn_5": 0.62042, "loss_dense_depth": 0.68703, "loss": 23.16678, "grad_norm": 35.18084, "time": 1.56452} -{"mode": "train", "epoch": 1, "iter": 426, "lr": 0.00027, "memory": 49164, "data_time": 0.10053, "loss_cls_0": 0.7185, "loss_box_0": 1.54792, "loss_cns_0": 0.64483, "loss_yns_0": 0.13813, "loss_cls_1": 0.77392, "loss_box_1": 1.4158, "loss_cns_1": 0.66655, "loss_yns_1": 0.13419, "loss_cls_2": 0.78922, "loss_box_2": 1.37712, "loss_cns_2": 0.66452, "loss_yns_2": 0.13266, "loss_cls_3": 0.78971, "loss_box_3": 1.38927, "loss_cns_3": 0.66314, "loss_yns_3": 0.13483, "loss_cls_4": 0.79227, "loss_box_4": 1.37481, "loss_cns_4": 0.66424, "loss_yns_4": 0.13264, "loss_cls_5": 0.81049, "loss_box_5": 1.3806, "loss_cns_5": 0.66228, "loss_yns_5": 0.13156, "loss_cls_dn_0": 0.1089, "loss_box_dn_0": 0.71028, "loss_cls_dn_1": 0.09631, "loss_box_dn_1": 0.63547, "loss_cls_dn_2": 0.09969, "loss_box_dn_2": 0.61812, "loss_cls_dn_3": 0.10036, "loss_box_dn_3": 0.61785, "loss_cls_dn_4": 0.09795, "loss_box_dn_4": 0.61543, "loss_cls_dn_5": 0.10088, "loss_box_dn_5": 0.62321, "loss_dense_depth": 0.6526, "loss": 23.00624, "grad_norm": 42.51623, "time": 1.57643} -{"mode": "train", "epoch": 1, "iter": 427, "lr": 0.00027, "memory": 49164, "data_time": 0.07214, "loss_cls_0": 0.72035, "loss_box_0": 1.56196, "loss_cns_0": 0.63828, "loss_yns_0": 0.13834, "loss_cls_1": 0.7675, "loss_box_1": 1.40206, "loss_cns_1": 0.66367, "loss_yns_1": 0.13566, "loss_cls_2": 0.7844, "loss_box_2": 1.37075, "loss_cns_2": 0.66162, "loss_yns_2": 0.13559, "loss_cls_3": 0.78436, "loss_box_3": 1.3757, "loss_cns_3": 0.66312, "loss_yns_3": 0.13591, "loss_cls_4": 0.79454, "loss_box_4": 1.36104, "loss_cns_4": 0.66287, "loss_yns_4": 0.13415, "loss_cls_5": 0.80869, "loss_box_5": 1.35105, "loss_cns_5": 0.65811, "loss_yns_5": 0.13429, "loss_cls_dn_0": 0.11433, "loss_box_dn_0": 0.707, "loss_cls_dn_1": 0.09601, "loss_box_dn_1": 0.62737, "loss_cls_dn_2": 0.09798, "loss_box_dn_2": 0.6121, "loss_cls_dn_3": 0.09795, "loss_box_dn_3": 0.61829, "loss_cls_dn_4": 0.09363, "loss_box_dn_4": 0.62137, "loss_cls_dn_5": 0.09633, "loss_box_dn_5": 0.62584, "loss_dense_depth": 0.68083, "loss": 22.93301, "grad_norm": 36.85945, "time": 1.57919} diff --git a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20251112_151550.log b/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20251112_151550.log deleted file mode 100644 index 4a5407e44ccdf0ae8fdd20abf69f562843f26dcf..0000000000000000000000000000000000000000 --- a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20251112_151550.log +++ /dev/null @@ -1,3220 +0,0 @@ -2025-11-12 15:15:50,982 - mmdet - INFO - Environment info: ------------------------------------------------------------- -sys.platform: linux -Python: 3.10.12 (main, May 27 2025, 17:12:29) [GCC 11.4.0] -CUDA available: True -GPU 0,1,2,3,4,5,6,7: BW200, UBB BW1000 -CUDA_HOME: /opt/dtk -NVCC: Not Available -GCC: x86_64-linux-gnu-gcc (Ubuntu 11.4.0-1ubuntu1~22.04) 11.4.0 -PyTorch: 2.4.1 -PyTorch compiling details: PyTorch built with: - - GCC 10.3 - - C++ Version: 201703 - - Intel(R) Math Kernel Library Version 2020.0.4 Product Build 20200917 for Intel(R) 64 architecture applications - - OpenMP 201511 (a.k.a. OpenMP 4.5) - - LAPACK is enabled (usually provided by MKL) - - NNPACK is enabled - - CPU capability usage: AVX512 - - HIP Runtime 6.3.25211 - - MIOpen 2.17.0 - - Magma 2.8.0 - - Build settings: BLAS_INFO=mkl, BUILD_TYPE=Release, CXX_COMPILER=/opt/rh/gcc-toolset-10/root/usr/bin/c++, CXX_FLAGS= -D_GLIBCXX_USE_CXX11_ABI=1 -fvisibility-inlines-hidden -DUSE_PTHREADPOOL -DNDEBUG -DUSE_KINETO -DLIBKINETO_NOCUPTI -DUSE_FBGEMM -DUSE_PYTORCH_QNNPACK -DUSE_XNNPACK -DSYMBOLICATE_MOBILE_DEBUG_HANDLE -O2 -fPIC -Wall -Wextra -Werror=return-type -Werror=non-virtual-dtor -Werror=bool-operation -Wnarrowing -Wno-missing-field-initializers -Wno-type-limits -Wno-array-bounds -Wno-unknown-pragmas -Wno-unused-parameter -Wno-unused-function -Wno-unused-result -Wno-strict-overflow -Wno-strict-aliasing -Wno-stringop-overflow -Wsuggest-override -Wno-psabi -Wno-error=pedantic -Wno-error=old-style-cast -Wno-missing-braces -fdiagnostics-color=always -faligned-new -Wno-unused-but-set-variable -Wno-maybe-uninitialized -fno-math-errno -fno-trapping-math -Werror=format -Wno-stringop-overflow, FORCE_FALLBACK_CUDA_MPI=1, LAPACK_INFO=mkl, PERF_WITH_AVX=1, PERF_WITH_AVX2=1, PERF_WITH_AVX512=1, TORCH_VERSION=2.4.1, USE_CUDA=0, USE_CUDNN=OFF, USE_CUSPARSELT=OFF, USE_EXCEPTION_PTR=1, USE_GFLAGS=1, USE_GLOG=1, USE_GLOO=1, USE_MKL=ON, USE_MKLDNN=0, USE_MPI=1, USE_NCCL=1, USE_NNPACK=ON, USE_OPENMP=1, USE_ROCM=ON, USE_ROCM_KERNEL_ASSERT=OFF, - -TorchVision: 0.19.1 -OpenCV: 4.11.0 -MMCV: 1.6.1 -MMCV Compiler: GCC 11.4 -MMCV CUDA Compiler: rocm not available -MMDetection: 2.26.0+c41df4b ------------------------------------------------------------- - -2025-11-12 15:15:51,912 - mmdet - INFO - Distributed training: True -2025-11-12 15:15:52,636 - mmdet - INFO - Config: -plugin = True -plugin_dir = 'projects/mmdet3d_plugin/' -dist_params = dict(backend='nccl') -log_level = 'INFO' -work_dir = './work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704' -total_batch_size = 160 -num_gpus = 8 -batch_size = 20 -num_iters_per_epoch = 175 -num_epochs = 100 -checkpoint_epoch_interval = 20 -checkpoint_config = dict(interval=3500) -log_config = dict( - interval=1, - hooks=[ - dict(type='TextLoggerHook', by_epoch=False), - dict(type='TensorboardLoggerHook') - ]) -load_from = None -resume_from = None -workflow = [('train', 1)] -fp16 = dict(loss_scale=32.0) -input_shape = (704, 256) -tracking_test = True -tracking_threshold = 0.2 -class_names = [ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier', - 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' -] -num_classes = 10 -embed_dims = 256 -num_groups = 8 -num_decoder = 6 -num_single_frame_decoder = 1 -use_deformable_func = True -strides = [4, 8, 16, 32] -num_levels = 4 -num_depth_layers = 3 -drop_out = 0.1 -temporal = True -decouple_attn = True -with_quality_estimation = True -model = dict( - type='Sparse4D', - use_grid_mask=True, - use_deformable_func=True, - img_backbone=dict( - type='ResNet', - depth=50, - num_stages=4, - frozen_stages=-1, - norm_eval=False, - style='pytorch', - with_cp=True, - out_indices=(0, 1, 2, 3), - norm_cfg=dict(type='BN', requires_grad=True), - pretrained='ckpt/resnet50-19c8e357.pth'), - img_neck=dict( - type='FPN', - num_outs=4, - start_level=0, - out_channels=256, - add_extra_convs='on_output', - relu_before_extra_convs=True, - in_channels=[256, 512, 1024, 2048]), - depth_branch=dict( - type='DenseDepthNet', - embed_dims=256, - num_depth_layers=3, - loss_weight=0.2), - head=dict( - type='Sparse4DHead', - cls_threshold_to_reg=0.05, - decouple_attn=True, - instance_bank=dict( - type='InstanceBank', - num_anchor=900, - embed_dims=256, - anchor='nuscenes_kmeans900.npy', - anchor_handler=dict(type='SparseBox3DKeyPointsGenerator'), - num_temp_instances=600, - confidence_decay=0.6, - feat_grad=False), - anchor_encoder=dict( - type='SparseBox3DEncoder', - vel_dims=3, - embed_dims=[128, 32, 32, 64], - mode='cat', - output_fc=False, - in_loops=1, - out_loops=4), - num_single_frame_decoder=1, - operation_order=[ - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine' - ], - temp_graph_model=dict( - type='MultiheadAttention', - embed_dims=512, - num_heads=8, - batch_first=True, - dropout=0.1), - graph_model=dict( - type='MultiheadAttention', - embed_dims=512, - num_heads=8, - batch_first=True, - dropout=0.1), - norm_layer=dict(type='LN', normalized_shape=256), - ffn=dict( - type='AsymmetricFFN', - in_channels=512, - pre_norm=dict(type='LN'), - embed_dims=256, - feedforward_channels=1024, - num_fcs=2, - ffn_drop=0.1, - act_cfg=dict(type='ReLU', inplace=True)), - deformable_model=dict( - type='DeformableFeatureAggregation', - embed_dims=256, - num_groups=8, - num_levels=4, - num_cams=6, - attn_drop=0.15, - use_deformable_func=True, - use_camera_embed=True, - residual_mode='cat', - kps_generator=dict( - type='SparseBox3DKeyPointsGenerator', - num_learnable_pts=6, - fix_scale=[[0, 0, 0], [0.45, 0, 0], [-0.45, 0, 0], - [0, 0.45, 0], [0, -0.45, 0], [0, 0, 0.45], - [0, 0, -0.45]])), - refine_layer=dict( - type='SparseBox3DRefinementModule', - embed_dims=256, - num_cls=10, - refine_yaw=True, - with_quality_estimation=True), - sampler=dict( - type='SparseBox3DTarget', - num_dn_groups=5, - num_temp_dn_groups=3, - dn_noise_scale=[2.0, 2.0, 2.0, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5], - max_dn_gt=32, - add_neg_dn=True, - cls_weight=2.0, - box_weight=0.25, - reg_weights=[2.0, 2.0, 2.0, 0.5, 0.5, 0.5, 0.0, 0.0, 0.0, 0.0], - cls_wise_reg_weights=dict( - {9: [2.0, 2.0, 2.0, 1.0, 1.0, 1.0, 0.0, 0.0, 1.0, 1.0]})), - loss_cls=dict( - type='FocalLoss', - use_sigmoid=True, - gamma=2.0, - alpha=0.25, - loss_weight=2.0), - loss_reg=dict( - type='SparseBox3DLoss', - loss_box=dict(type='L1Loss', loss_weight=0.25), - loss_centerness=dict(type='CrossEntropyLoss', use_sigmoid=True), - loss_yawness=dict(type='GaussianFocalLoss'), - cls_allow_reverse=[5]), - decoder=dict(type='SparseBox3DDecoder'), - reg_weights=[2.0, 2.0, 2.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0])) -dataset_type = 'NuScenes3DDetTrackDataset' -data_root = 'data/nuscenes/' -anno_root = 'data/nuscenes_anno_pkls/' -file_client_args = dict(backend='disk') -img_norm_cfg = dict( - mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375], to_rgb=True) -train_pipeline = [ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict( - type='LoadPointsFromFile', - coord_type='LIDAR', - load_dim=5, - use_dim=5, - file_client_args=dict(backend='disk')), - dict(type='ResizeCropFlipImage'), - dict(type='MultiScaleDepthMapGenerator', downsample=[4, 8, 16]), - dict(type='BBoxRotation'), - dict(type='PhotoMetricDistortionMultiViewImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict( - type='CircleObjectRangeFilter', - class_dist_thred=[55, 55, 55, 55, 55, 55, 55, 55, 55, 55]), - dict( - type='InstanceNameFilter', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ]), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=[ - 'img', 'timestamp', 'projection_mat', 'image_wh', 'gt_depth', - 'focal', 'gt_bboxes_3d', 'gt_labels_3d' - ], - meta_keys=['T_global', 'T_global_inv', 'timestamp', 'instance_id']) -] -test_pipeline = [ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='ResizeCropFlipImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=['img', 'timestamp', 'projection_mat', 'image_wh'], - meta_keys=['T_global', 'T_global_inv', 'timestamp']) -] -input_modality = dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False) -data_basic_config = dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier', - 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval') -data_aug_conf = dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]) -data = dict( - samples_per_gpu=20, - workers_per_gpu=20, - train=dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval', - ann_file='data/nuscenes_anno_pkls/nuscenes_infos_train.pkl', - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict( - type='LoadPointsFromFile', - coord_type='LIDAR', - load_dim=5, - use_dim=5, - file_client_args=dict(backend='disk')), - dict(type='ResizeCropFlipImage'), - dict(type='MultiScaleDepthMapGenerator', downsample=[4, 8, 16]), - dict(type='BBoxRotation'), - dict(type='PhotoMetricDistortionMultiViewImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict( - type='CircleObjectRangeFilter', - class_dist_thred=[55, 55, 55, 55, 55, 55, 55, 55, 55, 55]), - dict( - type='InstanceNameFilter', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', - 'traffic_cone' - ]), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=[ - 'img', 'timestamp', 'projection_mat', 'image_wh', - 'gt_depth', 'focal', 'gt_bboxes_3d', 'gt_labels_3d' - ], - meta_keys=[ - 'T_global', 'T_global_inv', 'timestamp', 'instance_id' - ]) - ], - test_mode=False, - data_aug_conf=dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]), - with_seq_flag=True, - sequences_split_num=2, - keep_consistent_seq_aug=True), - val=dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval', - ann_file='data/nuscenes_anno_pkls/nuscenes_infos_val.pkl', - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='ResizeCropFlipImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=['img', 'timestamp', 'projection_mat', 'image_wh'], - meta_keys=['T_global', 'T_global_inv', 'timestamp']) - ], - data_aug_conf=dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]), - test_mode=True, - tracking=True, - tracking_threshold=0.2), - test=dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval', - ann_file='data/nuscenes_anno_pkls/nuscenes_infos_val.pkl', - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='ResizeCropFlipImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=['img', 'timestamp', 'projection_mat', 'image_wh'], - meta_keys=['T_global', 'T_global_inv', 'timestamp']) - ], - data_aug_conf=dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]), - test_mode=True, - tracking=True, - tracking_threshold=0.2)) -optimizer = dict( - type='Miopen_AdamW', - lr=0.0006, - weight_decay=0.001, - paramwise_cfg=dict(custom_keys=dict(img_backbone=dict(lr_mult=0.5)))) -optimizer_config = dict(grad_clip=dict(max_norm=25, norm_type=2)) -lr_config = dict( - policy='CosineAnnealing', - warmup='linear', - warmup_iters=500, - warmup_ratio=0.3333333333333333, - min_lr_ratio=0.001) -runner = dict(type='IterBasedRunner', max_iters=17500) -vis_pipeline = [ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='Collect', keys=['img'], meta_keys=['timestamp', 'lidar2img']) -] -evaluation = dict( - interval=3500, - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict( - type='Collect', keys=['img'], meta_keys=['timestamp', 'lidar2img']) - ]) -gpu_ids = range(0, 8) - -2025-11-12 15:15:52,637 - mmdet - INFO - Set random seed to 0, deterministic: False -2025-11-12 15:15:52,933 - mmdet - INFO - initialize ResNet with init_cfg {'type': 'Pretrained', 'checkpoint': 'ckpt/resnet50-19c8e357.pth'} -2025-11-12 15:15:53,143 - mmdet - INFO - initialize FPN with init_cfg {'type': 'Xavier', 'layer': 'Conv2d', 'distribution': 'uniform'} -Name of parameter - Initialization information - -img_backbone.conv1.weight - torch.Size([64, 3, 7, 7]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.bn1.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.bn1.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.conv1.weight - torch.Size([64, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn1.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn1.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.conv2.weight - torch.Size([64, 64, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn2.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn2.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.conv3.weight - torch.Size([256, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn3.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn3.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.downsample.0.weight - torch.Size([256, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.downsample.1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.downsample.1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.conv1.weight - torch.Size([64, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn1.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn1.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.conv2.weight - torch.Size([64, 64, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn2.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn2.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.conv3.weight - torch.Size([256, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn3.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn3.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.conv1.weight - torch.Size([64, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn1.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn1.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.conv2.weight - torch.Size([64, 64, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn2.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn2.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.conv3.weight - torch.Size([256, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn3.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn3.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.conv1.weight - torch.Size([128, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn1.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn1.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.conv2.weight - torch.Size([128, 128, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn2.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn2.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.conv3.weight - torch.Size([512, 128, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn3.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn3.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.downsample.0.weight - torch.Size([512, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.downsample.1.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.downsample.1.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.conv1.weight - torch.Size([128, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn1.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn1.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.conv2.weight - torch.Size([128, 128, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn2.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn2.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.conv3.weight - torch.Size([512, 128, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn3.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn3.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.conv1.weight - torch.Size([128, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn1.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn1.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.conv2.weight - torch.Size([128, 128, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn2.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn2.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.conv3.weight - torch.Size([512, 128, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn3.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn3.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.conv1.weight - torch.Size([128, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn1.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn1.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.conv2.weight - torch.Size([128, 128, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn2.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn2.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.conv3.weight - torch.Size([512, 128, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn3.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn3.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.conv1.weight - torch.Size([256, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.downsample.0.weight - torch.Size([1024, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.downsample.1.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.downsample.1.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.conv1.weight - torch.Size([512, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn1.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn1.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.conv2.weight - torch.Size([512, 512, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn2.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn2.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.conv3.weight - torch.Size([2048, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn3.weight - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn3.bias - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.downsample.0.weight - torch.Size([2048, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.downsample.1.weight - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.downsample.1.bias - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.conv1.weight - torch.Size([512, 2048, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn1.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn1.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.conv2.weight - torch.Size([512, 512, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn2.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn2.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.conv3.weight - torch.Size([2048, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn3.weight - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn3.bias - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.conv1.weight - torch.Size([512, 2048, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn1.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn1.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.conv2.weight - torch.Size([512, 512, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn2.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn2.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.conv3.weight - torch.Size([2048, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn3.weight - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn3.bias - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_neck.lateral_convs.0.conv.weight - torch.Size([256, 256, 1, 1]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.lateral_convs.0.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.lateral_convs.1.conv.weight - torch.Size([256, 512, 1, 1]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.lateral_convs.1.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.lateral_convs.2.conv.weight - torch.Size([256, 1024, 1, 1]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.lateral_convs.2.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.lateral_convs.3.conv.weight - torch.Size([256, 2048, 1, 1]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.lateral_convs.3.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.fpn_convs.0.conv.weight - torch.Size([256, 256, 3, 3]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.fpn_convs.0.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.fpn_convs.1.conv.weight - torch.Size([256, 256, 3, 3]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.fpn_convs.1.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.fpn_convs.2.conv.weight - torch.Size([256, 256, 3, 3]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.fpn_convs.2.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.fpn_convs.3.conv.weight - torch.Size([256, 256, 3, 3]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.fpn_convs.3.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.instance_bank.anchor - torch.Size([900, 11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.instance_bank.instance_feature - torch.Size([900, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.instance_bank.anchor_handler.fix_scale - torch.Size([1, 3]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.0.weight - torch.Size([128, 3]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.0.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.2.weight - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.2.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.3.weight - torch.Size([128, 128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.3.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.5.weight - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.5.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.6.weight - torch.Size([128, 128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.6.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.8.weight - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.8.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.9.weight - torch.Size([128, 128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.9.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.11.weight - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.11.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.0.weight - torch.Size([32, 3]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.0.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.2.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.2.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.3.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.3.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.5.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.5.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.6.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.6.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.8.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.8.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.9.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.9.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.11.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.11.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.0.weight - torch.Size([32, 2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.0.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.2.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.2.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.3.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.3.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.5.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.5.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.6.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.6.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.8.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.8.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.9.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.9.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.11.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.11.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.0.weight - torch.Size([64, 3]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.0.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.2.weight - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.2.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.3.weight - torch.Size([64, 64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.3.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.5.weight - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.5.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.6.weight - torch.Size([64, 64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.6.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.8.weight - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.8.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.9.weight - torch.Size([64, 64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.9.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.11.weight - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.11.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.1.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.1.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.1.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.1.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.1.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.1.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.1.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.1.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.3.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.4.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.4.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.4.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.4.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.5.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.5.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.5.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.5.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.6.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.6.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.8.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.8.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.8.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.8.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.8.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.8.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.8.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.8.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.10.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.11.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.11.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.11.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.11.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.12.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.12.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.12.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.12.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.13.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.13.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.15.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.15.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.15.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.15.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.15.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.15.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.15.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.15.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.16.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.16.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.17.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.18.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.18.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.18.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.18.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.19.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.19.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.19.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.19.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.20.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.20.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.22.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.22.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.22.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.22.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.22.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.22.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.22.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.22.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.23.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.23.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.24.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.25.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.25.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.25.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.25.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.26.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.26.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.26.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.26.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.27.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.27.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.29.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.29.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.29.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.29.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.29.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.29.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.29.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.29.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.30.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.30.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.31.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.32.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.32.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.32.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.32.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.33.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.33.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.33.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.33.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.34.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.34.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.36.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.36.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.36.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.36.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.36.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.36.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.36.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.36.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.37.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.37.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.38.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.fc_before.weight - torch.Size([512, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.fc_after.weight - torch.Size([256, 512]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.0.weight - torch.Size([1, 256, 1, 1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.0.bias - torch.Size([1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.1.weight - torch.Size([1, 256, 1, 1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.1.bias - torch.Size([1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.2.weight - torch.Size([1, 256, 1, 1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.2.bias - torch.Size([1]): -The value is the same before and after calling `init_weights` of Sparse4D -2025-11-12 15:15:53,233 - mmdet - INFO - Model: -Sparse4D( - (img_backbone): ResNet( - (conv1): Conv2d(3, 64, kernel_size=(7, 7), stride=(2, 2), padding=(3, 3), bias=False) - (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (maxpool): MaxPool2d(kernel_size=3, stride=2, padding=1, dilation=1, ceil_mode=False) - (layer1): ResLayer( - (0): Bottleneck( - (conv1): Conv2d(64, 64, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (downsample): Sequential( - (0): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - ) - ) - (1): Bottleneck( - (conv1): Conv2d(256, 64, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (2): Bottleneck( - (conv1): Conv2d(256, 64, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - ) - (layer2): ResLayer( - (0): Bottleneck( - (conv1): Conv2d(256, 128, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (downsample): Sequential( - (0): Conv2d(256, 512, kernel_size=(1, 1), stride=(2, 2), bias=False) - (1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - ) - ) - (1): Bottleneck( - (conv1): Conv2d(512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (2): Bottleneck( - (conv1): Conv2d(512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (3): Bottleneck( - (conv1): Conv2d(512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - ) - (layer3): ResLayer( - (0): Bottleneck( - (conv1): Conv2d(512, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (downsample): Sequential( - (0): Conv2d(512, 1024, kernel_size=(1, 1), stride=(2, 2), bias=False) - (1): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - ) - ) - (1): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (2): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (3): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (4): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (5): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - ) - (layer4): ResLayer( - (0): Bottleneck( - (conv1): Conv2d(1024, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(512, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (downsample): Sequential( - (0): Conv2d(1024, 2048, kernel_size=(1, 1), stride=(2, 2), bias=False) - (1): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - ) - ) - (1): Bottleneck( - (conv1): Conv2d(2048, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(512, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (2): Bottleneck( - (conv1): Conv2d(2048, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(512, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - ) - ) - init_cfg={'type': 'Pretrained', 'checkpoint': 'ckpt/resnet50-19c8e357.pth'} - (img_neck): FPN( - (lateral_convs): ModuleList( - (0): ConvModule( - (conv): Conv2d(256, 256, kernel_size=(1, 1), stride=(1, 1)) - ) - (1): ConvModule( - (conv): Conv2d(512, 256, kernel_size=(1, 1), stride=(1, 1)) - ) - (2): ConvModule( - (conv): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1)) - ) - (3): ConvModule( - (conv): Conv2d(2048, 256, kernel_size=(1, 1), stride=(1, 1)) - ) - ) - (fpn_convs): ModuleList( - (0-3): 4 x ConvModule( - (conv): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1)) - ) - ) - ) - init_cfg={'type': 'Xavier', 'layer': 'Conv2d', 'distribution': 'uniform'} - (head): Sparse4DHead( - (instance_bank): InstanceBank( - (anchor_handler): SparseBox3DKeyPointsGenerator() - ) - (anchor_encoder): SparseBox3DEncoder( - (pos_fc): Sequential( - (0): Linear(in_features=3, out_features=128, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((128,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=128, out_features=128, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((128,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=128, out_features=128, bias=True) - (7): ReLU(inplace=True) - (8): LayerNorm((128,), eps=1e-05, elementwise_affine=True) - (9): Linear(in_features=128, out_features=128, bias=True) - (10): ReLU(inplace=True) - (11): LayerNorm((128,), eps=1e-05, elementwise_affine=True) - ) - (size_fc): Sequential( - (0): Linear(in_features=3, out_features=32, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=32, out_features=32, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=32, out_features=32, bias=True) - (7): ReLU(inplace=True) - (8): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (9): Linear(in_features=32, out_features=32, bias=True) - (10): ReLU(inplace=True) - (11): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - ) - (yaw_fc): Sequential( - (0): Linear(in_features=2, out_features=32, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=32, out_features=32, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=32, out_features=32, bias=True) - (7): ReLU(inplace=True) - (8): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (9): Linear(in_features=32, out_features=32, bias=True) - (10): ReLU(inplace=True) - (11): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - ) - (vel_fc): Sequential( - (0): Linear(in_features=3, out_features=64, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((64,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=64, out_features=64, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((64,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=64, out_features=64, bias=True) - (7): ReLU(inplace=True) - (8): LayerNorm((64,), eps=1e-05, elementwise_affine=True) - (9): Linear(in_features=64, out_features=64, bias=True) - (10): ReLU(inplace=True) - (11): LayerNorm((64,), eps=1e-05, elementwise_affine=True) - ) - ) - (loss_cls): FocalLoss() - (loss_reg): SparseBox3DLoss( - (loss_box): L1Loss() - (loss_cns): CrossEntropyLoss(avg_non_ignore=False) - (loss_yns): GaussianFocalLoss() - ) - (layers): ModuleList( - (0): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (1): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (4-5): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (6): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (7): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (8): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (11-12): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (13): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (14): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (15): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (16): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (17): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (18-19): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (20): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (21): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (22): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (23): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (24): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (25-26): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (27): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (28): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (29): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (30): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (31): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (32-33): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (34): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (35): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (36): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (37): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (38): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - ) - (fc_before): Linear(in_features=256, out_features=512, bias=False) - (fc_after): Linear(in_features=512, out_features=256, bias=False) - ) - (depth_branch): DenseDepthNet( - (depth_layers): ModuleList( - (0-2): 3 x Conv2d(256, 1, kernel_size=(1, 1), stride=(1, 1)) - ) - ) - (grid_mask): GridMask() -) -2025-11-12 15:16:05,726 - mmdet - INFO - Start running, host: root@VM-120-96-tencentos, work_dir: /cfs/auto/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704 -2025-11-12 15:16:05,726 - mmdet - INFO - Hooks will be executed in the following order: -before_run: -(VERY_HIGH ) CosineAnnealingLrUpdaterHook -(ABOVE_NORMAL) Fp16OptimizerHook -(NORMAL ) CheckpointHook -(NORMAL ) CustomDistEvalHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -before_train_epoch: -(VERY_HIGH ) CosineAnnealingLrUpdaterHook -(NORMAL ) CustomDistEvalHook -(LOW ) IterTimerHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -before_train_iter: -(VERY_HIGH ) CosineAnnealingLrUpdaterHook -(NORMAL ) CustomDistEvalHook -(LOW ) IterTimerHook - -------------------- -after_train_iter: -(ABOVE_NORMAL) Fp16OptimizerHook -(NORMAL ) CheckpointHook -(NORMAL ) CustomDistEvalHook -(LOW ) IterTimerHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -after_train_epoch: -(NORMAL ) CheckpointHook -(NORMAL ) CustomDistEvalHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -before_val_epoch: -(LOW ) IterTimerHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -before_val_iter: -(LOW ) IterTimerHook - -------------------- -after_val_iter: -(LOW ) IterTimerHook - -------------------- -after_val_epoch: -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -after_run: -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -2025-11-12 15:16:05,727 - mmdet - INFO - workflow: [('train', 1)], max: 17500 iters -2025-11-12 15:16:05,729 - mmdet - INFO - Checkpoints will be saved to /cfs/auto/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704 by HardDiskBackend. diff --git a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20251112_151550.log.json b/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20251112_151550.log.json deleted file mode 100644 index 1fda3e213098b4a1f6239f0e36f9459d0a88f9c1..0000000000000000000000000000000000000000 --- a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20251112_151550.log.json +++ /dev/null @@ -1 +0,0 @@ -{"env_info": "sys.platform: linux\nPython: 3.10.12 (main, May 27 2025, 17:12:29) [GCC 11.4.0]\nCUDA available: True\nGPU 0,1,2,3,4,5,6,7: BW200, UBB BW1000\nCUDA_HOME: /opt/dtk\nNVCC: Not Available\nGCC: x86_64-linux-gnu-gcc (Ubuntu 11.4.0-1ubuntu1~22.04) 11.4.0\nPyTorch: 2.4.1\nPyTorch compiling details: PyTorch built with:\n - GCC 10.3\n - C++ Version: 201703\n - Intel(R) Math Kernel Library Version 2020.0.4 Product Build 20200917 for Intel(R) 64 architecture applications\n - OpenMP 201511 (a.k.a. OpenMP 4.5)\n - LAPACK is enabled (usually provided by MKL)\n - NNPACK is enabled\n - CPU capability usage: AVX512\n - HIP Runtime 6.3.25211\n - MIOpen 2.17.0\n - Magma 2.8.0\n - Build settings: BLAS_INFO=mkl, BUILD_TYPE=Release, CXX_COMPILER=/opt/rh/gcc-toolset-10/root/usr/bin/c++, CXX_FLAGS= -D_GLIBCXX_USE_CXX11_ABI=1 -fvisibility-inlines-hidden -DUSE_PTHREADPOOL -DNDEBUG -DUSE_KINETO -DLIBKINETO_NOCUPTI -DUSE_FBGEMM -DUSE_PYTORCH_QNNPACK -DUSE_XNNPACK -DSYMBOLICATE_MOBILE_DEBUG_HANDLE -O2 -fPIC -Wall -Wextra -Werror=return-type -Werror=non-virtual-dtor -Werror=bool-operation -Wnarrowing -Wno-missing-field-initializers -Wno-type-limits -Wno-array-bounds -Wno-unknown-pragmas -Wno-unused-parameter -Wno-unused-function -Wno-unused-result -Wno-strict-overflow -Wno-strict-aliasing -Wno-stringop-overflow -Wsuggest-override -Wno-psabi -Wno-error=pedantic -Wno-error=old-style-cast -Wno-missing-braces -fdiagnostics-color=always -faligned-new -Wno-unused-but-set-variable -Wno-maybe-uninitialized -fno-math-errno -fno-trapping-math -Werror=format -Wno-stringop-overflow, FORCE_FALLBACK_CUDA_MPI=1, LAPACK_INFO=mkl, PERF_WITH_AVX=1, PERF_WITH_AVX2=1, PERF_WITH_AVX512=1, TORCH_VERSION=2.4.1, USE_CUDA=0, USE_CUDNN=OFF, USE_CUSPARSELT=OFF, USE_EXCEPTION_PTR=1, USE_GFLAGS=1, USE_GLOG=1, USE_GLOO=1, USE_MKL=ON, USE_MKLDNN=0, USE_MPI=1, USE_NCCL=1, USE_NNPACK=ON, USE_OPENMP=1, USE_ROCM=ON, USE_ROCM_KERNEL_ASSERT=OFF, \n\nTorchVision: 0.19.1\nOpenCV: 4.11.0\nMMCV: 1.6.1\nMMCV Compiler: GCC 11.4\nMMCV CUDA Compiler: rocm not available\nMMDetection: 2.26.0+c41df4b", "config": "plugin = True\nplugin_dir = 'projects/mmdet3d_plugin/'\ndist_params = dict(backend='nccl')\nlog_level = 'INFO'\nwork_dir = './work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704'\ntotal_batch_size = 160\nnum_gpus = 8\nbatch_size = 20\nnum_iters_per_epoch = 175\nnum_epochs = 100\ncheckpoint_epoch_interval = 20\ncheckpoint_config = dict(interval=3500)\nlog_config = dict(\n interval=1,\n hooks=[\n dict(type='TextLoggerHook', by_epoch=False),\n dict(type='TensorboardLoggerHook')\n ])\nload_from = None\nresume_from = None\nworkflow = [('train', 1)]\nfp16 = dict(loss_scale=32.0)\ninput_shape = (704, 256)\ntracking_test = True\ntracking_threshold = 0.2\nclass_names = [\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier',\n 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n]\nnum_classes = 10\nembed_dims = 256\nnum_groups = 8\nnum_decoder = 6\nnum_single_frame_decoder = 1\nuse_deformable_func = True\nstrides = [4, 8, 16, 32]\nnum_levels = 4\nnum_depth_layers = 3\ndrop_out = 0.1\ntemporal = True\ndecouple_attn = True\nwith_quality_estimation = True\nmodel = dict(\n type='Sparse4D',\n use_grid_mask=True,\n use_deformable_func=True,\n img_backbone=dict(\n type='ResNet',\n depth=50,\n num_stages=4,\n frozen_stages=-1,\n norm_eval=False,\n style='pytorch',\n with_cp=True,\n out_indices=(0, 1, 2, 3),\n norm_cfg=dict(type='BN', requires_grad=True),\n pretrained='ckpt/resnet50-19c8e357.pth'),\n img_neck=dict(\n type='FPN',\n num_outs=4,\n start_level=0,\n out_channels=256,\n add_extra_convs='on_output',\n relu_before_extra_convs=True,\n in_channels=[256, 512, 1024, 2048]),\n depth_branch=dict(\n type='DenseDepthNet',\n embed_dims=256,\n num_depth_layers=3,\n loss_weight=0.2),\n head=dict(\n type='Sparse4DHead',\n cls_threshold_to_reg=0.05,\n decouple_attn=True,\n instance_bank=dict(\n type='InstanceBank',\n num_anchor=900,\n embed_dims=256,\n anchor='nuscenes_kmeans900.npy',\n anchor_handler=dict(type='SparseBox3DKeyPointsGenerator'),\n num_temp_instances=600,\n confidence_decay=0.6,\n feat_grad=False),\n anchor_encoder=dict(\n type='SparseBox3DEncoder',\n vel_dims=3,\n embed_dims=[128, 32, 32, 64],\n mode='cat',\n output_fc=False,\n in_loops=1,\n out_loops=4),\n num_single_frame_decoder=1,\n operation_order=[\n 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm',\n 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm',\n 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm',\n 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm',\n 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm',\n 'deformable', 'ffn', 'norm', 'refine'\n ],\n temp_graph_model=dict(\n type='MultiheadAttention',\n embed_dims=512,\n num_heads=8,\n batch_first=True,\n dropout=0.1),\n graph_model=dict(\n type='MultiheadAttention',\n embed_dims=512,\n num_heads=8,\n batch_first=True,\n dropout=0.1),\n norm_layer=dict(type='LN', normalized_shape=256),\n ffn=dict(\n type='AsymmetricFFN',\n in_channels=512,\n pre_norm=dict(type='LN'),\n embed_dims=256,\n feedforward_channels=1024,\n num_fcs=2,\n ffn_drop=0.1,\n act_cfg=dict(type='ReLU', inplace=True)),\n deformable_model=dict(\n type='DeformableFeatureAggregation',\n embed_dims=256,\n num_groups=8,\n num_levels=4,\n num_cams=6,\n attn_drop=0.15,\n use_deformable_func=True,\n use_camera_embed=True,\n residual_mode='cat',\n kps_generator=dict(\n type='SparseBox3DKeyPointsGenerator',\n num_learnable_pts=6,\n fix_scale=[[0, 0, 0], [0.45, 0, 0], [-0.45, 0, 0],\n [0, 0.45, 0], [0, -0.45, 0], [0, 0, 0.45],\n [0, 0, -0.45]])),\n refine_layer=dict(\n type='SparseBox3DRefinementModule',\n embed_dims=256,\n num_cls=10,\n refine_yaw=True,\n with_quality_estimation=True),\n sampler=dict(\n type='SparseBox3DTarget',\n num_dn_groups=5,\n num_temp_dn_groups=3,\n dn_noise_scale=[2.0, 2.0, 2.0, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5],\n max_dn_gt=32,\n add_neg_dn=True,\n cls_weight=2.0,\n box_weight=0.25,\n reg_weights=[2.0, 2.0, 2.0, 0.5, 0.5, 0.5, 0.0, 0.0, 0.0, 0.0],\n cls_wise_reg_weights=dict(\n {9: [2.0, 2.0, 2.0, 1.0, 1.0, 1.0, 0.0, 0.0, 1.0, 1.0]})),\n loss_cls=dict(\n type='FocalLoss',\n use_sigmoid=True,\n gamma=2.0,\n alpha=0.25,\n loss_weight=2.0),\n loss_reg=dict(\n type='SparseBox3DLoss',\n loss_box=dict(type='L1Loss', loss_weight=0.25),\n loss_centerness=dict(type='CrossEntropyLoss', use_sigmoid=True),\n loss_yawness=dict(type='GaussianFocalLoss'),\n cls_allow_reverse=[5]),\n decoder=dict(type='SparseBox3DDecoder'),\n reg_weights=[2.0, 2.0, 2.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0]))\ndataset_type = 'NuScenes3DDetTrackDataset'\ndata_root = 'data/nuscenes/'\nanno_root = 'data/nuscenes_anno_pkls/'\nfile_client_args = dict(backend='disk')\nimg_norm_cfg = dict(\n mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375], to_rgb=True)\ntrain_pipeline = [\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(\n type='LoadPointsFromFile',\n coord_type='LIDAR',\n load_dim=5,\n use_dim=5,\n file_client_args=dict(backend='disk')),\n dict(type='ResizeCropFlipImage'),\n dict(type='MultiScaleDepthMapGenerator', downsample=[4, 8, 16]),\n dict(type='BBoxRotation'),\n dict(type='PhotoMetricDistortionMultiViewImage'),\n dict(\n type='NormalizeMultiviewImage',\n mean=[123.675, 116.28, 103.53],\n std=[58.395, 57.12, 57.375],\n to_rgb=True),\n dict(\n type='CircleObjectRangeFilter',\n class_dist_thred=[55, 55, 55, 55, 55, 55, 55, 55, 55, 55]),\n dict(\n type='InstanceNameFilter',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ]),\n dict(type='NuScenesSparse4DAdaptor'),\n dict(\n type='Collect',\n keys=[\n 'img', 'timestamp', 'projection_mat', 'image_wh', 'gt_depth',\n 'focal', 'gt_bboxes_3d', 'gt_labels_3d'\n ],\n meta_keys=['T_global', 'T_global_inv', 'timestamp', 'instance_id'])\n]\ntest_pipeline = [\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(type='ResizeCropFlipImage'),\n dict(\n type='NormalizeMultiviewImage',\n mean=[123.675, 116.28, 103.53],\n std=[58.395, 57.12, 57.375],\n to_rgb=True),\n dict(type='NuScenesSparse4DAdaptor'),\n dict(\n type='Collect',\n keys=['img', 'timestamp', 'projection_mat', 'image_wh'],\n meta_keys=['T_global', 'T_global_inv', 'timestamp'])\n]\ninput_modality = dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False)\ndata_basic_config = dict(\n type='NuScenes3DDetTrackDataset',\n data_root='data/nuscenes/',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier',\n 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ],\n modality=dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False),\n version='v1.0-trainval')\ndata_aug_conf = dict(\n resize_lim=(0.4, 0.47),\n final_dim=(256, 704),\n bot_pct_lim=(0.0, 0.0),\n rot_lim=(-5.4, 5.4),\n H=900,\n W=1600,\n rand_flip=True,\n rot3d_range=[-0.3925, 0.3925])\ndata = dict(\n samples_per_gpu=20,\n workers_per_gpu=20,\n train=dict(\n type='NuScenes3DDetTrackDataset',\n data_root='data/nuscenes/',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ],\n modality=dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False),\n version='v1.0-trainval',\n ann_file='data/nuscenes_anno_pkls/nuscenes_infos_train.pkl',\n pipeline=[\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(\n type='LoadPointsFromFile',\n coord_type='LIDAR',\n load_dim=5,\n use_dim=5,\n file_client_args=dict(backend='disk')),\n dict(type='ResizeCropFlipImage'),\n dict(type='MultiScaleDepthMapGenerator', downsample=[4, 8, 16]),\n dict(type='BBoxRotation'),\n dict(type='PhotoMetricDistortionMultiViewImage'),\n dict(\n type='NormalizeMultiviewImage',\n mean=[123.675, 116.28, 103.53],\n std=[58.395, 57.12, 57.375],\n to_rgb=True),\n dict(\n type='CircleObjectRangeFilter',\n class_dist_thred=[55, 55, 55, 55, 55, 55, 55, 55, 55, 55]),\n dict(\n type='InstanceNameFilter',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian',\n 'traffic_cone'\n ]),\n dict(type='NuScenesSparse4DAdaptor'),\n dict(\n type='Collect',\n keys=[\n 'img', 'timestamp', 'projection_mat', 'image_wh',\n 'gt_depth', 'focal', 'gt_bboxes_3d', 'gt_labels_3d'\n ],\n meta_keys=[\n 'T_global', 'T_global_inv', 'timestamp', 'instance_id'\n ])\n ],\n test_mode=False,\n data_aug_conf=dict(\n resize_lim=(0.4, 0.47),\n final_dim=(256, 704),\n bot_pct_lim=(0.0, 0.0),\n rot_lim=(-5.4, 5.4),\n H=900,\n W=1600,\n rand_flip=True,\n rot3d_range=[-0.3925, 0.3925]),\n with_seq_flag=True,\n sequences_split_num=2,\n keep_consistent_seq_aug=True),\n val=dict(\n type='NuScenes3DDetTrackDataset',\n data_root='data/nuscenes/',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ],\n modality=dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False),\n version='v1.0-trainval',\n ann_file='data/nuscenes_anno_pkls/nuscenes_infos_val.pkl',\n pipeline=[\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(type='ResizeCropFlipImage'),\n dict(\n type='NormalizeMultiviewImage',\n mean=[123.675, 116.28, 103.53],\n std=[58.395, 57.12, 57.375],\n to_rgb=True),\n dict(type='NuScenesSparse4DAdaptor'),\n dict(\n type='Collect',\n keys=['img', 'timestamp', 'projection_mat', 'image_wh'],\n meta_keys=['T_global', 'T_global_inv', 'timestamp'])\n ],\n data_aug_conf=dict(\n resize_lim=(0.4, 0.47),\n final_dim=(256, 704),\n bot_pct_lim=(0.0, 0.0),\n rot_lim=(-5.4, 5.4),\n H=900,\n W=1600,\n rand_flip=True,\n rot3d_range=[-0.3925, 0.3925]),\n test_mode=True,\n tracking=True,\n tracking_threshold=0.2),\n test=dict(\n type='NuScenes3DDetTrackDataset',\n data_root='data/nuscenes/',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ],\n modality=dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False),\n version='v1.0-trainval',\n ann_file='data/nuscenes_anno_pkls/nuscenes_infos_val.pkl',\n pipeline=[\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(type='ResizeCropFlipImage'),\n dict(\n type='NormalizeMultiviewImage',\n mean=[123.675, 116.28, 103.53],\n std=[58.395, 57.12, 57.375],\n to_rgb=True),\n dict(type='NuScenesSparse4DAdaptor'),\n dict(\n type='Collect',\n keys=['img', 'timestamp', 'projection_mat', 'image_wh'],\n meta_keys=['T_global', 'T_global_inv', 'timestamp'])\n ],\n data_aug_conf=dict(\n resize_lim=(0.4, 0.47),\n final_dim=(256, 704),\n bot_pct_lim=(0.0, 0.0),\n rot_lim=(-5.4, 5.4),\n H=900,\n W=1600,\n rand_flip=True,\n rot3d_range=[-0.3925, 0.3925]),\n test_mode=True,\n tracking=True,\n tracking_threshold=0.2))\noptimizer = dict(\n type='Miopen_AdamW',\n lr=0.0006,\n weight_decay=0.001,\n paramwise_cfg=dict(custom_keys=dict(img_backbone=dict(lr_mult=0.5))))\noptimizer_config = dict(grad_clip=dict(max_norm=25, norm_type=2))\nlr_config = dict(\n policy='CosineAnnealing',\n warmup='linear',\n warmup_iters=500,\n warmup_ratio=0.3333333333333333,\n min_lr_ratio=0.001)\nrunner = dict(type='IterBasedRunner', max_iters=17500)\nvis_pipeline = [\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(type='Collect', keys=['img'], meta_keys=['timestamp', 'lidar2img'])\n]\nevaluation = dict(\n interval=3500,\n pipeline=[\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(\n type='Collect', keys=['img'], meta_keys=['timestamp', 'lidar2img'])\n ])\ngpu_ids = range(0, 8)\n", "seed": 0, "exp_name": "sparse4dv3_temporal_r50_1x8_bs6_256x704.py"} diff --git a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20251112_195656.log b/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20251112_195656.log deleted file mode 100644 index f970378a588e1fdd3cee7614783b56243d072a7b..0000000000000000000000000000000000000000 --- a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20251112_195656.log +++ /dev/null @@ -1,3368 +0,0 @@ -2025-11-12 19:56:56,560 - mmdet - INFO - Environment info: ------------------------------------------------------------- -sys.platform: linux -Python: 3.10.12 (main, May 27 2025, 17:12:29) [GCC 11.4.0] -CUDA available: True -GPU 0,1,2,3,4,5,6,7: BW200, UBB BW1000 -CUDA_HOME: /opt/dtk -NVCC: Not Available -GCC: x86_64-linux-gnu-gcc (Ubuntu 11.4.0-1ubuntu1~22.04) 11.4.0 -PyTorch: 2.4.1 -PyTorch compiling details: PyTorch built with: - - GCC 10.3 - - C++ Version: 201703 - - Intel(R) Math Kernel Library Version 2020.0.4 Product Build 20200917 for Intel(R) 64 architecture applications - - OpenMP 201511 (a.k.a. OpenMP 4.5) - - LAPACK is enabled (usually provided by MKL) - - NNPACK is enabled - - CPU capability usage: AVX512 - - HIP Runtime 6.3.25211 - - MIOpen 2.17.0 - - Magma 2.8.0 - - Build settings: BLAS_INFO=mkl, BUILD_TYPE=Release, CXX_COMPILER=/opt/rh/gcc-toolset-10/root/usr/bin/c++, CXX_FLAGS= -D_GLIBCXX_USE_CXX11_ABI=1 -fvisibility-inlines-hidden -DUSE_PTHREADPOOL -DNDEBUG -DUSE_KINETO -DLIBKINETO_NOCUPTI -DUSE_FBGEMM -DUSE_PYTORCH_QNNPACK -DUSE_XNNPACK -DSYMBOLICATE_MOBILE_DEBUG_HANDLE -O2 -fPIC -Wall -Wextra -Werror=return-type -Werror=non-virtual-dtor -Werror=bool-operation -Wnarrowing -Wno-missing-field-initializers -Wno-type-limits -Wno-array-bounds -Wno-unknown-pragmas -Wno-unused-parameter -Wno-unused-function -Wno-unused-result -Wno-strict-overflow -Wno-strict-aliasing -Wno-stringop-overflow -Wsuggest-override -Wno-psabi -Wno-error=pedantic -Wno-error=old-style-cast -Wno-missing-braces -fdiagnostics-color=always -faligned-new -Wno-unused-but-set-variable -Wno-maybe-uninitialized -fno-math-errno -fno-trapping-math -Werror=format -Wno-stringop-overflow, FORCE_FALLBACK_CUDA_MPI=1, LAPACK_INFO=mkl, PERF_WITH_AVX=1, PERF_WITH_AVX2=1, PERF_WITH_AVX512=1, TORCH_VERSION=2.4.1, USE_CUDA=0, USE_CUDNN=OFF, USE_CUSPARSELT=OFF, USE_EXCEPTION_PTR=1, USE_GFLAGS=1, USE_GLOG=1, USE_GLOO=1, USE_MKL=ON, USE_MKLDNN=0, USE_MPI=1, USE_NCCL=1, USE_NNPACK=ON, USE_OPENMP=1, USE_ROCM=ON, USE_ROCM_KERNEL_ASSERT=OFF, - -TorchVision: 0.19.1 -OpenCV: 4.11.0 -MMCV: 1.6.1 -MMCV Compiler: GCC 11.4 -MMCV CUDA Compiler: rocm not available -MMDetection: 2.26.0+c41df4b ------------------------------------------------------------- - -2025-11-12 19:56:57,492 - mmdet - INFO - Distributed training: True -2025-11-12 19:56:58,198 - mmdet - INFO - Config: -plugin = True -plugin_dir = 'projects/mmdet3d_plugin/' -dist_params = dict(backend='nccl') -log_level = 'INFO' -work_dir = './work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704' -total_batch_size = 160 -num_gpus = 8 -batch_size = 20 -num_iters_per_epoch = 175 -num_epochs = 100 -checkpoint_epoch_interval = 20 -checkpoint_config = dict(interval=3500) -log_config = dict( - interval=1, - hooks=[ - dict(type='TextLoggerHook', by_epoch=False), - dict(type='TensorboardLoggerHook') - ]) -load_from = None -resume_from = None -workflow = [('train', 1)] -fp16 = dict(loss_scale=32.0) -input_shape = (704, 256) -tracking_test = True -tracking_threshold = 0.2 -class_names = [ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier', - 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' -] -num_classes = 10 -embed_dims = 256 -num_groups = 8 -num_decoder = 6 -num_single_frame_decoder = 1 -use_deformable_func = True -strides = [4, 8, 16, 32] -num_levels = 4 -num_depth_layers = 3 -drop_out = 0.1 -temporal = True -decouple_attn = True -with_quality_estimation = True -model = dict( - type='Sparse4D', - use_grid_mask=True, - use_deformable_func=True, - img_backbone=dict( - type='ResNet', - depth=50, - num_stages=4, - frozen_stages=-1, - norm_eval=False, - style='pytorch', - with_cp=True, - out_indices=(0, 1, 2, 3), - norm_cfg=dict(type='BN', requires_grad=True), - pretrained='ckpt/resnet50-19c8e357.pth'), - img_neck=dict( - type='FPN', - num_outs=4, - start_level=0, - out_channels=256, - add_extra_convs='on_output', - relu_before_extra_convs=True, - in_channels=[256, 512, 1024, 2048]), - depth_branch=dict( - type='DenseDepthNet', - embed_dims=256, - num_depth_layers=3, - loss_weight=0.2), - head=dict( - type='Sparse4DHead', - cls_threshold_to_reg=0.05, - decouple_attn=True, - instance_bank=dict( - type='InstanceBank', - num_anchor=900, - embed_dims=256, - anchor='nuscenes_kmeans900.npy', - anchor_handler=dict(type='SparseBox3DKeyPointsGenerator'), - num_temp_instances=600, - confidence_decay=0.6, - feat_grad=False), - anchor_encoder=dict( - type='SparseBox3DEncoder', - vel_dims=3, - embed_dims=[128, 32, 32, 64], - mode='cat', - output_fc=False, - in_loops=1, - out_loops=4), - num_single_frame_decoder=1, - operation_order=[ - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine' - ], - temp_graph_model=dict( - type='MultiheadAttention', - embed_dims=512, - num_heads=8, - batch_first=True, - dropout=0.1), - graph_model=dict( - type='MultiheadAttention', - embed_dims=512, - num_heads=8, - batch_first=True, - dropout=0.1), - norm_layer=dict(type='LN', normalized_shape=256), - ffn=dict( - type='AsymmetricFFN', - in_channels=512, - pre_norm=dict(type='LN'), - embed_dims=256, - feedforward_channels=1024, - num_fcs=2, - ffn_drop=0.1, - act_cfg=dict(type='ReLU', inplace=True)), - deformable_model=dict( - type='DeformableFeatureAggregation', - embed_dims=256, - num_groups=8, - num_levels=4, - num_cams=6, - attn_drop=0.15, - use_deformable_func=True, - use_camera_embed=True, - residual_mode='cat', - kps_generator=dict( - type='SparseBox3DKeyPointsGenerator', - num_learnable_pts=6, - fix_scale=[[0, 0, 0], [0.45, 0, 0], [-0.45, 0, 0], - [0, 0.45, 0], [0, -0.45, 0], [0, 0, 0.45], - [0, 0, -0.45]])), - refine_layer=dict( - type='SparseBox3DRefinementModule', - embed_dims=256, - num_cls=10, - refine_yaw=True, - with_quality_estimation=True), - sampler=dict( - type='SparseBox3DTarget', - num_dn_groups=5, - num_temp_dn_groups=3, - dn_noise_scale=[2.0, 2.0, 2.0, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5], - max_dn_gt=32, - add_neg_dn=True, - cls_weight=2.0, - box_weight=0.25, - reg_weights=[2.0, 2.0, 2.0, 0.5, 0.5, 0.5, 0.0, 0.0, 0.0, 0.0], - cls_wise_reg_weights=dict( - {9: [2.0, 2.0, 2.0, 1.0, 1.0, 1.0, 0.0, 0.0, 1.0, 1.0]})), - loss_cls=dict( - type='FocalLoss', - use_sigmoid=True, - gamma=2.0, - alpha=0.25, - loss_weight=2.0), - loss_reg=dict( - type='SparseBox3DLoss', - loss_box=dict(type='L1Loss', loss_weight=0.25), - loss_centerness=dict(type='CrossEntropyLoss', use_sigmoid=True), - loss_yawness=dict(type='GaussianFocalLoss'), - cls_allow_reverse=[5]), - decoder=dict(type='SparseBox3DDecoder'), - reg_weights=[2.0, 2.0, 2.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0])) -dataset_type = 'NuScenes3DDetTrackDataset' -data_root = 'data/nuscenes/' -anno_root = 'data/nuscenes_anno_pkls/' -file_client_args = dict(backend='disk') -img_norm_cfg = dict( - mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375], to_rgb=True) -train_pipeline = [ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict( - type='LoadPointsFromFile', - coord_type='LIDAR', - load_dim=5, - use_dim=5, - file_client_args=dict(backend='disk')), - dict(type='ResizeCropFlipImage'), - dict(type='MultiScaleDepthMapGenerator', downsample=[4, 8, 16]), - dict(type='BBoxRotation'), - dict(type='PhotoMetricDistortionMultiViewImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict( - type='CircleObjectRangeFilter', - class_dist_thred=[55, 55, 55, 55, 55, 55, 55, 55, 55, 55]), - dict( - type='InstanceNameFilter', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ]), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=[ - 'img', 'timestamp', 'projection_mat', 'image_wh', 'gt_depth', - 'focal', 'gt_bboxes_3d', 'gt_labels_3d' - ], - meta_keys=['T_global', 'T_global_inv', 'timestamp', 'instance_id']) -] -test_pipeline = [ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='ResizeCropFlipImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=['img', 'timestamp', 'projection_mat', 'image_wh'], - meta_keys=['T_global', 'T_global_inv', 'timestamp']) -] -input_modality = dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False) -data_basic_config = dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier', - 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval') -data_aug_conf = dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]) -data = dict( - samples_per_gpu=20, - workers_per_gpu=20, - train=dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval', - ann_file='data/nuscenes_anno_pkls/nuscenes_infos_train.pkl', - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict( - type='LoadPointsFromFile', - coord_type='LIDAR', - load_dim=5, - use_dim=5, - file_client_args=dict(backend='disk')), - dict(type='ResizeCropFlipImage'), - dict(type='MultiScaleDepthMapGenerator', downsample=[4, 8, 16]), - dict(type='BBoxRotation'), - dict(type='PhotoMetricDistortionMultiViewImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict( - type='CircleObjectRangeFilter', - class_dist_thred=[55, 55, 55, 55, 55, 55, 55, 55, 55, 55]), - dict( - type='InstanceNameFilter', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', - 'traffic_cone' - ]), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=[ - 'img', 'timestamp', 'projection_mat', 'image_wh', - 'gt_depth', 'focal', 'gt_bboxes_3d', 'gt_labels_3d' - ], - meta_keys=[ - 'T_global', 'T_global_inv', 'timestamp', 'instance_id' - ]) - ], - test_mode=False, - data_aug_conf=dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]), - with_seq_flag=True, - sequences_split_num=2, - keep_consistent_seq_aug=True), - val=dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval', - ann_file='data/nuscenes_anno_pkls/nuscenes_infos_val.pkl', - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='ResizeCropFlipImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=['img', 'timestamp', 'projection_mat', 'image_wh'], - meta_keys=['T_global', 'T_global_inv', 'timestamp']) - ], - data_aug_conf=dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]), - test_mode=True, - tracking=True, - tracking_threshold=0.2), - test=dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval', - ann_file='data/nuscenes_anno_pkls/nuscenes_infos_val.pkl', - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='ResizeCropFlipImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=['img', 'timestamp', 'projection_mat', 'image_wh'], - meta_keys=['T_global', 'T_global_inv', 'timestamp']) - ], - data_aug_conf=dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]), - test_mode=True, - tracking=True, - tracking_threshold=0.2)) -optimizer = dict( - type='AdamW', - lr=0.0006, - weight_decay=0.001, - paramwise_cfg=dict(custom_keys=dict(img_backbone=dict(lr_mult=0.5)))) -optimizer_config = dict(grad_clip=dict(max_norm=25, norm_type=2)) -lr_config = dict( - policy='CosineAnnealing', - warmup='linear', - warmup_iters=500, - warmup_ratio=0.3333333333333333, - min_lr_ratio=0.001) -runner = dict(type='IterBasedRunner', max_iters=17500) -vis_pipeline = [ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='Collect', keys=['img'], meta_keys=['timestamp', 'lidar2img']) -] -evaluation = dict( - interval=3500, - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict( - type='Collect', keys=['img'], meta_keys=['timestamp', 'lidar2img']) - ]) -gpu_ids = range(0, 8) - -2025-11-12 19:56:58,198 - mmdet - INFO - Set random seed to 0, deterministic: False -2025-11-12 19:56:58,494 - mmdet - INFO - initialize ResNet with init_cfg {'type': 'Pretrained', 'checkpoint': 'ckpt/resnet50-19c8e357.pth'} -2025-11-12 19:56:58,820 - mmdet - INFO - initialize FPN with init_cfg {'type': 'Xavier', 'layer': 'Conv2d', 'distribution': 'uniform'} -Name of parameter - Initialization information - -img_backbone.conv1.weight - torch.Size([64, 3, 7, 7]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.bn1.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.bn1.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.conv1.weight - torch.Size([64, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn1.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn1.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.conv2.weight - torch.Size([64, 64, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn2.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn2.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.conv3.weight - torch.Size([256, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn3.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn3.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.downsample.0.weight - torch.Size([256, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.downsample.1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.downsample.1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.conv1.weight - torch.Size([64, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn1.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn1.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.conv2.weight - torch.Size([64, 64, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn2.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn2.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.conv3.weight - torch.Size([256, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn3.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn3.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.conv1.weight - torch.Size([64, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn1.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn1.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.conv2.weight - torch.Size([64, 64, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn2.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn2.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.conv3.weight - torch.Size([256, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn3.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn3.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.conv1.weight - torch.Size([128, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn1.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn1.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.conv2.weight - torch.Size([128, 128, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn2.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn2.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.conv3.weight - torch.Size([512, 128, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn3.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn3.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.downsample.0.weight - torch.Size([512, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.downsample.1.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.downsample.1.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.conv1.weight - torch.Size([128, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn1.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn1.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.conv2.weight - torch.Size([128, 128, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn2.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn2.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.conv3.weight - torch.Size([512, 128, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn3.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn3.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.conv1.weight - torch.Size([128, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn1.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn1.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.conv2.weight - torch.Size([128, 128, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn2.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn2.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.conv3.weight - torch.Size([512, 128, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn3.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn3.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.conv1.weight - torch.Size([128, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn1.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn1.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.conv2.weight - torch.Size([128, 128, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn2.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn2.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.conv3.weight - torch.Size([512, 128, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn3.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn3.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.conv1.weight - torch.Size([256, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.downsample.0.weight - torch.Size([1024, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.downsample.1.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.downsample.1.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.conv1.weight - torch.Size([512, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn1.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn1.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.conv2.weight - torch.Size([512, 512, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn2.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn2.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.conv3.weight - torch.Size([2048, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn3.weight - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn3.bias - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.downsample.0.weight - torch.Size([2048, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.downsample.1.weight - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.downsample.1.bias - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.conv1.weight - torch.Size([512, 2048, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn1.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn1.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.conv2.weight - torch.Size([512, 512, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn2.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn2.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.conv3.weight - torch.Size([2048, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn3.weight - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn3.bias - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.conv1.weight - torch.Size([512, 2048, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn1.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn1.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.conv2.weight - torch.Size([512, 512, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn2.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn2.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.conv3.weight - torch.Size([2048, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn3.weight - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn3.bias - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_neck.lateral_convs.0.conv.weight - torch.Size([256, 256, 1, 1]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.lateral_convs.0.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.lateral_convs.1.conv.weight - torch.Size([256, 512, 1, 1]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.lateral_convs.1.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.lateral_convs.2.conv.weight - torch.Size([256, 1024, 1, 1]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.lateral_convs.2.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.lateral_convs.3.conv.weight - torch.Size([256, 2048, 1, 1]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.lateral_convs.3.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.fpn_convs.0.conv.weight - torch.Size([256, 256, 3, 3]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.fpn_convs.0.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.fpn_convs.1.conv.weight - torch.Size([256, 256, 3, 3]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.fpn_convs.1.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.fpn_convs.2.conv.weight - torch.Size([256, 256, 3, 3]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.fpn_convs.2.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.fpn_convs.3.conv.weight - torch.Size([256, 256, 3, 3]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.fpn_convs.3.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.instance_bank.anchor - torch.Size([900, 11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.instance_bank.instance_feature - torch.Size([900, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.instance_bank.anchor_handler.fix_scale - torch.Size([1, 3]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.0.weight - torch.Size([128, 3]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.0.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.2.weight - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.2.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.3.weight - torch.Size([128, 128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.3.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.5.weight - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.5.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.6.weight - torch.Size([128, 128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.6.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.8.weight - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.8.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.9.weight - torch.Size([128, 128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.9.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.11.weight - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.11.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.0.weight - torch.Size([32, 3]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.0.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.2.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.2.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.3.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.3.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.5.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.5.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.6.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.6.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.8.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.8.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.9.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.9.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.11.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.11.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.0.weight - torch.Size([32, 2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.0.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.2.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.2.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.3.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.3.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.5.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.5.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.6.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.6.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.8.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.8.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.9.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.9.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.11.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.11.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.0.weight - torch.Size([64, 3]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.0.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.2.weight - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.2.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.3.weight - torch.Size([64, 64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.3.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.5.weight - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.5.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.6.weight - torch.Size([64, 64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.6.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.8.weight - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.8.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.9.weight - torch.Size([64, 64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.9.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.11.weight - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.11.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.1.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.1.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.1.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.1.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.1.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.1.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.1.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.1.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.3.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.4.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.4.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.4.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.4.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.5.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.5.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.5.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.5.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.6.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.6.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.8.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.8.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.8.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.8.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.8.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.8.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.8.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.8.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.10.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.11.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.11.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.11.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.11.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.12.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.12.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.12.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.12.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.13.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.13.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.15.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.15.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.15.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.15.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.15.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.15.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.15.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.15.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.16.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.16.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.17.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.18.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.18.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.18.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.18.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.19.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.19.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.19.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.19.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.20.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.20.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.22.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.22.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.22.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.22.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.22.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.22.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.22.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.22.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.23.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.23.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.24.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.25.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.25.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.25.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.25.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.26.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.26.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.26.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.26.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.27.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.27.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.29.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.29.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.29.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.29.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.29.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.29.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.29.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.29.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.30.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.30.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.31.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.32.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.32.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.32.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.32.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.33.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.33.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.33.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.33.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.34.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.34.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.36.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.36.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.36.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.36.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.36.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.36.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.36.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.36.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.37.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.37.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.38.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.fc_before.weight - torch.Size([512, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.fc_after.weight - torch.Size([256, 512]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.0.weight - torch.Size([1, 256, 1, 1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.0.bias - torch.Size([1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.1.weight - torch.Size([1, 256, 1, 1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.1.bias - torch.Size([1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.2.weight - torch.Size([1, 256, 1, 1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.2.bias - torch.Size([1]): -The value is the same before and after calling `init_weights` of Sparse4D -2025-11-12 19:56:58,910 - mmdet - INFO - Model: -Sparse4D( - (img_backbone): ResNet( - (conv1): Conv2d(3, 64, kernel_size=(7, 7), stride=(2, 2), padding=(3, 3), bias=False) - (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (maxpool): MaxPool2d(kernel_size=3, stride=2, padding=1, dilation=1, ceil_mode=False) - (layer1): ResLayer( - (0): Bottleneck( - (conv1): Conv2d(64, 64, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (downsample): Sequential( - (0): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - ) - ) - (1): Bottleneck( - (conv1): Conv2d(256, 64, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (2): Bottleneck( - (conv1): Conv2d(256, 64, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - ) - (layer2): ResLayer( - (0): Bottleneck( - (conv1): Conv2d(256, 128, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (downsample): Sequential( - (0): Conv2d(256, 512, kernel_size=(1, 1), stride=(2, 2), bias=False) - (1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - ) - ) - (1): Bottleneck( - (conv1): Conv2d(512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (2): Bottleneck( - (conv1): Conv2d(512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (3): Bottleneck( - (conv1): Conv2d(512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - ) - (layer3): ResLayer( - (0): Bottleneck( - (conv1): Conv2d(512, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (downsample): Sequential( - (0): Conv2d(512, 1024, kernel_size=(1, 1), stride=(2, 2), bias=False) - (1): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - ) - ) - (1): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (2): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (3): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (4): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (5): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - ) - (layer4): ResLayer( - (0): Bottleneck( - (conv1): Conv2d(1024, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(512, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (downsample): Sequential( - (0): Conv2d(1024, 2048, kernel_size=(1, 1), stride=(2, 2), bias=False) - (1): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - ) - ) - (1): Bottleneck( - (conv1): Conv2d(2048, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(512, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (2): Bottleneck( - (conv1): Conv2d(2048, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(512, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - ) - ) - init_cfg={'type': 'Pretrained', 'checkpoint': 'ckpt/resnet50-19c8e357.pth'} - (img_neck): FPN( - (lateral_convs): ModuleList( - (0): ConvModule( - (conv): Conv2d(256, 256, kernel_size=(1, 1), stride=(1, 1)) - ) - (1): ConvModule( - (conv): Conv2d(512, 256, kernel_size=(1, 1), stride=(1, 1)) - ) - (2): ConvModule( - (conv): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1)) - ) - (3): ConvModule( - (conv): Conv2d(2048, 256, kernel_size=(1, 1), stride=(1, 1)) - ) - ) - (fpn_convs): ModuleList( - (0-3): 4 x ConvModule( - (conv): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1)) - ) - ) - ) - init_cfg={'type': 'Xavier', 'layer': 'Conv2d', 'distribution': 'uniform'} - (head): Sparse4DHead( - (instance_bank): InstanceBank( - (anchor_handler): SparseBox3DKeyPointsGenerator() - ) - (anchor_encoder): SparseBox3DEncoder( - (pos_fc): Sequential( - (0): Linear(in_features=3, out_features=128, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((128,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=128, out_features=128, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((128,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=128, out_features=128, bias=True) - (7): ReLU(inplace=True) - (8): LayerNorm((128,), eps=1e-05, elementwise_affine=True) - (9): Linear(in_features=128, out_features=128, bias=True) - (10): ReLU(inplace=True) - (11): LayerNorm((128,), eps=1e-05, elementwise_affine=True) - ) - (size_fc): Sequential( - (0): Linear(in_features=3, out_features=32, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=32, out_features=32, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=32, out_features=32, bias=True) - (7): ReLU(inplace=True) - (8): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (9): Linear(in_features=32, out_features=32, bias=True) - (10): ReLU(inplace=True) - (11): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - ) - (yaw_fc): Sequential( - (0): Linear(in_features=2, out_features=32, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=32, out_features=32, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=32, out_features=32, bias=True) - (7): ReLU(inplace=True) - (8): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (9): Linear(in_features=32, out_features=32, bias=True) - (10): ReLU(inplace=True) - (11): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - ) - (vel_fc): Sequential( - (0): Linear(in_features=3, out_features=64, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((64,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=64, out_features=64, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((64,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=64, out_features=64, bias=True) - (7): ReLU(inplace=True) - (8): LayerNorm((64,), eps=1e-05, elementwise_affine=True) - (9): Linear(in_features=64, out_features=64, bias=True) - (10): ReLU(inplace=True) - (11): LayerNorm((64,), eps=1e-05, elementwise_affine=True) - ) - ) - (loss_cls): FocalLoss() - (loss_reg): SparseBox3DLoss( - (loss_box): L1Loss() - (loss_cns): CrossEntropyLoss(avg_non_ignore=False) - (loss_yns): GaussianFocalLoss() - ) - (layers): ModuleList( - (0): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (1): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (4-5): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (6): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (7): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (8): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (11-12): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (13): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (14): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (15): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (16): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (17): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (18-19): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (20): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (21): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (22): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (23): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (24): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (25-26): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (27): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (28): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (29): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (30): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (31): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (32-33): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (34): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (35): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (36): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (37): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (38): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - ) - (fc_before): Linear(in_features=256, out_features=512, bias=False) - (fc_after): Linear(in_features=512, out_features=256, bias=False) - ) - (depth_branch): DenseDepthNet( - (depth_layers): ModuleList( - (0-2): 3 x Conv2d(256, 1, kernel_size=(1, 1), stride=(1, 1)) - ) - ) - (grid_mask): GridMask() -) -2025-11-12 19:57:11,751 - mmdet - INFO - Start running, host: root@VM-120-96-tencentos, work_dir: /cfs/auto/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704 -2025-11-12 19:57:11,751 - mmdet - INFO - Hooks will be executed in the following order: -before_run: -(VERY_HIGH ) CosineAnnealingLrUpdaterHook -(ABOVE_NORMAL) Fp16OptimizerHook -(NORMAL ) CheckpointHook -(NORMAL ) CustomDistEvalHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -before_train_epoch: -(VERY_HIGH ) CosineAnnealingLrUpdaterHook -(NORMAL ) CustomDistEvalHook -(LOW ) IterTimerHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -before_train_iter: -(VERY_HIGH ) CosineAnnealingLrUpdaterHook -(NORMAL ) CustomDistEvalHook -(LOW ) IterTimerHook - -------------------- -after_train_iter: -(ABOVE_NORMAL) Fp16OptimizerHook -(NORMAL ) CheckpointHook -(NORMAL ) CustomDistEvalHook -(LOW ) IterTimerHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -after_train_epoch: -(NORMAL ) CheckpointHook -(NORMAL ) CustomDistEvalHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -before_val_epoch: -(LOW ) IterTimerHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -before_val_iter: -(LOW ) IterTimerHook - -------------------- -after_val_iter: -(LOW ) IterTimerHook - -------------------- -after_val_epoch: -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -after_run: -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -2025-11-12 19:57:11,751 - mmdet - INFO - workflow: [('train', 1)], max: 17500 iters -2025-11-12 19:57:11,754 - mmdet - INFO - Checkpoints will be saved to /cfs/auto/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704 by HardDiskBackend. -2025-11-12 19:59:08,461 - mmdet - INFO - Iter [1/17500] lr: 1.000e-04, eta: 23 days, 10:15:14, time: 115.670, data_time: 9.157, memory: 49167, loss_cls_0: 2.3613, loss_box_0: 0.0138, loss_cns_0: 0.0027, loss_yns_0: 0.0008, loss_cls_1: 2.1544, loss_box_1: 0.1081, loss_cns_1: 0.0245, loss_yns_1: 0.0067, loss_cls_2: 2.3120, loss_box_2: 0.0050, loss_cns_2: 0.0006, loss_yns_2: 0.0003, loss_cls_3: 2.3903, loss_box_3: 0.0295, loss_cns_3: 0.0050, loss_yns_3: 0.0014, loss_cls_4: 2.0281, loss_box_4: 0.4164, loss_cns_4: 0.0535, loss_yns_4: 0.0254, loss_cls_5: 2.4244, loss_box_5: 0.0180, loss_cns_5: 0.0022, loss_yns_5: 0.0016, loss_cls_dn_0: 1.1980, loss_box_dn_0: 1.4603, loss_cls_dn_1: 1.1102, loss_box_dn_1: 1.7318, loss_cls_dn_2: 1.1741, loss_box_dn_2: 1.9719, loss_cls_dn_3: 1.1721, loss_box_dn_3: 2.2419, loss_cls_dn_4: 1.0528, loss_box_dn_4: 2.4269, loss_cls_dn_5: 1.2387, loss_box_dn_5: 2.6774, loss_dense_depth: 1.8643, loss: 35.7063, grad_norm: 270.0987 -2025-11-12 19:59:10,503 - mmdet - INFO - Iter [2/17500] lr: 1.004e-04, eta: 11 days, 22:04:43, time: 2.044, data_time: 0.088, memory: 49167, loss_cls_0: 2.0402, loss_box_0: 0.0099, loss_cns_0: 0.0028, loss_yns_0: 0.0010, loss_cls_1: 2.0228, loss_box_1: 0.1235, loss_cns_1: 0.0234, loss_yns_1: 0.0063, loss_cls_2: 2.1048, loss_box_2: 0.2293, loss_cns_2: 0.0208, loss_yns_2: 0.0092, loss_cls_3: 1.9507, loss_box_3: 0.3958, loss_cns_3: 0.0531, loss_yns_3: 0.0192, loss_cls_4: 1.7974, loss_box_4: 1.5609, loss_cns_4: 0.1555, loss_yns_4: 0.0555, loss_cls_5: 2.0567, loss_box_5: 0.5088, loss_cns_5: 0.0577, loss_yns_5: 0.0178, loss_cls_dn_0: 1.0243, loss_box_dn_0: 1.2597, loss_cls_dn_1: 0.9560, loss_box_dn_1: 2.4088, loss_cls_dn_2: 0.9705, loss_box_dn_2: 2.5263, loss_cls_dn_3: 0.9119, loss_box_dn_3: 2.6115, loss_cls_dn_4: 0.8416, loss_box_dn_4: 2.8775, loss_cls_dn_5: 0.9863, loss_box_dn_5: 3.1170, loss_dense_depth: 1.7055, loss: 37.4201, grad_norm: 67.0644 -2025-11-12 19:59:12,154 - mmdet - INFO - Iter [3/17500] lr: 1.008e-04, eta: 8 days, 1:22:59, time: 1.651, data_time: 0.185, memory: 49167, loss_cls_0: 1.4687, loss_box_0: 2.5402, loss_cns_0: 0.6160, loss_yns_0: 0.2146, loss_cls_1: 1.7707, loss_box_1: 1.8194, loss_cns_1: 0.2881, loss_yns_1: 0.1072, loss_cls_2: 1.7881, loss_box_2: 3.8549, loss_cns_2: 0.3539, loss_yns_2: 0.1896, loss_cls_3: 1.6205, loss_box_3: 4.8938, loss_cns_3: 0.4457, loss_yns_3: 0.2107, loss_cls_4: 1.5748, loss_box_4: 4.0875, loss_cns_4: 0.3851, loss_yns_4: 0.1648, loss_cls_5: 1.6861, loss_box_5: 2.8913, loss_cns_5: 0.2246, loss_yns_5: 0.0969, loss_cls_dn_0: 0.7114, loss_box_dn_0: 1.1802, loss_cls_dn_1: 0.8338, loss_box_dn_1: 2.4062, loss_cls_dn_2: 0.8065, loss_box_dn_2: 2.6183, loss_cls_dn_3: 0.7092, loss_box_dn_3: 2.8204, loss_cls_dn_4: 0.7209, loss_box_dn_4: 3.0763, loss_cls_dn_5: 0.8051, loss_box_dn_5: 3.3182, loss_dense_depth: 1.5943, loss: 54.8942, grad_norm: 103.7646 -2025-11-12 19:59:13,713 - mmdet - INFO - Iter [4/17500] lr: 1.012e-04, eta: 6 days, 2:54:57, time: 1.553, data_time: 0.081, memory: 49167, loss_cls_0: 1.3838, loss_box_0: 2.5337, loss_cns_0: 0.5547, loss_yns_0: 0.1801, loss_cls_1: 1.6130, loss_box_1: 3.0731, loss_cns_1: 0.4630, loss_yns_1: 0.1971, loss_cls_2: 1.7034, loss_box_2: 3.5914, loss_cns_2: 0.4577, loss_yns_2: 0.1874, loss_cls_3: 1.5155, loss_box_3: 4.1199, loss_cns_3: 0.4607, loss_yns_3: 0.2141, loss_cls_4: 1.4750, loss_box_4: 4.6860, loss_cns_4: 0.3790, loss_yns_4: 0.1986, loss_cls_5: 1.5025, loss_box_5: 4.8823, loss_cns_5: 0.4498, loss_yns_5: 0.1983, loss_cls_dn_0: 0.5591, loss_box_dn_0: 1.1780, loss_cls_dn_1: 0.7276, loss_box_dn_1: 2.6139, loss_cls_dn_2: 0.6977, loss_box_dn_2: 2.6730, loss_cls_dn_3: 0.6181, loss_box_dn_3: 2.8683, loss_cls_dn_4: 0.5981, loss_box_dn_4: 3.0763, loss_cls_dn_5: 0.6719, loss_box_dn_5: 3.2494, loss_dense_depth: 1.5227, loss: 57.0743, grad_norm: 117.0771 -2025-11-12 19:59:15,311 - mmdet - INFO - Iter [5/17500] lr: 1.016e-04, eta: 4 days, 23:04:49, time: 1.599, data_time: 0.111, memory: 49167, loss_cls_0: 1.3224, loss_box_0: 2.9003, loss_cns_0: 0.4871, loss_yns_0: 0.1900, loss_cls_1: 1.5693, loss_box_1: 3.9507, loss_cns_1: 0.4075, loss_yns_1: 0.2125, loss_cls_2: 1.6129, loss_box_2: 3.9523, loss_cns_2: 0.3926, loss_yns_2: 0.1940, loss_cls_3: 1.4597, loss_box_3: 4.0404, loss_cns_3: 0.3962, loss_yns_3: 0.1972, loss_cls_4: 1.3982, loss_box_4: 4.2180, loss_cns_4: 0.3793, loss_yns_4: 0.1967, loss_cls_5: 1.3988, loss_box_5: 4.4569, loss_cns_5: 0.4207, loss_yns_5: 0.1978, loss_cls_dn_0: 0.5290, loss_box_dn_0: 1.2604, loss_cls_dn_1: 0.6606, loss_box_dn_1: 2.2608, loss_cls_dn_2: 0.6474, loss_box_dn_2: 2.3486, loss_cls_dn_3: 0.5612, loss_box_dn_3: 2.4627, loss_cls_dn_4: 0.5603, loss_box_dn_4: 2.6284, loss_cls_dn_5: 0.5867, loss_box_dn_5: 2.7265, loss_dense_depth: 1.3980, loss: 54.5822, grad_norm: 110.9343 -2025-11-12 19:59:16,953 - mmdet - INFO - Iter [6/17500] lr: 1.020e-04, eta: 4 days, 4:33:09, time: 1.635, data_time: 0.084, memory: 49167, loss_cls_0: 1.2923, loss_box_0: 2.6159, loss_cns_0: 0.5666, loss_yns_0: 0.1805, loss_cls_1: 1.4888, loss_box_1: 3.9561, loss_cns_1: 0.3785, loss_yns_1: 0.1964, loss_cls_2: 1.4741, loss_box_2: 4.0452, loss_cns_2: 0.3728, loss_yns_2: 0.1941, loss_cls_3: 1.3573, loss_box_3: 4.0083, loss_cns_3: 0.3527, loss_yns_3: 0.1996, loss_cls_4: 1.3318, loss_box_4: 4.2542, loss_cns_4: 0.3149, loss_yns_4: 0.1979, loss_cls_5: 1.3394, loss_box_5: 4.3757, loss_cns_5: 0.3105, loss_yns_5: 0.2079, loss_cls_dn_0: 0.5268, loss_box_dn_0: 1.1785, loss_cls_dn_1: 0.5899, loss_box_dn_1: 2.4008, loss_cls_dn_2: 0.5804, loss_box_dn_2: 2.4350, loss_cls_dn_3: 0.5180, loss_box_dn_3: 2.4804, loss_cls_dn_4: 0.4876, loss_box_dn_4: 2.6996, loss_cls_dn_5: 0.4955, loss_box_dn_5: 2.7633, loss_dense_depth: 1.4364, loss: 53.6041, grad_norm: 113.0488 -2025-11-12 19:59:18,515 - mmdet - INFO - Iter [7/17500] lr: 1.024e-04, eta: 3 days, 15:16:29, time: 1.573, data_time: 0.086, memory: 49167, loss_cls_0: 1.2451, loss_box_0: 2.3830, loss_cns_0: 0.6708, loss_yns_0: 0.1775, loss_cls_1: 1.3739, loss_box_1: 3.6689, loss_cns_1: 0.4545, loss_yns_1: 0.1934, loss_cls_2: 1.3933, loss_box_2: 3.6643, loss_cns_2: 0.4529, loss_yns_2: 0.1860, loss_cls_3: 1.3057, loss_box_3: 3.4899, loss_cns_3: 0.4799, loss_yns_3: 0.1884, loss_cls_4: 1.2951, loss_box_4: 3.7630, loss_cns_4: 0.4484, loss_yns_4: 0.1850, loss_cls_5: 1.3390, loss_box_5: 3.9668, loss_cns_5: 0.4373, loss_yns_5: 0.1910, loss_cls_dn_0: 0.5318, loss_box_dn_0: 1.1014, loss_cls_dn_1: 0.5245, loss_box_dn_1: 2.4483, loss_cls_dn_2: 0.5260, loss_box_dn_2: 2.4184, loss_cls_dn_3: 0.4805, loss_box_dn_3: 2.4243, loss_cls_dn_4: 0.4441, loss_box_dn_4: 2.6239, loss_cls_dn_5: 0.4350, loss_box_dn_5: 2.7024, loss_dense_depth: 1.4265, loss: 51.0402, grad_norm: 107.6412 -2025-11-12 19:59:20,082 - mmdet - INFO - Iter [8/17500] lr: 1.028e-04, eta: 3 days, 5:18:39, time: 1.564, data_time: 0.086, memory: 49167, loss_cls_0: 1.2332, loss_box_0: 2.2753, loss_cns_0: 0.6568, loss_yns_0: 0.1807, loss_cls_1: 1.2919, loss_box_1: 3.5719, loss_cns_1: 0.5116, loss_yns_1: 0.1853, loss_cls_2: 1.3893, loss_box_2: 3.5660, loss_cns_2: 0.4553, loss_yns_2: 0.1900, loss_cls_3: 1.2977, loss_box_3: 3.5557, loss_cns_3: 0.4852, loss_yns_3: 0.1894, loss_cls_4: 1.3053, loss_box_4: 3.5895, loss_cns_4: 0.4893, loss_yns_4: 0.1832, loss_cls_5: 1.3283, loss_box_5: 3.6328, loss_cns_5: 0.5266, loss_yns_5: 0.1965, loss_cls_dn_0: 0.5205, loss_box_dn_0: 1.0236, loss_cls_dn_1: 0.5289, loss_box_dn_1: 1.8191, loss_cls_dn_2: 0.5432, loss_box_dn_2: 1.7136, loss_cls_dn_3: 0.4978, loss_box_dn_3: 1.7433, loss_cls_dn_4: 0.4635, loss_box_dn_4: 1.8404, loss_cls_dn_5: 0.4380, loss_box_dn_5: 1.8836, loss_dense_depth: 1.4037, loss: 46.7061, grad_norm: 81.7312 -2025-11-12 19:59:21,644 - mmdet - INFO - Iter [9/17500] lr: 1.032e-04, eta: 2 days, 21:33:21, time: 1.554, data_time: 0.083, memory: 49167, loss_cls_0: 1.2400, loss_box_0: 2.2421, loss_cns_0: 0.6178, loss_yns_0: 0.1766, loss_cls_1: 1.2558, loss_box_1: 3.2785, loss_cns_1: 0.5313, loss_yns_1: 0.1799, loss_cls_2: 1.3671, loss_box_2: 3.3431, loss_cns_2: 0.4683, loss_yns_2: 0.1803, loss_cls_3: 1.2814, loss_box_3: 3.4756, loss_cns_3: 0.5099, loss_yns_3: 0.2073, loss_cls_4: 1.2726, loss_box_4: 3.3490, loss_cns_4: 0.5142, loss_yns_4: 0.1817, loss_cls_5: 1.3105, loss_box_5: 3.3867, loss_cns_5: 0.5408, loss_yns_5: 0.1854, loss_cls_dn_0: 0.4971, loss_box_dn_0: 1.0132, loss_cls_dn_1: 0.4881, loss_box_dn_1: 1.4834, loss_cls_dn_2: 0.5200, loss_box_dn_2: 1.4741, loss_cls_dn_3: 0.4748, loss_box_dn_3: 1.5814, loss_cls_dn_4: 0.4547, loss_box_dn_4: 1.5234, loss_cls_dn_5: 0.4332, loss_box_dn_5: 1.6065, loss_dense_depth: 1.3013, loss: 43.9470, grad_norm: 66.9323 -2025-11-12 19:59:23,202 - mmdet - INFO - Iter [10/17500] lr: 1.036e-04, eta: 2 days, 15:21:15, time: 1.559, data_time: 0.084, memory: 49167, loss_cls_0: 1.2206, loss_box_0: 2.2377, loss_cns_0: 0.6160, loss_yns_0: 0.1729, loss_cls_1: 1.2627, loss_box_1: 3.0808, loss_cns_1: 0.5300, loss_yns_1: 0.1801, loss_cls_2: 1.2943, loss_box_2: 3.1318, loss_cns_2: 0.5070, loss_yns_2: 0.1821, loss_cls_3: 1.2602, loss_box_3: 3.2280, loss_cns_3: 0.5532, loss_yns_3: 0.1864, loss_cls_4: 1.2466, loss_box_4: 3.2064, loss_cns_4: 0.5289, loss_yns_4: 0.1838, loss_cls_5: 1.2751, loss_box_5: 3.3120, loss_cns_5: 0.5283, loss_yns_5: 0.1832, loss_cls_dn_0: 0.4766, loss_box_dn_0: 1.0320, loss_cls_dn_1: 0.4451, loss_box_dn_1: 1.5997, loss_cls_dn_2: 0.4827, loss_box_dn_2: 1.6582, loss_cls_dn_3: 0.4384, loss_box_dn_3: 1.7302, loss_cls_dn_4: 0.4275, loss_box_dn_4: 1.7284, loss_cls_dn_5: 0.4272, loss_box_dn_5: 1.8590, loss_dense_depth: 1.5603, loss: 43.9734, grad_norm: 67.5922 -2025-11-12 19:59:24,768 - mmdet - INFO - Iter [11/17500] lr: 1.040e-04, eta: 2 days, 10:17:13, time: 1.575, data_time: 0.089, memory: 49167, loss_cls_0: 1.2371, loss_box_0: 2.2408, loss_cns_0: 0.6270, loss_yns_0: 0.1766, loss_cls_1: 1.2714, loss_box_1: 2.9951, loss_cns_1: 0.5062, loss_yns_1: 0.1832, loss_cls_2: 1.2743, loss_box_2: 2.9895, loss_cns_2: 0.5142, loss_yns_2: 0.1874, loss_cls_3: 1.2662, loss_box_3: 2.9821, loss_cns_3: 0.5653, loss_yns_3: 0.1803, loss_cls_4: 1.2533, loss_box_4: 3.0299, loss_cns_4: 0.5337, loss_yns_4: 0.1827, loss_cls_5: 1.2772, loss_box_5: 3.2671, loss_cns_5: 0.5464, loss_yns_5: 0.1800, loss_cls_dn_0: 0.4658, loss_box_dn_0: 1.0447, loss_cls_dn_1: 0.4152, loss_box_dn_1: 1.8825, loss_cls_dn_2: 0.4530, loss_box_dn_2: 1.9242, loss_cls_dn_3: 0.4099, loss_box_dn_3: 1.9304, loss_cls_dn_4: 0.3967, loss_box_dn_4: 1.9670, loss_cls_dn_5: 0.4123, loss_box_dn_5: 2.1230, loss_dense_depth: 1.4462, loss: 44.3380, grad_norm: 70.2180 -2025-11-12 19:59:26,340 - mmdet - INFO - Iter [12/17500] lr: 1.044e-04, eta: 2 days, 6:03:37, time: 1.565, data_time: 0.089, memory: 49167, loss_cls_0: 1.2292, loss_box_0: 2.2301, loss_cns_0: 0.6333, loss_yns_0: 0.1738, loss_cls_1: 1.2459, loss_box_1: 2.9448, loss_cns_1: 0.4904, loss_yns_1: 0.1755, loss_cls_2: 1.2735, loss_box_2: 2.9343, loss_cns_2: 0.4984, loss_yns_2: 0.1833, loss_cls_3: 1.2450, loss_box_3: 2.9005, loss_cns_3: 0.5340, loss_yns_3: 0.1766, loss_cls_4: 1.2428, loss_box_4: 2.9318, loss_cns_4: 0.5309, loss_yns_4: 0.1818, loss_cls_5: 1.2805, loss_box_5: 3.0701, loss_cns_5: 0.5384, loss_yns_5: 0.1836, loss_cls_dn_0: 0.4635, loss_box_dn_0: 1.0499, loss_cls_dn_1: 0.3938, loss_box_dn_1: 2.1803, loss_cls_dn_2: 0.4344, loss_box_dn_2: 2.1753, loss_cls_dn_3: 0.3959, loss_box_dn_3: 2.1438, loss_cls_dn_4: 0.3816, loss_box_dn_4: 2.1894, loss_cls_dn_5: 0.3934, loss_box_dn_5: 2.2528, loss_dense_depth: 1.5168, loss: 44.7993, grad_norm: 55.7055 -2025-11-12 19:59:27,911 - mmdet - INFO - Iter [13/17500] lr: 1.048e-04, eta: 2 days, 2:29:05, time: 1.568, data_time: 0.088, memory: 49167, loss_cls_0: 1.1975, loss_box_0: 2.2556, loss_cns_0: 0.6112, loss_yns_0: 0.1705, loss_cls_1: 1.2194, loss_box_1: 2.8349, loss_cns_1: 0.5191, loss_yns_1: 0.1758, loss_cls_2: 1.2620, loss_box_2: 2.8835, loss_cns_2: 0.5316, loss_yns_2: 0.1794, loss_cls_3: 1.2249, loss_box_3: 2.8761, loss_cns_3: 0.5346, loss_yns_3: 0.1750, loss_cls_4: 1.2627, loss_box_4: 2.9050, loss_cns_4: 0.5193, loss_yns_4: 0.1805, loss_cls_5: 1.2791, loss_box_5: 2.8966, loss_cns_5: 0.5324, loss_yns_5: 0.1805, loss_cls_dn_0: 0.4633, loss_box_dn_0: 1.0524, loss_cls_dn_1: 0.4269, loss_box_dn_1: 1.6521, loss_cls_dn_2: 0.4598, loss_box_dn_2: 1.6655, loss_cls_dn_3: 0.4484, loss_box_dn_3: 1.6950, loss_cls_dn_4: 0.4330, loss_box_dn_4: 1.8179, loss_cls_dn_5: 0.4440, loss_box_dn_5: 1.8001, loss_dense_depth: 1.3876, loss: 42.1532, grad_norm: 71.3796 -2025-11-12 19:59:29,473 - mmdet - INFO - Iter [14/17500] lr: 1.052e-04, eta: 1 day, 23:25:09, time: 1.565, data_time: 0.088, memory: 49167, loss_cls_0: 1.2054, loss_box_0: 2.2837, loss_cns_0: 0.5919, loss_yns_0: 0.1714, loss_cls_1: 1.2857, loss_box_1: 2.7326, loss_cns_1: 0.5420, loss_yns_1: 0.1833, loss_cls_2: 1.2958, loss_box_2: 2.7597, loss_cns_2: 0.5523, loss_yns_2: 0.1794, loss_cls_3: 1.2611, loss_box_3: 2.8065, loss_cns_3: 0.5559, loss_yns_3: 0.1744, loss_cls_4: 1.2868, loss_box_4: 2.8265, loss_cns_4: 0.5421, loss_yns_4: 0.1757, loss_cls_5: 1.2873, loss_box_5: 2.8968, loss_cns_5: 0.5478, loss_yns_5: 0.1751, loss_cls_dn_0: 0.4675, loss_box_dn_0: 1.0365, loss_cls_dn_1: 0.4419, loss_box_dn_1: 1.3915, loss_cls_dn_2: 0.4666, loss_box_dn_2: 1.3914, loss_cls_dn_3: 0.4672, loss_box_dn_3: 1.4909, loss_cls_dn_4: 0.4550, loss_box_dn_4: 1.6218, loss_cls_dn_5: 0.4695, loss_box_dn_5: 1.6285, loss_dense_depth: 1.3411, loss: 40.9886, grad_norm: 79.5052 -2025-11-12 19:59:31,035 - mmdet - INFO - Iter [15/17500] lr: 1.056e-04, eta: 1 day, 20:45:43, time: 1.564, data_time: 0.085, memory: 49167, loss_cls_0: 1.2013, loss_box_0: 2.3496, loss_cns_0: 0.5783, loss_yns_0: 0.1713, loss_cls_1: 1.2992, loss_box_1: 2.7144, loss_cns_1: 0.5618, loss_yns_1: 0.1816, loss_cls_2: 1.2949, loss_box_2: 2.6824, loss_cns_2: 0.5718, loss_yns_2: 0.1787, loss_cls_3: 1.2939, loss_box_3: 2.7549, loss_cns_3: 0.5727, loss_yns_3: 0.1752, loss_cls_4: 1.2760, loss_box_4: 2.8019, loss_cns_4: 0.5708, loss_yns_4: 0.1752, loss_cls_5: 1.2773, loss_box_5: 2.8361, loss_cns_5: 0.5770, loss_yns_5: 0.1737, loss_cls_dn_0: 0.4699, loss_box_dn_0: 1.0323, loss_cls_dn_1: 0.4447, loss_box_dn_1: 1.4736, loss_cls_dn_2: 0.4571, loss_box_dn_2: 1.4204, loss_cls_dn_3: 0.4520, loss_box_dn_3: 1.5090, loss_cls_dn_4: 0.4512, loss_box_dn_4: 1.5914, loss_cls_dn_5: 0.4725, loss_box_dn_5: 1.5962, loss_dense_depth: 1.2846, loss: 40.9249, grad_norm: 78.4620 -2025-11-12 19:59:32,609 - mmdet - INFO - Iter [16/17500] lr: 1.060e-04, eta: 1 day, 18:26:18, time: 1.570, data_time: 0.088, memory: 49167, loss_cls_0: 1.1819, loss_box_0: 2.3423, loss_cns_0: 0.5706, loss_yns_0: 0.1731, loss_cls_1: 1.2798, loss_box_1: 2.7372, loss_cns_1: 0.5661, loss_yns_1: 0.1778, loss_cls_2: 1.2695, loss_box_2: 2.7819, loss_cns_2: 0.5637, loss_yns_2: 0.1793, loss_cls_3: 1.3143, loss_box_3: 2.8185, loss_cns_3: 0.5812, loss_yns_3: 0.1785, loss_cls_4: 1.2593, loss_box_4: 2.8109, loss_cns_4: 0.5632, loss_yns_4: 0.1773, loss_cls_5: 1.2648, loss_box_5: 2.7962, loss_cns_5: 0.5941, loss_yns_5: 0.1819, loss_cls_dn_0: 0.4892, loss_box_dn_0: 1.0103, loss_cls_dn_1: 0.4678, loss_box_dn_1: 1.4303, loss_cls_dn_2: 0.4854, loss_box_dn_2: 1.3996, loss_cls_dn_3: 0.4420, loss_box_dn_3: 1.4728, loss_cls_dn_4: 0.4585, loss_box_dn_4: 1.5006, loss_cls_dn_5: 0.4844, loss_box_dn_5: 1.5369, loss_dense_depth: 1.3600, loss: 40.9013, grad_norm: 68.9904 -2025-11-12 19:59:34,166 - mmdet - INFO - Iter [17/17500] lr: 1.064e-04, eta: 1 day, 16:23:06, time: 1.559, data_time: 0.088, memory: 49167, loss_cls_0: 1.1737, loss_box_0: 2.3391, loss_cns_0: 0.5742, loss_yns_0: 0.1709, loss_cls_1: 1.2611, loss_box_1: 2.9443, loss_cns_1: 0.5365, loss_yns_1: 0.1746, loss_cls_2: 1.2722, loss_box_2: 2.9796, loss_cns_2: 0.5386, loss_yns_2: 0.1762, loss_cls_3: 1.2962, loss_box_3: 2.9489, loss_cns_3: 0.5639, loss_yns_3: 0.1752, loss_cls_4: 1.2508, loss_box_4: 2.9161, loss_cns_4: 0.5528, loss_yns_4: 0.1771, loss_cls_5: 1.2611, loss_box_5: 2.9201, loss_cns_5: 0.5664, loss_yns_5: 0.1799, loss_cls_dn_0: 0.4836, loss_box_dn_0: 1.0191, loss_cls_dn_1: 0.4642, loss_box_dn_1: 1.6060, loss_cls_dn_2: 0.4793, loss_box_dn_2: 1.5932, loss_cls_dn_3: 0.4267, loss_box_dn_3: 1.6267, loss_cls_dn_4: 0.4398, loss_box_dn_4: 1.6360, loss_cls_dn_5: 0.4679, loss_box_dn_5: 1.7077, loss_dense_depth: 1.2157, loss: 42.1153, grad_norm: 61.7145 -2025-11-12 19:59:35,722 - mmdet - INFO - Iter [18/17500] lr: 1.068e-04, eta: 1 day, 14:33:30, time: 1.554, data_time: 0.086, memory: 49167, loss_cls_0: 1.1848, loss_box_0: 2.2552, loss_cns_0: 0.5957, loss_yns_0: 0.1706, loss_cls_1: 1.2427, loss_box_1: 2.9356, loss_cns_1: 0.5261, loss_yns_1: 0.1819, loss_cls_2: 1.2557, loss_box_2: 2.9427, loss_cns_2: 0.5299, loss_yns_2: 0.1762, loss_cls_3: 1.2617, loss_box_3: 2.9017, loss_cns_3: 0.5525, loss_yns_3: 0.1774, loss_cls_4: 1.2429, loss_box_4: 2.9633, loss_cns_4: 0.5629, loss_yns_4: 0.1764, loss_cls_5: 1.2611, loss_box_5: 3.0103, loss_cns_5: 0.5477, loss_yns_5: 0.1711, loss_cls_dn_0: 0.4716, loss_box_dn_0: 1.0100, loss_cls_dn_1: 0.4569, loss_box_dn_1: 1.5867, loss_cls_dn_2: 0.4658, loss_box_dn_2: 1.6103, loss_cls_dn_3: 0.4254, loss_box_dn_3: 1.6268, loss_cls_dn_4: 0.4297, loss_box_dn_4: 1.6713, loss_cls_dn_5: 0.4567, loss_box_dn_5: 1.7987, loss_dense_depth: 1.2204, loss: 42.0563, grad_norm: 53.8407 -2025-11-12 19:59:37,281 - mmdet - INFO - Iter [19/17500] lr: 1.072e-04, eta: 1 day, 12:55:33, time: 1.560, data_time: 0.083, memory: 49167, loss_cls_0: 1.1837, loss_box_0: 2.2342, loss_cns_0: 0.5948, loss_yns_0: 0.1713, loss_cls_1: 1.2344, loss_box_1: 2.9653, loss_cns_1: 0.5319, loss_yns_1: 0.1767, loss_cls_2: 1.2549, loss_box_2: 2.9676, loss_cns_2: 0.5321, loss_yns_2: 0.1761, loss_cls_3: 1.2529, loss_box_3: 2.9341, loss_cns_3: 0.5381, loss_yns_3: 0.1753, loss_cls_4: 1.2497, loss_box_4: 3.0025, loss_cns_4: 0.5526, loss_yns_4: 0.1744, loss_cls_5: 1.2634, loss_box_5: 3.0434, loss_cns_5: 0.5409, loss_yns_5: 0.1726, loss_cls_dn_0: 0.4754, loss_box_dn_0: 1.0182, loss_cls_dn_1: 0.4652, loss_box_dn_1: 1.2448, loss_cls_dn_2: 0.4568, loss_box_dn_2: 1.3432, loss_cls_dn_3: 0.4407, loss_box_dn_3: 1.3858, loss_cls_dn_4: 0.4319, loss_box_dn_4: 1.4455, loss_cls_dn_5: 0.4599, loss_box_dn_5: 1.5607, loss_dense_depth: 1.1981, loss: 40.8492, grad_norm: 83.4206 -2025-11-12 19:59:38,840 - mmdet - INFO - Iter [20/17500] lr: 1.076e-04, eta: 1 day, 11:27:22, time: 1.560, data_time: 0.080, memory: 49167, loss_cls_0: 1.1773, loss_box_0: 2.2308, loss_cns_0: 0.5958, loss_yns_0: 0.1699, loss_cls_1: 1.2353, loss_box_1: 2.9129, loss_cns_1: 0.5440, loss_yns_1: 0.1755, loss_cls_2: 1.2622, loss_box_2: 2.8768, loss_cns_2: 0.5484, loss_yns_2: 0.1784, loss_cls_3: 1.2413, loss_box_3: 2.8387, loss_cns_3: 0.5466, loss_yns_3: 0.1726, loss_cls_4: 1.2577, loss_box_4: 2.8708, loss_cns_4: 0.5535, loss_yns_4: 0.1727, loss_cls_5: 1.2688, loss_box_5: 2.9410, loss_cns_5: 0.5497, loss_yns_5: 0.1811, loss_cls_dn_0: 0.4698, loss_box_dn_0: 1.0081, loss_cls_dn_1: 0.4209, loss_box_dn_1: 1.4225, loss_cls_dn_2: 0.4170, loss_box_dn_2: 1.4684, loss_cls_dn_3: 0.4126, loss_box_dn_3: 1.4667, loss_cls_dn_4: 0.3968, loss_box_dn_4: 1.4720, loss_cls_dn_5: 0.4157, loss_box_dn_5: 1.5336, loss_dense_depth: 1.1991, loss: 40.6049, grad_norm: 54.2283 -2025-11-12 19:59:40,521 - mmdet - INFO - Iter [21/17500] lr: 1.080e-04, eta: 1 day, 10:09:14, time: 1.678, data_time: 0.094, memory: 49167, loss_cls_0: 1.1561, loss_box_0: 2.1948, loss_cns_0: 0.6019, loss_yns_0: 0.1694, loss_cls_1: 1.2581, loss_box_1: 2.8328, loss_cns_1: 0.5817, loss_yns_1: 0.1727, loss_cls_2: 1.2704, loss_box_2: 2.7871, loss_cns_2: 0.5823, loss_yns_2: 0.1775, loss_cls_3: 1.2587, loss_box_3: 2.7940, loss_cns_3: 0.5753, loss_yns_3: 0.1737, loss_cls_4: 1.2582, loss_box_4: 2.8504, loss_cns_4: 0.5744, loss_yns_4: 0.1730, loss_cls_5: 1.2826, loss_box_5: 2.8955, loss_cns_5: 0.5647, loss_yns_5: 0.1814, loss_cls_dn_0: 0.4711, loss_box_dn_0: 1.0238, loss_cls_dn_1: 0.4293, loss_box_dn_1: 1.2106, loss_cls_dn_2: 0.4236, loss_box_dn_2: 1.2289, loss_cls_dn_3: 0.4276, loss_box_dn_3: 1.2331, loss_cls_dn_4: 0.4200, loss_box_dn_4: 1.2652, loss_cls_dn_5: 0.4294, loss_box_dn_5: 1.2815, loss_dense_depth: 1.0929, loss: 39.3035, grad_norm: 49.2117 -2025-11-12 19:59:42,163 - mmdet - INFO - Iter [22/17500] lr: 1.084e-04, eta: 1 day, 8:57:46, time: 1.646, data_time: 0.083, memory: 49167, loss_cls_0: 1.1743, loss_box_0: 2.1630, loss_cns_0: 0.6081, loss_yns_0: 0.1680, loss_cls_1: 1.2753, loss_box_1: 2.8052, loss_cns_1: 0.5754, loss_yns_1: 0.1771, loss_cls_2: 1.2720, loss_box_2: 2.7756, loss_cns_2: 0.5786, loss_yns_2: 0.1776, loss_cls_3: 1.2599, loss_box_3: 2.7547, loss_cns_3: 0.6000, loss_yns_3: 0.1796, loss_cls_4: 1.2500, loss_box_4: 2.8207, loss_cns_4: 0.6106, loss_yns_4: 0.1755, loss_cls_5: 1.2716, loss_box_5: 2.8057, loss_cns_5: 0.6136, loss_yns_5: 0.1889, loss_cls_dn_0: 0.4650, loss_box_dn_0: 1.0104, loss_cls_dn_1: 0.4361, loss_box_dn_1: 1.1003, loss_cls_dn_2: 0.4347, loss_box_dn_2: 1.1360, loss_cls_dn_3: 0.4435, loss_box_dn_3: 1.1808, loss_cls_dn_4: 0.4486, loss_box_dn_4: 1.2919, loss_cls_dn_5: 0.4438, loss_box_dn_5: 1.2873, loss_dense_depth: 1.1253, loss: 39.0848, grad_norm: 66.9158 -2025-11-12 19:59:43,763 - mmdet - INFO - Iter [23/17500] lr: 1.088e-04, eta: 1 day, 7:51:55, time: 1.598, data_time: 0.136, memory: 49167, loss_cls_0: 1.1512, loss_box_0: 2.1198, loss_cns_0: 0.6146, loss_yns_0: 0.1687, loss_cls_1: 1.2409, loss_box_1: 2.8219, loss_cns_1: 0.5575, loss_yns_1: 0.1743, loss_cls_2: 1.2622, loss_box_2: 2.7317, loss_cns_2: 0.5774, loss_yns_2: 0.1809, loss_cls_3: 1.2419, loss_box_3: 2.7427, loss_cns_3: 0.6014, loss_yns_3: 0.1749, loss_cls_4: 1.2300, loss_box_4: 2.7740, loss_cns_4: 0.6062, loss_yns_4: 0.1734, loss_cls_5: 1.2549, loss_box_5: 2.8033, loss_cns_5: 0.6192, loss_yns_5: 0.1771, loss_cls_dn_0: 0.4662, loss_box_dn_0: 0.9759, loss_cls_dn_1: 0.4395, loss_box_dn_1: 1.1561, loss_cls_dn_2: 0.4471, loss_box_dn_2: 1.1650, loss_cls_dn_3: 0.4560, loss_box_dn_3: 1.2546, loss_cls_dn_4: 0.4548, loss_box_dn_4: 1.4108, loss_cls_dn_5: 0.4503, loss_box_dn_5: 1.4294, loss_dense_depth: 1.1586, loss: 39.2643, grad_norm: 64.6627 -2025-11-12 19:59:45,346 - mmdet - INFO - Iter [24/17500] lr: 1.092e-04, eta: 1 day, 6:51:23, time: 1.586, data_time: 0.081, memory: 49167, loss_cls_0: 1.1629, loss_box_0: 2.1168, loss_cns_0: 0.6190, loss_yns_0: 0.1697, loss_cls_1: 1.2477, loss_box_1: 2.7689, loss_cns_1: 0.5598, loss_yns_1: 0.1760, loss_cls_2: 1.2668, loss_box_2: 2.7131, loss_cns_2: 0.5782, loss_yns_2: 0.1766, loss_cls_3: 1.2463, loss_box_3: 2.7138, loss_cns_3: 0.5926, loss_yns_3: 0.1704, loss_cls_4: 1.2341, loss_box_4: 2.7120, loss_cns_4: 0.5885, loss_yns_4: 0.1706, loss_cls_5: 1.2555, loss_box_5: 2.7746, loss_cns_5: 0.6040, loss_yns_5: 0.1734, loss_cls_dn_0: 0.4749, loss_box_dn_0: 0.9839, loss_cls_dn_1: 0.4282, loss_box_dn_1: 1.3321, loss_cls_dn_2: 0.4406, loss_box_dn_2: 1.3544, loss_cls_dn_3: 0.4467, loss_box_dn_3: 1.4424, loss_cls_dn_4: 0.4309, loss_box_dn_4: 1.5748, loss_cls_dn_5: 0.4406, loss_box_dn_5: 1.6020, loss_dense_depth: 1.1946, loss: 39.9372, grad_norm: 53.9247 -2025-11-12 19:59:46,932 - mmdet - INFO - Iter [25/17500] lr: 1.096e-04, eta: 1 day, 5:55:43, time: 1.586, data_time: 0.109, memory: 49167, loss_cls_0: 1.1700, loss_box_0: 2.1089, loss_cns_0: 0.6223, loss_yns_0: 0.1689, loss_cls_1: 1.2284, loss_box_1: 2.7691, loss_cns_1: 0.5582, loss_yns_1: 0.1709, loss_cls_2: 1.2339, loss_box_2: 2.7409, loss_cns_2: 0.5811, loss_yns_2: 0.1712, loss_cls_3: 1.2348, loss_box_3: 2.6968, loss_cns_3: 0.5794, loss_yns_3: 0.1708, loss_cls_4: 1.2454, loss_box_4: 2.7244, loss_cns_4: 0.5729, loss_yns_4: 0.1696, loss_cls_5: 1.2394, loss_box_5: 2.7590, loss_cns_5: 0.5691, loss_yns_5: 0.1771, loss_cls_dn_0: 0.4641, loss_box_dn_0: 0.9726, loss_cls_dn_1: 0.4161, loss_box_dn_1: 1.4225, loss_cls_dn_2: 0.4287, loss_box_dn_2: 1.4394, loss_cls_dn_3: 0.4289, loss_box_dn_3: 1.4687, loss_cls_dn_4: 0.4010, loss_box_dn_4: 1.5560, loss_cls_dn_5: 0.4307, loss_box_dn_5: 1.5603, loss_dense_depth: 1.1048, loss: 39.7564, grad_norm: 59.2715 -2025-11-12 19:59:48,511 - mmdet - INFO - Iter [26/17500] lr: 1.100e-04, eta: 1 day, 5:04:13, time: 1.578, data_time: 0.079, memory: 49167, loss_cls_0: 1.1557, loss_box_0: 2.0793, loss_cns_0: 0.6252, loss_yns_0: 0.1699, loss_cls_1: 1.2179, loss_box_1: 2.7988, loss_cns_1: 0.5593, loss_yns_1: 0.1690, loss_cls_2: 1.2317, loss_box_2: 2.7758, loss_cns_2: 0.5845, loss_yns_2: 0.1705, loss_cls_3: 1.2427, loss_box_3: 2.7398, loss_cns_3: 0.5689, loss_yns_3: 0.1710, loss_cls_4: 1.2807, loss_box_4: 2.7542, loss_cns_4: 0.5659, loss_yns_4: 0.1715, loss_cls_5: 1.2453, loss_box_5: 2.7224, loss_cns_5: 0.5720, loss_yns_5: 0.1747, loss_cls_dn_0: 0.4446, loss_box_dn_0: 0.9798, loss_cls_dn_1: 0.4353, loss_box_dn_1: 1.2259, loss_cls_dn_2: 0.4387, loss_box_dn_2: 1.2211, loss_cls_dn_3: 0.4370, loss_box_dn_3: 1.2276, loss_cls_dn_4: 0.4083, loss_box_dn_4: 1.2746, loss_cls_dn_5: 0.4471, loss_box_dn_5: 1.2605, loss_dense_depth: 1.1188, loss: 38.6661, grad_norm: 69.6562 -2025-11-12 19:59:50,088 - mmdet - INFO - Iter [27/17500] lr: 1.104e-04, eta: 1 day, 4:16:30, time: 1.575, data_time: 0.079, memory: 49167, loss_cls_0: 1.1426, loss_box_0: 2.1006, loss_cns_0: 0.6206, loss_yns_0: 0.1722, loss_cls_1: 1.2178, loss_box_1: 2.7567, loss_cns_1: 0.5619, loss_yns_1: 0.1738, loss_cls_2: 1.2464, loss_box_2: 2.7024, loss_cns_2: 0.5809, loss_yns_2: 0.1748, loss_cls_3: 1.2305, loss_box_3: 2.6607, loss_cns_3: 0.5780, loss_yns_3: 0.1713, loss_cls_4: 1.2485, loss_box_4: 2.6568, loss_cns_4: 0.5852, loss_yns_4: 0.1709, loss_cls_5: 1.2493, loss_box_5: 2.6795, loss_cns_5: 0.5829, loss_yns_5: 0.1722, loss_cls_dn_0: 0.4423, loss_box_dn_0: 0.9758, loss_cls_dn_1: 0.4218, loss_box_dn_1: 1.1952, loss_cls_dn_2: 0.4300, loss_box_dn_2: 1.1511, loss_cls_dn_3: 0.4264, loss_box_dn_3: 1.1509, loss_cls_dn_4: 0.4104, loss_box_dn_4: 1.1624, loss_cls_dn_5: 0.4345, loss_box_dn_5: 1.2098, loss_dense_depth: 1.0970, loss: 37.9443, grad_norm: 49.7721 -2025-11-12 19:59:51,643 - mmdet - INFO - Iter [28/17500] lr: 1.108e-04, eta: 1 day, 3:32:00, time: 1.555, data_time: 0.086, memory: 49167, loss_cls_0: 1.1191, loss_box_0: 2.1261, loss_cns_0: 0.6182, loss_yns_0: 0.1706, loss_cls_1: 1.2123, loss_box_1: 2.5238, loss_cns_1: 0.5817, loss_yns_1: 0.1746, loss_cls_2: 1.2090, loss_box_2: 2.5081, loss_cns_2: 0.5912, loss_yns_2: 0.1760, loss_cls_3: 1.2126, loss_box_3: 2.4887, loss_cns_3: 0.5893, loss_yns_3: 0.1689, loss_cls_4: 1.2186, loss_box_4: 2.5189, loss_cns_4: 0.5839, loss_yns_4: 0.1716, loss_cls_5: 1.2298, loss_box_5: 2.5991, loss_cns_5: 0.5776, loss_yns_5: 0.1746, loss_cls_dn_0: 0.4422, loss_box_dn_0: 0.9753, loss_cls_dn_1: 0.4105, loss_box_dn_1: 1.1795, loss_cls_dn_2: 0.4274, loss_box_dn_2: 1.1610, loss_cls_dn_3: 0.4150, loss_box_dn_3: 1.1975, loss_cls_dn_4: 0.4194, loss_box_dn_4: 1.2225, loss_cls_dn_5: 0.4181, loss_box_dn_5: 1.3379, loss_dense_depth: 1.0541, loss: 37.2047, grad_norm: 63.1198 -2025-11-12 19:59:53,213 - mmdet - INFO - Iter [29/17500] lr: 1.112e-04, eta: 1 day, 2:50:47, time: 1.579, data_time: 0.089, memory: 49167, loss_cls_0: 1.0941, loss_box_0: 2.1184, loss_cns_0: 0.6190, loss_yns_0: 0.1713, loss_cls_1: 1.1976, loss_box_1: 2.5242, loss_cns_1: 0.5858, loss_yns_1: 0.1729, loss_cls_2: 1.1889, loss_box_2: 2.5014, loss_cns_2: 0.6025, loss_yns_2: 0.1742, loss_cls_3: 1.2053, loss_box_3: 2.4858, loss_cns_3: 0.5983, loss_yns_3: 0.1709, loss_cls_4: 1.2224, loss_box_4: 2.5062, loss_cns_4: 0.5960, loss_yns_4: 0.1708, loss_cls_5: 1.2063, loss_box_5: 2.5266, loss_cns_5: 0.5972, loss_yns_5: 0.1724, loss_cls_dn_0: 0.4422, loss_box_dn_0: 0.9835, loss_cls_dn_1: 0.4114, loss_box_dn_1: 1.1577, loss_cls_dn_2: 0.4378, loss_box_dn_2: 1.1659, loss_cls_dn_3: 0.4141, loss_box_dn_3: 1.2182, loss_cls_dn_4: 0.4266, loss_box_dn_4: 1.2427, loss_cls_dn_5: 0.4221, loss_box_dn_5: 1.3577, loss_dense_depth: 1.0969, loss: 37.1851, grad_norm: 60.5716 -2025-11-12 19:59:54,789 - mmdet - INFO - Iter [30/17500] lr: 1.116e-04, eta: 1 day, 2:12:16, time: 1.572, data_time: 0.070, memory: 49167, loss_cls_0: 1.0870, loss_box_0: 2.0733, loss_cns_0: 0.6198, loss_yns_0: 0.1713, loss_cls_1: 1.1607, loss_box_1: 2.5797, loss_cns_1: 0.5889, loss_yns_1: 0.1707, loss_cls_2: 1.1686, loss_box_2: 2.5384, loss_cns_2: 0.6035, loss_yns_2: 0.1750, loss_cls_3: 1.1936, loss_box_3: 2.5238, loss_cns_3: 0.5984, loss_yns_3: 0.1710, loss_cls_4: 1.1984, loss_box_4: 2.5409, loss_cns_4: 0.6018, loss_yns_4: 0.1705, loss_cls_5: 1.1935, loss_box_5: 2.5620, loss_cns_5: 0.6069, loss_yns_5: 0.1753, loss_cls_dn_0: 0.4287, loss_box_dn_0: 0.9859, loss_cls_dn_1: 0.4001, loss_box_dn_1: 1.2511, loss_cls_dn_2: 0.4310, loss_box_dn_2: 1.2625, loss_cls_dn_3: 0.3928, loss_box_dn_3: 1.2920, loss_cls_dn_4: 0.4086, loss_box_dn_4: 1.3082, loss_cls_dn_5: 0.4080, loss_box_dn_5: 1.4108, loss_dense_depth: 1.0629, loss: 37.5157, grad_norm: 64.5793 -2025-11-12 19:59:56,344 - mmdet - INFO - Iter [31/17500] lr: 1.120e-04, eta: 1 day, 1:36:07, time: 1.560, data_time: 0.076, memory: 49167, loss_cls_0: 1.0927, loss_box_0: 2.0728, loss_cns_0: 0.6209, loss_yns_0: 0.1722, loss_cls_1: 1.1412, loss_box_1: 2.5913, loss_cns_1: 0.5765, loss_yns_1: 0.1718, loss_cls_2: 1.1579, loss_box_2: 2.5135, loss_cns_2: 0.5956, loss_yns_2: 0.1741, loss_cls_3: 1.1867, loss_box_3: 2.4674, loss_cns_3: 0.5974, loss_yns_3: 0.1713, loss_cls_4: 1.1618, loss_box_4: 2.4990, loss_cns_4: 0.6031, loss_yns_4: 0.1694, loss_cls_5: 1.1761, loss_box_5: 2.5330, loss_cns_5: 0.6043, loss_yns_5: 0.1716, loss_cls_dn_0: 0.4205, loss_box_dn_0: 0.9759, loss_cls_dn_1: 0.3915, loss_box_dn_1: 1.2773, loss_cls_dn_2: 0.4208, loss_box_dn_2: 1.2632, loss_cls_dn_3: 0.3766, loss_box_dn_3: 1.2544, loss_cls_dn_4: 0.3851, loss_box_dn_4: 1.2595, loss_cls_dn_5: 0.3957, loss_box_dn_5: 1.3384, loss_dense_depth: 1.0291, loss: 37.0096, grad_norm: 59.7434 -2025-11-12 19:59:57,901 - mmdet - INFO - Iter [32/17500] lr: 1.124e-04, eta: 1 day, 1:02:11, time: 1.557, data_time: 0.078, memory: 49167, loss_cls_0: 1.0611, loss_box_0: 2.0607, loss_cns_0: 0.6208, loss_yns_0: 0.1710, loss_cls_1: 1.1248, loss_box_1: 2.5011, loss_cns_1: 0.5802, loss_yns_1: 0.1759, loss_cls_2: 1.1349, loss_box_2: 2.3984, loss_cns_2: 0.6029, loss_yns_2: 0.1739, loss_cls_3: 1.1741, loss_box_3: 2.3949, loss_cns_3: 0.6081, loss_yns_3: 0.1705, loss_cls_4: 1.1743, loss_box_4: 2.4031, loss_cns_4: 0.6179, loss_yns_4: 0.1688, loss_cls_5: 1.1605, loss_box_5: 2.4130, loss_cns_5: 0.6083, loss_yns_5: 0.1695, loss_cls_dn_0: 0.4124, loss_box_dn_0: 0.9667, loss_cls_dn_1: 0.3787, loss_box_dn_1: 1.1654, loss_cls_dn_2: 0.4057, loss_box_dn_2: 1.1492, loss_cls_dn_3: 0.3672, loss_box_dn_3: 1.1549, loss_cls_dn_4: 0.3646, loss_box_dn_4: 1.1449, loss_cls_dn_5: 0.3815, loss_box_dn_5: 1.1892, loss_dense_depth: 1.0086, loss: 35.7577, grad_norm: 49.9411 -2025-11-12 19:59:59,482 - mmdet - INFO - Iter [33/17500] lr: 1.128e-04, eta: 1 day, 0:30:26, time: 1.571, data_time: 0.074, memory: 49167, loss_cls_0: 1.0583, loss_box_0: 2.0109, loss_cns_0: 0.6206, loss_yns_0: 0.1697, loss_cls_1: 1.1312, loss_box_1: 2.4353, loss_cns_1: 0.5825, loss_yns_1: 0.1741, loss_cls_2: 1.1424, loss_box_2: 2.3226, loss_cns_2: 0.6067, loss_yns_2: 0.1721, loss_cls_3: 1.1658, loss_box_3: 2.3618, loss_cns_3: 0.6102, loss_yns_3: 0.1688, loss_cls_4: 1.1713, loss_box_4: 2.3551, loss_cns_4: 0.6149, loss_yns_4: 0.1692, loss_cls_5: 1.1727, loss_box_5: 2.3515, loss_cns_5: 0.6183, loss_yns_5: 0.1687, loss_cls_dn_0: 0.4329, loss_box_dn_0: 0.9667, loss_cls_dn_1: 0.3513, loss_box_dn_1: 1.1861, loss_cls_dn_2: 0.3738, loss_box_dn_2: 1.1632, loss_cls_dn_3: 0.3594, loss_box_dn_3: 1.1763, loss_cls_dn_4: 0.3525, loss_box_dn_4: 1.1577, loss_cls_dn_5: 0.3563, loss_box_dn_5: 1.1713, loss_dense_depth: 0.9755, loss: 35.3779, grad_norm: 51.0906 -2025-11-12 20:00:01,053 - mmdet - INFO - Iter [34/17500] lr: 1.132e-04, eta: 1 day, 0:00:35, time: 1.573, data_time: 0.083, memory: 49167, loss_cls_0: 1.0590, loss_box_0: 2.0014, loss_cns_0: 0.6217, loss_yns_0: 0.1709, loss_cls_1: 1.1314, loss_box_1: 2.4095, loss_cns_1: 0.5917, loss_yns_1: 0.1708, loss_cls_2: 1.1492, loss_box_2: 2.3009, loss_cns_2: 0.6120, loss_yns_2: 0.1708, loss_cls_3: 1.1559, loss_box_3: 2.2913, loss_cns_3: 0.6139, loss_yns_3: 0.1704, loss_cls_4: 1.1438, loss_box_4: 2.2881, loss_cns_4: 0.6130, loss_yns_4: 0.1683, loss_cls_5: 1.1614, loss_box_5: 2.3011, loss_cns_5: 0.6163, loss_yns_5: 0.1703, loss_cls_dn_0: 0.4444, loss_box_dn_0: 0.9694, loss_cls_dn_1: 0.3360, loss_box_dn_1: 1.1787, loss_cls_dn_2: 0.3478, loss_box_dn_2: 1.1345, loss_cls_dn_3: 0.3607, loss_box_dn_3: 1.1264, loss_cls_dn_4: 0.3556, loss_box_dn_4: 1.1290, loss_cls_dn_5: 0.3528, loss_box_dn_5: 1.1290, loss_dense_depth: 1.0123, loss: 34.9597, grad_norm: 36.2309 -2025-11-12 20:00:02,619 - mmdet - INFO - Iter [35/17500] lr: 1.136e-04, eta: 23:32:23, time: 1.569, data_time: 0.083, memory: 49167, loss_cls_0: 1.0498, loss_box_0: 1.9701, loss_cns_0: 0.6250, loss_yns_0: 0.1718, loss_cls_1: 1.1430, loss_box_1: 2.3712, loss_cns_1: 0.5986, loss_yns_1: 0.1769, loss_cls_2: 1.1758, loss_box_2: 2.3405, loss_cns_2: 0.6069, loss_yns_2: 0.1713, loss_cls_3: 1.1547, loss_box_3: 2.3616, loss_cns_3: 0.6110, loss_yns_3: 0.1681, loss_cls_4: 1.1525, loss_box_4: 2.4045, loss_cns_4: 0.6085, loss_yns_4: 0.1692, loss_cls_5: 1.1546, loss_box_5: 2.3884, loss_cns_5: 0.6086, loss_yns_5: 0.1712, loss_cls_dn_0: 0.4290, loss_box_dn_0: 0.9622, loss_cls_dn_1: 0.3324, loss_box_dn_1: 1.1412, loss_cls_dn_2: 0.3374, loss_box_dn_2: 1.1189, loss_cls_dn_3: 0.3573, loss_box_dn_3: 1.1390, loss_cls_dn_4: 0.3572, loss_box_dn_4: 1.2084, loss_cls_dn_5: 0.3634, loss_box_dn_5: 1.2028, loss_dense_depth: 1.0939, loss: 35.3967, grad_norm: 50.0200 -2025-11-12 20:00:04,180 - mmdet - INFO - Iter [36/17500] lr: 1.140e-04, eta: 23:05:41, time: 1.560, data_time: 0.082, memory: 49167, loss_cls_0: 1.0561, loss_box_0: 1.9733, loss_cns_0: 0.6204, loss_yns_0: 0.1699, loss_cls_1: 1.1335, loss_box_1: 2.4384, loss_cns_1: 0.5902, loss_yns_1: 0.1757, loss_cls_2: 1.1571, loss_box_2: 2.4350, loss_cns_2: 0.6036, loss_yns_2: 0.1716, loss_cls_3: 1.1546, loss_box_3: 2.4638, loss_cns_3: 0.6082, loss_yns_3: 0.1676, loss_cls_4: 1.1457, loss_box_4: 2.5078, loss_cns_4: 0.6083, loss_yns_4: 0.1705, loss_cls_5: 1.1395, loss_box_5: 2.4816, loss_cns_5: 0.6091, loss_yns_5: 0.1739, loss_cls_dn_0: 0.4069, loss_box_dn_0: 0.9547, loss_cls_dn_1: 0.3579, loss_box_dn_1: 1.0891, loss_cls_dn_2: 0.3628, loss_box_dn_2: 1.0831, loss_cls_dn_3: 0.3720, loss_box_dn_3: 1.1334, loss_cls_dn_4: 0.3795, loss_box_dn_4: 1.2492, loss_cls_dn_5: 0.3970, loss_box_dn_5: 1.2409, loss_dense_depth: 1.0071, loss: 35.7894, grad_norm: 60.3766 -2025-11-12 20:00:05,744 - mmdet - INFO - Iter [37/17500] lr: 1.144e-04, eta: 22:40:31, time: 1.570, data_time: 0.080, memory: 49167, loss_cls_0: 1.0555, loss_box_0: 1.9825, loss_cns_0: 0.6234, loss_yns_0: 0.1723, loss_cls_1: 1.1177, loss_box_1: 2.5157, loss_cns_1: 0.5893, loss_yns_1: 0.1740, loss_cls_2: 1.1511, loss_box_2: 2.4645, loss_cns_2: 0.6096, loss_yns_2: 0.1733, loss_cls_3: 1.1590, loss_box_3: 2.4950, loss_cns_3: 0.6087, loss_yns_3: 0.1687, loss_cls_4: 1.1372, loss_box_4: 2.5313, loss_cns_4: 0.6055, loss_yns_4: 0.1716, loss_cls_5: 1.1362, loss_box_5: 2.5193, loss_cns_5: 0.6071, loss_yns_5: 0.1718, loss_cls_dn_0: 0.4072, loss_box_dn_0: 0.9538, loss_cls_dn_1: 0.3484, loss_box_dn_1: 1.1695, loss_cls_dn_2: 0.3620, loss_box_dn_2: 1.1420, loss_cls_dn_3: 0.3558, loss_box_dn_3: 1.1803, loss_cls_dn_4: 0.3697, loss_box_dn_4: 1.2697, loss_cls_dn_5: 0.3838, loss_box_dn_5: 1.2617, loss_dense_depth: 1.0354, loss: 36.1794, grad_norm: 56.6507 -2025-11-12 20:00:07,292 - mmdet - INFO - Iter [38/17500] lr: 1.148e-04, eta: 22:16:29, time: 1.549, data_time: 0.072, memory: 49167, loss_cls_0: 1.0260, loss_box_0: 1.9724, loss_cns_0: 0.6234, loss_yns_0: 0.1682, loss_cls_1: 1.0949, loss_box_1: 2.4260, loss_cns_1: 0.5977, loss_yns_1: 0.1723, loss_cls_2: 1.1536, loss_box_2: 2.3624, loss_cns_2: 0.6177, loss_yns_2: 0.1727, loss_cls_3: 1.1403, loss_box_3: 2.3593, loss_cns_3: 0.6209, loss_yns_3: 0.1680, loss_cls_4: 1.1204, loss_box_4: 2.3718, loss_cns_4: 0.6210, loss_yns_4: 0.1691, loss_cls_5: 1.1327, loss_box_5: 2.3967, loss_cns_5: 0.6294, loss_yns_5: 0.1690, loss_cls_dn_0: 0.4152, loss_box_dn_0: 0.9504, loss_cls_dn_1: 0.3549, loss_box_dn_1: 1.1152, loss_cls_dn_2: 0.3736, loss_box_dn_2: 1.0640, loss_cls_dn_3: 0.3632, loss_box_dn_3: 1.0769, loss_cls_dn_4: 0.3758, loss_box_dn_4: 1.1312, loss_cls_dn_5: 0.3828, loss_box_dn_5: 1.1431, loss_dense_depth: 0.9712, loss: 35.0032, grad_norm: 51.0901 -2025-11-12 20:00:08,846 - mmdet - INFO - Iter [39/17500] lr: 1.152e-04, eta: 21:53:45, time: 1.554, data_time: 0.069, memory: 49167, loss_cls_0: 1.0514, loss_box_0: 1.9353, loss_cns_0: 0.6261, loss_yns_0: 0.1677, loss_cls_1: 1.1122, loss_box_1: 2.3126, loss_cns_1: 0.6033, loss_yns_1: 0.1723, loss_cls_2: 1.1687, loss_box_2: 2.2523, loss_cns_2: 0.6261, loss_yns_2: 0.1717, loss_cls_3: 1.1392, loss_box_3: 2.2419, loss_cns_3: 0.6298, loss_yns_3: 0.1687, loss_cls_4: 1.1319, loss_box_4: 2.2482, loss_cns_4: 0.6310, loss_yns_4: 0.1684, loss_cls_5: 1.1399, loss_box_5: 2.2457, loss_cns_5: 0.6459, loss_yns_5: 0.1688, loss_cls_dn_0: 0.4198, loss_box_dn_0: 0.9302, loss_cls_dn_1: 0.3570, loss_box_dn_1: 1.0258, loss_cls_dn_2: 0.3812, loss_box_dn_2: 0.9802, loss_cls_dn_3: 0.3775, loss_box_dn_3: 0.9873, loss_cls_dn_4: 0.3737, loss_box_dn_4: 1.0105, loss_cls_dn_5: 0.3826, loss_box_dn_5: 1.0171, loss_dense_depth: 0.9869, loss: 33.9890, grad_norm: 49.2662 -2025-11-12 20:00:10,391 - mmdet - INFO - Iter [40/17500] lr: 1.156e-04, eta: 21:32:04, time: 1.546, data_time: 0.069, memory: 49167, loss_cls_0: 1.0469, loss_box_0: 1.9052, loss_cns_0: 0.6274, loss_yns_0: 0.1665, loss_cls_1: 1.1027, loss_box_1: 2.2471, loss_cns_1: 0.6116, loss_yns_1: 0.1720, loss_cls_2: 1.1435, loss_box_2: 2.2061, loss_cns_2: 0.6295, loss_yns_2: 0.1703, loss_cls_3: 1.1348, loss_box_3: 2.1954, loss_cns_3: 0.6331, loss_yns_3: 0.1676, loss_cls_4: 1.1369, loss_box_4: 2.2021, loss_cns_4: 0.6416, loss_yns_4: 0.1685, loss_cls_5: 1.1396, loss_box_5: 2.2371, loss_cns_5: 0.6403, loss_yns_5: 0.1678, loss_cls_dn_0: 0.4047, loss_box_dn_0: 0.9239, loss_cls_dn_1: 0.3523, loss_box_dn_1: 0.9870, loss_cls_dn_2: 0.3743, loss_box_dn_2: 0.9548, loss_cls_dn_3: 0.3758, loss_box_dn_3: 0.9554, loss_cls_dn_4: 0.3582, loss_box_dn_4: 0.9626, loss_cls_dn_5: 0.3730, loss_box_dn_5: 0.9863, loss_dense_depth: 1.0296, loss: 33.5313, grad_norm: 46.4490 -2025-11-12 20:00:12,018 - mmdet - INFO - Iter [41/17500] lr: 1.160e-04, eta: 21:12:01, time: 1.624, data_time: 0.082, memory: 49167, loss_cls_0: 1.0202, loss_box_0: 1.8832, loss_cns_0: 0.6280, loss_yns_0: 0.1647, loss_cls_1: 1.0944, loss_box_1: 2.1996, loss_cns_1: 0.6196, loss_yns_1: 0.1659, loss_cls_2: 1.1250, loss_box_2: 2.1826, loss_cns_2: 0.6302, loss_yns_2: 0.1673, loss_cls_3: 1.1211, loss_box_3: 2.1771, loss_cns_3: 0.6317, loss_yns_3: 0.1643, loss_cls_4: 1.1375, loss_box_4: 2.1816, loss_cns_4: 0.6339, loss_yns_4: 0.1636, loss_cls_5: 1.1373, loss_box_5: 2.2102, loss_cns_5: 0.6362, loss_yns_5: 0.1647, loss_cls_dn_0: 0.3950, loss_box_dn_0: 0.9147, loss_cls_dn_1: 0.3259, loss_box_dn_1: 0.9411, loss_cls_dn_2: 0.3431, loss_box_dn_2: 0.9293, loss_cls_dn_3: 0.3483, loss_box_dn_3: 0.9349, loss_cls_dn_4: 0.3327, loss_box_dn_4: 0.9457, loss_cls_dn_5: 0.3506, loss_box_dn_5: 0.9883, loss_dense_depth: 0.9574, loss: 32.9468, grad_norm: 45.9855 -2025-11-12 20:00:13,688 - mmdet - INFO - Iter [42/17500] lr: 1.164e-04, eta: 20:53:13, time: 1.669, data_time: 0.071, memory: 49167, loss_cls_0: 1.0290, loss_box_0: 1.8954, loss_cns_0: 0.6274, loss_yns_0: 0.1631, loss_cls_1: 1.1010, loss_box_1: 2.1681, loss_cns_1: 0.6233, loss_yns_1: 0.1653, loss_cls_2: 1.1691, loss_box_2: 2.1576, loss_cns_2: 0.6262, loss_yns_2: 0.1667, loss_cls_3: 1.1294, loss_box_3: 2.1397, loss_cns_3: 0.6280, loss_yns_3: 0.1653, loss_cls_4: 1.1451, loss_box_4: 2.1561, loss_cns_4: 0.6231, loss_yns_4: 0.1637, loss_cls_5: 1.1395, loss_box_5: 2.1735, loss_cns_5: 0.6249, loss_yns_5: 0.1630, loss_cls_dn_0: 0.3975, loss_box_dn_0: 0.9162, loss_cls_dn_1: 0.3305, loss_box_dn_1: 0.9442, loss_cls_dn_2: 0.3422, loss_box_dn_2: 0.9671, loss_cls_dn_3: 0.3517, loss_box_dn_3: 0.9958, loss_cls_dn_4: 0.3442, loss_box_dn_4: 1.0207, loss_cls_dn_5: 0.3680, loss_box_dn_5: 1.0882, loss_dense_depth: 1.0203, loss: 33.2300, grad_norm: 42.0883 -2025-11-12 20:00:15,300 - mmdet - INFO - Iter [43/17500] lr: 1.168e-04, eta: 20:34:54, time: 1.610, data_time: 0.132, memory: 49167, loss_cls_0: 1.0286, loss_box_0: 1.9438, loss_cns_0: 0.6211, loss_yns_0: 0.1619, loss_cls_1: 1.1103, loss_box_1: 2.1972, loss_cns_1: 0.6195, loss_yns_1: 0.1662, loss_cls_2: 1.1674, loss_box_2: 2.1838, loss_cns_2: 0.6265, loss_yns_2: 0.1661, loss_cls_3: 1.1407, loss_box_3: 2.1832, loss_cns_3: 0.6256, loss_yns_3: 0.1660, loss_cls_4: 1.1383, loss_box_4: 2.1704, loss_cns_4: 0.6270, loss_yns_4: 0.1648, loss_cls_5: 1.1307, loss_box_5: 2.2028, loss_cns_5: 0.6250, loss_yns_5: 0.1630, loss_cls_dn_0: 0.4194, loss_box_dn_0: 0.9153, loss_cls_dn_1: 0.3484, loss_box_dn_1: 0.9905, loss_cls_dn_2: 0.3659, loss_box_dn_2: 1.0364, loss_cls_dn_3: 0.3678, loss_box_dn_3: 1.0839, loss_cls_dn_4: 0.3759, loss_box_dn_4: 1.0996, loss_cls_dn_5: 0.3947, loss_box_dn_5: 1.1768, loss_dense_depth: 0.9598, loss: 33.8642, grad_norm: 59.1596 -2025-11-12 20:00:16,860 - mmdet - INFO - Iter [44/17500] lr: 1.172e-04, eta: 20:17:06, time: 1.565, data_time: 0.078, memory: 49167, loss_cls_0: 1.0244, loss_box_0: 1.9732, loss_cns_0: 0.6126, loss_yns_0: 0.1605, loss_cls_1: 1.1089, loss_box_1: 2.2159, loss_cns_1: 0.6169, loss_yns_1: 0.1656, loss_cls_2: 1.1169, loss_box_2: 2.2159, loss_cns_2: 0.6230, loss_yns_2: 0.1677, loss_cls_3: 1.1215, loss_box_3: 2.2189, loss_cns_3: 0.6253, loss_yns_3: 0.1668, loss_cls_4: 1.1128, loss_box_4: 2.2072, loss_cns_4: 0.6282, loss_yns_4: 0.1670, loss_cls_5: 1.1179, loss_box_5: 2.2201, loss_cns_5: 0.6286, loss_yns_5: 0.1644, loss_cls_dn_0: 0.4316, loss_box_dn_0: 0.9146, loss_cls_dn_1: 0.3571, loss_box_dn_1: 1.0473, loss_cls_dn_2: 0.3838, loss_box_dn_2: 1.0892, loss_cls_dn_3: 0.3700, loss_box_dn_3: 1.1349, loss_cls_dn_4: 0.3933, loss_box_dn_4: 1.1463, loss_cls_dn_5: 0.4002, loss_box_dn_5: 1.1985, loss_dense_depth: 1.0261, loss: 34.2727, grad_norm: 61.5323 -2025-11-12 20:00:18,464 - mmdet - INFO - Iter [45/17500] lr: 1.176e-04, eta: 20:00:22, time: 1.604, data_time: 0.101, memory: 49167, loss_cls_0: 1.0444, loss_box_0: 1.9593, loss_cns_0: 0.6142, loss_yns_0: 0.1627, loss_cls_1: 1.1533, loss_box_1: 2.2667, loss_cns_1: 0.6126, loss_yns_1: 0.1671, loss_cls_2: 1.1376, loss_box_2: 2.2160, loss_cns_2: 0.6265, loss_yns_2: 0.1713, loss_cls_3: 1.1693, loss_box_3: 2.2256, loss_cns_3: 0.6308, loss_yns_3: 0.1691, loss_cls_4: 1.1538, loss_box_4: 2.2106, loss_cns_4: 0.6318, loss_yns_4: 0.1721, loss_cls_5: 1.1803, loss_box_5: 2.2316, loss_cns_5: 0.6340, loss_yns_5: 0.1730, loss_cls_dn_0: 0.4250, loss_box_dn_0: 0.9210, loss_cls_dn_1: 0.3599, loss_box_dn_1: 1.0577, loss_cls_dn_2: 0.3971, loss_box_dn_2: 1.0790, loss_cls_dn_3: 0.3715, loss_box_dn_3: 1.1080, loss_cls_dn_4: 0.3969, loss_box_dn_4: 1.1120, loss_cls_dn_5: 0.4016, loss_box_dn_5: 1.1527, loss_dense_depth: 1.0155, loss: 34.5113, grad_norm: 46.6722 -2025-11-12 20:00:20,035 - mmdet - INFO - Iter [46/17500] lr: 1.180e-04, eta: 19:44:07, time: 1.568, data_time: 0.076, memory: 49167, loss_cls_0: 1.0204, loss_box_0: 1.9482, loss_cns_0: 0.6162, loss_yns_0: 0.1600, loss_cls_1: 1.0828, loss_box_1: 2.3261, loss_cns_1: 0.6048, loss_yns_1: 0.1639, loss_cls_2: 1.1167, loss_box_2: 2.2600, loss_cns_2: 0.6213, loss_yns_2: 0.1689, loss_cls_3: 1.1430, loss_box_3: 2.2506, loss_cns_3: 0.6304, loss_yns_3: 0.1663, loss_cls_4: 1.1524, loss_box_4: 2.2428, loss_cns_4: 0.6321, loss_yns_4: 0.1690, loss_cls_5: 1.1646, loss_box_5: 2.2795, loss_cns_5: 0.6423, loss_yns_5: 0.1706, loss_cls_dn_0: 0.4007, loss_box_dn_0: 0.9144, loss_cls_dn_1: 0.3376, loss_box_dn_1: 1.0259, loss_cls_dn_2: 0.3889, loss_box_dn_2: 1.0053, loss_cls_dn_3: 0.3565, loss_box_dn_3: 1.0053, loss_cls_dn_4: 0.3731, loss_box_dn_4: 1.0001, loss_cls_dn_5: 0.3802, loss_box_dn_5: 1.0271, loss_dense_depth: 0.9423, loss: 33.8903, grad_norm: 48.5977 -2025-11-12 20:00:21,619 - mmdet - INFO - Iter [47/17500] lr: 1.184e-04, eta: 19:28:40, time: 1.587, data_time: 0.078, memory: 49167, loss_cls_0: 1.0533, loss_box_0: 1.9661, loss_cns_0: 0.6191, loss_yns_0: 0.1585, loss_cls_1: 1.1349, loss_box_1: 2.2856, loss_cns_1: 0.6141, loss_yns_1: 0.1634, loss_cls_2: 1.1087, loss_box_2: 2.2517, loss_cns_2: 0.6254, loss_yns_2: 0.1650, loss_cls_3: 1.1236, loss_box_3: 2.2441, loss_cns_3: 0.6348, loss_yns_3: 0.1651, loss_cls_4: 1.1175, loss_box_4: 2.2228, loss_cns_4: 0.6377, loss_yns_4: 0.1676, loss_cls_5: 1.1282, loss_box_5: 2.2192, loss_cns_5: 0.6450, loss_yns_5: 0.1674, loss_cls_dn_0: 0.3899, loss_box_dn_0: 0.9188, loss_cls_dn_1: 0.3309, loss_box_dn_1: 0.9618, loss_cls_dn_2: 0.3835, loss_box_dn_2: 0.9416, loss_cls_dn_3: 0.3607, loss_box_dn_3: 0.9322, loss_cls_dn_4: 0.3668, loss_box_dn_4: 0.9183, loss_cls_dn_5: 0.3743, loss_box_dn_5: 0.9215, loss_dense_depth: 1.0053, loss: 33.4244, grad_norm: 50.2602 -2025-11-12 20:00:23,200 - mmdet - INFO - Iter [48/17500] lr: 1.188e-04, eta: 19:13:50, time: 1.581, data_time: 0.081, memory: 49167, loss_cls_0: 1.0392, loss_box_0: 1.9625, loss_cns_0: 0.6164, loss_yns_0: 0.1600, loss_cls_1: 1.1452, loss_box_1: 2.4016, loss_cns_1: 0.6005, loss_yns_1: 0.1600, loss_cls_2: 1.1109, loss_box_2: 2.3428, loss_cns_2: 0.6185, loss_yns_2: 0.1637, loss_cls_3: 1.1544, loss_box_3: 2.3381, loss_cns_3: 0.6300, loss_yns_3: 0.1609, loss_cls_4: 1.1173, loss_box_4: 2.3219, loss_cns_4: 0.6327, loss_yns_4: 0.1649, loss_cls_5: 1.1596, loss_box_5: 2.3102, loss_cns_5: 0.6317, loss_yns_5: 0.1645, loss_cls_dn_0: 0.3954, loss_box_dn_0: 0.9058, loss_cls_dn_1: 0.3293, loss_box_dn_1: 0.9774, loss_cls_dn_2: 0.3643, loss_box_dn_2: 0.9375, loss_cls_dn_3: 0.3584, loss_box_dn_3: 0.9300, loss_cls_dn_4: 0.3516, loss_box_dn_4: 0.9233, loss_cls_dn_5: 0.3635, loss_box_dn_5: 0.9283, loss_dense_depth: 0.9419, loss: 33.8143, grad_norm: 44.1979 -2025-11-12 20:00:24,759 - mmdet - INFO - Iter [49/17500] lr: 1.192e-04, eta: 18:59:29, time: 1.561, data_time: 0.080, memory: 49167, loss_cls_0: 1.0197, loss_box_0: 1.9833, loss_cns_0: 0.6110, loss_yns_0: 0.1615, loss_cls_1: 1.1023, loss_box_1: 2.3828, loss_cns_1: 0.6042, loss_yns_1: 0.1622, loss_cls_2: 1.1121, loss_box_2: 2.3274, loss_cns_2: 0.6221, loss_yns_2: 0.1621, loss_cls_3: 1.1333, loss_box_3: 2.3078, loss_cns_3: 0.6269, loss_yns_3: 0.1613, loss_cls_4: 1.1249, loss_box_4: 2.3248, loss_cns_4: 0.6245, loss_yns_4: 0.1603, loss_cls_5: 1.1448, loss_box_5: 2.3499, loss_cns_5: 0.6211, loss_yns_5: 0.1613, loss_cls_dn_0: 0.4100, loss_box_dn_0: 0.8936, loss_cls_dn_1: 0.3331, loss_box_dn_1: 0.9722, loss_cls_dn_2: 0.3340, loss_box_dn_2: 0.9346, loss_cls_dn_3: 0.3534, loss_box_dn_3: 0.9411, loss_cls_dn_4: 0.3311, loss_box_dn_4: 0.9646, loss_cls_dn_5: 0.3566, loss_box_dn_5: 0.9968, loss_dense_depth: 1.0116, loss: 33.8244, grad_norm: 52.5162 -2025-11-12 20:00:26,319 - mmdet - INFO - Iter [50/17500] lr: 1.196e-04, eta: 18:45:41, time: 1.555, data_time: 0.073, memory: 49167, loss_cls_0: 0.9952, loss_box_0: 1.9756, loss_cns_0: 0.6171, loss_yns_0: 0.1602, loss_cls_1: 1.0815, loss_box_1: 2.3838, loss_cns_1: 0.6088, loss_yns_1: 0.1633, loss_cls_2: 1.1178, loss_box_2: 2.3106, loss_cns_2: 0.6286, loss_yns_2: 0.1617, loss_cls_3: 1.0915, loss_box_3: 2.3028, loss_cns_3: 0.6299, loss_yns_3: 0.1602, loss_cls_4: 1.1148, loss_box_4: 2.3335, loss_cns_4: 0.6269, loss_yns_4: 0.1602, loss_cls_5: 1.1005, loss_box_5: 2.3374, loss_cns_5: 0.6280, loss_yns_5: 0.1604, loss_cls_dn_0: 0.4006, loss_box_dn_0: 0.8954, loss_cls_dn_1: 0.3380, loss_box_dn_1: 1.0314, loss_cls_dn_2: 0.3224, loss_box_dn_2: 0.9930, loss_cls_dn_3: 0.3561, loss_box_dn_3: 1.0309, loss_cls_dn_4: 0.3276, loss_box_dn_4: 1.0869, loss_cls_dn_5: 0.3655, loss_box_dn_5: 1.1213, loss_dense_depth: 0.9462, loss: 34.0655, grad_norm: 61.1904 -2025-11-12 20:00:27,883 - mmdet - INFO - Iter [51/17500] lr: 1.200e-04, eta: 18:32:27, time: 1.561, data_time: 0.079, memory: 49167, loss_cls_0: 1.0065, loss_box_0: 1.9318, loss_cns_0: 0.6198, loss_yns_0: 0.1600, loss_cls_1: 1.1072, loss_box_1: 2.4215, loss_cns_1: 0.5947, loss_yns_1: 0.1646, loss_cls_2: 1.1219, loss_box_2: 2.3037, loss_cns_2: 0.6235, loss_yns_2: 0.1623, loss_cls_3: 1.0984, loss_box_3: 2.3322, loss_cns_3: 0.6265, loss_yns_3: 0.1603, loss_cls_4: 1.1289, loss_box_4: 2.3675, loss_cns_4: 0.6240, loss_yns_4: 0.1623, loss_cls_5: 1.1110, loss_box_5: 2.3477, loss_cns_5: 0.6287, loss_yns_5: 0.1636, loss_cls_dn_0: 0.3917, loss_box_dn_0: 0.8769, loss_cls_dn_1: 0.3172, loss_box_dn_1: 1.0969, loss_cls_dn_2: 0.3131, loss_box_dn_2: 1.0481, loss_cls_dn_3: 0.3346, loss_box_dn_3: 1.0920, loss_cls_dn_4: 0.3174, loss_box_dn_4: 1.1491, loss_cls_dn_5: 0.3569, loss_box_dn_5: 1.1681, loss_dense_depth: 0.8892, loss: 34.3196, grad_norm: 53.9870 -2025-11-12 20:00:29,480 - mmdet - INFO - Iter [52/17500] lr: 1.204e-04, eta: 18:19:54, time: 1.592, data_time: 0.087, memory: 49167, loss_cls_0: 1.0140, loss_box_0: 1.9297, loss_cns_0: 0.6193, loss_yns_0: 0.1632, loss_cls_1: 1.1267, loss_box_1: 2.3728, loss_cns_1: 0.5998, loss_yns_1: 0.1669, loss_cls_2: 1.1227, loss_box_2: 2.3005, loss_cns_2: 0.6233, loss_yns_2: 0.1645, loss_cls_3: 1.1347, loss_box_3: 2.3092, loss_cns_3: 0.6308, loss_yns_3: 0.1599, loss_cls_4: 1.1184, loss_box_4: 2.3150, loss_cns_4: 0.6347, loss_yns_4: 0.1618, loss_cls_5: 1.1227, loss_box_5: 2.3154, loss_cns_5: 0.6394, loss_yns_5: 0.1647, loss_cls_dn_0: 0.3902, loss_box_dn_0: 0.8725, loss_cls_dn_1: 0.3012, loss_box_dn_1: 1.0700, loss_cls_dn_2: 0.3154, loss_box_dn_2: 1.0309, loss_cls_dn_3: 0.3202, loss_box_dn_3: 1.0558, loss_cls_dn_4: 0.3247, loss_box_dn_4: 1.0972, loss_cls_dn_5: 0.3614, loss_box_dn_5: 1.1181, loss_dense_depth: 0.9253, loss: 34.0932, grad_norm: 58.3463 -2025-11-12 20:00:31,048 - mmdet - INFO - Iter [53/17500] lr: 1.208e-04, eta: 18:07:41, time: 1.568, data_time: 0.085, memory: 49167, loss_cls_0: 1.0235, loss_box_0: 1.9784, loss_cns_0: 0.6151, loss_yns_0: 0.1649, loss_cls_1: 1.0775, loss_box_1: 2.3914, loss_cns_1: 0.5971, loss_yns_1: 0.1680, loss_cls_2: 1.1117, loss_box_2: 2.3218, loss_cns_2: 0.6243, loss_yns_2: 0.1660, loss_cls_3: 1.1521, loss_box_3: 2.3446, loss_cns_3: 0.6265, loss_yns_3: 0.1649, loss_cls_4: 1.1246, loss_box_4: 2.3384, loss_cns_4: 0.6295, loss_yns_4: 0.1624, loss_cls_5: 1.1232, loss_box_5: 2.3378, loss_cns_5: 0.6354, loss_yns_5: 0.1645, loss_cls_dn_0: 0.3958, loss_box_dn_0: 0.8733, loss_cls_dn_1: 0.2879, loss_box_dn_1: 1.0014, loss_cls_dn_2: 0.3220, loss_box_dn_2: 0.9602, loss_cls_dn_3: 0.3124, loss_box_dn_3: 0.9727, loss_cls_dn_4: 0.3278, loss_box_dn_4: 0.9910, loss_cls_dn_5: 0.3641, loss_box_dn_5: 1.0103, loss_dense_depth: 0.9365, loss: 33.7990, grad_norm: 61.7795 -2025-11-12 20:00:32,622 - mmdet - INFO - Iter [54/17500] lr: 1.212e-04, eta: 17:55:57, time: 1.573, data_time: 0.083, memory: 49167, loss_cls_0: 1.0081, loss_box_0: 1.9194, loss_cns_0: 0.6222, loss_yns_0: 0.1622, loss_cls_1: 1.0979, loss_box_1: 2.3170, loss_cns_1: 0.6065, loss_yns_1: 0.1644, loss_cls_2: 1.1089, loss_box_2: 2.2375, loss_cns_2: 0.6307, loss_yns_2: 0.1655, loss_cls_3: 1.1066, loss_box_3: 2.2496, loss_cns_3: 0.6371, loss_yns_3: 0.1628, loss_cls_4: 1.1067, loss_box_4: 2.2641, loss_cns_4: 0.6362, loss_yns_4: 0.1611, loss_cls_5: 1.1114, loss_box_5: 2.2343, loss_cns_5: 0.6400, loss_yns_5: 0.1599, loss_cls_dn_0: 0.3779, loss_box_dn_0: 0.8740, loss_cls_dn_1: 0.2761, loss_box_dn_1: 0.9879, loss_cls_dn_2: 0.3225, loss_box_dn_2: 0.9409, loss_cls_dn_3: 0.3106, loss_box_dn_3: 0.9413, loss_cls_dn_4: 0.3168, loss_box_dn_4: 0.9521, loss_cls_dn_5: 0.3446, loss_box_dn_5: 0.9519, loss_dense_depth: 0.9624, loss: 33.0691, grad_norm: 46.9881 -2025-11-12 20:00:34,194 - mmdet - INFO - Iter [55/17500] lr: 1.216e-04, eta: 17:44:39, time: 1.574, data_time: 0.089, memory: 49167, loss_cls_0: 1.0034, loss_box_0: 1.9422, loss_cns_0: 0.6192, loss_yns_0: 0.1648, loss_cls_1: 1.0738, loss_box_1: 2.2913, loss_cns_1: 0.6048, loss_yns_1: 0.1701, loss_cls_2: 1.0900, loss_box_2: 2.1869, loss_cns_2: 0.6323, loss_yns_2: 0.1678, loss_cls_3: 1.0870, loss_box_3: 2.1977, loss_cns_3: 0.6439, loss_yns_3: 0.1663, loss_cls_4: 1.0891, loss_box_4: 2.2201, loss_cns_4: 0.6442, loss_yns_4: 0.1668, loss_cls_5: 1.1046, loss_box_5: 2.1972, loss_cns_5: 0.6413, loss_yns_5: 0.1692, loss_cls_dn_0: 0.3901, loss_box_dn_0: 0.8652, loss_cls_dn_1: 0.2961, loss_box_dn_1: 0.8959, loss_cls_dn_2: 0.3543, loss_box_dn_2: 0.8450, loss_cls_dn_3: 0.3499, loss_box_dn_3: 0.8457, loss_cls_dn_4: 0.3331, loss_box_dn_4: 0.8533, loss_cls_dn_5: 0.3531, loss_box_dn_5: 0.8528, loss_dense_depth: 0.9016, loss: 32.4102, grad_norm: 55.9026 -2025-11-12 20:00:35,756 - mmdet - INFO - Iter [56/17500] lr: 1.220e-04, eta: 17:33:41, time: 1.561, data_time: 0.084, memory: 49167, loss_cls_0: 0.9725, loss_box_0: 1.9098, loss_cns_0: 0.6148, loss_yns_0: 0.1643, loss_cls_1: 1.0375, loss_box_1: 2.1540, loss_cns_1: 0.6081, loss_yns_1: 0.1657, loss_cls_2: 1.0886, loss_box_2: 2.0679, loss_cns_2: 0.6361, loss_yns_2: 0.1668, loss_cls_3: 1.1132, loss_box_3: 2.0958, loss_cns_3: 0.6425, loss_yns_3: 0.1648, loss_cls_4: 1.1281, loss_box_4: 2.0999, loss_cns_4: 0.6431, loss_yns_4: 0.1649, loss_cls_5: 1.1115, loss_box_5: 2.1059, loss_cns_5: 0.6376, loss_yns_5: 0.1639, loss_cls_dn_0: 0.3901, loss_box_dn_0: 0.8620, loss_cls_dn_1: 0.2947, loss_box_dn_1: 0.9011, loss_cls_dn_2: 0.3498, loss_box_dn_2: 0.8670, loss_cls_dn_3: 0.3542, loss_box_dn_3: 0.8823, loss_cls_dn_4: 0.3346, loss_box_dn_4: 0.8782, loss_cls_dn_5: 0.3404, loss_box_dn_5: 0.9018, loss_dense_depth: 0.9221, loss: 31.9356, grad_norm: 72.7321 -2025-11-12 20:00:37,320 - mmdet - INFO - Iter [57/17500] lr: 1.224e-04, eta: 17:23:06, time: 1.563, data_time: 0.083, memory: 49167, loss_cls_0: 0.9746, loss_box_0: 1.8797, loss_cns_0: 0.6180, loss_yns_0: 0.1632, loss_cls_1: 1.0465, loss_box_1: 2.1486, loss_cns_1: 0.6153, loss_yns_1: 0.1652, loss_cls_2: 1.0944, loss_box_2: 2.1185, loss_cns_2: 0.6331, loss_yns_2: 0.1649, loss_cls_3: 1.1179, loss_box_3: 2.1352, loss_cns_3: 0.6375, loss_yns_3: 0.1646, loss_cls_4: 1.1565, loss_box_4: 2.1245, loss_cns_4: 0.6401, loss_yns_4: 0.1639, loss_cls_5: 1.1310, loss_box_5: 2.1524, loss_cns_5: 0.6382, loss_yns_5: 0.1651, loss_cls_dn_0: 0.3933, loss_box_dn_0: 0.8569, loss_cls_dn_1: 0.2959, loss_box_dn_1: 0.8968, loss_cls_dn_2: 0.3399, loss_box_dn_2: 0.8889, loss_cls_dn_3: 0.3473, loss_box_dn_3: 0.9062, loss_cls_dn_4: 0.3404, loss_box_dn_4: 0.8987, loss_cls_dn_5: 0.3369, loss_box_dn_5: 0.9382, loss_dense_depth: 0.9215, loss: 32.2099, grad_norm: 65.8056 -2025-11-12 20:00:38,880 - mmdet - INFO - Iter [58/17500] lr: 1.228e-04, eta: 17:12:54, time: 1.565, data_time: 0.088, memory: 49167, loss_cls_0: 0.9729, loss_box_0: 1.8852, loss_cns_0: 0.6193, loss_yns_0: 0.1636, loss_cls_1: 1.0460, loss_box_1: 2.0913, loss_cns_1: 0.6191, loss_yns_1: 0.1655, loss_cls_2: 1.0722, loss_box_2: 2.0961, loss_cns_2: 0.6277, loss_yns_2: 0.1662, loss_cls_3: 1.0756, loss_box_3: 2.0694, loss_cns_3: 0.6381, loss_yns_3: 0.1657, loss_cls_4: 1.0889, loss_box_4: 2.0770, loss_cns_4: 0.6382, loss_yns_4: 0.1650, loss_cls_5: 1.0993, loss_box_5: 2.0744, loss_cns_5: 0.6392, loss_yns_5: 0.1652, loss_cls_dn_0: 0.3814, loss_box_dn_0: 0.8530, loss_cls_dn_1: 0.2958, loss_box_dn_1: 0.9083, loss_cls_dn_2: 0.3208, loss_box_dn_2: 0.9270, loss_cls_dn_3: 0.3253, loss_box_dn_3: 0.9335, loss_cls_dn_4: 0.3381, loss_box_dn_4: 0.9439, loss_cls_dn_5: 0.3349, loss_box_dn_5: 0.9773, loss_dense_depth: 0.8851, loss: 31.8454, grad_norm: 53.4634 -2025-11-12 20:00:40,437 - mmdet - INFO - Iter [59/17500] lr: 1.232e-04, eta: 17:03:01, time: 1.557, data_time: 0.084, memory: 49167, loss_cls_0: 0.9903, loss_box_0: 1.9088, loss_cns_0: 0.6187, loss_yns_0: 0.1629, loss_cls_1: 1.0435, loss_box_1: 2.1531, loss_cns_1: 0.6239, loss_yns_1: 0.1645, loss_cls_2: 1.1179, loss_box_2: 2.1274, loss_cns_2: 0.6352, loss_yns_2: 0.1649, loss_cls_3: 1.2346, loss_box_3: 2.0977, loss_cns_3: 0.6419, loss_yns_3: 0.1647, loss_cls_4: 1.1146, loss_box_4: 2.1310, loss_cns_4: 0.6434, loss_yns_4: 0.1632, loss_cls_5: 1.1511, loss_box_5: 2.1090, loss_cns_5: 0.6471, loss_yns_5: 0.1675, loss_cls_dn_0: 0.3658, loss_box_dn_0: 0.8542, loss_cls_dn_1: 0.2916, loss_box_dn_1: 0.9346, loss_cls_dn_2: 0.3063, loss_box_dn_2: 0.9590, loss_cls_dn_3: 0.3153, loss_box_dn_3: 0.9603, loss_cls_dn_4: 0.3281, loss_box_dn_4: 0.9841, loss_cls_dn_5: 0.3380, loss_box_dn_5: 1.0085, loss_dense_depth: 0.8708, loss: 32.4942, grad_norm: 72.0871 -2025-11-12 20:00:41,996 - mmdet - INFO - Iter [60/17500] lr: 1.236e-04, eta: 16:53:27, time: 1.560, data_time: 0.080, memory: 49167, loss_cls_0: 0.9954, loss_box_0: 1.8656, loss_cns_0: 0.6223, loss_yns_0: 0.1592, loss_cls_1: 1.0602, loss_box_1: 2.1364, loss_cns_1: 0.6332, loss_yns_1: 0.1631, loss_cls_2: 1.1744, loss_box_2: 2.1209, loss_cns_2: 0.6447, loss_yns_2: 0.1637, loss_cls_3: 1.1717, loss_box_3: 2.1092, loss_cns_3: 0.6432, loss_yns_3: 0.1634, loss_cls_4: 1.1971, loss_box_4: 2.1412, loss_cns_4: 0.6466, loss_yns_4: 0.1621, loss_cls_5: 1.1435, loss_box_5: 2.1166, loss_cns_5: 0.6489, loss_yns_5: 0.1658, loss_cls_dn_0: 0.3581, loss_box_dn_0: 0.8614, loss_cls_dn_1: 0.2758, loss_box_dn_1: 0.9625, loss_cls_dn_2: 0.2984, loss_box_dn_2: 0.9806, loss_cls_dn_3: 0.3269, loss_box_dn_3: 0.9786, loss_cls_dn_4: 0.3221, loss_box_dn_4: 0.9967, loss_cls_dn_5: 0.3479, loss_box_dn_5: 1.0120, loss_dense_depth: 0.8907, loss: 32.6601, grad_norm: 88.4225 -2025-11-12 20:00:43,632 - mmdet - INFO - Iter [61/17500] lr: 1.240e-04, eta: 16:44:36, time: 1.641, data_time: 0.094, memory: 49167, loss_cls_0: 0.9892, loss_box_0: 1.8519, loss_cns_0: 0.6203, loss_yns_0: 0.1584, loss_cls_1: 1.0546, loss_box_1: 2.1289, loss_cns_1: 0.6309, loss_yns_1: 0.1624, loss_cls_2: 1.0805, loss_box_2: 2.0891, loss_cns_2: 0.6444, loss_yns_2: 0.1626, loss_cls_3: 1.0989, loss_box_3: 2.0858, loss_cns_3: 0.6430, loss_yns_3: 0.1626, loss_cls_4: 1.1310, loss_box_4: 2.0783, loss_cns_4: 0.6478, loss_yns_4: 0.1614, loss_cls_5: 1.0915, loss_box_5: 2.0452, loss_cns_5: 0.6494, loss_yns_5: 0.1632, loss_cls_dn_0: 0.3632, loss_box_dn_0: 0.8567, loss_cls_dn_1: 0.2683, loss_box_dn_1: 0.9806, loss_cls_dn_2: 0.3077, loss_box_dn_2: 0.9695, loss_cls_dn_3: 0.3490, loss_box_dn_3: 0.9619, loss_cls_dn_4: 0.3342, loss_box_dn_4: 0.9594, loss_cls_dn_5: 0.3637, loss_box_dn_5: 0.9635, loss_dense_depth: 0.9309, loss: 32.1400, grad_norm: 57.3364 -2025-11-12 20:00:45,281 - mmdet - INFO - Iter [62/17500] lr: 1.244e-04, eta: 16:36:05, time: 1.652, data_time: 0.119, memory: 49167, loss_cls_0: 0.9831, loss_box_0: 1.8675, loss_cns_0: 0.6191, loss_yns_0: 0.1600, loss_cls_1: 1.0601, loss_box_1: 2.1890, loss_cns_1: 0.6282, loss_yns_1: 0.1612, loss_cls_2: 1.0871, loss_box_2: 2.1084, loss_cns_2: 0.6443, loss_yns_2: 0.1621, loss_cls_3: 1.0954, loss_box_3: 2.1152, loss_cns_3: 0.6473, loss_yns_3: 0.1612, loss_cls_4: 1.0844, loss_box_4: 2.0844, loss_cns_4: 0.6506, loss_yns_4: 0.1602, loss_cls_5: 1.0902, loss_box_5: 2.0934, loss_cns_5: 0.6493, loss_yns_5: 0.1611, loss_cls_dn_0: 0.3795, loss_box_dn_0: 0.8679, loss_cls_dn_1: 0.2723, loss_box_dn_1: 0.8994, loss_cls_dn_2: 0.3270, loss_box_dn_2: 0.8669, loss_cls_dn_3: 0.3571, loss_box_dn_3: 0.8668, loss_cls_dn_4: 0.3422, loss_box_dn_4: 0.8566, loss_cls_dn_5: 0.3773, loss_box_dn_5: 0.8702, loss_dense_depth: 0.8932, loss: 31.8394, grad_norm: 43.4314 -2025-11-12 20:00:46,895 - mmdet - INFO - Iter [63/17500] lr: 1.248e-04, eta: 16:27:40, time: 1.614, data_time: 0.127, memory: 49167, loss_cls_0: 0.9750, loss_box_0: 1.8810, loss_cns_0: 0.6134, loss_yns_0: 0.1605, loss_cls_1: 1.0716, loss_box_1: 2.1940, loss_cns_1: 0.6279, loss_yns_1: 0.1620, loss_cls_2: 1.1344, loss_box_2: 2.1553, loss_cns_2: 0.6426, loss_yns_2: 0.1630, loss_cls_3: 1.1343, loss_box_3: 2.1358, loss_cns_3: 0.6493, loss_yns_3: 0.1616, loss_cls_4: 1.0979, loss_box_4: 2.1336, loss_cns_4: 0.6539, loss_yns_4: 0.1607, loss_cls_5: 1.1358, loss_box_5: 2.1461, loss_cns_5: 0.6529, loss_yns_5: 0.1612, loss_cls_dn_0: 0.3871, loss_box_dn_0: 0.8748, loss_cls_dn_1: 0.2651, loss_box_dn_1: 0.8779, loss_cls_dn_2: 0.3095, loss_box_dn_2: 0.8633, loss_cls_dn_3: 0.3161, loss_box_dn_3: 0.8698, loss_cls_dn_4: 0.3008, loss_box_dn_4: 0.8769, loss_cls_dn_5: 0.3375, loss_box_dn_5: 0.8916, loss_dense_depth: 0.8968, loss: 32.0707, grad_norm: 63.7674 -2025-11-12 20:00:48,459 - mmdet - INFO - Iter [64/17500] lr: 1.252e-04, eta: 16:19:17, time: 1.564, data_time: 0.076, memory: 49167, loss_cls_0: 0.9526, loss_box_0: 1.8526, loss_cns_0: 0.6163, loss_yns_0: 0.1606, loss_cls_1: 1.0582, loss_box_1: 2.0976, loss_cns_1: 0.6352, loss_yns_1: 0.1654, loss_cls_2: 1.1599, loss_box_2: 2.1203, loss_cns_2: 0.6451, loss_yns_2: 0.1651, loss_cls_3: 1.1822, loss_box_3: 2.0967, loss_cns_3: 0.6486, loss_yns_3: 0.1619, loss_cls_4: 1.1880, loss_box_4: 2.1249, loss_cns_4: 0.6528, loss_yns_4: 0.1632, loss_cls_5: 1.2009, loss_box_5: 2.1211, loss_cns_5: 0.6502, loss_yns_5: 0.1657, loss_cls_dn_0: 0.3793, loss_box_dn_0: 0.8591, loss_cls_dn_1: 0.2466, loss_box_dn_1: 0.8673, loss_cls_dn_2: 0.2746, loss_box_dn_2: 0.8720, loss_cls_dn_3: 0.2806, loss_box_dn_3: 0.8790, loss_cls_dn_4: 0.2673, loss_box_dn_4: 0.9050, loss_cls_dn_5: 0.2986, loss_box_dn_5: 0.9183, loss_dense_depth: 0.8714, loss: 31.9040, grad_norm: 85.0247 -2025-11-12 20:00:50,040 - mmdet - INFO - Iter [65/17500] lr: 1.256e-04, eta: 16:11:14, time: 1.582, data_time: 0.098, memory: 49167, loss_cls_0: 0.9623, loss_box_0: 1.8511, loss_cns_0: 0.6148, loss_yns_0: 0.1619, loss_cls_1: 1.0468, loss_box_1: 2.0390, loss_cns_1: 0.6354, loss_yns_1: 0.1628, loss_cls_2: 1.0938, loss_box_2: 2.0272, loss_cns_2: 0.6471, loss_yns_2: 0.1617, loss_cls_3: 1.1059, loss_box_3: 2.0173, loss_cns_3: 0.6494, loss_yns_3: 0.1610, loss_cls_4: 1.0940, loss_box_4: 2.0122, loss_cns_4: 0.6525, loss_yns_4: 0.1619, loss_cls_5: 1.1004, loss_box_5: 2.0140, loss_cns_5: 0.6489, loss_yns_5: 0.1626, loss_cls_dn_0: 0.3801, loss_box_dn_0: 0.8537, loss_cls_dn_1: 0.2359, loss_box_dn_1: 0.9254, loss_cls_dn_2: 0.2579, loss_box_dn_2: 0.9183, loss_cls_dn_3: 0.2670, loss_box_dn_3: 0.9274, loss_cls_dn_4: 0.2676, loss_box_dn_4: 0.9451, loss_cls_dn_5: 0.2875, loss_box_dn_5: 0.9647, loss_dense_depth: 0.8596, loss: 31.2741, grad_norm: 53.9865 -2025-11-12 20:00:51,611 - mmdet - INFO - Iter [66/17500] lr: 1.260e-04, eta: 16:03:22, time: 1.571, data_time: 0.073, memory: 49167, loss_cls_0: 0.9698, loss_box_0: 1.8414, loss_cns_0: 0.6227, loss_yns_0: 0.1640, loss_cls_1: 1.0761, loss_box_1: 2.0565, loss_cns_1: 0.6332, loss_yns_1: 0.1630, loss_cls_2: 1.0840, loss_box_2: 2.0309, loss_cns_2: 0.6505, loss_yns_2: 0.1647, loss_cls_3: 1.0854, loss_box_3: 2.0689, loss_cns_3: 0.6499, loss_yns_3: 0.1623, loss_cls_4: 1.0976, loss_box_4: 2.0124, loss_cns_4: 0.6509, loss_yns_4: 0.1636, loss_cls_5: 1.0922, loss_box_5: 2.0198, loss_cns_5: 0.6507, loss_yns_5: 0.1669, loss_cls_dn_0: 0.3658, loss_box_dn_0: 0.8394, loss_cls_dn_1: 0.2327, loss_box_dn_1: 0.9747, loss_cls_dn_2: 0.2655, loss_box_dn_2: 0.9550, loss_cls_dn_3: 0.2821, loss_box_dn_3: 0.9734, loss_cls_dn_4: 0.3064, loss_box_dn_4: 0.9769, loss_cls_dn_5: 0.3143, loss_box_dn_5: 0.9940, loss_dense_depth: 0.8443, loss: 31.6018, grad_norm: 48.0229 -2025-11-12 20:00:53,186 - mmdet - INFO - Iter [67/17500] lr: 1.264e-04, eta: 15:55:46, time: 1.575, data_time: 0.073, memory: 49167, loss_cls_0: 0.9704, loss_box_0: 1.8692, loss_cns_0: 0.6201, loss_yns_0: 0.1599, loss_cls_1: 1.1341, loss_box_1: 2.1628, loss_cns_1: 0.6215, loss_yns_1: 0.1637, loss_cls_2: 1.1494, loss_box_2: 2.1448, loss_cns_2: 0.6442, loss_yns_2: 0.1629, loss_cls_3: 1.1399, loss_box_3: 2.1809, loss_cns_3: 0.6451, loss_yns_3: 0.1599, loss_cls_4: 1.1314, loss_box_4: 2.1411, loss_cns_4: 0.6457, loss_yns_4: 0.1618, loss_cls_5: 1.1375, loss_box_5: 2.1556, loss_cns_5: 0.6461, loss_yns_5: 0.1628, loss_cls_dn_0: 0.3660, loss_box_dn_0: 0.8442, loss_cls_dn_1: 0.2535, loss_box_dn_1: 0.9505, loss_cls_dn_2: 0.2994, loss_box_dn_2: 0.9260, loss_cls_dn_3: 0.3187, loss_box_dn_3: 0.9483, loss_cls_dn_4: 0.3280, loss_box_dn_4: 0.9557, loss_cls_dn_5: 0.3501, loss_box_dn_5: 0.9754, loss_dense_depth: 0.8750, loss: 32.5012, grad_norm: 63.5033 -2025-11-12 20:00:54,773 - mmdet - INFO - Iter [68/17500] lr: 1.268e-04, eta: 15:48:23, time: 1.573, data_time: 0.080, memory: 49167, loss_cls_0: 0.9695, loss_box_0: 1.8757, loss_cns_0: 0.6227, loss_yns_0: 0.1587, loss_cls_1: 1.1035, loss_box_1: 2.1332, loss_cns_1: 0.6208, loss_yns_1: 0.1633, loss_cls_2: 1.1329, loss_box_2: 2.1026, loss_cns_2: 0.6379, loss_yns_2: 0.1655, loss_cls_3: 1.1253, loss_box_3: 2.1199, loss_cns_3: 0.6421, loss_yns_3: 0.1609, loss_cls_4: 1.1037, loss_box_4: 2.0956, loss_cns_4: 0.6434, loss_yns_4: 0.1645, loss_cls_5: 1.1359, loss_box_5: 2.1185, loss_cns_5: 0.6437, loss_yns_5: 0.1608, loss_cls_dn_0: 0.3557, loss_box_dn_0: 0.8621, loss_cls_dn_1: 0.2665, loss_box_dn_1: 0.9264, loss_cls_dn_2: 0.3176, loss_box_dn_2: 0.8965, loss_cls_dn_3: 0.3352, loss_box_dn_3: 0.9110, loss_cls_dn_4: 0.3128, loss_box_dn_4: 0.9216, loss_cls_dn_5: 0.3592, loss_box_dn_5: 0.9463, loss_dense_depth: 0.8530, loss: 32.0646, grad_norm: 60.6957 -2025-11-12 20:00:56,323 - mmdet - INFO - Iter [69/17500] lr: 1.272e-04, eta: 15:41:08, time: 1.557, data_time: 0.087, memory: 49167, loss_cls_0: 0.9818, loss_box_0: 1.8757, loss_cns_0: 0.6212, loss_yns_0: 0.1574, loss_cls_1: 1.0330, loss_box_1: 2.0471, loss_cns_1: 0.6225, loss_yns_1: 0.1599, loss_cls_2: 1.0844, loss_box_2: 2.0002, loss_cns_2: 0.6381, loss_yns_2: 0.1636, loss_cls_3: 1.0813, loss_box_3: 1.9644, loss_cns_3: 0.6477, loss_yns_3: 0.1607, loss_cls_4: 1.1563, loss_box_4: 1.9455, loss_cns_4: 0.6516, loss_yns_4: 0.1643, loss_cls_5: 1.0974, loss_box_5: 1.9662, loss_cns_5: 0.6500, loss_yns_5: 0.1599, loss_cls_dn_0: 0.3364, loss_box_dn_0: 0.8675, loss_cls_dn_1: 0.2500, loss_box_dn_1: 0.9101, loss_cls_dn_2: 0.2981, loss_box_dn_2: 0.8758, loss_cls_dn_3: 0.3148, loss_box_dn_3: 0.8730, loss_cls_dn_4: 0.2752, loss_box_dn_4: 0.8835, loss_cls_dn_5: 0.3367, loss_box_dn_5: 0.9034, loss_dense_depth: 0.8154, loss: 30.9701, grad_norm: 45.4076 -2025-11-12 20:00:57,893 - mmdet - INFO - Iter [70/17500] lr: 1.276e-04, eta: 15:34:07, time: 1.562, data_time: 0.079, memory: 49167, loss_cls_0: 0.9771, loss_box_0: 1.8632, loss_cns_0: 0.6200, loss_yns_0: 0.1584, loss_cls_1: 0.9889, loss_box_1: 2.0560, loss_cns_1: 0.6205, loss_yns_1: 0.1580, loss_cls_2: 1.0352, loss_box_2: 2.0143, loss_cns_2: 0.6364, loss_yns_2: 0.1598, loss_cls_3: 1.0433, loss_box_3: 1.9854, loss_cns_3: 0.6427, loss_yns_3: 0.1588, loss_cls_4: 1.0504, loss_box_4: 1.9926, loss_cns_4: 0.6495, loss_yns_4: 0.1599, loss_cls_5: 1.0477, loss_box_5: 2.0021, loss_cns_5: 0.6444, loss_yns_5: 0.1613, loss_cls_dn_0: 0.3582, loss_box_dn_0: 0.8761, loss_cls_dn_1: 0.2476, loss_box_dn_1: 0.8564, loss_cls_dn_2: 0.2944, loss_box_dn_2: 0.8302, loss_cls_dn_3: 0.3188, loss_box_dn_3: 0.8266, loss_cls_dn_4: 0.2948, loss_box_dn_4: 0.8452, loss_cls_dn_5: 0.3511, loss_box_dn_5: 0.8604, loss_dense_depth: 0.8644, loss: 30.6501, grad_norm: 50.1670 -2025-11-12 20:00:59,463 - mmdet - INFO - Iter [71/17500] lr: 1.280e-04, eta: 15:27:22, time: 1.577, data_time: 0.086, memory: 49167, loss_cls_0: 0.9327, loss_box_0: 1.8095, loss_cns_0: 0.6236, loss_yns_0: 0.1582, loss_cls_1: 1.0197, loss_box_1: 2.0951, loss_cns_1: 0.6191, loss_yns_1: 0.1585, loss_cls_2: 1.0325, loss_box_2: 2.0404, loss_cns_2: 0.6394, loss_yns_2: 0.1564, loss_cls_3: 1.0441, loss_box_3: 2.0265, loss_cns_3: 0.6487, loss_yns_3: 0.1585, loss_cls_4: 1.0572, loss_box_4: 2.0151, loss_cns_4: 0.6554, loss_yns_4: 0.1598, loss_cls_5: 1.0585, loss_box_5: 2.0277, loss_cns_5: 0.6488, loss_yns_5: 0.1583, loss_cls_dn_0: 0.3580, loss_box_dn_0: 0.8719, loss_cls_dn_1: 0.2375, loss_box_dn_1: 0.8927, loss_cls_dn_2: 0.2773, loss_box_dn_2: 0.8616, loss_cls_dn_3: 0.2989, loss_box_dn_3: 0.8619, loss_cls_dn_4: 0.2940, loss_box_dn_4: 0.8662, loss_cls_dn_5: 0.3353, loss_box_dn_5: 0.8755, loss_dense_depth: 0.8632, loss: 30.8379, grad_norm: 59.4133 -2025-11-12 20:01:01,021 - mmdet - INFO - Iter [72/17500] lr: 1.284e-04, eta: 15:20:43, time: 1.558, data_time: 0.083, memory: 49167, loss_cls_0: 0.9386, loss_box_0: 1.8005, loss_cns_0: 0.6247, loss_yns_0: 0.1589, loss_cls_1: 1.0488, loss_box_1: 2.0741, loss_cns_1: 0.6185, loss_yns_1: 0.1635, loss_cls_2: 1.0774, loss_box_2: 1.9829, loss_cns_2: 0.6397, loss_yns_2: 0.1607, loss_cls_3: 1.0660, loss_box_3: 1.9683, loss_cns_3: 0.6471, loss_yns_3: 0.1590, loss_cls_4: 1.0612, loss_box_4: 1.9306, loss_cns_4: 0.6504, loss_yns_4: 0.1639, loss_cls_5: 1.0702, loss_box_5: 1.9551, loss_cns_5: 0.6481, loss_yns_5: 0.1580, loss_cls_dn_0: 0.3645, loss_box_dn_0: 0.8624, loss_cls_dn_1: 0.2417, loss_box_dn_1: 0.8758, loss_cls_dn_2: 0.2704, loss_box_dn_2: 0.8485, loss_cls_dn_3: 0.2864, loss_box_dn_3: 0.8473, loss_cls_dn_4: 0.2881, loss_box_dn_4: 0.8335, loss_cls_dn_5: 0.3189, loss_box_dn_5: 0.8505, loss_dense_depth: 0.8557, loss: 30.5096, grad_norm: 59.1347 -2025-11-12 20:01:02,597 - mmdet - INFO - Iter [73/17500] lr: 1.288e-04, eta: 15:14:19, time: 1.572, data_time: 0.080, memory: 49167, loss_cls_0: 0.9639, loss_box_0: 1.8603, loss_cns_0: 0.6136, loss_yns_0: 0.1591, loss_cls_1: 1.0214, loss_box_1: 2.0704, loss_cns_1: 0.6175, loss_yns_1: 0.1642, loss_cls_2: 1.0574, loss_box_2: 1.9665, loss_cns_2: 0.6417, loss_yns_2: 0.1633, loss_cls_3: 1.0754, loss_box_3: 1.9613, loss_cns_3: 0.6480, loss_yns_3: 0.1592, loss_cls_4: 1.0502, loss_box_4: 1.9518, loss_cns_4: 0.6499, loss_yns_4: 0.1663, loss_cls_5: 1.0899, loss_box_5: 1.9478, loss_cns_5: 0.6480, loss_yns_5: 0.1595, loss_cls_dn_0: 0.3687, loss_box_dn_0: 0.8652, loss_cls_dn_1: 0.2371, loss_box_dn_1: 0.8702, loss_cls_dn_2: 0.2644, loss_box_dn_2: 0.8453, loss_cls_dn_3: 0.2757, loss_box_dn_3: 0.8495, loss_cls_dn_4: 0.2820, loss_box_dn_4: 0.8483, loss_cls_dn_5: 0.2972, loss_box_dn_5: 0.8633, loss_dense_depth: 0.8609, loss: 30.5343, grad_norm: 48.8488 -2025-11-12 20:01:04,165 - mmdet - INFO - Iter [74/17500] lr: 1.292e-04, eta: 15:08:04, time: 1.572, data_time: 0.085, memory: 49167, loss_cls_0: 0.9532, loss_box_0: 1.8510, loss_cns_0: 0.6158, loss_yns_0: 0.1584, loss_cls_1: 1.0313, loss_box_1: 2.0762, loss_cns_1: 0.6275, loss_yns_1: 0.1622, loss_cls_2: 1.0577, loss_box_2: 2.0011, loss_cns_2: 0.6490, loss_yns_2: 0.1637, loss_cls_3: 1.0572, loss_box_3: 2.0096, loss_cns_3: 0.6561, loss_yns_3: 0.1625, loss_cls_4: 1.0543, loss_box_4: 2.0340, loss_cns_4: 0.6544, loss_yns_4: 0.1629, loss_cls_5: 1.0653, loss_box_5: 2.0025, loss_cns_5: 0.6504, loss_yns_5: 0.1648, loss_cls_dn_0: 0.3773, loss_box_dn_0: 0.8577, loss_cls_dn_1: 0.2308, loss_box_dn_1: 0.8823, loss_cls_dn_2: 0.2617, loss_box_dn_2: 0.8698, loss_cls_dn_3: 0.2816, loss_box_dn_3: 0.8740, loss_cls_dn_4: 0.2964, loss_box_dn_4: 0.8886, loss_cls_dn_5: 0.3008, loss_box_dn_5: 0.9023, loss_dense_depth: 0.8579, loss: 30.9023, grad_norm: 46.4073 -2025-11-12 20:01:05,732 - mmdet - INFO - Iter [75/17500] lr: 1.296e-04, eta: 15:01:58, time: 1.563, data_time: 0.082, memory: 49167, loss_cls_0: 0.9487, loss_box_0: 1.8401, loss_cns_0: 0.6138, loss_yns_0: 0.1595, loss_cls_1: 1.0308, loss_box_1: 2.1078, loss_cns_1: 0.6345, loss_yns_1: 0.1631, loss_cls_2: 1.0616, loss_box_2: 2.0683, loss_cns_2: 0.6449, loss_yns_2: 0.1627, loss_cls_3: 1.0675, loss_box_3: 2.0386, loss_cns_3: 0.6525, loss_yns_3: 0.1642, loss_cls_4: 1.0655, loss_box_4: 2.0403, loss_cns_4: 0.6542, loss_yns_4: 0.1615, loss_cls_5: 1.0577, loss_box_5: 2.0152, loss_cns_5: 0.6503, loss_yns_5: 0.1650, loss_cls_dn_0: 0.3609, loss_box_dn_0: 0.8435, loss_cls_dn_1: 0.2258, loss_box_dn_1: 0.8956, loss_cls_dn_2: 0.2549, loss_box_dn_2: 0.8906, loss_cls_dn_3: 0.2794, loss_box_dn_3: 0.8771, loss_cls_dn_4: 0.2908, loss_box_dn_4: 0.8899, loss_cls_dn_5: 0.2956, loss_box_dn_5: 0.9036, loss_dense_depth: 0.8265, loss: 31.0025, grad_norm: 45.0613 -2025-11-12 20:01:07,292 - mmdet - INFO - Iter [76/17500] lr: 1.300e-04, eta: 14:56:00, time: 1.557, data_time: 0.084, memory: 49167, loss_cls_0: 0.9641, loss_box_0: 1.8354, loss_cns_0: 0.6191, loss_yns_0: 0.1617, loss_cls_1: 1.0222, loss_box_1: 2.1513, loss_cns_1: 0.6340, loss_yns_1: 0.1652, loss_cls_2: 1.0575, loss_box_2: 2.1060, loss_cns_2: 0.6450, loss_yns_2: 0.1629, loss_cls_3: 1.0662, loss_box_3: 2.0733, loss_cns_3: 0.6513, loss_yns_3: 0.1639, loss_cls_4: 1.0551, loss_box_4: 2.0424, loss_cns_4: 0.6561, loss_yns_4: 0.1676, loss_cls_5: 1.0810, loss_box_5: 2.0273, loss_cns_5: 0.6528, loss_yns_5: 0.1641, loss_cls_dn_0: 0.3390, loss_box_dn_0: 0.8443, loss_cls_dn_1: 0.2280, loss_box_dn_1: 0.8636, loss_cls_dn_2: 0.2485, loss_box_dn_2: 0.8515, loss_cls_dn_3: 0.2622, loss_box_dn_3: 0.8358, loss_cls_dn_4: 0.2617, loss_box_dn_4: 0.8403, loss_cls_dn_5: 0.2780, loss_box_dn_5: 0.8445, loss_dense_depth: 0.8240, loss: 30.8468, grad_norm: 40.6588 -2025-11-12 20:01:08,857 - mmdet - INFO - Iter [77/17500] lr: 1.304e-04, eta: 14:50:14, time: 1.571, data_time: 0.085, memory: 49167, loss_cls_0: 0.9611, loss_box_0: 1.8749, loss_cns_0: 0.6157, loss_yns_0: 0.1601, loss_cls_1: 1.0454, loss_box_1: 2.1013, loss_cns_1: 0.6413, loss_yns_1: 0.1650, loss_cls_2: 1.0494, loss_box_2: 2.0534, loss_cns_2: 0.6523, loss_yns_2: 0.1620, loss_cls_3: 1.0702, loss_box_3: 2.0396, loss_cns_3: 0.6530, loss_yns_3: 0.1618, loss_cls_4: 1.0690, loss_box_4: 2.0102, loss_cns_4: 0.6569, loss_yns_4: 0.1675, loss_cls_5: 1.0594, loss_box_5: 2.0032, loss_cns_5: 0.6543, loss_yns_5: 0.1633, loss_cls_dn_0: 0.3379, loss_box_dn_0: 0.8332, loss_cls_dn_1: 0.2288, loss_box_dn_1: 0.8147, loss_cls_dn_2: 0.2439, loss_box_dn_2: 0.7967, loss_cls_dn_3: 0.2569, loss_box_dn_3: 0.7859, loss_cls_dn_4: 0.2504, loss_box_dn_4: 0.7870, loss_cls_dn_5: 0.2804, loss_box_dn_5: 0.7899, loss_dense_depth: 0.8662, loss: 30.4623, grad_norm: 45.8695 -2025-11-12 20:01:10,421 - mmdet - INFO - Iter [78/17500] lr: 1.308e-04, eta: 14:44:36, time: 1.567, data_time: 0.083, memory: 49167, loss_cls_0: 0.9710, loss_box_0: 1.8733, loss_cns_0: 0.6188, loss_yns_0: 0.1629, loss_cls_1: 1.0274, loss_box_1: 2.1506, loss_cns_1: 0.6307, loss_yns_1: 0.1651, loss_cls_2: 1.0453, loss_box_2: 2.0824, loss_cns_2: 0.6450, loss_yns_2: 0.1636, loss_cls_3: 1.0556, loss_box_3: 2.0854, loss_cns_3: 0.6518, loss_yns_3: 0.1647, loss_cls_4: 1.0562, loss_box_4: 2.0686, loss_cns_4: 0.6511, loss_yns_4: 0.1680, loss_cls_5: 1.0673, loss_box_5: 2.0697, loss_cns_5: 0.6501, loss_yns_5: 0.1659, loss_cls_dn_0: 0.3532, loss_box_dn_0: 0.8250, loss_cls_dn_1: 0.2160, loss_box_dn_1: 0.8115, loss_cls_dn_2: 0.2383, loss_box_dn_2: 0.7879, loss_cls_dn_3: 0.2532, loss_box_dn_3: 0.7949, loss_cls_dn_4: 0.2440, loss_box_dn_4: 0.8002, loss_cls_dn_5: 0.2868, loss_box_dn_5: 0.8087, loss_dense_depth: 0.8462, loss: 30.6565, grad_norm: 36.6567 -2025-11-12 20:01:11,990 - mmdet - INFO - Iter [79/17500] lr: 1.312e-04, eta: 14:39:07, time: 1.566, data_time: 0.080, memory: 49167, loss_cls_0: 0.9316, loss_box_0: 1.8561, loss_cns_0: 0.6259, loss_yns_0: 0.1606, loss_cls_1: 1.0100, loss_box_1: 2.0743, loss_cns_1: 0.6377, loss_yns_1: 0.1630, loss_cls_2: 1.0293, loss_box_2: 2.0562, loss_cns_2: 0.6506, loss_yns_2: 0.1621, loss_cls_3: 1.0364, loss_box_3: 2.0386, loss_cns_3: 0.6557, loss_yns_3: 0.1639, loss_cls_4: 1.0437, loss_box_4: 2.0152, loss_cns_4: 0.6525, loss_yns_4: 0.1635, loss_cls_5: 1.0390, loss_box_5: 2.0048, loss_cns_5: 0.6523, loss_yns_5: 0.1657, loss_cls_dn_0: 0.3402, loss_box_dn_0: 0.8268, loss_cls_dn_1: 0.2110, loss_box_dn_1: 0.8207, loss_cls_dn_2: 0.2324, loss_box_dn_2: 0.8126, loss_cls_dn_3: 0.2529, loss_box_dn_3: 0.8203, loss_cls_dn_4: 0.2356, loss_box_dn_4: 0.8306, loss_cls_dn_5: 0.2687, loss_box_dn_5: 0.8408, loss_dense_depth: 0.8313, loss: 30.3124, grad_norm: 47.2708 -2025-11-12 20:01:13,549 - mmdet - INFO - Iter [80/17500] lr: 1.316e-04, eta: 14:33:43, time: 1.557, data_time: 0.080, memory: 49167, loss_cls_0: 0.9308, loss_box_0: 1.8531, loss_cns_0: 0.6259, loss_yns_0: 0.1598, loss_cls_1: 0.9990, loss_box_1: 2.0677, loss_cns_1: 0.6413, loss_yns_1: 0.1611, loss_cls_2: 1.0630, loss_box_2: 2.0634, loss_cns_2: 0.6504, loss_yns_2: 0.1615, loss_cls_3: 1.0581, loss_box_3: 2.0128, loss_cns_3: 0.6546, loss_yns_3: 0.1613, loss_cls_4: 1.0659, loss_box_4: 2.0061, loss_cns_4: 0.6535, loss_yns_4: 0.1699, loss_cls_5: 1.0749, loss_box_5: 1.9926, loss_cns_5: 0.6516, loss_yns_5: 0.1622, loss_cls_dn_0: 0.3311, loss_box_dn_0: 0.8242, loss_cls_dn_1: 0.2068, loss_box_dn_1: 0.8454, loss_cls_dn_2: 0.2294, loss_box_dn_2: 0.8453, loss_cls_dn_3: 0.2489, loss_box_dn_3: 0.8328, loss_cls_dn_4: 0.2291, loss_box_dn_4: 0.8522, loss_cls_dn_5: 0.2497, loss_box_dn_5: 0.8652, loss_dense_depth: 0.8021, loss: 30.4026, grad_norm: 55.6575 -2025-11-12 20:01:15,164 - mmdet - INFO - Iter [81/17500] lr: 1.320e-04, eta: 14:28:42, time: 1.623, data_time: 0.093, memory: 49167, loss_cls_0: 0.9470, loss_box_0: 1.8764, loss_cns_0: 0.6211, loss_yns_0: 0.1619, loss_cls_1: 1.0123, loss_box_1: 2.0277, loss_cns_1: 0.6403, loss_yns_1: 0.1607, loss_cls_2: 1.0587, loss_box_2: 2.0151, loss_cns_2: 0.6474, loss_yns_2: 0.1623, loss_cls_3: 1.0631, loss_box_3: 1.9948, loss_cns_3: 0.6512, loss_yns_3: 0.1613, loss_cls_4: 1.1071, loss_box_4: 1.9908, loss_cns_4: 0.6473, loss_yns_4: 0.1762, loss_cls_5: 1.0512, loss_box_5: 1.9966, loss_cns_5: 0.6493, loss_yns_5: 0.1622, loss_cls_dn_0: 0.3379, loss_box_dn_0: 0.8272, loss_cls_dn_1: 0.2042, loss_box_dn_1: 0.8782, loss_cls_dn_2: 0.2213, loss_box_dn_2: 0.8697, loss_cls_dn_3: 0.2371, loss_box_dn_3: 0.8623, loss_cls_dn_4: 0.2319, loss_box_dn_4: 0.8877, loss_cls_dn_5: 0.2433, loss_box_dn_5: 0.8937, loss_dense_depth: 0.8642, loss: 30.5410, grad_norm: 43.9515 -2025-11-12 20:01:16,799 - mmdet - INFO - Iter [82/17500] lr: 1.324e-04, eta: 14:23:51, time: 1.635, data_time: 0.073, memory: 49167, loss_cls_0: 0.9573, loss_box_0: 1.8511, loss_cns_0: 0.6197, loss_yns_0: 0.1571, loss_cls_1: 1.0114, loss_box_1: 2.0612, loss_cns_1: 0.6364, loss_yns_1: 0.1582, loss_cls_2: 1.0596, loss_box_2: 2.0237, loss_cns_2: 0.6483, loss_yns_2: 0.1592, loss_cls_3: 1.0829, loss_box_3: 2.0541, loss_cns_3: 0.6536, loss_yns_3: 0.1576, loss_cls_4: 1.1267, loss_box_4: 2.0582, loss_cns_4: 0.6472, loss_yns_4: 0.1636, loss_cls_5: 1.0635, loss_box_5: 2.0370, loss_cns_5: 0.6494, loss_yns_5: 0.1579, loss_cls_dn_0: 0.3411, loss_box_dn_0: 0.8344, loss_cls_dn_1: 0.1995, loss_box_dn_1: 0.8891, loss_cls_dn_2: 0.2145, loss_box_dn_2: 0.8755, loss_cls_dn_3: 0.2273, loss_box_dn_3: 0.8876, loss_cls_dn_4: 0.2276, loss_box_dn_4: 0.9039, loss_cls_dn_5: 0.2353, loss_box_dn_5: 0.9004, loss_dense_depth: 0.9078, loss: 30.8389, grad_norm: 45.2964 -2025-11-12 20:01:18,405 - mmdet - INFO - Iter [83/17500] lr: 1.328e-04, eta: 14:19:01, time: 1.608, data_time: 0.136, memory: 49167, loss_cls_0: 0.9516, loss_box_0: 1.8003, loss_cns_0: 0.6261, loss_yns_0: 0.1594, loss_cls_1: 1.0274, loss_box_1: 1.9819, loss_cns_1: 0.6326, loss_yns_1: 0.1591, loss_cls_2: 1.0393, loss_box_2: 1.9375, loss_cns_2: 0.6450, loss_yns_2: 0.1586, loss_cls_3: 1.0540, loss_box_3: 1.9535, loss_cns_3: 0.6517, loss_yns_3: 0.1589, loss_cls_4: 1.0633, loss_box_4: 1.9401, loss_cns_4: 0.6519, loss_yns_4: 0.1598, loss_cls_5: 1.0744, loss_box_5: 1.9422, loss_cns_5: 0.6477, loss_yns_5: 0.1595, loss_cls_dn_0: 0.3290, loss_box_dn_0: 0.8341, loss_cls_dn_1: 0.2018, loss_box_dn_1: 0.8485, loss_cls_dn_2: 0.2182, loss_box_dn_2: 0.8349, loss_cls_dn_3: 0.2196, loss_box_dn_3: 0.8394, loss_cls_dn_4: 0.2216, loss_box_dn_4: 0.8407, loss_cls_dn_5: 0.2363, loss_box_dn_5: 0.8502, loss_dense_depth: 0.8428, loss: 29.8929, grad_norm: 53.7565 -2025-11-12 20:01:19,951 - mmdet - INFO - Iter [84/17500] lr: 1.332e-04, eta: 14:14:03, time: 1.539, data_time: 0.073, memory: 49167, loss_cls_0: 0.9027, loss_box_0: 1.7919, loss_cns_0: 0.6269, loss_yns_0: 0.1558, loss_cls_1: 1.0404, loss_box_1: 1.9365, loss_cns_1: 0.6438, loss_yns_1: 0.1590, loss_cls_2: 1.0593, loss_box_2: 1.9153, loss_cns_2: 0.6500, loss_yns_2: 0.1571, loss_cls_3: 1.0588, loss_box_3: 1.9099, loss_cns_3: 0.6530, loss_yns_3: 0.1584, loss_cls_4: 1.0768, loss_box_4: 1.9030, loss_cns_4: 0.6555, loss_yns_4: 0.1670, loss_cls_5: 1.0374, loss_box_5: 1.9080, loss_cns_5: 0.6518, loss_yns_5: 0.1597, loss_cls_dn_0: 0.3190, loss_box_dn_0: 0.8238, loss_cls_dn_1: 0.2036, loss_box_dn_1: 0.7904, loss_cls_dn_2: 0.2211, loss_box_dn_2: 0.7803, loss_cls_dn_3: 0.2213, loss_box_dn_3: 0.7734, loss_cls_dn_4: 0.2228, loss_box_dn_4: 0.7800, loss_cls_dn_5: 0.2408, loss_box_dn_5: 0.7973, loss_dense_depth: 0.8411, loss: 29.3930, grad_norm: 59.7771 -2025-11-12 20:01:21,518 - mmdet - INFO - Iter [85/17500] lr: 1.336e-04, eta: 14:09:19, time: 1.568, data_time: 0.105, memory: 49167, loss_cls_0: 0.9407, loss_box_0: 1.8265, loss_cns_0: 0.6196, loss_yns_0: 0.1569, loss_cls_1: 1.0071, loss_box_1: 1.9469, loss_cns_1: 0.6422, loss_yns_1: 0.1602, loss_cls_2: 1.0445, loss_box_2: 1.9330, loss_cns_2: 0.6467, loss_yns_2: 0.1584, loss_cls_3: 1.0484, loss_box_3: 1.9507, loss_cns_3: 0.6479, loss_yns_3: 0.1610, loss_cls_4: 1.0636, loss_box_4: 1.9467, loss_cns_4: 0.6535, loss_yns_4: 0.1686, loss_cls_5: 1.1149, loss_box_5: 1.9339, loss_cns_5: 0.6479, loss_yns_5: 0.1568, loss_cls_dn_0: 0.3499, loss_box_dn_0: 0.8424, loss_cls_dn_1: 0.2075, loss_box_dn_1: 0.8003, loss_cls_dn_2: 0.2286, loss_box_dn_2: 0.7928, loss_cls_dn_3: 0.2356, loss_box_dn_3: 0.8040, loss_cls_dn_4: 0.2345, loss_box_dn_4: 0.8187, loss_cls_dn_5: 0.2599, loss_box_dn_5: 0.8188, loss_dense_depth: 0.8415, loss: 29.8112, grad_norm: 52.5056 -2025-11-12 20:01:23,099 - mmdet - INFO - Iter [86/17500] lr: 1.340e-04, eta: 14:04:44, time: 1.582, data_time: 0.079, memory: 49167, loss_cls_0: 0.9228, loss_box_0: 1.8070, loss_cns_0: 0.6223, loss_yns_0: 0.1604, loss_cls_1: 1.0442, loss_box_1: 1.8829, loss_cns_1: 0.6275, loss_yns_1: 0.1584, loss_cls_2: 1.0705, loss_box_2: 1.8766, loss_cns_2: 0.6407, loss_yns_2: 0.1622, loss_cls_3: 1.0407, loss_box_3: 1.9014, loss_cns_3: 0.6471, loss_yns_3: 0.1609, loss_cls_4: 1.0665, loss_box_4: 1.8984, loss_cns_4: 0.6506, loss_yns_4: 0.1645, loss_cls_5: 1.0818, loss_box_5: 1.8805, loss_cns_5: 0.6471, loss_yns_5: 0.1633, loss_cls_dn_0: 0.3234, loss_box_dn_0: 0.8295, loss_cls_dn_1: 0.2101, loss_box_dn_1: 0.7980, loss_cls_dn_2: 0.2305, loss_box_dn_2: 0.7956, loss_cls_dn_3: 0.2376, loss_box_dn_3: 0.8070, loss_cls_dn_4: 0.2432, loss_box_dn_4: 0.8180, loss_cls_dn_5: 0.2584, loss_box_dn_5: 0.8162, loss_dense_depth: 0.8666, loss: 29.5123, grad_norm: 49.0762 -2025-11-12 20:01:24,675 - mmdet - INFO - Iter [87/17500] lr: 1.344e-04, eta: 14:00:12, time: 1.569, data_time: 0.080, memory: 49167, loss_cls_0: 0.9547, loss_box_0: 1.7970, loss_cns_0: 0.6236, loss_yns_0: 0.1604, loss_cls_1: 0.9962, loss_box_1: 1.9390, loss_cns_1: 0.6288, loss_yns_1: 0.1594, loss_cls_2: 1.0317, loss_box_2: 1.9108, loss_cns_2: 0.6457, loss_yns_2: 0.1622, loss_cls_3: 1.0250, loss_box_3: 1.8875, loss_cns_3: 0.6514, loss_yns_3: 0.1596, loss_cls_4: 1.0278, loss_box_4: 1.8879, loss_cns_4: 0.6524, loss_yns_4: 0.1590, loss_cls_5: 1.0491, loss_box_5: 1.8998, loss_cns_5: 0.6503, loss_yns_5: 0.1619, loss_cls_dn_0: 0.2883, loss_box_dn_0: 0.8160, loss_cls_dn_1: 0.2069, loss_box_dn_1: 0.7990, loss_cls_dn_2: 0.2228, loss_box_dn_2: 0.7963, loss_cls_dn_3: 0.2295, loss_box_dn_3: 0.7885, loss_cls_dn_4: 0.2297, loss_box_dn_4: 0.7936, loss_cls_dn_5: 0.2342, loss_box_dn_5: 0.8112, loss_dense_depth: 0.8195, loss: 29.2567, grad_norm: 43.3283 -2025-11-12 20:01:26,239 - mmdet - INFO - Iter [88/17500] lr: 1.348e-04, eta: 13:55:47, time: 1.566, data_time: 0.086, memory: 49167, loss_cls_0: 0.9504, loss_box_0: 1.7785, loss_cns_0: 0.6207, loss_yns_0: 0.1578, loss_cls_1: 0.9778, loss_box_1: 1.9413, loss_cns_1: 0.6328, loss_yns_1: 0.1603, loss_cls_2: 1.0282, loss_box_2: 1.9179, loss_cns_2: 0.6487, loss_yns_2: 0.1578, loss_cls_3: 1.0711, loss_box_3: 1.8851, loss_cns_3: 0.6531, loss_yns_3: 0.1572, loss_cls_4: 1.0569, loss_box_4: 1.8645, loss_cns_4: 0.6550, loss_yns_4: 0.1593, loss_cls_5: 1.1341, loss_box_5: 1.8818, loss_cns_5: 0.6521, loss_yns_5: 0.1586, loss_cls_dn_0: 0.2917, loss_box_dn_0: 0.8225, loss_cls_dn_1: 0.2017, loss_box_dn_1: 0.7978, loss_cls_dn_2: 0.2122, loss_box_dn_2: 0.8028, loss_cls_dn_3: 0.2230, loss_box_dn_3: 0.7886, loss_cls_dn_4: 0.2234, loss_box_dn_4: 0.7863, loss_cls_dn_5: 0.2337, loss_box_dn_5: 0.8106, loss_dense_depth: 0.8688, loss: 29.3641, grad_norm: 71.4081 -2025-11-12 20:01:27,804 - mmdet - INFO - Iter [89/17500] lr: 1.352e-04, eta: 13:51:28, time: 1.572, data_time: 0.085, memory: 49167, loss_cls_0: 0.9081, loss_box_0: 1.7808, loss_cns_0: 0.6187, loss_yns_0: 0.1588, loss_cls_1: 0.9863, loss_box_1: 1.9240, loss_cns_1: 0.6300, loss_yns_1: 0.1586, loss_cls_2: 1.0260, loss_box_2: 1.8929, loss_cns_2: 0.6467, loss_yns_2: 0.1598, loss_cls_3: 1.0198, loss_box_3: 1.8831, loss_cns_3: 0.6501, loss_yns_3: 0.1571, loss_cls_4: 1.0330, loss_box_4: 1.8602, loss_cns_4: 0.6525, loss_yns_4: 0.1576, loss_cls_5: 1.0342, loss_box_5: 1.8953, loss_cns_5: 0.6481, loss_yns_5: 0.1628, loss_cls_dn_0: 0.3086, loss_box_dn_0: 0.8243, loss_cls_dn_1: 0.2019, loss_box_dn_1: 0.8118, loss_cls_dn_2: 0.2083, loss_box_dn_2: 0.8099, loss_cls_dn_3: 0.2223, loss_box_dn_3: 0.8010, loss_cls_dn_4: 0.2222, loss_box_dn_4: 0.7951, loss_cls_dn_5: 0.2409, loss_box_dn_5: 0.8194, loss_dense_depth: 0.8515, loss: 29.1620, grad_norm: 56.4350 -2025-11-12 20:01:29,361 - mmdet - INFO - Iter [90/17500] lr: 1.356e-04, eta: 13:47:12, time: 1.558, data_time: 0.071, memory: 49167, loss_cls_0: 0.9307, loss_box_0: 1.7955, loss_cns_0: 0.6141, loss_yns_0: 0.1555, loss_cls_1: 0.9929, loss_box_1: 1.9256, loss_cns_1: 0.6366, loss_yns_1: 0.1578, loss_cls_2: 1.0228, loss_box_2: 1.8662, loss_cns_2: 0.6527, loss_yns_2: 0.1605, loss_cls_3: 1.0338, loss_box_3: 1.8928, loss_cns_3: 0.6564, loss_yns_3: 0.1569, loss_cls_4: 1.0352, loss_box_4: 1.9220, loss_cns_4: 0.6532, loss_yns_4: 0.1626, loss_cls_5: 1.0545, loss_box_5: 1.8990, loss_cns_5: 0.6525, loss_yns_5: 0.1590, loss_cls_dn_0: 0.3312, loss_box_dn_0: 0.8424, loss_cls_dn_1: 0.2014, loss_box_dn_1: 0.8154, loss_cls_dn_2: 0.2093, loss_box_dn_2: 0.7924, loss_cls_dn_3: 0.2220, loss_box_dn_3: 0.7963, loss_cls_dn_4: 0.2212, loss_box_dn_4: 0.8141, loss_cls_dn_5: 0.2483, loss_box_dn_5: 0.8119, loss_dense_depth: 0.8398, loss: 29.3345, grad_norm: 45.0907 -2025-11-12 20:01:30,938 - mmdet - INFO - Iter [91/17500] lr: 1.360e-04, eta: 13:43:04, time: 1.571, data_time: 0.069, memory: 49167, loss_cls_0: 0.9112, loss_box_0: 1.7506, loss_cns_0: 0.6148, loss_yns_0: 0.1534, loss_cls_1: 0.9751, loss_box_1: 1.9613, loss_cns_1: 0.6365, loss_yns_1: 0.1564, loss_cls_2: 1.0148, loss_box_2: 1.8760, loss_cns_2: 0.6538, loss_yns_2: 0.1575, loss_cls_3: 1.0249, loss_box_3: 1.8761, loss_cns_3: 0.6597, loss_yns_3: 0.1557, loss_cls_4: 1.0561, loss_box_4: 1.9191, loss_cns_4: 0.6583, loss_yns_4: 0.1608, loss_cls_5: 1.0327, loss_box_5: 1.8950, loss_cns_5: 0.6498, loss_yns_5: 0.1550, loss_cls_dn_0: 0.3117, loss_box_dn_0: 0.8180, loss_cls_dn_1: 0.1997, loss_box_dn_1: 0.8119, loss_cls_dn_2: 0.2108, loss_box_dn_2: 0.7804, loss_cls_dn_3: 0.2151, loss_box_dn_3: 0.7830, loss_cls_dn_4: 0.2169, loss_box_dn_4: 0.8152, loss_cls_dn_5: 0.2363, loss_box_dn_5: 0.8063, loss_dense_depth: 0.8241, loss: 29.1337, grad_norm: 50.4439 -2025-11-12 20:01:32,500 - mmdet - INFO - Iter [92/17500] lr: 1.364e-04, eta: 13:39:00, time: 1.562, data_time: 0.082, memory: 49167, loss_cls_0: 0.9178, loss_box_0: 1.7611, loss_cns_0: 0.6176, loss_yns_0: 0.1517, loss_cls_1: 0.9754, loss_box_1: 2.0381, loss_cns_1: 0.6317, loss_yns_1: 0.1553, loss_cls_2: 1.0102, loss_box_2: 1.9532, loss_cns_2: 0.6471, loss_yns_2: 0.1562, loss_cls_3: 1.0322, loss_box_3: 1.9294, loss_cns_3: 0.6531, loss_yns_3: 0.1572, loss_cls_4: 1.0602, loss_box_4: 1.9675, loss_cns_4: 0.6562, loss_yns_4: 0.1588, loss_cls_5: 1.0416, loss_box_5: 1.9479, loss_cns_5: 0.6490, loss_yns_5: 0.1547, loss_cls_dn_0: 0.2912, loss_box_dn_0: 0.8120, loss_cls_dn_1: 0.1975, loss_box_dn_1: 0.8139, loss_cls_dn_2: 0.2040, loss_box_dn_2: 0.7918, loss_cls_dn_3: 0.2111, loss_box_dn_3: 0.7880, loss_cls_dn_4: 0.2152, loss_box_dn_4: 0.8202, loss_cls_dn_5: 0.2279, loss_box_dn_5: 0.8174, loss_dense_depth: 0.8302, loss: 29.4437, grad_norm: 61.3712 -2025-11-12 20:01:34,061 - mmdet - INFO - Iter [93/17500] lr: 1.368e-04, eta: 13:35:01, time: 1.557, data_time: 0.077, memory: 49167, loss_cls_0: 0.9252, loss_box_0: 1.7475, loss_cns_0: 0.6225, loss_yns_0: 0.1503, loss_cls_1: 0.9941, loss_box_1: 1.9831, loss_cns_1: 0.6382, loss_yns_1: 0.1542, loss_cls_2: 1.0197, loss_box_2: 1.9263, loss_cns_2: 0.6500, loss_yns_2: 0.1574, loss_cls_3: 1.0293, loss_box_3: 1.9087, loss_cns_3: 0.6506, loss_yns_3: 0.1551, loss_cls_4: 1.0335, loss_box_4: 1.9077, loss_cns_4: 0.6554, loss_yns_4: 0.1565, loss_cls_5: 1.0396, loss_box_5: 1.9043, loss_cns_5: 0.6526, loss_yns_5: 0.1554, loss_cls_dn_0: 0.2794, loss_box_dn_0: 0.8072, loss_cls_dn_1: 0.2018, loss_box_dn_1: 0.8165, loss_cls_dn_2: 0.2025, loss_box_dn_2: 0.8051, loss_cls_dn_3: 0.2085, loss_box_dn_3: 0.8095, loss_cls_dn_4: 0.2148, loss_box_dn_4: 0.8212, loss_cls_dn_5: 0.2251, loss_box_dn_5: 0.8313, loss_dense_depth: 0.8148, loss: 29.2548, grad_norm: 42.2335 -2025-11-12 20:01:35,622 - mmdet - INFO - Iter [94/17500] lr: 1.372e-04, eta: 13:31:08, time: 1.569, data_time: 0.078, memory: 49167, loss_cls_0: 0.9230, loss_box_0: 1.7813, loss_cns_0: 0.6208, loss_yns_0: 0.1519, loss_cls_1: 1.0076, loss_box_1: 1.9923, loss_cns_1: 0.6390, loss_yns_1: 0.1555, loss_cls_2: 1.0428, loss_box_2: 1.9463, loss_cns_2: 0.6541, loss_yns_2: 0.1565, loss_cls_3: 1.0374, loss_box_3: 1.9195, loss_cns_3: 0.6599, loss_yns_3: 0.1559, loss_cls_4: 1.0703, loss_box_4: 1.8935, loss_cns_4: 0.6593, loss_yns_4: 0.1637, loss_cls_5: 1.0388, loss_box_5: 1.8825, loss_cns_5: 0.6531, loss_yns_5: 0.1539, loss_cls_dn_0: 0.2884, loss_box_dn_0: 0.8042, loss_cls_dn_1: 0.2055, loss_box_dn_1: 0.8061, loss_cls_dn_2: 0.2097, loss_box_dn_2: 0.7980, loss_cls_dn_3: 0.2107, loss_box_dn_3: 0.8052, loss_cls_dn_4: 0.2237, loss_box_dn_4: 0.8031, loss_cls_dn_5: 0.2260, loss_box_dn_5: 0.8149, loss_dense_depth: 0.8311, loss: 29.3853, grad_norm: 52.8006 -2025-11-12 20:01:37,180 - mmdet - INFO - Iter [95/17500] lr: 1.376e-04, eta: 13:27:18, time: 1.555, data_time: 0.073, memory: 49167, loss_cls_0: 0.9131, loss_box_0: 1.8008, loss_cns_0: 0.6182, loss_yns_0: 0.1521, loss_cls_1: 1.0013, loss_box_1: 1.9255, loss_cns_1: 0.6454, loss_yns_1: 0.1551, loss_cls_2: 1.0309, loss_box_2: 1.8997, loss_cns_2: 0.6579, loss_yns_2: 0.1548, loss_cls_3: 1.0453, loss_box_3: 1.8945, loss_cns_3: 0.6623, loss_yns_3: 0.1556, loss_cls_4: 1.0669, loss_box_4: 1.8903, loss_cns_4: 0.6598, loss_yns_4: 0.1630, loss_cls_5: 1.0503, loss_box_5: 1.8521, loss_cns_5: 0.6540, loss_yns_5: 0.1555, loss_cls_dn_0: 0.2857, loss_box_dn_0: 0.8084, loss_cls_dn_1: 0.1943, loss_box_dn_1: 0.7924, loss_cls_dn_2: 0.2026, loss_box_dn_2: 0.7874, loss_cls_dn_3: 0.2063, loss_box_dn_3: 0.8010, loss_cls_dn_4: 0.2163, loss_box_dn_4: 0.8041, loss_cls_dn_5: 0.2174, loss_box_dn_5: 0.8042, loss_dense_depth: 0.8510, loss: 29.1760, grad_norm: 55.7436 -2025-11-12 20:01:38,788 - mmdet - INFO - Iter [96/17500] lr: 1.380e-04, eta: 13:23:42, time: 1.610, data_time: 0.074, memory: 49167, loss_cls_0: 0.9435, loss_box_0: 1.8359, loss_cns_0: 0.6180, loss_yns_0: 0.1563, loss_cls_1: 1.0092, loss_box_1: 1.8896, loss_cns_1: 0.6405, loss_yns_1: 0.1572, loss_cls_2: 1.0359, loss_box_2: 1.8562, loss_cns_2: 0.6499, loss_yns_2: 0.1576, loss_cls_3: 1.0861, loss_box_3: 1.8895, loss_cns_3: 0.6536, loss_yns_3: 0.1557, loss_cls_4: 1.0620, loss_box_4: 1.8609, loss_cns_4: 0.6554, loss_yns_4: 0.1578, loss_cls_5: 1.0899, loss_box_5: 1.8543, loss_cns_5: 0.6530, loss_yns_5: 0.1593, loss_cls_dn_0: 0.2870, loss_box_dn_0: 0.8196, loss_cls_dn_1: 0.1920, loss_box_dn_1: 0.7915, loss_cls_dn_2: 0.2022, loss_box_dn_2: 0.7801, loss_cls_dn_3: 0.2106, loss_box_dn_3: 0.8006, loss_cls_dn_4: 0.2125, loss_box_dn_4: 0.7961, loss_cls_dn_5: 0.2216, loss_box_dn_5: 0.8080, loss_dense_depth: 0.8435, loss: 29.1924, grad_norm: 48.8072 -2025-11-12 20:01:40,366 - mmdet - INFO - Iter [97/17500] lr: 1.384e-04, eta: 13:20:06, time: 1.577, data_time: 0.077, memory: 49167, loss_cls_0: 0.9431, loss_box_0: 1.7997, loss_cns_0: 0.6222, loss_yns_0: 0.1538, loss_cls_1: 1.0257, loss_box_1: 1.8327, loss_cns_1: 0.6442, loss_yns_1: 0.1569, loss_cls_2: 1.0340, loss_box_2: 1.7986, loss_cns_2: 0.6508, loss_yns_2: 0.1587, loss_cls_3: 1.0621, loss_box_3: 1.8189, loss_cns_3: 0.6559, loss_yns_3: 0.1568, loss_cls_4: 1.0487, loss_box_4: 1.7832, loss_cns_4: 0.6591, loss_yns_4: 0.1567, loss_cls_5: 1.0477, loss_box_5: 1.7972, loss_cns_5: 0.6575, loss_yns_5: 0.1568, loss_cls_dn_0: 0.2890, loss_box_dn_0: 0.8179, loss_cls_dn_1: 0.1905, loss_box_dn_1: 0.7855, loss_cls_dn_2: 0.1972, loss_box_dn_2: 0.7724, loss_cls_dn_3: 0.2065, loss_box_dn_3: 0.7871, loss_cls_dn_4: 0.2066, loss_box_dn_4: 0.7781, loss_cls_dn_5: 0.2147, loss_box_dn_5: 0.7958, loss_dense_depth: 0.8457, loss: 28.7082, grad_norm: 43.6435 -2025-11-12 20:01:41,927 - mmdet - INFO - Iter [98/17500] lr: 1.388e-04, eta: 13:16:29, time: 1.556, data_time: 0.079, memory: 49167, loss_cls_0: 0.9183, loss_box_0: 1.7548, loss_cns_0: 0.6224, loss_yns_0: 0.1532, loss_cls_1: 0.9910, loss_box_1: 1.8383, loss_cns_1: 0.6476, loss_yns_1: 0.1563, loss_cls_2: 1.0127, loss_box_2: 1.7944, loss_cns_2: 0.6563, loss_yns_2: 0.1553, loss_cls_3: 1.0432, loss_box_3: 1.8005, loss_cns_3: 0.6607, loss_yns_3: 0.1591, loss_cls_4: 1.0328, loss_box_4: 1.8213, loss_cns_4: 0.6610, loss_yns_4: 0.1615, loss_cls_5: 1.0884, loss_box_5: 1.7952, loss_cns_5: 0.6616, loss_yns_5: 0.1586, loss_cls_dn_0: 0.2863, loss_box_dn_0: 0.8110, loss_cls_dn_1: 0.1946, loss_box_dn_1: 0.7725, loss_cls_dn_2: 0.2002, loss_box_dn_2: 0.7616, loss_cls_dn_3: 0.2106, loss_box_dn_3: 0.7734, loss_cls_dn_4: 0.2135, loss_box_dn_4: 0.7889, loss_cls_dn_5: 0.2301, loss_box_dn_5: 0.7893, loss_dense_depth: 0.8035, loss: 28.5800, grad_norm: 43.3053 -2025-11-12 20:01:43,499 - mmdet - INFO - Iter [99/17500] lr: 1.392e-04, eta: 13:13:00, time: 1.571, data_time: 0.083, memory: 49167, loss_cls_0: 0.9179, loss_box_0: 1.7614, loss_cns_0: 0.6196, loss_yns_0: 0.1536, loss_cls_1: 0.9995, loss_box_1: 1.8742, loss_cns_1: 0.6401, loss_yns_1: 0.1546, loss_cls_2: 1.0194, loss_box_2: 1.8396, loss_cns_2: 0.6540, loss_yns_2: 0.1584, loss_cls_3: 1.0593, loss_box_3: 1.8199, loss_cns_3: 0.6535, loss_yns_3: 0.1606, loss_cls_4: 1.0453, loss_box_4: 1.8598, loss_cns_4: 0.6529, loss_yns_4: 0.1656, loss_cls_5: 1.0740, loss_box_5: 1.8319, loss_cns_5: 0.6558, loss_yns_5: 0.1608, loss_cls_dn_0: 0.2761, loss_box_dn_0: 0.8061, loss_cls_dn_1: 0.1961, loss_box_dn_1: 0.7881, loss_cls_dn_2: 0.1998, loss_box_dn_2: 0.7877, loss_cls_dn_3: 0.2160, loss_box_dn_3: 0.7933, loss_cls_dn_4: 0.2173, loss_box_dn_4: 0.8216, loss_cls_dn_5: 0.2280, loss_box_dn_5: 0.8217, loss_dense_depth: 0.8282, loss: 28.9117, grad_norm: 54.4568 -2025-11-12 20:01:45,064 - mmdet - INFO - Iter [100/17500] lr: 1.396e-04, eta: 13:09:34, time: 1.567, data_time: 0.080, memory: 49167, loss_cls_0: 0.9320, loss_box_0: 1.8012, loss_cns_0: 0.6213, loss_yns_0: 0.1555, loss_cls_1: 0.9905, loss_box_1: 1.9087, loss_cns_1: 0.6356, loss_yns_1: 0.1562, loss_cls_2: 1.0221, loss_box_2: 1.8817, loss_cns_2: 0.6447, loss_yns_2: 0.1625, loss_cls_3: 1.0485, loss_box_3: 1.8579, loss_cns_3: 0.6509, loss_yns_3: 0.1592, loss_cls_4: 1.0378, loss_box_4: 1.8675, loss_cns_4: 0.6507, loss_yns_4: 0.1597, loss_cls_5: 1.0461, loss_box_5: 1.8951, loss_cns_5: 0.6558, loss_yns_5: 0.1593, loss_cls_dn_0: 0.2648, loss_box_dn_0: 0.8147, loss_cls_dn_1: 0.1930, loss_box_dn_1: 0.8075, loss_cls_dn_2: 0.1983, loss_box_dn_2: 0.7973, loss_cls_dn_3: 0.2073, loss_box_dn_3: 0.8033, loss_cls_dn_4: 0.2115, loss_box_dn_4: 0.8284, loss_cls_dn_5: 0.2173, loss_box_dn_5: 0.8540, loss_dense_depth: 0.7995, loss: 29.0976, grad_norm: 45.5039 -2025-11-12 20:01:46,709 - mmdet - INFO - Iter [101/17500] lr: 1.400e-04, eta: 13:06:26, time: 1.645, data_time: 0.097, memory: 49167, loss_cls_0: 0.9188, loss_box_0: 1.7858, loss_cns_0: 0.6234, loss_yns_0: 0.1588, loss_cls_1: 0.9991, loss_box_1: 1.9519, loss_cns_1: 0.6417, loss_yns_1: 0.1583, loss_cls_2: 1.0129, loss_box_2: 1.9194, loss_cns_2: 0.6473, loss_yns_2: 0.1640, loss_cls_3: 1.0469, loss_box_3: 1.8986, loss_cns_3: 0.6534, loss_yns_3: 0.1637, loss_cls_4: 1.0238, loss_box_4: 1.9064, loss_cns_4: 0.6527, loss_yns_4: 0.1613, loss_cls_5: 1.0389, loss_box_5: 1.9078, loss_cns_5: 0.6559, loss_yns_5: 0.1591, loss_cls_dn_0: 0.2711, loss_box_dn_0: 0.8055, loss_cls_dn_1: 0.1930, loss_box_dn_1: 0.8294, loss_cls_dn_2: 0.1984, loss_box_dn_2: 0.8157, loss_cls_dn_3: 0.2030, loss_box_dn_3: 0.8180, loss_cls_dn_4: 0.2057, loss_box_dn_4: 0.8315, loss_cls_dn_5: 0.2131, loss_box_dn_5: 0.8499, loss_dense_depth: 0.8185, loss: 29.3029, grad_norm: 44.8799 -2025-11-12 20:01:48,373 - mmdet - INFO - Iter [102/17500] lr: 1.404e-04, eta: 13:03:25, time: 1.668, data_time: 0.084, memory: 49167, loss_cls_0: 0.9143, loss_box_0: 1.8157, loss_cns_0: 0.6173, loss_yns_0: 0.1562, loss_cls_1: 0.9970, loss_box_1: 1.9442, loss_cns_1: 0.6370, loss_yns_1: 0.1592, loss_cls_2: 1.0318, loss_box_2: 1.9093, loss_cns_2: 0.6436, loss_yns_2: 0.1579, loss_cls_3: 1.0386, loss_box_3: 1.9029, loss_cns_3: 0.6483, loss_yns_3: 0.1606, loss_cls_4: 1.0249, loss_box_4: 1.8865, loss_cns_4: 0.6524, loss_yns_4: 0.1598, loss_cls_5: 1.0271, loss_box_5: 1.8685, loss_cns_5: 0.6490, loss_yns_5: 0.1578, loss_cls_dn_0: 0.2666, loss_box_dn_0: 0.8134, loss_cls_dn_1: 0.1913, loss_box_dn_1: 0.8254, loss_cls_dn_2: 0.2037, loss_box_dn_2: 0.8197, loss_cls_dn_3: 0.2103, loss_box_dn_3: 0.8224, loss_cls_dn_4: 0.2075, loss_box_dn_4: 0.8269, loss_cls_dn_5: 0.2120, loss_box_dn_5: 0.8387, loss_dense_depth: 0.8013, loss: 29.1990, grad_norm: 60.0014 -2025-11-12 20:01:49,969 - mmdet - INFO - Iter [103/17500] lr: 1.408e-04, eta: 13:00:15, time: 1.595, data_time: 0.129, memory: 49167, loss_cls_0: 0.9237, loss_box_0: 1.7796, loss_cns_0: 0.6235, loss_yns_0: 0.1583, loss_cls_1: 0.9604, loss_box_1: 1.9234, loss_cns_1: 0.6359, loss_yns_1: 0.1614, loss_cls_2: 0.9965, loss_box_2: 1.8690, loss_cns_2: 0.6495, loss_yns_2: 0.1622, loss_cls_3: 1.0123, loss_box_3: 1.8455, loss_cns_3: 0.6564, loss_yns_3: 0.1609, loss_cls_4: 1.0037, loss_box_4: 1.8426, loss_cns_4: 0.6566, loss_yns_4: 0.1596, loss_cls_5: 1.0173, loss_box_5: 1.8313, loss_cns_5: 0.6578, loss_yns_5: 0.1582, loss_cls_dn_0: 0.2529, loss_box_dn_0: 0.8019, loss_cls_dn_1: 0.1813, loss_box_dn_1: 0.8249, loss_cls_dn_2: 0.1938, loss_box_dn_2: 0.8025, loss_cls_dn_3: 0.2017, loss_box_dn_3: 0.8051, loss_cls_dn_4: 0.1989, loss_box_dn_4: 0.8167, loss_cls_dn_5: 0.2060, loss_box_dn_5: 0.8272, loss_dense_depth: 0.8182, loss: 28.7765, grad_norm: 44.8026 -2025-11-12 20:01:51,527 - mmdet - INFO - Iter [104/17500] lr: 1.412e-04, eta: 12:57:03, time: 1.558, data_time: 0.075, memory: 49167, loss_cls_0: 0.9012, loss_box_0: 1.8031, loss_cns_0: 0.6144, loss_yns_0: 0.1536, loss_cls_1: 0.9631, loss_box_1: 1.9124, loss_cns_1: 0.6384, loss_yns_1: 0.1584, loss_cls_2: 0.9926, loss_box_2: 1.8588, loss_cns_2: 0.6479, loss_yns_2: 0.1634, loss_cls_3: 1.0617, loss_box_3: 1.8573, loss_cns_3: 0.6595, loss_yns_3: 0.1594, loss_cls_4: 1.0185, loss_box_4: 1.8937, loss_cns_4: 0.6546, loss_yns_4: 0.1582, loss_cls_5: 1.0201, loss_box_5: 1.8536, loss_cns_5: 0.6722, loss_yns_5: 0.1568, loss_cls_dn_0: 0.2590, loss_box_dn_0: 0.8244, loss_cls_dn_1: 0.1758, loss_box_dn_1: 0.8153, loss_cls_dn_2: 0.1820, loss_box_dn_2: 0.7962, loss_cls_dn_3: 0.1892, loss_box_dn_3: 0.8133, loss_cls_dn_4: 0.1884, loss_box_dn_4: 0.8412, loss_cls_dn_5: 0.2032, loss_box_dn_5: 0.8418, loss_dense_depth: 0.7974, loss: 28.9001, grad_norm: 49.4417 -2025-11-12 20:01:53,095 - mmdet - INFO - Iter [105/17500] lr: 1.416e-04, eta: 12:53:56, time: 1.564, data_time: 0.106, memory: 49167, loss_cls_0: 0.8935, loss_box_0: 1.7272, loss_cns_0: 0.6164, loss_yns_0: 0.1522, loss_cls_1: 0.9572, loss_box_1: 1.8774, loss_cns_1: 0.6471, loss_yns_1: 0.1572, loss_cls_2: 0.9878, loss_box_2: 1.8252, loss_cns_2: 0.6557, loss_yns_2: 0.1616, loss_cls_3: 1.0300, loss_box_3: 1.8227, loss_cns_3: 0.6650, loss_yns_3: 0.1618, loss_cls_4: 1.0142, loss_box_4: 1.8277, loss_cns_4: 0.6632, loss_yns_4: 0.1580, loss_cls_5: 1.0148, loss_box_5: 1.8357, loss_cns_5: 0.6716, loss_yns_5: 0.1570, loss_cls_dn_0: 0.2598, loss_box_dn_0: 0.8108, loss_cls_dn_1: 0.1741, loss_box_dn_1: 0.8298, loss_cls_dn_2: 0.1819, loss_box_dn_2: 0.8140, loss_cls_dn_3: 0.1903, loss_box_dn_3: 0.8331, loss_cls_dn_4: 0.1923, loss_box_dn_4: 0.8526, loss_cls_dn_5: 0.2085, loss_box_dn_5: 0.8680, loss_dense_depth: 0.7779, loss: 28.6733, grad_norm: 58.7201 -2025-11-12 20:01:54,655 - mmdet - INFO - Iter [106/17500] lr: 1.420e-04, eta: 12:50:51, time: 1.560, data_time: 0.081, memory: 49167, loss_cls_0: 0.9130, loss_box_0: 1.7141, loss_cns_0: 0.6069, loss_yns_0: 0.1519, loss_cls_1: 0.9503, loss_box_1: 1.9239, loss_cns_1: 0.6420, loss_yns_1: 0.1572, loss_cls_2: 0.9897, loss_box_2: 1.8785, loss_cns_2: 0.6511, loss_yns_2: 0.1571, loss_cls_3: 1.0066, loss_box_3: 1.8780, loss_cns_3: 0.6525, loss_yns_3: 0.1596, loss_cls_4: 1.0046, loss_box_4: 1.8543, loss_cns_4: 0.6555, loss_yns_4: 0.1565, loss_cls_5: 1.0153, loss_box_5: 1.9188, loss_cns_5: 0.6442, loss_yns_5: 0.1581, loss_cls_dn_0: 0.2582, loss_box_dn_0: 0.7979, loss_cls_dn_1: 0.1745, loss_box_dn_1: 0.8188, loss_cls_dn_2: 0.1822, loss_box_dn_2: 0.8182, loss_cls_dn_3: 0.1971, loss_box_dn_3: 0.8355, loss_cls_dn_4: 0.1921, loss_box_dn_4: 0.8458, loss_cls_dn_5: 0.2035, loss_box_dn_5: 0.8818, loss_dense_depth: 0.8067, loss: 28.8520, grad_norm: 58.1730 -2025-11-12 20:01:56,223 - mmdet - INFO - Iter [107/17500] lr: 1.424e-04, eta: 12:47:51, time: 1.569, data_time: 0.079, memory: 49167, loss_cls_0: 0.8885, loss_box_0: 1.7417, loss_cns_0: 0.6188, loss_yns_0: 0.1564, loss_cls_1: 0.9410, loss_box_1: 1.9016, loss_cns_1: 0.6363, loss_yns_1: 0.1586, loss_cls_2: 0.9826, loss_box_2: 1.8361, loss_cns_2: 0.6520, loss_yns_2: 0.1569, loss_cls_3: 1.0342, loss_box_3: 1.8417, loss_cns_3: 0.6547, loss_yns_3: 0.1563, loss_cls_4: 0.9969, loss_box_4: 1.8485, loss_cns_4: 0.6565, loss_yns_4: 0.1569, loss_cls_5: 1.0121, loss_box_5: 1.8663, loss_cns_5: 0.6536, loss_yns_5: 0.1563, loss_cls_dn_0: 0.2487, loss_box_dn_0: 0.8136, loss_cls_dn_1: 0.1780, loss_box_dn_1: 0.8186, loss_cls_dn_2: 0.1841, loss_box_dn_2: 0.8178, loss_cls_dn_3: 0.2039, loss_box_dn_3: 0.8387, loss_cls_dn_4: 0.1953, loss_box_dn_4: 0.8519, loss_cls_dn_5: 0.2086, loss_box_dn_5: 0.8734, loss_dense_depth: 0.7952, loss: 28.7324, grad_norm: 41.5342 -2025-11-12 20:01:57,781 - mmdet - INFO - Iter [108/17500] lr: 1.428e-04, eta: 12:44:52, time: 1.557, data_time: 0.093, memory: 49167, loss_cls_0: 0.9159, loss_box_0: 1.7620, loss_cns_0: 0.6248, loss_yns_0: 0.1577, loss_cls_1: 0.9420, loss_box_1: 1.8766, loss_cns_1: 0.6340, loss_yns_1: 0.1573, loss_cls_2: 0.9799, loss_box_2: 1.8457, loss_cns_2: 0.6484, loss_yns_2: 0.1607, loss_cls_3: 1.0313, loss_box_3: 1.8702, loss_cns_3: 0.6536, loss_yns_3: 0.1586, loss_cls_4: 1.0124, loss_box_4: 1.8997, loss_cns_4: 0.6526, loss_yns_4: 0.1584, loss_cls_5: 1.0420, loss_box_5: 1.8923, loss_cns_5: 0.6538, loss_yns_5: 0.1587, loss_cls_dn_0: 0.2430, loss_box_dn_0: 0.8094, loss_cls_dn_1: 0.1806, loss_box_dn_1: 0.8230, loss_cls_dn_2: 0.1844, loss_box_dn_2: 0.8295, loss_cls_dn_3: 0.1988, loss_box_dn_3: 0.8538, loss_cls_dn_4: 0.1961, loss_box_dn_4: 0.8694, loss_cls_dn_5: 0.2085, loss_box_dn_5: 0.8786, loss_dense_depth: 0.8213, loss: 28.9850, grad_norm: 73.3042 -2025-11-12 20:01:59,348 - mmdet - INFO - Iter [109/17500] lr: 1.432e-04, eta: 12:42:00, time: 1.571, data_time: 0.087, memory: 49167, loss_cls_0: 0.9161, loss_box_0: 1.7982, loss_cns_0: 0.6185, loss_yns_0: 0.1593, loss_cls_1: 0.9542, loss_box_1: 1.8900, loss_cns_1: 0.6280, loss_yns_1: 0.1570, loss_cls_2: 0.9771, loss_box_2: 1.8459, loss_cns_2: 0.6423, loss_yns_2: 0.1662, loss_cls_3: 1.0308, loss_box_3: 1.8466, loss_cns_3: 0.6508, loss_yns_3: 0.1629, loss_cls_4: 1.0214, loss_box_4: 1.8736, loss_cns_4: 0.6493, loss_yns_4: 0.1594, loss_cls_5: 1.0186, loss_box_5: 1.8603, loss_cns_5: 0.6515, loss_yns_5: 0.1599, loss_cls_dn_0: 0.2426, loss_box_dn_0: 0.8091, loss_cls_dn_1: 0.1792, loss_box_dn_1: 0.8337, loss_cls_dn_2: 0.1830, loss_box_dn_2: 0.8175, loss_cls_dn_3: 0.1936, loss_box_dn_3: 0.8240, loss_cls_dn_4: 0.1966, loss_box_dn_4: 0.8333, loss_cls_dn_5: 0.2026, loss_box_dn_5: 0.8373, loss_dense_depth: 0.8162, loss: 28.8063, grad_norm: 50.5346 -2025-11-12 20:02:00,914 - mmdet - INFO - Iter [110/17500] lr: 1.436e-04, eta: 12:39:09, time: 1.567, data_time: 0.077, memory: 49167, loss_cls_0: 0.9016, loss_box_0: 1.8032, loss_cns_0: 0.6131, loss_yns_0: 0.1559, loss_cls_1: 0.9634, loss_box_1: 1.8641, loss_cns_1: 0.6268, loss_yns_1: 0.1563, loss_cls_2: 0.9842, loss_box_2: 1.8788, loss_cns_2: 0.6350, loss_yns_2: 0.1614, loss_cls_3: 1.0216, loss_box_3: 1.8422, loss_cns_3: 0.6417, loss_yns_3: 0.1613, loss_cls_4: 1.0325, loss_box_4: 1.8613, loss_cns_4: 0.6376, loss_yns_4: 0.1572, loss_cls_5: 1.0178, loss_box_5: 1.8573, loss_cns_5: 0.6401, loss_yns_5: 0.1586, loss_cls_dn_0: 0.2519, loss_box_dn_0: 0.8041, loss_cls_dn_1: 0.1784, loss_box_dn_1: 0.7707, loss_cls_dn_2: 0.1821, loss_box_dn_2: 0.7710, loss_cls_dn_3: 0.1865, loss_box_dn_3: 0.7652, loss_cls_dn_4: 0.1942, loss_box_dn_4: 0.7799, loss_cls_dn_5: 0.1998, loss_box_dn_5: 0.7875, loss_dense_depth: 0.8336, loss: 28.4778, grad_norm: 52.2304 -2025-11-12 20:02:02,474 - mmdet - INFO - Iter [111/17500] lr: 1.440e-04, eta: 12:36:20, time: 1.555, data_time: 0.078, memory: 49167, loss_cls_0: 0.9060, loss_box_0: 1.7863, loss_cns_0: 0.6172, loss_yns_0: 0.1556, loss_cls_1: 0.9549, loss_box_1: 1.8900, loss_cns_1: 0.6310, loss_yns_1: 0.1591, loss_cls_2: 0.9860, loss_box_2: 1.8514, loss_cns_2: 0.6457, loss_yns_2: 0.1560, loss_cls_3: 1.0300, loss_box_3: 1.8420, loss_cns_3: 0.6467, loss_yns_3: 0.1578, loss_cls_4: 1.0403, loss_box_4: 1.8206, loss_cns_4: 0.6488, loss_yns_4: 0.1566, loss_cls_5: 1.0364, loss_box_5: 1.8288, loss_cns_5: 0.6465, loss_yns_5: 0.1570, loss_cls_dn_0: 0.2550, loss_box_dn_0: 0.8031, loss_cls_dn_1: 0.1775, loss_box_dn_1: 0.7662, loss_cls_dn_2: 0.1841, loss_box_dn_2: 0.7634, loss_cls_dn_3: 0.1968, loss_box_dn_3: 0.7731, loss_cls_dn_4: 0.1934, loss_box_dn_4: 0.7858, loss_cls_dn_5: 0.2046, loss_box_dn_5: 0.8001, loss_dense_depth: 0.8521, loss: 28.5057, grad_norm: 48.5197 -2025-11-12 20:02:04,049 - mmdet - INFO - Iter [112/17500] lr: 1.444e-04, eta: 12:33:37, time: 1.576, data_time: 0.086, memory: 49167, loss_cls_0: 0.9114, loss_box_0: 1.8202, loss_cns_0: 0.6136, loss_yns_0: 0.1588, loss_cls_1: 0.9675, loss_box_1: 1.9538, loss_cns_1: 0.6273, loss_yns_1: 0.1576, loss_cls_2: 0.9878, loss_box_2: 1.9045, loss_cns_2: 0.6401, loss_yns_2: 0.1577, loss_cls_3: 1.0237, loss_box_3: 1.9422, loss_cns_3: 0.6384, loss_yns_3: 0.1547, loss_cls_4: 1.0280, loss_box_4: 1.9087, loss_cns_4: 0.6444, loss_yns_4: 0.1604, loss_cls_5: 1.0383, loss_box_5: 1.9711, loss_cns_5: 0.6352, loss_yns_5: 0.1599, loss_cls_dn_0: 0.2588, loss_box_dn_0: 0.8083, loss_cls_dn_1: 0.1814, loss_box_dn_1: 0.8071, loss_cls_dn_2: 0.1882, loss_box_dn_2: 0.7956, loss_cls_dn_3: 0.2072, loss_box_dn_3: 0.8288, loss_cls_dn_4: 0.2013, loss_box_dn_4: 0.8436, loss_cls_dn_5: 0.2119, loss_box_dn_5: 0.8795, loss_dense_depth: 0.8799, loss: 29.2969, grad_norm: 71.8072 -2025-11-12 20:02:05,603 - mmdet - INFO - Iter [113/17500] lr: 1.448e-04, eta: 12:30:53, time: 1.558, data_time: 0.078, memory: 49167, loss_cls_0: 0.8829, loss_box_0: 1.7890, loss_cns_0: 0.6205, loss_yns_0: 0.1576, loss_cls_1: 0.9531, loss_box_1: 1.9001, loss_cns_1: 0.6327, loss_yns_1: 0.1532, loss_cls_2: 0.9601, loss_box_2: 1.8406, loss_cns_2: 0.6407, loss_yns_2: 0.1620, loss_cls_3: 0.9849, loss_box_3: 1.8558, loss_cns_3: 0.6423, loss_yns_3: 0.1550, loss_cls_4: 0.9763, loss_box_4: 1.8254, loss_cns_4: 0.6481, loss_yns_4: 0.1561, loss_cls_5: 0.9915, loss_box_5: 1.8676, loss_cns_5: 0.6404, loss_yns_5: 0.1554, loss_cls_dn_0: 0.2415, loss_box_dn_0: 0.8055, loss_cls_dn_1: 0.1816, loss_box_dn_1: 0.8529, loss_cls_dn_2: 0.1861, loss_box_dn_2: 0.8377, loss_cls_dn_3: 0.2023, loss_box_dn_3: 0.8654, loss_cls_dn_4: 0.1962, loss_box_dn_4: 0.8828, loss_cls_dn_5: 0.2063, loss_box_dn_5: 0.9124, loss_dense_depth: 0.8468, loss: 28.8091, grad_norm: 53.0554 -2025-11-12 20:02:07,165 - mmdet - INFO - Iter [114/17500] lr: 1.452e-04, eta: 12:28:14, time: 1.565, data_time: 0.073, memory: 49167, loss_cls_0: 0.8922, loss_box_0: 1.8290, loss_cns_0: 0.6164, loss_yns_0: 0.1603, loss_cls_1: 0.9640, loss_box_1: 1.8890, loss_cns_1: 0.6437, loss_yns_1: 0.1542, loss_cls_2: 0.9893, loss_box_2: 1.8414, loss_cns_2: 0.6464, loss_yns_2: 0.1637, loss_cls_3: 0.9940, loss_box_3: 1.8713, loss_cns_3: 0.6468, loss_yns_3: 0.1589, loss_cls_4: 0.9977, loss_box_4: 1.8865, loss_cns_4: 0.6404, loss_yns_4: 0.1587, loss_cls_5: 0.9936, loss_box_5: 1.8429, loss_cns_5: 0.6488, loss_yns_5: 0.1578, loss_cls_dn_0: 0.2488, loss_box_dn_0: 0.8093, loss_cls_dn_1: 0.1729, loss_box_dn_1: 0.8269, loss_cls_dn_2: 0.1827, loss_box_dn_2: 0.8247, loss_cls_dn_3: 0.1886, loss_box_dn_3: 0.8462, loss_cls_dn_4: 0.1920, loss_box_dn_4: 0.8764, loss_cls_dn_5: 0.1988, loss_box_dn_5: 0.8789, loss_dense_depth: 0.8512, loss: 28.8844, grad_norm: 58.7725 -2025-11-12 20:02:08,734 - mmdet - INFO - Iter [115/17500] lr: 1.456e-04, eta: 12:25:39, time: 1.571, data_time: 0.068, memory: 49167, loss_cls_0: 0.8538, loss_box_0: 1.7831, loss_cns_0: 0.6158, loss_yns_0: 0.1563, loss_cls_1: 0.9550, loss_box_1: 1.9113, loss_cns_1: 0.6439, loss_yns_1: 0.1563, loss_cls_2: 0.9750, loss_box_2: 1.8668, loss_cns_2: 0.6459, loss_yns_2: 0.1576, loss_cls_3: 1.0037, loss_box_3: 1.8836, loss_cns_3: 0.6479, loss_yns_3: 0.1561, loss_cls_4: 0.9739, loss_box_4: 1.9038, loss_cns_4: 0.6413, loss_yns_4: 0.1596, loss_cls_5: 1.0006, loss_box_5: 1.8489, loss_cns_5: 0.6511, loss_yns_5: 0.1582, loss_cls_dn_0: 0.2399, loss_box_dn_0: 0.8117, loss_cls_dn_1: 0.1696, loss_box_dn_1: 0.8429, loss_cls_dn_2: 0.1806, loss_box_dn_2: 0.8408, loss_cls_dn_3: 0.1868, loss_box_dn_3: 0.8559, loss_cls_dn_4: 0.1913, loss_box_dn_4: 0.8736, loss_cls_dn_5: 0.1978, loss_box_dn_5: 0.8672, loss_dense_depth: 0.8531, loss: 28.8607, grad_norm: 68.6848 -2025-11-12 20:02:10,301 - mmdet - INFO - Iter [116/17500] lr: 1.460e-04, eta: 12:23:06, time: 1.568, data_time: 0.073, memory: 49167, loss_cls_0: 0.8758, loss_box_0: 1.7507, loss_cns_0: 0.6136, loss_yns_0: 0.1544, loss_cls_1: 0.9625, loss_box_1: 1.8896, loss_cns_1: 0.6422, loss_yns_1: 0.1587, loss_cls_2: 0.9637, loss_box_2: 1.8258, loss_cns_2: 0.6477, loss_yns_2: 0.1548, loss_cls_3: 0.9990, loss_box_3: 1.8101, loss_cns_3: 0.6540, loss_yns_3: 0.1554, loss_cls_4: 0.9778, loss_box_4: 1.8159, loss_cns_4: 0.6537, loss_yns_4: 0.1606, loss_cls_5: 0.9996, loss_box_5: 1.8203, loss_cns_5: 0.6525, loss_yns_5: 0.1589, loss_cls_dn_0: 0.2436, loss_box_dn_0: 0.7992, loss_cls_dn_1: 0.1727, loss_box_dn_1: 0.7992, loss_cls_dn_2: 0.1796, loss_box_dn_2: 0.7747, loss_cls_dn_3: 0.1860, loss_box_dn_3: 0.7705, loss_cls_dn_4: 0.1922, loss_box_dn_4: 0.7756, loss_cls_dn_5: 0.1950, loss_box_dn_5: 0.7816, loss_dense_depth: 0.8460, loss: 28.2131, grad_norm: 48.5982 -2025-11-12 20:02:11,874 - mmdet - INFO - Iter [117/17500] lr: 1.464e-04, eta: 12:20:34, time: 1.564, data_time: 0.073, memory: 49167, loss_cls_0: 0.8779, loss_box_0: 1.7495, loss_cns_0: 0.6181, loss_yns_0: 0.1539, loss_cls_1: 0.9407, loss_box_1: 1.8425, loss_cns_1: 0.6451, loss_yns_1: 0.1540, loss_cls_2: 0.9889, loss_box_2: 1.8347, loss_cns_2: 0.6435, loss_yns_2: 0.1548, loss_cls_3: 0.9867, loss_box_3: 1.8262, loss_cns_3: 0.6467, loss_yns_3: 0.1529, loss_cls_4: 1.0165, loss_box_4: 1.8278, loss_cns_4: 0.6519, loss_yns_4: 0.1554, loss_cls_5: 0.9952, loss_box_5: 1.8516, loss_cns_5: 0.6464, loss_yns_5: 0.1573, loss_cls_dn_0: 0.2390, loss_box_dn_0: 0.8059, loss_cls_dn_1: 0.1703, loss_box_dn_1: 0.7716, loss_cls_dn_2: 0.1736, loss_box_dn_2: 0.7577, loss_cls_dn_3: 0.1815, loss_box_dn_3: 0.7595, loss_cls_dn_4: 0.1893, loss_box_dn_4: 0.7686, loss_cls_dn_5: 0.1895, loss_box_dn_5: 0.7819, loss_dense_depth: 0.8609, loss: 28.1673, grad_norm: 64.0381 -2025-11-12 20:02:13,439 - mmdet - INFO - Iter [118/17500] lr: 1.468e-04, eta: 12:18:07, time: 1.570, data_time: 0.080, memory: 49167, loss_cls_0: 0.8949, loss_box_0: 1.7658, loss_cns_0: 0.6187, loss_yns_0: 0.1572, loss_cls_1: 0.9636, loss_box_1: 1.8336, loss_cns_1: 0.6481, loss_yns_1: 0.1560, loss_cls_2: 1.0040, loss_box_2: 1.8523, loss_cns_2: 0.6471, loss_yns_2: 0.1595, loss_cls_3: 1.0052, loss_box_3: 1.8482, loss_cns_3: 0.6513, loss_yns_3: 0.1567, loss_cls_4: 1.0026, loss_box_4: 1.8393, loss_cns_4: 0.6553, loss_yns_4: 0.1565, loss_cls_5: 1.0054, loss_box_5: 1.8504, loss_cns_5: 0.6534, loss_yns_5: 0.1581, loss_cls_dn_0: 0.2421, loss_box_dn_0: 0.8047, loss_cls_dn_1: 0.1720, loss_box_dn_1: 0.7585, loss_cls_dn_2: 0.1761, loss_box_dn_2: 0.7566, loss_cls_dn_3: 0.1826, loss_box_dn_3: 0.7645, loss_cls_dn_4: 0.1871, loss_box_dn_4: 0.7713, loss_cls_dn_5: 0.1907, loss_box_dn_5: 0.7878, loss_dense_depth: 0.8861, loss: 28.3631, grad_norm: 57.4909 -2025-11-12 20:02:15,038 - mmdet - INFO - Iter [119/17500] lr: 1.472e-04, eta: 12:15:45, time: 1.597, data_time: 0.074, memory: 49167, loss_cls_0: 0.8837, loss_box_0: 1.7262, loss_cns_0: 0.6120, loss_yns_0: 0.1532, loss_cls_1: 0.9744, loss_box_1: 1.8954, loss_cns_1: 0.6467, loss_yns_1: 0.1563, loss_cls_2: 0.9913, loss_box_2: 1.8651, loss_cns_2: 0.6503, loss_yns_2: 0.1578, loss_cls_3: 1.0164, loss_box_3: 1.8705, loss_cns_3: 0.6517, loss_yns_3: 0.1556, loss_cls_4: 1.0038, loss_box_4: 1.9008, loss_cns_4: 0.6485, loss_yns_4: 0.1555, loss_cls_5: 0.9938, loss_box_5: 1.9243, loss_cns_5: 0.6444, loss_yns_5: 0.1540, loss_cls_dn_0: 0.2377, loss_box_dn_0: 0.7899, loss_cls_dn_1: 0.1736, loss_box_dn_1: 0.7566, loss_cls_dn_2: 0.1775, loss_box_dn_2: 0.7483, loss_cls_dn_3: 0.1834, loss_box_dn_3: 0.7576, loss_cls_dn_4: 0.1833, loss_box_dn_4: 0.7811, loss_cls_dn_5: 0.1898, loss_box_dn_5: 0.8061, loss_dense_depth: 0.8241, loss: 28.4408, grad_norm: 61.8301 -2025-11-12 20:02:16,591 - mmdet - INFO - Iter [120/17500] lr: 1.476e-04, eta: 12:13:19, time: 1.551, data_time: 0.077, memory: 49167, loss_cls_0: 0.8733, loss_box_0: 1.7294, loss_cns_0: 0.6074, loss_yns_0: 0.1508, loss_cls_1: 0.9838, loss_box_1: 1.8892, loss_cns_1: 0.6405, loss_yns_1: 0.1572, loss_cls_2: 1.0080, loss_box_2: 1.8298, loss_cns_2: 0.6497, loss_yns_2: 0.1574, loss_cls_3: 1.0199, loss_box_3: 1.8303, loss_cns_3: 0.6502, loss_yns_3: 0.1563, loss_cls_4: 1.0322, loss_box_4: 1.8555, loss_cns_4: 0.6484, loss_yns_4: 0.1601, loss_cls_5: 0.9989, loss_box_5: 1.9046, loss_cns_5: 0.6422, loss_yns_5: 0.1569, loss_cls_dn_0: 0.2383, loss_box_dn_0: 0.7965, loss_cls_dn_1: 0.1786, loss_box_dn_1: 0.7896, loss_cls_dn_2: 0.1873, loss_box_dn_2: 0.7825, loss_cls_dn_3: 0.1950, loss_box_dn_3: 0.7949, loss_cls_dn_4: 0.2102, loss_box_dn_4: 0.8227, loss_cls_dn_5: 0.2138, loss_box_dn_5: 0.8615, loss_dense_depth: 0.8749, loss: 28.6780, grad_norm: 64.2354 -2025-11-12 20:02:18,212 - mmdet - INFO - Iter [121/17500] lr: 1.480e-04, eta: 12:11:06, time: 1.620, data_time: 0.093, memory: 49167, loss_cls_0: 0.8849, loss_box_0: 1.7674, loss_cns_0: 0.6138, loss_yns_0: 0.1524, loss_cls_1: 0.9896, loss_box_1: 1.8939, loss_cns_1: 0.6426, loss_yns_1: 0.1551, loss_cls_2: 1.0351, loss_box_2: 1.8529, loss_cns_2: 0.6509, loss_yns_2: 0.1554, loss_cls_3: 1.0217, loss_box_3: 1.8619, loss_cns_3: 0.6543, loss_yns_3: 0.1556, loss_cls_4: 1.0480, loss_box_4: 1.8606, loss_cns_4: 0.6534, loss_yns_4: 0.1610, loss_cls_5: 1.0176, loss_box_5: 1.8775, loss_cns_5: 0.6521, loss_yns_5: 0.1600, loss_cls_dn_0: 0.2393, loss_box_dn_0: 0.8030, loss_cls_dn_1: 0.1761, loss_box_dn_1: 0.8004, loss_cls_dn_2: 0.1955, loss_box_dn_2: 0.8005, loss_cls_dn_3: 0.2014, loss_box_dn_3: 0.8148, loss_cls_dn_4: 0.2318, loss_box_dn_4: 0.8223, loss_cls_dn_5: 0.2324, loss_box_dn_5: 0.8547, loss_dense_depth: 0.8554, loss: 28.9451, grad_norm: 63.3764 -2025-11-12 20:02:19,855 - mmdet - INFO - Iter [122/17500] lr: 1.484e-04, eta: 12:08:58, time: 1.643, data_time: 0.078, memory: 49167, loss_cls_0: 0.9164, loss_box_0: 1.7685, loss_cns_0: 0.6224, loss_yns_0: 0.1568, loss_cls_1: 0.9531, loss_box_1: 1.8790, loss_cns_1: 0.6403, loss_yns_1: 0.1588, loss_cls_2: 0.9947, loss_box_2: 1.8477, loss_cns_2: 0.6471, loss_yns_2: 0.1596, loss_cls_3: 1.0015, loss_box_3: 1.8583, loss_cns_3: 0.6566, loss_yns_3: 0.1575, loss_cls_4: 1.0027, loss_box_4: 1.8494, loss_cns_4: 0.6504, loss_yns_4: 0.1605, loss_cls_5: 0.9957, loss_box_5: 1.8744, loss_cns_5: 0.6502, loss_yns_5: 0.1618, loss_cls_dn_0: 0.2383, loss_box_dn_0: 0.7949, loss_cls_dn_1: 0.1718, loss_box_dn_1: 0.8140, loss_cls_dn_2: 0.1896, loss_box_dn_2: 0.8145, loss_cls_dn_3: 0.1926, loss_box_dn_3: 0.8296, loss_cls_dn_4: 0.2094, loss_box_dn_4: 0.8304, loss_cls_dn_5: 0.2087, loss_box_dn_5: 0.8575, loss_dense_depth: 0.7984, loss: 28.7131, grad_norm: 66.2274 -2025-11-12 20:02:21,469 - mmdet - INFO - Iter [123/17500] lr: 1.488e-04, eta: 12:06:48, time: 1.617, data_time: 0.129, memory: 49167, loss_cls_0: 0.9410, loss_box_0: 1.7570, loss_cns_0: 0.6239, loss_yns_0: 0.1567, loss_cls_1: 1.0384, loss_box_1: 1.8542, loss_cns_1: 0.6437, loss_yns_1: 0.1588, loss_cls_2: 1.0034, loss_box_2: 1.7993, loss_cns_2: 0.6518, loss_yns_2: 0.1604, loss_cls_3: 1.0567, loss_box_3: 1.8031, loss_cns_3: 0.6572, loss_yns_3: 0.1576, loss_cls_4: 1.0206, loss_box_4: 1.7967, loss_cns_4: 0.6537, loss_yns_4: 0.1582, loss_cls_5: 1.0356, loss_box_5: 1.8022, loss_cns_5: 0.6542, loss_yns_5: 0.1608, loss_cls_dn_0: 0.2341, loss_box_dn_0: 0.7976, loss_cls_dn_1: 0.1754, loss_box_dn_1: 0.8017, loss_cls_dn_2: 0.1810, loss_box_dn_2: 0.7905, loss_cls_dn_3: 0.1831, loss_box_dn_3: 0.8010, loss_cls_dn_4: 0.1884, loss_box_dn_4: 0.8073, loss_cls_dn_5: 0.1930, loss_box_dn_5: 0.8202, loss_dense_depth: 0.8670, loss: 28.5853, grad_norm: 49.3492 -2025-11-12 20:02:23,029 - mmdet - INFO - Iter [124/17500] lr: 1.492e-04, eta: 12:04:32, time: 1.556, data_time: 0.076, memory: 49167, loss_cls_0: 0.8991, loss_box_0: 1.7454, loss_cns_0: 0.6187, loss_yns_0: 0.1536, loss_cls_1: 0.9557, loss_box_1: 1.7942, loss_cns_1: 0.6525, loss_yns_1: 0.1599, loss_cls_2: 0.9977, loss_box_2: 1.7570, loss_cns_2: 0.6574, loss_yns_2: 0.1570, loss_cls_3: 1.0251, loss_box_3: 1.7697, loss_cns_3: 0.6544, loss_yns_3: 0.1582, loss_cls_4: 1.0311, loss_box_4: 1.7923, loss_cns_4: 0.6557, loss_yns_4: 0.1603, loss_cls_5: 1.0096, loss_box_5: 1.7915, loss_cns_5: 0.6586, loss_yns_5: 0.1621, loss_cls_dn_0: 0.2410, loss_box_dn_0: 0.8055, loss_cls_dn_1: 0.1696, loss_box_dn_1: 0.7709, loss_cls_dn_2: 0.1712, loss_box_dn_2: 0.7572, loss_cls_dn_3: 0.1760, loss_box_dn_3: 0.7638, loss_cls_dn_4: 0.1775, loss_box_dn_4: 0.7770, loss_cls_dn_5: 0.1900, loss_box_dn_5: 0.7883, loss_dense_depth: 0.7779, loss: 27.9825, grad_norm: 51.3189 -2025-11-12 20:02:24,587 - mmdet - INFO - Iter [125/17500] lr: 1.496e-04, eta: 12:02:19, time: 1.566, data_time: 0.104, memory: 49167, loss_cls_0: 0.9829, loss_box_0: 1.7023, loss_cns_0: 0.6035, loss_yns_0: 0.1526, loss_cls_1: 0.9968, loss_box_1: 1.7694, loss_cns_1: 0.6524, loss_yns_1: 0.1579, loss_cls_2: 1.0036, loss_box_2: 1.7388, loss_cns_2: 0.6615, loss_yns_2: 0.1572, loss_cls_3: 1.0139, loss_box_3: 1.7365, loss_cns_3: 0.6658, loss_yns_3: 0.1586, loss_cls_4: 1.0076, loss_box_4: 1.7606, loss_cns_4: 0.6630, loss_yns_4: 0.1610, loss_cls_5: 1.0188, loss_box_5: 1.7707, loss_cns_5: 0.6587, loss_yns_5: 0.1577, loss_cls_dn_0: 0.2486, loss_box_dn_0: 0.7882, loss_cls_dn_1: 0.1706, loss_box_dn_1: 0.7626, loss_cls_dn_2: 0.1711, loss_box_dn_2: 0.7485, loss_cls_dn_3: 0.1758, loss_box_dn_3: 0.7492, loss_cls_dn_4: 0.1760, loss_box_dn_4: 0.7564, loss_cls_dn_5: 0.1863, loss_box_dn_5: 0.7673, loss_dense_depth: 0.8321, loss: 27.8845, grad_norm: 43.3002 -2025-11-12 20:02:26,169 - mmdet - INFO - Iter [126/17500] lr: 1.500e-04, eta: 12:00:11, time: 1.581, data_time: 0.074, memory: 49167, loss_cls_0: 0.9140, loss_box_0: 1.7768, loss_cns_0: 0.6188, loss_yns_0: 0.1576, loss_cls_1: 0.9937, loss_box_1: 1.8707, loss_cns_1: 0.6447, loss_yns_1: 0.1606, loss_cls_2: 1.0106, loss_box_2: 1.8256, loss_cns_2: 0.6523, loss_yns_2: 0.1617, loss_cls_3: 1.0236, loss_box_3: 1.8179, loss_cns_3: 0.6576, loss_yns_3: 0.1632, loss_cls_4: 1.0092, loss_box_4: 1.8170, loss_cns_4: 0.6529, loss_yns_4: 0.1611, loss_cls_5: 1.0251, loss_box_5: 1.8243, loss_cns_5: 0.6524, loss_yns_5: 0.1581, loss_cls_dn_0: 0.2405, loss_box_dn_0: 0.7892, loss_cls_dn_1: 0.1709, loss_box_dn_1: 0.7572, loss_cls_dn_2: 0.1766, loss_box_dn_2: 0.7435, loss_cls_dn_3: 0.1819, loss_box_dn_3: 0.7502, loss_cls_dn_4: 0.1841, loss_box_dn_4: 0.7576, loss_cls_dn_5: 0.1897, loss_box_dn_5: 0.7682, loss_dense_depth: 0.7978, loss: 28.2569, grad_norm: 45.4092 -2025-11-12 20:02:27,725 - mmdet - INFO - Iter [127/17500] lr: 1.504e-04, eta: 11:58:01, time: 1.554, data_time: 0.076, memory: 49167, loss_cls_0: 0.9314, loss_box_0: 1.7830, loss_cns_0: 0.6215, loss_yns_0: 0.1553, loss_cls_1: 0.9874, loss_box_1: 1.8161, loss_cns_1: 0.6449, loss_yns_1: 0.1617, loss_cls_2: 1.0101, loss_box_2: 1.7723, loss_cns_2: 0.6554, loss_yns_2: 0.1576, loss_cls_3: 1.0191, loss_box_3: 1.7655, loss_cns_3: 0.6581, loss_yns_3: 0.1579, loss_cls_4: 1.0035, loss_box_4: 1.7676, loss_cns_4: 0.6583, loss_yns_4: 0.1573, loss_cls_5: 1.0099, loss_box_5: 1.7813, loss_cns_5: 0.6611, loss_yns_5: 0.1583, loss_cls_dn_0: 0.2360, loss_box_dn_0: 0.7850, loss_cls_dn_1: 0.1695, loss_box_dn_1: 0.7719, loss_cls_dn_2: 0.1762, loss_box_dn_2: 0.7629, loss_cls_dn_3: 0.1823, loss_box_dn_3: 0.7759, loss_cls_dn_4: 0.1877, loss_box_dn_4: 0.7914, loss_cls_dn_5: 0.1932, loss_box_dn_5: 0.8110, loss_dense_depth: 0.8059, loss: 28.1437, grad_norm: 60.5699 -2025-11-12 20:02:29,285 - mmdet - INFO - Iter [128/17500] lr: 1.508e-04, eta: 11:55:54, time: 1.563, data_time: 0.080, memory: 49167, loss_cls_0: 0.9679, loss_box_0: 1.7627, loss_cns_0: 0.6262, loss_yns_0: 0.1553, loss_cls_1: 0.9724, loss_box_1: 1.8112, loss_cns_1: 0.6444, loss_yns_1: 0.1649, loss_cls_2: 0.9902, loss_box_2: 1.7635, loss_cns_2: 0.6565, loss_yns_2: 0.1569, loss_cls_3: 1.0055, loss_box_3: 1.7587, loss_cns_3: 0.6579, loss_yns_3: 0.1573, loss_cls_4: 1.0158, loss_box_4: 1.7421, loss_cns_4: 0.6572, loss_yns_4: 0.1586, loss_cls_5: 0.9961, loss_box_5: 1.7655, loss_cns_5: 0.6551, loss_yns_5: 0.1623, loss_cls_dn_0: 0.2356, loss_box_dn_0: 0.7865, loss_cls_dn_1: 0.1678, loss_box_dn_1: 0.7736, loss_cls_dn_2: 0.1696, loss_box_dn_2: 0.7687, loss_cls_dn_3: 0.1783, loss_box_dn_3: 0.7818, loss_cls_dn_4: 0.1808, loss_box_dn_4: 0.7928, loss_cls_dn_5: 0.1850, loss_box_dn_5: 0.8151, loss_dense_depth: 0.8528, loss: 28.0924, grad_norm: 52.2682 -2025-11-12 20:02:30,839 - mmdet - INFO - Iter [129/17500] lr: 1.512e-04, eta: 11:53:47, time: 1.550, data_time: 0.078, memory: 49167, loss_cls_0: 0.8986, loss_box_0: 1.7599, loss_cns_0: 0.6209, loss_yns_0: 0.1556, loss_cls_1: 0.9614, loss_box_1: 1.8107, loss_cns_1: 0.6429, loss_yns_1: 0.1564, loss_cls_2: 0.9860, loss_box_2: 1.7435, loss_cns_2: 0.6585, loss_yns_2: 0.1585, loss_cls_3: 1.0011, loss_box_3: 1.7546, loss_cns_3: 0.6619, loss_yns_3: 0.1573, loss_cls_4: 1.0344, loss_box_4: 1.7265, loss_cns_4: 0.6565, loss_yns_4: 0.1583, loss_cls_5: 1.0120, loss_box_5: 1.7516, loss_cns_5: 0.6552, loss_yns_5: 0.1595, loss_cls_dn_0: 0.2344, loss_box_dn_0: 0.7927, loss_cls_dn_1: 0.1654, loss_box_dn_1: 0.7877, loss_cls_dn_2: 0.1672, loss_box_dn_2: 0.7675, loss_cls_dn_3: 0.1772, loss_box_dn_3: 0.7770, loss_cls_dn_4: 0.1798, loss_box_dn_4: 0.7807, loss_cls_dn_5: 0.1834, loss_box_dn_5: 0.7964, loss_dense_depth: 0.8019, loss: 27.8934, grad_norm: 45.5645 -2025-11-12 20:02:32,415 - mmdet - INFO - Iter [130/17500] lr: 1.516e-04, eta: 11:51:47, time: 1.580, data_time: 0.073, memory: 49167, loss_cls_0: 0.9143, loss_box_0: 1.7544, loss_cns_0: 0.6146, loss_yns_0: 0.1525, loss_cls_1: 0.9908, loss_box_1: 1.8559, loss_cns_1: 0.6432, loss_yns_1: 0.1558, loss_cls_2: 1.0271, loss_box_2: 1.8085, loss_cns_2: 0.6534, loss_yns_2: 0.1598, loss_cls_3: 1.0225, loss_box_3: 1.8094, loss_cns_3: 0.6547, loss_yns_3: 0.1586, loss_cls_4: 1.0520, loss_box_4: 1.7813, loss_cns_4: 0.6512, loss_yns_4: 0.1573, loss_cls_5: 1.0446, loss_box_5: 1.7976, loss_cns_5: 0.6516, loss_yns_5: 0.1578, loss_cls_dn_0: 0.2464, loss_box_dn_0: 0.7966, loss_cls_dn_1: 0.1661, loss_box_dn_1: 0.8022, loss_cls_dn_2: 0.1767, loss_box_dn_2: 0.7853, loss_cls_dn_3: 0.1836, loss_box_dn_3: 0.7848, loss_cls_dn_4: 0.1901, loss_box_dn_4: 0.7877, loss_cls_dn_5: 0.1912, loss_box_dn_5: 0.7966, loss_dense_depth: 0.8114, loss: 28.3877, grad_norm: 53.2485 -2025-11-12 20:02:33,971 - mmdet - INFO - Iter [131/17500] lr: 1.520e-04, eta: 11:49:43, time: 1.548, data_time: 0.070, memory: 49167, loss_cls_0: 0.9205, loss_box_0: 1.7237, loss_cns_0: 0.6190, loss_yns_0: 0.1507, loss_cls_1: 0.9704, loss_box_1: 1.9007, loss_cns_1: 0.6416, loss_yns_1: 0.1576, loss_cls_2: 1.0163, loss_box_2: 1.8501, loss_cns_2: 0.6507, loss_yns_2: 0.1565, loss_cls_3: 1.0222, loss_box_3: 1.8294, loss_cns_3: 0.6505, loss_yns_3: 0.1576, loss_cls_4: 1.0252, loss_box_4: 1.8422, loss_cns_4: 0.6489, loss_yns_4: 0.1560, loss_cls_5: 1.0388, loss_box_5: 1.8332, loss_cns_5: 0.6505, loss_yns_5: 0.1575, loss_cls_dn_0: 0.2489, loss_box_dn_0: 0.7991, loss_cls_dn_1: 0.1672, loss_box_dn_1: 0.7908, loss_cls_dn_2: 0.1759, loss_box_dn_2: 0.7749, loss_cls_dn_3: 0.1806, loss_box_dn_3: 0.7679, loss_cls_dn_4: 0.1819, loss_box_dn_4: 0.7775, loss_cls_dn_5: 0.1862, loss_box_dn_5: 0.7797, loss_dense_depth: 0.8766, loss: 28.4768, grad_norm: 56.4857 -2025-11-12 20:02:35,541 - mmdet - INFO - Iter [132/17500] lr: 1.524e-04, eta: 11:47:45, time: 1.571, data_time: 0.087, memory: 49167, loss_cls_0: 0.8917, loss_box_0: 1.7309, loss_cns_0: 0.6186, loss_yns_0: 0.1496, loss_cls_1: 0.9613, loss_box_1: 1.8401, loss_cns_1: 0.6412, loss_yns_1: 0.1550, loss_cls_2: 0.9871, loss_box_2: 1.7756, loss_cns_2: 0.6509, loss_yns_2: 0.1507, loss_cls_3: 1.0181, loss_box_3: 1.7714, loss_cns_3: 0.6534, loss_yns_3: 0.1529, loss_cls_4: 1.0144, loss_box_4: 1.7963, loss_cns_4: 0.6496, loss_yns_4: 0.1531, loss_cls_5: 1.0178, loss_box_5: 1.7744, loss_cns_5: 0.6557, loss_yns_5: 0.1540, loss_cls_dn_0: 0.2476, loss_box_dn_0: 0.7854, loss_cls_dn_1: 0.1638, loss_box_dn_1: 0.7774, loss_cls_dn_2: 0.1657, loss_box_dn_2: 0.7667, loss_cls_dn_3: 0.1737, loss_box_dn_3: 0.7749, loss_cls_dn_4: 0.1774, loss_box_dn_4: 0.7930, loss_cls_dn_5: 0.1836, loss_box_dn_5: 0.8048, loss_dense_depth: 0.8288, loss: 28.0065, grad_norm: 57.0472 -2025-11-12 20:02:37,132 - mmdet - INFO - Iter [133/17500] lr: 1.528e-04, eta: 11:45:51, time: 1.587, data_time: 0.083, memory: 49167, loss_cls_0: 0.9232, loss_box_0: 1.7280, loss_cns_0: 0.6145, loss_yns_0: 0.1513, loss_cls_1: 0.9781, loss_box_1: 1.8137, loss_cns_1: 0.6446, loss_yns_1: 0.1540, loss_cls_2: 0.9892, loss_box_2: 1.8115, loss_cns_2: 0.6466, loss_yns_2: 0.1560, loss_cls_3: 1.0110, loss_box_3: 1.8045, loss_cns_3: 0.6595, loss_yns_3: 0.1541, loss_cls_4: 1.0087, loss_box_4: 1.7991, loss_cns_4: 0.6601, loss_yns_4: 0.1560, loss_cls_5: 1.0007, loss_box_5: 1.8174, loss_cns_5: 0.6549, loss_yns_5: 0.1561, loss_cls_dn_0: 0.2492, loss_box_dn_0: 0.7906, loss_cls_dn_1: 0.1646, loss_box_dn_1: 0.7770, loss_cls_dn_2: 0.1666, loss_box_dn_2: 0.7905, loss_cls_dn_3: 0.1736, loss_box_dn_3: 0.8075, loss_cls_dn_4: 0.1809, loss_box_dn_4: 0.8205, loss_cls_dn_5: 0.1918, loss_box_dn_5: 0.8558, loss_dense_depth: 0.8193, loss: 28.2806, grad_norm: 55.1862 -2025-11-12 20:02:38,713 - mmdet - INFO - Iter [134/17500] lr: 1.532e-04, eta: 11:43:57, time: 1.581, data_time: 0.084, memory: 49167, loss_cls_0: 0.8853, loss_box_0: 1.7228, loss_cns_0: 0.6226, loss_yns_0: 0.1527, loss_cls_1: 0.9574, loss_box_1: 1.8462, loss_cns_1: 0.6487, loss_yns_1: 0.1521, loss_cls_2: 0.9769, loss_box_2: 1.8398, loss_cns_2: 0.6481, loss_yns_2: 0.1557, loss_cls_3: 0.9951, loss_box_3: 1.8132, loss_cns_3: 0.6594, loss_yns_3: 0.1540, loss_cls_4: 1.0012, loss_box_4: 1.8002, loss_cns_4: 0.6582, loss_yns_4: 0.1546, loss_cls_5: 0.9899, loss_box_5: 1.8221, loss_cns_5: 0.6532, loss_yns_5: 0.1546, loss_cls_dn_0: 0.2349, loss_box_dn_0: 0.7886, loss_cls_dn_1: 0.1669, loss_box_dn_1: 0.8012, loss_cls_dn_2: 0.1722, loss_box_dn_2: 0.8117, loss_cls_dn_3: 0.1766, loss_box_dn_3: 0.8189, loss_cls_dn_4: 0.1821, loss_box_dn_4: 0.8276, loss_cls_dn_5: 0.1945, loss_box_dn_5: 0.8586, loss_dense_depth: 0.8179, loss: 28.3159, grad_norm: 55.6087 -2025-11-12 20:02:40,295 - mmdet - INFO - Iter [135/17500] lr: 1.536e-04, eta: 11:42:05, time: 1.583, data_time: 0.081, memory: 49167, loss_cls_0: 0.8789, loss_box_0: 1.7556, loss_cns_0: 0.6233, loss_yns_0: 0.1528, loss_cls_1: 0.9621, loss_box_1: 1.9052, loss_cns_1: 0.6464, loss_yns_1: 0.1561, loss_cls_2: 0.9881, loss_box_2: 1.8520, loss_cns_2: 0.6487, loss_yns_2: 0.1556, loss_cls_3: 1.0092, loss_box_3: 1.8389, loss_cns_3: 0.6574, loss_yns_3: 0.1552, loss_cls_4: 0.9997, loss_box_4: 1.8351, loss_cns_4: 0.6519, loss_yns_4: 0.1543, loss_cls_5: 1.0097, loss_box_5: 1.8376, loss_cns_5: 0.6530, loss_yns_5: 0.1544, loss_cls_dn_0: 0.2405, loss_box_dn_0: 0.7876, loss_cls_dn_1: 0.1663, loss_box_dn_1: 0.8050, loss_cls_dn_2: 0.1749, loss_box_dn_2: 0.7956, loss_cls_dn_3: 0.1859, loss_box_dn_3: 0.7907, loss_cls_dn_4: 0.1831, loss_box_dn_4: 0.8010, loss_cls_dn_5: 0.1986, loss_box_dn_5: 0.8109, loss_dense_depth: 0.8143, loss: 28.4359, grad_norm: 37.5335 -2025-11-12 20:02:41,857 - mmdet - INFO - Iter [136/17500] lr: 1.540e-04, eta: 11:40:13, time: 1.563, data_time: 0.080, memory: 49167, loss_cls_0: 0.8513, loss_box_0: 1.7166, loss_cns_0: 0.6256, loss_yns_0: 0.1511, loss_cls_1: 0.9253, loss_box_1: 1.8320, loss_cns_1: 0.6443, loss_yns_1: 0.1485, loss_cls_2: 0.9508, loss_box_2: 1.7631, loss_cns_2: 0.6538, loss_yns_2: 0.1501, loss_cls_3: 0.9911, loss_box_3: 1.7673, loss_cns_3: 0.6578, loss_yns_3: 0.1499, loss_cls_4: 0.9694, loss_box_4: 1.7530, loss_cns_4: 0.6594, loss_yns_4: 0.1501, loss_cls_5: 0.9677, loss_box_5: 1.7531, loss_cns_5: 0.6594, loss_yns_5: 0.1496, loss_cls_dn_0: 0.2362, loss_box_dn_0: 0.7763, loss_cls_dn_1: 0.1594, loss_box_dn_1: 0.7795, loss_cls_dn_2: 0.1634, loss_box_dn_2: 0.7588, loss_cls_dn_3: 0.1733, loss_box_dn_3: 0.7593, loss_cls_dn_4: 0.1718, loss_box_dn_4: 0.7650, loss_cls_dn_5: 0.1812, loss_box_dn_5: 0.7687, loss_dense_depth: 0.8238, loss: 27.5569, grad_norm: 43.7536 -2025-11-12 20:02:43,420 - mmdet - INFO - Iter [137/17500] lr: 1.544e-04, eta: 11:38:22, time: 1.561, data_time: 0.083, memory: 49167, loss_cls_0: 0.8636, loss_box_0: 1.7245, loss_cns_0: 0.6208, loss_yns_0: 0.1519, loss_cls_1: 0.9274, loss_box_1: 1.8560, loss_cns_1: 0.6459, loss_yns_1: 0.1507, loss_cls_2: 0.9478, loss_box_2: 1.7747, loss_cns_2: 0.6508, loss_yns_2: 0.1523, loss_cls_3: 0.9798, loss_box_3: 1.7762, loss_cns_3: 0.6556, loss_yns_3: 0.1511, loss_cls_4: 0.9709, loss_box_4: 1.7606, loss_cns_4: 0.6566, loss_yns_4: 0.1511, loss_cls_5: 0.9629, loss_box_5: 1.7664, loss_cns_5: 0.6588, loss_yns_5: 0.1512, loss_cls_dn_0: 0.2458, loss_box_dn_0: 0.7845, loss_cls_dn_1: 0.1608, loss_box_dn_1: 0.7626, loss_cls_dn_2: 0.1616, loss_box_dn_2: 0.7380, loss_cls_dn_3: 0.1669, loss_box_dn_3: 0.7361, loss_cls_dn_4: 0.1684, loss_box_dn_4: 0.7374, loss_cls_dn_5: 0.1757, loss_box_dn_5: 0.7414, loss_dense_depth: 0.8122, loss: 27.4989, grad_norm: 37.1297 -2025-11-12 20:02:44,998 - mmdet - INFO - Iter [138/17500] lr: 1.548e-04, eta: 11:36:34, time: 1.578, data_time: 0.083, memory: 49167, loss_cls_0: 0.8585, loss_box_0: 1.7259, loss_cns_0: 0.6282, loss_yns_0: 0.1489, loss_cls_1: 0.9235, loss_box_1: 1.8052, loss_cns_1: 0.6495, loss_yns_1: 0.1511, loss_cls_2: 0.9527, loss_box_2: 1.7544, loss_cns_2: 0.6504, loss_yns_2: 0.1518, loss_cls_3: 0.9639, loss_box_3: 1.7276, loss_cns_3: 0.6567, loss_yns_3: 0.1509, loss_cls_4: 0.9660, loss_box_4: 1.7325, loss_cns_4: 0.6563, loss_yns_4: 0.1510, loss_cls_5: 0.9971, loss_box_5: 1.7387, loss_cns_5: 0.6568, loss_yns_5: 0.1502, loss_cls_dn_0: 0.2403, loss_box_dn_0: 0.7928, loss_cls_dn_1: 0.1599, loss_box_dn_1: 0.7532, loss_cls_dn_2: 0.1607, loss_box_dn_2: 0.7396, loss_cls_dn_3: 0.1629, loss_box_dn_3: 0.7369, loss_cls_dn_4: 0.1672, loss_box_dn_4: 0.7452, loss_cls_dn_5: 0.1769, loss_box_dn_5: 0.7531, loss_dense_depth: 0.8459, loss: 27.3825, grad_norm: 33.6138 -2025-11-12 20:02:46,568 - mmdet - INFO - Iter [139/17500] lr: 1.552e-04, eta: 11:34:47, time: 1.569, data_time: 0.080, memory: 49167, loss_cls_0: 0.8131, loss_box_0: 1.6894, loss_cns_0: 0.6263, loss_yns_0: 0.1479, loss_cls_1: 0.8942, loss_box_1: 1.7806, loss_cns_1: 0.6460, loss_yns_1: 0.1490, loss_cls_2: 0.9209, loss_box_2: 1.7408, loss_cns_2: 0.6535, loss_yns_2: 0.1489, loss_cls_3: 0.9412, loss_box_3: 1.7125, loss_cns_3: 0.6567, loss_yns_3: 0.1488, loss_cls_4: 0.9356, loss_box_4: 1.7237, loss_cns_4: 0.6539, loss_yns_4: 0.1494, loss_cls_5: 0.9305, loss_box_5: 1.7290, loss_cns_5: 0.6530, loss_yns_5: 0.1478, loss_cls_dn_0: 0.2280, loss_box_dn_0: 0.7895, loss_cls_dn_1: 0.1554, loss_box_dn_1: 0.7623, loss_cls_dn_2: 0.1630, loss_box_dn_2: 0.7607, loss_cls_dn_3: 0.1609, loss_box_dn_3: 0.7627, loss_cls_dn_4: 0.1668, loss_box_dn_4: 0.7783, loss_cls_dn_5: 0.1748, loss_box_dn_5: 0.7973, loss_dense_depth: 0.8434, loss: 27.1360, grad_norm: 40.8107 -2025-11-12 20:02:48,141 - mmdet - INFO - Iter [140/17500] lr: 1.556e-04, eta: 11:33:02, time: 1.576, data_time: 0.083, memory: 49167, loss_cls_0: 0.8268, loss_box_0: 1.6812, loss_cns_0: 0.6245, loss_yns_0: 0.1480, loss_cls_1: 0.8882, loss_box_1: 1.7896, loss_cns_1: 0.6463, loss_yns_1: 0.1474, loss_cls_2: 0.9355, loss_box_2: 1.7326, loss_cns_2: 0.6535, loss_yns_2: 0.1469, loss_cls_3: 0.9527, loss_box_3: 1.7124, loss_cns_3: 0.6580, loss_yns_3: 0.1472, loss_cls_4: 0.9382, loss_box_4: 1.7156, loss_cns_4: 0.6586, loss_yns_4: 0.1485, loss_cls_5: 0.9454, loss_box_5: 1.7128, loss_cns_5: 0.6593, loss_yns_5: 0.1470, loss_cls_dn_0: 0.2268, loss_box_dn_0: 0.7874, loss_cls_dn_1: 0.1495, loss_box_dn_1: 0.7895, loss_cls_dn_2: 0.1639, loss_box_dn_2: 0.7841, loss_cls_dn_3: 0.1589, loss_box_dn_3: 0.7880, loss_cls_dn_4: 0.1655, loss_box_dn_4: 0.8003, loss_cls_dn_5: 0.1772, loss_box_dn_5: 0.8153, loss_dense_depth: 0.8065, loss: 27.2293, grad_norm: 44.5335 -2025-11-12 20:02:49,821 - mmdet - INFO - Iter [141/17500] lr: 1.560e-04, eta: 11:31:32, time: 1.680, data_time: 0.097, memory: 49167, loss_cls_0: 0.8878, loss_box_0: 1.7677, loss_cns_0: 0.6283, loss_yns_0: 0.1528, loss_cls_1: 0.9357, loss_box_1: 1.8552, loss_cns_1: 0.6543, loss_yns_1: 0.1488, loss_cls_2: 0.9684, loss_box_2: 1.7805, loss_cns_2: 0.6558, loss_yns_2: 0.1501, loss_cls_3: 0.9647, loss_box_3: 1.7870, loss_cns_3: 0.6576, loss_yns_3: 0.1502, loss_cls_4: 0.9581, loss_box_4: 1.7813, loss_cns_4: 0.6598, loss_yns_4: 0.1524, loss_cls_5: 0.9745, loss_box_5: 1.7696, loss_cns_5: 0.6615, loss_yns_5: 0.1504, loss_cls_dn_0: 0.2429, loss_box_dn_0: 0.7860, loss_cls_dn_1: 0.1620, loss_box_dn_1: 0.7896, loss_cls_dn_2: 0.1717, loss_box_dn_2: 0.7723, loss_cls_dn_3: 0.1674, loss_box_dn_3: 0.7800, loss_cls_dn_4: 0.1736, loss_box_dn_4: 0.7847, loss_cls_dn_5: 0.1820, loss_box_dn_5: 0.7902, loss_dense_depth: 0.8367, loss: 27.8916, grad_norm: 38.7864 -2025-11-12 20:02:51,508 - mmdet - INFO - Iter [142/17500] lr: 1.564e-04, eta: 11:30:03, time: 1.682, data_time: 0.080, memory: 49167, loss_cls_0: 0.8857, loss_box_0: 1.7896, loss_cns_0: 0.6205, loss_yns_0: 0.1518, loss_cls_1: 0.9520, loss_box_1: 1.8670, loss_cns_1: 0.6466, loss_yns_1: 0.1508, loss_cls_2: 0.9702, loss_box_2: 1.8005, loss_cns_2: 0.6542, loss_yns_2: 0.1528, loss_cls_3: 0.9699, loss_box_3: 1.8124, loss_cns_3: 0.6545, loss_yns_3: 0.1517, loss_cls_4: 0.9747, loss_box_4: 1.7922, loss_cns_4: 0.6535, loss_yns_4: 0.1509, loss_cls_5: 0.9709, loss_box_5: 1.7992, loss_cns_5: 0.6558, loss_yns_5: 0.1510, loss_cls_dn_0: 0.2363, loss_box_dn_0: 0.7895, loss_cls_dn_1: 0.1625, loss_box_dn_1: 0.7792, loss_cls_dn_2: 0.1642, loss_box_dn_2: 0.7545, loss_cls_dn_3: 0.1653, loss_box_dn_3: 0.7668, loss_cls_dn_4: 0.1766, loss_box_dn_4: 0.7667, loss_cls_dn_5: 0.1787, loss_box_dn_5: 0.7734, loss_dense_depth: 0.8184, loss: 27.9108, grad_norm: 42.2026 -2025-11-12 20:02:53,140 - mmdet - INFO - Iter [143/17500] lr: 1.568e-04, eta: 11:28:29, time: 1.635, data_time: 0.142, memory: 49167, loss_cls_0: 0.8384, loss_box_0: 1.7761, loss_cns_0: 0.6238, loss_yns_0: 0.1478, loss_cls_1: 0.9259, loss_box_1: 1.8029, loss_cns_1: 0.6511, loss_yns_1: 0.1481, loss_cls_2: 0.9511, loss_box_2: 1.7613, loss_cns_2: 0.6577, loss_yns_2: 0.1490, loss_cls_3: 0.9636, loss_box_3: 1.7710, loss_cns_3: 0.6573, loss_yns_3: 0.1493, loss_cls_4: 0.9551, loss_box_4: 1.7388, loss_cns_4: 0.6579, loss_yns_4: 0.1490, loss_cls_5: 0.9572, loss_box_5: 1.7481, loss_cns_5: 0.6594, loss_yns_5: 0.1489, loss_cls_dn_0: 0.2229, loss_box_dn_0: 0.7918, loss_cls_dn_1: 0.1561, loss_box_dn_1: 0.7483, loss_cls_dn_2: 0.1642, loss_box_dn_2: 0.7325, loss_cls_dn_3: 0.1651, loss_box_dn_3: 0.7445, loss_cls_dn_4: 0.1793, loss_box_dn_4: 0.7418, loss_cls_dn_5: 0.1818, loss_box_dn_5: 0.7476, loss_dense_depth: 0.8046, loss: 27.3692, grad_norm: 42.7491 -2025-11-12 20:02:54,696 - mmdet - INFO - Iter [144/17500] lr: 1.572e-04, eta: 11:26:48, time: 1.557, data_time: 0.082, memory: 49167, loss_cls_0: 0.8533, loss_box_0: 1.7751, loss_cns_0: 0.6186, loss_yns_0: 0.1480, loss_cls_1: 0.9368, loss_box_1: 1.8074, loss_cns_1: 0.6462, loss_yns_1: 0.1480, loss_cls_2: 0.9586, loss_box_2: 1.7788, loss_cns_2: 0.6540, loss_yns_2: 0.1508, loss_cls_3: 1.0126, loss_box_3: 1.7808, loss_cns_3: 0.6536, loss_yns_3: 0.1487, loss_cls_4: 0.9808, loss_box_4: 1.7632, loss_cns_4: 0.6540, loss_yns_4: 0.1508, loss_cls_5: 0.9789, loss_box_5: 1.7521, loss_cns_5: 0.6556, loss_yns_5: 0.1489, loss_cls_dn_0: 0.2318, loss_box_dn_0: 0.7968, loss_cls_dn_1: 0.1572, loss_box_dn_1: 0.7569, loss_cls_dn_2: 0.1659, loss_box_dn_2: 0.7505, loss_cls_dn_3: 0.1697, loss_box_dn_3: 0.7582, loss_cls_dn_4: 0.1751, loss_box_dn_4: 0.7698, loss_cls_dn_5: 0.1849, loss_box_dn_5: 0.7734, loss_dense_depth: 0.7876, loss: 27.6332, grad_norm: 47.7396 -2025-11-12 20:02:56,280 - mmdet - INFO - Iter [145/17500] lr: 1.576e-04, eta: 11:25:11, time: 1.585, data_time: 0.109, memory: 49167, loss_cls_0: 0.8458, loss_box_0: 1.7856, loss_cns_0: 0.6234, loss_yns_0: 0.1512, loss_cls_1: 0.9391, loss_box_1: 1.7978, loss_cns_1: 0.6488, loss_yns_1: 0.1520, loss_cls_2: 0.9543, loss_box_2: 1.7645, loss_cns_2: 0.6522, loss_yns_2: 0.1515, loss_cls_3: 0.9964, loss_box_3: 1.7618, loss_cns_3: 0.6529, loss_yns_3: 0.1497, loss_cls_4: 0.9804, loss_box_4: 1.7574, loss_cns_4: 0.6533, loss_yns_4: 0.1483, loss_cls_5: 0.9628, loss_box_5: 1.7683, loss_cns_5: 0.6590, loss_yns_5: 0.1499, loss_cls_dn_0: 0.2285, loss_box_dn_0: 0.7924, loss_cls_dn_1: 0.1545, loss_box_dn_1: 0.7639, loss_cls_dn_2: 0.1597, loss_box_dn_2: 0.7579, loss_cls_dn_3: 0.1637, loss_box_dn_3: 0.7693, loss_cls_dn_4: 0.1670, loss_box_dn_4: 0.7831, loss_cls_dn_5: 0.1735, loss_box_dn_5: 0.7975, loss_dense_depth: 0.8122, loss: 27.6296, grad_norm: 47.2437 -2025-11-12 20:02:57,869 - mmdet - INFO - Iter [146/17500] lr: 1.580e-04, eta: 11:23:36, time: 1.589, data_time: 0.080, memory: 49167, loss_cls_0: 0.8501, loss_box_0: 1.7848, loss_cns_0: 0.6253, loss_yns_0: 0.1486, loss_cls_1: 0.9226, loss_box_1: 1.7894, loss_cns_1: 0.6497, loss_yns_1: 0.1514, loss_cls_2: 0.9460, loss_box_2: 1.7570, loss_cns_2: 0.6520, loss_yns_2: 0.1506, loss_cls_3: 0.9539, loss_box_3: 1.7559, loss_cns_3: 0.6529, loss_yns_3: 0.1516, loss_cls_4: 0.9646, loss_box_4: 1.7506, loss_cns_4: 0.6537, loss_yns_4: 0.1488, loss_cls_5: 0.9589, loss_box_5: 1.7604, loss_cns_5: 0.6586, loss_yns_5: 0.1500, loss_cls_dn_0: 0.2212, loss_box_dn_0: 0.7900, loss_cls_dn_1: 0.1538, loss_box_dn_1: 0.7791, loss_cls_dn_2: 0.1554, loss_box_dn_2: 0.7752, loss_cls_dn_3: 0.1566, loss_box_dn_3: 0.7959, loss_cls_dn_4: 0.1682, loss_box_dn_4: 0.8051, loss_cls_dn_5: 0.1736, loss_box_dn_5: 0.8283, loss_dense_depth: 0.7765, loss: 27.5667, grad_norm: 44.4904 -2025-11-12 20:02:59,447 - mmdet - INFO - Iter [147/17500] lr: 1.584e-04, eta: 11:22:01, time: 1.579, data_time: 0.080, memory: 49167, loss_cls_0: 0.8574, loss_box_0: 1.7830, loss_cns_0: 0.6246, loss_yns_0: 0.1520, loss_cls_1: 0.9374, loss_box_1: 1.8256, loss_cns_1: 0.6506, loss_yns_1: 0.1505, loss_cls_2: 0.9637, loss_box_2: 1.7797, loss_cns_2: 0.6552, loss_yns_2: 0.1492, loss_cls_3: 0.9924, loss_box_3: 1.7604, loss_cns_3: 0.6556, loss_yns_3: 0.1509, loss_cls_4: 0.9691, loss_box_4: 1.7587, loss_cns_4: 0.6568, loss_yns_4: 0.1496, loss_cls_5: 0.9735, loss_box_5: 1.7567, loss_cns_5: 0.6584, loss_yns_5: 0.1488, loss_cls_dn_0: 0.2212, loss_box_dn_0: 0.7834, loss_cls_dn_1: 0.1532, loss_box_dn_1: 0.7998, loss_cls_dn_2: 0.1551, loss_box_dn_2: 0.8045, loss_cls_dn_3: 0.1624, loss_box_dn_3: 0.8158, loss_cls_dn_4: 0.1687, loss_box_dn_4: 0.8239, loss_cls_dn_5: 0.1754, loss_box_dn_5: 0.8432, loss_dense_depth: 0.7843, loss: 27.8508, grad_norm: 45.6622 -2025-11-12 20:03:00,997 - mmdet - INFO - Iter [148/17500] lr: 1.588e-04, eta: 11:20:24, time: 1.549, data_time: 0.083, memory: 49167, loss_cls_0: 0.8347, loss_box_0: 1.7400, loss_cns_0: 0.6257, loss_yns_0: 0.1478, loss_cls_1: 0.9302, loss_box_1: 1.7545, loss_cns_1: 0.6539, loss_yns_1: 0.1494, loss_cls_2: 0.9438, loss_box_2: 1.7146, loss_cns_2: 0.6602, loss_yns_2: 0.1473, loss_cls_3: 0.9834, loss_box_3: 1.6887, loss_cns_3: 0.6621, loss_yns_3: 0.1467, loss_cls_4: 0.9611, loss_box_4: 1.6959, loss_cns_4: 0.6618, loss_yns_4: 0.1506, loss_cls_5: 0.9647, loss_box_5: 1.6876, loss_cns_5: 0.6622, loss_yns_5: 0.1479, loss_cls_dn_0: 0.2184, loss_box_dn_0: 0.7816, loss_cls_dn_1: 0.1529, loss_box_dn_1: 0.7951, loss_cls_dn_2: 0.1571, loss_box_dn_2: 0.7940, loss_cls_dn_3: 0.1661, loss_box_dn_3: 0.7958, loss_cls_dn_4: 0.1689, loss_box_dn_4: 0.8049, loss_cls_dn_5: 0.1725, loss_box_dn_5: 0.8188, loss_dense_depth: 0.7592, loss: 27.3003, grad_norm: 44.2248 diff --git a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20251112_195656.log.json b/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20251112_195656.log.json deleted file mode 100644 index 1d91bf6aec8bea3ce6e937dc5842417f8815bf87..0000000000000000000000000000000000000000 --- a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20251112_195656.log.json +++ /dev/null @@ -1,149 +0,0 @@ -{"env_info": "sys.platform: linux\nPython: 3.10.12 (main, May 27 2025, 17:12:29) [GCC 11.4.0]\nCUDA available: True\nGPU 0,1,2,3,4,5,6,7: BW200, UBB BW1000\nCUDA_HOME: /opt/dtk\nNVCC: Not Available\nGCC: x86_64-linux-gnu-gcc (Ubuntu 11.4.0-1ubuntu1~22.04) 11.4.0\nPyTorch: 2.4.1\nPyTorch compiling details: PyTorch built with:\n - GCC 10.3\n - C++ Version: 201703\n - Intel(R) Math Kernel Library Version 2020.0.4 Product Build 20200917 for Intel(R) 64 architecture applications\n - OpenMP 201511 (a.k.a. OpenMP 4.5)\n - LAPACK is enabled (usually provided by MKL)\n - NNPACK is enabled\n - CPU capability usage: AVX512\n - HIP Runtime 6.3.25211\n - MIOpen 2.17.0\n - Magma 2.8.0\n - Build settings: BLAS_INFO=mkl, BUILD_TYPE=Release, CXX_COMPILER=/opt/rh/gcc-toolset-10/root/usr/bin/c++, CXX_FLAGS= -D_GLIBCXX_USE_CXX11_ABI=1 -fvisibility-inlines-hidden -DUSE_PTHREADPOOL -DNDEBUG -DUSE_KINETO -DLIBKINETO_NOCUPTI -DUSE_FBGEMM -DUSE_PYTORCH_QNNPACK -DUSE_XNNPACK -DSYMBOLICATE_MOBILE_DEBUG_HANDLE -O2 -fPIC -Wall -Wextra -Werror=return-type -Werror=non-virtual-dtor -Werror=bool-operation -Wnarrowing -Wno-missing-field-initializers -Wno-type-limits -Wno-array-bounds -Wno-unknown-pragmas -Wno-unused-parameter -Wno-unused-function -Wno-unused-result -Wno-strict-overflow -Wno-strict-aliasing -Wno-stringop-overflow -Wsuggest-override -Wno-psabi -Wno-error=pedantic -Wno-error=old-style-cast -Wno-missing-braces -fdiagnostics-color=always -faligned-new -Wno-unused-but-set-variable -Wno-maybe-uninitialized -fno-math-errno -fno-trapping-math -Werror=format -Wno-stringop-overflow, FORCE_FALLBACK_CUDA_MPI=1, LAPACK_INFO=mkl, PERF_WITH_AVX=1, PERF_WITH_AVX2=1, PERF_WITH_AVX512=1, TORCH_VERSION=2.4.1, USE_CUDA=0, USE_CUDNN=OFF, USE_CUSPARSELT=OFF, USE_EXCEPTION_PTR=1, USE_GFLAGS=1, USE_GLOG=1, USE_GLOO=1, USE_MKL=ON, USE_MKLDNN=0, USE_MPI=1, USE_NCCL=1, USE_NNPACK=ON, USE_OPENMP=1, USE_ROCM=ON, USE_ROCM_KERNEL_ASSERT=OFF, \n\nTorchVision: 0.19.1\nOpenCV: 4.11.0\nMMCV: 1.6.1\nMMCV Compiler: GCC 11.4\nMMCV CUDA Compiler: rocm not available\nMMDetection: 2.26.0+c41df4b", "config": "plugin = True\nplugin_dir = 'projects/mmdet3d_plugin/'\ndist_params = dict(backend='nccl')\nlog_level = 'INFO'\nwork_dir = './work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704'\ntotal_batch_size = 160\nnum_gpus = 8\nbatch_size = 20\nnum_iters_per_epoch = 175\nnum_epochs = 100\ncheckpoint_epoch_interval = 20\ncheckpoint_config = dict(interval=3500)\nlog_config = dict(\n interval=1,\n hooks=[\n dict(type='TextLoggerHook', by_epoch=False),\n dict(type='TensorboardLoggerHook')\n ])\nload_from = None\nresume_from = None\nworkflow = [('train', 1)]\nfp16 = dict(loss_scale=32.0)\ninput_shape = (704, 256)\ntracking_test = True\ntracking_threshold = 0.2\nclass_names = [\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier',\n 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n]\nnum_classes = 10\nembed_dims = 256\nnum_groups = 8\nnum_decoder = 6\nnum_single_frame_decoder = 1\nuse_deformable_func = True\nstrides = [4, 8, 16, 32]\nnum_levels = 4\nnum_depth_layers = 3\ndrop_out = 0.1\ntemporal = True\ndecouple_attn = True\nwith_quality_estimation = True\nmodel = dict(\n type='Sparse4D',\n use_grid_mask=True,\n use_deformable_func=True,\n img_backbone=dict(\n type='ResNet',\n depth=50,\n num_stages=4,\n frozen_stages=-1,\n norm_eval=False,\n style='pytorch',\n with_cp=True,\n out_indices=(0, 1, 2, 3),\n norm_cfg=dict(type='BN', requires_grad=True),\n pretrained='ckpt/resnet50-19c8e357.pth'),\n img_neck=dict(\n type='FPN',\n num_outs=4,\n start_level=0,\n out_channels=256,\n add_extra_convs='on_output',\n relu_before_extra_convs=True,\n in_channels=[256, 512, 1024, 2048]),\n depth_branch=dict(\n type='DenseDepthNet',\n embed_dims=256,\n num_depth_layers=3,\n loss_weight=0.2),\n head=dict(\n type='Sparse4DHead',\n cls_threshold_to_reg=0.05,\n decouple_attn=True,\n instance_bank=dict(\n type='InstanceBank',\n num_anchor=900,\n embed_dims=256,\n anchor='nuscenes_kmeans900.npy',\n anchor_handler=dict(type='SparseBox3DKeyPointsGenerator'),\n num_temp_instances=600,\n confidence_decay=0.6,\n feat_grad=False),\n anchor_encoder=dict(\n type='SparseBox3DEncoder',\n vel_dims=3,\n embed_dims=[128, 32, 32, 64],\n mode='cat',\n output_fc=False,\n in_loops=1,\n out_loops=4),\n num_single_frame_decoder=1,\n operation_order=[\n 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm',\n 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm',\n 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm',\n 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm',\n 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm',\n 'deformable', 'ffn', 'norm', 'refine'\n ],\n temp_graph_model=dict(\n type='MultiheadAttention',\n embed_dims=512,\n num_heads=8,\n batch_first=True,\n dropout=0.1),\n graph_model=dict(\n type='MultiheadAttention',\n embed_dims=512,\n num_heads=8,\n batch_first=True,\n dropout=0.1),\n norm_layer=dict(type='LN', normalized_shape=256),\n ffn=dict(\n type='AsymmetricFFN',\n in_channels=512,\n pre_norm=dict(type='LN'),\n embed_dims=256,\n feedforward_channels=1024,\n num_fcs=2,\n ffn_drop=0.1,\n act_cfg=dict(type='ReLU', inplace=True)),\n deformable_model=dict(\n type='DeformableFeatureAggregation',\n embed_dims=256,\n num_groups=8,\n num_levels=4,\n num_cams=6,\n attn_drop=0.15,\n use_deformable_func=True,\n use_camera_embed=True,\n residual_mode='cat',\n kps_generator=dict(\n type='SparseBox3DKeyPointsGenerator',\n num_learnable_pts=6,\n fix_scale=[[0, 0, 0], [0.45, 0, 0], [-0.45, 0, 0],\n [0, 0.45, 0], [0, -0.45, 0], [0, 0, 0.45],\n [0, 0, -0.45]])),\n refine_layer=dict(\n type='SparseBox3DRefinementModule',\n embed_dims=256,\n num_cls=10,\n refine_yaw=True,\n with_quality_estimation=True),\n sampler=dict(\n type='SparseBox3DTarget',\n num_dn_groups=5,\n num_temp_dn_groups=3,\n dn_noise_scale=[2.0, 2.0, 2.0, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5],\n max_dn_gt=32,\n add_neg_dn=True,\n cls_weight=2.0,\n box_weight=0.25,\n reg_weights=[2.0, 2.0, 2.0, 0.5, 0.5, 0.5, 0.0, 0.0, 0.0, 0.0],\n cls_wise_reg_weights=dict(\n {9: [2.0, 2.0, 2.0, 1.0, 1.0, 1.0, 0.0, 0.0, 1.0, 1.0]})),\n loss_cls=dict(\n type='FocalLoss',\n use_sigmoid=True,\n gamma=2.0,\n alpha=0.25,\n loss_weight=2.0),\n loss_reg=dict(\n type='SparseBox3DLoss',\n loss_box=dict(type='L1Loss', loss_weight=0.25),\n loss_centerness=dict(type='CrossEntropyLoss', use_sigmoid=True),\n loss_yawness=dict(type='GaussianFocalLoss'),\n cls_allow_reverse=[5]),\n decoder=dict(type='SparseBox3DDecoder'),\n reg_weights=[2.0, 2.0, 2.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0]))\ndataset_type = 'NuScenes3DDetTrackDataset'\ndata_root = 'data/nuscenes/'\nanno_root = 'data/nuscenes_anno_pkls/'\nfile_client_args = dict(backend='disk')\nimg_norm_cfg = dict(\n mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375], to_rgb=True)\ntrain_pipeline = [\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(\n type='LoadPointsFromFile',\n coord_type='LIDAR',\n load_dim=5,\n use_dim=5,\n file_client_args=dict(backend='disk')),\n dict(type='ResizeCropFlipImage'),\n dict(type='MultiScaleDepthMapGenerator', downsample=[4, 8, 16]),\n dict(type='BBoxRotation'),\n dict(type='PhotoMetricDistortionMultiViewImage'),\n dict(\n type='NormalizeMultiviewImage',\n mean=[123.675, 116.28, 103.53],\n std=[58.395, 57.12, 57.375],\n to_rgb=True),\n dict(\n type='CircleObjectRangeFilter',\n class_dist_thred=[55, 55, 55, 55, 55, 55, 55, 55, 55, 55]),\n dict(\n type='InstanceNameFilter',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ]),\n dict(type='NuScenesSparse4DAdaptor'),\n dict(\n type='Collect',\n keys=[\n 'img', 'timestamp', 'projection_mat', 'image_wh', 'gt_depth',\n 'focal', 'gt_bboxes_3d', 'gt_labels_3d'\n ],\n meta_keys=['T_global', 'T_global_inv', 'timestamp', 'instance_id'])\n]\ntest_pipeline = [\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(type='ResizeCropFlipImage'),\n dict(\n type='NormalizeMultiviewImage',\n mean=[123.675, 116.28, 103.53],\n std=[58.395, 57.12, 57.375],\n to_rgb=True),\n dict(type='NuScenesSparse4DAdaptor'),\n dict(\n type='Collect',\n keys=['img', 'timestamp', 'projection_mat', 'image_wh'],\n meta_keys=['T_global', 'T_global_inv', 'timestamp'])\n]\ninput_modality = dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False)\ndata_basic_config = dict(\n type='NuScenes3DDetTrackDataset',\n data_root='data/nuscenes/',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier',\n 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ],\n modality=dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False),\n version='v1.0-trainval')\ndata_aug_conf = dict(\n resize_lim=(0.4, 0.47),\n final_dim=(256, 704),\n bot_pct_lim=(0.0, 0.0),\n rot_lim=(-5.4, 5.4),\n H=900,\n W=1600,\n rand_flip=True,\n rot3d_range=[-0.3925, 0.3925])\ndata = dict(\n samples_per_gpu=20,\n workers_per_gpu=20,\n train=dict(\n type='NuScenes3DDetTrackDataset',\n data_root='data/nuscenes/',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ],\n modality=dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False),\n version='v1.0-trainval',\n ann_file='data/nuscenes_anno_pkls/nuscenes_infos_train.pkl',\n pipeline=[\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(\n type='LoadPointsFromFile',\n coord_type='LIDAR',\n load_dim=5,\n use_dim=5,\n file_client_args=dict(backend='disk')),\n dict(type='ResizeCropFlipImage'),\n dict(type='MultiScaleDepthMapGenerator', downsample=[4, 8, 16]),\n dict(type='BBoxRotation'),\n dict(type='PhotoMetricDistortionMultiViewImage'),\n dict(\n type='NormalizeMultiviewImage',\n mean=[123.675, 116.28, 103.53],\n std=[58.395, 57.12, 57.375],\n to_rgb=True),\n dict(\n type='CircleObjectRangeFilter',\n class_dist_thred=[55, 55, 55, 55, 55, 55, 55, 55, 55, 55]),\n dict(\n type='InstanceNameFilter',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian',\n 'traffic_cone'\n ]),\n dict(type='NuScenesSparse4DAdaptor'),\n dict(\n type='Collect',\n keys=[\n 'img', 'timestamp', 'projection_mat', 'image_wh',\n 'gt_depth', 'focal', 'gt_bboxes_3d', 'gt_labels_3d'\n ],\n meta_keys=[\n 'T_global', 'T_global_inv', 'timestamp', 'instance_id'\n ])\n ],\n test_mode=False,\n data_aug_conf=dict(\n resize_lim=(0.4, 0.47),\n final_dim=(256, 704),\n bot_pct_lim=(0.0, 0.0),\n rot_lim=(-5.4, 5.4),\n H=900,\n W=1600,\n rand_flip=True,\n rot3d_range=[-0.3925, 0.3925]),\n with_seq_flag=True,\n sequences_split_num=2,\n keep_consistent_seq_aug=True),\n val=dict(\n type='NuScenes3DDetTrackDataset',\n data_root='data/nuscenes/',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ],\n modality=dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False),\n version='v1.0-trainval',\n ann_file='data/nuscenes_anno_pkls/nuscenes_infos_val.pkl',\n pipeline=[\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(type='ResizeCropFlipImage'),\n dict(\n type='NormalizeMultiviewImage',\n mean=[123.675, 116.28, 103.53],\n std=[58.395, 57.12, 57.375],\n to_rgb=True),\n dict(type='NuScenesSparse4DAdaptor'),\n dict(\n type='Collect',\n keys=['img', 'timestamp', 'projection_mat', 'image_wh'],\n meta_keys=['T_global', 'T_global_inv', 'timestamp'])\n ],\n data_aug_conf=dict(\n resize_lim=(0.4, 0.47),\n final_dim=(256, 704),\n bot_pct_lim=(0.0, 0.0),\n rot_lim=(-5.4, 5.4),\n H=900,\n W=1600,\n rand_flip=True,\n rot3d_range=[-0.3925, 0.3925]),\n test_mode=True,\n tracking=True,\n tracking_threshold=0.2),\n test=dict(\n type='NuScenes3DDetTrackDataset',\n data_root='data/nuscenes/',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ],\n modality=dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False),\n version='v1.0-trainval',\n ann_file='data/nuscenes_anno_pkls/nuscenes_infos_val.pkl',\n pipeline=[\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(type='ResizeCropFlipImage'),\n dict(\n type='NormalizeMultiviewImage',\n mean=[123.675, 116.28, 103.53],\n std=[58.395, 57.12, 57.375],\n to_rgb=True),\n dict(type='NuScenesSparse4DAdaptor'),\n dict(\n type='Collect',\n keys=['img', 'timestamp', 'projection_mat', 'image_wh'],\n meta_keys=['T_global', 'T_global_inv', 'timestamp'])\n ],\n data_aug_conf=dict(\n resize_lim=(0.4, 0.47),\n final_dim=(256, 704),\n bot_pct_lim=(0.0, 0.0),\n rot_lim=(-5.4, 5.4),\n H=900,\n W=1600,\n rand_flip=True,\n rot3d_range=[-0.3925, 0.3925]),\n test_mode=True,\n tracking=True,\n tracking_threshold=0.2))\noptimizer = dict(\n type='AdamW',\n lr=0.0006,\n weight_decay=0.001,\n paramwise_cfg=dict(custom_keys=dict(img_backbone=dict(lr_mult=0.5))))\noptimizer_config = dict(grad_clip=dict(max_norm=25, norm_type=2))\nlr_config = dict(\n policy='CosineAnnealing',\n warmup='linear',\n warmup_iters=500,\n warmup_ratio=0.3333333333333333,\n min_lr_ratio=0.001)\nrunner = dict(type='IterBasedRunner', max_iters=17500)\nvis_pipeline = [\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(type='Collect', keys=['img'], meta_keys=['timestamp', 'lidar2img'])\n]\nevaluation = dict(\n interval=3500,\n pipeline=[\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(\n type='Collect', keys=['img'], meta_keys=['timestamp', 'lidar2img'])\n ])\ngpu_ids = range(0, 8)\n", "seed": 0, "exp_name": "sparse4dv3_temporal_r50_1x8_bs6_256x704.py"} -{"mode": "train", "epoch": 1, "iter": 1, "lr": 0.0001, "memory": 49167, "data_time": 9.15678, "loss_cls_0": 2.36126, "loss_box_0": 0.01384, "loss_cns_0": 0.0027, "loss_yns_0": 0.00079, "loss_cls_1": 2.15443, "loss_box_1": 0.10814, "loss_cns_1": 0.02455, "loss_yns_1": 0.00669, "loss_cls_2": 2.31196, "loss_box_2": 0.00504, "loss_cns_2": 0.00059, "loss_yns_2": 0.00029, "loss_cls_3": 2.3903, "loss_box_3": 0.02946, "loss_cns_3": 0.00504, "loss_yns_3": 0.00144, "loss_cls_4": 2.02813, "loss_box_4": 0.41638, "loss_cns_4": 0.05352, "loss_yns_4": 0.02542, "loss_cls_5": 2.42438, "loss_box_5": 0.01799, "loss_cns_5": 0.00217, "loss_yns_5": 0.00161, "loss_cls_dn_0": 1.19803, "loss_box_dn_0": 1.46028, "loss_cls_dn_1": 1.11018, "loss_box_dn_1": 1.73176, "loss_cls_dn_2": 1.17412, "loss_box_dn_2": 1.97185, "loss_cls_dn_3": 1.17207, "loss_box_dn_3": 2.24186, "loss_cls_dn_4": 1.05278, "loss_box_dn_4": 2.4269, "loss_cls_dn_5": 1.23869, "loss_box_dn_5": 2.67736, "loss_dense_depth": 1.86432, "loss": 35.7063, "grad_norm": 270.09869, "time": 115.67029} -{"mode": "train", "epoch": 1, "iter": 2, "lr": 0.0001, "memory": 49167, "data_time": 0.08767, "loss_cls_0": 2.04017, "loss_box_0": 0.0099, "loss_cns_0": 0.00282, "loss_yns_0": 0.00099, "loss_cls_1": 2.02281, "loss_box_1": 0.12352, "loss_cns_1": 0.02341, "loss_yns_1": 0.00634, "loss_cls_2": 2.10479, "loss_box_2": 0.22932, "loss_cns_2": 0.02082, "loss_yns_2": 0.00919, "loss_cls_3": 1.95071, "loss_box_3": 0.39584, "loss_cns_3": 0.05309, "loss_yns_3": 0.01917, "loss_cls_4": 1.79737, "loss_box_4": 1.56093, "loss_cns_4": 0.15548, "loss_yns_4": 0.05545, "loss_cls_5": 2.05673, "loss_box_5": 0.50885, "loss_cns_5": 0.05772, "loss_yns_5": 0.0178, "loss_cls_dn_0": 1.02427, "loss_box_dn_0": 1.25972, "loss_cls_dn_1": 0.95601, "loss_box_dn_1": 2.40878, "loss_cls_dn_2": 0.9705, "loss_box_dn_2": 2.52627, "loss_cls_dn_3": 0.91194, "loss_box_dn_3": 2.61151, "loss_cls_dn_4": 0.84158, "loss_box_dn_4": 2.87751, "loss_cls_dn_5": 0.98631, "loss_box_dn_5": 3.11696, "loss_dense_depth": 1.70552, "loss": 37.42012, "grad_norm": 67.06441, "time": 2.04416} -{"mode": "train", "epoch": 1, "iter": 3, "lr": 0.0001, "memory": 49167, "data_time": 0.18547, "loss_cls_0": 1.46873, "loss_box_0": 2.54025, "loss_cns_0": 0.61601, "loss_yns_0": 0.2146, "loss_cls_1": 1.77067, "loss_box_1": 1.81938, "loss_cns_1": 0.28812, "loss_yns_1": 0.10716, "loss_cls_2": 1.78815, "loss_box_2": 3.85493, "loss_cns_2": 0.35395, "loss_yns_2": 0.18959, "loss_cls_3": 1.62048, "loss_box_3": 4.89383, "loss_cns_3": 0.44575, "loss_yns_3": 0.21074, "loss_cls_4": 1.57476, "loss_box_4": 4.08754, "loss_cns_4": 0.38511, "loss_yns_4": 0.16477, "loss_cls_5": 1.68611, "loss_box_5": 2.89129, "loss_cns_5": 0.22458, "loss_yns_5": 0.09689, "loss_cls_dn_0": 0.71143, "loss_box_dn_0": 1.18017, "loss_cls_dn_1": 0.83382, "loss_box_dn_1": 2.40616, "loss_cls_dn_2": 0.8065, "loss_box_dn_2": 2.61828, "loss_cls_dn_3": 0.70916, "loss_box_dn_3": 2.82038, "loss_cls_dn_4": 0.72087, "loss_box_dn_4": 3.07632, "loss_cls_dn_5": 0.80513, "loss_box_dn_5": 3.31822, "loss_dense_depth": 1.59434, "loss": 54.89415, "grad_norm": 103.76462, "time": 1.65112} -{"mode": "train", "epoch": 1, "iter": 4, "lr": 0.0001, "memory": 49167, "data_time": 0.0812, "loss_cls_0": 1.38377, "loss_box_0": 2.53374, "loss_cns_0": 0.55469, "loss_yns_0": 0.18007, "loss_cls_1": 1.61305, "loss_box_1": 3.07313, "loss_cns_1": 0.46304, "loss_yns_1": 0.19711, "loss_cls_2": 1.70337, "loss_box_2": 3.59143, "loss_cns_2": 0.45775, "loss_yns_2": 0.18739, "loss_cls_3": 1.51546, "loss_box_3": 4.11987, "loss_cns_3": 0.46072, "loss_yns_3": 0.21414, "loss_cls_4": 1.47503, "loss_box_4": 4.68597, "loss_cns_4": 0.37902, "loss_yns_4": 0.19859, "loss_cls_5": 1.5025, "loss_box_5": 4.88227, "loss_cns_5": 0.44976, "loss_yns_5": 0.19829, "loss_cls_dn_0": 0.55909, "loss_box_dn_0": 1.17795, "loss_cls_dn_1": 0.72765, "loss_box_dn_1": 2.61393, "loss_cls_dn_2": 0.69769, "loss_box_dn_2": 2.67305, "loss_cls_dn_3": 0.61813, "loss_box_dn_3": 2.86825, "loss_cls_dn_4": 0.59814, "loss_box_dn_4": 3.07633, "loss_cls_dn_5": 0.67187, "loss_box_dn_5": 3.24938, "loss_dense_depth": 1.52273, "loss": 57.07434, "grad_norm": 117.07706, "time": 1.55286} -{"mode": "train", "epoch": 1, "iter": 5, "lr": 0.0001, "memory": 49167, "data_time": 0.11115, "loss_cls_0": 1.32244, "loss_box_0": 2.90033, "loss_cns_0": 0.48712, "loss_yns_0": 0.19001, "loss_cls_1": 1.56926, "loss_box_1": 3.95065, "loss_cns_1": 0.40754, "loss_yns_1": 0.21255, "loss_cls_2": 1.61288, "loss_box_2": 3.95234, "loss_cns_2": 0.39257, "loss_yns_2": 0.194, "loss_cls_3": 1.45966, "loss_box_3": 4.04037, "loss_cns_3": 0.39617, "loss_yns_3": 0.19717, "loss_cls_4": 1.39821, "loss_box_4": 4.218, "loss_cns_4": 0.37931, "loss_yns_4": 0.19667, "loss_cls_5": 1.39878, "loss_box_5": 4.45694, "loss_cns_5": 0.42071, "loss_yns_5": 0.1978, "loss_cls_dn_0": 0.52898, "loss_box_dn_0": 1.26038, "loss_cls_dn_1": 0.66062, "loss_box_dn_1": 2.26076, "loss_cls_dn_2": 0.6474, "loss_box_dn_2": 2.34857, "loss_cls_dn_3": 0.56125, "loss_box_dn_3": 2.46268, "loss_cls_dn_4": 0.56033, "loss_box_dn_4": 2.62845, "loss_cls_dn_5": 0.58675, "loss_box_dn_5": 2.72654, "loss_dense_depth": 1.398, "loss": 54.58221, "grad_norm": 110.93433, "time": 1.59936} -{"mode": "train", "epoch": 1, "iter": 6, "lr": 0.0001, "memory": 49167, "data_time": 0.08436, "loss_cls_0": 1.29231, "loss_box_0": 2.61595, "loss_cns_0": 0.56664, "loss_yns_0": 0.18048, "loss_cls_1": 1.48877, "loss_box_1": 3.95614, "loss_cns_1": 0.37852, "loss_yns_1": 0.19643, "loss_cls_2": 1.4741, "loss_box_2": 4.04516, "loss_cns_2": 0.3728, "loss_yns_2": 0.19411, "loss_cls_3": 1.35731, "loss_box_3": 4.00833, "loss_cns_3": 0.3527, "loss_yns_3": 0.19956, "loss_cls_4": 1.33185, "loss_box_4": 4.25421, "loss_cns_4": 0.31494, "loss_yns_4": 0.19789, "loss_cls_5": 1.33944, "loss_box_5": 4.37565, "loss_cns_5": 0.31052, "loss_yns_5": 0.20789, "loss_cls_dn_0": 0.52684, "loss_box_dn_0": 1.17846, "loss_cls_dn_1": 0.58992, "loss_box_dn_1": 2.40081, "loss_cls_dn_2": 0.58043, "loss_box_dn_2": 2.43505, "loss_cls_dn_3": 0.51804, "loss_box_dn_3": 2.48044, "loss_cls_dn_4": 0.48758, "loss_box_dn_4": 2.69959, "loss_cls_dn_5": 0.49555, "loss_box_dn_5": 2.76335, "loss_dense_depth": 1.4364, "loss": 53.60415, "grad_norm": 113.04875, "time": 1.63549} -{"mode": "train", "epoch": 1, "iter": 7, "lr": 0.0001, "memory": 49167, "data_time": 0.08596, "loss_cls_0": 1.24511, "loss_box_0": 2.38295, "loss_cns_0": 0.67085, "loss_yns_0": 0.1775, "loss_cls_1": 1.37394, "loss_box_1": 3.66888, "loss_cns_1": 0.45447, "loss_yns_1": 0.19342, "loss_cls_2": 1.39331, "loss_box_2": 3.66434, "loss_cns_2": 0.45288, "loss_yns_2": 0.186, "loss_cls_3": 1.30575, "loss_box_3": 3.4899, "loss_cns_3": 0.47987, "loss_yns_3": 0.18837, "loss_cls_4": 1.29507, "loss_box_4": 3.76305, "loss_cns_4": 0.44842, "loss_yns_4": 0.18498, "loss_cls_5": 1.33896, "loss_box_5": 3.96676, "loss_cns_5": 0.43728, "loss_yns_5": 0.19104, "loss_cls_dn_0": 0.5318, "loss_box_dn_0": 1.10136, "loss_cls_dn_1": 0.5245, "loss_box_dn_1": 2.4483, "loss_cls_dn_2": 0.52597, "loss_box_dn_2": 2.41845, "loss_cls_dn_3": 0.48047, "loss_box_dn_3": 2.42434, "loss_cls_dn_4": 0.44409, "loss_box_dn_4": 2.62393, "loss_cls_dn_5": 0.43498, "loss_box_dn_5": 2.70245, "loss_dense_depth": 1.42651, "loss": 51.04022, "grad_norm": 107.64119, "time": 1.57283} -{"mode": "train", "epoch": 1, "iter": 8, "lr": 0.0001, "memory": 49167, "data_time": 0.08624, "loss_cls_0": 1.23321, "loss_box_0": 2.27531, "loss_cns_0": 0.65684, "loss_yns_0": 0.1807, "loss_cls_1": 1.2919, "loss_box_1": 3.57193, "loss_cns_1": 0.51159, "loss_yns_1": 0.18528, "loss_cls_2": 1.3893, "loss_box_2": 3.56599, "loss_cns_2": 0.45532, "loss_yns_2": 0.19004, "loss_cls_3": 1.29765, "loss_box_3": 3.55574, "loss_cns_3": 0.48523, "loss_yns_3": 0.18937, "loss_cls_4": 1.30526, "loss_box_4": 3.58955, "loss_cns_4": 0.48927, "loss_yns_4": 0.1832, "loss_cls_5": 1.32829, "loss_box_5": 3.63275, "loss_cns_5": 0.52665, "loss_yns_5": 0.1965, "loss_cls_dn_0": 0.52048, "loss_box_dn_0": 1.02356, "loss_cls_dn_1": 0.52893, "loss_box_dn_1": 1.81913, "loss_cls_dn_2": 0.54323, "loss_box_dn_2": 1.7136, "loss_cls_dn_3": 0.49781, "loss_box_dn_3": 1.74334, "loss_cls_dn_4": 0.46345, "loss_box_dn_4": 1.84039, "loss_cls_dn_5": 0.43803, "loss_box_dn_5": 1.88355, "loss_dense_depth": 1.4037, "loss": 46.70608, "grad_norm": 81.73125, "time": 1.56396} -{"mode": "train", "epoch": 1, "iter": 9, "lr": 0.0001, "memory": 49167, "data_time": 0.08292, "loss_cls_0": 1.23998, "loss_box_0": 2.24214, "loss_cns_0": 0.61781, "loss_yns_0": 0.17664, "loss_cls_1": 1.25577, "loss_box_1": 3.27852, "loss_cns_1": 0.53125, "loss_yns_1": 0.17986, "loss_cls_2": 1.36715, "loss_box_2": 3.34312, "loss_cns_2": 0.46825, "loss_yns_2": 0.18034, "loss_cls_3": 1.28139, "loss_box_3": 3.47556, "loss_cns_3": 0.50988, "loss_yns_3": 0.20731, "loss_cls_4": 1.27257, "loss_box_4": 3.34903, "loss_cns_4": 0.51416, "loss_yns_4": 0.18173, "loss_cls_5": 1.31052, "loss_box_5": 3.3867, "loss_cns_5": 0.54082, "loss_yns_5": 0.1854, "loss_cls_dn_0": 0.49709, "loss_box_dn_0": 1.0132, "loss_cls_dn_1": 0.48807, "loss_box_dn_1": 1.48339, "loss_cls_dn_2": 0.52, "loss_box_dn_2": 1.4741, "loss_cls_dn_3": 0.47484, "loss_box_dn_3": 1.5814, "loss_cls_dn_4": 0.45471, "loss_box_dn_4": 1.52339, "loss_cls_dn_5": 0.43318, "loss_box_dn_5": 1.60645, "loss_dense_depth": 1.30126, "loss": 43.94699, "grad_norm": 66.93229, "time": 1.55423} -{"mode": "train", "epoch": 1, "iter": 10, "lr": 0.0001, "memory": 49167, "data_time": 0.08402, "loss_cls_0": 1.2206, "loss_box_0": 2.23775, "loss_cns_0": 0.61602, "loss_yns_0": 0.17293, "loss_cls_1": 1.26274, "loss_box_1": 3.08076, "loss_cns_1": 0.53, "loss_yns_1": 0.18006, "loss_cls_2": 1.29435, "loss_box_2": 3.13178, "loss_cns_2": 0.50701, "loss_yns_2": 0.18214, "loss_cls_3": 1.26024, "loss_box_3": 3.22797, "loss_cns_3": 0.55316, "loss_yns_3": 0.18643, "loss_cls_4": 1.24659, "loss_box_4": 3.20639, "loss_cns_4": 0.52886, "loss_yns_4": 0.18376, "loss_cls_5": 1.27513, "loss_box_5": 3.31197, "loss_cns_5": 0.52829, "loss_yns_5": 0.18323, "loss_cls_dn_0": 0.47663, "loss_box_dn_0": 1.03202, "loss_cls_dn_1": 0.44508, "loss_box_dn_1": 1.59967, "loss_cls_dn_2": 0.48269, "loss_box_dn_2": 1.65819, "loss_cls_dn_3": 0.43842, "loss_box_dn_3": 1.73018, "loss_cls_dn_4": 0.42751, "loss_box_dn_4": 1.72845, "loss_cls_dn_5": 0.42718, "loss_box_dn_5": 1.85896, "loss_dense_depth": 1.56028, "loss": 43.97341, "grad_norm": 67.59219, "time": 1.55888} -{"mode": "train", "epoch": 1, "iter": 11, "lr": 0.0001, "memory": 49167, "data_time": 0.08873, "loss_cls_0": 1.23714, "loss_box_0": 2.24083, "loss_cns_0": 0.62703, "loss_yns_0": 0.17659, "loss_cls_1": 1.27135, "loss_box_1": 2.99506, "loss_cns_1": 0.50615, "loss_yns_1": 0.18318, "loss_cls_2": 1.27428, "loss_box_2": 2.98955, "loss_cns_2": 0.51419, "loss_yns_2": 0.18741, "loss_cls_3": 1.26621, "loss_box_3": 2.98208, "loss_cns_3": 0.56529, "loss_yns_3": 0.18033, "loss_cls_4": 1.2533, "loss_box_4": 3.02995, "loss_cns_4": 0.53372, "loss_yns_4": 0.1827, "loss_cls_5": 1.27717, "loss_box_5": 3.26709, "loss_cns_5": 0.54637, "loss_yns_5": 0.18004, "loss_cls_dn_0": 0.4658, "loss_box_dn_0": 1.04469, "loss_cls_dn_1": 0.41525, "loss_box_dn_1": 1.88247, "loss_cls_dn_2": 0.45298, "loss_box_dn_2": 1.92423, "loss_cls_dn_3": 0.4099, "loss_box_dn_3": 1.93038, "loss_cls_dn_4": 0.39673, "loss_box_dn_4": 1.96705, "loss_cls_dn_5": 0.41228, "loss_box_dn_5": 2.12296, "loss_dense_depth": 1.44622, "loss": 44.33796, "grad_norm": 70.21803, "time": 1.5753} -{"mode": "train", "epoch": 1, "iter": 12, "lr": 0.0001, "memory": 49167, "data_time": 0.0894, "loss_cls_0": 1.22923, "loss_box_0": 2.23006, "loss_cns_0": 0.63332, "loss_yns_0": 0.17381, "loss_cls_1": 1.24594, "loss_box_1": 2.9448, "loss_cns_1": 0.49041, "loss_yns_1": 0.17552, "loss_cls_2": 1.27351, "loss_box_2": 2.93429, "loss_cns_2": 0.49837, "loss_yns_2": 0.18326, "loss_cls_3": 1.24496, "loss_box_3": 2.90054, "loss_cns_3": 0.53403, "loss_yns_3": 0.17659, "loss_cls_4": 1.24284, "loss_box_4": 2.93176, "loss_cns_4": 0.53089, "loss_yns_4": 0.18179, "loss_cls_5": 1.28053, "loss_box_5": 3.07008, "loss_cns_5": 0.53837, "loss_yns_5": 0.18356, "loss_cls_dn_0": 0.46347, "loss_box_dn_0": 1.04993, "loss_cls_dn_1": 0.39382, "loss_box_dn_1": 2.18033, "loss_cls_dn_2": 0.43441, "loss_box_dn_2": 2.17528, "loss_cls_dn_3": 0.3959, "loss_box_dn_3": 2.1438, "loss_cls_dn_4": 0.38159, "loss_box_dn_4": 2.18942, "loss_cls_dn_5": 0.39339, "loss_box_dn_5": 2.25277, "loss_dense_depth": 1.51675, "loss": 44.7993, "grad_norm": 55.70547, "time": 1.56506} -{"mode": "train", "epoch": 1, "iter": 13, "lr": 0.0001, "memory": 49167, "data_time": 0.08845, "loss_cls_0": 1.1975, "loss_box_0": 2.25564, "loss_cns_0": 0.61115, "loss_yns_0": 0.17054, "loss_cls_1": 1.21937, "loss_box_1": 2.8349, "loss_cns_1": 0.51912, "loss_yns_1": 0.17576, "loss_cls_2": 1.26197, "loss_box_2": 2.88352, "loss_cns_2": 0.53163, "loss_yns_2": 0.17944, "loss_cls_3": 1.22487, "loss_box_3": 2.87609, "loss_cns_3": 0.53459, "loss_yns_3": 0.17495, "loss_cls_4": 1.26266, "loss_box_4": 2.90503, "loss_cns_4": 0.51935, "loss_yns_4": 0.18055, "loss_cls_5": 1.2791, "loss_box_5": 2.89663, "loss_cns_5": 0.53242, "loss_yns_5": 0.1805, "loss_cls_dn_0": 0.46326, "loss_box_dn_0": 1.0524, "loss_cls_dn_1": 0.42688, "loss_box_dn_1": 1.65211, "loss_cls_dn_2": 0.45982, "loss_box_dn_2": 1.66552, "loss_cls_dn_3": 0.44841, "loss_box_dn_3": 1.69497, "loss_cls_dn_4": 0.43297, "loss_box_dn_4": 1.81789, "loss_cls_dn_5": 0.444, "loss_box_dn_5": 1.80006, "loss_dense_depth": 1.38764, "loss": 42.15321, "grad_norm": 71.37956, "time": 1.56782} -{"mode": "train", "epoch": 1, "iter": 14, "lr": 0.00011, "memory": 49167, "data_time": 0.08789, "loss_cls_0": 1.20543, "loss_box_0": 2.28368, "loss_cns_0": 0.59191, "loss_yns_0": 0.17138, "loss_cls_1": 1.28568, "loss_box_1": 2.73262, "loss_cns_1": 0.54202, "loss_yns_1": 0.18333, "loss_cls_2": 1.29576, "loss_box_2": 2.75968, "loss_cns_2": 0.55231, "loss_yns_2": 0.17938, "loss_cls_3": 1.26111, "loss_box_3": 2.80651, "loss_cns_3": 0.55589, "loss_yns_3": 0.17442, "loss_cls_4": 1.28678, "loss_box_4": 2.82654, "loss_cns_4": 0.54213, "loss_yns_4": 0.17575, "loss_cls_5": 1.28727, "loss_box_5": 2.89682, "loss_cns_5": 0.54783, "loss_yns_5": 0.1751, "loss_cls_dn_0": 0.46751, "loss_box_dn_0": 1.03647, "loss_cls_dn_1": 0.44194, "loss_box_dn_1": 1.39152, "loss_cls_dn_2": 0.46657, "loss_box_dn_2": 1.39135, "loss_cls_dn_3": 0.46718, "loss_box_dn_3": 1.49086, "loss_cls_dn_4": 0.45496, "loss_box_dn_4": 1.62177, "loss_cls_dn_5": 0.4695, "loss_box_dn_5": 1.62848, "loss_dense_depth": 1.34112, "loss": 40.98857, "grad_norm": 79.50517, "time": 1.5655} -{"mode": "train", "epoch": 1, "iter": 15, "lr": 0.00011, "memory": 49167, "data_time": 0.08451, "loss_cls_0": 1.20129, "loss_box_0": 2.34963, "loss_cns_0": 0.57826, "loss_yns_0": 0.17132, "loss_cls_1": 1.29918, "loss_box_1": 2.71436, "loss_cns_1": 0.5618, "loss_yns_1": 0.18163, "loss_cls_2": 1.29494, "loss_box_2": 2.68241, "loss_cns_2": 0.57178, "loss_yns_2": 0.17869, "loss_cls_3": 1.29391, "loss_box_3": 2.75485, "loss_cns_3": 0.57275, "loss_yns_3": 0.17519, "loss_cls_4": 1.27601, "loss_box_4": 2.80194, "loss_cns_4": 0.57083, "loss_yns_4": 0.17517, "loss_cls_5": 1.27731, "loss_box_5": 2.83606, "loss_cns_5": 0.57696, "loss_yns_5": 0.17368, "loss_cls_dn_0": 0.46988, "loss_box_dn_0": 1.03233, "loss_cls_dn_1": 0.44469, "loss_box_dn_1": 1.47358, "loss_cls_dn_2": 0.45706, "loss_box_dn_2": 1.42041, "loss_cls_dn_3": 0.452, "loss_box_dn_3": 1.50899, "loss_cls_dn_4": 0.45117, "loss_box_dn_4": 1.59142, "loss_cls_dn_5": 0.47254, "loss_box_dn_5": 1.59624, "loss_dense_depth": 1.28463, "loss": 40.9249, "grad_norm": 78.46203, "time": 1.5644} -{"mode": "train", "epoch": 1, "iter": 16, "lr": 0.00011, "memory": 49167, "data_time": 0.08752, "loss_cls_0": 1.18189, "loss_box_0": 2.34234, "loss_cns_0": 0.57063, "loss_yns_0": 0.17311, "loss_cls_1": 1.2798, "loss_box_1": 2.73724, "loss_cns_1": 0.56608, "loss_yns_1": 0.17785, "loss_cls_2": 1.26947, "loss_box_2": 2.78191, "loss_cns_2": 0.56367, "loss_yns_2": 0.17927, "loss_cls_3": 1.31427, "loss_box_3": 2.81854, "loss_cns_3": 0.58116, "loss_yns_3": 0.17849, "loss_cls_4": 1.25931, "loss_box_4": 2.81092, "loss_cns_4": 0.56317, "loss_yns_4": 0.17733, "loss_cls_5": 1.26485, "loss_box_5": 2.7962, "loss_cns_5": 0.59409, "loss_yns_5": 0.18193, "loss_cls_dn_0": 0.48917, "loss_box_dn_0": 1.01025, "loss_cls_dn_1": 0.46784, "loss_box_dn_1": 1.43028, "loss_cls_dn_2": 0.48543, "loss_box_dn_2": 1.3996, "loss_cls_dn_3": 0.44204, "loss_box_dn_3": 1.47284, "loss_cls_dn_4": 0.45851, "loss_box_dn_4": 1.50055, "loss_cls_dn_5": 0.48441, "loss_box_dn_5": 1.53694, "loss_dense_depth": 1.35996, "loss": 40.90134, "grad_norm": 68.99036, "time": 1.56975} -{"mode": "train", "epoch": 1, "iter": 17, "lr": 0.00011, "memory": 49167, "data_time": 0.08762, "loss_cls_0": 1.1737, "loss_box_0": 2.33911, "loss_cns_0": 0.57421, "loss_yns_0": 0.17087, "loss_cls_1": 1.26107, "loss_box_1": 2.94426, "loss_cns_1": 0.53649, "loss_yns_1": 0.17459, "loss_cls_2": 1.27223, "loss_box_2": 2.97964, "loss_cns_2": 0.53864, "loss_yns_2": 0.17619, "loss_cls_3": 1.29616, "loss_box_3": 2.94892, "loss_cns_3": 0.56388, "loss_yns_3": 0.17523, "loss_cls_4": 1.25081, "loss_box_4": 2.91614, "loss_cns_4": 0.55278, "loss_yns_4": 0.17711, "loss_cls_5": 1.26114, "loss_box_5": 2.92006, "loss_cns_5": 0.5664, "loss_yns_5": 0.17989, "loss_cls_dn_0": 0.48359, "loss_box_dn_0": 1.0191, "loss_cls_dn_1": 0.46415, "loss_box_dn_1": 1.60601, "loss_cls_dn_2": 0.47927, "loss_box_dn_2": 1.59315, "loss_cls_dn_3": 0.42673, "loss_box_dn_3": 1.62666, "loss_cls_dn_4": 0.43983, "loss_box_dn_4": 1.63596, "loss_cls_dn_5": 0.4679, "loss_box_dn_5": 1.70769, "loss_dense_depth": 1.21575, "loss": 42.11535, "grad_norm": 61.71445, "time": 1.55869} -{"mode": "train", "epoch": 1, "iter": 18, "lr": 0.00011, "memory": 49167, "data_time": 0.08568, "loss_cls_0": 1.18483, "loss_box_0": 2.2552, "loss_cns_0": 0.59568, "loss_yns_0": 0.17058, "loss_cls_1": 1.24268, "loss_box_1": 2.93561, "loss_cns_1": 0.52612, "loss_yns_1": 0.18188, "loss_cls_2": 1.25572, "loss_box_2": 2.9427, "loss_cns_2": 0.52987, "loss_yns_2": 0.17617, "loss_cls_3": 1.26171, "loss_box_3": 2.9017, "loss_cns_3": 0.55251, "loss_yns_3": 0.17735, "loss_cls_4": 1.24293, "loss_box_4": 2.96328, "loss_cns_4": 0.56286, "loss_yns_4": 0.17638, "loss_cls_5": 1.26113, "loss_box_5": 3.01026, "loss_cns_5": 0.54768, "loss_yns_5": 0.1711, "loss_cls_dn_0": 0.47159, "loss_box_dn_0": 1.01002, "loss_cls_dn_1": 0.45688, "loss_box_dn_1": 1.58671, "loss_cls_dn_2": 0.4658, "loss_box_dn_2": 1.61032, "loss_cls_dn_3": 0.42544, "loss_box_dn_3": 1.62679, "loss_cls_dn_4": 0.42966, "loss_box_dn_4": 1.67134, "loss_cls_dn_5": 0.45668, "loss_box_dn_5": 1.79866, "loss_dense_depth": 1.22042, "loss": 42.05628, "grad_norm": 53.84068, "time": 1.55411} -{"mode": "train", "epoch": 1, "iter": 19, "lr": 0.00011, "memory": 49167, "data_time": 0.08314, "loss_cls_0": 1.18374, "loss_box_0": 2.23423, "loss_cns_0": 0.5948, "loss_yns_0": 0.17135, "loss_cls_1": 1.23436, "loss_box_1": 2.96529, "loss_cns_1": 0.53192, "loss_yns_1": 0.17665, "loss_cls_2": 1.25487, "loss_box_2": 2.96757, "loss_cns_2": 0.53211, "loss_yns_2": 0.1761, "loss_cls_3": 1.25289, "loss_box_3": 2.93415, "loss_cns_3": 0.53815, "loss_yns_3": 0.17529, "loss_cls_4": 1.24967, "loss_box_4": 3.00254, "loss_cns_4": 0.55259, "loss_yns_4": 0.17436, "loss_cls_5": 1.26343, "loss_box_5": 3.04337, "loss_cns_5": 0.5409, "loss_yns_5": 0.17256, "loss_cls_dn_0": 0.47539, "loss_box_dn_0": 1.01825, "loss_cls_dn_1": 0.46522, "loss_box_dn_1": 1.24485, "loss_cls_dn_2": 0.45675, "loss_box_dn_2": 1.34325, "loss_cls_dn_3": 0.44066, "loss_box_dn_3": 1.38581, "loss_cls_dn_4": 0.43188, "loss_box_dn_4": 1.44555, "loss_cls_dn_5": 0.4599, "loss_box_dn_5": 1.5607, "loss_dense_depth": 1.19806, "loss": 40.84916, "grad_norm": 83.42062, "time": 1.56031} -{"mode": "train", "epoch": 1, "iter": 20, "lr": 0.00011, "memory": 49167, "data_time": 0.08046, "loss_cls_0": 1.17729, "loss_box_0": 2.23076, "loss_cns_0": 0.59575, "loss_yns_0": 0.16992, "loss_cls_1": 1.23527, "loss_box_1": 2.91285, "loss_cns_1": 0.54399, "loss_yns_1": 0.17548, "loss_cls_2": 1.26224, "loss_box_2": 2.87683, "loss_cns_2": 0.54841, "loss_yns_2": 0.1784, "loss_cls_3": 1.24128, "loss_box_3": 2.83868, "loss_cns_3": 0.5466, "loss_yns_3": 0.17255, "loss_cls_4": 1.25773, "loss_box_4": 2.87085, "loss_cns_4": 0.5535, "loss_yns_4": 0.17266, "loss_cls_5": 1.2688, "loss_box_5": 2.94101, "loss_cns_5": 0.5497, "loss_yns_5": 0.18111, "loss_cls_dn_0": 0.46982, "loss_box_dn_0": 1.0081, "loss_cls_dn_1": 0.42086, "loss_box_dn_1": 1.42252, "loss_cls_dn_2": 0.41699, "loss_box_dn_2": 1.46837, "loss_cls_dn_3": 0.41262, "loss_box_dn_3": 1.46674, "loss_cls_dn_4": 0.39677, "loss_box_dn_4": 1.47202, "loss_cls_dn_5": 0.41569, "loss_box_dn_5": 1.53362, "loss_dense_depth": 1.19913, "loss": 40.60488, "grad_norm": 54.22835, "time": 1.55991} -{"mode": "train", "epoch": 1, "iter": 21, "lr": 0.00011, "memory": 49167, "data_time": 0.09441, "loss_cls_0": 1.15614, "loss_box_0": 2.19479, "loss_cns_0": 0.60185, "loss_yns_0": 0.16937, "loss_cls_1": 1.25814, "loss_box_1": 2.83275, "loss_cns_1": 0.58168, "loss_yns_1": 0.17271, "loss_cls_2": 1.27036, "loss_box_2": 2.78707, "loss_cns_2": 0.5823, "loss_yns_2": 0.17747, "loss_cls_3": 1.25872, "loss_box_3": 2.79396, "loss_cns_3": 0.57534, "loss_yns_3": 0.17375, "loss_cls_4": 1.25817, "loss_box_4": 2.8504, "loss_cns_4": 0.57437, "loss_yns_4": 0.17297, "loss_cls_5": 1.28258, "loss_box_5": 2.89552, "loss_cns_5": 0.56471, "loss_yns_5": 0.18139, "loss_cls_dn_0": 0.47112, "loss_box_dn_0": 1.02385, "loss_cls_dn_1": 0.42934, "loss_box_dn_1": 1.21058, "loss_cls_dn_2": 0.42361, "loss_box_dn_2": 1.2289, "loss_cls_dn_3": 0.42757, "loss_box_dn_3": 1.23311, "loss_cls_dn_4": 0.41996, "loss_box_dn_4": 1.26517, "loss_cls_dn_5": 0.42935, "loss_box_dn_5": 1.2815, "loss_dense_depth": 1.09291, "loss": 39.30351, "grad_norm": 49.21168, "time": 1.67816} -{"mode": "train", "epoch": 1, "iter": 22, "lr": 0.00011, "memory": 49167, "data_time": 0.08258, "loss_cls_0": 1.17431, "loss_box_0": 2.16297, "loss_cns_0": 0.60811, "loss_yns_0": 0.16801, "loss_cls_1": 1.27527, "loss_box_1": 2.80518, "loss_cns_1": 0.57541, "loss_yns_1": 0.17715, "loss_cls_2": 1.27199, "loss_box_2": 2.77562, "loss_cns_2": 0.57862, "loss_yns_2": 0.17761, "loss_cls_3": 1.25993, "loss_box_3": 2.7547, "loss_cns_3": 0.59997, "loss_yns_3": 0.17956, "loss_cls_4": 1.24998, "loss_box_4": 2.82068, "loss_cns_4": 0.61059, "loss_yns_4": 0.17549, "loss_cls_5": 1.27159, "loss_box_5": 2.80572, "loss_cns_5": 0.61363, "loss_yns_5": 0.18888, "loss_cls_dn_0": 0.46497, "loss_box_dn_0": 1.0104, "loss_cls_dn_1": 0.43611, "loss_box_dn_1": 1.1003, "loss_cls_dn_2": 0.43469, "loss_box_dn_2": 1.13601, "loss_cls_dn_3": 0.44349, "loss_box_dn_3": 1.18084, "loss_cls_dn_4": 0.4486, "loss_box_dn_4": 1.29191, "loss_cls_dn_5": 0.44382, "loss_box_dn_5": 1.28733, "loss_dense_depth": 1.1253, "loss": 39.08477, "grad_norm": 66.91582, "time": 1.64645} -{"mode": "train", "epoch": 1, "iter": 23, "lr": 0.00011, "memory": 49167, "data_time": 0.13563, "loss_cls_0": 1.15125, "loss_box_0": 2.11982, "loss_cns_0": 0.61465, "loss_yns_0": 0.16866, "loss_cls_1": 1.24091, "loss_box_1": 2.82188, "loss_cns_1": 0.5575, "loss_yns_1": 0.17428, "loss_cls_2": 1.26222, "loss_box_2": 2.73169, "loss_cns_2": 0.57744, "loss_yns_2": 0.18087, "loss_cls_3": 1.24193, "loss_box_3": 2.74268, "loss_cns_3": 0.60135, "loss_yns_3": 0.1749, "loss_cls_4": 1.22997, "loss_box_4": 2.77403, "loss_cns_4": 0.60622, "loss_yns_4": 0.17335, "loss_cls_5": 1.2549, "loss_box_5": 2.80328, "loss_cns_5": 0.61916, "loss_yns_5": 0.17712, "loss_cls_dn_0": 0.46618, "loss_box_dn_0": 0.97591, "loss_cls_dn_1": 0.43946, "loss_box_dn_1": 1.1561, "loss_cls_dn_2": 0.44707, "loss_box_dn_2": 1.16502, "loss_cls_dn_3": 0.45597, "loss_box_dn_3": 1.25462, "loss_cls_dn_4": 0.45481, "loss_box_dn_4": 1.41084, "loss_cls_dn_5": 0.4503, "loss_box_dn_5": 1.42937, "loss_dense_depth": 1.15859, "loss": 39.26433, "grad_norm": 64.66269, "time": 1.59819} -{"mode": "train", "epoch": 1, "iter": 24, "lr": 0.00011, "memory": 49167, "data_time": 0.08068, "loss_cls_0": 1.16287, "loss_box_0": 2.11677, "loss_cns_0": 0.61895, "loss_yns_0": 0.16973, "loss_cls_1": 1.24769, "loss_box_1": 2.76891, "loss_cns_1": 0.55976, "loss_yns_1": 0.17601, "loss_cls_2": 1.26679, "loss_box_2": 2.71313, "loss_cns_2": 0.57817, "loss_yns_2": 0.17663, "loss_cls_3": 1.24631, "loss_box_3": 2.71378, "loss_cns_3": 0.59255, "loss_yns_3": 0.17042, "loss_cls_4": 1.23407, "loss_box_4": 2.71201, "loss_cns_4": 0.58848, "loss_yns_4": 0.17061, "loss_cls_5": 1.25551, "loss_box_5": 2.77464, "loss_cns_5": 0.60396, "loss_yns_5": 0.1734, "loss_cls_dn_0": 0.47486, "loss_box_dn_0": 0.98393, "loss_cls_dn_1": 0.42821, "loss_box_dn_1": 1.33215, "loss_cls_dn_2": 0.44057, "loss_box_dn_2": 1.35438, "loss_cls_dn_3": 0.44667, "loss_box_dn_3": 1.44237, "loss_cls_dn_4": 0.43094, "loss_box_dn_4": 1.57483, "loss_cls_dn_5": 0.44057, "loss_box_dn_5": 1.60195, "loss_dense_depth": 1.19458, "loss": 39.93718, "grad_norm": 53.92472, "time": 1.58594} -{"mode": "train", "epoch": 1, "iter": 25, "lr": 0.00011, "memory": 49167, "data_time": 0.10866, "loss_cls_0": 1.17004, "loss_box_0": 2.10893, "loss_cns_0": 0.62226, "loss_yns_0": 0.16893, "loss_cls_1": 1.22841, "loss_box_1": 2.76907, "loss_cns_1": 0.55821, "loss_yns_1": 0.17094, "loss_cls_2": 1.23391, "loss_box_2": 2.74088, "loss_cns_2": 0.58113, "loss_yns_2": 0.17115, "loss_cls_3": 1.23476, "loss_box_3": 2.69684, "loss_cns_3": 0.57941, "loss_yns_3": 0.17083, "loss_cls_4": 1.24537, "loss_box_4": 2.72443, "loss_cns_4": 0.57289, "loss_yns_4": 0.16964, "loss_cls_5": 1.23935, "loss_box_5": 2.759, "loss_cns_5": 0.56913, "loss_yns_5": 0.17711, "loss_cls_dn_0": 0.46412, "loss_box_dn_0": 0.97259, "loss_cls_dn_1": 0.41613, "loss_box_dn_1": 1.42255, "loss_cls_dn_2": 0.42867, "loss_box_dn_2": 1.43936, "loss_cls_dn_3": 0.42886, "loss_box_dn_3": 1.46866, "loss_cls_dn_4": 0.40102, "loss_box_dn_4": 1.55603, "loss_cls_dn_5": 0.43069, "loss_box_dn_5": 1.56028, "loss_dense_depth": 1.10483, "loss": 39.75642, "grad_norm": 59.27147, "time": 1.5862} -{"mode": "train", "epoch": 1, "iter": 26, "lr": 0.00011, "memory": 49167, "data_time": 0.07863, "loss_cls_0": 1.15574, "loss_box_0": 2.07935, "loss_cns_0": 0.62523, "loss_yns_0": 0.16992, "loss_cls_1": 1.21791, "loss_box_1": 2.79883, "loss_cns_1": 0.55932, "loss_yns_1": 0.16896, "loss_cls_2": 1.23166, "loss_box_2": 2.77585, "loss_cns_2": 0.58449, "loss_yns_2": 0.17048, "loss_cls_3": 1.24272, "loss_box_3": 2.73979, "loss_cns_3": 0.56889, "loss_yns_3": 0.171, "loss_cls_4": 1.28071, "loss_box_4": 2.75418, "loss_cns_4": 0.56587, "loss_yns_4": 0.17145, "loss_cls_5": 1.2453, "loss_box_5": 2.72238, "loss_cns_5": 0.57196, "loss_yns_5": 0.17474, "loss_cls_dn_0": 0.44457, "loss_box_dn_0": 0.97981, "loss_cls_dn_1": 0.43532, "loss_box_dn_1": 1.22595, "loss_cls_dn_2": 0.43874, "loss_box_dn_2": 1.2211, "loss_cls_dn_3": 0.43702, "loss_box_dn_3": 1.22759, "loss_cls_dn_4": 0.40831, "loss_box_dn_4": 1.27455, "loss_cls_dn_5": 0.44709, "loss_box_dn_5": 1.26053, "loss_dense_depth": 1.1188, "loss": 38.66608, "grad_norm": 69.65617, "time": 1.57768} -{"mode": "train", "epoch": 1, "iter": 27, "lr": 0.00011, "memory": 49167, "data_time": 0.07856, "loss_cls_0": 1.14259, "loss_box_0": 2.10063, "loss_cns_0": 0.62061, "loss_yns_0": 0.17224, "loss_cls_1": 1.21783, "loss_box_1": 2.75669, "loss_cns_1": 0.56194, "loss_yns_1": 0.17379, "loss_cls_2": 1.24644, "loss_box_2": 2.70243, "loss_cns_2": 0.58091, "loss_yns_2": 0.17483, "loss_cls_3": 1.23051, "loss_box_3": 2.66065, "loss_cns_3": 0.57797, "loss_yns_3": 0.17129, "loss_cls_4": 1.24848, "loss_box_4": 2.65676, "loss_cns_4": 0.58519, "loss_yns_4": 0.1709, "loss_cls_5": 1.24932, "loss_box_5": 2.67951, "loss_cns_5": 0.58286, "loss_yns_5": 0.17223, "loss_cls_dn_0": 0.44233, "loss_box_dn_0": 0.9758, "loss_cls_dn_1": 0.42177, "loss_box_dn_1": 1.19519, "loss_cls_dn_2": 0.43002, "loss_box_dn_2": 1.15113, "loss_cls_dn_3": 0.42644, "loss_box_dn_3": 1.15085, "loss_cls_dn_4": 0.41041, "loss_box_dn_4": 1.16235, "loss_cls_dn_5": 0.43447, "loss_box_dn_5": 1.20984, "loss_dense_depth": 1.09704, "loss": 37.94426, "grad_norm": 49.77213, "time": 1.5747} -{"mode": "train", "epoch": 1, "iter": 28, "lr": 0.00011, "memory": 49167, "data_time": 0.08594, "loss_cls_0": 1.11911, "loss_box_0": 2.1261, "loss_cns_0": 0.6182, "loss_yns_0": 0.17056, "loss_cls_1": 1.21233, "loss_box_1": 2.52381, "loss_cns_1": 0.58172, "loss_yns_1": 0.1746, "loss_cls_2": 1.20899, "loss_box_2": 2.50812, "loss_cns_2": 0.59117, "loss_yns_2": 0.17604, "loss_cls_3": 1.21261, "loss_box_3": 2.4887, "loss_cns_3": 0.58932, "loss_yns_3": 0.1689, "loss_cls_4": 1.21858, "loss_box_4": 2.51889, "loss_cns_4": 0.58393, "loss_yns_4": 0.17157, "loss_cls_5": 1.22982, "loss_box_5": 2.59911, "loss_cns_5": 0.57759, "loss_yns_5": 0.17459, "loss_cls_dn_0": 0.44222, "loss_box_dn_0": 0.97533, "loss_cls_dn_1": 0.4105, "loss_box_dn_1": 1.17945, "loss_cls_dn_2": 0.42736, "loss_box_dn_2": 1.16102, "loss_cls_dn_3": 0.41505, "loss_box_dn_3": 1.19749, "loss_cls_dn_4": 0.41939, "loss_box_dn_4": 1.22245, "loss_cls_dn_5": 0.41807, "loss_box_dn_5": 1.33786, "loss_dense_depth": 1.05409, "loss": 37.20468, "grad_norm": 63.11984, "time": 1.55484} -{"mode": "train", "epoch": 1, "iter": 29, "lr": 0.00011, "memory": 49167, "data_time": 0.0888, "loss_cls_0": 1.09413, "loss_box_0": 2.1184, "loss_cns_0": 0.61903, "loss_yns_0": 0.17129, "loss_cls_1": 1.19763, "loss_box_1": 2.5242, "loss_cns_1": 0.58581, "loss_yns_1": 0.17287, "loss_cls_2": 1.18887, "loss_box_2": 2.50141, "loss_cns_2": 0.6025, "loss_yns_2": 0.17422, "loss_cls_3": 1.20528, "loss_box_3": 2.4858, "loss_cns_3": 0.59826, "loss_yns_3": 0.17088, "loss_cls_4": 1.22235, "loss_box_4": 2.50618, "loss_cns_4": 0.59595, "loss_yns_4": 0.17077, "loss_cls_5": 1.20629, "loss_box_5": 2.52657, "loss_cns_5": 0.59721, "loss_yns_5": 0.17244, "loss_cls_dn_0": 0.44219, "loss_box_dn_0": 0.9835, "loss_cls_dn_1": 0.41141, "loss_box_dn_1": 1.1577, "loss_cls_dn_2": 0.43782, "loss_box_dn_2": 1.16588, "loss_cls_dn_3": 0.4141, "loss_box_dn_3": 1.21819, "loss_cls_dn_4": 0.42663, "loss_box_dn_4": 1.2427, "loss_cls_dn_5": 0.42208, "loss_box_dn_5": 1.35769, "loss_dense_depth": 1.09688, "loss": 37.18514, "grad_norm": 60.57161, "time": 1.57877} -{"mode": "train", "epoch": 1, "iter": 30, "lr": 0.00011, "memory": 49167, "data_time": 0.06971, "loss_cls_0": 1.08699, "loss_box_0": 2.07331, "loss_cns_0": 0.61976, "loss_yns_0": 0.1713, "loss_cls_1": 1.16069, "loss_box_1": 2.57966, "loss_cns_1": 0.58889, "loss_yns_1": 0.17065, "loss_cls_2": 1.1686, "loss_box_2": 2.5384, "loss_cns_2": 0.60346, "loss_yns_2": 0.175, "loss_cls_3": 1.19364, "loss_box_3": 2.52379, "loss_cns_3": 0.59842, "loss_yns_3": 0.171, "loss_cls_4": 1.19838, "loss_box_4": 2.54095, "loss_cns_4": 0.60182, "loss_yns_4": 0.17047, "loss_cls_5": 1.19351, "loss_box_5": 2.56199, "loss_cns_5": 0.60695, "loss_yns_5": 0.17532, "loss_cls_dn_0": 0.42875, "loss_box_dn_0": 0.9859, "loss_cls_dn_1": 0.40012, "loss_box_dn_1": 1.25106, "loss_cls_dn_2": 0.43104, "loss_box_dn_2": 1.26255, "loss_cls_dn_3": 0.39283, "loss_box_dn_3": 1.29198, "loss_cls_dn_4": 0.40857, "loss_box_dn_4": 1.30822, "loss_cls_dn_5": 0.40798, "loss_box_dn_5": 1.41081, "loss_dense_depth": 1.0629, "loss": 37.51566, "grad_norm": 64.57935, "time": 1.57215} -{"mode": "train", "epoch": 1, "iter": 31, "lr": 0.00011, "memory": 49167, "data_time": 0.07589, "loss_cls_0": 1.09275, "loss_box_0": 2.07282, "loss_cns_0": 0.62091, "loss_yns_0": 0.17223, "loss_cls_1": 1.14119, "loss_box_1": 2.59129, "loss_cns_1": 0.57655, "loss_yns_1": 0.17177, "loss_cls_2": 1.15787, "loss_box_2": 2.51354, "loss_cns_2": 0.59558, "loss_yns_2": 0.17413, "loss_cls_3": 1.18666, "loss_box_3": 2.46735, "loss_cns_3": 0.59739, "loss_yns_3": 0.17135, "loss_cls_4": 1.16184, "loss_box_4": 2.49901, "loss_cns_4": 0.60312, "loss_yns_4": 0.16937, "loss_cls_5": 1.17609, "loss_box_5": 2.533, "loss_cns_5": 0.60428, "loss_yns_5": 0.1716, "loss_cls_dn_0": 0.42055, "loss_box_dn_0": 0.97585, "loss_cls_dn_1": 0.39145, "loss_box_dn_1": 1.27732, "loss_cls_dn_2": 0.42075, "loss_box_dn_2": 1.26316, "loss_cls_dn_3": 0.37663, "loss_box_dn_3": 1.25439, "loss_cls_dn_4": 0.38506, "loss_box_dn_4": 1.25951, "loss_cls_dn_5": 0.39571, "loss_box_dn_5": 1.33837, "loss_dense_depth": 1.02914, "loss": 37.00957, "grad_norm": 59.74335, "time": 1.56024} -{"mode": "train", "epoch": 1, "iter": 32, "lr": 0.00011, "memory": 49167, "data_time": 0.07836, "loss_cls_0": 1.06109, "loss_box_0": 2.06068, "loss_cns_0": 0.62082, "loss_yns_0": 0.17099, "loss_cls_1": 1.12479, "loss_box_1": 2.50113, "loss_cns_1": 0.58023, "loss_yns_1": 0.17587, "loss_cls_2": 1.13488, "loss_box_2": 2.39841, "loss_cns_2": 0.60286, "loss_yns_2": 0.17392, "loss_cls_3": 1.17408, "loss_box_3": 2.39494, "loss_cns_3": 0.60812, "loss_yns_3": 0.17054, "loss_cls_4": 1.17435, "loss_box_4": 2.40308, "loss_cns_4": 0.6179, "loss_yns_4": 0.16884, "loss_cls_5": 1.16049, "loss_box_5": 2.413, "loss_cns_5": 0.60834, "loss_yns_5": 0.16952, "loss_cls_dn_0": 0.41239, "loss_box_dn_0": 0.96666, "loss_cls_dn_1": 0.37869, "loss_box_dn_1": 1.16539, "loss_cls_dn_2": 0.40572, "loss_box_dn_2": 1.14919, "loss_cls_dn_3": 0.36715, "loss_box_dn_3": 1.15486, "loss_cls_dn_4": 0.36456, "loss_box_dn_4": 1.14492, "loss_cls_dn_5": 0.38149, "loss_box_dn_5": 1.18922, "loss_dense_depth": 1.00861, "loss": 35.75771, "grad_norm": 49.94112, "time": 1.55715} -{"mode": "train", "epoch": 1, "iter": 33, "lr": 0.00011, "memory": 49167, "data_time": 0.07353, "loss_cls_0": 1.05829, "loss_box_0": 2.01088, "loss_cns_0": 0.62062, "loss_yns_0": 0.16974, "loss_cls_1": 1.1312, "loss_box_1": 2.43531, "loss_cns_1": 0.58254, "loss_yns_1": 0.17415, "loss_cls_2": 1.14236, "loss_box_2": 2.32259, "loss_cns_2": 0.60671, "loss_yns_2": 0.17208, "loss_cls_3": 1.16581, "loss_box_3": 2.36176, "loss_cns_3": 0.61018, "loss_yns_3": 0.16883, "loss_cls_4": 1.17127, "loss_box_4": 2.35508, "loss_cns_4": 0.61491, "loss_yns_4": 0.16917, "loss_cls_5": 1.17274, "loss_box_5": 2.35154, "loss_cns_5": 0.61834, "loss_yns_5": 0.16867, "loss_cls_dn_0": 0.43292, "loss_box_dn_0": 0.96672, "loss_cls_dn_1": 0.35135, "loss_box_dn_1": 1.18615, "loss_cls_dn_2": 0.37375, "loss_box_dn_2": 1.16322, "loss_cls_dn_3": 0.35937, "loss_box_dn_3": 1.17631, "loss_cls_dn_4": 0.35249, "loss_box_dn_4": 1.15768, "loss_cls_dn_5": 0.35629, "loss_box_dn_5": 1.17132, "loss_dense_depth": 0.97555, "loss": 35.37789, "grad_norm": 51.09058, "time": 1.57061} -{"mode": "train", "epoch": 1, "iter": 34, "lr": 0.00011, "memory": 49167, "data_time": 0.08254, "loss_cls_0": 1.059, "loss_box_0": 2.00141, "loss_cns_0": 0.62169, "loss_yns_0": 0.17087, "loss_cls_1": 1.13137, "loss_box_1": 2.40951, "loss_cns_1": 0.59167, "loss_yns_1": 0.17082, "loss_cls_2": 1.14915, "loss_box_2": 2.30091, "loss_cns_2": 0.612, "loss_yns_2": 0.17077, "loss_cls_3": 1.15591, "loss_box_3": 2.29133, "loss_cns_3": 0.61386, "loss_yns_3": 0.1704, "loss_cls_4": 1.14375, "loss_box_4": 2.28811, "loss_cns_4": 0.61302, "loss_yns_4": 0.16832, "loss_cls_5": 1.16138, "loss_box_5": 2.30111, "loss_cns_5": 0.61632, "loss_yns_5": 0.17032, "loss_cls_dn_0": 0.44443, "loss_box_dn_0": 0.96937, "loss_cls_dn_1": 0.33604, "loss_box_dn_1": 1.17872, "loss_cls_dn_2": 0.34784, "loss_box_dn_2": 1.13451, "loss_cls_dn_3": 0.3607, "loss_box_dn_3": 1.12645, "loss_cls_dn_4": 0.35556, "loss_box_dn_4": 1.12904, "loss_cls_dn_5": 0.35282, "loss_box_dn_5": 1.12898, "loss_dense_depth": 1.01226, "loss": 34.95972, "grad_norm": 36.23091, "time": 1.57288} -{"mode": "train", "epoch": 1, "iter": 35, "lr": 0.00011, "memory": 49167, "data_time": 0.08323, "loss_cls_0": 1.04979, "loss_box_0": 1.97014, "loss_cns_0": 0.62501, "loss_yns_0": 0.17179, "loss_cls_1": 1.14302, "loss_box_1": 2.37118, "loss_cns_1": 0.59863, "loss_yns_1": 0.17691, "loss_cls_2": 1.17578, "loss_box_2": 2.3405, "loss_cns_2": 0.60689, "loss_yns_2": 0.17131, "loss_cls_3": 1.15468, "loss_box_3": 2.36162, "loss_cns_3": 0.61102, "loss_yns_3": 0.16808, "loss_cls_4": 1.15252, "loss_box_4": 2.40452, "loss_cns_4": 0.6085, "loss_yns_4": 0.16916, "loss_cls_5": 1.15455, "loss_box_5": 2.38839, "loss_cns_5": 0.60859, "loss_yns_5": 0.17116, "loss_cls_dn_0": 0.429, "loss_box_dn_0": 0.96216, "loss_cls_dn_1": 0.33241, "loss_box_dn_1": 1.14119, "loss_cls_dn_2": 0.33737, "loss_box_dn_2": 1.11887, "loss_cls_dn_3": 0.35728, "loss_box_dn_3": 1.13895, "loss_cls_dn_4": 0.35724, "loss_box_dn_4": 1.20838, "loss_cls_dn_5": 0.3634, "loss_box_dn_5": 1.20277, "loss_dense_depth": 1.09392, "loss": 35.39666, "grad_norm": 50.01997, "time": 1.56931} -{"mode": "train", "epoch": 1, "iter": 36, "lr": 0.00011, "memory": 49167, "data_time": 0.08157, "loss_cls_0": 1.0561, "loss_box_0": 1.97333, "loss_cns_0": 0.62044, "loss_yns_0": 0.1699, "loss_cls_1": 1.13349, "loss_box_1": 2.43844, "loss_cns_1": 0.5902, "loss_yns_1": 0.1757, "loss_cls_2": 1.15706, "loss_box_2": 2.43503, "loss_cns_2": 0.60363, "loss_yns_2": 0.17162, "loss_cls_3": 1.15464, "loss_box_3": 2.46384, "loss_cns_3": 0.60817, "loss_yns_3": 0.16763, "loss_cls_4": 1.14572, "loss_box_4": 2.50784, "loss_cns_4": 0.60826, "loss_yns_4": 0.17051, "loss_cls_5": 1.13954, "loss_box_5": 2.48157, "loss_cns_5": 0.60906, "loss_yns_5": 0.17388, "loss_cls_dn_0": 0.40693, "loss_box_dn_0": 0.95472, "loss_cls_dn_1": 0.35794, "loss_box_dn_1": 1.08914, "loss_cls_dn_2": 0.36283, "loss_box_dn_2": 1.08307, "loss_cls_dn_3": 0.37196, "loss_box_dn_3": 1.13344, "loss_cls_dn_4": 0.37954, "loss_box_dn_4": 1.24921, "loss_cls_dn_5": 0.39699, "loss_box_dn_5": 1.24093, "loss_dense_depth": 1.00711, "loss": 35.78943, "grad_norm": 60.37665, "time": 1.55983} -{"mode": "train", "epoch": 1, "iter": 37, "lr": 0.00011, "memory": 49167, "data_time": 0.08027, "loss_cls_0": 1.05551, "loss_box_0": 1.98245, "loss_cns_0": 0.62337, "loss_yns_0": 0.17234, "loss_cls_1": 1.11771, "loss_box_1": 2.51567, "loss_cns_1": 0.58928, "loss_yns_1": 0.17399, "loss_cls_2": 1.15111, "loss_box_2": 2.46449, "loss_cns_2": 0.60959, "loss_yns_2": 0.17328, "loss_cls_3": 1.15897, "loss_box_3": 2.49504, "loss_cns_3": 0.60869, "loss_yns_3": 0.16868, "loss_cls_4": 1.13719, "loss_box_4": 2.5313, "loss_cns_4": 0.60545, "loss_yns_4": 0.17164, "loss_cls_5": 1.13623, "loss_box_5": 2.51929, "loss_cns_5": 0.60712, "loss_yns_5": 0.17176, "loss_cls_dn_0": 0.40716, "loss_box_dn_0": 0.95381, "loss_cls_dn_1": 0.34844, "loss_box_dn_1": 1.16955, "loss_cls_dn_2": 0.36196, "loss_box_dn_2": 1.14201, "loss_cls_dn_3": 0.35579, "loss_box_dn_3": 1.18031, "loss_cls_dn_4": 0.36965, "loss_box_dn_4": 1.26969, "loss_cls_dn_5": 0.38384, "loss_box_dn_5": 1.26167, "loss_dense_depth": 1.03539, "loss": 36.17944, "grad_norm": 56.65066, "time": 1.56992} -{"mode": "train", "epoch": 1, "iter": 38, "lr": 0.00011, "memory": 49167, "data_time": 0.0716, "loss_cls_0": 1.02596, "loss_box_0": 1.97241, "loss_cns_0": 0.62337, "loss_yns_0": 0.16818, "loss_cls_1": 1.09491, "loss_box_1": 2.426, "loss_cns_1": 0.59768, "loss_yns_1": 0.17231, "loss_cls_2": 1.15364, "loss_box_2": 2.36235, "loss_cns_2": 0.61771, "loss_yns_2": 0.17265, "loss_cls_3": 1.14027, "loss_box_3": 2.35928, "loss_cns_3": 0.62094, "loss_yns_3": 0.168, "loss_cls_4": 1.12037, "loss_box_4": 2.37179, "loss_cns_4": 0.62103, "loss_yns_4": 0.16913, "loss_cls_5": 1.13267, "loss_box_5": 2.39668, "loss_cns_5": 0.6294, "loss_yns_5": 0.16899, "loss_cls_dn_0": 0.41517, "loss_box_dn_0": 0.95038, "loss_cls_dn_1": 0.35489, "loss_box_dn_1": 1.11523, "loss_cls_dn_2": 0.37363, "loss_box_dn_2": 1.06398, "loss_cls_dn_3": 0.36318, "loss_box_dn_3": 1.0769, "loss_cls_dn_4": 0.3758, "loss_box_dn_4": 1.13117, "loss_cls_dn_5": 0.3828, "loss_box_dn_5": 1.14307, "loss_dense_depth": 0.97125, "loss": 35.00318, "grad_norm": 51.09012, "time": 1.54856} -{"mode": "train", "epoch": 1, "iter": 39, "lr": 0.00012, "memory": 49167, "data_time": 0.06913, "loss_cls_0": 1.05139, "loss_box_0": 1.93529, "loss_cns_0": 0.62609, "loss_yns_0": 0.1677, "loss_cls_1": 1.11216, "loss_box_1": 2.31261, "loss_cns_1": 0.60325, "loss_yns_1": 0.17229, "loss_cls_2": 1.16867, "loss_box_2": 2.25229, "loss_cns_2": 0.62614, "loss_yns_2": 0.17171, "loss_cls_3": 1.1392, "loss_box_3": 2.24193, "loss_cns_3": 0.62976, "loss_yns_3": 0.16874, "loss_cls_4": 1.13188, "loss_box_4": 2.24825, "loss_cns_4": 0.63101, "loss_yns_4": 0.16837, "loss_cls_5": 1.13993, "loss_box_5": 2.24572, "loss_cns_5": 0.64589, "loss_yns_5": 0.16882, "loss_cls_dn_0": 0.41983, "loss_box_dn_0": 0.93023, "loss_cls_dn_1": 0.35697, "loss_box_dn_1": 1.02579, "loss_cls_dn_2": 0.38119, "loss_box_dn_2": 0.98018, "loss_cls_dn_3": 0.37751, "loss_box_dn_3": 0.98734, "loss_cls_dn_4": 0.37372, "loss_box_dn_4": 1.01047, "loss_cls_dn_5": 0.38259, "loss_box_dn_5": 1.01715, "loss_dense_depth": 0.98694, "loss": 33.989, "grad_norm": 49.26616, "time": 1.55402} -{"mode": "train", "epoch": 1, "iter": 40, "lr": 0.00012, "memory": 49167, "data_time": 0.06933, "loss_cls_0": 1.04686, "loss_box_0": 1.9052, "loss_cns_0": 0.62735, "loss_yns_0": 0.16649, "loss_cls_1": 1.10267, "loss_box_1": 2.24715, "loss_cns_1": 0.61165, "loss_yns_1": 0.17199, "loss_cls_2": 1.14351, "loss_box_2": 2.20608, "loss_cns_2": 0.62954, "loss_yns_2": 0.17025, "loss_cls_3": 1.13475, "loss_box_3": 2.19545, "loss_cns_3": 0.63311, "loss_yns_3": 0.1676, "loss_cls_4": 1.13695, "loss_box_4": 2.20214, "loss_cns_4": 0.64163, "loss_yns_4": 0.16847, "loss_cls_5": 1.13957, "loss_box_5": 2.23715, "loss_cns_5": 0.64026, "loss_yns_5": 0.16776, "loss_cls_dn_0": 0.40471, "loss_box_dn_0": 0.92387, "loss_cls_dn_1": 0.35227, "loss_box_dn_1": 0.98697, "loss_cls_dn_2": 0.37426, "loss_box_dn_2": 0.95476, "loss_cls_dn_3": 0.3758, "loss_box_dn_3": 0.95542, "loss_cls_dn_4": 0.35821, "loss_box_dn_4": 0.96263, "loss_cls_dn_5": 0.37299, "loss_box_dn_5": 0.98626, "loss_dense_depth": 1.02956, "loss": 33.53126, "grad_norm": 46.44902, "time": 1.54579} -{"mode": "train", "epoch": 1, "iter": 41, "lr": 0.00012, "memory": 49167, "data_time": 0.08212, "loss_cls_0": 1.02016, "loss_box_0": 1.88316, "loss_cns_0": 0.62798, "loss_yns_0": 0.16474, "loss_cls_1": 1.09439, "loss_box_1": 2.19958, "loss_cns_1": 0.61957, "loss_yns_1": 0.16593, "loss_cls_2": 1.12501, "loss_box_2": 2.18258, "loss_cns_2": 0.63025, "loss_yns_2": 0.1673, "loss_cls_3": 1.12115, "loss_box_3": 2.17708, "loss_cns_3": 0.63168, "loss_yns_3": 0.16434, "loss_cls_4": 1.13752, "loss_box_4": 2.18157, "loss_cns_4": 0.63393, "loss_yns_4": 0.16357, "loss_cls_5": 1.13726, "loss_box_5": 2.21015, "loss_cns_5": 0.63623, "loss_yns_5": 0.16468, "loss_cls_dn_0": 0.39504, "loss_box_dn_0": 0.91472, "loss_cls_dn_1": 0.32585, "loss_box_dn_1": 0.9411, "loss_cls_dn_2": 0.34312, "loss_box_dn_2": 0.92932, "loss_cls_dn_3": 0.3483, "loss_box_dn_3": 0.93489, "loss_cls_dn_4": 0.33267, "loss_box_dn_4": 0.94567, "loss_cls_dn_5": 0.35062, "loss_box_dn_5": 0.98831, "loss_dense_depth": 0.9574, "loss": 32.94684, "grad_norm": 45.98553, "time": 1.62447} -{"mode": "train", "epoch": 1, "iter": 42, "lr": 0.00012, "memory": 49167, "data_time": 0.07054, "loss_cls_0": 1.02903, "loss_box_0": 1.89543, "loss_cns_0": 0.62739, "loss_yns_0": 0.16306, "loss_cls_1": 1.10104, "loss_box_1": 2.16809, "loss_cns_1": 0.62328, "loss_yns_1": 0.16529, "loss_cls_2": 1.16909, "loss_box_2": 2.15757, "loss_cns_2": 0.62622, "loss_yns_2": 0.16668, "loss_cls_3": 1.12939, "loss_box_3": 2.13968, "loss_cns_3": 0.62798, "loss_yns_3": 0.16534, "loss_cls_4": 1.14514, "loss_box_4": 2.15613, "loss_cns_4": 0.62315, "loss_yns_4": 0.16371, "loss_cls_5": 1.1395, "loss_box_5": 2.17347, "loss_cns_5": 0.62493, "loss_yns_5": 0.16297, "loss_cls_dn_0": 0.39751, "loss_box_dn_0": 0.91619, "loss_cls_dn_1": 0.33046, "loss_box_dn_1": 0.94421, "loss_cls_dn_2": 0.34221, "loss_box_dn_2": 0.96714, "loss_cls_dn_3": 0.35167, "loss_box_dn_3": 0.99578, "loss_cls_dn_4": 0.3442, "loss_box_dn_4": 1.02067, "loss_cls_dn_5": 0.36796, "loss_box_dn_5": 1.08817, "loss_dense_depth": 1.02026, "loss": 33.23, "grad_norm": 42.08831, "time": 1.66857} -{"mode": "train", "epoch": 1, "iter": 43, "lr": 0.00012, "memory": 49167, "data_time": 0.13224, "loss_cls_0": 1.02858, "loss_box_0": 1.94382, "loss_cns_0": 0.62111, "loss_yns_0": 0.16189, "loss_cls_1": 1.11035, "loss_box_1": 2.19715, "loss_cns_1": 0.61953, "loss_yns_1": 0.16622, "loss_cls_2": 1.16738, "loss_box_2": 2.18375, "loss_cns_2": 0.62655, "loss_yns_2": 0.16608, "loss_cls_3": 1.14073, "loss_box_3": 2.18316, "loss_cns_3": 0.6256, "loss_yns_3": 0.16602, "loss_cls_4": 1.13826, "loss_box_4": 2.17042, "loss_cns_4": 0.62697, "loss_yns_4": 0.16485, "loss_cls_5": 1.13067, "loss_box_5": 2.20277, "loss_cns_5": 0.62498, "loss_yns_5": 0.16296, "loss_cls_dn_0": 0.41943, "loss_box_dn_0": 0.91535, "loss_cls_dn_1": 0.34836, "loss_box_dn_1": 0.99045, "loss_cls_dn_2": 0.36587, "loss_box_dn_2": 1.03644, "loss_cls_dn_3": 0.36784, "loss_box_dn_3": 1.08387, "loss_cls_dn_4": 0.37592, "loss_box_dn_4": 1.09961, "loss_cls_dn_5": 0.39469, "loss_box_dn_5": 1.17675, "loss_dense_depth": 0.95983, "loss": 33.8642, "grad_norm": 59.15962, "time": 1.60988} -{"mode": "train", "epoch": 1, "iter": 44, "lr": 0.00012, "memory": 49167, "data_time": 0.07796, "loss_cls_0": 1.02444, "loss_box_0": 1.97317, "loss_cns_0": 0.61259, "loss_yns_0": 0.16053, "loss_cls_1": 1.10888, "loss_box_1": 2.21592, "loss_cns_1": 0.61687, "loss_yns_1": 0.16561, "loss_cls_2": 1.11694, "loss_box_2": 2.21585, "loss_cns_2": 0.62297, "loss_yns_2": 0.16768, "loss_cls_3": 1.12148, "loss_box_3": 2.2189, "loss_cns_3": 0.62531, "loss_yns_3": 0.1668, "loss_cls_4": 1.11279, "loss_box_4": 2.20717, "loss_cns_4": 0.62816, "loss_yns_4": 0.16697, "loss_cls_5": 1.11794, "loss_box_5": 2.22006, "loss_cns_5": 0.62858, "loss_yns_5": 0.16442, "loss_cls_dn_0": 0.43157, "loss_box_dn_0": 0.91456, "loss_cls_dn_1": 0.35706, "loss_box_dn_1": 1.04726, "loss_cls_dn_2": 0.3838, "loss_box_dn_2": 1.08917, "loss_cls_dn_3": 0.36996, "loss_box_dn_3": 1.13494, "loss_cls_dn_4": 0.39331, "loss_box_dn_4": 1.14629, "loss_cls_dn_5": 0.40022, "loss_box_dn_5": 1.19848, "loss_dense_depth": 1.02611, "loss": 34.27275, "grad_norm": 61.5323, "time": 1.56511} -{"mode": "train", "epoch": 1, "iter": 45, "lr": 0.00012, "memory": 49167, "data_time": 0.10074, "loss_cls_0": 1.0444, "loss_box_0": 1.95926, "loss_cns_0": 0.61416, "loss_yns_0": 0.16269, "loss_cls_1": 1.15331, "loss_box_1": 2.26674, "loss_cns_1": 0.61262, "loss_yns_1": 0.16714, "loss_cls_2": 1.13758, "loss_box_2": 2.21604, "loss_cns_2": 0.6265, "loss_yns_2": 0.1713, "loss_cls_3": 1.16926, "loss_box_3": 2.22557, "loss_cns_3": 0.6308, "loss_yns_3": 0.16911, "loss_cls_4": 1.15382, "loss_box_4": 2.21056, "loss_cns_4": 0.63182, "loss_yns_4": 0.17207, "loss_cls_5": 1.18027, "loss_box_5": 2.23164, "loss_cns_5": 0.63396, "loss_yns_5": 0.17296, "loss_cls_dn_0": 0.42499, "loss_box_dn_0": 0.92102, "loss_cls_dn_1": 0.35986, "loss_box_dn_1": 1.05774, "loss_cls_dn_2": 0.3971, "loss_box_dn_2": 1.07899, "loss_cls_dn_3": 0.37147, "loss_box_dn_3": 1.10796, "loss_cls_dn_4": 0.39687, "loss_box_dn_4": 1.11196, "loss_cls_dn_5": 0.40158, "loss_box_dn_5": 1.15273, "loss_dense_depth": 1.01551, "loss": 34.51134, "grad_norm": 46.67218, "time": 1.60364} -{"mode": "train", "epoch": 1, "iter": 46, "lr": 0.00012, "memory": 49167, "data_time": 0.07634, "loss_cls_0": 1.02036, "loss_box_0": 1.94818, "loss_cns_0": 0.61622, "loss_yns_0": 0.15998, "loss_cls_1": 1.08284, "loss_box_1": 2.32607, "loss_cns_1": 0.60476, "loss_yns_1": 0.16388, "loss_cls_2": 1.11667, "loss_box_2": 2.25997, "loss_cns_2": 0.62125, "loss_yns_2": 0.16887, "loss_cls_3": 1.14302, "loss_box_3": 2.25059, "loss_cns_3": 0.63038, "loss_yns_3": 0.16628, "loss_cls_4": 1.15241, "loss_box_4": 2.24276, "loss_cns_4": 0.63214, "loss_yns_4": 0.16898, "loss_cls_5": 1.16461, "loss_box_5": 2.27945, "loss_cns_5": 0.64235, "loss_yns_5": 0.17062, "loss_cls_dn_0": 0.40074, "loss_box_dn_0": 0.91443, "loss_cls_dn_1": 0.33764, "loss_box_dn_1": 1.0259, "loss_cls_dn_2": 0.38888, "loss_box_dn_2": 1.00529, "loss_cls_dn_3": 0.35654, "loss_box_dn_3": 1.00534, "loss_cls_dn_4": 0.37312, "loss_box_dn_4": 1.0001, "loss_cls_dn_5": 0.38024, "loss_box_dn_5": 1.02708, "loss_dense_depth": 0.9423, "loss": 33.89025, "grad_norm": 48.59766, "time": 1.56761} -{"mode": "train", "epoch": 1, "iter": 47, "lr": 0.00012, "memory": 49167, "data_time": 0.07798, "loss_cls_0": 1.05334, "loss_box_0": 1.96611, "loss_cns_0": 0.61908, "loss_yns_0": 0.15853, "loss_cls_1": 1.13493, "loss_box_1": 2.28555, "loss_cns_1": 0.61414, "loss_yns_1": 0.16336, "loss_cls_2": 1.10872, "loss_box_2": 2.25172, "loss_cns_2": 0.62537, "loss_yns_2": 0.16504, "loss_cls_3": 1.12355, "loss_box_3": 2.24414, "loss_cns_3": 0.63481, "loss_yns_3": 0.1651, "loss_cls_4": 1.1175, "loss_box_4": 2.22277, "loss_cns_4": 0.63775, "loss_yns_4": 0.16758, "loss_cls_5": 1.12825, "loss_box_5": 2.2192, "loss_cns_5": 0.64496, "loss_yns_5": 0.16742, "loss_cls_dn_0": 0.38991, "loss_box_dn_0": 0.91881, "loss_cls_dn_1": 0.33086, "loss_box_dn_1": 0.96184, "loss_cls_dn_2": 0.38351, "loss_box_dn_2": 0.94163, "loss_cls_dn_3": 0.36066, "loss_box_dn_3": 0.93218, "loss_cls_dn_4": 0.36681, "loss_box_dn_4": 0.9183, "loss_cls_dn_5": 0.37428, "loss_box_dn_5": 0.92149, "loss_dense_depth": 1.00525, "loss": 33.42443, "grad_norm": 50.26023, "time": 1.5871} -{"mode": "train", "epoch": 1, "iter": 48, "lr": 0.00012, "memory": 49167, "data_time": 0.08097, "loss_cls_0": 1.03919, "loss_box_0": 1.96248, "loss_cns_0": 0.61641, "loss_yns_0": 0.16004, "loss_cls_1": 1.14523, "loss_box_1": 2.40165, "loss_cns_1": 0.60047, "loss_yns_1": 0.16002, "loss_cls_2": 1.11091, "loss_box_2": 2.34275, "loss_cns_2": 0.61847, "loss_yns_2": 0.16373, "loss_cls_3": 1.15444, "loss_box_3": 2.33811, "loss_cns_3": 0.63001, "loss_yns_3": 0.16087, "loss_cls_4": 1.11734, "loss_box_4": 2.32187, "loss_cns_4": 0.63269, "loss_yns_4": 0.16491, "loss_cls_5": 1.15963, "loss_box_5": 2.31018, "loss_cns_5": 0.63165, "loss_yns_5": 0.16447, "loss_cls_dn_0": 0.3954, "loss_box_dn_0": 0.90581, "loss_cls_dn_1": 0.32929, "loss_box_dn_1": 0.97743, "loss_cls_dn_2": 0.36429, "loss_box_dn_2": 0.93749, "loss_cls_dn_3": 0.35844, "loss_box_dn_3": 0.92995, "loss_cls_dn_4": 0.35162, "loss_box_dn_4": 0.92333, "loss_cls_dn_5": 0.36355, "loss_box_dn_5": 0.92832, "loss_dense_depth": 0.94187, "loss": 33.81432, "grad_norm": 44.19788, "time": 1.58068} -{"mode": "train", "epoch": 1, "iter": 49, "lr": 0.00012, "memory": 49167, "data_time": 0.07989, "loss_cls_0": 1.01975, "loss_box_0": 1.98331, "loss_cns_0": 0.611, "loss_yns_0": 0.16145, "loss_cls_1": 1.10226, "loss_box_1": 2.38282, "loss_cns_1": 0.60419, "loss_yns_1": 0.16221, "loss_cls_2": 1.11209, "loss_box_2": 2.32737, "loss_cns_2": 0.62211, "loss_yns_2": 0.16206, "loss_cls_3": 1.13332, "loss_box_3": 2.30782, "loss_cns_3": 0.62689, "loss_yns_3": 0.16133, "loss_cls_4": 1.12493, "loss_box_4": 2.32478, "loss_cns_4": 0.62449, "loss_yns_4": 0.16033, "loss_cls_5": 1.14479, "loss_box_5": 2.34988, "loss_cns_5": 0.62112, "loss_yns_5": 0.1613, "loss_cls_dn_0": 0.41005, "loss_box_dn_0": 0.89357, "loss_cls_dn_1": 0.33314, "loss_box_dn_1": 0.9722, "loss_cls_dn_2": 0.334, "loss_box_dn_2": 0.93465, "loss_cls_dn_3": 0.35337, "loss_box_dn_3": 0.94114, "loss_cls_dn_4": 0.33115, "loss_box_dn_4": 0.9646, "loss_cls_dn_5": 0.3566, "loss_box_dn_5": 0.99679, "loss_dense_depth": 1.01156, "loss": 33.8244, "grad_norm": 52.51621, "time": 1.56054} -{"mode": "train", "epoch": 1, "iter": 50, "lr": 0.00012, "memory": 49167, "data_time": 0.0732, "loss_cls_0": 0.99516, "loss_box_0": 1.97556, "loss_cns_0": 0.61715, "loss_yns_0": 0.16018, "loss_cls_1": 1.08147, "loss_box_1": 2.38375, "loss_cns_1": 0.60881, "loss_yns_1": 0.16334, "loss_cls_2": 1.11784, "loss_box_2": 2.31058, "loss_cns_2": 0.62862, "loss_yns_2": 0.16174, "loss_cls_3": 1.09146, "loss_box_3": 2.30275, "loss_cns_3": 0.62991, "loss_yns_3": 0.16017, "loss_cls_4": 1.11481, "loss_box_4": 2.33351, "loss_cns_4": 0.62688, "loss_yns_4": 0.16023, "loss_cls_5": 1.10054, "loss_box_5": 2.33742, "loss_cns_5": 0.62798, "loss_yns_5": 0.16036, "loss_cls_dn_0": 0.40056, "loss_box_dn_0": 0.89543, "loss_cls_dn_1": 0.33802, "loss_box_dn_1": 1.03138, "loss_cls_dn_2": 0.32244, "loss_box_dn_2": 0.993, "loss_cls_dn_3": 0.35607, "loss_box_dn_3": 1.0309, "loss_cls_dn_4": 0.32755, "loss_box_dn_4": 1.08685, "loss_cls_dn_5": 0.36551, "loss_box_dn_5": 1.12131, "loss_dense_depth": 0.94624, "loss": 34.0655, "grad_norm": 61.1904, "time": 1.55499} -{"mode": "train", "epoch": 1, "iter": 51, "lr": 0.00012, "memory": 49167, "data_time": 0.07859, "loss_cls_0": 1.00651, "loss_box_0": 1.93175, "loss_cns_0": 0.61977, "loss_yns_0": 0.16002, "loss_cls_1": 1.10723, "loss_box_1": 2.42152, "loss_cns_1": 0.59465, "loss_yns_1": 0.16458, "loss_cls_2": 1.12189, "loss_box_2": 2.3037, "loss_cns_2": 0.62348, "loss_yns_2": 0.16231, "loss_cls_3": 1.09845, "loss_box_3": 2.33219, "loss_cns_3": 0.62646, "loss_yns_3": 0.16026, "loss_cls_4": 1.12895, "loss_box_4": 2.36754, "loss_cns_4": 0.62397, "loss_yns_4": 0.16232, "loss_cls_5": 1.11104, "loss_box_5": 2.34769, "loss_cns_5": 0.62867, "loss_yns_5": 0.16362, "loss_cls_dn_0": 0.39167, "loss_box_dn_0": 0.87695, "loss_cls_dn_1": 0.31715, "loss_box_dn_1": 1.09692, "loss_cls_dn_2": 0.31308, "loss_box_dn_2": 1.04807, "loss_cls_dn_3": 0.33457, "loss_box_dn_3": 1.092, "loss_cls_dn_4": 0.31743, "loss_box_dn_4": 1.14909, "loss_cls_dn_5": 0.35687, "loss_box_dn_5": 1.16809, "loss_dense_depth": 0.88915, "loss": 34.31961, "grad_norm": 53.98701, "time": 1.56128} -{"mode": "train", "epoch": 1, "iter": 52, "lr": 0.00012, "memory": 49167, "data_time": 0.08658, "loss_cls_0": 1.01403, "loss_box_0": 1.92974, "loss_cns_0": 0.61935, "loss_yns_0": 0.16324, "loss_cls_1": 1.12667, "loss_box_1": 2.37276, "loss_cns_1": 0.59985, "loss_yns_1": 0.16693, "loss_cls_2": 1.12271, "loss_box_2": 2.30053, "loss_cns_2": 0.62332, "loss_yns_2": 0.16447, "loss_cls_3": 1.13472, "loss_box_3": 2.30919, "loss_cns_3": 0.63078, "loss_yns_3": 0.15994, "loss_cls_4": 1.11839, "loss_box_4": 2.315, "loss_cns_4": 0.63472, "loss_yns_4": 0.16182, "loss_cls_5": 1.12265, "loss_box_5": 2.31543, "loss_cns_5": 0.63944, "loss_yns_5": 0.16471, "loss_cls_dn_0": 0.39021, "loss_box_dn_0": 0.87247, "loss_cls_dn_1": 0.30115, "loss_box_dn_1": 1.07003, "loss_cls_dn_2": 0.31538, "loss_box_dn_2": 1.03094, "loss_cls_dn_3": 0.32019, "loss_box_dn_3": 1.05578, "loss_cls_dn_4": 0.32471, "loss_box_dn_4": 1.09718, "loss_cls_dn_5": 0.36144, "loss_box_dn_5": 1.11805, "loss_dense_depth": 0.92531, "loss": 34.09322, "grad_norm": 58.34633, "time": 1.59228} -{"mode": "train", "epoch": 1, "iter": 53, "lr": 0.00012, "memory": 49167, "data_time": 0.08507, "loss_cls_0": 1.02347, "loss_box_0": 1.97844, "loss_cns_0": 0.61506, "loss_yns_0": 0.1649, "loss_cls_1": 1.07747, "loss_box_1": 2.39145, "loss_cns_1": 0.59711, "loss_yns_1": 0.16795, "loss_cls_2": 1.11175, "loss_box_2": 2.32179, "loss_cns_2": 0.62427, "loss_yns_2": 0.16602, "loss_cls_3": 1.15212, "loss_box_3": 2.34463, "loss_cns_3": 0.62655, "loss_yns_3": 0.1649, "loss_cls_4": 1.12462, "loss_box_4": 2.33838, "loss_cns_4": 0.62951, "loss_yns_4": 0.16239, "loss_cls_5": 1.12315, "loss_box_5": 2.33777, "loss_cns_5": 0.63539, "loss_yns_5": 0.16446, "loss_cls_dn_0": 0.39577, "loss_box_dn_0": 0.87326, "loss_cls_dn_1": 0.28785, "loss_box_dn_1": 1.00145, "loss_cls_dn_2": 0.32202, "loss_box_dn_2": 0.96022, "loss_cls_dn_3": 0.31239, "loss_box_dn_3": 0.97273, "loss_cls_dn_4": 0.32776, "loss_box_dn_4": 0.99097, "loss_cls_dn_5": 0.36413, "loss_box_dn_5": 1.01033, "loss_dense_depth": 0.93652, "loss": 33.79897, "grad_norm": 61.77947, "time": 1.5682} -{"mode": "train", "epoch": 1, "iter": 54, "lr": 0.00012, "memory": 49167, "data_time": 0.08326, "loss_cls_0": 1.00805, "loss_box_0": 1.91945, "loss_cns_0": 0.62219, "loss_yns_0": 0.16221, "loss_cls_1": 1.09791, "loss_box_1": 2.31697, "loss_cns_1": 0.60646, "loss_yns_1": 0.16438, "loss_cls_2": 1.10887, "loss_box_2": 2.23751, "loss_cns_2": 0.63069, "loss_yns_2": 0.1655, "loss_cls_3": 1.10659, "loss_box_3": 2.24957, "loss_cns_3": 0.63707, "loss_yns_3": 0.16279, "loss_cls_4": 1.10669, "loss_box_4": 2.26407, "loss_cns_4": 0.63623, "loss_yns_4": 0.1611, "loss_cls_5": 1.11143, "loss_box_5": 2.2343, "loss_cns_5": 0.64001, "loss_yns_5": 0.15995, "loss_cls_dn_0": 0.37788, "loss_box_dn_0": 0.87396, "loss_cls_dn_1": 0.27614, "loss_box_dn_1": 0.98792, "loss_cls_dn_2": 0.32248, "loss_box_dn_2": 0.94095, "loss_cls_dn_3": 0.31064, "loss_box_dn_3": 0.94129, "loss_cls_dn_4": 0.31684, "loss_box_dn_4": 0.95207, "loss_cls_dn_5": 0.34462, "loss_box_dn_5": 0.95188, "loss_dense_depth": 0.96244, "loss": 33.06912, "grad_norm": 46.98807, "time": 1.57256} -{"mode": "train", "epoch": 1, "iter": 55, "lr": 0.00012, "memory": 49167, "data_time": 0.08932, "loss_cls_0": 1.00344, "loss_box_0": 1.9422, "loss_cns_0": 0.61922, "loss_yns_0": 0.16476, "loss_cls_1": 1.07377, "loss_box_1": 2.29131, "loss_cns_1": 0.60478, "loss_yns_1": 0.17012, "loss_cls_2": 1.08996, "loss_box_2": 2.18691, "loss_cns_2": 0.63229, "loss_yns_2": 0.16777, "loss_cls_3": 1.08697, "loss_box_3": 2.19772, "loss_cns_3": 0.64394, "loss_yns_3": 0.1663, "loss_cls_4": 1.08905, "loss_box_4": 2.22014, "loss_cns_4": 0.64422, "loss_yns_4": 0.16675, "loss_cls_5": 1.1046, "loss_box_5": 2.19719, "loss_cns_5": 0.64135, "loss_yns_5": 0.16918, "loss_cls_dn_0": 0.39013, "loss_box_dn_0": 0.86515, "loss_cls_dn_1": 0.29615, "loss_box_dn_1": 0.89591, "loss_cls_dn_2": 0.35433, "loss_box_dn_2": 0.84497, "loss_cls_dn_3": 0.34986, "loss_box_dn_3": 0.84571, "loss_cls_dn_4": 0.33314, "loss_box_dn_4": 0.85332, "loss_cls_dn_5": 0.35313, "loss_box_dn_5": 0.85284, "loss_dense_depth": 0.90163, "loss": 32.41021, "grad_norm": 55.90255, "time": 1.574} -{"mode": "train", "epoch": 1, "iter": 56, "lr": 0.00012, "memory": 49167, "data_time": 0.08359, "loss_cls_0": 0.97249, "loss_box_0": 1.90979, "loss_cns_0": 0.61484, "loss_yns_0": 0.16431, "loss_cls_1": 1.03752, "loss_box_1": 2.15399, "loss_cns_1": 0.60815, "loss_yns_1": 0.16566, "loss_cls_2": 1.08861, "loss_box_2": 2.06793, "loss_cns_2": 0.63613, "loss_yns_2": 0.16678, "loss_cls_3": 1.11315, "loss_box_3": 2.09581, "loss_cns_3": 0.64247, "loss_yns_3": 0.16478, "loss_cls_4": 1.12808, "loss_box_4": 2.09994, "loss_cns_4": 0.64308, "loss_yns_4": 0.1649, "loss_cls_5": 1.11146, "loss_box_5": 2.10593, "loss_cns_5": 0.63758, "loss_yns_5": 0.16392, "loss_cls_dn_0": 0.3901, "loss_box_dn_0": 0.86202, "loss_cls_dn_1": 0.29467, "loss_box_dn_1": 0.90107, "loss_cls_dn_2": 0.34981, "loss_box_dn_2": 0.867, "loss_cls_dn_3": 0.35415, "loss_box_dn_3": 0.88235, "loss_cls_dn_4": 0.33463, "loss_box_dn_4": 0.87824, "loss_cls_dn_5": 0.34036, "loss_box_dn_5": 0.90178, "loss_dense_depth": 0.92212, "loss": 31.93559, "grad_norm": 72.73215, "time": 1.56071} -{"mode": "train", "epoch": 1, "iter": 57, "lr": 0.00012, "memory": 49167, "data_time": 0.08289, "loss_cls_0": 0.97462, "loss_box_0": 1.87971, "loss_cns_0": 0.618, "loss_yns_0": 0.16317, "loss_cls_1": 1.04649, "loss_box_1": 2.14862, "loss_cns_1": 0.61534, "loss_yns_1": 0.1652, "loss_cls_2": 1.0944, "loss_box_2": 2.1185, "loss_cns_2": 0.63315, "loss_yns_2": 0.16487, "loss_cls_3": 1.1179, "loss_box_3": 2.13524, "loss_cns_3": 0.63748, "loss_yns_3": 0.16457, "loss_cls_4": 1.15648, "loss_box_4": 2.12453, "loss_cns_4": 0.64006, "loss_yns_4": 0.16387, "loss_cls_5": 1.13105, "loss_box_5": 2.15243, "loss_cns_5": 0.63824, "loss_yns_5": 0.16513, "loss_cls_dn_0": 0.3933, "loss_box_dn_0": 0.85693, "loss_cls_dn_1": 0.29594, "loss_box_dn_1": 0.89677, "loss_cls_dn_2": 0.33986, "loss_box_dn_2": 0.88889, "loss_cls_dn_3": 0.34727, "loss_box_dn_3": 0.90619, "loss_cls_dn_4": 0.34039, "loss_box_dn_4": 0.8987, "loss_cls_dn_5": 0.33692, "loss_box_dn_5": 0.93817, "loss_dense_depth": 0.92149, "loss": 32.20989, "grad_norm": 65.80556, "time": 1.56261} -{"mode": "train", "epoch": 1, "iter": 58, "lr": 0.00012, "memory": 49167, "data_time": 0.08837, "loss_cls_0": 0.97294, "loss_box_0": 1.88522, "loss_cns_0": 0.61934, "loss_yns_0": 0.16356, "loss_cls_1": 1.04595, "loss_box_1": 2.09127, "loss_cns_1": 0.61906, "loss_yns_1": 0.1655, "loss_cls_2": 1.07224, "loss_box_2": 2.0961, "loss_cns_2": 0.62775, "loss_yns_2": 0.1662, "loss_cls_3": 1.07565, "loss_box_3": 2.06941, "loss_cns_3": 0.63805, "loss_yns_3": 0.16572, "loss_cls_4": 1.08886, "loss_box_4": 2.07697, "loss_cns_4": 0.63819, "loss_yns_4": 0.16496, "loss_cls_5": 1.09932, "loss_box_5": 2.07437, "loss_cns_5": 0.63921, "loss_yns_5": 0.16517, "loss_cls_dn_0": 0.38144, "loss_box_dn_0": 0.853, "loss_cls_dn_1": 0.29578, "loss_box_dn_1": 0.90827, "loss_cls_dn_2": 0.32076, "loss_box_dn_2": 0.92701, "loss_cls_dn_3": 0.3253, "loss_box_dn_3": 0.93354, "loss_cls_dn_4": 0.33808, "loss_box_dn_4": 0.9439, "loss_cls_dn_5": 0.33487, "loss_box_dn_5": 0.97734, "loss_dense_depth": 0.8851, "loss": 31.84541, "grad_norm": 53.46335, "time": 1.56488} -{"mode": "train", "epoch": 1, "iter": 59, "lr": 0.00012, "memory": 49167, "data_time": 0.08372, "loss_cls_0": 0.99031, "loss_box_0": 1.90883, "loss_cns_0": 0.61875, "loss_yns_0": 0.16295, "loss_cls_1": 1.04355, "loss_box_1": 2.15314, "loss_cns_1": 0.62387, "loss_yns_1": 0.16453, "loss_cls_2": 1.11788, "loss_box_2": 2.1274, "loss_cns_2": 0.63524, "loss_yns_2": 0.16492, "loss_cls_3": 1.23457, "loss_box_3": 2.09774, "loss_cns_3": 0.64194, "loss_yns_3": 0.16472, "loss_cls_4": 1.11464, "loss_box_4": 2.13097, "loss_cns_4": 0.64343, "loss_yns_4": 0.16322, "loss_cls_5": 1.15114, "loss_box_5": 2.10904, "loss_cns_5": 0.64714, "loss_yns_5": 0.16752, "loss_cls_dn_0": 0.36584, "loss_box_dn_0": 0.85416, "loss_cls_dn_1": 0.29162, "loss_box_dn_1": 0.93464, "loss_cls_dn_2": 0.30628, "loss_box_dn_2": 0.95905, "loss_cls_dn_3": 0.31531, "loss_box_dn_3": 0.96032, "loss_cls_dn_4": 0.32813, "loss_box_dn_4": 0.98411, "loss_cls_dn_5": 0.33801, "loss_box_dn_5": 1.0085, "loss_dense_depth": 0.87081, "loss": 32.4942, "grad_norm": 72.08705, "time": 1.55722} -{"mode": "train", "epoch": 1, "iter": 60, "lr": 0.00012, "memory": 49167, "data_time": 0.07992, "loss_cls_0": 0.99538, "loss_box_0": 1.86557, "loss_cns_0": 0.62227, "loss_yns_0": 0.1592, "loss_cls_1": 1.06016, "loss_box_1": 2.1364, "loss_cns_1": 0.63321, "loss_yns_1": 0.1631, "loss_cls_2": 1.17445, "loss_box_2": 2.12091, "loss_cns_2": 0.64468, "loss_yns_2": 0.16368, "loss_cls_3": 1.1717, "loss_box_3": 2.10922, "loss_cns_3": 0.64316, "loss_yns_3": 0.16335, "loss_cls_4": 1.1971, "loss_box_4": 2.14124, "loss_cns_4": 0.64663, "loss_yns_4": 0.16208, "loss_cls_5": 1.14347, "loss_box_5": 2.11655, "loss_cns_5": 0.64893, "loss_yns_5": 0.1658, "loss_cls_dn_0": 0.35814, "loss_box_dn_0": 0.86137, "loss_cls_dn_1": 0.27576, "loss_box_dn_1": 0.96249, "loss_cls_dn_2": 0.29844, "loss_box_dn_2": 0.98064, "loss_cls_dn_3": 0.32693, "loss_box_dn_3": 0.97865, "loss_cls_dn_4": 0.3221, "loss_box_dn_4": 0.99666, "loss_cls_dn_5": 0.3479, "loss_box_dn_5": 1.01201, "loss_dense_depth": 0.89075, "loss": 32.66007, "grad_norm": 88.42252, "time": 1.55982} -{"mode": "train", "epoch": 1, "iter": 61, "lr": 0.00012, "memory": 49167, "data_time": 0.0939, "loss_cls_0": 0.98924, "loss_box_0": 1.85195, "loss_cns_0": 0.62028, "loss_yns_0": 0.15839, "loss_cls_1": 1.05463, "loss_box_1": 2.12888, "loss_cns_1": 0.63088, "loss_yns_1": 0.16239, "loss_cls_2": 1.08052, "loss_box_2": 2.0891, "loss_cns_2": 0.64437, "loss_yns_2": 0.16263, "loss_cls_3": 1.09894, "loss_box_3": 2.08577, "loss_cns_3": 0.64302, "loss_yns_3": 0.16257, "loss_cls_4": 1.13104, "loss_box_4": 2.07826, "loss_cns_4": 0.64777, "loss_yns_4": 0.1614, "loss_cls_5": 1.09155, "loss_box_5": 2.04522, "loss_cns_5": 0.64938, "loss_yns_5": 0.16321, "loss_cls_dn_0": 0.36318, "loss_box_dn_0": 0.85666, "loss_cls_dn_1": 0.26833, "loss_box_dn_1": 0.98063, "loss_cls_dn_2": 0.30765, "loss_box_dn_2": 0.96949, "loss_cls_dn_3": 0.34903, "loss_box_dn_3": 0.96192, "loss_cls_dn_4": 0.33424, "loss_box_dn_4": 0.9594, "loss_cls_dn_5": 0.36372, "loss_box_dn_5": 0.96346, "loss_dense_depth": 0.93089, "loss": 32.13999, "grad_norm": 57.33644, "time": 1.6408} -{"mode": "train", "epoch": 1, "iter": 62, "lr": 0.00012, "memory": 49167, "data_time": 0.11919, "loss_cls_0": 0.98308, "loss_box_0": 1.86751, "loss_cns_0": 0.61909, "loss_yns_0": 0.16003, "loss_cls_1": 1.06015, "loss_box_1": 2.18903, "loss_cns_1": 0.62818, "loss_yns_1": 0.16125, "loss_cls_2": 1.08714, "loss_box_2": 2.10841, "loss_cns_2": 0.64431, "loss_yns_2": 0.16207, "loss_cls_3": 1.09544, "loss_box_3": 2.1152, "loss_cns_3": 0.64731, "loss_yns_3": 0.16115, "loss_cls_4": 1.08437, "loss_box_4": 2.08436, "loss_cns_4": 0.65063, "loss_yns_4": 0.16023, "loss_cls_5": 1.0902, "loss_box_5": 2.09339, "loss_cns_5": 0.64931, "loss_yns_5": 0.16112, "loss_cls_dn_0": 0.37953, "loss_box_dn_0": 0.86794, "loss_cls_dn_1": 0.27225, "loss_box_dn_1": 0.89941, "loss_cls_dn_2": 0.32696, "loss_box_dn_2": 0.86695, "loss_cls_dn_3": 0.35706, "loss_box_dn_3": 0.86682, "loss_cls_dn_4": 0.34221, "loss_box_dn_4": 0.85656, "loss_cls_dn_5": 0.37732, "loss_box_dn_5": 0.87021, "loss_dense_depth": 0.89325, "loss": 31.83941, "grad_norm": 43.43135, "time": 1.65181} -{"mode": "train", "epoch": 1, "iter": 63, "lr": 0.00012, "memory": 49167, "data_time": 0.12736, "loss_cls_0": 0.97497, "loss_box_0": 1.88104, "loss_cns_0": 0.61345, "loss_yns_0": 0.16047, "loss_cls_1": 1.07157, "loss_box_1": 2.19401, "loss_cns_1": 0.62785, "loss_yns_1": 0.16196, "loss_cls_2": 1.13436, "loss_box_2": 2.15526, "loss_cns_2": 0.64263, "loss_yns_2": 0.16295, "loss_cls_3": 1.13427, "loss_box_3": 2.13575, "loss_cns_3": 0.64927, "loss_yns_3": 0.16161, "loss_cls_4": 1.09786, "loss_box_4": 2.1336, "loss_cns_4": 0.65393, "loss_yns_4": 0.16073, "loss_cls_5": 1.13583, "loss_box_5": 2.14609, "loss_cns_5": 0.65294, "loss_yns_5": 0.16118, "loss_cls_dn_0": 0.38709, "loss_box_dn_0": 0.87478, "loss_cls_dn_1": 0.26511, "loss_box_dn_1": 0.87789, "loss_cls_dn_2": 0.30955, "loss_box_dn_2": 0.8633, "loss_cls_dn_3": 0.31609, "loss_box_dn_3": 0.86982, "loss_cls_dn_4": 0.30075, "loss_box_dn_4": 0.8769, "loss_cls_dn_5": 0.33746, "loss_box_dn_5": 0.89163, "loss_dense_depth": 0.89678, "loss": 32.07074, "grad_norm": 63.76735, "time": 1.61391} -{"mode": "train", "epoch": 1, "iter": 64, "lr": 0.00013, "memory": 49167, "data_time": 0.07649, "loss_cls_0": 0.95257, "loss_box_0": 1.8526, "loss_cns_0": 0.61632, "loss_yns_0": 0.16062, "loss_cls_1": 1.05815, "loss_box_1": 2.09759, "loss_cns_1": 0.63524, "loss_yns_1": 0.16535, "loss_cls_2": 1.15989, "loss_box_2": 2.1203, "loss_cns_2": 0.64514, "loss_yns_2": 0.16512, "loss_cls_3": 1.18215, "loss_box_3": 2.0967, "loss_cns_3": 0.64857, "loss_yns_3": 0.16192, "loss_cls_4": 1.18797, "loss_box_4": 2.12489, "loss_cns_4": 0.65279, "loss_yns_4": 0.16315, "loss_cls_5": 1.20093, "loss_box_5": 2.12115, "loss_cns_5": 0.65015, "loss_yns_5": 0.16575, "loss_cls_dn_0": 0.37929, "loss_box_dn_0": 0.85908, "loss_cls_dn_1": 0.24659, "loss_box_dn_1": 0.86726, "loss_cls_dn_2": 0.2746, "loss_box_dn_2": 0.87196, "loss_cls_dn_3": 0.2806, "loss_box_dn_3": 0.87902, "loss_cls_dn_4": 0.26731, "loss_box_dn_4": 0.90501, "loss_cls_dn_5": 0.29864, "loss_box_dn_5": 0.91826, "loss_dense_depth": 0.87138, "loss": 31.90401, "grad_norm": 85.02466, "time": 1.56425} -{"mode": "train", "epoch": 1, "iter": 65, "lr": 0.00013, "memory": 49167, "data_time": 0.09811, "loss_cls_0": 0.96227, "loss_box_0": 1.85105, "loss_cns_0": 0.6148, "loss_yns_0": 0.16187, "loss_cls_1": 1.04682, "loss_box_1": 2.03895, "loss_cns_1": 0.63538, "loss_yns_1": 0.16282, "loss_cls_2": 1.0938, "loss_box_2": 2.02722, "loss_cns_2": 0.6471, "loss_yns_2": 0.16172, "loss_cls_3": 1.10588, "loss_box_3": 2.01734, "loss_cns_3": 0.6494, "loss_yns_3": 0.161, "loss_cls_4": 1.09396, "loss_box_4": 2.01223, "loss_cns_4": 0.65251, "loss_yns_4": 0.16186, "loss_cls_5": 1.10044, "loss_box_5": 2.01405, "loss_cns_5": 0.64889, "loss_yns_5": 0.16263, "loss_cls_dn_0": 0.38013, "loss_box_dn_0": 0.85368, "loss_cls_dn_1": 0.23591, "loss_box_dn_1": 0.92538, "loss_cls_dn_2": 0.25789, "loss_box_dn_2": 0.91825, "loss_cls_dn_3": 0.267, "loss_box_dn_3": 0.92737, "loss_cls_dn_4": 0.26762, "loss_box_dn_4": 0.94514, "loss_cls_dn_5": 0.28749, "loss_box_dn_5": 0.96467, "loss_dense_depth": 0.85958, "loss": 31.27407, "grad_norm": 53.98653, "time": 1.58155} -{"mode": "train", "epoch": 1, "iter": 66, "lr": 0.00013, "memory": 49167, "data_time": 0.07326, "loss_cls_0": 0.96978, "loss_box_0": 1.8414, "loss_cns_0": 0.6227, "loss_yns_0": 0.16397, "loss_cls_1": 1.07607, "loss_box_1": 2.05655, "loss_cns_1": 0.63319, "loss_yns_1": 0.163, "loss_cls_2": 1.08397, "loss_box_2": 2.03093, "loss_cns_2": 0.65054, "loss_yns_2": 0.16471, "loss_cls_3": 1.0854, "loss_box_3": 2.06893, "loss_cns_3": 0.64994, "loss_yns_3": 0.16228, "loss_cls_4": 1.09756, "loss_box_4": 2.01239, "loss_cns_4": 0.65093, "loss_yns_4": 0.16362, "loss_cls_5": 1.09217, "loss_box_5": 2.01978, "loss_cns_5": 0.65075, "loss_yns_5": 0.16687, "loss_cls_dn_0": 0.36579, "loss_box_dn_0": 0.83939, "loss_cls_dn_1": 0.2327, "loss_box_dn_1": 0.97466, "loss_cls_dn_2": 0.26546, "loss_box_dn_2": 0.95503, "loss_cls_dn_3": 0.28207, "loss_box_dn_3": 0.97344, "loss_cls_dn_4": 0.30638, "loss_box_dn_4": 0.97693, "loss_cls_dn_5": 0.3143, "loss_box_dn_5": 0.994, "loss_dense_depth": 0.84427, "loss": 31.60182, "grad_norm": 48.02287, "time": 1.57083} -{"mode": "train", "epoch": 1, "iter": 67, "lr": 0.00013, "memory": 49167, "data_time": 0.07269, "loss_cls_0": 0.97042, "loss_box_0": 1.86916, "loss_cns_0": 0.62011, "loss_yns_0": 0.15985, "loss_cls_1": 1.13407, "loss_box_1": 2.16278, "loss_cns_1": 0.62149, "loss_yns_1": 0.16367, "loss_cls_2": 1.14942, "loss_box_2": 2.14481, "loss_cns_2": 0.64417, "loss_yns_2": 0.16287, "loss_cls_3": 1.13992, "loss_box_3": 2.18087, "loss_cns_3": 0.64507, "loss_yns_3": 0.15989, "loss_cls_4": 1.13135, "loss_box_4": 2.14107, "loss_cns_4": 0.64566, "loss_yns_4": 0.16182, "loss_cls_5": 1.13746, "loss_box_5": 2.15564, "loss_cns_5": 0.64609, "loss_yns_5": 0.1628, "loss_cls_dn_0": 0.36601, "loss_box_dn_0": 0.84417, "loss_cls_dn_1": 0.25348, "loss_box_dn_1": 0.95051, "loss_cls_dn_2": 0.29938, "loss_box_dn_2": 0.92605, "loss_cls_dn_3": 0.31871, "loss_box_dn_3": 0.94834, "loss_cls_dn_4": 0.32796, "loss_box_dn_4": 0.95568, "loss_cls_dn_5": 0.3501, "loss_box_dn_5": 0.97541, "loss_dense_depth": 0.87497, "loss": 32.50124, "grad_norm": 63.50334, "time": 1.57466} -{"mode": "train", "epoch": 1, "iter": 68, "lr": 0.00013, "memory": 49167, "data_time": 0.07987, "loss_cls_0": 0.9695, "loss_box_0": 1.87565, "loss_cns_0": 0.62266, "loss_yns_0": 0.1587, "loss_cls_1": 1.10353, "loss_box_1": 2.13323, "loss_cns_1": 0.6208, "loss_yns_1": 0.16326, "loss_cls_2": 1.13292, "loss_box_2": 2.1026, "loss_cns_2": 0.63794, "loss_yns_2": 0.16553, "loss_cls_3": 1.12531, "loss_box_3": 2.11987, "loss_cns_3": 0.64211, "loss_yns_3": 0.16091, "loss_cls_4": 1.1037, "loss_box_4": 2.09558, "loss_cns_4": 0.64339, "loss_yns_4": 0.16446, "loss_cls_5": 1.13594, "loss_box_5": 2.11849, "loss_cns_5": 0.64375, "loss_yns_5": 0.16085, "loss_cls_dn_0": 0.35574, "loss_box_dn_0": 0.86207, "loss_cls_dn_1": 0.26652, "loss_box_dn_1": 0.92639, "loss_cls_dn_2": 0.31762, "loss_box_dn_2": 0.89653, "loss_cls_dn_3": 0.33518, "loss_box_dn_3": 0.91096, "loss_cls_dn_4": 0.31277, "loss_box_dn_4": 0.92163, "loss_cls_dn_5": 0.35918, "loss_box_dn_5": 0.94627, "loss_dense_depth": 0.85304, "loss": 32.06458, "grad_norm": 60.69571, "time": 1.57329} -{"mode": "train", "epoch": 1, "iter": 69, "lr": 0.00013, "memory": 49167, "data_time": 0.08693, "loss_cls_0": 0.98181, "loss_box_0": 1.87566, "loss_cns_0": 0.62119, "loss_yns_0": 0.15744, "loss_cls_1": 1.033, "loss_box_1": 2.04707, "loss_cns_1": 0.62252, "loss_yns_1": 0.1599, "loss_cls_2": 1.0844, "loss_box_2": 2.0002, "loss_cns_2": 0.63814, "loss_yns_2": 0.1636, "loss_cls_3": 1.08133, "loss_box_3": 1.96437, "loss_cns_3": 0.64766, "loss_yns_3": 0.16066, "loss_cls_4": 1.15634, "loss_box_4": 1.94547, "loss_cns_4": 0.65162, "loss_yns_4": 0.16431, "loss_cls_5": 1.09744, "loss_box_5": 1.96623, "loss_cns_5": 0.65002, "loss_yns_5": 0.15993, "loss_cls_dn_0": 0.33639, "loss_box_dn_0": 0.86746, "loss_cls_dn_1": 0.24998, "loss_box_dn_1": 0.91014, "loss_cls_dn_2": 0.2981, "loss_box_dn_2": 0.87579, "loss_cls_dn_3": 0.31477, "loss_box_dn_3": 0.87302, "loss_cls_dn_4": 0.27518, "loss_box_dn_4": 0.88353, "loss_cls_dn_5": 0.33667, "loss_box_dn_5": 0.90335, "loss_dense_depth": 0.81535, "loss": 30.97008, "grad_norm": 45.40764, "time": 1.5573} -{"mode": "train", "epoch": 1, "iter": 70, "lr": 0.00013, "memory": 49167, "data_time": 0.07873, "loss_cls_0": 0.97713, "loss_box_0": 1.86321, "loss_cns_0": 0.61999, "loss_yns_0": 0.1584, "loss_cls_1": 0.98888, "loss_box_1": 2.05602, "loss_cns_1": 0.6205, "loss_yns_1": 0.15798, "loss_cls_2": 1.03519, "loss_box_2": 2.01432, "loss_cns_2": 0.63642, "loss_yns_2": 0.15976, "loss_cls_3": 1.04326, "loss_box_3": 1.98538, "loss_cns_3": 0.64269, "loss_yns_3": 0.1588, "loss_cls_4": 1.05038, "loss_box_4": 1.99259, "loss_cns_4": 0.6495, "loss_yns_4": 0.15993, "loss_cls_5": 1.04768, "loss_box_5": 2.00206, "loss_cns_5": 0.64445, "loss_yns_5": 0.16134, "loss_cls_dn_0": 0.35822, "loss_box_dn_0": 0.87608, "loss_cls_dn_1": 0.24758, "loss_box_dn_1": 0.8564, "loss_cls_dn_2": 0.29442, "loss_box_dn_2": 0.83019, "loss_cls_dn_3": 0.31876, "loss_box_dn_3": 0.82659, "loss_cls_dn_4": 0.29479, "loss_box_dn_4": 0.84523, "loss_cls_dn_5": 0.35109, "loss_box_dn_5": 0.86041, "loss_dense_depth": 0.86444, "loss": 30.65006, "grad_norm": 50.16703, "time": 1.56207} -{"mode": "train", "epoch": 1, "iter": 71, "lr": 0.00013, "memory": 49167, "data_time": 0.08594, "loss_cls_0": 0.93272, "loss_box_0": 1.80948, "loss_cns_0": 0.6236, "loss_yns_0": 0.15823, "loss_cls_1": 1.01974, "loss_box_1": 2.09514, "loss_cns_1": 0.61911, "loss_yns_1": 0.15853, "loss_cls_2": 1.03247, "loss_box_2": 2.04039, "loss_cns_2": 0.63939, "loss_yns_2": 0.15637, "loss_cls_3": 1.04412, "loss_box_3": 2.02645, "loss_cns_3": 0.64873, "loss_yns_3": 0.15851, "loss_cls_4": 1.0572, "loss_box_4": 2.01515, "loss_cns_4": 0.65542, "loss_yns_4": 0.15977, "loss_cls_5": 1.05848, "loss_box_5": 2.02765, "loss_cns_5": 0.64882, "loss_yns_5": 0.1583, "loss_cls_dn_0": 0.35801, "loss_box_dn_0": 0.87189, "loss_cls_dn_1": 0.23747, "loss_box_dn_1": 0.8927, "loss_cls_dn_2": 0.27734, "loss_box_dn_2": 0.86163, "loss_cls_dn_3": 0.29891, "loss_box_dn_3": 0.86189, "loss_cls_dn_4": 0.29397, "loss_box_dn_4": 0.86624, "loss_cls_dn_5": 0.3353, "loss_box_dn_5": 0.87554, "loss_dense_depth": 0.86321, "loss": 30.83786, "grad_norm": 59.41325, "time": 1.57744} -{"mode": "train", "epoch": 1, "iter": 72, "lr": 0.00013, "memory": 49167, "data_time": 0.08305, "loss_cls_0": 0.93863, "loss_box_0": 1.80051, "loss_cns_0": 0.62471, "loss_yns_0": 0.15889, "loss_cls_1": 1.04885, "loss_box_1": 2.07408, "loss_cns_1": 0.61846, "loss_yns_1": 0.16352, "loss_cls_2": 1.07739, "loss_box_2": 1.98291, "loss_cns_2": 0.63969, "loss_yns_2": 0.16068, "loss_cls_3": 1.06601, "loss_box_3": 1.96826, "loss_cns_3": 0.64714, "loss_yns_3": 0.159, "loss_cls_4": 1.06117, "loss_box_4": 1.93063, "loss_cns_4": 0.65037, "loss_yns_4": 0.16392, "loss_cls_5": 1.07017, "loss_box_5": 1.95507, "loss_cns_5": 0.64814, "loss_yns_5": 0.15797, "loss_cls_dn_0": 0.36445, "loss_box_dn_0": 0.86236, "loss_cls_dn_1": 0.24166, "loss_box_dn_1": 0.87577, "loss_cls_dn_2": 0.27035, "loss_box_dn_2": 0.84851, "loss_cls_dn_3": 0.28642, "loss_box_dn_3": 0.84726, "loss_cls_dn_4": 0.28805, "loss_box_dn_4": 0.83353, "loss_cls_dn_5": 0.31888, "loss_box_dn_5": 0.85048, "loss_dense_depth": 0.85574, "loss": 30.50964, "grad_norm": 59.13472, "time": 1.55814} -{"mode": "train", "epoch": 1, "iter": 73, "lr": 0.00013, "memory": 49167, "data_time": 0.0801, "loss_cls_0": 0.96393, "loss_box_0": 1.86028, "loss_cns_0": 0.61359, "loss_yns_0": 0.1591, "loss_cls_1": 1.02139, "loss_box_1": 2.07044, "loss_cns_1": 0.61748, "loss_yns_1": 0.16416, "loss_cls_2": 1.05738, "loss_box_2": 1.96651, "loss_cns_2": 0.64175, "loss_yns_2": 0.16329, "loss_cls_3": 1.07542, "loss_box_3": 1.96132, "loss_cns_3": 0.64804, "loss_yns_3": 0.15916, "loss_cls_4": 1.0502, "loss_box_4": 1.95184, "loss_cns_4": 0.64987, "loss_yns_4": 0.16626, "loss_cls_5": 1.08985, "loss_box_5": 1.94781, "loss_cns_5": 0.64798, "loss_yns_5": 0.1595, "loss_cls_dn_0": 0.36868, "loss_box_dn_0": 0.86516, "loss_cls_dn_1": 0.23714, "loss_box_dn_1": 0.87018, "loss_cls_dn_2": 0.26445, "loss_box_dn_2": 0.84532, "loss_cls_dn_3": 0.27572, "loss_box_dn_3": 0.8495, "loss_cls_dn_4": 0.28195, "loss_box_dn_4": 0.84828, "loss_cls_dn_5": 0.29715, "loss_box_dn_5": 0.86329, "loss_dense_depth": 0.86093, "loss": 30.5343, "grad_norm": 48.84884, "time": 1.57245} -{"mode": "train", "epoch": 1, "iter": 74, "lr": 0.00013, "memory": 49167, "data_time": 0.08463, "loss_cls_0": 0.95325, "loss_box_0": 1.85096, "loss_cns_0": 0.61578, "loss_yns_0": 0.1584, "loss_cls_1": 1.03128, "loss_box_1": 2.07619, "loss_cns_1": 0.62748, "loss_yns_1": 0.16224, "loss_cls_2": 1.05772, "loss_box_2": 2.00105, "loss_cns_2": 0.64902, "loss_yns_2": 0.16375, "loss_cls_3": 1.05721, "loss_box_3": 2.0096, "loss_cns_3": 0.65605, "loss_yns_3": 0.16246, "loss_cls_4": 1.05434, "loss_box_4": 2.03397, "loss_cns_4": 0.65443, "loss_yns_4": 0.16286, "loss_cls_5": 1.06535, "loss_box_5": 2.00247, "loss_cns_5": 0.65044, "loss_yns_5": 0.16478, "loss_cls_dn_0": 0.37727, "loss_box_dn_0": 0.85769, "loss_cls_dn_1": 0.23082, "loss_box_dn_1": 0.8823, "loss_cls_dn_2": 0.26165, "loss_box_dn_2": 0.8698, "loss_cls_dn_3": 0.28165, "loss_box_dn_3": 0.87402, "loss_cls_dn_4": 0.29641, "loss_box_dn_4": 0.88855, "loss_cls_dn_5": 0.30085, "loss_box_dn_5": 0.90234, "loss_dense_depth": 0.85793, "loss": 30.90233, "grad_norm": 46.40733, "time": 1.57159} -{"mode": "train", "epoch": 1, "iter": 75, "lr": 0.00013, "memory": 49167, "data_time": 0.08194, "loss_cls_0": 0.94866, "loss_box_0": 1.84008, "loss_cns_0": 0.61377, "loss_yns_0": 0.15954, "loss_cls_1": 1.03079, "loss_box_1": 2.10776, "loss_cns_1": 0.63452, "loss_yns_1": 0.16314, "loss_cls_2": 1.06156, "loss_box_2": 2.06833, "loss_cns_2": 0.64487, "loss_yns_2": 0.1627, "loss_cls_3": 1.06754, "loss_box_3": 2.03861, "loss_cns_3": 0.65247, "loss_yns_3": 0.16424, "loss_cls_4": 1.06552, "loss_box_4": 2.04027, "loss_cns_4": 0.65422, "loss_yns_4": 0.16149, "loss_cls_5": 1.05773, "loss_box_5": 2.01517, "loss_cns_5": 0.65025, "loss_yns_5": 0.165, "loss_cls_dn_0": 0.36095, "loss_box_dn_0": 0.84347, "loss_cls_dn_1": 0.22584, "loss_box_dn_1": 0.89565, "loss_cls_dn_2": 0.25491, "loss_box_dn_2": 0.89056, "loss_cls_dn_3": 0.27942, "loss_box_dn_3": 0.87711, "loss_cls_dn_4": 0.29078, "loss_box_dn_4": 0.88988, "loss_cls_dn_5": 0.29564, "loss_box_dn_5": 0.90359, "loss_dense_depth": 0.82648, "loss": 31.00249, "grad_norm": 45.06128, "time": 1.56349} -{"mode": "train", "epoch": 1, "iter": 76, "lr": 0.00013, "memory": 49167, "data_time": 0.08421, "loss_cls_0": 0.96407, "loss_box_0": 1.83543, "loss_cns_0": 0.61913, "loss_yns_0": 0.16169, "loss_cls_1": 1.02218, "loss_box_1": 2.15129, "loss_cns_1": 0.63405, "loss_yns_1": 0.1652, "loss_cls_2": 1.05747, "loss_box_2": 2.10597, "loss_cns_2": 0.64505, "loss_yns_2": 0.16293, "loss_cls_3": 1.06616, "loss_box_3": 2.07335, "loss_cns_3": 0.65129, "loss_yns_3": 0.16385, "loss_cls_4": 1.05511, "loss_box_4": 2.0424, "loss_cns_4": 0.65611, "loss_yns_4": 0.16761, "loss_cls_5": 1.08103, "loss_box_5": 2.02729, "loss_cns_5": 0.65283, "loss_yns_5": 0.16407, "loss_cls_dn_0": 0.33903, "loss_box_dn_0": 0.8443, "loss_cls_dn_1": 0.22804, "loss_box_dn_1": 0.86356, "loss_cls_dn_2": 0.24847, "loss_box_dn_2": 0.85147, "loss_cls_dn_3": 0.2622, "loss_box_dn_3": 0.83576, "loss_cls_dn_4": 0.26169, "loss_box_dn_4": 0.84028, "loss_cls_dn_5": 0.27802, "loss_box_dn_5": 0.8445, "loss_dense_depth": 0.82398, "loss": 30.84685, "grad_norm": 40.65885, "time": 1.55682} -{"mode": "train", "epoch": 1, "iter": 77, "lr": 0.00013, "memory": 49167, "data_time": 0.08454, "loss_cls_0": 0.96113, "loss_box_0": 1.87487, "loss_cns_0": 0.61569, "loss_yns_0": 0.16011, "loss_cls_1": 1.0454, "loss_box_1": 2.10127, "loss_cns_1": 0.64125, "loss_yns_1": 0.16496, "loss_cls_2": 1.04938, "loss_box_2": 2.05342, "loss_cns_2": 0.65234, "loss_yns_2": 0.16204, "loss_cls_3": 1.07022, "loss_box_3": 2.03958, "loss_cns_3": 0.653, "loss_yns_3": 0.1618, "loss_cls_4": 1.06905, "loss_box_4": 2.01019, "loss_cns_4": 0.65694, "loss_yns_4": 0.16754, "loss_cls_5": 1.05942, "loss_box_5": 2.00318, "loss_cns_5": 0.65427, "loss_yns_5": 0.16328, "loss_cls_dn_0": 0.33787, "loss_box_dn_0": 0.83319, "loss_cls_dn_1": 0.22876, "loss_box_dn_1": 0.81475, "loss_cls_dn_2": 0.24394, "loss_box_dn_2": 0.79672, "loss_cls_dn_3": 0.25688, "loss_box_dn_3": 0.7859, "loss_cls_dn_4": 0.25037, "loss_box_dn_4": 0.78699, "loss_cls_dn_5": 0.28041, "loss_box_dn_5": 0.78995, "loss_dense_depth": 0.86624, "loss": 30.46233, "grad_norm": 45.8695, "time": 1.57098} -{"mode": "train", "epoch": 1, "iter": 78, "lr": 0.00013, "memory": 49167, "data_time": 0.08285, "loss_cls_0": 0.97103, "loss_box_0": 1.87328, "loss_cns_0": 0.61877, "loss_yns_0": 0.16294, "loss_cls_1": 1.02738, "loss_box_1": 2.15064, "loss_cns_1": 0.63067, "loss_yns_1": 0.16514, "loss_cls_2": 1.04534, "loss_box_2": 2.08237, "loss_cns_2": 0.64503, "loss_yns_2": 0.16362, "loss_cls_3": 1.05562, "loss_box_3": 2.08539, "loss_cns_3": 0.65179, "loss_yns_3": 0.16472, "loss_cls_4": 1.05616, "loss_box_4": 2.06863, "loss_cns_4": 0.6511, "loss_yns_4": 0.168, "loss_cls_5": 1.06732, "loss_box_5": 2.06967, "loss_cns_5": 0.65014, "loss_yns_5": 0.16587, "loss_cls_dn_0": 0.35324, "loss_box_dn_0": 0.82502, "loss_cls_dn_1": 0.21602, "loss_box_dn_1": 0.81154, "loss_cls_dn_2": 0.23832, "loss_box_dn_2": 0.78786, "loss_cls_dn_3": 0.25316, "loss_box_dn_3": 0.79486, "loss_cls_dn_4": 0.24395, "loss_box_dn_4": 0.80025, "loss_cls_dn_5": 0.28676, "loss_box_dn_5": 0.80873, "loss_dense_depth": 0.84617, "loss": 30.6565, "grad_norm": 36.65675, "time": 1.56672} -{"mode": "train", "epoch": 1, "iter": 79, "lr": 0.00013, "memory": 49167, "data_time": 0.07989, "loss_cls_0": 0.93162, "loss_box_0": 1.85607, "loss_cns_0": 0.62589, "loss_yns_0": 0.16055, "loss_cls_1": 1.00997, "loss_box_1": 2.07427, "loss_cns_1": 0.63774, "loss_yns_1": 0.16297, "loss_cls_2": 1.02929, "loss_box_2": 2.0562, "loss_cns_2": 0.65058, "loss_yns_2": 0.16215, "loss_cls_3": 1.0364, "loss_box_3": 2.03858, "loss_cns_3": 0.65574, "loss_yns_3": 0.16385, "loss_cls_4": 1.04367, "loss_box_4": 2.0152, "loss_cns_4": 0.65253, "loss_yns_4": 0.16346, "loss_cls_5": 1.03897, "loss_box_5": 2.00484, "loss_cns_5": 0.65228, "loss_yns_5": 0.16566, "loss_cls_dn_0": 0.34018, "loss_box_dn_0": 0.82678, "loss_cls_dn_1": 0.21102, "loss_box_dn_1": 0.82067, "loss_cls_dn_2": 0.23244, "loss_box_dn_2": 0.81264, "loss_cls_dn_3": 0.2529, "loss_box_dn_3": 0.82029, "loss_cls_dn_4": 0.23563, "loss_box_dn_4": 0.83059, "loss_cls_dn_5": 0.26872, "loss_box_dn_5": 0.84082, "loss_dense_depth": 0.83127, "loss": 30.31244, "grad_norm": 47.27075, "time": 1.5662} -{"mode": "train", "epoch": 1, "iter": 80, "lr": 0.00013, "memory": 49167, "data_time": 0.0796, "loss_cls_0": 0.93075, "loss_box_0": 1.85307, "loss_cns_0": 0.62588, "loss_yns_0": 0.15982, "loss_cls_1": 0.99898, "loss_box_1": 2.0677, "loss_cns_1": 0.64128, "loss_yns_1": 0.16113, "loss_cls_2": 1.06298, "loss_box_2": 2.06338, "loss_cns_2": 0.65044, "loss_yns_2": 0.16151, "loss_cls_3": 1.05805, "loss_box_3": 2.01281, "loss_cns_3": 0.65464, "loss_yns_3": 0.16131, "loss_cls_4": 1.06592, "loss_box_4": 2.0061, "loss_cns_4": 0.65352, "loss_yns_4": 0.16987, "loss_cls_5": 1.07491, "loss_box_5": 1.99256, "loss_cns_5": 0.65155, "loss_yns_5": 0.16225, "loss_cls_dn_0": 0.33111, "loss_box_dn_0": 0.82421, "loss_cls_dn_1": 0.20682, "loss_box_dn_1": 0.84542, "loss_cls_dn_2": 0.22937, "loss_box_dn_2": 0.84527, "loss_cls_dn_3": 0.24893, "loss_box_dn_3": 0.83284, "loss_cls_dn_4": 0.22911, "loss_box_dn_4": 0.85215, "loss_cls_dn_5": 0.24968, "loss_box_dn_5": 0.8652, "loss_dense_depth": 0.80209, "loss": 30.40263, "grad_norm": 55.65752, "time": 1.55688} -{"mode": "train", "epoch": 1, "iter": 81, "lr": 0.00013, "memory": 49167, "data_time": 0.09291, "loss_cls_0": 0.94704, "loss_box_0": 1.87641, "loss_cns_0": 0.62107, "loss_yns_0": 0.16192, "loss_cls_1": 1.01226, "loss_box_1": 2.02772, "loss_cns_1": 0.6403, "loss_yns_1": 0.16068, "loss_cls_2": 1.0587, "loss_box_2": 2.01515, "loss_cns_2": 0.64742, "loss_yns_2": 0.16227, "loss_cls_3": 1.06312, "loss_box_3": 1.99475, "loss_cns_3": 0.65118, "loss_yns_3": 0.16134, "loss_cls_4": 1.10714, "loss_box_4": 1.99078, "loss_cns_4": 0.64732, "loss_yns_4": 0.17622, "loss_cls_5": 1.05118, "loss_box_5": 1.99664, "loss_cns_5": 0.64927, "loss_yns_5": 0.16223, "loss_cls_dn_0": 0.33786, "loss_box_dn_0": 0.82724, "loss_cls_dn_1": 0.20425, "loss_box_dn_1": 0.87822, "loss_cls_dn_2": 0.22127, "loss_box_dn_2": 0.86973, "loss_cls_dn_3": 0.23707, "loss_box_dn_3": 0.86234, "loss_cls_dn_4": 0.23194, "loss_box_dn_4": 0.88771, "loss_cls_dn_5": 0.24331, "loss_box_dn_5": 0.89371, "loss_dense_depth": 0.86419, "loss": 30.54097, "grad_norm": 43.95145, "time": 1.6225} -{"mode": "train", "epoch": 1, "iter": 82, "lr": 0.00013, "memory": 49167, "data_time": 0.07313, "loss_cls_0": 0.95727, "loss_box_0": 1.85105, "loss_cns_0": 0.61973, "loss_yns_0": 0.15706, "loss_cls_1": 1.01136, "loss_box_1": 2.06121, "loss_cns_1": 0.63641, "loss_yns_1": 0.15822, "loss_cls_2": 1.05963, "loss_box_2": 2.02372, "loss_cns_2": 0.64833, "loss_yns_2": 0.15919, "loss_cls_3": 1.08293, "loss_box_3": 2.05411, "loss_cns_3": 0.65363, "loss_yns_3": 0.1576, "loss_cls_4": 1.12672, "loss_box_4": 2.05821, "loss_cns_4": 0.64722, "loss_yns_4": 0.16364, "loss_cls_5": 1.06346, "loss_box_5": 2.03696, "loss_cns_5": 0.6494, "loss_yns_5": 0.15793, "loss_cls_dn_0": 0.3411, "loss_box_dn_0": 0.83441, "loss_cls_dn_1": 0.19949, "loss_box_dn_1": 0.88909, "loss_cls_dn_2": 0.21451, "loss_box_dn_2": 0.8755, "loss_cls_dn_3": 0.22732, "loss_box_dn_3": 0.88761, "loss_cls_dn_4": 0.22757, "loss_box_dn_4": 0.90385, "loss_cls_dn_5": 0.23531, "loss_box_dn_5": 0.90038, "loss_dense_depth": 0.90776, "loss": 30.83891, "grad_norm": 45.29645, "time": 1.63493} -{"mode": "train", "epoch": 1, "iter": 83, "lr": 0.00013, "memory": 49167, "data_time": 0.1357, "loss_cls_0": 0.95159, "loss_box_0": 1.80033, "loss_cns_0": 0.62613, "loss_yns_0": 0.1594, "loss_cls_1": 1.0274, "loss_box_1": 1.98188, "loss_cns_1": 0.63256, "loss_yns_1": 0.15911, "loss_cls_2": 1.03934, "loss_box_2": 1.93749, "loss_cns_2": 0.645, "loss_yns_2": 0.15858, "loss_cls_3": 1.05397, "loss_box_3": 1.95346, "loss_cns_3": 0.65172, "loss_yns_3": 0.15889, "loss_cls_4": 1.06333, "loss_box_4": 1.94012, "loss_cns_4": 0.65194, "loss_yns_4": 0.15978, "loss_cls_5": 1.07441, "loss_box_5": 1.94218, "loss_cns_5": 0.64766, "loss_yns_5": 0.15954, "loss_cls_dn_0": 0.32904, "loss_box_dn_0": 0.83408, "loss_cls_dn_1": 0.2018, "loss_box_dn_1": 0.8485, "loss_cls_dn_2": 0.21819, "loss_box_dn_2": 0.8349, "loss_cls_dn_3": 0.21963, "loss_box_dn_3": 0.83944, "loss_cls_dn_4": 0.22161, "loss_box_dn_4": 0.84067, "loss_cls_dn_5": 0.23627, "loss_box_dn_5": 0.85016, "loss_dense_depth": 0.84281, "loss": 29.89293, "grad_norm": 53.75653, "time": 1.60798} -{"mode": "train", "epoch": 1, "iter": 84, "lr": 0.00013, "memory": 49167, "data_time": 0.07298, "loss_cls_0": 0.90271, "loss_box_0": 1.79188, "loss_cns_0": 0.62694, "loss_yns_0": 0.1558, "loss_cls_1": 1.0404, "loss_box_1": 1.93647, "loss_cns_1": 0.64384, "loss_yns_1": 0.15905, "loss_cls_2": 1.05925, "loss_box_2": 1.9153, "loss_cns_2": 0.65003, "loss_yns_2": 0.1571, "loss_cls_3": 1.05879, "loss_box_3": 1.90985, "loss_cns_3": 0.65299, "loss_yns_3": 0.15841, "loss_cls_4": 1.07682, "loss_box_4": 1.90302, "loss_cns_4": 0.65548, "loss_yns_4": 0.16699, "loss_cls_5": 1.03737, "loss_box_5": 1.90798, "loss_cns_5": 0.65182, "loss_yns_5": 0.15968, "loss_cls_dn_0": 0.31903, "loss_box_dn_0": 0.82377, "loss_cls_dn_1": 0.20364, "loss_box_dn_1": 0.79042, "loss_cls_dn_2": 0.22108, "loss_box_dn_2": 0.78035, "loss_cls_dn_3": 0.2213, "loss_box_dn_3": 0.7734, "loss_cls_dn_4": 0.22285, "loss_box_dn_4": 0.77998, "loss_cls_dn_5": 0.24083, "loss_box_dn_5": 0.79728, "loss_dense_depth": 0.8411, "loss": 29.39299, "grad_norm": 59.77712, "time": 1.53932} -{"mode": "train", "epoch": 1, "iter": 85, "lr": 0.00013, "memory": 49167, "data_time": 0.10511, "loss_cls_0": 0.9407, "loss_box_0": 1.82652, "loss_cns_0": 0.61961, "loss_yns_0": 0.15685, "loss_cls_1": 1.00711, "loss_box_1": 1.94688, "loss_cns_1": 0.64218, "loss_yns_1": 0.16018, "loss_cls_2": 1.04447, "loss_box_2": 1.93295, "loss_cns_2": 0.64671, "loss_yns_2": 0.15844, "loss_cls_3": 1.04838, "loss_box_3": 1.95068, "loss_cns_3": 0.64789, "loss_yns_3": 0.161, "loss_cls_4": 1.06361, "loss_box_4": 1.94672, "loss_cns_4": 0.65349, "loss_yns_4": 0.16864, "loss_cls_5": 1.11493, "loss_box_5": 1.93394, "loss_cns_5": 0.64791, "loss_yns_5": 0.1568, "loss_cls_dn_0": 0.34994, "loss_box_dn_0": 0.84244, "loss_cls_dn_1": 0.20751, "loss_box_dn_1": 0.8003, "loss_cls_dn_2": 0.22861, "loss_box_dn_2": 0.79285, "loss_cls_dn_3": 0.23561, "loss_box_dn_3": 0.80396, "loss_cls_dn_4": 0.2345, "loss_box_dn_4": 0.81868, "loss_cls_dn_5": 0.25992, "loss_box_dn_5": 0.8188, "loss_dense_depth": 0.84149, "loss": 29.81119, "grad_norm": 52.50562, "time": 1.56778} -{"mode": "train", "epoch": 1, "iter": 86, "lr": 0.00013, "memory": 49167, "data_time": 0.07925, "loss_cls_0": 0.92283, "loss_box_0": 1.80699, "loss_cns_0": 0.62227, "loss_yns_0": 0.16044, "loss_cls_1": 1.04425, "loss_box_1": 1.88288, "loss_cns_1": 0.62751, "loss_yns_1": 0.15842, "loss_cls_2": 1.07053, "loss_box_2": 1.87659, "loss_cns_2": 0.64069, "loss_yns_2": 0.16217, "loss_cls_3": 1.04071, "loss_box_3": 1.90145, "loss_cns_3": 0.64705, "loss_yns_3": 0.16093, "loss_cls_4": 1.0665, "loss_box_4": 1.89844, "loss_cns_4": 0.6506, "loss_yns_4": 0.16448, "loss_cls_5": 1.08179, "loss_box_5": 1.88046, "loss_cns_5": 0.64707, "loss_yns_5": 0.16328, "loss_cls_dn_0": 0.32342, "loss_box_dn_0": 0.82947, "loss_cls_dn_1": 0.2101, "loss_box_dn_1": 0.79797, "loss_cls_dn_2": 0.23048, "loss_box_dn_2": 0.79557, "loss_cls_dn_3": 0.23763, "loss_box_dn_3": 0.80699, "loss_cls_dn_4": 0.24316, "loss_box_dn_4": 0.81803, "loss_cls_dn_5": 0.25835, "loss_box_dn_5": 0.81623, "loss_dense_depth": 0.86657, "loss": 29.51226, "grad_norm": 49.07616, "time": 1.5825} -{"mode": "train", "epoch": 1, "iter": 87, "lr": 0.00013, "memory": 49167, "data_time": 0.07962, "loss_cls_0": 0.95466, "loss_box_0": 1.79695, "loss_cns_0": 0.62358, "loss_yns_0": 0.16044, "loss_cls_1": 0.9962, "loss_box_1": 1.93895, "loss_cns_1": 0.62882, "loss_yns_1": 0.15941, "loss_cls_2": 1.03172, "loss_box_2": 1.91081, "loss_cns_2": 0.64567, "loss_yns_2": 0.16219, "loss_cls_3": 1.02503, "loss_box_3": 1.8875, "loss_cns_3": 0.65144, "loss_yns_3": 0.15961, "loss_cls_4": 1.02779, "loss_box_4": 1.88787, "loss_cns_4": 0.65245, "loss_yns_4": 0.15902, "loss_cls_5": 1.04908, "loss_box_5": 1.89985, "loss_cns_5": 0.65029, "loss_yns_5": 0.16191, "loss_cls_dn_0": 0.28835, "loss_box_dn_0": 0.81605, "loss_cls_dn_1": 0.20689, "loss_box_dn_1": 0.79904, "loss_cls_dn_2": 0.22275, "loss_box_dn_2": 0.79627, "loss_cls_dn_3": 0.22945, "loss_box_dn_3": 0.78852, "loss_cls_dn_4": 0.22971, "loss_box_dn_4": 0.79355, "loss_cls_dn_5": 0.23418, "loss_box_dn_5": 0.81124, "loss_dense_depth": 0.81949, "loss": 29.25672, "grad_norm": 43.3283, "time": 1.56854} -{"mode": "train", "epoch": 1, "iter": 88, "lr": 0.00013, "memory": 49167, "data_time": 0.08588, "loss_cls_0": 0.95036, "loss_box_0": 1.77848, "loss_cns_0": 0.62065, "loss_yns_0": 0.15782, "loss_cls_1": 0.97783, "loss_box_1": 1.94132, "loss_cns_1": 0.63278, "loss_yns_1": 0.1603, "loss_cls_2": 1.02819, "loss_box_2": 1.91789, "loss_cns_2": 0.64872, "loss_yns_2": 0.15783, "loss_cls_3": 1.07105, "loss_box_3": 1.8851, "loss_cns_3": 0.65309, "loss_yns_3": 0.15723, "loss_cls_4": 1.0569, "loss_box_4": 1.86455, "loss_cns_4": 0.65502, "loss_yns_4": 0.15933, "loss_cls_5": 1.13414, "loss_box_5": 1.88178, "loss_cns_5": 0.65209, "loss_yns_5": 0.15859, "loss_cls_dn_0": 0.29165, "loss_box_dn_0": 0.82253, "loss_cls_dn_1": 0.20171, "loss_box_dn_1": 0.79783, "loss_cls_dn_2": 0.21223, "loss_box_dn_2": 0.8028, "loss_cls_dn_3": 0.22302, "loss_box_dn_3": 0.78858, "loss_cls_dn_4": 0.22337, "loss_box_dn_4": 0.78626, "loss_cls_dn_5": 0.23367, "loss_box_dn_5": 0.81056, "loss_dense_depth": 0.86881, "loss": 29.36407, "grad_norm": 71.40807, "time": 1.5661} -{"mode": "train", "epoch": 1, "iter": 89, "lr": 0.00014, "memory": 49167, "data_time": 0.08544, "loss_cls_0": 0.90805, "loss_box_0": 1.7808, "loss_cns_0": 0.61872, "loss_yns_0": 0.15883, "loss_cls_1": 0.98634, "loss_box_1": 1.92405, "loss_cns_1": 0.63004, "loss_yns_1": 0.15856, "loss_cls_2": 1.02603, "loss_box_2": 1.89287, "loss_cns_2": 0.64672, "loss_yns_2": 0.15978, "loss_cls_3": 1.01983, "loss_box_3": 1.88312, "loss_cns_3": 0.65011, "loss_yns_3": 0.15713, "loss_cls_4": 1.03298, "loss_box_4": 1.86016, "loss_cns_4": 0.6525, "loss_yns_4": 0.15763, "loss_cls_5": 1.0342, "loss_box_5": 1.8953, "loss_cns_5": 0.64813, "loss_yns_5": 0.16278, "loss_cls_dn_0": 0.30864, "loss_box_dn_0": 0.82429, "loss_cls_dn_1": 0.20191, "loss_box_dn_1": 0.81179, "loss_cls_dn_2": 0.20831, "loss_box_dn_2": 0.8099, "loss_cls_dn_3": 0.22233, "loss_box_dn_3": 0.801, "loss_cls_dn_4": 0.22224, "loss_box_dn_4": 0.7951, "loss_cls_dn_5": 0.24093, "loss_box_dn_5": 0.81938, "loss_dense_depth": 0.85153, "loss": 29.16203, "grad_norm": 56.435, "time": 1.5723} -{"mode": "train", "epoch": 1, "iter": 90, "lr": 0.00014, "memory": 49167, "data_time": 0.07134, "loss_cls_0": 0.93066, "loss_box_0": 1.79548, "loss_cns_0": 0.61414, "loss_yns_0": 0.15547, "loss_cls_1": 0.99294, "loss_box_1": 1.92557, "loss_cns_1": 0.63664, "loss_yns_1": 0.15781, "loss_cls_2": 1.02276, "loss_box_2": 1.86617, "loss_cns_2": 0.65266, "loss_yns_2": 0.16054, "loss_cls_3": 1.03375, "loss_box_3": 1.89283, "loss_cns_3": 0.65638, "loss_yns_3": 0.15686, "loss_cls_4": 1.0352, "loss_box_4": 1.92196, "loss_cns_4": 0.65317, "loss_yns_4": 0.16258, "loss_cls_5": 1.05452, "loss_box_5": 1.89898, "loss_cns_5": 0.65252, "loss_yns_5": 0.15903, "loss_cls_dn_0": 0.33119, "loss_box_dn_0": 0.84244, "loss_cls_dn_1": 0.20141, "loss_box_dn_1": 0.81543, "loss_cls_dn_2": 0.20933, "loss_box_dn_2": 0.79244, "loss_cls_dn_3": 0.22204, "loss_box_dn_3": 0.79632, "loss_cls_dn_4": 0.22118, "loss_box_dn_4": 0.81407, "loss_cls_dn_5": 0.24832, "loss_box_dn_5": 0.81189, "loss_dense_depth": 0.83981, "loss": 29.33448, "grad_norm": 45.09066, "time": 1.55834} -{"mode": "train", "epoch": 1, "iter": 91, "lr": 0.00014, "memory": 49167, "data_time": 0.06885, "loss_cls_0": 0.9112, "loss_box_0": 1.75058, "loss_cns_0": 0.61478, "loss_yns_0": 0.15338, "loss_cls_1": 0.9751, "loss_box_1": 1.96131, "loss_cns_1": 0.63648, "loss_yns_1": 0.15639, "loss_cls_2": 1.01475, "loss_box_2": 1.87599, "loss_cns_2": 0.65376, "loss_yns_2": 0.15749, "loss_cls_3": 1.02494, "loss_box_3": 1.87612, "loss_cns_3": 0.6597, "loss_yns_3": 0.15571, "loss_cls_4": 1.05605, "loss_box_4": 1.9191, "loss_cns_4": 0.65832, "loss_yns_4": 0.16081, "loss_cls_5": 1.03266, "loss_box_5": 1.89504, "loss_cns_5": 0.64977, "loss_yns_5": 0.15495, "loss_cls_dn_0": 0.31169, "loss_box_dn_0": 0.81802, "loss_cls_dn_1": 0.19965, "loss_box_dn_1": 0.81193, "loss_cls_dn_2": 0.2108, "loss_box_dn_2": 0.78041, "loss_cls_dn_3": 0.2151, "loss_box_dn_3": 0.78295, "loss_cls_dn_4": 0.21686, "loss_box_dn_4": 0.81519, "loss_cls_dn_5": 0.23628, "loss_box_dn_5": 0.80633, "loss_dense_depth": 0.82407, "loss": 29.13369, "grad_norm": 50.4439, "time": 1.57067} -{"mode": "train", "epoch": 1, "iter": 92, "lr": 0.00014, "memory": 49167, "data_time": 0.08206, "loss_cls_0": 0.91783, "loss_box_0": 1.76111, "loss_cns_0": 0.61759, "loss_yns_0": 0.15172, "loss_cls_1": 0.97545, "loss_box_1": 2.03806, "loss_cns_1": 0.63173, "loss_yns_1": 0.15535, "loss_cls_2": 1.01024, "loss_box_2": 1.95317, "loss_cns_2": 0.64707, "loss_yns_2": 0.15624, "loss_cls_3": 1.03224, "loss_box_3": 1.92942, "loss_cns_3": 0.65306, "loss_yns_3": 0.15721, "loss_cls_4": 1.0602, "loss_box_4": 1.96747, "loss_cns_4": 0.65618, "loss_yns_4": 0.15883, "loss_cls_5": 1.0416, "loss_box_5": 1.94794, "loss_cns_5": 0.64896, "loss_yns_5": 0.15465, "loss_cls_dn_0": 0.29117, "loss_box_dn_0": 0.81198, "loss_cls_dn_1": 0.19751, "loss_box_dn_1": 0.81392, "loss_cls_dn_2": 0.20402, "loss_box_dn_2": 0.79182, "loss_cls_dn_3": 0.21107, "loss_box_dn_3": 0.78801, "loss_cls_dn_4": 0.21517, "loss_box_dn_4": 0.82024, "loss_cls_dn_5": 0.2279, "loss_box_dn_5": 0.81741, "loss_dense_depth": 0.83018, "loss": 29.44374, "grad_norm": 61.37122, "time": 1.56222} -{"mode": "train", "epoch": 1, "iter": 93, "lr": 0.00014, "memory": 49167, "data_time": 0.07684, "loss_cls_0": 0.92522, "loss_box_0": 1.74753, "loss_cns_0": 0.62249, "loss_yns_0": 0.15028, "loss_cls_1": 0.99407, "loss_box_1": 1.98306, "loss_cns_1": 0.63818, "loss_yns_1": 0.15419, "loss_cls_2": 1.01974, "loss_box_2": 1.9263, "loss_cns_2": 0.64996, "loss_yns_2": 0.15739, "loss_cls_3": 1.02926, "loss_box_3": 1.90866, "loss_cns_3": 0.65063, "loss_yns_3": 0.15512, "loss_cls_4": 1.03353, "loss_box_4": 1.90767, "loss_cns_4": 0.65544, "loss_yns_4": 0.15654, "loss_cls_5": 1.03955, "loss_box_5": 1.90431, "loss_cns_5": 0.65264, "loss_yns_5": 0.1554, "loss_cls_dn_0": 0.27937, "loss_box_dn_0": 0.8072, "loss_cls_dn_1": 0.20182, "loss_box_dn_1": 0.81647, "loss_cls_dn_2": 0.20248, "loss_box_dn_2": 0.80508, "loss_cls_dn_3": 0.20854, "loss_box_dn_3": 0.80952, "loss_cls_dn_4": 0.21477, "loss_box_dn_4": 0.82119, "loss_cls_dn_5": 0.2251, "loss_box_dn_5": 0.83126, "loss_dense_depth": 0.81481, "loss": 29.25476, "grad_norm": 42.23354, "time": 1.55712} -{"mode": "train", "epoch": 1, "iter": 94, "lr": 0.00014, "memory": 49167, "data_time": 0.07842, "loss_cls_0": 0.92297, "loss_box_0": 1.78132, "loss_cns_0": 0.62076, "loss_yns_0": 0.15191, "loss_cls_1": 1.00759, "loss_box_1": 1.99227, "loss_cns_1": 0.63899, "loss_yns_1": 0.15553, "loss_cls_2": 1.04276, "loss_box_2": 1.94627, "loss_cns_2": 0.65407, "loss_yns_2": 0.15647, "loss_cls_3": 1.03744, "loss_box_3": 1.9195, "loss_cns_3": 0.65995, "loss_yns_3": 0.15592, "loss_cls_4": 1.07028, "loss_box_4": 1.89346, "loss_cns_4": 0.65931, "loss_yns_4": 0.16367, "loss_cls_5": 1.03881, "loss_box_5": 1.88254, "loss_cns_5": 0.6531, "loss_yns_5": 0.15394, "loss_cls_dn_0": 0.28836, "loss_box_dn_0": 0.80423, "loss_cls_dn_1": 0.20546, "loss_box_dn_1": 0.80611, "loss_cls_dn_2": 0.20972, "loss_box_dn_2": 0.79802, "loss_cls_dn_3": 0.21072, "loss_box_dn_3": 0.80518, "loss_cls_dn_4": 0.2237, "loss_box_dn_4": 0.80306, "loss_cls_dn_5": 0.22598, "loss_box_dn_5": 0.81489, "loss_dense_depth": 0.83106, "loss": 29.3853, "grad_norm": 52.80065, "time": 1.56884} -{"mode": "train", "epoch": 1, "iter": 95, "lr": 0.00014, "memory": 49167, "data_time": 0.07254, "loss_cls_0": 0.91307, "loss_box_0": 1.80081, "loss_cns_0": 0.61825, "loss_yns_0": 0.15214, "loss_cls_1": 1.00134, "loss_box_1": 1.92552, "loss_cns_1": 0.64539, "loss_yns_1": 0.15506, "loss_cls_2": 1.03091, "loss_box_2": 1.89975, "loss_cns_2": 0.65791, "loss_yns_2": 0.15478, "loss_cls_3": 1.04531, "loss_box_3": 1.89447, "loss_cns_3": 0.66234, "loss_yns_3": 0.15562, "loss_cls_4": 1.06695, "loss_box_4": 1.89034, "loss_cns_4": 0.6598, "loss_yns_4": 0.16304, "loss_cls_5": 1.05033, "loss_box_5": 1.8521, "loss_cns_5": 0.65401, "loss_yns_5": 0.15552, "loss_cls_dn_0": 0.28575, "loss_box_dn_0": 0.8084, "loss_cls_dn_1": 0.19434, "loss_box_dn_1": 0.79242, "loss_cls_dn_2": 0.20261, "loss_box_dn_2": 0.78736, "loss_cls_dn_3": 0.20632, "loss_box_dn_3": 0.80104, "loss_cls_dn_4": 0.21633, "loss_box_dn_4": 0.80407, "loss_cls_dn_5": 0.21737, "loss_box_dn_5": 0.8042, "loss_dense_depth": 0.85101, "loss": 29.17598, "grad_norm": 55.74363, "time": 1.55458} -{"mode": "train", "epoch": 1, "iter": 96, "lr": 0.00014, "memory": 49167, "data_time": 0.07446, "loss_cls_0": 0.94351, "loss_box_0": 1.83591, "loss_cns_0": 0.61801, "loss_yns_0": 0.15628, "loss_cls_1": 1.0092, "loss_box_1": 1.8896, "loss_cns_1": 0.64053, "loss_yns_1": 0.15719, "loss_cls_2": 1.03594, "loss_box_2": 1.85616, "loss_cns_2": 0.64986, "loss_yns_2": 0.15756, "loss_cls_3": 1.08611, "loss_box_3": 1.88952, "loss_cns_3": 0.65359, "loss_yns_3": 0.15568, "loss_cls_4": 1.06204, "loss_box_4": 1.86086, "loss_cns_4": 0.65539, "loss_yns_4": 0.15783, "loss_cls_5": 1.08986, "loss_box_5": 1.85429, "loss_cns_5": 0.65298, "loss_yns_5": 0.15932, "loss_cls_dn_0": 0.28701, "loss_box_dn_0": 0.81964, "loss_cls_dn_1": 0.19198, "loss_box_dn_1": 0.79149, "loss_cls_dn_2": 0.20217, "loss_box_dn_2": 0.78008, "loss_cls_dn_3": 0.21057, "loss_box_dn_3": 0.80057, "loss_cls_dn_4": 0.21249, "loss_box_dn_4": 0.79608, "loss_cls_dn_5": 0.22159, "loss_box_dn_5": 0.80803, "loss_dense_depth": 0.84349, "loss": 29.1924, "grad_norm": 48.80716, "time": 1.61007} -{"mode": "train", "epoch": 1, "iter": 97, "lr": 0.00014, "memory": 49167, "data_time": 0.07749, "loss_cls_0": 0.94313, "loss_box_0": 1.79975, "loss_cns_0": 0.62224, "loss_yns_0": 0.15376, "loss_cls_1": 1.02568, "loss_box_1": 1.83271, "loss_cns_1": 0.64421, "loss_yns_1": 0.15691, "loss_cls_2": 1.03401, "loss_box_2": 1.79862, "loss_cns_2": 0.65081, "loss_yns_2": 0.15866, "loss_cls_3": 1.06205, "loss_box_3": 1.81894, "loss_cns_3": 0.65595, "loss_yns_3": 0.15683, "loss_cls_4": 1.04869, "loss_box_4": 1.78325, "loss_cns_4": 0.65906, "loss_yns_4": 0.15674, "loss_cls_5": 1.04771, "loss_box_5": 1.79717, "loss_cns_5": 0.65753, "loss_yns_5": 0.15676, "loss_cls_dn_0": 0.28895, "loss_box_dn_0": 0.81788, "loss_cls_dn_1": 0.19046, "loss_box_dn_1": 0.78551, "loss_cls_dn_2": 0.19723, "loss_box_dn_2": 0.77241, "loss_cls_dn_3": 0.20653, "loss_box_dn_3": 0.78707, "loss_cls_dn_4": 0.20661, "loss_box_dn_4": 0.77811, "loss_cls_dn_5": 0.21469, "loss_box_dn_5": 0.79585, "loss_dense_depth": 0.84575, "loss": 28.70821, "grad_norm": 43.64354, "time": 1.57721} -{"mode": "train", "epoch": 1, "iter": 98, "lr": 0.00014, "memory": 49167, "data_time": 0.07932, "loss_cls_0": 0.91835, "loss_box_0": 1.75475, "loss_cns_0": 0.6224, "loss_yns_0": 0.15317, "loss_cls_1": 0.991, "loss_box_1": 1.83828, "loss_cns_1": 0.64763, "loss_yns_1": 0.15632, "loss_cls_2": 1.01272, "loss_box_2": 1.79444, "loss_cns_2": 0.65631, "loss_yns_2": 0.15534, "loss_cls_3": 1.0432, "loss_box_3": 1.80047, "loss_cns_3": 0.66074, "loss_yns_3": 0.1591, "loss_cls_4": 1.03279, "loss_box_4": 1.82127, "loss_cns_4": 0.66099, "loss_yns_4": 0.16145, "loss_cls_5": 1.08842, "loss_box_5": 1.79519, "loss_cns_5": 0.66159, "loss_yns_5": 0.15856, "loss_cls_dn_0": 0.2863, "loss_box_dn_0": 0.81099, "loss_cls_dn_1": 0.19457, "loss_box_dn_1": 0.7725, "loss_cls_dn_2": 0.20019, "loss_box_dn_2": 0.76162, "loss_cls_dn_3": 0.21062, "loss_box_dn_3": 0.77345, "loss_cls_dn_4": 0.21351, "loss_box_dn_4": 0.78887, "loss_cls_dn_5": 0.23012, "loss_box_dn_5": 0.78926, "loss_dense_depth": 0.80346, "loss": 28.57999, "grad_norm": 43.30529, "time": 1.55607} -{"mode": "train", "epoch": 1, "iter": 99, "lr": 0.00014, "memory": 49167, "data_time": 0.08274, "loss_cls_0": 0.91788, "loss_box_0": 1.7614, "loss_cns_0": 0.61957, "loss_yns_0": 0.15359, "loss_cls_1": 0.99955, "loss_box_1": 1.8742, "loss_cns_1": 0.64012, "loss_yns_1": 0.15464, "loss_cls_2": 1.01937, "loss_box_2": 1.83961, "loss_cns_2": 0.65397, "loss_yns_2": 0.15844, "loss_cls_3": 1.05933, "loss_box_3": 1.81995, "loss_cns_3": 0.65347, "loss_yns_3": 0.16063, "loss_cls_4": 1.04531, "loss_box_4": 1.8598, "loss_cns_4": 0.65291, "loss_yns_4": 0.16559, "loss_cls_5": 1.07403, "loss_box_5": 1.83186, "loss_cns_5": 0.65579, "loss_yns_5": 0.16081, "loss_cls_dn_0": 0.27609, "loss_box_dn_0": 0.80613, "loss_cls_dn_1": 0.19606, "loss_box_dn_1": 0.78806, "loss_cls_dn_2": 0.19982, "loss_box_dn_2": 0.78769, "loss_cls_dn_3": 0.21596, "loss_box_dn_3": 0.79327, "loss_cls_dn_4": 0.21729, "loss_box_dn_4": 0.82157, "loss_cls_dn_5": 0.22805, "loss_box_dn_5": 0.82173, "loss_dense_depth": 0.82818, "loss": 28.91169, "grad_norm": 54.45678, "time": 1.57117} -{"mode": "train", "epoch": 1, "iter": 100, "lr": 0.00014, "memory": 49167, "data_time": 0.08026, "loss_cls_0": 0.93203, "loss_box_0": 1.80123, "loss_cns_0": 0.62126, "loss_yns_0": 0.1555, "loss_cls_1": 0.99054, "loss_box_1": 1.90873, "loss_cns_1": 0.63564, "loss_yns_1": 0.15619, "loss_cls_2": 1.02207, "loss_box_2": 1.88173, "loss_cns_2": 0.64468, "loss_yns_2": 0.16254, "loss_cls_3": 1.04853, "loss_box_3": 1.85791, "loss_cns_3": 0.65094, "loss_yns_3": 0.15921, "loss_cls_4": 1.0378, "loss_box_4": 1.86746, "loss_cns_4": 0.65069, "loss_yns_4": 0.15966, "loss_cls_5": 1.04606, "loss_box_5": 1.89507, "loss_cns_5": 0.65583, "loss_yns_5": 0.15933, "loss_cls_dn_0": 0.26481, "loss_box_dn_0": 0.81475, "loss_cls_dn_1": 0.19297, "loss_box_dn_1": 0.80748, "loss_cls_dn_2": 0.19832, "loss_box_dn_2": 0.79731, "loss_cls_dn_3": 0.20734, "loss_box_dn_3": 0.80328, "loss_cls_dn_4": 0.21152, "loss_box_dn_4": 0.82838, "loss_cls_dn_5": 0.21727, "loss_box_dn_5": 0.85403, "loss_dense_depth": 0.79947, "loss": 29.09755, "grad_norm": 45.50394, "time": 1.56656} -{"mode": "train", "epoch": 1, "iter": 101, "lr": 0.00014, "memory": 49167, "data_time": 0.09715, "loss_cls_0": 0.91878, "loss_box_0": 1.78582, "loss_cns_0": 0.62337, "loss_yns_0": 0.15885, "loss_cls_1": 0.9991, "loss_box_1": 1.95193, "loss_cns_1": 0.64171, "loss_yns_1": 0.15832, "loss_cls_2": 1.0129, "loss_box_2": 1.91941, "loss_cns_2": 0.64728, "loss_yns_2": 0.16404, "loss_cls_3": 1.04691, "loss_box_3": 1.89857, "loss_cns_3": 0.65337, "loss_yns_3": 0.16368, "loss_cls_4": 1.02384, "loss_box_4": 1.90645, "loss_cns_4": 0.65267, "loss_yns_4": 0.16128, "loss_cls_5": 1.03887, "loss_box_5": 1.90779, "loss_cns_5": 0.65586, "loss_yns_5": 0.15912, "loss_cls_dn_0": 0.27115, "loss_box_dn_0": 0.80554, "loss_cls_dn_1": 0.19301, "loss_box_dn_1": 0.82935, "loss_cls_dn_2": 0.1984, "loss_box_dn_2": 0.81574, "loss_cls_dn_3": 0.20302, "loss_box_dn_3": 0.81802, "loss_cls_dn_4": 0.20575, "loss_box_dn_4": 0.8315, "loss_cls_dn_5": 0.21305, "loss_box_dn_5": 0.8499, "loss_dense_depth": 0.81855, "loss": 29.30292, "grad_norm": 44.87988, "time": 1.64504} -{"mode": "train", "epoch": 1, "iter": 102, "lr": 0.00014, "memory": 49167, "data_time": 0.08388, "loss_cls_0": 0.91434, "loss_box_0": 1.81568, "loss_cns_0": 0.61733, "loss_yns_0": 0.15621, "loss_cls_1": 0.99697, "loss_box_1": 1.94418, "loss_cns_1": 0.637, "loss_yns_1": 0.1592, "loss_cls_2": 1.03185, "loss_box_2": 1.90931, "loss_cns_2": 0.64364, "loss_yns_2": 0.15789, "loss_cls_3": 1.03859, "loss_box_3": 1.90294, "loss_cns_3": 0.64829, "loss_yns_3": 0.16063, "loss_cls_4": 1.02494, "loss_box_4": 1.8865, "loss_cns_4": 0.6524, "loss_yns_4": 0.15983, "loss_cls_5": 1.02712, "loss_box_5": 1.86845, "loss_cns_5": 0.64897, "loss_yns_5": 0.15783, "loss_cls_dn_0": 0.26656, "loss_box_dn_0": 0.81337, "loss_cls_dn_1": 0.19132, "loss_box_dn_1": 0.82541, "loss_cls_dn_2": 0.20367, "loss_box_dn_2": 0.81965, "loss_cls_dn_3": 0.21026, "loss_box_dn_3": 0.82236, "loss_cls_dn_4": 0.20745, "loss_box_dn_4": 0.82691, "loss_cls_dn_5": 0.21195, "loss_box_dn_5": 0.83871, "loss_dense_depth": 0.80129, "loss": 29.19901, "grad_norm": 60.00145, "time": 1.6678} -{"mode": "train", "epoch": 1, "iter": 103, "lr": 0.00014, "memory": 49167, "data_time": 0.12854, "loss_cls_0": 0.92374, "loss_box_0": 1.77963, "loss_cns_0": 0.6235, "loss_yns_0": 0.15827, "loss_cls_1": 0.96039, "loss_box_1": 1.92339, "loss_cns_1": 0.63586, "loss_yns_1": 0.16142, "loss_cls_2": 0.99652, "loss_box_2": 1.86899, "loss_cns_2": 0.64946, "loss_yns_2": 0.16222, "loss_cls_3": 1.01226, "loss_box_3": 1.8455, "loss_cns_3": 0.65643, "loss_yns_3": 0.16092, "loss_cls_4": 1.00372, "loss_box_4": 1.84258, "loss_cns_4": 0.65658, "loss_yns_4": 0.15957, "loss_cls_5": 1.01734, "loss_box_5": 1.8313, "loss_cns_5": 0.65777, "loss_yns_5": 0.15815, "loss_cls_dn_0": 0.25287, "loss_box_dn_0": 0.80191, "loss_cls_dn_1": 0.18127, "loss_box_dn_1": 0.82488, "loss_cls_dn_2": 0.19382, "loss_box_dn_2": 0.80255, "loss_cls_dn_3": 0.20166, "loss_box_dn_3": 0.80505, "loss_cls_dn_4": 0.19889, "loss_box_dn_4": 0.81668, "loss_cls_dn_5": 0.206, "loss_box_dn_5": 0.82717, "loss_dense_depth": 0.81817, "loss": 28.77646, "grad_norm": 44.80261, "time": 1.59534} -{"mode": "train", "epoch": 1, "iter": 104, "lr": 0.00014, "memory": 49167, "data_time": 0.07534, "loss_cls_0": 0.90124, "loss_box_0": 1.80314, "loss_cns_0": 0.61444, "loss_yns_0": 0.15358, "loss_cls_1": 0.96307, "loss_box_1": 1.91243, "loss_cns_1": 0.63842, "loss_yns_1": 0.15841, "loss_cls_2": 0.99258, "loss_box_2": 1.85884, "loss_cns_2": 0.64786, "loss_yns_2": 0.16341, "loss_cls_3": 1.06167, "loss_box_3": 1.85726, "loss_cns_3": 0.65952, "loss_yns_3": 0.1594, "loss_cls_4": 1.01846, "loss_box_4": 1.89368, "loss_cns_4": 0.65462, "loss_yns_4": 0.1582, "loss_cls_5": 1.02007, "loss_box_5": 1.85362, "loss_cns_5": 0.6722, "loss_yns_5": 0.1568, "loss_cls_dn_0": 0.25905, "loss_box_dn_0": 0.82436, "loss_cls_dn_1": 0.17575, "loss_box_dn_1": 0.81534, "loss_cls_dn_2": 0.18204, "loss_box_dn_2": 0.79616, "loss_cls_dn_3": 0.18918, "loss_box_dn_3": 0.81329, "loss_cls_dn_4": 0.18839, "loss_box_dn_4": 0.84121, "loss_cls_dn_5": 0.20319, "loss_box_dn_5": 0.84182, "loss_dense_depth": 0.79742, "loss": 28.90012, "grad_norm": 49.44173, "time": 1.55766} -{"mode": "train", "epoch": 1, "iter": 105, "lr": 0.00014, "memory": 49167, "data_time": 0.10607, "loss_cls_0": 0.89355, "loss_box_0": 1.72725, "loss_cns_0": 0.61641, "loss_yns_0": 0.15221, "loss_cls_1": 0.95716, "loss_box_1": 1.8774, "loss_cns_1": 0.64714, "loss_yns_1": 0.15718, "loss_cls_2": 0.98782, "loss_box_2": 1.82524, "loss_cns_2": 0.6557, "loss_yns_2": 0.16158, "loss_cls_3": 1.02996, "loss_box_3": 1.8227, "loss_cns_3": 0.66501, "loss_yns_3": 0.16175, "loss_cls_4": 1.01424, "loss_box_4": 1.8277, "loss_cns_4": 0.6632, "loss_yns_4": 0.15799, "loss_cls_5": 1.01476, "loss_box_5": 1.83568, "loss_cns_5": 0.67157, "loss_yns_5": 0.15697, "loss_cls_dn_0": 0.2598, "loss_box_dn_0": 0.81075, "loss_cls_dn_1": 0.17413, "loss_box_dn_1": 0.82984, "loss_cls_dn_2": 0.18189, "loss_box_dn_2": 0.81401, "loss_cls_dn_3": 0.19027, "loss_box_dn_3": 0.8331, "loss_cls_dn_4": 0.19234, "loss_box_dn_4": 0.85264, "loss_cls_dn_5": 0.20847, "loss_box_dn_5": 0.86796, "loss_dense_depth": 0.77793, "loss": 28.6733, "grad_norm": 58.72005, "time": 1.56444} -{"mode": "train", "epoch": 1, "iter": 106, "lr": 0.00014, "memory": 49167, "data_time": 0.08133, "loss_cls_0": 0.91298, "loss_box_0": 1.71407, "loss_cns_0": 0.60689, "loss_yns_0": 0.1519, "loss_cls_1": 0.95031, "loss_box_1": 1.92388, "loss_cns_1": 0.64201, "loss_yns_1": 0.15723, "loss_cls_2": 0.98972, "loss_box_2": 1.87852, "loss_cns_2": 0.6511, "loss_yns_2": 0.15713, "loss_cls_3": 1.00663, "loss_box_3": 1.87796, "loss_cns_3": 0.65246, "loss_yns_3": 0.15958, "loss_cls_4": 1.00462, "loss_box_4": 1.85435, "loss_cns_4": 0.65549, "loss_yns_4": 0.15654, "loss_cls_5": 1.01531, "loss_box_5": 1.91875, "loss_cns_5": 0.64423, "loss_yns_5": 0.15806, "loss_cls_dn_0": 0.25819, "loss_box_dn_0": 0.79791, "loss_cls_dn_1": 0.17451, "loss_box_dn_1": 0.81882, "loss_cls_dn_2": 0.1822, "loss_box_dn_2": 0.81825, "loss_cls_dn_3": 0.19711, "loss_box_dn_3": 0.83547, "loss_cls_dn_4": 0.19206, "loss_box_dn_4": 0.84575, "loss_cls_dn_5": 0.20351, "loss_box_dn_5": 0.88178, "loss_dense_depth": 0.80669, "loss": 28.85199, "grad_norm": 58.17303, "time": 1.56044} -{"mode": "train", "epoch": 1, "iter": 107, "lr": 0.00014, "memory": 49167, "data_time": 0.07949, "loss_cls_0": 0.88849, "loss_box_0": 1.74168, "loss_cns_0": 0.61884, "loss_yns_0": 0.15636, "loss_cls_1": 0.94105, "loss_box_1": 1.90159, "loss_cns_1": 0.63629, "loss_yns_1": 0.15863, "loss_cls_2": 0.98262, "loss_box_2": 1.83609, "loss_cns_2": 0.65202, "loss_yns_2": 0.15689, "loss_cls_3": 1.03417, "loss_box_3": 1.8417, "loss_cns_3": 0.65467, "loss_yns_3": 0.15629, "loss_cls_4": 0.99694, "loss_box_4": 1.84853, "loss_cns_4": 0.6565, "loss_yns_4": 0.15691, "loss_cls_5": 1.01211, "loss_box_5": 1.86633, "loss_cns_5": 0.65361, "loss_yns_5": 0.15634, "loss_cls_dn_0": 0.24872, "loss_box_dn_0": 0.81363, "loss_cls_dn_1": 0.17799, "loss_box_dn_1": 0.81858, "loss_cls_dn_2": 0.18405, "loss_box_dn_2": 0.81776, "loss_cls_dn_3": 0.20389, "loss_box_dn_3": 0.83868, "loss_cls_dn_4": 0.19529, "loss_box_dn_4": 0.85195, "loss_cls_dn_5": 0.20862, "loss_box_dn_5": 0.87335, "loss_dense_depth": 0.79523, "loss": 28.73239, "grad_norm": 41.53421, "time": 1.56882} -{"mode": "train", "epoch": 1, "iter": 108, "lr": 0.00014, "memory": 49167, "data_time": 0.09264, "loss_cls_0": 0.91594, "loss_box_0": 1.76197, "loss_cns_0": 0.62484, "loss_yns_0": 0.15766, "loss_cls_1": 0.94203, "loss_box_1": 1.87664, "loss_cns_1": 0.63397, "loss_yns_1": 0.15726, "loss_cls_2": 0.97988, "loss_box_2": 1.84573, "loss_cns_2": 0.64842, "loss_yns_2": 0.16069, "loss_cls_3": 1.0313, "loss_box_3": 1.87017, "loss_cns_3": 0.65361, "loss_yns_3": 0.15859, "loss_cls_4": 1.01237, "loss_box_4": 1.89966, "loss_cns_4": 0.65263, "loss_yns_4": 0.1584, "loss_cls_5": 1.04197, "loss_box_5": 1.89229, "loss_cns_5": 0.6538, "loss_yns_5": 0.15872, "loss_cls_dn_0": 0.24302, "loss_box_dn_0": 0.80937, "loss_cls_dn_1": 0.18058, "loss_box_dn_1": 0.82305, "loss_cls_dn_2": 0.18436, "loss_box_dn_2": 0.82953, "loss_cls_dn_3": 0.19882, "loss_box_dn_3": 0.85384, "loss_cls_dn_4": 0.19613, "loss_box_dn_4": 0.86936, "loss_cls_dn_5": 0.20847, "loss_box_dn_5": 0.87863, "loss_dense_depth": 0.82132, "loss": 28.985, "grad_norm": 73.30416, "time": 1.55698} -{"mode": "train", "epoch": 1, "iter": 109, "lr": 0.00014, "memory": 49167, "data_time": 0.08732, "loss_cls_0": 0.91611, "loss_box_0": 1.79816, "loss_cns_0": 0.61847, "loss_yns_0": 0.15932, "loss_cls_1": 0.95419, "loss_box_1": 1.89003, "loss_cns_1": 0.62798, "loss_yns_1": 0.15699, "loss_cls_2": 0.97706, "loss_box_2": 1.84585, "loss_cns_2": 0.64228, "loss_yns_2": 0.16615, "loss_cls_3": 1.03083, "loss_box_3": 1.84657, "loss_cns_3": 0.65082, "loss_yns_3": 0.16287, "loss_cls_4": 1.02144, "loss_box_4": 1.87358, "loss_cns_4": 0.64934, "loss_yns_4": 0.15937, "loss_cls_5": 1.01861, "loss_box_5": 1.86032, "loss_cns_5": 0.65145, "loss_yns_5": 0.15987, "loss_cls_dn_0": 0.24256, "loss_box_dn_0": 0.80912, "loss_cls_dn_1": 0.17923, "loss_box_dn_1": 0.83368, "loss_cls_dn_2": 0.18295, "loss_box_dn_2": 0.81751, "loss_cls_dn_3": 0.19363, "loss_box_dn_3": 0.82401, "loss_cls_dn_4": 0.19662, "loss_box_dn_4": 0.83326, "loss_cls_dn_5": 0.20259, "loss_box_dn_5": 0.83729, "loss_dense_depth": 0.81615, "loss": 28.80628, "grad_norm": 50.53461, "time": 1.57119} -{"mode": "train", "epoch": 1, "iter": 110, "lr": 0.00014, "memory": 49167, "data_time": 0.07709, "loss_cls_0": 0.90159, "loss_box_0": 1.80319, "loss_cns_0": 0.61308, "loss_yns_0": 0.15585, "loss_cls_1": 0.96336, "loss_box_1": 1.86408, "loss_cns_1": 0.62678, "loss_yns_1": 0.15627, "loss_cls_2": 0.98418, "loss_box_2": 1.87882, "loss_cns_2": 0.63505, "loss_yns_2": 0.16136, "loss_cls_3": 1.0216, "loss_box_3": 1.84223, "loss_cns_3": 0.64171, "loss_yns_3": 0.16127, "loss_cls_4": 1.03251, "loss_box_4": 1.86128, "loss_cns_4": 0.63764, "loss_yns_4": 0.1572, "loss_cls_5": 1.01776, "loss_box_5": 1.85729, "loss_cns_5": 0.64013, "loss_yns_5": 0.15864, "loss_cls_dn_0": 0.25194, "loss_box_dn_0": 0.8041, "loss_cls_dn_1": 0.17841, "loss_box_dn_1": 0.77065, "loss_cls_dn_2": 0.18209, "loss_box_dn_2": 0.77102, "loss_cls_dn_3": 0.18655, "loss_box_dn_3": 0.76519, "loss_cls_dn_4": 0.1942, "loss_box_dn_4": 0.77994, "loss_cls_dn_5": 0.19975, "loss_box_dn_5": 0.78752, "loss_dense_depth": 0.83358, "loss": 28.47781, "grad_norm": 52.23039, "time": 1.56724} -{"mode": "train", "epoch": 1, "iter": 111, "lr": 0.00014, "memory": 49167, "data_time": 0.07796, "loss_cls_0": 0.90603, "loss_box_0": 1.7863, "loss_cns_0": 0.61718, "loss_yns_0": 0.15559, "loss_cls_1": 0.95493, "loss_box_1": 1.88999, "loss_cns_1": 0.63104, "loss_yns_1": 0.15914, "loss_cls_2": 0.986, "loss_box_2": 1.85138, "loss_cns_2": 0.64565, "loss_yns_2": 0.15598, "loss_cls_3": 1.02997, "loss_box_3": 1.84201, "loss_cns_3": 0.64669, "loss_yns_3": 0.15777, "loss_cls_4": 1.04029, "loss_box_4": 1.8206, "loss_cns_4": 0.64876, "loss_yns_4": 0.15657, "loss_cls_5": 1.03639, "loss_box_5": 1.82877, "loss_cns_5": 0.64655, "loss_yns_5": 0.157, "loss_cls_dn_0": 0.25499, "loss_box_dn_0": 0.80311, "loss_cls_dn_1": 0.17754, "loss_box_dn_1": 0.76621, "loss_cls_dn_2": 0.18408, "loss_box_dn_2": 0.76335, "loss_cls_dn_3": 0.19681, "loss_box_dn_3": 0.77307, "loss_cls_dn_4": 0.19336, "loss_box_dn_4": 0.78584, "loss_cls_dn_5": 0.20457, "loss_box_dn_5": 0.8001, "loss_dense_depth": 0.85211, "loss": 28.50573, "grad_norm": 48.51966, "time": 1.55458} -{"mode": "train", "epoch": 1, "iter": 112, "lr": 0.00014, "memory": 49167, "data_time": 0.08573, "loss_cls_0": 0.91136, "loss_box_0": 1.82021, "loss_cns_0": 0.61362, "loss_yns_0": 0.1588, "loss_cls_1": 0.96754, "loss_box_1": 1.95382, "loss_cns_1": 0.62727, "loss_yns_1": 0.15761, "loss_cls_2": 0.98784, "loss_box_2": 1.90451, "loss_cns_2": 0.64012, "loss_yns_2": 0.15768, "loss_cls_3": 1.02374, "loss_box_3": 1.94217, "loss_cns_3": 0.63839, "loss_yns_3": 0.15468, "loss_cls_4": 1.028, "loss_box_4": 1.9087, "loss_cns_4": 0.64439, "loss_yns_4": 0.16039, "loss_cls_5": 1.03829, "loss_box_5": 1.97106, "loss_cns_5": 0.63522, "loss_yns_5": 0.15986, "loss_cls_dn_0": 0.25882, "loss_box_dn_0": 0.8083, "loss_cls_dn_1": 0.18139, "loss_box_dn_1": 0.80715, "loss_cls_dn_2": 0.1882, "loss_box_dn_2": 0.79563, "loss_cls_dn_3": 0.20716, "loss_box_dn_3": 0.82881, "loss_cls_dn_4": 0.20126, "loss_box_dn_4": 0.84362, "loss_cls_dn_5": 0.21193, "loss_box_dn_5": 0.8795, "loss_dense_depth": 0.87991, "loss": 29.29692, "grad_norm": 71.8072, "time": 1.57602} -{"mode": "train", "epoch": 1, "iter": 113, "lr": 0.00014, "memory": 49167, "data_time": 0.07848, "loss_cls_0": 0.88291, "loss_box_0": 1.78901, "loss_cns_0": 0.62055, "loss_yns_0": 0.15761, "loss_cls_1": 0.95309, "loss_box_1": 1.9001, "loss_cns_1": 0.63266, "loss_yns_1": 0.15323, "loss_cls_2": 0.96014, "loss_box_2": 1.84059, "loss_cns_2": 0.64066, "loss_yns_2": 0.162, "loss_cls_3": 0.98486, "loss_box_3": 1.85582, "loss_cns_3": 0.64227, "loss_yns_3": 0.15504, "loss_cls_4": 0.97632, "loss_box_4": 1.82543, "loss_cns_4": 0.64813, "loss_yns_4": 0.15608, "loss_cls_5": 0.99154, "loss_box_5": 1.86763, "loss_cns_5": 0.64044, "loss_yns_5": 0.15542, "loss_cls_dn_0": 0.2415, "loss_box_dn_0": 0.80552, "loss_cls_dn_1": 0.18159, "loss_box_dn_1": 0.85293, "loss_cls_dn_2": 0.18606, "loss_box_dn_2": 0.83768, "loss_cls_dn_3": 0.2023, "loss_box_dn_3": 0.86543, "loss_cls_dn_4": 0.19624, "loss_box_dn_4": 0.88278, "loss_cls_dn_5": 0.20627, "loss_box_dn_5": 0.91245, "loss_dense_depth": 0.84683, "loss": 28.80913, "grad_norm": 53.05536, "time": 1.55759} -{"mode": "train", "epoch": 1, "iter": 114, "lr": 0.00015, "memory": 49167, "data_time": 0.07284, "loss_cls_0": 0.89221, "loss_box_0": 1.82904, "loss_cns_0": 0.61636, "loss_yns_0": 0.16034, "loss_cls_1": 0.96401, "loss_box_1": 1.88896, "loss_cns_1": 0.64373, "loss_yns_1": 0.15422, "loss_cls_2": 0.98932, "loss_box_2": 1.84137, "loss_cns_2": 0.6464, "loss_yns_2": 0.16368, "loss_cls_3": 0.99404, "loss_box_3": 1.87126, "loss_cns_3": 0.6468, "loss_yns_3": 0.15891, "loss_cls_4": 0.99774, "loss_box_4": 1.88647, "loss_cns_4": 0.64041, "loss_yns_4": 0.15867, "loss_cls_5": 0.99364, "loss_box_5": 1.84286, "loss_cns_5": 0.6488, "loss_yns_5": 0.15784, "loss_cls_dn_0": 0.24877, "loss_box_dn_0": 0.8093, "loss_cls_dn_1": 0.17292, "loss_box_dn_1": 0.82688, "loss_cls_dn_2": 0.18271, "loss_box_dn_2": 0.82472, "loss_cls_dn_3": 0.18855, "loss_box_dn_3": 0.84624, "loss_cls_dn_4": 0.19202, "loss_box_dn_4": 0.87642, "loss_cls_dn_5": 0.19876, "loss_box_dn_5": 0.87887, "loss_dense_depth": 0.85116, "loss": 28.8844, "grad_norm": 58.77251, "time": 1.56465} -{"mode": "train", "epoch": 1, "iter": 115, "lr": 0.00015, "memory": 49167, "data_time": 0.0684, "loss_cls_0": 0.85385, "loss_box_0": 1.78305, "loss_cns_0": 0.61576, "loss_yns_0": 0.15627, "loss_cls_1": 0.955, "loss_box_1": 1.91135, "loss_cns_1": 0.64388, "loss_yns_1": 0.1563, "loss_cls_2": 0.97501, "loss_box_2": 1.86684, "loss_cns_2": 0.6459, "loss_yns_2": 0.15759, "loss_cls_3": 1.0037, "loss_box_3": 1.88358, "loss_cns_3": 0.64795, "loss_yns_3": 0.15613, "loss_cls_4": 0.97395, "loss_box_4": 1.90375, "loss_cns_4": 0.64132, "loss_yns_4": 0.15958, "loss_cls_5": 1.00058, "loss_box_5": 1.84892, "loss_cns_5": 0.65112, "loss_yns_5": 0.15821, "loss_cls_dn_0": 0.2399, "loss_box_dn_0": 0.81175, "loss_cls_dn_1": 0.16964, "loss_box_dn_1": 0.84288, "loss_cls_dn_2": 0.18062, "loss_box_dn_2": 0.84077, "loss_cls_dn_3": 0.18676, "loss_box_dn_3": 0.85587, "loss_cls_dn_4": 0.19127, "loss_box_dn_4": 0.8736, "loss_cls_dn_5": 0.19781, "loss_box_dn_5": 0.86722, "loss_dense_depth": 0.85308, "loss": 28.86073, "grad_norm": 68.68478, "time": 1.57061} -{"mode": "train", "epoch": 1, "iter": 116, "lr": 0.00015, "memory": 49167, "data_time": 0.07307, "loss_cls_0": 0.87584, "loss_box_0": 1.75074, "loss_cns_0": 0.61359, "loss_yns_0": 0.15441, "loss_cls_1": 0.96251, "loss_box_1": 1.88956, "loss_cns_1": 0.6422, "loss_yns_1": 0.15867, "loss_cls_2": 0.96371, "loss_box_2": 1.82583, "loss_cns_2": 0.64765, "loss_yns_2": 0.15481, "loss_cls_3": 0.99902, "loss_box_3": 1.81008, "loss_cns_3": 0.65402, "loss_yns_3": 0.15537, "loss_cls_4": 0.97775, "loss_box_4": 1.81591, "loss_cns_4": 0.65375, "loss_yns_4": 0.16057, "loss_cls_5": 0.99957, "loss_box_5": 1.82033, "loss_cns_5": 0.65251, "loss_yns_5": 0.15891, "loss_cls_dn_0": 0.24358, "loss_box_dn_0": 0.79924, "loss_cls_dn_1": 0.17265, "loss_box_dn_1": 0.79921, "loss_cls_dn_2": 0.17957, "loss_box_dn_2": 0.7747, "loss_cls_dn_3": 0.18598, "loss_box_dn_3": 0.77052, "loss_cls_dn_4": 0.19217, "loss_box_dn_4": 0.77559, "loss_cls_dn_5": 0.19503, "loss_box_dn_5": 0.78162, "loss_dense_depth": 0.84596, "loss": 28.21312, "grad_norm": 48.59819, "time": 1.56799} -{"mode": "train", "epoch": 1, "iter": 117, "lr": 0.00015, "memory": 49167, "data_time": 0.07303, "loss_cls_0": 0.87788, "loss_box_0": 1.74947, "loss_cns_0": 0.61807, "loss_yns_0": 0.15389, "loss_cls_1": 0.94074, "loss_box_1": 1.84251, "loss_cns_1": 0.64508, "loss_yns_1": 0.15398, "loss_cls_2": 0.98888, "loss_box_2": 1.8347, "loss_cns_2": 0.64354, "loss_yns_2": 0.15481, "loss_cls_3": 0.98672, "loss_box_3": 1.82621, "loss_cns_3": 0.64674, "loss_yns_3": 0.15288, "loss_cls_4": 1.01646, "loss_box_4": 1.82776, "loss_cns_4": 0.65187, "loss_yns_4": 0.15543, "loss_cls_5": 0.99521, "loss_box_5": 1.85159, "loss_cns_5": 0.64639, "loss_yns_5": 0.15731, "loss_cls_dn_0": 0.23897, "loss_box_dn_0": 0.80592, "loss_cls_dn_1": 0.17031, "loss_box_dn_1": 0.77161, "loss_cls_dn_2": 0.17358, "loss_box_dn_2": 0.75767, "loss_cls_dn_3": 0.18147, "loss_box_dn_3": 0.75948, "loss_cls_dn_4": 0.18927, "loss_box_dn_4": 0.7686, "loss_cls_dn_5": 0.18949, "loss_box_dn_5": 0.78186, "loss_dense_depth": 0.86095, "loss": 28.16731, "grad_norm": 64.03812, "time": 1.56375} -{"mode": "train", "epoch": 1, "iter": 118, "lr": 0.00015, "memory": 49167, "data_time": 0.08033, "loss_cls_0": 0.89495, "loss_box_0": 1.76583, "loss_cns_0": 0.6187, "loss_yns_0": 0.15722, "loss_cls_1": 0.9636, "loss_box_1": 1.83359, "loss_cns_1": 0.6481, "loss_yns_1": 0.156, "loss_cls_2": 1.00397, "loss_box_2": 1.85232, "loss_cns_2": 0.64707, "loss_yns_2": 0.15949, "loss_cls_3": 1.00521, "loss_box_3": 1.84817, "loss_cns_3": 0.65134, "loss_yns_3": 0.15672, "loss_cls_4": 1.00256, "loss_box_4": 1.83932, "loss_cns_4": 0.6553, "loss_yns_4": 0.15646, "loss_cls_5": 1.00544, "loss_box_5": 1.8504, "loss_cns_5": 0.65337, "loss_yns_5": 0.1581, "loss_cls_dn_0": 0.24211, "loss_box_dn_0": 0.80467, "loss_cls_dn_1": 0.17196, "loss_box_dn_1": 0.75846, "loss_cls_dn_2": 0.17608, "loss_box_dn_2": 0.75663, "loss_cls_dn_3": 0.18258, "loss_box_dn_3": 0.7645, "loss_cls_dn_4": 0.18709, "loss_box_dn_4": 0.77126, "loss_cls_dn_5": 0.19066, "loss_box_dn_5": 0.78785, "loss_dense_depth": 0.88609, "loss": 28.36314, "grad_norm": 57.49092, "time": 1.57047} -{"mode": "train", "epoch": 1, "iter": 119, "lr": 0.00015, "memory": 49167, "data_time": 0.07444, "loss_cls_0": 0.88374, "loss_box_0": 1.72619, "loss_cns_0": 0.61201, "loss_yns_0": 0.15318, "loss_cls_1": 0.97444, "loss_box_1": 1.89543, "loss_cns_1": 0.64669, "loss_yns_1": 0.15632, "loss_cls_2": 0.99134, "loss_box_2": 1.86508, "loss_cns_2": 0.65031, "loss_yns_2": 0.15783, "loss_cls_3": 1.01643, "loss_box_3": 1.87047, "loss_cns_3": 0.65169, "loss_yns_3": 0.1556, "loss_cls_4": 1.00378, "loss_box_4": 1.90079, "loss_cns_4": 0.64855, "loss_yns_4": 0.15546, "loss_cls_5": 0.99378, "loss_box_5": 1.92431, "loss_cns_5": 0.64436, "loss_yns_5": 0.15397, "loss_cls_dn_0": 0.23774, "loss_box_dn_0": 0.78987, "loss_cls_dn_1": 0.17356, "loss_box_dn_1": 0.75659, "loss_cls_dn_2": 0.17753, "loss_box_dn_2": 0.74835, "loss_cls_dn_3": 0.18338, "loss_box_dn_3": 0.75758, "loss_cls_dn_4": 0.18335, "loss_box_dn_4": 0.78112, "loss_cls_dn_5": 0.18977, "loss_box_dn_5": 0.80614, "loss_dense_depth": 0.82408, "loss": 28.44078, "grad_norm": 61.83014, "time": 1.59721} -{"mode": "train", "epoch": 1, "iter": 120, "lr": 0.00015, "memory": 49167, "data_time": 0.0768, "loss_cls_0": 0.87328, "loss_box_0": 1.7294, "loss_cns_0": 0.60743, "loss_yns_0": 0.15082, "loss_cls_1": 0.98385, "loss_box_1": 1.88919, "loss_cns_1": 0.64055, "loss_yns_1": 0.15716, "loss_cls_2": 1.00796, "loss_box_2": 1.82976, "loss_cns_2": 0.64968, "loss_yns_2": 0.15739, "loss_cls_3": 1.01994, "loss_box_3": 1.83031, "loss_cns_3": 0.65021, "loss_yns_3": 0.15628, "loss_cls_4": 1.03222, "loss_box_4": 1.85549, "loss_cns_4": 0.6484, "loss_yns_4": 0.16009, "loss_cls_5": 0.99888, "loss_box_5": 1.90462, "loss_cns_5": 0.64219, "loss_yns_5": 0.15692, "loss_cls_dn_0": 0.23835, "loss_box_dn_0": 0.7965, "loss_cls_dn_1": 0.17859, "loss_box_dn_1": 0.78957, "loss_cls_dn_2": 0.1873, "loss_box_dn_2": 0.78245, "loss_cls_dn_3": 0.19498, "loss_box_dn_3": 0.79492, "loss_cls_dn_4": 0.21025, "loss_box_dn_4": 0.82274, "loss_cls_dn_5": 0.21383, "loss_box_dn_5": 0.86155, "loss_dense_depth": 0.87492, "loss": 28.67795, "grad_norm": 64.23539, "time": 1.55146} -{"mode": "train", "epoch": 1, "iter": 121, "lr": 0.00015, "memory": 49167, "data_time": 0.09256, "loss_cls_0": 0.88494, "loss_box_0": 1.76736, "loss_cns_0": 0.61376, "loss_yns_0": 0.15244, "loss_cls_1": 0.98959, "loss_box_1": 1.89393, "loss_cns_1": 0.64257, "loss_yns_1": 0.15509, "loss_cls_2": 1.03513, "loss_box_2": 1.85286, "loss_cns_2": 0.65089, "loss_yns_2": 0.15541, "loss_cls_3": 1.02172, "loss_box_3": 1.86186, "loss_cns_3": 0.65428, "loss_yns_3": 0.15558, "loss_cls_4": 1.04796, "loss_box_4": 1.86058, "loss_cns_4": 0.65337, "loss_yns_4": 0.16103, "loss_cls_5": 1.01763, "loss_box_5": 1.87749, "loss_cns_5": 0.65207, "loss_yns_5": 0.15997, "loss_cls_dn_0": 0.23935, "loss_box_dn_0": 0.80301, "loss_cls_dn_1": 0.17607, "loss_box_dn_1": 0.80044, "loss_cls_dn_2": 0.19549, "loss_box_dn_2": 0.80047, "loss_cls_dn_3": 0.20138, "loss_box_dn_3": 0.81476, "loss_cls_dn_4": 0.23184, "loss_box_dn_4": 0.8223, "loss_cls_dn_5": 0.2324, "loss_box_dn_5": 0.85474, "loss_dense_depth": 0.85536, "loss": 28.94514, "grad_norm": 63.37638, "time": 1.61981} -{"mode": "train", "epoch": 1, "iter": 122, "lr": 0.00015, "memory": 49167, "data_time": 0.07799, "loss_cls_0": 0.91642, "loss_box_0": 1.76849, "loss_cns_0": 0.62237, "loss_yns_0": 0.15681, "loss_cls_1": 0.95312, "loss_box_1": 1.87903, "loss_cns_1": 0.64026, "loss_yns_1": 0.15875, "loss_cls_2": 0.99474, "loss_box_2": 1.84772, "loss_cns_2": 0.64709, "loss_yns_2": 0.15964, "loss_cls_3": 1.00146, "loss_box_3": 1.85831, "loss_cns_3": 0.65662, "loss_yns_3": 0.15754, "loss_cls_4": 1.00269, "loss_box_4": 1.84939, "loss_cns_4": 0.65042, "loss_yns_4": 0.1605, "loss_cls_5": 0.99571, "loss_box_5": 1.87439, "loss_cns_5": 0.65017, "loss_yns_5": 0.16181, "loss_cls_dn_0": 0.2383, "loss_box_dn_0": 0.79488, "loss_cls_dn_1": 0.17181, "loss_box_dn_1": 0.81403, "loss_cls_dn_2": 0.18963, "loss_box_dn_2": 0.81454, "loss_cls_dn_3": 0.1926, "loss_box_dn_3": 0.8296, "loss_cls_dn_4": 0.20935, "loss_box_dn_4": 0.83038, "loss_cls_dn_5": 0.2087, "loss_box_dn_5": 0.85749, "loss_dense_depth": 0.79835, "loss": 28.71312, "grad_norm": 66.22742, "time": 1.64262} -{"mode": "train", "epoch": 1, "iter": 123, "lr": 0.00015, "memory": 49167, "data_time": 0.1293, "loss_cls_0": 0.94101, "loss_box_0": 1.75696, "loss_cns_0": 0.62386, "loss_yns_0": 0.1567, "loss_cls_1": 1.03837, "loss_box_1": 1.85416, "loss_cns_1": 0.64374, "loss_yns_1": 0.15881, "loss_cls_2": 1.00335, "loss_box_2": 1.79932, "loss_cns_2": 0.65181, "loss_yns_2": 0.16045, "loss_cls_3": 1.05666, "loss_box_3": 1.80308, "loss_cns_3": 0.65718, "loss_yns_3": 0.15759, "loss_cls_4": 1.02061, "loss_box_4": 1.79666, "loss_cns_4": 0.65371, "loss_yns_4": 0.15817, "loss_cls_5": 1.03563, "loss_box_5": 1.80216, "loss_cns_5": 0.65419, "loss_yns_5": 0.1608, "loss_cls_dn_0": 0.2341, "loss_box_dn_0": 0.79764, "loss_cls_dn_1": 0.17535, "loss_box_dn_1": 0.80173, "loss_cls_dn_2": 0.18097, "loss_box_dn_2": 0.79054, "loss_cls_dn_3": 0.18312, "loss_box_dn_3": 0.80098, "loss_cls_dn_4": 0.18839, "loss_box_dn_4": 0.80729, "loss_cls_dn_5": 0.19298, "loss_box_dn_5": 0.82021, "loss_dense_depth": 0.86697, "loss": 28.58526, "grad_norm": 49.34915, "time": 1.61679} -{"mode": "train", "epoch": 1, "iter": 124, "lr": 0.00015, "memory": 49167, "data_time": 0.07614, "loss_cls_0": 0.89907, "loss_box_0": 1.74536, "loss_cns_0": 0.61869, "loss_yns_0": 0.15365, "loss_cls_1": 0.95575, "loss_box_1": 1.79422, "loss_cns_1": 0.65253, "loss_yns_1": 0.15994, "loss_cls_2": 0.99771, "loss_box_2": 1.75696, "loss_cns_2": 0.65738, "loss_yns_2": 0.15696, "loss_cls_3": 1.0251, "loss_box_3": 1.76965, "loss_cns_3": 0.65438, "loss_yns_3": 0.15821, "loss_cls_4": 1.03108, "loss_box_4": 1.79227, "loss_cns_4": 0.65566, "loss_yns_4": 0.16034, "loss_cls_5": 1.00959, "loss_box_5": 1.79148, "loss_cns_5": 0.65864, "loss_yns_5": 0.16206, "loss_cls_dn_0": 0.24097, "loss_box_dn_0": 0.80547, "loss_cls_dn_1": 0.16956, "loss_box_dn_1": 0.77089, "loss_cls_dn_2": 0.17124, "loss_box_dn_2": 0.75723, "loss_cls_dn_3": 0.17603, "loss_box_dn_3": 0.7638, "loss_cls_dn_4": 0.17748, "loss_box_dn_4": 0.77699, "loss_cls_dn_5": 0.19005, "loss_box_dn_5": 0.78828, "loss_dense_depth": 0.77786, "loss": 27.98253, "grad_norm": 51.31892, "time": 1.55647} -{"mode": "train", "epoch": 1, "iter": 125, "lr": 0.00015, "memory": 49167, "data_time": 0.10394, "loss_cls_0": 0.98294, "loss_box_0": 1.70231, "loss_cns_0": 0.60347, "loss_yns_0": 0.15263, "loss_cls_1": 0.99677, "loss_box_1": 1.76938, "loss_cns_1": 0.6524, "loss_yns_1": 0.15785, "loss_cls_2": 1.00358, "loss_box_2": 1.73884, "loss_cns_2": 0.66151, "loss_yns_2": 0.15717, "loss_cls_3": 1.01392, "loss_box_3": 1.73646, "loss_cns_3": 0.66575, "loss_yns_3": 0.15863, "loss_cls_4": 1.00764, "loss_box_4": 1.76057, "loss_cns_4": 0.66304, "loss_yns_4": 0.16099, "loss_cls_5": 1.01877, "loss_box_5": 1.77071, "loss_cns_5": 0.65873, "loss_yns_5": 0.15773, "loss_cls_dn_0": 0.2486, "loss_box_dn_0": 0.78815, "loss_cls_dn_1": 0.17061, "loss_box_dn_1": 0.7626, "loss_cls_dn_2": 0.17107, "loss_box_dn_2": 0.74853, "loss_cls_dn_3": 0.17585, "loss_box_dn_3": 0.74923, "loss_cls_dn_4": 0.17597, "loss_box_dn_4": 0.75641, "loss_cls_dn_5": 0.18633, "loss_box_dn_5": 0.7673, "loss_dense_depth": 0.83206, "loss": 27.8845, "grad_norm": 43.30021, "time": 1.5659} -{"mode": "train", "epoch": 1, "iter": 126, "lr": 0.00015, "memory": 49167, "data_time": 0.07418, "loss_cls_0": 0.91397, "loss_box_0": 1.77681, "loss_cns_0": 0.61883, "loss_yns_0": 0.15763, "loss_cls_1": 0.9937, "loss_box_1": 1.87069, "loss_cns_1": 0.64469, "loss_yns_1": 0.16057, "loss_cls_2": 1.01065, "loss_box_2": 1.82563, "loss_cns_2": 0.6523, "loss_yns_2": 0.16172, "loss_cls_3": 1.02364, "loss_box_3": 1.8179, "loss_cns_3": 0.65756, "loss_yns_3": 0.16318, "loss_cls_4": 1.00922, "loss_box_4": 1.81696, "loss_cns_4": 0.65293, "loss_yns_4": 0.1611, "loss_cls_5": 1.02506, "loss_box_5": 1.82429, "loss_cns_5": 0.65238, "loss_yns_5": 0.1581, "loss_cls_dn_0": 0.24051, "loss_box_dn_0": 0.78921, "loss_cls_dn_1": 0.17088, "loss_box_dn_1": 0.75721, "loss_cls_dn_2": 0.17659, "loss_box_dn_2": 0.74348, "loss_cls_dn_3": 0.18189, "loss_box_dn_3": 0.75023, "loss_cls_dn_4": 0.18412, "loss_box_dn_4": 0.7576, "loss_cls_dn_5": 0.1897, "loss_box_dn_5": 0.7682, "loss_dense_depth": 0.7978, "loss": 28.25695, "grad_norm": 45.40922, "time": 1.58119} -{"mode": "train", "epoch": 1, "iter": 127, "lr": 0.00015, "memory": 49167, "data_time": 0.07597, "loss_cls_0": 0.93143, "loss_box_0": 1.78305, "loss_cns_0": 0.62154, "loss_yns_0": 0.15526, "loss_cls_1": 0.98745, "loss_box_1": 1.81606, "loss_cns_1": 0.6449, "loss_yns_1": 0.16174, "loss_cls_2": 1.01012, "loss_box_2": 1.77229, "loss_cns_2": 0.65544, "loss_yns_2": 0.1576, "loss_cls_3": 1.01907, "loss_box_3": 1.76547, "loss_cns_3": 0.65811, "loss_yns_3": 0.15793, "loss_cls_4": 1.00352, "loss_box_4": 1.7676, "loss_cns_4": 0.6583, "loss_yns_4": 0.1573, "loss_cls_5": 1.00987, "loss_box_5": 1.78127, "loss_cns_5": 0.66108, "loss_yns_5": 0.15831, "loss_cls_dn_0": 0.23603, "loss_box_dn_0": 0.78501, "loss_cls_dn_1": 0.16952, "loss_box_dn_1": 0.77193, "loss_cls_dn_2": 0.17622, "loss_box_dn_2": 0.76295, "loss_cls_dn_3": 0.18226, "loss_box_dn_3": 0.77587, "loss_cls_dn_4": 0.1877, "loss_box_dn_4": 0.79141, "loss_cls_dn_5": 0.19323, "loss_box_dn_5": 0.81101, "loss_dense_depth": 0.80587, "loss": 28.14372, "grad_norm": 60.56989, "time": 1.55416} -{"mode": "train", "epoch": 1, "iter": 128, "lr": 0.00015, "memory": 49167, "data_time": 0.08026, "loss_cls_0": 0.96786, "loss_box_0": 1.76272, "loss_cns_0": 0.62616, "loss_yns_0": 0.15526, "loss_cls_1": 0.9724, "loss_box_1": 1.81118, "loss_cns_1": 0.64436, "loss_yns_1": 0.16489, "loss_cls_2": 0.99017, "loss_box_2": 1.76348, "loss_cns_2": 0.65651, "loss_yns_2": 0.15694, "loss_cls_3": 1.00546, "loss_box_3": 1.75869, "loss_cns_3": 0.65788, "loss_yns_3": 0.15735, "loss_cls_4": 1.01585, "loss_box_4": 1.74206, "loss_cns_4": 0.65721, "loss_yns_4": 0.15864, "loss_cls_5": 0.99605, "loss_box_5": 1.76551, "loss_cns_5": 0.65508, "loss_yns_5": 0.1623, "loss_cls_dn_0": 0.23563, "loss_box_dn_0": 0.78648, "loss_cls_dn_1": 0.1678, "loss_box_dn_1": 0.77364, "loss_cls_dn_2": 0.16959, "loss_box_dn_2": 0.76874, "loss_cls_dn_3": 0.17829, "loss_box_dn_3": 0.78179, "loss_cls_dn_4": 0.18075, "loss_box_dn_4": 0.79278, "loss_cls_dn_5": 0.18499, "loss_box_dn_5": 0.81507, "loss_dense_depth": 0.85283, "loss": 28.09236, "grad_norm": 52.26822, "time": 1.56348} -{"mode": "train", "epoch": 1, "iter": 129, "lr": 0.00015, "memory": 49167, "data_time": 0.07806, "loss_cls_0": 0.8986, "loss_box_0": 1.75994, "loss_cns_0": 0.62093, "loss_yns_0": 0.15563, "loss_cls_1": 0.96144, "loss_box_1": 1.81072, "loss_cns_1": 0.64293, "loss_yns_1": 0.15641, "loss_cls_2": 0.98604, "loss_box_2": 1.74353, "loss_cns_2": 0.65846, "loss_yns_2": 0.15852, "loss_cls_3": 1.00113, "loss_box_3": 1.75463, "loss_cns_3": 0.66193, "loss_yns_3": 0.15726, "loss_cls_4": 1.03439, "loss_box_4": 1.7265, "loss_cns_4": 0.65651, "loss_yns_4": 0.15832, "loss_cls_5": 1.01196, "loss_box_5": 1.75155, "loss_cns_5": 0.65523, "loss_yns_5": 0.15946, "loss_cls_dn_0": 0.23443, "loss_box_dn_0": 0.79275, "loss_cls_dn_1": 0.16542, "loss_box_dn_1": 0.78771, "loss_cls_dn_2": 0.16719, "loss_box_dn_2": 0.76749, "loss_cls_dn_3": 0.17724, "loss_box_dn_3": 0.77701, "loss_cls_dn_4": 0.1798, "loss_box_dn_4": 0.78067, "loss_cls_dn_5": 0.18343, "loss_box_dn_5": 0.79636, "loss_dense_depth": 0.80189, "loss": 27.8934, "grad_norm": 45.56451, "time": 1.55018} -{"mode": "train", "epoch": 1, "iter": 130, "lr": 0.00015, "memory": 49167, "data_time": 0.07294, "loss_cls_0": 0.91427, "loss_box_0": 1.75445, "loss_cns_0": 0.61458, "loss_yns_0": 0.15255, "loss_cls_1": 0.99082, "loss_box_1": 1.8559, "loss_cns_1": 0.64315, "loss_yns_1": 0.15581, "loss_cls_2": 1.02712, "loss_box_2": 1.8085, "loss_cns_2": 0.6534, "loss_yns_2": 0.15983, "loss_cls_3": 1.02252, "loss_box_3": 1.80945, "loss_cns_3": 0.65468, "loss_yns_3": 0.15857, "loss_cls_4": 1.05199, "loss_box_4": 1.78132, "loss_cns_4": 0.6512, "loss_yns_4": 0.15729, "loss_cls_5": 1.04459, "loss_box_5": 1.7976, "loss_cns_5": 0.65156, "loss_yns_5": 0.15782, "loss_cls_dn_0": 0.24642, "loss_box_dn_0": 0.79659, "loss_cls_dn_1": 0.16613, "loss_box_dn_1": 0.8022, "loss_cls_dn_2": 0.17666, "loss_box_dn_2": 0.78526, "loss_cls_dn_3": 0.18361, "loss_box_dn_3": 0.78485, "loss_cls_dn_4": 0.19005, "loss_box_dn_4": 0.78767, "loss_cls_dn_5": 0.19124, "loss_box_dn_5": 0.79663, "loss_dense_depth": 0.81144, "loss": 28.38773, "grad_norm": 53.24852, "time": 1.58014} -{"mode": "train", "epoch": 1, "iter": 131, "lr": 0.00015, "memory": 49167, "data_time": 0.07049, "loss_cls_0": 0.92047, "loss_box_0": 1.72367, "loss_cns_0": 0.61897, "loss_yns_0": 0.15071, "loss_cls_1": 0.97037, "loss_box_1": 1.90069, "loss_cns_1": 0.64159, "loss_yns_1": 0.15756, "loss_cls_2": 1.01625, "loss_box_2": 1.85013, "loss_cns_2": 0.65066, "loss_yns_2": 0.15653, "loss_cls_3": 1.02224, "loss_box_3": 1.82939, "loss_cns_3": 0.65053, "loss_yns_3": 0.15756, "loss_cls_4": 1.02518, "loss_box_4": 1.84224, "loss_cns_4": 0.64887, "loss_yns_4": 0.15604, "loss_cls_5": 1.03879, "loss_box_5": 1.83317, "loss_cns_5": 0.65046, "loss_yns_5": 0.15753, "loss_cls_dn_0": 0.24888, "loss_box_dn_0": 0.79906, "loss_cls_dn_1": 0.1672, "loss_box_dn_1": 0.79078, "loss_cls_dn_2": 0.17589, "loss_box_dn_2": 0.77493, "loss_cls_dn_3": 0.1806, "loss_box_dn_3": 0.76794, "loss_cls_dn_4": 0.18191, "loss_box_dn_4": 0.77749, "loss_cls_dn_5": 0.18616, "loss_box_dn_5": 0.77973, "loss_dense_depth": 0.87658, "loss": 28.47678, "grad_norm": 56.48566, "time": 1.54803} -{"mode": "train", "epoch": 1, "iter": 132, "lr": 0.00015, "memory": 49167, "data_time": 0.08663, "loss_cls_0": 0.89165, "loss_box_0": 1.73085, "loss_cns_0": 0.61862, "loss_yns_0": 0.14957, "loss_cls_1": 0.96129, "loss_box_1": 1.84007, "loss_cns_1": 0.6412, "loss_yns_1": 0.15502, "loss_cls_2": 0.98709, "loss_box_2": 1.77556, "loss_cns_2": 0.65093, "loss_yns_2": 0.15071, "loss_cls_3": 1.01806, "loss_box_3": 1.77138, "loss_cns_3": 0.65338, "loss_yns_3": 0.1529, "loss_cls_4": 1.01443, "loss_box_4": 1.79633, "loss_cns_4": 0.64957, "loss_yns_4": 0.15311, "loss_cls_5": 1.01776, "loss_box_5": 1.77444, "loss_cns_5": 0.65574, "loss_yns_5": 0.154, "loss_cls_dn_0": 0.24759, "loss_box_dn_0": 0.78535, "loss_cls_dn_1": 0.16376, "loss_box_dn_1": 0.77743, "loss_cls_dn_2": 0.16575, "loss_box_dn_2": 0.76666, "loss_cls_dn_3": 0.17373, "loss_box_dn_3": 0.7749, "loss_cls_dn_4": 0.17737, "loss_box_dn_4": 0.79305, "loss_cls_dn_5": 0.18357, "loss_box_dn_5": 0.80483, "loss_dense_depth": 0.82878, "loss": 28.00645, "grad_norm": 57.04716, "time": 1.57067} -{"mode": "train", "epoch": 1, "iter": 133, "lr": 0.00015, "memory": 49167, "data_time": 0.08301, "loss_cls_0": 0.92322, "loss_box_0": 1.72803, "loss_cns_0": 0.61447, "loss_yns_0": 0.1513, "loss_cls_1": 0.97806, "loss_box_1": 1.81371, "loss_cns_1": 0.64461, "loss_yns_1": 0.15402, "loss_cls_2": 0.98924, "loss_box_2": 1.81148, "loss_cns_2": 0.64656, "loss_yns_2": 0.156, "loss_cls_3": 1.01103, "loss_box_3": 1.80448, "loss_cns_3": 0.65945, "loss_yns_3": 0.15413, "loss_cls_4": 1.00868, "loss_box_4": 1.79911, "loss_cns_4": 0.66015, "loss_yns_4": 0.15598, "loss_cls_5": 1.00065, "loss_box_5": 1.81737, "loss_cns_5": 0.65493, "loss_yns_5": 0.15611, "loss_cls_dn_0": 0.24921, "loss_box_dn_0": 0.79062, "loss_cls_dn_1": 0.16457, "loss_box_dn_1": 0.77696, "loss_cls_dn_2": 0.16657, "loss_box_dn_2": 0.79053, "loss_cls_dn_3": 0.17362, "loss_box_dn_3": 0.80745, "loss_cls_dn_4": 0.18088, "loss_box_dn_4": 0.82048, "loss_cls_dn_5": 0.19179, "loss_box_dn_5": 0.85583, "loss_dense_depth": 0.81927, "loss": 28.28057, "grad_norm": 55.18621, "time": 1.58733} -{"mode": "train", "epoch": 1, "iter": 134, "lr": 0.00015, "memory": 49167, "data_time": 0.08407, "loss_cls_0": 0.88526, "loss_box_0": 1.72283, "loss_cns_0": 0.62256, "loss_yns_0": 0.1527, "loss_cls_1": 0.95736, "loss_box_1": 1.84624, "loss_cns_1": 0.64874, "loss_yns_1": 0.15207, "loss_cls_2": 0.97693, "loss_box_2": 1.83981, "loss_cns_2": 0.64814, "loss_yns_2": 0.15575, "loss_cls_3": 0.99506, "loss_box_3": 1.81321, "loss_cns_3": 0.65942, "loss_yns_3": 0.15396, "loss_cls_4": 1.0012, "loss_box_4": 1.80021, "loss_cns_4": 0.65823, "loss_yns_4": 0.15458, "loss_cls_5": 0.98993, "loss_box_5": 1.82208, "loss_cns_5": 0.65325, "loss_yns_5": 0.15457, "loss_cls_dn_0": 0.23494, "loss_box_dn_0": 0.78859, "loss_cls_dn_1": 0.16695, "loss_box_dn_1": 0.80119, "loss_cls_dn_2": 0.17223, "loss_box_dn_2": 0.81172, "loss_cls_dn_3": 0.17661, "loss_box_dn_3": 0.81888, "loss_cls_dn_4": 0.18208, "loss_box_dn_4": 0.82757, "loss_cls_dn_5": 0.1945, "loss_box_dn_5": 0.85864, "loss_dense_depth": 0.81792, "loss": 28.31592, "grad_norm": 55.60872, "time": 1.58086} -{"mode": "train", "epoch": 1, "iter": 135, "lr": 0.00015, "memory": 49167, "data_time": 0.08081, "loss_cls_0": 0.87892, "loss_box_0": 1.75562, "loss_cns_0": 0.62331, "loss_yns_0": 0.15284, "loss_cls_1": 0.9621, "loss_box_1": 1.90521, "loss_cns_1": 0.64638, "loss_yns_1": 0.15612, "loss_cls_2": 0.98805, "loss_box_2": 1.85199, "loss_cns_2": 0.64871, "loss_yns_2": 0.15565, "loss_cls_3": 1.00923, "loss_box_3": 1.83893, "loss_cns_3": 0.65737, "loss_yns_3": 0.15517, "loss_cls_4": 0.99969, "loss_box_4": 1.83507, "loss_cns_4": 0.6519, "loss_yns_4": 0.15435, "loss_cls_5": 1.00971, "loss_box_5": 1.83759, "loss_cns_5": 0.65301, "loss_yns_5": 0.1544, "loss_cls_dn_0": 0.24051, "loss_box_dn_0": 0.78764, "loss_cls_dn_1": 0.16632, "loss_box_dn_1": 0.80499, "loss_cls_dn_2": 0.17492, "loss_box_dn_2": 0.79565, "loss_cls_dn_3": 0.18592, "loss_box_dn_3": 0.7907, "loss_cls_dn_4": 0.1831, "loss_box_dn_4": 0.80101, "loss_cls_dn_5": 0.19863, "loss_box_dn_5": 0.81091, "loss_dense_depth": 0.81429, "loss": 28.43589, "grad_norm": 37.5335, "time": 1.58302} -{"mode": "train", "epoch": 1, "iter": 136, "lr": 0.00015, "memory": 49167, "data_time": 0.07962, "loss_cls_0": 0.85127, "loss_box_0": 1.71656, "loss_cns_0": 0.62563, "loss_yns_0": 0.15114, "loss_cls_1": 0.9253, "loss_box_1": 1.83198, "loss_cns_1": 0.64429, "loss_yns_1": 0.14848, "loss_cls_2": 0.95082, "loss_box_2": 1.76313, "loss_cns_2": 0.65383, "loss_yns_2": 0.15008, "loss_cls_3": 0.99111, "loss_box_3": 1.76727, "loss_cns_3": 0.65782, "loss_yns_3": 0.14986, "loss_cls_4": 0.96943, "loss_box_4": 1.75303, "loss_cns_4": 0.65939, "loss_yns_4": 0.15008, "loss_cls_5": 0.96774, "loss_box_5": 1.75308, "loss_cns_5": 0.65937, "loss_yns_5": 0.14956, "loss_cls_dn_0": 0.23617, "loss_box_dn_0": 0.77631, "loss_cls_dn_1": 0.15941, "loss_box_dn_1": 0.77948, "loss_cls_dn_2": 0.16345, "loss_box_dn_2": 0.75878, "loss_cls_dn_3": 0.17333, "loss_box_dn_3": 0.75931, "loss_cls_dn_4": 0.17178, "loss_box_dn_4": 0.76496, "loss_cls_dn_5": 0.18119, "loss_box_dn_5": 0.76871, "loss_dense_depth": 0.82375, "loss": 27.55689, "grad_norm": 43.75362, "time": 1.56294} -{"mode": "train", "epoch": 1, "iter": 137, "lr": 0.00015, "memory": 49167, "data_time": 0.08317, "loss_cls_0": 0.86358, "loss_box_0": 1.72449, "loss_cns_0": 0.62079, "loss_yns_0": 0.15194, "loss_cls_1": 0.92743, "loss_box_1": 1.85602, "loss_cns_1": 0.64585, "loss_yns_1": 0.15071, "loss_cls_2": 0.94775, "loss_box_2": 1.77468, "loss_cns_2": 0.65082, "loss_yns_2": 0.15234, "loss_cls_3": 0.97977, "loss_box_3": 1.77622, "loss_cns_3": 0.65563, "loss_yns_3": 0.15111, "loss_cls_4": 0.97088, "loss_box_4": 1.76063, "loss_cns_4": 0.65665, "loss_yns_4": 0.15114, "loss_cls_5": 0.96291, "loss_box_5": 1.76635, "loss_cns_5": 0.65876, "loss_yns_5": 0.15115, "loss_cls_dn_0": 0.24582, "loss_box_dn_0": 0.78451, "loss_cls_dn_1": 0.16075, "loss_box_dn_1": 0.76263, "loss_cls_dn_2": 0.16158, "loss_box_dn_2": 0.73802, "loss_cls_dn_3": 0.16685, "loss_box_dn_3": 0.73605, "loss_cls_dn_4": 0.1684, "loss_box_dn_4": 0.73743, "loss_cls_dn_5": 0.17566, "loss_box_dn_5": 0.74141, "loss_dense_depth": 0.81216, "loss": 27.49889, "grad_norm": 37.12968, "time": 1.56124} -{"mode": "train", "epoch": 1, "iter": 138, "lr": 0.00015, "memory": 49167, "data_time": 0.08315, "loss_cls_0": 0.85847, "loss_box_0": 1.72592, "loss_cns_0": 0.62822, "loss_yns_0": 0.14887, "loss_cls_1": 0.92354, "loss_box_1": 1.80522, "loss_cns_1": 0.64949, "loss_yns_1": 0.15111, "loss_cls_2": 0.95268, "loss_box_2": 1.7544, "loss_cns_2": 0.65042, "loss_yns_2": 0.15176, "loss_cls_3": 0.9639, "loss_box_3": 1.72761, "loss_cns_3": 0.6567, "loss_yns_3": 0.15088, "loss_cls_4": 0.96596, "loss_box_4": 1.73248, "loss_cns_4": 0.65627, "loss_yns_4": 0.15104, "loss_cls_5": 0.99707, "loss_box_5": 1.73875, "loss_cns_5": 0.65682, "loss_yns_5": 0.15017, "loss_cls_dn_0": 0.24025, "loss_box_dn_0": 0.79282, "loss_cls_dn_1": 0.15995, "loss_box_dn_1": 0.75321, "loss_cls_dn_2": 0.16075, "loss_box_dn_2": 0.73958, "loss_cls_dn_3": 0.1629, "loss_box_dn_3": 0.73693, "loss_cls_dn_4": 0.16717, "loss_box_dn_4": 0.74523, "loss_cls_dn_5": 0.17693, "loss_box_dn_5": 0.75314, "loss_dense_depth": 0.8459, "loss": 27.38249, "grad_norm": 33.61383, "time": 1.57803} -{"mode": "train", "epoch": 1, "iter": 139, "lr": 0.00016, "memory": 49167, "data_time": 0.07977, "loss_cls_0": 0.81313, "loss_box_0": 1.68943, "loss_cns_0": 0.62633, "loss_yns_0": 0.1479, "loss_cls_1": 0.89418, "loss_box_1": 1.78058, "loss_cns_1": 0.64597, "loss_yns_1": 0.14903, "loss_cls_2": 0.92092, "loss_box_2": 1.74082, "loss_cns_2": 0.6535, "loss_yns_2": 0.14886, "loss_cls_3": 0.9412, "loss_box_3": 1.71246, "loss_cns_3": 0.65669, "loss_yns_3": 0.14879, "loss_cls_4": 0.93565, "loss_box_4": 1.72369, "loss_cns_4": 0.65392, "loss_yns_4": 0.14936, "loss_cls_5": 0.93051, "loss_box_5": 1.72899, "loss_cns_5": 0.65301, "loss_yns_5": 0.14784, "loss_cls_dn_0": 0.22801, "loss_box_dn_0": 0.78953, "loss_cls_dn_1": 0.15537, "loss_box_dn_1": 0.76231, "loss_cls_dn_2": 0.16305, "loss_box_dn_2": 0.76066, "loss_cls_dn_3": 0.16093, "loss_box_dn_3": 0.76271, "loss_cls_dn_4": 0.16678, "loss_box_dn_4": 0.77831, "loss_cls_dn_5": 0.17484, "loss_box_dn_5": 0.79732, "loss_dense_depth": 0.8434, "loss": 27.13596, "grad_norm": 40.81067, "time": 1.56922} -{"mode": "train", "epoch": 1, "iter": 140, "lr": 0.00016, "memory": 49167, "data_time": 0.08277, "loss_cls_0": 0.82676, "loss_box_0": 1.68124, "loss_cns_0": 0.62449, "loss_yns_0": 0.14799, "loss_cls_1": 0.88818, "loss_box_1": 1.78959, "loss_cns_1": 0.64632, "loss_yns_1": 0.14742, "loss_cls_2": 0.93554, "loss_box_2": 1.73261, "loss_cns_2": 0.65346, "loss_yns_2": 0.14695, "loss_cls_3": 0.95271, "loss_box_3": 1.71241, "loss_cns_3": 0.65804, "loss_yns_3": 0.14722, "loss_cls_4": 0.9382, "loss_box_4": 1.71555, "loss_cns_4": 0.65855, "loss_yns_4": 0.14854, "loss_cls_5": 0.94543, "loss_box_5": 1.71282, "loss_cns_5": 0.65927, "loss_yns_5": 0.14703, "loss_cls_dn_0": 0.22675, "loss_box_dn_0": 0.78741, "loss_cls_dn_1": 0.14954, "loss_box_dn_1": 0.78953, "loss_cls_dn_2": 0.16394, "loss_box_dn_2": 0.78412, "loss_cls_dn_3": 0.15886, "loss_box_dn_3": 0.78802, "loss_cls_dn_4": 0.16552, "loss_box_dn_4": 0.80032, "loss_cls_dn_5": 0.17719, "loss_box_dn_5": 0.81528, "loss_dense_depth": 0.80647, "loss": 27.22926, "grad_norm": 44.53352, "time": 1.57647} -{"mode": "train", "epoch": 1, "iter": 141, "lr": 0.00016, "memory": 49167, "data_time": 0.09699, "loss_cls_0": 0.8878, "loss_box_0": 1.76767, "loss_cns_0": 0.62833, "loss_yns_0": 0.15275, "loss_cls_1": 0.93568, "loss_box_1": 1.85522, "loss_cns_1": 0.6543, "loss_yns_1": 0.14884, "loss_cls_2": 0.96845, "loss_box_2": 1.78049, "loss_cns_2": 0.6558, "loss_yns_2": 0.15006, "loss_cls_3": 0.96473, "loss_box_3": 1.78698, "loss_cns_3": 0.65763, "loss_yns_3": 0.15024, "loss_cls_4": 0.95809, "loss_box_4": 1.78131, "loss_cns_4": 0.65975, "loss_yns_4": 0.15239, "loss_cls_5": 0.97454, "loss_box_5": 1.7696, "loss_cns_5": 0.66155, "loss_yns_5": 0.15039, "loss_cls_dn_0": 0.24287, "loss_box_dn_0": 0.78598, "loss_cls_dn_1": 0.16195, "loss_box_dn_1": 0.78962, "loss_cls_dn_2": 0.17173, "loss_box_dn_2": 0.7723, "loss_cls_dn_3": 0.16737, "loss_box_dn_3": 0.77995, "loss_cls_dn_4": 0.17358, "loss_box_dn_4": 0.78471, "loss_cls_dn_5": 0.18195, "loss_box_dn_5": 0.79022, "loss_dense_depth": 0.83674, "loss": 27.89157, "grad_norm": 38.78638, "time": 1.68019} -{"mode": "train", "epoch": 1, "iter": 142, "lr": 0.00016, "memory": 49167, "data_time": 0.0799, "loss_cls_0": 0.8857, "loss_box_0": 1.78962, "loss_cns_0": 0.62053, "loss_yns_0": 0.15182, "loss_cls_1": 0.95201, "loss_box_1": 1.86703, "loss_cns_1": 0.64657, "loss_yns_1": 0.15075, "loss_cls_2": 0.97023, "loss_box_2": 1.80052, "loss_cns_2": 0.65419, "loss_yns_2": 0.15281, "loss_cls_3": 0.96991, "loss_box_3": 1.81241, "loss_cns_3": 0.65454, "loss_yns_3": 0.15171, "loss_cls_4": 0.97473, "loss_box_4": 1.79219, "loss_cns_4": 0.65351, "loss_yns_4": 0.15094, "loss_cls_5": 0.97092, "loss_box_5": 1.79923, "loss_cns_5": 0.65575, "loss_yns_5": 0.15099, "loss_cls_dn_0": 0.2363, "loss_box_dn_0": 0.78948, "loss_cls_dn_1": 0.16253, "loss_box_dn_1": 0.77916, "loss_cls_dn_2": 0.16423, "loss_box_dn_2": 0.75455, "loss_cls_dn_3": 0.16532, "loss_box_dn_3": 0.76685, "loss_cls_dn_4": 0.17661, "loss_box_dn_4": 0.76674, "loss_cls_dn_5": 0.17874, "loss_box_dn_5": 0.77338, "loss_dense_depth": 0.81835, "loss": 27.91084, "grad_norm": 42.2026, "time": 1.68223} -{"mode": "train", "epoch": 1, "iter": 143, "lr": 0.00016, "memory": 49167, "data_time": 0.1417, "loss_cls_0": 0.83842, "loss_box_0": 1.77605, "loss_cns_0": 0.62381, "loss_yns_0": 0.14777, "loss_cls_1": 0.92589, "loss_box_1": 1.80289, "loss_cns_1": 0.65105, "loss_yns_1": 0.14812, "loss_cls_2": 0.95115, "loss_box_2": 1.76128, "loss_cns_2": 0.6577, "loss_yns_2": 0.14901, "loss_cls_3": 0.96356, "loss_box_3": 1.77103, "loss_cns_3": 0.65727, "loss_yns_3": 0.14928, "loss_cls_4": 0.95515, "loss_box_4": 1.73878, "loss_cns_4": 0.65786, "loss_yns_4": 0.14898, "loss_cls_5": 0.95717, "loss_box_5": 1.74808, "loss_cns_5": 0.65941, "loss_yns_5": 0.14888, "loss_cls_dn_0": 0.22292, "loss_box_dn_0": 0.79181, "loss_cls_dn_1": 0.15605, "loss_box_dn_1": 0.74834, "loss_cls_dn_2": 0.1642, "loss_box_dn_2": 0.73249, "loss_cls_dn_3": 0.16515, "loss_box_dn_3": 0.74447, "loss_cls_dn_4": 0.1793, "loss_box_dn_4": 0.74183, "loss_cls_dn_5": 0.18184, "loss_box_dn_5": 0.74756, "loss_dense_depth": 0.80461, "loss": 27.36915, "grad_norm": 42.74908, "time": 1.63489} -{"mode": "train", "epoch": 1, "iter": 144, "lr": 0.00016, "memory": 49167, "data_time": 0.08176, "loss_cls_0": 0.85327, "loss_box_0": 1.7751, "loss_cns_0": 0.61855, "loss_yns_0": 0.14803, "loss_cls_1": 0.93685, "loss_box_1": 1.80743, "loss_cns_1": 0.64624, "loss_yns_1": 0.14795, "loss_cls_2": 0.95856, "loss_box_2": 1.77882, "loss_cns_2": 0.65396, "loss_yns_2": 0.15079, "loss_cls_3": 1.01259, "loss_box_3": 1.78084, "loss_cns_3": 0.6536, "loss_yns_3": 0.1487, "loss_cls_4": 0.98076, "loss_box_4": 1.76319, "loss_cns_4": 0.65396, "loss_yns_4": 0.15081, "loss_cls_5": 0.9789, "loss_box_5": 1.75207, "loss_cns_5": 0.65563, "loss_yns_5": 0.1489, "loss_cls_dn_0": 0.23178, "loss_box_dn_0": 0.79678, "loss_cls_dn_1": 0.1572, "loss_box_dn_1": 0.75688, "loss_cls_dn_2": 0.16591, "loss_box_dn_2": 0.75048, "loss_cls_dn_3": 0.16965, "loss_box_dn_3": 0.75824, "loss_cls_dn_4": 0.17506, "loss_box_dn_4": 0.76982, "loss_cls_dn_5": 0.18494, "loss_box_dn_5": 0.77339, "loss_dense_depth": 0.78757, "loss": 27.6332, "grad_norm": 47.73961, "time": 1.55701} -{"mode": "train", "epoch": 1, "iter": 145, "lr": 0.00016, "memory": 49167, "data_time": 0.10872, "loss_cls_0": 0.84582, "loss_box_0": 1.7856, "loss_cns_0": 0.62342, "loss_yns_0": 0.15121, "loss_cls_1": 0.93908, "loss_box_1": 1.7978, "loss_cns_1": 0.64877, "loss_yns_1": 0.15202, "loss_cls_2": 0.95433, "loss_box_2": 1.76452, "loss_cns_2": 0.65215, "loss_yns_2": 0.15148, "loss_cls_3": 0.99637, "loss_box_3": 1.76178, "loss_cns_3": 0.65288, "loss_yns_3": 0.14973, "loss_cls_4": 0.98042, "loss_box_4": 1.75741, "loss_cns_4": 0.6533, "loss_yns_4": 0.14831, "loss_cls_5": 0.96276, "loss_box_5": 1.76834, "loss_cns_5": 0.65902, "loss_yns_5": 0.1499, "loss_cls_dn_0": 0.2285, "loss_box_dn_0": 0.79235, "loss_cls_dn_1": 0.15453, "loss_box_dn_1": 0.76394, "loss_cls_dn_2": 0.15965, "loss_box_dn_2": 0.75786, "loss_cls_dn_3": 0.16369, "loss_box_dn_3": 0.7693, "loss_cls_dn_4": 0.16697, "loss_box_dn_4": 0.78311, "loss_cls_dn_5": 0.17349, "loss_box_dn_5": 0.79748, "loss_dense_depth": 0.81225, "loss": 27.62956, "grad_norm": 47.24368, "time": 1.58479} -{"mode": "train", "epoch": 1, "iter": 146, "lr": 0.00016, "memory": 49167, "data_time": 0.0798, "loss_cls_0": 0.85014, "loss_box_0": 1.78483, "loss_cns_0": 0.62527, "loss_yns_0": 0.1486, "loss_cls_1": 0.92258, "loss_box_1": 1.78937, "loss_cns_1": 0.64971, "loss_yns_1": 0.15145, "loss_cls_2": 0.94595, "loss_box_2": 1.75703, "loss_cns_2": 0.65203, "loss_yns_2": 0.15064, "loss_cls_3": 0.95391, "loss_box_3": 1.7559, "loss_cns_3": 0.65289, "loss_yns_3": 0.15163, "loss_cls_4": 0.96464, "loss_box_4": 1.75061, "loss_cns_4": 0.65374, "loss_yns_4": 0.1488, "loss_cls_5": 0.95887, "loss_box_5": 1.76042, "loss_cns_5": 0.65858, "loss_yns_5": 0.14996, "loss_cls_dn_0": 0.22123, "loss_box_dn_0": 0.79001, "loss_cls_dn_1": 0.15384, "loss_box_dn_1": 0.77905, "loss_cls_dn_2": 0.15538, "loss_box_dn_2": 0.77524, "loss_cls_dn_3": 0.15663, "loss_box_dn_3": 0.79594, "loss_cls_dn_4": 0.16822, "loss_box_dn_4": 0.80512, "loss_cls_dn_5": 0.1736, "loss_box_dn_5": 0.82835, "loss_dense_depth": 0.77654, "loss": 27.56671, "grad_norm": 44.49037, "time": 1.58941} -{"mode": "train", "epoch": 1, "iter": 147, "lr": 0.00016, "memory": 49167, "data_time": 0.07971, "loss_cls_0": 0.8574, "loss_box_0": 1.78301, "loss_cns_0": 0.6246, "loss_yns_0": 0.15195, "loss_cls_1": 0.9374, "loss_box_1": 1.82556, "loss_cns_1": 0.65063, "loss_yns_1": 0.15046, "loss_cls_2": 0.96367, "loss_box_2": 1.77975, "loss_cns_2": 0.65525, "loss_yns_2": 0.1492, "loss_cls_3": 0.99243, "loss_box_3": 1.7604, "loss_cns_3": 0.6556, "loss_yns_3": 0.15087, "loss_cls_4": 0.96906, "loss_box_4": 1.75871, "loss_cns_4": 0.65676, "loss_yns_4": 0.14963, "loss_cls_5": 0.97351, "loss_box_5": 1.75672, "loss_cns_5": 0.65845, "loss_yns_5": 0.14879, "loss_cls_dn_0": 0.22122, "loss_box_dn_0": 0.78342, "loss_cls_dn_1": 0.15323, "loss_box_dn_1": 0.79976, "loss_cls_dn_2": 0.15509, "loss_box_dn_2": 0.8045, "loss_cls_dn_3": 0.16244, "loss_box_dn_3": 0.8158, "loss_cls_dn_4": 0.16871, "loss_box_dn_4": 0.82394, "loss_cls_dn_5": 0.1754, "loss_box_dn_5": 0.84322, "loss_dense_depth": 0.78431, "loss": 27.85085, "grad_norm": 45.6622, "time": 1.57901} -{"mode": "train", "epoch": 1, "iter": 148, "lr": 0.00016, "memory": 49167, "data_time": 0.08309, "loss_cls_0": 0.83469, "loss_box_0": 1.73995, "loss_cns_0": 0.62569, "loss_yns_0": 0.14784, "loss_cls_1": 0.93022, "loss_box_1": 1.75452, "loss_cns_1": 0.65387, "loss_yns_1": 0.14937, "loss_cls_2": 0.9438, "loss_box_2": 1.71455, "loss_cns_2": 0.66023, "loss_yns_2": 0.14733, "loss_cls_3": 0.9834, "loss_box_3": 1.68872, "loss_cns_3": 0.66215, "loss_yns_3": 0.14673, "loss_cls_4": 0.9611, "loss_box_4": 1.69586, "loss_cns_4": 0.66182, "loss_yns_4": 0.15063, "loss_cls_5": 0.96474, "loss_box_5": 1.68756, "loss_cns_5": 0.66224, "loss_yns_5": 0.14785, "loss_cls_dn_0": 0.21844, "loss_box_dn_0": 0.78161, "loss_cls_dn_1": 0.15287, "loss_box_dn_1": 0.79506, "loss_cls_dn_2": 0.15709, "loss_box_dn_2": 0.794, "loss_cls_dn_3": 0.16611, "loss_box_dn_3": 0.79584, "loss_cls_dn_4": 0.16891, "loss_box_dn_4": 0.80495, "loss_cls_dn_5": 0.17247, "loss_box_dn_5": 0.81885, "loss_dense_depth": 0.7592, "loss": 27.30026, "grad_norm": 44.22478, "time": 1.54901} diff --git a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20251112_200501.log b/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20251112_200501.log deleted file mode 100644 index 829fa048f16d340c12cf0e7eb95aecb9500bebcd..0000000000000000000000000000000000000000 --- a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20251112_200501.log +++ /dev/null @@ -1,3451 +0,0 @@ -2025-11-12 20:05:01,175 - mmdet - INFO - Environment info: ------------------------------------------------------------- -sys.platform: linux -Python: 3.10.12 (main, May 27 2025, 17:12:29) [GCC 11.4.0] -CUDA available: True -GPU 0,1,2,3,4,5,6,7: BW200, UBB BW1000 -CUDA_HOME: /opt/dtk -NVCC: Not Available -GCC: x86_64-linux-gnu-gcc (Ubuntu 11.4.0-1ubuntu1~22.04) 11.4.0 -PyTorch: 2.4.1 -PyTorch compiling details: PyTorch built with: - - GCC 10.3 - - C++ Version: 201703 - - Intel(R) Math Kernel Library Version 2020.0.4 Product Build 20200917 for Intel(R) 64 architecture applications - - OpenMP 201511 (a.k.a. OpenMP 4.5) - - LAPACK is enabled (usually provided by MKL) - - NNPACK is enabled - - CPU capability usage: AVX512 - - HIP Runtime 6.3.25211 - - MIOpen 2.17.0 - - Magma 2.8.0 - - Build settings: BLAS_INFO=mkl, BUILD_TYPE=Release, CXX_COMPILER=/opt/rh/gcc-toolset-10/root/usr/bin/c++, CXX_FLAGS= -D_GLIBCXX_USE_CXX11_ABI=1 -fvisibility-inlines-hidden -DUSE_PTHREADPOOL -DNDEBUG -DUSE_KINETO -DLIBKINETO_NOCUPTI -DUSE_FBGEMM -DUSE_PYTORCH_QNNPACK -DUSE_XNNPACK -DSYMBOLICATE_MOBILE_DEBUG_HANDLE -O2 -fPIC -Wall -Wextra -Werror=return-type -Werror=non-virtual-dtor -Werror=bool-operation -Wnarrowing -Wno-missing-field-initializers -Wno-type-limits -Wno-array-bounds -Wno-unknown-pragmas -Wno-unused-parameter -Wno-unused-function -Wno-unused-result -Wno-strict-overflow -Wno-strict-aliasing -Wno-stringop-overflow -Wsuggest-override -Wno-psabi -Wno-error=pedantic -Wno-error=old-style-cast -Wno-missing-braces -fdiagnostics-color=always -faligned-new -Wno-unused-but-set-variable -Wno-maybe-uninitialized -fno-math-errno -fno-trapping-math -Werror=format -Wno-stringop-overflow, FORCE_FALLBACK_CUDA_MPI=1, LAPACK_INFO=mkl, PERF_WITH_AVX=1, PERF_WITH_AVX2=1, PERF_WITH_AVX512=1, TORCH_VERSION=2.4.1, USE_CUDA=0, USE_CUDNN=OFF, USE_CUSPARSELT=OFF, USE_EXCEPTION_PTR=1, USE_GFLAGS=1, USE_GLOG=1, USE_GLOO=1, USE_MKL=ON, USE_MKLDNN=0, USE_MPI=1, USE_NCCL=1, USE_NNPACK=ON, USE_OPENMP=1, USE_ROCM=ON, USE_ROCM_KERNEL_ASSERT=OFF, - -TorchVision: 0.19.1 -OpenCV: 4.11.0 -MMCV: 1.6.1 -MMCV Compiler: GCC 11.4 -MMCV CUDA Compiler: rocm not available -MMDetection: 2.26.0+c41df4b ------------------------------------------------------------- - -2025-11-12 20:05:02,102 - mmdet - INFO - Distributed training: True -2025-11-12 20:05:02,815 - mmdet - INFO - Config: -plugin = True -plugin_dir = 'projects/mmdet3d_plugin/' -dist_params = dict(backend='nccl') -log_level = 'INFO' -work_dir = './work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704' -total_batch_size = 160 -num_gpus = 8 -batch_size = 20 -num_iters_per_epoch = 175 -num_epochs = 100 -checkpoint_epoch_interval = 20 -checkpoint_config = dict(interval=3500) -log_config = dict( - interval=1, - hooks=[ - dict(type='TextLoggerHook', by_epoch=False), - dict(type='TensorboardLoggerHook') - ]) -load_from = None -resume_from = None -workflow = [('train', 1)] -fp16 = dict(loss_scale=32.0) -input_shape = (704, 256) -tracking_test = True -tracking_threshold = 0.2 -class_names = [ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier', - 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' -] -num_classes = 10 -embed_dims = 256 -num_groups = 8 -num_decoder = 6 -num_single_frame_decoder = 1 -use_deformable_func = True -strides = [4, 8, 16, 32] -num_levels = 4 -num_depth_layers = 3 -drop_out = 0.1 -temporal = True -decouple_attn = True -with_quality_estimation = True -model = dict( - type='Sparse4D', - use_grid_mask=True, - use_deformable_func=True, - img_backbone=dict( - type='ResNet', - depth=50, - num_stages=4, - frozen_stages=-1, - norm_eval=False, - style='pytorch', - with_cp=True, - out_indices=(0, 1, 2, 3), - norm_cfg=dict(type='BN', requires_grad=True), - pretrained='ckpt/resnet50-19c8e357.pth'), - img_neck=dict( - type='FPN', - num_outs=4, - start_level=0, - out_channels=256, - add_extra_convs='on_output', - relu_before_extra_convs=True, - in_channels=[256, 512, 1024, 2048]), - depth_branch=dict( - type='DenseDepthNet', - embed_dims=256, - num_depth_layers=3, - loss_weight=0.2), - head=dict( - type='Sparse4DHead', - cls_threshold_to_reg=0.05, - decouple_attn=True, - instance_bank=dict( - type='InstanceBank', - num_anchor=900, - embed_dims=256, - anchor='nuscenes_kmeans900.npy', - anchor_handler=dict(type='SparseBox3DKeyPointsGenerator'), - num_temp_instances=600, - confidence_decay=0.6, - feat_grad=False), - anchor_encoder=dict( - type='SparseBox3DEncoder', - vel_dims=3, - embed_dims=[128, 32, 32, 64], - mode='cat', - output_fc=False, - in_loops=1, - out_loops=4), - num_single_frame_decoder=1, - operation_order=[ - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine' - ], - temp_graph_model=dict( - type='MultiheadAttention', - embed_dims=512, - num_heads=8, - batch_first=True, - dropout=0.1), - graph_model=dict( - type='MultiheadAttention', - embed_dims=512, - num_heads=8, - batch_first=True, - dropout=0.1), - norm_layer=dict(type='LN', normalized_shape=256), - ffn=dict( - type='AsymmetricFFN', - in_channels=512, - pre_norm=dict(type='LN'), - embed_dims=256, - feedforward_channels=1024, - num_fcs=2, - ffn_drop=0.1, - act_cfg=dict(type='ReLU', inplace=True)), - deformable_model=dict( - type='DeformableFeatureAggregation', - embed_dims=256, - num_groups=8, - num_levels=4, - num_cams=6, - attn_drop=0.15, - use_deformable_func=True, - use_camera_embed=True, - residual_mode='cat', - kps_generator=dict( - type='SparseBox3DKeyPointsGenerator', - num_learnable_pts=6, - fix_scale=[[0, 0, 0], [0.45, 0, 0], [-0.45, 0, 0], - [0, 0.45, 0], [0, -0.45, 0], [0, 0, 0.45], - [0, 0, -0.45]])), - refine_layer=dict( - type='SparseBox3DRefinementModule', - embed_dims=256, - num_cls=10, - refine_yaw=True, - with_quality_estimation=True), - sampler=dict( - type='SparseBox3DTarget', - num_dn_groups=5, - num_temp_dn_groups=3, - dn_noise_scale=[2.0, 2.0, 2.0, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5], - max_dn_gt=32, - add_neg_dn=True, - cls_weight=2.0, - box_weight=0.25, - reg_weights=[2.0, 2.0, 2.0, 0.5, 0.5, 0.5, 0.0, 0.0, 0.0, 0.0], - cls_wise_reg_weights=dict( - {9: [2.0, 2.0, 2.0, 1.0, 1.0, 1.0, 0.0, 0.0, 1.0, 1.0]})), - loss_cls=dict( - type='FocalLoss', - use_sigmoid=True, - gamma=2.0, - alpha=0.25, - loss_weight=2.0), - loss_reg=dict( - type='SparseBox3DLoss', - loss_box=dict(type='L1Loss', loss_weight=0.25), - loss_centerness=dict(type='CrossEntropyLoss', use_sigmoid=True), - loss_yawness=dict(type='GaussianFocalLoss'), - cls_allow_reverse=[5]), - decoder=dict(type='SparseBox3DDecoder'), - reg_weights=[2.0, 2.0, 2.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0])) -dataset_type = 'NuScenes3DDetTrackDataset' -data_root = 'data/nuscenes/' -anno_root = 'data/nuscenes_anno_pkls/' -file_client_args = dict(backend='disk') -img_norm_cfg = dict( - mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375], to_rgb=True) -train_pipeline = [ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict( - type='LoadPointsFromFile', - coord_type='LIDAR', - load_dim=5, - use_dim=5, - file_client_args=dict(backend='disk')), - dict(type='ResizeCropFlipImage'), - dict(type='MultiScaleDepthMapGenerator', downsample=[4, 8, 16]), - dict(type='BBoxRotation'), - dict(type='PhotoMetricDistortionMultiViewImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict( - type='CircleObjectRangeFilter', - class_dist_thred=[55, 55, 55, 55, 55, 55, 55, 55, 55, 55]), - dict( - type='InstanceNameFilter', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ]), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=[ - 'img', 'timestamp', 'projection_mat', 'image_wh', 'gt_depth', - 'focal', 'gt_bboxes_3d', 'gt_labels_3d' - ], - meta_keys=['T_global', 'T_global_inv', 'timestamp', 'instance_id']) -] -test_pipeline = [ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='ResizeCropFlipImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=['img', 'timestamp', 'projection_mat', 'image_wh'], - meta_keys=['T_global', 'T_global_inv', 'timestamp']) -] -input_modality = dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False) -data_basic_config = dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier', - 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval') -data_aug_conf = dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]) -data = dict( - samples_per_gpu=20, - workers_per_gpu=20, - train=dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval', - ann_file='data/nuscenes_anno_pkls/nuscenes_infos_train.pkl', - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict( - type='LoadPointsFromFile', - coord_type='LIDAR', - load_dim=5, - use_dim=5, - file_client_args=dict(backend='disk')), - dict(type='ResizeCropFlipImage'), - dict(type='MultiScaleDepthMapGenerator', downsample=[4, 8, 16]), - dict(type='BBoxRotation'), - dict(type='PhotoMetricDistortionMultiViewImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict( - type='CircleObjectRangeFilter', - class_dist_thred=[55, 55, 55, 55, 55, 55, 55, 55, 55, 55]), - dict( - type='InstanceNameFilter', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', - 'traffic_cone' - ]), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=[ - 'img', 'timestamp', 'projection_mat', 'image_wh', - 'gt_depth', 'focal', 'gt_bboxes_3d', 'gt_labels_3d' - ], - meta_keys=[ - 'T_global', 'T_global_inv', 'timestamp', 'instance_id' - ]) - ], - test_mode=False, - data_aug_conf=dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]), - with_seq_flag=True, - sequences_split_num=2, - keep_consistent_seq_aug=True), - val=dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval', - ann_file='data/nuscenes_anno_pkls/nuscenes_infos_val.pkl', - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='ResizeCropFlipImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=['img', 'timestamp', 'projection_mat', 'image_wh'], - meta_keys=['T_global', 'T_global_inv', 'timestamp']) - ], - data_aug_conf=dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]), - test_mode=True, - tracking=True, - tracking_threshold=0.2), - test=dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval', - ann_file='data/nuscenes_anno_pkls/nuscenes_infos_val.pkl', - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='ResizeCropFlipImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=['img', 'timestamp', 'projection_mat', 'image_wh'], - meta_keys=['T_global', 'T_global_inv', 'timestamp']) - ], - data_aug_conf=dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]), - test_mode=True, - tracking=True, - tracking_threshold=0.2)) -optimizer = dict( - type='Miopen_AdamW', - lr=0.0006, - weight_decay=0.001, - paramwise_cfg=dict(custom_keys=dict(img_backbone=dict(lr_mult=0.5)))) -optimizer_config = dict(grad_clip=dict(max_norm=25, norm_type=2)) -lr_config = dict( - policy='CosineAnnealing', - warmup='linear', - warmup_iters=500, - warmup_ratio=0.3333333333333333, - min_lr_ratio=0.001) -runner = dict(type='IterBasedRunner', max_iters=17500) -vis_pipeline = [ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='Collect', keys=['img'], meta_keys=['timestamp', 'lidar2img']) -] -evaluation = dict( - interval=3500, - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict( - type='Collect', keys=['img'], meta_keys=['timestamp', 'lidar2img']) - ]) -gpu_ids = range(0, 8) - -2025-11-12 20:05:02,815 - mmdet - INFO - Set random seed to 0, deterministic: False -2025-11-12 20:05:03,123 - mmdet - INFO - initialize ResNet with init_cfg {'type': 'Pretrained', 'checkpoint': 'ckpt/resnet50-19c8e357.pth'} -2025-11-12 20:05:03,324 - mmdet - INFO - initialize FPN with init_cfg {'type': 'Xavier', 'layer': 'Conv2d', 'distribution': 'uniform'} -Name of parameter - Initialization information - -img_backbone.conv1.weight - torch.Size([64, 3, 7, 7]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.bn1.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.bn1.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.conv1.weight - torch.Size([64, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn1.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn1.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.conv2.weight - torch.Size([64, 64, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn2.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn2.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.conv3.weight - torch.Size([256, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn3.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn3.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.downsample.0.weight - torch.Size([256, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.downsample.1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.downsample.1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.conv1.weight - torch.Size([64, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn1.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn1.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.conv2.weight - torch.Size([64, 64, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn2.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn2.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.conv3.weight - torch.Size([256, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn3.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn3.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.conv1.weight - torch.Size([64, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn1.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn1.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.conv2.weight - torch.Size([64, 64, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn2.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn2.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.conv3.weight - torch.Size([256, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn3.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn3.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.conv1.weight - torch.Size([128, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn1.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn1.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.conv2.weight - torch.Size([128, 128, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn2.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn2.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.conv3.weight - torch.Size([512, 128, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn3.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn3.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.downsample.0.weight - torch.Size([512, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.downsample.1.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.downsample.1.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.conv1.weight - torch.Size([128, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn1.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn1.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.conv2.weight - torch.Size([128, 128, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn2.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn2.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.conv3.weight - torch.Size([512, 128, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn3.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn3.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.conv1.weight - torch.Size([128, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn1.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn1.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.conv2.weight - torch.Size([128, 128, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn2.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn2.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.conv3.weight - torch.Size([512, 128, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn3.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn3.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.conv1.weight - torch.Size([128, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn1.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn1.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.conv2.weight - torch.Size([128, 128, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn2.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn2.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.conv3.weight - torch.Size([512, 128, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn3.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn3.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.conv1.weight - torch.Size([256, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.downsample.0.weight - torch.Size([1024, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.downsample.1.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.downsample.1.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.conv1.weight - torch.Size([512, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn1.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn1.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.conv2.weight - torch.Size([512, 512, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn2.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn2.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.conv3.weight - torch.Size([2048, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn3.weight - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn3.bias - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.downsample.0.weight - torch.Size([2048, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.downsample.1.weight - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.downsample.1.bias - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.conv1.weight - torch.Size([512, 2048, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn1.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn1.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.conv2.weight - torch.Size([512, 512, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn2.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn2.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.conv3.weight - torch.Size([2048, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn3.weight - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn3.bias - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.conv1.weight - torch.Size([512, 2048, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn1.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn1.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.conv2.weight - torch.Size([512, 512, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn2.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn2.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.conv3.weight - torch.Size([2048, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn3.weight - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn3.bias - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_neck.lateral_convs.0.conv.weight - torch.Size([256, 256, 1, 1]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.lateral_convs.0.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.lateral_convs.1.conv.weight - torch.Size([256, 512, 1, 1]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.lateral_convs.1.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.lateral_convs.2.conv.weight - torch.Size([256, 1024, 1, 1]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.lateral_convs.2.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.lateral_convs.3.conv.weight - torch.Size([256, 2048, 1, 1]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.lateral_convs.3.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.fpn_convs.0.conv.weight - torch.Size([256, 256, 3, 3]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.fpn_convs.0.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.fpn_convs.1.conv.weight - torch.Size([256, 256, 3, 3]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.fpn_convs.1.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.fpn_convs.2.conv.weight - torch.Size([256, 256, 3, 3]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.fpn_convs.2.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.fpn_convs.3.conv.weight - torch.Size([256, 256, 3, 3]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.fpn_convs.3.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.instance_bank.anchor - torch.Size([900, 11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.instance_bank.instance_feature - torch.Size([900, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.instance_bank.anchor_handler.fix_scale - torch.Size([1, 3]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.0.weight - torch.Size([128, 3]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.0.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.2.weight - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.2.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.3.weight - torch.Size([128, 128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.3.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.5.weight - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.5.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.6.weight - torch.Size([128, 128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.6.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.8.weight - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.8.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.9.weight - torch.Size([128, 128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.9.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.11.weight - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.11.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.0.weight - torch.Size([32, 3]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.0.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.2.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.2.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.3.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.3.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.5.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.5.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.6.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.6.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.8.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.8.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.9.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.9.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.11.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.11.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.0.weight - torch.Size([32, 2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.0.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.2.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.2.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.3.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.3.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.5.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.5.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.6.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.6.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.8.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.8.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.9.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.9.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.11.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.11.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.0.weight - torch.Size([64, 3]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.0.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.2.weight - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.2.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.3.weight - torch.Size([64, 64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.3.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.5.weight - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.5.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.6.weight - torch.Size([64, 64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.6.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.8.weight - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.8.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.9.weight - torch.Size([64, 64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.9.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.11.weight - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.11.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.1.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.1.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.1.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.1.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.1.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.1.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.1.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.1.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.3.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.4.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.4.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.4.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.4.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.5.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.5.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.5.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.5.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.6.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.6.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.8.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.8.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.8.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.8.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.8.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.8.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.8.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.8.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.10.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.11.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.11.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.11.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.11.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.12.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.12.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.12.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.12.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.13.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.13.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.15.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.15.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.15.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.15.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.15.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.15.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.15.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.15.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.16.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.16.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.17.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.18.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.18.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.18.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.18.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.19.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.19.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.19.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.19.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.20.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.20.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.22.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.22.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.22.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.22.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.22.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.22.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.22.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.22.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.23.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.23.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.24.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.25.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.25.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.25.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.25.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.26.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.26.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.26.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.26.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.27.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.27.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.29.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.29.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.29.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.29.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.29.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.29.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.29.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.29.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.30.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.30.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.31.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.32.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.32.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.32.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.32.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.33.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.33.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.33.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.33.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.34.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.34.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.36.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.36.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.36.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.36.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.36.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.36.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.36.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.36.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.37.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.37.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.38.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.fc_before.weight - torch.Size([512, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.fc_after.weight - torch.Size([256, 512]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.0.weight - torch.Size([1, 256, 1, 1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.0.bias - torch.Size([1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.1.weight - torch.Size([1, 256, 1, 1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.1.bias - torch.Size([1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.2.weight - torch.Size([1, 256, 1, 1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.2.bias - torch.Size([1]): -The value is the same before and after calling `init_weights` of Sparse4D -2025-11-12 20:05:03,415 - mmdet - INFO - Model: -Sparse4D( - (img_backbone): ResNet( - (conv1): Conv2d(3, 64, kernel_size=(7, 7), stride=(2, 2), padding=(3, 3), bias=False) - (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (maxpool): MaxPool2d(kernel_size=3, stride=2, padding=1, dilation=1, ceil_mode=False) - (layer1): ResLayer( - (0): Bottleneck( - (conv1): Conv2d(64, 64, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (downsample): Sequential( - (0): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - ) - ) - (1): Bottleneck( - (conv1): Conv2d(256, 64, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (2): Bottleneck( - (conv1): Conv2d(256, 64, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - ) - (layer2): ResLayer( - (0): Bottleneck( - (conv1): Conv2d(256, 128, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (downsample): Sequential( - (0): Conv2d(256, 512, kernel_size=(1, 1), stride=(2, 2), bias=False) - (1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - ) - ) - (1): Bottleneck( - (conv1): Conv2d(512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (2): Bottleneck( - (conv1): Conv2d(512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (3): Bottleneck( - (conv1): Conv2d(512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - ) - (layer3): ResLayer( - (0): Bottleneck( - (conv1): Conv2d(512, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (downsample): Sequential( - (0): Conv2d(512, 1024, kernel_size=(1, 1), stride=(2, 2), bias=False) - (1): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - ) - ) - (1): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (2): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (3): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (4): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (5): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - ) - (layer4): ResLayer( - (0): Bottleneck( - (conv1): Conv2d(1024, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(512, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (downsample): Sequential( - (0): Conv2d(1024, 2048, kernel_size=(1, 1), stride=(2, 2), bias=False) - (1): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - ) - ) - (1): Bottleneck( - (conv1): Conv2d(2048, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(512, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (2): Bottleneck( - (conv1): Conv2d(2048, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(512, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - ) - ) - init_cfg={'type': 'Pretrained', 'checkpoint': 'ckpt/resnet50-19c8e357.pth'} - (img_neck): FPN( - (lateral_convs): ModuleList( - (0): ConvModule( - (conv): Conv2d(256, 256, kernel_size=(1, 1), stride=(1, 1)) - ) - (1): ConvModule( - (conv): Conv2d(512, 256, kernel_size=(1, 1), stride=(1, 1)) - ) - (2): ConvModule( - (conv): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1)) - ) - (3): ConvModule( - (conv): Conv2d(2048, 256, kernel_size=(1, 1), stride=(1, 1)) - ) - ) - (fpn_convs): ModuleList( - (0-3): 4 x ConvModule( - (conv): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1)) - ) - ) - ) - init_cfg={'type': 'Xavier', 'layer': 'Conv2d', 'distribution': 'uniform'} - (head): Sparse4DHead( - (instance_bank): InstanceBank( - (anchor_handler): SparseBox3DKeyPointsGenerator() - ) - (anchor_encoder): SparseBox3DEncoder( - (pos_fc): Sequential( - (0): Linear(in_features=3, out_features=128, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((128,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=128, out_features=128, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((128,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=128, out_features=128, bias=True) - (7): ReLU(inplace=True) - (8): LayerNorm((128,), eps=1e-05, elementwise_affine=True) - (9): Linear(in_features=128, out_features=128, bias=True) - (10): ReLU(inplace=True) - (11): LayerNorm((128,), eps=1e-05, elementwise_affine=True) - ) - (size_fc): Sequential( - (0): Linear(in_features=3, out_features=32, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=32, out_features=32, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=32, out_features=32, bias=True) - (7): ReLU(inplace=True) - (8): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (9): Linear(in_features=32, out_features=32, bias=True) - (10): ReLU(inplace=True) - (11): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - ) - (yaw_fc): Sequential( - (0): Linear(in_features=2, out_features=32, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=32, out_features=32, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=32, out_features=32, bias=True) - (7): ReLU(inplace=True) - (8): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (9): Linear(in_features=32, out_features=32, bias=True) - (10): ReLU(inplace=True) - (11): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - ) - (vel_fc): Sequential( - (0): Linear(in_features=3, out_features=64, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((64,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=64, out_features=64, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((64,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=64, out_features=64, bias=True) - (7): ReLU(inplace=True) - (8): LayerNorm((64,), eps=1e-05, elementwise_affine=True) - (9): Linear(in_features=64, out_features=64, bias=True) - (10): ReLU(inplace=True) - (11): LayerNorm((64,), eps=1e-05, elementwise_affine=True) - ) - ) - (loss_cls): FocalLoss() - (loss_reg): SparseBox3DLoss( - (loss_box): L1Loss() - (loss_cns): CrossEntropyLoss(avg_non_ignore=False) - (loss_yns): GaussianFocalLoss() - ) - (layers): ModuleList( - (0): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (1): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (4-5): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (6): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (7): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (8): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (11-12): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (13): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (14): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (15): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (16): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (17): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (18-19): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (20): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (21): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (22): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (23): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (24): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (25-26): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (27): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (28): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (29): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (30): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (31): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (32-33): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (34): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (35): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (36): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (37): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (38): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - ) - (fc_before): Linear(in_features=256, out_features=512, bias=False) - (fc_after): Linear(in_features=512, out_features=256, bias=False) - ) - (depth_branch): DenseDepthNet( - (depth_layers): ModuleList( - (0-2): 3 x Conv2d(256, 1, kernel_size=(1, 1), stride=(1, 1)) - ) - ) - (grid_mask): GridMask() -) -2025-11-12 20:05:15,999 - mmdet - INFO - Start running, host: root@VM-120-96-tencentos, work_dir: /cfs/auto/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704 -2025-11-12 20:05:15,999 - mmdet - INFO - Hooks will be executed in the following order: -before_run: -(VERY_HIGH ) CosineAnnealingLrUpdaterHook -(ABOVE_NORMAL) Fp16OptimizerHook -(NORMAL ) CheckpointHook -(NORMAL ) CustomDistEvalHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -before_train_epoch: -(VERY_HIGH ) CosineAnnealingLrUpdaterHook -(NORMAL ) CustomDistEvalHook -(LOW ) IterTimerHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -before_train_iter: -(VERY_HIGH ) CosineAnnealingLrUpdaterHook -(NORMAL ) CustomDistEvalHook -(LOW ) IterTimerHook - -------------------- -after_train_iter: -(ABOVE_NORMAL) Fp16OptimizerHook -(NORMAL ) CheckpointHook -(NORMAL ) CustomDistEvalHook -(LOW ) IterTimerHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -after_train_epoch: -(NORMAL ) CheckpointHook -(NORMAL ) CustomDistEvalHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -before_val_epoch: -(LOW ) IterTimerHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -before_val_iter: -(LOW ) IterTimerHook - -------------------- -after_val_iter: -(LOW ) IterTimerHook - -------------------- -after_val_epoch: -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -after_run: -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -2025-11-12 20:05:16,000 - mmdet - INFO - workflow: [('train', 1)], max: 17500 iters -2025-11-12 20:05:16,002 - mmdet - INFO - Checkpoints will be saved to /cfs/auto/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704 by HardDiskBackend. -2025-11-12 20:07:13,228 - mmdet - INFO - Iter [1/17500] lr: 1.000e-04, eta: 23 days, 13:36:20, time: 116.360, data_time: 10.763, memory: 49163, loss_cls_0: 2.3613, loss_box_0: 0.0138, loss_cns_0: 0.0027, loss_yns_0: 0.0008, loss_cls_1: 2.1544, loss_box_1: 0.1078, loss_cns_1: 0.0245, loss_yns_1: 0.0066, loss_cls_2: 2.3121, loss_box_2: 0.0041, loss_cns_2: 0.0005, loss_yns_2: 0.0003, loss_cls_3: 2.3905, loss_box_3: 0.0294, loss_cns_3: 0.0050, loss_yns_3: 0.0014, loss_cls_4: 2.0286, loss_box_4: 0.4134, loss_cns_4: 0.0530, loss_yns_4: 0.0250, loss_cls_5: 2.4250, loss_box_5: 0.0190, loss_cns_5: 0.0024, loss_yns_5: 0.0017, loss_cls_dn_0: 1.1980, loss_box_dn_0: 1.4603, loss_cls_dn_1: 1.1102, loss_box_dn_1: 1.7318, loss_cls_dn_2: 1.1741, loss_box_dn_2: 1.9718, loss_cls_dn_3: 1.1721, loss_box_dn_3: 2.2418, loss_cls_dn_4: 1.0528, loss_box_dn_4: 2.4268, loss_cls_dn_5: 1.2387, loss_box_dn_5: 2.6773, loss_dense_depth: 1.8643, loss: 35.7037, grad_norm: 273.3042 -2025-11-12 20:07:15,157 - mmdet - INFO - Iter [2/17500] lr: 1.004e-04, eta: 11 days, 23:33:32, time: 1.964, data_time: 0.103, memory: 49163, loss_cls_0: 2.0910, loss_box_0: 0.0234, loss_cns_0: 0.0060, loss_yns_0: 0.0023, loss_cls_1: 2.0875, loss_box_1: 0.0790, loss_cns_1: 0.0173, loss_yns_1: 0.0046, loss_cls_2: 2.1739, loss_box_2: 0.0944, loss_cns_2: 0.0084, loss_yns_2: 0.0039, loss_cls_3: 2.0273, loss_box_3: 0.1372, loss_cns_3: 0.0205, loss_yns_3: 0.0064, loss_cls_4: 1.8416, loss_box_4: 0.9727, loss_cns_4: 0.1054, loss_yns_4: 0.0371, loss_cls_5: 2.1238, loss_box_5: 0.2947, loss_cns_5: 0.0334, loss_yns_5: 0.0117, loss_cls_dn_0: 1.0513, loss_box_dn_0: 1.2540, loss_cls_dn_1: 0.9779, loss_box_dn_1: 2.3923, loss_cls_dn_2: 0.9979, loss_box_dn_2: 2.4732, loss_cls_dn_3: 0.9459, loss_box_dn_3: 2.5040, loss_cls_dn_4: 0.8587, loss_box_dn_4: 2.6549, loss_cls_dn_5: 1.0106, loss_box_dn_5: 2.8527, loss_dense_depth: 1.7266, loss: 35.9033, grad_norm: 60.8044 -2025-11-12 20:07:16,702 - mmdet - INFO - Iter [3/17500] lr: 1.008e-04, eta: 8 days, 2:11:43, time: 1.543, data_time: 0.076, memory: 49163, loss_cls_0: 1.5284, loss_box_0: 2.4217, loss_cns_0: 0.5859, loss_yns_0: 0.2238, loss_cls_1: 1.8416, loss_box_1: 0.9230, loss_cns_1: 0.1672, loss_yns_1: 0.0551, loss_cls_2: 1.8635, loss_box_2: 2.8887, loss_cns_2: 0.2831, loss_yns_2: 0.1489, loss_cls_3: 1.6627, loss_box_3: 3.9145, loss_cns_3: 0.4384, loss_yns_3: 0.1831, loss_cls_4: 1.6174, loss_box_4: 3.2098, loss_cns_4: 0.3387, loss_yns_4: 0.1381, loss_cls_5: 1.7637, loss_box_5: 1.7525, loss_cns_5: 0.1627, loss_yns_5: 0.0626, loss_cls_dn_0: 0.7473, loss_box_dn_0: 1.1840, loss_cls_dn_1: 0.8658, loss_box_dn_1: 2.1939, loss_cls_dn_2: 0.8436, loss_box_dn_2: 2.3596, loss_cls_dn_3: 0.7448, loss_box_dn_3: 2.5418, loss_cls_dn_4: 0.7465, loss_box_dn_4: 2.7795, loss_cls_dn_5: 0.8380, loss_box_dn_5: 2.9764, loss_dense_depth: 1.7304, loss: 48.7267, grad_norm: 90.0063 -2025-11-12 20:07:18,225 - mmdet - INFO - Iter [4/17500] lr: 1.012e-04, eta: 6 days, 3:29:21, time: 1.523, data_time: 0.090, memory: 49163, loss_cls_0: 1.3740, loss_box_0: 2.5241, loss_cns_0: 0.5720, loss_yns_0: 0.1767, loss_cls_1: 1.6826, loss_box_1: 2.5736, loss_cns_1: 0.4099, loss_yns_1: 0.1545, loss_cls_2: 1.7257, loss_box_2: 3.5218, loss_cns_2: 0.4378, loss_yns_2: 0.1865, loss_cls_3: 1.5319, loss_box_3: 3.9277, loss_cns_3: 0.4852, loss_yns_3: 0.2096, loss_cls_4: 1.4623, loss_box_4: 4.4370, loss_cns_4: 0.3999, loss_yns_4: 0.1940, loss_cls_5: 1.5618, loss_box_5: 4.2082, loss_cns_5: 0.4431, loss_yns_5: 0.1779, loss_cls_dn_0: 0.6044, loss_box_dn_0: 1.1794, loss_cls_dn_1: 0.7645, loss_box_dn_1: 2.3304, loss_cls_dn_2: 0.7229, loss_box_dn_2: 2.3941, loss_cls_dn_3: 0.6345, loss_box_dn_3: 2.5691, loss_cls_dn_4: 0.6279, loss_box_dn_4: 2.7675, loss_cls_dn_5: 0.7071, loss_box_dn_5: 2.9333, loss_dense_depth: 1.6199, loss: 54.2331, grad_norm: 100.7352 -2025-11-12 20:07:19,748 - mmdet - INFO - Iter [5/17500] lr: 1.016e-04, eta: 4 days, 23:27:52, time: 1.523, data_time: 0.078, memory: 49163, loss_cls_0: 1.3696, loss_box_0: 2.7170, loss_cns_0: 0.5137, loss_yns_0: 0.1892, loss_cls_1: 1.5650, loss_box_1: 3.6089, loss_cns_1: 0.4283, loss_yns_1: 0.2004, loss_cls_2: 1.6857, loss_box_2: 3.7582, loss_cns_2: 0.4020, loss_yns_2: 0.1945, loss_cls_3: 1.4914, loss_box_3: 3.8409, loss_cns_3: 0.4151, loss_yns_3: 0.1916, loss_cls_4: 1.4211, loss_box_4: 4.0110, loss_cns_4: 0.3898, loss_yns_4: 0.1914, loss_cls_5: 1.4078, loss_box_5: 4.1709, loss_cns_5: 0.4286, loss_yns_5: 0.1971, loss_cls_dn_0: 0.5291, loss_box_dn_0: 1.2160, loss_cls_dn_1: 0.6923, loss_box_dn_1: 2.0878, loss_cls_dn_2: 0.6645, loss_box_dn_2: 2.1535, loss_cls_dn_3: 0.5719, loss_box_dn_3: 2.2582, loss_cls_dn_4: 0.5605, loss_box_dn_4: 2.4089, loss_cls_dn_5: 0.6058, loss_box_dn_5: 2.5008, loss_dense_depth: 1.5783, loss: 52.6171, grad_norm: 100.9737 -2025-11-12 20:07:21,305 - mmdet - INFO - Iter [6/17500] lr: 1.020e-04, eta: 4 days, 4:48:29, time: 1.556, data_time: 0.080, memory: 49163, loss_cls_0: 1.3456, loss_box_0: 2.4624, loss_cns_0: 0.6064, loss_yns_0: 0.1835, loss_cls_1: 1.4826, loss_box_1: 3.6179, loss_cns_1: 0.4145, loss_yns_1: 0.1909, loss_cls_2: 1.5617, loss_box_2: 3.8708, loss_cns_2: 0.3751, loss_yns_2: 0.1984, loss_cls_3: 1.3797, loss_box_3: 3.8665, loss_cns_3: 0.3669, loss_yns_3: 0.1892, loss_cls_4: 1.3772, loss_box_4: 4.1179, loss_cns_4: 0.3193, loss_yns_4: 0.1927, loss_cls_5: 1.3252, loss_box_5: 4.2906, loss_cns_5: 0.3186, loss_yns_5: 0.1985, loss_cls_dn_0: 0.5129, loss_box_dn_0: 1.1262, loss_cls_dn_1: 0.6140, loss_box_dn_1: 2.1841, loss_cls_dn_2: 0.5927, loss_box_dn_2: 2.2243, loss_cls_dn_3: 0.5182, loss_box_dn_3: 2.2543, loss_cls_dn_4: 0.4825, loss_box_dn_4: 2.4340, loss_cls_dn_5: 0.5042, loss_box_dn_5: 2.5306, loss_dense_depth: 1.4534, loss: 51.6832, grad_norm: 110.0105 -2025-11-12 20:07:22,838 - mmdet - INFO - Iter [7/17500] lr: 1.024e-04, eta: 3 days, 15:28:03, time: 1.535, data_time: 0.083, memory: 49163, loss_cls_0: 1.2958, loss_box_0: 2.3196, loss_cns_0: 0.6814, loss_yns_0: 0.1756, loss_cls_1: 1.3899, loss_box_1: 3.5972, loss_cns_1: 0.4405, loss_yns_1: 0.1918, loss_cls_2: 1.4334, loss_box_2: 3.7346, loss_cns_2: 0.4329, loss_yns_2: 0.1915, loss_cls_3: 1.3133, loss_box_3: 3.5864, loss_cns_3: 0.4414, loss_yns_3: 0.1797, loss_cls_4: 1.2872, loss_box_4: 3.7354, loss_cns_4: 0.4443, loss_yns_4: 0.1848, loss_cls_5: 1.3239, loss_box_5: 4.0127, loss_cns_5: 0.4445, loss_yns_5: 0.1900, loss_cls_dn_0: 0.5166, loss_box_dn_0: 1.0581, loss_cls_dn_1: 0.5441, loss_box_dn_1: 2.2512, loss_cls_dn_2: 0.5463, loss_box_dn_2: 2.2655, loss_cls_dn_3: 0.4810, loss_box_dn_3: 2.2494, loss_cls_dn_4: 0.4504, loss_box_dn_4: 2.3683, loss_cls_dn_5: 0.4409, loss_box_dn_5: 2.4838, loss_dense_depth: 1.4796, loss: 50.1631, grad_norm: 99.4865 -2025-11-12 20:07:24,383 - mmdet - INFO - Iter [8/17500] lr: 1.028e-04, eta: 3 days, 5:28:02, time: 1.543, data_time: 0.079, memory: 49163, loss_cls_0: 1.2736, loss_box_0: 2.2404, loss_cns_0: 0.6592, loss_yns_0: 0.1772, loss_cls_1: 1.3118, loss_box_1: 3.5004, loss_cns_1: 0.4688, loss_yns_1: 0.1901, loss_cls_2: 1.3739, loss_box_2: 3.6015, loss_cns_2: 0.4532, loss_yns_2: 0.1845, loss_cls_3: 1.3032, loss_box_3: 3.5334, loss_cns_3: 0.4701, loss_yns_3: 0.1857, loss_cls_4: 1.3015, loss_box_4: 3.5316, loss_cns_4: 0.4629, loss_yns_4: 0.1822, loss_cls_5: 1.3243, loss_box_5: 3.6177, loss_cns_5: 0.4911, loss_yns_5: 0.1928, loss_cls_dn_0: 0.5191, loss_box_dn_0: 1.0035, loss_cls_dn_1: 0.5426, loss_box_dn_1: 1.6633, loss_cls_dn_2: 0.5692, loss_box_dn_2: 1.6930, loss_cls_dn_3: 0.4990, loss_box_dn_3: 1.6901, loss_cls_dn_4: 0.4787, loss_box_dn_4: 1.7049, loss_cls_dn_5: 0.4424, loss_box_dn_5: 1.7689, loss_dense_depth: 1.3827, loss: 45.9885, grad_norm: 78.6110 -2025-11-12 20:07:25,891 - mmdet - INFO - Iter [9/17500] lr: 1.032e-04, eta: 2 days, 21:40:16, time: 1.510, data_time: 0.078, memory: 49163, loss_cls_0: 1.2588, loss_box_0: 2.2329, loss_cns_0: 0.6263, loss_yns_0: 0.1762, loss_cls_1: 1.2664, loss_box_1: 3.2705, loss_cns_1: 0.5048, loss_yns_1: 0.1796, loss_cls_2: 1.3743, loss_box_2: 3.2958, loss_cns_2: 0.4714, loss_yns_2: 0.1818, loss_cls_3: 1.2671, loss_box_3: 3.3872, loss_cns_3: 0.5000, loss_yns_3: 0.2175, loss_cls_4: 1.2574, loss_box_4: 3.4021, loss_cns_4: 0.5006, loss_yns_4: 0.1938, loss_cls_5: 1.3052, loss_box_5: 3.4520, loss_cns_5: 0.5044, loss_yns_5: 0.1812, loss_cls_dn_0: 0.5123, loss_box_dn_0: 1.0187, loss_cls_dn_1: 0.4949, loss_box_dn_1: 1.4901, loss_cls_dn_2: 0.5498, loss_box_dn_2: 1.5429, loss_cls_dn_3: 0.4786, loss_box_dn_3: 1.6353, loss_cls_dn_4: 0.4616, loss_box_dn_4: 1.6412, loss_cls_dn_5: 0.4305, loss_box_dn_5: 1.7394, loss_dense_depth: 1.3672, loss: 44.3697, grad_norm: 71.3364 -2025-11-12 20:07:27,421 - mmdet - INFO - Iter [10/17500] lr: 1.036e-04, eta: 2 days, 15:26:36, time: 1.529, data_time: 0.077, memory: 49163, loss_cls_0: 1.2264, loss_box_0: 2.2520, loss_cns_0: 0.6152, loss_yns_0: 0.1727, loss_cls_1: 1.2647, loss_box_1: 3.1757, loss_cns_1: 0.5159, loss_yns_1: 0.1830, loss_cls_2: 1.2915, loss_box_2: 3.1361, loss_cns_2: 0.4893, loss_yns_2: 0.1887, loss_cls_3: 1.2443, loss_box_3: 3.2282, loss_cns_3: 0.5523, loss_yns_3: 0.1876, loss_cls_4: 1.2518, loss_box_4: 3.2652, loss_cns_4: 0.5498, loss_yns_4: 0.1820, loss_cls_5: 1.2774, loss_box_5: 3.2760, loss_cns_5: 0.5379, loss_yns_5: 0.1834, loss_cls_dn_0: 0.5011, loss_box_dn_0: 1.0593, loss_cls_dn_1: 0.4490, loss_box_dn_1: 1.8082, loss_cls_dn_2: 0.5021, loss_box_dn_2: 1.8527, loss_cls_dn_3: 0.4539, loss_box_dn_3: 1.9158, loss_cls_dn_4: 0.4272, loss_box_dn_4: 1.9409, loss_cls_dn_5: 0.4285, loss_box_dn_5: 2.0431, loss_dense_depth: 1.3385, loss: 44.9675, grad_norm: 64.4576 -2025-11-12 20:07:28,941 - mmdet - INFO - Iter [11/17500] lr: 1.040e-04, eta: 2 days, 10:20:37, time: 1.520, data_time: 0.078, memory: 49163, loss_cls_0: 1.2167, loss_box_0: 2.2917, loss_cns_0: 0.6272, loss_yns_0: 0.1735, loss_cls_1: 1.2707, loss_box_1: 3.1006, loss_cns_1: 0.5204, loss_yns_1: 0.1779, loss_cls_2: 1.2733, loss_box_2: 3.0365, loss_cns_2: 0.4940, loss_yns_2: 0.1859, loss_cls_3: 1.2468, loss_box_3: 3.0196, loss_cns_3: 0.5733, loss_yns_3: 0.1794, loss_cls_4: 1.2390, loss_box_4: 3.1132, loss_cns_4: 0.5562, loss_yns_4: 0.1811, loss_cls_5: 1.2896, loss_box_5: 3.3394, loss_cns_5: 0.5228, loss_yns_5: 0.1808, loss_cls_dn_0: 0.4886, loss_box_dn_0: 1.0852, loss_cls_dn_1: 0.4218, loss_box_dn_1: 2.1759, loss_cls_dn_2: 0.4470, loss_box_dn_2: 2.1771, loss_cls_dn_3: 0.4227, loss_box_dn_3: 2.1623, loss_cls_dn_4: 0.4032, loss_box_dn_4: 2.1758, loss_cls_dn_5: 0.4142, loss_box_dn_5: 2.3095, loss_dense_depth: 1.3814, loss: 45.8744, grad_norm: 78.4923 -2025-11-12 20:07:30,453 - mmdet - INFO - Iter [12/17500] lr: 1.044e-04, eta: 2 days, 6:05:26, time: 1.512, data_time: 0.079, memory: 49163, loss_cls_0: 1.2331, loss_box_0: 2.2982, loss_cns_0: 0.6264, loss_yns_0: 0.1732, loss_cls_1: 1.2656, loss_box_1: 2.9224, loss_cns_1: 0.5136, loss_yns_1: 0.1786, loss_cls_2: 1.3124, loss_box_2: 2.8859, loss_cns_2: 0.5022, loss_yns_2: 0.1865, loss_cls_3: 1.2604, loss_box_3: 2.9510, loss_cns_3: 0.5424, loss_yns_3: 0.1790, loss_cls_4: 1.2383, loss_box_4: 3.0226, loss_cns_4: 0.5266, loss_yns_4: 0.1804, loss_cls_5: 1.2872, loss_box_5: 3.1802, loss_cns_5: 0.5157, loss_yns_5: 0.1789, loss_cls_dn_0: 0.4795, loss_box_dn_0: 1.0860, loss_cls_dn_1: 0.3976, loss_box_dn_1: 2.4937, loss_cls_dn_2: 0.4091, loss_box_dn_2: 2.4443, loss_cls_dn_3: 0.3931, loss_box_dn_3: 2.4248, loss_cls_dn_4: 0.3819, loss_box_dn_4: 2.4263, loss_cls_dn_5: 0.3969, loss_box_dn_5: 2.4998, loss_dense_depth: 1.2749, loss: 46.2689, grad_norm: 63.2691 -2025-11-12 20:07:31,976 - mmdet - INFO - Iter [13/17500] lr: 1.048e-04, eta: 2 days, 2:29:46, time: 1.524, data_time: 0.081, memory: 49163, loss_cls_0: 1.2401, loss_box_0: 2.3143, loss_cns_0: 0.5970, loss_yns_0: 0.1710, loss_cls_1: 1.2444, loss_box_1: 2.7622, loss_cns_1: 0.5293, loss_yns_1: 0.1772, loss_cls_2: 1.3162, loss_box_2: 2.8380, loss_cns_2: 0.5288, loss_yns_2: 0.1809, loss_cls_3: 1.2460, loss_box_3: 2.9518, loss_cns_3: 0.5421, loss_yns_3: 0.1773, loss_cls_4: 1.2531, loss_box_4: 2.9179, loss_cns_4: 0.5209, loss_yns_4: 0.1790, loss_cls_5: 1.2726, loss_box_5: 2.9110, loss_cns_5: 0.5205, loss_yns_5: 0.1757, loss_cls_dn_0: 0.4721, loss_box_dn_0: 1.0747, loss_cls_dn_1: 0.4261, loss_box_dn_1: 1.7728, loss_cls_dn_2: 0.4396, loss_box_dn_2: 1.8014, loss_cls_dn_3: 0.4329, loss_box_dn_3: 1.9251, loss_cls_dn_4: 0.4216, loss_box_dn_4: 2.0190, loss_cls_dn_5: 0.4491, loss_box_dn_5: 2.0121, loss_dense_depth: 1.3757, loss: 43.1898, grad_norm: 77.9191 -2025-11-12 20:07:33,501 - mmdet - INFO - Iter [14/17500] lr: 1.052e-04, eta: 1 day, 23:24:56, time: 1.525, data_time: 0.080, memory: 49163, loss_cls_0: 1.2528, loss_box_0: 2.3667, loss_cns_0: 0.5710, loss_yns_0: 0.1714, loss_cls_1: 1.2837, loss_box_1: 2.6549, loss_cns_1: 0.5442, loss_yns_1: 0.1777, loss_cls_2: 1.3127, loss_box_2: 2.7457, loss_cns_2: 0.5444, loss_yns_2: 0.1758, loss_cls_3: 1.2796, loss_box_3: 2.8546, loss_cns_3: 0.5530, loss_yns_3: 0.1745, loss_cls_4: 1.2975, loss_box_4: 2.8779, loss_cns_4: 0.5368, loss_yns_4: 0.1773, loss_cls_5: 1.2991, loss_box_5: 2.8649, loss_cns_5: 0.5533, loss_yns_5: 0.1772, loss_cls_dn_0: 0.4652, loss_box_dn_0: 1.0501, loss_cls_dn_1: 0.4429, loss_box_dn_1: 1.4028, loss_cls_dn_2: 0.4649, loss_box_dn_2: 1.4409, loss_cls_dn_3: 0.4524, loss_box_dn_3: 1.6012, loss_cls_dn_4: 0.4386, loss_box_dn_4: 1.7401, loss_cls_dn_5: 0.4680, loss_box_dn_5: 1.7073, loss_dense_depth: 1.3661, loss: 41.4872, grad_norm: 79.7711 -2025-11-12 20:07:35,012 - mmdet - INFO - Iter [15/17500] lr: 1.056e-04, eta: 1 day, 20:44:28, time: 1.510, data_time: 0.078, memory: 49163, loss_cls_0: 1.2566, loss_box_0: 2.4275, loss_cns_0: 0.5659, loss_yns_0: 0.1711, loss_cls_1: 1.3199, loss_box_1: 2.6553, loss_cns_1: 0.5723, loss_yns_1: 0.1783, loss_cls_2: 1.2747, loss_box_2: 2.6997, loss_cns_2: 0.5601, loss_yns_2: 0.1787, loss_cls_3: 1.3120, loss_box_3: 2.7069, loss_cns_3: 0.5709, loss_yns_3: 0.1782, loss_cls_4: 1.2924, loss_box_4: 2.7519, loss_cns_4: 0.5636, loss_yns_4: 0.1798, loss_cls_5: 1.2977, loss_box_5: 2.7531, loss_cns_5: 0.5798, loss_yns_5: 0.1737, loss_cls_dn_0: 0.4640, loss_box_dn_0: 1.0547, loss_cls_dn_1: 0.4434, loss_box_dn_1: 1.5012, loss_cls_dn_2: 0.4695, loss_box_dn_2: 1.4920, loss_cls_dn_3: 0.4450, loss_box_dn_3: 1.5769, loss_cls_dn_4: 0.4351, loss_box_dn_4: 1.6633, loss_cls_dn_5: 0.4629, loss_box_dn_5: 1.6342, loss_dense_depth: 1.4539, loss: 41.3163, grad_norm: 65.6057 -2025-11-12 20:07:36,524 - mmdet - INFO - Iter [16/17500] lr: 1.060e-04, eta: 1 day, 18:24:04, time: 1.512, data_time: 0.077, memory: 49163, loss_cls_0: 1.2246, loss_box_0: 2.3874, loss_cns_0: 0.5799, loss_yns_0: 0.1730, loss_cls_1: 1.2901, loss_box_1: 2.7004, loss_cns_1: 0.5887, loss_yns_1: 0.1737, loss_cls_2: 1.2678, loss_box_2: 2.7553, loss_cns_2: 0.5565, loss_yns_2: 0.1774, loss_cls_3: 1.3110, loss_box_3: 2.7956, loss_cns_3: 0.5759, loss_yns_3: 0.1748, loss_cls_4: 1.2686, loss_box_4: 2.7711, loss_cns_4: 0.5600, loss_yns_4: 0.1812, loss_cls_5: 1.2767, loss_box_5: 2.8043, loss_cns_5: 0.5712, loss_yns_5: 0.1781, loss_cls_dn_0: 0.4831, loss_box_dn_0: 1.0273, loss_cls_dn_1: 0.4647, loss_box_dn_1: 1.3963, loss_cls_dn_2: 0.4855, loss_box_dn_2: 1.3814, loss_cls_dn_3: 0.4420, loss_box_dn_3: 1.4548, loss_cls_dn_4: 0.4432, loss_box_dn_4: 1.4875, loss_cls_dn_5: 0.4750, loss_box_dn_5: 1.5146, loss_dense_depth: 1.2915, loss: 40.6902, grad_norm: 87.2290 -2025-11-12 20:07:38,038 - mmdet - INFO - Iter [17/17500] lr: 1.064e-04, eta: 1 day, 16:20:15, time: 1.515, data_time: 0.076, memory: 49163, loss_cls_0: 1.2104, loss_box_0: 2.3208, loss_cns_0: 0.5986, loss_yns_0: 0.1720, loss_cls_1: 1.2754, loss_box_1: 2.8831, loss_cns_1: 0.5668, loss_yns_1: 0.1739, loss_cls_2: 1.2749, loss_box_2: 2.9018, loss_cns_2: 0.5432, loss_yns_2: 0.1783, loss_cls_3: 1.2982, loss_box_3: 2.9393, loss_cns_3: 0.5725, loss_yns_3: 0.1738, loss_cls_4: 1.2452, loss_box_4: 2.9326, loss_cns_4: 0.5630, loss_yns_4: 0.1763, loss_cls_5: 1.2658, loss_box_5: 2.9757, loss_cns_5: 0.5481, loss_yns_5: 0.1744, loss_cls_dn_0: 0.4917, loss_box_dn_0: 1.0140, loss_cls_dn_1: 0.4613, loss_box_dn_1: 1.5142, loss_cls_dn_2: 0.4670, loss_box_dn_2: 1.5102, loss_cls_dn_3: 0.4245, loss_box_dn_3: 1.5730, loss_cls_dn_4: 0.4388, loss_box_dn_4: 1.6056, loss_cls_dn_5: 0.4741, loss_box_dn_5: 1.7165, loss_dense_depth: 1.3616, loss: 42.0166, grad_norm: 73.0541 -2025-11-12 20:07:39,563 - mmdet - INFO - Iter [18/17500] lr: 1.068e-04, eta: 1 day, 14:30:21, time: 1.525, data_time: 0.078, memory: 49163, loss_cls_0: 1.2060, loss_box_0: 2.2289, loss_cns_0: 0.6304, loss_yns_0: 0.1724, loss_cls_1: 1.2474, loss_box_1: 2.8837, loss_cns_1: 0.5501, loss_yns_1: 0.1776, loss_cls_2: 1.2721, loss_box_2: 2.8731, loss_cns_2: 0.5452, loss_yns_2: 0.1772, loss_cls_3: 1.2670, loss_box_3: 2.8404, loss_cns_3: 0.5635, loss_yns_3: 0.1722, loss_cls_4: 1.2339, loss_box_4: 2.8674, loss_cns_4: 0.5693, loss_yns_4: 0.1726, loss_cls_5: 1.2636, loss_box_5: 2.8980, loss_cns_5: 0.5384, loss_yns_5: 0.1721, loss_cls_dn_0: 0.4920, loss_box_dn_0: 0.9991, loss_cls_dn_1: 0.4569, loss_box_dn_1: 1.5619, loss_cls_dn_2: 0.4601, loss_box_dn_2: 1.5868, loss_cls_dn_3: 0.4235, loss_box_dn_3: 1.6273, loss_cls_dn_4: 0.4423, loss_box_dn_4: 1.6708, loss_cls_dn_5: 0.4708, loss_box_dn_5: 1.8353, loss_dense_depth: 1.2024, loss: 41.7516, grad_norm: 63.5673 -2025-11-12 20:07:41,086 - mmdet - INFO - Iter [19/17500] lr: 1.072e-04, eta: 1 day, 12:51:58, time: 1.523, data_time: 0.074, memory: 49163, loss_cls_0: 1.2223, loss_box_0: 2.1801, loss_cns_0: 0.6306, loss_yns_0: 0.1719, loss_cls_1: 1.2421, loss_box_1: 2.9120, loss_cns_1: 0.5287, loss_yns_1: 0.1736, loss_cls_2: 1.2718, loss_box_2: 2.9304, loss_cns_2: 0.5370, loss_yns_2: 0.1746, loss_cls_3: 1.2547, loss_box_3: 2.9187, loss_cns_3: 0.5460, loss_yns_3: 0.1734, loss_cls_4: 1.2439, loss_box_4: 2.9297, loss_cns_4: 0.5698, loss_yns_4: 0.1740, loss_cls_5: 1.2683, loss_box_5: 3.0522, loss_cns_5: 0.5274, loss_yns_5: 0.1740, loss_cls_dn_0: 0.4857, loss_box_dn_0: 1.0004, loss_cls_dn_1: 0.4666, loss_box_dn_1: 1.3304, loss_cls_dn_2: 0.4676, loss_box_dn_2: 1.4162, loss_cls_dn_3: 0.4469, loss_box_dn_3: 1.4507, loss_cls_dn_4: 0.4560, loss_box_dn_4: 1.4991, loss_cls_dn_5: 0.4775, loss_box_dn_5: 1.6979, loss_dense_depth: 1.3290, loss: 41.3309, grad_norm: 78.5918 -2025-11-12 20:07:42,614 - mmdet - INFO - Iter [20/17500] lr: 1.076e-04, eta: 1 day, 11:23:30, time: 1.527, data_time: 0.072, memory: 49163, loss_cls_0: 1.2257, loss_box_0: 2.1540, loss_cns_0: 0.6269, loss_yns_0: 0.1713, loss_cls_1: 1.2343, loss_box_1: 3.0547, loss_cns_1: 0.5126, loss_yns_1: 0.1745, loss_cls_2: 1.2671, loss_box_2: 3.0790, loss_cns_2: 0.5252, loss_yns_2: 0.1740, loss_cls_3: 1.2453, loss_box_3: 3.1077, loss_cns_3: 0.5291, loss_yns_3: 0.1723, loss_cls_4: 1.2429, loss_box_4: 3.1049, loss_cns_4: 0.5460, loss_yns_4: 0.1750, loss_cls_5: 1.2674, loss_box_5: 3.1993, loss_cns_5: 0.5255, loss_yns_5: 0.1760, loss_cls_dn_0: 0.4760, loss_box_dn_0: 0.9861, loss_cls_dn_1: 0.4303, loss_box_dn_1: 1.5404, loss_cls_dn_2: 0.4302, loss_box_dn_2: 1.5863, loss_cls_dn_3: 0.4188, loss_box_dn_3: 1.5813, loss_cls_dn_4: 0.4201, loss_box_dn_4: 1.5927, loss_cls_dn_5: 0.4309, loss_box_dn_5: 1.6992, loss_dense_depth: 1.2269, loss: 42.3101, grad_norm: 77.4070 -2025-11-12 20:07:44,256 - mmdet - INFO - Iter [21/17500] lr: 1.080e-04, eta: 1 day, 10:05:03, time: 1.643, data_time: 0.218, memory: 49163, loss_cls_0: 1.2075, loss_box_0: 2.1197, loss_cns_0: 0.6235, loss_yns_0: 0.1704, loss_cls_1: 1.2398, loss_box_1: 3.0260, loss_cns_1: 0.5312, loss_yns_1: 0.1762, loss_cls_2: 1.2760, loss_box_2: 3.0101, loss_cns_2: 0.5366, loss_yns_2: 0.1754, loss_cls_3: 1.2512, loss_box_3: 3.0168, loss_cns_3: 0.5364, loss_yns_3: 0.1739, loss_cls_4: 1.2506, loss_box_4: 3.0013, loss_cns_4: 0.5436, loss_yns_4: 0.1734, loss_cls_5: 1.2852, loss_box_5: 3.0737, loss_cns_5: 0.5385, loss_yns_5: 0.1788, loss_cls_dn_0: 0.4803, loss_box_dn_0: 1.0027, loss_cls_dn_1: 0.4422, loss_box_dn_1: 1.2792, loss_cls_dn_2: 0.4385, loss_box_dn_2: 1.2910, loss_cls_dn_3: 0.4338, loss_box_dn_3: 1.2878, loss_cls_dn_4: 0.4260, loss_box_dn_4: 1.2966, loss_cls_dn_5: 0.4366, loss_box_dn_5: 1.3487, loss_dense_depth: 1.1828, loss: 40.4621, grad_norm: 60.0931 -2025-11-12 20:07:45,804 - mmdet - INFO - Iter [22/17500] lr: 1.084e-04, eta: 1 day, 8:52:28, time: 1.547, data_time: 0.080, memory: 49163, loss_cls_0: 1.2080, loss_box_0: 2.1187, loss_cns_0: 0.6192, loss_yns_0: 0.1692, loss_cls_1: 1.2684, loss_box_1: 2.8878, loss_cns_1: 0.5644, loss_yns_1: 0.1737, loss_cls_2: 1.2880, loss_box_2: 2.8171, loss_cns_2: 0.5561, loss_yns_2: 0.1762, loss_cls_3: 1.2727, loss_box_3: 2.8296, loss_cns_3: 0.5585, loss_yns_3: 0.1726, loss_cls_4: 1.2789, loss_box_4: 2.8720, loss_cns_4: 0.5395, loss_yns_4: 0.1734, loss_cls_5: 1.3176, loss_box_5: 2.9179, loss_cns_5: 0.5408, loss_yns_5: 0.1748, loss_cls_dn_0: 0.4851, loss_box_dn_0: 1.0012, loss_cls_dn_1: 0.4404, loss_box_dn_1: 1.1063, loss_cls_dn_2: 0.4365, loss_box_dn_2: 1.1044, loss_cls_dn_3: 0.4392, loss_box_dn_3: 1.1397, loss_cls_dn_4: 0.4245, loss_box_dn_4: 1.2017, loss_cls_dn_5: 0.4359, loss_box_dn_5: 1.2151, loss_dense_depth: 1.1993, loss: 39.1244, grad_norm: 52.6254 -2025-11-12 20:07:47,350 - mmdet - INFO - Iter [23/17500] lr: 1.088e-04, eta: 1 day, 7:46:11, time: 1.546, data_time: 0.078, memory: 49163, loss_cls_0: 1.1881, loss_box_0: 2.1182, loss_cns_0: 0.6213, loss_yns_0: 0.1683, loss_cls_1: 1.2498, loss_box_1: 2.9260, loss_cns_1: 0.5393, loss_yns_1: 0.1721, loss_cls_2: 1.2669, loss_box_2: 2.8618, loss_cns_2: 0.5486, loss_yns_2: 0.1759, loss_cls_3: 1.2477, loss_box_3: 2.9340, loss_cns_3: 0.5532, loss_yns_3: 0.1698, loss_cls_4: 1.2453, loss_box_4: 3.0126, loss_cns_4: 0.5386, loss_yns_4: 0.1776, loss_cls_5: 1.2673, loss_box_5: 3.0735, loss_cns_5: 0.5519, loss_yns_5: 0.1727, loss_cls_dn_0: 0.4849, loss_box_dn_0: 0.9816, loss_cls_dn_1: 0.4349, loss_box_dn_1: 1.1556, loss_cls_dn_2: 0.4350, loss_box_dn_2: 1.1458, loss_cls_dn_3: 0.4495, loss_box_dn_3: 1.2281, loss_cls_dn_4: 0.4322, loss_box_dn_4: 1.3430, loss_cls_dn_5: 0.4459, loss_box_dn_5: 1.3477, loss_dense_depth: 1.1172, loss: 39.7819, grad_norm: 79.1846 -2025-11-12 20:07:48,892 - mmdet - INFO - Iter [24/17500] lr: 1.092e-04, eta: 1 day, 6:45:23, time: 1.543, data_time: 0.092, memory: 49163, loss_cls_0: 1.2014, loss_box_0: 2.1203, loss_cns_0: 0.6277, loss_yns_0: 0.1684, loss_cls_1: 1.2480, loss_box_1: 3.0219, loss_cns_1: 0.5346, loss_yns_1: 0.1731, loss_cls_2: 1.2696, loss_box_2: 2.9877, loss_cns_2: 0.5486, loss_yns_2: 0.1718, loss_cls_3: 1.2453, loss_box_3: 3.0424, loss_cns_3: 0.5637, loss_yns_3: 0.1699, loss_cls_4: 1.2336, loss_box_4: 3.0716, loss_cns_4: 0.5581, loss_yns_4: 0.1718, loss_cls_5: 1.2540, loss_box_5: 3.1175, loss_cns_5: 0.5753, loss_yns_5: 0.1703, loss_cls_dn_0: 0.4909, loss_box_dn_0: 0.9891, loss_cls_dn_1: 0.4270, loss_box_dn_1: 1.3022, loss_cls_dn_2: 0.4290, loss_box_dn_2: 1.3156, loss_cls_dn_3: 0.4498, loss_box_dn_3: 1.4266, loss_cls_dn_4: 0.4345, loss_box_dn_4: 1.5347, loss_cls_dn_5: 0.4473, loss_box_dn_5: 1.5379, loss_dense_depth: 1.2491, loss: 41.2802, grad_norm: 80.5127 -2025-11-12 20:07:50,434 - mmdet - INFO - Iter [25/17500] lr: 1.096e-04, eta: 1 day, 5:49:25, time: 1.541, data_time: 0.079, memory: 49163, loss_cls_0: 1.2079, loss_box_0: 2.1216, loss_cns_0: 0.6335, loss_yns_0: 0.1696, loss_cls_1: 1.2499, loss_box_1: 2.9552, loss_cns_1: 0.5343, loss_yns_1: 0.1743, loss_cls_2: 1.2751, loss_box_2: 2.9484, loss_cns_2: 0.5485, loss_yns_2: 0.1747, loss_cls_3: 1.2388, loss_box_3: 2.9775, loss_cns_3: 0.5520, loss_yns_3: 0.1711, loss_cls_4: 1.2283, loss_box_4: 3.0214, loss_cns_4: 0.5377, loss_yns_4: 0.1710, loss_cls_5: 1.2497, loss_box_5: 3.0619, loss_cns_5: 0.5298, loss_yns_5: 0.1730, loss_cls_dn_0: 0.4809, loss_box_dn_0: 0.9803, loss_cls_dn_1: 0.4201, loss_box_dn_1: 1.3778, loss_cls_dn_2: 0.4226, loss_box_dn_2: 1.4089, loss_cls_dn_3: 0.4398, loss_box_dn_3: 1.5178, loss_cls_dn_4: 0.4278, loss_box_dn_4: 1.6059, loss_cls_dn_5: 0.4444, loss_box_dn_5: 1.5993, loss_dense_depth: 1.1931, loss: 41.2243, grad_norm: 76.3189 -2025-11-12 20:07:51,970 - mmdet - INFO - Iter [26/17500] lr: 1.100e-04, eta: 1 day, 4:57:43, time: 1.538, data_time: 0.078, memory: 49163, loss_cls_0: 1.2057, loss_box_0: 2.1302, loss_cns_0: 0.6356, loss_yns_0: 0.1692, loss_cls_1: 1.2517, loss_box_1: 2.8271, loss_cns_1: 0.5525, loss_yns_1: 0.1715, loss_cls_2: 1.2887, loss_box_2: 2.8186, loss_cns_2: 0.5641, loss_yns_2: 0.1755, loss_cls_3: 1.2490, loss_box_3: 2.8633, loss_cns_3: 0.5658, loss_yns_3: 0.1730, loss_cls_4: 1.2415, loss_box_4: 2.8859, loss_cns_4: 0.5622, loss_yns_4: 0.1751, loss_cls_5: 1.2526, loss_box_5: 2.8634, loss_cns_5: 0.5461, loss_yns_5: 0.1719, loss_cls_dn_0: 0.4673, loss_box_dn_0: 0.9921, loss_cls_dn_1: 0.4414, loss_box_dn_1: 1.2015, loss_cls_dn_2: 0.4414, loss_box_dn_2: 1.2365, loss_cls_dn_3: 0.4497, loss_box_dn_3: 1.3520, loss_cls_dn_4: 0.4391, loss_box_dn_4: 1.4205, loss_cls_dn_5: 0.4703, loss_box_dn_5: 1.3906, loss_dense_depth: 1.1131, loss: 39.7553, grad_norm: 74.6831 -2025-11-12 20:07:53,493 - mmdet - INFO - Iter [27/17500] lr: 1.104e-04, eta: 1 day, 4:09:41, time: 1.522, data_time: 0.079, memory: 49163, loss_cls_0: 1.1726, loss_box_0: 2.1653, loss_cns_0: 0.6303, loss_yns_0: 0.1688, loss_cls_1: 1.2378, loss_box_1: 2.7664, loss_cns_1: 0.5607, loss_yns_1: 0.1722, loss_cls_2: 1.2863, loss_box_2: 2.7984, loss_cns_2: 0.5696, loss_yns_2: 0.1779, loss_cls_3: 1.2465, loss_box_3: 2.8064, loss_cns_3: 0.5761, loss_yns_3: 0.1709, loss_cls_4: 1.2465, loss_box_4: 2.7747, loss_cns_4: 0.5977, loss_yns_4: 0.1729, loss_cls_5: 1.2385, loss_box_5: 2.8050, loss_cns_5: 0.5692, loss_yns_5: 0.1688, loss_cls_dn_0: 0.4749, loss_box_dn_0: 0.9970, loss_cls_dn_1: 0.4279, loss_box_dn_1: 1.2012, loss_cls_dn_2: 0.4293, loss_box_dn_2: 1.1914, loss_cls_dn_3: 0.4342, loss_box_dn_3: 1.2476, loss_cls_dn_4: 0.4231, loss_box_dn_4: 1.2579, loss_cls_dn_5: 0.4610, loss_box_dn_5: 1.2592, loss_dense_depth: 1.1425, loss: 39.0269, grad_norm: 66.4882 -2025-11-12 20:07:55,008 - mmdet - INFO - Iter [28/17500] lr: 1.108e-04, eta: 1 day, 3:24:59, time: 1.515, data_time: 0.078, memory: 49163, loss_cls_0: 1.1527, loss_box_0: 2.1655, loss_cns_0: 0.6235, loss_yns_0: 0.1682, loss_cls_1: 1.2185, loss_box_1: 2.6154, loss_cns_1: 0.5707, loss_yns_1: 0.1715, loss_cls_2: 1.2551, loss_box_2: 2.7533, loss_cns_2: 0.5627, loss_yns_2: 0.1757, loss_cls_3: 1.2392, loss_box_3: 2.7324, loss_cns_3: 0.5590, loss_yns_3: 0.1713, loss_cls_4: 1.2507, loss_box_4: 2.6985, loss_cns_4: 0.5679, loss_yns_4: 0.1721, loss_cls_5: 1.2318, loss_box_5: 2.7579, loss_cns_5: 0.5680, loss_yns_5: 0.1695, loss_cls_dn_0: 0.4758, loss_box_dn_0: 0.9865, loss_cls_dn_1: 0.4131, loss_box_dn_1: 1.2271, loss_cls_dn_2: 0.4184, loss_box_dn_2: 1.2187, loss_cls_dn_3: 0.4137, loss_box_dn_3: 1.2322, loss_cls_dn_4: 0.4013, loss_box_dn_4: 1.2240, loss_cls_dn_5: 0.4333, loss_box_dn_5: 1.2633, loss_dense_depth: 1.0908, loss: 38.3494, grad_norm: 84.2650 -2025-11-12 20:07:56,545 - mmdet - INFO - Iter [29/17500] lr: 1.112e-04, eta: 1 day, 2:43:37, time: 1.537, data_time: 0.077, memory: 49163, loss_cls_0: 1.1311, loss_box_0: 2.1338, loss_cns_0: 0.6203, loss_yns_0: 0.1683, loss_cls_1: 1.2132, loss_box_1: 2.5998, loss_cns_1: 0.5676, loss_yns_1: 0.1712, loss_cls_2: 1.2295, loss_box_2: 2.6814, loss_cns_2: 0.5674, loss_yns_2: 0.1789, loss_cls_3: 1.2257, loss_box_3: 2.6437, loss_cns_3: 0.5708, loss_yns_3: 0.1726, loss_cls_4: 1.2498, loss_box_4: 2.6078, loss_cns_4: 0.5770, loss_yns_4: 0.1807, loss_cls_5: 1.2497, loss_box_5: 2.6332, loss_cns_5: 0.5768, loss_yns_5: 0.1739, loss_cls_dn_0: 0.4580, loss_box_dn_0: 0.9798, loss_cls_dn_1: 0.4112, loss_box_dn_1: 1.1437, loss_cls_dn_2: 0.4254, loss_box_dn_2: 1.1375, loss_cls_dn_3: 0.4122, loss_box_dn_3: 1.1399, loss_cls_dn_4: 0.3959, loss_box_dn_4: 1.1461, loss_cls_dn_5: 0.4200, loss_box_dn_5: 1.2153, loss_dense_depth: 1.0854, loss: 37.4948, grad_norm: 73.0367 -2025-11-12 20:07:58,066 - mmdet - INFO - Iter [30/17500] lr: 1.116e-04, eta: 1 day, 2:04:50, time: 1.521, data_time: 0.079, memory: 49163, loss_cls_0: 1.1260, loss_box_0: 2.0741, loss_cns_0: 0.6209, loss_yns_0: 0.1689, loss_cls_1: 1.1836, loss_box_1: 2.5584, loss_cns_1: 0.5723, loss_yns_1: 0.1708, loss_cls_2: 1.1846, loss_box_2: 2.5519, loss_cns_2: 0.5804, loss_yns_2: 0.1809, loss_cls_3: 1.2047, loss_box_3: 2.5254, loss_cns_3: 0.5818, loss_yns_3: 0.1748, loss_cls_4: 1.2197, loss_box_4: 2.5436, loss_cns_4: 0.5857, loss_yns_4: 0.1766, loss_cls_5: 1.2520, loss_box_5: 2.5915, loss_cns_5: 0.5730, loss_yns_5: 0.1785, loss_cls_dn_0: 0.4329, loss_box_dn_0: 0.9773, loss_cls_dn_1: 0.3985, loss_box_dn_1: 1.1639, loss_cls_dn_2: 0.4241, loss_box_dn_2: 1.1455, loss_cls_dn_3: 0.3991, loss_box_dn_3: 1.1626, loss_cls_dn_4: 0.3866, loss_box_dn_4: 1.2031, loss_cls_dn_5: 0.3961, loss_box_dn_5: 1.3061, loss_dense_depth: 1.0910, loss: 37.0671, grad_norm: 50.2350 -2025-11-12 20:07:59,601 - mmdet - INFO - Iter [31/17500] lr: 1.120e-04, eta: 1 day, 1:28:41, time: 1.535, data_time: 0.076, memory: 49163, loss_cls_0: 1.1219, loss_box_0: 2.0568, loss_cns_0: 0.6208, loss_yns_0: 0.1714, loss_cls_1: 1.1707, loss_box_1: 2.5591, loss_cns_1: 0.5718, loss_yns_1: 0.1703, loss_cls_2: 1.1784, loss_box_2: 2.5514, loss_cns_2: 0.5737, loss_yns_2: 0.1781, loss_cls_3: 1.1916, loss_box_3: 2.5766, loss_cns_3: 0.5682, loss_yns_3: 0.1723, loss_cls_4: 1.1893, loss_box_4: 2.6386, loss_cns_4: 0.5537, loss_yns_4: 0.1718, loss_cls_5: 1.2150, loss_box_5: 2.7141, loss_cns_5: 0.5498, loss_yns_5: 0.1717, loss_cls_dn_0: 0.4264, loss_box_dn_0: 0.9666, loss_cls_dn_1: 0.3941, loss_box_dn_1: 1.2301, loss_cls_dn_2: 0.4224, loss_box_dn_2: 1.2305, loss_cls_dn_3: 0.3895, loss_box_dn_3: 1.2718, loss_cls_dn_4: 0.3893, loss_box_dn_4: 1.3230, loss_cls_dn_5: 0.3974, loss_box_dn_5: 1.4541, loss_dense_depth: 1.0558, loss: 37.5878, grad_norm: 76.9091 -2025-11-12 20:08:01,117 - mmdet - INFO - Iter [32/17500] lr: 1.124e-04, eta: 1 day, 0:54:36, time: 1.515, data_time: 0.078, memory: 49163, loss_cls_0: 1.1031, loss_box_0: 2.0548, loss_cns_0: 0.6215, loss_yns_0: 0.1706, loss_cls_1: 1.1538, loss_box_1: 2.7122, loss_cns_1: 0.5662, loss_yns_1: 0.1723, loss_cls_2: 1.1650, loss_box_2: 2.6925, loss_cns_2: 0.5684, loss_yns_2: 0.1778, loss_cls_3: 1.1846, loss_box_3: 2.7244, loss_cns_3: 0.5674, loss_yns_3: 0.1725, loss_cls_4: 1.1609, loss_box_4: 2.7824, loss_cns_4: 0.5597, loss_yns_4: 0.1713, loss_cls_5: 1.1852, loss_box_5: 2.8242, loss_cns_5: 0.5569, loss_yns_5: 0.1715, loss_cls_dn_0: 0.4153, loss_box_dn_0: 0.9700, loss_cls_dn_1: 0.3851, loss_box_dn_1: 1.2555, loss_cls_dn_2: 0.4124, loss_box_dn_2: 1.2630, loss_cls_dn_3: 0.3789, loss_box_dn_3: 1.2977, loss_cls_dn_4: 0.3875, loss_box_dn_4: 1.3443, loss_cls_dn_5: 0.4018, loss_box_dn_5: 1.4577, loss_dense_depth: 1.0137, loss: 38.2022, grad_norm: 74.1020 -2025-11-12 20:08:02,630 - mmdet - INFO - Iter [33/17500] lr: 1.128e-04, eta: 1 day, 0:22:36, time: 1.516, data_time: 0.077, memory: 49163, loss_cls_0: 1.0945, loss_box_0: 2.0611, loss_cns_0: 0.6241, loss_yns_0: 0.1709, loss_cls_1: 1.1565, loss_box_1: 2.7349, loss_cns_1: 0.5672, loss_yns_1: 0.1741, loss_cls_2: 1.1648, loss_box_2: 2.6772, loss_cns_2: 0.5793, loss_yns_2: 0.1774, loss_cls_3: 1.1921, loss_box_3: 2.7046, loss_cns_3: 0.5820, loss_yns_3: 0.1715, loss_cls_4: 1.1594, loss_box_4: 2.7570, loss_cns_4: 0.5801, loss_yns_4: 0.1702, loss_cls_5: 1.1802, loss_box_5: 2.7618, loss_cns_5: 0.5755, loss_yns_5: 0.1745, loss_cls_dn_0: 0.4311, loss_box_dn_0: 0.9889, loss_cls_dn_1: 0.3614, loss_box_dn_1: 1.3875, loss_cls_dn_2: 0.3868, loss_box_dn_2: 1.3515, loss_cls_dn_3: 0.3584, loss_box_dn_3: 1.3571, loss_cls_dn_4: 0.3684, loss_box_dn_4: 1.3749, loss_cls_dn_5: 0.3842, loss_box_dn_5: 1.4382, loss_dense_depth: 1.0395, loss: 38.4188, grad_norm: 56.0135 -2025-11-12 20:08:04,146 - mmdet - INFO - Iter [34/17500] lr: 1.132e-04, eta: 23:52:28, time: 1.515, data_time: 0.078, memory: 49163, loss_cls_0: 1.0850, loss_box_0: 2.0518, loss_cns_0: 0.6246, loss_yns_0: 0.1706, loss_cls_1: 1.1434, loss_box_1: 2.6773, loss_cns_1: 0.5675, loss_yns_1: 0.1741, loss_cls_2: 1.1570, loss_box_2: 2.6908, loss_cns_2: 0.5735, loss_yns_2: 0.1744, loss_cls_3: 1.1707, loss_box_3: 2.7071, loss_cns_3: 0.5720, loss_yns_3: 0.1712, loss_cls_4: 1.1484, loss_box_4: 2.7378, loss_cns_4: 0.5673, loss_yns_4: 0.1709, loss_cls_5: 1.1757, loss_box_5: 2.7483, loss_cns_5: 0.5563, loss_yns_5: 0.1778, loss_cls_dn_0: 0.4429, loss_box_dn_0: 0.9972, loss_cls_dn_1: 0.3499, loss_box_dn_1: 1.3728, loss_cls_dn_2: 0.3702, loss_box_dn_2: 1.3260, loss_cls_dn_3: 0.3635, loss_box_dn_3: 1.3069, loss_cls_dn_4: 0.3605, loss_box_dn_4: 1.2898, loss_cls_dn_5: 0.3809, loss_box_dn_5: 1.3201, loss_dense_depth: 1.0068, loss: 37.8810, grad_norm: 60.0180 -2025-11-12 20:08:05,712 - mmdet - INFO - Iter [35/17500] lr: 1.136e-04, eta: 23:24:29, time: 1.566, data_time: 0.078, memory: 49163, loss_cls_0: 1.0794, loss_box_0: 2.0076, loss_cns_0: 0.6284, loss_yns_0: 0.1714, loss_cls_1: 1.1507, loss_box_1: 2.5446, loss_cns_1: 0.5726, loss_yns_1: 0.1780, loss_cls_2: 1.1643, loss_box_2: 2.5885, loss_cns_2: 0.5802, loss_yns_2: 0.1758, loss_cls_3: 1.1730, loss_box_3: 2.5520, loss_cns_3: 0.5759, loss_yns_3: 0.1715, loss_cls_4: 1.1754, loss_box_4: 2.5649, loss_cns_4: 0.5759, loss_yns_4: 0.1728, loss_cls_5: 1.1811, loss_box_5: 2.6029, loss_cns_5: 0.5712, loss_yns_5: 0.1750, loss_cls_dn_0: 0.4420, loss_box_dn_0: 0.9875, loss_cls_dn_1: 0.3476, loss_box_dn_1: 1.2575, loss_cls_dn_2: 0.3641, loss_box_dn_2: 1.2491, loss_cls_dn_3: 0.3781, loss_box_dn_3: 1.2368, loss_cls_dn_4: 0.3639, loss_box_dn_4: 1.2352, loss_cls_dn_5: 0.3807, loss_box_dn_5: 1.2538, loss_dense_depth: 1.0183, loss: 36.8480, grad_norm: 58.5398 -2025-11-12 20:08:07,234 - mmdet - INFO - Iter [36/17500] lr: 1.140e-04, eta: 22:57:42, time: 1.522, data_time: 0.077, memory: 49163, loss_cls_0: 1.0763, loss_box_0: 1.9905, loss_cns_0: 0.6223, loss_yns_0: 0.1711, loss_cls_1: 1.1606, loss_box_1: 2.4930, loss_cns_1: 0.5765, loss_yns_1: 0.1789, loss_cls_2: 1.1689, loss_box_2: 2.5190, loss_cns_2: 0.5836, loss_yns_2: 0.1801, loss_cls_3: 1.1585, loss_box_3: 2.5148, loss_cns_3: 0.5788, loss_yns_3: 0.1707, loss_cls_4: 1.1810, loss_box_4: 2.5853, loss_cns_4: 0.5792, loss_yns_4: 0.1716, loss_cls_5: 1.1759, loss_box_5: 2.5722, loss_cns_5: 0.5810, loss_yns_5: 0.1731, loss_cls_dn_0: 0.4294, loss_box_dn_0: 0.9847, loss_cls_dn_1: 0.3695, loss_box_dn_1: 1.0910, loss_cls_dn_2: 0.3806, loss_box_dn_2: 1.1322, loss_cls_dn_3: 0.3961, loss_box_dn_3: 1.1865, loss_cls_dn_4: 0.3735, loss_box_dn_4: 1.2733, loss_cls_dn_5: 0.3971, loss_box_dn_5: 1.2650, loss_dense_depth: 0.9781, loss: 36.4198, grad_norm: 73.8795 -2025-11-12 20:08:08,751 - mmdet - INFO - Iter [37/17500] lr: 1.144e-04, eta: 22:32:19, time: 1.516, data_time: 0.074, memory: 49163, loss_cls_0: 1.0905, loss_box_0: 1.9941, loss_cns_0: 0.6196, loss_yns_0: 0.1713, loss_cls_1: 1.1532, loss_box_1: 2.5840, loss_cns_1: 0.5737, loss_yns_1: 0.1753, loss_cls_2: 1.1748, loss_box_2: 2.5896, loss_cns_2: 0.5766, loss_yns_2: 0.1768, loss_cls_3: 1.1717, loss_box_3: 2.6092, loss_cns_3: 0.5704, loss_yns_3: 0.1688, loss_cls_4: 1.1933, loss_box_4: 2.6817, loss_cns_4: 0.5700, loss_yns_4: 0.1706, loss_cls_5: 1.1886, loss_box_5: 2.6293, loss_cns_5: 0.5726, loss_yns_5: 0.1799, loss_cls_dn_0: 0.4232, loss_box_dn_0: 0.9904, loss_cls_dn_1: 0.3527, loss_box_dn_1: 1.2022, loss_cls_dn_2: 0.3668, loss_box_dn_2: 1.2401, loss_cls_dn_3: 0.3653, loss_box_dn_3: 1.3190, loss_cls_dn_4: 0.3570, loss_box_dn_4: 1.4194, loss_cls_dn_5: 0.3730, loss_box_dn_5: 1.3986, loss_dense_depth: 0.9931, loss: 37.3865, grad_norm: 76.6383 -2025-11-12 20:08:10,284 - mmdet - INFO - Iter [38/17500] lr: 1.148e-04, eta: 22:08:23, time: 1.532, data_time: 0.076, memory: 49163, loss_cls_0: 1.0845, loss_box_0: 1.9953, loss_cns_0: 0.6174, loss_yns_0: 0.1686, loss_cls_1: 1.1383, loss_box_1: 2.6087, loss_cns_1: 0.5736, loss_yns_1: 0.1710, loss_cls_2: 1.1670, loss_box_2: 2.5810, loss_cns_2: 0.5803, loss_yns_2: 0.1734, loss_cls_3: 1.2061, loss_box_3: 2.5730, loss_cns_3: 0.5783, loss_yns_3: 0.1699, loss_cls_4: 1.1764, loss_box_4: 2.6104, loss_cns_4: 0.5811, loss_yns_4: 0.1741, loss_cls_5: 1.1931, loss_box_5: 2.6210, loss_cns_5: 0.5890, loss_yns_5: 0.1808, loss_cls_dn_0: 0.4216, loss_box_dn_0: 0.9952, loss_cls_dn_1: 0.3470, loss_box_dn_1: 1.2100, loss_cls_dn_2: 0.3664, loss_box_dn_2: 1.2117, loss_cls_dn_3: 0.3539, loss_box_dn_3: 1.2833, loss_cls_dn_4: 0.3631, loss_box_dn_4: 1.3754, loss_cls_dn_5: 0.3701, loss_box_dn_5: 1.3714, loss_dense_depth: 0.9793, loss: 37.1604, grad_norm: 62.1415 -2025-11-12 20:08:11,803 - mmdet - INFO - Iter [39/17500] lr: 1.152e-04, eta: 21:45:36, time: 1.520, data_time: 0.076, memory: 49163, loss_cls_0: 1.0862, loss_box_0: 1.9777, loss_cns_0: 0.6171, loss_yns_0: 0.1674, loss_cls_1: 1.1502, loss_box_1: 2.4725, loss_cns_1: 0.5892, loss_yns_1: 0.1682, loss_cls_2: 1.1772, loss_box_2: 2.4697, loss_cns_2: 0.5934, loss_yns_2: 0.1747, loss_cls_3: 1.1679, loss_box_3: 2.4885, loss_cns_3: 0.5885, loss_yns_3: 0.1697, loss_cls_4: 1.1588, loss_box_4: 2.5263, loss_cns_4: 0.5880, loss_yns_4: 0.1712, loss_cls_5: 1.1862, loss_box_5: 2.6369, loss_cns_5: 0.5774, loss_yns_5: 0.1692, loss_cls_dn_0: 0.4258, loss_box_dn_0: 0.9806, loss_cls_dn_1: 0.3504, loss_box_dn_1: 1.1764, loss_cls_dn_2: 0.3770, loss_box_dn_2: 1.1620, loss_cls_dn_3: 0.3744, loss_box_dn_3: 1.2317, loss_cls_dn_4: 0.3936, loss_box_dn_4: 1.3002, loss_cls_dn_5: 0.3793, loss_box_dn_5: 1.3273, loss_dense_depth: 0.9711, loss: 36.5219, grad_norm: 71.3989 -2025-11-12 20:08:13,331 - mmdet - INFO - Iter [40/17500] lr: 1.156e-04, eta: 21:24:00, time: 1.528, data_time: 0.076, memory: 49163, loss_cls_0: 1.0765, loss_box_0: 1.9610, loss_cns_0: 0.6214, loss_yns_0: 0.1690, loss_cls_1: 1.1292, loss_box_1: 2.3854, loss_cns_1: 0.5944, loss_yns_1: 0.1736, loss_cls_2: 1.1689, loss_box_2: 2.3601, loss_cns_2: 0.5988, loss_yns_2: 0.1767, loss_cls_3: 1.1525, loss_box_3: 2.4172, loss_cns_3: 0.5944, loss_yns_3: 0.1703, loss_cls_4: 1.1492, loss_box_4: 2.4632, loss_cns_4: 0.5897, loss_yns_4: 0.1693, loss_cls_5: 1.1620, loss_box_5: 2.5270, loss_cns_5: 0.5822, loss_yns_5: 0.1703, loss_cls_dn_0: 0.4303, loss_box_dn_0: 0.9719, loss_cls_dn_1: 0.3571, loss_box_dn_1: 1.1968, loss_cls_dn_2: 0.3855, loss_box_dn_2: 1.1648, loss_cls_dn_3: 0.3991, loss_box_dn_3: 1.2300, loss_cls_dn_4: 0.4149, loss_box_dn_4: 1.2664, loss_cls_dn_5: 0.3871, loss_box_dn_5: 1.2732, loss_dense_depth: 1.0194, loss: 36.0585, grad_norm: 68.3254 -2025-11-12 20:08:14,940 - mmdet - INFO - Iter [41/17500] lr: 1.160e-04, eta: 21:04:02, time: 1.610, data_time: 0.157, memory: 49163, loss_cls_0: 1.0401, loss_box_0: 1.9935, loss_cns_0: 0.6203, loss_yns_0: 0.1684, loss_cls_1: 1.1087, loss_box_1: 2.3757, loss_cns_1: 0.5892, loss_yns_1: 0.1718, loss_cls_2: 1.1510, loss_box_2: 2.2904, loss_cns_2: 0.6083, loss_yns_2: 0.1750, loss_cls_3: 1.1400, loss_box_3: 2.3438, loss_cns_3: 0.6109, loss_yns_3: 0.1672, loss_cls_4: 1.1286, loss_box_4: 2.3334, loss_cns_4: 0.6064, loss_yns_4: 0.1695, loss_cls_5: 1.1451, loss_box_5: 2.3281, loss_cns_5: 0.6136, loss_yns_5: 0.1707, loss_cls_dn_0: 0.4313, loss_box_dn_0: 0.9681, loss_cls_dn_1: 0.3445, loss_box_dn_1: 1.2272, loss_cls_dn_2: 0.3696, loss_box_dn_2: 1.1631, loss_cls_dn_3: 0.3853, loss_box_dn_3: 1.2011, loss_cls_dn_4: 0.3931, loss_box_dn_4: 1.1961, loss_cls_dn_5: 0.3746, loss_box_dn_5: 1.1773, loss_dense_depth: 0.9758, loss: 35.2564, grad_norm: 57.7562 -2025-11-12 20:08:16,480 - mmdet - INFO - Iter [42/17500] lr: 1.164e-04, eta: 20:44:32, time: 1.540, data_time: 0.074, memory: 49163, loss_cls_0: 1.0325, loss_box_0: 1.9951, loss_cns_0: 0.6193, loss_yns_0: 0.1675, loss_cls_1: 1.1107, loss_box_1: 2.3868, loss_cns_1: 0.5893, loss_yns_1: 0.1723, loss_cls_2: 1.1451, loss_box_2: 2.3134, loss_cns_2: 0.6085, loss_yns_2: 0.1703, loss_cls_3: 1.1347, loss_box_3: 2.3455, loss_cns_3: 0.6089, loss_yns_3: 0.1686, loss_cls_4: 1.1205, loss_box_4: 2.3241, loss_cns_4: 0.6041, loss_yns_4: 0.1706, loss_cls_5: 1.1442, loss_box_5: 2.3779, loss_cns_5: 0.6022, loss_yns_5: 0.1715, loss_cls_dn_0: 0.4206, loss_box_dn_0: 0.9626, loss_cls_dn_1: 0.3468, loss_box_dn_1: 1.1099, loss_cls_dn_2: 0.3590, loss_box_dn_2: 1.0605, loss_cls_dn_3: 0.3788, loss_box_dn_3: 1.0732, loss_cls_dn_4: 0.3745, loss_box_dn_4: 1.0548, loss_cls_dn_5: 0.3821, loss_box_dn_5: 1.0831, loss_dense_depth: 0.9570, loss: 34.6462, grad_norm: 67.6462 -2025-11-12 20:08:18,041 - mmdet - INFO - Iter [43/17500] lr: 1.168e-04, eta: 20:26:05, time: 1.560, data_time: 0.076, memory: 49163, loss_cls_0: 1.0712, loss_box_0: 1.9982, loss_cns_0: 0.6212, loss_yns_0: 0.1646, loss_cls_1: 1.1418, loss_box_1: 2.4385, loss_cns_1: 0.5908, loss_yns_1: 0.1668, loss_cls_2: 1.1880, loss_box_2: 2.3790, loss_cns_2: 0.6075, loss_yns_2: 0.1706, loss_cls_3: 1.1661, loss_box_3: 2.3703, loss_cns_3: 0.6059, loss_yns_3: 0.1667, loss_cls_4: 1.1661, loss_box_4: 2.4015, loss_cns_4: 0.5990, loss_yns_4: 0.1705, loss_cls_5: 1.1640, loss_box_5: 2.4723, loss_cns_5: 0.5897, loss_yns_5: 0.1719, loss_cls_dn_0: 0.4306, loss_box_dn_0: 0.9486, loss_cls_dn_1: 0.3552, loss_box_dn_1: 1.0275, loss_cls_dn_2: 0.3588, loss_box_dn_2: 1.0127, loss_cls_dn_3: 0.3756, loss_box_dn_3: 1.0229, loss_cls_dn_4: 0.3674, loss_box_dn_4: 1.0437, loss_cls_dn_5: 0.3990, loss_box_dn_5: 1.1334, loss_dense_depth: 0.9869, loss: 35.0446, grad_norm: 51.0860 -2025-11-12 20:08:19,611 - mmdet - INFO - Iter [44/17500] lr: 1.172e-04, eta: 20:08:32, time: 1.571, data_time: 0.097, memory: 49163, loss_cls_0: 1.0646, loss_box_0: 2.0591, loss_cns_0: 0.6142, loss_yns_0: 0.1620, loss_cls_1: 1.1198, loss_box_1: 2.3757, loss_cns_1: 0.5973, loss_yns_1: 0.1652, loss_cls_2: 1.1755, loss_box_2: 2.3963, loss_cns_2: 0.6031, loss_yns_2: 0.1680, loss_cls_3: 1.1719, loss_box_3: 2.3827, loss_cns_3: 0.6041, loss_yns_3: 0.1650, loss_cls_4: 1.1856, loss_box_4: 2.4258, loss_cns_4: 0.6029, loss_yns_4: 0.1670, loss_cls_5: 1.1547, loss_box_5: 2.4095, loss_cns_5: 0.6001, loss_yns_5: 0.1659, loss_cls_dn_0: 0.4404, loss_box_dn_0: 0.9465, loss_cls_dn_1: 0.3606, loss_box_dn_1: 1.0625, loss_cls_dn_2: 0.3630, loss_box_dn_2: 1.0933, loss_cls_dn_3: 0.3661, loss_box_dn_3: 1.1250, loss_cls_dn_4: 0.3592, loss_box_dn_4: 1.1775, loss_cls_dn_5: 0.4015, loss_box_dn_5: 1.2586, loss_dense_depth: 1.0245, loss: 35.5147, grad_norm: 69.2449 -2025-11-12 20:08:21,164 - mmdet - INFO - Iter [45/17500] lr: 1.176e-04, eta: 19:51:38, time: 1.552, data_time: 0.074, memory: 49163, loss_cls_0: 1.0698, loss_box_0: 2.0831, loss_cns_0: 0.6069, loss_yns_0: 0.1640, loss_cls_1: 1.1311, loss_box_1: 2.3636, loss_cns_1: 0.5923, loss_yns_1: 0.1688, loss_cls_2: 1.1815, loss_box_2: 2.4581, loss_cns_2: 0.5833, loss_yns_2: 0.1740, loss_cls_3: 1.1933, loss_box_3: 2.4835, loss_cns_3: 0.5763, loss_yns_3: 0.1716, loss_cls_4: 1.2170, loss_box_4: 2.5128, loss_cns_4: 0.5740, loss_yns_4: 0.1679, loss_cls_5: 1.1689, loss_box_5: 2.4888, loss_cns_5: 0.5857, loss_yns_5: 0.1689, loss_cls_dn_0: 0.4370, loss_box_dn_0: 0.9523, loss_cls_dn_1: 0.3614, loss_box_dn_1: 1.1425, loss_cls_dn_2: 0.3672, loss_box_dn_2: 1.2215, loss_cls_dn_3: 0.3667, loss_box_dn_3: 1.2694, loss_cls_dn_4: 0.3583, loss_box_dn_4: 1.3317, loss_cls_dn_5: 0.4041, loss_box_dn_5: 1.4120, loss_dense_depth: 1.0157, loss: 36.5251, grad_norm: 80.6518 -2025-11-12 20:08:22,727 - mmdet - INFO - Iter [46/17500] lr: 1.180e-04, eta: 19:35:33, time: 1.563, data_time: 0.075, memory: 49163, loss_cls_0: 1.0664, loss_box_0: 2.0914, loss_cns_0: 0.6095, loss_yns_0: 0.1618, loss_cls_1: 1.1103, loss_box_1: 2.4779, loss_cns_1: 0.5893, loss_yns_1: 0.1679, loss_cls_2: 1.1553, loss_box_2: 2.5233, loss_cns_2: 0.5810, loss_yns_2: 0.1687, loss_cls_3: 1.1641, loss_box_3: 2.5188, loss_cns_3: 0.5836, loss_yns_3: 0.1694, loss_cls_4: 1.1790, loss_box_4: 2.5241, loss_cns_4: 0.5835, loss_yns_4: 0.1690, loss_cls_5: 1.1506, loss_box_5: 2.5031, loss_cns_5: 0.5915, loss_yns_5: 0.1692, loss_cls_dn_0: 0.4198, loss_box_dn_0: 0.9395, loss_cls_dn_1: 0.3488, loss_box_dn_1: 1.2019, loss_cls_dn_2: 0.3643, loss_box_dn_2: 1.2572, loss_cls_dn_3: 0.3585, loss_box_dn_3: 1.2807, loss_cls_dn_4: 0.3518, loss_box_dn_4: 1.3101, loss_cls_dn_5: 0.3894, loss_box_dn_5: 1.3721, loss_dense_depth: 0.9680, loss: 36.5709, grad_norm: 72.1898 -2025-11-12 20:08:24,251 - mmdet - INFO - Iter [47/17500] lr: 1.184e-04, eta: 19:19:55, time: 1.526, data_time: 0.073, memory: 49163, loss_cls_0: 1.0675, loss_box_0: 2.0608, loss_cns_0: 0.6170, loss_yns_0: 0.1639, loss_cls_1: 1.1308, loss_box_1: 2.4497, loss_cns_1: 0.6028, loss_yns_1: 0.1673, loss_cls_2: 1.1459, loss_box_2: 2.4364, loss_cns_2: 0.6064, loss_yns_2: 0.1661, loss_cls_3: 1.1556, loss_box_3: 2.4080, loss_cns_3: 0.6089, loss_yns_3: 0.1662, loss_cls_4: 1.1586, loss_box_4: 2.4074, loss_cns_4: 0.6104, loss_yns_4: 0.1691, loss_cls_5: 1.1624, loss_box_5: 2.4199, loss_cns_5: 0.6108, loss_yns_5: 0.1688, loss_cls_dn_0: 0.4183, loss_box_dn_0: 0.9285, loss_cls_dn_1: 0.3427, loss_box_dn_1: 1.1604, loss_cls_dn_2: 0.3752, loss_box_dn_2: 1.1850, loss_cls_dn_3: 0.3642, loss_box_dn_3: 1.1764, loss_cls_dn_4: 0.3614, loss_box_dn_4: 1.1852, loss_cls_dn_5: 0.3848, loss_box_dn_5: 1.2377, loss_dense_depth: 0.9708, loss: 35.7514, grad_norm: 45.1827 -2025-11-12 20:08:25,790 - mmdet - INFO - Iter [48/17500] lr: 1.188e-04, eta: 19:05:01, time: 1.539, data_time: 0.072, memory: 49163, loss_cls_0: 1.0621, loss_box_0: 2.0368, loss_cns_0: 0.6131, loss_yns_0: 0.1627, loss_cls_1: 1.1330, loss_box_1: 2.4920, loss_cns_1: 0.5928, loss_yns_1: 0.1660, loss_cls_2: 1.1413, loss_box_2: 2.4750, loss_cns_2: 0.5982, loss_yns_2: 0.1681, loss_cls_3: 1.1449, loss_box_3: 2.4952, loss_cns_3: 0.5948, loss_yns_3: 0.1670, loss_cls_4: 1.1425, loss_box_4: 2.5350, loss_cns_4: 0.5859, loss_yns_4: 0.1680, loss_cls_5: 1.1643, loss_box_5: 2.5572, loss_cns_5: 0.5906, loss_yns_5: 0.1659, loss_cls_dn_0: 0.4166, loss_box_dn_0: 0.9139, loss_cls_dn_1: 0.3288, loss_box_dn_1: 1.1940, loss_cls_dn_2: 0.3789, loss_box_dn_2: 1.1777, loss_cls_dn_3: 0.3610, loss_box_dn_3: 1.1673, loss_cls_dn_4: 0.3591, loss_box_dn_4: 1.1761, loss_cls_dn_5: 0.3705, loss_box_dn_5: 1.2031, loss_dense_depth: 1.0221, loss: 36.0214, grad_norm: 56.8425 -2025-11-12 20:08:27,328 - mmdet - INFO - Iter [49/17500] lr: 1.192e-04, eta: 18:50:42, time: 1.536, data_time: 0.073, memory: 49163, loss_cls_0: 1.0593, loss_box_0: 2.0572, loss_cns_0: 0.6122, loss_yns_0: 0.1636, loss_cls_1: 1.1091, loss_box_1: 2.5129, loss_cns_1: 0.5885, loss_yns_1: 0.1659, loss_cls_2: 1.1248, loss_box_2: 2.4705, loss_cns_2: 0.6006, loss_yns_2: 0.1723, loss_cls_3: 1.1358, loss_box_3: 2.5061, loss_cns_3: 0.5939, loss_yns_3: 0.1668, loss_cls_4: 1.1267, loss_box_4: 2.5411, loss_cns_4: 0.5940, loss_yns_4: 0.1690, loss_cls_5: 1.1482, loss_box_5: 2.4969, loss_cns_5: 0.6025, loss_yns_5: 0.1659, loss_cls_dn_0: 0.4170, loss_box_dn_0: 0.9147, loss_cls_dn_1: 0.3198, loss_box_dn_1: 1.1933, loss_cls_dn_2: 0.3729, loss_box_dn_2: 1.1591, loss_cls_dn_3: 0.3562, loss_box_dn_3: 1.1515, loss_cls_dn_4: 0.3556, loss_box_dn_4: 1.1582, loss_cls_dn_5: 0.3592, loss_box_dn_5: 1.1454, loss_dense_depth: 1.0000, loss: 35.7868, grad_norm: 52.9440 -2025-11-12 20:08:28,844 - mmdet - INFO - Iter [50/17500] lr: 1.196e-04, eta: 18:36:51, time: 1.516, data_time: 0.074, memory: 49163, loss_cls_0: 1.0298, loss_box_0: 2.0569, loss_cns_0: 0.6165, loss_yns_0: 0.1628, loss_cls_1: 1.0849, loss_box_1: 2.4537, loss_cns_1: 0.6009, loss_yns_1: 0.1660, loss_cls_2: 1.1076, loss_box_2: 2.3967, loss_cns_2: 0.6283, loss_yns_2: 0.1697, loss_cls_3: 1.1213, loss_box_3: 2.4072, loss_cns_3: 0.6203, loss_yns_3: 0.1652, loss_cls_4: 1.1092, loss_box_4: 2.4484, loss_cns_4: 0.6185, loss_yns_4: 0.1647, loss_cls_5: 1.1346, loss_box_5: 2.4715, loss_cns_5: 0.6203, loss_yns_5: 0.1638, loss_cls_dn_0: 0.4080, loss_box_dn_0: 0.9329, loss_cls_dn_1: 0.3238, loss_box_dn_1: 1.0762, loss_cls_dn_2: 0.3672, loss_box_dn_2: 1.0317, loss_cls_dn_3: 0.3557, loss_box_dn_3: 1.0390, loss_cls_dn_4: 0.3590, loss_box_dn_4: 1.0687, loss_cls_dn_5: 0.3636, loss_box_dn_5: 1.0844, loss_dense_depth: 1.0065, loss: 34.9355, grad_norm: 46.2172 -2025-11-12 20:08:30,378 - mmdet - INFO - Iter [51/17500] lr: 1.200e-04, eta: 18:23:38, time: 1.534, data_time: 0.072, memory: 49163, loss_cls_0: 1.0175, loss_box_0: 2.0364, loss_cns_0: 0.6172, loss_yns_0: 0.1626, loss_cls_1: 1.0685, loss_box_1: 2.4817, loss_cns_1: 0.5878, loss_yns_1: 0.1682, loss_cls_2: 1.1023, loss_box_2: 2.4365, loss_cns_2: 0.6015, loss_yns_2: 0.1670, loss_cls_3: 1.1223, loss_box_3: 2.4381, loss_cns_3: 0.6033, loss_yns_3: 0.1642, loss_cls_4: 1.1080, loss_box_4: 2.4848, loss_cns_4: 0.6008, loss_yns_4: 0.1638, loss_cls_5: 1.1285, loss_box_5: 2.5797, loss_cns_5: 0.5828, loss_yns_5: 0.1629, loss_cls_dn_0: 0.4066, loss_box_dn_0: 0.9272, loss_cls_dn_1: 0.3089, loss_box_dn_1: 1.1344, loss_cls_dn_2: 0.3377, loss_box_dn_2: 1.0710, loss_cls_dn_3: 0.3330, loss_box_dn_3: 1.0907, loss_cls_dn_4: 0.3417, loss_box_dn_4: 1.1356, loss_cls_dn_5: 0.3548, loss_box_dn_5: 1.1871, loss_dense_depth: 0.9332, loss: 35.1484, grad_norm: 53.9445 -2025-11-12 20:08:31,902 - mmdet - INFO - Iter [52/17500] lr: 1.204e-04, eta: 18:10:52, time: 1.523, data_time: 0.074, memory: 49163, loss_cls_0: 1.0359, loss_box_0: 2.0419, loss_cns_0: 0.6142, loss_yns_0: 0.1641, loss_cls_1: 1.0859, loss_box_1: 2.4431, loss_cns_1: 0.5861, loss_yns_1: 0.1701, loss_cls_2: 1.1331, loss_box_2: 2.3642, loss_cns_2: 0.6112, loss_yns_2: 0.1653, loss_cls_3: 1.1388, loss_box_3: 2.3652, loss_cns_3: 0.6119, loss_yns_3: 0.1637, loss_cls_4: 1.1267, loss_box_4: 2.4017, loss_cns_4: 0.6124, loss_yns_4: 0.1626, loss_cls_5: 1.1327, loss_box_5: 2.4594, loss_cns_5: 0.6164, loss_yns_5: 0.1639, loss_cls_dn_0: 0.4120, loss_box_dn_0: 0.9219, loss_cls_dn_1: 0.3075, loss_box_dn_1: 1.1627, loss_cls_dn_2: 0.3257, loss_box_dn_2: 1.1000, loss_cls_dn_3: 0.3325, loss_box_dn_3: 1.1386, loss_cls_dn_4: 0.3377, loss_box_dn_4: 1.1963, loss_cls_dn_5: 0.3653, loss_box_dn_5: 1.2459, loss_dense_depth: 0.9845, loss: 35.2010, grad_norm: 54.4458 -2025-11-12 20:08:33,423 - mmdet - INFO - Iter [53/17500] lr: 1.208e-04, eta: 17:58:34, time: 1.523, data_time: 0.074, memory: 49163, loss_cls_0: 1.0405, loss_box_0: 2.0625, loss_cns_0: 0.6115, loss_yns_0: 0.1653, loss_cls_1: 1.0899, loss_box_1: 2.4853, loss_cns_1: 0.5860, loss_yns_1: 0.1655, loss_cls_2: 1.1421, loss_box_2: 2.4222, loss_cns_2: 0.6074, loss_yns_2: 0.1657, loss_cls_3: 1.1503, loss_box_3: 2.4400, loss_cns_3: 0.6034, loss_yns_3: 0.1634, loss_cls_4: 1.1523, loss_box_4: 2.4908, loss_cns_4: 0.6056, loss_yns_4: 0.1630, loss_cls_5: 1.1465, loss_box_5: 2.4835, loss_cns_5: 0.6207, loss_yns_5: 0.1625, loss_cls_dn_0: 0.4188, loss_box_dn_0: 0.9082, loss_cls_dn_1: 0.3112, loss_box_dn_1: 1.1038, loss_cls_dn_2: 0.3242, loss_box_dn_2: 1.0688, loss_cls_dn_3: 0.3369, loss_box_dn_3: 1.1156, loss_cls_dn_4: 0.3315, loss_box_dn_4: 1.1818, loss_cls_dn_5: 0.3716, loss_box_dn_5: 1.2103, loss_dense_depth: 1.0185, loss: 35.4268, grad_norm: 62.1471 -2025-11-12 20:08:34,959 - mmdet - INFO - Iter [54/17500] lr: 1.212e-04, eta: 17:46:48, time: 1.535, data_time: 0.074, memory: 49163, loss_cls_0: 1.0241, loss_box_0: 1.9906, loss_cns_0: 0.6166, loss_yns_0: 0.1623, loss_cls_1: 1.0683, loss_box_1: 2.4162, loss_cns_1: 0.5985, loss_yns_1: 0.1612, loss_cls_2: 1.1086, loss_box_2: 2.3911, loss_cns_2: 0.6086, loss_yns_2: 0.1622, loss_cls_3: 1.1231, loss_box_3: 2.4075, loss_cns_3: 0.6008, loss_yns_3: 0.1622, loss_cls_4: 1.1441, loss_box_4: 2.4382, loss_cns_4: 0.6047, loss_yns_4: 0.1617, loss_cls_5: 1.1330, loss_box_5: 2.4415, loss_cns_5: 0.6155, loss_yns_5: 0.1608, loss_cls_dn_0: 0.4035, loss_box_dn_0: 0.8999, loss_cls_dn_1: 0.3041, loss_box_dn_1: 1.1330, loss_cls_dn_2: 0.3218, loss_box_dn_2: 1.1156, loss_cls_dn_3: 0.3318, loss_box_dn_3: 1.1542, loss_cls_dn_4: 0.3169, loss_box_dn_4: 1.2007, loss_cls_dn_5: 0.3575, loss_box_dn_5: 1.2195, loss_dense_depth: 0.9967, loss: 35.0566, grad_norm: 57.8820 -2025-11-12 20:08:36,565 - mmdet - INFO - Iter [55/17500] lr: 1.216e-04, eta: 17:35:50, time: 1.606, data_time: 0.075, memory: 49163, loss_cls_0: 1.0333, loss_box_0: 1.9848, loss_cns_0: 0.6133, loss_yns_0: 0.1657, loss_cls_1: 1.0588, loss_box_1: 2.2850, loss_cns_1: 0.6104, loss_yns_1: 0.1655, loss_cls_2: 1.1005, loss_box_2: 2.2276, loss_cns_2: 0.6221, loss_yns_2: 0.1658, loss_cls_3: 1.1178, loss_box_3: 2.2352, loss_cns_3: 0.6253, loss_yns_3: 0.1657, loss_cls_4: 1.1423, loss_box_4: 2.2337, loss_cns_4: 0.6241, loss_yns_4: 0.1649, loss_cls_5: 1.1249, loss_box_5: 2.2475, loss_cns_5: 0.6310, loss_yns_5: 0.1645, loss_cls_dn_0: 0.4119, loss_box_dn_0: 0.8937, loss_cls_dn_1: 0.3173, loss_box_dn_1: 0.9791, loss_cls_dn_2: 0.3516, loss_box_dn_2: 0.9498, loss_cls_dn_3: 0.3557, loss_box_dn_3: 0.9708, loss_cls_dn_4: 0.3364, loss_box_dn_4: 0.9909, loss_cls_dn_5: 0.3757, loss_box_dn_5: 1.0029, loss_dense_depth: 0.9958, loss: 33.4415, grad_norm: 44.4383 -2025-11-12 20:08:38,105 - mmdet - INFO - Iter [56/17500] lr: 1.220e-04, eta: 17:24:55, time: 1.540, data_time: 0.075, memory: 49163, loss_cls_0: 1.0033, loss_box_0: 1.9729, loss_cns_0: 0.6064, loss_yns_0: 0.1640, loss_cls_1: 1.0524, loss_box_1: 2.1856, loss_cns_1: 0.6115, loss_yns_1: 0.1662, loss_cls_2: 1.0827, loss_box_2: 2.1457, loss_cns_2: 0.6235, loss_yns_2: 0.1651, loss_cls_3: 1.0914, loss_box_3: 2.1499, loss_cns_3: 0.6282, loss_yns_3: 0.1643, loss_cls_4: 1.0975, loss_box_4: 2.2139, loss_cns_4: 0.6108, loss_yns_4: 0.1654, loss_cls_5: 1.1110, loss_box_5: 2.2300, loss_cns_5: 0.6105, loss_yns_5: 0.1658, loss_cls_dn_0: 0.3964, loss_box_dn_0: 0.8998, loss_cls_dn_1: 0.2948, loss_box_dn_1: 0.9790, loss_cls_dn_2: 0.3442, loss_box_dn_2: 0.9423, loss_cls_dn_3: 0.3403, loss_box_dn_3: 0.9440, loss_cls_dn_4: 0.3293, loss_box_dn_4: 0.9749, loss_cls_dn_5: 0.3615, loss_box_dn_5: 0.9802, loss_dense_depth: 0.9625, loss: 32.7672, grad_norm: 52.8701 -2025-11-12 20:08:39,650 - mmdet - INFO - Iter [57/17500] lr: 1.224e-04, eta: 17:14:24, time: 1.543, data_time: 0.077, memory: 49163, loss_cls_0: 1.0193, loss_box_0: 1.9478, loss_cns_0: 0.6100, loss_yns_0: 0.1642, loss_cls_1: 1.0627, loss_box_1: 2.2186, loss_cns_1: 0.6106, loss_yns_1: 0.1660, loss_cls_2: 1.0878, loss_box_2: 2.1837, loss_cns_2: 0.6235, loss_yns_2: 0.1643, loss_cls_3: 1.1251, loss_box_3: 2.1904, loss_cns_3: 0.6257, loss_yns_3: 0.1670, loss_cls_4: 1.1272, loss_box_4: 2.2516, loss_cns_4: 0.6130, loss_yns_4: 0.1671, loss_cls_5: 1.1237, loss_box_5: 2.2631, loss_cns_5: 0.6185, loss_yns_5: 0.1658, loss_cls_dn_0: 0.3846, loss_box_dn_0: 0.8946, loss_cls_dn_1: 0.2773, loss_box_dn_1: 0.9284, loss_cls_dn_2: 0.3290, loss_box_dn_2: 0.8982, loss_cls_dn_3: 0.3240, loss_box_dn_3: 0.8963, loss_cls_dn_4: 0.3217, loss_box_dn_4: 0.9261, loss_cls_dn_5: 0.3473, loss_box_dn_5: 0.9449, loss_dense_depth: 1.0145, loss: 32.7836, grad_norm: 50.2643 -2025-11-12 20:08:41,162 - mmdet - INFO - Iter [58/17500] lr: 1.228e-04, eta: 17:04:06, time: 1.513, data_time: 0.075, memory: 49163, loss_cls_0: 1.0027, loss_box_0: 1.9122, loss_cns_0: 0.6138, loss_yns_0: 0.1642, loss_cls_1: 1.0463, loss_box_1: 2.1937, loss_cns_1: 0.6091, loss_yns_1: 0.1669, loss_cls_2: 1.0838, loss_box_2: 2.1615, loss_cns_2: 0.6220, loss_yns_2: 0.1665, loss_cls_3: 1.0948, loss_box_3: 2.1615, loss_cns_3: 0.6253, loss_yns_3: 0.1663, loss_cls_4: 1.1014, loss_box_4: 2.1507, loss_cns_4: 0.6293, loss_yns_4: 0.1655, loss_cls_5: 1.1053, loss_box_5: 2.1663, loss_cns_5: 0.6302, loss_yns_5: 0.1672, loss_cls_dn_0: 0.3888, loss_box_dn_0: 0.8812, loss_cls_dn_1: 0.2718, loss_box_dn_1: 0.9125, loss_cls_dn_2: 0.3188, loss_box_dn_2: 0.9006, loss_cls_dn_3: 0.3253, loss_box_dn_3: 0.9098, loss_cls_dn_4: 0.3261, loss_box_dn_4: 0.9196, loss_cls_dn_5: 0.3380, loss_box_dn_5: 0.9552, loss_dense_depth: 0.9850, loss: 32.3390, grad_norm: 40.9287 -2025-11-12 20:08:42,687 - mmdet - INFO - Iter [59/17500] lr: 1.232e-04, eta: 16:54:11, time: 1.524, data_time: 0.079, memory: 49163, loss_cls_0: 0.9835, loss_box_0: 1.9230, loss_cns_0: 0.6146, loss_yns_0: 0.1632, loss_cls_1: 1.0508, loss_box_1: 2.1900, loss_cns_1: 0.6100, loss_yns_1: 0.1647, loss_cls_2: 1.0785, loss_box_2: 2.1485, loss_cns_2: 0.6235, loss_yns_2: 0.1676, loss_cls_3: 1.0795, loss_box_3: 2.1619, loss_cns_3: 0.6222, loss_yns_3: 0.1659, loss_cls_4: 1.0798, loss_box_4: 2.1491, loss_cns_4: 0.6275, loss_yns_4: 0.1647, loss_cls_5: 1.1048, loss_box_5: 2.2120, loss_cns_5: 0.6254, loss_yns_5: 0.1746, loss_cls_dn_0: 0.3975, loss_box_dn_0: 0.8745, loss_cls_dn_1: 0.2723, loss_box_dn_1: 0.9528, loss_cls_dn_2: 0.3123, loss_box_dn_2: 0.9574, loss_cls_dn_3: 0.3378, loss_box_dn_3: 0.9799, loss_cls_dn_4: 0.3485, loss_box_dn_4: 0.9952, loss_cls_dn_5: 0.3423, loss_box_dn_5: 1.0596, loss_dense_depth: 1.0311, loss: 32.7466, grad_norm: 57.0302 -2025-11-12 20:08:44,212 - mmdet - INFO - Iter [60/17500] lr: 1.236e-04, eta: 16:44:37, time: 1.526, data_time: 0.080, memory: 49163, loss_cls_0: 0.9946, loss_box_0: 1.8841, loss_cns_0: 0.6191, loss_yns_0: 0.1619, loss_cls_1: 1.0921, loss_box_1: 2.0867, loss_cns_1: 0.6258, loss_yns_1: 0.1640, loss_cls_2: 1.1142, loss_box_2: 2.0418, loss_cns_2: 0.6391, loss_yns_2: 0.1662, loss_cls_3: 1.1144, loss_box_3: 2.0452, loss_cns_3: 0.6352, loss_yns_3: 0.1637, loss_cls_4: 1.1034, loss_box_4: 2.0471, loss_cns_4: 0.6392, loss_yns_4: 0.1638, loss_cls_5: 1.1372, loss_box_5: 2.1089, loss_cns_5: 0.6310, loss_yns_5: 0.1710, loss_cls_dn_0: 0.3966, loss_box_dn_0: 0.8817, loss_cls_dn_1: 0.2616, loss_box_dn_1: 0.9871, loss_cls_dn_2: 0.2991, loss_box_dn_2: 0.9937, loss_cls_dn_3: 0.3368, loss_box_dn_3: 1.0112, loss_cls_dn_4: 0.3566, loss_box_dn_4: 1.0273, loss_cls_dn_5: 0.3560, loss_box_dn_5: 1.0824, loss_dense_depth: 0.9752, loss: 32.5148, grad_norm: 52.2218 -2025-11-12 20:08:45,810 - mmdet - INFO - Iter [61/17500] lr: 1.240e-04, eta: 16:35:42, time: 1.597, data_time: 0.159, memory: 49163, loss_cls_0: 1.0027, loss_box_0: 1.8968, loss_cns_0: 0.6168, loss_yns_0: 0.1602, loss_cls_1: 1.0899, loss_box_1: 2.0769, loss_cns_1: 0.6265, loss_yns_1: 0.1656, loss_cls_2: 1.1200, loss_box_2: 2.0552, loss_cns_2: 0.6328, loss_yns_2: 0.1625, loss_cls_3: 1.1088, loss_box_3: 2.0738, loss_cns_3: 0.6342, loss_yns_3: 0.1618, loss_cls_4: 1.0999, loss_box_4: 2.0919, loss_cns_4: 0.6352, loss_yns_4: 0.1639, loss_cls_5: 1.1074, loss_box_5: 2.0677, loss_cns_5: 0.6387, loss_yns_5: 0.1670, loss_cls_dn_0: 0.3863, loss_box_dn_0: 0.8774, loss_cls_dn_1: 0.2524, loss_box_dn_1: 1.0389, loss_cls_dn_2: 0.2943, loss_box_dn_2: 1.0351, loss_cls_dn_3: 0.3175, loss_box_dn_3: 1.0474, loss_cls_dn_4: 0.3384, loss_box_dn_4: 1.0620, loss_cls_dn_5: 0.3589, loss_box_dn_5: 1.0765, loss_dense_depth: 1.0285, loss: 32.6698, grad_norm: 49.4540 -2025-11-12 20:08:47,363 - mmdet - INFO - Iter [62/17500] lr: 1.244e-04, eta: 16:26:52, time: 1.553, data_time: 0.118, memory: 49163, loss_cls_0: 1.0275, loss_box_0: 1.9023, loss_cns_0: 0.6158, loss_yns_0: 0.1585, loss_cls_1: 1.0789, loss_box_1: 2.1172, loss_cns_1: 0.6272, loss_yns_1: 0.1648, loss_cls_2: 1.0935, loss_box_2: 2.0769, loss_cns_2: 0.6346, loss_yns_2: 0.1629, loss_cls_3: 1.1144, loss_box_3: 2.0914, loss_cns_3: 0.6384, loss_yns_3: 0.1621, loss_cls_4: 1.1092, loss_box_4: 2.1132, loss_cns_4: 0.6363, loss_yns_4: 0.1647, loss_cls_5: 1.0982, loss_box_5: 2.0687, loss_cns_5: 0.6425, loss_yns_5: 0.1665, loss_cls_dn_0: 0.3747, loss_box_dn_0: 0.8732, loss_cls_dn_1: 0.2616, loss_box_dn_1: 0.9756, loss_cls_dn_2: 0.3097, loss_box_dn_2: 0.9591, loss_cls_dn_3: 0.3066, loss_box_dn_3: 0.9636, loss_cls_dn_4: 0.3070, loss_box_dn_4: 0.9773, loss_cls_dn_5: 0.3556, loss_box_dn_5: 0.9735, loss_dense_depth: 0.9717, loss: 32.2751, grad_norm: 50.4716 -2025-11-12 20:08:48,941 - mmdet - INFO - Iter [63/17500] lr: 1.248e-04, eta: 16:18:26, time: 1.579, data_time: 0.074, memory: 49163, loss_cls_0: 1.0140, loss_box_0: 1.9007, loss_cns_0: 0.6140, loss_yns_0: 0.1607, loss_cls_1: 1.0944, loss_box_1: 2.1524, loss_cns_1: 0.6257, loss_yns_1: 0.1636, loss_cls_2: 1.1011, loss_box_2: 2.0960, loss_cns_2: 0.6392, loss_yns_2: 0.1683, loss_cls_3: 1.1326, loss_box_3: 2.0866, loss_cns_3: 0.6445, loss_yns_3: 0.1628, loss_cls_4: 1.1785, loss_box_4: 2.0883, loss_cns_4: 0.6442, loss_yns_4: 0.1637, loss_cls_5: 1.1445, loss_box_5: 2.0720, loss_cns_5: 0.6481, loss_yns_5: 0.1707, loss_cls_dn_0: 0.3809, loss_box_dn_0: 0.8724, loss_cls_dn_1: 0.2583, loss_box_dn_1: 0.9580, loss_cls_dn_2: 0.3092, loss_box_dn_2: 0.9156, loss_cls_dn_3: 0.2981, loss_box_dn_3: 0.9054, loss_cls_dn_4: 0.2773, loss_box_dn_4: 0.9120, loss_cls_dn_5: 0.3265, loss_box_dn_5: 0.9136, loss_dense_depth: 1.0311, loss: 32.2248, grad_norm: 34.3502 -2025-11-12 20:08:50,514 - mmdet - INFO - Iter [64/17500] lr: 1.252e-04, eta: 16:10:13, time: 1.572, data_time: 0.098, memory: 49163, loss_cls_0: 0.9808, loss_box_0: 1.8916, loss_cns_0: 0.6176, loss_yns_0: 0.1611, loss_cls_1: 1.0703, loss_box_1: 2.1828, loss_cns_1: 0.6214, loss_yns_1: 0.1650, loss_cls_2: 1.0924, loss_box_2: 2.1348, loss_cns_2: 0.6398, loss_yns_2: 0.1665, loss_cls_3: 1.0959, loss_box_3: 2.1536, loss_cns_3: 0.6387, loss_yns_3: 0.1642, loss_cls_4: 1.0858, loss_box_4: 2.1496, loss_cns_4: 0.6345, loss_yns_4: 0.1638, loss_cls_5: 1.0974, loss_box_5: 2.1458, loss_cns_5: 0.6375, loss_yns_5: 0.1743, loss_cls_dn_0: 0.3878, loss_box_dn_0: 0.8691, loss_cls_dn_1: 0.2366, loss_box_dn_1: 0.9309, loss_cls_dn_2: 0.2870, loss_box_dn_2: 0.8954, loss_cls_dn_3: 0.3111, loss_box_dn_3: 0.8998, loss_cls_dn_4: 0.3024, loss_box_dn_4: 0.9066, loss_cls_dn_5: 0.3182, loss_box_dn_5: 0.9116, loss_dense_depth: 0.9363, loss: 32.0580, grad_norm: 56.6977 -2025-11-12 20:08:52,044 - mmdet - INFO - Iter [65/17500] lr: 1.256e-04, eta: 16:02:05, time: 1.531, data_time: 0.076, memory: 49163, loss_cls_0: 0.9799, loss_box_0: 1.8974, loss_cns_0: 0.6177, loss_yns_0: 0.1618, loss_cls_1: 1.0316, loss_box_1: 2.1906, loss_cns_1: 0.6046, loss_yns_1: 0.1625, loss_cls_2: 1.0655, loss_box_2: 2.1535, loss_cns_2: 0.6329, loss_yns_2: 0.1624, loss_cls_3: 1.1005, loss_box_3: 2.1788, loss_cns_3: 0.6304, loss_yns_3: 0.1621, loss_cls_4: 1.1082, loss_box_4: 2.1983, loss_cns_4: 0.6274, loss_yns_4: 0.1634, loss_cls_5: 1.0758, loss_box_5: 2.1748, loss_cns_5: 0.6332, loss_yns_5: 0.1684, loss_cls_dn_0: 0.3944, loss_box_dn_0: 0.8793, loss_cls_dn_1: 0.2211, loss_box_dn_1: 0.9543, loss_cls_dn_2: 0.2660, loss_box_dn_2: 0.9200, loss_cls_dn_3: 0.3131, loss_box_dn_3: 0.9358, loss_cls_dn_4: 0.3234, loss_box_dn_4: 0.9610, loss_cls_dn_5: 0.3073, loss_box_dn_5: 0.9547, loss_dense_depth: 0.9663, loss: 32.2784, grad_norm: 57.5303 -2025-11-12 20:08:53,583 - mmdet - INFO - Iter [66/17500] lr: 1.260e-04, eta: 15:54:14, time: 1.539, data_time: 0.076, memory: 49163, loss_cls_0: 0.9730, loss_box_0: 1.8774, loss_cns_0: 0.6206, loss_yns_0: 0.1626, loss_cls_1: 1.0313, loss_box_1: 2.1201, loss_cns_1: 0.6072, loss_yns_1: 0.1625, loss_cls_2: 1.0648, loss_box_2: 2.0600, loss_cns_2: 0.6350, loss_yns_2: 0.1624, loss_cls_3: 1.0642, loss_box_3: 2.0731, loss_cns_3: 0.6366, loss_yns_3: 0.1618, loss_cls_4: 1.0833, loss_box_4: 2.0910, loss_cns_4: 0.6356, loss_yns_4: 0.1630, loss_cls_5: 1.0702, loss_box_5: 2.0576, loss_cns_5: 0.6374, loss_yns_5: 0.1631, loss_cls_dn_0: 0.3722, loss_box_dn_0: 0.8709, loss_cls_dn_1: 0.2108, loss_box_dn_1: 0.9807, loss_cls_dn_2: 0.2420, loss_box_dn_2: 0.9434, loss_cls_dn_3: 0.2877, loss_box_dn_3: 0.9638, loss_cls_dn_4: 0.3116, loss_box_dn_4: 0.9948, loss_cls_dn_5: 0.2824, loss_box_dn_5: 0.9893, loss_dense_depth: 0.9204, loss: 31.6838, grad_norm: 40.7845 -2025-11-12 20:08:55,097 - mmdet - INFO - Iter [67/17500] lr: 1.264e-04, eta: 15:46:30, time: 1.514, data_time: 0.073, memory: 49163, loss_cls_0: 0.9987, loss_box_0: 1.8971, loss_cns_0: 0.6176, loss_yns_0: 0.1627, loss_cls_1: 1.0308, loss_box_1: 2.1031, loss_cns_1: 0.6080, loss_yns_1: 0.1598, loss_cls_2: 1.0997, loss_box_2: 2.0777, loss_cns_2: 0.6377, loss_yns_2: 0.1656, loss_cls_3: 1.0899, loss_box_3: 2.1012, loss_cns_3: 0.6383, loss_yns_3: 0.1663, loss_cls_4: 1.0793, loss_box_4: 2.1018, loss_cns_4: 0.6377, loss_yns_4: 0.1648, loss_cls_5: 1.1096, loss_box_5: 2.1550, loss_cns_5: 0.6249, loss_yns_5: 0.1717, loss_cls_dn_0: 0.3673, loss_box_dn_0: 0.8716, loss_cls_dn_1: 0.2222, loss_box_dn_1: 0.9798, loss_cls_dn_2: 0.2428, loss_box_dn_2: 0.9615, loss_cls_dn_3: 0.2692, loss_box_dn_3: 0.9957, loss_cls_dn_4: 0.2995, loss_box_dn_4: 1.0326, loss_cls_dn_5: 0.2879, loss_box_dn_5: 1.0622, loss_dense_depth: 0.9464, loss: 32.1377, grad_norm: 66.0553 -2025-11-12 20:08:56,626 - mmdet - INFO - Iter [68/17500] lr: 1.268e-04, eta: 15:39:04, time: 1.529, data_time: 0.075, memory: 49163, loss_cls_0: 0.9976, loss_box_0: 1.8779, loss_cns_0: 0.6196, loss_yns_0: 0.1611, loss_cls_1: 1.0429, loss_box_1: 2.0865, loss_cns_1: 0.6075, loss_yns_1: 0.1604, loss_cls_2: 1.0650, loss_box_2: 2.0650, loss_cns_2: 0.6325, loss_yns_2: 0.1658, loss_cls_3: 1.1189, loss_box_3: 2.0680, loss_cns_3: 0.6358, loss_yns_3: 0.1638, loss_cls_4: 1.0789, loss_box_4: 2.0791, loss_cns_4: 0.6333, loss_yns_4: 0.1630, loss_cls_5: 1.0870, loss_box_5: 2.1624, loss_cns_5: 0.6199, loss_yns_5: 0.1742, loss_cls_dn_0: 0.3632, loss_box_dn_0: 0.8793, loss_cls_dn_1: 0.2421, loss_box_dn_1: 0.9994, loss_cls_dn_2: 0.2581, loss_box_dn_2: 0.9860, loss_cls_dn_3: 0.2754, loss_box_dn_3: 1.0252, loss_cls_dn_4: 0.2987, loss_box_dn_4: 1.0660, loss_cls_dn_5: 0.3157, loss_box_dn_5: 1.1163, loss_dense_depth: 0.9040, loss: 32.1955, grad_norm: 62.4889 -2025-11-12 20:08:58,151 - mmdet - INFO - Iter [69/17500] lr: 1.272e-04, eta: 15:31:49, time: 1.526, data_time: 0.077, memory: 49163, loss_cls_0: 0.9686, loss_box_0: 1.8236, loss_cns_0: 0.6236, loss_yns_0: 0.1598, loss_cls_1: 1.0290, loss_box_1: 2.0990, loss_cns_1: 0.6239, loss_yns_1: 0.1603, loss_cls_2: 1.0570, loss_box_2: 2.0606, loss_cns_2: 0.6408, loss_yns_2: 0.1637, loss_cls_3: 1.0866, loss_box_3: 2.0331, loss_cns_3: 0.6484, loss_yns_3: 0.1604, loss_cls_4: 1.0928, loss_box_4: 2.0499, loss_cns_4: 0.6451, loss_yns_4: 0.1612, loss_cls_5: 1.0742, loss_box_5: 2.0566, loss_cns_5: 0.6449, loss_yns_5: 0.1724, loss_cls_dn_0: 0.3522, loss_box_dn_0: 0.8643, loss_cls_dn_1: 0.2341, loss_box_dn_1: 0.9797, loss_cls_dn_2: 0.2614, loss_box_dn_2: 0.9644, loss_cls_dn_3: 0.2725, loss_box_dn_3: 0.9841, loss_cls_dn_4: 0.2735, loss_box_dn_4: 1.0191, loss_cls_dn_5: 0.3177, loss_box_dn_5: 1.0352, loss_dense_depth: 0.9066, loss: 31.7004, grad_norm: 51.5597 -2025-11-12 20:08:59,663 - mmdet - INFO - Iter [70/17500] lr: 1.276e-04, eta: 15:24:44, time: 1.512, data_time: 0.074, memory: 49163, loss_cls_0: 0.9620, loss_box_0: 1.8225, loss_cns_0: 0.6237, loss_yns_0: 0.1627, loss_cls_1: 1.0002, loss_box_1: 2.1084, loss_cns_1: 0.6193, loss_yns_1: 0.1618, loss_cls_2: 1.0508, loss_box_2: 2.0709, loss_cns_2: 0.6308, loss_yns_2: 0.1610, loss_cls_3: 1.0575, loss_box_3: 2.0663, loss_cns_3: 0.6354, loss_yns_3: 0.1600, loss_cls_4: 1.1015, loss_box_4: 2.0484, loss_cns_4: 0.6418, loss_yns_4: 0.1618, loss_cls_5: 1.0714, loss_box_5: 2.0279, loss_cns_5: 0.6453, loss_yns_5: 0.1627, loss_cls_dn_0: 0.3717, loss_box_dn_0: 0.8592, loss_cls_dn_1: 0.2326, loss_box_dn_1: 0.9384, loss_cls_dn_2: 0.2737, loss_box_dn_2: 0.9254, loss_cls_dn_3: 0.2879, loss_box_dn_3: 0.9373, loss_cls_dn_4: 0.2717, loss_box_dn_4: 0.9437, loss_cls_dn_5: 0.3205, loss_box_dn_5: 0.9389, loss_dense_depth: 0.8940, loss: 31.3493, grad_norm: 50.1703 -2025-11-12 20:09:01,185 - mmdet - INFO - Iter [71/17500] lr: 1.280e-04, eta: 15:17:53, time: 1.521, data_time: 0.075, memory: 49163, loss_cls_0: 0.9470, loss_box_0: 1.8611, loss_cns_0: 0.6196, loss_yns_0: 0.1615, loss_cls_1: 1.0419, loss_box_1: 2.1404, loss_cns_1: 0.6104, loss_yns_1: 0.1607, loss_cls_2: 1.0229, loss_box_2: 2.0723, loss_cns_2: 0.6296, loss_yns_2: 0.1577, loss_cls_3: 1.0401, loss_box_3: 2.0811, loss_cns_3: 0.6323, loss_yns_3: 0.1578, loss_cls_4: 1.0740, loss_box_4: 2.0268, loss_cns_4: 0.6425, loss_yns_4: 0.1574, loss_cls_5: 1.0544, loss_box_5: 2.0636, loss_cns_5: 0.6438, loss_yns_5: 0.1581, loss_cls_dn_0: 0.3690, loss_box_dn_0: 0.8642, loss_cls_dn_1: 0.2212, loss_box_dn_1: 0.9514, loss_cls_dn_2: 0.2540, loss_box_dn_2: 0.9224, loss_cls_dn_3: 0.2766, loss_box_dn_3: 0.9246, loss_cls_dn_4: 0.2634, loss_box_dn_4: 0.9043, loss_cls_dn_5: 0.2906, loss_box_dn_5: 0.9120, loss_dense_depth: 0.9696, loss: 31.2803, grad_norm: 56.4383 -2025-11-12 20:09:02,700 - mmdet - INFO - Iter [72/17500] lr: 1.284e-04, eta: 15:11:12, time: 1.515, data_time: 0.076, memory: 49163, loss_cls_0: 0.9640, loss_box_0: 1.8711, loss_cns_0: 0.6216, loss_yns_0: 0.1599, loss_cls_1: 1.0479, loss_box_1: 2.0843, loss_cns_1: 0.6197, loss_yns_1: 0.1617, loss_cls_2: 1.0525, loss_box_2: 1.9881, loss_cns_2: 0.6368, loss_yns_2: 0.1626, loss_cls_3: 1.0588, loss_box_3: 1.9656, loss_cns_3: 0.6418, loss_yns_3: 0.1622, loss_cls_4: 1.0604, loss_box_4: 1.9558, loss_cns_4: 0.6458, loss_yns_4: 0.1602, loss_cls_5: 1.0719, loss_box_5: 1.9873, loss_cns_5: 0.6444, loss_yns_5: 0.1668, loss_cls_dn_0: 0.3741, loss_box_dn_0: 0.8600, loss_cls_dn_1: 0.2285, loss_box_dn_1: 0.8564, loss_cls_dn_2: 0.2516, loss_box_dn_2: 0.8175, loss_cls_dn_3: 0.2807, loss_box_dn_3: 0.8081, loss_cls_dn_4: 0.2821, loss_box_dn_4: 0.8102, loss_cls_dn_5: 0.2886, loss_box_dn_5: 0.8304, loss_dense_depth: 0.9096, loss: 30.4892, grad_norm: 42.2750 -2025-11-12 20:09:04,229 - mmdet - INFO - Iter [73/17500] lr: 1.288e-04, eta: 15:04:45, time: 1.529, data_time: 0.073, memory: 49163, loss_cls_0: 0.9793, loss_box_0: 1.8961, loss_cns_0: 0.6166, loss_yns_0: 0.1605, loss_cls_1: 1.0519, loss_box_1: 2.1377, loss_cns_1: 0.6144, loss_yns_1: 0.1614, loss_cls_2: 1.0762, loss_box_2: 2.0979, loss_cns_2: 0.6262, loss_yns_2: 0.1618, loss_cls_3: 1.0941, loss_box_3: 2.0747, loss_cns_3: 0.6353, loss_yns_3: 0.1607, loss_cls_4: 1.0972, loss_box_4: 2.0948, loss_cns_4: 0.6342, loss_yns_4: 0.1587, loss_cls_5: 1.1082, loss_box_5: 2.0701, loss_cns_5: 0.6357, loss_yns_5: 0.1732, loss_cls_dn_0: 0.3667, loss_box_dn_0: 0.8626, loss_cls_dn_1: 0.2291, loss_box_dn_1: 0.8676, loss_cls_dn_2: 0.2474, loss_box_dn_2: 0.8413, loss_cls_dn_3: 0.2820, loss_box_dn_3: 0.8383, loss_cls_dn_4: 0.3010, loss_box_dn_4: 0.8675, loss_cls_dn_5: 0.2844, loss_box_dn_5: 0.8874, loss_dense_depth: 0.9399, loss: 31.3320, grad_norm: 50.2276 -2025-11-12 20:09:05,756 - mmdet - INFO - Iter [74/17500] lr: 1.292e-04, eta: 14:58:27, time: 1.526, data_time: 0.077, memory: 49163, loss_cls_0: 0.9816, loss_box_0: 1.8673, loss_cns_0: 0.6185, loss_yns_0: 0.1595, loss_cls_1: 1.0548, loss_box_1: 2.1361, loss_cns_1: 0.6114, loss_yns_1: 0.1600, loss_cls_2: 1.0640, loss_box_2: 2.1196, loss_cns_2: 0.6217, loss_yns_2: 0.1600, loss_cls_3: 1.0792, loss_box_3: 2.0865, loss_cns_3: 0.6361, loss_yns_3: 0.1599, loss_cls_4: 1.0822, loss_box_4: 2.0899, loss_cns_4: 0.6351, loss_yns_4: 0.1603, loss_cls_5: 1.1090, loss_box_5: 2.0858, loss_cns_5: 0.6359, loss_yns_5: 0.1704, loss_cls_dn_0: 0.3496, loss_box_dn_0: 0.8566, loss_cls_dn_1: 0.2226, loss_box_dn_1: 0.9130, loss_cls_dn_2: 0.2353, loss_box_dn_2: 0.8939, loss_cls_dn_3: 0.2638, loss_box_dn_3: 0.9031, loss_cls_dn_4: 0.2873, loss_box_dn_4: 0.9330, loss_cls_dn_5: 0.2583, loss_box_dn_5: 0.9672, loss_dense_depth: 0.9280, loss: 31.4965, grad_norm: 51.0422 -2025-11-12 20:09:07,340 - mmdet - INFO - Iter [75/17500] lr: 1.296e-04, eta: 14:52:34, time: 1.585, data_time: 0.077, memory: 49163, loss_cls_0: 1.0160, loss_box_0: 1.8745, loss_cns_0: 0.6152, loss_yns_0: 0.1623, loss_cls_1: 1.0386, loss_box_1: 2.1588, loss_cns_1: 0.6159, loss_yns_1: 0.1640, loss_cls_2: 1.0584, loss_box_2: 2.1072, loss_cns_2: 0.6289, loss_yns_2: 0.1636, loss_cls_3: 1.0756, loss_box_3: 2.0822, loss_cns_3: 0.6385, loss_yns_3: 0.1642, loss_cls_4: 1.0614, loss_box_4: 2.0600, loss_cns_4: 0.6385, loss_yns_4: 0.1635, loss_cls_5: 1.1535, loss_box_5: 2.0966, loss_cns_5: 0.6323, loss_yns_5: 0.1640, loss_cls_dn_0: 0.3313, loss_box_dn_0: 0.8510, loss_cls_dn_1: 0.2157, loss_box_dn_1: 0.9401, loss_cls_dn_2: 0.2318, loss_box_dn_2: 0.9253, loss_cls_dn_3: 0.2428, loss_box_dn_3: 0.9404, loss_cls_dn_4: 0.2525, loss_box_dn_4: 0.9583, loss_cls_dn_5: 0.2419, loss_box_dn_5: 0.9989, loss_dense_depth: 0.9005, loss: 31.5640, grad_norm: 51.5188 -2025-11-12 20:09:08,865 - mmdet - INFO - Iter [76/17500] lr: 1.300e-04, eta: 14:46:36, time: 1.525, data_time: 0.074, memory: 49163, loss_cls_0: 0.9944, loss_box_0: 1.8635, loss_cns_0: 0.6175, loss_yns_0: 0.1640, loss_cls_1: 1.0614, loss_box_1: 2.0830, loss_cns_1: 0.6294, loss_yns_1: 0.1638, loss_cls_2: 1.0484, loss_box_2: 2.0555, loss_cns_2: 0.6389, loss_yns_2: 0.1632, loss_cls_3: 1.0693, loss_box_3: 2.0575, loss_cns_3: 0.6401, loss_yns_3: 0.1668, loss_cls_4: 1.0673, loss_box_4: 2.0553, loss_cns_4: 0.6401, loss_yns_4: 0.1649, loss_cls_5: 1.0689, loss_box_5: 2.0964, loss_cns_5: 0.6310, loss_yns_5: 0.1654, loss_cls_dn_0: 0.3481, loss_box_dn_0: 0.8526, loss_cls_dn_1: 0.2161, loss_box_dn_1: 0.9228, loss_cls_dn_2: 0.2331, loss_box_dn_2: 0.9181, loss_cls_dn_3: 0.2401, loss_box_dn_3: 0.9320, loss_cls_dn_4: 0.2482, loss_box_dn_4: 0.9479, loss_cls_dn_5: 0.2669, loss_box_dn_5: 0.9836, loss_dense_depth: 0.9212, loss: 31.3367, grad_norm: 54.4007 -2025-11-12 20:09:10,389 - mmdet - INFO - Iter [77/17500] lr: 1.304e-04, eta: 14:40:47, time: 1.524, data_time: 0.073, memory: 49163, loss_cls_0: 0.9598, loss_box_0: 1.8817, loss_cns_0: 0.6129, loss_yns_0: 0.1608, loss_cls_1: 1.0388, loss_box_1: 2.0297, loss_cns_1: 0.6313, loss_yns_1: 0.1637, loss_cls_2: 1.0609, loss_box_2: 1.9919, loss_cns_2: 0.6457, loss_yns_2: 0.1669, loss_cls_3: 1.0648, loss_box_3: 1.9897, loss_cns_3: 0.6477, loss_yns_3: 0.1677, loss_cls_4: 1.0723, loss_box_4: 1.9989, loss_cns_4: 0.6496, loss_yns_4: 0.1630, loss_cls_5: 1.1229, loss_box_5: 2.0223, loss_cns_5: 0.6478, loss_yns_5: 0.1753, loss_cls_dn_0: 0.3671, loss_box_dn_0: 0.8395, loss_cls_dn_1: 0.2178, loss_box_dn_1: 0.8895, loss_cls_dn_2: 0.2348, loss_box_dn_2: 0.8688, loss_cls_dn_3: 0.2418, loss_box_dn_3: 0.8700, loss_cls_dn_4: 0.2463, loss_box_dn_4: 0.8810, loss_cls_dn_5: 0.2996, loss_box_dn_5: 0.8990, loss_dense_depth: 0.8949, loss: 30.8162, grad_norm: 39.4893 -2025-11-12 20:09:11,926 - mmdet - INFO - Iter [78/17500] lr: 1.308e-04, eta: 14:35:08, time: 1.532, data_time: 0.073, memory: 49163, loss_cls_0: 0.9795, loss_box_0: 1.8896, loss_cns_0: 0.6122, loss_yns_0: 0.1625, loss_cls_1: 1.0458, loss_box_1: 2.0992, loss_cns_1: 0.6256, loss_yns_1: 0.1648, loss_cls_2: 1.0793, loss_box_2: 2.0619, loss_cns_2: 0.6433, loss_yns_2: 0.1676, loss_cls_3: 1.0900, loss_box_3: 2.0650, loss_cns_3: 0.6459, loss_yns_3: 0.1660, loss_cls_4: 1.1545, loss_box_4: 2.0650, loss_cns_4: 0.6472, loss_yns_4: 0.1657, loss_cls_5: 1.0846, loss_box_5: 2.0625, loss_cns_5: 0.6496, loss_yns_5: 0.1712, loss_cls_dn_0: 0.3626, loss_box_dn_0: 0.8331, loss_cls_dn_1: 0.2158, loss_box_dn_1: 0.8921, loss_cls_dn_2: 0.2310, loss_box_dn_2: 0.8692, loss_cls_dn_3: 0.2265, loss_box_dn_3: 0.8670, loss_cls_dn_4: 0.2260, loss_box_dn_4: 0.8723, loss_cls_dn_5: 0.2536, loss_box_dn_5: 0.8762, loss_dense_depth: 0.9375, loss: 31.1612, grad_norm: 74.7759 -2025-11-12 20:09:13,451 - mmdet - INFO - Iter [79/17500] lr: 1.312e-04, eta: 14:29:38, time: 1.530, data_time: 0.083, memory: 49163, loss_cls_0: 0.9728, loss_box_0: 1.8845, loss_cns_0: 0.6183, loss_yns_0: 0.1635, loss_cls_1: 1.0533, loss_box_1: 2.1230, loss_cns_1: 0.6300, loss_yns_1: 0.1662, loss_cls_2: 1.0800, loss_box_2: 2.1018, loss_cns_2: 0.6459, loss_yns_2: 0.1659, loss_cls_3: 1.0580, loss_box_3: 2.1238, loss_cns_3: 0.6442, loss_yns_3: 0.1693, loss_cls_4: 1.0536, loss_box_4: 2.1001, loss_cns_4: 0.6462, loss_yns_4: 0.1677, loss_cls_5: 1.0785, loss_box_5: 2.1044, loss_cns_5: 0.6438, loss_yns_5: 0.1680, loss_cls_dn_0: 0.3380, loss_box_dn_0: 0.8363, loss_cls_dn_1: 0.2152, loss_box_dn_1: 0.8637, loss_cls_dn_2: 0.2248, loss_box_dn_2: 0.8463, loss_cls_dn_3: 0.2224, loss_box_dn_3: 0.8492, loss_cls_dn_4: 0.2268, loss_box_dn_4: 0.8505, loss_cls_dn_5: 0.2433, loss_box_dn_5: 0.8606, loss_dense_depth: 0.9176, loss: 31.0578, grad_norm: 58.8276 -2025-11-12 20:09:14,976 - mmdet - INFO - Iter [80/17500] lr: 1.316e-04, eta: 14:24:14, time: 1.523, data_time: 0.081, memory: 49163, loss_cls_0: 0.9560, loss_box_0: 1.8852, loss_cns_0: 0.6229, loss_yns_0: 0.1623, loss_cls_1: 1.0281, loss_box_1: 2.1651, loss_cns_1: 0.6291, loss_yns_1: 0.1646, loss_cls_2: 1.0756, loss_box_2: 2.0935, loss_cns_2: 0.6423, loss_yns_2: 0.1670, loss_cls_3: 1.1124, loss_box_3: 2.1229, loss_cns_3: 0.6462, loss_yns_3: 0.1744, loss_cls_4: 1.0787, loss_box_4: 2.0915, loss_cns_4: 0.6466, loss_yns_4: 0.1664, loss_cls_5: 1.0975, loss_box_5: 2.0837, loss_cns_5: 0.6464, loss_yns_5: 0.1657, loss_cls_dn_0: 0.3387, loss_box_dn_0: 0.8296, loss_cls_dn_1: 0.2086, loss_box_dn_1: 0.8599, loss_cls_dn_2: 0.2181, loss_box_dn_2: 0.8343, loss_cls_dn_3: 0.2296, loss_box_dn_3: 0.8456, loss_cls_dn_4: 0.2381, loss_box_dn_4: 0.8553, loss_cls_dn_5: 0.2423, loss_box_dn_5: 0.8721, loss_dense_depth: 0.8889, loss: 31.0852, grad_norm: 57.5498 -2025-11-12 20:09:16,569 - mmdet - INFO - Iter [81/17500] lr: 1.320e-04, eta: 14:19:14, time: 1.594, data_time: 0.151, memory: 49163, loss_cls_0: 0.9787, loss_box_0: 1.9047, loss_cns_0: 0.6166, loss_yns_0: 0.1614, loss_cls_1: 1.0301, loss_box_1: 2.1511, loss_cns_1: 0.6222, loss_yns_1: 0.1632, loss_cls_2: 1.0908, loss_box_2: 2.1120, loss_cns_2: 0.6337, loss_yns_2: 0.1656, loss_cls_3: 1.1086, loss_box_3: 2.1295, loss_cns_3: 0.6372, loss_yns_3: 0.1697, loss_cls_4: 1.0896, loss_box_4: 2.1234, loss_cns_4: 0.6335, loss_yns_4: 0.1643, loss_cls_5: 1.1246, loss_box_5: 2.1226, loss_cns_5: 0.6355, loss_yns_5: 0.1643, loss_cls_dn_0: 0.3396, loss_box_dn_0: 0.8372, loss_cls_dn_1: 0.2045, loss_box_dn_1: 0.8998, loss_cls_dn_2: 0.2191, loss_box_dn_2: 0.8968, loss_cls_dn_3: 0.2279, loss_box_dn_3: 0.9108, loss_cls_dn_4: 0.2250, loss_box_dn_4: 0.9457, loss_cls_dn_5: 0.2283, loss_box_dn_5: 0.9687, loss_dense_depth: 0.9094, loss: 31.5453, grad_norm: 66.6203 -2025-11-12 20:09:18,118 - mmdet - INFO - Iter [82/17500] lr: 1.324e-04, eta: 14:14:12, time: 1.550, data_time: 0.074, memory: 49163, loss_cls_0: 1.0072, loss_box_0: 1.9079, loss_cns_0: 0.6190, loss_yns_0: 0.1575, loss_cls_1: 1.0206, loss_box_1: 2.1184, loss_cns_1: 0.6260, loss_yns_1: 0.1612, loss_cls_2: 1.0775, loss_box_2: 2.0762, loss_cns_2: 0.6336, loss_yns_2: 0.1612, loss_cls_3: 1.0809, loss_box_3: 2.0716, loss_cns_3: 0.6353, loss_yns_3: 0.1626, loss_cls_4: 1.0847, loss_box_4: 2.0864, loss_cns_4: 0.6321, loss_yns_4: 0.1613, loss_cls_5: 1.1621, loss_box_5: 2.0776, loss_cns_5: 0.6379, loss_yns_5: 0.1665, loss_cls_dn_0: 0.3304, loss_box_dn_0: 0.8488, loss_cls_dn_1: 0.2057, loss_box_dn_1: 0.9509, loss_cls_dn_2: 0.2181, loss_box_dn_2: 0.9613, loss_cls_dn_3: 0.2228, loss_box_dn_3: 0.9760, loss_cls_dn_4: 0.2187, loss_box_dn_4: 1.0240, loss_cls_dn_5: 0.2209, loss_box_dn_5: 1.0409, loss_dense_depth: 0.9458, loss: 31.6897, grad_norm: 64.4139 -2025-11-12 20:09:19,699 - mmdet - INFO - Iter [83/17500] lr: 1.328e-04, eta: 14:09:23, time: 1.580, data_time: 0.073, memory: 49163, loss_cls_0: 0.9896, loss_box_0: 1.8734, loss_cns_0: 0.6250, loss_yns_0: 0.1600, loss_cls_1: 1.0340, loss_box_1: 2.0461, loss_cns_1: 0.6308, loss_yns_1: 0.1618, loss_cls_2: 1.0584, loss_box_2: 1.9976, loss_cns_2: 0.6423, loss_yns_2: 0.1623, loss_cls_3: 1.0639, loss_box_3: 2.0036, loss_cns_3: 0.6429, loss_yns_3: 0.1632, loss_cls_4: 1.0681, loss_box_4: 2.0058, loss_cns_4: 0.6437, loss_yns_4: 0.1640, loss_cls_5: 1.0837, loss_box_5: 1.9775, loss_cns_5: 0.6454, loss_yns_5: 0.1633, loss_cls_dn_0: 0.3242, loss_box_dn_0: 0.8498, loss_cls_dn_1: 0.2065, loss_box_dn_1: 0.9004, loss_cls_dn_2: 0.2138, loss_box_dn_2: 0.9033, loss_cls_dn_3: 0.2157, loss_box_dn_3: 0.9265, loss_cls_dn_4: 0.2249, loss_box_dn_4: 0.9654, loss_cls_dn_5: 0.2350, loss_box_dn_5: 0.9673, loss_dense_depth: 0.9086, loss: 30.8480, grad_norm: 53.6553 -2025-11-12 20:09:21,274 - mmdet - INFO - Iter [84/17500] lr: 1.332e-04, eta: 14:04:39, time: 1.570, data_time: 0.098, memory: 49163, loss_cls_0: 0.9301, loss_box_0: 1.8642, loss_cns_0: 0.6246, loss_yns_0: 0.1594, loss_cls_1: 1.0062, loss_box_1: 2.0732, loss_cns_1: 0.6286, loss_yns_1: 0.1616, loss_cls_2: 1.0690, loss_box_2: 2.0093, loss_cns_2: 0.6442, loss_yns_2: 0.1609, loss_cls_3: 1.0769, loss_box_3: 2.0226, loss_cns_3: 0.6465, loss_yns_3: 0.1620, loss_cls_4: 1.0875, loss_box_4: 2.0031, loss_cns_4: 0.6493, loss_yns_4: 0.1620, loss_cls_5: 1.1105, loss_box_5: 1.9895, loss_cns_5: 0.6461, loss_yns_5: 0.1597, loss_cls_dn_0: 0.3151, loss_box_dn_0: 0.8363, loss_cls_dn_1: 0.2056, loss_box_dn_1: 0.8779, loss_cls_dn_2: 0.2145, loss_box_dn_2: 0.8720, loss_cls_dn_3: 0.2232, loss_box_dn_3: 0.9012, loss_cls_dn_4: 0.2446, loss_box_dn_4: 0.9251, loss_cls_dn_5: 0.2723, loss_box_dn_5: 0.9339, loss_dense_depth: 0.8726, loss: 30.7412, grad_norm: 67.6508 -2025-11-12 20:09:22,806 - mmdet - INFO - Iter [85/17500] lr: 1.336e-04, eta: 13:59:54, time: 1.535, data_time: 0.078, memory: 49163, loss_cls_0: 0.9687, loss_box_0: 1.8628, loss_cns_0: 0.6213, loss_yns_0: 0.1602, loss_cls_1: 1.0040, loss_box_1: 2.1236, loss_cns_1: 0.6150, loss_yns_1: 0.1624, loss_cls_2: 1.0667, loss_box_2: 2.0575, loss_cns_2: 0.6390, loss_yns_2: 0.1607, loss_cls_3: 1.0834, loss_box_3: 2.0586, loss_cns_3: 0.6411, loss_yns_3: 0.1647, loss_cls_4: 1.0962, loss_box_4: 2.0326, loss_cns_4: 0.6430, loss_yns_4: 0.1610, loss_cls_5: 1.1241, loss_box_5: 2.0710, loss_cns_5: 0.6378, loss_yns_5: 0.1612, loss_cls_dn_0: 0.3398, loss_box_dn_0: 0.8544, loss_cls_dn_1: 0.2099, loss_box_dn_1: 0.8877, loss_cls_dn_2: 0.2190, loss_box_dn_2: 0.8645, loss_cls_dn_3: 0.2269, loss_box_dn_3: 0.8766, loss_cls_dn_4: 0.2415, loss_box_dn_4: 0.8859, loss_cls_dn_5: 0.2800, loss_box_dn_5: 0.9158, loss_dense_depth: 0.9521, loss: 31.0704, grad_norm: 51.4632 -2025-11-12 20:09:24,337 - mmdet - INFO - Iter [86/17500] lr: 1.340e-04, eta: 13:55:16, time: 1.533, data_time: 0.074, memory: 49163, loss_cls_0: 0.9393, loss_box_0: 1.8680, loss_cns_0: 0.6202, loss_yns_0: 0.1619, loss_cls_1: 0.9993, loss_box_1: 2.0376, loss_cns_1: 0.6173, loss_yns_1: 0.1625, loss_cls_2: 1.0449, loss_box_2: 1.9877, loss_cns_2: 0.6412, loss_yns_2: 0.1614, loss_cls_3: 1.0352, loss_box_3: 1.9725, loss_cns_3: 0.6413, loss_yns_3: 0.1633, loss_cls_4: 1.0632, loss_box_4: 1.9435, loss_cns_4: 0.6466, loss_yns_4: 0.1617, loss_cls_5: 1.0903, loss_box_5: 1.9683, loss_cns_5: 0.6436, loss_yns_5: 0.1613, loss_cls_dn_0: 0.3184, loss_box_dn_0: 0.8496, loss_cls_dn_1: 0.2081, loss_box_dn_1: 0.8492, loss_cls_dn_2: 0.2193, loss_box_dn_2: 0.8316, loss_cls_dn_3: 0.2168, loss_box_dn_3: 0.8204, loss_cls_dn_4: 0.2220, loss_box_dn_4: 0.8206, loss_cls_dn_5: 0.2577, loss_box_dn_5: 0.8431, loss_dense_depth: 0.9542, loss: 30.1433, grad_norm: 57.5528 -2025-11-12 20:09:25,844 - mmdet - INFO - Iter [87/17500] lr: 1.344e-04, eta: 13:50:39, time: 1.506, data_time: 0.076, memory: 49163, loss_cls_0: 0.9458, loss_box_0: 1.8731, loss_cns_0: 0.6208, loss_yns_0: 0.1639, loss_cls_1: 0.9973, loss_box_1: 2.0788, loss_cns_1: 0.6276, loss_yns_1: 0.1630, loss_cls_2: 1.0468, loss_box_2: 2.0477, loss_cns_2: 0.6443, loss_yns_2: 0.1592, loss_cls_3: 1.0329, loss_box_3: 2.0733, loss_cns_3: 0.6426, loss_yns_3: 0.1602, loss_cls_4: 1.0504, loss_box_4: 2.0373, loss_cns_4: 0.6474, loss_yns_4: 0.1614, loss_cls_5: 1.0696, loss_box_5: 2.0383, loss_cns_5: 0.6477, loss_yns_5: 0.1601, loss_cls_dn_0: 0.3001, loss_box_dn_0: 0.8437, loss_cls_dn_1: 0.2074, loss_box_dn_1: 0.8222, loss_cls_dn_2: 0.2196, loss_box_dn_2: 0.8155, loss_cls_dn_3: 0.2172, loss_box_dn_3: 0.8074, loss_cls_dn_4: 0.2179, loss_box_dn_4: 0.8044, loss_cls_dn_5: 0.2445, loss_box_dn_5: 0.8051, loss_dense_depth: 0.9301, loss: 30.3246, grad_norm: 62.1981 -2025-11-12 20:09:27,351 - mmdet - INFO - Iter [88/17500] lr: 1.348e-04, eta: 13:46:07, time: 1.506, data_time: 0.072, memory: 49163, loss_cls_0: 0.9705, loss_box_0: 1.8445, loss_cns_0: 0.6164, loss_yns_0: 0.1623, loss_cls_1: 0.9938, loss_box_1: 2.0298, loss_cns_1: 0.6254, loss_yns_1: 0.1604, loss_cls_2: 1.0376, loss_box_2: 1.9902, loss_cns_2: 0.6413, loss_yns_2: 0.1594, loss_cls_3: 1.0413, loss_box_3: 2.0138, loss_cns_3: 0.6400, loss_yns_3: 0.1595, loss_cls_4: 1.0483, loss_box_4: 1.9963, loss_cns_4: 0.6425, loss_yns_4: 0.1603, loss_cls_5: 1.0630, loss_box_5: 2.0009, loss_cns_5: 0.6420, loss_yns_5: 0.1614, loss_cls_dn_0: 0.3000, loss_box_dn_0: 0.8432, loss_cls_dn_1: 0.2043, loss_box_dn_1: 0.8005, loss_cls_dn_2: 0.2124, loss_box_dn_2: 0.7932, loss_cls_dn_3: 0.2149, loss_box_dn_3: 0.7964, loss_cls_dn_4: 0.2192, loss_box_dn_4: 0.8029, loss_cls_dn_5: 0.2333, loss_box_dn_5: 0.8090, loss_dense_depth: 0.8711, loss: 29.9010, grad_norm: 57.8700 -2025-11-12 20:09:28,870 - mmdet - INFO - Iter [89/17500] lr: 1.352e-04, eta: 13:41:45, time: 1.520, data_time: 0.077, memory: 49163, loss_cls_0: 0.9302, loss_box_0: 1.8140, loss_cns_0: 0.6148, loss_yns_0: 0.1612, loss_cls_1: 0.9928, loss_box_1: 2.0098, loss_cns_1: 0.6223, loss_yns_1: 0.1593, loss_cls_2: 1.0246, loss_box_2: 1.9484, loss_cns_2: 0.6424, loss_yns_2: 0.1581, loss_cls_3: 1.0237, loss_box_3: 1.9457, loss_cns_3: 0.6444, loss_yns_3: 0.1582, loss_cls_4: 1.0357, loss_box_4: 1.9523, loss_cns_4: 0.6470, loss_yns_4: 0.1607, loss_cls_5: 1.0758, loss_box_5: 1.9648, loss_cns_5: 0.6467, loss_yns_5: 0.1625, loss_cls_dn_0: 0.3057, loss_box_dn_0: 0.8297, loss_cls_dn_1: 0.1998, loss_box_dn_1: 0.8134, loss_cls_dn_2: 0.2080, loss_box_dn_2: 0.8068, loss_cls_dn_3: 0.2092, loss_box_dn_3: 0.8167, loss_cls_dn_4: 0.2153, loss_box_dn_4: 0.8392, loss_cls_dn_5: 0.2270, loss_box_dn_5: 0.8663, loss_dense_depth: 0.8902, loss: 29.7225, grad_norm: 51.6057 -2025-11-12 20:09:30,400 - mmdet - INFO - Iter [90/17500] lr: 1.356e-04, eta: 13:37:30, time: 1.530, data_time: 0.076, memory: 49163, loss_cls_0: 0.9596, loss_box_0: 1.8187, loss_cns_0: 0.6138, loss_yns_0: 0.1578, loss_cls_1: 1.0083, loss_box_1: 2.0361, loss_cns_1: 0.6318, loss_yns_1: 0.1591, loss_cls_2: 1.0407, loss_box_2: 2.0204, loss_cns_2: 0.6430, loss_yns_2: 0.1580, loss_cls_3: 1.0396, loss_box_3: 2.0135, loss_cns_3: 0.6467, loss_yns_3: 0.1603, loss_cls_4: 1.0613, loss_box_4: 2.0447, loss_cns_4: 0.6471, loss_yns_4: 0.1588, loss_cls_5: 1.0727, loss_box_5: 2.0434, loss_cns_5: 0.6479, loss_yns_5: 0.1603, loss_cls_dn_0: 0.3262, loss_box_dn_0: 0.8426, loss_cls_dn_1: 0.2012, loss_box_dn_1: 0.8537, loss_cls_dn_2: 0.2131, loss_box_dn_2: 0.8575, loss_cls_dn_3: 0.2142, loss_box_dn_3: 0.8716, loss_cls_dn_4: 0.2183, loss_box_dn_4: 0.9100, loss_cls_dn_5: 0.2381, loss_box_dn_5: 0.9406, loss_dense_depth: 0.8871, loss: 30.5181, grad_norm: 67.6334 -2025-11-12 20:09:31,934 - mmdet - INFO - Iter [91/17500] lr: 1.360e-04, eta: 13:33:22, time: 1.534, data_time: 0.074, memory: 49163, loss_cls_0: 0.9526, loss_box_0: 1.7954, loss_cns_0: 0.6090, loss_yns_0: 0.1552, loss_cls_1: 0.9870, loss_box_1: 2.0691, loss_cns_1: 0.6268, loss_yns_1: 0.1558, loss_cls_2: 1.0258, loss_box_2: 2.0571, loss_cns_2: 0.6383, loss_yns_2: 0.1568, loss_cls_3: 1.0319, loss_box_3: 2.0433, loss_cns_3: 0.6454, loss_yns_3: 0.1601, loss_cls_4: 1.0541, loss_box_4: 2.0752, loss_cns_4: 0.6446, loss_yns_4: 0.1546, loss_cls_5: 1.1126, loss_box_5: 2.0592, loss_cns_5: 0.6465, loss_yns_5: 0.1549, loss_cls_dn_0: 0.3086, loss_box_dn_0: 0.8268, loss_cls_dn_1: 0.2080, loss_box_dn_1: 0.8809, loss_cls_dn_2: 0.2209, loss_box_dn_2: 0.8715, loss_cls_dn_3: 0.2162, loss_box_dn_3: 0.8826, loss_cls_dn_4: 0.2194, loss_box_dn_4: 0.9256, loss_cls_dn_5: 0.2381, loss_box_dn_5: 0.9509, loss_dense_depth: 0.8968, loss: 30.6575, grad_norm: 69.8527 -2025-11-12 20:09:33,474 - mmdet - INFO - Iter [92/17500] lr: 1.364e-04, eta: 13:29:20, time: 1.540, data_time: 0.074, memory: 49163, loss_cls_0: 0.9603, loss_box_0: 1.8061, loss_cns_0: 0.6157, loss_yns_0: 0.1545, loss_cls_1: 0.9935, loss_box_1: 2.0941, loss_cns_1: 0.6254, loss_yns_1: 0.1560, loss_cls_2: 1.0158, loss_box_2: 2.0376, loss_cns_2: 0.6409, loss_yns_2: 0.1562, loss_cls_3: 1.0506, loss_box_3: 2.0322, loss_cns_3: 0.6475, loss_yns_3: 0.1606, loss_cls_4: 1.0446, loss_box_4: 2.0454, loss_cns_4: 0.6465, loss_yns_4: 0.1573, loss_cls_5: 1.1325, loss_box_5: 2.0123, loss_cns_5: 0.6495, loss_yns_5: 0.1546, loss_cls_dn_0: 0.2892, loss_box_dn_0: 0.8200, loss_cls_dn_1: 0.2018, loss_box_dn_1: 0.8491, loss_cls_dn_2: 0.2094, loss_box_dn_2: 0.8164, loss_cls_dn_3: 0.2056, loss_box_dn_3: 0.8206, loss_cls_dn_4: 0.2084, loss_box_dn_4: 0.8443, loss_cls_dn_5: 0.2249, loss_box_dn_5: 0.8577, loss_dense_depth: 0.9016, loss: 30.2387, grad_norm: 43.8536 -2025-11-12 20:09:35,003 - mmdet - INFO - Iter [93/17500] lr: 1.368e-04, eta: 13:25:22, time: 1.530, data_time: 0.076, memory: 49163, loss_cls_0: 0.9444, loss_box_0: 1.8053, loss_cns_0: 0.6208, loss_yns_0: 0.1529, loss_cls_1: 0.9865, loss_box_1: 2.0355, loss_cns_1: 0.6198, loss_yns_1: 0.1554, loss_cls_2: 1.0255, loss_box_2: 1.9526, loss_cns_2: 0.6429, loss_yns_2: 0.1578, loss_cls_3: 1.0350, loss_box_3: 1.9818, loss_cns_3: 0.6454, loss_yns_3: 0.1574, loss_cls_4: 1.0414, loss_box_4: 1.9768, loss_cns_4: 0.6454, loss_yns_4: 0.1567, loss_cls_5: 1.1017, loss_box_5: 1.9743, loss_cns_5: 0.6459, loss_yns_5: 0.1572, loss_cls_dn_0: 0.2798, loss_box_dn_0: 0.8203, loss_cls_dn_1: 0.1983, loss_box_dn_1: 0.8342, loss_cls_dn_2: 0.2064, loss_box_dn_2: 0.7956, loss_cls_dn_3: 0.2045, loss_box_dn_3: 0.8025, loss_cls_dn_4: 0.2117, loss_box_dn_4: 0.8030, loss_cls_dn_5: 0.2275, loss_box_dn_5: 0.8106, loss_dense_depth: 0.8427, loss: 29.6557, grad_norm: 57.4118 -2025-11-12 20:09:36,530 - mmdet - INFO - Iter [94/17500] lr: 1.372e-04, eta: 13:21:27, time: 1.526, data_time: 0.078, memory: 49163, loss_cls_0: 0.9483, loss_box_0: 1.8325, loss_cns_0: 0.6186, loss_yns_0: 0.1528, loss_cls_1: 1.0025, loss_box_1: 2.0596, loss_cns_1: 0.6206, loss_yns_1: 0.1577, loss_cls_2: 1.0095, loss_box_2: 1.9972, loss_cns_2: 0.6380, loss_yns_2: 0.1560, loss_cls_3: 1.0275, loss_box_3: 2.0280, loss_cns_3: 0.6408, loss_yns_3: 0.1564, loss_cls_4: 1.0293, loss_box_4: 2.0286, loss_cns_4: 0.6406, loss_yns_4: 0.1591, loss_cls_5: 1.0415, loss_box_5: 2.0331, loss_cns_5: 0.6465, loss_yns_5: 0.1560, loss_cls_dn_0: 0.2904, loss_box_dn_0: 0.8174, loss_cls_dn_1: 0.2072, loss_box_dn_1: 0.8254, loss_cls_dn_2: 0.2115, loss_box_dn_2: 0.8007, loss_cls_dn_3: 0.2141, loss_box_dn_3: 0.8151, loss_cls_dn_4: 0.2194, loss_box_dn_4: 0.8166, loss_cls_dn_5: 0.2315, loss_box_dn_5: 0.8176, loss_dense_depth: 0.8943, loss: 29.9417, grad_norm: 64.9000 -2025-11-12 20:09:38,103 - mmdet - INFO - Iter [95/17500] lr: 1.376e-04, eta: 13:17:47, time: 1.573, data_time: 0.077, memory: 49163, loss_cls_0: 0.9573, loss_box_0: 1.8383, loss_cns_0: 0.6201, loss_yns_0: 0.1538, loss_cls_1: 1.0150, loss_box_1: 1.9959, loss_cns_1: 0.6309, loss_yns_1: 0.1592, loss_cls_2: 1.0251, loss_box_2: 1.9434, loss_cns_2: 0.6419, loss_yns_2: 0.1587, loss_cls_3: 1.0615, loss_box_3: 1.9517, loss_cns_3: 0.6475, loss_yns_3: 0.1587, loss_cls_4: 1.0796, loss_box_4: 1.9740, loss_cns_4: 0.6467, loss_yns_4: 0.1602, loss_cls_5: 1.1078, loss_box_5: 1.9384, loss_cns_5: 0.6530, loss_yns_5: 0.1551, loss_cls_dn_0: 0.2946, loss_box_dn_0: 0.8160, loss_cls_dn_1: 0.2090, loss_box_dn_1: 0.7980, loss_cls_dn_2: 0.2078, loss_box_dn_2: 0.7777, loss_cls_dn_3: 0.2119, loss_box_dn_3: 0.7924, loss_cls_dn_4: 0.2152, loss_box_dn_4: 0.8058, loss_cls_dn_5: 0.2219, loss_box_dn_5: 0.7902, loss_dense_depth: 0.8817, loss: 29.6958, grad_norm: 62.3698 -2025-11-12 20:09:39,637 - mmdet - INFO - Iter [96/17500] lr: 1.380e-04, eta: 13:14:03, time: 1.534, data_time: 0.079, memory: 49163, loss_cls_0: 0.9728, loss_box_0: 1.8956, loss_cns_0: 0.6153, loss_yns_0: 0.1555, loss_cls_1: 1.0303, loss_box_1: 2.0163, loss_cns_1: 0.6278, loss_yns_1: 0.1579, loss_cls_2: 1.0531, loss_box_2: 1.9822, loss_cns_2: 0.6351, loss_yns_2: 0.1607, loss_cls_3: 1.1133, loss_box_3: 1.9501, loss_cns_3: 0.6433, loss_yns_3: 0.1612, loss_cls_4: 1.0831, loss_box_4: 1.9696, loss_cns_4: 0.6437, loss_yns_4: 0.1594, loss_cls_5: 1.1354, loss_box_5: 1.9526, loss_cns_5: 0.6468, loss_yns_5: 0.1590, loss_cls_dn_0: 0.2966, loss_box_dn_0: 0.8283, loss_cls_dn_1: 0.2049, loss_box_dn_1: 0.7977, loss_cls_dn_2: 0.2064, loss_box_dn_2: 0.7846, loss_cls_dn_3: 0.2134, loss_box_dn_3: 0.7939, loss_cls_dn_4: 0.2152, loss_box_dn_4: 0.8085, loss_cls_dn_5: 0.2246, loss_box_dn_5: 0.8070, loss_dense_depth: 0.8784, loss: 29.9795, grad_norm: 38.1399 -2025-11-12 20:09:41,177 - mmdet - INFO - Iter [97/17500] lr: 1.384e-04, eta: 13:10:26, time: 1.541, data_time: 0.076, memory: 49163, loss_cls_0: 0.9747, loss_box_0: 1.8513, loss_cns_0: 0.6193, loss_yns_0: 0.1551, loss_cls_1: 1.0045, loss_box_1: 2.0419, loss_cns_1: 0.6202, loss_yns_1: 0.1575, loss_cls_2: 1.0373, loss_box_2: 2.0084, loss_cns_2: 0.6341, loss_yns_2: 0.1577, loss_cls_3: 1.0881, loss_box_3: 1.9587, loss_cns_3: 0.6431, loss_yns_3: 0.1585, loss_cls_4: 1.0890, loss_box_4: 1.9446, loss_cns_4: 0.6447, loss_yns_4: 0.1592, loss_cls_5: 1.1469, loss_box_5: 1.9480, loss_cns_5: 0.6444, loss_yns_5: 0.1600, loss_cls_dn_0: 0.2812, loss_box_dn_0: 0.8225, loss_cls_dn_1: 0.2004, loss_box_dn_1: 0.8065, loss_cls_dn_2: 0.2034, loss_box_dn_2: 0.7931, loss_cls_dn_3: 0.2096, loss_box_dn_3: 0.7901, loss_cls_dn_4: 0.2107, loss_box_dn_4: 0.7988, loss_cls_dn_5: 0.2270, loss_box_dn_5: 0.8154, loss_dense_depth: 0.8814, loss: 29.8873, grad_norm: 46.1717 -2025-11-12 20:09:42,699 - mmdet - INFO - Iter [98/17500] lr: 1.388e-04, eta: 13:06:49, time: 1.521, data_time: 0.074, memory: 49163, loss_cls_0: 0.9840, loss_box_0: 1.8167, loss_cns_0: 0.6254, loss_yns_0: 0.1542, loss_cls_1: 1.0143, loss_box_1: 1.9675, loss_cns_1: 0.6271, loss_yns_1: 0.1583, loss_cls_2: 1.0703, loss_box_2: 1.9404, loss_cns_2: 0.6413, loss_yns_2: 0.1582, loss_cls_3: 1.0338, loss_box_3: 1.9081, loss_cns_3: 0.6445, loss_yns_3: 0.1568, loss_cls_4: 1.0502, loss_box_4: 1.8924, loss_cns_4: 0.6475, loss_yns_4: 0.1624, loss_cls_5: 1.1062, loss_box_5: 1.9144, loss_cns_5: 0.6487, loss_yns_5: 0.1576, loss_cls_dn_0: 0.2709, loss_box_dn_0: 0.8233, loss_cls_dn_1: 0.1973, loss_box_dn_1: 0.8461, loss_cls_dn_2: 0.2048, loss_box_dn_2: 0.8287, loss_cls_dn_3: 0.2066, loss_box_dn_3: 0.8254, loss_cls_dn_4: 0.2120, loss_box_dn_4: 0.8368, loss_cls_dn_5: 0.2306, loss_box_dn_5: 0.8559, loss_dense_depth: 0.8557, loss: 29.6746, grad_norm: 42.1515 -2025-11-12 20:09:44,228 - mmdet - INFO - Iter [99/17500] lr: 1.392e-04, eta: 13:03:19, time: 1.529, data_time: 0.081, memory: 49163, loss_cls_0: 0.9707, loss_box_0: 1.8212, loss_cns_0: 0.6244, loss_yns_0: 0.1550, loss_cls_1: 1.0166, loss_box_1: 2.0186, loss_cns_1: 0.6251, loss_yns_1: 0.1588, loss_cls_2: 1.0685, loss_box_2: 1.9486, loss_cns_2: 0.6404, loss_yns_2: 0.1577, loss_cls_3: 1.0652, loss_box_3: 1.9446, loss_cns_3: 0.6424, loss_yns_3: 0.1580, loss_cls_4: 1.0748, loss_box_4: 1.9281, loss_cns_4: 0.6461, loss_yns_4: 0.1642, loss_cls_5: 1.0764, loss_box_5: 1.9408, loss_cns_5: 0.6470, loss_yns_5: 0.1561, loss_cls_dn_0: 0.2783, loss_box_dn_0: 0.8190, loss_cls_dn_1: 0.1932, loss_box_dn_1: 0.8423, loss_cls_dn_2: 0.1999, loss_box_dn_2: 0.8176, loss_cls_dn_3: 0.2008, loss_box_dn_3: 0.8228, loss_cls_dn_4: 0.2048, loss_box_dn_4: 0.8307, loss_cls_dn_5: 0.2195, loss_box_dn_5: 0.8358, loss_dense_depth: 0.8726, loss: 29.7867, grad_norm: 42.8888 -2025-11-12 20:09:45,761 - mmdet - INFO - Iter [100/17500] lr: 1.396e-04, eta: 12:59:53, time: 1.534, data_time: 0.078, memory: 49163, loss_cls_0: 0.9421, loss_box_0: 1.8364, loss_cns_0: 0.6214, loss_yns_0: 0.1557, loss_cls_1: 1.0011, loss_box_1: 2.0468, loss_cns_1: 0.6249, loss_yns_1: 0.1590, loss_cls_2: 1.0292, loss_box_2: 1.9658, loss_cns_2: 0.6380, loss_yns_2: 0.1583, loss_cls_3: 1.0405, loss_box_3: 1.9577, loss_cns_3: 0.6406, loss_yns_3: 0.1585, loss_cls_4: 1.0523, loss_box_4: 1.9393, loss_cns_4: 0.6431, loss_yns_4: 0.1576, loss_cls_5: 1.0676, loss_box_5: 1.9402, loss_cns_5: 0.6431, loss_yns_5: 0.1568, loss_cls_dn_0: 0.2758, loss_box_dn_0: 0.8266, loss_cls_dn_1: 0.1917, loss_box_dn_1: 0.8456, loss_cls_dn_2: 0.1978, loss_box_dn_2: 0.8133, loss_cls_dn_3: 0.1971, loss_box_dn_3: 0.8190, loss_cls_dn_4: 0.2039, loss_box_dn_4: 0.8187, loss_cls_dn_5: 0.2197, loss_box_dn_5: 0.8270, loss_dense_depth: 0.8480, loss: 29.6605, grad_norm: 31.2682 -2025-11-12 20:09:47,362 - mmdet - INFO - Iter [101/17500] lr: 1.400e-04, eta: 12:56:43, time: 1.601, data_time: 0.156, memory: 49163, loss_cls_0: 0.9741, loss_box_0: 1.8128, loss_cns_0: 0.6198, loss_yns_0: 0.1563, loss_cls_1: 1.0117, loss_box_1: 2.0007, loss_cns_1: 0.6278, loss_yns_1: 0.1583, loss_cls_2: 1.0428, loss_box_2: 1.9407, loss_cns_2: 0.6383, loss_yns_2: 0.1582, loss_cls_3: 1.0434, loss_box_3: 1.9191, loss_cns_3: 0.6432, loss_yns_3: 0.1582, loss_cls_4: 1.0437, loss_box_4: 1.9030, loss_cns_4: 0.6464, loss_yns_4: 0.1614, loss_cls_5: 1.0615, loss_box_5: 1.9232, loss_cns_5: 0.6441, loss_yns_5: 0.1573, loss_cls_dn_0: 0.2702, loss_box_dn_0: 0.8216, loss_cls_dn_1: 0.1949, loss_box_dn_1: 0.8514, loss_cls_dn_2: 0.2015, loss_box_dn_2: 0.8192, loss_cls_dn_3: 0.1983, loss_box_dn_3: 0.8136, loss_cls_dn_4: 0.2035, loss_box_dn_4: 0.8116, loss_cls_dn_5: 0.2137, loss_box_dn_5: 0.8338, loss_dense_depth: 0.9000, loss: 29.5793, grad_norm: 42.8565 -2025-11-12 20:09:48,913 - mmdet - INFO - Iter [102/17500] lr: 1.404e-04, eta: 12:53:28, time: 1.551, data_time: 0.073, memory: 49163, loss_cls_0: 0.9546, loss_box_0: 1.8300, loss_cns_0: 0.6125, loss_yns_0: 0.1569, loss_cls_1: 1.0227, loss_box_1: 1.9930, loss_cns_1: 0.6356, loss_yns_1: 0.1609, loss_cls_2: 1.0771, loss_box_2: 1.9595, loss_cns_2: 0.6420, loss_yns_2: 0.1611, loss_cls_3: 1.0467, loss_box_3: 1.9258, loss_cns_3: 0.6476, loss_yns_3: 0.1606, loss_cls_4: 1.0664, loss_box_4: 1.9241, loss_cns_4: 0.6477, loss_yns_4: 0.1691, loss_cls_5: 1.0809, loss_box_5: 1.9353, loss_cns_5: 0.6461, loss_yns_5: 0.1598, loss_cls_dn_0: 0.2757, loss_box_dn_0: 0.8294, loss_cls_dn_1: 0.1987, loss_box_dn_1: 0.8139, loss_cls_dn_2: 0.2058, loss_box_dn_2: 0.7935, loss_cls_dn_3: 0.2017, loss_box_dn_3: 0.7841, loss_cls_dn_4: 0.2058, loss_box_dn_4: 0.7935, loss_cls_dn_5: 0.2139, loss_box_dn_5: 0.8215, loss_dense_depth: 0.8527, loss: 29.6062, grad_norm: 44.2340 -2025-11-12 20:09:50,505 - mmdet - INFO - Iter [103/17500] lr: 1.408e-04, eta: 12:50:23, time: 1.591, data_time: 0.075, memory: 49163, loss_cls_0: 0.9562, loss_box_0: 1.7957, loss_cns_0: 0.6064, loss_yns_0: 0.1561, loss_cls_1: 1.0225, loss_box_1: 1.9489, loss_cns_1: 0.6425, loss_yns_1: 0.1630, loss_cls_2: 1.0536, loss_box_2: 1.8996, loss_cns_2: 0.6487, loss_yns_2: 0.1631, loss_cls_3: 1.0550, loss_box_3: 1.8896, loss_cns_3: 0.6537, loss_yns_3: 0.1626, loss_cls_4: 1.0467, loss_box_4: 1.9086, loss_cns_4: 0.6527, loss_yns_4: 0.1697, loss_cls_5: 1.0522, loss_box_5: 1.9002, loss_cns_5: 0.6514, loss_yns_5: 0.1615, loss_cls_dn_0: 0.2698, loss_box_dn_0: 0.8205, loss_cls_dn_1: 0.1850, loss_box_dn_1: 0.8290, loss_cls_dn_2: 0.1962, loss_box_dn_2: 0.8100, loss_cls_dn_3: 0.2011, loss_box_dn_3: 0.8092, loss_cls_dn_4: 0.2071, loss_box_dn_4: 0.8246, loss_cls_dn_5: 0.2178, loss_box_dn_5: 0.8450, loss_dense_depth: 0.8541, loss: 29.4298, grad_norm: 41.6286 -2025-11-12 20:09:52,089 - mmdet - INFO - Iter [104/17500] lr: 1.412e-04, eta: 12:47:21, time: 1.583, data_time: 0.097, memory: 49163, loss_cls_0: 0.9408, loss_box_0: 1.8455, loss_cns_0: 0.6137, loss_yns_0: 0.1560, loss_cls_1: 1.0052, loss_box_1: 1.9504, loss_cns_1: 0.6369, loss_yns_1: 0.1597, loss_cls_2: 1.0454, loss_box_2: 1.9135, loss_cns_2: 0.6508, loss_yns_2: 0.1594, loss_cls_3: 1.0517, loss_box_3: 1.9144, loss_cns_3: 0.6534, loss_yns_3: 0.1592, loss_cls_4: 1.0518, loss_box_4: 1.9105, loss_cns_4: 0.6544, loss_yns_4: 0.1602, loss_cls_5: 1.0680, loss_box_5: 1.9135, loss_cns_5: 0.6559, loss_yns_5: 0.1606, loss_cls_dn_0: 0.2593, loss_box_dn_0: 0.8274, loss_cls_dn_1: 0.1783, loss_box_dn_1: 0.8361, loss_cls_dn_2: 0.1890, loss_box_dn_2: 0.8298, loss_cls_dn_3: 0.1898, loss_box_dn_3: 0.8363, loss_cls_dn_4: 0.1966, loss_box_dn_4: 0.8441, loss_cls_dn_5: 0.2086, loss_box_dn_5: 0.8694, loss_dense_depth: 0.8511, loss: 29.5465, grad_norm: 43.0032 -2025-11-12 20:09:53,640 - mmdet - INFO - Iter [105/17500] lr: 1.416e-04, eta: 12:44:17, time: 1.551, data_time: 0.073, memory: 49163, loss_cls_0: 0.9689, loss_box_0: 1.7925, loss_cns_0: 0.6240, loss_yns_0: 0.1567, loss_cls_1: 0.9890, loss_box_1: 1.9137, loss_cns_1: 0.6399, loss_yns_1: 0.1599, loss_cls_2: 1.0221, loss_box_2: 1.8789, loss_cns_2: 0.6560, loss_yns_2: 0.1590, loss_cls_3: 1.0207, loss_box_3: 1.8539, loss_cns_3: 0.6551, loss_yns_3: 0.1590, loss_cls_4: 1.0208, loss_box_4: 1.8613, loss_cns_4: 0.6574, loss_yns_4: 0.1670, loss_cls_5: 1.0359, loss_box_5: 1.8731, loss_cns_5: 0.6574, loss_yns_5: 0.1618, loss_cls_dn_0: 0.2484, loss_box_dn_0: 0.8163, loss_cls_dn_1: 0.1798, loss_box_dn_1: 0.8593, loss_cls_dn_2: 0.1853, loss_box_dn_2: 0.8534, loss_cls_dn_3: 0.1847, loss_box_dn_3: 0.8492, loss_cls_dn_4: 0.1875, loss_box_dn_4: 0.8577, loss_cls_dn_5: 0.1947, loss_box_dn_5: 0.8836, loss_dense_depth: 0.8164, loss: 29.2001, grad_norm: 38.3072 -2025-11-12 20:09:55,180 - mmdet - INFO - Iter [106/17500] lr: 1.420e-04, eta: 12:41:14, time: 1.541, data_time: 0.077, memory: 49163, loss_cls_0: 0.9415, loss_box_0: 1.7285, loss_cns_0: 0.6059, loss_yns_0: 0.1517, loss_cls_1: 0.9764, loss_box_1: 1.9259, loss_cns_1: 0.6357, loss_yns_1: 0.1580, loss_cls_2: 1.0191, loss_box_2: 1.8590, loss_cns_2: 0.6526, loss_yns_2: 0.1595, loss_cls_3: 1.0273, loss_box_3: 1.8416, loss_cns_3: 0.6561, loss_yns_3: 0.1577, loss_cls_4: 1.0331, loss_box_4: 1.8750, loss_cns_4: 0.6583, loss_yns_4: 0.1701, loss_cls_5: 1.0293, loss_box_5: 1.8653, loss_cns_5: 0.6575, loss_yns_5: 0.1594, loss_cls_dn_0: 0.2585, loss_box_dn_0: 0.7999, loss_cls_dn_1: 0.1779, loss_box_dn_1: 0.8299, loss_cls_dn_2: 0.1834, loss_box_dn_2: 0.8137, loss_cls_dn_3: 0.1844, loss_box_dn_3: 0.8135, loss_cls_dn_4: 0.1887, loss_box_dn_4: 0.8318, loss_cls_dn_5: 0.1946, loss_box_dn_5: 0.8461, loss_dense_depth: 0.8289, loss: 28.8960, grad_norm: 39.4209 -2025-11-12 20:09:56,696 - mmdet - INFO - Iter [107/17500] lr: 1.424e-04, eta: 12:38:11, time: 1.516, data_time: 0.077, memory: 49163, loss_cls_0: 0.9559, loss_box_0: 1.6922, loss_cns_0: 0.5980, loss_yns_0: 0.1507, loss_cls_1: 0.9647, loss_box_1: 1.9582, loss_cns_1: 0.6310, loss_yns_1: 0.1554, loss_cls_2: 1.0420, loss_box_2: 1.8925, loss_cns_2: 0.6489, loss_yns_2: 0.1569, loss_cls_3: 1.0292, loss_box_3: 1.8717, loss_cns_3: 0.6560, loss_yns_3: 0.1553, loss_cls_4: 1.0345, loss_box_4: 1.8893, loss_cns_4: 0.6567, loss_yns_4: 0.1612, loss_cls_5: 1.0376, loss_box_5: 1.8885, loss_cns_5: 0.6555, loss_yns_5: 0.1570, loss_cls_dn_0: 0.2640, loss_box_dn_0: 0.8185, loss_cls_dn_1: 0.1785, loss_box_dn_1: 0.8009, loss_cls_dn_2: 0.1881, loss_box_dn_2: 0.7788, loss_cls_dn_3: 0.1858, loss_box_dn_3: 0.7805, loss_cls_dn_4: 0.1916, loss_box_dn_4: 0.7881, loss_cls_dn_5: 0.2043, loss_box_dn_5: 0.7986, loss_dense_depth: 0.8529, loss: 28.8695, grad_norm: 39.6909 -2025-11-12 20:09:58,211 - mmdet - INFO - Iter [108/17500] lr: 1.428e-04, eta: 12:35:11, time: 1.514, data_time: 0.075, memory: 49163, loss_cls_0: 0.9055, loss_box_0: 1.7688, loss_cns_0: 0.6177, loss_yns_0: 0.1571, loss_cls_1: 0.9854, loss_box_1: 1.9363, loss_cns_1: 0.6320, loss_yns_1: 0.1567, loss_cls_2: 1.0212, loss_box_2: 1.8658, loss_cns_2: 0.6491, loss_yns_2: 0.1577, loss_cls_3: 1.0215, loss_box_3: 1.8417, loss_cns_3: 0.6541, loss_yns_3: 0.1585, loss_cls_4: 1.0362, loss_box_4: 1.8306, loss_cns_4: 0.6545, loss_yns_4: 0.1592, loss_cls_5: 1.0384, loss_box_5: 1.8482, loss_cns_5: 0.6541, loss_yns_5: 0.1614, loss_cls_dn_0: 0.2506, loss_box_dn_0: 0.8175, loss_cls_dn_1: 0.1818, loss_box_dn_1: 0.7720, loss_cls_dn_2: 0.1886, loss_box_dn_2: 0.7529, loss_cls_dn_3: 0.1841, loss_box_dn_3: 0.7526, loss_cls_dn_4: 0.1901, loss_box_dn_4: 0.7546, loss_cls_dn_5: 0.2028, loss_box_dn_5: 0.7750, loss_dense_depth: 0.8159, loss: 28.5506, grad_norm: 38.1971 -2025-11-12 20:09:59,747 - mmdet - INFO - Iter [109/17500] lr: 1.432e-04, eta: 12:32:18, time: 1.537, data_time: 0.077, memory: 49163, loss_cls_0: 0.9568, loss_box_0: 1.8256, loss_cns_0: 0.6156, loss_yns_0: 0.1603, loss_cls_1: 0.9681, loss_box_1: 2.0092, loss_cns_1: 0.6173, loss_yns_1: 0.1574, loss_cls_2: 1.0219, loss_box_2: 1.9320, loss_cns_2: 0.6429, loss_yns_2: 0.1601, loss_cls_3: 1.0173, loss_box_3: 1.9277, loss_cns_3: 0.6444, loss_yns_3: 0.1598, loss_cls_4: 1.0300, loss_box_4: 1.9225, loss_cns_4: 0.6464, loss_yns_4: 0.1698, loss_cls_5: 1.0566, loss_box_5: 1.9410, loss_cns_5: 0.6502, loss_yns_5: 0.1650, loss_cls_dn_0: 0.2515, loss_box_dn_0: 0.8229, loss_cls_dn_1: 0.1784, loss_box_dn_1: 0.8055, loss_cls_dn_2: 0.1823, loss_box_dn_2: 0.7843, loss_cls_dn_3: 0.1812, loss_box_dn_3: 0.7867, loss_cls_dn_4: 0.1876, loss_box_dn_4: 0.8006, loss_cls_dn_5: 0.1993, loss_box_dn_5: 0.8238, loss_dense_depth: 0.8353, loss: 29.2373, grad_norm: 47.1765 -2025-11-12 20:10:01,260 - mmdet - INFO - Iter [110/17500] lr: 1.436e-04, eta: 12:29:25, time: 1.513, data_time: 0.076, memory: 49163, loss_cls_0: 0.9385, loss_box_0: 1.8305, loss_cns_0: 0.6170, loss_yns_0: 0.1595, loss_cls_1: 1.0069, loss_box_1: 1.9816, loss_cns_1: 0.6122, loss_yns_1: 0.1562, loss_cls_2: 1.0242, loss_box_2: 1.9508, loss_cns_2: 0.6440, loss_yns_2: 0.1595, loss_cls_3: 1.0213, loss_box_3: 1.9860, loss_cns_3: 0.6425, loss_yns_3: 0.1608, loss_cls_4: 1.0387, loss_box_4: 1.9619, loss_cns_4: 0.6450, loss_yns_4: 0.1743, loss_cls_5: 1.0596, loss_box_5: 1.9453, loss_cns_5: 0.6504, loss_yns_5: 0.1614, loss_cls_dn_0: 0.2540, loss_box_dn_0: 0.8111, loss_cls_dn_1: 0.1909, loss_box_dn_1: 0.8321, loss_cls_dn_2: 0.1853, loss_box_dn_2: 0.8205, loss_cls_dn_3: 0.1865, loss_box_dn_3: 0.8419, loss_cls_dn_4: 0.1955, loss_box_dn_4: 0.8594, loss_cls_dn_5: 0.2052, loss_box_dn_5: 0.8758, loss_dense_depth: 0.8410, loss: 29.6272, grad_norm: 45.6766 -2025-11-12 20:10:02,772 - mmdet - INFO - Iter [111/17500] lr: 1.440e-04, eta: 12:26:34, time: 1.512, data_time: 0.076, memory: 49163, loss_cls_0: 0.9246, loss_box_0: 1.8355, loss_cns_0: 0.6179, loss_yns_0: 0.1604, loss_cls_1: 0.9805, loss_box_1: 1.9667, loss_cns_1: 0.6238, loss_yns_1: 0.1553, loss_cls_2: 1.0385, loss_box_2: 1.9271, loss_cns_2: 0.6441, loss_yns_2: 0.1574, loss_cls_3: 1.0310, loss_box_3: 1.9582, loss_cns_3: 0.6490, loss_yns_3: 0.1567, loss_cls_4: 1.0309, loss_box_4: 1.9307, loss_cns_4: 0.6475, loss_yns_4: 0.1599, loss_cls_5: 1.0333, loss_box_5: 1.9292, loss_cns_5: 0.6467, loss_yns_5: 0.1576, loss_cls_dn_0: 0.2552, loss_box_dn_0: 0.8119, loss_cls_dn_1: 0.1851, loss_box_dn_1: 0.8428, loss_cls_dn_2: 0.1900, loss_box_dn_2: 0.8359, loss_cls_dn_3: 0.1902, loss_box_dn_3: 0.8627, loss_cls_dn_4: 0.1960, loss_box_dn_4: 0.8743, loss_cls_dn_5: 0.2057, loss_box_dn_5: 0.8911, loss_dense_depth: 0.8726, loss: 29.5761, grad_norm: 48.9672 -2025-11-12 20:10:04,294 - mmdet - INFO - Iter [112/17500] lr: 1.444e-04, eta: 12:23:48, time: 1.522, data_time: 0.073, memory: 49163, loss_cls_0: 0.9485, loss_box_0: 1.8583, loss_cns_0: 0.6127, loss_yns_0: 0.1613, loss_cls_1: 0.9944, loss_box_1: 2.0035, loss_cns_1: 0.6177, loss_yns_1: 0.1580, loss_cls_2: 1.0159, loss_box_2: 1.9501, loss_cns_2: 0.6381, loss_yns_2: 0.1580, loss_cls_3: 1.0237, loss_box_3: 1.9689, loss_cns_3: 0.6446, loss_yns_3: 0.1554, loss_cls_4: 1.0261, loss_box_4: 1.9326, loss_cns_4: 0.6449, loss_yns_4: 0.1568, loss_cls_5: 1.0402, loss_box_5: 1.9418, loss_cns_5: 0.6422, loss_yns_5: 0.1584, loss_cls_dn_0: 0.2658, loss_box_dn_0: 0.8170, loss_cls_dn_1: 0.1817, loss_box_dn_1: 0.8738, loss_cls_dn_2: 0.1910, loss_box_dn_2: 0.8491, loss_cls_dn_3: 0.1936, loss_box_dn_3: 0.8675, loss_cls_dn_4: 0.1941, loss_box_dn_4: 0.8687, loss_cls_dn_5: 0.2059, loss_box_dn_5: 0.8842, loss_dense_depth: 0.8840, loss: 29.7283, grad_norm: 50.2108 -2025-11-12 20:10:05,821 - mmdet - INFO - Iter [113/17500] lr: 1.448e-04, eta: 12:21:05, time: 1.527, data_time: 0.073, memory: 49163, loss_cls_0: 0.9146, loss_box_0: 1.8312, loss_cns_0: 0.6144, loss_yns_0: 0.1566, loss_cls_1: 0.9499, loss_box_1: 2.0115, loss_cns_1: 0.6180, loss_yns_1: 0.1574, loss_cls_2: 0.9957, loss_box_2: 1.9292, loss_cns_2: 0.6390, loss_yns_2: 0.1536, loss_cls_3: 1.0075, loss_box_3: 1.9181, loss_cns_3: 0.6449, loss_yns_3: 0.1531, loss_cls_4: 1.0029, loss_box_4: 1.9179, loss_cns_4: 0.6464, loss_yns_4: 0.1550, loss_cls_5: 1.0190, loss_box_5: 1.8887, loss_cns_5: 0.6450, loss_yns_5: 0.1576, loss_cls_dn_0: 0.2540, loss_box_dn_0: 0.8132, loss_cls_dn_1: 0.1765, loss_box_dn_1: 0.8572, loss_cls_dn_2: 0.1876, loss_box_dn_2: 0.8118, loss_cls_dn_3: 0.1872, loss_box_dn_3: 0.8107, loss_cls_dn_4: 0.1880, loss_box_dn_4: 0.8149, loss_cls_dn_5: 0.2022, loss_box_dn_5: 0.8099, loss_dense_depth: 0.8371, loss: 29.0776, grad_norm: 49.7811 -2025-11-12 20:10:07,350 - mmdet - INFO - Iter [114/17500] lr: 1.452e-04, eta: 12:18:26, time: 1.530, data_time: 0.077, memory: 49163, loss_cls_0: 0.9167, loss_box_0: 1.8415, loss_cns_0: 0.6187, loss_yns_0: 0.1582, loss_cls_1: 0.9759, loss_box_1: 1.9795, loss_cns_1: 0.6257, loss_yns_1: 0.1606, loss_cls_2: 1.0147, loss_box_2: 1.9018, loss_cns_2: 0.6466, loss_yns_2: 0.1591, loss_cls_3: 1.0098, loss_box_3: 1.8908, loss_cns_3: 0.6458, loss_yns_3: 0.1573, loss_cls_4: 1.0111, loss_box_4: 1.9094, loss_cns_4: 0.6489, loss_yns_4: 0.1627, loss_cls_5: 1.0289, loss_box_5: 1.8689, loss_cns_5: 0.6505, loss_yns_5: 0.1602, loss_cls_dn_0: 0.2582, loss_box_dn_0: 0.8037, loss_cls_dn_1: 0.1736, loss_box_dn_1: 0.8069, loss_cls_dn_2: 0.1822, loss_box_dn_2: 0.7651, loss_cls_dn_3: 0.1814, loss_box_dn_3: 0.7587, loss_cls_dn_4: 0.1867, loss_box_dn_4: 0.7733, loss_cls_dn_5: 0.2011, loss_box_dn_5: 0.7638, loss_dense_depth: 0.8366, loss: 28.8345, grad_norm: 34.9486 -2025-11-12 20:10:08,930 - mmdet - INFO - Iter [115/17500] lr: 1.456e-04, eta: 12:15:57, time: 1.579, data_time: 0.075, memory: 49163, loss_cls_0: 0.8950, loss_box_0: 1.8246, loss_cns_0: 0.6220, loss_yns_0: 0.1574, loss_cls_1: 0.9649, loss_box_1: 1.9764, loss_cns_1: 0.6306, loss_yns_1: 0.1570, loss_cls_2: 1.0008, loss_box_2: 1.9221, loss_cns_2: 0.6507, loss_yns_2: 0.1562, loss_cls_3: 1.0009, loss_box_3: 1.9106, loss_cns_3: 0.6469, loss_yns_3: 0.1549, loss_cls_4: 1.0131, loss_box_4: 1.9316, loss_cns_4: 0.6485, loss_yns_4: 0.1654, loss_cls_5: 1.0449, loss_box_5: 1.9142, loss_cns_5: 0.6510, loss_yns_5: 0.1576, loss_cls_dn_0: 0.2432, loss_box_dn_0: 0.8126, loss_cls_dn_1: 0.1734, loss_box_dn_1: 0.7789, loss_cls_dn_2: 0.1796, loss_box_dn_2: 0.7578, loss_cls_dn_3: 0.1773, loss_box_dn_3: 0.7532, loss_cls_dn_4: 0.1826, loss_box_dn_4: 0.7685, loss_cls_dn_5: 0.1940, loss_box_dn_5: 0.7735, loss_dense_depth: 0.8238, loss: 28.8159, grad_norm: 60.3746 -2025-11-12 20:10:10,456 - mmdet - INFO - Iter [116/17500] lr: 1.460e-04, eta: 12:13:22, time: 1.527, data_time: 0.076, memory: 49163, loss_cls_0: 0.9283, loss_box_0: 1.8413, loss_cns_0: 0.6146, loss_yns_0: 0.1584, loss_cls_1: 0.9679, loss_box_1: 1.9555, loss_cns_1: 0.6330, loss_yns_1: 0.1569, loss_cls_2: 0.9979, loss_box_2: 1.8827, loss_cns_2: 0.6459, loss_yns_2: 0.1565, loss_cls_3: 1.0003, loss_box_3: 1.8682, loss_cns_3: 0.6476, loss_yns_3: 0.1558, loss_cls_4: 1.0071, loss_box_4: 1.8640, loss_cns_4: 0.6489, loss_yns_4: 0.1592, loss_cls_5: 1.0122, loss_box_5: 1.8969, loss_cns_5: 0.6471, loss_yns_5: 0.1572, loss_cls_dn_0: 0.2457, loss_box_dn_0: 0.8141, loss_cls_dn_1: 0.1784, loss_box_dn_1: 0.7966, loss_cls_dn_2: 0.1839, loss_box_dn_2: 0.7864, loss_cls_dn_3: 0.1828, loss_box_dn_3: 0.7879, loss_cls_dn_4: 0.1909, loss_box_dn_4: 0.8035, loss_cls_dn_5: 0.2036, loss_box_dn_5: 0.8306, loss_dense_depth: 0.8489, loss: 28.8567, grad_norm: 46.3308 -2025-11-12 20:10:11,970 - mmdet - INFO - Iter [117/17500] lr: 1.464e-04, eta: 12:10:48, time: 1.513, data_time: 0.074, memory: 49163, loss_cls_0: 0.8986, loss_box_0: 1.8058, loss_cns_0: 0.6143, loss_yns_0: 0.1532, loss_cls_1: 0.9699, loss_box_1: 1.9552, loss_cns_1: 0.6336, loss_yns_1: 0.1585, loss_cls_2: 0.9927, loss_box_2: 1.8684, loss_cns_2: 0.6473, loss_yns_2: 0.1588, loss_cls_3: 0.9987, loss_box_3: 1.8432, loss_cns_3: 0.6480, loss_yns_3: 0.1590, loss_cls_4: 1.0063, loss_box_4: 1.8422, loss_cns_4: 0.6504, loss_yns_4: 0.1580, loss_cls_5: 1.0111, loss_box_5: 1.8510, loss_cns_5: 0.6473, loss_yns_5: 0.1566, loss_cls_dn_0: 0.2409, loss_box_dn_0: 0.8122, loss_cls_dn_1: 0.1788, loss_box_dn_1: 0.8189, loss_cls_dn_2: 0.1807, loss_box_dn_2: 0.8069, loss_cls_dn_3: 0.1834, loss_box_dn_3: 0.8071, loss_cls_dn_4: 0.1910, loss_box_dn_4: 0.8270, loss_cls_dn_5: 0.2034, loss_box_dn_5: 0.8499, loss_dense_depth: 0.8346, loss: 28.7628, grad_norm: 36.8538 -2025-11-12 20:10:13,509 - mmdet - INFO - Iter [118/17500] lr: 1.468e-04, eta: 12:08:21, time: 1.540, data_time: 0.076, memory: 49163, loss_cls_0: 0.9050, loss_box_0: 1.7673, loss_cns_0: 0.6193, loss_yns_0: 0.1533, loss_cls_1: 0.9749, loss_box_1: 1.9207, loss_cns_1: 0.6330, loss_yns_1: 0.1568, loss_cls_2: 1.0108, loss_box_2: 1.8502, loss_cns_2: 0.6511, loss_yns_2: 0.1571, loss_cls_3: 1.0150, loss_box_3: 1.8369, loss_cns_3: 0.6486, loss_yns_3: 0.1585, loss_cls_4: 1.0196, loss_box_4: 1.8521, loss_cns_4: 0.6487, loss_yns_4: 0.1579, loss_cls_5: 1.0059, loss_box_5: 1.8587, loss_cns_5: 0.6505, loss_yns_5: 0.1581, loss_cls_dn_0: 0.2454, loss_box_dn_0: 0.8012, loss_cls_dn_1: 0.1754, loss_box_dn_1: 0.8275, loss_cls_dn_2: 0.1778, loss_box_dn_2: 0.8154, loss_cls_dn_3: 0.1819, loss_box_dn_3: 0.8277, loss_cls_dn_4: 0.1882, loss_box_dn_4: 0.8545, loss_cls_dn_5: 0.1966, loss_box_dn_5: 0.8703, loss_dense_depth: 0.8705, loss: 28.8423, grad_norm: 49.3583 -2025-11-12 20:10:15,035 - mmdet - INFO - Iter [119/17500] lr: 1.472e-04, eta: 12:05:54, time: 1.525, data_time: 0.080, memory: 49163, loss_cls_0: 0.9067, loss_box_0: 1.7562, loss_cns_0: 0.6108, loss_yns_0: 0.1527, loss_cls_1: 0.9692, loss_box_1: 1.9229, loss_cns_1: 0.6351, loss_yns_1: 0.1561, loss_cls_2: 1.0371, loss_box_2: 1.8771, loss_cns_2: 0.6460, loss_yns_2: 0.1557, loss_cls_3: 1.0085, loss_box_3: 1.8708, loss_cns_3: 0.6462, loss_yns_3: 0.1577, loss_cls_4: 1.0240, loss_box_4: 1.8701, loss_cns_4: 0.6493, loss_yns_4: 0.1616, loss_cls_5: 1.0061, loss_box_5: 1.8861, loss_cns_5: 0.6524, loss_yns_5: 0.1600, loss_cls_dn_0: 0.2454, loss_box_dn_0: 0.8002, loss_cls_dn_1: 0.1727, loss_box_dn_1: 0.8336, loss_cls_dn_2: 0.1801, loss_box_dn_2: 0.8214, loss_cls_dn_3: 0.1808, loss_box_dn_3: 0.8311, loss_cls_dn_4: 0.1850, loss_box_dn_4: 0.8445, loss_cls_dn_5: 0.1889, loss_box_dn_5: 0.8522, loss_dense_depth: 0.8364, loss: 28.8905, grad_norm: 53.8683 -2025-11-12 20:10:16,547 - mmdet - INFO - Iter [120/17500] lr: 1.476e-04, eta: 12:03:28, time: 1.512, data_time: 0.080, memory: 49163, loss_cls_0: 0.9041, loss_box_0: 1.7715, loss_cns_0: 0.6072, loss_yns_0: 0.1530, loss_cls_1: 0.9631, loss_box_1: 1.9164, loss_cns_1: 0.6388, loss_yns_1: 0.1569, loss_cls_2: 0.9887, loss_box_2: 1.8624, loss_cns_2: 0.6459, loss_yns_2: 0.1559, loss_cls_3: 1.0179, loss_box_3: 1.8377, loss_cns_3: 0.6492, loss_yns_3: 0.1565, loss_cls_4: 1.0299, loss_box_4: 1.8236, loss_cns_4: 0.6527, loss_yns_4: 0.1654, loss_cls_5: 1.0122, loss_box_5: 1.8380, loss_cns_5: 0.6528, loss_yns_5: 0.1598, loss_cls_dn_0: 0.2480, loss_box_dn_0: 0.8130, loss_cls_dn_1: 0.1722, loss_box_dn_1: 0.7973, loss_cls_dn_2: 0.1755, loss_box_dn_2: 0.7740, loss_cls_dn_3: 0.1793, loss_box_dn_3: 0.7661, loss_cls_dn_4: 0.1813, loss_box_dn_4: 0.7689, loss_cls_dn_5: 0.1895, loss_box_dn_5: 0.7717, loss_dense_depth: 0.8523, loss: 28.4488, grad_norm: 40.7181 -2025-11-12 20:10:18,148 - mmdet - INFO - Iter [121/17500] lr: 1.480e-04, eta: 12:01:17, time: 1.602, data_time: 0.156, memory: 49163, loss_cls_0: 0.9185, loss_box_0: 1.7811, loss_cns_0: 0.6180, loss_yns_0: 0.1551, loss_cls_1: 0.9686, loss_box_1: 1.9959, loss_cns_1: 0.6341, loss_yns_1: 0.1584, loss_cls_2: 1.0246, loss_box_2: 1.9120, loss_cns_2: 0.6409, loss_yns_2: 0.1581, loss_cls_3: 1.0542, loss_box_3: 1.9063, loss_cns_3: 0.6466, loss_yns_3: 0.1566, loss_cls_4: 1.0474, loss_box_4: 1.9147, loss_cns_4: 0.6486, loss_yns_4: 0.1600, loss_cls_5: 1.0298, loss_box_5: 1.9085, loss_cns_5: 0.6467, loss_yns_5: 0.1568, loss_cls_dn_0: 0.2412, loss_box_dn_0: 0.8099, loss_cls_dn_1: 0.1676, loss_box_dn_1: 0.8024, loss_cls_dn_2: 0.1768, loss_box_dn_2: 0.7781, loss_cls_dn_3: 0.1835, loss_box_dn_3: 0.7768, loss_cls_dn_4: 0.1838, loss_box_dn_4: 0.7907, loss_cls_dn_5: 0.1878, loss_box_dn_5: 0.7958, loss_dense_depth: 0.8525, loss: 28.9886, grad_norm: 48.8396 -2025-11-12 20:10:19,688 - mmdet - INFO - Iter [122/17500] lr: 1.484e-04, eta: 11:58:59, time: 1.539, data_time: 0.078, memory: 49163, loss_cls_0: 0.9548, loss_box_0: 1.7735, loss_cns_0: 0.6235, loss_yns_0: 0.1562, loss_cls_1: 0.9605, loss_box_1: 1.9370, loss_cns_1: 0.6366, loss_yns_1: 0.1609, loss_cls_2: 1.0120, loss_box_2: 1.8754, loss_cns_2: 0.6458, loss_yns_2: 0.1645, loss_cls_3: 1.0191, loss_box_3: 1.8908, loss_cns_3: 0.6497, loss_yns_3: 0.1630, loss_cls_4: 1.0165, loss_box_4: 1.8969, loss_cns_4: 0.6528, loss_yns_4: 0.1623, loss_cls_5: 1.0177, loss_box_5: 1.8919, loss_cns_5: 0.6504, loss_yns_5: 0.1621, loss_cls_dn_0: 0.2343, loss_box_dn_0: 0.8005, loss_cls_dn_1: 0.1640, loss_box_dn_1: 0.7892, loss_cls_dn_2: 0.1733, loss_box_dn_2: 0.7726, loss_cls_dn_3: 0.1767, loss_box_dn_3: 0.7880, loss_cls_dn_4: 0.1818, loss_box_dn_4: 0.8068, loss_cls_dn_5: 0.1856, loss_box_dn_5: 0.8201, loss_dense_depth: 0.8304, loss: 28.7971, grad_norm: 52.4539 -2025-11-12 20:10:21,274 - mmdet - INFO - Iter [123/17500] lr: 1.488e-04, eta: 11:56:50, time: 1.586, data_time: 0.076, memory: 49163, loss_cls_0: 0.8991, loss_box_0: 1.7762, loss_cns_0: 0.6223, loss_yns_0: 0.1562, loss_cls_1: 0.9841, loss_box_1: 1.9392, loss_cns_1: 0.6365, loss_yns_1: 0.1598, loss_cls_2: 1.0204, loss_box_2: 1.8844, loss_cns_2: 0.6460, loss_yns_2: 0.1628, loss_cls_3: 1.0054, loss_box_3: 1.8850, loss_cns_3: 0.6459, loss_yns_3: 0.1618, loss_cls_4: 1.0315, loss_box_4: 1.8776, loss_cns_4: 0.6505, loss_yns_4: 0.1688, loss_cls_5: 1.0250, loss_box_5: 1.8765, loss_cns_5: 0.6475, loss_yns_5: 0.1667, loss_cls_dn_0: 0.2278, loss_box_dn_0: 0.7939, loss_cls_dn_1: 0.1629, loss_box_dn_1: 0.8131, loss_cls_dn_2: 0.1716, loss_box_dn_2: 0.8048, loss_cls_dn_3: 0.1731, loss_box_dn_3: 0.8263, loss_cls_dn_4: 0.1793, loss_box_dn_4: 0.8466, loss_cls_dn_5: 0.1857, loss_box_dn_5: 0.8640, loss_dense_depth: 0.8585, loss: 28.9366, grad_norm: 61.0067 -2025-11-12 20:10:22,845 - mmdet - INFO - Iter [124/17500] lr: 1.492e-04, eta: 11:54:40, time: 1.571, data_time: 0.103, memory: 49163, loss_cls_0: 0.9098, loss_box_0: 1.7791, loss_cns_0: 0.6209, loss_yns_0: 0.1554, loss_cls_1: 0.9816, loss_box_1: 1.9067, loss_cns_1: 0.6472, loss_yns_1: 0.1583, loss_cls_2: 1.0251, loss_box_2: 1.8600, loss_cns_2: 0.6517, loss_yns_2: 0.1605, loss_cls_3: 1.0283, loss_box_3: 1.8562, loss_cns_3: 0.6509, loss_yns_3: 0.1590, loss_cls_4: 1.0347, loss_box_4: 1.8391, loss_cns_4: 0.6533, loss_yns_4: 0.1655, loss_cls_5: 1.0246, loss_box_5: 1.8441, loss_cns_5: 0.6520, loss_yns_5: 0.1617, loss_cls_dn_0: 0.2347, loss_box_dn_0: 0.7988, loss_cls_dn_1: 0.1619, loss_box_dn_1: 0.8041, loss_cls_dn_2: 0.1675, loss_box_dn_2: 0.7896, loss_cls_dn_3: 0.1706, loss_box_dn_3: 0.8015, loss_cls_dn_4: 0.1740, loss_box_dn_4: 0.8141, loss_cls_dn_5: 0.1851, loss_box_dn_5: 0.8331, loss_dense_depth: 0.8117, loss: 28.6726, grad_norm: 46.2065 -2025-11-12 20:10:24,368 - mmdet - INFO - Iter [125/17500] lr: 1.496e-04, eta: 11:52:27, time: 1.522, data_time: 0.074, memory: 49163, loss_cls_0: 0.9636, loss_box_0: 1.7800, loss_cns_0: 0.6147, loss_yns_0: 0.1558, loss_cls_1: 1.0100, loss_box_1: 1.9343, loss_cns_1: 0.6474, loss_yns_1: 0.1615, loss_cls_2: 1.0340, loss_box_2: 1.8809, loss_cns_2: 0.6506, loss_yns_2: 0.1620, loss_cls_3: 1.0382, loss_box_3: 1.8849, loss_cns_3: 0.6493, loss_yns_3: 0.1614, loss_cls_4: 1.0420, loss_box_4: 1.8735, loss_cns_4: 0.6505, loss_yns_4: 0.1595, loss_cls_5: 1.0414, loss_box_5: 1.8927, loss_cns_5: 0.6484, loss_yns_5: 0.1573, loss_cls_dn_0: 0.2442, loss_box_dn_0: 0.7924, loss_cls_dn_1: 0.1645, loss_box_dn_1: 0.8071, loss_cls_dn_2: 0.1707, loss_box_dn_2: 0.7917, loss_cls_dn_3: 0.1733, loss_box_dn_3: 0.8015, loss_cls_dn_4: 0.1777, loss_box_dn_4: 0.8167, loss_cls_dn_5: 0.1915, loss_box_dn_5: 0.8383, loss_dense_depth: 0.8440, loss: 29.0074, grad_norm: 43.0024 -2025-11-12 20:10:25,892 - mmdet - INFO - Iter [126/17500] lr: 1.500e-04, eta: 11:50:15, time: 1.524, data_time: 0.075, memory: 49163, loss_cls_0: 0.9295, loss_box_0: 1.7865, loss_cns_0: 0.6198, loss_yns_0: 0.1552, loss_cls_1: 0.9864, loss_box_1: 1.9961, loss_cns_1: 0.6354, loss_yns_1: 0.1646, loss_cls_2: 1.0211, loss_box_2: 1.9302, loss_cns_2: 0.6499, loss_yns_2: 0.1640, loss_cls_3: 1.0276, loss_box_3: 1.9203, loss_cns_3: 0.6506, loss_yns_3: 0.1595, loss_cls_4: 1.0392, loss_box_4: 1.9180, loss_cns_4: 0.6512, loss_yns_4: 0.1583, loss_cls_5: 1.0357, loss_box_5: 1.9363, loss_cns_5: 0.6484, loss_yns_5: 0.1562, loss_cls_dn_0: 0.2395, loss_box_dn_0: 0.7966, loss_cls_dn_1: 0.1702, loss_box_dn_1: 0.8079, loss_cls_dn_2: 0.1783, loss_box_dn_2: 0.7906, loss_cls_dn_3: 0.1796, loss_box_dn_3: 0.7905, loss_cls_dn_4: 0.1837, loss_box_dn_4: 0.8059, loss_cls_dn_5: 0.1925, loss_box_dn_5: 0.8232, loss_dense_depth: 0.7861, loss: 29.0845, grad_norm: 45.7278 -2025-11-12 20:10:27,410 - mmdet - INFO - Iter [127/17500] lr: 1.504e-04, eta: 11:48:05, time: 1.517, data_time: 0.076, memory: 49163, loss_cls_0: 0.9198, loss_box_0: 1.7729, loss_cns_0: 0.6203, loss_yns_0: 0.1582, loss_cls_1: 0.9758, loss_box_1: 1.9486, loss_cns_1: 0.6278, loss_yns_1: 0.1612, loss_cls_2: 1.0388, loss_box_2: 1.8665, loss_cns_2: 0.6454, loss_yns_2: 0.1600, loss_cls_3: 1.0224, loss_box_3: 1.8545, loss_cns_3: 0.6468, loss_yns_3: 0.1578, loss_cls_4: 1.0300, loss_box_4: 1.8529, loss_cns_4: 0.6474, loss_yns_4: 0.1587, loss_cls_5: 1.0221, loss_box_5: 1.8644, loss_cns_5: 0.6424, loss_yns_5: 0.1572, loss_cls_dn_0: 0.2339, loss_box_dn_0: 0.7960, loss_cls_dn_1: 0.1670, loss_box_dn_1: 0.8057, loss_cls_dn_2: 0.1771, loss_box_dn_2: 0.7839, loss_cls_dn_3: 0.1758, loss_box_dn_3: 0.7805, loss_cls_dn_4: 0.1773, loss_box_dn_4: 0.7928, loss_cls_dn_5: 0.1879, loss_box_dn_5: 0.8042, loss_dense_depth: 0.8116, loss: 28.6457, grad_norm: 53.1452 -2025-11-12 20:10:28,922 - mmdet - INFO - Iter [128/17500] lr: 1.508e-04, eta: 11:45:56, time: 1.514, data_time: 0.075, memory: 49163, loss_cls_0: 0.9461, loss_box_0: 1.7770, loss_cns_0: 0.6249, loss_yns_0: 0.1631, loss_cls_1: 0.9692, loss_box_1: 1.8977, loss_cns_1: 0.6304, loss_yns_1: 0.1677, loss_cls_2: 1.0112, loss_box_2: 1.8364, loss_cns_2: 0.6463, loss_yns_2: 0.1656, loss_cls_3: 1.0181, loss_box_3: 1.8177, loss_cns_3: 0.6478, loss_yns_3: 0.1648, loss_cls_4: 1.0253, loss_box_4: 1.8147, loss_cns_4: 0.6530, loss_yns_4: 0.1608, loss_cls_5: 1.0221, loss_box_5: 1.8170, loss_cns_5: 0.6467, loss_yns_5: 0.1588, loss_cls_dn_0: 0.2343, loss_box_dn_0: 0.7933, loss_cls_dn_1: 0.1612, loss_box_dn_1: 0.7746, loss_cls_dn_2: 0.1685, loss_box_dn_2: 0.7670, loss_cls_dn_3: 0.1691, loss_box_dn_3: 0.7699, loss_cls_dn_4: 0.1736, loss_box_dn_4: 0.7837, loss_cls_dn_5: 0.1872, loss_box_dn_5: 0.7952, loss_dense_depth: 0.8046, loss: 28.3645, grad_norm: 45.3803 -2025-11-12 20:10:30,439 - mmdet - INFO - Iter [129/17500] lr: 1.512e-04, eta: 11:43:49, time: 1.516, data_time: 0.073, memory: 49163, loss_cls_0: 0.9076, loss_box_0: 1.7859, loss_cns_0: 0.6216, loss_yns_0: 0.1605, loss_cls_1: 0.9648, loss_box_1: 1.8910, loss_cns_1: 0.6362, loss_yns_1: 0.1645, loss_cls_2: 1.0043, loss_box_2: 1.8342, loss_cns_2: 0.6488, loss_yns_2: 0.1599, loss_cls_3: 1.0195, loss_box_3: 1.8087, loss_cns_3: 0.6479, loss_yns_3: 0.1586, loss_cls_4: 1.0174, loss_box_4: 1.8113, loss_cns_4: 0.6532, loss_yns_4: 0.1608, loss_cls_5: 1.0139, loss_box_5: 1.8448, loss_cns_5: 0.6472, loss_yns_5: 0.1596, loss_cls_dn_0: 0.2313, loss_box_dn_0: 0.7987, loss_cls_dn_1: 0.1630, loss_box_dn_1: 0.7780, loss_cls_dn_2: 0.1652, loss_box_dn_2: 0.7689, loss_cls_dn_3: 0.1684, loss_box_dn_3: 0.7685, loss_cls_dn_4: 0.1715, loss_box_dn_4: 0.7779, loss_cls_dn_5: 0.1848, loss_box_dn_5: 0.7986, loss_dense_depth: 0.7815, loss: 28.2786, grad_norm: 43.5310 -2025-11-12 20:10:31,957 - mmdet - INFO - Iter [130/17500] lr: 1.516e-04, eta: 11:41:44, time: 1.517, data_time: 0.075, memory: 49163, loss_cls_0: 0.9106, loss_box_0: 1.7851, loss_cns_0: 0.6175, loss_yns_0: 0.1539, loss_cls_1: 0.9694, loss_box_1: 1.8719, loss_cns_1: 0.6366, loss_yns_1: 0.1569, loss_cls_2: 1.0106, loss_box_2: 1.8016, loss_cns_2: 0.6500, loss_yns_2: 0.1555, loss_cls_3: 1.0277, loss_box_3: 1.7739, loss_cns_3: 0.6479, loss_yns_3: 0.1554, loss_cls_4: 1.0258, loss_box_4: 1.7939, loss_cns_4: 0.6498, loss_yns_4: 0.1598, loss_cls_5: 1.0178, loss_box_5: 1.8231, loss_cns_5: 0.6489, loss_yns_5: 0.1580, loss_cls_dn_0: 0.2346, loss_box_dn_0: 0.7969, loss_cls_dn_1: 0.1619, loss_box_dn_1: 0.7882, loss_cls_dn_2: 0.1657, loss_box_dn_2: 0.7751, loss_cls_dn_3: 0.1733, loss_box_dn_3: 0.7751, loss_cls_dn_4: 0.1724, loss_box_dn_4: 0.7889, loss_cls_dn_5: 0.1836, loss_box_dn_5: 0.8036, loss_dense_depth: 0.8295, loss: 28.2502, grad_norm: 40.3491 -2025-11-12 20:10:33,495 - mmdet - INFO - Iter [131/17500] lr: 1.520e-04, eta: 11:39:45, time: 1.538, data_time: 0.075, memory: 49163, loss_cls_0: 0.9144, loss_box_0: 1.7537, loss_cns_0: 0.6157, loss_yns_0: 0.1541, loss_cls_1: 0.9594, loss_box_1: 1.9142, loss_cns_1: 0.6363, loss_yns_1: 0.1563, loss_cls_2: 1.0005, loss_box_2: 1.8599, loss_cns_2: 0.6474, loss_yns_2: 0.1563, loss_cls_3: 1.0049, loss_box_3: 1.8646, loss_cns_3: 0.6481, loss_yns_3: 0.1565, loss_cls_4: 1.0180, loss_box_4: 1.8709, loss_cns_4: 0.6527, loss_yns_4: 0.1575, loss_cls_5: 1.0142, loss_box_5: 1.8505, loss_cns_5: 0.6499, loss_yns_5: 0.1575, loss_cls_dn_0: 0.2356, loss_box_dn_0: 0.7977, loss_cls_dn_1: 0.1619, loss_box_dn_1: 0.7822, loss_cls_dn_2: 0.1650, loss_box_dn_2: 0.7789, loss_cls_dn_3: 0.1697, loss_box_dn_3: 0.7893, loss_cls_dn_4: 0.1731, loss_box_dn_4: 0.8056, loss_cls_dn_5: 0.1824, loss_box_dn_5: 0.8081, loss_dense_depth: 0.8333, loss: 28.4962, grad_norm: 44.6403 -2025-11-12 20:10:35,017 - mmdet - INFO - Iter [132/17500] lr: 1.524e-04, eta: 11:37:45, time: 1.523, data_time: 0.077, memory: 49163, loss_cls_0: 0.9021, loss_box_0: 1.7628, loss_cns_0: 0.6155, loss_yns_0: 0.1538, loss_cls_1: 0.9525, loss_box_1: 1.9124, loss_cns_1: 0.6331, loss_yns_1: 0.1574, loss_cls_2: 0.9857, loss_box_2: 1.8573, loss_cns_2: 0.6447, loss_yns_2: 0.1539, loss_cls_3: 1.0182, loss_box_3: 1.8582, loss_cns_3: 0.6456, loss_yns_3: 0.1549, loss_cls_4: 1.0184, loss_box_4: 1.8471, loss_cns_4: 0.6487, loss_yns_4: 0.1537, loss_cls_5: 1.0079, loss_box_5: 1.8510, loss_cns_5: 0.6453, loss_yns_5: 0.1545, loss_cls_dn_0: 0.2388, loss_box_dn_0: 0.7864, loss_cls_dn_1: 0.1646, loss_box_dn_1: 0.7755, loss_cls_dn_2: 0.1671, loss_box_dn_2: 0.7763, loss_cls_dn_3: 0.1669, loss_box_dn_3: 0.7935, loss_cls_dn_4: 0.1754, loss_box_dn_4: 0.8082, loss_cls_dn_5: 0.1858, loss_box_dn_5: 0.8288, loss_dense_depth: 0.8209, loss: 28.4230, grad_norm: 46.3797 -2025-11-12 20:10:36,554 - mmdet - INFO - Iter [133/17500] lr: 1.528e-04, eta: 11:35:48, time: 1.537, data_time: 0.076, memory: 49163, loss_cls_0: 0.9199, loss_box_0: 1.7349, loss_cns_0: 0.6089, loss_yns_0: 0.1513, loss_cls_1: 0.9551, loss_box_1: 1.9438, loss_cns_1: 0.6269, loss_yns_1: 0.1589, loss_cls_2: 0.9872, loss_box_2: 1.8862, loss_cns_2: 0.6390, loss_yns_2: 0.1546, loss_cls_3: 1.0204, loss_box_3: 1.8810, loss_cns_3: 0.6401, loss_yns_3: 0.1543, loss_cls_4: 1.0108, loss_box_4: 1.8870, loss_cns_4: 0.6410, loss_yns_4: 0.1527, loss_cls_5: 1.0057, loss_box_5: 1.9206, loss_cns_5: 0.6345, loss_yns_5: 0.1563, loss_cls_dn_0: 0.2454, loss_box_dn_0: 0.7924, loss_cls_dn_1: 0.1662, loss_box_dn_1: 0.8010, loss_cls_dn_2: 0.1704, loss_box_dn_2: 0.7937, loss_cls_dn_3: 0.1676, loss_box_dn_3: 0.8081, loss_cls_dn_4: 0.1738, loss_box_dn_4: 0.8337, loss_cls_dn_5: 0.1930, loss_box_dn_5: 0.8675, loss_dense_depth: 0.8289, loss: 28.7127, grad_norm: 54.3624 -2025-11-12 20:10:38,065 - mmdet - INFO - Iter [134/17500] lr: 1.532e-04, eta: 11:33:50, time: 1.510, data_time: 0.078, memory: 49163, loss_cls_0: 0.8807, loss_box_0: 1.7092, loss_cns_0: 0.6199, loss_yns_0: 0.1498, loss_cls_1: 0.9457, loss_box_1: 1.8663, loss_cns_1: 0.6405, loss_yns_1: 0.1571, loss_cls_2: 0.9656, loss_box_2: 1.8260, loss_cns_2: 0.6468, loss_yns_2: 0.1529, loss_cls_3: 0.9844, loss_box_3: 1.8129, loss_cns_3: 0.6538, loss_yns_3: 0.1517, loss_cls_4: 0.9959, loss_box_4: 1.8286, loss_cns_4: 0.6493, loss_yns_4: 0.1571, loss_cls_5: 1.0035, loss_box_5: 1.8404, loss_cns_5: 0.6491, loss_yns_5: 0.1531, loss_cls_dn_0: 0.2323, loss_box_dn_0: 0.7897, loss_cls_dn_1: 0.1672, loss_box_dn_1: 0.8403, loss_cls_dn_2: 0.1710, loss_box_dn_2: 0.8296, loss_cls_dn_3: 0.1698, loss_box_dn_3: 0.8388, loss_cls_dn_4: 0.1723, loss_box_dn_4: 0.8668, loss_cls_dn_5: 0.1867, loss_box_dn_5: 0.8854, loss_dense_depth: 0.7984, loss: 28.3886, grad_norm: 50.3075 -2025-11-12 20:10:39,643 - mmdet - INFO - Iter [135/17500] lr: 1.536e-04, eta: 11:32:02, time: 1.579, data_time: 0.077, memory: 49163, loss_cls_0: 0.9016, loss_box_0: 1.7677, loss_cns_0: 0.6155, loss_yns_0: 0.1569, loss_cls_1: 0.9590, loss_box_1: 1.8948, loss_cns_1: 0.6457, loss_yns_1: 0.1548, loss_cls_2: 0.9888, loss_box_2: 1.8628, loss_cns_2: 0.6485, loss_yns_2: 0.1546, loss_cls_3: 0.9826, loss_box_3: 1.8458, loss_cns_3: 0.6513, loss_yns_3: 0.1557, loss_cls_4: 1.0028, loss_box_4: 1.8681, loss_cns_4: 0.6477, loss_yns_4: 0.1726, loss_cls_5: 1.0064, loss_box_5: 1.8811, loss_cns_5: 0.6443, loss_yns_5: 0.1608, loss_cls_dn_0: 0.2398, loss_box_dn_0: 0.7962, loss_cls_dn_1: 0.1653, loss_box_dn_1: 0.8397, loss_cls_dn_2: 0.1724, loss_box_dn_2: 0.8235, loss_cls_dn_3: 0.1719, loss_box_dn_3: 0.8202, loss_cls_dn_4: 0.1736, loss_box_dn_4: 0.8365, loss_cls_dn_5: 0.1805, loss_box_dn_5: 0.8451, loss_dense_depth: 0.8091, loss: 28.6435, grad_norm: 49.5323 -2025-11-12 20:10:41,167 - mmdet - INFO - Iter [136/17500] lr: 1.540e-04, eta: 11:30:09, time: 1.524, data_time: 0.077, memory: 49163, loss_cls_0: 0.8647, loss_box_0: 1.7534, loss_cns_0: 0.6220, loss_yns_0: 0.1526, loss_cls_1: 0.9349, loss_box_1: 1.8678, loss_cns_1: 0.6410, loss_yns_1: 0.1512, loss_cls_2: 0.9669, loss_box_2: 1.8153, loss_cns_2: 0.6510, loss_yns_2: 0.1525, loss_cls_3: 0.9784, loss_box_3: 1.7969, loss_cns_3: 0.6516, loss_yns_3: 0.1535, loss_cls_4: 0.9873, loss_box_4: 1.8056, loss_cns_4: 0.6516, loss_yns_4: 0.1611, loss_cls_5: 0.9737, loss_box_5: 1.8247, loss_cns_5: 0.6476, loss_yns_5: 0.1563, loss_cls_dn_0: 0.2330, loss_box_dn_0: 0.7902, loss_cls_dn_1: 0.1582, loss_box_dn_1: 0.8133, loss_cls_dn_2: 0.1627, loss_box_dn_2: 0.7876, loss_cls_dn_3: 0.1644, loss_box_dn_3: 0.7809, loss_cls_dn_4: 0.1707, loss_box_dn_4: 0.7853, loss_cls_dn_5: 0.1751, loss_box_dn_5: 0.7956, loss_dense_depth: 0.7885, loss: 27.9673, grad_norm: 37.7789 -2025-11-12 20:10:42,684 - mmdet - INFO - Iter [137/17500] lr: 1.544e-04, eta: 11:28:17, time: 1.518, data_time: 0.076, memory: 49163, loss_cls_0: 0.8657, loss_box_0: 1.7338, loss_cns_0: 0.6218, loss_yns_0: 0.1494, loss_cls_1: 0.9378, loss_box_1: 1.8967, loss_cns_1: 0.6374, loss_yns_1: 0.1533, loss_cls_2: 0.9646, loss_box_2: 1.8164, loss_cns_2: 0.6487, loss_yns_2: 0.1517, loss_cls_3: 0.9863, loss_box_3: 1.8132, loss_cns_3: 0.6490, loss_yns_3: 0.1517, loss_cls_4: 0.9993, loss_box_4: 1.8274, loss_cns_4: 0.6532, loss_yns_4: 0.1527, loss_cls_5: 0.9918, loss_box_5: 1.8308, loss_cns_5: 0.6503, loss_yns_5: 0.1519, loss_cls_dn_0: 0.2383, loss_box_dn_0: 0.7883, loss_cls_dn_1: 0.1548, loss_box_dn_1: 0.7906, loss_cls_dn_2: 0.1557, loss_box_dn_2: 0.7589, loss_cls_dn_3: 0.1580, loss_box_dn_3: 0.7607, loss_cls_dn_4: 0.1612, loss_box_dn_4: 0.7759, loss_cls_dn_5: 0.1749, loss_box_dn_5: 0.7839, loss_dense_depth: 0.7916, loss: 27.9278, grad_norm: 45.1021 -2025-11-12 20:10:44,223 - mmdet - INFO - Iter [138/17500] lr: 1.548e-04, eta: 11:26:29, time: 1.537, data_time: 0.078, memory: 49163, loss_cls_0: 0.8646, loss_box_0: 1.7251, loss_cns_0: 0.6248, loss_yns_0: 0.1487, loss_cls_1: 0.9362, loss_box_1: 1.8887, loss_cns_1: 0.6375, loss_yns_1: 0.1525, loss_cls_2: 0.9561, loss_box_2: 1.8337, loss_cns_2: 0.6454, loss_yns_2: 0.1498, loss_cls_3: 0.9582, loss_box_3: 1.8302, loss_cns_3: 0.6446, loss_yns_3: 0.1501, loss_cls_4: 0.9759, loss_box_4: 1.8440, loss_cns_4: 0.6447, loss_yns_4: 0.1546, loss_cls_5: 0.9818, loss_box_5: 1.8550, loss_cns_5: 0.6440, loss_yns_5: 0.1506, loss_cls_dn_0: 0.2341, loss_box_dn_0: 0.7927, loss_cls_dn_1: 0.1530, loss_box_dn_1: 0.7855, loss_cls_dn_2: 0.1559, loss_box_dn_2: 0.7733, loss_cls_dn_3: 0.1591, loss_box_dn_3: 0.7862, loss_cls_dn_4: 0.1600, loss_box_dn_4: 0.8170, loss_cls_dn_5: 0.1758, loss_box_dn_5: 0.8362, loss_dense_depth: 0.8291, loss: 28.0549, grad_norm: 42.8241 -2025-11-12 20:10:45,747 - mmdet - INFO - Iter [139/17500] lr: 1.552e-04, eta: 11:24:40, time: 1.525, data_time: 0.081, memory: 49163, loss_cls_0: 0.8384, loss_box_0: 1.7281, loss_cns_0: 0.6275, loss_yns_0: 0.1497, loss_cls_1: 0.9215, loss_box_1: 1.8586, loss_cns_1: 0.6416, loss_yns_1: 0.1511, loss_cls_2: 0.9496, loss_box_2: 1.8381, loss_cns_2: 0.6469, loss_yns_2: 0.1499, loss_cls_3: 0.9464, loss_box_3: 1.8244, loss_cns_3: 0.6477, loss_yns_3: 0.1492, loss_cls_4: 0.9603, loss_box_4: 1.8284, loss_cns_4: 0.6440, loss_yns_4: 0.1592, loss_cls_5: 0.9493, loss_box_5: 1.8722, loss_cns_5: 0.6412, loss_yns_5: 0.1511, loss_cls_dn_0: 0.2278, loss_box_dn_0: 0.7899, loss_cls_dn_1: 0.1559, loss_box_dn_1: 0.8149, loss_cls_dn_2: 0.1622, loss_box_dn_2: 0.8283, loss_cls_dn_3: 0.1622, loss_box_dn_3: 0.8497, loss_cls_dn_4: 0.1685, loss_box_dn_4: 0.8778, loss_cls_dn_5: 0.1738, loss_box_dn_5: 0.9134, loss_dense_depth: 0.8184, loss: 28.2171, grad_norm: 54.1202 -2025-11-12 20:10:47,289 - mmdet - INFO - Iter [140/17500] lr: 1.556e-04, eta: 11:22:56, time: 1.541, data_time: 0.083, memory: 49163, loss_cls_0: 0.8377, loss_box_0: 1.7270, loss_cns_0: 0.6299, loss_yns_0: 0.1486, loss_cls_1: 0.9331, loss_box_1: 1.8417, loss_cns_1: 0.6476, loss_yns_1: 0.1492, loss_cls_2: 0.9625, loss_box_2: 1.8231, loss_cns_2: 0.6502, loss_yns_2: 0.1494, loss_cls_3: 0.9606, loss_box_3: 1.8060, loss_cns_3: 0.6508, loss_yns_3: 0.1491, loss_cls_4: 0.9644, loss_box_4: 1.8080, loss_cns_4: 0.6490, loss_yns_4: 0.1595, loss_cls_5: 0.9537, loss_box_5: 1.8391, loss_cns_5: 0.6435, loss_yns_5: 0.1519, loss_cls_dn_0: 0.2255, loss_box_dn_0: 0.7913, loss_cls_dn_1: 0.1519, loss_box_dn_1: 0.8689, loss_cls_dn_2: 0.1565, loss_box_dn_2: 0.8850, loss_cls_dn_3: 0.1583, loss_box_dn_3: 0.9024, loss_cls_dn_4: 0.1651, loss_box_dn_4: 0.9225, loss_cls_dn_5: 0.1703, loss_box_dn_5: 0.9561, loss_dense_depth: 0.8114, loss: 28.4010, grad_norm: 61.7888 -2025-11-12 20:10:48,889 - mmdet - INFO - Iter [141/17500] lr: 1.560e-04, eta: 11:21:20, time: 1.602, data_time: 0.150, memory: 49163, loss_cls_0: 0.8827, loss_box_0: 1.7924, loss_cns_0: 0.6255, loss_yns_0: 0.1502, loss_cls_1: 0.9427, loss_box_1: 1.9050, loss_cns_1: 0.6441, loss_yns_1: 0.1499, loss_cls_2: 0.9638, loss_box_2: 1.8532, loss_cns_2: 0.6481, loss_yns_2: 0.1508, loss_cls_3: 0.9619, loss_box_3: 1.8393, loss_cns_3: 0.6492, loss_yns_3: 0.1499, loss_cls_4: 0.9752, loss_box_4: 1.8370, loss_cns_4: 0.6518, loss_yns_4: 0.1567, loss_cls_5: 0.9724, loss_box_5: 1.8396, loss_cns_5: 0.6495, loss_yns_5: 0.1525, loss_cls_dn_0: 0.2355, loss_box_dn_0: 0.7935, loss_cls_dn_1: 0.1561, loss_box_dn_1: 0.8625, loss_cls_dn_2: 0.1588, loss_box_dn_2: 0.8663, loss_cls_dn_3: 0.1583, loss_box_dn_3: 0.8750, loss_cls_dn_4: 0.1637, loss_box_dn_4: 0.8882, loss_cls_dn_5: 0.1744, loss_box_dn_5: 0.9033, loss_dense_depth: 0.8407, loss: 28.6196, grad_norm: 42.7018 -2025-11-12 20:10:50,443 - mmdet - INFO - Iter [142/17500] lr: 1.564e-04, eta: 11:19:40, time: 1.554, data_time: 0.077, memory: 49163, loss_cls_0: 0.8948, loss_box_0: 1.8174, loss_cns_0: 0.6198, loss_yns_0: 0.1514, loss_cls_1: 0.9528, loss_box_1: 1.8798, loss_cns_1: 0.6389, loss_yns_1: 0.1505, loss_cls_2: 0.9712, loss_box_2: 1.8643, loss_cns_2: 0.6410, loss_yns_2: 0.1503, loss_cls_3: 0.9692, loss_box_3: 1.8849, loss_cns_3: 0.6397, loss_yns_3: 0.1519, loss_cls_4: 0.9914, loss_box_4: 1.8496, loss_cns_4: 0.6446, loss_yns_4: 0.1518, loss_cls_5: 0.9751, loss_box_5: 1.8501, loss_cns_5: 0.6427, loss_yns_5: 0.1525, loss_cls_dn_0: 0.2339, loss_box_dn_0: 0.7905, loss_cls_dn_1: 0.1560, loss_box_dn_1: 0.8456, loss_cls_dn_2: 0.1599, loss_box_dn_2: 0.8455, loss_cls_dn_3: 0.1576, loss_box_dn_3: 0.8541, loss_cls_dn_4: 0.1642, loss_box_dn_4: 0.8492, loss_cls_dn_5: 0.1724, loss_box_dn_5: 0.8509, loss_dense_depth: 0.8810, loss: 28.5965, grad_norm: 49.4567 -2025-11-12 20:10:52,008 - mmdet - INFO - Iter [143/17500] lr: 1.568e-04, eta: 11:18:02, time: 1.563, data_time: 0.075, memory: 49163, loss_cls_0: 0.8674, loss_box_0: 1.7815, loss_cns_0: 0.6228, loss_yns_0: 0.1502, loss_cls_1: 0.9642, loss_box_1: 1.8939, loss_cns_1: 0.6483, loss_yns_1: 0.1504, loss_cls_2: 0.9720, loss_box_2: 1.8591, loss_cns_2: 0.6502, loss_yns_2: 0.1482, loss_cls_3: 0.9782, loss_box_3: 1.8477, loss_cns_3: 0.6497, loss_yns_3: 0.1504, loss_cls_4: 0.9849, loss_box_4: 1.8302, loss_cns_4: 0.6526, loss_yns_4: 0.1528, loss_cls_5: 0.9804, loss_box_5: 1.8436, loss_cns_5: 0.6504, loss_yns_5: 0.1511, loss_cls_dn_0: 0.2252, loss_box_dn_0: 0.7808, loss_cls_dn_1: 0.1575, loss_box_dn_1: 0.7715, loss_cls_dn_2: 0.1612, loss_box_dn_2: 0.7507, loss_cls_dn_3: 0.1595, loss_box_dn_3: 0.7496, loss_cls_dn_4: 0.1682, loss_box_dn_4: 0.7488, loss_cls_dn_5: 0.1747, loss_box_dn_5: 0.7625, loss_dense_depth: 0.8319, loss: 28.0220, grad_norm: 34.7378 -2025-11-12 20:10:53,565 - mmdet - INFO - Iter [144/17500] lr: 1.572e-04, eta: 11:16:25, time: 1.558, data_time: 0.095, memory: 49163, loss_cls_0: 0.8913, loss_box_0: 1.7957, loss_cns_0: 0.6144, loss_yns_0: 0.1516, loss_cls_1: 0.9706, loss_box_1: 1.9071, loss_cns_1: 0.6403, loss_yns_1: 0.1510, loss_cls_2: 0.9756, loss_box_2: 1.8699, loss_cns_2: 0.6464, loss_yns_2: 0.1505, loss_cls_3: 0.9757, loss_box_3: 1.8641, loss_cns_3: 0.6462, loss_yns_3: 0.1502, loss_cls_4: 0.9966, loss_box_4: 1.8700, loss_cns_4: 0.6447, loss_yns_4: 0.1588, loss_cls_5: 0.9826, loss_box_5: 1.8764, loss_cns_5: 0.6459, loss_yns_5: 0.1515, loss_cls_dn_0: 0.2339, loss_box_dn_0: 0.7927, loss_cls_dn_1: 0.1651, loss_box_dn_1: 0.7849, loss_cls_dn_2: 0.1643, loss_box_dn_2: 0.7690, loss_cls_dn_3: 0.1620, loss_box_dn_3: 0.7722, loss_cls_dn_4: 0.1740, loss_box_dn_4: 0.7866, loss_cls_dn_5: 0.1772, loss_box_dn_5: 0.8141, loss_dense_depth: 0.8432, loss: 28.3666, grad_norm: 55.8502 -2025-11-12 20:10:55,088 - mmdet - INFO - Iter [145/17500] lr: 1.576e-04, eta: 11:14:45, time: 1.524, data_time: 0.073, memory: 49163, loss_cls_0: 0.8830, loss_box_0: 1.7870, loss_cns_0: 0.6175, loss_yns_0: 0.1516, loss_cls_1: 0.9483, loss_box_1: 1.9077, loss_cns_1: 0.6408, loss_yns_1: 0.1526, loss_cls_2: 0.9624, loss_box_2: 1.8780, loss_cns_2: 0.6485, loss_yns_2: 0.1547, loss_cls_3: 0.9679, loss_box_3: 1.8837, loss_cns_3: 0.6484, loss_yns_3: 0.1536, loss_cls_4: 1.0054, loss_box_4: 1.8738, loss_cns_4: 0.6463, loss_yns_4: 0.1618, loss_cls_5: 0.9761, loss_box_5: 1.8916, loss_cns_5: 0.6477, loss_yns_5: 0.1531, loss_cls_dn_0: 0.2337, loss_box_dn_0: 0.7881, loss_cls_dn_1: 0.1637, loss_box_dn_1: 0.8019, loss_cls_dn_2: 0.1598, loss_box_dn_2: 0.7944, loss_cls_dn_3: 0.1599, loss_box_dn_3: 0.8057, loss_cls_dn_4: 0.1703, loss_box_dn_4: 0.8270, loss_cls_dn_5: 0.1689, loss_box_dn_5: 0.8642, loss_dense_depth: 0.9064, loss: 28.5856, grad_norm: 49.6157 -2025-11-12 20:10:56,623 - mmdet - INFO - Iter [146/17500] lr: 1.580e-04, eta: 11:13:08, time: 1.535, data_time: 0.076, memory: 49163, loss_cls_0: 0.8519, loss_box_0: 1.7793, loss_cns_0: 0.6189, loss_yns_0: 0.1489, loss_cls_1: 0.9302, loss_box_1: 1.8635, loss_cns_1: 0.6386, loss_yns_1: 0.1505, loss_cls_2: 0.9640, loss_box_2: 1.8189, loss_cns_2: 0.6460, loss_yns_2: 0.1496, loss_cls_3: 0.9673, loss_box_3: 1.8185, loss_cns_3: 0.6476, loss_yns_3: 0.1511, loss_cls_4: 1.0037, loss_box_4: 1.8084, loss_cns_4: 0.6501, loss_yns_4: 0.1526, loss_cls_5: 0.9827, loss_box_5: 1.8426, loss_cns_5: 0.6414, loss_yns_5: 0.1501, loss_cls_dn_0: 0.2314, loss_box_dn_0: 0.7853, loss_cls_dn_1: 0.1545, loss_box_dn_1: 0.8354, loss_cls_dn_2: 0.1562, loss_box_dn_2: 0.8365, loss_cls_dn_3: 0.1593, loss_box_dn_3: 0.8509, loss_cls_dn_4: 0.1709, loss_box_dn_4: 0.8911, loss_cls_dn_5: 0.1738, loss_box_dn_5: 0.9396, loss_dense_depth: 0.8715, loss: 28.4327, grad_norm: 55.7992 -2025-11-12 20:10:58,140 - mmdet - INFO - Iter [147/17500] lr: 1.584e-04, eta: 11:11:30, time: 1.516, data_time: 0.076, memory: 49163, loss_cls_0: 0.8965, loss_box_0: 1.8005, loss_cns_0: 0.6204, loss_yns_0: 0.1507, loss_cls_1: 0.9488, loss_box_1: 1.8606, loss_cns_1: 0.6408, loss_yns_1: 0.1506, loss_cls_2: 0.9655, loss_box_2: 1.8023, loss_cns_2: 0.6498, loss_yns_2: 0.1502, loss_cls_3: 0.9749, loss_box_3: 1.8038, loss_cns_3: 0.6545, loss_yns_3: 0.1519, loss_cls_4: 1.0065, loss_box_4: 1.7948, loss_cns_4: 0.6604, loss_yns_4: 0.1506, loss_cls_5: 0.9904, loss_box_5: 1.8253, loss_cns_5: 0.6524, loss_yns_5: 0.1519, loss_cls_dn_0: 0.2316, loss_box_dn_0: 0.7845, loss_cls_dn_1: 0.1551, loss_box_dn_1: 0.8807, loss_cls_dn_2: 0.1602, loss_box_dn_2: 0.8869, loss_cls_dn_3: 0.1639, loss_box_dn_3: 0.8994, loss_cls_dn_4: 0.1728, loss_box_dn_4: 0.9284, loss_cls_dn_5: 0.1751, loss_box_dn_5: 0.9681, loss_dense_depth: 0.8385, loss: 28.6992, grad_norm: 51.8454 -2025-11-12 20:10:59,650 - mmdet - INFO - Iter [148/17500] lr: 1.588e-04, eta: 11:09:52, time: 1.511, data_time: 0.076, memory: 49163, loss_cls_0: 0.8800, loss_box_0: 1.7884, loss_cns_0: 0.6204, loss_yns_0: 0.1501, loss_cls_1: 0.9337, loss_box_1: 1.8406, loss_cns_1: 0.6457, loss_yns_1: 0.1474, loss_cls_2: 0.9600, loss_box_2: 1.7823, loss_cns_2: 0.6519, loss_yns_2: 0.1484, loss_cls_3: 0.9631, loss_box_3: 1.7982, loss_cns_3: 0.6545, loss_yns_3: 0.1502, loss_cls_4: 0.9816, loss_box_4: 1.7891, loss_cns_4: 0.6536, loss_yns_4: 0.1493, loss_cls_5: 0.9835, loss_box_5: 1.7864, loss_cns_5: 0.6594, loss_yns_5: 0.1496, loss_cls_dn_0: 0.2231, loss_box_dn_0: 0.7812, loss_cls_dn_1: 0.1531, loss_box_dn_1: 0.8706, loss_cls_dn_2: 0.1565, loss_box_dn_2: 0.8536, loss_cls_dn_3: 0.1592, loss_box_dn_3: 0.8674, loss_cls_dn_4: 0.1635, loss_box_dn_4: 0.8760, loss_cls_dn_5: 0.1709, loss_box_dn_5: 0.8957, loss_dense_depth: 0.8684, loss: 28.3066, grad_norm: 46.7675 -2025-11-12 20:11:01,167 - mmdet - INFO - Iter [149/17500] lr: 1.592e-04, eta: 11:08:17, time: 1.519, data_time: 0.078, memory: 49163, loss_cls_0: 0.8470, loss_box_0: 1.7763, loss_cns_0: 0.6163, loss_yns_0: 0.1473, loss_cls_1: 0.9245, loss_box_1: 1.8524, loss_cns_1: 0.6410, loss_yns_1: 0.1477, loss_cls_2: 0.9645, loss_box_2: 1.7979, loss_cns_2: 0.6483, loss_yns_2: 0.1494, loss_cls_3: 0.9723, loss_box_3: 1.8009, loss_cns_3: 0.6493, loss_yns_3: 0.1499, loss_cls_4: 1.0046, loss_box_4: 1.7880, loss_cns_4: 0.6481, loss_yns_4: 0.1489, loss_cls_5: 0.9800, loss_box_5: 1.7824, loss_cns_5: 0.6517, loss_yns_5: 0.1507, loss_cls_dn_0: 0.2199, loss_box_dn_0: 0.7769, loss_cls_dn_1: 0.1475, loss_box_dn_1: 0.8047, loss_cls_dn_2: 0.1521, loss_box_dn_2: 0.7765, loss_cls_dn_3: 0.1531, loss_box_dn_3: 0.7780, loss_cls_dn_4: 0.1608, loss_box_dn_4: 0.7735, loss_cls_dn_5: 0.1657, loss_box_dn_5: 0.7746, loss_dense_depth: 0.8222, loss: 27.7447, grad_norm: 41.9171 -2025-11-12 20:11:02,698 - mmdet - INFO - Iter [150/17500] lr: 1.596e-04, eta: 11:06:44, time: 1.528, data_time: 0.077, memory: 49163, loss_cls_0: 0.8796, loss_box_0: 1.7387, loss_cns_0: 0.6158, loss_yns_0: 0.1461, loss_cls_1: 0.9263, loss_box_1: 1.8684, loss_cns_1: 0.6369, loss_yns_1: 0.1477, loss_cls_2: 0.9720, loss_box_2: 1.7820, loss_cns_2: 0.6485, loss_yns_2: 0.1496, loss_cls_3: 0.9653, loss_box_3: 1.7713, loss_cns_3: 0.6530, loss_yns_3: 0.1491, loss_cls_4: 0.9851, loss_box_4: 1.7697, loss_cns_4: 0.6540, loss_yns_4: 0.1506, loss_cls_5: 0.9772, loss_box_5: 1.7817, loss_cns_5: 0.6501, loss_yns_5: 0.1479, loss_cls_dn_0: 0.2278, loss_box_dn_0: 0.7748, loss_cls_dn_1: 0.1465, loss_box_dn_1: 0.7263, loss_cls_dn_2: 0.1503, loss_box_dn_2: 0.7025, loss_cls_dn_3: 0.1523, loss_box_dn_3: 0.6977, loss_cls_dn_4: 0.1607, loss_box_dn_4: 0.7041, loss_cls_dn_5: 0.1630, loss_box_dn_5: 0.7126, loss_dense_depth: 0.8808, loss: 27.3660, grad_norm: 35.5880 -2025-11-12 20:11:04,224 - mmdet - INFO - Iter [151/17500] lr: 1.600e-04, eta: 11:05:12, time: 1.527, data_time: 0.075, memory: 49163, loss_cls_0: 0.8735, loss_box_0: 1.7649, loss_cns_0: 0.6230, loss_yns_0: 0.1468, loss_cls_1: 0.9264, loss_box_1: 1.8203, loss_cns_1: 0.6415, loss_yns_1: 0.1485, loss_cls_2: 0.9554, loss_box_2: 1.7697, loss_cns_2: 0.6486, loss_yns_2: 0.1486, loss_cls_3: 0.9647, loss_box_3: 1.7595, loss_cns_3: 0.6480, loss_yns_3: 0.1505, loss_cls_4: 0.9769, loss_box_4: 1.7623, loss_cns_4: 0.6515, loss_yns_4: 0.1508, loss_cls_5: 0.9767, loss_box_5: 1.7695, loss_cns_5: 0.6484, loss_yns_5: 0.1490, loss_cls_dn_0: 0.2285, loss_box_dn_0: 0.7807, loss_cls_dn_1: 0.1443, loss_box_dn_1: 0.7458, loss_cls_dn_2: 0.1454, loss_box_dn_2: 0.7446, loss_cls_dn_3: 0.1478, loss_box_dn_3: 0.7626, loss_cls_dn_4: 0.1539, loss_box_dn_4: 0.7844, loss_cls_dn_5: 0.1612, loss_box_dn_5: 0.8079, loss_dense_depth: 0.8302, loss: 27.5124, grad_norm: 44.2607 -2025-11-12 20:11:05,732 - mmdet - INFO - Iter [152/17500] lr: 1.604e-04, eta: 11:03:40, time: 1.509, data_time: 0.076, memory: 49163, loss_cls_0: 0.8642, loss_box_0: 1.7691, loss_cns_0: 0.6201, loss_yns_0: 0.1503, loss_cls_1: 0.9486, loss_box_1: 1.8487, loss_cns_1: 0.6397, loss_yns_1: 0.1499, loss_cls_2: 0.9723, loss_box_2: 1.8267, loss_cns_2: 0.6458, loss_yns_2: 0.1504, loss_cls_3: 0.9722, loss_box_3: 1.8100, loss_cns_3: 0.6464, loss_yns_3: 0.1522, loss_cls_4: 0.9839, loss_box_4: 1.8150, loss_cns_4: 0.6479, loss_yns_4: 0.1537, loss_cls_5: 0.9839, loss_box_5: 1.8155, loss_cns_5: 0.6477, loss_yns_5: 0.1539, loss_cls_dn_0: 0.2251, loss_box_dn_0: 0.7892, loss_cls_dn_1: 0.1471, loss_box_dn_1: 0.7978, loss_cls_dn_2: 0.1479, loss_box_dn_2: 0.8126, loss_cls_dn_3: 0.1502, loss_box_dn_3: 0.8401, loss_cls_dn_4: 0.1554, loss_box_dn_4: 0.8651, loss_cls_dn_5: 0.1616, loss_box_dn_5: 0.8926, loss_dense_depth: 0.7998, loss: 28.1525, grad_norm: 51.5854 -2025-11-12 20:11:07,255 - mmdet - INFO - Iter [153/17500] lr: 1.608e-04, eta: 11:02:10, time: 1.523, data_time: 0.076, memory: 49163, loss_cls_0: 0.8811, loss_box_0: 1.7512, loss_cns_0: 0.6218, loss_yns_0: 0.1509, loss_cls_1: 0.9524, loss_box_1: 1.8549, loss_cns_1: 0.6402, loss_yns_1: 0.1530, loss_cls_2: 0.9679, loss_box_2: 1.8243, loss_cns_2: 0.6471, loss_yns_2: 0.1521, loss_cls_3: 0.9790, loss_box_3: 1.8018, loss_cns_3: 0.6476, loss_yns_3: 0.1528, loss_cls_4: 1.0038, loss_box_4: 1.8164, loss_cns_4: 0.6461, loss_yns_4: 0.1541, loss_cls_5: 0.9989, loss_box_5: 1.8304, loss_cns_5: 0.6435, loss_yns_5: 0.1550, loss_cls_dn_0: 0.2337, loss_box_dn_0: 0.7866, loss_cls_dn_1: 0.1483, loss_box_dn_1: 0.8273, loss_cls_dn_2: 0.1481, loss_box_dn_2: 0.8371, loss_cls_dn_3: 0.1474, loss_box_dn_3: 0.8541, loss_cls_dn_4: 0.1544, loss_box_dn_4: 0.8762, loss_cls_dn_5: 0.1626, loss_box_dn_5: 0.9043, loss_dense_depth: 0.8576, loss: 28.3641, grad_norm: 43.5913 -2025-11-12 20:11:08,768 - mmdet - INFO - Iter [154/17500] lr: 1.612e-04, eta: 11:00:40, time: 1.512, data_time: 0.076, memory: 49163, loss_cls_0: 0.8602, loss_box_0: 1.7412, loss_cns_0: 0.6205, loss_yns_0: 0.1496, loss_cls_1: 0.9453, loss_box_1: 1.7883, loss_cns_1: 0.6407, loss_yns_1: 0.1494, loss_cls_2: 0.9492, loss_box_2: 1.7426, loss_cns_2: 0.6486, loss_yns_2: 0.1497, loss_cls_3: 0.9655, loss_box_3: 1.7288, loss_cns_3: 0.6494, loss_yns_3: 0.1497, loss_cls_4: 1.0001, loss_box_4: 1.7308, loss_cns_4: 0.6522, loss_yns_4: 0.1506, loss_cls_5: 0.9687, loss_box_5: 1.7395, loss_cns_5: 0.6497, loss_yns_5: 0.1512, loss_cls_dn_0: 0.2263, loss_box_dn_0: 0.7760, loss_cls_dn_1: 0.1485, loss_box_dn_1: 0.8310, loss_cls_dn_2: 0.1468, loss_box_dn_2: 0.8221, loss_cls_dn_3: 0.1484, loss_box_dn_3: 0.8239, loss_cls_dn_4: 0.1533, loss_box_dn_4: 0.8333, loss_cls_dn_5: 0.1601, loss_box_dn_5: 0.8454, loss_dense_depth: 0.7687, loss: 27.6052, grad_norm: 40.7751 -2025-11-12 20:11:10,345 - mmdet - INFO - Iter [155/17500] lr: 1.616e-04, eta: 10:59:18, time: 1.577, data_time: 0.078, memory: 49163, loss_cls_0: 0.8341, loss_box_0: 1.7547, loss_cns_0: 0.6184, loss_yns_0: 0.1497, loss_cls_1: 0.9026, loss_box_1: 1.7456, loss_cns_1: 0.6429, loss_yns_1: 0.1498, loss_cls_2: 0.9291, loss_box_2: 1.7201, loss_cns_2: 0.6490, loss_yns_2: 0.1488, loss_cls_3: 0.9317, loss_box_3: 1.7123, loss_cns_3: 0.6513, loss_yns_3: 0.1489, loss_cls_4: 0.9546, loss_box_4: 1.6953, loss_cns_4: 0.6530, loss_yns_4: 0.1500, loss_cls_5: 0.9538, loss_box_5: 1.6831, loss_cns_5: 0.6524, loss_yns_5: 0.1495, loss_cls_dn_0: 0.2204, loss_box_dn_0: 0.7895, loss_cls_dn_1: 0.1472, loss_box_dn_1: 0.8276, loss_cls_dn_2: 0.1429, loss_box_dn_2: 0.8134, loss_cls_dn_3: 0.1440, loss_box_dn_3: 0.8100, loss_cls_dn_4: 0.1508, loss_box_dn_4: 0.8083, loss_cls_dn_5: 0.1553, loss_box_dn_5: 0.8049, loss_dense_depth: 0.7999, loss: 27.1950, grad_norm: 39.8959 -2025-11-12 20:11:11,864 - mmdet - INFO - Iter [156/17500] lr: 1.620e-04, eta: 10:57:51, time: 1.519, data_time: 0.075, memory: 49163, loss_cls_0: 0.8585, loss_box_0: 1.7425, loss_cns_0: 0.6221, loss_yns_0: 0.1514, loss_cls_1: 0.9386, loss_box_1: 1.7687, loss_cns_1: 0.6388, loss_yns_1: 0.1505, loss_cls_2: 0.9588, loss_box_2: 1.7289, loss_cns_2: 0.6497, loss_yns_2: 0.1516, loss_cls_3: 0.9723, loss_box_3: 1.6985, loss_cns_3: 0.6508, loss_yns_3: 0.1530, loss_cls_4: 0.9556, loss_box_4: 1.6950, loss_cns_4: 0.6511, loss_yns_4: 0.1518, loss_cls_5: 0.9640, loss_box_5: 1.6907, loss_cns_5: 0.6489, loss_yns_5: 0.1522, loss_cls_dn_0: 0.2248, loss_box_dn_0: 0.7794, loss_cls_dn_1: 0.1468, loss_box_dn_1: 0.8117, loss_cls_dn_2: 0.1436, loss_box_dn_2: 0.7919, loss_cls_dn_3: 0.1461, loss_box_dn_3: 0.7784, loss_cls_dn_4: 0.1485, loss_box_dn_4: 0.7789, loss_cls_dn_5: 0.1548, loss_box_dn_5: 0.7802, loss_dense_depth: 0.8016, loss: 27.2309, grad_norm: 45.5358 -2025-11-12 20:11:13,386 - mmdet - INFO - Iter [157/17500] lr: 1.624e-04, eta: 10:56:26, time: 1.521, data_time: 0.074, memory: 49163, loss_cls_0: 0.8509, loss_box_0: 1.7277, loss_cns_0: 0.6200, loss_yns_0: 0.1538, loss_cls_1: 0.9141, loss_box_1: 1.8236, loss_cns_1: 0.6355, loss_yns_1: 0.1539, loss_cls_2: 0.9510, loss_box_2: 1.7637, loss_cns_2: 0.6480, loss_yns_2: 0.1518, loss_cls_3: 0.9547, loss_box_3: 1.7439, loss_cns_3: 0.6457, loss_yns_3: 0.1560, loss_cls_4: 0.9546, loss_box_4: 1.7491, loss_cns_4: 0.6467, loss_yns_4: 0.1531, loss_cls_5: 0.9705, loss_box_5: 1.7597, loss_cns_5: 0.6431, loss_yns_5: 0.1541, loss_cls_dn_0: 0.2214, loss_box_dn_0: 0.7793, loss_cls_dn_1: 0.1427, loss_box_dn_1: 0.7668, loss_cls_dn_2: 0.1435, loss_box_dn_2: 0.7515, loss_cls_dn_3: 0.1484, loss_box_dn_3: 0.7586, loss_cls_dn_4: 0.1499, loss_box_dn_4: 0.7765, loss_cls_dn_5: 0.1632, loss_box_dn_5: 0.8038, loss_dense_depth: 0.7928, loss: 27.3236, grad_norm: 49.0621 -2025-11-12 20:11:14,930 - mmdet - INFO - Iter [158/17500] lr: 1.628e-04, eta: 10:55:04, time: 1.544, data_time: 0.076, memory: 49163, loss_cls_0: 0.8264, loss_box_0: 1.7290, loss_cns_0: 0.6149, loss_yns_0: 0.1530, loss_cls_1: 0.9120, loss_box_1: 1.8002, loss_cns_1: 0.6400, loss_yns_1: 0.1533, loss_cls_2: 0.9519, loss_box_2: 1.7459, loss_cns_2: 0.6449, loss_yns_2: 0.1509, loss_cls_3: 0.9505, loss_box_3: 1.7532, loss_cns_3: 0.6459, loss_yns_3: 0.1534, loss_cls_4: 0.9427, loss_box_4: 1.7676, loss_cns_4: 0.6502, loss_yns_4: 0.1525, loss_cls_5: 0.9458, loss_box_5: 1.7643, loss_cns_5: 0.6447, loss_yns_5: 0.1534, loss_cls_dn_0: 0.2144, loss_box_dn_0: 0.7817, loss_cls_dn_1: 0.1424, loss_box_dn_1: 0.8267, loss_cls_dn_2: 0.1451, loss_box_dn_2: 0.8314, loss_cls_dn_3: 0.1502, loss_box_dn_3: 0.8677, loss_cls_dn_4: 0.1498, loss_box_dn_4: 0.9143, loss_cls_dn_5: 0.1586, loss_box_dn_5: 0.9554, loss_dense_depth: 0.7477, loss: 27.7318, grad_norm: 55.4528 -2025-11-12 20:11:16,451 - mmdet - INFO - Iter [159/17500] lr: 1.632e-04, eta: 10:53:40, time: 1.521, data_time: 0.081, memory: 49163, loss_cls_0: 0.8319, loss_box_0: 1.7369, loss_cns_0: 0.6162, loss_yns_0: 0.1547, loss_cls_1: 0.9228, loss_box_1: 1.8009, loss_cns_1: 0.6429, loss_yns_1: 0.1532, loss_cls_2: 0.9445, loss_box_2: 1.7800, loss_cns_2: 0.6464, loss_yns_2: 0.1522, loss_cls_3: 0.9391, loss_box_3: 1.7687, loss_cns_3: 0.6489, loss_yns_3: 0.1537, loss_cls_4: 0.9607, loss_box_4: 1.7765, loss_cns_4: 0.6540, loss_yns_4: 0.1534, loss_cls_5: 0.9434, loss_box_5: 1.7767, loss_cns_5: 0.6486, loss_yns_5: 0.1539, loss_cls_dn_0: 0.2147, loss_box_dn_0: 0.7745, loss_cls_dn_1: 0.1422, loss_box_dn_1: 0.8796, loss_cls_dn_2: 0.1448, loss_box_dn_2: 0.8964, loss_cls_dn_3: 0.1478, loss_box_dn_3: 0.9240, loss_cls_dn_4: 0.1498, loss_box_dn_4: 0.9739, loss_cls_dn_5: 0.1570, loss_box_dn_5: 1.0114, loss_dense_depth: 0.7701, loss: 28.1465, grad_norm: 61.8078 -2025-11-12 20:11:17,969 - mmdet - INFO - Iter [160/17500] lr: 1.636e-04, eta: 10:52:17, time: 1.513, data_time: 0.080, memory: 49163, loss_cls_0: 0.8212, loss_box_0: 1.7412, loss_cns_0: 0.6159, loss_yns_0: 0.1545, loss_cls_1: 0.8982, loss_box_1: 1.7940, loss_cns_1: 0.6431, loss_yns_1: 0.1549, loss_cls_2: 0.9245, loss_box_2: 1.7673, loss_cns_2: 0.6488, loss_yns_2: 0.1546, loss_cls_3: 0.9265, loss_box_3: 1.7308, loss_cns_3: 0.6514, loss_yns_3: 0.1555, loss_cls_4: 0.9486, loss_box_4: 1.7373, loss_cns_4: 0.6520, loss_yns_4: 0.1538, loss_cls_5: 0.9325, loss_box_5: 1.7453, loss_cns_5: 0.6490, loss_yns_5: 0.1549, loss_cls_dn_0: 0.2139, loss_box_dn_0: 0.7713, loss_cls_dn_1: 0.1408, loss_box_dn_1: 0.8651, loss_cls_dn_2: 0.1425, loss_box_dn_2: 0.8757, loss_cls_dn_3: 0.1461, loss_box_dn_3: 0.8765, loss_cls_dn_4: 0.1517, loss_box_dn_4: 0.9080, loss_cls_dn_5: 0.1566, loss_box_dn_5: 0.9377, loss_dense_depth: 0.7814, loss: 27.7229, grad_norm: 37.1007 -2025-11-12 20:11:19,535 - mmdet - INFO - Iter [161/17500] lr: 1.640e-04, eta: 10:51:01, time: 1.571, data_time: 0.155, memory: 49163, loss_cls_0: 0.8144, loss_box_0: 1.7663, loss_cns_0: 0.6185, loss_yns_0: 0.1548, loss_cls_1: 0.8904, loss_box_1: 1.7935, loss_cns_1: 0.6438, loss_yns_1: 0.1577, loss_cls_2: 0.9299, loss_box_2: 1.7446, loss_cns_2: 0.6523, loss_yns_2: 0.1553, loss_cls_3: 0.9545, loss_box_3: 1.7188, loss_cns_3: 0.6563, loss_yns_3: 0.1577, loss_cls_4: 0.9450, loss_box_4: 1.7227, loss_cns_4: 0.6540, loss_yns_4: 0.1560, loss_cls_5: 0.9466, loss_box_5: 1.7021, loss_cns_5: 0.6529, loss_yns_5: 0.1567, loss_cls_dn_0: 0.2130, loss_box_dn_0: 0.7786, loss_cls_dn_1: 0.1405, loss_box_dn_1: 0.8208, loss_cls_dn_2: 0.1436, loss_box_dn_2: 0.8130, loss_cls_dn_3: 0.1501, loss_box_dn_3: 0.8026, loss_cls_dn_4: 0.1516, loss_box_dn_4: 0.8130, loss_cls_dn_5: 0.1575, loss_box_dn_5: 0.8191, loss_dense_depth: 0.7684, loss: 27.3165, grad_norm: 55.7444 -2025-11-12 20:11:21,081 - mmdet - INFO - Iter [162/17500] lr: 1.644e-04, eta: 10:49:43, time: 1.546, data_time: 0.074, memory: 49163, loss_cls_0: 0.8291, loss_box_0: 1.7508, loss_cns_0: 0.6149, loss_yns_0: 0.1561, loss_cls_1: 0.8806, loss_box_1: 1.8077, loss_cns_1: 0.6374, loss_yns_1: 0.1584, loss_cls_2: 0.9304, loss_box_2: 1.7304, loss_cns_2: 0.6513, loss_yns_2: 0.1578, loss_cls_3: 0.9440, loss_box_3: 1.7162, loss_cns_3: 0.6573, loss_yns_3: 0.1572, loss_cls_4: 0.9734, loss_box_4: 1.7056, loss_cns_4: 0.6539, loss_yns_4: 0.1579, loss_cls_5: 0.9591, loss_box_5: 1.6792, loss_cns_5: 0.6560, loss_yns_5: 0.1569, loss_cls_dn_0: 0.2161, loss_box_dn_0: 0.7815, loss_cls_dn_1: 0.1392, loss_box_dn_1: 0.7637, loss_cls_dn_2: 0.1398, loss_box_dn_2: 0.7395, loss_cls_dn_3: 0.1447, loss_box_dn_3: 0.7267, loss_cls_dn_4: 0.1547, loss_box_dn_4: 0.7188, loss_cls_dn_5: 0.1585, loss_box_dn_5: 0.7184, loss_dense_depth: 0.7539, loss: 26.8772, grad_norm: 40.0695 -2025-11-12 20:11:22,652 - mmdet - INFO - Iter [163/17500] lr: 1.648e-04, eta: 10:48:28, time: 1.571, data_time: 0.075, memory: 49163, loss_cls_0: 0.8206, loss_box_0: 1.7471, loss_cns_0: 0.6181, loss_yns_0: 0.1569, loss_cls_1: 0.8795, loss_box_1: 1.8367, loss_cns_1: 0.6308, loss_yns_1: 0.1577, loss_cls_2: 0.9204, loss_box_2: 1.7384, loss_cns_2: 0.6529, loss_yns_2: 0.1579, loss_cls_3: 0.9255, loss_box_3: 1.7102, loss_cns_3: 0.6575, loss_yns_3: 0.1595, loss_cls_4: 0.9594, loss_box_4: 1.7054, loss_cns_4: 0.6555, loss_yns_4: 0.1580, loss_cls_5: 0.9342, loss_box_5: 1.7093, loss_cns_5: 0.6551, loss_yns_5: 0.1579, loss_cls_dn_0: 0.2132, loss_box_dn_0: 0.7735, loss_cls_dn_1: 0.1428, loss_box_dn_1: 0.7395, loss_cls_dn_2: 0.1437, loss_box_dn_2: 0.7177, loss_cls_dn_3: 0.1457, loss_box_dn_3: 0.7058, loss_cls_dn_4: 0.1506, loss_box_dn_4: 0.7084, loss_cls_dn_5: 0.1516, loss_box_dn_5: 0.7155, loss_dense_depth: 0.7623, loss: 26.7745, grad_norm: 42.4519 -2025-11-12 20:11:24,226 - mmdet - INFO - Iter [164/17500] lr: 1.652e-04, eta: 10:47:15, time: 1.575, data_time: 0.097, memory: 49163, loss_cls_0: 0.8242, loss_box_0: 1.7760, loss_cns_0: 0.6225, loss_yns_0: 0.1585, loss_cls_1: 0.8813, loss_box_1: 1.8528, loss_cns_1: 0.6374, loss_yns_1: 0.1604, loss_cls_2: 0.9294, loss_box_2: 1.7440, loss_cns_2: 0.6547, loss_yns_2: 0.1590, loss_cls_3: 0.9322, loss_box_3: 1.7346, loss_cns_3: 0.6556, loss_yns_3: 0.1613, loss_cls_4: 0.9520, loss_box_4: 1.7422, loss_cns_4: 0.6552, loss_yns_4: 0.1606, loss_cls_5: 0.9438, loss_box_5: 1.7378, loss_cns_5: 0.6532, loss_yns_5: 0.1612, loss_cls_dn_0: 0.2127, loss_box_dn_0: 0.7806, loss_cls_dn_1: 0.1373, loss_box_dn_1: 0.7566, loss_cls_dn_2: 0.1388, loss_box_dn_2: 0.7477, loss_cls_dn_3: 0.1405, loss_box_dn_3: 0.7572, loss_cls_dn_4: 0.1448, loss_box_dn_4: 0.7765, loss_cls_dn_5: 0.1501, loss_box_dn_5: 0.7955, loss_dense_depth: 0.7487, loss: 27.1772, grad_norm: 46.0969 -2025-11-12 20:11:25,761 - mmdet - INFO - Iter [165/17500] lr: 1.656e-04, eta: 10:45:59, time: 1.534, data_time: 0.077, memory: 49163, loss_cls_0: 0.8321, loss_box_0: 1.8038, loss_cns_0: 0.6153, loss_yns_0: 0.1588, loss_cls_1: 0.8978, loss_box_1: 1.8053, loss_cns_1: 0.6394, loss_yns_1: 0.1609, loss_cls_2: 0.9449, loss_box_2: 1.7439, loss_cns_2: 0.6523, loss_yns_2: 0.1604, loss_cls_3: 0.9436, loss_box_3: 1.7301, loss_cns_3: 0.6547, loss_yns_3: 0.1613, loss_cls_4: 0.9429, loss_box_4: 1.7295, loss_cns_4: 0.6557, loss_yns_4: 0.1629, loss_cls_5: 0.9486, loss_box_5: 1.7484, loss_cns_5: 0.6522, loss_yns_5: 0.1607, loss_cls_dn_0: 0.2148, loss_box_dn_0: 0.7814, loss_cls_dn_1: 0.1350, loss_box_dn_1: 0.7997, loss_cls_dn_2: 0.1369, loss_box_dn_2: 0.8002, loss_cls_dn_3: 0.1389, loss_box_dn_3: 0.8201, loss_cls_dn_4: 0.1450, loss_box_dn_4: 0.8386, loss_cls_dn_5: 0.1533, loss_box_dn_5: 0.8832, loss_dense_depth: 0.7741, loss: 27.5266, grad_norm: 37.6536 -2025-11-12 20:11:27,280 - mmdet - INFO - Iter [166/17500] lr: 1.660e-04, eta: 10:44:42, time: 1.519, data_time: 0.077, memory: 49163, loss_cls_0: 0.8196, loss_box_0: 1.7857, loss_cns_0: 0.6118, loss_yns_0: 0.1563, loss_cls_1: 0.9021, loss_box_1: 1.7693, loss_cns_1: 0.6495, loss_yns_1: 0.1598, loss_cls_2: 0.9409, loss_box_2: 1.7096, loss_cns_2: 0.6590, loss_yns_2: 0.1602, loss_cls_3: 0.9403, loss_box_3: 1.6839, loss_cns_3: 0.6616, loss_yns_3: 0.1603, loss_cls_4: 0.9444, loss_box_4: 1.6807, loss_cns_4: 0.6614, loss_yns_4: 0.1601, loss_cls_5: 0.9336, loss_box_5: 1.7026, loss_cns_5: 0.6591, loss_yns_5: 0.1610, loss_cls_dn_0: 0.2091, loss_box_dn_0: 0.7721, loss_cls_dn_1: 0.1359, loss_box_dn_1: 0.8396, loss_cls_dn_2: 0.1387, loss_box_dn_2: 0.8410, loss_cls_dn_3: 0.1411, loss_box_dn_3: 0.8472, loss_cls_dn_4: 0.1471, loss_box_dn_4: 0.8598, loss_cls_dn_5: 0.1517, loss_box_dn_5: 0.9013, loss_dense_depth: 0.7793, loss: 27.4369, grad_norm: 50.6362 -2025-11-12 20:11:34,021 - mmdet - INFO - Iter [167/17500] lr: 1.664e-04, eta: 10:52:28, time: 6.743, data_time: 0.076, memory: 49163, loss_cls_0: 0.8702, loss_box_0: 1.7808, loss_cns_0: 0.6096, loss_yns_0: 0.1596, loss_cls_1: 0.9352, loss_box_1: 1.7456, loss_cns_1: 0.6501, loss_yns_1: 0.1613, loss_cls_2: 0.9562, loss_box_2: 1.6947, loss_cns_2: 0.6559, loss_yns_2: 0.1597, loss_cls_3: 0.9548, loss_box_3: 1.6673, loss_cns_3: 0.6579, loss_yns_3: 0.1617, loss_cls_4: 0.9737, loss_box_4: 1.6594, loss_cns_4: 0.6591, loss_yns_4: 0.1582, loss_cls_5: 0.9635, loss_box_5: 1.6614, loss_cns_5: 0.6593, loss_yns_5: 0.1608, loss_cls_dn_0: 0.2216, loss_box_dn_0: 0.7740, loss_cls_dn_1: 0.1448, loss_box_dn_1: 0.8764, loss_cls_dn_2: 0.1463, loss_box_dn_2: 0.8683, loss_cls_dn_3: 0.1478, loss_box_dn_3: 0.8638, loss_cls_dn_4: 0.1524, loss_box_dn_4: 0.8649, loss_cls_dn_5: 0.1550, loss_box_dn_5: 0.8848, loss_dense_depth: 0.8108, loss: 27.6267, grad_norm: 33.6713 -2025-11-12 20:11:35,517 - mmdet - INFO - Iter [168/17500] lr: 1.668e-04, eta: 10:51:07, time: 1.495, data_time: 0.073, memory: 49163, loss_cls_0: 0.8510, loss_box_0: 1.7518, loss_cns_0: 0.6218, loss_yns_0: 0.1609, loss_cls_1: 0.9261, loss_box_1: 1.7285, loss_cns_1: 0.6478, loss_yns_1: 0.1624, loss_cls_2: 0.9499, loss_box_2: 1.6722, loss_cns_2: 0.6552, loss_yns_2: 0.1604, loss_cls_3: 0.9505, loss_box_3: 1.6711, loss_cns_3: 0.6557, loss_yns_3: 0.1617, loss_cls_4: 0.9649, loss_box_4: 1.6500, loss_cns_4: 0.6591, loss_yns_4: 0.1618, loss_cls_5: 0.9682, loss_box_5: 1.6469, loss_cns_5: 0.6579, loss_yns_5: 0.1619, loss_cls_dn_0: 0.2136, loss_box_dn_0: 0.7664, loss_cls_dn_1: 0.1446, loss_box_dn_1: 0.8185, loss_cls_dn_2: 0.1437, loss_box_dn_2: 0.7949, loss_cls_dn_3: 0.1447, loss_box_dn_3: 0.7869, loss_cls_dn_4: 0.1482, loss_box_dn_4: 0.7774, loss_cls_dn_5: 0.1539, loss_box_dn_5: 0.7811, loss_dense_depth: 0.8241, loss: 27.0955, grad_norm: 33.4088 -2025-11-12 20:11:37,022 - mmdet - INFO - Iter [169/17500] lr: 1.672e-04, eta: 10:49:48, time: 1.506, data_time: 0.076, memory: 49163, loss_cls_0: 0.8963, loss_box_0: 1.7815, loss_cns_0: 0.6176, loss_yns_0: 0.1621, loss_cls_1: 0.9429, loss_box_1: 1.7568, loss_cns_1: 0.6442, loss_yns_1: 0.1613, loss_cls_2: 0.9621, loss_box_2: 1.7016, loss_cns_2: 0.6479, loss_yns_2: 0.1611, loss_cls_3: 0.9734, loss_box_3: 1.6949, loss_cns_3: 0.6459, loss_yns_3: 0.1607, loss_cls_4: 0.9987, loss_box_4: 1.6575, loss_cns_4: 0.6467, loss_yns_4: 0.1597, loss_cls_5: 0.9839, loss_box_5: 1.6761, loss_cns_5: 0.6521, loss_yns_5: 0.1623, loss_cls_dn_0: 0.2178, loss_box_dn_0: 0.7726, loss_cls_dn_1: 0.1396, loss_box_dn_1: 0.7691, loss_cls_dn_2: 0.1380, loss_box_dn_2: 0.7453, loss_cls_dn_3: 0.1391, loss_box_dn_3: 0.7451, loss_cls_dn_4: 0.1454, loss_box_dn_4: 0.7326, loss_cls_dn_5: 0.1502, loss_box_dn_5: 0.7349, loss_dense_depth: 0.8872, loss: 27.1644, grad_norm: 31.9031 -2025-11-12 20:11:38,551 - mmdet - INFO - Iter [170/17500] lr: 1.676e-04, eta: 10:48:32, time: 1.528, data_time: 0.076, memory: 49163, loss_cls_0: 0.8586, loss_box_0: 1.7350, loss_cns_0: 0.6211, loss_yns_0: 0.1596, loss_cls_1: 0.9366, loss_box_1: 1.7234, loss_cns_1: 0.6482, loss_yns_1: 0.1611, loss_cls_2: 0.9581, loss_box_2: 1.6609, loss_cns_2: 0.6519, loss_yns_2: 0.1618, loss_cls_3: 0.9650, loss_box_3: 1.6587, loss_cns_3: 0.6526, loss_yns_3: 0.1595, loss_cls_4: 0.9739, loss_box_4: 1.6484, loss_cns_4: 0.6557, loss_yns_4: 0.1613, loss_cls_5: 0.9912, loss_box_5: 1.6573, loss_cns_5: 0.6532, loss_yns_5: 0.1613, loss_cls_dn_0: 0.2155, loss_box_dn_0: 0.7661, loss_cls_dn_1: 0.1374, loss_box_dn_1: 0.7342, loss_cls_dn_2: 0.1376, loss_box_dn_2: 0.7205, loss_cls_dn_3: 0.1360, loss_box_dn_3: 0.7368, loss_cls_dn_4: 0.1401, loss_box_dn_4: 0.7461, loss_cls_dn_5: 0.1518, loss_box_dn_5: 0.7665, loss_dense_depth: 0.8203, loss: 26.8233, grad_norm: 33.7774 -2025-11-12 20:11:40,078 - mmdet - INFO - Iter [171/17500] lr: 1.680e-04, eta: 10:47:17, time: 1.526, data_time: 0.076, memory: 49163, loss_cls_0: 0.8752, loss_box_0: 1.7044, loss_cns_0: 0.6121, loss_yns_0: 0.1578, loss_cls_1: 0.9405, loss_box_1: 1.7174, loss_cns_1: 0.6493, loss_yns_1: 0.1593, loss_cls_2: 0.9622, loss_box_2: 1.6726, loss_cns_2: 0.6538, loss_yns_2: 0.1581, loss_cls_3: 0.9649, loss_box_3: 1.6662, loss_cns_3: 0.6553, loss_yns_3: 0.1581, loss_cls_4: 0.9788, loss_box_4: 1.6749, loss_cns_4: 0.6578, loss_yns_4: 0.1611, loss_cls_5: 0.9741, loss_box_5: 1.6729, loss_cns_5: 0.6549, loss_yns_5: 0.1587, loss_cls_dn_0: 0.2172, loss_box_dn_0: 0.7647, loss_cls_dn_1: 0.1393, loss_box_dn_1: 0.7631, loss_cls_dn_2: 0.1415, loss_box_dn_2: 0.7580, loss_cls_dn_3: 0.1426, loss_box_dn_3: 0.7768, loss_cls_dn_4: 0.1455, loss_box_dn_4: 0.7995, loss_cls_dn_5: 0.1488, loss_box_dn_5: 0.8226, loss_dense_depth: 0.7790, loss: 27.0392, grad_norm: 40.2900 -2025-11-12 20:11:41,597 - mmdet - INFO - Iter [172/17500] lr: 1.684e-04, eta: 10:46:02, time: 1.519, data_time: 0.074, memory: 49163, loss_cls_0: 0.8748, loss_box_0: 1.7375, loss_cns_0: 0.6114, loss_yns_0: 0.1578, loss_cls_1: 0.9402, loss_box_1: 1.7536, loss_cns_1: 0.6489, loss_yns_1: 0.1590, loss_cls_2: 0.9565, loss_box_2: 1.7394, loss_cns_2: 0.6495, loss_yns_2: 0.1595, loss_cls_3: 0.9664, loss_box_3: 1.7279, loss_cns_3: 0.6534, loss_yns_3: 0.1592, loss_cls_4: 0.9843, loss_box_4: 1.7259, loss_cns_4: 0.6545, loss_yns_4: 0.1605, loss_cls_5: 0.9732, loss_box_5: 1.7323, loss_cns_5: 0.6522, loss_yns_5: 0.1601, loss_cls_dn_0: 0.2211, loss_box_dn_0: 0.7635, loss_cls_dn_1: 0.1390, loss_box_dn_1: 0.7829, loss_cls_dn_2: 0.1357, loss_box_dn_2: 0.7882, loss_cls_dn_3: 0.1418, loss_box_dn_3: 0.7963, loss_cls_dn_4: 0.1472, loss_box_dn_4: 0.8118, loss_cls_dn_5: 0.1510, loss_box_dn_5: 0.8345, loss_dense_depth: 0.8221, loss: 27.4732, grad_norm: 44.7740 -2025-11-12 20:11:43,122 - mmdet - INFO - Iter [173/17500] lr: 1.688e-04, eta: 10:44:49, time: 1.527, data_time: 0.073, memory: 49163, loss_cls_0: 0.8385, loss_box_0: 1.7227, loss_cns_0: 0.6216, loss_yns_0: 0.1590, loss_cls_1: 0.9286, loss_box_1: 1.7178, loss_cns_1: 0.6509, loss_yns_1: 0.1570, loss_cls_2: 0.9579, loss_box_2: 1.7123, loss_cns_2: 0.6520, loss_yns_2: 0.1586, loss_cls_3: 0.9672, loss_box_3: 1.6757, loss_cns_3: 0.6579, loss_yns_3: 0.1583, loss_cls_4: 0.9690, loss_box_4: 1.6754, loss_cns_4: 0.6572, loss_yns_4: 0.1581, loss_cls_5: 0.9681, loss_box_5: 1.6821, loss_cns_5: 0.6562, loss_yns_5: 0.1581, loss_cls_dn_0: 0.2184, loss_box_dn_0: 0.7572, loss_cls_dn_1: 0.1389, loss_box_dn_1: 0.7844, loss_cls_dn_2: 0.1368, loss_box_dn_2: 0.7882, loss_cls_dn_3: 0.1418, loss_box_dn_3: 0.7714, loss_cls_dn_4: 0.1422, loss_box_dn_4: 0.7818, loss_cls_dn_5: 0.1477, loss_box_dn_5: 0.7976, loss_dense_depth: 0.7862, loss: 27.0532, grad_norm: 36.0276 -2025-11-12 20:11:44,653 - mmdet - INFO - Iter [174/17500] lr: 1.692e-04, eta: 10:43:36, time: 1.529, data_time: 0.071, memory: 49163, loss_cls_0: 0.8737, loss_box_0: 1.7717, loss_cns_0: 0.6150, loss_yns_0: 0.1588, loss_cls_1: 0.9310, loss_box_1: 1.7431, loss_cns_1: 0.6420, loss_yns_1: 0.1568, loss_cls_2: 0.9629, loss_box_2: 1.7202, loss_cns_2: 0.6462, loss_yns_2: 0.1549, loss_cls_3: 0.9664, loss_box_3: 1.7132, loss_cns_3: 0.6535, loss_yns_3: 0.1564, loss_cls_4: 0.9785, loss_box_4: 1.7174, loss_cns_4: 0.6557, loss_yns_4: 0.1578, loss_cls_5: 0.9823, loss_box_5: 1.6951, loss_cns_5: 0.6528, loss_yns_5: 0.1556, loss_cls_dn_0: 0.2153, loss_box_dn_0: 0.7676, loss_cls_dn_1: 0.1382, loss_box_dn_1: 0.7626, loss_cls_dn_2: 0.1413, loss_box_dn_2: 0.7569, loss_cls_dn_3: 0.1420, loss_box_dn_3: 0.7417, loss_cls_dn_4: 0.1460, loss_box_dn_4: 0.7493, loss_cls_dn_5: 0.1614, loss_box_dn_5: 0.7468, loss_dense_depth: 0.8696, loss: 27.1997, grad_norm: 48.6802 -2025-11-12 20:11:46,227 - mmdet - INFO - Iter [175/17500] lr: 1.696e-04, eta: 10:42:29, time: 1.575, data_time: 0.081, memory: 49163, loss_cls_0: 0.8575, loss_box_0: 1.7462, loss_cns_0: 0.6207, loss_yns_0: 0.1584, loss_cls_1: 0.9293, loss_box_1: 1.7342, loss_cns_1: 0.6471, loss_yns_1: 0.1569, loss_cls_2: 0.9528, loss_box_2: 1.6930, loss_cns_2: 0.6502, loss_yns_2: 0.1569, loss_cls_3: 0.9615, loss_box_3: 1.6949, loss_cns_3: 0.6550, loss_yns_3: 0.1584, loss_cls_4: 0.9789, loss_box_4: 1.6777, loss_cns_4: 0.6571, loss_yns_4: 0.1598, loss_cls_5: 0.9695, loss_box_5: 1.6747, loss_cns_5: 0.6556, loss_yns_5: 0.1562, loss_cls_dn_0: 0.2177, loss_box_dn_0: 0.7596, loss_cls_dn_1: 0.1356, loss_box_dn_1: 0.7430, loss_cls_dn_2: 0.1387, loss_box_dn_2: 0.7310, loss_cls_dn_3: 0.1390, loss_box_dn_3: 0.7249, loss_cls_dn_4: 0.1431, loss_box_dn_4: 0.7225, loss_cls_dn_5: 0.1446, loss_box_dn_5: 0.7245, loss_dense_depth: 0.7818, loss: 26.8088, grad_norm: 34.9613 -2025-11-12 20:11:47,741 - mmdet - INFO - Iter [176/17500] lr: 1.700e-04, eta: 10:41:17, time: 1.514, data_time: 0.076, memory: 49163, loss_cls_0: 0.8444, loss_box_0: 1.7129, loss_cns_0: 0.6198, loss_yns_0: 0.1547, loss_cls_1: 0.9326, loss_box_1: 1.6853, loss_cns_1: 0.6509, loss_yns_1: 0.1520, loss_cls_2: 0.9491, loss_box_2: 1.6677, loss_cns_2: 0.6553, loss_yns_2: 0.1512, loss_cls_3: 0.9778, loss_box_3: 1.6681, loss_cns_3: 0.6596, loss_yns_3: 0.1519, loss_cls_4: 0.9835, loss_box_4: 1.6730, loss_cns_4: 0.6621, loss_yns_4: 0.1528, loss_cls_5: 0.9721, loss_box_5: 1.6622, loss_cns_5: 0.6582, loss_yns_5: 0.1525, loss_cls_dn_0: 0.2194, loss_box_dn_0: 0.7698, loss_cls_dn_1: 0.1359, loss_box_dn_1: 0.7310, loss_cls_dn_2: 0.1355, loss_box_dn_2: 0.7224, loss_cls_dn_3: 0.1390, loss_box_dn_3: 0.7325, loss_cls_dn_4: 0.1426, loss_box_dn_4: 0.7400, loss_cls_dn_5: 0.1562, loss_box_dn_5: 0.7492, loss_dense_depth: 0.8591, loss: 26.7823, grad_norm: 43.9797 -2025-11-12 20:11:49,240 - mmdet - INFO - Iter [177/17500] lr: 1.704e-04, eta: 10:40:04, time: 1.500, data_time: 0.072, memory: 49163, loss_cls_0: 0.8397, loss_box_0: 1.7128, loss_cns_0: 0.6202, loss_yns_0: 0.1511, loss_cls_1: 0.9101, loss_box_1: 1.6657, loss_cns_1: 0.6490, loss_yns_1: 0.1496, loss_cls_2: 0.9431, loss_box_2: 1.6537, loss_cns_2: 0.6553, loss_yns_2: 0.1500, loss_cls_3: 0.9722, loss_box_3: 1.6708, loss_cns_3: 0.6608, loss_yns_3: 0.1503, loss_cls_4: 0.9723, loss_box_4: 1.6744, loss_cns_4: 0.6610, loss_yns_4: 0.1500, loss_cls_5: 0.9631, loss_box_5: 1.6623, loss_cns_5: 0.6571, loss_yns_5: 0.1487, loss_cls_dn_0: 0.2169, loss_box_dn_0: 0.7657, loss_cls_dn_1: 0.1380, loss_box_dn_1: 0.7452, loss_cls_dn_2: 0.1370, loss_box_dn_2: 0.7425, loss_cls_dn_3: 0.1419, loss_box_dn_3: 0.7642, loss_cls_dn_4: 0.1454, loss_box_dn_4: 0.7780, loss_cls_dn_5: 0.1581, loss_box_dn_5: 0.7958, loss_dense_depth: 0.8200, loss: 26.7921, grad_norm: 50.2052 -2025-11-12 20:11:50,757 - mmdet - INFO - Iter [178/17500] lr: 1.708e-04, eta: 10:38:54, time: 1.516, data_time: 0.071, memory: 49163, loss_cls_0: 0.8241, loss_box_0: 1.7494, loss_cns_0: 0.6173, loss_yns_0: 0.1509, loss_cls_1: 0.9042, loss_box_1: 1.7167, loss_cns_1: 0.6453, loss_yns_1: 0.1491, loss_cls_2: 0.9378, loss_box_2: 1.6812, loss_cns_2: 0.6521, loss_yns_2: 0.1493, loss_cls_3: 0.9376, loss_box_3: 1.6655, loss_cns_3: 0.6532, loss_yns_3: 0.1496, loss_cls_4: 0.9488, loss_box_4: 1.6659, loss_cns_4: 0.6529, loss_yns_4: 0.1503, loss_cls_5: 0.9475, loss_box_5: 1.6785, loss_cns_5: 0.6503, loss_yns_5: 0.1505, loss_cls_dn_0: 0.2172, loss_box_dn_0: 0.7740, loss_cls_dn_1: 0.1415, loss_box_dn_1: 0.7762, loss_cls_dn_2: 0.1414, loss_box_dn_2: 0.7732, loss_cls_dn_3: 0.1433, loss_box_dn_3: 0.7944, loss_cls_dn_4: 0.1493, loss_box_dn_4: 0.8079, loss_cls_dn_5: 0.1539, loss_box_dn_5: 0.8394, loss_dense_depth: 0.8092, loss: 26.9490, grad_norm: 41.5059 -2025-11-12 20:11:52,290 - mmdet - INFO - Iter [179/17500] lr: 1.712e-04, eta: 10:37:46, time: 1.534, data_time: 0.082, memory: 49163, loss_cls_0: 0.8225, loss_box_0: 1.7573, loss_cns_0: 0.6192, loss_yns_0: 0.1515, loss_cls_1: 0.9079, loss_box_1: 1.7417, loss_cns_1: 0.6479, loss_yns_1: 0.1512, loss_cls_2: 0.9291, loss_box_2: 1.7298, loss_cns_2: 0.6504, loss_yns_2: 0.1515, loss_cls_3: 0.9367, loss_box_3: 1.7019, loss_cns_3: 0.6541, loss_yns_3: 0.1515, loss_cls_4: 0.9461, loss_box_4: 1.6869, loss_cns_4: 0.6546, loss_yns_4: 0.1517, loss_cls_5: 0.9452, loss_box_5: 1.6811, loss_cns_5: 0.6535, loss_yns_5: 0.1527, loss_cls_dn_0: 0.2182, loss_box_dn_0: 0.7810, loss_cls_dn_1: 0.1391, loss_box_dn_1: 0.7976, loss_cls_dn_2: 0.1422, loss_box_dn_2: 0.7985, loss_cls_dn_3: 0.1491, loss_box_dn_3: 0.8016, loss_cls_dn_4: 0.1546, loss_box_dn_4: 0.8032, loss_cls_dn_5: 0.1559, loss_box_dn_5: 0.8229, loss_dense_depth: 0.8104, loss: 27.1501, grad_norm: 49.1817 -2025-11-12 20:11:53,806 - mmdet - INFO - Iter [180/17500] lr: 1.716e-04, eta: 10:36:37, time: 1.516, data_time: 0.082, memory: 49163, loss_cls_0: 0.8172, loss_box_0: 1.6991, loss_cns_0: 0.6215, loss_yns_0: 0.1487, loss_cls_1: 0.9010, loss_box_1: 1.7057, loss_cns_1: 0.6469, loss_yns_1: 0.1480, loss_cls_2: 0.9231, loss_box_2: 1.6989, loss_cns_2: 0.6563, loss_yns_2: 0.1500, loss_cls_3: 0.9423, loss_box_3: 1.6572, loss_cns_3: 0.6598, loss_yns_3: 0.1500, loss_cls_4: 0.9395, loss_box_4: 1.6421, loss_cns_4: 0.6583, loss_yns_4: 0.1500, loss_cls_5: 0.9393, loss_box_5: 1.6313, loss_cns_5: 0.6604, loss_yns_5: 0.1521, loss_cls_dn_0: 0.2184, loss_box_dn_0: 0.7722, loss_cls_dn_1: 0.1381, loss_box_dn_1: 0.7853, loss_cls_dn_2: 0.1412, loss_box_dn_2: 0.7835, loss_cls_dn_3: 0.1511, loss_box_dn_3: 0.7665, loss_cls_dn_4: 0.1497, loss_box_dn_4: 0.7591, loss_cls_dn_5: 0.1523, loss_box_dn_5: 0.7642, loss_dense_depth: 0.7977, loss: 26.6778, grad_norm: 44.5319 -2025-11-12 20:11:55,392 - mmdet - INFO - Iter [181/17500] lr: 1.720e-04, eta: 10:35:35, time: 1.586, data_time: 0.149, memory: 49163, loss_cls_0: 0.8129, loss_box_0: 1.7200, loss_cns_0: 0.6185, loss_yns_0: 0.1494, loss_cls_1: 0.9004, loss_box_1: 1.7366, loss_cns_1: 0.6436, loss_yns_1: 0.1500, loss_cls_2: 0.9292, loss_box_2: 1.6897, loss_cns_2: 0.6547, loss_yns_2: 0.1511, loss_cls_3: 0.9283, loss_box_3: 1.6593, loss_cns_3: 0.6555, loss_yns_3: 0.1507, loss_cls_4: 0.9323, loss_box_4: 1.6438, loss_cns_4: 0.6570, loss_yns_4: 0.1515, loss_cls_5: 0.9372, loss_box_5: 1.6355, loss_cns_5: 0.6563, loss_yns_5: 0.1509, loss_cls_dn_0: 0.2195, loss_box_dn_0: 0.7726, loss_cls_dn_1: 0.1393, loss_box_dn_1: 0.7374, loss_cls_dn_2: 0.1404, loss_box_dn_2: 0.7203, loss_cls_dn_3: 0.1445, loss_box_dn_3: 0.7028, loss_cls_dn_4: 0.1440, loss_box_dn_4: 0.7032, loss_cls_dn_5: 0.1487, loss_box_dn_5: 0.7046, loss_dense_depth: 0.8129, loss: 26.4044, grad_norm: 34.1696 -2025-11-12 20:11:56,922 - mmdet - INFO - Iter [182/17500] lr: 1.724e-04, eta: 10:34:29, time: 1.530, data_time: 0.079, memory: 49163, loss_cls_0: 0.8008, loss_box_0: 1.7038, loss_cns_0: 0.6253, loss_yns_0: 0.1489, loss_cls_1: 0.8813, loss_box_1: 1.7318, loss_cns_1: 0.6465, loss_yns_1: 0.1480, loss_cls_2: 0.9150, loss_box_2: 1.6689, loss_cns_2: 0.6550, loss_yns_2: 0.1507, loss_cls_3: 0.9277, loss_box_3: 1.6798, loss_cns_3: 0.6560, loss_yns_3: 0.1479, loss_cls_4: 0.9419, loss_box_4: 1.6696, loss_cns_4: 0.6571, loss_yns_4: 0.1487, loss_cls_5: 0.9463, loss_box_5: 1.6532, loss_cns_5: 0.6543, loss_yns_5: 0.1476, loss_cls_dn_0: 0.2145, loss_box_dn_0: 0.7699, loss_cls_dn_1: 0.1371, loss_box_dn_1: 0.7267, loss_cls_dn_2: 0.1384, loss_box_dn_2: 0.7036, loss_cls_dn_3: 0.1414, loss_box_dn_3: 0.7069, loss_cls_dn_4: 0.1435, loss_box_dn_4: 0.7121, loss_cls_dn_5: 0.1464, loss_box_dn_5: 0.7121, loss_dense_depth: 0.7892, loss: 26.3478, grad_norm: 47.5843 -2025-11-12 20:11:58,501 - mmdet - INFO - Iter [183/17500] lr: 1.728e-04, eta: 10:33:28, time: 1.579, data_time: 0.076, memory: 49163, loss_cls_0: 0.7929, loss_box_0: 1.6715, loss_cns_0: 0.6246, loss_yns_0: 0.1468, loss_cls_1: 0.8856, loss_box_1: 1.6801, loss_cns_1: 0.6520, loss_yns_1: 0.1482, loss_cls_2: 0.9200, loss_box_2: 1.6220, loss_cns_2: 0.6579, loss_yns_2: 0.1495, loss_cls_3: 0.9350, loss_box_3: 1.6463, loss_cns_3: 0.6606, loss_yns_3: 0.1478, loss_cls_4: 0.9499, loss_box_4: 1.6322, loss_cns_4: 0.6602, loss_yns_4: 0.1503, loss_cls_5: 0.9323, loss_box_5: 1.6259, loss_cns_5: 0.6585, loss_yns_5: 0.1488, loss_cls_dn_0: 0.2111, loss_box_dn_0: 0.7565, loss_cls_dn_1: 0.1316, loss_box_dn_1: 0.7429, loss_cls_dn_2: 0.1353, loss_box_dn_2: 0.7229, loss_cls_dn_3: 0.1380, loss_box_dn_3: 0.7386, loss_cls_dn_4: 0.1420, loss_box_dn_4: 0.7476, loss_cls_dn_5: 0.1538, loss_box_dn_5: 0.7534, loss_dense_depth: 0.7592, loss: 26.2319, grad_norm: 54.8835 -2025-11-12 20:12:00,070 - mmdet - INFO - Iter [184/17500] lr: 1.732e-04, eta: 10:32:27, time: 1.568, data_time: 0.099, memory: 49163, loss_cls_0: 0.8112, loss_box_0: 1.6943, loss_cns_0: 0.6214, loss_yns_0: 0.1441, loss_cls_1: 0.8851, loss_box_1: 1.7086, loss_cns_1: 0.6521, loss_yns_1: 0.1455, loss_cls_2: 0.9104, loss_box_2: 1.6699, loss_cns_2: 0.6584, loss_yns_2: 0.1455, loss_cls_3: 0.9267, loss_box_3: 1.6779, loss_cns_3: 0.6609, loss_yns_3: 0.1465, loss_cls_4: 0.9381, loss_box_4: 1.6673, loss_cns_4: 0.6618, loss_yns_4: 0.1473, loss_cls_5: 0.9359, loss_box_5: 1.6620, loss_cns_5: 0.6621, loss_yns_5: 0.1487, loss_cls_dn_0: 0.2161, loss_box_dn_0: 0.7609, loss_cls_dn_1: 0.1367, loss_box_dn_1: 0.7726, loss_cls_dn_2: 0.1397, loss_box_dn_2: 0.7631, loss_cls_dn_3: 0.1427, loss_box_dn_3: 0.7818, loss_cls_dn_4: 0.1455, loss_box_dn_4: 0.7913, loss_cls_dn_5: 0.1556, loss_box_dn_5: 0.8070, loss_dense_depth: 0.8114, loss: 26.7060, grad_norm: 48.7295 -2025-11-12 20:12:01,592 - mmdet - INFO - Iter [185/17500] lr: 1.736e-04, eta: 10:31:22, time: 1.522, data_time: 0.077, memory: 49163, loss_cls_0: 0.8125, loss_box_0: 1.7027, loss_cns_0: 0.6179, loss_yns_0: 0.1474, loss_cls_1: 0.8882, loss_box_1: 1.7085, loss_cns_1: 0.6435, loss_yns_1: 0.1472, loss_cls_2: 0.9388, loss_box_2: 1.6655, loss_cns_2: 0.6514, loss_yns_2: 0.1491, loss_cls_3: 0.9307, loss_box_3: 1.6678, loss_cns_3: 0.6506, loss_yns_3: 0.1481, loss_cls_4: 0.9343, loss_box_4: 1.6783, loss_cns_4: 0.6546, loss_yns_4: 0.1479, loss_cls_5: 0.9339, loss_box_5: 1.6584, loss_cns_5: 0.6541, loss_yns_5: 0.1474, loss_cls_dn_0: 0.2171, loss_box_dn_0: 0.7649, loss_cls_dn_1: 0.1423, loss_box_dn_1: 0.8169, loss_cls_dn_2: 0.1467, loss_box_dn_2: 0.7995, loss_cls_dn_3: 0.1482, loss_box_dn_3: 0.8113, loss_cls_dn_4: 0.1492, loss_box_dn_4: 0.8257, loss_cls_dn_5: 0.1561, loss_box_dn_5: 0.8323, loss_dense_depth: 0.8078, loss: 26.8968, grad_norm: 45.9521 -2025-11-12 20:12:03,125 - mmdet - INFO - Iter [186/17500] lr: 1.740e-04, eta: 10:30:19, time: 1.533, data_time: 0.077, memory: 49163, loss_cls_0: 0.8304, loss_box_0: 1.6622, loss_cns_0: 0.6153, loss_yns_0: 0.1484, loss_cls_1: 0.9136, loss_box_1: 1.6920, loss_cns_1: 0.6515, loss_yns_1: 0.1498, loss_cls_2: 0.9482, loss_box_2: 1.6459, loss_cns_2: 0.6577, loss_yns_2: 0.1534, loss_cls_3: 0.9614, loss_box_3: 1.6515, loss_cns_3: 0.6548, loss_yns_3: 0.1500, loss_cls_4: 0.9664, loss_box_4: 1.6607, loss_cns_4: 0.6620, loss_yns_4: 0.1519, loss_cls_5: 0.9557, loss_box_5: 1.6276, loss_cns_5: 0.6610, loss_yns_5: 0.1522, loss_cls_dn_0: 0.2119, loss_box_dn_0: 0.7655, loss_cls_dn_1: 0.1441, loss_box_dn_1: 0.8103, loss_cls_dn_2: 0.1455, loss_box_dn_2: 0.7892, loss_cls_dn_3: 0.1491, loss_box_dn_3: 0.7992, loss_cls_dn_4: 0.1585, loss_box_dn_4: 0.8082, loss_cls_dn_5: 0.1627, loss_box_dn_5: 0.8067, loss_dense_depth: 0.7646, loss: 26.8391, grad_norm: 51.8268 -2025-11-12 20:12:04,648 - mmdet - INFO - Iter [187/17500] lr: 1.744e-04, eta: 10:29:16, time: 1.523, data_time: 0.077, memory: 49163, loss_cls_0: 0.8303, loss_box_0: 1.6645, loss_cns_0: 0.6201, loss_yns_0: 0.1478, loss_cls_1: 0.9235, loss_box_1: 1.7062, loss_cns_1: 0.6532, loss_yns_1: 0.1521, loss_cls_2: 0.9703, loss_box_2: 1.6565, loss_cns_2: 0.6609, loss_yns_2: 0.1573, loss_cls_3: 0.9655, loss_box_3: 1.6436, loss_cns_3: 0.6596, loss_yns_3: 0.1548, loss_cls_4: 0.9710, loss_box_4: 1.6551, loss_cns_4: 0.6634, loss_yns_4: 0.1544, loss_cls_5: 0.9551, loss_box_5: 1.6510, loss_cns_5: 0.6596, loss_yns_5: 0.1559, loss_cls_dn_0: 0.2145, loss_box_dn_0: 0.7622, loss_cls_dn_1: 0.1390, loss_box_dn_1: 0.7840, loss_cls_dn_2: 0.1409, loss_box_dn_2: 0.7659, loss_cls_dn_3: 0.1442, loss_box_dn_3: 0.7627, loss_cls_dn_4: 0.1546, loss_box_dn_4: 0.7706, loss_cls_dn_5: 0.1611, loss_box_dn_5: 0.7790, loss_dense_depth: 0.8093, loss: 26.8195, grad_norm: 46.3366 -2025-11-12 20:12:06,162 - mmdet - INFO - Iter [188/17500] lr: 1.748e-04, eta: 10:28:12, time: 1.515, data_time: 0.075, memory: 49163, loss_cls_0: 0.8243, loss_box_0: 1.6939, loss_cns_0: 0.6206, loss_yns_0: 0.1476, loss_cls_1: 0.9089, loss_box_1: 1.7428, loss_cns_1: 0.6420, loss_yns_1: 0.1496, loss_cls_2: 0.9596, loss_box_2: 1.6659, loss_cns_2: 0.6548, loss_yns_2: 0.1498, loss_cls_3: 0.9532, loss_box_3: 1.6333, loss_cns_3: 0.6533, loss_yns_3: 0.1492, loss_cls_4: 0.9778, loss_box_4: 1.6369, loss_cns_4: 0.6564, loss_yns_4: 0.1501, loss_cls_5: 0.9637, loss_box_5: 1.6578, loss_cns_5: 0.6525, loss_yns_5: 0.1492, loss_cls_dn_0: 0.2118, loss_box_dn_0: 0.7648, loss_cls_dn_1: 0.1445, loss_box_dn_1: 0.7454, loss_cls_dn_2: 0.1469, loss_box_dn_2: 0.7179, loss_cls_dn_3: 0.1470, loss_box_dn_3: 0.7101, loss_cls_dn_4: 0.1526, loss_box_dn_4: 0.7129, loss_cls_dn_5: 0.1570, loss_box_dn_5: 0.7273, loss_dense_depth: 0.7825, loss: 26.5141, grad_norm: 40.9754 -2025-11-12 20:12:07,676 - mmdet - INFO - Iter [189/17500] lr: 1.752e-04, eta: 10:27:09, time: 1.515, data_time: 0.076, memory: 49163, loss_cls_0: 0.8273, loss_box_0: 1.6966, loss_cns_0: 0.6283, loss_yns_0: 0.1519, loss_cls_1: 0.9073, loss_box_1: 1.6940, loss_cns_1: 0.6500, loss_yns_1: 0.1525, loss_cls_2: 0.9618, loss_box_2: 1.6680, loss_cns_2: 0.6564, loss_yns_2: 0.1520, loss_cls_3: 0.9579, loss_box_3: 1.6859, loss_cns_3: 0.6543, loss_yns_3: 0.1520, loss_cls_4: 0.9717, loss_box_4: 1.6713, loss_cns_4: 0.6574, loss_yns_4: 0.1577, loss_cls_5: 0.9535, loss_box_5: 1.6805, loss_cns_5: 0.6542, loss_yns_5: 0.1508, loss_cls_dn_0: 0.2126, loss_box_dn_0: 0.7743, loss_cls_dn_1: 0.1447, loss_box_dn_1: 0.7328, loss_cls_dn_2: 0.1539, loss_box_dn_2: 0.7290, loss_cls_dn_3: 0.1501, loss_box_dn_3: 0.7447, loss_cls_dn_4: 0.1546, loss_box_dn_4: 0.7424, loss_cls_dn_5: 0.1624, loss_box_dn_5: 0.7529, loss_dense_depth: 0.8263, loss: 26.7739, grad_norm: 58.1075 -2025-11-12 20:12:09,187 - mmdet - INFO - Iter [190/17500] lr: 1.755e-04, eta: 10:26:07, time: 1.509, data_time: 0.077, memory: 49163, loss_cls_0: 0.8406, loss_box_0: 1.7233, loss_cns_0: 0.6234, loss_yns_0: 0.1539, loss_cls_1: 0.9166, loss_box_1: 1.7179, loss_cns_1: 0.6487, loss_yns_1: 0.1547, loss_cls_2: 0.9755, loss_box_2: 1.6996, loss_cns_2: 0.6545, loss_yns_2: 0.1571, loss_cls_3: 0.9719, loss_box_3: 1.7333, loss_cns_3: 0.6532, loss_yns_3: 0.1559, loss_cls_4: 0.9863, loss_box_4: 1.7292, loss_cns_4: 0.6556, loss_yns_4: 0.1578, loss_cls_5: 0.9701, loss_box_5: 1.7146, loss_cns_5: 0.6552, loss_yns_5: 0.1562, loss_cls_dn_0: 0.2184, loss_box_dn_0: 0.7708, loss_cls_dn_1: 0.1450, loss_box_dn_1: 0.7530, loss_cls_dn_2: 0.1588, loss_box_dn_2: 0.7657, loss_cls_dn_3: 0.1553, loss_box_dn_3: 0.7966, loss_cls_dn_4: 0.1659, loss_box_dn_4: 0.7990, loss_cls_dn_5: 0.1739, loss_box_dn_5: 0.8146, loss_dense_depth: 0.8172, loss: 27.3396, grad_norm: 68.6053 -2025-11-12 20:12:10,716 - mmdet - INFO - Iter [191/17500] lr: 1.759e-04, eta: 10:25:07, time: 1.530, data_time: 0.074, memory: 49163, loss_cls_0: 0.8278, loss_box_0: 1.7340, loss_cns_0: 0.6172, loss_yns_0: 0.1520, loss_cls_1: 0.9058, loss_box_1: 1.7240, loss_cns_1: 0.6497, loss_yns_1: 0.1525, loss_cls_2: 0.9533, loss_box_2: 1.6888, loss_cns_2: 0.6548, loss_yns_2: 0.1560, loss_cls_3: 0.9633, loss_box_3: 1.7212, loss_cns_3: 0.6529, loss_yns_3: 0.1556, loss_cls_4: 0.9698, loss_box_4: 1.7207, loss_cns_4: 0.6539, loss_yns_4: 0.1563, loss_cls_5: 0.9626, loss_box_5: 1.7104, loss_cns_5: 0.6541, loss_yns_5: 0.1577, loss_cls_dn_0: 0.2169, loss_box_dn_0: 0.7604, loss_cls_dn_1: 0.1423, loss_box_dn_1: 0.7734, loss_cls_dn_2: 0.1495, loss_box_dn_2: 0.7705, loss_cls_dn_3: 0.1483, loss_box_dn_3: 0.7987, loss_cls_dn_4: 0.1543, loss_box_dn_4: 0.8060, loss_cls_dn_5: 0.1587, loss_box_dn_5: 0.8225, loss_dense_depth: 0.7973, loss: 27.1933, grad_norm: 48.2247 -2025-11-12 20:12:12,242 - mmdet - INFO - Iter [192/17500] lr: 1.763e-04, eta: 10:24:07, time: 1.526, data_time: 0.075, memory: 49163, loss_cls_0: 0.8251, loss_box_0: 1.7294, loss_cns_0: 0.6187, loss_yns_0: 0.1532, loss_cls_1: 0.9101, loss_box_1: 1.7188, loss_cns_1: 0.6509, loss_yns_1: 0.1533, loss_cls_2: 0.9513, loss_box_2: 1.6808, loss_cns_2: 0.6547, loss_yns_2: 0.1538, loss_cls_3: 0.9557, loss_box_3: 1.6953, loss_cns_3: 0.6543, loss_yns_3: 0.1565, loss_cls_4: 0.9553, loss_box_4: 1.7066, loss_cns_4: 0.6567, loss_yns_4: 0.1554, loss_cls_5: 0.9651, loss_box_5: 1.6857, loss_cns_5: 0.6543, loss_yns_5: 0.1561, loss_cls_dn_0: 0.2171, loss_box_dn_0: 0.7605, loss_cls_dn_1: 0.1402, loss_box_dn_1: 0.7886, loss_cls_dn_2: 0.1439, loss_box_dn_2: 0.7777, loss_cls_dn_3: 0.1444, loss_box_dn_3: 0.7896, loss_cls_dn_4: 0.1497, loss_box_dn_4: 0.8003, loss_cls_dn_5: 0.1550, loss_box_dn_5: 0.8049, loss_dense_depth: 0.7712, loss: 27.0402, grad_norm: 42.8027 -2025-11-12 20:12:13,751 - mmdet - INFO - Iter [193/17500] lr: 1.767e-04, eta: 10:23:06, time: 1.508, data_time: 0.072, memory: 49163, loss_cls_0: 0.8295, loss_box_0: 1.7254, loss_cns_0: 0.6217, loss_yns_0: 0.1514, loss_cls_1: 0.9122, loss_box_1: 1.7437, loss_cns_1: 0.6481, loss_yns_1: 0.1565, loss_cls_2: 0.9518, loss_box_2: 1.6681, loss_cns_2: 0.6573, loss_yns_2: 0.1542, loss_cls_3: 0.9639, loss_box_3: 1.6438, loss_cns_3: 0.6558, loss_yns_3: 0.1533, loss_cls_4: 0.9770, loss_box_4: 1.6674, loss_cns_4: 0.6592, loss_yns_4: 0.1575, loss_cls_5: 0.9791, loss_box_5: 1.6626, loss_cns_5: 0.6547, loss_yns_5: 0.1549, loss_cls_dn_0: 0.2151, loss_box_dn_0: 0.7683, loss_cls_dn_1: 0.1365, loss_box_dn_1: 0.7727, loss_cls_dn_2: 0.1425, loss_box_dn_2: 0.7599, loss_cls_dn_3: 0.1432, loss_box_dn_3: 0.7542, loss_cls_dn_4: 0.1464, loss_box_dn_4: 0.7648, loss_cls_dn_5: 0.1552, loss_box_dn_5: 0.7704, loss_dense_depth: 0.7769, loss: 26.8553, grad_norm: 48.5657 -2025-11-12 20:12:15,262 - mmdet - INFO - Iter [194/17500] lr: 1.771e-04, eta: 10:22:06, time: 1.512, data_time: 0.074, memory: 49163, loss_cls_0: 0.8422, loss_box_0: 1.7289, loss_cns_0: 0.6198, loss_yns_0: 0.1511, loss_cls_1: 0.9108, loss_box_1: 1.7937, loss_cns_1: 0.6378, loss_yns_1: 0.1549, loss_cls_2: 0.9516, loss_box_2: 1.7174, loss_cns_2: 0.6486, loss_yns_2: 0.1544, loss_cls_3: 0.9748, loss_box_3: 1.6967, loss_cns_3: 0.6470, loss_yns_3: 0.1530, loss_cls_4: 0.9999, loss_box_4: 1.7168, loss_cns_4: 0.6538, loss_yns_4: 0.1559, loss_cls_5: 0.9763, loss_box_5: 1.7214, loss_cns_5: 0.6543, loss_yns_5: 0.1545, loss_cls_dn_0: 0.2149, loss_box_dn_0: 0.7622, loss_cls_dn_1: 0.1366, loss_box_dn_1: 0.7755, loss_cls_dn_2: 0.1416, loss_box_dn_2: 0.7572, loss_cls_dn_3: 0.1436, loss_box_dn_3: 0.7530, loss_cls_dn_4: 0.1482, loss_box_dn_4: 0.7671, loss_cls_dn_5: 0.1551, loss_box_dn_5: 0.7687, loss_dense_depth: 0.7921, loss: 27.1317, grad_norm: 41.9178 -2025-11-12 20:12:16,826 - mmdet - INFO - Iter [195/17500] lr: 1.775e-04, eta: 10:21:11, time: 1.563, data_time: 0.074, memory: 49163, loss_cls_0: 0.8130, loss_box_0: 1.7140, loss_cns_0: 0.6311, loss_yns_0: 0.1483, loss_cls_1: 0.8735, loss_box_1: 1.6850, loss_cns_1: 0.6460, loss_yns_1: 0.1467, loss_cls_2: 0.9076, loss_box_2: 1.6493, loss_cns_2: 0.6493, loss_yns_2: 0.1495, loss_cls_3: 0.9110, loss_box_3: 1.6503, loss_cns_3: 0.6498, loss_yns_3: 0.1512, loss_cls_4: 0.9160, loss_box_4: 1.6526, loss_cns_4: 0.6584, loss_yns_4: 0.1505, loss_cls_5: 0.9199, loss_box_5: 1.6293, loss_cns_5: 0.6606, loss_yns_5: 0.1503, loss_cls_dn_0: 0.2009, loss_box_dn_0: 0.7634, loss_cls_dn_1: 0.1301, loss_box_dn_1: 0.7490, loss_cls_dn_2: 0.1348, loss_box_dn_2: 0.7361, loss_cls_dn_3: 0.1341, loss_box_dn_3: 0.7395, loss_cls_dn_4: 0.1397, loss_box_dn_4: 0.7480, loss_cls_dn_5: 0.1489, loss_box_dn_5: 0.7455, loss_dense_depth: 0.7635, loss: 26.2465, grad_norm: 51.1292 -2025-11-12 20:12:18,351 - mmdet - INFO - Iter [196/17500] lr: 1.779e-04, eta: 10:20:13, time: 1.527, data_time: 0.074, memory: 49163, loss_cls_0: 0.8085, loss_box_0: 1.7333, loss_cns_0: 0.6282, loss_yns_0: 0.1492, loss_cls_1: 0.8684, loss_box_1: 1.7009, loss_cns_1: 0.6505, loss_yns_1: 0.1488, loss_cls_2: 0.9025, loss_box_2: 1.6404, loss_cns_2: 0.6602, loss_yns_2: 0.1497, loss_cls_3: 0.9175, loss_box_3: 1.6382, loss_cns_3: 0.6580, loss_yns_3: 0.1507, loss_cls_4: 0.9277, loss_box_4: 1.6422, loss_cns_4: 0.6610, loss_yns_4: 0.1523, loss_cls_5: 0.9454, loss_box_5: 1.6278, loss_cns_5: 0.6596, loss_yns_5: 0.1514, loss_cls_dn_0: 0.2025, loss_box_dn_0: 0.7725, loss_cls_dn_1: 0.1308, loss_box_dn_1: 0.7497, loss_cls_dn_2: 0.1303, loss_box_dn_2: 0.7354, loss_cls_dn_3: 0.1337, loss_box_dn_3: 0.7376, loss_cls_dn_4: 0.1368, loss_box_dn_4: 0.7509, loss_cls_dn_5: 0.1525, loss_box_dn_5: 0.7562, loss_dense_depth: 0.7870, loss: 26.3483, grad_norm: 33.7100 -2025-11-12 20:12:19,866 - mmdet - INFO - Iter [197/17500] lr: 1.783e-04, eta: 10:19:15, time: 1.513, data_time: 0.072, memory: 49163, loss_cls_0: 0.8175, loss_box_0: 1.7766, loss_cns_0: 0.6241, loss_yns_0: 0.1498, loss_cls_1: 0.8955, loss_box_1: 1.7554, loss_cns_1: 0.6522, loss_yns_1: 0.1515, loss_cls_2: 0.9231, loss_box_2: 1.7216, loss_cns_2: 0.6543, loss_yns_2: 0.1502, loss_cls_3: 0.9195, loss_box_3: 1.7237, loss_cns_3: 0.6551, loss_yns_3: 0.1505, loss_cls_4: 0.9273, loss_box_4: 1.7639, loss_cns_4: 0.6585, loss_yns_4: 0.1551, loss_cls_5: 0.9290, loss_box_5: 1.7678, loss_cns_5: 0.6540, loss_yns_5: 0.1510, loss_cls_dn_0: 0.2125, loss_box_dn_0: 0.7605, loss_cls_dn_1: 0.1290, loss_box_dn_1: 0.7551, loss_cls_dn_2: 0.1286, loss_box_dn_2: 0.7539, loss_cls_dn_3: 0.1340, loss_box_dn_3: 0.7623, loss_cls_dn_4: 0.1378, loss_box_dn_4: 0.7993, loss_cls_dn_5: 0.1572, loss_box_dn_5: 0.8157, loss_dense_depth: 0.8353, loss: 27.1084, grad_norm: 53.5164 -2025-11-12 20:12:21,390 - mmdet - INFO - Iter [198/17500] lr: 1.787e-04, eta: 10:18:19, time: 1.524, data_time: 0.072, memory: 49163, loss_cls_0: 0.8382, loss_box_0: 1.7647, loss_cns_0: 0.6207, loss_yns_0: 0.1546, loss_cls_1: 0.8971, loss_box_1: 1.7146, loss_cns_1: 0.6466, loss_yns_1: 0.1527, loss_cls_2: 0.9149, loss_box_2: 1.6912, loss_cns_2: 0.6481, loss_yns_2: 0.1536, loss_cls_3: 0.9335, loss_box_3: 1.6948, loss_cns_3: 0.6534, loss_yns_3: 0.1556, loss_cls_4: 0.9359, loss_box_4: 1.7029, loss_cns_4: 0.6596, loss_yns_4: 0.1572, loss_cls_5: 0.9238, loss_box_5: 1.7002, loss_cns_5: 0.6626, loss_yns_5: 0.1605, loss_cls_dn_0: 0.2183, loss_box_dn_0: 0.7659, loss_cls_dn_1: 0.1340, loss_box_dn_1: 0.7886, loss_cls_dn_2: 0.1339, loss_box_dn_2: 0.7875, loss_cls_dn_3: 0.1379, loss_box_dn_3: 0.7996, loss_cls_dn_4: 0.1418, loss_box_dn_4: 0.8194, loss_cls_dn_5: 0.1544, loss_box_dn_5: 0.8351, loss_dense_depth: 0.8352, loss: 27.0889, grad_norm: 35.7431 -2025-11-12 20:12:22,941 - mmdet - INFO - Iter [199/17500] lr: 1.791e-04, eta: 10:17:25, time: 1.552, data_time: 0.082, memory: 49163, loss_cls_0: 0.8268, loss_box_0: 1.7441, loss_cns_0: 0.6166, loss_yns_0: 0.1575, loss_cls_1: 0.8779, loss_box_1: 1.7354, loss_cns_1: 0.6495, loss_yns_1: 0.1579, loss_cls_2: 0.8988, loss_box_2: 1.7098, loss_cns_2: 0.6557, loss_yns_2: 0.1581, loss_cls_3: 0.9119, loss_box_3: 1.6895, loss_cns_3: 0.6579, loss_yns_3: 0.1632, loss_cls_4: 0.9285, loss_box_4: 1.7101, loss_cns_4: 0.6586, loss_yns_4: 0.1592, loss_cls_5: 0.9264, loss_box_5: 1.7353, loss_cns_5: 0.6529, loss_yns_5: 0.1668, loss_cls_dn_0: 0.2075, loss_box_dn_0: 0.7538, loss_cls_dn_1: 0.1316, loss_box_dn_1: 0.7947, loss_cls_dn_2: 0.1314, loss_box_dn_2: 0.7853, loss_cls_dn_3: 0.1365, loss_box_dn_3: 0.7883, loss_cls_dn_4: 0.1472, loss_box_dn_4: 0.7931, loss_cls_dn_5: 0.1511, loss_box_dn_5: 0.8124, loss_dense_depth: 0.8653, loss: 27.0463, grad_norm: 51.9985 -2025-11-12 20:12:24,474 - mmdet - INFO - Iter [200/17500] lr: 1.795e-04, eta: 10:16:30, time: 1.533, data_time: 0.081, memory: 49163, loss_cls_0: 0.8235, loss_box_0: 1.7418, loss_cns_0: 0.6099, loss_yns_0: 0.1559, loss_cls_1: 0.8956, loss_box_1: 1.7761, loss_cns_1: 0.6465, loss_yns_1: 0.1562, loss_cls_2: 0.9045, loss_box_2: 1.7216, loss_cns_2: 0.6554, loss_yns_2: 0.1550, loss_cls_3: 0.9143, loss_box_3: 1.6891, loss_cns_3: 0.6541, loss_yns_3: 0.1584, loss_cls_4: 0.9381, loss_box_4: 1.6926, loss_cns_4: 0.6574, loss_yns_4: 0.1525, loss_cls_5: 0.9232, loss_box_5: 1.6987, loss_cns_5: 0.6528, loss_yns_5: 0.1546, loss_cls_dn_0: 0.2086, loss_box_dn_0: 0.7600, loss_cls_dn_1: 0.1309, loss_box_dn_1: 0.7899, loss_cls_dn_2: 0.1295, loss_box_dn_2: 0.7669, loss_cls_dn_3: 0.1346, loss_box_dn_3: 0.7610, loss_cls_dn_4: 0.1442, loss_box_dn_4: 0.7611, loss_cls_dn_5: 0.1446, loss_box_dn_5: 0.7676, loss_dense_depth: 0.8412, loss: 26.8680, grad_norm: 34.4243 -2025-11-12 20:12:26,074 - mmdet - INFO - Iter [201/17500] lr: 1.799e-04, eta: 10:15:42, time: 1.600, data_time: 0.146, memory: 49163, loss_cls_0: 0.8118, loss_box_0: 1.7404, loss_cns_0: 0.6126, loss_yns_0: 0.1547, loss_cls_1: 0.9004, loss_box_1: 1.7571, loss_cns_1: 0.6459, loss_yns_1: 0.1558, loss_cls_2: 0.9115, loss_box_2: 1.7506, loss_cns_2: 0.6471, loss_yns_2: 0.1542, loss_cls_3: 0.9167, loss_box_3: 1.7476, loss_cns_3: 0.6453, loss_yns_3: 0.1559, loss_cls_4: 0.9293, loss_box_4: 1.7469, loss_cns_4: 0.6519, loss_yns_4: 0.1568, loss_cls_5: 0.9243, loss_box_5: 1.7258, loss_cns_5: 0.6487, loss_yns_5: 0.1563, loss_cls_dn_0: 0.2078, loss_box_dn_0: 0.7666, loss_cls_dn_1: 0.1246, loss_box_dn_1: 0.7669, loss_cls_dn_2: 0.1246, loss_box_dn_2: 0.7544, loss_cls_dn_3: 0.1279, loss_box_dn_3: 0.7533, loss_cls_dn_4: 0.1338, loss_box_dn_4: 0.7637, loss_cls_dn_5: 0.1422, loss_box_dn_5: 0.7635, loss_dense_depth: 0.8442, loss: 26.9211, grad_norm: 53.4023 -2025-11-12 20:12:27,616 - mmdet - INFO - Iter [202/17500] lr: 1.803e-04, eta: 10:14:49, time: 1.542, data_time: 0.073, memory: 49163, loss_cls_0: 0.7979, loss_box_0: 1.7002, loss_cns_0: 0.6193, loss_yns_0: 0.1552, loss_cls_1: 0.9026, loss_box_1: 1.7223, loss_cns_1: 0.6505, loss_yns_1: 0.1557, loss_cls_2: 0.9069, loss_box_2: 1.7037, loss_cns_2: 0.6495, loss_yns_2: 0.1565, loss_cls_3: 0.9188, loss_box_3: 1.6837, loss_cns_3: 0.6529, loss_yns_3: 0.1550, loss_cls_4: 0.9340, loss_box_4: 1.6939, loss_cns_4: 0.6518, loss_yns_4: 0.1613, loss_cls_5: 0.9275, loss_box_5: 1.6935, loss_cns_5: 0.6501, loss_yns_5: 0.1565, loss_cls_dn_0: 0.2045, loss_box_dn_0: 0.7675, loss_cls_dn_1: 0.1297, loss_box_dn_1: 0.7354, loss_cls_dn_2: 0.1289, loss_box_dn_2: 0.7302, loss_cls_dn_3: 0.1333, loss_box_dn_3: 0.7302, loss_cls_dn_4: 0.1370, loss_box_dn_4: 0.7436, loss_cls_dn_5: 0.1414, loss_box_dn_5: 0.7574, loss_dense_depth: 0.9178, loss: 26.6563, grad_norm: 47.5516 -2025-11-12 20:12:29,210 - mmdet - INFO - Iter [203/17500] lr: 1.807e-04, eta: 10:14:01, time: 1.594, data_time: 0.073, memory: 49163, loss_cls_0: 0.8053, loss_box_0: 1.7109, loss_cns_0: 0.6138, loss_yns_0: 0.1550, loss_cls_1: 0.9042, loss_box_1: 1.7678, loss_cns_1: 0.6492, loss_yns_1: 0.1540, loss_cls_2: 0.9208, loss_box_2: 1.7458, loss_cns_2: 0.6494, loss_yns_2: 0.1529, loss_cls_3: 0.9199, loss_box_3: 1.7273, loss_cns_3: 0.6500, loss_yns_3: 0.1566, loss_cls_4: 0.9346, loss_box_4: 1.7368, loss_cns_4: 0.6494, loss_yns_4: 0.1545, loss_cls_5: 0.9187, loss_box_5: 1.7348, loss_cns_5: 0.6490, loss_yns_5: 0.1567, loss_cls_dn_0: 0.2032, loss_box_dn_0: 0.7689, loss_cls_dn_1: 0.1274, loss_box_dn_1: 0.7491, loss_cls_dn_2: 0.1296, loss_box_dn_2: 0.7562, loss_cls_dn_3: 0.1355, loss_box_dn_3: 0.7573, loss_cls_dn_4: 0.1386, loss_box_dn_4: 0.7614, loss_cls_dn_5: 0.1438, loss_box_dn_5: 0.7736, loss_dense_depth: 0.8315, loss: 26.8935, grad_norm: 54.8556 -2025-11-12 20:12:30,781 - mmdet - INFO - Iter [204/17500] lr: 1.811e-04, eta: 10:13:11, time: 1.572, data_time: 0.099, memory: 49163, loss_cls_0: 0.8040, loss_box_0: 1.7206, loss_cns_0: 0.6202, loss_yns_0: 0.1543, loss_cls_1: 0.8887, loss_box_1: 1.7606, loss_cns_1: 0.6508, loss_yns_1: 0.1532, loss_cls_2: 0.9073, loss_box_2: 1.7494, loss_cns_2: 0.6560, loss_yns_2: 0.1522, loss_cls_3: 0.9051, loss_box_3: 1.7321, loss_cns_3: 0.6529, loss_yns_3: 0.1608, loss_cls_4: 0.9181, loss_box_4: 1.7426, loss_cns_4: 0.6567, loss_yns_4: 0.1521, loss_cls_5: 0.9054, loss_box_5: 1.7239, loss_cns_5: 0.6555, loss_yns_5: 0.1558, loss_cls_dn_0: 0.1996, loss_box_dn_0: 0.7570, loss_cls_dn_1: 0.1223, loss_box_dn_1: 0.7398, loss_cls_dn_2: 0.1245, loss_box_dn_2: 0.7538, loss_cls_dn_3: 0.1328, loss_box_dn_3: 0.7661, loss_cls_dn_4: 0.1343, loss_box_dn_4: 0.7757, loss_cls_dn_5: 0.1415, loss_box_dn_5: 0.7856, loss_dense_depth: 0.9171, loss: 26.9285, grad_norm: 55.5944 -2025-11-12 20:12:32,336 - mmdet - INFO - Iter [205/17500] lr: 1.815e-04, eta: 10:12:21, time: 1.555, data_time: 0.074, memory: 49163, loss_cls_0: 0.8087, loss_box_0: 1.7320, loss_cns_0: 0.6212, loss_yns_0: 0.1537, loss_cls_1: 0.8870, loss_box_1: 1.7333, loss_cns_1: 0.6493, loss_yns_1: 0.1528, loss_cls_2: 0.9100, loss_box_2: 1.7077, loss_cns_2: 0.6555, loss_yns_2: 0.1520, loss_cls_3: 0.9120, loss_box_3: 1.6919, loss_cns_3: 0.6532, loss_yns_3: 0.1573, loss_cls_4: 0.9160, loss_box_4: 1.7209, loss_cns_4: 0.6542, loss_yns_4: 0.1534, loss_cls_5: 0.9293, loss_box_5: 1.7280, loss_cns_5: 0.6522, loss_yns_5: 0.1536, loss_cls_dn_0: 0.2064, loss_box_dn_0: 0.7616, loss_cls_dn_1: 0.1326, loss_box_dn_1: 0.7494, loss_cls_dn_2: 0.1315, loss_box_dn_2: 0.7507, loss_cls_dn_3: 0.1418, loss_box_dn_3: 0.7617, loss_cls_dn_4: 0.1448, loss_box_dn_4: 0.7756, loss_cls_dn_5: 0.1493, loss_box_dn_5: 0.7977, loss_dense_depth: 0.8547, loss: 26.8430, grad_norm: 42.7695 -2025-11-12 20:12:33,878 - mmdet - INFO - Iter [206/17500] lr: 1.819e-04, eta: 10:11:30, time: 1.541, data_time: 0.074, memory: 49163, loss_cls_0: 0.8040, loss_box_0: 1.7263, loss_cns_0: 0.6239, loss_yns_0: 0.1526, loss_cls_1: 0.8996, loss_box_1: 1.7785, loss_cns_1: 0.6519, loss_yns_1: 0.1535, loss_cls_2: 0.9158, loss_box_2: 1.7308, loss_cns_2: 0.6561, loss_yns_2: 0.1534, loss_cls_3: 0.9153, loss_box_3: 1.7242, loss_cns_3: 0.6568, loss_yns_3: 0.1532, loss_cls_4: 0.9233, loss_box_4: 1.7111, loss_cns_4: 0.6584, loss_yns_4: 0.1545, loss_cls_5: 0.9188, loss_box_5: 1.7255, loss_cns_5: 0.6570, loss_yns_5: 0.1525, loss_cls_dn_0: 0.2119, loss_box_dn_0: 0.7637, loss_cls_dn_1: 0.1300, loss_box_dn_1: 0.7537, loss_cls_dn_2: 0.1326, loss_box_dn_2: 0.7439, loss_cls_dn_3: 0.1359, loss_box_dn_3: 0.7515, loss_cls_dn_4: 0.1459, loss_box_dn_4: 0.7410, loss_cls_dn_5: 0.1452, loss_box_dn_5: 0.7590, loss_dense_depth: 0.9155, loss: 26.9268, grad_norm: 45.0954 -2025-11-12 20:12:35,425 - mmdet - INFO - Iter [207/17500] lr: 1.823e-04, eta: 10:10:40, time: 1.547, data_time: 0.076, memory: 49163, loss_cls_0: 0.8072, loss_box_0: 1.7251, loss_cns_0: 0.6162, loss_yns_0: 0.1528, loss_cls_1: 0.8936, loss_box_1: 1.7571, loss_cns_1: 0.6500, loss_yns_1: 0.1519, loss_cls_2: 0.9040, loss_box_2: 1.7210, loss_cns_2: 0.6560, loss_yns_2: 0.1525, loss_cls_3: 0.9174, loss_box_3: 1.7153, loss_cns_3: 0.6553, loss_yns_3: 0.1534, loss_cls_4: 0.9287, loss_box_4: 1.6811, loss_cns_4: 0.6563, loss_yns_4: 0.1529, loss_cls_5: 0.9470, loss_box_5: 1.6741, loss_cns_5: 0.6555, loss_yns_5: 0.1534, loss_cls_dn_0: 0.2104, loss_box_dn_0: 0.7506, loss_cls_dn_1: 0.1360, loss_box_dn_1: 0.7309, loss_cls_dn_2: 0.1367, loss_box_dn_2: 0.7244, loss_cls_dn_3: 0.1390, loss_box_dn_3: 0.7292, loss_cls_dn_4: 0.1497, loss_box_dn_4: 0.7067, loss_cls_dn_5: 0.1481, loss_box_dn_5: 0.7110, loss_dense_depth: 0.8213, loss: 26.5718, grad_norm: 40.1680 -2025-11-12 20:12:36,961 - mmdet - INFO - Iter [208/17500] lr: 1.827e-04, eta: 10:09:49, time: 1.536, data_time: 0.077, memory: 49163, loss_cls_0: 0.8137, loss_box_0: 1.7217, loss_cns_0: 0.6170, loss_yns_0: 0.1569, loss_cls_1: 0.8936, loss_box_1: 1.7561, loss_cns_1: 0.6454, loss_yns_1: 0.1562, loss_cls_2: 0.9069, loss_box_2: 1.7207, loss_cns_2: 0.6511, loss_yns_2: 0.1545, loss_cls_3: 0.9149, loss_box_3: 1.7074, loss_cns_3: 0.6541, loss_yns_3: 0.1547, loss_cls_4: 0.9356, loss_box_4: 1.7152, loss_cns_4: 0.6486, loss_yns_4: 0.1540, loss_cls_5: 0.9427, loss_box_5: 1.7183, loss_cns_5: 0.6475, loss_yns_5: 0.1587, loss_cls_dn_0: 0.2058, loss_box_dn_0: 0.7599, loss_cls_dn_1: 0.1313, loss_box_dn_1: 0.6947, loss_cls_dn_2: 0.1320, loss_box_dn_2: 0.6808, loss_cls_dn_3: 0.1342, loss_box_dn_3: 0.6802, loss_cls_dn_4: 0.1375, loss_box_dn_4: 0.6862, loss_cls_dn_5: 0.1403, loss_box_dn_5: 0.6917, loss_dense_depth: 0.8376, loss: 26.4578, grad_norm: 35.2503 -2025-11-12 20:12:38,494 - mmdet - INFO - Iter [209/17500] lr: 1.831e-04, eta: 10:08:59, time: 1.533, data_time: 0.079, memory: 49163, loss_cls_0: 0.8265, loss_box_0: 1.7624, loss_cns_0: 0.6175, loss_yns_0: 0.1520, loss_cls_1: 0.8934, loss_box_1: 1.7376, loss_cns_1: 0.6415, loss_yns_1: 0.1520, loss_cls_2: 0.9152, loss_box_2: 1.6909, loss_cns_2: 0.6476, loss_yns_2: 0.1523, loss_cls_3: 0.9173, loss_box_3: 1.6865, loss_cns_3: 0.6503, loss_yns_3: 0.1529, loss_cls_4: 0.9393, loss_box_4: 1.7339, loss_cns_4: 0.6517, loss_yns_4: 0.1523, loss_cls_5: 0.9340, loss_box_5: 1.6831, loss_cns_5: 0.6543, loss_yns_5: 0.1579, loss_cls_dn_0: 0.2076, loss_box_dn_0: 0.7694, loss_cls_dn_1: 0.1283, loss_box_dn_1: 0.6988, loss_cls_dn_2: 0.1285, loss_box_dn_2: 0.6761, loss_cls_dn_3: 0.1298, loss_box_dn_3: 0.6793, loss_cls_dn_4: 0.1335, loss_box_dn_4: 0.7047, loss_cls_dn_5: 0.1421, loss_box_dn_5: 0.6932, loss_dense_depth: 0.8696, loss: 26.4633, grad_norm: 44.8647 -2025-11-12 20:12:40,019 - mmdet - INFO - Iter [210/17500] lr: 1.835e-04, eta: 10:08:08, time: 1.525, data_time: 0.078, memory: 49163, loss_cls_0: 0.8260, loss_box_0: 1.7316, loss_cns_0: 0.6254, loss_yns_0: 0.1522, loss_cls_1: 0.8767, loss_box_1: 1.7083, loss_cns_1: 0.6435, loss_yns_1: 0.1509, loss_cls_2: 0.9018, loss_box_2: 1.6587, loss_cns_2: 0.6539, loss_yns_2: 0.1534, loss_cls_3: 0.9187, loss_box_3: 1.6513, loss_cns_3: 0.6540, loss_yns_3: 0.1531, loss_cls_4: 0.9226, loss_box_4: 1.6753, loss_cns_4: 0.6547, loss_yns_4: 0.1521, loss_cls_5: 0.9234, loss_box_5: 1.6440, loss_cns_5: 0.6532, loss_yns_5: 0.1534, loss_cls_dn_0: 0.1986, loss_box_dn_0: 0.7570, loss_cls_dn_1: 0.1247, loss_box_dn_1: 0.6984, loss_cls_dn_2: 0.1243, loss_box_dn_2: 0.6878, loss_cls_dn_3: 0.1253, loss_box_dn_3: 0.6956, loss_cls_dn_4: 0.1325, loss_box_dn_4: 0.7155, loss_cls_dn_5: 0.1378, loss_box_dn_5: 0.7135, loss_dense_depth: 0.7903, loss: 26.1396, grad_norm: 43.9011 -2025-11-12 20:12:41,562 - mmdet - INFO - Iter [211/17500] lr: 1.839e-04, eta: 10:07:20, time: 1.543, data_time: 0.076, memory: 49163, loss_cls_0: 0.7956, loss_box_0: 1.7071, loss_cns_0: 0.6236, loss_yns_0: 0.1526, loss_cls_1: 0.8902, loss_box_1: 1.6604, loss_cns_1: 0.6501, loss_yns_1: 0.1515, loss_cls_2: 0.9051, loss_box_2: 1.6305, loss_cns_2: 0.6559, loss_yns_2: 0.1525, loss_cls_3: 0.9052, loss_box_3: 1.6230, loss_cns_3: 0.6618, loss_yns_3: 0.1515, loss_cls_4: 0.9212, loss_box_4: 1.6358, loss_cns_4: 0.6561, loss_yns_4: 0.1499, loss_cls_5: 0.9105, loss_box_5: 1.6401, loss_cns_5: 0.6629, loss_yns_5: 0.1549, loss_cls_dn_0: 0.1933, loss_box_dn_0: 0.7584, loss_cls_dn_1: 0.1186, loss_box_dn_1: 0.7027, loss_cls_dn_2: 0.1184, loss_box_dn_2: 0.7014, loss_cls_dn_3: 0.1216, loss_box_dn_3: 0.7085, loss_cls_dn_4: 0.1241, loss_box_dn_4: 0.7285, loss_cls_dn_5: 0.1342, loss_box_dn_5: 0.7469, loss_dense_depth: 0.8125, loss: 26.0174, grad_norm: 46.6267 -2025-11-12 20:12:43,096 - mmdet - INFO - Iter [212/17500] lr: 1.843e-04, eta: 10:06:30, time: 1.529, data_time: 0.078, memory: 49163, loss_cls_0: 0.7935, loss_box_0: 1.6956, loss_cns_0: 0.6258, loss_yns_0: 0.1490, loss_cls_1: 0.8702, loss_box_1: 1.6853, loss_cns_1: 0.6528, loss_yns_1: 0.1497, loss_cls_2: 0.9016, loss_box_2: 1.6286, loss_cns_2: 0.6536, loss_yns_2: 0.1473, loss_cls_3: 0.8965, loss_box_3: 1.6353, loss_cns_3: 0.6641, loss_yns_3: 0.1492, loss_cls_4: 0.9156, loss_box_4: 1.6480, loss_cns_4: 0.6568, loss_yns_4: 0.1480, loss_cls_5: 0.9065, loss_box_5: 1.6479, loss_cns_5: 0.6654, loss_yns_5: 0.1513, loss_cls_dn_0: 0.1987, loss_box_dn_0: 0.7604, loss_cls_dn_1: 0.1265, loss_box_dn_1: 0.7291, loss_cls_dn_2: 0.1255, loss_box_dn_2: 0.7221, loss_cls_dn_3: 0.1261, loss_box_dn_3: 0.7258, loss_cls_dn_4: 0.1248, loss_box_dn_4: 0.7463, loss_cls_dn_5: 0.1335, loss_box_dn_5: 0.7595, loss_dense_depth: 0.8103, loss: 26.1261, grad_norm: 43.4355 -2025-11-12 20:12:44,623 - mmdet - INFO - Iter [213/17500] lr: 1.847e-04, eta: 10:05:42, time: 1.532, data_time: 0.082, memory: 49163, loss_cls_0: 0.7946, loss_box_0: 1.7234, loss_cns_0: 0.6257, loss_yns_0: 0.1525, loss_cls_1: 0.8631, loss_box_1: 1.6658, loss_cns_1: 0.6522, loss_yns_1: 0.1503, loss_cls_2: 0.8841, loss_box_2: 1.6173, loss_cns_2: 0.6531, loss_yns_2: 0.1512, loss_cls_3: 0.8954, loss_box_3: 1.6309, loss_cns_3: 0.6561, loss_yns_3: 0.1497, loss_cls_4: 0.9081, loss_box_4: 1.6187, loss_cns_4: 0.6550, loss_yns_4: 0.1517, loss_cls_5: 0.9071, loss_box_5: 1.6415, loss_cns_5: 0.6570, loss_yns_5: 0.1513, loss_cls_dn_0: 0.2049, loss_box_dn_0: 0.7555, loss_cls_dn_1: 0.1296, loss_box_dn_1: 0.7358, loss_cls_dn_2: 0.1291, loss_box_dn_2: 0.7306, loss_cls_dn_3: 0.1292, loss_box_dn_3: 0.7392, loss_cls_dn_4: 0.1312, loss_box_dn_4: 0.7431, loss_cls_dn_5: 0.1399, loss_box_dn_5: 0.7543, loss_dense_depth: 0.8354, loss: 26.1133, grad_norm: 49.2804 -2025-11-12 20:12:46,152 - mmdet - INFO - Iter [214/17500] lr: 1.851e-04, eta: 10:04:53, time: 1.530, data_time: 0.077, memory: 49163, loss_cls_0: 0.8204, loss_box_0: 1.7254, loss_cns_0: 0.6242, loss_yns_0: 0.1523, loss_cls_1: 0.8917, loss_box_1: 1.6753, loss_cns_1: 0.6501, loss_yns_1: 0.1521, loss_cls_2: 0.9180, loss_box_2: 1.6441, loss_cns_2: 0.6535, loss_yns_2: 0.1519, loss_cls_3: 0.9395, loss_box_3: 1.6502, loss_cns_3: 0.6555, loss_yns_3: 0.1540, loss_cls_4: 0.9304, loss_box_4: 1.6149, loss_cns_4: 0.6556, loss_yns_4: 0.1552, loss_cls_5: 0.9246, loss_box_5: 1.6274, loss_cns_5: 0.6546, loss_yns_5: 0.1523, loss_cls_dn_0: 0.2039, loss_box_dn_0: 0.7646, loss_cls_dn_1: 0.1259, loss_box_dn_1: 0.7271, loss_cls_dn_2: 0.1241, loss_box_dn_2: 0.7209, loss_cls_dn_3: 0.1251, loss_box_dn_3: 0.7258, loss_cls_dn_4: 0.1316, loss_box_dn_4: 0.7158, loss_cls_dn_5: 0.1352, loss_box_dn_5: 0.7199, loss_dense_depth: 0.8071, loss: 26.2005, grad_norm: 49.4748 -2025-11-12 20:12:47,733 - mmdet - INFO - Iter [215/17500] lr: 1.855e-04, eta: 10:04:09, time: 1.580, data_time: 0.078, memory: 49163, loss_cls_0: 0.8214, loss_box_0: 1.7220, loss_cns_0: 0.6249, loss_yns_0: 0.1520, loss_cls_1: 0.8839, loss_box_1: 1.6659, loss_cns_1: 0.6465, loss_yns_1: 0.1509, loss_cls_2: 0.9055, loss_box_2: 1.6040, loss_cns_2: 0.6539, loss_yns_2: 0.1525, loss_cls_3: 0.9151, loss_box_3: 1.5943, loss_cns_3: 0.6611, loss_yns_3: 0.1508, loss_cls_4: 0.9383, loss_box_4: 1.6022, loss_cns_4: 0.6550, loss_yns_4: 0.1502, loss_cls_5: 0.9595, loss_box_5: 1.6099, loss_cns_5: 0.6568, loss_yns_5: 0.1503, loss_cls_dn_0: 0.2063, loss_box_dn_0: 0.7553, loss_cls_dn_1: 0.1265, loss_box_dn_1: 0.7005, loss_cls_dn_2: 0.1262, loss_box_dn_2: 0.6854, loss_cls_dn_3: 0.1289, loss_box_dn_3: 0.6822, loss_cls_dn_4: 0.1351, loss_box_dn_4: 0.6845, loss_cls_dn_5: 0.1478, loss_box_dn_5: 0.6924, loss_dense_depth: 0.8428, loss: 25.9409, grad_norm: 35.1431 -2025-11-12 20:12:49,241 - mmdet - INFO - Iter [216/17500] lr: 1.859e-04, eta: 10:03:20, time: 1.508, data_time: 0.078, memory: 49163, loss_cls_0: 0.8364, loss_box_0: 1.7354, loss_cns_0: 0.6276, loss_yns_0: 0.1552, loss_cls_1: 0.8835, loss_box_1: 1.6937, loss_cns_1: 0.6475, loss_yns_1: 0.1524, loss_cls_2: 0.9208, loss_box_2: 1.6386, loss_cns_2: 0.6591, loss_yns_2: 0.1544, loss_cls_3: 0.9164, loss_box_3: 1.6149, loss_cns_3: 0.6651, loss_yns_3: 0.1534, loss_cls_4: 0.9283, loss_box_4: 1.6204, loss_cns_4: 0.6576, loss_yns_4: 0.1530, loss_cls_5: 0.9229, loss_box_5: 1.6339, loss_cns_5: 0.6555, loss_yns_5: 0.1531, loss_cls_dn_0: 0.2109, loss_box_dn_0: 0.7639, loss_cls_dn_1: 0.1287, loss_box_dn_1: 0.7010, loss_cls_dn_2: 0.1381, loss_box_dn_2: 0.6883, loss_cls_dn_3: 0.1410, loss_box_dn_3: 0.6847, loss_cls_dn_4: 0.1367, loss_box_dn_4: 0.6890, loss_cls_dn_5: 0.1410, loss_box_dn_5: 0.7019, loss_dense_depth: 0.8023, loss: 26.1067, grad_norm: 37.5188 -2025-11-12 20:12:50,790 - mmdet - INFO - Iter [217/17500] lr: 1.863e-04, eta: 10:02:35, time: 1.550, data_time: 0.072, memory: 49163, loss_cls_0: 0.7996, loss_box_0: 1.7276, loss_cns_0: 0.6237, loss_yns_0: 0.1556, loss_cls_1: 0.8718, loss_box_1: 1.6531, loss_cns_1: 0.6500, loss_yns_1: 0.1553, loss_cls_2: 0.8981, loss_box_2: 1.6216, loss_cns_2: 0.6532, loss_yns_2: 0.1549, loss_cls_3: 0.9009, loss_box_3: 1.6045, loss_cns_3: 0.6559, loss_yns_3: 0.1556, loss_cls_4: 0.9050, loss_box_4: 1.5996, loss_cns_4: 0.6579, loss_yns_4: 0.1552, loss_cls_5: 0.9070, loss_box_5: 1.6030, loss_cns_5: 0.6570, loss_yns_5: 0.1580, loss_cls_dn_0: 0.2024, loss_box_dn_0: 0.7604, loss_cls_dn_1: 0.1225, loss_box_dn_1: 0.7013, loss_cls_dn_2: 0.1252, loss_box_dn_2: 0.7009, loss_cls_dn_3: 0.1288, loss_box_dn_3: 0.7007, loss_cls_dn_4: 0.1290, loss_box_dn_4: 0.7072, loss_cls_dn_5: 0.1326, loss_box_dn_5: 0.7162, loss_dense_depth: 0.7724, loss: 25.8236, grad_norm: 36.5455 -2025-11-12 20:12:52,323 - mmdet - INFO - Iter [218/17500] lr: 1.867e-04, eta: 10:01:48, time: 1.532, data_time: 0.077, memory: 49163, loss_cls_0: 0.7872, loss_box_0: 1.6989, loss_cns_0: 0.6263, loss_yns_0: 0.1528, loss_cls_1: 0.8689, loss_box_1: 1.6578, loss_cns_1: 0.6562, loss_yns_1: 0.1546, loss_cls_2: 0.9106, loss_box_2: 1.6284, loss_cns_2: 0.6569, loss_yns_2: 0.1566, loss_cls_3: 0.8906, loss_box_3: 1.6119, loss_cns_3: 0.6591, loss_yns_3: 0.1548, loss_cls_4: 0.9021, loss_box_4: 1.6180, loss_cns_4: 0.6576, loss_yns_4: 0.1541, loss_cls_5: 0.9075, loss_box_5: 1.6135, loss_cns_5: 0.6567, loss_yns_5: 0.1571, loss_cls_dn_0: 0.1960, loss_box_dn_0: 0.7645, loss_cls_dn_1: 0.1258, loss_box_dn_1: 0.7056, loss_cls_dn_2: 0.1286, loss_box_dn_2: 0.7016, loss_cls_dn_3: 0.1296, loss_box_dn_3: 0.7005, loss_cls_dn_4: 0.1305, loss_box_dn_4: 0.7110, loss_cls_dn_5: 0.1342, loss_box_dn_5: 0.7199, loss_dense_depth: 0.7517, loss: 25.8378, grad_norm: 33.5584 -2025-11-12 20:12:53,852 - mmdet - INFO - Iter [219/17500] lr: 1.871e-04, eta: 10:01:02, time: 1.531, data_time: 0.080, memory: 49163, loss_cls_0: 0.8091, loss_box_0: 1.7316, loss_cns_0: 0.6184, loss_yns_0: 0.1524, loss_cls_1: 0.8720, loss_box_1: 1.7039, loss_cns_1: 0.6463, loss_yns_1: 0.1518, loss_cls_2: 0.9060, loss_box_2: 1.6636, loss_cns_2: 0.6511, loss_yns_2: 0.1529, loss_cls_3: 0.9078, loss_box_3: 1.6475, loss_cns_3: 0.6542, loss_yns_3: 0.1521, loss_cls_4: 0.9144, loss_box_4: 1.6273, loss_cns_4: 0.6537, loss_yns_4: 0.1520, loss_cls_5: 0.9140, loss_box_5: 1.6303, loss_cns_5: 0.6545, loss_yns_5: 0.1547, loss_cls_dn_0: 0.2056, loss_box_dn_0: 0.7575, loss_cls_dn_1: 0.1276, loss_box_dn_1: 0.7040, loss_cls_dn_2: 0.1289, loss_box_dn_2: 0.6962, loss_cls_dn_3: 0.1318, loss_box_dn_3: 0.6941, loss_cls_dn_4: 0.1344, loss_box_dn_4: 0.6908, loss_cls_dn_5: 0.1376, loss_box_dn_5: 0.6971, loss_dense_depth: 0.8003, loss: 26.0275, grad_norm: 31.7924 -2025-11-12 20:12:55,384 - mmdet - INFO - Iter [220/17500] lr: 1.875e-04, eta: 10:00:16, time: 1.532, data_time: 0.084, memory: 49163, loss_cls_0: 0.7855, loss_box_0: 1.7102, loss_cns_0: 0.6262, loss_yns_0: 0.1546, loss_cls_1: 0.8646, loss_box_1: 1.6867, loss_cns_1: 0.6489, loss_yns_1: 0.1528, loss_cls_2: 0.8710, loss_box_2: 1.6317, loss_cns_2: 0.6537, loss_yns_2: 0.1536, loss_cls_3: 0.8869, loss_box_3: 1.6143, loss_cns_3: 0.6555, loss_yns_3: 0.1524, loss_cls_4: 0.8923, loss_box_4: 1.6057, loss_cns_4: 0.6530, loss_yns_4: 0.1517, loss_cls_5: 0.8942, loss_box_5: 1.6143, loss_cns_5: 0.6571, loss_yns_5: 0.1543, loss_cls_dn_0: 0.2035, loss_box_dn_0: 0.7484, loss_cls_dn_1: 0.1217, loss_box_dn_1: 0.6762, loss_cls_dn_2: 0.1251, loss_box_dn_2: 0.6611, loss_cls_dn_3: 0.1245, loss_box_dn_3: 0.6587, loss_cls_dn_4: 0.1313, loss_box_dn_4: 0.6568, loss_cls_dn_5: 0.1342, loss_box_dn_5: 0.6616, loss_dense_depth: 0.7842, loss: 25.5588, grad_norm: 33.8379 -2025-11-12 20:12:56,992 - mmdet - INFO - Iter [221/17500] lr: 1.879e-04, eta: 9:59:37, time: 1.601, data_time: 0.155, memory: 49163, loss_cls_0: 0.7848, loss_box_0: 1.6987, loss_cns_0: 0.6257, loss_yns_0: 0.1543, loss_cls_1: 0.8606, loss_box_1: 1.7013, loss_cns_1: 0.6484, loss_yns_1: 0.1527, loss_cls_2: 0.8875, loss_box_2: 1.6543, loss_cns_2: 0.6553, loss_yns_2: 0.1522, loss_cls_3: 0.8946, loss_box_3: 1.6347, loss_cns_3: 0.6542, loss_yns_3: 0.1508, loss_cls_4: 0.8952, loss_box_4: 1.6414, loss_cns_4: 0.6548, loss_yns_4: 0.1504, loss_cls_5: 0.9012, loss_box_5: 1.6324, loss_cns_5: 0.6544, loss_yns_5: 0.1532, loss_cls_dn_0: 0.2005, loss_box_dn_0: 0.7527, loss_cls_dn_1: 0.1268, loss_box_dn_1: 0.6690, loss_cls_dn_2: 0.1275, loss_box_dn_2: 0.6520, loss_cls_dn_3: 0.1295, loss_box_dn_3: 0.6530, loss_cls_dn_4: 0.1312, loss_box_dn_4: 0.6607, loss_cls_dn_5: 0.1357, loss_box_dn_5: 0.6619, loss_dense_depth: 0.7678, loss: 25.6613, grad_norm: 25.8164 -2025-11-12 20:12:58,549 - mmdet - INFO - Iter [222/17500] lr: 1.883e-04, eta: 9:58:54, time: 1.564, data_time: 0.081, memory: 49163, loss_cls_0: 0.7904, loss_box_0: 1.7015, loss_cns_0: 0.6235, loss_yns_0: 0.1518, loss_cls_1: 0.8697, loss_box_1: 1.6639, loss_cns_1: 0.6507, loss_yns_1: 0.1501, loss_cls_2: 0.8926, loss_box_2: 1.6214, loss_cns_2: 0.6575, loss_yns_2: 0.1474, loss_cls_3: 0.8912, loss_box_3: 1.6001, loss_cns_3: 0.6548, loss_yns_3: 0.1494, loss_cls_4: 0.9007, loss_box_4: 1.6084, loss_cns_4: 0.6564, loss_yns_4: 0.1484, loss_cls_5: 0.8992, loss_box_5: 1.6069, loss_cns_5: 0.6559, loss_yns_5: 0.1510, loss_cls_dn_0: 0.1980, loss_box_dn_0: 0.7576, loss_cls_dn_1: 0.1255, loss_box_dn_1: 0.6866, loss_cls_dn_2: 0.1276, loss_box_dn_2: 0.6699, loss_cls_dn_3: 0.1300, loss_box_dn_3: 0.6749, loss_cls_dn_4: 0.1317, loss_box_dn_4: 0.6850, loss_cls_dn_5: 0.1364, loss_box_dn_5: 0.6936, loss_dense_depth: 0.8140, loss: 25.6734, grad_norm: 35.2019 -2025-11-12 20:13:00,144 - mmdet - INFO - Iter [223/17500] lr: 1.887e-04, eta: 9:58:14, time: 1.594, data_time: 0.075, memory: 49163, loss_cls_0: 0.8193, loss_box_0: 1.7369, loss_cns_0: 0.6223, loss_yns_0: 0.1537, loss_cls_1: 0.8858, loss_box_1: 1.6942, loss_cns_1: 0.6495, loss_yns_1: 0.1511, loss_cls_2: 0.8983, loss_box_2: 1.6394, loss_cns_2: 0.6528, loss_yns_2: 0.1498, loss_cls_3: 0.9037, loss_box_3: 1.6238, loss_cns_3: 0.6543, loss_yns_3: 0.1520, loss_cls_4: 0.9153, loss_box_4: 1.6397, loss_cns_4: 0.6609, loss_yns_4: 0.1520, loss_cls_5: 0.9066, loss_box_5: 1.6272, loss_cns_5: 0.6598, loss_yns_5: 0.1513, loss_cls_dn_0: 0.2082, loss_box_dn_0: 0.7606, loss_cls_dn_1: 0.1319, loss_box_dn_1: 0.7091, loss_cls_dn_2: 0.1336, loss_box_dn_2: 0.6959, loss_cls_dn_3: 0.1346, loss_box_dn_3: 0.6974, loss_cls_dn_4: 0.1412, loss_box_dn_4: 0.7099, loss_cls_dn_5: 0.1445, loss_box_dn_5: 0.7120, loss_dense_depth: 0.7964, loss: 26.0752, grad_norm: 31.7721 -2025-11-12 20:13:01,722 - mmdet - INFO - Iter [224/17500] lr: 1.891e-04, eta: 9:57:34, time: 1.578, data_time: 0.097, memory: 49163, loss_cls_0: 0.7849, loss_box_0: 1.7139, loss_cns_0: 0.6278, loss_yns_0: 0.1518, loss_cls_1: 0.8704, loss_box_1: 1.6976, loss_cns_1: 0.6526, loss_yns_1: 0.1512, loss_cls_2: 0.9014, loss_box_2: 1.6566, loss_cns_2: 0.6575, loss_yns_2: 0.1520, loss_cls_3: 0.8853, loss_box_3: 1.6316, loss_cns_3: 0.6568, loss_yns_3: 0.1526, loss_cls_4: 0.9029, loss_box_4: 1.6517, loss_cns_4: 0.6587, loss_yns_4: 0.1526, loss_cls_5: 0.8932, loss_box_5: 1.6288, loss_cns_5: 0.6582, loss_yns_5: 0.1532, loss_cls_dn_0: 0.1943, loss_box_dn_0: 0.7533, loss_cls_dn_1: 0.1247, loss_box_dn_1: 0.7062, loss_cls_dn_2: 0.1265, loss_box_dn_2: 0.6931, loss_cls_dn_3: 0.1236, loss_box_dn_3: 0.6868, loss_cls_dn_4: 0.1277, loss_box_dn_4: 0.6992, loss_cls_dn_5: 0.1386, loss_box_dn_5: 0.6944, loss_dense_depth: 0.8086, loss: 25.9202, grad_norm: 29.3641 -2025-11-12 20:13:03,263 - mmdet - INFO - Iter [225/17500] lr: 1.895e-04, eta: 9:56:51, time: 1.542, data_time: 0.079, memory: 49163, loss_cls_0: 0.7769, loss_box_0: 1.7256, loss_cns_0: 0.6246, loss_yns_0: 0.1535, loss_cls_1: 0.8768, loss_box_1: 1.6844, loss_cns_1: 0.6496, loss_yns_1: 0.1522, loss_cls_2: 0.8960, loss_box_2: 1.6527, loss_cns_2: 0.6520, loss_yns_2: 0.1518, loss_cls_3: 0.9021, loss_box_3: 1.6289, loss_cns_3: 0.6516, loss_yns_3: 0.1513, loss_cls_4: 0.9155, loss_box_4: 1.6314, loss_cns_4: 0.6549, loss_yns_4: 0.1506, loss_cls_5: 0.9025, loss_box_5: 1.6247, loss_cns_5: 0.6548, loss_yns_5: 0.1511, loss_cls_dn_0: 0.1899, loss_box_dn_0: 0.7471, loss_cls_dn_1: 0.1225, loss_box_dn_1: 0.6933, loss_cls_dn_2: 0.1264, loss_box_dn_2: 0.6838, loss_cls_dn_3: 0.1286, loss_box_dn_3: 0.6719, loss_cls_dn_4: 0.1335, loss_box_dn_4: 0.6733, loss_cls_dn_5: 0.1336, loss_box_dn_5: 0.6707, loss_dense_depth: 0.8454, loss: 25.8356, grad_norm: 29.6019 -2025-11-12 20:13:04,786 - mmdet - INFO - Iter [226/17500] lr: 1.899e-04, eta: 9:56:07, time: 1.523, data_time: 0.077, memory: 49163, loss_cls_0: 0.8043, loss_box_0: 1.7039, loss_cns_0: 0.6273, loss_yns_0: 0.1529, loss_cls_1: 0.8657, loss_box_1: 1.7261, loss_cns_1: 0.6467, loss_yns_1: 0.1510, loss_cls_2: 0.9014, loss_box_2: 1.6561, loss_cns_2: 0.6481, loss_yns_2: 0.1486, loss_cls_3: 0.8966, loss_box_3: 1.6610, loss_cns_3: 0.6514, loss_yns_3: 0.1502, loss_cls_4: 0.9056, loss_box_4: 1.6520, loss_cns_4: 0.6534, loss_yns_4: 0.1504, loss_cls_5: 0.9154, loss_box_5: 1.6874, loss_cns_5: 0.6522, loss_yns_5: 0.1493, loss_cls_dn_0: 0.1950, loss_box_dn_0: 0.7545, loss_cls_dn_1: 0.1242, loss_box_dn_1: 0.6870, loss_cls_dn_2: 0.1274, loss_box_dn_2: 0.6717, loss_cls_dn_3: 0.1254, loss_box_dn_3: 0.6693, loss_cls_dn_4: 0.1293, loss_box_dn_4: 0.6656, loss_cls_dn_5: 0.1305, loss_box_dn_5: 0.6761, loss_dense_depth: 0.7781, loss: 25.8909, grad_norm: 41.0973 -2025-11-12 20:13:06,299 - mmdet - INFO - Iter [227/17500] lr: 1.903e-04, eta: 9:55:22, time: 1.511, data_time: 0.076, memory: 49163, loss_cls_0: 0.7940, loss_box_0: 1.7235, loss_cns_0: 0.6179, loss_yns_0: 0.1492, loss_cls_1: 0.8681, loss_box_1: 1.7029, loss_cns_1: 0.6394, loss_yns_1: 0.1470, loss_cls_2: 0.9070, loss_box_2: 1.6471, loss_cns_2: 0.6366, loss_yns_2: 0.1452, loss_cls_3: 0.8939, loss_box_3: 1.6690, loss_cns_3: 0.6476, loss_yns_3: 0.1492, loss_cls_4: 0.8961, loss_box_4: 1.6597, loss_cns_4: 0.6487, loss_yns_4: 0.1488, loss_cls_5: 0.8910, loss_box_5: 1.6789, loss_cns_5: 0.6493, loss_yns_5: 0.1476, loss_cls_dn_0: 0.1964, loss_box_dn_0: 0.7623, loss_cls_dn_1: 0.1237, loss_box_dn_1: 0.6994, loss_cls_dn_2: 0.1294, loss_box_dn_2: 0.6938, loss_cls_dn_3: 0.1240, loss_box_dn_3: 0.6900, loss_cls_dn_4: 0.1274, loss_box_dn_4: 0.6907, loss_cls_dn_5: 0.1304, loss_box_dn_5: 0.7037, loss_dense_depth: 0.8126, loss: 25.9410, grad_norm: 42.1737 -2025-11-12 20:13:07,817 - mmdet - INFO - Iter [228/17500] lr: 1.907e-04, eta: 9:54:38, time: 1.520, data_time: 0.074, memory: 49163, loss_cls_0: 0.7958, loss_box_0: 1.7151, loss_cns_0: 0.6229, loss_yns_0: 0.1509, loss_cls_1: 0.8608, loss_box_1: 1.7073, loss_cns_1: 0.6471, loss_yns_1: 0.1491, loss_cls_2: 0.8873, loss_box_2: 1.6665, loss_cns_2: 0.6439, loss_yns_2: 0.1476, loss_cls_3: 0.9033, loss_box_3: 1.6732, loss_cns_3: 0.6548, loss_yns_3: 0.1537, loss_cls_4: 0.8970, loss_box_4: 1.6704, loss_cns_4: 0.6547, loss_yns_4: 0.1518, loss_cls_5: 0.8853, loss_box_5: 1.6716, loss_cns_5: 0.6551, loss_yns_5: 0.1503, loss_cls_dn_0: 0.1974, loss_box_dn_0: 0.7521, loss_cls_dn_1: 0.1226, loss_box_dn_1: 0.7064, loss_cls_dn_2: 0.1233, loss_box_dn_2: 0.7009, loss_cls_dn_3: 0.1282, loss_box_dn_3: 0.6951, loss_cls_dn_4: 0.1265, loss_box_dn_4: 0.7009, loss_cls_dn_5: 0.1264, loss_box_dn_5: 0.7096, loss_dense_depth: 0.7837, loss: 25.9885, grad_norm: 33.3405 -2025-11-12 20:13:09,344 - mmdet - INFO - Iter [229/17500] lr: 1.911e-04, eta: 9:53:56, time: 1.527, data_time: 0.078, memory: 49163, loss_cls_0: 0.7970, loss_box_0: 1.7067, loss_cns_0: 0.6199, loss_yns_0: 0.1472, loss_cls_1: 0.8572, loss_box_1: 1.6769, loss_cns_1: 0.6500, loss_yns_1: 0.1465, loss_cls_2: 0.8832, loss_box_2: 1.6595, loss_cns_2: 0.6530, loss_yns_2: 0.1475, loss_cls_3: 0.8849, loss_box_3: 1.6486, loss_cns_3: 0.6541, loss_yns_3: 0.1477, loss_cls_4: 0.8955, loss_box_4: 1.6438, loss_cns_4: 0.6549, loss_yns_4: 0.1477, loss_cls_5: 0.9034, loss_box_5: 1.6420, loss_cns_5: 0.6528, loss_yns_5: 0.1470, loss_cls_dn_0: 0.1897, loss_box_dn_0: 0.7414, loss_cls_dn_1: 0.1186, loss_box_dn_1: 0.7080, loss_cls_dn_2: 0.1199, loss_box_dn_2: 0.7054, loss_cls_dn_3: 0.1201, loss_box_dn_3: 0.7078, loss_cls_dn_4: 0.1257, loss_box_dn_4: 0.7120, loss_cls_dn_5: 0.1315, loss_box_dn_5: 0.7199, loss_dense_depth: 0.7790, loss: 25.8459, grad_norm: 41.0848 -2025-11-12 20:13:10,861 - mmdet - INFO - Iter [230/17500] lr: 1.915e-04, eta: 9:53:13, time: 1.517, data_time: 0.077, memory: 49163, loss_cls_0: 0.7991, loss_box_0: 1.7406, loss_cns_0: 0.6233, loss_yns_0: 0.1509, loss_cls_1: 0.8552, loss_box_1: 1.6985, loss_cns_1: 0.6493, loss_yns_1: 0.1468, loss_cls_2: 0.8916, loss_box_2: 1.6717, loss_cns_2: 0.6530, loss_yns_2: 0.1478, loss_cls_3: 0.8924, loss_box_3: 1.6385, loss_cns_3: 0.6511, loss_yns_3: 0.1481, loss_cls_4: 0.8936, loss_box_4: 1.6231, loss_cns_4: 0.6548, loss_yns_4: 0.1485, loss_cls_5: 0.8980, loss_box_5: 1.6524, loss_cns_5: 0.6571, loss_yns_5: 0.1473, loss_cls_dn_0: 0.1922, loss_box_dn_0: 0.7526, loss_cls_dn_1: 0.1170, loss_box_dn_1: 0.7065, loss_cls_dn_2: 0.1173, loss_box_dn_2: 0.7026, loss_cls_dn_3: 0.1181, loss_box_dn_3: 0.7035, loss_cls_dn_4: 0.1211, loss_box_dn_4: 0.7053, loss_cls_dn_5: 0.1262, loss_box_dn_5: 0.7184, loss_dense_depth: 0.7635, loss: 25.8770, grad_norm: 34.7219 -2025-11-12 20:13:12,387 - mmdet - INFO - Iter [231/17500] lr: 1.919e-04, eta: 9:52:31, time: 1.526, data_time: 0.074, memory: 49163, loss_cls_0: 0.7868, loss_box_0: 1.7075, loss_cns_0: 0.6254, loss_yns_0: 0.1508, loss_cls_1: 0.8453, loss_box_1: 1.6460, loss_cns_1: 0.6494, loss_yns_1: 0.1467, loss_cls_2: 0.8975, loss_box_2: 1.6076, loss_cns_2: 0.6547, loss_yns_2: 0.1450, loss_cls_3: 0.8647, loss_box_3: 1.5780, loss_cns_3: 0.6545, loss_yns_3: 0.1472, loss_cls_4: 0.8782, loss_box_4: 1.5916, loss_cns_4: 0.6563, loss_yns_4: 0.1477, loss_cls_5: 0.8794, loss_box_5: 1.5965, loss_cns_5: 0.6601, loss_yns_5: 0.1456, loss_cls_dn_0: 0.1858, loss_box_dn_0: 0.7480, loss_cls_dn_1: 0.1161, loss_box_dn_1: 0.6897, loss_cls_dn_2: 0.1167, loss_box_dn_2: 0.6779, loss_cls_dn_3: 0.1168, loss_box_dn_3: 0.6777, loss_cls_dn_4: 0.1184, loss_box_dn_4: 0.6857, loss_cls_dn_5: 0.1228, loss_box_dn_5: 0.6933, loss_dense_depth: 0.7343, loss: 25.3457, grad_norm: 37.4588 diff --git a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20251112_200501.log.json b/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20251112_200501.log.json deleted file mode 100644 index db2b845e5f0bf01d7f85ce9eae9fcd1fdc7da4e0..0000000000000000000000000000000000000000 --- a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20251112_200501.log.json +++ /dev/null @@ -1,232 +0,0 @@ -{"env_info": "sys.platform: linux\nPython: 3.10.12 (main, May 27 2025, 17:12:29) [GCC 11.4.0]\nCUDA available: True\nGPU 0,1,2,3,4,5,6,7: BW200, UBB BW1000\nCUDA_HOME: /opt/dtk\nNVCC: Not Available\nGCC: x86_64-linux-gnu-gcc (Ubuntu 11.4.0-1ubuntu1~22.04) 11.4.0\nPyTorch: 2.4.1\nPyTorch compiling details: PyTorch built with:\n - GCC 10.3\n - C++ Version: 201703\n - Intel(R) Math Kernel Library Version 2020.0.4 Product Build 20200917 for Intel(R) 64 architecture applications\n - OpenMP 201511 (a.k.a. OpenMP 4.5)\n - LAPACK is enabled (usually provided by MKL)\n - NNPACK is enabled\n - CPU capability usage: AVX512\n - HIP Runtime 6.3.25211\n - MIOpen 2.17.0\n - Magma 2.8.0\n - Build settings: BLAS_INFO=mkl, BUILD_TYPE=Release, CXX_COMPILER=/opt/rh/gcc-toolset-10/root/usr/bin/c++, CXX_FLAGS= -D_GLIBCXX_USE_CXX11_ABI=1 -fvisibility-inlines-hidden -DUSE_PTHREADPOOL -DNDEBUG -DUSE_KINETO -DLIBKINETO_NOCUPTI -DUSE_FBGEMM -DUSE_PYTORCH_QNNPACK -DUSE_XNNPACK -DSYMBOLICATE_MOBILE_DEBUG_HANDLE -O2 -fPIC -Wall -Wextra -Werror=return-type -Werror=non-virtual-dtor -Werror=bool-operation -Wnarrowing -Wno-missing-field-initializers -Wno-type-limits -Wno-array-bounds -Wno-unknown-pragmas -Wno-unused-parameter -Wno-unused-function -Wno-unused-result -Wno-strict-overflow -Wno-strict-aliasing -Wno-stringop-overflow -Wsuggest-override -Wno-psabi -Wno-error=pedantic -Wno-error=old-style-cast -Wno-missing-braces -fdiagnostics-color=always -faligned-new -Wno-unused-but-set-variable -Wno-maybe-uninitialized -fno-math-errno -fno-trapping-math -Werror=format -Wno-stringop-overflow, FORCE_FALLBACK_CUDA_MPI=1, LAPACK_INFO=mkl, PERF_WITH_AVX=1, PERF_WITH_AVX2=1, PERF_WITH_AVX512=1, TORCH_VERSION=2.4.1, USE_CUDA=0, USE_CUDNN=OFF, USE_CUSPARSELT=OFF, USE_EXCEPTION_PTR=1, USE_GFLAGS=1, USE_GLOG=1, USE_GLOO=1, USE_MKL=ON, USE_MKLDNN=0, USE_MPI=1, USE_NCCL=1, USE_NNPACK=ON, USE_OPENMP=1, USE_ROCM=ON, USE_ROCM_KERNEL_ASSERT=OFF, \n\nTorchVision: 0.19.1\nOpenCV: 4.11.0\nMMCV: 1.6.1\nMMCV Compiler: GCC 11.4\nMMCV CUDA Compiler: rocm not available\nMMDetection: 2.26.0+c41df4b", "config": "plugin = True\nplugin_dir = 'projects/mmdet3d_plugin/'\ndist_params = dict(backend='nccl')\nlog_level = 'INFO'\nwork_dir = './work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704'\ntotal_batch_size = 160\nnum_gpus = 8\nbatch_size = 20\nnum_iters_per_epoch = 175\nnum_epochs = 100\ncheckpoint_epoch_interval = 20\ncheckpoint_config = dict(interval=3500)\nlog_config = dict(\n interval=1,\n hooks=[\n dict(type='TextLoggerHook', by_epoch=False),\n dict(type='TensorboardLoggerHook')\n ])\nload_from = None\nresume_from = None\nworkflow = [('train', 1)]\nfp16 = dict(loss_scale=32.0)\ninput_shape = (704, 256)\ntracking_test = True\ntracking_threshold = 0.2\nclass_names = [\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier',\n 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n]\nnum_classes = 10\nembed_dims = 256\nnum_groups = 8\nnum_decoder = 6\nnum_single_frame_decoder = 1\nuse_deformable_func = True\nstrides = [4, 8, 16, 32]\nnum_levels = 4\nnum_depth_layers = 3\ndrop_out = 0.1\ntemporal = True\ndecouple_attn = True\nwith_quality_estimation = True\nmodel = dict(\n type='Sparse4D',\n use_grid_mask=True,\n use_deformable_func=True,\n img_backbone=dict(\n type='ResNet',\n depth=50,\n num_stages=4,\n frozen_stages=-1,\n norm_eval=False,\n style='pytorch',\n with_cp=True,\n out_indices=(0, 1, 2, 3),\n norm_cfg=dict(type='BN', requires_grad=True),\n pretrained='ckpt/resnet50-19c8e357.pth'),\n img_neck=dict(\n type='FPN',\n num_outs=4,\n start_level=0,\n out_channels=256,\n add_extra_convs='on_output',\n relu_before_extra_convs=True,\n in_channels=[256, 512, 1024, 2048]),\n depth_branch=dict(\n type='DenseDepthNet',\n embed_dims=256,\n num_depth_layers=3,\n loss_weight=0.2),\n head=dict(\n type='Sparse4DHead',\n cls_threshold_to_reg=0.05,\n decouple_attn=True,\n instance_bank=dict(\n type='InstanceBank',\n num_anchor=900,\n embed_dims=256,\n anchor='nuscenes_kmeans900.npy',\n anchor_handler=dict(type='SparseBox3DKeyPointsGenerator'),\n num_temp_instances=600,\n confidence_decay=0.6,\n feat_grad=False),\n anchor_encoder=dict(\n type='SparseBox3DEncoder',\n vel_dims=3,\n embed_dims=[128, 32, 32, 64],\n mode='cat',\n output_fc=False,\n in_loops=1,\n out_loops=4),\n num_single_frame_decoder=1,\n operation_order=[\n 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm',\n 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm',\n 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm',\n 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm',\n 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm',\n 'deformable', 'ffn', 'norm', 'refine'\n ],\n temp_graph_model=dict(\n type='MultiheadAttention',\n embed_dims=512,\n num_heads=8,\n batch_first=True,\n dropout=0.1),\n graph_model=dict(\n type='MultiheadAttention',\n embed_dims=512,\n num_heads=8,\n batch_first=True,\n dropout=0.1),\n norm_layer=dict(type='LN', normalized_shape=256),\n ffn=dict(\n type='AsymmetricFFN',\n in_channels=512,\n pre_norm=dict(type='LN'),\n embed_dims=256,\n feedforward_channels=1024,\n num_fcs=2,\n ffn_drop=0.1,\n act_cfg=dict(type='ReLU', inplace=True)),\n deformable_model=dict(\n type='DeformableFeatureAggregation',\n embed_dims=256,\n num_groups=8,\n num_levels=4,\n num_cams=6,\n attn_drop=0.15,\n use_deformable_func=True,\n use_camera_embed=True,\n residual_mode='cat',\n kps_generator=dict(\n type='SparseBox3DKeyPointsGenerator',\n num_learnable_pts=6,\n fix_scale=[[0, 0, 0], [0.45, 0, 0], [-0.45, 0, 0],\n [0, 0.45, 0], [0, -0.45, 0], [0, 0, 0.45],\n [0, 0, -0.45]])),\n refine_layer=dict(\n type='SparseBox3DRefinementModule',\n embed_dims=256,\n num_cls=10,\n refine_yaw=True,\n with_quality_estimation=True),\n sampler=dict(\n type='SparseBox3DTarget',\n num_dn_groups=5,\n num_temp_dn_groups=3,\n dn_noise_scale=[2.0, 2.0, 2.0, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5],\n max_dn_gt=32,\n add_neg_dn=True,\n cls_weight=2.0,\n box_weight=0.25,\n reg_weights=[2.0, 2.0, 2.0, 0.5, 0.5, 0.5, 0.0, 0.0, 0.0, 0.0],\n cls_wise_reg_weights=dict(\n {9: [2.0, 2.0, 2.0, 1.0, 1.0, 1.0, 0.0, 0.0, 1.0, 1.0]})),\n loss_cls=dict(\n type='FocalLoss',\n use_sigmoid=True,\n gamma=2.0,\n alpha=0.25,\n loss_weight=2.0),\n loss_reg=dict(\n type='SparseBox3DLoss',\n loss_box=dict(type='L1Loss', loss_weight=0.25),\n loss_centerness=dict(type='CrossEntropyLoss', use_sigmoid=True),\n loss_yawness=dict(type='GaussianFocalLoss'),\n cls_allow_reverse=[5]),\n decoder=dict(type='SparseBox3DDecoder'),\n reg_weights=[2.0, 2.0, 2.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0]))\ndataset_type = 'NuScenes3DDetTrackDataset'\ndata_root = 'data/nuscenes/'\nanno_root = 'data/nuscenes_anno_pkls/'\nfile_client_args = dict(backend='disk')\nimg_norm_cfg = dict(\n mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375], to_rgb=True)\ntrain_pipeline = [\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(\n type='LoadPointsFromFile',\n coord_type='LIDAR',\n load_dim=5,\n use_dim=5,\n file_client_args=dict(backend='disk')),\n dict(type='ResizeCropFlipImage'),\n dict(type='MultiScaleDepthMapGenerator', downsample=[4, 8, 16]),\n dict(type='BBoxRotation'),\n dict(type='PhotoMetricDistortionMultiViewImage'),\n dict(\n type='NormalizeMultiviewImage',\n mean=[123.675, 116.28, 103.53],\n std=[58.395, 57.12, 57.375],\n to_rgb=True),\n dict(\n type='CircleObjectRangeFilter',\n class_dist_thred=[55, 55, 55, 55, 55, 55, 55, 55, 55, 55]),\n dict(\n type='InstanceNameFilter',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ]),\n dict(type='NuScenesSparse4DAdaptor'),\n dict(\n type='Collect',\n keys=[\n 'img', 'timestamp', 'projection_mat', 'image_wh', 'gt_depth',\n 'focal', 'gt_bboxes_3d', 'gt_labels_3d'\n ],\n meta_keys=['T_global', 'T_global_inv', 'timestamp', 'instance_id'])\n]\ntest_pipeline = [\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(type='ResizeCropFlipImage'),\n dict(\n type='NormalizeMultiviewImage',\n mean=[123.675, 116.28, 103.53],\n std=[58.395, 57.12, 57.375],\n to_rgb=True),\n dict(type='NuScenesSparse4DAdaptor'),\n dict(\n type='Collect',\n keys=['img', 'timestamp', 'projection_mat', 'image_wh'],\n meta_keys=['T_global', 'T_global_inv', 'timestamp'])\n]\ninput_modality = dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False)\ndata_basic_config = dict(\n type='NuScenes3DDetTrackDataset',\n data_root='data/nuscenes/',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier',\n 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ],\n modality=dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False),\n version='v1.0-trainval')\ndata_aug_conf = dict(\n resize_lim=(0.4, 0.47),\n final_dim=(256, 704),\n bot_pct_lim=(0.0, 0.0),\n rot_lim=(-5.4, 5.4),\n H=900,\n W=1600,\n rand_flip=True,\n rot3d_range=[-0.3925, 0.3925])\ndata = dict(\n samples_per_gpu=20,\n workers_per_gpu=20,\n train=dict(\n type='NuScenes3DDetTrackDataset',\n data_root='data/nuscenes/',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ],\n modality=dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False),\n version='v1.0-trainval',\n ann_file='data/nuscenes_anno_pkls/nuscenes_infos_train.pkl',\n pipeline=[\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(\n type='LoadPointsFromFile',\n coord_type='LIDAR',\n load_dim=5,\n use_dim=5,\n file_client_args=dict(backend='disk')),\n dict(type='ResizeCropFlipImage'),\n dict(type='MultiScaleDepthMapGenerator', downsample=[4, 8, 16]),\n dict(type='BBoxRotation'),\n dict(type='PhotoMetricDistortionMultiViewImage'),\n dict(\n type='NormalizeMultiviewImage',\n mean=[123.675, 116.28, 103.53],\n std=[58.395, 57.12, 57.375],\n to_rgb=True),\n dict(\n type='CircleObjectRangeFilter',\n class_dist_thred=[55, 55, 55, 55, 55, 55, 55, 55, 55, 55]),\n dict(\n type='InstanceNameFilter',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian',\n 'traffic_cone'\n ]),\n dict(type='NuScenesSparse4DAdaptor'),\n dict(\n type='Collect',\n keys=[\n 'img', 'timestamp', 'projection_mat', 'image_wh',\n 'gt_depth', 'focal', 'gt_bboxes_3d', 'gt_labels_3d'\n ],\n meta_keys=[\n 'T_global', 'T_global_inv', 'timestamp', 'instance_id'\n ])\n ],\n test_mode=False,\n data_aug_conf=dict(\n resize_lim=(0.4, 0.47),\n final_dim=(256, 704),\n bot_pct_lim=(0.0, 0.0),\n rot_lim=(-5.4, 5.4),\n H=900,\n W=1600,\n rand_flip=True,\n rot3d_range=[-0.3925, 0.3925]),\n with_seq_flag=True,\n sequences_split_num=2,\n keep_consistent_seq_aug=True),\n val=dict(\n type='NuScenes3DDetTrackDataset',\n data_root='data/nuscenes/',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ],\n modality=dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False),\n version='v1.0-trainval',\n ann_file='data/nuscenes_anno_pkls/nuscenes_infos_val.pkl',\n pipeline=[\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(type='ResizeCropFlipImage'),\n dict(\n type='NormalizeMultiviewImage',\n mean=[123.675, 116.28, 103.53],\n std=[58.395, 57.12, 57.375],\n to_rgb=True),\n dict(type='NuScenesSparse4DAdaptor'),\n dict(\n type='Collect',\n keys=['img', 'timestamp', 'projection_mat', 'image_wh'],\n meta_keys=['T_global', 'T_global_inv', 'timestamp'])\n ],\n data_aug_conf=dict(\n resize_lim=(0.4, 0.47),\n final_dim=(256, 704),\n bot_pct_lim=(0.0, 0.0),\n rot_lim=(-5.4, 5.4),\n H=900,\n W=1600,\n rand_flip=True,\n rot3d_range=[-0.3925, 0.3925]),\n test_mode=True,\n tracking=True,\n tracking_threshold=0.2),\n test=dict(\n type='NuScenes3DDetTrackDataset',\n data_root='data/nuscenes/',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ],\n modality=dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False),\n version='v1.0-trainval',\n ann_file='data/nuscenes_anno_pkls/nuscenes_infos_val.pkl',\n pipeline=[\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(type='ResizeCropFlipImage'),\n dict(\n type='NormalizeMultiviewImage',\n mean=[123.675, 116.28, 103.53],\n std=[58.395, 57.12, 57.375],\n to_rgb=True),\n dict(type='NuScenesSparse4DAdaptor'),\n dict(\n type='Collect',\n keys=['img', 'timestamp', 'projection_mat', 'image_wh'],\n meta_keys=['T_global', 'T_global_inv', 'timestamp'])\n ],\n data_aug_conf=dict(\n resize_lim=(0.4, 0.47),\n final_dim=(256, 704),\n bot_pct_lim=(0.0, 0.0),\n rot_lim=(-5.4, 5.4),\n H=900,\n W=1600,\n rand_flip=True,\n rot3d_range=[-0.3925, 0.3925]),\n test_mode=True,\n tracking=True,\n tracking_threshold=0.2))\noptimizer = dict(\n type='Miopen_AdamW',\n lr=0.0006,\n weight_decay=0.001,\n paramwise_cfg=dict(custom_keys=dict(img_backbone=dict(lr_mult=0.5))))\noptimizer_config = dict(grad_clip=dict(max_norm=25, norm_type=2))\nlr_config = dict(\n policy='CosineAnnealing',\n warmup='linear',\n warmup_iters=500,\n warmup_ratio=0.3333333333333333,\n min_lr_ratio=0.001)\nrunner = dict(type='IterBasedRunner', max_iters=17500)\nvis_pipeline = [\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(type='Collect', keys=['img'], meta_keys=['timestamp', 'lidar2img'])\n]\nevaluation = dict(\n interval=3500,\n pipeline=[\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(\n type='Collect', keys=['img'], meta_keys=['timestamp', 'lidar2img'])\n ])\ngpu_ids = range(0, 8)\n", "seed": 0, "exp_name": "sparse4dv3_temporal_r50_1x8_bs6_256x704.py"} -{"mode": "train", "epoch": 1, "iter": 1, "lr": 0.0001, "memory": 49163, "data_time": 10.76317, "loss_cls_0": 2.36127, "loss_box_0": 0.01384, "loss_cns_0": 0.0027, "loss_yns_0": 0.00079, "loss_cls_1": 2.15442, "loss_box_1": 0.10777, "loss_cns_1": 0.02453, "loss_yns_1": 0.00663, "loss_cls_2": 2.31215, "loss_box_2": 0.00415, "loss_cns_2": 0.00052, "loss_yns_2": 0.00028, "loss_cls_3": 2.39054, "loss_box_3": 0.02944, "loss_cns_3": 0.00504, "loss_yns_3": 0.00144, "loss_cls_4": 2.02859, "loss_box_4": 0.41339, "loss_cns_4": 0.05305, "loss_yns_4": 0.02503, "loss_cls_5": 2.42502, "loss_box_5": 0.01903, "loss_cns_5": 0.00238, "loss_yns_5": 0.00168, "loss_cls_dn_0": 1.19803, "loss_box_dn_0": 1.46028, "loss_cls_dn_1": 1.11019, "loss_box_dn_1": 1.73177, "loss_cls_dn_2": 1.17413, "loss_box_dn_2": 1.97183, "loss_cls_dn_3": 1.17207, "loss_box_dn_3": 2.2418, "loss_cls_dn_4": 1.05279, "loss_box_dn_4": 2.42684, "loss_cls_dn_5": 1.23869, "loss_box_dn_5": 2.67728, "loss_dense_depth": 1.86433, "loss": 35.70368, "grad_norm": 273.3042, "time": 116.35982} -{"mode": "train", "epoch": 1, "iter": 2, "lr": 0.0001, "memory": 49163, "data_time": 0.10319, "loss_cls_0": 2.09103, "loss_box_0": 0.02337, "loss_cns_0": 0.00603, "loss_yns_0": 0.0023, "loss_cls_1": 2.08753, "loss_box_1": 0.07899, "loss_cns_1": 0.01735, "loss_yns_1": 0.00455, "loss_cls_2": 2.17393, "loss_box_2": 0.09441, "loss_cns_2": 0.0084, "loss_yns_2": 0.00387, "loss_cls_3": 2.02725, "loss_box_3": 0.13716, "loss_cns_3": 0.02054, "loss_yns_3": 0.00639, "loss_cls_4": 1.84159, "loss_box_4": 0.97269, "loss_cns_4": 0.1054, "loss_yns_4": 0.03711, "loss_cls_5": 2.12375, "loss_box_5": 0.2947, "loss_cns_5": 0.03336, "loss_yns_5": 0.01167, "loss_cls_dn_0": 1.05126, "loss_box_dn_0": 1.25399, "loss_cls_dn_1": 0.97791, "loss_box_dn_1": 2.39234, "loss_cls_dn_2": 0.99791, "loss_box_dn_2": 2.47317, "loss_cls_dn_3": 0.94585, "loss_box_dn_3": 2.50395, "loss_cls_dn_4": 0.85869, "loss_box_dn_4": 2.65493, "loss_cls_dn_5": 1.01063, "loss_box_dn_5": 2.8527, "loss_dense_depth": 1.72659, "loss": 35.90331, "grad_norm": 60.80439, "time": 1.96369} -{"mode": "train", "epoch": 1, "iter": 3, "lr": 0.0001, "memory": 49163, "data_time": 0.0764, "loss_cls_0": 1.52841, "loss_box_0": 2.42174, "loss_cns_0": 0.58587, "loss_yns_0": 0.22381, "loss_cls_1": 1.84158, "loss_box_1": 0.92298, "loss_cns_1": 0.16717, "loss_yns_1": 0.05509, "loss_cls_2": 1.8635, "loss_box_2": 2.88871, "loss_cns_2": 0.28307, "loss_yns_2": 0.1489, "loss_cls_3": 1.66275, "loss_box_3": 3.91452, "loss_cns_3": 0.43836, "loss_yns_3": 0.18313, "loss_cls_4": 1.61744, "loss_box_4": 3.2098, "loss_cns_4": 0.33872, "loss_yns_4": 0.1381, "loss_cls_5": 1.7637, "loss_box_5": 1.75249, "loss_cns_5": 0.16265, "loss_yns_5": 0.06263, "loss_cls_dn_0": 0.74727, "loss_box_dn_0": 1.18402, "loss_cls_dn_1": 0.86585, "loss_box_dn_1": 2.19392, "loss_cls_dn_2": 0.84358, "loss_box_dn_2": 2.3596, "loss_cls_dn_3": 0.74477, "loss_box_dn_3": 2.54183, "loss_cls_dn_4": 0.74646, "loss_box_dn_4": 2.77949, "loss_cls_dn_5": 0.83801, "loss_box_dn_5": 2.97637, "loss_dense_depth": 1.73041, "loss": 48.72673, "grad_norm": 90.00634, "time": 1.54339} -{"mode": "train", "epoch": 1, "iter": 4, "lr": 0.0001, "memory": 49163, "data_time": 0.09048, "loss_cls_0": 1.37404, "loss_box_0": 2.52414, "loss_cns_0": 0.57203, "loss_yns_0": 0.17669, "loss_cls_1": 1.68263, "loss_box_1": 2.57356, "loss_cns_1": 0.4099, "loss_yns_1": 0.15448, "loss_cls_2": 1.72565, "loss_box_2": 3.52183, "loss_cns_2": 0.43784, "loss_yns_2": 0.18651, "loss_cls_3": 1.53189, "loss_box_3": 3.92771, "loss_cns_3": 0.48516, "loss_yns_3": 0.2096, "loss_cls_4": 1.46234, "loss_box_4": 4.43698, "loss_cns_4": 0.39988, "loss_yns_4": 0.19405, "loss_cls_5": 1.56176, "loss_box_5": 4.20816, "loss_cns_5": 0.44309, "loss_yns_5": 0.17792, "loss_cls_dn_0": 0.60437, "loss_box_dn_0": 1.17945, "loss_cls_dn_1": 0.76453, "loss_box_dn_1": 2.33042, "loss_cls_dn_2": 0.72294, "loss_box_dn_2": 2.39414, "loss_cls_dn_3": 0.63451, "loss_box_dn_3": 2.56911, "loss_cls_dn_4": 0.62795, "loss_box_dn_4": 2.76748, "loss_cls_dn_5": 0.70712, "loss_box_dn_5": 2.93335, "loss_dense_depth": 1.6199, "loss": 54.2331, "grad_norm": 100.73523, "time": 1.52345} -{"mode": "train", "epoch": 1, "iter": 5, "lr": 0.0001, "memory": 49163, "data_time": 0.07826, "loss_cls_0": 1.36964, "loss_box_0": 2.71704, "loss_cns_0": 0.51371, "loss_yns_0": 0.18924, "loss_cls_1": 1.56502, "loss_box_1": 3.60893, "loss_cns_1": 0.42833, "loss_yns_1": 0.20037, "loss_cls_2": 1.68565, "loss_box_2": 3.75816, "loss_cns_2": 0.40204, "loss_yns_2": 0.19447, "loss_cls_3": 1.49137, "loss_box_3": 3.84093, "loss_cns_3": 0.41512, "loss_yns_3": 0.19156, "loss_cls_4": 1.42114, "loss_box_4": 4.01098, "loss_cns_4": 0.38982, "loss_yns_4": 0.19142, "loss_cls_5": 1.40777, "loss_box_5": 4.17091, "loss_cns_5": 0.42864, "loss_yns_5": 0.19709, "loss_cls_dn_0": 0.5291, "loss_box_dn_0": 1.21598, "loss_cls_dn_1": 0.69233, "loss_box_dn_1": 2.08779, "loss_cls_dn_2": 0.66454, "loss_box_dn_2": 2.1535, "loss_cls_dn_3": 0.5719, "loss_box_dn_3": 2.25815, "loss_cls_dn_4": 0.56053, "loss_box_dn_4": 2.40894, "loss_cls_dn_5": 0.60583, "loss_box_dn_5": 2.50081, "loss_dense_depth": 1.57832, "loss": 52.61709, "grad_norm": 100.97372, "time": 1.52265} -{"mode": "train", "epoch": 1, "iter": 6, "lr": 0.0001, "memory": 49163, "data_time": 0.07988, "loss_cls_0": 1.34558, "loss_box_0": 2.46237, "loss_cns_0": 0.60638, "loss_yns_0": 0.18348, "loss_cls_1": 1.48256, "loss_box_1": 3.61786, "loss_cns_1": 0.41451, "loss_yns_1": 0.19092, "loss_cls_2": 1.56167, "loss_box_2": 3.87083, "loss_cns_2": 0.37505, "loss_yns_2": 0.19837, "loss_cls_3": 1.37966, "loss_box_3": 3.86649, "loss_cns_3": 0.36688, "loss_yns_3": 0.18915, "loss_cls_4": 1.37723, "loss_box_4": 4.11791, "loss_cns_4": 0.31934, "loss_yns_4": 0.19271, "loss_cls_5": 1.32518, "loss_box_5": 4.29064, "loss_cns_5": 0.31858, "loss_yns_5": 0.19853, "loss_cls_dn_0": 0.51294, "loss_box_dn_0": 1.12617, "loss_cls_dn_1": 0.614, "loss_box_dn_1": 2.18408, "loss_cls_dn_2": 0.59267, "loss_box_dn_2": 2.22434, "loss_cls_dn_3": 0.51819, "loss_box_dn_3": 2.25426, "loss_cls_dn_4": 0.48253, "loss_box_dn_4": 2.434, "loss_cls_dn_5": 0.50417, "loss_box_dn_5": 2.53059, "loss_dense_depth": 1.45341, "loss": 51.68324, "grad_norm": 110.01046, "time": 1.5557} -{"mode": "train", "epoch": 1, "iter": 7, "lr": 0.0001, "memory": 49163, "data_time": 0.08266, "loss_cls_0": 1.29576, "loss_box_0": 2.31963, "loss_cns_0": 0.68143, "loss_yns_0": 0.17561, "loss_cls_1": 1.38985, "loss_box_1": 3.59721, "loss_cns_1": 0.4405, "loss_yns_1": 0.19177, "loss_cls_2": 1.43338, "loss_box_2": 3.73462, "loss_cns_2": 0.43291, "loss_yns_2": 0.19154, "loss_cls_3": 1.31329, "loss_box_3": 3.58642, "loss_cns_3": 0.44136, "loss_yns_3": 0.17975, "loss_cls_4": 1.28725, "loss_box_4": 3.73542, "loss_cns_4": 0.4443, "loss_yns_4": 0.18481, "loss_cls_5": 1.32389, "loss_box_5": 4.01265, "loss_cns_5": 0.44452, "loss_yns_5": 0.18999, "loss_cls_dn_0": 0.51662, "loss_box_dn_0": 1.05808, "loss_cls_dn_1": 0.54409, "loss_box_dn_1": 2.25119, "loss_cls_dn_2": 0.54626, "loss_box_dn_2": 2.26551, "loss_cls_dn_3": 0.48102, "loss_box_dn_3": 2.24945, "loss_cls_dn_4": 0.45038, "loss_box_dn_4": 2.36834, "loss_cls_dn_5": 0.44089, "loss_box_dn_5": 2.48376, "loss_dense_depth": 1.47963, "loss": 50.16308, "grad_norm": 99.4865, "time": 1.53513} -{"mode": "train", "epoch": 1, "iter": 8, "lr": 0.0001, "memory": 49163, "data_time": 0.07888, "loss_cls_0": 1.27362, "loss_box_0": 2.24044, "loss_cns_0": 0.65924, "loss_yns_0": 0.17717, "loss_cls_1": 1.31176, "loss_box_1": 3.50038, "loss_cns_1": 0.46882, "loss_yns_1": 0.19011, "loss_cls_2": 1.3739, "loss_box_2": 3.60151, "loss_cns_2": 0.45321, "loss_yns_2": 0.18449, "loss_cls_3": 1.30324, "loss_box_3": 3.53337, "loss_cns_3": 0.47012, "loss_yns_3": 0.18567, "loss_cls_4": 1.30154, "loss_box_4": 3.53165, "loss_cns_4": 0.4629, "loss_yns_4": 0.18218, "loss_cls_5": 1.32426, "loss_box_5": 3.61766, "loss_cns_5": 0.49105, "loss_yns_5": 0.19285, "loss_cls_dn_0": 0.51912, "loss_box_dn_0": 1.00351, "loss_cls_dn_1": 0.54257, "loss_box_dn_1": 1.66335, "loss_cls_dn_2": 0.56916, "loss_box_dn_2": 1.69296, "loss_cls_dn_3": 0.49904, "loss_box_dn_3": 1.69006, "loss_cls_dn_4": 0.47868, "loss_box_dn_4": 1.70491, "loss_cls_dn_5": 0.44235, "loss_box_dn_5": 1.76891, "loss_dense_depth": 1.38274, "loss": 45.98848, "grad_norm": 78.61095, "time": 1.54346} -{"mode": "train", "epoch": 1, "iter": 9, "lr": 0.0001, "memory": 49163, "data_time": 0.07762, "loss_cls_0": 1.25885, "loss_box_0": 2.23286, "loss_cns_0": 0.62631, "loss_yns_0": 0.17619, "loss_cls_1": 1.26641, "loss_box_1": 3.27047, "loss_cns_1": 0.50478, "loss_yns_1": 0.17962, "loss_cls_2": 1.37428, "loss_box_2": 3.2958, "loss_cns_2": 0.47145, "loss_yns_2": 0.18177, "loss_cls_3": 1.26709, "loss_box_3": 3.38722, "loss_cns_3": 0.5, "loss_yns_3": 0.21751, "loss_cls_4": 1.25744, "loss_box_4": 3.40206, "loss_cns_4": 0.50061, "loss_yns_4": 0.19378, "loss_cls_5": 1.30522, "loss_box_5": 3.45202, "loss_cns_5": 0.5044, "loss_yns_5": 0.18123, "loss_cls_dn_0": 0.5123, "loss_box_dn_0": 1.01867, "loss_cls_dn_1": 0.49486, "loss_box_dn_1": 1.49015, "loss_cls_dn_2": 0.54977, "loss_box_dn_2": 1.54286, "loss_cls_dn_3": 0.47858, "loss_box_dn_3": 1.63529, "loss_cls_dn_4": 0.46157, "loss_box_dn_4": 1.64121, "loss_cls_dn_5": 0.43048, "loss_box_dn_5": 1.73943, "loss_dense_depth": 1.3672, "loss": 44.36974, "grad_norm": 71.33639, "time": 1.51018} -{"mode": "train", "epoch": 1, "iter": 10, "lr": 0.0001, "memory": 49163, "data_time": 0.07686, "loss_cls_0": 1.22639, "loss_box_0": 2.25199, "loss_cns_0": 0.61524, "loss_yns_0": 0.17274, "loss_cls_1": 1.26474, "loss_box_1": 3.17565, "loss_cns_1": 0.51586, "loss_yns_1": 0.18302, "loss_cls_2": 1.29149, "loss_box_2": 3.13614, "loss_cns_2": 0.48927, "loss_yns_2": 0.18871, "loss_cls_3": 1.24429, "loss_box_3": 3.22816, "loss_cns_3": 0.55225, "loss_yns_3": 0.18763, "loss_cls_4": 1.2518, "loss_box_4": 3.26519, "loss_cns_4": 0.54977, "loss_yns_4": 0.18203, "loss_cls_5": 1.27744, "loss_box_5": 3.27596, "loss_cns_5": 0.53795, "loss_yns_5": 0.18342, "loss_cls_dn_0": 0.50113, "loss_box_dn_0": 1.05925, "loss_cls_dn_1": 0.44902, "loss_box_dn_1": 1.80822, "loss_cls_dn_2": 0.50211, "loss_box_dn_2": 1.85267, "loss_cls_dn_3": 0.45392, "loss_box_dn_3": 1.91579, "loss_cls_dn_4": 0.42724, "loss_box_dn_4": 1.9409, "loss_cls_dn_5": 0.42855, "loss_box_dn_5": 2.04306, "loss_dense_depth": 1.33851, "loss": 44.96749, "grad_norm": 64.45756, "time": 1.52941} -{"mode": "train", "epoch": 1, "iter": 11, "lr": 0.0001, "memory": 49163, "data_time": 0.07843, "loss_cls_0": 1.21673, "loss_box_0": 2.29169, "loss_cns_0": 0.6272, "loss_yns_0": 0.17346, "loss_cls_1": 1.27073, "loss_box_1": 3.10059, "loss_cns_1": 0.52037, "loss_yns_1": 0.17793, "loss_cls_2": 1.2733, "loss_box_2": 3.03646, "loss_cns_2": 0.49402, "loss_yns_2": 0.18593, "loss_cls_3": 1.2468, "loss_box_3": 3.01957, "loss_cns_3": 0.5733, "loss_yns_3": 0.17939, "loss_cls_4": 1.23902, "loss_box_4": 3.11321, "loss_cns_4": 0.55619, "loss_yns_4": 0.18113, "loss_cls_5": 1.28964, "loss_box_5": 3.33944, "loss_cns_5": 0.52279, "loss_yns_5": 0.18083, "loss_cls_dn_0": 0.48865, "loss_box_dn_0": 1.08518, "loss_cls_dn_1": 0.42178, "loss_box_dn_1": 2.17591, "loss_cls_dn_2": 0.44698, "loss_box_dn_2": 2.17708, "loss_cls_dn_3": 0.4227, "loss_box_dn_3": 2.16229, "loss_cls_dn_4": 0.40324, "loss_box_dn_4": 2.17575, "loss_cls_dn_5": 0.41425, "loss_box_dn_5": 2.30947, "loss_dense_depth": 1.38139, "loss": 45.87441, "grad_norm": 78.49229, "time": 1.51976} -{"mode": "train", "epoch": 1, "iter": 12, "lr": 0.0001, "memory": 49163, "data_time": 0.07934, "loss_cls_0": 1.23312, "loss_box_0": 2.29819, "loss_cns_0": 0.62643, "loss_yns_0": 0.17324, "loss_cls_1": 1.2656, "loss_box_1": 2.92237, "loss_cns_1": 0.51361, "loss_yns_1": 0.17861, "loss_cls_2": 1.31244, "loss_box_2": 2.88587, "loss_cns_2": 0.50223, "loss_yns_2": 0.18653, "loss_cls_3": 1.26043, "loss_box_3": 2.951, "loss_cns_3": 0.54241, "loss_yns_3": 0.17902, "loss_cls_4": 1.2383, "loss_box_4": 3.0226, "loss_cns_4": 0.52662, "loss_yns_4": 0.18041, "loss_cls_5": 1.28721, "loss_box_5": 3.18021, "loss_cns_5": 0.51567, "loss_yns_5": 0.17894, "loss_cls_dn_0": 0.47951, "loss_box_dn_0": 1.086, "loss_cls_dn_1": 0.39758, "loss_box_dn_1": 2.4937, "loss_cls_dn_2": 0.40906, "loss_box_dn_2": 2.44426, "loss_cls_dn_3": 0.39307, "loss_box_dn_3": 2.42481, "loss_cls_dn_4": 0.38191, "loss_box_dn_4": 2.42634, "loss_cls_dn_5": 0.39688, "loss_box_dn_5": 2.49981, "loss_dense_depth": 1.2749, "loss": 46.26889, "grad_norm": 63.26905, "time": 1.5116} -{"mode": "train", "epoch": 1, "iter": 13, "lr": 0.0001, "memory": 49163, "data_time": 0.08079, "loss_cls_0": 1.24012, "loss_box_0": 2.3143, "loss_cns_0": 0.59696, "loss_yns_0": 0.17104, "loss_cls_1": 1.24439, "loss_box_1": 2.76223, "loss_cns_1": 0.52932, "loss_yns_1": 0.17723, "loss_cls_2": 1.31624, "loss_box_2": 2.83802, "loss_cns_2": 0.52884, "loss_yns_2": 0.1809, "loss_cls_3": 1.24598, "loss_box_3": 2.95183, "loss_cns_3": 0.54212, "loss_yns_3": 0.17734, "loss_cls_4": 1.25315, "loss_box_4": 2.91788, "loss_cns_4": 0.52094, "loss_yns_4": 0.17896, "loss_cls_5": 1.27255, "loss_box_5": 2.91103, "loss_cns_5": 0.52055, "loss_yns_5": 0.1757, "loss_cls_dn_0": 0.47212, "loss_box_dn_0": 1.07472, "loss_cls_dn_1": 0.42611, "loss_box_dn_1": 1.77276, "loss_cls_dn_2": 0.43958, "loss_box_dn_2": 1.80136, "loss_cls_dn_3": 0.43288, "loss_box_dn_3": 1.92514, "loss_cls_dn_4": 0.42156, "loss_box_dn_4": 2.01896, "loss_cls_dn_5": 0.4491, "loss_box_dn_5": 2.01213, "loss_dense_depth": 1.37572, "loss": 43.18977, "grad_norm": 77.91913, "time": 1.52363} -{"mode": "train", "epoch": 1, "iter": 14, "lr": 0.00011, "memory": 49163, "data_time": 0.08035, "loss_cls_0": 1.25277, "loss_box_0": 2.3667, "loss_cns_0": 0.57098, "loss_yns_0": 0.17142, "loss_cls_1": 1.28368, "loss_box_1": 2.65487, "loss_cns_1": 0.54423, "loss_yns_1": 0.17768, "loss_cls_2": 1.31272, "loss_box_2": 2.7457, "loss_cns_2": 0.54436, "loss_yns_2": 0.1758, "loss_cls_3": 1.27964, "loss_box_3": 2.85458, "loss_cns_3": 0.55302, "loss_yns_3": 0.1745, "loss_cls_4": 1.29754, "loss_box_4": 2.87786, "loss_cns_4": 0.5368, "loss_yns_4": 0.17731, "loss_cls_5": 1.29913, "loss_box_5": 2.86486, "loss_cns_5": 0.55328, "loss_yns_5": 0.17721, "loss_cls_dn_0": 0.46522, "loss_box_dn_0": 1.05014, "loss_cls_dn_1": 0.44289, "loss_box_dn_1": 1.40277, "loss_cls_dn_2": 0.46492, "loss_box_dn_2": 1.4409, "loss_cls_dn_3": 0.45238, "loss_box_dn_3": 1.60124, "loss_cls_dn_4": 0.4386, "loss_box_dn_4": 1.74012, "loss_cls_dn_5": 0.46801, "loss_box_dn_5": 1.70729, "loss_dense_depth": 1.36609, "loss": 41.48721, "grad_norm": 79.77113, "time": 1.5247} -{"mode": "train", "epoch": 1, "iter": 15, "lr": 0.00011, "memory": 49163, "data_time": 0.07841, "loss_cls_0": 1.25662, "loss_box_0": 2.42749, "loss_cns_0": 0.56593, "loss_yns_0": 0.17115, "loss_cls_1": 1.31993, "loss_box_1": 2.65532, "loss_cns_1": 0.57226, "loss_yns_1": 0.17827, "loss_cls_2": 1.27474, "loss_box_2": 2.69969, "loss_cns_2": 0.5601, "loss_yns_2": 0.1787, "loss_cls_3": 1.31199, "loss_box_3": 2.7069, "loss_cns_3": 0.57093, "loss_yns_3": 0.17821, "loss_cls_4": 1.29245, "loss_box_4": 2.75192, "loss_cns_4": 0.56359, "loss_yns_4": 0.17985, "loss_cls_5": 1.29773, "loss_box_5": 2.7531, "loss_cns_5": 0.57983, "loss_yns_5": 0.17367, "loss_cls_dn_0": 0.46403, "loss_box_dn_0": 1.05466, "loss_cls_dn_1": 0.4434, "loss_box_dn_1": 1.50116, "loss_cls_dn_2": 0.46954, "loss_box_dn_2": 1.49197, "loss_cls_dn_3": 0.44497, "loss_box_dn_3": 1.57695, "loss_cls_dn_4": 0.43508, "loss_box_dn_4": 1.66328, "loss_cls_dn_5": 0.46287, "loss_box_dn_5": 1.63419, "loss_dense_depth": 1.45386, "loss": 41.31631, "grad_norm": 65.60568, "time": 1.51039} -{"mode": "train", "epoch": 1, "iter": 16, "lr": 0.00011, "memory": 49163, "data_time": 0.07699, "loss_cls_0": 1.22464, "loss_box_0": 2.38741, "loss_cns_0": 0.57993, "loss_yns_0": 0.17299, "loss_cls_1": 1.29006, "loss_box_1": 2.70035, "loss_cns_1": 0.58868, "loss_yns_1": 0.17365, "loss_cls_2": 1.26782, "loss_box_2": 2.75529, "loss_cns_2": 0.55649, "loss_yns_2": 0.17744, "loss_cls_3": 1.31102, "loss_box_3": 2.79557, "loss_cns_3": 0.57592, "loss_yns_3": 0.17481, "loss_cls_4": 1.26865, "loss_box_4": 2.77107, "loss_cns_4": 0.56, "loss_yns_4": 0.18117, "loss_cls_5": 1.27667, "loss_box_5": 2.80433, "loss_cns_5": 0.57123, "loss_yns_5": 0.1781, "loss_cls_dn_0": 0.48313, "loss_box_dn_0": 1.02728, "loss_cls_dn_1": 0.4647, "loss_box_dn_1": 1.39631, "loss_cls_dn_2": 0.48545, "loss_box_dn_2": 1.38137, "loss_cls_dn_3": 0.44196, "loss_box_dn_3": 1.4548, "loss_cls_dn_4": 0.44325, "loss_box_dn_4": 1.4875, "loss_cls_dn_5": 0.47503, "loss_box_dn_5": 1.5146, "loss_dense_depth": 1.2915, "loss": 40.69016, "grad_norm": 87.22899, "time": 1.51156} -{"mode": "train", "epoch": 1, "iter": 17, "lr": 0.00011, "memory": 49163, "data_time": 0.07589, "loss_cls_0": 1.21038, "loss_box_0": 2.32078, "loss_cns_0": 0.59857, "loss_yns_0": 0.17196, "loss_cls_1": 1.27537, "loss_box_1": 2.88312, "loss_cns_1": 0.56677, "loss_yns_1": 0.17392, "loss_cls_2": 1.27489, "loss_box_2": 2.90177, "loss_cns_2": 0.54316, "loss_yns_2": 0.17835, "loss_cls_3": 1.29817, "loss_box_3": 2.93934, "loss_cns_3": 0.57251, "loss_yns_3": 0.17384, "loss_cls_4": 1.2452, "loss_box_4": 2.93259, "loss_cns_4": 0.56304, "loss_yns_4": 0.17633, "loss_cls_5": 1.26579, "loss_box_5": 2.97566, "loss_cns_5": 0.54805, "loss_yns_5": 0.17442, "loss_cls_dn_0": 0.49172, "loss_box_dn_0": 1.01401, "loss_cls_dn_1": 0.46131, "loss_box_dn_1": 1.51419, "loss_cls_dn_2": 0.46705, "loss_box_dn_2": 1.51017, "loss_cls_dn_3": 0.42446, "loss_box_dn_3": 1.573, "loss_cls_dn_4": 0.43879, "loss_box_dn_4": 1.60563, "loss_cls_dn_5": 0.47406, "loss_box_dn_5": 1.71654, "loss_dense_depth": 1.36165, "loss": 42.01656, "grad_norm": 73.05412, "time": 1.51547} -{"mode": "train", "epoch": 1, "iter": 18, "lr": 0.00011, "memory": 49163, "data_time": 0.0776, "loss_cls_0": 1.20604, "loss_box_0": 2.22889, "loss_cns_0": 0.63043, "loss_yns_0": 0.17237, "loss_cls_1": 1.24736, "loss_box_1": 2.88374, "loss_cns_1": 0.55008, "loss_yns_1": 0.17757, "loss_cls_2": 1.27208, "loss_box_2": 2.87308, "loss_cns_2": 0.54518, "loss_yns_2": 0.17722, "loss_cls_3": 1.26702, "loss_box_3": 2.84036, "loss_cns_3": 0.56354, "loss_yns_3": 0.17222, "loss_cls_4": 1.23391, "loss_box_4": 2.86741, "loss_cns_4": 0.56927, "loss_yns_4": 0.17262, "loss_cls_5": 1.26356, "loss_box_5": 2.89798, "loss_cns_5": 0.53842, "loss_yns_5": 0.1721, "loss_cls_dn_0": 0.49205, "loss_box_dn_0": 0.99907, "loss_cls_dn_1": 0.45691, "loss_box_dn_1": 1.56187, "loss_cls_dn_2": 0.46013, "loss_box_dn_2": 1.58678, "loss_cls_dn_3": 0.42346, "loss_box_dn_3": 1.62729, "loss_cls_dn_4": 0.4423, "loss_box_dn_4": 1.67078, "loss_cls_dn_5": 0.47084, "loss_box_dn_5": 1.83527, "loss_dense_depth": 1.20244, "loss": 41.75163, "grad_norm": 63.56734, "time": 1.52479} -{"mode": "train", "epoch": 1, "iter": 19, "lr": 0.00011, "memory": 49163, "data_time": 0.07405, "loss_cls_0": 1.22232, "loss_box_0": 2.18013, "loss_cns_0": 0.63057, "loss_yns_0": 0.17192, "loss_cls_1": 1.24208, "loss_box_1": 2.91202, "loss_cns_1": 0.5287, "loss_yns_1": 0.1736, "loss_cls_2": 1.27182, "loss_box_2": 2.93042, "loss_cns_2": 0.53697, "loss_yns_2": 0.17458, "loss_cls_3": 1.25471, "loss_box_3": 2.91865, "loss_cns_3": 0.54603, "loss_yns_3": 0.1734, "loss_cls_4": 1.24387, "loss_box_4": 2.92974, "loss_cns_4": 0.56976, "loss_yns_4": 0.17398, "loss_cls_5": 1.26831, "loss_box_5": 3.05216, "loss_cns_5": 0.52742, "loss_yns_5": 0.17402, "loss_cls_dn_0": 0.48569, "loss_box_dn_0": 1.00042, "loss_cls_dn_1": 0.46656, "loss_box_dn_1": 1.33039, "loss_cls_dn_2": 0.46763, "loss_box_dn_2": 1.41616, "loss_cls_dn_3": 0.44686, "loss_box_dn_3": 1.45066, "loss_cls_dn_4": 0.45597, "loss_box_dn_4": 1.49906, "loss_cls_dn_5": 0.47746, "loss_box_dn_5": 1.69786, "loss_dense_depth": 1.329, "loss": 41.3309, "grad_norm": 78.59177, "time": 1.52252} -{"mode": "train", "epoch": 1, "iter": 20, "lr": 0.00011, "memory": 49163, "data_time": 0.0723, "loss_cls_0": 1.22572, "loss_box_0": 2.15402, "loss_cns_0": 0.62686, "loss_yns_0": 0.17129, "loss_cls_1": 1.23432, "loss_box_1": 3.05467, "loss_cns_1": 0.51262, "loss_yns_1": 0.17449, "loss_cls_2": 1.26712, "loss_box_2": 3.07901, "loss_cns_2": 0.52524, "loss_yns_2": 0.17397, "loss_cls_3": 1.24525, "loss_box_3": 3.10771, "loss_cns_3": 0.52913, "loss_yns_3": 0.17225, "loss_cls_4": 1.24295, "loss_box_4": 3.10491, "loss_cns_4": 0.546, "loss_yns_4": 0.17498, "loss_cls_5": 1.26742, "loss_box_5": 3.19928, "loss_cns_5": 0.52555, "loss_yns_5": 0.17605, "loss_cls_dn_0": 0.47604, "loss_box_dn_0": 0.98614, "loss_cls_dn_1": 0.43032, "loss_box_dn_1": 1.5404, "loss_cls_dn_2": 0.43025, "loss_box_dn_2": 1.58627, "loss_cls_dn_3": 0.41884, "loss_box_dn_3": 1.58131, "loss_cls_dn_4": 0.42007, "loss_box_dn_4": 1.59267, "loss_cls_dn_5": 0.4309, "loss_box_dn_5": 1.6992, "loss_dense_depth": 1.22691, "loss": 42.31011, "grad_norm": 77.40698, "time": 1.52741} -{"mode": "train", "epoch": 1, "iter": 21, "lr": 0.00011, "memory": 49163, "data_time": 0.21755, "loss_cls_0": 1.20752, "loss_box_0": 2.11967, "loss_cns_0": 0.62354, "loss_yns_0": 0.17043, "loss_cls_1": 1.23981, "loss_box_1": 3.02604, "loss_cns_1": 0.53123, "loss_yns_1": 0.17615, "loss_cls_2": 1.27604, "loss_box_2": 3.01015, "loss_cns_2": 0.5366, "loss_yns_2": 0.17539, "loss_cls_3": 1.25121, "loss_box_3": 3.01681, "loss_cns_3": 0.53641, "loss_yns_3": 0.17393, "loss_cls_4": 1.25058, "loss_box_4": 3.00134, "loss_cns_4": 0.54355, "loss_yns_4": 0.17342, "loss_cls_5": 1.28521, "loss_box_5": 3.07374, "loss_cns_5": 0.53853, "loss_yns_5": 0.17879, "loss_cls_dn_0": 0.4803, "loss_box_dn_0": 1.00267, "loss_cls_dn_1": 0.44219, "loss_box_dn_1": 1.27917, "loss_cls_dn_2": 0.43852, "loss_box_dn_2": 1.29099, "loss_cls_dn_3": 0.43376, "loss_box_dn_3": 1.28779, "loss_cls_dn_4": 0.42595, "loss_box_dn_4": 1.29658, "loss_cls_dn_5": 0.43664, "loss_box_dn_5": 1.34866, "loss_dense_depth": 1.18276, "loss": 40.46209, "grad_norm": 60.09308, "time": 1.64282} -{"mode": "train", "epoch": 1, "iter": 22, "lr": 0.00011, "memory": 49163, "data_time": 0.0799, "loss_cls_0": 1.20802, "loss_box_0": 2.11871, "loss_cns_0": 0.6192, "loss_yns_0": 0.16922, "loss_cls_1": 1.26839, "loss_box_1": 2.88778, "loss_cns_1": 0.56442, "loss_yns_1": 0.17371, "loss_cls_2": 1.28802, "loss_box_2": 2.81714, "loss_cns_2": 0.55613, "loss_yns_2": 0.17621, "loss_cls_3": 1.27272, "loss_box_3": 2.82961, "loss_cns_3": 0.55845, "loss_yns_3": 0.17258, "loss_cls_4": 1.27888, "loss_box_4": 2.87196, "loss_cns_4": 0.53954, "loss_yns_4": 0.17339, "loss_cls_5": 1.31761, "loss_box_5": 2.91793, "loss_cns_5": 0.54079, "loss_yns_5": 0.17481, "loss_cls_dn_0": 0.4851, "loss_box_dn_0": 1.00117, "loss_cls_dn_1": 0.44037, "loss_box_dn_1": 1.10627, "loss_cls_dn_2": 0.43649, "loss_box_dn_2": 1.1044, "loss_cls_dn_3": 0.43917, "loss_box_dn_3": 1.13967, "loss_cls_dn_4": 0.42448, "loss_box_dn_4": 1.2017, "loss_cls_dn_5": 0.43587, "loss_box_dn_5": 1.21511, "loss_dense_depth": 1.19935, "loss": 39.12436, "grad_norm": 52.62543, "time": 1.5473} -{"mode": "train", "epoch": 1, "iter": 23, "lr": 0.00011, "memory": 49163, "data_time": 0.07809, "loss_cls_0": 1.18806, "loss_box_0": 2.11821, "loss_cns_0": 0.62131, "loss_yns_0": 0.16834, "loss_cls_1": 1.2498, "loss_box_1": 2.92601, "loss_cns_1": 0.53926, "loss_yns_1": 0.17207, "loss_cls_2": 1.26692, "loss_box_2": 2.86182, "loss_cns_2": 0.54862, "loss_yns_2": 0.17586, "loss_cls_3": 1.24772, "loss_box_3": 2.93404, "loss_cns_3": 0.55324, "loss_yns_3": 0.16979, "loss_cls_4": 1.24527, "loss_box_4": 3.01256, "loss_cns_4": 0.53855, "loss_yns_4": 0.17763, "loss_cls_5": 1.26731, "loss_box_5": 3.07345, "loss_cns_5": 0.55194, "loss_yns_5": 0.17265, "loss_cls_dn_0": 0.48488, "loss_box_dn_0": 0.98158, "loss_cls_dn_1": 0.43495, "loss_box_dn_1": 1.15563, "loss_cls_dn_2": 0.43504, "loss_box_dn_2": 1.14584, "loss_cls_dn_3": 0.44949, "loss_box_dn_3": 1.22806, "loss_cls_dn_4": 0.43223, "loss_box_dn_4": 1.34304, "loss_cls_dn_5": 0.44594, "loss_box_dn_5": 1.34765, "loss_dense_depth": 1.1172, "loss": 39.78195, "grad_norm": 79.18465, "time": 1.54617} -{"mode": "train", "epoch": 1, "iter": 24, "lr": 0.00011, "memory": 49163, "data_time": 0.09235, "loss_cls_0": 1.20141, "loss_box_0": 2.12033, "loss_cns_0": 0.62767, "loss_yns_0": 0.16838, "loss_cls_1": 1.24797, "loss_box_1": 3.02191, "loss_cns_1": 0.53457, "loss_yns_1": 0.17315, "loss_cls_2": 1.2696, "loss_box_2": 2.98766, "loss_cns_2": 0.54859, "loss_yns_2": 0.1718, "loss_cls_3": 1.24527, "loss_box_3": 3.04244, "loss_cns_3": 0.56369, "loss_yns_3": 0.16993, "loss_cls_4": 1.23363, "loss_box_4": 3.07163, "loss_cns_4": 0.55806, "loss_yns_4": 0.17176, "loss_cls_5": 1.25396, "loss_box_5": 3.11746, "loss_cns_5": 0.57528, "loss_yns_5": 0.17026, "loss_cls_dn_0": 0.49089, "loss_box_dn_0": 0.98911, "loss_cls_dn_1": 0.42696, "loss_box_dn_1": 1.30221, "loss_cls_dn_2": 0.42895, "loss_box_dn_2": 1.31562, "loss_cls_dn_3": 0.44984, "loss_box_dn_3": 1.42663, "loss_cls_dn_4": 0.43451, "loss_box_dn_4": 1.53474, "loss_cls_dn_5": 0.4473, "loss_box_dn_5": 1.53788, "loss_dense_depth": 1.24913, "loss": 41.28017, "grad_norm": 80.51273, "time": 1.543} -{"mode": "train", "epoch": 1, "iter": 25, "lr": 0.00011, "memory": 49163, "data_time": 0.07866, "loss_cls_0": 1.20787, "loss_box_0": 2.12159, "loss_cns_0": 0.63348, "loss_yns_0": 0.16961, "loss_cls_1": 1.24993, "loss_box_1": 2.95524, "loss_cns_1": 0.53434, "loss_yns_1": 0.1743, "loss_cls_2": 1.27515, "loss_box_2": 2.94845, "loss_cns_2": 0.54853, "loss_yns_2": 0.17467, "loss_cls_3": 1.2388, "loss_box_3": 2.97748, "loss_cns_3": 0.55202, "loss_yns_3": 0.17114, "loss_cls_4": 1.22835, "loss_box_4": 3.02139, "loss_cns_4": 0.53772, "loss_yns_4": 0.17104, "loss_cls_5": 1.24966, "loss_box_5": 3.06192, "loss_cns_5": 0.52983, "loss_yns_5": 0.17296, "loss_cls_dn_0": 0.48095, "loss_box_dn_0": 0.98035, "loss_cls_dn_1": 0.42005, "loss_box_dn_1": 1.37778, "loss_cls_dn_2": 0.4226, "loss_box_dn_2": 1.40894, "loss_cls_dn_3": 0.43983, "loss_box_dn_3": 1.51782, "loss_cls_dn_4": 0.42777, "loss_box_dn_4": 1.60594, "loss_cls_dn_5": 0.44441, "loss_box_dn_5": 1.59932, "loss_dense_depth": 1.1931, "loss": 41.22432, "grad_norm": 76.31889, "time": 1.54062} -{"mode": "train", "epoch": 1, "iter": 26, "lr": 0.00011, "memory": 49163, "data_time": 0.07769, "loss_cls_0": 1.2057, "loss_box_0": 2.1302, "loss_cns_0": 0.63561, "loss_yns_0": 0.16918, "loss_cls_1": 1.25166, "loss_box_1": 2.82708, "loss_cns_1": 0.55252, "loss_yns_1": 0.17147, "loss_cls_2": 1.28868, "loss_box_2": 2.8186, "loss_cns_2": 0.56406, "loss_yns_2": 0.17546, "loss_cls_3": 1.24899, "loss_box_3": 2.8633, "loss_cns_3": 0.5658, "loss_yns_3": 0.173, "loss_cls_4": 1.24148, "loss_box_4": 2.88589, "loss_cns_4": 0.56222, "loss_yns_4": 0.17506, "loss_cls_5": 1.25257, "loss_box_5": 2.86336, "loss_cns_5": 0.54612, "loss_yns_5": 0.17194, "loss_cls_dn_0": 0.46725, "loss_box_dn_0": 0.99214, "loss_cls_dn_1": 0.44136, "loss_box_dn_1": 1.20151, "loss_cls_dn_2": 0.44139, "loss_box_dn_2": 1.23654, "loss_cls_dn_3": 0.44967, "loss_box_dn_3": 1.35197, "loss_cls_dn_4": 0.43912, "loss_box_dn_4": 1.42048, "loss_cls_dn_5": 0.47027, "loss_box_dn_5": 1.39058, "loss_dense_depth": 1.11305, "loss": 39.75527, "grad_norm": 74.68313, "time": 1.53782} -{"mode": "train", "epoch": 1, "iter": 27, "lr": 0.00011, "memory": 49163, "data_time": 0.07884, "loss_cls_0": 1.1726, "loss_box_0": 2.16535, "loss_cns_0": 0.63032, "loss_yns_0": 0.16881, "loss_cls_1": 1.23781, "loss_box_1": 2.7664, "loss_cns_1": 0.56073, "loss_yns_1": 0.17217, "loss_cls_2": 1.28634, "loss_box_2": 2.79842, "loss_cns_2": 0.56962, "loss_yns_2": 0.17788, "loss_cls_3": 1.24646, "loss_box_3": 2.8064, "loss_cns_3": 0.57614, "loss_yns_3": 0.17088, "loss_cls_4": 1.24646, "loss_box_4": 2.77475, "loss_cns_4": 0.59772, "loss_yns_4": 0.17294, "loss_cls_5": 1.23853, "loss_box_5": 2.80503, "loss_cns_5": 0.56917, "loss_yns_5": 0.16875, "loss_cls_dn_0": 0.47495, "loss_box_dn_0": 0.99698, "loss_cls_dn_1": 0.42794, "loss_box_dn_1": 1.20124, "loss_cls_dn_2": 0.42927, "loss_box_dn_2": 1.19139, "loss_cls_dn_3": 0.43424, "loss_box_dn_3": 1.24757, "loss_cls_dn_4": 0.42312, "loss_box_dn_4": 1.25794, "loss_cls_dn_5": 0.46098, "loss_box_dn_5": 1.25915, "loss_dense_depth": 1.1425, "loss": 39.02694, "grad_norm": 66.4882, "time": 1.52187} -{"mode": "train", "epoch": 1, "iter": 28, "lr": 0.00011, "memory": 49163, "data_time": 0.07782, "loss_cls_0": 1.15274, "loss_box_0": 2.16551, "loss_cns_0": 0.62351, "loss_yns_0": 0.16823, "loss_cls_1": 1.21851, "loss_box_1": 2.61541, "loss_cns_1": 0.57071, "loss_yns_1": 0.17151, "loss_cls_2": 1.25507, "loss_box_2": 2.75335, "loss_cns_2": 0.56271, "loss_yns_2": 0.17574, "loss_cls_3": 1.23924, "loss_box_3": 2.73242, "loss_cns_3": 0.55902, "loss_yns_3": 0.17126, "loss_cls_4": 1.25073, "loss_box_4": 2.69847, "loss_cns_4": 0.56792, "loss_yns_4": 0.17211, "loss_cls_5": 1.23176, "loss_box_5": 2.75788, "loss_cns_5": 0.56797, "loss_yns_5": 0.16947, "loss_cls_dn_0": 0.47577, "loss_box_dn_0": 0.98649, "loss_cls_dn_1": 0.41306, "loss_box_dn_1": 1.22707, "loss_cls_dn_2": 0.41842, "loss_box_dn_2": 1.21872, "loss_cls_dn_3": 0.41369, "loss_box_dn_3": 1.23218, "loss_cls_dn_4": 0.4013, "loss_box_dn_4": 1.22403, "loss_cls_dn_5": 0.43329, "loss_box_dn_5": 1.26331, "loss_dense_depth": 1.09082, "loss": 38.34941, "grad_norm": 84.26495, "time": 1.51461} -{"mode": "train", "epoch": 1, "iter": 29, "lr": 0.00011, "memory": 49163, "data_time": 0.07737, "loss_cls_0": 1.13113, "loss_box_0": 2.13384, "loss_cns_0": 0.62026, "loss_yns_0": 0.16831, "loss_cls_1": 1.21316, "loss_box_1": 2.5998, "loss_cns_1": 0.56765, "loss_yns_1": 0.17122, "loss_cls_2": 1.22948, "loss_box_2": 2.68144, "loss_cns_2": 0.56737, "loss_yns_2": 0.17894, "loss_cls_3": 1.22574, "loss_box_3": 2.64369, "loss_cns_3": 0.5708, "loss_yns_3": 0.17264, "loss_cls_4": 1.24985, "loss_box_4": 2.6078, "loss_cns_4": 0.57697, "loss_yns_4": 0.1807, "loss_cls_5": 1.24969, "loss_box_5": 2.63316, "loss_cns_5": 0.57682, "loss_yns_5": 0.17392, "loss_cls_dn_0": 0.45796, "loss_box_dn_0": 0.97977, "loss_cls_dn_1": 0.41122, "loss_box_dn_1": 1.14372, "loss_cls_dn_2": 0.42541, "loss_box_dn_2": 1.13748, "loss_cls_dn_3": 0.41224, "loss_box_dn_3": 1.13994, "loss_cls_dn_4": 0.39592, "loss_box_dn_4": 1.14614, "loss_cls_dn_5": 0.41997, "loss_box_dn_5": 1.21525, "loss_dense_depth": 1.08538, "loss": 37.49478, "grad_norm": 73.03674, "time": 1.53718} -{"mode": "train", "epoch": 1, "iter": 30, "lr": 0.00011, "memory": 49163, "data_time": 0.0791, "loss_cls_0": 1.12598, "loss_box_0": 2.07406, "loss_cns_0": 0.62094, "loss_yns_0": 0.16889, "loss_cls_1": 1.18365, "loss_box_1": 2.55839, "loss_cns_1": 0.5723, "loss_yns_1": 0.17083, "loss_cls_2": 1.18463, "loss_box_2": 2.55189, "loss_cns_2": 0.58039, "loss_yns_2": 0.18094, "loss_cls_3": 1.20475, "loss_box_3": 2.52538, "loss_cns_3": 0.58178, "loss_yns_3": 0.17477, "loss_cls_4": 1.21971, "loss_box_4": 2.54355, "loss_cns_4": 0.58571, "loss_yns_4": 0.17658, "loss_cls_5": 1.25201, "loss_box_5": 2.59155, "loss_cns_5": 0.57298, "loss_yns_5": 0.1785, "loss_cls_dn_0": 0.43291, "loss_box_dn_0": 0.97728, "loss_cls_dn_1": 0.3985, "loss_box_dn_1": 1.16394, "loss_cls_dn_2": 0.42413, "loss_box_dn_2": 1.14549, "loss_cls_dn_3": 0.39915, "loss_box_dn_3": 1.16259, "loss_cls_dn_4": 0.38657, "loss_box_dn_4": 1.20312, "loss_cls_dn_5": 0.39614, "loss_box_dn_5": 1.30611, "loss_dense_depth": 1.091, "loss": 37.06707, "grad_norm": 50.23505, "time": 1.52097} -{"mode": "train", "epoch": 1, "iter": 31, "lr": 0.00011, "memory": 49163, "data_time": 0.07564, "loss_cls_0": 1.12188, "loss_box_0": 2.05681, "loss_cns_0": 0.62081, "loss_yns_0": 0.17142, "loss_cls_1": 1.17066, "loss_box_1": 2.55905, "loss_cns_1": 0.57181, "loss_yns_1": 0.17033, "loss_cls_2": 1.17839, "loss_box_2": 2.55136, "loss_cns_2": 0.57374, "loss_yns_2": 0.17813, "loss_cls_3": 1.1916, "loss_box_3": 2.57655, "loss_cns_3": 0.56819, "loss_yns_3": 0.17228, "loss_cls_4": 1.18929, "loss_box_4": 2.63857, "loss_cns_4": 0.55366, "loss_yns_4": 0.17183, "loss_cls_5": 1.21499, "loss_box_5": 2.71409, "loss_cns_5": 0.54983, "loss_yns_5": 0.17173, "loss_cls_dn_0": 0.42635, "loss_box_dn_0": 0.96659, "loss_cls_dn_1": 0.39407, "loss_box_dn_1": 1.23006, "loss_cls_dn_2": 0.42238, "loss_box_dn_2": 1.23046, "loss_cls_dn_3": 0.38953, "loss_box_dn_3": 1.27177, "loss_cls_dn_4": 0.38925, "loss_box_dn_4": 1.32298, "loss_cls_dn_5": 0.3974, "loss_box_dn_5": 1.45413, "loss_dense_depth": 1.0558, "loss": 37.58777, "grad_norm": 76.90909, "time": 1.53482} -{"mode": "train", "epoch": 1, "iter": 32, "lr": 0.00011, "memory": 49163, "data_time": 0.07823, "loss_cls_0": 1.10306, "loss_box_0": 2.05476, "loss_cns_0": 0.62153, "loss_yns_0": 0.17059, "loss_cls_1": 1.15385, "loss_box_1": 2.71224, "loss_cns_1": 0.56615, "loss_yns_1": 0.17231, "loss_cls_2": 1.16499, "loss_box_2": 2.69252, "loss_cns_2": 0.56845, "loss_yns_2": 0.17776, "loss_cls_3": 1.1846, "loss_box_3": 2.72441, "loss_cns_3": 0.56736, "loss_yns_3": 0.17249, "loss_cls_4": 1.16092, "loss_box_4": 2.78244, "loss_cns_4": 0.55974, "loss_yns_4": 0.17126, "loss_cls_5": 1.18518, "loss_box_5": 2.82416, "loss_cns_5": 0.55693, "loss_yns_5": 0.17145, "loss_cls_dn_0": 0.41534, "loss_box_dn_0": 0.97004, "loss_cls_dn_1": 0.38506, "loss_box_dn_1": 1.25548, "loss_cls_dn_2": 0.41245, "loss_box_dn_2": 1.26301, "loss_cls_dn_3": 0.3789, "loss_box_dn_3": 1.29773, "loss_cls_dn_4": 0.38748, "loss_box_dn_4": 1.34433, "loss_cls_dn_5": 0.40181, "loss_box_dn_5": 1.45769, "loss_dense_depth": 1.0137, "loss": 38.20215, "grad_norm": 74.10196, "time": 1.51482} -{"mode": "train", "epoch": 1, "iter": 33, "lr": 0.00011, "memory": 49163, "data_time": 0.07712, "loss_cls_0": 1.0945, "loss_box_0": 2.06107, "loss_cns_0": 0.62408, "loss_yns_0": 0.17093, "loss_cls_1": 1.15652, "loss_box_1": 2.73491, "loss_cns_1": 0.56721, "loss_yns_1": 0.17409, "loss_cls_2": 1.16484, "loss_box_2": 2.67719, "loss_cns_2": 0.57932, "loss_yns_2": 0.17737, "loss_cls_3": 1.19207, "loss_box_3": 2.70456, "loss_cns_3": 0.58201, "loss_yns_3": 0.1715, "loss_cls_4": 1.15937, "loss_box_4": 2.757, "loss_cns_4": 0.58009, "loss_yns_4": 0.17025, "loss_cls_5": 1.18025, "loss_box_5": 2.76178, "loss_cns_5": 0.57548, "loss_yns_5": 0.17452, "loss_cls_dn_0": 0.43108, "loss_box_dn_0": 0.98888, "loss_cls_dn_1": 0.36143, "loss_box_dn_1": 1.38753, "loss_cls_dn_2": 0.38676, "loss_box_dn_2": 1.35149, "loss_cls_dn_3": 0.3584, "loss_box_dn_3": 1.35707, "loss_cls_dn_4": 0.36836, "loss_box_dn_4": 1.37492, "loss_cls_dn_5": 0.38422, "loss_box_dn_5": 1.43825, "loss_dense_depth": 1.0395, "loss": 38.4188, "grad_norm": 56.01353, "time": 1.51573} -{"mode": "train", "epoch": 1, "iter": 34, "lr": 0.00011, "memory": 49163, "data_time": 0.07774, "loss_cls_0": 1.08499, "loss_box_0": 2.05183, "loss_cns_0": 0.62459, "loss_yns_0": 0.17063, "loss_cls_1": 1.14342, "loss_box_1": 2.67731, "loss_cns_1": 0.56751, "loss_yns_1": 0.17414, "loss_cls_2": 1.15697, "loss_box_2": 2.69077, "loss_cns_2": 0.57354, "loss_yns_2": 0.17439, "loss_cls_3": 1.17068, "loss_box_3": 2.70706, "loss_cns_3": 0.57199, "loss_yns_3": 0.17116, "loss_cls_4": 1.14838, "loss_box_4": 2.73775, "loss_cns_4": 0.56733, "loss_yns_4": 0.17095, "loss_cls_5": 1.17568, "loss_box_5": 2.7483, "loss_cns_5": 0.55633, "loss_yns_5": 0.17781, "loss_cls_dn_0": 0.44288, "loss_box_dn_0": 0.99724, "loss_cls_dn_1": 0.34991, "loss_box_dn_1": 1.37283, "loss_cls_dn_2": 0.37023, "loss_box_dn_2": 1.32602, "loss_cls_dn_3": 0.36351, "loss_box_dn_3": 1.30687, "loss_cls_dn_4": 0.36048, "loss_box_dn_4": 1.28975, "loss_cls_dn_5": 0.3809, "loss_box_dn_5": 1.32007, "loss_dense_depth": 1.0068, "loss": 37.881, "grad_norm": 60.01796, "time": 1.51496} -{"mode": "train", "epoch": 1, "iter": 35, "lr": 0.00011, "memory": 49163, "data_time": 0.0781, "loss_cls_0": 1.07943, "loss_box_0": 2.00764, "loss_cns_0": 0.62836, "loss_yns_0": 0.17143, "loss_cls_1": 1.15074, "loss_box_1": 2.54463, "loss_cns_1": 0.57259, "loss_yns_1": 0.17804, "loss_cls_2": 1.16429, "loss_box_2": 2.5885, "loss_cns_2": 0.58023, "loss_yns_2": 0.17577, "loss_cls_3": 1.17296, "loss_box_3": 2.55204, "loss_cns_3": 0.5759, "loss_yns_3": 0.17154, "loss_cls_4": 1.17537, "loss_box_4": 2.56486, "loss_cns_4": 0.57595, "loss_yns_4": 0.17278, "loss_cls_5": 1.18115, "loss_box_5": 2.60287, "loss_cns_5": 0.57122, "loss_yns_5": 0.17501, "loss_cls_dn_0": 0.44197, "loss_box_dn_0": 0.98751, "loss_cls_dn_1": 0.34758, "loss_box_dn_1": 1.25747, "loss_cls_dn_2": 0.36412, "loss_box_dn_2": 1.24915, "loss_cls_dn_3": 0.37809, "loss_box_dn_3": 1.23682, "loss_cls_dn_4": 0.36389, "loss_box_dn_4": 1.23522, "loss_cls_dn_5": 0.38073, "loss_box_dn_5": 1.25383, "loss_dense_depth": 1.01835, "loss": 36.84805, "grad_norm": 58.53977, "time": 1.56624} -{"mode": "train", "epoch": 1, "iter": 36, "lr": 0.00011, "memory": 49163, "data_time": 0.07716, "loss_cls_0": 1.07628, "loss_box_0": 1.99052, "loss_cns_0": 0.62232, "loss_yns_0": 0.17106, "loss_cls_1": 1.16064, "loss_box_1": 2.49296, "loss_cns_1": 0.57649, "loss_yns_1": 0.17892, "loss_cls_2": 1.16888, "loss_box_2": 2.51903, "loss_cns_2": 0.58364, "loss_yns_2": 0.18012, "loss_cls_3": 1.15852, "loss_box_3": 2.51478, "loss_cns_3": 0.57877, "loss_yns_3": 0.17067, "loss_cls_4": 1.18096, "loss_box_4": 2.58526, "loss_cns_4": 0.57918, "loss_yns_4": 0.17157, "loss_cls_5": 1.17588, "loss_box_5": 2.57217, "loss_cns_5": 0.58101, "loss_yns_5": 0.17308, "loss_cls_dn_0": 0.42936, "loss_box_dn_0": 0.98473, "loss_cls_dn_1": 0.36946, "loss_box_dn_1": 1.09104, "loss_cls_dn_2": 0.38062, "loss_box_dn_2": 1.13222, "loss_cls_dn_3": 0.39613, "loss_box_dn_3": 1.18653, "loss_cls_dn_4": 0.37354, "loss_box_dn_4": 1.27332, "loss_cls_dn_5": 0.39712, "loss_box_dn_5": 1.26496, "loss_dense_depth": 0.97805, "loss": 36.41977, "grad_norm": 73.87955, "time": 1.5222} -{"mode": "train", "epoch": 1, "iter": 37, "lr": 0.00011, "memory": 49163, "data_time": 0.07409, "loss_cls_0": 1.0905, "loss_box_0": 1.99413, "loss_cns_0": 0.61962, "loss_yns_0": 0.17129, "loss_cls_1": 1.15325, "loss_box_1": 2.58399, "loss_cns_1": 0.57368, "loss_yns_1": 0.17527, "loss_cls_2": 1.17476, "loss_box_2": 2.58959, "loss_cns_2": 0.57661, "loss_yns_2": 0.17685, "loss_cls_3": 1.17166, "loss_box_3": 2.60922, "loss_cns_3": 0.57037, "loss_yns_3": 0.1688, "loss_cls_4": 1.19327, "loss_box_4": 2.68168, "loss_cns_4": 0.56999, "loss_yns_4": 0.17061, "loss_cls_5": 1.18859, "loss_box_5": 2.62934, "loss_cns_5": 0.57264, "loss_yns_5": 0.17995, "loss_cls_dn_0": 0.42323, "loss_box_dn_0": 0.99036, "loss_cls_dn_1": 0.35271, "loss_box_dn_1": 1.20221, "loss_cls_dn_2": 0.36675, "loss_box_dn_2": 1.24011, "loss_cls_dn_3": 0.36533, "loss_box_dn_3": 1.31897, "loss_cls_dn_4": 0.35703, "loss_box_dn_4": 1.41943, "loss_cls_dn_5": 0.37301, "loss_box_dn_5": 1.39857, "loss_dense_depth": 0.99313, "loss": 37.38649, "grad_norm": 76.63825, "time": 1.51562} -{"mode": "train", "epoch": 1, "iter": 38, "lr": 0.00011, "memory": 49163, "data_time": 0.07643, "loss_cls_0": 1.08451, "loss_box_0": 1.99526, "loss_cns_0": 0.61738, "loss_yns_0": 0.16863, "loss_cls_1": 1.1383, "loss_box_1": 2.60872, "loss_cns_1": 0.57362, "loss_yns_1": 0.171, "loss_cls_2": 1.16703, "loss_box_2": 2.58095, "loss_cns_2": 0.58025, "loss_yns_2": 0.17339, "loss_cls_3": 1.20613, "loss_box_3": 2.57296, "loss_cns_3": 0.57829, "loss_yns_3": 0.16994, "loss_cls_4": 1.17641, "loss_box_4": 2.61039, "loss_cns_4": 0.58107, "loss_yns_4": 0.17407, "loss_cls_5": 1.19313, "loss_box_5": 2.62096, "loss_cns_5": 0.58898, "loss_yns_5": 0.18078, "loss_cls_dn_0": 0.42155, "loss_box_dn_0": 0.99517, "loss_cls_dn_1": 0.34699, "loss_box_dn_1": 1.20996, "loss_cls_dn_2": 0.36636, "loss_box_dn_2": 1.21168, "loss_cls_dn_3": 0.35392, "loss_box_dn_3": 1.28331, "loss_cls_dn_4": 0.36312, "loss_box_dn_4": 1.3754, "loss_cls_dn_5": 0.37009, "loss_box_dn_5": 1.37138, "loss_dense_depth": 0.97934, "loss": 37.16045, "grad_norm": 62.14153, "time": 1.53216} -{"mode": "train", "epoch": 1, "iter": 39, "lr": 0.00012, "memory": 49163, "data_time": 0.07617, "loss_cls_0": 1.08623, "loss_box_0": 1.97768, "loss_cns_0": 0.61705, "loss_yns_0": 0.16743, "loss_cls_1": 1.15024, "loss_box_1": 2.47246, "loss_cns_1": 0.58923, "loss_yns_1": 0.16815, "loss_cls_2": 1.17722, "loss_box_2": 2.46968, "loss_cns_2": 0.5934, "loss_yns_2": 0.17469, "loss_cls_3": 1.1679, "loss_box_3": 2.48852, "loss_cns_3": 0.58847, "loss_yns_3": 0.16969, "loss_cls_4": 1.15881, "loss_box_4": 2.52633, "loss_cns_4": 0.58799, "loss_yns_4": 0.17116, "loss_cls_5": 1.18619, "loss_box_5": 2.63695, "loss_cns_5": 0.57743, "loss_yns_5": 0.16923, "loss_cls_dn_0": 0.42579, "loss_box_dn_0": 0.98056, "loss_cls_dn_1": 0.35037, "loss_box_dn_1": 1.17639, "loss_cls_dn_2": 0.37696, "loss_box_dn_2": 1.16203, "loss_cls_dn_3": 0.37445, "loss_box_dn_3": 1.23166, "loss_cls_dn_4": 0.39362, "loss_box_dn_4": 1.30022, "loss_cls_dn_5": 0.37932, "loss_box_dn_5": 1.32731, "loss_dense_depth": 0.97108, "loss": 36.52188, "grad_norm": 71.39887, "time": 1.52021} -{"mode": "train", "epoch": 1, "iter": 40, "lr": 0.00012, "memory": 49163, "data_time": 0.07646, "loss_cls_0": 1.0765, "loss_box_0": 1.96101, "loss_cns_0": 0.6214, "loss_yns_0": 0.16895, "loss_cls_1": 1.12916, "loss_box_1": 2.38536, "loss_cns_1": 0.59436, "loss_yns_1": 0.17362, "loss_cls_2": 1.16886, "loss_box_2": 2.36007, "loss_cns_2": 0.59876, "loss_yns_2": 0.17666, "loss_cls_3": 1.15245, "loss_box_3": 2.41716, "loss_cns_3": 0.59439, "loss_yns_3": 0.17029, "loss_cls_4": 1.14918, "loss_box_4": 2.46323, "loss_cns_4": 0.58968, "loss_yns_4": 0.16934, "loss_cls_5": 1.16204, "loss_box_5": 2.52697, "loss_cns_5": 0.58222, "loss_yns_5": 0.17026, "loss_cls_dn_0": 0.43029, "loss_box_dn_0": 0.97192, "loss_cls_dn_1": 0.35713, "loss_box_dn_1": 1.19677, "loss_cls_dn_2": 0.38553, "loss_box_dn_2": 1.16476, "loss_cls_dn_3": 0.39913, "loss_box_dn_3": 1.22997, "loss_cls_dn_4": 0.41489, "loss_box_dn_4": 1.2664, "loss_cls_dn_5": 0.38712, "loss_box_dn_5": 1.27321, "loss_dense_depth": 1.0194, "loss": 36.05845, "grad_norm": 68.32541, "time": 1.52788} -{"mode": "train", "epoch": 1, "iter": 41, "lr": 0.00012, "memory": 49163, "data_time": 0.15677, "loss_cls_0": 1.04006, "loss_box_0": 1.99346, "loss_cns_0": 0.62026, "loss_yns_0": 0.16843, "loss_cls_1": 1.10867, "loss_box_1": 2.37569, "loss_cns_1": 0.58919, "loss_yns_1": 0.17181, "loss_cls_2": 1.15095, "loss_box_2": 2.29036, "loss_cns_2": 0.60827, "loss_yns_2": 0.175, "loss_cls_3": 1.13996, "loss_box_3": 2.34383, "loss_cns_3": 0.61086, "loss_yns_3": 0.16715, "loss_cls_4": 1.12863, "loss_box_4": 2.33338, "loss_cns_4": 0.60642, "loss_yns_4": 0.16949, "loss_cls_5": 1.1451, "loss_box_5": 2.32812, "loss_cns_5": 0.61363, "loss_yns_5": 0.17069, "loss_cls_dn_0": 0.43126, "loss_box_dn_0": 0.96813, "loss_cls_dn_1": 0.34455, "loss_box_dn_1": 1.22719, "loss_cls_dn_2": 0.36958, "loss_box_dn_2": 1.16312, "loss_cls_dn_3": 0.38526, "loss_box_dn_3": 1.20106, "loss_cls_dn_4": 0.39311, "loss_box_dn_4": 1.19605, "loss_cls_dn_5": 0.37461, "loss_box_dn_5": 1.17725, "loss_dense_depth": 0.97576, "loss": 35.25635, "grad_norm": 57.75616, "time": 1.60994} -{"mode": "train", "epoch": 1, "iter": 42, "lr": 0.00012, "memory": 49163, "data_time": 0.07378, "loss_cls_0": 1.0325, "loss_box_0": 1.99514, "loss_cns_0": 0.6193, "loss_yns_0": 0.16748, "loss_cls_1": 1.11067, "loss_box_1": 2.38676, "loss_cns_1": 0.58927, "loss_yns_1": 0.17227, "loss_cls_2": 1.14506, "loss_box_2": 2.3134, "loss_cns_2": 0.60848, "loss_yns_2": 0.17027, "loss_cls_3": 1.13466, "loss_box_3": 2.34546, "loss_cns_3": 0.60894, "loss_yns_3": 0.16859, "loss_cls_4": 1.12046, "loss_box_4": 2.32408, "loss_cns_4": 0.60414, "loss_yns_4": 0.1706, "loss_cls_5": 1.14418, "loss_box_5": 2.37788, "loss_cns_5": 0.60219, "loss_yns_5": 0.17151, "loss_cls_dn_0": 0.42055, "loss_box_dn_0": 0.96261, "loss_cls_dn_1": 0.34676, "loss_box_dn_1": 1.1099, "loss_cls_dn_2": 0.35899, "loss_box_dn_2": 1.06053, "loss_cls_dn_3": 0.3788, "loss_box_dn_3": 1.07324, "loss_cls_dn_4": 0.3745, "loss_box_dn_4": 1.05484, "loss_cls_dn_5": 0.38208, "loss_box_dn_5": 1.08309, "loss_dense_depth": 0.95697, "loss": 34.64619, "grad_norm": 67.64624, "time": 1.54003} -{"mode": "train", "epoch": 1, "iter": 43, "lr": 0.00012, "memory": 49163, "data_time": 0.07606, "loss_cls_0": 1.07122, "loss_box_0": 1.99824, "loss_cns_0": 0.62125, "loss_yns_0": 0.16457, "loss_cls_1": 1.14178, "loss_box_1": 2.43847, "loss_cns_1": 0.59076, "loss_yns_1": 0.16681, "loss_cls_2": 1.188, "loss_box_2": 2.37896, "loss_cns_2": 0.60751, "loss_yns_2": 0.17061, "loss_cls_3": 1.16614, "loss_box_3": 2.37029, "loss_cns_3": 0.60592, "loss_yns_3": 0.1667, "loss_cls_4": 1.16613, "loss_box_4": 2.40154, "loss_cns_4": 0.59898, "loss_yns_4": 0.17051, "loss_cls_5": 1.16397, "loss_box_5": 2.47231, "loss_cns_5": 0.58971, "loss_yns_5": 0.17195, "loss_cls_dn_0": 0.43063, "loss_box_dn_0": 0.94863, "loss_cls_dn_1": 0.35518, "loss_box_dn_1": 1.02747, "loss_cls_dn_2": 0.35875, "loss_box_dn_2": 1.01267, "loss_cls_dn_3": 0.37561, "loss_box_dn_3": 1.02293, "loss_cls_dn_4": 0.36741, "loss_box_dn_4": 1.04367, "loss_cls_dn_5": 0.39905, "loss_box_dn_5": 1.13338, "loss_dense_depth": 0.98686, "loss": 35.04458, "grad_norm": 51.08602, "time": 1.55986} -{"mode": "train", "epoch": 1, "iter": 44, "lr": 0.00012, "memory": 49163, "data_time": 0.09684, "loss_cls_0": 1.06456, "loss_box_0": 2.05914, "loss_cns_0": 0.61416, "loss_yns_0": 0.16197, "loss_cls_1": 1.11977, "loss_box_1": 2.37571, "loss_cns_1": 0.59729, "loss_yns_1": 0.16524, "loss_cls_2": 1.17549, "loss_box_2": 2.39628, "loss_cns_2": 0.60313, "loss_yns_2": 0.16804, "loss_cls_3": 1.17194, "loss_box_3": 2.38269, "loss_cns_3": 0.60407, "loss_yns_3": 0.16501, "loss_cls_4": 1.1856, "loss_box_4": 2.42583, "loss_cns_4": 0.6029, "loss_yns_4": 0.167, "loss_cls_5": 1.15475, "loss_box_5": 2.4095, "loss_cns_5": 0.60008, "loss_yns_5": 0.16589, "loss_cls_dn_0": 0.44045, "loss_box_dn_0": 0.94647, "loss_cls_dn_1": 0.36059, "loss_box_dn_1": 1.06253, "loss_cls_dn_2": 0.36296, "loss_box_dn_2": 1.0933, "loss_cls_dn_3": 0.36605, "loss_box_dn_3": 1.12504, "loss_cls_dn_4": 0.3592, "loss_box_dn_4": 1.17747, "loss_cls_dn_5": 0.40146, "loss_box_dn_5": 1.25865, "loss_dense_depth": 1.02454, "loss": 35.51471, "grad_norm": 69.2449, "time": 1.57101} -{"mode": "train", "epoch": 1, "iter": 45, "lr": 0.00012, "memory": 49163, "data_time": 0.07416, "loss_cls_0": 1.06984, "loss_box_0": 2.08312, "loss_cns_0": 0.60691, "loss_yns_0": 0.16401, "loss_cls_1": 1.13111, "loss_box_1": 2.36361, "loss_cns_1": 0.59231, "loss_yns_1": 0.1688, "loss_cls_2": 1.18155, "loss_box_2": 2.45812, "loss_cns_2": 0.5833, "loss_yns_2": 0.17398, "loss_cls_3": 1.19325, "loss_box_3": 2.48353, "loss_cns_3": 0.57634, "loss_yns_3": 0.1716, "loss_cls_4": 1.21698, "loss_box_4": 2.51282, "loss_cns_4": 0.57404, "loss_yns_4": 0.1679, "loss_cls_5": 1.16889, "loss_box_5": 2.48883, "loss_cns_5": 0.58567, "loss_yns_5": 0.16889, "loss_cls_dn_0": 0.437, "loss_box_dn_0": 0.9523, "loss_cls_dn_1": 0.36143, "loss_box_dn_1": 1.14252, "loss_cls_dn_2": 0.36716, "loss_box_dn_2": 1.22154, "loss_cls_dn_3": 0.3667, "loss_box_dn_3": 1.26938, "loss_cls_dn_4": 0.35827, "loss_box_dn_4": 1.33166, "loss_cls_dn_5": 0.40407, "loss_box_dn_5": 1.412, "loss_dense_depth": 1.01566, "loss": 36.52512, "grad_norm": 80.65181, "time": 1.55182} -{"mode": "train", "epoch": 1, "iter": 46, "lr": 0.00012, "memory": 49163, "data_time": 0.0746, "loss_cls_0": 1.0664, "loss_box_0": 2.09141, "loss_cns_0": 0.6095, "loss_yns_0": 0.16176, "loss_cls_1": 1.11029, "loss_box_1": 2.47793, "loss_cns_1": 0.58933, "loss_yns_1": 0.16791, "loss_cls_2": 1.15534, "loss_box_2": 2.52328, "loss_cns_2": 0.58097, "loss_yns_2": 0.16873, "loss_cls_3": 1.16406, "loss_box_3": 2.51878, "loss_cns_3": 0.58357, "loss_yns_3": 0.16943, "loss_cls_4": 1.17901, "loss_box_4": 2.52411, "loss_cns_4": 0.58352, "loss_yns_4": 0.16896, "loss_cls_5": 1.15058, "loss_box_5": 2.5031, "loss_cns_5": 0.59151, "loss_yns_5": 0.16922, "loss_cls_dn_0": 0.41982, "loss_box_dn_0": 0.93954, "loss_cls_dn_1": 0.34879, "loss_box_dn_1": 1.20188, "loss_cls_dn_2": 0.36435, "loss_box_dn_2": 1.25722, "loss_cls_dn_3": 0.35847, "loss_box_dn_3": 1.28072, "loss_cls_dn_4": 0.35181, "loss_box_dn_4": 1.31011, "loss_cls_dn_5": 0.38941, "loss_box_dn_5": 1.37207, "loss_dense_depth": 0.96796, "loss": 36.57085, "grad_norm": 72.18983, "time": 1.56299} -{"mode": "train", "epoch": 1, "iter": 47, "lr": 0.00012, "memory": 49163, "data_time": 0.07292, "loss_cls_0": 1.06753, "loss_box_0": 2.06077, "loss_cns_0": 0.61701, "loss_yns_0": 0.16392, "loss_cls_1": 1.13077, "loss_box_1": 2.4497, "loss_cns_1": 0.60281, "loss_yns_1": 0.1673, "loss_cls_2": 1.14595, "loss_box_2": 2.43643, "loss_cns_2": 0.60642, "loss_yns_2": 0.16607, "loss_cls_3": 1.15561, "loss_box_3": 2.40804, "loss_cns_3": 0.60893, "loss_yns_3": 0.16623, "loss_cls_4": 1.15864, "loss_box_4": 2.40742, "loss_cns_4": 0.61039, "loss_yns_4": 0.1691, "loss_cls_5": 1.16236, "loss_box_5": 2.4199, "loss_cns_5": 0.61076, "loss_yns_5": 0.16879, "loss_cls_dn_0": 0.41827, "loss_box_dn_0": 0.92848, "loss_cls_dn_1": 0.34271, "loss_box_dn_1": 1.16036, "loss_cls_dn_2": 0.37522, "loss_box_dn_2": 1.18499, "loss_cls_dn_3": 0.36423, "loss_box_dn_3": 1.1764, "loss_cls_dn_4": 0.36137, "loss_box_dn_4": 1.18521, "loss_cls_dn_5": 0.38477, "loss_box_dn_5": 1.23774, "loss_dense_depth": 0.9708, "loss": 35.75142, "grad_norm": 45.18268, "time": 1.52563} -{"mode": "train", "epoch": 1, "iter": 48, "lr": 0.00012, "memory": 49163, "data_time": 0.0717, "loss_cls_0": 1.06213, "loss_box_0": 2.03677, "loss_cns_0": 0.61312, "loss_yns_0": 0.16275, "loss_cls_1": 1.13299, "loss_box_1": 2.492, "loss_cns_1": 0.59282, "loss_yns_1": 0.16599, "loss_cls_2": 1.1413, "loss_box_2": 2.47497, "loss_cns_2": 0.59822, "loss_yns_2": 0.16806, "loss_cls_3": 1.14489, "loss_box_3": 2.49518, "loss_cns_3": 0.59477, "loss_yns_3": 0.16697, "loss_cls_4": 1.1425, "loss_box_4": 2.53503, "loss_cns_4": 0.58589, "loss_yns_4": 0.16805, "loss_cls_5": 1.16425, "loss_box_5": 2.55717, "loss_cns_5": 0.59064, "loss_yns_5": 0.16585, "loss_cls_dn_0": 0.41661, "loss_box_dn_0": 0.91394, "loss_cls_dn_1": 0.32881, "loss_box_dn_1": 1.19402, "loss_cls_dn_2": 0.37891, "loss_box_dn_2": 1.17767, "loss_cls_dn_3": 0.36097, "loss_box_dn_3": 1.16727, "loss_cls_dn_4": 0.35908, "loss_box_dn_4": 1.17606, "loss_cls_dn_5": 0.37053, "loss_box_dn_5": 1.2031, "loss_dense_depth": 1.02214, "loss": 36.02142, "grad_norm": 56.84248, "time": 1.53909} -{"mode": "train", "epoch": 1, "iter": 49, "lr": 0.00012, "memory": 49163, "data_time": 0.07299, "loss_cls_0": 1.05927, "loss_box_0": 2.05725, "loss_cns_0": 0.61225, "loss_yns_0": 0.16364, "loss_cls_1": 1.10914, "loss_box_1": 2.51294, "loss_cns_1": 0.58846, "loss_yns_1": 0.1659, "loss_cls_2": 1.1248, "loss_box_2": 2.47055, "loss_cns_2": 0.60057, "loss_yns_2": 0.17229, "loss_cls_3": 1.13583, "loss_box_3": 2.50612, "loss_cns_3": 0.59387, "loss_yns_3": 0.16676, "loss_cls_4": 1.12672, "loss_box_4": 2.54107, "loss_cns_4": 0.59396, "loss_yns_4": 0.16902, "loss_cls_5": 1.14823, "loss_box_5": 2.49694, "loss_cns_5": 0.60248, "loss_yns_5": 0.16594, "loss_cls_dn_0": 0.417, "loss_box_dn_0": 0.91469, "loss_cls_dn_1": 0.3198, "loss_box_dn_1": 1.19328, "loss_cls_dn_2": 0.37292, "loss_box_dn_2": 1.15906, "loss_cls_dn_3": 0.35617, "loss_box_dn_3": 1.15154, "loss_cls_dn_4": 0.35557, "loss_box_dn_4": 1.15821, "loss_cls_dn_5": 0.35924, "loss_box_dn_5": 1.14539, "loss_dense_depth": 0.99999, "loss": 35.78685, "grad_norm": 52.94398, "time": 1.53649} -{"mode": "train", "epoch": 1, "iter": 50, "lr": 0.00012, "memory": 49163, "data_time": 0.074, "loss_cls_0": 1.02984, "loss_box_0": 2.05689, "loss_cns_0": 0.61647, "loss_yns_0": 0.16279, "loss_cls_1": 1.08489, "loss_box_1": 2.4537, "loss_cns_1": 0.60092, "loss_yns_1": 0.16598, "loss_cls_2": 1.10764, "loss_box_2": 2.39673, "loss_cns_2": 0.62833, "loss_yns_2": 0.16971, "loss_cls_3": 1.12131, "loss_box_3": 2.40722, "loss_cns_3": 0.62034, "loss_yns_3": 0.16519, "loss_cls_4": 1.1092, "loss_box_4": 2.44839, "loss_cns_4": 0.61853, "loss_yns_4": 0.16467, "loss_cls_5": 1.13464, "loss_box_5": 2.47151, "loss_cns_5": 0.62026, "loss_yns_5": 0.16381, "loss_cls_dn_0": 0.40799, "loss_box_dn_0": 0.93293, "loss_cls_dn_1": 0.32375, "loss_box_dn_1": 1.07615, "loss_cls_dn_2": 0.36721, "loss_box_dn_2": 1.03167, "loss_cls_dn_3": 0.35567, "loss_box_dn_3": 1.03905, "loss_cls_dn_4": 0.35899, "loss_box_dn_4": 1.06871, "loss_cls_dn_5": 0.36357, "loss_box_dn_5": 1.08436, "loss_dense_depth": 1.00649, "loss": 34.9355, "grad_norm": 46.21722, "time": 1.51621} -{"mode": "train", "epoch": 1, "iter": 51, "lr": 0.00012, "memory": 49163, "data_time": 0.07201, "loss_cls_0": 1.01747, "loss_box_0": 2.03636, "loss_cns_0": 0.61719, "loss_yns_0": 0.16263, "loss_cls_1": 1.06855, "loss_box_1": 2.48174, "loss_cns_1": 0.58781, "loss_yns_1": 0.16816, "loss_cls_2": 1.10229, "loss_box_2": 2.43647, "loss_cns_2": 0.60151, "loss_yns_2": 0.16705, "loss_cls_3": 1.12234, "loss_box_3": 2.43811, "loss_cns_3": 0.60331, "loss_yns_3": 0.16415, "loss_cls_4": 1.10802, "loss_box_4": 2.4848, "loss_cns_4": 0.60081, "loss_yns_4": 0.16381, "loss_cls_5": 1.12852, "loss_box_5": 2.57975, "loss_cns_5": 0.58278, "loss_yns_5": 0.16287, "loss_cls_dn_0": 0.40659, "loss_box_dn_0": 0.92722, "loss_cls_dn_1": 0.30893, "loss_box_dn_1": 1.13436, "loss_cls_dn_2": 0.33774, "loss_box_dn_2": 1.07103, "loss_cls_dn_3": 0.33299, "loss_box_dn_3": 1.09067, "loss_cls_dn_4": 0.34171, "loss_box_dn_4": 1.13558, "loss_cls_dn_5": 0.35483, "loss_box_dn_5": 1.1871, "loss_dense_depth": 0.93318, "loss": 35.14842, "grad_norm": 53.94448, "time": 1.53433} -{"mode": "train", "epoch": 1, "iter": 52, "lr": 0.00012, "memory": 49163, "data_time": 0.07446, "loss_cls_0": 1.03593, "loss_box_0": 2.0419, "loss_cns_0": 0.61418, "loss_yns_0": 0.16414, "loss_cls_1": 1.08587, "loss_box_1": 2.44307, "loss_cns_1": 0.5861, "loss_yns_1": 0.17007, "loss_cls_2": 1.1331, "loss_box_2": 2.36419, "loss_cns_2": 0.61118, "loss_yns_2": 0.16532, "loss_cls_3": 1.13878, "loss_box_3": 2.3652, "loss_cns_3": 0.61185, "loss_yns_3": 0.16368, "loss_cls_4": 1.12669, "loss_box_4": 2.40171, "loss_cns_4": 0.61238, "loss_yns_4": 0.16258, "loss_cls_5": 1.13274, "loss_box_5": 2.45942, "loss_cns_5": 0.61642, "loss_yns_5": 0.16393, "loss_cls_dn_0": 0.41195, "loss_box_dn_0": 0.92187, "loss_cls_dn_1": 0.30745, "loss_box_dn_1": 1.16275, "loss_cls_dn_2": 0.32574, "loss_box_dn_2": 1.10004, "loss_cls_dn_3": 0.33252, "loss_box_dn_3": 1.13859, "loss_cls_dn_4": 0.33766, "loss_box_dn_4": 1.19627, "loss_cls_dn_5": 0.36533, "loss_box_dn_5": 1.2459, "loss_dense_depth": 0.98451, "loss": 35.20101, "grad_norm": 54.44579, "time": 1.52264} -{"mode": "train", "epoch": 1, "iter": 53, "lr": 0.00012, "memory": 49163, "data_time": 0.07423, "loss_cls_0": 1.0405, "loss_box_0": 2.06247, "loss_cns_0": 0.61153, "loss_yns_0": 0.16527, "loss_cls_1": 1.0899, "loss_box_1": 2.48532, "loss_cns_1": 0.58599, "loss_yns_1": 0.16552, "loss_cls_2": 1.1421, "loss_box_2": 2.42223, "loss_cns_2": 0.60739, "loss_yns_2": 0.16573, "loss_cls_3": 1.15025, "loss_box_3": 2.43998, "loss_cns_3": 0.60337, "loss_yns_3": 0.16338, "loss_cls_4": 1.1523, "loss_box_4": 2.49082, "loss_cns_4": 0.60558, "loss_yns_4": 0.16299, "loss_cls_5": 1.1465, "loss_box_5": 2.48352, "loss_cns_5": 0.62069, "loss_yns_5": 0.16247, "loss_cls_dn_0": 0.4188, "loss_box_dn_0": 0.90818, "loss_cls_dn_1": 0.31116, "loss_box_dn_1": 1.10382, "loss_cls_dn_2": 0.32415, "loss_box_dn_2": 1.06881, "loss_cls_dn_3": 0.3369, "loss_box_dn_3": 1.11556, "loss_cls_dn_4": 0.33147, "loss_box_dn_4": 1.1818, "loss_cls_dn_5": 0.37161, "loss_box_dn_5": 1.21026, "loss_dense_depth": 1.01852, "loss": 35.42685, "grad_norm": 62.1471, "time": 1.52309} -{"mode": "train", "epoch": 1, "iter": 54, "lr": 0.00012, "memory": 49163, "data_time": 0.07402, "loss_cls_0": 1.02413, "loss_box_0": 1.99056, "loss_cns_0": 0.6166, "loss_yns_0": 0.16229, "loss_cls_1": 1.06826, "loss_box_1": 2.4162, "loss_cns_1": 0.59847, "loss_yns_1": 0.16115, "loss_cls_2": 1.1086, "loss_box_2": 2.39106, "loss_cns_2": 0.60864, "loss_yns_2": 0.16223, "loss_cls_3": 1.12308, "loss_box_3": 2.4075, "loss_cns_3": 0.60075, "loss_yns_3": 0.16216, "loss_cls_4": 1.14414, "loss_box_4": 2.43815, "loss_cns_4": 0.60474, "loss_yns_4": 0.16175, "loss_cls_5": 1.13298, "loss_box_5": 2.44154, "loss_cns_5": 0.61545, "loss_yns_5": 0.16081, "loss_cls_dn_0": 0.40351, "loss_box_dn_0": 0.89987, "loss_cls_dn_1": 0.30414, "loss_box_dn_1": 1.13296, "loss_cls_dn_2": 0.32179, "loss_box_dn_2": 1.11565, "loss_cls_dn_3": 0.33184, "loss_box_dn_3": 1.15423, "loss_cls_dn_4": 0.3169, "loss_box_dn_4": 1.2007, "loss_cls_dn_5": 0.35753, "loss_box_dn_5": 1.21951, "loss_dense_depth": 0.99674, "loss": 35.05661, "grad_norm": 57.88204, "time": 1.53534} -{"mode": "train", "epoch": 1, "iter": 55, "lr": 0.00012, "memory": 49163, "data_time": 0.07547, "loss_cls_0": 1.03326, "loss_box_0": 1.98483, "loss_cns_0": 0.61328, "loss_yns_0": 0.16569, "loss_cls_1": 1.05875, "loss_box_1": 2.28502, "loss_cns_1": 0.61038, "loss_yns_1": 0.16554, "loss_cls_2": 1.10055, "loss_box_2": 2.22759, "loss_cns_2": 0.62208, "loss_yns_2": 0.16583, "loss_cls_3": 1.11783, "loss_box_3": 2.23519, "loss_cns_3": 0.6253, "loss_yns_3": 0.16567, "loss_cls_4": 1.14231, "loss_box_4": 2.23374, "loss_cns_4": 0.62413, "loss_yns_4": 0.16493, "loss_cls_5": 1.12487, "loss_box_5": 2.24749, "loss_cns_5": 0.63099, "loss_yns_5": 0.16454, "loss_cls_dn_0": 0.41191, "loss_box_dn_0": 0.89373, "loss_cls_dn_1": 0.31731, "loss_box_dn_1": 0.97911, "loss_cls_dn_2": 0.35162, "loss_box_dn_2": 0.94975, "loss_cls_dn_3": 0.35572, "loss_box_dn_3": 0.97079, "loss_cls_dn_4": 0.33644, "loss_box_dn_4": 0.99091, "loss_cls_dn_5": 0.37575, "loss_box_dn_5": 1.00287, "loss_dense_depth": 0.99582, "loss": 33.44154, "grad_norm": 44.43835, "time": 1.60598} -{"mode": "train", "epoch": 1, "iter": 56, "lr": 0.00012, "memory": 49163, "data_time": 0.07539, "loss_cls_0": 1.00327, "loss_box_0": 1.97291, "loss_cns_0": 0.6064, "loss_yns_0": 0.16397, "loss_cls_1": 1.0524, "loss_box_1": 2.18559, "loss_cns_1": 0.61149, "loss_yns_1": 0.16619, "loss_cls_2": 1.08273, "loss_box_2": 2.14568, "loss_cns_2": 0.62351, "loss_yns_2": 0.16506, "loss_cls_3": 1.09135, "loss_box_3": 2.14995, "loss_cns_3": 0.62823, "loss_yns_3": 0.1643, "loss_cls_4": 1.09749, "loss_box_4": 2.21388, "loss_cns_4": 0.61084, "loss_yns_4": 0.16541, "loss_cls_5": 1.11099, "loss_box_5": 2.22998, "loss_cns_5": 0.61048, "loss_yns_5": 0.16578, "loss_cls_dn_0": 0.39644, "loss_box_dn_0": 0.89984, "loss_cls_dn_1": 0.29479, "loss_box_dn_1": 0.97901, "loss_cls_dn_2": 0.34418, "loss_box_dn_2": 0.94233, "loss_cls_dn_3": 0.34034, "loss_box_dn_3": 0.94396, "loss_cls_dn_4": 0.32926, "loss_box_dn_4": 0.9749, "loss_cls_dn_5": 0.36153, "loss_box_dn_5": 0.98023, "loss_dense_depth": 0.96251, "loss": 32.7672, "grad_norm": 52.87007, "time": 1.54045} -{"mode": "train", "epoch": 1, "iter": 57, "lr": 0.00012, "memory": 49163, "data_time": 0.07705, "loss_cls_0": 1.01931, "loss_box_0": 1.94776, "loss_cns_0": 0.60999, "loss_yns_0": 0.16423, "loss_cls_1": 1.0627, "loss_box_1": 2.21859, "loss_cns_1": 0.61059, "loss_yns_1": 0.16599, "loss_cls_2": 1.0878, "loss_box_2": 2.18372, "loss_cns_2": 0.62346, "loss_yns_2": 0.16434, "loss_cls_3": 1.1251, "loss_box_3": 2.19045, "loss_cns_3": 0.62567, "loss_yns_3": 0.167, "loss_cls_4": 1.12717, "loss_box_4": 2.25162, "loss_cns_4": 0.61302, "loss_yns_4": 0.16705, "loss_cls_5": 1.12366, "loss_box_5": 2.26313, "loss_cns_5": 0.61854, "loss_yns_5": 0.16584, "loss_cls_dn_0": 0.3846, "loss_box_dn_0": 0.89463, "loss_cls_dn_1": 0.27732, "loss_box_dn_1": 0.92841, "loss_cls_dn_2": 0.32896, "loss_box_dn_2": 0.89822, "loss_cls_dn_3": 0.32399, "loss_box_dn_3": 0.89632, "loss_cls_dn_4": 0.32167, "loss_box_dn_4": 0.92611, "loss_cls_dn_5": 0.34726, "loss_box_dn_5": 0.94492, "loss_dense_depth": 1.01449, "loss": 32.78362, "grad_norm": 50.26433, "time": 1.54342} -{"mode": "train", "epoch": 1, "iter": 58, "lr": 0.00012, "memory": 49163, "data_time": 0.07514, "loss_cls_0": 1.00274, "loss_box_0": 1.91222, "loss_cns_0": 0.61378, "loss_yns_0": 0.16417, "loss_cls_1": 1.04633, "loss_box_1": 2.1937, "loss_cns_1": 0.60911, "loss_yns_1": 0.16686, "loss_cls_2": 1.08377, "loss_box_2": 2.16151, "loss_cns_2": 0.62197, "loss_yns_2": 0.16647, "loss_cls_3": 1.09484, "loss_box_3": 2.1615, "loss_cns_3": 0.62534, "loss_yns_3": 0.16633, "loss_cls_4": 1.10144, "loss_box_4": 2.15067, "loss_cns_4": 0.62928, "loss_yns_4": 0.16551, "loss_cls_5": 1.10528, "loss_box_5": 2.1663, "loss_cns_5": 0.63016, "loss_yns_5": 0.16716, "loss_cls_dn_0": 0.38881, "loss_box_dn_0": 0.88118, "loss_cls_dn_1": 0.27177, "loss_box_dn_1": 0.91249, "loss_cls_dn_2": 0.3188, "loss_box_dn_2": 0.90057, "loss_cls_dn_3": 0.32528, "loss_box_dn_3": 0.90979, "loss_cls_dn_4": 0.3261, "loss_box_dn_4": 0.91955, "loss_cls_dn_5": 0.33802, "loss_box_dn_5": 0.95517, "loss_dense_depth": 0.98504, "loss": 32.339, "grad_norm": 40.92872, "time": 1.51348} -{"mode": "train", "epoch": 1, "iter": 59, "lr": 0.00012, "memory": 49163, "data_time": 0.07854, "loss_cls_0": 0.98355, "loss_box_0": 1.92303, "loss_cns_0": 0.61463, "loss_yns_0": 0.16318, "loss_cls_1": 1.05078, "loss_box_1": 2.18998, "loss_cns_1": 0.61, "loss_yns_1": 0.1647, "loss_cls_2": 1.07847, "loss_box_2": 2.14854, "loss_cns_2": 0.6235, "loss_yns_2": 0.1676, "loss_cls_3": 1.0795, "loss_box_3": 2.16195, "loss_cns_3": 0.62221, "loss_yns_3": 0.16588, "loss_cls_4": 1.07982, "loss_box_4": 2.14907, "loss_cns_4": 0.62754, "loss_yns_4": 0.16473, "loss_cls_5": 1.10476, "loss_box_5": 2.21205, "loss_cns_5": 0.62542, "loss_yns_5": 0.1746, "loss_cls_dn_0": 0.3975, "loss_box_dn_0": 0.87448, "loss_cls_dn_1": 0.27229, "loss_box_dn_1": 0.95282, "loss_cls_dn_2": 0.31229, "loss_box_dn_2": 0.95745, "loss_cls_dn_3": 0.33779, "loss_box_dn_3": 0.97988, "loss_cls_dn_4": 0.34846, "loss_box_dn_4": 0.99522, "loss_cls_dn_5": 0.34226, "loss_box_dn_5": 1.05963, "loss_dense_depth": 1.03105, "loss": 32.74661, "grad_norm": 57.03023, "time": 1.52413} -{"mode": "train", "epoch": 1, "iter": 60, "lr": 0.00012, "memory": 49163, "data_time": 0.08034, "loss_cls_0": 0.99461, "loss_box_0": 1.88407, "loss_cns_0": 0.61907, "loss_yns_0": 0.16188, "loss_cls_1": 1.09208, "loss_box_1": 2.08672, "loss_cns_1": 0.62583, "loss_yns_1": 0.16397, "loss_cls_2": 1.11424, "loss_box_2": 2.04177, "loss_cns_2": 0.63912, "loss_yns_2": 0.1662, "loss_cls_3": 1.11442, "loss_box_3": 2.04519, "loss_cns_3": 0.63523, "loss_yns_3": 0.1637, "loss_cls_4": 1.1034, "loss_box_4": 2.04706, "loss_cns_4": 0.63921, "loss_yns_4": 0.16383, "loss_cls_5": 1.13716, "loss_box_5": 2.1089, "loss_cns_5": 0.63102, "loss_yns_5": 0.171, "loss_cls_dn_0": 0.39658, "loss_box_dn_0": 0.88168, "loss_cls_dn_1": 0.26161, "loss_box_dn_1": 0.98709, "loss_cls_dn_2": 0.29906, "loss_box_dn_2": 0.99373, "loss_cls_dn_3": 0.33677, "loss_box_dn_3": 1.01117, "loss_cls_dn_4": 0.35661, "loss_box_dn_4": 1.02733, "loss_cls_dn_5": 0.35597, "loss_box_dn_5": 1.08235, "loss_dense_depth": 0.97517, "loss": 32.51479, "grad_norm": 52.2218, "time": 1.52586} -{"mode": "train", "epoch": 1, "iter": 61, "lr": 0.00012, "memory": 49163, "data_time": 0.15906, "loss_cls_0": 1.00273, "loss_box_0": 1.89676, "loss_cns_0": 0.61678, "loss_yns_0": 0.16019, "loss_cls_1": 1.08993, "loss_box_1": 2.07687, "loss_cns_1": 0.62654, "loss_yns_1": 0.16558, "loss_cls_2": 1.12005, "loss_box_2": 2.05518, "loss_cns_2": 0.63283, "loss_yns_2": 0.16254, "loss_cls_3": 1.10884, "loss_box_3": 2.0738, "loss_cns_3": 0.63424, "loss_yns_3": 0.16179, "loss_cls_4": 1.09986, "loss_box_4": 2.09189, "loss_cns_4": 0.63521, "loss_yns_4": 0.16389, "loss_cls_5": 1.10736, "loss_box_5": 2.06771, "loss_cns_5": 0.6387, "loss_yns_5": 0.16698, "loss_cls_dn_0": 0.38626, "loss_box_dn_0": 0.87739, "loss_cls_dn_1": 0.25239, "loss_box_dn_1": 1.03891, "loss_cls_dn_2": 0.2943, "loss_box_dn_2": 1.03513, "loss_cls_dn_3": 0.31746, "loss_box_dn_3": 1.04745, "loss_cls_dn_4": 0.3384, "loss_box_dn_4": 1.06198, "loss_cls_dn_5": 0.35887, "loss_box_dn_5": 1.07649, "loss_dense_depth": 1.02852, "loss": 32.66978, "grad_norm": 49.45399, "time": 1.59738} -{"mode": "train", "epoch": 1, "iter": 62, "lr": 0.00012, "memory": 49163, "data_time": 0.11816, "loss_cls_0": 1.02754, "loss_box_0": 1.90231, "loss_cns_0": 0.61576, "loss_yns_0": 0.15847, "loss_cls_1": 1.07894, "loss_box_1": 2.11716, "loss_cns_1": 0.62718, "loss_yns_1": 0.16484, "loss_cls_2": 1.09353, "loss_box_2": 2.07692, "loss_cns_2": 0.63457, "loss_yns_2": 0.16289, "loss_cls_3": 1.11436, "loss_box_3": 2.09144, "loss_cns_3": 0.63841, "loss_yns_3": 0.16215, "loss_cls_4": 1.10922, "loss_box_4": 2.11318, "loss_cns_4": 0.63628, "loss_yns_4": 0.16466, "loss_cls_5": 1.09825, "loss_box_5": 2.06874, "loss_cns_5": 0.64252, "loss_yns_5": 0.16649, "loss_cls_dn_0": 0.37467, "loss_box_dn_0": 0.87321, "loss_cls_dn_1": 0.26162, "loss_box_dn_1": 0.97565, "loss_cls_dn_2": 0.30966, "loss_box_dn_2": 0.95913, "loss_cls_dn_3": 0.30665, "loss_box_dn_3": 0.96357, "loss_cls_dn_4": 0.30697, "loss_box_dn_4": 0.97732, "loss_cls_dn_5": 0.35557, "loss_box_dn_5": 0.97351, "loss_dense_depth": 0.97173, "loss": 32.27505, "grad_norm": 50.47156, "time": 1.55302} -{"mode": "train", "epoch": 1, "iter": 63, "lr": 0.00012, "memory": 49163, "data_time": 0.07443, "loss_cls_0": 1.01396, "loss_box_0": 1.90068, "loss_cns_0": 0.614, "loss_yns_0": 0.16074, "loss_cls_1": 1.09445, "loss_box_1": 2.15236, "loss_cns_1": 0.62568, "loss_yns_1": 0.16365, "loss_cls_2": 1.10106, "loss_box_2": 2.09602, "loss_cns_2": 0.63919, "loss_yns_2": 0.16827, "loss_cls_3": 1.13263, "loss_box_3": 2.08657, "loss_cns_3": 0.64446, "loss_yns_3": 0.16277, "loss_cls_4": 1.17849, "loss_box_4": 2.08827, "loss_cns_4": 0.64423, "loss_yns_4": 0.16367, "loss_cls_5": 1.14449, "loss_box_5": 2.07199, "loss_cns_5": 0.64809, "loss_yns_5": 0.17068, "loss_cls_dn_0": 0.38087, "loss_box_dn_0": 0.87235, "loss_cls_dn_1": 0.25834, "loss_box_dn_1": 0.95804, "loss_cls_dn_2": 0.30924, "loss_box_dn_2": 0.91561, "loss_cls_dn_3": 0.29806, "loss_box_dn_3": 0.90538, "loss_cls_dn_4": 0.27735, "loss_box_dn_4": 0.91196, "loss_cls_dn_5": 0.32647, "loss_box_dn_5": 0.91362, "loss_dense_depth": 1.03112, "loss": 32.22479, "grad_norm": 34.35017, "time": 1.57861} -{"mode": "train", "epoch": 1, "iter": 64, "lr": 0.00013, "memory": 49163, "data_time": 0.09845, "loss_cls_0": 0.98078, "loss_box_0": 1.89158, "loss_cns_0": 0.61756, "loss_yns_0": 0.16105, "loss_cls_1": 1.07032, "loss_box_1": 2.18279, "loss_cns_1": 0.62141, "loss_yns_1": 0.16502, "loss_cls_2": 1.09242, "loss_box_2": 2.13483, "loss_cns_2": 0.63984, "loss_yns_2": 0.16653, "loss_cls_3": 1.09585, "loss_box_3": 2.15355, "loss_cns_3": 0.63875, "loss_yns_3": 0.16425, "loss_cls_4": 1.08579, "loss_box_4": 2.14955, "loss_cns_4": 0.63447, "loss_yns_4": 0.16381, "loss_cls_5": 1.09742, "loss_box_5": 2.14582, "loss_cns_5": 0.63749, "loss_yns_5": 0.17428, "loss_cls_dn_0": 0.38778, "loss_box_dn_0": 0.86909, "loss_cls_dn_1": 0.23658, "loss_box_dn_1": 0.93093, "loss_cls_dn_2": 0.28703, "loss_box_dn_2": 0.89538, "loss_cls_dn_3": 0.31112, "loss_box_dn_3": 0.89985, "loss_cls_dn_4": 0.3024, "loss_box_dn_4": 0.90662, "loss_cls_dn_5": 0.31818, "loss_box_dn_5": 0.9116, "loss_dense_depth": 0.93631, "loss": 32.05801, "grad_norm": 56.6977, "time": 1.57167} -{"mode": "train", "epoch": 1, "iter": 65, "lr": 0.00013, "memory": 49163, "data_time": 0.07609, "loss_cls_0": 0.97993, "loss_box_0": 1.89739, "loss_cns_0": 0.61773, "loss_yns_0": 0.1618, "loss_cls_1": 1.03158, "loss_box_1": 2.19058, "loss_cns_1": 0.60456, "loss_yns_1": 0.1625, "loss_cls_2": 1.06546, "loss_box_2": 2.15352, "loss_cns_2": 0.63288, "loss_yns_2": 0.16243, "loss_cls_3": 1.10046, "loss_box_3": 2.17882, "loss_cns_3": 0.63041, "loss_yns_3": 0.16214, "loss_cls_4": 1.10825, "loss_box_4": 2.19829, "loss_cns_4": 0.62738, "loss_yns_4": 0.1634, "loss_cls_5": 1.07577, "loss_box_5": 2.17478, "loss_cns_5": 0.63319, "loss_yns_5": 0.16843, "loss_cls_dn_0": 0.39438, "loss_box_dn_0": 0.87934, "loss_cls_dn_1": 0.2211, "loss_box_dn_1": 0.95428, "loss_cls_dn_2": 0.26599, "loss_box_dn_2": 0.92004, "loss_cls_dn_3": 0.31314, "loss_box_dn_3": 0.93582, "loss_cls_dn_4": 0.32337, "loss_box_dn_4": 0.96097, "loss_cls_dn_5": 0.30732, "loss_box_dn_5": 0.95471, "loss_dense_depth": 0.96625, "loss": 32.27837, "grad_norm": 57.53035, "time": 1.53077} -{"mode": "train", "epoch": 1, "iter": 66, "lr": 0.00013, "memory": 49163, "data_time": 0.07592, "loss_cls_0": 0.97299, "loss_box_0": 1.8774, "loss_cns_0": 0.62059, "loss_yns_0": 0.16263, "loss_cls_1": 1.03128, "loss_box_1": 2.12012, "loss_cns_1": 0.60717, "loss_yns_1": 0.16252, "loss_cls_2": 1.06481, "loss_box_2": 2.06005, "loss_cns_2": 0.63505, "loss_yns_2": 0.16237, "loss_cls_3": 1.06423, "loss_box_3": 2.07315, "loss_cns_3": 0.63656, "loss_yns_3": 0.16184, "loss_cls_4": 1.08328, "loss_box_4": 2.091, "loss_cns_4": 0.63561, "loss_yns_4": 0.16304, "loss_cls_5": 1.07017, "loss_box_5": 2.05762, "loss_cns_5": 0.63737, "loss_yns_5": 0.16313, "loss_cls_dn_0": 0.37225, "loss_box_dn_0": 0.87087, "loss_cls_dn_1": 0.21076, "loss_box_dn_1": 0.98066, "loss_cls_dn_2": 0.24197, "loss_box_dn_2": 0.94337, "loss_cls_dn_3": 0.28771, "loss_box_dn_3": 0.96376, "loss_cls_dn_4": 0.31162, "loss_box_dn_4": 0.99482, "loss_cls_dn_5": 0.28237, "loss_box_dn_5": 0.9893, "loss_dense_depth": 0.9204, "loss": 31.68384, "grad_norm": 40.78453, "time": 1.53943} -{"mode": "train", "epoch": 1, "iter": 67, "lr": 0.00013, "memory": 49163, "data_time": 0.07269, "loss_cls_0": 0.99873, "loss_box_0": 1.89706, "loss_cns_0": 0.61758, "loss_yns_0": 0.16273, "loss_cls_1": 1.03082, "loss_box_1": 2.10312, "loss_cns_1": 0.60797, "loss_yns_1": 0.15981, "loss_cls_2": 1.09972, "loss_box_2": 2.07766, "loss_cns_2": 0.63772, "loss_yns_2": 0.1656, "loss_cls_3": 1.08993, "loss_box_3": 2.10122, "loss_cns_3": 0.63833, "loss_yns_3": 0.16627, "loss_cls_4": 1.07926, "loss_box_4": 2.10183, "loss_cns_4": 0.63768, "loss_yns_4": 0.16476, "loss_cls_5": 1.10963, "loss_box_5": 2.155, "loss_cns_5": 0.62492, "loss_yns_5": 0.17172, "loss_cls_dn_0": 0.36725, "loss_box_dn_0": 0.87158, "loss_cls_dn_1": 0.22225, "loss_box_dn_1": 0.97984, "loss_cls_dn_2": 0.24279, "loss_box_dn_2": 0.96147, "loss_cls_dn_3": 0.26917, "loss_box_dn_3": 0.99572, "loss_cls_dn_4": 0.29954, "loss_box_dn_4": 1.03258, "loss_cls_dn_5": 0.28787, "loss_box_dn_5": 1.06222, "loss_dense_depth": 0.94636, "loss": 32.13771, "grad_norm": 66.05527, "time": 1.51397} -{"mode": "train", "epoch": 1, "iter": 68, "lr": 0.00013, "memory": 49163, "data_time": 0.07507, "loss_cls_0": 0.99756, "loss_box_0": 1.87791, "loss_cns_0": 0.61957, "loss_yns_0": 0.16114, "loss_cls_1": 1.04288, "loss_box_1": 2.08652, "loss_cns_1": 0.60751, "loss_yns_1": 0.16037, "loss_cls_2": 1.065, "loss_box_2": 2.06499, "loss_cns_2": 0.63254, "loss_yns_2": 0.16578, "loss_cls_3": 1.11892, "loss_box_3": 2.06805, "loss_cns_3": 0.63583, "loss_yns_3": 0.16381, "loss_cls_4": 1.07895, "loss_box_4": 2.07906, "loss_cns_4": 0.63328, "loss_yns_4": 0.16298, "loss_cls_5": 1.08701, "loss_box_5": 2.16241, "loss_cns_5": 0.61995, "loss_yns_5": 0.17419, "loss_cls_dn_0": 0.36318, "loss_box_dn_0": 0.87933, "loss_cls_dn_1": 0.24212, "loss_box_dn_1": 0.99935, "loss_cls_dn_2": 0.2581, "loss_box_dn_2": 0.986, "loss_cls_dn_3": 0.27545, "loss_box_dn_3": 1.02517, "loss_cls_dn_4": 0.29869, "loss_box_dn_4": 1.06597, "loss_cls_dn_5": 0.31565, "loss_box_dn_5": 1.11628, "loss_dense_depth": 0.90403, "loss": 32.19553, "grad_norm": 62.48892, "time": 1.52892} -{"mode": "train", "epoch": 1, "iter": 69, "lr": 0.00013, "memory": 49163, "data_time": 0.07708, "loss_cls_0": 0.96859, "loss_box_0": 1.82357, "loss_cns_0": 0.6236, "loss_yns_0": 0.15983, "loss_cls_1": 1.02895, "loss_box_1": 2.09904, "loss_cns_1": 0.62389, "loss_yns_1": 0.16033, "loss_cls_2": 1.05699, "loss_box_2": 2.06063, "loss_cns_2": 0.64079, "loss_yns_2": 0.16367, "loss_cls_3": 1.08658, "loss_box_3": 2.03309, "loss_cns_3": 0.64843, "loss_yns_3": 0.16038, "loss_cls_4": 1.09284, "loss_box_4": 2.04989, "loss_cns_4": 0.64511, "loss_yns_4": 0.16121, "loss_cls_5": 1.0742, "loss_box_5": 2.05662, "loss_cns_5": 0.6449, "loss_yns_5": 0.17239, "loss_cls_dn_0": 0.35223, "loss_box_dn_0": 0.86434, "loss_cls_dn_1": 0.2341, "loss_box_dn_1": 0.97966, "loss_cls_dn_2": 0.26145, "loss_box_dn_2": 0.96442, "loss_cls_dn_3": 0.27247, "loss_box_dn_3": 0.98411, "loss_cls_dn_4": 0.27354, "loss_box_dn_4": 1.01905, "loss_cls_dn_5": 0.31767, "loss_box_dn_5": 1.0352, "loss_dense_depth": 0.90659, "loss": 31.70037, "grad_norm": 51.55967, "time": 1.52558} -{"mode": "train", "epoch": 1, "iter": 70, "lr": 0.00013, "memory": 49163, "data_time": 0.07356, "loss_cls_0": 0.96195, "loss_box_0": 1.82248, "loss_cns_0": 0.62365, "loss_yns_0": 0.1627, "loss_cls_1": 1.00021, "loss_box_1": 2.10844, "loss_cns_1": 0.61926, "loss_yns_1": 0.1618, "loss_cls_2": 1.0508, "loss_box_2": 2.07091, "loss_cns_2": 0.63083, "loss_yns_2": 0.16099, "loss_cls_3": 1.05752, "loss_box_3": 2.06628, "loss_cns_3": 0.63541, "loss_yns_3": 0.16004, "loss_cls_4": 1.10151, "loss_box_4": 2.04839, "loss_cns_4": 0.64178, "loss_yns_4": 0.16184, "loss_cls_5": 1.07142, "loss_box_5": 2.02793, "loss_cns_5": 0.64525, "loss_yns_5": 0.16266, "loss_cls_dn_0": 0.37174, "loss_box_dn_0": 0.85917, "loss_cls_dn_1": 0.23263, "loss_box_dn_1": 0.9384, "loss_cls_dn_2": 0.27374, "loss_box_dn_2": 0.92541, "loss_cls_dn_3": 0.28794, "loss_box_dn_3": 0.93732, "loss_cls_dn_4": 0.27172, "loss_box_dn_4": 0.94375, "loss_cls_dn_5": 0.32052, "loss_box_dn_5": 0.9389, "loss_dense_depth": 0.89395, "loss": 31.34926, "grad_norm": 50.17034, "time": 1.5125} -{"mode": "train", "epoch": 1, "iter": 71, "lr": 0.00013, "memory": 49163, "data_time": 0.07508, "loss_cls_0": 0.94702, "loss_box_0": 1.86106, "loss_cns_0": 0.61958, "loss_yns_0": 0.16153, "loss_cls_1": 1.04193, "loss_box_1": 2.14039, "loss_cns_1": 0.61037, "loss_yns_1": 0.16074, "loss_cls_2": 1.02289, "loss_box_2": 2.0723, "loss_cns_2": 0.6296, "loss_yns_2": 0.15774, "loss_cls_3": 1.04005, "loss_box_3": 2.0811, "loss_cns_3": 0.63226, "loss_yns_3": 0.15779, "loss_cls_4": 1.074, "loss_box_4": 2.02676, "loss_cns_4": 0.6425, "loss_yns_4": 0.15742, "loss_cls_5": 1.05438, "loss_box_5": 2.06363, "loss_cns_5": 0.64384, "loss_yns_5": 0.15814, "loss_cls_dn_0": 0.36902, "loss_box_dn_0": 0.86421, "loss_cls_dn_1": 0.22118, "loss_box_dn_1": 0.9514, "loss_cls_dn_2": 0.254, "loss_box_dn_2": 0.92235, "loss_cls_dn_3": 0.27659, "loss_box_dn_3": 0.92461, "loss_cls_dn_4": 0.26336, "loss_box_dn_4": 0.90432, "loss_cls_dn_5": 0.29055, "loss_box_dn_5": 0.91202, "loss_dense_depth": 0.96959, "loss": 31.28025, "grad_norm": 56.43827, "time": 1.52139} -{"mode": "train", "epoch": 1, "iter": 72, "lr": 0.00013, "memory": 49163, "data_time": 0.07564, "loss_cls_0": 0.96395, "loss_box_0": 1.87114, "loss_cns_0": 0.62164, "loss_yns_0": 0.15992, "loss_cls_1": 1.04793, "loss_box_1": 2.08432, "loss_cns_1": 0.61968, "loss_yns_1": 0.16166, "loss_cls_2": 1.05254, "loss_box_2": 1.98814, "loss_cns_2": 0.63681, "loss_yns_2": 0.16263, "loss_cls_3": 1.05878, "loss_box_3": 1.96562, "loss_cns_3": 0.64184, "loss_yns_3": 0.16223, "loss_cls_4": 1.06036, "loss_box_4": 1.95582, "loss_cns_4": 0.64581, "loss_yns_4": 0.16021, "loss_cls_5": 1.07191, "loss_box_5": 1.98727, "loss_cns_5": 0.64438, "loss_yns_5": 0.16676, "loss_cls_dn_0": 0.37414, "loss_box_dn_0": 0.86004, "loss_cls_dn_1": 0.22855, "loss_box_dn_1": 0.85635, "loss_cls_dn_2": 0.25159, "loss_box_dn_2": 0.81749, "loss_cls_dn_3": 0.28068, "loss_box_dn_3": 0.80806, "loss_cls_dn_4": 0.28211, "loss_box_dn_4": 0.81025, "loss_cls_dn_5": 0.28857, "loss_box_dn_5": 0.83037, "loss_dense_depth": 0.90965, "loss": 30.4892, "grad_norm": 42.275, "time": 1.51548} -{"mode": "train", "epoch": 1, "iter": 73, "lr": 0.00013, "memory": 49163, "data_time": 0.07302, "loss_cls_0": 0.97927, "loss_box_0": 1.89606, "loss_cns_0": 0.61661, "loss_yns_0": 0.16047, "loss_cls_1": 1.05187, "loss_box_1": 2.13767, "loss_cns_1": 0.61436, "loss_yns_1": 0.16141, "loss_cls_2": 1.07625, "loss_box_2": 2.09794, "loss_cns_2": 0.62615, "loss_yns_2": 0.16185, "loss_cls_3": 1.09412, "loss_box_3": 2.07475, "loss_cns_3": 0.63527, "loss_yns_3": 0.16066, "loss_cls_4": 1.09723, "loss_box_4": 2.09475, "loss_cns_4": 0.63421, "loss_yns_4": 0.15868, "loss_cls_5": 1.10821, "loss_box_5": 2.07011, "loss_cns_5": 0.63566, "loss_yns_5": 0.17325, "loss_cls_dn_0": 0.36665, "loss_box_dn_0": 0.86255, "loss_cls_dn_1": 0.22915, "loss_box_dn_1": 0.8676, "loss_cls_dn_2": 0.24737, "loss_box_dn_2": 0.8413, "loss_cls_dn_3": 0.28204, "loss_box_dn_3": 0.83828, "loss_cls_dn_4": 0.30103, "loss_box_dn_4": 0.86752, "loss_cls_dn_5": 0.28435, "loss_box_dn_5": 0.88743, "loss_dense_depth": 0.93991, "loss": 31.33201, "grad_norm": 50.2276, "time": 1.5286} -{"mode": "train", "epoch": 1, "iter": 74, "lr": 0.00013, "memory": 49163, "data_time": 0.07698, "loss_cls_0": 0.98155, "loss_box_0": 1.86734, "loss_cns_0": 0.61851, "loss_yns_0": 0.15948, "loss_cls_1": 1.05481, "loss_box_1": 2.13607, "loss_cns_1": 0.61139, "loss_yns_1": 0.16, "loss_cls_2": 1.06403, "loss_box_2": 2.11959, "loss_cns_2": 0.62173, "loss_yns_2": 0.16, "loss_cls_3": 1.07917, "loss_box_3": 2.08646, "loss_cns_3": 0.63606, "loss_yns_3": 0.15995, "loss_cls_4": 1.08219, "loss_box_4": 2.08986, "loss_cns_4": 0.63513, "loss_yns_4": 0.16028, "loss_cls_5": 1.10901, "loss_box_5": 2.08581, "loss_cns_5": 0.63593, "loss_yns_5": 0.17045, "loss_cls_dn_0": 0.34963, "loss_box_dn_0": 0.85659, "loss_cls_dn_1": 0.22259, "loss_box_dn_1": 0.91301, "loss_cls_dn_2": 0.23526, "loss_box_dn_2": 0.89388, "loss_cls_dn_3": 0.26379, "loss_box_dn_3": 0.90306, "loss_cls_dn_4": 0.28728, "loss_box_dn_4": 0.93304, "loss_cls_dn_5": 0.25832, "loss_box_dn_5": 0.9672, "loss_dense_depth": 0.92805, "loss": 31.49649, "grad_norm": 51.04219, "time": 1.52568} -{"mode": "train", "epoch": 1, "iter": 75, "lr": 0.00013, "memory": 49163, "data_time": 0.07654, "loss_cls_0": 1.01602, "loss_box_0": 1.87446, "loss_cns_0": 0.61525, "loss_yns_0": 0.16235, "loss_cls_1": 1.03856, "loss_box_1": 2.15877, "loss_cns_1": 0.61588, "loss_yns_1": 0.16397, "loss_cls_2": 1.05844, "loss_box_2": 2.10718, "loss_cns_2": 0.62893, "loss_yns_2": 0.1636, "loss_cls_3": 1.07563, "loss_box_3": 2.08219, "loss_cns_3": 0.63849, "loss_yns_3": 0.16418, "loss_cls_4": 1.0614, "loss_box_4": 2.05995, "loss_cns_4": 0.6385, "loss_yns_4": 0.16345, "loss_cls_5": 1.15349, "loss_box_5": 2.09658, "loss_cns_5": 0.63233, "loss_yns_5": 0.16398, "loss_cls_dn_0": 0.33133, "loss_box_dn_0": 0.85097, "loss_cls_dn_1": 0.21572, "loss_box_dn_1": 0.94008, "loss_cls_dn_2": 0.2318, "loss_box_dn_2": 0.92535, "loss_cls_dn_3": 0.24276, "loss_box_dn_3": 0.94038, "loss_cls_dn_4": 0.25249, "loss_box_dn_4": 0.95825, "loss_cls_dn_5": 0.24185, "loss_box_dn_5": 0.99894, "loss_dense_depth": 0.90054, "loss": 31.56403, "grad_norm": 51.51884, "time": 1.58544} -{"mode": "train", "epoch": 1, "iter": 76, "lr": 0.00013, "memory": 49163, "data_time": 0.07354, "loss_cls_0": 0.99437, "loss_box_0": 1.86352, "loss_cns_0": 0.61754, "loss_yns_0": 0.16399, "loss_cls_1": 1.0614, "loss_box_1": 2.08303, "loss_cns_1": 0.62943, "loss_yns_1": 0.16377, "loss_cls_2": 1.04837, "loss_box_2": 2.0555, "loss_cns_2": 0.63889, "loss_yns_2": 0.1632, "loss_cls_3": 1.06929, "loss_box_3": 2.05749, "loss_cns_3": 0.64006, "loss_yns_3": 0.16684, "loss_cls_4": 1.06731, "loss_box_4": 2.05527, "loss_cns_4": 0.64013, "loss_yns_4": 0.16488, "loss_cls_5": 1.06892, "loss_box_5": 2.09644, "loss_cns_5": 0.63098, "loss_yns_5": 0.16539, "loss_cls_dn_0": 0.34809, "loss_box_dn_0": 0.85259, "loss_cls_dn_1": 0.21607, "loss_box_dn_1": 0.92278, "loss_cls_dn_2": 0.23306, "loss_box_dn_2": 0.91814, "loss_cls_dn_3": 0.2401, "loss_box_dn_3": 0.93201, "loss_cls_dn_4": 0.24821, "loss_box_dn_4": 0.94786, "loss_cls_dn_5": 0.26695, "loss_box_dn_5": 0.98362, "loss_dense_depth": 0.92118, "loss": 31.33666, "grad_norm": 54.40069, "time": 1.52467} -{"mode": "train", "epoch": 1, "iter": 77, "lr": 0.00013, "memory": 49163, "data_time": 0.07321, "loss_cls_0": 0.9598, "loss_box_0": 1.88174, "loss_cns_0": 0.61294, "loss_yns_0": 0.16078, "loss_cls_1": 1.03878, "loss_box_1": 2.02975, "loss_cns_1": 0.63126, "loss_yns_1": 0.16365, "loss_cls_2": 1.06086, "loss_box_2": 1.99191, "loss_cns_2": 0.64573, "loss_yns_2": 0.16695, "loss_cls_3": 1.06475, "loss_box_3": 1.98973, "loss_cns_3": 0.64769, "loss_yns_3": 0.16772, "loss_cls_4": 1.07229, "loss_box_4": 1.99887, "loss_cns_4": 0.64959, "loss_yns_4": 0.163, "loss_cls_5": 1.12291, "loss_box_5": 2.02227, "loss_cns_5": 0.64783, "loss_yns_5": 0.17529, "loss_cls_dn_0": 0.3671, "loss_box_dn_0": 0.83947, "loss_cls_dn_1": 0.21781, "loss_box_dn_1": 0.88948, "loss_cls_dn_2": 0.23483, "loss_box_dn_2": 0.86885, "loss_cls_dn_3": 0.24181, "loss_box_dn_3": 0.86998, "loss_cls_dn_4": 0.24633, "loss_box_dn_4": 0.88098, "loss_cls_dn_5": 0.29961, "loss_box_dn_5": 0.89904, "loss_dense_depth": 0.89488, "loss": 30.81624, "grad_norm": 39.48933, "time": 1.52384} -{"mode": "train", "epoch": 1, "iter": 78, "lr": 0.00013, "memory": 49163, "data_time": 0.07266, "loss_cls_0": 0.9795, "loss_box_0": 1.88956, "loss_cns_0": 0.61221, "loss_yns_0": 0.16248, "loss_cls_1": 1.0458, "loss_box_1": 2.09917, "loss_cns_1": 0.62565, "loss_yns_1": 0.1648, "loss_cls_2": 1.0793, "loss_box_2": 2.06193, "loss_cns_2": 0.64327, "loss_yns_2": 0.1676, "loss_cls_3": 1.08997, "loss_box_3": 2.06503, "loss_cns_3": 0.64588, "loss_yns_3": 0.16597, "loss_cls_4": 1.15446, "loss_box_4": 2.06501, "loss_cns_4": 0.64725, "loss_yns_4": 0.16573, "loss_cls_5": 1.08456, "loss_box_5": 2.06252, "loss_cns_5": 0.6496, "loss_yns_5": 0.1712, "loss_cls_dn_0": 0.36255, "loss_box_dn_0": 0.8331, "loss_cls_dn_1": 0.21583, "loss_box_dn_1": 0.89207, "loss_cls_dn_2": 0.23097, "loss_box_dn_2": 0.86921, "loss_cls_dn_3": 0.22654, "loss_box_dn_3": 0.86699, "loss_cls_dn_4": 0.22596, "loss_box_dn_4": 0.87225, "loss_cls_dn_5": 0.25357, "loss_box_dn_5": 0.87618, "loss_dense_depth": 0.93754, "loss": 31.16121, "grad_norm": 74.77588, "time": 1.53231} -{"mode": "train", "epoch": 1, "iter": 79, "lr": 0.00013, "memory": 49163, "data_time": 0.08274, "loss_cls_0": 0.97283, "loss_box_0": 1.88448, "loss_cns_0": 0.61829, "loss_yns_0": 0.16354, "loss_cls_1": 1.0533, "loss_box_1": 2.12303, "loss_cns_1": 0.62997, "loss_yns_1": 0.1662, "loss_cls_2": 1.07999, "loss_box_2": 2.10178, "loss_cns_2": 0.64592, "loss_yns_2": 0.16589, "loss_cls_3": 1.05797, "loss_box_3": 2.12385, "loss_cns_3": 0.64422, "loss_yns_3": 0.16934, "loss_cls_4": 1.05364, "loss_box_4": 2.10008, "loss_cns_4": 0.6462, "loss_yns_4": 0.16774, "loss_cls_5": 1.07847, "loss_box_5": 2.10438, "loss_cns_5": 0.64381, "loss_yns_5": 0.16799, "loss_cls_dn_0": 0.33801, "loss_box_dn_0": 0.83633, "loss_cls_dn_1": 0.21523, "loss_box_dn_1": 0.86372, "loss_cls_dn_2": 0.22479, "loss_box_dn_2": 0.84634, "loss_cls_dn_3": 0.22241, "loss_box_dn_3": 0.84923, "loss_cls_dn_4": 0.22679, "loss_box_dn_4": 0.85047, "loss_cls_dn_5": 0.24335, "loss_box_dn_5": 0.86058, "loss_dense_depth": 0.9176, "loss": 31.05775, "grad_norm": 58.82763, "time": 1.52973} -{"mode": "train", "epoch": 1, "iter": 80, "lr": 0.00013, "memory": 49163, "data_time": 0.08065, "loss_cls_0": 0.956, "loss_box_0": 1.88521, "loss_cns_0": 0.62286, "loss_yns_0": 0.16228, "loss_cls_1": 1.02809, "loss_box_1": 2.16508, "loss_cns_1": 0.62915, "loss_yns_1": 0.16456, "loss_cls_2": 1.07561, "loss_box_2": 2.09346, "loss_cns_2": 0.6423, "loss_yns_2": 0.16705, "loss_cls_3": 1.11245, "loss_box_3": 2.12289, "loss_cns_3": 0.64619, "loss_yns_3": 0.17445, "loss_cls_4": 1.07866, "loss_box_4": 2.09148, "loss_cns_4": 0.64662, "loss_yns_4": 0.16642, "loss_cls_5": 1.09754, "loss_box_5": 2.08372, "loss_cns_5": 0.64643, "loss_yns_5": 0.16571, "loss_cls_dn_0": 0.33872, "loss_box_dn_0": 0.82958, "loss_cls_dn_1": 0.20858, "loss_box_dn_1": 0.85995, "loss_cls_dn_2": 0.21809, "loss_box_dn_2": 0.83431, "loss_cls_dn_3": 0.22957, "loss_box_dn_3": 0.84556, "loss_cls_dn_4": 0.23806, "loss_box_dn_4": 0.85527, "loss_cls_dn_5": 0.24234, "loss_box_dn_5": 0.87209, "loss_dense_depth": 0.88889, "loss": 31.08518, "grad_norm": 57.54979, "time": 1.52345} -{"mode": "train", "epoch": 1, "iter": 81, "lr": 0.00013, "memory": 49163, "data_time": 0.15053, "loss_cls_0": 0.9787, "loss_box_0": 1.90468, "loss_cns_0": 0.61665, "loss_yns_0": 0.16136, "loss_cls_1": 1.03007, "loss_box_1": 2.1511, "loss_cns_1": 0.62216, "loss_yns_1": 0.16322, "loss_cls_2": 1.09078, "loss_box_2": 2.11197, "loss_cns_2": 0.63374, "loss_yns_2": 0.16564, "loss_cls_3": 1.10858, "loss_box_3": 2.12949, "loss_cns_3": 0.63724, "loss_yns_3": 0.16967, "loss_cls_4": 1.08957, "loss_box_4": 2.12336, "loss_cns_4": 0.63346, "loss_yns_4": 0.16427, "loss_cls_5": 1.12457, "loss_box_5": 2.12257, "loss_cns_5": 0.63546, "loss_yns_5": 0.16428, "loss_cls_dn_0": 0.33962, "loss_box_dn_0": 0.8372, "loss_cls_dn_1": 0.20447, "loss_box_dn_1": 0.89977, "loss_cls_dn_2": 0.21914, "loss_box_dn_2": 0.8968, "loss_cls_dn_3": 0.22789, "loss_box_dn_3": 0.91082, "loss_cls_dn_4": 0.22502, "loss_box_dn_4": 0.94568, "loss_cls_dn_5": 0.22825, "loss_box_dn_5": 0.96869, "loss_dense_depth": 0.90935, "loss": 31.54531, "grad_norm": 66.62029, "time": 1.5942} -{"mode": "train", "epoch": 1, "iter": 82, "lr": 0.00013, "memory": 49163, "data_time": 0.07425, "loss_cls_0": 1.00725, "loss_box_0": 1.90792, "loss_cns_0": 0.61904, "loss_yns_0": 0.15751, "loss_cls_1": 1.02057, "loss_box_1": 2.11842, "loss_cns_1": 0.62595, "loss_yns_1": 0.16118, "loss_cls_2": 1.07748, "loss_box_2": 2.07623, "loss_cns_2": 0.63361, "loss_yns_2": 0.16117, "loss_cls_3": 1.08094, "loss_box_3": 2.0716, "loss_cns_3": 0.63531, "loss_yns_3": 0.16256, "loss_cls_4": 1.08471, "loss_box_4": 2.08637, "loss_cns_4": 0.63212, "loss_yns_4": 0.16126, "loss_cls_5": 1.16211, "loss_box_5": 2.07761, "loss_cns_5": 0.63792, "loss_yns_5": 0.1665, "loss_cls_dn_0": 0.33037, "loss_box_dn_0": 0.84883, "loss_cls_dn_1": 0.20571, "loss_box_dn_1": 0.95088, "loss_cls_dn_2": 0.21814, "loss_box_dn_2": 0.96133, "loss_cls_dn_3": 0.22275, "loss_box_dn_3": 0.97603, "loss_cls_dn_4": 0.21867, "loss_box_dn_4": 1.024, "loss_cls_dn_5": 0.22093, "loss_box_dn_5": 1.04088, "loss_dense_depth": 0.9458, "loss": 31.68967, "grad_norm": 64.41386, "time": 1.55016} -{"mode": "train", "epoch": 1, "iter": 83, "lr": 0.00013, "memory": 49163, "data_time": 0.07339, "loss_cls_0": 0.9896, "loss_box_0": 1.87341, "loss_cns_0": 0.62503, "loss_yns_0": 0.16004, "loss_cls_1": 1.03403, "loss_box_1": 2.04611, "loss_cns_1": 0.63084, "loss_yns_1": 0.16182, "loss_cls_2": 1.0584, "loss_box_2": 1.99763, "loss_cns_2": 0.64235, "loss_yns_2": 0.16231, "loss_cls_3": 1.06388, "loss_box_3": 2.00356, "loss_cns_3": 0.64291, "loss_yns_3": 0.16321, "loss_cls_4": 1.06809, "loss_box_4": 2.00583, "loss_cns_4": 0.64368, "loss_yns_4": 0.16401, "loss_cls_5": 1.0837, "loss_box_5": 1.97754, "loss_cns_5": 0.64538, "loss_yns_5": 0.16331, "loss_cls_dn_0": 0.32418, "loss_box_dn_0": 0.84976, "loss_cls_dn_1": 0.20646, "loss_box_dn_1": 0.90043, "loss_cls_dn_2": 0.21381, "loss_box_dn_2": 0.90326, "loss_cls_dn_3": 0.21569, "loss_box_dn_3": 0.9265, "loss_cls_dn_4": 0.22493, "loss_box_dn_4": 0.96537, "loss_cls_dn_5": 0.23505, "loss_box_dn_5": 0.96728, "loss_dense_depth": 0.90856, "loss": 30.84796, "grad_norm": 53.65528, "time": 1.58038} -{"mode": "train", "epoch": 1, "iter": 84, "lr": 0.00013, "memory": 49163, "data_time": 0.09782, "loss_cls_0": 0.93011, "loss_box_0": 1.86417, "loss_cns_0": 0.62459, "loss_yns_0": 0.15938, "loss_cls_1": 1.00617, "loss_box_1": 2.07324, "loss_cns_1": 0.62861, "loss_yns_1": 0.1616, "loss_cls_2": 1.069, "loss_box_2": 2.00928, "loss_cns_2": 0.64423, "loss_yns_2": 0.16093, "loss_cls_3": 1.07693, "loss_box_3": 2.0226, "loss_cns_3": 0.6465, "loss_yns_3": 0.16205, "loss_cls_4": 1.08747, "loss_box_4": 2.00309, "loss_cns_4": 0.64925, "loss_yns_4": 0.16197, "loss_cls_5": 1.11052, "loss_box_5": 1.9895, "loss_cns_5": 0.64606, "loss_yns_5": 0.15969, "loss_cls_dn_0": 0.31515, "loss_box_dn_0": 0.83629, "loss_cls_dn_1": 0.20564, "loss_box_dn_1": 0.8779, "loss_cls_dn_2": 0.21449, "loss_box_dn_2": 0.87196, "loss_cls_dn_3": 0.2232, "loss_box_dn_3": 0.90116, "loss_cls_dn_4": 0.24461, "loss_box_dn_4": 0.92508, "loss_cls_dn_5": 0.27228, "loss_box_dn_5": 0.93391, "loss_dense_depth": 0.8726, "loss": 30.7412, "grad_norm": 67.65077, "time": 1.57014} -{"mode": "train", "epoch": 1, "iter": 85, "lr": 0.00013, "memory": 49163, "data_time": 0.07817, "loss_cls_0": 0.96869, "loss_box_0": 1.86283, "loss_cns_0": 0.62127, "loss_yns_0": 0.16016, "loss_cls_1": 1.00402, "loss_box_1": 2.12358, "loss_cns_1": 0.61501, "loss_yns_1": 0.1624, "loss_cls_2": 1.06668, "loss_box_2": 2.05748, "loss_cns_2": 0.63895, "loss_yns_2": 0.16068, "loss_cls_3": 1.08343, "loss_box_3": 2.05861, "loss_cns_3": 0.64114, "loss_yns_3": 0.16467, "loss_cls_4": 1.0962, "loss_box_4": 2.03258, "loss_cns_4": 0.64301, "loss_yns_4": 0.16103, "loss_cls_5": 1.12406, "loss_box_5": 2.07097, "loss_cns_5": 0.63781, "loss_yns_5": 0.16116, "loss_cls_dn_0": 0.33985, "loss_box_dn_0": 0.85443, "loss_cls_dn_1": 0.20987, "loss_box_dn_1": 0.88768, "loss_cls_dn_2": 0.21899, "loss_box_dn_2": 0.86447, "loss_cls_dn_3": 0.2269, "loss_box_dn_3": 0.87662, "loss_cls_dn_4": 0.24146, "loss_box_dn_4": 0.88588, "loss_cls_dn_5": 0.28004, "loss_box_dn_5": 0.91576, "loss_dense_depth": 0.95205, "loss": 31.07042, "grad_norm": 51.46324, "time": 1.53497} -{"mode": "train", "epoch": 1, "iter": 86, "lr": 0.00013, "memory": 49163, "data_time": 0.07408, "loss_cls_0": 0.93931, "loss_box_0": 1.86804, "loss_cns_0": 0.62018, "loss_yns_0": 0.16195, "loss_cls_1": 0.99933, "loss_box_1": 2.03759, "loss_cns_1": 0.61731, "loss_yns_1": 0.16253, "loss_cls_2": 1.04494, "loss_box_2": 1.98767, "loss_cns_2": 0.64119, "loss_yns_2": 0.16137, "loss_cls_3": 1.03522, "loss_box_3": 1.97246, "loss_cns_3": 0.64132, "loss_yns_3": 0.16332, "loss_cls_4": 1.06323, "loss_box_4": 1.94347, "loss_cns_4": 0.64663, "loss_yns_4": 0.16169, "loss_cls_5": 1.09034, "loss_box_5": 1.9683, "loss_cns_5": 0.64357, "loss_yns_5": 0.1613, "loss_cls_dn_0": 0.31844, "loss_box_dn_0": 0.8496, "loss_cls_dn_1": 0.20811, "loss_box_dn_1": 0.8492, "loss_cls_dn_2": 0.21931, "loss_box_dn_2": 0.83158, "loss_cls_dn_3": 0.21677, "loss_box_dn_3": 0.82041, "loss_cls_dn_4": 0.22202, "loss_box_dn_4": 0.82061, "loss_cls_dn_5": 0.25773, "loss_box_dn_5": 0.84306, "loss_dense_depth": 0.95418, "loss": 30.14329, "grad_norm": 57.5528, "time": 1.53337} -{"mode": "train", "epoch": 1, "iter": 87, "lr": 0.00013, "memory": 49163, "data_time": 0.07641, "loss_cls_0": 0.94578, "loss_box_0": 1.87308, "loss_cns_0": 0.62077, "loss_yns_0": 0.1639, "loss_cls_1": 0.99735, "loss_box_1": 2.07885, "loss_cns_1": 0.62758, "loss_yns_1": 0.16304, "loss_cls_2": 1.04684, "loss_box_2": 2.04768, "loss_cns_2": 0.64426, "loss_yns_2": 0.15918, "loss_cls_3": 1.03289, "loss_box_3": 2.07328, "loss_cns_3": 0.64263, "loss_yns_3": 0.16023, "loss_cls_4": 1.0504, "loss_box_4": 2.03731, "loss_cns_4": 0.6474, "loss_yns_4": 0.16137, "loss_cls_5": 1.0696, "loss_box_5": 2.03827, "loss_cns_5": 0.6477, "loss_yns_5": 0.16009, "loss_cls_dn_0": 0.30011, "loss_box_dn_0": 0.84369, "loss_cls_dn_1": 0.20743, "loss_box_dn_1": 0.82215, "loss_cls_dn_2": 0.21963, "loss_box_dn_2": 0.81554, "loss_cls_dn_3": 0.21719, "loss_box_dn_3": 0.80742, "loss_cls_dn_4": 0.21795, "loss_box_dn_4": 0.80435, "loss_cls_dn_5": 0.24451, "loss_box_dn_5": 0.80507, "loss_dense_depth": 0.93012, "loss": 30.32465, "grad_norm": 62.19807, "time": 1.50611} -{"mode": "train", "epoch": 1, "iter": 88, "lr": 0.00013, "memory": 49163, "data_time": 0.07202, "loss_cls_0": 0.97054, "loss_box_0": 1.8445, "loss_cns_0": 0.61644, "loss_yns_0": 0.16233, "loss_cls_1": 0.99376, "loss_box_1": 2.0298, "loss_cns_1": 0.62539, "loss_yns_1": 0.16036, "loss_cls_2": 1.03765, "loss_box_2": 1.99021, "loss_cns_2": 0.64127, "loss_yns_2": 0.15937, "loss_cls_3": 1.04127, "loss_box_3": 2.01376, "loss_cns_3": 0.63995, "loss_yns_3": 0.15951, "loss_cls_4": 1.04833, "loss_box_4": 1.99626, "loss_cns_4": 0.64252, "loss_yns_4": 0.16026, "loss_cls_5": 1.06304, "loss_box_5": 2.00092, "loss_cns_5": 0.64203, "loss_yns_5": 0.16136, "loss_cls_dn_0": 0.29997, "loss_box_dn_0": 0.84316, "loss_cls_dn_1": 0.20431, "loss_box_dn_1": 0.80046, "loss_cls_dn_2": 0.21238, "loss_box_dn_2": 0.79315, "loss_cls_dn_3": 0.21488, "loss_box_dn_3": 0.79641, "loss_cls_dn_4": 0.21919, "loss_box_dn_4": 0.80289, "loss_cls_dn_5": 0.2333, "loss_box_dn_5": 0.809, "loss_dense_depth": 0.87112, "loss": 29.90104, "grad_norm": 57.87001, "time": 1.50613} -{"mode": "train", "epoch": 1, "iter": 89, "lr": 0.00014, "memory": 49163, "data_time": 0.07665, "loss_cls_0": 0.9302, "loss_box_0": 1.81396, "loss_cns_0": 0.61483, "loss_yns_0": 0.16117, "loss_cls_1": 0.99284, "loss_box_1": 2.00979, "loss_cns_1": 0.62229, "loss_yns_1": 0.15931, "loss_cls_2": 1.02458, "loss_box_2": 1.94844, "loss_cns_2": 0.64242, "loss_yns_2": 0.15808, "loss_cls_3": 1.02368, "loss_box_3": 1.94575, "loss_cns_3": 0.64436, "loss_yns_3": 0.15816, "loss_cls_4": 1.03567, "loss_box_4": 1.95227, "loss_cns_4": 0.64696, "loss_yns_4": 0.1607, "loss_cls_5": 1.07584, "loss_box_5": 1.96484, "loss_cns_5": 0.64669, "loss_yns_5": 0.16252, "loss_cls_dn_0": 0.30569, "loss_box_dn_0": 0.8297, "loss_cls_dn_1": 0.19976, "loss_box_dn_1": 0.81338, "loss_cls_dn_2": 0.20799, "loss_box_dn_2": 0.80676, "loss_cls_dn_3": 0.2092, "loss_box_dn_3": 0.81672, "loss_cls_dn_4": 0.21529, "loss_box_dn_4": 0.83917, "loss_cls_dn_5": 0.22699, "loss_box_dn_5": 0.86627, "loss_dense_depth": 0.89024, "loss": 29.72252, "grad_norm": 51.60568, "time": 1.51987} -{"mode": "train", "epoch": 1, "iter": 90, "lr": 0.00014, "memory": 49163, "data_time": 0.07636, "loss_cls_0": 0.95961, "loss_box_0": 1.81866, "loss_cns_0": 0.61381, "loss_yns_0": 0.15778, "loss_cls_1": 1.00828, "loss_box_1": 2.03608, "loss_cns_1": 0.63183, "loss_yns_1": 0.15909, "loss_cls_2": 1.04071, "loss_box_2": 2.02042, "loss_cns_2": 0.64305, "loss_yns_2": 0.15802, "loss_cls_3": 1.03963, "loss_box_3": 2.01348, "loss_cns_3": 0.64669, "loss_yns_3": 0.16034, "loss_cls_4": 1.06129, "loss_box_4": 2.0447, "loss_cns_4": 0.64715, "loss_yns_4": 0.1588, "loss_cls_5": 1.07268, "loss_box_5": 2.04344, "loss_cns_5": 0.64788, "loss_yns_5": 0.16028, "loss_cls_dn_0": 0.32625, "loss_box_dn_0": 0.84263, "loss_cls_dn_1": 0.20124, "loss_box_dn_1": 0.85375, "loss_cls_dn_2": 0.21311, "loss_box_dn_2": 0.8575, "loss_cls_dn_3": 0.21424, "loss_box_dn_3": 0.87158, "loss_cls_dn_4": 0.21831, "loss_box_dn_4": 0.91004, "loss_cls_dn_5": 0.23809, "loss_box_dn_5": 0.94059, "loss_dense_depth": 0.88713, "loss": 30.51814, "grad_norm": 67.63343, "time": 1.52978} -{"mode": "train", "epoch": 1, "iter": 91, "lr": 0.00014, "memory": 49163, "data_time": 0.07359, "loss_cls_0": 0.95262, "loss_box_0": 1.79538, "loss_cns_0": 0.60895, "loss_yns_0": 0.15523, "loss_cls_1": 0.98702, "loss_box_1": 2.06908, "loss_cns_1": 0.62678, "loss_yns_1": 0.15583, "loss_cls_2": 1.02581, "loss_box_2": 2.05712, "loss_cns_2": 0.63827, "loss_yns_2": 0.15676, "loss_cls_3": 1.03188, "loss_box_3": 2.04329, "loss_cns_3": 0.64541, "loss_yns_3": 0.16012, "loss_cls_4": 1.05408, "loss_box_4": 2.07519, "loss_cns_4": 0.64459, "loss_yns_4": 0.15458, "loss_cls_5": 1.11257, "loss_box_5": 2.05917, "loss_cns_5": 0.64653, "loss_yns_5": 0.15491, "loss_cls_dn_0": 0.30862, "loss_box_dn_0": 0.82685, "loss_cls_dn_1": 0.208, "loss_box_dn_1": 0.8809, "loss_cls_dn_2": 0.22085, "loss_box_dn_2": 0.87146, "loss_cls_dn_3": 0.21618, "loss_box_dn_3": 0.88263, "loss_cls_dn_4": 0.21938, "loss_box_dn_4": 0.92565, "loss_cls_dn_5": 0.23812, "loss_box_dn_5": 0.95087, "loss_dense_depth": 0.89679, "loss": 30.65751, "grad_norm": 69.85265, "time": 1.53425} -{"mode": "train", "epoch": 1, "iter": 92, "lr": 0.00014, "memory": 49163, "data_time": 0.0741, "loss_cls_0": 0.96031, "loss_box_0": 1.80609, "loss_cns_0": 0.61575, "loss_yns_0": 0.15453, "loss_cls_1": 0.99345, "loss_box_1": 2.09415, "loss_cns_1": 0.62535, "loss_yns_1": 0.15597, "loss_cls_2": 1.01584, "loss_box_2": 2.03757, "loss_cns_2": 0.64086, "loss_yns_2": 0.15624, "loss_cls_3": 1.05056, "loss_box_3": 2.03218, "loss_cns_3": 0.64752, "loss_yns_3": 0.16065, "loss_cls_4": 1.04461, "loss_box_4": 2.04541, "loss_cns_4": 0.64655, "loss_yns_4": 0.15732, "loss_cls_5": 1.13248, "loss_box_5": 2.01231, "loss_cns_5": 0.64946, "loss_yns_5": 0.15458, "loss_cls_dn_0": 0.28916, "loss_box_dn_0": 0.81999, "loss_cls_dn_1": 0.2018, "loss_box_dn_1": 0.84906, "loss_cls_dn_2": 0.20939, "loss_box_dn_2": 0.81638, "loss_cls_dn_3": 0.20558, "loss_box_dn_3": 0.82064, "loss_cls_dn_4": 0.20838, "loss_box_dn_4": 0.84431, "loss_cls_dn_5": 0.22495, "loss_box_dn_5": 0.8577, "loss_dense_depth": 0.90161, "loss": 30.2387, "grad_norm": 43.8536, "time": 1.53997} -{"mode": "train", "epoch": 1, "iter": 93, "lr": 0.00014, "memory": 49163, "data_time": 0.07587, "loss_cls_0": 0.94445, "loss_box_0": 1.80528, "loss_cns_0": 0.62082, "loss_yns_0": 0.15293, "loss_cls_1": 0.98648, "loss_box_1": 2.03548, "loss_cns_1": 0.61981, "loss_yns_1": 0.15538, "loss_cls_2": 1.02547, "loss_box_2": 1.95256, "loss_cns_2": 0.64294, "loss_yns_2": 0.15775, "loss_cls_3": 1.03502, "loss_box_3": 1.98185, "loss_cns_3": 0.64537, "loss_yns_3": 0.15745, "loss_cls_4": 1.04144, "loss_box_4": 1.9768, "loss_cns_4": 0.64541, "loss_yns_4": 0.15675, "loss_cls_5": 1.1017, "loss_box_5": 1.97425, "loss_cns_5": 0.64592, "loss_yns_5": 0.15724, "loss_cls_dn_0": 0.27979, "loss_box_dn_0": 0.8203, "loss_cls_dn_1": 0.19827, "loss_box_dn_1": 0.83419, "loss_cls_dn_2": 0.20642, "loss_box_dn_2": 0.79561, "loss_cls_dn_3": 0.20449, "loss_box_dn_3": 0.8025, "loss_cls_dn_4": 0.2117, "loss_box_dn_4": 0.80302, "loss_cls_dn_5": 0.22755, "loss_box_dn_5": 0.81058, "loss_dense_depth": 0.84273, "loss": 29.65568, "grad_norm": 57.41175, "time": 1.5297} -{"mode": "train", "epoch": 1, "iter": 94, "lr": 0.00014, "memory": 49163, "data_time": 0.07788, "loss_cls_0": 0.94831, "loss_box_0": 1.83254, "loss_cns_0": 0.61862, "loss_yns_0": 0.15284, "loss_cls_1": 1.0025, "loss_box_1": 2.05962, "loss_cns_1": 0.62056, "loss_yns_1": 0.15766, "loss_cls_2": 1.00951, "loss_box_2": 1.99719, "loss_cns_2": 0.63803, "loss_yns_2": 0.15603, "loss_cls_3": 1.02747, "loss_box_3": 2.02805, "loss_cns_3": 0.64076, "loss_yns_3": 0.15644, "loss_cls_4": 1.02925, "loss_box_4": 2.0286, "loss_cns_4": 0.64058, "loss_yns_4": 0.15905, "loss_cls_5": 1.04151, "loss_box_5": 2.0331, "loss_cns_5": 0.6465, "loss_yns_5": 0.15605, "loss_cls_dn_0": 0.29039, "loss_box_dn_0": 0.81737, "loss_cls_dn_1": 0.20725, "loss_box_dn_1": 0.82535, "loss_cls_dn_2": 0.21147, "loss_box_dn_2": 0.8007, "loss_cls_dn_3": 0.21409, "loss_box_dn_3": 0.81506, "loss_cls_dn_4": 0.21936, "loss_box_dn_4": 0.81656, "loss_cls_dn_5": 0.23149, "loss_box_dn_5": 0.81756, "loss_dense_depth": 0.89434, "loss": 29.94175, "grad_norm": 64.89999, "time": 1.52635} -{"mode": "train", "epoch": 1, "iter": 95, "lr": 0.00014, "memory": 49163, "data_time": 0.07727, "loss_cls_0": 0.95731, "loss_box_0": 1.83829, "loss_cns_0": 0.62006, "loss_yns_0": 0.1538, "loss_cls_1": 1.01503, "loss_box_1": 1.99589, "loss_cns_1": 0.6309, "loss_yns_1": 0.15922, "loss_cls_2": 1.02512, "loss_box_2": 1.94344, "loss_cns_2": 0.6419, "loss_yns_2": 0.1587, "loss_cls_3": 1.06151, "loss_box_3": 1.95175, "loss_cns_3": 0.64752, "loss_yns_3": 0.15865, "loss_cls_4": 1.07957, "loss_box_4": 1.97398, "loss_cns_4": 0.64665, "loss_yns_4": 0.16019, "loss_cls_5": 1.10776, "loss_box_5": 1.93844, "loss_cns_5": 0.653, "loss_yns_5": 0.15506, "loss_cls_dn_0": 0.29462, "loss_box_dn_0": 0.81599, "loss_cls_dn_1": 0.209, "loss_box_dn_1": 0.79797, "loss_cls_dn_2": 0.20784, "loss_box_dn_2": 0.77769, "loss_cls_dn_3": 0.2119, "loss_box_dn_3": 0.79236, "loss_cls_dn_4": 0.21515, "loss_box_dn_4": 0.80578, "loss_cls_dn_5": 0.22188, "loss_box_dn_5": 0.79019, "loss_dense_depth": 0.88172, "loss": 29.69583, "grad_norm": 62.36982, "time": 1.5731} -{"mode": "train", "epoch": 1, "iter": 96, "lr": 0.00014, "memory": 49163, "data_time": 0.07859, "loss_cls_0": 0.97279, "loss_box_0": 1.89557, "loss_cns_0": 0.61531, "loss_yns_0": 0.15548, "loss_cls_1": 1.03028, "loss_box_1": 2.01632, "loss_cns_1": 0.62785, "loss_yns_1": 0.15785, "loss_cls_2": 1.05315, "loss_box_2": 1.98223, "loss_cns_2": 0.63508, "loss_yns_2": 0.16071, "loss_cls_3": 1.11331, "loss_box_3": 1.95008, "loss_cns_3": 0.64326, "loss_yns_3": 0.16123, "loss_cls_4": 1.08309, "loss_box_4": 1.96959, "loss_cns_4": 0.64372, "loss_yns_4": 0.15941, "loss_cls_5": 1.13537, "loss_box_5": 1.95257, "loss_cns_5": 0.64683, "loss_yns_5": 0.15895, "loss_cls_dn_0": 0.29661, "loss_box_dn_0": 0.8283, "loss_cls_dn_1": 0.20491, "loss_box_dn_1": 0.7977, "loss_cls_dn_2": 0.2064, "loss_box_dn_2": 0.78464, "loss_cls_dn_3": 0.21341, "loss_box_dn_3": 0.79393, "loss_cls_dn_4": 0.21518, "loss_box_dn_4": 0.80849, "loss_cls_dn_5": 0.22455, "loss_box_dn_5": 0.807, "loss_dense_depth": 0.87836, "loss": 29.9795, "grad_norm": 38.13988, "time": 1.53394} -{"mode": "train", "epoch": 1, "iter": 97, "lr": 0.00014, "memory": 49163, "data_time": 0.07591, "loss_cls_0": 0.97465, "loss_box_0": 1.85127, "loss_cns_0": 0.6193, "loss_yns_0": 0.15505, "loss_cls_1": 1.00447, "loss_box_1": 2.04193, "loss_cns_1": 0.62021, "loss_yns_1": 0.15751, "loss_cls_2": 1.03732, "loss_box_2": 2.0084, "loss_cns_2": 0.63412, "loss_yns_2": 0.15768, "loss_cls_3": 1.08806, "loss_box_3": 1.95875, "loss_cns_3": 0.64312, "loss_yns_3": 0.15846, "loss_cls_4": 1.08898, "loss_box_4": 1.94459, "loss_cns_4": 0.64472, "loss_yns_4": 0.15915, "loss_cls_5": 1.14686, "loss_box_5": 1.94802, "loss_cns_5": 0.64436, "loss_yns_5": 0.16003, "loss_cls_dn_0": 0.28117, "loss_box_dn_0": 0.82252, "loss_cls_dn_1": 0.20039, "loss_box_dn_1": 0.80654, "loss_cls_dn_2": 0.20344, "loss_box_dn_2": 0.79313, "loss_cls_dn_3": 0.20961, "loss_box_dn_3": 0.79012, "loss_cls_dn_4": 0.21067, "loss_box_dn_4": 0.79883, "loss_cls_dn_5": 0.22699, "loss_box_dn_5": 0.81543, "loss_dense_depth": 0.88141, "loss": 29.88726, "grad_norm": 46.1717, "time": 1.5406} -{"mode": "train", "epoch": 1, "iter": 98, "lr": 0.00014, "memory": 49163, "data_time": 0.07434, "loss_cls_0": 0.98404, "loss_box_0": 1.81672, "loss_cns_0": 0.62541, "loss_yns_0": 0.1542, "loss_cls_1": 1.01426, "loss_box_1": 1.96749, "loss_cns_1": 0.62709, "loss_yns_1": 0.15833, "loss_cls_2": 1.07034, "loss_box_2": 1.94044, "loss_cns_2": 0.64129, "loss_yns_2": 0.1582, "loss_cls_3": 1.03381, "loss_box_3": 1.90814, "loss_cns_3": 0.64451, "loss_yns_3": 0.15677, "loss_cls_4": 1.05022, "loss_box_4": 1.89237, "loss_cns_4": 0.64751, "loss_yns_4": 0.16245, "loss_cls_5": 1.10618, "loss_box_5": 1.91437, "loss_cns_5": 0.6487, "loss_yns_5": 0.15758, "loss_cls_dn_0": 0.27094, "loss_box_dn_0": 0.82325, "loss_cls_dn_1": 0.19731, "loss_box_dn_1": 0.84607, "loss_cls_dn_2": 0.20484, "loss_box_dn_2": 0.82865, "loss_cls_dn_3": 0.20658, "loss_box_dn_3": 0.82536, "loss_cls_dn_4": 0.21204, "loss_box_dn_4": 0.8368, "loss_cls_dn_5": 0.23065, "loss_box_dn_5": 0.85593, "loss_dense_depth": 0.85574, "loss": 29.67459, "grad_norm": 42.15149, "time": 1.52127} -{"mode": "train", "epoch": 1, "iter": 99, "lr": 0.00014, "memory": 49163, "data_time": 0.08057, "loss_cls_0": 0.97066, "loss_box_0": 1.82121, "loss_cns_0": 0.62436, "loss_yns_0": 0.15499, "loss_cls_1": 1.01661, "loss_box_1": 2.01861, "loss_cns_1": 0.62507, "loss_yns_1": 0.1588, "loss_cls_2": 1.06853, "loss_box_2": 1.94863, "loss_cns_2": 0.64043, "loss_yns_2": 0.15774, "loss_cls_3": 1.06517, "loss_box_3": 1.94465, "loss_cns_3": 0.64243, "loss_yns_3": 0.15803, "loss_cls_4": 1.07478, "loss_box_4": 1.92815, "loss_cns_4": 0.64614, "loss_yns_4": 0.16417, "loss_cls_5": 1.07637, "loss_box_5": 1.94081, "loss_cns_5": 0.64696, "loss_yns_5": 0.15605, "loss_cls_dn_0": 0.27831, "loss_box_dn_0": 0.81898, "loss_cls_dn_1": 0.19321, "loss_box_dn_1": 0.84235, "loss_cls_dn_2": 0.19991, "loss_box_dn_2": 0.81759, "loss_cls_dn_3": 0.20083, "loss_box_dn_3": 0.82283, "loss_cls_dn_4": 0.20479, "loss_box_dn_4": 0.83069, "loss_cls_dn_5": 0.21953, "loss_box_dn_5": 0.83575, "loss_dense_depth": 0.87263, "loss": 29.78673, "grad_norm": 42.8888, "time": 1.52864} -{"mode": "train", "epoch": 1, "iter": 100, "lr": 0.00014, "memory": 49163, "data_time": 0.07772, "loss_cls_0": 0.94213, "loss_box_0": 1.8364, "loss_cns_0": 0.62143, "loss_yns_0": 0.15571, "loss_cls_1": 1.00108, "loss_box_1": 2.04683, "loss_cns_1": 0.62491, "loss_yns_1": 0.15904, "loss_cls_2": 1.02915, "loss_box_2": 1.96582, "loss_cns_2": 0.638, "loss_yns_2": 0.15828, "loss_cls_3": 1.04048, "loss_box_3": 1.95771, "loss_cns_3": 0.6406, "loss_yns_3": 0.15853, "loss_cls_4": 1.05234, "loss_box_4": 1.93934, "loss_cns_4": 0.64311, "loss_yns_4": 0.15762, "loss_cls_5": 1.06761, "loss_box_5": 1.94023, "loss_cns_5": 0.64309, "loss_yns_5": 0.1568, "loss_cls_dn_0": 0.27581, "loss_box_dn_0": 0.82664, "loss_cls_dn_1": 0.1917, "loss_box_dn_1": 0.84561, "loss_cls_dn_2": 0.19778, "loss_box_dn_2": 0.81328, "loss_cls_dn_3": 0.19715, "loss_box_dn_3": 0.81898, "loss_cls_dn_4": 0.20393, "loss_box_dn_4": 0.81868, "loss_cls_dn_5": 0.21974, "loss_box_dn_5": 0.82697, "loss_dense_depth": 0.84801, "loss": 29.66052, "grad_norm": 31.2682, "time": 1.53364} -{"mode": "train", "epoch": 1, "iter": 101, "lr": 0.00014, "memory": 49163, "data_time": 0.15587, "loss_cls_0": 0.97415, "loss_box_0": 1.81283, "loss_cns_0": 0.61979, "loss_yns_0": 0.15633, "loss_cls_1": 1.01173, "loss_box_1": 2.00068, "loss_cns_1": 0.62776, "loss_yns_1": 0.15833, "loss_cls_2": 1.04283, "loss_box_2": 1.94065, "loss_cns_2": 0.63832, "loss_yns_2": 0.15824, "loss_cls_3": 1.04338, "loss_box_3": 1.91915, "loss_cns_3": 0.64317, "loss_yns_3": 0.15816, "loss_cls_4": 1.04369, "loss_box_4": 1.90297, "loss_cns_4": 0.64643, "loss_yns_4": 0.16143, "loss_cls_5": 1.06145, "loss_box_5": 1.92315, "loss_cns_5": 0.64415, "loss_yns_5": 0.15725, "loss_cls_dn_0": 0.2702, "loss_box_dn_0": 0.8216, "loss_cls_dn_1": 0.19486, "loss_box_dn_1": 0.85142, "loss_cls_dn_2": 0.20152, "loss_box_dn_2": 0.81917, "loss_cls_dn_3": 0.19827, "loss_box_dn_3": 0.81364, "loss_cls_dn_4": 0.20348, "loss_box_dn_4": 0.8116, "loss_cls_dn_5": 0.2137, "loss_box_dn_5": 0.83377, "loss_dense_depth": 0.90003, "loss": 29.57927, "grad_norm": 42.85645, "time": 1.60119} -{"mode": "train", "epoch": 1, "iter": 102, "lr": 0.00014, "memory": 49163, "data_time": 0.07343, "loss_cls_0": 0.95464, "loss_box_0": 1.82998, "loss_cns_0": 0.6125, "loss_yns_0": 0.15686, "loss_cls_1": 1.0227, "loss_box_1": 1.99303, "loss_cns_1": 0.63564, "loss_yns_1": 0.1609, "loss_cls_2": 1.07707, "loss_box_2": 1.95946, "loss_cns_2": 0.64197, "loss_yns_2": 0.16108, "loss_cls_3": 1.04673, "loss_box_3": 1.9258, "loss_cns_3": 0.6476, "loss_yns_3": 0.16059, "loss_cls_4": 1.06641, "loss_box_4": 1.92411, "loss_cns_4": 0.64774, "loss_yns_4": 0.16911, "loss_cls_5": 1.08093, "loss_box_5": 1.93529, "loss_cns_5": 0.64613, "loss_yns_5": 0.15979, "loss_cls_dn_0": 0.27572, "loss_box_dn_0": 0.82944, "loss_cls_dn_1": 0.19874, "loss_box_dn_1": 0.81385, "loss_cls_dn_2": 0.20583, "loss_box_dn_2": 0.79347, "loss_cls_dn_3": 0.20166, "loss_box_dn_3": 0.78405, "loss_cls_dn_4": 0.20582, "loss_box_dn_4": 0.7935, "loss_cls_dn_5": 0.21387, "loss_box_dn_5": 0.82147, "loss_dense_depth": 0.85273, "loss": 29.60624, "grad_norm": 44.23405, "time": 1.55147} -{"mode": "train", "epoch": 1, "iter": 103, "lr": 0.00014, "memory": 49163, "data_time": 0.07498, "loss_cls_0": 0.9562, "loss_box_0": 1.79571, "loss_cns_0": 0.60644, "loss_yns_0": 0.15613, "loss_cls_1": 1.02246, "loss_box_1": 1.94888, "loss_cns_1": 0.64253, "loss_yns_1": 0.16299, "loss_cls_2": 1.05363, "loss_box_2": 1.89959, "loss_cns_2": 0.64871, "loss_yns_2": 0.1631, "loss_cls_3": 1.05503, "loss_box_3": 1.88963, "loss_cns_3": 0.65365, "loss_yns_3": 0.16261, "loss_cls_4": 1.04672, "loss_box_4": 1.90861, "loss_cns_4": 0.65273, "loss_yns_4": 0.1697, "loss_cls_5": 1.05217, "loss_box_5": 1.90023, "loss_cns_5": 0.65138, "loss_yns_5": 0.16153, "loss_cls_dn_0": 0.26983, "loss_box_dn_0": 0.82055, "loss_cls_dn_1": 0.18502, "loss_box_dn_1": 0.82904, "loss_cls_dn_2": 0.19621, "loss_box_dn_2": 0.80996, "loss_cls_dn_3": 0.20105, "loss_box_dn_3": 0.80921, "loss_cls_dn_4": 0.2071, "loss_box_dn_4": 0.8246, "loss_cls_dn_5": 0.21777, "loss_box_dn_5": 0.84501, "loss_dense_depth": 0.85406, "loss": 29.42979, "grad_norm": 41.62861, "time": 1.59087} -{"mode": "train", "epoch": 1, "iter": 104, "lr": 0.00014, "memory": 49163, "data_time": 0.09688, "loss_cls_0": 0.94076, "loss_box_0": 1.84551, "loss_cns_0": 0.61371, "loss_yns_0": 0.15604, "loss_cls_1": 1.00518, "loss_box_1": 1.95038, "loss_cns_1": 0.63691, "loss_yns_1": 0.15974, "loss_cls_2": 1.04539, "loss_box_2": 1.91345, "loss_cns_2": 0.65081, "loss_yns_2": 0.15944, "loss_cls_3": 1.05168, "loss_box_3": 1.91442, "loss_cns_3": 0.65345, "loss_yns_3": 0.15917, "loss_cls_4": 1.0518, "loss_box_4": 1.91046, "loss_cns_4": 0.65436, "loss_yns_4": 0.16019, "loss_cls_5": 1.06798, "loss_box_5": 1.91354, "loss_cns_5": 0.6559, "loss_yns_5": 0.16055, "loss_cls_dn_0": 0.2593, "loss_box_dn_0": 0.82738, "loss_cls_dn_1": 0.17825, "loss_box_dn_1": 0.83612, "loss_cls_dn_2": 0.18897, "loss_box_dn_2": 0.8298, "loss_cls_dn_3": 0.18978, "loss_box_dn_3": 0.83633, "loss_cls_dn_4": 0.19662, "loss_box_dn_4": 0.84407, "loss_cls_dn_5": 0.20856, "loss_box_dn_5": 0.86939, "loss_dense_depth": 0.85107, "loss": 29.54646, "grad_norm": 43.00325, "time": 1.58323} -{"mode": "train", "epoch": 1, "iter": 105, "lr": 0.00014, "memory": 49163, "data_time": 0.07287, "loss_cls_0": 0.96889, "loss_box_0": 1.79254, "loss_cns_0": 0.62402, "loss_yns_0": 0.15674, "loss_cls_1": 0.98895, "loss_box_1": 1.9137, "loss_cns_1": 0.63987, "loss_yns_1": 0.15991, "loss_cls_2": 1.02214, "loss_box_2": 1.8789, "loss_cns_2": 0.65598, "loss_yns_2": 0.159, "loss_cls_3": 1.02066, "loss_box_3": 1.85394, "loss_cns_3": 0.65508, "loss_yns_3": 0.15897, "loss_cls_4": 1.02075, "loss_box_4": 1.86128, "loss_cns_4": 0.65737, "loss_yns_4": 0.16698, "loss_cls_5": 1.03593, "loss_box_5": 1.87314, "loss_cns_5": 0.6574, "loss_yns_5": 0.1618, "loss_cls_dn_0": 0.24841, "loss_box_dn_0": 0.81635, "loss_cls_dn_1": 0.17976, "loss_box_dn_1": 0.85925, "loss_cls_dn_2": 0.18526, "loss_box_dn_2": 0.85335, "loss_cls_dn_3": 0.18472, "loss_box_dn_3": 0.84924, "loss_cls_dn_4": 0.18754, "loss_box_dn_4": 0.85767, "loss_cls_dn_5": 0.1947, "loss_box_dn_5": 0.88357, "loss_dense_depth": 0.81637, "loss": 29.20012, "grad_norm": 38.30715, "time": 1.55118} -{"mode": "train", "epoch": 1, "iter": 106, "lr": 0.00014, "memory": 49163, "data_time": 0.07735, "loss_cls_0": 0.94153, "loss_box_0": 1.7285, "loss_cns_0": 0.60588, "loss_yns_0": 0.15169, "loss_cls_1": 0.97636, "loss_box_1": 1.92593, "loss_cns_1": 0.63574, "loss_yns_1": 0.15797, "loss_cls_2": 1.01907, "loss_box_2": 1.85901, "loss_cns_2": 0.65259, "loss_yns_2": 0.15949, "loss_cls_3": 1.02734, "loss_box_3": 1.84165, "loss_cns_3": 0.65613, "loss_yns_3": 0.15766, "loss_cls_4": 1.03312, "loss_box_4": 1.87502, "loss_cns_4": 0.65832, "loss_yns_4": 0.17012, "loss_cls_5": 1.02929, "loss_box_5": 1.86532, "loss_cns_5": 0.65755, "loss_yns_5": 0.15937, "loss_cls_dn_0": 0.25849, "loss_box_dn_0": 0.79988, "loss_cls_dn_1": 0.17786, "loss_box_dn_1": 0.82988, "loss_cls_dn_2": 0.18344, "loss_box_dn_2": 0.81371, "loss_cls_dn_3": 0.18441, "loss_box_dn_3": 0.81346, "loss_cls_dn_4": 0.18872, "loss_box_dn_4": 0.83179, "loss_cls_dn_5": 0.19461, "loss_box_dn_5": 0.84613, "loss_dense_depth": 0.82895, "loss": 28.89596, "grad_norm": 39.42088, "time": 1.54077} -{"mode": "train", "epoch": 1, "iter": 107, "lr": 0.00014, "memory": 49163, "data_time": 0.0765, "loss_cls_0": 0.95585, "loss_box_0": 1.69217, "loss_cns_0": 0.59803, "loss_yns_0": 0.15071, "loss_cls_1": 0.96474, "loss_box_1": 1.95818, "loss_cns_1": 0.63103, "loss_yns_1": 0.15542, "loss_cls_2": 1.04195, "loss_box_2": 1.89253, "loss_cns_2": 0.64888, "loss_yns_2": 0.15692, "loss_cls_3": 1.02918, "loss_box_3": 1.87167, "loss_cns_3": 0.65597, "loss_yns_3": 0.1553, "loss_cls_4": 1.0345, "loss_box_4": 1.88934, "loss_cns_4": 0.65666, "loss_yns_4": 0.16124, "loss_cls_5": 1.0376, "loss_box_5": 1.88846, "loss_cns_5": 0.65548, "loss_yns_5": 0.15704, "loss_cls_dn_0": 0.26397, "loss_box_dn_0": 0.8185, "loss_cls_dn_1": 0.17855, "loss_box_dn_1": 0.80092, "loss_cls_dn_2": 0.18808, "loss_box_dn_2": 0.77876, "loss_cls_dn_3": 0.18584, "loss_box_dn_3": 0.7805, "loss_cls_dn_4": 0.1916, "loss_box_dn_4": 0.78807, "loss_cls_dn_5": 0.20435, "loss_box_dn_5": 0.79864, "loss_dense_depth": 0.85289, "loss": 28.86953, "grad_norm": 39.69088, "time": 1.51629} -{"mode": "train", "epoch": 1, "iter": 108, "lr": 0.00014, "memory": 49163, "data_time": 0.07485, "loss_cls_0": 0.90554, "loss_box_0": 1.7688, "loss_cns_0": 0.61769, "loss_yns_0": 0.15707, "loss_cls_1": 0.9854, "loss_box_1": 1.93634, "loss_cns_1": 0.632, "loss_yns_1": 0.15674, "loss_cls_2": 1.02118, "loss_box_2": 1.8658, "loss_cns_2": 0.64913, "loss_yns_2": 0.15773, "loss_cls_3": 1.02147, "loss_box_3": 1.84169, "loss_cns_3": 0.65407, "loss_yns_3": 0.15854, "loss_cls_4": 1.03624, "loss_box_4": 1.8306, "loss_cns_4": 0.6545, "loss_yns_4": 0.15922, "loss_cls_5": 1.03842, "loss_box_5": 1.84816, "loss_cns_5": 0.65415, "loss_yns_5": 0.16145, "loss_cls_dn_0": 0.25065, "loss_box_dn_0": 0.81754, "loss_cls_dn_1": 0.18176, "loss_box_dn_1": 0.77202, "loss_cls_dn_2": 0.18863, "loss_box_dn_2": 0.75291, "loss_cls_dn_3": 0.18406, "loss_box_dn_3": 0.75257, "loss_cls_dn_4": 0.19014, "loss_box_dn_4": 0.75459, "loss_cls_dn_5": 0.20279, "loss_box_dn_5": 0.77504, "loss_dense_depth": 0.81595, "loss": 28.55055, "grad_norm": 38.19712, "time": 1.51414} -{"mode": "train", "epoch": 1, "iter": 109, "lr": 0.00014, "memory": 49163, "data_time": 0.07743, "loss_cls_0": 0.95676, "loss_box_0": 1.82556, "loss_cns_0": 0.61564, "loss_yns_0": 0.16034, "loss_cls_1": 0.96814, "loss_box_1": 2.00918, "loss_cns_1": 0.61733, "loss_yns_1": 0.15737, "loss_cls_2": 1.02192, "loss_box_2": 1.93198, "loss_cns_2": 0.64292, "loss_yns_2": 0.16007, "loss_cls_3": 1.01728, "loss_box_3": 1.92765, "loss_cns_3": 0.64444, "loss_yns_3": 0.15981, "loss_cls_4": 1.02998, "loss_box_4": 1.92251, "loss_cns_4": 0.64643, "loss_yns_4": 0.16983, "loss_cls_5": 1.05655, "loss_box_5": 1.94099, "loss_cns_5": 0.65023, "loss_yns_5": 0.16503, "loss_cls_dn_0": 0.25149, "loss_box_dn_0": 0.82295, "loss_cls_dn_1": 0.17841, "loss_box_dn_1": 0.80547, "loss_cls_dn_2": 0.18227, "loss_box_dn_2": 0.78432, "loss_cls_dn_3": 0.18116, "loss_box_dn_3": 0.7867, "loss_cls_dn_4": 0.18762, "loss_box_dn_4": 0.80055, "loss_cls_dn_5": 0.19933, "loss_box_dn_5": 0.82378, "loss_dense_depth": 0.83533, "loss": 29.23729, "grad_norm": 47.17654, "time": 1.53678} -{"mode": "train", "epoch": 1, "iter": 110, "lr": 0.00014, "memory": 49163, "data_time": 0.07615, "loss_cls_0": 0.93851, "loss_box_0": 1.83045, "loss_cns_0": 0.61701, "loss_yns_0": 0.15951, "loss_cls_1": 1.0069, "loss_box_1": 1.98156, "loss_cns_1": 0.61218, "loss_yns_1": 0.15625, "loss_cls_2": 1.02421, "loss_box_2": 1.95082, "loss_cns_2": 0.64399, "loss_yns_2": 0.15951, "loss_cls_3": 1.02132, "loss_box_3": 1.98603, "loss_cns_3": 0.64246, "loss_yns_3": 0.16084, "loss_cls_4": 1.03866, "loss_box_4": 1.96185, "loss_cns_4": 0.645, "loss_yns_4": 0.17431, "loss_cls_5": 1.0596, "loss_box_5": 1.9453, "loss_cns_5": 0.65037, "loss_yns_5": 0.16138, "loss_cls_dn_0": 0.25396, "loss_box_dn_0": 0.81109, "loss_cls_dn_1": 0.19091, "loss_box_dn_1": 0.83209, "loss_cls_dn_2": 0.1853, "loss_box_dn_2": 0.8205, "loss_cls_dn_3": 0.18655, "loss_box_dn_3": 0.84192, "loss_cls_dn_4": 0.19549, "loss_box_dn_4": 0.85939, "loss_cls_dn_5": 0.20519, "loss_box_dn_5": 0.87575, "loss_dense_depth": 0.84103, "loss": 29.62717, "grad_norm": 45.67657, "time": 1.51277} -{"mode": "train", "epoch": 1, "iter": 111, "lr": 0.00014, "memory": 49163, "data_time": 0.07599, "loss_cls_0": 0.9246, "loss_box_0": 1.83549, "loss_cns_0": 0.61793, "loss_yns_0": 0.1604, "loss_cls_1": 0.98055, "loss_box_1": 1.96673, "loss_cns_1": 0.62383, "loss_yns_1": 0.15532, "loss_cls_2": 1.03849, "loss_box_2": 1.92712, "loss_cns_2": 0.64413, "loss_yns_2": 0.15742, "loss_cls_3": 1.03101, "loss_box_3": 1.95824, "loss_cns_3": 0.64902, "loss_yns_3": 0.15674, "loss_cls_4": 1.03087, "loss_box_4": 1.93066, "loss_cns_4": 0.64749, "loss_yns_4": 0.1599, "loss_cls_5": 1.03332, "loss_box_5": 1.92919, "loss_cns_5": 0.64666, "loss_yns_5": 0.15758, "loss_cls_dn_0": 0.25515, "loss_box_dn_0": 0.81192, "loss_cls_dn_1": 0.18505, "loss_box_dn_1": 0.84283, "loss_cls_dn_2": 0.18998, "loss_box_dn_2": 0.83595, "loss_cls_dn_3": 0.19016, "loss_box_dn_3": 0.86268, "loss_cls_dn_4": 0.19598, "loss_box_dn_4": 0.87429, "loss_cls_dn_5": 0.20572, "loss_box_dn_5": 0.89106, "loss_dense_depth": 0.87261, "loss": 29.57608, "grad_norm": 48.96724, "time": 1.51186} -{"mode": "train", "epoch": 1, "iter": 112, "lr": 0.00014, "memory": 49163, "data_time": 0.07284, "loss_cls_0": 0.94848, "loss_box_0": 1.85833, "loss_cns_0": 0.61271, "loss_yns_0": 0.16129, "loss_cls_1": 0.99437, "loss_box_1": 2.0035, "loss_cns_1": 0.61774, "loss_yns_1": 0.15798, "loss_cls_2": 1.0159, "loss_box_2": 1.9501, "loss_cns_2": 0.63808, "loss_yns_2": 0.15798, "loss_cls_3": 1.02365, "loss_box_3": 1.96893, "loss_cns_3": 0.64456, "loss_yns_3": 0.15543, "loss_cls_4": 1.02607, "loss_box_4": 1.93256, "loss_cns_4": 0.64487, "loss_yns_4": 0.15676, "loss_cls_5": 1.04019, "loss_box_5": 1.9418, "loss_cns_5": 0.64225, "loss_yns_5": 0.15843, "loss_cls_dn_0": 0.26579, "loss_box_dn_0": 0.81701, "loss_cls_dn_1": 0.18166, "loss_box_dn_1": 0.87382, "loss_cls_dn_2": 0.19096, "loss_box_dn_2": 0.84907, "loss_cls_dn_3": 0.19358, "loss_box_dn_3": 0.86745, "loss_cls_dn_4": 0.1941, "loss_box_dn_4": 0.86875, "loss_cls_dn_5": 0.20594, "loss_box_dn_5": 0.88419, "loss_dense_depth": 0.88403, "loss": 29.72831, "grad_norm": 50.21082, "time": 1.52203} -{"mode": "train", "epoch": 1, "iter": 113, "lr": 0.00014, "memory": 49163, "data_time": 0.07322, "loss_cls_0": 0.91463, "loss_box_0": 1.83119, "loss_cns_0": 0.61445, "loss_yns_0": 0.15662, "loss_cls_1": 0.94994, "loss_box_1": 2.01149, "loss_cns_1": 0.61796, "loss_yns_1": 0.15744, "loss_cls_2": 0.99574, "loss_box_2": 1.92922, "loss_cns_2": 0.63903, "loss_yns_2": 0.15357, "loss_cls_3": 1.00749, "loss_box_3": 1.91807, "loss_cns_3": 0.64494, "loss_yns_3": 0.15305, "loss_cls_4": 1.00287, "loss_box_4": 1.91794, "loss_cns_4": 0.64636, "loss_yns_4": 0.15501, "loss_cls_5": 1.01897, "loss_box_5": 1.88873, "loss_cns_5": 0.64499, "loss_yns_5": 0.15762, "loss_cls_dn_0": 0.25403, "loss_box_dn_0": 0.81324, "loss_cls_dn_1": 0.17648, "loss_box_dn_1": 0.85715, "loss_cls_dn_2": 0.18755, "loss_box_dn_2": 0.81179, "loss_cls_dn_3": 0.1872, "loss_box_dn_3": 0.81069, "loss_cls_dn_4": 0.18803, "loss_box_dn_4": 0.81489, "loss_cls_dn_5": 0.20219, "loss_box_dn_5": 0.80991, "loss_dense_depth": 0.83708, "loss": 29.07755, "grad_norm": 49.78112, "time": 1.52665} -{"mode": "train", "epoch": 1, "iter": 114, "lr": 0.00015, "memory": 49163, "data_time": 0.07679, "loss_cls_0": 0.91674, "loss_box_0": 1.84153, "loss_cns_0": 0.61866, "loss_yns_0": 0.15819, "loss_cls_1": 0.97591, "loss_box_1": 1.97946, "loss_cns_1": 0.6257, "loss_yns_1": 0.16058, "loss_cls_2": 1.01468, "loss_box_2": 1.90175, "loss_cns_2": 0.64659, "loss_yns_2": 0.15907, "loss_cls_3": 1.00983, "loss_box_3": 1.89079, "loss_cns_3": 0.64578, "loss_yns_3": 0.15734, "loss_cls_4": 1.01108, "loss_box_4": 1.90943, "loss_cns_4": 0.64886, "loss_yns_4": 0.16268, "loss_cls_5": 1.02891, "loss_box_5": 1.8689, "loss_cns_5": 0.65054, "loss_yns_5": 0.16018, "loss_cls_dn_0": 0.2582, "loss_box_dn_0": 0.80368, "loss_cls_dn_1": 0.17359, "loss_box_dn_1": 0.80693, "loss_cls_dn_2": 0.18218, "loss_box_dn_2": 0.76514, "loss_cls_dn_3": 0.18139, "loss_box_dn_3": 0.75871, "loss_cls_dn_4": 0.18672, "loss_box_dn_4": 0.77329, "loss_cls_dn_5": 0.20108, "loss_box_dn_5": 0.76377, "loss_dense_depth": 0.83661, "loss": 28.83448, "grad_norm": 34.94857, "time": 1.53009} -{"mode": "train", "epoch": 1, "iter": 115, "lr": 0.00015, "memory": 49163, "data_time": 0.07505, "loss_cls_0": 0.89504, "loss_box_0": 1.82461, "loss_cns_0": 0.62205, "loss_yns_0": 0.15735, "loss_cls_1": 0.9649, "loss_box_1": 1.97638, "loss_cns_1": 0.63064, "loss_yns_1": 0.15702, "loss_cls_2": 1.00077, "loss_box_2": 1.92214, "loss_cns_2": 0.65065, "loss_yns_2": 0.15619, "loss_cls_3": 1.00087, "loss_box_3": 1.91064, "loss_cns_3": 0.64692, "loss_yns_3": 0.15486, "loss_cls_4": 1.01308, "loss_box_4": 1.9316, "loss_cns_4": 0.64851, "loss_yns_4": 0.16539, "loss_cls_5": 1.04494, "loss_box_5": 1.91422, "loss_cns_5": 0.65102, "loss_yns_5": 0.15756, "loss_cls_dn_0": 0.24322, "loss_box_dn_0": 0.81261, "loss_cls_dn_1": 0.17338, "loss_box_dn_1": 0.77891, "loss_cls_dn_2": 0.17957, "loss_box_dn_2": 0.75784, "loss_cls_dn_3": 0.17733, "loss_box_dn_3": 0.75321, "loss_cls_dn_4": 0.18263, "loss_box_dn_4": 0.76846, "loss_cls_dn_5": 0.19402, "loss_box_dn_5": 0.77353, "loss_dense_depth": 0.82384, "loss": 28.81589, "grad_norm": 60.3746, "time": 1.57874} -{"mode": "train", "epoch": 1, "iter": 116, "lr": 0.00015, "memory": 49163, "data_time": 0.07648, "loss_cls_0": 0.92833, "loss_box_0": 1.84134, "loss_cns_0": 0.61458, "loss_yns_0": 0.15844, "loss_cls_1": 0.96786, "loss_box_1": 1.95552, "loss_cns_1": 0.63303, "loss_yns_1": 0.15691, "loss_cls_2": 0.99791, "loss_box_2": 1.88269, "loss_cns_2": 0.64593, "loss_yns_2": 0.15649, "loss_cls_3": 1.00025, "loss_box_3": 1.86816, "loss_cns_3": 0.64756, "loss_yns_3": 0.15578, "loss_cls_4": 1.00708, "loss_box_4": 1.86399, "loss_cns_4": 0.64887, "loss_yns_4": 0.15916, "loss_cls_5": 1.01222, "loss_box_5": 1.89687, "loss_cns_5": 0.64711, "loss_yns_5": 0.15716, "loss_cls_dn_0": 0.24572, "loss_box_dn_0": 0.8141, "loss_cls_dn_1": 0.17842, "loss_box_dn_1": 0.79661, "loss_cls_dn_2": 0.18393, "loss_box_dn_2": 0.78643, "loss_cls_dn_3": 0.18278, "loss_box_dn_3": 0.78789, "loss_cls_dn_4": 0.19093, "loss_box_dn_4": 0.80351, "loss_cls_dn_5": 0.20358, "loss_box_dn_5": 0.83056, "loss_dense_depth": 0.84895, "loss": 28.85666, "grad_norm": 46.33076, "time": 1.52712} -{"mode": "train", "epoch": 1, "iter": 117, "lr": 0.00015, "memory": 49163, "data_time": 0.07437, "loss_cls_0": 0.8986, "loss_box_0": 1.80579, "loss_cns_0": 0.61429, "loss_yns_0": 0.15319, "loss_cls_1": 0.96989, "loss_box_1": 1.95516, "loss_cns_1": 0.63364, "loss_yns_1": 0.15853, "loss_cls_2": 0.99273, "loss_box_2": 1.86839, "loss_cns_2": 0.64726, "loss_yns_2": 0.15884, "loss_cls_3": 0.99871, "loss_box_3": 1.84317, "loss_cns_3": 0.64802, "loss_yns_3": 0.15905, "loss_cls_4": 1.00631, "loss_box_4": 1.84222, "loss_cns_4": 0.6504, "loss_yns_4": 0.158, "loss_cls_5": 1.01106, "loss_box_5": 1.85104, "loss_cns_5": 0.64728, "loss_yns_5": 0.15661, "loss_cls_dn_0": 0.24091, "loss_box_dn_0": 0.81218, "loss_cls_dn_1": 0.17875, "loss_box_dn_1": 0.8189, "loss_cls_dn_2": 0.18072, "loss_box_dn_2": 0.8069, "loss_cls_dn_3": 0.18336, "loss_box_dn_3": 0.80705, "loss_cls_dn_4": 0.19103, "loss_box_dn_4": 0.82699, "loss_cls_dn_5": 0.20341, "loss_box_dn_5": 0.84986, "loss_dense_depth": 0.83456, "loss": 28.76281, "grad_norm": 36.85384, "time": 1.51275} -{"mode": "train", "epoch": 1, "iter": 118, "lr": 0.00015, "memory": 49163, "data_time": 0.07589, "loss_cls_0": 0.90496, "loss_box_0": 1.76735, "loss_cns_0": 0.6193, "loss_yns_0": 0.15332, "loss_cls_1": 0.97489, "loss_box_1": 1.92067, "loss_cns_1": 0.633, "loss_yns_1": 0.15681, "loss_cls_2": 1.01078, "loss_box_2": 1.85019, "loss_cns_2": 0.65106, "loss_yns_2": 0.15709, "loss_cls_3": 1.01498, "loss_box_3": 1.83692, "loss_cns_3": 0.64864, "loss_yns_3": 0.15847, "loss_cls_4": 1.01958, "loss_box_4": 1.85206, "loss_cns_4": 0.64873, "loss_yns_4": 0.15787, "loss_cls_5": 1.00586, "loss_box_5": 1.85874, "loss_cns_5": 0.65049, "loss_yns_5": 0.15814, "loss_cls_dn_0": 0.24536, "loss_box_dn_0": 0.80117, "loss_cls_dn_1": 0.17542, "loss_box_dn_1": 0.82754, "loss_cls_dn_2": 0.17784, "loss_box_dn_2": 0.81536, "loss_cls_dn_3": 0.18194, "loss_box_dn_3": 0.82772, "loss_cls_dn_4": 0.1882, "loss_box_dn_4": 0.85448, "loss_cls_dn_5": 0.19659, "loss_box_dn_5": 0.87031, "loss_dense_depth": 0.8705, "loss": 28.84232, "grad_norm": 49.35835, "time": 1.53989} -{"mode": "train", "epoch": 1, "iter": 119, "lr": 0.00015, "memory": 49163, "data_time": 0.08022, "loss_cls_0": 0.90669, "loss_box_0": 1.75621, "loss_cns_0": 0.61076, "loss_yns_0": 0.15272, "loss_cls_1": 0.96925, "loss_box_1": 1.92287, "loss_cns_1": 0.63505, "loss_yns_1": 0.1561, "loss_cls_2": 1.03711, "loss_box_2": 1.87706, "loss_cns_2": 0.646, "loss_yns_2": 0.15569, "loss_cls_3": 1.00848, "loss_box_3": 1.87075, "loss_cns_3": 0.64618, "loss_yns_3": 0.15769, "loss_cls_4": 1.02396, "loss_box_4": 1.87011, "loss_cns_4": 0.64928, "loss_yns_4": 0.1616, "loss_cls_5": 1.00608, "loss_box_5": 1.88613, "loss_cns_5": 0.65243, "loss_yns_5": 0.15998, "loss_cls_dn_0": 0.24544, "loss_box_dn_0": 0.80024, "loss_cls_dn_1": 0.17267, "loss_box_dn_1": 0.83362, "loss_cls_dn_2": 0.18013, "loss_box_dn_2": 0.82138, "loss_cls_dn_3": 0.18085, "loss_box_dn_3": 0.83115, "loss_cls_dn_4": 0.18496, "loss_box_dn_4": 0.8445, "loss_cls_dn_5": 0.18885, "loss_box_dn_5": 0.8522, "loss_dense_depth": 0.83636, "loss": 28.89053, "grad_norm": 53.86833, "time": 1.52532} -{"mode": "train", "epoch": 1, "iter": 120, "lr": 0.00015, "memory": 49163, "data_time": 0.08046, "loss_cls_0": 0.90405, "loss_box_0": 1.77148, "loss_cns_0": 0.60723, "loss_yns_0": 0.153, "loss_cls_1": 0.96313, "loss_box_1": 1.91639, "loss_cns_1": 0.63885, "loss_yns_1": 0.15694, "loss_cls_2": 0.98869, "loss_box_2": 1.86242, "loss_cns_2": 0.6459, "loss_yns_2": 0.15593, "loss_cls_3": 1.01793, "loss_box_3": 1.83765, "loss_cns_3": 0.64924, "loss_yns_3": 0.15647, "loss_cls_4": 1.02991, "loss_box_4": 1.82362, "loss_cns_4": 0.65268, "loss_yns_4": 0.16539, "loss_cls_5": 1.01225, "loss_box_5": 1.83796, "loss_cns_5": 0.65282, "loss_yns_5": 0.15985, "loss_cls_dn_0": 0.24797, "loss_box_dn_0": 0.81296, "loss_cls_dn_1": 0.17216, "loss_box_dn_1": 0.7973, "loss_cls_dn_2": 0.17551, "loss_box_dn_2": 0.77395, "loss_cls_dn_3": 0.17934, "loss_box_dn_3": 0.76615, "loss_cls_dn_4": 0.18133, "loss_box_dn_4": 0.76893, "loss_cls_dn_5": 0.18948, "loss_box_dn_5": 0.77169, "loss_dense_depth": 0.85226, "loss": 28.44882, "grad_norm": 40.71808, "time": 1.51174} -{"mode": "train", "epoch": 1, "iter": 121, "lr": 0.00015, "memory": 49163, "data_time": 0.15554, "loss_cls_0": 0.91848, "loss_box_0": 1.78109, "loss_cns_0": 0.618, "loss_yns_0": 0.15515, "loss_cls_1": 0.96863, "loss_box_1": 1.99592, "loss_cns_1": 0.63412, "loss_yns_1": 0.15844, "loss_cls_2": 1.0246, "loss_box_2": 1.91197, "loss_cns_2": 0.64093, "loss_yns_2": 0.15812, "loss_cls_3": 1.05421, "loss_box_3": 1.90627, "loss_cns_3": 0.64659, "loss_yns_3": 0.15661, "loss_cls_4": 1.04737, "loss_box_4": 1.91473, "loss_cns_4": 0.64857, "loss_yns_4": 0.16002, "loss_cls_5": 1.02985, "loss_box_5": 1.90851, "loss_cns_5": 0.64673, "loss_yns_5": 0.15685, "loss_cls_dn_0": 0.2412, "loss_box_dn_0": 0.80991, "loss_cls_dn_1": 0.16764, "loss_box_dn_1": 0.80238, "loss_cls_dn_2": 0.17681, "loss_box_dn_2": 0.77812, "loss_cls_dn_3": 0.18351, "loss_box_dn_3": 0.7768, "loss_cls_dn_4": 0.18379, "loss_box_dn_4": 0.79068, "loss_cls_dn_5": 0.18775, "loss_box_dn_5": 0.7958, "loss_dense_depth": 0.85246, "loss": 28.9886, "grad_norm": 48.83958, "time": 1.60178} -{"mode": "train", "epoch": 1, "iter": 122, "lr": 0.00015, "memory": 49163, "data_time": 0.07782, "loss_cls_0": 0.95477, "loss_box_0": 1.77353, "loss_cns_0": 0.62345, "loss_yns_0": 0.15621, "loss_cls_1": 0.96052, "loss_box_1": 1.937, "loss_cns_1": 0.63656, "loss_yns_1": 0.16093, "loss_cls_2": 1.01202, "loss_box_2": 1.87543, "loss_cns_2": 0.6458, "loss_yns_2": 0.16449, "loss_cls_3": 1.01911, "loss_box_3": 1.8908, "loss_cns_3": 0.6497, "loss_yns_3": 0.16296, "loss_cls_4": 1.01646, "loss_box_4": 1.89694, "loss_cns_4": 0.65284, "loss_yns_4": 0.16227, "loss_cls_5": 1.01767, "loss_box_5": 1.89193, "loss_cns_5": 0.65039, "loss_yns_5": 0.16207, "loss_cls_dn_0": 0.23426, "loss_box_dn_0": 0.80049, "loss_cls_dn_1": 0.16405, "loss_box_dn_1": 0.78923, "loss_cls_dn_2": 0.17335, "loss_box_dn_2": 0.77258, "loss_cls_dn_3": 0.17666, "loss_box_dn_3": 0.78798, "loss_cls_dn_4": 0.18182, "loss_box_dn_4": 0.80679, "loss_cls_dn_5": 0.18557, "loss_box_dn_5": 0.82007, "loss_dense_depth": 0.83045, "loss": 28.79712, "grad_norm": 52.45389, "time": 1.53915} -{"mode": "train", "epoch": 1, "iter": 123, "lr": 0.00015, "memory": 49163, "data_time": 0.07563, "loss_cls_0": 0.89908, "loss_box_0": 1.77621, "loss_cns_0": 0.62227, "loss_yns_0": 0.15619, "loss_cls_1": 0.98412, "loss_box_1": 1.93921, "loss_cns_1": 0.63649, "loss_yns_1": 0.1598, "loss_cls_2": 1.02037, "loss_box_2": 1.88444, "loss_cns_2": 0.64598, "loss_yns_2": 0.16283, "loss_cls_3": 1.00542, "loss_box_3": 1.88496, "loss_cns_3": 0.64591, "loss_yns_3": 0.16175, "loss_cls_4": 1.0315, "loss_box_4": 1.87758, "loss_cns_4": 0.6505, "loss_yns_4": 0.16882, "loss_cls_5": 1.025, "loss_box_5": 1.87649, "loss_cns_5": 0.64746, "loss_yns_5": 0.1667, "loss_cls_dn_0": 0.22785, "loss_box_dn_0": 0.79389, "loss_cls_dn_1": 0.16291, "loss_box_dn_1": 0.81311, "loss_cls_dn_2": 0.17156, "loss_box_dn_2": 0.80477, "loss_cls_dn_3": 0.17308, "loss_box_dn_3": 0.82633, "loss_cls_dn_4": 0.17929, "loss_box_dn_4": 0.84655, "loss_cls_dn_5": 0.1857, "loss_box_dn_5": 0.86404, "loss_dense_depth": 0.85848, "loss": 28.93662, "grad_norm": 61.00671, "time": 1.58645} -{"mode": "train", "epoch": 1, "iter": 124, "lr": 0.00015, "memory": 49163, "data_time": 0.10269, "loss_cls_0": 0.90982, "loss_box_0": 1.77911, "loss_cns_0": 0.6209, "loss_yns_0": 0.15542, "loss_cls_1": 0.98156, "loss_box_1": 1.90665, "loss_cns_1": 0.64725, "loss_yns_1": 0.15829, "loss_cls_2": 1.02511, "loss_box_2": 1.86002, "loss_cns_2": 0.6517, "loss_yns_2": 0.16052, "loss_cls_3": 1.02833, "loss_box_3": 1.85615, "loss_cns_3": 0.65093, "loss_yns_3": 0.15901, "loss_cls_4": 1.03469, "loss_box_4": 1.83908, "loss_cns_4": 0.65332, "loss_yns_4": 0.16552, "loss_cls_5": 1.02457, "loss_box_5": 1.84411, "loss_cns_5": 0.65203, "loss_yns_5": 0.16173, "loss_cls_dn_0": 0.23465, "loss_box_dn_0": 0.79883, "loss_cls_dn_1": 0.16195, "loss_box_dn_1": 0.80411, "loss_cls_dn_2": 0.16749, "loss_box_dn_2": 0.78963, "loss_cls_dn_3": 0.17064, "loss_box_dn_3": 0.80149, "loss_cls_dn_4": 0.17404, "loss_box_dn_4": 0.81413, "loss_cls_dn_5": 0.18508, "loss_box_dn_5": 0.83307, "loss_dense_depth": 0.81169, "loss": 28.67262, "grad_norm": 46.20651, "time": 1.57141} -{"mode": "train", "epoch": 1, "iter": 125, "lr": 0.00015, "memory": 49163, "data_time": 0.07365, "loss_cls_0": 0.9636, "loss_box_0": 1.77997, "loss_cns_0": 0.61473, "loss_yns_0": 0.15584, "loss_cls_1": 1.00999, "loss_box_1": 1.93425, "loss_cns_1": 0.6474, "loss_yns_1": 0.16153, "loss_cls_2": 1.03402, "loss_box_2": 1.88086, "loss_cns_2": 0.65061, "loss_yns_2": 0.16199, "loss_cls_3": 1.03816, "loss_box_3": 1.88486, "loss_cns_3": 0.64929, "loss_yns_3": 0.16135, "loss_cls_4": 1.04197, "loss_box_4": 1.87348, "loss_cns_4": 0.65052, "loss_yns_4": 0.15947, "loss_cls_5": 1.04142, "loss_box_5": 1.89265, "loss_cns_5": 0.64841, "loss_yns_5": 0.15732, "loss_cls_dn_0": 0.24424, "loss_box_dn_0": 0.79239, "loss_cls_dn_1": 0.16449, "loss_box_dn_1": 0.80715, "loss_cls_dn_2": 0.17071, "loss_box_dn_2": 0.79171, "loss_cls_dn_3": 0.17327, "loss_box_dn_3": 0.80154, "loss_cls_dn_4": 0.17771, "loss_box_dn_4": 0.81671, "loss_cls_dn_5": 0.19153, "loss_box_dn_5": 0.83829, "loss_dense_depth": 0.84399, "loss": 29.00743, "grad_norm": 43.00239, "time": 1.52243} -{"mode": "train", "epoch": 1, "iter": 126, "lr": 0.00015, "memory": 49163, "data_time": 0.07527, "loss_cls_0": 0.92952, "loss_box_0": 1.7865, "loss_cns_0": 0.61982, "loss_yns_0": 0.1552, "loss_cls_1": 0.98639, "loss_box_1": 1.99613, "loss_cns_1": 0.63538, "loss_yns_1": 0.16461, "loss_cls_2": 1.02105, "loss_box_2": 1.93017, "loss_cns_2": 0.64992, "loss_yns_2": 0.16395, "loss_cls_3": 1.02761, "loss_box_3": 1.92026, "loss_cns_3": 0.65055, "loss_yns_3": 0.15948, "loss_cls_4": 1.03919, "loss_box_4": 1.91796, "loss_cns_4": 0.65118, "loss_yns_4": 0.15834, "loss_cls_5": 1.03568, "loss_box_5": 1.93634, "loss_cns_5": 0.64842, "loss_yns_5": 0.15617, "loss_cls_dn_0": 0.23949, "loss_box_dn_0": 0.7966, "loss_cls_dn_1": 0.17016, "loss_box_dn_1": 0.80787, "loss_cls_dn_2": 0.17832, "loss_box_dn_2": 0.79064, "loss_cls_dn_3": 0.17959, "loss_box_dn_3": 0.7905, "loss_cls_dn_4": 0.18372, "loss_box_dn_4": 0.80588, "loss_cls_dn_5": 0.19251, "loss_box_dn_5": 0.82323, "loss_dense_depth": 0.78613, "loss": 29.08448, "grad_norm": 45.72777, "time": 1.52413} -{"mode": "train", "epoch": 1, "iter": 127, "lr": 0.00015, "memory": 49163, "data_time": 0.07603, "loss_cls_0": 0.91981, "loss_box_0": 1.77293, "loss_cns_0": 0.62035, "loss_yns_0": 0.15818, "loss_cls_1": 0.9758, "loss_box_1": 1.94861, "loss_cns_1": 0.62775, "loss_yns_1": 0.1612, "loss_cls_2": 1.03882, "loss_box_2": 1.86651, "loss_cns_2": 0.64536, "loss_yns_2": 0.16001, "loss_cls_3": 1.0224, "loss_box_3": 1.85452, "loss_cns_3": 0.64675, "loss_yns_3": 0.15782, "loss_cls_4": 1.03002, "loss_box_4": 1.85288, "loss_cns_4": 0.64739, "loss_yns_4": 0.15872, "loss_cls_5": 1.02208, "loss_box_5": 1.86441, "loss_cns_5": 0.64236, "loss_yns_5": 0.15721, "loss_cls_dn_0": 0.23391, "loss_box_dn_0": 0.79602, "loss_cls_dn_1": 0.16697, "loss_box_dn_1": 0.80575, "loss_cls_dn_2": 0.17706, "loss_box_dn_2": 0.7839, "loss_cls_dn_3": 0.17578, "loss_box_dn_3": 0.78053, "loss_cls_dn_4": 0.17733, "loss_box_dn_4": 0.79282, "loss_cls_dn_5": 0.18791, "loss_box_dn_5": 0.80417, "loss_dense_depth": 0.81159, "loss": 28.64567, "grad_norm": 53.14521, "time": 1.51684} -{"mode": "train", "epoch": 1, "iter": 128, "lr": 0.00015, "memory": 49163, "data_time": 0.07454, "loss_cls_0": 0.94605, "loss_box_0": 1.77702, "loss_cns_0": 0.62494, "loss_yns_0": 0.16306, "loss_cls_1": 0.96924, "loss_box_1": 1.89765, "loss_cns_1": 0.63036, "loss_yns_1": 0.16771, "loss_cls_2": 1.01119, "loss_box_2": 1.83637, "loss_cns_2": 0.6463, "loss_yns_2": 0.16564, "loss_cls_3": 1.01808, "loss_box_3": 1.81768, "loss_cns_3": 0.64779, "loss_yns_3": 0.16477, "loss_cls_4": 1.0253, "loss_box_4": 1.81472, "loss_cns_4": 0.65299, "loss_yns_4": 0.16077, "loss_cls_5": 1.0221, "loss_box_5": 1.81698, "loss_cns_5": 0.64668, "loss_yns_5": 0.15878, "loss_cls_dn_0": 0.23433, "loss_box_dn_0": 0.79332, "loss_cls_dn_1": 0.16116, "loss_box_dn_1": 0.77465, "loss_cls_dn_2": 0.16854, "loss_box_dn_2": 0.767, "loss_cls_dn_3": 0.16911, "loss_box_dn_3": 0.76991, "loss_cls_dn_4": 0.17363, "loss_box_dn_4": 0.78371, "loss_cls_dn_5": 0.18719, "loss_box_dn_5": 0.79523, "loss_dense_depth": 0.80455, "loss": 28.3645, "grad_norm": 45.38033, "time": 1.51354} -{"mode": "train", "epoch": 1, "iter": 129, "lr": 0.00015, "memory": 49163, "data_time": 0.0731, "loss_cls_0": 0.90764, "loss_box_0": 1.78589, "loss_cns_0": 0.62162, "loss_yns_0": 0.16052, "loss_cls_1": 0.96484, "loss_box_1": 1.89096, "loss_cns_1": 0.63619, "loss_yns_1": 0.16454, "loss_cls_2": 1.00429, "loss_box_2": 1.83423, "loss_cns_2": 0.64881, "loss_yns_2": 0.15988, "loss_cls_3": 1.01946, "loss_box_3": 1.8087, "loss_cns_3": 0.64786, "loss_yns_3": 0.15863, "loss_cls_4": 1.01742, "loss_box_4": 1.81134, "loss_cns_4": 0.65317, "loss_yns_4": 0.16076, "loss_cls_5": 1.01388, "loss_box_5": 1.84479, "loss_cns_5": 0.64719, "loss_yns_5": 0.15962, "loss_cls_dn_0": 0.23126, "loss_box_dn_0": 0.79875, "loss_cls_dn_1": 0.16297, "loss_box_dn_1": 0.77801, "loss_cls_dn_2": 0.16524, "loss_box_dn_2": 0.76892, "loss_cls_dn_3": 0.16838, "loss_box_dn_3": 0.76846, "loss_cls_dn_4": 0.17154, "loss_box_dn_4": 0.77786, "loss_cls_dn_5": 0.18484, "loss_box_dn_5": 0.79858, "loss_dense_depth": 0.78152, "loss": 28.27857, "grad_norm": 43.531, "time": 1.51602} -{"mode": "train", "epoch": 1, "iter": 130, "lr": 0.00015, "memory": 49163, "data_time": 0.07484, "loss_cls_0": 0.91056, "loss_box_0": 1.78511, "loss_cns_0": 0.61751, "loss_yns_0": 0.1539, "loss_cls_1": 0.96937, "loss_box_1": 1.87186, "loss_cns_1": 0.63663, "loss_yns_1": 0.15686, "loss_cls_2": 1.01065, "loss_box_2": 1.80159, "loss_cns_2": 0.64995, "loss_yns_2": 0.15553, "loss_cls_3": 1.02766, "loss_box_3": 1.77388, "loss_cns_3": 0.6479, "loss_yns_3": 0.15537, "loss_cls_4": 1.02582, "loss_box_4": 1.79391, "loss_cns_4": 0.64975, "loss_yns_4": 0.15977, "loss_cls_5": 1.01781, "loss_box_5": 1.82308, "loss_cns_5": 0.64892, "loss_yns_5": 0.15805, "loss_cls_dn_0": 0.23459, "loss_box_dn_0": 0.79686, "loss_cls_dn_1": 0.16192, "loss_box_dn_1": 0.78823, "loss_cls_dn_2": 0.16571, "loss_box_dn_2": 0.77508, "loss_cls_dn_3": 0.17327, "loss_box_dn_3": 0.77511, "loss_cls_dn_4": 0.17239, "loss_box_dn_4": 0.7889, "loss_cls_dn_5": 0.18356, "loss_box_dn_5": 0.80362, "loss_dense_depth": 0.82952, "loss": 28.25021, "grad_norm": 40.34909, "time": 1.51745} -{"mode": "train", "epoch": 1, "iter": 131, "lr": 0.00015, "memory": 49163, "data_time": 0.0749, "loss_cls_0": 0.91442, "loss_box_0": 1.75369, "loss_cns_0": 0.61573, "loss_yns_0": 0.1541, "loss_cls_1": 0.9594, "loss_box_1": 1.9142, "loss_cns_1": 0.63633, "loss_yns_1": 0.15629, "loss_cls_2": 1.00053, "loss_box_2": 1.85989, "loss_cns_2": 0.64744, "loss_yns_2": 0.15629, "loss_cls_3": 1.00486, "loss_box_3": 1.86461, "loss_cns_3": 0.64809, "loss_yns_3": 0.15647, "loss_cls_4": 1.01797, "loss_box_4": 1.87086, "loss_cns_4": 0.65271, "loss_yns_4": 0.15745, "loss_cls_5": 1.01424, "loss_box_5": 1.85052, "loss_cns_5": 0.64994, "loss_yns_5": 0.15754, "loss_cls_dn_0": 0.23559, "loss_box_dn_0": 0.79768, "loss_cls_dn_1": 0.16185, "loss_box_dn_1": 0.78222, "loss_cls_dn_2": 0.16497, "loss_box_dn_2": 0.77894, "loss_cls_dn_3": 0.16968, "loss_box_dn_3": 0.78934, "loss_cls_dn_4": 0.17307, "loss_box_dn_4": 0.80556, "loss_cls_dn_5": 0.18236, "loss_box_dn_5": 0.80813, "loss_dense_depth": 0.83327, "loss": 28.49624, "grad_norm": 44.64025, "time": 1.53794} -{"mode": "train", "epoch": 1, "iter": 132, "lr": 0.00015, "memory": 49163, "data_time": 0.07726, "loss_cls_0": 0.90207, "loss_box_0": 1.76283, "loss_cns_0": 0.61552, "loss_yns_0": 0.15382, "loss_cls_1": 0.95246, "loss_box_1": 1.91241, "loss_cns_1": 0.63311, "loss_yns_1": 0.15739, "loss_cls_2": 0.9857, "loss_box_2": 1.85734, "loss_cns_2": 0.64471, "loss_yns_2": 0.15387, "loss_cls_3": 1.01819, "loss_box_3": 1.85821, "loss_cns_3": 0.64556, "loss_yns_3": 0.15493, "loss_cls_4": 1.01844, "loss_box_4": 1.84714, "loss_cns_4": 0.64869, "loss_yns_4": 0.15374, "loss_cls_5": 1.00792, "loss_box_5": 1.85097, "loss_cns_5": 0.64531, "loss_yns_5": 0.15453, "loss_cls_dn_0": 0.23881, "loss_box_dn_0": 0.78638, "loss_cls_dn_1": 0.16464, "loss_box_dn_1": 0.77551, "loss_cls_dn_2": 0.16707, "loss_box_dn_2": 0.77628, "loss_cls_dn_3": 0.16687, "loss_box_dn_3": 0.79348, "loss_cls_dn_4": 0.17541, "loss_box_dn_4": 0.80822, "loss_cls_dn_5": 0.18582, "loss_box_dn_5": 0.82879, "loss_dense_depth": 0.82088, "loss": 28.42299, "grad_norm": 46.37974, "time": 1.52327} -{"mode": "train", "epoch": 1, "iter": 133, "lr": 0.00015, "memory": 49163, "data_time": 0.0759, "loss_cls_0": 0.91991, "loss_box_0": 1.73495, "loss_cns_0": 0.60892, "loss_yns_0": 0.15135, "loss_cls_1": 0.95509, "loss_box_1": 1.94381, "loss_cns_1": 0.6269, "loss_yns_1": 0.15891, "loss_cls_2": 0.98721, "loss_box_2": 1.8862, "loss_cns_2": 0.63895, "loss_yns_2": 0.15462, "loss_cls_3": 1.02037, "loss_box_3": 1.88095, "loss_cns_3": 0.64009, "loss_yns_3": 0.15428, "loss_cls_4": 1.0108, "loss_box_4": 1.88702, "loss_cns_4": 0.64099, "loss_yns_4": 0.15269, "loss_cls_5": 1.0057, "loss_box_5": 1.92057, "loss_cns_5": 0.63451, "loss_yns_5": 0.15626, "loss_cls_dn_0": 0.24545, "loss_box_dn_0": 0.79238, "loss_cls_dn_1": 0.16615, "loss_box_dn_1": 0.80102, "loss_cls_dn_2": 0.17044, "loss_box_dn_2": 0.79375, "loss_cls_dn_3": 0.16764, "loss_box_dn_3": 0.80807, "loss_cls_dn_4": 0.17377, "loss_box_dn_4": 0.83374, "loss_cls_dn_5": 0.19299, "loss_box_dn_5": 0.86746, "loss_dense_depth": 0.82885, "loss": 28.71274, "grad_norm": 54.36243, "time": 1.5373} -{"mode": "train", "epoch": 1, "iter": 134, "lr": 0.00015, "memory": 49163, "data_time": 0.07793, "loss_cls_0": 0.88068, "loss_box_0": 1.70924, "loss_cns_0": 0.6199, "loss_yns_0": 0.14981, "loss_cls_1": 0.94569, "loss_box_1": 1.86626, "loss_cns_1": 0.64053, "loss_yns_1": 0.15708, "loss_cls_2": 0.96556, "loss_box_2": 1.82597, "loss_cns_2": 0.64679, "loss_yns_2": 0.15293, "loss_cls_3": 0.98439, "loss_box_3": 1.81288, "loss_cns_3": 0.65384, "loss_yns_3": 0.15171, "loss_cls_4": 0.99588, "loss_box_4": 1.82862, "loss_cns_4": 0.64933, "loss_yns_4": 0.15708, "loss_cls_5": 1.0035, "loss_box_5": 1.84043, "loss_cns_5": 0.64912, "loss_yns_5": 0.15305, "loss_cls_dn_0": 0.23234, "loss_box_dn_0": 0.78966, "loss_cls_dn_1": 0.16723, "loss_box_dn_1": 0.84027, "loss_cls_dn_2": 0.17096, "loss_box_dn_2": 0.82962, "loss_cls_dn_3": 0.16983, "loss_box_dn_3": 0.83879, "loss_cls_dn_4": 0.1723, "loss_box_dn_4": 0.86682, "loss_cls_dn_5": 0.18673, "loss_box_dn_5": 0.88539, "loss_dense_depth": 0.79837, "loss": 28.38858, "grad_norm": 50.30749, "time": 1.50955} -{"mode": "train", "epoch": 1, "iter": 135, "lr": 0.00015, "memory": 49163, "data_time": 0.07743, "loss_cls_0": 0.90163, "loss_box_0": 1.76772, "loss_cns_0": 0.61549, "loss_yns_0": 0.15689, "loss_cls_1": 0.95898, "loss_box_1": 1.89479, "loss_cns_1": 0.64567, "loss_yns_1": 0.15476, "loss_cls_2": 0.98876, "loss_box_2": 1.86278, "loss_cns_2": 0.64849, "loss_yns_2": 0.15456, "loss_cls_3": 0.9826, "loss_box_3": 1.84581, "loss_cns_3": 0.65127, "loss_yns_3": 0.1557, "loss_cls_4": 1.00283, "loss_box_4": 1.86813, "loss_cns_4": 0.64771, "loss_yns_4": 0.17263, "loss_cls_5": 1.0064, "loss_box_5": 1.88112, "loss_cns_5": 0.64425, "loss_yns_5": 0.16076, "loss_cls_dn_0": 0.2398, "loss_box_dn_0": 0.79623, "loss_cls_dn_1": 0.16532, "loss_box_dn_1": 0.83969, "loss_cls_dn_2": 0.17239, "loss_box_dn_2": 0.82347, "loss_cls_dn_3": 0.17186, "loss_box_dn_3": 0.82016, "loss_cls_dn_4": 0.17361, "loss_box_dn_4": 0.83652, "loss_cls_dn_5": 0.18049, "loss_box_dn_5": 0.84514, "loss_dense_depth": 0.80908, "loss": 28.64352, "grad_norm": 49.53232, "time": 1.57858} -{"mode": "train", "epoch": 1, "iter": 136, "lr": 0.00015, "memory": 49163, "data_time": 0.07666, "loss_cls_0": 0.86467, "loss_box_0": 1.75338, "loss_cns_0": 0.62195, "loss_yns_0": 0.15264, "loss_cls_1": 0.9349, "loss_box_1": 1.86784, "loss_cns_1": 0.64097, "loss_yns_1": 0.15123, "loss_cls_2": 0.96688, "loss_box_2": 1.81527, "loss_cns_2": 0.65102, "loss_yns_2": 0.15252, "loss_cls_3": 0.97842, "loss_box_3": 1.7969, "loss_cns_3": 0.65159, "loss_yns_3": 0.15347, "loss_cls_4": 0.98727, "loss_box_4": 1.80561, "loss_cns_4": 0.65165, "loss_yns_4": 0.1611, "loss_cls_5": 0.97366, "loss_box_5": 1.82475, "loss_cns_5": 0.64762, "loss_yns_5": 0.1563, "loss_cls_dn_0": 0.23303, "loss_box_dn_0": 0.79017, "loss_cls_dn_1": 0.15823, "loss_box_dn_1": 0.81335, "loss_cls_dn_2": 0.16273, "loss_box_dn_2": 0.78756, "loss_cls_dn_3": 0.16437, "loss_box_dn_3": 0.78093, "loss_cls_dn_4": 0.17068, "loss_box_dn_4": 0.78533, "loss_cls_dn_5": 0.17511, "loss_box_dn_5": 0.79561, "loss_dense_depth": 0.78854, "loss": 27.96725, "grad_norm": 37.77893, "time": 1.52412} -{"mode": "train", "epoch": 1, "iter": 137, "lr": 0.00015, "memory": 49163, "data_time": 0.0756, "loss_cls_0": 0.86575, "loss_box_0": 1.7338, "loss_cns_0": 0.62175, "loss_yns_0": 0.14942, "loss_cls_1": 0.93784, "loss_box_1": 1.89673, "loss_cns_1": 0.63738, "loss_yns_1": 0.15332, "loss_cls_2": 0.96462, "loss_box_2": 1.81643, "loss_cns_2": 0.64874, "loss_yns_2": 0.15171, "loss_cls_3": 0.98629, "loss_box_3": 1.81321, "loss_cns_3": 0.64899, "loss_yns_3": 0.15171, "loss_cls_4": 0.99927, "loss_box_4": 1.82737, "loss_cns_4": 0.65319, "loss_yns_4": 0.15269, "loss_cls_5": 0.99176, "loss_box_5": 1.83079, "loss_cns_5": 0.65029, "loss_yns_5": 0.15186, "loss_cls_dn_0": 0.23833, "loss_box_dn_0": 0.7883, "loss_cls_dn_1": 0.15478, "loss_box_dn_1": 0.7906, "loss_cls_dn_2": 0.15573, "loss_box_dn_2": 0.75889, "loss_cls_dn_3": 0.15805, "loss_box_dn_3": 0.76073, "loss_cls_dn_4": 0.16118, "loss_box_dn_4": 0.7759, "loss_cls_dn_5": 0.17494, "loss_box_dn_5": 0.7839, "loss_dense_depth": 0.79158, "loss": 27.92782, "grad_norm": 45.1021, "time": 1.5177} -{"mode": "train", "epoch": 1, "iter": 138, "lr": 0.00015, "memory": 49163, "data_time": 0.07758, "loss_cls_0": 0.86458, "loss_box_0": 1.72508, "loss_cns_0": 0.62477, "loss_yns_0": 0.14867, "loss_cls_1": 0.93618, "loss_box_1": 1.88872, "loss_cns_1": 0.63747, "loss_yns_1": 0.15248, "loss_cls_2": 0.95614, "loss_box_2": 1.83366, "loss_cns_2": 0.6454, "loss_yns_2": 0.14983, "loss_cls_3": 0.95821, "loss_box_3": 1.83021, "loss_cns_3": 0.64463, "loss_yns_3": 0.15008, "loss_cls_4": 0.97594, "loss_box_4": 1.84397, "loss_cns_4": 0.64471, "loss_yns_4": 0.15463, "loss_cls_5": 0.98178, "loss_box_5": 1.85501, "loss_cns_5": 0.64401, "loss_yns_5": 0.15063, "loss_cls_dn_0": 0.23413, "loss_box_dn_0": 0.79273, "loss_cls_dn_1": 0.15302, "loss_box_dn_1": 0.78554, "loss_cls_dn_2": 0.15587, "loss_box_dn_2": 0.77332, "loss_cls_dn_3": 0.15913, "loss_box_dn_3": 0.78615, "loss_cls_dn_4": 0.16002, "loss_box_dn_4": 0.81696, "loss_cls_dn_5": 0.17584, "loss_box_dn_5": 0.83624, "loss_dense_depth": 0.82914, "loss": 28.05486, "grad_norm": 42.82412, "time": 1.53739} -{"mode": "train", "epoch": 1, "iter": 139, "lr": 0.00016, "memory": 49163, "data_time": 0.08061, "loss_cls_0": 0.83836, "loss_box_0": 1.72805, "loss_cns_0": 0.62745, "loss_yns_0": 0.14974, "loss_cls_1": 0.92151, "loss_box_1": 1.85865, "loss_cns_1": 0.64159, "loss_yns_1": 0.1511, "loss_cls_2": 0.9496, "loss_box_2": 1.83807, "loss_cns_2": 0.64695, "loss_yns_2": 0.14987, "loss_cls_3": 0.94637, "loss_box_3": 1.82436, "loss_cns_3": 0.64766, "loss_yns_3": 0.14915, "loss_cls_4": 0.96027, "loss_box_4": 1.82842, "loss_cns_4": 0.64401, "loss_yns_4": 0.15924, "loss_cls_5": 0.94931, "loss_box_5": 1.87224, "loss_cns_5": 0.64119, "loss_yns_5": 0.15115, "loss_cls_dn_0": 0.22783, "loss_box_dn_0": 0.78986, "loss_cls_dn_1": 0.15591, "loss_box_dn_1": 0.81492, "loss_cls_dn_2": 0.16225, "loss_box_dn_2": 0.8283, "loss_cls_dn_3": 0.16221, "loss_box_dn_3": 0.8497, "loss_cls_dn_4": 0.16847, "loss_box_dn_4": 0.8778, "loss_cls_dn_5": 0.17376, "loss_box_dn_5": 0.91337, "loss_dense_depth": 0.81843, "loss": 28.21712, "grad_norm": 54.12024, "time": 1.52523} -{"mode": "train", "epoch": 1, "iter": 140, "lr": 0.00016, "memory": 49163, "data_time": 0.08258, "loss_cls_0": 0.83767, "loss_box_0": 1.72702, "loss_cns_0": 0.62994, "loss_yns_0": 0.14863, "loss_cls_1": 0.93314, "loss_box_1": 1.84171, "loss_cns_1": 0.64763, "loss_yns_1": 0.14918, "loss_cls_2": 0.96251, "loss_box_2": 1.82313, "loss_cns_2": 0.65021, "loss_yns_2": 0.14943, "loss_cls_3": 0.96059, "loss_box_3": 1.80597, "loss_cns_3": 0.65076, "loss_yns_3": 0.14907, "loss_cls_4": 0.96436, "loss_box_4": 1.80802, "loss_cns_4": 0.649, "loss_yns_4": 0.15946, "loss_cls_5": 0.95375, "loss_box_5": 1.83914, "loss_cns_5": 0.64345, "loss_yns_5": 0.15193, "loss_cls_dn_0": 0.22551, "loss_box_dn_0": 0.79133, "loss_cls_dn_1": 0.1519, "loss_box_dn_1": 0.86893, "loss_cls_dn_2": 0.15655, "loss_box_dn_2": 0.885, "loss_cls_dn_3": 0.15826, "loss_box_dn_3": 0.90242, "loss_cls_dn_4": 0.16511, "loss_box_dn_4": 0.92249, "loss_cls_dn_5": 0.17025, "loss_box_dn_5": 0.9561, "loss_dense_depth": 0.81145, "loss": 28.40101, "grad_norm": 61.78885, "time": 1.54139} -{"mode": "train", "epoch": 1, "iter": 141, "lr": 0.00016, "memory": 49163, "data_time": 0.1496, "loss_cls_0": 0.88268, "loss_box_0": 1.7924, "loss_cns_0": 0.62552, "loss_yns_0": 0.1502, "loss_cls_1": 0.94268, "loss_box_1": 1.90504, "loss_cns_1": 0.64411, "loss_yns_1": 0.14991, "loss_cls_2": 0.96378, "loss_box_2": 1.85316, "loss_cns_2": 0.64808, "loss_yns_2": 0.15079, "loss_cls_3": 0.96187, "loss_box_3": 1.83934, "loss_cns_3": 0.64916, "loss_yns_3": 0.14985, "loss_cls_4": 0.97518, "loss_box_4": 1.83699, "loss_cns_4": 0.65175, "loss_yns_4": 0.15666, "loss_cls_5": 0.97238, "loss_box_5": 1.83963, "loss_cns_5": 0.64947, "loss_yns_5": 0.1525, "loss_cls_dn_0": 0.23548, "loss_box_dn_0": 0.79354, "loss_cls_dn_1": 0.15614, "loss_box_dn_1": 0.86245, "loss_cls_dn_2": 0.15885, "loss_box_dn_2": 0.8663, "loss_cls_dn_3": 0.15833, "loss_box_dn_3": 0.87505, "loss_cls_dn_4": 0.16368, "loss_box_dn_4": 0.88825, "loss_cls_dn_5": 0.17439, "loss_box_dn_5": 0.90333, "loss_dense_depth": 0.84065, "loss": 28.61956, "grad_norm": 42.70182, "time": 1.60158} -{"mode": "train", "epoch": 1, "iter": 142, "lr": 0.00016, "memory": 49163, "data_time": 0.07686, "loss_cls_0": 0.89481, "loss_box_0": 1.81738, "loss_cns_0": 0.61982, "loss_yns_0": 0.15139, "loss_cls_1": 0.95279, "loss_box_1": 1.87982, "loss_cns_1": 0.63891, "loss_yns_1": 0.15049, "loss_cls_2": 0.97123, "loss_box_2": 1.86433, "loss_cns_2": 0.64098, "loss_yns_2": 0.15027, "loss_cls_3": 0.96919, "loss_box_3": 1.8849, "loss_cns_3": 0.63966, "loss_yns_3": 0.15193, "loss_cls_4": 0.99135, "loss_box_4": 1.84958, "loss_cns_4": 0.6446, "loss_yns_4": 0.15181, "loss_cls_5": 0.9751, "loss_box_5": 1.85011, "loss_cns_5": 0.64269, "loss_yns_5": 0.15254, "loss_cls_dn_0": 0.23386, "loss_box_dn_0": 0.79054, "loss_cls_dn_1": 0.15602, "loss_box_dn_1": 0.84562, "loss_cls_dn_2": 0.15985, "loss_box_dn_2": 0.84552, "loss_cls_dn_3": 0.15757, "loss_box_dn_3": 0.85412, "loss_cls_dn_4": 0.16423, "loss_box_dn_4": 0.8492, "loss_cls_dn_5": 0.17236, "loss_box_dn_5": 0.85088, "loss_dense_depth": 0.88102, "loss": 28.59649, "grad_norm": 49.45667, "time": 1.5537} -{"mode": "train", "epoch": 1, "iter": 143, "lr": 0.00016, "memory": 49163, "data_time": 0.07472, "loss_cls_0": 0.86738, "loss_box_0": 1.78145, "loss_cns_0": 0.62277, "loss_yns_0": 0.15021, "loss_cls_1": 0.96416, "loss_box_1": 1.89388, "loss_cns_1": 0.64828, "loss_yns_1": 0.15042, "loss_cls_2": 0.97199, "loss_box_2": 1.85907, "loss_cns_2": 0.65022, "loss_yns_2": 0.14821, "loss_cls_3": 0.97821, "loss_box_3": 1.8477, "loss_cns_3": 0.64965, "loss_yns_3": 0.15037, "loss_cls_4": 0.98487, "loss_box_4": 1.83017, "loss_cns_4": 0.65259, "loss_yns_4": 0.15281, "loss_cls_5": 0.98036, "loss_box_5": 1.84362, "loss_cns_5": 0.65045, "loss_yns_5": 0.15105, "loss_cls_dn_0": 0.22521, "loss_box_dn_0": 0.78078, "loss_cls_dn_1": 0.15747, "loss_box_dn_1": 0.77147, "loss_cls_dn_2": 0.16124, "loss_box_dn_2": 0.75071, "loss_cls_dn_3": 0.1595, "loss_box_dn_3": 0.74961, "loss_cls_dn_4": 0.16822, "loss_box_dn_4": 0.74876, "loss_cls_dn_5": 0.17467, "loss_box_dn_5": 0.76254, "loss_dense_depth": 0.83195, "loss": 28.02205, "grad_norm": 34.73779, "time": 1.56288} -{"mode": "train", "epoch": 1, "iter": 144, "lr": 0.00016, "memory": 49163, "data_time": 0.09455, "loss_cls_0": 0.8913, "loss_box_0": 1.7957, "loss_cns_0": 0.61442, "loss_yns_0": 0.15156, "loss_cls_1": 0.97064, "loss_box_1": 1.90713, "loss_cns_1": 0.64027, "loss_yns_1": 0.15105, "loss_cls_2": 0.97562, "loss_box_2": 1.86991, "loss_cns_2": 0.64643, "loss_yns_2": 0.15052, "loss_cls_3": 0.97573, "loss_box_3": 1.86407, "loss_cns_3": 0.64624, "loss_yns_3": 0.15019, "loss_cls_4": 0.9966, "loss_box_4": 1.87004, "loss_cns_4": 0.64473, "loss_yns_4": 0.1588, "loss_cls_5": 0.98263, "loss_box_5": 1.87639, "loss_cns_5": 0.64587, "loss_yns_5": 0.15153, "loss_cls_dn_0": 0.23393, "loss_box_dn_0": 0.79266, "loss_cls_dn_1": 0.1651, "loss_box_dn_1": 0.78489, "loss_cls_dn_2": 0.16433, "loss_box_dn_2": 0.76897, "loss_cls_dn_3": 0.16199, "loss_box_dn_3": 0.7722, "loss_cls_dn_4": 0.17403, "loss_box_dn_4": 0.78663, "loss_cls_dn_5": 0.17717, "loss_box_dn_5": 0.81409, "loss_dense_depth": 0.84322, "loss": 28.36658, "grad_norm": 55.85024, "time": 1.55816} -{"mode": "train", "epoch": 1, "iter": 145, "lr": 0.00016, "memory": 49163, "data_time": 0.07343, "loss_cls_0": 0.88305, "loss_box_0": 1.78704, "loss_cns_0": 0.61746, "loss_yns_0": 0.15161, "loss_cls_1": 0.94832, "loss_box_1": 1.90765, "loss_cns_1": 0.64077, "loss_yns_1": 0.15261, "loss_cls_2": 0.96242, "loss_box_2": 1.87795, "loss_cns_2": 0.64849, "loss_yns_2": 0.15466, "loss_cls_3": 0.96792, "loss_box_3": 1.88372, "loss_cns_3": 0.64844, "loss_yns_3": 0.15362, "loss_cls_4": 1.00536, "loss_box_4": 1.87376, "loss_cns_4": 0.64632, "loss_yns_4": 0.16183, "loss_cls_5": 0.9761, "loss_box_5": 1.89157, "loss_cns_5": 0.64774, "loss_yns_5": 0.15307, "loss_cls_dn_0": 0.2337, "loss_box_dn_0": 0.78809, "loss_cls_dn_1": 0.16368, "loss_box_dn_1": 0.80186, "loss_cls_dn_2": 0.15985, "loss_box_dn_2": 0.79445, "loss_cls_dn_3": 0.1599, "loss_box_dn_3": 0.80574, "loss_cls_dn_4": 0.17034, "loss_box_dn_4": 0.827, "loss_cls_dn_5": 0.16891, "loss_box_dn_5": 0.86416, "loss_dense_depth": 0.90643, "loss": 28.58558, "grad_norm": 49.61567, "time": 1.52361} -{"mode": "train", "epoch": 1, "iter": 146, "lr": 0.00016, "memory": 49163, "data_time": 0.07585, "loss_cls_0": 0.85194, "loss_box_0": 1.77933, "loss_cns_0": 0.61889, "loss_yns_0": 0.14892, "loss_cls_1": 0.93019, "loss_box_1": 1.86354, "loss_cns_1": 0.63857, "loss_yns_1": 0.15051, "loss_cls_2": 0.96396, "loss_box_2": 1.81892, "loss_cns_2": 0.64595, "loss_yns_2": 0.14957, "loss_cls_3": 0.9673, "loss_box_3": 1.81854, "loss_cns_3": 0.64757, "loss_yns_3": 0.1511, "loss_cls_4": 1.00372, "loss_box_4": 1.80842, "loss_cns_4": 0.65009, "loss_yns_4": 0.15265, "loss_cls_5": 0.98271, "loss_box_5": 1.84261, "loss_cns_5": 0.64143, "loss_yns_5": 0.15009, "loss_cls_dn_0": 0.23136, "loss_box_dn_0": 0.78527, "loss_cls_dn_1": 0.15446, "loss_box_dn_1": 0.83543, "loss_cls_dn_2": 0.15617, "loss_box_dn_2": 0.83652, "loss_cls_dn_3": 0.15927, "loss_box_dn_3": 0.85086, "loss_cls_dn_4": 0.1709, "loss_box_dn_4": 0.89111, "loss_cls_dn_5": 0.17377, "loss_box_dn_5": 0.93961, "loss_dense_depth": 0.87145, "loss": 28.43272, "grad_norm": 55.79919, "time": 1.53463} -{"mode": "train", "epoch": 1, "iter": 147, "lr": 0.00016, "memory": 49163, "data_time": 0.07584, "loss_cls_0": 0.89647, "loss_box_0": 1.80052, "loss_cns_0": 0.62039, "loss_yns_0": 0.15068, "loss_cls_1": 0.94883, "loss_box_1": 1.86058, "loss_cns_1": 0.64077, "loss_yns_1": 0.15056, "loss_cls_2": 0.96548, "loss_box_2": 1.80231, "loss_cns_2": 0.64977, "loss_yns_2": 0.15019, "loss_cls_3": 0.97495, "loss_box_3": 1.80378, "loss_cns_3": 0.65455, "loss_yns_3": 0.15194, "loss_cls_4": 1.00647, "loss_box_4": 1.79482, "loss_cns_4": 0.66042, "loss_yns_4": 0.15057, "loss_cls_5": 0.99045, "loss_box_5": 1.82527, "loss_cns_5": 0.65244, "loss_yns_5": 0.1519, "loss_cls_dn_0": 0.23157, "loss_box_dn_0": 0.78445, "loss_cls_dn_1": 0.15511, "loss_box_dn_1": 0.88072, "loss_cls_dn_2": 0.16015, "loss_box_dn_2": 0.88693, "loss_cls_dn_3": 0.16385, "loss_box_dn_3": 0.89942, "loss_cls_dn_4": 0.17282, "loss_box_dn_4": 0.92836, "loss_cls_dn_5": 0.17509, "loss_box_dn_5": 0.96806, "loss_dense_depth": 0.83851, "loss": 28.69915, "grad_norm": 51.84541, "time": 1.5155} -{"mode": "train", "epoch": 1, "iter": 148, "lr": 0.00016, "memory": 49163, "data_time": 0.07555, "loss_cls_0": 0.88002, "loss_box_0": 1.78843, "loss_cns_0": 0.62037, "loss_yns_0": 0.15006, "loss_cls_1": 0.93375, "loss_box_1": 1.84063, "loss_cns_1": 0.64569, "loss_yns_1": 0.14743, "loss_cls_2": 0.95995, "loss_box_2": 1.78229, "loss_cns_2": 0.65186, "loss_yns_2": 0.14839, "loss_cls_3": 0.96307, "loss_box_3": 1.79824, "loss_cns_3": 0.6545, "loss_yns_3": 0.15024, "loss_cls_4": 0.98158, "loss_box_4": 1.78912, "loss_cns_4": 0.65364, "loss_yns_4": 0.1493, "loss_cls_5": 0.98345, "loss_box_5": 1.78636, "loss_cns_5": 0.65939, "loss_yns_5": 0.14963, "loss_cls_dn_0": 0.22312, "loss_box_dn_0": 0.78119, "loss_cls_dn_1": 0.15307, "loss_box_dn_1": 0.87063, "loss_cls_dn_2": 0.15655, "loss_box_dn_2": 0.85357, "loss_cls_dn_3": 0.15924, "loss_box_dn_3": 0.86742, "loss_cls_dn_4": 0.16345, "loss_box_dn_4": 0.87598, "loss_cls_dn_5": 0.17087, "loss_box_dn_5": 0.89573, "loss_dense_depth": 0.86836, "loss": 28.30655, "grad_norm": 46.7675, "time": 1.51117} -{"mode": "train", "epoch": 1, "iter": 149, "lr": 0.00016, "memory": 49163, "data_time": 0.07831, "loss_cls_0": 0.84698, "loss_box_0": 1.77633, "loss_cns_0": 0.61627, "loss_yns_0": 0.14727, "loss_cls_1": 0.9245, "loss_box_1": 1.85237, "loss_cns_1": 0.641, "loss_yns_1": 0.14765, "loss_cls_2": 0.96447, "loss_box_2": 1.79788, "loss_cns_2": 0.64826, "loss_yns_2": 0.14937, "loss_cls_3": 0.97232, "loss_box_3": 1.80087, "loss_cns_3": 0.64933, "loss_yns_3": 0.14989, "loss_cls_4": 1.00464, "loss_box_4": 1.78802, "loss_cns_4": 0.64807, "loss_yns_4": 0.14889, "loss_cls_5": 0.98002, "loss_box_5": 1.78244, "loss_cns_5": 0.65171, "loss_yns_5": 0.15067, "loss_cls_dn_0": 0.2199, "loss_box_dn_0": 0.77689, "loss_cls_dn_1": 0.14753, "loss_box_dn_1": 0.80467, "loss_cls_dn_2": 0.15215, "loss_box_dn_2": 0.77649, "loss_cls_dn_3": 0.15309, "loss_box_dn_3": 0.77795, "loss_cls_dn_4": 0.16083, "loss_box_dn_4": 0.77345, "loss_cls_dn_5": 0.16571, "loss_box_dn_5": 0.77459, "loss_dense_depth": 0.82224, "loss": 27.74474, "grad_norm": 41.91705, "time": 1.51885} -{"mode": "train", "epoch": 1, "iter": 150, "lr": 0.00016, "memory": 49163, "data_time": 0.07725, "loss_cls_0": 0.87963, "loss_box_0": 1.73869, "loss_cns_0": 0.61575, "loss_yns_0": 0.14605, "loss_cls_1": 0.92626, "loss_box_1": 1.86841, "loss_cns_1": 0.63689, "loss_yns_1": 0.14768, "loss_cls_2": 0.97202, "loss_box_2": 1.78196, "loss_cns_2": 0.64852, "loss_yns_2": 0.14958, "loss_cls_3": 0.96531, "loss_box_3": 1.7713, "loss_cns_3": 0.65299, "loss_yns_3": 0.14912, "loss_cls_4": 0.98509, "loss_box_4": 1.76969, "loss_cns_4": 0.65403, "loss_yns_4": 0.15057, "loss_cls_5": 0.97716, "loss_box_5": 1.7817, "loss_cns_5": 0.65014, "loss_yns_5": 0.14794, "loss_cls_dn_0": 0.22784, "loss_box_dn_0": 0.77481, "loss_cls_dn_1": 0.14654, "loss_box_dn_1": 0.72632, "loss_cls_dn_2": 0.15035, "loss_box_dn_2": 0.70255, "loss_cls_dn_3": 0.1523, "loss_box_dn_3": 0.69765, "loss_cls_dn_4": 0.16067, "loss_box_dn_4": 0.70405, "loss_cls_dn_5": 0.16296, "loss_box_dn_5": 0.7126, "loss_dense_depth": 0.88083, "loss": 27.36597, "grad_norm": 35.58796, "time": 1.52781} -{"mode": "train", "epoch": 1, "iter": 151, "lr": 0.00016, "memory": 49163, "data_time": 0.07453, "loss_cls_0": 0.8735, "loss_box_0": 1.7649, "loss_cns_0": 0.62302, "loss_yns_0": 0.14679, "loss_cls_1": 0.92637, "loss_box_1": 1.82032, "loss_cns_1": 0.64151, "loss_yns_1": 0.14852, "loss_cls_2": 0.95542, "loss_box_2": 1.76966, "loss_cns_2": 0.64857, "loss_yns_2": 0.14859, "loss_cls_3": 0.96474, "loss_box_3": 1.75949, "loss_cns_3": 0.64805, "loss_yns_3": 0.15047, "loss_cls_4": 0.97691, "loss_box_4": 1.76235, "loss_cns_4": 0.6515, "loss_yns_4": 0.1508, "loss_cls_5": 0.97669, "loss_box_5": 1.76949, "loss_cns_5": 0.64836, "loss_yns_5": 0.14903, "loss_cls_dn_0": 0.22848, "loss_box_dn_0": 0.7807, "loss_cls_dn_1": 0.14431, "loss_box_dn_1": 0.74581, "loss_cls_dn_2": 0.14536, "loss_box_dn_2": 0.74465, "loss_cls_dn_3": 0.14785, "loss_box_dn_3": 0.76264, "loss_cls_dn_4": 0.15392, "loss_box_dn_4": 0.78436, "loss_cls_dn_5": 0.16123, "loss_box_dn_5": 0.8079, "loss_dense_depth": 0.83015, "loss": 27.51238, "grad_norm": 44.26065, "time": 1.52669} -{"mode": "train", "epoch": 1, "iter": 152, "lr": 0.00016, "memory": 49163, "data_time": 0.07581, "loss_cls_0": 0.86415, "loss_box_0": 1.76914, "loss_cns_0": 0.62012, "loss_yns_0": 0.15025, "loss_cls_1": 0.9486, "loss_box_1": 1.84873, "loss_cns_1": 0.63973, "loss_yns_1": 0.14993, "loss_cls_2": 0.97226, "loss_box_2": 1.82674, "loss_cns_2": 0.64581, "loss_yns_2": 0.15042, "loss_cls_3": 0.9722, "loss_box_3": 1.81, "loss_cns_3": 0.64636, "loss_yns_3": 0.15218, "loss_cls_4": 0.98393, "loss_box_4": 1.81499, "loss_cns_4": 0.6479, "loss_yns_4": 0.15365, "loss_cls_5": 0.98392, "loss_box_5": 1.8155, "loss_cns_5": 0.64767, "loss_yns_5": 0.15393, "loss_cls_dn_0": 0.22512, "loss_box_dn_0": 0.78916, "loss_cls_dn_1": 0.14711, "loss_box_dn_1": 0.79783, "loss_cls_dn_2": 0.14789, "loss_box_dn_2": 0.81256, "loss_cls_dn_3": 0.15017, "loss_box_dn_3": 0.84013, "loss_cls_dn_4": 0.15544, "loss_box_dn_4": 0.86505, "loss_cls_dn_5": 0.16157, "loss_box_dn_5": 0.89258, "loss_dense_depth": 0.7998, "loss": 28.1525, "grad_norm": 51.58538, "time": 1.50947} -{"mode": "train", "epoch": 1, "iter": 153, "lr": 0.00016, "memory": 49163, "data_time": 0.07578, "loss_cls_0": 0.88114, "loss_box_0": 1.75118, "loss_cns_0": 0.62176, "loss_yns_0": 0.15089, "loss_cls_1": 0.95242, "loss_box_1": 1.85487, "loss_cns_1": 0.64017, "loss_yns_1": 0.15297, "loss_cls_2": 0.96791, "loss_box_2": 1.82426, "loss_cns_2": 0.64713, "loss_yns_2": 0.15211, "loss_cls_3": 0.97902, "loss_box_3": 1.80182, "loss_cns_3": 0.64762, "loss_yns_3": 0.15284, "loss_cls_4": 1.00384, "loss_box_4": 1.81641, "loss_cns_4": 0.64613, "loss_yns_4": 0.1541, "loss_cls_5": 0.99893, "loss_box_5": 1.83039, "loss_cns_5": 0.6435, "loss_yns_5": 0.15496, "loss_cls_dn_0": 0.23366, "loss_box_dn_0": 0.78663, "loss_cls_dn_1": 0.14831, "loss_box_dn_1": 0.8273, "loss_cls_dn_2": 0.14807, "loss_box_dn_2": 0.83713, "loss_cls_dn_3": 0.14743, "loss_box_dn_3": 0.85411, "loss_cls_dn_4": 0.15444, "loss_box_dn_4": 0.87622, "loss_cls_dn_5": 0.16257, "loss_box_dn_5": 0.90427, "loss_dense_depth": 0.85762, "loss": 28.36412, "grad_norm": 43.59132, "time": 1.52307} -{"mode": "train", "epoch": 1, "iter": 154, "lr": 0.00016, "memory": 49163, "data_time": 0.0761, "loss_cls_0": 0.86019, "loss_box_0": 1.74117, "loss_cns_0": 0.62053, "loss_yns_0": 0.14959, "loss_cls_1": 0.94527, "loss_box_1": 1.78829, "loss_cns_1": 0.64066, "loss_yns_1": 0.14943, "loss_cls_2": 0.94916, "loss_box_2": 1.74256, "loss_cns_2": 0.64862, "loss_yns_2": 0.14971, "loss_cls_3": 0.96553, "loss_box_3": 1.72876, "loss_cns_3": 0.64937, "loss_yns_3": 0.14972, "loss_cls_4": 1.00009, "loss_box_4": 1.73076, "loss_cns_4": 0.65222, "loss_yns_4": 0.1506, "loss_cls_5": 0.96872, "loss_box_5": 1.73946, "loss_cns_5": 0.64971, "loss_yns_5": 0.15121, "loss_cls_dn_0": 0.22628, "loss_box_dn_0": 0.77597, "loss_cls_dn_1": 0.14848, "loss_box_dn_1": 0.83104, "loss_cls_dn_2": 0.14678, "loss_box_dn_2": 0.82209, "loss_cls_dn_3": 0.14842, "loss_box_dn_3": 0.82393, "loss_cls_dn_4": 0.15326, "loss_box_dn_4": 0.83328, "loss_cls_dn_5": 0.16015, "loss_box_dn_5": 0.84544, "loss_dense_depth": 0.76872, "loss": 27.60516, "grad_norm": 40.77505, "time": 1.51211} -{"mode": "train", "epoch": 1, "iter": 155, "lr": 0.00016, "memory": 49163, "data_time": 0.07752, "loss_cls_0": 0.83411, "loss_box_0": 1.75472, "loss_cns_0": 0.61842, "loss_yns_0": 0.14971, "loss_cls_1": 0.90261, "loss_box_1": 1.74561, "loss_cns_1": 0.64291, "loss_yns_1": 0.14985, "loss_cls_2": 0.9291, "loss_box_2": 1.72007, "loss_cns_2": 0.649, "loss_yns_2": 0.14884, "loss_cls_3": 0.93168, "loss_box_3": 1.71233, "loss_cns_3": 0.65127, "loss_yns_3": 0.14892, "loss_cls_4": 0.95464, "loss_box_4": 1.6953, "loss_cns_4": 0.653, "loss_yns_4": 0.14999, "loss_cls_5": 0.95377, "loss_box_5": 1.68306, "loss_cns_5": 0.65243, "loss_yns_5": 0.14947, "loss_cls_dn_0": 0.22038, "loss_box_dn_0": 0.7895, "loss_cls_dn_1": 0.14721, "loss_box_dn_1": 0.82759, "loss_cls_dn_2": 0.14291, "loss_box_dn_2": 0.81344, "loss_cls_dn_3": 0.14402, "loss_box_dn_3": 0.80999, "loss_cls_dn_4": 0.15078, "loss_box_dn_4": 0.80826, "loss_cls_dn_5": 0.15526, "loss_box_dn_5": 0.80495, "loss_dense_depth": 0.79989, "loss": 27.19501, "grad_norm": 39.89589, "time": 1.57666} -{"mode": "train", "epoch": 1, "iter": 156, "lr": 0.00016, "memory": 49163, "data_time": 0.07473, "loss_cls_0": 0.85855, "loss_box_0": 1.74252, "loss_cns_0": 0.62214, "loss_yns_0": 0.15136, "loss_cls_1": 0.93861, "loss_box_1": 1.76867, "loss_cns_1": 0.63885, "loss_yns_1": 0.15053, "loss_cls_2": 0.95878, "loss_box_2": 1.72889, "loss_cns_2": 0.64974, "loss_yns_2": 0.15158, "loss_cls_3": 0.97228, "loss_box_3": 1.69846, "loss_cns_3": 0.65082, "loss_yns_3": 0.15298, "loss_cls_4": 0.9556, "loss_box_4": 1.69501, "loss_cns_4": 0.65113, "loss_yns_4": 0.15183, "loss_cls_5": 0.96397, "loss_box_5": 1.69065, "loss_cns_5": 0.64885, "loss_yns_5": 0.15225, "loss_cls_dn_0": 0.22478, "loss_box_dn_0": 0.77941, "loss_cls_dn_1": 0.1468, "loss_box_dn_1": 0.81173, "loss_cls_dn_2": 0.14363, "loss_box_dn_2": 0.7919, "loss_cls_dn_3": 0.14615, "loss_box_dn_3": 0.7784, "loss_cls_dn_4": 0.14846, "loss_box_dn_4": 0.77894, "loss_cls_dn_5": 0.15479, "loss_box_dn_5": 0.78021, "loss_dense_depth": 0.80162, "loss": 27.23088, "grad_norm": 45.53585, "time": 1.51939} -{"mode": "train", "epoch": 1, "iter": 157, "lr": 0.00016, "memory": 49163, "data_time": 0.07433, "loss_cls_0": 0.85091, "loss_box_0": 1.72767, "loss_cns_0": 0.62003, "loss_yns_0": 0.15383, "loss_cls_1": 0.91413, "loss_box_1": 1.82361, "loss_cns_1": 0.6355, "loss_yns_1": 0.15394, "loss_cls_2": 0.95098, "loss_box_2": 1.76375, "loss_cns_2": 0.64797, "loss_yns_2": 0.15181, "loss_cls_3": 0.95475, "loss_box_3": 1.74393, "loss_cns_3": 0.6457, "loss_yns_3": 0.15596, "loss_cls_4": 0.95456, "loss_box_4": 1.74908, "loss_cns_4": 0.64673, "loss_yns_4": 0.15307, "loss_cls_5": 0.97048, "loss_box_5": 1.75965, "loss_cns_5": 0.64307, "loss_yns_5": 0.15411, "loss_cls_dn_0": 0.2214, "loss_box_dn_0": 0.77933, "loss_cls_dn_1": 0.14272, "loss_box_dn_1": 0.76675, "loss_cls_dn_2": 0.14348, "loss_box_dn_2": 0.75151, "loss_cls_dn_3": 0.14837, "loss_box_dn_3": 0.75858, "loss_cls_dn_4": 0.14989, "loss_box_dn_4": 0.77654, "loss_cls_dn_5": 0.16318, "loss_box_dn_5": 0.80382, "loss_dense_depth": 0.7928, "loss": 27.32361, "grad_norm": 49.06209, "time": 1.52132} -{"mode": "train", "epoch": 1, "iter": 158, "lr": 0.00016, "memory": 49163, "data_time": 0.07595, "loss_cls_0": 0.82638, "loss_box_0": 1.72898, "loss_cns_0": 0.61486, "loss_yns_0": 0.15302, "loss_cls_1": 0.91202, "loss_box_1": 1.80018, "loss_cns_1": 0.63999, "loss_yns_1": 0.15331, "loss_cls_2": 0.95193, "loss_box_2": 1.74587, "loss_cns_2": 0.64491, "loss_yns_2": 0.15091, "loss_cls_3": 0.95049, "loss_box_3": 1.75319, "loss_cns_3": 0.64585, "loss_yns_3": 0.15342, "loss_cls_4": 0.94265, "loss_box_4": 1.76759, "loss_cns_4": 0.65024, "loss_yns_4": 0.15249, "loss_cls_5": 0.94576, "loss_box_5": 1.76428, "loss_cns_5": 0.64474, "loss_yns_5": 0.15343, "loss_cls_dn_0": 0.21436, "loss_box_dn_0": 0.7817, "loss_cls_dn_1": 0.14241, "loss_box_dn_1": 0.8267, "loss_cls_dn_2": 0.14508, "loss_box_dn_2": 0.83139, "loss_cls_dn_3": 0.1502, "loss_box_dn_3": 0.86769, "loss_cls_dn_4": 0.14981, "loss_box_dn_4": 0.91427, "loss_cls_dn_5": 0.15859, "loss_box_dn_5": 0.9554, "loss_dense_depth": 0.74767, "loss": 27.73177, "grad_norm": 55.45279, "time": 1.54444} -{"mode": "train", "epoch": 1, "iter": 159, "lr": 0.00016, "memory": 49163, "data_time": 0.08127, "loss_cls_0": 0.83186, "loss_box_0": 1.73686, "loss_cns_0": 0.61624, "loss_yns_0": 0.15472, "loss_cls_1": 0.92279, "loss_box_1": 1.80088, "loss_cns_1": 0.64287, "loss_yns_1": 0.15323, "loss_cls_2": 0.94453, "loss_box_2": 1.78001, "loss_cns_2": 0.64639, "loss_yns_2": 0.15222, "loss_cls_3": 0.93912, "loss_box_3": 1.76867, "loss_cns_3": 0.64891, "loss_yns_3": 0.15371, "loss_cls_4": 0.96073, "loss_box_4": 1.77649, "loss_cns_4": 0.654, "loss_yns_4": 0.15341, "loss_cls_5": 0.94338, "loss_box_5": 1.77673, "loss_cns_5": 0.64859, "loss_yns_5": 0.15388, "loss_cls_dn_0": 0.21474, "loss_box_dn_0": 0.7745, "loss_cls_dn_1": 0.14224, "loss_box_dn_1": 0.87959, "loss_cls_dn_2": 0.14477, "loss_box_dn_2": 0.89643, "loss_cls_dn_3": 0.14785, "loss_box_dn_3": 0.924, "loss_cls_dn_4": 0.14979, "loss_box_dn_4": 0.97389, "loss_cls_dn_5": 0.15696, "loss_box_dn_5": 1.0114, "loss_dense_depth": 0.77014, "loss": 28.14654, "grad_norm": 61.80776, "time": 1.52096} -{"mode": "train", "epoch": 1, "iter": 160, "lr": 0.00016, "memory": 49163, "data_time": 0.07973, "loss_cls_0": 0.82125, "loss_box_0": 1.74118, "loss_cns_0": 0.61586, "loss_yns_0": 0.15449, "loss_cls_1": 0.89819, "loss_box_1": 1.79399, "loss_cns_1": 0.64307, "loss_yns_1": 0.15487, "loss_cls_2": 0.92447, "loss_box_2": 1.76727, "loss_cns_2": 0.64876, "loss_yns_2": 0.15463, "loss_cls_3": 0.92649, "loss_box_3": 1.73075, "loss_cns_3": 0.65137, "loss_yns_3": 0.15545, "loss_cls_4": 0.9486, "loss_box_4": 1.73734, "loss_cns_4": 0.652, "loss_yns_4": 0.15377, "loss_cls_5": 0.93249, "loss_box_5": 1.74526, "loss_cns_5": 0.64901, "loss_yns_5": 0.15487, "loss_cls_dn_0": 0.21391, "loss_box_dn_0": 0.77133, "loss_cls_dn_1": 0.14077, "loss_box_dn_1": 0.86514, "loss_cls_dn_2": 0.14254, "loss_box_dn_2": 0.87574, "loss_cls_dn_3": 0.14613, "loss_box_dn_3": 0.87652, "loss_cls_dn_4": 0.15172, "loss_box_dn_4": 0.90798, "loss_cls_dn_5": 0.15659, "loss_box_dn_5": 0.93771, "loss_dense_depth": 0.78144, "loss": 27.72293, "grad_norm": 37.1007, "time": 1.51301} -{"mode": "train", "epoch": 1, "iter": 161, "lr": 0.00016, "memory": 49163, "data_time": 0.15483, "loss_cls_0": 0.81441, "loss_box_0": 1.76632, "loss_cns_0": 0.61849, "loss_yns_0": 0.15478, "loss_cls_1": 0.89044, "loss_box_1": 1.79352, "loss_cns_1": 0.64379, "loss_yns_1": 0.15772, "loss_cls_2": 0.9299, "loss_box_2": 1.7446, "loss_cns_2": 0.65229, "loss_yns_2": 0.15526, "loss_cls_3": 0.95447, "loss_box_3": 1.71876, "loss_cns_3": 0.65627, "loss_yns_3": 0.1577, "loss_cls_4": 0.94502, "loss_box_4": 1.72267, "loss_cns_4": 0.654, "loss_yns_4": 0.15604, "loss_cls_5": 0.94657, "loss_box_5": 1.70208, "loss_cns_5": 0.65289, "loss_yns_5": 0.1567, "loss_cls_dn_0": 0.21296, "loss_box_dn_0": 0.77859, "loss_cls_dn_1": 0.14049, "loss_box_dn_1": 0.82084, "loss_cls_dn_2": 0.14361, "loss_box_dn_2": 0.81302, "loss_cls_dn_3": 0.15009, "loss_box_dn_3": 0.80257, "loss_cls_dn_4": 0.15162, "loss_box_dn_4": 0.81304, "loss_cls_dn_5": 0.15747, "loss_box_dn_5": 0.81911, "loss_dense_depth": 0.76841, "loss": 27.31654, "grad_norm": 55.74436, "time": 1.57103} -{"mode": "train", "epoch": 1, "iter": 162, "lr": 0.00016, "memory": 49163, "data_time": 0.07377, "loss_cls_0": 0.82906, "loss_box_0": 1.75075, "loss_cns_0": 0.61492, "loss_yns_0": 0.1561, "loss_cls_1": 0.88062, "loss_box_1": 1.80771, "loss_cns_1": 0.63742, "loss_yns_1": 0.15843, "loss_cls_2": 0.93041, "loss_box_2": 1.73045, "loss_cns_2": 0.65126, "loss_yns_2": 0.15781, "loss_cls_3": 0.944, "loss_box_3": 1.71617, "loss_cns_3": 0.65729, "loss_yns_3": 0.1572, "loss_cls_4": 0.9734, "loss_box_4": 1.70563, "loss_cns_4": 0.65389, "loss_yns_4": 0.15788, "loss_cls_5": 0.95912, "loss_box_5": 1.67919, "loss_cns_5": 0.65598, "loss_yns_5": 0.15691, "loss_cls_dn_0": 0.21614, "loss_box_dn_0": 0.7815, "loss_cls_dn_1": 0.13917, "loss_box_dn_1": 0.76371, "loss_cls_dn_2": 0.13975, "loss_box_dn_2": 0.73952, "loss_cls_dn_3": 0.14473, "loss_box_dn_3": 0.72668, "loss_cls_dn_4": 0.15473, "loss_box_dn_4": 0.71882, "loss_cls_dn_5": 0.1585, "loss_box_dn_5": 0.71842, "loss_dense_depth": 0.7539, "loss": 26.87717, "grad_norm": 40.0695, "time": 1.54614} -{"mode": "train", "epoch": 1, "iter": 163, "lr": 0.00016, "memory": 49163, "data_time": 0.07547, "loss_cls_0": 0.82057, "loss_box_0": 1.74706, "loss_cns_0": 0.61815, "loss_yns_0": 0.1569, "loss_cls_1": 0.87945, "loss_box_1": 1.83667, "loss_cns_1": 0.63082, "loss_yns_1": 0.15768, "loss_cls_2": 0.92039, "loss_box_2": 1.73843, "loss_cns_2": 0.65289, "loss_yns_2": 0.15792, "loss_cls_3": 0.92554, "loss_box_3": 1.71017, "loss_cns_3": 0.65745, "loss_yns_3": 0.15953, "loss_cls_4": 0.95939, "loss_box_4": 1.70537, "loss_cns_4": 0.65546, "loss_yns_4": 0.15801, "loss_cls_5": 0.93415, "loss_box_5": 1.7093, "loss_cns_5": 0.6551, "loss_yns_5": 0.15795, "loss_cls_dn_0": 0.21318, "loss_box_dn_0": 0.7735, "loss_cls_dn_1": 0.14279, "loss_box_dn_1": 0.73948, "loss_cls_dn_2": 0.14372, "loss_box_dn_2": 0.71765, "loss_cls_dn_3": 0.14565, "loss_box_dn_3": 0.7058, "loss_cls_dn_4": 0.15057, "loss_box_dn_4": 0.70838, "loss_cls_dn_5": 0.15156, "loss_box_dn_5": 0.71554, "loss_dense_depth": 0.76228, "loss": 26.77447, "grad_norm": 42.45192, "time": 1.57142} -{"mode": "train", "epoch": 1, "iter": 164, "lr": 0.00017, "memory": 49163, "data_time": 0.09678, "loss_cls_0": 0.82419, "loss_box_0": 1.77601, "loss_cns_0": 0.62249, "loss_yns_0": 0.1585, "loss_cls_1": 0.88134, "loss_box_1": 1.85282, "loss_cns_1": 0.63742, "loss_yns_1": 0.16036, "loss_cls_2": 0.92945, "loss_box_2": 1.74405, "loss_cns_2": 0.65475, "loss_yns_2": 0.15898, "loss_cls_3": 0.9322, "loss_box_3": 1.7346, "loss_cns_3": 0.65564, "loss_yns_3": 0.16125, "loss_cls_4": 0.95203, "loss_box_4": 1.74219, "loss_cns_4": 0.65521, "loss_yns_4": 0.1606, "loss_cls_5": 0.94385, "loss_box_5": 1.73778, "loss_cns_5": 0.65324, "loss_yns_5": 0.16117, "loss_cls_dn_0": 0.2127, "loss_box_dn_0": 0.78059, "loss_cls_dn_1": 0.13729, "loss_box_dn_1": 0.75658, "loss_cls_dn_2": 0.13876, "loss_box_dn_2": 0.74772, "loss_cls_dn_3": 0.14048, "loss_box_dn_3": 0.75724, "loss_cls_dn_4": 0.14479, "loss_box_dn_4": 0.77652, "loss_cls_dn_5": 0.15012, "loss_box_dn_5": 0.79555, "loss_dense_depth": 0.74871, "loss": 27.17717, "grad_norm": 46.09692, "time": 1.5746} -{"mode": "train", "epoch": 1, "iter": 165, "lr": 0.00017, "memory": 49163, "data_time": 0.07708, "loss_cls_0": 0.83213, "loss_box_0": 1.80375, "loss_cns_0": 0.61533, "loss_yns_0": 0.15881, "loss_cls_1": 0.89775, "loss_box_1": 1.80526, "loss_cns_1": 0.63942, "loss_yns_1": 0.16092, "loss_cls_2": 0.94485, "loss_box_2": 1.7439, "loss_cns_2": 0.65228, "loss_yns_2": 0.16043, "loss_cls_3": 0.94362, "loss_box_3": 1.7301, "loss_cns_3": 0.65468, "loss_yns_3": 0.16127, "loss_cls_4": 0.94287, "loss_box_4": 1.72952, "loss_cns_4": 0.65566, "loss_yns_4": 0.16289, "loss_cls_5": 0.94865, "loss_box_5": 1.74838, "loss_cns_5": 0.65225, "loss_yns_5": 0.16068, "loss_cls_dn_0": 0.21483, "loss_box_dn_0": 0.7814, "loss_cls_dn_1": 0.13496, "loss_box_dn_1": 0.79973, "loss_cls_dn_2": 0.13695, "loss_box_dn_2": 0.80024, "loss_cls_dn_3": 0.13889, "loss_box_dn_3": 0.8201, "loss_cls_dn_4": 0.14497, "loss_box_dn_4": 0.83858, "loss_cls_dn_5": 0.15334, "loss_box_dn_5": 0.88316, "loss_dense_depth": 0.7741, "loss": 27.52664, "grad_norm": 37.65358, "time": 1.53409} -{"mode": "train", "epoch": 1, "iter": 166, "lr": 0.00017, "memory": 49163, "data_time": 0.07652, "loss_cls_0": 0.81959, "loss_box_0": 1.78574, "loss_cns_0": 0.61179, "loss_yns_0": 0.15635, "loss_cls_1": 0.90212, "loss_box_1": 1.76926, "loss_cns_1": 0.64946, "loss_yns_1": 0.15983, "loss_cls_2": 0.94085, "loss_box_2": 1.7096, "loss_cns_2": 0.65903, "loss_yns_2": 0.16017, "loss_cls_3": 0.94029, "loss_box_3": 1.68394, "loss_cns_3": 0.66155, "loss_yns_3": 0.16032, "loss_cls_4": 0.94443, "loss_box_4": 1.6807, "loss_cns_4": 0.66143, "loss_yns_4": 0.16014, "loss_cls_5": 0.93357, "loss_box_5": 1.70263, "loss_cns_5": 0.65914, "loss_yns_5": 0.16098, "loss_cls_dn_0": 0.20913, "loss_box_dn_0": 0.77213, "loss_cls_dn_1": 0.13594, "loss_box_dn_1": 0.83963, "loss_cls_dn_2": 0.13867, "loss_box_dn_2": 0.84098, "loss_cls_dn_3": 0.14108, "loss_box_dn_3": 0.84724, "loss_cls_dn_4": 0.14712, "loss_box_dn_4": 0.85978, "loss_cls_dn_5": 0.15166, "loss_box_dn_5": 0.90133, "loss_dense_depth": 0.77927, "loss": 27.43688, "grad_norm": 50.63619, "time": 1.51874} -{"mode": "train", "epoch": 1, "iter": 167, "lr": 0.00017, "memory": 49163, "data_time": 0.07561, "loss_cls_0": 0.87024, "loss_box_0": 1.78085, "loss_cns_0": 0.60961, "loss_yns_0": 0.15957, "loss_cls_1": 0.93515, "loss_box_1": 1.74563, "loss_cns_1": 0.65008, "loss_yns_1": 0.1613, "loss_cls_2": 0.9562, "loss_box_2": 1.69475, "loss_cns_2": 0.65585, "loss_yns_2": 0.1597, "loss_cls_3": 0.9548, "loss_box_3": 1.66729, "loss_cns_3": 0.65787, "loss_yns_3": 0.16171, "loss_cls_4": 0.97373, "loss_box_4": 1.65937, "loss_cns_4": 0.65905, "loss_yns_4": 0.15822, "loss_cls_5": 0.96348, "loss_box_5": 1.66143, "loss_cns_5": 0.65929, "loss_yns_5": 0.16079, "loss_cls_dn_0": 0.22156, "loss_box_dn_0": 0.77399, "loss_cls_dn_1": 0.14482, "loss_box_dn_1": 0.87636, "loss_cls_dn_2": 0.14629, "loss_box_dn_2": 0.86825, "loss_cls_dn_3": 0.1478, "loss_box_dn_3": 0.86375, "loss_cls_dn_4": 0.15245, "loss_box_dn_4": 0.86485, "loss_cls_dn_5": 0.15497, "loss_box_dn_5": 0.88484, "loss_dense_depth": 0.8108, "loss": 27.6267, "grad_norm": 33.67125, "time": 6.74274} -{"mode": "train", "epoch": 1, "iter": 168, "lr": 0.00017, "memory": 49163, "data_time": 0.07336, "loss_cls_0": 0.85098, "loss_box_0": 1.75178, "loss_cns_0": 0.6218, "loss_yns_0": 0.16092, "loss_cls_1": 0.92605, "loss_box_1": 1.72854, "loss_cns_1": 0.6478, "loss_yns_1": 0.16244, "loss_cls_2": 0.94994, "loss_box_2": 1.67217, "loss_cns_2": 0.65515, "loss_yns_2": 0.16039, "loss_cls_3": 0.95048, "loss_box_3": 1.67115, "loss_cns_3": 0.65566, "loss_yns_3": 0.16166, "loss_cls_4": 0.96495, "loss_box_4": 1.64997, "loss_cns_4": 0.65915, "loss_yns_4": 0.16178, "loss_cls_5": 0.96818, "loss_box_5": 1.64694, "loss_cns_5": 0.65789, "loss_yns_5": 0.16186, "loss_cls_dn_0": 0.21361, "loss_box_dn_0": 0.76636, "loss_cls_dn_1": 0.14462, "loss_box_dn_1": 0.81848, "loss_cls_dn_2": 0.14368, "loss_box_dn_2": 0.79489, "loss_cls_dn_3": 0.14474, "loss_box_dn_3": 0.78687, "loss_cls_dn_4": 0.14823, "loss_box_dn_4": 0.77739, "loss_cls_dn_5": 0.15388, "loss_box_dn_5": 0.78108, "loss_dense_depth": 0.82406, "loss": 27.09553, "grad_norm": 33.40876, "time": 1.49457} -{"mode": "train", "epoch": 1, "iter": 169, "lr": 0.00017, "memory": 49163, "data_time": 0.07572, "loss_cls_0": 0.89632, "loss_box_0": 1.78148, "loss_cns_0": 0.61759, "loss_yns_0": 0.16211, "loss_cls_1": 0.94292, "loss_box_1": 1.75678, "loss_cns_1": 0.64416, "loss_yns_1": 0.16134, "loss_cls_2": 0.96213, "loss_box_2": 1.70157, "loss_cns_2": 0.64787, "loss_yns_2": 0.16109, "loss_cls_3": 0.97345, "loss_box_3": 1.69489, "loss_cns_3": 0.64594, "loss_yns_3": 0.16067, "loss_cls_4": 0.99868, "loss_box_4": 1.65749, "loss_cns_4": 0.6467, "loss_yns_4": 0.15968, "loss_cls_5": 0.98392, "loss_box_5": 1.6761, "loss_cns_5": 0.65212, "loss_yns_5": 0.16234, "loss_cls_dn_0": 0.21775, "loss_box_dn_0": 0.77259, "loss_cls_dn_1": 0.13959, "loss_box_dn_1": 0.76914, "loss_cls_dn_2": 0.13803, "loss_box_dn_2": 0.74535, "loss_cls_dn_3": 0.13913, "loss_box_dn_3": 0.74511, "loss_cls_dn_4": 0.14542, "loss_box_dn_4": 0.73261, "loss_cls_dn_5": 0.15024, "loss_box_dn_5": 0.73491, "loss_dense_depth": 0.88724, "loss": 27.16445, "grad_norm": 31.90306, "time": 1.50593} -{"mode": "train", "epoch": 1, "iter": 170, "lr": 0.00017, "memory": 49163, "data_time": 0.07631, "loss_cls_0": 0.85857, "loss_box_0": 1.73504, "loss_cns_0": 0.62112, "loss_yns_0": 0.15956, "loss_cls_1": 0.93655, "loss_box_1": 1.72342, "loss_cns_1": 0.64815, "loss_yns_1": 0.16114, "loss_cls_2": 0.95807, "loss_box_2": 1.66086, "loss_cns_2": 0.65189, "loss_yns_2": 0.1618, "loss_cls_3": 0.96496, "loss_box_3": 1.65865, "loss_cns_3": 0.6526, "loss_yns_3": 0.15954, "loss_cls_4": 0.97394, "loss_box_4": 1.64844, "loss_cns_4": 0.65565, "loss_yns_4": 0.16134, "loss_cls_5": 0.99123, "loss_box_5": 1.65727, "loss_cns_5": 0.6532, "loss_yns_5": 0.16127, "loss_cls_dn_0": 0.21554, "loss_box_dn_0": 0.76609, "loss_cls_dn_1": 0.13742, "loss_box_dn_1": 0.7342, "loss_cls_dn_2": 0.13756, "loss_box_dn_2": 0.72053, "loss_cls_dn_3": 0.13601, "loss_box_dn_3": 0.73684, "loss_cls_dn_4": 0.14007, "loss_box_dn_4": 0.74613, "loss_cls_dn_5": 0.15181, "loss_box_dn_5": 0.76653, "loss_dense_depth": 0.82028, "loss": 26.8233, "grad_norm": 33.77744, "time": 1.52844} -{"mode": "train", "epoch": 1, "iter": 171, "lr": 0.00017, "memory": 49163, "data_time": 0.07594, "loss_cls_0": 0.87523, "loss_box_0": 1.70437, "loss_cns_0": 0.61207, "loss_yns_0": 0.15778, "loss_cls_1": 0.94053, "loss_box_1": 1.71735, "loss_cns_1": 0.6493, "loss_yns_1": 0.15931, "loss_cls_2": 0.96215, "loss_box_2": 1.67261, "loss_cns_2": 0.65384, "loss_yns_2": 0.15808, "loss_cls_3": 0.96495, "loss_box_3": 1.66624, "loss_cns_3": 0.65533, "loss_yns_3": 0.15813, "loss_cls_4": 0.97878, "loss_box_4": 1.67493, "loss_cns_4": 0.65784, "loss_yns_4": 0.16114, "loss_cls_5": 0.97409, "loss_box_5": 1.67286, "loss_cns_5": 0.65494, "loss_yns_5": 0.15866, "loss_cls_dn_0": 0.21715, "loss_box_dn_0": 0.76474, "loss_cls_dn_1": 0.13933, "loss_box_dn_1": 0.76315, "loss_cls_dn_2": 0.1415, "loss_box_dn_2": 0.75797, "loss_cls_dn_3": 0.14258, "loss_box_dn_3": 0.77684, "loss_cls_dn_4": 0.14553, "loss_box_dn_4": 0.7995, "loss_cls_dn_5": 0.14878, "loss_box_dn_5": 0.82264, "loss_dense_depth": 0.77897, "loss": 27.0392, "grad_norm": 40.28996, "time": 1.52626} -{"mode": "train", "epoch": 1, "iter": 172, "lr": 0.00017, "memory": 49163, "data_time": 0.07355, "loss_cls_0": 0.87479, "loss_box_0": 1.73753, "loss_cns_0": 0.61135, "loss_yns_0": 0.1578, "loss_cls_1": 0.94019, "loss_box_1": 1.75361, "loss_cns_1": 0.64892, "loss_yns_1": 0.15895, "loss_cls_2": 0.9565, "loss_box_2": 1.7394, "loss_cns_2": 0.64952, "loss_yns_2": 0.15949, "loss_cls_3": 0.9664, "loss_box_3": 1.7279, "loss_cns_3": 0.65336, "loss_yns_3": 0.15925, "loss_cls_4": 0.98429, "loss_box_4": 1.72586, "loss_cns_4": 0.65453, "loss_yns_4": 0.16048, "loss_cls_5": 0.97324, "loss_box_5": 1.73234, "loss_cns_5": 0.65215, "loss_yns_5": 0.16012, "loss_cls_dn_0": 0.2211, "loss_box_dn_0": 0.76354, "loss_cls_dn_1": 0.13897, "loss_box_dn_1": 0.78287, "loss_cls_dn_2": 0.13572, "loss_box_dn_2": 0.78815, "loss_cls_dn_3": 0.14182, "loss_box_dn_3": 0.79633, "loss_cls_dn_4": 0.14724, "loss_box_dn_4": 0.81183, "loss_cls_dn_5": 0.15096, "loss_box_dn_5": 0.83451, "loss_dense_depth": 0.82214, "loss": 27.47315, "grad_norm": 44.774, "time": 1.51916} -{"mode": "train", "epoch": 1, "iter": 173, "lr": 0.00017, "memory": 49163, "data_time": 0.07328, "loss_cls_0": 0.8385, "loss_box_0": 1.72274, "loss_cns_0": 0.62159, "loss_yns_0": 0.15905, "loss_cls_1": 0.92855, "loss_box_1": 1.71781, "loss_cns_1": 0.65088, "loss_yns_1": 0.15698, "loss_cls_2": 0.95792, "loss_box_2": 1.71232, "loss_cns_2": 0.65201, "loss_yns_2": 0.15857, "loss_cls_3": 0.96722, "loss_box_3": 1.67574, "loss_cns_3": 0.65789, "loss_yns_3": 0.15831, "loss_cls_4": 0.96897, "loss_box_4": 1.67541, "loss_cns_4": 0.65725, "loss_yns_4": 0.15813, "loss_cls_5": 0.96814, "loss_box_5": 1.68208, "loss_cns_5": 0.65624, "loss_yns_5": 0.15807, "loss_cls_dn_0": 0.21844, "loss_box_dn_0": 0.75724, "loss_cls_dn_1": 0.13893, "loss_box_dn_1": 0.78439, "loss_cls_dn_2": 0.13682, "loss_box_dn_2": 0.78824, "loss_cls_dn_3": 0.14183, "loss_box_dn_3": 0.77144, "loss_cls_dn_4": 0.14216, "loss_box_dn_4": 0.78176, "loss_cls_dn_5": 0.14773, "loss_box_dn_5": 0.79761, "loss_dense_depth": 0.78621, "loss": 27.05316, "grad_norm": 36.02763, "time": 1.52657} -{"mode": "train", "epoch": 1, "iter": 174, "lr": 0.00017, "memory": 49163, "data_time": 0.07141, "loss_cls_0": 0.87365, "loss_box_0": 1.77165, "loss_cns_0": 0.61504, "loss_yns_0": 0.15881, "loss_cls_1": 0.93098, "loss_box_1": 1.74314, "loss_cns_1": 0.642, "loss_yns_1": 0.15685, "loss_cls_2": 0.96289, "loss_box_2": 1.72023, "loss_cns_2": 0.64619, "loss_yns_2": 0.15489, "loss_cls_3": 0.96644, "loss_box_3": 1.71316, "loss_cns_3": 0.65352, "loss_yns_3": 0.15643, "loss_cls_4": 0.9785, "loss_box_4": 1.71742, "loss_cns_4": 0.6557, "loss_yns_4": 0.15783, "loss_cls_5": 0.98228, "loss_box_5": 1.69513, "loss_cns_5": 0.65276, "loss_yns_5": 0.15559, "loss_cls_dn_0": 0.21529, "loss_box_dn_0": 0.76761, "loss_cls_dn_1": 0.13816, "loss_box_dn_1": 0.76258, "loss_cls_dn_2": 0.14131, "loss_box_dn_2": 0.75688, "loss_cls_dn_3": 0.14201, "loss_box_dn_3": 0.74168, "loss_cls_dn_4": 0.14601, "loss_box_dn_4": 0.74935, "loss_cls_dn_5": 0.16141, "loss_box_dn_5": 0.7468, "loss_dense_depth": 0.86955, "loss": 27.19975, "grad_norm": 48.68022, "time": 1.52857} -{"mode": "train", "epoch": 1, "iter": 175, "lr": 0.00017, "memory": 49163, "data_time": 0.08064, "loss_cls_0": 0.85746, "loss_box_0": 1.74624, "loss_cns_0": 0.62071, "loss_yns_0": 0.15843, "loss_cls_1": 0.92928, "loss_box_1": 1.73418, "loss_cns_1": 0.6471, "loss_yns_1": 0.15693, "loss_cls_2": 0.95279, "loss_box_2": 1.69305, "loss_cns_2": 0.6502, "loss_yns_2": 0.15691, "loss_cls_3": 0.96146, "loss_box_3": 1.69489, "loss_cns_3": 0.65504, "loss_yns_3": 0.15844, "loss_cls_4": 0.97892, "loss_box_4": 1.67771, "loss_cns_4": 0.65713, "loss_yns_4": 0.15985, "loss_cls_5": 0.96953, "loss_box_5": 1.67467, "loss_cns_5": 0.65559, "loss_yns_5": 0.15621, "loss_cls_dn_0": 0.21773, "loss_box_dn_0": 0.7596, "loss_cls_dn_1": 0.13557, "loss_box_dn_1": 0.74304, "loss_cls_dn_2": 0.13874, "loss_box_dn_2": 0.73104, "loss_cls_dn_3": 0.13901, "loss_box_dn_3": 0.72486, "loss_cls_dn_4": 0.14306, "loss_box_dn_4": 0.72253, "loss_cls_dn_5": 0.14464, "loss_box_dn_5": 0.72445, "loss_dense_depth": 0.78179, "loss": 26.80879, "grad_norm": 34.96129, "time": 1.57459} -{"mode": "train", "epoch": 1, "iter": 176, "lr": 0.00017, "memory": 49163, "data_time": 0.07611, "loss_cls_0": 0.84437, "loss_box_0": 1.71294, "loss_cns_0": 0.61981, "loss_yns_0": 0.15466, "loss_cls_1": 0.9326, "loss_box_1": 1.68531, "loss_cns_1": 0.6509, "loss_yns_1": 0.15205, "loss_cls_2": 0.9491, "loss_box_2": 1.66768, "loss_cns_2": 0.65533, "loss_yns_2": 0.15123, "loss_cls_3": 0.97784, "loss_box_3": 1.66811, "loss_cns_3": 0.65961, "loss_yns_3": 0.15187, "loss_cls_4": 0.98354, "loss_box_4": 1.67305, "loss_cns_4": 0.66211, "loss_yns_4": 0.15278, "loss_cls_5": 0.97212, "loss_box_5": 1.66216, "loss_cns_5": 0.65816, "loss_yns_5": 0.15251, "loss_cls_dn_0": 0.21938, "loss_box_dn_0": 0.76976, "loss_cls_dn_1": 0.13591, "loss_box_dn_1": 0.73103, "loss_cls_dn_2": 0.13553, "loss_box_dn_2": 0.72235, "loss_cls_dn_3": 0.13898, "loss_box_dn_3": 0.73251, "loss_cls_dn_4": 0.14259, "loss_box_dn_4": 0.73997, "loss_cls_dn_5": 0.15616, "loss_box_dn_5": 0.74919, "loss_dense_depth": 0.85907, "loss": 26.78226, "grad_norm": 43.97967, "time": 1.51408} -{"mode": "train", "epoch": 1, "iter": 177, "lr": 0.00017, "memory": 49163, "data_time": 0.07238, "loss_cls_0": 0.83966, "loss_box_0": 1.71282, "loss_cns_0": 0.62019, "loss_yns_0": 0.15112, "loss_cls_1": 0.91009, "loss_box_1": 1.66568, "loss_cns_1": 0.64897, "loss_yns_1": 0.14958, "loss_cls_2": 0.94311, "loss_box_2": 1.65372, "loss_cns_2": 0.65533, "loss_yns_2": 0.15002, "loss_cls_3": 0.97222, "loss_box_3": 1.67084, "loss_cns_3": 0.66075, "loss_yns_3": 0.15032, "loss_cls_4": 0.97232, "loss_box_4": 1.67438, "loss_cns_4": 0.66099, "loss_yns_4": 0.15, "loss_cls_5": 0.96311, "loss_box_5": 1.66233, "loss_cns_5": 0.65707, "loss_yns_5": 0.14867, "loss_cls_dn_0": 0.21686, "loss_box_dn_0": 0.7657, "loss_cls_dn_1": 0.13804, "loss_box_dn_1": 0.74524, "loss_cls_dn_2": 0.13702, "loss_box_dn_2": 0.74253, "loss_cls_dn_3": 0.14191, "loss_box_dn_3": 0.76415, "loss_cls_dn_4": 0.14539, "loss_box_dn_4": 0.77797, "loss_cls_dn_5": 0.15815, "loss_box_dn_5": 0.79584, "loss_dense_depth": 0.82002, "loss": 26.79208, "grad_norm": 50.20523, "time": 1.4998} -{"mode": "train", "epoch": 1, "iter": 178, "lr": 0.00017, "memory": 49163, "data_time": 0.07084, "loss_cls_0": 0.82407, "loss_box_0": 1.74944, "loss_cns_0": 0.61735, "loss_yns_0": 0.15092, "loss_cls_1": 0.90424, "loss_box_1": 1.71666, "loss_cns_1": 0.64526, "loss_yns_1": 0.14905, "loss_cls_2": 0.9378, "loss_box_2": 1.68115, "loss_cns_2": 0.65211, "loss_yns_2": 0.14931, "loss_cls_3": 0.9376, "loss_box_3": 1.66552, "loss_cns_3": 0.65315, "loss_yns_3": 0.14965, "loss_cls_4": 0.94879, "loss_box_4": 1.66586, "loss_cns_4": 0.65289, "loss_yns_4": 0.1503, "loss_cls_5": 0.94746, "loss_box_5": 1.67853, "loss_cns_5": 0.65035, "loss_yns_5": 0.15053, "loss_cls_dn_0": 0.21721, "loss_box_dn_0": 0.77404, "loss_cls_dn_1": 0.1415, "loss_box_dn_1": 0.77618, "loss_cls_dn_2": 0.14136, "loss_box_dn_2": 0.77323, "loss_cls_dn_3": 0.14332, "loss_box_dn_3": 0.7944, "loss_cls_dn_4": 0.14926, "loss_box_dn_4": 0.80794, "loss_cls_dn_5": 0.15394, "loss_box_dn_5": 0.83941, "loss_dense_depth": 0.80922, "loss": 26.949, "grad_norm": 41.50588, "time": 1.5161} -{"mode": "train", "epoch": 1, "iter": 179, "lr": 0.00017, "memory": 49163, "data_time": 0.08168, "loss_cls_0": 0.82252, "loss_box_0": 1.75728, "loss_cns_0": 0.61918, "loss_yns_0": 0.15149, "loss_cls_1": 0.90785, "loss_box_1": 1.7417, "loss_cns_1": 0.64793, "loss_yns_1": 0.15118, "loss_cls_2": 0.92914, "loss_box_2": 1.7298, "loss_cns_2": 0.65037, "loss_yns_2": 0.15146, "loss_cls_3": 0.93665, "loss_box_3": 1.70186, "loss_cns_3": 0.6541, "loss_yns_3": 0.15146, "loss_cls_4": 0.94609, "loss_box_4": 1.68691, "loss_cns_4": 0.65456, "loss_yns_4": 0.15166, "loss_cls_5": 0.94518, "loss_box_5": 1.68109, "loss_cns_5": 0.65351, "loss_yns_5": 0.15273, "loss_cls_dn_0": 0.21819, "loss_box_dn_0": 0.78099, "loss_cls_dn_1": 0.13911, "loss_box_dn_1": 0.79759, "loss_cls_dn_2": 0.14218, "loss_box_dn_2": 0.79855, "loss_cls_dn_3": 0.14912, "loss_box_dn_3": 0.80161, "loss_cls_dn_4": 0.15462, "loss_box_dn_4": 0.80323, "loss_cls_dn_5": 0.15587, "loss_box_dn_5": 0.82295, "loss_dense_depth": 0.81037, "loss": 27.15007, "grad_norm": 49.18174, "time": 1.53406} -{"mode": "train", "epoch": 1, "iter": 180, "lr": 0.00017, "memory": 49163, "data_time": 0.08153, "loss_cls_0": 0.81716, "loss_box_0": 1.69913, "loss_cns_0": 0.62146, "loss_yns_0": 0.1487, "loss_cls_1": 0.90096, "loss_box_1": 1.70567, "loss_cns_1": 0.64685, "loss_yns_1": 0.14804, "loss_cls_2": 0.92311, "loss_box_2": 1.69893, "loss_cns_2": 0.65632, "loss_yns_2": 0.14996, "loss_cls_3": 0.94226, "loss_box_3": 1.65724, "loss_cns_3": 0.65983, "loss_yns_3": 0.14997, "loss_cls_4": 0.93952, "loss_box_4": 1.64207, "loss_cns_4": 0.65832, "loss_yns_4": 0.14997, "loss_cls_5": 0.93931, "loss_box_5": 1.63135, "loss_cns_5": 0.66038, "loss_yns_5": 0.15206, "loss_cls_dn_0": 0.21838, "loss_box_dn_0": 0.77222, "loss_cls_dn_1": 0.13812, "loss_box_dn_1": 0.78526, "loss_cls_dn_2": 0.14117, "loss_box_dn_2": 0.78345, "loss_cls_dn_3": 0.15114, "loss_box_dn_3": 0.76649, "loss_cls_dn_4": 0.14968, "loss_box_dn_4": 0.75908, "loss_cls_dn_5": 0.15233, "loss_box_dn_5": 0.76424, "loss_dense_depth": 0.79767, "loss": 26.67783, "grad_norm": 44.53192, "time": 1.51567} -{"mode": "train", "epoch": 1, "iter": 181, "lr": 0.00017, "memory": 49163, "data_time": 0.14885, "loss_cls_0": 0.81292, "loss_box_0": 1.71997, "loss_cns_0": 0.61849, "loss_yns_0": 0.14938, "loss_cls_1": 0.90041, "loss_box_1": 1.73663, "loss_cns_1": 0.64359, "loss_yns_1": 0.15002, "loss_cls_2": 0.92923, "loss_box_2": 1.68968, "loss_cns_2": 0.65466, "loss_yns_2": 0.15109, "loss_cls_3": 0.92832, "loss_box_3": 1.65927, "loss_cns_3": 0.65547, "loss_yns_3": 0.15073, "loss_cls_4": 0.93228, "loss_box_4": 1.64378, "loss_cns_4": 0.65699, "loss_yns_4": 0.15151, "loss_cls_5": 0.93716, "loss_box_5": 1.63552, "loss_cns_5": 0.65626, "loss_yns_5": 0.15092, "loss_cls_dn_0": 0.21949, "loss_box_dn_0": 0.77261, "loss_cls_dn_1": 0.13933, "loss_box_dn_1": 0.73741, "loss_cls_dn_2": 0.14037, "loss_box_dn_2": 0.72026, "loss_cls_dn_3": 0.14449, "loss_box_dn_3": 0.70281, "loss_cls_dn_4": 0.14397, "loss_box_dn_4": 0.7032, "loss_cls_dn_5": 0.14866, "loss_box_dn_5": 0.70455, "loss_dense_depth": 0.81293, "loss": 26.40437, "grad_norm": 34.16962, "time": 1.58584} -{"mode": "train", "epoch": 1, "iter": 182, "lr": 0.00017, "memory": 49163, "data_time": 0.07855, "loss_cls_0": 0.80084, "loss_box_0": 1.7038, "loss_cns_0": 0.62528, "loss_yns_0": 0.14891, "loss_cls_1": 0.88125, "loss_box_1": 1.73176, "loss_cns_1": 0.64651, "loss_yns_1": 0.14797, "loss_cls_2": 0.91498, "loss_box_2": 1.66888, "loss_cns_2": 0.65502, "loss_yns_2": 0.1507, "loss_cls_3": 0.92767, "loss_box_3": 1.67983, "loss_cns_3": 0.65605, "loss_yns_3": 0.14787, "loss_cls_4": 0.94189, "loss_box_4": 1.66961, "loss_cns_4": 0.65711, "loss_yns_4": 0.14874, "loss_cls_5": 0.94632, "loss_box_5": 1.65319, "loss_cns_5": 0.6543, "loss_yns_5": 0.14758, "loss_cls_dn_0": 0.21447, "loss_box_dn_0": 0.76987, "loss_cls_dn_1": 0.13708, "loss_box_dn_1": 0.72667, "loss_cls_dn_2": 0.13841, "loss_box_dn_2": 0.70365, "loss_cls_dn_3": 0.14141, "loss_box_dn_3": 0.7069, "loss_cls_dn_4": 0.14355, "loss_box_dn_4": 0.71205, "loss_cls_dn_5": 0.14641, "loss_box_dn_5": 0.71208, "loss_dense_depth": 0.78918, "loss": 26.34779, "grad_norm": 47.58435, "time": 1.52981} -{"mode": "train", "epoch": 1, "iter": 183, "lr": 0.00017, "memory": 49163, "data_time": 0.07609, "loss_cls_0": 0.79294, "loss_box_0": 1.67154, "loss_cns_0": 0.6246, "loss_yns_0": 0.14676, "loss_cls_1": 0.88564, "loss_box_1": 1.68008, "loss_cns_1": 0.65203, "loss_yns_1": 0.14824, "loss_cls_2": 0.92004, "loss_box_2": 1.622, "loss_cns_2": 0.6579, "loss_yns_2": 0.14954, "loss_cls_3": 0.93498, "loss_box_3": 1.64629, "loss_cns_3": 0.66059, "loss_yns_3": 0.14779, "loss_cls_4": 0.94989, "loss_box_4": 1.63223, "loss_cns_4": 0.66017, "loss_yns_4": 0.15026, "loss_cls_5": 0.93226, "loss_box_5": 1.62591, "loss_cns_5": 0.65853, "loss_yns_5": 0.1488, "loss_cls_dn_0": 0.21112, "loss_box_dn_0": 0.75655, "loss_cls_dn_1": 0.1316, "loss_box_dn_1": 0.7429, "loss_cls_dn_2": 0.13529, "loss_box_dn_2": 0.7229, "loss_cls_dn_3": 0.13797, "loss_box_dn_3": 0.73864, "loss_cls_dn_4": 0.14198, "loss_box_dn_4": 0.74757, "loss_cls_dn_5": 0.15379, "loss_box_dn_5": 0.75339, "loss_dense_depth": 0.75918, "loss": 26.23189, "grad_norm": 54.88355, "time": 1.57924} -{"mode": "train", "epoch": 1, "iter": 184, "lr": 0.00017, "memory": 49163, "data_time": 0.09922, "loss_cls_0": 0.8112, "loss_box_0": 1.69429, "loss_cns_0": 0.62141, "loss_yns_0": 0.14408, "loss_cls_1": 0.88515, "loss_box_1": 1.70856, "loss_cns_1": 0.65207, "loss_yns_1": 0.1455, "loss_cls_2": 0.91036, "loss_box_2": 1.66991, "loss_cns_2": 0.65839, "loss_yns_2": 0.14547, "loss_cls_3": 0.92667, "loss_box_3": 1.67786, "loss_cns_3": 0.66094, "loss_yns_3": 0.14652, "loss_cls_4": 0.93812, "loss_box_4": 1.66734, "loss_cns_4": 0.66184, "loss_yns_4": 0.14735, "loss_cls_5": 0.93586, "loss_box_5": 1.66202, "loss_cns_5": 0.66206, "loss_yns_5": 0.14868, "loss_cls_dn_0": 0.21606, "loss_box_dn_0": 0.76092, "loss_cls_dn_1": 0.13673, "loss_box_dn_1": 0.77258, "loss_cls_dn_2": 0.13975, "loss_box_dn_2": 0.76305, "loss_cls_dn_3": 0.14272, "loss_box_dn_3": 0.7818, "loss_cls_dn_4": 0.14545, "loss_box_dn_4": 0.79128, "loss_cls_dn_5": 0.15564, "loss_box_dn_5": 0.80698, "loss_dense_depth": 0.81143, "loss": 26.70604, "grad_norm": 48.72953, "time": 1.5685} -{"mode": "train", "epoch": 1, "iter": 185, "lr": 0.00017, "memory": 49163, "data_time": 0.07708, "loss_cls_0": 0.81252, "loss_box_0": 1.70271, "loss_cns_0": 0.61788, "loss_yns_0": 0.14739, "loss_cls_1": 0.8882, "loss_box_1": 1.70849, "loss_cns_1": 0.64347, "loss_yns_1": 0.1472, "loss_cls_2": 0.93878, "loss_box_2": 1.66555, "loss_cns_2": 0.65138, "loss_yns_2": 0.14911, "loss_cls_3": 0.93074, "loss_box_3": 1.66784, "loss_cns_3": 0.65064, "loss_yns_3": 0.14807, "loss_cls_4": 0.93431, "loss_box_4": 1.67827, "loss_cns_4": 0.65459, "loss_yns_4": 0.14793, "loss_cls_5": 0.93386, "loss_box_5": 1.6584, "loss_cns_5": 0.6541, "loss_yns_5": 0.14737, "loss_cls_dn_0": 0.21709, "loss_box_dn_0": 0.76493, "loss_cls_dn_1": 0.14228, "loss_box_dn_1": 0.8169, "loss_cls_dn_2": 0.14671, "loss_box_dn_2": 0.7995, "loss_cls_dn_3": 0.14819, "loss_box_dn_3": 0.81127, "loss_cls_dn_4": 0.14917, "loss_box_dn_4": 0.82569, "loss_cls_dn_5": 0.15613, "loss_box_dn_5": 0.83227, "loss_dense_depth": 0.80784, "loss": 26.89676, "grad_norm": 45.95214, "time": 1.52159} -{"mode": "train", "epoch": 1, "iter": 186, "lr": 0.00017, "memory": 49163, "data_time": 0.07696, "loss_cls_0": 0.8304, "loss_box_0": 1.66221, "loss_cns_0": 0.61534, "loss_yns_0": 0.14836, "loss_cls_1": 0.91358, "loss_box_1": 1.69199, "loss_cns_1": 0.65153, "loss_yns_1": 0.14982, "loss_cls_2": 0.94821, "loss_box_2": 1.64589, "loss_cns_2": 0.65767, "loss_yns_2": 0.15345, "loss_cls_3": 0.96139, "loss_box_3": 1.65153, "loss_cns_3": 0.65476, "loss_yns_3": 0.15002, "loss_cls_4": 0.96637, "loss_box_4": 1.66072, "loss_cns_4": 0.66201, "loss_yns_4": 0.15192, "loss_cls_5": 0.9557, "loss_box_5": 1.62763, "loss_cns_5": 0.661, "loss_yns_5": 0.1522, "loss_cls_dn_0": 0.21186, "loss_box_dn_0": 0.76546, "loss_cls_dn_1": 0.14407, "loss_box_dn_1": 0.81033, "loss_cls_dn_2": 0.1455, "loss_box_dn_2": 0.78921, "loss_cls_dn_3": 0.14905, "loss_box_dn_3": 0.7992, "loss_cls_dn_4": 0.15846, "loss_box_dn_4": 0.80819, "loss_cls_dn_5": 0.16274, "loss_box_dn_5": 0.80674, "loss_dense_depth": 0.76456, "loss": 26.83909, "grad_norm": 51.82679, "time": 1.53342} -{"mode": "train", "epoch": 1, "iter": 187, "lr": 0.00017, "memory": 49163, "data_time": 0.07739, "loss_cls_0": 0.83027, "loss_box_0": 1.66455, "loss_cns_0": 0.62009, "loss_yns_0": 0.14778, "loss_cls_1": 0.92349, "loss_box_1": 1.70618, "loss_cns_1": 0.65317, "loss_yns_1": 0.15209, "loss_cls_2": 0.97031, "loss_box_2": 1.65649, "loss_cns_2": 0.66092, "loss_yns_2": 0.15732, "loss_cls_3": 0.96546, "loss_box_3": 1.64362, "loss_cns_3": 0.65956, "loss_yns_3": 0.15485, "loss_cls_4": 0.97099, "loss_box_4": 1.65509, "loss_cns_4": 0.66337, "loss_yns_4": 0.15439, "loss_cls_5": 0.95513, "loss_box_5": 1.65102, "loss_cns_5": 0.6596, "loss_yns_5": 0.15587, "loss_cls_dn_0": 0.21448, "loss_box_dn_0": 0.76219, "loss_cls_dn_1": 0.13903, "loss_box_dn_1": 0.78399, "loss_cls_dn_2": 0.14094, "loss_box_dn_2": 0.76588, "loss_cls_dn_3": 0.14422, "loss_box_dn_3": 0.76265, "loss_cls_dn_4": 0.15458, "loss_box_dn_4": 0.77059, "loss_cls_dn_5": 0.16113, "loss_box_dn_5": 0.77898, "loss_dense_depth": 0.8093, "loss": 26.81955, "grad_norm": 46.33662, "time": 1.52259} -{"mode": "train", "epoch": 1, "iter": 188, "lr": 0.00017, "memory": 49163, "data_time": 0.07461, "loss_cls_0": 0.82431, "loss_box_0": 1.69386, "loss_cns_0": 0.62057, "loss_yns_0": 0.14762, "loss_cls_1": 0.9089, "loss_box_1": 1.74282, "loss_cns_1": 0.64197, "loss_yns_1": 0.14961, "loss_cls_2": 0.95962, "loss_box_2": 1.66593, "loss_cns_2": 0.65479, "loss_yns_2": 0.14983, "loss_cls_3": 0.95315, "loss_box_3": 1.63327, "loss_cns_3": 0.65332, "loss_yns_3": 0.14924, "loss_cls_4": 0.9778, "loss_box_4": 1.63694, "loss_cns_4": 0.65641, "loss_yns_4": 0.1501, "loss_cls_5": 0.96367, "loss_box_5": 1.65783, "loss_cns_5": 0.65254, "loss_yns_5": 0.14922, "loss_cls_dn_0": 0.21181, "loss_box_dn_0": 0.76476, "loss_cls_dn_1": 0.14453, "loss_box_dn_1": 0.74541, "loss_cls_dn_2": 0.14694, "loss_box_dn_2": 0.71795, "loss_cls_dn_3": 0.14695, "loss_box_dn_3": 0.71012, "loss_cls_dn_4": 0.15256, "loss_box_dn_4": 0.7129, "loss_cls_dn_5": 0.157, "loss_box_dn_5": 0.72735, "loss_dense_depth": 0.78251, "loss": 26.51412, "grad_norm": 40.97542, "time": 1.51456} -{"mode": "train", "epoch": 1, "iter": 189, "lr": 0.00018, "memory": 49163, "data_time": 0.07557, "loss_cls_0": 0.82728, "loss_box_0": 1.69662, "loss_cns_0": 0.62828, "loss_yns_0": 0.15188, "loss_cls_1": 0.90727, "loss_box_1": 1.69399, "loss_cns_1": 0.65, "loss_yns_1": 0.15245, "loss_cls_2": 0.96176, "loss_box_2": 1.66796, "loss_cns_2": 0.65643, "loss_yns_2": 0.15198, "loss_cls_3": 0.95785, "loss_box_3": 1.68591, "loss_cns_3": 0.65426, "loss_yns_3": 0.15203, "loss_cls_4": 0.97175, "loss_box_4": 1.67129, "loss_cns_4": 0.65741, "loss_yns_4": 0.15774, "loss_cls_5": 0.95354, "loss_box_5": 1.68053, "loss_cns_5": 0.65416, "loss_yns_5": 0.15085, "loss_cls_dn_0": 0.21256, "loss_box_dn_0": 0.77427, "loss_cls_dn_1": 0.14471, "loss_box_dn_1": 0.73284, "loss_cls_dn_2": 0.15393, "loss_box_dn_2": 0.72897, "loss_cls_dn_3": 0.15008, "loss_box_dn_3": 0.74471, "loss_cls_dn_4": 0.15457, "loss_box_dn_4": 0.74244, "loss_cls_dn_5": 0.16237, "loss_box_dn_5": 0.75291, "loss_dense_depth": 0.82626, "loss": 26.77387, "grad_norm": 58.1075, "time": 1.51485} -{"mode": "train", "epoch": 1, "iter": 190, "lr": 0.00018, "memory": 49163, "data_time": 0.07668, "loss_cls_0": 0.84062, "loss_box_0": 1.72334, "loss_cns_0": 0.62344, "loss_yns_0": 0.15392, "loss_cls_1": 0.91658, "loss_box_1": 1.71794, "loss_cns_1": 0.64868, "loss_yns_1": 0.15473, "loss_cls_2": 0.97549, "loss_box_2": 1.69965, "loss_cns_2": 0.65449, "loss_yns_2": 0.15711, "loss_cls_3": 0.97189, "loss_box_3": 1.73326, "loss_cns_3": 0.65324, "loss_yns_3": 0.15585, "loss_cls_4": 0.98629, "loss_box_4": 1.72916, "loss_cns_4": 0.65565, "loss_yns_4": 0.15785, "loss_cls_5": 0.97009, "loss_box_5": 1.71461, "loss_cns_5": 0.65524, "loss_yns_5": 0.15624, "loss_cls_dn_0": 0.21843, "loss_box_dn_0": 0.77075, "loss_cls_dn_1": 0.14504, "loss_box_dn_1": 0.75299, "loss_cls_dn_2": 0.15884, "loss_box_dn_2": 0.76574, "loss_cls_dn_3": 0.15528, "loss_box_dn_3": 0.7966, "loss_cls_dn_4": 0.16595, "loss_box_dn_4": 0.79901, "loss_cls_dn_5": 0.17393, "loss_box_dn_5": 0.81457, "loss_dense_depth": 0.81718, "loss": 27.33965, "grad_norm": 68.6053, "time": 1.50931} -{"mode": "train", "epoch": 1, "iter": 191, "lr": 0.00018, "memory": 49163, "data_time": 0.07432, "loss_cls_0": 0.82776, "loss_box_0": 1.73397, "loss_cns_0": 0.61716, "loss_yns_0": 0.15203, "loss_cls_1": 0.90581, "loss_box_1": 1.72396, "loss_cns_1": 0.64968, "loss_yns_1": 0.15247, "loss_cls_2": 0.95327, "loss_box_2": 1.68881, "loss_cns_2": 0.65484, "loss_yns_2": 0.15604, "loss_cls_3": 0.9633, "loss_box_3": 1.72119, "loss_cns_3": 0.65286, "loss_yns_3": 0.1556, "loss_cls_4": 0.96975, "loss_box_4": 1.72075, "loss_cns_4": 0.65394, "loss_yns_4": 0.15635, "loss_cls_5": 0.96265, "loss_box_5": 1.7104, "loss_cns_5": 0.6541, "loss_yns_5": 0.15773, "loss_cls_dn_0": 0.21691, "loss_box_dn_0": 0.7604, "loss_cls_dn_1": 0.14228, "loss_box_dn_1": 0.77345, "loss_cls_dn_2": 0.14954, "loss_box_dn_2": 0.77048, "loss_cls_dn_3": 0.14832, "loss_box_dn_3": 0.79865, "loss_cls_dn_4": 0.15434, "loss_box_dn_4": 0.80601, "loss_cls_dn_5": 0.15874, "loss_box_dn_5": 0.82254, "loss_dense_depth": 0.79728, "loss": 27.19335, "grad_norm": 48.22474, "time": 1.52994} -{"mode": "train", "epoch": 1, "iter": 192, "lr": 0.00018, "memory": 49163, "data_time": 0.0752, "loss_cls_0": 0.82505, "loss_box_0": 1.72945, "loss_cns_0": 0.61869, "loss_yns_0": 0.15321, "loss_cls_1": 0.91007, "loss_box_1": 1.71881, "loss_cns_1": 0.6509, "loss_yns_1": 0.1533, "loss_cls_2": 0.95133, "loss_box_2": 1.68082, "loss_cns_2": 0.65467, "loss_yns_2": 0.15382, "loss_cls_3": 0.95567, "loss_box_3": 1.69528, "loss_cns_3": 0.65429, "loss_yns_3": 0.15654, "loss_cls_4": 0.95532, "loss_box_4": 1.7066, "loss_cns_4": 0.65668, "loss_yns_4": 0.1554, "loss_cls_5": 0.96513, "loss_box_5": 1.68572, "loss_cns_5": 0.65429, "loss_yns_5": 0.15607, "loss_cls_dn_0": 0.21711, "loss_box_dn_0": 0.76045, "loss_cls_dn_1": 0.1402, "loss_box_dn_1": 0.78856, "loss_cls_dn_2": 0.14393, "loss_box_dn_2": 0.77768, "loss_cls_dn_3": 0.1444, "loss_box_dn_3": 0.78964, "loss_cls_dn_4": 0.14972, "loss_box_dn_4": 0.80029, "loss_cls_dn_5": 0.15503, "loss_box_dn_5": 0.80489, "loss_dense_depth": 0.7712, "loss": 27.04021, "grad_norm": 42.80266, "time": 1.52643} -{"mode": "train", "epoch": 1, "iter": 193, "lr": 0.00018, "memory": 49163, "data_time": 0.07201, "loss_cls_0": 0.82951, "loss_box_0": 1.72544, "loss_cns_0": 0.62171, "loss_yns_0": 0.15142, "loss_cls_1": 0.91221, "loss_box_1": 1.74372, "loss_cns_1": 0.64807, "loss_yns_1": 0.15648, "loss_cls_2": 0.95179, "loss_box_2": 1.66811, "loss_cns_2": 0.65728, "loss_yns_2": 0.15422, "loss_cls_3": 0.96387, "loss_box_3": 1.64384, "loss_cns_3": 0.65578, "loss_yns_3": 0.15331, "loss_cls_4": 0.97698, "loss_box_4": 1.6674, "loss_cns_4": 0.65925, "loss_yns_4": 0.15752, "loss_cls_5": 0.97909, "loss_box_5": 1.66264, "loss_cns_5": 0.65468, "loss_yns_5": 0.15495, "loss_cls_dn_0": 0.21512, "loss_box_dn_0": 0.76833, "loss_cls_dn_1": 0.13651, "loss_box_dn_1": 0.77274, "loss_cls_dn_2": 0.14245, "loss_box_dn_2": 0.75985, "loss_cls_dn_3": 0.14319, "loss_box_dn_3": 0.75421, "loss_cls_dn_4": 0.14636, "loss_box_dn_4": 0.76478, "loss_cls_dn_5": 0.15525, "loss_box_dn_5": 0.77042, "loss_dense_depth": 0.77686, "loss": 26.85535, "grad_norm": 48.56566, "time": 1.50767} -{"mode": "train", "epoch": 1, "iter": 194, "lr": 0.00018, "memory": 49163, "data_time": 0.07364, "loss_cls_0": 0.84219, "loss_box_0": 1.72889, "loss_cns_0": 0.61982, "loss_yns_0": 0.15106, "loss_cls_1": 0.91083, "loss_box_1": 1.79369, "loss_cns_1": 0.63778, "loss_yns_1": 0.15491, "loss_cls_2": 0.95156, "loss_box_2": 1.71743, "loss_cns_2": 0.64862, "loss_yns_2": 0.15444, "loss_cls_3": 0.9748, "loss_box_3": 1.69668, "loss_cns_3": 0.64705, "loss_yns_3": 0.15301, "loss_cls_4": 0.99985, "loss_box_4": 1.71682, "loss_cns_4": 0.65378, "loss_yns_4": 0.15588, "loss_cls_5": 0.97627, "loss_box_5": 1.72145, "loss_cns_5": 0.65429, "loss_yns_5": 0.1545, "loss_cls_dn_0": 0.21495, "loss_box_dn_0": 0.76223, "loss_cls_dn_1": 0.13657, "loss_box_dn_1": 0.7755, "loss_cls_dn_2": 0.14163, "loss_box_dn_2": 0.75723, "loss_cls_dn_3": 0.14362, "loss_box_dn_3": 0.75305, "loss_cls_dn_4": 0.14823, "loss_box_dn_4": 0.76715, "loss_cls_dn_5": 0.15511, "loss_box_dn_5": 0.76875, "loss_dense_depth": 0.79207, "loss": 27.13169, "grad_norm": 41.91784, "time": 1.51156} -{"mode": "train", "epoch": 1, "iter": 195, "lr": 0.00018, "memory": 49163, "data_time": 0.07428, "loss_cls_0": 0.81298, "loss_box_0": 1.714, "loss_cns_0": 0.63112, "loss_yns_0": 0.14834, "loss_cls_1": 0.87348, "loss_box_1": 1.68496, "loss_cns_1": 0.64596, "loss_yns_1": 0.1467, "loss_cls_2": 0.90757, "loss_box_2": 1.6493, "loss_cns_2": 0.64925, "loss_yns_2": 0.14946, "loss_cls_3": 0.91098, "loss_box_3": 1.65033, "loss_cns_3": 0.6498, "loss_yns_3": 0.15119, "loss_cls_4": 0.91603, "loss_box_4": 1.65261, "loss_cns_4": 0.65836, "loss_yns_4": 0.15052, "loss_cls_5": 0.91994, "loss_box_5": 1.62927, "loss_cns_5": 0.66059, "loss_yns_5": 0.15029, "loss_cls_dn_0": 0.20095, "loss_box_dn_0": 0.7634, "loss_cls_dn_1": 0.13015, "loss_box_dn_1": 0.74905, "loss_cls_dn_2": 0.1348, "loss_box_dn_2": 0.73609, "loss_cls_dn_3": 0.13408, "loss_box_dn_3": 0.7395, "loss_cls_dn_4": 0.13965, "loss_box_dn_4": 0.74798, "loss_cls_dn_5": 0.14886, "loss_box_dn_5": 0.7455, "loss_dense_depth": 0.76345, "loss": 26.24651, "grad_norm": 51.12922, "time": 1.56286} -{"mode": "train", "epoch": 1, "iter": 196, "lr": 0.00018, "memory": 49163, "data_time": 0.07372, "loss_cls_0": 0.80848, "loss_box_0": 1.73328, "loss_cns_0": 0.62815, "loss_yns_0": 0.14921, "loss_cls_1": 0.86842, "loss_box_1": 1.70089, "loss_cns_1": 0.65054, "loss_yns_1": 0.14878, "loss_cls_2": 0.90251, "loss_box_2": 1.64043, "loss_cns_2": 0.66023, "loss_yns_2": 0.14973, "loss_cls_3": 0.91751, "loss_box_3": 1.63818, "loss_cns_3": 0.65796, "loss_yns_3": 0.1507, "loss_cls_4": 0.92769, "loss_box_4": 1.64221, "loss_cns_4": 0.66097, "loss_yns_4": 0.15225, "loss_cls_5": 0.9454, "loss_box_5": 1.62779, "loss_cns_5": 0.65957, "loss_yns_5": 0.15144, "loss_cls_dn_0": 0.2025, "loss_box_dn_0": 0.77245, "loss_cls_dn_1": 0.1308, "loss_box_dn_1": 0.74974, "loss_cls_dn_2": 0.13034, "loss_box_dn_2": 0.7354, "loss_cls_dn_3": 0.13369, "loss_box_dn_3": 0.73764, "loss_cls_dn_4": 0.13679, "loss_box_dn_4": 0.7509, "loss_cls_dn_5": 0.15251, "loss_box_dn_5": 0.75617, "loss_dense_depth": 0.78704, "loss": 26.34831, "grad_norm": 33.71003, "time": 1.52686} -{"mode": "train", "epoch": 1, "iter": 197, "lr": 0.00018, "memory": 49163, "data_time": 0.07242, "loss_cls_0": 0.81747, "loss_box_0": 1.77663, "loss_cns_0": 0.62405, "loss_yns_0": 0.14978, "loss_cls_1": 0.89555, "loss_box_1": 1.75537, "loss_cns_1": 0.65225, "loss_yns_1": 0.15147, "loss_cls_2": 0.92313, "loss_box_2": 1.72161, "loss_cns_2": 0.65432, "loss_yns_2": 0.15022, "loss_cls_3": 0.91948, "loss_box_3": 1.72365, "loss_cns_3": 0.65509, "loss_yns_3": 0.15054, "loss_cls_4": 0.92734, "loss_box_4": 1.76394, "loss_cns_4": 0.65851, "loss_yns_4": 0.15508, "loss_cls_5": 0.92898, "loss_box_5": 1.76778, "loss_cns_5": 0.65401, "loss_yns_5": 0.15104, "loss_cls_dn_0": 0.21248, "loss_box_dn_0": 0.76052, "loss_cls_dn_1": 0.129, "loss_box_dn_1": 0.75511, "loss_cls_dn_2": 0.12856, "loss_box_dn_2": 0.75385, "loss_cls_dn_3": 0.13403, "loss_box_dn_3": 0.76227, "loss_cls_dn_4": 0.13778, "loss_box_dn_4": 0.79929, "loss_cls_dn_5": 0.15717, "loss_box_dn_5": 0.81573, "loss_dense_depth": 0.83528, "loss": 27.10837, "grad_norm": 53.51642, "time": 1.51348} -{"mode": "train", "epoch": 1, "iter": 198, "lr": 0.00018, "memory": 49163, "data_time": 0.07194, "loss_cls_0": 0.83818, "loss_box_0": 1.76468, "loss_cns_0": 0.62068, "loss_yns_0": 0.15461, "loss_cls_1": 0.89714, "loss_box_1": 1.71461, "loss_cns_1": 0.64663, "loss_yns_1": 0.15271, "loss_cls_2": 0.91491, "loss_box_2": 1.69121, "loss_cns_2": 0.64814, "loss_yns_2": 0.15357, "loss_cls_3": 0.93348, "loss_box_3": 1.6948, "loss_cns_3": 0.65341, "loss_yns_3": 0.15562, "loss_cls_4": 0.93595, "loss_box_4": 1.70286, "loss_cns_4": 0.65962, "loss_yns_4": 0.15718, "loss_cls_5": 0.92385, "loss_box_5": 1.70021, "loss_cns_5": 0.66264, "loss_yns_5": 0.16046, "loss_cls_dn_0": 0.21834, "loss_box_dn_0": 0.76591, "loss_cls_dn_1": 0.13398, "loss_box_dn_1": 0.78861, "loss_cls_dn_2": 0.1339, "loss_box_dn_2": 0.78752, "loss_cls_dn_3": 0.13787, "loss_box_dn_3": 0.79962, "loss_cls_dn_4": 0.14175, "loss_box_dn_4": 0.81945, "loss_cls_dn_5": 0.15445, "loss_box_dn_5": 0.83509, "loss_dense_depth": 0.83521, "loss": 27.08886, "grad_norm": 35.74313, "time": 1.52424} -{"mode": "train", "epoch": 1, "iter": 199, "lr": 0.00018, "memory": 49163, "data_time": 0.08227, "loss_cls_0": 0.8268, "loss_box_0": 1.74409, "loss_cns_0": 0.61661, "loss_yns_0": 0.15751, "loss_cls_1": 0.87789, "loss_box_1": 1.73542, "loss_cns_1": 0.64952, "loss_yns_1": 0.15791, "loss_cls_2": 0.89876, "loss_box_2": 1.70979, "loss_cns_2": 0.65568, "loss_yns_2": 0.15809, "loss_cls_3": 0.91187, "loss_box_3": 1.68949, "loss_cns_3": 0.65795, "loss_yns_3": 0.16317, "loss_cls_4": 0.92848, "loss_box_4": 1.71007, "loss_cns_4": 0.65861, "loss_yns_4": 0.15922, "loss_cls_5": 0.92636, "loss_box_5": 1.73526, "loss_cns_5": 0.65291, "loss_yns_5": 0.16678, "loss_cls_dn_0": 0.20745, "loss_box_dn_0": 0.75378, "loss_cls_dn_1": 0.13164, "loss_box_dn_1": 0.79474, "loss_cls_dn_2": 0.13145, "loss_box_dn_2": 0.78525, "loss_cls_dn_3": 0.13648, "loss_box_dn_3": 0.78834, "loss_cls_dn_4": 0.14722, "loss_box_dn_4": 0.79306, "loss_cls_dn_5": 0.15105, "loss_box_dn_5": 0.81236, "loss_dense_depth": 0.86527, "loss": 27.04635, "grad_norm": 51.99848, "time": 1.55194} -{"mode": "train", "epoch": 1, "iter": 200, "lr": 0.00018, "memory": 49163, "data_time": 0.0809, "loss_cls_0": 0.82349, "loss_box_0": 1.74184, "loss_cns_0": 0.60991, "loss_yns_0": 0.15585, "loss_cls_1": 0.89559, "loss_box_1": 1.77606, "loss_cns_1": 0.64655, "loss_yns_1": 0.15623, "loss_cls_2": 0.90453, "loss_box_2": 1.72165, "loss_cns_2": 0.6554, "loss_yns_2": 0.15503, "loss_cls_3": 0.91429, "loss_box_3": 1.68908, "loss_cns_3": 0.65414, "loss_yns_3": 0.15838, "loss_cls_4": 0.93807, "loss_box_4": 1.69262, "loss_cns_4": 0.6574, "loss_yns_4": 0.15248, "loss_cls_5": 0.92316, "loss_box_5": 1.69872, "loss_cns_5": 0.65282, "loss_yns_5": 0.15462, "loss_cls_dn_0": 0.20855, "loss_box_dn_0": 0.76001, "loss_cls_dn_1": 0.13086, "loss_box_dn_1": 0.78994, "loss_cls_dn_2": 0.12954, "loss_box_dn_2": 0.76689, "loss_cls_dn_3": 0.13461, "loss_box_dn_3": 0.761, "loss_cls_dn_4": 0.14421, "loss_box_dn_4": 0.76115, "loss_cls_dn_5": 0.14459, "loss_box_dn_5": 0.76762, "loss_dense_depth": 0.84117, "loss": 26.86803, "grad_norm": 34.42434, "time": 1.53296} -{"mode": "train", "epoch": 1, "iter": 201, "lr": 0.00018, "memory": 49163, "data_time": 0.14625, "loss_cls_0": 0.81178, "loss_box_0": 1.74036, "loss_cns_0": 0.61257, "loss_yns_0": 0.15467, "loss_cls_1": 0.90038, "loss_box_1": 1.75709, "loss_cns_1": 0.6459, "loss_yns_1": 0.15582, "loss_cls_2": 0.91151, "loss_box_2": 1.75062, "loss_cns_2": 0.64714, "loss_yns_2": 0.15423, "loss_cls_3": 0.91671, "loss_box_3": 1.74758, "loss_cns_3": 0.64529, "loss_yns_3": 0.15593, "loss_cls_4": 0.92932, "loss_box_4": 1.74689, "loss_cns_4": 0.65186, "loss_yns_4": 0.15682, "loss_cls_5": 0.92435, "loss_box_5": 1.72583, "loss_cns_5": 0.6487, "loss_yns_5": 0.15628, "loss_cls_dn_0": 0.20783, "loss_box_dn_0": 0.76662, "loss_cls_dn_1": 0.12465, "loss_box_dn_1": 0.76685, "loss_cls_dn_2": 0.12458, "loss_box_dn_2": 0.75439, "loss_cls_dn_3": 0.1279, "loss_box_dn_3": 0.75332, "loss_cls_dn_4": 0.13378, "loss_box_dn_4": 0.76367, "loss_cls_dn_5": 0.14221, "loss_box_dn_5": 0.76348, "loss_dense_depth": 0.84418, "loss": 26.92106, "grad_norm": 53.40235, "time": 1.59994} -{"mode": "train", "epoch": 1, "iter": 202, "lr": 0.00018, "memory": 49163, "data_time": 0.07306, "loss_cls_0": 0.79795, "loss_box_0": 1.7002, "loss_cns_0": 0.61932, "loss_yns_0": 0.15515, "loss_cls_1": 0.90264, "loss_box_1": 1.72228, "loss_cns_1": 0.65049, "loss_yns_1": 0.15574, "loss_cls_2": 0.90685, "loss_box_2": 1.70375, "loss_cns_2": 0.64954, "loss_yns_2": 0.15647, "loss_cls_3": 0.91884, "loss_box_3": 1.6837, "loss_cns_3": 0.65289, "loss_yns_3": 0.15502, "loss_cls_4": 0.93404, "loss_box_4": 1.6939, "loss_cns_4": 0.65177, "loss_yns_4": 0.16133, "loss_cls_5": 0.92747, "loss_box_5": 1.69353, "loss_cns_5": 0.65005, "loss_yns_5": 0.1565, "loss_cls_dn_0": 0.20447, "loss_box_dn_0": 0.7675, "loss_cls_dn_1": 0.1297, "loss_box_dn_1": 0.73539, "loss_cls_dn_2": 0.12893, "loss_box_dn_2": 0.73021, "loss_cls_dn_3": 0.13329, "loss_box_dn_3": 0.73015, "loss_cls_dn_4": 0.13698, "loss_box_dn_4": 0.74357, "loss_cls_dn_5": 0.14141, "loss_box_dn_5": 0.75742, "loss_dense_depth": 0.9178, "loss": 26.65625, "grad_norm": 47.5516, "time": 1.54194} -{"mode": "train", "epoch": 1, "iter": 203, "lr": 0.00018, "memory": 49163, "data_time": 0.07287, "loss_cls_0": 0.80528, "loss_box_0": 1.71095, "loss_cns_0": 0.61382, "loss_yns_0": 0.15496, "loss_cls_1": 0.90419, "loss_box_1": 1.76776, "loss_cns_1": 0.64918, "loss_yns_1": 0.15396, "loss_cls_2": 0.92083, "loss_box_2": 1.74584, "loss_cns_2": 0.6494, "loss_yns_2": 0.15291, "loss_cls_3": 0.91987, "loss_box_3": 1.72729, "loss_cns_3": 0.65002, "loss_yns_3": 0.1566, "loss_cls_4": 0.9346, "loss_box_4": 1.73676, "loss_cns_4": 0.64945, "loss_yns_4": 0.15451, "loss_cls_5": 0.9187, "loss_box_5": 1.73481, "loss_cns_5": 0.64903, "loss_yns_5": 0.15666, "loss_cls_dn_0": 0.20321, "loss_box_dn_0": 0.76886, "loss_cls_dn_1": 0.12736, "loss_box_dn_1": 0.74909, "loss_cls_dn_2": 0.1296, "loss_box_dn_2": 0.75617, "loss_cls_dn_3": 0.13554, "loss_box_dn_3": 0.75732, "loss_cls_dn_4": 0.13856, "loss_box_dn_4": 0.76141, "loss_cls_dn_5": 0.14383, "loss_box_dn_5": 0.77361, "loss_dense_depth": 0.83151, "loss": 26.89345, "grad_norm": 54.85564, "time": 1.59381} -{"mode": "train", "epoch": 1, "iter": 204, "lr": 0.00018, "memory": 49163, "data_time": 0.09907, "loss_cls_0": 0.80399, "loss_box_0": 1.72062, "loss_cns_0": 0.62016, "loss_yns_0": 0.15432, "loss_cls_1": 0.88873, "loss_box_1": 1.76056, "loss_cns_1": 0.65083, "loss_yns_1": 0.15324, "loss_cls_2": 0.90733, "loss_box_2": 1.7494, "loss_cns_2": 0.65596, "loss_yns_2": 0.1522, "loss_cls_3": 0.90507, "loss_box_3": 1.73212, "loss_cns_3": 0.65285, "loss_yns_3": 0.16081, "loss_cls_4": 0.91807, "loss_box_4": 1.74261, "loss_cns_4": 0.65666, "loss_yns_4": 0.15207, "loss_cls_5": 0.90539, "loss_box_5": 1.72394, "loss_cns_5": 0.65549, "loss_yns_5": 0.15576, "loss_cls_dn_0": 0.1996, "loss_box_dn_0": 0.75704, "loss_cls_dn_1": 0.12229, "loss_box_dn_1": 0.73976, "loss_cls_dn_2": 0.12454, "loss_box_dn_2": 0.75383, "loss_cls_dn_3": 0.13285, "loss_box_dn_3": 0.76613, "loss_cls_dn_4": 0.13434, "loss_box_dn_4": 0.77569, "loss_cls_dn_5": 0.14151, "loss_box_dn_5": 0.78559, "loss_dense_depth": 0.91714, "loss": 26.92846, "grad_norm": 55.59443, "time": 1.5716} -{"mode": "train", "epoch": 1, "iter": 205, "lr": 0.00018, "memory": 49163, "data_time": 0.07371, "loss_cls_0": 0.80866, "loss_box_0": 1.73203, "loss_cns_0": 0.62119, "loss_yns_0": 0.15373, "loss_cls_1": 0.88698, "loss_box_1": 1.73328, "loss_cns_1": 0.64932, "loss_yns_1": 0.15282, "loss_cls_2": 0.91002, "loss_box_2": 1.7077, "loss_cns_2": 0.65554, "loss_yns_2": 0.15204, "loss_cls_3": 0.91196, "loss_box_3": 1.69191, "loss_cns_3": 0.65318, "loss_yns_3": 0.15725, "loss_cls_4": 0.91602, "loss_box_4": 1.72086, "loss_cns_4": 0.65418, "loss_yns_4": 0.15336, "loss_cls_5": 0.92932, "loss_box_5": 1.72801, "loss_cns_5": 0.65219, "loss_yns_5": 0.15355, "loss_cls_dn_0": 0.20635, "loss_box_dn_0": 0.76163, "loss_cls_dn_1": 0.1326, "loss_box_dn_1": 0.74936, "loss_cls_dn_2": 0.13154, "loss_box_dn_2": 0.75067, "loss_cls_dn_3": 0.14182, "loss_box_dn_3": 0.76171, "loss_cls_dn_4": 0.14482, "loss_box_dn_4": 0.77558, "loss_cls_dn_5": 0.14931, "loss_box_dn_5": 0.79775, "loss_dense_depth": 0.85475, "loss": 26.84299, "grad_norm": 42.76947, "time": 1.55474} -{"mode": "train", "epoch": 1, "iter": 206, "lr": 0.00018, "memory": 49163, "data_time": 0.07396, "loss_cls_0": 0.80397, "loss_box_0": 1.72631, "loss_cns_0": 0.62393, "loss_yns_0": 0.15263, "loss_cls_1": 0.89964, "loss_box_1": 1.77845, "loss_cns_1": 0.6519, "loss_yns_1": 0.15345, "loss_cls_2": 0.91575, "loss_box_2": 1.73079, "loss_cns_2": 0.65611, "loss_yns_2": 0.15343, "loss_cls_3": 0.9153, "loss_box_3": 1.72418, "loss_cns_3": 0.6568, "loss_yns_3": 0.15316, "loss_cls_4": 0.92326, "loss_box_4": 1.71108, "loss_cns_4": 0.65838, "loss_yns_4": 0.15448, "loss_cls_5": 0.91884, "loss_box_5": 1.72546, "loss_cns_5": 0.65704, "loss_yns_5": 0.15248, "loss_cls_dn_0": 0.21191, "loss_box_dn_0": 0.76367, "loss_cls_dn_1": 0.12999, "loss_box_dn_1": 0.75369, "loss_cls_dn_2": 0.13257, "loss_box_dn_2": 0.74394, "loss_cls_dn_3": 0.13594, "loss_box_dn_3": 0.7515, "loss_cls_dn_4": 0.14593, "loss_box_dn_4": 0.74101, "loss_cls_dn_5": 0.14523, "loss_box_dn_5": 0.759, "loss_dense_depth": 0.91554, "loss": 26.92676, "grad_norm": 45.09544, "time": 1.54149} -{"mode": "train", "epoch": 1, "iter": 207, "lr": 0.00018, "memory": 49163, "data_time": 0.07559, "loss_cls_0": 0.8072, "loss_box_0": 1.72505, "loss_cns_0": 0.61625, "loss_yns_0": 0.15282, "loss_cls_1": 0.89359, "loss_box_1": 1.75708, "loss_cns_1": 0.65003, "loss_yns_1": 0.1519, "loss_cls_2": 0.90403, "loss_box_2": 1.72098, "loss_cns_2": 0.65595, "loss_yns_2": 0.15254, "loss_cls_3": 0.91735, "loss_box_3": 1.7153, "loss_cns_3": 0.65534, "loss_yns_3": 0.1534, "loss_cls_4": 0.9287, "loss_box_4": 1.68113, "loss_cns_4": 0.65635, "loss_yns_4": 0.15293, "loss_cls_5": 0.94697, "loss_box_5": 1.67409, "loss_cns_5": 0.65554, "loss_yns_5": 0.15343, "loss_cls_dn_0": 0.2104, "loss_box_dn_0": 0.75057, "loss_cls_dn_1": 0.13595, "loss_box_dn_1": 0.73095, "loss_cls_dn_2": 0.13673, "loss_box_dn_2": 0.72436, "loss_cls_dn_3": 0.13905, "loss_box_dn_3": 0.72916, "loss_cls_dn_4": 0.1497, "loss_box_dn_4": 0.70669, "loss_cls_dn_5": 0.1481, "loss_box_dn_5": 0.711, "loss_dense_depth": 0.82125, "loss": 26.57183, "grad_norm": 40.168, "time": 1.54743} -{"mode": "train", "epoch": 1, "iter": 208, "lr": 0.00018, "memory": 49163, "data_time": 0.0771, "loss_cls_0": 0.81367, "loss_box_0": 1.72171, "loss_cns_0": 0.61698, "loss_yns_0": 0.15687, "loss_cls_1": 0.89355, "loss_box_1": 1.75609, "loss_cns_1": 0.64543, "loss_yns_1": 0.15624, "loss_cls_2": 0.90693, "loss_box_2": 1.72065, "loss_cns_2": 0.6511, "loss_yns_2": 0.15451, "loss_cls_3": 0.91491, "loss_box_3": 1.70743, "loss_cns_3": 0.65409, "loss_yns_3": 0.15472, "loss_cls_4": 0.9356, "loss_box_4": 1.71522, "loss_cns_4": 0.64865, "loss_yns_4": 0.154, "loss_cls_5": 0.94266, "loss_box_5": 1.71831, "loss_cns_5": 0.6475, "loss_yns_5": 0.15871, "loss_cls_dn_0": 0.20584, "loss_box_dn_0": 0.75987, "loss_cls_dn_1": 0.13135, "loss_box_dn_1": 0.69473, "loss_cls_dn_2": 0.13204, "loss_box_dn_2": 0.68078, "loss_cls_dn_3": 0.13417, "loss_box_dn_3": 0.68019, "loss_cls_dn_4": 0.13749, "loss_box_dn_4": 0.68618, "loss_cls_dn_5": 0.14026, "loss_box_dn_5": 0.69174, "loss_dense_depth": 0.83764, "loss": 26.4578, "grad_norm": 35.25033, "time": 1.53623} -{"mode": "train", "epoch": 1, "iter": 209, "lr": 0.00018, "memory": 49163, "data_time": 0.07904, "loss_cls_0": 0.82653, "loss_box_0": 1.76236, "loss_cns_0": 0.61751, "loss_yns_0": 0.15202, "loss_cls_1": 0.89338, "loss_box_1": 1.73756, "loss_cns_1": 0.64154, "loss_yns_1": 0.152, "loss_cls_2": 0.91522, "loss_box_2": 1.69086, "loss_cns_2": 0.64757, "loss_yns_2": 0.15234, "loss_cls_3": 0.91726, "loss_box_3": 1.68648, "loss_cns_3": 0.65029, "loss_yns_3": 0.1529, "loss_cls_4": 0.93934, "loss_box_4": 1.73387, "loss_cns_4": 0.65174, "loss_yns_4": 0.15234, "loss_cls_5": 0.93396, "loss_box_5": 1.68312, "loss_cns_5": 0.65425, "loss_yns_5": 0.1579, "loss_cls_dn_0": 0.20757, "loss_box_dn_0": 0.7694, "loss_cls_dn_1": 0.12834, "loss_box_dn_1": 0.69884, "loss_cls_dn_2": 0.12855, "loss_box_dn_2": 0.67613, "loss_cls_dn_3": 0.12979, "loss_box_dn_3": 0.67927, "loss_cls_dn_4": 0.13345, "loss_box_dn_4": 0.70468, "loss_cls_dn_5": 0.1421, "loss_box_dn_5": 0.69324, "loss_dense_depth": 0.86961, "loss": 26.4633, "grad_norm": 44.8647, "time": 1.53252} -{"mode": "train", "epoch": 1, "iter": 210, "lr": 0.00018, "memory": 49163, "data_time": 0.07817, "loss_cls_0": 0.826, "loss_box_0": 1.73163, "loss_cns_0": 0.62536, "loss_yns_0": 0.15222, "loss_cls_1": 0.87668, "loss_box_1": 1.70835, "loss_cns_1": 0.64347, "loss_yns_1": 0.15088, "loss_cls_2": 0.90181, "loss_box_2": 1.65871, "loss_cns_2": 0.65387, "loss_yns_2": 0.15344, "loss_cls_3": 0.91873, "loss_box_3": 1.65134, "loss_cns_3": 0.65399, "loss_yns_3": 0.1531, "loss_cls_4": 0.92257, "loss_box_4": 1.67533, "loss_cns_4": 0.65469, "loss_yns_4": 0.15208, "loss_cls_5": 0.92336, "loss_box_5": 1.64397, "loss_cns_5": 0.65324, "loss_yns_5": 0.1534, "loss_cls_dn_0": 0.19856, "loss_box_dn_0": 0.75704, "loss_cls_dn_1": 0.12466, "loss_box_dn_1": 0.69844, "loss_cls_dn_2": 0.12429, "loss_box_dn_2": 0.68778, "loss_cls_dn_3": 0.12533, "loss_box_dn_3": 0.69562, "loss_cls_dn_4": 0.13254, "loss_box_dn_4": 0.71547, "loss_cls_dn_5": 0.13776, "loss_box_dn_5": 0.71355, "loss_dense_depth": 0.79032, "loss": 26.13956, "grad_norm": 43.90113, "time": 1.52473} -{"mode": "train", "epoch": 1, "iter": 211, "lr": 0.00018, "memory": 49163, "data_time": 0.07604, "loss_cls_0": 0.79558, "loss_box_0": 1.70707, "loss_cns_0": 0.62355, "loss_yns_0": 0.1526, "loss_cls_1": 0.89024, "loss_box_1": 1.66036, "loss_cns_1": 0.65013, "loss_yns_1": 0.15148, "loss_cls_2": 0.90509, "loss_box_2": 1.63052, "loss_cns_2": 0.65595, "loss_yns_2": 0.15255, "loss_cls_3": 0.90518, "loss_box_3": 1.62298, "loss_cns_3": 0.6618, "loss_yns_3": 0.15153, "loss_cls_4": 0.92122, "loss_box_4": 1.63579, "loss_cns_4": 0.65614, "loss_yns_4": 0.14988, "loss_cls_5": 0.91054, "loss_box_5": 1.64013, "loss_cns_5": 0.66294, "loss_yns_5": 0.15491, "loss_cls_dn_0": 0.19335, "loss_box_dn_0": 0.7584, "loss_cls_dn_1": 0.11864, "loss_box_dn_1": 0.70273, "loss_cls_dn_2": 0.11844, "loss_box_dn_2": 0.70136, "loss_cls_dn_3": 0.12163, "loss_box_dn_3": 0.7085, "loss_cls_dn_4": 0.12408, "loss_box_dn_4": 0.72845, "loss_cls_dn_5": 0.13419, "loss_box_dn_5": 0.74693, "loss_dense_depth": 0.81253, "loss": 26.01739, "grad_norm": 46.62672, "time": 1.54313} -{"mode": "train", "epoch": 1, "iter": 212, "lr": 0.00018, "memory": 49163, "data_time": 0.0781, "loss_cls_0": 0.7935, "loss_box_0": 1.69557, "loss_cns_0": 0.62579, "loss_yns_0": 0.149, "loss_cls_1": 0.87023, "loss_box_1": 1.68531, "loss_cns_1": 0.65285, "loss_yns_1": 0.14967, "loss_cls_2": 0.9016, "loss_box_2": 1.62857, "loss_cns_2": 0.65364, "loss_yns_2": 0.14726, "loss_cls_3": 0.89654, "loss_box_3": 1.63526, "loss_cns_3": 0.66412, "loss_yns_3": 0.14921, "loss_cls_4": 0.91559, "loss_box_4": 1.64796, "loss_cns_4": 0.65681, "loss_yns_4": 0.148, "loss_cls_5": 0.9065, "loss_box_5": 1.6479, "loss_cns_5": 0.6654, "loss_yns_5": 0.15134, "loss_cls_dn_0": 0.19867, "loss_box_dn_0": 0.76037, "loss_cls_dn_1": 0.12649, "loss_box_dn_1": 0.72907, "loss_cls_dn_2": 0.12547, "loss_box_dn_2": 0.7221, "loss_cls_dn_3": 0.12613, "loss_box_dn_3": 0.72576, "loss_cls_dn_4": 0.12481, "loss_box_dn_4": 0.74627, "loss_cls_dn_5": 0.13349, "loss_box_dn_5": 0.75952, "loss_dense_depth": 0.81034, "loss": 26.12609, "grad_norm": 43.43552, "time": 1.52856} -{"mode": "train", "epoch": 1, "iter": 213, "lr": 0.00018, "memory": 49163, "data_time": 0.0824, "loss_cls_0": 0.79463, "loss_box_0": 1.72336, "loss_cns_0": 0.6257, "loss_yns_0": 0.15247, "loss_cls_1": 0.86307, "loss_box_1": 1.66585, "loss_cns_1": 0.65225, "loss_yns_1": 0.15025, "loss_cls_2": 0.88409, "loss_box_2": 1.61728, "loss_cns_2": 0.65311, "loss_yns_2": 0.15117, "loss_cls_3": 0.8954, "loss_box_3": 1.63086, "loss_cns_3": 0.65607, "loss_yns_3": 0.14966, "loss_cls_4": 0.90809, "loss_box_4": 1.61865, "loss_cns_4": 0.655, "loss_yns_4": 0.15168, "loss_cls_5": 0.90713, "loss_box_5": 1.64146, "loss_cns_5": 0.65703, "loss_yns_5": 0.15129, "loss_cls_dn_0": 0.20487, "loss_box_dn_0": 0.75552, "loss_cls_dn_1": 0.12957, "loss_box_dn_1": 0.73581, "loss_cls_dn_2": 0.1291, "loss_box_dn_2": 0.73057, "loss_cls_dn_3": 0.12919, "loss_box_dn_3": 0.73923, "loss_cls_dn_4": 0.13121, "loss_box_dn_4": 0.74307, "loss_cls_dn_5": 0.13986, "loss_box_dn_5": 0.75433, "loss_dense_depth": 0.83543, "loss": 26.11327, "grad_norm": 49.28038, "time": 1.53249} -{"mode": "train", "epoch": 1, "iter": 214, "lr": 0.00019, "memory": 49163, "data_time": 0.07722, "loss_cls_0": 0.82043, "loss_box_0": 1.7254, "loss_cns_0": 0.62419, "loss_yns_0": 0.15232, "loss_cls_1": 0.89173, "loss_box_1": 1.67528, "loss_cns_1": 0.65006, "loss_yns_1": 0.15206, "loss_cls_2": 0.91796, "loss_box_2": 1.64411, "loss_cns_2": 0.65354, "loss_yns_2": 0.15194, "loss_cls_3": 0.93954, "loss_box_3": 1.65024, "loss_cns_3": 0.65548, "loss_yns_3": 0.15396, "loss_cls_4": 0.93043, "loss_box_4": 1.61488, "loss_cns_4": 0.65563, "loss_yns_4": 0.15521, "loss_cls_5": 0.92464, "loss_box_5": 1.62738, "loss_cns_5": 0.65458, "loss_yns_5": 0.15229, "loss_cls_dn_0": 0.20395, "loss_box_dn_0": 0.76458, "loss_cls_dn_1": 0.12591, "loss_box_dn_1": 0.72705, "loss_cls_dn_2": 0.12414, "loss_box_dn_2": 0.72092, "loss_cls_dn_3": 0.12509, "loss_box_dn_3": 0.72584, "loss_cls_dn_4": 0.13162, "loss_box_dn_4": 0.71583, "loss_cls_dn_5": 0.13519, "loss_box_dn_5": 0.71991, "loss_dense_depth": 0.80715, "loss": 26.20049, "grad_norm": 49.47481, "time": 1.52983} -{"mode": "train", "epoch": 1, "iter": 215, "lr": 0.00019, "memory": 49163, "data_time": 0.07754, "loss_cls_0": 0.82139, "loss_box_0": 1.72198, "loss_cns_0": 0.6249, "loss_yns_0": 0.15203, "loss_cls_1": 0.88385, "loss_box_1": 1.66594, "loss_cns_1": 0.6465, "loss_yns_1": 0.1509, "loss_cls_2": 0.90547, "loss_box_2": 1.60398, "loss_cns_2": 0.65389, "loss_yns_2": 0.15253, "loss_cls_3": 0.91512, "loss_box_3": 1.59431, "loss_cns_3": 0.66114, "loss_yns_3": 0.15085, "loss_cls_4": 0.93833, "loss_box_4": 1.6022, "loss_cns_4": 0.65498, "loss_yns_4": 0.15022, "loss_cls_5": 0.95945, "loss_box_5": 1.6099, "loss_cns_5": 0.65681, "loss_yns_5": 0.1503, "loss_cls_dn_0": 0.20628, "loss_box_dn_0": 0.75535, "loss_cls_dn_1": 0.12647, "loss_box_dn_1": 0.70055, "loss_cls_dn_2": 0.12622, "loss_box_dn_2": 0.68539, "loss_cls_dn_3": 0.12888, "loss_box_dn_3": 0.68215, "loss_cls_dn_4": 0.13505, "loss_box_dn_4": 0.68454, "loss_cls_dn_5": 0.14785, "loss_box_dn_5": 0.69237, "loss_dense_depth": 0.84279, "loss": 25.94086, "grad_norm": 35.14315, "time": 1.57976} -{"mode": "train", "epoch": 1, "iter": 216, "lr": 0.00019, "memory": 49163, "data_time": 0.07792, "loss_cls_0": 0.8364, "loss_box_0": 1.73538, "loss_cns_0": 0.62758, "loss_yns_0": 0.15523, "loss_cls_1": 0.88354, "loss_box_1": 1.69373, "loss_cns_1": 0.64749, "loss_yns_1": 0.15244, "loss_cls_2": 0.92077, "loss_box_2": 1.63865, "loss_cns_2": 0.65914, "loss_yns_2": 0.15441, "loss_cls_3": 0.91635, "loss_box_3": 1.6149, "loss_cns_3": 0.66513, "loss_yns_3": 0.15336, "loss_cls_4": 0.92832, "loss_box_4": 1.62038, "loss_cns_4": 0.65763, "loss_yns_4": 0.15302, "loss_cls_5": 0.92295, "loss_box_5": 1.63388, "loss_cns_5": 0.65546, "loss_yns_5": 0.15314, "loss_cls_dn_0": 0.21089, "loss_box_dn_0": 0.76392, "loss_cls_dn_1": 0.12866, "loss_box_dn_1": 0.70095, "loss_cls_dn_2": 0.13814, "loss_box_dn_2": 0.68834, "loss_cls_dn_3": 0.141, "loss_box_dn_3": 0.68471, "loss_cls_dn_4": 0.13665, "loss_box_dn_4": 0.68896, "loss_cls_dn_5": 0.14099, "loss_box_dn_5": 0.70191, "loss_dense_depth": 0.80231, "loss": 26.10669, "grad_norm": 37.51883, "time": 1.5079} -{"mode": "train", "epoch": 1, "iter": 217, "lr": 0.00019, "memory": 49163, "data_time": 0.07238, "loss_cls_0": 0.79957, "loss_box_0": 1.72757, "loss_cns_0": 0.62369, "loss_yns_0": 0.15556, "loss_cls_1": 0.87184, "loss_box_1": 1.65314, "loss_cns_1": 0.65001, "loss_yns_1": 0.15535, "loss_cls_2": 0.89809, "loss_box_2": 1.62163, "loss_cns_2": 0.65317, "loss_yns_2": 0.1549, "loss_cls_3": 0.90088, "loss_box_3": 1.60449, "loss_cns_3": 0.65586, "loss_yns_3": 0.15561, "loss_cls_4": 0.90505, "loss_box_4": 1.59963, "loss_cns_4": 0.65786, "loss_yns_4": 0.15516, "loss_cls_5": 0.90702, "loss_box_5": 1.60297, "loss_cns_5": 0.65702, "loss_yns_5": 0.15796, "loss_cls_dn_0": 0.20236, "loss_box_dn_0": 0.76035, "loss_cls_dn_1": 0.12253, "loss_box_dn_1": 0.70134, "loss_cls_dn_2": 0.12517, "loss_box_dn_2": 0.70085, "loss_cls_dn_3": 0.1288, "loss_box_dn_3": 0.70075, "loss_cls_dn_4": 0.12901, "loss_box_dn_4": 0.70715, "loss_cls_dn_5": 0.1326, "loss_box_dn_5": 0.71622, "loss_dense_depth": 0.77244, "loss": 25.82358, "grad_norm": 36.54551, "time": 1.54994} -{"mode": "train", "epoch": 1, "iter": 218, "lr": 0.00019, "memory": 49163, "data_time": 0.07656, "loss_cls_0": 0.78718, "loss_box_0": 1.69891, "loss_cns_0": 0.62634, "loss_yns_0": 0.15277, "loss_cls_1": 0.86892, "loss_box_1": 1.65778, "loss_cns_1": 0.65623, "loss_yns_1": 0.15457, "loss_cls_2": 0.91062, "loss_box_2": 1.62844, "loss_cns_2": 0.65694, "loss_yns_2": 0.15663, "loss_cls_3": 0.8906, "loss_box_3": 1.61188, "loss_cns_3": 0.65907, "loss_yns_3": 0.15476, "loss_cls_4": 0.90206, "loss_box_4": 1.61804, "loss_cns_4": 0.65765, "loss_yns_4": 0.15412, "loss_cls_5": 0.9075, "loss_box_5": 1.61346, "loss_cns_5": 0.65665, "loss_yns_5": 0.15707, "loss_cls_dn_0": 0.196, "loss_box_dn_0": 0.76453, "loss_cls_dn_1": 0.12577, "loss_box_dn_1": 0.70562, "loss_cls_dn_2": 0.12864, "loss_box_dn_2": 0.7016, "loss_cls_dn_3": 0.12963, "loss_box_dn_3": 0.70048, "loss_cls_dn_4": 0.13055, "loss_box_dn_4": 0.71103, "loss_cls_dn_5": 0.1342, "loss_box_dn_5": 0.71986, "loss_dense_depth": 0.75173, "loss": 25.83784, "grad_norm": 33.55844, "time": 1.5318} -{"mode": "train", "epoch": 1, "iter": 219, "lr": 0.00019, "memory": 49163, "data_time": 0.08015, "loss_cls_0": 0.80906, "loss_box_0": 1.73158, "loss_cns_0": 0.61837, "loss_yns_0": 0.15241, "loss_cls_1": 0.87198, "loss_box_1": 1.70386, "loss_cns_1": 0.64632, "loss_yns_1": 0.15185, "loss_cls_2": 0.90602, "loss_box_2": 1.66359, "loss_cns_2": 0.65115, "loss_yns_2": 0.15286, "loss_cls_3": 0.90779, "loss_box_3": 1.64745, "loss_cns_3": 0.65421, "loss_yns_3": 0.15213, "loss_cls_4": 0.91445, "loss_box_4": 1.62735, "loss_cns_4": 0.65372, "loss_yns_4": 0.15197, "loss_cls_5": 0.914, "loss_box_5": 1.63029, "loss_cns_5": 0.65453, "loss_yns_5": 0.15475, "loss_cls_dn_0": 0.20557, "loss_box_dn_0": 0.75752, "loss_cls_dn_1": 0.1276, "loss_box_dn_1": 0.70397, "loss_cls_dn_2": 0.12887, "loss_box_dn_2": 0.69621, "loss_cls_dn_3": 0.13179, "loss_box_dn_3": 0.69411, "loss_cls_dn_4": 0.13441, "loss_box_dn_4": 0.69077, "loss_cls_dn_5": 0.13756, "loss_box_dn_5": 0.69713, "loss_dense_depth": 0.8003, "loss": 26.0275, "grad_norm": 31.79243, "time": 1.53066} -{"mode": "train", "epoch": 1, "iter": 220, "lr": 0.00019, "memory": 49163, "data_time": 0.08359, "loss_cls_0": 0.78552, "loss_box_0": 1.71018, "loss_cns_0": 0.62615, "loss_yns_0": 0.15463, "loss_cls_1": 0.86457, "loss_box_1": 1.68673, "loss_cns_1": 0.64888, "loss_yns_1": 0.15284, "loss_cls_2": 0.87102, "loss_box_2": 1.63165, "loss_cns_2": 0.65369, "loss_yns_2": 0.15359, "loss_cls_3": 0.88687, "loss_box_3": 1.61432, "loss_cns_3": 0.65554, "loss_yns_3": 0.15241, "loss_cls_4": 0.89233, "loss_box_4": 1.6057, "loss_cns_4": 0.653, "loss_yns_4": 0.15171, "loss_cls_5": 0.89423, "loss_box_5": 1.61434, "loss_cns_5": 0.6571, "loss_yns_5": 0.15434, "loss_cls_dn_0": 0.20353, "loss_box_dn_0": 0.74839, "loss_cls_dn_1": 0.12174, "loss_box_dn_1": 0.67618, "loss_cls_dn_2": 0.1251, "loss_box_dn_2": 0.6611, "loss_cls_dn_3": 0.12449, "loss_box_dn_3": 0.65875, "loss_cls_dn_4": 0.13134, "loss_box_dn_4": 0.65676, "loss_cls_dn_5": 0.13425, "loss_box_dn_5": 0.66157, "loss_dense_depth": 0.78424, "loss": 25.55878, "grad_norm": 33.83788, "time": 1.53212} -{"mode": "train", "epoch": 1, "iter": 221, "lr": 0.00019, "memory": 49163, "data_time": 0.15534, "loss_cls_0": 0.78484, "loss_box_0": 1.69868, "loss_cns_0": 0.62567, "loss_yns_0": 0.15433, "loss_cls_1": 0.86058, "loss_box_1": 1.70129, "loss_cns_1": 0.64838, "loss_yns_1": 0.15273, "loss_cls_2": 0.8875, "loss_box_2": 1.65427, "loss_cns_2": 0.6553, "loss_yns_2": 0.15219, "loss_cls_3": 0.89459, "loss_box_3": 1.63468, "loss_cns_3": 0.65417, "loss_yns_3": 0.15078, "loss_cls_4": 0.89525, "loss_box_4": 1.64142, "loss_cns_4": 0.65478, "loss_yns_4": 0.15037, "loss_cls_5": 0.90116, "loss_box_5": 1.63241, "loss_cns_5": 0.65444, "loss_yns_5": 0.15318, "loss_cls_dn_0": 0.20051, "loss_box_dn_0": 0.75269, "loss_cls_dn_1": 0.12679, "loss_box_dn_1": 0.66895, "loss_cls_dn_2": 0.12754, "loss_box_dn_2": 0.65202, "loss_cls_dn_3": 0.12948, "loss_box_dn_3": 0.65297, "loss_cls_dn_4": 0.13116, "loss_box_dn_4": 0.66075, "loss_cls_dn_5": 0.13568, "loss_box_dn_5": 0.66195, "loss_dense_depth": 0.76784, "loss": 25.6613, "grad_norm": 25.81645, "time": 1.60058} -{"mode": "train", "epoch": 1, "iter": 222, "lr": 0.00019, "memory": 49163, "data_time": 0.08129, "loss_cls_0": 0.79039, "loss_box_0": 1.7015, "loss_cns_0": 0.62353, "loss_yns_0": 0.15182, "loss_cls_1": 0.86967, "loss_box_1": 1.66393, "loss_cns_1": 0.65069, "loss_yns_1": 0.15009, "loss_cls_2": 0.89262, "loss_box_2": 1.6214, "loss_cns_2": 0.6575, "loss_yns_2": 0.14736, "loss_cls_3": 0.89117, "loss_box_3": 1.60005, "loss_cns_3": 0.65479, "loss_yns_3": 0.14941, "loss_cls_4": 0.90067, "loss_box_4": 1.60839, "loss_cns_4": 0.65636, "loss_yns_4": 0.14843, "loss_cls_5": 0.89918, "loss_box_5": 1.60691, "loss_cns_5": 0.65587, "loss_yns_5": 0.15096, "loss_cls_dn_0": 0.19804, "loss_box_dn_0": 0.75757, "loss_cls_dn_1": 0.12552, "loss_box_dn_1": 0.68657, "loss_cls_dn_2": 0.12761, "loss_box_dn_2": 0.66993, "loss_cls_dn_3": 0.13005, "loss_box_dn_3": 0.67486, "loss_cls_dn_4": 0.13166, "loss_box_dn_4": 0.68498, "loss_cls_dn_5": 0.13637, "loss_box_dn_5": 0.6936, "loss_dense_depth": 0.81397, "loss": 25.67343, "grad_norm": 35.20187, "time": 1.5638} -{"mode": "train", "epoch": 1, "iter": 223, "lr": 0.00019, "memory": 49163, "data_time": 0.07503, "loss_cls_0": 0.81934, "loss_box_0": 1.73694, "loss_cns_0": 0.62229, "loss_yns_0": 0.15368, "loss_cls_1": 0.8858, "loss_box_1": 1.69418, "loss_cns_1": 0.64955, "loss_yns_1": 0.15113, "loss_cls_2": 0.89831, "loss_box_2": 1.63941, "loss_cns_2": 0.65282, "loss_yns_2": 0.14981, "loss_cls_3": 0.90371, "loss_box_3": 1.62384, "loss_cns_3": 0.65433, "loss_yns_3": 0.15204, "loss_cls_4": 0.91526, "loss_box_4": 1.63973, "loss_cns_4": 0.66088, "loss_yns_4": 0.15199, "loss_cls_5": 0.90662, "loss_box_5": 1.62717, "loss_cns_5": 0.65981, "loss_yns_5": 0.15134, "loss_cls_dn_0": 0.2082, "loss_box_dn_0": 0.76056, "loss_cls_dn_1": 0.13192, "loss_box_dn_1": 0.70911, "loss_cls_dn_2": 0.13358, "loss_box_dn_2": 0.69595, "loss_cls_dn_3": 0.13456, "loss_box_dn_3": 0.69742, "loss_cls_dn_4": 0.14116, "loss_box_dn_4": 0.70988, "loss_cls_dn_5": 0.14448, "loss_box_dn_5": 0.71205, "loss_dense_depth": 0.79636, "loss": 26.0752, "grad_norm": 31.77205, "time": 1.59365} -{"mode": "train", "epoch": 1, "iter": 224, "lr": 0.00019, "memory": 49163, "data_time": 0.09657, "loss_cls_0": 0.7849, "loss_box_0": 1.71386, "loss_cns_0": 0.62785, "loss_yns_0": 0.15176, "loss_cls_1": 0.87044, "loss_box_1": 1.69756, "loss_cns_1": 0.65257, "loss_yns_1": 0.15116, "loss_cls_2": 0.90145, "loss_box_2": 1.65663, "loss_cns_2": 0.65749, "loss_yns_2": 0.15196, "loss_cls_3": 0.88532, "loss_box_3": 1.63156, "loss_cns_3": 0.6568, "loss_yns_3": 0.15264, "loss_cls_4": 0.90292, "loss_box_4": 1.65171, "loss_cns_4": 0.65865, "loss_yns_4": 0.1526, "loss_cls_5": 0.89324, "loss_box_5": 1.62884, "loss_cns_5": 0.65822, "loss_yns_5": 0.15319, "loss_cls_dn_0": 0.19428, "loss_box_dn_0": 0.75326, "loss_cls_dn_1": 0.12469, "loss_box_dn_1": 0.70616, "loss_cls_dn_2": 0.12646, "loss_box_dn_2": 0.69306, "loss_cls_dn_3": 0.12364, "loss_box_dn_3": 0.68683, "loss_cls_dn_4": 0.12766, "loss_box_dn_4": 0.69924, "loss_cls_dn_5": 0.13859, "loss_box_dn_5": 0.69438, "loss_dense_depth": 0.80861, "loss": 25.92018, "grad_norm": 29.36406, "time": 1.57824} -{"mode": "train", "epoch": 1, "iter": 225, "lr": 0.00019, "memory": 49163, "data_time": 0.07869, "loss_cls_0": 0.77694, "loss_box_0": 1.72561, "loss_cns_0": 0.62465, "loss_yns_0": 0.15352, "loss_cls_1": 0.87683, "loss_box_1": 1.68444, "loss_cns_1": 0.64962, "loss_yns_1": 0.15218, "loss_cls_2": 0.89603, "loss_box_2": 1.65269, "loss_cns_2": 0.65205, "loss_yns_2": 0.15176, "loss_cls_3": 0.90214, "loss_box_3": 1.62891, "loss_cns_3": 0.65162, "loss_yns_3": 0.15131, "loss_cls_4": 0.91548, "loss_box_4": 1.63141, "loss_cns_4": 0.65486, "loss_yns_4": 0.15064, "loss_cls_5": 0.90246, "loss_box_5": 1.62465, "loss_cns_5": 0.65481, "loss_yns_5": 0.15105, "loss_cls_dn_0": 0.18986, "loss_box_dn_0": 0.74713, "loss_cls_dn_1": 0.1225, "loss_box_dn_1": 0.69327, "loss_cls_dn_2": 0.12638, "loss_box_dn_2": 0.6838, "loss_cls_dn_3": 0.12864, "loss_box_dn_3": 0.67189, "loss_cls_dn_4": 0.13346, "loss_box_dn_4": 0.67326, "loss_cls_dn_5": 0.13358, "loss_box_dn_5": 0.67074, "loss_dense_depth": 0.84542, "loss": 25.8356, "grad_norm": 29.60189, "time": 1.54233} -{"mode": "train", "epoch": 1, "iter": 226, "lr": 0.00019, "memory": 49163, "data_time": 0.07742, "loss_cls_0": 0.80431, "loss_box_0": 1.7039, "loss_cns_0": 0.62728, "loss_yns_0": 0.15289, "loss_cls_1": 0.86567, "loss_box_1": 1.72607, "loss_cns_1": 0.64671, "loss_yns_1": 0.15099, "loss_cls_2": 0.90138, "loss_box_2": 1.65613, "loss_cns_2": 0.64811, "loss_yns_2": 0.14858, "loss_cls_3": 0.89663, "loss_box_3": 1.661, "loss_cns_3": 0.65139, "loss_yns_3": 0.15017, "loss_cls_4": 0.90563, "loss_box_4": 1.65196, "loss_cns_4": 0.65337, "loss_yns_4": 0.15045, "loss_cls_5": 0.91542, "loss_box_5": 1.68738, "loss_cns_5": 0.65215, "loss_yns_5": 0.14935, "loss_cls_dn_0": 0.19501, "loss_box_dn_0": 0.75446, "loss_cls_dn_1": 0.12417, "loss_box_dn_1": 0.68696, "loss_cls_dn_2": 0.12743, "loss_box_dn_2": 0.67168, "loss_cls_dn_3": 0.12535, "loss_box_dn_3": 0.66928, "loss_cls_dn_4": 0.12926, "loss_box_dn_4": 0.66562, "loss_cls_dn_5": 0.13051, "loss_box_dn_5": 0.67615, "loss_dense_depth": 0.77813, "loss": 25.89092, "grad_norm": 41.09734, "time": 1.52291} -{"mode": "train", "epoch": 1, "iter": 227, "lr": 0.00019, "memory": 49163, "data_time": 0.0757, "loss_cls_0": 0.79402, "loss_box_0": 1.72353, "loss_cns_0": 0.6179, "loss_yns_0": 0.14924, "loss_cls_1": 0.86808, "loss_box_1": 1.70293, "loss_cns_1": 0.6394, "loss_yns_1": 0.14695, "loss_cls_2": 0.90696, "loss_box_2": 1.64706, "loss_cns_2": 0.6366, "loss_yns_2": 0.14524, "loss_cls_3": 0.8939, "loss_box_3": 1.66897, "loss_cns_3": 0.64759, "loss_yns_3": 0.14917, "loss_cls_4": 0.89611, "loss_box_4": 1.65968, "loss_cns_4": 0.64869, "loss_yns_4": 0.14877, "loss_cls_5": 0.89102, "loss_box_5": 1.67885, "loss_cns_5": 0.64929, "loss_yns_5": 0.14756, "loss_cls_dn_0": 0.19635, "loss_box_dn_0": 0.76227, "loss_cls_dn_1": 0.12366, "loss_box_dn_1": 0.69937, "loss_cls_dn_2": 0.12938, "loss_box_dn_2": 0.69383, "loss_cls_dn_3": 0.12397, "loss_box_dn_3": 0.68995, "loss_cls_dn_4": 0.12739, "loss_box_dn_4": 0.69071, "loss_cls_dn_5": 0.13035, "loss_box_dn_5": 0.70368, "loss_dense_depth": 0.81262, "loss": 25.94105, "grad_norm": 42.17367, "time": 1.51087} -{"mode": "train", "epoch": 1, "iter": 228, "lr": 0.00019, "memory": 49163, "data_time": 0.07353, "loss_cls_0": 0.79579, "loss_box_0": 1.71506, "loss_cns_0": 0.62286, "loss_yns_0": 0.15092, "loss_cls_1": 0.86081, "loss_box_1": 1.70732, "loss_cns_1": 0.6471, "loss_yns_1": 0.14905, "loss_cls_2": 0.88726, "loss_box_2": 1.6665, "loss_cns_2": 0.64394, "loss_yns_2": 0.14757, "loss_cls_3": 0.90335, "loss_box_3": 1.67318, "loss_cns_3": 0.65484, "loss_yns_3": 0.15373, "loss_cls_4": 0.89697, "loss_box_4": 1.6704, "loss_cns_4": 0.65467, "loss_yns_4": 0.15183, "loss_cls_5": 0.88529, "loss_box_5": 1.67165, "loss_cns_5": 0.65506, "loss_yns_5": 0.15031, "loss_cls_dn_0": 0.19741, "loss_box_dn_0": 0.75207, "loss_cls_dn_1": 0.12259, "loss_box_dn_1": 0.70637, "loss_cls_dn_2": 0.12334, "loss_box_dn_2": 0.70089, "loss_cls_dn_3": 0.12817, "loss_box_dn_3": 0.69512, "loss_cls_dn_4": 0.12647, "loss_box_dn_4": 0.70094, "loss_cls_dn_5": 0.12643, "loss_box_dn_5": 0.7096, "loss_dense_depth": 0.7837, "loss": 25.98852, "grad_norm": 33.34052, "time": 1.5199} -{"mode": "train", "epoch": 1, "iter": 229, "lr": 0.00019, "memory": 49163, "data_time": 0.07806, "loss_cls_0": 0.79696, "loss_box_0": 1.70669, "loss_cns_0": 0.61988, "loss_yns_0": 0.14719, "loss_cls_1": 0.85719, "loss_box_1": 1.67688, "loss_cns_1": 0.65002, "loss_yns_1": 0.14648, "loss_cls_2": 0.88317, "loss_box_2": 1.65947, "loss_cns_2": 0.65296, "loss_yns_2": 0.1475, "loss_cls_3": 0.88491, "loss_box_3": 1.64859, "loss_cns_3": 0.65415, "loss_yns_3": 0.14772, "loss_cls_4": 0.89546, "loss_box_4": 1.64381, "loss_cns_4": 0.65489, "loss_yns_4": 0.1477, "loss_cls_5": 0.90336, "loss_box_5": 1.64201, "loss_cns_5": 0.65283, "loss_yns_5": 0.14698, "loss_cls_dn_0": 0.18971, "loss_box_dn_0": 0.74138, "loss_cls_dn_1": 0.11865, "loss_box_dn_1": 0.70803, "loss_cls_dn_2": 0.11985, "loss_box_dn_2": 0.70542, "loss_cls_dn_3": 0.1201, "loss_box_dn_3": 0.70784, "loss_cls_dn_4": 0.12573, "loss_box_dn_4": 0.71202, "loss_cls_dn_5": 0.1315, "loss_box_dn_5": 0.71988, "loss_dense_depth": 0.77904, "loss": 25.84594, "grad_norm": 41.08479, "time": 1.52659} -{"mode": "train", "epoch": 1, "iter": 230, "lr": 0.00019, "memory": 49163, "data_time": 0.07656, "loss_cls_0": 0.79911, "loss_box_0": 1.74064, "loss_cns_0": 0.62331, "loss_yns_0": 0.15094, "loss_cls_1": 0.85517, "loss_box_1": 1.69851, "loss_cns_1": 0.64928, "loss_yns_1": 0.14684, "loss_cls_2": 0.89159, "loss_box_2": 1.67172, "loss_cns_2": 0.65301, "loss_yns_2": 0.14778, "loss_cls_3": 0.89238, "loss_box_3": 1.63847, "loss_cns_3": 0.65107, "loss_yns_3": 0.14807, "loss_cls_4": 0.89361, "loss_box_4": 1.62306, "loss_cns_4": 0.65476, "loss_yns_4": 0.14849, "loss_cls_5": 0.89803, "loss_box_5": 1.65243, "loss_cns_5": 0.6571, "loss_yns_5": 0.14732, "loss_cls_dn_0": 0.19217, "loss_box_dn_0": 0.7526, "loss_cls_dn_1": 0.11697, "loss_box_dn_1": 0.70652, "loss_cls_dn_2": 0.11729, "loss_box_dn_2": 0.70261, "loss_cls_dn_3": 0.11809, "loss_box_dn_3": 0.70351, "loss_cls_dn_4": 0.12109, "loss_box_dn_4": 0.70531, "loss_cls_dn_5": 0.12621, "loss_box_dn_5": 0.71835, "loss_dense_depth": 0.76354, "loss": 25.87698, "grad_norm": 34.72187, "time": 1.51744} -{"mode": "train", "epoch": 1, "iter": 231, "lr": 0.00019, "memory": 49163, "data_time": 0.07375, "loss_cls_0": 0.78683, "loss_box_0": 1.70745, "loss_cns_0": 0.62545, "loss_yns_0": 0.15081, "loss_cls_1": 0.84532, "loss_box_1": 1.646, "loss_cns_1": 0.6494, "loss_yns_1": 0.1467, "loss_cls_2": 0.89749, "loss_box_2": 1.6076, "loss_cns_2": 0.65474, "loss_yns_2": 0.14497, "loss_cls_3": 0.86469, "loss_box_3": 1.57805, "loss_cns_3": 0.65452, "loss_yns_3": 0.14721, "loss_cls_4": 0.8782, "loss_box_4": 1.59159, "loss_cns_4": 0.65626, "loss_yns_4": 0.14766, "loss_cls_5": 0.87937, "loss_box_5": 1.59652, "loss_cns_5": 0.66006, "loss_yns_5": 0.14558, "loss_cls_dn_0": 0.1858, "loss_box_dn_0": 0.74805, "loss_cls_dn_1": 0.11612, "loss_box_dn_1": 0.6897, "loss_cls_dn_2": 0.11673, "loss_box_dn_2": 0.67792, "loss_cls_dn_3": 0.11682, "loss_box_dn_3": 0.67766, "loss_cls_dn_4": 0.11836, "loss_box_dn_4": 0.68572, "loss_cls_dn_5": 0.12278, "loss_box_dn_5": 0.69326, "loss_dense_depth": 0.73432, "loss": 25.3457, "grad_norm": 37.45884, "time": 1.52597} diff --git a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20251112_212046.log b/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20251112_212046.log deleted file mode 100644 index 8a1c5bd1ea79e842fdd658c652c3f75efec167d7..0000000000000000000000000000000000000000 --- a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20251112_212046.log +++ /dev/null @@ -1,3220 +0,0 @@ -2025-11-12 21:20:46,093 - mmdet - INFO - Environment info: ------------------------------------------------------------- -sys.platform: linux -Python: 3.10.12 (main, May 27 2025, 17:12:29) [GCC 11.4.0] -CUDA available: True -GPU 0,1,2,3,4,5,6,7: BW200, UBB BW1000 -CUDA_HOME: /opt/dtk -NVCC: Not Available -GCC: x86_64-linux-gnu-gcc (Ubuntu 11.4.0-1ubuntu1~22.04) 11.4.0 -PyTorch: 2.4.1 -PyTorch compiling details: PyTorch built with: - - GCC 10.3 - - C++ Version: 201703 - - Intel(R) Math Kernel Library Version 2020.0.4 Product Build 20200917 for Intel(R) 64 architecture applications - - OpenMP 201511 (a.k.a. OpenMP 4.5) - - LAPACK is enabled (usually provided by MKL) - - NNPACK is enabled - - CPU capability usage: AVX512 - - HIP Runtime 6.3.25211 - - MIOpen 2.17.0 - - Magma 2.8.0 - - Build settings: BLAS_INFO=mkl, BUILD_TYPE=Release, CXX_COMPILER=/opt/rh/gcc-toolset-10/root/usr/bin/c++, CXX_FLAGS= -D_GLIBCXX_USE_CXX11_ABI=1 -fvisibility-inlines-hidden -DUSE_PTHREADPOOL -DNDEBUG -DUSE_KINETO -DLIBKINETO_NOCUPTI -DUSE_FBGEMM -DUSE_PYTORCH_QNNPACK -DUSE_XNNPACK -DSYMBOLICATE_MOBILE_DEBUG_HANDLE -O2 -fPIC -Wall -Wextra -Werror=return-type -Werror=non-virtual-dtor -Werror=bool-operation -Wnarrowing -Wno-missing-field-initializers -Wno-type-limits -Wno-array-bounds -Wno-unknown-pragmas -Wno-unused-parameter -Wno-unused-function -Wno-unused-result -Wno-strict-overflow -Wno-strict-aliasing -Wno-stringop-overflow -Wsuggest-override -Wno-psabi -Wno-error=pedantic -Wno-error=old-style-cast -Wno-missing-braces -fdiagnostics-color=always -faligned-new -Wno-unused-but-set-variable -Wno-maybe-uninitialized -fno-math-errno -fno-trapping-math -Werror=format -Wno-stringop-overflow, FORCE_FALLBACK_CUDA_MPI=1, LAPACK_INFO=mkl, PERF_WITH_AVX=1, PERF_WITH_AVX2=1, PERF_WITH_AVX512=1, TORCH_VERSION=2.4.1, USE_CUDA=0, USE_CUDNN=OFF, USE_CUSPARSELT=OFF, USE_EXCEPTION_PTR=1, USE_GFLAGS=1, USE_GLOG=1, USE_GLOO=1, USE_MKL=ON, USE_MKLDNN=0, USE_MPI=1, USE_NCCL=1, USE_NNPACK=ON, USE_OPENMP=1, USE_ROCM=ON, USE_ROCM_KERNEL_ASSERT=OFF, - -TorchVision: 0.19.1 -OpenCV: 4.11.0 -MMCV: 1.6.1 -MMCV Compiler: GCC 11.4 -MMCV CUDA Compiler: rocm not available -MMDetection: 2.26.0+c41df4b ------------------------------------------------------------- - -2025-11-12 21:20:47,030 - mmdet - INFO - Distributed training: True -2025-11-12 21:20:47,752 - mmdet - INFO - Config: -plugin = True -plugin_dir = 'projects/mmdet3d_plugin/' -dist_params = dict(backend='nccl') -log_level = 'INFO' -work_dir = './work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704' -total_batch_size = 160 -num_gpus = 8 -batch_size = 20 -num_iters_per_epoch = 175 -num_epochs = 100 -checkpoint_epoch_interval = 20 -checkpoint_config = dict(interval=3500) -log_config = dict( - interval=1, - hooks=[ - dict(type='TextLoggerHook', by_epoch=False), - dict(type='TensorboardLoggerHook') - ]) -load_from = None -resume_from = None -workflow = [('train', 1)] -fp16 = dict(loss_scale=32.0) -input_shape = (704, 256) -tracking_test = True -tracking_threshold = 0.2 -class_names = [ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier', - 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' -] -num_classes = 10 -embed_dims = 256 -num_groups = 8 -num_decoder = 6 -num_single_frame_decoder = 1 -use_deformable_func = True -strides = [4, 8, 16, 32] -num_levels = 4 -num_depth_layers = 3 -drop_out = 0.1 -temporal = True -decouple_attn = True -with_quality_estimation = True -model = dict( - type='Sparse4D', - use_grid_mask=True, - use_deformable_func=True, - img_backbone=dict( - type='ResNet', - depth=50, - num_stages=4, - frozen_stages=-1, - norm_eval=False, - style='pytorch', - with_cp=True, - out_indices=(0, 1, 2, 3), - norm_cfg=dict(type='BN', requires_grad=True), - pretrained='ckpt/resnet50-19c8e357.pth'), - img_neck=dict( - type='FPN', - num_outs=4, - start_level=0, - out_channels=256, - add_extra_convs='on_output', - relu_before_extra_convs=True, - in_channels=[256, 512, 1024, 2048]), - depth_branch=dict( - type='DenseDepthNet', - embed_dims=256, - num_depth_layers=3, - loss_weight=0.2), - head=dict( - type='Sparse4DHead', - cls_threshold_to_reg=0.05, - decouple_attn=True, - instance_bank=dict( - type='InstanceBank', - num_anchor=900, - embed_dims=256, - anchor='nuscenes_kmeans900.npy', - anchor_handler=dict(type='SparseBox3DKeyPointsGenerator'), - num_temp_instances=600, - confidence_decay=0.6, - feat_grad=False), - anchor_encoder=dict( - type='SparseBox3DEncoder', - vel_dims=3, - embed_dims=[128, 32, 32, 64], - mode='cat', - output_fc=False, - in_loops=1, - out_loops=4), - num_single_frame_decoder=1, - operation_order=[ - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine' - ], - temp_graph_model=dict( - type='MultiheadAttention', - embed_dims=512, - num_heads=8, - batch_first=True, - dropout=0.1), - graph_model=dict( - type='MultiheadAttention', - embed_dims=512, - num_heads=8, - batch_first=True, - dropout=0.1), - norm_layer=dict(type='LN', normalized_shape=256), - ffn=dict( - type='AsymmetricFFN', - in_channels=512, - pre_norm=dict(type='LN'), - embed_dims=256, - feedforward_channels=1024, - num_fcs=2, - ffn_drop=0.1, - act_cfg=dict(type='ReLU', inplace=True)), - deformable_model=dict( - type='DeformableFeatureAggregation', - embed_dims=256, - num_groups=8, - num_levels=4, - num_cams=6, - attn_drop=0.15, - use_deformable_func=True, - use_camera_embed=True, - residual_mode='cat', - kps_generator=dict( - type='SparseBox3DKeyPointsGenerator', - num_learnable_pts=6, - fix_scale=[[0, 0, 0], [0.45, 0, 0], [-0.45, 0, 0], - [0, 0.45, 0], [0, -0.45, 0], [0, 0, 0.45], - [0, 0, -0.45]])), - refine_layer=dict( - type='SparseBox3DRefinementModule', - embed_dims=256, - num_cls=10, - refine_yaw=True, - with_quality_estimation=True), - sampler=dict( - type='SparseBox3DTarget', - num_dn_groups=5, - num_temp_dn_groups=3, - dn_noise_scale=[2.0, 2.0, 2.0, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5], - max_dn_gt=32, - add_neg_dn=True, - cls_weight=2.0, - box_weight=0.25, - reg_weights=[2.0, 2.0, 2.0, 0.5, 0.5, 0.5, 0.0, 0.0, 0.0, 0.0], - cls_wise_reg_weights=dict( - {9: [2.0, 2.0, 2.0, 1.0, 1.0, 1.0, 0.0, 0.0, 1.0, 1.0]})), - loss_cls=dict( - type='FocalLoss', - use_sigmoid=True, - gamma=2.0, - alpha=0.25, - loss_weight=2.0), - loss_reg=dict( - type='SparseBox3DLoss', - loss_box=dict(type='L1Loss', loss_weight=0.25), - loss_centerness=dict(type='CrossEntropyLoss', use_sigmoid=True), - loss_yawness=dict(type='GaussianFocalLoss'), - cls_allow_reverse=[5]), - decoder=dict(type='SparseBox3DDecoder'), - reg_weights=[2.0, 2.0, 2.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0])) -dataset_type = 'NuScenes3DDetTrackDataset' -data_root = 'data/nuscenes/' -anno_root = 'data/nuscenes_anno_pkls/' -file_client_args = dict(backend='disk') -img_norm_cfg = dict( - mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375], to_rgb=True) -train_pipeline = [ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict( - type='LoadPointsFromFile', - coord_type='LIDAR', - load_dim=5, - use_dim=5, - file_client_args=dict(backend='disk')), - dict(type='ResizeCropFlipImage'), - dict(type='MultiScaleDepthMapGenerator', downsample=[4, 8, 16]), - dict(type='BBoxRotation'), - dict(type='PhotoMetricDistortionMultiViewImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict( - type='CircleObjectRangeFilter', - class_dist_thred=[55, 55, 55, 55, 55, 55, 55, 55, 55, 55]), - dict( - type='InstanceNameFilter', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ]), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=[ - 'img', 'timestamp', 'projection_mat', 'image_wh', 'gt_depth', - 'focal', 'gt_bboxes_3d', 'gt_labels_3d' - ], - meta_keys=['T_global', 'T_global_inv', 'timestamp', 'instance_id']) -] -test_pipeline = [ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='ResizeCropFlipImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=['img', 'timestamp', 'projection_mat', 'image_wh'], - meta_keys=['T_global', 'T_global_inv', 'timestamp']) -] -input_modality = dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False) -data_basic_config = dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier', - 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval') -data_aug_conf = dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]) -data = dict( - samples_per_gpu=20, - workers_per_gpu=20, - train=dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval', - ann_file='data/nuscenes_anno_pkls/nuscenes_infos_train.pkl', - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict( - type='LoadPointsFromFile', - coord_type='LIDAR', - load_dim=5, - use_dim=5, - file_client_args=dict(backend='disk')), - dict(type='ResizeCropFlipImage'), - dict(type='MultiScaleDepthMapGenerator', downsample=[4, 8, 16]), - dict(type='BBoxRotation'), - dict(type='PhotoMetricDistortionMultiViewImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict( - type='CircleObjectRangeFilter', - class_dist_thred=[55, 55, 55, 55, 55, 55, 55, 55, 55, 55]), - dict( - type='InstanceNameFilter', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', - 'traffic_cone' - ]), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=[ - 'img', 'timestamp', 'projection_mat', 'image_wh', - 'gt_depth', 'focal', 'gt_bboxes_3d', 'gt_labels_3d' - ], - meta_keys=[ - 'T_global', 'T_global_inv', 'timestamp', 'instance_id' - ]) - ], - test_mode=False, - data_aug_conf=dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]), - with_seq_flag=True, - sequences_split_num=2, - keep_consistent_seq_aug=True), - val=dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval', - ann_file='data/nuscenes_anno_pkls/nuscenes_infos_val.pkl', - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='ResizeCropFlipImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=['img', 'timestamp', 'projection_mat', 'image_wh'], - meta_keys=['T_global', 'T_global_inv', 'timestamp']) - ], - data_aug_conf=dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]), - test_mode=True, - tracking=True, - tracking_threshold=0.2), - test=dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval', - ann_file='data/nuscenes_anno_pkls/nuscenes_infos_val.pkl', - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='ResizeCropFlipImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=['img', 'timestamp', 'projection_mat', 'image_wh'], - meta_keys=['T_global', 'T_global_inv', 'timestamp']) - ], - data_aug_conf=dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]), - test_mode=True, - tracking=True, - tracking_threshold=0.2)) -optimizer = dict( - type='Miopen_AdamW', - lr=0.0006, - weight_decay=0.001, - paramwise_cfg=dict(custom_keys=dict(img_backbone=dict(lr_mult=0.5)))) -optimizer_config = dict(grad_clip=dict(max_norm=25, norm_type=2)) -lr_config = dict( - policy='CosineAnnealing', - warmup='linear', - warmup_iters=500, - warmup_ratio=0.3333333333333333, - min_lr_ratio=0.001) -runner = dict(type='IterBasedRunner', max_iters=17500) -vis_pipeline = [ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='Collect', keys=['img'], meta_keys=['timestamp', 'lidar2img']) -] -evaluation = dict( - interval=3500, - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict( - type='Collect', keys=['img'], meta_keys=['timestamp', 'lidar2img']) - ]) -gpu_ids = range(0, 8) - -2025-11-12 21:20:47,752 - mmdet - INFO - Set random seed to 0, deterministic: False -2025-11-12 21:20:48,055 - mmdet - INFO - initialize ResNet with init_cfg {'type': 'Pretrained', 'checkpoint': 'ckpt/resnet50-19c8e357.pth'} -2025-11-12 21:20:48,306 - mmdet - INFO - initialize FPN with init_cfg {'type': 'Xavier', 'layer': 'Conv2d', 'distribution': 'uniform'} -Name of parameter - Initialization information - -img_backbone.conv1.weight - torch.Size([64, 3, 7, 7]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.bn1.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.bn1.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.conv1.weight - torch.Size([64, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn1.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn1.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.conv2.weight - torch.Size([64, 64, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn2.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn2.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.conv3.weight - torch.Size([256, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn3.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn3.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.downsample.0.weight - torch.Size([256, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.downsample.1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.downsample.1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.conv1.weight - torch.Size([64, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn1.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn1.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.conv2.weight - torch.Size([64, 64, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn2.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn2.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.conv3.weight - torch.Size([256, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn3.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn3.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.conv1.weight - torch.Size([64, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn1.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn1.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.conv2.weight - torch.Size([64, 64, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn2.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn2.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.conv3.weight - torch.Size([256, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn3.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn3.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.conv1.weight - torch.Size([128, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn1.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn1.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.conv2.weight - torch.Size([128, 128, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn2.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn2.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.conv3.weight - torch.Size([512, 128, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn3.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn3.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.downsample.0.weight - torch.Size([512, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.downsample.1.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.downsample.1.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.conv1.weight - torch.Size([128, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn1.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn1.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.conv2.weight - torch.Size([128, 128, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn2.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn2.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.conv3.weight - torch.Size([512, 128, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn3.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn3.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.conv1.weight - torch.Size([128, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn1.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn1.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.conv2.weight - torch.Size([128, 128, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn2.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn2.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.conv3.weight - torch.Size([512, 128, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn3.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn3.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.conv1.weight - torch.Size([128, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn1.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn1.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.conv2.weight - torch.Size([128, 128, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn2.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn2.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.conv3.weight - torch.Size([512, 128, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn3.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn3.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.conv1.weight - torch.Size([256, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.downsample.0.weight - torch.Size([1024, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.downsample.1.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.downsample.1.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.conv1.weight - torch.Size([512, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn1.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn1.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.conv2.weight - torch.Size([512, 512, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn2.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn2.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.conv3.weight - torch.Size([2048, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn3.weight - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn3.bias - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.downsample.0.weight - torch.Size([2048, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.downsample.1.weight - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.downsample.1.bias - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.conv1.weight - torch.Size([512, 2048, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn1.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn1.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.conv2.weight - torch.Size([512, 512, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn2.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn2.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.conv3.weight - torch.Size([2048, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn3.weight - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn3.bias - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.conv1.weight - torch.Size([512, 2048, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn1.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn1.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.conv2.weight - torch.Size([512, 512, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn2.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn2.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.conv3.weight - torch.Size([2048, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn3.weight - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn3.bias - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_neck.lateral_convs.0.conv.weight - torch.Size([256, 256, 1, 1]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.lateral_convs.0.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.lateral_convs.1.conv.weight - torch.Size([256, 512, 1, 1]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.lateral_convs.1.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.lateral_convs.2.conv.weight - torch.Size([256, 1024, 1, 1]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.lateral_convs.2.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.lateral_convs.3.conv.weight - torch.Size([256, 2048, 1, 1]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.lateral_convs.3.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.fpn_convs.0.conv.weight - torch.Size([256, 256, 3, 3]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.fpn_convs.0.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.fpn_convs.1.conv.weight - torch.Size([256, 256, 3, 3]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.fpn_convs.1.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.fpn_convs.2.conv.weight - torch.Size([256, 256, 3, 3]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.fpn_convs.2.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.fpn_convs.3.conv.weight - torch.Size([256, 256, 3, 3]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.fpn_convs.3.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.instance_bank.anchor - torch.Size([900, 11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.instance_bank.instance_feature - torch.Size([900, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.instance_bank.anchor_handler.fix_scale - torch.Size([1, 3]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.0.weight - torch.Size([128, 3]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.0.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.2.weight - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.2.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.3.weight - torch.Size([128, 128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.3.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.5.weight - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.5.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.6.weight - torch.Size([128, 128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.6.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.8.weight - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.8.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.9.weight - torch.Size([128, 128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.9.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.11.weight - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.11.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.0.weight - torch.Size([32, 3]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.0.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.2.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.2.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.3.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.3.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.5.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.5.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.6.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.6.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.8.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.8.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.9.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.9.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.11.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.11.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.0.weight - torch.Size([32, 2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.0.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.2.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.2.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.3.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.3.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.5.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.5.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.6.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.6.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.8.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.8.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.9.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.9.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.11.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.11.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.0.weight - torch.Size([64, 3]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.0.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.2.weight - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.2.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.3.weight - torch.Size([64, 64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.3.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.5.weight - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.5.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.6.weight - torch.Size([64, 64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.6.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.8.weight - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.8.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.9.weight - torch.Size([64, 64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.9.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.11.weight - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.11.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.1.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.1.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.1.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.1.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.1.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.1.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.1.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.1.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.3.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.4.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.4.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.4.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.4.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.5.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.5.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.5.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.5.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.6.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.6.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.8.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.8.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.8.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.8.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.8.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.8.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.8.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.8.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.10.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.11.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.11.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.11.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.11.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.12.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.12.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.12.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.12.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.13.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.13.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.15.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.15.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.15.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.15.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.15.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.15.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.15.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.15.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.16.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.16.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.17.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.18.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.18.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.18.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.18.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.19.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.19.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.19.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.19.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.20.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.20.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.22.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.22.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.22.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.22.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.22.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.22.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.22.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.22.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.23.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.23.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.24.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.25.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.25.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.25.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.25.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.26.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.26.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.26.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.26.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.27.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.27.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.29.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.29.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.29.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.29.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.29.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.29.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.29.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.29.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.30.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.30.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.31.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.32.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.32.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.32.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.32.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.33.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.33.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.33.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.33.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.34.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.34.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.36.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.36.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.36.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.36.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.36.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.36.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.36.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.36.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.37.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.37.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.38.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.fc_before.weight - torch.Size([512, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.fc_after.weight - torch.Size([256, 512]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.0.weight - torch.Size([1, 256, 1, 1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.0.bias - torch.Size([1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.1.weight - torch.Size([1, 256, 1, 1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.1.bias - torch.Size([1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.2.weight - torch.Size([1, 256, 1, 1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.2.bias - torch.Size([1]): -The value is the same before and after calling `init_weights` of Sparse4D -2025-11-12 21:20:48,397 - mmdet - INFO - Model: -Sparse4D( - (img_backbone): ResNet( - (conv1): Conv2d(3, 64, kernel_size=(7, 7), stride=(2, 2), padding=(3, 3), bias=False) - (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (maxpool): MaxPool2d(kernel_size=3, stride=2, padding=1, dilation=1, ceil_mode=False) - (layer1): ResLayer( - (0): Bottleneck( - (conv1): Conv2d(64, 64, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (downsample): Sequential( - (0): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - ) - ) - (1): Bottleneck( - (conv1): Conv2d(256, 64, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (2): Bottleneck( - (conv1): Conv2d(256, 64, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - ) - (layer2): ResLayer( - (0): Bottleneck( - (conv1): Conv2d(256, 128, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (downsample): Sequential( - (0): Conv2d(256, 512, kernel_size=(1, 1), stride=(2, 2), bias=False) - (1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - ) - ) - (1): Bottleneck( - (conv1): Conv2d(512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (2): Bottleneck( - (conv1): Conv2d(512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (3): Bottleneck( - (conv1): Conv2d(512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - ) - (layer3): ResLayer( - (0): Bottleneck( - (conv1): Conv2d(512, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (downsample): Sequential( - (0): Conv2d(512, 1024, kernel_size=(1, 1), stride=(2, 2), bias=False) - (1): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - ) - ) - (1): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (2): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (3): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (4): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (5): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - ) - (layer4): ResLayer( - (0): Bottleneck( - (conv1): Conv2d(1024, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(512, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (downsample): Sequential( - (0): Conv2d(1024, 2048, kernel_size=(1, 1), stride=(2, 2), bias=False) - (1): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - ) - ) - (1): Bottleneck( - (conv1): Conv2d(2048, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(512, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (2): Bottleneck( - (conv1): Conv2d(2048, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(512, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - ) - ) - init_cfg={'type': 'Pretrained', 'checkpoint': 'ckpt/resnet50-19c8e357.pth'} - (img_neck): FPN( - (lateral_convs): ModuleList( - (0): ConvModule( - (conv): Conv2d(256, 256, kernel_size=(1, 1), stride=(1, 1)) - ) - (1): ConvModule( - (conv): Conv2d(512, 256, kernel_size=(1, 1), stride=(1, 1)) - ) - (2): ConvModule( - (conv): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1)) - ) - (3): ConvModule( - (conv): Conv2d(2048, 256, kernel_size=(1, 1), stride=(1, 1)) - ) - ) - (fpn_convs): ModuleList( - (0-3): 4 x ConvModule( - (conv): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1)) - ) - ) - ) - init_cfg={'type': 'Xavier', 'layer': 'Conv2d', 'distribution': 'uniform'} - (head): Sparse4DHead( - (instance_bank): InstanceBank( - (anchor_handler): SparseBox3DKeyPointsGenerator() - ) - (anchor_encoder): SparseBox3DEncoder( - (pos_fc): Sequential( - (0): Linear(in_features=3, out_features=128, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((128,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=128, out_features=128, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((128,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=128, out_features=128, bias=True) - (7): ReLU(inplace=True) - (8): LayerNorm((128,), eps=1e-05, elementwise_affine=True) - (9): Linear(in_features=128, out_features=128, bias=True) - (10): ReLU(inplace=True) - (11): LayerNorm((128,), eps=1e-05, elementwise_affine=True) - ) - (size_fc): Sequential( - (0): Linear(in_features=3, out_features=32, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=32, out_features=32, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=32, out_features=32, bias=True) - (7): ReLU(inplace=True) - (8): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (9): Linear(in_features=32, out_features=32, bias=True) - (10): ReLU(inplace=True) - (11): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - ) - (yaw_fc): Sequential( - (0): Linear(in_features=2, out_features=32, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=32, out_features=32, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=32, out_features=32, bias=True) - (7): ReLU(inplace=True) - (8): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (9): Linear(in_features=32, out_features=32, bias=True) - (10): ReLU(inplace=True) - (11): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - ) - (vel_fc): Sequential( - (0): Linear(in_features=3, out_features=64, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((64,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=64, out_features=64, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((64,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=64, out_features=64, bias=True) - (7): ReLU(inplace=True) - (8): LayerNorm((64,), eps=1e-05, elementwise_affine=True) - (9): Linear(in_features=64, out_features=64, bias=True) - (10): ReLU(inplace=True) - (11): LayerNorm((64,), eps=1e-05, elementwise_affine=True) - ) - ) - (loss_cls): FocalLoss() - (loss_reg): SparseBox3DLoss( - (loss_box): L1Loss() - (loss_cns): CrossEntropyLoss(avg_non_ignore=False) - (loss_yns): GaussianFocalLoss() - ) - (layers): ModuleList( - (0): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (1): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (4-5): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (6): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (7): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (8): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (11-12): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (13): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (14): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (15): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (16): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (17): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (18-19): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (20): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (21): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (22): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (23): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (24): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (25-26): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (27): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (28): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (29): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (30): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (31): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (32-33): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (34): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (35): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (36): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (37): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (38): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - ) - (fc_before): Linear(in_features=256, out_features=512, bias=False) - (fc_after): Linear(in_features=512, out_features=256, bias=False) - ) - (depth_branch): DenseDepthNet( - (depth_layers): ModuleList( - (0-2): 3 x Conv2d(256, 1, kernel_size=(1, 1), stride=(1, 1)) - ) - ) - (grid_mask): GridMask() -) -2025-11-12 21:21:01,086 - mmdet - INFO - Start running, host: root@VM-120-96-tencentos, work_dir: /cfs/auto/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704 -2025-11-12 21:21:01,086 - mmdet - INFO - Hooks will be executed in the following order: -before_run: -(VERY_HIGH ) CosineAnnealingLrUpdaterHook -(ABOVE_NORMAL) Fp16OptimizerHook -(NORMAL ) CheckpointHook -(NORMAL ) CustomDistEvalHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -before_train_epoch: -(VERY_HIGH ) CosineAnnealingLrUpdaterHook -(NORMAL ) CustomDistEvalHook -(LOW ) IterTimerHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -before_train_iter: -(VERY_HIGH ) CosineAnnealingLrUpdaterHook -(NORMAL ) CustomDistEvalHook -(LOW ) IterTimerHook - -------------------- -after_train_iter: -(ABOVE_NORMAL) Fp16OptimizerHook -(NORMAL ) CheckpointHook -(NORMAL ) CustomDistEvalHook -(LOW ) IterTimerHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -after_train_epoch: -(NORMAL ) CheckpointHook -(NORMAL ) CustomDistEvalHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -before_val_epoch: -(LOW ) IterTimerHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -before_val_iter: -(LOW ) IterTimerHook - -------------------- -after_val_iter: -(LOW ) IterTimerHook - -------------------- -after_val_epoch: -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -after_run: -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -2025-11-12 21:21:01,086 - mmdet - INFO - workflow: [('train', 1)], max: 17500 iters -2025-11-12 21:21:01,088 - mmdet - INFO - Checkpoints will be saved to /cfs/auto/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704 by HardDiskBackend. diff --git a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20251112_212046.log.json b/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20251112_212046.log.json deleted file mode 100644 index 1fda3e213098b4a1f6239f0e36f9459d0a88f9c1..0000000000000000000000000000000000000000 --- a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20251112_212046.log.json +++ /dev/null @@ -1 +0,0 @@ -{"env_info": "sys.platform: linux\nPython: 3.10.12 (main, May 27 2025, 17:12:29) [GCC 11.4.0]\nCUDA available: True\nGPU 0,1,2,3,4,5,6,7: BW200, UBB BW1000\nCUDA_HOME: /opt/dtk\nNVCC: Not Available\nGCC: x86_64-linux-gnu-gcc (Ubuntu 11.4.0-1ubuntu1~22.04) 11.4.0\nPyTorch: 2.4.1\nPyTorch compiling details: PyTorch built with:\n - GCC 10.3\n - C++ Version: 201703\n - Intel(R) Math Kernel Library Version 2020.0.4 Product Build 20200917 for Intel(R) 64 architecture applications\n - OpenMP 201511 (a.k.a. OpenMP 4.5)\n - LAPACK is enabled (usually provided by MKL)\n - NNPACK is enabled\n - CPU capability usage: AVX512\n - HIP Runtime 6.3.25211\n - MIOpen 2.17.0\n - Magma 2.8.0\n - Build settings: BLAS_INFO=mkl, BUILD_TYPE=Release, CXX_COMPILER=/opt/rh/gcc-toolset-10/root/usr/bin/c++, CXX_FLAGS= -D_GLIBCXX_USE_CXX11_ABI=1 -fvisibility-inlines-hidden -DUSE_PTHREADPOOL -DNDEBUG -DUSE_KINETO -DLIBKINETO_NOCUPTI -DUSE_FBGEMM -DUSE_PYTORCH_QNNPACK -DUSE_XNNPACK -DSYMBOLICATE_MOBILE_DEBUG_HANDLE -O2 -fPIC -Wall -Wextra -Werror=return-type -Werror=non-virtual-dtor -Werror=bool-operation -Wnarrowing -Wno-missing-field-initializers -Wno-type-limits -Wno-array-bounds -Wno-unknown-pragmas -Wno-unused-parameter -Wno-unused-function -Wno-unused-result -Wno-strict-overflow -Wno-strict-aliasing -Wno-stringop-overflow -Wsuggest-override -Wno-psabi -Wno-error=pedantic -Wno-error=old-style-cast -Wno-missing-braces -fdiagnostics-color=always -faligned-new -Wno-unused-but-set-variable -Wno-maybe-uninitialized -fno-math-errno -fno-trapping-math -Werror=format -Wno-stringop-overflow, FORCE_FALLBACK_CUDA_MPI=1, LAPACK_INFO=mkl, PERF_WITH_AVX=1, PERF_WITH_AVX2=1, PERF_WITH_AVX512=1, TORCH_VERSION=2.4.1, USE_CUDA=0, USE_CUDNN=OFF, USE_CUSPARSELT=OFF, USE_EXCEPTION_PTR=1, USE_GFLAGS=1, USE_GLOG=1, USE_GLOO=1, USE_MKL=ON, USE_MKLDNN=0, USE_MPI=1, USE_NCCL=1, USE_NNPACK=ON, USE_OPENMP=1, USE_ROCM=ON, USE_ROCM_KERNEL_ASSERT=OFF, \n\nTorchVision: 0.19.1\nOpenCV: 4.11.0\nMMCV: 1.6.1\nMMCV Compiler: GCC 11.4\nMMCV CUDA Compiler: rocm not available\nMMDetection: 2.26.0+c41df4b", "config": "plugin = True\nplugin_dir = 'projects/mmdet3d_plugin/'\ndist_params = dict(backend='nccl')\nlog_level = 'INFO'\nwork_dir = './work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704'\ntotal_batch_size = 160\nnum_gpus = 8\nbatch_size = 20\nnum_iters_per_epoch = 175\nnum_epochs = 100\ncheckpoint_epoch_interval = 20\ncheckpoint_config = dict(interval=3500)\nlog_config = dict(\n interval=1,\n hooks=[\n dict(type='TextLoggerHook', by_epoch=False),\n dict(type='TensorboardLoggerHook')\n ])\nload_from = None\nresume_from = None\nworkflow = [('train', 1)]\nfp16 = dict(loss_scale=32.0)\ninput_shape = (704, 256)\ntracking_test = True\ntracking_threshold = 0.2\nclass_names = [\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier',\n 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n]\nnum_classes = 10\nembed_dims = 256\nnum_groups = 8\nnum_decoder = 6\nnum_single_frame_decoder = 1\nuse_deformable_func = True\nstrides = [4, 8, 16, 32]\nnum_levels = 4\nnum_depth_layers = 3\ndrop_out = 0.1\ntemporal = True\ndecouple_attn = True\nwith_quality_estimation = True\nmodel = dict(\n type='Sparse4D',\n use_grid_mask=True,\n use_deformable_func=True,\n img_backbone=dict(\n type='ResNet',\n depth=50,\n num_stages=4,\n frozen_stages=-1,\n norm_eval=False,\n style='pytorch',\n with_cp=True,\n out_indices=(0, 1, 2, 3),\n norm_cfg=dict(type='BN', requires_grad=True),\n pretrained='ckpt/resnet50-19c8e357.pth'),\n img_neck=dict(\n type='FPN',\n num_outs=4,\n start_level=0,\n out_channels=256,\n add_extra_convs='on_output',\n relu_before_extra_convs=True,\n in_channels=[256, 512, 1024, 2048]),\n depth_branch=dict(\n type='DenseDepthNet',\n embed_dims=256,\n num_depth_layers=3,\n loss_weight=0.2),\n head=dict(\n type='Sparse4DHead',\n cls_threshold_to_reg=0.05,\n decouple_attn=True,\n instance_bank=dict(\n type='InstanceBank',\n num_anchor=900,\n embed_dims=256,\n anchor='nuscenes_kmeans900.npy',\n anchor_handler=dict(type='SparseBox3DKeyPointsGenerator'),\n num_temp_instances=600,\n confidence_decay=0.6,\n feat_grad=False),\n anchor_encoder=dict(\n type='SparseBox3DEncoder',\n vel_dims=3,\n embed_dims=[128, 32, 32, 64],\n mode='cat',\n output_fc=False,\n in_loops=1,\n out_loops=4),\n num_single_frame_decoder=1,\n operation_order=[\n 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm',\n 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm',\n 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm',\n 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm',\n 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm',\n 'deformable', 'ffn', 'norm', 'refine'\n ],\n temp_graph_model=dict(\n type='MultiheadAttention',\n embed_dims=512,\n num_heads=8,\n batch_first=True,\n dropout=0.1),\n graph_model=dict(\n type='MultiheadAttention',\n embed_dims=512,\n num_heads=8,\n batch_first=True,\n dropout=0.1),\n norm_layer=dict(type='LN', normalized_shape=256),\n ffn=dict(\n type='AsymmetricFFN',\n in_channels=512,\n pre_norm=dict(type='LN'),\n embed_dims=256,\n feedforward_channels=1024,\n num_fcs=2,\n ffn_drop=0.1,\n act_cfg=dict(type='ReLU', inplace=True)),\n deformable_model=dict(\n type='DeformableFeatureAggregation',\n embed_dims=256,\n num_groups=8,\n num_levels=4,\n num_cams=6,\n attn_drop=0.15,\n use_deformable_func=True,\n use_camera_embed=True,\n residual_mode='cat',\n kps_generator=dict(\n type='SparseBox3DKeyPointsGenerator',\n num_learnable_pts=6,\n fix_scale=[[0, 0, 0], [0.45, 0, 0], [-0.45, 0, 0],\n [0, 0.45, 0], [0, -0.45, 0], [0, 0, 0.45],\n [0, 0, -0.45]])),\n refine_layer=dict(\n type='SparseBox3DRefinementModule',\n embed_dims=256,\n num_cls=10,\n refine_yaw=True,\n with_quality_estimation=True),\n sampler=dict(\n type='SparseBox3DTarget',\n num_dn_groups=5,\n num_temp_dn_groups=3,\n dn_noise_scale=[2.0, 2.0, 2.0, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5],\n max_dn_gt=32,\n add_neg_dn=True,\n cls_weight=2.0,\n box_weight=0.25,\n reg_weights=[2.0, 2.0, 2.0, 0.5, 0.5, 0.5, 0.0, 0.0, 0.0, 0.0],\n cls_wise_reg_weights=dict(\n {9: [2.0, 2.0, 2.0, 1.0, 1.0, 1.0, 0.0, 0.0, 1.0, 1.0]})),\n loss_cls=dict(\n type='FocalLoss',\n use_sigmoid=True,\n gamma=2.0,\n alpha=0.25,\n loss_weight=2.0),\n loss_reg=dict(\n type='SparseBox3DLoss',\n loss_box=dict(type='L1Loss', loss_weight=0.25),\n loss_centerness=dict(type='CrossEntropyLoss', use_sigmoid=True),\n loss_yawness=dict(type='GaussianFocalLoss'),\n cls_allow_reverse=[5]),\n decoder=dict(type='SparseBox3DDecoder'),\n reg_weights=[2.0, 2.0, 2.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0]))\ndataset_type = 'NuScenes3DDetTrackDataset'\ndata_root = 'data/nuscenes/'\nanno_root = 'data/nuscenes_anno_pkls/'\nfile_client_args = dict(backend='disk')\nimg_norm_cfg = dict(\n mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375], to_rgb=True)\ntrain_pipeline = [\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(\n type='LoadPointsFromFile',\n coord_type='LIDAR',\n load_dim=5,\n use_dim=5,\n file_client_args=dict(backend='disk')),\n dict(type='ResizeCropFlipImage'),\n dict(type='MultiScaleDepthMapGenerator', downsample=[4, 8, 16]),\n dict(type='BBoxRotation'),\n dict(type='PhotoMetricDistortionMultiViewImage'),\n dict(\n type='NormalizeMultiviewImage',\n mean=[123.675, 116.28, 103.53],\n std=[58.395, 57.12, 57.375],\n to_rgb=True),\n dict(\n type='CircleObjectRangeFilter',\n class_dist_thred=[55, 55, 55, 55, 55, 55, 55, 55, 55, 55]),\n dict(\n type='InstanceNameFilter',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ]),\n dict(type='NuScenesSparse4DAdaptor'),\n dict(\n type='Collect',\n keys=[\n 'img', 'timestamp', 'projection_mat', 'image_wh', 'gt_depth',\n 'focal', 'gt_bboxes_3d', 'gt_labels_3d'\n ],\n meta_keys=['T_global', 'T_global_inv', 'timestamp', 'instance_id'])\n]\ntest_pipeline = [\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(type='ResizeCropFlipImage'),\n dict(\n type='NormalizeMultiviewImage',\n mean=[123.675, 116.28, 103.53],\n std=[58.395, 57.12, 57.375],\n to_rgb=True),\n dict(type='NuScenesSparse4DAdaptor'),\n dict(\n type='Collect',\n keys=['img', 'timestamp', 'projection_mat', 'image_wh'],\n meta_keys=['T_global', 'T_global_inv', 'timestamp'])\n]\ninput_modality = dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False)\ndata_basic_config = dict(\n type='NuScenes3DDetTrackDataset',\n data_root='data/nuscenes/',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier',\n 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ],\n modality=dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False),\n version='v1.0-trainval')\ndata_aug_conf = dict(\n resize_lim=(0.4, 0.47),\n final_dim=(256, 704),\n bot_pct_lim=(0.0, 0.0),\n rot_lim=(-5.4, 5.4),\n H=900,\n W=1600,\n rand_flip=True,\n rot3d_range=[-0.3925, 0.3925])\ndata = dict(\n samples_per_gpu=20,\n workers_per_gpu=20,\n train=dict(\n type='NuScenes3DDetTrackDataset',\n data_root='data/nuscenes/',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ],\n modality=dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False),\n version='v1.0-trainval',\n ann_file='data/nuscenes_anno_pkls/nuscenes_infos_train.pkl',\n pipeline=[\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(\n type='LoadPointsFromFile',\n coord_type='LIDAR',\n load_dim=5,\n use_dim=5,\n file_client_args=dict(backend='disk')),\n dict(type='ResizeCropFlipImage'),\n dict(type='MultiScaleDepthMapGenerator', downsample=[4, 8, 16]),\n dict(type='BBoxRotation'),\n dict(type='PhotoMetricDistortionMultiViewImage'),\n dict(\n type='NormalizeMultiviewImage',\n mean=[123.675, 116.28, 103.53],\n std=[58.395, 57.12, 57.375],\n to_rgb=True),\n dict(\n type='CircleObjectRangeFilter',\n class_dist_thred=[55, 55, 55, 55, 55, 55, 55, 55, 55, 55]),\n dict(\n type='InstanceNameFilter',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian',\n 'traffic_cone'\n ]),\n dict(type='NuScenesSparse4DAdaptor'),\n dict(\n type='Collect',\n keys=[\n 'img', 'timestamp', 'projection_mat', 'image_wh',\n 'gt_depth', 'focal', 'gt_bboxes_3d', 'gt_labels_3d'\n ],\n meta_keys=[\n 'T_global', 'T_global_inv', 'timestamp', 'instance_id'\n ])\n ],\n test_mode=False,\n data_aug_conf=dict(\n resize_lim=(0.4, 0.47),\n final_dim=(256, 704),\n bot_pct_lim=(0.0, 0.0),\n rot_lim=(-5.4, 5.4),\n H=900,\n W=1600,\n rand_flip=True,\n rot3d_range=[-0.3925, 0.3925]),\n with_seq_flag=True,\n sequences_split_num=2,\n keep_consistent_seq_aug=True),\n val=dict(\n type='NuScenes3DDetTrackDataset',\n data_root='data/nuscenes/',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ],\n modality=dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False),\n version='v1.0-trainval',\n ann_file='data/nuscenes_anno_pkls/nuscenes_infos_val.pkl',\n pipeline=[\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(type='ResizeCropFlipImage'),\n dict(\n type='NormalizeMultiviewImage',\n mean=[123.675, 116.28, 103.53],\n std=[58.395, 57.12, 57.375],\n to_rgb=True),\n dict(type='NuScenesSparse4DAdaptor'),\n dict(\n type='Collect',\n keys=['img', 'timestamp', 'projection_mat', 'image_wh'],\n meta_keys=['T_global', 'T_global_inv', 'timestamp'])\n ],\n data_aug_conf=dict(\n resize_lim=(0.4, 0.47),\n final_dim=(256, 704),\n bot_pct_lim=(0.0, 0.0),\n rot_lim=(-5.4, 5.4),\n H=900,\n W=1600,\n rand_flip=True,\n rot3d_range=[-0.3925, 0.3925]),\n test_mode=True,\n tracking=True,\n tracking_threshold=0.2),\n test=dict(\n type='NuScenes3DDetTrackDataset',\n data_root='data/nuscenes/',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ],\n modality=dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False),\n version='v1.0-trainval',\n ann_file='data/nuscenes_anno_pkls/nuscenes_infos_val.pkl',\n pipeline=[\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(type='ResizeCropFlipImage'),\n dict(\n type='NormalizeMultiviewImage',\n mean=[123.675, 116.28, 103.53],\n std=[58.395, 57.12, 57.375],\n to_rgb=True),\n dict(type='NuScenesSparse4DAdaptor'),\n dict(\n type='Collect',\n keys=['img', 'timestamp', 'projection_mat', 'image_wh'],\n meta_keys=['T_global', 'T_global_inv', 'timestamp'])\n ],\n data_aug_conf=dict(\n resize_lim=(0.4, 0.47),\n final_dim=(256, 704),\n bot_pct_lim=(0.0, 0.0),\n rot_lim=(-5.4, 5.4),\n H=900,\n W=1600,\n rand_flip=True,\n rot3d_range=[-0.3925, 0.3925]),\n test_mode=True,\n tracking=True,\n tracking_threshold=0.2))\noptimizer = dict(\n type='Miopen_AdamW',\n lr=0.0006,\n weight_decay=0.001,\n paramwise_cfg=dict(custom_keys=dict(img_backbone=dict(lr_mult=0.5))))\noptimizer_config = dict(grad_clip=dict(max_norm=25, norm_type=2))\nlr_config = dict(\n policy='CosineAnnealing',\n warmup='linear',\n warmup_iters=500,\n warmup_ratio=0.3333333333333333,\n min_lr_ratio=0.001)\nrunner = dict(type='IterBasedRunner', max_iters=17500)\nvis_pipeline = [\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(type='Collect', keys=['img'], meta_keys=['timestamp', 'lidar2img'])\n]\nevaluation = dict(\n interval=3500,\n pipeline=[\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(\n type='Collect', keys=['img'], meta_keys=['timestamp', 'lidar2img'])\n ])\ngpu_ids = range(0, 8)\n", "seed": 0, "exp_name": "sparse4dv3_temporal_r50_1x8_bs6_256x704.py"} diff --git a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20251112_212736.log b/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20251112_212736.log deleted file mode 100644 index 2588749729ee403f4a7da55bf9159b4a53fc0d5e..0000000000000000000000000000000000000000 --- a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20251112_212736.log +++ /dev/null @@ -1,3367 +0,0 @@ -2025-11-12 21:27:36,161 - mmdet - INFO - Environment info: ------------------------------------------------------------- -sys.platform: linux -Python: 3.10.12 (main, May 27 2025, 17:12:29) [GCC 11.4.0] -CUDA available: True -GPU 0,1,2,3,4,5,6,7: BW200, UBB BW1000 -CUDA_HOME: /opt/dtk -NVCC: Not Available -GCC: x86_64-linux-gnu-gcc (Ubuntu 11.4.0-1ubuntu1~22.04) 11.4.0 -PyTorch: 2.4.1 -PyTorch compiling details: PyTorch built with: - - GCC 10.3 - - C++ Version: 201703 - - Intel(R) Math Kernel Library Version 2020.0.4 Product Build 20200917 for Intel(R) 64 architecture applications - - OpenMP 201511 (a.k.a. OpenMP 4.5) - - LAPACK is enabled (usually provided by MKL) - - NNPACK is enabled - - CPU capability usage: AVX512 - - HIP Runtime 6.3.25211 - - MIOpen 2.17.0 - - Magma 2.8.0 - - Build settings: BLAS_INFO=mkl, BUILD_TYPE=Release, CXX_COMPILER=/opt/rh/gcc-toolset-10/root/usr/bin/c++, CXX_FLAGS= -D_GLIBCXX_USE_CXX11_ABI=1 -fvisibility-inlines-hidden -DUSE_PTHREADPOOL -DNDEBUG -DUSE_KINETO -DLIBKINETO_NOCUPTI -DUSE_FBGEMM -DUSE_PYTORCH_QNNPACK -DUSE_XNNPACK -DSYMBOLICATE_MOBILE_DEBUG_HANDLE -O2 -fPIC -Wall -Wextra -Werror=return-type -Werror=non-virtual-dtor -Werror=bool-operation -Wnarrowing -Wno-missing-field-initializers -Wno-type-limits -Wno-array-bounds -Wno-unknown-pragmas -Wno-unused-parameter -Wno-unused-function -Wno-unused-result -Wno-strict-overflow -Wno-strict-aliasing -Wno-stringop-overflow -Wsuggest-override -Wno-psabi -Wno-error=pedantic -Wno-error=old-style-cast -Wno-missing-braces -fdiagnostics-color=always -faligned-new -Wno-unused-but-set-variable -Wno-maybe-uninitialized -fno-math-errno -fno-trapping-math -Werror=format -Wno-stringop-overflow, FORCE_FALLBACK_CUDA_MPI=1, LAPACK_INFO=mkl, PERF_WITH_AVX=1, PERF_WITH_AVX2=1, PERF_WITH_AVX512=1, TORCH_VERSION=2.4.1, USE_CUDA=0, USE_CUDNN=OFF, USE_CUSPARSELT=OFF, USE_EXCEPTION_PTR=1, USE_GFLAGS=1, USE_GLOG=1, USE_GLOO=1, USE_MKL=ON, USE_MKLDNN=0, USE_MPI=1, USE_NCCL=1, USE_NNPACK=ON, USE_OPENMP=1, USE_ROCM=ON, USE_ROCM_KERNEL_ASSERT=OFF, - -TorchVision: 0.19.1 -OpenCV: 4.11.0 -MMCV: 1.6.1 -MMCV Compiler: GCC 11.4 -MMCV CUDA Compiler: rocm not available -MMDetection: 2.26.0+c41df4b ------------------------------------------------------------- - -2025-11-12 21:27:37,087 - mmdet - INFO - Distributed training: True -2025-11-12 21:27:37,800 - mmdet - INFO - Config: -plugin = True -plugin_dir = 'projects/mmdet3d_plugin/' -dist_params = dict(backend='nccl') -log_level = 'INFO' -work_dir = './work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704' -total_batch_size = 160 -num_gpus = 8 -batch_size = 20 -num_iters_per_epoch = 175 -num_epochs = 100 -checkpoint_epoch_interval = 20 -checkpoint_config = dict(interval=3500) -log_config = dict( - interval=1, - hooks=[ - dict(type='TextLoggerHook', by_epoch=False), - dict(type='TensorboardLoggerHook') - ]) -load_from = None -resume_from = None -workflow = [('train', 1)] -fp16 = dict(loss_scale=32.0) -input_shape = (704, 256) -tracking_test = True -tracking_threshold = 0.2 -class_names = [ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier', - 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' -] -num_classes = 10 -embed_dims = 256 -num_groups = 8 -num_decoder = 6 -num_single_frame_decoder = 1 -use_deformable_func = True -strides = [4, 8, 16, 32] -num_levels = 4 -num_depth_layers = 3 -drop_out = 0.1 -temporal = True -decouple_attn = True -with_quality_estimation = True -model = dict( - type='Sparse4D', - use_grid_mask=True, - use_deformable_func=True, - img_backbone=dict( - type='ResNet', - depth=50, - num_stages=4, - frozen_stages=-1, - norm_eval=False, - style='pytorch', - with_cp=True, - out_indices=(0, 1, 2, 3), - norm_cfg=dict(type='BN', requires_grad=True), - pretrained='ckpt/resnet50-19c8e357.pth'), - img_neck=dict( - type='FPN', - num_outs=4, - start_level=0, - out_channels=256, - add_extra_convs='on_output', - relu_before_extra_convs=True, - in_channels=[256, 512, 1024, 2048]), - depth_branch=dict( - type='DenseDepthNet', - embed_dims=256, - num_depth_layers=3, - loss_weight=0.2), - head=dict( - type='Sparse4DHead', - cls_threshold_to_reg=0.05, - decouple_attn=True, - instance_bank=dict( - type='InstanceBank', - num_anchor=900, - embed_dims=256, - anchor='nuscenes_kmeans900.npy', - anchor_handler=dict(type='SparseBox3DKeyPointsGenerator'), - num_temp_instances=600, - confidence_decay=0.6, - feat_grad=False), - anchor_encoder=dict( - type='SparseBox3DEncoder', - vel_dims=3, - embed_dims=[128, 32, 32, 64], - mode='cat', - output_fc=False, - in_loops=1, - out_loops=4), - num_single_frame_decoder=1, - operation_order=[ - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine' - ], - temp_graph_model=dict( - type='MultiheadAttention', - embed_dims=512, - num_heads=8, - batch_first=True, - dropout=0.1), - graph_model=dict( - type='MultiheadAttention', - embed_dims=512, - num_heads=8, - batch_first=True, - dropout=0.1), - norm_layer=dict(type='LN', normalized_shape=256), - ffn=dict( - type='AsymmetricFFN', - in_channels=512, - pre_norm=dict(type='LN'), - embed_dims=256, - feedforward_channels=1024, - num_fcs=2, - ffn_drop=0.1, - act_cfg=dict(type='ReLU', inplace=True)), - deformable_model=dict( - type='DeformableFeatureAggregation', - embed_dims=256, - num_groups=8, - num_levels=4, - num_cams=6, - attn_drop=0.15, - use_deformable_func=True, - use_camera_embed=True, - residual_mode='cat', - kps_generator=dict( - type='SparseBox3DKeyPointsGenerator', - num_learnable_pts=6, - fix_scale=[[0, 0, 0], [0.45, 0, 0], [-0.45, 0, 0], - [0, 0.45, 0], [0, -0.45, 0], [0, 0, 0.45], - [0, 0, -0.45]])), - refine_layer=dict( - type='SparseBox3DRefinementModule', - embed_dims=256, - num_cls=10, - refine_yaw=True, - with_quality_estimation=True), - sampler=dict( - type='SparseBox3DTarget', - num_dn_groups=5, - num_temp_dn_groups=3, - dn_noise_scale=[2.0, 2.0, 2.0, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5], - max_dn_gt=32, - add_neg_dn=True, - cls_weight=2.0, - box_weight=0.25, - reg_weights=[2.0, 2.0, 2.0, 0.5, 0.5, 0.5, 0.0, 0.0, 0.0, 0.0], - cls_wise_reg_weights=dict( - {9: [2.0, 2.0, 2.0, 1.0, 1.0, 1.0, 0.0, 0.0, 1.0, 1.0]})), - loss_cls=dict( - type='FocalLoss', - use_sigmoid=True, - gamma=2.0, - alpha=0.25, - loss_weight=2.0), - loss_reg=dict( - type='SparseBox3DLoss', - loss_box=dict(type='L1Loss', loss_weight=0.25), - loss_centerness=dict(type='CrossEntropyLoss', use_sigmoid=True), - loss_yawness=dict(type='GaussianFocalLoss'), - cls_allow_reverse=[5]), - decoder=dict(type='SparseBox3DDecoder'), - reg_weights=[2.0, 2.0, 2.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0])) -dataset_type = 'NuScenes3DDetTrackDataset' -data_root = 'data/nuscenes/' -anno_root = 'data/nuscenes_anno_pkls/' -file_client_args = dict(backend='disk') -img_norm_cfg = dict( - mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375], to_rgb=True) -train_pipeline = [ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict( - type='LoadPointsFromFile', - coord_type='LIDAR', - load_dim=5, - use_dim=5, - file_client_args=dict(backend='disk')), - dict(type='ResizeCropFlipImage'), - dict(type='MultiScaleDepthMapGenerator', downsample=[4, 8, 16]), - dict(type='BBoxRotation'), - dict(type='PhotoMetricDistortionMultiViewImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict( - type='CircleObjectRangeFilter', - class_dist_thred=[55, 55, 55, 55, 55, 55, 55, 55, 55, 55]), - dict( - type='InstanceNameFilter', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ]), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=[ - 'img', 'timestamp', 'projection_mat', 'image_wh', 'gt_depth', - 'focal', 'gt_bboxes_3d', 'gt_labels_3d' - ], - meta_keys=['T_global', 'T_global_inv', 'timestamp', 'instance_id']) -] -test_pipeline = [ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='ResizeCropFlipImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=['img', 'timestamp', 'projection_mat', 'image_wh'], - meta_keys=['T_global', 'T_global_inv', 'timestamp']) -] -input_modality = dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False) -data_basic_config = dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier', - 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval') -data_aug_conf = dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]) -data = dict( - samples_per_gpu=20, - workers_per_gpu=20, - train=dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval', - ann_file='data/nuscenes_anno_pkls/nuscenes_infos_train.pkl', - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict( - type='LoadPointsFromFile', - coord_type='LIDAR', - load_dim=5, - use_dim=5, - file_client_args=dict(backend='disk')), - dict(type='ResizeCropFlipImage'), - dict(type='MultiScaleDepthMapGenerator', downsample=[4, 8, 16]), - dict(type='BBoxRotation'), - dict(type='PhotoMetricDistortionMultiViewImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict( - type='CircleObjectRangeFilter', - class_dist_thred=[55, 55, 55, 55, 55, 55, 55, 55, 55, 55]), - dict( - type='InstanceNameFilter', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', - 'traffic_cone' - ]), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=[ - 'img', 'timestamp', 'projection_mat', 'image_wh', - 'gt_depth', 'focal', 'gt_bboxes_3d', 'gt_labels_3d' - ], - meta_keys=[ - 'T_global', 'T_global_inv', 'timestamp', 'instance_id' - ]) - ], - test_mode=False, - data_aug_conf=dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]), - with_seq_flag=True, - sequences_split_num=2, - keep_consistent_seq_aug=True), - val=dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval', - ann_file='data/nuscenes_anno_pkls/nuscenes_infos_val.pkl', - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='ResizeCropFlipImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=['img', 'timestamp', 'projection_mat', 'image_wh'], - meta_keys=['T_global', 'T_global_inv', 'timestamp']) - ], - data_aug_conf=dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]), - test_mode=True, - tracking=True, - tracking_threshold=0.2), - test=dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval', - ann_file='data/nuscenes_anno_pkls/nuscenes_infos_val.pkl', - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='ResizeCropFlipImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=['img', 'timestamp', 'projection_mat', 'image_wh'], - meta_keys=['T_global', 'T_global_inv', 'timestamp']) - ], - data_aug_conf=dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]), - test_mode=True, - tracking=True, - tracking_threshold=0.2)) -optimizer = dict( - type='Miopen_AdamW', - lr=0.0006, - weight_decay=0.001, - paramwise_cfg=dict(custom_keys=dict(img_backbone=dict(lr_mult=0.5)))) -optimizer_config = dict(grad_clip=dict(max_norm=25, norm_type=2)) -lr_config = dict( - policy='CosineAnnealing', - warmup='linear', - warmup_iters=500, - warmup_ratio=0.3333333333333333, - min_lr_ratio=0.001) -runner = dict(type='IterBasedRunner', max_iters=17500) -vis_pipeline = [ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='Collect', keys=['img'], meta_keys=['timestamp', 'lidar2img']) -] -evaluation = dict( - interval=3500, - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict( - type='Collect', keys=['img'], meta_keys=['timestamp', 'lidar2img']) - ]) -gpu_ids = range(0, 8) - -2025-11-12 21:27:37,800 - mmdet - INFO - Set random seed to 0, deterministic: False -2025-11-12 21:27:38,100 - mmdet - INFO - initialize ResNet with init_cfg {'type': 'Pretrained', 'checkpoint': 'ckpt/resnet50-19c8e357.pth'} -2025-11-12 21:27:38,320 - mmdet - INFO - initialize FPN with init_cfg {'type': 'Xavier', 'layer': 'Conv2d', 'distribution': 'uniform'} -Name of parameter - Initialization information - -img_backbone.conv1.weight - torch.Size([64, 3, 7, 7]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.bn1.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.bn1.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.conv1.weight - torch.Size([64, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn1.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn1.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.conv2.weight - torch.Size([64, 64, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn2.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn2.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.conv3.weight - torch.Size([256, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn3.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn3.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.downsample.0.weight - torch.Size([256, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.downsample.1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.downsample.1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.conv1.weight - torch.Size([64, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn1.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn1.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.conv2.weight - torch.Size([64, 64, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn2.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn2.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.conv3.weight - torch.Size([256, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn3.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn3.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.conv1.weight - torch.Size([64, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn1.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn1.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.conv2.weight - torch.Size([64, 64, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn2.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn2.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.conv3.weight - torch.Size([256, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn3.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn3.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.conv1.weight - torch.Size([128, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn1.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn1.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.conv2.weight - torch.Size([128, 128, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn2.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn2.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.conv3.weight - torch.Size([512, 128, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn3.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn3.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.downsample.0.weight - torch.Size([512, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.downsample.1.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.downsample.1.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.conv1.weight - torch.Size([128, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn1.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn1.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.conv2.weight - torch.Size([128, 128, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn2.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn2.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.conv3.weight - torch.Size([512, 128, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn3.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn3.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.conv1.weight - torch.Size([128, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn1.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn1.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.conv2.weight - torch.Size([128, 128, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn2.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn2.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.conv3.weight - torch.Size([512, 128, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn3.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn3.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.conv1.weight - torch.Size([128, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn1.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn1.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.conv2.weight - torch.Size([128, 128, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn2.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn2.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.conv3.weight - torch.Size([512, 128, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn3.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn3.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.conv1.weight - torch.Size([256, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.downsample.0.weight - torch.Size([1024, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.downsample.1.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.downsample.1.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.conv1.weight - torch.Size([512, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn1.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn1.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.conv2.weight - torch.Size([512, 512, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn2.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn2.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.conv3.weight - torch.Size([2048, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn3.weight - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn3.bias - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.downsample.0.weight - torch.Size([2048, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.downsample.1.weight - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.downsample.1.bias - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.conv1.weight - torch.Size([512, 2048, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn1.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn1.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.conv2.weight - torch.Size([512, 512, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn2.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn2.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.conv3.weight - torch.Size([2048, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn3.weight - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn3.bias - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.conv1.weight - torch.Size([512, 2048, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn1.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn1.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.conv2.weight - torch.Size([512, 512, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn2.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn2.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.conv3.weight - torch.Size([2048, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn3.weight - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn3.bias - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_neck.lateral_convs.0.conv.weight - torch.Size([256, 256, 1, 1]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.lateral_convs.0.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.lateral_convs.1.conv.weight - torch.Size([256, 512, 1, 1]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.lateral_convs.1.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.lateral_convs.2.conv.weight - torch.Size([256, 1024, 1, 1]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.lateral_convs.2.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.lateral_convs.3.conv.weight - torch.Size([256, 2048, 1, 1]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.lateral_convs.3.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.fpn_convs.0.conv.weight - torch.Size([256, 256, 3, 3]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.fpn_convs.0.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.fpn_convs.1.conv.weight - torch.Size([256, 256, 3, 3]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.fpn_convs.1.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.fpn_convs.2.conv.weight - torch.Size([256, 256, 3, 3]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.fpn_convs.2.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.fpn_convs.3.conv.weight - torch.Size([256, 256, 3, 3]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.fpn_convs.3.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.instance_bank.anchor - torch.Size([900, 11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.instance_bank.instance_feature - torch.Size([900, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.instance_bank.anchor_handler.fix_scale - torch.Size([1, 3]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.0.weight - torch.Size([128, 3]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.0.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.2.weight - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.2.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.3.weight - torch.Size([128, 128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.3.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.5.weight - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.5.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.6.weight - torch.Size([128, 128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.6.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.8.weight - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.8.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.9.weight - torch.Size([128, 128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.9.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.11.weight - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.11.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.0.weight - torch.Size([32, 3]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.0.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.2.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.2.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.3.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.3.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.5.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.5.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.6.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.6.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.8.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.8.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.9.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.9.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.11.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.11.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.0.weight - torch.Size([32, 2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.0.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.2.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.2.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.3.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.3.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.5.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.5.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.6.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.6.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.8.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.8.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.9.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.9.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.11.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.11.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.0.weight - torch.Size([64, 3]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.0.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.2.weight - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.2.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.3.weight - torch.Size([64, 64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.3.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.5.weight - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.5.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.6.weight - torch.Size([64, 64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.6.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.8.weight - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.8.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.9.weight - torch.Size([64, 64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.9.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.11.weight - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.11.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.1.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.1.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.1.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.1.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.1.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.1.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.1.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.1.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.3.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.4.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.4.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.4.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.4.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.5.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.5.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.5.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.5.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.6.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.6.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.8.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.8.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.8.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.8.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.8.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.8.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.8.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.8.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.10.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.11.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.11.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.11.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.11.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.12.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.12.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.12.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.12.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.13.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.13.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.15.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.15.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.15.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.15.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.15.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.15.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.15.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.15.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.16.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.16.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.17.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.18.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.18.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.18.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.18.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.19.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.19.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.19.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.19.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.20.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.20.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.22.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.22.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.22.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.22.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.22.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.22.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.22.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.22.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.23.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.23.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.24.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.25.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.25.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.25.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.25.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.26.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.26.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.26.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.26.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.27.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.27.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.29.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.29.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.29.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.29.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.29.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.29.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.29.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.29.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.30.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.30.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.31.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.32.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.32.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.32.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.32.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.33.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.33.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.33.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.33.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.34.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.34.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.36.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.36.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.36.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.36.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.36.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.36.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.36.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.36.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.37.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.37.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.38.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.fc_before.weight - torch.Size([512, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.fc_after.weight - torch.Size([256, 512]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.0.weight - torch.Size([1, 256, 1, 1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.0.bias - torch.Size([1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.1.weight - torch.Size([1, 256, 1, 1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.1.bias - torch.Size([1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.2.weight - torch.Size([1, 256, 1, 1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.2.bias - torch.Size([1]): -The value is the same before and after calling `init_weights` of Sparse4D -2025-11-12 21:27:38,411 - mmdet - INFO - Model: -Sparse4D( - (img_backbone): ResNet( - (conv1): Conv2d(3, 64, kernel_size=(7, 7), stride=(2, 2), padding=(3, 3), bias=False) - (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (maxpool): MaxPool2d(kernel_size=3, stride=2, padding=1, dilation=1, ceil_mode=False) - (layer1): ResLayer( - (0): Bottleneck( - (conv1): Conv2d(64, 64, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (downsample): Sequential( - (0): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - ) - ) - (1): Bottleneck( - (conv1): Conv2d(256, 64, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (2): Bottleneck( - (conv1): Conv2d(256, 64, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - ) - (layer2): ResLayer( - (0): Bottleneck( - (conv1): Conv2d(256, 128, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (downsample): Sequential( - (0): Conv2d(256, 512, kernel_size=(1, 1), stride=(2, 2), bias=False) - (1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - ) - ) - (1): Bottleneck( - (conv1): Conv2d(512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (2): Bottleneck( - (conv1): Conv2d(512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (3): Bottleneck( - (conv1): Conv2d(512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - ) - (layer3): ResLayer( - (0): Bottleneck( - (conv1): Conv2d(512, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (downsample): Sequential( - (0): Conv2d(512, 1024, kernel_size=(1, 1), stride=(2, 2), bias=False) - (1): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - ) - ) - (1): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (2): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (3): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (4): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (5): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - ) - (layer4): ResLayer( - (0): Bottleneck( - (conv1): Conv2d(1024, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(512, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (downsample): Sequential( - (0): Conv2d(1024, 2048, kernel_size=(1, 1), stride=(2, 2), bias=False) - (1): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - ) - ) - (1): Bottleneck( - (conv1): Conv2d(2048, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(512, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (2): Bottleneck( - (conv1): Conv2d(2048, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(512, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - ) - ) - init_cfg={'type': 'Pretrained', 'checkpoint': 'ckpt/resnet50-19c8e357.pth'} - (img_neck): FPN( - (lateral_convs): ModuleList( - (0): ConvModule( - (conv): Conv2d(256, 256, kernel_size=(1, 1), stride=(1, 1)) - ) - (1): ConvModule( - (conv): Conv2d(512, 256, kernel_size=(1, 1), stride=(1, 1)) - ) - (2): ConvModule( - (conv): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1)) - ) - (3): ConvModule( - (conv): Conv2d(2048, 256, kernel_size=(1, 1), stride=(1, 1)) - ) - ) - (fpn_convs): ModuleList( - (0-3): 4 x ConvModule( - (conv): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1)) - ) - ) - ) - init_cfg={'type': 'Xavier', 'layer': 'Conv2d', 'distribution': 'uniform'} - (head): Sparse4DHead( - (instance_bank): InstanceBank( - (anchor_handler): SparseBox3DKeyPointsGenerator() - ) - (anchor_encoder): SparseBox3DEncoder( - (pos_fc): Sequential( - (0): Linear(in_features=3, out_features=128, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((128,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=128, out_features=128, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((128,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=128, out_features=128, bias=True) - (7): ReLU(inplace=True) - (8): LayerNorm((128,), eps=1e-05, elementwise_affine=True) - (9): Linear(in_features=128, out_features=128, bias=True) - (10): ReLU(inplace=True) - (11): LayerNorm((128,), eps=1e-05, elementwise_affine=True) - ) - (size_fc): Sequential( - (0): Linear(in_features=3, out_features=32, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=32, out_features=32, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=32, out_features=32, bias=True) - (7): ReLU(inplace=True) - (8): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (9): Linear(in_features=32, out_features=32, bias=True) - (10): ReLU(inplace=True) - (11): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - ) - (yaw_fc): Sequential( - (0): Linear(in_features=2, out_features=32, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=32, out_features=32, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=32, out_features=32, bias=True) - (7): ReLU(inplace=True) - (8): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (9): Linear(in_features=32, out_features=32, bias=True) - (10): ReLU(inplace=True) - (11): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - ) - (vel_fc): Sequential( - (0): Linear(in_features=3, out_features=64, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((64,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=64, out_features=64, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((64,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=64, out_features=64, bias=True) - (7): ReLU(inplace=True) - (8): LayerNorm((64,), eps=1e-05, elementwise_affine=True) - (9): Linear(in_features=64, out_features=64, bias=True) - (10): ReLU(inplace=True) - (11): LayerNorm((64,), eps=1e-05, elementwise_affine=True) - ) - ) - (loss_cls): FocalLoss() - (loss_reg): SparseBox3DLoss( - (loss_box): L1Loss() - (loss_cns): CrossEntropyLoss(avg_non_ignore=False) - (loss_yns): GaussianFocalLoss() - ) - (layers): ModuleList( - (0): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (1): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (4-5): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (6): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (7): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (8): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (11-12): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (13): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (14): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (15): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (16): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (17): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (18-19): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (20): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (21): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (22): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (23): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (24): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (25-26): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (27): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (28): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (29): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (30): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (31): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (32-33): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (34): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (35): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (36): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (37): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (38): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - ) - (fc_before): Linear(in_features=256, out_features=512, bias=False) - (fc_after): Linear(in_features=512, out_features=256, bias=False) - ) - (depth_branch): DenseDepthNet( - (depth_layers): ModuleList( - (0-2): 3 x Conv2d(256, 1, kernel_size=(1, 1), stride=(1, 1)) - ) - ) - (grid_mask): GridMask() -) -2025-11-12 21:27:51,029 - mmdet - INFO - Start running, host: root@VM-120-96-tencentos, work_dir: /cfs/auto/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704 -2025-11-12 21:27:51,029 - mmdet - INFO - Hooks will be executed in the following order: -before_run: -(VERY_HIGH ) CosineAnnealingLrUpdaterHook -(ABOVE_NORMAL) Fp16OptimizerHook -(NORMAL ) CheckpointHook -(NORMAL ) CustomDistEvalHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -before_train_epoch: -(VERY_HIGH ) CosineAnnealingLrUpdaterHook -(NORMAL ) CustomDistEvalHook -(LOW ) IterTimerHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -before_train_iter: -(VERY_HIGH ) CosineAnnealingLrUpdaterHook -(NORMAL ) CustomDistEvalHook -(LOW ) IterTimerHook - -------------------- -after_train_iter: -(ABOVE_NORMAL) Fp16OptimizerHook -(NORMAL ) CheckpointHook -(NORMAL ) CustomDistEvalHook -(LOW ) IterTimerHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -after_train_epoch: -(NORMAL ) CheckpointHook -(NORMAL ) CustomDistEvalHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -before_val_epoch: -(LOW ) IterTimerHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -before_val_iter: -(LOW ) IterTimerHook - -------------------- -after_val_iter: -(LOW ) IterTimerHook - -------------------- -after_val_epoch: -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -after_run: -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -2025-11-12 21:27:51,029 - mmdet - INFO - workflow: [('train', 1)], max: 17500 iters -2025-11-12 21:27:51,031 - mmdet - INFO - Checkpoints will be saved to /cfs/auto/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704 by HardDiskBackend. -2025-11-12 21:29:47,555 - mmdet - INFO - Iter [1/17500] lr: 1.000e-04, eta: 23 days, 10:04:57, time: 115.635, data_time: 10.291, memory: 49167, loss_cls_0: 2.3610, loss_box_0: 0.0138, loss_cns_0: 0.0027, loss_yns_0: 0.0008, loss_cls_1: 2.1543, loss_box_1: 0.1089, loss_cns_1: 0.0247, loss_yns_1: 0.0067, loss_cls_2: 2.3122, loss_box_2: 0.0041, loss_cns_2: 0.0005, loss_yns_2: 0.0003, loss_cls_3: 2.3899, loss_box_3: 0.0295, loss_cns_3: 0.0050, loss_yns_3: 0.0014, loss_cls_4: 2.0278, loss_box_4: 0.4160, loss_cns_4: 0.0534, loss_yns_4: 0.0252, loss_cls_5: 2.4246, loss_box_5: 0.0180, loss_cns_5: 0.0022, loss_yns_5: 0.0016, loss_cls_dn_0: 1.1980, loss_box_dn_0: 1.4603, loss_cls_dn_1: 1.1102, loss_box_dn_1: 1.7318, loss_cls_dn_2: 1.1741, loss_box_dn_2: 1.9718, loss_cls_dn_3: 1.1721, loss_box_dn_3: 2.2418, loss_cls_dn_4: 1.0528, loss_box_dn_4: 2.4268, loss_cls_dn_5: 1.2386, loss_box_dn_5: 2.6772, loss_dense_depth: 1.8643, loss: 35.7047, grad_norm: 267.6892 -2025-11-12 21:29:49,538 - mmdet - INFO - Iter [2/17500] lr: 1.004e-04, eta: 11 days, 21:53:19, time: 2.001, data_time: 0.089, memory: 49167, loss_cls_0: 2.0496, loss_box_0: 0.0102, loss_cns_0: 0.0025, loss_yns_0: 0.0010, loss_cls_1: 1.9947, loss_box_1: 0.2676, loss_cns_1: 0.0489, loss_yns_1: 0.0131, loss_cls_2: 2.1007, loss_box_2: 0.2280, loss_cns_2: 0.0216, loss_yns_2: 0.0104, loss_cls_3: 1.9497, loss_box_3: 0.4134, loss_cns_3: 0.0551, loss_yns_3: 0.0170, loss_cls_4: 1.7975, loss_box_4: 1.6069, loss_cns_4: 0.1638, loss_yns_4: 0.0575, loss_cls_5: 2.0481, loss_box_5: 0.6360, loss_cns_5: 0.0705, loss_yns_5: 0.0226, loss_cls_dn_0: 1.0295, loss_box_dn_0: 1.2464, loss_cls_dn_1: 0.9495, loss_box_dn_1: 2.4110, loss_cls_dn_2: 0.9680, loss_box_dn_2: 2.5316, loss_cls_dn_3: 0.9124, loss_box_dn_3: 2.6059, loss_cls_dn_4: 0.8389, loss_box_dn_4: 2.8708, loss_cls_dn_5: 0.9856, loss_box_dn_5: 3.1053, loss_dense_depth: 1.7162, loss: 37.7574, grad_norm: 66.3681 -2025-11-12 21:29:51,043 - mmdet - INFO - Iter [3/17500] lr: 1.008e-04, eta: 8 days, 1:01:08, time: 1.504, data_time: 0.082, memory: 49167, loss_cls_0: 1.4131, loss_box_0: 2.6173, loss_cns_0: 0.6192, loss_yns_0: 0.1982, loss_cls_1: 1.7096, loss_box_1: 2.5340, loss_cns_1: 0.3695, loss_yns_1: 0.1333, loss_cls_2: 1.7469, loss_box_2: 4.4270, loss_cns_2: 0.3867, loss_yns_2: 0.2135, loss_cls_3: 1.6062, loss_box_3: 5.1886, loss_cns_3: 0.4487, loss_yns_3: 0.2184, loss_cls_4: 1.5105, loss_box_4: 5.1191, loss_cns_4: 0.4270, loss_yns_4: 0.2043, loss_cls_5: 1.6210, loss_box_5: 4.1094, loss_cns_5: 0.3001, loss_yns_5: 0.1330, loss_cls_dn_0: 0.6678, loss_box_dn_0: 1.2149, loss_cls_dn_1: 0.7984, loss_box_dn_1: 2.3815, loss_cls_dn_2: 0.7669, loss_box_dn_2: 2.5801, loss_cls_dn_3: 0.6727, loss_box_dn_3: 2.7835, loss_cls_dn_4: 0.6828, loss_box_dn_4: 3.0332, loss_cls_dn_5: 0.7734, loss_box_dn_5: 3.3118, loss_dense_depth: 1.6569, loss: 58.5785, grad_norm: 108.3686 -2025-11-12 21:29:52,560 - mmdet - INFO - Iter [4/17500] lr: 1.012e-04, eta: 6 days, 2:35:56, time: 1.517, data_time: 0.082, memory: 49167, loss_cls_0: 1.3802, loss_box_0: 2.6899, loss_cns_0: 0.5008, loss_yns_0: 0.2283, loss_cls_1: 1.5604, loss_box_1: 3.7631, loss_cns_1: 0.4305, loss_yns_1: 0.2017, loss_cls_2: 1.6940, loss_box_2: 3.7766, loss_cns_2: 0.4334, loss_yns_2: 0.1976, loss_cls_3: 1.4653, loss_box_3: 4.2331, loss_cns_3: 0.4635, loss_yns_3: 0.2020, loss_cls_4: 1.4792, loss_box_4: 4.6513, loss_cns_4: 0.4368, loss_yns_4: 0.1889, loss_cls_5: 1.4037, loss_box_5: 5.1886, loss_cns_5: 0.4498, loss_yns_5: 0.1981, loss_cls_dn_0: 0.5233, loss_box_dn_0: 1.2461, loss_cls_dn_1: 0.6623, loss_box_dn_1: 2.5621, loss_cls_dn_2: 0.6438, loss_box_dn_2: 2.6686, loss_cls_dn_3: 0.5805, loss_box_dn_3: 2.9276, loss_cls_dn_4: 0.5414, loss_box_dn_4: 3.1710, loss_cls_dn_5: 0.5970, loss_box_dn_5: 3.4158, loss_dense_depth: 1.5682, loss: 58.3245, grad_norm: 114.8815 -2025-11-12 21:29:54,093 - mmdet - INFO - Iter [5/17500] lr: 1.016e-04, eta: 4 days, 22:45:45, time: 1.533, data_time: 0.082, memory: 49167, loss_cls_0: 1.3777, loss_box_0: 2.8098, loss_cns_0: 0.5585, loss_yns_0: 0.1842, loss_cls_1: 1.5733, loss_box_1: 4.1658, loss_cns_1: 0.4177, loss_yns_1: 0.2161, loss_cls_2: 1.4688, loss_box_2: 4.1779, loss_cns_2: 0.4229, loss_yns_2: 0.1923, loss_cls_3: 1.3635, loss_box_3: 4.2147, loss_cns_3: 0.4210, loss_yns_3: 0.2059, loss_cls_4: 1.3220, loss_box_4: 4.4737, loss_cns_4: 0.3968, loss_yns_4: 0.2018, loss_cls_5: 1.3638, loss_box_5: 4.5614, loss_cns_5: 0.3858, loss_yns_5: 0.2190, loss_cls_dn_0: 0.5589, loss_box_dn_0: 1.2822, loss_cls_dn_1: 0.5963, loss_box_dn_1: 2.4317, loss_cls_dn_2: 0.6315, loss_box_dn_2: 2.5607, loss_cls_dn_3: 0.5308, loss_box_dn_3: 2.6600, loss_cls_dn_4: 0.5404, loss_box_dn_4: 2.8850, loss_cls_dn_5: 0.5092, loss_box_dn_5: 2.8676, loss_dense_depth: 2.2072, loss: 56.9559, grad_norm: 126.3227 -2025-11-12 21:29:55,654 - mmdet - INFO - Iter [6/17500] lr: 1.020e-04, eta: 4 days, 4:13:38, time: 1.561, data_time: 0.138, memory: 49167, loss_cls_0: 1.2640, loss_box_0: 2.4030, loss_cns_0: 0.6907, loss_yns_0: 0.1939, loss_cls_1: 1.3402, loss_box_1: 3.6635, loss_cns_1: 0.5122, loss_yns_1: 0.1820, loss_cls_2: 1.3441, loss_box_2: 3.6079, loss_cns_2: 0.4820, loss_yns_2: 0.1918, loss_cls_3: 1.3176, loss_box_3: 3.4914, loss_cns_3: 0.5153, loss_yns_3: 0.1868, loss_cls_4: 1.3213, loss_box_4: 3.7886, loss_cns_4: 0.5041, loss_yns_4: 0.1857, loss_cls_5: 1.3701, loss_box_5: 4.0794, loss_cns_5: 0.5161, loss_yns_5: 0.2038, loss_cls_dn_0: 0.5590, loss_box_dn_0: 1.1371, loss_cls_dn_1: 0.5254, loss_box_dn_1: 2.4720, loss_cls_dn_2: 0.5570, loss_box_dn_2: 2.3767, loss_cls_dn_3: 0.4788, loss_box_dn_3: 2.4054, loss_cls_dn_4: 0.4722, loss_box_dn_4: 2.6162, loss_cls_dn_5: 0.4355, loss_box_dn_5: 2.7385, loss_dense_depth: 1.9786, loss: 52.1080, grad_norm: 101.3703 -2025-11-12 21:29:57,173 - mmdet - INFO - Iter [7/17500] lr: 1.024e-04, eta: 3 days, 14:57:29, time: 1.518, data_time: 0.078, memory: 49167, loss_cls_0: 1.2854, loss_box_0: 2.3049, loss_cns_0: 0.6474, loss_yns_0: 0.1795, loss_cls_1: 1.2752, loss_box_1: 3.5715, loss_cns_1: 0.5134, loss_yns_1: 0.1836, loss_cls_2: 1.4198, loss_box_2: 3.6326, loss_cns_2: 0.4419, loss_yns_2: 0.1919, loss_cls_3: 1.2793, loss_box_3: 3.5811, loss_cns_3: 0.4787, loss_yns_3: 0.1850, loss_cls_4: 1.3211, loss_box_4: 3.5843, loss_cns_4: 0.4846, loss_yns_4: 0.1894, loss_cls_5: 1.3639, loss_box_5: 3.7530, loss_cns_5: 0.5403, loss_yns_5: 0.2039, loss_cls_dn_0: 0.5140, loss_box_dn_0: 1.0523, loss_cls_dn_1: 0.4639, loss_box_dn_1: 2.3015, loss_cls_dn_2: 0.4693, loss_box_dn_2: 2.2466, loss_cls_dn_3: 0.4375, loss_box_dn_3: 2.2557, loss_cls_dn_4: 0.4202, loss_box_dn_4: 2.2787, loss_cls_dn_5: 0.3908, loss_box_dn_5: 2.3667, loss_dense_depth: 2.1374, loss: 49.9465, grad_norm: 87.0230 -2025-11-12 21:29:58,698 - mmdet - INFO - Iter [8/17500] lr: 1.028e-04, eta: 3 days, 5:00:37, time: 1.525, data_time: 0.079, memory: 49167, loss_cls_0: 1.2437, loss_box_0: 2.2878, loss_cns_0: 0.6141, loss_yns_0: 0.1837, loss_cls_1: 1.2769, loss_box_1: 3.2715, loss_cns_1: 0.5313, loss_yns_1: 0.1862, loss_cls_2: 1.2937, loss_box_2: 3.3042, loss_cns_2: 0.4794, loss_yns_2: 0.1795, loss_cls_3: 1.2600, loss_box_3: 3.3404, loss_cns_3: 0.5217, loss_yns_3: 0.2085, loss_cls_4: 1.2520, loss_box_4: 3.3004, loss_cns_4: 0.5123, loss_yns_4: 0.1852, loss_cls_5: 1.2924, loss_box_5: 3.3794, loss_cns_5: 0.5476, loss_yns_5: 0.1814, loss_cls_dn_0: 0.4926, loss_box_dn_0: 1.0665, loss_cls_dn_1: 0.4594, loss_box_dn_1: 1.5048, loss_cls_dn_2: 0.5018, loss_box_dn_2: 1.5585, loss_cls_dn_3: 0.4369, loss_box_dn_3: 1.6975, loss_cls_dn_4: 0.4554, loss_box_dn_4: 1.6901, loss_cls_dn_5: 0.4422, loss_box_dn_5: 1.8282, loss_dense_depth: 2.1013, loss: 45.0686, grad_norm: 69.0308 -2025-11-12 21:30:00,235 - mmdet - INFO - Iter [9/17500] lr: 1.032e-04, eta: 2 days, 21:16:47, time: 1.538, data_time: 0.085, memory: 49167, loss_cls_0: 1.2380, loss_box_0: 2.2966, loss_cns_0: 0.6303, loss_yns_0: 0.1744, loss_cls_1: 1.2855, loss_box_1: 3.0893, loss_cns_1: 0.4618, loss_yns_1: 0.1745, loss_cls_2: 1.2810, loss_box_2: 3.0859, loss_cns_2: 0.4729, loss_yns_2: 0.1931, loss_cls_3: 1.2675, loss_box_3: 3.1982, loss_cns_3: 0.4976, loss_yns_3: 0.1872, loss_cls_4: 1.2480, loss_box_4: 3.2000, loss_cns_4: 0.4960, loss_yns_4: 0.1827, loss_cls_5: 1.2530, loss_box_5: 3.4526, loss_cns_5: 0.4718, loss_yns_5: 0.1902, loss_cls_dn_0: 0.4898, loss_box_dn_0: 1.0646, loss_cls_dn_1: 0.4220, loss_box_dn_1: 1.6844, loss_cls_dn_2: 0.4809, loss_box_dn_2: 1.7738, loss_cls_dn_3: 0.4240, loss_box_dn_3: 1.8931, loss_cls_dn_4: 0.4286, loss_box_dn_4: 1.9040, loss_cls_dn_5: 0.4595, loss_box_dn_5: 2.1430, loss_dense_depth: 1.8964, loss: 45.0920, grad_norm: 80.0242 -2025-11-12 21:30:01,776 - mmdet - INFO - Iter [10/17500] lr: 1.036e-04, eta: 2 days, 15:05:50, time: 1.542, data_time: 0.080, memory: 49167, loss_cls_0: 1.2687, loss_box_0: 2.3473, loss_cns_0: 0.6282, loss_yns_0: 0.1729, loss_cls_1: 1.2766, loss_box_1: 3.0610, loss_cns_1: 0.4857, loss_yns_1: 0.1780, loss_cls_2: 1.2600, loss_box_2: 2.9934, loss_cns_2: 0.4814, loss_yns_2: 0.1949, loss_cls_3: 1.2275, loss_box_3: 3.0524, loss_cns_3: 0.5014, loss_yns_3: 0.1919, loss_cls_4: 1.2474, loss_box_4: 3.2717, loss_cns_4: 0.4934, loss_yns_4: 0.1935, loss_cls_5: 1.2728, loss_box_5: 3.5404, loss_cns_5: 0.5024, loss_yns_5: 0.1805, loss_cls_dn_0: 0.4531, loss_box_dn_0: 1.0570, loss_cls_dn_1: 0.4001, loss_box_dn_1: 1.9026, loss_cls_dn_2: 0.4509, loss_box_dn_2: 1.8495, loss_cls_dn_3: 0.4373, loss_box_dn_3: 1.8697, loss_cls_dn_4: 0.4079, loss_box_dn_4: 1.9602, loss_cls_dn_5: 0.4403, loss_box_dn_5: 2.1012, loss_dense_depth: 2.1337, loss: 45.4870, grad_norm: 98.1957 -2025-11-12 21:30:03,285 - mmdet - INFO - Iter [11/17500] lr: 1.040e-04, eta: 2 days, 10:01:24, time: 1.507, data_time: 0.080, memory: 49167, loss_cls_0: 1.2538, loss_box_0: 2.3375, loss_cns_0: 0.5992, loss_yns_0: 0.1732, loss_cls_1: 1.2208, loss_box_1: 2.9978, loss_cns_1: 0.5393, loss_yns_1: 0.1744, loss_cls_2: 1.2242, loss_box_2: 3.0713, loss_cns_2: 0.5327, loss_yns_2: 0.1823, loss_cls_3: 1.2450, loss_box_3: 3.0779, loss_cns_3: 0.5500, loss_yns_3: 0.1786, loss_cls_4: 1.2800, loss_box_4: 3.0370, loss_cns_4: 0.5111, loss_yns_4: 0.1733, loss_cls_5: 1.2693, loss_box_5: 3.2583, loss_cns_5: 0.5118, loss_yns_5: 0.1769, loss_cls_dn_0: 0.4335, loss_box_dn_0: 1.0527, loss_cls_dn_1: 0.3950, loss_box_dn_1: 1.9904, loss_cls_dn_2: 0.4292, loss_box_dn_2: 1.9634, loss_cls_dn_3: 0.4239, loss_box_dn_3: 1.9966, loss_cls_dn_4: 0.3956, loss_box_dn_4: 2.0446, loss_cls_dn_5: 0.4182, loss_box_dn_5: 2.1253, loss_dense_depth: 1.8813, loss: 45.1256, grad_norm: 56.4927 -2025-11-12 21:30:04,813 - mmdet - INFO - Iter [12/17500] lr: 1.044e-04, eta: 2 days, 5:48:15, time: 1.530, data_time: 0.077, memory: 49167, loss_cls_0: 1.1719, loss_box_0: 2.3840, loss_cns_0: 0.5622, loss_yns_0: 0.1717, loss_cls_1: 1.2069, loss_box_1: 3.0297, loss_cns_1: 0.5202, loss_yns_1: 0.1753, loss_cls_2: 1.2355, loss_box_2: 3.2076, loss_cns_2: 0.5315, loss_yns_2: 0.1877, loss_cls_3: 1.2419, loss_box_3: 3.3504, loss_cns_3: 0.5383, loss_yns_3: 0.1764, loss_cls_4: 1.2276, loss_box_4: 3.2954, loss_cns_4: 0.5133, loss_yns_4: 0.1961, loss_cls_5: 1.2752, loss_box_5: 3.3945, loss_cns_5: 0.5254, loss_yns_5: 0.1884, loss_cls_dn_0: 0.4834, loss_box_dn_0: 1.0678, loss_cls_dn_1: 0.3939, loss_box_dn_1: 2.1769, loss_cls_dn_2: 0.3903, loss_box_dn_2: 2.2152, loss_cls_dn_3: 0.3737, loss_box_dn_3: 2.2695, loss_cls_dn_4: 0.3799, loss_box_dn_4: 2.2890, loss_cls_dn_5: 0.3971, loss_box_dn_5: 2.3224, loss_dense_depth: 1.9536, loss: 47.0195, grad_norm: 105.3454 -2025-11-12 21:30:06,338 - mmdet - INFO - Iter [13/17500] lr: 1.048e-04, eta: 2 days, 2:13:57, time: 1.525, data_time: 0.089, memory: 49167, loss_cls_0: 1.1523, loss_box_0: 2.4870, loss_cns_0: 0.5483, loss_yns_0: 0.1703, loss_cls_1: 1.2194, loss_box_1: 3.0111, loss_cns_1: 0.5263, loss_yns_1: 0.1769, loss_cls_2: 1.3087, loss_box_2: 3.1313, loss_cns_2: 0.5382, loss_yns_2: 0.2009, loss_cls_3: 1.2974, loss_box_3: 3.2386, loss_cns_3: 0.5339, loss_yns_3: 0.1924, loss_cls_4: 1.2205, loss_box_4: 3.3106, loss_cns_4: 0.5312, loss_yns_4: 0.1788, loss_cls_5: 1.2461, loss_box_5: 3.4499, loss_cns_5: 0.5383, loss_yns_5: 0.1818, loss_cls_dn_0: 0.5005, loss_box_dn_0: 1.0970, loss_cls_dn_1: 0.4448, loss_box_dn_1: 1.6445, loss_cls_dn_2: 0.4064, loss_box_dn_2: 1.7112, loss_cls_dn_3: 0.3996, loss_box_dn_3: 1.7733, loss_cls_dn_4: 0.4388, loss_box_dn_4: 1.8492, loss_cls_dn_5: 0.4511, loss_box_dn_5: 1.9146, loss_dense_depth: 1.9407, loss: 44.9620, grad_norm: 94.9055 -2025-11-12 21:30:07,852 - mmdet - INFO - Iter [14/17500] lr: 1.052e-04, eta: 1 day, 23:10:00, time: 1.513, data_time: 0.080, memory: 49167, loss_cls_0: 1.1666, loss_box_0: 2.5910, loss_cns_0: 0.5540, loss_yns_0: 0.1727, loss_cls_1: 1.2476, loss_box_1: 2.9169, loss_cns_1: 0.5695, loss_yns_1: 0.1756, loss_cls_2: 1.3144, loss_box_2: 2.9155, loss_cns_2: 0.5758, loss_yns_2: 0.1851, loss_cls_3: 1.2655, loss_box_3: 2.9073, loss_cns_3: 0.5577, loss_yns_3: 0.1797, loss_cls_4: 1.2532, loss_box_4: 2.8813, loss_cns_4: 0.5535, loss_yns_4: 0.1791, loss_cls_5: 1.2452, loss_box_5: 3.1118, loss_cns_5: 0.5310, loss_yns_5: 0.1883, loss_cls_dn_0: 0.4802, loss_box_dn_0: 1.1271, loss_cls_dn_1: 0.4663, loss_box_dn_1: 1.3704, loss_cls_dn_2: 0.4338, loss_box_dn_2: 1.3429, loss_cls_dn_3: 0.4331, loss_box_dn_3: 1.3623, loss_cls_dn_4: 0.4658, loss_box_dn_4: 1.4119, loss_cls_dn_5: 0.4689, loss_box_dn_5: 1.5205, loss_dense_depth: 1.8407, loss: 41.9622, grad_norm: 56.1500 -2025-11-12 21:30:09,375 - mmdet - INFO - Iter [15/17500] lr: 1.056e-04, eta: 1 day, 20:30:46, time: 1.523, data_time: 0.079, memory: 49167, loss_cls_0: 1.2411, loss_box_0: 2.6196, loss_cns_0: 0.5580, loss_yns_0: 0.1727, loss_cls_1: 1.2464, loss_box_1: 2.7212, loss_cns_1: 0.5968, loss_yns_1: 0.1725, loss_cls_2: 1.2844, loss_box_2: 2.7919, loss_cns_2: 0.5836, loss_yns_2: 0.1817, loss_cls_3: 1.2538, loss_box_3: 2.8483, loss_cns_3: 0.5914, loss_yns_3: 0.1828, loss_cls_4: 1.2863, loss_box_4: 2.8292, loss_cns_4: 0.6031, loss_yns_4: 0.1739, loss_cls_5: 1.2689, loss_box_5: 2.8714, loss_cns_5: 0.5820, loss_yns_5: 0.1889, loss_cls_dn_0: 0.4338, loss_box_dn_0: 1.1433, loss_cls_dn_1: 0.4513, loss_box_dn_1: 1.2684, loss_cls_dn_2: 0.4411, loss_box_dn_2: 1.3144, loss_cls_dn_3: 0.4399, loss_box_dn_3: 1.3943, loss_cls_dn_4: 0.4455, loss_box_dn_4: 1.4018, loss_cls_dn_5: 0.4440, loss_box_dn_5: 1.5224, loss_dense_depth: 1.7473, loss: 41.2974, grad_norm: 115.7359 -2025-11-12 21:30:10,907 - mmdet - INFO - Iter [16/17500] lr: 1.060e-04, eta: 1 day, 18:11:36, time: 1.532, data_time: 0.078, memory: 49167, loss_cls_0: 1.1946, loss_box_0: 2.5122, loss_cns_0: 0.5541, loss_yns_0: 0.1753, loss_cls_1: 1.2304, loss_box_1: 2.7120, loss_cns_1: 0.5786, loss_yns_1: 0.1767, loss_cls_2: 1.2728, loss_box_2: 2.8029, loss_cns_2: 0.5737, loss_yns_2: 0.2063, loss_cls_3: 1.2581, loss_box_3: 2.8293, loss_cns_3: 0.5745, loss_yns_3: 0.1829, loss_cls_4: 1.2840, loss_box_4: 2.8009, loss_cns_4: 0.5701, loss_yns_4: 0.1753, loss_cls_5: 1.2705, loss_box_5: 2.9911, loss_cns_5: 0.5620, loss_yns_5: 0.1737, loss_cls_dn_0: 0.4597, loss_box_dn_0: 1.0866, loss_cls_dn_1: 0.4586, loss_box_dn_1: 1.3209, loss_cls_dn_2: 0.4532, loss_box_dn_2: 1.4406, loss_cls_dn_3: 0.4580, loss_box_dn_3: 1.5500, loss_cls_dn_4: 0.4533, loss_box_dn_4: 1.5611, loss_cls_dn_5: 0.4543, loss_box_dn_5: 1.7747, loss_dense_depth: 1.8348, loss: 41.9678, grad_norm: 79.6951 -2025-11-12 21:30:12,439 - mmdet - INFO - Iter [17/17500] lr: 1.064e-04, eta: 1 day, 16:08:48, time: 1.532, data_time: 0.078, memory: 49167, loss_cls_0: 1.1478, loss_box_0: 2.4150, loss_cns_0: 0.5731, loss_yns_0: 0.1729, loss_cls_1: 1.2161, loss_box_1: 2.7663, loss_cns_1: 0.5486, loss_yns_1: 0.1754, loss_cls_2: 1.2475, loss_box_2: 2.8066, loss_cns_2: 0.5708, loss_yns_2: 0.1844, loss_cls_3: 1.2314, loss_box_3: 2.8695, loss_cns_3: 0.5659, loss_yns_3: 0.1767, loss_cls_4: 1.2600, loss_box_4: 2.8095, loss_cns_4: 0.5601, loss_yns_4: 0.1744, loss_cls_5: 1.2500, loss_box_5: 2.9327, loss_cns_5: 0.5417, loss_yns_5: 0.1810, loss_cls_dn_0: 0.4681, loss_box_dn_0: 1.0534, loss_cls_dn_1: 0.4525, loss_box_dn_1: 1.4999, loss_cls_dn_2: 0.4414, loss_box_dn_2: 1.6130, loss_cls_dn_3: 0.4360, loss_box_dn_3: 1.7306, loss_cls_dn_4: 0.4328, loss_box_dn_4: 1.7174, loss_cls_dn_5: 0.4260, loss_box_dn_5: 1.8797, loss_dense_depth: 1.5740, loss: 42.1021, grad_norm: 74.7798 -2025-11-12 21:30:13,953 - mmdet - INFO - Iter [18/17500] lr: 1.068e-04, eta: 1 day, 14:19:22, time: 1.515, data_time: 0.073, memory: 49167, loss_cls_0: 1.1421, loss_box_0: 2.2901, loss_cns_0: 0.6007, loss_yns_0: 0.1721, loss_cls_1: 1.2295, loss_box_1: 2.7427, loss_cns_1: 0.5470, loss_yns_1: 0.1760, loss_cls_2: 1.2491, loss_box_2: 2.7815, loss_cns_2: 0.5669, loss_yns_2: 0.1763, loss_cls_3: 1.2742, loss_box_3: 2.7441, loss_cns_3: 0.5764, loss_yns_3: 0.1747, loss_cls_4: 1.2619, loss_box_4: 2.6907, loss_cns_4: 0.5572, loss_yns_4: 0.1750, loss_cls_5: 1.2761, loss_box_5: 2.6921, loss_cns_5: 0.5647, loss_yns_5: 0.1816, loss_cls_dn_0: 0.4742, loss_box_dn_0: 1.0270, loss_cls_dn_1: 0.4304, loss_box_dn_1: 1.4964, loss_cls_dn_2: 0.4362, loss_box_dn_2: 1.5828, loss_cls_dn_3: 0.4075, loss_box_dn_3: 1.6625, loss_cls_dn_4: 0.4135, loss_box_dn_4: 1.6488, loss_cls_dn_5: 0.4089, loss_box_dn_5: 1.7213, loss_dense_depth: 1.5206, loss: 41.0728, grad_norm: 66.5065 -2025-11-12 21:30:15,464 - mmdet - INFO - Iter [19/17500] lr: 1.072e-04, eta: 1 day, 12:41:23, time: 1.510, data_time: 0.080, memory: 49167, loss_cls_0: 1.1471, loss_box_0: 2.2575, loss_cns_0: 0.5996, loss_yns_0: 0.1696, loss_cls_1: 1.2959, loss_box_1: 2.7430, loss_cns_1: 0.5411, loss_yns_1: 0.1758, loss_cls_2: 1.2728, loss_box_2: 2.6301, loss_cns_2: 0.5625, loss_yns_2: 0.1738, loss_cls_3: 1.3190, loss_box_3: 2.6111, loss_cns_3: 0.5846, loss_yns_3: 0.1746, loss_cls_4: 1.2694, loss_box_4: 2.8122, loss_cns_4: 0.5799, loss_yns_4: 0.1716, loss_cls_5: 1.2863, loss_box_5: 2.7016, loss_cns_5: 0.5899, loss_yns_5: 0.1750, loss_cls_dn_0: 0.4735, loss_box_dn_0: 1.0344, loss_cls_dn_1: 0.4121, loss_box_dn_1: 1.2493, loss_cls_dn_2: 0.4442, loss_box_dn_2: 1.2503, loss_cls_dn_3: 0.4246, loss_box_dn_3: 1.3102, loss_cls_dn_4: 0.4351, loss_box_dn_4: 1.4463, loss_cls_dn_5: 0.4339, loss_box_dn_5: 1.4242, loss_dense_depth: 1.4426, loss: 39.6247, grad_norm: 78.1580 -2025-11-12 21:30:16,973 - mmdet - INFO - Iter [20/17500] lr: 1.076e-04, eta: 1 day, 11:13:12, time: 1.510, data_time: 0.082, memory: 49167, loss_cls_0: 1.1528, loss_box_0: 2.2442, loss_cns_0: 0.6059, loss_yns_0: 0.1692, loss_cls_1: 1.2822, loss_box_1: 2.8704, loss_cns_1: 0.5718, loss_yns_1: 0.1784, loss_cls_2: 1.2522, loss_box_2: 2.6847, loss_cns_2: 0.6016, loss_yns_2: 0.1759, loss_cls_3: 1.2502, loss_box_3: 2.7891, loss_cns_3: 0.5828, loss_yns_3: 0.1762, loss_cls_4: 1.2480, loss_box_4: 3.0489, loss_cns_4: 0.5808, loss_yns_4: 0.1733, loss_cls_5: 1.2569, loss_box_5: 3.2271, loss_cns_5: 0.5469, loss_yns_5: 0.1802, loss_cls_dn_0: 0.4585, loss_box_dn_0: 1.0253, loss_cls_dn_1: 0.3838, loss_box_dn_1: 1.3063, loss_cls_dn_2: 0.4207, loss_box_dn_2: 1.2584, loss_cls_dn_3: 0.4308, loss_box_dn_3: 1.3089, loss_cls_dn_4: 0.4278, loss_box_dn_4: 1.4402, loss_cls_dn_5: 0.4261, loss_box_dn_5: 1.4686, loss_dense_depth: 1.4210, loss: 40.6265, grad_norm: 73.0247 -2025-11-12 21:30:18,603 - mmdet - INFO - Iter [21/17500] lr: 1.080e-04, eta: 1 day, 9:55:04, time: 1.630, data_time: 0.112, memory: 49167, loss_cls_0: 1.1410, loss_box_0: 2.2045, loss_cns_0: 0.6167, loss_yns_0: 0.1683, loss_cls_1: 1.2318, loss_box_1: 3.0090, loss_cns_1: 0.5512, loss_yns_1: 0.1762, loss_cls_2: 1.2542, loss_box_2: 2.9306, loss_cns_2: 0.5439, loss_yns_2: 0.1761, loss_cls_3: 1.2572, loss_box_3: 3.0573, loss_cns_3: 0.5140, loss_yns_3: 0.1785, loss_cls_4: 1.2454, loss_box_4: 3.2745, loss_cns_4: 0.4882, loss_yns_4: 0.1709, loss_cls_5: 1.2613, loss_box_5: 3.4865, loss_cns_5: 0.4642, loss_yns_5: 0.1794, loss_cls_dn_0: 0.4526, loss_box_dn_0: 1.0405, loss_cls_dn_1: 0.4270, loss_box_dn_1: 1.3878, loss_cls_dn_2: 0.4584, loss_box_dn_2: 1.4172, loss_cls_dn_3: 0.4839, loss_box_dn_3: 1.5215, loss_cls_dn_4: 0.4707, loss_box_dn_4: 1.7322, loss_cls_dn_5: 0.4687, loss_box_dn_5: 1.8355, loss_dense_depth: 1.3141, loss: 42.5908, grad_norm: 99.5158 -2025-11-12 21:30:20,200 - mmdet - INFO - Iter [22/17500] lr: 1.084e-04, eta: 1 day, 8:43:36, time: 1.598, data_time: 0.075, memory: 49167, loss_cls_0: 1.1493, loss_box_0: 2.1524, loss_cns_0: 0.6234, loss_yns_0: 0.1686, loss_cls_1: 1.2334, loss_box_1: 3.0137, loss_cns_1: 0.5243, loss_yns_1: 0.1699, loss_cls_2: 1.2665, loss_box_2: 2.9229, loss_cns_2: 0.5356, loss_yns_2: 0.1760, loss_cls_3: 1.2644, loss_box_3: 2.8970, loss_cns_3: 0.5328, loss_yns_3: 0.1752, loss_cls_4: 1.2384, loss_box_4: 2.9448, loss_cns_4: 0.5299, loss_yns_4: 0.1701, loss_cls_5: 1.2628, loss_box_5: 2.9657, loss_cns_5: 0.5153, loss_yns_5: 0.1702, loss_cls_dn_0: 0.4519, loss_box_dn_0: 1.0209, loss_cls_dn_1: 0.4509, loss_box_dn_1: 1.6049, loss_cls_dn_2: 0.4716, loss_box_dn_2: 1.6404, loss_cls_dn_3: 0.4948, loss_box_dn_3: 1.6970, loss_cls_dn_4: 0.4692, loss_box_dn_4: 1.8663, loss_cls_dn_5: 0.4798, loss_box_dn_5: 1.8999, loss_dense_depth: 1.2828, loss: 42.4330, grad_norm: 66.8701 -2025-11-12 21:30:21,759 - mmdet - INFO - Iter [23/17500] lr: 1.088e-04, eta: 1 day, 7:37:51, time: 1.557, data_time: 0.108, memory: 49167, loss_cls_0: 1.1314, loss_box_0: 2.0894, loss_cns_0: 0.6236, loss_yns_0: 0.1701, loss_cls_1: 1.2205, loss_box_1: 3.0267, loss_cns_1: 0.5068, loss_yns_1: 0.1728, loss_cls_2: 1.2352, loss_box_2: 3.0106, loss_cns_2: 0.5187, loss_yns_2: 0.1744, loss_cls_3: 1.2439, loss_box_3: 2.9609, loss_cns_3: 0.5156, loss_yns_3: 0.1735, loss_cls_4: 1.2437, loss_box_4: 2.9817, loss_cns_4: 0.5160, loss_yns_4: 0.1733, loss_cls_5: 1.2457, loss_box_5: 2.9688, loss_cns_5: 0.5174, loss_yns_5: 0.1701, loss_cls_dn_0: 0.4533, loss_box_dn_0: 0.9861, loss_cls_dn_1: 0.4468, loss_box_dn_1: 1.5650, loss_cls_dn_2: 0.4556, loss_box_dn_2: 1.6104, loss_cls_dn_3: 0.4661, loss_box_dn_3: 1.6194, loss_cls_dn_4: 0.4286, loss_box_dn_4: 1.7318, loss_cls_dn_5: 0.4588, loss_box_dn_5: 1.7567, loss_dense_depth: 1.2861, loss: 41.8558, grad_norm: 68.0117 -2025-11-12 21:30:23,284 - mmdet - INFO - Iter [24/17500] lr: 1.092e-04, eta: 1 day, 6:37:11, time: 1.526, data_time: 0.076, memory: 49167, loss_cls_0: 1.1525, loss_box_0: 2.0700, loss_cns_0: 0.6174, loss_yns_0: 0.1681, loss_cls_1: 1.2235, loss_box_1: 2.9363, loss_cns_1: 0.5230, loss_yns_1: 0.1703, loss_cls_2: 1.2465, loss_box_2: 2.9982, loss_cns_2: 0.5295, loss_yns_2: 0.1746, loss_cls_3: 1.2586, loss_box_3: 3.0347, loss_cns_3: 0.4931, loss_yns_3: 0.1721, loss_cls_4: 1.3160, loss_box_4: 3.0949, loss_cns_4: 0.4797, loss_yns_4: 0.1724, loss_cls_5: 1.2632, loss_box_5: 3.1651, loss_cns_5: 0.4564, loss_yns_5: 0.1803, loss_cls_dn_0: 0.4623, loss_box_dn_0: 1.0057, loss_cls_dn_1: 0.4180, loss_box_dn_1: 1.6280, loss_cls_dn_2: 0.4170, loss_box_dn_2: 1.6909, loss_cls_dn_3: 0.4215, loss_box_dn_3: 1.6859, loss_cls_dn_4: 0.3894, loss_box_dn_4: 1.7586, loss_cls_dn_5: 0.4196, loss_box_dn_5: 1.8146, loss_dense_depth: 1.3095, loss: 42.3175, grad_norm: 74.8554 -2025-11-12 21:30:24,815 - mmdet - INFO - Iter [25/17500] lr: 1.096e-04, eta: 1 day, 5:41:25, time: 1.530, data_time: 0.077, memory: 49167, loss_cls_0: 1.1475, loss_box_0: 2.1002, loss_cns_0: 0.6124, loss_yns_0: 0.1675, loss_cls_1: 1.2277, loss_box_1: 2.8601, loss_cns_1: 0.5532, loss_yns_1: 0.1691, loss_cls_2: 1.2928, loss_box_2: 2.8895, loss_cns_2: 0.5528, loss_yns_2: 0.1750, loss_cls_3: 1.2857, loss_box_3: 2.9841, loss_cns_3: 0.5191, loss_yns_3: 0.1746, loss_cls_4: 1.2836, loss_box_4: 3.0832, loss_cns_4: 0.5084, loss_yns_4: 0.1775, loss_cls_5: 1.2787, loss_box_5: 3.1822, loss_cns_5: 0.4759, loss_yns_5: 0.1813, loss_cls_dn_0: 0.4653, loss_box_dn_0: 1.0130, loss_cls_dn_1: 0.3921, loss_box_dn_1: 1.5335, loss_cls_dn_2: 0.3909, loss_box_dn_2: 1.5614, loss_cls_dn_3: 0.3944, loss_box_dn_3: 1.5625, loss_cls_dn_4: 0.3847, loss_box_dn_4: 1.6089, loss_cls_dn_5: 0.3965, loss_box_dn_5: 1.6711, loss_dense_depth: 1.2654, loss: 41.5217, grad_norm: 68.1637 -2025-11-12 21:30:26,369 - mmdet - INFO - Iter [26/17500] lr: 1.100e-04, eta: 1 day, 4:50:13, time: 1.555, data_time: 0.108, memory: 49167, loss_cls_0: 1.1236, loss_box_0: 2.1531, loss_cns_0: 0.6049, loss_yns_0: 0.1709, loss_cls_1: 1.2364, loss_box_1: 2.7819, loss_cns_1: 0.5774, loss_yns_1: 0.1723, loss_cls_2: 1.2712, loss_box_2: 2.8121, loss_cns_2: 0.5628, loss_yns_2: 0.1702, loss_cls_3: 1.2796, loss_box_3: 2.7912, loss_cns_3: 0.5952, loss_yns_3: 0.1777, loss_cls_4: 1.2398, loss_box_4: 2.8817, loss_cns_4: 0.5688, loss_yns_4: 0.1745, loss_cls_5: 1.2758, loss_box_5: 2.9532, loss_cns_5: 0.5890, loss_yns_5: 0.1819, loss_cls_dn_0: 0.4603, loss_box_dn_0: 1.0463, loss_cls_dn_1: 0.4079, loss_box_dn_1: 1.1707, loss_cls_dn_2: 0.4199, loss_box_dn_2: 1.1830, loss_cls_dn_3: 0.4181, loss_box_dn_3: 1.2075, loss_cls_dn_4: 0.4273, loss_box_dn_4: 1.2278, loss_cls_dn_5: 0.4198, loss_box_dn_5: 1.3090, loss_dense_depth: 1.2373, loss: 39.2800, grad_norm: 64.9983 -2025-11-12 21:30:27,899 - mmdet - INFO - Iter [27/17500] lr: 1.104e-04, eta: 1 day, 4:02:33, time: 1.530, data_time: 0.082, memory: 49167, loss_cls_0: 1.1253, loss_box_0: 2.2554, loss_cns_0: 0.5927, loss_yns_0: 0.1717, loss_cls_1: 1.2021, loss_box_1: 2.9121, loss_cns_1: 0.5341, loss_yns_1: 0.1693, loss_cls_2: 1.1993, loss_box_2: 3.1098, loss_cns_2: 0.5148, loss_yns_2: 0.1743, loss_cls_3: 1.2373, loss_box_3: 2.9850, loss_cns_3: 0.5282, loss_yns_3: 0.1720, loss_cls_4: 1.2128, loss_box_4: 3.0246, loss_cns_4: 0.5082, loss_yns_4: 0.1789, loss_cls_5: 1.2399, loss_box_5: 2.9891, loss_cns_5: 0.5377, loss_yns_5: 0.1804, loss_cls_dn_0: 0.4526, loss_box_dn_0: 1.0743, loss_cls_dn_1: 0.4134, loss_box_dn_1: 1.2886, loss_cls_dn_2: 0.4432, loss_box_dn_2: 1.3111, loss_cls_dn_3: 0.4318, loss_box_dn_3: 1.3450, loss_cls_dn_4: 0.4577, loss_box_dn_4: 1.3413, loss_cls_dn_5: 0.4286, loss_box_dn_5: 1.3984, loss_dense_depth: 1.2231, loss: 40.3639, grad_norm: 90.5355 -2025-11-12 21:30:29,417 - mmdet - INFO - Iter [28/17500] lr: 1.108e-04, eta: 1 day, 3:18:08, time: 1.517, data_time: 0.077, memory: 49167, loss_cls_0: 1.1238, loss_box_0: 2.3058, loss_cns_0: 0.5979, loss_yns_0: 0.1711, loss_cls_1: 1.1591, loss_box_1: 2.8568, loss_cns_1: 0.5344, loss_yns_1: 0.1702, loss_cls_2: 1.1946, loss_box_2: 3.0237, loss_cns_2: 0.5119, loss_yns_2: 0.1818, loss_cls_3: 1.2160, loss_box_3: 2.8459, loss_cns_3: 0.5382, loss_yns_3: 0.1744, loss_cls_4: 1.2059, loss_box_4: 2.8256, loss_cns_4: 0.5307, loss_yns_4: 0.1730, loss_cls_5: 1.2190, loss_box_5: 2.8168, loss_cns_5: 0.5504, loss_yns_5: 0.1766, loss_cls_dn_0: 0.4235, loss_box_dn_0: 1.0999, loss_cls_dn_1: 0.4233, loss_box_dn_1: 1.4224, loss_cls_dn_2: 0.4599, loss_box_dn_2: 1.4690, loss_cls_dn_3: 0.4320, loss_box_dn_3: 1.5134, loss_cls_dn_4: 0.4661, loss_box_dn_4: 1.5106, loss_cls_dn_5: 0.4279, loss_box_dn_5: 1.5816, loss_dense_depth: 1.2237, loss: 40.5567, grad_norm: 90.6187 -2025-11-12 21:30:30,939 - mmdet - INFO - Iter [29/17500] lr: 1.112e-04, eta: 1 day, 2:36:51, time: 1.522, data_time: 0.090, memory: 49167, loss_cls_0: 1.1347, loss_box_0: 2.2303, loss_cns_0: 0.6122, loss_yns_0: 0.1703, loss_cls_1: 1.1529, loss_box_1: 2.7915, loss_cns_1: 0.5682, loss_yns_1: 0.1738, loss_cls_2: 1.1928, loss_box_2: 2.8346, loss_cns_2: 0.5634, loss_yns_2: 0.1715, loss_cls_3: 1.2007, loss_box_3: 2.6502, loss_cns_3: 0.5734, loss_yns_3: 0.1733, loss_cls_4: 1.1832, loss_box_4: 2.6391, loss_cns_4: 0.5715, loss_yns_4: 0.1748, loss_cls_5: 1.2029, loss_box_5: 2.6651, loss_cns_5: 0.5632, loss_yns_5: 0.1732, loss_cls_dn_0: 0.4094, loss_box_dn_0: 1.0907, loss_cls_dn_1: 0.4420, loss_box_dn_1: 1.4181, loss_cls_dn_2: 0.4695, loss_box_dn_2: 1.4416, loss_cls_dn_3: 0.4381, loss_box_dn_3: 1.4633, loss_cls_dn_4: 0.4617, loss_box_dn_4: 1.4854, loss_cls_dn_5: 0.4391, loss_box_dn_5: 1.5742, loss_dense_depth: 1.1923, loss: 39.6922, grad_norm: 94.9877 -2025-11-12 21:30:32,479 - mmdet - INFO - Iter [30/17500] lr: 1.116e-04, eta: 1 day, 1:58:29, time: 1.541, data_time: 0.078, memory: 49167, loss_cls_0: 1.1132, loss_box_0: 2.0799, loss_cns_0: 0.6228, loss_yns_0: 0.1718, loss_cls_1: 1.1322, loss_box_1: 2.6830, loss_cns_1: 0.5733, loss_yns_1: 0.1758, loss_cls_2: 1.1696, loss_box_2: 2.6476, loss_cns_2: 0.5984, loss_yns_2: 0.1748, loss_cls_3: 1.1793, loss_box_3: 2.6000, loss_cns_3: 0.5687, loss_yns_3: 0.1708, loss_cls_4: 1.1676, loss_box_4: 2.6160, loss_cns_4: 0.5683, loss_yns_4: 0.1756, loss_cls_5: 1.1825, loss_box_5: 2.7046, loss_cns_5: 0.5509, loss_yns_5: 0.1703, loss_cls_dn_0: 0.4132, loss_box_dn_0: 1.0572, loss_cls_dn_1: 0.4404, loss_box_dn_1: 1.5390, loss_cls_dn_2: 0.4518, loss_box_dn_2: 1.5264, loss_cls_dn_3: 0.4214, loss_box_dn_3: 1.5535, loss_cls_dn_4: 0.4243, loss_box_dn_4: 1.5956, loss_cls_dn_5: 0.4270, loss_box_dn_5: 1.6989, loss_dense_depth: 1.1426, loss: 39.4882, grad_norm: 77.9685 -2025-11-12 21:30:33,991 - mmdet - INFO - Iter [31/17500] lr: 1.120e-04, eta: 1 day, 1:22:19, time: 1.511, data_time: 0.078, memory: 49167, loss_cls_0: 1.0646, loss_box_0: 2.0642, loss_cns_0: 0.6218, loss_yns_0: 0.1726, loss_cls_1: 1.1190, loss_box_1: 2.5530, loss_cns_1: 0.5695, loss_yns_1: 0.1693, loss_cls_2: 1.1520, loss_box_2: 2.5007, loss_cns_2: 0.6016, loss_yns_2: 0.1790, loss_cls_3: 1.1709, loss_box_3: 2.6020, loss_cns_3: 0.5617, loss_yns_3: 0.1737, loss_cls_4: 1.2169, loss_box_4: 2.5629, loss_cns_4: 0.5799, loss_yns_4: 0.1783, loss_cls_5: 1.1629, loss_box_5: 2.6104, loss_cns_5: 0.5721, loss_yns_5: 0.1733, loss_cls_dn_0: 0.4426, loss_box_dn_0: 1.0258, loss_cls_dn_1: 0.4274, loss_box_dn_1: 1.5015, loss_cls_dn_2: 0.4232, loss_box_dn_2: 1.4726, loss_cls_dn_3: 0.4071, loss_box_dn_3: 1.5248, loss_cls_dn_4: 0.3896, loss_box_dn_4: 1.5360, loss_cls_dn_5: 0.4188, loss_box_dn_5: 1.6116, loss_dense_depth: 1.1252, loss: 38.6386, grad_norm: 65.3174 -2025-11-12 21:30:35,524 - mmdet - INFO - Iter [32/17500] lr: 1.124e-04, eta: 1 day, 0:48:34, time: 1.529, data_time: 0.076, memory: 49167, loss_cls_0: 1.0400, loss_box_0: 2.1181, loss_cns_0: 0.6155, loss_yns_0: 0.1701, loss_cls_1: 1.0930, loss_box_1: 2.5831, loss_cns_1: 0.5673, loss_yns_1: 0.1699, loss_cls_2: 1.1455, loss_box_2: 2.4907, loss_cns_2: 0.5949, loss_yns_2: 0.1760, loss_cls_3: 1.1605, loss_box_3: 2.5509, loss_cns_3: 0.5837, loss_yns_3: 0.1760, loss_cls_4: 1.1926, loss_box_4: 2.4975, loss_cns_4: 0.5914, loss_yns_4: 0.1721, loss_cls_5: 1.1459, loss_box_5: 2.4815, loss_cns_5: 0.5962, loss_yns_5: 0.1756, loss_cls_dn_0: 0.4567, loss_box_dn_0: 1.0284, loss_cls_dn_1: 0.3999, loss_box_dn_1: 1.3487, loss_cls_dn_2: 0.3832, loss_box_dn_2: 1.3215, loss_cls_dn_3: 0.3901, loss_box_dn_3: 1.3343, loss_cls_dn_4: 0.3798, loss_box_dn_4: 1.3136, loss_cls_dn_5: 0.4071, loss_box_dn_5: 1.3388, loss_dense_depth: 1.0581, loss: 37.2483, grad_norm: 53.6501 -2025-11-12 21:30:37,049 - mmdet - INFO - Iter [33/17500] lr: 1.128e-04, eta: 1 day, 0:16:52, time: 1.528, data_time: 0.086, memory: 49167, loss_cls_0: 1.0486, loss_box_0: 2.1099, loss_cns_0: 0.6164, loss_yns_0: 0.1696, loss_cls_1: 1.1080, loss_box_1: 2.6071, loss_cns_1: 0.5649, loss_yns_1: 0.1729, loss_cls_2: 1.2025, loss_box_2: 2.5475, loss_cns_2: 0.5815, loss_yns_2: 0.1756, loss_cls_3: 1.1656, loss_box_3: 2.5379, loss_cns_3: 0.5904, loss_yns_3: 0.1743, loss_cls_4: 1.1398, loss_box_4: 2.5814, loss_cns_4: 0.5803, loss_yns_4: 0.1771, loss_cls_5: 1.1489, loss_box_5: 2.6655, loss_cns_5: 0.5739, loss_yns_5: 0.1745, loss_cls_dn_0: 0.4526, loss_box_dn_0: 1.0248, loss_cls_dn_1: 0.3625, loss_box_dn_1: 1.4175, loss_cls_dn_2: 0.3531, loss_box_dn_2: 1.3669, loss_cls_dn_3: 0.3670, loss_box_dn_3: 1.3258, loss_cls_dn_4: 0.3835, loss_box_dn_4: 1.3271, loss_cls_dn_5: 0.3811, loss_box_dn_5: 1.3702, loss_dense_depth: 1.0697, loss: 37.6158, grad_norm: 56.6414 -2025-11-12 21:30:38,556 - mmdet - INFO - Iter [34/17500] lr: 1.132e-04, eta: 23:46:50, time: 1.507, data_time: 0.073, memory: 49167, loss_cls_0: 1.0561, loss_box_0: 2.0783, loss_cns_0: 0.6166, loss_yns_0: 0.1705, loss_cls_1: 1.1240, loss_box_1: 2.6220, loss_cns_1: 0.5604, loss_yns_1: 0.1740, loss_cls_2: 1.1915, loss_box_2: 2.5650, loss_cns_2: 0.5853, loss_yns_2: 0.1797, loss_cls_3: 1.1538, loss_box_3: 2.5339, loss_cns_3: 0.5887, loss_yns_3: 0.1789, loss_cls_4: 1.1322, loss_box_4: 2.6145, loss_cns_4: 0.5850, loss_yns_4: 0.1760, loss_cls_5: 1.1472, loss_box_5: 2.6980, loss_cns_5: 0.5729, loss_yns_5: 0.1725, loss_cls_dn_0: 0.4296, loss_box_dn_0: 1.0097, loss_cls_dn_1: 0.3427, loss_box_dn_1: 1.4860, loss_cls_dn_2: 0.3487, loss_box_dn_2: 1.4172, loss_cls_dn_3: 0.3646, loss_box_dn_3: 1.3885, loss_cls_dn_4: 0.3928, loss_box_dn_4: 1.4348, loss_cls_dn_5: 0.3752, loss_box_dn_5: 1.4906, loss_dense_depth: 1.0644, loss: 38.0218, grad_norm: 54.7594 -2025-11-12 21:30:40,071 - mmdet - INFO - Iter [35/17500] lr: 1.136e-04, eta: 23:18:36, time: 1.516, data_time: 0.080, memory: 49167, loss_cls_0: 1.0886, loss_box_0: 2.0250, loss_cns_0: 0.6224, loss_yns_0: 0.1720, loss_cls_1: 1.1521, loss_box_1: 2.5913, loss_cns_1: 0.5552, loss_yns_1: 0.1705, loss_cls_2: 1.1470, loss_box_2: 2.5060, loss_cns_2: 0.5906, loss_yns_2: 0.1760, loss_cls_3: 1.1483, loss_box_3: 2.5514, loss_cns_3: 0.5830, loss_yns_3: 0.1721, loss_cls_4: 1.1438, loss_box_4: 2.6038, loss_cns_4: 0.5800, loss_yns_4: 0.1744, loss_cls_5: 1.1572, loss_box_5: 2.6120, loss_cns_5: 0.5837, loss_yns_5: 0.1776, loss_cls_dn_0: 0.4036, loss_box_dn_0: 0.9933, loss_cls_dn_1: 0.3358, loss_box_dn_1: 1.5539, loss_cls_dn_2: 0.3635, loss_box_dn_2: 1.5128, loss_cls_dn_3: 0.3717, loss_box_dn_3: 1.5600, loss_cls_dn_4: 0.3944, loss_box_dn_4: 1.6412, loss_cls_dn_5: 0.3721, loss_box_dn_5: 1.7036, loss_dense_depth: 1.0776, loss: 38.5676, grad_norm: 67.9903 -2025-11-12 21:30:41,620 - mmdet - INFO - Iter [36/17500] lr: 1.140e-04, eta: 22:52:04, time: 1.533, data_time: 0.080, memory: 49167, loss_cls_0: 1.0924, loss_box_0: 2.0085, loss_cns_0: 0.6225, loss_yns_0: 0.1716, loss_cls_1: 1.1299, loss_box_1: 2.7283, loss_cns_1: 0.5479, loss_yns_1: 0.1714, loss_cls_2: 1.1169, loss_box_2: 2.6153, loss_cns_2: 0.5871, loss_yns_2: 0.1728, loss_cls_3: 1.1325, loss_box_3: 2.6482, loss_cns_3: 0.5747, loss_yns_3: 0.1716, loss_cls_4: 1.1289, loss_box_4: 2.6675, loss_cns_4: 0.5735, loss_yns_4: 0.1785, loss_cls_5: 1.1586, loss_box_5: 2.6809, loss_cns_5: 0.5801, loss_yns_5: 0.1766, loss_cls_dn_0: 0.3924, loss_box_dn_0: 0.9756, loss_cls_dn_1: 0.3551, loss_box_dn_1: 1.4028, loss_cls_dn_2: 0.4049, loss_box_dn_2: 1.4053, loss_cls_dn_3: 0.4032, loss_box_dn_3: 1.4860, loss_cls_dn_4: 0.4095, loss_box_dn_4: 1.5967, loss_cls_dn_5: 0.3917, loss_box_dn_5: 1.6870, loss_dense_depth: 1.0353, loss: 38.5817, grad_norm: 72.2057 -2025-11-12 21:30:43,131 - mmdet - INFO - Iter [37/17500] lr: 1.144e-04, eta: 22:26:51, time: 1.519, data_time: 0.089, memory: 49167, loss_cls_0: 1.0644, loss_box_0: 2.0148, loss_cns_0: 0.6233, loss_yns_0: 0.1724, loss_cls_1: 1.1180, loss_box_1: 2.7830, loss_cns_1: 0.5614, loss_yns_1: 0.1732, loss_cls_2: 1.1224, loss_box_2: 2.6971, loss_cns_2: 0.5973, loss_yns_2: 0.1765, loss_cls_3: 1.1338, loss_box_3: 2.6547, loss_cns_3: 0.5902, loss_yns_3: 0.1754, loss_cls_4: 1.1288, loss_box_4: 2.6797, loss_cns_4: 0.5910, loss_yns_4: 0.1721, loss_cls_5: 1.1700, loss_box_5: 2.7487, loss_cns_5: 0.6060, loss_yns_5: 0.1696, loss_cls_dn_0: 0.4055, loss_box_dn_0: 0.9780, loss_cls_dn_1: 0.3489, loss_box_dn_1: 1.5668, loss_cls_dn_2: 0.4084, loss_box_dn_2: 1.5437, loss_cls_dn_3: 0.3896, loss_box_dn_3: 1.5645, loss_cls_dn_4: 0.3863, loss_box_dn_4: 1.6434, loss_cls_dn_5: 0.3762, loss_box_dn_5: 1.7271, loss_dense_depth: 1.0753, loss: 39.3371, grad_norm: 63.5421 -2025-11-12 21:30:44,679 - mmdet - INFO - Iter [38/17500] lr: 1.148e-04, eta: 22:03:15, time: 1.557, data_time: 0.083, memory: 49167, loss_cls_0: 1.0339, loss_box_0: 2.0152, loss_cns_0: 0.6234, loss_yns_0: 0.1697, loss_cls_1: 1.0883, loss_box_1: 2.6544, loss_cns_1: 0.5685, loss_yns_1: 0.1705, loss_cls_2: 1.1109, loss_box_2: 2.6159, loss_cns_2: 0.5944, loss_yns_2: 0.1741, loss_cls_3: 1.1259, loss_box_3: 2.6224, loss_cns_3: 0.5901, loss_yns_3: 0.1703, loss_cls_4: 1.1261, loss_box_4: 2.6523, loss_cns_4: 0.5874, loss_yns_4: 0.1669, loss_cls_5: 1.1411, loss_box_5: 2.7444, loss_cns_5: 0.6020, loss_yns_5: 0.1733, loss_cls_dn_0: 0.4220, loss_box_dn_0: 0.9835, loss_cls_dn_1: 0.3594, loss_box_dn_1: 1.3724, loss_cls_dn_2: 0.4072, loss_box_dn_2: 1.3135, loss_cls_dn_3: 0.3807, loss_box_dn_3: 1.3113, loss_cls_dn_4: 0.3722, loss_box_dn_4: 1.3526, loss_cls_dn_5: 0.3790, loss_box_dn_5: 1.4130, loss_dense_depth: 1.0614, loss: 37.6495, grad_norm: 69.6355 -2025-11-12 21:30:46,202 - mmdet - INFO - Iter [39/17500] lr: 1.152e-04, eta: 21:40:36, time: 1.521, data_time: 0.076, memory: 49167, loss_cls_0: 1.0480, loss_box_0: 1.9949, loss_cns_0: 0.6212, loss_yns_0: 0.1668, loss_cls_1: 1.1374, loss_box_1: 2.4763, loss_cns_1: 0.5919, loss_yns_1: 0.1707, loss_cls_2: 1.1245, loss_box_2: 2.4566, loss_cns_2: 0.6077, loss_yns_2: 0.1717, loss_cls_3: 1.1538, loss_box_3: 2.4643, loss_cns_3: 0.6131, loss_yns_3: 0.1701, loss_cls_4: 1.1686, loss_box_4: 2.4642, loss_cns_4: 0.6090, loss_yns_4: 0.1725, loss_cls_5: 1.1464, loss_box_5: 2.5382, loss_cns_5: 0.6233, loss_yns_5: 0.1757, loss_cls_dn_0: 0.4352, loss_box_dn_0: 0.9770, loss_cls_dn_1: 0.3630, loss_box_dn_1: 1.1534, loss_cls_dn_2: 0.4000, loss_box_dn_2: 1.0883, loss_cls_dn_3: 0.3736, loss_box_dn_3: 1.0743, loss_cls_dn_4: 0.3598, loss_box_dn_4: 1.0866, loss_cls_dn_5: 0.3880, loss_box_dn_5: 1.1222, loss_dense_depth: 1.0301, loss: 35.7186, grad_norm: 60.8788 -2025-11-12 21:30:47,729 - mmdet - INFO - Iter [40/17500] lr: 1.156e-04, eta: 21:19:08, time: 1.527, data_time: 0.084, memory: 49167, loss_cls_0: 1.0503, loss_box_0: 1.9891, loss_cns_0: 0.6224, loss_yns_0: 0.1660, loss_cls_1: 1.1387, loss_box_1: 2.3253, loss_cns_1: 0.6090, loss_yns_1: 0.1702, loss_cls_2: 1.1359, loss_box_2: 2.3409, loss_cns_2: 0.6169, loss_yns_2: 0.1687, loss_cls_3: 1.1653, loss_box_3: 2.2632, loss_cns_3: 0.6287, loss_yns_3: 0.1710, loss_cls_4: 1.1910, loss_box_4: 2.2443, loss_cns_4: 0.6375, loss_yns_4: 0.1713, loss_cls_5: 1.1760, loss_box_5: 2.2666, loss_cns_5: 0.6386, loss_yns_5: 0.1716, loss_cls_dn_0: 0.4266, loss_box_dn_0: 0.9802, loss_cls_dn_1: 0.3583, loss_box_dn_1: 1.0919, loss_cls_dn_2: 0.3834, loss_box_dn_2: 1.0451, loss_cls_dn_3: 0.3652, loss_box_dn_3: 1.0179, loss_cls_dn_4: 0.3514, loss_box_dn_4: 1.0088, loss_cls_dn_5: 0.3900, loss_box_dn_5: 1.0242, loss_dense_depth: 1.0627, loss: 34.5641, grad_norm: 48.2083 -2025-11-12 21:30:49,325 - mmdet - INFO - Iter [41/17500] lr: 1.160e-04, eta: 20:59:11, time: 1.596, data_time: 0.114, memory: 49167, loss_cls_0: 1.0201, loss_box_0: 1.9583, loss_cns_0: 0.6216, loss_yns_0: 0.1626, loss_cls_1: 1.0948, loss_box_1: 2.2204, loss_cns_1: 0.6175, loss_yns_1: 0.1677, loss_cls_2: 1.1253, loss_box_2: 2.2616, loss_cns_2: 0.6217, loss_yns_2: 0.1664, loss_cls_3: 1.1351, loss_box_3: 2.2048, loss_cns_3: 0.6284, loss_yns_3: 0.1711, loss_cls_4: 1.1672, loss_box_4: 2.2311, loss_cns_4: 0.6384, loss_yns_4: 0.1661, loss_cls_5: 1.1445, loss_box_5: 2.2173, loss_cns_5: 0.6326, loss_yns_5: 0.1676, loss_cls_dn_0: 0.4080, loss_box_dn_0: 0.9686, loss_cls_dn_1: 0.3347, loss_box_dn_1: 1.0340, loss_cls_dn_2: 0.3507, loss_box_dn_2: 1.0100, loss_cls_dn_3: 0.3535, loss_box_dn_3: 0.9975, loss_cls_dn_4: 0.3426, loss_box_dn_4: 1.0100, loss_cls_dn_5: 0.3689, loss_box_dn_5: 1.0268, loss_dense_depth: 1.0205, loss: 33.7680, grad_norm: 59.4485 -2025-11-12 21:30:50,952 - mmdet - INFO - Iter [42/17500] lr: 1.164e-04, eta: 20:40:24, time: 1.627, data_time: 0.075, memory: 49167, loss_cls_0: 1.0588, loss_box_0: 1.9309, loss_cns_0: 0.6218, loss_yns_0: 0.1629, loss_cls_1: 1.1339, loss_box_1: 2.2629, loss_cns_1: 0.6122, loss_yns_1: 0.1681, loss_cls_2: 1.1754, loss_box_2: 2.2359, loss_cns_2: 0.6307, loss_yns_2: 0.1646, loss_cls_3: 1.1325, loss_box_3: 2.2689, loss_cns_3: 0.6262, loss_yns_3: 0.1679, loss_cls_4: 1.1435, loss_box_4: 2.3014, loss_cns_4: 0.6261, loss_yns_4: 0.1671, loss_cls_5: 1.1369, loss_box_5: 2.3283, loss_cns_5: 0.6311, loss_yns_5: 0.1710, loss_cls_dn_0: 0.3940, loss_box_dn_0: 0.9643, loss_cls_dn_1: 0.3409, loss_box_dn_1: 1.0288, loss_cls_dn_2: 0.3467, loss_box_dn_2: 1.0110, loss_cls_dn_3: 0.3766, loss_box_dn_3: 1.0610, loss_cls_dn_4: 0.3683, loss_box_dn_4: 1.0893, loss_cls_dn_5: 0.3851, loss_box_dn_5: 1.1484, loss_dense_depth: 1.0577, loss: 34.4311, grad_norm: 67.5417 -2025-11-12 21:30:52,509 - mmdet - INFO - Iter [43/17500] lr: 1.168e-04, eta: 20:22:01, time: 1.556, data_time: 0.108, memory: 49167, loss_cls_0: 1.0837, loss_box_0: 1.9453, loss_cns_0: 0.6183, loss_yns_0: 0.1652, loss_cls_1: 1.1856, loss_box_1: 2.3941, loss_cns_1: 0.5984, loss_yns_1: 0.1683, loss_cls_2: 1.2207, loss_box_2: 2.2916, loss_cns_2: 0.6197, loss_yns_2: 0.1669, loss_cls_3: 1.1670, loss_box_3: 2.3349, loss_cns_3: 0.6205, loss_yns_3: 0.1672, loss_cls_4: 1.1661, loss_box_4: 2.3262, loss_cns_4: 0.6198, loss_yns_4: 0.1682, loss_cls_5: 1.1919, loss_box_5: 2.4003, loss_cns_5: 0.6381, loss_yns_5: 0.1745, loss_cls_dn_0: 0.4080, loss_box_dn_0: 0.9546, loss_cls_dn_1: 0.3652, loss_box_dn_1: 1.0698, loss_cls_dn_2: 0.3636, loss_box_dn_2: 1.0561, loss_cls_dn_3: 0.4066, loss_box_dn_3: 1.1181, loss_cls_dn_4: 0.4097, loss_box_dn_4: 1.1305, loss_cls_dn_5: 0.4112, loss_box_dn_5: 1.2171, loss_dense_depth: 1.0187, loss: 35.3617, grad_norm: 62.8160 -2025-11-12 21:30:54,052 - mmdet - INFO - Iter [44/17500] lr: 1.172e-04, eta: 20:04:22, time: 1.542, data_time: 0.082, memory: 49167, loss_cls_0: 1.0497, loss_box_0: 1.9972, loss_cns_0: 0.6138, loss_yns_0: 0.1670, loss_cls_1: 1.1355, loss_box_1: 2.4197, loss_cns_1: 0.5995, loss_yns_1: 0.1681, loss_cls_2: 1.1311, loss_box_2: 2.3341, loss_cns_2: 0.6115, loss_yns_2: 0.1672, loss_cls_3: 1.1385, loss_box_3: 2.3338, loss_cns_3: 0.6157, loss_yns_3: 0.1666, loss_cls_4: 1.1421, loss_box_4: 2.2998, loss_cns_4: 0.6237, loss_yns_4: 0.1689, loss_cls_5: 1.1626, loss_box_5: 2.3580, loss_cns_5: 0.6337, loss_yns_5: 0.1738, loss_cls_dn_0: 0.4282, loss_box_dn_0: 0.9521, loss_cls_dn_1: 0.3736, loss_box_dn_1: 1.1453, loss_cls_dn_2: 0.3810, loss_box_dn_2: 1.1277, loss_cls_dn_3: 0.4142, loss_box_dn_3: 1.1684, loss_cls_dn_4: 0.4287, loss_box_dn_4: 1.1732, loss_cls_dn_5: 0.4132, loss_box_dn_5: 1.2439, loss_dense_depth: 1.0634, loss: 35.5243, grad_norm: 58.8151 -2025-11-12 21:30:55,580 - mmdet - INFO - Iter [45/17500] lr: 1.176e-04, eta: 19:47:26, time: 1.530, data_time: 0.081, memory: 49167, loss_cls_0: 1.0607, loss_box_0: 2.0066, loss_cns_0: 0.6124, loss_yns_0: 0.1676, loss_cls_1: 1.1105, loss_box_1: 2.3576, loss_cns_1: 0.5975, loss_yns_1: 0.1719, loss_cls_2: 1.1225, loss_box_2: 2.2637, loss_cns_2: 0.6148, loss_yns_2: 0.1709, loss_cls_3: 1.1446, loss_box_3: 2.2561, loss_cns_3: 0.6182, loss_yns_3: 0.1661, loss_cls_4: 1.1389, loss_box_4: 2.2708, loss_cns_4: 0.6260, loss_yns_4: 0.1666, loss_cls_5: 1.1428, loss_box_5: 2.2873, loss_cns_5: 0.6272, loss_yns_5: 0.1690, loss_cls_dn_0: 0.4421, loss_box_dn_0: 0.9558, loss_cls_dn_1: 0.3790, loss_box_dn_1: 1.1463, loss_cls_dn_2: 0.3940, loss_box_dn_2: 1.1315, loss_cls_dn_3: 0.4118, loss_box_dn_3: 1.1673, loss_cls_dn_4: 0.4327, loss_box_dn_4: 1.1817, loss_cls_dn_5: 0.4114, loss_box_dn_5: 1.2312, loss_dense_depth: 1.0629, loss: 35.2179, grad_norm: 63.5439 -2025-11-12 21:30:57,154 - mmdet - INFO - Iter [46/17500] lr: 1.180e-04, eta: 19:31:30, time: 1.573, data_time: 0.111, memory: 49167, loss_cls_0: 1.0572, loss_box_0: 1.9958, loss_cns_0: 0.6110, loss_yns_0: 0.1655, loss_cls_1: 1.1161, loss_box_1: 2.4176, loss_cns_1: 0.5920, loss_yns_1: 0.1671, loss_cls_2: 1.1315, loss_box_2: 2.3006, loss_cns_2: 0.6178, loss_yns_2: 0.1677, loss_cls_3: 1.1411, loss_box_3: 2.3039, loss_cns_3: 0.6253, loss_yns_3: 0.1628, loss_cls_4: 1.1315, loss_box_4: 2.3124, loss_cns_4: 0.6273, loss_yns_4: 0.1694, loss_cls_5: 1.1381, loss_box_5: 2.3492, loss_cns_5: 0.6273, loss_yns_5: 0.1672, loss_cls_dn_0: 0.4393, loss_box_dn_0: 0.9425, loss_cls_dn_1: 0.3641, loss_box_dn_1: 1.1327, loss_cls_dn_2: 0.3894, loss_box_dn_2: 1.0876, loss_cls_dn_3: 0.3873, loss_box_dn_3: 1.1137, loss_cls_dn_4: 0.4093, loss_box_dn_4: 1.1185, loss_cls_dn_5: 0.3955, loss_box_dn_5: 1.1633, loss_dense_depth: 1.0333, loss: 35.0720, grad_norm: 59.1166 -2025-11-12 21:30:58,681 - mmdet - INFO - Iter [47/17500] lr: 1.184e-04, eta: 19:15:58, time: 1.528, data_time: 0.078, memory: 49167, loss_cls_0: 1.0447, loss_box_0: 1.9863, loss_cns_0: 0.6134, loss_yns_0: 0.1675, loss_cls_1: 1.1722, loss_box_1: 2.3963, loss_cns_1: 0.6010, loss_yns_1: 0.1687, loss_cls_2: 1.1522, loss_box_2: 2.3031, loss_cns_2: 0.6248, loss_yns_2: 0.1662, loss_cls_3: 1.1439, loss_box_3: 2.2900, loss_cns_3: 0.6330, loss_yns_3: 0.1607, loss_cls_4: 1.1439, loss_box_4: 2.2866, loss_cns_4: 0.6347, loss_yns_4: 0.1689, loss_cls_5: 1.1586, loss_box_5: 2.3289, loss_cns_5: 0.6329, loss_yns_5: 0.1670, loss_cls_dn_0: 0.4319, loss_box_dn_0: 0.9411, loss_cls_dn_1: 0.3538, loss_box_dn_1: 1.0757, loss_cls_dn_2: 0.3877, loss_box_dn_2: 1.0253, loss_cls_dn_3: 0.3715, loss_box_dn_3: 1.0297, loss_cls_dn_4: 0.3847, loss_box_dn_4: 1.0241, loss_cls_dn_5: 0.3896, loss_box_dn_5: 1.0541, loss_dense_depth: 1.0424, loss: 34.6570, grad_norm: 46.6776 -2025-11-12 21:31:00,216 - mmdet - INFO - Iter [48/17500] lr: 1.188e-04, eta: 19:01:07, time: 1.535, data_time: 0.078, memory: 49167, loss_cls_0: 1.0457, loss_box_0: 1.9723, loss_cns_0: 0.6095, loss_yns_0: 0.1675, loss_cls_1: 1.1255, loss_box_1: 2.4347, loss_cns_1: 0.5819, loss_yns_1: 0.1681, loss_cls_2: 1.1265, loss_box_2: 2.3342, loss_cns_2: 0.6196, loss_yns_2: 0.1638, loss_cls_3: 1.1493, loss_box_3: 2.3122, loss_cns_3: 0.6291, loss_yns_3: 0.1619, loss_cls_4: 1.1517, loss_box_4: 2.3211, loss_cns_4: 0.6320, loss_yns_4: 0.1675, loss_cls_5: 1.1482, loss_box_5: 2.3601, loss_cns_5: 0.6309, loss_yns_5: 0.1661, loss_cls_dn_0: 0.4032, loss_box_dn_0: 0.9297, loss_cls_dn_1: 0.3328, loss_box_dn_1: 1.1207, loss_cls_dn_2: 0.3839, loss_box_dn_2: 1.0402, loss_cls_dn_3: 0.3524, loss_box_dn_3: 1.0262, loss_cls_dn_4: 0.3522, loss_box_dn_4: 1.0214, loss_cls_dn_5: 0.3730, loss_box_dn_5: 1.0419, loss_dense_depth: 1.1767, loss: 34.7338, grad_norm: 49.9471 -2025-11-12 21:31:01,782 - mmdet - INFO - Iter [49/17500] lr: 1.192e-04, eta: 18:47:03, time: 1.566, data_time: 0.091, memory: 49167, loss_cls_0: 1.0992, loss_box_0: 1.9773, loss_cns_0: 0.6102, loss_yns_0: 0.1712, loss_cls_1: 1.1124, loss_box_1: 2.4580, loss_cns_1: 0.5899, loss_yns_1: 0.1651, loss_cls_2: 1.1165, loss_box_2: 2.3756, loss_cns_2: 0.6208, loss_yns_2: 0.1679, loss_cls_3: 1.1769, loss_box_3: 2.3963, loss_cns_3: 0.6200, loss_yns_3: 0.1631, loss_cls_4: 1.1843, loss_box_4: 2.3802, loss_cns_4: 0.6245, loss_yns_4: 0.1679, loss_cls_5: 1.1488, loss_box_5: 2.3700, loss_cns_5: 0.6266, loss_yns_5: 0.1638, loss_cls_dn_0: 0.3863, loss_box_dn_0: 0.9254, loss_cls_dn_1: 0.3231, loss_box_dn_1: 1.1215, loss_cls_dn_2: 0.3731, loss_box_dn_2: 1.0557, loss_cls_dn_3: 0.3417, loss_box_dn_3: 1.0587, loss_cls_dn_4: 0.3355, loss_box_dn_4: 1.0584, loss_cls_dn_5: 0.3617, loss_box_dn_5: 1.0719, loss_dense_depth: 1.1708, loss: 35.0702, grad_norm: 59.7755 -2025-11-12 21:31:03,323 - mmdet - INFO - Iter [50/17500] lr: 1.196e-04, eta: 18:33:25, time: 1.542, data_time: 0.075, memory: 49167, loss_cls_0: 1.0612, loss_box_0: 2.0040, loss_cns_0: 0.6121, loss_yns_0: 0.1670, loss_cls_1: 1.1078, loss_box_1: 2.5322, loss_cns_1: 0.5932, loss_yns_1: 0.1640, loss_cls_2: 1.1132, loss_box_2: 2.4510, loss_cns_2: 0.6177, loss_yns_2: 0.1643, loss_cls_3: 1.1535, loss_box_3: 2.4813, loss_cns_3: 0.6137, loss_yns_3: 0.1624, loss_cls_4: 1.1539, loss_box_4: 2.4461, loss_cns_4: 0.6186, loss_yns_4: 0.1641, loss_cls_5: 1.1565, loss_box_5: 2.4362, loss_cns_5: 0.6203, loss_yns_5: 0.1618, loss_cls_dn_0: 0.3810, loss_box_dn_0: 0.9360, loss_cls_dn_1: 0.3326, loss_box_dn_1: 1.1266, loss_cls_dn_2: 0.3709, loss_box_dn_2: 1.0736, loss_cls_dn_3: 0.3512, loss_box_dn_3: 1.0992, loss_cls_dn_4: 0.3446, loss_box_dn_4: 1.1205, loss_cls_dn_5: 0.3649, loss_box_dn_5: 1.1562, loss_dense_depth: 1.1176, loss: 35.5311, grad_norm: 63.7948 -2025-11-12 21:31:04,838 - mmdet - INFO - Iter [51/17500] lr: 1.200e-04, eta: 18:20:10, time: 1.514, data_time: 0.076, memory: 49167, loss_cls_0: 1.0326, loss_box_0: 2.0291, loss_cns_0: 0.6081, loss_yns_0: 0.1650, loss_cls_1: 1.0713, loss_box_1: 2.4933, loss_cns_1: 0.5962, loss_yns_1: 0.1642, loss_cls_2: 1.1081, loss_box_2: 2.3954, loss_cns_2: 0.6195, loss_yns_2: 0.1631, loss_cls_3: 1.1054, loss_box_3: 2.4289, loss_cns_3: 0.6218, loss_yns_3: 0.1658, loss_cls_4: 1.1101, loss_box_4: 2.4266, loss_cns_4: 0.6243, loss_yns_4: 0.1649, loss_cls_5: 1.1524, loss_box_5: 2.4548, loss_cns_5: 0.6258, loss_yns_5: 0.1632, loss_cls_dn_0: 0.3922, loss_box_dn_0: 0.9370, loss_cls_dn_1: 0.3201, loss_box_dn_1: 1.1864, loss_cls_dn_2: 0.3494, loss_box_dn_2: 1.1038, loss_cls_dn_3: 0.3537, loss_box_dn_3: 1.1229, loss_cls_dn_4: 0.3502, loss_box_dn_4: 1.1515, loss_cls_dn_5: 0.3513, loss_box_dn_5: 1.1971, loss_dense_depth: 0.9497, loss: 35.2551, grad_norm: 52.5811 -2025-11-12 21:31:06,354 - mmdet - INFO - Iter [52/17500] lr: 1.204e-04, eta: 18:07:25, time: 1.515, data_time: 0.078, memory: 49167, loss_cls_0: 1.0288, loss_box_0: 2.0388, loss_cns_0: 0.6015, loss_yns_0: 0.1659, loss_cls_1: 1.0652, loss_box_1: 2.4356, loss_cns_1: 0.5952, loss_yns_1: 0.1658, loss_cls_2: 1.1188, loss_box_2: 2.3886, loss_cns_2: 0.6191, loss_yns_2: 0.1639, loss_cls_3: 1.1181, loss_box_3: 2.4097, loss_cns_3: 0.6225, loss_yns_3: 0.1651, loss_cls_4: 1.1101, loss_box_4: 2.3966, loss_cns_4: 0.6220, loss_yns_4: 0.1652, loss_cls_5: 1.1287, loss_box_5: 2.4204, loss_cns_5: 0.6256, loss_yns_5: 0.1632, loss_cls_dn_0: 0.4143, loss_box_dn_0: 0.9380, loss_cls_dn_1: 0.3200, loss_box_dn_1: 1.1394, loss_cls_dn_2: 0.3352, loss_box_dn_2: 1.0702, loss_cls_dn_3: 0.3623, loss_box_dn_3: 1.0860, loss_cls_dn_4: 0.3664, loss_box_dn_4: 1.1097, loss_cls_dn_5: 0.3568, loss_box_dn_5: 1.1530, loss_dense_depth: 1.1471, loss: 35.1326, grad_norm: 55.1888 -2025-11-12 21:31:07,887 - mmdet - INFO - Iter [53/17500] lr: 1.208e-04, eta: 17:55:15, time: 1.534, data_time: 0.078, memory: 49167, loss_cls_0: 1.0340, loss_box_0: 2.0406, loss_cns_0: 0.6060, loss_yns_0: 0.1668, loss_cls_1: 1.0729, loss_box_1: 2.5123, loss_cns_1: 0.5883, loss_yns_1: 0.1646, loss_cls_2: 1.1507, loss_box_2: 2.4682, loss_cns_2: 0.6165, loss_yns_2: 0.1628, loss_cls_3: 1.1526, loss_box_3: 2.4608, loss_cns_3: 0.6204, loss_yns_3: 0.1664, loss_cls_4: 1.1341, loss_box_4: 2.4378, loss_cns_4: 0.6183, loss_yns_4: 0.1671, loss_cls_5: 1.1372, loss_box_5: 2.4241, loss_cns_5: 0.6225, loss_yns_5: 0.1646, loss_cls_dn_0: 0.4269, loss_box_dn_0: 0.9289, loss_cls_dn_1: 0.3186, loss_box_dn_1: 1.0884, loss_cls_dn_2: 0.3243, loss_box_dn_2: 1.0389, loss_cls_dn_3: 0.3645, loss_box_dn_3: 1.0404, loss_cls_dn_4: 0.3721, loss_box_dn_4: 1.0605, loss_cls_dn_5: 0.3693, loss_box_dn_5: 1.0828, loss_dense_depth: 1.0739, loss: 35.1790, grad_norm: 59.1450 -2025-11-12 21:31:09,423 - mmdet - INFO - Iter [54/17500] lr: 1.212e-04, eta: 17:43:32, time: 1.533, data_time: 0.074, memory: 49167, loss_cls_0: 1.0082, loss_box_0: 1.9529, loss_cns_0: 0.6188, loss_yns_0: 0.1622, loss_cls_1: 1.0697, loss_box_1: 2.3705, loss_cns_1: 0.6050, loss_yns_1: 0.1609, loss_cls_2: 1.1446, loss_box_2: 2.3451, loss_cns_2: 0.6280, loss_yns_2: 0.1599, loss_cls_3: 1.1222, loss_box_3: 2.2976, loss_cns_3: 0.6330, loss_yns_3: 0.1626, loss_cls_4: 1.1197, loss_box_4: 2.2646, loss_cns_4: 0.6340, loss_yns_4: 0.1649, loss_cls_5: 1.1352, loss_box_5: 2.2583, loss_cns_5: 0.6361, loss_yns_5: 0.1609, loss_cls_dn_0: 0.4098, loss_box_dn_0: 0.9155, loss_cls_dn_1: 0.3050, loss_box_dn_1: 1.0830, loss_cls_dn_2: 0.3096, loss_box_dn_2: 1.0464, loss_cls_dn_3: 0.3497, loss_box_dn_3: 1.0259, loss_cls_dn_4: 0.3511, loss_box_dn_4: 1.0327, loss_cls_dn_5: 0.3653, loss_box_dn_5: 1.0458, loss_dense_depth: 1.1207, loss: 34.1754, grad_norm: 54.1458 -2025-11-12 21:31:10,968 - mmdet - INFO - Iter [55/17500] lr: 1.216e-04, eta: 17:32:19, time: 1.546, data_time: 0.079, memory: 49167, loss_cls_0: 1.0111, loss_box_0: 1.9476, loss_cns_0: 0.6140, loss_yns_0: 0.1641, loss_cls_1: 1.0576, loss_box_1: 2.2848, loss_cns_1: 0.6124, loss_yns_1: 0.1642, loss_cls_2: 1.1251, loss_box_2: 2.2746, loss_cns_2: 0.6302, loss_yns_2: 0.1649, loss_cls_3: 1.0986, loss_box_3: 2.2282, loss_cns_3: 0.6319, loss_yns_3: 0.1695, loss_cls_4: 1.1373, loss_box_4: 2.2054, loss_cns_4: 0.6378, loss_yns_4: 0.1722, loss_cls_5: 1.1283, loss_box_5: 2.2077, loss_cns_5: 0.6347, loss_yns_5: 0.1670, loss_cls_dn_0: 0.4075, loss_box_dn_0: 0.8991, loss_cls_dn_1: 0.3311, loss_box_dn_1: 0.9843, loss_cls_dn_2: 0.3361, loss_box_dn_2: 0.9612, loss_cls_dn_3: 0.3673, loss_box_dn_3: 0.9387, loss_cls_dn_4: 0.3576, loss_box_dn_4: 0.9447, loss_cls_dn_5: 0.4003, loss_box_dn_5: 0.9527, loss_dense_depth: 1.0696, loss: 33.4193, grad_norm: 61.5542 -2025-11-12 21:31:12,515 - mmdet - INFO - Iter [56/17500] lr: 1.220e-04, eta: 17:21:30, time: 1.548, data_time: 0.080, memory: 49167, loss_cls_0: 0.9863, loss_box_0: 1.9008, loss_cns_0: 0.6094, loss_yns_0: 0.1631, loss_cls_1: 1.0403, loss_box_1: 2.1834, loss_cns_1: 0.6082, loss_yns_1: 0.1648, loss_cls_2: 1.1069, loss_box_2: 2.1648, loss_cns_2: 0.6310, loss_yns_2: 0.1654, loss_cls_3: 1.1384, loss_box_3: 2.1418, loss_cns_3: 0.6311, loss_yns_3: 0.1688, loss_cls_4: 1.1633, loss_box_4: 2.1163, loss_cns_4: 0.6390, loss_yns_4: 0.1764, loss_cls_5: 1.1050, loss_box_5: 2.0865, loss_cns_5: 0.6352, loss_yns_5: 0.1685, loss_cls_dn_0: 0.3918, loss_box_dn_0: 0.8970, loss_cls_dn_1: 0.3175, loss_box_dn_1: 0.9777, loss_cls_dn_2: 0.3364, loss_box_dn_2: 0.9598, loss_cls_dn_3: 0.3448, loss_box_dn_3: 0.9494, loss_cls_dn_4: 0.3358, loss_box_dn_4: 0.9512, loss_cls_dn_5: 0.3903, loss_box_dn_5: 0.9426, loss_dense_depth: 1.0187, loss: 32.7077, grad_norm: 62.4904 -2025-11-12 21:31:14,064 - mmdet - INFO - Iter [57/17500] lr: 1.224e-04, eta: 17:11:04, time: 1.548, data_time: 0.078, memory: 49167, loss_cls_0: 1.0007, loss_box_0: 1.8714, loss_cns_0: 0.6107, loss_yns_0: 0.1622, loss_cls_1: 1.0509, loss_box_1: 2.1810, loss_cns_1: 0.6117, loss_yns_1: 0.1662, loss_cls_2: 1.0958, loss_box_2: 2.1648, loss_cns_2: 0.6316, loss_yns_2: 0.1677, loss_cls_3: 1.1796, loss_box_3: 2.1862, loss_cns_3: 0.6313, loss_yns_3: 0.1703, loss_cls_4: 1.1903, loss_box_4: 2.1483, loss_cns_4: 0.6391, loss_yns_4: 0.1754, loss_cls_5: 1.1034, loss_box_5: 2.1130, loss_cns_5: 0.6357, loss_yns_5: 0.1690, loss_cls_dn_0: 0.3821, loss_box_dn_0: 0.8925, loss_cls_dn_1: 0.3130, loss_box_dn_1: 0.9423, loss_cls_dn_2: 0.3412, loss_box_dn_2: 0.9263, loss_cls_dn_3: 0.3335, loss_box_dn_3: 0.9488, loss_cls_dn_4: 0.3296, loss_box_dn_4: 0.9424, loss_cls_dn_5: 0.3786, loss_box_dn_5: 0.9377, loss_dense_depth: 1.1001, loss: 32.8244, grad_norm: 62.0057 -2025-11-12 21:31:15,607 - mmdet - INFO - Iter [58/17500] lr: 1.228e-04, eta: 17:00:58, time: 1.543, data_time: 0.079, memory: 49167, loss_cls_0: 1.0036, loss_box_0: 1.8841, loss_cns_0: 0.6129, loss_yns_0: 0.1630, loss_cls_1: 1.0396, loss_box_1: 2.1462, loss_cns_1: 0.6143, loss_yns_1: 0.1659, loss_cls_2: 1.0902, loss_box_2: 2.0704, loss_cns_2: 0.6364, loss_yns_2: 0.1701, loss_cls_3: 1.1199, loss_box_3: 2.1379, loss_cns_3: 0.6379, loss_yns_3: 0.1705, loss_cls_4: 1.1177, loss_box_4: 2.0948, loss_cns_4: 0.6399, loss_yns_4: 0.1690, loss_cls_5: 1.1020, loss_box_5: 2.0942, loss_cns_5: 0.6416, loss_yns_5: 0.1669, loss_cls_dn_0: 0.3712, loss_box_dn_0: 0.8871, loss_cls_dn_1: 0.3114, loss_box_dn_1: 0.9532, loss_cls_dn_2: 0.3494, loss_box_dn_2: 0.9218, loss_cls_dn_3: 0.3417, loss_box_dn_3: 0.9603, loss_cls_dn_4: 0.3474, loss_box_dn_4: 0.9418, loss_cls_dn_5: 0.3719, loss_box_dn_5: 0.9590, loss_dense_depth: 0.9346, loss: 32.3398, grad_norm: 49.4023 -2025-11-12 21:31:17,146 - mmdet - INFO - Iter [59/17500] lr: 1.232e-04, eta: 16:51:11, time: 1.540, data_time: 0.078, memory: 49167, loss_cls_0: 0.9876, loss_box_0: 1.8869, loss_cns_0: 0.6138, loss_yns_0: 0.1592, loss_cls_1: 1.0260, loss_box_1: 2.1742, loss_cns_1: 0.6150, loss_yns_1: 0.1637, loss_cls_2: 1.0802, loss_box_2: 2.1133, loss_cns_2: 0.6363, loss_yns_2: 0.1694, loss_cls_3: 1.0840, loss_box_3: 2.1671, loss_cns_3: 0.6404, loss_yns_3: 0.1679, loss_cls_4: 1.0834, loss_box_4: 2.1203, loss_cns_4: 0.6415, loss_yns_4: 0.1685, loss_cls_5: 1.1045, loss_box_5: 2.1218, loss_cns_5: 0.6436, loss_yns_5: 0.1639, loss_cls_dn_0: 0.3694, loss_box_dn_0: 0.8847, loss_cls_dn_1: 0.3134, loss_box_dn_1: 0.9679, loss_cls_dn_2: 0.3490, loss_box_dn_2: 0.9363, loss_cls_dn_3: 0.3565, loss_box_dn_3: 0.9665, loss_cls_dn_4: 0.3777, loss_box_dn_4: 0.9418, loss_cls_dn_5: 0.3602, loss_box_dn_5: 0.9664, loss_dense_depth: 1.0414, loss: 32.5636, grad_norm: 55.7274 -2025-11-12 21:31:18,694 - mmdet - INFO - Iter [60/17500] lr: 1.236e-04, eta: 16:41:46, time: 1.546, data_time: 0.087, memory: 49167, loss_cls_0: 0.9853, loss_box_0: 1.8426, loss_cns_0: 0.6155, loss_yns_0: 0.1587, loss_cls_1: 1.0366, loss_box_1: 2.2058, loss_cns_1: 0.6136, loss_yns_1: 0.1663, loss_cls_2: 1.0817, loss_box_2: 2.1608, loss_cns_2: 0.6349, loss_yns_2: 0.1648, loss_cls_3: 1.0975, loss_box_3: 2.1847, loss_cns_3: 0.6390, loss_yns_3: 0.1641, loss_cls_4: 1.1061, loss_box_4: 2.1408, loss_cns_4: 0.6446, loss_yns_4: 0.1696, loss_cls_5: 1.1116, loss_box_5: 2.1591, loss_cns_5: 0.6402, loss_yns_5: 0.1677, loss_cls_dn_0: 0.3792, loss_box_dn_0: 0.8901, loss_cls_dn_1: 0.2943, loss_box_dn_1: 0.9748, loss_cls_dn_2: 0.3366, loss_box_dn_2: 0.9502, loss_cls_dn_3: 0.3625, loss_box_dn_3: 0.9670, loss_cls_dn_4: 0.3916, loss_box_dn_4: 0.9471, loss_cls_dn_5: 0.3500, loss_box_dn_5: 0.9805, loss_dense_depth: 0.9865, loss: 32.7018, grad_norm: 50.0290 -2025-11-12 21:31:20,302 - mmdet - INFO - Iter [61/17500] lr: 1.240e-04, eta: 16:32:57, time: 1.609, data_time: 0.119, memory: 49167, loss_cls_0: 0.9869, loss_box_0: 1.8515, loss_cns_0: 0.6109, loss_yns_0: 0.1590, loss_cls_1: 1.0639, loss_box_1: 2.1404, loss_cns_1: 0.6183, loss_yns_1: 0.1641, loss_cls_2: 1.0792, loss_box_2: 2.0908, loss_cns_2: 0.6366, loss_yns_2: 0.1635, loss_cls_3: 1.1163, loss_box_3: 2.1072, loss_cns_3: 0.6421, loss_yns_3: 0.1617, loss_cls_4: 1.1140, loss_box_4: 2.1131, loss_cns_4: 0.6465, loss_yns_4: 0.1673, loss_cls_5: 1.1220, loss_box_5: 2.1451, loss_cns_5: 0.6354, loss_yns_5: 0.1681, loss_cls_dn_0: 0.3899, loss_box_dn_0: 0.8813, loss_cls_dn_1: 0.2861, loss_box_dn_1: 0.9935, loss_cls_dn_2: 0.3237, loss_box_dn_2: 0.9571, loss_cls_dn_3: 0.3544, loss_box_dn_3: 0.9622, loss_cls_dn_4: 0.3852, loss_box_dn_4: 0.9681, loss_cls_dn_5: 0.3432, loss_box_dn_5: 0.9941, loss_dense_depth: 1.0002, loss: 32.5428, grad_norm: 45.8227 -2025-11-12 21:31:21,907 - mmdet - INFO - Iter [62/17500] lr: 1.244e-04, eta: 16:24:25, time: 1.607, data_time: 0.123, memory: 49167, loss_cls_0: 0.9887, loss_box_0: 1.8867, loss_cns_0: 0.6117, loss_yns_0: 0.1582, loss_cls_1: 1.0402, loss_box_1: 2.1589, loss_cns_1: 0.6236, loss_yns_1: 0.1624, loss_cls_2: 1.0745, loss_box_2: 2.1059, loss_cns_2: 0.6354, loss_yns_2: 0.1622, loss_cls_3: 1.0931, loss_box_3: 2.1127, loss_cns_3: 0.6412, loss_yns_3: 0.1625, loss_cls_4: 1.0912, loss_box_4: 2.1258, loss_cns_4: 0.6468, loss_yns_4: 0.1642, loss_cls_5: 1.1027, loss_box_5: 2.1483, loss_cns_5: 0.6375, loss_yns_5: 0.1627, loss_cls_dn_0: 0.3875, loss_box_dn_0: 0.8831, loss_cls_dn_1: 0.2961, loss_box_dn_1: 0.9254, loss_cls_dn_2: 0.3206, loss_box_dn_2: 0.8840, loss_cls_dn_3: 0.3497, loss_box_dn_3: 0.8843, loss_cls_dn_4: 0.3787, loss_box_dn_4: 0.8982, loss_cls_dn_5: 0.3628, loss_box_dn_5: 0.9131, loss_dense_depth: 0.9866, loss: 32.1673, grad_norm: 33.4032 -2025-11-12 21:31:23,465 - mmdet - INFO - Iter [63/17500] lr: 1.248e-04, eta: 16:15:54, time: 1.553, data_time: 0.103, memory: 49167, loss_cls_0: 0.9966, loss_box_0: 1.8921, loss_cns_0: 0.6136, loss_yns_0: 0.1612, loss_cls_1: 1.0432, loss_box_1: 2.1365, loss_cns_1: 0.6264, loss_yns_1: 0.1650, loss_cls_2: 1.1328, loss_box_2: 2.0767, loss_cns_2: 0.6379, loss_yns_2: 0.1631, loss_cls_3: 1.1258, loss_box_3: 2.0589, loss_cns_3: 0.6447, loss_yns_3: 0.1634, loss_cls_4: 1.1156, loss_box_4: 2.0721, loss_cns_4: 0.6461, loss_yns_4: 0.1626, loss_cls_5: 1.0945, loss_box_5: 2.0899, loss_cns_5: 0.6463, loss_yns_5: 0.1624, loss_cls_dn_0: 0.3675, loss_box_dn_0: 0.8759, loss_cls_dn_1: 0.2936, loss_box_dn_1: 0.9152, loss_cls_dn_2: 0.3054, loss_box_dn_2: 0.8769, loss_cls_dn_3: 0.3157, loss_box_dn_3: 0.8735, loss_cls_dn_4: 0.3386, loss_box_dn_4: 0.8879, loss_cls_dn_5: 0.3626, loss_box_dn_5: 0.9000, loss_dense_depth: 1.0145, loss: 31.9545, grad_norm: 42.7332 -2025-11-12 21:31:24,971 - mmdet - INFO - Iter [64/17500] lr: 1.252e-04, eta: 16:07:27, time: 1.510, data_time: 0.078, memory: 49167, loss_cls_0: 0.9877, loss_box_0: 1.8515, loss_cns_0: 0.6197, loss_yns_0: 0.1629, loss_cls_1: 1.0560, loss_box_1: 2.1314, loss_cns_1: 0.6277, loss_yns_1: 0.1682, loss_cls_2: 1.0953, loss_box_2: 2.0942, loss_cns_2: 0.6365, loss_yns_2: 0.1623, loss_cls_3: 1.1487, loss_box_3: 2.0839, loss_cns_3: 0.6400, loss_yns_3: 0.1662, loss_cls_4: 1.1513, loss_box_4: 2.0524, loss_cns_4: 0.6466, loss_yns_4: 0.1693, loss_cls_5: 1.0914, loss_box_5: 2.1017, loss_cns_5: 0.6466, loss_yns_5: 0.1673, loss_cls_dn_0: 0.3489, loss_box_dn_0: 0.8537, loss_cls_dn_1: 0.2746, loss_box_dn_1: 0.9154, loss_cls_dn_2: 0.2961, loss_box_dn_2: 0.8849, loss_cls_dn_3: 0.2942, loss_box_dn_3: 0.8828, loss_cls_dn_4: 0.3057, loss_box_dn_4: 0.8839, loss_cls_dn_5: 0.3474, loss_box_dn_5: 0.9061, loss_dense_depth: 0.9695, loss: 31.8221, grad_norm: 41.3911 -2025-11-12 21:31:26,501 - mmdet - INFO - Iter [65/17500] lr: 1.256e-04, eta: 15:59:21, time: 1.531, data_time: 0.076, memory: 49167, loss_cls_0: 0.9779, loss_box_0: 1.8577, loss_cns_0: 0.6194, loss_yns_0: 0.1662, loss_cls_1: 1.0545, loss_box_1: 2.1584, loss_cns_1: 0.6291, loss_yns_1: 0.1661, loss_cls_2: 1.0685, loss_box_2: 2.1423, loss_cns_2: 0.6288, loss_yns_2: 0.1617, loss_cls_3: 1.0810, loss_box_3: 2.1493, loss_cns_3: 0.6339, loss_yns_3: 0.1639, loss_cls_4: 1.1116, loss_box_4: 2.0815, loss_cns_4: 0.6496, loss_yns_4: 0.1711, loss_cls_5: 1.0945, loss_box_5: 2.1009, loss_cns_5: 0.6514, loss_yns_5: 0.1702, loss_cls_dn_0: 0.3557, loss_box_dn_0: 0.8563, loss_cls_dn_1: 0.2553, loss_box_dn_1: 0.9593, loss_cls_dn_2: 0.2985, loss_box_dn_2: 0.9350, loss_cls_dn_3: 0.2991, loss_box_dn_3: 0.9372, loss_cls_dn_4: 0.2991, loss_box_dn_4: 0.9262, loss_cls_dn_5: 0.3343, loss_box_dn_5: 0.9412, loss_dense_depth: 0.9812, loss: 32.0682, grad_norm: 36.7776 -2025-11-12 21:31:28,064 - mmdet - INFO - Iter [66/17500] lr: 1.260e-04, eta: 15:51:39, time: 1.563, data_time: 0.109, memory: 49167, loss_cls_0: 0.9630, loss_box_0: 1.8607, loss_cns_0: 0.6205, loss_yns_0: 0.1651, loss_cls_1: 1.0624, loss_box_1: 2.0911, loss_cns_1: 0.6339, loss_yns_1: 0.1641, loss_cls_2: 1.0732, loss_box_2: 2.0350, loss_cns_2: 0.6437, loss_yns_2: 0.1628, loss_cls_3: 1.0919, loss_box_3: 2.0372, loss_cns_3: 0.6472, loss_yns_3: 0.1648, loss_cls_4: 1.0974, loss_box_4: 2.0227, loss_cns_4: 0.6539, loss_yns_4: 0.1681, loss_cls_5: 1.1026, loss_box_5: 2.0411, loss_cns_5: 0.6489, loss_yns_5: 0.1673, loss_cls_dn_0: 0.3516, loss_box_dn_0: 0.8498, loss_cls_dn_1: 0.2353, loss_box_dn_1: 0.9926, loss_cls_dn_2: 0.2925, loss_box_dn_2: 0.9523, loss_cls_dn_3: 0.3024, loss_box_dn_3: 0.9531, loss_cls_dn_4: 0.2988, loss_box_dn_4: 0.9586, loss_cls_dn_5: 0.3143, loss_box_dn_5: 0.9741, loss_dense_depth: 0.9332, loss: 31.7275, grad_norm: 31.4052 -2025-11-12 21:31:29,575 - mmdet - INFO - Iter [67/17500] lr: 1.264e-04, eta: 15:43:56, time: 1.511, data_time: 0.074, memory: 49167, loss_cls_0: 0.9844, loss_box_0: 1.8835, loss_cns_0: 0.6160, loss_yns_0: 0.1660, loss_cls_1: 1.0705, loss_box_1: 2.1130, loss_cns_1: 0.6245, loss_yns_1: 0.1657, loss_cls_2: 1.0702, loss_box_2: 2.0384, loss_cns_2: 0.6453, loss_yns_2: 0.1622, loss_cls_3: 1.0784, loss_box_3: 2.0445, loss_cns_3: 0.6479, loss_yns_3: 0.1621, loss_cls_4: 1.0812, loss_box_4: 2.0502, loss_cns_4: 0.6495, loss_yns_4: 0.1616, loss_cls_5: 1.1040, loss_box_5: 2.0555, loss_cns_5: 0.6464, loss_yns_5: 0.1617, loss_cls_dn_0: 0.3504, loss_box_dn_0: 0.8565, loss_cls_dn_1: 0.2313, loss_box_dn_1: 0.9556, loss_cls_dn_2: 0.2823, loss_box_dn_2: 0.9103, loss_cls_dn_3: 0.3026, loss_box_dn_3: 0.9174, loss_cls_dn_4: 0.3084, loss_box_dn_4: 0.9274, loss_cls_dn_5: 0.3075, loss_box_dn_5: 0.9410, loss_dense_depth: 0.9203, loss: 31.5937, grad_norm: 35.9473 -2025-11-12 21:31:31,095 - mmdet - INFO - Iter [68/17500] lr: 1.268e-04, eta: 15:36:30, time: 1.520, data_time: 0.073, memory: 49167, loss_cls_0: 0.9831, loss_box_0: 1.8441, loss_cns_0: 0.6208, loss_yns_0: 0.1642, loss_cls_1: 1.0467, loss_box_1: 2.0864, loss_cns_1: 0.6203, loss_yns_1: 0.1645, loss_cls_2: 1.0778, loss_box_2: 2.0135, loss_cns_2: 0.6445, loss_yns_2: 0.1632, loss_cls_3: 1.0742, loss_box_3: 2.0132, loss_cns_3: 0.6485, loss_yns_3: 0.1662, loss_cls_4: 1.0847, loss_box_4: 2.0230, loss_cns_4: 0.6499, loss_yns_4: 0.1668, loss_cls_5: 1.1394, loss_box_5: 2.0181, loss_cns_5: 0.6522, loss_yns_5: 0.1693, loss_cls_dn_0: 0.3480, loss_box_dn_0: 0.8656, loss_cls_dn_1: 0.2424, loss_box_dn_1: 0.8953, loss_cls_dn_2: 0.2795, loss_box_dn_2: 0.8490, loss_cls_dn_3: 0.3023, loss_box_dn_3: 0.8519, loss_cls_dn_4: 0.3235, loss_box_dn_4: 0.8609, loss_cls_dn_5: 0.3170, loss_box_dn_5: 0.8714, loss_dense_depth: 0.9206, loss: 31.1622, grad_norm: 37.2480 -2025-11-12 21:31:32,605 - mmdet - INFO - Iter [69/17500] lr: 1.272e-04, eta: 15:29:14, time: 1.510, data_time: 0.084, memory: 49167, loss_cls_0: 0.9494, loss_box_0: 1.7985, loss_cns_0: 0.6237, loss_yns_0: 0.1592, loss_cls_1: 1.0015, loss_box_1: 2.0777, loss_cns_1: 0.6208, loss_yns_1: 0.1627, loss_cls_2: 1.0647, loss_box_2: 1.9843, loss_cns_2: 0.6462, loss_yns_2: 0.1605, loss_cls_3: 1.0821, loss_box_3: 1.9762, loss_cns_3: 0.6506, loss_yns_3: 0.1654, loss_cls_4: 1.0763, loss_box_4: 1.9959, loss_cns_4: 0.6508, loss_yns_4: 0.1638, loss_cls_5: 1.0853, loss_box_5: 2.0013, loss_cns_5: 0.6473, loss_yns_5: 0.1643, loss_cls_dn_0: 0.3305, loss_box_dn_0: 0.8554, loss_cls_dn_1: 0.2545, loss_box_dn_1: 0.8978, loss_cls_dn_2: 0.2703, loss_box_dn_2: 0.8375, loss_cls_dn_3: 0.2890, loss_box_dn_3: 0.8347, loss_cls_dn_4: 0.3225, loss_box_dn_4: 0.8445, loss_cls_dn_5: 0.3346, loss_box_dn_5: 0.8587, loss_dense_depth: 0.8600, loss: 30.6983, grad_norm: 34.4999 -2025-11-12 21:31:34,152 - mmdet - INFO - Iter [70/17500] lr: 1.276e-04, eta: 15:22:20, time: 1.547, data_time: 0.073, memory: 49167, loss_cls_0: 0.9550, loss_box_0: 1.8347, loss_cns_0: 0.6175, loss_yns_0: 0.1586, loss_cls_1: 0.9952, loss_box_1: 2.0251, loss_cns_1: 0.6144, loss_yns_1: 0.1597, loss_cls_2: 1.0415, loss_box_2: 1.9736, loss_cns_2: 0.6358, loss_yns_2: 0.1592, loss_cls_3: 1.0679, loss_box_3: 1.9624, loss_cns_3: 0.6449, loss_yns_3: 0.1613, loss_cls_4: 1.0584, loss_box_4: 1.9592, loss_cns_4: 0.6466, loss_yns_4: 0.1605, loss_cls_5: 1.0555, loss_box_5: 1.9598, loss_cns_5: 0.6451, loss_yns_5: 0.1624, loss_cls_dn_0: 0.3324, loss_box_dn_0: 0.8679, loss_cls_dn_1: 0.2770, loss_box_dn_1: 0.8407, loss_cls_dn_2: 0.2854, loss_box_dn_2: 0.7965, loss_cls_dn_3: 0.2933, loss_box_dn_3: 0.8005, loss_cls_dn_4: 0.3300, loss_box_dn_4: 0.8099, loss_cls_dn_5: 0.3586, loss_box_dn_5: 0.8275, loss_dense_depth: 0.8908, loss: 30.3647, grad_norm: 37.3496 -2025-11-12 21:31:35,663 - mmdet - INFO - Iter [71/17500] lr: 1.280e-04, eta: 15:15:28, time: 1.511, data_time: 0.078, memory: 49167, loss_cls_0: 0.9573, loss_box_0: 1.8413, loss_cns_0: 0.6212, loss_yns_0: 0.1605, loss_cls_1: 1.0049, loss_box_1: 2.0521, loss_cns_1: 0.6152, loss_yns_1: 0.1576, loss_cls_2: 1.0408, loss_box_2: 1.9675, loss_cns_2: 0.6396, loss_yns_2: 0.1580, loss_cls_3: 1.0655, loss_box_3: 1.9698, loss_cns_3: 0.6484, loss_yns_3: 0.1584, loss_cls_4: 1.0578, loss_box_4: 1.9675, loss_cns_4: 0.6493, loss_yns_4: 0.1596, loss_cls_5: 1.0529, loss_box_5: 1.9730, loss_cns_5: 0.6460, loss_yns_5: 0.1618, loss_cls_dn_0: 0.3142, loss_box_dn_0: 0.8764, loss_cls_dn_1: 0.2645, loss_box_dn_1: 0.8533, loss_cls_dn_2: 0.2729, loss_box_dn_2: 0.8157, loss_cls_dn_3: 0.2695, loss_box_dn_3: 0.8225, loss_cls_dn_4: 0.3058, loss_box_dn_4: 0.8352, loss_cls_dn_5: 0.3300, loss_box_dn_5: 0.8546, loss_dense_depth: 0.8893, loss: 30.4297, grad_norm: 42.5130 -2025-11-12 21:31:37,187 - mmdet - INFO - Iter [72/17500] lr: 1.284e-04, eta: 15:08:51, time: 1.524, data_time: 0.077, memory: 49167, loss_cls_0: 0.9567, loss_box_0: 1.8181, loss_cns_0: 0.6205, loss_yns_0: 0.1625, loss_cls_1: 1.0070, loss_box_1: 2.0451, loss_cns_1: 0.6178, loss_yns_1: 0.1604, loss_cls_2: 1.0341, loss_box_2: 1.9297, loss_cns_2: 0.6427, loss_yns_2: 0.1593, loss_cls_3: 1.0557, loss_box_3: 1.9533, loss_cns_3: 0.6459, loss_yns_3: 0.1622, loss_cls_4: 1.0703, loss_box_4: 1.9333, loss_cns_4: 0.6494, loss_yns_4: 0.1622, loss_cls_5: 1.0664, loss_box_5: 1.9466, loss_cns_5: 0.6459, loss_yns_5: 0.1636, loss_cls_dn_0: 0.3127, loss_box_dn_0: 0.8644, loss_cls_dn_1: 0.2577, loss_box_dn_1: 0.8455, loss_cls_dn_2: 0.2721, loss_box_dn_2: 0.8138, loss_cls_dn_3: 0.2674, loss_box_dn_3: 0.8266, loss_cls_dn_4: 0.2971, loss_box_dn_4: 0.8400, loss_cls_dn_5: 0.3175, loss_box_dn_5: 0.8601, loss_dense_depth: 0.8908, loss: 30.2746, grad_norm: 41.9050 -2025-11-12 21:31:38,711 - mmdet - INFO - Iter [73/17500] lr: 1.288e-04, eta: 15:02:24, time: 1.523, data_time: 0.074, memory: 49167, loss_cls_0: 0.9640, loss_box_0: 1.8327, loss_cns_0: 0.6112, loss_yns_0: 0.1585, loss_cls_1: 1.0011, loss_box_1: 2.0543, loss_cns_1: 0.6198, loss_yns_1: 0.1625, loss_cls_2: 1.0543, loss_box_2: 1.9560, loss_cns_2: 0.6432, loss_yns_2: 0.1593, loss_cls_3: 1.0690, loss_box_3: 1.9710, loss_cns_3: 0.6466, loss_yns_3: 0.1639, loss_cls_4: 1.0604, loss_box_4: 1.9595, loss_cns_4: 0.6494, loss_yns_4: 0.1623, loss_cls_5: 1.0576, loss_box_5: 1.9611, loss_cns_5: 0.6479, loss_yns_5: 0.1619, loss_cls_dn_0: 0.3151, loss_box_dn_0: 0.8588, loss_cls_dn_1: 0.2525, loss_box_dn_1: 0.8571, loss_cls_dn_2: 0.2846, loss_box_dn_2: 0.8286, loss_cls_dn_3: 0.2865, loss_box_dn_3: 0.8396, loss_cls_dn_4: 0.2923, loss_box_dn_4: 0.8539, loss_cls_dn_5: 0.3134, loss_box_dn_5: 0.8729, loss_dense_depth: 0.8760, loss: 30.4590, grad_norm: 32.6876 -2025-11-12 21:31:40,220 - mmdet - INFO - Iter [74/17500] lr: 1.292e-04, eta: 14:56:05, time: 1.509, data_time: 0.076, memory: 49167, loss_cls_0: 0.9561, loss_box_0: 1.8109, loss_cns_0: 0.6163, loss_yns_0: 0.1588, loss_cls_1: 1.0080, loss_box_1: 2.0658, loss_cns_1: 0.6262, loss_yns_1: 0.1642, loss_cls_2: 1.0377, loss_box_2: 1.9984, loss_cns_2: 0.6444, loss_yns_2: 0.1622, loss_cls_3: 1.0677, loss_box_3: 1.9895, loss_cns_3: 0.6517, loss_yns_3: 0.1620, loss_cls_4: 1.0530, loss_box_4: 2.0158, loss_cns_4: 0.6492, loss_yns_4: 0.1633, loss_cls_5: 1.0489, loss_box_5: 2.0061, loss_cns_5: 0.6497, loss_yns_5: 0.1610, loss_cls_dn_0: 0.3119, loss_box_dn_0: 0.8519, loss_cls_dn_1: 0.2355, loss_box_dn_1: 0.8800, loss_cls_dn_2: 0.2661, loss_box_dn_2: 0.8492, loss_cls_dn_3: 0.2799, loss_box_dn_3: 0.8495, loss_cls_dn_4: 0.2765, loss_box_dn_4: 0.8665, loss_cls_dn_5: 0.2966, loss_box_dn_5: 0.8783, loss_dense_depth: 0.8787, loss: 30.5876, grad_norm: 39.8087 -2025-11-12 21:31:41,762 - mmdet - INFO - Iter [75/17500] lr: 1.296e-04, eta: 14:50:03, time: 1.541, data_time: 0.076, memory: 49167, loss_cls_0: 0.9604, loss_box_0: 1.8343, loss_cns_0: 0.6143, loss_yns_0: 0.1634, loss_cls_1: 1.0441, loss_box_1: 2.1091, loss_cns_1: 0.6223, loss_yns_1: 0.1638, loss_cls_2: 1.0984, loss_box_2: 2.0460, loss_cns_2: 0.6403, loss_yns_2: 0.1627, loss_cls_3: 1.0509, loss_box_3: 2.0126, loss_cns_3: 0.6501, loss_yns_3: 0.1620, loss_cls_4: 1.0853, loss_box_4: 2.0072, loss_cns_4: 0.6513, loss_yns_4: 0.1669, loss_cls_5: 1.1647, loss_box_5: 2.0010, loss_cns_5: 0.6534, loss_yns_5: 0.1650, loss_cls_dn_0: 0.3014, loss_box_dn_0: 0.8432, loss_cls_dn_1: 0.2147, loss_box_dn_1: 0.8823, loss_cls_dn_2: 0.2379, loss_box_dn_2: 0.8470, loss_cls_dn_3: 0.2491, loss_box_dn_3: 0.8357, loss_cls_dn_4: 0.2599, loss_box_dn_4: 0.8347, loss_cls_dn_5: 0.2712, loss_box_dn_5: 0.8428, loss_dense_depth: 0.8640, loss: 30.7132, grad_norm: 53.8116 -2025-11-12 21:31:43,282 - mmdet - INFO - Iter [76/17500] lr: 1.300e-04, eta: 14:44:06, time: 1.520, data_time: 0.079, memory: 49167, loss_cls_0: 0.9693, loss_box_0: 1.8252, loss_cns_0: 0.6144, loss_yns_0: 0.1640, loss_cls_1: 1.0193, loss_box_1: 2.0735, loss_cns_1: 0.6256, loss_yns_1: 0.1626, loss_cls_2: 1.0606, loss_box_2: 2.0341, loss_cns_2: 0.6427, loss_yns_2: 0.1631, loss_cls_3: 1.0682, loss_box_3: 2.0097, loss_cns_3: 0.6488, loss_yns_3: 0.1639, loss_cls_4: 1.0699, loss_box_4: 2.0019, loss_cns_4: 0.6476, loss_yns_4: 0.1699, loss_cls_5: 1.0534, loss_box_5: 2.0033, loss_cns_5: 0.6493, loss_yns_5: 0.1649, loss_cls_dn_0: 0.3004, loss_box_dn_0: 0.8426, loss_cls_dn_1: 0.2117, loss_box_dn_1: 0.8365, loss_cls_dn_2: 0.2317, loss_box_dn_2: 0.8093, loss_cls_dn_3: 0.2373, loss_box_dn_3: 0.7998, loss_cls_dn_4: 0.2678, loss_box_dn_4: 0.8018, loss_cls_dn_5: 0.2856, loss_box_dn_5: 0.8077, loss_dense_depth: 0.8594, loss: 30.2967, grad_norm: 34.5344 -2025-11-12 21:31:44,813 - mmdet - INFO - Iter [77/17500] lr: 1.304e-04, eta: 14:38:20, time: 1.530, data_time: 0.080, memory: 49167, loss_cls_0: 0.9474, loss_box_0: 1.8597, loss_cns_0: 0.6084, loss_yns_0: 0.1592, loss_cls_1: 1.0499, loss_box_1: 2.0349, loss_cns_1: 0.6326, loss_yns_1: 0.1614, loss_cls_2: 1.0612, loss_box_2: 2.0068, loss_cns_2: 0.6487, loss_yns_2: 0.1626, loss_cls_3: 1.0536, loss_box_3: 1.9771, loss_cns_3: 0.6532, loss_yns_3: 0.1611, loss_cls_4: 1.0759, loss_box_4: 1.9588, loss_cns_4: 0.6529, loss_yns_4: 0.1614, loss_cls_5: 1.0854, loss_box_5: 1.9513, loss_cns_5: 0.6544, loss_yns_5: 0.1623, loss_cls_dn_0: 0.2958, loss_box_dn_0: 0.8337, loss_cls_dn_1: 0.2164, loss_box_dn_1: 0.7963, loss_cls_dn_2: 0.2395, loss_box_dn_2: 0.7775, loss_cls_dn_3: 0.2359, loss_box_dn_3: 0.7779, loss_cls_dn_4: 0.2657, loss_box_dn_4: 0.7821, loss_cls_dn_5: 0.2976, loss_box_dn_5: 0.7902, loss_dense_depth: 0.8508, loss: 30.0397, grad_norm: 35.5354 -2025-11-12 21:31:46,327 - mmdet - INFO - Iter [78/17500] lr: 1.308e-04, eta: 14:32:40, time: 1.516, data_time: 0.078, memory: 49167, loss_cls_0: 0.9717, loss_box_0: 1.8933, loss_cns_0: 0.6133, loss_yns_0: 0.1615, loss_cls_1: 1.0497, loss_box_1: 2.1390, loss_cns_1: 0.6205, loss_yns_1: 0.1688, loss_cls_2: 1.0693, loss_box_2: 2.0594, loss_cns_2: 0.6429, loss_yns_2: 0.1668, loss_cls_3: 1.1056, loss_box_3: 2.0483, loss_cns_3: 0.6483, loss_yns_3: 0.1641, loss_cls_4: 1.1194, loss_box_4: 2.0400, loss_cns_4: 0.6494, loss_yns_4: 0.1669, loss_cls_5: 1.0865, loss_box_5: 2.0475, loss_cns_5: 0.6494, loss_yns_5: 0.1649, loss_cls_dn_0: 0.2820, loss_box_dn_0: 0.8308, loss_cls_dn_1: 0.2059, loss_box_dn_1: 0.8380, loss_cls_dn_2: 0.2256, loss_box_dn_2: 0.8007, loss_cls_dn_3: 0.2241, loss_box_dn_3: 0.8125, loss_cls_dn_4: 0.2347, loss_box_dn_4: 0.8204, loss_cls_dn_5: 0.2595, loss_box_dn_5: 0.8353, loss_dense_depth: 0.8597, loss: 30.6756, grad_norm: 59.3089 -2025-11-12 21:31:47,857 - mmdet - INFO - Iter [79/17500] lr: 1.312e-04, eta: 14:27:12, time: 1.530, data_time: 0.079, memory: 49167, loss_cls_0: 0.9809, loss_box_0: 1.8580, loss_cns_0: 0.6221, loss_yns_0: 0.1607, loss_cls_1: 1.0395, loss_box_1: 2.0844, loss_cns_1: 0.6327, loss_yns_1: 0.1668, loss_cls_2: 1.0553, loss_box_2: 2.0141, loss_cns_2: 0.6469, loss_yns_2: 0.1620, loss_cls_3: 1.0835, loss_box_3: 2.0035, loss_cns_3: 0.6519, loss_yns_3: 0.1633, loss_cls_4: 1.1140, loss_box_4: 2.0014, loss_cns_4: 0.6534, loss_yns_4: 0.1695, loss_cls_5: 1.0676, loss_box_5: 2.0002, loss_cns_5: 0.6521, loss_yns_5: 0.1635, loss_cls_dn_0: 0.2640, loss_box_dn_0: 0.8261, loss_cls_dn_1: 0.2020, loss_box_dn_1: 0.8500, loss_cls_dn_2: 0.2174, loss_box_dn_2: 0.8211, loss_cls_dn_3: 0.2163, loss_box_dn_3: 0.8272, loss_cls_dn_4: 0.2247, loss_box_dn_4: 0.8331, loss_cls_dn_5: 0.2397, loss_box_dn_5: 0.8430, loss_dense_depth: 0.8241, loss: 30.3358, grad_norm: 53.1106 -2025-11-12 21:31:49,385 - mmdet - INFO - Iter [80/17500] lr: 1.316e-04, eta: 14:21:51, time: 1.528, data_time: 0.079, memory: 49167, loss_cls_0: 0.9457, loss_box_0: 1.8627, loss_cns_0: 0.6188, loss_yns_0: 0.1594, loss_cls_1: 1.0241, loss_box_1: 2.1347, loss_cns_1: 0.6306, loss_yns_1: 0.1620, loss_cls_2: 1.0650, loss_box_2: 2.1025, loss_cns_2: 0.6428, loss_yns_2: 0.1623, loss_cls_3: 1.0872, loss_box_3: 2.0601, loss_cns_3: 0.6533, loss_yns_3: 0.1618, loss_cls_4: 1.0939, loss_box_4: 2.0496, loss_cns_4: 0.6547, loss_yns_4: 0.1645, loss_cls_5: 1.0491, loss_box_5: 2.0461, loss_cns_5: 0.6552, loss_yns_5: 0.1616, loss_cls_dn_0: 0.2645, loss_box_dn_0: 0.8188, loss_cls_dn_1: 0.2024, loss_box_dn_1: 0.8573, loss_cls_dn_2: 0.2193, loss_box_dn_2: 0.8286, loss_cls_dn_3: 0.2259, loss_box_dn_3: 0.8159, loss_cls_dn_4: 0.2411, loss_box_dn_4: 0.8149, loss_cls_dn_5: 0.2530, loss_box_dn_5: 0.8213, loss_dense_depth: 0.8233, loss: 30.5339, grad_norm: 45.4088 -2025-11-12 21:31:50,994 - mmdet - INFO - Iter [81/17500] lr: 1.320e-04, eta: 14:16:56, time: 1.609, data_time: 0.112, memory: 49167, loss_cls_0: 0.9605, loss_box_0: 1.8858, loss_cns_0: 0.6178, loss_yns_0: 0.1607, loss_cls_1: 1.0376, loss_box_1: 2.1423, loss_cns_1: 0.6255, loss_yns_1: 0.1610, loss_cls_2: 1.0860, loss_box_2: 2.1026, loss_cns_2: 0.6467, loss_yns_2: 0.1648, loss_cls_3: 1.1163, loss_box_3: 2.0610, loss_cns_3: 0.6542, loss_yns_3: 0.1623, loss_cls_4: 1.1381, loss_box_4: 2.0483, loss_cns_4: 0.6542, loss_yns_4: 0.1607, loss_cls_5: 1.0663, loss_box_5: 2.0377, loss_cns_5: 0.6549, loss_yns_5: 0.1610, loss_cls_dn_0: 0.2706, loss_box_dn_0: 0.8193, loss_cls_dn_1: 0.2039, loss_box_dn_1: 0.8679, loss_cls_dn_2: 0.2178, loss_box_dn_2: 0.8286, loss_cls_dn_3: 0.2326, loss_box_dn_3: 0.8138, loss_cls_dn_4: 0.2533, loss_box_dn_4: 0.8130, loss_cls_dn_5: 0.2512, loss_box_dn_5: 0.8140, loss_dense_depth: 0.8516, loss: 30.7440, grad_norm: 48.8783 -2025-11-12 21:31:52,584 - mmdet - INFO - Iter [82/17500] lr: 1.324e-04, eta: 14:12:04, time: 1.590, data_time: 0.077, memory: 49167, loss_cls_0: 0.9728, loss_box_0: 1.8861, loss_cns_0: 0.6180, loss_yns_0: 0.1588, loss_cls_1: 1.0429, loss_box_1: 2.1276, loss_cns_1: 0.6217, loss_yns_1: 0.1618, loss_cls_2: 1.0758, loss_box_2: 2.0642, loss_cns_2: 0.6430, loss_yns_2: 0.1654, loss_cls_3: 1.0902, loss_box_3: 2.0409, loss_cns_3: 0.6464, loss_yns_3: 0.1621, loss_cls_4: 1.1022, loss_box_4: 2.0498, loss_cns_4: 0.6432, loss_yns_4: 0.1621, loss_cls_5: 1.0780, loss_box_5: 2.0256, loss_cns_5: 0.6413, loss_yns_5: 0.1596, loss_cls_dn_0: 0.2688, loss_box_dn_0: 0.8335, loss_cls_dn_1: 0.1961, loss_box_dn_1: 0.8539, loss_cls_dn_2: 0.2070, loss_box_dn_2: 0.8108, loss_cls_dn_3: 0.2209, loss_box_dn_3: 0.8043, loss_cls_dn_4: 0.2318, loss_box_dn_4: 0.8153, loss_cls_dn_5: 0.2311, loss_box_dn_5: 0.8144, loss_dense_depth: 0.8673, loss: 30.4948, grad_norm: 44.0004 -2025-11-12 21:31:54,125 - mmdet - INFO - Iter [83/17500] lr: 1.328e-04, eta: 14:07:08, time: 1.541, data_time: 0.104, memory: 49167, loss_cls_0: 0.9725, loss_box_0: 1.8386, loss_cns_0: 0.6230, loss_yns_0: 0.1607, loss_cls_1: 1.0278, loss_box_1: 2.0515, loss_cns_1: 0.6191, loss_yns_1: 0.1628, loss_cls_2: 1.0860, loss_box_2: 1.9608, loss_cns_2: 0.6406, loss_yns_2: 0.1612, loss_cls_3: 1.0685, loss_box_3: 1.9328, loss_cns_3: 0.6456, loss_yns_3: 0.1608, loss_cls_4: 1.0922, loss_box_4: 1.9693, loss_cns_4: 0.6426, loss_yns_4: 0.1640, loss_cls_5: 1.1058, loss_box_5: 1.9747, loss_cns_5: 0.6407, loss_yns_5: 0.1593, loss_cls_dn_0: 0.2639, loss_box_dn_0: 0.8385, loss_cls_dn_1: 0.1996, loss_box_dn_1: 0.8395, loss_cls_dn_2: 0.2138, loss_box_dn_2: 0.8058, loss_cls_dn_3: 0.2167, loss_box_dn_3: 0.8063, loss_cls_dn_4: 0.2198, loss_box_dn_4: 0.8266, loss_cls_dn_5: 0.2343, loss_box_dn_5: 0.8372, loss_dense_depth: 0.8510, loss: 30.0141, grad_norm: 46.1550 -2025-11-12 21:31:55,653 - mmdet - INFO - Iter [84/17500] lr: 1.332e-04, eta: 14:02:17, time: 1.527, data_time: 0.080, memory: 49167, loss_cls_0: 0.9085, loss_box_0: 1.8107, loss_cns_0: 0.6250, loss_yns_0: 0.1595, loss_cls_1: 0.9739, loss_box_1: 2.0144, loss_cns_1: 0.6238, loss_yns_1: 0.1600, loss_cls_2: 1.0646, loss_box_2: 1.9188, loss_cns_2: 0.6485, loss_yns_2: 0.1605, loss_cls_3: 1.0581, loss_box_3: 1.9062, loss_cns_3: 0.6514, loss_yns_3: 0.1604, loss_cls_4: 1.1009, loss_box_4: 1.9313, loss_cns_4: 0.6526, loss_yns_4: 0.1605, loss_cls_5: 1.0428, loss_box_5: 1.9501, loss_cns_5: 0.6539, loss_yns_5: 0.1597, loss_cls_dn_0: 0.2527, loss_box_dn_0: 0.8280, loss_cls_dn_1: 0.2029, loss_box_dn_1: 0.8248, loss_cls_dn_2: 0.2168, loss_box_dn_2: 0.7893, loss_cls_dn_3: 0.2159, loss_box_dn_3: 0.7944, loss_cls_dn_4: 0.2228, loss_box_dn_4: 0.8101, loss_cls_dn_5: 0.2373, loss_box_dn_5: 0.8248, loss_dense_depth: 0.8022, loss: 29.5180, grad_norm: 51.0390 -2025-11-12 21:31:57,172 - mmdet - INFO - Iter [85/17500] lr: 1.336e-04, eta: 13:57:31, time: 1.519, data_time: 0.079, memory: 49167, loss_cls_0: 0.9581, loss_box_0: 1.8088, loss_cns_0: 0.6163, loss_yns_0: 0.1596, loss_cls_1: 1.0001, loss_box_1: 1.9658, loss_cns_1: 0.6166, loss_yns_1: 0.1571, loss_cls_2: 1.0430, loss_box_2: 1.9078, loss_cns_2: 0.6498, loss_yns_2: 0.1603, loss_cls_3: 1.0662, loss_box_3: 1.9129, loss_cns_3: 0.6513, loss_yns_3: 0.1585, loss_cls_4: 1.0694, loss_box_4: 1.9435, loss_cns_4: 0.6522, loss_yns_4: 0.1584, loss_cls_5: 1.0883, loss_box_5: 1.9559, loss_cns_5: 0.6520, loss_yns_5: 0.1596, loss_cls_dn_0: 0.2630, loss_box_dn_0: 0.8478, loss_cls_dn_1: 0.2044, loss_box_dn_1: 0.8575, loss_cls_dn_2: 0.2180, loss_box_dn_2: 0.8191, loss_cls_dn_3: 0.2164, loss_box_dn_3: 0.8241, loss_cls_dn_4: 0.2303, loss_box_dn_4: 0.8385, loss_cls_dn_5: 0.2541, loss_box_dn_5: 0.8469, loss_dense_depth: 0.8507, loss: 29.7824, grad_norm: 45.9071 -2025-11-12 21:31:58,720 - mmdet - INFO - Iter [86/17500] lr: 1.340e-04, eta: 13:52:57, time: 1.548, data_time: 0.112, memory: 49167, loss_cls_0: 0.9409, loss_box_0: 1.8086, loss_cns_0: 0.6182, loss_yns_0: 0.1614, loss_cls_1: 1.0091, loss_box_1: 1.8815, loss_cns_1: 0.6048, loss_yns_1: 0.1553, loss_cls_2: 1.0532, loss_box_2: 1.8974, loss_cns_2: 0.6421, loss_yns_2: 0.1645, loss_cls_3: 1.0556, loss_box_3: 1.8669, loss_cns_3: 0.6487, loss_yns_3: 0.1609, loss_cls_4: 1.0559, loss_box_4: 1.8824, loss_cns_4: 0.6504, loss_yns_4: 0.1624, loss_cls_5: 1.0983, loss_box_5: 1.8636, loss_cns_5: 0.6501, loss_yns_5: 0.1623, loss_cls_dn_0: 0.2510, loss_box_dn_0: 0.8351, loss_cls_dn_1: 0.2010, loss_box_dn_1: 0.8283, loss_cls_dn_2: 0.2137, loss_box_dn_2: 0.7941, loss_cls_dn_3: 0.2139, loss_box_dn_3: 0.7846, loss_cls_dn_4: 0.2337, loss_box_dn_4: 0.7887, loss_cls_dn_5: 0.2568, loss_box_dn_5: 0.7904, loss_dense_depth: 0.8425, loss: 29.2286, grad_norm: 34.1936 -2025-11-12 21:32:00,241 - mmdet - INFO - Iter [87/17500] lr: 1.344e-04, eta: 13:48:24, time: 1.521, data_time: 0.081, memory: 49167, loss_cls_0: 0.9208, loss_box_0: 1.8066, loss_cns_0: 0.6191, loss_yns_0: 0.1591, loss_cls_1: 0.9727, loss_box_1: 1.9846, loss_cns_1: 0.6034, loss_yns_1: 0.1551, loss_cls_2: 1.0388, loss_box_2: 2.0400, loss_cns_2: 0.6325, loss_yns_2: 0.1600, loss_cls_3: 1.0561, loss_box_3: 1.9751, loss_cns_3: 0.6466, loss_yns_3: 0.1611, loss_cls_4: 1.0722, loss_box_4: 1.9432, loss_cns_4: 0.6506, loss_yns_4: 0.1605, loss_cls_5: 1.0687, loss_box_5: 1.9425, loss_cns_5: 0.6508, loss_yns_5: 0.1615, loss_cls_dn_0: 0.2473, loss_box_dn_0: 0.8230, loss_cls_dn_1: 0.1970, loss_box_dn_1: 0.7984, loss_cls_dn_2: 0.2113, loss_box_dn_2: 0.7822, loss_cls_dn_3: 0.2164, loss_box_dn_3: 0.7645, loss_cls_dn_4: 0.2294, loss_box_dn_4: 0.7543, loss_cls_dn_5: 0.2410, loss_box_dn_5: 0.7689, loss_dense_depth: 0.8366, loss: 29.4518, grad_norm: 42.9300 -2025-11-12 21:32:01,752 - mmdet - INFO - Iter [88/17500] lr: 1.348e-04, eta: 13:43:56, time: 1.511, data_time: 0.078, memory: 49167, loss_cls_0: 0.9280, loss_box_0: 1.7917, loss_cns_0: 0.6155, loss_yns_0: 0.1572, loss_cls_1: 0.9640, loss_box_1: 1.9653, loss_cns_1: 0.5951, loss_yns_1: 0.1538, loss_cls_2: 1.0306, loss_box_2: 1.9852, loss_cns_2: 0.6368, loss_yns_2: 0.1593, loss_cls_3: 1.0622, loss_box_3: 1.9437, loss_cns_3: 0.6469, loss_yns_3: 0.1601, loss_cls_4: 1.0720, loss_box_4: 1.9068, loss_cns_4: 0.6507, loss_yns_4: 0.1596, loss_cls_5: 1.1334, loss_box_5: 1.9128, loss_cns_5: 0.6498, loss_yns_5: 0.1599, loss_cls_dn_0: 0.2523, loss_box_dn_0: 0.8247, loss_cls_dn_1: 0.1955, loss_box_dn_1: 0.7924, loss_cls_dn_2: 0.2112, loss_box_dn_2: 0.7668, loss_cls_dn_3: 0.2158, loss_box_dn_3: 0.7555, loss_cls_dn_4: 0.2260, loss_box_dn_4: 0.7598, loss_cls_dn_5: 0.2366, loss_box_dn_5: 0.7677, loss_dense_depth: 0.8099, loss: 29.2544, grad_norm: 50.1196 -2025-11-12 21:32:03,284 - mmdet - INFO - Iter [89/17500] lr: 1.352e-04, eta: 13:39:37, time: 1.533, data_time: 0.089, memory: 49167, loss_cls_0: 0.9520, loss_box_0: 1.7834, loss_cns_0: 0.6130, loss_yns_0: 0.1556, loss_cls_1: 0.9644, loss_box_1: 1.9619, loss_cns_1: 0.5926, loss_yns_1: 0.1546, loss_cls_2: 1.0153, loss_box_2: 1.9145, loss_cns_2: 0.6391, loss_yns_2: 0.1614, loss_cls_3: 1.0502, loss_box_3: 1.8865, loss_cns_3: 0.6482, loss_yns_3: 0.1598, loss_cls_4: 1.0602, loss_box_4: 1.8729, loss_cns_4: 0.6491, loss_yns_4: 0.1594, loss_cls_5: 1.0689, loss_box_5: 1.8557, loss_cns_5: 0.6491, loss_yns_5: 0.1591, loss_cls_dn_0: 0.2566, loss_box_dn_0: 0.8235, loss_cls_dn_1: 0.1984, loss_box_dn_1: 0.7962, loss_cls_dn_2: 0.2094, loss_box_dn_2: 0.7542, loss_cls_dn_3: 0.2133, loss_box_dn_3: 0.7497, loss_cls_dn_4: 0.2303, loss_box_dn_4: 0.7667, loss_cls_dn_5: 0.2305, loss_box_dn_5: 0.7647, loss_dense_depth: 0.8165, loss: 28.9370, grad_norm: 42.5916 -2025-11-12 21:32:04,828 - mmdet - INFO - Iter [90/17500] lr: 1.356e-04, eta: 13:35:27, time: 1.544, data_time: 0.077, memory: 49167, loss_cls_0: 0.9486, loss_box_0: 1.7677, loss_cns_0: 0.6069, loss_yns_0: 0.1546, loss_cls_1: 0.9973, loss_box_1: 2.0467, loss_cns_1: 0.6139, loss_yns_1: 0.1588, loss_cls_2: 1.0220, loss_box_2: 1.9489, loss_cns_2: 0.6440, loss_yns_2: 0.1602, loss_cls_3: 1.0572, loss_box_3: 1.9215, loss_cns_3: 0.6517, loss_yns_3: 0.1572, loss_cls_4: 1.0666, loss_box_4: 1.9234, loss_cns_4: 0.6523, loss_yns_4: 0.1574, loss_cls_5: 1.0643, loss_box_5: 1.9245, loss_cns_5: 0.6519, loss_yns_5: 0.1569, loss_cls_dn_0: 0.2566, loss_box_dn_0: 0.8360, loss_cls_dn_1: 0.1950, loss_box_dn_1: 0.7984, loss_cls_dn_2: 0.2026, loss_box_dn_2: 0.7662, loss_cls_dn_3: 0.2072, loss_box_dn_3: 0.7627, loss_cls_dn_4: 0.2352, loss_box_dn_4: 0.7785, loss_cls_dn_5: 0.2358, loss_box_dn_5: 0.7803, loss_dense_depth: 0.8756, loss: 29.3847, grad_norm: 37.8157 -2025-11-12 21:32:06,340 - mmdet - INFO - Iter [91/17500] lr: 1.360e-04, eta: 13:31:15, time: 1.511, data_time: 0.077, memory: 49167, loss_cls_0: 0.9320, loss_box_0: 1.7606, loss_cns_0: 0.6145, loss_yns_0: 0.1535, loss_cls_1: 0.9722, loss_box_1: 2.0753, loss_cns_1: 0.6130, loss_yns_1: 0.1556, loss_cls_2: 1.0355, loss_box_2: 1.9504, loss_cns_2: 0.6467, loss_yns_2: 0.1565, loss_cls_3: 1.0431, loss_box_3: 1.9296, loss_cns_3: 0.6536, loss_yns_3: 0.1560, loss_cls_4: 1.0554, loss_box_4: 1.9225, loss_cns_4: 0.6524, loss_yns_4: 0.1569, loss_cls_5: 1.0724, loss_box_5: 1.9685, loss_cns_5: 0.6496, loss_yns_5: 0.1553, loss_cls_dn_0: 0.2457, loss_box_dn_0: 0.8116, loss_cls_dn_1: 0.1925, loss_box_dn_1: 0.8109, loss_cls_dn_2: 0.2000, loss_box_dn_2: 0.7770, loss_cls_dn_3: 0.1997, loss_box_dn_3: 0.7768, loss_cls_dn_4: 0.2309, loss_box_dn_4: 0.7816, loss_cls_dn_5: 0.2462, loss_box_dn_5: 0.8008, loss_dense_depth: 0.8275, loss: 29.3821, grad_norm: 42.9594 -2025-11-12 21:32:07,848 - mmdet - INFO - Iter [92/17500] lr: 1.364e-04, eta: 13:27:09, time: 1.509, data_time: 0.077, memory: 49167, loss_cls_0: 0.9434, loss_box_0: 1.7435, loss_cns_0: 0.6083, loss_yns_0: 0.1497, loss_cls_1: 0.9517, loss_box_1: 2.0452, loss_cns_1: 0.6147, loss_yns_1: 0.1550, loss_cls_2: 1.0321, loss_box_2: 1.9386, loss_cns_2: 0.6430, loss_yns_2: 0.1557, loss_cls_3: 1.0413, loss_box_3: 1.9064, loss_cns_3: 0.6506, loss_yns_3: 0.1563, loss_cls_4: 1.0530, loss_box_4: 1.8978, loss_cns_4: 0.6535, loss_yns_4: 0.1563, loss_cls_5: 1.0701, loss_box_5: 1.8975, loss_cns_5: 0.6508, loss_yns_5: 0.1549, loss_cls_dn_0: 0.2470, loss_box_dn_0: 0.8106, loss_cls_dn_1: 0.1925, loss_box_dn_1: 0.8081, loss_cls_dn_2: 0.1979, loss_box_dn_2: 0.7686, loss_cls_dn_3: 0.1971, loss_box_dn_3: 0.7617, loss_cls_dn_4: 0.2242, loss_box_dn_4: 0.7600, loss_cls_dn_5: 0.2498, loss_box_dn_5: 0.7664, loss_dense_depth: 0.8226, loss: 29.0761, grad_norm: 43.1454 -2025-11-12 21:32:09,371 - mmdet - INFO - Iter [93/17500] lr: 1.368e-04, eta: 13:23:11, time: 1.523, data_time: 0.079, memory: 49167, loss_cls_0: 0.9423, loss_box_0: 1.7621, loss_cns_0: 0.6138, loss_yns_0: 0.1504, loss_cls_1: 0.9896, loss_box_1: 1.9732, loss_cns_1: 0.6287, loss_yns_1: 0.1532, loss_cls_2: 1.0314, loss_box_2: 1.9067, loss_cns_2: 0.6445, loss_yns_2: 0.1551, loss_cls_3: 1.0492, loss_box_3: 1.8622, loss_cns_3: 0.6548, loss_yns_3: 0.1533, loss_cls_4: 1.0539, loss_box_4: 1.8555, loss_cns_4: 0.6574, loss_yns_4: 0.1555, loss_cls_5: 1.0692, loss_box_5: 1.8492, loss_cns_5: 0.6567, loss_yns_5: 0.1563, loss_cls_dn_0: 0.2473, loss_box_dn_0: 0.8124, loss_cls_dn_1: 0.1948, loss_box_dn_1: 0.7856, loss_cls_dn_2: 0.1975, loss_box_dn_2: 0.7527, loss_cls_dn_3: 0.1990, loss_box_dn_3: 0.7431, loss_cls_dn_4: 0.2152, loss_box_dn_4: 0.7446, loss_cls_dn_5: 0.2404, loss_box_dn_5: 0.7507, loss_dense_depth: 0.8249, loss: 28.8324, grad_norm: 38.3038 -2025-11-12 21:32:10,894 - mmdet - INFO - Iter [94/17500] lr: 1.372e-04, eta: 13:19:17, time: 1.522, data_time: 0.078, memory: 49167, loss_cls_0: 0.9398, loss_box_0: 1.8006, loss_cns_0: 0.6175, loss_yns_0: 0.1540, loss_cls_1: 1.0024, loss_box_1: 1.9943, loss_cns_1: 0.6242, loss_yns_1: 0.1557, loss_cls_2: 1.0478, loss_box_2: 1.9434, loss_cns_2: 0.6432, loss_yns_2: 0.1573, loss_cls_3: 1.0527, loss_box_3: 1.9117, loss_cns_3: 0.6532, loss_yns_3: 0.1560, loss_cls_4: 1.0681, loss_box_4: 1.9040, loss_cns_4: 0.6543, loss_yns_4: 0.1614, loss_cls_5: 1.0696, loss_box_5: 1.9262, loss_cns_5: 0.6524, loss_yns_5: 0.1591, loss_cls_dn_0: 0.2445, loss_box_dn_0: 0.8081, loss_cls_dn_1: 0.2001, loss_box_dn_1: 0.7850, loss_cls_dn_2: 0.2054, loss_box_dn_2: 0.7586, loss_cls_dn_3: 0.2040, loss_box_dn_3: 0.7568, loss_cls_dn_4: 0.2132, loss_box_dn_4: 0.7698, loss_cls_dn_5: 0.2247, loss_box_dn_5: 0.7933, loss_dense_depth: 0.8060, loss: 29.2184, grad_norm: 39.1741 -2025-11-12 21:32:12,406 - mmdet - INFO - Iter [95/17500] lr: 1.376e-04, eta: 13:15:26, time: 1.512, data_time: 0.077, memory: 49167, loss_cls_0: 0.9350, loss_box_0: 1.8257, loss_cns_0: 0.6154, loss_yns_0: 0.1541, loss_cls_1: 0.9851, loss_box_1: 2.0134, loss_cns_1: 0.6300, loss_yns_1: 0.1577, loss_cls_2: 1.0530, loss_box_2: 1.9386, loss_cns_2: 0.6507, loss_yns_2: 0.1567, loss_cls_3: 1.0866, loss_box_3: 1.9393, loss_cns_3: 0.6575, loss_yns_3: 0.1572, loss_cls_4: 1.1451, loss_box_4: 1.9557, loss_cns_4: 0.6547, loss_yns_4: 0.1615, loss_cls_5: 1.0822, loss_box_5: 1.9563, loss_cns_5: 0.6531, loss_yns_5: 0.1601, loss_cls_dn_0: 0.2401, loss_box_dn_0: 0.8106, loss_cls_dn_1: 0.1928, loss_box_dn_1: 0.7841, loss_cls_dn_2: 0.2055, loss_box_dn_2: 0.7687, loss_cls_dn_3: 0.2024, loss_box_dn_3: 0.7838, loss_cls_dn_4: 0.2134, loss_box_dn_4: 0.8038, loss_cls_dn_5: 0.2180, loss_box_dn_5: 0.8180, loss_dense_depth: 0.8248, loss: 29.5908, grad_norm: 56.7011 -2025-11-12 21:32:13,927 - mmdet - INFO - Iter [96/17500] lr: 1.380e-04, eta: 13:11:42, time: 1.521, data_time: 0.079, memory: 49167, loss_cls_0: 0.9620, loss_box_0: 1.8692, loss_cns_0: 0.6133, loss_yns_0: 0.1583, loss_cls_1: 1.0085, loss_box_1: 1.9956, loss_cns_1: 0.6280, loss_yns_1: 0.1571, loss_cls_2: 1.0768, loss_box_2: 1.9462, loss_cns_2: 0.6463, loss_yns_2: 0.1547, loss_cls_3: 1.0509, loss_box_3: 1.9436, loss_cns_3: 0.6511, loss_yns_3: 0.1559, loss_cls_4: 1.0662, loss_box_4: 1.9698, loss_cns_4: 0.6508, loss_yns_4: 0.1584, loss_cls_5: 1.0761, loss_box_5: 1.9380, loss_cns_5: 0.6500, loss_yns_5: 0.1619, loss_cls_dn_0: 0.2471, loss_box_dn_0: 0.8226, loss_cls_dn_1: 0.1904, loss_box_dn_1: 0.8032, loss_cls_dn_2: 0.2097, loss_box_dn_2: 0.8008, loss_cls_dn_3: 0.2076, loss_box_dn_3: 0.8140, loss_cls_dn_4: 0.2155, loss_box_dn_4: 0.8269, loss_cls_dn_5: 0.2206, loss_box_dn_5: 0.8239, loss_dense_depth: 0.8428, loss: 29.7136, grad_norm: 53.4177 -2025-11-12 21:32:15,449 - mmdet - INFO - Iter [97/17500] lr: 1.384e-04, eta: 13:08:03, time: 1.521, data_time: 0.079, memory: 49167, loss_cls_0: 0.9589, loss_box_0: 1.8175, loss_cns_0: 0.6226, loss_yns_0: 0.1561, loss_cls_1: 1.0181, loss_box_1: 2.0158, loss_cns_1: 0.6308, loss_yns_1: 0.1571, loss_cls_2: 1.1264, loss_box_2: 1.9424, loss_cns_2: 0.6462, loss_yns_2: 0.1542, loss_cls_3: 1.0698, loss_box_3: 1.9360, loss_cns_3: 0.6546, loss_yns_3: 0.1566, loss_cls_4: 1.0733, loss_box_4: 1.9609, loss_cns_4: 0.6567, loss_yns_4: 0.1573, loss_cls_5: 1.0677, loss_box_5: 1.9443, loss_cns_5: 0.6519, loss_yns_5: 0.1575, loss_cls_dn_0: 0.2502, loss_box_dn_0: 0.8110, loss_cls_dn_1: 0.1898, loss_box_dn_1: 0.7991, loss_cls_dn_2: 0.2127, loss_box_dn_2: 0.7866, loss_cls_dn_3: 0.2059, loss_box_dn_3: 0.7900, loss_cls_dn_4: 0.2116, loss_box_dn_4: 0.7951, loss_cls_dn_5: 0.2165, loss_box_dn_5: 0.7950, loss_dense_depth: 0.8153, loss: 29.6117, grad_norm: 50.8466 -2025-11-12 21:32:16,951 - mmdet - INFO - Iter [98/17500] lr: 1.388e-04, eta: 13:04:24, time: 1.503, data_time: 0.077, memory: 49167, loss_cls_0: 0.9358, loss_box_0: 1.7887, loss_cns_0: 0.6254, loss_yns_0: 0.1558, loss_cls_1: 0.9976, loss_box_1: 1.9807, loss_cns_1: 0.6285, loss_yns_1: 0.1567, loss_cls_2: 1.0943, loss_box_2: 1.8745, loss_cns_2: 0.6440, loss_yns_2: 0.1553, loss_cls_3: 1.0633, loss_box_3: 1.8813, loss_cns_3: 0.6584, loss_yns_3: 0.1577, loss_cls_4: 1.0799, loss_box_4: 1.8829, loss_cns_4: 0.6559, loss_yns_4: 0.1577, loss_cls_5: 1.0727, loss_box_5: 1.8979, loss_cns_5: 0.6483, loss_yns_5: 0.1565, loss_cls_dn_0: 0.2478, loss_box_dn_0: 0.8087, loss_cls_dn_1: 0.1883, loss_box_dn_1: 0.8056, loss_cls_dn_2: 0.2089, loss_box_dn_2: 0.7731, loss_cls_dn_3: 0.1977, loss_box_dn_3: 0.7785, loss_cls_dn_4: 0.2113, loss_box_dn_4: 0.7758, loss_cls_dn_5: 0.2162, loss_box_dn_5: 0.7887, loss_dense_depth: 0.8261, loss: 29.1765, grad_norm: 49.0576 -2025-11-12 21:32:18,475 - mmdet - INFO - Iter [99/17500] lr: 1.392e-04, eta: 13:00:54, time: 1.524, data_time: 0.078, memory: 49167, loss_cls_0: 0.9355, loss_box_0: 1.7785, loss_cns_0: 0.6220, loss_yns_0: 0.1561, loss_cls_1: 1.0109, loss_box_1: 1.9817, loss_cns_1: 0.6283, loss_yns_1: 0.1586, loss_cls_2: 1.0573, loss_box_2: 1.9049, loss_cns_2: 0.6446, loss_yns_2: 0.1582, loss_cls_3: 1.0643, loss_box_3: 1.9011, loss_cns_3: 0.6549, loss_yns_3: 0.1579, loss_cls_4: 1.0751, loss_box_4: 1.8876, loss_cns_4: 0.6535, loss_yns_4: 0.1610, loss_cls_5: 1.0752, loss_box_5: 1.9145, loss_cns_5: 0.6500, loss_yns_5: 0.1632, loss_cls_dn_0: 0.2490, loss_box_dn_0: 0.8057, loss_cls_dn_1: 0.1845, loss_box_dn_1: 0.7929, loss_cls_dn_2: 0.1955, loss_box_dn_2: 0.7685, loss_cls_dn_3: 0.1957, loss_box_dn_3: 0.7739, loss_cls_dn_4: 0.2153, loss_box_dn_4: 0.7701, loss_cls_dn_5: 0.2237, loss_box_dn_5: 0.7883, loss_dense_depth: 0.8295, loss: 29.1876, grad_norm: 42.7610 -2025-11-12 21:32:20,001 - mmdet - INFO - Iter [100/17500] lr: 1.396e-04, eta: 12:57:29, time: 1.526, data_time: 0.087, memory: 49167, loss_cls_0: 0.9425, loss_box_0: 1.7968, loss_cns_0: 0.6212, loss_yns_0: 0.1578, loss_cls_1: 1.0225, loss_box_1: 2.0424, loss_cns_1: 0.6258, loss_yns_1: 0.1597, loss_cls_2: 1.0367, loss_box_2: 2.0087, loss_cns_2: 0.6361, loss_yns_2: 0.1586, loss_cls_3: 1.0609, loss_box_3: 1.9790, loss_cns_3: 0.6418, loss_yns_3: 0.1596, loss_cls_4: 1.0848, loss_box_4: 1.9717, loss_cns_4: 0.6437, loss_yns_4: 0.1601, loss_cls_5: 1.0777, loss_box_5: 1.9843, loss_cns_5: 0.6418, loss_yns_5: 0.1635, loss_cls_dn_0: 0.2477, loss_box_dn_0: 0.8158, loss_cls_dn_1: 0.1872, loss_box_dn_1: 0.7953, loss_cls_dn_2: 0.1971, loss_box_dn_2: 0.7777, loss_cls_dn_3: 0.2027, loss_box_dn_3: 0.7778, loss_cls_dn_4: 0.2261, loss_box_dn_4: 0.7773, loss_cls_dn_5: 0.2339, loss_box_dn_5: 0.7911, loss_dense_depth: 0.8022, loss: 29.6095, grad_norm: 48.3821 -2025-11-12 21:32:21,604 - mmdet - INFO - Iter [101/17500] lr: 1.400e-04, eta: 12:54:20, time: 1.603, data_time: 0.119, memory: 49167, loss_cls_0: 0.9424, loss_box_0: 1.7869, loss_cns_0: 0.6221, loss_yns_0: 0.1596, loss_cls_1: 1.0031, loss_box_1: 2.0905, loss_cns_1: 0.6234, loss_yns_1: 0.1607, loss_cls_2: 1.0376, loss_box_2: 2.0417, loss_cns_2: 0.6380, loss_yns_2: 0.1598, loss_cls_3: 1.0522, loss_box_3: 2.0128, loss_cns_3: 0.6425, loss_yns_3: 0.1603, loss_cls_4: 1.0852, loss_box_4: 1.9952, loss_cns_4: 0.6429, loss_yns_4: 0.1604, loss_cls_5: 1.0665, loss_box_5: 2.0076, loss_cns_5: 0.6432, loss_yns_5: 0.1596, loss_cls_dn_0: 0.2462, loss_box_dn_0: 0.8115, loss_cls_dn_1: 0.1954, loss_box_dn_1: 0.8267, loss_cls_dn_2: 0.2076, loss_box_dn_2: 0.7940, loss_cls_dn_3: 0.2069, loss_box_dn_3: 0.7959, loss_cls_dn_4: 0.2261, loss_box_dn_4: 0.7962, loss_cls_dn_5: 0.2338, loss_box_dn_5: 0.8080, loss_dense_depth: 0.8190, loss: 29.8619, grad_norm: 50.1735 -2025-11-12 21:32:23,193 - mmdet - INFO - Iter [102/17500] lr: 1.404e-04, eta: 12:51:13, time: 1.589, data_time: 0.081, memory: 49167, loss_cls_0: 0.9278, loss_box_0: 1.8186, loss_cns_0: 0.6149, loss_yns_0: 0.1604, loss_cls_1: 0.9861, loss_box_1: 2.0421, loss_cns_1: 0.6266, loss_yns_1: 0.1630, loss_cls_2: 1.0386, loss_box_2: 1.9783, loss_cns_2: 0.6401, loss_yns_2: 0.1622, loss_cls_3: 1.0469, loss_box_3: 1.9632, loss_cns_3: 0.6456, loss_yns_3: 0.1630, loss_cls_4: 1.0752, loss_box_4: 1.9511, loss_cns_4: 0.6480, loss_yns_4: 0.1640, loss_cls_5: 1.0474, loss_box_5: 1.9535, loss_cns_5: 0.6478, loss_yns_5: 0.1612, loss_cls_dn_0: 0.2467, loss_box_dn_0: 0.8229, loss_cls_dn_1: 0.1971, loss_box_dn_1: 0.8132, loss_cls_dn_2: 0.2096, loss_box_dn_2: 0.7803, loss_cls_dn_3: 0.2015, loss_box_dn_3: 0.7937, loss_cls_dn_4: 0.2195, loss_box_dn_4: 0.8033, loss_cls_dn_5: 0.2264, loss_box_dn_5: 0.8119, loss_dense_depth: 0.8086, loss: 29.5602, grad_norm: 40.4903 -2025-11-12 21:32:24,730 - mmdet - INFO - Iter [103/17500] lr: 1.408e-04, eta: 12:48:01, time: 1.538, data_time: 0.100, memory: 49167, loss_cls_0: 0.9159, loss_box_0: 1.8087, loss_cns_0: 0.6209, loss_yns_0: 0.1607, loss_cls_1: 0.9929, loss_box_1: 1.9625, loss_cns_1: 0.6352, loss_yns_1: 0.1644, loss_cls_2: 1.0262, loss_box_2: 1.8948, loss_cns_2: 0.6476, loss_yns_2: 0.1638, loss_cls_3: 1.0599, loss_box_3: 1.8875, loss_cns_3: 0.6525, loss_yns_3: 0.1641, loss_cls_4: 1.0565, loss_box_4: 1.9193, loss_cns_4: 0.6551, loss_yns_4: 0.1671, loss_cls_5: 1.0447, loss_box_5: 1.9122, loss_cns_5: 0.6510, loss_yns_5: 0.1638, loss_cls_dn_0: 0.2402, loss_box_dn_0: 0.8100, loss_cls_dn_1: 0.1863, loss_box_dn_1: 0.8046, loss_cls_dn_2: 0.1948, loss_box_dn_2: 0.7773, loss_cls_dn_3: 0.1951, loss_box_dn_3: 0.7887, loss_cls_dn_4: 0.2072, loss_box_dn_4: 0.8141, loss_cls_dn_5: 0.2119, loss_box_dn_5: 0.8126, loss_dense_depth: 0.8035, loss: 29.1737, grad_norm: 47.3091 -2025-11-12 21:32:26,251 - mmdet - INFO - Iter [104/17500] lr: 1.412e-04, eta: 12:44:49, time: 1.519, data_time: 0.074, memory: 49167, loss_cls_0: 0.9130, loss_box_0: 1.8160, loss_cns_0: 0.6213, loss_yns_0: 0.1584, loss_cls_1: 0.9968, loss_box_1: 1.9807, loss_cns_1: 0.6327, loss_yns_1: 0.1608, loss_cls_2: 1.0204, loss_box_2: 1.9295, loss_cns_2: 0.6467, loss_yns_2: 0.1627, loss_cls_3: 1.0757, loss_box_3: 1.9116, loss_cns_3: 0.6532, loss_yns_3: 0.1625, loss_cls_4: 1.0557, loss_box_4: 1.9536, loss_cns_4: 0.6547, loss_yns_4: 0.1633, loss_cls_5: 1.0737, loss_box_5: 1.9622, loss_cns_5: 0.6505, loss_yns_5: 0.1614, loss_cls_dn_0: 0.2427, loss_box_dn_0: 0.8157, loss_cls_dn_1: 0.1817, loss_box_dn_1: 0.8150, loss_cls_dn_2: 0.1879, loss_box_dn_2: 0.7956, loss_cls_dn_3: 0.1952, loss_box_dn_3: 0.7961, loss_cls_dn_4: 0.1987, loss_box_dn_4: 0.8198, loss_cls_dn_5: 0.2031, loss_box_dn_5: 0.8171, loss_dense_depth: 0.7914, loss: 29.3772, grad_norm: 66.3213 -2025-11-12 21:32:27,792 - mmdet - INFO - Iter [105/17500] lr: 1.416e-04, eta: 12:41:45, time: 1.543, data_time: 0.081, memory: 49167, loss_cls_0: 0.9099, loss_box_0: 1.7737, loss_cns_0: 0.6264, loss_yns_0: 0.1568, loss_cls_1: 0.9612, loss_box_1: 1.9402, loss_cns_1: 0.6327, loss_yns_1: 0.1580, loss_cls_2: 1.0106, loss_box_2: 1.8865, loss_cns_2: 0.6501, loss_yns_2: 0.1606, loss_cls_3: 1.0437, loss_box_3: 1.8492, loss_cns_3: 0.6559, loss_yns_3: 0.1603, loss_cls_4: 1.0488, loss_box_4: 1.8700, loss_cns_4: 0.6572, loss_yns_4: 0.1607, loss_cls_5: 1.0551, loss_box_5: 1.8760, loss_cns_5: 0.6538, loss_yns_5: 0.1588, loss_cls_dn_0: 0.2374, loss_box_dn_0: 0.8097, loss_cls_dn_1: 0.1802, loss_box_dn_1: 0.8289, loss_cls_dn_2: 0.1883, loss_box_dn_2: 0.8120, loss_cls_dn_3: 0.1945, loss_box_dn_3: 0.7985, loss_cls_dn_4: 0.1998, loss_box_dn_4: 0.8044, loss_cls_dn_5: 0.2007, loss_box_dn_5: 0.8047, loss_dense_depth: 0.7765, loss: 28.8917, grad_norm: 45.3372 -2025-11-12 21:32:29,348 - mmdet - INFO - Iter [106/17500] lr: 1.420e-04, eta: 12:38:46, time: 1.555, data_time: 0.112, memory: 49167, loss_cls_0: 0.8980, loss_box_0: 1.7475, loss_cns_0: 0.6189, loss_yns_0: 0.1554, loss_cls_1: 0.9532, loss_box_1: 1.9304, loss_cns_1: 0.6349, loss_yns_1: 0.1620, loss_cls_2: 1.0113, loss_box_2: 1.8711, loss_cns_2: 0.6503, loss_yns_2: 0.1647, loss_cls_3: 1.0255, loss_box_3: 1.8387, loss_cns_3: 0.6550, loss_yns_3: 0.1614, loss_cls_4: 1.0609, loss_box_4: 1.8476, loss_cns_4: 0.6574, loss_yns_4: 0.1618, loss_cls_5: 1.0306, loss_box_5: 1.8408, loss_cns_5: 0.6531, loss_yns_5: 0.1596, loss_cls_dn_0: 0.2366, loss_box_dn_0: 0.7929, loss_cls_dn_1: 0.1779, loss_box_dn_1: 0.7821, loss_cls_dn_2: 0.1880, loss_box_dn_2: 0.7714, loss_cls_dn_3: 0.1865, loss_box_dn_3: 0.7560, loss_cls_dn_4: 0.1971, loss_box_dn_4: 0.7536, loss_cls_dn_5: 0.2024, loss_box_dn_5: 0.7597, loss_dense_depth: 0.7818, loss: 28.4759, grad_norm: 37.6388 -2025-11-12 21:32:30,870 - mmdet - INFO - Iter [107/17500] lr: 1.424e-04, eta: 12:35:46, time: 1.524, data_time: 0.080, memory: 49167, loss_cls_0: 0.9038, loss_box_0: 1.7091, loss_cns_0: 0.6079, loss_yns_0: 0.1532, loss_cls_1: 0.9640, loss_box_1: 1.9290, loss_cns_1: 0.6349, loss_yns_1: 0.1598, loss_cls_2: 1.0135, loss_box_2: 1.8616, loss_cns_2: 0.6507, loss_yns_2: 0.1609, loss_cls_3: 1.0458, loss_box_3: 1.8401, loss_cns_3: 0.6567, loss_yns_3: 0.1592, loss_cls_4: 1.0714, loss_box_4: 1.8551, loss_cns_4: 0.6584, loss_yns_4: 0.1618, loss_cls_5: 1.0299, loss_box_5: 1.8374, loss_cns_5: 0.6543, loss_yns_5: 0.1577, loss_cls_dn_0: 0.2395, loss_box_dn_0: 0.8141, loss_cls_dn_1: 0.1818, loss_box_dn_1: 0.7670, loss_cls_dn_2: 0.1897, loss_box_dn_2: 0.7555, loss_cls_dn_3: 0.1925, loss_box_dn_3: 0.7570, loss_cls_dn_4: 0.2035, loss_box_dn_4: 0.7655, loss_cls_dn_5: 0.2183, loss_box_dn_5: 0.7807, loss_dense_depth: 0.7910, loss: 28.5322, grad_norm: 50.5213 -2025-11-12 21:32:32,387 - mmdet - INFO - Iter [108/17500] lr: 1.428e-04, eta: 12:32:48, time: 1.515, data_time: 0.078, memory: 49167, loss_cls_0: 0.9102, loss_box_0: 1.7200, loss_cns_0: 0.6080, loss_yns_0: 0.1535, loss_cls_1: 0.9607, loss_box_1: 1.9588, loss_cns_1: 0.6262, loss_yns_1: 0.1644, loss_cls_2: 1.0134, loss_box_2: 1.9025, loss_cns_2: 0.6477, loss_yns_2: 0.1636, loss_cls_3: 1.0610, loss_box_3: 1.8938, loss_cns_3: 0.6550, loss_yns_3: 0.1646, loss_cls_4: 1.0747, loss_box_4: 1.9123, loss_cns_4: 0.6544, loss_yns_4: 0.1628, loss_cls_5: 1.0346, loss_box_5: 1.9218, loss_cns_5: 0.6506, loss_yns_5: 0.1586, loss_cls_dn_0: 0.2433, loss_box_dn_0: 0.8126, loss_cls_dn_1: 0.1897, loss_box_dn_1: 0.7761, loss_cls_dn_2: 0.1938, loss_box_dn_2: 0.7622, loss_cls_dn_3: 0.2039, loss_box_dn_3: 0.7728, loss_cls_dn_4: 0.2102, loss_box_dn_4: 0.7935, loss_cls_dn_5: 0.2238, loss_box_dn_5: 0.8194, loss_dense_depth: 0.7677, loss: 28.9424, grad_norm: 57.4224 -2025-11-12 21:32:33,911 - mmdet - INFO - Iter [109/17500] lr: 1.432e-04, eta: 12:29:54, time: 1.526, data_time: 0.087, memory: 49167, loss_cls_0: 0.9300, loss_box_0: 1.7506, loss_cns_0: 0.6015, loss_yns_0: 0.1546, loss_cls_1: 0.9810, loss_box_1: 2.0279, loss_cns_1: 0.6143, loss_yns_1: 0.1604, loss_cls_2: 1.0139, loss_box_2: 1.9894, loss_cns_2: 0.6404, loss_yns_2: 0.1600, loss_cls_3: 1.0676, loss_box_3: 1.9516, loss_cns_3: 0.6451, loss_yns_3: 0.1628, loss_cls_4: 1.0507, loss_box_4: 1.9597, loss_cns_4: 0.6466, loss_yns_4: 0.1605, loss_cls_5: 1.0434, loss_box_5: 1.9838, loss_cns_5: 0.6447, loss_yns_5: 0.1600, loss_cls_dn_0: 0.2437, loss_box_dn_0: 0.8149, loss_cls_dn_1: 0.1900, loss_box_dn_1: 0.8085, loss_cls_dn_2: 0.1932, loss_box_dn_2: 0.7886, loss_cls_dn_3: 0.2006, loss_box_dn_3: 0.7907, loss_cls_dn_4: 0.2028, loss_box_dn_4: 0.8074, loss_cls_dn_5: 0.2144, loss_box_dn_5: 0.8315, loss_dense_depth: 0.7931, loss: 29.3796, grad_norm: 52.5225 -2025-11-12 21:32:35,454 - mmdet - INFO - Iter [110/17500] lr: 1.436e-04, eta: 12:27:06, time: 1.543, data_time: 0.076, memory: 49167, loss_cls_0: 0.9274, loss_box_0: 1.7849, loss_cns_0: 0.6080, loss_yns_0: 0.1561, loss_cls_1: 0.9755, loss_box_1: 2.0615, loss_cns_1: 0.6151, loss_yns_1: 0.1588, loss_cls_2: 1.0175, loss_box_2: 1.9958, loss_cns_2: 0.6388, loss_yns_2: 0.1629, loss_cls_3: 1.0631, loss_box_3: 1.9516, loss_cns_3: 0.6395, loss_yns_3: 0.1592, loss_cls_4: 1.0442, loss_box_4: 1.9480, loss_cns_4: 0.6433, loss_yns_4: 0.1609, loss_cls_5: 1.0414, loss_box_5: 1.9642, loss_cns_5: 0.6412, loss_yns_5: 0.1587, loss_cls_dn_0: 0.2440, loss_box_dn_0: 0.8045, loss_cls_dn_1: 0.1853, loss_box_dn_1: 0.7920, loss_cls_dn_2: 0.1905, loss_box_dn_2: 0.7723, loss_cls_dn_3: 0.1865, loss_box_dn_3: 0.7678, loss_cls_dn_4: 0.1982, loss_box_dn_4: 0.7711, loss_cls_dn_5: 0.2112, loss_box_dn_5: 0.7815, loss_dense_depth: 0.7940, loss: 29.2165, grad_norm: 33.1229 -2025-11-12 21:32:36,973 - mmdet - INFO - Iter [111/17500] lr: 1.440e-04, eta: 12:24:18, time: 1.517, data_time: 0.077, memory: 49167, loss_cls_0: 0.9143, loss_box_0: 1.8065, loss_cns_0: 0.6132, loss_yns_0: 0.1587, loss_cls_1: 0.9679, loss_box_1: 1.9975, loss_cns_1: 0.6222, loss_yns_1: 0.1601, loss_cls_2: 1.0272, loss_box_2: 1.9254, loss_cns_2: 0.6430, loss_yns_2: 0.1658, loss_cls_3: 1.0494, loss_box_3: 1.9164, loss_cns_3: 0.6458, loss_yns_3: 0.1599, loss_cls_4: 1.0602, loss_box_4: 1.9284, loss_cns_4: 0.6463, loss_yns_4: 0.1590, loss_cls_5: 1.0261, loss_box_5: 1.9186, loss_cns_5: 0.6439, loss_yns_5: 0.1586, loss_cls_dn_0: 0.2433, loss_box_dn_0: 0.8063, loss_cls_dn_1: 0.1788, loss_box_dn_1: 0.7727, loss_cls_dn_2: 0.1940, loss_box_dn_2: 0.7553, loss_cls_dn_3: 0.1888, loss_box_dn_3: 0.7579, loss_cls_dn_4: 0.2085, loss_box_dn_4: 0.7640, loss_cls_dn_5: 0.2174, loss_box_dn_5: 0.7589, loss_dense_depth: 0.8466, loss: 29.0070, grad_norm: 47.8034 -2025-11-12 21:32:38,489 - mmdet - INFO - Iter [112/17500] lr: 1.444e-04, eta: 12:21:32, time: 1.517, data_time: 0.073, memory: 49167, loss_cls_0: 0.9396, loss_box_0: 1.8256, loss_cns_0: 0.6096, loss_yns_0: 0.1590, loss_cls_1: 0.9910, loss_box_1: 1.9623, loss_cns_1: 0.6221, loss_yns_1: 0.1594, loss_cls_2: 1.0323, loss_box_2: 1.8974, loss_cns_2: 0.6417, loss_yns_2: 0.1644, loss_cls_3: 1.0472, loss_box_3: 1.8975, loss_cns_3: 0.6458, loss_yns_3: 0.1613, loss_cls_4: 1.0893, loss_box_4: 1.9136, loss_cns_4: 0.6469, loss_yns_4: 0.1582, loss_cls_5: 1.0327, loss_box_5: 1.9083, loss_cns_5: 0.6433, loss_yns_5: 0.1585, loss_cls_dn_0: 0.2508, loss_box_dn_0: 0.8046, loss_cls_dn_1: 0.1789, loss_box_dn_1: 0.7700, loss_cls_dn_2: 0.1965, loss_box_dn_2: 0.7578, loss_cls_dn_3: 0.1957, loss_box_dn_3: 0.7655, loss_cls_dn_4: 0.2118, loss_box_dn_4: 0.7750, loss_cls_dn_5: 0.2176, loss_box_dn_5: 0.7723, loss_dense_depth: 0.8419, loss: 29.0459, grad_norm: 54.8380 -2025-11-12 21:32:40,029 - mmdet - INFO - Iter [113/17500] lr: 1.448e-04, eta: 12:18:52, time: 1.539, data_time: 0.074, memory: 49167, loss_cls_0: 0.8944, loss_box_0: 1.8155, loss_cns_0: 0.6179, loss_yns_0: 0.1568, loss_cls_1: 0.9512, loss_box_1: 1.9335, loss_cns_1: 0.6281, loss_yns_1: 0.1567, loss_cls_2: 0.9813, loss_box_2: 1.8682, loss_cns_2: 0.6413, loss_yns_2: 0.1584, loss_cls_3: 1.0012, loss_box_3: 1.8547, loss_cns_3: 0.6468, loss_yns_3: 0.1574, loss_cls_4: 1.0513, loss_box_4: 1.8563, loss_cns_4: 0.6482, loss_yns_4: 0.1572, loss_cls_5: 0.9943, loss_box_5: 1.8547, loss_cns_5: 0.6452, loss_yns_5: 0.1559, loss_cls_dn_0: 0.2369, loss_box_dn_0: 0.8061, loss_cls_dn_1: 0.1776, loss_box_dn_1: 0.7935, loss_cls_dn_2: 0.1878, loss_box_dn_2: 0.7760, loss_cls_dn_3: 0.1911, loss_box_dn_3: 0.7762, loss_cls_dn_4: 0.1977, loss_box_dn_4: 0.7818, loss_cls_dn_5: 0.2031, loss_box_dn_5: 0.7829, loss_dense_depth: 0.8075, loss: 28.5450, grad_norm: 44.4574 -2025-11-12 21:32:41,551 - mmdet - INFO - Iter [114/17500] lr: 1.452e-04, eta: 12:16:13, time: 1.522, data_time: 0.076, memory: 49167, loss_cls_0: 0.9098, loss_box_0: 1.8222, loss_cns_0: 0.6209, loss_yns_0: 0.1583, loss_cls_1: 0.9603, loss_box_1: 1.9538, loss_cns_1: 0.6320, loss_yns_1: 0.1572, loss_cls_2: 0.9929, loss_box_2: 1.8612, loss_cns_2: 0.6457, loss_yns_2: 0.1565, loss_cls_3: 1.0060, loss_box_3: 1.8411, loss_cns_3: 0.6497, loss_yns_3: 0.1569, loss_cls_4: 1.0338, loss_box_4: 1.8316, loss_cns_4: 0.6518, loss_yns_4: 0.1585, loss_cls_5: 1.0068, loss_box_5: 1.8482, loss_cns_5: 0.6526, loss_yns_5: 0.1592, loss_cls_dn_0: 0.2415, loss_box_dn_0: 0.7991, loss_cls_dn_1: 0.1734, loss_box_dn_1: 0.7853, loss_cls_dn_2: 0.1825, loss_box_dn_2: 0.7591, loss_cls_dn_3: 0.1834, loss_box_dn_3: 0.7547, loss_cls_dn_4: 0.1931, loss_box_dn_4: 0.7608, loss_cls_dn_5: 0.2012, loss_box_dn_5: 0.7744, loss_dense_depth: 0.8240, loss: 28.4991, grad_norm: 42.3551 -2025-11-12 21:32:43,077 - mmdet - INFO - Iter [115/17500] lr: 1.456e-04, eta: 12:13:37, time: 1.527, data_time: 0.074, memory: 49167, loss_cls_0: 0.8894, loss_box_0: 1.7853, loss_cns_0: 0.6225, loss_yns_0: 0.1559, loss_cls_1: 0.9425, loss_box_1: 1.9421, loss_cns_1: 0.6332, loss_yns_1: 0.1547, loss_cls_2: 0.9840, loss_box_2: 1.8543, loss_cns_2: 0.6487, loss_yns_2: 0.1611, loss_cls_3: 1.0039, loss_box_3: 1.8548, loss_cns_3: 0.6486, loss_yns_3: 0.1593, loss_cls_4: 1.0023, loss_box_4: 1.8453, loss_cns_4: 0.6518, loss_yns_4: 0.1594, loss_cls_5: 1.0157, loss_box_5: 1.8468, loss_cns_5: 0.6518, loss_yns_5: 0.1631, loss_cls_dn_0: 0.2328, loss_box_dn_0: 0.8014, loss_cls_dn_1: 0.1735, loss_box_dn_1: 0.7767, loss_cls_dn_2: 0.1817, loss_box_dn_2: 0.7503, loss_cls_dn_3: 0.1822, loss_box_dn_3: 0.7534, loss_cls_dn_4: 0.2006, loss_box_dn_4: 0.7603, loss_cls_dn_5: 0.2060, loss_box_dn_5: 0.7676, loss_dense_depth: 0.7769, loss: 28.3399, grad_norm: 39.9046 -2025-11-12 21:32:44,605 - mmdet - INFO - Iter [116/17500] lr: 1.460e-04, eta: 12:11:04, time: 1.526, data_time: 0.076, memory: 49167, loss_cls_0: 0.9026, loss_box_0: 1.8104, loss_cns_0: 0.6187, loss_yns_0: 0.1558, loss_cls_1: 0.9666, loss_box_1: 1.9541, loss_cns_1: 0.6313, loss_yns_1: 0.1562, loss_cls_2: 0.9992, loss_box_2: 1.8794, loss_cns_2: 0.6467, loss_yns_2: 0.1638, loss_cls_3: 1.0272, loss_box_3: 1.8745, loss_cns_3: 0.6476, loss_yns_3: 0.1588, loss_cls_4: 1.0185, loss_box_4: 1.8589, loss_cns_4: 0.6495, loss_yns_4: 0.1574, loss_cls_5: 1.0126, loss_box_5: 1.8658, loss_cns_5: 0.6482, loss_yns_5: 0.1578, loss_cls_dn_0: 0.2368, loss_box_dn_0: 0.8030, loss_cls_dn_1: 0.1807, loss_box_dn_1: 0.7741, loss_cls_dn_2: 0.1888, loss_box_dn_2: 0.7498, loss_cls_dn_3: 0.1873, loss_box_dn_3: 0.7541, loss_cls_dn_4: 0.2058, loss_box_dn_4: 0.7573, loss_cls_dn_5: 0.2100, loss_box_dn_5: 0.7689, loss_dense_depth: 0.8147, loss: 28.5931, grad_norm: 32.3170 -2025-11-12 21:32:46,113 - mmdet - INFO - Iter [117/17500] lr: 1.464e-04, eta: 12:08:31, time: 1.508, data_time: 0.076, memory: 49167, loss_cls_0: 0.8938, loss_box_0: 1.7874, loss_cns_0: 0.6214, loss_yns_0: 0.1531, loss_cls_1: 0.9555, loss_box_1: 1.9679, loss_cns_1: 0.6336, loss_yns_1: 0.1561, loss_cls_2: 0.9854, loss_box_2: 1.8958, loss_cns_2: 0.6482, loss_yns_2: 0.1629, loss_cls_3: 1.0186, loss_box_3: 1.8744, loss_cns_3: 0.6515, loss_yns_3: 0.1566, loss_cls_4: 1.0103, loss_box_4: 1.8803, loss_cns_4: 0.6500, loss_yns_4: 0.1553, loss_cls_5: 1.0029, loss_box_5: 1.9166, loss_cns_5: 0.6433, loss_yns_5: 0.1547, loss_cls_dn_0: 0.2338, loss_box_dn_0: 0.8067, loss_cls_dn_1: 0.1801, loss_box_dn_1: 0.7788, loss_cls_dn_2: 0.1875, loss_box_dn_2: 0.7521, loss_cls_dn_3: 0.1867, loss_box_dn_3: 0.7501, loss_cls_dn_4: 0.1958, loss_box_dn_4: 0.7571, loss_cls_dn_5: 0.2037, loss_box_dn_5: 0.7766, loss_dense_depth: 0.7969, loss: 28.5814, grad_norm: 37.9723 -2025-11-12 21:32:47,637 - mmdet - INFO - Iter [118/17500] lr: 1.468e-04, eta: 12:06:02, time: 1.523, data_time: 0.075, memory: 49167, loss_cls_0: 0.9082, loss_box_0: 1.7725, loss_cns_0: 0.6220, loss_yns_0: 0.1556, loss_cls_1: 0.9704, loss_box_1: 1.9135, loss_cns_1: 0.6379, loss_yns_1: 0.1564, loss_cls_2: 0.9851, loss_box_2: 1.8435, loss_cns_2: 0.6485, loss_yns_2: 0.1566, loss_cls_3: 1.0121, loss_box_3: 1.8226, loss_cns_3: 0.6539, loss_yns_3: 0.1552, loss_cls_4: 1.0078, loss_box_4: 1.8263, loss_cns_4: 0.6544, loss_yns_4: 0.1565, loss_cls_5: 1.0312, loss_box_5: 1.8364, loss_cns_5: 0.6468, loss_yns_5: 0.1571, loss_cls_dn_0: 0.2353, loss_box_dn_0: 0.8041, loss_cls_dn_1: 0.1760, loss_box_dn_1: 0.7842, loss_cls_dn_2: 0.1834, loss_box_dn_2: 0.7571, loss_cls_dn_3: 0.1848, loss_box_dn_3: 0.7547, loss_cls_dn_4: 0.1887, loss_box_dn_4: 0.7639, loss_cls_dn_5: 0.2093, loss_box_dn_5: 0.7732, loss_dense_depth: 0.7949, loss: 28.3400, grad_norm: 36.1398 -2025-11-12 21:32:49,146 - mmdet - INFO - Iter [119/17500] lr: 1.472e-04, eta: 12:03:34, time: 1.509, data_time: 0.079, memory: 49167, loss_cls_0: 0.8925, loss_box_0: 1.7553, loss_cns_0: 0.6219, loss_yns_0: 0.1552, loss_cls_1: 0.9580, loss_box_1: 1.9453, loss_cns_1: 0.6332, loss_yns_1: 0.1544, loss_cls_2: 0.9773, loss_box_2: 1.8745, loss_cns_2: 0.6445, loss_yns_2: 0.1549, loss_cls_3: 1.0025, loss_box_3: 1.8507, loss_cns_3: 0.6492, loss_yns_3: 0.1547, loss_cls_4: 1.0012, loss_box_4: 1.8451, loss_cns_4: 0.6503, loss_yns_4: 0.1565, loss_cls_5: 1.0213, loss_box_5: 1.8475, loss_cns_5: 0.6456, loss_yns_5: 0.1565, loss_cls_dn_0: 0.2305, loss_box_dn_0: 0.7874, loss_cls_dn_1: 0.1688, loss_box_dn_1: 0.7734, loss_cls_dn_2: 0.1770, loss_box_dn_2: 0.7463, loss_cls_dn_3: 0.1783, loss_box_dn_3: 0.7416, loss_cls_dn_4: 0.1851, loss_box_dn_4: 0.7472, loss_cls_dn_5: 0.2004, loss_box_dn_5: 0.7499, loss_dense_depth: 0.7943, loss: 28.2285, grad_norm: 30.2528 -2025-11-12 21:32:50,657 - mmdet - INFO - Iter [120/17500] lr: 1.476e-04, eta: 12:01:08, time: 1.511, data_time: 0.082, memory: 49167, loss_cls_0: 0.8894, loss_box_0: 1.7635, loss_cns_0: 0.6201, loss_yns_0: 0.1549, loss_cls_1: 0.9712, loss_box_1: 1.9658, loss_cns_1: 0.6331, loss_yns_1: 0.1549, loss_cls_2: 0.9756, loss_box_2: 1.8940, loss_cns_2: 0.6472, loss_yns_2: 0.1585, loss_cls_3: 0.9916, loss_box_3: 1.8646, loss_cns_3: 0.6485, loss_yns_3: 0.1551, loss_cls_4: 0.9959, loss_box_4: 1.8559, loss_cns_4: 0.6507, loss_yns_4: 0.1565, loss_cls_5: 0.9999, loss_box_5: 1.8665, loss_cns_5: 0.6501, loss_yns_5: 0.1564, loss_cls_dn_0: 0.2286, loss_box_dn_0: 0.7948, loss_cls_dn_1: 0.1688, loss_box_dn_1: 0.7632, loss_cls_dn_2: 0.1753, loss_box_dn_2: 0.7313, loss_cls_dn_3: 0.1767, loss_box_dn_3: 0.7249, loss_cls_dn_4: 0.1895, loss_box_dn_4: 0.7268, loss_cls_dn_5: 0.1961, loss_box_dn_5: 0.7349, loss_dense_depth: 0.8103, loss: 28.2409, grad_norm: 31.4791 -2025-11-12 21:32:52,294 - mmdet - INFO - Iter [121/17500] lr: 1.480e-04, eta: 11:59:04, time: 1.638, data_time: 0.118, memory: 49167, loss_cls_0: 0.8982, loss_box_0: 1.7790, loss_cns_0: 0.6213, loss_yns_0: 0.1545, loss_cls_1: 0.9604, loss_box_1: 1.9855, loss_cns_1: 0.6326, loss_yns_1: 0.1569, loss_cls_2: 0.9876, loss_box_2: 1.9117, loss_cns_2: 0.6456, loss_yns_2: 0.1630, loss_cls_3: 1.0065, loss_box_3: 1.8863, loss_cns_3: 0.6510, loss_yns_3: 0.1560, loss_cls_4: 1.0097, loss_box_4: 1.8815, loss_cns_4: 0.6524, loss_yns_4: 0.1546, loss_cls_5: 1.0129, loss_box_5: 1.8899, loss_cns_5: 0.6490, loss_yns_5: 0.1542, loss_cls_dn_0: 0.2324, loss_box_dn_0: 0.8024, loss_cls_dn_1: 0.1636, loss_box_dn_1: 0.7675, loss_cls_dn_2: 0.1710, loss_box_dn_2: 0.7387, loss_cls_dn_3: 0.1782, loss_box_dn_3: 0.7342, loss_cls_dn_4: 0.1865, loss_box_dn_4: 0.7368, loss_cls_dn_5: 0.1893, loss_box_dn_5: 0.7477, loss_dense_depth: 0.8519, loss: 28.5006, grad_norm: 35.4250 -2025-11-12 21:32:53,891 - mmdet - INFO - Iter [122/17500] lr: 1.484e-04, eta: 11:56:55, time: 1.599, data_time: 0.081, memory: 49167, loss_cls_0: 0.9061, loss_box_0: 1.7599, loss_cns_0: 0.6210, loss_yns_0: 0.1578, loss_cls_1: 0.9594, loss_box_1: 1.9020, loss_cns_1: 0.6349, loss_yns_1: 0.1574, loss_cls_2: 0.9889, loss_box_2: 1.8435, loss_cns_2: 0.6459, loss_yns_2: 0.1615, loss_cls_3: 1.0111, loss_box_3: 1.8155, loss_cns_3: 0.6494, loss_yns_3: 0.1595, loss_cls_4: 1.0192, loss_box_4: 1.8104, loss_cns_4: 0.6503, loss_yns_4: 0.1603, loss_cls_5: 1.0349, loss_box_5: 1.8115, loss_cns_5: 0.6461, loss_yns_5: 0.1592, loss_cls_dn_0: 0.2342, loss_box_dn_0: 0.7988, loss_cls_dn_1: 0.1644, loss_box_dn_1: 0.7691, loss_cls_dn_2: 0.1709, loss_box_dn_2: 0.7519, loss_cls_dn_3: 0.1817, loss_box_dn_3: 0.7519, loss_cls_dn_4: 0.1806, loss_box_dn_4: 0.7559, loss_cls_dn_5: 0.1867, loss_box_dn_5: 0.7689, loss_dense_depth: 0.8186, loss: 28.1994, grad_norm: 40.2633 -2025-11-12 21:32:55,439 - mmdet - INFO - Iter [123/17500] lr: 1.488e-04, eta: 11:54:42, time: 1.546, data_time: 0.105, memory: 49167, loss_cls_0: 0.8991, loss_box_0: 1.7457, loss_cns_0: 0.6207, loss_yns_0: 0.1586, loss_cls_1: 0.9749, loss_box_1: 1.8677, loss_cns_1: 0.6401, loss_yns_1: 0.1581, loss_cls_2: 0.9831, loss_box_2: 1.8096, loss_cns_2: 0.6493, loss_yns_2: 0.1589, loss_cls_3: 1.0030, loss_box_3: 1.7935, loss_cns_3: 0.6515, loss_yns_3: 0.1590, loss_cls_4: 1.0120, loss_box_4: 1.7950, loss_cns_4: 0.6519, loss_yns_4: 0.1585, loss_cls_5: 1.0078, loss_box_5: 1.8042, loss_cns_5: 0.6477, loss_yns_5: 0.1598, loss_cls_dn_0: 0.2269, loss_box_dn_0: 0.7930, loss_cls_dn_1: 0.1665, loss_box_dn_1: 0.8029, loss_cls_dn_2: 0.1721, loss_box_dn_2: 0.7889, loss_cls_dn_3: 0.1835, loss_box_dn_3: 0.7978, loss_cls_dn_4: 0.1809, loss_box_dn_4: 0.8082, loss_cls_dn_5: 0.1882, loss_box_dn_5: 0.8254, loss_dense_depth: 0.7938, loss: 28.2375, grad_norm: 35.6131 -2025-11-12 21:32:56,963 - mmdet - INFO - Iter [124/17500] lr: 1.492e-04, eta: 11:52:27, time: 1.526, data_time: 0.077, memory: 49167, loss_cls_0: 0.8882, loss_box_0: 1.7526, loss_cns_0: 0.6249, loss_yns_0: 0.1571, loss_cls_1: 0.9693, loss_box_1: 1.8882, loss_cns_1: 0.6408, loss_yns_1: 0.1581, loss_cls_2: 0.9803, loss_box_2: 1.8295, loss_cns_2: 0.6505, loss_yns_2: 0.1563, loss_cls_3: 1.0074, loss_box_3: 1.8138, loss_cns_3: 0.6529, loss_yns_3: 0.1565, loss_cls_4: 1.0164, loss_box_4: 1.8073, loss_cns_4: 0.6517, loss_yns_4: 0.1556, loss_cls_5: 1.0020, loss_box_5: 1.8147, loss_cns_5: 0.6522, loss_yns_5: 0.1566, loss_cls_dn_0: 0.2264, loss_box_dn_0: 0.7914, loss_cls_dn_1: 0.1640, loss_box_dn_1: 0.8011, loss_cls_dn_2: 0.1661, loss_box_dn_2: 0.7825, loss_cls_dn_3: 0.1753, loss_box_dn_3: 0.7848, loss_cls_dn_4: 0.1797, loss_box_dn_4: 0.7865, loss_cls_dn_5: 0.1850, loss_box_dn_5: 0.7926, loss_dense_depth: 0.8188, loss: 28.2371, grad_norm: 34.2729 -2025-11-12 21:32:58,490 - mmdet - INFO - Iter [125/17500] lr: 1.496e-04, eta: 11:50:15, time: 1.526, data_time: 0.079, memory: 49167, loss_cls_0: 0.9054, loss_box_0: 1.7865, loss_cns_0: 0.6191, loss_yns_0: 0.1585, loss_cls_1: 0.9703, loss_box_1: 1.8551, loss_cns_1: 0.6414, loss_yns_1: 0.1593, loss_cls_2: 0.9958, loss_box_2: 1.8031, loss_cns_2: 0.6498, loss_yns_2: 0.1592, loss_cls_3: 1.0168, loss_box_3: 1.7829, loss_cns_3: 0.6524, loss_yns_3: 0.1576, loss_cls_4: 1.0228, loss_box_4: 1.7752, loss_cns_4: 0.6523, loss_yns_4: 0.1572, loss_cls_5: 1.0139, loss_box_5: 1.7816, loss_cns_5: 0.6562, loss_yns_5: 0.1554, loss_cls_dn_0: 0.2329, loss_box_dn_0: 0.7837, loss_cls_dn_1: 0.1636, loss_box_dn_1: 0.7967, loss_cls_dn_2: 0.1665, loss_box_dn_2: 0.7761, loss_cls_dn_3: 0.1703, loss_box_dn_3: 0.7705, loss_cls_dn_4: 0.1835, loss_box_dn_4: 0.7680, loss_cls_dn_5: 0.1856, loss_box_dn_5: 0.7713, loss_dense_depth: 0.8295, loss: 28.1261, grad_norm: 26.9423 -2025-11-12 21:33:00,033 - mmdet - INFO - Iter [126/17500] lr: 1.500e-04, eta: 11:48:07, time: 1.543, data_time: 0.113, memory: 49167, loss_cls_0: 0.8988, loss_box_0: 1.7671, loss_cns_0: 0.6157, loss_yns_0: 0.1557, loss_cls_1: 0.9615, loss_box_1: 1.8874, loss_cns_1: 0.6399, loss_yns_1: 0.1576, loss_cls_2: 0.9990, loss_box_2: 1.8207, loss_cns_2: 0.6469, loss_yns_2: 0.1588, loss_cls_3: 1.0173, loss_box_3: 1.8051, loss_cns_3: 0.6526, loss_yns_3: 0.1569, loss_cls_4: 1.0153, loss_box_4: 1.8041, loss_cns_4: 0.6523, loss_yns_4: 0.1569, loss_cls_5: 1.0139, loss_box_5: 1.7973, loss_cns_5: 0.6529, loss_yns_5: 0.1547, loss_cls_dn_0: 0.2337, loss_box_dn_0: 0.7857, loss_cls_dn_1: 0.1666, loss_box_dn_1: 0.7502, loss_cls_dn_2: 0.1732, loss_box_dn_2: 0.7286, loss_cls_dn_3: 0.1756, loss_box_dn_3: 0.7228, loss_cls_dn_4: 0.1883, loss_box_dn_4: 0.7232, loss_cls_dn_5: 0.1894, loss_box_dn_5: 0.7228, loss_dense_depth: 0.7870, loss: 27.9357, grad_norm: 30.2363 -2025-11-12 21:33:01,556 - mmdet - INFO - Iter [127/17500] lr: 1.504e-04, eta: 11:45:59, time: 1.525, data_time: 0.078, memory: 49167, loss_cls_0: 0.8993, loss_box_0: 1.7445, loss_cns_0: 0.6146, loss_yns_0: 0.1571, loss_cls_1: 0.9761, loss_box_1: 1.8479, loss_cns_1: 0.6399, loss_yns_1: 0.1545, loss_cls_2: 0.9966, loss_box_2: 1.7710, loss_cns_2: 0.6494, loss_yns_2: 0.1563, loss_cls_3: 1.0164, loss_box_3: 1.7580, loss_cns_3: 0.6542, loss_yns_3: 0.1553, loss_cls_4: 1.0215, loss_box_4: 1.7614, loss_cns_4: 0.6523, loss_yns_4: 0.1564, loss_cls_5: 1.0107, loss_box_5: 1.7651, loss_cns_5: 0.6504, loss_yns_5: 0.1543, loss_cls_dn_0: 0.2319, loss_box_dn_0: 0.7854, loss_cls_dn_1: 0.1629, loss_box_dn_1: 0.7363, loss_cls_dn_2: 0.1688, loss_box_dn_2: 0.7167, loss_cls_dn_3: 0.1748, loss_box_dn_3: 0.7167, loss_cls_dn_4: 0.1852, loss_box_dn_4: 0.7234, loss_cls_dn_5: 0.1852, loss_box_dn_5: 0.7309, loss_dense_depth: 0.7716, loss: 27.6528, grad_norm: 29.0152 -2025-11-12 21:33:03,080 - mmdet - INFO - Iter [128/17500] lr: 1.508e-04, eta: 11:43:52, time: 1.525, data_time: 0.076, memory: 49167, loss_cls_0: 0.9064, loss_box_0: 1.7147, loss_cns_0: 0.6178, loss_yns_0: 0.1580, loss_cls_1: 0.9702, loss_box_1: 1.8359, loss_cns_1: 0.6378, loss_yns_1: 0.1582, loss_cls_2: 1.0049, loss_box_2: 1.7764, loss_cns_2: 0.6459, loss_yns_2: 0.1567, loss_cls_3: 1.0166, loss_box_3: 1.7646, loss_cns_3: 0.6495, loss_yns_3: 0.1595, loss_cls_4: 1.0197, loss_box_4: 1.7651, loss_cns_4: 0.6520, loss_yns_4: 0.1586, loss_cls_5: 1.0117, loss_box_5: 1.7761, loss_cns_5: 0.6500, loss_yns_5: 0.1563, loss_cls_dn_0: 0.2310, loss_box_dn_0: 0.7860, loss_cls_dn_1: 0.1590, loss_box_dn_1: 0.7404, loss_cls_dn_2: 0.1657, loss_box_dn_2: 0.7335, loss_cls_dn_3: 0.1727, loss_box_dn_3: 0.7385, loss_cls_dn_4: 0.1768, loss_box_dn_4: 0.7492, loss_cls_dn_5: 0.1828, loss_box_dn_5: 0.7669, loss_dense_depth: 0.8288, loss: 27.7939, grad_norm: 33.8043 -2025-11-12 21:33:04,607 - mmdet - INFO - Iter [129/17500] lr: 1.512e-04, eta: 11:41:48, time: 1.527, data_time: 0.087, memory: 49167, loss_cls_0: 0.8934, loss_box_0: 1.7358, loss_cns_0: 0.6201, loss_yns_0: 0.1597, loss_cls_1: 0.9565, loss_box_1: 1.8273, loss_cns_1: 0.6398, loss_yns_1: 0.1590, loss_cls_2: 0.9972, loss_box_2: 1.7763, loss_cns_2: 0.6482, loss_yns_2: 0.1570, loss_cls_3: 1.0107, loss_box_3: 1.7609, loss_cns_3: 0.6518, loss_yns_3: 0.1573, loss_cls_4: 1.0164, loss_box_4: 1.7670, loss_cns_4: 0.6512, loss_yns_4: 0.1585, loss_cls_5: 0.9983, loss_box_5: 1.7843, loss_cns_5: 0.6532, loss_yns_5: 0.1570, loss_cls_dn_0: 0.2267, loss_box_dn_0: 0.7873, loss_cls_dn_1: 0.1586, loss_box_dn_1: 0.7608, loss_cls_dn_2: 0.1624, loss_box_dn_2: 0.7537, loss_cls_dn_3: 0.1680, loss_box_dn_3: 0.7594, loss_cls_dn_4: 0.1753, loss_box_dn_4: 0.7707, loss_cls_dn_5: 0.1832, loss_box_dn_5: 0.7924, loss_dense_depth: 0.7713, loss: 27.8068, grad_norm: 39.2003 -2025-11-12 21:33:06,131 - mmdet - INFO - Iter [130/17500] lr: 1.516e-04, eta: 11:39:45, time: 1.523, data_time: 0.076, memory: 49167, loss_cls_0: 0.9088, loss_box_0: 1.7669, loss_cns_0: 0.6172, loss_yns_0: 0.1558, loss_cls_1: 0.9764, loss_box_1: 1.8433, loss_cns_1: 0.6454, loss_yns_1: 0.1549, loss_cls_2: 1.0093, loss_box_2: 1.8027, loss_cns_2: 0.6525, loss_yns_2: 0.1551, loss_cls_3: 1.0151, loss_box_3: 1.7725, loss_cns_3: 0.6546, loss_yns_3: 0.1536, loss_cls_4: 1.0177, loss_box_4: 1.7741, loss_cns_4: 0.6541, loss_yns_4: 0.1568, loss_cls_5: 1.0104, loss_box_5: 1.7943, loss_cns_5: 0.6548, loss_yns_5: 0.1538, loss_cls_dn_0: 0.2299, loss_box_dn_0: 0.7916, loss_cls_dn_1: 0.1598, loss_box_dn_1: 0.7855, loss_cls_dn_2: 0.1604, loss_box_dn_2: 0.7767, loss_cls_dn_3: 0.1654, loss_box_dn_3: 0.7766, loss_cls_dn_4: 0.1741, loss_box_dn_4: 0.7832, loss_cls_dn_5: 0.1806, loss_box_dn_5: 0.8021, loss_dense_depth: 0.8215, loss: 28.1076, grad_norm: 34.6155 -2025-11-12 21:33:07,660 - mmdet - INFO - Iter [131/17500] lr: 1.520e-04, eta: 11:37:45, time: 1.528, data_time: 0.076, memory: 49167, loss_cls_0: 0.9110, loss_box_0: 1.7436, loss_cns_0: 0.6208, loss_yns_0: 0.1530, loss_cls_1: 0.9750, loss_box_1: 1.8560, loss_cns_1: 0.6477, loss_yns_1: 0.1536, loss_cls_2: 0.9953, loss_box_2: 1.8243, loss_cns_2: 0.6509, loss_yns_2: 0.1556, loss_cls_3: 1.0156, loss_box_3: 1.7931, loss_cns_3: 0.6554, loss_yns_3: 0.1536, loss_cls_4: 1.0170, loss_box_4: 1.7884, loss_cns_4: 0.6540, loss_yns_4: 0.1539, loss_cls_5: 1.0247, loss_box_5: 1.7915, loss_cns_5: 0.6533, loss_yns_5: 0.1529, loss_cls_dn_0: 0.2304, loss_box_dn_0: 0.7954, loss_cls_dn_1: 0.1574, loss_box_dn_1: 0.7636, loss_cls_dn_2: 0.1571, loss_box_dn_2: 0.7485, loss_cls_dn_3: 0.1627, loss_box_dn_3: 0.7403, loss_cls_dn_4: 0.1727, loss_box_dn_4: 0.7388, loss_cls_dn_5: 0.1789, loss_box_dn_5: 0.7457, loss_dense_depth: 0.8258, loss: 27.9576, grad_norm: 26.5820 -2025-11-12 21:33:09,187 - mmdet - INFO - Iter [132/17500] lr: 1.524e-04, eta: 11:35:44, time: 1.513, data_time: 0.075, memory: 49167, loss_cls_0: 0.8933, loss_box_0: 1.7553, loss_cns_0: 0.6222, loss_yns_0: 0.1504, loss_cls_1: 0.9534, loss_box_1: 1.8524, loss_cns_1: 0.6408, loss_yns_1: 0.1527, loss_cls_2: 0.9829, loss_box_2: 1.8162, loss_cns_2: 0.6479, loss_yns_2: 0.1527, loss_cls_3: 0.9996, loss_box_3: 1.7973, loss_cns_3: 0.6487, loss_yns_3: 0.1531, loss_cls_4: 1.0067, loss_box_4: 1.7977, loss_cns_4: 0.6503, loss_yns_4: 0.1514, loss_cls_5: 1.0026, loss_box_5: 1.7985, loss_cns_5: 0.6467, loss_yns_5: 0.1522, loss_cls_dn_0: 0.2325, loss_box_dn_0: 0.7844, loss_cls_dn_1: 0.1568, loss_box_dn_1: 0.7310, loss_cls_dn_2: 0.1594, loss_box_dn_2: 0.7129, loss_cls_dn_3: 0.1678, loss_box_dn_3: 0.7076, loss_cls_dn_4: 0.1764, loss_box_dn_4: 0.7100, loss_cls_dn_5: 0.1843, loss_box_dn_5: 0.7129, loss_dense_depth: 0.8270, loss: 27.6881, grad_norm: 35.5540 -2025-11-12 21:33:10,720 - mmdet - INFO - Iter [133/17500] lr: 1.528e-04, eta: 11:33:50, time: 1.548, data_time: 0.092, memory: 49167, loss_cls_0: 0.9009, loss_box_0: 1.7533, loss_cns_0: 0.6179, loss_yns_0: 0.1518, loss_cls_1: 0.9516, loss_box_1: 1.8658, loss_cns_1: 0.6360, loss_yns_1: 0.1526, loss_cls_2: 0.9799, loss_box_2: 1.8068, loss_cns_2: 0.6464, loss_yns_2: 0.1529, loss_cls_3: 0.9989, loss_box_3: 1.7871, loss_cns_3: 0.6472, loss_yns_3: 0.1526, loss_cls_4: 1.0009, loss_box_4: 1.7907, loss_cns_4: 0.6469, loss_yns_4: 0.1528, loss_cls_5: 1.0136, loss_box_5: 1.7966, loss_cns_5: 0.6452, loss_yns_5: 0.1538, loss_cls_dn_0: 0.2315, loss_box_dn_0: 0.7899, loss_cls_dn_1: 0.1581, loss_box_dn_1: 0.7350, loss_cls_dn_2: 0.1618, loss_box_dn_2: 0.7149, loss_cls_dn_3: 0.1658, loss_box_dn_3: 0.7107, loss_cls_dn_4: 0.1748, loss_box_dn_4: 0.7214, loss_cls_dn_5: 0.1818, loss_box_dn_5: 0.7303, loss_dense_depth: 0.8311, loss: 27.7093, grad_norm: 30.0114 -2025-11-12 21:33:12,238 - mmdet - INFO - Iter [134/17500] lr: 1.532e-04, eta: 11:31:54, time: 1.516, data_time: 0.078, memory: 49167, loss_cls_0: 0.8716, loss_box_0: 1.7176, loss_cns_0: 0.6223, loss_yns_0: 0.1531, loss_cls_1: 0.9411, loss_box_1: 1.8624, loss_cns_1: 0.6390, loss_yns_1: 0.1513, loss_cls_2: 0.9686, loss_box_2: 1.8105, loss_cns_2: 0.6479, loss_yns_2: 0.1534, loss_cls_3: 0.9842, loss_box_3: 1.7823, loss_cns_3: 0.6515, loss_yns_3: 0.1519, loss_cls_4: 0.9877, loss_box_4: 1.7833, loss_cns_4: 0.6491, loss_yns_4: 0.1525, loss_cls_5: 0.9976, loss_box_5: 1.7912, loss_cns_5: 0.6479, loss_yns_5: 0.1539, loss_cls_dn_0: 0.2255, loss_box_dn_0: 0.7861, loss_cls_dn_1: 0.1566, loss_box_dn_1: 0.7325, loss_cls_dn_2: 0.1626, loss_box_dn_2: 0.7146, loss_cls_dn_3: 0.1631, loss_box_dn_3: 0.7105, loss_cls_dn_4: 0.1736, loss_box_dn_4: 0.7251, loss_cls_dn_5: 0.1780, loss_box_dn_5: 0.7399, loss_dense_depth: 0.8059, loss: 27.5459, grad_norm: 35.4680 -2025-11-12 21:33:13,753 - mmdet - INFO - Iter [135/17500] lr: 1.536e-04, eta: 11:29:59, time: 1.517, data_time: 0.079, memory: 49167, loss_cls_0: 0.8929, loss_box_0: 1.7461, loss_cns_0: 0.6174, loss_yns_0: 0.1550, loss_cls_1: 0.9485, loss_box_1: 1.9041, loss_cns_1: 0.6389, loss_yns_1: 0.1544, loss_cls_2: 0.9799, loss_box_2: 1.8530, loss_cns_2: 0.6489, loss_yns_2: 0.1547, loss_cls_3: 0.9930, loss_box_3: 1.8260, loss_cns_3: 0.6490, loss_yns_3: 0.1549, loss_cls_4: 0.9949, loss_box_4: 1.8282, loss_cns_4: 0.6488, loss_yns_4: 0.1540, loss_cls_5: 0.9923, loss_box_5: 1.8449, loss_cns_5: 0.6470, loss_yns_5: 0.1540, loss_cls_dn_0: 0.2356, loss_box_dn_0: 0.7864, loss_cls_dn_1: 0.1558, loss_box_dn_1: 0.7561, loss_cls_dn_2: 0.1609, loss_box_dn_2: 0.7395, loss_cls_dn_3: 0.1647, loss_box_dn_3: 0.7337, loss_cls_dn_4: 0.1758, loss_box_dn_4: 0.7469, loss_cls_dn_5: 0.1824, loss_box_dn_5: 0.7634, loss_dense_depth: 0.8071, loss: 27.9889, grad_norm: 41.5121 -2025-11-12 21:33:15,272 - mmdet - INFO - Iter [136/17500] lr: 1.540e-04, eta: 11:28:06, time: 1.519, data_time: 0.077, memory: 49167, loss_cls_0: 0.8613, loss_box_0: 1.7161, loss_cns_0: 0.6212, loss_yns_0: 0.1525, loss_cls_1: 0.9174, loss_box_1: 1.8513, loss_cns_1: 0.6398, loss_yns_1: 0.1502, loss_cls_2: 0.9429, loss_box_2: 1.7955, loss_cns_2: 0.6512, loss_yns_2: 0.1512, loss_cls_3: 0.9641, loss_box_3: 1.7688, loss_cns_3: 0.6518, loss_yns_3: 0.1525, loss_cls_4: 0.9629, loss_box_4: 1.7659, loss_cns_4: 0.6536, loss_yns_4: 0.1520, loss_cls_5: 0.9623, loss_box_5: 1.7781, loss_cns_5: 0.6523, loss_yns_5: 0.1512, loss_cls_dn_0: 0.2302, loss_box_dn_0: 0.7801, loss_cls_dn_1: 0.1519, loss_box_dn_1: 0.7559, loss_cls_dn_2: 0.1533, loss_box_dn_2: 0.7406, loss_cls_dn_3: 0.1589, loss_box_dn_3: 0.7383, loss_cls_dn_4: 0.1662, loss_box_dn_4: 0.7477, loss_cls_dn_5: 0.1726, loss_box_dn_5: 0.7610, loss_dense_depth: 0.7726, loss: 27.3957, grad_norm: 31.1621 -2025-11-12 21:33:16,825 - mmdet - INFO - Iter [137/17500] lr: 1.544e-04, eta: 11:26:19, time: 1.553, data_time: 0.078, memory: 49167, loss_cls_0: 0.8785, loss_box_0: 1.7414, loss_cns_0: 0.6200, loss_yns_0: 0.1523, loss_cls_1: 0.9432, loss_box_1: 1.8958, loss_cns_1: 0.6364, loss_yns_1: 0.1515, loss_cls_2: 0.9522, loss_box_2: 1.8306, loss_cns_2: 0.6454, loss_yns_2: 0.1505, loss_cls_3: 0.9721, loss_box_3: 1.8176, loss_cns_3: 0.6479, loss_yns_3: 0.1514, loss_cls_4: 0.9729, loss_box_4: 1.8085, loss_cns_4: 0.6474, loss_yns_4: 0.1517, loss_cls_5: 0.9713, loss_box_5: 1.8063, loss_cns_5: 0.6490, loss_yns_5: 0.1512, loss_cls_dn_0: 0.2363, loss_box_dn_0: 0.7827, loss_cls_dn_1: 0.1514, loss_box_dn_1: 0.7624, loss_cls_dn_2: 0.1545, loss_box_dn_2: 0.7449, loss_cls_dn_3: 0.1597, loss_box_dn_3: 0.7458, loss_cls_dn_4: 0.1678, loss_box_dn_4: 0.7500, loss_cls_dn_5: 0.1677, loss_box_dn_5: 0.7557, loss_dense_depth: 0.7945, loss: 27.7187, grad_norm: 39.2674 -2025-11-12 21:33:18,362 - mmdet - INFO - Iter [138/17500] lr: 1.548e-04, eta: 11:24:32, time: 1.536, data_time: 0.077, memory: 49167, loss_cls_0: 0.8672, loss_box_0: 1.7083, loss_cns_0: 0.6253, loss_yns_0: 0.1505, loss_cls_1: 0.9345, loss_box_1: 1.8386, loss_cns_1: 0.6396, loss_yns_1: 0.1474, loss_cls_2: 0.9550, loss_box_2: 1.7856, loss_cns_2: 0.6486, loss_yns_2: 0.1487, loss_cls_3: 0.9606, loss_box_3: 1.7700, loss_cns_3: 0.6542, loss_yns_3: 0.1497, loss_cls_4: 0.9748, loss_box_4: 1.7696, loss_cns_4: 0.6527, loss_yns_4: 0.1490, loss_cls_5: 0.9808, loss_box_5: 1.7777, loss_cns_5: 0.6532, loss_yns_5: 0.1484, loss_cls_dn_0: 0.2311, loss_box_dn_0: 0.7824, loss_cls_dn_1: 0.1481, loss_box_dn_1: 0.7588, loss_cls_dn_2: 0.1510, loss_box_dn_2: 0.7356, loss_cls_dn_3: 0.1561, loss_box_dn_3: 0.7315, loss_cls_dn_4: 0.1650, loss_box_dn_4: 0.7358, loss_cls_dn_5: 0.1654, loss_box_dn_5: 0.7439, loss_dense_depth: 0.7963, loss: 27.3908, grad_norm: 39.1360 -2025-11-12 21:33:19,886 - mmdet - INFO - Iter [139/17500] lr: 1.552e-04, eta: 11:22:44, time: 1.526, data_time: 0.076, memory: 49167, loss_cls_0: 0.8369, loss_box_0: 1.6897, loss_cns_0: 0.6276, loss_yns_0: 0.1488, loss_cls_1: 0.9147, loss_box_1: 1.8086, loss_cns_1: 0.6421, loss_yns_1: 0.1470, loss_cls_2: 0.9240, loss_box_2: 1.7665, loss_cns_2: 0.6462, loss_yns_2: 0.1492, loss_cls_3: 0.9396, loss_box_3: 1.7551, loss_cns_3: 0.6506, loss_yns_3: 0.1500, loss_cls_4: 0.9410, loss_box_4: 1.7458, loss_cns_4: 0.6487, loss_yns_4: 0.1487, loss_cls_5: 0.9388, loss_box_5: 1.7619, loss_cns_5: 0.6502, loss_yns_5: 0.1484, loss_cls_dn_0: 0.2241, loss_box_dn_0: 0.7775, loss_cls_dn_1: 0.1493, loss_box_dn_1: 0.7328, loss_cls_dn_2: 0.1535, loss_box_dn_2: 0.7104, loss_cls_dn_3: 0.1553, loss_box_dn_3: 0.7078, loss_cls_dn_4: 0.1670, loss_box_dn_4: 0.7098, loss_cls_dn_5: 0.1695, loss_box_dn_5: 0.7189, loss_dense_depth: 0.7870, loss: 26.9430, grad_norm: 30.5983 -2025-11-12 21:33:21,403 - mmdet - INFO - Iter [140/17500] lr: 1.556e-04, eta: 11:20:57, time: 1.517, data_time: 0.082, memory: 49167, loss_cls_0: 0.8480, loss_box_0: 1.7080, loss_cns_0: 0.6264, loss_yns_0: 0.1492, loss_cls_1: 0.9130, loss_box_1: 1.7858, loss_cns_1: 0.6457, loss_yns_1: 0.1475, loss_cls_2: 0.9428, loss_box_2: 1.7269, loss_cns_2: 0.6527, loss_yns_2: 0.1480, loss_cls_3: 0.9612, loss_box_3: 1.7493, loss_cns_3: 0.6535, loss_yns_3: 0.1482, loss_cls_4: 0.9623, loss_box_4: 1.7419, loss_cns_4: 0.6552, loss_yns_4: 0.1496, loss_cls_5: 0.9587, loss_box_5: 1.7444, loss_cns_5: 0.6556, loss_yns_5: 0.1476, loss_cls_dn_0: 0.2228, loss_box_dn_0: 0.7851, loss_cls_dn_1: 0.1446, loss_box_dn_1: 0.7253, loss_cls_dn_2: 0.1484, loss_box_dn_2: 0.7065, loss_cls_dn_3: 0.1492, loss_box_dn_3: 0.7179, loss_cls_dn_4: 0.1592, loss_box_dn_4: 0.7232, loss_cls_dn_5: 0.1656, loss_box_dn_5: 0.7288, loss_dense_depth: 0.7678, loss: 26.9662, grad_norm: 50.6606 -2025-11-12 21:33:23,016 - mmdet - INFO - Iter [141/17500] lr: 1.560e-04, eta: 11:19:24, time: 1.613, data_time: 0.111, memory: 49167, loss_cls_0: 0.8957, loss_box_0: 1.7710, loss_cns_0: 0.6274, loss_yns_0: 0.1526, loss_cls_1: 0.9555, loss_box_1: 1.8980, loss_cns_1: 0.6441, loss_yns_1: 0.1488, loss_cls_2: 0.9605, loss_box_2: 1.8341, loss_cns_2: 0.6514, loss_yns_2: 0.1507, loss_cls_3: 0.9630, loss_box_3: 1.8435, loss_cns_3: 0.6527, loss_yns_3: 0.1510, loss_cls_4: 0.9797, loss_box_4: 1.8497, loss_cns_4: 0.6534, loss_yns_4: 0.1512, loss_cls_5: 0.9842, loss_box_5: 1.8507, loss_cns_5: 0.6535, loss_yns_5: 0.1525, loss_cls_dn_0: 0.2372, loss_box_dn_0: 0.7864, loss_cls_dn_1: 0.1533, loss_box_dn_1: 0.7587, loss_cls_dn_2: 0.1543, loss_box_dn_2: 0.7487, loss_cls_dn_3: 0.1585, loss_box_dn_3: 0.7639, loss_cls_dn_4: 0.1692, loss_box_dn_4: 0.7804, loss_cls_dn_5: 0.1774, loss_box_dn_5: 0.7954, loss_dense_depth: 0.8195, loss: 28.0783, grad_norm: 47.3646 -2025-11-12 21:33:24,594 - mmdet - INFO - Iter [142/17500] lr: 1.564e-04, eta: 11:17:47, time: 1.577, data_time: 0.071, memory: 49167, loss_cls_0: 0.9002, loss_box_0: 1.7885, loss_cns_0: 0.6189, loss_yns_0: 0.1538, loss_cls_1: 0.9742, loss_box_1: 1.8928, loss_cns_1: 0.6391, loss_yns_1: 0.1496, loss_cls_2: 0.9705, loss_box_2: 1.8615, loss_cns_2: 0.6403, loss_yns_2: 0.1532, loss_cls_3: 0.9834, loss_box_3: 1.8524, loss_cns_3: 0.6443, loss_yns_3: 0.1522, loss_cls_4: 0.9784, loss_box_4: 1.8591, loss_cns_4: 0.6441, loss_yns_4: 0.1523, loss_cls_5: 0.9758, loss_box_5: 1.8640, loss_cns_5: 0.6413, loss_yns_5: 0.1521, loss_cls_dn_0: 0.2337, loss_box_dn_0: 0.7888, loss_cls_dn_1: 0.1553, loss_box_dn_1: 0.7802, loss_cls_dn_2: 0.1572, loss_box_dn_2: 0.7869, loss_cls_dn_3: 0.1648, loss_box_dn_3: 0.7972, loss_cls_dn_4: 0.1749, loss_box_dn_4: 0.8154, loss_cls_dn_5: 0.1806, loss_box_dn_5: 0.8377, loss_dense_depth: 0.8056, loss: 28.3202, grad_norm: 57.4089 -2025-11-12 21:33:26,125 - mmdet - INFO - Iter [143/17500] lr: 1.568e-04, eta: 11:16:07, time: 1.531, data_time: 0.096, memory: 49167, loss_cls_0: 0.8577, loss_box_0: 1.7690, loss_cns_0: 0.6199, loss_yns_0: 0.1508, loss_cls_1: 0.9424, loss_box_1: 1.8977, loss_cns_1: 0.6451, loss_yns_1: 0.1477, loss_cls_2: 0.9549, loss_box_2: 1.8625, loss_cns_2: 0.6434, loss_yns_2: 0.1509, loss_cls_3: 0.9778, loss_box_3: 1.8496, loss_cns_3: 0.6476, loss_yns_3: 0.1492, loss_cls_4: 0.9604, loss_box_4: 1.8393, loss_cns_4: 0.6462, loss_yns_4: 0.1497, loss_cls_5: 0.9645, loss_box_5: 1.8543, loss_cns_5: 0.6448, loss_yns_5: 0.1491, loss_cls_dn_0: 0.2245, loss_box_dn_0: 0.7811, loss_cls_dn_1: 0.1540, loss_box_dn_1: 0.7867, loss_cls_dn_2: 0.1567, loss_box_dn_2: 0.7918, loss_cls_dn_3: 0.1649, loss_box_dn_3: 0.7991, loss_cls_dn_4: 0.1776, loss_box_dn_4: 0.8092, loss_cls_dn_5: 0.1835, loss_box_dn_5: 0.8340, loss_dense_depth: 0.7885, loss: 28.1260, grad_norm: 54.8365 -2025-11-12 21:33:27,638 - mmdet - INFO - Iter [144/17500] lr: 1.572e-04, eta: 11:14:25, time: 1.512, data_time: 0.074, memory: 49167, loss_cls_0: 0.8826, loss_box_0: 1.7966, loss_cns_0: 0.6182, loss_yns_0: 0.1513, loss_cls_1: 0.9660, loss_box_1: 1.9046, loss_cns_1: 0.6445, loss_yns_1: 0.1482, loss_cls_2: 0.9830, loss_box_2: 1.8551, loss_cns_2: 0.6510, loss_yns_2: 0.1516, loss_cls_3: 0.9880, loss_box_3: 1.8475, loss_cns_3: 0.6504, loss_yns_3: 0.1495, loss_cls_4: 0.9847, loss_box_4: 1.8483, loss_cns_4: 0.6504, loss_yns_4: 0.1495, loss_cls_5: 0.9883, loss_box_5: 1.8635, loss_cns_5: 0.6468, loss_yns_5: 0.1496, loss_cls_dn_0: 0.2345, loss_box_dn_0: 0.7924, loss_cls_dn_1: 0.1575, loss_box_dn_1: 0.7966, loss_cls_dn_2: 0.1568, loss_box_dn_2: 0.7864, loss_cls_dn_3: 0.1620, loss_box_dn_3: 0.7919, loss_cls_dn_4: 0.1732, loss_box_dn_4: 0.7991, loss_cls_dn_5: 0.1819, loss_box_dn_5: 0.8188, loss_dense_depth: 0.7774, loss: 28.2979, grad_norm: 41.9817 -2025-11-12 21:33:29,170 - mmdet - INFO - Iter [145/17500] lr: 1.576e-04, eta: 11:12:47, time: 1.532, data_time: 0.071, memory: 49167, loss_cls_0: 0.8724, loss_box_0: 1.7824, loss_cns_0: 0.6199, loss_yns_0: 0.1521, loss_cls_1: 0.9584, loss_box_1: 1.9575, loss_cns_1: 0.6414, loss_yns_1: 0.1504, loss_cls_2: 0.9792, loss_box_2: 1.9151, loss_cns_2: 0.6499, loss_yns_2: 0.1510, loss_cls_3: 0.9802, loss_box_3: 1.8999, loss_cns_3: 0.6499, loss_yns_3: 0.1507, loss_cls_4: 0.9831, loss_box_4: 1.9068, loss_cns_4: 0.6492, loss_yns_4: 0.1504, loss_cls_5: 0.9787, loss_box_5: 1.9158, loss_cns_5: 0.6452, loss_yns_5: 0.1510, loss_cls_dn_0: 0.2279, loss_box_dn_0: 0.7898, loss_cls_dn_1: 0.1554, loss_box_dn_1: 0.8157, loss_cls_dn_2: 0.1535, loss_box_dn_2: 0.8021, loss_cls_dn_3: 0.1588, loss_box_dn_3: 0.7994, loss_cls_dn_4: 0.1645, loss_box_dn_4: 0.8059, loss_cls_dn_5: 0.1722, loss_box_dn_5: 0.8142, loss_dense_depth: 0.7824, loss: 28.5322, grad_norm: 46.7076 -2025-11-12 21:33:30,734 - mmdet - INFO - Iter [146/17500] lr: 1.580e-04, eta: 11:11:14, time: 1.564, data_time: 0.111, memory: 49167, loss_cls_0: 0.8579, loss_box_0: 1.7580, loss_cns_0: 0.6191, loss_yns_0: 0.1520, loss_cls_1: 0.9281, loss_box_1: 1.8852, loss_cns_1: 0.6438, loss_yns_1: 0.1496, loss_cls_2: 0.9545, loss_box_2: 1.8211, loss_cns_2: 0.6515, loss_yns_2: 0.1515, loss_cls_3: 0.9622, loss_box_3: 1.8097, loss_cns_3: 0.6555, loss_yns_3: 0.1506, loss_cls_4: 0.9651, loss_box_4: 1.8103, loss_cns_4: 0.6595, loss_yns_4: 0.1521, loss_cls_5: 0.9614, loss_box_5: 1.8078, loss_cns_5: 0.6640, loss_yns_5: 0.1524, loss_cls_dn_0: 0.2221, loss_box_dn_0: 0.7783, loss_cls_dn_1: 0.1457, loss_box_dn_1: 0.7620, loss_cls_dn_2: 0.1468, loss_box_dn_2: 0.7412, loss_cls_dn_3: 0.1503, loss_box_dn_3: 0.7368, loss_cls_dn_4: 0.1569, loss_box_dn_4: 0.7438, loss_cls_dn_5: 0.1642, loss_box_dn_5: 0.7462, loss_dense_depth: 0.7621, loss: 27.5791, grad_norm: 31.9522 -2025-11-12 21:33:32,252 - mmdet - INFO - Iter [147/17500] lr: 1.584e-04, eta: 11:09:37, time: 1.517, data_time: 0.074, memory: 49167, loss_cls_0: 0.8744, loss_box_0: 1.7921, loss_cns_0: 0.6153, loss_yns_0: 0.1520, loss_cls_1: 0.9490, loss_box_1: 1.8484, loss_cns_1: 0.6451, loss_yns_1: 0.1513, loss_cls_2: 0.9726, loss_box_2: 1.7878, loss_cns_2: 0.6518, loss_yns_2: 0.1537, loss_cls_3: 0.9758, loss_box_3: 1.7922, loss_cns_3: 0.6523, loss_yns_3: 0.1532, loss_cls_4: 0.9740, loss_box_4: 1.8099, loss_cns_4: 0.6549, loss_yns_4: 0.1540, loss_cls_5: 0.9701, loss_box_5: 1.8143, loss_cns_5: 0.6542, loss_yns_5: 0.1545, loss_cls_dn_0: 0.2268, loss_box_dn_0: 0.7829, loss_cls_dn_1: 0.1470, loss_box_dn_1: 0.7501, loss_cls_dn_2: 0.1480, loss_box_dn_2: 0.7384, loss_cls_dn_3: 0.1532, loss_box_dn_3: 0.7403, loss_cls_dn_4: 0.1578, loss_box_dn_4: 0.7527, loss_cls_dn_5: 0.1637, loss_box_dn_5: 0.7567, loss_dense_depth: 0.7945, loss: 27.6650, grad_norm: 47.1476 diff --git a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20251112_212736.log.json b/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20251112_212736.log.json deleted file mode 100644 index 9863299d1ef28bd9330cd1169c18ab627f8b2d84..0000000000000000000000000000000000000000 --- a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20251112_212736.log.json +++ /dev/null @@ -1,148 +0,0 @@ -{"env_info": "sys.platform: linux\nPython: 3.10.12 (main, May 27 2025, 17:12:29) [GCC 11.4.0]\nCUDA available: True\nGPU 0,1,2,3,4,5,6,7: BW200, UBB BW1000\nCUDA_HOME: /opt/dtk\nNVCC: Not Available\nGCC: x86_64-linux-gnu-gcc (Ubuntu 11.4.0-1ubuntu1~22.04) 11.4.0\nPyTorch: 2.4.1\nPyTorch compiling details: PyTorch built with:\n - GCC 10.3\n - C++ Version: 201703\n - Intel(R) Math Kernel Library Version 2020.0.4 Product Build 20200917 for Intel(R) 64 architecture applications\n - OpenMP 201511 (a.k.a. OpenMP 4.5)\n - LAPACK is enabled (usually provided by MKL)\n - NNPACK is enabled\n - CPU capability usage: AVX512\n - HIP Runtime 6.3.25211\n - MIOpen 2.17.0\n - Magma 2.8.0\n - Build settings: BLAS_INFO=mkl, BUILD_TYPE=Release, CXX_COMPILER=/opt/rh/gcc-toolset-10/root/usr/bin/c++, CXX_FLAGS= -D_GLIBCXX_USE_CXX11_ABI=1 -fvisibility-inlines-hidden -DUSE_PTHREADPOOL -DNDEBUG -DUSE_KINETO -DLIBKINETO_NOCUPTI -DUSE_FBGEMM -DUSE_PYTORCH_QNNPACK -DUSE_XNNPACK -DSYMBOLICATE_MOBILE_DEBUG_HANDLE -O2 -fPIC -Wall -Wextra -Werror=return-type -Werror=non-virtual-dtor -Werror=bool-operation -Wnarrowing -Wno-missing-field-initializers -Wno-type-limits -Wno-array-bounds -Wno-unknown-pragmas -Wno-unused-parameter -Wno-unused-function -Wno-unused-result -Wno-strict-overflow -Wno-strict-aliasing -Wno-stringop-overflow -Wsuggest-override -Wno-psabi -Wno-error=pedantic -Wno-error=old-style-cast -Wno-missing-braces -fdiagnostics-color=always -faligned-new -Wno-unused-but-set-variable -Wno-maybe-uninitialized -fno-math-errno -fno-trapping-math -Werror=format -Wno-stringop-overflow, FORCE_FALLBACK_CUDA_MPI=1, LAPACK_INFO=mkl, PERF_WITH_AVX=1, PERF_WITH_AVX2=1, PERF_WITH_AVX512=1, TORCH_VERSION=2.4.1, USE_CUDA=0, USE_CUDNN=OFF, USE_CUSPARSELT=OFF, USE_EXCEPTION_PTR=1, USE_GFLAGS=1, USE_GLOG=1, USE_GLOO=1, USE_MKL=ON, USE_MKLDNN=0, USE_MPI=1, USE_NCCL=1, USE_NNPACK=ON, USE_OPENMP=1, USE_ROCM=ON, USE_ROCM_KERNEL_ASSERT=OFF, \n\nTorchVision: 0.19.1\nOpenCV: 4.11.0\nMMCV: 1.6.1\nMMCV Compiler: GCC 11.4\nMMCV CUDA Compiler: rocm not available\nMMDetection: 2.26.0+c41df4b", "config": "plugin = True\nplugin_dir = 'projects/mmdet3d_plugin/'\ndist_params = dict(backend='nccl')\nlog_level = 'INFO'\nwork_dir = './work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704'\ntotal_batch_size = 160\nnum_gpus = 8\nbatch_size = 20\nnum_iters_per_epoch = 175\nnum_epochs = 100\ncheckpoint_epoch_interval = 20\ncheckpoint_config = dict(interval=3500)\nlog_config = dict(\n interval=1,\n hooks=[\n dict(type='TextLoggerHook', by_epoch=False),\n dict(type='TensorboardLoggerHook')\n ])\nload_from = None\nresume_from = None\nworkflow = [('train', 1)]\nfp16 = dict(loss_scale=32.0)\ninput_shape = (704, 256)\ntracking_test = True\ntracking_threshold = 0.2\nclass_names = [\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier',\n 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n]\nnum_classes = 10\nembed_dims = 256\nnum_groups = 8\nnum_decoder = 6\nnum_single_frame_decoder = 1\nuse_deformable_func = True\nstrides = [4, 8, 16, 32]\nnum_levels = 4\nnum_depth_layers = 3\ndrop_out = 0.1\ntemporal = True\ndecouple_attn = True\nwith_quality_estimation = True\nmodel = dict(\n type='Sparse4D',\n use_grid_mask=True,\n use_deformable_func=True,\n img_backbone=dict(\n type='ResNet',\n depth=50,\n num_stages=4,\n frozen_stages=-1,\n norm_eval=False,\n style='pytorch',\n with_cp=True,\n out_indices=(0, 1, 2, 3),\n norm_cfg=dict(type='BN', requires_grad=True),\n pretrained='ckpt/resnet50-19c8e357.pth'),\n img_neck=dict(\n type='FPN',\n num_outs=4,\n start_level=0,\n out_channels=256,\n add_extra_convs='on_output',\n relu_before_extra_convs=True,\n in_channels=[256, 512, 1024, 2048]),\n depth_branch=dict(\n type='DenseDepthNet',\n embed_dims=256,\n num_depth_layers=3,\n loss_weight=0.2),\n head=dict(\n type='Sparse4DHead',\n cls_threshold_to_reg=0.05,\n decouple_attn=True,\n instance_bank=dict(\n type='InstanceBank',\n num_anchor=900,\n embed_dims=256,\n anchor='nuscenes_kmeans900.npy',\n anchor_handler=dict(type='SparseBox3DKeyPointsGenerator'),\n num_temp_instances=600,\n confidence_decay=0.6,\n feat_grad=False),\n anchor_encoder=dict(\n type='SparseBox3DEncoder',\n vel_dims=3,\n embed_dims=[128, 32, 32, 64],\n mode='cat',\n output_fc=False,\n in_loops=1,\n out_loops=4),\n num_single_frame_decoder=1,\n operation_order=[\n 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm',\n 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm',\n 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm',\n 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm',\n 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm',\n 'deformable', 'ffn', 'norm', 'refine'\n ],\n temp_graph_model=dict(\n type='MultiheadAttention',\n embed_dims=512,\n num_heads=8,\n batch_first=True,\n dropout=0.1),\n graph_model=dict(\n type='MultiheadAttention',\n embed_dims=512,\n num_heads=8,\n batch_first=True,\n dropout=0.1),\n norm_layer=dict(type='LN', normalized_shape=256),\n ffn=dict(\n type='AsymmetricFFN',\n in_channels=512,\n pre_norm=dict(type='LN'),\n embed_dims=256,\n feedforward_channels=1024,\n num_fcs=2,\n ffn_drop=0.1,\n act_cfg=dict(type='ReLU', inplace=True)),\n deformable_model=dict(\n type='DeformableFeatureAggregation',\n embed_dims=256,\n num_groups=8,\n num_levels=4,\n num_cams=6,\n attn_drop=0.15,\n use_deformable_func=True,\n use_camera_embed=True,\n residual_mode='cat',\n kps_generator=dict(\n type='SparseBox3DKeyPointsGenerator',\n num_learnable_pts=6,\n fix_scale=[[0, 0, 0], [0.45, 0, 0], [-0.45, 0, 0],\n [0, 0.45, 0], [0, -0.45, 0], [0, 0, 0.45],\n [0, 0, -0.45]])),\n refine_layer=dict(\n type='SparseBox3DRefinementModule',\n embed_dims=256,\n num_cls=10,\n refine_yaw=True,\n with_quality_estimation=True),\n sampler=dict(\n type='SparseBox3DTarget',\n num_dn_groups=5,\n num_temp_dn_groups=3,\n dn_noise_scale=[2.0, 2.0, 2.0, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5],\n max_dn_gt=32,\n add_neg_dn=True,\n cls_weight=2.0,\n box_weight=0.25,\n reg_weights=[2.0, 2.0, 2.0, 0.5, 0.5, 0.5, 0.0, 0.0, 0.0, 0.0],\n cls_wise_reg_weights=dict(\n {9: [2.0, 2.0, 2.0, 1.0, 1.0, 1.0, 0.0, 0.0, 1.0, 1.0]})),\n loss_cls=dict(\n type='FocalLoss',\n use_sigmoid=True,\n gamma=2.0,\n alpha=0.25,\n loss_weight=2.0),\n loss_reg=dict(\n type='SparseBox3DLoss',\n loss_box=dict(type='L1Loss', loss_weight=0.25),\n loss_centerness=dict(type='CrossEntropyLoss', use_sigmoid=True),\n loss_yawness=dict(type='GaussianFocalLoss'),\n cls_allow_reverse=[5]),\n decoder=dict(type='SparseBox3DDecoder'),\n reg_weights=[2.0, 2.0, 2.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0]))\ndataset_type = 'NuScenes3DDetTrackDataset'\ndata_root = 'data/nuscenes/'\nanno_root = 'data/nuscenes_anno_pkls/'\nfile_client_args = dict(backend='disk')\nimg_norm_cfg = dict(\n mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375], to_rgb=True)\ntrain_pipeline = [\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(\n type='LoadPointsFromFile',\n coord_type='LIDAR',\n load_dim=5,\n use_dim=5,\n file_client_args=dict(backend='disk')),\n dict(type='ResizeCropFlipImage'),\n dict(type='MultiScaleDepthMapGenerator', downsample=[4, 8, 16]),\n dict(type='BBoxRotation'),\n dict(type='PhotoMetricDistortionMultiViewImage'),\n dict(\n type='NormalizeMultiviewImage',\n mean=[123.675, 116.28, 103.53],\n std=[58.395, 57.12, 57.375],\n to_rgb=True),\n dict(\n type='CircleObjectRangeFilter',\n class_dist_thred=[55, 55, 55, 55, 55, 55, 55, 55, 55, 55]),\n dict(\n type='InstanceNameFilter',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ]),\n dict(type='NuScenesSparse4DAdaptor'),\n dict(\n type='Collect',\n keys=[\n 'img', 'timestamp', 'projection_mat', 'image_wh', 'gt_depth',\n 'focal', 'gt_bboxes_3d', 'gt_labels_3d'\n ],\n meta_keys=['T_global', 'T_global_inv', 'timestamp', 'instance_id'])\n]\ntest_pipeline = [\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(type='ResizeCropFlipImage'),\n dict(\n type='NormalizeMultiviewImage',\n mean=[123.675, 116.28, 103.53],\n std=[58.395, 57.12, 57.375],\n to_rgb=True),\n dict(type='NuScenesSparse4DAdaptor'),\n dict(\n type='Collect',\n keys=['img', 'timestamp', 'projection_mat', 'image_wh'],\n meta_keys=['T_global', 'T_global_inv', 'timestamp'])\n]\ninput_modality = dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False)\ndata_basic_config = dict(\n type='NuScenes3DDetTrackDataset',\n data_root='data/nuscenes/',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier',\n 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ],\n modality=dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False),\n version='v1.0-trainval')\ndata_aug_conf = dict(\n resize_lim=(0.4, 0.47),\n final_dim=(256, 704),\n bot_pct_lim=(0.0, 0.0),\n rot_lim=(-5.4, 5.4),\n H=900,\n W=1600,\n rand_flip=True,\n rot3d_range=[-0.3925, 0.3925])\ndata = dict(\n samples_per_gpu=20,\n workers_per_gpu=20,\n train=dict(\n type='NuScenes3DDetTrackDataset',\n data_root='data/nuscenes/',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ],\n modality=dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False),\n version='v1.0-trainval',\n ann_file='data/nuscenes_anno_pkls/nuscenes_infos_train.pkl',\n pipeline=[\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(\n type='LoadPointsFromFile',\n coord_type='LIDAR',\n load_dim=5,\n use_dim=5,\n file_client_args=dict(backend='disk')),\n dict(type='ResizeCropFlipImage'),\n dict(type='MultiScaleDepthMapGenerator', downsample=[4, 8, 16]),\n dict(type='BBoxRotation'),\n dict(type='PhotoMetricDistortionMultiViewImage'),\n dict(\n type='NormalizeMultiviewImage',\n mean=[123.675, 116.28, 103.53],\n std=[58.395, 57.12, 57.375],\n to_rgb=True),\n dict(\n type='CircleObjectRangeFilter',\n class_dist_thred=[55, 55, 55, 55, 55, 55, 55, 55, 55, 55]),\n dict(\n type='InstanceNameFilter',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian',\n 'traffic_cone'\n ]),\n dict(type='NuScenesSparse4DAdaptor'),\n dict(\n type='Collect',\n keys=[\n 'img', 'timestamp', 'projection_mat', 'image_wh',\n 'gt_depth', 'focal', 'gt_bboxes_3d', 'gt_labels_3d'\n ],\n meta_keys=[\n 'T_global', 'T_global_inv', 'timestamp', 'instance_id'\n ])\n ],\n test_mode=False,\n data_aug_conf=dict(\n resize_lim=(0.4, 0.47),\n final_dim=(256, 704),\n bot_pct_lim=(0.0, 0.0),\n rot_lim=(-5.4, 5.4),\n H=900,\n W=1600,\n rand_flip=True,\n rot3d_range=[-0.3925, 0.3925]),\n with_seq_flag=True,\n sequences_split_num=2,\n keep_consistent_seq_aug=True),\n val=dict(\n type='NuScenes3DDetTrackDataset',\n data_root='data/nuscenes/',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ],\n modality=dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False),\n version='v1.0-trainval',\n ann_file='data/nuscenes_anno_pkls/nuscenes_infos_val.pkl',\n pipeline=[\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(type='ResizeCropFlipImage'),\n dict(\n type='NormalizeMultiviewImage',\n mean=[123.675, 116.28, 103.53],\n std=[58.395, 57.12, 57.375],\n to_rgb=True),\n dict(type='NuScenesSparse4DAdaptor'),\n dict(\n type='Collect',\n keys=['img', 'timestamp', 'projection_mat', 'image_wh'],\n meta_keys=['T_global', 'T_global_inv', 'timestamp'])\n ],\n data_aug_conf=dict(\n resize_lim=(0.4, 0.47),\n final_dim=(256, 704),\n bot_pct_lim=(0.0, 0.0),\n rot_lim=(-5.4, 5.4),\n H=900,\n W=1600,\n rand_flip=True,\n rot3d_range=[-0.3925, 0.3925]),\n test_mode=True,\n tracking=True,\n tracking_threshold=0.2),\n test=dict(\n type='NuScenes3DDetTrackDataset',\n data_root='data/nuscenes/',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ],\n modality=dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False),\n version='v1.0-trainval',\n ann_file='data/nuscenes_anno_pkls/nuscenes_infos_val.pkl',\n pipeline=[\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(type='ResizeCropFlipImage'),\n dict(\n type='NormalizeMultiviewImage',\n mean=[123.675, 116.28, 103.53],\n std=[58.395, 57.12, 57.375],\n to_rgb=True),\n dict(type='NuScenesSparse4DAdaptor'),\n dict(\n type='Collect',\n keys=['img', 'timestamp', 'projection_mat', 'image_wh'],\n meta_keys=['T_global', 'T_global_inv', 'timestamp'])\n ],\n data_aug_conf=dict(\n resize_lim=(0.4, 0.47),\n final_dim=(256, 704),\n bot_pct_lim=(0.0, 0.0),\n rot_lim=(-5.4, 5.4),\n H=900,\n W=1600,\n rand_flip=True,\n rot3d_range=[-0.3925, 0.3925]),\n test_mode=True,\n tracking=True,\n tracking_threshold=0.2))\noptimizer = dict(\n type='Miopen_AdamW',\n lr=0.0006,\n weight_decay=0.001,\n paramwise_cfg=dict(custom_keys=dict(img_backbone=dict(lr_mult=0.5))))\noptimizer_config = dict(grad_clip=dict(max_norm=25, norm_type=2))\nlr_config = dict(\n policy='CosineAnnealing',\n warmup='linear',\n warmup_iters=500,\n warmup_ratio=0.3333333333333333,\n min_lr_ratio=0.001)\nrunner = dict(type='IterBasedRunner', max_iters=17500)\nvis_pipeline = [\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(type='Collect', keys=['img'], meta_keys=['timestamp', 'lidar2img'])\n]\nevaluation = dict(\n interval=3500,\n pipeline=[\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(\n type='Collect', keys=['img'], meta_keys=['timestamp', 'lidar2img'])\n ])\ngpu_ids = range(0, 8)\n", "seed": 0, "exp_name": "sparse4dv3_temporal_r50_1x8_bs6_256x704.py"} -{"mode": "train", "epoch": 1, "iter": 1, "lr": 0.0001, "memory": 49167, "data_time": 10.29133, "loss_cls_0": 2.36103, "loss_box_0": 0.01385, "loss_cns_0": 0.0027, "loss_yns_0": 0.00079, "loss_cls_1": 2.15435, "loss_box_1": 0.10895, "loss_cns_1": 0.02472, "loss_yns_1": 0.00672, "loss_cls_2": 2.31219, "loss_box_2": 0.00414, "loss_cns_2": 0.00052, "loss_yns_2": 0.00027, "loss_cls_3": 2.38992, "loss_box_3": 0.02945, "loss_cns_3": 0.00504, "loss_yns_3": 0.00144, "loss_cls_4": 2.02781, "loss_box_4": 0.41601, "loss_cns_4": 0.05344, "loss_yns_4": 0.02517, "loss_cls_5": 2.42458, "loss_box_5": 0.01797, "loss_cns_5": 0.00217, "loss_yns_5": 0.00161, "loss_cls_dn_0": 1.19802, "loss_box_dn_0": 1.46029, "loss_cls_dn_1": 1.11019, "loss_box_dn_1": 1.73175, "loss_cls_dn_2": 1.17413, "loss_box_dn_2": 1.97184, "loss_cls_dn_3": 1.17206, "loss_box_dn_3": 2.24179, "loss_cls_dn_4": 1.05279, "loss_box_dn_4": 2.42683, "loss_cls_dn_5": 1.23865, "loss_box_dn_5": 2.67724, "loss_dense_depth": 1.86432, "loss": 35.70474, "grad_norm": 267.68921, "time": 115.63506} -{"mode": "train", "epoch": 1, "iter": 2, "lr": 0.0001, "memory": 49167, "data_time": 0.08859, "loss_cls_0": 2.04959, "loss_box_0": 0.01016, "loss_cns_0": 0.00251, "loss_yns_0": 0.00102, "loss_cls_1": 1.99472, "loss_box_1": 0.2676, "loss_cns_1": 0.04895, "loss_yns_1": 0.01313, "loss_cls_2": 2.10069, "loss_box_2": 0.22799, "loss_cns_2": 0.0216, "loss_yns_2": 0.01038, "loss_cls_3": 1.94967, "loss_box_3": 0.41338, "loss_cns_3": 0.05512, "loss_yns_3": 0.01699, "loss_cls_4": 1.79749, "loss_box_4": 1.60688, "loss_cns_4": 0.16379, "loss_yns_4": 0.05751, "loss_cls_5": 2.0481, "loss_box_5": 0.63597, "loss_cns_5": 0.07048, "loss_yns_5": 0.02256, "loss_cls_dn_0": 1.02954, "loss_box_dn_0": 1.24643, "loss_cls_dn_1": 0.94946, "loss_box_dn_1": 2.41103, "loss_cls_dn_2": 0.96799, "loss_box_dn_2": 2.53158, "loss_cls_dn_3": 0.9124, "loss_box_dn_3": 2.60585, "loss_cls_dn_4": 0.83892, "loss_box_dn_4": 2.87083, "loss_cls_dn_5": 0.98559, "loss_box_dn_5": 3.10529, "loss_dense_depth": 1.71618, "loss": 37.7574, "grad_norm": 66.36814, "time": 2.00116} -{"mode": "train", "epoch": 1, "iter": 3, "lr": 0.0001, "memory": 49167, "data_time": 0.082, "loss_cls_0": 1.41315, "loss_box_0": 2.61729, "loss_cns_0": 0.61923, "loss_yns_0": 0.19824, "loss_cls_1": 1.70961, "loss_box_1": 2.53403, "loss_cns_1": 0.36945, "loss_yns_1": 0.13329, "loss_cls_2": 1.74693, "loss_box_2": 4.42698, "loss_cns_2": 0.38668, "loss_yns_2": 0.21346, "loss_cls_3": 1.60623, "loss_box_3": 5.18864, "loss_cns_3": 0.44866, "loss_yns_3": 0.21839, "loss_cls_4": 1.5105, "loss_box_4": 5.11905, "loss_cns_4": 0.42697, "loss_yns_4": 0.20427, "loss_cls_5": 1.62095, "loss_box_5": 4.10937, "loss_cns_5": 0.30011, "loss_yns_5": 0.13297, "loss_cls_dn_0": 0.66783, "loss_box_dn_0": 1.21488, "loss_cls_dn_1": 0.79842, "loss_box_dn_1": 2.38154, "loss_cls_dn_2": 0.76693, "loss_box_dn_2": 2.58008, "loss_cls_dn_3": 0.67273, "loss_box_dn_3": 2.7835, "loss_cls_dn_4": 0.68279, "loss_box_dn_4": 3.03319, "loss_cls_dn_5": 0.7734, "loss_box_dn_5": 3.31178, "loss_dense_depth": 1.65695, "loss": 58.57849, "grad_norm": 108.36858, "time": 1.50449} -{"mode": "train", "epoch": 1, "iter": 4, "lr": 0.0001, "memory": 49167, "data_time": 0.08209, "loss_cls_0": 1.3802, "loss_box_0": 2.6899, "loss_cns_0": 0.50076, "loss_yns_0": 0.22833, "loss_cls_1": 1.56039, "loss_box_1": 3.76313, "loss_cns_1": 0.43048, "loss_yns_1": 0.20171, "loss_cls_2": 1.69398, "loss_box_2": 3.77664, "loss_cns_2": 0.43337, "loss_yns_2": 0.19762, "loss_cls_3": 1.46527, "loss_box_3": 4.23309, "loss_cns_3": 0.46352, "loss_yns_3": 0.20197, "loss_cls_4": 1.47916, "loss_box_4": 4.65133, "loss_cns_4": 0.43682, "loss_yns_4": 0.18892, "loss_cls_5": 1.40369, "loss_box_5": 5.18859, "loss_cns_5": 0.4498, "loss_yns_5": 0.1981, "loss_cls_dn_0": 0.52332, "loss_box_dn_0": 1.24609, "loss_cls_dn_1": 0.66232, "loss_box_dn_1": 2.56212, "loss_cls_dn_2": 0.64381, "loss_box_dn_2": 2.66859, "loss_cls_dn_3": 0.58051, "loss_box_dn_3": 2.92762, "loss_cls_dn_4": 0.5414, "loss_box_dn_4": 3.17099, "loss_cls_dn_5": 0.59696, "loss_box_dn_5": 3.41576, "loss_dense_depth": 1.56823, "loss": 58.32451, "grad_norm": 114.88149, "time": 1.51688} -{"mode": "train", "epoch": 1, "iter": 5, "lr": 0.0001, "memory": 49167, "data_time": 0.08246, "loss_cls_0": 1.37766, "loss_box_0": 2.80981, "loss_cns_0": 0.55854, "loss_yns_0": 0.18423, "loss_cls_1": 1.57334, "loss_box_1": 4.16582, "loss_cns_1": 0.41773, "loss_yns_1": 0.21609, "loss_cls_2": 1.46878, "loss_box_2": 4.17793, "loss_cns_2": 0.42287, "loss_yns_2": 0.19235, "loss_cls_3": 1.36351, "loss_box_3": 4.21469, "loss_cns_3": 0.42096, "loss_yns_3": 0.20586, "loss_cls_4": 1.32204, "loss_box_4": 4.47368, "loss_cns_4": 0.3968, "loss_yns_4": 0.20179, "loss_cls_5": 1.36381, "loss_box_5": 4.56141, "loss_cns_5": 0.38582, "loss_yns_5": 0.21901, "loss_cls_dn_0": 0.55894, "loss_box_dn_0": 1.28222, "loss_cls_dn_1": 0.59626, "loss_box_dn_1": 2.43165, "loss_cls_dn_2": 0.63145, "loss_box_dn_2": 2.56069, "loss_cls_dn_3": 0.53079, "loss_box_dn_3": 2.65999, "loss_cls_dn_4": 0.5404, "loss_box_dn_4": 2.88496, "loss_cls_dn_5": 0.50918, "loss_box_dn_5": 2.86756, "loss_dense_depth": 2.20723, "loss": 56.95585, "grad_norm": 126.32269, "time": 1.53313} -{"mode": "train", "epoch": 1, "iter": 6, "lr": 0.0001, "memory": 49167, "data_time": 0.13783, "loss_cls_0": 1.26396, "loss_box_0": 2.40301, "loss_cns_0": 0.69073, "loss_yns_0": 0.19385, "loss_cls_1": 1.34019, "loss_box_1": 3.66351, "loss_cns_1": 0.51219, "loss_yns_1": 0.18197, "loss_cls_2": 1.34407, "loss_box_2": 3.60785, "loss_cns_2": 0.48197, "loss_yns_2": 0.19183, "loss_cls_3": 1.31765, "loss_box_3": 3.49139, "loss_cns_3": 0.51534, "loss_yns_3": 0.18675, "loss_cls_4": 1.32133, "loss_box_4": 3.78863, "loss_cns_4": 0.50415, "loss_yns_4": 0.18573, "loss_cls_5": 1.37013, "loss_box_5": 4.07942, "loss_cns_5": 0.51611, "loss_yns_5": 0.20379, "loss_cls_dn_0": 0.559, "loss_box_dn_0": 1.13714, "loss_cls_dn_1": 0.52544, "loss_box_dn_1": 2.47198, "loss_cls_dn_2": 0.55701, "loss_box_dn_2": 2.37671, "loss_cls_dn_3": 0.47879, "loss_box_dn_3": 2.40537, "loss_cls_dn_4": 0.47224, "loss_box_dn_4": 2.61616, "loss_cls_dn_5": 0.43546, "loss_box_dn_5": 2.73854, "loss_dense_depth": 1.97857, "loss": 52.10796, "grad_norm": 101.37029, "time": 1.56085} -{"mode": "train", "epoch": 1, "iter": 7, "lr": 0.0001, "memory": 49167, "data_time": 0.07825, "loss_cls_0": 1.28535, "loss_box_0": 2.30493, "loss_cns_0": 0.64739, "loss_yns_0": 0.17952, "loss_cls_1": 1.2752, "loss_box_1": 3.57152, "loss_cns_1": 0.51341, "loss_yns_1": 0.1836, "loss_cls_2": 1.41981, "loss_box_2": 3.63258, "loss_cns_2": 0.44188, "loss_yns_2": 0.19189, "loss_cls_3": 1.2793, "loss_box_3": 3.5811, "loss_cns_3": 0.47869, "loss_yns_3": 0.18503, "loss_cls_4": 1.32111, "loss_box_4": 3.58432, "loss_cns_4": 0.48458, "loss_yns_4": 0.18938, "loss_cls_5": 1.36391, "loss_box_5": 3.75304, "loss_cns_5": 0.54031, "loss_yns_5": 0.20392, "loss_cls_dn_0": 0.51403, "loss_box_dn_0": 1.05234, "loss_cls_dn_1": 0.4639, "loss_box_dn_1": 2.30151, "loss_cls_dn_2": 0.46929, "loss_box_dn_2": 2.24657, "loss_cls_dn_3": 0.43751, "loss_box_dn_3": 2.25568, "loss_cls_dn_4": 0.42024, "loss_box_dn_4": 2.27872, "loss_cls_dn_5": 0.39078, "loss_box_dn_5": 2.36675, "loss_dense_depth": 2.13741, "loss": 49.94651, "grad_norm": 87.02299, "time": 1.51848} -{"mode": "train", "epoch": 1, "iter": 8, "lr": 0.0001, "memory": 49167, "data_time": 0.07876, "loss_cls_0": 1.2437, "loss_box_0": 2.28785, "loss_cns_0": 0.61411, "loss_yns_0": 0.18368, "loss_cls_1": 1.27694, "loss_box_1": 3.27147, "loss_cns_1": 0.53133, "loss_yns_1": 0.18621, "loss_cls_2": 1.29373, "loss_box_2": 3.3042, "loss_cns_2": 0.4794, "loss_yns_2": 0.17951, "loss_cls_3": 1.26003, "loss_box_3": 3.34035, "loss_cns_3": 0.52174, "loss_yns_3": 0.20854, "loss_cls_4": 1.25196, "loss_box_4": 3.30045, "loss_cns_4": 0.51225, "loss_yns_4": 0.18518, "loss_cls_5": 1.29241, "loss_box_5": 3.37944, "loss_cns_5": 0.54761, "loss_yns_5": 0.18144, "loss_cls_dn_0": 0.49256, "loss_box_dn_0": 1.06648, "loss_cls_dn_1": 0.45937, "loss_box_dn_1": 1.50478, "loss_cls_dn_2": 0.50177, "loss_box_dn_2": 1.55848, "loss_cls_dn_3": 0.43691, "loss_box_dn_3": 1.69751, "loss_cls_dn_4": 0.45542, "loss_box_dn_4": 1.69009, "loss_cls_dn_5": 0.44219, "loss_box_dn_5": 1.82817, "loss_dense_depth": 2.10135, "loss": 45.06858, "grad_norm": 69.03079, "time": 1.52506} -{"mode": "train", "epoch": 1, "iter": 9, "lr": 0.0001, "memory": 49167, "data_time": 0.08522, "loss_cls_0": 1.23804, "loss_box_0": 2.29662, "loss_cns_0": 0.63031, "loss_yns_0": 0.1744, "loss_cls_1": 1.28552, "loss_box_1": 3.08928, "loss_cns_1": 0.46184, "loss_yns_1": 0.17453, "loss_cls_2": 1.28101, "loss_box_2": 3.0859, "loss_cns_2": 0.47286, "loss_yns_2": 0.19305, "loss_cls_3": 1.26751, "loss_box_3": 3.19818, "loss_cns_3": 0.49757, "loss_yns_3": 0.18717, "loss_cls_4": 1.24802, "loss_box_4": 3.20002, "loss_cns_4": 0.49602, "loss_yns_4": 0.18268, "loss_cls_5": 1.25297, "loss_box_5": 3.45255, "loss_cns_5": 0.47179, "loss_yns_5": 0.1902, "loss_cls_dn_0": 0.48978, "loss_box_dn_0": 1.06457, "loss_cls_dn_1": 0.42195, "loss_box_dn_1": 1.68441, "loss_cls_dn_2": 0.48089, "loss_box_dn_2": 1.77377, "loss_cls_dn_3": 0.42396, "loss_box_dn_3": 1.89313, "loss_cls_dn_4": 0.42861, "loss_box_dn_4": 1.90401, "loss_cls_dn_5": 0.45953, "loss_box_dn_5": 2.14298, "loss_dense_depth": 1.89637, "loss": 45.09202, "grad_norm": 80.02423, "time": 1.5377} -{"mode": "train", "epoch": 1, "iter": 10, "lr": 0.0001, "memory": 49167, "data_time": 0.07972, "loss_cls_0": 1.26868, "loss_box_0": 2.34726, "loss_cns_0": 0.62817, "loss_yns_0": 0.17288, "loss_cls_1": 1.27663, "loss_box_1": 3.06102, "loss_cns_1": 0.48573, "loss_yns_1": 0.17802, "loss_cls_2": 1.26, "loss_box_2": 2.99337, "loss_cns_2": 0.48135, "loss_yns_2": 0.19486, "loss_cls_3": 1.22752, "loss_box_3": 3.05243, "loss_cns_3": 0.50141, "loss_yns_3": 0.1919, "loss_cls_4": 1.24737, "loss_box_4": 3.27168, "loss_cns_4": 0.49342, "loss_yns_4": 0.19349, "loss_cls_5": 1.2728, "loss_box_5": 3.54037, "loss_cns_5": 0.50241, "loss_yns_5": 0.18054, "loss_cls_dn_0": 0.45306, "loss_box_dn_0": 1.05704, "loss_cls_dn_1": 0.40013, "loss_box_dn_1": 1.90255, "loss_cls_dn_2": 0.4509, "loss_box_dn_2": 1.84954, "loss_cls_dn_3": 0.43733, "loss_box_dn_3": 1.8697, "loss_cls_dn_4": 0.40791, "loss_box_dn_4": 1.96023, "loss_cls_dn_5": 0.44032, "loss_box_dn_5": 2.10123, "loss_dense_depth": 2.13372, "loss": 45.48695, "grad_norm": 98.19573, "time": 1.54186} -{"mode": "train", "epoch": 1, "iter": 11, "lr": 0.0001, "memory": 49167, "data_time": 0.07956, "loss_cls_0": 1.25378, "loss_box_0": 2.33747, "loss_cns_0": 0.59922, "loss_yns_0": 0.1732, "loss_cls_1": 1.22079, "loss_box_1": 2.99779, "loss_cns_1": 0.53933, "loss_yns_1": 0.17438, "loss_cls_2": 1.22421, "loss_box_2": 3.07127, "loss_cns_2": 0.53269, "loss_yns_2": 0.18233, "loss_cls_3": 1.245, "loss_box_3": 3.0779, "loss_cns_3": 0.55002, "loss_yns_3": 0.17858, "loss_cls_4": 1.28004, "loss_box_4": 3.03701, "loss_cns_4": 0.51114, "loss_yns_4": 0.17331, "loss_cls_5": 1.26934, "loss_box_5": 3.25831, "loss_cns_5": 0.51179, "loss_yns_5": 0.17694, "loss_cls_dn_0": 0.43351, "loss_box_dn_0": 1.05274, "loss_cls_dn_1": 0.39498, "loss_box_dn_1": 1.99041, "loss_cls_dn_2": 0.42924, "loss_box_dn_2": 1.96337, "loss_cls_dn_3": 0.42386, "loss_box_dn_3": 1.99665, "loss_cls_dn_4": 0.39556, "loss_box_dn_4": 2.04464, "loss_cls_dn_5": 0.41824, "loss_box_dn_5": 2.12528, "loss_dense_depth": 1.88127, "loss": 45.12558, "grad_norm": 56.49272, "time": 1.50661} -{"mode": "train", "epoch": 1, "iter": 12, "lr": 0.0001, "memory": 49167, "data_time": 0.07744, "loss_cls_0": 1.17193, "loss_box_0": 2.38395, "loss_cns_0": 0.56221, "loss_yns_0": 0.17169, "loss_cls_1": 1.2069, "loss_box_1": 3.02966, "loss_cns_1": 0.5202, "loss_yns_1": 0.17529, "loss_cls_2": 1.23552, "loss_box_2": 3.20763, "loss_cns_2": 0.53147, "loss_yns_2": 0.18766, "loss_cls_3": 1.24187, "loss_box_3": 3.3504, "loss_cns_3": 0.5383, "loss_yns_3": 0.17639, "loss_cls_4": 1.22758, "loss_box_4": 3.29543, "loss_cns_4": 0.51331, "loss_yns_4": 0.19606, "loss_cls_5": 1.27519, "loss_box_5": 3.39449, "loss_cns_5": 0.52539, "loss_yns_5": 0.18841, "loss_cls_dn_0": 0.48344, "loss_box_dn_0": 1.06776, "loss_cls_dn_1": 0.39388, "loss_box_dn_1": 2.17692, "loss_cls_dn_2": 0.39027, "loss_box_dn_2": 2.21518, "loss_cls_dn_3": 0.37369, "loss_box_dn_3": 2.26949, "loss_cls_dn_4": 0.37989, "loss_box_dn_4": 2.28897, "loss_cls_dn_5": 0.39706, "loss_box_dn_5": 2.3224, "loss_dense_depth": 1.9536, "loss": 47.01948, "grad_norm": 105.3454, "time": 1.52972} -{"mode": "train", "epoch": 1, "iter": 13, "lr": 0.0001, "memory": 49167, "data_time": 0.08857, "loss_cls_0": 1.15231, "loss_box_0": 2.48695, "loss_cns_0": 0.54831, "loss_yns_0": 0.17029, "loss_cls_1": 1.21942, "loss_box_1": 3.01109, "loss_cns_1": 0.52631, "loss_yns_1": 0.1769, "loss_cls_2": 1.30874, "loss_box_2": 3.1313, "loss_cns_2": 0.53818, "loss_yns_2": 0.20094, "loss_cls_3": 1.29743, "loss_box_3": 3.23863, "loss_cns_3": 0.53385, "loss_yns_3": 0.19243, "loss_cls_4": 1.22048, "loss_box_4": 3.31063, "loss_cns_4": 0.5312, "loss_yns_4": 0.17881, "loss_cls_5": 1.24614, "loss_box_5": 3.44994, "loss_cns_5": 0.53826, "loss_yns_5": 0.18177, "loss_cls_dn_0": 0.50048, "loss_box_dn_0": 1.09701, "loss_cls_dn_1": 0.44476, "loss_box_dn_1": 1.64453, "loss_cls_dn_2": 0.4064, "loss_box_dn_2": 1.71123, "loss_cls_dn_3": 0.39961, "loss_box_dn_3": 1.77331, "loss_cls_dn_4": 0.43875, "loss_box_dn_4": 1.84922, "loss_cls_dn_5": 0.45106, "loss_box_dn_5": 1.91464, "loss_dense_depth": 1.94072, "loss": 44.96202, "grad_norm": 94.90555, "time": 1.52496} -{"mode": "train", "epoch": 1, "iter": 14, "lr": 0.00011, "memory": 49167, "data_time": 0.08039, "loss_cls_0": 1.16657, "loss_box_0": 2.59098, "loss_cns_0": 0.55404, "loss_yns_0": 0.17268, "loss_cls_1": 1.24757, "loss_box_1": 2.91694, "loss_cns_1": 0.56953, "loss_yns_1": 0.17557, "loss_cls_2": 1.31442, "loss_box_2": 2.91552, "loss_cns_2": 0.57582, "loss_yns_2": 0.18509, "loss_cls_3": 1.26547, "loss_box_3": 2.90727, "loss_cns_3": 0.55768, "loss_yns_3": 0.17973, "loss_cls_4": 1.2532, "loss_box_4": 2.88128, "loss_cns_4": 0.55352, "loss_yns_4": 0.17914, "loss_cls_5": 1.24516, "loss_box_5": 3.1118, "loss_cns_5": 0.53096, "loss_yns_5": 0.18835, "loss_cls_dn_0": 0.4802, "loss_box_dn_0": 1.12715, "loss_cls_dn_1": 0.46633, "loss_box_dn_1": 1.37042, "loss_cls_dn_2": 0.43377, "loss_box_dn_2": 1.34287, "loss_cls_dn_3": 0.43308, "loss_box_dn_3": 1.36231, "loss_cls_dn_4": 0.46584, "loss_box_dn_4": 1.41189, "loss_cls_dn_5": 0.46889, "loss_box_dn_5": 1.52052, "loss_dense_depth": 1.84069, "loss": 41.96223, "grad_norm": 56.15, "time": 1.51342} -{"mode": "train", "epoch": 1, "iter": 15, "lr": 0.00011, "memory": 49167, "data_time": 0.07901, "loss_cls_0": 1.24109, "loss_box_0": 2.61963, "loss_cns_0": 0.55803, "loss_yns_0": 0.1727, "loss_cls_1": 1.24637, "loss_box_1": 2.72124, "loss_cns_1": 0.59684, "loss_yns_1": 0.17247, "loss_cls_2": 1.28436, "loss_box_2": 2.79189, "loss_cns_2": 0.58355, "loss_yns_2": 0.18165, "loss_cls_3": 1.2538, "loss_box_3": 2.84828, "loss_cns_3": 0.59135, "loss_yns_3": 0.1828, "loss_cls_4": 1.28627, "loss_box_4": 2.82916, "loss_cns_4": 0.60314, "loss_yns_4": 0.17393, "loss_cls_5": 1.26889, "loss_box_5": 2.87143, "loss_cns_5": 0.58201, "loss_yns_5": 0.18889, "loss_cls_dn_0": 0.43381, "loss_box_dn_0": 1.14332, "loss_cls_dn_1": 0.45133, "loss_box_dn_1": 1.26836, "loss_cls_dn_2": 0.44111, "loss_box_dn_2": 1.3144, "loss_cls_dn_3": 0.43991, "loss_box_dn_3": 1.3943, "loss_cls_dn_4": 0.44549, "loss_box_dn_4": 1.40185, "loss_cls_dn_5": 0.44403, "loss_box_dn_5": 1.52235, "loss_dense_depth": 1.74734, "loss": 41.29739, "grad_norm": 115.73591, "time": 1.52269} -{"mode": "train", "epoch": 1, "iter": 16, "lr": 0.00011, "memory": 49167, "data_time": 0.07779, "loss_cls_0": 1.19465, "loss_box_0": 2.51218, "loss_cns_0": 0.55406, "loss_yns_0": 0.17528, "loss_cls_1": 1.23039, "loss_box_1": 2.71198, "loss_cns_1": 0.57864, "loss_yns_1": 0.17667, "loss_cls_2": 1.2728, "loss_box_2": 2.80292, "loss_cns_2": 0.57369, "loss_yns_2": 0.20627, "loss_cls_3": 1.25812, "loss_box_3": 2.82933, "loss_cns_3": 0.57449, "loss_yns_3": 0.1829, "loss_cls_4": 1.28397, "loss_box_4": 2.80085, "loss_cns_4": 0.57013, "loss_yns_4": 0.17535, "loss_cls_5": 1.27049, "loss_box_5": 2.99112, "loss_cns_5": 0.56199, "loss_yns_5": 0.17372, "loss_cls_dn_0": 0.45966, "loss_box_dn_0": 1.08663, "loss_cls_dn_1": 0.45861, "loss_box_dn_1": 1.32085, "loss_cls_dn_2": 0.45319, "loss_box_dn_2": 1.44063, "loss_cls_dn_3": 0.45797, "loss_box_dn_3": 1.54998, "loss_cls_dn_4": 0.4533, "loss_box_dn_4": 1.56115, "loss_cls_dn_5": 0.45432, "loss_box_dn_5": 1.77469, "loss_dense_depth": 1.83484, "loss": 41.96781, "grad_norm": 79.69514, "time": 1.53205} -{"mode": "train", "epoch": 1, "iter": 17, "lr": 0.00011, "memory": 49167, "data_time": 0.07787, "loss_cls_0": 1.1478, "loss_box_0": 2.41497, "loss_cns_0": 0.57314, "loss_yns_0": 0.17292, "loss_cls_1": 1.21609, "loss_box_1": 2.76627, "loss_cns_1": 0.5486, "loss_yns_1": 0.17544, "loss_cls_2": 1.24753, "loss_box_2": 2.80664, "loss_cns_2": 0.57076, "loss_yns_2": 0.18437, "loss_cls_3": 1.23138, "loss_box_3": 2.86954, "loss_cns_3": 0.56593, "loss_yns_3": 0.17667, "loss_cls_4": 1.26003, "loss_box_4": 2.80951, "loss_cns_4": 0.56007, "loss_yns_4": 0.17437, "loss_cls_5": 1.25, "loss_box_5": 2.93266, "loss_cns_5": 0.54169, "loss_yns_5": 0.18098, "loss_cls_dn_0": 0.46807, "loss_box_dn_0": 1.05342, "loss_cls_dn_1": 0.45247, "loss_box_dn_1": 1.49992, "loss_cls_dn_2": 0.44139, "loss_box_dn_2": 1.61297, "loss_cls_dn_3": 0.43599, "loss_box_dn_3": 1.73062, "loss_cls_dn_4": 0.43279, "loss_box_dn_4": 1.71741, "loss_cls_dn_5": 0.42603, "loss_box_dn_5": 1.87968, "loss_dense_depth": 1.57399, "loss": 42.10211, "grad_norm": 74.77982, "time": 1.53155} -{"mode": "train", "epoch": 1, "iter": 18, "lr": 0.00011, "memory": 49167, "data_time": 0.07262, "loss_cls_0": 1.14213, "loss_box_0": 2.29015, "loss_cns_0": 0.60071, "loss_yns_0": 0.17211, "loss_cls_1": 1.22952, "loss_box_1": 2.74273, "loss_cns_1": 0.54695, "loss_yns_1": 0.17602, "loss_cls_2": 1.24907, "loss_box_2": 2.78149, "loss_cns_2": 0.5669, "loss_yns_2": 0.17629, "loss_cls_3": 1.27419, "loss_box_3": 2.74406, "loss_cns_3": 0.57637, "loss_yns_3": 0.17472, "loss_cls_4": 1.26188, "loss_box_4": 2.69069, "loss_cns_4": 0.55721, "loss_yns_4": 0.17496, "loss_cls_5": 1.27606, "loss_box_5": 2.69211, "loss_cns_5": 0.5647, "loss_yns_5": 0.18164, "loss_cls_dn_0": 0.47415, "loss_box_dn_0": 1.02699, "loss_cls_dn_1": 0.43036, "loss_box_dn_1": 1.49644, "loss_cls_dn_2": 0.43622, "loss_box_dn_2": 1.58276, "loss_cls_dn_3": 0.40753, "loss_box_dn_3": 1.66255, "loss_cls_dn_4": 0.41348, "loss_box_dn_4": 1.64882, "loss_cls_dn_5": 0.4089, "loss_box_dn_5": 1.72132, "loss_dense_depth": 1.52057, "loss": 41.07278, "grad_norm": 66.50645, "time": 1.51491} -{"mode": "train", "epoch": 1, "iter": 19, "lr": 0.00011, "memory": 49167, "data_time": 0.07985, "loss_cls_0": 1.14712, "loss_box_0": 2.25752, "loss_cns_0": 0.59964, "loss_yns_0": 0.16961, "loss_cls_1": 1.29585, "loss_box_1": 2.74304, "loss_cns_1": 0.54105, "loss_yns_1": 0.17581, "loss_cls_2": 1.27281, "loss_box_2": 2.63005, "loss_cns_2": 0.56254, "loss_yns_2": 0.17384, "loss_cls_3": 1.31898, "loss_box_3": 2.61108, "loss_cns_3": 0.58465, "loss_yns_3": 0.17456, "loss_cls_4": 1.2694, "loss_box_4": 2.81221, "loss_cns_4": 0.57986, "loss_yns_4": 0.17156, "loss_cls_5": 1.2863, "loss_box_5": 2.70161, "loss_cns_5": 0.58987, "loss_yns_5": 0.17501, "loss_cls_dn_0": 0.47352, "loss_box_dn_0": 1.03438, "loss_cls_dn_1": 0.4121, "loss_box_dn_1": 1.24925, "loss_cls_dn_2": 0.44416, "loss_box_dn_2": 1.25032, "loss_cls_dn_3": 0.42463, "loss_box_dn_3": 1.31016, "loss_cls_dn_4": 0.43513, "loss_box_dn_4": 1.44628, "loss_cls_dn_5": 0.43391, "loss_box_dn_5": 1.42424, "loss_dense_depth": 1.44263, "loss": 39.62473, "grad_norm": 78.15804, "time": 1.50989} -{"mode": "train", "epoch": 1, "iter": 20, "lr": 0.00011, "memory": 49167, "data_time": 0.08154, "loss_cls_0": 1.15283, "loss_box_0": 2.24418, "loss_cns_0": 0.60593, "loss_yns_0": 0.16925, "loss_cls_1": 1.2822, "loss_box_1": 2.87042, "loss_cns_1": 0.57183, "loss_yns_1": 0.17836, "loss_cls_2": 1.25224, "loss_box_2": 2.68472, "loss_cns_2": 0.60164, "loss_yns_2": 0.17585, "loss_cls_3": 1.25018, "loss_box_3": 2.78906, "loss_cns_3": 0.58282, "loss_yns_3": 0.17622, "loss_cls_4": 1.24805, "loss_box_4": 3.04889, "loss_cns_4": 0.58083, "loss_yns_4": 0.17334, "loss_cls_5": 1.25689, "loss_box_5": 3.22708, "loss_cns_5": 0.54695, "loss_yns_5": 0.18023, "loss_cls_dn_0": 0.45846, "loss_box_dn_0": 1.02527, "loss_cls_dn_1": 0.38385, "loss_box_dn_1": 1.30635, "loss_cls_dn_2": 0.42075, "loss_box_dn_2": 1.25837, "loss_cls_dn_3": 0.43082, "loss_box_dn_3": 1.30888, "loss_cls_dn_4": 0.42781, "loss_box_dn_4": 1.44024, "loss_cls_dn_5": 0.42607, "loss_box_dn_5": 1.4686, "loss_dense_depth": 1.42103, "loss": 40.62649, "grad_norm": 73.02473, "time": 1.51049} -{"mode": "train", "epoch": 1, "iter": 21, "lr": 0.00011, "memory": 49167, "data_time": 0.1116, "loss_cls_0": 1.14104, "loss_box_0": 2.20449, "loss_cns_0": 0.61667, "loss_yns_0": 0.16826, "loss_cls_1": 1.2318, "loss_box_1": 3.00898, "loss_cns_1": 0.55116, "loss_yns_1": 0.17622, "loss_cls_2": 1.25417, "loss_box_2": 2.93061, "loss_cns_2": 0.54391, "loss_yns_2": 0.17609, "loss_cls_3": 1.25715, "loss_box_3": 3.05728, "loss_cns_3": 0.51404, "loss_yns_3": 0.17846, "loss_cls_4": 1.24536, "loss_box_4": 3.27446, "loss_cns_4": 0.48821, "loss_yns_4": 0.17093, "loss_cls_5": 1.26126, "loss_box_5": 3.48647, "loss_cns_5": 0.46421, "loss_yns_5": 0.17939, "loss_cls_dn_0": 0.45265, "loss_box_dn_0": 1.04055, "loss_cls_dn_1": 0.42703, "loss_box_dn_1": 1.3878, "loss_cls_dn_2": 0.45842, "loss_box_dn_2": 1.41721, "loss_cls_dn_3": 0.48388, "loss_box_dn_3": 1.52148, "loss_cls_dn_4": 0.47067, "loss_box_dn_4": 1.7322, "loss_cls_dn_5": 0.46874, "loss_box_dn_5": 1.83547, "loss_dense_depth": 1.31411, "loss": 42.59084, "grad_norm": 99.51576, "time": 1.62997} -{"mode": "train", "epoch": 1, "iter": 22, "lr": 0.00011, "memory": 49167, "data_time": 0.07455, "loss_cls_0": 1.14932, "loss_box_0": 2.15239, "loss_cns_0": 0.62339, "loss_yns_0": 0.16855, "loss_cls_1": 1.23341, "loss_box_1": 3.01367, "loss_cns_1": 0.52427, "loss_yns_1": 0.16986, "loss_cls_2": 1.26645, "loss_box_2": 2.92286, "loss_cns_2": 0.53559, "loss_yns_2": 0.17598, "loss_cls_3": 1.26443, "loss_box_3": 2.89704, "loss_cns_3": 0.53278, "loss_yns_3": 0.17522, "loss_cls_4": 1.23844, "loss_box_4": 2.9448, "loss_cns_4": 0.52987, "loss_yns_4": 0.17007, "loss_cls_5": 1.26285, "loss_box_5": 2.96569, "loss_cns_5": 0.51533, "loss_yns_5": 0.1702, "loss_cls_dn_0": 0.45194, "loss_box_dn_0": 1.02092, "loss_cls_dn_1": 0.4509, "loss_box_dn_1": 1.60489, "loss_cls_dn_2": 0.47156, "loss_box_dn_2": 1.64036, "loss_cls_dn_3": 0.4948, "loss_box_dn_3": 1.69704, "loss_cls_dn_4": 0.46924, "loss_box_dn_4": 1.86634, "loss_cls_dn_5": 0.47984, "loss_box_dn_5": 1.89989, "loss_dense_depth": 1.28279, "loss": 42.43299, "grad_norm": 66.87009, "time": 1.5978} -{"mode": "train", "epoch": 1, "iter": 23, "lr": 0.00011, "memory": 49167, "data_time": 0.10821, "loss_cls_0": 1.13142, "loss_box_0": 2.08943, "loss_cns_0": 0.62363, "loss_yns_0": 0.17014, "loss_cls_1": 1.22055, "loss_box_1": 3.02672, "loss_cns_1": 0.50682, "loss_yns_1": 0.17279, "loss_cls_2": 1.23524, "loss_box_2": 3.01062, "loss_cns_2": 0.51868, "loss_yns_2": 0.17439, "loss_cls_3": 1.24395, "loss_box_3": 2.96091, "loss_cns_3": 0.51556, "loss_yns_3": 0.17351, "loss_cls_4": 1.24373, "loss_box_4": 2.98165, "loss_cns_4": 0.51599, "loss_yns_4": 0.17329, "loss_cls_5": 1.24574, "loss_box_5": 2.96878, "loss_cns_5": 0.51738, "loss_yns_5": 0.1701, "loss_cls_dn_0": 0.45326, "loss_box_dn_0": 0.98613, "loss_cls_dn_1": 0.4468, "loss_box_dn_1": 1.56503, "loss_cls_dn_2": 0.45558, "loss_box_dn_2": 1.61041, "loss_cls_dn_3": 0.46608, "loss_box_dn_3": 1.61941, "loss_cls_dn_4": 0.42863, "loss_box_dn_4": 1.7318, "loss_cls_dn_5": 0.45878, "loss_box_dn_5": 1.75675, "loss_dense_depth": 1.28611, "loss": 41.85578, "grad_norm": 68.01165, "time": 1.55736} -{"mode": "train", "epoch": 1, "iter": 24, "lr": 0.00011, "memory": 49167, "data_time": 0.07627, "loss_cls_0": 1.1525, "loss_box_0": 2.06996, "loss_cns_0": 0.61738, "loss_yns_0": 0.16809, "loss_cls_1": 1.22352, "loss_box_1": 2.9363, "loss_cns_1": 0.523, "loss_yns_1": 0.17025, "loss_cls_2": 1.24653, "loss_box_2": 2.99822, "loss_cns_2": 0.52949, "loss_yns_2": 0.17463, "loss_cls_3": 1.2586, "loss_box_3": 3.03467, "loss_cns_3": 0.49315, "loss_yns_3": 0.17212, "loss_cls_4": 1.31604, "loss_box_4": 3.09485, "loss_cns_4": 0.47968, "loss_yns_4": 0.17241, "loss_cls_5": 1.2632, "loss_box_5": 3.16513, "loss_cns_5": 0.45643, "loss_yns_5": 0.18028, "loss_cls_dn_0": 0.46233, "loss_box_dn_0": 1.00569, "loss_cls_dn_1": 0.41799, "loss_box_dn_1": 1.62798, "loss_cls_dn_2": 0.41696, "loss_box_dn_2": 1.6909, "loss_cls_dn_3": 0.42152, "loss_box_dn_3": 1.68593, "loss_cls_dn_4": 0.38944, "loss_box_dn_4": 1.75864, "loss_cls_dn_5": 0.41958, "loss_box_dn_5": 1.81464, "loss_dense_depth": 1.30947, "loss": 42.3175, "grad_norm": 74.85537, "time": 1.52592} -{"mode": "train", "epoch": 1, "iter": 25, "lr": 0.00011, "memory": 49167, "data_time": 0.07743, "loss_cls_0": 1.14749, "loss_box_0": 2.10016, "loss_cns_0": 0.61239, "loss_yns_0": 0.16747, "loss_cls_1": 1.22767, "loss_box_1": 2.86009, "loss_cns_1": 0.55325, "loss_yns_1": 0.16905, "loss_cls_2": 1.29279, "loss_box_2": 2.88947, "loss_cns_2": 0.55282, "loss_yns_2": 0.17504, "loss_cls_3": 1.28571, "loss_box_3": 2.98408, "loss_cns_3": 0.51911, "loss_yns_3": 0.1746, "loss_cls_4": 1.2836, "loss_box_4": 3.08322, "loss_cns_4": 0.50839, "loss_yns_4": 0.17755, "loss_cls_5": 1.27868, "loss_box_5": 3.18223, "loss_cns_5": 0.47589, "loss_yns_5": 0.1813, "loss_cls_dn_0": 0.46531, "loss_box_dn_0": 1.01302, "loss_cls_dn_1": 0.39205, "loss_box_dn_1": 1.53354, "loss_cls_dn_2": 0.39088, "loss_box_dn_2": 1.56141, "loss_cls_dn_3": 0.39436, "loss_box_dn_3": 1.5625, "loss_cls_dn_4": 0.3847, "loss_box_dn_4": 1.60893, "loss_cls_dn_5": 0.39653, "loss_box_dn_5": 1.6711, "loss_dense_depth": 1.26535, "loss": 41.52173, "grad_norm": 68.16371, "time": 1.52993} -{"mode": "train", "epoch": 1, "iter": 26, "lr": 0.00011, "memory": 49167, "data_time": 0.1079, "loss_cls_0": 1.1236, "loss_box_0": 2.1531, "loss_cns_0": 0.6049, "loss_yns_0": 0.17086, "loss_cls_1": 1.23643, "loss_box_1": 2.78188, "loss_cns_1": 0.57744, "loss_yns_1": 0.17226, "loss_cls_2": 1.27119, "loss_box_2": 2.81209, "loss_cns_2": 0.56282, "loss_yns_2": 0.17021, "loss_cls_3": 1.27962, "loss_box_3": 2.79121, "loss_cns_3": 0.5952, "loss_yns_3": 0.17766, "loss_cls_4": 1.23982, "loss_box_4": 2.88168, "loss_cns_4": 0.56879, "loss_yns_4": 0.1745, "loss_cls_5": 1.27581, "loss_box_5": 2.9532, "loss_cns_5": 0.589, "loss_yns_5": 0.18188, "loss_cls_dn_0": 0.46027, "loss_box_dn_0": 1.04627, "loss_cls_dn_1": 0.40793, "loss_box_dn_1": 1.17068, "loss_cls_dn_2": 0.41992, "loss_box_dn_2": 1.18302, "loss_cls_dn_3": 0.41806, "loss_box_dn_3": 1.20747, "loss_cls_dn_4": 0.42732, "loss_box_dn_4": 1.22782, "loss_cls_dn_5": 0.41979, "loss_box_dn_5": 1.309, "loss_dense_depth": 1.23728, "loss": 39.27996, "grad_norm": 64.99834, "time": 1.55461} -{"mode": "train", "epoch": 1, "iter": 27, "lr": 0.00011, "memory": 49167, "data_time": 0.0816, "loss_cls_0": 1.12526, "loss_box_0": 2.2554, "loss_cns_0": 0.59268, "loss_yns_0": 0.17168, "loss_cls_1": 1.2021, "loss_box_1": 2.91213, "loss_cns_1": 0.5341, "loss_yns_1": 0.16925, "loss_cls_2": 1.1993, "loss_box_2": 3.10976, "loss_cns_2": 0.51475, "loss_yns_2": 0.17434, "loss_cls_3": 1.23732, "loss_box_3": 2.98498, "loss_cns_3": 0.52818, "loss_yns_3": 0.17196, "loss_cls_4": 1.21276, "loss_box_4": 3.02465, "loss_cns_4": 0.50817, "loss_yns_4": 0.17889, "loss_cls_5": 1.23994, "loss_box_5": 2.98911, "loss_cns_5": 0.53768, "loss_yns_5": 0.1804, "loss_cls_dn_0": 0.45256, "loss_box_dn_0": 1.07429, "loss_cls_dn_1": 0.4134, "loss_box_dn_1": 1.28861, "loss_cls_dn_2": 0.4432, "loss_box_dn_2": 1.31114, "loss_cls_dn_3": 0.43182, "loss_box_dn_3": 1.34501, "loss_cls_dn_4": 0.45766, "loss_box_dn_4": 1.34128, "loss_cls_dn_5": 0.42859, "loss_box_dn_5": 1.39843, "loss_dense_depth": 1.22314, "loss": 40.36392, "grad_norm": 90.53548, "time": 1.5302} -{"mode": "train", "epoch": 1, "iter": 28, "lr": 0.00011, "memory": 49167, "data_time": 0.07714, "loss_cls_0": 1.12376, "loss_box_0": 2.30576, "loss_cns_0": 0.59787, "loss_yns_0": 0.17108, "loss_cls_1": 1.15907, "loss_box_1": 2.8568, "loss_cns_1": 0.53444, "loss_yns_1": 0.17022, "loss_cls_2": 1.19462, "loss_box_2": 3.02373, "loss_cns_2": 0.51187, "loss_yns_2": 0.1818, "loss_cls_3": 1.21601, "loss_box_3": 2.84586, "loss_cns_3": 0.53818, "loss_yns_3": 0.17444, "loss_cls_4": 1.20594, "loss_box_4": 2.82555, "loss_cns_4": 0.53069, "loss_yns_4": 0.17298, "loss_cls_5": 1.21901, "loss_box_5": 2.81678, "loss_cns_5": 0.55039, "loss_yns_5": 0.17656, "loss_cls_dn_0": 0.42352, "loss_box_dn_0": 1.09985, "loss_cls_dn_1": 0.42334, "loss_box_dn_1": 1.4224, "loss_cls_dn_2": 0.45995, "loss_box_dn_2": 1.46898, "loss_cls_dn_3": 0.43195, "loss_box_dn_3": 1.51345, "loss_cls_dn_4": 0.4661, "loss_box_dn_4": 1.51058, "loss_cls_dn_5": 0.42788, "loss_box_dn_5": 1.58161, "loss_dense_depth": 1.22369, "loss": 40.55673, "grad_norm": 90.61868, "time": 1.51726} -{"mode": "train", "epoch": 1, "iter": 29, "lr": 0.00011, "memory": 49167, "data_time": 0.08976, "loss_cls_0": 1.13474, "loss_box_0": 2.23033, "loss_cns_0": 0.61221, "loss_yns_0": 0.17028, "loss_cls_1": 1.15287, "loss_box_1": 2.79154, "loss_cns_1": 0.56823, "loss_yns_1": 0.17377, "loss_cls_2": 1.19281, "loss_box_2": 2.83458, "loss_cns_2": 0.56343, "loss_yns_2": 0.17146, "loss_cls_3": 1.2007, "loss_box_3": 2.65023, "loss_cns_3": 0.57335, "loss_yns_3": 0.17331, "loss_cls_4": 1.18324, "loss_box_4": 2.63906, "loss_cns_4": 0.57153, "loss_yns_4": 0.17477, "loss_cls_5": 1.20292, "loss_box_5": 2.66514, "loss_cns_5": 0.56318, "loss_yns_5": 0.17324, "loss_cls_dn_0": 0.40936, "loss_box_dn_0": 1.09067, "loss_cls_dn_1": 0.44204, "loss_box_dn_1": 1.41811, "loss_cls_dn_2": 0.46947, "loss_box_dn_2": 1.44164, "loss_cls_dn_3": 0.43808, "loss_box_dn_3": 1.46326, "loss_cls_dn_4": 0.46168, "loss_box_dn_4": 1.48539, "loss_cls_dn_5": 0.4391, "loss_box_dn_5": 1.57422, "loss_dense_depth": 1.19231, "loss": 39.69225, "grad_norm": 94.98769, "time": 1.52238} -{"mode": "train", "epoch": 1, "iter": 30, "lr": 0.00011, "memory": 49167, "data_time": 0.07806, "loss_cls_0": 1.11316, "loss_box_0": 2.07992, "loss_cns_0": 0.62279, "loss_yns_0": 0.17181, "loss_cls_1": 1.13219, "loss_box_1": 2.68305, "loss_cns_1": 0.5733, "loss_yns_1": 0.17584, "loss_cls_2": 1.16956, "loss_box_2": 2.64761, "loss_cns_2": 0.59836, "loss_yns_2": 0.17479, "loss_cls_3": 1.17926, "loss_box_3": 2.60001, "loss_cns_3": 0.56865, "loss_yns_3": 0.17075, "loss_cls_4": 1.1676, "loss_box_4": 2.616, "loss_cns_4": 0.56834, "loss_yns_4": 0.17563, "loss_cls_5": 1.18247, "loss_box_5": 2.7046, "loss_cns_5": 0.55092, "loss_yns_5": 0.17033, "loss_cls_dn_0": 0.41317, "loss_box_dn_0": 1.05724, "loss_cls_dn_1": 0.44041, "loss_box_dn_1": 1.53899, "loss_cls_dn_2": 0.45177, "loss_box_dn_2": 1.52644, "loss_cls_dn_3": 0.42138, "loss_box_dn_3": 1.55347, "loss_cls_dn_4": 0.42427, "loss_box_dn_4": 1.5956, "loss_cls_dn_5": 0.42704, "loss_box_dn_5": 1.69886, "loss_dense_depth": 1.14259, "loss": 39.48819, "grad_norm": 77.96847, "time": 1.54084} -{"mode": "train", "epoch": 1, "iter": 31, "lr": 0.00011, "memory": 49167, "data_time": 0.07808, "loss_cls_0": 1.06464, "loss_box_0": 2.06416, "loss_cns_0": 0.62183, "loss_yns_0": 0.17256, "loss_cls_1": 1.11902, "loss_box_1": 2.55297, "loss_cns_1": 0.56951, "loss_yns_1": 0.1693, "loss_cls_2": 1.15197, "loss_box_2": 2.50066, "loss_cns_2": 0.60158, "loss_yns_2": 0.17904, "loss_cls_3": 1.17086, "loss_box_3": 2.60196, "loss_cns_3": 0.56169, "loss_yns_3": 0.17372, "loss_cls_4": 1.21688, "loss_box_4": 2.56287, "loss_cns_4": 0.57991, "loss_yns_4": 0.17834, "loss_cls_5": 1.16294, "loss_box_5": 2.61041, "loss_cns_5": 0.5721, "loss_yns_5": 0.17332, "loss_cls_dn_0": 0.44265, "loss_box_dn_0": 1.02577, "loss_cls_dn_1": 0.42741, "loss_box_dn_1": 1.50146, "loss_cls_dn_2": 0.42324, "loss_box_dn_2": 1.47262, "loss_cls_dn_3": 0.40715, "loss_box_dn_3": 1.52484, "loss_cls_dn_4": 0.38959, "loss_box_dn_4": 1.53601, "loss_cls_dn_5": 0.4188, "loss_box_dn_5": 1.61158, "loss_dense_depth": 1.12522, "loss": 38.63859, "grad_norm": 65.3174, "time": 1.51102} -{"mode": "train", "epoch": 1, "iter": 32, "lr": 0.00011, "memory": 49167, "data_time": 0.07648, "loss_cls_0": 1.04004, "loss_box_0": 2.11815, "loss_cns_0": 0.6155, "loss_yns_0": 0.17012, "loss_cls_1": 1.09295, "loss_box_1": 2.58314, "loss_cns_1": 0.5673, "loss_yns_1": 0.16988, "loss_cls_2": 1.14547, "loss_box_2": 2.49074, "loss_cns_2": 0.59493, "loss_yns_2": 0.17602, "loss_cls_3": 1.16053, "loss_box_3": 2.55093, "loss_cns_3": 0.58368, "loss_yns_3": 0.17595, "loss_cls_4": 1.19259, "loss_box_4": 2.49752, "loss_cns_4": 0.59135, "loss_yns_4": 0.17214, "loss_cls_5": 1.14587, "loss_box_5": 2.48148, "loss_cns_5": 0.5962, "loss_yns_5": 0.17564, "loss_cls_dn_0": 0.45674, "loss_box_dn_0": 1.02838, "loss_cls_dn_1": 0.39986, "loss_box_dn_1": 1.34869, "loss_cls_dn_2": 0.38316, "loss_box_dn_2": 1.32147, "loss_cls_dn_3": 0.39011, "loss_box_dn_3": 1.33433, "loss_cls_dn_4": 0.37979, "loss_box_dn_4": 1.31362, "loss_cls_dn_5": 0.40706, "loss_box_dn_5": 1.33882, "loss_dense_depth": 1.05814, "loss": 37.24825, "grad_norm": 53.65015, "time": 1.52891} -{"mode": "train", "epoch": 1, "iter": 33, "lr": 0.00011, "memory": 49167, "data_time": 0.08618, "loss_cls_0": 1.04859, "loss_box_0": 2.10993, "loss_cns_0": 0.61643, "loss_yns_0": 0.16959, "loss_cls_1": 1.10798, "loss_box_1": 2.60708, "loss_cns_1": 0.56488, "loss_yns_1": 0.17293, "loss_cls_2": 1.20251, "loss_box_2": 2.54749, "loss_cns_2": 0.58147, "loss_yns_2": 0.1756, "loss_cls_3": 1.16562, "loss_box_3": 2.53792, "loss_cns_3": 0.59041, "loss_yns_3": 0.17433, "loss_cls_4": 1.1398, "loss_box_4": 2.5814, "loss_cns_4": 0.58033, "loss_yns_4": 0.17709, "loss_cls_5": 1.14887, "loss_box_5": 2.66554, "loss_cns_5": 0.57387, "loss_yns_5": 0.17446, "loss_cls_dn_0": 0.45258, "loss_box_dn_0": 1.02482, "loss_cls_dn_1": 0.36252, "loss_box_dn_1": 1.41747, "loss_cls_dn_2": 0.35314, "loss_box_dn_2": 1.36688, "loss_cls_dn_3": 0.36695, "loss_box_dn_3": 1.32582, "loss_cls_dn_4": 0.38347, "loss_box_dn_4": 1.32713, "loss_cls_dn_5": 0.38109, "loss_box_dn_5": 1.37017, "loss_dense_depth": 1.06968, "loss": 37.61585, "grad_norm": 56.64138, "time": 1.52834} -{"mode": "train", "epoch": 1, "iter": 34, "lr": 0.00011, "memory": 49167, "data_time": 0.07341, "loss_cls_0": 1.05607, "loss_box_0": 2.07834, "loss_cns_0": 0.61665, "loss_yns_0": 0.17046, "loss_cls_1": 1.12402, "loss_box_1": 2.62197, "loss_cns_1": 0.56039, "loss_yns_1": 0.17398, "loss_cls_2": 1.1915, "loss_box_2": 2.56505, "loss_cns_2": 0.58533, "loss_yns_2": 0.17967, "loss_cls_3": 1.15377, "loss_box_3": 2.53386, "loss_cns_3": 0.58873, "loss_yns_3": 0.17885, "loss_cls_4": 1.13217, "loss_box_4": 2.61455, "loss_cns_4": 0.58502, "loss_yns_4": 0.17604, "loss_cls_5": 1.1472, "loss_box_5": 2.69798, "loss_cns_5": 0.57292, "loss_yns_5": 0.17246, "loss_cls_dn_0": 0.42956, "loss_box_dn_0": 1.00973, "loss_cls_dn_1": 0.34268, "loss_box_dn_1": 1.48598, "loss_cls_dn_2": 0.34873, "loss_box_dn_2": 1.41716, "loss_cls_dn_3": 0.36463, "loss_box_dn_3": 1.38851, "loss_cls_dn_4": 0.3928, "loss_box_dn_4": 1.43482, "loss_cls_dn_5": 0.37525, "loss_box_dn_5": 1.49058, "loss_dense_depth": 1.06437, "loss": 38.02177, "grad_norm": 54.75942, "time": 1.50738} -{"mode": "train", "epoch": 1, "iter": 35, "lr": 0.00011, "memory": 49167, "data_time": 0.07995, "loss_cls_0": 1.08864, "loss_box_0": 2.02505, "loss_cns_0": 0.62235, "loss_yns_0": 0.17199, "loss_cls_1": 1.15212, "loss_box_1": 2.59135, "loss_cns_1": 0.55524, "loss_yns_1": 0.17053, "loss_cls_2": 1.14697, "loss_box_2": 2.50601, "loss_cns_2": 0.59056, "loss_yns_2": 0.17603, "loss_cls_3": 1.14835, "loss_box_3": 2.55142, "loss_cns_3": 0.58301, "loss_yns_3": 0.17212, "loss_cls_4": 1.1438, "loss_box_4": 2.60385, "loss_cns_4": 0.57996, "loss_yns_4": 0.17437, "loss_cls_5": 1.15716, "loss_box_5": 2.61205, "loss_cns_5": 0.58368, "loss_yns_5": 0.17756, "loss_cls_dn_0": 0.40363, "loss_box_dn_0": 0.9933, "loss_cls_dn_1": 0.33577, "loss_box_dn_1": 1.5539, "loss_cls_dn_2": 0.36351, "loss_box_dn_2": 1.51284, "loss_cls_dn_3": 0.37166, "loss_box_dn_3": 1.55996, "loss_cls_dn_4": 0.39444, "loss_box_dn_4": 1.64123, "loss_cls_dn_5": 0.37208, "loss_box_dn_5": 1.70359, "loss_dense_depth": 1.07756, "loss": 38.56762, "grad_norm": 67.99028, "time": 1.51562} -{"mode": "train", "epoch": 1, "iter": 36, "lr": 0.00011, "memory": 49167, "data_time": 0.07963, "loss_cls_0": 1.0924, "loss_box_0": 2.00851, "loss_cns_0": 0.62252, "loss_yns_0": 0.17157, "loss_cls_1": 1.12989, "loss_box_1": 2.72828, "loss_cns_1": 0.54788, "loss_yns_1": 0.17144, "loss_cls_2": 1.11687, "loss_box_2": 2.61532, "loss_cns_2": 0.58706, "loss_yns_2": 0.17284, "loss_cls_3": 1.13246, "loss_box_3": 2.64818, "loss_cns_3": 0.57473, "loss_yns_3": 0.17162, "loss_cls_4": 1.12886, "loss_box_4": 2.66753, "loss_cns_4": 0.57351, "loss_yns_4": 0.17849, "loss_cls_5": 1.15862, "loss_box_5": 2.68094, "loss_cns_5": 0.5801, "loss_yns_5": 0.17664, "loss_cls_dn_0": 0.39236, "loss_box_dn_0": 0.9756, "loss_cls_dn_1": 0.35506, "loss_box_dn_1": 1.40275, "loss_cls_dn_2": 0.40488, "loss_box_dn_2": 1.40529, "loss_cls_dn_3": 0.40321, "loss_box_dn_3": 1.48598, "loss_cls_dn_4": 0.4095, "loss_box_dn_4": 1.59673, "loss_cls_dn_5": 0.39171, "loss_box_dn_5": 1.68701, "loss_dense_depth": 1.03531, "loss": 38.58166, "grad_norm": 72.20571, "time": 1.53347} -{"mode": "train", "epoch": 1, "iter": 37, "lr": 0.00011, "memory": 49167, "data_time": 0.08879, "loss_cls_0": 1.06443, "loss_box_0": 2.01478, "loss_cns_0": 0.6233, "loss_yns_0": 0.17243, "loss_cls_1": 1.11802, "loss_box_1": 2.78297, "loss_cns_1": 0.56138, "loss_yns_1": 0.17316, "loss_cls_2": 1.12237, "loss_box_2": 2.69706, "loss_cns_2": 0.59726, "loss_yns_2": 0.17649, "loss_cls_3": 1.13384, "loss_box_3": 2.65473, "loss_cns_3": 0.59019, "loss_yns_3": 0.17542, "loss_cls_4": 1.12878, "loss_box_4": 2.67967, "loss_cns_4": 0.59099, "loss_yns_4": 0.1721, "loss_cls_5": 1.17001, "loss_box_5": 2.74865, "loss_cns_5": 0.60602, "loss_yns_5": 0.1696, "loss_cls_dn_0": 0.40553, "loss_box_dn_0": 0.97802, "loss_cls_dn_1": 0.34891, "loss_box_dn_1": 1.56682, "loss_cls_dn_2": 0.40835, "loss_box_dn_2": 1.54365, "loss_cls_dn_3": 0.38955, "loss_box_dn_3": 1.56452, "loss_cls_dn_4": 0.38626, "loss_box_dn_4": 1.64338, "loss_cls_dn_5": 0.37616, "loss_box_dn_5": 1.72708, "loss_dense_depth": 1.07525, "loss": 39.33714, "grad_norm": 63.54215, "time": 1.51859} -{"mode": "train", "epoch": 1, "iter": 38, "lr": 0.00011, "memory": 49167, "data_time": 0.08329, "loss_cls_0": 1.03391, "loss_box_0": 2.0152, "loss_cns_0": 0.62339, "loss_yns_0": 0.16967, "loss_cls_1": 1.08828, "loss_box_1": 2.65441, "loss_cns_1": 0.56853, "loss_yns_1": 0.17047, "loss_cls_2": 1.1109, "loss_box_2": 2.61587, "loss_cns_2": 0.59437, "loss_yns_2": 0.17407, "loss_cls_3": 1.12594, "loss_box_3": 2.62242, "loss_cns_3": 0.59015, "loss_yns_3": 0.17035, "loss_cls_4": 1.12612, "loss_box_4": 2.65225, "loss_cns_4": 0.58738, "loss_yns_4": 0.16686, "loss_cls_5": 1.14109, "loss_box_5": 2.74435, "loss_cns_5": 0.602, "loss_yns_5": 0.17332, "loss_cls_dn_0": 0.42196, "loss_box_dn_0": 0.98348, "loss_cls_dn_1": 0.35938, "loss_box_dn_1": 1.3724, "loss_cls_dn_2": 0.40721, "loss_box_dn_2": 1.31349, "loss_cls_dn_3": 0.38071, "loss_box_dn_3": 1.31133, "loss_cls_dn_4": 0.37216, "loss_box_dn_4": 1.35262, "loss_cls_dn_5": 0.37897, "loss_box_dn_5": 1.41303, "loss_dense_depth": 1.0614, "loss": 37.64946, "grad_norm": 69.63551, "time": 1.55693} -{"mode": "train", "epoch": 1, "iter": 39, "lr": 0.00012, "memory": 49167, "data_time": 0.07634, "loss_cls_0": 1.04803, "loss_box_0": 1.99495, "loss_cns_0": 0.62118, "loss_yns_0": 0.16685, "loss_cls_1": 1.13745, "loss_box_1": 2.47627, "loss_cns_1": 0.59194, "loss_yns_1": 0.17068, "loss_cls_2": 1.12455, "loss_box_2": 2.45657, "loss_cns_2": 0.6077, "loss_yns_2": 0.17173, "loss_cls_3": 1.1538, "loss_box_3": 2.4643, "loss_cns_3": 0.61312, "loss_yns_3": 0.17007, "loss_cls_4": 1.16863, "loss_box_4": 2.46419, "loss_cns_4": 0.60897, "loss_yns_4": 0.17245, "loss_cls_5": 1.14637, "loss_box_5": 2.53824, "loss_cns_5": 0.62331, "loss_yns_5": 0.17569, "loss_cls_dn_0": 0.43519, "loss_box_dn_0": 0.977, "loss_cls_dn_1": 0.36299, "loss_box_dn_1": 1.15342, "loss_cls_dn_2": 0.40003, "loss_box_dn_2": 1.08833, "loss_cls_dn_3": 0.37359, "loss_box_dn_3": 1.07426, "loss_cls_dn_4": 0.35983, "loss_box_dn_4": 1.08664, "loss_cls_dn_5": 0.38802, "loss_box_dn_5": 1.12216, "loss_dense_depth": 1.03015, "loss": 35.71863, "grad_norm": 60.87881, "time": 1.52147} -{"mode": "train", "epoch": 1, "iter": 40, "lr": 0.00012, "memory": 49167, "data_time": 0.0841, "loss_cls_0": 1.05027, "loss_box_0": 1.98911, "loss_cns_0": 0.62236, "loss_yns_0": 0.16598, "loss_cls_1": 1.13869, "loss_box_1": 2.32533, "loss_cns_1": 0.60901, "loss_yns_1": 0.17019, "loss_cls_2": 1.13593, "loss_box_2": 2.34088, "loss_cns_2": 0.61686, "loss_yns_2": 0.16866, "loss_cls_3": 1.16529, "loss_box_3": 2.26318, "loss_cns_3": 0.62867, "loss_yns_3": 0.17099, "loss_cls_4": 1.19097, "loss_box_4": 2.24433, "loss_cns_4": 0.63752, "loss_yns_4": 0.17126, "loss_cls_5": 1.17598, "loss_box_5": 2.26658, "loss_cns_5": 0.63861, "loss_yns_5": 0.17162, "loss_cls_dn_0": 0.42663, "loss_box_dn_0": 0.98024, "loss_cls_dn_1": 0.35827, "loss_box_dn_1": 1.09194, "loss_cls_dn_2": 0.38342, "loss_box_dn_2": 1.04513, "loss_cls_dn_3": 0.3652, "loss_box_dn_3": 1.0179, "loss_cls_dn_4": 0.35144, "loss_box_dn_4": 1.00878, "loss_cls_dn_5": 0.38996, "loss_box_dn_5": 1.02425, "loss_dense_depth": 1.06265, "loss": 34.56408, "grad_norm": 48.20831, "time": 1.52732} -{"mode": "train", "epoch": 1, "iter": 41, "lr": 0.00012, "memory": 49167, "data_time": 0.11442, "loss_cls_0": 1.02009, "loss_box_0": 1.95835, "loss_cns_0": 0.6216, "loss_yns_0": 0.16258, "loss_cls_1": 1.09478, "loss_box_1": 2.22037, "loss_cns_1": 0.61752, "loss_yns_1": 0.16772, "loss_cls_2": 1.12528, "loss_box_2": 2.26156, "loss_cns_2": 0.62173, "loss_yns_2": 0.1664, "loss_cls_3": 1.13505, "loss_box_3": 2.2048, "loss_cns_3": 0.62841, "loss_yns_3": 0.17107, "loss_cls_4": 1.16718, "loss_box_4": 2.23112, "loss_cns_4": 0.6384, "loss_yns_4": 0.16615, "loss_cls_5": 1.14452, "loss_box_5": 2.21728, "loss_cns_5": 0.63257, "loss_yns_5": 0.1676, "loss_cls_dn_0": 0.40796, "loss_box_dn_0": 0.96859, "loss_cls_dn_1": 0.33474, "loss_box_dn_1": 1.03405, "loss_cls_dn_2": 0.35071, "loss_box_dn_2": 1.00998, "loss_cls_dn_3": 0.35352, "loss_box_dn_3": 0.99754, "loss_cls_dn_4": 0.34257, "loss_box_dn_4": 1.01003, "loss_cls_dn_5": 0.36885, "loss_box_dn_5": 1.02682, "loss_dense_depth": 1.02054, "loss": 33.76803, "grad_norm": 59.44854, "time": 1.59587} -{"mode": "train", "epoch": 1, "iter": 42, "lr": 0.00012, "memory": 49167, "data_time": 0.07451, "loss_cls_0": 1.0588, "loss_box_0": 1.93093, "loss_cns_0": 0.62183, "loss_yns_0": 0.16285, "loss_cls_1": 1.13391, "loss_box_1": 2.26291, "loss_cns_1": 0.61219, "loss_yns_1": 0.16806, "loss_cls_2": 1.17535, "loss_box_2": 2.2359, "loss_cns_2": 0.63071, "loss_yns_2": 0.16459, "loss_cls_3": 1.13252, "loss_box_3": 2.26894, "loss_cns_3": 0.62623, "loss_yns_3": 0.16788, "loss_cls_4": 1.14352, "loss_box_4": 2.30139, "loss_cns_4": 0.62611, "loss_yns_4": 0.16712, "loss_cls_5": 1.13688, "loss_box_5": 2.32828, "loss_cns_5": 0.6311, "loss_yns_5": 0.17099, "loss_cls_dn_0": 0.39402, "loss_box_dn_0": 0.9643, "loss_cls_dn_1": 0.34094, "loss_box_dn_1": 1.02879, "loss_cls_dn_2": 0.34668, "loss_box_dn_2": 1.01101, "loss_cls_dn_3": 0.37658, "loss_box_dn_3": 1.06096, "loss_cls_dn_4": 0.36834, "loss_box_dn_4": 1.08931, "loss_cls_dn_5": 0.3851, "loss_box_dn_5": 1.14843, "loss_dense_depth": 1.05767, "loss": 34.43111, "grad_norm": 67.54173, "time": 1.62724} -{"mode": "train", "epoch": 1, "iter": 43, "lr": 0.00012, "memory": 49167, "data_time": 0.10765, "loss_cls_0": 1.08373, "loss_box_0": 1.94529, "loss_cns_0": 0.61827, "loss_yns_0": 0.16515, "loss_cls_1": 1.18563, "loss_box_1": 2.39414, "loss_cns_1": 0.59837, "loss_yns_1": 0.16825, "loss_cls_2": 1.22068, "loss_box_2": 2.29156, "loss_cns_2": 0.6197, "loss_yns_2": 0.16691, "loss_cls_3": 1.16697, "loss_box_3": 2.33493, "loss_cns_3": 0.62051, "loss_yns_3": 0.16715, "loss_cls_4": 1.1661, "loss_box_4": 2.3262, "loss_cns_4": 0.61983, "loss_yns_4": 0.16821, "loss_cls_5": 1.19195, "loss_box_5": 2.40032, "loss_cns_5": 0.63811, "loss_yns_5": 0.1745, "loss_cls_dn_0": 0.40798, "loss_box_dn_0": 0.95457, "loss_cls_dn_1": 0.36518, "loss_box_dn_1": 1.0698, "loss_cls_dn_2": 0.36361, "loss_box_dn_2": 1.05611, "loss_cls_dn_3": 0.40656, "loss_box_dn_3": 1.11812, "loss_cls_dn_4": 0.40975, "loss_box_dn_4": 1.13054, "loss_cls_dn_5": 0.41125, "loss_box_dn_5": 1.21709, "loss_dense_depth": 1.0187, "loss": 35.36171, "grad_norm": 62.81604, "time": 1.55627} -{"mode": "train", "epoch": 1, "iter": 44, "lr": 0.00012, "memory": 49167, "data_time": 0.08193, "loss_cls_0": 1.04974, "loss_box_0": 1.99723, "loss_cns_0": 0.61378, "loss_yns_0": 0.16695, "loss_cls_1": 1.13546, "loss_box_1": 2.41966, "loss_cns_1": 0.59946, "loss_yns_1": 0.16813, "loss_cls_2": 1.13108, "loss_box_2": 2.33414, "loss_cns_2": 0.61147, "loss_yns_2": 0.16719, "loss_cls_3": 1.13852, "loss_box_3": 2.33378, "loss_cns_3": 0.61569, "loss_yns_3": 0.16662, "loss_cls_4": 1.14215, "loss_box_4": 2.29978, "loss_cns_4": 0.62373, "loss_yns_4": 0.16892, "loss_cls_5": 1.16263, "loss_box_5": 2.35799, "loss_cns_5": 0.63367, "loss_yns_5": 0.17375, "loss_cls_dn_0": 0.42821, "loss_box_dn_0": 0.95211, "loss_cls_dn_1": 0.37358, "loss_box_dn_1": 1.14528, "loss_cls_dn_2": 0.38101, "loss_box_dn_2": 1.12772, "loss_cls_dn_3": 0.41423, "loss_box_dn_3": 1.16836, "loss_cls_dn_4": 0.4287, "loss_box_dn_4": 1.17316, "loss_cls_dn_5": 0.41317, "loss_box_dn_5": 1.24392, "loss_dense_depth": 1.06337, "loss": 35.52434, "grad_norm": 58.81508, "time": 1.54197} -{"mode": "train", "epoch": 1, "iter": 45, "lr": 0.00012, "memory": 49167, "data_time": 0.08061, "loss_cls_0": 1.06073, "loss_box_0": 2.00658, "loss_cns_0": 0.6124, "loss_yns_0": 0.16764, "loss_cls_1": 1.11054, "loss_box_1": 2.3576, "loss_cns_1": 0.59749, "loss_yns_1": 0.17186, "loss_cls_2": 1.1225, "loss_box_2": 2.26369, "loss_cns_2": 0.61482, "loss_yns_2": 0.17088, "loss_cls_3": 1.14461, "loss_box_3": 2.25608, "loss_cns_3": 0.61819, "loss_yns_3": 0.16611, "loss_cls_4": 1.13887, "loss_box_4": 2.27077, "loss_cns_4": 0.62598, "loss_yns_4": 0.16662, "loss_cls_5": 1.14285, "loss_box_5": 2.28732, "loss_cns_5": 0.62719, "loss_yns_5": 0.16901, "loss_cls_dn_0": 0.4421, "loss_box_dn_0": 0.95579, "loss_cls_dn_1": 0.37897, "loss_box_dn_1": 1.14629, "loss_cls_dn_2": 0.39399, "loss_box_dn_2": 1.13146, "loss_cls_dn_3": 0.41176, "loss_box_dn_3": 1.16734, "loss_cls_dn_4": 0.43275, "loss_box_dn_4": 1.18166, "loss_cls_dn_5": 0.41136, "loss_box_dn_5": 1.23118, "loss_dense_depth": 1.06294, "loss": 35.21793, "grad_norm": 63.54389, "time": 1.52952} -{"mode": "train", "epoch": 1, "iter": 46, "lr": 0.00012, "memory": 49167, "data_time": 0.11056, "loss_cls_0": 1.05715, "loss_box_0": 1.99583, "loss_cns_0": 0.61098, "loss_yns_0": 0.16547, "loss_cls_1": 1.11608, "loss_box_1": 2.4176, "loss_cns_1": 0.59205, "loss_yns_1": 0.16713, "loss_cls_2": 1.1315, "loss_box_2": 2.30062, "loss_cns_2": 0.61775, "loss_yns_2": 0.16774, "loss_cls_3": 1.14111, "loss_box_3": 2.30393, "loss_cns_3": 0.62529, "loss_yns_3": 0.16277, "loss_cls_4": 1.13147, "loss_box_4": 2.31241, "loss_cns_4": 0.62733, "loss_yns_4": 0.16943, "loss_cls_5": 1.13811, "loss_box_5": 2.34921, "loss_cns_5": 0.62733, "loss_yns_5": 0.16717, "loss_cls_dn_0": 0.43932, "loss_box_dn_0": 0.94246, "loss_cls_dn_1": 0.36414, "loss_box_dn_1": 1.13271, "loss_cls_dn_2": 0.38936, "loss_box_dn_2": 1.08755, "loss_cls_dn_3": 0.38731, "loss_box_dn_3": 1.11373, "loss_cls_dn_4": 0.40931, "loss_box_dn_4": 1.11854, "loss_cls_dn_5": 0.39549, "loss_box_dn_5": 1.1633, "loss_dense_depth": 1.03326, "loss": 35.07196, "grad_norm": 59.1166, "time": 1.57317} -{"mode": "train", "epoch": 1, "iter": 47, "lr": 0.00012, "memory": 49167, "data_time": 0.07812, "loss_cls_0": 1.04471, "loss_box_0": 1.98629, "loss_cns_0": 0.61339, "loss_yns_0": 0.1675, "loss_cls_1": 1.17219, "loss_box_1": 2.39632, "loss_cns_1": 0.60103, "loss_yns_1": 0.16869, "loss_cls_2": 1.15218, "loss_box_2": 2.30305, "loss_cns_2": 0.62477, "loss_yns_2": 0.16616, "loss_cls_3": 1.14395, "loss_box_3": 2.28996, "loss_cns_3": 0.63303, "loss_yns_3": 0.16071, "loss_cls_4": 1.14389, "loss_box_4": 2.28664, "loss_cns_4": 0.63475, "loss_yns_4": 0.16886, "loss_cls_5": 1.15862, "loss_box_5": 2.3289, "loss_cns_5": 0.63293, "loss_yns_5": 0.16697, "loss_cls_dn_0": 0.43193, "loss_box_dn_0": 0.94111, "loss_cls_dn_1": 0.35379, "loss_box_dn_1": 1.07565, "loss_cls_dn_2": 0.38768, "loss_box_dn_2": 1.02532, "loss_cls_dn_3": 0.37146, "loss_box_dn_3": 1.02971, "loss_cls_dn_4": 0.38467, "loss_box_dn_4": 1.02408, "loss_cls_dn_5": 0.38956, "loss_box_dn_5": 1.05414, "loss_dense_depth": 1.0424, "loss": 34.65697, "grad_norm": 46.6776, "time": 1.52758} -{"mode": "train", "epoch": 1, "iter": 48, "lr": 0.00012, "memory": 49167, "data_time": 0.07825, "loss_cls_0": 1.04573, "loss_box_0": 1.97233, "loss_cns_0": 0.60945, "loss_yns_0": 0.16754, "loss_cls_1": 1.12549, "loss_box_1": 2.43472, "loss_cns_1": 0.5819, "loss_yns_1": 0.16811, "loss_cls_2": 1.12646, "loss_box_2": 2.33417, "loss_cns_2": 0.61959, "loss_yns_2": 0.16376, "loss_cls_3": 1.14934, "loss_box_3": 2.31224, "loss_cns_3": 0.62905, "loss_yns_3": 0.16191, "loss_cls_4": 1.15172, "loss_box_4": 2.32112, "loss_cns_4": 0.63196, "loss_yns_4": 0.16752, "loss_cls_5": 1.14821, "loss_box_5": 2.36014, "loss_cns_5": 0.63092, "loss_yns_5": 0.16611, "loss_cls_dn_0": 0.40323, "loss_box_dn_0": 0.92968, "loss_cls_dn_1": 0.33279, "loss_box_dn_1": 1.12065, "loss_cls_dn_2": 0.38394, "loss_box_dn_2": 1.04021, "loss_cls_dn_3": 0.35241, "loss_box_dn_3": 1.02618, "loss_cls_dn_4": 0.35218, "loss_box_dn_4": 1.02142, "loss_cls_dn_5": 0.373, "loss_box_dn_5": 1.04188, "loss_dense_depth": 1.17672, "loss": 34.7338, "grad_norm": 49.94709, "time": 1.53464} -{"mode": "train", "epoch": 1, "iter": 49, "lr": 0.00012, "memory": 49167, "data_time": 0.09107, "loss_cls_0": 1.09915, "loss_box_0": 1.97728, "loss_cns_0": 0.6102, "loss_yns_0": 0.17118, "loss_cls_1": 1.11237, "loss_box_1": 2.45797, "loss_cns_1": 0.58995, "loss_yns_1": 0.16515, "loss_cls_2": 1.11645, "loss_box_2": 2.37556, "loss_cns_2": 0.6208, "loss_yns_2": 0.16794, "loss_cls_3": 1.1769, "loss_box_3": 2.39627, "loss_cns_3": 0.61998, "loss_yns_3": 0.1631, "loss_cls_4": 1.18435, "loss_box_4": 2.38016, "loss_cns_4": 0.62452, "loss_yns_4": 0.16792, "loss_cls_5": 1.14882, "loss_box_5": 2.37001, "loss_cns_5": 0.62665, "loss_yns_5": 0.1638, "loss_cls_dn_0": 0.38633, "loss_box_dn_0": 0.9254, "loss_cls_dn_1": 0.32312, "loss_box_dn_1": 1.12147, "loss_cls_dn_2": 0.37306, "loss_box_dn_2": 1.05565, "loss_cls_dn_3": 0.3417, "loss_box_dn_3": 1.05874, "loss_cls_dn_4": 0.33551, "loss_box_dn_4": 1.05837, "loss_cls_dn_5": 0.36167, "loss_box_dn_5": 1.07193, "loss_dense_depth": 1.17079, "loss": 35.07022, "grad_norm": 59.77552, "time": 1.56603} -{"mode": "train", "epoch": 1, "iter": 50, "lr": 0.00012, "memory": 49167, "data_time": 0.07469, "loss_cls_0": 1.06124, "loss_box_0": 2.00401, "loss_cns_0": 0.6121, "loss_yns_0": 0.16704, "loss_cls_1": 1.10776, "loss_box_1": 2.53224, "loss_cns_1": 0.5932, "loss_yns_1": 0.16403, "loss_cls_2": 1.11316, "loss_box_2": 2.45105, "loss_cns_2": 0.61773, "loss_yns_2": 0.16431, "loss_cls_3": 1.15346, "loss_box_3": 2.48134, "loss_cns_3": 0.61367, "loss_yns_3": 0.16239, "loss_cls_4": 1.15392, "loss_box_4": 2.4461, "loss_cns_4": 0.61858, "loss_yns_4": 0.16408, "loss_cls_5": 1.15648, "loss_box_5": 2.4362, "loss_cns_5": 0.62035, "loss_yns_5": 0.16183, "loss_cls_dn_0": 0.381, "loss_box_dn_0": 0.93599, "loss_cls_dn_1": 0.33257, "loss_box_dn_1": 1.12658, "loss_cls_dn_2": 0.37085, "loss_box_dn_2": 1.07362, "loss_cls_dn_3": 0.35117, "loss_box_dn_3": 1.09921, "loss_cls_dn_4": 0.34458, "loss_box_dn_4": 1.12046, "loss_cls_dn_5": 0.36488, "loss_box_dn_5": 1.15625, "loss_dense_depth": 1.11762, "loss": 35.53106, "grad_norm": 63.79477, "time": 1.54205} -{"mode": "train", "epoch": 1, "iter": 51, "lr": 0.00012, "memory": 49167, "data_time": 0.07612, "loss_cls_0": 1.03259, "loss_box_0": 2.02913, "loss_cns_0": 0.60812, "loss_yns_0": 0.165, "loss_cls_1": 1.07129, "loss_box_1": 2.49334, "loss_cns_1": 0.5962, "loss_yns_1": 0.16417, "loss_cls_2": 1.10812, "loss_box_2": 2.39536, "loss_cns_2": 0.61949, "loss_yns_2": 0.16309, "loss_cls_3": 1.1054, "loss_box_3": 2.42886, "loss_cns_3": 0.62177, "loss_yns_3": 0.16579, "loss_cls_4": 1.11014, "loss_box_4": 2.42658, "loss_cns_4": 0.62433, "loss_yns_4": 0.16494, "loss_cls_5": 1.15241, "loss_box_5": 2.4548, "loss_cns_5": 0.6258, "loss_yns_5": 0.16323, "loss_cls_dn_0": 0.39222, "loss_box_dn_0": 0.93701, "loss_cls_dn_1": 0.3201, "loss_box_dn_1": 1.18636, "loss_cls_dn_2": 0.34936, "loss_box_dn_2": 1.10376, "loss_cls_dn_3": 0.35369, "loss_box_dn_3": 1.12286, "loss_cls_dn_4": 0.35015, "loss_box_dn_4": 1.15149, "loss_cls_dn_5": 0.35129, "loss_box_dn_5": 1.19713, "loss_dense_depth": 0.9497, "loss": 35.25509, "grad_norm": 52.58107, "time": 1.51387} -{"mode": "train", "epoch": 1, "iter": 52, "lr": 0.00012, "memory": 49167, "data_time": 0.07814, "loss_cls_0": 1.0288, "loss_box_0": 2.0388, "loss_cns_0": 0.60149, "loss_yns_0": 0.16589, "loss_cls_1": 1.06525, "loss_box_1": 2.43562, "loss_cns_1": 0.59517, "loss_yns_1": 0.16583, "loss_cls_2": 1.11881, "loss_box_2": 2.38857, "loss_cns_2": 0.61908, "loss_yns_2": 0.1639, "loss_cls_3": 1.11809, "loss_box_3": 2.40974, "loss_cns_3": 0.62249, "loss_yns_3": 0.16514, "loss_cls_4": 1.11014, "loss_box_4": 2.39656, "loss_cns_4": 0.62197, "loss_yns_4": 0.1652, "loss_cls_5": 1.12867, "loss_box_5": 2.42037, "loss_cns_5": 0.62557, "loss_yns_5": 0.16321, "loss_cls_dn_0": 0.41425, "loss_box_dn_0": 0.93798, "loss_cls_dn_1": 0.31995, "loss_box_dn_1": 1.13938, "loss_cls_dn_2": 0.33519, "loss_box_dn_2": 1.07015, "loss_cls_dn_3": 0.36234, "loss_box_dn_3": 1.08604, "loss_cls_dn_4": 0.36642, "loss_box_dn_4": 1.10967, "loss_cls_dn_5": 0.35678, "loss_box_dn_5": 1.15305, "loss_dense_depth": 1.14708, "loss": 35.13264, "grad_norm": 55.18881, "time": 1.51506} -{"mode": "train", "epoch": 1, "iter": 53, "lr": 0.00012, "memory": 49167, "data_time": 0.07831, "loss_cls_0": 1.03404, "loss_box_0": 2.04058, "loss_cns_0": 0.606, "loss_yns_0": 0.16679, "loss_cls_1": 1.0729, "loss_box_1": 2.51227, "loss_cns_1": 0.58829, "loss_yns_1": 0.16459, "loss_cls_2": 1.15073, "loss_box_2": 2.46824, "loss_cns_2": 0.61649, "loss_yns_2": 0.16281, "loss_cls_3": 1.15257, "loss_box_3": 2.46075, "loss_cns_3": 0.62039, "loss_yns_3": 0.16642, "loss_cls_4": 1.13405, "loss_box_4": 2.43783, "loss_cns_4": 0.61833, "loss_yns_4": 0.16711, "loss_cls_5": 1.13716, "loss_box_5": 2.4241, "loss_cns_5": 0.62254, "loss_yns_5": 0.16456, "loss_cls_dn_0": 0.42687, "loss_box_dn_0": 0.92888, "loss_cls_dn_1": 0.31861, "loss_box_dn_1": 1.08838, "loss_cls_dn_2": 0.32428, "loss_box_dn_2": 1.03894, "loss_cls_dn_3": 0.36448, "loss_box_dn_3": 1.04037, "loss_cls_dn_4": 0.37212, "loss_box_dn_4": 1.06052, "loss_cls_dn_5": 0.3693, "loss_box_dn_5": 1.08277, "loss_dense_depth": 1.0739, "loss": 35.17898, "grad_norm": 59.145, "time": 1.53437} -{"mode": "train", "epoch": 1, "iter": 54, "lr": 0.00012, "memory": 49167, "data_time": 0.07447, "loss_cls_0": 1.00818, "loss_box_0": 1.95285, "loss_cns_0": 0.61877, "loss_yns_0": 0.16222, "loss_cls_1": 1.06966, "loss_box_1": 2.37054, "loss_cns_1": 0.60501, "loss_yns_1": 0.16086, "loss_cls_2": 1.1446, "loss_box_2": 2.34511, "loss_cns_2": 0.628, "loss_yns_2": 0.1599, "loss_cls_3": 1.12219, "loss_box_3": 2.29764, "loss_cns_3": 0.63303, "loss_yns_3": 0.16259, "loss_cls_4": 1.11975, "loss_box_4": 2.26461, "loss_cns_4": 0.63402, "loss_yns_4": 0.16489, "loss_cls_5": 1.13518, "loss_box_5": 2.25827, "loss_cns_5": 0.63608, "loss_yns_5": 0.1609, "loss_cls_dn_0": 0.40979, "loss_box_dn_0": 0.91551, "loss_cls_dn_1": 0.305, "loss_box_dn_1": 1.08298, "loss_cls_dn_2": 0.30963, "loss_box_dn_2": 1.04642, "loss_cls_dn_3": 0.34972, "loss_box_dn_3": 1.02593, "loss_cls_dn_4": 0.35109, "loss_box_dn_4": 1.03274, "loss_cls_dn_5": 0.36526, "loss_box_dn_5": 1.04581, "loss_dense_depth": 1.12067, "loss": 34.17538, "grad_norm": 54.14576, "time": 1.53324} -{"mode": "train", "epoch": 1, "iter": 55, "lr": 0.00012, "memory": 49167, "data_time": 0.07903, "loss_cls_0": 1.01111, "loss_box_0": 1.94755, "loss_cns_0": 0.61396, "loss_yns_0": 0.16411, "loss_cls_1": 1.05764, "loss_box_1": 2.28481, "loss_cns_1": 0.61236, "loss_yns_1": 0.16422, "loss_cls_2": 1.12507, "loss_box_2": 2.27459, "loss_cns_2": 0.63024, "loss_yns_2": 0.16485, "loss_cls_3": 1.09864, "loss_box_3": 2.2282, "loss_cns_3": 0.63186, "loss_yns_3": 0.16949, "loss_cls_4": 1.13732, "loss_box_4": 2.20543, "loss_cns_4": 0.63783, "loss_yns_4": 0.17218, "loss_cls_5": 1.1283, "loss_box_5": 2.20767, "loss_cns_5": 0.63467, "loss_yns_5": 0.16699, "loss_cls_dn_0": 0.40745, "loss_box_dn_0": 0.89909, "loss_cls_dn_1": 0.33113, "loss_box_dn_1": 0.98427, "loss_cls_dn_2": 0.33614, "loss_box_dn_2": 0.96119, "loss_cls_dn_3": 0.36732, "loss_box_dn_3": 0.93872, "loss_cls_dn_4": 0.35761, "loss_box_dn_4": 0.94469, "loss_cls_dn_5": 0.40031, "loss_box_dn_5": 0.95271, "loss_dense_depth": 1.0696, "loss": 33.41932, "grad_norm": 61.55417, "time": 1.54594} -{"mode": "train", "epoch": 1, "iter": 56, "lr": 0.00012, "memory": 49167, "data_time": 0.08019, "loss_cls_0": 0.98631, "loss_box_0": 1.90076, "loss_cns_0": 0.60942, "loss_yns_0": 0.16312, "loss_cls_1": 1.0403, "loss_box_1": 2.18339, "loss_cns_1": 0.60819, "loss_yns_1": 0.16478, "loss_cls_2": 1.10691, "loss_box_2": 2.16475, "loss_cns_2": 0.63101, "loss_yns_2": 0.16537, "loss_cls_3": 1.13835, "loss_box_3": 2.14183, "loss_cns_3": 0.63115, "loss_yns_3": 0.1688, "loss_cls_4": 1.16334, "loss_box_4": 2.11631, "loss_cns_4": 0.63898, "loss_yns_4": 0.17644, "loss_cls_5": 1.10502, "loss_box_5": 2.08654, "loss_cns_5": 0.63515, "loss_yns_5": 0.16847, "loss_cls_dn_0": 0.39182, "loss_box_dn_0": 0.89702, "loss_cls_dn_1": 0.31747, "loss_box_dn_1": 0.97768, "loss_cls_dn_2": 0.33641, "loss_box_dn_2": 0.9598, "loss_cls_dn_3": 0.3448, "loss_box_dn_3": 0.94942, "loss_cls_dn_4": 0.33576, "loss_box_dn_4": 0.95121, "loss_cls_dn_5": 0.39033, "loss_box_dn_5": 0.94259, "loss_dense_depth": 1.01872, "loss": 32.70774, "grad_norm": 62.49037, "time": 1.54811} -{"mode": "train", "epoch": 1, "iter": 57, "lr": 0.00012, "memory": 49167, "data_time": 0.07845, "loss_cls_0": 1.00067, "loss_box_0": 1.87145, "loss_cns_0": 0.61071, "loss_yns_0": 0.16216, "loss_cls_1": 1.05095, "loss_box_1": 2.18104, "loss_cns_1": 0.61168, "loss_yns_1": 0.16621, "loss_cls_2": 1.09576, "loss_box_2": 2.16475, "loss_cns_2": 0.63159, "loss_yns_2": 0.16773, "loss_cls_3": 1.17958, "loss_box_3": 2.18623, "loss_cns_3": 0.63135, "loss_yns_3": 0.1703, "loss_cls_4": 1.19033, "loss_box_4": 2.14826, "loss_cns_4": 0.63914, "loss_yns_4": 0.17542, "loss_cls_5": 1.10337, "loss_box_5": 2.11295, "loss_cns_5": 0.63572, "loss_yns_5": 0.16899, "loss_cls_dn_0": 0.38206, "loss_box_dn_0": 0.89248, "loss_cls_dn_1": 0.31295, "loss_box_dn_1": 0.94231, "loss_cls_dn_2": 0.34121, "loss_box_dn_2": 0.92629, "loss_cls_dn_3": 0.33354, "loss_box_dn_3": 0.94885, "loss_cls_dn_4": 0.32959, "loss_box_dn_4": 0.94244, "loss_cls_dn_5": 0.37857, "loss_box_dn_5": 0.93767, "loss_dense_depth": 1.10006, "loss": 32.82435, "grad_norm": 62.00573, "time": 1.54795} -{"mode": "train", "epoch": 1, "iter": 58, "lr": 0.00012, "memory": 49167, "data_time": 0.07935, "loss_cls_0": 1.00359, "loss_box_0": 1.8841, "loss_cns_0": 0.61286, "loss_yns_0": 0.16297, "loss_cls_1": 1.03957, "loss_box_1": 2.14616, "loss_cns_1": 0.61434, "loss_yns_1": 0.16592, "loss_cls_2": 1.09024, "loss_box_2": 2.0704, "loss_cns_2": 0.63641, "loss_yns_2": 0.17014, "loss_cls_3": 1.11992, "loss_box_3": 2.13794, "loss_cns_3": 0.6379, "loss_yns_3": 0.17052, "loss_cls_4": 1.11765, "loss_box_4": 2.09482, "loss_cns_4": 0.63985, "loss_yns_4": 0.16905, "loss_cls_5": 1.10198, "loss_box_5": 2.09423, "loss_cns_5": 0.6416, "loss_yns_5": 0.16689, "loss_cls_dn_0": 0.37117, "loss_box_dn_0": 0.88712, "loss_cls_dn_1": 0.31138, "loss_box_dn_1": 0.9532, "loss_cls_dn_2": 0.34938, "loss_box_dn_2": 0.92178, "loss_cls_dn_3": 0.34173, "loss_box_dn_3": 0.96029, "loss_cls_dn_4": 0.34744, "loss_box_dn_4": 0.94177, "loss_cls_dn_5": 0.37194, "loss_box_dn_5": 0.95897, "loss_dense_depth": 0.93455, "loss": 32.33976, "grad_norm": 49.40229, "time": 1.54289} -{"mode": "train", "epoch": 1, "iter": 59, "lr": 0.00012, "memory": 49167, "data_time": 0.07838, "loss_cls_0": 0.98755, "loss_box_0": 1.88691, "loss_cns_0": 0.61384, "loss_yns_0": 0.15921, "loss_cls_1": 1.02603, "loss_box_1": 2.17419, "loss_cns_1": 0.615, "loss_yns_1": 0.16369, "loss_cls_2": 1.08016, "loss_box_2": 2.11327, "loss_cns_2": 0.63625, "loss_yns_2": 0.16941, "loss_cls_3": 1.08404, "loss_box_3": 2.16709, "loss_cns_3": 0.64035, "loss_yns_3": 0.16785, "loss_cls_4": 1.08341, "loss_box_4": 2.12031, "loss_cns_4": 0.64152, "loss_yns_4": 0.16853, "loss_cls_5": 1.10447, "loss_box_5": 2.1218, "loss_cns_5": 0.64357, "loss_yns_5": 0.16391, "loss_cls_dn_0": 0.3694, "loss_box_dn_0": 0.88471, "loss_cls_dn_1": 0.3134, "loss_box_dn_1": 0.9679, "loss_cls_dn_2": 0.34899, "loss_box_dn_2": 0.93631, "loss_cls_dn_3": 0.35647, "loss_box_dn_3": 0.96649, "loss_cls_dn_4": 0.37773, "loss_box_dn_4": 0.94182, "loss_cls_dn_5": 0.36018, "loss_box_dn_5": 0.96639, "loss_dense_depth": 1.0414, "loss": 32.56358, "grad_norm": 55.7274, "time": 1.53957} -{"mode": "train", "epoch": 1, "iter": 60, "lr": 0.00012, "memory": 49167, "data_time": 0.08657, "loss_cls_0": 0.98527, "loss_box_0": 1.84257, "loss_cns_0": 0.6155, "loss_yns_0": 0.15871, "loss_cls_1": 1.03664, "loss_box_1": 2.20579, "loss_cns_1": 0.61361, "loss_yns_1": 0.16629, "loss_cls_2": 1.08173, "loss_box_2": 2.16075, "loss_cns_2": 0.63487, "loss_yns_2": 0.16479, "loss_cls_3": 1.09751, "loss_box_3": 2.18473, "loss_cns_3": 0.63901, "loss_yns_3": 0.1641, "loss_cls_4": 1.10609, "loss_box_4": 2.14079, "loss_cns_4": 0.64455, "loss_yns_4": 0.16962, "loss_cls_5": 1.11164, "loss_box_5": 2.15913, "loss_cns_5": 0.64019, "loss_yns_5": 0.1677, "loss_cls_dn_0": 0.37923, "loss_box_dn_0": 0.89005, "loss_cls_dn_1": 0.29426, "loss_box_dn_1": 0.97477, "loss_cls_dn_2": 0.33655, "loss_box_dn_2": 0.95022, "loss_cls_dn_3": 0.36246, "loss_box_dn_3": 0.96702, "loss_cls_dn_4": 0.39165, "loss_box_dn_4": 0.94705, "loss_cls_dn_5": 0.35001, "loss_box_dn_5": 0.98047, "loss_dense_depth": 0.98649, "loss": 32.7018, "grad_norm": 50.02901, "time": 1.54584} -{"mode": "train", "epoch": 1, "iter": 61, "lr": 0.00012, "memory": 49167, "data_time": 0.11854, "loss_cls_0": 0.98686, "loss_box_0": 1.85149, "loss_cns_0": 0.61092, "loss_yns_0": 0.15904, "loss_cls_1": 1.0639, "loss_box_1": 2.1404, "loss_cns_1": 0.61827, "loss_yns_1": 0.16414, "loss_cls_2": 1.07918, "loss_box_2": 2.09082, "loss_cns_2": 0.6366, "loss_yns_2": 0.16347, "loss_cls_3": 1.11625, "loss_box_3": 2.1072, "loss_cns_3": 0.64209, "loss_yns_3": 0.16171, "loss_cls_4": 1.11401, "loss_box_4": 2.11315, "loss_cns_4": 0.64647, "loss_yns_4": 0.16732, "loss_cls_5": 1.122, "loss_box_5": 2.14511, "loss_cns_5": 0.63542, "loss_yns_5": 0.16805, "loss_cls_dn_0": 0.38991, "loss_box_dn_0": 0.88129, "loss_cls_dn_1": 0.28607, "loss_box_dn_1": 0.99353, "loss_cls_dn_2": 0.32367, "loss_box_dn_2": 0.95707, "loss_cls_dn_3": 0.35437, "loss_box_dn_3": 0.96217, "loss_cls_dn_4": 0.38521, "loss_box_dn_4": 0.96809, "loss_cls_dn_5": 0.3432, "loss_box_dn_5": 0.99414, "loss_dense_depth": 1.00024, "loss": 32.54284, "grad_norm": 45.82271, "time": 1.60923} -{"mode": "train", "epoch": 1, "iter": 62, "lr": 0.00012, "memory": 49167, "data_time": 0.12313, "loss_cls_0": 0.98873, "loss_box_0": 1.88667, "loss_cns_0": 0.61168, "loss_yns_0": 0.15822, "loss_cls_1": 1.04017, "loss_box_1": 2.15894, "loss_cns_1": 0.62363, "loss_yns_1": 0.16242, "loss_cls_2": 1.07452, "loss_box_2": 2.10588, "loss_cns_2": 0.63539, "loss_yns_2": 0.16216, "loss_cls_3": 1.09308, "loss_box_3": 2.11274, "loss_cns_3": 0.64125, "loss_yns_3": 0.16251, "loss_cls_4": 1.09124, "loss_box_4": 2.12578, "loss_cns_4": 0.64683, "loss_yns_4": 0.16418, "loss_cls_5": 1.10275, "loss_box_5": 2.1483, "loss_cns_5": 0.63754, "loss_yns_5": 0.16273, "loss_cls_dn_0": 0.3875, "loss_box_dn_0": 0.8831, "loss_cls_dn_1": 0.29608, "loss_box_dn_1": 0.9254, "loss_cls_dn_2": 0.32061, "loss_box_dn_2": 0.884, "loss_cls_dn_3": 0.34965, "loss_box_dn_3": 0.88428, "loss_cls_dn_4": 0.37869, "loss_box_dn_4": 0.8982, "loss_cls_dn_5": 0.36277, "loss_box_dn_5": 0.91315, "loss_dense_depth": 0.98656, "loss": 32.16731, "grad_norm": 33.40319, "time": 1.60685} -{"mode": "train", "epoch": 1, "iter": 63, "lr": 0.00012, "memory": 49167, "data_time": 0.10313, "loss_cls_0": 0.99657, "loss_box_0": 1.89213, "loss_cns_0": 0.61361, "loss_yns_0": 0.16118, "loss_cls_1": 1.04321, "loss_box_1": 2.13647, "loss_cns_1": 0.62636, "loss_yns_1": 0.16503, "loss_cls_2": 1.13278, "loss_box_2": 2.07667, "loss_cns_2": 0.63786, "loss_yns_2": 0.16308, "loss_cls_3": 1.12584, "loss_box_3": 2.05889, "loss_cns_3": 0.64471, "loss_yns_3": 0.16343, "loss_cls_4": 1.11561, "loss_box_4": 2.07208, "loss_cns_4": 0.6461, "loss_yns_4": 0.16263, "loss_cls_5": 1.09447, "loss_box_5": 2.08993, "loss_cns_5": 0.64626, "loss_yns_5": 0.1624, "loss_cls_dn_0": 0.36746, "loss_box_dn_0": 0.8759, "loss_cls_dn_1": 0.29361, "loss_box_dn_1": 0.91519, "loss_cls_dn_2": 0.30535, "loss_box_dn_2": 0.87686, "loss_cls_dn_3": 0.31569, "loss_box_dn_3": 0.87348, "loss_cls_dn_4": 0.33865, "loss_box_dn_4": 0.88789, "loss_cls_dn_5": 0.36263, "loss_box_dn_5": 0.89998, "loss_dense_depth": 1.01451, "loss": 31.95449, "grad_norm": 42.73325, "time": 1.55316} -{"mode": "train", "epoch": 1, "iter": 64, "lr": 0.00013, "memory": 49167, "data_time": 0.0782, "loss_cls_0": 0.9877, "loss_box_0": 1.85148, "loss_cns_0": 0.61965, "loss_yns_0": 0.16292, "loss_cls_1": 1.056, "loss_box_1": 2.13144, "loss_cns_1": 0.62773, "loss_yns_1": 0.16821, "loss_cls_2": 1.09527, "loss_box_2": 2.09419, "loss_cns_2": 0.63653, "loss_yns_2": 0.16231, "loss_cls_3": 1.14873, "loss_box_3": 2.08393, "loss_cns_3": 0.64002, "loss_yns_3": 0.16623, "loss_cls_4": 1.15128, "loss_box_4": 2.05237, "loss_cns_4": 0.64661, "loss_yns_4": 0.16932, "loss_cls_5": 1.09136, "loss_box_5": 2.10174, "loss_cns_5": 0.64657, "loss_yns_5": 0.16726, "loss_cls_dn_0": 0.34886, "loss_box_dn_0": 0.85371, "loss_cls_dn_1": 0.27459, "loss_box_dn_1": 0.9154, "loss_cls_dn_2": 0.29614, "loss_box_dn_2": 0.88487, "loss_cls_dn_3": 0.29416, "loss_box_dn_3": 0.88284, "loss_cls_dn_4": 0.30573, "loss_box_dn_4": 0.88391, "loss_cls_dn_5": 0.34739, "loss_box_dn_5": 0.90614, "loss_dense_depth": 0.96953, "loss": 31.82213, "grad_norm": 41.39115, "time": 1.51009} -{"mode": "train", "epoch": 1, "iter": 65, "lr": 0.00013, "memory": 49167, "data_time": 0.0761, "loss_cls_0": 0.97795, "loss_box_0": 1.85774, "loss_cns_0": 0.61944, "loss_yns_0": 0.16623, "loss_cls_1": 1.05454, "loss_box_1": 2.15839, "loss_cns_1": 0.62909, "loss_yns_1": 0.16605, "loss_cls_2": 1.06848, "loss_box_2": 2.14231, "loss_cns_2": 0.62879, "loss_yns_2": 0.16168, "loss_cls_3": 1.08096, "loss_box_3": 2.14933, "loss_cns_3": 0.63392, "loss_yns_3": 0.16394, "loss_cls_4": 1.11157, "loss_box_4": 2.08154, "loss_cns_4": 0.64965, "loss_yns_4": 0.17113, "loss_cls_5": 1.09448, "loss_box_5": 2.10092, "loss_cns_5": 0.65137, "loss_yns_5": 0.17024, "loss_cls_dn_0": 0.35567, "loss_box_dn_0": 0.85627, "loss_cls_dn_1": 0.25533, "loss_box_dn_1": 0.95934, "loss_cls_dn_2": 0.29855, "loss_box_dn_2": 0.93498, "loss_cls_dn_3": 0.29907, "loss_box_dn_3": 0.93723, "loss_cls_dn_4": 0.29905, "loss_box_dn_4": 0.92621, "loss_cls_dn_5": 0.33428, "loss_box_dn_5": 0.9412, "loss_dense_depth": 0.98123, "loss": 32.06815, "grad_norm": 36.77763, "time": 1.53146} -{"mode": "train", "epoch": 1, "iter": 66, "lr": 0.00013, "memory": 49167, "data_time": 0.10883, "loss_cls_0": 0.96302, "loss_box_0": 1.86074, "loss_cns_0": 0.62054, "loss_yns_0": 0.16509, "loss_cls_1": 1.06236, "loss_box_1": 2.09113, "loss_cns_1": 0.63395, "loss_yns_1": 0.16411, "loss_cls_2": 1.07323, "loss_box_2": 2.03497, "loss_cns_2": 0.64375, "loss_yns_2": 0.16283, "loss_cls_3": 1.09188, "loss_box_3": 2.03723, "loss_cns_3": 0.64718, "loss_yns_3": 0.16478, "loss_cls_4": 1.09744, "loss_box_4": 2.02273, "loss_cns_4": 0.65391, "loss_yns_4": 0.16809, "loss_cls_5": 1.10256, "loss_box_5": 2.04108, "loss_cns_5": 0.64894, "loss_yns_5": 0.16735, "loss_cls_dn_0": 0.35162, "loss_box_dn_0": 0.84984, "loss_cls_dn_1": 0.23526, "loss_box_dn_1": 0.99256, "loss_cls_dn_2": 0.29247, "loss_box_dn_2": 0.95233, "loss_cls_dn_3": 0.30237, "loss_box_dn_3": 0.95311, "loss_cls_dn_4": 0.29879, "loss_box_dn_4": 0.9586, "loss_cls_dn_5": 0.31432, "loss_box_dn_5": 0.97408, "loss_dense_depth": 0.93324, "loss": 31.7275, "grad_norm": 31.4052, "time": 1.5631} -{"mode": "train", "epoch": 1, "iter": 67, "lr": 0.00013, "memory": 49167, "data_time": 0.07436, "loss_cls_0": 0.98438, "loss_box_0": 1.88351, "loss_cns_0": 0.61595, "loss_yns_0": 0.16597, "loss_cls_1": 1.07046, "loss_box_1": 2.11302, "loss_cns_1": 0.62447, "loss_yns_1": 0.16571, "loss_cls_2": 1.07017, "loss_box_2": 2.03842, "loss_cns_2": 0.64533, "loss_yns_2": 0.16218, "loss_cls_3": 1.0784, "loss_box_3": 2.04451, "loss_cns_3": 0.64786, "loss_yns_3": 0.16208, "loss_cls_4": 1.08124, "loss_box_4": 2.05024, "loss_cns_4": 0.64952, "loss_yns_4": 0.16159, "loss_cls_5": 1.104, "loss_box_5": 2.05553, "loss_cns_5": 0.64638, "loss_yns_5": 0.16174, "loss_cls_dn_0": 0.35044, "loss_box_dn_0": 0.85646, "loss_cls_dn_1": 0.23127, "loss_box_dn_1": 0.95557, "loss_cls_dn_2": 0.28226, "loss_box_dn_2": 0.91029, "loss_cls_dn_3": 0.30257, "loss_box_dn_3": 0.91745, "loss_cls_dn_4": 0.30843, "loss_box_dn_4": 0.92745, "loss_cls_dn_5": 0.3075, "loss_box_dn_5": 0.941, "loss_dense_depth": 0.92035, "loss": 31.59371, "grad_norm": 35.94728, "time": 1.51073} -{"mode": "train", "epoch": 1, "iter": 68, "lr": 0.00013, "memory": 49167, "data_time": 0.07257, "loss_cls_0": 0.98309, "loss_box_0": 1.84412, "loss_cns_0": 0.62084, "loss_yns_0": 0.1642, "loss_cls_1": 1.04675, "loss_box_1": 2.08637, "loss_cns_1": 0.62034, "loss_yns_1": 0.16446, "loss_cls_2": 1.07783, "loss_box_2": 2.01353, "loss_cns_2": 0.64452, "loss_yns_2": 0.16322, "loss_cls_3": 1.07422, "loss_box_3": 2.01324, "loss_cns_3": 0.6485, "loss_yns_3": 0.16618, "loss_cls_4": 1.08465, "loss_box_4": 2.02304, "loss_cns_4": 0.6499, "loss_yns_4": 0.1668, "loss_cls_5": 1.13944, "loss_box_5": 2.01809, "loss_cns_5": 0.65223, "loss_yns_5": 0.16928, "loss_cls_dn_0": 0.34803, "loss_box_dn_0": 0.86557, "loss_cls_dn_1": 0.2424, "loss_box_dn_1": 0.89534, "loss_cls_dn_2": 0.27949, "loss_box_dn_2": 0.84896, "loss_cls_dn_3": 0.30229, "loss_box_dn_3": 0.85194, "loss_cls_dn_4": 0.32347, "loss_box_dn_4": 0.86093, "loss_cls_dn_5": 0.31696, "loss_box_dn_5": 0.87141, "loss_dense_depth": 0.92061, "loss": 31.16224, "grad_norm": 37.248, "time": 1.52041} -{"mode": "train", "epoch": 1, "iter": 69, "lr": 0.00013, "memory": 49167, "data_time": 0.08366, "loss_cls_0": 0.94936, "loss_box_0": 1.79845, "loss_cns_0": 0.62366, "loss_yns_0": 0.15922, "loss_cls_1": 1.00147, "loss_box_1": 2.07773, "loss_cns_1": 0.62083, "loss_yns_1": 0.16267, "loss_cls_2": 1.06474, "loss_box_2": 1.98433, "loss_cns_2": 0.64625, "loss_yns_2": 0.16047, "loss_cls_3": 1.08207, "loss_box_3": 1.97617, "loss_cns_3": 0.65062, "loss_yns_3": 0.16538, "loss_cls_4": 1.07627, "loss_box_4": 1.99588, "loss_cns_4": 0.65078, "loss_yns_4": 0.16378, "loss_cls_5": 1.08525, "loss_box_5": 2.00134, "loss_cns_5": 0.64726, "loss_yns_5": 0.1643, "loss_cls_dn_0": 0.3305, "loss_box_dn_0": 0.85536, "loss_cls_dn_1": 0.25453, "loss_box_dn_1": 0.89779, "loss_cls_dn_2": 0.27026, "loss_box_dn_2": 0.83748, "loss_cls_dn_3": 0.28904, "loss_box_dn_3": 0.83465, "loss_cls_dn_4": 0.32254, "loss_box_dn_4": 0.84455, "loss_cls_dn_5": 0.33462, "loss_box_dn_5": 0.85873, "loss_dense_depth": 0.86001, "loss": 30.69832, "grad_norm": 34.49989, "time": 1.50987} -{"mode": "train", "epoch": 1, "iter": 70, "lr": 0.00013, "memory": 49167, "data_time": 0.0733, "loss_cls_0": 0.95496, "loss_box_0": 1.83469, "loss_cns_0": 0.61755, "loss_yns_0": 0.1586, "loss_cls_1": 0.99518, "loss_box_1": 2.02509, "loss_cns_1": 0.61437, "loss_yns_1": 0.1597, "loss_cls_2": 1.04153, "loss_box_2": 1.97359, "loss_cns_2": 0.63576, "loss_yns_2": 0.15925, "loss_cls_3": 1.06785, "loss_box_3": 1.96236, "loss_cns_3": 0.64494, "loss_yns_3": 0.16129, "loss_cls_4": 1.05836, "loss_box_4": 1.95918, "loss_cns_4": 0.64661, "loss_yns_4": 0.16046, "loss_cls_5": 1.0555, "loss_box_5": 1.95982, "loss_cns_5": 0.64506, "loss_yns_5": 0.16237, "loss_cls_dn_0": 0.33244, "loss_box_dn_0": 0.8679, "loss_cls_dn_1": 0.27702, "loss_box_dn_1": 0.84074, "loss_cls_dn_2": 0.28539, "loss_box_dn_2": 0.79651, "loss_cls_dn_3": 0.2933, "loss_box_dn_3": 0.80053, "loss_cls_dn_4": 0.32998, "loss_box_dn_4": 0.80992, "loss_cls_dn_5": 0.35856, "loss_box_dn_5": 0.82751, "loss_dense_depth": 0.89079, "loss": 30.36469, "grad_norm": 37.34963, "time": 1.54673} -{"mode": "train", "epoch": 1, "iter": 71, "lr": 0.00013, "memory": 49167, "data_time": 0.07829, "loss_cls_0": 0.95733, "loss_box_0": 1.84133, "loss_cns_0": 0.6212, "loss_yns_0": 0.16049, "loss_cls_1": 1.00486, "loss_box_1": 2.05206, "loss_cns_1": 0.61517, "loss_yns_1": 0.15759, "loss_cls_2": 1.04083, "loss_box_2": 1.96751, "loss_cns_2": 0.6396, "loss_yns_2": 0.15803, "loss_cls_3": 1.06553, "loss_box_3": 1.96984, "loss_cns_3": 0.64841, "loss_yns_3": 0.15835, "loss_cls_4": 1.05777, "loss_box_4": 1.96747, "loss_cns_4": 0.64935, "loss_yns_4": 0.15964, "loss_cls_5": 1.05288, "loss_box_5": 1.97295, "loss_cns_5": 0.646, "loss_yns_5": 0.16177, "loss_cls_dn_0": 0.31424, "loss_box_dn_0": 0.87637, "loss_cls_dn_1": 0.26447, "loss_box_dn_1": 0.85327, "loss_cls_dn_2": 0.27285, "loss_box_dn_2": 0.81573, "loss_cls_dn_3": 0.26951, "loss_box_dn_3": 0.82246, "loss_cls_dn_4": 0.3058, "loss_box_dn_4": 0.83519, "loss_cls_dn_5": 0.32996, "loss_box_dn_5": 0.85463, "loss_dense_depth": 0.8893, "loss": 30.42975, "grad_norm": 42.51303, "time": 1.51133} -{"mode": "train", "epoch": 1, "iter": 72, "lr": 0.00013, "memory": 49167, "data_time": 0.07707, "loss_cls_0": 0.95669, "loss_box_0": 1.81808, "loss_cns_0": 0.62052, "loss_yns_0": 0.16248, "loss_cls_1": 1.00698, "loss_box_1": 2.04505, "loss_cns_1": 0.61784, "loss_yns_1": 0.16035, "loss_cls_2": 1.03405, "loss_box_2": 1.92971, "loss_cns_2": 0.64269, "loss_yns_2": 0.1593, "loss_cls_3": 1.05571, "loss_box_3": 1.9533, "loss_cns_3": 0.64588, "loss_yns_3": 0.16222, "loss_cls_4": 1.07033, "loss_box_4": 1.93328, "loss_cns_4": 0.64945, "loss_yns_4": 0.1622, "loss_cls_5": 1.06642, "loss_box_5": 1.94661, "loss_cns_5": 0.64594, "loss_yns_5": 0.16364, "loss_cls_dn_0": 0.31268, "loss_box_dn_0": 0.86435, "loss_cls_dn_1": 0.25765, "loss_box_dn_1": 0.84551, "loss_cls_dn_2": 0.27212, "loss_box_dn_2": 0.81385, "loss_cls_dn_3": 0.26742, "loss_box_dn_3": 0.82663, "loss_cls_dn_4": 0.29714, "loss_box_dn_4": 0.84001, "loss_cls_dn_5": 0.31752, "loss_box_dn_5": 0.86011, "loss_dense_depth": 0.89082, "loss": 30.27455, "grad_norm": 41.905, "time": 1.52416} -{"mode": "train", "epoch": 1, "iter": 73, "lr": 0.00013, "memory": 49167, "data_time": 0.07443, "loss_cls_0": 0.964, "loss_box_0": 1.8327, "loss_cns_0": 0.61118, "loss_yns_0": 0.15853, "loss_cls_1": 1.00106, "loss_box_1": 2.05427, "loss_cns_1": 0.61975, "loss_yns_1": 0.16253, "loss_cls_2": 1.0543, "loss_box_2": 1.95605, "loss_cns_2": 0.64316, "loss_yns_2": 0.1593, "loss_cls_3": 1.06899, "loss_box_3": 1.97098, "loss_cns_3": 0.64665, "loss_yns_3": 0.16389, "loss_cls_4": 1.06038, "loss_box_4": 1.95952, "loss_cns_4": 0.64945, "loss_yns_4": 0.16229, "loss_cls_5": 1.0576, "loss_box_5": 1.96105, "loss_cns_5": 0.6479, "loss_yns_5": 0.1619, "loss_cls_dn_0": 0.31511, "loss_box_dn_0": 0.85885, "loss_cls_dn_1": 0.25253, "loss_box_dn_1": 0.85714, "loss_cls_dn_2": 0.28465, "loss_box_dn_2": 0.82861, "loss_cls_dn_3": 0.28649, "loss_box_dn_3": 0.83959, "loss_cls_dn_4": 0.29229, "loss_box_dn_4": 0.85394, "loss_cls_dn_5": 0.31345, "loss_box_dn_5": 0.87293, "loss_dense_depth": 0.87604, "loss": 30.45903, "grad_norm": 32.68764, "time": 1.52298} -{"mode": "train", "epoch": 1, "iter": 74, "lr": 0.00013, "memory": 49167, "data_time": 0.07591, "loss_cls_0": 0.95607, "loss_box_0": 1.81087, "loss_cns_0": 0.61635, "loss_yns_0": 0.15875, "loss_cls_1": 1.00802, "loss_box_1": 2.06583, "loss_cns_1": 0.62623, "loss_yns_1": 0.16419, "loss_cls_2": 1.03766, "loss_box_2": 1.99844, "loss_cns_2": 0.64444, "loss_yns_2": 0.16217, "loss_cls_3": 1.06774, "loss_box_3": 1.98949, "loss_cns_3": 0.6517, "loss_yns_3": 0.16202, "loss_cls_4": 1.05305, "loss_box_4": 2.01579, "loss_cns_4": 0.64925, "loss_yns_4": 0.16328, "loss_cls_5": 1.04891, "loss_box_5": 2.0061, "loss_cns_5": 0.64971, "loss_yns_5": 0.16099, "loss_cls_dn_0": 0.3119, "loss_box_dn_0": 0.85188, "loss_cls_dn_1": 0.23553, "loss_box_dn_1": 0.87998, "loss_cls_dn_2": 0.26608, "loss_box_dn_2": 0.84917, "loss_cls_dn_3": 0.27993, "loss_box_dn_3": 0.84953, "loss_cls_dn_4": 0.2765, "loss_box_dn_4": 0.86651, "loss_cls_dn_5": 0.29658, "loss_box_dn_5": 0.87833, "loss_dense_depth": 0.87865, "loss": 30.58763, "grad_norm": 39.80866, "time": 1.5093} -{"mode": "train", "epoch": 1, "iter": 75, "lr": 0.00013, "memory": 49167, "data_time": 0.07553, "loss_cls_0": 0.96037, "loss_box_0": 1.83429, "loss_cns_0": 0.61428, "loss_yns_0": 0.16336, "loss_cls_1": 1.04411, "loss_box_1": 2.10911, "loss_cns_1": 0.62235, "loss_yns_1": 0.16375, "loss_cls_2": 1.09842, "loss_box_2": 2.04601, "loss_cns_2": 0.64032, "loss_yns_2": 0.16273, "loss_cls_3": 1.05086, "loss_box_3": 2.01265, "loss_cns_3": 0.65011, "loss_yns_3": 0.16197, "loss_cls_4": 1.08525, "loss_box_4": 2.00722, "loss_cns_4": 0.65132, "loss_yns_4": 0.16685, "loss_cls_5": 1.16469, "loss_box_5": 2.00101, "loss_cns_5": 0.65341, "loss_yns_5": 0.16499, "loss_cls_dn_0": 0.30138, "loss_box_dn_0": 0.84318, "loss_cls_dn_1": 0.21469, "loss_box_dn_1": 0.88228, "loss_cls_dn_2": 0.23785, "loss_box_dn_2": 0.84698, "loss_cls_dn_3": 0.24914, "loss_box_dn_3": 0.83571, "loss_cls_dn_4": 0.25993, "loss_box_dn_4": 0.83466, "loss_cls_dn_5": 0.27117, "loss_box_dn_5": 0.84281, "loss_dense_depth": 0.86399, "loss": 30.71322, "grad_norm": 53.81155, "time": 1.54119} -{"mode": "train", "epoch": 1, "iter": 76, "lr": 0.00013, "memory": 49167, "data_time": 0.07916, "loss_cls_0": 0.96935, "loss_box_0": 1.82522, "loss_cns_0": 0.61435, "loss_yns_0": 0.16398, "loss_cls_1": 1.01925, "loss_box_1": 2.07347, "loss_cns_1": 0.6256, "loss_yns_1": 0.16255, "loss_cls_2": 1.06062, "loss_box_2": 2.03406, "loss_cns_2": 0.64271, "loss_yns_2": 0.16315, "loss_cls_3": 1.06815, "loss_box_3": 2.00966, "loss_cns_3": 0.64877, "loss_yns_3": 0.16389, "loss_cls_4": 1.0699, "loss_box_4": 2.00191, "loss_cns_4": 0.64765, "loss_yns_4": 0.16993, "loss_cls_5": 1.05342, "loss_box_5": 2.00334, "loss_cns_5": 0.64928, "loss_yns_5": 0.16492, "loss_cls_dn_0": 0.30036, "loss_box_dn_0": 0.84263, "loss_cls_dn_1": 0.21168, "loss_box_dn_1": 0.83651, "loss_cls_dn_2": 0.23172, "loss_box_dn_2": 0.80926, "loss_cls_dn_3": 0.23728, "loss_box_dn_3": 0.79982, "loss_cls_dn_4": 0.26776, "loss_box_dn_4": 0.80185, "loss_cls_dn_5": 0.28565, "loss_box_dn_5": 0.80771, "loss_dense_depth": 0.85938, "loss": 30.29672, "grad_norm": 34.53444, "time": 1.52021} -{"mode": "train", "epoch": 1, "iter": 77, "lr": 0.00013, "memory": 49167, "data_time": 0.07957, "loss_cls_0": 0.94739, "loss_box_0": 1.85971, "loss_cns_0": 0.60844, "loss_yns_0": 0.15922, "loss_cls_1": 1.04992, "loss_box_1": 2.03487, "loss_cns_1": 0.63262, "loss_yns_1": 0.16143, "loss_cls_2": 1.06116, "loss_box_2": 2.00677, "loss_cns_2": 0.64868, "loss_yns_2": 0.16263, "loss_cls_3": 1.05356, "loss_box_3": 1.97705, "loss_cns_3": 0.65323, "loss_yns_3": 0.16115, "loss_cls_4": 1.07589, "loss_box_4": 1.95876, "loss_cns_4": 0.65291, "loss_yns_4": 0.16139, "loss_cls_5": 1.08542, "loss_box_5": 1.95129, "loss_cns_5": 0.65443, "loss_yns_5": 0.1623, "loss_cls_dn_0": 0.29576, "loss_box_dn_0": 0.83367, "loss_cls_dn_1": 0.21642, "loss_box_dn_1": 0.79633, "loss_cls_dn_2": 0.23946, "loss_box_dn_2": 0.77752, "loss_cls_dn_3": 0.23591, "loss_box_dn_3": 0.77794, "loss_cls_dn_4": 0.26572, "loss_box_dn_4": 0.78215, "loss_cls_dn_5": 0.29758, "loss_box_dn_5": 0.79018, "loss_dense_depth": 0.85079, "loss": 30.03968, "grad_norm": 35.53545, "time": 1.53041} -{"mode": "train", "epoch": 1, "iter": 78, "lr": 0.00013, "memory": 49167, "data_time": 0.07771, "loss_cls_0": 0.97167, "loss_box_0": 1.8933, "loss_cns_0": 0.61327, "loss_yns_0": 0.16147, "loss_cls_1": 1.04969, "loss_box_1": 2.13902, "loss_cns_1": 0.62046, "loss_yns_1": 0.16877, "loss_cls_2": 1.06931, "loss_box_2": 2.05939, "loss_cns_2": 0.64294, "loss_yns_2": 0.16678, "loss_cls_3": 1.10556, "loss_box_3": 2.04834, "loss_cns_3": 0.64833, "loss_yns_3": 0.16412, "loss_cls_4": 1.11942, "loss_box_4": 2.03998, "loss_cns_4": 0.64937, "loss_yns_4": 0.16686, "loss_cls_5": 1.08648, "loss_box_5": 2.04752, "loss_cns_5": 0.64938, "loss_yns_5": 0.16491, "loss_cls_dn_0": 0.28197, "loss_box_dn_0": 0.83075, "loss_cls_dn_1": 0.20594, "loss_box_dn_1": 0.83796, "loss_cls_dn_2": 0.22563, "loss_box_dn_2": 0.80074, "loss_cls_dn_3": 0.22412, "loss_box_dn_3": 0.81254, "loss_cls_dn_4": 0.23471, "loss_box_dn_4": 0.82038, "loss_cls_dn_5": 0.25948, "loss_box_dn_5": 0.83534, "loss_dense_depth": 0.85968, "loss": 30.67556, "grad_norm": 59.30885, "time": 1.51577} -{"mode": "train", "epoch": 1, "iter": 79, "lr": 0.00013, "memory": 49167, "data_time": 0.07918, "loss_cls_0": 0.98092, "loss_box_0": 1.85799, "loss_cns_0": 0.62212, "loss_yns_0": 0.16071, "loss_cls_1": 1.03949, "loss_box_1": 2.08436, "loss_cns_1": 0.63268, "loss_yns_1": 0.16675, "loss_cls_2": 1.05532, "loss_box_2": 2.01411, "loss_cns_2": 0.64685, "loss_yns_2": 0.16203, "loss_cls_3": 1.08354, "loss_box_3": 2.00347, "loss_cns_3": 0.6519, "loss_yns_3": 0.16331, "loss_cls_4": 1.11397, "loss_box_4": 2.00143, "loss_cns_4": 0.65342, "loss_yns_4": 0.16951, "loss_cls_5": 1.06759, "loss_box_5": 2.00017, "loss_cns_5": 0.65209, "loss_yns_5": 0.16347, "loss_cls_dn_0": 0.26404, "loss_box_dn_0": 0.82607, "loss_cls_dn_1": 0.20203, "loss_box_dn_1": 0.85004, "loss_cls_dn_2": 0.21738, "loss_box_dn_2": 0.82112, "loss_cls_dn_3": 0.21628, "loss_box_dn_3": 0.82717, "loss_cls_dn_4": 0.22467, "loss_box_dn_4": 0.83306, "loss_cls_dn_5": 0.23965, "loss_box_dn_5": 0.84301, "loss_dense_depth": 0.82409, "loss": 30.33581, "grad_norm": 53.11056, "time": 1.52975} -{"mode": "train", "epoch": 1, "iter": 80, "lr": 0.00013, "memory": 49167, "data_time": 0.07929, "loss_cls_0": 0.94567, "loss_box_0": 1.86274, "loss_cns_0": 0.61876, "loss_yns_0": 0.1594, "loss_cls_1": 1.02414, "loss_box_1": 2.1347, "loss_cns_1": 0.63061, "loss_yns_1": 0.16201, "loss_cls_2": 1.06501, "loss_box_2": 2.10245, "loss_cns_2": 0.64284, "loss_yns_2": 0.16231, "loss_cls_3": 1.08719, "loss_box_3": 2.0601, "loss_cns_3": 0.65326, "loss_yns_3": 0.16176, "loss_cls_4": 1.09389, "loss_box_4": 2.04964, "loss_cns_4": 0.65474, "loss_yns_4": 0.1645, "loss_cls_5": 1.04913, "loss_box_5": 2.04608, "loss_cns_5": 0.65516, "loss_yns_5": 0.16157, "loss_cls_dn_0": 0.26452, "loss_box_dn_0": 0.81876, "loss_cls_dn_1": 0.20239, "loss_box_dn_1": 0.85731, "loss_cls_dn_2": 0.21928, "loss_box_dn_2": 0.82857, "loss_cls_dn_3": 0.22589, "loss_box_dn_3": 0.81585, "loss_cls_dn_4": 0.24107, "loss_box_dn_4": 0.81493, "loss_cls_dn_5": 0.25304, "loss_box_dn_5": 0.82133, "loss_dense_depth": 0.82326, "loss": 30.53387, "grad_norm": 45.40878, "time": 1.52802} -{"mode": "train", "epoch": 1, "iter": 81, "lr": 0.00013, "memory": 49167, "data_time": 0.11222, "loss_cls_0": 0.96053, "loss_box_0": 1.88575, "loss_cns_0": 0.61775, "loss_yns_0": 0.16074, "loss_cls_1": 1.03759, "loss_box_1": 2.14233, "loss_cns_1": 0.62549, "loss_yns_1": 0.16102, "loss_cls_2": 1.08603, "loss_box_2": 2.10257, "loss_cns_2": 0.64668, "loss_yns_2": 0.16481, "loss_cls_3": 1.11633, "loss_box_3": 2.06104, "loss_cns_3": 0.6542, "loss_yns_3": 0.1623, "loss_cls_4": 1.13811, "loss_box_4": 2.04834, "loss_cns_4": 0.65423, "loss_yns_4": 0.1607, "loss_cls_5": 1.06631, "loss_box_5": 2.03768, "loss_cns_5": 0.65487, "loss_yns_5": 0.16102, "loss_cls_dn_0": 0.27064, "loss_box_dn_0": 0.81928, "loss_cls_dn_1": 0.2039, "loss_box_dn_1": 0.86792, "loss_cls_dn_2": 0.21784, "loss_box_dn_2": 0.82862, "loss_cls_dn_3": 0.2326, "loss_box_dn_3": 0.81381, "loss_cls_dn_4": 0.25325, "loss_box_dn_4": 0.81304, "loss_cls_dn_5": 0.25116, "loss_box_dn_5": 0.81397, "loss_dense_depth": 0.85156, "loss": 30.74401, "grad_norm": 48.87827, "time": 1.60931} -{"mode": "train", "epoch": 1, "iter": 82, "lr": 0.00013, "memory": 49167, "data_time": 0.07748, "loss_cls_0": 0.97281, "loss_box_0": 1.88613, "loss_cns_0": 0.61801, "loss_yns_0": 0.15881, "loss_cls_1": 1.04293, "loss_box_1": 2.12763, "loss_cns_1": 0.62167, "loss_yns_1": 0.16176, "loss_cls_2": 1.07583, "loss_box_2": 2.06423, "loss_cns_2": 0.64302, "loss_yns_2": 0.16542, "loss_cls_3": 1.09017, "loss_box_3": 2.04094, "loss_cns_3": 0.6464, "loss_yns_3": 0.16209, "loss_cls_4": 1.10221, "loss_box_4": 2.04976, "loss_cns_4": 0.64324, "loss_yns_4": 0.16211, "loss_cls_5": 1.07801, "loss_box_5": 2.02562, "loss_cns_5": 0.64126, "loss_yns_5": 0.15958, "loss_cls_dn_0": 0.26881, "loss_box_dn_0": 0.83352, "loss_cls_dn_1": 0.19612, "loss_box_dn_1": 0.85388, "loss_cls_dn_2": 0.20701, "loss_box_dn_2": 0.81077, "loss_cls_dn_3": 0.22086, "loss_box_dn_3": 0.80431, "loss_cls_dn_4": 0.2318, "loss_box_dn_4": 0.81526, "loss_cls_dn_5": 0.23111, "loss_box_dn_5": 0.81443, "loss_dense_depth": 0.86725, "loss": 30.49478, "grad_norm": 44.00044, "time": 1.5896} -{"mode": "train", "epoch": 1, "iter": 83, "lr": 0.00013, "memory": 49167, "data_time": 0.10377, "loss_cls_0": 0.9725, "loss_box_0": 1.83864, "loss_cns_0": 0.62302, "loss_yns_0": 0.16071, "loss_cls_1": 1.02776, "loss_box_1": 2.05155, "loss_cns_1": 0.61913, "loss_yns_1": 0.16279, "loss_cls_2": 1.08604, "loss_box_2": 1.96082, "loss_cns_2": 0.6406, "loss_yns_2": 0.16119, "loss_cls_3": 1.06852, "loss_box_3": 1.93283, "loss_cns_3": 0.64558, "loss_yns_3": 0.16078, "loss_cls_4": 1.09219, "loss_box_4": 1.96931, "loss_cns_4": 0.64256, "loss_yns_4": 0.16403, "loss_cls_5": 1.10577, "loss_box_5": 1.97474, "loss_cns_5": 0.64066, "loss_yns_5": 0.1593, "loss_cls_dn_0": 0.26395, "loss_box_dn_0": 0.8385, "loss_cls_dn_1": 0.19963, "loss_box_dn_1": 0.83952, "loss_cls_dn_2": 0.21378, "loss_box_dn_2": 0.80578, "loss_cls_dn_3": 0.21671, "loss_box_dn_3": 0.80633, "loss_cls_dn_4": 0.21979, "loss_box_dn_4": 0.82664, "loss_cls_dn_5": 0.23427, "loss_box_dn_5": 0.83723, "loss_dense_depth": 0.85096, "loss": 30.01408, "grad_norm": 46.15497, "time": 1.5412} -{"mode": "train", "epoch": 1, "iter": 84, "lr": 0.00013, "memory": 49167, "data_time": 0.08022, "loss_cls_0": 0.90848, "loss_box_0": 1.81069, "loss_cns_0": 0.62501, "loss_yns_0": 0.15955, "loss_cls_1": 0.97386, "loss_box_1": 2.01443, "loss_cns_1": 0.6238, "loss_yns_1": 0.16001, "loss_cls_2": 1.06458, "loss_box_2": 1.91877, "loss_cns_2": 0.64846, "loss_yns_2": 0.16049, "loss_cls_3": 1.0581, "loss_box_3": 1.90618, "loss_cns_3": 0.65143, "loss_yns_3": 0.16038, "loss_cls_4": 1.10086, "loss_box_4": 1.93132, "loss_cns_4": 0.65258, "loss_yns_4": 0.16046, "loss_cls_5": 1.04278, "loss_box_5": 1.95012, "loss_cns_5": 0.65391, "loss_yns_5": 0.15969, "loss_cls_dn_0": 0.2527, "loss_box_dn_0": 0.828, "loss_cls_dn_1": 0.20294, "loss_box_dn_1": 0.82481, "loss_cls_dn_2": 0.21684, "loss_box_dn_2": 0.78933, "loss_cls_dn_3": 0.21591, "loss_box_dn_3": 0.79437, "loss_cls_dn_4": 0.22283, "loss_box_dn_4": 0.81009, "loss_cls_dn_5": 0.23729, "loss_box_dn_5": 0.8248, "loss_dense_depth": 0.80219, "loss": 29.518, "grad_norm": 51.03897, "time": 1.52747} -{"mode": "train", "epoch": 1, "iter": 85, "lr": 0.00013, "memory": 49167, "data_time": 0.07943, "loss_cls_0": 0.95806, "loss_box_0": 1.80875, "loss_cns_0": 0.61629, "loss_yns_0": 0.15957, "loss_cls_1": 1.00007, "loss_box_1": 1.9658, "loss_cns_1": 0.61662, "loss_yns_1": 0.15708, "loss_cls_2": 1.04299, "loss_box_2": 1.90778, "loss_cns_2": 0.64977, "loss_yns_2": 0.16035, "loss_cls_3": 1.0662, "loss_box_3": 1.91292, "loss_cns_3": 0.65133, "loss_yns_3": 0.15847, "loss_cls_4": 1.06941, "loss_box_4": 1.9435, "loss_cns_4": 0.65222, "loss_yns_4": 0.15839, "loss_cls_5": 1.08827, "loss_box_5": 1.95593, "loss_cns_5": 0.65204, "loss_yns_5": 0.15956, "loss_cls_dn_0": 0.26303, "loss_box_dn_0": 0.8478, "loss_cls_dn_1": 0.20443, "loss_box_dn_1": 0.85753, "loss_cls_dn_2": 0.21803, "loss_box_dn_2": 0.81915, "loss_cls_dn_3": 0.21636, "loss_box_dn_3": 0.8241, "loss_cls_dn_4": 0.23029, "loss_box_dn_4": 0.83851, "loss_cls_dn_5": 0.25412, "loss_box_dn_5": 0.84694, "loss_dense_depth": 0.85072, "loss": 29.7824, "grad_norm": 45.9071, "time": 1.51882} -{"mode": "train", "epoch": 1, "iter": 86, "lr": 0.00013, "memory": 49167, "data_time": 0.11207, "loss_cls_0": 0.94088, "loss_box_0": 1.80864, "loss_cns_0": 0.6182, "loss_yns_0": 0.16141, "loss_cls_1": 1.00914, "loss_box_1": 1.88148, "loss_cns_1": 0.60484, "loss_yns_1": 0.15535, "loss_cls_2": 1.05316, "loss_box_2": 1.89735, "loss_cns_2": 0.64209, "loss_yns_2": 0.16455, "loss_cls_3": 1.05555, "loss_box_3": 1.86691, "loss_cns_3": 0.64867, "loss_yns_3": 0.16091, "loss_cls_4": 1.05589, "loss_box_4": 1.88245, "loss_cns_4": 0.65045, "loss_yns_4": 0.16245, "loss_cls_5": 1.09826, "loss_box_5": 1.86363, "loss_cns_5": 0.65009, "loss_yns_5": 0.16228, "loss_cls_dn_0": 0.25104, "loss_box_dn_0": 0.83511, "loss_cls_dn_1": 0.20105, "loss_box_dn_1": 0.82833, "loss_cls_dn_2": 0.21369, "loss_box_dn_2": 0.79414, "loss_cls_dn_3": 0.21394, "loss_box_dn_3": 0.78463, "loss_cls_dn_4": 0.23375, "loss_box_dn_4": 0.78866, "loss_cls_dn_5": 0.25682, "loss_box_dn_5": 0.79035, "loss_dense_depth": 0.84249, "loss": 29.22862, "grad_norm": 34.19355, "time": 1.54845} -{"mode": "train", "epoch": 1, "iter": 87, "lr": 0.00013, "memory": 49167, "data_time": 0.08104, "loss_cls_0": 0.92076, "loss_box_0": 1.80663, "loss_cns_0": 0.61913, "loss_yns_0": 0.15908, "loss_cls_1": 0.97267, "loss_box_1": 1.98455, "loss_cns_1": 0.60337, "loss_yns_1": 0.15505, "loss_cls_2": 1.03881, "loss_box_2": 2.04003, "loss_cns_2": 0.63251, "loss_yns_2": 0.16005, "loss_cls_3": 1.05608, "loss_box_3": 1.97506, "loss_cns_3": 0.64659, "loss_yns_3": 0.16113, "loss_cls_4": 1.07217, "loss_box_4": 1.94325, "loss_cns_4": 0.65058, "loss_yns_4": 0.16051, "loss_cls_5": 1.06875, "loss_box_5": 1.9425, "loss_cns_5": 0.65084, "loss_yns_5": 0.16152, "loss_cls_dn_0": 0.24725, "loss_box_dn_0": 0.82299, "loss_cls_dn_1": 0.19703, "loss_box_dn_1": 0.79839, "loss_cls_dn_2": 0.21132, "loss_box_dn_2": 0.78217, "loss_cls_dn_3": 0.21639, "loss_box_dn_3": 0.76449, "loss_cls_dn_4": 0.22936, "loss_box_dn_4": 0.75429, "loss_cls_dn_5": 0.24099, "loss_box_dn_5": 0.76893, "loss_dense_depth": 0.83661, "loss": 29.45185, "grad_norm": 42.93005, "time": 1.52119} -{"mode": "train", "epoch": 1, "iter": 88, "lr": 0.00013, "memory": 49167, "data_time": 0.07754, "loss_cls_0": 0.92798, "loss_box_0": 1.79171, "loss_cns_0": 0.61549, "loss_yns_0": 0.15724, "loss_cls_1": 0.96396, "loss_box_1": 1.96527, "loss_cns_1": 0.59506, "loss_yns_1": 0.15383, "loss_cls_2": 1.03059, "loss_box_2": 1.98525, "loss_cns_2": 0.6368, "loss_yns_2": 0.15926, "loss_cls_3": 1.06219, "loss_box_3": 1.94372, "loss_cns_3": 0.64691, "loss_yns_3": 0.1601, "loss_cls_4": 1.07195, "loss_box_4": 1.90681, "loss_cns_4": 0.65066, "loss_yns_4": 0.15963, "loss_cls_5": 1.13342, "loss_box_5": 1.91284, "loss_cns_5": 0.64977, "loss_yns_5": 0.15987, "loss_cls_dn_0": 0.25226, "loss_box_dn_0": 0.82474, "loss_cls_dn_1": 0.19549, "loss_box_dn_1": 0.79243, "loss_cls_dn_2": 0.21116, "loss_box_dn_2": 0.76676, "loss_cls_dn_3": 0.21575, "loss_box_dn_3": 0.75551, "loss_cls_dn_4": 0.22602, "loss_box_dn_4": 0.75977, "loss_cls_dn_5": 0.23663, "loss_box_dn_5": 0.76765, "loss_dense_depth": 0.8099, "loss": 29.25442, "grad_norm": 50.11963, "time": 1.51065} -{"mode": "train", "epoch": 1, "iter": 89, "lr": 0.00014, "memory": 49167, "data_time": 0.08853, "loss_cls_0": 0.952, "loss_box_0": 1.7834, "loss_cns_0": 0.613, "loss_yns_0": 0.15564, "loss_cls_1": 0.96436, "loss_box_1": 1.96187, "loss_cns_1": 0.5926, "loss_yns_1": 0.15462, "loss_cls_2": 1.01529, "loss_box_2": 1.91447, "loss_cns_2": 0.63914, "loss_yns_2": 0.16141, "loss_cls_3": 1.05024, "loss_box_3": 1.8865, "loss_cns_3": 0.64819, "loss_yns_3": 0.15979, "loss_cls_4": 1.06019, "loss_box_4": 1.87295, "loss_cns_4": 0.64914, "loss_yns_4": 0.15938, "loss_cls_5": 1.06895, "loss_box_5": 1.85566, "loss_cns_5": 0.64915, "loss_yns_5": 0.15914, "loss_cls_dn_0": 0.25657, "loss_box_dn_0": 0.8235, "loss_cls_dn_1": 0.19835, "loss_box_dn_1": 0.7962, "loss_cls_dn_2": 0.20944, "loss_box_dn_2": 0.75418, "loss_cls_dn_3": 0.21327, "loss_box_dn_3": 0.74968, "loss_cls_dn_4": 0.23029, "loss_box_dn_4": 0.76674, "loss_cls_dn_5": 0.23047, "loss_box_dn_5": 0.76472, "loss_dense_depth": 0.8165, "loss": 28.937, "grad_norm": 42.59156, "time": 1.53257} -{"mode": "train", "epoch": 1, "iter": 90, "lr": 0.00014, "memory": 49167, "data_time": 0.07738, "loss_cls_0": 0.94865, "loss_box_0": 1.76769, "loss_cns_0": 0.60695, "loss_yns_0": 0.15458, "loss_cls_1": 0.99733, "loss_box_1": 2.04669, "loss_cns_1": 0.61391, "loss_yns_1": 0.15875, "loss_cls_2": 1.02196, "loss_box_2": 1.94893, "loss_cns_2": 0.64399, "loss_yns_2": 0.16019, "loss_cls_3": 1.05724, "loss_box_3": 1.92148, "loss_cns_3": 0.6517, "loss_yns_3": 0.15723, "loss_cls_4": 1.06655, "loss_box_4": 1.92342, "loss_cns_4": 0.65226, "loss_yns_4": 0.15744, "loss_cls_5": 1.06428, "loss_box_5": 1.92453, "loss_cns_5": 0.65188, "loss_yns_5": 0.15687, "loss_cls_dn_0": 0.25657, "loss_box_dn_0": 0.83597, "loss_cls_dn_1": 0.19504, "loss_box_dn_1": 0.79838, "loss_cls_dn_2": 0.20265, "loss_box_dn_2": 0.76618, "loss_cls_dn_3": 0.20722, "loss_box_dn_3": 0.76273, "loss_cls_dn_4": 0.23523, "loss_box_dn_4": 0.77853, "loss_cls_dn_5": 0.23583, "loss_box_dn_5": 0.7803, "loss_dense_depth": 0.87562, "loss": 29.38474, "grad_norm": 37.81572, "time": 1.54442} -{"mode": "train", "epoch": 1, "iter": 91, "lr": 0.00014, "memory": 49167, "data_time": 0.07703, "loss_cls_0": 0.93202, "loss_box_0": 1.7606, "loss_cns_0": 0.61449, "loss_yns_0": 0.15347, "loss_cls_1": 0.9722, "loss_box_1": 2.07531, "loss_cns_1": 0.61298, "loss_yns_1": 0.15564, "loss_cls_2": 1.03546, "loss_box_2": 1.95041, "loss_cns_2": 0.64668, "loss_yns_2": 0.15648, "loss_cls_3": 1.04306, "loss_box_3": 1.92957, "loss_cns_3": 0.65363, "loss_yns_3": 0.15604, "loss_cls_4": 1.05536, "loss_box_4": 1.92246, "loss_cns_4": 0.65239, "loss_yns_4": 0.15686, "loss_cls_5": 1.07245, "loss_box_5": 1.96846, "loss_cns_5": 0.64965, "loss_yns_5": 0.15527, "loss_cls_dn_0": 0.24569, "loss_box_dn_0": 0.81159, "loss_cls_dn_1": 0.19255, "loss_box_dn_1": 0.81091, "loss_cls_dn_2": 0.19997, "loss_box_dn_2": 0.77695, "loss_cls_dn_3": 0.19965, "loss_box_dn_3": 0.77683, "loss_cls_dn_4": 0.23091, "loss_box_dn_4": 0.78161, "loss_cls_dn_5": 0.24616, "loss_box_dn_5": 0.80084, "loss_dense_depth": 0.82752, "loss": 29.38212, "grad_norm": 42.95937, "time": 1.51079} -{"mode": "train", "epoch": 1, "iter": 92, "lr": 0.00014, "memory": 49167, "data_time": 0.07699, "loss_cls_0": 0.94339, "loss_box_0": 1.74354, "loss_cns_0": 0.6083, "loss_yns_0": 0.14972, "loss_cls_1": 0.95165, "loss_box_1": 2.04525, "loss_cns_1": 0.6147, "loss_yns_1": 0.15501, "loss_cls_2": 1.03206, "loss_box_2": 1.93863, "loss_cns_2": 0.64303, "loss_yns_2": 0.15574, "loss_cls_3": 1.04126, "loss_box_3": 1.90642, "loss_cns_3": 0.65065, "loss_yns_3": 0.15634, "loss_cls_4": 1.05301, "loss_box_4": 1.8978, "loss_cns_4": 0.65349, "loss_yns_4": 0.15629, "loss_cls_5": 1.07009, "loss_box_5": 1.89753, "loss_cns_5": 0.6508, "loss_yns_5": 0.15493, "loss_cls_dn_0": 0.24699, "loss_box_dn_0": 0.81055, "loss_cls_dn_1": 0.1925, "loss_box_dn_1": 0.80808, "loss_cls_dn_2": 0.19788, "loss_box_dn_2": 0.76861, "loss_cls_dn_3": 0.19715, "loss_box_dn_3": 0.76175, "loss_cls_dn_4": 0.22417, "loss_box_dn_4": 0.76003, "loss_cls_dn_5": 0.24976, "loss_box_dn_5": 0.76639, "loss_dense_depth": 0.82261, "loss": 29.0761, "grad_norm": 43.14536, "time": 1.50907} -{"mode": "train", "epoch": 1, "iter": 93, "lr": 0.00014, "memory": 49167, "data_time": 0.07888, "loss_cls_0": 0.94232, "loss_box_0": 1.76213, "loss_cns_0": 0.61379, "loss_yns_0": 0.15041, "loss_cls_1": 0.9896, "loss_box_1": 1.97315, "loss_cns_1": 0.62872, "loss_yns_1": 0.15318, "loss_cls_2": 1.03137, "loss_box_2": 1.90666, "loss_cns_2": 0.64454, "loss_yns_2": 0.15512, "loss_cls_3": 1.04918, "loss_box_3": 1.86217, "loss_cns_3": 0.65479, "loss_yns_3": 0.15334, "loss_cls_4": 1.05387, "loss_box_4": 1.85547, "loss_cns_4": 0.65742, "loss_yns_4": 0.15548, "loss_cls_5": 1.06915, "loss_box_5": 1.8492, "loss_cns_5": 0.65671, "loss_yns_5": 0.15633, "loss_cls_dn_0": 0.24734, "loss_box_dn_0": 0.81239, "loss_cls_dn_1": 0.19476, "loss_box_dn_1": 0.78559, "loss_cls_dn_2": 0.19746, "loss_box_dn_2": 0.75267, "loss_cls_dn_3": 0.19905, "loss_box_dn_3": 0.74309, "loss_cls_dn_4": 0.21522, "loss_box_dn_4": 0.74464, "loss_cls_dn_5": 0.24038, "loss_box_dn_5": 0.75073, "loss_dense_depth": 0.82493, "loss": 28.83236, "grad_norm": 38.30383, "time": 1.52326} -{"mode": "train", "epoch": 1, "iter": 94, "lr": 0.00014, "memory": 49167, "data_time": 0.07795, "loss_cls_0": 0.9398, "loss_box_0": 1.80061, "loss_cns_0": 0.61745, "loss_yns_0": 0.15395, "loss_cls_1": 1.00239, "loss_box_1": 1.9943, "loss_cns_1": 0.6242, "loss_yns_1": 0.15575, "loss_cls_2": 1.0478, "loss_box_2": 1.94342, "loss_cns_2": 0.64319, "loss_yns_2": 0.15727, "loss_cls_3": 1.05266, "loss_box_3": 1.91165, "loss_cns_3": 0.65317, "loss_yns_3": 0.15599, "loss_cls_4": 1.06811, "loss_box_4": 1.90402, "loss_cns_4": 0.65433, "loss_yns_4": 0.1614, "loss_cls_5": 1.06964, "loss_box_5": 1.92623, "loss_cns_5": 0.65238, "loss_yns_5": 0.15907, "loss_cls_dn_0": 0.24447, "loss_box_dn_0": 0.80815, "loss_cls_dn_1": 0.20012, "loss_box_dn_1": 0.78501, "loss_cls_dn_2": 0.20541, "loss_box_dn_2": 0.75862, "loss_cls_dn_3": 0.20399, "loss_box_dn_3": 0.75685, "loss_cls_dn_4": 0.21315, "loss_box_dn_4": 0.76984, "loss_cls_dn_5": 0.22473, "loss_box_dn_5": 0.79326, "loss_dense_depth": 0.80599, "loss": 29.2184, "grad_norm": 39.17413, "time": 1.52198} -{"mode": "train", "epoch": 1, "iter": 95, "lr": 0.00014, "memory": 49167, "data_time": 0.07741, "loss_cls_0": 0.93505, "loss_box_0": 1.82572, "loss_cns_0": 0.61541, "loss_yns_0": 0.15413, "loss_cls_1": 0.9851, "loss_box_1": 2.01336, "loss_cns_1": 0.63, "loss_yns_1": 0.15772, "loss_cls_2": 1.05298, "loss_box_2": 1.93859, "loss_cns_2": 0.6507, "loss_yns_2": 0.1567, "loss_cls_3": 1.08659, "loss_box_3": 1.93925, "loss_cns_3": 0.65747, "loss_yns_3": 0.15718, "loss_cls_4": 1.1451, "loss_box_4": 1.95569, "loss_cns_4": 0.65471, "loss_yns_4": 0.16148, "loss_cls_5": 1.08217, "loss_box_5": 1.95631, "loss_cns_5": 0.65314, "loss_yns_5": 0.16012, "loss_cls_dn_0": 0.24012, "loss_box_dn_0": 0.81062, "loss_cls_dn_1": 0.19279, "loss_box_dn_1": 0.78412, "loss_cls_dn_2": 0.20548, "loss_box_dn_2": 0.76874, "loss_cls_dn_3": 0.20244, "loss_box_dn_3": 0.78376, "loss_cls_dn_4": 0.2134, "loss_box_dn_4": 0.80383, "loss_cls_dn_5": 0.21804, "loss_box_dn_5": 0.81798, "loss_dense_depth": 0.82477, "loss": 29.59075, "grad_norm": 56.70111, "time": 1.51244} -{"mode": "train", "epoch": 1, "iter": 96, "lr": 0.00014, "memory": 49167, "data_time": 0.07917, "loss_cls_0": 0.96199, "loss_box_0": 1.86921, "loss_cns_0": 0.61332, "loss_yns_0": 0.15826, "loss_cls_1": 1.00848, "loss_box_1": 1.9956, "loss_cns_1": 0.62796, "loss_yns_1": 0.1571, "loss_cls_2": 1.07677, "loss_box_2": 1.94623, "loss_cns_2": 0.64629, "loss_yns_2": 0.1547, "loss_cls_3": 1.05085, "loss_box_3": 1.94362, "loss_cns_3": 0.6511, "loss_yns_3": 0.15588, "loss_cls_4": 1.06617, "loss_box_4": 1.96977, "loss_cns_4": 0.65078, "loss_yns_4": 0.15842, "loss_cls_5": 1.07606, "loss_box_5": 1.93804, "loss_cns_5": 0.65004, "loss_yns_5": 0.16187, "loss_cls_dn_0": 0.24711, "loss_box_dn_0": 0.82255, "loss_cls_dn_1": 0.19037, "loss_box_dn_1": 0.80318, "loss_cls_dn_2": 0.20971, "loss_box_dn_2": 0.80081, "loss_cls_dn_3": 0.20763, "loss_box_dn_3": 0.81401, "loss_cls_dn_4": 0.21547, "loss_box_dn_4": 0.82695, "loss_cls_dn_5": 0.22063, "loss_box_dn_5": 0.82387, "loss_dense_depth": 0.84279, "loss": 29.71361, "grad_norm": 53.4177, "time": 1.5209} -{"mode": "train", "epoch": 1, "iter": 97, "lr": 0.00014, "memory": 49167, "data_time": 0.07899, "loss_cls_0": 0.95892, "loss_box_0": 1.81748, "loss_cns_0": 0.6226, "loss_yns_0": 0.1561, "loss_cls_1": 1.0181, "loss_box_1": 2.0158, "loss_cns_1": 0.63077, "loss_yns_1": 0.1571, "loss_cls_2": 1.12642, "loss_box_2": 1.9424, "loss_cns_2": 0.64621, "loss_yns_2": 0.15421, "loss_cls_3": 1.06978, "loss_box_3": 1.93605, "loss_cns_3": 0.65465, "loss_yns_3": 0.15664, "loss_cls_4": 1.07335, "loss_box_4": 1.96094, "loss_cns_4": 0.65665, "loss_yns_4": 0.15728, "loss_cls_5": 1.06765, "loss_box_5": 1.94429, "loss_cns_5": 0.65195, "loss_yns_5": 0.15748, "loss_cls_dn_0": 0.25016, "loss_box_dn_0": 0.81096, "loss_cls_dn_1": 0.18984, "loss_box_dn_1": 0.79911, "loss_cls_dn_2": 0.21267, "loss_box_dn_2": 0.78665, "loss_cls_dn_3": 0.2059, "loss_box_dn_3": 0.79004, "loss_cls_dn_4": 0.2116, "loss_box_dn_4": 0.79512, "loss_cls_dn_5": 0.21647, "loss_box_dn_5": 0.79504, "loss_dense_depth": 0.81528, "loss": 29.61169, "grad_norm": 50.84659, "time": 1.52097} -{"mode": "train", "epoch": 1, "iter": 98, "lr": 0.00014, "memory": 49167, "data_time": 0.07686, "loss_cls_0": 0.93582, "loss_box_0": 1.78873, "loss_cns_0": 0.62542, "loss_yns_0": 0.15578, "loss_cls_1": 0.99762, "loss_box_1": 1.9807, "loss_cns_1": 0.62848, "loss_yns_1": 0.15674, "loss_cls_2": 1.09433, "loss_box_2": 1.87448, "loss_cns_2": 0.64403, "loss_yns_2": 0.15526, "loss_cls_3": 1.06326, "loss_box_3": 1.88134, "loss_cns_3": 0.65838, "loss_yns_3": 0.15768, "loss_cls_4": 1.0799, "loss_box_4": 1.88286, "loss_cns_4": 0.65593, "loss_yns_4": 0.15766, "loss_cls_5": 1.07274, "loss_box_5": 1.89795, "loss_cns_5": 0.64835, "loss_yns_5": 0.15651, "loss_cls_dn_0": 0.24778, "loss_box_dn_0": 0.80871, "loss_cls_dn_1": 0.1883, "loss_box_dn_1": 0.80555, "loss_cls_dn_2": 0.20886, "loss_box_dn_2": 0.77307, "loss_cls_dn_3": 0.19774, "loss_box_dn_3": 0.77849, "loss_cls_dn_4": 0.21127, "loss_box_dn_4": 0.77577, "loss_cls_dn_5": 0.21615, "loss_box_dn_5": 0.78868, "loss_dense_depth": 0.82613, "loss": 29.17646, "grad_norm": 49.05758, "time": 1.50263} -{"mode": "train", "epoch": 1, "iter": 99, "lr": 0.00014, "memory": 49167, "data_time": 0.07801, "loss_cls_0": 0.93551, "loss_box_0": 1.7785, "loss_cns_0": 0.62204, "loss_yns_0": 0.15606, "loss_cls_1": 1.0109, "loss_box_1": 1.98174, "loss_cns_1": 0.62827, "loss_yns_1": 0.15864, "loss_cls_2": 1.05726, "loss_box_2": 1.90489, "loss_cns_2": 0.64462, "loss_yns_2": 0.15821, "loss_cls_3": 1.06433, "loss_box_3": 1.90105, "loss_cns_3": 0.65487, "loss_yns_3": 0.1579, "loss_cls_4": 1.07507, "loss_box_4": 1.88763, "loss_cns_4": 0.65353, "loss_yns_4": 0.16099, "loss_cls_5": 1.07523, "loss_box_5": 1.91449, "loss_cns_5": 0.65004, "loss_yns_5": 0.16321, "loss_cls_dn_0": 0.24899, "loss_box_dn_0": 0.80573, "loss_cls_dn_1": 0.18446, "loss_box_dn_1": 0.79294, "loss_cls_dn_2": 0.19545, "loss_box_dn_2": 0.76848, "loss_cls_dn_3": 0.19569, "loss_box_dn_3": 0.77392, "loss_cls_dn_4": 0.21535, "loss_box_dn_4": 0.77006, "loss_cls_dn_5": 0.22374, "loss_box_dn_5": 0.78828, "loss_dense_depth": 0.82954, "loss": 29.1876, "grad_norm": 42.76099, "time": 1.52392} -{"mode": "train", "epoch": 1, "iter": 100, "lr": 0.00014, "memory": 49167, "data_time": 0.08652, "loss_cls_0": 0.94252, "loss_box_0": 1.79678, "loss_cns_0": 0.6212, "loss_yns_0": 0.1578, "loss_cls_1": 1.0225, "loss_box_1": 2.04235, "loss_cns_1": 0.62578, "loss_yns_1": 0.15966, "loss_cls_2": 1.03669, "loss_box_2": 2.00871, "loss_cns_2": 0.63613, "loss_yns_2": 0.15864, "loss_cls_3": 1.06091, "loss_box_3": 1.97901, "loss_cns_3": 0.64183, "loss_yns_3": 0.15956, "loss_cls_4": 1.08482, "loss_box_4": 1.97166, "loss_cns_4": 0.64373, "loss_yns_4": 0.16008, "loss_cls_5": 1.07765, "loss_box_5": 1.98434, "loss_cns_5": 0.64177, "loss_yns_5": 0.16354, "loss_cls_dn_0": 0.24769, "loss_box_dn_0": 0.81581, "loss_cls_dn_1": 0.18722, "loss_box_dn_1": 0.79532, "loss_cls_dn_2": 0.19711, "loss_box_dn_2": 0.77769, "loss_cls_dn_3": 0.20266, "loss_box_dn_3": 0.77781, "loss_cls_dn_4": 0.22606, "loss_box_dn_4": 0.7773, "loss_cls_dn_5": 0.23392, "loss_box_dn_5": 0.79108, "loss_dense_depth": 0.80217, "loss": 29.6095, "grad_norm": 48.38213, "time": 1.52595} -{"mode": "train", "epoch": 1, "iter": 101, "lr": 0.00014, "memory": 49167, "data_time": 0.11944, "loss_cls_0": 0.94245, "loss_box_0": 1.78688, "loss_cns_0": 0.62214, "loss_yns_0": 0.15965, "loss_cls_1": 1.00312, "loss_box_1": 2.09055, "loss_cns_1": 0.62342, "loss_yns_1": 0.16066, "loss_cls_2": 1.03764, "loss_box_2": 2.04174, "loss_cns_2": 0.63797, "loss_yns_2": 0.1598, "loss_cls_3": 1.05216, "loss_box_3": 2.01281, "loss_cns_3": 0.6425, "loss_yns_3": 0.16034, "loss_cls_4": 1.08521, "loss_box_4": 1.99516, "loss_cns_4": 0.64295, "loss_yns_4": 0.16035, "loss_cls_5": 1.06645, "loss_box_5": 2.00764, "loss_cns_5": 0.64323, "loss_yns_5": 0.1596, "loss_cls_dn_0": 0.2462, "loss_box_dn_0": 0.81152, "loss_cls_dn_1": 0.19542, "loss_box_dn_1": 0.82674, "loss_cls_dn_2": 0.20764, "loss_box_dn_2": 0.794, "loss_cls_dn_3": 0.20689, "loss_box_dn_3": 0.79594, "loss_cls_dn_4": 0.22612, "loss_box_dn_4": 0.79624, "loss_cls_dn_5": 0.23376, "loss_box_dn_5": 0.80801, "loss_dense_depth": 0.81897, "loss": 29.86185, "grad_norm": 50.17352, "time": 1.60327} -{"mode": "train", "epoch": 1, "iter": 102, "lr": 0.00014, "memory": 49167, "data_time": 0.08067, "loss_cls_0": 0.92777, "loss_box_0": 1.81859, "loss_cns_0": 0.6149, "loss_yns_0": 0.16037, "loss_cls_1": 0.98611, "loss_box_1": 2.04215, "loss_cns_1": 0.62655, "loss_yns_1": 0.16302, "loss_cls_2": 1.03856, "loss_box_2": 1.97828, "loss_cns_2": 0.64012, "loss_yns_2": 0.16216, "loss_cls_3": 1.04694, "loss_box_3": 1.96316, "loss_cns_3": 0.64563, "loss_yns_3": 0.16301, "loss_cls_4": 1.07522, "loss_box_4": 1.95107, "loss_cns_4": 0.64804, "loss_yns_4": 0.16397, "loss_cls_5": 1.04737, "loss_box_5": 1.95348, "loss_cns_5": 0.64778, "loss_yns_5": 0.1612, "loss_cls_dn_0": 0.24667, "loss_box_dn_0": 0.82294, "loss_cls_dn_1": 0.19711, "loss_box_dn_1": 0.81316, "loss_cls_dn_2": 0.20958, "loss_box_dn_2": 0.78034, "loss_cls_dn_3": 0.20152, "loss_box_dn_3": 0.79371, "loss_cls_dn_4": 0.21949, "loss_box_dn_4": 0.80333, "loss_cls_dn_5": 0.22642, "loss_box_dn_5": 0.81189, "loss_dense_depth": 0.80857, "loss": 29.56022, "grad_norm": 40.49029, "time": 1.58878} -{"mode": "train", "epoch": 1, "iter": 103, "lr": 0.00014, "memory": 49167, "data_time": 0.09972, "loss_cls_0": 0.91589, "loss_box_0": 1.80872, "loss_cns_0": 0.62089, "loss_yns_0": 0.16074, "loss_cls_1": 0.9929, "loss_box_1": 1.96255, "loss_cns_1": 0.63523, "loss_yns_1": 0.16444, "loss_cls_2": 1.02617, "loss_box_2": 1.89484, "loss_cns_2": 0.64765, "loss_yns_2": 0.16376, "loss_cls_3": 1.05987, "loss_box_3": 1.88754, "loss_cns_3": 0.65252, "loss_yns_3": 0.16407, "loss_cls_4": 1.05649, "loss_box_4": 1.91927, "loss_cns_4": 0.65513, "loss_yns_4": 0.16707, "loss_cls_5": 1.04468, "loss_box_5": 1.91221, "loss_cns_5": 0.65097, "loss_yns_5": 0.16385, "loss_cls_dn_0": 0.24016, "loss_box_dn_0": 0.81004, "loss_cls_dn_1": 0.18629, "loss_box_dn_1": 0.80456, "loss_cls_dn_2": 0.19481, "loss_box_dn_2": 0.77735, "loss_cls_dn_3": 0.19507, "loss_box_dn_3": 0.78865, "loss_cls_dn_4": 0.20724, "loss_box_dn_4": 0.81411, "loss_cls_dn_5": 0.21193, "loss_box_dn_5": 0.81263, "loss_dense_depth": 0.80346, "loss": 29.17373, "grad_norm": 47.3091, "time": 1.53805} -{"mode": "train", "epoch": 1, "iter": 104, "lr": 0.00014, "memory": 49167, "data_time": 0.07424, "loss_cls_0": 0.91298, "loss_box_0": 1.81604, "loss_cns_0": 0.62129, "loss_yns_0": 0.15841, "loss_cls_1": 0.99678, "loss_box_1": 1.98068, "loss_cns_1": 0.63268, "loss_yns_1": 0.16079, "loss_cls_2": 1.02044, "loss_box_2": 1.92945, "loss_cns_2": 0.64674, "loss_yns_2": 0.16271, "loss_cls_3": 1.07574, "loss_box_3": 1.91156, "loss_cns_3": 0.65324, "loss_yns_3": 0.16253, "loss_cls_4": 1.05571, "loss_box_4": 1.95362, "loss_cns_4": 0.65467, "loss_yns_4": 0.16326, "loss_cls_5": 1.07375, "loss_box_5": 1.96217, "loss_cns_5": 0.65051, "loss_yns_5": 0.16143, "loss_cls_dn_0": 0.24273, "loss_box_dn_0": 0.81571, "loss_cls_dn_1": 0.18173, "loss_box_dn_1": 0.815, "loss_cls_dn_2": 0.18791, "loss_box_dn_2": 0.79556, "loss_cls_dn_3": 0.19518, "loss_box_dn_3": 0.79605, "loss_cls_dn_4": 0.19869, "loss_box_dn_4": 0.81981, "loss_cls_dn_5": 0.20312, "loss_box_dn_5": 0.81711, "loss_dense_depth": 0.79138, "loss": 29.37718, "grad_norm": 66.32126, "time": 1.51869} -{"mode": "train", "epoch": 1, "iter": 105, "lr": 0.00014, "memory": 49167, "data_time": 0.0808, "loss_cls_0": 0.90986, "loss_box_0": 1.77375, "loss_cns_0": 0.62639, "loss_yns_0": 0.15676, "loss_cls_1": 0.96121, "loss_box_1": 1.94021, "loss_cns_1": 0.63275, "loss_yns_1": 0.15799, "loss_cls_2": 1.01061, "loss_box_2": 1.8865, "loss_cns_2": 0.65014, "loss_yns_2": 0.16059, "loss_cls_3": 1.04369, "loss_box_3": 1.84916, "loss_cns_3": 0.65586, "loss_yns_3": 0.16033, "loss_cls_4": 1.0488, "loss_box_4": 1.86997, "loss_cns_4": 0.65719, "loss_yns_4": 0.16071, "loss_cls_5": 1.05507, "loss_box_5": 1.87603, "loss_cns_5": 0.65376, "loss_yns_5": 0.15879, "loss_cls_dn_0": 0.23743, "loss_box_dn_0": 0.80972, "loss_cls_dn_1": 0.18015, "loss_box_dn_1": 0.82889, "loss_cls_dn_2": 0.18826, "loss_box_dn_2": 0.81201, "loss_cls_dn_3": 0.19453, "loss_box_dn_3": 0.79846, "loss_cls_dn_4": 0.19984, "loss_box_dn_4": 0.80437, "loss_cls_dn_5": 0.20074, "loss_box_dn_5": 0.8047, "loss_dense_depth": 0.77649, "loss": 28.89172, "grad_norm": 45.33723, "time": 1.54252} -{"mode": "train", "epoch": 1, "iter": 106, "lr": 0.00014, "memory": 49167, "data_time": 0.11228, "loss_cls_0": 0.89798, "loss_box_0": 1.7475, "loss_cns_0": 0.61893, "loss_yns_0": 0.1554, "loss_cls_1": 0.95324, "loss_box_1": 1.93035, "loss_cns_1": 0.63493, "loss_yns_1": 0.16197, "loss_cls_2": 1.01125, "loss_box_2": 1.8711, "loss_cns_2": 0.65031, "loss_yns_2": 0.16468, "loss_cls_3": 1.02549, "loss_box_3": 1.83871, "loss_cns_3": 0.65497, "loss_yns_3": 0.16138, "loss_cls_4": 1.06087, "loss_box_4": 1.84758, "loss_cns_4": 0.65744, "loss_yns_4": 0.16183, "loss_cls_5": 1.03063, "loss_box_5": 1.84082, "loss_cns_5": 0.65306, "loss_yns_5": 0.15955, "loss_cls_dn_0": 0.23663, "loss_box_dn_0": 0.79291, "loss_cls_dn_1": 0.17787, "loss_box_dn_1": 0.78208, "loss_cls_dn_2": 0.18804, "loss_box_dn_2": 0.77137, "loss_cls_dn_3": 0.18648, "loss_box_dn_3": 0.75603, "loss_cls_dn_4": 0.19705, "loss_box_dn_4": 0.75359, "loss_cls_dn_5": 0.20245, "loss_box_dn_5": 0.75967, "loss_dense_depth": 0.78175, "loss": 28.47589, "grad_norm": 37.63884, "time": 1.55467} -{"mode": "train", "epoch": 1, "iter": 107, "lr": 0.00014, "memory": 49167, "data_time": 0.07985, "loss_cls_0": 0.90377, "loss_box_0": 1.7091, "loss_cns_0": 0.60785, "loss_yns_0": 0.15323, "loss_cls_1": 0.96403, "loss_box_1": 1.92902, "loss_cns_1": 0.6349, "loss_yns_1": 0.15981, "loss_cls_2": 1.01347, "loss_box_2": 1.86162, "loss_cns_2": 0.65067, "loss_yns_2": 0.16093, "loss_cls_3": 1.04585, "loss_box_3": 1.84007, "loss_cns_3": 0.65668, "loss_yns_3": 0.15919, "loss_cls_4": 1.0714, "loss_box_4": 1.85506, "loss_cns_4": 0.65843, "loss_yns_4": 0.16181, "loss_cls_5": 1.02989, "loss_box_5": 1.83742, "loss_cns_5": 0.65425, "loss_yns_5": 0.15775, "loss_cls_dn_0": 0.23954, "loss_box_dn_0": 0.81407, "loss_cls_dn_1": 0.1818, "loss_box_dn_1": 0.76697, "loss_cls_dn_2": 0.18974, "loss_box_dn_2": 0.75547, "loss_cls_dn_3": 0.19245, "loss_box_dn_3": 0.75703, "loss_cls_dn_4": 0.20347, "loss_box_dn_4": 0.76548, "loss_cls_dn_5": 0.21832, "loss_box_dn_5": 0.78071, "loss_dense_depth": 0.79098, "loss": 28.53222, "grad_norm": 50.52134, "time": 1.52385} -{"mode": "train", "epoch": 1, "iter": 108, "lr": 0.00014, "memory": 49167, "data_time": 0.0779, "loss_cls_0": 0.91017, "loss_box_0": 1.72001, "loss_cns_0": 0.60795, "loss_yns_0": 0.15352, "loss_cls_1": 0.96069, "loss_box_1": 1.95877, "loss_cns_1": 0.62617, "loss_yns_1": 0.16443, "loss_cls_2": 1.01342, "loss_box_2": 1.90249, "loss_cns_2": 0.64772, "loss_yns_2": 0.16362, "loss_cls_3": 1.06104, "loss_box_3": 1.89383, "loss_cns_3": 0.65498, "loss_yns_3": 0.16458, "loss_cls_4": 1.07473, "loss_box_4": 1.91232, "loss_cns_4": 0.65443, "loss_yns_4": 0.16275, "loss_cls_5": 1.03456, "loss_box_5": 1.92179, "loss_cns_5": 0.65058, "loss_yns_5": 0.15861, "loss_cls_dn_0": 0.2433, "loss_box_dn_0": 0.81264, "loss_cls_dn_1": 0.18974, "loss_box_dn_1": 0.77607, "loss_cls_dn_2": 0.19384, "loss_box_dn_2": 0.7622, "loss_cls_dn_3": 0.20389, "loss_box_dn_3": 0.77284, "loss_cls_dn_4": 0.21022, "loss_box_dn_4": 0.79353, "loss_cls_dn_5": 0.2238, "loss_box_dn_5": 0.81944, "loss_dense_depth": 0.76773, "loss": 28.94237, "grad_norm": 57.42239, "time": 1.51531} -{"mode": "train", "epoch": 1, "iter": 109, "lr": 0.00014, "memory": 49167, "data_time": 0.08695, "loss_cls_0": 0.92995, "loss_box_0": 1.75061, "loss_cns_0": 0.60153, "loss_yns_0": 0.15455, "loss_cls_1": 0.98098, "loss_box_1": 2.02789, "loss_cns_1": 0.61432, "loss_yns_1": 0.16039, "loss_cls_2": 1.01394, "loss_box_2": 1.98938, "loss_cns_2": 0.64041, "loss_yns_2": 0.16003, "loss_cls_3": 1.06757, "loss_box_3": 1.95156, "loss_cns_3": 0.64513, "loss_yns_3": 0.16281, "loss_cls_4": 1.0507, "loss_box_4": 1.95966, "loss_cns_4": 0.64655, "loss_yns_4": 0.1605, "loss_cls_5": 1.04335, "loss_box_5": 1.98378, "loss_cns_5": 0.64473, "loss_yns_5": 0.15997, "loss_cls_dn_0": 0.24372, "loss_box_dn_0": 0.81486, "loss_cls_dn_1": 0.19001, "loss_box_dn_1": 0.80852, "loss_cls_dn_2": 0.19324, "loss_box_dn_2": 0.78857, "loss_cls_dn_3": 0.20058, "loss_box_dn_3": 0.79071, "loss_cls_dn_4": 0.20277, "loss_box_dn_4": 0.80739, "loss_cls_dn_5": 0.21439, "loss_box_dn_5": 0.83145, "loss_dense_depth": 0.79308, "loss": 29.37959, "grad_norm": 52.52246, "time": 1.52589} -{"mode": "train", "epoch": 1, "iter": 110, "lr": 0.00014, "memory": 49167, "data_time": 0.07637, "loss_cls_0": 0.92736, "loss_box_0": 1.78489, "loss_cns_0": 0.60801, "loss_yns_0": 0.15611, "loss_cls_1": 0.97549, "loss_box_1": 2.06152, "loss_cns_1": 0.61505, "loss_yns_1": 0.15885, "loss_cls_2": 1.01754, "loss_box_2": 1.99575, "loss_cns_2": 0.63882, "loss_yns_2": 0.16287, "loss_cls_3": 1.0631, "loss_box_3": 1.95155, "loss_cns_3": 0.63946, "loss_yns_3": 0.1592, "loss_cls_4": 1.04424, "loss_box_4": 1.948, "loss_cns_4": 0.64331, "loss_yns_4": 0.1609, "loss_cls_5": 1.04139, "loss_box_5": 1.96425, "loss_cns_5": 0.64122, "loss_yns_5": 0.15874, "loss_cls_dn_0": 0.24398, "loss_box_dn_0": 0.80452, "loss_cls_dn_1": 0.18528, "loss_box_dn_1": 0.79201, "loss_cls_dn_2": 0.19047, "loss_box_dn_2": 0.77234, "loss_cls_dn_3": 0.18651, "loss_box_dn_3": 0.76777, "loss_cls_dn_4": 0.19817, "loss_box_dn_4": 0.77113, "loss_cls_dn_5": 0.21117, "loss_box_dn_5": 0.78154, "loss_dense_depth": 0.79396, "loss": 29.21648, "grad_norm": 33.1229, "time": 1.54287} -{"mode": "train", "epoch": 1, "iter": 111, "lr": 0.00014, "memory": 49167, "data_time": 0.0766, "loss_cls_0": 0.91434, "loss_box_0": 1.80648, "loss_cns_0": 0.6132, "loss_yns_0": 0.15869, "loss_cls_1": 0.96788, "loss_box_1": 1.99755, "loss_cns_1": 0.62219, "loss_yns_1": 0.1601, "loss_cls_2": 1.02725, "loss_box_2": 1.92541, "loss_cns_2": 0.64296, "loss_yns_2": 0.16578, "loss_cls_3": 1.04945, "loss_box_3": 1.91635, "loss_cns_3": 0.64583, "loss_yns_3": 0.15992, "loss_cls_4": 1.06025, "loss_box_4": 1.9284, "loss_cns_4": 0.64633, "loss_yns_4": 0.15901, "loss_cls_5": 1.02612, "loss_box_5": 1.91858, "loss_cns_5": 0.64394, "loss_yns_5": 0.15858, "loss_cls_dn_0": 0.24329, "loss_box_dn_0": 0.80629, "loss_cls_dn_1": 0.17878, "loss_box_dn_1": 0.77268, "loss_cls_dn_2": 0.19403, "loss_box_dn_2": 0.7553, "loss_cls_dn_3": 0.18882, "loss_box_dn_3": 0.75794, "loss_cls_dn_4": 0.20847, "loss_box_dn_4": 0.76398, "loss_cls_dn_5": 0.21738, "loss_box_dn_5": 0.7589, "loss_dense_depth": 0.84657, "loss": 29.00704, "grad_norm": 47.80338, "time": 1.51723} -{"mode": "train", "epoch": 1, "iter": 112, "lr": 0.00014, "memory": 49167, "data_time": 0.07304, "loss_cls_0": 0.9396, "loss_box_0": 1.82558, "loss_cns_0": 0.60963, "loss_yns_0": 0.15903, "loss_cls_1": 0.99102, "loss_box_1": 1.96231, "loss_cns_1": 0.62211, "loss_yns_1": 0.15937, "loss_cls_2": 1.03231, "loss_box_2": 1.89745, "loss_cns_2": 0.64173, "loss_yns_2": 0.16442, "loss_cls_3": 1.04724, "loss_box_3": 1.8975, "loss_cns_3": 0.64579, "loss_yns_3": 0.16131, "loss_cls_4": 1.08931, "loss_box_4": 1.91361, "loss_cns_4": 0.64688, "loss_yns_4": 0.15822, "loss_cls_5": 1.0327, "loss_box_5": 1.90833, "loss_cns_5": 0.64332, "loss_yns_5": 0.15855, "loss_cls_dn_0": 0.25083, "loss_box_dn_0": 0.80461, "loss_cls_dn_1": 0.17895, "loss_box_dn_1": 0.76999, "loss_cls_dn_2": 0.19648, "loss_box_dn_2": 0.75779, "loss_cls_dn_3": 0.1957, "loss_box_dn_3": 0.76554, "loss_cls_dn_4": 0.21179, "loss_box_dn_4": 0.77504, "loss_cls_dn_5": 0.21764, "loss_box_dn_5": 0.77231, "loss_dense_depth": 0.84188, "loss": 29.04587, "grad_norm": 54.83796, "time": 1.51665} -{"mode": "train", "epoch": 1, "iter": 113, "lr": 0.00014, "memory": 49167, "data_time": 0.07412, "loss_cls_0": 0.8944, "loss_box_0": 1.81551, "loss_cns_0": 0.61794, "loss_yns_0": 0.1568, "loss_cls_1": 0.95122, "loss_box_1": 1.93348, "loss_cns_1": 0.62806, "loss_yns_1": 0.15672, "loss_cls_2": 0.98134, "loss_box_2": 1.86823, "loss_cns_2": 0.64134, "loss_yns_2": 0.1584, "loss_cls_3": 1.00115, "loss_box_3": 1.85466, "loss_cns_3": 0.64681, "loss_yns_3": 0.15741, "loss_cls_4": 1.05128, "loss_box_4": 1.85632, "loss_cns_4": 0.64825, "loss_yns_4": 0.1572, "loss_cls_5": 0.99429, "loss_box_5": 1.85474, "loss_cns_5": 0.64523, "loss_yns_5": 0.15589, "loss_cls_dn_0": 0.23694, "loss_box_dn_0": 0.80608, "loss_cls_dn_1": 0.17764, "loss_box_dn_1": 0.79351, "loss_cls_dn_2": 0.18778, "loss_box_dn_2": 0.776, "loss_cls_dn_3": 0.19106, "loss_box_dn_3": 0.77622, "loss_cls_dn_4": 0.1977, "loss_box_dn_4": 0.78178, "loss_cls_dn_5": 0.20313, "loss_box_dn_5": 0.78295, "loss_dense_depth": 0.80755, "loss": 28.54502, "grad_norm": 44.45742, "time": 1.53928} -{"mode": "train", "epoch": 1, "iter": 114, "lr": 0.00015, "memory": 49167, "data_time": 0.07568, "loss_cls_0": 0.90977, "loss_box_0": 1.82216, "loss_cns_0": 0.62092, "loss_yns_0": 0.15827, "loss_cls_1": 0.96025, "loss_box_1": 1.9538, "loss_cns_1": 0.63202, "loss_yns_1": 0.15722, "loss_cls_2": 0.9929, "loss_box_2": 1.86115, "loss_cns_2": 0.64572, "loss_yns_2": 0.15648, "loss_cls_3": 1.006, "loss_box_3": 1.84106, "loss_cns_3": 0.64968, "loss_yns_3": 0.15689, "loss_cls_4": 1.03382, "loss_box_4": 1.8316, "loss_cns_4": 0.65179, "loss_yns_4": 0.15847, "loss_cls_5": 1.00677, "loss_box_5": 1.84818, "loss_cns_5": 0.65265, "loss_yns_5": 0.15916, "loss_cls_dn_0": 0.24149, "loss_box_dn_0": 0.79907, "loss_cls_dn_1": 0.17336, "loss_box_dn_1": 0.78528, "loss_cls_dn_2": 0.18246, "loss_box_dn_2": 0.75912, "loss_cls_dn_3": 0.1834, "loss_box_dn_3": 0.75466, "loss_cls_dn_4": 0.19314, "loss_box_dn_4": 0.76081, "loss_cls_dn_5": 0.20123, "loss_box_dn_5": 0.77442, "loss_dense_depth": 0.82397, "loss": 28.49915, "grad_norm": 42.35509, "time": 1.52236} -{"mode": "train", "epoch": 1, "iter": 115, "lr": 0.00015, "memory": 49167, "data_time": 0.07367, "loss_cls_0": 0.88945, "loss_box_0": 1.78529, "loss_cns_0": 0.62254, "loss_yns_0": 0.15593, "loss_cls_1": 0.94245, "loss_box_1": 1.94213, "loss_cns_1": 0.6332, "loss_yns_1": 0.15472, "loss_cls_2": 0.98396, "loss_box_2": 1.85429, "loss_cns_2": 0.64866, "loss_yns_2": 0.16114, "loss_cls_3": 1.00392, "loss_box_3": 1.85482, "loss_cns_3": 0.64858, "loss_yns_3": 0.15927, "loss_cls_4": 1.00229, "loss_box_4": 1.84527, "loss_cns_4": 0.65177, "loss_yns_4": 0.15941, "loss_cls_5": 1.01571, "loss_box_5": 1.84683, "loss_cns_5": 0.65182, "loss_yns_5": 0.16309, "loss_cls_dn_0": 0.23277, "loss_box_dn_0": 0.80139, "loss_cls_dn_1": 0.17352, "loss_box_dn_1": 0.77669, "loss_cls_dn_2": 0.1817, "loss_box_dn_2": 0.75031, "loss_cls_dn_3": 0.1822, "loss_box_dn_3": 0.75339, "loss_cls_dn_4": 0.2006, "loss_box_dn_4": 0.76033, "loss_cls_dn_5": 0.20599, "loss_box_dn_5": 0.76759, "loss_dense_depth": 0.77689, "loss": 28.3399, "grad_norm": 39.90459, "time": 1.52689} -{"mode": "train", "epoch": 1, "iter": 116, "lr": 0.00015, "memory": 49167, "data_time": 0.07562, "loss_cls_0": 0.90256, "loss_box_0": 1.81038, "loss_cns_0": 0.6187, "loss_yns_0": 0.15584, "loss_cls_1": 0.96661, "loss_box_1": 1.9541, "loss_cns_1": 0.63127, "loss_yns_1": 0.15615, "loss_cls_2": 0.99924, "loss_box_2": 1.87943, "loss_cns_2": 0.64673, "loss_yns_2": 0.16384, "loss_cls_3": 1.02723, "loss_box_3": 1.87452, "loss_cns_3": 0.6476, "loss_yns_3": 0.15879, "loss_cls_4": 1.0185, "loss_box_4": 1.85889, "loss_cns_4": 0.6495, "loss_yns_4": 0.15745, "loss_cls_5": 1.0126, "loss_box_5": 1.86581, "loss_cns_5": 0.64821, "loss_yns_5": 0.15781, "loss_cls_dn_0": 0.23683, "loss_box_dn_0": 0.80299, "loss_cls_dn_1": 0.18069, "loss_box_dn_1": 0.77414, "loss_cls_dn_2": 0.18877, "loss_box_dn_2": 0.74985, "loss_cls_dn_3": 0.18726, "loss_box_dn_3": 0.75411, "loss_cls_dn_4": 0.20576, "loss_box_dn_4": 0.75733, "loss_cls_dn_5": 0.21002, "loss_box_dn_5": 0.76895, "loss_dense_depth": 0.8147, "loss": 28.59314, "grad_norm": 32.31699, "time": 1.52574} -{"mode": "train", "epoch": 1, "iter": 117, "lr": 0.00015, "memory": 49167, "data_time": 0.07644, "loss_cls_0": 0.8938, "loss_box_0": 1.78736, "loss_cns_0": 0.62138, "loss_yns_0": 0.15311, "loss_cls_1": 0.95554, "loss_box_1": 1.96787, "loss_cns_1": 0.63361, "loss_yns_1": 0.15607, "loss_cls_2": 0.98545, "loss_box_2": 1.89584, "loss_cns_2": 0.6482, "loss_yns_2": 0.16294, "loss_cls_3": 1.01861, "loss_box_3": 1.87438, "loss_cns_3": 0.65147, "loss_yns_3": 0.15663, "loss_cls_4": 1.01028, "loss_box_4": 1.88033, "loss_cns_4": 0.64998, "loss_yns_4": 0.1553, "loss_cls_5": 1.00285, "loss_box_5": 1.91656, "loss_cns_5": 0.64329, "loss_yns_5": 0.15471, "loss_cls_dn_0": 0.23379, "loss_box_dn_0": 0.80669, "loss_cls_dn_1": 0.1801, "loss_box_dn_1": 0.77882, "loss_cls_dn_2": 0.18746, "loss_box_dn_2": 0.75212, "loss_cls_dn_3": 0.18671, "loss_box_dn_3": 0.75012, "loss_cls_dn_4": 0.19576, "loss_box_dn_4": 0.7571, "loss_cls_dn_5": 0.20367, "loss_box_dn_5": 0.7766, "loss_dense_depth": 0.79692, "loss": 28.58142, "grad_norm": 37.97226, "time": 1.5082} -{"mode": "train", "epoch": 1, "iter": 118, "lr": 0.00015, "memory": 49167, "data_time": 0.075, "loss_cls_0": 0.90816, "loss_box_0": 1.77252, "loss_cns_0": 0.62199, "loss_yns_0": 0.15563, "loss_cls_1": 0.97041, "loss_box_1": 1.91351, "loss_cns_1": 0.63793, "loss_yns_1": 0.15641, "loss_cls_2": 0.98512, "loss_box_2": 1.84345, "loss_cns_2": 0.64848, "loss_yns_2": 0.15655, "loss_cls_3": 1.01206, "loss_box_3": 1.82258, "loss_cns_3": 0.65393, "loss_yns_3": 0.15519, "loss_cls_4": 1.00776, "loss_box_4": 1.82634, "loss_cns_4": 0.65438, "loss_yns_4": 0.15651, "loss_cls_5": 1.0312, "loss_box_5": 1.83637, "loss_cns_5": 0.64679, "loss_yns_5": 0.15706, "loss_cls_dn_0": 0.23531, "loss_box_dn_0": 0.8041, "loss_cls_dn_1": 0.17596, "loss_box_dn_1": 0.78423, "loss_cls_dn_2": 0.18345, "loss_box_dn_2": 0.75706, "loss_cls_dn_3": 0.1848, "loss_box_dn_3": 0.75472, "loss_cls_dn_4": 0.1887, "loss_box_dn_4": 0.7639, "loss_cls_dn_5": 0.20929, "loss_box_dn_5": 0.77325, "loss_dense_depth": 0.79493, "loss": 28.34001, "grad_norm": 36.13984, "time": 1.52259} -{"mode": "train", "epoch": 1, "iter": 119, "lr": 0.00015, "memory": 49167, "data_time": 0.07857, "loss_cls_0": 0.89249, "loss_box_0": 1.75526, "loss_cns_0": 0.62192, "loss_yns_0": 0.1552, "loss_cls_1": 0.95803, "loss_box_1": 1.94534, "loss_cns_1": 0.63323, "loss_yns_1": 0.15441, "loss_cls_2": 0.97733, "loss_box_2": 1.87455, "loss_cns_2": 0.64449, "loss_yns_2": 0.15495, "loss_cls_3": 1.00254, "loss_box_3": 1.85071, "loss_cns_3": 0.64918, "loss_yns_3": 0.1547, "loss_cls_4": 1.00119, "loss_box_4": 1.84508, "loss_cns_4": 0.65032, "loss_yns_4": 0.1565, "loss_cls_5": 1.02132, "loss_box_5": 1.84747, "loss_cns_5": 0.6456, "loss_yns_5": 0.15645, "loss_cls_dn_0": 0.23052, "loss_box_dn_0": 0.78739, "loss_cls_dn_1": 0.16883, "loss_box_dn_1": 0.77343, "loss_cls_dn_2": 0.17703, "loss_box_dn_2": 0.7463, "loss_cls_dn_3": 0.1783, "loss_box_dn_3": 0.74156, "loss_cls_dn_4": 0.18514, "loss_box_dn_4": 0.7472, "loss_cls_dn_5": 0.20039, "loss_box_dn_5": 0.74989, "loss_dense_depth": 0.79428, "loss": 28.2285, "grad_norm": 30.25276, "time": 1.50889} -{"mode": "train", "epoch": 1, "iter": 120, "lr": 0.00015, "memory": 49167, "data_time": 0.08222, "loss_cls_0": 0.88937, "loss_box_0": 1.76345, "loss_cns_0": 0.62005, "loss_yns_0": 0.15489, "loss_cls_1": 0.97124, "loss_box_1": 1.96582, "loss_cns_1": 0.63309, "loss_yns_1": 0.15494, "loss_cls_2": 0.97556, "loss_box_2": 1.89397, "loss_cns_2": 0.64716, "loss_yns_2": 0.15852, "loss_cls_3": 0.99157, "loss_box_3": 1.86456, "loss_cns_3": 0.64851, "loss_yns_3": 0.15515, "loss_cls_4": 0.99593, "loss_box_4": 1.85591, "loss_cns_4": 0.65072, "loss_yns_4": 0.15651, "loss_cls_5": 0.9999, "loss_box_5": 1.86653, "loss_cns_5": 0.65007, "loss_yns_5": 0.15637, "loss_cls_dn_0": 0.22859, "loss_box_dn_0": 0.79476, "loss_cls_dn_1": 0.1688, "loss_box_dn_1": 0.7632, "loss_cls_dn_2": 0.17531, "loss_box_dn_2": 0.73132, "loss_cls_dn_3": 0.17668, "loss_box_dn_3": 0.72488, "loss_cls_dn_4": 0.18953, "loss_box_dn_4": 0.72675, "loss_cls_dn_5": 0.19608, "loss_box_dn_5": 0.73489, "loss_dense_depth": 0.81034, "loss": 28.24093, "grad_norm": 31.47913, "time": 1.51065} -{"mode": "train", "epoch": 1, "iter": 121, "lr": 0.00015, "memory": 49167, "data_time": 0.11759, "loss_cls_0": 0.89823, "loss_box_0": 1.77903, "loss_cns_0": 0.62129, "loss_yns_0": 0.15449, "loss_cls_1": 0.96044, "loss_box_1": 1.98553, "loss_cns_1": 0.63256, "loss_yns_1": 0.15692, "loss_cls_2": 0.98763, "loss_box_2": 1.91171, "loss_cns_2": 0.64562, "loss_yns_2": 0.16296, "loss_cls_3": 1.00652, "loss_box_3": 1.88632, "loss_cns_3": 0.65097, "loss_yns_3": 0.15604, "loss_cls_4": 1.00971, "loss_box_4": 1.8815, "loss_cns_4": 0.65241, "loss_yns_4": 0.15464, "loss_cls_5": 1.01294, "loss_box_5": 1.88995, "loss_cns_5": 0.64901, "loss_yns_5": 0.15417, "loss_cls_dn_0": 0.23238, "loss_box_dn_0": 0.80238, "loss_cls_dn_1": 0.16362, "loss_box_dn_1": 0.76746, "loss_cls_dn_2": 0.17102, "loss_box_dn_2": 0.73868, "loss_cls_dn_3": 0.17823, "loss_box_dn_3": 0.73415, "loss_cls_dn_4": 0.18651, "loss_box_dn_4": 0.73675, "loss_cls_dn_5": 0.18929, "loss_box_dn_5": 0.74768, "loss_dense_depth": 0.85188, "loss": 28.50064, "grad_norm": 35.42499, "time": 1.63843} -{"mode": "train", "epoch": 1, "iter": 122, "lr": 0.00015, "memory": 49167, "data_time": 0.0815, "loss_cls_0": 0.90614, "loss_box_0": 1.75988, "loss_cns_0": 0.62097, "loss_yns_0": 0.15782, "loss_cls_1": 0.95941, "loss_box_1": 1.90198, "loss_cns_1": 0.63486, "loss_yns_1": 0.1574, "loss_cls_2": 0.98895, "loss_box_2": 1.84345, "loss_cns_2": 0.64586, "loss_yns_2": 0.16155, "loss_cls_3": 1.01113, "loss_box_3": 1.8155, "loss_cns_3": 0.64944, "loss_yns_3": 0.15955, "loss_cls_4": 1.01917, "loss_box_4": 1.81041, "loss_cns_4": 0.65026, "loss_yns_4": 0.16026, "loss_cls_5": 1.0349, "loss_box_5": 1.81152, "loss_cns_5": 0.64607, "loss_yns_5": 0.15924, "loss_cls_dn_0": 0.23417, "loss_box_dn_0": 0.79879, "loss_cls_dn_1": 0.16443, "loss_box_dn_1": 0.76908, "loss_cls_dn_2": 0.1709, "loss_box_dn_2": 0.75191, "loss_cls_dn_3": 0.18171, "loss_box_dn_3": 0.75193, "loss_cls_dn_4": 0.18062, "loss_box_dn_4": 0.75591, "loss_cls_dn_5": 0.18672, "loss_box_dn_5": 0.76893, "loss_dense_depth": 0.8186, "loss": 28.19939, "grad_norm": 40.2633, "time": 1.59855} -{"mode": "train", "epoch": 1, "iter": 123, "lr": 0.00015, "memory": 49167, "data_time": 0.10534, "loss_cls_0": 0.89914, "loss_box_0": 1.74566, "loss_cns_0": 0.62072, "loss_yns_0": 0.1586, "loss_cls_1": 0.97487, "loss_box_1": 1.86766, "loss_cns_1": 0.64006, "loss_yns_1": 0.15805, "loss_cls_2": 0.98311, "loss_box_2": 1.80958, "loss_cns_2": 0.64926, "loss_yns_2": 0.15892, "loss_cls_3": 1.00304, "loss_box_3": 1.79345, "loss_cns_3": 0.65146, "loss_yns_3": 0.15898, "loss_cls_4": 1.01204, "loss_box_4": 1.79497, "loss_cns_4": 0.65191, "loss_yns_4": 0.15854, "loss_cls_5": 1.00778, "loss_box_5": 1.80416, "loss_cns_5": 0.64766, "loss_yns_5": 0.15981, "loss_cls_dn_0": 0.22691, "loss_box_dn_0": 0.79301, "loss_cls_dn_1": 0.16648, "loss_box_dn_1": 0.8029, "loss_cls_dn_2": 0.17212, "loss_box_dn_2": 0.78894, "loss_cls_dn_3": 0.18352, "loss_box_dn_3": 0.79783, "loss_cls_dn_4": 0.18091, "loss_box_dn_4": 0.80818, "loss_cls_dn_5": 0.18816, "loss_box_dn_5": 0.82538, "loss_dense_depth": 0.79378, "loss": 28.23755, "grad_norm": 35.61314, "time": 1.54586} -{"mode": "train", "epoch": 1, "iter": 124, "lr": 0.00015, "memory": 49167, "data_time": 0.07715, "loss_cls_0": 0.8882, "loss_box_0": 1.75258, "loss_cns_0": 0.62486, "loss_yns_0": 0.15714, "loss_cls_1": 0.96929, "loss_box_1": 1.88817, "loss_cns_1": 0.6408, "loss_yns_1": 0.15811, "loss_cls_2": 0.98028, "loss_box_2": 1.82945, "loss_cns_2": 0.65047, "loss_yns_2": 0.15629, "loss_cls_3": 1.00741, "loss_box_3": 1.81381, "loss_cns_3": 0.65295, "loss_yns_3": 0.15648, "loss_cls_4": 1.01635, "loss_box_4": 1.80734, "loss_cns_4": 0.65174, "loss_yns_4": 0.15559, "loss_cls_5": 1.00197, "loss_box_5": 1.81474, "loss_cns_5": 0.65224, "loss_yns_5": 0.15655, "loss_cls_dn_0": 0.22645, "loss_box_dn_0": 0.7914, "loss_cls_dn_1": 0.16405, "loss_box_dn_1": 0.80107, "loss_cls_dn_2": 0.16611, "loss_box_dn_2": 0.78249, "loss_cls_dn_3": 0.17525, "loss_box_dn_3": 0.78484, "loss_cls_dn_4": 0.17974, "loss_box_dn_4": 0.78647, "loss_cls_dn_5": 0.18503, "loss_box_dn_5": 0.79264, "loss_dense_depth": 0.81877, "loss": 28.23712, "grad_norm": 34.27293, "time": 1.52557} -{"mode": "train", "epoch": 1, "iter": 125, "lr": 0.00015, "memory": 49167, "data_time": 0.07884, "loss_cls_0": 0.90544, "loss_box_0": 1.78653, "loss_cns_0": 0.61905, "loss_yns_0": 0.15847, "loss_cls_1": 0.97034, "loss_box_1": 1.85513, "loss_cns_1": 0.64137, "loss_yns_1": 0.15929, "loss_cls_2": 0.99578, "loss_box_2": 1.80313, "loss_cns_2": 0.64982, "loss_yns_2": 0.15922, "loss_cls_3": 1.01679, "loss_box_3": 1.78291, "loss_cns_3": 0.65239, "loss_yns_3": 0.15761, "loss_cls_4": 1.02278, "loss_box_4": 1.77516, "loss_cns_4": 0.65231, "loss_yns_4": 0.15721, "loss_cls_5": 1.01387, "loss_box_5": 1.78159, "loss_cns_5": 0.65617, "loss_yns_5": 0.15545, "loss_cls_dn_0": 0.23287, "loss_box_dn_0": 0.7837, "loss_cls_dn_1": 0.16364, "loss_box_dn_1": 0.79667, "loss_cls_dn_2": 0.1665, "loss_box_dn_2": 0.77613, "loss_cls_dn_3": 0.17032, "loss_box_dn_3": 0.77054, "loss_cls_dn_4": 0.18349, "loss_box_dn_4": 0.76798, "loss_cls_dn_5": 0.18563, "loss_box_dn_5": 0.77129, "loss_dense_depth": 0.82951, "loss": 28.12608, "grad_norm": 26.94234, "time": 1.52562} -{"mode": "train", "epoch": 1, "iter": 126, "lr": 0.00015, "memory": 49167, "data_time": 0.11336, "loss_cls_0": 0.89879, "loss_box_0": 1.76709, "loss_cns_0": 0.61574, "loss_yns_0": 0.15572, "loss_cls_1": 0.96148, "loss_box_1": 1.88741, "loss_cns_1": 0.6399, "loss_yns_1": 0.15763, "loss_cls_2": 0.99902, "loss_box_2": 1.82073, "loss_cns_2": 0.64685, "loss_yns_2": 0.1588, "loss_cls_3": 1.01733, "loss_box_3": 1.80515, "loss_cns_3": 0.65261, "loss_yns_3": 0.15689, "loss_cls_4": 1.01526, "loss_box_4": 1.80412, "loss_cns_4": 0.65232, "loss_yns_4": 0.15692, "loss_cls_5": 1.0139, "loss_box_5": 1.79734, "loss_cns_5": 0.65292, "loss_yns_5": 0.15471, "loss_cls_dn_0": 0.23374, "loss_box_dn_0": 0.78566, "loss_cls_dn_1": 0.16665, "loss_box_dn_1": 0.75021, "loss_cls_dn_2": 0.17318, "loss_box_dn_2": 0.72855, "loss_cls_dn_3": 0.17556, "loss_box_dn_3": 0.72282, "loss_cls_dn_4": 0.18826, "loss_box_dn_4": 0.72319, "loss_cls_dn_5": 0.18942, "loss_box_dn_5": 0.7228, "loss_dense_depth": 0.78702, "loss": 27.9357, "grad_norm": 30.23627, "time": 1.54294} -{"mode": "train", "epoch": 1, "iter": 127, "lr": 0.00015, "memory": 49167, "data_time": 0.07841, "loss_cls_0": 0.89928, "loss_box_0": 1.7445, "loss_cns_0": 0.61459, "loss_yns_0": 0.15714, "loss_cls_1": 0.97609, "loss_box_1": 1.84786, "loss_cns_1": 0.63994, "loss_yns_1": 0.15451, "loss_cls_2": 0.99661, "loss_box_2": 1.77098, "loss_cns_2": 0.64943, "loss_yns_2": 0.15628, "loss_cls_3": 1.01635, "loss_box_3": 1.75795, "loss_cns_3": 0.65421, "loss_yns_3": 0.15529, "loss_cls_4": 1.02147, "loss_box_4": 1.7614, "loss_cns_4": 0.65226, "loss_yns_4": 0.15638, "loss_cls_5": 1.01074, "loss_box_5": 1.76508, "loss_cns_5": 0.65039, "loss_yns_5": 0.15429, "loss_cls_dn_0": 0.23191, "loss_box_dn_0": 0.78544, "loss_cls_dn_1": 0.16289, "loss_box_dn_1": 0.73631, "loss_cls_dn_2": 0.16879, "loss_box_dn_2": 0.71668, "loss_cls_dn_3": 0.17481, "loss_box_dn_3": 0.71668, "loss_cls_dn_4": 0.18521, "loss_box_dn_4": 0.7234, "loss_cls_dn_5": 0.18523, "loss_box_dn_5": 0.73091, "loss_dense_depth": 0.77157, "loss": 27.65284, "grad_norm": 29.01523, "time": 1.52536} -{"mode": "train", "epoch": 1, "iter": 128, "lr": 0.00015, "memory": 49167, "data_time": 0.076, "loss_cls_0": 0.90639, "loss_box_0": 1.71472, "loss_cns_0": 0.61778, "loss_yns_0": 0.15797, "loss_cls_1": 0.97016, "loss_box_1": 1.83585, "loss_cns_1": 0.63776, "loss_yns_1": 0.15818, "loss_cls_2": 1.00488, "loss_box_2": 1.77644, "loss_cns_2": 0.64588, "loss_yns_2": 0.15675, "loss_cls_3": 1.01665, "loss_box_3": 1.76457, "loss_cns_3": 0.64954, "loss_yns_3": 0.15948, "loss_cls_4": 1.01972, "loss_box_4": 1.76509, "loss_cns_4": 0.65198, "loss_yns_4": 0.15857, "loss_cls_5": 1.01167, "loss_box_5": 1.77612, "loss_cns_5": 0.65002, "loss_yns_5": 0.15628, "loss_cls_dn_0": 0.23105, "loss_box_dn_0": 0.78604, "loss_cls_dn_1": 0.15901, "loss_box_dn_1": 0.74039, "loss_cls_dn_2": 0.16575, "loss_box_dn_2": 0.73349, "loss_cls_dn_3": 0.17271, "loss_box_dn_3": 0.73846, "loss_cls_dn_4": 0.17677, "loss_box_dn_4": 0.74921, "loss_cls_dn_5": 0.18283, "loss_box_dn_5": 0.76694, "loss_dense_depth": 0.82882, "loss": 27.79391, "grad_norm": 33.80426, "time": 1.52475} -{"mode": "train", "epoch": 1, "iter": 129, "lr": 0.00015, "memory": 49167, "data_time": 0.08658, "loss_cls_0": 0.89338, "loss_box_0": 1.73579, "loss_cns_0": 0.62007, "loss_yns_0": 0.15972, "loss_cls_1": 0.95648, "loss_box_1": 1.82733, "loss_cns_1": 0.63983, "loss_yns_1": 0.15896, "loss_cls_2": 0.99724, "loss_box_2": 1.77632, "loss_cns_2": 0.64822, "loss_yns_2": 0.157, "loss_cls_3": 1.01071, "loss_box_3": 1.76089, "loss_cns_3": 0.65181, "loss_yns_3": 0.15727, "loss_cls_4": 1.01642, "loss_box_4": 1.76702, "loss_cns_4": 0.65115, "loss_yns_4": 0.15849, "loss_cls_5": 0.99832, "loss_box_5": 1.78432, "loss_cns_5": 0.65322, "loss_yns_5": 0.15704, "loss_cls_dn_0": 0.22666, "loss_box_dn_0": 0.78735, "loss_cls_dn_1": 0.15857, "loss_box_dn_1": 0.76082, "loss_cls_dn_2": 0.16238, "loss_box_dn_2": 0.75365, "loss_cls_dn_3": 0.168, "loss_box_dn_3": 0.75937, "loss_cls_dn_4": 0.17533, "loss_box_dn_4": 0.7707, "loss_cls_dn_5": 0.18318, "loss_box_dn_5": 0.79243, "loss_dense_depth": 0.77134, "loss": 27.80678, "grad_norm": 39.20029, "time": 1.52744} -{"mode": "train", "epoch": 1, "iter": 130, "lr": 0.00015, "memory": 49167, "data_time": 0.0764, "loss_cls_0": 0.9088, "loss_box_0": 1.76687, "loss_cns_0": 0.61721, "loss_yns_0": 0.15583, "loss_cls_1": 0.97639, "loss_box_1": 1.8433, "loss_cns_1": 0.64543, "loss_yns_1": 0.15486, "loss_cls_2": 1.00934, "loss_box_2": 1.80267, "loss_cns_2": 0.65248, "loss_yns_2": 0.15509, "loss_cls_3": 1.01513, "loss_box_3": 1.77247, "loss_cns_3": 0.65463, "loss_yns_3": 0.15359, "loss_cls_4": 1.01769, "loss_box_4": 1.77406, "loss_cns_4": 0.65413, "loss_yns_4": 0.15678, "loss_cls_5": 1.01041, "loss_box_5": 1.79431, "loss_cns_5": 0.65479, "loss_yns_5": 0.15381, "loss_cls_dn_0": 0.22993, "loss_box_dn_0": 0.79161, "loss_cls_dn_1": 0.1598, "loss_box_dn_1": 0.78552, "loss_cls_dn_2": 0.16041, "loss_box_dn_2": 0.77669, "loss_cls_dn_3": 0.16544, "loss_box_dn_3": 0.77664, "loss_cls_dn_4": 0.17405, "loss_box_dn_4": 0.78322, "loss_cls_dn_5": 0.18056, "loss_box_dn_5": 0.80214, "loss_dense_depth": 0.82146, "loss": 28.10757, "grad_norm": 34.61551, "time": 1.52281} -{"mode": "train", "epoch": 1, "iter": 131, "lr": 0.00015, "memory": 49167, "data_time": 0.07601, "loss_cls_0": 0.91096, "loss_box_0": 1.74359, "loss_cns_0": 0.62081, "loss_yns_0": 0.15302, "loss_cls_1": 0.975, "loss_box_1": 1.85604, "loss_cns_1": 0.64771, "loss_yns_1": 0.15358, "loss_cls_2": 0.99529, "loss_box_2": 1.82428, "loss_cns_2": 0.65093, "loss_yns_2": 0.15565, "loss_cls_3": 1.01555, "loss_box_3": 1.79308, "loss_cns_3": 0.65542, "loss_yns_3": 0.15359, "loss_cls_4": 1.01699, "loss_box_4": 1.7884, "loss_cns_4": 0.65403, "loss_yns_4": 0.15388, "loss_cls_5": 1.02475, "loss_box_5": 1.79151, "loss_cns_5": 0.65334, "loss_yns_5": 0.15291, "loss_cls_dn_0": 0.23041, "loss_box_dn_0": 0.79543, "loss_cls_dn_1": 0.15736, "loss_box_dn_1": 0.7636, "loss_cls_dn_2": 0.15709, "loss_box_dn_2": 0.74849, "loss_cls_dn_3": 0.16273, "loss_box_dn_3": 0.74034, "loss_cls_dn_4": 0.17269, "loss_box_dn_4": 0.73878, "loss_cls_dn_5": 0.17887, "loss_box_dn_5": 0.74572, "loss_dense_depth": 0.82576, "loss": 27.95757, "grad_norm": 26.582, "time": 1.52813} -{"mode": "train", "epoch": 1, "iter": 132, "lr": 0.00015, "memory": 49167, "data_time": 0.07466, "loss_cls_0": 0.8933, "loss_box_0": 1.75531, "loss_cns_0": 0.62222, "loss_yns_0": 0.15038, "loss_cls_1": 0.95339, "loss_box_1": 1.85245, "loss_cns_1": 0.64079, "loss_yns_1": 0.15266, "loss_cls_2": 0.98292, "loss_box_2": 1.81623, "loss_cns_2": 0.64787, "loss_yns_2": 0.1527, "loss_cls_3": 0.99963, "loss_box_3": 1.79733, "loss_cns_3": 0.64874, "loss_yns_3": 0.1531, "loss_cls_4": 1.00665, "loss_box_4": 1.79765, "loss_cns_4": 0.65032, "loss_yns_4": 0.15137, "loss_cls_5": 1.00263, "loss_box_5": 1.79849, "loss_cns_5": 0.6467, "loss_yns_5": 0.15219, "loss_cls_dn_0": 0.23248, "loss_box_dn_0": 0.78439, "loss_cls_dn_1": 0.15682, "loss_box_dn_1": 0.73103, "loss_cls_dn_2": 0.1594, "loss_box_dn_2": 0.71291, "loss_cls_dn_3": 0.16781, "loss_box_dn_3": 0.70763, "loss_cls_dn_4": 0.17641, "loss_box_dn_4": 0.71, "loss_cls_dn_5": 0.18432, "loss_box_dn_5": 0.71289, "loss_dense_depth": 0.82698, "loss": 27.68812, "grad_norm": 35.55404, "time": 1.51295} -{"mode": "train", "epoch": 1, "iter": 133, "lr": 0.00015, "memory": 49167, "data_time": 0.09225, "loss_cls_0": 0.90087, "loss_box_0": 1.75332, "loss_cns_0": 0.6179, "loss_yns_0": 0.1518, "loss_cls_1": 0.95159, "loss_box_1": 1.86578, "loss_cns_1": 0.63604, "loss_yns_1": 0.15257, "loss_cls_2": 0.97995, "loss_box_2": 1.80677, "loss_cns_2": 0.64637, "loss_yns_2": 0.15293, "loss_cls_3": 0.99885, "loss_box_3": 1.78711, "loss_cns_3": 0.64715, "loss_yns_3": 0.15262, "loss_cls_4": 1.00093, "loss_box_4": 1.79075, "loss_cns_4": 0.64688, "loss_yns_4": 0.1528, "loss_cls_5": 1.01365, "loss_box_5": 1.7966, "loss_cns_5": 0.64517, "loss_yns_5": 0.1538, "loss_cls_dn_0": 0.23151, "loss_box_dn_0": 0.7899, "loss_cls_dn_1": 0.15812, "loss_box_dn_1": 0.73503, "loss_cls_dn_2": 0.16185, "loss_box_dn_2": 0.71492, "loss_cls_dn_3": 0.16576, "loss_box_dn_3": 0.71066, "loss_cls_dn_4": 0.17476, "loss_box_dn_4": 0.72142, "loss_cls_dn_5": 0.18181, "loss_box_dn_5": 0.73026, "loss_dense_depth": 0.83108, "loss": 27.70928, "grad_norm": 30.0114, "time": 1.54752} -{"mode": "train", "epoch": 1, "iter": 134, "lr": 0.00015, "memory": 49167, "data_time": 0.07815, "loss_cls_0": 0.87156, "loss_box_0": 1.71764, "loss_cns_0": 0.62227, "loss_yns_0": 0.15313, "loss_cls_1": 0.94114, "loss_box_1": 1.86245, "loss_cns_1": 0.63899, "loss_yns_1": 0.15128, "loss_cls_2": 0.96857, "loss_box_2": 1.81047, "loss_cns_2": 0.64786, "loss_yns_2": 0.15341, "loss_cls_3": 0.98418, "loss_box_3": 1.78235, "loss_cns_3": 0.65153, "loss_yns_3": 0.15195, "loss_cls_4": 0.98767, "loss_box_4": 1.78331, "loss_cns_4": 0.64914, "loss_yns_4": 0.15249, "loss_cls_5": 0.99758, "loss_box_5": 1.79116, "loss_cns_5": 0.64794, "loss_yns_5": 0.15387, "loss_cls_dn_0": 0.22546, "loss_box_dn_0": 0.78607, "loss_cls_dn_1": 0.15663, "loss_box_dn_1": 0.73246, "loss_cls_dn_2": 0.16259, "loss_box_dn_2": 0.71461, "loss_cls_dn_3": 0.16309, "loss_box_dn_3": 0.71052, "loss_cls_dn_4": 0.17361, "loss_box_dn_4": 0.72514, "loss_cls_dn_5": 0.17804, "loss_box_dn_5": 0.73987, "loss_dense_depth": 0.80585, "loss": 27.54586, "grad_norm": 35.46795, "time": 1.51624} -{"mode": "train", "epoch": 1, "iter": 135, "lr": 0.00015, "memory": 49167, "data_time": 0.0787, "loss_cls_0": 0.89291, "loss_box_0": 1.74608, "loss_cns_0": 0.61742, "loss_yns_0": 0.15496, "loss_cls_1": 0.94846, "loss_box_1": 1.90413, "loss_cns_1": 0.63894, "loss_yns_1": 0.15437, "loss_cls_2": 0.97986, "loss_box_2": 1.853, "loss_cns_2": 0.6489, "loss_yns_2": 0.1547, "loss_cls_3": 0.99304, "loss_box_3": 1.82601, "loss_cns_3": 0.64905, "loss_yns_3": 0.15488, "loss_cls_4": 0.99493, "loss_box_4": 1.82815, "loss_cns_4": 0.64882, "loss_yns_4": 0.154, "loss_cls_5": 0.9923, "loss_box_5": 1.84486, "loss_cns_5": 0.64698, "loss_yns_5": 0.154, "loss_cls_dn_0": 0.23563, "loss_box_dn_0": 0.78636, "loss_cls_dn_1": 0.15576, "loss_box_dn_1": 0.75614, "loss_cls_dn_2": 0.16086, "loss_box_dn_2": 0.73949, "loss_cls_dn_3": 0.16469, "loss_box_dn_3": 0.73371, "loss_cls_dn_4": 0.17582, "loss_box_dn_4": 0.74686, "loss_cls_dn_5": 0.18238, "loss_box_dn_5": 0.76338, "loss_dense_depth": 0.80706, "loss": 27.98891, "grad_norm": 41.51213, "time": 1.51687} -{"mode": "train", "epoch": 1, "iter": 136, "lr": 0.00015, "memory": 49167, "data_time": 0.07696, "loss_cls_0": 0.86133, "loss_box_0": 1.71606, "loss_cns_0": 0.62116, "loss_yns_0": 0.15254, "loss_cls_1": 0.91744, "loss_box_1": 1.85127, "loss_cns_1": 0.63984, "loss_yns_1": 0.15021, "loss_cls_2": 0.94293, "loss_box_2": 1.79554, "loss_cns_2": 0.65118, "loss_yns_2": 0.15121, "loss_cls_3": 0.96413, "loss_box_3": 1.76878, "loss_cns_3": 0.65183, "loss_yns_3": 0.15254, "loss_cls_4": 0.9629, "loss_box_4": 1.76592, "loss_cns_4": 0.65362, "loss_yns_4": 0.152, "loss_cls_5": 0.96226, "loss_box_5": 1.77809, "loss_cns_5": 0.65228, "loss_yns_5": 0.15123, "loss_cls_dn_0": 0.23018, "loss_box_dn_0": 0.78011, "loss_cls_dn_1": 0.15195, "loss_box_dn_1": 0.7559, "loss_cls_dn_2": 0.15334, "loss_box_dn_2": 0.74063, "loss_cls_dn_3": 0.1589, "loss_box_dn_3": 0.73829, "loss_cls_dn_4": 0.16619, "loss_box_dn_4": 0.74773, "loss_cls_dn_5": 0.17265, "loss_box_dn_5": 0.761, "loss_dense_depth": 0.77257, "loss": 27.39574, "grad_norm": 31.16214, "time": 1.51855} -{"mode": "train", "epoch": 1, "iter": 137, "lr": 0.00015, "memory": 49167, "data_time": 0.07822, "loss_cls_0": 0.87848, "loss_box_0": 1.74144, "loss_cns_0": 0.62004, "loss_yns_0": 0.15234, "loss_cls_1": 0.94324, "loss_box_1": 1.89577, "loss_cns_1": 0.63642, "loss_yns_1": 0.15149, "loss_cls_2": 0.95218, "loss_box_2": 1.83064, "loss_cns_2": 0.6454, "loss_yns_2": 0.15046, "loss_cls_3": 0.97214, "loss_box_3": 1.81763, "loss_cns_3": 0.64792, "loss_yns_3": 0.15145, "loss_cls_4": 0.97287, "loss_box_4": 1.80846, "loss_cns_4": 0.6474, "loss_yns_4": 0.15172, "loss_cls_5": 0.97127, "loss_box_5": 1.80631, "loss_cns_5": 0.64898, "loss_yns_5": 0.1512, "loss_cls_dn_0": 0.23625, "loss_box_dn_0": 0.78266, "loss_cls_dn_1": 0.15144, "loss_box_dn_1": 0.76242, "loss_cls_dn_2": 0.15451, "loss_box_dn_2": 0.74486, "loss_cls_dn_3": 0.15968, "loss_box_dn_3": 0.74583, "loss_cls_dn_4": 0.1678, "loss_box_dn_4": 0.75, "loss_cls_dn_5": 0.16774, "loss_box_dn_5": 0.75573, "loss_dense_depth": 0.79452, "loss": 27.7187, "grad_norm": 39.26744, "time": 1.55299} -{"mode": "train", "epoch": 1, "iter": 138, "lr": 0.00015, "memory": 49167, "data_time": 0.07741, "loss_cls_0": 0.86716, "loss_box_0": 1.70827, "loss_cns_0": 0.62535, "loss_yns_0": 0.15051, "loss_cls_1": 0.93445, "loss_box_1": 1.83857, "loss_cns_1": 0.63958, "loss_yns_1": 0.14736, "loss_cls_2": 0.95497, "loss_box_2": 1.78562, "loss_cns_2": 0.64857, "loss_yns_2": 0.14871, "loss_cls_3": 0.96065, "loss_box_3": 1.76999, "loss_cns_3": 0.65415, "loss_yns_3": 0.14975, "loss_cls_4": 0.9748, "loss_box_4": 1.76964, "loss_cns_4": 0.65266, "loss_yns_4": 0.14895, "loss_cls_5": 0.98082, "loss_box_5": 1.77773, "loss_cns_5": 0.65325, "loss_yns_5": 0.14845, "loss_cls_dn_0": 0.23109, "loss_box_dn_0": 0.78238, "loss_cls_dn_1": 0.14808, "loss_box_dn_1": 0.75884, "loss_cls_dn_2": 0.15102, "loss_box_dn_2": 0.73555, "loss_cls_dn_3": 0.15606, "loss_box_dn_3": 0.73149, "loss_cls_dn_4": 0.16498, "loss_box_dn_4": 0.73576, "loss_cls_dn_5": 0.16536, "loss_box_dn_5": 0.74394, "loss_dense_depth": 0.7963, "loss": 27.39081, "grad_norm": 39.13601, "time": 1.5364} -{"mode": "train", "epoch": 1, "iter": 139, "lr": 0.00016, "memory": 49167, "data_time": 0.07551, "loss_cls_0": 0.8369, "loss_box_0": 1.68973, "loss_cns_0": 0.6276, "loss_yns_0": 0.14884, "loss_cls_1": 0.91465, "loss_box_1": 1.8086, "loss_cns_1": 0.64214, "loss_yns_1": 0.14704, "loss_cls_2": 0.92401, "loss_box_2": 1.76649, "loss_cns_2": 0.64615, "loss_yns_2": 0.14923, "loss_cls_3": 0.93956, "loss_box_3": 1.75511, "loss_cns_3": 0.6506, "loss_yns_3": 0.14998, "loss_cls_4": 0.94096, "loss_box_4": 1.74582, "loss_cns_4": 0.64871, "loss_yns_4": 0.14873, "loss_cls_5": 0.93881, "loss_box_5": 1.76186, "loss_cns_5": 0.65024, "loss_yns_5": 0.14844, "loss_cls_dn_0": 0.22408, "loss_box_dn_0": 0.77747, "loss_cls_dn_1": 0.1493, "loss_box_dn_1": 0.73282, "loss_cls_dn_2": 0.15348, "loss_box_dn_2": 0.71037, "loss_cls_dn_3": 0.15526, "loss_box_dn_3": 0.70777, "loss_cls_dn_4": 0.16699, "loss_box_dn_4": 0.70984, "loss_cls_dn_5": 0.16952, "loss_box_dn_5": 0.71894, "loss_dense_depth": 0.78696, "loss": 26.94302, "grad_norm": 30.59834, "time": 1.52607} -{"mode": "train", "epoch": 1, "iter": 140, "lr": 0.00016, "memory": 49167, "data_time": 0.08246, "loss_cls_0": 0.84804, "loss_box_0": 1.70802, "loss_cns_0": 0.62643, "loss_yns_0": 0.14915, "loss_cls_1": 0.91303, "loss_box_1": 1.78581, "loss_cns_1": 0.64572, "loss_yns_1": 0.14746, "loss_cls_2": 0.94282, "loss_box_2": 1.72695, "loss_cns_2": 0.65271, "loss_yns_2": 0.14798, "loss_cls_3": 0.96118, "loss_box_3": 1.74928, "loss_cns_3": 0.65353, "loss_yns_3": 0.14816, "loss_cls_4": 0.96232, "loss_box_4": 1.74194, "loss_cns_4": 0.65521, "loss_yns_4": 0.14959, "loss_cls_5": 0.95875, "loss_box_5": 1.74442, "loss_cns_5": 0.65558, "loss_yns_5": 0.14761, "loss_cls_dn_0": 0.22285, "loss_box_dn_0": 0.78514, "loss_cls_dn_1": 0.14464, "loss_box_dn_1": 0.72531, "loss_cls_dn_2": 0.14835, "loss_box_dn_2": 0.70652, "loss_cls_dn_3": 0.14923, "loss_box_dn_3": 0.71792, "loss_cls_dn_4": 0.15923, "loss_box_dn_4": 0.72318, "loss_cls_dn_5": 0.16561, "loss_box_dn_5": 0.7288, "loss_dense_depth": 0.76778, "loss": 26.96625, "grad_norm": 50.66061, "time": 1.51668} -{"mode": "train", "epoch": 1, "iter": 141, "lr": 0.00016, "memory": 49167, "data_time": 0.11069, "loss_cls_0": 0.8957, "loss_box_0": 1.77104, "loss_cns_0": 0.62744, "loss_yns_0": 0.15256, "loss_cls_1": 0.95551, "loss_box_1": 1.89795, "loss_cns_1": 0.64412, "loss_yns_1": 0.14879, "loss_cls_2": 0.96053, "loss_box_2": 1.83413, "loss_cns_2": 0.65144, "loss_yns_2": 0.15067, "loss_cls_3": 0.96299, "loss_box_3": 1.84353, "loss_cns_3": 0.65271, "loss_yns_3": 0.15102, "loss_cls_4": 0.97967, "loss_box_4": 1.8497, "loss_cns_4": 0.65338, "loss_yns_4": 0.15125, "loss_cls_5": 0.98423, "loss_box_5": 1.85073, "loss_cns_5": 0.65345, "loss_yns_5": 0.15254, "loss_cls_dn_0": 0.23722, "loss_box_dn_0": 0.78639, "loss_cls_dn_1": 0.15334, "loss_box_dn_1": 0.75873, "loss_cls_dn_2": 0.1543, "loss_box_dn_2": 0.74874, "loss_cls_dn_3": 0.15851, "loss_box_dn_3": 0.76395, "loss_cls_dn_4": 0.16923, "loss_box_dn_4": 0.78044, "loss_cls_dn_5": 0.17742, "loss_box_dn_5": 0.79536, "loss_dense_depth": 0.81952, "loss": 28.07826, "grad_norm": 47.36462, "time": 1.61343} -{"mode": "train", "epoch": 1, "iter": 142, "lr": 0.00016, "memory": 49167, "data_time": 0.07131, "loss_cls_0": 0.90016, "loss_box_0": 1.7885, "loss_cns_0": 0.61889, "loss_yns_0": 0.15379, "loss_cls_1": 0.97417, "loss_box_1": 1.89281, "loss_cns_1": 0.63906, "loss_yns_1": 0.14956, "loss_cls_2": 0.97047, "loss_box_2": 1.86151, "loss_cns_2": 0.6403, "loss_yns_2": 0.15317, "loss_cls_3": 0.98338, "loss_box_3": 1.85241, "loss_cns_3": 0.64431, "loss_yns_3": 0.1522, "loss_cls_4": 0.9784, "loss_box_4": 1.85912, "loss_cns_4": 0.64407, "loss_yns_4": 0.1523, "loss_cls_5": 0.97578, "loss_box_5": 1.86396, "loss_cns_5": 0.64128, "loss_yns_5": 0.15214, "loss_cls_dn_0": 0.23368, "loss_box_dn_0": 0.78884, "loss_cls_dn_1": 0.15526, "loss_box_dn_1": 0.78024, "loss_cls_dn_2": 0.15722, "loss_box_dn_2": 0.78686, "loss_cls_dn_3": 0.16484, "loss_box_dn_3": 0.79721, "loss_cls_dn_4": 0.17491, "loss_box_dn_4": 0.81538, "loss_cls_dn_5": 0.18065, "loss_box_dn_5": 0.83772, "loss_dense_depth": 0.80563, "loss": 28.32019, "grad_norm": 57.40893, "time": 1.57745} -{"mode": "train", "epoch": 1, "iter": 143, "lr": 0.00016, "memory": 49167, "data_time": 0.09569, "loss_cls_0": 0.85773, "loss_box_0": 1.769, "loss_cns_0": 0.61989, "loss_yns_0": 0.15078, "loss_cls_1": 0.94237, "loss_box_1": 1.89774, "loss_cns_1": 0.64507, "loss_yns_1": 0.14767, "loss_cls_2": 0.95491, "loss_box_2": 1.86254, "loss_cns_2": 0.64337, "loss_yns_2": 0.15092, "loss_cls_3": 0.97783, "loss_box_3": 1.84957, "loss_cns_3": 0.64765, "loss_yns_3": 0.14917, "loss_cls_4": 0.96037, "loss_box_4": 1.83935, "loss_cns_4": 0.64623, "loss_yns_4": 0.14969, "loss_cls_5": 0.96447, "loss_box_5": 1.85431, "loss_cns_5": 0.64477, "loss_yns_5": 0.14912, "loss_cls_dn_0": 0.2245, "loss_box_dn_0": 0.78112, "loss_cls_dn_1": 0.15399, "loss_box_dn_1": 0.78667, "loss_cls_dn_2": 0.15672, "loss_box_dn_2": 0.79177, "loss_cls_dn_3": 0.16488, "loss_box_dn_3": 0.79909, "loss_cls_dn_4": 0.17756, "loss_box_dn_4": 0.80918, "loss_cls_dn_5": 0.18346, "loss_box_dn_5": 0.83405, "loss_dense_depth": 0.78848, "loss": 28.12598, "grad_norm": 54.83649, "time": 1.5313} -{"mode": "train", "epoch": 1, "iter": 144, "lr": 0.00016, "memory": 49167, "data_time": 0.07351, "loss_cls_0": 0.88265, "loss_box_0": 1.79662, "loss_cns_0": 0.61822, "loss_yns_0": 0.15129, "loss_cls_1": 0.96605, "loss_box_1": 1.90463, "loss_cns_1": 0.6445, "loss_yns_1": 0.14822, "loss_cls_2": 0.98303, "loss_box_2": 1.85508, "loss_cns_2": 0.65096, "loss_yns_2": 0.15161, "loss_cls_3": 0.98799, "loss_box_3": 1.84753, "loss_cns_3": 0.65035, "loss_yns_3": 0.14952, "loss_cls_4": 0.98471, "loss_box_4": 1.84828, "loss_cns_4": 0.65043, "loss_yns_4": 0.1495, "loss_cls_5": 0.98831, "loss_box_5": 1.86355, "loss_cns_5": 0.64681, "loss_yns_5": 0.14957, "loss_cls_dn_0": 0.23447, "loss_box_dn_0": 0.79236, "loss_cls_dn_1": 0.15749, "loss_box_dn_1": 0.79663, "loss_cls_dn_2": 0.15678, "loss_box_dn_2": 0.78645, "loss_cls_dn_3": 0.16204, "loss_box_dn_3": 0.79189, "loss_cls_dn_4": 0.1732, "loss_box_dn_4": 0.79912, "loss_cls_dn_5": 0.18186, "loss_box_dn_5": 0.81882, "loss_dense_depth": 0.77737, "loss": 28.29788, "grad_norm": 41.98172, "time": 1.51194} -{"mode": "train", "epoch": 1, "iter": 145, "lr": 0.00016, "memory": 49167, "data_time": 0.07143, "loss_cls_0": 0.87241, "loss_box_0": 1.78236, "loss_cns_0": 0.61985, "loss_yns_0": 0.15205, "loss_cls_1": 0.95838, "loss_box_1": 1.95754, "loss_cns_1": 0.64144, "loss_yns_1": 0.15045, "loss_cls_2": 0.97918, "loss_box_2": 1.91505, "loss_cns_2": 0.6499, "loss_yns_2": 0.15103, "loss_cls_3": 0.98023, "loss_box_3": 1.8999, "loss_cns_3": 0.64988, "loss_yns_3": 0.15068, "loss_cls_4": 0.98306, "loss_box_4": 1.90676, "loss_cns_4": 0.64922, "loss_yns_4": 0.15038, "loss_cls_5": 0.9787, "loss_box_5": 1.91579, "loss_cns_5": 0.64515, "loss_yns_5": 0.15101, "loss_cls_dn_0": 0.22791, "loss_box_dn_0": 0.7898, "loss_cls_dn_1": 0.15536, "loss_box_dn_1": 0.81566, "loss_cls_dn_2": 0.1535, "loss_box_dn_2": 0.80213, "loss_cls_dn_3": 0.1588, "loss_box_dn_3": 0.79939, "loss_cls_dn_4": 0.16454, "loss_box_dn_4": 0.80591, "loss_cls_dn_5": 0.17216, "loss_box_dn_5": 0.81416, "loss_dense_depth": 0.78244, "loss": 28.53217, "grad_norm": 46.70765, "time": 1.53189} -{"mode": "train", "epoch": 1, "iter": 146, "lr": 0.00016, "memory": 49167, "data_time": 0.11111, "loss_cls_0": 0.85788, "loss_box_0": 1.75798, "loss_cns_0": 0.61906, "loss_yns_0": 0.15201, "loss_cls_1": 0.92815, "loss_box_1": 1.88519, "loss_cns_1": 0.64379, "loss_yns_1": 0.14959, "loss_cls_2": 0.95447, "loss_box_2": 1.82108, "loss_cns_2": 0.65153, "loss_yns_2": 0.15153, "loss_cls_3": 0.96219, "loss_box_3": 1.80973, "loss_cns_3": 0.65546, "loss_yns_3": 0.15056, "loss_cls_4": 0.96508, "loss_box_4": 1.8103, "loss_cns_4": 0.65949, "loss_yns_4": 0.15211, "loss_cls_5": 0.96138, "loss_box_5": 1.80785, "loss_cns_5": 0.66397, "loss_yns_5": 0.1524, "loss_cls_dn_0": 0.22207, "loss_box_dn_0": 0.77828, "loss_cls_dn_1": 0.14571, "loss_box_dn_1": 0.76196, "loss_cls_dn_2": 0.14684, "loss_box_dn_2": 0.74124, "loss_cls_dn_3": 0.15027, "loss_box_dn_3": 0.73676, "loss_cls_dn_4": 0.15686, "loss_box_dn_4": 0.74381, "loss_cls_dn_5": 0.16425, "loss_box_dn_5": 0.74622, "loss_dense_depth": 0.76207, "loss": 27.57914, "grad_norm": 31.95215, "time": 1.56442} -{"mode": "train", "epoch": 1, "iter": 147, "lr": 0.00016, "memory": 49167, "data_time": 0.07358, "loss_cls_0": 0.8744, "loss_box_0": 1.79206, "loss_cns_0": 0.61533, "loss_yns_0": 0.15202, "loss_cls_1": 0.94903, "loss_box_1": 1.84836, "loss_cns_1": 0.64509, "loss_yns_1": 0.15131, "loss_cls_2": 0.97255, "loss_box_2": 1.78781, "loss_cns_2": 0.6518, "loss_yns_2": 0.15371, "loss_cls_3": 0.97581, "loss_box_3": 1.79221, "loss_cns_3": 0.65235, "loss_yns_3": 0.1532, "loss_cls_4": 0.97405, "loss_box_4": 1.80992, "loss_cns_4": 0.65492, "loss_yns_4": 0.15398, "loss_cls_5": 0.97012, "loss_box_5": 1.81432, "loss_cns_5": 0.65416, "loss_yns_5": 0.15454, "loss_cls_dn_0": 0.22682, "loss_box_dn_0": 0.78291, "loss_cls_dn_1": 0.14701, "loss_box_dn_1": 0.75006, "loss_cls_dn_2": 0.14799, "loss_box_dn_2": 0.7384, "loss_cls_dn_3": 0.15316, "loss_box_dn_3": 0.74033, "loss_cls_dn_4": 0.15776, "loss_box_dn_4": 0.75269, "loss_cls_dn_5": 0.1637, "loss_box_dn_5": 0.75666, "loss_dense_depth": 0.79445, "loss": 27.66503, "grad_norm": 47.1476, "time": 1.51726} diff --git a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20251113_104840.log b/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20251113_104840.log deleted file mode 100644 index 758be1fe2da0adfbc4bee6fcdd6d2c4c8f250b00..0000000000000000000000000000000000000000 --- a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20251113_104840.log +++ /dev/null @@ -1,3278 +0,0 @@ -2025-11-13 10:48:40,473 - mmdet - INFO - Environment info: ------------------------------------------------------------- -sys.platform: linux -Python: 3.10.12 (main, May 27 2025, 17:12:29) [GCC 11.4.0] -CUDA available: True -GPU 0,1,2,3,4,5,6,7: BW200, UBB BW1000 -CUDA_HOME: /opt/dtk -NVCC: Not Available -GCC: x86_64-linux-gnu-gcc (Ubuntu 11.4.0-1ubuntu1~22.04) 11.4.0 -PyTorch: 2.4.1 -PyTorch compiling details: PyTorch built with: - - GCC 10.3 - - C++ Version: 201703 - - Intel(R) Math Kernel Library Version 2020.0.4 Product Build 20200917 for Intel(R) 64 architecture applications - - OpenMP 201511 (a.k.a. OpenMP 4.5) - - LAPACK is enabled (usually provided by MKL) - - NNPACK is enabled - - CPU capability usage: AVX512 - - HIP Runtime 6.3.25211 - - MIOpen 2.17.0 - - Magma 2.8.0 - - Build settings: BLAS_INFO=mkl, BUILD_TYPE=Release, CXX_COMPILER=/opt/rh/gcc-toolset-10/root/usr/bin/c++, CXX_FLAGS= -D_GLIBCXX_USE_CXX11_ABI=1 -fvisibility-inlines-hidden -DUSE_PTHREADPOOL -DNDEBUG -DUSE_KINETO -DLIBKINETO_NOCUPTI -DUSE_FBGEMM -DUSE_PYTORCH_QNNPACK -DUSE_XNNPACK -DSYMBOLICATE_MOBILE_DEBUG_HANDLE -O2 -fPIC -Wall -Wextra -Werror=return-type -Werror=non-virtual-dtor -Werror=bool-operation -Wnarrowing -Wno-missing-field-initializers -Wno-type-limits -Wno-array-bounds -Wno-unknown-pragmas -Wno-unused-parameter -Wno-unused-function -Wno-unused-result -Wno-strict-overflow -Wno-strict-aliasing -Wno-stringop-overflow -Wsuggest-override -Wno-psabi -Wno-error=pedantic -Wno-error=old-style-cast -Wno-missing-braces -fdiagnostics-color=always -faligned-new -Wno-unused-but-set-variable -Wno-maybe-uninitialized -fno-math-errno -fno-trapping-math -Werror=format -Wno-stringop-overflow, FORCE_FALLBACK_CUDA_MPI=1, LAPACK_INFO=mkl, PERF_WITH_AVX=1, PERF_WITH_AVX2=1, PERF_WITH_AVX512=1, TORCH_VERSION=2.4.1, USE_CUDA=0, USE_CUDNN=OFF, USE_CUSPARSELT=OFF, USE_EXCEPTION_PTR=1, USE_GFLAGS=1, USE_GLOG=1, USE_GLOO=1, USE_MKL=ON, USE_MKLDNN=0, USE_MPI=1, USE_NCCL=1, USE_NNPACK=ON, USE_OPENMP=1, USE_ROCM=ON, USE_ROCM_KERNEL_ASSERT=OFF, - -TorchVision: 0.19.1 -OpenCV: 4.11.0 -MMCV: 1.6.1 -MMCV Compiler: GCC 11.4 -MMCV CUDA Compiler: rocm not available -MMDetection: 2.26.0+c41df4b ------------------------------------------------------------- - -2025-11-13 10:48:41,423 - mmdet - INFO - Distributed training: True -2025-11-13 10:48:42,142 - mmdet - INFO - Config: -plugin = True -plugin_dir = 'projects/mmdet3d_plugin/' -dist_params = dict(backend='nccl') -log_level = 'INFO' -work_dir = './work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704' -total_batch_size = 160 -num_gpus = 8 -batch_size = 20 -num_iters_per_epoch = 175 -num_epochs = 100 -checkpoint_epoch_interval = 20 -checkpoint_config = dict(interval=3500) -log_config = dict( - interval=1, - hooks=[ - dict(type='TextLoggerHook', by_epoch=False), - dict(type='TensorboardLoggerHook') - ]) -load_from = None -resume_from = None -workflow = [('train', 1)] -fp16 = dict(loss_scale=32.0) -input_shape = (704, 256) -tracking_test = True -tracking_threshold = 0.2 -class_names = [ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier', - 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' -] -num_classes = 10 -embed_dims = 256 -num_groups = 8 -num_decoder = 6 -num_single_frame_decoder = 1 -use_deformable_func = True -strides = [4, 8, 16, 32] -num_levels = 4 -num_depth_layers = 3 -drop_out = 0.1 -temporal = True -decouple_attn = True -with_quality_estimation = True -model = dict( - type='Sparse4D', - use_grid_mask=True, - use_deformable_func=True, - img_backbone=dict( - type='ResNet', - depth=50, - num_stages=4, - frozen_stages=-1, - norm_eval=False, - style='pytorch', - with_cp=True, - out_indices=(0, 1, 2, 3), - norm_cfg=dict(type='BN', requires_grad=True), - pretrained='ckpt/resnet50-19c8e357.pth'), - img_neck=dict( - type='FPN', - num_outs=4, - start_level=0, - out_channels=256, - add_extra_convs='on_output', - relu_before_extra_convs=True, - in_channels=[256, 512, 1024, 2048]), - depth_branch=dict( - type='DenseDepthNet', - embed_dims=256, - num_depth_layers=3, - loss_weight=0.2), - head=dict( - type='Sparse4DHead', - cls_threshold_to_reg=0.05, - decouple_attn=True, - instance_bank=dict( - type='InstanceBank', - num_anchor=900, - embed_dims=256, - anchor='nuscenes_kmeans900.npy', - anchor_handler=dict(type='SparseBox3DKeyPointsGenerator'), - num_temp_instances=600, - confidence_decay=0.6, - feat_grad=False), - anchor_encoder=dict( - type='SparseBox3DEncoder', - vel_dims=3, - embed_dims=[128, 32, 32, 64], - mode='cat', - output_fc=False, - in_loops=1, - out_loops=4), - num_single_frame_decoder=1, - operation_order=[ - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine' - ], - temp_graph_model=dict( - type='MultiheadAttention', - embed_dims=512, - num_heads=8, - batch_first=True, - dropout=0.1), - graph_model=dict( - type='MultiheadAttention', - embed_dims=512, - num_heads=8, - batch_first=True, - dropout=0.1), - norm_layer=dict(type='LN', normalized_shape=256), - ffn=dict( - type='AsymmetricFFN', - in_channels=512, - pre_norm=dict(type='LN'), - embed_dims=256, - feedforward_channels=1024, - num_fcs=2, - ffn_drop=0.1, - act_cfg=dict(type='ReLU', inplace=True)), - deformable_model=dict( - type='DeformableFeatureAggregation', - embed_dims=256, - num_groups=8, - num_levels=4, - num_cams=6, - attn_drop=0.15, - use_deformable_func=True, - use_camera_embed=True, - residual_mode='cat', - kps_generator=dict( - type='SparseBox3DKeyPointsGenerator', - num_learnable_pts=6, - fix_scale=[[0, 0, 0], [0.45, 0, 0], [-0.45, 0, 0], - [0, 0.45, 0], [0, -0.45, 0], [0, 0, 0.45], - [0, 0, -0.45]])), - refine_layer=dict( - type='SparseBox3DRefinementModule', - embed_dims=256, - num_cls=10, - refine_yaw=True, - with_quality_estimation=True), - sampler=dict( - type='SparseBox3DTarget', - num_dn_groups=5, - num_temp_dn_groups=3, - dn_noise_scale=[2.0, 2.0, 2.0, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5], - max_dn_gt=32, - add_neg_dn=True, - cls_weight=2.0, - box_weight=0.25, - reg_weights=[2.0, 2.0, 2.0, 0.5, 0.5, 0.5, 0.0, 0.0, 0.0, 0.0], - cls_wise_reg_weights=dict( - {9: [2.0, 2.0, 2.0, 1.0, 1.0, 1.0, 0.0, 0.0, 1.0, 1.0]})), - loss_cls=dict( - type='FocalLoss', - use_sigmoid=True, - gamma=2.0, - alpha=0.25, - loss_weight=2.0), - loss_reg=dict( - type='SparseBox3DLoss', - loss_box=dict(type='L1Loss', loss_weight=0.25), - loss_centerness=dict(type='CrossEntropyLoss', use_sigmoid=True), - loss_yawness=dict(type='GaussianFocalLoss'), - cls_allow_reverse=[5]), - decoder=dict(type='SparseBox3DDecoder'), - reg_weights=[2.0, 2.0, 2.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0])) -dataset_type = 'NuScenes3DDetTrackDataset' -data_root = 'data/nuscenes/' -anno_root = 'data/nuscenes_anno_pkls/' -file_client_args = dict(backend='disk') -img_norm_cfg = dict( - mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375], to_rgb=True) -train_pipeline = [ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict( - type='LoadPointsFromFile', - coord_type='LIDAR', - load_dim=5, - use_dim=5, - file_client_args=dict(backend='disk')), - dict(type='ResizeCropFlipImage'), - dict(type='MultiScaleDepthMapGenerator', downsample=[4, 8, 16]), - dict(type='BBoxRotation'), - dict(type='PhotoMetricDistortionMultiViewImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict( - type='CircleObjectRangeFilter', - class_dist_thred=[55, 55, 55, 55, 55, 55, 55, 55, 55, 55]), - dict( - type='InstanceNameFilter', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ]), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=[ - 'img', 'timestamp', 'projection_mat', 'image_wh', 'gt_depth', - 'focal', 'gt_bboxes_3d', 'gt_labels_3d' - ], - meta_keys=['T_global', 'T_global_inv', 'timestamp', 'instance_id']) -] -test_pipeline = [ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='ResizeCropFlipImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=['img', 'timestamp', 'projection_mat', 'image_wh'], - meta_keys=['T_global', 'T_global_inv', 'timestamp']) -] -input_modality = dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False) -data_basic_config = dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier', - 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval') -data_aug_conf = dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]) -data = dict( - samples_per_gpu=20, - workers_per_gpu=20, - train=dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval', - ann_file='data/nuscenes_anno_pkls/nuscenes_infos_train.pkl', - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict( - type='LoadPointsFromFile', - coord_type='LIDAR', - load_dim=5, - use_dim=5, - file_client_args=dict(backend='disk')), - dict(type='ResizeCropFlipImage'), - dict(type='MultiScaleDepthMapGenerator', downsample=[4, 8, 16]), - dict(type='BBoxRotation'), - dict(type='PhotoMetricDistortionMultiViewImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict( - type='CircleObjectRangeFilter', - class_dist_thred=[55, 55, 55, 55, 55, 55, 55, 55, 55, 55]), - dict( - type='InstanceNameFilter', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', - 'traffic_cone' - ]), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=[ - 'img', 'timestamp', 'projection_mat', 'image_wh', - 'gt_depth', 'focal', 'gt_bboxes_3d', 'gt_labels_3d' - ], - meta_keys=[ - 'T_global', 'T_global_inv', 'timestamp', 'instance_id' - ]) - ], - test_mode=False, - data_aug_conf=dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]), - with_seq_flag=True, - sequences_split_num=2, - keep_consistent_seq_aug=True), - val=dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval', - ann_file='data/nuscenes_anno_pkls/nuscenes_infos_val.pkl', - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='ResizeCropFlipImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=['img', 'timestamp', 'projection_mat', 'image_wh'], - meta_keys=['T_global', 'T_global_inv', 'timestamp']) - ], - data_aug_conf=dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]), - test_mode=True, - tracking=True, - tracking_threshold=0.2), - test=dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval', - ann_file='data/nuscenes_anno_pkls/nuscenes_infos_val.pkl', - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='ResizeCropFlipImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=['img', 'timestamp', 'projection_mat', 'image_wh'], - meta_keys=['T_global', 'T_global_inv', 'timestamp']) - ], - data_aug_conf=dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]), - test_mode=True, - tracking=True, - tracking_threshold=0.2)) -optimizer = dict( - type='Miopen_AdamW', - lr=0.0006, - weight_decay=0.001, - paramwise_cfg=dict(custom_keys=dict(img_backbone=dict(lr_mult=0.5)))) -optimizer_config = dict(grad_clip=dict(max_norm=25, norm_type=2)) -lr_config = dict( - policy='CosineAnnealing', - warmup='linear', - warmup_iters=500, - warmup_ratio=0.3333333333333333, - min_lr_ratio=0.001) -runner = dict(type='IterBasedRunner', max_iters=17500) -vis_pipeline = [ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='Collect', keys=['img'], meta_keys=['timestamp', 'lidar2img']) -] -evaluation = dict( - interval=3500, - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict( - type='Collect', keys=['img'], meta_keys=['timestamp', 'lidar2img']) - ]) -gpu_ids = range(0, 8) - -2025-11-13 10:48:42,142 - mmdet - INFO - Set random seed to 0, deterministic: False -2025-11-13 10:48:42,448 - mmdet - INFO - initialize ResNet with init_cfg {'type': 'Pretrained', 'checkpoint': 'ckpt/resnet50-19c8e357.pth'} -2025-11-13 10:48:42,705 - mmdet - INFO - initialize FPN with init_cfg {'type': 'Xavier', 'layer': 'Conv2d', 'distribution': 'uniform'} -Name of parameter - Initialization information - -img_backbone.conv1.weight - torch.Size([64, 3, 7, 7]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.bn1.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.bn1.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.conv1.weight - torch.Size([64, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn1.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn1.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.conv2.weight - torch.Size([64, 64, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn2.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn2.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.conv3.weight - torch.Size([256, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn3.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn3.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.downsample.0.weight - torch.Size([256, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.downsample.1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.downsample.1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.conv1.weight - torch.Size([64, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn1.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn1.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.conv2.weight - torch.Size([64, 64, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn2.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn2.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.conv3.weight - torch.Size([256, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn3.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn3.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.conv1.weight - torch.Size([64, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn1.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn1.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.conv2.weight - torch.Size([64, 64, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn2.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn2.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.conv3.weight - torch.Size([256, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn3.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn3.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.conv1.weight - torch.Size([128, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn1.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn1.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.conv2.weight - torch.Size([128, 128, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn2.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn2.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.conv3.weight - torch.Size([512, 128, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn3.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn3.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.downsample.0.weight - torch.Size([512, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.downsample.1.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.downsample.1.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.conv1.weight - torch.Size([128, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn1.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn1.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.conv2.weight - torch.Size([128, 128, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn2.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn2.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.conv3.weight - torch.Size([512, 128, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn3.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn3.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.conv1.weight - torch.Size([128, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn1.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn1.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.conv2.weight - torch.Size([128, 128, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn2.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn2.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.conv3.weight - torch.Size([512, 128, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn3.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn3.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.conv1.weight - torch.Size([128, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn1.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn1.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.conv2.weight - torch.Size([128, 128, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn2.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn2.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.conv3.weight - torch.Size([512, 128, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn3.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn3.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.conv1.weight - torch.Size([256, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.downsample.0.weight - torch.Size([1024, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.downsample.1.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.downsample.1.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.conv1.weight - torch.Size([512, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn1.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn1.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.conv2.weight - torch.Size([512, 512, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn2.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn2.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.conv3.weight - torch.Size([2048, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn3.weight - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn3.bias - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.downsample.0.weight - torch.Size([2048, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.downsample.1.weight - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.downsample.1.bias - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.conv1.weight - torch.Size([512, 2048, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn1.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn1.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.conv2.weight - torch.Size([512, 512, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn2.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn2.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.conv3.weight - torch.Size([2048, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn3.weight - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn3.bias - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.conv1.weight - torch.Size([512, 2048, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn1.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn1.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.conv2.weight - torch.Size([512, 512, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn2.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn2.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.conv3.weight - torch.Size([2048, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn3.weight - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn3.bias - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_neck.lateral_convs.0.conv.weight - torch.Size([256, 256, 1, 1]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.lateral_convs.0.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.lateral_convs.1.conv.weight - torch.Size([256, 512, 1, 1]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.lateral_convs.1.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.lateral_convs.2.conv.weight - torch.Size([256, 1024, 1, 1]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.lateral_convs.2.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.lateral_convs.3.conv.weight - torch.Size([256, 2048, 1, 1]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.lateral_convs.3.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.fpn_convs.0.conv.weight - torch.Size([256, 256, 3, 3]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.fpn_convs.0.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.fpn_convs.1.conv.weight - torch.Size([256, 256, 3, 3]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.fpn_convs.1.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.fpn_convs.2.conv.weight - torch.Size([256, 256, 3, 3]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.fpn_convs.2.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.fpn_convs.3.conv.weight - torch.Size([256, 256, 3, 3]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.fpn_convs.3.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.instance_bank.anchor - torch.Size([900, 11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.instance_bank.instance_feature - torch.Size([900, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.instance_bank.anchor_handler.fix_scale - torch.Size([1, 3]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.0.weight - torch.Size([128, 3]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.0.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.2.weight - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.2.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.3.weight - torch.Size([128, 128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.3.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.5.weight - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.5.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.6.weight - torch.Size([128, 128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.6.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.8.weight - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.8.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.9.weight - torch.Size([128, 128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.9.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.11.weight - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.11.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.0.weight - torch.Size([32, 3]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.0.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.2.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.2.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.3.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.3.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.5.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.5.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.6.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.6.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.8.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.8.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.9.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.9.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.11.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.11.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.0.weight - torch.Size([32, 2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.0.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.2.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.2.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.3.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.3.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.5.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.5.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.6.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.6.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.8.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.8.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.9.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.9.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.11.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.11.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.0.weight - torch.Size([64, 3]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.0.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.2.weight - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.2.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.3.weight - torch.Size([64, 64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.3.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.5.weight - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.5.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.6.weight - torch.Size([64, 64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.6.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.8.weight - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.8.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.9.weight - torch.Size([64, 64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.9.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.11.weight - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.11.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.1.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.1.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.1.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.1.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.1.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.1.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.1.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.1.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.3.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.4.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.4.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.4.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.4.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.5.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.5.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.5.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.5.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.6.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.6.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.8.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.8.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.8.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.8.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.8.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.8.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.8.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.8.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.10.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.11.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.11.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.11.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.11.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.12.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.12.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.12.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.12.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.13.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.13.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.15.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.15.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.15.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.15.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.15.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.15.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.15.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.15.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.16.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.16.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.17.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.18.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.18.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.18.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.18.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.19.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.19.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.19.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.19.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.20.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.20.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.22.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.22.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.22.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.22.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.22.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.22.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.22.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.22.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.23.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.23.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.24.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.25.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.25.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.25.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.25.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.26.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.26.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.26.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.26.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.27.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.27.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.29.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.29.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.29.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.29.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.29.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.29.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.29.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.29.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.30.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.30.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.31.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.32.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.32.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.32.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.32.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.33.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.33.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.33.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.33.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.34.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.34.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.36.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.36.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.36.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.36.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.36.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.36.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.36.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.36.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.37.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.37.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.38.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.fc_before.weight - torch.Size([512, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.fc_after.weight - torch.Size([256, 512]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.0.weight - torch.Size([1, 256, 1, 1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.0.bias - torch.Size([1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.1.weight - torch.Size([1, 256, 1, 1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.1.bias - torch.Size([1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.2.weight - torch.Size([1, 256, 1, 1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.2.bias - torch.Size([1]): -The value is the same before and after calling `init_weights` of Sparse4D -2025-11-13 10:48:42,797 - mmdet - INFO - Model: -Sparse4D( - (img_backbone): ResNet( - (conv1): Conv2d(3, 64, kernel_size=(7, 7), stride=(2, 2), padding=(3, 3), bias=False) - (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (maxpool): MaxPool2d(kernel_size=3, stride=2, padding=1, dilation=1, ceil_mode=False) - (layer1): ResLayer( - (0): Bottleneck( - (conv1): Conv2d(64, 64, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (downsample): Sequential( - (0): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - ) - ) - (1): Bottleneck( - (conv1): Conv2d(256, 64, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (2): Bottleneck( - (conv1): Conv2d(256, 64, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - ) - (layer2): ResLayer( - (0): Bottleneck( - (conv1): Conv2d(256, 128, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (downsample): Sequential( - (0): Conv2d(256, 512, kernel_size=(1, 1), stride=(2, 2), bias=False) - (1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - ) - ) - (1): Bottleneck( - (conv1): Conv2d(512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (2): Bottleneck( - (conv1): Conv2d(512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (3): Bottleneck( - (conv1): Conv2d(512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - ) - (layer3): ResLayer( - (0): Bottleneck( - (conv1): Conv2d(512, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (downsample): Sequential( - (0): Conv2d(512, 1024, kernel_size=(1, 1), stride=(2, 2), bias=False) - (1): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - ) - ) - (1): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (2): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (3): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (4): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (5): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - ) - (layer4): ResLayer( - (0): Bottleneck( - (conv1): Conv2d(1024, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(512, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (downsample): Sequential( - (0): Conv2d(1024, 2048, kernel_size=(1, 1), stride=(2, 2), bias=False) - (1): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - ) - ) - (1): Bottleneck( - (conv1): Conv2d(2048, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(512, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (2): Bottleneck( - (conv1): Conv2d(2048, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(512, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - ) - ) - init_cfg={'type': 'Pretrained', 'checkpoint': 'ckpt/resnet50-19c8e357.pth'} - (img_neck): FPN( - (lateral_convs): ModuleList( - (0): ConvModule( - (conv): Conv2d(256, 256, kernel_size=(1, 1), stride=(1, 1)) - ) - (1): ConvModule( - (conv): Conv2d(512, 256, kernel_size=(1, 1), stride=(1, 1)) - ) - (2): ConvModule( - (conv): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1)) - ) - (3): ConvModule( - (conv): Conv2d(2048, 256, kernel_size=(1, 1), stride=(1, 1)) - ) - ) - (fpn_convs): ModuleList( - (0-3): 4 x ConvModule( - (conv): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1)) - ) - ) - ) - init_cfg={'type': 'Xavier', 'layer': 'Conv2d', 'distribution': 'uniform'} - (head): Sparse4DHead( - (instance_bank): InstanceBank( - (anchor_handler): SparseBox3DKeyPointsGenerator() - ) - (anchor_encoder): SparseBox3DEncoder( - (pos_fc): Sequential( - (0): Linear(in_features=3, out_features=128, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((128,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=128, out_features=128, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((128,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=128, out_features=128, bias=True) - (7): ReLU(inplace=True) - (8): LayerNorm((128,), eps=1e-05, elementwise_affine=True) - (9): Linear(in_features=128, out_features=128, bias=True) - (10): ReLU(inplace=True) - (11): LayerNorm((128,), eps=1e-05, elementwise_affine=True) - ) - (size_fc): Sequential( - (0): Linear(in_features=3, out_features=32, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=32, out_features=32, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=32, out_features=32, bias=True) - (7): ReLU(inplace=True) - (8): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (9): Linear(in_features=32, out_features=32, bias=True) - (10): ReLU(inplace=True) - (11): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - ) - (yaw_fc): Sequential( - (0): Linear(in_features=2, out_features=32, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=32, out_features=32, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=32, out_features=32, bias=True) - (7): ReLU(inplace=True) - (8): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (9): Linear(in_features=32, out_features=32, bias=True) - (10): ReLU(inplace=True) - (11): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - ) - (vel_fc): Sequential( - (0): Linear(in_features=3, out_features=64, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((64,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=64, out_features=64, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((64,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=64, out_features=64, bias=True) - (7): ReLU(inplace=True) - (8): LayerNorm((64,), eps=1e-05, elementwise_affine=True) - (9): Linear(in_features=64, out_features=64, bias=True) - (10): ReLU(inplace=True) - (11): LayerNorm((64,), eps=1e-05, elementwise_affine=True) - ) - ) - (loss_cls): FocalLoss() - (loss_reg): SparseBox3DLoss( - (loss_box): L1Loss() - (loss_cns): CrossEntropyLoss(avg_non_ignore=False) - (loss_yns): GaussianFocalLoss() - ) - (layers): ModuleList( - (0): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (1): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (4-5): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (6): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (7): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (8): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (11-12): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (13): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (14): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (15): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (16): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (17): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (18-19): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (20): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (21): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (22): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (23): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (24): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (25-26): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (27): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (28): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (29): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (30): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (31): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (32-33): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (34): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (35): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (36): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (37): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (38): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - ) - (fc_before): Linear(in_features=256, out_features=512, bias=False) - (fc_after): Linear(in_features=512, out_features=256, bias=False) - ) - (depth_branch): DenseDepthNet( - (depth_layers): ModuleList( - (0-2): 3 x Conv2d(256, 1, kernel_size=(1, 1), stride=(1, 1)) - ) - ) - (grid_mask): GridMask() -) -2025-11-13 10:48:55,290 - mmdet - INFO - Start running, host: root@VM-120-96-tencentos, work_dir: /cfs/auto/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704 -2025-11-13 10:48:55,291 - mmdet - INFO - Hooks will be executed in the following order: -before_run: -(VERY_HIGH ) CosineAnnealingLrUpdaterHook -(ABOVE_NORMAL) Fp16OptimizerHook -(NORMAL ) CheckpointHook -(NORMAL ) CustomDistEvalHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -before_train_epoch: -(VERY_HIGH ) CosineAnnealingLrUpdaterHook -(NORMAL ) CustomDistEvalHook -(LOW ) IterTimerHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -before_train_iter: -(VERY_HIGH ) CosineAnnealingLrUpdaterHook -(NORMAL ) CustomDistEvalHook -(LOW ) IterTimerHook - -------------------- -after_train_iter: -(ABOVE_NORMAL) Fp16OptimizerHook -(NORMAL ) CheckpointHook -(NORMAL ) CustomDistEvalHook -(NORMAL ) ProfilerHook -(LOW ) IterTimerHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -after_train_epoch: -(NORMAL ) CheckpointHook -(NORMAL ) CustomDistEvalHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -before_val_epoch: -(LOW ) IterTimerHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -before_val_iter: -(LOW ) IterTimerHook - -------------------- -after_val_iter: -(LOW ) IterTimerHook - -------------------- -after_val_epoch: -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -after_run: -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -2025-11-13 10:48:55,291 - mmdet - INFO - workflow: [('train', 1)], max: 17500 iters -2025-11-13 10:48:55,293 - mmdet - INFO - Checkpoints will be saved to /cfs/auto/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704 by HardDiskBackend. -2025-11-13 10:50:51,791 - mmdet - INFO - Iter [1/17500] lr: 1.000e-04, eta: 23 days, 9:46:15, time: 115.571, data_time: 10.387, memory: 49163, loss_cls_0: 2.3613, loss_box_0: 0.0138, loss_cns_0: 0.0027, loss_yns_0: 0.0008, loss_cls_1: 2.1544, loss_box_1: 0.1089, loss_cns_1: 0.0249, loss_yns_1: 0.0067, loss_cls_2: 2.3119, loss_box_2: 0.0050, loss_cns_2: 0.0006, loss_yns_2: 0.0003, loss_cls_3: 2.3903, loss_box_3: 0.0294, loss_cns_3: 0.0050, loss_yns_3: 0.0014, loss_cls_4: 2.0275, loss_box_4: 0.4213, loss_cns_4: 0.0542, loss_yns_4: 0.0257, loss_cls_5: 2.4248, loss_box_5: 0.0180, loss_cns_5: 0.0022, loss_yns_5: 0.0016, loss_cls_dn_0: 1.1980, loss_box_dn_0: 1.4603, loss_cls_dn_1: 1.1102, loss_box_dn_1: 1.7318, loss_cls_dn_2: 1.1741, loss_box_dn_2: 1.9719, loss_cls_dn_3: 1.1721, loss_box_dn_3: 2.2418, loss_cls_dn_4: 1.0528, loss_box_dn_4: 2.4269, loss_cls_dn_5: 1.2387, loss_box_dn_5: 2.6773, loss_dense_depth: 1.8643, loss: 35.7130, grad_norm: 271.3412 -2025-11-13 10:50:53,677 - mmdet - INFO - Iter [2/17500] lr: 1.004e-04, eta: 11 days, 21:27:07, time: 1.886, data_time: 0.068, memory: 49163, loss_cls_0: 2.0900, loss_box_0: 0.0102, loss_cns_0: 0.0027, loss_yns_0: 0.0009, loss_cls_1: 2.0373, loss_box_1: 0.1042, loss_cns_1: 0.0203, loss_yns_1: 0.0051, loss_cls_2: 2.1145, loss_box_2: 0.2272, loss_cns_2: 0.0206, loss_yns_2: 0.0094, loss_cls_3: 1.9607, loss_box_3: 0.4292, loss_cns_3: 0.0549, loss_yns_3: 0.0192, loss_cls_4: 1.8013, loss_box_4: 1.5444, loss_cns_4: 0.1539, loss_yns_4: 0.0561, loss_cls_5: 2.0597, loss_box_5: 0.5300, loss_cns_5: 0.0582, loss_yns_5: 0.0190, loss_cls_dn_0: 1.0482, loss_box_dn_0: 1.3109, loss_cls_dn_1: 0.9594, loss_box_dn_1: 2.4244, loss_cls_dn_2: 0.9745, loss_box_dn_2: 2.5379, loss_cls_dn_3: 0.9134, loss_box_dn_3: 2.6210, loss_cls_dn_4: 0.8410, loss_box_dn_4: 2.8801, loss_cls_dn_5: 0.9862, loss_box_dn_5: 3.1205, loss_dense_depth: 1.7123, loss: 37.6586, grad_norm: 66.1444 -2025-11-13 10:50:55,302 - mmdet - INFO - Iter [3/17500] lr: 1.008e-04, eta: 8 days, 0:55:15, time: 1.624, data_time: 0.080, memory: 49163, loss_cls_0: 1.4182, loss_box_0: 2.5811, loss_cns_0: 0.6100, loss_yns_0: 0.2379, loss_cls_1: 1.7319, loss_box_1: 2.1585, loss_cns_1: 0.3113, loss_yns_1: 0.1191, loss_cls_2: 1.7541, loss_box_2: 4.4303, loss_cns_2: 0.3755, loss_yns_2: 0.2147, loss_cls_3: 1.6043, loss_box_3: 5.2067, loss_cns_3: 0.4302, loss_yns_3: 0.2128, loss_cls_4: 1.5240, loss_box_4: 4.9937, loss_cns_4: 0.4127, loss_yns_4: 0.1979, loss_cls_5: 1.6218, loss_box_5: 4.0515, loss_cns_5: 0.2919, loss_yns_5: 0.1300, loss_cls_dn_0: 0.6716, loss_box_dn_0: 1.1847, loss_cls_dn_1: 0.8076, loss_box_dn_1: 2.4520, loss_cls_dn_2: 0.7705, loss_box_dn_2: 2.6752, loss_cls_dn_3: 0.6736, loss_box_dn_3: 2.8808, loss_cls_dn_4: 0.6894, loss_box_dn_4: 3.1615, loss_cls_dn_5: 0.7764, loss_box_dn_5: 3.4483, loss_dense_depth: 1.6223, loss: 58.4339, grad_norm: 105.2519 -2025-11-13 10:50:56,827 - mmdet - INFO - Iter [4/17500] lr: 1.012e-04, eta: 6 days, 2:32:07, time: 1.525, data_time: 0.078, memory: 49163, loss_cls_0: 1.4022, loss_box_0: 2.7160, loss_cns_0: 0.5235, loss_yns_0: 0.2210, loss_cls_1: 1.5669, loss_box_1: 3.6494, loss_cns_1: 0.4607, loss_yns_1: 0.2172, loss_cls_2: 1.7355, loss_box_2: 3.7819, loss_cns_2: 0.4450, loss_yns_2: 0.2018, loss_cls_3: 1.4723, loss_box_3: 4.2721, loss_cns_3: 0.4693, loss_yns_3: 0.1982, loss_cls_4: 1.5360, loss_box_4: 4.6699, loss_cns_4: 0.4215, loss_yns_4: 0.1984, loss_cls_5: 1.3972, loss_box_5: 5.1989, loss_cns_5: 0.4438, loss_yns_5: 0.1912, loss_cls_dn_0: 0.5293, loss_box_dn_0: 1.2393, loss_cls_dn_1: 0.6709, loss_box_dn_1: 2.6800, loss_cls_dn_2: 0.6413, loss_box_dn_2: 2.7869, loss_cls_dn_3: 0.5750, loss_box_dn_3: 3.0409, loss_cls_dn_4: 0.5365, loss_box_dn_4: 3.2523, loss_cls_dn_5: 0.5951, loss_box_dn_5: 3.5115, loss_dense_depth: 1.6985, loss: 59.1476, grad_norm: 120.4662 -2025-11-13 10:50:58,375 - mmdet - INFO - Iter [5/17500] lr: 1.016e-04, eta: 4 days, 22:43:33, time: 1.548, data_time: 0.072, memory: 49163, loss_cls_0: 1.4064, loss_box_0: 2.7322, loss_cns_0: 0.5357, loss_yns_0: 0.2204, loss_cls_1: 1.5113, loss_box_1: 4.1121, loss_cns_1: 0.4204, loss_yns_1: 0.2346, loss_cls_2: 1.4997, loss_box_2: 4.2147, loss_cns_2: 0.3986, loss_yns_2: 0.1872, loss_cls_3: 1.3518, loss_box_3: 4.3599, loss_cns_3: 0.4020, loss_yns_3: 0.1909, loss_cls_4: 1.3275, loss_box_4: 4.6983, loss_cns_4: 0.3668, loss_yns_4: 0.2048, loss_cls_5: 1.3590, loss_box_5: 4.8862, loss_cns_5: 0.3469, loss_yns_5: 0.2041, loss_cls_dn_0: 0.5533, loss_box_dn_0: 1.2579, loss_cls_dn_1: 0.5919, loss_box_dn_1: 2.3890, loss_cls_dn_2: 0.6497, loss_box_dn_2: 2.5744, loss_cls_dn_3: 0.5336, loss_box_dn_3: 2.6802, loss_cls_dn_4: 0.5346, loss_box_dn_4: 2.9078, loss_cls_dn_5: 0.4954, loss_box_dn_5: 2.9207, loss_dense_depth: 1.7784, loss: 57.0386, grad_norm: 124.6013 -2025-11-13 10:50:59,944 - mmdet - INFO - Iter [6/17500] lr: 1.020e-04, eta: 4 days, 4:12:12, time: 1.569, data_time: 0.078, memory: 49163, loss_cls_0: 1.2691, loss_box_0: 2.3441, loss_cns_0: 0.6616, loss_yns_0: 0.1812, loss_cls_1: 1.3245, loss_box_1: 3.7259, loss_cns_1: 0.4490, loss_yns_1: 0.1973, loss_cls_2: 1.3757, loss_box_2: 3.8435, loss_cns_2: 0.4525, loss_yns_2: 0.1971, loss_cls_3: 1.2978, loss_box_3: 3.6684, loss_cns_3: 0.4791, loss_yns_3: 0.2091, loss_cls_4: 1.3122, loss_box_4: 3.9393, loss_cns_4: 0.4516, loss_yns_4: 0.1982, loss_cls_5: 1.3488, loss_box_5: 4.3078, loss_cns_5: 0.4087, loss_yns_5: 0.1888, loss_cls_dn_0: 0.5653, loss_box_dn_0: 1.1328, loss_cls_dn_1: 0.5228, loss_box_dn_1: 2.5266, loss_cls_dn_2: 0.5704, loss_box_dn_2: 2.5205, loss_cls_dn_3: 0.4778, loss_box_dn_3: 2.5446, loss_cls_dn_4: 0.4885, loss_box_dn_4: 2.7416, loss_cls_dn_5: 0.4368, loss_box_dn_5: 2.8730, loss_dense_depth: 2.0182, loss: 53.2502, grad_norm: 106.4340 -2025-11-13 10:51:01,456 - mmdet - INFO - Iter [7/17500] lr: 1.024e-04, eta: 3 days, 14:56:01, time: 1.513, data_time: 0.082, memory: 49163, loss_cls_0: 1.2454, loss_box_0: 2.2830, loss_cns_0: 0.6218, loss_yns_0: 0.1963, loss_cls_1: 1.2698, loss_box_1: 3.6721, loss_cns_1: 0.4250, loss_yns_1: 0.1850, loss_cls_2: 1.4273, loss_box_2: 3.8404, loss_cns_2: 0.3712, loss_yns_2: 0.1854, loss_cls_3: 1.2820, loss_box_3: 3.6974, loss_cns_3: 0.3861, loss_yns_3: 0.1831, loss_cls_4: 1.3090, loss_box_4: 3.7493, loss_cns_4: 0.3988, loss_yns_4: 0.2022, loss_cls_5: 1.3276, loss_box_5: 3.8927, loss_cns_5: 0.4430, loss_yns_5: 0.1842, loss_cls_dn_0: 0.5441, loss_box_dn_0: 1.0490, loss_cls_dn_1: 0.4558, loss_box_dn_1: 2.3741, loss_cls_dn_2: 0.4744, loss_box_dn_2: 2.3715, loss_cls_dn_3: 0.4287, loss_box_dn_3: 2.3307, loss_cls_dn_4: 0.4371, loss_box_dn_4: 2.3735, loss_cls_dn_5: 0.3992, loss_box_dn_5: 2.4605, loss_dense_depth: 1.7117, loss: 50.1882, grad_norm: 89.6524 -2025-11-13 10:51:02,975 - mmdet - INFO - Iter [8/17500] lr: 1.028e-04, eta: 3 days, 4:59:04, time: 1.518, data_time: 0.078, memory: 49163, loss_cls_0: 1.3159, loss_box_0: 2.1988, loss_cns_0: 0.6176, loss_yns_0: 0.1781, loss_cls_1: 1.2563, loss_box_1: 3.3025, loss_cns_1: 0.5097, loss_yns_1: 0.1803, loss_cls_2: 1.3170, loss_box_2: 3.3727, loss_cns_2: 0.4602, loss_yns_2: 0.1830, loss_cls_3: 1.2608, loss_box_3: 3.4652, loss_cns_3: 0.5008, loss_yns_3: 0.2158, loss_cls_4: 1.2611, loss_box_4: 3.4037, loss_cns_4: 0.4878, loss_yns_4: 0.1928, loss_cls_5: 1.2833, loss_box_5: 3.4675, loss_cns_5: 0.5083, loss_yns_5: 0.1819, loss_cls_dn_0: 0.4833, loss_box_dn_0: 1.0423, loss_cls_dn_1: 0.4498, loss_box_dn_1: 1.5895, loss_cls_dn_2: 0.4948, loss_box_dn_2: 1.6261, loss_cls_dn_3: 0.4453, loss_box_dn_3: 1.7564, loss_cls_dn_4: 0.4540, loss_box_dn_4: 1.7770, loss_cls_dn_5: 0.4415, loss_box_dn_5: 1.9066, loss_dense_depth: 2.6516, loss: 46.2393, grad_norm: 89.8605 -2025-11-13 10:51:04,502 - mmdet - INFO - Iter [9/17500] lr: 1.032e-04, eta: 2 days, 21:15:02, time: 1.526, data_time: 0.080, memory: 49163, loss_cls_0: 1.2200, loss_box_0: 2.3047, loss_cns_0: 0.6283, loss_yns_0: 0.1810, loss_cls_1: 1.2736, loss_box_1: 3.1716, loss_cns_1: 0.4992, loss_yns_1: 0.1786, loss_cls_2: 1.2924, loss_box_2: 3.1391, loss_cns_2: 0.4926, loss_yns_2: 0.1882, loss_cls_3: 1.2639, loss_box_3: 3.4129, loss_cns_3: 0.4980, loss_yns_3: 0.1820, loss_cls_4: 1.2709, loss_box_4: 3.3573, loss_cns_4: 0.4968, loss_yns_4: 0.1820, loss_cls_5: 1.2566, loss_box_5: 3.5793, loss_cns_5: 0.4730, loss_yns_5: 0.1844, loss_cls_dn_0: 0.4937, loss_box_dn_0: 1.0833, loss_cls_dn_1: 0.4179, loss_box_dn_1: 1.5916, loss_cls_dn_2: 0.4765, loss_box_dn_2: 1.6233, loss_cls_dn_3: 0.4313, loss_box_dn_3: 1.8892, loss_cls_dn_4: 0.4323, loss_box_dn_4: 1.9222, loss_cls_dn_5: 0.4585, loss_box_dn_5: 2.1507, loss_dense_depth: 2.1237, loss: 45.8204, grad_norm: 86.9061 -2025-11-13 10:51:06,027 - mmdet - INFO - Iter [10/17500] lr: 1.036e-04, eta: 2 days, 15:03:49, time: 1.526, data_time: 0.091, memory: 49163, loss_cls_0: 1.2441, loss_box_0: 2.3086, loss_cns_0: 0.6028, loss_yns_0: 0.1778, loss_cls_1: 1.2516, loss_box_1: 3.1771, loss_cns_1: 0.4686, loss_yns_1: 0.1768, loss_cls_2: 1.2780, loss_box_2: 3.0368, loss_cns_2: 0.4716, loss_yns_2: 0.1810, loss_cls_3: 1.2258, loss_box_3: 3.0905, loss_cns_3: 0.4798, loss_yns_3: 0.1913, loss_cls_4: 1.2596, loss_box_4: 3.0744, loss_cns_4: 0.4878, loss_yns_4: 0.2034, loss_cls_5: 1.2446, loss_box_5: 3.3420, loss_cns_5: 0.4790, loss_yns_5: 0.1814, loss_cls_dn_0: 0.4890, loss_box_dn_0: 1.0820, loss_cls_dn_1: 0.4029, loss_box_dn_1: 1.8343, loss_cls_dn_2: 0.4399, loss_box_dn_2: 1.7838, loss_cls_dn_3: 0.4241, loss_box_dn_3: 1.8979, loss_cls_dn_4: 0.4066, loss_box_dn_4: 1.9061, loss_cls_dn_5: 0.4354, loss_box_dn_5: 2.0909, loss_dense_depth: 2.3932, loss: 45.2206, grad_norm: 69.2462 -2025-11-13 10:51:07,555 - mmdet - INFO - Iter [11/17500] lr: 1.040e-04, eta: 2 days, 10:00:06, time: 1.527, data_time: 0.077, memory: 49163, loss_cls_0: 1.2266, loss_box_0: 2.2797, loss_cns_0: 0.5975, loss_yns_0: 0.1726, loss_cls_1: 1.2343, loss_box_1: 2.8842, loss_cns_1: 0.5224, loss_yns_1: 0.1774, loss_cls_2: 1.2441, loss_box_2: 2.9626, loss_cns_2: 0.5260, loss_yns_2: 0.1854, loss_cls_3: 1.2618, loss_box_3: 3.0106, loss_cns_3: 0.5267, loss_yns_3: 0.1739, loss_cls_4: 1.2736, loss_box_4: 3.0564, loss_cns_4: 0.4905, loss_yns_4: 0.1808, loss_cls_5: 1.2846, loss_box_5: 3.0689, loss_cns_5: 0.5038, loss_yns_5: 0.1766, loss_cls_dn_0: 0.4630, loss_box_dn_0: 1.0657, loss_cls_dn_1: 0.4003, loss_box_dn_1: 1.7705, loss_cls_dn_2: 0.4175, loss_box_dn_2: 1.7621, loss_cls_dn_3: 0.4064, loss_box_dn_3: 1.8353, loss_cls_dn_4: 0.3935, loss_box_dn_4: 1.9520, loss_cls_dn_5: 0.4024, loss_box_dn_5: 1.9803, loss_dense_depth: 2.4526, loss: 44.3228, grad_norm: 74.0453 -2025-11-13 10:51:09,074 - mmdet - INFO - Iter [12/17500] lr: 1.044e-04, eta: 2 days, 5:46:51, time: 1.521, data_time: 0.081, memory: 49163, loss_cls_0: 1.2337, loss_box_0: 2.2271, loss_cns_0: 0.6095, loss_yns_0: 0.1755, loss_cls_1: 1.2421, loss_box_1: 2.8049, loss_cns_1: 0.5428, loss_yns_1: 0.1741, loss_cls_2: 1.2728, loss_box_2: 3.0144, loss_cns_2: 0.5371, loss_yns_2: 0.1777, loss_cls_3: 1.2577, loss_box_3: 3.1014, loss_cns_3: 0.5233, loss_yns_3: 0.1841, loss_cls_4: 1.2601, loss_box_4: 3.0602, loss_cns_4: 0.5028, loss_yns_4: 0.2078, loss_cls_5: 1.2808, loss_box_5: 3.2005, loss_cns_5: 0.4887, loss_yns_5: 0.1791, loss_cls_dn_0: 0.4501, loss_box_dn_0: 1.0433, loss_cls_dn_1: 0.3848, loss_box_dn_1: 1.7897, loss_cls_dn_2: 0.3920, loss_box_dn_2: 1.8391, loss_cls_dn_3: 0.3689, loss_box_dn_3: 1.8744, loss_cls_dn_4: 0.3845, loss_box_dn_4: 1.9621, loss_cls_dn_5: 0.3857, loss_box_dn_5: 2.0122, loss_dense_depth: 2.4308, loss: 44.5759, grad_norm: 82.5912 -2025-11-13 10:51:10,632 - mmdet - INFO - Iter [13/17500] lr: 1.048e-04, eta: 2 days, 2:13:12, time: 1.550, data_time: 0.081, memory: 49163, loss_cls_0: 1.1723, loss_box_0: 2.3456, loss_cns_0: 0.5848, loss_yns_0: 0.1735, loss_cls_1: 1.2267, loss_box_1: 2.9209, loss_cns_1: 0.5452, loss_yns_1: 0.1724, loss_cls_2: 1.2517, loss_box_2: 3.0752, loss_cns_2: 0.5375, loss_yns_2: 0.1734, loss_cls_3: 1.2518, loss_box_3: 3.1146, loss_cns_3: 0.5498, loss_yns_3: 0.1795, loss_cls_4: 1.2404, loss_box_4: 3.2219, loss_cns_4: 0.5298, loss_yns_4: 0.1833, loss_cls_5: 1.2420, loss_box_5: 3.2253, loss_cns_5: 0.5391, loss_yns_5: 0.1759, loss_cls_dn_0: 0.4670, loss_box_dn_0: 1.0596, loss_cls_dn_1: 0.4283, loss_box_dn_1: 1.4861, loss_cls_dn_2: 0.4377, loss_box_dn_2: 1.6043, loss_cls_dn_3: 0.4144, loss_box_dn_3: 1.6621, loss_cls_dn_4: 0.4438, loss_box_dn_4: 1.8122, loss_cls_dn_5: 0.4493, loss_box_dn_5: 1.8844, loss_dense_depth: 2.3335, loss: 44.1154, grad_norm: 107.1947 -2025-11-13 10:51:12,151 - mmdet - INFO - Iter [14/17500] lr: 1.052e-04, eta: 1 day, 23:09:35, time: 1.526, data_time: 0.081, memory: 49163, loss_cls_0: 1.1726, loss_box_0: 2.4337, loss_cns_0: 0.5807, loss_yns_0: 0.1744, loss_cls_1: 1.2268, loss_box_1: 2.8979, loss_cns_1: 0.5310, loss_yns_1: 0.1756, loss_cls_2: 1.2557, loss_box_2: 3.0320, loss_cns_2: 0.5084, loss_yns_2: 0.1772, loss_cls_3: 1.2613, loss_box_3: 3.0626, loss_cns_3: 0.5100, loss_yns_3: 0.1745, loss_cls_4: 1.2620, loss_box_4: 3.1273, loss_cns_4: 0.4986, loss_yns_4: 0.1799, loss_cls_5: 1.2749, loss_box_5: 3.1921, loss_cns_5: 0.5037, loss_yns_5: 0.1742, loss_cls_dn_0: 0.4823, loss_box_dn_0: 1.0846, loss_cls_dn_1: 0.4470, loss_box_dn_1: 1.4008, loss_cls_dn_2: 0.4578, loss_box_dn_2: 1.5233, loss_cls_dn_3: 0.4570, loss_box_dn_3: 1.6049, loss_cls_dn_4: 0.4725, loss_box_dn_4: 1.7198, loss_cls_dn_5: 0.4791, loss_box_dn_5: 1.8439, loss_dense_depth: 2.2152, loss: 43.5752, grad_norm: 104.9707 -2025-11-13 10:51:13,669 - mmdet - INFO - Iter [15/17500] lr: 1.056e-04, eta: 1 day, 20:30:14, time: 1.515, data_time: 0.076, memory: 49163, loss_cls_0: 1.1849, loss_box_0: 2.4341, loss_cns_0: 0.5998, loss_yns_0: 0.1806, loss_cls_1: 1.2320, loss_box_1: 2.7131, loss_cns_1: 0.5594, loss_yns_1: 0.1806, loss_cls_2: 1.2722, loss_box_2: 2.7978, loss_cns_2: 0.5600, loss_yns_2: 0.1840, loss_cls_3: 1.2629, loss_box_3: 2.8538, loss_cns_3: 0.5688, loss_yns_3: 0.1798, loss_cls_4: 1.2796, loss_box_4: 2.8516, loss_cns_4: 0.5619, loss_yns_4: 0.1872, loss_cls_5: 1.2858, loss_box_5: 2.8595, loss_cns_5: 0.5662, loss_yns_5: 0.1874, loss_cls_dn_0: 0.4680, loss_box_dn_0: 1.1021, loss_cls_dn_1: 0.4459, loss_box_dn_1: 1.4226, loss_cls_dn_2: 0.4440, loss_box_dn_2: 1.5077, loss_cls_dn_3: 0.4551, loss_box_dn_3: 1.5769, loss_cls_dn_4: 0.4519, loss_box_dn_4: 1.6111, loss_cls_dn_5: 0.4703, loss_box_dn_5: 1.7323, loss_dense_depth: 2.0984, loss: 42.3294, grad_norm: 81.3708 -2025-11-13 10:51:15,188 - mmdet - INFO - Iter [16/17500] lr: 1.060e-04, eta: 1 day, 18:10:54, time: 1.521, data_time: 0.079, memory: 49163, loss_cls_0: 1.1925, loss_box_0: 2.3173, loss_cns_0: 0.6035, loss_yns_0: 0.1758, loss_cls_1: 1.2354, loss_box_1: 2.6750, loss_cns_1: 0.5852, loss_yns_1: 0.1785, loss_cls_2: 1.2697, loss_box_2: 2.7144, loss_cns_2: 0.5606, loss_yns_2: 0.1794, loss_cls_3: 1.2542, loss_box_3: 2.7140, loss_cns_3: 0.5573, loss_yns_3: 0.1826, loss_cls_4: 1.2676, loss_box_4: 2.6900, loss_cns_4: 0.5478, loss_yns_4: 0.1752, loss_cls_5: 1.2468, loss_box_5: 2.8814, loss_cns_5: 0.5315, loss_yns_5: 0.1742, loss_cls_dn_0: 0.4758, loss_box_dn_0: 1.0615, loss_cls_dn_1: 0.4553, loss_box_dn_1: 1.3886, loss_cls_dn_2: 0.4353, loss_box_dn_2: 1.4681, loss_cls_dn_3: 0.4378, loss_box_dn_3: 1.5364, loss_cls_dn_4: 0.4438, loss_box_dn_4: 1.5314, loss_cls_dn_5: 0.4533, loss_box_dn_5: 1.7349, loss_dense_depth: 2.1296, loss: 41.4617, grad_norm: 83.2742 -2025-11-13 10:51:16,713 - mmdet - INFO - Iter [17/17500] lr: 1.064e-04, eta: 1 day, 16:08:03, time: 1.526, data_time: 0.079, memory: 49163, loss_cls_0: 1.1745, loss_box_0: 2.1966, loss_cns_0: 0.6099, loss_yns_0: 0.1724, loss_cls_1: 1.2263, loss_box_1: 2.6704, loss_cns_1: 0.5819, loss_yns_1: 0.1794, loss_cls_2: 1.2731, loss_box_2: 2.7330, loss_cns_2: 0.5519, loss_yns_2: 0.1765, loss_cls_3: 1.3303, loss_box_3: 2.7622, loss_cns_3: 0.5471, loss_yns_3: 0.1768, loss_cls_4: 1.2834, loss_box_4: 2.7774, loss_cns_4: 0.5484, loss_yns_4: 0.1801, loss_cls_5: 1.2747, loss_box_5: 2.8777, loss_cns_5: 0.5433, loss_yns_5: 0.1809, loss_cls_dn_0: 0.4564, loss_box_dn_0: 1.0232, loss_cls_dn_1: 0.4347, loss_box_dn_1: 1.4402, loss_cls_dn_2: 0.4121, loss_box_dn_2: 1.4948, loss_cls_dn_3: 0.4028, loss_box_dn_3: 1.5840, loss_cls_dn_4: 0.4136, loss_box_dn_4: 1.5678, loss_cls_dn_5: 0.4196, loss_box_dn_5: 1.7040, loss_dense_depth: 1.9324, loss: 41.3139, grad_norm: 75.4876 -2025-11-13 10:51:18,237 - mmdet - INFO - Iter [18/17500] lr: 1.068e-04, eta: 1 day, 14:18:47, time: 1.523, data_time: 0.078, memory: 49163, loss_cls_0: 1.1619, loss_box_0: 2.1464, loss_cns_0: 0.6153, loss_yns_0: 0.1746, loss_cls_1: 1.2407, loss_box_1: 2.5975, loss_cns_1: 0.5785, loss_yns_1: 0.1782, loss_cls_2: 1.2710, loss_box_2: 2.6637, loss_cns_2: 0.5725, loss_yns_2: 0.1756, loss_cls_3: 1.2548, loss_box_3: 2.6447, loss_cns_3: 0.5819, loss_yns_3: 0.1796, loss_cls_4: 1.2834, loss_box_4: 2.6968, loss_cns_4: 0.5762, loss_yns_4: 0.1753, loss_cls_5: 1.3002, loss_box_5: 2.6893, loss_cns_5: 0.5950, loss_yns_5: 0.1772, loss_cls_dn_0: 0.4732, loss_box_dn_0: 1.0262, loss_cls_dn_1: 0.4116, loss_box_dn_1: 1.3516, loss_cls_dn_2: 0.4092, loss_box_dn_2: 1.3608, loss_cls_dn_3: 0.4205, loss_box_dn_3: 1.4523, loss_cls_dn_4: 0.4135, loss_box_dn_4: 1.4731, loss_cls_dn_5: 0.4096, loss_box_dn_5: 1.5290, loss_dense_depth: 1.8190, loss: 40.0799, grad_norm: 59.8329 -2025-11-13 10:51:19,756 - mmdet - INFO - Iter [19/17500] lr: 1.072e-04, eta: 1 day, 12:41:00, time: 1.521, data_time: 0.082, memory: 49163, loss_cls_0: 1.1580, loss_box_0: 2.1780, loss_cns_0: 0.6087, loss_yns_0: 0.1697, loss_cls_1: 1.2528, loss_box_1: 2.6160, loss_cns_1: 0.5887, loss_yns_1: 0.1770, loss_cls_2: 1.2699, loss_box_2: 2.6121, loss_cns_2: 0.5837, loss_yns_2: 0.1774, loss_cls_3: 1.2543, loss_box_3: 2.6818, loss_cns_3: 0.5697, loss_yns_3: 0.1749, loss_cls_4: 1.2453, loss_box_4: 2.9931, loss_cns_4: 0.5411, loss_yns_4: 0.1714, loss_cls_5: 1.2690, loss_box_5: 3.1341, loss_cns_5: 0.4825, loss_yns_5: 0.1752, loss_cls_dn_0: 0.4785, loss_box_dn_0: 1.0503, loss_cls_dn_1: 0.4263, loss_box_dn_1: 1.1680, loss_cls_dn_2: 0.4309, loss_box_dn_2: 1.1453, loss_cls_dn_3: 0.4628, loss_box_dn_3: 1.2510, loss_cls_dn_4: 0.4626, loss_box_dn_4: 1.4586, loss_cls_dn_5: 0.4397, loss_box_dn_5: 1.5492, loss_dense_depth: 1.7322, loss: 40.1398, grad_norm: 83.7490 -2025-11-13 10:51:21,283 - mmdet - INFO - Iter [20/17500] lr: 1.076e-04, eta: 1 day, 11:13:03, time: 1.526, data_time: 0.075, memory: 49163, loss_cls_0: 1.1614, loss_box_0: 2.1881, loss_cns_0: 0.6091, loss_yns_0: 0.1703, loss_cls_1: 1.2257, loss_box_1: 2.8231, loss_cns_1: 0.5725, loss_yns_1: 0.1811, loss_cls_2: 1.2551, loss_box_2: 2.8194, loss_cns_2: 0.5558, loss_yns_2: 0.1747, loss_cls_3: 1.2610, loss_box_3: 2.9282, loss_cns_3: 0.5348, loss_yns_3: 0.1774, loss_cls_4: 1.2426, loss_box_4: 3.2070, loss_cns_4: 0.4880, loss_yns_4: 0.1692, loss_cls_5: 1.2611, loss_box_5: 3.4366, loss_cns_5: 0.4390, loss_yns_5: 0.1721, loss_cls_dn_0: 0.4635, loss_box_dn_0: 1.0460, loss_cls_dn_1: 0.4071, loss_box_dn_1: 1.2991, loss_cls_dn_2: 0.4135, loss_box_dn_2: 1.2851, loss_cls_dn_3: 0.4393, loss_box_dn_3: 1.3895, loss_cls_dn_4: 0.4346, loss_box_dn_4: 1.5733, loss_cls_dn_5: 0.4233, loss_box_dn_5: 1.6546, loss_dense_depth: 1.5896, loss: 41.4718, grad_norm: 97.8135 -2025-11-13 10:51:22,912 - mmdet - INFO - Iter [21/17500] lr: 1.080e-04, eta: 1 day, 9:54:55, time: 1.629, data_time: 0.130, memory: 49163, loss_cls_0: 1.1727, loss_box_0: 2.1958, loss_cns_0: 0.6172, loss_yns_0: 0.1720, loss_cls_1: 1.2273, loss_box_1: 2.9926, loss_cns_1: 0.5737, loss_yns_1: 0.1739, loss_cls_2: 1.2492, loss_box_2: 2.9781, loss_cns_2: 0.5459, loss_yns_2: 0.1736, loss_cls_3: 1.2535, loss_box_3: 2.9992, loss_cns_3: 0.5438, loss_yns_3: 0.1738, loss_cls_4: 1.2475, loss_box_4: 2.9818, loss_cns_4: 0.5274, loss_yns_4: 0.1701, loss_cls_5: 1.2705, loss_box_5: 3.0781, loss_cns_5: 0.5076, loss_yns_5: 0.1731, loss_cls_dn_0: 0.4569, loss_box_dn_0: 1.0599, loss_cls_dn_1: 0.4395, loss_box_dn_1: 1.3719, loss_cls_dn_2: 0.4512, loss_box_dn_2: 1.4225, loss_cls_dn_3: 0.4543, loss_box_dn_3: 1.5280, loss_cls_dn_4: 0.4499, loss_box_dn_4: 1.6980, loss_cls_dn_5: 0.4586, loss_box_dn_5: 1.7122, loss_dense_depth: 1.5916, loss: 42.0928, grad_norm: 82.5243 -2025-11-13 10:51:24,466 - mmdet - INFO - Iter [22/17500] lr: 1.084e-04, eta: 1 day, 8:42:52, time: 1.553, data_time: 0.111, memory: 49163, loss_cls_0: 1.1454, loss_box_0: 2.1764, loss_cns_0: 0.6172, loss_yns_0: 0.1700, loss_cls_1: 1.2567, loss_box_1: 2.8038, loss_cns_1: 0.5672, loss_yns_1: 0.1702, loss_cls_2: 1.2506, loss_box_2: 2.8049, loss_cns_2: 0.5584, loss_yns_2: 0.1700, loss_cls_3: 1.2620, loss_box_3: 2.7900, loss_cns_3: 0.5540, loss_yns_3: 0.1722, loss_cls_4: 1.2563, loss_box_4: 2.8072, loss_cns_4: 0.5599, loss_yns_4: 0.1711, loss_cls_5: 1.2813, loss_box_5: 2.7376, loss_cns_5: 0.5659, loss_yns_5: 0.1808, loss_cls_dn_0: 0.4645, loss_box_dn_0: 1.0329, loss_cls_dn_1: 0.4437, loss_box_dn_1: 1.3775, loss_cls_dn_2: 0.4559, loss_box_dn_2: 1.4355, loss_cls_dn_3: 0.4359, loss_box_dn_3: 1.4689, loss_cls_dn_4: 0.4444, loss_box_dn_4: 1.6223, loss_cls_dn_5: 0.4651, loss_box_dn_5: 1.5486, loss_dense_depth: 1.4749, loss: 40.6991, grad_norm: 66.1829 -2025-11-13 10:51:26,045 - mmdet - INFO - Iter [23/17500] lr: 1.088e-04, eta: 1 day, 7:37:27, time: 1.581, data_time: 0.119, memory: 49163, loss_cls_0: 1.1221, loss_box_0: 2.1063, loss_cns_0: 0.6166, loss_yns_0: 0.1689, loss_cls_1: 1.2415, loss_box_1: 2.6870, loss_cns_1: 0.5632, loss_yns_1: 0.1727, loss_cls_2: 1.2465, loss_box_2: 2.7662, loss_cns_2: 0.5663, loss_yns_2: 0.1745, loss_cls_3: 1.2803, loss_box_3: 2.7998, loss_cns_3: 0.5511, loss_yns_3: 0.1772, loss_cls_4: 1.2672, loss_box_4: 2.9099, loss_cns_4: 0.5473, loss_yns_4: 0.1708, loss_cls_5: 1.2702, loss_box_5: 2.9101, loss_cns_5: 0.5422, loss_yns_5: 0.1813, loss_cls_dn_0: 0.4666, loss_box_dn_0: 0.9851, loss_cls_dn_1: 0.4174, loss_box_dn_1: 1.2651, loss_cls_dn_2: 0.4366, loss_box_dn_2: 1.3792, loss_cls_dn_3: 0.4159, loss_box_dn_3: 1.3839, loss_cls_dn_4: 0.4191, loss_box_dn_4: 1.5211, loss_cls_dn_5: 0.4552, loss_box_dn_5: 1.4892, loss_dense_depth: 1.4200, loss: 40.0939, grad_norm: 81.8851 -2025-11-13 10:51:27,595 - mmdet - INFO - Iter [24/17500] lr: 1.092e-04, eta: 1 day, 6:37:05, time: 1.549, data_time: 0.076, memory: 49163, loss_cls_0: 1.1558, loss_box_0: 2.1341, loss_cns_0: 0.6171, loss_yns_0: 0.1700, loss_cls_1: 1.2732, loss_box_1: 2.7864, loss_cns_1: 0.5620, loss_yns_1: 0.1735, loss_cls_2: 1.2832, loss_box_2: 2.9229, loss_cns_2: 0.5779, loss_yns_2: 0.1707, loss_cls_3: 1.2942, loss_box_3: 2.8879, loss_cns_3: 0.5644, loss_yns_3: 0.1762, loss_cls_4: 1.3153, loss_box_4: 2.8732, loss_cns_4: 0.5699, loss_yns_4: 0.1781, loss_cls_5: 1.2762, loss_box_5: 2.9807, loss_cns_5: 0.5599, loss_yns_5: 0.1787, loss_cls_dn_0: 0.4670, loss_box_dn_0: 0.9965, loss_cls_dn_1: 0.3845, loss_box_dn_1: 1.3443, loss_cls_dn_2: 0.4084, loss_box_dn_2: 1.4495, loss_cls_dn_3: 0.4018, loss_box_dn_3: 1.3952, loss_cls_dn_4: 0.3920, loss_box_dn_4: 1.4328, loss_cls_dn_5: 0.4277, loss_box_dn_5: 1.4644, loss_dense_depth: 1.4467, loss: 40.6921, grad_norm: 80.9888 -2025-11-13 10:51:29,114 - mmdet - INFO - Iter [25/17500] lr: 1.096e-04, eta: 1 day, 5:41:10, time: 1.517, data_time: 0.075, memory: 49163, loss_cls_0: 1.1749, loss_box_0: 2.1574, loss_cns_0: 0.6126, loss_yns_0: 0.1699, loss_cls_1: 1.2690, loss_box_1: 2.8658, loss_cns_1: 0.5762, loss_yns_1: 0.1738, loss_cls_2: 1.2996, loss_box_2: 2.9688, loss_cns_2: 0.6019, loss_yns_2: 0.1708, loss_cls_3: 1.2637, loss_box_3: 2.8943, loss_cns_3: 0.5938, loss_yns_3: 0.1771, loss_cls_4: 1.3104, loss_box_4: 2.8950, loss_cns_4: 0.5905, loss_yns_4: 0.1806, loss_cls_5: 1.2828, loss_box_5: 2.9659, loss_cns_5: 0.5914, loss_yns_5: 0.1857, loss_cls_dn_0: 0.4539, loss_box_dn_0: 0.9958, loss_cls_dn_1: 0.3769, loss_box_dn_1: 1.2782, loss_cls_dn_2: 0.3953, loss_box_dn_2: 1.3535, loss_cls_dn_3: 0.4001, loss_box_dn_3: 1.3139, loss_cls_dn_4: 0.3829, loss_box_dn_4: 1.3253, loss_cls_dn_5: 0.4050, loss_box_dn_5: 1.4283, loss_dense_depth: 1.4282, loss: 40.5093, grad_norm: 69.0364 -2025-11-13 10:51:30,668 - mmdet - INFO - Iter [26/17500] lr: 1.100e-04, eta: 1 day, 4:49:58, time: 1.554, data_time: 0.079, memory: 49163, loss_cls_0: 1.1516, loss_box_0: 2.1991, loss_cns_0: 0.6072, loss_yns_0: 0.1708, loss_cls_1: 1.2110, loss_box_1: 2.7677, loss_cns_1: 0.5718, loss_yns_1: 0.1707, loss_cls_2: 1.2466, loss_box_2: 2.8632, loss_cns_2: 0.5791, loss_yns_2: 0.1767, loss_cls_3: 1.2409, loss_box_3: 2.8566, loss_cns_3: 0.5667, loss_yns_3: 0.1825, loss_cls_4: 1.2511, loss_box_4: 2.9212, loss_cns_4: 0.5674, loss_yns_4: 0.1811, loss_cls_5: 1.3226, loss_box_5: 2.9317, loss_cns_5: 0.5756, loss_yns_5: 0.2005, loss_cls_dn_0: 0.4400, loss_box_dn_0: 1.0220, loss_cls_dn_1: 0.4159, loss_box_dn_1: 1.2191, loss_cls_dn_2: 0.4336, loss_box_dn_2: 1.3086, loss_cls_dn_3: 0.4379, loss_box_dn_3: 1.3889, loss_cls_dn_4: 0.4289, loss_box_dn_4: 1.4489, loss_cls_dn_5: 0.4241, loss_box_dn_5: 1.6196, loss_dense_depth: 1.3086, loss: 40.4097, grad_norm: 87.3186 -2025-11-13 10:51:32,182 - mmdet - INFO - Iter [27/17500] lr: 1.104e-04, eta: 1 day, 4:02:08, time: 1.515, data_time: 0.078, memory: 49163, loss_cls_0: 1.1482, loss_box_0: 2.2523, loss_cns_0: 0.6014, loss_yns_0: 0.1724, loss_cls_1: 1.1824, loss_box_1: 2.7953, loss_cns_1: 0.5560, loss_yns_1: 0.1729, loss_cls_2: 1.2131, loss_box_2: 2.8165, loss_cns_2: 0.5627, loss_yns_2: 0.1820, loss_cls_3: 1.2292, loss_box_3: 2.8038, loss_cns_3: 0.5638, loss_yns_3: 0.1878, loss_cls_4: 1.2153, loss_box_4: 2.9239, loss_cns_4: 0.5707, loss_yns_4: 0.1954, loss_cls_5: 1.2981, loss_box_5: 2.9298, loss_cns_5: 0.5706, loss_yns_5: 0.1978, loss_cls_dn_0: 0.4387, loss_box_dn_0: 1.0391, loss_cls_dn_1: 0.4380, loss_box_dn_1: 1.3492, loss_cls_dn_2: 0.4498, loss_box_dn_2: 1.3832, loss_cls_dn_3: 0.4501, loss_box_dn_3: 1.5355, loss_cls_dn_4: 0.4595, loss_box_dn_4: 1.6323, loss_cls_dn_5: 0.4298, loss_box_dn_5: 1.8034, loss_dense_depth: 1.3192, loss: 41.0690, grad_norm: 87.8562 -2025-11-13 10:51:33,722 - mmdet - INFO - Iter [28/17500] lr: 1.108e-04, eta: 1 day, 3:17:58, time: 1.539, data_time: 0.078, memory: 49163, loss_cls_0: 1.1303, loss_box_0: 2.2604, loss_cns_0: 0.6007, loss_yns_0: 0.1715, loss_cls_1: 1.1719, loss_box_1: 2.7193, loss_cns_1: 0.5657, loss_yns_1: 0.1760, loss_cls_2: 1.2092, loss_box_2: 2.6830, loss_cns_2: 0.5703, loss_yns_2: 0.1741, loss_cls_3: 1.2200, loss_box_3: 2.7444, loss_cns_3: 0.5767, loss_yns_3: 0.1841, loss_cls_4: 1.2045, loss_box_4: 2.9092, loss_cns_4: 0.5710, loss_yns_4: 0.1778, loss_cls_5: 1.2376, loss_box_5: 2.9045, loss_cns_5: 0.5681, loss_yns_5: 0.1754, loss_cls_dn_0: 0.4377, loss_box_dn_0: 1.0502, loss_cls_dn_1: 0.4476, loss_box_dn_1: 1.5048, loss_cls_dn_2: 0.4520, loss_box_dn_2: 1.5051, loss_cls_dn_3: 0.4476, loss_box_dn_3: 1.6837, loss_cls_dn_4: 0.4638, loss_box_dn_4: 1.7730, loss_cls_dn_5: 0.4322, loss_box_dn_5: 1.9243, loss_dense_depth: 1.2812, loss: 41.3088, grad_norm: 84.1629 -2025-11-13 10:51:35,262 - mmdet - INFO - Iter [29/17500] lr: 1.112e-04, eta: 1 day, 2:36:52, time: 1.540, data_time: 0.080, memory: 49163, loss_cls_0: 1.1037, loss_box_0: 2.2601, loss_cns_0: 0.6014, loss_yns_0: 0.1715, loss_cls_1: 1.1547, loss_box_1: 2.7161, loss_cns_1: 0.5817, loss_yns_1: 0.1752, loss_cls_2: 1.1938, loss_box_2: 2.6880, loss_cns_2: 0.5824, loss_yns_2: 0.1770, loss_cls_3: 1.1975, loss_box_3: 2.6999, loss_cns_3: 0.5799, loss_yns_3: 0.1780, loss_cls_4: 1.1895, loss_box_4: 2.8307, loss_cns_4: 0.5687, loss_yns_4: 0.1789, loss_cls_5: 1.2101, loss_box_5: 2.8585, loss_cns_5: 0.5683, loss_yns_5: 0.1738, loss_cls_dn_0: 0.4440, loss_box_dn_0: 1.0464, loss_cls_dn_1: 0.4465, loss_box_dn_1: 1.4218, loss_cls_dn_2: 0.4505, loss_box_dn_2: 1.3938, loss_cls_dn_3: 0.4459, loss_box_dn_3: 1.5091, loss_cls_dn_4: 0.4602, loss_box_dn_4: 1.5585, loss_cls_dn_5: 0.4510, loss_box_dn_5: 1.6912, loss_dense_depth: 1.2617, loss: 40.2201, grad_norm: 93.7117 -2025-11-13 10:51:36,788 - mmdet - INFO - Iter [30/17500] lr: 1.116e-04, eta: 1 day, 1:58:22, time: 1.528, data_time: 0.079, memory: 49163, loss_cls_0: 1.0881, loss_box_0: 2.1852, loss_cns_0: 0.6026, loss_yns_0: 0.1738, loss_cls_1: 1.1145, loss_box_1: 2.6642, loss_cns_1: 0.5775, loss_yns_1: 0.1750, loss_cls_2: 1.1646, loss_box_2: 2.6644, loss_cns_2: 0.5851, loss_yns_2: 0.1753, loss_cls_3: 1.1701, loss_box_3: 2.5813, loss_cns_3: 0.5846, loss_yns_3: 0.1713, loss_cls_4: 1.1733, loss_box_4: 2.6040, loss_cns_4: 0.5808, loss_yns_4: 0.1785, loss_cls_5: 1.1924, loss_box_5: 2.6559, loss_cns_5: 0.5818, loss_yns_5: 0.1854, loss_cls_dn_0: 0.4445, loss_box_dn_0: 1.0234, loss_cls_dn_1: 0.4224, loss_box_dn_1: 1.5234, loss_cls_dn_2: 0.4242, loss_box_dn_2: 1.4824, loss_cls_dn_3: 0.4203, loss_box_dn_3: 1.4848, loss_cls_dn_4: 0.4308, loss_box_dn_4: 1.4740, loss_cls_dn_5: 0.4440, loss_box_dn_5: 1.5705, loss_dense_depth: 1.1879, loss: 39.1625, grad_norm: 71.9034 -2025-11-13 10:51:38,373 - mmdet - INFO - Iter [31/17500] lr: 1.120e-04, eta: 1 day, 1:22:54, time: 1.586, data_time: 0.078, memory: 49163, loss_cls_0: 1.0800, loss_box_0: 2.1170, loss_cns_0: 0.6046, loss_yns_0: 0.1740, loss_cls_1: 1.1083, loss_box_1: 2.5783, loss_cns_1: 0.5656, loss_yns_1: 0.1745, loss_cls_2: 1.1516, loss_box_2: 2.5944, loss_cns_2: 0.5736, loss_yns_2: 0.1723, loss_cls_3: 1.1617, loss_box_3: 2.5099, loss_cns_3: 0.5747, loss_yns_3: 0.1755, loss_cls_4: 1.1790, loss_box_4: 2.5267, loss_cns_4: 0.5794, loss_yns_4: 0.1723, loss_cls_5: 1.1801, loss_box_5: 2.5341, loss_cns_5: 0.5817, loss_yns_5: 0.1844, loss_cls_dn_0: 0.4430, loss_box_dn_0: 0.9863, loss_cls_dn_1: 0.3954, loss_box_dn_1: 1.4291, loss_cls_dn_2: 0.3964, loss_box_dn_2: 1.3872, loss_cls_dn_3: 0.3982, loss_box_dn_3: 1.3266, loss_cls_dn_4: 0.4057, loss_box_dn_4: 1.3061, loss_cls_dn_5: 0.4365, loss_box_dn_5: 1.3386, loss_dense_depth: 1.1698, loss: 37.6724, grad_norm: 41.9387 -2025-11-13 10:51:39,899 - mmdet - INFO - Iter [32/17500] lr: 1.124e-04, eta: 1 day, 0:49:06, time: 1.524, data_time: 0.080, memory: 49163, loss_cls_0: 1.0755, loss_box_0: 2.0865, loss_cns_0: 0.6122, loss_yns_0: 0.1712, loss_cls_1: 1.1262, loss_box_1: 2.6263, loss_cns_1: 0.5502, loss_yns_1: 0.1721, loss_cls_2: 1.1657, loss_box_2: 2.6335, loss_cns_2: 0.5626, loss_yns_2: 0.1717, loss_cls_3: 1.1755, loss_box_3: 2.6613, loss_cns_3: 0.5593, loss_yns_3: 0.1847, loss_cls_4: 1.1606, loss_box_4: 2.7021, loss_cns_4: 0.5572, loss_yns_4: 0.1703, loss_cls_5: 1.1529, loss_box_5: 2.6842, loss_cns_5: 0.5657, loss_yns_5: 0.1724, loss_cls_dn_0: 0.4231, loss_box_dn_0: 0.9687, loss_cls_dn_1: 0.3683, loss_box_dn_1: 1.2845, loss_cls_dn_2: 0.3681, loss_box_dn_2: 1.2884, loss_cls_dn_3: 0.3750, loss_box_dn_3: 1.3175, loss_cls_dn_4: 0.3840, loss_box_dn_4: 1.3796, loss_cls_dn_5: 0.4197, loss_box_dn_5: 1.3826, loss_dense_depth: 1.2388, loss: 37.8981, grad_norm: 56.2962 -2025-11-13 10:51:41,418 - mmdet - INFO - Iter [33/17500] lr: 1.128e-04, eta: 1 day, 0:17:17, time: 1.518, data_time: 0.080, memory: 49163, loss_cls_0: 1.0945, loss_box_0: 2.0954, loss_cns_0: 0.6172, loss_yns_0: 0.1726, loss_cls_1: 1.1211, loss_box_1: 2.6951, loss_cns_1: 0.5404, loss_yns_1: 0.1741, loss_cls_2: 1.1839, loss_box_2: 2.6556, loss_cns_2: 0.5594, loss_yns_2: 0.1737, loss_cls_3: 1.1797, loss_box_3: 2.6961, loss_cns_3: 0.5583, loss_yns_3: 0.1775, loss_cls_4: 1.1581, loss_box_4: 2.7185, loss_cns_4: 0.5589, loss_yns_4: 0.1746, loss_cls_5: 1.1531, loss_box_5: 2.6792, loss_cns_5: 0.5669, loss_yns_5: 0.1729, loss_cls_dn_0: 0.4204, loss_box_dn_0: 0.9735, loss_cls_dn_1: 0.3462, loss_box_dn_1: 1.5156, loss_cls_dn_2: 0.3533, loss_box_dn_2: 1.5222, loss_cls_dn_3: 0.3595, loss_box_dn_3: 1.5846, loss_cls_dn_4: 0.3683, loss_box_dn_4: 1.6643, loss_cls_dn_5: 0.3822, loss_box_dn_5: 1.6668, loss_dense_depth: 1.2340, loss: 39.2676, grad_norm: 65.3435 -2025-11-13 10:51:42,938 - mmdet - INFO - Iter [34/17500] lr: 1.132e-04, eta: 23:47:21, time: 1.520, data_time: 0.079, memory: 49163, loss_cls_0: 1.0841, loss_box_0: 2.0852, loss_cns_0: 0.6202, loss_yns_0: 0.1752, loss_cls_1: 1.1034, loss_box_1: 2.6092, loss_cns_1: 0.5499, loss_yns_1: 0.1798, loss_cls_2: 1.1449, loss_box_2: 2.5719, loss_cns_2: 0.5687, loss_yns_2: 0.1751, loss_cls_3: 1.1579, loss_box_3: 2.5726, loss_cns_3: 0.5692, loss_yns_3: 0.1702, loss_cls_4: 1.1636, loss_box_4: 2.5753, loss_cns_4: 0.5688, loss_yns_4: 0.1746, loss_cls_5: 1.1667, loss_box_5: 2.5545, loss_cns_5: 0.5818, loss_yns_5: 0.1791, loss_cls_dn_0: 0.4328, loss_box_dn_0: 0.9729, loss_cls_dn_1: 0.3458, loss_box_dn_1: 1.6622, loss_cls_dn_2: 0.3571, loss_box_dn_2: 1.6680, loss_cls_dn_3: 0.3644, loss_box_dn_3: 1.7144, loss_cls_dn_4: 0.3634, loss_box_dn_4: 1.7884, loss_cls_dn_5: 0.3629, loss_box_dn_5: 1.7797, loss_dense_depth: 1.2448, loss: 39.3588, grad_norm: 61.2490 -2025-11-13 10:51:44,452 - mmdet - INFO - Iter [35/17500] lr: 1.136e-04, eta: 23:19:07, time: 1.517, data_time: 0.077, memory: 49163, loss_cls_0: 1.0672, loss_box_0: 2.0570, loss_cns_0: 0.6204, loss_yns_0: 0.1753, loss_cls_1: 1.1271, loss_box_1: 2.5331, loss_cns_1: 0.5633, loss_yns_1: 0.1811, loss_cls_2: 1.1353, loss_box_2: 2.5231, loss_cns_2: 0.5795, loss_yns_2: 0.1721, loss_cls_3: 1.1495, loss_box_3: 2.4930, loss_cns_3: 0.5853, loss_yns_3: 0.1745, loss_cls_4: 1.1681, loss_box_4: 2.5191, loss_cns_4: 0.5839, loss_yns_4: 0.1719, loss_cls_5: 1.1973, loss_box_5: 2.5471, loss_cns_5: 0.5913, loss_yns_5: 0.1811, loss_cls_dn_0: 0.4342, loss_box_dn_0: 0.9620, loss_cls_dn_1: 0.3481, loss_box_dn_1: 1.6042, loss_cls_dn_2: 0.3629, loss_box_dn_2: 1.6207, loss_cls_dn_3: 0.3726, loss_box_dn_3: 1.6345, loss_cls_dn_4: 0.3635, loss_box_dn_4: 1.7035, loss_cls_dn_5: 0.3556, loss_box_dn_5: 1.7052, loss_dense_depth: 1.2529, loss: 38.8162, grad_norm: 62.4994 -2025-11-13 10:51:45,960 - mmdet - INFO - Iter [36/17500] lr: 1.140e-04, eta: 22:52:21, time: 1.507, data_time: 0.074, memory: 49163, loss_cls_0: 1.0540, loss_box_0: 2.0953, loss_cns_0: 0.6165, loss_yns_0: 0.1723, loss_cls_1: 1.0979, loss_box_1: 2.5893, loss_cns_1: 0.5632, loss_yns_1: 0.1794, loss_cls_2: 1.1204, loss_box_2: 2.5572, loss_cns_2: 0.5782, loss_yns_2: 0.1715, loss_cls_3: 1.1411, loss_box_3: 2.5524, loss_cns_3: 0.5878, loss_yns_3: 0.1856, loss_cls_4: 1.1531, loss_box_4: 2.5946, loss_cns_4: 0.5802, loss_yns_4: 0.1709, loss_cls_5: 1.1874, loss_box_5: 2.6282, loss_cns_5: 0.5840, loss_yns_5: 0.1766, loss_cls_dn_0: 0.4241, loss_box_dn_0: 0.9617, loss_cls_dn_1: 0.3710, loss_box_dn_1: 1.1443, loss_cls_dn_2: 0.3844, loss_box_dn_2: 1.1458, loss_cls_dn_3: 0.3921, loss_box_dn_3: 1.1437, loss_cls_dn_4: 0.3830, loss_box_dn_4: 1.1971, loss_cls_dn_5: 0.3801, loss_box_dn_5: 1.1960, loss_dense_depth: 1.2294, loss: 36.6897, grad_norm: 50.5562 -2025-11-13 10:51:47,484 - mmdet - INFO - Iter [37/17500] lr: 1.144e-04, eta: 22:27:10, time: 1.523, data_time: 0.077, memory: 49163, loss_cls_0: 1.0689, loss_box_0: 2.0939, loss_cns_0: 0.6195, loss_yns_0: 0.1738, loss_cls_1: 1.1219, loss_box_1: 2.7261, loss_cns_1: 0.5598, loss_yns_1: 0.1718, loss_cls_2: 1.1248, loss_box_2: 2.6705, loss_cns_2: 0.5731, loss_yns_2: 0.1756, loss_cls_3: 1.1566, loss_box_3: 2.6700, loss_cns_3: 0.5791, loss_yns_3: 0.1818, loss_cls_4: 1.1466, loss_box_4: 2.6820, loss_cns_4: 0.5794, loss_yns_4: 0.1721, loss_cls_5: 1.1711, loss_box_5: 2.6593, loss_cns_5: 0.5851, loss_yns_5: 0.1757, loss_cls_dn_0: 0.4125, loss_box_dn_0: 0.9747, loss_cls_dn_1: 0.3513, loss_box_dn_1: 1.1808, loss_cls_dn_2: 0.3738, loss_box_dn_2: 1.1297, loss_cls_dn_3: 0.3700, loss_box_dn_3: 1.1037, loss_cls_dn_4: 0.3745, loss_box_dn_4: 1.1204, loss_cls_dn_5: 0.3717, loss_box_dn_5: 1.1003, loss_dense_depth: 1.2167, loss: 36.9186, grad_norm: 52.3418 -2025-11-13 10:51:49,006 - mmdet - INFO - Iter [38/17500] lr: 1.148e-04, eta: 22:03:19, time: 1.523, data_time: 0.075, memory: 49163, loss_cls_0: 1.0706, loss_box_0: 2.0607, loss_cns_0: 0.6216, loss_yns_0: 0.1736, loss_cls_1: 1.1135, loss_box_1: 2.7599, loss_cns_1: 0.5599, loss_yns_1: 0.1715, loss_cls_2: 1.1129, loss_box_2: 2.6809, loss_cns_2: 0.5770, loss_yns_2: 0.1755, loss_cls_3: 1.1492, loss_box_3: 2.6819, loss_cns_3: 0.5799, loss_yns_3: 0.1705, loss_cls_4: 1.1326, loss_box_4: 2.7240, loss_cns_4: 0.5817, loss_yns_4: 0.1724, loss_cls_5: 1.1557, loss_box_5: 2.7533, loss_cns_5: 0.5821, loss_yns_5: 0.1746, loss_cls_dn_0: 0.4013, loss_box_dn_0: 0.9698, loss_cls_dn_1: 0.3448, loss_box_dn_1: 1.1070, loss_cls_dn_2: 0.3738, loss_box_dn_2: 1.0467, loss_cls_dn_3: 0.3708, loss_box_dn_3: 1.0587, loss_cls_dn_4: 0.3776, loss_box_dn_4: 1.0657, loss_cls_dn_5: 0.3824, loss_box_dn_5: 1.1061, loss_dense_depth: 1.2020, loss: 36.7422, grad_norm: 66.6528 -2025-11-13 10:51:50,523 - mmdet - INFO - Iter [39/17500] lr: 1.152e-04, eta: 21:40:37, time: 1.515, data_time: 0.075, memory: 49163, loss_cls_0: 1.0774, loss_box_0: 1.9870, loss_cns_0: 0.6233, loss_yns_0: 0.1732, loss_cls_1: 1.1278, loss_box_1: 2.5066, loss_cns_1: 0.5808, loss_yns_1: 0.1779, loss_cls_2: 1.1167, loss_box_2: 2.4225, loss_cns_2: 0.5960, loss_yns_2: 0.1714, loss_cls_3: 1.1441, loss_box_3: 2.4314, loss_cns_3: 0.5939, loss_yns_3: 0.1697, loss_cls_4: 1.1385, loss_box_4: 2.4644, loss_cns_4: 0.5939, loss_yns_4: 0.1699, loss_cls_5: 1.1616, loss_box_5: 2.5587, loss_cns_5: 0.5877, loss_yns_5: 0.1715, loss_cls_dn_0: 0.4088, loss_box_dn_0: 0.9447, loss_cls_dn_1: 0.3460, loss_box_dn_1: 1.1413, loss_cls_dn_2: 0.3798, loss_box_dn_2: 1.1078, loss_cls_dn_3: 0.3814, loss_box_dn_3: 1.1767, loss_cls_dn_4: 0.3819, loss_box_dn_4: 1.1979, loss_cls_dn_5: 0.3959, loss_box_dn_5: 1.3252, loss_dense_depth: 1.1933, loss: 36.1267, grad_norm: 59.9087 -2025-11-13 10:51:52,036 - mmdet - INFO - Iter [40/17500] lr: 1.156e-04, eta: 21:19:03, time: 1.515, data_time: 0.074, memory: 49163, loss_cls_0: 1.0634, loss_box_0: 1.9480, loss_cns_0: 0.6233, loss_yns_0: 0.1679, loss_cls_1: 1.1083, loss_box_1: 2.3979, loss_cns_1: 0.5917, loss_yns_1: 0.1827, loss_cls_2: 1.1161, loss_box_2: 2.3229, loss_cns_2: 0.6043, loss_yns_2: 0.1685, loss_cls_3: 1.1456, loss_box_3: 2.3297, loss_cns_3: 0.6020, loss_yns_3: 0.1759, loss_cls_4: 1.1322, loss_box_4: 2.3533, loss_cns_4: 0.6033, loss_yns_4: 0.1682, loss_cls_5: 1.1542, loss_box_5: 2.4307, loss_cns_5: 0.5987, loss_yns_5: 0.1763, loss_cls_dn_0: 0.4192, loss_box_dn_0: 0.9357, loss_cls_dn_1: 0.3556, loss_box_dn_1: 1.2828, loss_cls_dn_2: 0.3850, loss_box_dn_2: 1.2479, loss_cls_dn_3: 0.3857, loss_box_dn_3: 1.3375, loss_cls_dn_4: 0.3837, loss_box_dn_4: 1.3630, loss_cls_dn_5: 0.4056, loss_box_dn_5: 1.4909, loss_dense_depth: 1.2174, loss: 36.3752, grad_norm: 53.9123 -2025-11-13 10:51:53,641 - mmdet - INFO - Iter [41/17500] lr: 1.160e-04, eta: 20:59:10, time: 1.604, data_time: 0.134, memory: 49163, loss_cls_0: 1.0208, loss_box_0: 1.9444, loss_cns_0: 0.6236, loss_yns_0: 0.1635, loss_cls_1: 1.0814, loss_box_1: 2.3183, loss_cns_1: 0.6007, loss_yns_1: 0.1783, loss_cls_2: 1.0980, loss_box_2: 2.2732, loss_cns_2: 0.6131, loss_yns_2: 0.1670, loss_cls_3: 1.1216, loss_box_3: 2.3255, loss_cns_3: 0.6135, loss_yns_3: 0.1772, loss_cls_4: 1.1236, loss_box_4: 2.3674, loss_cns_4: 0.6122, loss_yns_4: 0.1715, loss_cls_5: 1.1313, loss_box_5: 2.3702, loss_cns_5: 0.6150, loss_yns_5: 0.1861, loss_cls_dn_0: 0.4196, loss_box_dn_0: 0.9312, loss_cls_dn_1: 0.3430, loss_box_dn_1: 1.3814, loss_cls_dn_2: 0.3563, loss_box_dn_2: 1.3263, loss_cls_dn_3: 0.3552, loss_box_dn_3: 1.4107, loss_cls_dn_4: 0.3632, loss_box_dn_4: 1.4329, loss_cls_dn_5: 0.3871, loss_box_dn_5: 1.5043, loss_dense_depth: 1.1988, loss: 36.3074, grad_norm: 63.6657 -2025-11-13 10:51:55,190 - mmdet - INFO - Iter [42/17500] lr: 1.164e-04, eta: 20:39:51, time: 1.549, data_time: 0.108, memory: 49163, loss_cls_0: 1.0284, loss_box_0: 1.9596, loss_cns_0: 0.6217, loss_yns_0: 0.1661, loss_cls_1: 1.0863, loss_box_1: 2.3099, loss_cns_1: 0.6037, loss_yns_1: 0.1695, loss_cls_2: 1.1279, loss_box_2: 2.2757, loss_cns_2: 0.6138, loss_yns_2: 0.1700, loss_cls_3: 1.1498, loss_box_3: 2.3443, loss_cns_3: 0.6161, loss_yns_3: 0.1782, loss_cls_4: 1.1271, loss_box_4: 2.3520, loss_cns_4: 0.6174, loss_yns_4: 0.1744, loss_cls_5: 1.1202, loss_box_5: 2.3048, loss_cns_5: 0.6221, loss_yns_5: 0.1836, loss_cls_dn_0: 0.4089, loss_box_dn_0: 0.9289, loss_cls_dn_1: 0.3375, loss_box_dn_1: 1.2890, loss_cls_dn_2: 0.3443, loss_box_dn_2: 1.2248, loss_cls_dn_3: 0.3512, loss_box_dn_3: 1.2957, loss_cls_dn_4: 0.3594, loss_box_dn_4: 1.2891, loss_cls_dn_5: 0.3926, loss_box_dn_5: 1.3142, loss_dense_depth: 1.1835, loss: 35.6418, grad_norm: 66.6666 -2025-11-13 10:51:56,737 - mmdet - INFO - Iter [43/17500] lr: 1.168e-04, eta: 20:21:25, time: 1.548, data_time: 0.121, memory: 49163, loss_cls_0: 1.0751, loss_box_0: 1.9995, loss_cns_0: 0.6158, loss_yns_0: 0.1690, loss_cls_1: 1.1381, loss_box_1: 2.3084, loss_cns_1: 0.6019, loss_yns_1: 0.1708, loss_cls_2: 1.1664, loss_box_2: 2.2408, loss_cns_2: 0.6123, loss_yns_2: 0.1667, loss_cls_3: 1.1733, loss_box_3: 2.2726, loss_cns_3: 0.6191, loss_yns_3: 0.1675, loss_cls_4: 1.1586, loss_box_4: 2.2647, loss_cns_4: 0.6193, loss_yns_4: 0.1716, loss_cls_5: 1.1535, loss_box_5: 2.2811, loss_cns_5: 0.6188, loss_yns_5: 0.1747, loss_cls_dn_0: 0.4131, loss_box_dn_0: 0.9244, loss_cls_dn_1: 0.3445, loss_box_dn_1: 1.1381, loss_cls_dn_2: 0.3572, loss_box_dn_2: 1.0738, loss_cls_dn_3: 0.3760, loss_box_dn_3: 1.1018, loss_cls_dn_4: 0.3689, loss_box_dn_4: 1.0805, loss_cls_dn_5: 0.4059, loss_box_dn_5: 1.1087, loss_dense_depth: 1.1291, loss: 34.7618, grad_norm: 42.2290 -2025-11-13 10:51:58,261 - mmdet - INFO - Iter [44/17500] lr: 1.172e-04, eta: 20:03:39, time: 1.523, data_time: 0.076, memory: 49163, loss_cls_0: 1.0517, loss_box_0: 2.0397, loss_cns_0: 0.6111, loss_yns_0: 0.1636, loss_cls_1: 1.1149, loss_box_1: 2.3427, loss_cns_1: 0.6039, loss_yns_1: 0.1726, loss_cls_2: 1.1134, loss_box_2: 2.2702, loss_cns_2: 0.6169, loss_yns_2: 0.1680, loss_cls_3: 1.1349, loss_box_3: 2.2528, loss_cns_3: 0.6215, loss_yns_3: 0.1636, loss_cls_4: 1.1334, loss_box_4: 2.3285, loss_cns_4: 0.6196, loss_yns_4: 0.1716, loss_cls_5: 1.1634, loss_box_5: 2.3653, loss_cns_5: 0.6186, loss_yns_5: 0.1742, loss_cls_dn_0: 0.4265, loss_box_dn_0: 0.9177, loss_cls_dn_1: 0.3456, loss_box_dn_1: 1.1311, loss_cls_dn_2: 0.3788, loss_box_dn_2: 1.0587, loss_cls_dn_3: 0.4097, loss_box_dn_3: 1.0480, loss_cls_dn_4: 0.3902, loss_box_dn_4: 1.0763, loss_cls_dn_5: 0.4079, loss_box_dn_5: 1.0975, loss_dense_depth: 1.2142, loss: 34.9182, grad_norm: 55.8051 -2025-11-13 10:51:59,791 - mmdet - INFO - Iter [45/17500] lr: 1.176e-04, eta: 19:46:44, time: 1.530, data_time: 0.077, memory: 49163, loss_cls_0: 1.0678, loss_box_0: 2.0030, loss_cns_0: 0.6145, loss_yns_0: 0.1628, loss_cls_1: 1.1096, loss_box_1: 2.3754, loss_cns_1: 0.5934, loss_yns_1: 0.1760, loss_cls_2: 1.1250, loss_box_2: 2.3008, loss_cns_2: 0.6077, loss_yns_2: 0.1695, loss_cls_3: 1.1475, loss_box_3: 2.2754, loss_cns_3: 0.6165, loss_yns_3: 0.1677, loss_cls_4: 1.1325, loss_box_4: 2.3483, loss_cns_4: 0.6152, loss_yns_4: 0.1692, loss_cls_5: 1.1689, loss_box_5: 2.3329, loss_cns_5: 0.6179, loss_yns_5: 0.1855, loss_cls_dn_0: 0.4279, loss_box_dn_0: 0.9133, loss_cls_dn_1: 0.3466, loss_box_dn_1: 1.0566, loss_cls_dn_2: 0.3814, loss_box_dn_2: 0.9944, loss_cls_dn_3: 0.4152, loss_box_dn_3: 0.9846, loss_cls_dn_4: 0.3893, loss_box_dn_4: 1.0467, loss_cls_dn_5: 0.3978, loss_box_dn_5: 1.0498, loss_dense_depth: 1.1495, loss: 34.6360, grad_norm: 61.4185 -2025-11-13 10:52:01,323 - mmdet - INFO - Iter [46/17500] lr: 1.180e-04, eta: 19:30:33, time: 1.531, data_time: 0.076, memory: 49163, loss_cls_0: 1.0583, loss_box_0: 1.9779, loss_cns_0: 0.6155, loss_yns_0: 0.1650, loss_cls_1: 1.0994, loss_box_1: 2.4310, loss_cns_1: 0.5897, loss_yns_1: 0.1714, loss_cls_2: 1.1219, loss_box_2: 2.3166, loss_cns_2: 0.6071, loss_yns_2: 0.1675, loss_cls_3: 1.1391, loss_box_3: 2.2904, loss_cns_3: 0.6161, loss_yns_3: 0.1669, loss_cls_4: 1.1289, loss_box_4: 2.3306, loss_cns_4: 0.6174, loss_yns_4: 0.1645, loss_cls_5: 1.1573, loss_box_5: 2.3451, loss_cns_5: 0.6218, loss_yns_5: 0.1884, loss_cls_dn_0: 0.4219, loss_box_dn_0: 0.9064, loss_cls_dn_1: 0.3397, loss_box_dn_1: 1.0595, loss_cls_dn_2: 0.3641, loss_box_dn_2: 1.0017, loss_cls_dn_3: 0.3859, loss_box_dn_3: 0.9968, loss_cls_dn_4: 0.3576, loss_box_dn_4: 1.0627, loss_cls_dn_5: 0.3721, loss_box_dn_5: 1.0858, loss_dense_depth: 1.1468, loss: 34.5888, grad_norm: 46.7259 -2025-11-13 10:52:02,851 - mmdet - INFO - Iter [47/17500] lr: 1.184e-04, eta: 19:15:02, time: 1.529, data_time: 0.078, memory: 49163, loss_cls_0: 1.0478, loss_box_0: 1.9625, loss_cns_0: 0.6145, loss_yns_0: 0.1656, loss_cls_1: 1.0942, loss_box_1: 2.5202, loss_cns_1: 0.5977, loss_yns_1: 0.1673, loss_cls_2: 1.1273, loss_box_2: 2.4208, loss_cns_2: 0.6081, loss_yns_2: 0.1648, loss_cls_3: 1.1560, loss_box_3: 2.3929, loss_cns_3: 0.6180, loss_yns_3: 0.1645, loss_cls_4: 1.1742, loss_box_4: 2.4097, loss_cns_4: 0.6184, loss_yns_4: 0.1663, loss_cls_5: 1.1741, loss_box_5: 2.4561, loss_cns_5: 0.6238, loss_yns_5: 0.1762, loss_cls_dn_0: 0.4220, loss_box_dn_0: 0.9099, loss_cls_dn_1: 0.3401, loss_box_dn_1: 1.0879, loss_cls_dn_2: 0.3468, loss_box_dn_2: 1.0687, loss_cls_dn_3: 0.3551, loss_box_dn_3: 1.0573, loss_cls_dn_4: 0.3422, loss_box_dn_4: 1.1195, loss_cls_dn_5: 0.3674, loss_box_dn_5: 1.1612, loss_dense_depth: 1.1713, loss: 35.3703, grad_norm: 55.2099 -2025-11-13 10:52:04,372 - mmdet - INFO - Iter [48/17500] lr: 1.188e-04, eta: 19:00:08, time: 1.521, data_time: 0.076, memory: 49163, loss_cls_0: 1.0467, loss_box_0: 1.9592, loss_cns_0: 0.6070, loss_yns_0: 0.1621, loss_cls_1: 1.0710, loss_box_1: 2.5555, loss_cns_1: 0.5803, loss_yns_1: 0.1673, loss_cls_2: 1.1198, loss_box_2: 2.4253, loss_cns_2: 0.6073, loss_yns_2: 0.1628, loss_cls_3: 1.1657, loss_box_3: 2.4321, loss_cns_3: 0.6185, loss_yns_3: 0.1625, loss_cls_4: 1.1393, loss_box_4: 2.4678, loss_cns_4: 0.6208, loss_yns_4: 0.1654, loss_cls_5: 1.1573, loss_box_5: 2.5252, loss_cns_5: 0.6234, loss_yns_5: 0.1657, loss_cls_dn_0: 0.4176, loss_box_dn_0: 0.9027, loss_cls_dn_1: 0.3259, loss_box_dn_1: 1.1746, loss_cls_dn_2: 0.3428, loss_box_dn_2: 1.1240, loss_cls_dn_3: 0.3367, loss_box_dn_3: 1.1093, loss_cls_dn_4: 0.3467, loss_box_dn_4: 1.1627, loss_cls_dn_5: 0.3634, loss_box_dn_5: 1.1963, loss_dense_depth: 1.0826, loss: 35.5934, grad_norm: 57.6643 -2025-11-13 10:52:05,907 - mmdet - INFO - Iter [49/17500] lr: 1.192e-04, eta: 18:45:54, time: 1.534, data_time: 0.078, memory: 49163, loss_cls_0: 1.0740, loss_box_0: 1.9646, loss_cns_0: 0.6045, loss_yns_0: 0.1602, loss_cls_1: 1.0848, loss_box_1: 2.5473, loss_cns_1: 0.5912, loss_yns_1: 0.1688, loss_cls_2: 1.1096, loss_box_2: 2.4652, loss_cns_2: 0.6136, loss_yns_2: 0.1632, loss_cls_3: 1.1505, loss_box_3: 2.4953, loss_cns_3: 0.6234, loss_yns_3: 0.1638, loss_cls_4: 1.1179, loss_box_4: 2.5610, loss_cns_4: 0.6204, loss_yns_4: 0.1658, loss_cls_5: 1.1413, loss_box_5: 2.5983, loss_cns_5: 0.6221, loss_yns_5: 0.1710, loss_cls_dn_0: 0.4094, loss_box_dn_0: 0.8930, loss_cls_dn_1: 0.3133, loss_box_dn_1: 1.1860, loss_cls_dn_2: 0.3396, loss_box_dn_2: 1.1360, loss_cls_dn_3: 0.3352, loss_box_dn_3: 1.1208, loss_cls_dn_4: 0.3508, loss_box_dn_4: 1.1607, loss_cls_dn_5: 0.3634, loss_box_dn_5: 1.1743, loss_dense_depth: 1.2293, loss: 35.9896, grad_norm: 59.2040 -2025-11-13 10:52:07,463 - mmdet - INFO - Iter [50/17500] lr: 1.196e-04, eta: 18:32:23, time: 1.557, data_time: 0.076, memory: 49163, loss_cls_0: 1.0589, loss_box_0: 1.9645, loss_cns_0: 0.6116, loss_yns_0: 0.1615, loss_cls_1: 1.0902, loss_box_1: 2.5463, loss_cns_1: 0.5856, loss_yns_1: 0.1674, loss_cls_2: 1.0979, loss_box_2: 2.4116, loss_cns_2: 0.6188, loss_yns_2: 0.1648, loss_cls_3: 1.1321, loss_box_3: 2.4104, loss_cns_3: 0.6259, loss_yns_3: 0.1629, loss_cls_4: 1.1098, loss_box_4: 2.4287, loss_cns_4: 0.6271, loss_yns_4: 0.1633, loss_cls_5: 1.1293, loss_box_5: 2.4435, loss_cns_5: 0.6317, loss_yns_5: 0.1765, loss_cls_dn_0: 0.3872, loss_box_dn_0: 0.8918, loss_cls_dn_1: 0.3119, loss_box_dn_1: 1.0873, loss_cls_dn_2: 0.3430, loss_box_dn_2: 1.0299, loss_cls_dn_3: 0.3435, loss_box_dn_3: 1.0094, loss_cls_dn_4: 0.3461, loss_box_dn_4: 1.0198, loss_cls_dn_5: 0.3677, loss_box_dn_5: 1.0248, loss_dense_depth: 1.0205, loss: 34.7032, grad_norm: 44.4188 -2025-11-13 10:52:09,103 - mmdet - INFO - Iter [51/17500] lr: 1.200e-04, eta: 18:19:51, time: 1.637, data_time: 0.076, memory: 49163, loss_cls_0: 1.0271, loss_box_0: 1.9604, loss_cns_0: 0.6168, loss_yns_0: 0.1643, loss_cls_1: 1.0836, loss_box_1: 2.4921, loss_cns_1: 0.5904, loss_yns_1: 0.1675, loss_cls_2: 1.1036, loss_box_2: 2.3684, loss_cns_2: 0.6228, loss_yns_2: 0.1665, loss_cls_3: 1.1233, loss_box_3: 2.3486, loss_cns_3: 0.6273, loss_yns_3: 0.1656, loss_cls_4: 1.1305, loss_box_4: 2.3545, loss_cns_4: 0.6307, loss_yns_4: 0.1655, loss_cls_5: 1.1399, loss_box_5: 2.3310, loss_cns_5: 0.6358, loss_yns_5: 0.1692, loss_cls_dn_0: 0.3854, loss_box_dn_0: 0.8833, loss_cls_dn_1: 0.2937, loss_box_dn_1: 1.0734, loss_cls_dn_2: 0.3267, loss_box_dn_2: 1.0010, loss_cls_dn_3: 0.3366, loss_box_dn_3: 0.9868, loss_cls_dn_4: 0.3263, loss_box_dn_4: 0.9845, loss_cls_dn_5: 0.3513, loss_box_dn_5: 0.9753, loss_dense_depth: 1.2163, loss: 34.3260, grad_norm: 34.4872 -2025-11-13 10:52:10,806 - mmdet - INFO - Iter [52/17500] lr: 1.204e-04, eta: 18:08:10, time: 1.704, data_time: 0.082, memory: 49163, loss_cls_0: 1.0370, loss_box_0: 1.9884, loss_cns_0: 0.6140, loss_yns_0: 0.1646, loss_cls_1: 1.1024, loss_box_1: 2.4120, loss_cns_1: 0.5987, loss_yns_1: 0.1668, loss_cls_2: 1.1331, loss_box_2: 2.3251, loss_cns_2: 0.6268, loss_yns_2: 0.1653, loss_cls_3: 1.1376, loss_box_3: 2.3362, loss_cns_3: 0.6322, loss_yns_3: 0.1667, loss_cls_4: 1.1371, loss_box_4: 2.3741, loss_cns_4: 0.6339, loss_yns_4: 0.1674, loss_cls_5: 1.1456, loss_box_5: 2.3295, loss_cns_5: 0.6350, loss_yns_5: 0.1676, loss_cls_dn_0: 0.4086, loss_box_dn_0: 0.8837, loss_cls_dn_1: 0.2853, loss_box_dn_1: 1.0144, loss_cls_dn_2: 0.3183, loss_box_dn_2: 0.9542, loss_cls_dn_3: 0.3444, loss_box_dn_3: 0.9769, loss_cls_dn_4: 0.3361, loss_box_dn_4: 0.9915, loss_cls_dn_5: 0.3609, loss_box_dn_5: 0.9886, loss_dense_depth: 1.2210, loss: 34.2806, grad_norm: 52.0058 -2025-11-13 10:52:12,497 - mmdet - INFO - Iter [53/17500] lr: 1.208e-04, eta: 17:56:50, time: 1.687, data_time: 0.081, memory: 49163, loss_cls_0: 1.0524, loss_box_0: 2.0295, loss_cns_0: 0.6084, loss_yns_0: 0.1673, loss_cls_1: 1.1015, loss_box_1: 2.4087, loss_cns_1: 0.5998, loss_yns_1: 0.1708, loss_cls_2: 1.1206, loss_box_2: 2.3872, loss_cns_2: 0.6149, loss_yns_2: 0.1677, loss_cls_3: 1.1363, loss_box_3: 2.3925, loss_cns_3: 0.6195, loss_yns_3: 0.1678, loss_cls_4: 1.1249, loss_box_4: 2.4225, loss_cns_4: 0.6167, loss_yns_4: 0.1658, loss_cls_5: 1.1420, loss_box_5: 2.4393, loss_cns_5: 0.6193, loss_yns_5: 0.1677, loss_cls_dn_0: 0.4288, loss_box_dn_0: 0.8909, loss_cls_dn_1: 0.2879, loss_box_dn_1: 1.0095, loss_cls_dn_2: 0.3181, loss_box_dn_2: 0.9674, loss_cls_dn_3: 0.3499, loss_box_dn_3: 0.9994, loss_cls_dn_4: 0.3506, loss_box_dn_4: 1.0111, loss_cls_dn_5: 0.3775, loss_box_dn_5: 1.0383, loss_dense_depth: 1.1920, loss: 34.6644, grad_norm: 51.5358 -2025-11-13 10:52:44,835 - mmdet - INFO - Iter [54/17500] lr: 1.212e-04, eta: 20:28:44, time: 31.926, data_time: 0.083, memory: 49163, loss_cls_0: 1.0142, loss_box_0: 1.9471, loss_cns_0: 0.6187, loss_yns_0: 0.1647, loss_cls_1: 1.0570, loss_box_1: 2.3871, loss_cns_1: 0.6090, loss_yns_1: 0.1705, loss_cls_2: 1.0885, loss_box_2: 2.3535, loss_cns_2: 0.6192, loss_yns_2: 0.1641, loss_cls_3: 1.1126, loss_box_3: 2.3283, loss_cns_3: 0.6245, loss_yns_3: 0.1637, loss_cls_4: 1.1094, loss_box_4: 2.3359, loss_cns_4: 0.6234, loss_yns_4: 0.1626, loss_cls_5: 1.1360, loss_box_5: 2.4328, loss_cns_5: 0.6185, loss_yns_5: 0.1661, loss_cls_dn_0: 0.4094, loss_box_dn_0: 0.8988, loss_cls_dn_1: 0.2778, loss_box_dn_1: 1.0192, loss_cls_dn_2: 0.3040, loss_box_dn_2: 0.9742, loss_cls_dn_3: 0.3351, loss_box_dn_3: 0.9992, loss_cls_dn_4: 0.3393, loss_box_dn_4: 0.9972, loss_cls_dn_5: 0.3740, loss_box_dn_5: 1.0590, loss_dense_depth: 1.2150, loss: 34.2093, grad_norm: 48.3951 -2025-11-13 10:52:46,339 - mmdet - INFO - Iter [55/17500] lr: 1.216e-04, eta: 20:16:28, time: 1.919, data_time: 0.471, memory: 49163, loss_cls_0: 1.0095, loss_box_0: 1.9594, loss_cns_0: 0.6139, loss_yns_0: 0.1682, loss_cls_1: 1.0557, loss_box_1: 2.2797, loss_cns_1: 0.6173, loss_yns_1: 0.1737, loss_cls_2: 1.1016, loss_box_2: 2.2285, loss_cns_2: 0.6282, loss_yns_2: 0.1656, loss_cls_3: 1.1160, loss_box_3: 2.2055, loss_cns_3: 0.6280, loss_yns_3: 0.1656, loss_cls_4: 1.1143, loss_box_4: 2.1904, loss_cns_4: 0.6292, loss_yns_4: 0.1649, loss_cls_5: 1.1346, loss_box_5: 2.2825, loss_cns_5: 0.6208, loss_yns_5: 0.1677, loss_cls_dn_0: 0.4044, loss_box_dn_0: 0.8996, loss_cls_dn_1: 0.2993, loss_box_dn_1: 0.9795, loss_cls_dn_2: 0.3143, loss_box_dn_2: 0.9180, loss_cls_dn_3: 0.3458, loss_box_dn_3: 0.9437, loss_cls_dn_4: 0.3533, loss_box_dn_4: 0.9352, loss_cls_dn_5: 0.4049, loss_box_dn_5: 1.0006, loss_dense_depth: 1.0904, loss: 33.3099, grad_norm: 50.8328 -2025-11-13 10:52:47,867 - mmdet - INFO - Iter [56/17500] lr: 1.220e-04, eta: 20:02:37, time: 1.529, data_time: 0.074, memory: 49163, loss_cls_0: 0.9968, loss_box_0: 1.9215, loss_cns_0: 0.6102, loss_yns_0: 0.1649, loss_cls_1: 1.0376, loss_box_1: 2.1293, loss_cns_1: 0.6146, loss_yns_1: 0.1725, loss_cls_2: 1.0797, loss_box_2: 2.1281, loss_cns_2: 0.6209, loss_yns_2: 0.1646, loss_cls_3: 1.0952, loss_box_3: 2.1005, loss_cns_3: 0.6297, loss_yns_3: 0.1632, loss_cls_4: 1.0737, loss_box_4: 2.0962, loss_cns_4: 0.6302, loss_yns_4: 0.1647, loss_cls_5: 1.0933, loss_box_5: 2.1043, loss_cns_5: 0.6284, loss_yns_5: 0.1661, loss_cls_dn_0: 0.3798, loss_box_dn_0: 0.9001, loss_cls_dn_1: 0.2884, loss_box_dn_1: 0.9726, loss_cls_dn_2: 0.3019, loss_box_dn_2: 0.9271, loss_cls_dn_3: 0.3270, loss_box_dn_3: 0.9377, loss_cls_dn_4: 0.3306, loss_box_dn_4: 0.9310, loss_cls_dn_5: 0.3765, loss_box_dn_5: 0.9535, loss_dense_depth: 1.0917, loss: 32.3041, grad_norm: 44.3987 -2025-11-13 10:52:49,386 - mmdet - INFO - Iter [57/17500] lr: 1.224e-04, eta: 19:49:12, time: 1.521, data_time: 0.073, memory: 49163, loss_cls_0: 1.0372, loss_box_0: 1.8958, loss_cns_0: 0.6165, loss_yns_0: 0.1628, loss_cls_1: 1.0485, loss_box_1: 2.1617, loss_cns_1: 0.6110, loss_yns_1: 0.1681, loss_cls_2: 1.0630, loss_box_2: 2.1939, loss_cns_2: 0.6204, loss_yns_2: 0.1647, loss_cls_3: 1.0921, loss_box_3: 2.1593, loss_cns_3: 0.6323, loss_yns_3: 0.1645, loss_cls_4: 1.0944, loss_box_4: 2.1624, loss_cns_4: 0.6311, loss_yns_4: 0.1669, loss_cls_5: 1.1030, loss_box_5: 2.1461, loss_cns_5: 0.6366, loss_yns_5: 0.1658, loss_cls_dn_0: 0.3643, loss_box_dn_0: 0.8854, loss_cls_dn_1: 0.2772, loss_box_dn_1: 0.9241, loss_cls_dn_2: 0.2962, loss_box_dn_2: 0.9053, loss_cls_dn_3: 0.3117, loss_box_dn_3: 0.8958, loss_cls_dn_4: 0.3004, loss_box_dn_4: 0.8904, loss_cls_dn_5: 0.3486, loss_box_dn_5: 0.8942, loss_dense_depth: 1.0739, loss: 32.2657, grad_norm: 56.0448 diff --git a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20251113_104840.log.json b/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20251113_104840.log.json deleted file mode 100644 index e6465de064c232cd6137e49616116637b6f4d90a..0000000000000000000000000000000000000000 --- a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20251113_104840.log.json +++ /dev/null @@ -1,58 +0,0 @@ -{"env_info": "sys.platform: linux\nPython: 3.10.12 (main, May 27 2025, 17:12:29) [GCC 11.4.0]\nCUDA available: True\nGPU 0,1,2,3,4,5,6,7: BW200, UBB BW1000\nCUDA_HOME: /opt/dtk\nNVCC: Not Available\nGCC: x86_64-linux-gnu-gcc (Ubuntu 11.4.0-1ubuntu1~22.04) 11.4.0\nPyTorch: 2.4.1\nPyTorch compiling details: PyTorch built with:\n - GCC 10.3\n - C++ Version: 201703\n - Intel(R) Math Kernel Library Version 2020.0.4 Product Build 20200917 for Intel(R) 64 architecture applications\n - OpenMP 201511 (a.k.a. OpenMP 4.5)\n - LAPACK is enabled (usually provided by MKL)\n - NNPACK is enabled\n - CPU capability usage: AVX512\n - HIP Runtime 6.3.25211\n - MIOpen 2.17.0\n - Magma 2.8.0\n - Build settings: BLAS_INFO=mkl, BUILD_TYPE=Release, CXX_COMPILER=/opt/rh/gcc-toolset-10/root/usr/bin/c++, CXX_FLAGS= -D_GLIBCXX_USE_CXX11_ABI=1 -fvisibility-inlines-hidden -DUSE_PTHREADPOOL -DNDEBUG -DUSE_KINETO -DLIBKINETO_NOCUPTI -DUSE_FBGEMM -DUSE_PYTORCH_QNNPACK -DUSE_XNNPACK -DSYMBOLICATE_MOBILE_DEBUG_HANDLE -O2 -fPIC -Wall -Wextra -Werror=return-type -Werror=non-virtual-dtor -Werror=bool-operation -Wnarrowing -Wno-missing-field-initializers -Wno-type-limits -Wno-array-bounds -Wno-unknown-pragmas -Wno-unused-parameter -Wno-unused-function -Wno-unused-result -Wno-strict-overflow -Wno-strict-aliasing -Wno-stringop-overflow -Wsuggest-override -Wno-psabi -Wno-error=pedantic -Wno-error=old-style-cast -Wno-missing-braces -fdiagnostics-color=always -faligned-new -Wno-unused-but-set-variable -Wno-maybe-uninitialized -fno-math-errno -fno-trapping-math -Werror=format -Wno-stringop-overflow, FORCE_FALLBACK_CUDA_MPI=1, LAPACK_INFO=mkl, PERF_WITH_AVX=1, PERF_WITH_AVX2=1, PERF_WITH_AVX512=1, TORCH_VERSION=2.4.1, USE_CUDA=0, USE_CUDNN=OFF, USE_CUSPARSELT=OFF, USE_EXCEPTION_PTR=1, USE_GFLAGS=1, USE_GLOG=1, USE_GLOO=1, USE_MKL=ON, USE_MKLDNN=0, USE_MPI=1, USE_NCCL=1, USE_NNPACK=ON, USE_OPENMP=1, USE_ROCM=ON, USE_ROCM_KERNEL_ASSERT=OFF, \n\nTorchVision: 0.19.1\nOpenCV: 4.11.0\nMMCV: 1.6.1\nMMCV Compiler: GCC 11.4\nMMCV CUDA Compiler: rocm not available\nMMDetection: 2.26.0+c41df4b", "config": "plugin = True\nplugin_dir = 'projects/mmdet3d_plugin/'\ndist_params = dict(backend='nccl')\nlog_level = 'INFO'\nwork_dir = './work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704'\ntotal_batch_size = 160\nnum_gpus = 8\nbatch_size = 20\nnum_iters_per_epoch = 175\nnum_epochs = 100\ncheckpoint_epoch_interval = 20\ncheckpoint_config = dict(interval=3500)\nlog_config = dict(\n interval=1,\n hooks=[\n dict(type='TextLoggerHook', by_epoch=False),\n dict(type='TensorboardLoggerHook')\n ])\nload_from = None\nresume_from = None\nworkflow = [('train', 1)]\nfp16 = dict(loss_scale=32.0)\ninput_shape = (704, 256)\ntracking_test = True\ntracking_threshold = 0.2\nclass_names = [\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier',\n 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n]\nnum_classes = 10\nembed_dims = 256\nnum_groups = 8\nnum_decoder = 6\nnum_single_frame_decoder = 1\nuse_deformable_func = True\nstrides = [4, 8, 16, 32]\nnum_levels = 4\nnum_depth_layers = 3\ndrop_out = 0.1\ntemporal = True\ndecouple_attn = True\nwith_quality_estimation = True\nmodel = dict(\n type='Sparse4D',\n use_grid_mask=True,\n use_deformable_func=True,\n img_backbone=dict(\n type='ResNet',\n depth=50,\n num_stages=4,\n frozen_stages=-1,\n norm_eval=False,\n style='pytorch',\n with_cp=True,\n out_indices=(0, 1, 2, 3),\n norm_cfg=dict(type='BN', requires_grad=True),\n pretrained='ckpt/resnet50-19c8e357.pth'),\n img_neck=dict(\n type='FPN',\n num_outs=4,\n start_level=0,\n out_channels=256,\n add_extra_convs='on_output',\n relu_before_extra_convs=True,\n in_channels=[256, 512, 1024, 2048]),\n depth_branch=dict(\n type='DenseDepthNet',\n embed_dims=256,\n num_depth_layers=3,\n loss_weight=0.2),\n head=dict(\n type='Sparse4DHead',\n cls_threshold_to_reg=0.05,\n decouple_attn=True,\n instance_bank=dict(\n type='InstanceBank',\n num_anchor=900,\n embed_dims=256,\n anchor='nuscenes_kmeans900.npy',\n anchor_handler=dict(type='SparseBox3DKeyPointsGenerator'),\n num_temp_instances=600,\n confidence_decay=0.6,\n feat_grad=False),\n anchor_encoder=dict(\n type='SparseBox3DEncoder',\n vel_dims=3,\n embed_dims=[128, 32, 32, 64],\n mode='cat',\n output_fc=False,\n in_loops=1,\n out_loops=4),\n num_single_frame_decoder=1,\n operation_order=[\n 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm',\n 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm',\n 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm',\n 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm',\n 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm',\n 'deformable', 'ffn', 'norm', 'refine'\n ],\n temp_graph_model=dict(\n type='MultiheadAttention',\n embed_dims=512,\n num_heads=8,\n batch_first=True,\n dropout=0.1),\n graph_model=dict(\n type='MultiheadAttention',\n embed_dims=512,\n num_heads=8,\n batch_first=True,\n dropout=0.1),\n norm_layer=dict(type='LN', normalized_shape=256),\n ffn=dict(\n type='AsymmetricFFN',\n in_channels=512,\n pre_norm=dict(type='LN'),\n embed_dims=256,\n feedforward_channels=1024,\n num_fcs=2,\n ffn_drop=0.1,\n act_cfg=dict(type='ReLU', inplace=True)),\n deformable_model=dict(\n type='DeformableFeatureAggregation',\n embed_dims=256,\n num_groups=8,\n num_levels=4,\n num_cams=6,\n attn_drop=0.15,\n use_deformable_func=True,\n use_camera_embed=True,\n residual_mode='cat',\n kps_generator=dict(\n type='SparseBox3DKeyPointsGenerator',\n num_learnable_pts=6,\n fix_scale=[[0, 0, 0], [0.45, 0, 0], [-0.45, 0, 0],\n [0, 0.45, 0], [0, -0.45, 0], [0, 0, 0.45],\n [0, 0, -0.45]])),\n refine_layer=dict(\n type='SparseBox3DRefinementModule',\n embed_dims=256,\n num_cls=10,\n refine_yaw=True,\n with_quality_estimation=True),\n sampler=dict(\n type='SparseBox3DTarget',\n num_dn_groups=5,\n num_temp_dn_groups=3,\n dn_noise_scale=[2.0, 2.0, 2.0, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5],\n max_dn_gt=32,\n add_neg_dn=True,\n cls_weight=2.0,\n box_weight=0.25,\n reg_weights=[2.0, 2.0, 2.0, 0.5, 0.5, 0.5, 0.0, 0.0, 0.0, 0.0],\n cls_wise_reg_weights=dict(\n {9: [2.0, 2.0, 2.0, 1.0, 1.0, 1.0, 0.0, 0.0, 1.0, 1.0]})),\n loss_cls=dict(\n type='FocalLoss',\n use_sigmoid=True,\n gamma=2.0,\n alpha=0.25,\n loss_weight=2.0),\n loss_reg=dict(\n type='SparseBox3DLoss',\n loss_box=dict(type='L1Loss', loss_weight=0.25),\n loss_centerness=dict(type='CrossEntropyLoss', use_sigmoid=True),\n loss_yawness=dict(type='GaussianFocalLoss'),\n cls_allow_reverse=[5]),\n decoder=dict(type='SparseBox3DDecoder'),\n reg_weights=[2.0, 2.0, 2.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0]))\ndataset_type = 'NuScenes3DDetTrackDataset'\ndata_root = 'data/nuscenes/'\nanno_root = 'data/nuscenes_anno_pkls/'\nfile_client_args = dict(backend='disk')\nimg_norm_cfg = dict(\n mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375], to_rgb=True)\ntrain_pipeline = [\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(\n type='LoadPointsFromFile',\n coord_type='LIDAR',\n load_dim=5,\n use_dim=5,\n file_client_args=dict(backend='disk')),\n dict(type='ResizeCropFlipImage'),\n dict(type='MultiScaleDepthMapGenerator', downsample=[4, 8, 16]),\n dict(type='BBoxRotation'),\n dict(type='PhotoMetricDistortionMultiViewImage'),\n dict(\n type='NormalizeMultiviewImage',\n mean=[123.675, 116.28, 103.53],\n std=[58.395, 57.12, 57.375],\n to_rgb=True),\n dict(\n type='CircleObjectRangeFilter',\n class_dist_thred=[55, 55, 55, 55, 55, 55, 55, 55, 55, 55]),\n dict(\n type='InstanceNameFilter',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ]),\n dict(type='NuScenesSparse4DAdaptor'),\n dict(\n type='Collect',\n keys=[\n 'img', 'timestamp', 'projection_mat', 'image_wh', 'gt_depth',\n 'focal', 'gt_bboxes_3d', 'gt_labels_3d'\n ],\n meta_keys=['T_global', 'T_global_inv', 'timestamp', 'instance_id'])\n]\ntest_pipeline = [\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(type='ResizeCropFlipImage'),\n dict(\n type='NormalizeMultiviewImage',\n mean=[123.675, 116.28, 103.53],\n std=[58.395, 57.12, 57.375],\n to_rgb=True),\n dict(type='NuScenesSparse4DAdaptor'),\n dict(\n type='Collect',\n keys=['img', 'timestamp', 'projection_mat', 'image_wh'],\n meta_keys=['T_global', 'T_global_inv', 'timestamp'])\n]\ninput_modality = dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False)\ndata_basic_config = dict(\n type='NuScenes3DDetTrackDataset',\n data_root='data/nuscenes/',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier',\n 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ],\n modality=dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False),\n version='v1.0-trainval')\ndata_aug_conf = dict(\n resize_lim=(0.4, 0.47),\n final_dim=(256, 704),\n bot_pct_lim=(0.0, 0.0),\n rot_lim=(-5.4, 5.4),\n H=900,\n W=1600,\n rand_flip=True,\n rot3d_range=[-0.3925, 0.3925])\ndata = dict(\n samples_per_gpu=20,\n workers_per_gpu=20,\n train=dict(\n type='NuScenes3DDetTrackDataset',\n data_root='data/nuscenes/',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ],\n modality=dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False),\n version='v1.0-trainval',\n ann_file='data/nuscenes_anno_pkls/nuscenes_infos_train.pkl',\n pipeline=[\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(\n type='LoadPointsFromFile',\n coord_type='LIDAR',\n load_dim=5,\n use_dim=5,\n file_client_args=dict(backend='disk')),\n dict(type='ResizeCropFlipImage'),\n dict(type='MultiScaleDepthMapGenerator', downsample=[4, 8, 16]),\n dict(type='BBoxRotation'),\n dict(type='PhotoMetricDistortionMultiViewImage'),\n dict(\n type='NormalizeMultiviewImage',\n mean=[123.675, 116.28, 103.53],\n std=[58.395, 57.12, 57.375],\n to_rgb=True),\n dict(\n type='CircleObjectRangeFilter',\n class_dist_thred=[55, 55, 55, 55, 55, 55, 55, 55, 55, 55]),\n dict(\n type='InstanceNameFilter',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian',\n 'traffic_cone'\n ]),\n dict(type='NuScenesSparse4DAdaptor'),\n dict(\n type='Collect',\n keys=[\n 'img', 'timestamp', 'projection_mat', 'image_wh',\n 'gt_depth', 'focal', 'gt_bboxes_3d', 'gt_labels_3d'\n ],\n meta_keys=[\n 'T_global', 'T_global_inv', 'timestamp', 'instance_id'\n ])\n ],\n test_mode=False,\n data_aug_conf=dict(\n resize_lim=(0.4, 0.47),\n final_dim=(256, 704),\n bot_pct_lim=(0.0, 0.0),\n rot_lim=(-5.4, 5.4),\n H=900,\n W=1600,\n rand_flip=True,\n rot3d_range=[-0.3925, 0.3925]),\n with_seq_flag=True,\n sequences_split_num=2,\n keep_consistent_seq_aug=True),\n val=dict(\n type='NuScenes3DDetTrackDataset',\n data_root='data/nuscenes/',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ],\n modality=dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False),\n version='v1.0-trainval',\n ann_file='data/nuscenes_anno_pkls/nuscenes_infos_val.pkl',\n pipeline=[\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(type='ResizeCropFlipImage'),\n dict(\n type='NormalizeMultiviewImage',\n mean=[123.675, 116.28, 103.53],\n std=[58.395, 57.12, 57.375],\n to_rgb=True),\n dict(type='NuScenesSparse4DAdaptor'),\n dict(\n type='Collect',\n keys=['img', 'timestamp', 'projection_mat', 'image_wh'],\n meta_keys=['T_global', 'T_global_inv', 'timestamp'])\n ],\n data_aug_conf=dict(\n resize_lim=(0.4, 0.47),\n final_dim=(256, 704),\n bot_pct_lim=(0.0, 0.0),\n rot_lim=(-5.4, 5.4),\n H=900,\n W=1600,\n rand_flip=True,\n rot3d_range=[-0.3925, 0.3925]),\n test_mode=True,\n tracking=True,\n tracking_threshold=0.2),\n test=dict(\n type='NuScenes3DDetTrackDataset',\n data_root='data/nuscenes/',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ],\n modality=dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False),\n version='v1.0-trainval',\n ann_file='data/nuscenes_anno_pkls/nuscenes_infos_val.pkl',\n pipeline=[\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(type='ResizeCropFlipImage'),\n dict(\n type='NormalizeMultiviewImage',\n mean=[123.675, 116.28, 103.53],\n std=[58.395, 57.12, 57.375],\n to_rgb=True),\n dict(type='NuScenesSparse4DAdaptor'),\n dict(\n type='Collect',\n keys=['img', 'timestamp', 'projection_mat', 'image_wh'],\n meta_keys=['T_global', 'T_global_inv', 'timestamp'])\n ],\n data_aug_conf=dict(\n resize_lim=(0.4, 0.47),\n final_dim=(256, 704),\n bot_pct_lim=(0.0, 0.0),\n rot_lim=(-5.4, 5.4),\n H=900,\n W=1600,\n rand_flip=True,\n rot3d_range=[-0.3925, 0.3925]),\n test_mode=True,\n tracking=True,\n tracking_threshold=0.2))\noptimizer = dict(\n type='Miopen_AdamW',\n lr=0.0006,\n weight_decay=0.001,\n paramwise_cfg=dict(custom_keys=dict(img_backbone=dict(lr_mult=0.5))))\noptimizer_config = dict(grad_clip=dict(max_norm=25, norm_type=2))\nlr_config = dict(\n policy='CosineAnnealing',\n warmup='linear',\n warmup_iters=500,\n warmup_ratio=0.3333333333333333,\n min_lr_ratio=0.001)\nrunner = dict(type='IterBasedRunner', max_iters=17500)\nvis_pipeline = [\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(type='Collect', keys=['img'], meta_keys=['timestamp', 'lidar2img'])\n]\nevaluation = dict(\n interval=3500,\n pipeline=[\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(\n type='Collect', keys=['img'], meta_keys=['timestamp', 'lidar2img'])\n ])\ngpu_ids = range(0, 8)\n", "seed": 0, "exp_name": "sparse4dv3_temporal_r50_1x8_bs6_256x704.py"} -{"mode": "train", "epoch": 1, "iter": 1, "lr": 0.0001, "memory": 49163, "data_time": 10.38666, "loss_cls_0": 2.36126, "loss_box_0": 0.01384, "loss_cns_0": 0.0027, "loss_yns_0": 0.00079, "loss_cls_1": 2.1544, "loss_box_1": 0.10891, "loss_cns_1": 0.02493, "loss_yns_1": 0.0067, "loss_cls_2": 2.31194, "loss_box_2": 0.00503, "loss_cns_2": 0.00059, "loss_yns_2": 0.00029, "loss_cls_3": 2.39029, "loss_box_3": 0.02944, "loss_cns_3": 0.00504, "loss_yns_3": 0.00144, "loss_cls_4": 2.02754, "loss_box_4": 0.42132, "loss_cns_4": 0.05423, "loss_yns_4": 0.02569, "loss_cls_5": 2.42479, "loss_box_5": 0.018, "loss_cns_5": 0.00217, "loss_yns_5": 0.00161, "loss_cls_dn_0": 1.19803, "loss_box_dn_0": 1.46028, "loss_cls_dn_1": 1.11019, "loss_box_dn_1": 1.73176, "loss_cls_dn_2": 1.17413, "loss_box_dn_2": 1.97186, "loss_cls_dn_3": 1.17207, "loss_box_dn_3": 2.24177, "loss_cls_dn_4": 1.05279, "loss_box_dn_4": 2.42686, "loss_cls_dn_5": 1.23867, "loss_box_dn_5": 2.67735, "loss_dense_depth": 1.86432, "loss": 35.71299, "grad_norm": 271.34122, "time": 115.5709} -{"mode": "train", "epoch": 1, "iter": 2, "lr": 0.0001, "memory": 49163, "data_time": 0.06849, "loss_cls_0": 2.08996, "loss_box_0": 0.01018, "loss_cns_0": 0.0027, "loss_yns_0": 0.00087, "loss_cls_1": 2.03729, "loss_box_1": 0.10421, "loss_cns_1": 0.02027, "loss_yns_1": 0.00514, "loss_cls_2": 2.11448, "loss_box_2": 0.22716, "loss_cns_2": 0.02058, "loss_yns_2": 0.00936, "loss_cls_3": 1.96068, "loss_box_3": 0.42916, "loss_cns_3": 0.0549, "loss_yns_3": 0.01916, "loss_cls_4": 1.80132, "loss_box_4": 1.54439, "loss_cns_4": 0.15391, "loss_yns_4": 0.05609, "loss_cls_5": 2.05975, "loss_box_5": 0.53001, "loss_cns_5": 0.05825, "loss_yns_5": 0.01904, "loss_cls_dn_0": 1.04823, "loss_box_dn_0": 1.31086, "loss_cls_dn_1": 0.95938, "loss_box_dn_1": 2.42438, "loss_cls_dn_2": 0.97452, "loss_box_dn_2": 2.53788, "loss_cls_dn_3": 0.91335, "loss_box_dn_3": 2.62098, "loss_cls_dn_4": 0.84104, "loss_box_dn_4": 2.88015, "loss_cls_dn_5": 0.98618, "loss_box_dn_5": 3.12045, "loss_dense_depth": 1.71234, "loss": 37.6586, "grad_norm": 66.14442, "time": 1.88562} -{"mode": "train", "epoch": 1, "iter": 3, "lr": 0.0001, "memory": 49163, "data_time": 0.07952, "loss_cls_0": 1.41819, "loss_box_0": 2.58109, "loss_cns_0": 0.61005, "loss_yns_0": 0.23792, "loss_cls_1": 1.73189, "loss_box_1": 2.15855, "loss_cns_1": 0.31132, "loss_yns_1": 0.11905, "loss_cls_2": 1.75409, "loss_box_2": 4.43034, "loss_cns_2": 0.37553, "loss_yns_2": 0.21473, "loss_cls_3": 1.60427, "loss_box_3": 5.20675, "loss_cns_3": 0.43019, "loss_yns_3": 0.21277, "loss_cls_4": 1.52403, "loss_box_4": 4.99373, "loss_cns_4": 0.41271, "loss_yns_4": 0.19786, "loss_cls_5": 1.62182, "loss_box_5": 4.0515, "loss_cns_5": 0.2919, "loss_yns_5": 0.12999, "loss_cls_dn_0": 0.6716, "loss_box_dn_0": 1.18468, "loss_cls_dn_1": 0.80755, "loss_box_dn_1": 2.45199, "loss_cls_dn_2": 0.77046, "loss_box_dn_2": 2.67515, "loss_cls_dn_3": 0.67363, "loss_box_dn_3": 2.88081, "loss_cls_dn_4": 0.68936, "loss_box_dn_4": 3.16147, "loss_cls_dn_5": 0.77643, "loss_box_dn_5": 3.44829, "loss_dense_depth": 1.62226, "loss": 58.43395, "grad_norm": 105.25195, "time": 1.62363} -{"mode": "train", "epoch": 1, "iter": 4, "lr": 0.0001, "memory": 49163, "data_time": 0.078, "loss_cls_0": 1.40216, "loss_box_0": 2.71598, "loss_cns_0": 0.52351, "loss_yns_0": 0.22102, "loss_cls_1": 1.56695, "loss_box_1": 3.64943, "loss_cns_1": 0.46073, "loss_yns_1": 0.21718, "loss_cls_2": 1.73552, "loss_box_2": 3.7819, "loss_cns_2": 0.44496, "loss_yns_2": 0.20184, "loss_cls_3": 1.47229, "loss_box_3": 4.27206, "loss_cns_3": 0.46926, "loss_yns_3": 0.19817, "loss_cls_4": 1.53604, "loss_box_4": 4.66993, "loss_cns_4": 0.4215, "loss_yns_4": 0.19844, "loss_cls_5": 1.39722, "loss_box_5": 5.19893, "loss_cns_5": 0.4438, "loss_yns_5": 0.19121, "loss_cls_dn_0": 0.52926, "loss_box_dn_0": 1.23931, "loss_cls_dn_1": 0.6709, "loss_box_dn_1": 2.68004, "loss_cls_dn_2": 0.6413, "loss_box_dn_2": 2.78695, "loss_cls_dn_3": 0.57503, "loss_box_dn_3": 3.04095, "loss_cls_dn_4": 0.5365, "loss_box_dn_4": 3.25233, "loss_cls_dn_5": 0.59505, "loss_box_dn_5": 3.51146, "loss_dense_depth": 1.69852, "loss": 59.14764, "grad_norm": 120.46623, "time": 1.52517} -{"mode": "train", "epoch": 1, "iter": 5, "lr": 0.0001, "memory": 49163, "data_time": 0.07237, "loss_cls_0": 1.40642, "loss_box_0": 2.73225, "loss_cns_0": 0.53574, "loss_yns_0": 0.22043, "loss_cls_1": 1.5113, "loss_box_1": 4.11211, "loss_cns_1": 0.42041, "loss_yns_1": 0.23462, "loss_cls_2": 1.4997, "loss_box_2": 4.21473, "loss_cns_2": 0.39865, "loss_yns_2": 0.18718, "loss_cls_3": 1.35175, "loss_box_3": 4.35992, "loss_cns_3": 0.40201, "loss_yns_3": 0.19086, "loss_cls_4": 1.32751, "loss_box_4": 4.69833, "loss_cns_4": 0.36683, "loss_yns_4": 0.20478, "loss_cls_5": 1.35898, "loss_box_5": 4.88619, "loss_cns_5": 0.34692, "loss_yns_5": 0.20412, "loss_cls_dn_0": 0.55325, "loss_box_dn_0": 1.25794, "loss_cls_dn_1": 0.5919, "loss_box_dn_1": 2.38904, "loss_cls_dn_2": 0.64966, "loss_box_dn_2": 2.57439, "loss_cls_dn_3": 0.53355, "loss_box_dn_3": 2.68023, "loss_cls_dn_4": 0.53458, "loss_box_dn_4": 2.9078, "loss_cls_dn_5": 0.49541, "loss_box_dn_5": 2.92068, "loss_dense_depth": 1.77843, "loss": 57.0386, "grad_norm": 124.6013, "time": 1.54766} -{"mode": "train", "epoch": 1, "iter": 6, "lr": 0.0001, "memory": 49163, "data_time": 0.07751, "loss_cls_0": 1.26907, "loss_box_0": 2.34412, "loss_cns_0": 0.66158, "loss_yns_0": 0.18123, "loss_cls_1": 1.32452, "loss_box_1": 3.72592, "loss_cns_1": 0.44901, "loss_yns_1": 0.19731, "loss_cls_2": 1.3757, "loss_box_2": 3.84354, "loss_cns_2": 0.45254, "loss_yns_2": 0.19713, "loss_cls_3": 1.29781, "loss_box_3": 3.66836, "loss_cns_3": 0.47913, "loss_yns_3": 0.20911, "loss_cls_4": 1.31216, "loss_box_4": 3.93929, "loss_cns_4": 0.4516, "loss_yns_4": 0.19816, "loss_cls_5": 1.3488, "loss_box_5": 4.30778, "loss_cns_5": 0.40874, "loss_yns_5": 0.18876, "loss_cls_dn_0": 0.56529, "loss_box_dn_0": 1.13281, "loss_cls_dn_1": 0.52278, "loss_box_dn_1": 2.52663, "loss_cls_dn_2": 0.57036, "loss_box_dn_2": 2.52046, "loss_cls_dn_3": 0.4778, "loss_box_dn_3": 2.5446, "loss_cls_dn_4": 0.4885, "loss_box_dn_4": 2.74164, "loss_cls_dn_5": 0.43682, "loss_box_dn_5": 2.873, "loss_dense_depth": 2.0182, "loss": 53.25024, "grad_norm": 106.43399, "time": 1.56927} -{"mode": "train", "epoch": 1, "iter": 7, "lr": 0.0001, "memory": 49163, "data_time": 0.08165, "loss_cls_0": 1.24535, "loss_box_0": 2.28299, "loss_cns_0": 0.62184, "loss_yns_0": 0.19629, "loss_cls_1": 1.26977, "loss_box_1": 3.67209, "loss_cns_1": 0.42501, "loss_yns_1": 0.18496, "loss_cls_2": 1.42726, "loss_box_2": 3.84039, "loss_cns_2": 0.37117, "loss_yns_2": 0.1854, "loss_cls_3": 1.28204, "loss_box_3": 3.69735, "loss_cns_3": 0.38609, "loss_yns_3": 0.18307, "loss_cls_4": 1.30904, "loss_box_4": 3.74931, "loss_cns_4": 0.39881, "loss_yns_4": 0.20223, "loss_cls_5": 1.32759, "loss_box_5": 3.89267, "loss_cns_5": 0.44296, "loss_yns_5": 0.18424, "loss_cls_dn_0": 0.54412, "loss_box_dn_0": 1.04902, "loss_cls_dn_1": 0.45582, "loss_box_dn_1": 2.37408, "loss_cls_dn_2": 0.47441, "loss_box_dn_2": 2.37148, "loss_cls_dn_3": 0.42867, "loss_box_dn_3": 2.33071, "loss_cls_dn_4": 0.43713, "loss_box_dn_4": 2.37347, "loss_cls_dn_5": 0.39915, "loss_box_dn_5": 2.46052, "loss_dense_depth": 1.71168, "loss": 50.18822, "grad_norm": 89.6524, "time": 1.51262} -{"mode": "train", "epoch": 1, "iter": 8, "lr": 0.0001, "memory": 49163, "data_time": 0.07764, "loss_cls_0": 1.31587, "loss_box_0": 2.19876, "loss_cns_0": 0.6176, "loss_yns_0": 0.1781, "loss_cls_1": 1.25633, "loss_box_1": 3.30249, "loss_cns_1": 0.5097, "loss_yns_1": 0.18025, "loss_cls_2": 1.31703, "loss_box_2": 3.37275, "loss_cns_2": 0.46021, "loss_yns_2": 0.18304, "loss_cls_3": 1.26081, "loss_box_3": 3.4652, "loss_cns_3": 0.50077, "loss_yns_3": 0.21577, "loss_cls_4": 1.26107, "loss_box_4": 3.4037, "loss_cns_4": 0.48776, "loss_yns_4": 0.19278, "loss_cls_5": 1.28334, "loss_box_5": 3.46749, "loss_cns_5": 0.50828, "loss_yns_5": 0.18187, "loss_cls_dn_0": 0.48331, "loss_box_dn_0": 1.04228, "loss_cls_dn_1": 0.44978, "loss_box_dn_1": 1.58953, "loss_cls_dn_2": 0.49477, "loss_box_dn_2": 1.62611, "loss_cls_dn_3": 0.4453, "loss_box_dn_3": 1.75643, "loss_cls_dn_4": 0.45404, "loss_box_dn_4": 1.77701, "loss_cls_dn_5": 0.44147, "loss_box_dn_5": 1.90663, "loss_dense_depth": 2.65163, "loss": 46.23927, "grad_norm": 89.86046, "time": 1.51778} -{"mode": "train", "epoch": 1, "iter": 9, "lr": 0.0001, "memory": 49163, "data_time": 0.08, "loss_cls_0": 1.21998, "loss_box_0": 2.30472, "loss_cns_0": 0.62826, "loss_yns_0": 0.18097, "loss_cls_1": 1.27364, "loss_box_1": 3.1716, "loss_cns_1": 0.4992, "loss_yns_1": 0.17865, "loss_cls_2": 1.29241, "loss_box_2": 3.13907, "loss_cns_2": 0.49262, "loss_yns_2": 0.18817, "loss_cls_3": 1.26389, "loss_box_3": 3.41295, "loss_cns_3": 0.49797, "loss_yns_3": 0.18197, "loss_cls_4": 1.27091, "loss_box_4": 3.35725, "loss_cns_4": 0.49678, "loss_yns_4": 0.182, "loss_cls_5": 1.25665, "loss_box_5": 3.57929, "loss_cns_5": 0.47298, "loss_yns_5": 0.1844, "loss_cls_dn_0": 0.49373, "loss_box_dn_0": 1.08329, "loss_cls_dn_1": 0.41786, "loss_box_dn_1": 1.59159, "loss_cls_dn_2": 0.47653, "loss_box_dn_2": 1.62326, "loss_cls_dn_3": 0.43129, "loss_box_dn_3": 1.88916, "loss_cls_dn_4": 0.43227, "loss_box_dn_4": 1.92224, "loss_cls_dn_5": 0.45849, "loss_box_dn_5": 2.1507, "loss_dense_depth": 2.12369, "loss": 45.82043, "grad_norm": 86.90611, "time": 1.52624} -{"mode": "train", "epoch": 1, "iter": 10, "lr": 0.0001, "memory": 49163, "data_time": 0.09115, "loss_cls_0": 1.24408, "loss_box_0": 2.30863, "loss_cns_0": 0.60282, "loss_yns_0": 0.17783, "loss_cls_1": 1.25164, "loss_box_1": 3.17714, "loss_cns_1": 0.46857, "loss_yns_1": 0.17679, "loss_cls_2": 1.27801, "loss_box_2": 3.03677, "loss_cns_2": 0.47165, "loss_yns_2": 0.18101, "loss_cls_3": 1.22576, "loss_box_3": 3.09048, "loss_cns_3": 0.47978, "loss_yns_3": 0.19126, "loss_cls_4": 1.25957, "loss_box_4": 3.07444, "loss_cns_4": 0.48784, "loss_yns_4": 0.20335, "loss_cls_5": 1.24463, "loss_box_5": 3.34201, "loss_cns_5": 0.47899, "loss_yns_5": 0.18142, "loss_cls_dn_0": 0.48899, "loss_box_dn_0": 1.08199, "loss_cls_dn_1": 0.40289, "loss_box_dn_1": 1.83428, "loss_cls_dn_2": 0.43994, "loss_box_dn_2": 1.78385, "loss_cls_dn_3": 0.42411, "loss_box_dn_3": 1.89787, "loss_cls_dn_4": 0.40664, "loss_box_dn_4": 1.90609, "loss_cls_dn_5": 0.43542, "loss_box_dn_5": 2.09087, "loss_dense_depth": 2.3932, "loss": 45.22062, "grad_norm": 69.24619, "time": 1.52647} -{"mode": "train", "epoch": 1, "iter": 11, "lr": 0.0001, "memory": 49163, "data_time": 0.07669, "loss_cls_0": 1.22662, "loss_box_0": 2.27969, "loss_cns_0": 0.59749, "loss_yns_0": 0.17259, "loss_cls_1": 1.2343, "loss_box_1": 2.88421, "loss_cns_1": 0.52238, "loss_yns_1": 0.17744, "loss_cls_2": 1.24407, "loss_box_2": 2.96259, "loss_cns_2": 0.526, "loss_yns_2": 0.18537, "loss_cls_3": 1.26179, "loss_box_3": 3.01058, "loss_cns_3": 0.52675, "loss_yns_3": 0.17393, "loss_cls_4": 1.27363, "loss_box_4": 3.05642, "loss_cns_4": 0.49053, "loss_yns_4": 0.18077, "loss_cls_5": 1.28461, "loss_box_5": 3.06891, "loss_cns_5": 0.50385, "loss_yns_5": 0.17656, "loss_cls_dn_0": 0.46304, "loss_box_dn_0": 1.06572, "loss_cls_dn_1": 0.40032, "loss_box_dn_1": 1.77053, "loss_cls_dn_2": 0.41752, "loss_box_dn_2": 1.76212, "loss_cls_dn_3": 0.40641, "loss_box_dn_3": 1.83526, "loss_cls_dn_4": 0.39349, "loss_box_dn_4": 1.95201, "loss_cls_dn_5": 0.40242, "loss_box_dn_5": 1.98029, "loss_dense_depth": 2.4526, "loss": 44.32278, "grad_norm": 74.04531, "time": 1.52689} -{"mode": "train", "epoch": 1, "iter": 12, "lr": 0.0001, "memory": 49163, "data_time": 0.0813, "loss_cls_0": 1.23372, "loss_box_0": 2.22706, "loss_cns_0": 0.60954, "loss_yns_0": 0.17552, "loss_cls_1": 1.24214, "loss_box_1": 2.80488, "loss_cns_1": 0.5428, "loss_yns_1": 0.17409, "loss_cls_2": 1.27277, "loss_box_2": 3.01442, "loss_cns_2": 0.53708, "loss_yns_2": 0.17773, "loss_cls_3": 1.25772, "loss_box_3": 3.10137, "loss_cns_3": 0.52329, "loss_yns_3": 0.1841, "loss_cls_4": 1.26011, "loss_box_4": 3.06016, "loss_cns_4": 0.50284, "loss_yns_4": 0.20778, "loss_cls_5": 1.28076, "loss_box_5": 3.20045, "loss_cns_5": 0.48874, "loss_yns_5": 0.17911, "loss_cls_dn_0": 0.45008, "loss_box_dn_0": 1.04332, "loss_cls_dn_1": 0.38484, "loss_box_dn_1": 1.78973, "loss_cls_dn_2": 0.392, "loss_box_dn_2": 1.8391, "loss_cls_dn_3": 0.36889, "loss_box_dn_3": 1.87437, "loss_cls_dn_4": 0.38452, "loss_box_dn_4": 1.96214, "loss_cls_dn_5": 0.3857, "loss_box_dn_5": 2.01222, "loss_dense_depth": 2.43077, "loss": 44.57586, "grad_norm": 82.59117, "time": 1.52111} -{"mode": "train", "epoch": 1, "iter": 13, "lr": 0.0001, "memory": 49163, "data_time": 0.08121, "loss_cls_0": 1.17226, "loss_box_0": 2.34563, "loss_cns_0": 0.58477, "loss_yns_0": 0.17348, "loss_cls_1": 1.22668, "loss_box_1": 2.92089, "loss_cns_1": 0.54516, "loss_yns_1": 0.17244, "loss_cls_2": 1.2517, "loss_box_2": 3.07524, "loss_cns_2": 0.53754, "loss_yns_2": 0.17344, "loss_cls_3": 1.25179, "loss_box_3": 3.11464, "loss_cns_3": 0.54976, "loss_yns_3": 0.17953, "loss_cls_4": 1.24043, "loss_box_4": 3.22191, "loss_cns_4": 0.52976, "loss_yns_4": 0.18328, "loss_cls_5": 1.24198, "loss_box_5": 3.22531, "loss_cns_5": 0.53911, "loss_yns_5": 0.17587, "loss_cls_dn_0": 0.46701, "loss_box_dn_0": 1.05965, "loss_cls_dn_1": 0.42834, "loss_box_dn_1": 1.48612, "loss_cls_dn_2": 0.43769, "loss_box_dn_2": 1.6043, "loss_cls_dn_3": 0.4144, "loss_box_dn_3": 1.66213, "loss_cls_dn_4": 0.44382, "loss_box_dn_4": 1.81222, "loss_cls_dn_5": 0.44926, "loss_box_dn_5": 1.88438, "loss_dense_depth": 2.33347, "loss": 44.11539, "grad_norm": 107.19473, "time": 1.54961} -{"mode": "train", "epoch": 1, "iter": 14, "lr": 0.00011, "memory": 49163, "data_time": 0.08056, "loss_cls_0": 1.17259, "loss_box_0": 2.43369, "loss_cns_0": 0.58065, "loss_yns_0": 0.17444, "loss_cls_1": 1.2268, "loss_box_1": 2.89793, "loss_cns_1": 0.53103, "loss_yns_1": 0.17556, "loss_cls_2": 1.25565, "loss_box_2": 3.03199, "loss_cns_2": 0.50841, "loss_yns_2": 0.17716, "loss_cls_3": 1.2613, "loss_box_3": 3.06259, "loss_cns_3": 0.51003, "loss_yns_3": 0.1745, "loss_cls_4": 1.26201, "loss_box_4": 3.12728, "loss_cns_4": 0.49857, "loss_yns_4": 0.17993, "loss_cls_5": 1.27487, "loss_box_5": 3.1921, "loss_cns_5": 0.50371, "loss_yns_5": 0.17421, "loss_cls_dn_0": 0.48232, "loss_box_dn_0": 1.08461, "loss_cls_dn_1": 0.447, "loss_box_dn_1": 1.40075, "loss_cls_dn_2": 0.45782, "loss_box_dn_2": 1.52331, "loss_cls_dn_3": 0.45697, "loss_box_dn_3": 1.60491, "loss_cls_dn_4": 0.47246, "loss_box_dn_4": 1.71983, "loss_cls_dn_5": 0.47907, "loss_box_dn_5": 1.84392, "loss_dense_depth": 2.21522, "loss": 43.57521, "grad_norm": 104.9707, "time": 1.52618} -{"mode": "train", "epoch": 1, "iter": 15, "lr": 0.00011, "memory": 49163, "data_time": 0.0764, "loss_cls_0": 1.18486, "loss_box_0": 2.4341, "loss_cns_0": 0.59978, "loss_yns_0": 0.18062, "loss_cls_1": 1.23198, "loss_box_1": 2.71308, "loss_cns_1": 0.55943, "loss_yns_1": 0.18055, "loss_cls_2": 1.27223, "loss_box_2": 2.79776, "loss_cns_2": 0.56002, "loss_yns_2": 0.18399, "loss_cls_3": 1.26294, "loss_box_3": 2.85383, "loss_cns_3": 0.56875, "loss_yns_3": 0.1798, "loss_cls_4": 1.27956, "loss_box_4": 2.8516, "loss_cns_4": 0.56195, "loss_yns_4": 0.18718, "loss_cls_5": 1.28584, "loss_box_5": 2.85952, "loss_cns_5": 0.56621, "loss_yns_5": 0.18737, "loss_cls_dn_0": 0.468, "loss_box_dn_0": 1.10209, "loss_cls_dn_1": 0.44595, "loss_box_dn_1": 1.42259, "loss_cls_dn_2": 0.444, "loss_box_dn_2": 1.50769, "loss_cls_dn_3": 0.45512, "loss_box_dn_3": 1.57689, "loss_cls_dn_4": 0.45187, "loss_box_dn_4": 1.6111, "loss_cls_dn_5": 0.47035, "loss_box_dn_5": 1.73234, "loss_dense_depth": 2.09841, "loss": 42.32936, "grad_norm": 81.37077, "time": 1.51532} -{"mode": "train", "epoch": 1, "iter": 16, "lr": 0.00011, "memory": 49163, "data_time": 0.07905, "loss_cls_0": 1.19252, "loss_box_0": 2.31732, "loss_cns_0": 0.60351, "loss_yns_0": 0.17582, "loss_cls_1": 1.23544, "loss_box_1": 2.67501, "loss_cns_1": 0.58518, "loss_yns_1": 0.17853, "loss_cls_2": 1.26965, "loss_box_2": 2.71438, "loss_cns_2": 0.56057, "loss_yns_2": 0.17938, "loss_cls_3": 1.2542, "loss_box_3": 2.71403, "loss_cns_3": 0.55728, "loss_yns_3": 0.18258, "loss_cls_4": 1.26761, "loss_box_4": 2.69004, "loss_cns_4": 0.54779, "loss_yns_4": 0.17521, "loss_cls_5": 1.24676, "loss_box_5": 2.88142, "loss_cns_5": 0.53151, "loss_yns_5": 0.17417, "loss_cls_dn_0": 0.4758, "loss_box_dn_0": 1.06146, "loss_cls_dn_1": 0.45528, "loss_box_dn_1": 1.38861, "loss_cls_dn_2": 0.43525, "loss_box_dn_2": 1.46812, "loss_cls_dn_3": 0.43777, "loss_box_dn_3": 1.53643, "loss_cls_dn_4": 0.44382, "loss_box_dn_4": 1.53136, "loss_cls_dn_5": 0.45334, "loss_box_dn_5": 1.73488, "loss_dense_depth": 2.12965, "loss": 41.46165, "grad_norm": 83.27417, "time": 1.52093} -{"mode": "train", "epoch": 1, "iter": 17, "lr": 0.00011, "memory": 49163, "data_time": 0.07883, "loss_cls_0": 1.17446, "loss_box_0": 2.19656, "loss_cns_0": 0.60986, "loss_yns_0": 0.17242, "loss_cls_1": 1.22634, "loss_box_1": 2.67043, "loss_cns_1": 0.58188, "loss_yns_1": 0.17936, "loss_cls_2": 1.27315, "loss_box_2": 2.73302, "loss_cns_2": 0.55194, "loss_yns_2": 0.17646, "loss_cls_3": 1.3303, "loss_box_3": 2.76225, "loss_cns_3": 0.54715, "loss_yns_3": 0.17682, "loss_cls_4": 1.2834, "loss_box_4": 2.77745, "loss_cns_4": 0.54836, "loss_yns_4": 0.18011, "loss_cls_5": 1.27474, "loss_box_5": 2.87765, "loss_cns_5": 0.54332, "loss_yns_5": 0.18089, "loss_cls_dn_0": 0.45641, "loss_box_dn_0": 1.02321, "loss_cls_dn_1": 0.43473, "loss_box_dn_1": 1.44018, "loss_cls_dn_2": 0.4121, "loss_box_dn_2": 1.49479, "loss_cls_dn_3": 0.40281, "loss_box_dn_3": 1.58402, "loss_cls_dn_4": 0.41356, "loss_box_dn_4": 1.56776, "loss_cls_dn_5": 0.41961, "loss_box_dn_5": 1.70397, "loss_dense_depth": 1.9324, "loss": 41.31389, "grad_norm": 75.48756, "time": 1.52605} -{"mode": "train", "epoch": 1, "iter": 18, "lr": 0.00011, "memory": 49163, "data_time": 0.07814, "loss_cls_0": 1.16191, "loss_box_0": 2.14638, "loss_cns_0": 0.61527, "loss_yns_0": 0.17463, "loss_cls_1": 1.24071, "loss_box_1": 2.59748, "loss_cns_1": 0.57851, "loss_yns_1": 0.1782, "loss_cls_2": 1.27105, "loss_box_2": 2.66366, "loss_cns_2": 0.57252, "loss_yns_2": 0.17562, "loss_cls_3": 1.25479, "loss_box_3": 2.64466, "loss_cns_3": 0.58193, "loss_yns_3": 0.17964, "loss_cls_4": 1.28341, "loss_box_4": 2.69676, "loss_cns_4": 0.57619, "loss_yns_4": 0.17527, "loss_cls_5": 1.30016, "loss_box_5": 2.68933, "loss_cns_5": 0.59499, "loss_yns_5": 0.17725, "loss_cls_dn_0": 0.47321, "loss_box_dn_0": 1.02619, "loss_cls_dn_1": 0.41157, "loss_box_dn_1": 1.35162, "loss_cls_dn_2": 0.40925, "loss_box_dn_2": 1.36079, "loss_cls_dn_3": 0.42047, "loss_box_dn_3": 1.45226, "loss_cls_dn_4": 0.41353, "loss_box_dn_4": 1.47309, "loss_cls_dn_5": 0.40964, "loss_box_dn_5": 1.52897, "loss_dense_depth": 1.81897, "loss": 40.07987, "grad_norm": 59.83289, "time": 1.5233} -{"mode": "train", "epoch": 1, "iter": 19, "lr": 0.00011, "memory": 49163, "data_time": 0.08159, "loss_cls_0": 1.15796, "loss_box_0": 2.17803, "loss_cns_0": 0.60868, "loss_yns_0": 0.16972, "loss_cls_1": 1.25275, "loss_box_1": 2.61602, "loss_cns_1": 0.58873, "loss_yns_1": 0.177, "loss_cls_2": 1.26989, "loss_box_2": 2.61208, "loss_cns_2": 0.58374, "loss_yns_2": 0.17737, "loss_cls_3": 1.25426, "loss_box_3": 2.68177, "loss_cns_3": 0.56975, "loss_yns_3": 0.17489, "loss_cls_4": 1.24527, "loss_box_4": 2.99308, "loss_cns_4": 0.54111, "loss_yns_4": 0.1714, "loss_cls_5": 1.26897, "loss_box_5": 3.13408, "loss_cns_5": 0.48252, "loss_yns_5": 0.17523, "loss_cls_dn_0": 0.47846, "loss_box_dn_0": 1.05026, "loss_cls_dn_1": 0.42634, "loss_box_dn_1": 1.16797, "loss_cls_dn_2": 0.43093, "loss_box_dn_2": 1.14529, "loss_cls_dn_3": 0.46281, "loss_box_dn_3": 1.25102, "loss_cls_dn_4": 0.46264, "loss_box_dn_4": 1.45864, "loss_cls_dn_5": 0.43969, "loss_box_dn_5": 1.54919, "loss_dense_depth": 1.73222, "loss": 40.13976, "grad_norm": 83.74898, "time": 1.52059} -{"mode": "train", "epoch": 1, "iter": 20, "lr": 0.00011, "memory": 49163, "data_time": 0.0747, "loss_cls_0": 1.16144, "loss_box_0": 2.1881, "loss_cns_0": 0.60914, "loss_yns_0": 0.17032, "loss_cls_1": 1.22566, "loss_box_1": 2.8231, "loss_cns_1": 0.57255, "loss_yns_1": 0.18109, "loss_cls_2": 1.25507, "loss_box_2": 2.81935, "loss_cns_2": 0.55585, "loss_yns_2": 0.17473, "loss_cls_3": 1.26104, "loss_box_3": 2.92818, "loss_cns_3": 0.53483, "loss_yns_3": 0.17742, "loss_cls_4": 1.24261, "loss_box_4": 3.20703, "loss_cns_4": 0.48795, "loss_yns_4": 0.16921, "loss_cls_5": 1.2611, "loss_box_5": 3.43662, "loss_cns_5": 0.43904, "loss_yns_5": 0.17205, "loss_cls_dn_0": 0.46349, "loss_box_dn_0": 1.04604, "loss_cls_dn_1": 0.40706, "loss_box_dn_1": 1.29911, "loss_cls_dn_2": 0.41346, "loss_box_dn_2": 1.28509, "loss_cls_dn_3": 0.43926, "loss_box_dn_3": 1.38946, "loss_cls_dn_4": 0.43465, "loss_box_dn_4": 1.57327, "loss_cls_dn_5": 0.42332, "loss_box_dn_5": 1.65458, "loss_dense_depth": 1.58957, "loss": 41.47182, "grad_norm": 97.81349, "time": 1.52611} -{"mode": "train", "epoch": 1, "iter": 21, "lr": 0.00011, "memory": 49163, "data_time": 0.12994, "loss_cls_0": 1.17271, "loss_box_0": 2.1958, "loss_cns_0": 0.61716, "loss_yns_0": 0.17199, "loss_cls_1": 1.22727, "loss_box_1": 2.99259, "loss_cns_1": 0.57375, "loss_yns_1": 0.17393, "loss_cls_2": 1.2492, "loss_box_2": 2.97806, "loss_cns_2": 0.54594, "loss_yns_2": 0.1736, "loss_cls_3": 1.2535, "loss_box_3": 2.99924, "loss_cns_3": 0.54383, "loss_yns_3": 0.17381, "loss_cls_4": 1.24749, "loss_box_4": 2.98177, "loss_cns_4": 0.52744, "loss_yns_4": 0.17006, "loss_cls_5": 1.27054, "loss_box_5": 3.0781, "loss_cns_5": 0.50762, "loss_yns_5": 0.17313, "loss_cls_dn_0": 0.45687, "loss_box_dn_0": 1.05986, "loss_cls_dn_1": 0.43947, "loss_box_dn_1": 1.37187, "loss_cls_dn_2": 0.4512, "loss_box_dn_2": 1.42255, "loss_cls_dn_3": 0.45428, "loss_box_dn_3": 1.52796, "loss_cls_dn_4": 0.44991, "loss_box_dn_4": 1.69796, "loss_cls_dn_5": 0.45861, "loss_box_dn_5": 1.71217, "loss_dense_depth": 1.59163, "loss": 42.09285, "grad_norm": 82.52425, "time": 1.62922} -{"mode": "train", "epoch": 1, "iter": 22, "lr": 0.00011, "memory": 49163, "data_time": 0.11086, "loss_cls_0": 1.14537, "loss_box_0": 2.17644, "loss_cns_0": 0.61719, "loss_yns_0": 0.16999, "loss_cls_1": 1.25665, "loss_box_1": 2.80385, "loss_cns_1": 0.5672, "loss_yns_1": 0.17016, "loss_cls_2": 1.25062, "loss_box_2": 2.8049, "loss_cns_2": 0.55836, "loss_yns_2": 0.17004, "loss_cls_3": 1.26199, "loss_box_3": 2.79003, "loss_cns_3": 0.55396, "loss_yns_3": 0.17224, "loss_cls_4": 1.25626, "loss_box_4": 2.80716, "loss_cns_4": 0.55992, "loss_yns_4": 0.17114, "loss_cls_5": 1.28134, "loss_box_5": 2.73762, "loss_cns_5": 0.56586, "loss_yns_5": 0.18079, "loss_cls_dn_0": 0.46449, "loss_box_dn_0": 1.03288, "loss_cls_dn_1": 0.44365, "loss_box_dn_1": 1.37745, "loss_cls_dn_2": 0.45591, "loss_box_dn_2": 1.43553, "loss_cls_dn_3": 0.43593, "loss_box_dn_3": 1.46892, "loss_cls_dn_4": 0.44437, "loss_box_dn_4": 1.62227, "loss_cls_dn_5": 0.46509, "loss_box_dn_5": 1.54864, "loss_dense_depth": 1.47488, "loss": 40.69907, "grad_norm": 66.18288, "time": 1.55316} -{"mode": "train", "epoch": 1, "iter": 23, "lr": 0.00011, "memory": 49163, "data_time": 0.11932, "loss_cls_0": 1.12212, "loss_box_0": 2.10631, "loss_cns_0": 0.61662, "loss_yns_0": 0.16888, "loss_cls_1": 1.24154, "loss_box_1": 2.68701, "loss_cns_1": 0.56325, "loss_yns_1": 0.17274, "loss_cls_2": 1.24654, "loss_box_2": 2.76624, "loss_cns_2": 0.56635, "loss_yns_2": 0.17447, "loss_cls_3": 1.28029, "loss_box_3": 2.79979, "loss_cns_3": 0.55112, "loss_yns_3": 0.17717, "loss_cls_4": 1.26717, "loss_box_4": 2.90993, "loss_cns_4": 0.5473, "loss_yns_4": 0.17076, "loss_cls_5": 1.27025, "loss_box_5": 2.91015, "loss_cns_5": 0.5422, "loss_yns_5": 0.18133, "loss_cls_dn_0": 0.4666, "loss_box_dn_0": 0.98512, "loss_cls_dn_1": 0.41737, "loss_box_dn_1": 1.26507, "loss_cls_dn_2": 0.43658, "loss_box_dn_2": 1.37922, "loss_cls_dn_3": 0.41588, "loss_box_dn_3": 1.38389, "loss_cls_dn_4": 0.41914, "loss_box_dn_4": 1.52114, "loss_cls_dn_5": 0.45521, "loss_box_dn_5": 1.48924, "loss_dense_depth": 1.41996, "loss": 40.09393, "grad_norm": 81.88515, "time": 1.58073} -{"mode": "train", "epoch": 1, "iter": 24, "lr": 0.00011, "memory": 49163, "data_time": 0.07617, "loss_cls_0": 1.15583, "loss_box_0": 2.13408, "loss_cns_0": 0.61713, "loss_yns_0": 0.17005, "loss_cls_1": 1.2732, "loss_box_1": 2.78639, "loss_cns_1": 0.56199, "loss_yns_1": 0.17346, "loss_cls_2": 1.28319, "loss_box_2": 2.92291, "loss_cns_2": 0.57789, "loss_yns_2": 0.17067, "loss_cls_3": 1.29416, "loss_box_3": 2.88789, "loss_cns_3": 0.5644, "loss_yns_3": 0.17624, "loss_cls_4": 1.31534, "loss_box_4": 2.87317, "loss_cns_4": 0.56991, "loss_yns_4": 0.17815, "loss_cls_5": 1.27617, "loss_box_5": 2.98072, "loss_cns_5": 0.55989, "loss_yns_5": 0.17871, "loss_cls_dn_0": 0.46697, "loss_box_dn_0": 0.99647, "loss_cls_dn_1": 0.38447, "loss_box_dn_1": 1.34425, "loss_cls_dn_2": 0.40837, "loss_box_dn_2": 1.44947, "loss_cls_dn_3": 0.4018, "loss_box_dn_3": 1.39516, "loss_cls_dn_4": 0.39199, "loss_box_dn_4": 1.43279, "loss_cls_dn_5": 0.42771, "loss_box_dn_5": 1.4644, "loss_dense_depth": 1.44673, "loss": 40.69212, "grad_norm": 80.98878, "time": 1.54926} -{"mode": "train", "epoch": 1, "iter": 25, "lr": 0.00011, "memory": 49163, "data_time": 0.07512, "loss_cls_0": 1.17487, "loss_box_0": 2.15741, "loss_cns_0": 0.61261, "loss_yns_0": 0.16987, "loss_cls_1": 1.26899, "loss_box_1": 2.86583, "loss_cns_1": 0.5762, "loss_yns_1": 0.17376, "loss_cls_2": 1.29956, "loss_box_2": 2.96882, "loss_cns_2": 0.60191, "loss_yns_2": 0.17084, "loss_cls_3": 1.26371, "loss_box_3": 2.89432, "loss_cns_3": 0.59375, "loss_yns_3": 0.17707, "loss_cls_4": 1.31041, "loss_box_4": 2.89501, "loss_cns_4": 0.5905, "loss_yns_4": 0.18059, "loss_cls_5": 1.28279, "loss_box_5": 2.96592, "loss_cns_5": 0.59142, "loss_yns_5": 0.18569, "loss_cls_dn_0": 0.45391, "loss_box_dn_0": 0.99575, "loss_cls_dn_1": 0.3769, "loss_box_dn_1": 1.27819, "loss_cls_dn_2": 0.39532, "loss_box_dn_2": 1.35355, "loss_cls_dn_3": 0.40014, "loss_box_dn_3": 1.31388, "loss_cls_dn_4": 0.38294, "loss_box_dn_4": 1.32533, "loss_cls_dn_5": 0.40498, "loss_box_dn_5": 1.42834, "loss_dense_depth": 1.42823, "loss": 40.50929, "grad_norm": 69.03638, "time": 1.51672} -{"mode": "train", "epoch": 1, "iter": 26, "lr": 0.00011, "memory": 49163, "data_time": 0.07851, "loss_cls_0": 1.15159, "loss_box_0": 2.19909, "loss_cns_0": 0.60719, "loss_yns_0": 0.17076, "loss_cls_1": 1.21096, "loss_box_1": 2.76769, "loss_cns_1": 0.57185, "loss_yns_1": 0.1707, "loss_cls_2": 1.2466, "loss_box_2": 2.86324, "loss_cns_2": 0.57905, "loss_yns_2": 0.17666, "loss_cls_3": 1.24093, "loss_box_3": 2.85664, "loss_cns_3": 0.56673, "loss_yns_3": 0.18247, "loss_cls_4": 1.25114, "loss_box_4": 2.92123, "loss_cns_4": 0.56745, "loss_yns_4": 0.18112, "loss_cls_5": 1.32263, "loss_box_5": 2.93175, "loss_cns_5": 0.5756, "loss_yns_5": 0.20051, "loss_cls_dn_0": 0.44, "loss_box_dn_0": 1.02203, "loss_cls_dn_1": 0.41592, "loss_box_dn_1": 1.21909, "loss_cls_dn_2": 0.43362, "loss_box_dn_2": 1.30857, "loss_cls_dn_3": 0.43793, "loss_box_dn_3": 1.38891, "loss_cls_dn_4": 0.42892, "loss_box_dn_4": 1.44886, "loss_cls_dn_5": 0.42408, "loss_box_dn_5": 1.61959, "loss_dense_depth": 1.30857, "loss": 40.40967, "grad_norm": 87.31863, "time": 1.55355} -{"mode": "train", "epoch": 1, "iter": 27, "lr": 0.00011, "memory": 49163, "data_time": 0.07788, "loss_cls_0": 1.14818, "loss_box_0": 2.25235, "loss_cns_0": 0.60144, "loss_yns_0": 0.17242, "loss_cls_1": 1.18237, "loss_box_1": 2.79529, "loss_cns_1": 0.55596, "loss_yns_1": 0.17287, "loss_cls_2": 1.21315, "loss_box_2": 2.81648, "loss_cns_2": 0.56269, "loss_yns_2": 0.18195, "loss_cls_3": 1.22924, "loss_box_3": 2.80379, "loss_cns_3": 0.5638, "loss_yns_3": 0.18777, "loss_cls_4": 1.2153, "loss_box_4": 2.9239, "loss_cns_4": 0.57068, "loss_yns_4": 0.19539, "loss_cls_5": 1.29807, "loss_box_5": 2.92976, "loss_cns_5": 0.57055, "loss_yns_5": 0.19784, "loss_cls_dn_0": 0.43874, "loss_box_dn_0": 1.03906, "loss_cls_dn_1": 0.43797, "loss_box_dn_1": 1.34917, "loss_cls_dn_2": 0.44978, "loss_box_dn_2": 1.3832, "loss_cls_dn_3": 0.4501, "loss_box_dn_3": 1.5355, "loss_cls_dn_4": 0.45948, "loss_box_dn_4": 1.63226, "loss_cls_dn_5": 0.4298, "loss_box_dn_5": 1.80344, "loss_dense_depth": 1.31919, "loss": 41.06895, "grad_norm": 87.85618, "time": 1.51479} -{"mode": "train", "epoch": 1, "iter": 28, "lr": 0.00011, "memory": 49163, "data_time": 0.07794, "loss_cls_0": 1.13028, "loss_box_0": 2.2604, "loss_cns_0": 0.60068, "loss_yns_0": 0.17153, "loss_cls_1": 1.17192, "loss_box_1": 2.71929, "loss_cns_1": 0.56566, "loss_yns_1": 0.17604, "loss_cls_2": 1.20923, "loss_box_2": 2.68303, "loss_cns_2": 0.5703, "loss_yns_2": 0.17414, "loss_cls_3": 1.22002, "loss_box_3": 2.74439, "loss_cns_3": 0.57667, "loss_yns_3": 0.18406, "loss_cls_4": 1.20446, "loss_box_4": 2.90921, "loss_cns_4": 0.57104, "loss_yns_4": 0.17778, "loss_cls_5": 1.23761, "loss_box_5": 2.90448, "loss_cns_5": 0.56812, "loss_yns_5": 0.17541, "loss_cls_dn_0": 0.43769, "loss_box_dn_0": 1.05018, "loss_cls_dn_1": 0.4476, "loss_box_dn_1": 1.50483, "loss_cls_dn_2": 0.45196, "loss_box_dn_2": 1.50508, "loss_cls_dn_3": 0.44759, "loss_box_dn_3": 1.6837, "loss_cls_dn_4": 0.46379, "loss_box_dn_4": 1.77296, "loss_cls_dn_5": 0.43223, "loss_box_dn_5": 1.92425, "loss_dense_depth": 1.28123, "loss": 41.30884, "grad_norm": 84.16286, "time": 1.53898} -{"mode": "train", "epoch": 1, "iter": 29, "lr": 0.00011, "memory": 49163, "data_time": 0.07965, "loss_cls_0": 1.10368, "loss_box_0": 2.26006, "loss_cns_0": 0.60144, "loss_yns_0": 0.1715, "loss_cls_1": 1.15466, "loss_box_1": 2.71613, "loss_cns_1": 0.58172, "loss_yns_1": 0.17517, "loss_cls_2": 1.19383, "loss_box_2": 2.68802, "loss_cns_2": 0.58242, "loss_yns_2": 0.17702, "loss_cls_3": 1.19751, "loss_box_3": 2.69986, "loss_cns_3": 0.57993, "loss_yns_3": 0.17799, "loss_cls_4": 1.18951, "loss_box_4": 2.83072, "loss_cns_4": 0.56872, "loss_yns_4": 0.17889, "loss_cls_5": 1.21014, "loss_box_5": 2.85852, "loss_cns_5": 0.56826, "loss_yns_5": 0.1738, "loss_cls_dn_0": 0.44399, "loss_box_dn_0": 1.04636, "loss_cls_dn_1": 0.44653, "loss_box_dn_1": 1.42178, "loss_cls_dn_2": 0.45048, "loss_box_dn_2": 1.39376, "loss_cls_dn_3": 0.4459, "loss_box_dn_3": 1.50911, "loss_cls_dn_4": 0.46021, "loss_box_dn_4": 1.55852, "loss_cls_dn_5": 0.45104, "loss_box_dn_5": 1.69125, "loss_dense_depth": 1.26166, "loss": 40.22013, "grad_norm": 93.71168, "time": 1.54012} -{"mode": "train", "epoch": 1, "iter": 30, "lr": 0.00011, "memory": 49163, "data_time": 0.07919, "loss_cls_0": 1.08811, "loss_box_0": 2.18517, "loss_cns_0": 0.60261, "loss_yns_0": 0.17385, "loss_cls_1": 1.11446, "loss_box_1": 2.66423, "loss_cns_1": 0.57751, "loss_yns_1": 0.17502, "loss_cls_2": 1.16463, "loss_box_2": 2.66437, "loss_cns_2": 0.5851, "loss_yns_2": 0.17531, "loss_cls_3": 1.17014, "loss_box_3": 2.58127, "loss_cns_3": 0.58458, "loss_yns_3": 0.1713, "loss_cls_4": 1.1733, "loss_box_4": 2.60398, "loss_cns_4": 0.58078, "loss_yns_4": 0.17853, "loss_cls_5": 1.19241, "loss_box_5": 2.6559, "loss_cns_5": 0.58181, "loss_yns_5": 0.18541, "loss_cls_dn_0": 0.44449, "loss_box_dn_0": 1.02338, "loss_cls_dn_1": 0.42244, "loss_box_dn_1": 1.52341, "loss_cls_dn_2": 0.42418, "loss_box_dn_2": 1.48236, "loss_cls_dn_3": 0.4203, "loss_box_dn_3": 1.48484, "loss_cls_dn_4": 0.43085, "loss_box_dn_4": 1.47399, "loss_cls_dn_5": 0.44403, "loss_box_dn_5": 1.57054, "loss_dense_depth": 1.18788, "loss": 39.16249, "grad_norm": 71.9034, "time": 1.52764} -{"mode": "train", "epoch": 1, "iter": 31, "lr": 0.00011, "memory": 49163, "data_time": 0.07762, "loss_cls_0": 1.07995, "loss_box_0": 2.11699, "loss_cns_0": 0.60459, "loss_yns_0": 0.174, "loss_cls_1": 1.10831, "loss_box_1": 2.57826, "loss_cns_1": 0.56564, "loss_yns_1": 0.17448, "loss_cls_2": 1.15161, "loss_box_2": 2.59444, "loss_cns_2": 0.57356, "loss_yns_2": 0.17225, "loss_cls_3": 1.16165, "loss_box_3": 2.50991, "loss_cns_3": 0.5747, "loss_yns_3": 0.17552, "loss_cls_4": 1.17901, "loss_box_4": 2.52668, "loss_cns_4": 0.57942, "loss_yns_4": 0.17227, "loss_cls_5": 1.18014, "loss_box_5": 2.53409, "loss_cns_5": 0.58168, "loss_yns_5": 0.18439, "loss_cls_dn_0": 0.44299, "loss_box_dn_0": 0.98631, "loss_cls_dn_1": 0.39536, "loss_box_dn_1": 1.42911, "loss_cls_dn_2": 0.39641, "loss_box_dn_2": 1.38716, "loss_cls_dn_3": 0.39819, "loss_box_dn_3": 1.32663, "loss_cls_dn_4": 0.4057, "loss_box_dn_4": 1.3061, "loss_cls_dn_5": 0.43652, "loss_box_dn_5": 1.33857, "loss_dense_depth": 1.16979, "loss": 37.67237, "grad_norm": 41.93866, "time": 1.58564} -{"mode": "train", "epoch": 1, "iter": 32, "lr": 0.00011, "memory": 49163, "data_time": 0.07964, "loss_cls_0": 1.07548, "loss_box_0": 2.08649, "loss_cns_0": 0.61217, "loss_yns_0": 0.17122, "loss_cls_1": 1.12623, "loss_box_1": 2.6263, "loss_cns_1": 0.55018, "loss_yns_1": 0.17214, "loss_cls_2": 1.16575, "loss_box_2": 2.63345, "loss_cns_2": 0.56261, "loss_yns_2": 0.17168, "loss_cls_3": 1.17548, "loss_box_3": 2.66128, "loss_cns_3": 0.55934, "loss_yns_3": 0.1847, "loss_cls_4": 1.16056, "loss_box_4": 2.7021, "loss_cns_4": 0.55719, "loss_yns_4": 0.17031, "loss_cls_5": 1.15291, "loss_box_5": 2.68416, "loss_cns_5": 0.5657, "loss_yns_5": 0.17238, "loss_cls_dn_0": 0.42307, "loss_box_dn_0": 0.96874, "loss_cls_dn_1": 0.36832, "loss_box_dn_1": 1.28446, "loss_cls_dn_2": 0.3681, "loss_box_dn_2": 1.28836, "loss_cls_dn_3": 0.37497, "loss_box_dn_3": 1.31755, "loss_cls_dn_4": 0.38399, "loss_box_dn_4": 1.37958, "loss_cls_dn_5": 0.41974, "loss_box_dn_5": 1.38262, "loss_dense_depth": 1.23881, "loss": 37.89813, "grad_norm": 56.2962, "time": 1.52412} -{"mode": "train", "epoch": 1, "iter": 33, "lr": 0.00011, "memory": 49163, "data_time": 0.08044, "loss_cls_0": 1.09449, "loss_box_0": 2.09536, "loss_cns_0": 0.6172, "loss_yns_0": 0.1726, "loss_cls_1": 1.12109, "loss_box_1": 2.69514, "loss_cns_1": 0.54039, "loss_yns_1": 0.17405, "loss_cls_2": 1.18393, "loss_box_2": 2.65556, "loss_cns_2": 0.55937, "loss_yns_2": 0.17366, "loss_cls_3": 1.17975, "loss_box_3": 2.69612, "loss_cns_3": 0.55833, "loss_yns_3": 0.17749, "loss_cls_4": 1.15812, "loss_box_4": 2.71847, "loss_cns_4": 0.55891, "loss_yns_4": 0.17462, "loss_cls_5": 1.15307, "loss_box_5": 2.67923, "loss_cns_5": 0.56688, "loss_yns_5": 0.17291, "loss_cls_dn_0": 0.42037, "loss_box_dn_0": 0.97346, "loss_cls_dn_1": 0.34619, "loss_box_dn_1": 1.51557, "loss_cls_dn_2": 0.35329, "loss_box_dn_2": 1.5222, "loss_cls_dn_3": 0.35946, "loss_box_dn_3": 1.58459, "loss_cls_dn_4": 0.36835, "loss_box_dn_4": 1.66429, "loss_cls_dn_5": 0.3822, "loss_box_dn_5": 1.66683, "loss_dense_depth": 1.23405, "loss": 39.2676, "grad_norm": 65.3435, "time": 1.51824} -{"mode": "train", "epoch": 1, "iter": 34, "lr": 0.00011, "memory": 49163, "data_time": 0.07931, "loss_cls_0": 1.08413, "loss_box_0": 2.08517, "loss_cns_0": 0.62015, "loss_yns_0": 0.17524, "loss_cls_1": 1.10344, "loss_box_1": 2.60917, "loss_cns_1": 0.54985, "loss_yns_1": 0.17977, "loss_cls_2": 1.14489, "loss_box_2": 2.57186, "loss_cns_2": 0.56875, "loss_yns_2": 0.17507, "loss_cls_3": 1.15787, "loss_box_3": 2.57263, "loss_cns_3": 0.5692, "loss_yns_3": 0.17019, "loss_cls_4": 1.16358, "loss_box_4": 2.57532, "loss_cns_4": 0.5688, "loss_yns_4": 0.17463, "loss_cls_5": 1.16674, "loss_box_5": 2.5545, "loss_cns_5": 0.58183, "loss_yns_5": 0.17912, "loss_cls_dn_0": 0.43279, "loss_box_dn_0": 0.97288, "loss_cls_dn_1": 0.34585, "loss_box_dn_1": 1.6622, "loss_cls_dn_2": 0.35711, "loss_box_dn_2": 1.66801, "loss_cls_dn_3": 0.36439, "loss_box_dn_3": 1.71439, "loss_cls_dn_4": 0.36344, "loss_box_dn_4": 1.78836, "loss_cls_dn_5": 0.36293, "loss_box_dn_5": 1.77973, "loss_dense_depth": 1.24479, "loss": 39.35878, "grad_norm": 61.24897, "time": 1.51977} -{"mode": "train", "epoch": 1, "iter": 35, "lr": 0.00011, "memory": 49163, "data_time": 0.07717, "loss_cls_0": 1.06718, "loss_box_0": 2.057, "loss_cns_0": 0.62041, "loss_yns_0": 0.17532, "loss_cls_1": 1.12709, "loss_box_1": 2.53313, "loss_cns_1": 0.56332, "loss_yns_1": 0.18106, "loss_cls_2": 1.13531, "loss_box_2": 2.52309, "loss_cns_2": 0.5795, "loss_yns_2": 0.17206, "loss_cls_3": 1.14945, "loss_box_3": 2.49297, "loss_cns_3": 0.58526, "loss_yns_3": 0.1745, "loss_cls_4": 1.16811, "loss_box_4": 2.51906, "loss_cns_4": 0.58386, "loss_yns_4": 0.17188, "loss_cls_5": 1.19726, "loss_box_5": 2.54706, "loss_cns_5": 0.5913, "loss_yns_5": 0.18113, "loss_cls_dn_0": 0.43417, "loss_box_dn_0": 0.962, "loss_cls_dn_1": 0.3481, "loss_box_dn_1": 1.60416, "loss_cls_dn_2": 0.36293, "loss_box_dn_2": 1.62068, "loss_cls_dn_3": 0.37257, "loss_box_dn_3": 1.63448, "loss_cls_dn_4": 0.36354, "loss_box_dn_4": 1.70354, "loss_cls_dn_5": 0.35556, "loss_box_dn_5": 1.70521, "loss_dense_depth": 1.25293, "loss": 38.8162, "grad_norm": 62.49941, "time": 1.51718} -{"mode": "train", "epoch": 1, "iter": 36, "lr": 0.00011, "memory": 49163, "data_time": 0.07378, "loss_cls_0": 1.05399, "loss_box_0": 2.09534, "loss_cns_0": 0.61654, "loss_yns_0": 0.17227, "loss_cls_1": 1.09789, "loss_box_1": 2.5893, "loss_cns_1": 0.56317, "loss_yns_1": 0.17945, "loss_cls_2": 1.12042, "loss_box_2": 2.55721, "loss_cns_2": 0.57818, "loss_yns_2": 0.17153, "loss_cls_3": 1.14109, "loss_box_3": 2.55239, "loss_cns_3": 0.58783, "loss_yns_3": 0.18564, "loss_cls_4": 1.15309, "loss_box_4": 2.59463, "loss_cns_4": 0.58022, "loss_yns_4": 0.17087, "loss_cls_5": 1.18738, "loss_box_5": 2.6282, "loss_cns_5": 0.58397, "loss_yns_5": 0.1766, "loss_cls_dn_0": 0.42408, "loss_box_dn_0": 0.96165, "loss_cls_dn_1": 0.37096, "loss_box_dn_1": 1.14425, "loss_cls_dn_2": 0.38443, "loss_box_dn_2": 1.14575, "loss_cls_dn_3": 0.39207, "loss_box_dn_3": 1.14367, "loss_cls_dn_4": 0.38297, "loss_box_dn_4": 1.19715, "loss_cls_dn_5": 0.3801, "loss_box_dn_5": 1.19603, "loss_dense_depth": 1.22942, "loss": 36.68972, "grad_norm": 50.55619, "time": 1.50742} -{"mode": "train", "epoch": 1, "iter": 37, "lr": 0.00011, "memory": 49163, "data_time": 0.0773, "loss_cls_0": 1.0689, "loss_box_0": 2.09394, "loss_cns_0": 0.6195, "loss_yns_0": 0.17377, "loss_cls_1": 1.12192, "loss_box_1": 2.72612, "loss_cns_1": 0.55979, "loss_yns_1": 0.17182, "loss_cls_2": 1.12481, "loss_box_2": 2.67055, "loss_cns_2": 0.57305, "loss_yns_2": 0.17565, "loss_cls_3": 1.15661, "loss_box_3": 2.66997, "loss_cns_3": 0.57908, "loss_yns_3": 0.18178, "loss_cls_4": 1.1466, "loss_box_4": 2.68202, "loss_cns_4": 0.5794, "loss_yns_4": 0.17207, "loss_cls_5": 1.17113, "loss_box_5": 2.65929, "loss_cns_5": 0.58508, "loss_yns_5": 0.1757, "loss_cls_dn_0": 0.41253, "loss_box_dn_0": 0.97475, "loss_cls_dn_1": 0.3513, "loss_box_dn_1": 1.18075, "loss_cls_dn_2": 0.37377, "loss_box_dn_2": 1.12968, "loss_cls_dn_3": 0.36998, "loss_box_dn_3": 1.10374, "loss_cls_dn_4": 0.37451, "loss_box_dn_4": 1.12036, "loss_cls_dn_5": 0.37169, "loss_box_dn_5": 1.10031, "loss_dense_depth": 1.21669, "loss": 36.9186, "grad_norm": 52.3418, "time": 1.5234} -{"mode": "train", "epoch": 1, "iter": 38, "lr": 0.00011, "memory": 49163, "data_time": 0.07545, "loss_cls_0": 1.07062, "loss_box_0": 2.06066, "loss_cns_0": 0.62163, "loss_yns_0": 0.17355, "loss_cls_1": 1.1135, "loss_box_1": 2.75989, "loss_cns_1": 0.55993, "loss_yns_1": 0.17155, "loss_cls_2": 1.11291, "loss_box_2": 2.68089, "loss_cns_2": 0.57701, "loss_yns_2": 0.17552, "loss_cls_3": 1.14921, "loss_box_3": 2.68194, "loss_cns_3": 0.5799, "loss_yns_3": 0.17052, "loss_cls_4": 1.13259, "loss_box_4": 2.72401, "loss_cns_4": 0.58172, "loss_yns_4": 0.17243, "loss_cls_5": 1.1557, "loss_box_5": 2.75329, "loss_cns_5": 0.58207, "loss_yns_5": 0.17458, "loss_cls_dn_0": 0.40126, "loss_box_dn_0": 0.96975, "loss_cls_dn_1": 0.34485, "loss_box_dn_1": 1.10701, "loss_cls_dn_2": 0.37378, "loss_box_dn_2": 1.04668, "loss_cls_dn_3": 0.37078, "loss_box_dn_3": 1.0587, "loss_cls_dn_4": 0.37755, "loss_box_dn_4": 1.06575, "loss_cls_dn_5": 0.38243, "loss_box_dn_5": 1.10607, "loss_dense_depth": 1.20198, "loss": 36.7422, "grad_norm": 66.6528, "time": 1.52349} -{"mode": "train", "epoch": 1, "iter": 39, "lr": 0.00012, "memory": 49163, "data_time": 0.07475, "loss_cls_0": 1.07745, "loss_box_0": 1.98699, "loss_cns_0": 0.62334, "loss_yns_0": 0.17325, "loss_cls_1": 1.1278, "loss_box_1": 2.50658, "loss_cns_1": 0.58078, "loss_yns_1": 0.17791, "loss_cls_2": 1.11669, "loss_box_2": 2.42249, "loss_cns_2": 0.59602, "loss_yns_2": 0.17144, "loss_cls_3": 1.1441, "loss_box_3": 2.4314, "loss_cns_3": 0.59387, "loss_yns_3": 0.16968, "loss_cls_4": 1.13847, "loss_box_4": 2.46442, "loss_cns_4": 0.59393, "loss_yns_4": 0.16995, "loss_cls_5": 1.16155, "loss_box_5": 2.55869, "loss_cns_5": 0.58774, "loss_yns_5": 0.17151, "loss_cls_dn_0": 0.40879, "loss_box_dn_0": 0.94465, "loss_cls_dn_1": 0.346, "loss_box_dn_1": 1.14126, "loss_cls_dn_2": 0.37983, "loss_box_dn_2": 1.10776, "loss_cls_dn_3": 0.3814, "loss_box_dn_3": 1.17672, "loss_cls_dn_4": 0.38192, "loss_box_dn_4": 1.19786, "loss_cls_dn_5": 0.39594, "loss_box_dn_5": 1.32515, "loss_dense_depth": 1.19332, "loss": 36.12668, "grad_norm": 59.90871, "time": 1.51488} -{"mode": "train", "epoch": 1, "iter": 40, "lr": 0.00012, "memory": 49163, "data_time": 0.07359, "loss_cls_0": 1.06338, "loss_box_0": 1.94804, "loss_cns_0": 0.62331, "loss_yns_0": 0.16788, "loss_cls_1": 1.10833, "loss_box_1": 2.39793, "loss_cns_1": 0.59169, "loss_yns_1": 0.18269, "loss_cls_2": 1.11606, "loss_box_2": 2.32289, "loss_cns_2": 0.60428, "loss_yns_2": 0.16853, "loss_cls_3": 1.14558, "loss_box_3": 2.32971, "loss_cns_3": 0.60199, "loss_yns_3": 0.1759, "loss_cls_4": 1.13215, "loss_box_4": 2.35333, "loss_cns_4": 0.60332, "loss_yns_4": 0.16821, "loss_cls_5": 1.15424, "loss_box_5": 2.4307, "loss_cns_5": 0.59868, "loss_yns_5": 0.17625, "loss_cls_dn_0": 0.41916, "loss_box_dn_0": 0.9357, "loss_cls_dn_1": 0.35564, "loss_box_dn_1": 1.28279, "loss_cls_dn_2": 0.38504, "loss_box_dn_2": 1.24792, "loss_cls_dn_3": 0.38571, "loss_box_dn_3": 1.3375, "loss_cls_dn_4": 0.38374, "loss_box_dn_4": 1.36305, "loss_cls_dn_5": 0.40562, "loss_box_dn_5": 1.49094, "loss_dense_depth": 1.21736, "loss": 36.37524, "grad_norm": 53.9123, "time": 1.51482} -{"mode": "train", "epoch": 1, "iter": 41, "lr": 0.00012, "memory": 49163, "data_time": 0.13376, "loss_cls_0": 1.02081, "loss_box_0": 1.94444, "loss_cns_0": 0.62362, "loss_yns_0": 0.16352, "loss_cls_1": 1.08144, "loss_box_1": 2.31831, "loss_cns_1": 0.60066, "loss_yns_1": 0.17831, "loss_cls_2": 1.09798, "loss_box_2": 2.27319, "loss_cns_2": 0.61314, "loss_yns_2": 0.16698, "loss_cls_3": 1.12156, "loss_box_3": 2.32549, "loss_cns_3": 0.61348, "loss_yns_3": 0.17724, "loss_cls_4": 1.12359, "loss_box_4": 2.36739, "loss_cns_4": 0.61224, "loss_yns_4": 0.17149, "loss_cls_5": 1.13125, "loss_box_5": 2.37021, "loss_cns_5": 0.61502, "loss_yns_5": 0.18606, "loss_cls_dn_0": 0.41958, "loss_box_dn_0": 0.93116, "loss_cls_dn_1": 0.34304, "loss_box_dn_1": 1.3814, "loss_cls_dn_2": 0.35629, "loss_box_dn_2": 1.32626, "loss_cls_dn_3": 0.35523, "loss_box_dn_3": 1.41071, "loss_cls_dn_4": 0.36323, "loss_box_dn_4": 1.43286, "loss_cls_dn_5": 0.38709, "loss_box_dn_5": 1.50434, "loss_dense_depth": 1.19879, "loss": 36.30739, "grad_norm": 63.6657, "time": 1.60444} -{"mode": "train", "epoch": 1, "iter": 42, "lr": 0.00012, "memory": 49163, "data_time": 0.10816, "loss_cls_0": 1.02837, "loss_box_0": 1.95965, "loss_cns_0": 0.62167, "loss_yns_0": 0.16614, "loss_cls_1": 1.08625, "loss_box_1": 2.30988, "loss_cns_1": 0.60369, "loss_yns_1": 0.1695, "loss_cls_2": 1.12793, "loss_box_2": 2.27566, "loss_cns_2": 0.61378, "loss_yns_2": 0.17002, "loss_cls_3": 1.14981, "loss_box_3": 2.34434, "loss_cns_3": 0.61609, "loss_yns_3": 0.17818, "loss_cls_4": 1.12714, "loss_box_4": 2.35198, "loss_cns_4": 0.61739, "loss_yns_4": 0.1744, "loss_cls_5": 1.12017, "loss_box_5": 2.3048, "loss_cns_5": 0.62209, "loss_yns_5": 0.18357, "loss_cls_dn_0": 0.40893, "loss_box_dn_0": 0.92893, "loss_cls_dn_1": 0.33749, "loss_box_dn_1": 1.28899, "loss_cls_dn_2": 0.34435, "loss_box_dn_2": 1.2248, "loss_cls_dn_3": 0.35124, "loss_box_dn_3": 1.29573, "loss_cls_dn_4": 0.3594, "loss_box_dn_4": 1.28908, "loss_cls_dn_5": 0.39264, "loss_box_dn_5": 1.31419, "loss_dense_depth": 1.18354, "loss": 35.64182, "grad_norm": 66.66664, "time": 1.54893} -{"mode": "train", "epoch": 1, "iter": 43, "lr": 0.00012, "memory": 49163, "data_time": 0.12076, "loss_cls_0": 1.07514, "loss_box_0": 1.99952, "loss_cns_0": 0.61582, "loss_yns_0": 0.169, "loss_cls_1": 1.13814, "loss_box_1": 2.30837, "loss_cns_1": 0.60194, "loss_yns_1": 0.17082, "loss_cls_2": 1.16643, "loss_box_2": 2.24078, "loss_cns_2": 0.61231, "loss_yns_2": 0.16671, "loss_cls_3": 1.1733, "loss_box_3": 2.27259, "loss_cns_3": 0.61906, "loss_yns_3": 0.16754, "loss_cls_4": 1.15858, "loss_box_4": 2.26472, "loss_cns_4": 0.61933, "loss_yns_4": 0.17162, "loss_cls_5": 1.1535, "loss_box_5": 2.28106, "loss_cns_5": 0.61879, "loss_yns_5": 0.1747, "loss_cls_dn_0": 0.41312, "loss_box_dn_0": 0.9244, "loss_cls_dn_1": 0.34455, "loss_box_dn_1": 1.13807, "loss_cls_dn_2": 0.35721, "loss_box_dn_2": 1.07378, "loss_cls_dn_3": 0.37596, "loss_box_dn_3": 1.10179, "loss_cls_dn_4": 0.3689, "loss_box_dn_4": 1.08047, "loss_cls_dn_5": 0.40593, "loss_box_dn_5": 1.10869, "loss_dense_depth": 1.12915, "loss": 34.76181, "grad_norm": 42.22898, "time": 1.54789} -{"mode": "train", "epoch": 1, "iter": 44, "lr": 0.00012, "memory": 49163, "data_time": 0.07644, "loss_cls_0": 1.05167, "loss_box_0": 2.03968, "loss_cns_0": 0.61112, "loss_yns_0": 0.16359, "loss_cls_1": 1.11489, "loss_box_1": 2.3427, "loss_cns_1": 0.60391, "loss_yns_1": 0.17258, "loss_cls_2": 1.11337, "loss_box_2": 2.27025, "loss_cns_2": 0.6169, "loss_yns_2": 0.16797, "loss_cls_3": 1.13487, "loss_box_3": 2.25279, "loss_cns_3": 0.6215, "loss_yns_3": 0.16357, "loss_cls_4": 1.13337, "loss_box_4": 2.3285, "loss_cns_4": 0.61961, "loss_yns_4": 0.17165, "loss_cls_5": 1.16345, "loss_box_5": 2.3653, "loss_cns_5": 0.61863, "loss_yns_5": 0.17421, "loss_cls_dn_0": 0.42651, "loss_box_dn_0": 0.91769, "loss_cls_dn_1": 0.34559, "loss_box_dn_1": 1.13108, "loss_cls_dn_2": 0.37876, "loss_box_dn_2": 1.05867, "loss_cls_dn_3": 0.40965, "loss_box_dn_3": 1.04804, "loss_cls_dn_4": 0.39025, "loss_box_dn_4": 1.07627, "loss_cls_dn_5": 0.4079, "loss_box_dn_5": 1.09753, "loss_dense_depth": 1.21417, "loss": 34.9182, "grad_norm": 55.80508, "time": 1.52298} -{"mode": "train", "epoch": 1, "iter": 45, "lr": 0.00012, "memory": 49163, "data_time": 0.07669, "loss_cls_0": 1.06779, "loss_box_0": 2.00303, "loss_cns_0": 0.61448, "loss_yns_0": 0.16284, "loss_cls_1": 1.10961, "loss_box_1": 2.37536, "loss_cns_1": 0.59336, "loss_yns_1": 0.17605, "loss_cls_2": 1.12503, "loss_box_2": 2.30083, "loss_cns_2": 0.6077, "loss_yns_2": 0.1695, "loss_cls_3": 1.14746, "loss_box_3": 2.27536, "loss_cns_3": 0.61646, "loss_yns_3": 0.16769, "loss_cls_4": 1.13246, "loss_box_4": 2.34833, "loss_cns_4": 0.61522, "loss_yns_4": 0.16918, "loss_cls_5": 1.16885, "loss_box_5": 2.33289, "loss_cns_5": 0.61794, "loss_yns_5": 0.18547, "loss_cls_dn_0": 0.42794, "loss_box_dn_0": 0.9133, "loss_cls_dn_1": 0.34663, "loss_box_dn_1": 1.05663, "loss_cls_dn_2": 0.38138, "loss_box_dn_2": 0.99444, "loss_cls_dn_3": 0.41522, "loss_box_dn_3": 0.98455, "loss_cls_dn_4": 0.3893, "loss_box_dn_4": 1.04671, "loss_cls_dn_5": 0.39776, "loss_box_dn_5": 1.04981, "loss_dense_depth": 1.14946, "loss": 34.63602, "grad_norm": 61.41852, "time": 1.53037} -{"mode": "train", "epoch": 1, "iter": 46, "lr": 0.00012, "memory": 49163, "data_time": 0.0761, "loss_cls_0": 1.05833, "loss_box_0": 1.9779, "loss_cns_0": 0.61551, "loss_yns_0": 0.16497, "loss_cls_1": 1.09944, "loss_box_1": 2.431, "loss_cns_1": 0.58974, "loss_yns_1": 0.17136, "loss_cls_2": 1.12192, "loss_box_2": 2.31664, "loss_cns_2": 0.60715, "loss_yns_2": 0.16754, "loss_cls_3": 1.13909, "loss_box_3": 2.29045, "loss_cns_3": 0.61608, "loss_yns_3": 0.16691, "loss_cls_4": 1.12886, "loss_box_4": 2.33062, "loss_cns_4": 0.6174, "loss_yns_4": 0.16446, "loss_cls_5": 1.15727, "loss_box_5": 2.34514, "loss_cns_5": 0.62176, "loss_yns_5": 0.18838, "loss_cls_dn_0": 0.42185, "loss_box_dn_0": 0.90637, "loss_cls_dn_1": 0.33966, "loss_box_dn_1": 1.05946, "loss_cls_dn_2": 0.36413, "loss_box_dn_2": 1.00169, "loss_cls_dn_3": 0.38591, "loss_box_dn_3": 0.99685, "loss_cls_dn_4": 0.35758, "loss_box_dn_4": 1.0627, "loss_cls_dn_5": 0.3721, "loss_box_dn_5": 1.08583, "loss_dense_depth": 1.14681, "loss": 34.58885, "grad_norm": 46.72591, "time": 1.53062} -{"mode": "train", "epoch": 1, "iter": 47, "lr": 0.00012, "memory": 49163, "data_time": 0.07751, "loss_cls_0": 1.04778, "loss_box_0": 1.96247, "loss_cns_0": 0.61446, "loss_yns_0": 0.16563, "loss_cls_1": 1.09419, "loss_box_1": 2.52021, "loss_cns_1": 0.59774, "loss_yns_1": 0.16733, "loss_cls_2": 1.12732, "loss_box_2": 2.42079, "loss_cns_2": 0.60806, "loss_yns_2": 0.16477, "loss_cls_3": 1.15596, "loss_box_3": 2.39293, "loss_cns_3": 0.61795, "loss_yns_3": 0.16448, "loss_cls_4": 1.17417, "loss_box_4": 2.40972, "loss_cns_4": 0.61844, "loss_yns_4": 0.1663, "loss_cls_5": 1.17413, "loss_box_5": 2.45611, "loss_cns_5": 0.62378, "loss_yns_5": 0.17617, "loss_cls_dn_0": 0.42203, "loss_box_dn_0": 0.90991, "loss_cls_dn_1": 0.34014, "loss_box_dn_1": 1.08791, "loss_cls_dn_2": 0.34684, "loss_box_dn_2": 1.06868, "loss_cls_dn_3": 0.35508, "loss_box_dn_3": 1.05733, "loss_cls_dn_4": 0.34219, "loss_box_dn_4": 1.11945, "loss_cls_dn_5": 0.36739, "loss_box_dn_5": 1.16117, "loss_dense_depth": 1.17126, "loss": 35.37031, "grad_norm": 55.20985, "time": 1.52938} -{"mode": "train", "epoch": 1, "iter": 48, "lr": 0.00012, "memory": 49163, "data_time": 0.07586, "loss_cls_0": 1.04673, "loss_box_0": 1.95921, "loss_cns_0": 0.60696, "loss_yns_0": 0.16206, "loss_cls_1": 1.07101, "loss_box_1": 2.55548, "loss_cns_1": 0.58032, "loss_yns_1": 0.16734, "loss_cls_2": 1.11982, "loss_box_2": 2.42526, "loss_cns_2": 0.60732, "loss_yns_2": 0.16279, "loss_cls_3": 1.16568, "loss_box_3": 2.43213, "loss_cns_3": 0.61851, "loss_yns_3": 0.16254, "loss_cls_4": 1.13933, "loss_box_4": 2.46782, "loss_cns_4": 0.62079, "loss_yns_4": 0.16535, "loss_cls_5": 1.15727, "loss_box_5": 2.52523, "loss_cns_5": 0.62345, "loss_yns_5": 0.16571, "loss_cls_dn_0": 0.41759, "loss_box_dn_0": 0.90265, "loss_cls_dn_1": 0.3259, "loss_box_dn_1": 1.17457, "loss_cls_dn_2": 0.34281, "loss_box_dn_2": 1.12401, "loss_cls_dn_3": 0.33673, "loss_box_dn_3": 1.10934, "loss_cls_dn_4": 0.34668, "loss_box_dn_4": 1.16267, "loss_cls_dn_5": 0.36339, "loss_box_dn_5": 1.19634, "loss_dense_depth": 1.08259, "loss": 35.59338, "grad_norm": 57.66426, "time": 1.52128} -{"mode": "train", "epoch": 1, "iter": 49, "lr": 0.00012, "memory": 49163, "data_time": 0.0777, "loss_cls_0": 1.07403, "loss_box_0": 1.96464, "loss_cns_0": 0.60445, "loss_yns_0": 0.16021, "loss_cls_1": 1.08483, "loss_box_1": 2.54729, "loss_cns_1": 0.59124, "loss_yns_1": 0.16884, "loss_cls_2": 1.10959, "loss_box_2": 2.46518, "loss_cns_2": 0.6136, "loss_yns_2": 0.16316, "loss_cls_3": 1.15053, "loss_box_3": 2.49526, "loss_cns_3": 0.62345, "loss_yns_3": 0.16377, "loss_cls_4": 1.11792, "loss_box_4": 2.561, "loss_cns_4": 0.62036, "loss_yns_4": 0.1658, "loss_cls_5": 1.14128, "loss_box_5": 2.59826, "loss_cns_5": 0.62208, "loss_yns_5": 0.17101, "loss_cls_dn_0": 0.40941, "loss_box_dn_0": 0.89299, "loss_cls_dn_1": 0.31333, "loss_box_dn_1": 1.18599, "loss_cls_dn_2": 0.33956, "loss_box_dn_2": 1.13602, "loss_cls_dn_3": 0.33521, "loss_box_dn_3": 1.12085, "loss_cls_dn_4": 0.35083, "loss_box_dn_4": 1.16069, "loss_cls_dn_5": 0.36342, "loss_box_dn_5": 1.17431, "loss_dense_depth": 1.22926, "loss": 35.98964, "grad_norm": 59.20397, "time": 1.53433} -{"mode": "train", "epoch": 1, "iter": 50, "lr": 0.00012, "memory": 49163, "data_time": 0.07595, "loss_cls_0": 1.05895, "loss_box_0": 1.96454, "loss_cns_0": 0.61156, "loss_yns_0": 0.1615, "loss_cls_1": 1.09021, "loss_box_1": 2.54632, "loss_cns_1": 0.58561, "loss_yns_1": 0.16737, "loss_cls_2": 1.09788, "loss_box_2": 2.41158, "loss_cns_2": 0.6188, "loss_yns_2": 0.16476, "loss_cls_3": 1.13211, "loss_box_3": 2.41043, "loss_cns_3": 0.62593, "loss_yns_3": 0.16288, "loss_cls_4": 1.10978, "loss_box_4": 2.4287, "loss_cns_4": 0.62711, "loss_yns_4": 0.16327, "loss_cls_5": 1.12929, "loss_box_5": 2.44347, "loss_cns_5": 0.63171, "loss_yns_5": 0.17655, "loss_cls_dn_0": 0.38717, "loss_box_dn_0": 0.89178, "loss_cls_dn_1": 0.31192, "loss_box_dn_1": 1.08733, "loss_cls_dn_2": 0.34297, "loss_box_dn_2": 1.02992, "loss_cls_dn_3": 0.3435, "loss_box_dn_3": 1.00943, "loss_cls_dn_4": 0.34613, "loss_box_dn_4": 1.01975, "loss_cls_dn_5": 0.3677, "loss_box_dn_5": 1.02479, "loss_dense_depth": 1.02048, "loss": 34.70316, "grad_norm": 44.41881, "time": 1.55742} -{"mode": "train", "epoch": 1, "iter": 51, "lr": 0.00012, "memory": 49163, "data_time": 0.0756, "loss_cls_0": 1.02711, "loss_box_0": 1.96043, "loss_cns_0": 0.61679, "loss_yns_0": 0.16426, "loss_cls_1": 1.08362, "loss_box_1": 2.4921, "loss_cns_1": 0.5904, "loss_yns_1": 0.16751, "loss_cls_2": 1.10364, "loss_box_2": 2.3684, "loss_cns_2": 0.62276, "loss_yns_2": 0.16651, "loss_cls_3": 1.1233, "loss_box_3": 2.3486, "loss_cns_3": 0.62735, "loss_yns_3": 0.16563, "loss_cls_4": 1.13047, "loss_box_4": 2.35453, "loss_cns_4": 0.63073, "loss_yns_4": 0.16545, "loss_cls_5": 1.13988, "loss_box_5": 2.33103, "loss_cns_5": 0.63578, "loss_yns_5": 0.16922, "loss_cls_dn_0": 0.38541, "loss_box_dn_0": 0.88328, "loss_cls_dn_1": 0.29373, "loss_box_dn_1": 1.07343, "loss_cls_dn_2": 0.32668, "loss_box_dn_2": 1.00103, "loss_cls_dn_3": 0.33656, "loss_box_dn_3": 0.9868, "loss_cls_dn_4": 0.3263, "loss_box_dn_4": 0.98445, "loss_cls_dn_5": 0.35125, "loss_box_dn_5": 0.97533, "loss_dense_depth": 1.21628, "loss": 34.32603, "grad_norm": 34.4872, "time": 1.63735} -{"mode": "train", "epoch": 1, "iter": 52, "lr": 0.00012, "memory": 49163, "data_time": 0.08176, "loss_cls_0": 1.03704, "loss_box_0": 1.98839, "loss_cns_0": 0.61397, "loss_yns_0": 0.16464, "loss_cls_1": 1.10237, "loss_box_1": 2.41199, "loss_cns_1": 0.59865, "loss_yns_1": 0.1668, "loss_cls_2": 1.13309, "loss_box_2": 2.32507, "loss_cns_2": 0.62677, "loss_yns_2": 0.16528, "loss_cls_3": 1.13757, "loss_box_3": 2.33625, "loss_cns_3": 0.63223, "loss_yns_3": 0.16673, "loss_cls_4": 1.13709, "loss_box_4": 2.37408, "loss_cns_4": 0.63387, "loss_yns_4": 0.16743, "loss_cls_5": 1.14557, "loss_box_5": 2.32954, "loss_cns_5": 0.63497, "loss_yns_5": 0.1676, "loss_cls_dn_0": 0.40861, "loss_box_dn_0": 0.88366, "loss_cls_dn_1": 0.28527, "loss_box_dn_1": 1.01437, "loss_cls_dn_2": 0.31829, "loss_box_dn_2": 0.95423, "loss_cls_dn_3": 0.34435, "loss_box_dn_3": 0.97687, "loss_cls_dn_4": 0.33606, "loss_box_dn_4": 0.9915, "loss_cls_dn_5": 0.36091, "loss_box_dn_5": 0.98857, "loss_dense_depth": 1.22095, "loss": 34.28063, "grad_norm": 52.00579, "time": 1.70431} -{"mode": "train", "epoch": 1, "iter": 53, "lr": 0.00012, "memory": 49163, "data_time": 0.08087, "loss_cls_0": 1.05239, "loss_box_0": 2.02952, "loss_cns_0": 0.60839, "loss_yns_0": 0.16731, "loss_cls_1": 1.10154, "loss_box_1": 2.40873, "loss_cns_1": 0.59976, "loss_yns_1": 0.17077, "loss_cls_2": 1.1206, "loss_box_2": 2.38721, "loss_cns_2": 0.61488, "loss_yns_2": 0.16766, "loss_cls_3": 1.1363, "loss_box_3": 2.39254, "loss_cns_3": 0.6195, "loss_yns_3": 0.16779, "loss_cls_4": 1.1249, "loss_box_4": 2.42251, "loss_cns_4": 0.61671, "loss_yns_4": 0.16581, "loss_cls_5": 1.14199, "loss_box_5": 2.43928, "loss_cns_5": 0.61928, "loss_yns_5": 0.16766, "loss_cls_dn_0": 0.42877, "loss_box_dn_0": 0.89091, "loss_cls_dn_1": 0.28793, "loss_box_dn_1": 1.00946, "loss_cls_dn_2": 0.31806, "loss_box_dn_2": 0.96742, "loss_cls_dn_3": 0.34994, "loss_box_dn_3": 0.99944, "loss_cls_dn_4": 0.35059, "loss_box_dn_4": 1.01109, "loss_cls_dn_5": 0.3775, "loss_box_dn_5": 1.03825, "loss_dense_depth": 1.19203, "loss": 34.66444, "grad_norm": 51.53578, "time": 1.68712} -{"mode": "train", "epoch": 1, "iter": 54, "lr": 0.00012, "memory": 49163, "data_time": 0.08313, "loss_cls_0": 1.01421, "loss_box_0": 1.94707, "loss_cns_0": 0.61873, "loss_yns_0": 0.16468, "loss_cls_1": 1.05696, "loss_box_1": 2.38706, "loss_cns_1": 0.60903, "loss_yns_1": 0.1705, "loss_cls_2": 1.08846, "loss_box_2": 2.35346, "loss_cns_2": 0.61923, "loss_yns_2": 0.16412, "loss_cls_3": 1.11259, "loss_box_3": 2.32829, "loss_cns_3": 0.62453, "loss_yns_3": 0.16371, "loss_cls_4": 1.10937, "loss_box_4": 2.33585, "loss_cns_4": 0.62337, "loss_yns_4": 0.16256, "loss_cls_5": 1.13601, "loss_box_5": 2.43278, "loss_cns_5": 0.61849, "loss_yns_5": 0.16608, "loss_cls_dn_0": 0.40936, "loss_box_dn_0": 0.8988, "loss_cls_dn_1": 0.27783, "loss_box_dn_1": 1.01924, "loss_cls_dn_2": 0.30404, "loss_box_dn_2": 0.9742, "loss_cls_dn_3": 0.33507, "loss_box_dn_3": 0.99915, "loss_cls_dn_4": 0.33929, "loss_box_dn_4": 0.99716, "loss_cls_dn_5": 0.37401, "loss_box_dn_5": 1.05899, "loss_dense_depth": 1.21498, "loss": 34.2093, "grad_norm": 48.3951, "time": 31.92593} -{"mode": "train", "epoch": 1, "iter": 55, "lr": 0.00012, "memory": 49163, "data_time": 0.4706, "loss_cls_0": 1.00949, "loss_box_0": 1.95936, "loss_cns_0": 0.61386, "loss_yns_0": 0.16821, "loss_cls_1": 1.05574, "loss_box_1": 2.27968, "loss_cns_1": 0.61731, "loss_yns_1": 0.17372, "loss_cls_2": 1.1016, "loss_box_2": 2.22853, "loss_cns_2": 0.62816, "loss_yns_2": 0.16561, "loss_cls_3": 1.11604, "loss_box_3": 2.20549, "loss_cns_3": 0.62805, "loss_yns_3": 0.16558, "loss_cls_4": 1.11427, "loss_box_4": 2.1904, "loss_cns_4": 0.62923, "loss_yns_4": 0.16494, "loss_cls_5": 1.13463, "loss_box_5": 2.28252, "loss_cns_5": 0.62076, "loss_yns_5": 0.16768, "loss_cls_dn_0": 0.40442, "loss_box_dn_0": 0.89958, "loss_cls_dn_1": 0.2993, "loss_box_dn_1": 0.97954, "loss_cls_dn_2": 0.3143, "loss_box_dn_2": 0.91797, "loss_cls_dn_3": 0.34585, "loss_box_dn_3": 0.94365, "loss_cls_dn_4": 0.35326, "loss_box_dn_4": 0.93516, "loss_cls_dn_5": 0.40495, "loss_box_dn_5": 1.00063, "loss_dense_depth": 1.09039, "loss": 33.30988, "grad_norm": 50.83277, "time": 1.91875} -{"mode": "train", "epoch": 1, "iter": 56, "lr": 0.00012, "memory": 49163, "data_time": 0.0736, "loss_cls_0": 0.99683, "loss_box_0": 1.92148, "loss_cns_0": 0.61021, "loss_yns_0": 0.16495, "loss_cls_1": 1.03757, "loss_box_1": 2.12926, "loss_cns_1": 0.61462, "loss_yns_1": 0.17253, "loss_cls_2": 1.0797, "loss_box_2": 2.12808, "loss_cns_2": 0.62088, "loss_yns_2": 0.16462, "loss_cls_3": 1.09521, "loss_box_3": 2.10047, "loss_cns_3": 0.62969, "loss_yns_3": 0.16319, "loss_cls_4": 1.07371, "loss_box_4": 2.0962, "loss_cns_4": 0.63024, "loss_yns_4": 0.16472, "loss_cls_5": 1.09328, "loss_box_5": 2.10431, "loss_cns_5": 0.62842, "loss_yns_5": 0.16608, "loss_cls_dn_0": 0.3798, "loss_box_dn_0": 0.90006, "loss_cls_dn_1": 0.28837, "loss_box_dn_1": 0.97264, "loss_cls_dn_2": 0.30194, "loss_box_dn_2": 0.92711, "loss_cls_dn_3": 0.32696, "loss_box_dn_3": 0.93774, "loss_cls_dn_4": 0.33062, "loss_box_dn_4": 0.93101, "loss_cls_dn_5": 0.37647, "loss_box_dn_5": 0.9535, "loss_dense_depth": 1.09168, "loss": 32.30415, "grad_norm": 44.39869, "time": 1.52853} -{"mode": "train", "epoch": 1, "iter": 57, "lr": 0.00012, "memory": 49163, "data_time": 0.07319, "loss_cls_0": 1.03719, "loss_box_0": 1.89584, "loss_cns_0": 0.61651, "loss_yns_0": 0.16279, "loss_cls_1": 1.04847, "loss_box_1": 2.16173, "loss_cns_1": 0.61101, "loss_yns_1": 0.16812, "loss_cls_2": 1.06299, "loss_box_2": 2.19387, "loss_cns_2": 0.62041, "loss_yns_2": 0.1647, "loss_cls_3": 1.0921, "loss_box_3": 2.15934, "loss_cns_3": 0.63227, "loss_yns_3": 0.16453, "loss_cls_4": 1.09435, "loss_box_4": 2.1624, "loss_cns_4": 0.63113, "loss_yns_4": 0.16694, "loss_cls_5": 1.10304, "loss_box_5": 2.14615, "loss_cns_5": 0.6366, "loss_yns_5": 0.16581, "loss_cls_dn_0": 0.36433, "loss_box_dn_0": 0.88535, "loss_cls_dn_1": 0.27719, "loss_box_dn_1": 0.92409, "loss_cls_dn_2": 0.29618, "loss_box_dn_2": 0.90526, "loss_cls_dn_3": 0.31173, "loss_box_dn_3": 0.89583, "loss_cls_dn_4": 0.30036, "loss_box_dn_4": 0.8904, "loss_cls_dn_5": 0.34858, "loss_box_dn_5": 0.89418, "loss_dense_depth": 1.07392, "loss": 32.26571, "grad_norm": 56.04476, "time": 1.52097} diff --git a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20251113_144114.log b/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20251113_144114.log deleted file mode 100644 index 0196231c5b8ef933d812cdacaec5d124b6a3ab5b..0000000000000000000000000000000000000000 --- a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20251113_144114.log +++ /dev/null @@ -1,3220 +0,0 @@ -2025-11-13 14:41:14,771 - mmdet - INFO - Environment info: ------------------------------------------------------------- -sys.platform: linux -Python: 3.10.12 (main, May 27 2025, 17:12:29) [GCC 11.4.0] -CUDA available: True -GPU 0,1,2,3,4,5,6,7: BW200, UBB BW1000 -CUDA_HOME: /opt/dtk -NVCC: Not Available -GCC: x86_64-linux-gnu-gcc (Ubuntu 11.4.0-1ubuntu1~22.04) 11.4.0 -PyTorch: 2.4.1 -PyTorch compiling details: PyTorch built with: - - GCC 10.3 - - C++ Version: 201703 - - Intel(R) Math Kernel Library Version 2020.0.4 Product Build 20200917 for Intel(R) 64 architecture applications - - OpenMP 201511 (a.k.a. OpenMP 4.5) - - LAPACK is enabled (usually provided by MKL) - - NNPACK is enabled - - CPU capability usage: AVX512 - - HIP Runtime 6.3.25211 - - MIOpen 2.17.0 - - Magma 2.8.0 - - Build settings: BLAS_INFO=mkl, BUILD_TYPE=Release, CXX_COMPILER=/opt/rh/gcc-toolset-10/root/usr/bin/c++, CXX_FLAGS= -D_GLIBCXX_USE_CXX11_ABI=1 -fvisibility-inlines-hidden -DUSE_PTHREADPOOL -DNDEBUG -DUSE_KINETO -DLIBKINETO_NOCUPTI -DUSE_FBGEMM -DUSE_PYTORCH_QNNPACK -DUSE_XNNPACK -DSYMBOLICATE_MOBILE_DEBUG_HANDLE -O2 -fPIC -Wall -Wextra -Werror=return-type -Werror=non-virtual-dtor -Werror=bool-operation -Wnarrowing -Wno-missing-field-initializers -Wno-type-limits -Wno-array-bounds -Wno-unknown-pragmas -Wno-unused-parameter -Wno-unused-function -Wno-unused-result -Wno-strict-overflow -Wno-strict-aliasing -Wno-stringop-overflow -Wsuggest-override -Wno-psabi -Wno-error=pedantic -Wno-error=old-style-cast -Wno-missing-braces -fdiagnostics-color=always -faligned-new -Wno-unused-but-set-variable -Wno-maybe-uninitialized -fno-math-errno -fno-trapping-math -Werror=format -Wno-stringop-overflow, FORCE_FALLBACK_CUDA_MPI=1, LAPACK_INFO=mkl, PERF_WITH_AVX=1, PERF_WITH_AVX2=1, PERF_WITH_AVX512=1, TORCH_VERSION=2.4.1, USE_CUDA=0, USE_CUDNN=OFF, USE_CUSPARSELT=OFF, USE_EXCEPTION_PTR=1, USE_GFLAGS=1, USE_GLOG=1, USE_GLOO=1, USE_MKL=ON, USE_MKLDNN=0, USE_MPI=1, USE_NCCL=1, USE_NNPACK=ON, USE_OPENMP=1, USE_ROCM=ON, USE_ROCM_KERNEL_ASSERT=OFF, - -TorchVision: 0.19.1 -OpenCV: 4.11.0 -MMCV: 1.6.1 -MMCV Compiler: GCC 11.4 -MMCV CUDA Compiler: rocm not available -MMDetection: 2.26.0+c41df4b ------------------------------------------------------------- - -2025-11-13 14:41:15,708 - mmdet - INFO - Distributed training: True -2025-11-13 14:41:16,429 - mmdet - INFO - Config: -plugin = True -plugin_dir = 'projects/mmdet3d_plugin/' -dist_params = dict(backend='nccl') -log_level = 'INFO' -work_dir = './work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704' -total_batch_size = 160 -num_gpus = 8 -batch_size = 20 -num_iters_per_epoch = 175 -num_epochs = 100 -checkpoint_epoch_interval = 20 -checkpoint_config = dict(interval=3500) -log_config = dict( - interval=1, - hooks=[ - dict(type='TextLoggerHook', by_epoch=False), - dict(type='TensorboardLoggerHook') - ]) -load_from = None -resume_from = None -workflow = [('train', 1)] -fp16 = dict(loss_scale=32.0) -input_shape = (704, 256) -tracking_test = True -tracking_threshold = 0.2 -class_names = [ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier', - 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' -] -num_classes = 10 -embed_dims = 256 -num_groups = 8 -num_decoder = 6 -num_single_frame_decoder = 1 -use_deformable_func = True -strides = [4, 8, 16, 32] -num_levels = 4 -num_depth_layers = 3 -drop_out = 0.1 -temporal = True -decouple_attn = True -with_quality_estimation = True -model = dict( - type='Sparse4D', - use_grid_mask=True, - use_deformable_func=True, - img_backbone=dict( - type='ResNet', - depth=50, - num_stages=4, - frozen_stages=-1, - norm_eval=False, - style='pytorch', - with_cp=True, - out_indices=(0, 1, 2, 3), - norm_cfg=dict(type='BN', requires_grad=True), - pretrained='ckpt/resnet50-19c8e357.pth'), - img_neck=dict( - type='FPN', - num_outs=4, - start_level=0, - out_channels=256, - add_extra_convs='on_output', - relu_before_extra_convs=True, - in_channels=[256, 512, 1024, 2048]), - depth_branch=dict( - type='DenseDepthNet', - embed_dims=256, - num_depth_layers=3, - loss_weight=0.2), - head=dict( - type='Sparse4DHead', - cls_threshold_to_reg=0.05, - decouple_attn=True, - instance_bank=dict( - type='InstanceBank', - num_anchor=900, - embed_dims=256, - anchor='nuscenes_kmeans900.npy', - anchor_handler=dict(type='SparseBox3DKeyPointsGenerator'), - num_temp_instances=600, - confidence_decay=0.6, - feat_grad=False), - anchor_encoder=dict( - type='SparseBox3DEncoder', - vel_dims=3, - embed_dims=[128, 32, 32, 64], - mode='cat', - output_fc=False, - in_loops=1, - out_loops=4), - num_single_frame_decoder=1, - operation_order=[ - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine' - ], - temp_graph_model=dict( - type='MultiheadAttention', - embed_dims=512, - num_heads=8, - batch_first=True, - dropout=0.1), - graph_model=dict( - type='MultiheadAttention', - embed_dims=512, - num_heads=8, - batch_first=True, - dropout=0.1), - norm_layer=dict(type='LN', normalized_shape=256), - ffn=dict( - type='AsymmetricFFN', - in_channels=512, - pre_norm=dict(type='LN'), - embed_dims=256, - feedforward_channels=1024, - num_fcs=2, - ffn_drop=0.1, - act_cfg=dict(type='ReLU', inplace=True)), - deformable_model=dict( - type='DeformableFeatureAggregation', - embed_dims=256, - num_groups=8, - num_levels=4, - num_cams=6, - attn_drop=0.15, - use_deformable_func=True, - use_camera_embed=True, - residual_mode='cat', - kps_generator=dict( - type='SparseBox3DKeyPointsGenerator', - num_learnable_pts=6, - fix_scale=[[0, 0, 0], [0.45, 0, 0], [-0.45, 0, 0], - [0, 0.45, 0], [0, -0.45, 0], [0, 0, 0.45], - [0, 0, -0.45]])), - refine_layer=dict( - type='SparseBox3DRefinementModule', - embed_dims=256, - num_cls=10, - refine_yaw=True, - with_quality_estimation=True), - sampler=dict( - type='SparseBox3DTarget', - num_dn_groups=5, - num_temp_dn_groups=3, - dn_noise_scale=[2.0, 2.0, 2.0, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5], - max_dn_gt=32, - add_neg_dn=True, - cls_weight=2.0, - box_weight=0.25, - reg_weights=[2.0, 2.0, 2.0, 0.5, 0.5, 0.5, 0.0, 0.0, 0.0, 0.0], - cls_wise_reg_weights=dict( - {9: [2.0, 2.0, 2.0, 1.0, 1.0, 1.0, 0.0, 0.0, 1.0, 1.0]})), - loss_cls=dict( - type='FocalLoss', - use_sigmoid=True, - gamma=2.0, - alpha=0.25, - loss_weight=2.0), - loss_reg=dict( - type='SparseBox3DLoss', - loss_box=dict(type='L1Loss', loss_weight=0.25), - loss_centerness=dict(type='CrossEntropyLoss', use_sigmoid=True), - loss_yawness=dict(type='GaussianFocalLoss'), - cls_allow_reverse=[5]), - decoder=dict(type='SparseBox3DDecoder'), - reg_weights=[2.0, 2.0, 2.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0])) -dataset_type = 'NuScenes3DDetTrackDataset' -data_root = 'data/nuscenes/' -anno_root = 'data/nuscenes_anno_pkls/' -file_client_args = dict(backend='disk') -img_norm_cfg = dict( - mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375], to_rgb=True) -train_pipeline = [ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict( - type='LoadPointsFromFile', - coord_type='LIDAR', - load_dim=5, - use_dim=5, - file_client_args=dict(backend='disk')), - dict(type='ResizeCropFlipImage'), - dict(type='MultiScaleDepthMapGenerator', downsample=[4, 8, 16]), - dict(type='BBoxRotation'), - dict(type='PhotoMetricDistortionMultiViewImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict( - type='CircleObjectRangeFilter', - class_dist_thred=[55, 55, 55, 55, 55, 55, 55, 55, 55, 55]), - dict( - type='InstanceNameFilter', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ]), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=[ - 'img', 'timestamp', 'projection_mat', 'image_wh', 'gt_depth', - 'focal', 'gt_bboxes_3d', 'gt_labels_3d' - ], - meta_keys=['T_global', 'T_global_inv', 'timestamp', 'instance_id']) -] -test_pipeline = [ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='ResizeCropFlipImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=['img', 'timestamp', 'projection_mat', 'image_wh'], - meta_keys=['T_global', 'T_global_inv', 'timestamp']) -] -input_modality = dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False) -data_basic_config = dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier', - 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval') -data_aug_conf = dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]) -data = dict( - samples_per_gpu=20, - workers_per_gpu=20, - train=dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval', - ann_file='data/nuscenes_anno_pkls/nuscenes_infos_train.pkl', - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict( - type='LoadPointsFromFile', - coord_type='LIDAR', - load_dim=5, - use_dim=5, - file_client_args=dict(backend='disk')), - dict(type='ResizeCropFlipImage'), - dict(type='MultiScaleDepthMapGenerator', downsample=[4, 8, 16]), - dict(type='BBoxRotation'), - dict(type='PhotoMetricDistortionMultiViewImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict( - type='CircleObjectRangeFilter', - class_dist_thred=[55, 55, 55, 55, 55, 55, 55, 55, 55, 55]), - dict( - type='InstanceNameFilter', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', - 'traffic_cone' - ]), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=[ - 'img', 'timestamp', 'projection_mat', 'image_wh', - 'gt_depth', 'focal', 'gt_bboxes_3d', 'gt_labels_3d' - ], - meta_keys=[ - 'T_global', 'T_global_inv', 'timestamp', 'instance_id' - ]) - ], - test_mode=False, - data_aug_conf=dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]), - with_seq_flag=True, - sequences_split_num=2, - keep_consistent_seq_aug=True), - val=dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval', - ann_file='data/nuscenes_anno_pkls/nuscenes_infos_val.pkl', - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='ResizeCropFlipImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=['img', 'timestamp', 'projection_mat', 'image_wh'], - meta_keys=['T_global', 'T_global_inv', 'timestamp']) - ], - data_aug_conf=dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]), - test_mode=True, - tracking=True, - tracking_threshold=0.2), - test=dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval', - ann_file='data/nuscenes_anno_pkls/nuscenes_infos_val.pkl', - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='ResizeCropFlipImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=['img', 'timestamp', 'projection_mat', 'image_wh'], - meta_keys=['T_global', 'T_global_inv', 'timestamp']) - ], - data_aug_conf=dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]), - test_mode=True, - tracking=True, - tracking_threshold=0.2)) -optimizer = dict( - type='Miopen_AdamW', - lr=0.0006, - weight_decay=0.001, - paramwise_cfg=dict(custom_keys=dict(img_backbone=dict(lr_mult=0.5)))) -optimizer_config = dict(grad_clip=dict(max_norm=25, norm_type=2)) -lr_config = dict( - policy='CosineAnnealing', - warmup='linear', - warmup_iters=500, - warmup_ratio=0.3333333333333333, - min_lr_ratio=0.001) -runner = dict(type='IterBasedRunner', max_iters=17500) -vis_pipeline = [ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='Collect', keys=['img'], meta_keys=['timestamp', 'lidar2img']) -] -evaluation = dict( - interval=3500, - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict( - type='Collect', keys=['img'], meta_keys=['timestamp', 'lidar2img']) - ]) -gpu_ids = range(0, 8) - -2025-11-13 14:41:16,429 - mmdet - INFO - Set random seed to 0, deterministic: False -2025-11-13 14:41:16,733 - mmdet - INFO - initialize ResNet with init_cfg {'type': 'Pretrained', 'checkpoint': 'ckpt/resnet50-19c8e357.pth'} -2025-11-13 14:41:16,978 - mmdet - INFO - initialize FPN with init_cfg {'type': 'Xavier', 'layer': 'Conv2d', 'distribution': 'uniform'} -Name of parameter - Initialization information - -img_backbone.conv1.weight - torch.Size([64, 3, 7, 7]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.bn1.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.bn1.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.conv1.weight - torch.Size([64, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn1.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn1.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.conv2.weight - torch.Size([64, 64, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn2.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn2.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.conv3.weight - torch.Size([256, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn3.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn3.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.downsample.0.weight - torch.Size([256, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.downsample.1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.downsample.1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.conv1.weight - torch.Size([64, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn1.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn1.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.conv2.weight - torch.Size([64, 64, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn2.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn2.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.conv3.weight - torch.Size([256, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn3.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn3.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.conv1.weight - torch.Size([64, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn1.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn1.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.conv2.weight - torch.Size([64, 64, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn2.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn2.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.conv3.weight - torch.Size([256, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn3.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn3.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.conv1.weight - torch.Size([128, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn1.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn1.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.conv2.weight - torch.Size([128, 128, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn2.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn2.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.conv3.weight - torch.Size([512, 128, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn3.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn3.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.downsample.0.weight - torch.Size([512, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.downsample.1.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.downsample.1.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.conv1.weight - torch.Size([128, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn1.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn1.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.conv2.weight - torch.Size([128, 128, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn2.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn2.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.conv3.weight - torch.Size([512, 128, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn3.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn3.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.conv1.weight - torch.Size([128, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn1.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn1.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.conv2.weight - torch.Size([128, 128, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn2.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn2.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.conv3.weight - torch.Size([512, 128, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn3.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn3.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.conv1.weight - torch.Size([128, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn1.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn1.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.conv2.weight - torch.Size([128, 128, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn2.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn2.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.conv3.weight - torch.Size([512, 128, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn3.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn3.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.conv1.weight - torch.Size([256, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.downsample.0.weight - torch.Size([1024, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.downsample.1.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.downsample.1.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.conv1.weight - torch.Size([512, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn1.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn1.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.conv2.weight - torch.Size([512, 512, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn2.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn2.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.conv3.weight - torch.Size([2048, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn3.weight - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn3.bias - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.downsample.0.weight - torch.Size([2048, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.downsample.1.weight - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.downsample.1.bias - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.conv1.weight - torch.Size([512, 2048, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn1.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn1.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.conv2.weight - torch.Size([512, 512, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn2.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn2.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.conv3.weight - torch.Size([2048, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn3.weight - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn3.bias - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.conv1.weight - torch.Size([512, 2048, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn1.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn1.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.conv2.weight - torch.Size([512, 512, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn2.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn2.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.conv3.weight - torch.Size([2048, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn3.weight - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn3.bias - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_neck.lateral_convs.0.conv.weight - torch.Size([256, 256, 1, 1]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.lateral_convs.0.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.lateral_convs.1.conv.weight - torch.Size([256, 512, 1, 1]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.lateral_convs.1.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.lateral_convs.2.conv.weight - torch.Size([256, 1024, 1, 1]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.lateral_convs.2.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.lateral_convs.3.conv.weight - torch.Size([256, 2048, 1, 1]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.lateral_convs.3.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.fpn_convs.0.conv.weight - torch.Size([256, 256, 3, 3]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.fpn_convs.0.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.fpn_convs.1.conv.weight - torch.Size([256, 256, 3, 3]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.fpn_convs.1.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.fpn_convs.2.conv.weight - torch.Size([256, 256, 3, 3]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.fpn_convs.2.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.fpn_convs.3.conv.weight - torch.Size([256, 256, 3, 3]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.fpn_convs.3.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.instance_bank.anchor - torch.Size([900, 11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.instance_bank.instance_feature - torch.Size([900, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.instance_bank.anchor_handler.fix_scale - torch.Size([1, 3]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.0.weight - torch.Size([128, 3]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.0.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.2.weight - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.2.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.3.weight - torch.Size([128, 128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.3.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.5.weight - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.5.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.6.weight - torch.Size([128, 128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.6.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.8.weight - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.8.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.9.weight - torch.Size([128, 128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.9.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.11.weight - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.11.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.0.weight - torch.Size([32, 3]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.0.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.2.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.2.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.3.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.3.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.5.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.5.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.6.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.6.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.8.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.8.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.9.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.9.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.11.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.11.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.0.weight - torch.Size([32, 2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.0.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.2.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.2.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.3.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.3.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.5.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.5.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.6.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.6.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.8.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.8.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.9.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.9.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.11.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.11.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.0.weight - torch.Size([64, 3]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.0.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.2.weight - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.2.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.3.weight - torch.Size([64, 64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.3.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.5.weight - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.5.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.6.weight - torch.Size([64, 64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.6.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.8.weight - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.8.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.9.weight - torch.Size([64, 64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.9.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.11.weight - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.11.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.1.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.1.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.1.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.1.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.1.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.1.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.1.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.1.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.3.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.4.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.4.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.4.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.4.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.5.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.5.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.5.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.5.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.6.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.6.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.8.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.8.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.8.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.8.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.8.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.8.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.8.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.8.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.10.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.11.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.11.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.11.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.11.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.12.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.12.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.12.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.12.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.13.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.13.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.15.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.15.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.15.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.15.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.15.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.15.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.15.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.15.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.16.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.16.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.17.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.18.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.18.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.18.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.18.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.19.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.19.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.19.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.19.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.20.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.20.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.22.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.22.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.22.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.22.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.22.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.22.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.22.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.22.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.23.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.23.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.24.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.25.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.25.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.25.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.25.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.26.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.26.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.26.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.26.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.27.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.27.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.29.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.29.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.29.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.29.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.29.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.29.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.29.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.29.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.30.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.30.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.31.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.32.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.32.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.32.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.32.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.33.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.33.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.33.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.33.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.34.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.34.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.36.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.36.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.36.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.36.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.36.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.36.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.36.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.36.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.37.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.37.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.38.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.fc_before.weight - torch.Size([512, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.fc_after.weight - torch.Size([256, 512]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.0.weight - torch.Size([1, 256, 1, 1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.0.bias - torch.Size([1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.1.weight - torch.Size([1, 256, 1, 1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.1.bias - torch.Size([1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.2.weight - torch.Size([1, 256, 1, 1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.2.bias - torch.Size([1]): -The value is the same before and after calling `init_weights` of Sparse4D -2025-11-13 14:41:17,068 - mmdet - INFO - Model: -Sparse4D( - (img_backbone): ResNet( - (conv1): Conv2d(3, 64, kernel_size=(7, 7), stride=(2, 2), padding=(3, 3), bias=False) - (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (maxpool): MaxPool2d(kernel_size=3, stride=2, padding=1, dilation=1, ceil_mode=False) - (layer1): ResLayer( - (0): Bottleneck( - (conv1): Conv2d(64, 64, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (downsample): Sequential( - (0): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - ) - ) - (1): Bottleneck( - (conv1): Conv2d(256, 64, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (2): Bottleneck( - (conv1): Conv2d(256, 64, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - ) - (layer2): ResLayer( - (0): Bottleneck( - (conv1): Conv2d(256, 128, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (downsample): Sequential( - (0): Conv2d(256, 512, kernel_size=(1, 1), stride=(2, 2), bias=False) - (1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - ) - ) - (1): Bottleneck( - (conv1): Conv2d(512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (2): Bottleneck( - (conv1): Conv2d(512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (3): Bottleneck( - (conv1): Conv2d(512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - ) - (layer3): ResLayer( - (0): Bottleneck( - (conv1): Conv2d(512, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (downsample): Sequential( - (0): Conv2d(512, 1024, kernel_size=(1, 1), stride=(2, 2), bias=False) - (1): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - ) - ) - (1): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (2): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (3): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (4): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (5): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - ) - (layer4): ResLayer( - (0): Bottleneck( - (conv1): Conv2d(1024, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(512, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (downsample): Sequential( - (0): Conv2d(1024, 2048, kernel_size=(1, 1), stride=(2, 2), bias=False) - (1): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - ) - ) - (1): Bottleneck( - (conv1): Conv2d(2048, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(512, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (2): Bottleneck( - (conv1): Conv2d(2048, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(512, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - ) - ) - init_cfg={'type': 'Pretrained', 'checkpoint': 'ckpt/resnet50-19c8e357.pth'} - (img_neck): FPN( - (lateral_convs): ModuleList( - (0): ConvModule( - (conv): Conv2d(256, 256, kernel_size=(1, 1), stride=(1, 1)) - ) - (1): ConvModule( - (conv): Conv2d(512, 256, kernel_size=(1, 1), stride=(1, 1)) - ) - (2): ConvModule( - (conv): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1)) - ) - (3): ConvModule( - (conv): Conv2d(2048, 256, kernel_size=(1, 1), stride=(1, 1)) - ) - ) - (fpn_convs): ModuleList( - (0-3): 4 x ConvModule( - (conv): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1)) - ) - ) - ) - init_cfg={'type': 'Xavier', 'layer': 'Conv2d', 'distribution': 'uniform'} - (head): Sparse4DHead( - (instance_bank): InstanceBank( - (anchor_handler): SparseBox3DKeyPointsGenerator() - ) - (anchor_encoder): SparseBox3DEncoder( - (pos_fc): Sequential( - (0): Linear(in_features=3, out_features=128, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((128,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=128, out_features=128, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((128,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=128, out_features=128, bias=True) - (7): ReLU(inplace=True) - (8): LayerNorm((128,), eps=1e-05, elementwise_affine=True) - (9): Linear(in_features=128, out_features=128, bias=True) - (10): ReLU(inplace=True) - (11): LayerNorm((128,), eps=1e-05, elementwise_affine=True) - ) - (size_fc): Sequential( - (0): Linear(in_features=3, out_features=32, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=32, out_features=32, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=32, out_features=32, bias=True) - (7): ReLU(inplace=True) - (8): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (9): Linear(in_features=32, out_features=32, bias=True) - (10): ReLU(inplace=True) - (11): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - ) - (yaw_fc): Sequential( - (0): Linear(in_features=2, out_features=32, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=32, out_features=32, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=32, out_features=32, bias=True) - (7): ReLU(inplace=True) - (8): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (9): Linear(in_features=32, out_features=32, bias=True) - (10): ReLU(inplace=True) - (11): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - ) - (vel_fc): Sequential( - (0): Linear(in_features=3, out_features=64, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((64,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=64, out_features=64, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((64,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=64, out_features=64, bias=True) - (7): ReLU(inplace=True) - (8): LayerNorm((64,), eps=1e-05, elementwise_affine=True) - (9): Linear(in_features=64, out_features=64, bias=True) - (10): ReLU(inplace=True) - (11): LayerNorm((64,), eps=1e-05, elementwise_affine=True) - ) - ) - (loss_cls): FocalLoss() - (loss_reg): SparseBox3DLoss( - (loss_box): L1Loss() - (loss_cns): CrossEntropyLoss(avg_non_ignore=False) - (loss_yns): GaussianFocalLoss() - ) - (layers): ModuleList( - (0): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (1): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (4-5): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (6): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (7): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (8): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (11-12): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (13): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (14): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (15): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (16): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (17): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (18-19): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (20): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (21): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (22): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (23): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (24): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (25-26): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (27): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (28): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (29): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (30): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (31): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (32-33): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (34): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (35): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (36): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (37): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (38): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - ) - (fc_before): Linear(in_features=256, out_features=512, bias=False) - (fc_after): Linear(in_features=512, out_features=256, bias=False) - ) - (depth_branch): DenseDepthNet( - (depth_layers): ModuleList( - (0-2): 3 x Conv2d(256, 1, kernel_size=(1, 1), stride=(1, 1)) - ) - ) - (grid_mask): GridMask() -) -2025-11-13 14:41:29,737 - mmdet - INFO - Start running, host: root@VM-120-96-tencentos, work_dir: /cfs/auto/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704 -2025-11-13 14:41:29,737 - mmdet - INFO - Hooks will be executed in the following order: -before_run: -(VERY_HIGH ) CosineAnnealingLrUpdaterHook -(ABOVE_NORMAL) Fp16OptimizerHook -(NORMAL ) CheckpointHook -(NORMAL ) CustomDistEvalHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -before_train_epoch: -(VERY_HIGH ) CosineAnnealingLrUpdaterHook -(NORMAL ) CustomDistEvalHook -(LOW ) IterTimerHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -before_train_iter: -(VERY_HIGH ) CosineAnnealingLrUpdaterHook -(NORMAL ) CustomDistEvalHook -(LOW ) IterTimerHook - -------------------- -after_train_iter: -(ABOVE_NORMAL) Fp16OptimizerHook -(NORMAL ) CheckpointHook -(NORMAL ) CustomDistEvalHook -(LOW ) IterTimerHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -after_train_epoch: -(NORMAL ) CheckpointHook -(NORMAL ) CustomDistEvalHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -before_val_epoch: -(LOW ) IterTimerHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -before_val_iter: -(LOW ) IterTimerHook - -------------------- -after_val_iter: -(LOW ) IterTimerHook - -------------------- -after_val_epoch: -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -after_run: -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -2025-11-13 14:41:29,737 - mmdet - INFO - workflow: [('train', 1)], max: 17500 iters -2025-11-13 14:41:29,740 - mmdet - INFO - Checkpoints will be saved to /cfs/auto/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704 by HardDiskBackend. diff --git a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20251113_144114.log.json b/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20251113_144114.log.json deleted file mode 100644 index 1fda3e213098b4a1f6239f0e36f9459d0a88f9c1..0000000000000000000000000000000000000000 --- a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20251113_144114.log.json +++ /dev/null @@ -1 +0,0 @@ -{"env_info": "sys.platform: linux\nPython: 3.10.12 (main, May 27 2025, 17:12:29) [GCC 11.4.0]\nCUDA available: True\nGPU 0,1,2,3,4,5,6,7: BW200, UBB BW1000\nCUDA_HOME: /opt/dtk\nNVCC: Not Available\nGCC: x86_64-linux-gnu-gcc (Ubuntu 11.4.0-1ubuntu1~22.04) 11.4.0\nPyTorch: 2.4.1\nPyTorch compiling details: PyTorch built with:\n - GCC 10.3\n - C++ Version: 201703\n - Intel(R) Math Kernel Library Version 2020.0.4 Product Build 20200917 for Intel(R) 64 architecture applications\n - OpenMP 201511 (a.k.a. OpenMP 4.5)\n - LAPACK is enabled (usually provided by MKL)\n - NNPACK is enabled\n - CPU capability usage: AVX512\n - HIP Runtime 6.3.25211\n - MIOpen 2.17.0\n - Magma 2.8.0\n - Build settings: BLAS_INFO=mkl, BUILD_TYPE=Release, CXX_COMPILER=/opt/rh/gcc-toolset-10/root/usr/bin/c++, CXX_FLAGS= -D_GLIBCXX_USE_CXX11_ABI=1 -fvisibility-inlines-hidden -DUSE_PTHREADPOOL -DNDEBUG -DUSE_KINETO -DLIBKINETO_NOCUPTI -DUSE_FBGEMM -DUSE_PYTORCH_QNNPACK -DUSE_XNNPACK -DSYMBOLICATE_MOBILE_DEBUG_HANDLE -O2 -fPIC -Wall -Wextra -Werror=return-type -Werror=non-virtual-dtor -Werror=bool-operation -Wnarrowing -Wno-missing-field-initializers -Wno-type-limits -Wno-array-bounds -Wno-unknown-pragmas -Wno-unused-parameter -Wno-unused-function -Wno-unused-result -Wno-strict-overflow -Wno-strict-aliasing -Wno-stringop-overflow -Wsuggest-override -Wno-psabi -Wno-error=pedantic -Wno-error=old-style-cast -Wno-missing-braces -fdiagnostics-color=always -faligned-new -Wno-unused-but-set-variable -Wno-maybe-uninitialized -fno-math-errno -fno-trapping-math -Werror=format -Wno-stringop-overflow, FORCE_FALLBACK_CUDA_MPI=1, LAPACK_INFO=mkl, PERF_WITH_AVX=1, PERF_WITH_AVX2=1, PERF_WITH_AVX512=1, TORCH_VERSION=2.4.1, USE_CUDA=0, USE_CUDNN=OFF, USE_CUSPARSELT=OFF, USE_EXCEPTION_PTR=1, USE_GFLAGS=1, USE_GLOG=1, USE_GLOO=1, USE_MKL=ON, USE_MKLDNN=0, USE_MPI=1, USE_NCCL=1, USE_NNPACK=ON, USE_OPENMP=1, USE_ROCM=ON, USE_ROCM_KERNEL_ASSERT=OFF, \n\nTorchVision: 0.19.1\nOpenCV: 4.11.0\nMMCV: 1.6.1\nMMCV Compiler: GCC 11.4\nMMCV CUDA Compiler: rocm not available\nMMDetection: 2.26.0+c41df4b", "config": "plugin = True\nplugin_dir = 'projects/mmdet3d_plugin/'\ndist_params = dict(backend='nccl')\nlog_level = 'INFO'\nwork_dir = './work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704'\ntotal_batch_size = 160\nnum_gpus = 8\nbatch_size = 20\nnum_iters_per_epoch = 175\nnum_epochs = 100\ncheckpoint_epoch_interval = 20\ncheckpoint_config = dict(interval=3500)\nlog_config = dict(\n interval=1,\n hooks=[\n dict(type='TextLoggerHook', by_epoch=False),\n dict(type='TensorboardLoggerHook')\n ])\nload_from = None\nresume_from = None\nworkflow = [('train', 1)]\nfp16 = dict(loss_scale=32.0)\ninput_shape = (704, 256)\ntracking_test = True\ntracking_threshold = 0.2\nclass_names = [\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier',\n 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n]\nnum_classes = 10\nembed_dims = 256\nnum_groups = 8\nnum_decoder = 6\nnum_single_frame_decoder = 1\nuse_deformable_func = True\nstrides = [4, 8, 16, 32]\nnum_levels = 4\nnum_depth_layers = 3\ndrop_out = 0.1\ntemporal = True\ndecouple_attn = True\nwith_quality_estimation = True\nmodel = dict(\n type='Sparse4D',\n use_grid_mask=True,\n use_deformable_func=True,\n img_backbone=dict(\n type='ResNet',\n depth=50,\n num_stages=4,\n frozen_stages=-1,\n norm_eval=False,\n style='pytorch',\n with_cp=True,\n out_indices=(0, 1, 2, 3),\n norm_cfg=dict(type='BN', requires_grad=True),\n pretrained='ckpt/resnet50-19c8e357.pth'),\n img_neck=dict(\n type='FPN',\n num_outs=4,\n start_level=0,\n out_channels=256,\n add_extra_convs='on_output',\n relu_before_extra_convs=True,\n in_channels=[256, 512, 1024, 2048]),\n depth_branch=dict(\n type='DenseDepthNet',\n embed_dims=256,\n num_depth_layers=3,\n loss_weight=0.2),\n head=dict(\n type='Sparse4DHead',\n cls_threshold_to_reg=0.05,\n decouple_attn=True,\n instance_bank=dict(\n type='InstanceBank',\n num_anchor=900,\n embed_dims=256,\n anchor='nuscenes_kmeans900.npy',\n anchor_handler=dict(type='SparseBox3DKeyPointsGenerator'),\n num_temp_instances=600,\n confidence_decay=0.6,\n feat_grad=False),\n anchor_encoder=dict(\n type='SparseBox3DEncoder',\n vel_dims=3,\n embed_dims=[128, 32, 32, 64],\n mode='cat',\n output_fc=False,\n in_loops=1,\n out_loops=4),\n num_single_frame_decoder=1,\n operation_order=[\n 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm',\n 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm',\n 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm',\n 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm',\n 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm',\n 'deformable', 'ffn', 'norm', 'refine'\n ],\n temp_graph_model=dict(\n type='MultiheadAttention',\n embed_dims=512,\n num_heads=8,\n batch_first=True,\n dropout=0.1),\n graph_model=dict(\n type='MultiheadAttention',\n embed_dims=512,\n num_heads=8,\n batch_first=True,\n dropout=0.1),\n norm_layer=dict(type='LN', normalized_shape=256),\n ffn=dict(\n type='AsymmetricFFN',\n in_channels=512,\n pre_norm=dict(type='LN'),\n embed_dims=256,\n feedforward_channels=1024,\n num_fcs=2,\n ffn_drop=0.1,\n act_cfg=dict(type='ReLU', inplace=True)),\n deformable_model=dict(\n type='DeformableFeatureAggregation',\n embed_dims=256,\n num_groups=8,\n num_levels=4,\n num_cams=6,\n attn_drop=0.15,\n use_deformable_func=True,\n use_camera_embed=True,\n residual_mode='cat',\n kps_generator=dict(\n type='SparseBox3DKeyPointsGenerator',\n num_learnable_pts=6,\n fix_scale=[[0, 0, 0], [0.45, 0, 0], [-0.45, 0, 0],\n [0, 0.45, 0], [0, -0.45, 0], [0, 0, 0.45],\n [0, 0, -0.45]])),\n refine_layer=dict(\n type='SparseBox3DRefinementModule',\n embed_dims=256,\n num_cls=10,\n refine_yaw=True,\n with_quality_estimation=True),\n sampler=dict(\n type='SparseBox3DTarget',\n num_dn_groups=5,\n num_temp_dn_groups=3,\n dn_noise_scale=[2.0, 2.0, 2.0, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5],\n max_dn_gt=32,\n add_neg_dn=True,\n cls_weight=2.0,\n box_weight=0.25,\n reg_weights=[2.0, 2.0, 2.0, 0.5, 0.5, 0.5, 0.0, 0.0, 0.0, 0.0],\n cls_wise_reg_weights=dict(\n {9: [2.0, 2.0, 2.0, 1.0, 1.0, 1.0, 0.0, 0.0, 1.0, 1.0]})),\n loss_cls=dict(\n type='FocalLoss',\n use_sigmoid=True,\n gamma=2.0,\n alpha=0.25,\n loss_weight=2.0),\n loss_reg=dict(\n type='SparseBox3DLoss',\n loss_box=dict(type='L1Loss', loss_weight=0.25),\n loss_centerness=dict(type='CrossEntropyLoss', use_sigmoid=True),\n loss_yawness=dict(type='GaussianFocalLoss'),\n cls_allow_reverse=[5]),\n decoder=dict(type='SparseBox3DDecoder'),\n reg_weights=[2.0, 2.0, 2.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0]))\ndataset_type = 'NuScenes3DDetTrackDataset'\ndata_root = 'data/nuscenes/'\nanno_root = 'data/nuscenes_anno_pkls/'\nfile_client_args = dict(backend='disk')\nimg_norm_cfg = dict(\n mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375], to_rgb=True)\ntrain_pipeline = [\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(\n type='LoadPointsFromFile',\n coord_type='LIDAR',\n load_dim=5,\n use_dim=5,\n file_client_args=dict(backend='disk')),\n dict(type='ResizeCropFlipImage'),\n dict(type='MultiScaleDepthMapGenerator', downsample=[4, 8, 16]),\n dict(type='BBoxRotation'),\n dict(type='PhotoMetricDistortionMultiViewImage'),\n dict(\n type='NormalizeMultiviewImage',\n mean=[123.675, 116.28, 103.53],\n std=[58.395, 57.12, 57.375],\n to_rgb=True),\n dict(\n type='CircleObjectRangeFilter',\n class_dist_thred=[55, 55, 55, 55, 55, 55, 55, 55, 55, 55]),\n dict(\n type='InstanceNameFilter',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ]),\n dict(type='NuScenesSparse4DAdaptor'),\n dict(\n type='Collect',\n keys=[\n 'img', 'timestamp', 'projection_mat', 'image_wh', 'gt_depth',\n 'focal', 'gt_bboxes_3d', 'gt_labels_3d'\n ],\n meta_keys=['T_global', 'T_global_inv', 'timestamp', 'instance_id'])\n]\ntest_pipeline = [\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(type='ResizeCropFlipImage'),\n dict(\n type='NormalizeMultiviewImage',\n mean=[123.675, 116.28, 103.53],\n std=[58.395, 57.12, 57.375],\n to_rgb=True),\n dict(type='NuScenesSparse4DAdaptor'),\n dict(\n type='Collect',\n keys=['img', 'timestamp', 'projection_mat', 'image_wh'],\n meta_keys=['T_global', 'T_global_inv', 'timestamp'])\n]\ninput_modality = dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False)\ndata_basic_config = dict(\n type='NuScenes3DDetTrackDataset',\n data_root='data/nuscenes/',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier',\n 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ],\n modality=dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False),\n version='v1.0-trainval')\ndata_aug_conf = dict(\n resize_lim=(0.4, 0.47),\n final_dim=(256, 704),\n bot_pct_lim=(0.0, 0.0),\n rot_lim=(-5.4, 5.4),\n H=900,\n W=1600,\n rand_flip=True,\n rot3d_range=[-0.3925, 0.3925])\ndata = dict(\n samples_per_gpu=20,\n workers_per_gpu=20,\n train=dict(\n type='NuScenes3DDetTrackDataset',\n data_root='data/nuscenes/',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ],\n modality=dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False),\n version='v1.0-trainval',\n ann_file='data/nuscenes_anno_pkls/nuscenes_infos_train.pkl',\n pipeline=[\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(\n type='LoadPointsFromFile',\n coord_type='LIDAR',\n load_dim=5,\n use_dim=5,\n file_client_args=dict(backend='disk')),\n dict(type='ResizeCropFlipImage'),\n dict(type='MultiScaleDepthMapGenerator', downsample=[4, 8, 16]),\n dict(type='BBoxRotation'),\n dict(type='PhotoMetricDistortionMultiViewImage'),\n dict(\n type='NormalizeMultiviewImage',\n mean=[123.675, 116.28, 103.53],\n std=[58.395, 57.12, 57.375],\n to_rgb=True),\n dict(\n type='CircleObjectRangeFilter',\n class_dist_thred=[55, 55, 55, 55, 55, 55, 55, 55, 55, 55]),\n dict(\n type='InstanceNameFilter',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian',\n 'traffic_cone'\n ]),\n dict(type='NuScenesSparse4DAdaptor'),\n dict(\n type='Collect',\n keys=[\n 'img', 'timestamp', 'projection_mat', 'image_wh',\n 'gt_depth', 'focal', 'gt_bboxes_3d', 'gt_labels_3d'\n ],\n meta_keys=[\n 'T_global', 'T_global_inv', 'timestamp', 'instance_id'\n ])\n ],\n test_mode=False,\n data_aug_conf=dict(\n resize_lim=(0.4, 0.47),\n final_dim=(256, 704),\n bot_pct_lim=(0.0, 0.0),\n rot_lim=(-5.4, 5.4),\n H=900,\n W=1600,\n rand_flip=True,\n rot3d_range=[-0.3925, 0.3925]),\n with_seq_flag=True,\n sequences_split_num=2,\n keep_consistent_seq_aug=True),\n val=dict(\n type='NuScenes3DDetTrackDataset',\n data_root='data/nuscenes/',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ],\n modality=dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False),\n version='v1.0-trainval',\n ann_file='data/nuscenes_anno_pkls/nuscenes_infos_val.pkl',\n pipeline=[\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(type='ResizeCropFlipImage'),\n dict(\n type='NormalizeMultiviewImage',\n mean=[123.675, 116.28, 103.53],\n std=[58.395, 57.12, 57.375],\n to_rgb=True),\n dict(type='NuScenesSparse4DAdaptor'),\n dict(\n type='Collect',\n keys=['img', 'timestamp', 'projection_mat', 'image_wh'],\n meta_keys=['T_global', 'T_global_inv', 'timestamp'])\n ],\n data_aug_conf=dict(\n resize_lim=(0.4, 0.47),\n final_dim=(256, 704),\n bot_pct_lim=(0.0, 0.0),\n rot_lim=(-5.4, 5.4),\n H=900,\n W=1600,\n rand_flip=True,\n rot3d_range=[-0.3925, 0.3925]),\n test_mode=True,\n tracking=True,\n tracking_threshold=0.2),\n test=dict(\n type='NuScenes3DDetTrackDataset',\n data_root='data/nuscenes/',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ],\n modality=dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False),\n version='v1.0-trainval',\n ann_file='data/nuscenes_anno_pkls/nuscenes_infos_val.pkl',\n pipeline=[\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(type='ResizeCropFlipImage'),\n dict(\n type='NormalizeMultiviewImage',\n mean=[123.675, 116.28, 103.53],\n std=[58.395, 57.12, 57.375],\n to_rgb=True),\n dict(type='NuScenesSparse4DAdaptor'),\n dict(\n type='Collect',\n keys=['img', 'timestamp', 'projection_mat', 'image_wh'],\n meta_keys=['T_global', 'T_global_inv', 'timestamp'])\n ],\n data_aug_conf=dict(\n resize_lim=(0.4, 0.47),\n final_dim=(256, 704),\n bot_pct_lim=(0.0, 0.0),\n rot_lim=(-5.4, 5.4),\n H=900,\n W=1600,\n rand_flip=True,\n rot3d_range=[-0.3925, 0.3925]),\n test_mode=True,\n tracking=True,\n tracking_threshold=0.2))\noptimizer = dict(\n type='Miopen_AdamW',\n lr=0.0006,\n weight_decay=0.001,\n paramwise_cfg=dict(custom_keys=dict(img_backbone=dict(lr_mult=0.5))))\noptimizer_config = dict(grad_clip=dict(max_norm=25, norm_type=2))\nlr_config = dict(\n policy='CosineAnnealing',\n warmup='linear',\n warmup_iters=500,\n warmup_ratio=0.3333333333333333,\n min_lr_ratio=0.001)\nrunner = dict(type='IterBasedRunner', max_iters=17500)\nvis_pipeline = [\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(type='Collect', keys=['img'], meta_keys=['timestamp', 'lidar2img'])\n]\nevaluation = dict(\n interval=3500,\n pipeline=[\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(\n type='Collect', keys=['img'], meta_keys=['timestamp', 'lidar2img'])\n ])\ngpu_ids = range(0, 8)\n", "seed": 0, "exp_name": "sparse4dv3_temporal_r50_1x8_bs6_256x704.py"} diff --git a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20251113_151351.log b/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20251113_151351.log deleted file mode 100644 index 039ebc6622d3c4c34c1b8427edce24f7d73959ce..0000000000000000000000000000000000000000 --- a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20251113_151351.log +++ /dev/null @@ -1,3220 +0,0 @@ -2025-11-13 15:13:51,420 - mmdet - INFO - Environment info: ------------------------------------------------------------- -sys.platform: linux -Python: 3.10.12 (main, May 27 2025, 17:12:29) [GCC 11.4.0] -CUDA available: True -GPU 0,1,2,3,4,5,6,7: BW200, UBB BW1000 -CUDA_HOME: /opt/dtk -NVCC: Not Available -GCC: x86_64-linux-gnu-gcc (Ubuntu 11.4.0-1ubuntu1~22.04) 11.4.0 -PyTorch: 2.4.1 -PyTorch compiling details: PyTorch built with: - - GCC 10.3 - - C++ Version: 201703 - - Intel(R) Math Kernel Library Version 2020.0.4 Product Build 20200917 for Intel(R) 64 architecture applications - - OpenMP 201511 (a.k.a. OpenMP 4.5) - - LAPACK is enabled (usually provided by MKL) - - NNPACK is enabled - - CPU capability usage: AVX512 - - HIP Runtime 6.3.25211 - - MIOpen 2.17.0 - - Magma 2.8.0 - - Build settings: BLAS_INFO=mkl, BUILD_TYPE=Release, CXX_COMPILER=/opt/rh/gcc-toolset-10/root/usr/bin/c++, CXX_FLAGS= -D_GLIBCXX_USE_CXX11_ABI=1 -fvisibility-inlines-hidden -DUSE_PTHREADPOOL -DNDEBUG -DUSE_KINETO -DLIBKINETO_NOCUPTI -DUSE_FBGEMM -DUSE_PYTORCH_QNNPACK -DUSE_XNNPACK -DSYMBOLICATE_MOBILE_DEBUG_HANDLE -O2 -fPIC -Wall -Wextra -Werror=return-type -Werror=non-virtual-dtor -Werror=bool-operation -Wnarrowing -Wno-missing-field-initializers -Wno-type-limits -Wno-array-bounds -Wno-unknown-pragmas -Wno-unused-parameter -Wno-unused-function -Wno-unused-result -Wno-strict-overflow -Wno-strict-aliasing -Wno-stringop-overflow -Wsuggest-override -Wno-psabi -Wno-error=pedantic -Wno-error=old-style-cast -Wno-missing-braces -fdiagnostics-color=always -faligned-new -Wno-unused-but-set-variable -Wno-maybe-uninitialized -fno-math-errno -fno-trapping-math -Werror=format -Wno-stringop-overflow, FORCE_FALLBACK_CUDA_MPI=1, LAPACK_INFO=mkl, PERF_WITH_AVX=1, PERF_WITH_AVX2=1, PERF_WITH_AVX512=1, TORCH_VERSION=2.4.1, USE_CUDA=0, USE_CUDNN=OFF, USE_CUSPARSELT=OFF, USE_EXCEPTION_PTR=1, USE_GFLAGS=1, USE_GLOG=1, USE_GLOO=1, USE_MKL=ON, USE_MKLDNN=0, USE_MPI=1, USE_NCCL=1, USE_NNPACK=ON, USE_OPENMP=1, USE_ROCM=ON, USE_ROCM_KERNEL_ASSERT=OFF, - -TorchVision: 0.19.1 -OpenCV: 4.11.0 -MMCV: 1.6.1 -MMCV Compiler: GCC 11.4 -MMCV CUDA Compiler: rocm not available -MMDetection: 2.26.0+c41df4b ------------------------------------------------------------- - -2025-11-13 15:13:52,359 - mmdet - INFO - Distributed training: True -2025-11-13 15:13:53,084 - mmdet - INFO - Config: -plugin = True -plugin_dir = 'projects/mmdet3d_plugin/' -dist_params = dict(backend='nccl') -log_level = 'INFO' -work_dir = './work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704' -total_batch_size = 160 -num_gpus = 8 -batch_size = 20 -num_iters_per_epoch = 175 -num_epochs = 100 -checkpoint_epoch_interval = 20 -checkpoint_config = dict(interval=3500) -log_config = dict( - interval=1, - hooks=[ - dict(type='TextLoggerHook', by_epoch=False), - dict(type='TensorboardLoggerHook') - ]) -load_from = None -resume_from = None -workflow = [('train', 1)] -fp16 = dict(loss_scale=32.0) -input_shape = (704, 256) -tracking_test = True -tracking_threshold = 0.2 -class_names = [ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier', - 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' -] -num_classes = 10 -embed_dims = 256 -num_groups = 8 -num_decoder = 6 -num_single_frame_decoder = 1 -use_deformable_func = True -strides = [4, 8, 16, 32] -num_levels = 4 -num_depth_layers = 3 -drop_out = 0.1 -temporal = True -decouple_attn = True -with_quality_estimation = True -model = dict( - type='Sparse4D', - use_grid_mask=True, - use_deformable_func=True, - img_backbone=dict( - type='ResNet', - depth=50, - num_stages=4, - frozen_stages=-1, - norm_eval=False, - style='pytorch', - with_cp=True, - out_indices=(0, 1, 2, 3), - norm_cfg=dict(type='BN', requires_grad=True), - pretrained='ckpt/resnet50-19c8e357.pth'), - img_neck=dict( - type='FPN', - num_outs=4, - start_level=0, - out_channels=256, - add_extra_convs='on_output', - relu_before_extra_convs=True, - in_channels=[256, 512, 1024, 2048]), - depth_branch=dict( - type='DenseDepthNet', - embed_dims=256, - num_depth_layers=3, - loss_weight=0.2), - head=dict( - type='Sparse4DHead', - cls_threshold_to_reg=0.05, - decouple_attn=True, - instance_bank=dict( - type='InstanceBank', - num_anchor=900, - embed_dims=256, - anchor='nuscenes_kmeans900.npy', - anchor_handler=dict(type='SparseBox3DKeyPointsGenerator'), - num_temp_instances=600, - confidence_decay=0.6, - feat_grad=False), - anchor_encoder=dict( - type='SparseBox3DEncoder', - vel_dims=3, - embed_dims=[128, 32, 32, 64], - mode='cat', - output_fc=False, - in_loops=1, - out_loops=4), - num_single_frame_decoder=1, - operation_order=[ - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine' - ], - temp_graph_model=dict( - type='MultiheadAttention', - embed_dims=512, - num_heads=8, - batch_first=True, - dropout=0.1), - graph_model=dict( - type='MultiheadAttention', - embed_dims=512, - num_heads=8, - batch_first=True, - dropout=0.1), - norm_layer=dict(type='LN', normalized_shape=256), - ffn=dict( - type='AsymmetricFFN', - in_channels=512, - pre_norm=dict(type='LN'), - embed_dims=256, - feedforward_channels=1024, - num_fcs=2, - ffn_drop=0.1, - act_cfg=dict(type='ReLU', inplace=True)), - deformable_model=dict( - type='DeformableFeatureAggregation', - embed_dims=256, - num_groups=8, - num_levels=4, - num_cams=6, - attn_drop=0.15, - use_deformable_func=True, - use_camera_embed=True, - residual_mode='cat', - kps_generator=dict( - type='SparseBox3DKeyPointsGenerator', - num_learnable_pts=6, - fix_scale=[[0, 0, 0], [0.45, 0, 0], [-0.45, 0, 0], - [0, 0.45, 0], [0, -0.45, 0], [0, 0, 0.45], - [0, 0, -0.45]])), - refine_layer=dict( - type='SparseBox3DRefinementModule', - embed_dims=256, - num_cls=10, - refine_yaw=True, - with_quality_estimation=True), - sampler=dict( - type='SparseBox3DTarget', - num_dn_groups=5, - num_temp_dn_groups=3, - dn_noise_scale=[2.0, 2.0, 2.0, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5], - max_dn_gt=32, - add_neg_dn=True, - cls_weight=2.0, - box_weight=0.25, - reg_weights=[2.0, 2.0, 2.0, 0.5, 0.5, 0.5, 0.0, 0.0, 0.0, 0.0], - cls_wise_reg_weights=dict( - {9: [2.0, 2.0, 2.0, 1.0, 1.0, 1.0, 0.0, 0.0, 1.0, 1.0]})), - loss_cls=dict( - type='FocalLoss', - use_sigmoid=True, - gamma=2.0, - alpha=0.25, - loss_weight=2.0), - loss_reg=dict( - type='SparseBox3DLoss', - loss_box=dict(type='L1Loss', loss_weight=0.25), - loss_centerness=dict(type='CrossEntropyLoss', use_sigmoid=True), - loss_yawness=dict(type='GaussianFocalLoss'), - cls_allow_reverse=[5]), - decoder=dict(type='SparseBox3DDecoder'), - reg_weights=[2.0, 2.0, 2.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0])) -dataset_type = 'NuScenes3DDetTrackDataset' -data_root = 'data/nuscenes/' -anno_root = 'data/nuscenes_anno_pkls/' -file_client_args = dict(backend='disk') -img_norm_cfg = dict( - mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375], to_rgb=True) -train_pipeline = [ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict( - type='LoadPointsFromFile', - coord_type='LIDAR', - load_dim=5, - use_dim=5, - file_client_args=dict(backend='disk')), - dict(type='ResizeCropFlipImage'), - dict(type='MultiScaleDepthMapGenerator', downsample=[4, 8, 16]), - dict(type='BBoxRotation'), - dict(type='PhotoMetricDistortionMultiViewImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict( - type='CircleObjectRangeFilter', - class_dist_thred=[55, 55, 55, 55, 55, 55, 55, 55, 55, 55]), - dict( - type='InstanceNameFilter', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ]), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=[ - 'img', 'timestamp', 'projection_mat', 'image_wh', 'gt_depth', - 'focal', 'gt_bboxes_3d', 'gt_labels_3d' - ], - meta_keys=['T_global', 'T_global_inv', 'timestamp', 'instance_id']) -] -test_pipeline = [ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='ResizeCropFlipImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=['img', 'timestamp', 'projection_mat', 'image_wh'], - meta_keys=['T_global', 'T_global_inv', 'timestamp']) -] -input_modality = dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False) -data_basic_config = dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier', - 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval') -data_aug_conf = dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]) -data = dict( - samples_per_gpu=20, - workers_per_gpu=20, - train=dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval', - ann_file='data/nuscenes_anno_pkls/nuscenes_infos_train.pkl', - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict( - type='LoadPointsFromFile', - coord_type='LIDAR', - load_dim=5, - use_dim=5, - file_client_args=dict(backend='disk')), - dict(type='ResizeCropFlipImage'), - dict(type='MultiScaleDepthMapGenerator', downsample=[4, 8, 16]), - dict(type='BBoxRotation'), - dict(type='PhotoMetricDistortionMultiViewImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict( - type='CircleObjectRangeFilter', - class_dist_thred=[55, 55, 55, 55, 55, 55, 55, 55, 55, 55]), - dict( - type='InstanceNameFilter', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', - 'traffic_cone' - ]), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=[ - 'img', 'timestamp', 'projection_mat', 'image_wh', - 'gt_depth', 'focal', 'gt_bboxes_3d', 'gt_labels_3d' - ], - meta_keys=[ - 'T_global', 'T_global_inv', 'timestamp', 'instance_id' - ]) - ], - test_mode=False, - data_aug_conf=dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]), - with_seq_flag=True, - sequences_split_num=2, - keep_consistent_seq_aug=True), - val=dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval', - ann_file='data/nuscenes_anno_pkls/nuscenes_infos_val.pkl', - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='ResizeCropFlipImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=['img', 'timestamp', 'projection_mat', 'image_wh'], - meta_keys=['T_global', 'T_global_inv', 'timestamp']) - ], - data_aug_conf=dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]), - test_mode=True, - tracking=True, - tracking_threshold=0.2), - test=dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval', - ann_file='data/nuscenes_anno_pkls/nuscenes_infos_val.pkl', - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='ResizeCropFlipImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=['img', 'timestamp', 'projection_mat', 'image_wh'], - meta_keys=['T_global', 'T_global_inv', 'timestamp']) - ], - data_aug_conf=dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]), - test_mode=True, - tracking=True, - tracking_threshold=0.2)) -optimizer = dict( - type='Miopen_AdamW', - lr=0.0006, - weight_decay=0.001, - paramwise_cfg=dict(custom_keys=dict(img_backbone=dict(lr_mult=0.5)))) -optimizer_config = dict(grad_clip=dict(max_norm=25, norm_type=2)) -lr_config = dict( - policy='CosineAnnealing', - warmup='linear', - warmup_iters=500, - warmup_ratio=0.3333333333333333, - min_lr_ratio=0.001) -runner = dict(type='IterBasedRunner', max_iters=17500) -vis_pipeline = [ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='Collect', keys=['img'], meta_keys=['timestamp', 'lidar2img']) -] -evaluation = dict( - interval=3500, - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict( - type='Collect', keys=['img'], meta_keys=['timestamp', 'lidar2img']) - ]) -gpu_ids = range(0, 8) - -2025-11-13 15:13:53,084 - mmdet - INFO - Set random seed to 0, deterministic: False -2025-11-13 15:13:53,387 - mmdet - INFO - initialize ResNet with init_cfg {'type': 'Pretrained', 'checkpoint': 'ckpt/resnet50-19c8e357.pth'} -2025-11-13 15:13:53,667 - mmdet - INFO - initialize FPN with init_cfg {'type': 'Xavier', 'layer': 'Conv2d', 'distribution': 'uniform'} -Name of parameter - Initialization information - -img_backbone.conv1.weight - torch.Size([64, 3, 7, 7]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.bn1.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.bn1.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.conv1.weight - torch.Size([64, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn1.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn1.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.conv2.weight - torch.Size([64, 64, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn2.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn2.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.conv3.weight - torch.Size([256, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn3.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn3.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.downsample.0.weight - torch.Size([256, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.downsample.1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.downsample.1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.conv1.weight - torch.Size([64, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn1.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn1.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.conv2.weight - torch.Size([64, 64, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn2.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn2.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.conv3.weight - torch.Size([256, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn3.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn3.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.conv1.weight - torch.Size([64, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn1.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn1.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.conv2.weight - torch.Size([64, 64, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn2.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn2.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.conv3.weight - torch.Size([256, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn3.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn3.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.conv1.weight - torch.Size([128, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn1.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn1.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.conv2.weight - torch.Size([128, 128, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn2.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn2.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.conv3.weight - torch.Size([512, 128, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn3.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn3.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.downsample.0.weight - torch.Size([512, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.downsample.1.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.downsample.1.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.conv1.weight - torch.Size([128, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn1.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn1.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.conv2.weight - torch.Size([128, 128, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn2.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn2.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.conv3.weight - torch.Size([512, 128, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn3.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn3.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.conv1.weight - torch.Size([128, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn1.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn1.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.conv2.weight - torch.Size([128, 128, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn2.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn2.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.conv3.weight - torch.Size([512, 128, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn3.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn3.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.conv1.weight - torch.Size([128, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn1.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn1.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.conv2.weight - torch.Size([128, 128, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn2.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn2.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.conv3.weight - torch.Size([512, 128, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn3.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn3.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.conv1.weight - torch.Size([256, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.downsample.0.weight - torch.Size([1024, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.downsample.1.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.downsample.1.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.conv1.weight - torch.Size([512, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn1.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn1.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.conv2.weight - torch.Size([512, 512, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn2.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn2.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.conv3.weight - torch.Size([2048, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn3.weight - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn3.bias - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.downsample.0.weight - torch.Size([2048, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.downsample.1.weight - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.downsample.1.bias - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.conv1.weight - torch.Size([512, 2048, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn1.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn1.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.conv2.weight - torch.Size([512, 512, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn2.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn2.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.conv3.weight - torch.Size([2048, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn3.weight - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn3.bias - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.conv1.weight - torch.Size([512, 2048, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn1.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn1.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.conv2.weight - torch.Size([512, 512, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn2.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn2.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.conv3.weight - torch.Size([2048, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn3.weight - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn3.bias - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_neck.lateral_convs.0.conv.weight - torch.Size([256, 256, 1, 1]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.lateral_convs.0.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.lateral_convs.1.conv.weight - torch.Size([256, 512, 1, 1]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.lateral_convs.1.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.lateral_convs.2.conv.weight - torch.Size([256, 1024, 1, 1]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.lateral_convs.2.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.lateral_convs.3.conv.weight - torch.Size([256, 2048, 1, 1]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.lateral_convs.3.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.fpn_convs.0.conv.weight - torch.Size([256, 256, 3, 3]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.fpn_convs.0.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.fpn_convs.1.conv.weight - torch.Size([256, 256, 3, 3]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.fpn_convs.1.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.fpn_convs.2.conv.weight - torch.Size([256, 256, 3, 3]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.fpn_convs.2.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.fpn_convs.3.conv.weight - torch.Size([256, 256, 3, 3]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.fpn_convs.3.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.instance_bank.anchor - torch.Size([900, 11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.instance_bank.instance_feature - torch.Size([900, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.instance_bank.anchor_handler.fix_scale - torch.Size([1, 3]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.0.weight - torch.Size([128, 3]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.0.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.2.weight - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.2.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.3.weight - torch.Size([128, 128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.3.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.5.weight - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.5.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.6.weight - torch.Size([128, 128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.6.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.8.weight - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.8.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.9.weight - torch.Size([128, 128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.9.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.11.weight - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.11.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.0.weight - torch.Size([32, 3]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.0.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.2.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.2.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.3.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.3.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.5.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.5.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.6.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.6.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.8.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.8.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.9.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.9.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.11.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.11.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.0.weight - torch.Size([32, 2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.0.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.2.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.2.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.3.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.3.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.5.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.5.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.6.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.6.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.8.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.8.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.9.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.9.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.11.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.11.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.0.weight - torch.Size([64, 3]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.0.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.2.weight - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.2.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.3.weight - torch.Size([64, 64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.3.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.5.weight - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.5.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.6.weight - torch.Size([64, 64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.6.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.8.weight - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.8.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.9.weight - torch.Size([64, 64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.9.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.11.weight - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.11.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.1.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.1.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.1.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.1.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.1.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.1.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.1.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.1.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.3.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.4.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.4.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.4.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.4.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.5.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.5.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.5.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.5.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.6.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.6.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.8.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.8.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.8.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.8.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.8.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.8.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.8.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.8.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.10.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.11.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.11.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.11.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.11.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.12.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.12.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.12.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.12.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.13.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.13.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.15.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.15.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.15.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.15.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.15.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.15.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.15.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.15.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.16.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.16.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.17.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.18.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.18.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.18.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.18.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.19.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.19.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.19.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.19.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.20.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.20.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.22.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.22.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.22.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.22.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.22.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.22.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.22.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.22.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.23.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.23.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.24.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.25.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.25.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.25.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.25.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.26.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.26.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.26.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.26.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.27.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.27.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.29.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.29.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.29.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.29.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.29.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.29.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.29.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.29.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.30.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.30.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.31.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.32.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.32.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.32.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.32.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.33.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.33.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.33.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.33.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.34.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.34.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.36.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.36.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.36.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.36.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.36.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.36.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.36.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.36.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.37.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.37.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.38.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.fc_before.weight - torch.Size([512, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.fc_after.weight - torch.Size([256, 512]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.0.weight - torch.Size([1, 256, 1, 1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.0.bias - torch.Size([1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.1.weight - torch.Size([1, 256, 1, 1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.1.bias - torch.Size([1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.2.weight - torch.Size([1, 256, 1, 1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.2.bias - torch.Size([1]): -The value is the same before and after calling `init_weights` of Sparse4D -2025-11-13 15:13:53,757 - mmdet - INFO - Model: -Sparse4D( - (img_backbone): ResNet( - (conv1): Conv2d(3, 64, kernel_size=(7, 7), stride=(2, 2), padding=(3, 3), bias=False) - (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (maxpool): MaxPool2d(kernel_size=3, stride=2, padding=1, dilation=1, ceil_mode=False) - (layer1): ResLayer( - (0): Bottleneck( - (conv1): Conv2d(64, 64, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (downsample): Sequential( - (0): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - ) - ) - (1): Bottleneck( - (conv1): Conv2d(256, 64, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (2): Bottleneck( - (conv1): Conv2d(256, 64, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - ) - (layer2): ResLayer( - (0): Bottleneck( - (conv1): Conv2d(256, 128, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (downsample): Sequential( - (0): Conv2d(256, 512, kernel_size=(1, 1), stride=(2, 2), bias=False) - (1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - ) - ) - (1): Bottleneck( - (conv1): Conv2d(512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (2): Bottleneck( - (conv1): Conv2d(512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (3): Bottleneck( - (conv1): Conv2d(512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - ) - (layer3): ResLayer( - (0): Bottleneck( - (conv1): Conv2d(512, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (downsample): Sequential( - (0): Conv2d(512, 1024, kernel_size=(1, 1), stride=(2, 2), bias=False) - (1): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - ) - ) - (1): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (2): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (3): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (4): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (5): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - ) - (layer4): ResLayer( - (0): Bottleneck( - (conv1): Conv2d(1024, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(512, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (downsample): Sequential( - (0): Conv2d(1024, 2048, kernel_size=(1, 1), stride=(2, 2), bias=False) - (1): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - ) - ) - (1): Bottleneck( - (conv1): Conv2d(2048, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(512, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (2): Bottleneck( - (conv1): Conv2d(2048, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(512, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - ) - ) - init_cfg={'type': 'Pretrained', 'checkpoint': 'ckpt/resnet50-19c8e357.pth'} - (img_neck): FPN( - (lateral_convs): ModuleList( - (0): ConvModule( - (conv): Conv2d(256, 256, kernel_size=(1, 1), stride=(1, 1)) - ) - (1): ConvModule( - (conv): Conv2d(512, 256, kernel_size=(1, 1), stride=(1, 1)) - ) - (2): ConvModule( - (conv): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1)) - ) - (3): ConvModule( - (conv): Conv2d(2048, 256, kernel_size=(1, 1), stride=(1, 1)) - ) - ) - (fpn_convs): ModuleList( - (0-3): 4 x ConvModule( - (conv): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1)) - ) - ) - ) - init_cfg={'type': 'Xavier', 'layer': 'Conv2d', 'distribution': 'uniform'} - (head): Sparse4DHead( - (instance_bank): InstanceBank( - (anchor_handler): SparseBox3DKeyPointsGenerator() - ) - (anchor_encoder): SparseBox3DEncoder( - (pos_fc): Sequential( - (0): Linear(in_features=3, out_features=128, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((128,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=128, out_features=128, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((128,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=128, out_features=128, bias=True) - (7): ReLU(inplace=True) - (8): LayerNorm((128,), eps=1e-05, elementwise_affine=True) - (9): Linear(in_features=128, out_features=128, bias=True) - (10): ReLU(inplace=True) - (11): LayerNorm((128,), eps=1e-05, elementwise_affine=True) - ) - (size_fc): Sequential( - (0): Linear(in_features=3, out_features=32, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=32, out_features=32, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=32, out_features=32, bias=True) - (7): ReLU(inplace=True) - (8): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (9): Linear(in_features=32, out_features=32, bias=True) - (10): ReLU(inplace=True) - (11): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - ) - (yaw_fc): Sequential( - (0): Linear(in_features=2, out_features=32, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=32, out_features=32, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=32, out_features=32, bias=True) - (7): ReLU(inplace=True) - (8): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (9): Linear(in_features=32, out_features=32, bias=True) - (10): ReLU(inplace=True) - (11): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - ) - (vel_fc): Sequential( - (0): Linear(in_features=3, out_features=64, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((64,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=64, out_features=64, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((64,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=64, out_features=64, bias=True) - (7): ReLU(inplace=True) - (8): LayerNorm((64,), eps=1e-05, elementwise_affine=True) - (9): Linear(in_features=64, out_features=64, bias=True) - (10): ReLU(inplace=True) - (11): LayerNorm((64,), eps=1e-05, elementwise_affine=True) - ) - ) - (loss_cls): FocalLoss() - (loss_reg): SparseBox3DLoss( - (loss_box): L1Loss() - (loss_cns): CrossEntropyLoss(avg_non_ignore=False) - (loss_yns): GaussianFocalLoss() - ) - (layers): ModuleList( - (0): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (1): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (4-5): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (6): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (7): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (8): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (11-12): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (13): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (14): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (15): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (16): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (17): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (18-19): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (20): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (21): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (22): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (23): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (24): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (25-26): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (27): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (28): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (29): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (30): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (31): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (32-33): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (34): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (35): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (36): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (37): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (38): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - ) - (fc_before): Linear(in_features=256, out_features=512, bias=False) - (fc_after): Linear(in_features=512, out_features=256, bias=False) - ) - (depth_branch): DenseDepthNet( - (depth_layers): ModuleList( - (0-2): 3 x Conv2d(256, 1, kernel_size=(1, 1), stride=(1, 1)) - ) - ) - (grid_mask): GridMask() -) -2025-11-13 15:14:06,399 - mmdet - INFO - Start running, host: root@VM-120-96-tencentos, work_dir: /cfs/auto/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704 -2025-11-13 15:14:06,400 - mmdet - INFO - Hooks will be executed in the following order: -before_run: -(VERY_HIGH ) CosineAnnealingLrUpdaterHook -(ABOVE_NORMAL) Fp16OptimizerHook -(NORMAL ) CheckpointHook -(NORMAL ) CustomDistEvalHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -before_train_epoch: -(VERY_HIGH ) CosineAnnealingLrUpdaterHook -(NORMAL ) CustomDistEvalHook -(LOW ) IterTimerHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -before_train_iter: -(VERY_HIGH ) CosineAnnealingLrUpdaterHook -(NORMAL ) CustomDistEvalHook -(LOW ) IterTimerHook - -------------------- -after_train_iter: -(ABOVE_NORMAL) Fp16OptimizerHook -(NORMAL ) CheckpointHook -(NORMAL ) CustomDistEvalHook -(LOW ) IterTimerHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -after_train_epoch: -(NORMAL ) CheckpointHook -(NORMAL ) CustomDistEvalHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -before_val_epoch: -(LOW ) IterTimerHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -before_val_iter: -(LOW ) IterTimerHook - -------------------- -after_val_iter: -(LOW ) IterTimerHook - -------------------- -after_val_epoch: -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -after_run: -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -2025-11-13 15:14:06,400 - mmdet - INFO - workflow: [('train', 1)], max: 17500 iters -2025-11-13 15:14:06,402 - mmdet - INFO - Checkpoints will be saved to /cfs/auto/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704 by HardDiskBackend. diff --git a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20251113_151351.log.json b/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20251113_151351.log.json deleted file mode 100644 index 1fda3e213098b4a1f6239f0e36f9459d0a88f9c1..0000000000000000000000000000000000000000 --- a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20251113_151351.log.json +++ /dev/null @@ -1 +0,0 @@ -{"env_info": "sys.platform: linux\nPython: 3.10.12 (main, May 27 2025, 17:12:29) [GCC 11.4.0]\nCUDA available: True\nGPU 0,1,2,3,4,5,6,7: BW200, UBB BW1000\nCUDA_HOME: /opt/dtk\nNVCC: Not Available\nGCC: x86_64-linux-gnu-gcc (Ubuntu 11.4.0-1ubuntu1~22.04) 11.4.0\nPyTorch: 2.4.1\nPyTorch compiling details: PyTorch built with:\n - GCC 10.3\n - C++ Version: 201703\n - Intel(R) Math Kernel Library Version 2020.0.4 Product Build 20200917 for Intel(R) 64 architecture applications\n - OpenMP 201511 (a.k.a. OpenMP 4.5)\n - LAPACK is enabled (usually provided by MKL)\n - NNPACK is enabled\n - CPU capability usage: AVX512\n - HIP Runtime 6.3.25211\n - MIOpen 2.17.0\n - Magma 2.8.0\n - Build settings: BLAS_INFO=mkl, BUILD_TYPE=Release, CXX_COMPILER=/opt/rh/gcc-toolset-10/root/usr/bin/c++, CXX_FLAGS= -D_GLIBCXX_USE_CXX11_ABI=1 -fvisibility-inlines-hidden -DUSE_PTHREADPOOL -DNDEBUG -DUSE_KINETO -DLIBKINETO_NOCUPTI -DUSE_FBGEMM -DUSE_PYTORCH_QNNPACK -DUSE_XNNPACK -DSYMBOLICATE_MOBILE_DEBUG_HANDLE -O2 -fPIC -Wall -Wextra -Werror=return-type -Werror=non-virtual-dtor -Werror=bool-operation -Wnarrowing -Wno-missing-field-initializers -Wno-type-limits -Wno-array-bounds -Wno-unknown-pragmas -Wno-unused-parameter -Wno-unused-function -Wno-unused-result -Wno-strict-overflow -Wno-strict-aliasing -Wno-stringop-overflow -Wsuggest-override -Wno-psabi -Wno-error=pedantic -Wno-error=old-style-cast -Wno-missing-braces -fdiagnostics-color=always -faligned-new -Wno-unused-but-set-variable -Wno-maybe-uninitialized -fno-math-errno -fno-trapping-math -Werror=format -Wno-stringop-overflow, FORCE_FALLBACK_CUDA_MPI=1, LAPACK_INFO=mkl, PERF_WITH_AVX=1, PERF_WITH_AVX2=1, PERF_WITH_AVX512=1, TORCH_VERSION=2.4.1, USE_CUDA=0, USE_CUDNN=OFF, USE_CUSPARSELT=OFF, USE_EXCEPTION_PTR=1, USE_GFLAGS=1, USE_GLOG=1, USE_GLOO=1, USE_MKL=ON, USE_MKLDNN=0, USE_MPI=1, USE_NCCL=1, USE_NNPACK=ON, USE_OPENMP=1, USE_ROCM=ON, USE_ROCM_KERNEL_ASSERT=OFF, \n\nTorchVision: 0.19.1\nOpenCV: 4.11.0\nMMCV: 1.6.1\nMMCV Compiler: GCC 11.4\nMMCV CUDA Compiler: rocm not available\nMMDetection: 2.26.0+c41df4b", "config": "plugin = True\nplugin_dir = 'projects/mmdet3d_plugin/'\ndist_params = dict(backend='nccl')\nlog_level = 'INFO'\nwork_dir = './work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704'\ntotal_batch_size = 160\nnum_gpus = 8\nbatch_size = 20\nnum_iters_per_epoch = 175\nnum_epochs = 100\ncheckpoint_epoch_interval = 20\ncheckpoint_config = dict(interval=3500)\nlog_config = dict(\n interval=1,\n hooks=[\n dict(type='TextLoggerHook', by_epoch=False),\n dict(type='TensorboardLoggerHook')\n ])\nload_from = None\nresume_from = None\nworkflow = [('train', 1)]\nfp16 = dict(loss_scale=32.0)\ninput_shape = (704, 256)\ntracking_test = True\ntracking_threshold = 0.2\nclass_names = [\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier',\n 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n]\nnum_classes = 10\nembed_dims = 256\nnum_groups = 8\nnum_decoder = 6\nnum_single_frame_decoder = 1\nuse_deformable_func = True\nstrides = [4, 8, 16, 32]\nnum_levels = 4\nnum_depth_layers = 3\ndrop_out = 0.1\ntemporal = True\ndecouple_attn = True\nwith_quality_estimation = True\nmodel = dict(\n type='Sparse4D',\n use_grid_mask=True,\n use_deformable_func=True,\n img_backbone=dict(\n type='ResNet',\n depth=50,\n num_stages=4,\n frozen_stages=-1,\n norm_eval=False,\n style='pytorch',\n with_cp=True,\n out_indices=(0, 1, 2, 3),\n norm_cfg=dict(type='BN', requires_grad=True),\n pretrained='ckpt/resnet50-19c8e357.pth'),\n img_neck=dict(\n type='FPN',\n num_outs=4,\n start_level=0,\n out_channels=256,\n add_extra_convs='on_output',\n relu_before_extra_convs=True,\n in_channels=[256, 512, 1024, 2048]),\n depth_branch=dict(\n type='DenseDepthNet',\n embed_dims=256,\n num_depth_layers=3,\n loss_weight=0.2),\n head=dict(\n type='Sparse4DHead',\n cls_threshold_to_reg=0.05,\n decouple_attn=True,\n instance_bank=dict(\n type='InstanceBank',\n num_anchor=900,\n embed_dims=256,\n anchor='nuscenes_kmeans900.npy',\n anchor_handler=dict(type='SparseBox3DKeyPointsGenerator'),\n num_temp_instances=600,\n confidence_decay=0.6,\n feat_grad=False),\n anchor_encoder=dict(\n type='SparseBox3DEncoder',\n vel_dims=3,\n embed_dims=[128, 32, 32, 64],\n mode='cat',\n output_fc=False,\n in_loops=1,\n out_loops=4),\n num_single_frame_decoder=1,\n operation_order=[\n 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm',\n 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm',\n 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm',\n 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm',\n 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm',\n 'deformable', 'ffn', 'norm', 'refine'\n ],\n temp_graph_model=dict(\n type='MultiheadAttention',\n embed_dims=512,\n num_heads=8,\n batch_first=True,\n dropout=0.1),\n graph_model=dict(\n type='MultiheadAttention',\n embed_dims=512,\n num_heads=8,\n batch_first=True,\n dropout=0.1),\n norm_layer=dict(type='LN', normalized_shape=256),\n ffn=dict(\n type='AsymmetricFFN',\n in_channels=512,\n pre_norm=dict(type='LN'),\n embed_dims=256,\n feedforward_channels=1024,\n num_fcs=2,\n ffn_drop=0.1,\n act_cfg=dict(type='ReLU', inplace=True)),\n deformable_model=dict(\n type='DeformableFeatureAggregation',\n embed_dims=256,\n num_groups=8,\n num_levels=4,\n num_cams=6,\n attn_drop=0.15,\n use_deformable_func=True,\n use_camera_embed=True,\n residual_mode='cat',\n kps_generator=dict(\n type='SparseBox3DKeyPointsGenerator',\n num_learnable_pts=6,\n fix_scale=[[0, 0, 0], [0.45, 0, 0], [-0.45, 0, 0],\n [0, 0.45, 0], [0, -0.45, 0], [0, 0, 0.45],\n [0, 0, -0.45]])),\n refine_layer=dict(\n type='SparseBox3DRefinementModule',\n embed_dims=256,\n num_cls=10,\n refine_yaw=True,\n with_quality_estimation=True),\n sampler=dict(\n type='SparseBox3DTarget',\n num_dn_groups=5,\n num_temp_dn_groups=3,\n dn_noise_scale=[2.0, 2.0, 2.0, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5],\n max_dn_gt=32,\n add_neg_dn=True,\n cls_weight=2.0,\n box_weight=0.25,\n reg_weights=[2.0, 2.0, 2.0, 0.5, 0.5, 0.5, 0.0, 0.0, 0.0, 0.0],\n cls_wise_reg_weights=dict(\n {9: [2.0, 2.0, 2.0, 1.0, 1.0, 1.0, 0.0, 0.0, 1.0, 1.0]})),\n loss_cls=dict(\n type='FocalLoss',\n use_sigmoid=True,\n gamma=2.0,\n alpha=0.25,\n loss_weight=2.0),\n loss_reg=dict(\n type='SparseBox3DLoss',\n loss_box=dict(type='L1Loss', loss_weight=0.25),\n loss_centerness=dict(type='CrossEntropyLoss', use_sigmoid=True),\n loss_yawness=dict(type='GaussianFocalLoss'),\n cls_allow_reverse=[5]),\n decoder=dict(type='SparseBox3DDecoder'),\n reg_weights=[2.0, 2.0, 2.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0]))\ndataset_type = 'NuScenes3DDetTrackDataset'\ndata_root = 'data/nuscenes/'\nanno_root = 'data/nuscenes_anno_pkls/'\nfile_client_args = dict(backend='disk')\nimg_norm_cfg = dict(\n mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375], to_rgb=True)\ntrain_pipeline = [\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(\n type='LoadPointsFromFile',\n coord_type='LIDAR',\n load_dim=5,\n use_dim=5,\n file_client_args=dict(backend='disk')),\n dict(type='ResizeCropFlipImage'),\n dict(type='MultiScaleDepthMapGenerator', downsample=[4, 8, 16]),\n dict(type='BBoxRotation'),\n dict(type='PhotoMetricDistortionMultiViewImage'),\n dict(\n type='NormalizeMultiviewImage',\n mean=[123.675, 116.28, 103.53],\n std=[58.395, 57.12, 57.375],\n to_rgb=True),\n dict(\n type='CircleObjectRangeFilter',\n class_dist_thred=[55, 55, 55, 55, 55, 55, 55, 55, 55, 55]),\n dict(\n type='InstanceNameFilter',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ]),\n dict(type='NuScenesSparse4DAdaptor'),\n dict(\n type='Collect',\n keys=[\n 'img', 'timestamp', 'projection_mat', 'image_wh', 'gt_depth',\n 'focal', 'gt_bboxes_3d', 'gt_labels_3d'\n ],\n meta_keys=['T_global', 'T_global_inv', 'timestamp', 'instance_id'])\n]\ntest_pipeline = [\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(type='ResizeCropFlipImage'),\n dict(\n type='NormalizeMultiviewImage',\n mean=[123.675, 116.28, 103.53],\n std=[58.395, 57.12, 57.375],\n to_rgb=True),\n dict(type='NuScenesSparse4DAdaptor'),\n dict(\n type='Collect',\n keys=['img', 'timestamp', 'projection_mat', 'image_wh'],\n meta_keys=['T_global', 'T_global_inv', 'timestamp'])\n]\ninput_modality = dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False)\ndata_basic_config = dict(\n type='NuScenes3DDetTrackDataset',\n data_root='data/nuscenes/',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier',\n 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ],\n modality=dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False),\n version='v1.0-trainval')\ndata_aug_conf = dict(\n resize_lim=(0.4, 0.47),\n final_dim=(256, 704),\n bot_pct_lim=(0.0, 0.0),\n rot_lim=(-5.4, 5.4),\n H=900,\n W=1600,\n rand_flip=True,\n rot3d_range=[-0.3925, 0.3925])\ndata = dict(\n samples_per_gpu=20,\n workers_per_gpu=20,\n train=dict(\n type='NuScenes3DDetTrackDataset',\n data_root='data/nuscenes/',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ],\n modality=dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False),\n version='v1.0-trainval',\n ann_file='data/nuscenes_anno_pkls/nuscenes_infos_train.pkl',\n pipeline=[\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(\n type='LoadPointsFromFile',\n coord_type='LIDAR',\n load_dim=5,\n use_dim=5,\n file_client_args=dict(backend='disk')),\n dict(type='ResizeCropFlipImage'),\n dict(type='MultiScaleDepthMapGenerator', downsample=[4, 8, 16]),\n dict(type='BBoxRotation'),\n dict(type='PhotoMetricDistortionMultiViewImage'),\n dict(\n type='NormalizeMultiviewImage',\n mean=[123.675, 116.28, 103.53],\n std=[58.395, 57.12, 57.375],\n to_rgb=True),\n dict(\n type='CircleObjectRangeFilter',\n class_dist_thred=[55, 55, 55, 55, 55, 55, 55, 55, 55, 55]),\n dict(\n type='InstanceNameFilter',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian',\n 'traffic_cone'\n ]),\n dict(type='NuScenesSparse4DAdaptor'),\n dict(\n type='Collect',\n keys=[\n 'img', 'timestamp', 'projection_mat', 'image_wh',\n 'gt_depth', 'focal', 'gt_bboxes_3d', 'gt_labels_3d'\n ],\n meta_keys=[\n 'T_global', 'T_global_inv', 'timestamp', 'instance_id'\n ])\n ],\n test_mode=False,\n data_aug_conf=dict(\n resize_lim=(0.4, 0.47),\n final_dim=(256, 704),\n bot_pct_lim=(0.0, 0.0),\n rot_lim=(-5.4, 5.4),\n H=900,\n W=1600,\n rand_flip=True,\n rot3d_range=[-0.3925, 0.3925]),\n with_seq_flag=True,\n sequences_split_num=2,\n keep_consistent_seq_aug=True),\n val=dict(\n type='NuScenes3DDetTrackDataset',\n data_root='data/nuscenes/',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ],\n modality=dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False),\n version='v1.0-trainval',\n ann_file='data/nuscenes_anno_pkls/nuscenes_infos_val.pkl',\n pipeline=[\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(type='ResizeCropFlipImage'),\n dict(\n type='NormalizeMultiviewImage',\n mean=[123.675, 116.28, 103.53],\n std=[58.395, 57.12, 57.375],\n to_rgb=True),\n dict(type='NuScenesSparse4DAdaptor'),\n dict(\n type='Collect',\n keys=['img', 'timestamp', 'projection_mat', 'image_wh'],\n meta_keys=['T_global', 'T_global_inv', 'timestamp'])\n ],\n data_aug_conf=dict(\n resize_lim=(0.4, 0.47),\n final_dim=(256, 704),\n bot_pct_lim=(0.0, 0.0),\n rot_lim=(-5.4, 5.4),\n H=900,\n W=1600,\n rand_flip=True,\n rot3d_range=[-0.3925, 0.3925]),\n test_mode=True,\n tracking=True,\n tracking_threshold=0.2),\n test=dict(\n type='NuScenes3DDetTrackDataset',\n data_root='data/nuscenes/',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ],\n modality=dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False),\n version='v1.0-trainval',\n ann_file='data/nuscenes_anno_pkls/nuscenes_infos_val.pkl',\n pipeline=[\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(type='ResizeCropFlipImage'),\n dict(\n type='NormalizeMultiviewImage',\n mean=[123.675, 116.28, 103.53],\n std=[58.395, 57.12, 57.375],\n to_rgb=True),\n dict(type='NuScenesSparse4DAdaptor'),\n dict(\n type='Collect',\n keys=['img', 'timestamp', 'projection_mat', 'image_wh'],\n meta_keys=['T_global', 'T_global_inv', 'timestamp'])\n ],\n data_aug_conf=dict(\n resize_lim=(0.4, 0.47),\n final_dim=(256, 704),\n bot_pct_lim=(0.0, 0.0),\n rot_lim=(-5.4, 5.4),\n H=900,\n W=1600,\n rand_flip=True,\n rot3d_range=[-0.3925, 0.3925]),\n test_mode=True,\n tracking=True,\n tracking_threshold=0.2))\noptimizer = dict(\n type='Miopen_AdamW',\n lr=0.0006,\n weight_decay=0.001,\n paramwise_cfg=dict(custom_keys=dict(img_backbone=dict(lr_mult=0.5))))\noptimizer_config = dict(grad_clip=dict(max_norm=25, norm_type=2))\nlr_config = dict(\n policy='CosineAnnealing',\n warmup='linear',\n warmup_iters=500,\n warmup_ratio=0.3333333333333333,\n min_lr_ratio=0.001)\nrunner = dict(type='IterBasedRunner', max_iters=17500)\nvis_pipeline = [\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(type='Collect', keys=['img'], meta_keys=['timestamp', 'lidar2img'])\n]\nevaluation = dict(\n interval=3500,\n pipeline=[\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(\n type='Collect', keys=['img'], meta_keys=['timestamp', 'lidar2img'])\n ])\ngpu_ids = range(0, 8)\n", "seed": 0, "exp_name": "sparse4dv3_temporal_r50_1x8_bs6_256x704.py"} diff --git a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20251113_151710.log b/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20251113_151710.log deleted file mode 100644 index 255831af003321e14242b320e8559f262d3cddb0..0000000000000000000000000000000000000000 --- a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20251113_151710.log +++ /dev/null @@ -1,3220 +0,0 @@ -2025-11-13 15:17:10,982 - mmdet - INFO - Environment info: ------------------------------------------------------------- -sys.platform: linux -Python: 3.10.12 (main, May 27 2025, 17:12:29) [GCC 11.4.0] -CUDA available: True -GPU 0,1,2,3,4,5,6,7: BW200, UBB BW1000 -CUDA_HOME: /opt/dtk -NVCC: Not Available -GCC: x86_64-linux-gnu-gcc (Ubuntu 11.4.0-1ubuntu1~22.04) 11.4.0 -PyTorch: 2.4.1 -PyTorch compiling details: PyTorch built with: - - GCC 10.3 - - C++ Version: 201703 - - Intel(R) Math Kernel Library Version 2020.0.4 Product Build 20200917 for Intel(R) 64 architecture applications - - OpenMP 201511 (a.k.a. OpenMP 4.5) - - LAPACK is enabled (usually provided by MKL) - - NNPACK is enabled - - CPU capability usage: AVX512 - - HIP Runtime 6.3.25211 - - MIOpen 2.17.0 - - Magma 2.8.0 - - Build settings: BLAS_INFO=mkl, BUILD_TYPE=Release, CXX_COMPILER=/opt/rh/gcc-toolset-10/root/usr/bin/c++, CXX_FLAGS= -D_GLIBCXX_USE_CXX11_ABI=1 -fvisibility-inlines-hidden -DUSE_PTHREADPOOL -DNDEBUG -DUSE_KINETO -DLIBKINETO_NOCUPTI -DUSE_FBGEMM -DUSE_PYTORCH_QNNPACK -DUSE_XNNPACK -DSYMBOLICATE_MOBILE_DEBUG_HANDLE -O2 -fPIC -Wall -Wextra -Werror=return-type -Werror=non-virtual-dtor -Werror=bool-operation -Wnarrowing -Wno-missing-field-initializers -Wno-type-limits -Wno-array-bounds -Wno-unknown-pragmas -Wno-unused-parameter -Wno-unused-function -Wno-unused-result -Wno-strict-overflow -Wno-strict-aliasing -Wno-stringop-overflow -Wsuggest-override -Wno-psabi -Wno-error=pedantic -Wno-error=old-style-cast -Wno-missing-braces -fdiagnostics-color=always -faligned-new -Wno-unused-but-set-variable -Wno-maybe-uninitialized -fno-math-errno -fno-trapping-math -Werror=format -Wno-stringop-overflow, FORCE_FALLBACK_CUDA_MPI=1, LAPACK_INFO=mkl, PERF_WITH_AVX=1, PERF_WITH_AVX2=1, PERF_WITH_AVX512=1, TORCH_VERSION=2.4.1, USE_CUDA=0, USE_CUDNN=OFF, USE_CUSPARSELT=OFF, USE_EXCEPTION_PTR=1, USE_GFLAGS=1, USE_GLOG=1, USE_GLOO=1, USE_MKL=ON, USE_MKLDNN=0, USE_MPI=1, USE_NCCL=1, USE_NNPACK=ON, USE_OPENMP=1, USE_ROCM=ON, USE_ROCM_KERNEL_ASSERT=OFF, - -TorchVision: 0.19.1 -OpenCV: 4.11.0 -MMCV: 1.6.1 -MMCV Compiler: GCC 11.4 -MMCV CUDA Compiler: rocm not available -MMDetection: 2.26.0+c41df4b ------------------------------------------------------------- - -2025-11-13 15:17:11,922 - mmdet - INFO - Distributed training: True -2025-11-13 15:17:12,643 - mmdet - INFO - Config: -plugin = True -plugin_dir = 'projects/mmdet3d_plugin/' -dist_params = dict(backend='nccl') -log_level = 'INFO' -work_dir = './work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704' -total_batch_size = 160 -num_gpus = 8 -batch_size = 20 -num_iters_per_epoch = 175 -num_epochs = 100 -checkpoint_epoch_interval = 20 -checkpoint_config = dict(interval=3500) -log_config = dict( - interval=1, - hooks=[ - dict(type='TextLoggerHook', by_epoch=False), - dict(type='TensorboardLoggerHook') - ]) -load_from = None -resume_from = None -workflow = [('train', 1)] -fp16 = dict(loss_scale=32.0) -input_shape = (704, 256) -tracking_test = True -tracking_threshold = 0.2 -class_names = [ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier', - 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' -] -num_classes = 10 -embed_dims = 256 -num_groups = 8 -num_decoder = 6 -num_single_frame_decoder = 1 -use_deformable_func = True -strides = [4, 8, 16, 32] -num_levels = 4 -num_depth_layers = 3 -drop_out = 0.1 -temporal = True -decouple_attn = True -with_quality_estimation = True -model = dict( - type='Sparse4D', - use_grid_mask=True, - use_deformable_func=True, - img_backbone=dict( - type='ResNet', - depth=50, - num_stages=4, - frozen_stages=-1, - norm_eval=False, - style='pytorch', - with_cp=True, - out_indices=(0, 1, 2, 3), - norm_cfg=dict(type='BN', requires_grad=True), - pretrained='ckpt/resnet50-19c8e357.pth'), - img_neck=dict( - type='FPN', - num_outs=4, - start_level=0, - out_channels=256, - add_extra_convs='on_output', - relu_before_extra_convs=True, - in_channels=[256, 512, 1024, 2048]), - depth_branch=dict( - type='DenseDepthNet', - embed_dims=256, - num_depth_layers=3, - loss_weight=0.2), - head=dict( - type='Sparse4DHead', - cls_threshold_to_reg=0.05, - decouple_attn=True, - instance_bank=dict( - type='InstanceBank', - num_anchor=900, - embed_dims=256, - anchor='nuscenes_kmeans900.npy', - anchor_handler=dict(type='SparseBox3DKeyPointsGenerator'), - num_temp_instances=600, - confidence_decay=0.6, - feat_grad=False), - anchor_encoder=dict( - type='SparseBox3DEncoder', - vel_dims=3, - embed_dims=[128, 32, 32, 64], - mode='cat', - output_fc=False, - in_loops=1, - out_loops=4), - num_single_frame_decoder=1, - operation_order=[ - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine' - ], - temp_graph_model=dict( - type='MultiheadAttention', - embed_dims=512, - num_heads=8, - batch_first=True, - dropout=0.1), - graph_model=dict( - type='MultiheadAttention', - embed_dims=512, - num_heads=8, - batch_first=True, - dropout=0.1), - norm_layer=dict(type='LN', normalized_shape=256), - ffn=dict( - type='AsymmetricFFN', - in_channels=512, - pre_norm=dict(type='LN'), - embed_dims=256, - feedforward_channels=1024, - num_fcs=2, - ffn_drop=0.1, - act_cfg=dict(type='ReLU', inplace=True)), - deformable_model=dict( - type='DeformableFeatureAggregation', - embed_dims=256, - num_groups=8, - num_levels=4, - num_cams=6, - attn_drop=0.15, - use_deformable_func=True, - use_camera_embed=True, - residual_mode='cat', - kps_generator=dict( - type='SparseBox3DKeyPointsGenerator', - num_learnable_pts=6, - fix_scale=[[0, 0, 0], [0.45, 0, 0], [-0.45, 0, 0], - [0, 0.45, 0], [0, -0.45, 0], [0, 0, 0.45], - [0, 0, -0.45]])), - refine_layer=dict( - type='SparseBox3DRefinementModule', - embed_dims=256, - num_cls=10, - refine_yaw=True, - with_quality_estimation=True), - sampler=dict( - type='SparseBox3DTarget', - num_dn_groups=5, - num_temp_dn_groups=3, - dn_noise_scale=[2.0, 2.0, 2.0, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5], - max_dn_gt=32, - add_neg_dn=True, - cls_weight=2.0, - box_weight=0.25, - reg_weights=[2.0, 2.0, 2.0, 0.5, 0.5, 0.5, 0.0, 0.0, 0.0, 0.0], - cls_wise_reg_weights=dict( - {9: [2.0, 2.0, 2.0, 1.0, 1.0, 1.0, 0.0, 0.0, 1.0, 1.0]})), - loss_cls=dict( - type='FocalLoss', - use_sigmoid=True, - gamma=2.0, - alpha=0.25, - loss_weight=2.0), - loss_reg=dict( - type='SparseBox3DLoss', - loss_box=dict(type='L1Loss', loss_weight=0.25), - loss_centerness=dict(type='CrossEntropyLoss', use_sigmoid=True), - loss_yawness=dict(type='GaussianFocalLoss'), - cls_allow_reverse=[5]), - decoder=dict(type='SparseBox3DDecoder'), - reg_weights=[2.0, 2.0, 2.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0])) -dataset_type = 'NuScenes3DDetTrackDataset' -data_root = 'data/nuscenes/' -anno_root = 'data/nuscenes_anno_pkls/' -file_client_args = dict(backend='disk') -img_norm_cfg = dict( - mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375], to_rgb=True) -train_pipeline = [ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict( - type='LoadPointsFromFile', - coord_type='LIDAR', - load_dim=5, - use_dim=5, - file_client_args=dict(backend='disk')), - dict(type='ResizeCropFlipImage'), - dict(type='MultiScaleDepthMapGenerator', downsample=[4, 8, 16]), - dict(type='BBoxRotation'), - dict(type='PhotoMetricDistortionMultiViewImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict( - type='CircleObjectRangeFilter', - class_dist_thred=[55, 55, 55, 55, 55, 55, 55, 55, 55, 55]), - dict( - type='InstanceNameFilter', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ]), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=[ - 'img', 'timestamp', 'projection_mat', 'image_wh', 'gt_depth', - 'focal', 'gt_bboxes_3d', 'gt_labels_3d' - ], - meta_keys=['T_global', 'T_global_inv', 'timestamp', 'instance_id']) -] -test_pipeline = [ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='ResizeCropFlipImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=['img', 'timestamp', 'projection_mat', 'image_wh'], - meta_keys=['T_global', 'T_global_inv', 'timestamp']) -] -input_modality = dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False) -data_basic_config = dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier', - 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval') -data_aug_conf = dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]) -data = dict( - samples_per_gpu=20, - workers_per_gpu=20, - train=dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval', - ann_file='data/nuscenes_anno_pkls/nuscenes_infos_train.pkl', - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict( - type='LoadPointsFromFile', - coord_type='LIDAR', - load_dim=5, - use_dim=5, - file_client_args=dict(backend='disk')), - dict(type='ResizeCropFlipImage'), - dict(type='MultiScaleDepthMapGenerator', downsample=[4, 8, 16]), - dict(type='BBoxRotation'), - dict(type='PhotoMetricDistortionMultiViewImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict( - type='CircleObjectRangeFilter', - class_dist_thred=[55, 55, 55, 55, 55, 55, 55, 55, 55, 55]), - dict( - type='InstanceNameFilter', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', - 'traffic_cone' - ]), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=[ - 'img', 'timestamp', 'projection_mat', 'image_wh', - 'gt_depth', 'focal', 'gt_bboxes_3d', 'gt_labels_3d' - ], - meta_keys=[ - 'T_global', 'T_global_inv', 'timestamp', 'instance_id' - ]) - ], - test_mode=False, - data_aug_conf=dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]), - with_seq_flag=True, - sequences_split_num=2, - keep_consistent_seq_aug=True), - val=dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval', - ann_file='data/nuscenes_anno_pkls/nuscenes_infos_val.pkl', - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='ResizeCropFlipImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=['img', 'timestamp', 'projection_mat', 'image_wh'], - meta_keys=['T_global', 'T_global_inv', 'timestamp']) - ], - data_aug_conf=dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]), - test_mode=True, - tracking=True, - tracking_threshold=0.2), - test=dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval', - ann_file='data/nuscenes_anno_pkls/nuscenes_infos_val.pkl', - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='ResizeCropFlipImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=['img', 'timestamp', 'projection_mat', 'image_wh'], - meta_keys=['T_global', 'T_global_inv', 'timestamp']) - ], - data_aug_conf=dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]), - test_mode=True, - tracking=True, - tracking_threshold=0.2)) -optimizer = dict( - type='Miopen_AdamW', - lr=0.0006, - weight_decay=0.001, - paramwise_cfg=dict(custom_keys=dict(img_backbone=dict(lr_mult=0.5)))) -optimizer_config = dict(grad_clip=dict(max_norm=25, norm_type=2)) -lr_config = dict( - policy='CosineAnnealing', - warmup='linear', - warmup_iters=500, - warmup_ratio=0.3333333333333333, - min_lr_ratio=0.001) -runner = dict(type='IterBasedRunner', max_iters=17500) -vis_pipeline = [ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='Collect', keys=['img'], meta_keys=['timestamp', 'lidar2img']) -] -evaluation = dict( - interval=3500, - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict( - type='Collect', keys=['img'], meta_keys=['timestamp', 'lidar2img']) - ]) -gpu_ids = range(0, 8) - -2025-11-13 15:17:12,643 - mmdet - INFO - Set random seed to 0, deterministic: False -2025-11-13 15:17:12,946 - mmdet - INFO - initialize ResNet with init_cfg {'type': 'Pretrained', 'checkpoint': 'ckpt/resnet50-19c8e357.pth'} -2025-11-13 15:17:13,218 - mmdet - INFO - initialize FPN with init_cfg {'type': 'Xavier', 'layer': 'Conv2d', 'distribution': 'uniform'} -Name of parameter - Initialization information - -img_backbone.conv1.weight - torch.Size([64, 3, 7, 7]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.bn1.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.bn1.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.conv1.weight - torch.Size([64, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn1.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn1.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.conv2.weight - torch.Size([64, 64, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn2.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn2.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.conv3.weight - torch.Size([256, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn3.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn3.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.downsample.0.weight - torch.Size([256, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.downsample.1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.downsample.1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.conv1.weight - torch.Size([64, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn1.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn1.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.conv2.weight - torch.Size([64, 64, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn2.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn2.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.conv3.weight - torch.Size([256, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn3.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn3.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.conv1.weight - torch.Size([64, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn1.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn1.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.conv2.weight - torch.Size([64, 64, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn2.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn2.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.conv3.weight - torch.Size([256, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn3.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn3.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.conv1.weight - torch.Size([128, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn1.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn1.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.conv2.weight - torch.Size([128, 128, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn2.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn2.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.conv3.weight - torch.Size([512, 128, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn3.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn3.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.downsample.0.weight - torch.Size([512, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.downsample.1.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.downsample.1.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.conv1.weight - torch.Size([128, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn1.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn1.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.conv2.weight - torch.Size([128, 128, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn2.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn2.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.conv3.weight - torch.Size([512, 128, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn3.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn3.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.conv1.weight - torch.Size([128, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn1.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn1.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.conv2.weight - torch.Size([128, 128, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn2.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn2.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.conv3.weight - torch.Size([512, 128, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn3.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn3.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.conv1.weight - torch.Size([128, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn1.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn1.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.conv2.weight - torch.Size([128, 128, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn2.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn2.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.conv3.weight - torch.Size([512, 128, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn3.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn3.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.conv1.weight - torch.Size([256, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.downsample.0.weight - torch.Size([1024, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.downsample.1.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.downsample.1.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.conv1.weight - torch.Size([512, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn1.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn1.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.conv2.weight - torch.Size([512, 512, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn2.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn2.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.conv3.weight - torch.Size([2048, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn3.weight - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn3.bias - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.downsample.0.weight - torch.Size([2048, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.downsample.1.weight - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.downsample.1.bias - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.conv1.weight - torch.Size([512, 2048, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn1.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn1.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.conv2.weight - torch.Size([512, 512, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn2.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn2.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.conv3.weight - torch.Size([2048, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn3.weight - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn3.bias - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.conv1.weight - torch.Size([512, 2048, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn1.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn1.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.conv2.weight - torch.Size([512, 512, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn2.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn2.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.conv3.weight - torch.Size([2048, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn3.weight - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn3.bias - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_neck.lateral_convs.0.conv.weight - torch.Size([256, 256, 1, 1]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.lateral_convs.0.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.lateral_convs.1.conv.weight - torch.Size([256, 512, 1, 1]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.lateral_convs.1.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.lateral_convs.2.conv.weight - torch.Size([256, 1024, 1, 1]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.lateral_convs.2.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.lateral_convs.3.conv.weight - torch.Size([256, 2048, 1, 1]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.lateral_convs.3.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.fpn_convs.0.conv.weight - torch.Size([256, 256, 3, 3]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.fpn_convs.0.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.fpn_convs.1.conv.weight - torch.Size([256, 256, 3, 3]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.fpn_convs.1.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.fpn_convs.2.conv.weight - torch.Size([256, 256, 3, 3]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.fpn_convs.2.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.fpn_convs.3.conv.weight - torch.Size([256, 256, 3, 3]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.fpn_convs.3.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.instance_bank.anchor - torch.Size([900, 11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.instance_bank.instance_feature - torch.Size([900, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.instance_bank.anchor_handler.fix_scale - torch.Size([1, 3]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.0.weight - torch.Size([128, 3]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.0.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.2.weight - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.2.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.3.weight - torch.Size([128, 128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.3.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.5.weight - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.5.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.6.weight - torch.Size([128, 128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.6.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.8.weight - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.8.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.9.weight - torch.Size([128, 128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.9.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.11.weight - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.11.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.0.weight - torch.Size([32, 3]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.0.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.2.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.2.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.3.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.3.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.5.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.5.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.6.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.6.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.8.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.8.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.9.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.9.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.11.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.11.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.0.weight - torch.Size([32, 2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.0.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.2.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.2.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.3.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.3.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.5.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.5.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.6.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.6.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.8.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.8.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.9.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.9.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.11.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.11.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.0.weight - torch.Size([64, 3]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.0.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.2.weight - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.2.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.3.weight - torch.Size([64, 64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.3.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.5.weight - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.5.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.6.weight - torch.Size([64, 64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.6.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.8.weight - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.8.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.9.weight - torch.Size([64, 64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.9.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.11.weight - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.11.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.1.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.1.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.1.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.1.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.1.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.1.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.1.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.1.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.3.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.4.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.4.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.4.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.4.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.5.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.5.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.5.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.5.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.6.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.6.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.8.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.8.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.8.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.8.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.8.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.8.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.8.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.8.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.10.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.11.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.11.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.11.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.11.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.12.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.12.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.12.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.12.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.13.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.13.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.15.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.15.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.15.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.15.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.15.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.15.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.15.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.15.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.16.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.16.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.17.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.18.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.18.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.18.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.18.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.19.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.19.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.19.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.19.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.20.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.20.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.22.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.22.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.22.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.22.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.22.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.22.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.22.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.22.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.23.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.23.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.24.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.25.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.25.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.25.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.25.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.26.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.26.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.26.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.26.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.27.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.27.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.29.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.29.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.29.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.29.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.29.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.29.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.29.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.29.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.30.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.30.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.31.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.32.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.32.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.32.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.32.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.33.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.33.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.33.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.33.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.34.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.34.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.36.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.36.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.36.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.36.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.36.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.36.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.36.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.36.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.37.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.37.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.38.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.fc_before.weight - torch.Size([512, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.fc_after.weight - torch.Size([256, 512]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.0.weight - torch.Size([1, 256, 1, 1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.0.bias - torch.Size([1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.1.weight - torch.Size([1, 256, 1, 1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.1.bias - torch.Size([1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.2.weight - torch.Size([1, 256, 1, 1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.2.bias - torch.Size([1]): -The value is the same before and after calling `init_weights` of Sparse4D -2025-11-13 15:17:13,309 - mmdet - INFO - Model: -Sparse4D( - (img_backbone): ResNet( - (conv1): Conv2d(3, 64, kernel_size=(7, 7), stride=(2, 2), padding=(3, 3), bias=False) - (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (maxpool): MaxPool2d(kernel_size=3, stride=2, padding=1, dilation=1, ceil_mode=False) - (layer1): ResLayer( - (0): Bottleneck( - (conv1): Conv2d(64, 64, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (downsample): Sequential( - (0): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - ) - ) - (1): Bottleneck( - (conv1): Conv2d(256, 64, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (2): Bottleneck( - (conv1): Conv2d(256, 64, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - ) - (layer2): ResLayer( - (0): Bottleneck( - (conv1): Conv2d(256, 128, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (downsample): Sequential( - (0): Conv2d(256, 512, kernel_size=(1, 1), stride=(2, 2), bias=False) - (1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - ) - ) - (1): Bottleneck( - (conv1): Conv2d(512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (2): Bottleneck( - (conv1): Conv2d(512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (3): Bottleneck( - (conv1): Conv2d(512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - ) - (layer3): ResLayer( - (0): Bottleneck( - (conv1): Conv2d(512, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (downsample): Sequential( - (0): Conv2d(512, 1024, kernel_size=(1, 1), stride=(2, 2), bias=False) - (1): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - ) - ) - (1): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (2): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (3): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (4): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (5): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - ) - (layer4): ResLayer( - (0): Bottleneck( - (conv1): Conv2d(1024, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(512, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (downsample): Sequential( - (0): Conv2d(1024, 2048, kernel_size=(1, 1), stride=(2, 2), bias=False) - (1): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - ) - ) - (1): Bottleneck( - (conv1): Conv2d(2048, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(512, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (2): Bottleneck( - (conv1): Conv2d(2048, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(512, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - ) - ) - init_cfg={'type': 'Pretrained', 'checkpoint': 'ckpt/resnet50-19c8e357.pth'} - (img_neck): FPN( - (lateral_convs): ModuleList( - (0): ConvModule( - (conv): Conv2d(256, 256, kernel_size=(1, 1), stride=(1, 1)) - ) - (1): ConvModule( - (conv): Conv2d(512, 256, kernel_size=(1, 1), stride=(1, 1)) - ) - (2): ConvModule( - (conv): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1)) - ) - (3): ConvModule( - (conv): Conv2d(2048, 256, kernel_size=(1, 1), stride=(1, 1)) - ) - ) - (fpn_convs): ModuleList( - (0-3): 4 x ConvModule( - (conv): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1)) - ) - ) - ) - init_cfg={'type': 'Xavier', 'layer': 'Conv2d', 'distribution': 'uniform'} - (head): Sparse4DHead( - (instance_bank): InstanceBank( - (anchor_handler): SparseBox3DKeyPointsGenerator() - ) - (anchor_encoder): SparseBox3DEncoder( - (pos_fc): Sequential( - (0): Linear(in_features=3, out_features=128, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((128,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=128, out_features=128, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((128,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=128, out_features=128, bias=True) - (7): ReLU(inplace=True) - (8): LayerNorm((128,), eps=1e-05, elementwise_affine=True) - (9): Linear(in_features=128, out_features=128, bias=True) - (10): ReLU(inplace=True) - (11): LayerNorm((128,), eps=1e-05, elementwise_affine=True) - ) - (size_fc): Sequential( - (0): Linear(in_features=3, out_features=32, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=32, out_features=32, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=32, out_features=32, bias=True) - (7): ReLU(inplace=True) - (8): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (9): Linear(in_features=32, out_features=32, bias=True) - (10): ReLU(inplace=True) - (11): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - ) - (yaw_fc): Sequential( - (0): Linear(in_features=2, out_features=32, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=32, out_features=32, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=32, out_features=32, bias=True) - (7): ReLU(inplace=True) - (8): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (9): Linear(in_features=32, out_features=32, bias=True) - (10): ReLU(inplace=True) - (11): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - ) - (vel_fc): Sequential( - (0): Linear(in_features=3, out_features=64, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((64,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=64, out_features=64, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((64,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=64, out_features=64, bias=True) - (7): ReLU(inplace=True) - (8): LayerNorm((64,), eps=1e-05, elementwise_affine=True) - (9): Linear(in_features=64, out_features=64, bias=True) - (10): ReLU(inplace=True) - (11): LayerNorm((64,), eps=1e-05, elementwise_affine=True) - ) - ) - (loss_cls): FocalLoss() - (loss_reg): SparseBox3DLoss( - (loss_box): L1Loss() - (loss_cns): CrossEntropyLoss(avg_non_ignore=False) - (loss_yns): GaussianFocalLoss() - ) - (layers): ModuleList( - (0): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (1): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (4-5): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (6): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (7): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (8): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (11-12): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (13): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (14): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (15): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (16): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (17): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (18-19): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (20): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (21): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (22): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (23): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (24): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (25-26): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (27): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (28): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (29): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (30): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (31): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (32-33): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (34): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (35): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (36): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (37): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (38): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - ) - (fc_before): Linear(in_features=256, out_features=512, bias=False) - (fc_after): Linear(in_features=512, out_features=256, bias=False) - ) - (depth_branch): DenseDepthNet( - (depth_layers): ModuleList( - (0-2): 3 x Conv2d(256, 1, kernel_size=(1, 1), stride=(1, 1)) - ) - ) - (grid_mask): GridMask() -) -2025-11-13 15:17:26,003 - mmdet - INFO - Start running, host: root@VM-120-96-tencentos, work_dir: /cfs/auto/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704 -2025-11-13 15:17:26,003 - mmdet - INFO - Hooks will be executed in the following order: -before_run: -(VERY_HIGH ) CosineAnnealingLrUpdaterHook -(ABOVE_NORMAL) Fp16OptimizerHook -(NORMAL ) CheckpointHook -(NORMAL ) CustomDistEvalHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -before_train_epoch: -(VERY_HIGH ) CosineAnnealingLrUpdaterHook -(NORMAL ) CustomDistEvalHook -(LOW ) IterTimerHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -before_train_iter: -(VERY_HIGH ) CosineAnnealingLrUpdaterHook -(NORMAL ) CustomDistEvalHook -(LOW ) IterTimerHook - -------------------- -after_train_iter: -(ABOVE_NORMAL) Fp16OptimizerHook -(NORMAL ) CheckpointHook -(NORMAL ) CustomDistEvalHook -(LOW ) IterTimerHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -after_train_epoch: -(NORMAL ) CheckpointHook -(NORMAL ) CustomDistEvalHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -before_val_epoch: -(LOW ) IterTimerHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -before_val_iter: -(LOW ) IterTimerHook - -------------------- -after_val_iter: -(LOW ) IterTimerHook - -------------------- -after_val_epoch: -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -after_run: -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -2025-11-13 15:17:26,003 - mmdet - INFO - workflow: [('train', 1)], max: 17500 iters -2025-11-13 15:17:26,005 - mmdet - INFO - Checkpoints will be saved to /cfs/auto/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704 by HardDiskBackend. diff --git a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20251113_151710.log.json b/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20251113_151710.log.json deleted file mode 100644 index 1fda3e213098b4a1f6239f0e36f9459d0a88f9c1..0000000000000000000000000000000000000000 --- a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20251113_151710.log.json +++ /dev/null @@ -1 +0,0 @@ -{"env_info": "sys.platform: linux\nPython: 3.10.12 (main, May 27 2025, 17:12:29) [GCC 11.4.0]\nCUDA available: True\nGPU 0,1,2,3,4,5,6,7: BW200, UBB BW1000\nCUDA_HOME: /opt/dtk\nNVCC: Not Available\nGCC: x86_64-linux-gnu-gcc (Ubuntu 11.4.0-1ubuntu1~22.04) 11.4.0\nPyTorch: 2.4.1\nPyTorch compiling details: PyTorch built with:\n - GCC 10.3\n - C++ Version: 201703\n - Intel(R) Math Kernel Library Version 2020.0.4 Product Build 20200917 for Intel(R) 64 architecture applications\n - OpenMP 201511 (a.k.a. OpenMP 4.5)\n - LAPACK is enabled (usually provided by MKL)\n - NNPACK is enabled\n - CPU capability usage: AVX512\n - HIP Runtime 6.3.25211\n - MIOpen 2.17.0\n - Magma 2.8.0\n - Build settings: BLAS_INFO=mkl, BUILD_TYPE=Release, CXX_COMPILER=/opt/rh/gcc-toolset-10/root/usr/bin/c++, CXX_FLAGS= -D_GLIBCXX_USE_CXX11_ABI=1 -fvisibility-inlines-hidden -DUSE_PTHREADPOOL -DNDEBUG -DUSE_KINETO -DLIBKINETO_NOCUPTI -DUSE_FBGEMM -DUSE_PYTORCH_QNNPACK -DUSE_XNNPACK -DSYMBOLICATE_MOBILE_DEBUG_HANDLE -O2 -fPIC -Wall -Wextra -Werror=return-type -Werror=non-virtual-dtor -Werror=bool-operation -Wnarrowing -Wno-missing-field-initializers -Wno-type-limits -Wno-array-bounds -Wno-unknown-pragmas -Wno-unused-parameter -Wno-unused-function -Wno-unused-result -Wno-strict-overflow -Wno-strict-aliasing -Wno-stringop-overflow -Wsuggest-override -Wno-psabi -Wno-error=pedantic -Wno-error=old-style-cast -Wno-missing-braces -fdiagnostics-color=always -faligned-new -Wno-unused-but-set-variable -Wno-maybe-uninitialized -fno-math-errno -fno-trapping-math -Werror=format -Wno-stringop-overflow, FORCE_FALLBACK_CUDA_MPI=1, LAPACK_INFO=mkl, PERF_WITH_AVX=1, PERF_WITH_AVX2=1, PERF_WITH_AVX512=1, TORCH_VERSION=2.4.1, USE_CUDA=0, USE_CUDNN=OFF, USE_CUSPARSELT=OFF, USE_EXCEPTION_PTR=1, USE_GFLAGS=1, USE_GLOG=1, USE_GLOO=1, USE_MKL=ON, USE_MKLDNN=0, USE_MPI=1, USE_NCCL=1, USE_NNPACK=ON, USE_OPENMP=1, USE_ROCM=ON, USE_ROCM_KERNEL_ASSERT=OFF, \n\nTorchVision: 0.19.1\nOpenCV: 4.11.0\nMMCV: 1.6.1\nMMCV Compiler: GCC 11.4\nMMCV CUDA Compiler: rocm not available\nMMDetection: 2.26.0+c41df4b", "config": "plugin = True\nplugin_dir = 'projects/mmdet3d_plugin/'\ndist_params = dict(backend='nccl')\nlog_level = 'INFO'\nwork_dir = './work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704'\ntotal_batch_size = 160\nnum_gpus = 8\nbatch_size = 20\nnum_iters_per_epoch = 175\nnum_epochs = 100\ncheckpoint_epoch_interval = 20\ncheckpoint_config = dict(interval=3500)\nlog_config = dict(\n interval=1,\n hooks=[\n dict(type='TextLoggerHook', by_epoch=False),\n dict(type='TensorboardLoggerHook')\n ])\nload_from = None\nresume_from = None\nworkflow = [('train', 1)]\nfp16 = dict(loss_scale=32.0)\ninput_shape = (704, 256)\ntracking_test = True\ntracking_threshold = 0.2\nclass_names = [\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier',\n 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n]\nnum_classes = 10\nembed_dims = 256\nnum_groups = 8\nnum_decoder = 6\nnum_single_frame_decoder = 1\nuse_deformable_func = True\nstrides = [4, 8, 16, 32]\nnum_levels = 4\nnum_depth_layers = 3\ndrop_out = 0.1\ntemporal = True\ndecouple_attn = True\nwith_quality_estimation = True\nmodel = dict(\n type='Sparse4D',\n use_grid_mask=True,\n use_deformable_func=True,\n img_backbone=dict(\n type='ResNet',\n depth=50,\n num_stages=4,\n frozen_stages=-1,\n norm_eval=False,\n style='pytorch',\n with_cp=True,\n out_indices=(0, 1, 2, 3),\n norm_cfg=dict(type='BN', requires_grad=True),\n pretrained='ckpt/resnet50-19c8e357.pth'),\n img_neck=dict(\n type='FPN',\n num_outs=4,\n start_level=0,\n out_channels=256,\n add_extra_convs='on_output',\n relu_before_extra_convs=True,\n in_channels=[256, 512, 1024, 2048]),\n depth_branch=dict(\n type='DenseDepthNet',\n embed_dims=256,\n num_depth_layers=3,\n loss_weight=0.2),\n head=dict(\n type='Sparse4DHead',\n cls_threshold_to_reg=0.05,\n decouple_attn=True,\n instance_bank=dict(\n type='InstanceBank',\n num_anchor=900,\n embed_dims=256,\n anchor='nuscenes_kmeans900.npy',\n anchor_handler=dict(type='SparseBox3DKeyPointsGenerator'),\n num_temp_instances=600,\n confidence_decay=0.6,\n feat_grad=False),\n anchor_encoder=dict(\n type='SparseBox3DEncoder',\n vel_dims=3,\n embed_dims=[128, 32, 32, 64],\n mode='cat',\n output_fc=False,\n in_loops=1,\n out_loops=4),\n num_single_frame_decoder=1,\n operation_order=[\n 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm',\n 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm',\n 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm',\n 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm',\n 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm',\n 'deformable', 'ffn', 'norm', 'refine'\n ],\n temp_graph_model=dict(\n type='MultiheadAttention',\n embed_dims=512,\n num_heads=8,\n batch_first=True,\n dropout=0.1),\n graph_model=dict(\n type='MultiheadAttention',\n embed_dims=512,\n num_heads=8,\n batch_first=True,\n dropout=0.1),\n norm_layer=dict(type='LN', normalized_shape=256),\n ffn=dict(\n type='AsymmetricFFN',\n in_channels=512,\n pre_norm=dict(type='LN'),\n embed_dims=256,\n feedforward_channels=1024,\n num_fcs=2,\n ffn_drop=0.1,\n act_cfg=dict(type='ReLU', inplace=True)),\n deformable_model=dict(\n type='DeformableFeatureAggregation',\n embed_dims=256,\n num_groups=8,\n num_levels=4,\n num_cams=6,\n attn_drop=0.15,\n use_deformable_func=True,\n use_camera_embed=True,\n residual_mode='cat',\n kps_generator=dict(\n type='SparseBox3DKeyPointsGenerator',\n num_learnable_pts=6,\n fix_scale=[[0, 0, 0], [0.45, 0, 0], [-0.45, 0, 0],\n [0, 0.45, 0], [0, -0.45, 0], [0, 0, 0.45],\n [0, 0, -0.45]])),\n refine_layer=dict(\n type='SparseBox3DRefinementModule',\n embed_dims=256,\n num_cls=10,\n refine_yaw=True,\n with_quality_estimation=True),\n sampler=dict(\n type='SparseBox3DTarget',\n num_dn_groups=5,\n num_temp_dn_groups=3,\n dn_noise_scale=[2.0, 2.0, 2.0, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5],\n max_dn_gt=32,\n add_neg_dn=True,\n cls_weight=2.0,\n box_weight=0.25,\n reg_weights=[2.0, 2.0, 2.0, 0.5, 0.5, 0.5, 0.0, 0.0, 0.0, 0.0],\n cls_wise_reg_weights=dict(\n {9: [2.0, 2.0, 2.0, 1.0, 1.0, 1.0, 0.0, 0.0, 1.0, 1.0]})),\n loss_cls=dict(\n type='FocalLoss',\n use_sigmoid=True,\n gamma=2.0,\n alpha=0.25,\n loss_weight=2.0),\n loss_reg=dict(\n type='SparseBox3DLoss',\n loss_box=dict(type='L1Loss', loss_weight=0.25),\n loss_centerness=dict(type='CrossEntropyLoss', use_sigmoid=True),\n loss_yawness=dict(type='GaussianFocalLoss'),\n cls_allow_reverse=[5]),\n decoder=dict(type='SparseBox3DDecoder'),\n reg_weights=[2.0, 2.0, 2.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0]))\ndataset_type = 'NuScenes3DDetTrackDataset'\ndata_root = 'data/nuscenes/'\nanno_root = 'data/nuscenes_anno_pkls/'\nfile_client_args = dict(backend='disk')\nimg_norm_cfg = dict(\n mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375], to_rgb=True)\ntrain_pipeline = [\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(\n type='LoadPointsFromFile',\n coord_type='LIDAR',\n load_dim=5,\n use_dim=5,\n file_client_args=dict(backend='disk')),\n dict(type='ResizeCropFlipImage'),\n dict(type='MultiScaleDepthMapGenerator', downsample=[4, 8, 16]),\n dict(type='BBoxRotation'),\n dict(type='PhotoMetricDistortionMultiViewImage'),\n dict(\n type='NormalizeMultiviewImage',\n mean=[123.675, 116.28, 103.53],\n std=[58.395, 57.12, 57.375],\n to_rgb=True),\n dict(\n type='CircleObjectRangeFilter',\n class_dist_thred=[55, 55, 55, 55, 55, 55, 55, 55, 55, 55]),\n dict(\n type='InstanceNameFilter',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ]),\n dict(type='NuScenesSparse4DAdaptor'),\n dict(\n type='Collect',\n keys=[\n 'img', 'timestamp', 'projection_mat', 'image_wh', 'gt_depth',\n 'focal', 'gt_bboxes_3d', 'gt_labels_3d'\n ],\n meta_keys=['T_global', 'T_global_inv', 'timestamp', 'instance_id'])\n]\ntest_pipeline = [\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(type='ResizeCropFlipImage'),\n dict(\n type='NormalizeMultiviewImage',\n mean=[123.675, 116.28, 103.53],\n std=[58.395, 57.12, 57.375],\n to_rgb=True),\n dict(type='NuScenesSparse4DAdaptor'),\n dict(\n type='Collect',\n keys=['img', 'timestamp', 'projection_mat', 'image_wh'],\n meta_keys=['T_global', 'T_global_inv', 'timestamp'])\n]\ninput_modality = dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False)\ndata_basic_config = dict(\n type='NuScenes3DDetTrackDataset',\n data_root='data/nuscenes/',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier',\n 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ],\n modality=dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False),\n version='v1.0-trainval')\ndata_aug_conf = dict(\n resize_lim=(0.4, 0.47),\n final_dim=(256, 704),\n bot_pct_lim=(0.0, 0.0),\n rot_lim=(-5.4, 5.4),\n H=900,\n W=1600,\n rand_flip=True,\n rot3d_range=[-0.3925, 0.3925])\ndata = dict(\n samples_per_gpu=20,\n workers_per_gpu=20,\n train=dict(\n type='NuScenes3DDetTrackDataset',\n data_root='data/nuscenes/',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ],\n modality=dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False),\n version='v1.0-trainval',\n ann_file='data/nuscenes_anno_pkls/nuscenes_infos_train.pkl',\n pipeline=[\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(\n type='LoadPointsFromFile',\n coord_type='LIDAR',\n load_dim=5,\n use_dim=5,\n file_client_args=dict(backend='disk')),\n dict(type='ResizeCropFlipImage'),\n dict(type='MultiScaleDepthMapGenerator', downsample=[4, 8, 16]),\n dict(type='BBoxRotation'),\n dict(type='PhotoMetricDistortionMultiViewImage'),\n dict(\n type='NormalizeMultiviewImage',\n mean=[123.675, 116.28, 103.53],\n std=[58.395, 57.12, 57.375],\n to_rgb=True),\n dict(\n type='CircleObjectRangeFilter',\n class_dist_thred=[55, 55, 55, 55, 55, 55, 55, 55, 55, 55]),\n dict(\n type='InstanceNameFilter',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian',\n 'traffic_cone'\n ]),\n dict(type='NuScenesSparse4DAdaptor'),\n dict(\n type='Collect',\n keys=[\n 'img', 'timestamp', 'projection_mat', 'image_wh',\n 'gt_depth', 'focal', 'gt_bboxes_3d', 'gt_labels_3d'\n ],\n meta_keys=[\n 'T_global', 'T_global_inv', 'timestamp', 'instance_id'\n ])\n ],\n test_mode=False,\n data_aug_conf=dict(\n resize_lim=(0.4, 0.47),\n final_dim=(256, 704),\n bot_pct_lim=(0.0, 0.0),\n rot_lim=(-5.4, 5.4),\n H=900,\n W=1600,\n rand_flip=True,\n rot3d_range=[-0.3925, 0.3925]),\n with_seq_flag=True,\n sequences_split_num=2,\n keep_consistent_seq_aug=True),\n val=dict(\n type='NuScenes3DDetTrackDataset',\n data_root='data/nuscenes/',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ],\n modality=dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False),\n version='v1.0-trainval',\n ann_file='data/nuscenes_anno_pkls/nuscenes_infos_val.pkl',\n pipeline=[\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(type='ResizeCropFlipImage'),\n dict(\n type='NormalizeMultiviewImage',\n mean=[123.675, 116.28, 103.53],\n std=[58.395, 57.12, 57.375],\n to_rgb=True),\n dict(type='NuScenesSparse4DAdaptor'),\n dict(\n type='Collect',\n keys=['img', 'timestamp', 'projection_mat', 'image_wh'],\n meta_keys=['T_global', 'T_global_inv', 'timestamp'])\n ],\n data_aug_conf=dict(\n resize_lim=(0.4, 0.47),\n final_dim=(256, 704),\n bot_pct_lim=(0.0, 0.0),\n rot_lim=(-5.4, 5.4),\n H=900,\n W=1600,\n rand_flip=True,\n rot3d_range=[-0.3925, 0.3925]),\n test_mode=True,\n tracking=True,\n tracking_threshold=0.2),\n test=dict(\n type='NuScenes3DDetTrackDataset',\n data_root='data/nuscenes/',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ],\n modality=dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False),\n version='v1.0-trainval',\n ann_file='data/nuscenes_anno_pkls/nuscenes_infos_val.pkl',\n pipeline=[\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(type='ResizeCropFlipImage'),\n dict(\n type='NormalizeMultiviewImage',\n mean=[123.675, 116.28, 103.53],\n std=[58.395, 57.12, 57.375],\n to_rgb=True),\n dict(type='NuScenesSparse4DAdaptor'),\n dict(\n type='Collect',\n keys=['img', 'timestamp', 'projection_mat', 'image_wh'],\n meta_keys=['T_global', 'T_global_inv', 'timestamp'])\n ],\n data_aug_conf=dict(\n resize_lim=(0.4, 0.47),\n final_dim=(256, 704),\n bot_pct_lim=(0.0, 0.0),\n rot_lim=(-5.4, 5.4),\n H=900,\n W=1600,\n rand_flip=True,\n rot3d_range=[-0.3925, 0.3925]),\n test_mode=True,\n tracking=True,\n tracking_threshold=0.2))\noptimizer = dict(\n type='Miopen_AdamW',\n lr=0.0006,\n weight_decay=0.001,\n paramwise_cfg=dict(custom_keys=dict(img_backbone=dict(lr_mult=0.5))))\noptimizer_config = dict(grad_clip=dict(max_norm=25, norm_type=2))\nlr_config = dict(\n policy='CosineAnnealing',\n warmup='linear',\n warmup_iters=500,\n warmup_ratio=0.3333333333333333,\n min_lr_ratio=0.001)\nrunner = dict(type='IterBasedRunner', max_iters=17500)\nvis_pipeline = [\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(type='Collect', keys=['img'], meta_keys=['timestamp', 'lidar2img'])\n]\nevaluation = dict(\n interval=3500,\n pipeline=[\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(\n type='Collect', keys=['img'], meta_keys=['timestamp', 'lidar2img'])\n ])\ngpu_ids = range(0, 8)\n", "seed": 0, "exp_name": "sparse4dv3_temporal_r50_1x8_bs6_256x704.py"} diff --git a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20251113_152314.log b/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20251113_152314.log deleted file mode 100644 index 9bd4289896e2e1883ecdf52bbee74b9b3ad4e1d3..0000000000000000000000000000000000000000 --- a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20251113_152314.log +++ /dev/null @@ -1,3220 +0,0 @@ -2025-11-13 15:23:14,483 - mmdet - INFO - Environment info: ------------------------------------------------------------- -sys.platform: linux -Python: 3.10.12 (main, May 27 2025, 17:12:29) [GCC 11.4.0] -CUDA available: True -GPU 0,1,2,3,4,5,6,7: BW200, UBB BW1000 -CUDA_HOME: /opt/dtk -NVCC: Not Available -GCC: x86_64-linux-gnu-gcc (Ubuntu 11.4.0-1ubuntu1~22.04) 11.4.0 -PyTorch: 2.4.1 -PyTorch compiling details: PyTorch built with: - - GCC 10.3 - - C++ Version: 201703 - - Intel(R) Math Kernel Library Version 2020.0.4 Product Build 20200917 for Intel(R) 64 architecture applications - - OpenMP 201511 (a.k.a. OpenMP 4.5) - - LAPACK is enabled (usually provided by MKL) - - NNPACK is enabled - - CPU capability usage: AVX512 - - HIP Runtime 6.3.25211 - - MIOpen 2.17.0 - - Magma 2.8.0 - - Build settings: BLAS_INFO=mkl, BUILD_TYPE=Release, CXX_COMPILER=/opt/rh/gcc-toolset-10/root/usr/bin/c++, CXX_FLAGS= -D_GLIBCXX_USE_CXX11_ABI=1 -fvisibility-inlines-hidden -DUSE_PTHREADPOOL -DNDEBUG -DUSE_KINETO -DLIBKINETO_NOCUPTI -DUSE_FBGEMM -DUSE_PYTORCH_QNNPACK -DUSE_XNNPACK -DSYMBOLICATE_MOBILE_DEBUG_HANDLE -O2 -fPIC -Wall -Wextra -Werror=return-type -Werror=non-virtual-dtor -Werror=bool-operation -Wnarrowing -Wno-missing-field-initializers -Wno-type-limits -Wno-array-bounds -Wno-unknown-pragmas -Wno-unused-parameter -Wno-unused-function -Wno-unused-result -Wno-strict-overflow -Wno-strict-aliasing -Wno-stringop-overflow -Wsuggest-override -Wno-psabi -Wno-error=pedantic -Wno-error=old-style-cast -Wno-missing-braces -fdiagnostics-color=always -faligned-new -Wno-unused-but-set-variable -Wno-maybe-uninitialized -fno-math-errno -fno-trapping-math -Werror=format -Wno-stringop-overflow, FORCE_FALLBACK_CUDA_MPI=1, LAPACK_INFO=mkl, PERF_WITH_AVX=1, PERF_WITH_AVX2=1, PERF_WITH_AVX512=1, TORCH_VERSION=2.4.1, USE_CUDA=0, USE_CUDNN=OFF, USE_CUSPARSELT=OFF, USE_EXCEPTION_PTR=1, USE_GFLAGS=1, USE_GLOG=1, USE_GLOO=1, USE_MKL=ON, USE_MKLDNN=0, USE_MPI=1, USE_NCCL=1, USE_NNPACK=ON, USE_OPENMP=1, USE_ROCM=ON, USE_ROCM_KERNEL_ASSERT=OFF, - -TorchVision: 0.19.1 -OpenCV: 4.11.0 -MMCV: 1.6.1 -MMCV Compiler: GCC 11.4 -MMCV CUDA Compiler: rocm not available -MMDetection: 2.26.0+c41df4b ------------------------------------------------------------- - -2025-11-13 15:23:15,412 - mmdet - INFO - Distributed training: True -2025-11-13 15:23:16,125 - mmdet - INFO - Config: -plugin = True -plugin_dir = 'projects/mmdet3d_plugin/' -dist_params = dict(backend='nccl') -log_level = 'INFO' -work_dir = './work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704' -total_batch_size = 160 -num_gpus = 8 -batch_size = 20 -num_iters_per_epoch = 175 -num_epochs = 100 -checkpoint_epoch_interval = 20 -checkpoint_config = dict(interval=3500) -log_config = dict( - interval=1, - hooks=[ - dict(type='TextLoggerHook', by_epoch=False), - dict(type='TensorboardLoggerHook') - ]) -load_from = None -resume_from = None -workflow = [('train', 1)] -fp16 = dict(loss_scale=32.0) -input_shape = (704, 256) -tracking_test = True -tracking_threshold = 0.2 -class_names = [ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier', - 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' -] -num_classes = 10 -embed_dims = 256 -num_groups = 8 -num_decoder = 6 -num_single_frame_decoder = 1 -use_deformable_func = True -strides = [4, 8, 16, 32] -num_levels = 4 -num_depth_layers = 3 -drop_out = 0.1 -temporal = True -decouple_attn = True -with_quality_estimation = True -model = dict( - type='Sparse4D', - use_grid_mask=True, - use_deformable_func=True, - img_backbone=dict( - type='ResNet', - depth=50, - num_stages=4, - frozen_stages=-1, - norm_eval=False, - style='pytorch', - with_cp=True, - out_indices=(0, 1, 2, 3), - norm_cfg=dict(type='BN', requires_grad=True), - pretrained='ckpt/resnet50-19c8e357.pth'), - img_neck=dict( - type='FPN', - num_outs=4, - start_level=0, - out_channels=256, - add_extra_convs='on_output', - relu_before_extra_convs=True, - in_channels=[256, 512, 1024, 2048]), - depth_branch=dict( - type='DenseDepthNet', - embed_dims=256, - num_depth_layers=3, - loss_weight=0.2), - head=dict( - type='Sparse4DHead', - cls_threshold_to_reg=0.05, - decouple_attn=True, - instance_bank=dict( - type='InstanceBank', - num_anchor=900, - embed_dims=256, - anchor='nuscenes_kmeans900.npy', - anchor_handler=dict(type='SparseBox3DKeyPointsGenerator'), - num_temp_instances=600, - confidence_decay=0.6, - feat_grad=False), - anchor_encoder=dict( - type='SparseBox3DEncoder', - vel_dims=3, - embed_dims=[128, 32, 32, 64], - mode='cat', - output_fc=False, - in_loops=1, - out_loops=4), - num_single_frame_decoder=1, - operation_order=[ - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine' - ], - temp_graph_model=dict( - type='MultiheadAttention', - embed_dims=512, - num_heads=8, - batch_first=True, - dropout=0.1), - graph_model=dict( - type='MultiheadAttention', - embed_dims=512, - num_heads=8, - batch_first=True, - dropout=0.1), - norm_layer=dict(type='LN', normalized_shape=256), - ffn=dict( - type='AsymmetricFFN', - in_channels=512, - pre_norm=dict(type='LN'), - embed_dims=256, - feedforward_channels=1024, - num_fcs=2, - ffn_drop=0.1, - act_cfg=dict(type='ReLU', inplace=True)), - deformable_model=dict( - type='DeformableFeatureAggregation', - embed_dims=256, - num_groups=8, - num_levels=4, - num_cams=6, - attn_drop=0.15, - use_deformable_func=True, - use_camera_embed=True, - residual_mode='cat', - kps_generator=dict( - type='SparseBox3DKeyPointsGenerator', - num_learnable_pts=6, - fix_scale=[[0, 0, 0], [0.45, 0, 0], [-0.45, 0, 0], - [0, 0.45, 0], [0, -0.45, 0], [0, 0, 0.45], - [0, 0, -0.45]])), - refine_layer=dict( - type='SparseBox3DRefinementModule', - embed_dims=256, - num_cls=10, - refine_yaw=True, - with_quality_estimation=True), - sampler=dict( - type='SparseBox3DTarget', - num_dn_groups=5, - num_temp_dn_groups=3, - dn_noise_scale=[2.0, 2.0, 2.0, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5], - max_dn_gt=32, - add_neg_dn=True, - cls_weight=2.0, - box_weight=0.25, - reg_weights=[2.0, 2.0, 2.0, 0.5, 0.5, 0.5, 0.0, 0.0, 0.0, 0.0], - cls_wise_reg_weights=dict( - {9: [2.0, 2.0, 2.0, 1.0, 1.0, 1.0, 0.0, 0.0, 1.0, 1.0]})), - loss_cls=dict( - type='FocalLoss', - use_sigmoid=True, - gamma=2.0, - alpha=0.25, - loss_weight=2.0), - loss_reg=dict( - type='SparseBox3DLoss', - loss_box=dict(type='L1Loss', loss_weight=0.25), - loss_centerness=dict(type='CrossEntropyLoss', use_sigmoid=True), - loss_yawness=dict(type='GaussianFocalLoss'), - cls_allow_reverse=[5]), - decoder=dict(type='SparseBox3DDecoder'), - reg_weights=[2.0, 2.0, 2.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0])) -dataset_type = 'NuScenes3DDetTrackDataset' -data_root = 'data/nuscenes/' -anno_root = 'data/nuscenes_anno_pkls/' -file_client_args = dict(backend='disk') -img_norm_cfg = dict( - mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375], to_rgb=True) -train_pipeline = [ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict( - type='LoadPointsFromFile', - coord_type='LIDAR', - load_dim=5, - use_dim=5, - file_client_args=dict(backend='disk')), - dict(type='ResizeCropFlipImage'), - dict(type='MultiScaleDepthMapGenerator', downsample=[4, 8, 16]), - dict(type='BBoxRotation'), - dict(type='PhotoMetricDistortionMultiViewImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict( - type='CircleObjectRangeFilter', - class_dist_thred=[55, 55, 55, 55, 55, 55, 55, 55, 55, 55]), - dict( - type='InstanceNameFilter', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ]), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=[ - 'img', 'timestamp', 'projection_mat', 'image_wh', 'gt_depth', - 'focal', 'gt_bboxes_3d', 'gt_labels_3d' - ], - meta_keys=['T_global', 'T_global_inv', 'timestamp', 'instance_id']) -] -test_pipeline = [ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='ResizeCropFlipImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=['img', 'timestamp', 'projection_mat', 'image_wh'], - meta_keys=['T_global', 'T_global_inv', 'timestamp']) -] -input_modality = dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False) -data_basic_config = dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier', - 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval') -data_aug_conf = dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]) -data = dict( - samples_per_gpu=20, - workers_per_gpu=20, - train=dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval', - ann_file='data/nuscenes_anno_pkls/nuscenes_infos_train.pkl', - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict( - type='LoadPointsFromFile', - coord_type='LIDAR', - load_dim=5, - use_dim=5, - file_client_args=dict(backend='disk')), - dict(type='ResizeCropFlipImage'), - dict(type='MultiScaleDepthMapGenerator', downsample=[4, 8, 16]), - dict(type='BBoxRotation'), - dict(type='PhotoMetricDistortionMultiViewImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict( - type='CircleObjectRangeFilter', - class_dist_thred=[55, 55, 55, 55, 55, 55, 55, 55, 55, 55]), - dict( - type='InstanceNameFilter', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', - 'traffic_cone' - ]), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=[ - 'img', 'timestamp', 'projection_mat', 'image_wh', - 'gt_depth', 'focal', 'gt_bboxes_3d', 'gt_labels_3d' - ], - meta_keys=[ - 'T_global', 'T_global_inv', 'timestamp', 'instance_id' - ]) - ], - test_mode=False, - data_aug_conf=dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]), - with_seq_flag=True, - sequences_split_num=2, - keep_consistent_seq_aug=True), - val=dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval', - ann_file='data/nuscenes_anno_pkls/nuscenes_infos_val.pkl', - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='ResizeCropFlipImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=['img', 'timestamp', 'projection_mat', 'image_wh'], - meta_keys=['T_global', 'T_global_inv', 'timestamp']) - ], - data_aug_conf=dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]), - test_mode=True, - tracking=True, - tracking_threshold=0.2), - test=dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval', - ann_file='data/nuscenes_anno_pkls/nuscenes_infos_val.pkl', - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='ResizeCropFlipImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=['img', 'timestamp', 'projection_mat', 'image_wh'], - meta_keys=['T_global', 'T_global_inv', 'timestamp']) - ], - data_aug_conf=dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]), - test_mode=True, - tracking=True, - tracking_threshold=0.2)) -optimizer = dict( - type='Miopen_AdamW', - lr=0.0006, - weight_decay=0.001, - paramwise_cfg=dict(custom_keys=dict(img_backbone=dict(lr_mult=0.5)))) -optimizer_config = dict(grad_clip=dict(max_norm=25, norm_type=2)) -lr_config = dict( - policy='CosineAnnealing', - warmup='linear', - warmup_iters=500, - warmup_ratio=0.3333333333333333, - min_lr_ratio=0.001) -runner = dict(type='IterBasedRunner', max_iters=17500) -vis_pipeline = [ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='Collect', keys=['img'], meta_keys=['timestamp', 'lidar2img']) -] -evaluation = dict( - interval=3500, - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict( - type='Collect', keys=['img'], meta_keys=['timestamp', 'lidar2img']) - ]) -gpu_ids = range(0, 8) - -2025-11-13 15:23:16,125 - mmdet - INFO - Set random seed to 0, deterministic: False -2025-11-13 15:23:16,426 - mmdet - INFO - initialize ResNet with init_cfg {'type': 'Pretrained', 'checkpoint': 'ckpt/resnet50-19c8e357.pth'} -2025-11-13 15:23:16,672 - mmdet - INFO - initialize FPN with init_cfg {'type': 'Xavier', 'layer': 'Conv2d', 'distribution': 'uniform'} -Name of parameter - Initialization information - -img_backbone.conv1.weight - torch.Size([64, 3, 7, 7]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.bn1.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.bn1.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.conv1.weight - torch.Size([64, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn1.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn1.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.conv2.weight - torch.Size([64, 64, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn2.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn2.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.conv3.weight - torch.Size([256, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn3.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn3.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.downsample.0.weight - torch.Size([256, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.downsample.1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.downsample.1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.conv1.weight - torch.Size([64, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn1.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn1.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.conv2.weight - torch.Size([64, 64, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn2.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn2.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.conv3.weight - torch.Size([256, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn3.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn3.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.conv1.weight - torch.Size([64, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn1.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn1.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.conv2.weight - torch.Size([64, 64, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn2.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn2.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.conv3.weight - torch.Size([256, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn3.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn3.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.conv1.weight - torch.Size([128, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn1.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn1.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.conv2.weight - torch.Size([128, 128, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn2.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn2.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.conv3.weight - torch.Size([512, 128, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn3.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn3.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.downsample.0.weight - torch.Size([512, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.downsample.1.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.downsample.1.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.conv1.weight - torch.Size([128, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn1.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn1.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.conv2.weight - torch.Size([128, 128, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn2.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn2.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.conv3.weight - torch.Size([512, 128, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn3.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn3.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.conv1.weight - torch.Size([128, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn1.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn1.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.conv2.weight - torch.Size([128, 128, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn2.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn2.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.conv3.weight - torch.Size([512, 128, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn3.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn3.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.conv1.weight - torch.Size([128, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn1.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn1.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.conv2.weight - torch.Size([128, 128, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn2.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn2.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.conv3.weight - torch.Size([512, 128, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn3.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn3.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.conv1.weight - torch.Size([256, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.downsample.0.weight - torch.Size([1024, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.downsample.1.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.downsample.1.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.conv1.weight - torch.Size([512, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn1.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn1.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.conv2.weight - torch.Size([512, 512, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn2.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn2.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.conv3.weight - torch.Size([2048, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn3.weight - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn3.bias - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.downsample.0.weight - torch.Size([2048, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.downsample.1.weight - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.downsample.1.bias - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.conv1.weight - torch.Size([512, 2048, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn1.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn1.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.conv2.weight - torch.Size([512, 512, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn2.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn2.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.conv3.weight - torch.Size([2048, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn3.weight - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn3.bias - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.conv1.weight - torch.Size([512, 2048, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn1.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn1.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.conv2.weight - torch.Size([512, 512, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn2.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn2.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.conv3.weight - torch.Size([2048, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn3.weight - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn3.bias - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_neck.lateral_convs.0.conv.weight - torch.Size([256, 256, 1, 1]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.lateral_convs.0.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.lateral_convs.1.conv.weight - torch.Size([256, 512, 1, 1]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.lateral_convs.1.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.lateral_convs.2.conv.weight - torch.Size([256, 1024, 1, 1]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.lateral_convs.2.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.lateral_convs.3.conv.weight - torch.Size([256, 2048, 1, 1]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.lateral_convs.3.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.fpn_convs.0.conv.weight - torch.Size([256, 256, 3, 3]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.fpn_convs.0.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.fpn_convs.1.conv.weight - torch.Size([256, 256, 3, 3]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.fpn_convs.1.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.fpn_convs.2.conv.weight - torch.Size([256, 256, 3, 3]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.fpn_convs.2.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.fpn_convs.3.conv.weight - torch.Size([256, 256, 3, 3]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.fpn_convs.3.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.instance_bank.anchor - torch.Size([900, 11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.instance_bank.instance_feature - torch.Size([900, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.instance_bank.anchor_handler.fix_scale - torch.Size([1, 3]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.0.weight - torch.Size([128, 3]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.0.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.2.weight - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.2.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.3.weight - torch.Size([128, 128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.3.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.5.weight - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.5.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.6.weight - torch.Size([128, 128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.6.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.8.weight - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.8.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.9.weight - torch.Size([128, 128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.9.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.11.weight - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.11.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.0.weight - torch.Size([32, 3]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.0.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.2.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.2.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.3.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.3.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.5.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.5.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.6.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.6.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.8.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.8.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.9.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.9.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.11.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.11.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.0.weight - torch.Size([32, 2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.0.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.2.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.2.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.3.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.3.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.5.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.5.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.6.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.6.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.8.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.8.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.9.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.9.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.11.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.11.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.0.weight - torch.Size([64, 3]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.0.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.2.weight - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.2.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.3.weight - torch.Size([64, 64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.3.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.5.weight - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.5.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.6.weight - torch.Size([64, 64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.6.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.8.weight - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.8.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.9.weight - torch.Size([64, 64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.9.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.11.weight - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.11.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.1.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.1.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.1.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.1.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.1.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.1.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.1.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.1.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.3.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.4.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.4.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.4.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.4.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.5.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.5.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.5.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.5.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.6.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.6.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.8.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.8.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.8.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.8.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.8.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.8.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.8.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.8.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.10.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.11.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.11.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.11.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.11.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.12.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.12.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.12.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.12.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.13.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.13.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.15.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.15.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.15.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.15.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.15.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.15.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.15.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.15.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.16.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.16.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.17.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.18.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.18.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.18.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.18.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.19.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.19.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.19.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.19.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.20.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.20.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.22.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.22.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.22.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.22.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.22.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.22.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.22.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.22.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.23.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.23.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.24.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.25.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.25.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.25.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.25.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.26.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.26.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.26.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.26.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.27.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.27.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.29.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.29.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.29.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.29.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.29.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.29.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.29.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.29.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.30.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.30.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.31.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.32.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.32.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.32.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.32.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.33.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.33.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.33.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.33.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.34.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.34.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.36.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.36.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.36.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.36.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.36.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.36.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.36.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.36.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.37.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.37.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.38.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.fc_before.weight - torch.Size([512, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.fc_after.weight - torch.Size([256, 512]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.0.weight - torch.Size([1, 256, 1, 1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.0.bias - torch.Size([1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.1.weight - torch.Size([1, 256, 1, 1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.1.bias - torch.Size([1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.2.weight - torch.Size([1, 256, 1, 1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.2.bias - torch.Size([1]): -The value is the same before and after calling `init_weights` of Sparse4D -2025-11-13 15:23:16,762 - mmdet - INFO - Model: -Sparse4D( - (img_backbone): ResNet( - (conv1): Conv2d(3, 64, kernel_size=(7, 7), stride=(2, 2), padding=(3, 3), bias=False) - (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (maxpool): MaxPool2d(kernel_size=3, stride=2, padding=1, dilation=1, ceil_mode=False) - (layer1): ResLayer( - (0): Bottleneck( - (conv1): Conv2d(64, 64, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (downsample): Sequential( - (0): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - ) - ) - (1): Bottleneck( - (conv1): Conv2d(256, 64, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (2): Bottleneck( - (conv1): Conv2d(256, 64, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - ) - (layer2): ResLayer( - (0): Bottleneck( - (conv1): Conv2d(256, 128, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (downsample): Sequential( - (0): Conv2d(256, 512, kernel_size=(1, 1), stride=(2, 2), bias=False) - (1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - ) - ) - (1): Bottleneck( - (conv1): Conv2d(512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (2): Bottleneck( - (conv1): Conv2d(512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (3): Bottleneck( - (conv1): Conv2d(512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - ) - (layer3): ResLayer( - (0): Bottleneck( - (conv1): Conv2d(512, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (downsample): Sequential( - (0): Conv2d(512, 1024, kernel_size=(1, 1), stride=(2, 2), bias=False) - (1): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - ) - ) - (1): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (2): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (3): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (4): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (5): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - ) - (layer4): ResLayer( - (0): Bottleneck( - (conv1): Conv2d(1024, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(512, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (downsample): Sequential( - (0): Conv2d(1024, 2048, kernel_size=(1, 1), stride=(2, 2), bias=False) - (1): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - ) - ) - (1): Bottleneck( - (conv1): Conv2d(2048, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(512, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (2): Bottleneck( - (conv1): Conv2d(2048, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(512, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - ) - ) - init_cfg={'type': 'Pretrained', 'checkpoint': 'ckpt/resnet50-19c8e357.pth'} - (img_neck): FPN( - (lateral_convs): ModuleList( - (0): ConvModule( - (conv): Conv2d(256, 256, kernel_size=(1, 1), stride=(1, 1)) - ) - (1): ConvModule( - (conv): Conv2d(512, 256, kernel_size=(1, 1), stride=(1, 1)) - ) - (2): ConvModule( - (conv): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1)) - ) - (3): ConvModule( - (conv): Conv2d(2048, 256, kernel_size=(1, 1), stride=(1, 1)) - ) - ) - (fpn_convs): ModuleList( - (0-3): 4 x ConvModule( - (conv): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1)) - ) - ) - ) - init_cfg={'type': 'Xavier', 'layer': 'Conv2d', 'distribution': 'uniform'} - (head): Sparse4DHead( - (instance_bank): InstanceBank( - (anchor_handler): SparseBox3DKeyPointsGenerator() - ) - (anchor_encoder): SparseBox3DEncoder( - (pos_fc): Sequential( - (0): Linear(in_features=3, out_features=128, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((128,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=128, out_features=128, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((128,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=128, out_features=128, bias=True) - (7): ReLU(inplace=True) - (8): LayerNorm((128,), eps=1e-05, elementwise_affine=True) - (9): Linear(in_features=128, out_features=128, bias=True) - (10): ReLU(inplace=True) - (11): LayerNorm((128,), eps=1e-05, elementwise_affine=True) - ) - (size_fc): Sequential( - (0): Linear(in_features=3, out_features=32, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=32, out_features=32, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=32, out_features=32, bias=True) - (7): ReLU(inplace=True) - (8): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (9): Linear(in_features=32, out_features=32, bias=True) - (10): ReLU(inplace=True) - (11): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - ) - (yaw_fc): Sequential( - (0): Linear(in_features=2, out_features=32, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=32, out_features=32, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=32, out_features=32, bias=True) - (7): ReLU(inplace=True) - (8): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (9): Linear(in_features=32, out_features=32, bias=True) - (10): ReLU(inplace=True) - (11): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - ) - (vel_fc): Sequential( - (0): Linear(in_features=3, out_features=64, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((64,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=64, out_features=64, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((64,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=64, out_features=64, bias=True) - (7): ReLU(inplace=True) - (8): LayerNorm((64,), eps=1e-05, elementwise_affine=True) - (9): Linear(in_features=64, out_features=64, bias=True) - (10): ReLU(inplace=True) - (11): LayerNorm((64,), eps=1e-05, elementwise_affine=True) - ) - ) - (loss_cls): FocalLoss() - (loss_reg): SparseBox3DLoss( - (loss_box): L1Loss() - (loss_cns): CrossEntropyLoss(avg_non_ignore=False) - (loss_yns): GaussianFocalLoss() - ) - (layers): ModuleList( - (0): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (1): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (4-5): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (6): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (7): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (8): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (11-12): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (13): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (14): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (15): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (16): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (17): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (18-19): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (20): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (21): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (22): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (23): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (24): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (25-26): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (27): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (28): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (29): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (30): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (31): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (32-33): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (34): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (35): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (36): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (37): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (38): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - ) - (fc_before): Linear(in_features=256, out_features=512, bias=False) - (fc_after): Linear(in_features=512, out_features=256, bias=False) - ) - (depth_branch): DenseDepthNet( - (depth_layers): ModuleList( - (0-2): 3 x Conv2d(256, 1, kernel_size=(1, 1), stride=(1, 1)) - ) - ) - (grid_mask): GridMask() -) -2025-11-13 15:23:29,232 - mmdet - INFO - Start running, host: root@VM-120-96-tencentos, work_dir: /cfs/auto/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704 -2025-11-13 15:23:29,232 - mmdet - INFO - Hooks will be executed in the following order: -before_run: -(VERY_HIGH ) CosineAnnealingLrUpdaterHook -(ABOVE_NORMAL) Fp16OptimizerHook -(NORMAL ) CheckpointHook -(NORMAL ) CustomDistEvalHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -before_train_epoch: -(VERY_HIGH ) CosineAnnealingLrUpdaterHook -(NORMAL ) CustomDistEvalHook -(LOW ) IterTimerHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -before_train_iter: -(VERY_HIGH ) CosineAnnealingLrUpdaterHook -(NORMAL ) CustomDistEvalHook -(LOW ) IterTimerHook - -------------------- -after_train_iter: -(ABOVE_NORMAL) Fp16OptimizerHook -(NORMAL ) CheckpointHook -(NORMAL ) CustomDistEvalHook -(LOW ) IterTimerHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -after_train_epoch: -(NORMAL ) CheckpointHook -(NORMAL ) CustomDistEvalHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -before_val_epoch: -(LOW ) IterTimerHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -before_val_iter: -(LOW ) IterTimerHook - -------------------- -after_val_iter: -(LOW ) IterTimerHook - -------------------- -after_val_epoch: -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -after_run: -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -2025-11-13 15:23:29,233 - mmdet - INFO - workflow: [('train', 1)], max: 17500 iters -2025-11-13 15:23:29,235 - mmdet - INFO - Checkpoints will be saved to /cfs/auto/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704 by HardDiskBackend. diff --git a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20251113_152314.log.json b/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20251113_152314.log.json deleted file mode 100644 index 1fda3e213098b4a1f6239f0e36f9459d0a88f9c1..0000000000000000000000000000000000000000 --- a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20251113_152314.log.json +++ /dev/null @@ -1 +0,0 @@ -{"env_info": "sys.platform: linux\nPython: 3.10.12 (main, May 27 2025, 17:12:29) [GCC 11.4.0]\nCUDA available: True\nGPU 0,1,2,3,4,5,6,7: BW200, UBB BW1000\nCUDA_HOME: /opt/dtk\nNVCC: Not Available\nGCC: x86_64-linux-gnu-gcc (Ubuntu 11.4.0-1ubuntu1~22.04) 11.4.0\nPyTorch: 2.4.1\nPyTorch compiling details: PyTorch built with:\n - GCC 10.3\n - C++ Version: 201703\n - Intel(R) Math Kernel Library Version 2020.0.4 Product Build 20200917 for Intel(R) 64 architecture applications\n - OpenMP 201511 (a.k.a. OpenMP 4.5)\n - LAPACK is enabled (usually provided by MKL)\n - NNPACK is enabled\n - CPU capability usage: AVX512\n - HIP Runtime 6.3.25211\n - MIOpen 2.17.0\n - Magma 2.8.0\n - Build settings: BLAS_INFO=mkl, BUILD_TYPE=Release, CXX_COMPILER=/opt/rh/gcc-toolset-10/root/usr/bin/c++, CXX_FLAGS= -D_GLIBCXX_USE_CXX11_ABI=1 -fvisibility-inlines-hidden -DUSE_PTHREADPOOL -DNDEBUG -DUSE_KINETO -DLIBKINETO_NOCUPTI -DUSE_FBGEMM -DUSE_PYTORCH_QNNPACK -DUSE_XNNPACK -DSYMBOLICATE_MOBILE_DEBUG_HANDLE -O2 -fPIC -Wall -Wextra -Werror=return-type -Werror=non-virtual-dtor -Werror=bool-operation -Wnarrowing -Wno-missing-field-initializers -Wno-type-limits -Wno-array-bounds -Wno-unknown-pragmas -Wno-unused-parameter -Wno-unused-function -Wno-unused-result -Wno-strict-overflow -Wno-strict-aliasing -Wno-stringop-overflow -Wsuggest-override -Wno-psabi -Wno-error=pedantic -Wno-error=old-style-cast -Wno-missing-braces -fdiagnostics-color=always -faligned-new -Wno-unused-but-set-variable -Wno-maybe-uninitialized -fno-math-errno -fno-trapping-math -Werror=format -Wno-stringop-overflow, FORCE_FALLBACK_CUDA_MPI=1, LAPACK_INFO=mkl, PERF_WITH_AVX=1, PERF_WITH_AVX2=1, PERF_WITH_AVX512=1, TORCH_VERSION=2.4.1, USE_CUDA=0, USE_CUDNN=OFF, USE_CUSPARSELT=OFF, USE_EXCEPTION_PTR=1, USE_GFLAGS=1, USE_GLOG=1, USE_GLOO=1, USE_MKL=ON, USE_MKLDNN=0, USE_MPI=1, USE_NCCL=1, USE_NNPACK=ON, USE_OPENMP=1, USE_ROCM=ON, USE_ROCM_KERNEL_ASSERT=OFF, \n\nTorchVision: 0.19.1\nOpenCV: 4.11.0\nMMCV: 1.6.1\nMMCV Compiler: GCC 11.4\nMMCV CUDA Compiler: rocm not available\nMMDetection: 2.26.0+c41df4b", "config": "plugin = True\nplugin_dir = 'projects/mmdet3d_plugin/'\ndist_params = dict(backend='nccl')\nlog_level = 'INFO'\nwork_dir = './work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704'\ntotal_batch_size = 160\nnum_gpus = 8\nbatch_size = 20\nnum_iters_per_epoch = 175\nnum_epochs = 100\ncheckpoint_epoch_interval = 20\ncheckpoint_config = dict(interval=3500)\nlog_config = dict(\n interval=1,\n hooks=[\n dict(type='TextLoggerHook', by_epoch=False),\n dict(type='TensorboardLoggerHook')\n ])\nload_from = None\nresume_from = None\nworkflow = [('train', 1)]\nfp16 = dict(loss_scale=32.0)\ninput_shape = (704, 256)\ntracking_test = True\ntracking_threshold = 0.2\nclass_names = [\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier',\n 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n]\nnum_classes = 10\nembed_dims = 256\nnum_groups = 8\nnum_decoder = 6\nnum_single_frame_decoder = 1\nuse_deformable_func = True\nstrides = [4, 8, 16, 32]\nnum_levels = 4\nnum_depth_layers = 3\ndrop_out = 0.1\ntemporal = True\ndecouple_attn = True\nwith_quality_estimation = True\nmodel = dict(\n type='Sparse4D',\n use_grid_mask=True,\n use_deformable_func=True,\n img_backbone=dict(\n type='ResNet',\n depth=50,\n num_stages=4,\n frozen_stages=-1,\n norm_eval=False,\n style='pytorch',\n with_cp=True,\n out_indices=(0, 1, 2, 3),\n norm_cfg=dict(type='BN', requires_grad=True),\n pretrained='ckpt/resnet50-19c8e357.pth'),\n img_neck=dict(\n type='FPN',\n num_outs=4,\n start_level=0,\n out_channels=256,\n add_extra_convs='on_output',\n relu_before_extra_convs=True,\n in_channels=[256, 512, 1024, 2048]),\n depth_branch=dict(\n type='DenseDepthNet',\n embed_dims=256,\n num_depth_layers=3,\n loss_weight=0.2),\n head=dict(\n type='Sparse4DHead',\n cls_threshold_to_reg=0.05,\n decouple_attn=True,\n instance_bank=dict(\n type='InstanceBank',\n num_anchor=900,\n embed_dims=256,\n anchor='nuscenes_kmeans900.npy',\n anchor_handler=dict(type='SparseBox3DKeyPointsGenerator'),\n num_temp_instances=600,\n confidence_decay=0.6,\n feat_grad=False),\n anchor_encoder=dict(\n type='SparseBox3DEncoder',\n vel_dims=3,\n embed_dims=[128, 32, 32, 64],\n mode='cat',\n output_fc=False,\n in_loops=1,\n out_loops=4),\n num_single_frame_decoder=1,\n operation_order=[\n 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm',\n 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm',\n 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm',\n 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm',\n 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm',\n 'deformable', 'ffn', 'norm', 'refine'\n ],\n temp_graph_model=dict(\n type='MultiheadAttention',\n embed_dims=512,\n num_heads=8,\n batch_first=True,\n dropout=0.1),\n graph_model=dict(\n type='MultiheadAttention',\n embed_dims=512,\n num_heads=8,\n batch_first=True,\n dropout=0.1),\n norm_layer=dict(type='LN', normalized_shape=256),\n ffn=dict(\n type='AsymmetricFFN',\n in_channels=512,\n pre_norm=dict(type='LN'),\n embed_dims=256,\n feedforward_channels=1024,\n num_fcs=2,\n ffn_drop=0.1,\n act_cfg=dict(type='ReLU', inplace=True)),\n deformable_model=dict(\n type='DeformableFeatureAggregation',\n embed_dims=256,\n num_groups=8,\n num_levels=4,\n num_cams=6,\n attn_drop=0.15,\n use_deformable_func=True,\n use_camera_embed=True,\n residual_mode='cat',\n kps_generator=dict(\n type='SparseBox3DKeyPointsGenerator',\n num_learnable_pts=6,\n fix_scale=[[0, 0, 0], [0.45, 0, 0], [-0.45, 0, 0],\n [0, 0.45, 0], [0, -0.45, 0], [0, 0, 0.45],\n [0, 0, -0.45]])),\n refine_layer=dict(\n type='SparseBox3DRefinementModule',\n embed_dims=256,\n num_cls=10,\n refine_yaw=True,\n with_quality_estimation=True),\n sampler=dict(\n type='SparseBox3DTarget',\n num_dn_groups=5,\n num_temp_dn_groups=3,\n dn_noise_scale=[2.0, 2.0, 2.0, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5],\n max_dn_gt=32,\n add_neg_dn=True,\n cls_weight=2.0,\n box_weight=0.25,\n reg_weights=[2.0, 2.0, 2.0, 0.5, 0.5, 0.5, 0.0, 0.0, 0.0, 0.0],\n cls_wise_reg_weights=dict(\n {9: [2.0, 2.0, 2.0, 1.0, 1.0, 1.0, 0.0, 0.0, 1.0, 1.0]})),\n loss_cls=dict(\n type='FocalLoss',\n use_sigmoid=True,\n gamma=2.0,\n alpha=0.25,\n loss_weight=2.0),\n loss_reg=dict(\n type='SparseBox3DLoss',\n loss_box=dict(type='L1Loss', loss_weight=0.25),\n loss_centerness=dict(type='CrossEntropyLoss', use_sigmoid=True),\n loss_yawness=dict(type='GaussianFocalLoss'),\n cls_allow_reverse=[5]),\n decoder=dict(type='SparseBox3DDecoder'),\n reg_weights=[2.0, 2.0, 2.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0]))\ndataset_type = 'NuScenes3DDetTrackDataset'\ndata_root = 'data/nuscenes/'\nanno_root = 'data/nuscenes_anno_pkls/'\nfile_client_args = dict(backend='disk')\nimg_norm_cfg = dict(\n mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375], to_rgb=True)\ntrain_pipeline = [\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(\n type='LoadPointsFromFile',\n coord_type='LIDAR',\n load_dim=5,\n use_dim=5,\n file_client_args=dict(backend='disk')),\n dict(type='ResizeCropFlipImage'),\n dict(type='MultiScaleDepthMapGenerator', downsample=[4, 8, 16]),\n dict(type='BBoxRotation'),\n dict(type='PhotoMetricDistortionMultiViewImage'),\n dict(\n type='NormalizeMultiviewImage',\n mean=[123.675, 116.28, 103.53],\n std=[58.395, 57.12, 57.375],\n to_rgb=True),\n dict(\n type='CircleObjectRangeFilter',\n class_dist_thred=[55, 55, 55, 55, 55, 55, 55, 55, 55, 55]),\n dict(\n type='InstanceNameFilter',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ]),\n dict(type='NuScenesSparse4DAdaptor'),\n dict(\n type='Collect',\n keys=[\n 'img', 'timestamp', 'projection_mat', 'image_wh', 'gt_depth',\n 'focal', 'gt_bboxes_3d', 'gt_labels_3d'\n ],\n meta_keys=['T_global', 'T_global_inv', 'timestamp', 'instance_id'])\n]\ntest_pipeline = [\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(type='ResizeCropFlipImage'),\n dict(\n type='NormalizeMultiviewImage',\n mean=[123.675, 116.28, 103.53],\n std=[58.395, 57.12, 57.375],\n to_rgb=True),\n dict(type='NuScenesSparse4DAdaptor'),\n dict(\n type='Collect',\n keys=['img', 'timestamp', 'projection_mat', 'image_wh'],\n meta_keys=['T_global', 'T_global_inv', 'timestamp'])\n]\ninput_modality = dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False)\ndata_basic_config = dict(\n type='NuScenes3DDetTrackDataset',\n data_root='data/nuscenes/',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier',\n 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ],\n modality=dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False),\n version='v1.0-trainval')\ndata_aug_conf = dict(\n resize_lim=(0.4, 0.47),\n final_dim=(256, 704),\n bot_pct_lim=(0.0, 0.0),\n rot_lim=(-5.4, 5.4),\n H=900,\n W=1600,\n rand_flip=True,\n rot3d_range=[-0.3925, 0.3925])\ndata = dict(\n samples_per_gpu=20,\n workers_per_gpu=20,\n train=dict(\n type='NuScenes3DDetTrackDataset',\n data_root='data/nuscenes/',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ],\n modality=dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False),\n version='v1.0-trainval',\n ann_file='data/nuscenes_anno_pkls/nuscenes_infos_train.pkl',\n pipeline=[\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(\n type='LoadPointsFromFile',\n coord_type='LIDAR',\n load_dim=5,\n use_dim=5,\n file_client_args=dict(backend='disk')),\n dict(type='ResizeCropFlipImage'),\n dict(type='MultiScaleDepthMapGenerator', downsample=[4, 8, 16]),\n dict(type='BBoxRotation'),\n dict(type='PhotoMetricDistortionMultiViewImage'),\n dict(\n type='NormalizeMultiviewImage',\n mean=[123.675, 116.28, 103.53],\n std=[58.395, 57.12, 57.375],\n to_rgb=True),\n dict(\n type='CircleObjectRangeFilter',\n class_dist_thred=[55, 55, 55, 55, 55, 55, 55, 55, 55, 55]),\n dict(\n type='InstanceNameFilter',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian',\n 'traffic_cone'\n ]),\n dict(type='NuScenesSparse4DAdaptor'),\n dict(\n type='Collect',\n keys=[\n 'img', 'timestamp', 'projection_mat', 'image_wh',\n 'gt_depth', 'focal', 'gt_bboxes_3d', 'gt_labels_3d'\n ],\n meta_keys=[\n 'T_global', 'T_global_inv', 'timestamp', 'instance_id'\n ])\n ],\n test_mode=False,\n data_aug_conf=dict(\n resize_lim=(0.4, 0.47),\n final_dim=(256, 704),\n bot_pct_lim=(0.0, 0.0),\n rot_lim=(-5.4, 5.4),\n H=900,\n W=1600,\n rand_flip=True,\n rot3d_range=[-0.3925, 0.3925]),\n with_seq_flag=True,\n sequences_split_num=2,\n keep_consistent_seq_aug=True),\n val=dict(\n type='NuScenes3DDetTrackDataset',\n data_root='data/nuscenes/',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ],\n modality=dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False),\n version='v1.0-trainval',\n ann_file='data/nuscenes_anno_pkls/nuscenes_infos_val.pkl',\n pipeline=[\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(type='ResizeCropFlipImage'),\n dict(\n type='NormalizeMultiviewImage',\n mean=[123.675, 116.28, 103.53],\n std=[58.395, 57.12, 57.375],\n to_rgb=True),\n dict(type='NuScenesSparse4DAdaptor'),\n dict(\n type='Collect',\n keys=['img', 'timestamp', 'projection_mat', 'image_wh'],\n meta_keys=['T_global', 'T_global_inv', 'timestamp'])\n ],\n data_aug_conf=dict(\n resize_lim=(0.4, 0.47),\n final_dim=(256, 704),\n bot_pct_lim=(0.0, 0.0),\n rot_lim=(-5.4, 5.4),\n H=900,\n W=1600,\n rand_flip=True,\n rot3d_range=[-0.3925, 0.3925]),\n test_mode=True,\n tracking=True,\n tracking_threshold=0.2),\n test=dict(\n type='NuScenes3DDetTrackDataset',\n data_root='data/nuscenes/',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ],\n modality=dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False),\n version='v1.0-trainval',\n ann_file='data/nuscenes_anno_pkls/nuscenes_infos_val.pkl',\n pipeline=[\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(type='ResizeCropFlipImage'),\n dict(\n type='NormalizeMultiviewImage',\n mean=[123.675, 116.28, 103.53],\n std=[58.395, 57.12, 57.375],\n to_rgb=True),\n dict(type='NuScenesSparse4DAdaptor'),\n dict(\n type='Collect',\n keys=['img', 'timestamp', 'projection_mat', 'image_wh'],\n meta_keys=['T_global', 'T_global_inv', 'timestamp'])\n ],\n data_aug_conf=dict(\n resize_lim=(0.4, 0.47),\n final_dim=(256, 704),\n bot_pct_lim=(0.0, 0.0),\n rot_lim=(-5.4, 5.4),\n H=900,\n W=1600,\n rand_flip=True,\n rot3d_range=[-0.3925, 0.3925]),\n test_mode=True,\n tracking=True,\n tracking_threshold=0.2))\noptimizer = dict(\n type='Miopen_AdamW',\n lr=0.0006,\n weight_decay=0.001,\n paramwise_cfg=dict(custom_keys=dict(img_backbone=dict(lr_mult=0.5))))\noptimizer_config = dict(grad_clip=dict(max_norm=25, norm_type=2))\nlr_config = dict(\n policy='CosineAnnealing',\n warmup='linear',\n warmup_iters=500,\n warmup_ratio=0.3333333333333333,\n min_lr_ratio=0.001)\nrunner = dict(type='IterBasedRunner', max_iters=17500)\nvis_pipeline = [\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(type='Collect', keys=['img'], meta_keys=['timestamp', 'lidar2img'])\n]\nevaluation = dict(\n interval=3500,\n pipeline=[\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(\n type='Collect', keys=['img'], meta_keys=['timestamp', 'lidar2img'])\n ])\ngpu_ids = range(0, 8)\n", "seed": 0, "exp_name": "sparse4dv3_temporal_r50_1x8_bs6_256x704.py"} diff --git a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20251113_153143.log b/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20251113_153143.log deleted file mode 100644 index 22b63764b356db0aaddaa573c7069875d07dcf2d..0000000000000000000000000000000000000000 --- a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20251113_153143.log +++ /dev/null @@ -1,3402 +0,0 @@ -2025-11-13 15:31:43,333 - mmdet - INFO - Environment info: ------------------------------------------------------------- -sys.platform: linux -Python: 3.10.12 (main, May 27 2025, 17:12:29) [GCC 11.4.0] -CUDA available: True -GPU 0,1,2,3,4,5,6,7: BW200, UBB BW1000 -CUDA_HOME: /opt/dtk -NVCC: Not Available -GCC: x86_64-linux-gnu-gcc (Ubuntu 11.4.0-1ubuntu1~22.04) 11.4.0 -PyTorch: 2.4.1 -PyTorch compiling details: PyTorch built with: - - GCC 10.3 - - C++ Version: 201703 - - Intel(R) Math Kernel Library Version 2020.0.4 Product Build 20200917 for Intel(R) 64 architecture applications - - OpenMP 201511 (a.k.a. OpenMP 4.5) - - LAPACK is enabled (usually provided by MKL) - - NNPACK is enabled - - CPU capability usage: AVX512 - - HIP Runtime 6.3.25211 - - MIOpen 2.17.0 - - Magma 2.8.0 - - Build settings: BLAS_INFO=mkl, BUILD_TYPE=Release, CXX_COMPILER=/opt/rh/gcc-toolset-10/root/usr/bin/c++, CXX_FLAGS= -D_GLIBCXX_USE_CXX11_ABI=1 -fvisibility-inlines-hidden -DUSE_PTHREADPOOL -DNDEBUG -DUSE_KINETO -DLIBKINETO_NOCUPTI -DUSE_FBGEMM -DUSE_PYTORCH_QNNPACK -DUSE_XNNPACK -DSYMBOLICATE_MOBILE_DEBUG_HANDLE -O2 -fPIC -Wall -Wextra -Werror=return-type -Werror=non-virtual-dtor -Werror=bool-operation -Wnarrowing -Wno-missing-field-initializers -Wno-type-limits -Wno-array-bounds -Wno-unknown-pragmas -Wno-unused-parameter -Wno-unused-function -Wno-unused-result -Wno-strict-overflow -Wno-strict-aliasing -Wno-stringop-overflow -Wsuggest-override -Wno-psabi -Wno-error=pedantic -Wno-error=old-style-cast -Wno-missing-braces -fdiagnostics-color=always -faligned-new -Wno-unused-but-set-variable -Wno-maybe-uninitialized -fno-math-errno -fno-trapping-math -Werror=format -Wno-stringop-overflow, FORCE_FALLBACK_CUDA_MPI=1, LAPACK_INFO=mkl, PERF_WITH_AVX=1, PERF_WITH_AVX2=1, PERF_WITH_AVX512=1, TORCH_VERSION=2.4.1, USE_CUDA=0, USE_CUDNN=OFF, USE_CUSPARSELT=OFF, USE_EXCEPTION_PTR=1, USE_GFLAGS=1, USE_GLOG=1, USE_GLOO=1, USE_MKL=ON, USE_MKLDNN=0, USE_MPI=1, USE_NCCL=1, USE_NNPACK=ON, USE_OPENMP=1, USE_ROCM=ON, USE_ROCM_KERNEL_ASSERT=OFF, - -TorchVision: 0.19.1 -OpenCV: 4.11.0 -MMCV: 1.6.1 -MMCV Compiler: GCC 11.4 -MMCV CUDA Compiler: rocm not available -MMDetection: 2.26.0+c41df4b ------------------------------------------------------------- - -2025-11-13 15:31:44,268 - mmdet - INFO - Distributed training: True -2025-11-13 15:31:44,987 - mmdet - INFO - Config: -plugin = True -plugin_dir = 'projects/mmdet3d_plugin/' -dist_params = dict(backend='nccl') -log_level = 'INFO' -work_dir = './work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704' -total_batch_size = 160 -num_gpus = 8 -batch_size = 20 -num_iters_per_epoch = 175 -num_epochs = 100 -checkpoint_epoch_interval = 20 -checkpoint_config = dict(interval=3500) -log_config = dict( - interval=1, - hooks=[ - dict(type='TextLoggerHook', by_epoch=False), - dict(type='TensorboardLoggerHook') - ]) -load_from = None -resume_from = None -workflow = [('train', 1)] -fp16 = dict(loss_scale=32.0) -input_shape = (704, 256) -tracking_test = True -tracking_threshold = 0.2 -class_names = [ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier', - 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' -] -num_classes = 10 -embed_dims = 256 -num_groups = 8 -num_decoder = 6 -num_single_frame_decoder = 1 -use_deformable_func = True -strides = [4, 8, 16, 32] -num_levels = 4 -num_depth_layers = 3 -drop_out = 0.1 -temporal = True -decouple_attn = True -with_quality_estimation = True -model = dict( - type='Sparse4D', - use_grid_mask=True, - use_deformable_func=True, - img_backbone=dict( - type='ResNet', - depth=50, - num_stages=4, - frozen_stages=-1, - norm_eval=False, - style='pytorch', - with_cp=True, - out_indices=(0, 1, 2, 3), - norm_cfg=dict(type='BN', requires_grad=True), - pretrained='ckpt/resnet50-19c8e357.pth'), - img_neck=dict( - type='FPN', - num_outs=4, - start_level=0, - out_channels=256, - add_extra_convs='on_output', - relu_before_extra_convs=True, - in_channels=[256, 512, 1024, 2048]), - depth_branch=dict( - type='DenseDepthNet', - embed_dims=256, - num_depth_layers=3, - loss_weight=0.2), - head=dict( - type='Sparse4DHead', - cls_threshold_to_reg=0.05, - decouple_attn=True, - instance_bank=dict( - type='InstanceBank', - num_anchor=900, - embed_dims=256, - anchor='nuscenes_kmeans900.npy', - anchor_handler=dict(type='SparseBox3DKeyPointsGenerator'), - num_temp_instances=600, - confidence_decay=0.6, - feat_grad=False), - anchor_encoder=dict( - type='SparseBox3DEncoder', - vel_dims=3, - embed_dims=[128, 32, 32, 64], - mode='cat', - output_fc=False, - in_loops=1, - out_loops=4), - num_single_frame_decoder=1, - operation_order=[ - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine' - ], - temp_graph_model=dict( - type='MultiheadAttention', - embed_dims=512, - num_heads=8, - batch_first=True, - dropout=0.1), - graph_model=dict( - type='MultiheadAttention', - embed_dims=512, - num_heads=8, - batch_first=True, - dropout=0.1), - norm_layer=dict(type='LN', normalized_shape=256), - ffn=dict( - type='AsymmetricFFN', - in_channels=512, - pre_norm=dict(type='LN'), - embed_dims=256, - feedforward_channels=1024, - num_fcs=2, - ffn_drop=0.1, - act_cfg=dict(type='ReLU', inplace=True)), - deformable_model=dict( - type='DeformableFeatureAggregation', - embed_dims=256, - num_groups=8, - num_levels=4, - num_cams=6, - attn_drop=0.15, - use_deformable_func=True, - use_camera_embed=True, - residual_mode='cat', - kps_generator=dict( - type='SparseBox3DKeyPointsGenerator', - num_learnable_pts=6, - fix_scale=[[0, 0, 0], [0.45, 0, 0], [-0.45, 0, 0], - [0, 0.45, 0], [0, -0.45, 0], [0, 0, 0.45], - [0, 0, -0.45]])), - refine_layer=dict( - type='SparseBox3DRefinementModule', - embed_dims=256, - num_cls=10, - refine_yaw=True, - with_quality_estimation=True), - sampler=dict( - type='SparseBox3DTarget', - num_dn_groups=5, - num_temp_dn_groups=3, - dn_noise_scale=[2.0, 2.0, 2.0, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5], - max_dn_gt=32, - add_neg_dn=True, - cls_weight=2.0, - box_weight=0.25, - reg_weights=[2.0, 2.0, 2.0, 0.5, 0.5, 0.5, 0.0, 0.0, 0.0, 0.0], - cls_wise_reg_weights=dict( - {9: [2.0, 2.0, 2.0, 1.0, 1.0, 1.0, 0.0, 0.0, 1.0, 1.0]})), - loss_cls=dict( - type='FocalLoss', - use_sigmoid=True, - gamma=2.0, - alpha=0.25, - loss_weight=2.0), - loss_reg=dict( - type='SparseBox3DLoss', - loss_box=dict(type='L1Loss', loss_weight=0.25), - loss_centerness=dict(type='CrossEntropyLoss', use_sigmoid=True), - loss_yawness=dict(type='GaussianFocalLoss'), - cls_allow_reverse=[5]), - decoder=dict(type='SparseBox3DDecoder'), - reg_weights=[2.0, 2.0, 2.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0])) -dataset_type = 'NuScenes3DDetTrackDataset' -data_root = 'data/nuscenes/' -anno_root = 'data/nuscenes_anno_pkls/' -file_client_args = dict(backend='disk') -img_norm_cfg = dict( - mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375], to_rgb=True) -train_pipeline = [ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict( - type='LoadPointsFromFile', - coord_type='LIDAR', - load_dim=5, - use_dim=5, - file_client_args=dict(backend='disk')), - dict(type='ResizeCropFlipImage'), - dict(type='MultiScaleDepthMapGenerator', downsample=[4, 8, 16]), - dict(type='BBoxRotation'), - dict(type='PhotoMetricDistortionMultiViewImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict( - type='CircleObjectRangeFilter', - class_dist_thred=[55, 55, 55, 55, 55, 55, 55, 55, 55, 55]), - dict( - type='InstanceNameFilter', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ]), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=[ - 'img', 'timestamp', 'projection_mat', 'image_wh', 'gt_depth', - 'focal', 'gt_bboxes_3d', 'gt_labels_3d' - ], - meta_keys=['T_global', 'T_global_inv', 'timestamp', 'instance_id']) -] -test_pipeline = [ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='ResizeCropFlipImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=['img', 'timestamp', 'projection_mat', 'image_wh'], - meta_keys=['T_global', 'T_global_inv', 'timestamp']) -] -input_modality = dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False) -data_basic_config = dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier', - 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval') -data_aug_conf = dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]) -data = dict( - samples_per_gpu=20, - workers_per_gpu=20, - train=dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval', - ann_file='data/nuscenes_anno_pkls/nuscenes_infos_train.pkl', - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict( - type='LoadPointsFromFile', - coord_type='LIDAR', - load_dim=5, - use_dim=5, - file_client_args=dict(backend='disk')), - dict(type='ResizeCropFlipImage'), - dict(type='MultiScaleDepthMapGenerator', downsample=[4, 8, 16]), - dict(type='BBoxRotation'), - dict(type='PhotoMetricDistortionMultiViewImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict( - type='CircleObjectRangeFilter', - class_dist_thred=[55, 55, 55, 55, 55, 55, 55, 55, 55, 55]), - dict( - type='InstanceNameFilter', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', - 'traffic_cone' - ]), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=[ - 'img', 'timestamp', 'projection_mat', 'image_wh', - 'gt_depth', 'focal', 'gt_bboxes_3d', 'gt_labels_3d' - ], - meta_keys=[ - 'T_global', 'T_global_inv', 'timestamp', 'instance_id' - ]) - ], - test_mode=False, - data_aug_conf=dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]), - with_seq_flag=True, - sequences_split_num=2, - keep_consistent_seq_aug=True), - val=dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval', - ann_file='data/nuscenes_anno_pkls/nuscenes_infos_val.pkl', - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='ResizeCropFlipImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=['img', 'timestamp', 'projection_mat', 'image_wh'], - meta_keys=['T_global', 'T_global_inv', 'timestamp']) - ], - data_aug_conf=dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]), - test_mode=True, - tracking=True, - tracking_threshold=0.2), - test=dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval', - ann_file='data/nuscenes_anno_pkls/nuscenes_infos_val.pkl', - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='ResizeCropFlipImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=['img', 'timestamp', 'projection_mat', 'image_wh'], - meta_keys=['T_global', 'T_global_inv', 'timestamp']) - ], - data_aug_conf=dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]), - test_mode=True, - tracking=True, - tracking_threshold=0.2)) -optimizer = dict( - type='Miopen_AdamW', - lr=0.0006, - weight_decay=0.001, - paramwise_cfg=dict(custom_keys=dict(img_backbone=dict(lr_mult=0.5)))) -optimizer_config = dict(grad_clip=dict(max_norm=25, norm_type=2)) -lr_config = dict( - policy='CosineAnnealing', - warmup='linear', - warmup_iters=500, - warmup_ratio=0.3333333333333333, - min_lr_ratio=0.001) -runner = dict(type='IterBasedRunner', max_iters=17500) -vis_pipeline = [ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='Collect', keys=['img'], meta_keys=['timestamp', 'lidar2img']) -] -evaluation = dict( - interval=3500, - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict( - type='Collect', keys=['img'], meta_keys=['timestamp', 'lidar2img']) - ]) -gpu_ids = range(0, 8) - -2025-11-13 15:31:44,987 - mmdet - INFO - Set random seed to 0, deterministic: False -2025-11-13 15:31:45,290 - mmdet - INFO - initialize ResNet with init_cfg {'type': 'Pretrained', 'checkpoint': 'ckpt/resnet50-19c8e357.pth'} -2025-11-13 15:31:45,549 - mmdet - INFO - initialize FPN with init_cfg {'type': 'Xavier', 'layer': 'Conv2d', 'distribution': 'uniform'} -Name of parameter - Initialization information - -img_backbone.conv1.weight - torch.Size([64, 3, 7, 7]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.bn1.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.bn1.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.conv1.weight - torch.Size([64, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn1.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn1.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.conv2.weight - torch.Size([64, 64, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn2.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn2.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.conv3.weight - torch.Size([256, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn3.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn3.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.downsample.0.weight - torch.Size([256, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.downsample.1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.downsample.1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.conv1.weight - torch.Size([64, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn1.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn1.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.conv2.weight - torch.Size([64, 64, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn2.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn2.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.conv3.weight - torch.Size([256, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn3.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn3.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.conv1.weight - torch.Size([64, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn1.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn1.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.conv2.weight - torch.Size([64, 64, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn2.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn2.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.conv3.weight - torch.Size([256, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn3.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn3.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.conv1.weight - torch.Size([128, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn1.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn1.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.conv2.weight - torch.Size([128, 128, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn2.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn2.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.conv3.weight - torch.Size([512, 128, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn3.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn3.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.downsample.0.weight - torch.Size([512, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.downsample.1.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.downsample.1.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.conv1.weight - torch.Size([128, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn1.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn1.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.conv2.weight - torch.Size([128, 128, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn2.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn2.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.conv3.weight - torch.Size([512, 128, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn3.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn3.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.conv1.weight - torch.Size([128, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn1.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn1.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.conv2.weight - torch.Size([128, 128, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn2.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn2.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.conv3.weight - torch.Size([512, 128, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn3.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn3.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.conv1.weight - torch.Size([128, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn1.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn1.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.conv2.weight - torch.Size([128, 128, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn2.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn2.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.conv3.weight - torch.Size([512, 128, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn3.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn3.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.conv1.weight - torch.Size([256, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.downsample.0.weight - torch.Size([1024, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.downsample.1.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.downsample.1.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.conv1.weight - torch.Size([512, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn1.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn1.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.conv2.weight - torch.Size([512, 512, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn2.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn2.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.conv3.weight - torch.Size([2048, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn3.weight - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn3.bias - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.downsample.0.weight - torch.Size([2048, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.downsample.1.weight - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.downsample.1.bias - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.conv1.weight - torch.Size([512, 2048, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn1.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn1.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.conv2.weight - torch.Size([512, 512, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn2.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn2.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.conv3.weight - torch.Size([2048, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn3.weight - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn3.bias - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.conv1.weight - torch.Size([512, 2048, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn1.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn1.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.conv2.weight - torch.Size([512, 512, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn2.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn2.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.conv3.weight - torch.Size([2048, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn3.weight - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn3.bias - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_neck.lateral_convs.0.conv.weight - torch.Size([256, 256, 1, 1]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.lateral_convs.0.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.lateral_convs.1.conv.weight - torch.Size([256, 512, 1, 1]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.lateral_convs.1.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.lateral_convs.2.conv.weight - torch.Size([256, 1024, 1, 1]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.lateral_convs.2.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.lateral_convs.3.conv.weight - torch.Size([256, 2048, 1, 1]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.lateral_convs.3.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.fpn_convs.0.conv.weight - torch.Size([256, 256, 3, 3]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.fpn_convs.0.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.fpn_convs.1.conv.weight - torch.Size([256, 256, 3, 3]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.fpn_convs.1.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.fpn_convs.2.conv.weight - torch.Size([256, 256, 3, 3]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.fpn_convs.2.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.fpn_convs.3.conv.weight - torch.Size([256, 256, 3, 3]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.fpn_convs.3.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.instance_bank.anchor - torch.Size([900, 11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.instance_bank.instance_feature - torch.Size([900, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.instance_bank.anchor_handler.fix_scale - torch.Size([1, 3]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.0.weight - torch.Size([128, 3]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.0.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.2.weight - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.2.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.3.weight - torch.Size([128, 128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.3.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.5.weight - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.5.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.6.weight - torch.Size([128, 128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.6.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.8.weight - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.8.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.9.weight - torch.Size([128, 128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.9.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.11.weight - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.11.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.0.weight - torch.Size([32, 3]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.0.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.2.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.2.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.3.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.3.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.5.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.5.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.6.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.6.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.8.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.8.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.9.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.9.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.11.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.11.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.0.weight - torch.Size([32, 2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.0.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.2.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.2.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.3.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.3.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.5.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.5.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.6.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.6.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.8.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.8.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.9.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.9.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.11.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.11.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.0.weight - torch.Size([64, 3]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.0.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.2.weight - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.2.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.3.weight - torch.Size([64, 64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.3.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.5.weight - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.5.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.6.weight - torch.Size([64, 64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.6.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.8.weight - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.8.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.9.weight - torch.Size([64, 64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.9.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.11.weight - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.11.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.1.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.1.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.1.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.1.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.1.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.1.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.1.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.1.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.3.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.4.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.4.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.4.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.4.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.5.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.5.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.5.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.5.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.6.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.6.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.8.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.8.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.8.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.8.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.8.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.8.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.8.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.8.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.10.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.11.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.11.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.11.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.11.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.12.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.12.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.12.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.12.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.13.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.13.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.15.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.15.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.15.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.15.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.15.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.15.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.15.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.15.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.16.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.16.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.17.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.18.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.18.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.18.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.18.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.19.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.19.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.19.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.19.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.20.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.20.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.22.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.22.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.22.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.22.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.22.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.22.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.22.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.22.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.23.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.23.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.24.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.25.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.25.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.25.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.25.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.26.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.26.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.26.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.26.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.27.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.27.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.29.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.29.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.29.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.29.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.29.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.29.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.29.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.29.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.30.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.30.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.31.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.32.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.32.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.32.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.32.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.33.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.33.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.33.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.33.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.34.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.34.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.36.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.36.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.36.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.36.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.36.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.36.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.36.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.36.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.37.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.37.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.38.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.fc_before.weight - torch.Size([512, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.fc_after.weight - torch.Size([256, 512]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.0.weight - torch.Size([1, 256, 1, 1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.0.bias - torch.Size([1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.1.weight - torch.Size([1, 256, 1, 1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.1.bias - torch.Size([1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.2.weight - torch.Size([1, 256, 1, 1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.2.bias - torch.Size([1]): -The value is the same before and after calling `init_weights` of Sparse4D -2025-11-13 15:31:45,639 - mmdet - INFO - Model: -Sparse4D( - (img_backbone): ResNet( - (conv1): Conv2d(3, 64, kernel_size=(7, 7), stride=(2, 2), padding=(3, 3), bias=False) - (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (maxpool): MaxPool2d(kernel_size=3, stride=2, padding=1, dilation=1, ceil_mode=False) - (layer1): ResLayer( - (0): Bottleneck( - (conv1): Conv2d(64, 64, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (downsample): Sequential( - (0): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - ) - ) - (1): Bottleneck( - (conv1): Conv2d(256, 64, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (2): Bottleneck( - (conv1): Conv2d(256, 64, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - ) - (layer2): ResLayer( - (0): Bottleneck( - (conv1): Conv2d(256, 128, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (downsample): Sequential( - (0): Conv2d(256, 512, kernel_size=(1, 1), stride=(2, 2), bias=False) - (1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - ) - ) - (1): Bottleneck( - (conv1): Conv2d(512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (2): Bottleneck( - (conv1): Conv2d(512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (3): Bottleneck( - (conv1): Conv2d(512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - ) - (layer3): ResLayer( - (0): Bottleneck( - (conv1): Conv2d(512, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (downsample): Sequential( - (0): Conv2d(512, 1024, kernel_size=(1, 1), stride=(2, 2), bias=False) - (1): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - ) - ) - (1): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (2): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (3): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (4): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (5): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - ) - (layer4): ResLayer( - (0): Bottleneck( - (conv1): Conv2d(1024, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(512, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (downsample): Sequential( - (0): Conv2d(1024, 2048, kernel_size=(1, 1), stride=(2, 2), bias=False) - (1): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - ) - ) - (1): Bottleneck( - (conv1): Conv2d(2048, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(512, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (2): Bottleneck( - (conv1): Conv2d(2048, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(512, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - ) - ) - init_cfg={'type': 'Pretrained', 'checkpoint': 'ckpt/resnet50-19c8e357.pth'} - (img_neck): FPN( - (lateral_convs): ModuleList( - (0): ConvModule( - (conv): Conv2d(256, 256, kernel_size=(1, 1), stride=(1, 1)) - ) - (1): ConvModule( - (conv): Conv2d(512, 256, kernel_size=(1, 1), stride=(1, 1)) - ) - (2): ConvModule( - (conv): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1)) - ) - (3): ConvModule( - (conv): Conv2d(2048, 256, kernel_size=(1, 1), stride=(1, 1)) - ) - ) - (fpn_convs): ModuleList( - (0-3): 4 x ConvModule( - (conv): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1)) - ) - ) - ) - init_cfg={'type': 'Xavier', 'layer': 'Conv2d', 'distribution': 'uniform'} - (head): Sparse4DHead( - (instance_bank): InstanceBank( - (anchor_handler): SparseBox3DKeyPointsGenerator() - ) - (anchor_encoder): SparseBox3DEncoder( - (pos_fc): Sequential( - (0): Linear(in_features=3, out_features=128, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((128,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=128, out_features=128, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((128,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=128, out_features=128, bias=True) - (7): ReLU(inplace=True) - (8): LayerNorm((128,), eps=1e-05, elementwise_affine=True) - (9): Linear(in_features=128, out_features=128, bias=True) - (10): ReLU(inplace=True) - (11): LayerNorm((128,), eps=1e-05, elementwise_affine=True) - ) - (size_fc): Sequential( - (0): Linear(in_features=3, out_features=32, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=32, out_features=32, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=32, out_features=32, bias=True) - (7): ReLU(inplace=True) - (8): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (9): Linear(in_features=32, out_features=32, bias=True) - (10): ReLU(inplace=True) - (11): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - ) - (yaw_fc): Sequential( - (0): Linear(in_features=2, out_features=32, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=32, out_features=32, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=32, out_features=32, bias=True) - (7): ReLU(inplace=True) - (8): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (9): Linear(in_features=32, out_features=32, bias=True) - (10): ReLU(inplace=True) - (11): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - ) - (vel_fc): Sequential( - (0): Linear(in_features=3, out_features=64, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((64,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=64, out_features=64, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((64,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=64, out_features=64, bias=True) - (7): ReLU(inplace=True) - (8): LayerNorm((64,), eps=1e-05, elementwise_affine=True) - (9): Linear(in_features=64, out_features=64, bias=True) - (10): ReLU(inplace=True) - (11): LayerNorm((64,), eps=1e-05, elementwise_affine=True) - ) - ) - (loss_cls): FocalLoss() - (loss_reg): SparseBox3DLoss( - (loss_box): L1Loss() - (loss_cns): CrossEntropyLoss(avg_non_ignore=False) - (loss_yns): GaussianFocalLoss() - ) - (layers): ModuleList( - (0): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (1): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (4-5): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (6): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (7): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (8): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (11-12): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (13): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (14): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (15): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (16): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (17): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (18-19): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (20): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (21): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (22): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (23): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (24): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (25-26): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (27): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (28): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (29): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (30): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (31): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (32-33): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (34): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (35): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (36): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (37): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (38): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - ) - (fc_before): Linear(in_features=256, out_features=512, bias=False) - (fc_after): Linear(in_features=512, out_features=256, bias=False) - ) - (depth_branch): DenseDepthNet( - (depth_layers): ModuleList( - (0-2): 3 x Conv2d(256, 1, kernel_size=(1, 1), stride=(1, 1)) - ) - ) - (grid_mask): GridMask() -) -2025-11-13 15:31:58,180 - mmdet - INFO - Start running, host: root@VM-120-96-tencentos, work_dir: /cfs/auto/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704 -2025-11-13 15:31:58,181 - mmdet - INFO - Hooks will be executed in the following order: -before_run: -(VERY_HIGH ) CosineAnnealingLrUpdaterHook -(ABOVE_NORMAL) Fp16OptimizerHook -(NORMAL ) CheckpointHook -(NORMAL ) CustomDistEvalHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -before_train_epoch: -(VERY_HIGH ) CosineAnnealingLrUpdaterHook -(NORMAL ) CustomDistEvalHook -(LOW ) IterTimerHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -before_train_iter: -(VERY_HIGH ) CosineAnnealingLrUpdaterHook -(NORMAL ) CustomDistEvalHook -(LOW ) IterTimerHook - -------------------- -after_train_iter: -(ABOVE_NORMAL) Fp16OptimizerHook -(NORMAL ) CheckpointHook -(NORMAL ) CustomDistEvalHook -(LOW ) IterTimerHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -after_train_epoch: -(NORMAL ) CheckpointHook -(NORMAL ) CustomDistEvalHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -before_val_epoch: -(LOW ) IterTimerHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -before_val_iter: -(LOW ) IterTimerHook - -------------------- -after_val_iter: -(LOW ) IterTimerHook - -------------------- -after_val_epoch: -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -after_run: -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -2025-11-13 15:31:58,181 - mmdet - INFO - workflow: [('train', 1)], max: 17500 iters -2025-11-13 15:31:58,183 - mmdet - INFO - Checkpoints will be saved to /cfs/auto/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704 by HardDiskBackend. -2025-11-13 15:33:54,591 - mmdet - INFO - Iter [1/17500] lr: 1.000e-04, eta: 23 days, 9:42:26, time: 115.558, data_time: 10.995, memory: 49163, loss_cls_0: 2.3610, loss_box_0: 0.0138, loss_cns_0: 0.0027, loss_yns_0: 0.0008, loss_cls_1: 2.1545, loss_box_1: 0.1071, loss_cns_1: 0.0243, loss_yns_1: 0.0066, loss_cls_2: 2.3120, loss_box_2: 0.0050, loss_cns_2: 0.0006, loss_yns_2: 0.0003, loss_cls_3: 2.3900, loss_box_3: 0.0295, loss_cns_3: 0.0050, loss_yns_3: 0.0014, loss_cls_4: 2.0282, loss_box_4: 0.4187, loss_cns_4: 0.0537, loss_yns_4: 0.0252, loss_cls_5: 2.4248, loss_box_5: 0.0191, loss_cns_5: 0.0024, loss_yns_5: 0.0017, loss_cls_dn_0: 1.1980, loss_box_dn_0: 1.4603, loss_cls_dn_1: 1.1102, loss_box_dn_1: 1.7318, loss_cls_dn_2: 1.1741, loss_box_dn_2: 1.9719, loss_cls_dn_3: 1.1721, loss_box_dn_3: 2.2418, loss_cls_dn_4: 1.0528, loss_box_dn_4: 2.4268, loss_cls_dn_5: 1.2387, loss_box_dn_5: 2.6773, loss_dense_depth: 1.8643, loss: 35.7084, grad_norm: 273.3050 -2025-11-13 15:33:56,552 - mmdet - INFO - Iter [2/17500] lr: 1.004e-04, eta: 11 days, 21:35:57, time: 1.959, data_time: 0.080, memory: 49163, loss_cls_0: 2.0556, loss_box_0: 0.0234, loss_cns_0: 0.0060, loss_yns_0: 0.0025, loss_cls_1: 2.0240, loss_box_1: 0.1440, loss_cns_1: 0.0255, loss_yns_1: 0.0077, loss_cls_2: 2.1057, loss_box_2: 0.2128, loss_cns_2: 0.0197, loss_yns_2: 0.0082, loss_cls_3: 1.9365, loss_box_3: 0.4639, loss_cns_3: 0.0612, loss_yns_3: 0.0210, loss_cls_4: 1.7994, loss_box_4: 1.5556, loss_cns_4: 0.1545, loss_yns_4: 0.0565, loss_cls_5: 2.0486, loss_box_5: 0.5749, loss_cns_5: 0.0629, loss_yns_5: 0.0196, loss_cls_dn_0: 1.0314, loss_box_dn_0: 1.2550, loss_cls_dn_1: 0.9545, loss_box_dn_1: 2.4088, loss_cls_dn_2: 0.9733, loss_box_dn_2: 2.5322, loss_cls_dn_3: 0.9077, loss_box_dn_3: 2.6107, loss_cls_dn_4: 0.8420, loss_box_dn_4: 2.8655, loss_cls_dn_5: 0.9861, loss_box_dn_5: 3.1056, loss_dense_depth: 1.7117, loss: 37.5741, grad_norm: 66.6678 -2025-11-13 15:33:58,069 - mmdet - INFO - Iter [3/17500] lr: 1.008e-04, eta: 8 days, 0:50:36, time: 1.515, data_time: 0.080, memory: 49163, loss_cls_0: 1.4134, loss_box_0: 2.7913, loss_cns_0: 0.6225, loss_yns_0: 0.2149, loss_cls_1: 1.7284, loss_box_1: 2.4857, loss_cns_1: 0.3514, loss_yns_1: 0.1318, loss_cls_2: 1.7499, loss_box_2: 4.4120, loss_cns_2: 0.3749, loss_yns_2: 0.2071, loss_cls_3: 1.5999, loss_box_3: 5.2129, loss_cns_3: 0.4558, loss_yns_3: 0.2134, loss_cls_4: 1.5143, loss_box_4: 5.1283, loss_cns_4: 0.4273, loss_yns_4: 0.2063, loss_cls_5: 1.6220, loss_box_5: 3.9920, loss_cns_5: 0.2873, loss_yns_5: 0.1296, loss_cls_dn_0: 0.6730, loss_box_dn_0: 1.2771, loss_cls_dn_1: 0.8058, loss_box_dn_1: 2.3997, loss_cls_dn_2: 0.7691, loss_box_dn_2: 2.6260, loss_cls_dn_3: 0.6697, loss_box_dn_3: 2.8185, loss_cls_dn_4: 0.6853, loss_box_dn_4: 3.0807, loss_cls_dn_5: 0.7753, loss_box_dn_5: 3.3671, loss_dense_depth: 1.6421, loss: 58.8616, grad_norm: 111.5939 -2025-11-13 15:33:59,571 - mmdet - INFO - Iter [4/17500] lr: 1.012e-04, eta: 6 days, 2:27:12, time: 1.505, data_time: 0.080, memory: 49163, loss_cls_0: 1.3976, loss_box_0: 2.5862, loss_cns_0: 0.5401, loss_yns_0: 0.2084, loss_cls_1: 1.5554, loss_box_1: 3.7094, loss_cns_1: 0.4545, loss_yns_1: 0.2245, loss_cls_2: 1.6974, loss_box_2: 3.8338, loss_cns_2: 0.4466, loss_yns_2: 0.2085, loss_cls_3: 1.4649, loss_box_3: 4.2736, loss_cns_3: 0.4590, loss_yns_3: 0.1987, loss_cls_4: 1.4906, loss_box_4: 4.6657, loss_cns_4: 0.4197, loss_yns_4: 0.1952, loss_cls_5: 1.4044, loss_box_5: 5.1883, loss_cns_5: 0.4363, loss_yns_5: 0.2013, loss_cls_dn_0: 0.5249, loss_box_dn_0: 1.2323, loss_cls_dn_1: 0.6637, loss_box_dn_1: 2.5984, loss_cls_dn_2: 0.6475, loss_box_dn_2: 2.7212, loss_cls_dn_3: 0.5828, loss_box_dn_3: 2.9838, loss_cls_dn_4: 0.5417, loss_box_dn_4: 3.2147, loss_cls_dn_5: 0.5956, loss_box_dn_5: 3.4518, loss_dense_depth: 1.7341, loss: 58.7524, grad_norm: 114.0531 -2025-11-13 15:34:01,111 - mmdet - INFO - Iter [5/17500] lr: 1.016e-04, eta: 4 days, 22:39:09, time: 1.540, data_time: 0.081, memory: 49163, loss_cls_0: 1.3759, loss_box_0: 2.7667, loss_cns_0: 0.5580, loss_yns_0: 0.1987, loss_cls_1: 1.5302, loss_box_1: 4.2303, loss_cns_1: 0.4065, loss_yns_1: 0.2178, loss_cls_2: 1.4521, loss_box_2: 4.2222, loss_cns_2: 0.3930, loss_yns_2: 0.1896, loss_cls_3: 1.3716, loss_box_3: 4.3039, loss_cns_3: 0.3818, loss_yns_3: 0.2158, loss_cls_4: 1.3347, loss_box_4: 4.6129, loss_cns_4: 0.3414, loss_yns_4: 0.2095, loss_cls_5: 1.3502, loss_box_5: 4.7917, loss_cns_5: 0.3311, loss_yns_5: 0.2115, loss_cls_dn_0: 0.5524, loss_box_dn_0: 1.2473, loss_cls_dn_1: 0.5830, loss_box_dn_1: 2.4446, loss_cls_dn_2: 0.6194, loss_box_dn_2: 2.5643, loss_cls_dn_3: 0.5179, loss_box_dn_3: 2.6632, loss_cls_dn_4: 0.5387, loss_box_dn_4: 2.9146, loss_cls_dn_5: 0.5024, loss_box_dn_5: 2.9571, loss_dense_depth: 1.9076, loss: 57.0097, grad_norm: 125.0766 -2025-11-13 15:34:02,624 - mmdet - INFO - Iter [6/17500] lr: 1.020e-04, eta: 4 days, 4:05:49, time: 1.513, data_time: 0.079, memory: 49163, loss_cls_0: 1.2577, loss_box_0: 2.3655, loss_cns_0: 0.6837, loss_yns_0: 0.1851, loss_cls_1: 1.3278, loss_box_1: 3.7018, loss_cns_1: 0.4756, loss_yns_1: 0.1930, loss_cls_2: 1.3424, loss_box_2: 3.7290, loss_cns_2: 0.4792, loss_yns_2: 0.1861, loss_cls_3: 1.3368, loss_box_3: 3.6119, loss_cns_3: 0.5060, loss_yns_3: 0.1867, loss_cls_4: 1.2959, loss_box_4: 3.9520, loss_cns_4: 0.4623, loss_yns_4: 0.1846, loss_cls_5: 1.3566, loss_box_5: 4.3228, loss_cns_5: 0.4319, loss_yns_5: 0.1952, loss_cls_dn_0: 0.5596, loss_box_dn_0: 1.1394, loss_cls_dn_1: 0.5118, loss_box_dn_1: 2.5260, loss_cls_dn_2: 0.5430, loss_box_dn_2: 2.4745, loss_cls_dn_3: 0.4855, loss_box_dn_3: 2.5256, loss_cls_dn_4: 0.4765, loss_box_dn_4: 2.7684, loss_cls_dn_5: 0.4370, loss_box_dn_5: 2.9192, loss_dense_depth: 1.7859, loss: 52.9220, grad_norm: 115.3840 -2025-11-13 15:34:04,108 - mmdet - INFO - Iter [7/17500] lr: 1.024e-04, eta: 3 days, 14:49:18, time: 1.483, data_time: 0.077, memory: 49163, loss_cls_0: 1.2530, loss_box_0: 2.3083, loss_cns_0: 0.6326, loss_yns_0: 0.1837, loss_cls_1: 1.2613, loss_box_1: 3.7170, loss_cns_1: 0.4546, loss_yns_1: 0.1836, loss_cls_2: 1.4248, loss_box_2: 3.7220, loss_cns_2: 0.4025, loss_yns_2: 0.1827, loss_cls_3: 1.2784, loss_box_3: 3.6005, loss_cns_3: 0.4495, loss_yns_3: 0.1839, loss_cls_4: 1.3173, loss_box_4: 3.6501, loss_cns_4: 0.4515, loss_yns_4: 0.1893, loss_cls_5: 1.3357, loss_box_5: 3.8667, loss_cns_5: 0.4932, loss_yns_5: 0.1855, loss_cls_dn_0: 0.5209, loss_box_dn_0: 1.0539, loss_cls_dn_1: 0.4632, loss_box_dn_1: 2.4456, loss_cls_dn_2: 0.4641, loss_box_dn_2: 2.3740, loss_cls_dn_3: 0.4467, loss_box_dn_3: 2.3725, loss_cls_dn_4: 0.4276, loss_box_dn_4: 2.4258, loss_cls_dn_5: 0.3958, loss_box_dn_5: 2.5152, loss_dense_depth: 1.9392, loss: 50.5723, grad_norm: 86.5918 -2025-11-13 15:34:05,618 - mmdet - INFO - Iter [8/17500] lr: 1.028e-04, eta: 3 days, 4:52:55, time: 1.511, data_time: 0.082, memory: 49163, loss_cls_0: 1.2686, loss_box_0: 2.2942, loss_cns_0: 0.6144, loss_yns_0: 0.1789, loss_cls_1: 1.2770, loss_box_1: 3.3836, loss_cns_1: 0.5029, loss_yns_1: 0.1803, loss_cls_2: 1.3054, loss_box_2: 3.3635, loss_cns_2: 0.4427, loss_yns_2: 0.1824, loss_cls_3: 1.2696, loss_box_3: 3.4625, loss_cns_3: 0.4756, loss_yns_3: 0.1955, loss_cls_4: 1.2476, loss_box_4: 3.4086, loss_cns_4: 0.5021, loss_yns_4: 0.1840, loss_cls_5: 1.2791, loss_box_5: 3.4134, loss_cns_5: 0.5576, loss_yns_5: 0.1943, loss_cls_dn_0: 0.4754, loss_box_dn_0: 1.0434, loss_cls_dn_1: 0.4651, loss_box_dn_1: 1.5743, loss_cls_dn_2: 0.5001, loss_box_dn_2: 1.6280, loss_cls_dn_3: 0.4356, loss_box_dn_3: 1.7691, loss_cls_dn_4: 0.4521, loss_box_dn_4: 1.7603, loss_cls_dn_5: 0.4533, loss_box_dn_5: 1.8718, loss_dense_depth: 2.1584, loss: 45.7709, grad_norm: 85.3385 -2025-11-13 15:34:07,131 - mmdet - INFO - Iter [9/17500] lr: 1.032e-04, eta: 2 days, 21:09:07, time: 1.512, data_time: 0.080, memory: 49163, loss_cls_0: 1.2341, loss_box_0: 2.3211, loss_cns_0: 0.6299, loss_yns_0: 0.1776, loss_cls_1: 1.2641, loss_box_1: 3.0627, loss_cns_1: 0.4948, loss_yns_1: 0.1910, loss_cls_2: 1.2525, loss_box_2: 3.0330, loss_cns_2: 0.5076, loss_yns_2: 0.1849, loss_cls_3: 1.2533, loss_box_3: 3.1835, loss_cns_3: 0.5101, loss_yns_3: 0.1810, loss_cls_4: 1.2544, loss_box_4: 3.2169, loss_cns_4: 0.4902, loss_yns_4: 0.1897, loss_cls_5: 1.2523, loss_box_5: 3.7742, loss_cns_5: 0.4737, loss_yns_5: 0.1810, loss_cls_dn_0: 0.4896, loss_box_dn_0: 1.0517, loss_cls_dn_1: 0.4303, loss_box_dn_1: 1.5695, loss_cls_dn_2: 0.4851, loss_box_dn_2: 1.7022, loss_cls_dn_3: 0.4297, loss_box_dn_3: 1.8660, loss_cls_dn_4: 0.4307, loss_box_dn_4: 1.8868, loss_cls_dn_5: 0.4629, loss_box_dn_5: 2.2965, loss_dense_depth: 2.1580, loss: 45.5724, grad_norm: 85.8332 -2025-11-13 15:34:08,628 - mmdet - INFO - Iter [10/17500] lr: 1.036e-04, eta: 2 days, 14:57:40, time: 1.498, data_time: 0.081, memory: 49163, loss_cls_0: 1.2425, loss_box_0: 2.3688, loss_cns_0: 0.6227, loss_yns_0: 0.1743, loss_cls_1: 1.2822, loss_box_1: 3.1470, loss_cns_1: 0.4638, loss_yns_1: 0.1811, loss_cls_2: 1.2275, loss_box_2: 3.0992, loss_cns_2: 0.4525, loss_yns_2: 0.1821, loss_cls_3: 1.2109, loss_box_3: 3.1210, loss_cns_3: 0.4689, loss_yns_3: 0.1737, loss_cls_4: 1.2515, loss_box_4: 3.1642, loss_cns_4: 0.4638, loss_yns_4: 0.1797, loss_cls_5: 1.2616, loss_box_5: 3.4433, loss_cns_5: 0.4761, loss_yns_5: 0.1749, loss_cls_dn_0: 0.4841, loss_box_dn_0: 1.0674, loss_cls_dn_1: 0.4024, loss_box_dn_1: 1.8894, loss_cls_dn_2: 0.4508, loss_box_dn_2: 1.8590, loss_cls_dn_3: 0.4416, loss_box_dn_3: 1.8954, loss_cls_dn_4: 0.4019, loss_box_dn_4: 1.9403, loss_cls_dn_5: 0.4297, loss_box_dn_5: 2.1501, loss_dense_depth: 2.4524, loss: 45.6978, grad_norm: 83.6780 -2025-11-13 15:34:10,108 - mmdet - INFO - Iter [11/17500] lr: 1.040e-04, eta: 2 days, 9:53:16, time: 1.480, data_time: 0.081, memory: 49163, loss_cls_0: 1.2126, loss_box_0: 2.3315, loss_cns_0: 0.6007, loss_yns_0: 0.1747, loss_cls_1: 1.2528, loss_box_1: 2.9545, loss_cns_1: 0.5217, loss_yns_1: 0.1828, loss_cls_2: 1.2298, loss_box_2: 3.0694, loss_cns_2: 0.5007, loss_yns_2: 0.1904, loss_cls_3: 1.2310, loss_box_3: 3.0868, loss_cns_3: 0.5204, loss_yns_3: 0.1771, loss_cls_4: 1.2738, loss_box_4: 3.2061, loss_cns_4: 0.4593, loss_yns_4: 0.1812, loss_cls_5: 1.2982, loss_box_5: 3.5345, loss_cns_5: 0.4701, loss_yns_5: 0.1738, loss_cls_dn_0: 0.4508, loss_box_dn_0: 1.0697, loss_cls_dn_1: 0.3941, loss_box_dn_1: 1.9727, loss_cls_dn_2: 0.4258, loss_box_dn_2: 1.9198, loss_cls_dn_3: 0.4194, loss_box_dn_3: 1.9163, loss_cls_dn_4: 0.3958, loss_box_dn_4: 2.0410, loss_cls_dn_5: 0.4064, loss_box_dn_5: 2.1210, loss_dense_depth: 2.2203, loss: 45.5871, grad_norm: 79.4470 -2025-11-13 15:34:11,597 - mmdet - INFO - Iter [12/17500] lr: 1.044e-04, eta: 2 days, 5:39:49, time: 1.489, data_time: 0.080, memory: 49163, loss_cls_0: 1.2471, loss_box_0: 2.3212, loss_cns_0: 0.5908, loss_yns_0: 0.1727, loss_cls_1: 1.2425, loss_box_1: 2.9994, loss_cns_1: 0.5278, loss_yns_1: 0.1726, loss_cls_2: 1.2459, loss_box_2: 3.2447, loss_cns_2: 0.5094, loss_yns_2: 0.1854, loss_cls_3: 1.2643, loss_box_3: 3.2968, loss_cns_3: 0.4954, loss_yns_3: 0.1746, loss_cls_4: 1.2212, loss_box_4: 3.4783, loss_cns_4: 0.4706, loss_yns_4: 0.1834, loss_cls_5: 1.2440, loss_box_5: 3.5985, loss_cns_5: 0.4202, loss_yns_5: 0.1758, loss_cls_dn_0: 0.4459, loss_box_dn_0: 1.0709, loss_cls_dn_1: 0.3981, loss_box_dn_1: 2.0634, loss_cls_dn_2: 0.3936, loss_box_dn_2: 2.0679, loss_cls_dn_3: 0.3773, loss_box_dn_3: 2.0628, loss_cls_dn_4: 0.3953, loss_box_dn_4: 2.1831, loss_cls_dn_5: 0.4004, loss_box_dn_5: 2.2145, loss_dense_depth: 2.1753, loss: 46.7309, grad_norm: 86.8314 -2025-11-13 15:34:13,094 - mmdet - INFO - Iter [13/17500] lr: 1.048e-04, eta: 2 days, 2:05:30, time: 1.495, data_time: 0.078, memory: 49163, loss_cls_0: 1.1664, loss_box_0: 2.3587, loss_cns_0: 0.5579, loss_yns_0: 0.1715, loss_cls_1: 1.2217, loss_box_1: 3.0542, loss_cns_1: 0.5306, loss_yns_1: 0.1712, loss_cls_2: 1.2571, loss_box_2: 3.0828, loss_cns_2: 0.5394, loss_yns_2: 0.1890, loss_cls_3: 1.2570, loss_box_3: 3.0942, loss_cns_3: 0.5371, loss_yns_3: 0.1754, loss_cls_4: 1.2184, loss_box_4: 3.1878, loss_cns_4: 0.5383, loss_yns_4: 0.1800, loss_cls_5: 1.2265, loss_box_5: 3.2220, loss_cns_5: 0.5316, loss_yns_5: 0.1732, loss_cls_dn_0: 0.4743, loss_box_dn_0: 1.0691, loss_cls_dn_1: 0.4520, loss_box_dn_1: 1.6965, loss_cls_dn_2: 0.4225, loss_box_dn_2: 1.7220, loss_cls_dn_3: 0.4055, loss_box_dn_3: 1.7638, loss_cls_dn_4: 0.4310, loss_box_dn_4: 1.9258, loss_cls_dn_5: 0.4536, loss_box_dn_5: 1.9676, loss_dense_depth: 2.3750, loss: 44.8006, grad_norm: 92.7331 -2025-11-13 15:34:14,597 - mmdet - INFO - Iter [14/17500] lr: 1.052e-04, eta: 1 day, 23:01:59, time: 1.505, data_time: 0.080, memory: 49163, loss_cls_0: 1.1698, loss_box_0: 2.4698, loss_cns_0: 0.5419, loss_yns_0: 0.1719, loss_cls_1: 1.2323, loss_box_1: 2.8053, loss_cns_1: 0.5555, loss_yns_1: 0.1740, loss_cls_2: 1.2889, loss_box_2: 2.8667, loss_cns_2: 0.5374, loss_yns_2: 0.1844, loss_cls_3: 1.2638, loss_box_3: 3.0747, loss_cns_3: 0.5521, loss_yns_3: 0.1749, loss_cls_4: 1.2782, loss_box_4: 2.9514, loss_cns_4: 0.5601, loss_yns_4: 0.1836, loss_cls_5: 1.2510, loss_box_5: 2.8564, loss_cns_5: 0.5723, loss_yns_5: 0.1752, loss_cls_dn_0: 0.4922, loss_box_dn_0: 1.0867, loss_cls_dn_1: 0.4688, loss_box_dn_1: 1.4187, loss_cls_dn_2: 0.4374, loss_box_dn_2: 1.4397, loss_cls_dn_3: 0.4317, loss_box_dn_3: 1.5861, loss_cls_dn_4: 0.4328, loss_box_dn_4: 1.6142, loss_cls_dn_5: 0.4617, loss_box_dn_5: 1.6727, loss_dense_depth: 2.0732, loss: 42.5074, grad_norm: 80.2924 -2025-11-13 15:34:16,071 - mmdet - INFO - Iter [15/17500] lr: 1.056e-04, eta: 1 day, 20:22:18, time: 1.472, data_time: 0.078, memory: 49163, loss_cls_0: 1.1814, loss_box_0: 2.5390, loss_cns_0: 0.5635, loss_yns_0: 0.1719, loss_cls_1: 1.2449, loss_box_1: 2.5555, loss_cns_1: 0.5957, loss_yns_1: 0.1758, loss_cls_2: 1.2987, loss_box_2: 2.6573, loss_cns_2: 0.5814, loss_yns_2: 0.1764, loss_cls_3: 1.2603, loss_box_3: 2.8082, loss_cns_3: 0.6096, loss_yns_3: 0.1756, loss_cls_4: 1.2715, loss_box_4: 2.6902, loss_cns_4: 0.6134, loss_yns_4: 0.1856, loss_cls_5: 1.2715, loss_box_5: 2.6813, loss_cns_5: 0.5975, loss_yns_5: 0.1765, loss_cls_dn_0: 0.4642, loss_box_dn_0: 1.1089, loss_cls_dn_1: 0.4581, loss_box_dn_1: 1.3919, loss_cls_dn_2: 0.4286, loss_box_dn_2: 1.4699, loss_cls_dn_3: 0.4253, loss_box_dn_3: 1.6174, loss_cls_dn_4: 0.4455, loss_box_dn_4: 1.5813, loss_cls_dn_5: 0.4402, loss_box_dn_5: 1.7021, loss_dense_depth: 2.0431, loss: 41.6593, grad_norm: 80.1467 -2025-11-13 15:34:17,555 - mmdet - INFO - Iter [16/17500] lr: 1.060e-04, eta: 1 day, 18:02:47, time: 1.484, data_time: 0.080, memory: 49163, loss_cls_0: 1.2079, loss_box_0: 2.4230, loss_cns_0: 0.5878, loss_yns_0: 0.1719, loss_cls_1: 1.2401, loss_box_1: 2.5730, loss_cns_1: 0.5819, loss_yns_1: 0.1775, loss_cls_2: 1.2602, loss_box_2: 2.6407, loss_cns_2: 0.5820, loss_yns_2: 0.1800, loss_cls_3: 1.2367, loss_box_3: 2.6312, loss_cns_3: 0.5867, loss_yns_3: 0.1806, loss_cls_4: 1.2473, loss_box_4: 2.6519, loss_cns_4: 0.5840, loss_yns_4: 0.1766, loss_cls_5: 1.2608, loss_box_5: 2.6167, loss_cns_5: 0.5722, loss_yns_5: 0.1759, loss_cls_dn_0: 0.4524, loss_box_dn_0: 1.0572, loss_cls_dn_1: 0.4593, loss_box_dn_1: 1.3150, loss_cls_dn_2: 0.4330, loss_box_dn_2: 1.3930, loss_cls_dn_3: 0.4480, loss_box_dn_3: 1.5004, loss_cls_dn_4: 0.4739, loss_box_dn_4: 1.5075, loss_cls_dn_5: 0.4371, loss_box_dn_5: 1.6120, loss_dense_depth: 1.9480, loss: 40.5836, grad_norm: 69.0845 -2025-11-13 15:34:19,046 - mmdet - INFO - Iter [17/17500] lr: 1.064e-04, eta: 1 day, 15:59:49, time: 1.491, data_time: 0.079, memory: 49163, loss_cls_0: 1.1642, loss_box_0: 2.2692, loss_cns_0: 0.6024, loss_yns_0: 0.1708, loss_cls_1: 1.2518, loss_box_1: 2.6355, loss_cns_1: 0.5801, loss_yns_1: 0.1754, loss_cls_2: 1.2662, loss_box_2: 2.6716, loss_cns_2: 0.5733, loss_yns_2: 0.1805, loss_cls_3: 1.2532, loss_box_3: 2.6922, loss_cns_3: 0.5880, loss_yns_3: 0.1942, loss_cls_4: 1.2489, loss_box_4: 2.7760, loss_cns_4: 0.5791, loss_yns_4: 0.1760, loss_cls_5: 1.2537, loss_box_5: 2.8255, loss_cns_5: 0.5751, loss_yns_5: 0.1767, loss_cls_dn_0: 0.4523, loss_box_dn_0: 1.0160, loss_cls_dn_1: 0.4362, loss_box_dn_1: 1.3434, loss_cls_dn_2: 0.4179, loss_box_dn_2: 1.4005, loss_cls_dn_3: 0.4251, loss_box_dn_3: 1.4893, loss_cls_dn_4: 0.4324, loss_box_dn_4: 1.5361, loss_cls_dn_5: 0.4242, loss_box_dn_5: 1.6396, loss_dense_depth: 1.8963, loss: 40.7888, grad_norm: 63.6702 -2025-11-13 15:34:20,536 - mmdet - INFO - Iter [18/17500] lr: 1.068e-04, eta: 1 day, 14:10:30, time: 1.491, data_time: 0.080, memory: 49163, loss_cls_0: 1.1355, loss_box_0: 2.1483, loss_cns_0: 0.6161, loss_yns_0: 0.1709, loss_cls_1: 1.2568, loss_box_1: 2.7479, loss_cns_1: 0.5779, loss_yns_1: 0.1780, loss_cls_2: 1.2464, loss_box_2: 2.7608, loss_cns_2: 0.5713, loss_yns_2: 0.1792, loss_cls_3: 1.2434, loss_box_3: 2.7937, loss_cns_3: 0.6019, loss_yns_3: 0.1768, loss_cls_4: 1.2728, loss_box_4: 2.8449, loss_cns_4: 0.5854, loss_yns_4: 0.1799, loss_cls_5: 1.2419, loss_box_5: 3.0438, loss_cns_5: 0.5841, loss_yns_5: 0.1753, loss_cls_dn_0: 0.4848, loss_box_dn_0: 0.9903, loss_cls_dn_1: 0.4198, loss_box_dn_1: 1.3332, loss_cls_dn_2: 0.4343, loss_box_dn_2: 1.3678, loss_cls_dn_3: 0.4287, loss_box_dn_3: 1.4520, loss_cls_dn_4: 0.4091, loss_box_dn_4: 1.5182, loss_cls_dn_5: 0.4241, loss_box_dn_5: 1.6745, loss_dense_depth: 1.6611, loss: 40.9309, grad_norm: 81.7362 -2025-11-13 15:34:22,026 - mmdet - INFO - Iter [19/17500] lr: 1.072e-04, eta: 1 day, 12:32:38, time: 1.488, data_time: 0.079, memory: 49163, loss_cls_0: 1.1449, loss_box_0: 2.1890, loss_cns_0: 0.6167, loss_yns_0: 0.1681, loss_cls_1: 1.2615, loss_box_1: 2.7842, loss_cns_1: 0.5874, loss_yns_1: 0.1779, loss_cls_2: 1.2567, loss_box_2: 2.6715, loss_cns_2: 0.5903, loss_yns_2: 0.1781, loss_cls_3: 1.2514, loss_box_3: 2.6364, loss_cns_3: 0.6130, loss_yns_3: 0.1734, loss_cls_4: 1.2640, loss_box_4: 2.6779, loss_cns_4: 0.5939, loss_yns_4: 0.1706, loss_cls_5: 1.2573, loss_box_5: 2.8063, loss_cns_5: 0.5851, loss_yns_5: 0.1776, loss_cls_dn_0: 0.4981, loss_box_dn_0: 1.0053, loss_cls_dn_1: 0.4268, loss_box_dn_1: 1.3039, loss_cls_dn_2: 0.4567, loss_box_dn_2: 1.2937, loss_cls_dn_3: 0.4466, loss_box_dn_3: 1.3578, loss_cls_dn_4: 0.4347, loss_box_dn_4: 1.4779, loss_cls_dn_5: 0.4365, loss_box_dn_5: 1.6082, loss_dense_depth: 1.5214, loss: 40.1009, grad_norm: 70.3858 -2025-11-13 15:34:23,508 - mmdet - INFO - Iter [20/17500] lr: 1.076e-04, eta: 1 day, 11:04:29, time: 1.483, data_time: 0.079, memory: 49163, loss_cls_0: 1.1470, loss_box_0: 2.2218, loss_cns_0: 0.6218, loss_yns_0: 0.1694, loss_cls_1: 1.2603, loss_box_1: 2.7152, loss_cns_1: 0.5970, loss_yns_1: 0.1746, loss_cls_2: 1.2630, loss_box_2: 2.6228, loss_cns_2: 0.5967, loss_yns_2: 0.1814, loss_cls_3: 1.2491, loss_box_3: 2.6686, loss_cns_3: 0.6017, loss_yns_3: 0.1769, loss_cls_4: 1.2557, loss_box_4: 2.7083, loss_cns_4: 0.5904, loss_yns_4: 0.1784, loss_cls_5: 1.2661, loss_box_5: 2.7391, loss_cns_5: 0.5858, loss_yns_5: 0.1764, loss_cls_dn_0: 0.4602, loss_box_dn_0: 1.0039, loss_cls_dn_1: 0.3970, loss_box_dn_1: 1.3494, loss_cls_dn_2: 0.4174, loss_box_dn_2: 1.3441, loss_cls_dn_3: 0.4045, loss_box_dn_3: 1.3867, loss_cls_dn_4: 0.4179, loss_box_dn_4: 1.4675, loss_cls_dn_5: 0.4080, loss_box_dn_5: 1.5179, loss_dense_depth: 1.4831, loss: 39.8253, grad_norm: 52.6002 -2025-11-13 15:34:25,088 - mmdet - INFO - Iter [21/17500] lr: 1.080e-04, eta: 1 day, 9:46:05, time: 1.581, data_time: 0.097, memory: 49163, loss_cls_0: 1.1857, loss_box_0: 2.1879, loss_cns_0: 0.6221, loss_yns_0: 0.1679, loss_cls_1: 1.2332, loss_box_1: 2.7931, loss_cns_1: 0.5871, loss_yns_1: 0.1758, loss_cls_2: 1.2640, loss_box_2: 2.8123, loss_cns_2: 0.5839, loss_yns_2: 0.1763, loss_cls_3: 1.2447, loss_box_3: 2.9034, loss_cns_3: 0.5944, loss_yns_3: 0.1742, loss_cls_4: 1.2485, loss_box_4: 3.0020, loss_cns_4: 0.5896, loss_yns_4: 0.1707, loss_cls_5: 1.2508, loss_box_5: 3.0383, loss_cns_5: 0.5888, loss_yns_5: 0.1712, loss_cls_dn_0: 0.4254, loss_box_dn_0: 1.0239, loss_cls_dn_1: 0.4245, loss_box_dn_1: 1.2838, loss_cls_dn_2: 0.4301, loss_box_dn_2: 1.3672, loss_cls_dn_3: 0.4142, loss_box_dn_3: 1.4456, loss_cls_dn_4: 0.4424, loss_box_dn_4: 1.6049, loss_cls_dn_5: 0.4407, loss_box_dn_5: 1.6187, loss_dense_depth: 1.4227, loss: 41.1099, grad_norm: 94.1380 -2025-11-13 15:34:26,682 - mmdet - INFO - Iter [22/17500] lr: 1.084e-04, eta: 1 day, 8:34:59, time: 1.593, data_time: 0.185, memory: 49163, loss_cls_0: 1.2243, loss_box_0: 2.1820, loss_cns_0: 0.6208, loss_yns_0: 0.1667, loss_cls_1: 1.2192, loss_box_1: 2.7875, loss_cns_1: 0.5707, loss_yns_1: 0.1759, loss_cls_2: 1.2578, loss_box_2: 2.8943, loss_cns_2: 0.5687, loss_yns_2: 0.1749, loss_cls_3: 1.2624, loss_box_3: 2.9568, loss_cns_3: 0.5768, loss_yns_3: 0.1694, loss_cls_4: 1.2536, loss_box_4: 3.0819, loss_cns_4: 0.5690, loss_yns_4: 0.1778, loss_cls_5: 1.2481, loss_box_5: 3.1506, loss_cns_5: 0.5593, loss_yns_5: 0.1846, loss_cls_dn_0: 0.4191, loss_box_dn_0: 1.0295, loss_cls_dn_1: 0.4365, loss_box_dn_1: 1.3444, loss_cls_dn_2: 0.4361, loss_box_dn_2: 1.4343, loss_cls_dn_3: 0.4150, loss_box_dn_3: 1.4943, loss_cls_dn_4: 0.4410, loss_box_dn_4: 1.6663, loss_cls_dn_5: 0.4457, loss_box_dn_5: 1.6842, loss_dense_depth: 1.4161, loss: 41.6958, grad_norm: 98.1124 -2025-11-13 15:34:28,189 - mmdet - INFO - Iter [23/17500] lr: 1.088e-04, eta: 1 day, 7:28:58, time: 1.508, data_time: 0.077, memory: 49163, loss_cls_0: 1.1422, loss_box_0: 2.1802, loss_cns_0: 0.6189, loss_yns_0: 0.1685, loss_cls_1: 1.2017, loss_box_1: 2.7091, loss_cns_1: 0.5761, loss_yns_1: 0.1732, loss_cls_2: 1.2325, loss_box_2: 2.7373, loss_cns_2: 0.5855, loss_yns_2: 0.1744, loss_cls_3: 1.2538, loss_box_3: 2.7332, loss_cns_3: 0.6021, loss_yns_3: 0.1707, loss_cls_4: 1.2637, loss_box_4: 2.8552, loss_cns_4: 0.5785, loss_yns_4: 0.1848, loss_cls_5: 1.2434, loss_box_5: 2.9533, loss_cns_5: 0.5838, loss_yns_5: 0.1896, loss_cls_dn_0: 0.4465, loss_box_dn_0: 1.0230, loss_cls_dn_1: 0.4318, loss_box_dn_1: 1.3066, loss_cls_dn_2: 0.4361, loss_box_dn_2: 1.3601, loss_cls_dn_3: 0.4102, loss_box_dn_3: 1.3809, loss_cls_dn_4: 0.4156, loss_box_dn_4: 1.5275, loss_cls_dn_5: 0.4340, loss_box_dn_5: 1.5869, loss_dense_depth: 1.3743, loss: 40.2451, grad_norm: 61.8290 -2025-11-13 15:34:29,697 - mmdet - INFO - Iter [24/17500] lr: 1.092e-04, eta: 1 day, 6:28:27, time: 1.507, data_time: 0.076, memory: 49163, loss_cls_0: 1.1420, loss_box_0: 2.1428, loss_cns_0: 0.5992, loss_yns_0: 0.1673, loss_cls_1: 1.2196, loss_box_1: 2.7375, loss_cns_1: 0.5697, loss_yns_1: 0.1747, loss_cls_2: 1.2277, loss_box_2: 2.7954, loss_cns_2: 0.5686, loss_yns_2: 0.1805, loss_cls_3: 1.2483, loss_box_3: 2.7853, loss_cns_3: 0.5899, loss_yns_3: 0.1724, loss_cls_4: 1.2781, loss_box_4: 2.8453, loss_cns_4: 0.5635, loss_yns_4: 0.1804, loss_cls_5: 1.2597, loss_box_5: 2.8597, loss_cns_5: 0.5822, loss_yns_5: 0.1777, loss_cls_dn_0: 0.5037, loss_box_dn_0: 1.0625, loss_cls_dn_1: 0.4160, loss_box_dn_1: 1.4195, loss_cls_dn_2: 0.4296, loss_box_dn_2: 1.4381, loss_cls_dn_3: 0.4097, loss_box_dn_3: 1.4188, loss_cls_dn_4: 0.3966, loss_box_dn_4: 1.4939, loss_cls_dn_5: 0.4144, loss_box_dn_5: 1.5212, loss_dense_depth: 1.3570, loss: 40.3486, grad_norm: 69.8439 -2025-11-13 15:34:31,187 - mmdet - INFO - Iter [25/17500] lr: 1.096e-04, eta: 1 day, 5:32:34, time: 1.489, data_time: 0.084, memory: 49163, loss_cls_0: 1.1449, loss_box_0: 2.1187, loss_cns_0: 0.6011, loss_yns_0: 0.1625, loss_cls_1: 1.1985, loss_box_1: 2.8900, loss_cns_1: 0.5488, loss_yns_1: 0.1756, loss_cls_2: 1.2044, loss_box_2: 2.9377, loss_cns_2: 0.5420, loss_yns_2: 0.1759, loss_cls_3: 1.2345, loss_box_3: 3.0111, loss_cns_3: 0.5603, loss_yns_3: 0.1724, loss_cls_4: 1.2399, loss_box_4: 3.0420, loss_cns_4: 0.5527, loss_yns_4: 0.1889, loss_cls_5: 1.2565, loss_box_5: 2.9955, loss_cns_5: 0.5666, loss_yns_5: 0.1795, loss_cls_dn_0: 0.5072, loss_box_dn_0: 1.0573, loss_cls_dn_1: 0.4078, loss_box_dn_1: 1.4096, loss_cls_dn_2: 0.4292, loss_box_dn_2: 1.4098, loss_cls_dn_3: 0.4193, loss_box_dn_3: 1.4302, loss_cls_dn_4: 0.4061, loss_box_dn_4: 1.4464, loss_cls_dn_5: 0.4085, loss_box_dn_5: 1.4711, loss_dense_depth: 1.3399, loss: 40.8427, grad_norm: 84.2850 -2025-11-13 15:34:32,684 - mmdet - INFO - Iter [26/17500] lr: 1.100e-04, eta: 1 day, 4:41:03, time: 1.497, data_time: 0.079, memory: 49163, loss_cls_0: 1.1238, loss_box_0: 2.0969, loss_cns_0: 0.6139, loss_yns_0: 0.1695, loss_cls_1: 1.1762, loss_box_1: 2.8686, loss_cns_1: 0.5516, loss_yns_1: 0.1722, loss_cls_2: 1.2018, loss_box_2: 2.8539, loss_cns_2: 0.5538, loss_yns_2: 0.1752, loss_cls_3: 1.2291, loss_box_3: 3.0461, loss_cns_3: 0.5617, loss_yns_3: 0.1704, loss_cls_4: 1.2256, loss_box_4: 3.1284, loss_cns_4: 0.5603, loss_yns_4: 0.1751, loss_cls_5: 1.2487, loss_box_5: 3.1969, loss_cns_5: 0.5456, loss_yns_5: 0.1748, loss_cls_dn_0: 0.4661, loss_box_dn_0: 1.0496, loss_cls_dn_1: 0.4384, loss_box_dn_1: 1.2705, loss_cls_dn_2: 0.4551, loss_box_dn_2: 1.2656, loss_cls_dn_3: 0.4506, loss_box_dn_3: 1.4045, loss_cls_dn_4: 0.4509, loss_box_dn_4: 1.4247, loss_cls_dn_5: 0.4405, loss_box_dn_5: 1.5267, loss_dense_depth: 1.2721, loss: 40.7355, grad_norm: 98.0892 -2025-11-13 15:34:34,177 - mmdet - INFO - Iter [27/17500] lr: 1.104e-04, eta: 1 day, 3:53:19, time: 1.493, data_time: 0.078, memory: 49163, loss_cls_0: 1.1416, loss_box_0: 2.1044, loss_cns_0: 0.6105, loss_yns_0: 0.1725, loss_cls_1: 1.1672, loss_box_1: 2.7989, loss_cns_1: 0.5588, loss_yns_1: 0.1729, loss_cls_2: 1.1935, loss_box_2: 2.7734, loss_cns_2: 0.5647, loss_yns_2: 0.1763, loss_cls_3: 1.2169, loss_box_3: 2.8903, loss_cns_3: 0.5618, loss_yns_3: 0.1698, loss_cls_4: 1.2133, loss_box_4: 2.9502, loss_cns_4: 0.5515, loss_yns_4: 0.1726, loss_cls_5: 1.2341, loss_box_5: 3.1313, loss_cns_5: 0.5265, loss_yns_5: 0.1726, loss_cls_dn_0: 0.4520, loss_box_dn_0: 1.0358, loss_cls_dn_1: 0.4276, loss_box_dn_1: 1.3126, loss_cls_dn_2: 0.4374, loss_box_dn_2: 1.2858, loss_cls_dn_3: 0.4396, loss_box_dn_3: 1.4075, loss_cls_dn_4: 0.4551, loss_box_dn_4: 1.4062, loss_cls_dn_5: 0.4454, loss_box_dn_5: 1.5478, loss_dense_depth: 1.2604, loss: 40.1388, grad_norm: 80.4135 -2025-11-13 15:34:35,660 - mmdet - INFO - Iter [28/17500] lr: 1.108e-04, eta: 1 day, 3:08:53, time: 1.483, data_time: 0.078, memory: 49163, loss_cls_0: 1.1282, loss_box_0: 2.1171, loss_cns_0: 0.6099, loss_yns_0: 0.1722, loss_cls_1: 1.1694, loss_box_1: 2.5647, loss_cns_1: 0.5753, loss_yns_1: 0.1722, loss_cls_2: 1.2100, loss_box_2: 2.5706, loss_cns_2: 0.5807, loss_yns_2: 0.1775, loss_cls_3: 1.2137, loss_box_3: 2.6163, loss_cns_3: 0.5756, loss_yns_3: 0.1703, loss_cls_4: 1.2073, loss_box_4: 2.6867, loss_cns_4: 0.5726, loss_yns_4: 0.1700, loss_cls_5: 1.2253, loss_box_5: 2.7523, loss_cns_5: 0.5602, loss_yns_5: 0.1747, loss_cls_dn_0: 0.4414, loss_box_dn_0: 1.0307, loss_cls_dn_1: 0.4172, loss_box_dn_1: 1.3544, loss_cls_dn_2: 0.4057, loss_box_dn_2: 1.3547, loss_cls_dn_3: 0.4116, loss_box_dn_3: 1.4462, loss_cls_dn_4: 0.4343, loss_box_dn_4: 1.4638, loss_cls_dn_5: 0.4372, loss_box_dn_5: 1.5572, loss_dense_depth: 1.2418, loss: 38.9689, grad_norm: 69.8118 -2025-11-13 15:34:37,164 - mmdet - INFO - Iter [29/17500] lr: 1.112e-04, eta: 1 day, 2:27:45, time: 1.506, data_time: 0.078, memory: 49163, loss_cls_0: 1.1112, loss_box_0: 2.1305, loss_cns_0: 0.6081, loss_yns_0: 0.1721, loss_cls_1: 1.1761, loss_box_1: 2.5372, loss_cns_1: 0.5786, loss_yns_1: 0.1722, loss_cls_2: 1.2549, loss_box_2: 2.5498, loss_cns_2: 0.5820, loss_yns_2: 0.1713, loss_cls_3: 1.2138, loss_box_3: 2.6028, loss_cns_3: 0.5877, loss_yns_3: 0.1689, loss_cls_4: 1.2118, loss_box_4: 2.7145, loss_cns_4: 0.5840, loss_yns_4: 0.1693, loss_cls_5: 1.2045, loss_box_5: 2.6348, loss_cns_5: 0.5930, loss_yns_5: 0.1724, loss_cls_dn_0: 0.4375, loss_box_dn_0: 1.0374, loss_cls_dn_1: 0.4035, loss_box_dn_1: 1.2679, loss_cls_dn_2: 0.3884, loss_box_dn_2: 1.2947, loss_cls_dn_3: 0.4029, loss_box_dn_3: 1.3397, loss_cls_dn_4: 0.4220, loss_box_dn_4: 1.3759, loss_cls_dn_5: 0.4373, loss_box_dn_5: 1.3925, loss_dense_depth: 1.2079, loss: 38.3091, grad_norm: 91.7043 -2025-11-13 15:34:38,651 - mmdet - INFO - Iter [30/17500] lr: 1.116e-04, eta: 1 day, 1:49:10, time: 1.486, data_time: 0.074, memory: 49163, loss_cls_0: 1.1079, loss_box_0: 2.0848, loss_cns_0: 0.6100, loss_yns_0: 0.1756, loss_cls_1: 1.1440, loss_box_1: 2.6299, loss_cns_1: 0.5670, loss_yns_1: 0.1763, loss_cls_2: 1.1840, loss_box_2: 2.7298, loss_cns_2: 0.5728, loss_yns_2: 0.1787, loss_cls_3: 1.1707, loss_box_3: 2.6507, loss_cns_3: 0.5903, loss_yns_3: 0.1693, loss_cls_4: 1.1828, loss_box_4: 2.7364, loss_cns_4: 0.5781, loss_yns_4: 0.1711, loss_cls_5: 1.1814, loss_box_5: 2.6866, loss_cns_5: 0.5810, loss_yns_5: 0.1715, loss_cls_dn_0: 0.4255, loss_box_dn_0: 1.0395, loss_cls_dn_1: 0.3889, loss_box_dn_1: 1.3480, loss_cls_dn_2: 0.3888, loss_box_dn_2: 1.3911, loss_cls_dn_3: 0.3934, loss_box_dn_3: 1.3516, loss_cls_dn_4: 0.3988, loss_box_dn_4: 1.3874, loss_cls_dn_5: 0.4081, loss_box_dn_5: 1.3959, loss_dense_depth: 1.2134, loss: 38.5612, grad_norm: 82.2647 -2025-11-13 15:34:40,138 - mmdet - INFO - Iter [31/17500] lr: 1.120e-04, eta: 1 day, 1:13:04, time: 1.486, data_time: 0.078, memory: 49163, loss_cls_0: 1.1015, loss_box_0: 2.0573, loss_cns_0: 0.6142, loss_yns_0: 0.1744, loss_cls_1: 1.1215, loss_box_1: 2.7292, loss_cns_1: 0.5570, loss_yns_1: 0.1741, loss_cls_2: 1.1435, loss_box_2: 2.8138, loss_cns_2: 0.5697, loss_yns_2: 0.1785, loss_cls_3: 1.1578, loss_box_3: 2.6836, loss_cns_3: 0.5852, loss_yns_3: 0.1692, loss_cls_4: 1.1655, loss_box_4: 2.6539, loss_cns_4: 0.5791, loss_yns_4: 0.1776, loss_cls_5: 1.1802, loss_box_5: 2.7055, loss_cns_5: 0.5810, loss_yns_5: 0.1728, loss_cls_dn_0: 0.4248, loss_box_dn_0: 1.0222, loss_cls_dn_1: 0.3856, loss_box_dn_1: 1.3597, loss_cls_dn_2: 0.4116, loss_box_dn_2: 1.3986, loss_cls_dn_3: 0.3938, loss_box_dn_3: 1.3174, loss_cls_dn_4: 0.3890, loss_box_dn_4: 1.3468, loss_cls_dn_5: 0.3948, loss_box_dn_5: 1.3707, loss_dense_depth: 1.1409, loss: 38.4019, grad_norm: 64.7050 -2025-11-13 15:34:41,624 - mmdet - INFO - Iter [32/17500] lr: 1.124e-04, eta: 1 day, 0:39:13, time: 1.487, data_time: 0.077, memory: 49163, loss_cls_0: 1.0654, loss_box_0: 2.0483, loss_cns_0: 0.6143, loss_yns_0: 0.1721, loss_cls_1: 1.1059, loss_box_1: 2.7373, loss_cns_1: 0.5573, loss_yns_1: 0.1713, loss_cls_2: 1.1356, loss_box_2: 2.8071, loss_cns_2: 0.5656, loss_yns_2: 0.1727, loss_cls_3: 1.1519, loss_box_3: 2.7541, loss_cns_3: 0.5758, loss_yns_3: 0.1712, loss_cls_4: 1.1556, loss_box_4: 2.7247, loss_cns_4: 0.5756, loss_yns_4: 0.1806, loss_cls_5: 1.1696, loss_box_5: 2.7891, loss_cns_5: 0.5839, loss_yns_5: 0.1763, loss_cls_dn_0: 0.4148, loss_box_dn_0: 1.0036, loss_cls_dn_1: 0.3798, loss_box_dn_1: 1.2653, loss_cls_dn_2: 0.4151, loss_box_dn_2: 1.3082, loss_cls_dn_3: 0.3792, loss_box_dn_3: 1.2934, loss_cls_dn_4: 0.3803, loss_box_dn_4: 1.3599, loss_cls_dn_5: 0.3911, loss_box_dn_5: 1.3896, loss_dense_depth: 1.1322, loss: 38.2737, grad_norm: 76.8151 -2025-11-13 15:34:43,122 - mmdet - INFO - Iter [33/17500] lr: 1.128e-04, eta: 1 day, 0:07:32, time: 1.498, data_time: 0.076, memory: 49163, loss_cls_0: 1.0602, loss_box_0: 2.0423, loss_cns_0: 0.6166, loss_yns_0: 0.1726, loss_cls_1: 1.1016, loss_box_1: 2.6795, loss_cns_1: 0.5698, loss_yns_1: 0.1732, loss_cls_2: 1.1371, loss_box_2: 2.7106, loss_cns_2: 0.5823, loss_yns_2: 0.1722, loss_cls_3: 1.1472, loss_box_3: 2.7341, loss_cns_3: 0.5866, loss_yns_3: 0.1730, loss_cls_4: 1.1484, loss_box_4: 2.7685, loss_cns_4: 0.5829, loss_yns_4: 0.1747, loss_cls_5: 1.1468, loss_box_5: 2.7782, loss_cns_5: 0.5896, loss_yns_5: 0.1792, loss_cls_dn_0: 0.4202, loss_box_dn_0: 1.0046, loss_cls_dn_1: 0.3624, loss_box_dn_1: 1.3823, loss_cls_dn_2: 0.3978, loss_box_dn_2: 1.4044, loss_cls_dn_3: 0.3703, loss_box_dn_3: 1.4285, loss_cls_dn_4: 0.3696, loss_box_dn_4: 1.5126, loss_cls_dn_5: 0.3854, loss_box_dn_5: 1.5262, loss_dense_depth: 1.1212, loss: 38.7129, grad_norm: 68.5190 -2025-11-13 15:34:44,616 - mmdet - INFO - Iter [34/17500] lr: 1.132e-04, eta: 23:37:38, time: 1.489, data_time: 0.079, memory: 49163, loss_cls_0: 1.0558, loss_box_0: 2.0321, loss_cns_0: 0.6180, loss_yns_0: 0.1717, loss_cls_1: 1.0996, loss_box_1: 2.4972, loss_cns_1: 0.5902, loss_yns_1: 0.1775, loss_cls_2: 1.1295, loss_box_2: 2.4887, loss_cns_2: 0.6017, loss_yns_2: 0.1760, loss_cls_3: 1.1390, loss_box_3: 2.5750, loss_cns_3: 0.6017, loss_yns_3: 0.1743, loss_cls_4: 1.1487, loss_box_4: 2.6204, loss_cns_4: 0.5894, loss_yns_4: 0.1793, loss_cls_5: 1.1484, loss_box_5: 2.6373, loss_cns_5: 0.5882, loss_yns_5: 0.1778, loss_cls_dn_0: 0.4281, loss_box_dn_0: 1.0009, loss_cls_dn_1: 0.3542, loss_box_dn_1: 1.4794, loss_cls_dn_2: 0.3729, loss_box_dn_2: 1.4641, loss_cls_dn_3: 0.3700, loss_box_dn_3: 1.4975, loss_cls_dn_4: 0.3622, loss_box_dn_4: 1.5619, loss_cls_dn_5: 0.3786, loss_box_dn_5: 1.5762, loss_dense_depth: 1.1443, loss: 38.2079, grad_norm: 73.3976 -2025-11-13 15:34:46,106 - mmdet - INFO - Iter [35/17500] lr: 1.136e-04, eta: 23:09:28, time: 1.494, data_time: 0.087, memory: 49163, loss_cls_0: 1.0644, loss_box_0: 1.9813, loss_cns_0: 0.6221, loss_yns_0: 0.1702, loss_cls_1: 1.1355, loss_box_1: 2.3777, loss_cns_1: 0.5970, loss_yns_1: 0.1757, loss_cls_2: 1.1488, loss_box_2: 2.3482, loss_cns_2: 0.6086, loss_yns_2: 0.1821, loss_cls_3: 1.1560, loss_box_3: 2.4227, loss_cns_3: 0.6181, loss_yns_3: 0.1789, loss_cls_4: 1.1636, loss_box_4: 2.4361, loss_cns_4: 0.6053, loss_yns_4: 0.1828, loss_cls_5: 1.1611, loss_box_5: 2.5031, loss_cns_5: 0.6010, loss_yns_5: 0.1756, loss_cls_dn_0: 0.4274, loss_box_dn_0: 0.9836, loss_cls_dn_1: 0.3437, loss_box_dn_1: 1.4530, loss_cls_dn_2: 0.3504, loss_box_dn_2: 1.4325, loss_cls_dn_3: 0.3635, loss_box_dn_3: 1.4591, loss_cls_dn_4: 0.3621, loss_box_dn_4: 1.5003, loss_cls_dn_5: 0.3730, loss_box_dn_5: 1.5282, loss_dense_depth: 1.0551, loss: 37.2480, grad_norm: 52.5054 -2025-11-13 15:34:47,591 - mmdet - INFO - Iter [36/17500] lr: 1.140e-04, eta: 22:42:48, time: 1.484, data_time: 0.076, memory: 49163, loss_cls_0: 1.0389, loss_box_0: 1.9681, loss_cns_0: 0.6193, loss_yns_0: 0.1714, loss_cls_1: 1.1255, loss_box_1: 2.4291, loss_cns_1: 0.5939, loss_yns_1: 0.1721, loss_cls_2: 1.1538, loss_box_2: 2.4229, loss_cns_2: 0.6086, loss_yns_2: 0.1766, loss_cls_3: 1.1516, loss_box_3: 2.4815, loss_cns_3: 0.6200, loss_yns_3: 0.1759, loss_cls_4: 1.1551, loss_box_4: 2.5263, loss_cns_4: 0.6091, loss_yns_4: 0.1746, loss_cls_5: 1.1488, loss_box_5: 2.5049, loss_cns_5: 0.6134, loss_yns_5: 0.1771, loss_cls_dn_0: 0.4365, loss_box_dn_0: 0.9695, loss_cls_dn_1: 0.3579, loss_box_dn_1: 1.1786, loss_cls_dn_2: 0.3605, loss_box_dn_2: 1.1750, loss_cls_dn_3: 0.3822, loss_box_dn_3: 1.2124, loss_cls_dn_4: 0.3919, loss_box_dn_4: 1.2692, loss_cls_dn_5: 0.3943, loss_box_dn_5: 1.2681, loss_dense_depth: 1.0567, loss: 36.2708, grad_norm: 72.9751 -2025-11-13 15:34:49,087 - mmdet - INFO - Iter [37/17500] lr: 1.144e-04, eta: 22:17:39, time: 1.495, data_time: 0.077, memory: 49163, loss_cls_0: 1.0354, loss_box_0: 1.9616, loss_cns_0: 0.6182, loss_yns_0: 0.1710, loss_cls_1: 1.1201, loss_box_1: 2.4624, loss_cns_1: 0.5943, loss_yns_1: 0.1739, loss_cls_2: 1.1665, loss_box_2: 2.4736, loss_cns_2: 0.6020, loss_yns_2: 0.1722, loss_cls_3: 1.1391, loss_box_3: 2.5140, loss_cns_3: 0.6113, loss_yns_3: 0.1699, loss_cls_4: 1.1427, loss_box_4: 2.5512, loss_cns_4: 0.6066, loss_yns_4: 0.1728, loss_cls_5: 1.1505, loss_box_5: 2.5172, loss_cns_5: 0.6121, loss_yns_5: 0.1714, loss_cls_dn_0: 0.4377, loss_box_dn_0: 0.9681, loss_cls_dn_1: 0.3406, loss_box_dn_1: 1.2076, loss_cls_dn_2: 0.3590, loss_box_dn_2: 1.1978, loss_cls_dn_3: 0.3693, loss_box_dn_3: 1.2305, loss_cls_dn_4: 0.3814, loss_box_dn_4: 1.2727, loss_cls_dn_5: 0.3733, loss_box_dn_5: 1.2767, loss_dense_depth: 1.0438, loss: 36.3684, grad_norm: 72.5426 -2025-11-13 15:34:50,586 - mmdet - INFO - Iter [38/17500] lr: 1.148e-04, eta: 21:53:52, time: 1.500, data_time: 0.078, memory: 49163, loss_cls_0: 1.0372, loss_box_0: 1.9809, loss_cns_0: 0.6197, loss_yns_0: 0.1701, loss_cls_1: 1.1208, loss_box_1: 2.4676, loss_cns_1: 0.5999, loss_yns_1: 0.1814, loss_cls_2: 1.1474, loss_box_2: 2.4991, loss_cns_2: 0.6140, loss_yns_2: 0.1763, loss_cls_3: 1.1368, loss_box_3: 2.5432, loss_cns_3: 0.6069, loss_yns_3: 0.1705, loss_cls_4: 1.1357, loss_box_4: 2.5725, loss_cns_4: 0.6030, loss_yns_4: 0.1725, loss_cls_5: 1.1575, loss_box_5: 2.5497, loss_cns_5: 0.6068, loss_yns_5: 0.1718, loss_cls_dn_0: 0.4188, loss_box_dn_0: 0.9666, loss_cls_dn_1: 0.3329, loss_box_dn_1: 1.0894, loss_cls_dn_2: 0.3598, loss_box_dn_2: 1.0910, loss_cls_dn_3: 0.3631, loss_box_dn_3: 1.1494, loss_cls_dn_4: 0.3723, loss_box_dn_4: 1.1761, loss_cls_dn_5: 0.3705, loss_box_dn_5: 1.2237, loss_dense_depth: 1.0279, loss: 35.9828, grad_norm: 54.4470 -2025-11-13 15:34:52,087 - mmdet - INFO - Iter [39/17500] lr: 1.152e-04, eta: 21:31:18, time: 1.501, data_time: 0.075, memory: 49163, loss_cls_0: 1.0733, loss_box_0: 1.9380, loss_cns_0: 0.6209, loss_yns_0: 0.1693, loss_cls_1: 1.1230, loss_box_1: 2.4263, loss_cns_1: 0.6013, loss_yns_1: 0.1758, loss_cls_2: 1.1370, loss_box_2: 2.4931, loss_cns_2: 0.6134, loss_yns_2: 0.1788, loss_cls_3: 1.1510, loss_box_3: 2.5411, loss_cns_3: 0.6018, loss_yns_3: 0.1724, loss_cls_4: 1.1517, loss_box_4: 2.5562, loss_cns_4: 0.5947, loss_yns_4: 0.1757, loss_cls_5: 1.1478, loss_box_5: 2.5564, loss_cns_5: 0.6006, loss_yns_5: 0.1672, loss_cls_dn_0: 0.4040, loss_box_dn_0: 0.9415, loss_cls_dn_1: 0.3351, loss_box_dn_1: 1.0528, loss_cls_dn_2: 0.3621, loss_box_dn_2: 1.0886, loss_cls_dn_3: 0.3670, loss_box_dn_3: 1.1708, loss_cls_dn_4: 0.3686, loss_box_dn_4: 1.1967, loss_cls_dn_5: 0.3882, loss_box_dn_5: 1.2898, loss_dense_depth: 1.0093, loss: 35.9413, grad_norm: 69.7383 -2025-11-13 15:34:53,582 - mmdet - INFO - Iter [40/17500] lr: 1.156e-04, eta: 21:09:49, time: 1.494, data_time: 0.079, memory: 49163, loss_cls_0: 1.0703, loss_box_0: 1.9506, loss_cns_0: 0.6218, loss_yns_0: 0.1689, loss_cls_1: 1.1071, loss_box_1: 2.3115, loss_cns_1: 0.6063, loss_yns_1: 0.1697, loss_cls_2: 1.1180, loss_box_2: 2.3249, loss_cns_2: 0.6135, loss_yns_2: 0.1744, loss_cls_3: 1.1256, loss_box_3: 2.3525, loss_cns_3: 0.6103, loss_yns_3: 0.1723, loss_cls_4: 1.1306, loss_box_4: 2.4019, loss_cns_4: 0.6006, loss_yns_4: 0.1715, loss_cls_5: 1.1291, loss_box_5: 2.5206, loss_cns_5: 0.5912, loss_yns_5: 0.1671, loss_cls_dn_0: 0.3985, loss_box_dn_0: 0.9354, loss_cls_dn_1: 0.3425, loss_box_dn_1: 1.1349, loss_cls_dn_2: 0.3663, loss_box_dn_2: 1.1736, loss_cls_dn_3: 0.3772, loss_box_dn_3: 1.2474, loss_cls_dn_4: 0.3711, loss_box_dn_4: 1.2870, loss_cls_dn_5: 0.4054, loss_box_dn_5: 1.4117, loss_dense_depth: 1.0360, loss: 35.6975, grad_norm: 71.7268 -2025-11-13 15:34:55,137 - mmdet - INFO - Iter [41/17500] lr: 1.160e-04, eta: 20:49:49, time: 1.555, data_time: 0.083, memory: 49163, loss_cls_0: 1.0176, loss_box_0: 1.9836, loss_cns_0: 0.6192, loss_yns_0: 0.1673, loss_cls_1: 1.1033, loss_box_1: 2.2813, loss_cns_1: 0.6106, loss_yns_1: 0.1695, loss_cls_2: 1.1145, loss_box_2: 2.2328, loss_cns_2: 0.6194, loss_yns_2: 0.1696, loss_cls_3: 1.1134, loss_box_3: 2.2691, loss_cns_3: 0.6181, loss_yns_3: 0.1694, loss_cls_4: 1.1102, loss_box_4: 2.3082, loss_cns_4: 0.6117, loss_yns_4: 0.1650, loss_cls_5: 1.1211, loss_box_5: 2.3949, loss_cns_5: 0.6012, loss_yns_5: 0.1689, loss_cls_dn_0: 0.4166, loss_box_dn_0: 0.9361, loss_cls_dn_1: 0.3334, loss_box_dn_1: 1.2301, loss_cls_dn_2: 0.3613, loss_box_dn_2: 1.2472, loss_cls_dn_3: 0.3735, loss_box_dn_3: 1.3043, loss_cls_dn_4: 0.3706, loss_box_dn_4: 1.3380, loss_cls_dn_5: 0.3938, loss_box_dn_5: 1.4423, loss_dense_depth: 1.0117, loss: 35.4989, grad_norm: 67.7564 -2025-11-13 15:34:56,707 - mmdet - INFO - Iter [42/17500] lr: 1.164e-04, eta: 20:30:52, time: 1.570, data_time: 0.158, memory: 49163, loss_cls_0: 1.0170, loss_box_0: 1.9702, loss_cns_0: 0.6216, loss_yns_0: 0.1665, loss_cls_1: 1.0955, loss_box_1: 2.2839, loss_cns_1: 0.6081, loss_yns_1: 0.1712, loss_cls_2: 1.1161, loss_box_2: 2.2599, loss_cns_2: 0.6111, loss_yns_2: 0.1689, loss_cls_3: 1.1095, loss_box_3: 2.3024, loss_cns_3: 0.6181, loss_yns_3: 0.1672, loss_cls_4: 1.1121, loss_box_4: 2.2686, loss_cns_4: 0.6183, loss_yns_4: 0.1707, loss_cls_5: 1.1209, loss_box_5: 2.2882, loss_cns_5: 0.6169, loss_yns_5: 0.1671, loss_cls_dn_0: 0.4246, loss_box_dn_0: 0.9343, loss_cls_dn_1: 0.3356, loss_box_dn_1: 1.2083, loss_cls_dn_2: 0.3641, loss_box_dn_2: 1.2092, loss_cls_dn_3: 0.3731, loss_box_dn_3: 1.2659, loss_cls_dn_4: 0.3698, loss_box_dn_4: 1.2745, loss_cls_dn_5: 0.3892, loss_box_dn_5: 1.3446, loss_dense_depth: 0.9734, loss: 35.1167, grad_norm: 69.4645 -2025-11-13 15:34:58,235 - mmdet - INFO - Iter [43/17500] lr: 1.168e-04, eta: 20:12:30, time: 1.527, data_time: 0.077, memory: 49163, loss_cls_0: 1.0413, loss_box_0: 1.9866, loss_cns_0: 0.6153, loss_yns_0: 0.1646, loss_cls_1: 1.1158, loss_box_1: 2.3849, loss_cns_1: 0.5882, loss_yns_1: 0.1679, loss_cls_2: 1.1293, loss_box_2: 2.3343, loss_cns_2: 0.5983, loss_yns_2: 0.1686, loss_cls_3: 1.1352, loss_box_3: 2.3912, loss_cns_3: 0.6053, loss_yns_3: 0.1677, loss_cls_4: 1.1340, loss_box_4: 2.3099, loss_cns_4: 0.6089, loss_yns_4: 0.1694, loss_cls_5: 1.1433, loss_box_5: 2.3259, loss_cns_5: 0.6102, loss_yns_5: 0.1675, loss_cls_dn_0: 0.4353, loss_box_dn_0: 0.9286, loss_cls_dn_1: 0.3352, loss_box_dn_1: 1.1496, loss_cls_dn_2: 0.3738, loss_box_dn_2: 1.1295, loss_cls_dn_3: 0.3720, loss_box_dn_3: 1.1809, loss_cls_dn_4: 0.3743, loss_box_dn_4: 1.1677, loss_cls_dn_5: 0.3872, loss_box_dn_5: 1.2173, loss_dense_depth: 0.9831, loss: 35.0981, grad_norm: 61.0776 -2025-11-13 15:34:59,754 - mmdet - INFO - Iter [44/17500] lr: 1.172e-04, eta: 19:54:56, time: 1.520, data_time: 0.075, memory: 49163, loss_cls_0: 1.0432, loss_box_0: 2.0708, loss_cns_0: 0.6045, loss_yns_0: 0.1652, loss_cls_1: 1.1176, loss_box_1: 2.3966, loss_cns_1: 0.5841, loss_yns_1: 0.1644, loss_cls_2: 1.1134, loss_box_2: 2.2947, loss_cns_2: 0.6152, loss_yns_2: 0.1693, loss_cls_3: 1.1609, loss_box_3: 2.3188, loss_cns_3: 0.6064, loss_yns_3: 0.1707, loss_cls_4: 1.1391, loss_box_4: 2.3050, loss_cns_4: 0.6058, loss_yns_4: 0.1721, loss_cls_5: 1.1401, loss_box_5: 2.2715, loss_cns_5: 0.6134, loss_yns_5: 0.1664, loss_cls_dn_0: 0.4355, loss_box_dn_0: 0.9413, loss_cls_dn_1: 0.3292, loss_box_dn_1: 1.1917, loss_cls_dn_2: 0.3745, loss_box_dn_2: 1.1395, loss_cls_dn_3: 0.3708, loss_box_dn_3: 1.1570, loss_cls_dn_4: 0.3775, loss_box_dn_4: 1.1658, loss_cls_dn_5: 0.3842, loss_box_dn_5: 1.1721, loss_dense_depth: 1.0327, loss: 35.0812, grad_norm: 47.3139 -2025-11-13 15:35:01,265 - mmdet - INFO - Iter [45/17500] lr: 1.176e-04, eta: 19:38:05, time: 1.511, data_time: 0.087, memory: 49163, loss_cls_0: 1.0628, loss_box_0: 2.0031, loss_cns_0: 0.6133, loss_yns_0: 0.1665, loss_cls_1: 1.1095, loss_box_1: 2.3762, loss_cns_1: 0.5876, loss_yns_1: 0.1705, loss_cls_2: 1.1257, loss_box_2: 2.3080, loss_cns_2: 0.6070, loss_yns_2: 0.1705, loss_cls_3: 1.1371, loss_box_3: 2.2920, loss_cns_3: 0.6057, loss_yns_3: 0.1715, loss_cls_4: 1.1336, loss_box_4: 2.3343, loss_cns_4: 0.6045, loss_yns_4: 0.1735, loss_cls_5: 1.1538, loss_box_5: 2.3158, loss_cns_5: 0.6085, loss_yns_5: 0.1724, loss_cls_dn_0: 0.4201, loss_box_dn_0: 0.9193, loss_cls_dn_1: 0.3228, loss_box_dn_1: 1.1115, loss_cls_dn_2: 0.3649, loss_box_dn_2: 1.0497, loss_cls_dn_3: 0.3711, loss_box_dn_3: 1.0433, loss_cls_dn_4: 0.3738, loss_box_dn_4: 1.0756, loss_cls_dn_5: 0.3797, loss_box_dn_5: 1.0806, loss_dense_depth: 1.0384, loss: 34.5544, grad_norm: 41.7758 -2025-11-13 15:35:02,791 - mmdet - INFO - Iter [46/17500] lr: 1.180e-04, eta: 19:22:03, time: 1.526, data_time: 0.073, memory: 49163, loss_cls_0: 1.0549, loss_box_0: 1.9900, loss_cns_0: 0.6127, loss_yns_0: 0.1649, loss_cls_1: 1.0885, loss_box_1: 2.4284, loss_cns_1: 0.5770, loss_yns_1: 0.1672, loss_cls_2: 1.1160, loss_box_2: 2.3519, loss_cns_2: 0.5971, loss_yns_2: 0.1656, loss_cls_3: 1.1508, loss_box_3: 2.3412, loss_cns_3: 0.6033, loss_yns_3: 0.1639, loss_cls_4: 1.1248, loss_box_4: 2.3543, loss_cns_4: 0.6019, loss_yns_4: 0.1715, loss_cls_5: 1.1459, loss_box_5: 2.3773, loss_cns_5: 0.6048, loss_yns_5: 0.1733, loss_cls_dn_0: 0.4015, loss_box_dn_0: 0.9021, loss_cls_dn_1: 0.3147, loss_box_dn_1: 1.0585, loss_cls_dn_2: 0.3412, loss_box_dn_2: 0.9773, loss_cls_dn_3: 0.3522, loss_box_dn_3: 0.9679, loss_cls_dn_4: 0.3525, loss_box_dn_4: 0.9923, loss_cls_dn_5: 0.3607, loss_box_dn_5: 1.0127, loss_dense_depth: 1.0731, loss: 34.2340, grad_norm: 35.4757 -2025-11-13 15:35:04,275 - mmdet - INFO - Iter [47/17500] lr: 1.184e-04, eta: 19:06:27, time: 1.484, data_time: 0.073, memory: 49163, loss_cls_0: 1.0493, loss_box_0: 2.0128, loss_cns_0: 0.6154, loss_yns_0: 0.1641, loss_cls_1: 1.0750, loss_box_1: 2.4528, loss_cns_1: 0.5886, loss_yns_1: 0.1663, loss_cls_2: 1.1159, loss_box_2: 2.3742, loss_cns_2: 0.6020, loss_yns_2: 0.1668, loss_cls_3: 1.1331, loss_box_3: 2.3862, loss_cns_3: 0.6033, loss_yns_3: 0.1651, loss_cls_4: 1.1363, loss_box_4: 2.3950, loss_cns_4: 0.6041, loss_yns_4: 0.1779, loss_cls_5: 1.1471, loss_box_5: 2.4136, loss_cns_5: 0.6069, loss_yns_5: 0.1666, loss_cls_dn_0: 0.4010, loss_box_dn_0: 0.9087, loss_cls_dn_1: 0.3261, loss_box_dn_1: 1.0416, loss_cls_dn_2: 0.3397, loss_box_dn_2: 0.9660, loss_cls_dn_3: 0.3657, loss_box_dn_3: 0.9740, loss_cls_dn_4: 0.3609, loss_box_dn_4: 1.0006, loss_cls_dn_5: 0.3702, loss_box_dn_5: 1.0252, loss_dense_depth: 1.2851, loss: 34.6834, grad_norm: 49.2144 -2025-11-13 15:35:05,778 - mmdet - INFO - Iter [48/17500] lr: 1.188e-04, eta: 18:51:36, time: 1.502, data_time: 0.074, memory: 49163, loss_cls_0: 1.0325, loss_box_0: 2.0307, loss_cns_0: 0.6083, loss_yns_0: 0.1649, loss_cls_1: 1.0738, loss_box_1: 2.4781, loss_cns_1: 0.5755, loss_yns_1: 0.1635, loss_cls_2: 1.0854, loss_box_2: 2.4063, loss_cns_2: 0.5983, loss_yns_2: 0.1689, loss_cls_3: 1.1103, loss_box_3: 2.4357, loss_cns_3: 0.5950, loss_yns_3: 0.1675, loss_cls_4: 1.1067, loss_box_4: 2.4117, loss_cns_4: 0.6008, loss_yns_4: 0.1727, loss_cls_5: 1.1229, loss_box_5: 2.4056, loss_cns_5: 0.6072, loss_yns_5: 0.1632, loss_cls_dn_0: 0.4047, loss_box_dn_0: 0.9161, loss_cls_dn_1: 0.3225, loss_box_dn_1: 1.1626, loss_cls_dn_2: 0.3512, loss_box_dn_2: 1.0665, loss_cls_dn_3: 0.3694, loss_box_dn_3: 1.0736, loss_cls_dn_4: 0.3567, loss_box_dn_4: 1.0832, loss_cls_dn_5: 0.3694, loss_box_dn_5: 1.1004, loss_dense_depth: 1.1709, loss: 35.0329, grad_norm: 40.5199 -2025-11-13 15:35:07,287 - mmdet - INFO - Iter [49/17500] lr: 1.192e-04, eta: 18:37:25, time: 1.511, data_time: 0.075, memory: 49163, loss_cls_0: 1.0494, loss_box_0: 2.0220, loss_cns_0: 0.6052, loss_yns_0: 0.1672, loss_cls_1: 1.0605, loss_box_1: 2.3832, loss_cns_1: 0.5844, loss_yns_1: 0.1678, loss_cls_2: 1.0997, loss_box_2: 2.3402, loss_cns_2: 0.6067, loss_yns_2: 0.1670, loss_cls_3: 1.1118, loss_box_3: 2.3529, loss_cns_3: 0.6094, loss_yns_3: 0.1653, loss_cls_4: 1.1090, loss_box_4: 2.3037, loss_cns_4: 0.6163, loss_yns_4: 0.1628, loss_cls_5: 1.1265, loss_box_5: 2.2962, loss_cns_5: 0.6166, loss_yns_5: 0.1626, loss_cls_dn_0: 0.4071, loss_box_dn_0: 0.9209, loss_cls_dn_1: 0.3005, loss_box_dn_1: 1.2119, loss_cls_dn_2: 0.3448, loss_box_dn_2: 1.1299, loss_cls_dn_3: 0.3475, loss_box_dn_3: 1.1333, loss_cls_dn_4: 0.3428, loss_box_dn_4: 1.1297, loss_cls_dn_5: 0.3676, loss_box_dn_5: 1.1472, loss_dense_depth: 1.2296, loss: 34.8993, grad_norm: 48.3981 -2025-11-13 15:35:08,798 - mmdet - INFO - Iter [50/17500] lr: 1.196e-04, eta: 18:23:47, time: 1.510, data_time: 0.074, memory: 49163, loss_cls_0: 1.0186, loss_box_0: 1.9960, loss_cns_0: 0.6070, loss_yns_0: 0.1681, loss_cls_1: 1.0638, loss_box_1: 2.4772, loss_cns_1: 0.5854, loss_yns_1: 0.1683, loss_cls_2: 1.0981, loss_box_2: 2.3780, loss_cns_2: 0.6120, loss_yns_2: 0.1652, loss_cls_3: 1.1329, loss_box_3: 2.3534, loss_cns_3: 0.6217, loss_yns_3: 0.1638, loss_cls_4: 1.1316, loss_box_4: 2.3474, loss_cns_4: 0.6238, loss_yns_4: 0.1667, loss_cls_5: 1.1381, loss_box_5: 2.3681, loss_cns_5: 0.6173, loss_yns_5: 0.1625, loss_cls_dn_0: 0.4003, loss_box_dn_0: 0.9371, loss_cls_dn_1: 0.2883, loss_box_dn_1: 1.1565, loss_cls_dn_2: 0.3351, loss_box_dn_2: 1.0785, loss_cls_dn_3: 0.3245, loss_box_dn_3: 1.0699, loss_cls_dn_4: 0.3395, loss_box_dn_4: 1.0652, loss_cls_dn_5: 0.3702, loss_box_dn_5: 1.0859, loss_dense_depth: 1.1804, loss: 34.7962, grad_norm: 49.1250 -2025-11-13 15:35:10,299 - mmdet - INFO - Iter [51/17500] lr: 1.200e-04, eta: 18:10:38, time: 1.501, data_time: 0.077, memory: 49163, loss_cls_0: 0.9850, loss_box_0: 1.9261, loss_cns_0: 0.6106, loss_yns_0: 0.1691, loss_cls_1: 1.0718, loss_box_1: 2.4805, loss_cns_1: 0.5799, loss_yns_1: 0.1706, loss_cls_2: 1.0836, loss_box_2: 2.3230, loss_cns_2: 0.6128, loss_yns_2: 0.1681, loss_cls_3: 1.1295, loss_box_3: 2.3070, loss_cns_3: 0.6240, loss_yns_3: 0.1658, loss_cls_4: 1.1134, loss_box_4: 2.3371, loss_cns_4: 0.6262, loss_yns_4: 0.1714, loss_cls_5: 1.1236, loss_box_5: 2.3205, loss_cns_5: 0.6236, loss_yns_5: 0.1625, loss_cls_dn_0: 0.4025, loss_box_dn_0: 0.9215, loss_cls_dn_1: 0.2657, loss_box_dn_1: 1.2166, loss_cls_dn_2: 0.3206, loss_box_dn_2: 1.1012, loss_cls_dn_3: 0.3066, loss_box_dn_3: 1.0749, loss_cls_dn_4: 0.3309, loss_box_dn_4: 1.0717, loss_cls_dn_5: 0.3483, loss_box_dn_5: 1.0791, loss_dense_depth: 1.0272, loss: 34.3523, grad_norm: 43.5466 -2025-11-13 15:35:11,790 - mmdet - INFO - Iter [52/17500] lr: 1.204e-04, eta: 17:57:56, time: 1.491, data_time: 0.079, memory: 49163, loss_cls_0: 1.0164, loss_box_0: 1.9355, loss_cns_0: 0.6022, loss_yns_0: 0.1682, loss_cls_1: 1.0689, loss_box_1: 2.4252, loss_cns_1: 0.5748, loss_yns_1: 0.1708, loss_cls_2: 1.0954, loss_box_2: 2.2777, loss_cns_2: 0.6099, loss_yns_2: 0.1673, loss_cls_3: 1.1121, loss_box_3: 2.2942, loss_cns_3: 0.6169, loss_yns_3: 0.1672, loss_cls_4: 1.1055, loss_box_4: 2.2899, loss_cns_4: 0.6241, loss_yns_4: 0.1684, loss_cls_5: 1.1288, loss_box_5: 2.2488, loss_cns_5: 0.6223, loss_yns_5: 0.1639, loss_cls_dn_0: 0.4138, loss_box_dn_0: 0.9151, loss_cls_dn_1: 0.2651, loss_box_dn_1: 1.1514, loss_cls_dn_2: 0.3239, loss_box_dn_2: 1.0577, loss_cls_dn_3: 0.3140, loss_box_dn_3: 1.0466, loss_cls_dn_4: 0.3307, loss_box_dn_4: 1.0332, loss_cls_dn_5: 0.3422, loss_box_dn_5: 1.0356, loss_dense_depth: 1.1041, loss: 33.9875, grad_norm: 43.0108 -2025-11-13 15:35:13,284 - mmdet - INFO - Iter [53/17500] lr: 1.208e-04, eta: 17:45:44, time: 1.494, data_time: 0.079, memory: 49163, loss_cls_0: 1.0459, loss_box_0: 2.0058, loss_cns_0: 0.6040, loss_yns_0: 0.1664, loss_cls_1: 1.0675, loss_box_1: 2.4725, loss_cns_1: 0.5719, loss_yns_1: 0.1693, loss_cls_2: 1.0925, loss_box_2: 2.3343, loss_cns_2: 0.6079, loss_yns_2: 0.1680, loss_cls_3: 1.1090, loss_box_3: 2.3466, loss_cns_3: 0.6162, loss_yns_3: 0.1733, loss_cls_4: 1.1092, loss_box_4: 2.3053, loss_cns_4: 0.6198, loss_yns_4: 0.1677, loss_cls_5: 1.1529, loss_box_5: 2.3470, loss_cns_5: 0.6187, loss_yns_5: 0.1649, loss_cls_dn_0: 0.4136, loss_box_dn_0: 0.9073, loss_cls_dn_1: 0.2642, loss_box_dn_1: 1.1153, loss_cls_dn_2: 0.3215, loss_box_dn_2: 1.0388, loss_cls_dn_3: 0.3209, loss_box_dn_3: 1.0337, loss_cls_dn_4: 0.3189, loss_box_dn_4: 1.0209, loss_cls_dn_5: 0.3461, loss_box_dn_5: 1.0590, loss_dense_depth: 1.0474, loss: 34.2444, grad_norm: 48.1013 -2025-11-13 15:35:14,785 - mmdet - INFO - Iter [54/17500] lr: 1.212e-04, eta: 17:34:01, time: 1.501, data_time: 0.077, memory: 49163, loss_cls_0: 1.0203, loss_box_0: 1.9643, loss_cns_0: 0.6185, loss_yns_0: 0.1634, loss_cls_1: 1.0467, loss_box_1: 2.3628, loss_cns_1: 0.6000, loss_yns_1: 0.1641, loss_cls_2: 1.0651, loss_box_2: 2.2840, loss_cns_2: 0.6201, loss_yns_2: 0.1635, loss_cls_3: 1.0882, loss_box_3: 2.2922, loss_cns_3: 0.6198, loss_yns_3: 0.1658, loss_cls_4: 1.1039, loss_box_4: 2.2757, loss_cns_4: 0.6173, loss_yns_4: 0.1630, loss_cls_5: 1.1317, loss_box_5: 2.3384, loss_cns_5: 0.6215, loss_yns_5: 0.1628, loss_cls_dn_0: 0.3827, loss_box_dn_0: 0.9057, loss_cls_dn_1: 0.2555, loss_box_dn_1: 1.1295, loss_cls_dn_2: 0.3013, loss_box_dn_2: 1.0726, loss_cls_dn_3: 0.3164, loss_box_dn_3: 1.0724, loss_cls_dn_4: 0.3015, loss_box_dn_4: 1.0868, loss_cls_dn_5: 0.3350, loss_box_dn_5: 1.1351, loss_dense_depth: 1.0815, loss: 34.0291, grad_norm: 54.7420 -2025-11-13 15:35:16,298 - mmdet - INFO - Iter [55/17500] lr: 1.216e-04, eta: 17:22:48, time: 1.514, data_time: 0.085, memory: 49163, loss_cls_0: 1.0196, loss_box_0: 1.9739, loss_cns_0: 0.6155, loss_yns_0: 0.1664, loss_cls_1: 1.0514, loss_box_1: 2.2830, loss_cns_1: 0.6052, loss_yns_1: 0.1695, loss_cls_2: 1.0702, loss_box_2: 2.2378, loss_cns_2: 0.6171, loss_yns_2: 0.1664, loss_cls_3: 1.0821, loss_box_3: 2.2351, loss_cns_3: 0.6194, loss_yns_3: 0.1671, loss_cls_4: 1.0954, loss_box_4: 2.2255, loss_cns_4: 0.6203, loss_yns_4: 0.1678, loss_cls_5: 1.0935, loss_box_5: 2.2021, loss_cns_5: 0.6225, loss_yns_5: 0.1649, loss_cls_dn_0: 0.3780, loss_box_dn_0: 0.8882, loss_cls_dn_1: 0.2639, loss_box_dn_1: 1.0287, loss_cls_dn_2: 0.3088, loss_box_dn_2: 0.9850, loss_cls_dn_3: 0.3345, loss_box_dn_3: 0.9999, loss_cls_dn_4: 0.3203, loss_box_dn_4: 1.0266, loss_cls_dn_5: 0.3616, loss_box_dn_5: 1.0496, loss_dense_depth: 1.0403, loss: 33.2571, grad_norm: 45.6365 -2025-11-13 15:35:17,811 - mmdet - INFO - Iter [56/17500] lr: 1.220e-04, eta: 17:11:58, time: 1.513, data_time: 0.078, memory: 49163, loss_cls_0: 1.0063, loss_box_0: 1.9257, loss_cns_0: 0.6115, loss_yns_0: 0.1652, loss_cls_1: 1.0401, loss_box_1: 2.1755, loss_cns_1: 0.5956, loss_yns_1: 0.1674, loss_cls_2: 1.0781, loss_box_2: 2.1342, loss_cns_2: 0.6217, loss_yns_2: 0.1672, loss_cls_3: 1.0581, loss_box_3: 2.1185, loss_cns_3: 0.6232, loss_yns_3: 0.1676, loss_cls_4: 1.0621, loss_box_4: 2.1328, loss_cns_4: 0.6218, loss_yns_4: 0.1695, loss_cls_5: 1.0879, loss_box_5: 2.1008, loss_cns_5: 0.6201, loss_yns_5: 0.1670, loss_cls_dn_0: 0.3685, loss_box_dn_0: 0.8850, loss_cls_dn_1: 0.2501, loss_box_dn_1: 1.0471, loss_cls_dn_2: 0.2904, loss_box_dn_2: 1.0055, loss_cls_dn_3: 0.3073, loss_box_dn_3: 1.0082, loss_cls_dn_4: 0.3147, loss_box_dn_4: 1.0338, loss_cls_dn_5: 0.3532, loss_box_dn_5: 1.0407, loss_dense_depth: 1.0036, loss: 32.5259, grad_norm: 56.0491 -2025-11-13 15:35:19,307 - mmdet - INFO - Iter [57/17500] lr: 1.224e-04, eta: 17:01:26, time: 1.495, data_time: 0.076, memory: 49163, loss_cls_0: 1.0053, loss_box_0: 1.8870, loss_cns_0: 0.6158, loss_yns_0: 0.1643, loss_cls_1: 1.0442, loss_box_1: 2.1670, loss_cns_1: 0.5992, loss_yns_1: 0.1658, loss_cls_2: 1.0949, loss_box_2: 2.1488, loss_cns_2: 0.6238, loss_yns_2: 0.1678, loss_cls_3: 1.1094, loss_box_3: 2.1090, loss_cns_3: 0.6231, loss_yns_3: 0.1689, loss_cls_4: 1.0607, loss_box_4: 2.1231, loss_cns_4: 0.6238, loss_yns_4: 0.1671, loss_cls_5: 1.0897, loss_box_5: 2.1294, loss_cns_5: 0.6211, loss_yns_5: 0.1727, loss_cls_dn_0: 0.3723, loss_box_dn_0: 0.8745, loss_cls_dn_1: 0.2397, loss_box_dn_1: 0.9694, loss_cls_dn_2: 0.2713, loss_box_dn_2: 0.9464, loss_cls_dn_3: 0.2788, loss_box_dn_3: 0.9335, loss_cls_dn_4: 0.2984, loss_box_dn_4: 0.9595, loss_cls_dn_5: 0.3356, loss_box_dn_5: 0.9786, loss_dense_depth: 1.0106, loss: 32.1507, grad_norm: 48.7233 -2025-11-13 15:35:20,791 - mmdet - INFO - Iter [58/17500] lr: 1.228e-04, eta: 16:51:13, time: 1.485, data_time: 0.077, memory: 49163, loss_cls_0: 0.9810, loss_box_0: 1.8962, loss_cns_0: 0.6174, loss_yns_0: 0.1641, loss_cls_1: 1.0243, loss_box_1: 2.1268, loss_cns_1: 0.5996, loss_yns_1: 0.1656, loss_cls_2: 1.0487, loss_box_2: 2.0974, loss_cns_2: 0.6269, loss_yns_2: 0.1687, loss_cls_3: 1.0539, loss_box_3: 2.0803, loss_cns_3: 0.6286, loss_yns_3: 0.1675, loss_cls_4: 1.0454, loss_box_4: 2.0994, loss_cns_4: 0.6333, loss_yns_4: 0.1689, loss_cls_5: 1.0964, loss_box_5: 2.0692, loss_cns_5: 0.6327, loss_yns_5: 0.1707, loss_cls_dn_0: 0.3752, loss_box_dn_0: 0.8668, loss_cls_dn_1: 0.2367, loss_box_dn_1: 0.9669, loss_cls_dn_2: 0.2695, loss_box_dn_2: 0.9311, loss_cls_dn_3: 0.2844, loss_box_dn_3: 0.9286, loss_cls_dn_4: 0.2933, loss_box_dn_4: 0.9620, loss_cls_dn_5: 0.3211, loss_box_dn_5: 0.9625, loss_dense_depth: 0.9787, loss: 31.7400, grad_norm: 58.3179 -2025-11-13 15:35:22,275 - mmdet - INFO - Iter [59/17500] lr: 1.232e-04, eta: 16:41:19, time: 1.483, data_time: 0.077, memory: 49163, loss_cls_0: 0.9631, loss_box_0: 1.9181, loss_cns_0: 0.6154, loss_yns_0: 0.1629, loss_cls_1: 1.0201, loss_box_1: 2.1508, loss_cns_1: 0.5969, loss_yns_1: 0.1646, loss_cls_2: 1.0543, loss_box_2: 2.1008, loss_cns_2: 0.6324, loss_yns_2: 0.1670, loss_cls_3: 1.0522, loss_box_3: 2.1093, loss_cns_3: 0.6350, loss_yns_3: 0.1656, loss_cls_4: 1.0650, loss_box_4: 2.0866, loss_cns_4: 0.6398, loss_yns_4: 0.1638, loss_cls_5: 1.0904, loss_box_5: 2.0718, loss_cns_5: 0.6346, loss_yns_5: 0.1642, loss_cls_dn_0: 0.3740, loss_box_dn_0: 0.8611, loss_cls_dn_1: 0.2478, loss_box_dn_1: 0.9737, loss_cls_dn_2: 0.2760, loss_box_dn_2: 0.9212, loss_cls_dn_3: 0.3068, loss_box_dn_3: 0.9331, loss_cls_dn_4: 0.3068, loss_box_dn_4: 0.9434, loss_cls_dn_5: 0.3258, loss_box_dn_5: 0.9461, loss_dense_depth: 0.9604, loss: 31.8008, grad_norm: 43.8469 -2025-11-13 15:35:23,766 - mmdet - INFO - Iter [60/17500] lr: 1.236e-04, eta: 16:31:48, time: 1.491, data_time: 0.087, memory: 49163, loss_cls_0: 0.9747, loss_box_0: 1.8747, loss_cns_0: 0.6193, loss_yns_0: 0.1603, loss_cls_1: 1.0183, loss_box_1: 2.1576, loss_cns_1: 0.6043, loss_yns_1: 0.1638, loss_cls_2: 1.0621, loss_box_2: 2.1059, loss_cns_2: 0.6341, loss_yns_2: 0.1640, loss_cls_3: 1.0622, loss_box_3: 2.1102, loss_cns_3: 0.6367, loss_yns_3: 0.1642, loss_cls_4: 1.0867, loss_box_4: 2.0785, loss_cns_4: 0.6384, loss_yns_4: 0.1663, loss_cls_5: 1.0753, loss_box_5: 2.1336, loss_cns_5: 0.6380, loss_yns_5: 0.1676, loss_cls_dn_0: 0.3634, loss_box_dn_0: 0.8636, loss_cls_dn_1: 0.2396, loss_box_dn_1: 0.9919, loss_cls_dn_2: 0.2661, loss_box_dn_2: 0.9334, loss_cls_dn_3: 0.2997, loss_box_dn_3: 0.9425, loss_cls_dn_4: 0.2993, loss_box_dn_4: 0.9435, loss_cls_dn_5: 0.3405, loss_box_dn_5: 0.9749, loss_dense_depth: 0.9439, loss: 31.8990, grad_norm: 56.0584 -2025-11-13 15:35:25,322 - mmdet - INFO - Iter [61/17500] lr: 1.240e-04, eta: 16:22:54, time: 1.556, data_time: 0.089, memory: 49163, loss_cls_0: 1.0042, loss_box_0: 1.8637, loss_cns_0: 0.6158, loss_yns_0: 0.1603, loss_cls_1: 1.0248, loss_box_1: 2.1617, loss_cns_1: 0.6066, loss_yns_1: 0.1637, loss_cls_2: 1.0717, loss_box_2: 2.1051, loss_cns_2: 0.6281, loss_yns_2: 0.1637, loss_cls_3: 1.0654, loss_box_3: 2.0943, loss_cns_3: 0.6273, loss_yns_3: 0.1633, loss_cls_4: 1.0990, loss_box_4: 2.0719, loss_cns_4: 0.6352, loss_yns_4: 0.1667, loss_cls_5: 1.0888, loss_box_5: 2.1165, loss_cns_5: 0.6374, loss_yns_5: 0.1664, loss_cls_dn_0: 0.3550, loss_box_dn_0: 0.8583, loss_cls_dn_1: 0.2286, loss_box_dn_1: 1.0681, loss_cls_dn_2: 0.2524, loss_box_dn_2: 1.0002, loss_cls_dn_3: 0.2704, loss_box_dn_3: 1.0037, loss_cls_dn_4: 0.2710, loss_box_dn_4: 1.0056, loss_cls_dn_5: 0.3349, loss_box_dn_5: 1.0355, loss_dense_depth: 0.9545, loss: 32.1398, grad_norm: 56.7002 -2025-11-13 15:35:26,952 - mmdet - INFO - Iter [62/17500] lr: 1.244e-04, eta: 16:14:38, time: 1.632, data_time: 0.238, memory: 49163, loss_cls_0: 1.0028, loss_box_0: 1.8681, loss_cns_0: 0.6158, loss_yns_0: 0.1594, loss_cls_1: 1.0383, loss_box_1: 2.1820, loss_cns_1: 0.6105, loss_yns_1: 0.1619, loss_cls_2: 1.0491, loss_box_2: 2.1304, loss_cns_2: 0.6302, loss_yns_2: 0.1623, loss_cls_3: 1.0833, loss_box_3: 2.1355, loss_cns_3: 0.6260, loss_yns_3: 0.1639, loss_cls_4: 1.1329, loss_box_4: 2.1030, loss_cns_4: 0.6402, loss_yns_4: 0.1642, loss_cls_5: 1.0748, loss_box_5: 2.0850, loss_cns_5: 0.6362, loss_yns_5: 0.1618, loss_cls_dn_0: 0.3538, loss_box_dn_0: 0.8592, loss_cls_dn_1: 0.2326, loss_box_dn_1: 1.0390, loss_cls_dn_2: 0.2560, loss_box_dn_2: 0.9753, loss_cls_dn_3: 0.2651, loss_box_dn_3: 0.9870, loss_cls_dn_4: 0.2771, loss_box_dn_4: 0.9865, loss_cls_dn_5: 0.3397, loss_box_dn_5: 1.0020, loss_dense_depth: 0.9856, loss: 32.1766, grad_norm: 47.4545 -2025-11-13 15:35:28,466 - mmdet - INFO - Iter [63/17500] lr: 1.248e-04, eta: 16:06:06, time: 1.514, data_time: 0.074, memory: 49163, loss_cls_0: 0.9798, loss_box_0: 1.8600, loss_cns_0: 0.6103, loss_yns_0: 0.1594, loss_cls_1: 1.0111, loss_box_1: 2.1957, loss_cns_1: 0.6088, loss_yns_1: 0.1629, loss_cls_2: 1.0874, loss_box_2: 2.1392, loss_cns_2: 0.6346, loss_yns_2: 0.1637, loss_cls_3: 1.0683, loss_box_3: 2.1065, loss_cns_3: 0.6388, loss_yns_3: 0.1640, loss_cls_4: 1.1249, loss_box_4: 2.1023, loss_cns_4: 0.6464, loss_yns_4: 0.1619, loss_cls_5: 1.0901, loss_box_5: 2.0916, loss_cns_5: 0.6433, loss_yns_5: 0.1625, loss_cls_dn_0: 0.3632, loss_box_dn_0: 0.8588, loss_cls_dn_1: 0.2302, loss_box_dn_1: 1.0555, loss_cls_dn_2: 0.2604, loss_box_dn_2: 0.9712, loss_cls_dn_3: 0.2697, loss_box_dn_3: 0.9689, loss_cls_dn_4: 0.3099, loss_box_dn_4: 0.9776, loss_cls_dn_5: 0.3201, loss_box_dn_5: 0.9909, loss_dense_depth: 0.9806, loss: 32.1703, grad_norm: 44.3072 -2025-11-13 15:35:30,091 - mmdet - INFO - Iter [64/17500] lr: 1.252e-04, eta: 15:58:19, time: 1.623, data_time: 0.076, memory: 49163, loss_cls_0: 0.9555, loss_box_0: 1.8432, loss_cns_0: 0.6148, loss_yns_0: 0.1594, loss_cls_1: 0.9843, loss_box_1: 2.1551, loss_cns_1: 0.6106, loss_yns_1: 0.1617, loss_cls_2: 1.0465, loss_box_2: 2.1144, loss_cns_2: 0.6332, loss_yns_2: 0.1649, loss_cls_3: 1.0578, loss_box_3: 2.0806, loss_cns_3: 0.6415, loss_yns_3: 0.1654, loss_cls_4: 1.0888, loss_box_4: 2.0993, loss_cns_4: 0.6472, loss_yns_4: 0.1645, loss_cls_5: 1.1253, loss_box_5: 2.0870, loss_cns_5: 0.6438, loss_yns_5: 0.1645, loss_cls_dn_0: 0.3526, loss_box_dn_0: 0.8542, loss_cls_dn_1: 0.2262, loss_box_dn_1: 1.0478, loss_cls_dn_2: 0.2382, loss_box_dn_2: 0.9477, loss_cls_dn_3: 0.2521, loss_box_dn_3: 0.9300, loss_cls_dn_4: 0.2708, loss_box_dn_4: 0.9353, loss_cls_dn_5: 0.2713, loss_box_dn_5: 0.9461, loss_dense_depth: 0.9459, loss: 31.6275, grad_norm: 36.8902 -2025-11-13 15:35:31,598 - mmdet - INFO - Iter [65/17500] lr: 1.256e-04, eta: 15:50:16, time: 1.509, data_time: 0.091, memory: 49163, loss_cls_0: 0.9649, loss_box_0: 1.8580, loss_cns_0: 0.6187, loss_yns_0: 0.1610, loss_cls_1: 0.9892, loss_box_1: 2.1172, loss_cns_1: 0.6063, loss_yns_1: 0.1606, loss_cls_2: 1.0781, loss_box_2: 2.0662, loss_cns_2: 0.6351, loss_yns_2: 0.1629, loss_cls_3: 1.0726, loss_box_3: 2.0607, loss_cns_3: 0.6426, loss_yns_3: 0.1642, loss_cls_4: 1.0927, loss_box_4: 2.0289, loss_cns_4: 0.6501, loss_yns_4: 0.1651, loss_cls_5: 1.1171, loss_box_5: 2.0429, loss_cns_5: 0.6470, loss_yns_5: 0.1625, loss_cls_dn_0: 0.3404, loss_box_dn_0: 0.8567, loss_cls_dn_1: 0.2214, loss_box_dn_1: 1.0504, loss_cls_dn_2: 0.2318, loss_box_dn_2: 0.9437, loss_cls_dn_3: 0.2388, loss_box_dn_3: 0.9322, loss_cls_dn_4: 0.2347, loss_box_dn_4: 0.9147, loss_cls_dn_5: 0.2618, loss_box_dn_5: 0.9253, loss_dense_depth: 0.9113, loss: 31.3280, grad_norm: 57.2736 -2025-11-13 15:35:33,095 - mmdet - INFO - Iter [66/17500] lr: 1.260e-04, eta: 15:42:24, time: 1.496, data_time: 0.074, memory: 49163, loss_cls_0: 0.9660, loss_box_0: 1.8311, loss_cns_0: 0.6173, loss_yns_0: 0.1616, loss_cls_1: 0.9753, loss_box_1: 2.0599, loss_cns_1: 0.5971, loss_yns_1: 0.1602, loss_cls_2: 1.0145, loss_box_2: 2.0514, loss_cns_2: 0.6357, loss_yns_2: 0.1633, loss_cls_3: 1.0684, loss_box_3: 2.0549, loss_cns_3: 0.6389, loss_yns_3: 0.1637, loss_cls_4: 1.0949, loss_box_4: 2.0042, loss_cns_4: 0.6476, loss_yns_4: 0.1654, loss_cls_5: 1.1253, loss_box_5: 2.0147, loss_cns_5: 0.6453, loss_yns_5: 0.1619, loss_cls_dn_0: 0.3244, loss_box_dn_0: 0.8390, loss_cls_dn_1: 0.2066, loss_box_dn_1: 1.0407, loss_cls_dn_2: 0.2180, loss_box_dn_2: 0.9485, loss_cls_dn_3: 0.2425, loss_box_dn_3: 0.9424, loss_cls_dn_4: 0.2480, loss_box_dn_4: 0.9217, loss_cls_dn_5: 0.2873, loss_box_dn_5: 0.9354, loss_dense_depth: 0.9290, loss: 31.1023, grad_norm: 62.5720 -2025-11-13 15:35:34,578 - mmdet - INFO - Iter [67/17500] lr: 1.264e-04, eta: 15:34:42, time: 1.482, data_time: 0.078, memory: 49163, loss_cls_0: 0.9746, loss_box_0: 1.8363, loss_cns_0: 0.6142, loss_yns_0: 0.1627, loss_cls_1: 0.9986, loss_box_1: 2.0272, loss_cns_1: 0.5866, loss_yns_1: 0.1565, loss_cls_2: 1.0063, loss_box_2: 2.0570, loss_cns_2: 0.6282, loss_yns_2: 0.1659, loss_cls_3: 1.0336, loss_box_3: 2.0436, loss_cns_3: 0.6329, loss_yns_3: 0.1616, loss_cls_4: 1.0529, loss_box_4: 2.0055, loss_cns_4: 0.6375, loss_yns_4: 0.1625, loss_cls_5: 1.0897, loss_box_5: 2.0009, loss_cns_5: 0.6377, loss_yns_5: 0.1630, loss_cls_dn_0: 0.3239, loss_box_dn_0: 0.8393, loss_cls_dn_1: 0.2051, loss_box_dn_1: 0.9743, loss_cls_dn_2: 0.2211, loss_box_dn_2: 0.9119, loss_cls_dn_3: 0.2320, loss_box_dn_3: 0.9018, loss_cls_dn_4: 0.2344, loss_box_dn_4: 0.9004, loss_cls_dn_5: 0.2862, loss_box_dn_5: 0.9195, loss_dense_depth: 0.9390, loss: 30.7242, grad_norm: 38.0973 -2025-11-13 15:35:36,069 - mmdet - INFO - Iter [68/17500] lr: 1.268e-04, eta: 15:27:17, time: 1.492, data_time: 0.077, memory: 49163, loss_cls_0: 0.9647, loss_box_0: 1.8443, loss_cns_0: 0.6190, loss_yns_0: 0.1615, loss_cls_1: 0.9853, loss_box_1: 1.9999, loss_cns_1: 0.6063, loss_yns_1: 0.1651, loss_cls_2: 1.0826, loss_box_2: 1.9469, loss_cns_2: 0.6282, loss_yns_2: 0.1647, loss_cls_3: 1.2954, loss_box_3: 1.9385, loss_cns_3: 0.6338, loss_yns_3: 0.1631, loss_cls_4: 1.5197, loss_box_4: 1.9445, loss_cns_4: 0.6385, loss_yns_4: 0.1625, loss_cls_5: 1.0884, loss_box_5: 1.9208, loss_cns_5: 0.6410, loss_yns_5: 0.1651, loss_cls_dn_0: 0.3267, loss_box_dn_0: 0.8547, loss_cls_dn_1: 0.2099, loss_box_dn_1: 0.9375, loss_cls_dn_2: 0.2216, loss_box_dn_2: 0.8947, loss_cls_dn_3: 0.2270, loss_box_dn_3: 0.9014, loss_cls_dn_4: 0.2358, loss_box_dn_4: 0.9405, loss_cls_dn_5: 0.2566, loss_box_dn_5: 0.9652, loss_dense_depth: 0.9267, loss: 31.1781, grad_norm: 150.5974 -2025-11-13 15:35:37,560 - mmdet - INFO - Iter [69/17500] lr: 1.272e-04, eta: 15:20:04, time: 1.491, data_time: 0.073, memory: 49163, loss_cls_0: 0.9310, loss_box_0: 1.8111, loss_cns_0: 0.6176, loss_yns_0: 0.1626, loss_cls_1: 0.9720, loss_box_1: 1.9892, loss_cns_1: 0.6189, loss_yns_1: 0.1654, loss_cls_2: 1.0267, loss_box_2: 1.9329, loss_cns_2: 0.6363, loss_yns_2: 0.1595, loss_cls_3: 1.0936, loss_box_3: 1.9573, loss_cns_3: 0.6408, loss_yns_3: 0.1601, loss_cls_4: 1.1492, loss_box_4: 1.9950, loss_cns_4: 0.6458, loss_yns_4: 0.1617, loss_cls_5: 1.1045, loss_box_5: 1.9970, loss_cns_5: 0.6513, loss_yns_5: 0.1630, loss_cls_dn_0: 0.3219, loss_box_dn_0: 0.8483, loss_cls_dn_1: 0.2108, loss_box_dn_1: 0.9670, loss_cls_dn_2: 0.2203, loss_box_dn_2: 0.9384, loss_cls_dn_3: 0.2257, loss_box_dn_3: 0.9734, loss_cls_dn_4: 0.2416, loss_box_dn_4: 1.0310, loss_cls_dn_5: 0.2834, loss_box_dn_5: 1.0615, loss_dense_depth: 1.0247, loss: 31.0903, grad_norm: 75.7302 -2025-11-13 15:35:39,081 - mmdet - INFO - Iter [70/17500] lr: 1.276e-04, eta: 15:13:11, time: 1.522, data_time: 0.073, memory: 49163, loss_cls_0: 0.9437, loss_box_0: 1.7869, loss_cns_0: 0.6114, loss_yns_0: 0.1612, loss_cls_1: 0.9850, loss_box_1: 2.0390, loss_cns_1: 0.6130, loss_yns_1: 0.1640, loss_cls_2: 1.0414, loss_box_2: 1.9933, loss_cns_2: 0.6307, loss_yns_2: 0.1593, loss_cls_3: 1.0878, loss_box_3: 1.9782, loss_cns_3: 0.6385, loss_yns_3: 0.1603, loss_cls_4: 1.1106, loss_box_4: 2.0094, loss_cns_4: 0.6440, loss_yns_4: 0.1597, loss_cls_5: 1.0954, loss_box_5: 2.0319, loss_cns_5: 0.6477, loss_yns_5: 0.1616, loss_cls_dn_0: 0.3296, loss_box_dn_0: 0.8444, loss_cls_dn_1: 0.2213, loss_box_dn_1: 0.9911, loss_cls_dn_2: 0.2410, loss_box_dn_2: 0.9601, loss_cls_dn_3: 0.2548, loss_box_dn_3: 0.9939, loss_cls_dn_4: 0.2843, loss_box_dn_4: 1.0517, loss_cls_dn_5: 0.3109, loss_box_dn_5: 1.0906, loss_dense_depth: 0.9251, loss: 31.3529, grad_norm: 59.4815 -2025-11-13 15:35:40,562 - mmdet - INFO - Iter [71/17500] lr: 1.280e-04, eta: 15:06:20, time: 1.480, data_time: 0.075, memory: 49163, loss_cls_0: 0.9327, loss_box_0: 1.7917, loss_cns_0: 0.6155, loss_yns_0: 0.1570, loss_cls_1: 0.9811, loss_box_1: 2.1341, loss_cns_1: 0.6083, loss_yns_1: 0.1611, loss_cls_2: 1.0192, loss_box_2: 2.0554, loss_cns_2: 0.6338, loss_yns_2: 0.1589, loss_cls_3: 1.0516, loss_box_3: 1.9916, loss_cns_3: 0.6415, loss_yns_3: 0.1597, loss_cls_4: 1.1157, loss_box_4: 2.0031, loss_cns_4: 0.6476, loss_yns_4: 0.1589, loss_cls_5: 1.0759, loss_box_5: 2.0136, loss_cns_5: 0.6456, loss_yns_5: 0.1593, loss_cls_dn_0: 0.3119, loss_box_dn_0: 0.8459, loss_cls_dn_1: 0.2166, loss_box_dn_1: 1.0571, loss_cls_dn_2: 0.2337, loss_box_dn_2: 1.0108, loss_cls_dn_3: 0.2481, loss_box_dn_3: 1.0203, loss_cls_dn_4: 0.2865, loss_box_dn_4: 1.0616, loss_cls_dn_5: 0.2729, loss_box_dn_5: 1.0790, loss_dense_depth: 0.9203, loss: 31.4776, grad_norm: 46.6870 -2025-11-13 15:35:42,052 - mmdet - INFO - Iter [72/17500] lr: 1.284e-04, eta: 14:59:42, time: 1.491, data_time: 0.076, memory: 49163, loss_cls_0: 0.9711, loss_box_0: 1.7903, loss_cns_0: 0.6181, loss_yns_0: 0.1569, loss_cls_1: 1.0191, loss_box_1: 2.1128, loss_cns_1: 0.6124, loss_yns_1: 0.1606, loss_cls_2: 1.0132, loss_box_2: 2.0399, loss_cns_2: 0.6361, loss_yns_2: 0.1606, loss_cls_3: 1.0291, loss_box_3: 1.9693, loss_cns_3: 0.6454, loss_yns_3: 0.1586, loss_cls_4: 1.0570, loss_box_4: 1.9685, loss_cns_4: 0.6468, loss_yns_4: 0.1587, loss_cls_5: 1.0883, loss_box_5: 1.9543, loss_cns_5: 0.6444, loss_yns_5: 0.1677, loss_cls_dn_0: 0.3007, loss_box_dn_0: 0.8426, loss_cls_dn_1: 0.2197, loss_box_dn_1: 0.9221, loss_cls_dn_2: 0.2279, loss_box_dn_2: 0.8741, loss_cls_dn_3: 0.2303, loss_box_dn_3: 0.8691, loss_cls_dn_4: 0.2549, loss_box_dn_4: 0.8875, loss_cls_dn_5: 0.2513, loss_box_dn_5: 0.8821, loss_dense_depth: 0.9366, loss: 30.4781, grad_norm: 57.8962 -2025-11-13 15:35:43,542 - mmdet - INFO - Iter [73/17500] lr: 1.288e-04, eta: 14:53:16, time: 1.491, data_time: 0.078, memory: 49163, loss_cls_0: 0.9809, loss_box_0: 1.8135, loss_cns_0: 0.6101, loss_yns_0: 0.1573, loss_cls_1: 1.0087, loss_box_1: 2.1152, loss_cns_1: 0.6145, loss_yns_1: 0.1610, loss_cls_2: 1.0251, loss_box_2: 2.0673, loss_cns_2: 0.6351, loss_yns_2: 0.1601, loss_cls_3: 1.0437, loss_box_3: 2.0039, loss_cns_3: 0.6429, loss_yns_3: 0.1587, loss_cls_4: 1.0482, loss_box_4: 1.9972, loss_cns_4: 0.6457, loss_yns_4: 0.1582, loss_cls_5: 1.0521, loss_box_5: 1.9957, loss_cns_5: 0.6452, loss_yns_5: 0.1681, loss_cls_dn_0: 0.2995, loss_box_dn_0: 0.8519, loss_cls_dn_1: 0.2166, loss_box_dn_1: 0.9363, loss_cls_dn_2: 0.2273, loss_box_dn_2: 0.8904, loss_cls_dn_3: 0.2326, loss_box_dn_3: 0.8790, loss_cls_dn_4: 0.2507, loss_box_dn_4: 0.8819, loss_cls_dn_5: 0.2535, loss_box_dn_5: 0.8816, loss_dense_depth: 0.9016, loss: 30.6115, grad_norm: 52.2542 -2025-11-13 15:35:45,046 - mmdet - INFO - Iter [74/17500] lr: 1.292e-04, eta: 14:47:02, time: 1.503, data_time: 0.076, memory: 49163, loss_cls_0: 0.9542, loss_box_0: 1.8162, loss_cns_0: 0.6073, loss_yns_0: 0.1580, loss_cls_1: 0.9994, loss_box_1: 2.1087, loss_cns_1: 0.6153, loss_yns_1: 0.1605, loss_cls_2: 1.0370, loss_box_2: 2.0406, loss_cns_2: 0.6336, loss_yns_2: 0.1617, loss_cls_3: 1.0427, loss_box_3: 2.0278, loss_cns_3: 0.6411, loss_yns_3: 0.1613, loss_cls_4: 1.0504, loss_box_4: 2.0104, loss_cns_4: 0.6448, loss_yns_4: 0.1601, loss_cls_5: 1.0496, loss_box_5: 2.0093, loss_cns_5: 0.6450, loss_yns_5: 0.1644, loss_cls_dn_0: 0.3073, loss_box_dn_0: 0.8555, loss_cls_dn_1: 0.2207, loss_box_dn_1: 0.9572, loss_cls_dn_2: 0.2285, loss_box_dn_2: 0.9026, loss_cls_dn_3: 0.2254, loss_box_dn_3: 0.8953, loss_cls_dn_4: 0.2339, loss_box_dn_4: 0.8877, loss_cls_dn_5: 0.2467, loss_box_dn_5: 0.8921, loss_dense_depth: 0.9113, loss: 30.6634, grad_norm: 44.9302 -2025-11-13 15:35:46,536 - mmdet - INFO - Iter [75/17500] lr: 1.296e-04, eta: 14:40:55, time: 1.488, data_time: 0.081, memory: 49163, loss_cls_0: 0.9580, loss_box_0: 1.8314, loss_cns_0: 0.6055, loss_yns_0: 0.1590, loss_cls_1: 1.0176, loss_box_1: 2.0849, loss_cns_1: 0.6204, loss_yns_1: 0.1631, loss_cls_2: 1.0397, loss_box_2: 2.0009, loss_cns_2: 0.6407, loss_yns_2: 0.1614, loss_cls_3: 1.0479, loss_box_3: 2.0060, loss_cns_3: 0.6473, loss_yns_3: 0.1638, loss_cls_4: 1.1024, loss_box_4: 1.9767, loss_cns_4: 0.6509, loss_yns_4: 0.1631, loss_cls_5: 1.0578, loss_box_5: 1.9666, loss_cns_5: 0.6493, loss_yns_5: 0.1617, loss_cls_dn_0: 0.3055, loss_box_dn_0: 0.8461, loss_cls_dn_1: 0.2201, loss_box_dn_1: 0.9372, loss_cls_dn_2: 0.2241, loss_box_dn_2: 0.8909, loss_cls_dn_3: 0.2175, loss_box_dn_3: 0.8864, loss_cls_dn_4: 0.2257, loss_box_dn_4: 0.8765, loss_cls_dn_5: 0.2342, loss_box_dn_5: 0.8809, loss_dense_depth: 0.8775, loss: 30.4988, grad_norm: 47.6216 -2025-11-13 15:35:48,039 - mmdet - INFO - Iter [76/17500] lr: 1.300e-04, eta: 14:35:02, time: 1.505, data_time: 0.076, memory: 49163, loss_cls_0: 0.9621, loss_box_0: 1.8323, loss_cns_0: 0.6127, loss_yns_0: 0.1627, loss_cls_1: 1.0259, loss_box_1: 2.1040, loss_cns_1: 0.6214, loss_yns_1: 0.1684, loss_cls_2: 1.0380, loss_box_2: 2.0274, loss_cns_2: 0.6442, loss_yns_2: 0.1630, loss_cls_3: 1.0597, loss_box_3: 2.0497, loss_cns_3: 0.6492, loss_yns_3: 0.1658, loss_cls_4: 1.1086, loss_box_4: 2.0175, loss_cns_4: 0.6544, loss_yns_4: 0.1652, loss_cls_5: 1.0679, loss_box_5: 2.0156, loss_cns_5: 0.6531, loss_yns_5: 0.1735, loss_cls_dn_0: 0.2947, loss_box_dn_0: 0.8450, loss_cls_dn_1: 0.2157, loss_box_dn_1: 0.8849, loss_cls_dn_2: 0.2223, loss_box_dn_2: 0.8521, loss_cls_dn_3: 0.2168, loss_box_dn_3: 0.8561, loss_cls_dn_4: 0.2261, loss_box_dn_4: 0.8524, loss_cls_dn_5: 0.2312, loss_box_dn_5: 0.8626, loss_dense_depth: 0.8836, loss: 30.5860, grad_norm: 44.4542 -2025-11-13 15:35:49,528 - mmdet - INFO - Iter [77/17500] lr: 1.304e-04, eta: 14:29:14, time: 1.489, data_time: 0.076, memory: 49163, loss_cls_0: 0.9634, loss_box_0: 1.8995, loss_cns_0: 0.6103, loss_yns_0: 0.1627, loss_cls_1: 1.0141, loss_box_1: 2.0622, loss_cns_1: 0.6286, loss_yns_1: 0.1664, loss_cls_2: 1.0578, loss_box_2: 2.0236, loss_cns_2: 0.6468, loss_yns_2: 0.1628, loss_cls_3: 1.0768, loss_box_3: 2.0394, loss_cns_3: 0.6491, loss_yns_3: 0.1642, loss_cls_4: 1.0691, loss_box_4: 1.9993, loss_cns_4: 0.6577, loss_yns_4: 0.1646, loss_cls_5: 1.0638, loss_box_5: 2.0110, loss_cns_5: 0.6552, loss_yns_5: 0.1813, loss_cls_dn_0: 0.2817, loss_box_dn_0: 0.8400, loss_cls_dn_1: 0.2178, loss_box_dn_1: 0.8258, loss_cls_dn_2: 0.2247, loss_box_dn_2: 0.8100, loss_cls_dn_3: 0.2236, loss_box_dn_3: 0.8167, loss_cls_dn_4: 0.2305, loss_box_dn_4: 0.8178, loss_cls_dn_5: 0.2318, loss_box_dn_5: 0.8366, loss_dense_depth: 0.8757, loss: 30.3626, grad_norm: 43.1709 -2025-11-13 15:35:51,012 - mmdet - INFO - Iter [78/17500] lr: 1.308e-04, eta: 14:23:34, time: 1.484, data_time: 0.074, memory: 49163, loss_cls_0: 0.9658, loss_box_0: 1.9181, loss_cns_0: 0.6114, loss_yns_0: 0.1636, loss_cls_1: 1.0017, loss_box_1: 2.1603, loss_cns_1: 0.6152, loss_yns_1: 0.1649, loss_cls_2: 1.0692, loss_box_2: 2.0687, loss_cns_2: 0.6425, loss_yns_2: 0.1635, loss_cls_3: 1.0717, loss_box_3: 2.0530, loss_cns_3: 0.6462, loss_yns_3: 0.1653, loss_cls_4: 1.0681, loss_box_4: 2.0283, loss_cns_4: 0.6531, loss_yns_4: 0.1682, loss_cls_5: 1.0636, loss_box_5: 2.0512, loss_cns_5: 0.6477, loss_yns_5: 0.1737, loss_cls_dn_0: 0.2815, loss_box_dn_0: 0.8385, loss_cls_dn_1: 0.2086, loss_box_dn_1: 0.8631, loss_cls_dn_2: 0.2167, loss_box_dn_2: 0.8243, loss_cls_dn_3: 0.2140, loss_box_dn_3: 0.8272, loss_cls_dn_4: 0.2190, loss_box_dn_4: 0.8324, loss_cls_dn_5: 0.2221, loss_box_dn_5: 0.8528, loss_dense_depth: 0.8793, loss: 30.6146, grad_norm: 39.1610 -2025-11-13 15:35:52,511 - mmdet - INFO - Iter [79/17500] lr: 1.312e-04, eta: 14:18:05, time: 1.498, data_time: 0.076, memory: 49163, loss_cls_0: 0.9348, loss_box_0: 1.8735, loss_cns_0: 0.6208, loss_yns_0: 0.1625, loss_cls_1: 0.9940, loss_box_1: 2.0852, loss_cns_1: 0.6277, loss_yns_1: 0.1678, loss_cls_2: 1.0417, loss_box_2: 2.0499, loss_cns_2: 0.6427, loss_yns_2: 0.1628, loss_cls_3: 1.0537, loss_box_3: 2.0465, loss_cns_3: 0.6443, loss_yns_3: 0.1651, loss_cls_4: 1.0991, loss_box_4: 1.9940, loss_cns_4: 0.6548, loss_yns_4: 0.1651, loss_cls_5: 1.0591, loss_box_5: 1.9910, loss_cns_5: 0.6519, loss_yns_5: 0.1648, loss_cls_dn_0: 0.2764, loss_box_dn_0: 0.8344, loss_cls_dn_1: 0.2001, loss_box_dn_1: 0.8600, loss_cls_dn_2: 0.2048, loss_box_dn_2: 0.8417, loss_cls_dn_3: 0.2036, loss_box_dn_3: 0.8514, loss_cls_dn_4: 0.2214, loss_box_dn_4: 0.8429, loss_cls_dn_5: 0.2215, loss_box_dn_5: 0.8545, loss_dense_depth: 0.8287, loss: 30.2942, grad_norm: 38.7498 -2025-11-13 15:35:53,999 - mmdet - INFO - Iter [80/17500] lr: 1.316e-04, eta: 14:12:43, time: 1.489, data_time: 0.086, memory: 49163, loss_cls_0: 0.9374, loss_box_0: 1.8429, loss_cns_0: 0.6172, loss_yns_0: 0.1588, loss_cls_1: 0.9807, loss_box_1: 2.1007, loss_cns_1: 0.6303, loss_yns_1: 0.1661, loss_cls_2: 1.0211, loss_box_2: 2.0762, loss_cns_2: 0.6428, loss_yns_2: 0.1617, loss_cls_3: 1.0446, loss_box_3: 2.0734, loss_cns_3: 0.6499, loss_yns_3: 0.1643, loss_cls_4: 1.0997, loss_box_4: 2.0451, loss_cns_4: 0.6571, loss_yns_4: 0.1634, loss_cls_5: 1.0534, loss_box_5: 2.0355, loss_cns_5: 0.6572, loss_yns_5: 0.1725, loss_cls_dn_0: 0.2720, loss_box_dn_0: 0.8243, loss_cls_dn_1: 0.1996, loss_box_dn_1: 0.8617, loss_cls_dn_2: 0.2028, loss_box_dn_2: 0.8522, loss_cls_dn_3: 0.2057, loss_box_dn_3: 0.8543, loss_cls_dn_4: 0.2199, loss_box_dn_4: 0.8489, loss_cls_dn_5: 0.2193, loss_box_dn_5: 0.8564, loss_dense_depth: 0.8426, loss: 30.4116, grad_norm: 47.7075 -2025-11-13 15:35:55,565 - mmdet - INFO - Iter [81/17500] lr: 1.320e-04, eta: 14:07:45, time: 1.565, data_time: 0.072, memory: 49163, loss_cls_0: 0.9497, loss_box_0: 1.8366, loss_cns_0: 0.6112, loss_yns_0: 0.1578, loss_cls_1: 0.9869, loss_box_1: 2.1008, loss_cns_1: 0.6251, loss_yns_1: 0.1632, loss_cls_2: 1.0523, loss_box_2: 2.0599, loss_cns_2: 0.6451, loss_yns_2: 0.1624, loss_cls_3: 1.0760, loss_box_3: 2.0482, loss_cns_3: 0.6495, loss_yns_3: 0.1629, loss_cls_4: 1.0532, loss_box_4: 2.0460, loss_cns_4: 0.6555, loss_yns_4: 0.1645, loss_cls_5: 1.0545, loss_box_5: 2.0498, loss_cns_5: 0.6537, loss_yns_5: 0.1743, loss_cls_dn_0: 0.2697, loss_box_dn_0: 0.8236, loss_cls_dn_1: 0.2015, loss_box_dn_1: 0.8854, loss_cls_dn_2: 0.2028, loss_box_dn_2: 0.8633, loss_cls_dn_3: 0.2063, loss_box_dn_3: 0.8577, loss_cls_dn_4: 0.2139, loss_box_dn_4: 0.8585, loss_cls_dn_5: 0.2178, loss_box_dn_5: 0.8665, loss_dense_depth: 0.8715, loss: 30.4777, grad_norm: 57.8432 -2025-11-13 15:35:57,113 - mmdet - INFO - Iter [82/17500] lr: 1.324e-04, eta: 14:02:51, time: 1.548, data_time: 0.150, memory: 49163, loss_cls_0: 0.9601, loss_box_0: 1.8253, loss_cns_0: 0.6109, loss_yns_0: 0.1558, loss_cls_1: 1.0084, loss_box_1: 2.0775, loss_cns_1: 0.6225, loss_yns_1: 0.1621, loss_cls_2: 1.0397, loss_box_2: 2.0402, loss_cns_2: 0.6443, loss_yns_2: 0.1617, loss_cls_3: 1.0786, loss_box_3: 2.0447, loss_cns_3: 0.6434, loss_yns_3: 0.1605, loss_cls_4: 1.0662, loss_box_4: 2.0164, loss_cns_4: 0.6514, loss_yns_4: 0.1653, loss_cls_5: 1.0571, loss_box_5: 2.0252, loss_cns_5: 0.6462, loss_yns_5: 0.1677, loss_cls_dn_0: 0.2679, loss_box_dn_0: 0.8292, loss_cls_dn_1: 0.2012, loss_box_dn_1: 0.8877, loss_cls_dn_2: 0.1945, loss_box_dn_2: 0.8562, loss_cls_dn_3: 0.1983, loss_box_dn_3: 0.8542, loss_cls_dn_4: 0.2075, loss_box_dn_4: 0.8480, loss_cls_dn_5: 0.2120, loss_box_dn_5: 0.8566, loss_dense_depth: 0.8984, loss: 30.3429, grad_norm: 38.3140 -2025-11-13 15:35:58,645 - mmdet - INFO - Iter [83/17500] lr: 1.328e-04, eta: 13:58:00, time: 1.533, data_time: 0.076, memory: 49163, loss_cls_0: 0.9529, loss_box_0: 1.8245, loss_cns_0: 0.6175, loss_yns_0: 0.1616, loss_cls_1: 1.0130, loss_box_1: 2.0441, loss_cns_1: 0.6175, loss_yns_1: 0.1641, loss_cls_2: 1.0387, loss_box_2: 1.9692, loss_cns_2: 0.6397, loss_yns_2: 0.1641, loss_cls_3: 1.0579, loss_box_3: 1.9527, loss_cns_3: 0.6456, loss_yns_3: 0.1635, loss_cls_4: 1.0562, loss_box_4: 1.9429, loss_cns_4: 0.6477, loss_yns_4: 0.1635, loss_cls_5: 1.0657, loss_box_5: 1.9594, loss_cns_5: 0.6448, loss_yns_5: 0.1601, loss_cls_dn_0: 0.2634, loss_box_dn_0: 0.8308, loss_cls_dn_1: 0.1968, loss_box_dn_1: 0.8837, loss_cls_dn_2: 0.1952, loss_box_dn_2: 0.8339, loss_cls_dn_3: 0.1954, loss_box_dn_3: 0.8364, loss_cls_dn_4: 0.2086, loss_box_dn_4: 0.8437, loss_cls_dn_5: 0.2194, loss_box_dn_5: 0.8513, loss_dense_depth: 0.8641, loss: 29.8895, grad_norm: 31.7391 -2025-11-13 15:36:00,168 - mmdet - INFO - Iter [84/17500] lr: 1.332e-04, eta: 13:53:15, time: 1.523, data_time: 0.079, memory: 49163, loss_cls_0: 0.9040, loss_box_0: 1.8197, loss_cns_0: 0.6225, loss_yns_0: 0.1582, loss_cls_1: 0.9467, loss_box_1: 2.0124, loss_cns_1: 0.6252, loss_yns_1: 0.1626, loss_cls_2: 1.0134, loss_box_2: 1.9447, loss_cns_2: 0.6435, loss_yns_2: 0.1631, loss_cls_3: 1.0282, loss_box_3: 1.9205, loss_cns_3: 0.6499, loss_yns_3: 0.1647, loss_cls_4: 1.0157, loss_box_4: 1.9395, loss_cns_4: 0.6483, loss_yns_4: 0.1611, loss_cls_5: 1.0227, loss_box_5: 1.9747, loss_cns_5: 0.6474, loss_yns_5: 0.1575, loss_cls_dn_0: 0.2481, loss_box_dn_0: 0.8212, loss_cls_dn_1: 0.1977, loss_box_dn_1: 0.8309, loss_cls_dn_2: 0.2057, loss_box_dn_2: 0.7921, loss_cls_dn_3: 0.2056, loss_box_dn_3: 0.8022, loss_cls_dn_4: 0.2160, loss_box_dn_4: 0.8267, loss_cls_dn_5: 0.2265, loss_box_dn_5: 0.8388, loss_dense_depth: 0.8515, loss: 29.4093, grad_norm: 46.0256 -2025-11-13 15:36:01,691 - mmdet - INFO - Iter [85/17500] lr: 1.336e-04, eta: 13:48:36, time: 1.523, data_time: 0.089, memory: 49163, loss_cls_0: 0.9418, loss_box_0: 1.8304, loss_cns_0: 0.6190, loss_yns_0: 0.1594, loss_cls_1: 0.9934, loss_box_1: 2.0576, loss_cns_1: 0.6180, loss_yns_1: 0.1608, loss_cls_2: 1.0186, loss_box_2: 1.9941, loss_cns_2: 0.6363, loss_yns_2: 0.1644, loss_cls_3: 1.0356, loss_box_3: 1.9553, loss_cns_3: 0.6436, loss_yns_3: 0.1637, loss_cls_4: 1.0374, loss_box_4: 1.9570, loss_cns_4: 0.6454, loss_yns_4: 0.1612, loss_cls_5: 1.0335, loss_box_5: 1.9732, loss_cns_5: 0.6435, loss_yns_5: 0.1592, loss_cls_dn_0: 0.2618, loss_box_dn_0: 0.8360, loss_cls_dn_1: 0.1988, loss_box_dn_1: 0.8683, loss_cls_dn_2: 0.2058, loss_box_dn_2: 0.8245, loss_cls_dn_3: 0.2057, loss_box_dn_3: 0.8240, loss_cls_dn_4: 0.2164, loss_box_dn_4: 0.8366, loss_cls_dn_5: 0.2227, loss_box_dn_5: 0.8401, loss_dense_depth: 0.8723, loss: 29.8155, grad_norm: 43.0271 -2025-11-13 15:36:03,223 - mmdet - INFO - Iter [86/17500] lr: 1.340e-04, eta: 13:44:05, time: 1.531, data_time: 0.077, memory: 49163, loss_cls_0: 0.9274, loss_box_0: 1.8079, loss_cns_0: 0.6161, loss_yns_0: 0.1600, loss_cls_1: 0.9765, loss_box_1: 1.9257, loss_cns_1: 0.6218, loss_yns_1: 0.1639, loss_cls_2: 1.0015, loss_box_2: 1.8589, loss_cns_2: 0.6412, loss_yns_2: 0.1676, loss_cls_3: 1.0227, loss_box_3: 1.8376, loss_cns_3: 0.6458, loss_yns_3: 0.1637, loss_cls_4: 1.0253, loss_box_4: 1.8156, loss_cns_4: 0.6536, loss_yns_4: 0.1642, loss_cls_5: 1.0289, loss_box_5: 1.8319, loss_cns_5: 0.6465, loss_yns_5: 0.1671, loss_cls_dn_0: 0.2541, loss_box_dn_0: 0.8291, loss_cls_dn_1: 0.1965, loss_box_dn_1: 0.8423, loss_cls_dn_2: 0.2019, loss_box_dn_2: 0.7972, loss_cls_dn_3: 0.2029, loss_box_dn_3: 0.7921, loss_cls_dn_4: 0.2132, loss_box_dn_4: 0.7926, loss_cls_dn_5: 0.2170, loss_box_dn_5: 0.8046, loss_dense_depth: 0.8644, loss: 28.8791, grad_norm: 29.6587 -2025-11-13 15:36:04,721 - mmdet - INFO - Iter [87/17500] lr: 1.344e-04, eta: 13:39:34, time: 1.499, data_time: 0.077, memory: 49163, loss_cls_0: 0.9078, loss_box_0: 1.8213, loss_cns_0: 0.6175, loss_yns_0: 0.1602, loss_cls_1: 0.9599, loss_box_1: 1.9598, loss_cns_1: 0.6198, loss_yns_1: 0.1635, loss_cls_2: 1.0156, loss_box_2: 1.8494, loss_cns_2: 0.6488, loss_yns_2: 0.1629, loss_cls_3: 1.0224, loss_box_3: 1.8403, loss_cns_3: 0.6511, loss_yns_3: 0.1608, loss_cls_4: 1.0321, loss_box_4: 1.8240, loss_cns_4: 0.6624, loss_yns_4: 0.1645, loss_cls_5: 1.0237, loss_box_5: 1.8366, loss_cns_5: 0.6548, loss_yns_5: 0.1647, loss_cls_dn_0: 0.2495, loss_box_dn_0: 0.8218, loss_cls_dn_1: 0.1977, loss_box_dn_1: 0.8337, loss_cls_dn_2: 0.2029, loss_box_dn_2: 0.7809, loss_cls_dn_3: 0.2011, loss_box_dn_3: 0.7733, loss_cls_dn_4: 0.2142, loss_box_dn_4: 0.7689, loss_cls_dn_5: 0.2160, loss_box_dn_5: 0.7854, loss_dense_depth: 0.8805, loss: 28.8497, grad_norm: 38.7696 -2025-11-13 15:36:06,210 - mmdet - INFO - Iter [88/17500] lr: 1.348e-04, eta: 13:35:06, time: 1.488, data_time: 0.078, memory: 49163, loss_cls_0: 0.9378, loss_box_0: 1.8284, loss_cns_0: 0.6166, loss_yns_0: 0.1594, loss_cls_1: 0.9831, loss_box_1: 1.9701, loss_cns_1: 0.6212, loss_yns_1: 0.1618, loss_cls_2: 1.0226, loss_box_2: 1.8744, loss_cns_2: 0.6450, loss_yns_2: 0.1600, loss_cls_3: 1.0234, loss_box_3: 1.8499, loss_cns_3: 0.6505, loss_yns_3: 0.1602, loss_cls_4: 1.0392, loss_box_4: 1.8398, loss_cns_4: 0.6590, loss_yns_4: 0.1624, loss_cls_5: 1.0357, loss_box_5: 1.8482, loss_cns_5: 0.6545, loss_yns_5: 0.1642, loss_cls_dn_0: 0.2502, loss_box_dn_0: 0.8313, loss_cls_dn_1: 0.1980, loss_box_dn_1: 0.8315, loss_cls_dn_2: 0.2007, loss_box_dn_2: 0.7828, loss_cls_dn_3: 0.1972, loss_box_dn_3: 0.7672, loss_cls_dn_4: 0.2103, loss_box_dn_4: 0.7656, loss_cls_dn_5: 0.2149, loss_box_dn_5: 0.7801, loss_dense_depth: 0.8361, loss: 28.9331, grad_norm: 39.4127 -2025-11-13 15:36:07,713 - mmdet - INFO - Iter [89/17500] lr: 1.352e-04, eta: 13:30:48, time: 1.505, data_time: 0.077, memory: 49163, loss_cls_0: 0.9214, loss_box_0: 1.8004, loss_cns_0: 0.6120, loss_yns_0: 0.1577, loss_cls_1: 0.9970, loss_box_1: 1.9618, loss_cns_1: 0.6196, loss_yns_1: 0.1591, loss_cls_2: 1.0153, loss_box_2: 1.8823, loss_cns_2: 0.6397, loss_yns_2: 0.1581, loss_cls_3: 1.0228, loss_box_3: 1.8513, loss_cns_3: 0.6470, loss_yns_3: 0.1584, loss_cls_4: 1.0296, loss_box_4: 1.8656, loss_cns_4: 0.6476, loss_yns_4: 0.1602, loss_cls_5: 1.0303, loss_box_5: 1.8997, loss_cns_5: 0.6436, loss_yns_5: 0.1588, loss_cls_dn_0: 0.2524, loss_box_dn_0: 0.8324, loss_cls_dn_1: 0.2015, loss_box_dn_1: 0.8240, loss_cls_dn_2: 0.1998, loss_box_dn_2: 0.7767, loss_cls_dn_3: 0.1958, loss_box_dn_3: 0.7633, loss_cls_dn_4: 0.2085, loss_box_dn_4: 0.7751, loss_cls_dn_5: 0.2147, loss_box_dn_5: 0.7945, loss_dense_depth: 0.8544, loss: 28.9324, grad_norm: 36.6414 -2025-11-13 15:36:09,214 - mmdet - INFO - Iter [90/17500] lr: 1.356e-04, eta: 13:26:35, time: 1.501, data_time: 0.073, memory: 49163, loss_cls_0: 0.9368, loss_box_0: 1.7710, loss_cns_0: 0.6084, loss_yns_0: 0.1555, loss_cls_1: 1.0033, loss_box_1: 1.9855, loss_cns_1: 0.6289, loss_yns_1: 0.1593, loss_cls_2: 1.0228, loss_box_2: 1.9046, loss_cns_2: 0.6450, loss_yns_2: 0.1638, loss_cls_3: 1.0360, loss_box_3: 1.8960, loss_cns_3: 0.6493, loss_yns_3: 0.1581, loss_cls_4: 1.0302, loss_box_4: 1.8762, loss_cns_4: 0.6530, loss_yns_4: 0.1606, loss_cls_5: 1.0333, loss_box_5: 1.9081, loss_cns_5: 0.6507, loss_yns_5: 0.1587, loss_cls_dn_0: 0.2538, loss_box_dn_0: 0.8333, loss_cls_dn_1: 0.2001, loss_box_dn_1: 0.8281, loss_cls_dn_2: 0.1981, loss_box_dn_2: 0.7857, loss_cls_dn_3: 0.1932, loss_box_dn_3: 0.7860, loss_cls_dn_4: 0.2039, loss_box_dn_4: 0.7867, loss_cls_dn_5: 0.2127, loss_box_dn_5: 0.8043, loss_dense_depth: 0.8719, loss: 29.1530, grad_norm: 32.8989 -2025-11-13 15:36:10,698 - mmdet - INFO - Iter [91/17500] lr: 1.360e-04, eta: 13:22:25, time: 1.484, data_time: 0.074, memory: 49163, loss_cls_0: 0.9440, loss_box_0: 1.7798, loss_cns_0: 0.6195, loss_yns_0: 0.1547, loss_cls_1: 0.9840, loss_box_1: 1.9601, loss_cns_1: 0.6324, loss_yns_1: 0.1571, loss_cls_2: 1.0200, loss_box_2: 1.8919, loss_cns_2: 0.6455, loss_yns_2: 0.1585, loss_cls_3: 1.0253, loss_box_3: 1.8926, loss_cns_3: 0.6482, loss_yns_3: 0.1545, loss_cls_4: 1.0252, loss_box_4: 1.8768, loss_cns_4: 0.6524, loss_yns_4: 0.1570, loss_cls_5: 1.0329, loss_box_5: 1.8840, loss_cns_5: 0.6475, loss_yns_5: 0.1614, loss_cls_dn_0: 0.2449, loss_box_dn_0: 0.8099, loss_cls_dn_1: 0.1954, loss_box_dn_1: 0.8297, loss_cls_dn_2: 0.1963, loss_box_dn_2: 0.8004, loss_cls_dn_3: 0.1942, loss_box_dn_3: 0.8038, loss_cls_dn_4: 0.2038, loss_box_dn_4: 0.8039, loss_cls_dn_5: 0.2105, loss_box_dn_5: 0.8167, loss_dense_depth: 0.8484, loss: 29.0632, grad_norm: 35.9735 -2025-11-13 15:36:12,184 - mmdet - INFO - Iter [92/17500] lr: 1.364e-04, eta: 13:18:20, time: 1.486, data_time: 0.073, memory: 49163, loss_cls_0: 0.9178, loss_box_0: 1.7770, loss_cns_0: 0.6177, loss_yns_0: 0.1509, loss_cls_1: 0.9901, loss_box_1: 1.9398, loss_cns_1: 0.6338, loss_yns_1: 0.1560, loss_cls_2: 1.0175, loss_box_2: 1.8854, loss_cns_2: 0.6456, loss_yns_2: 0.1550, loss_cls_3: 1.0022, loss_box_3: 1.8725, loss_cns_3: 0.6502, loss_yns_3: 0.1543, loss_cls_4: 1.0098, loss_box_4: 1.8668, loss_cns_4: 0.6540, loss_yns_4: 0.1558, loss_cls_5: 1.0174, loss_box_5: 1.8563, loss_cns_5: 0.6499, loss_yns_5: 0.1647, loss_cls_dn_0: 0.2483, loss_box_dn_0: 0.8121, loss_cls_dn_1: 0.1900, loss_box_dn_1: 0.8196, loss_cls_dn_2: 0.1911, loss_box_dn_2: 0.7829, loss_cls_dn_3: 0.1888, loss_box_dn_3: 0.7764, loss_cls_dn_4: 0.1985, loss_box_dn_4: 0.7778, loss_cls_dn_5: 0.2051, loss_box_dn_5: 0.7862, loss_dense_depth: 0.8807, loss: 28.7981, grad_norm: 33.3827 -2025-11-13 15:36:13,665 - mmdet - INFO - Iter [93/17500] lr: 1.368e-04, eta: 13:14:19, time: 1.481, data_time: 0.076, memory: 49163, loss_cls_0: 0.9367, loss_box_0: 1.7515, loss_cns_0: 0.6184, loss_yns_0: 0.1495, loss_cls_1: 1.0051, loss_box_1: 1.9388, loss_cns_1: 0.6387, loss_yns_1: 0.1559, loss_cls_2: 1.0321, loss_box_2: 1.8768, loss_cns_2: 0.6488, loss_yns_2: 0.1538, loss_cls_3: 1.0291, loss_box_3: 1.8642, loss_cns_3: 0.6558, loss_yns_3: 0.1542, loss_cls_4: 1.0242, loss_box_4: 1.8481, loss_cns_4: 0.6548, loss_yns_4: 0.1551, loss_cls_5: 1.0257, loss_box_5: 1.8355, loss_cns_5: 0.6521, loss_yns_5: 0.1573, loss_cls_dn_0: 0.2488, loss_box_dn_0: 0.8119, loss_cls_dn_1: 0.1867, loss_box_dn_1: 0.8137, loss_cls_dn_2: 0.1874, loss_box_dn_2: 0.7700, loss_cls_dn_3: 0.1865, loss_box_dn_3: 0.7588, loss_cls_dn_4: 0.1957, loss_box_dn_4: 0.7573, loss_cls_dn_5: 0.2007, loss_box_dn_5: 0.7607, loss_dense_depth: 0.8127, loss: 28.6533, grad_norm: 32.3101 -2025-11-13 15:36:15,155 - mmdet - INFO - Iter [94/17500] lr: 1.372e-04, eta: 13:10:25, time: 1.490, data_time: 0.071, memory: 49163, loss_cls_0: 0.9407, loss_box_0: 1.7807, loss_cns_0: 0.6187, loss_yns_0: 0.1511, loss_cls_1: 1.0026, loss_box_1: 1.9439, loss_cns_1: 0.6358, loss_yns_1: 0.1550, loss_cls_2: 1.0438, loss_box_2: 1.8882, loss_cns_2: 0.6460, loss_yns_2: 0.1568, loss_cls_3: 1.0379, loss_box_3: 1.8776, loss_cns_3: 0.6532, loss_yns_3: 0.1546, loss_cls_4: 1.0329, loss_box_4: 1.8611, loss_cns_4: 0.6532, loss_yns_4: 0.1564, loss_cls_5: 1.0417, loss_box_5: 1.8618, loss_cns_5: 0.6507, loss_yns_5: 0.1562, loss_cls_dn_0: 0.2478, loss_box_dn_0: 0.8055, loss_cls_dn_1: 0.1876, loss_box_dn_1: 0.7976, loss_cls_dn_2: 0.1927, loss_box_dn_2: 0.7617, loss_cls_dn_3: 0.1886, loss_box_dn_3: 0.7564, loss_cls_dn_4: 0.1998, loss_box_dn_4: 0.7583, loss_cls_dn_5: 0.2079, loss_box_dn_5: 0.7641, loss_dense_depth: 0.8664, loss: 28.8350, grad_norm: 35.6308 -2025-11-13 15:36:16,653 - mmdet - INFO - Iter [95/17500] lr: 1.376e-04, eta: 13:06:38, time: 1.498, data_time: 0.082, memory: 49163, loss_cls_0: 0.9410, loss_box_0: 1.8026, loss_cns_0: 0.6203, loss_yns_0: 0.1529, loss_cls_1: 1.0073, loss_box_1: 1.8759, loss_cns_1: 0.6425, loss_yns_1: 0.1555, loss_cls_2: 1.0538, loss_box_2: 1.8174, loss_cns_2: 0.6543, loss_yns_2: 0.1548, loss_cls_3: 1.0388, loss_box_3: 1.8179, loss_cns_3: 0.6557, loss_yns_3: 0.1541, loss_cls_4: 1.0392, loss_box_4: 1.8083, loss_cns_4: 0.6577, loss_yns_4: 0.1557, loss_cls_5: 1.0564, loss_box_5: 1.8044, loss_cns_5: 0.6570, loss_yns_5: 0.1607, loss_cls_dn_0: 0.2418, loss_box_dn_0: 0.8065, loss_cls_dn_1: 0.1838, loss_box_dn_1: 0.7791, loss_cls_dn_2: 0.1851, loss_box_dn_2: 0.7537, loss_cls_dn_3: 0.1858, loss_box_dn_3: 0.7651, loss_cls_dn_4: 0.1969, loss_box_dn_4: 0.7731, loss_cls_dn_5: 0.2036, loss_box_dn_5: 0.7819, loss_dense_depth: 0.8567, loss: 28.5971, grad_norm: 32.8336 -2025-11-13 15:36:18,144 - mmdet - INFO - Iter [96/17500] lr: 1.380e-04, eta: 13:02:54, time: 1.491, data_time: 0.078, memory: 49163, loss_cls_0: 0.9491, loss_box_0: 1.8277, loss_cns_0: 0.6164, loss_yns_0: 0.1558, loss_cls_1: 1.0026, loss_box_1: 1.8769, loss_cns_1: 0.6391, loss_yns_1: 0.1546, loss_cls_2: 1.0436, loss_box_2: 1.8273, loss_cns_2: 0.6485, loss_yns_2: 0.1558, loss_cls_3: 1.0356, loss_box_3: 1.8289, loss_cns_3: 0.6504, loss_yns_3: 0.1577, loss_cls_4: 1.0688, loss_box_4: 1.8105, loss_cns_4: 0.6515, loss_yns_4: 0.1584, loss_cls_5: 1.0531, loss_box_5: 1.8088, loss_cns_5: 0.6502, loss_yns_5: 0.1665, loss_cls_dn_0: 0.2481, loss_box_dn_0: 0.8144, loss_cls_dn_1: 0.1832, loss_box_dn_1: 0.7987, loss_cls_dn_2: 0.1826, loss_box_dn_2: 0.7734, loss_cls_dn_3: 0.1828, loss_box_dn_3: 0.7867, loss_cls_dn_4: 0.1980, loss_box_dn_4: 0.7910, loss_cls_dn_5: 0.2063, loss_box_dn_5: 0.8004, loss_dense_depth: 0.8591, loss: 28.7622, grad_norm: 35.4920 -2025-11-13 15:36:19,633 - mmdet - INFO - Iter [97/17500] lr: 1.384e-04, eta: 12:59:14, time: 1.490, data_time: 0.075, memory: 49163, loss_cls_0: 0.9537, loss_box_0: 1.7809, loss_cns_0: 0.6169, loss_yns_0: 0.1522, loss_cls_1: 0.9958, loss_box_1: 1.8513, loss_cns_1: 0.6326, loss_yns_1: 0.1555, loss_cls_2: 1.0452, loss_box_2: 1.7998, loss_cns_2: 0.6499, loss_yns_2: 0.1566, loss_cls_3: 1.0475, loss_box_3: 1.7917, loss_cns_3: 0.6532, loss_yns_3: 0.1578, loss_cls_4: 1.0586, loss_box_4: 1.7824, loss_cns_4: 0.6525, loss_yns_4: 0.1568, loss_cls_5: 1.0523, loss_box_5: 1.7873, loss_cns_5: 0.6496, loss_yns_5: 0.1609, loss_cls_dn_0: 0.2532, loss_box_dn_0: 0.8125, loss_cls_dn_1: 0.1850, loss_box_dn_1: 0.7994, loss_cls_dn_2: 0.1833, loss_box_dn_2: 0.7711, loss_cls_dn_3: 0.1848, loss_box_dn_3: 0.7740, loss_cls_dn_4: 0.1956, loss_box_dn_4: 0.7779, loss_cls_dn_5: 0.2026, loss_box_dn_5: 0.7871, loss_dense_depth: 0.8599, loss: 28.5275, grad_norm: 32.4983 -2025-11-13 15:36:21,103 - mmdet - INFO - Iter [98/17500] lr: 1.388e-04, eta: 12:55:35, time: 1.469, data_time: 0.071, memory: 49163, loss_cls_0: 0.9379, loss_box_0: 1.7566, loss_cns_0: 0.6195, loss_yns_0: 0.1512, loss_cls_1: 0.9876, loss_box_1: 1.8520, loss_cns_1: 0.6324, loss_yns_1: 0.1529, loss_cls_2: 1.0290, loss_box_2: 1.8049, loss_cns_2: 0.6507, loss_yns_2: 0.1565, loss_cls_3: 1.0419, loss_box_3: 1.8109, loss_cns_3: 0.6535, loss_yns_3: 0.1560, loss_cls_4: 1.0539, loss_box_4: 1.7838, loss_cns_4: 0.6535, loss_yns_4: 0.1566, loss_cls_5: 1.0572, loss_box_5: 1.7769, loss_cns_5: 0.6531, loss_yns_5: 0.1566, loss_cls_dn_0: 0.2501, loss_box_dn_0: 0.8085, loss_cls_dn_1: 0.1843, loss_box_dn_1: 0.7984, loss_cls_dn_2: 0.1886, loss_box_dn_2: 0.7613, loss_cls_dn_3: 0.1892, loss_box_dn_3: 0.7663, loss_cls_dn_4: 0.2005, loss_box_dn_4: 0.7617, loss_cls_dn_5: 0.2054, loss_box_dn_5: 0.7619, loss_dense_depth: 0.8140, loss: 28.3753, grad_norm: 37.3257 -2025-11-13 15:36:22,580 - mmdet - INFO - Iter [99/17500] lr: 1.392e-04, eta: 12:52:02, time: 1.477, data_time: 0.073, memory: 49163, loss_cls_0: 0.9344, loss_box_0: 1.7747, loss_cns_0: 0.6183, loss_yns_0: 0.1526, loss_cls_1: 0.9930, loss_box_1: 1.8984, loss_cns_1: 0.6246, loss_yns_1: 0.1532, loss_cls_2: 1.0427, loss_box_2: 1.8555, loss_cns_2: 0.6446, loss_yns_2: 0.1558, loss_cls_3: 1.0505, loss_box_3: 1.8621, loss_cns_3: 0.6481, loss_yns_3: 0.1574, loss_cls_4: 1.0587, loss_box_4: 1.8341, loss_cns_4: 0.6491, loss_yns_4: 0.1572, loss_cls_5: 1.0501, loss_box_5: 1.8406, loss_cns_5: 0.6462, loss_yns_5: 0.1554, loss_cls_dn_0: 0.2489, loss_box_dn_0: 0.8048, loss_cls_dn_1: 0.1810, loss_box_dn_1: 0.7829, loss_cls_dn_2: 0.1848, loss_box_dn_2: 0.7520, loss_cls_dn_3: 0.1862, loss_box_dn_3: 0.7588, loss_cls_dn_4: 0.1969, loss_box_dn_4: 0.7518, loss_cls_dn_5: 0.2006, loss_box_dn_5: 0.7568, loss_dense_depth: 0.8446, loss: 28.6076, grad_norm: 35.2298 -2025-11-13 15:36:24,069 - mmdet - INFO - Iter [100/17500] lr: 1.396e-04, eta: 12:48:35, time: 1.488, data_time: 0.085, memory: 49163, loss_cls_0: 0.9361, loss_box_0: 1.8134, loss_cns_0: 0.6215, loss_yns_0: 0.1557, loss_cls_1: 0.9672, loss_box_1: 1.9397, loss_cns_1: 0.6234, loss_yns_1: 0.1566, loss_cls_2: 1.0318, loss_box_2: 1.8827, loss_cns_2: 0.6422, loss_yns_2: 0.1581, loss_cls_3: 1.0451, loss_box_3: 1.8684, loss_cns_3: 0.6457, loss_yns_3: 0.1594, loss_cls_4: 1.0345, loss_box_4: 1.8662, loss_cns_4: 0.6460, loss_yns_4: 0.1585, loss_cls_5: 1.0384, loss_box_5: 1.8632, loss_cns_5: 0.6455, loss_yns_5: 0.1607, loss_cls_dn_0: 0.2461, loss_box_dn_0: 0.8114, loss_cls_dn_1: 0.1779, loss_box_dn_1: 0.7774, loss_cls_dn_2: 0.1817, loss_box_dn_2: 0.7455, loss_cls_dn_3: 0.1843, loss_box_dn_3: 0.7442, loss_cls_dn_4: 0.1899, loss_box_dn_4: 0.7523, loss_cls_dn_5: 0.1957, loss_box_dn_5: 0.7582, loss_dense_depth: 0.8145, loss: 28.6391, grad_norm: 33.4845 -2025-11-13 15:36:25,627 - mmdet - INFO - Iter [101/17500] lr: 1.400e-04, eta: 12:45:25, time: 1.558, data_time: 0.079, memory: 49163, loss_cls_0: 0.9534, loss_box_0: 1.8070, loss_cns_0: 0.6256, loss_yns_0: 0.1581, loss_cls_1: 0.9941, loss_box_1: 1.9594, loss_cns_1: 0.6248, loss_yns_1: 0.1586, loss_cls_2: 1.0205, loss_box_2: 1.9035, loss_cns_2: 0.6459, loss_yns_2: 0.1609, loss_cls_3: 1.0562, loss_box_3: 1.8972, loss_cns_3: 0.6491, loss_yns_3: 0.1591, loss_cls_4: 1.0568, loss_box_4: 1.8963, loss_cns_4: 0.6498, loss_yns_4: 0.1602, loss_cls_5: 1.0393, loss_box_5: 1.8851, loss_cns_5: 0.6481, loss_yns_5: 0.1628, loss_cls_dn_0: 0.2456, loss_box_dn_0: 0.8043, loss_cls_dn_1: 0.1819, loss_box_dn_1: 0.8133, loss_cls_dn_2: 0.1831, loss_box_dn_2: 0.7805, loss_cls_dn_3: 0.1851, loss_box_dn_3: 0.7857, loss_cls_dn_4: 0.1937, loss_box_dn_4: 0.7967, loss_cls_dn_5: 0.1982, loss_box_dn_5: 0.8046, loss_dense_depth: 0.8343, loss: 29.0786, grad_norm: 55.4085 -2025-11-13 15:36:27,177 - mmdet - INFO - Iter [102/17500] lr: 1.404e-04, eta: 12:42:16, time: 1.551, data_time: 0.158, memory: 49163, loss_cls_0: 0.9254, loss_box_0: 1.8200, loss_cns_0: 0.6191, loss_yns_0: 0.1584, loss_cls_1: 0.9774, loss_box_1: 1.9382, loss_cns_1: 0.6285, loss_yns_1: 0.1595, loss_cls_2: 1.0100, loss_box_2: 1.8760, loss_cns_2: 0.6439, loss_yns_2: 0.1596, loss_cls_3: 1.0229, loss_box_3: 1.8754, loss_cns_3: 0.6469, loss_yns_3: 0.1594, loss_cls_4: 1.0399, loss_box_4: 1.8628, loss_cns_4: 0.6501, loss_yns_4: 0.1599, loss_cls_5: 1.0258, loss_box_5: 1.8614, loss_cns_5: 0.6478, loss_yns_5: 0.1620, loss_cls_dn_0: 0.2456, loss_box_dn_0: 0.8118, loss_cls_dn_1: 0.1837, loss_box_dn_1: 0.8129, loss_cls_dn_2: 0.1826, loss_box_dn_2: 0.7865, loss_cls_dn_3: 0.1835, loss_box_dn_3: 0.7998, loss_cls_dn_4: 0.1977, loss_box_dn_4: 0.8128, loss_cls_dn_5: 0.2020, loss_box_dn_5: 0.8330, loss_dense_depth: 0.8258, loss: 28.9081, grad_norm: 29.8936 -2025-11-13 15:36:28,707 - mmdet - INFO - Iter [103/17500] lr: 1.408e-04, eta: 12:39:08, time: 1.528, data_time: 0.069, memory: 49163, loss_cls_0: 0.9083, loss_box_0: 1.7802, loss_cns_0: 0.6187, loss_yns_0: 0.1588, loss_cls_1: 0.9485, loss_box_1: 1.9338, loss_cns_1: 0.6309, loss_yns_1: 0.1585, loss_cls_2: 1.0071, loss_box_2: 1.8688, loss_cns_2: 0.6461, loss_yns_2: 0.1594, loss_cls_3: 1.0133, loss_box_3: 1.8632, loss_cns_3: 0.6493, loss_yns_3: 0.1603, loss_cls_4: 1.0216, loss_box_4: 1.8713, loss_cns_4: 0.6481, loss_yns_4: 0.1599, loss_cls_5: 1.0103, loss_box_5: 1.8989, loss_cns_5: 0.6500, loss_yns_5: 0.1614, loss_cls_dn_0: 0.2389, loss_box_dn_0: 0.8012, loss_cls_dn_1: 0.1760, loss_box_dn_1: 0.8428, loss_cls_dn_2: 0.1774, loss_box_dn_2: 0.8118, loss_cls_dn_3: 0.1789, loss_box_dn_3: 0.8174, loss_cls_dn_4: 0.1940, loss_box_dn_4: 0.8368, loss_cls_dn_5: 0.1976, loss_box_dn_5: 0.8645, loss_dense_depth: 0.8300, loss: 28.8940, grad_norm: 54.4737 -2025-11-13 15:36:30,267 - mmdet - INFO - Iter [104/17500] lr: 1.412e-04, eta: 12:36:08, time: 1.561, data_time: 0.070, memory: 49163, loss_cls_0: 0.9146, loss_box_0: 1.7744, loss_cns_0: 0.6145, loss_yns_0: 0.1573, loss_cls_1: 0.9645, loss_box_1: 1.9967, loss_cns_1: 0.6295, loss_yns_1: 0.1579, loss_cls_2: 1.0106, loss_box_2: 1.9167, loss_cns_2: 0.6497, loss_yns_2: 0.1569, loss_cls_3: 1.0366, loss_box_3: 1.9178, loss_cns_3: 0.6523, loss_yns_3: 0.1593, loss_cls_4: 1.0296, loss_box_4: 1.9252, loss_cns_4: 0.6517, loss_yns_4: 0.1585, loss_cls_5: 1.0198, loss_box_5: 1.9374, loss_cns_5: 0.6532, loss_yns_5: 0.1596, loss_cls_dn_0: 0.2423, loss_box_dn_0: 0.8105, loss_cls_dn_1: 0.1755, loss_box_dn_1: 0.8744, loss_cls_dn_2: 0.1775, loss_box_dn_2: 0.8257, loss_cls_dn_3: 0.1846, loss_box_dn_3: 0.8198, loss_cls_dn_4: 0.1941, loss_box_dn_4: 0.8342, loss_cls_dn_5: 0.1974, loss_box_dn_5: 0.8517, loss_dense_depth: 0.8251, loss: 29.2572, grad_norm: 55.9340 -2025-11-13 15:36:31,781 - mmdet - INFO - Iter [105/17500] lr: 1.416e-04, eta: 12:33:04, time: 1.513, data_time: 0.083, memory: 49163, loss_cls_0: 0.8987, loss_box_0: 1.7462, loss_cns_0: 0.6154, loss_yns_0: 0.1556, loss_cls_1: 0.9653, loss_box_1: 1.9874, loss_cns_1: 0.6270, loss_yns_1: 0.1562, loss_cls_2: 0.9944, loss_box_2: 1.9254, loss_cns_2: 0.6487, loss_yns_2: 0.1574, loss_cls_3: 1.0248, loss_box_3: 1.9228, loss_cns_3: 0.6516, loss_yns_3: 0.1586, loss_cls_4: 1.0267, loss_box_4: 1.8881, loss_cns_4: 0.6560, loss_yns_4: 0.1578, loss_cls_5: 1.0323, loss_box_5: 1.8908, loss_cns_5: 0.6510, loss_yns_5: 0.1605, loss_cls_dn_0: 0.2385, loss_box_dn_0: 0.8163, loss_cls_dn_1: 0.1772, loss_box_dn_1: 0.8910, loss_cls_dn_2: 0.1785, loss_box_dn_2: 0.8241, loss_cls_dn_3: 0.1850, loss_box_dn_3: 0.8104, loss_cls_dn_4: 0.1917, loss_box_dn_4: 0.8073, loss_cls_dn_5: 0.1919, loss_box_dn_5: 0.8123, loss_dense_depth: 0.7840, loss: 29.0069, grad_norm: 45.7376 -2025-11-13 15:36:33,288 - mmdet - INFO - Iter [106/17500] lr: 1.420e-04, eta: 12:30:03, time: 1.508, data_time: 0.070, memory: 49163, loss_cls_0: 0.9016, loss_box_0: 1.7473, loss_cns_0: 0.6103, loss_yns_0: 0.1566, loss_cls_1: 0.9606, loss_box_1: 2.0161, loss_cns_1: 0.6292, loss_yns_1: 0.1585, loss_cls_2: 0.9961, loss_box_2: 1.9579, loss_cns_2: 0.6439, loss_yns_2: 0.1579, loss_cls_3: 1.0124, loss_box_3: 1.9545, loss_cns_3: 0.6485, loss_yns_3: 0.1578, loss_cls_4: 1.0201, loss_box_4: 1.9092, loss_cns_4: 0.6526, loss_yns_4: 0.1568, loss_cls_5: 1.0207, loss_box_5: 1.9120, loss_cns_5: 0.6510, loss_yns_5: 0.1575, loss_cls_dn_0: 0.2368, loss_box_dn_0: 0.8016, loss_cls_dn_1: 0.1767, loss_box_dn_1: 0.8166, loss_cls_dn_2: 0.1782, loss_box_dn_2: 0.7708, loss_cls_dn_3: 0.1781, loss_box_dn_3: 0.7620, loss_cls_dn_4: 0.1862, loss_box_dn_4: 0.7513, loss_cls_dn_5: 0.1875, loss_box_dn_5: 0.7566, loss_dense_depth: 0.8010, loss: 28.7923, grad_norm: 32.8557 -2025-11-13 15:36:34,758 - mmdet - INFO - Iter [107/17500] lr: 1.424e-04, eta: 12:26:59, time: 1.470, data_time: 0.072, memory: 49163, loss_cls_0: 0.8971, loss_box_0: 1.7389, loss_cns_0: 0.6145, loss_yns_0: 0.1569, loss_cls_1: 0.9421, loss_box_1: 1.9663, loss_cns_1: 0.6306, loss_yns_1: 0.1581, loss_cls_2: 1.0012, loss_box_2: 1.8980, loss_cns_2: 0.6471, loss_yns_2: 0.1553, loss_cls_3: 1.0054, loss_box_3: 1.9218, loss_cns_3: 0.6477, loss_yns_3: 0.1574, loss_cls_4: 1.0168, loss_box_4: 1.9242, loss_cns_4: 0.6498, loss_yns_4: 0.1569, loss_cls_5: 1.0438, loss_box_5: 1.9276, loss_cns_5: 0.6454, loss_yns_5: 0.1557, loss_cls_dn_0: 0.2377, loss_box_dn_0: 0.8145, loss_cls_dn_1: 0.1749, loss_box_dn_1: 0.7868, loss_cls_dn_2: 0.1773, loss_box_dn_2: 0.7599, loss_cls_dn_3: 0.1738, loss_box_dn_3: 0.7747, loss_cls_dn_4: 0.1908, loss_box_dn_4: 0.7830, loss_cls_dn_5: 0.2041, loss_box_dn_5: 0.7941, loss_dense_depth: 0.8008, loss: 28.7311, grad_norm: 65.1560 -2025-11-13 15:36:36,236 - mmdet - INFO - Iter [108/17500] lr: 1.428e-04, eta: 12:23:59, time: 1.477, data_time: 0.070, memory: 49163, loss_cls_0: 0.9012, loss_box_0: 1.7526, loss_cns_0: 0.6178, loss_yns_0: 0.1576, loss_cls_1: 0.9254, loss_box_1: 1.9682, loss_cns_1: 0.6208, loss_yns_1: 0.1584, loss_cls_2: 0.9930, loss_box_2: 1.9195, loss_cns_2: 0.6411, loss_yns_2: 0.1575, loss_cls_3: 1.0026, loss_box_3: 1.9221, loss_cns_3: 0.6465, loss_yns_3: 0.1605, loss_cls_4: 1.0143, loss_box_4: 1.9275, loss_cns_4: 0.6489, loss_yns_4: 0.1604, loss_cls_5: 1.0381, loss_box_5: 1.9232, loss_cns_5: 0.6413, loss_yns_5: 0.1603, loss_cls_dn_0: 0.2397, loss_box_dn_0: 0.8143, loss_cls_dn_1: 0.1783, loss_box_dn_1: 0.8259, loss_cls_dn_2: 0.1843, loss_box_dn_2: 0.8058, loss_cls_dn_3: 0.1861, loss_box_dn_3: 0.8214, loss_cls_dn_4: 0.1940, loss_box_dn_4: 0.8395, loss_cls_dn_5: 0.2060, loss_box_dn_5: 0.8486, loss_dense_depth: 0.7997, loss: 29.0023, grad_norm: 72.5293 -2025-11-13 15:36:37,755 - mmdet - INFO - Iter [109/17500] lr: 1.432e-04, eta: 12:21:10, time: 1.521, data_time: 0.075, memory: 49163, loss_cls_0: 0.9204, loss_box_0: 1.7676, loss_cns_0: 0.6081, loss_yns_0: 0.1557, loss_cls_1: 0.9430, loss_box_1: 1.9688, loss_cns_1: 0.6255, loss_yns_1: 0.1595, loss_cls_2: 0.9772, loss_box_2: 1.9255, loss_cns_2: 0.6388, loss_yns_2: 0.1582, loss_cls_3: 0.9966, loss_box_3: 1.9022, loss_cns_3: 0.6428, loss_yns_3: 0.1604, loss_cls_4: 1.0070, loss_box_4: 1.9225, loss_cns_4: 0.6441, loss_yns_4: 0.1605, loss_cls_5: 1.0567, loss_box_5: 1.8811, loss_cns_5: 0.6406, loss_yns_5: 0.1599, loss_cls_dn_0: 0.2415, loss_box_dn_0: 0.8140, loss_cls_dn_1: 0.1750, loss_box_dn_1: 0.8487, loss_cls_dn_2: 0.1774, loss_box_dn_2: 0.8301, loss_cls_dn_3: 0.1819, loss_box_dn_3: 0.8405, loss_cls_dn_4: 0.1867, loss_box_dn_4: 0.8668, loss_cls_dn_5: 0.1943, loss_box_dn_5: 0.8715, loss_dense_depth: 0.8244, loss: 29.0758, grad_norm: 51.3208 -2025-11-13 15:36:39,253 - mmdet - INFO - Iter [110/17500] lr: 1.436e-04, eta: 12:18:20, time: 1.497, data_time: 0.075, memory: 49163, loss_cls_0: 0.9239, loss_box_0: 1.8086, loss_cns_0: 0.6096, loss_yns_0: 0.1555, loss_cls_1: 0.9553, loss_box_1: 2.0075, loss_cns_1: 0.6214, loss_yns_1: 0.1583, loss_cls_2: 1.0032, loss_box_2: 1.9199, loss_cns_2: 0.6420, loss_yns_2: 0.1566, loss_cls_3: 1.0088, loss_box_3: 1.8940, loss_cns_3: 0.6456, loss_yns_3: 0.1598, loss_cls_4: 1.0250, loss_box_4: 1.9318, loss_cns_4: 0.6429, loss_yns_4: 0.1585, loss_cls_5: 1.0774, loss_box_5: 1.8963, loss_cns_5: 0.6380, loss_yns_5: 0.1589, loss_cls_dn_0: 0.2456, loss_box_dn_0: 0.8077, loss_cls_dn_1: 0.1704, loss_box_dn_1: 0.8244, loss_cls_dn_2: 0.1707, loss_box_dn_2: 0.7878, loss_cls_dn_3: 0.1753, loss_box_dn_3: 0.7965, loss_cls_dn_4: 0.1860, loss_box_dn_4: 0.8274, loss_cls_dn_5: 0.1951, loss_box_dn_5: 0.8397, loss_dense_depth: 0.8112, loss: 29.0365, grad_norm: 53.6247 -2025-11-13 15:36:40,742 - mmdet - INFO - Iter [111/17500] lr: 1.440e-04, eta: 12:15:30, time: 1.484, data_time: 0.074, memory: 49163, loss_cls_0: 0.9097, loss_box_0: 1.8267, loss_cns_0: 0.6122, loss_yns_0: 0.1553, loss_cls_1: 0.9575, loss_box_1: 1.9662, loss_cns_1: 0.6274, loss_yns_1: 0.1568, loss_cls_2: 1.0221, loss_box_2: 1.8877, loss_cns_2: 0.6424, loss_yns_2: 0.1576, loss_cls_3: 1.0116, loss_box_3: 1.8732, loss_cns_3: 0.6466, loss_yns_3: 0.1588, loss_cls_4: 1.0160, loss_box_4: 1.8836, loss_cns_4: 0.6470, loss_yns_4: 0.1587, loss_cls_5: 1.0527, loss_box_5: 1.8865, loss_cns_5: 0.6479, loss_yns_5: 0.1566, loss_cls_dn_0: 0.2453, loss_box_dn_0: 0.8092, loss_cls_dn_1: 0.1731, loss_box_dn_1: 0.8291, loss_cls_dn_2: 0.1827, loss_box_dn_2: 0.7934, loss_cls_dn_3: 0.1827, loss_box_dn_3: 0.8064, loss_cls_dn_4: 0.1919, loss_box_dn_4: 0.8258, loss_cls_dn_5: 0.2050, loss_box_dn_5: 0.8419, loss_dense_depth: 0.8472, loss: 28.9946, grad_norm: 57.6136 -2025-11-13 15:36:42,239 - mmdet - INFO - Iter [112/17500] lr: 1.444e-04, eta: 12:12:47, time: 1.500, data_time: 0.074, memory: 49163, loss_cls_0: 0.9361, loss_box_0: 1.8441, loss_cns_0: 0.6088, loss_yns_0: 0.1567, loss_cls_1: 0.9809, loss_box_1: 1.9691, loss_cns_1: 0.6246, loss_yns_1: 0.1591, loss_cls_2: 1.0175, loss_box_2: 1.9086, loss_cns_2: 0.6386, loss_yns_2: 0.1584, loss_cls_3: 1.0136, loss_box_3: 1.9220, loss_cns_3: 0.6419, loss_yns_3: 0.1591, loss_cls_4: 1.0197, loss_box_4: 1.9015, loss_cns_4: 0.6442, loss_yns_4: 0.1601, loss_cls_5: 1.0347, loss_box_5: 1.8996, loss_cns_5: 0.6434, loss_yns_5: 0.1590, loss_cls_dn_0: 0.2524, loss_box_dn_0: 0.8125, loss_cls_dn_1: 0.1799, loss_box_dn_1: 0.8463, loss_cls_dn_2: 0.1860, loss_box_dn_2: 0.8178, loss_cls_dn_3: 0.1856, loss_box_dn_3: 0.8341, loss_cls_dn_4: 0.1965, loss_box_dn_4: 0.8362, loss_cls_dn_5: 0.2050, loss_box_dn_5: 0.8441, loss_dense_depth: 0.8614, loss: 29.2592, grad_norm: 59.8679 -2025-11-13 15:36:43,723 - mmdet - INFO - Iter [113/17500] lr: 1.448e-04, eta: 12:10:04, time: 1.486, data_time: 0.074, memory: 49163, loss_cls_0: 0.8879, loss_box_0: 1.8012, loss_cns_0: 0.6124, loss_yns_0: 0.1547, loss_cls_1: 0.9441, loss_box_1: 1.9011, loss_cns_1: 0.6322, loss_yns_1: 0.1542, loss_cls_2: 1.0016, loss_box_2: 1.8580, loss_cns_2: 0.6465, loss_yns_2: 0.1552, loss_cls_3: 0.9913, loss_box_3: 1.8815, loss_cns_3: 0.6436, loss_yns_3: 0.1553, loss_cls_4: 0.9973, loss_box_4: 1.8513, loss_cns_4: 0.6431, loss_yns_4: 0.1562, loss_cls_5: 1.0087, loss_box_5: 1.8574, loss_cns_5: 0.6407, loss_yns_5: 0.1560, loss_cls_dn_0: 0.2385, loss_box_dn_0: 0.8117, loss_cls_dn_1: 0.1786, loss_box_dn_1: 0.8200, loss_cls_dn_2: 0.1767, loss_box_dn_2: 0.7922, loss_cls_dn_3: 0.1760, loss_box_dn_3: 0.8032, loss_cls_dn_4: 0.1865, loss_box_dn_4: 0.7918, loss_cls_dn_5: 0.1908, loss_box_dn_5: 0.7965, loss_dense_depth: 0.8077, loss: 28.5020, grad_norm: 54.7609 -2025-11-13 15:36:45,218 - mmdet - INFO - Iter [114/17500] lr: 1.452e-04, eta: 12:07:25, time: 1.493, data_time: 0.073, memory: 49163, loss_cls_0: 0.9035, loss_box_0: 1.8099, loss_cns_0: 0.6159, loss_yns_0: 0.1557, loss_cls_1: 0.9753, loss_box_1: 1.9057, loss_cns_1: 0.6376, loss_yns_1: 0.1581, loss_cls_2: 1.0261, loss_box_2: 1.8325, loss_cns_2: 0.6566, loss_yns_2: 0.1567, loss_cls_3: 1.0124, loss_box_3: 1.8294, loss_cns_3: 0.6538, loss_yns_3: 0.1573, loss_cls_4: 1.0070, loss_box_4: 1.8333, loss_cns_4: 0.6558, loss_yns_4: 0.1579, loss_cls_5: 1.0078, loss_box_5: 1.8288, loss_cns_5: 0.6472, loss_yns_5: 0.1578, loss_cls_dn_0: 0.2445, loss_box_dn_0: 0.7980, loss_cls_dn_1: 0.1706, loss_box_dn_1: 0.7843, loss_cls_dn_2: 0.1675, loss_box_dn_2: 0.7516, loss_cls_dn_3: 0.1697, loss_box_dn_3: 0.7525, loss_cls_dn_4: 0.1799, loss_box_dn_4: 0.7490, loss_cls_dn_5: 0.1852, loss_box_dn_5: 0.7524, loss_dense_depth: 0.8299, loss: 28.3172, grad_norm: 37.7503 -2025-11-13 15:36:46,705 - mmdet - INFO - Iter [115/17500] lr: 1.456e-04, eta: 12:04:48, time: 1.487, data_time: 0.081, memory: 49163, loss_cls_0: 0.8829, loss_box_0: 1.7912, loss_cns_0: 0.6207, loss_yns_0: 0.1548, loss_cls_1: 0.9589, loss_box_1: 1.9641, loss_cns_1: 0.6364, loss_yns_1: 0.1565, loss_cls_2: 0.9826, loss_box_2: 1.8887, loss_cns_2: 0.6507, loss_yns_2: 0.1557, loss_cls_3: 0.9847, loss_box_3: 1.8692, loss_cns_3: 0.6481, loss_yns_3: 0.1548, loss_cls_4: 0.9926, loss_box_4: 1.9108, loss_cns_4: 0.6522, loss_yns_4: 0.1576, loss_cls_5: 0.9957, loss_box_5: 1.8810, loss_cns_5: 0.6472, loss_yns_5: 0.1553, loss_cls_dn_0: 0.2333, loss_box_dn_0: 0.8018, loss_cls_dn_1: 0.1701, loss_box_dn_1: 0.7817, loss_cls_dn_2: 0.1689, loss_box_dn_2: 0.7497, loss_cls_dn_3: 0.1686, loss_box_dn_3: 0.7430, loss_cls_dn_4: 0.1778, loss_box_dn_4: 0.7647, loss_cls_dn_5: 0.1875, loss_box_dn_5: 0.7606, loss_dense_depth: 0.7782, loss: 28.3783, grad_norm: 60.6305 -2025-11-13 15:36:48,202 - mmdet - INFO - Iter [116/17500] lr: 1.460e-04, eta: 12:02:15, time: 1.498, data_time: 0.079, memory: 49163, loss_cls_0: 0.9351, loss_box_0: 1.7964, loss_cns_0: 0.6197, loss_yns_0: 0.1556, loss_cls_1: 0.9572, loss_box_1: 1.9494, loss_cns_1: 0.6348, loss_yns_1: 0.1566, loss_cls_2: 0.9845, loss_box_2: 1.8899, loss_cns_2: 0.6474, loss_yns_2: 0.1558, loss_cls_3: 0.9989, loss_box_3: 1.8686, loss_cns_3: 0.6444, loss_yns_3: 0.1560, loss_cls_4: 0.9999, loss_box_4: 1.9035, loss_cns_4: 0.6462, loss_yns_4: 0.1581, loss_cls_5: 1.0039, loss_box_5: 1.8812, loss_cns_5: 0.6502, loss_yns_5: 0.1564, loss_cls_dn_0: 0.2361, loss_box_dn_0: 0.7973, loss_cls_dn_1: 0.1709, loss_box_dn_1: 0.7980, loss_cls_dn_2: 0.1691, loss_box_dn_2: 0.7820, loss_cls_dn_3: 0.1695, loss_box_dn_3: 0.7851, loss_cls_dn_4: 0.1792, loss_box_dn_4: 0.8131, loss_cls_dn_5: 0.1956, loss_box_dn_5: 0.8149, loss_dense_depth: 0.8322, loss: 28.6928, grad_norm: 57.6318 -2025-11-13 15:36:49,700 - mmdet - INFO - Iter [117/17500] lr: 1.464e-04, eta: 11:59:44, time: 1.498, data_time: 0.076, memory: 49163, loss_cls_0: 0.8989, loss_box_0: 1.7579, loss_cns_0: 0.6238, loss_yns_0: 0.1537, loss_cls_1: 0.9457, loss_box_1: 1.9227, loss_cns_1: 0.6361, loss_yns_1: 0.1560, loss_cls_2: 0.9889, loss_box_2: 1.8744, loss_cns_2: 0.6527, loss_yns_2: 0.1575, loss_cls_3: 0.9892, loss_box_3: 1.8614, loss_cns_3: 0.6484, loss_yns_3: 0.1566, loss_cls_4: 1.0008, loss_box_4: 1.8677, loss_cns_4: 0.6468, loss_yns_4: 0.1549, loss_cls_5: 0.9980, loss_box_5: 1.8889, loss_cns_5: 0.6538, loss_yns_5: 0.1555, loss_cls_dn_0: 0.2337, loss_box_dn_0: 0.7974, loss_cls_dn_1: 0.1700, loss_box_dn_1: 0.8213, loss_cls_dn_2: 0.1688, loss_box_dn_2: 0.8042, loss_cls_dn_3: 0.1700, loss_box_dn_3: 0.8115, loss_cls_dn_4: 0.1796, loss_box_dn_4: 0.8301, loss_cls_dn_5: 0.1939, loss_box_dn_5: 0.8521, loss_dense_depth: 0.8044, loss: 28.6272, grad_norm: 44.4085 -2025-11-13 15:36:51,206 - mmdet - INFO - Iter [118/17500] lr: 1.468e-04, eta: 11:57:18, time: 1.506, data_time: 0.075, memory: 49163, loss_cls_0: 0.9021, loss_box_0: 1.7444, loss_cns_0: 0.6194, loss_yns_0: 0.1561, loss_cls_1: 0.9558, loss_box_1: 1.8706, loss_cns_1: 0.6387, loss_yns_1: 0.1563, loss_cls_2: 0.9861, loss_box_2: 1.8153, loss_cns_2: 0.6563, loss_yns_2: 0.1550, loss_cls_3: 0.9988, loss_box_3: 1.8111, loss_cns_3: 0.6533, loss_yns_3: 0.1557, loss_cls_4: 1.0257, loss_box_4: 1.8024, loss_cns_4: 0.6519, loss_yns_4: 0.1552, loss_cls_5: 1.0075, loss_box_5: 1.8319, loss_cns_5: 0.6504, loss_yns_5: 0.1555, loss_cls_dn_0: 0.2375, loss_box_dn_0: 0.7968, loss_cls_dn_1: 0.1705, loss_box_dn_1: 0.8391, loss_cls_dn_2: 0.1675, loss_box_dn_2: 0.8164, loss_cls_dn_3: 0.1699, loss_box_dn_3: 0.8232, loss_cls_dn_4: 0.1830, loss_box_dn_4: 0.8322, loss_cls_dn_5: 0.1925, loss_box_dn_5: 0.8556, loss_dense_depth: 0.8134, loss: 28.4531, grad_norm: 50.7244 -2025-11-13 15:36:52,697 - mmdet - INFO - Iter [119/17500] lr: 1.472e-04, eta: 11:54:51, time: 1.489, data_time: 0.070, memory: 49163, loss_cls_0: 0.9212, loss_box_0: 1.7296, loss_cns_0: 0.6150, loss_yns_0: 0.1542, loss_cls_1: 0.9534, loss_box_1: 1.9030, loss_cns_1: 0.6402, loss_yns_1: 0.1551, loss_cls_2: 0.9803, loss_box_2: 1.8479, loss_cns_2: 0.6520, loss_yns_2: 0.1542, loss_cls_3: 0.9901, loss_box_3: 1.8244, loss_cns_3: 0.6511, loss_yns_3: 0.1546, loss_cls_4: 1.0134, loss_box_4: 1.8073, loss_cns_4: 0.6507, loss_yns_4: 0.1537, loss_cls_5: 0.9966, loss_box_5: 1.8212, loss_cns_5: 0.6482, loss_yns_5: 0.1537, loss_cls_dn_0: 0.2364, loss_box_dn_0: 0.7867, loss_cls_dn_1: 0.1694, loss_box_dn_1: 0.8523, loss_cls_dn_2: 0.1683, loss_box_dn_2: 0.8206, loss_cls_dn_3: 0.1692, loss_box_dn_3: 0.8113, loss_cls_dn_4: 0.1815, loss_box_dn_4: 0.8119, loss_cls_dn_5: 0.1821, loss_box_dn_5: 0.8224, loss_dense_depth: 0.8002, loss: 28.3831, grad_norm: 47.9274 -2025-11-13 15:36:54,189 - mmdet - INFO - Iter [120/17500] lr: 1.476e-04, eta: 11:52:28, time: 1.493, data_time: 0.083, memory: 49163, loss_cls_0: 0.8856, loss_box_0: 1.7601, loss_cns_0: 0.6141, loss_yns_0: 0.1544, loss_cls_1: 0.9538, loss_box_1: 1.9211, loss_cns_1: 0.6379, loss_yns_1: 0.1561, loss_cls_2: 0.9758, loss_box_2: 1.8713, loss_cns_2: 0.6469, loss_yns_2: 0.1558, loss_cls_3: 0.9823, loss_box_3: 1.8315, loss_cns_3: 0.6490, loss_yns_3: 0.1550, loss_cls_4: 0.9942, loss_box_4: 1.8185, loss_cns_4: 0.6495, loss_yns_4: 0.1559, loss_cls_5: 0.9918, loss_box_5: 1.8037, loss_cns_5: 0.6482, loss_yns_5: 0.1561, loss_cls_dn_0: 0.2330, loss_box_dn_0: 0.7977, loss_cls_dn_1: 0.1677, loss_box_dn_1: 0.7980, loss_cls_dn_2: 0.1670, loss_box_dn_2: 0.7645, loss_cls_dn_3: 0.1653, loss_box_dn_3: 0.7491, loss_cls_dn_4: 0.1775, loss_box_dn_4: 0.7430, loss_cls_dn_5: 0.1799, loss_box_dn_5: 0.7393, loss_dense_depth: 0.7874, loss: 28.0379, grad_norm: 45.2170 -2025-11-13 15:36:55,757 - mmdet - INFO - Iter [121/17500] lr: 1.480e-04, eta: 11:50:17, time: 1.568, data_time: 0.077, memory: 49163, loss_cls_0: 0.8984, loss_box_0: 1.7718, loss_cns_0: 0.6222, loss_yns_0: 0.1559, loss_cls_1: 0.9552, loss_box_1: 1.9613, loss_cns_1: 0.6336, loss_yns_1: 0.1550, loss_cls_2: 0.9868, loss_box_2: 1.8807, loss_cns_2: 0.6449, loss_yns_2: 0.1575, loss_cls_3: 0.9972, loss_box_3: 1.8798, loss_cns_3: 0.6469, loss_yns_3: 0.1548, loss_cls_4: 1.0182, loss_box_4: 1.8640, loss_cns_4: 0.6480, loss_yns_4: 0.1541, loss_cls_5: 1.0091, loss_box_5: 1.8629, loss_cns_5: 0.6488, loss_yns_5: 0.1552, loss_cls_dn_0: 0.2325, loss_box_dn_0: 0.7999, loss_cls_dn_1: 0.1630, loss_box_dn_1: 0.7846, loss_cls_dn_2: 0.1623, loss_box_dn_2: 0.7494, loss_cls_dn_3: 0.1610, loss_box_dn_3: 0.7533, loss_cls_dn_4: 0.1709, loss_box_dn_4: 0.7499, loss_cls_dn_5: 0.1780, loss_box_dn_5: 0.7549, loss_dense_depth: 0.7791, loss: 28.3009, grad_norm: 45.2341 -2025-11-13 15:36:57,319 - mmdet - INFO - Iter [122/17500] lr: 1.484e-04, eta: 11:48:08, time: 1.562, data_time: 0.161, memory: 49163, loss_cls_0: 0.9041, loss_box_0: 1.7613, loss_cns_0: 0.6239, loss_yns_0: 0.1568, loss_cls_1: 0.9414, loss_box_1: 1.9176, loss_cns_1: 0.6326, loss_yns_1: 0.1558, loss_cls_2: 0.9920, loss_box_2: 1.8544, loss_cns_2: 0.6509, loss_yns_2: 0.1592, loss_cls_3: 0.9956, loss_box_3: 1.8606, loss_cns_3: 0.6498, loss_yns_3: 0.1567, loss_cls_4: 1.0160, loss_box_4: 1.8343, loss_cns_4: 0.6485, loss_yns_4: 0.1588, loss_cls_5: 1.0115, loss_box_5: 1.8405, loss_cns_5: 0.6461, loss_yns_5: 0.1579, loss_cls_dn_0: 0.2331, loss_box_dn_0: 0.7919, loss_cls_dn_1: 0.1629, loss_box_dn_1: 0.7722, loss_cls_dn_2: 0.1647, loss_box_dn_2: 0.7446, loss_cls_dn_3: 0.1626, loss_box_dn_3: 0.7587, loss_cls_dn_4: 0.1717, loss_box_dn_4: 0.7604, loss_cls_dn_5: 0.1806, loss_box_dn_5: 0.7685, loss_dense_depth: 0.7766, loss: 28.1749, grad_norm: 35.6969 -2025-11-13 15:36:58,844 - mmdet - INFO - Iter [123/17500] lr: 1.488e-04, eta: 11:45:55, time: 1.525, data_time: 0.079, memory: 49163, loss_cls_0: 0.8823, loss_box_0: 1.7588, loss_cns_0: 0.6202, loss_yns_0: 0.1558, loss_cls_1: 0.9548, loss_box_1: 1.8861, loss_cns_1: 0.6315, loss_yns_1: 0.1553, loss_cls_2: 1.0024, loss_box_2: 1.8425, loss_cns_2: 0.6521, loss_yns_2: 0.1577, loss_cls_3: 0.9913, loss_box_3: 1.8308, loss_cns_3: 0.6487, loss_yns_3: 0.1559, loss_cls_4: 1.0008, loss_box_4: 1.8205, loss_cns_4: 0.6481, loss_yns_4: 0.1573, loss_cls_5: 0.9947, loss_box_5: 1.7990, loss_cns_5: 0.6457, loss_yns_5: 0.1569, loss_cls_dn_0: 0.2284, loss_box_dn_0: 0.7928, loss_cls_dn_1: 0.1704, loss_box_dn_1: 0.8033, loss_cls_dn_2: 0.1712, loss_box_dn_2: 0.7827, loss_cls_dn_3: 0.1682, loss_box_dn_3: 0.7990, loss_cls_dn_4: 0.1778, loss_box_dn_4: 0.8098, loss_cls_dn_5: 0.1847, loss_box_dn_5: 0.8100, loss_dense_depth: 0.7915, loss: 28.2393, grad_norm: 47.9780 -2025-11-13 15:37:00,414 - mmdet - INFO - Iter [124/17500] lr: 1.492e-04, eta: 11:43:51, time: 1.570, data_time: 0.079, memory: 49163, loss_cls_0: 0.8826, loss_box_0: 1.7486, loss_cns_0: 0.6220, loss_yns_0: 0.1572, loss_cls_1: 0.9313, loss_box_1: 1.9169, loss_cns_1: 0.6327, loss_yns_1: 0.1536, loss_cls_2: 0.9811, loss_box_2: 1.8799, loss_cns_2: 0.6490, loss_yns_2: 0.1553, loss_cls_3: 0.9858, loss_box_3: 1.8577, loss_cns_3: 0.6489, loss_yns_3: 0.1567, loss_cls_4: 0.9974, loss_box_4: 1.8748, loss_cns_4: 0.6513, loss_yns_4: 0.1567, loss_cls_5: 0.9929, loss_box_5: 1.8405, loss_cns_5: 0.6533, loss_yns_5: 0.1562, loss_cls_dn_0: 0.2289, loss_box_dn_0: 0.7907, loss_cls_dn_1: 0.1645, loss_box_dn_1: 0.8018, loss_cls_dn_2: 0.1595, loss_box_dn_2: 0.7781, loss_cls_dn_3: 0.1614, loss_box_dn_3: 0.7819, loss_cls_dn_4: 0.1719, loss_box_dn_4: 0.7946, loss_cls_dn_5: 0.1784, loss_box_dn_5: 0.7891, loss_dense_depth: 0.7597, loss: 28.2428, grad_norm: 42.3821 -2025-11-13 15:37:01,915 - mmdet - INFO - Iter [125/17500] lr: 1.496e-04, eta: 11:41:40, time: 1.503, data_time: 0.087, memory: 49163, loss_cls_0: 0.9007, loss_box_0: 1.7574, loss_cns_0: 0.6168, loss_yns_0: 0.1553, loss_cls_1: 0.9351, loss_box_1: 1.8705, loss_cns_1: 0.6268, loss_yns_1: 0.1547, loss_cls_2: 0.9971, loss_box_2: 1.8169, loss_cns_2: 0.6475, loss_yns_2: 0.1555, loss_cls_3: 1.0009, loss_box_3: 1.8020, loss_cns_3: 0.6492, loss_yns_3: 0.1562, loss_cls_4: 1.0094, loss_box_4: 1.8070, loss_cns_4: 0.6526, loss_yns_4: 0.1586, loss_cls_5: 1.0060, loss_box_5: 1.7910, loss_cns_5: 0.6509, loss_yns_5: 0.1575, loss_cls_dn_0: 0.2344, loss_box_dn_0: 0.7830, loss_cls_dn_1: 0.1651, loss_box_dn_1: 0.8107, loss_cls_dn_2: 0.1603, loss_box_dn_2: 0.7770, loss_cls_dn_3: 0.1621, loss_box_dn_3: 0.7710, loss_cls_dn_4: 0.1707, loss_box_dn_4: 0.7805, loss_cls_dn_5: 0.1802, loss_box_dn_5: 0.7820, loss_dense_depth: 0.7644, loss: 28.0171, grad_norm: 34.5191 -2025-11-13 15:37:03,412 - mmdet - INFO - Iter [126/17500] lr: 1.500e-04, eta: 11:39:30, time: 1.495, data_time: 0.072, memory: 49163, loss_cls_0: 0.8940, loss_box_0: 1.7689, loss_cns_0: 0.6132, loss_yns_0: 0.1550, loss_cls_1: 0.9262, loss_box_1: 1.9075, loss_cns_1: 0.6233, loss_yns_1: 0.1538, loss_cls_2: 0.9868, loss_box_2: 1.8372, loss_cns_2: 0.6481, loss_yns_2: 0.1563, loss_cls_3: 1.0036, loss_box_3: 1.8105, loss_cns_3: 0.6504, loss_yns_3: 0.1554, loss_cls_4: 1.0149, loss_box_4: 1.8173, loss_cns_4: 0.6479, loss_yns_4: 0.1551, loss_cls_5: 1.0188, loss_box_5: 1.8363, loss_cns_5: 0.6471, loss_yns_5: 0.1553, loss_cls_dn_0: 0.2333, loss_box_dn_0: 0.7906, loss_cls_dn_1: 0.1686, loss_box_dn_1: 0.7791, loss_cls_dn_2: 0.1643, loss_box_dn_2: 0.7433, loss_cls_dn_3: 0.1657, loss_box_dn_3: 0.7361, loss_cls_dn_4: 0.1725, loss_box_dn_4: 0.7419, loss_cls_dn_5: 0.1815, loss_box_dn_5: 0.7532, loss_dense_depth: 0.7514, loss: 27.9645, grad_norm: 40.5527 -2025-11-13 15:37:04,901 - mmdet - INFO - Iter [127/17500] lr: 1.504e-04, eta: 11:37:20, time: 1.489, data_time: 0.077, memory: 49163, loss_cls_0: 0.8899, loss_box_0: 1.7509, loss_cns_0: 0.6161, loss_yns_0: 0.1554, loss_cls_1: 0.9384, loss_box_1: 1.8503, loss_cns_1: 0.6304, loss_yns_1: 0.1569, loss_cls_2: 0.9847, loss_box_2: 1.7899, loss_cns_2: 0.6508, loss_yns_2: 0.1598, loss_cls_3: 1.0056, loss_box_3: 1.7546, loss_cns_3: 0.6518, loss_yns_3: 0.1574, loss_cls_4: 1.0095, loss_box_4: 1.7411, loss_cns_4: 0.6526, loss_yns_4: 0.1557, loss_cls_5: 1.0011, loss_box_5: 1.7400, loss_cns_5: 0.6514, loss_yns_5: 0.1552, loss_cls_dn_0: 0.2321, loss_box_dn_0: 0.7903, loss_cls_dn_1: 0.1633, loss_box_dn_1: 0.7553, loss_cls_dn_2: 0.1634, loss_box_dn_2: 0.7262, loss_cls_dn_3: 0.1623, loss_box_dn_3: 0.7138, loss_cls_dn_4: 0.1692, loss_box_dn_4: 0.7129, loss_cls_dn_5: 0.1782, loss_box_dn_5: 0.7192, loss_dense_depth: 0.7575, loss: 27.4935, grad_norm: 33.8374 -2025-11-13 15:37:06,394 - mmdet - INFO - Iter [128/17500] lr: 1.508e-04, eta: 11:35:14, time: 1.493, data_time: 0.074, memory: 49163, loss_cls_0: 0.9003, loss_box_0: 1.7342, loss_cns_0: 0.6176, loss_yns_0: 0.1559, loss_cls_1: 0.9597, loss_box_1: 1.8158, loss_cns_1: 0.6289, loss_yns_1: 0.1580, loss_cls_2: 0.9959, loss_box_2: 1.7805, loss_cns_2: 0.6463, loss_yns_2: 0.1592, loss_cls_3: 1.0071, loss_box_3: 1.7878, loss_cns_3: 0.6457, loss_yns_3: 0.1563, loss_cls_4: 1.0168, loss_box_4: 1.7757, loss_cns_4: 0.6448, loss_yns_4: 0.1584, loss_cls_5: 0.9994, loss_box_5: 1.7810, loss_cns_5: 0.6417, loss_yns_5: 0.1564, loss_cls_dn_0: 0.2319, loss_box_dn_0: 0.7960, loss_cls_dn_1: 0.1584, loss_box_dn_1: 0.7295, loss_cls_dn_2: 0.1563, loss_box_dn_2: 0.7116, loss_cls_dn_3: 0.1581, loss_box_dn_3: 0.7123, loss_cls_dn_4: 0.1687, loss_box_dn_4: 0.7161, loss_cls_dn_5: 0.1769, loss_box_dn_5: 0.7291, loss_dense_depth: 0.7759, loss: 27.5443, grad_norm: 46.3229 -2025-11-13 15:37:07,901 - mmdet - INFO - Iter [129/17500] lr: 1.512e-04, eta: 11:33:11, time: 1.508, data_time: 0.079, memory: 49163, loss_cls_0: 0.8776, loss_box_0: 1.7204, loss_cns_0: 0.6174, loss_yns_0: 0.1558, loss_cls_1: 0.9449, loss_box_1: 1.8271, loss_cns_1: 0.6344, loss_yns_1: 0.1571, loss_cls_2: 0.9869, loss_box_2: 1.7748, loss_cns_2: 0.6488, loss_yns_2: 0.1585, loss_cls_3: 0.9900, loss_box_3: 1.7649, loss_cns_3: 0.6493, loss_yns_3: 0.1571, loss_cls_4: 0.9991, loss_box_4: 1.7627, loss_cns_4: 0.6522, loss_yns_4: 0.1593, loss_cls_5: 0.9846, loss_box_5: 1.7649, loss_cns_5: 0.6469, loss_yns_5: 0.1555, loss_cls_dn_0: 0.2283, loss_box_dn_0: 0.7867, loss_cls_dn_1: 0.1594, loss_box_dn_1: 0.7356, loss_cls_dn_2: 0.1575, loss_box_dn_2: 0.7168, loss_cls_dn_3: 0.1580, loss_box_dn_3: 0.7152, loss_cls_dn_4: 0.1682, loss_box_dn_4: 0.7235, loss_cls_dn_5: 0.1729, loss_box_dn_5: 0.7391, loss_dense_depth: 0.7468, loss: 27.3984, grad_norm: 34.3571 -2025-11-13 15:37:09,401 - mmdet - INFO - Iter [130/17500] lr: 1.516e-04, eta: 11:31:09, time: 1.500, data_time: 0.076, memory: 49163, loss_cls_0: 0.8978, loss_box_0: 1.7274, loss_cns_0: 0.6114, loss_yns_0: 0.1517, loss_cls_1: 0.9616, loss_box_1: 1.8424, loss_cns_1: 0.6323, loss_yns_1: 0.1548, loss_cls_2: 1.0050, loss_box_2: 1.8016, loss_cns_2: 0.6457, loss_yns_2: 0.1538, loss_cls_3: 1.0076, loss_box_3: 1.7836, loss_cns_3: 0.6465, loss_yns_3: 0.1574, loss_cls_4: 1.0133, loss_box_4: 1.7991, loss_cns_4: 0.6479, loss_yns_4: 0.1582, loss_cls_5: 1.0029, loss_box_5: 1.7872, loss_cns_5: 0.6440, loss_yns_5: 0.1537, loss_cls_dn_0: 0.2312, loss_box_dn_0: 0.7862, loss_cls_dn_1: 0.1582, loss_box_dn_1: 0.7607, loss_cls_dn_2: 0.1579, loss_box_dn_2: 0.7494, loss_cls_dn_3: 0.1592, loss_box_dn_3: 0.7446, loss_cls_dn_4: 0.1663, loss_box_dn_4: 0.7562, loss_cls_dn_5: 0.1736, loss_box_dn_5: 0.7647, loss_dense_depth: 0.7835, loss: 27.7787, grad_norm: 51.0696 -2025-11-13 15:37:10,886 - mmdet - INFO - Iter [131/17500] lr: 1.520e-04, eta: 11:29:07, time: 1.484, data_time: 0.076, memory: 49163, loss_cls_0: 0.9135, loss_box_0: 1.7254, loss_cns_0: 0.6173, loss_yns_0: 0.1519, loss_cls_1: 0.9565, loss_box_1: 1.8530, loss_cns_1: 0.6386, loss_yns_1: 0.1562, loss_cls_2: 0.9860, loss_box_2: 1.8153, loss_cns_2: 0.6472, loss_yns_2: 0.1568, loss_cls_3: 1.0129, loss_box_3: 1.8072, loss_cns_3: 0.6484, loss_yns_3: 0.1562, loss_cls_4: 1.0182, loss_box_4: 1.8129, loss_cns_4: 0.6505, loss_yns_4: 0.1563, loss_cls_5: 0.9963, loss_box_5: 1.7911, loss_cns_5: 0.6469, loss_yns_5: 0.1539, loss_cls_dn_0: 0.2314, loss_box_dn_0: 0.7913, loss_cls_dn_1: 0.1588, loss_box_dn_1: 0.7651, loss_cls_dn_2: 0.1599, loss_box_dn_2: 0.7467, loss_cls_dn_3: 0.1578, loss_box_dn_3: 0.7395, loss_cls_dn_4: 0.1653, loss_box_dn_4: 0.7478, loss_cls_dn_5: 0.1749, loss_box_dn_5: 0.7510, loss_dense_depth: 0.7982, loss: 27.8561, grad_norm: 46.0806 -2025-11-13 15:37:12,364 - mmdet - INFO - Iter [132/17500] lr: 1.524e-04, eta: 11:27:06, time: 1.478, data_time: 0.077, memory: 49163, loss_cls_0: 0.8972, loss_box_0: 1.7560, loss_cns_0: 0.6185, loss_yns_0: 0.1494, loss_cls_1: 0.9530, loss_box_1: 1.8321, loss_cns_1: 0.6385, loss_yns_1: 0.1535, loss_cls_2: 0.9906, loss_box_2: 1.7911, loss_cns_2: 0.6456, loss_yns_2: 0.1538, loss_cls_3: 0.9977, loss_box_3: 1.8053, loss_cns_3: 0.6476, loss_yns_3: 0.1524, loss_cls_4: 0.9979, loss_box_4: 1.8110, loss_cns_4: 0.6603, loss_yns_4: 0.1532, loss_cls_5: 0.9842, loss_box_5: 1.8087, loss_cns_5: 0.6485, loss_yns_5: 0.1528, loss_cls_dn_0: 0.2334, loss_box_dn_0: 0.7874, loss_cls_dn_1: 0.1592, loss_box_dn_1: 0.7451, loss_cls_dn_2: 0.1585, loss_box_dn_2: 0.7264, loss_cls_dn_3: 0.1580, loss_box_dn_3: 0.7261, loss_cls_dn_4: 0.1673, loss_box_dn_4: 0.7326, loss_cls_dn_5: 0.1747, loss_box_dn_5: 0.7442, loss_dense_depth: 0.7848, loss: 27.6964, grad_norm: 46.5559 -2025-11-13 15:37:13,845 - mmdet - INFO - Iter [133/17500] lr: 1.528e-04, eta: 11:25:07, time: 1.483, data_time: 0.076, memory: 49163, loss_cls_0: 0.8986, loss_box_0: 1.7330, loss_cns_0: 0.6103, loss_yns_0: 0.1490, loss_cls_1: 0.9449, loss_box_1: 1.8257, loss_cns_1: 0.6363, loss_yns_1: 0.1494, loss_cls_2: 0.9848, loss_box_2: 1.7704, loss_cns_2: 0.6454, loss_yns_2: 0.1526, loss_cls_3: 0.9937, loss_box_3: 1.7777, loss_cns_3: 0.6467, loss_yns_3: 0.1514, loss_cls_4: 1.0082, loss_box_4: 1.7655, loss_cns_4: 0.6509, loss_yns_4: 0.1537, loss_cls_5: 0.9985, loss_box_5: 1.7771, loss_cns_5: 0.6477, loss_yns_5: 0.1521, loss_cls_dn_0: 0.2340, loss_box_dn_0: 0.7915, loss_cls_dn_1: 0.1605, loss_box_dn_1: 0.7387, loss_cls_dn_2: 0.1567, loss_box_dn_2: 0.7195, loss_cls_dn_3: 0.1622, loss_box_dn_3: 0.7180, loss_cls_dn_4: 0.1692, loss_box_dn_4: 0.7228, loss_cls_dn_5: 0.1755, loss_box_dn_5: 0.7271, loss_dense_depth: 0.7893, loss: 27.4887, grad_norm: 42.3410 -2025-11-13 15:37:15,351 - mmdet - INFO - Iter [134/17500] lr: 1.532e-04, eta: 11:23:13, time: 1.504, data_time: 0.076, memory: 49163, loss_cls_0: 0.8777, loss_box_0: 1.6900, loss_cns_0: 0.6147, loss_yns_0: 0.1489, loss_cls_1: 0.9493, loss_box_1: 1.8472, loss_cns_1: 0.6414, loss_yns_1: 0.1507, loss_cls_2: 0.9784, loss_box_2: 1.8068, loss_cns_2: 0.6492, loss_yns_2: 0.1504, loss_cls_3: 0.9906, loss_box_3: 1.8108, loss_cns_3: 0.6480, loss_yns_3: 0.1534, loss_cls_4: 0.9980, loss_box_4: 1.8000, loss_cns_4: 0.6488, loss_yns_4: 0.1532, loss_cls_5: 0.9925, loss_box_5: 1.8191, loss_cns_5: 0.6455, loss_yns_5: 0.1518, loss_cls_dn_0: 0.2278, loss_box_dn_0: 0.7881, loss_cls_dn_1: 0.1584, loss_box_dn_1: 0.7145, loss_cls_dn_2: 0.1565, loss_box_dn_2: 0.6987, loss_cls_dn_3: 0.1598, loss_box_dn_3: 0.6980, loss_cls_dn_4: 0.1668, loss_box_dn_4: 0.7041, loss_cls_dn_5: 0.1711, loss_box_dn_5: 0.7034, loss_dense_depth: 0.7671, loss: 27.4306, grad_norm: 46.0035 -2025-11-13 15:37:16,838 - mmdet - INFO - Iter [135/17500] lr: 1.536e-04, eta: 11:21:18, time: 1.489, data_time: 0.083, memory: 49163, loss_cls_0: 0.8877, loss_box_0: 1.7352, loss_cns_0: 0.6199, loss_yns_0: 0.1546, loss_cls_1: 0.9600, loss_box_1: 1.8778, loss_cns_1: 0.6434, loss_yns_1: 0.1559, loss_cls_2: 0.9812, loss_box_2: 1.8522, loss_cns_2: 0.6485, loss_yns_2: 0.1523, loss_cls_3: 0.9904, loss_box_3: 1.8483, loss_cns_3: 0.6466, loss_yns_3: 0.1531, loss_cls_4: 1.0038, loss_box_4: 1.8504, loss_cns_4: 0.6501, loss_yns_4: 0.1542, loss_cls_5: 0.9873, loss_box_5: 1.8597, loss_cns_5: 0.6439, loss_yns_5: 0.1548, loss_cls_dn_0: 0.2356, loss_box_dn_0: 0.7877, loss_cls_dn_1: 0.1612, loss_box_dn_1: 0.7259, loss_cls_dn_2: 0.1598, loss_box_dn_2: 0.7076, loss_cls_dn_3: 0.1576, loss_box_dn_3: 0.7089, loss_cls_dn_4: 0.1671, loss_box_dn_4: 0.7148, loss_cls_dn_5: 0.1744, loss_box_dn_5: 0.7171, loss_dense_depth: 0.7779, loss: 27.8069, grad_norm: 34.0763 -2025-11-13 15:37:18,339 - mmdet - INFO - Iter [136/17500] lr: 1.540e-04, eta: 11:19:27, time: 1.501, data_time: 0.077, memory: 49163, loss_cls_0: 0.8704, loss_box_0: 1.7269, loss_cns_0: 0.6248, loss_yns_0: 0.1494, loss_cls_1: 0.9333, loss_box_1: 1.8594, loss_cns_1: 0.6439, loss_yns_1: 0.1503, loss_cls_2: 0.9519, loss_box_2: 1.8372, loss_cns_2: 0.6479, loss_yns_2: 0.1500, loss_cls_3: 0.9623, loss_box_3: 1.8385, loss_cns_3: 0.6479, loss_yns_3: 0.1492, loss_cls_4: 0.9864, loss_box_4: 1.8241, loss_cns_4: 0.6484, loss_yns_4: 0.1511, loss_cls_5: 0.9752, loss_box_5: 1.8345, loss_cns_5: 0.6468, loss_yns_5: 0.1522, loss_cls_dn_0: 0.2299, loss_box_dn_0: 0.7815, loss_cls_dn_1: 0.1517, loss_box_dn_1: 0.7222, loss_cls_dn_2: 0.1506, loss_box_dn_2: 0.7130, loss_cls_dn_3: 0.1519, loss_box_dn_3: 0.7193, loss_cls_dn_4: 0.1603, loss_box_dn_4: 0.7209, loss_cls_dn_5: 0.1663, loss_box_dn_5: 0.7343, loss_dense_depth: 0.7648, loss: 27.5288, grad_norm: 60.7846 -2025-11-13 15:37:19,843 - mmdet - INFO - Iter [137/17500] lr: 1.544e-04, eta: 11:17:38, time: 1.502, data_time: 0.076, memory: 49163, loss_cls_0: 0.8748, loss_box_0: 1.7450, loss_cns_0: 0.6211, loss_yns_0: 0.1505, loss_cls_1: 0.9456, loss_box_1: 1.8815, loss_cns_1: 0.6397, loss_yns_1: 0.1492, loss_cls_2: 0.9763, loss_box_2: 1.8534, loss_cns_2: 0.6475, loss_yns_2: 0.1495, loss_cls_3: 0.9778, loss_box_3: 1.8659, loss_cns_3: 0.6454, loss_yns_3: 0.1498, loss_cls_4: 0.9869, loss_box_4: 1.8507, loss_cns_4: 0.6469, loss_yns_4: 0.1518, loss_cls_5: 0.9930, loss_box_5: 1.8660, loss_cns_5: 0.6452, loss_yns_5: 0.1545, loss_cls_dn_0: 0.2387, loss_box_dn_0: 0.7838, loss_cls_dn_1: 0.1504, loss_box_dn_1: 0.7443, loss_cls_dn_2: 0.1514, loss_box_dn_2: 0.7309, loss_cls_dn_3: 0.1534, loss_box_dn_3: 0.7390, loss_cls_dn_4: 0.1611, loss_box_dn_4: 0.7406, loss_cls_dn_5: 0.1680, loss_box_dn_5: 0.7545, loss_dense_depth: 0.7652, loss: 27.8494, grad_norm: 62.8316 -2025-11-13 15:37:21,335 - mmdet - INFO - Iter [138/17500] lr: 1.548e-04, eta: 11:15:48, time: 1.492, data_time: 0.075, memory: 49163, loss_cls_0: 0.8639, loss_box_0: 1.7361, loss_cns_0: 0.6237, loss_yns_0: 0.1494, loss_cls_1: 0.9384, loss_box_1: 1.8297, loss_cns_1: 0.6437, loss_yns_1: 0.1493, loss_cls_2: 0.9814, loss_box_2: 1.7733, loss_cns_2: 0.6530, loss_yns_2: 0.1496, loss_cls_3: 0.9782, loss_box_3: 1.7829, loss_cns_3: 0.6534, loss_yns_3: 0.1496, loss_cls_4: 0.9811, loss_box_4: 1.7554, loss_cns_4: 0.6546, loss_yns_4: 0.1504, loss_cls_5: 0.9958, loss_box_5: 1.7584, loss_cns_5: 0.6523, loss_yns_5: 0.1533, loss_cls_dn_0: 0.2351, loss_box_dn_0: 0.7921, loss_cls_dn_1: 0.1504, loss_box_dn_1: 0.7584, loss_cls_dn_2: 0.1505, loss_box_dn_2: 0.7345, loss_cls_dn_3: 0.1509, loss_box_dn_3: 0.7420, loss_cls_dn_4: 0.1616, loss_box_dn_4: 0.7385, loss_cls_dn_5: 0.1663, loss_box_dn_5: 0.7434, loss_dense_depth: 0.7887, loss: 27.4691, grad_norm: 34.4174 -2025-11-13 15:37:22,827 - mmdet - INFO - Iter [139/17500] lr: 1.552e-04, eta: 11:14:01, time: 1.492, data_time: 0.076, memory: 49163, loss_cls_0: 0.8271, loss_box_0: 1.7341, loss_cns_0: 0.6243, loss_yns_0: 0.1491, loss_cls_1: 0.8958, loss_box_1: 1.8198, loss_cns_1: 0.6428, loss_yns_1: 0.1503, loss_cls_2: 0.9373, loss_box_2: 1.7718, loss_cns_2: 0.6510, loss_yns_2: 0.1533, loss_cls_3: 0.9429, loss_box_3: 1.7716, loss_cns_3: 0.6505, loss_yns_3: 0.1490, loss_cls_4: 0.9476, loss_box_4: 1.7613, loss_cns_4: 0.6501, loss_yns_4: 0.1506, loss_cls_5: 0.9471, loss_box_5: 1.7624, loss_cns_5: 0.6475, loss_yns_5: 0.1502, loss_cls_dn_0: 0.2273, loss_box_dn_0: 0.7945, loss_cls_dn_1: 0.1498, loss_box_dn_1: 0.7474, loss_cls_dn_2: 0.1470, loss_box_dn_2: 0.7293, loss_cls_dn_3: 0.1495, loss_box_dn_3: 0.7349, loss_cls_dn_4: 0.1617, loss_box_dn_4: 0.7343, loss_cls_dn_5: 0.1669, loss_box_dn_5: 0.7316, loss_dense_depth: 0.7844, loss: 27.1461, grad_norm: 62.9380 -2025-11-13 15:37:24,351 - mmdet - INFO - Iter [140/17500] lr: 1.556e-04, eta: 11:12:17, time: 1.507, data_time: 0.087, memory: 49163, loss_cls_0: 0.8432, loss_box_0: 1.7095, loss_cns_0: 0.6258, loss_yns_0: 0.1487, loss_cls_1: 0.9175, loss_box_1: 1.7956, loss_cns_1: 0.6445, loss_yns_1: 0.1469, loss_cls_2: 0.9511, loss_box_2: 1.7609, loss_cns_2: 0.6542, loss_yns_2: 0.1533, loss_cls_3: 0.9568, loss_box_3: 1.7679, loss_cns_3: 0.6539, loss_yns_3: 0.1490, loss_cls_4: 0.9713, loss_box_4: 1.7527, loss_cns_4: 0.6558, loss_yns_4: 0.1485, loss_cls_5: 0.9477, loss_box_5: 1.7646, loss_cns_5: 0.6515, loss_yns_5: 0.1476, loss_cls_dn_0: 0.2239, loss_box_dn_0: 0.7927, loss_cls_dn_1: 0.1475, loss_box_dn_1: 0.7414, loss_cls_dn_2: 0.1460, loss_box_dn_2: 0.7297, loss_cls_dn_3: 0.1500, loss_box_dn_3: 0.7365, loss_cls_dn_4: 0.1593, loss_box_dn_4: 0.7354, loss_cls_dn_5: 0.1661, loss_box_dn_5: 0.7369, loss_dense_depth: 0.7728, loss: 27.1568, grad_norm: 72.0648 -2025-11-13 15:37:25,891 - mmdet - INFO - Iter [141/17500] lr: 1.560e-04, eta: 11:10:40, time: 1.558, data_time: 0.086, memory: 49163, loss_cls_0: 0.8928, loss_box_0: 1.7694, loss_cns_0: 0.6261, loss_yns_0: 0.1515, loss_cls_1: 0.9475, loss_box_1: 1.8594, loss_cns_1: 0.6457, loss_yns_1: 0.1495, loss_cls_2: 0.9795, loss_box_2: 1.8242, loss_cns_2: 0.6562, loss_yns_2: 0.1531, loss_cls_3: 0.9804, loss_box_3: 1.8401, loss_cns_3: 0.6531, loss_yns_3: 0.1527, loss_cls_4: 0.9995, loss_box_4: 1.8143, loss_cns_4: 0.6561, loss_yns_4: 0.1502, loss_cls_5: 0.9859, loss_box_5: 1.8315, loss_cns_5: 0.6499, loss_yns_5: 0.1543, loss_cls_dn_0: 0.2368, loss_box_dn_0: 0.7874, loss_cls_dn_1: 0.1599, loss_box_dn_1: 0.7477, loss_cls_dn_2: 0.1567, loss_box_dn_2: 0.7368, loss_cls_dn_3: 0.1589, loss_box_dn_3: 0.7440, loss_cls_dn_4: 0.1688, loss_box_dn_4: 0.7417, loss_cls_dn_5: 0.1762, loss_box_dn_5: 0.7475, loss_dense_depth: 0.7936, loss: 27.8790, grad_norm: 59.0515 -2025-11-13 15:37:27,450 - mmdet - INFO - Iter [142/17500] lr: 1.564e-04, eta: 11:09:05, time: 1.558, data_time: 0.157, memory: 49163, loss_cls_0: 0.8955, loss_box_0: 1.8108, loss_cns_0: 0.6200, loss_yns_0: 0.1540, loss_cls_1: 0.9494, loss_box_1: 1.9036, loss_cns_1: 0.6407, loss_yns_1: 0.1525, loss_cls_2: 0.9785, loss_box_2: 1.8407, loss_cns_2: 0.6492, loss_yns_2: 0.1552, loss_cls_3: 0.9782, loss_box_3: 1.8508, loss_cns_3: 0.6477, loss_yns_3: 0.1532, loss_cls_4: 0.9835, loss_box_4: 1.8290, loss_cns_4: 0.6497, loss_yns_4: 0.1527, loss_cls_5: 0.9807, loss_box_5: 1.8188, loss_cns_5: 0.6458, loss_yns_5: 0.1555, loss_cls_dn_0: 0.2354, loss_box_dn_0: 0.7914, loss_cls_dn_1: 0.1579, loss_box_dn_1: 0.7437, loss_cls_dn_2: 0.1558, loss_box_dn_2: 0.7249, loss_cls_dn_3: 0.1573, loss_box_dn_3: 0.7299, loss_cls_dn_4: 0.1629, loss_box_dn_4: 0.7290, loss_cls_dn_5: 0.1704, loss_box_dn_5: 0.7283, loss_dense_depth: 0.8023, loss: 27.8850, grad_norm: 45.1395 -2025-11-13 15:37:28,963 - mmdet - INFO - Iter [143/17500] lr: 1.568e-04, eta: 11:07:25, time: 1.514, data_time: 0.071, memory: 49163, loss_cls_0: 0.8671, loss_box_0: 1.8056, loss_cns_0: 0.6194, loss_yns_0: 0.1483, loss_cls_1: 0.9289, loss_box_1: 1.8767, loss_cns_1: 0.6462, loss_yns_1: 0.1496, loss_cls_2: 0.9663, loss_box_2: 1.8020, loss_cns_2: 0.6538, loss_yns_2: 0.1546, loss_cls_3: 0.9843, loss_box_3: 1.8049, loss_cns_3: 0.6539, loss_yns_3: 0.1521, loss_cls_4: 0.9716, loss_box_4: 1.7993, loss_cns_4: 0.6531, loss_yns_4: 0.1510, loss_cls_5: 0.9538, loss_box_5: 1.7867, loss_cns_5: 0.6502, loss_yns_5: 0.1504, loss_cls_dn_0: 0.2269, loss_box_dn_0: 0.7907, loss_cls_dn_1: 0.1532, loss_box_dn_1: 0.7516, loss_cls_dn_2: 0.1546, loss_box_dn_2: 0.7294, loss_cls_dn_3: 0.1622, loss_box_dn_3: 0.7346, loss_cls_dn_4: 0.1672, loss_box_dn_4: 0.7412, loss_cls_dn_5: 0.1720, loss_box_dn_5: 0.7430, loss_dense_depth: 0.7630, loss: 27.6193, grad_norm: 57.7700 -2025-11-13 15:37:30,529 - mmdet - INFO - Iter [144/17500] lr: 1.572e-04, eta: 11:05:54, time: 1.566, data_time: 0.074, memory: 49163, loss_cls_0: 0.8887, loss_box_0: 1.8198, loss_cns_0: 0.6179, loss_yns_0: 0.1514, loss_cls_1: 0.9533, loss_box_1: 1.9113, loss_cns_1: 0.6405, loss_yns_1: 0.1504, loss_cls_2: 0.9823, loss_box_2: 1.8597, loss_cns_2: 0.6471, loss_yns_2: 0.1556, loss_cls_3: 1.0064, loss_box_3: 1.8463, loss_cns_3: 0.6492, loss_yns_3: 0.1540, loss_cls_4: 1.0028, loss_box_4: 1.8451, loss_cns_4: 0.6507, loss_yns_4: 0.1524, loss_cls_5: 0.9907, loss_box_5: 1.8491, loss_cns_5: 0.6497, loss_yns_5: 0.1509, loss_cls_dn_0: 0.2361, loss_box_dn_0: 0.7923, loss_cls_dn_1: 0.1609, loss_box_dn_1: 0.7529, loss_cls_dn_2: 0.1629, loss_box_dn_2: 0.7311, loss_cls_dn_3: 0.1723, loss_box_dn_3: 0.7342, loss_cls_dn_4: 0.1731, loss_box_dn_4: 0.7426, loss_cls_dn_5: 0.1770, loss_box_dn_5: 0.7515, loss_dense_depth: 0.7689, loss: 28.0812, grad_norm: 56.3030 -2025-11-13 15:37:32,028 - mmdet - INFO - Iter [145/17500] lr: 1.576e-04, eta: 11:04:15, time: 1.499, data_time: 0.082, memory: 49163, loss_cls_0: 0.8756, loss_box_0: 1.7896, loss_cns_0: 0.6216, loss_yns_0: 0.1515, loss_cls_1: 0.9542, loss_box_1: 1.9122, loss_cns_1: 0.6411, loss_yns_1: 0.1518, loss_cls_2: 0.9846, loss_box_2: 1.8580, loss_cns_2: 0.6488, loss_yns_2: 0.1535, loss_cls_3: 0.9854, loss_box_3: 1.8473, loss_cns_3: 0.6515, loss_yns_3: 0.1548, loss_cls_4: 0.9926, loss_box_4: 1.8426, loss_cns_4: 0.6516, loss_yns_4: 0.1536, loss_cls_5: 0.9920, loss_box_5: 1.8479, loss_cns_5: 0.6501, loss_yns_5: 0.1532, loss_cls_dn_0: 0.2302, loss_box_dn_0: 0.7868, loss_cls_dn_1: 0.1597, loss_box_dn_1: 0.7586, loss_cls_dn_2: 0.1651, loss_box_dn_2: 0.7357, loss_cls_dn_3: 0.1640, loss_box_dn_3: 0.7363, loss_cls_dn_4: 0.1667, loss_box_dn_4: 0.7453, loss_cls_dn_5: 0.1698, loss_box_dn_5: 0.7544, loss_dense_depth: 0.7604, loss: 27.9981, grad_norm: 41.0506 -2025-11-13 15:37:33,538 - mmdet - INFO - Iter [146/17500] lr: 1.580e-04, eta: 11:02:39, time: 1.510, data_time: 0.076, memory: 49163, loss_cls_0: 0.8569, loss_box_0: 1.7657, loss_cns_0: 0.6210, loss_yns_0: 0.1499, loss_cls_1: 0.9229, loss_box_1: 1.8438, loss_cns_1: 0.6443, loss_yns_1: 0.1501, loss_cls_2: 0.9543, loss_box_2: 1.7948, loss_cns_2: 0.6515, loss_yns_2: 0.1521, loss_cls_3: 0.9626, loss_box_3: 1.7893, loss_cns_3: 0.6527, loss_yns_3: 0.1515, loss_cls_4: 0.9730, loss_box_4: 1.7982, loss_cns_4: 0.6519, loss_yns_4: 0.1536, loss_cls_5: 0.9668, loss_box_5: 1.8075, loss_cns_5: 0.6490, loss_yns_5: 0.1551, loss_cls_dn_0: 0.2270, loss_box_dn_0: 0.7836, loss_cls_dn_1: 0.1519, loss_box_dn_1: 0.7635, loss_cls_dn_2: 0.1570, loss_box_dn_2: 0.7406, loss_cls_dn_3: 0.1520, loss_box_dn_3: 0.7366, loss_cls_dn_4: 0.1613, loss_box_dn_4: 0.7470, loss_cls_dn_5: 0.1664, loss_box_dn_5: 0.7543, loss_dense_depth: 0.7599, loss: 27.5196, grad_norm: 43.9756 -2025-11-13 15:37:35,019 - mmdet - INFO - Iter [147/17500] lr: 1.584e-04, eta: 11:01:02, time: 1.482, data_time: 0.075, memory: 49163, loss_cls_0: 0.8823, loss_box_0: 1.8003, loss_cns_0: 0.6187, loss_yns_0: 0.1502, loss_cls_1: 0.9533, loss_box_1: 1.8304, loss_cns_1: 0.6454, loss_yns_1: 0.1499, loss_cls_2: 0.9730, loss_box_2: 1.7754, loss_cns_2: 0.6545, loss_yns_2: 0.1505, loss_cls_3: 0.9918, loss_box_3: 1.7636, loss_cns_3: 0.6555, loss_yns_3: 0.1512, loss_cls_4: 0.9781, loss_box_4: 1.7671, loss_cns_4: 0.6552, loss_yns_4: 0.1516, loss_cls_5: 0.9911, loss_box_5: 1.7600, loss_cns_5: 0.6538, loss_yns_5: 0.1530, loss_cls_dn_0: 0.2303, loss_box_dn_0: 0.7858, loss_cls_dn_1: 0.1503, loss_box_dn_1: 0.7523, loss_cls_dn_2: 0.1557, loss_box_dn_2: 0.7333, loss_cls_dn_3: 0.1564, loss_box_dn_3: 0.7254, loss_cls_dn_4: 0.1632, loss_box_dn_4: 0.7303, loss_cls_dn_5: 0.1697, loss_box_dn_5: 0.7318, loss_dense_depth: 0.7765, loss: 27.5170, grad_norm: 44.5809 -2025-11-13 15:37:36,508 - mmdet - INFO - Iter [148/17500] lr: 1.588e-04, eta: 10:59:26, time: 1.487, data_time: 0.074, memory: 49163, loss_cls_0: 0.8560, loss_box_0: 1.7803, loss_cns_0: 0.6218, loss_yns_0: 0.1490, loss_cls_1: 0.9605, loss_box_1: 1.8171, loss_cns_1: 0.6464, loss_yns_1: 0.1488, loss_cls_2: 0.9808, loss_box_2: 1.7600, loss_cns_2: 0.6534, loss_yns_2: 0.1508, loss_cls_3: 0.9862, loss_box_3: 1.7731, loss_cns_3: 0.6541, loss_yns_3: 0.1510, loss_cls_4: 0.9756, loss_box_4: 1.7593, loss_cns_4: 0.6578, loss_yns_4: 0.1484, loss_cls_5: 0.9823, loss_box_5: 1.7388, loss_cns_5: 0.6603, loss_yns_5: 0.1494, loss_cls_dn_0: 0.2240, loss_box_dn_0: 0.7870, loss_cls_dn_1: 0.1510, loss_box_dn_1: 0.7350, loss_cls_dn_2: 0.1518, loss_box_dn_2: 0.7144, loss_cls_dn_3: 0.1572, loss_box_dn_3: 0.7153, loss_cls_dn_4: 0.1637, loss_box_dn_4: 0.7117, loss_cls_dn_5: 0.1710, loss_box_dn_5: 0.7106, loss_dense_depth: 0.7469, loss: 27.3012, grad_norm: 46.2465 -2025-11-13 15:37:38,041 - mmdet - INFO - Iter [149/17500] lr: 1.592e-04, eta: 10:57:57, time: 1.533, data_time: 0.075, memory: 49163, loss_cls_0: 0.8543, loss_box_0: 1.7616, loss_cns_0: 0.6211, loss_yns_0: 0.1484, loss_cls_1: 0.9381, loss_box_1: 1.8275, loss_cns_1: 0.6430, loss_yns_1: 0.1508, loss_cls_2: 0.9689, loss_box_2: 1.7668, loss_cns_2: 0.6498, loss_yns_2: 0.1546, loss_cls_3: 0.9652, loss_box_3: 1.8010, loss_cns_3: 0.6508, loss_yns_3: 0.1546, loss_cls_4: 0.9704, loss_box_4: 1.7801, loss_cns_4: 0.6534, loss_yns_4: 0.1499, loss_cls_5: 0.9720, loss_box_5: 1.7945, loss_cns_5: 0.6527, loss_yns_5: 0.1510, loss_cls_dn_0: 0.2212, loss_box_dn_0: 0.7784, loss_cls_dn_1: 0.1505, loss_box_dn_1: 0.7442, loss_cls_dn_2: 0.1485, loss_box_dn_2: 0.7214, loss_cls_dn_3: 0.1535, loss_box_dn_3: 0.7366, loss_cls_dn_4: 0.1609, loss_box_dn_4: 0.7322, loss_cls_dn_5: 0.1775, loss_box_dn_5: 0.7501, loss_dense_depth: 0.7231, loss: 27.3785, grad_norm: 43.6179 -2025-11-13 15:37:39,539 - mmdet - INFO - Iter [150/17500] lr: 1.596e-04, eta: 10:56:25, time: 1.499, data_time: 0.070, memory: 49163, loss_cls_0: 0.8742, loss_box_0: 1.7488, loss_cns_0: 0.6186, loss_yns_0: 0.1467, loss_cls_1: 0.9361, loss_box_1: 1.8150, loss_cns_1: 0.6474, loss_yns_1: 0.1487, loss_cls_2: 0.9646, loss_box_2: 1.7665, loss_cns_2: 0.6523, loss_yns_2: 0.1501, loss_cls_3: 0.9650, loss_box_3: 1.7878, loss_cns_3: 0.6550, loss_yns_3: 0.1519, loss_cls_4: 0.9779, loss_box_4: 1.7780, loss_cns_4: 0.6558, loss_yns_4: 0.1494, loss_cls_5: 0.9767, loss_box_5: 1.8057, loss_cns_5: 0.6534, loss_yns_5: 0.1521, loss_cls_dn_0: 0.2224, loss_box_dn_0: 0.7741, loss_cls_dn_1: 0.1549, loss_box_dn_1: 0.7473, loss_cls_dn_2: 0.1548, loss_box_dn_2: 0.7299, loss_cls_dn_3: 0.1602, loss_box_dn_3: 0.7492, loss_cls_dn_4: 0.1670, loss_box_dn_4: 0.7557, loss_cls_dn_5: 0.1841, loss_box_dn_5: 0.7795, loss_dense_depth: 0.7585, loss: 27.5154, grad_norm: 51.5686 -2025-11-13 15:37:41,038 - mmdet - INFO - Iter [151/17500] lr: 1.600e-04, eta: 10:54:54, time: 1.499, data_time: 0.075, memory: 49163, loss_cls_0: 0.8767, loss_box_0: 1.7347, loss_cns_0: 0.6206, loss_yns_0: 0.1461, loss_cls_1: 0.9455, loss_box_1: 1.8246, loss_cns_1: 0.6427, loss_yns_1: 0.1479, loss_cls_2: 0.9535, loss_box_2: 1.7841, loss_cns_2: 0.6508, loss_yns_2: 0.1493, loss_cls_3: 0.9619, loss_box_3: 1.7766, loss_cns_3: 0.6512, loss_yns_3: 0.1499, loss_cls_4: 0.9772, loss_box_4: 1.7944, loss_cns_4: 0.6514, loss_yns_4: 0.1495, loss_cls_5: 0.9680, loss_box_5: 1.7938, loss_cns_5: 0.6492, loss_yns_5: 0.1537, loss_cls_dn_0: 0.2218, loss_box_dn_0: 0.7789, loss_cls_dn_1: 0.1548, loss_box_dn_1: 0.7633, loss_cls_dn_2: 0.1555, loss_box_dn_2: 0.7433, loss_cls_dn_3: 0.1566, loss_box_dn_3: 0.7484, loss_cls_dn_4: 0.1605, loss_box_dn_4: 0.7624, loss_cls_dn_5: 0.1675, loss_box_dn_5: 0.7705, loss_dense_depth: 0.7589, loss: 27.4958, grad_norm: 51.4416 -2025-11-13 15:37:42,527 - mmdet - INFO - Iter [152/17500] lr: 1.604e-04, eta: 10:53:23, time: 1.489, data_time: 0.076, memory: 49163, loss_cls_0: 0.8703, loss_box_0: 1.7040, loss_cns_0: 0.6150, loss_yns_0: 0.1479, loss_cls_1: 0.9573, loss_box_1: 1.7988, loss_cns_1: 0.6459, loss_yns_1: 0.1503, loss_cls_2: 0.9513, loss_box_2: 1.7585, loss_cns_2: 0.6530, loss_yns_2: 0.1512, loss_cls_3: 0.9697, loss_box_3: 1.7431, loss_cns_3: 0.6535, loss_yns_3: 0.1524, loss_cls_4: 0.9834, loss_box_4: 1.7574, loss_cns_4: 0.6543, loss_yns_4: 0.1518, loss_cls_5: 0.9894, loss_box_5: 1.7398, loss_cns_5: 0.6519, loss_yns_5: 0.1518, loss_cls_dn_0: 0.2211, loss_box_dn_0: 0.7776, loss_cls_dn_1: 0.1554, loss_box_dn_1: 0.7687, loss_cls_dn_2: 0.1513, loss_box_dn_2: 0.7458, loss_cls_dn_3: 0.1509, loss_box_dn_3: 0.7447, loss_cls_dn_4: 0.1564, loss_box_dn_4: 0.7564, loss_cls_dn_5: 0.1646, loss_box_dn_5: 0.7494, loss_dense_depth: 0.7318, loss: 27.2759, grad_norm: 38.6280 -2025-11-13 15:37:44,013 - mmdet - INFO - Iter [153/17500] lr: 1.608e-04, eta: 10:51:53, time: 1.485, data_time: 0.072, memory: 49163, loss_cls_0: 0.9007, loss_box_0: 1.7186, loss_cns_0: 0.6072, loss_yns_0: 0.1472, loss_cls_1: 0.9643, loss_box_1: 1.7796, loss_cns_1: 0.6471, loss_yns_1: 0.1510, loss_cls_2: 0.9635, loss_box_2: 1.7587, loss_cns_2: 0.6537, loss_yns_2: 0.1529, loss_cls_3: 0.9862, loss_box_3: 1.7488, loss_cns_3: 0.6542, loss_yns_3: 0.1538, loss_cls_4: 0.9808, loss_box_4: 1.7563, loss_cns_4: 0.6556, loss_yns_4: 0.1532, loss_cls_5: 0.9859, loss_box_5: 1.7477, loss_cns_5: 0.6539, loss_yns_5: 0.1511, loss_cls_dn_0: 0.2341, loss_box_dn_0: 0.7853, loss_cls_dn_1: 0.1481, loss_box_dn_1: 0.7448, loss_cls_dn_2: 0.1445, loss_box_dn_2: 0.7298, loss_cls_dn_3: 0.1498, loss_box_dn_3: 0.7328, loss_cls_dn_4: 0.1578, loss_box_dn_4: 0.7424, loss_cls_dn_5: 0.1685, loss_box_dn_5: 0.7371, loss_dense_depth: 0.7649, loss: 27.3120, grad_norm: 45.0183 -2025-11-13 15:37:45,497 - mmdet - INFO - Iter [154/17500] lr: 1.612e-04, eta: 10:50:24, time: 1.484, data_time: 0.074, memory: 49163, loss_cls_0: 0.8570, loss_box_0: 1.7050, loss_cns_0: 0.6136, loss_yns_0: 0.1498, loss_cls_1: 0.9193, loss_box_1: 1.7571, loss_cns_1: 0.6438, loss_yns_1: 0.1489, loss_cls_2: 0.9551, loss_box_2: 1.7452, loss_cns_2: 0.6506, loss_yns_2: 0.1517, loss_cls_3: 0.9521, loss_box_3: 1.7350, loss_cns_3: 0.6522, loss_yns_3: 0.1524, loss_cls_4: 0.9626, loss_box_4: 1.7269, loss_cns_4: 0.6595, loss_yns_4: 0.1515, loss_cls_5: 0.9515, loss_box_5: 1.7467, loss_cns_5: 0.6565, loss_yns_5: 0.1531, loss_cls_dn_0: 0.2247, loss_box_dn_0: 0.7747, loss_cls_dn_1: 0.1434, loss_box_dn_1: 0.7399, loss_cls_dn_2: 0.1459, loss_box_dn_2: 0.7308, loss_cls_dn_3: 0.1569, loss_box_dn_3: 0.7317, loss_cls_dn_4: 0.1596, loss_box_dn_4: 0.7380, loss_cls_dn_5: 0.1686, loss_box_dn_5: 0.7451, loss_dense_depth: 0.7227, loss: 26.9792, grad_norm: 44.3251 -2025-11-13 15:37:46,982 - mmdet - INFO - Iter [155/17500] lr: 1.616e-04, eta: 10:48:56, time: 1.487, data_time: 0.080, memory: 49163, loss_cls_0: 0.8424, loss_box_0: 1.7159, loss_cns_0: 0.6202, loss_yns_0: 0.1524, loss_cls_1: 0.8815, loss_box_1: 1.7386, loss_cns_1: 0.6427, loss_yns_1: 0.1493, loss_cls_2: 0.9354, loss_box_2: 1.7140, loss_cns_2: 0.6493, loss_yns_2: 0.1506, loss_cls_3: 0.9256, loss_box_3: 1.6967, loss_cns_3: 0.6534, loss_yns_3: 0.1518, loss_cls_4: 0.9475, loss_box_4: 1.6906, loss_cns_4: 0.6595, loss_yns_4: 0.1528, loss_cls_5: 0.9368, loss_box_5: 1.7064, loss_cns_5: 0.6559, loss_yns_5: 0.1514, loss_cls_dn_0: 0.2192, loss_box_dn_0: 0.7807, loss_cls_dn_1: 0.1436, loss_box_dn_1: 0.7401, loss_cls_dn_2: 0.1465, loss_box_dn_2: 0.7346, loss_cls_dn_3: 0.1530, loss_box_dn_3: 0.7316, loss_cls_dn_4: 0.1556, loss_box_dn_4: 0.7396, loss_cls_dn_5: 0.1602, loss_box_dn_5: 0.7472, loss_dense_depth: 0.7305, loss: 26.7030, grad_norm: 53.4709 -2025-11-13 15:37:48,469 - mmdet - INFO - Iter [156/17500] lr: 1.620e-04, eta: 10:47:29, time: 1.484, data_time: 0.071, memory: 49163, loss_cls_0: 0.8633, loss_box_0: 1.7329, loss_cns_0: 0.6174, loss_yns_0: 0.1513, loss_cls_1: 0.8948, loss_box_1: 1.7757, loss_cns_1: 0.6379, loss_yns_1: 0.1502, loss_cls_2: 0.9244, loss_box_2: 1.7373, loss_cns_2: 0.6449, loss_yns_2: 0.1530, loss_cls_3: 0.9419, loss_box_3: 1.7203, loss_cns_3: 0.6491, loss_yns_3: 0.1527, loss_cls_4: 0.9753, loss_box_4: 1.7017, loss_cns_4: 0.6483, loss_yns_4: 0.1511, loss_cls_5: 0.9623, loss_box_5: 1.7148, loss_cns_5: 0.6500, loss_yns_5: 0.1528, loss_cls_dn_0: 0.2248, loss_box_dn_0: 0.7804, loss_cls_dn_1: 0.1453, loss_box_dn_1: 0.7634, loss_cls_dn_2: 0.1433, loss_box_dn_2: 0.7481, loss_cls_dn_3: 0.1472, loss_box_dn_3: 0.7436, loss_cls_dn_4: 0.1519, loss_box_dn_4: 0.7514, loss_cls_dn_5: 0.1571, loss_box_dn_5: 0.7539, loss_dense_depth: 0.7431, loss: 26.9568, grad_norm: 38.7660 -2025-11-13 15:37:49,958 - mmdet - INFO - Iter [157/17500] lr: 1.624e-04, eta: 10:46:04, time: 1.491, data_time: 0.072, memory: 49163, loss_cls_0: 0.8567, loss_box_0: 1.7190, loss_cns_0: 0.6199, loss_yns_0: 0.1521, loss_cls_1: 0.9061, loss_box_1: 1.8093, loss_cns_1: 0.6339, loss_yns_1: 0.1510, loss_cls_2: 0.9666, loss_box_2: 1.7368, loss_cns_2: 0.6501, loss_yns_2: 0.1493, loss_cls_3: 0.9655, loss_box_3: 1.7380, loss_cns_3: 0.6518, loss_yns_3: 0.1512, loss_cls_4: 0.9859, loss_box_4: 1.7351, loss_cns_4: 0.6520, loss_yns_4: 0.1502, loss_cls_5: 0.9897, loss_box_5: 1.7613, loss_cns_5: 0.6498, loss_yns_5: 0.1541, loss_cls_dn_0: 0.2199, loss_box_dn_0: 0.7750, loss_cls_dn_1: 0.1437, loss_box_dn_1: 0.7328, loss_cls_dn_2: 0.1418, loss_box_dn_2: 0.7144, loss_cls_dn_3: 0.1438, loss_box_dn_3: 0.7154, loss_cls_dn_4: 0.1495, loss_box_dn_4: 0.7271, loss_cls_dn_5: 0.1634, loss_box_dn_5: 0.7357, loss_dense_depth: 0.7393, loss: 27.0370, grad_norm: 41.8592 -2025-11-13 15:37:51,441 - mmdet - INFO - Iter [158/17500] lr: 1.628e-04, eta: 10:44:39, time: 1.483, data_time: 0.071, memory: 49163, loss_cls_0: 0.8255, loss_box_0: 1.7200, loss_cns_0: 0.6165, loss_yns_0: 0.1526, loss_cls_1: 0.8812, loss_box_1: 1.7764, loss_cns_1: 0.6372, loss_yns_1: 0.1521, loss_cls_2: 0.9449, loss_box_2: 1.7193, loss_cns_2: 0.6508, loss_yns_2: 0.1529, loss_cls_3: 0.9360, loss_box_3: 1.7095, loss_cns_3: 0.6511, loss_yns_3: 0.1524, loss_cls_4: 0.9398, loss_box_4: 1.7212, loss_cns_4: 0.6579, loss_yns_4: 0.1526, loss_cls_5: 0.9516, loss_box_5: 1.7371, loss_cns_5: 0.6513, loss_yns_5: 0.1534, loss_cls_dn_0: 0.2136, loss_box_dn_0: 0.7777, loss_cls_dn_1: 0.1454, loss_box_dn_1: 0.7488, loss_cls_dn_2: 0.1440, loss_box_dn_2: 0.7322, loss_cls_dn_3: 0.1450, loss_box_dn_3: 0.7340, loss_cls_dn_4: 0.1527, loss_box_dn_4: 0.7431, loss_cls_dn_5: 0.1659, loss_box_dn_5: 0.7566, loss_dense_depth: 0.7134, loss: 26.8155, grad_norm: 49.8675 -2025-11-13 15:37:52,934 - mmdet - INFO - Iter [159/17500] lr: 1.632e-04, eta: 10:43:17, time: 1.493, data_time: 0.071, memory: 49163, loss_cls_0: 0.8402, loss_box_0: 1.7511, loss_cns_0: 0.6171, loss_yns_0: 0.1543, loss_cls_1: 0.8921, loss_box_1: 1.7762, loss_cns_1: 0.6430, loss_yns_1: 0.1545, loss_cls_2: 0.9225, loss_box_2: 1.7193, loss_cns_2: 0.6533, loss_yns_2: 0.1563, loss_cls_3: 0.9468, loss_box_3: 1.7141, loss_cns_3: 0.6529, loss_yns_3: 0.1571, loss_cls_4: 0.9629, loss_box_4: 1.7119, loss_cns_4: 0.6577, loss_yns_4: 0.1564, loss_cls_5: 0.9381, loss_box_5: 1.7216, loss_cns_5: 0.6527, loss_yns_5: 0.1557, loss_cls_dn_0: 0.2142, loss_box_dn_0: 0.7790, loss_cls_dn_1: 0.1418, loss_box_dn_1: 0.7466, loss_cls_dn_2: 0.1401, loss_box_dn_2: 0.7285, loss_cls_dn_3: 0.1431, loss_box_dn_3: 0.7312, loss_cls_dn_4: 0.1500, loss_box_dn_4: 0.7326, loss_cls_dn_5: 0.1569, loss_box_dn_5: 0.7403, loss_dense_depth: 0.7344, loss: 26.8463, grad_norm: 43.5872 -2025-11-13 15:37:54,437 - mmdet - INFO - Iter [160/17500] lr: 1.636e-04, eta: 10:41:56, time: 1.502, data_time: 0.090, memory: 49163, loss_cls_0: 0.8378, loss_box_0: 1.7477, loss_cns_0: 0.6169, loss_yns_0: 0.1551, loss_cls_1: 0.8812, loss_box_1: 1.7683, loss_cns_1: 0.6447, loss_yns_1: 0.1547, loss_cls_2: 0.9200, loss_box_2: 1.7033, loss_cns_2: 0.6525, loss_yns_2: 0.1542, loss_cls_3: 0.9331, loss_box_3: 1.7060, loss_cns_3: 0.6514, loss_yns_3: 0.1555, loss_cls_4: 0.9558, loss_box_4: 1.7055, loss_cns_4: 0.6512, loss_yns_4: 0.1549, loss_cls_5: 0.9414, loss_box_5: 1.7346, loss_cns_5: 0.6480, loss_yns_5: 0.1546, loss_cls_dn_0: 0.2144, loss_box_dn_0: 0.7763, loss_cls_dn_1: 0.1394, loss_box_dn_1: 0.7409, loss_cls_dn_2: 0.1373, loss_box_dn_2: 0.7196, loss_cls_dn_3: 0.1398, loss_box_dn_3: 0.7232, loss_cls_dn_4: 0.1465, loss_box_dn_4: 0.7254, loss_cls_dn_5: 0.1513, loss_box_dn_5: 0.7399, loss_dense_depth: 0.7228, loss: 26.7051, grad_norm: 39.8546 -2025-11-13 15:37:55,995 - mmdet - INFO - Iter [161/17500] lr: 1.640e-04, eta: 10:40:43, time: 1.558, data_time: 0.077, memory: 49163, loss_cls_0: 0.8448, loss_box_0: 1.7329, loss_cns_0: 0.6162, loss_yns_0: 0.1547, loss_cls_1: 0.8878, loss_box_1: 1.7595, loss_cns_1: 0.6454, loss_yns_1: 0.1558, loss_cls_2: 0.9243, loss_box_2: 1.6970, loss_cns_2: 0.6545, loss_yns_2: 0.1547, loss_cls_3: 0.9310, loss_box_3: 1.6809, loss_cns_3: 0.6561, loss_yns_3: 0.1563, loss_cls_4: 0.9410, loss_box_4: 1.6823, loss_cns_4: 0.6567, loss_yns_4: 0.1572, loss_cls_5: 0.9349, loss_box_5: 1.7080, loss_cns_5: 0.6547, loss_yns_5: 0.1558, loss_cls_dn_0: 0.2146, loss_box_dn_0: 0.7774, loss_cls_dn_1: 0.1392, loss_box_dn_1: 0.7165, loss_cls_dn_2: 0.1378, loss_box_dn_2: 0.6914, loss_cls_dn_3: 0.1424, loss_box_dn_3: 0.6911, loss_cls_dn_4: 0.1459, loss_box_dn_4: 0.6979, loss_cls_dn_5: 0.1507, loss_box_dn_5: 0.7141, loss_dense_depth: 0.7189, loss: 26.4804, grad_norm: 30.6485 -2025-11-13 15:37:57,558 - mmdet - INFO - Iter [162/17500] lr: 1.644e-04, eta: 10:39:30, time: 1.563, data_time: 0.154, memory: 49163, loss_cls_0: 0.8443, loss_box_0: 1.7362, loss_cns_0: 0.6153, loss_yns_0: 0.1564, loss_cls_1: 0.8976, loss_box_1: 1.7577, loss_cns_1: 0.6435, loss_yns_1: 0.1574, loss_cls_2: 0.9295, loss_box_2: 1.6954, loss_cns_2: 0.6534, loss_yns_2: 0.1569, loss_cls_3: 0.9351, loss_box_3: 1.6820, loss_cns_3: 0.6546, loss_yns_3: 0.1563, loss_cls_4: 0.9447, loss_box_4: 1.6842, loss_cns_4: 0.6545, loss_yns_4: 0.1572, loss_cls_5: 0.9451, loss_box_5: 1.6846, loss_cns_5: 0.6521, loss_yns_5: 0.1553, loss_cls_dn_0: 0.2164, loss_box_dn_0: 0.7809, loss_cls_dn_1: 0.1352, loss_box_dn_1: 0.7296, loss_cls_dn_2: 0.1333, loss_box_dn_2: 0.7101, loss_cls_dn_3: 0.1407, loss_box_dn_3: 0.7122, loss_cls_dn_4: 0.1451, loss_box_dn_4: 0.7223, loss_cls_dn_5: 0.1488, loss_box_dn_5: 0.7291, loss_dense_depth: 0.7312, loss: 26.5842, grad_norm: 37.7171 -2025-11-13 15:37:59,086 - mmdet - INFO - Iter [163/17500] lr: 1.648e-04, eta: 10:38:15, time: 1.527, data_time: 0.076, memory: 49163, loss_cls_0: 0.8317, loss_box_0: 1.7583, loss_cns_0: 0.6208, loss_yns_0: 0.1574, loss_cls_1: 0.9035, loss_box_1: 1.7626, loss_cns_1: 0.6436, loss_yns_1: 0.1584, loss_cls_2: 0.9472, loss_box_2: 1.7098, loss_cns_2: 0.6526, loss_yns_2: 0.1574, loss_cls_3: 0.9355, loss_box_3: 1.7000, loss_cns_3: 0.6521, loss_yns_3: 0.1572, loss_cls_4: 0.9467, loss_box_4: 1.6942, loss_cns_4: 0.6553, loss_yns_4: 0.1576, loss_cls_5: 0.9432, loss_box_5: 1.7135, loss_cns_5: 0.6514, loss_yns_5: 0.1577, loss_cls_dn_0: 0.2129, loss_box_dn_0: 0.7682, loss_cls_dn_1: 0.1365, loss_box_dn_1: 0.7182, loss_cls_dn_2: 0.1355, loss_box_dn_2: 0.7057, loss_cls_dn_3: 0.1367, loss_box_dn_3: 0.7081, loss_cls_dn_4: 0.1453, loss_box_dn_4: 0.7151, loss_cls_dn_5: 0.1482, loss_box_dn_5: 0.7267, loss_dense_depth: 0.7031, loss: 26.6277, grad_norm: 30.1368 -2025-11-13 15:38:00,651 - mmdet - INFO - Iter [164/17500] lr: 1.652e-04, eta: 10:37:05, time: 1.566, data_time: 0.078, memory: 49163, loss_cls_0: 0.8379, loss_box_0: 1.7761, loss_cns_0: 0.6249, loss_yns_0: 0.1600, loss_cls_1: 0.9065, loss_box_1: 1.7696, loss_cns_1: 0.6497, loss_yns_1: 0.1592, loss_cls_2: 0.9334, loss_box_2: 1.7130, loss_cns_2: 0.6564, loss_yns_2: 0.1585, loss_cls_3: 0.9370, loss_box_3: 1.6981, loss_cns_3: 0.6585, loss_yns_3: 0.1584, loss_cls_4: 0.9510, loss_box_4: 1.6966, loss_cns_4: 0.6587, loss_yns_4: 0.1579, loss_cls_5: 0.9431, loss_box_5: 1.7266, loss_cns_5: 0.6562, loss_yns_5: 0.1578, loss_cls_dn_0: 0.2131, loss_box_dn_0: 0.7732, loss_cls_dn_1: 0.1361, loss_box_dn_1: 0.7424, loss_cls_dn_2: 0.1333, loss_box_dn_2: 0.7227, loss_cls_dn_3: 0.1348, loss_box_dn_3: 0.7199, loss_cls_dn_4: 0.1407, loss_box_dn_4: 0.7233, loss_cls_dn_5: 0.1462, loss_box_dn_5: 0.7405, loss_dense_depth: 0.7221, loss: 26.7932, grad_norm: 30.1308 -2025-11-13 15:38:02,156 - mmdet - INFO - Iter [165/17500] lr: 1.656e-04, eta: 10:35:49, time: 1.504, data_time: 0.091, memory: 49163, loss_cls_0: 0.8409, loss_box_0: 1.7802, loss_cns_0: 0.6178, loss_yns_0: 0.1607, loss_cls_1: 0.9010, loss_box_1: 1.7730, loss_cns_1: 0.6476, loss_yns_1: 0.1596, loss_cls_2: 0.9394, loss_box_2: 1.7177, loss_cns_2: 0.6546, loss_yns_2: 0.1610, loss_cls_3: 0.9390, loss_box_3: 1.6981, loss_cns_3: 0.6573, loss_yns_3: 0.1611, loss_cls_4: 0.9548, loss_box_4: 1.6922, loss_cns_4: 0.6565, loss_yns_4: 0.1607, loss_cls_5: 0.9477, loss_box_5: 1.7112, loss_cns_5: 0.6561, loss_yns_5: 0.1607, loss_cls_dn_0: 0.2165, loss_box_dn_0: 0.7692, loss_cls_dn_1: 0.1345, loss_box_dn_1: 0.7459, loss_cls_dn_2: 0.1325, loss_box_dn_2: 0.7175, loss_cls_dn_3: 0.1354, loss_box_dn_3: 0.7146, loss_cls_dn_4: 0.1374, loss_box_dn_4: 0.7123, loss_cls_dn_5: 0.1450, loss_box_dn_5: 0.7253, loss_dense_depth: 0.7515, loss: 26.7867, grad_norm: 32.6227 -2025-11-13 15:38:03,657 - mmdet - INFO - Iter [166/17500] lr: 1.660e-04, eta: 10:34:34, time: 1.502, data_time: 0.074, memory: 49163, loss_cls_0: 0.8226, loss_box_0: 1.7822, loss_cns_0: 0.6131, loss_yns_0: 0.1586, loss_cls_1: 0.9023, loss_box_1: 1.7230, loss_cns_1: 0.6522, loss_yns_1: 0.1602, loss_cls_2: 0.9234, loss_box_2: 1.6857, loss_cns_2: 0.6565, loss_yns_2: 0.1608, loss_cls_3: 0.9378, loss_box_3: 1.6736, loss_cns_3: 0.6592, loss_yns_3: 0.1615, loss_cls_4: 0.9450, loss_box_4: 1.6672, loss_cns_4: 0.6581, loss_yns_4: 0.1625, loss_cls_5: 0.9361, loss_box_5: 1.6855, loss_cns_5: 0.6578, loss_yns_5: 0.1606, loss_cls_dn_0: 0.2101, loss_box_dn_0: 0.7653, loss_cls_dn_1: 0.1325, loss_box_dn_1: 0.7299, loss_cls_dn_2: 0.1321, loss_box_dn_2: 0.7124, loss_cls_dn_3: 0.1409, loss_box_dn_3: 0.7157, loss_cls_dn_4: 0.1399, loss_box_dn_4: 0.7129, loss_cls_dn_5: 0.1466, loss_box_dn_5: 0.7233, loss_dense_depth: 0.7366, loss: 26.5440, grad_norm: 32.1477 -2025-11-13 15:38:10,360 - mmdet - INFO - Iter [167/17500] lr: 1.664e-04, eta: 10:42:20, time: 6.704, data_time: 0.076, memory: 49163, loss_cls_0: 0.8665, loss_box_0: 1.7733, loss_cns_0: 0.6099, loss_yns_0: 0.1602, loss_cls_1: 0.9298, loss_box_1: 1.7324, loss_cns_1: 0.6480, loss_yns_1: 0.1617, loss_cls_2: 0.9465, loss_box_2: 1.6863, loss_cns_2: 0.6536, loss_yns_2: 0.1636, loss_cls_3: 0.9638, loss_box_3: 1.6826, loss_cns_3: 0.6545, loss_yns_3: 0.1642, loss_cls_4: 0.9709, loss_box_4: 1.6845, loss_cns_4: 0.6529, loss_yns_4: 0.1629, loss_cls_5: 0.9626, loss_box_5: 1.6908, loss_cns_5: 0.6509, loss_yns_5: 0.1610, loss_cls_dn_0: 0.2194, loss_box_dn_0: 0.7719, loss_cls_dn_1: 0.1380, loss_box_dn_1: 0.7389, loss_cls_dn_2: 0.1396, loss_box_dn_2: 0.7226, loss_cls_dn_3: 0.1437, loss_box_dn_3: 0.7296, loss_cls_dn_4: 0.1466, loss_box_dn_4: 0.7342, loss_cls_dn_5: 0.1524, loss_box_dn_5: 0.7399, loss_dense_depth: 0.7769, loss: 26.8873, grad_norm: 34.9921 -2025-11-13 15:38:11,826 - mmdet - INFO - Iter [168/17500] lr: 1.668e-04, eta: 10:40:59, time: 1.466, data_time: 0.072, memory: 49163, loss_cls_0: 0.8541, loss_box_0: 1.7328, loss_cns_0: 0.6126, loss_yns_0: 0.1605, loss_cls_1: 0.9137, loss_box_1: 1.7183, loss_cns_1: 0.6493, loss_yns_1: 0.1607, loss_cls_2: 0.9392, loss_box_2: 1.6744, loss_cns_2: 0.6549, loss_yns_2: 0.1629, loss_cls_3: 0.9514, loss_box_3: 1.6518, loss_cns_3: 0.6559, loss_yns_3: 0.1630, loss_cls_4: 0.9636, loss_box_4: 1.6652, loss_cns_4: 0.6556, loss_yns_4: 0.1618, loss_cls_5: 0.9539, loss_box_5: 1.6677, loss_cns_5: 0.6544, loss_yns_5: 0.1614, loss_cls_dn_0: 0.2143, loss_box_dn_0: 0.7635, loss_cls_dn_1: 0.1377, loss_box_dn_1: 0.7358, loss_cls_dn_2: 0.1375, loss_box_dn_2: 0.7210, loss_cls_dn_3: 0.1395, loss_box_dn_3: 0.7247, loss_cls_dn_4: 0.1461, loss_box_dn_4: 0.7359, loss_cls_dn_5: 0.1529, loss_box_dn_5: 0.7471, loss_dense_depth: 0.7491, loss: 26.6444, grad_norm: 28.8601 -2025-11-13 15:38:13,308 - mmdet - INFO - Iter [169/17500] lr: 1.672e-04, eta: 10:39:41, time: 1.480, data_time: 0.074, memory: 49163, loss_cls_0: 0.8868, loss_box_0: 1.7646, loss_cns_0: 0.6101, loss_yns_0: 0.1631, loss_cls_1: 0.9427, loss_box_1: 1.7206, loss_cns_1: 0.6372, loss_yns_1: 0.1603, loss_cls_2: 0.9590, loss_box_2: 1.6880, loss_cns_2: 0.6493, loss_yns_2: 0.1626, loss_cls_3: 0.9808, loss_box_3: 1.6694, loss_cns_3: 0.6517, loss_yns_3: 0.1618, loss_cls_4: 0.9750, loss_box_4: 1.6849, loss_cns_4: 0.6525, loss_yns_4: 0.1642, loss_cls_5: 0.9718, loss_box_5: 1.6933, loss_cns_5: 0.6496, loss_yns_5: 0.1643, loss_cls_dn_0: 0.2229, loss_box_dn_0: 0.7699, loss_cls_dn_1: 0.1379, loss_box_dn_1: 0.7505, loss_cls_dn_2: 0.1336, loss_box_dn_2: 0.7379, loss_cls_dn_3: 0.1367, loss_box_dn_3: 0.7418, loss_cls_dn_4: 0.1420, loss_box_dn_4: 0.7562, loss_cls_dn_5: 0.1493, loss_box_dn_5: 0.7708, loss_dense_depth: 0.8114, loss: 27.0245, grad_norm: 38.2000 -2025-11-13 15:38:14,803 - mmdet - INFO - Iter [170/17500] lr: 1.676e-04, eta: 10:38:26, time: 1.496, data_time: 0.077, memory: 49163, loss_cls_0: 0.8702, loss_box_0: 1.7471, loss_cns_0: 0.6210, loss_yns_0: 0.1622, loss_cls_1: 0.9285, loss_box_1: 1.6945, loss_cns_1: 0.6446, loss_yns_1: 0.1591, loss_cls_2: 0.9636, loss_box_2: 1.6490, loss_cns_2: 0.6563, loss_yns_2: 0.1616, loss_cls_3: 0.9674, loss_box_3: 1.6390, loss_cns_3: 0.6600, loss_yns_3: 0.1625, loss_cls_4: 0.9866, loss_box_4: 1.6380, loss_cns_4: 0.6584, loss_yns_4: 0.1625, loss_cls_5: 0.9747, loss_box_5: 1.6458, loss_cns_5: 0.6566, loss_yns_5: 0.1634, loss_cls_dn_0: 0.2220, loss_box_dn_0: 0.7661, loss_cls_dn_1: 0.1388, loss_box_dn_1: 0.7407, loss_cls_dn_2: 0.1348, loss_box_dn_2: 0.7281, loss_cls_dn_3: 0.1381, loss_box_dn_3: 0.7325, loss_cls_dn_4: 0.1451, loss_box_dn_4: 0.7456, loss_cls_dn_5: 0.1503, loss_box_dn_5: 0.7597, loss_dense_depth: 0.7551, loss: 26.7292, grad_norm: 34.0476 -2025-11-13 15:38:16,292 - mmdet - INFO - Iter [171/17500] lr: 1.680e-04, eta: 10:37:10, time: 1.488, data_time: 0.076, memory: 49163, loss_cls_0: 0.8589, loss_box_0: 1.7349, loss_cns_0: 0.6167, loss_yns_0: 0.1593, loss_cls_1: 0.9239, loss_box_1: 1.7025, loss_cns_1: 0.6464, loss_yns_1: 0.1584, loss_cls_2: 0.9600, loss_box_2: 1.6640, loss_cns_2: 0.6546, loss_yns_2: 0.1603, loss_cls_3: 0.9637, loss_box_3: 1.6654, loss_cns_3: 0.6534, loss_yns_3: 0.1604, loss_cls_4: 0.9711, loss_box_4: 1.6646, loss_cns_4: 0.6539, loss_yns_4: 0.1600, loss_cls_5: 0.9621, loss_box_5: 1.6605, loss_cns_5: 0.6533, loss_yns_5: 0.1615, loss_cls_dn_0: 0.2187, loss_box_dn_0: 0.7671, loss_cls_dn_1: 0.1368, loss_box_dn_1: 0.7432, loss_cls_dn_2: 0.1361, loss_box_dn_2: 0.7285, loss_cls_dn_3: 0.1378, loss_box_dn_3: 0.7328, loss_cls_dn_4: 0.1460, loss_box_dn_4: 0.7454, loss_cls_dn_5: 0.1528, loss_box_dn_5: 0.7522, loss_dense_depth: 0.7261, loss: 26.6932, grad_norm: 41.4219 -2025-11-13 15:38:17,779 - mmdet - INFO - Iter [172/17500] lr: 1.684e-04, eta: 10:35:56, time: 1.486, data_time: 0.075, memory: 49163, loss_cls_0: 0.8786, loss_box_0: 1.7674, loss_cns_0: 0.6127, loss_yns_0: 0.1603, loss_cls_1: 0.9222, loss_box_1: 1.7288, loss_cns_1: 0.6425, loss_yns_1: 0.1594, loss_cls_2: 0.9605, loss_box_2: 1.6803, loss_cns_2: 0.6511, loss_yns_2: 0.1604, loss_cls_3: 0.9660, loss_box_3: 1.6765, loss_cns_3: 0.6539, loss_yns_3: 0.1615, loss_cls_4: 0.9851, loss_box_4: 1.6699, loss_cns_4: 0.6501, loss_yns_4: 0.1597, loss_cls_5: 0.9710, loss_box_5: 1.6768, loss_cns_5: 0.6508, loss_yns_5: 0.1612, loss_cls_dn_0: 0.2237, loss_box_dn_0: 0.7614, loss_cls_dn_1: 0.1353, loss_box_dn_1: 0.7469, loss_cls_dn_2: 0.1328, loss_box_dn_2: 0.7275, loss_cls_dn_3: 0.1335, loss_box_dn_3: 0.7270, loss_cls_dn_4: 0.1410, loss_box_dn_4: 0.7358, loss_cls_dn_5: 0.1498, loss_box_dn_5: 0.7368, loss_dense_depth: 0.7433, loss: 26.8016, grad_norm: 37.3451 -2025-11-13 15:38:19,263 - mmdet - INFO - Iter [173/17500] lr: 1.688e-04, eta: 10:34:42, time: 1.486, data_time: 0.079, memory: 49163, loss_cls_0: 0.8611, loss_box_0: 1.7575, loss_cns_0: 0.6199, loss_yns_0: 0.1614, loss_cls_1: 0.9151, loss_box_1: 1.7367, loss_cns_1: 0.6445, loss_yns_1: 0.1606, loss_cls_2: 0.9549, loss_box_2: 1.6894, loss_cns_2: 0.6538, loss_yns_2: 0.1595, loss_cls_3: 0.9592, loss_box_3: 1.6790, loss_cns_3: 0.6547, loss_yns_3: 0.1595, loss_cls_4: 0.9841, loss_box_4: 1.6706, loss_cns_4: 0.6523, loss_yns_4: 0.1591, loss_cls_5: 0.9572, loss_box_5: 1.7092, loss_cns_5: 0.6524, loss_yns_5: 0.1597, loss_cls_dn_0: 0.2232, loss_box_dn_0: 0.7633, loss_cls_dn_1: 0.1367, loss_box_dn_1: 0.7295, loss_cls_dn_2: 0.1356, loss_box_dn_2: 0.7122, loss_cls_dn_3: 0.1341, loss_box_dn_3: 0.7103, loss_cls_dn_4: 0.1420, loss_box_dn_4: 0.7116, loss_cls_dn_5: 0.1458, loss_box_dn_5: 0.7226, loss_dense_depth: 0.7416, loss: 26.7195, grad_norm: 39.6440 -2025-11-13 15:38:20,755 - mmdet - INFO - Iter [174/17500] lr: 1.692e-04, eta: 10:33:29, time: 1.491, data_time: 0.076, memory: 49163, loss_cls_0: 0.8576, loss_box_0: 1.7725, loss_cns_0: 0.6151, loss_yns_0: 0.1588, loss_cls_1: 0.9313, loss_box_1: 1.7410, loss_cns_1: 0.6455, loss_yns_1: 0.1590, loss_cls_2: 0.9383, loss_box_2: 1.7050, loss_cns_2: 0.6492, loss_yns_2: 0.1574, loss_cls_3: 0.9564, loss_box_3: 1.6993, loss_cns_3: 0.6515, loss_yns_3: 0.1578, loss_cls_4: 0.9568, loss_box_4: 1.6891, loss_cns_4: 0.6502, loss_yns_4: 0.1590, loss_cls_5: 0.9556, loss_box_5: 1.7147, loss_cns_5: 0.6481, loss_yns_5: 0.1598, loss_cls_dn_0: 0.2185, loss_box_dn_0: 0.7719, loss_cls_dn_1: 0.1382, loss_box_dn_1: 0.7315, loss_cls_dn_2: 0.1380, loss_box_dn_2: 0.7158, loss_cls_dn_3: 0.1375, loss_box_dn_3: 0.7206, loss_cls_dn_4: 0.1455, loss_box_dn_4: 0.7209, loss_cls_dn_5: 0.1507, loss_box_dn_5: 0.7345, loss_dense_depth: 0.7482, loss: 26.8005, grad_norm: 39.0259 -2025-11-13 15:38:22,242 - mmdet - INFO - Iter [175/17500] lr: 1.696e-04, eta: 10:32:17, time: 1.486, data_time: 0.085, memory: 49163, loss_cls_0: 0.8558, loss_box_0: 1.7449, loss_cns_0: 0.6191, loss_yns_0: 0.1594, loss_cls_1: 0.9232, loss_box_1: 1.7100, loss_cns_1: 0.6481, loss_yns_1: 0.1578, loss_cls_2: 0.9462, loss_box_2: 1.6793, loss_cns_2: 0.6518, loss_yns_2: 0.1582, loss_cls_3: 0.9612, loss_box_3: 1.6768, loss_cns_3: 0.6536, loss_yns_3: 0.1588, loss_cls_4: 0.9603, loss_box_4: 1.6745, loss_cns_4: 0.6530, loss_yns_4: 0.1577, loss_cls_5: 0.9537, loss_box_5: 1.6991, loss_cns_5: 0.6516, loss_yns_5: 0.1611, loss_cls_dn_0: 0.2212, loss_box_dn_0: 0.7637, loss_cls_dn_1: 0.1361, loss_box_dn_1: 0.7368, loss_cls_dn_2: 0.1322, loss_box_dn_2: 0.7269, loss_cls_dn_3: 0.1322, loss_box_dn_3: 0.7378, loss_cls_dn_4: 0.1404, loss_box_dn_4: 0.7488, loss_cls_dn_5: 0.1491, loss_box_dn_5: 0.7629, loss_dense_depth: 0.7558, loss: 26.7590, grad_norm: 46.0513 -2025-11-13 15:38:23,739 - mmdet - INFO - Iter [176/17500] lr: 1.700e-04, eta: 10:31:06, time: 1.497, data_time: 0.076, memory: 49163, loss_cls_0: 0.8529, loss_box_0: 1.7307, loss_cns_0: 0.6161, loss_yns_0: 0.1542, loss_cls_1: 0.9371, loss_box_1: 1.7120, loss_cns_1: 0.6458, loss_yns_1: 0.1538, loss_cls_2: 0.9486, loss_box_2: 1.6766, loss_cns_2: 0.6525, loss_yns_2: 0.1525, loss_cls_3: 0.9526, loss_box_3: 1.6706, loss_cns_3: 0.6524, loss_yns_3: 0.1535, loss_cls_4: 0.9609, loss_box_4: 1.6763, loss_cns_4: 0.6539, loss_yns_4: 0.1533, loss_cls_5: 0.9540, loss_box_5: 1.6853, loss_cns_5: 0.6511, loss_yns_5: 0.1539, loss_cls_dn_0: 0.2236, loss_box_dn_0: 0.7718, loss_cls_dn_1: 0.1392, loss_box_dn_1: 0.7386, loss_cls_dn_2: 0.1344, loss_box_dn_2: 0.7300, loss_cls_dn_3: 0.1341, loss_box_dn_3: 0.7389, loss_cls_dn_4: 0.1409, loss_box_dn_4: 0.7510, loss_cls_dn_5: 0.1496, loss_box_dn_5: 0.7633, loss_dense_depth: 0.7808, loss: 26.7469, grad_norm: 38.0834 -2025-11-13 15:38:25,230 - mmdet - INFO - Iter [177/17500] lr: 1.704e-04, eta: 10:29:56, time: 1.491, data_time: 0.075, memory: 49163, loss_cls_0: 0.8469, loss_box_0: 1.7167, loss_cns_0: 0.6154, loss_yns_0: 0.1521, loss_cls_1: 0.9342, loss_box_1: 1.6743, loss_cns_1: 0.6468, loss_yns_1: 0.1526, loss_cls_2: 0.9426, loss_box_2: 1.6582, loss_cns_2: 0.6546, loss_yns_2: 0.1501, loss_cls_3: 0.9491, loss_box_3: 1.6576, loss_cns_3: 0.6581, loss_yns_3: 0.1513, loss_cls_4: 0.9577, loss_box_4: 1.6590, loss_cns_4: 0.6553, loss_yns_4: 0.1517, loss_cls_5: 0.9453, loss_box_5: 1.6668, loss_cns_5: 0.6526, loss_yns_5: 0.1512, loss_cls_dn_0: 0.2235, loss_box_dn_0: 0.7642, loss_cls_dn_1: 0.1370, loss_box_dn_1: 0.7511, loss_cls_dn_2: 0.1349, loss_box_dn_2: 0.7378, loss_cls_dn_3: 0.1374, loss_box_dn_3: 0.7448, loss_cls_dn_4: 0.1405, loss_box_dn_4: 0.7520, loss_cls_dn_5: 0.1482, loss_box_dn_5: 0.7622, loss_dense_depth: 0.7566, loss: 26.5901, grad_norm: 39.2311 -2025-11-13 15:38:26,704 - mmdet - INFO - Iter [178/17500] lr: 1.708e-04, eta: 10:28:45, time: 1.473, data_time: 0.070, memory: 49163, loss_cls_0: 0.8406, loss_box_0: 1.7298, loss_cns_0: 0.6168, loss_yns_0: 0.1521, loss_cls_1: 0.9080, loss_box_1: 1.7055, loss_cns_1: 0.6469, loss_yns_1: 0.1514, loss_cls_2: 0.9282, loss_box_2: 1.6965, loss_cns_2: 0.6501, loss_yns_2: 0.1521, loss_cls_3: 0.9428, loss_box_3: 1.6910, loss_cns_3: 0.6570, loss_yns_3: 0.1520, loss_cls_4: 0.9674, loss_box_4: 1.6846, loss_cns_4: 0.6544, loss_yns_4: 0.1520, loss_cls_5: 0.9417, loss_box_5: 1.6933, loss_cns_5: 0.6553, loss_yns_5: 0.1521, loss_cls_dn_0: 0.2209, loss_box_dn_0: 0.7638, loss_cls_dn_1: 0.1377, loss_box_dn_1: 0.7514, loss_cls_dn_2: 0.1367, loss_box_dn_2: 0.7380, loss_cls_dn_3: 0.1393, loss_box_dn_3: 0.7440, loss_cls_dn_4: 0.1426, loss_box_dn_4: 0.7463, loss_cls_dn_5: 0.1488, loss_box_dn_5: 0.7548, loss_dense_depth: 0.7583, loss: 26.7041, grad_norm: 43.5950 -2025-11-13 15:38:28,180 - mmdet - INFO - Iter [179/17500] lr: 1.712e-04, eta: 10:27:35, time: 1.475, data_time: 0.071, memory: 49163, loss_cls_0: 0.8321, loss_box_0: 1.7260, loss_cns_0: 0.6169, loss_yns_0: 0.1525, loss_cls_1: 0.9038, loss_box_1: 1.7160, loss_cns_1: 0.6465, loss_yns_1: 0.1518, loss_cls_2: 0.9215, loss_box_2: 1.6812, loss_cns_2: 0.6500, loss_yns_2: 0.1515, loss_cls_3: 0.9355, loss_box_3: 1.6705, loss_cns_3: 0.6564, loss_yns_3: 0.1518, loss_cls_4: 0.9548, loss_box_4: 1.6600, loss_cns_4: 0.6561, loss_yns_4: 0.1523, loss_cls_5: 0.9402, loss_box_5: 1.6632, loss_cns_5: 0.6529, loss_yns_5: 0.1522, loss_cls_dn_0: 0.2201, loss_box_dn_0: 0.7688, loss_cls_dn_1: 0.1373, loss_box_dn_1: 0.7398, loss_cls_dn_2: 0.1364, loss_box_dn_2: 0.7214, loss_cls_dn_3: 0.1377, loss_box_dn_3: 0.7190, loss_cls_dn_4: 0.1401, loss_box_dn_4: 0.7184, loss_cls_dn_5: 0.1443, loss_box_dn_5: 0.7217, loss_dense_depth: 0.7465, loss: 26.4473, grad_norm: 27.2417 -2025-11-13 15:38:29,680 - mmdet - INFO - Iter [180/17500] lr: 1.716e-04, eta: 10:26:27, time: 1.495, data_time: 0.084, memory: 49163, loss_cls_0: 0.8327, loss_box_0: 1.6858, loss_cns_0: 0.6155, loss_yns_0: 0.1492, loss_cls_1: 0.8885, loss_box_1: 1.6819, loss_cns_1: 0.6465, loss_yns_1: 0.1504, loss_cls_2: 0.9291, loss_box_2: 1.6240, loss_cns_2: 0.6488, loss_yns_2: 0.1500, loss_cls_3: 0.9258, loss_box_3: 1.6245, loss_cns_3: 0.6549, loss_yns_3: 0.1502, loss_cls_4: 0.9362, loss_box_4: 1.6271, loss_cns_4: 0.6541, loss_yns_4: 0.1506, loss_cls_5: 0.9307, loss_box_5: 1.6316, loss_cns_5: 0.6531, loss_yns_5: 0.1505, loss_cls_dn_0: 0.2208, loss_box_dn_0: 0.7696, loss_cls_dn_1: 0.1341, loss_box_dn_1: 0.7337, loss_cls_dn_2: 0.1321, loss_box_dn_2: 0.7120, loss_cls_dn_3: 0.1330, loss_box_dn_3: 0.7054, loss_cls_dn_4: 0.1398, loss_box_dn_4: 0.7083, loss_cls_dn_5: 0.1427, loss_box_dn_5: 0.7103, loss_dense_depth: 0.7685, loss: 26.1024, grad_norm: 33.5476 -2025-11-13 15:38:31,242 - mmdet - INFO - Iter [181/17500] lr: 1.720e-04, eta: 10:25:28, time: 1.569, data_time: 0.072, memory: 49163, loss_cls_0: 0.8190, loss_box_0: 1.7352, loss_cns_0: 0.6145, loss_yns_0: 0.1512, loss_cls_1: 0.8784, loss_box_1: 1.7062, loss_cns_1: 0.6388, loss_yns_1: 0.1490, loss_cls_2: 0.9190, loss_box_2: 1.6527, loss_cns_2: 0.6477, loss_yns_2: 0.1496, loss_cls_3: 0.9253, loss_box_3: 1.6433, loss_cns_3: 0.6537, loss_yns_3: 0.1495, loss_cls_4: 0.9401, loss_box_4: 1.6319, loss_cns_4: 0.6533, loss_yns_4: 0.1502, loss_cls_5: 0.9299, loss_box_5: 1.6255, loss_cns_5: 0.6512, loss_yns_5: 0.1499, loss_cls_dn_0: 0.2200, loss_box_dn_0: 0.7738, loss_cls_dn_1: 0.1319, loss_box_dn_1: 0.7195, loss_cls_dn_2: 0.1317, loss_box_dn_2: 0.7018, loss_cls_dn_3: 0.1338, loss_box_dn_3: 0.7062, loss_cls_dn_4: 0.1425, loss_box_dn_4: 0.7111, loss_cls_dn_5: 0.1451, loss_box_dn_5: 0.7171, loss_dense_depth: 0.7940, loss: 26.1938, grad_norm: 38.4079 -2025-11-13 15:38:32,826 - mmdet - INFO - Iter [182/17500] lr: 1.724e-04, eta: 10:24:30, time: 1.586, data_time: 0.177, memory: 49163, loss_cls_0: 0.8215, loss_box_0: 1.7112, loss_cns_0: 0.6210, loss_yns_0: 0.1483, loss_cls_1: 0.8757, loss_box_1: 1.7216, loss_cns_1: 0.6436, loss_yns_1: 0.1483, loss_cls_2: 0.9166, loss_box_2: 1.6692, loss_cns_2: 0.6501, loss_yns_2: 0.1503, loss_cls_3: 0.9296, loss_box_3: 1.6516, loss_cns_3: 0.6554, loss_yns_3: 0.1488, loss_cls_4: 0.9257, loss_box_4: 1.6414, loss_cns_4: 0.6538, loss_yns_4: 0.1494, loss_cls_5: 0.9106, loss_box_5: 1.6482, loss_cns_5: 0.6516, loss_yns_5: 0.1490, loss_cls_dn_0: 0.2171, loss_box_dn_0: 0.7749, loss_cls_dn_1: 0.1333, loss_box_dn_1: 0.7292, loss_cls_dn_2: 0.1340, loss_box_dn_2: 0.7243, loss_cls_dn_3: 0.1383, loss_box_dn_3: 0.7349, loss_cls_dn_4: 0.1415, loss_box_dn_4: 0.7448, loss_cls_dn_5: 0.1447, loss_box_dn_5: 0.7644, loss_dense_depth: 0.7445, loss: 26.3182, grad_norm: 42.4381 diff --git a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20251113_153143.log.json b/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20251113_153143.log.json deleted file mode 100644 index 9b29f7f86e2043b940df7acc8c14070c5d8e2606..0000000000000000000000000000000000000000 --- a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20251113_153143.log.json +++ /dev/null @@ -1,183 +0,0 @@ -{"env_info": "sys.platform: linux\nPython: 3.10.12 (main, May 27 2025, 17:12:29) [GCC 11.4.0]\nCUDA available: True\nGPU 0,1,2,3,4,5,6,7: BW200, UBB BW1000\nCUDA_HOME: /opt/dtk\nNVCC: Not Available\nGCC: x86_64-linux-gnu-gcc (Ubuntu 11.4.0-1ubuntu1~22.04) 11.4.0\nPyTorch: 2.4.1\nPyTorch compiling details: PyTorch built with:\n - GCC 10.3\n - C++ Version: 201703\n - Intel(R) Math Kernel Library Version 2020.0.4 Product Build 20200917 for Intel(R) 64 architecture applications\n - OpenMP 201511 (a.k.a. OpenMP 4.5)\n - LAPACK is enabled (usually provided by MKL)\n - NNPACK is enabled\n - CPU capability usage: AVX512\n - HIP Runtime 6.3.25211\n - MIOpen 2.17.0\n - Magma 2.8.0\n - Build settings: BLAS_INFO=mkl, BUILD_TYPE=Release, CXX_COMPILER=/opt/rh/gcc-toolset-10/root/usr/bin/c++, CXX_FLAGS= -D_GLIBCXX_USE_CXX11_ABI=1 -fvisibility-inlines-hidden -DUSE_PTHREADPOOL -DNDEBUG -DUSE_KINETO -DLIBKINETO_NOCUPTI -DUSE_FBGEMM -DUSE_PYTORCH_QNNPACK -DUSE_XNNPACK -DSYMBOLICATE_MOBILE_DEBUG_HANDLE -O2 -fPIC -Wall -Wextra -Werror=return-type -Werror=non-virtual-dtor -Werror=bool-operation -Wnarrowing -Wno-missing-field-initializers -Wno-type-limits -Wno-array-bounds -Wno-unknown-pragmas -Wno-unused-parameter -Wno-unused-function -Wno-unused-result -Wno-strict-overflow -Wno-strict-aliasing -Wno-stringop-overflow -Wsuggest-override -Wno-psabi -Wno-error=pedantic -Wno-error=old-style-cast -Wno-missing-braces -fdiagnostics-color=always -faligned-new -Wno-unused-but-set-variable -Wno-maybe-uninitialized -fno-math-errno -fno-trapping-math -Werror=format -Wno-stringop-overflow, FORCE_FALLBACK_CUDA_MPI=1, LAPACK_INFO=mkl, PERF_WITH_AVX=1, PERF_WITH_AVX2=1, PERF_WITH_AVX512=1, TORCH_VERSION=2.4.1, USE_CUDA=0, USE_CUDNN=OFF, USE_CUSPARSELT=OFF, USE_EXCEPTION_PTR=1, USE_GFLAGS=1, USE_GLOG=1, USE_GLOO=1, USE_MKL=ON, USE_MKLDNN=0, USE_MPI=1, USE_NCCL=1, USE_NNPACK=ON, USE_OPENMP=1, USE_ROCM=ON, USE_ROCM_KERNEL_ASSERT=OFF, \n\nTorchVision: 0.19.1\nOpenCV: 4.11.0\nMMCV: 1.6.1\nMMCV Compiler: GCC 11.4\nMMCV CUDA Compiler: rocm not available\nMMDetection: 2.26.0+c41df4b", "config": "plugin = True\nplugin_dir = 'projects/mmdet3d_plugin/'\ndist_params = dict(backend='nccl')\nlog_level = 'INFO'\nwork_dir = './work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704'\ntotal_batch_size = 160\nnum_gpus = 8\nbatch_size = 20\nnum_iters_per_epoch = 175\nnum_epochs = 100\ncheckpoint_epoch_interval = 20\ncheckpoint_config = dict(interval=3500)\nlog_config = dict(\n interval=1,\n hooks=[\n dict(type='TextLoggerHook', by_epoch=False),\n dict(type='TensorboardLoggerHook')\n ])\nload_from = None\nresume_from = None\nworkflow = [('train', 1)]\nfp16 = dict(loss_scale=32.0)\ninput_shape = (704, 256)\ntracking_test = True\ntracking_threshold = 0.2\nclass_names = [\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier',\n 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n]\nnum_classes = 10\nembed_dims = 256\nnum_groups = 8\nnum_decoder = 6\nnum_single_frame_decoder = 1\nuse_deformable_func = True\nstrides = [4, 8, 16, 32]\nnum_levels = 4\nnum_depth_layers = 3\ndrop_out = 0.1\ntemporal = True\ndecouple_attn = True\nwith_quality_estimation = True\nmodel = dict(\n type='Sparse4D',\n use_grid_mask=True,\n use_deformable_func=True,\n img_backbone=dict(\n type='ResNet',\n depth=50,\n num_stages=4,\n frozen_stages=-1,\n norm_eval=False,\n style='pytorch',\n with_cp=True,\n out_indices=(0, 1, 2, 3),\n norm_cfg=dict(type='BN', requires_grad=True),\n pretrained='ckpt/resnet50-19c8e357.pth'),\n img_neck=dict(\n type='FPN',\n num_outs=4,\n start_level=0,\n out_channels=256,\n add_extra_convs='on_output',\n relu_before_extra_convs=True,\n in_channels=[256, 512, 1024, 2048]),\n depth_branch=dict(\n type='DenseDepthNet',\n embed_dims=256,\n num_depth_layers=3,\n loss_weight=0.2),\n head=dict(\n type='Sparse4DHead',\n cls_threshold_to_reg=0.05,\n decouple_attn=True,\n instance_bank=dict(\n type='InstanceBank',\n num_anchor=900,\n embed_dims=256,\n anchor='nuscenes_kmeans900.npy',\n anchor_handler=dict(type='SparseBox3DKeyPointsGenerator'),\n num_temp_instances=600,\n confidence_decay=0.6,\n feat_grad=False),\n anchor_encoder=dict(\n type='SparseBox3DEncoder',\n vel_dims=3,\n embed_dims=[128, 32, 32, 64],\n mode='cat',\n output_fc=False,\n in_loops=1,\n out_loops=4),\n num_single_frame_decoder=1,\n operation_order=[\n 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm',\n 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm',\n 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm',\n 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm',\n 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm',\n 'deformable', 'ffn', 'norm', 'refine'\n ],\n temp_graph_model=dict(\n type='MultiheadAttention',\n embed_dims=512,\n num_heads=8,\n batch_first=True,\n dropout=0.1),\n graph_model=dict(\n type='MultiheadAttention',\n embed_dims=512,\n num_heads=8,\n batch_first=True,\n dropout=0.1),\n norm_layer=dict(type='LN', normalized_shape=256),\n ffn=dict(\n type='AsymmetricFFN',\n in_channels=512,\n pre_norm=dict(type='LN'),\n embed_dims=256,\n feedforward_channels=1024,\n num_fcs=2,\n ffn_drop=0.1,\n act_cfg=dict(type='ReLU', inplace=True)),\n deformable_model=dict(\n type='DeformableFeatureAggregation',\n embed_dims=256,\n num_groups=8,\n num_levels=4,\n num_cams=6,\n attn_drop=0.15,\n use_deformable_func=True,\n use_camera_embed=True,\n residual_mode='cat',\n kps_generator=dict(\n type='SparseBox3DKeyPointsGenerator',\n num_learnable_pts=6,\n fix_scale=[[0, 0, 0], [0.45, 0, 0], [-0.45, 0, 0],\n [0, 0.45, 0], [0, -0.45, 0], [0, 0, 0.45],\n [0, 0, -0.45]])),\n refine_layer=dict(\n type='SparseBox3DRefinementModule',\n embed_dims=256,\n num_cls=10,\n refine_yaw=True,\n with_quality_estimation=True),\n sampler=dict(\n type='SparseBox3DTarget',\n num_dn_groups=5,\n num_temp_dn_groups=3,\n dn_noise_scale=[2.0, 2.0, 2.0, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5],\n max_dn_gt=32,\n add_neg_dn=True,\n cls_weight=2.0,\n box_weight=0.25,\n reg_weights=[2.0, 2.0, 2.0, 0.5, 0.5, 0.5, 0.0, 0.0, 0.0, 0.0],\n cls_wise_reg_weights=dict(\n {9: [2.0, 2.0, 2.0, 1.0, 1.0, 1.0, 0.0, 0.0, 1.0, 1.0]})),\n loss_cls=dict(\n type='FocalLoss',\n use_sigmoid=True,\n gamma=2.0,\n alpha=0.25,\n loss_weight=2.0),\n loss_reg=dict(\n type='SparseBox3DLoss',\n loss_box=dict(type='L1Loss', loss_weight=0.25),\n loss_centerness=dict(type='CrossEntropyLoss', use_sigmoid=True),\n loss_yawness=dict(type='GaussianFocalLoss'),\n cls_allow_reverse=[5]),\n decoder=dict(type='SparseBox3DDecoder'),\n reg_weights=[2.0, 2.0, 2.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0]))\ndataset_type = 'NuScenes3DDetTrackDataset'\ndata_root = 'data/nuscenes/'\nanno_root = 'data/nuscenes_anno_pkls/'\nfile_client_args = dict(backend='disk')\nimg_norm_cfg = dict(\n mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375], to_rgb=True)\ntrain_pipeline = [\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(\n type='LoadPointsFromFile',\n coord_type='LIDAR',\n load_dim=5,\n use_dim=5,\n file_client_args=dict(backend='disk')),\n dict(type='ResizeCropFlipImage'),\n dict(type='MultiScaleDepthMapGenerator', downsample=[4, 8, 16]),\n dict(type='BBoxRotation'),\n dict(type='PhotoMetricDistortionMultiViewImage'),\n dict(\n type='NormalizeMultiviewImage',\n mean=[123.675, 116.28, 103.53],\n std=[58.395, 57.12, 57.375],\n to_rgb=True),\n dict(\n type='CircleObjectRangeFilter',\n class_dist_thred=[55, 55, 55, 55, 55, 55, 55, 55, 55, 55]),\n dict(\n type='InstanceNameFilter',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ]),\n dict(type='NuScenesSparse4DAdaptor'),\n dict(\n type='Collect',\n keys=[\n 'img', 'timestamp', 'projection_mat', 'image_wh', 'gt_depth',\n 'focal', 'gt_bboxes_3d', 'gt_labels_3d'\n ],\n meta_keys=['T_global', 'T_global_inv', 'timestamp', 'instance_id'])\n]\ntest_pipeline = [\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(type='ResizeCropFlipImage'),\n dict(\n type='NormalizeMultiviewImage',\n mean=[123.675, 116.28, 103.53],\n std=[58.395, 57.12, 57.375],\n to_rgb=True),\n dict(type='NuScenesSparse4DAdaptor'),\n dict(\n type='Collect',\n keys=['img', 'timestamp', 'projection_mat', 'image_wh'],\n meta_keys=['T_global', 'T_global_inv', 'timestamp'])\n]\ninput_modality = dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False)\ndata_basic_config = dict(\n type='NuScenes3DDetTrackDataset',\n data_root='data/nuscenes/',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier',\n 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ],\n modality=dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False),\n version='v1.0-trainval')\ndata_aug_conf = dict(\n resize_lim=(0.4, 0.47),\n final_dim=(256, 704),\n bot_pct_lim=(0.0, 0.0),\n rot_lim=(-5.4, 5.4),\n H=900,\n W=1600,\n rand_flip=True,\n rot3d_range=[-0.3925, 0.3925])\ndata = dict(\n samples_per_gpu=20,\n workers_per_gpu=20,\n train=dict(\n type='NuScenes3DDetTrackDataset',\n data_root='data/nuscenes/',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ],\n modality=dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False),\n version='v1.0-trainval',\n ann_file='data/nuscenes_anno_pkls/nuscenes_infos_train.pkl',\n pipeline=[\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(\n type='LoadPointsFromFile',\n coord_type='LIDAR',\n load_dim=5,\n use_dim=5,\n file_client_args=dict(backend='disk')),\n dict(type='ResizeCropFlipImage'),\n dict(type='MultiScaleDepthMapGenerator', downsample=[4, 8, 16]),\n dict(type='BBoxRotation'),\n dict(type='PhotoMetricDistortionMultiViewImage'),\n dict(\n type='NormalizeMultiviewImage',\n mean=[123.675, 116.28, 103.53],\n std=[58.395, 57.12, 57.375],\n to_rgb=True),\n dict(\n type='CircleObjectRangeFilter',\n class_dist_thred=[55, 55, 55, 55, 55, 55, 55, 55, 55, 55]),\n dict(\n type='InstanceNameFilter',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian',\n 'traffic_cone'\n ]),\n dict(type='NuScenesSparse4DAdaptor'),\n dict(\n type='Collect',\n keys=[\n 'img', 'timestamp', 'projection_mat', 'image_wh',\n 'gt_depth', 'focal', 'gt_bboxes_3d', 'gt_labels_3d'\n ],\n meta_keys=[\n 'T_global', 'T_global_inv', 'timestamp', 'instance_id'\n ])\n ],\n test_mode=False,\n data_aug_conf=dict(\n resize_lim=(0.4, 0.47),\n final_dim=(256, 704),\n bot_pct_lim=(0.0, 0.0),\n rot_lim=(-5.4, 5.4),\n H=900,\n W=1600,\n rand_flip=True,\n rot3d_range=[-0.3925, 0.3925]),\n with_seq_flag=True,\n sequences_split_num=2,\n keep_consistent_seq_aug=True),\n val=dict(\n type='NuScenes3DDetTrackDataset',\n data_root='data/nuscenes/',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ],\n modality=dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False),\n version='v1.0-trainval',\n ann_file='data/nuscenes_anno_pkls/nuscenes_infos_val.pkl',\n pipeline=[\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(type='ResizeCropFlipImage'),\n dict(\n type='NormalizeMultiviewImage',\n mean=[123.675, 116.28, 103.53],\n std=[58.395, 57.12, 57.375],\n to_rgb=True),\n dict(type='NuScenesSparse4DAdaptor'),\n dict(\n type='Collect',\n keys=['img', 'timestamp', 'projection_mat', 'image_wh'],\n meta_keys=['T_global', 'T_global_inv', 'timestamp'])\n ],\n data_aug_conf=dict(\n resize_lim=(0.4, 0.47),\n final_dim=(256, 704),\n bot_pct_lim=(0.0, 0.0),\n rot_lim=(-5.4, 5.4),\n H=900,\n W=1600,\n rand_flip=True,\n rot3d_range=[-0.3925, 0.3925]),\n test_mode=True,\n tracking=True,\n tracking_threshold=0.2),\n test=dict(\n type='NuScenes3DDetTrackDataset',\n data_root='data/nuscenes/',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ],\n modality=dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False),\n version='v1.0-trainval',\n ann_file='data/nuscenes_anno_pkls/nuscenes_infos_val.pkl',\n pipeline=[\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(type='ResizeCropFlipImage'),\n dict(\n type='NormalizeMultiviewImage',\n mean=[123.675, 116.28, 103.53],\n std=[58.395, 57.12, 57.375],\n to_rgb=True),\n dict(type='NuScenesSparse4DAdaptor'),\n dict(\n type='Collect',\n keys=['img', 'timestamp', 'projection_mat', 'image_wh'],\n meta_keys=['T_global', 'T_global_inv', 'timestamp'])\n ],\n data_aug_conf=dict(\n resize_lim=(0.4, 0.47),\n final_dim=(256, 704),\n bot_pct_lim=(0.0, 0.0),\n rot_lim=(-5.4, 5.4),\n H=900,\n W=1600,\n rand_flip=True,\n rot3d_range=[-0.3925, 0.3925]),\n test_mode=True,\n tracking=True,\n tracking_threshold=0.2))\noptimizer = dict(\n type='Miopen_AdamW',\n lr=0.0006,\n weight_decay=0.001,\n paramwise_cfg=dict(custom_keys=dict(img_backbone=dict(lr_mult=0.5))))\noptimizer_config = dict(grad_clip=dict(max_norm=25, norm_type=2))\nlr_config = dict(\n policy='CosineAnnealing',\n warmup='linear',\n warmup_iters=500,\n warmup_ratio=0.3333333333333333,\n min_lr_ratio=0.001)\nrunner = dict(type='IterBasedRunner', max_iters=17500)\nvis_pipeline = [\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(type='Collect', keys=['img'], meta_keys=['timestamp', 'lidar2img'])\n]\nevaluation = dict(\n interval=3500,\n pipeline=[\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(\n type='Collect', keys=['img'], meta_keys=['timestamp', 'lidar2img'])\n ])\ngpu_ids = range(0, 8)\n", "seed": 0, "exp_name": "sparse4dv3_temporal_r50_1x8_bs6_256x704.py"} -{"mode": "train", "epoch": 1, "iter": 1, "lr": 0.0001, "memory": 49163, "data_time": 10.99523, "loss_cls_0": 2.361, "loss_box_0": 0.01384, "loss_cns_0": 0.0027, "loss_yns_0": 0.00079, "loss_cls_1": 2.15449, "loss_box_1": 0.10705, "loss_cns_1": 0.02433, "loss_yns_1": 0.00662, "loss_cls_2": 2.31203, "loss_box_2": 0.00504, "loss_cns_2": 0.00059, "loss_yns_2": 0.00029, "loss_cls_3": 2.39, "loss_box_3": 0.02945, "loss_cns_3": 0.00504, "loss_yns_3": 0.00144, "loss_cls_4": 2.02821, "loss_box_4": 0.41867, "loss_cns_4": 0.05371, "loss_yns_4": 0.02521, "loss_cls_5": 2.42482, "loss_box_5": 0.01905, "loss_cns_5": 0.00239, "loss_yns_5": 0.00168, "loss_cls_dn_0": 1.19803, "loss_box_dn_0": 1.4603, "loss_cls_dn_1": 1.11018, "loss_box_dn_1": 1.73178, "loss_cls_dn_2": 1.17413, "loss_box_dn_2": 1.97185, "loss_cls_dn_3": 1.17206, "loss_box_dn_3": 2.2418, "loss_cls_dn_4": 1.05277, "loss_box_dn_4": 2.42681, "loss_cls_dn_5": 1.23868, "loss_box_dn_5": 2.67725, "loss_dense_depth": 1.86433, "loss": 35.70841, "grad_norm": 273.30502, "time": 115.55782} -{"mode": "train", "epoch": 1, "iter": 2, "lr": 0.0001, "memory": 49163, "data_time": 0.08011, "loss_cls_0": 2.05557, "loss_box_0": 0.02344, "loss_cns_0": 0.00596, "loss_yns_0": 0.00245, "loss_cls_1": 2.02404, "loss_box_1": 0.14399, "loss_cns_1": 0.02549, "loss_yns_1": 0.00767, "loss_cls_2": 2.10568, "loss_box_2": 0.21279, "loss_cns_2": 0.01973, "loss_yns_2": 0.00822, "loss_cls_3": 1.93649, "loss_box_3": 0.4639, "loss_cns_3": 0.06124, "loss_yns_3": 0.02103, "loss_cls_4": 1.79942, "loss_box_4": 1.55555, "loss_cns_4": 0.15454, "loss_yns_4": 0.05654, "loss_cls_5": 2.04857, "loss_box_5": 0.5749, "loss_cns_5": 0.06293, "loss_yns_5": 0.01963, "loss_cls_dn_0": 1.03136, "loss_box_dn_0": 1.25504, "loss_cls_dn_1": 0.95445, "loss_box_dn_1": 2.40876, "loss_cls_dn_2": 0.9733, "loss_box_dn_2": 2.53217, "loss_cls_dn_3": 0.90774, "loss_box_dn_3": 2.61067, "loss_cls_dn_4": 0.84198, "loss_box_dn_4": 2.8655, "loss_cls_dn_5": 0.98614, "loss_box_dn_5": 3.10555, "loss_dense_depth": 1.71169, "loss": 37.57413, "grad_norm": 66.66781, "time": 1.95929} -{"mode": "train", "epoch": 1, "iter": 3, "lr": 0.0001, "memory": 49163, "data_time": 0.08001, "loss_cls_0": 1.41337, "loss_box_0": 2.7913, "loss_cns_0": 0.62245, "loss_yns_0": 0.21489, "loss_cls_1": 1.72844, "loss_box_1": 2.48574, "loss_cns_1": 0.35136, "loss_yns_1": 0.13175, "loss_cls_2": 1.74986, "loss_box_2": 4.41202, "loss_cns_2": 0.37492, "loss_yns_2": 0.20711, "loss_cls_3": 1.59993, "loss_box_3": 5.21287, "loss_cns_3": 0.45577, "loss_yns_3": 0.21339, "loss_cls_4": 1.51431, "loss_box_4": 5.12831, "loss_cns_4": 0.42729, "loss_yns_4": 0.20632, "loss_cls_5": 1.62196, "loss_box_5": 3.99205, "loss_cns_5": 0.28733, "loss_yns_5": 0.12964, "loss_cls_dn_0": 0.67296, "loss_box_dn_0": 1.27709, "loss_cls_dn_1": 0.80583, "loss_box_dn_1": 2.3997, "loss_cls_dn_2": 0.76907, "loss_box_dn_2": 2.62599, "loss_cls_dn_3": 0.6697, "loss_box_dn_3": 2.81849, "loss_cls_dn_4": 0.68529, "loss_box_dn_4": 3.08067, "loss_cls_dn_5": 0.77532, "loss_box_dn_5": 3.36706, "loss_dense_depth": 1.64205, "loss": 58.86163, "grad_norm": 111.59391, "time": 1.51523} -{"mode": "train", "epoch": 1, "iter": 4, "lr": 0.0001, "memory": 49163, "data_time": 0.07976, "loss_cls_0": 1.39761, "loss_box_0": 2.58616, "loss_cns_0": 0.54009, "loss_yns_0": 0.20843, "loss_cls_1": 1.55536, "loss_box_1": 3.70938, "loss_cns_1": 0.45454, "loss_yns_1": 0.22453, "loss_cls_2": 1.69743, "loss_box_2": 3.83377, "loss_cns_2": 0.44659, "loss_yns_2": 0.20849, "loss_cls_3": 1.46485, "loss_box_3": 4.27359, "loss_cns_3": 0.45895, "loss_yns_3": 0.19872, "loss_cls_4": 1.49057, "loss_box_4": 4.66574, "loss_cns_4": 0.41972, "loss_yns_4": 0.19515, "loss_cls_5": 1.40438, "loss_box_5": 5.18834, "loss_cns_5": 0.4363, "loss_yns_5": 0.2013, "loss_cls_dn_0": 0.52491, "loss_box_dn_0": 1.23227, "loss_cls_dn_1": 0.66367, "loss_box_dn_1": 2.59843, "loss_cls_dn_2": 0.64754, "loss_box_dn_2": 2.72115, "loss_cls_dn_3": 0.58279, "loss_box_dn_3": 2.98376, "loss_cls_dn_4": 0.54175, "loss_box_dn_4": 3.21468, "loss_cls_dn_5": 0.59557, "loss_box_dn_5": 3.4518, "loss_dense_depth": 1.73412, "loss": 58.75245, "grad_norm": 114.05312, "time": 1.50544} -{"mode": "train", "epoch": 1, "iter": 5, "lr": 0.0001, "memory": 49163, "data_time": 0.0811, "loss_cls_0": 1.37589, "loss_box_0": 2.76673, "loss_cns_0": 0.55798, "loss_yns_0": 0.19871, "loss_cls_1": 1.53024, "loss_box_1": 4.23032, "loss_cns_1": 0.40653, "loss_yns_1": 0.21781, "loss_cls_2": 1.45209, "loss_box_2": 4.22224, "loss_cns_2": 0.393, "loss_yns_2": 0.18958, "loss_cls_3": 1.37158, "loss_box_3": 4.30395, "loss_cns_3": 0.38184, "loss_yns_3": 0.21581, "loss_cls_4": 1.33471, "loss_box_4": 4.61286, "loss_cns_4": 0.34143, "loss_yns_4": 0.20946, "loss_cls_5": 1.35015, "loss_box_5": 4.79173, "loss_cns_5": 0.33112, "loss_yns_5": 0.21151, "loss_cls_dn_0": 0.55238, "loss_box_dn_0": 1.24726, "loss_cls_dn_1": 0.58303, "loss_box_dn_1": 2.44457, "loss_cls_dn_2": 0.61938, "loss_box_dn_2": 2.56427, "loss_cls_dn_3": 0.51789, "loss_box_dn_3": 2.6632, "loss_cls_dn_4": 0.53872, "loss_box_dn_4": 2.91462, "loss_cls_dn_5": 0.5024, "loss_box_dn_5": 2.95709, "loss_dense_depth": 1.90759, "loss": 57.00967, "grad_norm": 125.07663, "time": 1.53968} -{"mode": "train", "epoch": 1, "iter": 6, "lr": 0.0001, "memory": 49163, "data_time": 0.0792, "loss_cls_0": 1.25772, "loss_box_0": 2.36554, "loss_cns_0": 0.68373, "loss_yns_0": 0.18512, "loss_cls_1": 1.32783, "loss_box_1": 3.70182, "loss_cns_1": 0.47556, "loss_yns_1": 0.19296, "loss_cls_2": 1.34243, "loss_box_2": 3.72899, "loss_cns_2": 0.47922, "loss_yns_2": 0.18607, "loss_cls_3": 1.33681, "loss_box_3": 3.61195, "loss_cns_3": 0.50602, "loss_yns_3": 0.18669, "loss_cls_4": 1.29588, "loss_box_4": 3.95196, "loss_cns_4": 0.46231, "loss_yns_4": 0.18456, "loss_cls_5": 1.35656, "loss_box_5": 4.32283, "loss_cns_5": 0.43191, "loss_yns_5": 0.19517, "loss_cls_dn_0": 0.5596, "loss_box_dn_0": 1.13936, "loss_cls_dn_1": 0.51179, "loss_box_dn_1": 2.52599, "loss_cls_dn_2": 0.54304, "loss_box_dn_2": 2.47451, "loss_cls_dn_3": 0.48552, "loss_box_dn_3": 2.52558, "loss_cls_dn_4": 0.47651, "loss_box_dn_4": 2.76838, "loss_cls_dn_5": 0.43698, "loss_box_dn_5": 2.91915, "loss_dense_depth": 1.78593, "loss": 52.92196, "grad_norm": 115.38404, "time": 1.51327} -{"mode": "train", "epoch": 1, "iter": 7, "lr": 0.0001, "memory": 49163, "data_time": 0.07665, "loss_cls_0": 1.25304, "loss_box_0": 2.30834, "loss_cns_0": 0.63262, "loss_yns_0": 0.18365, "loss_cls_1": 1.26129, "loss_box_1": 3.71702, "loss_cns_1": 0.45459, "loss_yns_1": 0.18357, "loss_cls_2": 1.42482, "loss_box_2": 3.72196, "loss_cns_2": 0.40249, "loss_yns_2": 0.1827, "loss_cls_3": 1.27844, "loss_box_3": 3.60048, "loss_cns_3": 0.44953, "loss_yns_3": 0.1839, "loss_cls_4": 1.31726, "loss_box_4": 3.65012, "loss_cns_4": 0.45154, "loss_yns_4": 0.18934, "loss_cls_5": 1.33569, "loss_box_5": 3.86666, "loss_cns_5": 0.49316, "loss_yns_5": 0.18555, "loss_cls_dn_0": 0.52085, "loss_box_dn_0": 1.05388, "loss_cls_dn_1": 0.46322, "loss_box_dn_1": 2.44563, "loss_cls_dn_2": 0.46406, "loss_box_dn_2": 2.37405, "loss_cls_dn_3": 0.44671, "loss_box_dn_3": 2.37254, "loss_cls_dn_4": 0.42758, "loss_box_dn_4": 2.42579, "loss_cls_dn_5": 0.39582, "loss_box_dn_5": 2.51521, "loss_dense_depth": 1.93918, "loss": 50.57228, "grad_norm": 86.59176, "time": 1.4826} -{"mode": "train", "epoch": 1, "iter": 8, "lr": 0.0001, "memory": 49163, "data_time": 0.08194, "loss_cls_0": 1.26858, "loss_box_0": 2.29416, "loss_cns_0": 0.61437, "loss_yns_0": 0.17893, "loss_cls_1": 1.27697, "loss_box_1": 3.3836, "loss_cns_1": 0.50294, "loss_yns_1": 0.18034, "loss_cls_2": 1.3054, "loss_box_2": 3.36353, "loss_cns_2": 0.44274, "loss_yns_2": 0.18244, "loss_cls_3": 1.26959, "loss_box_3": 3.46251, "loss_cns_3": 0.47555, "loss_yns_3": 0.19552, "loss_cls_4": 1.24758, "loss_box_4": 3.40859, "loss_cns_4": 0.50214, "loss_yns_4": 0.18399, "loss_cls_5": 1.27911, "loss_box_5": 3.41339, "loss_cns_5": 0.55763, "loss_yns_5": 0.19429, "loss_cls_dn_0": 0.47543, "loss_box_dn_0": 1.04336, "loss_cls_dn_1": 0.46513, "loss_box_dn_1": 1.57428, "loss_cls_dn_2": 0.50005, "loss_box_dn_2": 1.62802, "loss_cls_dn_3": 0.4356, "loss_box_dn_3": 1.76906, "loss_cls_dn_4": 0.4521, "loss_box_dn_4": 1.76032, "loss_cls_dn_5": 0.45334, "loss_box_dn_5": 1.87182, "loss_dense_depth": 2.15843, "loss": 45.77086, "grad_norm": 85.33852, "time": 1.5106} -{"mode": "train", "epoch": 1, "iter": 9, "lr": 0.0001, "memory": 49163, "data_time": 0.07992, "loss_cls_0": 1.2341, "loss_box_0": 2.32115, "loss_cns_0": 0.62985, "loss_yns_0": 0.17755, "loss_cls_1": 1.26412, "loss_box_1": 3.06266, "loss_cns_1": 0.49476, "loss_yns_1": 0.19097, "loss_cls_2": 1.25248, "loss_box_2": 3.03302, "loss_cns_2": 0.50755, "loss_yns_2": 0.1849, "loss_cls_3": 1.25326, "loss_box_3": 3.18348, "loss_cns_3": 0.51013, "loss_yns_3": 0.18098, "loss_cls_4": 1.25435, "loss_box_4": 3.21687, "loss_cns_4": 0.49016, "loss_yns_4": 0.18969, "loss_cls_5": 1.25232, "loss_box_5": 3.77418, "loss_cns_5": 0.47372, "loss_yns_5": 0.18096, "loss_cls_dn_0": 0.48963, "loss_box_dn_0": 1.0517, "loss_cls_dn_1": 0.43031, "loss_box_dn_1": 1.56953, "loss_cls_dn_2": 0.48513, "loss_box_dn_2": 1.70218, "loss_cls_dn_3": 0.42974, "loss_box_dn_3": 1.86602, "loss_cls_dn_4": 0.43075, "loss_box_dn_4": 1.88681, "loss_cls_dn_5": 0.46292, "loss_box_dn_5": 2.29645, "loss_dense_depth": 2.15805, "loss": 45.57241, "grad_norm": 85.83323, "time": 1.51198} -{"mode": "train", "epoch": 1, "iter": 10, "lr": 0.0001, "memory": 49163, "data_time": 0.08102, "loss_cls_0": 1.24254, "loss_box_0": 2.36883, "loss_cns_0": 0.62268, "loss_yns_0": 0.17428, "loss_cls_1": 1.28221, "loss_box_1": 3.14702, "loss_cns_1": 0.46377, "loss_yns_1": 0.18112, "loss_cls_2": 1.22748, "loss_box_2": 3.09918, "loss_cns_2": 0.45245, "loss_yns_2": 0.18213, "loss_cls_3": 1.2109, "loss_box_3": 3.12101, "loss_cns_3": 0.46889, "loss_yns_3": 0.1737, "loss_cls_4": 1.25155, "loss_box_4": 3.16418, "loss_cns_4": 0.46385, "loss_yns_4": 0.1797, "loss_cls_5": 1.2616, "loss_box_5": 3.44331, "loss_cns_5": 0.47608, "loss_yns_5": 0.17491, "loss_cls_dn_0": 0.4841, "loss_box_dn_0": 1.06743, "loss_cls_dn_1": 0.40236, "loss_box_dn_1": 1.8894, "loss_cls_dn_2": 0.45082, "loss_box_dn_2": 1.859, "loss_cls_dn_3": 0.44157, "loss_box_dn_3": 1.89537, "loss_cls_dn_4": 0.40188, "loss_box_dn_4": 1.94031, "loss_cls_dn_5": 0.42974, "loss_box_dn_5": 2.15006, "loss_dense_depth": 2.45241, "loss": 45.69784, "grad_norm": 83.678, "time": 1.49827} -{"mode": "train", "epoch": 1, "iter": 11, "lr": 0.0001, "memory": 49163, "data_time": 0.08078, "loss_cls_0": 1.21264, "loss_box_0": 2.33153, "loss_cns_0": 0.60075, "loss_yns_0": 0.1747, "loss_cls_1": 1.25279, "loss_box_1": 2.95454, "loss_cns_1": 0.52165, "loss_yns_1": 0.18279, "loss_cls_2": 1.22983, "loss_box_2": 3.06941, "loss_cns_2": 0.50075, "loss_yns_2": 0.19039, "loss_cls_3": 1.23099, "loss_box_3": 3.08676, "loss_cns_3": 0.52036, "loss_yns_3": 0.17709, "loss_cls_4": 1.27381, "loss_box_4": 3.20612, "loss_cns_4": 0.45931, "loss_yns_4": 0.18116, "loss_cls_5": 1.29817, "loss_box_5": 3.53446, "loss_cns_5": 0.47015, "loss_yns_5": 0.17378, "loss_cls_dn_0": 0.45084, "loss_box_dn_0": 1.06973, "loss_cls_dn_1": 0.39406, "loss_box_dn_1": 1.97274, "loss_cls_dn_2": 0.42578, "loss_box_dn_2": 1.91979, "loss_cls_dn_3": 0.41939, "loss_box_dn_3": 1.91635, "loss_cls_dn_4": 0.39584, "loss_box_dn_4": 2.04096, "loss_cls_dn_5": 0.4064, "loss_box_dn_5": 2.12105, "loss_dense_depth": 2.22029, "loss": 45.58714, "grad_norm": 79.44703, "time": 1.48017} -{"mode": "train", "epoch": 1, "iter": 12, "lr": 0.0001, "memory": 49163, "data_time": 0.07967, "loss_cls_0": 1.24714, "loss_box_0": 2.32118, "loss_cns_0": 0.59076, "loss_yns_0": 0.17266, "loss_cls_1": 1.24254, "loss_box_1": 2.99935, "loss_cns_1": 0.52779, "loss_yns_1": 0.17258, "loss_cls_2": 1.24586, "loss_box_2": 3.24468, "loss_cns_2": 0.50936, "loss_yns_2": 0.18535, "loss_cls_3": 1.26434, "loss_box_3": 3.29681, "loss_cns_3": 0.49544, "loss_yns_3": 0.17458, "loss_cls_4": 1.22122, "loss_box_4": 3.47826, "loss_cns_4": 0.47061, "loss_yns_4": 0.18345, "loss_cls_5": 1.24404, "loss_box_5": 3.59851, "loss_cns_5": 0.42017, "loss_yns_5": 0.17581, "loss_cls_dn_0": 0.44595, "loss_box_dn_0": 1.07093, "loss_cls_dn_1": 0.39806, "loss_box_dn_1": 2.06335, "loss_cls_dn_2": 0.39356, "loss_box_dn_2": 2.06794, "loss_cls_dn_3": 0.37733, "loss_box_dn_3": 2.06279, "loss_cls_dn_4": 0.39531, "loss_box_dn_4": 2.18308, "loss_cls_dn_5": 0.40037, "loss_box_dn_5": 2.21451, "loss_dense_depth": 2.17527, "loss": 46.73092, "grad_norm": 86.83142, "time": 1.48945} -{"mode": "train", "epoch": 1, "iter": 13, "lr": 0.0001, "memory": 49163, "data_time": 0.07784, "loss_cls_0": 1.16638, "loss_box_0": 2.35873, "loss_cns_0": 0.55792, "loss_yns_0": 0.17146, "loss_cls_1": 1.22169, "loss_box_1": 3.05422, "loss_cns_1": 0.53056, "loss_yns_1": 0.17116, "loss_cls_2": 1.2571, "loss_box_2": 3.08285, "loss_cns_2": 0.53943, "loss_yns_2": 0.18897, "loss_cls_3": 1.25697, "loss_box_3": 3.0942, "loss_cns_3": 0.53706, "loss_yns_3": 0.17545, "loss_cls_4": 1.21835, "loss_box_4": 3.18775, "loss_cns_4": 0.53834, "loss_yns_4": 0.18004, "loss_cls_5": 1.22648, "loss_box_5": 3.22204, "loss_cns_5": 0.53161, "loss_yns_5": 0.17323, "loss_cls_dn_0": 0.47429, "loss_box_dn_0": 1.06908, "loss_cls_dn_1": 0.45197, "loss_box_dn_1": 1.69651, "loss_cls_dn_2": 0.42251, "loss_box_dn_2": 1.72197, "loss_cls_dn_3": 0.40547, "loss_box_dn_3": 1.76377, "loss_cls_dn_4": 0.43096, "loss_box_dn_4": 1.92585, "loss_cls_dn_5": 0.45362, "loss_box_dn_5": 1.96758, "loss_dense_depth": 2.37501, "loss": 44.80056, "grad_norm": 92.73308, "time": 1.49524} -{"mode": "train", "epoch": 1, "iter": 14, "lr": 0.00011, "memory": 49163, "data_time": 0.08035, "loss_cls_0": 1.16976, "loss_box_0": 2.46976, "loss_cns_0": 0.54187, "loss_yns_0": 0.17191, "loss_cls_1": 1.23229, "loss_box_1": 2.80535, "loss_cns_1": 0.55545, "loss_yns_1": 0.17395, "loss_cls_2": 1.28887, "loss_box_2": 2.86674, "loss_cns_2": 0.5374, "loss_yns_2": 0.18438, "loss_cls_3": 1.26383, "loss_box_3": 3.07467, "loss_cns_3": 0.55215, "loss_yns_3": 0.1749, "loss_cls_4": 1.27821, "loss_box_4": 2.95142, "loss_cns_4": 0.56012, "loss_yns_4": 0.18363, "loss_cls_5": 1.25097, "loss_box_5": 2.85635, "loss_cns_5": 0.57233, "loss_yns_5": 0.17517, "loss_cls_dn_0": 0.49217, "loss_box_dn_0": 1.08671, "loss_cls_dn_1": 0.46885, "loss_box_dn_1": 1.41873, "loss_cls_dn_2": 0.43737, "loss_box_dn_2": 1.43973, "loss_cls_dn_3": 0.43166, "loss_box_dn_3": 1.58612, "loss_cls_dn_4": 0.43285, "loss_box_dn_4": 1.61416, "loss_cls_dn_5": 0.46171, "loss_box_dn_5": 1.6727, "loss_dense_depth": 2.07318, "loss": 42.50741, "grad_norm": 80.29237, "time": 1.50477} -{"mode": "train", "epoch": 1, "iter": 15, "lr": 0.00011, "memory": 49163, "data_time": 0.07833, "loss_cls_0": 1.18137, "loss_box_0": 2.53904, "loss_cns_0": 0.56347, "loss_yns_0": 0.17186, "loss_cls_1": 1.24488, "loss_box_1": 2.55554, "loss_cns_1": 0.59571, "loss_yns_1": 0.17578, "loss_cls_2": 1.29867, "loss_box_2": 2.65734, "loss_cns_2": 0.58138, "loss_yns_2": 0.17645, "loss_cls_3": 1.26029, "loss_box_3": 2.80817, "loss_cns_3": 0.60964, "loss_yns_3": 0.17558, "loss_cls_4": 1.2715, "loss_box_4": 2.69017, "loss_cns_4": 0.61338, "loss_yns_4": 0.18555, "loss_cls_5": 1.27152, "loss_box_5": 2.68134, "loss_cns_5": 0.59751, "loss_yns_5": 0.17649, "loss_cls_dn_0": 0.46422, "loss_box_dn_0": 1.10894, "loss_cls_dn_1": 0.45809, "loss_box_dn_1": 1.39187, "loss_cls_dn_2": 0.42858, "loss_box_dn_2": 1.46991, "loss_cls_dn_3": 0.42534, "loss_box_dn_3": 1.61743, "loss_cls_dn_4": 0.44546, "loss_box_dn_4": 1.58134, "loss_cls_dn_5": 0.4402, "loss_box_dn_5": 1.70212, "loss_dense_depth": 2.04313, "loss": 41.65931, "grad_norm": 80.14674, "time": 1.4722} -{"mode": "train", "epoch": 1, "iter": 16, "lr": 0.00011, "memory": 49163, "data_time": 0.08044, "loss_cls_0": 1.20791, "loss_box_0": 2.42303, "loss_cns_0": 0.58778, "loss_yns_0": 0.17187, "loss_cls_1": 1.24011, "loss_box_1": 2.57296, "loss_cns_1": 0.58189, "loss_yns_1": 0.17745, "loss_cls_2": 1.26024, "loss_box_2": 2.6407, "loss_cns_2": 0.58202, "loss_yns_2": 0.18002, "loss_cls_3": 1.23674, "loss_box_3": 2.63117, "loss_cns_3": 0.58671, "loss_yns_3": 0.18059, "loss_cls_4": 1.24735, "loss_box_4": 2.65189, "loss_cns_4": 0.58404, "loss_yns_4": 0.17658, "loss_cls_5": 1.26081, "loss_box_5": 2.61671, "loss_cns_5": 0.57217, "loss_yns_5": 0.17595, "loss_cls_dn_0": 0.45241, "loss_box_dn_0": 1.0572, "loss_cls_dn_1": 0.45929, "loss_box_dn_1": 1.31504, "loss_cls_dn_2": 0.43297, "loss_box_dn_2": 1.393, "loss_cls_dn_3": 0.44798, "loss_box_dn_3": 1.50042, "loss_cls_dn_4": 0.47395, "loss_box_dn_4": 1.50752, "loss_cls_dn_5": 0.43712, "loss_box_dn_5": 1.61203, "loss_dense_depth": 1.94799, "loss": 40.58358, "grad_norm": 69.08453, "time": 1.48389} -{"mode": "train", "epoch": 1, "iter": 17, "lr": 0.00011, "memory": 49163, "data_time": 0.07944, "loss_cls_0": 1.16425, "loss_box_0": 2.26918, "loss_cns_0": 0.60245, "loss_yns_0": 0.17081, "loss_cls_1": 1.25175, "loss_box_1": 2.6355, "loss_cns_1": 0.58011, "loss_yns_1": 0.17539, "loss_cls_2": 1.26621, "loss_box_2": 2.67161, "loss_cns_2": 0.57329, "loss_yns_2": 0.18049, "loss_cls_3": 1.25322, "loss_box_3": 2.69217, "loss_cns_3": 0.588, "loss_yns_3": 0.19417, "loss_cls_4": 1.24886, "loss_box_4": 2.77604, "loss_cns_4": 0.57906, "loss_yns_4": 0.17604, "loss_cls_5": 1.25374, "loss_box_5": 2.8255, "loss_cns_5": 0.57513, "loss_yns_5": 0.17666, "loss_cls_dn_0": 0.45234, "loss_box_dn_0": 1.01598, "loss_cls_dn_1": 0.43617, "loss_box_dn_1": 1.34337, "loss_cls_dn_2": 0.41789, "loss_box_dn_2": 1.40046, "loss_cls_dn_3": 0.42507, "loss_box_dn_3": 1.48933, "loss_cls_dn_4": 0.43239, "loss_box_dn_4": 1.53605, "loss_cls_dn_5": 0.42425, "loss_box_dn_5": 1.63957, "loss_dense_depth": 1.89632, "loss": 40.78881, "grad_norm": 63.67016, "time": 1.49123} -{"mode": "train", "epoch": 1, "iter": 18, "lr": 0.00011, "memory": 49163, "data_time": 0.08041, "loss_cls_0": 1.13547, "loss_box_0": 2.1483, "loss_cns_0": 0.61615, "loss_yns_0": 0.17091, "loss_cls_1": 1.25684, "loss_box_1": 2.74788, "loss_cns_1": 0.57793, "loss_yns_1": 0.17799, "loss_cls_2": 1.24635, "loss_box_2": 2.76078, "loss_cns_2": 0.57132, "loss_yns_2": 0.17921, "loss_cls_3": 1.24335, "loss_box_3": 2.7937, "loss_cns_3": 0.60193, "loss_yns_3": 0.17685, "loss_cls_4": 1.27281, "loss_box_4": 2.84486, "loss_cns_4": 0.58541, "loss_yns_4": 0.17987, "loss_cls_5": 1.24188, "loss_box_5": 3.0438, "loss_cns_5": 0.58413, "loss_yns_5": 0.17529, "loss_cls_dn_0": 0.48484, "loss_box_dn_0": 0.99027, "loss_cls_dn_1": 0.41983, "loss_box_dn_1": 1.33324, "loss_cls_dn_2": 0.43432, "loss_box_dn_2": 1.3678, "loss_cls_dn_3": 0.42866, "loss_box_dn_3": 1.452, "loss_cls_dn_4": 0.40914, "loss_box_dn_4": 1.51821, "loss_cls_dn_5": 0.42408, "loss_box_dn_5": 1.67449, "loss_dense_depth": 1.66108, "loss": 40.93093, "grad_norm": 81.73623, "time": 1.49116} -{"mode": "train", "epoch": 1, "iter": 19, "lr": 0.00011, "memory": 49163, "data_time": 0.07906, "loss_cls_0": 1.14491, "loss_box_0": 2.18904, "loss_cns_0": 0.61673, "loss_yns_0": 0.16811, "loss_cls_1": 1.26154, "loss_box_1": 2.78422, "loss_cns_1": 0.58742, "loss_yns_1": 0.17791, "loss_cls_2": 1.25673, "loss_box_2": 2.67151, "loss_cns_2": 0.59026, "loss_yns_2": 0.17808, "loss_cls_3": 1.25138, "loss_box_3": 2.63635, "loss_cns_3": 0.61299, "loss_yns_3": 0.1734, "loss_cls_4": 1.26395, "loss_box_4": 2.67793, "loss_cns_4": 0.59389, "loss_yns_4": 0.17063, "loss_cls_5": 1.2573, "loss_box_5": 2.80633, "loss_cns_5": 0.5851, "loss_yns_5": 0.17761, "loss_cls_dn_0": 0.4981, "loss_box_dn_0": 1.0053, "loss_cls_dn_1": 0.42682, "loss_box_dn_1": 1.30394, "loss_cls_dn_2": 0.45669, "loss_box_dn_2": 1.29372, "loss_cls_dn_3": 0.44656, "loss_box_dn_3": 1.35779, "loss_cls_dn_4": 0.43466, "loss_box_dn_4": 1.47792, "loss_cls_dn_5": 0.43655, "loss_box_dn_5": 1.60818, "loss_dense_depth": 1.5214, "loss": 40.10092, "grad_norm": 70.3858, "time": 1.48766} -{"mode": "train", "epoch": 1, "iter": 20, "lr": 0.00011, "memory": 49163, "data_time": 0.07925, "loss_cls_0": 1.14704, "loss_box_0": 2.22175, "loss_cns_0": 0.62185, "loss_yns_0": 0.16937, "loss_cls_1": 1.26028, "loss_box_1": 2.71522, "loss_cns_1": 0.59703, "loss_yns_1": 0.17464, "loss_cls_2": 1.26305, "loss_box_2": 2.62283, "loss_cns_2": 0.59669, "loss_yns_2": 0.18143, "loss_cls_3": 1.24911, "loss_box_3": 2.66855, "loss_cns_3": 0.60167, "loss_yns_3": 0.17687, "loss_cls_4": 1.25571, "loss_box_4": 2.70832, "loss_cns_4": 0.59041, "loss_yns_4": 0.17843, "loss_cls_5": 1.26615, "loss_box_5": 2.7391, "loss_cns_5": 0.5858, "loss_yns_5": 0.17639, "loss_cls_dn_0": 0.4602, "loss_box_dn_0": 1.0039, "loss_cls_dn_1": 0.39703, "loss_box_dn_1": 1.34938, "loss_cls_dn_2": 0.41742, "loss_box_dn_2": 1.34413, "loss_cls_dn_3": 0.40454, "loss_box_dn_3": 1.38668, "loss_cls_dn_4": 0.41788, "loss_box_dn_4": 1.46747, "loss_cls_dn_5": 0.40804, "loss_box_dn_5": 1.51785, "loss_dense_depth": 1.48305, "loss": 39.82529, "grad_norm": 52.60023, "time": 1.48301} -{"mode": "train", "epoch": 1, "iter": 21, "lr": 0.00011, "memory": 49163, "data_time": 0.09731, "loss_cls_0": 1.18567, "loss_box_0": 2.1879, "loss_cns_0": 0.62207, "loss_yns_0": 0.16794, "loss_cls_1": 1.23317, "loss_box_1": 2.79307, "loss_cns_1": 0.58709, "loss_yns_1": 0.17579, "loss_cls_2": 1.26399, "loss_box_2": 2.81226, "loss_cns_2": 0.58393, "loss_yns_2": 0.17628, "loss_cls_3": 1.2447, "loss_box_3": 2.90339, "loss_cns_3": 0.59439, "loss_yns_3": 0.17417, "loss_cls_4": 1.24848, "loss_box_4": 3.00202, "loss_cns_4": 0.58965, "loss_yns_4": 0.17067, "loss_cls_5": 1.25077, "loss_box_5": 3.03832, "loss_cns_5": 0.58875, "loss_yns_5": 0.17119, "loss_cls_dn_0": 0.42539, "loss_box_dn_0": 1.02391, "loss_cls_dn_1": 0.42451, "loss_box_dn_1": 1.28383, "loss_cls_dn_2": 0.4301, "loss_box_dn_2": 1.3672, "loss_cls_dn_3": 0.41423, "loss_box_dn_3": 1.44565, "loss_cls_dn_4": 0.44244, "loss_box_dn_4": 1.60488, "loss_cls_dn_5": 0.4407, "loss_box_dn_5": 1.61871, "loss_dense_depth": 1.42273, "loss": 41.10995, "grad_norm": 94.138, "time": 1.58118} -{"mode": "train", "epoch": 1, "iter": 22, "lr": 0.00011, "memory": 49163, "data_time": 0.1853, "loss_cls_0": 1.22425, "loss_box_0": 2.18201, "loss_cns_0": 0.62081, "loss_yns_0": 0.16671, "loss_cls_1": 1.21921, "loss_box_1": 2.78751, "loss_cns_1": 0.57075, "loss_yns_1": 0.17589, "loss_cls_2": 1.25778, "loss_box_2": 2.89431, "loss_cns_2": 0.56867, "loss_yns_2": 0.17494, "loss_cls_3": 1.26242, "loss_box_3": 2.95685, "loss_cns_3": 0.57676, "loss_yns_3": 0.1694, "loss_cls_4": 1.25362, "loss_box_4": 3.08187, "loss_cns_4": 0.56897, "loss_yns_4": 0.17783, "loss_cls_5": 1.24811, "loss_box_5": 3.15058, "loss_cns_5": 0.55935, "loss_yns_5": 0.18462, "loss_cls_dn_0": 0.41907, "loss_box_dn_0": 1.02949, "loss_cls_dn_1": 0.43653, "loss_box_dn_1": 1.34438, "loss_cls_dn_2": 0.43609, "loss_box_dn_2": 1.43432, "loss_cls_dn_3": 0.41498, "loss_box_dn_3": 1.4943, "loss_cls_dn_4": 0.44104, "loss_box_dn_4": 1.66632, "loss_cls_dn_5": 0.44574, "loss_box_dn_5": 1.68424, "loss_dense_depth": 1.41612, "loss": 41.69582, "grad_norm": 98.11241, "time": 1.59339} -{"mode": "train", "epoch": 1, "iter": 23, "lr": 0.00011, "memory": 49163, "data_time": 0.0766, "loss_cls_0": 1.14221, "loss_box_0": 2.18016, "loss_cns_0": 0.61894, "loss_yns_0": 0.16854, "loss_cls_1": 1.20172, "loss_box_1": 2.70911, "loss_cns_1": 0.57615, "loss_yns_1": 0.17321, "loss_cls_2": 1.23247, "loss_box_2": 2.73726, "loss_cns_2": 0.58554, "loss_yns_2": 0.17443, "loss_cls_3": 1.25376, "loss_box_3": 2.73316, "loss_cns_3": 0.60215, "loss_yns_3": 0.17067, "loss_cls_4": 1.26369, "loss_box_4": 2.85516, "loss_cns_4": 0.57853, "loss_yns_4": 0.18483, "loss_cls_5": 1.24342, "loss_box_5": 2.95327, "loss_cns_5": 0.58379, "loss_yns_5": 0.1896, "loss_cls_dn_0": 0.44646, "loss_box_dn_0": 1.02304, "loss_cls_dn_1": 0.43178, "loss_box_dn_1": 1.30655, "loss_cls_dn_2": 0.43609, "loss_box_dn_2": 1.36006, "loss_cls_dn_3": 0.41017, "loss_box_dn_3": 1.38092, "loss_cls_dn_4": 0.41565, "loss_box_dn_4": 1.52748, "loss_cls_dn_5": 0.43396, "loss_box_dn_5": 1.58685, "loss_dense_depth": 1.37433, "loss": 40.24515, "grad_norm": 61.82904, "time": 1.50784} -{"mode": "train", "epoch": 1, "iter": 24, "lr": 0.00011, "memory": 49163, "data_time": 0.07584, "loss_cls_0": 1.14201, "loss_box_0": 2.14282, "loss_cns_0": 0.59916, "loss_yns_0": 0.16727, "loss_cls_1": 1.21956, "loss_box_1": 2.73755, "loss_cns_1": 0.56974, "loss_yns_1": 0.17466, "loss_cls_2": 1.22769, "loss_box_2": 2.79538, "loss_cns_2": 0.56862, "loss_yns_2": 0.18052, "loss_cls_3": 1.24829, "loss_box_3": 2.78532, "loss_cns_3": 0.58993, "loss_yns_3": 0.17244, "loss_cls_4": 1.2781, "loss_box_4": 2.84525, "loss_cns_4": 0.56354, "loss_yns_4": 0.18037, "loss_cls_5": 1.25969, "loss_box_5": 2.8597, "loss_cns_5": 0.5822, "loss_yns_5": 0.17773, "loss_cls_dn_0": 0.50366, "loss_box_dn_0": 1.06252, "loss_cls_dn_1": 0.41599, "loss_box_dn_1": 1.4195, "loss_cls_dn_2": 0.42965, "loss_box_dn_2": 1.43805, "loss_cls_dn_3": 0.40974, "loss_box_dn_3": 1.41876, "loss_cls_dn_4": 0.39665, "loss_box_dn_4": 1.49391, "loss_cls_dn_5": 0.41443, "loss_box_dn_5": 1.52121, "loss_dense_depth": 1.35699, "loss": 40.34855, "grad_norm": 69.84389, "time": 1.50719} -{"mode": "train", "epoch": 1, "iter": 25, "lr": 0.00011, "memory": 49163, "data_time": 0.08365, "loss_cls_0": 1.1449, "loss_box_0": 2.11869, "loss_cns_0": 0.60108, "loss_yns_0": 0.16252, "loss_cls_1": 1.19848, "loss_box_1": 2.89003, "loss_cns_1": 0.54879, "loss_yns_1": 0.17556, "loss_cls_2": 1.20442, "loss_box_2": 2.93774, "loss_cns_2": 0.54195, "loss_yns_2": 0.17589, "loss_cls_3": 1.23448, "loss_box_3": 3.01114, "loss_cns_3": 0.56031, "loss_yns_3": 0.17236, "loss_cls_4": 1.23989, "loss_box_4": 3.04204, "loss_cns_4": 0.55272, "loss_yns_4": 0.18894, "loss_cls_5": 1.25651, "loss_box_5": 2.99548, "loss_cns_5": 0.56664, "loss_yns_5": 0.17954, "loss_cls_dn_0": 0.50725, "loss_box_dn_0": 1.05729, "loss_cls_dn_1": 0.40784, "loss_box_dn_1": 1.40959, "loss_cls_dn_2": 0.42923, "loss_box_dn_2": 1.40984, "loss_cls_dn_3": 0.41929, "loss_box_dn_3": 1.43019, "loss_cls_dn_4": 0.40605, "loss_box_dn_4": 1.44643, "loss_cls_dn_5": 0.40852, "loss_box_dn_5": 1.47114, "loss_dense_depth": 1.33995, "loss": 40.84271, "grad_norm": 84.28497, "time": 1.48949} -{"mode": "train", "epoch": 1, "iter": 26, "lr": 0.00011, "memory": 49163, "data_time": 0.0788, "loss_cls_0": 1.12377, "loss_box_0": 2.09686, "loss_cns_0": 0.6139, "loss_yns_0": 0.16948, "loss_cls_1": 1.17615, "loss_box_1": 2.86863, "loss_cns_1": 0.5516, "loss_yns_1": 0.17217, "loss_cls_2": 1.20184, "loss_box_2": 2.85386, "loss_cns_2": 0.55377, "loss_yns_2": 0.17522, "loss_cls_3": 1.22915, "loss_box_3": 3.04608, "loss_cns_3": 0.56175, "loss_yns_3": 0.17037, "loss_cls_4": 1.22558, "loss_box_4": 3.12842, "loss_cns_4": 0.56031, "loss_yns_4": 0.17514, "loss_cls_5": 1.24871, "loss_box_5": 3.19691, "loss_cns_5": 0.54561, "loss_yns_5": 0.17482, "loss_cls_dn_0": 0.46613, "loss_box_dn_0": 1.04958, "loss_cls_dn_1": 0.43841, "loss_box_dn_1": 1.27045, "loss_cls_dn_2": 0.45507, "loss_box_dn_2": 1.2656, "loss_cls_dn_3": 0.45059, "loss_box_dn_3": 1.40451, "loss_cls_dn_4": 0.45089, "loss_box_dn_4": 1.42472, "loss_cls_dn_5": 0.44053, "loss_box_dn_5": 1.52671, "loss_dense_depth": 1.27215, "loss": 40.73547, "grad_norm": 98.0892, "time": 1.49678} -{"mode": "train", "epoch": 1, "iter": 27, "lr": 0.00011, "memory": 49163, "data_time": 0.07776, "loss_cls_0": 1.1416, "loss_box_0": 2.10443, "loss_cns_0": 0.61049, "loss_yns_0": 0.17246, "loss_cls_1": 1.1672, "loss_box_1": 2.79891, "loss_cns_1": 0.55882, "loss_yns_1": 0.17291, "loss_cls_2": 1.19354, "loss_box_2": 2.77335, "loss_cns_2": 0.56466, "loss_yns_2": 0.17631, "loss_cls_3": 1.21691, "loss_box_3": 2.89033, "loss_cns_3": 0.5618, "loss_yns_3": 0.16976, "loss_cls_4": 1.21332, "loss_box_4": 2.95017, "loss_cns_4": 0.55153, "loss_yns_4": 0.1726, "loss_cls_5": 1.23414, "loss_box_5": 3.13133, "loss_cns_5": 0.5265, "loss_yns_5": 0.17265, "loss_cls_dn_0": 0.452, "loss_box_dn_0": 1.03577, "loss_cls_dn_1": 0.42764, "loss_box_dn_1": 1.31256, "loss_cls_dn_2": 0.4374, "loss_box_dn_2": 1.2858, "loss_cls_dn_3": 0.43963, "loss_box_dn_3": 1.40749, "loss_cls_dn_4": 0.45507, "loss_box_dn_4": 1.4062, "loss_cls_dn_5": 0.44538, "loss_box_dn_5": 1.5478, "loss_dense_depth": 1.26036, "loss": 40.13883, "grad_norm": 80.41345, "time": 1.4933} -{"mode": "train", "epoch": 1, "iter": 28, "lr": 0.00011, "memory": 49163, "data_time": 0.07839, "loss_cls_0": 1.1282, "loss_box_0": 2.11714, "loss_cns_0": 0.60987, "loss_yns_0": 0.1722, "loss_cls_1": 1.16943, "loss_box_1": 2.5647, "loss_cns_1": 0.57527, "loss_yns_1": 0.17223, "loss_cls_2": 1.20996, "loss_box_2": 2.57064, "loss_cns_2": 0.58066, "loss_yns_2": 0.17746, "loss_cls_3": 1.21365, "loss_box_3": 2.61626, "loss_cns_3": 0.57561, "loss_yns_3": 0.17031, "loss_cls_4": 1.20726, "loss_box_4": 2.68675, "loss_cns_4": 0.57265, "loss_yns_4": 0.17004, "loss_cls_5": 1.22532, "loss_box_5": 2.75231, "loss_cns_5": 0.56015, "loss_yns_5": 0.17475, "loss_cls_dn_0": 0.44141, "loss_box_dn_0": 1.03072, "loss_cls_dn_1": 0.41716, "loss_box_dn_1": 1.35441, "loss_cls_dn_2": 0.40567, "loss_box_dn_2": 1.35469, "loss_cls_dn_3": 0.41157, "loss_box_dn_3": 1.4462, "loss_cls_dn_4": 0.43431, "loss_box_dn_4": 1.46377, "loss_cls_dn_5": 0.43724, "loss_box_dn_5": 1.55721, "loss_dense_depth": 1.24175, "loss": 38.9689, "grad_norm": 69.81181, "time": 1.48258} -{"mode": "train", "epoch": 1, "iter": 29, "lr": 0.00011, "memory": 49163, "data_time": 0.07838, "loss_cls_0": 1.11123, "loss_box_0": 2.13047, "loss_cns_0": 0.60808, "loss_yns_0": 0.1721, "loss_cls_1": 1.17605, "loss_box_1": 2.53723, "loss_cns_1": 0.57861, "loss_yns_1": 0.17217, "loss_cls_2": 1.25492, "loss_box_2": 2.54978, "loss_cns_2": 0.58199, "loss_yns_2": 0.17127, "loss_cls_3": 1.21375, "loss_box_3": 2.60277, "loss_cns_3": 0.58773, "loss_yns_3": 0.16891, "loss_cls_4": 1.2118, "loss_box_4": 2.71452, "loss_cns_4": 0.58397, "loss_yns_4": 0.16933, "loss_cls_5": 1.20454, "loss_box_5": 2.63476, "loss_cns_5": 0.59298, "loss_yns_5": 0.17242, "loss_cls_dn_0": 0.43748, "loss_box_dn_0": 1.03744, "loss_cls_dn_1": 0.40353, "loss_box_dn_1": 1.26792, "loss_cls_dn_2": 0.3884, "loss_box_dn_2": 1.2947, "loss_cls_dn_3": 0.40292, "loss_box_dn_3": 1.33971, "loss_cls_dn_4": 0.42198, "loss_box_dn_4": 1.37592, "loss_cls_dn_5": 0.43728, "loss_box_dn_5": 1.3925, "loss_dense_depth": 1.20793, "loss": 38.30909, "grad_norm": 91.70428, "time": 1.50612} -{"mode": "train", "epoch": 1, "iter": 30, "lr": 0.00011, "memory": 49163, "data_time": 0.07416, "loss_cls_0": 1.10794, "loss_box_0": 2.08483, "loss_cns_0": 0.61003, "loss_yns_0": 0.17558, "loss_cls_1": 1.14396, "loss_box_1": 2.62988, "loss_cns_1": 0.56698, "loss_yns_1": 0.17633, "loss_cls_2": 1.18404, "loss_box_2": 2.72976, "loss_cns_2": 0.57277, "loss_yns_2": 0.17872, "loss_cls_3": 1.17066, "loss_box_3": 2.65072, "loss_cns_3": 0.59031, "loss_yns_3": 0.16931, "loss_cls_4": 1.18281, "loss_box_4": 2.73638, "loss_cns_4": 0.57808, "loss_yns_4": 0.17107, "loss_cls_5": 1.18137, "loss_box_5": 2.68663, "loss_cns_5": 0.58105, "loss_yns_5": 0.17147, "loss_cls_dn_0": 0.42553, "loss_box_dn_0": 1.03945, "loss_cls_dn_1": 0.38887, "loss_box_dn_1": 1.34803, "loss_cls_dn_2": 0.38881, "loss_box_dn_2": 1.39108, "loss_cls_dn_3": 0.39344, "loss_box_dn_3": 1.35164, "loss_cls_dn_4": 0.39883, "loss_box_dn_4": 1.38735, "loss_cls_dn_5": 0.40813, "loss_box_dn_5": 1.39588, "loss_dense_depth": 1.21345, "loss": 38.56115, "grad_norm": 82.2647, "time": 1.48646} -{"mode": "train", "epoch": 1, "iter": 31, "lr": 0.00011, "memory": 49163, "data_time": 0.0777, "loss_cls_0": 1.10146, "loss_box_0": 2.05727, "loss_cns_0": 0.61419, "loss_yns_0": 0.17439, "loss_cls_1": 1.12149, "loss_box_1": 2.72923, "loss_cns_1": 0.55698, "loss_yns_1": 0.17413, "loss_cls_2": 1.14346, "loss_box_2": 2.81378, "loss_cns_2": 0.56967, "loss_yns_2": 0.17853, "loss_cls_3": 1.15783, "loss_box_3": 2.6836, "loss_cns_3": 0.58525, "loss_yns_3": 0.1692, "loss_cls_4": 1.1655, "loss_box_4": 2.65387, "loss_cns_4": 0.57911, "loss_yns_4": 0.17761, "loss_cls_5": 1.18021, "loss_box_5": 2.70547, "loss_cns_5": 0.58102, "loss_yns_5": 0.1728, "loss_cls_dn_0": 0.42478, "loss_box_dn_0": 1.02223, "loss_cls_dn_1": 0.38559, "loss_box_dn_1": 1.35973, "loss_cls_dn_2": 0.41155, "loss_box_dn_2": 1.39862, "loss_cls_dn_3": 0.39379, "loss_box_dn_3": 1.31738, "loss_cls_dn_4": 0.38901, "loss_box_dn_4": 1.3468, "loss_cls_dn_5": 0.39476, "loss_box_dn_5": 1.3707, "loss_dense_depth": 1.14092, "loss": 38.40193, "grad_norm": 64.70496, "time": 1.48606} -{"mode": "train", "epoch": 1, "iter": 32, "lr": 0.00011, "memory": 49163, "data_time": 0.07734, "loss_cls_0": 1.06536, "loss_box_0": 2.04833, "loss_cns_0": 0.61434, "loss_yns_0": 0.17208, "loss_cls_1": 1.10593, "loss_box_1": 2.73733, "loss_cns_1": 0.55726, "loss_yns_1": 0.17126, "loss_cls_2": 1.13561, "loss_box_2": 2.80708, "loss_cns_2": 0.56559, "loss_yns_2": 0.17272, "loss_cls_3": 1.15186, "loss_box_3": 2.75412, "loss_cns_3": 0.57577, "loss_yns_3": 0.17121, "loss_cls_4": 1.15557, "loss_box_4": 2.72474, "loss_cns_4": 0.57561, "loss_yns_4": 0.18056, "loss_cls_5": 1.1696, "loss_box_5": 2.78905, "loss_cns_5": 0.58388, "loss_yns_5": 0.17625, "loss_cls_dn_0": 0.41483, "loss_box_dn_0": 1.00356, "loss_cls_dn_1": 0.37979, "loss_box_dn_1": 1.26527, "loss_cls_dn_2": 0.41508, "loss_box_dn_2": 1.30822, "loss_cls_dn_3": 0.37924, "loss_box_dn_3": 1.29335, "loss_cls_dn_4": 0.38033, "loss_box_dn_4": 1.35995, "loss_cls_dn_5": 0.39113, "loss_box_dn_5": 1.38961, "loss_dense_depth": 1.1322, "loss": 38.27367, "grad_norm": 76.81515, "time": 1.48685} -{"mode": "train", "epoch": 1, "iter": 33, "lr": 0.00011, "memory": 49163, "data_time": 0.07601, "loss_cls_0": 1.06023, "loss_box_0": 2.04228, "loss_cns_0": 0.61657, "loss_yns_0": 0.17257, "loss_cls_1": 1.10164, "loss_box_1": 2.67954, "loss_cns_1": 0.56977, "loss_yns_1": 0.17321, "loss_cls_2": 1.13711, "loss_box_2": 2.71057, "loss_cns_2": 0.58229, "loss_yns_2": 0.1722, "loss_cls_3": 1.14722, "loss_box_3": 2.73413, "loss_cns_3": 0.58662, "loss_yns_3": 0.17299, "loss_cls_4": 1.14842, "loss_box_4": 2.76853, "loss_cns_4": 0.58289, "loss_yns_4": 0.1747, "loss_cls_5": 1.14682, "loss_box_5": 2.7782, "loss_cns_5": 0.58964, "loss_yns_5": 0.17922, "loss_cls_dn_0": 0.42016, "loss_box_dn_0": 1.00457, "loss_cls_dn_1": 0.36244, "loss_box_dn_1": 1.38234, "loss_cls_dn_2": 0.3978, "loss_box_dn_2": 1.40445, "loss_cls_dn_3": 0.37027, "loss_box_dn_3": 1.42853, "loss_cls_dn_4": 0.3696, "loss_box_dn_4": 1.51259, "loss_cls_dn_5": 0.38542, "loss_box_dn_5": 1.52616, "loss_dense_depth": 1.12124, "loss": 38.71292, "grad_norm": 68.519, "time": 1.4982} -{"mode": "train", "epoch": 1, "iter": 34, "lr": 0.00011, "memory": 49163, "data_time": 0.07937, "loss_cls_0": 1.0558, "loss_box_0": 2.03214, "loss_cns_0": 0.618, "loss_yns_0": 0.17174, "loss_cls_1": 1.09964, "loss_box_1": 2.49715, "loss_cns_1": 0.59015, "loss_yns_1": 0.17749, "loss_cls_2": 1.12954, "loss_box_2": 2.48872, "loss_cns_2": 0.60167, "loss_yns_2": 0.17602, "loss_cls_3": 1.13904, "loss_box_3": 2.57505, "loss_cns_3": 0.60167, "loss_yns_3": 0.17427, "loss_cls_4": 1.14865, "loss_box_4": 2.62043, "loss_cns_4": 0.58944, "loss_yns_4": 0.17933, "loss_cls_5": 1.14841, "loss_box_5": 2.6373, "loss_cns_5": 0.58816, "loss_yns_5": 0.17779, "loss_cls_dn_0": 0.42813, "loss_box_dn_0": 1.00089, "loss_cls_dn_1": 0.35425, "loss_box_dn_1": 1.47939, "loss_cls_dn_2": 0.37291, "loss_box_dn_2": 1.46409, "loss_cls_dn_3": 0.37, "loss_box_dn_3": 1.49749, "loss_cls_dn_4": 0.36221, "loss_box_dn_4": 1.5619, "loss_cls_dn_5": 0.37859, "loss_box_dn_5": 1.57618, "loss_dense_depth": 1.14431, "loss": 38.20794, "grad_norm": 73.39757, "time": 1.48891} -{"mode": "train", "epoch": 1, "iter": 35, "lr": 0.00011, "memory": 49163, "data_time": 0.08685, "loss_cls_0": 1.06442, "loss_box_0": 1.98133, "loss_cns_0": 0.62215, "loss_yns_0": 0.17024, "loss_cls_1": 1.13548, "loss_box_1": 2.37765, "loss_cns_1": 0.59704, "loss_yns_1": 0.17571, "loss_cls_2": 1.14882, "loss_box_2": 2.3482, "loss_cns_2": 0.60858, "loss_yns_2": 0.18206, "loss_cls_3": 1.15603, "loss_box_3": 2.42273, "loss_cns_3": 0.6181, "loss_yns_3": 0.17894, "loss_cls_4": 1.1636, "loss_box_4": 2.43614, "loss_cns_4": 0.60529, "loss_yns_4": 0.18277, "loss_cls_5": 1.16108, "loss_box_5": 2.50307, "loss_cns_5": 0.60105, "loss_yns_5": 0.17555, "loss_cls_dn_0": 0.42739, "loss_box_dn_0": 0.98364, "loss_cls_dn_1": 0.3437, "loss_box_dn_1": 1.45299, "loss_cls_dn_2": 0.35035, "loss_box_dn_2": 1.43253, "loss_cls_dn_3": 0.36355, "loss_box_dn_3": 1.45913, "loss_cls_dn_4": 0.36209, "loss_box_dn_4": 1.50033, "loss_cls_dn_5": 0.373, "loss_box_dn_5": 1.52818, "loss_dense_depth": 1.05512, "loss": 37.24803, "grad_norm": 52.50537, "time": 1.4941} -{"mode": "train", "epoch": 1, "iter": 36, "lr": 0.00011, "memory": 49163, "data_time": 0.07616, "loss_cls_0": 1.03888, "loss_box_0": 1.96805, "loss_cns_0": 0.61928, "loss_yns_0": 0.17142, "loss_cls_1": 1.12547, "loss_box_1": 2.42908, "loss_cns_1": 0.59388, "loss_yns_1": 0.17207, "loss_cls_2": 1.15375, "loss_box_2": 2.42292, "loss_cns_2": 0.60858, "loss_yns_2": 0.17657, "loss_cls_3": 1.15155, "loss_box_3": 2.48152, "loss_cns_3": 0.62001, "loss_yns_3": 0.17588, "loss_cls_4": 1.15509, "loss_box_4": 2.5263, "loss_cns_4": 0.60906, "loss_yns_4": 0.17459, "loss_cls_5": 1.14876, "loss_box_5": 2.50485, "loss_cns_5": 0.61343, "loss_yns_5": 0.17709, "loss_cls_dn_0": 0.43646, "loss_box_dn_0": 0.96949, "loss_cls_dn_1": 0.35786, "loss_box_dn_1": 1.17861, "loss_cls_dn_2": 0.36052, "loss_box_dn_2": 1.17496, "loss_cls_dn_3": 0.38223, "loss_box_dn_3": 1.21241, "loss_cls_dn_4": 0.3919, "loss_box_dn_4": 1.26919, "loss_cls_dn_5": 0.39432, "loss_box_dn_5": 1.26813, "loss_dense_depth": 1.05668, "loss": 36.27085, "grad_norm": 72.9751, "time": 1.48437} -{"mode": "train", "epoch": 1, "iter": 37, "lr": 0.00011, "memory": 49163, "data_time": 0.07691, "loss_cls_0": 1.03539, "loss_box_0": 1.96161, "loss_cns_0": 0.61824, "loss_yns_0": 0.17099, "loss_cls_1": 1.12008, "loss_box_1": 2.46245, "loss_cns_1": 0.59431, "loss_yns_1": 0.17388, "loss_cls_2": 1.16653, "loss_box_2": 2.47357, "loss_cns_2": 0.60197, "loss_yns_2": 0.17216, "loss_cls_3": 1.13912, "loss_box_3": 2.51397, "loss_cns_3": 0.6113, "loss_yns_3": 0.16995, "loss_cls_4": 1.1427, "loss_box_4": 2.55122, "loss_cns_4": 0.60658, "loss_yns_4": 0.17285, "loss_cls_5": 1.15049, "loss_box_5": 2.51724, "loss_cns_5": 0.61206, "loss_yns_5": 0.17144, "loss_cls_dn_0": 0.43768, "loss_box_dn_0": 0.96814, "loss_cls_dn_1": 0.34059, "loss_box_dn_1": 1.20755, "loss_cls_dn_2": 0.35897, "loss_box_dn_2": 1.1978, "loss_cls_dn_3": 0.36934, "loss_box_dn_3": 1.23045, "loss_cls_dn_4": 0.38135, "loss_box_dn_4": 1.27269, "loss_cls_dn_5": 0.37334, "loss_box_dn_5": 1.27666, "loss_dense_depth": 1.04378, "loss": 36.36844, "grad_norm": 72.54264, "time": 1.49547} -{"mode": "train", "epoch": 1, "iter": 38, "lr": 0.00011, "memory": 49163, "data_time": 0.07811, "loss_cls_0": 1.03719, "loss_box_0": 1.98091, "loss_cns_0": 0.61966, "loss_yns_0": 0.1701, "loss_cls_1": 1.1208, "loss_box_1": 2.46764, "loss_cns_1": 0.5999, "loss_yns_1": 0.18141, "loss_cls_2": 1.1474, "loss_box_2": 2.49909, "loss_cns_2": 0.61397, "loss_yns_2": 0.17628, "loss_cls_3": 1.13676, "loss_box_3": 2.54318, "loss_cns_3": 0.60694, "loss_yns_3": 0.17054, "loss_cls_4": 1.1357, "loss_box_4": 2.57245, "loss_cns_4": 0.60296, "loss_yns_4": 0.17247, "loss_cls_5": 1.15748, "loss_box_5": 2.54973, "loss_cns_5": 0.60681, "loss_yns_5": 0.17177, "loss_cls_dn_0": 0.41884, "loss_box_dn_0": 0.9666, "loss_cls_dn_1": 0.3329, "loss_box_dn_1": 1.08942, "loss_cls_dn_2": 0.35985, "loss_box_dn_2": 1.09098, "loss_cls_dn_3": 0.36309, "loss_box_dn_3": 1.14941, "loss_cls_dn_4": 0.37233, "loss_box_dn_4": 1.17614, "loss_cls_dn_5": 0.37053, "loss_box_dn_5": 1.22374, "loss_dense_depth": 1.02791, "loss": 35.98285, "grad_norm": 54.44703, "time": 1.49962} -{"mode": "train", "epoch": 1, "iter": 39, "lr": 0.00012, "memory": 49163, "data_time": 0.07547, "loss_cls_0": 1.07326, "loss_box_0": 1.93796, "loss_cns_0": 0.62092, "loss_yns_0": 0.1693, "loss_cls_1": 1.12296, "loss_box_1": 2.42631, "loss_cns_1": 0.60135, "loss_yns_1": 0.17583, "loss_cls_2": 1.13698, "loss_box_2": 2.49306, "loss_cns_2": 0.61337, "loss_yns_2": 0.17882, "loss_cls_3": 1.151, "loss_box_3": 2.54109, "loss_cns_3": 0.60176, "loss_yns_3": 0.17243, "loss_cls_4": 1.15172, "loss_box_4": 2.5562, "loss_cns_4": 0.59469, "loss_yns_4": 0.17571, "loss_cls_5": 1.14782, "loss_box_5": 2.55639, "loss_cns_5": 0.60055, "loss_yns_5": 0.16722, "loss_cls_dn_0": 0.40403, "loss_box_dn_0": 0.94146, "loss_cls_dn_1": 0.33505, "loss_box_dn_1": 1.05282, "loss_cls_dn_2": 0.36214, "loss_box_dn_2": 1.0886, "loss_cls_dn_3": 0.36703, "loss_box_dn_3": 1.17082, "loss_cls_dn_4": 0.36858, "loss_box_dn_4": 1.19666, "loss_cls_dn_5": 0.38822, "loss_box_dn_5": 1.28984, "loss_dense_depth": 1.00934, "loss": 35.94129, "grad_norm": 69.73831, "time": 1.50146} -{"mode": "train", "epoch": 1, "iter": 40, "lr": 0.00012, "memory": 49163, "data_time": 0.07851, "loss_cls_0": 1.07034, "loss_box_0": 1.95064, "loss_cns_0": 0.62178, "loss_yns_0": 0.16892, "loss_cls_1": 1.1071, "loss_box_1": 2.31153, "loss_cns_1": 0.6063, "loss_yns_1": 0.16971, "loss_cls_2": 1.11796, "loss_box_2": 2.32491, "loss_cns_2": 0.61345, "loss_yns_2": 0.17443, "loss_cls_3": 1.12565, "loss_box_3": 2.35251, "loss_cns_3": 0.61031, "loss_yns_3": 0.17231, "loss_cls_4": 1.13057, "loss_box_4": 2.40195, "loss_cns_4": 0.60064, "loss_yns_4": 0.17152, "loss_cls_5": 1.12914, "loss_box_5": 2.52062, "loss_cns_5": 0.59116, "loss_yns_5": 0.1671, "loss_cls_dn_0": 0.39847, "loss_box_dn_0": 0.93539, "loss_cls_dn_1": 0.34248, "loss_box_dn_1": 1.13486, "loss_cls_dn_2": 0.36632, "loss_box_dn_2": 1.17359, "loss_cls_dn_3": 0.37722, "loss_box_dn_3": 1.24737, "loss_cls_dn_4": 0.37114, "loss_box_dn_4": 1.28702, "loss_cls_dn_5": 0.40536, "loss_box_dn_5": 1.41168, "loss_dense_depth": 1.03603, "loss": 35.69746, "grad_norm": 71.72676, "time": 1.49433} -{"mode": "train", "epoch": 1, "iter": 41, "lr": 0.00012, "memory": 49163, "data_time": 0.08251, "loss_cls_0": 1.01763, "loss_box_0": 1.98363, "loss_cns_0": 0.61921, "loss_yns_0": 0.16734, "loss_cls_1": 1.10328, "loss_box_1": 2.28134, "loss_cns_1": 0.61057, "loss_yns_1": 0.16953, "loss_cls_2": 1.11447, "loss_box_2": 2.2328, "loss_cns_2": 0.61942, "loss_yns_2": 0.16958, "loss_cls_3": 1.1134, "loss_box_3": 2.26912, "loss_cns_3": 0.61811, "loss_yns_3": 0.16944, "loss_cls_4": 1.11016, "loss_box_4": 2.30818, "loss_cns_4": 0.61175, "loss_yns_4": 0.165, "loss_cls_5": 1.12108, "loss_box_5": 2.39493, "loss_cns_5": 0.60116, "loss_yns_5": 0.16888, "loss_cls_dn_0": 0.41663, "loss_box_dn_0": 0.9361, "loss_cls_dn_1": 0.33343, "loss_box_dn_1": 1.23009, "loss_cls_dn_2": 0.36134, "loss_box_dn_2": 1.24715, "loss_cls_dn_3": 0.37346, "loss_box_dn_3": 1.3043, "loss_cls_dn_4": 0.37058, "loss_box_dn_4": 1.338, "loss_cls_dn_5": 0.39376, "loss_box_dn_5": 1.44235, "loss_dense_depth": 1.01171, "loss": 35.49891, "grad_norm": 67.75639, "time": 1.5553} -{"mode": "train", "epoch": 1, "iter": 42, "lr": 0.00012, "memory": 49163, "data_time": 0.15803, "loss_cls_0": 1.01698, "loss_box_0": 1.97018, "loss_cns_0": 0.62161, "loss_yns_0": 0.16647, "loss_cls_1": 1.09547, "loss_box_1": 2.28386, "loss_cns_1": 0.60811, "loss_yns_1": 0.17116, "loss_cls_2": 1.11614, "loss_box_2": 2.25989, "loss_cns_2": 0.61113, "loss_yns_2": 0.16893, "loss_cls_3": 1.10949, "loss_box_3": 2.30244, "loss_cns_3": 0.61815, "loss_yns_3": 0.16721, "loss_cls_4": 1.11208, "loss_box_4": 2.26857, "loss_cns_4": 0.61826, "loss_yns_4": 0.17072, "loss_cls_5": 1.12089, "loss_box_5": 2.28824, "loss_cns_5": 0.61694, "loss_yns_5": 0.16714, "loss_cls_dn_0": 0.42459, "loss_box_dn_0": 0.93432, "loss_cls_dn_1": 0.33564, "loss_box_dn_1": 1.20827, "loss_cls_dn_2": 0.36409, "loss_box_dn_2": 1.20921, "loss_cls_dn_3": 0.37313, "loss_box_dn_3": 1.26588, "loss_cls_dn_4": 0.36983, "loss_box_dn_4": 1.27451, "loss_cls_dn_5": 0.38917, "loss_box_dn_5": 1.34463, "loss_dense_depth": 0.97338, "loss": 35.11669, "grad_norm": 69.46452, "time": 1.57036} -{"mode": "train", "epoch": 1, "iter": 43, "lr": 0.00012, "memory": 49163, "data_time": 0.07669, "loss_cls_0": 1.04132, "loss_box_0": 1.98658, "loss_cns_0": 0.6153, "loss_yns_0": 0.16463, "loss_cls_1": 1.11584, "loss_box_1": 2.38488, "loss_cns_1": 0.58822, "loss_yns_1": 0.16791, "loss_cls_2": 1.12928, "loss_box_2": 2.33427, "loss_cns_2": 0.59827, "loss_yns_2": 0.16857, "loss_cls_3": 1.13521, "loss_box_3": 2.39121, "loss_cns_3": 0.60531, "loss_yns_3": 0.16767, "loss_cls_4": 1.134, "loss_box_4": 2.30991, "loss_cns_4": 0.60892, "loss_yns_4": 0.16939, "loss_cls_5": 1.14335, "loss_box_5": 2.32591, "loss_cns_5": 0.61016, "loss_yns_5": 0.16746, "loss_cls_dn_0": 0.4353, "loss_box_dn_0": 0.92858, "loss_cls_dn_1": 0.33519, "loss_box_dn_1": 1.14963, "loss_cls_dn_2": 0.3738, "loss_box_dn_2": 1.12952, "loss_cls_dn_3": 0.37202, "loss_box_dn_3": 1.18088, "loss_cls_dn_4": 0.37434, "loss_box_dn_4": 1.16769, "loss_cls_dn_5": 0.38722, "loss_box_dn_5": 1.21733, "loss_dense_depth": 0.98307, "loss": 35.09813, "grad_norm": 61.07757, "time": 1.52682} -{"mode": "train", "epoch": 1, "iter": 44, "lr": 0.00012, "memory": 49163, "data_time": 0.07539, "loss_cls_0": 1.04316, "loss_box_0": 2.07083, "loss_cns_0": 0.60452, "loss_yns_0": 0.16523, "loss_cls_1": 1.11764, "loss_box_1": 2.39656, "loss_cns_1": 0.58414, "loss_yns_1": 0.16442, "loss_cls_2": 1.11341, "loss_box_2": 2.29472, "loss_cns_2": 0.6152, "loss_yns_2": 0.16926, "loss_cls_3": 1.16091, "loss_box_3": 2.31877, "loss_cns_3": 0.60644, "loss_yns_3": 0.17075, "loss_cls_4": 1.13912, "loss_box_4": 2.30497, "loss_cns_4": 0.60578, "loss_yns_4": 0.1721, "loss_cls_5": 1.14008, "loss_box_5": 2.27146, "loss_cns_5": 0.6134, "loss_yns_5": 0.16644, "loss_cls_dn_0": 0.43555, "loss_box_dn_0": 0.94135, "loss_cls_dn_1": 0.32917, "loss_box_dn_1": 1.19167, "loss_cls_dn_2": 0.37448, "loss_box_dn_2": 1.13951, "loss_cls_dn_3": 0.3708, "loss_box_dn_3": 1.15705, "loss_cls_dn_4": 0.37754, "loss_box_dn_4": 1.16583, "loss_cls_dn_5": 0.3842, "loss_box_dn_5": 1.17206, "loss_dense_depth": 1.03272, "loss": 35.08123, "grad_norm": 47.31389, "time": 1.51984} -{"mode": "train", "epoch": 1, "iter": 45, "lr": 0.00012, "memory": 49163, "data_time": 0.08698, "loss_cls_0": 1.06284, "loss_box_0": 2.00305, "loss_cns_0": 0.6133, "loss_yns_0": 0.16654, "loss_cls_1": 1.10949, "loss_box_1": 2.3762, "loss_cns_1": 0.58763, "loss_yns_1": 0.17051, "loss_cls_2": 1.12572, "loss_box_2": 2.30797, "loss_cns_2": 0.607, "loss_yns_2": 0.17046, "loss_cls_3": 1.13709, "loss_box_3": 2.29201, "loss_cns_3": 0.60569, "loss_yns_3": 0.17147, "loss_cls_4": 1.1336, "loss_box_4": 2.33433, "loss_cns_4": 0.60446, "loss_yns_4": 0.17355, "loss_cls_5": 1.15384, "loss_box_5": 2.3158, "loss_cns_5": 0.60853, "loss_yns_5": 0.17242, "loss_cls_dn_0": 0.42006, "loss_box_dn_0": 0.91933, "loss_cls_dn_1": 0.32285, "loss_box_dn_1": 1.11152, "loss_cls_dn_2": 0.36486, "loss_box_dn_2": 1.04968, "loss_cls_dn_3": 0.37108, "loss_box_dn_3": 1.04332, "loss_cls_dn_4": 0.37377, "loss_box_dn_4": 1.07557, "loss_cls_dn_5": 0.37969, "loss_box_dn_5": 1.08064, "loss_dense_depth": 1.03844, "loss": 34.55438, "grad_norm": 41.77581, "time": 1.51112} -{"mode": "train", "epoch": 1, "iter": 46, "lr": 0.00012, "memory": 49163, "data_time": 0.07286, "loss_cls_0": 1.05486, "loss_box_0": 1.99003, "loss_cns_0": 0.61269, "loss_yns_0": 0.16491, "loss_cls_1": 1.08851, "loss_box_1": 2.42843, "loss_cns_1": 0.57704, "loss_yns_1": 0.16721, "loss_cls_2": 1.11601, "loss_box_2": 2.35193, "loss_cns_2": 0.59709, "loss_yns_2": 0.16558, "loss_cls_3": 1.15077, "loss_box_3": 2.34124, "loss_cns_3": 0.60332, "loss_yns_3": 0.16391, "loss_cls_4": 1.12481, "loss_box_4": 2.35429, "loss_cns_4": 0.60194, "loss_yns_4": 0.1715, "loss_cls_5": 1.14592, "loss_box_5": 2.37729, "loss_cns_5": 0.60484, "loss_yns_5": 0.17327, "loss_cls_dn_0": 0.40147, "loss_box_dn_0": 0.90208, "loss_cls_dn_1": 0.31465, "loss_box_dn_1": 1.05855, "loss_cls_dn_2": 0.34116, "loss_box_dn_2": 0.97732, "loss_cls_dn_3": 0.35216, "loss_box_dn_3": 0.9679, "loss_cls_dn_4": 0.35246, "loss_box_dn_4": 0.99235, "loss_cls_dn_5": 0.36072, "loss_box_dn_5": 1.01267, "loss_dense_depth": 1.07314, "loss": 34.23403, "grad_norm": 35.47575, "time": 1.52606} -{"mode": "train", "epoch": 1, "iter": 47, "lr": 0.00012, "memory": 49163, "data_time": 0.07305, "loss_cls_0": 1.04932, "loss_box_0": 2.01276, "loss_cns_0": 0.61538, "loss_yns_0": 0.16406, "loss_cls_1": 1.07496, "loss_box_1": 2.45284, "loss_cns_1": 0.58865, "loss_yns_1": 0.16628, "loss_cls_2": 1.11595, "loss_box_2": 2.37423, "loss_cns_2": 0.60197, "loss_yns_2": 0.16683, "loss_cls_3": 1.13309, "loss_box_3": 2.38624, "loss_cns_3": 0.6033, "loss_yns_3": 0.16509, "loss_cls_4": 1.13635, "loss_box_4": 2.39496, "loss_cns_4": 0.60407, "loss_yns_4": 0.1779, "loss_cls_5": 1.14715, "loss_box_5": 2.41358, "loss_cns_5": 0.60693, "loss_yns_5": 0.16665, "loss_cls_dn_0": 0.401, "loss_box_dn_0": 0.90868, "loss_cls_dn_1": 0.32615, "loss_box_dn_1": 1.04162, "loss_cls_dn_2": 0.33974, "loss_box_dn_2": 0.96605, "loss_cls_dn_3": 0.36571, "loss_box_dn_3": 0.97395, "loss_cls_dn_4": 0.36092, "loss_box_dn_4": 1.00058, "loss_cls_dn_5": 0.37022, "loss_box_dn_5": 1.02518, "loss_dense_depth": 1.28512, "loss": 34.68344, "grad_norm": 49.21441, "time": 1.48415} -{"mode": "train", "epoch": 1, "iter": 48, "lr": 0.00012, "memory": 49163, "data_time": 0.07448, "loss_cls_0": 1.03245, "loss_box_0": 2.03073, "loss_cns_0": 0.60833, "loss_yns_0": 0.16492, "loss_cls_1": 1.07375, "loss_box_1": 2.47807, "loss_cns_1": 0.57551, "loss_yns_1": 0.16349, "loss_cls_2": 1.08542, "loss_box_2": 2.40629, "loss_cns_2": 0.59828, "loss_yns_2": 0.16893, "loss_cls_3": 1.11035, "loss_box_3": 2.43571, "loss_cns_3": 0.59501, "loss_yns_3": 0.16749, "loss_cls_4": 1.10669, "loss_box_4": 2.41175, "loss_cns_4": 0.60077, "loss_yns_4": 0.17273, "loss_cls_5": 1.12291, "loss_box_5": 2.40565, "loss_cns_5": 0.60723, "loss_yns_5": 0.16319, "loss_cls_dn_0": 0.40472, "loss_box_dn_0": 0.91613, "loss_cls_dn_1": 0.32248, "loss_box_dn_1": 1.16261, "loss_cls_dn_2": 0.35116, "loss_box_dn_2": 1.0665, "loss_cls_dn_3": 0.36937, "loss_box_dn_3": 1.07359, "loss_cls_dn_4": 0.35674, "loss_box_dn_4": 1.08319, "loss_cls_dn_5": 0.36942, "loss_box_dn_5": 1.10044, "loss_dense_depth": 1.17089, "loss": 35.03291, "grad_norm": 40.51993, "time": 1.50205} -{"mode": "train", "epoch": 1, "iter": 49, "lr": 0.00012, "memory": 49163, "data_time": 0.07466, "loss_cls_0": 1.04943, "loss_box_0": 2.02202, "loss_cns_0": 0.60523, "loss_yns_0": 0.16719, "loss_cls_1": 1.06053, "loss_box_1": 2.38317, "loss_cns_1": 0.58436, "loss_yns_1": 0.1678, "loss_cls_2": 1.09967, "loss_box_2": 2.34018, "loss_cns_2": 0.60668, "loss_yns_2": 0.16704, "loss_cls_3": 1.11184, "loss_box_3": 2.35289, "loss_cns_3": 0.60942, "loss_yns_3": 0.16535, "loss_cls_4": 1.10899, "loss_box_4": 2.3037, "loss_cns_4": 0.61627, "loss_yns_4": 0.16277, "loss_cls_5": 1.12652, "loss_box_5": 2.29619, "loss_cns_5": 0.61661, "loss_yns_5": 0.16262, "loss_cls_dn_0": 0.40708, "loss_box_dn_0": 0.92091, "loss_cls_dn_1": 0.30051, "loss_box_dn_1": 1.2119, "loss_cls_dn_2": 0.34483, "loss_box_dn_2": 1.1299, "loss_cls_dn_3": 0.34748, "loss_box_dn_3": 1.13335, "loss_cls_dn_4": 0.34281, "loss_box_dn_4": 1.12967, "loss_cls_dn_5": 0.3676, "loss_box_dn_5": 1.14722, "loss_dense_depth": 1.22956, "loss": 34.89928, "grad_norm": 48.39806, "time": 1.51119} -{"mode": "train", "epoch": 1, "iter": 50, "lr": 0.00012, "memory": 49163, "data_time": 0.07422, "loss_cls_0": 1.01858, "loss_box_0": 1.99596, "loss_cns_0": 0.60704, "loss_yns_0": 0.16809, "loss_cls_1": 1.06376, "loss_box_1": 2.47718, "loss_cns_1": 0.58537, "loss_yns_1": 0.16833, "loss_cls_2": 1.09814, "loss_box_2": 2.37797, "loss_cns_2": 0.61197, "loss_yns_2": 0.16524, "loss_cls_3": 1.13286, "loss_box_3": 2.35344, "loss_cns_3": 0.62168, "loss_yns_3": 0.1638, "loss_cls_4": 1.13157, "loss_box_4": 2.34743, "loss_cns_4": 0.62376, "loss_yns_4": 0.16672, "loss_cls_5": 1.13814, "loss_box_5": 2.36811, "loss_cns_5": 0.61732, "loss_yns_5": 0.1625, "loss_cls_dn_0": 0.40026, "loss_box_dn_0": 0.9371, "loss_cls_dn_1": 0.28829, "loss_box_dn_1": 1.15646, "loss_cls_dn_2": 0.33509, "loss_box_dn_2": 1.07854, "loss_cls_dn_3": 0.32452, "loss_box_dn_3": 1.06992, "loss_cls_dn_4": 0.33948, "loss_box_dn_4": 1.06518, "loss_cls_dn_5": 0.37015, "loss_box_dn_5": 1.08586, "loss_dense_depth": 1.18045, "loss": 34.79624, "grad_norm": 49.12504, "time": 1.51016} -{"mode": "train", "epoch": 1, "iter": 51, "lr": 0.00012, "memory": 49163, "data_time": 0.07692, "loss_cls_0": 0.98501, "loss_box_0": 1.92615, "loss_cns_0": 0.61057, "loss_yns_0": 0.16905, "loss_cls_1": 1.07175, "loss_box_1": 2.48048, "loss_cns_1": 0.57987, "loss_yns_1": 0.17064, "loss_cls_2": 1.0836, "loss_box_2": 2.32302, "loss_cns_2": 0.6128, "loss_yns_2": 0.16806, "loss_cls_3": 1.12946, "loss_box_3": 2.30705, "loss_cns_3": 0.62401, "loss_yns_3": 0.16575, "loss_cls_4": 1.11336, "loss_box_4": 2.33712, "loss_cns_4": 0.6262, "loss_yns_4": 0.17142, "loss_cls_5": 1.12361, "loss_box_5": 2.32048, "loss_cns_5": 0.62357, "loss_yns_5": 0.16248, "loss_cls_dn_0": 0.40251, "loss_box_dn_0": 0.92153, "loss_cls_dn_1": 0.26571, "loss_box_dn_1": 1.21658, "loss_cls_dn_2": 0.32061, "loss_box_dn_2": 1.10122, "loss_cls_dn_3": 0.30657, "loss_box_dn_3": 1.07489, "loss_cls_dn_4": 0.33088, "loss_box_dn_4": 1.07173, "loss_cls_dn_5": 0.34828, "loss_box_dn_5": 1.07914, "loss_dense_depth": 1.02717, "loss": 34.35235, "grad_norm": 43.54663, "time": 1.50075} -{"mode": "train", "epoch": 1, "iter": 52, "lr": 0.00012, "memory": 49163, "data_time": 0.07921, "loss_cls_0": 1.01641, "loss_box_0": 1.9355, "loss_cns_0": 0.60223, "loss_yns_0": 0.16819, "loss_cls_1": 1.06893, "loss_box_1": 2.4252, "loss_cns_1": 0.57483, "loss_yns_1": 0.17081, "loss_cls_2": 1.09535, "loss_box_2": 2.27768, "loss_cns_2": 0.60991, "loss_yns_2": 0.1673, "loss_cls_3": 1.11208, "loss_box_3": 2.29419, "loss_cns_3": 0.61688, "loss_yns_3": 0.16723, "loss_cls_4": 1.10548, "loss_box_4": 2.28989, "loss_cns_4": 0.62406, "loss_yns_4": 0.16836, "loss_cls_5": 1.1288, "loss_box_5": 2.24876, "loss_cns_5": 0.62231, "loss_yns_5": 0.1639, "loss_cls_dn_0": 0.4138, "loss_box_dn_0": 0.91508, "loss_cls_dn_1": 0.26511, "loss_box_dn_1": 1.15138, "loss_cls_dn_2": 0.32389, "loss_box_dn_2": 1.05766, "loss_cls_dn_3": 0.31403, "loss_box_dn_3": 1.04657, "loss_cls_dn_4": 0.33067, "loss_box_dn_4": 1.03317, "loss_cls_dn_5": 0.34222, "loss_box_dn_5": 1.03559, "loss_dense_depth": 1.10406, "loss": 33.98752, "grad_norm": 43.01084, "time": 1.49069} -{"mode": "train", "epoch": 1, "iter": 53, "lr": 0.00012, "memory": 49163, "data_time": 0.07876, "loss_cls_0": 1.04591, "loss_box_0": 2.00584, "loss_cns_0": 0.60398, "loss_yns_0": 0.16644, "loss_cls_1": 1.06754, "loss_box_1": 2.47254, "loss_cns_1": 0.57189, "loss_yns_1": 0.16934, "loss_cls_2": 1.09248, "loss_box_2": 2.33435, "loss_cns_2": 0.60792, "loss_yns_2": 0.16805, "loss_cls_3": 1.10897, "loss_box_3": 2.34656, "loss_cns_3": 0.61618, "loss_yns_3": 0.17326, "loss_cls_4": 1.1092, "loss_box_4": 2.30531, "loss_cns_4": 0.61979, "loss_yns_4": 0.16771, "loss_cls_5": 1.15286, "loss_box_5": 2.34704, "loss_cns_5": 0.61872, "loss_yns_5": 0.16494, "loss_cls_dn_0": 0.41361, "loss_box_dn_0": 0.90729, "loss_cls_dn_1": 0.26417, "loss_box_dn_1": 1.11529, "loss_cls_dn_2": 0.32148, "loss_box_dn_2": 1.03883, "loss_cls_dn_3": 0.32088, "loss_box_dn_3": 1.0337, "loss_cls_dn_4": 0.31892, "loss_box_dn_4": 1.02091, "loss_cls_dn_5": 0.34607, "loss_box_dn_5": 1.05903, "loss_dense_depth": 1.04738, "loss": 34.24436, "grad_norm": 48.10125, "time": 1.49394} -{"mode": "train", "epoch": 1, "iter": 54, "lr": 0.00012, "memory": 49163, "data_time": 0.07728, "loss_cls_0": 1.02027, "loss_box_0": 1.9643, "loss_cns_0": 0.61846, "loss_yns_0": 0.16337, "loss_cls_1": 1.04672, "loss_box_1": 2.36281, "loss_cns_1": 0.59995, "loss_yns_1": 0.16406, "loss_cls_2": 1.06513, "loss_box_2": 2.28395, "loss_cns_2": 0.6201, "loss_yns_2": 0.1635, "loss_cls_3": 1.08817, "loss_box_3": 2.29224, "loss_cns_3": 0.61981, "loss_yns_3": 0.16582, "loss_cls_4": 1.10394, "loss_box_4": 2.27567, "loss_cns_4": 0.61727, "loss_yns_4": 0.16303, "loss_cls_5": 1.13174, "loss_box_5": 2.33839, "loss_cns_5": 0.62153, "loss_yns_5": 0.16279, "loss_cls_dn_0": 0.38266, "loss_box_dn_0": 0.9057, "loss_cls_dn_1": 0.25548, "loss_box_dn_1": 1.12949, "loss_cls_dn_2": 0.30133, "loss_box_dn_2": 1.07265, "loss_cls_dn_3": 0.31641, "loss_box_dn_3": 1.07244, "loss_cls_dn_4": 0.30152, "loss_box_dn_4": 1.08677, "loss_cls_dn_5": 0.335, "loss_box_dn_5": 1.13509, "loss_dense_depth": 1.08152, "loss": 34.02909, "grad_norm": 54.74203, "time": 1.50139} -{"mode": "train", "epoch": 1, "iter": 55, "lr": 0.00012, "memory": 49163, "data_time": 0.08516, "loss_cls_0": 1.01962, "loss_box_0": 1.97391, "loss_cns_0": 0.61554, "loss_yns_0": 0.16636, "loss_cls_1": 1.05145, "loss_box_1": 2.28296, "loss_cns_1": 0.60517, "loss_yns_1": 0.16949, "loss_cls_2": 1.07017, "loss_box_2": 2.23776, "loss_cns_2": 0.61712, "loss_yns_2": 0.16641, "loss_cls_3": 1.0821, "loss_box_3": 2.23514, "loss_cns_3": 0.61941, "loss_yns_3": 0.16711, "loss_cls_4": 1.09545, "loss_box_4": 2.22553, "loss_cns_4": 0.62029, "loss_yns_4": 0.16784, "loss_cls_5": 1.09346, "loss_box_5": 2.20209, "loss_cns_5": 0.62252, "loss_yns_5": 0.16486, "loss_cls_dn_0": 0.37797, "loss_box_dn_0": 0.88819, "loss_cls_dn_1": 0.26395, "loss_box_dn_1": 1.02866, "loss_cls_dn_2": 0.30881, "loss_box_dn_2": 0.98498, "loss_cls_dn_3": 0.33453, "loss_box_dn_3": 0.99986, "loss_cls_dn_4": 0.32033, "loss_box_dn_4": 1.02656, "loss_cls_dn_5": 0.36163, "loss_box_dn_5": 1.04961, "loss_dense_depth": 1.04025, "loss": 33.25706, "grad_norm": 45.63651, "time": 1.51379} -{"mode": "train", "epoch": 1, "iter": 56, "lr": 0.00012, "memory": 49163, "data_time": 0.07776, "loss_cls_0": 1.00631, "loss_box_0": 1.92566, "loss_cns_0": 0.61146, "loss_yns_0": 0.16516, "loss_cls_1": 1.04007, "loss_box_1": 2.17545, "loss_cns_1": 0.59561, "loss_yns_1": 0.16742, "loss_cls_2": 1.07814, "loss_box_2": 2.13422, "loss_cns_2": 0.62169, "loss_yns_2": 0.16719, "loss_cls_3": 1.05807, "loss_box_3": 2.11847, "loss_cns_3": 0.62323, "loss_yns_3": 0.16763, "loss_cls_4": 1.0621, "loss_box_4": 2.13279, "loss_cns_4": 0.62178, "loss_yns_4": 0.16947, "loss_cls_5": 1.08793, "loss_box_5": 2.10078, "loss_cns_5": 0.62009, "loss_yns_5": 0.16699, "loss_cls_dn_0": 0.36853, "loss_box_dn_0": 0.88505, "loss_cls_dn_1": 0.25009, "loss_box_dn_1": 1.04712, "loss_cls_dn_2": 0.29038, "loss_box_dn_2": 1.00552, "loss_cls_dn_3": 0.30729, "loss_box_dn_3": 1.0082, "loss_cls_dn_4": 0.3147, "loss_box_dn_4": 1.03377, "loss_cls_dn_5": 0.35324, "loss_box_dn_5": 1.04073, "loss_dense_depth": 1.00357, "loss": 32.52589, "grad_norm": 56.04913, "time": 1.51255} -{"mode": "train", "epoch": 1, "iter": 57, "lr": 0.00012, "memory": 49163, "data_time": 0.07631, "loss_cls_0": 1.0053, "loss_box_0": 1.88699, "loss_cns_0": 0.61582, "loss_yns_0": 0.16434, "loss_cls_1": 1.04423, "loss_box_1": 2.16697, "loss_cns_1": 0.59915, "loss_yns_1": 0.1658, "loss_cls_2": 1.09493, "loss_box_2": 2.14884, "loss_cns_2": 0.62376, "loss_yns_2": 0.1678, "loss_cls_3": 1.1094, "loss_box_3": 2.10899, "loss_cns_3": 0.62314, "loss_yns_3": 0.16886, "loss_cls_4": 1.06074, "loss_box_4": 2.12306, "loss_cns_4": 0.62381, "loss_yns_4": 0.16711, "loss_cls_5": 1.08974, "loss_box_5": 2.12944, "loss_cns_5": 0.62109, "loss_yns_5": 0.17269, "loss_cls_dn_0": 0.3723, "loss_box_dn_0": 0.87449, "loss_cls_dn_1": 0.2397, "loss_box_dn_1": 0.96936, "loss_cls_dn_2": 0.2713, "loss_box_dn_2": 0.94643, "loss_cls_dn_3": 0.27884, "loss_box_dn_3": 0.93353, "loss_cls_dn_4": 0.29842, "loss_box_dn_4": 0.95951, "loss_cls_dn_5": 0.3356, "loss_box_dn_5": 0.97855, "loss_dense_depth": 1.01062, "loss": 32.15067, "grad_norm": 48.72332, "time": 1.49476} -{"mode": "train", "epoch": 1, "iter": 58, "lr": 0.00012, "memory": 49163, "data_time": 0.07714, "loss_cls_0": 0.98104, "loss_box_0": 1.89619, "loss_cns_0": 0.61738, "loss_yns_0": 0.16406, "loss_cls_1": 1.0243, "loss_box_1": 2.12679, "loss_cns_1": 0.5996, "loss_yns_1": 0.1656, "loss_cls_2": 1.0487, "loss_box_2": 2.09735, "loss_cns_2": 0.6269, "loss_yns_2": 0.16871, "loss_cls_3": 1.05394, "loss_box_3": 2.08027, "loss_cns_3": 0.62857, "loss_yns_3": 0.1675, "loss_cls_4": 1.04544, "loss_box_4": 2.09943, "loss_cns_4": 0.63334, "loss_yns_4": 0.16889, "loss_cls_5": 1.09636, "loss_box_5": 2.06925, "loss_cns_5": 0.63272, "loss_yns_5": 0.17072, "loss_cls_dn_0": 0.37521, "loss_box_dn_0": 0.86681, "loss_cls_dn_1": 0.23674, "loss_box_dn_1": 0.96693, "loss_cls_dn_2": 0.26949, "loss_box_dn_2": 0.93107, "loss_cls_dn_3": 0.2844, "loss_box_dn_3": 0.92859, "loss_cls_dn_4": 0.29334, "loss_box_dn_4": 0.962, "loss_cls_dn_5": 0.3211, "loss_box_dn_5": 0.96254, "loss_dense_depth": 0.97868, "loss": 31.73996, "grad_norm": 58.31791, "time": 1.48546} -{"mode": "train", "epoch": 1, "iter": 59, "lr": 0.00012, "memory": 49163, "data_time": 0.07693, "loss_cls_0": 0.96305, "loss_box_0": 1.91805, "loss_cns_0": 0.61544, "loss_yns_0": 0.16286, "loss_cls_1": 1.02014, "loss_box_1": 2.15079, "loss_cns_1": 0.59694, "loss_yns_1": 0.16462, "loss_cls_2": 1.05432, "loss_box_2": 2.10078, "loss_cns_2": 0.63236, "loss_yns_2": 0.16695, "loss_cls_3": 1.05218, "loss_box_3": 2.10931, "loss_cns_3": 0.63503, "loss_yns_3": 0.16564, "loss_cls_4": 1.06504, "loss_box_4": 2.08663, "loss_cns_4": 0.63979, "loss_yns_4": 0.16378, "loss_cls_5": 1.09043, "loss_box_5": 2.07178, "loss_cns_5": 0.63462, "loss_yns_5": 0.16424, "loss_cls_dn_0": 0.37401, "loss_box_dn_0": 0.86108, "loss_cls_dn_1": 0.2478, "loss_box_dn_1": 0.97365, "loss_cls_dn_2": 0.27602, "loss_box_dn_2": 0.92116, "loss_cls_dn_3": 0.30679, "loss_box_dn_3": 0.93305, "loss_cls_dn_4": 0.30681, "loss_box_dn_4": 0.94336, "loss_cls_dn_5": 0.3258, "loss_box_dn_5": 0.94614, "loss_dense_depth": 0.96045, "loss": 31.80085, "grad_norm": 43.8469, "time": 1.48269} -{"mode": "train", "epoch": 1, "iter": 60, "lr": 0.00012, "memory": 49163, "data_time": 0.08662, "loss_cls_0": 0.97475, "loss_box_0": 1.87468, "loss_cns_0": 0.61928, "loss_yns_0": 0.16028, "loss_cls_1": 1.0183, "loss_box_1": 2.15757, "loss_cns_1": 0.60434, "loss_yns_1": 0.16382, "loss_cls_2": 1.06208, "loss_box_2": 2.10585, "loss_cns_2": 0.63408, "loss_yns_2": 0.16399, "loss_cls_3": 1.06223, "loss_box_3": 2.11024, "loss_cns_3": 0.63674, "loss_yns_3": 0.16417, "loss_cls_4": 1.0867, "loss_box_4": 2.07848, "loss_cns_4": 0.63839, "loss_yns_4": 0.16625, "loss_cls_5": 1.07533, "loss_box_5": 2.13357, "loss_cns_5": 0.63798, "loss_yns_5": 0.16759, "loss_cls_dn_0": 0.36339, "loss_box_dn_0": 0.86362, "loss_cls_dn_1": 0.23961, "loss_box_dn_1": 0.9919, "loss_cls_dn_2": 0.26608, "loss_box_dn_2": 0.93344, "loss_cls_dn_3": 0.29973, "loss_box_dn_3": 0.94254, "loss_cls_dn_4": 0.29925, "loss_box_dn_4": 0.94354, "loss_cls_dn_5": 0.34053, "loss_box_dn_5": 0.97486, "loss_dense_depth": 0.94386, "loss": 31.89902, "grad_norm": 56.05843, "time": 1.49068} -{"mode": "train", "epoch": 1, "iter": 61, "lr": 0.00012, "memory": 49163, "data_time": 0.08924, "loss_cls_0": 1.00418, "loss_box_0": 1.86366, "loss_cns_0": 0.61576, "loss_yns_0": 0.16033, "loss_cls_1": 1.02481, "loss_box_1": 2.16165, "loss_cns_1": 0.60664, "loss_yns_1": 0.16372, "loss_cls_2": 1.07167, "loss_box_2": 2.10515, "loss_cns_2": 0.62813, "loss_yns_2": 0.16366, "loss_cls_3": 1.06542, "loss_box_3": 2.09431, "loss_cns_3": 0.62726, "loss_yns_3": 0.1633, "loss_cls_4": 1.09903, "loss_box_4": 2.07188, "loss_cns_4": 0.63523, "loss_yns_4": 0.16673, "loss_cls_5": 1.08884, "loss_box_5": 2.11653, "loss_cns_5": 0.63739, "loss_yns_5": 0.16641, "loss_cls_dn_0": 0.35497, "loss_box_dn_0": 0.85829, "loss_cls_dn_1": 0.22859, "loss_box_dn_1": 1.06809, "loss_cls_dn_2": 0.25237, "loss_box_dn_2": 1.00021, "loss_cls_dn_3": 0.27042, "loss_box_dn_3": 1.00367, "loss_cls_dn_4": 0.27097, "loss_box_dn_4": 1.00564, "loss_cls_dn_5": 0.33489, "loss_box_dn_5": 1.03553, "loss_dense_depth": 0.95452, "loss": 32.13984, "grad_norm": 56.70023, "time": 1.55554} -{"mode": "train", "epoch": 1, "iter": 62, "lr": 0.00012, "memory": 49163, "data_time": 0.23841, "loss_cls_0": 1.00278, "loss_box_0": 1.8681, "loss_cns_0": 0.61576, "loss_yns_0": 0.15939, "loss_cls_1": 1.03828, "loss_box_1": 2.18199, "loss_cns_1": 0.61054, "loss_yns_1": 0.16193, "loss_cls_2": 1.04912, "loss_box_2": 2.13038, "loss_cns_2": 0.6302, "loss_yns_2": 0.16228, "loss_cls_3": 1.08331, "loss_box_3": 2.13548, "loss_cns_3": 0.62601, "loss_yns_3": 0.16393, "loss_cls_4": 1.13291, "loss_box_4": 2.10303, "loss_cns_4": 0.64017, "loss_yns_4": 0.1642, "loss_cls_5": 1.07482, "loss_box_5": 2.08497, "loss_cns_5": 0.63622, "loss_yns_5": 0.1618, "loss_cls_dn_0": 0.35384, "loss_box_dn_0": 0.85923, "loss_cls_dn_1": 0.23263, "loss_box_dn_1": 1.03902, "loss_cls_dn_2": 0.256, "loss_box_dn_2": 0.97534, "loss_cls_dn_3": 0.26505, "loss_box_dn_3": 0.98699, "loss_cls_dn_4": 0.2771, "loss_box_dn_4": 0.98654, "loss_cls_dn_5": 0.33969, "loss_box_dn_5": 1.00204, "loss_dense_depth": 0.98559, "loss": 32.17665, "grad_norm": 47.45448, "time": 1.63209} -{"mode": "train", "epoch": 1, "iter": 63, "lr": 0.00012, "memory": 49163, "data_time": 0.0737, "loss_cls_0": 0.97983, "loss_box_0": 1.85996, "loss_cns_0": 0.61029, "loss_yns_0": 0.15936, "loss_cls_1": 1.01107, "loss_box_1": 2.19573, "loss_cns_1": 0.60877, "loss_yns_1": 0.1629, "loss_cls_2": 1.08742, "loss_box_2": 2.13919, "loss_cns_2": 0.6346, "loss_yns_2": 0.1637, "loss_cls_3": 1.06833, "loss_box_3": 2.10646, "loss_cns_3": 0.63876, "loss_yns_3": 0.16395, "loss_cls_4": 1.12491, "loss_box_4": 2.10232, "loss_cns_4": 0.64639, "loss_yns_4": 0.16187, "loss_cls_5": 1.09005, "loss_box_5": 2.0916, "loss_cns_5": 0.64327, "loss_yns_5": 0.16254, "loss_cls_dn_0": 0.36316, "loss_box_dn_0": 0.85881, "loss_cls_dn_1": 0.2302, "loss_box_dn_1": 1.05553, "loss_cls_dn_2": 0.26042, "loss_box_dn_2": 0.97123, "loss_cls_dn_3": 0.2697, "loss_box_dn_3": 0.9689, "loss_cls_dn_4": 0.30987, "loss_box_dn_4": 0.97757, "loss_cls_dn_5": 0.32014, "loss_box_dn_5": 0.99088, "loss_dense_depth": 0.98059, "loss": 32.17029, "grad_norm": 44.30724, "time": 1.51431} -{"mode": "train", "epoch": 1, "iter": 64, "lr": 0.00013, "memory": 49163, "data_time": 0.07571, "loss_cls_0": 0.95549, "loss_box_0": 1.84322, "loss_cns_0": 0.61478, "loss_yns_0": 0.15936, "loss_cls_1": 0.98433, "loss_box_1": 2.15508, "loss_cns_1": 0.6106, "loss_yns_1": 0.1617, "loss_cls_2": 1.04646, "loss_box_2": 2.11439, "loss_cns_2": 0.63318, "loss_yns_2": 0.1649, "loss_cls_3": 1.05779, "loss_box_3": 2.08056, "loss_cns_3": 0.64152, "loss_yns_3": 0.16545, "loss_cls_4": 1.08883, "loss_box_4": 2.09932, "loss_cns_4": 0.64721, "loss_yns_4": 0.16451, "loss_cls_5": 1.12535, "loss_box_5": 2.08699, "loss_cns_5": 0.64378, "loss_yns_5": 0.16453, "loss_cls_dn_0": 0.35259, "loss_box_dn_0": 0.85417, "loss_cls_dn_1": 0.22622, "loss_box_dn_1": 1.04777, "loss_cls_dn_2": 0.23819, "loss_box_dn_2": 0.94772, "loss_cls_dn_3": 0.25212, "loss_box_dn_3": 0.92995, "loss_cls_dn_4": 0.27079, "loss_box_dn_4": 0.93534, "loss_cls_dn_5": 0.2713, "loss_box_dn_5": 0.94611, "loss_dense_depth": 0.94592, "loss": 31.6275, "grad_norm": 36.89016, "time": 1.62271} -{"mode": "train", "epoch": 1, "iter": 65, "lr": 0.00013, "memory": 49163, "data_time": 0.09083, "loss_cls_0": 0.96485, "loss_box_0": 1.85803, "loss_cns_0": 0.61871, "loss_yns_0": 0.16104, "loss_cls_1": 0.98921, "loss_box_1": 2.11722, "loss_cns_1": 0.60631, "loss_yns_1": 0.16064, "loss_cls_2": 1.07806, "loss_box_2": 2.06623, "loss_cns_2": 0.63513, "loss_yns_2": 0.16287, "loss_cls_3": 1.07263, "loss_box_3": 2.06075, "loss_cns_3": 0.6426, "loss_yns_3": 0.16422, "loss_cls_4": 1.09265, "loss_box_4": 2.02891, "loss_cns_4": 0.65008, "loss_yns_4": 0.16514, "loss_cls_5": 1.11708, "loss_box_5": 2.04295, "loss_cns_5": 0.64697, "loss_yns_5": 0.16254, "loss_cls_dn_0": 0.34039, "loss_box_dn_0": 0.85667, "loss_cls_dn_1": 0.22141, "loss_box_dn_1": 1.05042, "loss_cls_dn_2": 0.23183, "loss_box_dn_2": 0.94371, "loss_cls_dn_3": 0.23876, "loss_box_dn_3": 0.93223, "loss_cls_dn_4": 0.23468, "loss_box_dn_4": 0.9147, "loss_cls_dn_5": 0.26175, "loss_box_dn_5": 0.9253, "loss_dense_depth": 0.91131, "loss": 31.32796, "grad_norm": 57.27365, "time": 1.50856} -{"mode": "train", "epoch": 1, "iter": 66, "lr": 0.00013, "memory": 49163, "data_time": 0.07369, "loss_cls_0": 0.96602, "loss_box_0": 1.83107, "loss_cns_0": 0.61726, "loss_yns_0": 0.16164, "loss_cls_1": 0.97526, "loss_box_1": 2.05994, "loss_cns_1": 0.59713, "loss_yns_1": 0.16022, "loss_cls_2": 1.01451, "loss_box_2": 2.05142, "loss_cns_2": 0.6357, "loss_yns_2": 0.1633, "loss_cls_3": 1.0684, "loss_box_3": 2.0549, "loss_cns_3": 0.6389, "loss_yns_3": 0.16369, "loss_cls_4": 1.0949, "loss_box_4": 2.00423, "loss_cns_4": 0.64759, "loss_yns_4": 0.16544, "loss_cls_5": 1.1253, "loss_box_5": 2.0147, "loss_cns_5": 0.64532, "loss_yns_5": 0.16189, "loss_cls_dn_0": 0.32442, "loss_box_dn_0": 0.839, "loss_cls_dn_1": 0.20661, "loss_box_dn_1": 1.0407, "loss_cls_dn_2": 0.21802, "loss_box_dn_2": 0.94852, "loss_cls_dn_3": 0.24254, "loss_box_dn_3": 0.94243, "loss_cls_dn_4": 0.248, "loss_box_dn_4": 0.92171, "loss_cls_dn_5": 0.28727, "loss_box_dn_5": 0.9354, "loss_dense_depth": 0.92897, "loss": 31.10233, "grad_norm": 62.57196, "time": 1.49619} -{"mode": "train", "epoch": 1, "iter": 67, "lr": 0.00013, "memory": 49163, "data_time": 0.07797, "loss_cls_0": 0.97459, "loss_box_0": 1.83631, "loss_cns_0": 0.61417, "loss_yns_0": 0.16267, "loss_cls_1": 0.99859, "loss_box_1": 2.02717, "loss_cns_1": 0.58662, "loss_yns_1": 0.15651, "loss_cls_2": 1.00631, "loss_box_2": 2.05702, "loss_cns_2": 0.62824, "loss_yns_2": 0.1659, "loss_cls_3": 1.0336, "loss_box_3": 2.0436, "loss_cns_3": 0.63287, "loss_yns_3": 0.16156, "loss_cls_4": 1.05285, "loss_box_4": 2.00549, "loss_cns_4": 0.63749, "loss_yns_4": 0.16249, "loss_cls_5": 1.08972, "loss_box_5": 2.0009, "loss_cns_5": 0.63766, "loss_yns_5": 0.16304, "loss_cls_dn_0": 0.32392, "loss_box_dn_0": 0.83933, "loss_cls_dn_1": 0.20509, "loss_box_dn_1": 0.97428, "loss_cls_dn_2": 0.22105, "loss_box_dn_2": 0.91191, "loss_cls_dn_3": 0.23195, "loss_box_dn_3": 0.90183, "loss_cls_dn_4": 0.23441, "loss_box_dn_4": 0.90041, "loss_cls_dn_5": 0.28618, "loss_box_dn_5": 0.91946, "loss_dense_depth": 0.93898, "loss": 30.72416, "grad_norm": 38.0973, "time": 1.48214} -{"mode": "train", "epoch": 1, "iter": 68, "lr": 0.00013, "memory": 49163, "data_time": 0.07658, "loss_cls_0": 0.96474, "loss_box_0": 1.84433, "loss_cns_0": 0.61904, "loss_yns_0": 0.1615, "loss_cls_1": 0.98528, "loss_box_1": 1.99989, "loss_cns_1": 0.60628, "loss_yns_1": 0.16508, "loss_cls_2": 1.08258, "loss_box_2": 1.94691, "loss_cns_2": 0.6282, "loss_yns_2": 0.16468, "loss_cls_3": 1.29544, "loss_box_3": 1.93854, "loss_cns_3": 0.63381, "loss_yns_3": 0.16312, "loss_cls_4": 1.51974, "loss_box_4": 1.9445, "loss_cns_4": 0.63853, "loss_yns_4": 0.16251, "loss_cls_5": 1.08837, "loss_box_5": 1.92075, "loss_cns_5": 0.64101, "loss_yns_5": 0.16508, "loss_cls_dn_0": 0.3267, "loss_box_dn_0": 0.85468, "loss_cls_dn_1": 0.20993, "loss_box_dn_1": 0.93749, "loss_cls_dn_2": 0.22158, "loss_box_dn_2": 0.89466, "loss_cls_dn_3": 0.227, "loss_box_dn_3": 0.90138, "loss_cls_dn_4": 0.2358, "loss_box_dn_4": 0.94055, "loss_cls_dn_5": 0.2566, "loss_box_dn_5": 0.96516, "loss_dense_depth": 0.92668, "loss": 31.17812, "grad_norm": 150.59735, "time": 1.49189} -{"mode": "train", "epoch": 1, "iter": 69, "lr": 0.00013, "memory": 49163, "data_time": 0.07265, "loss_cls_0": 0.93104, "loss_box_0": 1.8111, "loss_cns_0": 0.61759, "loss_yns_0": 0.16255, "loss_cls_1": 0.97205, "loss_box_1": 1.98917, "loss_cns_1": 0.61887, "loss_yns_1": 0.16539, "loss_cls_2": 1.02669, "loss_box_2": 1.93285, "loss_cns_2": 0.63629, "loss_yns_2": 0.15947, "loss_cls_3": 1.09357, "loss_box_3": 1.95733, "loss_cns_3": 0.64082, "loss_yns_3": 0.16014, "loss_cls_4": 1.14924, "loss_box_4": 1.99503, "loss_cns_4": 0.64582, "loss_yns_4": 0.16169, "loss_cls_5": 1.10453, "loss_box_5": 1.997, "loss_cns_5": 0.65126, "loss_yns_5": 0.16302, "loss_cls_dn_0": 0.32191, "loss_box_dn_0": 0.84827, "loss_cls_dn_1": 0.21076, "loss_box_dn_1": 0.96701, "loss_cls_dn_2": 0.22029, "loss_box_dn_2": 0.93836, "loss_cls_dn_3": 0.22574, "loss_box_dn_3": 0.97338, "loss_cls_dn_4": 0.2416, "loss_box_dn_4": 1.03098, "loss_cls_dn_5": 0.28335, "loss_box_dn_5": 1.06147, "loss_dense_depth": 1.0247, "loss": 31.09034, "grad_norm": 75.73018, "time": 1.49089} -{"mode": "train", "epoch": 1, "iter": 70, "lr": 0.00013, "memory": 49163, "data_time": 0.07275, "loss_cls_0": 0.94366, "loss_box_0": 1.78687, "loss_cns_0": 0.61137, "loss_yns_0": 0.16115, "loss_cls_1": 0.98503, "loss_box_1": 2.03897, "loss_cns_1": 0.613, "loss_yns_1": 0.16402, "loss_cls_2": 1.04143, "loss_box_2": 1.99326, "loss_cns_2": 0.63071, "loss_yns_2": 0.15927, "loss_cls_3": 1.0878, "loss_box_3": 1.97817, "loss_cns_3": 0.63852, "loss_yns_3": 0.16034, "loss_cls_4": 1.1106, "loss_box_4": 2.00944, "loss_cns_4": 0.64396, "loss_yns_4": 0.15973, "loss_cls_5": 1.09544, "loss_box_5": 2.03192, "loss_cns_5": 0.6477, "loss_yns_5": 0.16157, "loss_cls_dn_0": 0.32962, "loss_box_dn_0": 0.84438, "loss_cls_dn_1": 0.22127, "loss_box_dn_1": 0.99113, "loss_cls_dn_2": 0.24097, "loss_box_dn_2": 0.96014, "loss_cls_dn_3": 0.2548, "loss_box_dn_3": 0.99394, "loss_cls_dn_4": 0.28425, "loss_box_dn_4": 1.05172, "loss_cls_dn_5": 0.3109, "loss_box_dn_5": 1.09065, "loss_dense_depth": 0.92514, "loss": 31.35286, "grad_norm": 59.48146, "time": 1.52218} -{"mode": "train", "epoch": 1, "iter": 71, "lr": 0.00013, "memory": 49163, "data_time": 0.07544, "loss_cls_0": 0.93273, "loss_box_0": 1.79172, "loss_cns_0": 0.61546, "loss_yns_0": 0.15702, "loss_cls_1": 0.98111, "loss_box_1": 2.13412, "loss_cns_1": 0.60829, "loss_yns_1": 0.16109, "loss_cls_2": 1.01924, "loss_box_2": 2.05545, "loss_cns_2": 0.63379, "loss_yns_2": 0.15886, "loss_cls_3": 1.0516, "loss_box_3": 1.99165, "loss_cns_3": 0.64149, "loss_yns_3": 0.15973, "loss_cls_4": 1.11566, "loss_box_4": 2.00305, "loss_cns_4": 0.64762, "loss_yns_4": 0.15894, "loss_cls_5": 1.07593, "loss_box_5": 2.01356, "loss_cns_5": 0.6456, "loss_yns_5": 0.15926, "loss_cls_dn_0": 0.31185, "loss_box_dn_0": 0.84594, "loss_cls_dn_1": 0.21656, "loss_box_dn_1": 1.05705, "loss_cls_dn_2": 0.23365, "loss_box_dn_2": 1.01081, "loss_cls_dn_3": 0.24812, "loss_box_dn_3": 1.02027, "loss_cls_dn_4": 0.28653, "loss_box_dn_4": 1.06165, "loss_cls_dn_5": 0.27289, "loss_box_dn_5": 1.07897, "loss_dense_depth": 0.92031, "loss": 31.47758, "grad_norm": 46.68696, "time": 1.48041} -{"mode": "train", "epoch": 1, "iter": 72, "lr": 0.00013, "memory": 49163, "data_time": 0.07562, "loss_cls_0": 0.97114, "loss_box_0": 1.79027, "loss_cns_0": 0.61812, "loss_yns_0": 0.15688, "loss_cls_1": 1.01907, "loss_box_1": 2.11282, "loss_cns_1": 0.61236, "loss_yns_1": 0.16062, "loss_cls_2": 1.01319, "loss_box_2": 2.03987, "loss_cns_2": 0.63612, "loss_yns_2": 0.16055, "loss_cls_3": 1.02912, "loss_box_3": 1.96934, "loss_cns_3": 0.64544, "loss_yns_3": 0.15857, "loss_cls_4": 1.057, "loss_box_4": 1.96849, "loss_cns_4": 0.64675, "loss_yns_4": 0.15875, "loss_cls_5": 1.08829, "loss_box_5": 1.9543, "loss_cns_5": 0.64436, "loss_yns_5": 0.16774, "loss_cls_dn_0": 0.30066, "loss_box_dn_0": 0.84257, "loss_cls_dn_1": 0.21968, "loss_box_dn_1": 0.92214, "loss_cls_dn_2": 0.22794, "loss_box_dn_2": 0.87407, "loss_cls_dn_3": 0.23033, "loss_box_dn_3": 0.86909, "loss_cls_dn_4": 0.25494, "loss_box_dn_4": 0.88747, "loss_cls_dn_5": 0.25131, "loss_box_dn_5": 0.88214, "loss_dense_depth": 0.93659, "loss": 30.47809, "grad_norm": 57.89616, "time": 1.49059} -{"mode": "train", "epoch": 1, "iter": 73, "lr": 0.00013, "memory": 49163, "data_time": 0.07786, "loss_cls_0": 0.98091, "loss_box_0": 1.81351, "loss_cns_0": 0.61008, "loss_yns_0": 0.1573, "loss_cls_1": 1.00873, "loss_box_1": 2.11523, "loss_cns_1": 0.61446, "loss_yns_1": 0.16098, "loss_cls_2": 1.02511, "loss_box_2": 2.06733, "loss_cns_2": 0.63511, "loss_yns_2": 0.16006, "loss_cls_3": 1.04373, "loss_box_3": 2.00395, "loss_cns_3": 0.64293, "loss_yns_3": 0.15873, "loss_cls_4": 1.04817, "loss_box_4": 1.99721, "loss_cns_4": 0.64567, "loss_yns_4": 0.15823, "loss_cls_5": 1.05212, "loss_box_5": 1.99574, "loss_cns_5": 0.64519, "loss_yns_5": 0.16812, "loss_cls_dn_0": 0.29949, "loss_box_dn_0": 0.85188, "loss_cls_dn_1": 0.21658, "loss_box_dn_1": 0.93633, "loss_cls_dn_2": 0.22733, "loss_box_dn_2": 0.89041, "loss_cls_dn_3": 0.23257, "loss_box_dn_3": 0.87901, "loss_cls_dn_4": 0.25075, "loss_box_dn_4": 0.88194, "loss_cls_dn_5": 0.25348, "loss_box_dn_5": 0.88158, "loss_dense_depth": 0.90159, "loss": 30.61153, "grad_norm": 52.25419, "time": 1.49102} -{"mode": "train", "epoch": 1, "iter": 74, "lr": 0.00013, "memory": 49163, "data_time": 0.07632, "loss_cls_0": 0.95418, "loss_box_0": 1.81617, "loss_cns_0": 0.60725, "loss_yns_0": 0.15801, "loss_cls_1": 0.99935, "loss_box_1": 2.10866, "loss_cns_1": 0.61534, "loss_yns_1": 0.16053, "loss_cls_2": 1.03698, "loss_box_2": 2.04061, "loss_cns_2": 0.63363, "loss_yns_2": 0.16174, "loss_cls_3": 1.04268, "loss_box_3": 2.02782, "loss_cns_3": 0.64107, "loss_yns_3": 0.16131, "loss_cls_4": 1.05038, "loss_box_4": 2.01044, "loss_cns_4": 0.64481, "loss_yns_4": 0.16012, "loss_cls_5": 1.04961, "loss_box_5": 2.00926, "loss_cns_5": 0.64502, "loss_yns_5": 0.16438, "loss_cls_dn_0": 0.30726, "loss_box_dn_0": 0.85554, "loss_cls_dn_1": 0.22068, "loss_box_dn_1": 0.95724, "loss_cls_dn_2": 0.22845, "loss_box_dn_2": 0.90262, "loss_cls_dn_3": 0.22537, "loss_box_dn_3": 0.89529, "loss_cls_dn_4": 0.2339, "loss_box_dn_4": 0.88767, "loss_cls_dn_5": 0.24667, "loss_box_dn_5": 0.89211, "loss_dense_depth": 0.91128, "loss": 30.66343, "grad_norm": 44.93022, "time": 1.50286} -{"mode": "train", "epoch": 1, "iter": 75, "lr": 0.00013, "memory": 49163, "data_time": 0.08052, "loss_cls_0": 0.95797, "loss_box_0": 1.83139, "loss_cns_0": 0.60548, "loss_yns_0": 0.15904, "loss_cls_1": 1.01762, "loss_box_1": 2.0849, "loss_cns_1": 0.62043, "loss_yns_1": 0.1631, "loss_cls_2": 1.0397, "loss_box_2": 2.00092, "loss_cns_2": 0.64071, "loss_yns_2": 0.16142, "loss_cls_3": 1.04792, "loss_box_3": 2.00604, "loss_cns_3": 0.6473, "loss_yns_3": 0.16381, "loss_cls_4": 1.1024, "loss_box_4": 1.9767, "loss_cns_4": 0.6509, "loss_yns_4": 0.16311, "loss_cls_5": 1.05777, "loss_box_5": 1.96662, "loss_cns_5": 0.64933, "loss_yns_5": 0.1617, "loss_cls_dn_0": 0.30549, "loss_box_dn_0": 0.8461, "loss_cls_dn_1": 0.22007, "loss_box_dn_1": 0.93717, "loss_cls_dn_2": 0.2241, "loss_box_dn_2": 0.89089, "loss_cls_dn_3": 0.21753, "loss_box_dn_3": 0.88637, "loss_cls_dn_4": 0.22569, "loss_box_dn_4": 0.87654, "loss_cls_dn_5": 0.23416, "loss_box_dn_5": 0.88089, "loss_dense_depth": 0.87751, "loss": 30.49879, "grad_norm": 47.62163, "time": 1.48821} -{"mode": "train", "epoch": 1, "iter": 76, "lr": 0.00013, "memory": 49163, "data_time": 0.07576, "loss_cls_0": 0.96213, "loss_box_0": 1.83229, "loss_cns_0": 0.61267, "loss_yns_0": 0.16267, "loss_cls_1": 1.02588, "loss_box_1": 2.10399, "loss_cns_1": 0.62136, "loss_yns_1": 0.1684, "loss_cls_2": 1.03802, "loss_box_2": 2.02741, "loss_cns_2": 0.64423, "loss_yns_2": 0.16303, "loss_cls_3": 1.05968, "loss_box_3": 2.0497, "loss_cns_3": 0.64921, "loss_yns_3": 0.16579, "loss_cls_4": 1.10863, "loss_box_4": 2.01751, "loss_cns_4": 0.65443, "loss_yns_4": 0.16524, "loss_cls_5": 1.06791, "loss_box_5": 2.01563, "loss_cns_5": 0.65313, "loss_yns_5": 0.17351, "loss_cls_dn_0": 0.29473, "loss_box_dn_0": 0.84502, "loss_cls_dn_1": 0.21567, "loss_box_dn_1": 0.88486, "loss_cls_dn_2": 0.22229, "loss_box_dn_2": 0.85215, "loss_cls_dn_3": 0.2168, "loss_box_dn_3": 0.85612, "loss_cls_dn_4": 0.22614, "loss_box_dn_4": 0.85244, "loss_cls_dn_5": 0.23116, "loss_box_dn_5": 0.86263, "loss_dense_depth": 0.8836, "loss": 30.58604, "grad_norm": 44.4542, "time": 1.50482} -{"mode": "train", "epoch": 1, "iter": 77, "lr": 0.00013, "memory": 49163, "data_time": 0.07564, "loss_cls_0": 0.96337, "loss_box_0": 1.89949, "loss_cns_0": 0.61034, "loss_yns_0": 0.16273, "loss_cls_1": 1.01412, "loss_box_1": 2.06216, "loss_cns_1": 0.62864, "loss_yns_1": 0.16643, "loss_cls_2": 1.05783, "loss_box_2": 2.02357, "loss_cns_2": 0.6468, "loss_yns_2": 0.1628, "loss_cls_3": 1.07681, "loss_box_3": 2.03939, "loss_cns_3": 0.64907, "loss_yns_3": 0.16422, "loss_cls_4": 1.0691, "loss_box_4": 1.99933, "loss_cns_4": 0.65773, "loss_yns_4": 0.1646, "loss_cls_5": 1.06383, "loss_box_5": 2.01098, "loss_cns_5": 0.65523, "loss_yns_5": 0.18132, "loss_cls_dn_0": 0.28174, "loss_box_dn_0": 0.83996, "loss_cls_dn_1": 0.21785, "loss_box_dn_1": 0.82581, "loss_cls_dn_2": 0.22471, "loss_box_dn_2": 0.80996, "loss_cls_dn_3": 0.22365, "loss_box_dn_3": 0.81667, "loss_cls_dn_4": 0.23045, "loss_box_dn_4": 0.81784, "loss_cls_dn_5": 0.23181, "loss_box_dn_5": 0.83663, "loss_dense_depth": 0.87568, "loss": 30.36264, "grad_norm": 43.17088, "time": 1.48859} -{"mode": "train", "epoch": 1, "iter": 78, "lr": 0.00013, "memory": 49163, "data_time": 0.07377, "loss_cls_0": 0.96578, "loss_box_0": 1.91807, "loss_cns_0": 0.61136, "loss_yns_0": 0.16361, "loss_cls_1": 1.00168, "loss_box_1": 2.16031, "loss_cns_1": 0.61523, "loss_yns_1": 0.16487, "loss_cls_2": 1.06925, "loss_box_2": 2.06873, "loss_cns_2": 0.64252, "loss_yns_2": 0.16345, "loss_cls_3": 1.07172, "loss_box_3": 2.053, "loss_cns_3": 0.64617, "loss_yns_3": 0.16532, "loss_cls_4": 1.06815, "loss_box_4": 2.02833, "loss_cns_4": 0.65308, "loss_yns_4": 0.16816, "loss_cls_5": 1.06357, "loss_box_5": 2.05117, "loss_cns_5": 0.64775, "loss_yns_5": 0.17367, "loss_cls_dn_0": 0.2815, "loss_box_dn_0": 0.83848, "loss_cls_dn_1": 0.20859, "loss_box_dn_1": 0.86314, "loss_cls_dn_2": 0.21667, "loss_box_dn_2": 0.82435, "loss_cls_dn_3": 0.21404, "loss_box_dn_3": 0.82717, "loss_cls_dn_4": 0.21901, "loss_box_dn_4": 0.8324, "loss_cls_dn_5": 0.22211, "loss_box_dn_5": 0.85283, "loss_dense_depth": 0.87933, "loss": 30.61457, "grad_norm": 39.16104, "time": 1.48421} -{"mode": "train", "epoch": 1, "iter": 79, "lr": 0.00013, "memory": 49163, "data_time": 0.07622, "loss_cls_0": 0.9348, "loss_box_0": 1.87348, "loss_cns_0": 0.62079, "loss_yns_0": 0.16253, "loss_cls_1": 0.99401, "loss_box_1": 2.0852, "loss_cns_1": 0.62765, "loss_yns_1": 0.16777, "loss_cls_2": 1.04169, "loss_box_2": 2.0499, "loss_cns_2": 0.6427, "loss_yns_2": 0.16277, "loss_cls_3": 1.05371, "loss_box_3": 2.04654, "loss_cns_3": 0.64432, "loss_yns_3": 0.16513, "loss_cls_4": 1.09907, "loss_box_4": 1.99404, "loss_cns_4": 0.65484, "loss_yns_4": 0.16513, "loss_cls_5": 1.05908, "loss_box_5": 1.991, "loss_cns_5": 0.65195, "loss_yns_5": 0.16479, "loss_cls_dn_0": 0.27636, "loss_box_dn_0": 0.83443, "loss_cls_dn_1": 0.20007, "loss_box_dn_1": 0.85997, "loss_cls_dn_2": 0.20475, "loss_box_dn_2": 0.84172, "loss_cls_dn_3": 0.20359, "loss_box_dn_3": 0.85141, "loss_cls_dn_4": 0.22144, "loss_box_dn_4": 0.84289, "loss_cls_dn_5": 0.22145, "loss_box_dn_5": 0.8545, "loss_dense_depth": 0.8287, "loss": 30.2942, "grad_norm": 38.74979, "time": 1.49834} -{"mode": "train", "epoch": 1, "iter": 80, "lr": 0.00013, "memory": 49163, "data_time": 0.08556, "loss_cls_0": 0.93735, "loss_box_0": 1.84295, "loss_cns_0": 0.61717, "loss_yns_0": 0.15882, "loss_cls_1": 0.98074, "loss_box_1": 2.1007, "loss_cns_1": 0.63032, "loss_yns_1": 0.16605, "loss_cls_2": 1.02109, "loss_box_2": 2.07621, "loss_cns_2": 0.64281, "loss_yns_2": 0.16166, "loss_cls_3": 1.04459, "loss_box_3": 2.07336, "loss_cns_3": 0.64988, "loss_yns_3": 0.1643, "loss_cls_4": 1.09973, "loss_box_4": 2.04514, "loss_cns_4": 0.65707, "loss_yns_4": 0.16341, "loss_cls_5": 1.05336, "loss_box_5": 2.0355, "loss_cns_5": 0.65725, "loss_yns_5": 0.17247, "loss_cls_dn_0": 0.272, "loss_box_dn_0": 0.82427, "loss_cls_dn_1": 0.19959, "loss_box_dn_1": 0.86175, "loss_cls_dn_2": 0.20277, "loss_box_dn_2": 0.8522, "loss_cls_dn_3": 0.20571, "loss_box_dn_3": 0.85432, "loss_cls_dn_4": 0.21987, "loss_box_dn_4": 0.84886, "loss_cls_dn_5": 0.21933, "loss_box_dn_5": 0.85638, "loss_dense_depth": 0.8426, "loss": 30.41158, "grad_norm": 47.70745, "time": 1.48946} -{"mode": "train", "epoch": 1, "iter": 81, "lr": 0.00013, "memory": 49163, "data_time": 0.07183, "loss_cls_0": 0.94972, "loss_box_0": 1.83662, "loss_cns_0": 0.61122, "loss_yns_0": 0.15779, "loss_cls_1": 0.98691, "loss_box_1": 2.10083, "loss_cns_1": 0.62515, "loss_yns_1": 0.16325, "loss_cls_2": 1.05229, "loss_box_2": 2.05991, "loss_cns_2": 0.64513, "loss_yns_2": 0.16239, "loss_cls_3": 1.07602, "loss_box_3": 2.04818, "loss_cns_3": 0.64949, "loss_yns_3": 0.16286, "loss_cls_4": 1.05317, "loss_box_4": 2.04596, "loss_cns_4": 0.65554, "loss_yns_4": 0.16448, "loss_cls_5": 1.05451, "loss_box_5": 2.04979, "loss_cns_5": 0.65368, "loss_yns_5": 0.17429, "loss_cls_dn_0": 0.26969, "loss_box_dn_0": 0.82364, "loss_cls_dn_1": 0.20152, "loss_box_dn_1": 0.88542, "loss_cls_dn_2": 0.20276, "loss_box_dn_2": 0.86333, "loss_cls_dn_3": 0.20631, "loss_box_dn_3": 0.85766, "loss_cls_dn_4": 0.21386, "loss_box_dn_4": 0.85853, "loss_cls_dn_5": 0.2178, "loss_box_dn_5": 0.86653, "loss_dense_depth": 0.87146, "loss": 30.47771, "grad_norm": 57.84317, "time": 1.56462} -{"mode": "train", "epoch": 1, "iter": 82, "lr": 0.00013, "memory": 49163, "data_time": 0.14985, "loss_cls_0": 0.96008, "loss_box_0": 1.82526, "loss_cns_0": 0.61089, "loss_yns_0": 0.15578, "loss_cls_1": 1.00839, "loss_box_1": 2.07754, "loss_cns_1": 0.62252, "loss_yns_1": 0.16211, "loss_cls_2": 1.03972, "loss_box_2": 2.04022, "loss_cns_2": 0.64427, "loss_yns_2": 0.16165, "loss_cls_3": 1.07864, "loss_box_3": 2.04471, "loss_cns_3": 0.64337, "loss_yns_3": 0.16053, "loss_cls_4": 1.0662, "loss_box_4": 2.01641, "loss_cns_4": 0.65144, "loss_yns_4": 0.16528, "loss_cls_5": 1.05714, "loss_box_5": 2.0252, "loss_cns_5": 0.64618, "loss_yns_5": 0.1677, "loss_cls_dn_0": 0.26789, "loss_box_dn_0": 0.82917, "loss_cls_dn_1": 0.20124, "loss_box_dn_1": 0.88774, "loss_cls_dn_2": 0.1945, "loss_box_dn_2": 0.85619, "loss_cls_dn_3": 0.19832, "loss_box_dn_3": 0.8542, "loss_cls_dn_4": 0.20748, "loss_box_dn_4": 0.84795, "loss_cls_dn_5": 0.21197, "loss_box_dn_5": 0.85663, "loss_dense_depth": 0.89842, "loss": 30.34295, "grad_norm": 38.31398, "time": 1.54806} -{"mode": "train", "epoch": 1, "iter": 83, "lr": 0.00013, "memory": 49163, "data_time": 0.07643, "loss_cls_0": 0.9529, "loss_box_0": 1.82452, "loss_cns_0": 0.61753, "loss_yns_0": 0.16158, "loss_cls_1": 1.013, "loss_box_1": 2.0441, "loss_cns_1": 0.61751, "loss_yns_1": 0.16409, "loss_cls_2": 1.03867, "loss_box_2": 1.96917, "loss_cns_2": 0.63972, "loss_yns_2": 0.1641, "loss_cls_3": 1.05788, "loss_box_3": 1.9527, "loss_cns_3": 0.64558, "loss_yns_3": 0.16349, "loss_cls_4": 1.05624, "loss_box_4": 1.94289, "loss_cns_4": 0.6477, "loss_yns_4": 0.1635, "loss_cls_5": 1.06569, "loss_box_5": 1.95935, "loss_cns_5": 0.64478, "loss_yns_5": 0.16013, "loss_cls_dn_0": 0.26341, "loss_box_dn_0": 0.83077, "loss_cls_dn_1": 0.19681, "loss_box_dn_1": 0.88366, "loss_cls_dn_2": 0.19518, "loss_box_dn_2": 0.83388, "loss_cls_dn_3": 0.19543, "loss_box_dn_3": 0.83644, "loss_cls_dn_4": 0.2086, "loss_box_dn_4": 0.8437, "loss_cls_dn_5": 0.21939, "loss_box_dn_5": 0.85134, "loss_dense_depth": 0.86407, "loss": 29.88952, "grad_norm": 31.73909, "time": 1.53263} -{"mode": "train", "epoch": 1, "iter": 84, "lr": 0.00013, "memory": 49163, "data_time": 0.07895, "loss_cls_0": 0.90398, "loss_box_0": 1.81966, "loss_cns_0": 0.62254, "loss_yns_0": 0.15825, "loss_cls_1": 0.94675, "loss_box_1": 2.0124, "loss_cns_1": 0.62521, "loss_yns_1": 0.16257, "loss_cls_2": 1.01339, "loss_box_2": 1.94466, "loss_cns_2": 0.64348, "loss_yns_2": 0.16315, "loss_cls_3": 1.02817, "loss_box_3": 1.9205, "loss_cns_3": 0.64995, "loss_yns_3": 0.16469, "loss_cls_4": 1.01571, "loss_box_4": 1.93953, "loss_cns_4": 0.64833, "loss_yns_4": 0.16108, "loss_cls_5": 1.02275, "loss_box_5": 1.97474, "loss_cns_5": 0.64737, "loss_yns_5": 0.15747, "loss_cls_dn_0": 0.24809, "loss_box_dn_0": 0.82119, "loss_cls_dn_1": 0.19768, "loss_box_dn_1": 0.83085, "loss_cls_dn_2": 0.20574, "loss_box_dn_2": 0.7921, "loss_cls_dn_3": 0.20564, "loss_box_dn_3": 0.80222, "loss_cls_dn_4": 0.21604, "loss_box_dn_4": 0.82674, "loss_cls_dn_5": 0.22653, "loss_box_dn_5": 0.83877, "loss_dense_depth": 0.85146, "loss": 29.40933, "grad_norm": 46.02559, "time": 1.52321} -{"mode": "train", "epoch": 1, "iter": 85, "lr": 0.00013, "memory": 49163, "data_time": 0.0888, "loss_cls_0": 0.94182, "loss_box_0": 1.83045, "loss_cns_0": 0.61895, "loss_yns_0": 0.15943, "loss_cls_1": 0.9934, "loss_box_1": 2.05762, "loss_cns_1": 0.61799, "loss_yns_1": 0.16078, "loss_cls_2": 1.01863, "loss_box_2": 1.99407, "loss_cns_2": 0.6363, "loss_yns_2": 0.16442, "loss_cls_3": 1.0356, "loss_box_3": 1.9553, "loss_cns_3": 0.64361, "loss_yns_3": 0.16368, "loss_cls_4": 1.03738, "loss_box_4": 1.95705, "loss_cns_4": 0.64539, "loss_yns_4": 0.16121, "loss_cls_5": 1.03353, "loss_box_5": 1.97321, "loss_cns_5": 0.64349, "loss_yns_5": 0.15923, "loss_cls_dn_0": 0.26184, "loss_box_dn_0": 0.83596, "loss_cls_dn_1": 0.19879, "loss_box_dn_1": 0.86833, "loss_cls_dn_2": 0.20585, "loss_box_dn_2": 0.82445, "loss_cls_dn_3": 0.2057, "loss_box_dn_3": 0.82397, "loss_cls_dn_4": 0.21637, "loss_box_dn_4": 0.83663, "loss_cls_dn_5": 0.22267, "loss_box_dn_5": 0.84011, "loss_dense_depth": 0.87231, "loss": 29.8155, "grad_norm": 43.02705, "time": 1.52259} -{"mode": "train", "epoch": 1, "iter": 86, "lr": 0.00013, "memory": 49163, "data_time": 0.07652, "loss_cls_0": 0.92736, "loss_box_0": 1.80786, "loss_cns_0": 0.61612, "loss_yns_0": 0.15996, "loss_cls_1": 0.9765, "loss_box_1": 1.92575, "loss_cns_1": 0.62177, "loss_yns_1": 0.16385, "loss_cls_2": 1.00148, "loss_box_2": 1.85886, "loss_cns_2": 0.64122, "loss_yns_2": 0.16759, "loss_cls_3": 1.02265, "loss_box_3": 1.8376, "loss_cns_3": 0.64584, "loss_yns_3": 0.16371, "loss_cls_4": 1.02531, "loss_box_4": 1.81562, "loss_cns_4": 0.65364, "loss_yns_4": 0.16415, "loss_cls_5": 1.02889, "loss_box_5": 1.83186, "loss_cns_5": 0.64647, "loss_yns_5": 0.16715, "loss_cls_dn_0": 0.25407, "loss_box_dn_0": 0.82911, "loss_cls_dn_1": 0.19646, "loss_box_dn_1": 0.84228, "loss_cls_dn_2": 0.20188, "loss_box_dn_2": 0.79725, "loss_cls_dn_3": 0.20292, "loss_box_dn_3": 0.79208, "loss_cls_dn_4": 0.21319, "loss_box_dn_4": 0.79264, "loss_cls_dn_5": 0.21699, "loss_box_dn_5": 0.80459, "loss_dense_depth": 0.86444, "loss": 28.87911, "grad_norm": 29.65866, "time": 1.53124} -{"mode": "train", "epoch": 1, "iter": 87, "lr": 0.00013, "memory": 49163, "data_time": 0.07742, "loss_cls_0": 0.90777, "loss_box_0": 1.82126, "loss_cns_0": 0.61745, "loss_yns_0": 0.16019, "loss_cls_1": 0.95988, "loss_box_1": 1.95977, "loss_cns_1": 0.61976, "loss_yns_1": 0.16352, "loss_cls_2": 1.01555, "loss_box_2": 1.84938, "loss_cns_2": 0.64885, "loss_yns_2": 0.16286, "loss_cls_3": 1.0224, "loss_box_3": 1.84035, "loss_cns_3": 0.6511, "loss_yns_3": 0.16085, "loss_cls_4": 1.03212, "loss_box_4": 1.82402, "loss_cns_4": 0.66237, "loss_yns_4": 0.16454, "loss_cls_5": 1.02368, "loss_box_5": 1.83661, "loss_cns_5": 0.65483, "loss_yns_5": 0.16468, "loss_cls_dn_0": 0.24955, "loss_box_dn_0": 0.82179, "loss_cls_dn_1": 0.19767, "loss_box_dn_1": 0.83373, "loss_cls_dn_2": 0.20293, "loss_box_dn_2": 0.78095, "loss_cls_dn_3": 0.20114, "loss_box_dn_3": 0.77326, "loss_cls_dn_4": 0.2142, "loss_box_dn_4": 0.76887, "loss_cls_dn_5": 0.21596, "loss_box_dn_5": 0.78538, "loss_dense_depth": 0.88055, "loss": 28.84975, "grad_norm": 38.76965, "time": 1.49885} -{"mode": "train", "epoch": 1, "iter": 88, "lr": 0.00013, "memory": 49163, "data_time": 0.07788, "loss_cls_0": 0.93776, "loss_box_0": 1.8284, "loss_cns_0": 0.61661, "loss_yns_0": 0.15935, "loss_cls_1": 0.98309, "loss_box_1": 1.97012, "loss_cns_1": 0.62117, "loss_yns_1": 0.1618, "loss_cls_2": 1.0226, "loss_box_2": 1.8744, "loss_cns_2": 0.64498, "loss_yns_2": 0.15998, "loss_cls_3": 1.02336, "loss_box_3": 1.8499, "loss_cns_3": 0.65052, "loss_yns_3": 0.16022, "loss_cls_4": 1.03919, "loss_box_4": 1.83983, "loss_cns_4": 0.65902, "loss_yns_4": 0.16238, "loss_cls_5": 1.03571, "loss_box_5": 1.84816, "loss_cns_5": 0.65454, "loss_yns_5": 0.16422, "loss_cls_dn_0": 0.25025, "loss_box_dn_0": 0.83125, "loss_cls_dn_1": 0.19803, "loss_box_dn_1": 0.83147, "loss_cls_dn_2": 0.20074, "loss_box_dn_2": 0.78284, "loss_cls_dn_3": 0.19718, "loss_box_dn_3": 0.76716, "loss_cls_dn_4": 0.21028, "loss_box_dn_4": 0.76556, "loss_cls_dn_5": 0.21486, "loss_box_dn_5": 0.78011, "loss_dense_depth": 0.83609, "loss": 28.93313, "grad_norm": 39.41268, "time": 1.4879} -{"mode": "train", "epoch": 1, "iter": 89, "lr": 0.00014, "memory": 49163, "data_time": 0.07744, "loss_cls_0": 0.92142, "loss_box_0": 1.80041, "loss_cns_0": 0.61195, "loss_yns_0": 0.15769, "loss_cls_1": 0.99698, "loss_box_1": 1.96177, "loss_cns_1": 0.6196, "loss_yns_1": 0.15906, "loss_cls_2": 1.01532, "loss_box_2": 1.88234, "loss_cns_2": 0.63973, "loss_yns_2": 0.15806, "loss_cls_3": 1.02279, "loss_box_3": 1.85126, "loss_cns_3": 0.64701, "loss_yns_3": 0.15839, "loss_cls_4": 1.02964, "loss_box_4": 1.8656, "loss_cns_4": 0.64764, "loss_yns_4": 0.16025, "loss_cls_5": 1.03032, "loss_box_5": 1.8997, "loss_cns_5": 0.6436, "loss_yns_5": 0.15881, "loss_cls_dn_0": 0.25239, "loss_box_dn_0": 0.83235, "loss_cls_dn_1": 0.20151, "loss_box_dn_1": 0.82397, "loss_cls_dn_2": 0.19978, "loss_box_dn_2": 0.77668, "loss_cls_dn_3": 0.19582, "loss_box_dn_3": 0.76329, "loss_cls_dn_4": 0.20851, "loss_box_dn_4": 0.77511, "loss_cls_dn_5": 0.21475, "loss_box_dn_5": 0.79446, "loss_dense_depth": 0.85439, "loss": 28.93237, "grad_norm": 36.6414, "time": 1.50455} -{"mode": "train", "epoch": 1, "iter": 90, "lr": 0.00014, "memory": 49163, "data_time": 0.0727, "loss_cls_0": 0.93682, "loss_box_0": 1.77102, "loss_cns_0": 0.60844, "loss_yns_0": 0.15549, "loss_cls_1": 1.00328, "loss_box_1": 1.98553, "loss_cns_1": 0.6289, "loss_yns_1": 0.15931, "loss_cls_2": 1.0228, "loss_box_2": 1.90462, "loss_cns_2": 0.64505, "loss_yns_2": 0.16379, "loss_cls_3": 1.03605, "loss_box_3": 1.89598, "loss_cns_3": 0.64934, "loss_yns_3": 0.15811, "loss_cls_4": 1.03019, "loss_box_4": 1.87623, "loss_cns_4": 0.65297, "loss_yns_4": 0.16059, "loss_cls_5": 1.03335, "loss_box_5": 1.90808, "loss_cns_5": 0.65067, "loss_yns_5": 0.15871, "loss_cls_dn_0": 0.25375, "loss_box_dn_0": 0.83334, "loss_cls_dn_1": 0.20011, "loss_box_dn_1": 0.82811, "loss_cls_dn_2": 0.19807, "loss_box_dn_2": 0.78568, "loss_cls_dn_3": 0.1932, "loss_box_dn_3": 0.78601, "loss_cls_dn_4": 0.20389, "loss_box_dn_4": 0.78671, "loss_cls_dn_5": 0.21266, "loss_box_dn_5": 0.80431, "loss_dense_depth": 0.87185, "loss": 29.15302, "grad_norm": 32.89888, "time": 1.50107} -{"mode": "train", "epoch": 1, "iter": 91, "lr": 0.00014, "memory": 49163, "data_time": 0.07396, "loss_cls_0": 0.944, "loss_box_0": 1.77983, "loss_cns_0": 0.61947, "loss_yns_0": 0.1547, "loss_cls_1": 0.98402, "loss_box_1": 1.96011, "loss_cns_1": 0.63241, "loss_yns_1": 0.15712, "loss_cls_2": 1.01999, "loss_box_2": 1.89194, "loss_cns_2": 0.64552, "loss_yns_2": 0.15852, "loss_cls_3": 1.02529, "loss_box_3": 1.89258, "loss_cns_3": 0.64821, "loss_yns_3": 0.1545, "loss_cls_4": 1.02524, "loss_box_4": 1.87678, "loss_cns_4": 0.65237, "loss_yns_4": 0.15699, "loss_cls_5": 1.03285, "loss_box_5": 1.88405, "loss_cns_5": 0.64745, "loss_yns_5": 0.16137, "loss_cls_dn_0": 0.24491, "loss_box_dn_0": 0.80994, "loss_cls_dn_1": 0.19536, "loss_box_dn_1": 0.82969, "loss_cls_dn_2": 0.19627, "loss_box_dn_2": 0.80035, "loss_cls_dn_3": 0.19424, "loss_box_dn_3": 0.80384, "loss_cls_dn_4": 0.20379, "loss_box_dn_4": 0.80393, "loss_cls_dn_5": 0.2105, "loss_box_dn_5": 0.81667, "loss_dense_depth": 0.8484, "loss": 29.06319, "grad_norm": 35.9735, "time": 1.48367} -{"mode": "train", "epoch": 1, "iter": 92, "lr": 0.00014, "memory": 49163, "data_time": 0.07309, "loss_cls_0": 0.9178, "loss_box_0": 1.77695, "loss_cns_0": 0.61772, "loss_yns_0": 0.15085, "loss_cls_1": 0.99011, "loss_box_1": 1.93984, "loss_cns_1": 0.63379, "loss_yns_1": 0.15602, "loss_cls_2": 1.0175, "loss_box_2": 1.88543, "loss_cns_2": 0.64556, "loss_yns_2": 0.15498, "loss_cls_3": 1.00223, "loss_box_3": 1.87246, "loss_cns_3": 0.65016, "loss_yns_3": 0.15427, "loss_cls_4": 1.00978, "loss_box_4": 1.86681, "loss_cns_4": 0.65395, "loss_yns_4": 0.15584, "loss_cls_5": 1.01745, "loss_box_5": 1.85634, "loss_cns_5": 0.64985, "loss_yns_5": 0.16473, "loss_cls_dn_0": 0.2483, "loss_box_dn_0": 0.81207, "loss_cls_dn_1": 0.19002, "loss_box_dn_1": 0.81961, "loss_cls_dn_2": 0.1911, "loss_box_dn_2": 0.78293, "loss_cls_dn_3": 0.18885, "loss_box_dn_3": 0.77642, "loss_cls_dn_4": 0.19853, "loss_box_dn_4": 0.77779, "loss_cls_dn_5": 0.20512, "loss_box_dn_5": 0.78616, "loss_dense_depth": 0.88075, "loss": 28.79808, "grad_norm": 33.38271, "time": 1.48638} -{"mode": "train", "epoch": 1, "iter": 93, "lr": 0.00014, "memory": 49163, "data_time": 0.0763, "loss_cls_0": 0.93674, "loss_box_0": 1.75147, "loss_cns_0": 0.61843, "loss_yns_0": 0.14954, "loss_cls_1": 1.00511, "loss_box_1": 1.93878, "loss_cns_1": 0.63875, "loss_yns_1": 0.15595, "loss_cls_2": 1.03213, "loss_box_2": 1.87684, "loss_cns_2": 0.64882, "loss_yns_2": 0.15381, "loss_cls_3": 1.02909, "loss_box_3": 1.86419, "loss_cns_3": 0.6558, "loss_yns_3": 0.15422, "loss_cls_4": 1.02421, "loss_box_4": 1.84812, "loss_cns_4": 0.65475, "loss_yns_4": 0.15507, "loss_cls_5": 1.02565, "loss_box_5": 1.83553, "loss_cns_5": 0.65209, "loss_yns_5": 0.1573, "loss_cls_dn_0": 0.24876, "loss_box_dn_0": 0.81194, "loss_cls_dn_1": 0.18673, "loss_box_dn_1": 0.81369, "loss_cls_dn_2": 0.18743, "loss_box_dn_2": 0.76996, "loss_cls_dn_3": 0.18648, "loss_box_dn_3": 0.7588, "loss_cls_dn_4": 0.19569, "loss_box_dn_4": 0.75732, "loss_cls_dn_5": 0.20073, "loss_box_dn_5": 0.76067, "loss_dense_depth": 0.81273, "loss": 28.65333, "grad_norm": 32.31005, "time": 1.48085} -{"mode": "train", "epoch": 1, "iter": 94, "lr": 0.00014, "memory": 49163, "data_time": 0.07135, "loss_cls_0": 0.94074, "loss_box_0": 1.78075, "loss_cns_0": 0.61875, "loss_yns_0": 0.15107, "loss_cls_1": 1.00261, "loss_box_1": 1.94389, "loss_cns_1": 0.63585, "loss_yns_1": 0.15497, "loss_cls_2": 1.04384, "loss_box_2": 1.88823, "loss_cns_2": 0.64603, "loss_yns_2": 0.15681, "loss_cls_3": 1.03787, "loss_box_3": 1.87757, "loss_cns_3": 0.65315, "loss_yns_3": 0.15458, "loss_cls_4": 1.0329, "loss_box_4": 1.86105, "loss_cns_4": 0.65321, "loss_yns_4": 0.15637, "loss_cls_5": 1.04165, "loss_box_5": 1.86178, "loss_cns_5": 0.65068, "loss_yns_5": 0.15618, "loss_cls_dn_0": 0.24782, "loss_box_dn_0": 0.80547, "loss_cls_dn_1": 0.18761, "loss_box_dn_1": 0.79762, "loss_cls_dn_2": 0.19269, "loss_box_dn_2": 0.76174, "loss_cls_dn_3": 0.18864, "loss_box_dn_3": 0.75644, "loss_cls_dn_4": 0.19978, "loss_box_dn_4": 0.75825, "loss_cls_dn_5": 0.2079, "loss_box_dn_5": 0.76415, "loss_dense_depth": 0.86638, "loss": 28.83503, "grad_norm": 35.63077, "time": 1.48966} -{"mode": "train", "epoch": 1, "iter": 95, "lr": 0.00014, "memory": 49163, "data_time": 0.08178, "loss_cls_0": 0.94099, "loss_box_0": 1.80258, "loss_cns_0": 0.62031, "loss_yns_0": 0.15294, "loss_cls_1": 1.00729, "loss_box_1": 1.87591, "loss_cns_1": 0.64254, "loss_yns_1": 0.15545, "loss_cls_2": 1.05376, "loss_box_2": 1.81739, "loss_cns_2": 0.65425, "loss_yns_2": 0.15476, "loss_cls_3": 1.03879, "loss_box_3": 1.8179, "loss_cns_3": 0.65565, "loss_yns_3": 0.15406, "loss_cls_4": 1.0392, "loss_box_4": 1.8083, "loss_cns_4": 0.65766, "loss_yns_4": 0.15571, "loss_cls_5": 1.05638, "loss_box_5": 1.80439, "loss_cns_5": 0.65699, "loss_yns_5": 0.16066, "loss_cls_dn_0": 0.24182, "loss_box_dn_0": 0.80654, "loss_cls_dn_1": 0.18377, "loss_box_dn_1": 0.77915, "loss_cls_dn_2": 0.18508, "loss_box_dn_2": 0.75368, "loss_cls_dn_3": 0.18576, "loss_box_dn_3": 0.76515, "loss_cls_dn_4": 0.19692, "loss_box_dn_4": 0.77314, "loss_cls_dn_5": 0.20361, "loss_box_dn_5": 0.78187, "loss_dense_depth": 0.85674, "loss": 28.59708, "grad_norm": 32.83364, "time": 1.49752} -{"mode": "train", "epoch": 1, "iter": 96, "lr": 0.00014, "memory": 49163, "data_time": 0.07817, "loss_cls_0": 0.94914, "loss_box_0": 1.82767, "loss_cns_0": 0.61639, "loss_yns_0": 0.15582, "loss_cls_1": 1.00261, "loss_box_1": 1.87689, "loss_cns_1": 0.63906, "loss_yns_1": 0.1546, "loss_cls_2": 1.04359, "loss_box_2": 1.82727, "loss_cns_2": 0.64852, "loss_yns_2": 0.15576, "loss_cls_3": 1.03558, "loss_box_3": 1.82895, "loss_cns_3": 0.65037, "loss_yns_3": 0.1577, "loss_cls_4": 1.06885, "loss_box_4": 1.8105, "loss_cns_4": 0.65145, "loss_yns_4": 0.1584, "loss_cls_5": 1.05308, "loss_box_5": 1.8088, "loss_cns_5": 0.65016, "loss_yns_5": 0.16647, "loss_cls_dn_0": 0.24814, "loss_box_dn_0": 0.81441, "loss_cls_dn_1": 0.18316, "loss_box_dn_1": 0.79871, "loss_cls_dn_2": 0.18259, "loss_box_dn_2": 0.77342, "loss_cls_dn_3": 0.18279, "loss_box_dn_3": 0.78668, "loss_cls_dn_4": 0.19796, "loss_box_dn_4": 0.79098, "loss_cls_dn_5": 0.20632, "loss_box_dn_5": 0.80035, "loss_dense_depth": 0.85908, "loss": 28.76225, "grad_norm": 35.49201, "time": 1.49109} -{"mode": "train", "epoch": 1, "iter": 97, "lr": 0.00014, "memory": 49163, "data_time": 0.07523, "loss_cls_0": 0.95374, "loss_box_0": 1.78092, "loss_cns_0": 0.61691, "loss_yns_0": 0.15223, "loss_cls_1": 0.99585, "loss_box_1": 1.85134, "loss_cns_1": 0.63256, "loss_yns_1": 0.15552, "loss_cls_2": 1.04524, "loss_box_2": 1.79979, "loss_cns_2": 0.64988, "loss_yns_2": 0.15656, "loss_cls_3": 1.04752, "loss_box_3": 1.79169, "loss_cns_3": 0.65322, "loss_yns_3": 0.15776, "loss_cls_4": 1.05864, "loss_box_4": 1.78244, "loss_cns_4": 0.65249, "loss_yns_4": 0.15677, "loss_cls_5": 1.0523, "loss_box_5": 1.78731, "loss_cns_5": 0.64955, "loss_yns_5": 0.16093, "loss_cls_dn_0": 0.2532, "loss_box_dn_0": 0.81246, "loss_cls_dn_1": 0.18502, "loss_box_dn_1": 0.7994, "loss_cls_dn_2": 0.18332, "loss_box_dn_2": 0.77108, "loss_cls_dn_3": 0.1848, "loss_box_dn_3": 0.77399, "loss_cls_dn_4": 0.19558, "loss_box_dn_4": 0.77787, "loss_cls_dn_5": 0.20257, "loss_box_dn_5": 0.78711, "loss_dense_depth": 0.85993, "loss": 28.52748, "grad_norm": 32.49829, "time": 1.49002} -{"mode": "train", "epoch": 1, "iter": 98, "lr": 0.00014, "memory": 49163, "data_time": 0.07088, "loss_cls_0": 0.93787, "loss_box_0": 1.75659, "loss_cns_0": 0.61955, "loss_yns_0": 0.15117, "loss_cls_1": 0.98761, "loss_box_1": 1.85201, "loss_cns_1": 0.63239, "loss_yns_1": 0.15285, "loss_cls_2": 1.02901, "loss_box_2": 1.80493, "loss_cns_2": 0.65068, "loss_yns_2": 0.15652, "loss_cls_3": 1.0419, "loss_box_3": 1.81095, "loss_cns_3": 0.65351, "loss_yns_3": 0.15603, "loss_cls_4": 1.05394, "loss_box_4": 1.78381, "loss_cns_4": 0.65348, "loss_yns_4": 0.15663, "loss_cls_5": 1.0572, "loss_box_5": 1.77688, "loss_cns_5": 0.65313, "loss_yns_5": 0.15665, "loss_cls_dn_0": 0.25014, "loss_box_dn_0": 0.80853, "loss_cls_dn_1": 0.18427, "loss_box_dn_1": 0.79839, "loss_cls_dn_2": 0.18856, "loss_box_dn_2": 0.76128, "loss_cls_dn_3": 0.18916, "loss_box_dn_3": 0.76631, "loss_cls_dn_4": 0.20048, "loss_box_dn_4": 0.7617, "loss_cls_dn_5": 0.20538, "loss_box_dn_5": 0.76188, "loss_dense_depth": 0.81398, "loss": 28.37535, "grad_norm": 37.32568, "time": 1.46899} -{"mode": "train", "epoch": 1, "iter": 99, "lr": 0.00014, "memory": 49163, "data_time": 0.07253, "loss_cls_0": 0.93442, "loss_box_0": 1.77474, "loss_cns_0": 0.61828, "loss_yns_0": 0.15263, "loss_cls_1": 0.99296, "loss_box_1": 1.89843, "loss_cns_1": 0.62463, "loss_yns_1": 0.15323, "loss_cls_2": 1.0427, "loss_box_2": 1.85552, "loss_cns_2": 0.64457, "loss_yns_2": 0.15583, "loss_cls_3": 1.05052, "loss_box_3": 1.86214, "loss_cns_3": 0.64809, "loss_yns_3": 0.15738, "loss_cls_4": 1.05866, "loss_box_4": 1.83415, "loss_cns_4": 0.64913, "loss_yns_4": 0.1572, "loss_cls_5": 1.05008, "loss_box_5": 1.84061, "loss_cns_5": 0.64615, "loss_yns_5": 0.15541, "loss_cls_dn_0": 0.24892, "loss_box_dn_0": 0.80479, "loss_cls_dn_1": 0.18096, "loss_box_dn_1": 0.78294, "loss_cls_dn_2": 0.18479, "loss_box_dn_2": 0.75204, "loss_cls_dn_3": 0.18617, "loss_box_dn_3": 0.75885, "loss_cls_dn_4": 0.1969, "loss_box_dn_4": 0.75176, "loss_cls_dn_5": 0.2006, "loss_box_dn_5": 0.75683, "loss_dense_depth": 0.84457, "loss": 28.60756, "grad_norm": 35.22978, "time": 1.47687} -{"mode": "train", "epoch": 1, "iter": 100, "lr": 0.00014, "memory": 49163, "data_time": 0.08509, "loss_cls_0": 0.93609, "loss_box_0": 1.81338, "loss_cns_0": 0.62146, "loss_yns_0": 0.15566, "loss_cls_1": 0.96723, "loss_box_1": 1.93969, "loss_cns_1": 0.62345, "loss_yns_1": 0.15663, "loss_cls_2": 1.03177, "loss_box_2": 1.88273, "loss_cns_2": 0.64222, "loss_yns_2": 0.15808, "loss_cls_3": 1.04512, "loss_box_3": 1.86842, "loss_cns_3": 0.64575, "loss_yns_3": 0.15935, "loss_cls_4": 1.03451, "loss_box_4": 1.86619, "loss_cns_4": 0.64602, "loss_yns_4": 0.15853, "loss_cls_5": 1.0384, "loss_box_5": 1.86317, "loss_cns_5": 0.64547, "loss_yns_5": 0.16065, "loss_cls_dn_0": 0.24614, "loss_box_dn_0": 0.81141, "loss_cls_dn_1": 0.17785, "loss_box_dn_1": 0.77744, "loss_cls_dn_2": 0.18166, "loss_box_dn_2": 0.7455, "loss_cls_dn_3": 0.1843, "loss_box_dn_3": 0.74421, "loss_cls_dn_4": 0.18988, "loss_box_dn_4": 0.75228, "loss_cls_dn_5": 0.19569, "loss_box_dn_5": 0.75823, "loss_dense_depth": 0.81448, "loss": 28.63906, "grad_norm": 33.48449, "time": 1.48817} -{"mode": "train", "epoch": 1, "iter": 101, "lr": 0.00014, "memory": 49163, "data_time": 0.07914, "loss_cls_0": 0.9534, "loss_box_0": 1.807, "loss_cns_0": 0.62559, "loss_yns_0": 0.15812, "loss_cls_1": 0.99409, "loss_box_1": 1.95941, "loss_cns_1": 0.62479, "loss_yns_1": 0.1586, "loss_cls_2": 1.02054, "loss_box_2": 1.90346, "loss_cns_2": 0.64588, "loss_yns_2": 0.16087, "loss_cls_3": 1.05622, "loss_box_3": 1.89721, "loss_cns_3": 0.6491, "loss_yns_3": 0.15906, "loss_cls_4": 1.05682, "loss_box_4": 1.89629, "loss_cns_4": 0.6498, "loss_yns_4": 0.16021, "loss_cls_5": 1.03925, "loss_box_5": 1.88509, "loss_cns_5": 0.64815, "loss_yns_5": 0.16276, "loss_cls_dn_0": 0.24558, "loss_box_dn_0": 0.80433, "loss_cls_dn_1": 0.18191, "loss_box_dn_1": 0.81334, "loss_cls_dn_2": 0.18306, "loss_box_dn_2": 0.78048, "loss_cls_dn_3": 0.1851, "loss_box_dn_3": 0.78566, "loss_cls_dn_4": 0.19368, "loss_box_dn_4": 0.79665, "loss_cls_dn_5": 0.19817, "loss_box_dn_5": 0.80459, "loss_dense_depth": 0.83433, "loss": 29.07862, "grad_norm": 55.40846, "time": 1.55836} -{"mode": "train", "epoch": 1, "iter": 102, "lr": 0.00014, "memory": 49163, "data_time": 0.15781, "loss_cls_0": 0.92542, "loss_box_0": 1.82003, "loss_cns_0": 0.61907, "loss_yns_0": 0.15841, "loss_cls_1": 0.97738, "loss_box_1": 1.9382, "loss_cns_1": 0.62855, "loss_yns_1": 0.15955, "loss_cls_2": 1.01001, "loss_box_2": 1.87598, "loss_cns_2": 0.64389, "loss_yns_2": 0.15964, "loss_cls_3": 1.02288, "loss_box_3": 1.87539, "loss_cns_3": 0.64695, "loss_yns_3": 0.15942, "loss_cls_4": 1.03987, "loss_box_4": 1.86281, "loss_cns_4": 0.6501, "loss_yns_4": 0.15987, "loss_cls_5": 1.0258, "loss_box_5": 1.8614, "loss_cns_5": 0.64784, "loss_yns_5": 0.16203, "loss_cls_dn_0": 0.24557, "loss_box_dn_0": 0.81177, "loss_cls_dn_1": 0.18366, "loss_box_dn_1": 0.8129, "loss_cls_dn_2": 0.18263, "loss_box_dn_2": 0.78654, "loss_cls_dn_3": 0.18351, "loss_box_dn_3": 0.79982, "loss_cls_dn_4": 0.19767, "loss_box_dn_4": 0.81282, "loss_cls_dn_5": 0.20201, "loss_box_dn_5": 0.833, "loss_dense_depth": 0.82575, "loss": 28.90815, "grad_norm": 29.89357, "time": 1.55099} -{"mode": "train", "epoch": 1, "iter": 103, "lr": 0.00014, "memory": 49163, "data_time": 0.06922, "loss_cls_0": 0.90826, "loss_box_0": 1.78016, "loss_cns_0": 0.61875, "loss_yns_0": 0.15885, "loss_cls_1": 0.94849, "loss_box_1": 1.9338, "loss_cns_1": 0.63094, "loss_yns_1": 0.15854, "loss_cls_2": 1.00708, "loss_box_2": 1.86876, "loss_cns_2": 0.6461, "loss_yns_2": 0.15936, "loss_cls_3": 1.01331, "loss_box_3": 1.86316, "loss_cns_3": 0.64931, "loss_yns_3": 0.16027, "loss_cls_4": 1.02159, "loss_box_4": 1.87131, "loss_cns_4": 0.6481, "loss_yns_4": 0.1599, "loss_cls_5": 1.01035, "loss_box_5": 1.89891, "loss_cns_5": 0.64997, "loss_yns_5": 0.16142, "loss_cls_dn_0": 0.23893, "loss_box_dn_0": 0.80115, "loss_cls_dn_1": 0.17596, "loss_box_dn_1": 0.84284, "loss_cls_dn_2": 0.17741, "loss_box_dn_2": 0.81177, "loss_cls_dn_3": 0.17887, "loss_box_dn_3": 0.81744, "loss_cls_dn_4": 0.19401, "loss_box_dn_4": 0.83683, "loss_cls_dn_5": 0.19763, "loss_box_dn_5": 0.86454, "loss_dense_depth": 0.82995, "loss": 28.89398, "grad_norm": 54.47373, "time": 1.52833} -{"mode": "train", "epoch": 1, "iter": 104, "lr": 0.00014, "memory": 49163, "data_time": 0.07011, "loss_cls_0": 0.91463, "loss_box_0": 1.77441, "loss_cns_0": 0.61445, "loss_yns_0": 0.15734, "loss_cls_1": 0.96446, "loss_box_1": 1.99673, "loss_cns_1": 0.6295, "loss_yns_1": 0.15795, "loss_cls_2": 1.01058, "loss_box_2": 1.9167, "loss_cns_2": 0.64965, "loss_yns_2": 0.15687, "loss_cls_3": 1.03655, "loss_box_3": 1.91785, "loss_cns_3": 0.65235, "loss_yns_3": 0.1593, "loss_cls_4": 1.02957, "loss_box_4": 1.9252, "loss_cns_4": 0.65167, "loss_yns_4": 0.15855, "loss_cls_5": 1.01981, "loss_box_5": 1.93743, "loss_cns_5": 0.65319, "loss_yns_5": 0.15961, "loss_cls_dn_0": 0.24228, "loss_box_dn_0": 0.81053, "loss_cls_dn_1": 0.1755, "loss_box_dn_1": 0.87437, "loss_cls_dn_2": 0.17749, "loss_box_dn_2": 0.82569, "loss_cls_dn_3": 0.18462, "loss_box_dn_3": 0.8198, "loss_cls_dn_4": 0.19411, "loss_box_dn_4": 0.83423, "loss_cls_dn_5": 0.19739, "loss_box_dn_5": 0.85174, "loss_dense_depth": 0.82508, "loss": 29.25716, "grad_norm": 55.93401, "time": 1.56141} -{"mode": "train", "epoch": 1, "iter": 105, "lr": 0.00014, "memory": 49163, "data_time": 0.08311, "loss_cls_0": 0.89875, "loss_box_0": 1.74616, "loss_cns_0": 0.61539, "loss_yns_0": 0.15563, "loss_cls_1": 0.96534, "loss_box_1": 1.98738, "loss_cns_1": 0.62704, "loss_yns_1": 0.15622, "loss_cls_2": 0.99442, "loss_box_2": 1.92539, "loss_cns_2": 0.64871, "loss_yns_2": 0.15739, "loss_cls_3": 1.02476, "loss_box_3": 1.92275, "loss_cns_3": 0.65164, "loss_yns_3": 0.15855, "loss_cls_4": 1.02667, "loss_box_4": 1.88815, "loss_cns_4": 0.65601, "loss_yns_4": 0.15775, "loss_cls_5": 1.03229, "loss_box_5": 1.89082, "loss_cns_5": 0.65102, "loss_yns_5": 0.1605, "loss_cls_dn_0": 0.23853, "loss_box_dn_0": 0.81627, "loss_cls_dn_1": 0.17723, "loss_box_dn_1": 0.89098, "loss_cls_dn_2": 0.17845, "loss_box_dn_2": 0.82407, "loss_cls_dn_3": 0.18503, "loss_box_dn_3": 0.81044, "loss_cls_dn_4": 0.19174, "loss_box_dn_4": 0.80727, "loss_cls_dn_5": 0.19186, "loss_box_dn_5": 0.81231, "loss_dense_depth": 0.78405, "loss": 29.00695, "grad_norm": 45.73758, "time": 1.51318} -{"mode": "train", "epoch": 1, "iter": 106, "lr": 0.00014, "memory": 49163, "data_time": 0.07011, "loss_cls_0": 0.90155, "loss_box_0": 1.74734, "loss_cns_0": 0.61028, "loss_yns_0": 0.15656, "loss_cls_1": 0.96064, "loss_box_1": 2.0161, "loss_cns_1": 0.62921, "loss_yns_1": 0.15848, "loss_cls_2": 0.9961, "loss_box_2": 1.95791, "loss_cns_2": 0.64391, "loss_yns_2": 0.15786, "loss_cls_3": 1.01236, "loss_box_3": 1.95455, "loss_cns_3": 0.64851, "loss_yns_3": 0.15776, "loss_cls_4": 1.02006, "loss_box_4": 1.90915, "loss_cns_4": 0.65255, "loss_yns_4": 0.1568, "loss_cls_5": 1.02074, "loss_box_5": 1.91201, "loss_cns_5": 0.65097, "loss_yns_5": 0.15752, "loss_cls_dn_0": 0.23677, "loss_box_dn_0": 0.8016, "loss_cls_dn_1": 0.17674, "loss_box_dn_1": 0.81656, "loss_cls_dn_2": 0.17819, "loss_box_dn_2": 0.77083, "loss_cls_dn_3": 0.17806, "loss_box_dn_3": 0.76198, "loss_cls_dn_4": 0.18618, "loss_box_dn_4": 0.75128, "loss_cls_dn_5": 0.18751, "loss_box_dn_5": 0.75662, "loss_dense_depth": 0.80102, "loss": 28.79226, "grad_norm": 32.85572, "time": 1.50792} -{"mode": "train", "epoch": 1, "iter": 107, "lr": 0.00014, "memory": 49163, "data_time": 0.07174, "loss_cls_0": 0.89712, "loss_box_0": 1.73888, "loss_cns_0": 0.6145, "loss_yns_0": 0.15694, "loss_cls_1": 0.94209, "loss_box_1": 1.96635, "loss_cns_1": 0.63063, "loss_yns_1": 0.15808, "loss_cls_2": 1.00124, "loss_box_2": 1.89799, "loss_cns_2": 0.64708, "loss_yns_2": 0.15532, "loss_cls_3": 1.00543, "loss_box_3": 1.92177, "loss_cns_3": 0.64769, "loss_yns_3": 0.15742, "loss_cls_4": 1.01681, "loss_box_4": 1.92415, "loss_cns_4": 0.64984, "loss_yns_4": 0.15689, "loss_cls_5": 1.04379, "loss_box_5": 1.92757, "loss_cns_5": 0.6454, "loss_yns_5": 0.15575, "loss_cls_dn_0": 0.23768, "loss_box_dn_0": 0.8145, "loss_cls_dn_1": 0.17495, "loss_box_dn_1": 0.7868, "loss_cls_dn_2": 0.17732, "loss_box_dn_2": 0.75986, "loss_cls_dn_3": 0.17382, "loss_box_dn_3": 0.77474, "loss_cls_dn_4": 0.19084, "loss_box_dn_4": 0.78295, "loss_cls_dn_5": 0.20408, "loss_box_dn_5": 0.79411, "loss_dense_depth": 0.80076, "loss": 28.73112, "grad_norm": 65.15598, "time": 1.47012} -{"mode": "train", "epoch": 1, "iter": 108, "lr": 0.00014, "memory": 49163, "data_time": 0.06971, "loss_cls_0": 0.90118, "loss_box_0": 1.75261, "loss_cns_0": 0.61785, "loss_yns_0": 0.15763, "loss_cls_1": 0.92544, "loss_box_1": 1.96818, "loss_cns_1": 0.62077, "loss_yns_1": 0.15835, "loss_cls_2": 0.99299, "loss_box_2": 1.91949, "loss_cns_2": 0.6411, "loss_yns_2": 0.15754, "loss_cls_3": 1.0026, "loss_box_3": 1.9221, "loss_cns_3": 0.6465, "loss_yns_3": 0.16048, "loss_cls_4": 1.01429, "loss_box_4": 1.92745, "loss_cns_4": 0.64886, "loss_yns_4": 0.16038, "loss_cls_5": 1.03813, "loss_box_5": 1.92323, "loss_cns_5": 0.64127, "loss_yns_5": 0.1603, "loss_cls_dn_0": 0.23974, "loss_box_dn_0": 0.81428, "loss_cls_dn_1": 0.17833, "loss_box_dn_1": 0.82591, "loss_cls_dn_2": 0.18425, "loss_box_dn_2": 0.80578, "loss_cls_dn_3": 0.18614, "loss_box_dn_3": 0.82144, "loss_cls_dn_4": 0.19404, "loss_box_dn_4": 0.83945, "loss_cls_dn_5": 0.20597, "loss_box_dn_5": 0.84859, "loss_dense_depth": 0.7997, "loss": 29.00235, "grad_norm": 72.52928, "time": 1.47707} -{"mode": "train", "epoch": 1, "iter": 109, "lr": 0.00014, "memory": 49163, "data_time": 0.07521, "loss_cls_0": 0.92035, "loss_box_0": 1.76757, "loss_cns_0": 0.60814, "loss_yns_0": 0.15566, "loss_cls_1": 0.94299, "loss_box_1": 1.96884, "loss_cns_1": 0.62548, "loss_yns_1": 0.15951, "loss_cls_2": 0.97723, "loss_box_2": 1.92555, "loss_cns_2": 0.63882, "loss_yns_2": 0.15819, "loss_cls_3": 0.99655, "loss_box_3": 1.90225, "loss_cns_3": 0.64284, "loss_yns_3": 0.16035, "loss_cls_4": 1.00703, "loss_box_4": 1.92255, "loss_cns_4": 0.64411, "loss_yns_4": 0.16049, "loss_cls_5": 1.0567, "loss_box_5": 1.88108, "loss_cns_5": 0.6406, "loss_yns_5": 0.15995, "loss_cls_dn_0": 0.2415, "loss_box_dn_0": 0.81397, "loss_cls_dn_1": 0.17504, "loss_box_dn_1": 0.84871, "loss_cls_dn_2": 0.17737, "loss_box_dn_2": 0.83009, "loss_cls_dn_3": 0.18194, "loss_box_dn_3": 0.84052, "loss_cls_dn_4": 0.18672, "loss_box_dn_4": 0.86684, "loss_cls_dn_5": 0.19428, "loss_box_dn_5": 0.87154, "loss_dense_depth": 0.82442, "loss": 29.07578, "grad_norm": 51.32078, "time": 1.52071} -{"mode": "train", "epoch": 1, "iter": 110, "lr": 0.00014, "memory": 49163, "data_time": 0.07535, "loss_cls_0": 0.92391, "loss_box_0": 1.80856, "loss_cns_0": 0.60958, "loss_yns_0": 0.15547, "loss_cls_1": 0.95528, "loss_box_1": 2.00752, "loss_cns_1": 0.62139, "loss_yns_1": 0.15835, "loss_cls_2": 1.00315, "loss_box_2": 1.91987, "loss_cns_2": 0.64196, "loss_yns_2": 0.15656, "loss_cls_3": 1.00875, "loss_box_3": 1.894, "loss_cns_3": 0.64564, "loss_yns_3": 0.15982, "loss_cls_4": 1.02501, "loss_box_4": 1.93183, "loss_cns_4": 0.64289, "loss_yns_4": 0.15848, "loss_cls_5": 1.07738, "loss_box_5": 1.89631, "loss_cns_5": 0.63802, "loss_yns_5": 0.15889, "loss_cls_dn_0": 0.24558, "loss_box_dn_0": 0.80773, "loss_cls_dn_1": 0.17041, "loss_box_dn_1": 0.82439, "loss_cls_dn_2": 0.17071, "loss_box_dn_2": 0.78779, "loss_cls_dn_3": 0.17529, "loss_box_dn_3": 0.79649, "loss_cls_dn_4": 0.18604, "loss_box_dn_4": 0.82742, "loss_cls_dn_5": 0.19509, "loss_box_dn_5": 0.83971, "loss_dense_depth": 0.81121, "loss": 29.0365, "grad_norm": 53.62466, "time": 1.49748} -{"mode": "train", "epoch": 1, "iter": 111, "lr": 0.00014, "memory": 49163, "data_time": 0.07376, "loss_cls_0": 0.90965, "loss_box_0": 1.8267, "loss_cns_0": 0.6122, "loss_yns_0": 0.15528, "loss_cls_1": 0.95746, "loss_box_1": 1.96615, "loss_cns_1": 0.62742, "loss_yns_1": 0.15678, "loss_cls_2": 1.02213, "loss_box_2": 1.8877, "loss_cns_2": 0.6424, "loss_yns_2": 0.1576, "loss_cls_3": 1.01161, "loss_box_3": 1.87319, "loss_cns_3": 0.64663, "loss_yns_3": 0.15876, "loss_cls_4": 1.01604, "loss_box_4": 1.88357, "loss_cns_4": 0.647, "loss_yns_4": 0.15875, "loss_cls_5": 1.05271, "loss_box_5": 1.88648, "loss_cns_5": 0.64793, "loss_yns_5": 0.1566, "loss_cls_dn_0": 0.24535, "loss_box_dn_0": 0.80917, "loss_cls_dn_1": 0.17313, "loss_box_dn_1": 0.82912, "loss_cls_dn_2": 0.1827, "loss_box_dn_2": 0.79336, "loss_cls_dn_3": 0.1827, "loss_box_dn_3": 0.80643, "loss_cls_dn_4": 0.19195, "loss_box_dn_4": 0.82583, "loss_cls_dn_5": 0.20502, "loss_box_dn_5": 0.8419, "loss_dense_depth": 0.84724, "loss": 28.99464, "grad_norm": 57.61362, "time": 1.48401} -{"mode": "train", "epoch": 1, "iter": 112, "lr": 0.00014, "memory": 49163, "data_time": 0.07445, "loss_cls_0": 0.93609, "loss_box_0": 1.84412, "loss_cns_0": 0.60875, "loss_yns_0": 0.15669, "loss_cls_1": 0.98088, "loss_box_1": 1.96909, "loss_cns_1": 0.62457, "loss_yns_1": 0.15906, "loss_cls_2": 1.01754, "loss_box_2": 1.90863, "loss_cns_2": 0.63863, "loss_yns_2": 0.15839, "loss_cls_3": 1.01364, "loss_box_3": 1.92196, "loss_cns_3": 0.64191, "loss_yns_3": 0.15914, "loss_cls_4": 1.01974, "loss_box_4": 1.90153, "loss_cns_4": 0.64416, "loss_yns_4": 0.16007, "loss_cls_5": 1.03472, "loss_box_5": 1.89965, "loss_cns_5": 0.6434, "loss_yns_5": 0.159, "loss_cls_dn_0": 0.25245, "loss_box_dn_0": 0.81252, "loss_cls_dn_1": 0.17988, "loss_box_dn_1": 0.84626, "loss_cls_dn_2": 0.18597, "loss_box_dn_2": 0.8178, "loss_cls_dn_3": 0.18563, "loss_box_dn_3": 0.83406, "loss_cls_dn_4": 0.19655, "loss_box_dn_4": 0.83616, "loss_cls_dn_5": 0.20497, "loss_box_dn_5": 0.84414, "loss_dense_depth": 0.86142, "loss": 29.25915, "grad_norm": 59.86788, "time": 1.49989} -{"mode": "train", "epoch": 1, "iter": 113, "lr": 0.00014, "memory": 49163, "data_time": 0.07388, "loss_cls_0": 0.8879, "loss_box_0": 1.8012, "loss_cns_0": 0.61244, "loss_yns_0": 0.15467, "loss_cls_1": 0.94407, "loss_box_1": 1.90106, "loss_cns_1": 0.63215, "loss_yns_1": 0.15421, "loss_cls_2": 1.00162, "loss_box_2": 1.85799, "loss_cns_2": 0.64648, "loss_yns_2": 0.15518, "loss_cls_3": 0.99135, "loss_box_3": 1.88155, "loss_cns_3": 0.64364, "loss_yns_3": 0.15529, "loss_cls_4": 0.9973, "loss_box_4": 1.85132, "loss_cns_4": 0.64311, "loss_yns_4": 0.15623, "loss_cls_5": 1.0087, "loss_box_5": 1.85744, "loss_cns_5": 0.64067, "loss_yns_5": 0.15602, "loss_cls_dn_0": 0.23847, "loss_box_dn_0": 0.81168, "loss_cls_dn_1": 0.17864, "loss_box_dn_1": 0.82002, "loss_cls_dn_2": 0.17673, "loss_box_dn_2": 0.79225, "loss_cls_dn_3": 0.17598, "loss_box_dn_3": 0.80323, "loss_cls_dn_4": 0.18652, "loss_box_dn_4": 0.79184, "loss_cls_dn_5": 0.19081, "loss_box_dn_5": 0.79649, "loss_dense_depth": 0.80772, "loss": 28.50196, "grad_norm": 54.76088, "time": 1.48594} -{"mode": "train", "epoch": 1, "iter": 114, "lr": 0.00015, "memory": 49163, "data_time": 0.07278, "loss_cls_0": 0.9035, "loss_box_0": 1.80991, "loss_cns_0": 0.61589, "loss_yns_0": 0.15572, "loss_cls_1": 0.97527, "loss_box_1": 1.90574, "loss_cns_1": 0.63758, "loss_yns_1": 0.15813, "loss_cls_2": 1.02612, "loss_box_2": 1.8325, "loss_cns_2": 0.65658, "loss_yns_2": 0.15667, "loss_cls_3": 1.0124, "loss_box_3": 1.82944, "loss_cns_3": 0.65383, "loss_yns_3": 0.15727, "loss_cls_4": 1.007, "loss_box_4": 1.83331, "loss_cns_4": 0.65579, "loss_yns_4": 0.15787, "loss_cls_5": 1.00775, "loss_box_5": 1.82879, "loss_cns_5": 0.64723, "loss_yns_5": 0.15775, "loss_cls_dn_0": 0.24454, "loss_box_dn_0": 0.79804, "loss_cls_dn_1": 0.17055, "loss_box_dn_1": 0.78433, "loss_cls_dn_2": 0.16751, "loss_box_dn_2": 0.75164, "loss_cls_dn_3": 0.16968, "loss_box_dn_3": 0.75245, "loss_cls_dn_4": 0.17989, "loss_box_dn_4": 0.749, "loss_cls_dn_5": 0.1852, "loss_box_dn_5": 0.75239, "loss_dense_depth": 0.82994, "loss": 28.31718, "grad_norm": 37.75026, "time": 1.4931} -{"mode": "train", "epoch": 1, "iter": 115, "lr": 0.00015, "memory": 49163, "data_time": 0.08054, "loss_cls_0": 0.88292, "loss_box_0": 1.79119, "loss_cns_0": 0.62073, "loss_yns_0": 0.15481, "loss_cls_1": 0.95887, "loss_box_1": 1.96415, "loss_cns_1": 0.63637, "loss_yns_1": 0.1565, "loss_cls_2": 0.98256, "loss_box_2": 1.88865, "loss_cns_2": 0.65066, "loss_yns_2": 0.15569, "loss_cls_3": 0.98473, "loss_box_3": 1.86918, "loss_cns_3": 0.64805, "loss_yns_3": 0.15482, "loss_cls_4": 0.99256, "loss_box_4": 1.91081, "loss_cns_4": 0.65222, "loss_yns_4": 0.15759, "loss_cls_5": 0.99572, "loss_box_5": 1.88102, "loss_cns_5": 0.64723, "loss_yns_5": 0.15533, "loss_cls_dn_0": 0.23333, "loss_box_dn_0": 0.80176, "loss_cls_dn_1": 0.1701, "loss_box_dn_1": 0.78171, "loss_cls_dn_2": 0.16889, "loss_box_dn_2": 0.74975, "loss_cls_dn_3": 0.16863, "loss_box_dn_3": 0.74304, "loss_cls_dn_4": 0.17785, "loss_box_dn_4": 0.76466, "loss_cls_dn_5": 0.1875, "loss_box_dn_5": 0.76057, "loss_dense_depth": 0.77818, "loss": 28.37831, "grad_norm": 60.63051, "time": 1.48719} -{"mode": "train", "epoch": 1, "iter": 116, "lr": 0.00015, "memory": 49163, "data_time": 0.0788, "loss_cls_0": 0.93506, "loss_box_0": 1.79643, "loss_cns_0": 0.61967, "loss_yns_0": 0.15562, "loss_cls_1": 0.95718, "loss_box_1": 1.9494, "loss_cns_1": 0.63482, "loss_yns_1": 0.15662, "loss_cls_2": 0.9845, "loss_box_2": 1.88987, "loss_cns_2": 0.64742, "loss_yns_2": 0.15578, "loss_cls_3": 0.99889, "loss_box_3": 1.86856, "loss_cns_3": 0.64444, "loss_yns_3": 0.15601, "loss_cls_4": 0.99994, "loss_box_4": 1.90355, "loss_cns_4": 0.64622, "loss_yns_4": 0.15815, "loss_cls_5": 1.00387, "loss_box_5": 1.88118, "loss_cns_5": 0.6502, "loss_yns_5": 0.1564, "loss_cls_dn_0": 0.23613, "loss_box_dn_0": 0.79729, "loss_cls_dn_1": 0.17091, "loss_box_dn_1": 0.79796, "loss_cls_dn_2": 0.16913, "loss_box_dn_2": 0.78201, "loss_cls_dn_3": 0.16951, "loss_box_dn_3": 0.78508, "loss_cls_dn_4": 0.17921, "loss_box_dn_4": 0.81311, "loss_cls_dn_5": 0.19557, "loss_box_dn_5": 0.81489, "loss_dense_depth": 0.83217, "loss": 28.69276, "grad_norm": 57.63181, "time": 1.49812} -{"mode": "train", "epoch": 1, "iter": 117, "lr": 0.00015, "memory": 49163, "data_time": 0.0759, "loss_cls_0": 0.89894, "loss_box_0": 1.75787, "loss_cns_0": 0.62377, "loss_yns_0": 0.15366, "loss_cls_1": 0.94566, "loss_box_1": 1.92272, "loss_cns_1": 0.6361, "loss_yns_1": 0.15601, "loss_cls_2": 0.98894, "loss_box_2": 1.87444, "loss_cns_2": 0.65268, "loss_yns_2": 0.15748, "loss_cls_3": 0.98923, "loss_box_3": 1.86135, "loss_cns_3": 0.6484, "loss_yns_3": 0.15657, "loss_cls_4": 1.00077, "loss_box_4": 1.86772, "loss_cns_4": 0.64678, "loss_yns_4": 0.15488, "loss_cls_5": 0.99797, "loss_box_5": 1.88888, "loss_cns_5": 0.65385, "loss_yns_5": 0.1555, "loss_cls_dn_0": 0.23373, "loss_box_dn_0": 0.79744, "loss_cls_dn_1": 0.16998, "loss_box_dn_1": 0.82131, "loss_cls_dn_2": 0.16876, "loss_box_dn_2": 0.80417, "loss_cls_dn_3": 0.16995, "loss_box_dn_3": 0.81155, "loss_cls_dn_4": 0.17961, "loss_box_dn_4": 0.83015, "loss_cls_dn_5": 0.1939, "loss_box_dn_5": 0.85213, "loss_dense_depth": 0.80439, "loss": 28.62724, "grad_norm": 44.4085, "time": 1.49815} -{"mode": "train", "epoch": 1, "iter": 118, "lr": 0.00015, "memory": 49163, "data_time": 0.07536, "loss_cls_0": 0.90208, "loss_box_0": 1.74445, "loss_cns_0": 0.61939, "loss_yns_0": 0.15614, "loss_cls_1": 0.95578, "loss_box_1": 1.87058, "loss_cns_1": 0.63866, "loss_yns_1": 0.15631, "loss_cls_2": 0.98613, "loss_box_2": 1.81532, "loss_cns_2": 0.65626, "loss_yns_2": 0.15497, "loss_cls_3": 0.99882, "loss_box_3": 1.81113, "loss_cns_3": 0.65333, "loss_yns_3": 0.15568, "loss_cls_4": 1.02574, "loss_box_4": 1.80238, "loss_cns_4": 0.65186, "loss_yns_4": 0.15524, "loss_cls_5": 1.0075, "loss_box_5": 1.83185, "loss_cns_5": 0.65037, "loss_yns_5": 0.15547, "loss_cls_dn_0": 0.23753, "loss_box_dn_0": 0.79677, "loss_cls_dn_1": 0.17047, "loss_box_dn_1": 0.83907, "loss_cls_dn_2": 0.1675, "loss_box_dn_2": 0.81643, "loss_cls_dn_3": 0.16986, "loss_box_dn_3": 0.82321, "loss_cls_dn_4": 0.18305, "loss_box_dn_4": 0.83223, "loss_cls_dn_5": 0.19254, "loss_box_dn_5": 0.85562, "loss_dense_depth": 0.81343, "loss": 28.45313, "grad_norm": 50.72441, "time": 1.50647} -{"mode": "train", "epoch": 1, "iter": 119, "lr": 0.00015, "memory": 49163, "data_time": 0.07023, "loss_cls_0": 0.92118, "loss_box_0": 1.72955, "loss_cns_0": 0.615, "loss_yns_0": 0.15416, "loss_cls_1": 0.95341, "loss_box_1": 1.903, "loss_cns_1": 0.64016, "loss_yns_1": 0.15514, "loss_cls_2": 0.98033, "loss_box_2": 1.84786, "loss_cns_2": 0.65203, "loss_yns_2": 0.15418, "loss_cls_3": 0.99013, "loss_box_3": 1.82436, "loss_cns_3": 0.65107, "loss_yns_3": 0.15456, "loss_cls_4": 1.01342, "loss_box_4": 1.8073, "loss_cns_4": 0.65067, "loss_yns_4": 0.15369, "loss_cls_5": 0.99661, "loss_box_5": 1.82124, "loss_cns_5": 0.64819, "loss_yns_5": 0.15368, "loss_cls_dn_0": 0.23643, "loss_box_dn_0": 0.78666, "loss_cls_dn_1": 0.16938, "loss_box_dn_1": 0.85228, "loss_cls_dn_2": 0.16833, "loss_box_dn_2": 0.82055, "loss_cls_dn_3": 0.16917, "loss_box_dn_3": 0.81127, "loss_cls_dn_4": 0.1815, "loss_box_dn_4": 0.81188, "loss_cls_dn_5": 0.18213, "loss_box_dn_5": 0.82242, "loss_dense_depth": 0.80018, "loss": 28.38312, "grad_norm": 47.92738, "time": 1.4894} -{"mode": "train", "epoch": 1, "iter": 120, "lr": 0.00015, "memory": 49163, "data_time": 0.08309, "loss_cls_0": 0.88562, "loss_box_0": 1.76009, "loss_cns_0": 0.61407, "loss_yns_0": 0.15439, "loss_cls_1": 0.95377, "loss_box_1": 1.92114, "loss_cns_1": 0.63788, "loss_yns_1": 0.15607, "loss_cls_2": 0.97577, "loss_box_2": 1.87127, "loss_cns_2": 0.64689, "loss_yns_2": 0.15584, "loss_cls_3": 0.98233, "loss_box_3": 1.83153, "loss_cns_3": 0.64904, "loss_yns_3": 0.15502, "loss_cls_4": 0.99423, "loss_box_4": 1.81848, "loss_cns_4": 0.64947, "loss_yns_4": 0.15595, "loss_cls_5": 0.99183, "loss_box_5": 1.80365, "loss_cns_5": 0.64822, "loss_yns_5": 0.1561, "loss_cls_dn_0": 0.23295, "loss_box_dn_0": 0.79768, "loss_cls_dn_1": 0.16769, "loss_box_dn_1": 0.79804, "loss_cls_dn_2": 0.16697, "loss_box_dn_2": 0.76446, "loss_cls_dn_3": 0.16527, "loss_box_dn_3": 0.74913, "loss_cls_dn_4": 0.17747, "loss_box_dn_4": 0.74301, "loss_cls_dn_5": 0.17986, "loss_box_dn_5": 0.73933, "loss_dense_depth": 0.78737, "loss": 28.03786, "grad_norm": 45.21702, "time": 1.49336} -{"mode": "train", "epoch": 1, "iter": 121, "lr": 0.00015, "memory": 49163, "data_time": 0.0775, "loss_cls_0": 0.89838, "loss_box_0": 1.77181, "loss_cns_0": 0.62219, "loss_yns_0": 0.15591, "loss_cls_1": 0.95517, "loss_box_1": 1.96126, "loss_cns_1": 0.63362, "loss_yns_1": 0.15504, "loss_cls_2": 0.9868, "loss_box_2": 1.88072, "loss_cns_2": 0.64491, "loss_yns_2": 0.15749, "loss_cls_3": 0.99721, "loss_box_3": 1.87978, "loss_cns_3": 0.64689, "loss_yns_3": 0.15484, "loss_cls_4": 1.01817, "loss_box_4": 1.86399, "loss_cns_4": 0.64796, "loss_yns_4": 0.15406, "loss_cls_5": 1.00912, "loss_box_5": 1.8629, "loss_cns_5": 0.64883, "loss_yns_5": 0.15519, "loss_cls_dn_0": 0.23247, "loss_box_dn_0": 0.79986, "loss_cls_dn_1": 0.16296, "loss_box_dn_1": 0.78463, "loss_cls_dn_2": 0.16233, "loss_box_dn_2": 0.74938, "loss_cls_dn_3": 0.16103, "loss_box_dn_3": 0.75328, "loss_cls_dn_4": 0.17088, "loss_box_dn_4": 0.74986, "loss_cls_dn_5": 0.17801, "loss_box_dn_5": 0.75487, "loss_dense_depth": 0.77908, "loss": 28.30089, "grad_norm": 45.23407, "time": 1.56796} -{"mode": "train", "epoch": 1, "iter": 122, "lr": 0.00015, "memory": 49163, "data_time": 0.16147, "loss_cls_0": 0.90411, "loss_box_0": 1.76131, "loss_cns_0": 0.62385, "loss_yns_0": 0.15682, "loss_cls_1": 0.94137, "loss_box_1": 1.91759, "loss_cns_1": 0.63262, "loss_yns_1": 0.15578, "loss_cls_2": 0.992, "loss_box_2": 1.8544, "loss_cns_2": 0.65093, "loss_yns_2": 0.15924, "loss_cls_3": 0.99563, "loss_box_3": 1.86062, "loss_cns_3": 0.64982, "loss_yns_3": 0.1567, "loss_cls_4": 1.016, "loss_box_4": 1.83435, "loss_cns_4": 0.64848, "loss_yns_4": 0.15877, "loss_cls_5": 1.0115, "loss_box_5": 1.84054, "loss_cns_5": 0.6461, "loss_yns_5": 0.15795, "loss_cls_dn_0": 0.23309, "loss_box_dn_0": 0.79187, "loss_cls_dn_1": 0.16288, "loss_box_dn_1": 0.77216, "loss_cls_dn_2": 0.16466, "loss_box_dn_2": 0.74458, "loss_cls_dn_3": 0.16265, "loss_box_dn_3": 0.75866, "loss_cls_dn_4": 0.17171, "loss_box_dn_4": 0.76037, "loss_cls_dn_5": 0.18055, "loss_box_dn_5": 0.76854, "loss_dense_depth": 0.77663, "loss": 28.17487, "grad_norm": 35.69691, "time": 1.56168} -{"mode": "train", "epoch": 1, "iter": 123, "lr": 0.00015, "memory": 49163, "data_time": 0.07875, "loss_cls_0": 0.88235, "loss_box_0": 1.75879, "loss_cns_0": 0.62019, "loss_yns_0": 0.15576, "loss_cls_1": 0.95484, "loss_box_1": 1.88608, "loss_cns_1": 0.63152, "loss_yns_1": 0.15529, "loss_cls_2": 1.00243, "loss_box_2": 1.84249, "loss_cns_2": 0.65213, "loss_yns_2": 0.15773, "loss_cls_3": 0.99134, "loss_box_3": 1.83079, "loss_cns_3": 0.64871, "loss_yns_3": 0.15591, "loss_cls_4": 1.00081, "loss_box_4": 1.8205, "loss_cns_4": 0.64812, "loss_yns_4": 0.15729, "loss_cls_5": 0.9947, "loss_box_5": 1.79904, "loss_cns_5": 0.64572, "loss_yns_5": 0.15688, "loss_cls_dn_0": 0.22844, "loss_box_dn_0": 0.79277, "loss_cls_dn_1": 0.17036, "loss_box_dn_1": 0.80331, "loss_cls_dn_2": 0.17115, "loss_box_dn_2": 0.78274, "loss_cls_dn_3": 0.16825, "loss_box_dn_3": 0.79901, "loss_cls_dn_4": 0.17779, "loss_box_dn_4": 0.80983, "loss_cls_dn_5": 0.18475, "loss_box_dn_5": 0.81004, "loss_dense_depth": 0.79148, "loss": 28.23932, "grad_norm": 47.97798, "time": 1.52459} -{"mode": "train", "epoch": 1, "iter": 124, "lr": 0.00015, "memory": 49163, "data_time": 0.07933, "loss_cls_0": 0.88258, "loss_box_0": 1.74856, "loss_cns_0": 0.62199, "loss_yns_0": 0.15717, "loss_cls_1": 0.93135, "loss_box_1": 1.91687, "loss_cns_1": 0.63269, "loss_yns_1": 0.15357, "loss_cls_2": 0.98108, "loss_box_2": 1.87988, "loss_cns_2": 0.64903, "loss_yns_2": 0.15528, "loss_cls_3": 0.98584, "loss_box_3": 1.85767, "loss_cns_3": 0.64895, "loss_yns_3": 0.15671, "loss_cls_4": 0.99743, "loss_box_4": 1.87481, "loss_cns_4": 0.65135, "loss_yns_4": 0.15668, "loss_cls_5": 0.9929, "loss_box_5": 1.84049, "loss_cns_5": 0.65333, "loss_yns_5": 0.15618, "loss_cls_dn_0": 0.22894, "loss_box_dn_0": 0.79066, "loss_cls_dn_1": 0.16447, "loss_box_dn_1": 0.80179, "loss_cls_dn_2": 0.15953, "loss_box_dn_2": 0.77808, "loss_cls_dn_3": 0.16135, "loss_box_dn_3": 0.78189, "loss_cls_dn_4": 0.17185, "loss_box_dn_4": 0.79464, "loss_cls_dn_5": 0.17844, "loss_box_dn_5": 0.78908, "loss_dense_depth": 0.75972, "loss": 28.24281, "grad_norm": 42.38208, "time": 1.56982} -{"mode": "train", "epoch": 1, "iter": 125, "lr": 0.00015, "memory": 49163, "data_time": 0.08685, "loss_cls_0": 0.9007, "loss_box_0": 1.75737, "loss_cns_0": 0.6168, "loss_yns_0": 0.1553, "loss_cls_1": 0.93509, "loss_box_1": 1.87048, "loss_cns_1": 0.62677, "loss_yns_1": 0.15469, "loss_cls_2": 0.99707, "loss_box_2": 1.81694, "loss_cns_2": 0.64748, "loss_yns_2": 0.1555, "loss_cls_3": 1.00094, "loss_box_3": 1.80195, "loss_cns_3": 0.6492, "loss_yns_3": 0.15624, "loss_cls_4": 1.0094, "loss_box_4": 1.80697, "loss_cns_4": 0.6526, "loss_yns_4": 0.15865, "loss_cls_5": 1.00602, "loss_box_5": 1.79101, "loss_cns_5": 0.65086, "loss_yns_5": 0.1575, "loss_cls_dn_0": 0.23443, "loss_box_dn_0": 0.78304, "loss_cls_dn_1": 0.16508, "loss_box_dn_1": 0.81072, "loss_cls_dn_2": 0.16033, "loss_box_dn_2": 0.77702, "loss_cls_dn_3": 0.16215, "loss_box_dn_3": 0.77104, "loss_cls_dn_4": 0.17071, "loss_box_dn_4": 0.78051, "loss_cls_dn_5": 0.18021, "loss_box_dn_5": 0.78198, "loss_dense_depth": 0.76437, "loss": 28.01714, "grad_norm": 34.51907, "time": 1.50252} -{"mode": "train", "epoch": 1, "iter": 126, "lr": 0.00015, "memory": 49163, "data_time": 0.07238, "loss_cls_0": 0.89397, "loss_box_0": 1.76886, "loss_cns_0": 0.6132, "loss_yns_0": 0.155, "loss_cls_1": 0.92617, "loss_box_1": 1.90747, "loss_cns_1": 0.62335, "loss_yns_1": 0.15381, "loss_cls_2": 0.98684, "loss_box_2": 1.83721, "loss_cns_2": 0.64815, "loss_yns_2": 0.15631, "loss_cls_3": 1.00361, "loss_box_3": 1.81054, "loss_cns_3": 0.65045, "loss_yns_3": 0.15542, "loss_cls_4": 1.01486, "loss_box_4": 1.81729, "loss_cns_4": 0.64793, "loss_yns_4": 0.15509, "loss_cls_5": 1.01882, "loss_box_5": 1.83635, "loss_cns_5": 0.64709, "loss_yns_5": 0.15529, "loss_cls_dn_0": 0.23334, "loss_box_dn_0": 0.79064, "loss_cls_dn_1": 0.16855, "loss_box_dn_1": 0.77908, "loss_cls_dn_2": 0.16426, "loss_box_dn_2": 0.74334, "loss_cls_dn_3": 0.16571, "loss_box_dn_3": 0.73614, "loss_cls_dn_4": 0.17245, "loss_box_dn_4": 0.74186, "loss_cls_dn_5": 0.18153, "loss_box_dn_5": 0.75316, "loss_dense_depth": 0.75135, "loss": 27.96448, "grad_norm": 40.55269, "time": 1.49511} -{"mode": "train", "epoch": 1, "iter": 127, "lr": 0.00015, "memory": 49163, "data_time": 0.07663, "loss_cls_0": 0.88993, "loss_box_0": 1.75093, "loss_cns_0": 0.61612, "loss_yns_0": 0.15542, "loss_cls_1": 0.93844, "loss_box_1": 1.85033, "loss_cns_1": 0.6304, "loss_yns_1": 0.15694, "loss_cls_2": 0.98465, "loss_box_2": 1.7899, "loss_cns_2": 0.65076, "loss_yns_2": 0.15984, "loss_cls_3": 1.00557, "loss_box_3": 1.75461, "loss_cns_3": 0.6518, "loss_yns_3": 0.15738, "loss_cls_4": 1.00953, "loss_box_4": 1.74113, "loss_cns_4": 0.65259, "loss_yns_4": 0.1557, "loss_cls_5": 1.00115, "loss_box_5": 1.74002, "loss_cns_5": 0.65139, "loss_yns_5": 0.15522, "loss_cls_dn_0": 0.23212, "loss_box_dn_0": 0.79031, "loss_cls_dn_1": 0.16331, "loss_box_dn_1": 0.75532, "loss_cls_dn_2": 0.1634, "loss_box_dn_2": 0.72618, "loss_cls_dn_3": 0.16226, "loss_box_dn_3": 0.71383, "loss_cls_dn_4": 0.16924, "loss_box_dn_4": 0.71287, "loss_cls_dn_5": 0.17824, "loss_box_dn_5": 0.7192, "loss_dense_depth": 0.75752, "loss": 27.49353, "grad_norm": 33.83744, "time": 1.48907} -{"mode": "train", "epoch": 1, "iter": 128, "lr": 0.00015, "memory": 49163, "data_time": 0.0744, "loss_cls_0": 0.90033, "loss_box_0": 1.73419, "loss_cns_0": 0.61755, "loss_yns_0": 0.15589, "loss_cls_1": 0.95969, "loss_box_1": 1.81581, "loss_cns_1": 0.62891, "loss_yns_1": 0.15804, "loss_cls_2": 0.99586, "loss_box_2": 1.78048, "loss_cns_2": 0.64632, "loss_yns_2": 0.15925, "loss_cls_3": 1.00709, "loss_box_3": 1.78781, "loss_cns_3": 0.64568, "loss_yns_3": 0.15634, "loss_cls_4": 1.01682, "loss_box_4": 1.77573, "loss_cns_4": 0.64477, "loss_yns_4": 0.15837, "loss_cls_5": 0.99937, "loss_box_5": 1.78099, "loss_cns_5": 0.64175, "loss_yns_5": 0.15637, "loss_cls_dn_0": 0.23191, "loss_box_dn_0": 0.79598, "loss_cls_dn_1": 0.1584, "loss_box_dn_1": 0.7295, "loss_cls_dn_2": 0.15629, "loss_box_dn_2": 0.71163, "loss_cls_dn_3": 0.15813, "loss_box_dn_3": 0.71229, "loss_cls_dn_4": 0.1687, "loss_box_dn_4": 0.7161, "loss_cls_dn_5": 0.17694, "loss_box_dn_5": 0.72909, "loss_dense_depth": 0.77588, "loss": 27.54428, "grad_norm": 46.32293, "time": 1.49297} -{"mode": "train", "epoch": 1, "iter": 129, "lr": 0.00015, "memory": 49163, "data_time": 0.07933, "loss_cls_0": 0.8776, "loss_box_0": 1.72044, "loss_cns_0": 0.61741, "loss_yns_0": 0.15582, "loss_cls_1": 0.94488, "loss_box_1": 1.82715, "loss_cns_1": 0.63438, "loss_yns_1": 0.15708, "loss_cls_2": 0.98694, "loss_box_2": 1.77483, "loss_cns_2": 0.64878, "loss_yns_2": 0.15846, "loss_cls_3": 0.99001, "loss_box_3": 1.76492, "loss_cns_3": 0.64934, "loss_yns_3": 0.15707, "loss_cls_4": 0.9991, "loss_box_4": 1.76271, "loss_cns_4": 0.65224, "loss_yns_4": 0.15934, "loss_cls_5": 0.98462, "loss_box_5": 1.76488, "loss_cns_5": 0.64685, "loss_yns_5": 0.15549, "loss_cls_dn_0": 0.22827, "loss_box_dn_0": 0.78669, "loss_cls_dn_1": 0.15939, "loss_box_dn_1": 0.73564, "loss_cls_dn_2": 0.15753, "loss_box_dn_2": 0.7168, "loss_cls_dn_3": 0.158, "loss_box_dn_3": 0.71522, "loss_cls_dn_4": 0.16822, "loss_box_dn_4": 0.72346, "loss_cls_dn_5": 0.17287, "loss_box_dn_5": 0.7391, "loss_dense_depth": 0.74683, "loss": 27.39838, "grad_norm": 34.35713, "time": 1.50812} -{"mode": "train", "epoch": 1, "iter": 130, "lr": 0.00015, "memory": 49163, "data_time": 0.07561, "loss_cls_0": 0.89783, "loss_box_0": 1.72738, "loss_cns_0": 0.61141, "loss_yns_0": 0.15169, "loss_cls_1": 0.96156, "loss_box_1": 1.84239, "loss_cns_1": 0.63232, "loss_yns_1": 0.15476, "loss_cls_2": 1.00505, "loss_box_2": 1.80162, "loss_cns_2": 0.64566, "loss_yns_2": 0.15383, "loss_cls_3": 1.00761, "loss_box_3": 1.78356, "loss_cns_3": 0.64649, "loss_yns_3": 0.15737, "loss_cls_4": 1.01334, "loss_box_4": 1.79909, "loss_cns_4": 0.64786, "loss_yns_4": 0.15823, "loss_cls_5": 1.00287, "loss_box_5": 1.78721, "loss_cns_5": 0.64404, "loss_yns_5": 0.15372, "loss_cls_dn_0": 0.23124, "loss_box_dn_0": 0.78622, "loss_cls_dn_1": 0.15818, "loss_box_dn_1": 0.76074, "loss_cls_dn_2": 0.15795, "loss_box_dn_2": 0.74936, "loss_cls_dn_3": 0.15918, "loss_box_dn_3": 0.74458, "loss_cls_dn_4": 0.16632, "loss_box_dn_4": 0.75625, "loss_cls_dn_5": 0.17359, "loss_box_dn_5": 0.76475, "loss_dense_depth": 0.78346, "loss": 27.77869, "grad_norm": 51.06961, "time": 1.49979} -{"mode": "train", "epoch": 1, "iter": 131, "lr": 0.00015, "memory": 49163, "data_time": 0.07557, "loss_cls_0": 0.91353, "loss_box_0": 1.72541, "loss_cns_0": 0.61732, "loss_yns_0": 0.15189, "loss_cls_1": 0.95651, "loss_box_1": 1.85296, "loss_cns_1": 0.63864, "loss_yns_1": 0.15618, "loss_cls_2": 0.98597, "loss_box_2": 1.81532, "loss_cns_2": 0.64716, "loss_yns_2": 0.15675, "loss_cls_3": 1.01286, "loss_box_3": 1.80721, "loss_cns_3": 0.64843, "loss_yns_3": 0.15618, "loss_cls_4": 1.01823, "loss_box_4": 1.8129, "loss_cns_4": 0.65047, "loss_yns_4": 0.15632, "loss_cls_5": 0.99634, "loss_box_5": 1.79108, "loss_cns_5": 0.6469, "loss_yns_5": 0.15387, "loss_cls_dn_0": 0.23139, "loss_box_dn_0": 0.7913, "loss_cls_dn_1": 0.15882, "loss_box_dn_1": 0.76505, "loss_cls_dn_2": 0.15994, "loss_box_dn_2": 0.74667, "loss_cls_dn_3": 0.1578, "loss_box_dn_3": 0.73951, "loss_cls_dn_4": 0.16531, "loss_box_dn_4": 0.74785, "loss_cls_dn_5": 0.17487, "loss_box_dn_5": 0.75098, "loss_dense_depth": 0.79819, "loss": 27.8561, "grad_norm": 46.08059, "time": 1.48442} -{"mode": "train", "epoch": 1, "iter": 132, "lr": 0.00015, "memory": 49163, "data_time": 0.07695, "loss_cls_0": 0.8972, "loss_box_0": 1.756, "loss_cns_0": 0.61845, "loss_yns_0": 0.14941, "loss_cls_1": 0.95296, "loss_box_1": 1.83206, "loss_cns_1": 0.63851, "loss_yns_1": 0.15348, "loss_cls_2": 0.99063, "loss_box_2": 1.79106, "loss_cns_2": 0.64561, "loss_yns_2": 0.15377, "loss_cls_3": 0.99774, "loss_box_3": 1.80527, "loss_cns_3": 0.64764, "loss_yns_3": 0.15237, "loss_cls_4": 0.99788, "loss_box_4": 1.81096, "loss_cns_4": 0.66031, "loss_yns_4": 0.15317, "loss_cls_5": 0.98417, "loss_box_5": 1.80867, "loss_cns_5": 0.64846, "loss_yns_5": 0.1528, "loss_cls_dn_0": 0.23341, "loss_box_dn_0": 0.78742, "loss_cls_dn_1": 0.15919, "loss_box_dn_1": 0.74508, "loss_cls_dn_2": 0.15849, "loss_box_dn_2": 0.72645, "loss_cls_dn_3": 0.15796, "loss_box_dn_3": 0.72613, "loss_cls_dn_4": 0.1673, "loss_box_dn_4": 0.73261, "loss_cls_dn_5": 0.1747, "loss_box_dn_5": 0.74424, "loss_dense_depth": 0.78481, "loss": 27.69639, "grad_norm": 46.5559, "time": 1.47819} -{"mode": "train", "epoch": 1, "iter": 133, "lr": 0.00015, "memory": 49163, "data_time": 0.07599, "loss_cls_0": 0.89865, "loss_box_0": 1.73299, "loss_cns_0": 0.6103, "loss_yns_0": 0.14903, "loss_cls_1": 0.94489, "loss_box_1": 1.82565, "loss_cns_1": 0.63632, "loss_yns_1": 0.14944, "loss_cls_2": 0.98484, "loss_box_2": 1.77043, "loss_cns_2": 0.64542, "loss_yns_2": 0.15258, "loss_cls_3": 0.99371, "loss_box_3": 1.77769, "loss_cns_3": 0.64672, "loss_yns_3": 0.15138, "loss_cls_4": 1.00821, "loss_box_4": 1.76553, "loss_cns_4": 0.65089, "loss_yns_4": 0.15371, "loss_cls_5": 0.99852, "loss_box_5": 1.77708, "loss_cns_5": 0.64772, "loss_yns_5": 0.15207, "loss_cls_dn_0": 0.23398, "loss_box_dn_0": 0.79148, "loss_cls_dn_1": 0.1605, "loss_box_dn_1": 0.73869, "loss_cls_dn_2": 0.15668, "loss_box_dn_2": 0.71952, "loss_cls_dn_3": 0.16221, "loss_box_dn_3": 0.71803, "loss_cls_dn_4": 0.1692, "loss_box_dn_4": 0.72277, "loss_cls_dn_5": 0.17546, "loss_box_dn_5": 0.72709, "loss_dense_depth": 0.78933, "loss": 27.48868, "grad_norm": 42.34099, "time": 1.48254} -{"mode": "train", "epoch": 1, "iter": 134, "lr": 0.00015, "memory": 49163, "data_time": 0.07615, "loss_cls_0": 0.8777, "loss_box_0": 1.68999, "loss_cns_0": 0.61466, "loss_yns_0": 0.14893, "loss_cls_1": 0.94929, "loss_box_1": 1.84718, "loss_cns_1": 0.64139, "loss_yns_1": 0.15074, "loss_cls_2": 0.97841, "loss_box_2": 1.80676, "loss_cns_2": 0.64918, "loss_yns_2": 0.15044, "loss_cls_3": 0.99062, "loss_box_3": 1.81077, "loss_cns_3": 0.64796, "loss_yns_3": 0.15336, "loss_cls_4": 0.99796, "loss_box_4": 1.80004, "loss_cns_4": 0.6488, "loss_yns_4": 0.15319, "loss_cls_5": 0.99253, "loss_box_5": 1.81915, "loss_cns_5": 0.64554, "loss_yns_5": 0.15178, "loss_cls_dn_0": 0.22781, "loss_box_dn_0": 0.78813, "loss_cls_dn_1": 0.15843, "loss_box_dn_1": 0.71454, "loss_cls_dn_2": 0.1565, "loss_box_dn_2": 0.69874, "loss_cls_dn_3": 0.15976, "loss_box_dn_3": 0.69799, "loss_cls_dn_4": 0.16676, "loss_box_dn_4": 0.70406, "loss_cls_dn_5": 0.17106, "loss_box_dn_5": 0.70341, "loss_dense_depth": 0.76706, "loss": 27.4306, "grad_norm": 46.00354, "time": 1.50365} -{"mode": "train", "epoch": 1, "iter": 135, "lr": 0.00015, "memory": 49163, "data_time": 0.0829, "loss_cls_0": 0.88772, "loss_box_0": 1.73524, "loss_cns_0": 0.61991, "loss_yns_0": 0.15463, "loss_cls_1": 0.96002, "loss_box_1": 1.87783, "loss_cns_1": 0.64335, "loss_yns_1": 0.15589, "loss_cls_2": 0.98115, "loss_box_2": 1.85223, "loss_cns_2": 0.64854, "loss_yns_2": 0.15227, "loss_cls_3": 0.99041, "loss_box_3": 1.84831, "loss_cns_3": 0.64663, "loss_yns_3": 0.15307, "loss_cls_4": 1.00378, "loss_box_4": 1.85043, "loss_cns_4": 0.65008, "loss_yns_4": 0.15419, "loss_cls_5": 0.98732, "loss_box_5": 1.85967, "loss_cns_5": 0.64393, "loss_yns_5": 0.15483, "loss_cls_dn_0": 0.23565, "loss_box_dn_0": 0.78765, "loss_cls_dn_1": 0.16119, "loss_box_dn_1": 0.72594, "loss_cls_dn_2": 0.15979, "loss_box_dn_2": 0.70758, "loss_cls_dn_3": 0.15757, "loss_box_dn_3": 0.70891, "loss_cls_dn_4": 0.16706, "loss_box_dn_4": 0.71478, "loss_cls_dn_5": 0.17439, "loss_box_dn_5": 0.7171, "loss_dense_depth": 0.77792, "loss": 27.80693, "grad_norm": 34.0763, "time": 1.48865} -{"mode": "train", "epoch": 1, "iter": 136, "lr": 0.00015, "memory": 49163, "data_time": 0.07736, "loss_cls_0": 0.87039, "loss_box_0": 1.7269, "loss_cns_0": 0.62484, "loss_yns_0": 0.14945, "loss_cls_1": 0.9333, "loss_box_1": 1.85943, "loss_cns_1": 0.64391, "loss_yns_1": 0.15031, "loss_cls_2": 0.95187, "loss_box_2": 1.83723, "loss_cns_2": 0.64793, "loss_yns_2": 0.14995, "loss_cls_3": 0.96231, "loss_box_3": 1.83847, "loss_cns_3": 0.64787, "loss_yns_3": 0.14918, "loss_cls_4": 0.9864, "loss_box_4": 1.82405, "loss_cns_4": 0.64843, "loss_yns_4": 0.15108, "loss_cls_5": 0.97521, "loss_box_5": 1.83449, "loss_cns_5": 0.64682, "loss_yns_5": 0.15217, "loss_cls_dn_0": 0.22989, "loss_box_dn_0": 0.78149, "loss_cls_dn_1": 0.1517, "loss_box_dn_1": 0.7222, "loss_cls_dn_2": 0.1506, "loss_box_dn_2": 0.71299, "loss_cls_dn_3": 0.15194, "loss_box_dn_3": 0.71933, "loss_cls_dn_4": 0.16031, "loss_box_dn_4": 0.72092, "loss_cls_dn_5": 0.16635, "loss_box_dn_5": 0.73426, "loss_dense_depth": 0.7648, "loss": 27.52879, "grad_norm": 60.78459, "time": 1.5007} -{"mode": "train", "epoch": 1, "iter": 137, "lr": 0.00015, "memory": 49163, "data_time": 0.07644, "loss_cls_0": 0.87481, "loss_box_0": 1.74504, "loss_cns_0": 0.62114, "loss_yns_0": 0.15052, "loss_cls_1": 0.9456, "loss_box_1": 1.88146, "loss_cns_1": 0.63973, "loss_yns_1": 0.14923, "loss_cls_2": 0.9763, "loss_box_2": 1.85339, "loss_cns_2": 0.64745, "loss_yns_2": 0.14951, "loss_cls_3": 0.97781, "loss_box_3": 1.86593, "loss_cns_3": 0.64544, "loss_yns_3": 0.14983, "loss_cls_4": 0.98692, "loss_box_4": 1.85074, "loss_cns_4": 0.64692, "loss_yns_4": 0.15184, "loss_cls_5": 0.99295, "loss_box_5": 1.86601, "loss_cns_5": 0.64516, "loss_yns_5": 0.15452, "loss_cls_dn_0": 0.23871, "loss_box_dn_0": 0.78376, "loss_cls_dn_1": 0.1504, "loss_box_dn_1": 0.74428, "loss_cls_dn_2": 0.15136, "loss_box_dn_2": 0.73086, "loss_cls_dn_3": 0.15336, "loss_box_dn_3": 0.739, "loss_cls_dn_4": 0.16109, "loss_box_dn_4": 0.74055, "loss_cls_dn_5": 0.16802, "loss_box_dn_5": 0.75454, "loss_dense_depth": 0.76524, "loss": 27.84942, "grad_norm": 62.83163, "time": 1.50205} -{"mode": "train", "epoch": 1, "iter": 138, "lr": 0.00015, "memory": 49163, "data_time": 0.07507, "loss_cls_0": 0.86386, "loss_box_0": 1.73609, "loss_cns_0": 0.62367, "loss_yns_0": 0.14936, "loss_cls_1": 0.9384, "loss_box_1": 1.82972, "loss_cns_1": 0.64366, "loss_yns_1": 0.14929, "loss_cls_2": 0.98145, "loss_box_2": 1.77331, "loss_cns_2": 0.65301, "loss_yns_2": 0.14958, "loss_cls_3": 0.97817, "loss_box_3": 1.78286, "loss_cns_3": 0.65341, "loss_yns_3": 0.14961, "loss_cls_4": 0.98109, "loss_box_4": 1.75544, "loss_cns_4": 0.65459, "loss_yns_4": 0.15037, "loss_cls_5": 0.99577, "loss_box_5": 1.75842, "loss_cns_5": 0.65228, "loss_yns_5": 0.15331, "loss_cls_dn_0": 0.23506, "loss_box_dn_0": 0.79207, "loss_cls_dn_1": 0.15035, "loss_box_dn_1": 0.75843, "loss_cls_dn_2": 0.1505, "loss_box_dn_2": 0.73453, "loss_cls_dn_3": 0.15094, "loss_box_dn_3": 0.74199, "loss_cls_dn_4": 0.16161, "loss_box_dn_4": 0.73846, "loss_cls_dn_5": 0.16633, "loss_box_dn_5": 0.74335, "loss_dense_depth": 0.78873, "loss": 27.46906, "grad_norm": 34.41742, "time": 1.49238} -{"mode": "train", "epoch": 1, "iter": 139, "lr": 0.00016, "memory": 49163, "data_time": 0.07596, "loss_cls_0": 0.82712, "loss_box_0": 1.73411, "loss_cns_0": 0.62428, "loss_yns_0": 0.14914, "loss_cls_1": 0.8958, "loss_box_1": 1.81982, "loss_cns_1": 0.64275, "loss_yns_1": 0.15029, "loss_cls_2": 0.93726, "loss_box_2": 1.77181, "loss_cns_2": 0.65103, "loss_yns_2": 0.15331, "loss_cls_3": 0.94288, "loss_box_3": 1.77164, "loss_cns_3": 0.65051, "loss_yns_3": 0.14897, "loss_cls_4": 0.94758, "loss_box_4": 1.76133, "loss_cns_4": 0.65009, "loss_yns_4": 0.15061, "loss_cls_5": 0.94711, "loss_box_5": 1.76241, "loss_cns_5": 0.64746, "loss_yns_5": 0.1502, "loss_cls_dn_0": 0.22735, "loss_box_dn_0": 0.79452, "loss_cls_dn_1": 0.14975, "loss_box_dn_1": 0.74735, "loss_cls_dn_2": 0.14701, "loss_box_dn_2": 0.72929, "loss_cls_dn_3": 0.14945, "loss_box_dn_3": 0.73489, "loss_cls_dn_4": 0.16166, "loss_box_dn_4": 0.73428, "loss_cls_dn_5": 0.16691, "loss_box_dn_5": 0.73163, "loss_dense_depth": 0.78443, "loss": 27.14606, "grad_norm": 62.93803, "time": 1.49244} -{"mode": "train", "epoch": 1, "iter": 140, "lr": 0.00016, "memory": 49163, "data_time": 0.08685, "loss_cls_0": 0.84323, "loss_box_0": 1.70954, "loss_cns_0": 0.62576, "loss_yns_0": 0.14867, "loss_cls_1": 0.91747, "loss_box_1": 1.7956, "loss_cns_1": 0.6445, "loss_yns_1": 0.14692, "loss_cls_2": 0.95113, "loss_box_2": 1.76089, "loss_cns_2": 0.6542, "loss_yns_2": 0.15326, "loss_cls_3": 0.9568, "loss_box_3": 1.76791, "loss_cns_3": 0.65394, "loss_yns_3": 0.14898, "loss_cls_4": 0.97132, "loss_box_4": 1.75268, "loss_cns_4": 0.65578, "loss_yns_4": 0.14848, "loss_cls_5": 0.94773, "loss_box_5": 1.76455, "loss_cns_5": 0.65146, "loss_yns_5": 0.14759, "loss_cls_dn_0": 0.22393, "loss_box_dn_0": 0.79274, "loss_cls_dn_1": 0.14746, "loss_box_dn_1": 0.74143, "loss_cls_dn_2": 0.14603, "loss_box_dn_2": 0.72975, "loss_cls_dn_3": 0.14998, "loss_box_dn_3": 0.73648, "loss_cls_dn_4": 0.15933, "loss_box_dn_4": 0.73542, "loss_cls_dn_5": 0.16607, "loss_box_dn_5": 0.73695, "loss_dense_depth": 0.77277, "loss": 27.15675, "grad_norm": 72.06483, "time": 1.50748} -{"mode": "train", "epoch": 1, "iter": 141, "lr": 0.00016, "memory": 49163, "data_time": 0.08569, "loss_cls_0": 0.89283, "loss_box_0": 1.76943, "loss_cns_0": 0.62606, "loss_yns_0": 0.15152, "loss_cls_1": 0.94748, "loss_box_1": 1.85944, "loss_cns_1": 0.64575, "loss_yns_1": 0.14948, "loss_cls_2": 0.97955, "loss_box_2": 1.82424, "loss_cns_2": 0.65621, "loss_yns_2": 0.15309, "loss_cls_3": 0.98044, "loss_box_3": 1.84013, "loss_cns_3": 0.65306, "loss_yns_3": 0.15274, "loss_cls_4": 0.99949, "loss_box_4": 1.81426, "loss_cns_4": 0.6561, "loss_yns_4": 0.15016, "loss_cls_5": 0.98588, "loss_box_5": 1.83152, "loss_cns_5": 0.64992, "loss_yns_5": 0.15428, "loss_cls_dn_0": 0.23683, "loss_box_dn_0": 0.78741, "loss_cls_dn_1": 0.15987, "loss_box_dn_1": 0.74771, "loss_cls_dn_2": 0.15675, "loss_box_dn_2": 0.73681, "loss_cls_dn_3": 0.15891, "loss_box_dn_3": 0.74399, "loss_cls_dn_4": 0.16876, "loss_box_dn_4": 0.74171, "loss_cls_dn_5": 0.17615, "loss_box_dn_5": 0.74749, "loss_dense_depth": 0.79357, "loss": 27.87902, "grad_norm": 59.05149, "time": 1.55798} -{"mode": "train", "epoch": 1, "iter": 142, "lr": 0.00016, "memory": 49163, "data_time": 0.15665, "loss_cls_0": 0.89547, "loss_box_0": 1.81083, "loss_cns_0": 0.62001, "loss_yns_0": 0.15401, "loss_cls_1": 0.94944, "loss_box_1": 1.90361, "loss_cns_1": 0.64066, "loss_yns_1": 0.15248, "loss_cls_2": 0.97845, "loss_box_2": 1.84067, "loss_cns_2": 0.64919, "loss_yns_2": 0.15525, "loss_cls_3": 0.97822, "loss_box_3": 1.85078, "loss_cns_3": 0.64773, "loss_yns_3": 0.15321, "loss_cls_4": 0.98355, "loss_box_4": 1.82901, "loss_cns_4": 0.6497, "loss_yns_4": 0.15271, "loss_cls_5": 0.98067, "loss_box_5": 1.8188, "loss_cns_5": 0.64583, "loss_yns_5": 0.1555, "loss_cls_dn_0": 0.23535, "loss_box_dn_0": 0.79139, "loss_cls_dn_1": 0.15794, "loss_box_dn_1": 0.7437, "loss_cls_dn_2": 0.15582, "loss_box_dn_2": 0.72491, "loss_cls_dn_3": 0.15735, "loss_box_dn_3": 0.72988, "loss_cls_dn_4": 0.1629, "loss_box_dn_4": 0.729, "loss_cls_dn_5": 0.17044, "loss_box_dn_5": 0.72827, "loss_dense_depth": 0.80231, "loss": 27.88504, "grad_norm": 45.13947, "time": 1.55792} -{"mode": "train", "epoch": 1, "iter": 143, "lr": 0.00016, "memory": 49163, "data_time": 0.07082, "loss_cls_0": 0.8671, "loss_box_0": 1.80563, "loss_cns_0": 0.61945, "loss_yns_0": 0.14831, "loss_cls_1": 0.92889, "loss_box_1": 1.87665, "loss_cns_1": 0.6462, "loss_yns_1": 0.14962, "loss_cls_2": 0.96632, "loss_box_2": 1.80204, "loss_cns_2": 0.65378, "loss_yns_2": 0.15458, "loss_cls_3": 0.98428, "loss_box_3": 1.80485, "loss_cns_3": 0.6539, "loss_yns_3": 0.15214, "loss_cls_4": 0.97161, "loss_box_4": 1.79932, "loss_cns_4": 0.65305, "loss_yns_4": 0.15104, "loss_cls_5": 0.95377, "loss_box_5": 1.78665, "loss_cns_5": 0.65018, "loss_yns_5": 0.15039, "loss_cls_dn_0": 0.22688, "loss_box_dn_0": 0.79071, "loss_cls_dn_1": 0.15323, "loss_box_dn_1": 0.75155, "loss_cls_dn_2": 0.15457, "loss_box_dn_2": 0.72938, "loss_cls_dn_3": 0.16224, "loss_box_dn_3": 0.73461, "loss_cls_dn_4": 0.16717, "loss_box_dn_4": 0.74117, "loss_cls_dn_5": 0.17203, "loss_box_dn_5": 0.74297, "loss_dense_depth": 0.76305, "loss": 27.61933, "grad_norm": 57.77005, "time": 1.51359} -{"mode": "train", "epoch": 1, "iter": 144, "lr": 0.00016, "memory": 49163, "data_time": 0.07382, "loss_cls_0": 0.88866, "loss_box_0": 1.81979, "loss_cns_0": 0.61789, "loss_yns_0": 0.15142, "loss_cls_1": 0.95335, "loss_box_1": 1.91129, "loss_cns_1": 0.64051, "loss_yns_1": 0.1504, "loss_cls_2": 0.98234, "loss_box_2": 1.85969, "loss_cns_2": 0.64714, "loss_yns_2": 0.15558, "loss_cls_3": 1.00643, "loss_box_3": 1.84625, "loss_cns_3": 0.64921, "loss_yns_3": 0.15399, "loss_cls_4": 1.00279, "loss_box_4": 1.84512, "loss_cns_4": 0.65072, "loss_yns_4": 0.15236, "loss_cls_5": 0.99073, "loss_box_5": 1.84908, "loss_cns_5": 0.64968, "loss_yns_5": 0.15089, "loss_cls_dn_0": 0.23606, "loss_box_dn_0": 0.79231, "loss_cls_dn_1": 0.16091, "loss_box_dn_1": 0.75293, "loss_cls_dn_2": 0.16292, "loss_box_dn_2": 0.7311, "loss_cls_dn_3": 0.17229, "loss_box_dn_3": 0.73422, "loss_cls_dn_4": 0.1731, "loss_box_dn_4": 0.74261, "loss_cls_dn_5": 0.177, "loss_box_dn_5": 0.75149, "loss_dense_depth": 0.76893, "loss": 28.08117, "grad_norm": 56.30296, "time": 1.56557} -{"mode": "train", "epoch": 1, "iter": 145, "lr": 0.00016, "memory": 49163, "data_time": 0.08214, "loss_cls_0": 0.87559, "loss_box_0": 1.78961, "loss_cns_0": 0.62156, "loss_yns_0": 0.15154, "loss_cls_1": 0.95417, "loss_box_1": 1.91216, "loss_cns_1": 0.64107, "loss_yns_1": 0.15179, "loss_cls_2": 0.98458, "loss_box_2": 1.85805, "loss_cns_2": 0.64883, "loss_yns_2": 0.15349, "loss_cls_3": 0.98535, "loss_box_3": 1.84729, "loss_cns_3": 0.65154, "loss_yns_3": 0.15483, "loss_cls_4": 0.99258, "loss_box_4": 1.84263, "loss_cns_4": 0.65164, "loss_yns_4": 0.15359, "loss_cls_5": 0.992, "loss_box_5": 1.84786, "loss_cns_5": 0.65008, "loss_yns_5": 0.15319, "loss_cls_dn_0": 0.23021, "loss_box_dn_0": 0.78677, "loss_cls_dn_1": 0.15975, "loss_box_dn_1": 0.75859, "loss_cls_dn_2": 0.16511, "loss_box_dn_2": 0.73569, "loss_cls_dn_3": 0.16395, "loss_box_dn_3": 0.73629, "loss_cls_dn_4": 0.16671, "loss_box_dn_4": 0.74533, "loss_cls_dn_5": 0.16984, "loss_box_dn_5": 0.75443, "loss_dense_depth": 0.76042, "loss": 27.9981, "grad_norm": 41.05056, "time": 1.49884} -{"mode": "train", "epoch": 1, "iter": 146, "lr": 0.00016, "memory": 49163, "data_time": 0.07572, "loss_cls_0": 0.85694, "loss_box_0": 1.76567, "loss_cns_0": 0.62098, "loss_yns_0": 0.14986, "loss_cls_1": 0.92289, "loss_box_1": 1.84378, "loss_cns_1": 0.6443, "loss_yns_1": 0.15013, "loss_cls_2": 0.95435, "loss_box_2": 1.79475, "loss_cns_2": 0.65151, "loss_yns_2": 0.15208, "loss_cls_3": 0.96264, "loss_box_3": 1.78932, "loss_cns_3": 0.65267, "loss_yns_3": 0.15155, "loss_cls_4": 0.973, "loss_box_4": 1.7982, "loss_cns_4": 0.65191, "loss_yns_4": 0.1536, "loss_cls_5": 0.96676, "loss_box_5": 1.80749, "loss_cns_5": 0.64901, "loss_yns_5": 0.15514, "loss_cls_dn_0": 0.22701, "loss_box_dn_0": 0.78364, "loss_cls_dn_1": 0.15189, "loss_box_dn_1": 0.76347, "loss_cls_dn_2": 0.15698, "loss_box_dn_2": 0.74062, "loss_cls_dn_3": 0.15197, "loss_box_dn_3": 0.73657, "loss_cls_dn_4": 0.16132, "loss_box_dn_4": 0.74698, "loss_cls_dn_5": 0.16641, "loss_box_dn_5": 0.7543, "loss_dense_depth": 0.75993, "loss": 27.51961, "grad_norm": 43.97563, "time": 1.51004} -{"mode": "train", "epoch": 1, "iter": 147, "lr": 0.00016, "memory": 49163, "data_time": 0.0748, "loss_cls_0": 0.88227, "loss_box_0": 1.80034, "loss_cns_0": 0.61868, "loss_yns_0": 0.15017, "loss_cls_1": 0.95333, "loss_box_1": 1.83045, "loss_cns_1": 0.64537, "loss_yns_1": 0.14992, "loss_cls_2": 0.97298, "loss_box_2": 1.7754, "loss_cns_2": 0.65446, "loss_yns_2": 0.15051, "loss_cls_3": 0.99176, "loss_box_3": 1.76364, "loss_cns_3": 0.65553, "loss_yns_3": 0.15123, "loss_cls_4": 0.97813, "loss_box_4": 1.7671, "loss_cns_4": 0.65525, "loss_yns_4": 0.15155, "loss_cls_5": 0.99105, "loss_box_5": 1.76001, "loss_cns_5": 0.6538, "loss_yns_5": 0.153, "loss_cls_dn_0": 0.23028, "loss_box_dn_0": 0.78582, "loss_cls_dn_1": 0.15026, "loss_box_dn_1": 0.75233, "loss_cls_dn_2": 0.15569, "loss_box_dn_2": 0.73333, "loss_cls_dn_3": 0.15642, "loss_box_dn_3": 0.72537, "loss_cls_dn_4": 0.16322, "loss_box_dn_4": 0.73031, "loss_cls_dn_5": 0.16965, "loss_box_dn_5": 0.73179, "loss_dense_depth": 0.77652, "loss": 27.51695, "grad_norm": 44.58092, "time": 1.48195} -{"mode": "train", "epoch": 1, "iter": 148, "lr": 0.00016, "memory": 49163, "data_time": 0.07393, "loss_cls_0": 0.85599, "loss_box_0": 1.78034, "loss_cns_0": 0.62178, "loss_yns_0": 0.14905, "loss_cls_1": 0.96055, "loss_box_1": 1.81711, "loss_cns_1": 0.64636, "loss_yns_1": 0.1488, "loss_cls_2": 0.98081, "loss_box_2": 1.75997, "loss_cns_2": 0.65341, "loss_yns_2": 0.15083, "loss_cls_3": 0.98619, "loss_box_3": 1.77308, "loss_cns_3": 0.65409, "loss_yns_3": 0.15102, "loss_cls_4": 0.97562, "loss_box_4": 1.75926, "loss_cns_4": 0.65783, "loss_yns_4": 0.14841, "loss_cls_5": 0.98233, "loss_box_5": 1.73884, "loss_cns_5": 0.66034, "loss_yns_5": 0.14944, "loss_cls_dn_0": 0.22401, "loss_box_dn_0": 0.78704, "loss_cls_dn_1": 0.15096, "loss_box_dn_1": 0.73503, "loss_cls_dn_2": 0.15175, "loss_box_dn_2": 0.71444, "loss_cls_dn_3": 0.15722, "loss_box_dn_3": 0.71531, "loss_cls_dn_4": 0.16375, "loss_box_dn_4": 0.71173, "loss_cls_dn_5": 0.17104, "loss_box_dn_5": 0.7106, "loss_dense_depth": 0.74687, "loss": 27.30122, "grad_norm": 46.24647, "time": 1.4872} -{"mode": "train", "epoch": 1, "iter": 149, "lr": 0.00016, "memory": 49163, "data_time": 0.07478, "loss_cls_0": 0.85435, "loss_box_0": 1.76159, "loss_cns_0": 0.62114, "loss_yns_0": 0.14844, "loss_cls_1": 0.93812, "loss_box_1": 1.82749, "loss_cns_1": 0.643, "loss_yns_1": 0.15081, "loss_cls_2": 0.96889, "loss_box_2": 1.76676, "loss_cns_2": 0.6498, "loss_yns_2": 0.15459, "loss_cls_3": 0.96524, "loss_box_3": 1.80096, "loss_cns_3": 0.65081, "loss_yns_3": 0.15462, "loss_cls_4": 0.97037, "loss_box_4": 1.7801, "loss_cns_4": 0.65338, "loss_yns_4": 0.14991, "loss_cls_5": 0.972, "loss_box_5": 1.79446, "loss_cns_5": 0.65269, "loss_yns_5": 0.15101, "loss_cls_dn_0": 0.22118, "loss_box_dn_0": 0.77838, "loss_cls_dn_1": 0.15047, "loss_box_dn_1": 0.74418, "loss_cls_dn_2": 0.14852, "loss_box_dn_2": 0.7214, "loss_cls_dn_3": 0.15349, "loss_box_dn_3": 0.73663, "loss_cls_dn_4": 0.16086, "loss_box_dn_4": 0.73223, "loss_cls_dn_5": 0.17749, "loss_box_dn_5": 0.75011, "loss_dense_depth": 0.72307, "loss": 27.37853, "grad_norm": 43.61793, "time": 1.53336} -{"mode": "train", "epoch": 1, "iter": 150, "lr": 0.00016, "memory": 49163, "data_time": 0.06996, "loss_cls_0": 0.87417, "loss_box_0": 1.74882, "loss_cns_0": 0.61863, "loss_yns_0": 0.14672, "loss_cls_1": 0.9361, "loss_box_1": 1.81501, "loss_cns_1": 0.6474, "loss_yns_1": 0.14869, "loss_cls_2": 0.96462, "loss_box_2": 1.76653, "loss_cns_2": 0.6523, "loss_yns_2": 0.15008, "loss_cls_3": 0.96503, "loss_box_3": 1.78783, "loss_cns_3": 0.65497, "loss_yns_3": 0.15188, "loss_cls_4": 0.97794, "loss_box_4": 1.778, "loss_cns_4": 0.65578, "loss_yns_4": 0.1494, "loss_cls_5": 0.97666, "loss_box_5": 1.80567, "loss_cns_5": 0.65337, "loss_yns_5": 0.15214, "loss_cls_dn_0": 0.22245, "loss_box_dn_0": 0.77407, "loss_cls_dn_1": 0.1549, "loss_box_dn_1": 0.74725, "loss_cls_dn_2": 0.15484, "loss_box_dn_2": 0.72989, "loss_cls_dn_3": 0.16024, "loss_box_dn_3": 0.74918, "loss_cls_dn_4": 0.16705, "loss_box_dn_4": 0.75566, "loss_cls_dn_5": 0.18413, "loss_box_dn_5": 0.77951, "loss_dense_depth": 0.75849, "loss": 27.51539, "grad_norm": 51.56858, "time": 1.49932} -{"mode": "train", "epoch": 1, "iter": 151, "lr": 0.00016, "memory": 49163, "data_time": 0.07471, "loss_cls_0": 0.87668, "loss_box_0": 1.73472, "loss_cns_0": 0.6206, "loss_yns_0": 0.14612, "loss_cls_1": 0.94554, "loss_box_1": 1.82457, "loss_cns_1": 0.64274, "loss_yns_1": 0.14789, "loss_cls_2": 0.95351, "loss_box_2": 1.78412, "loss_cns_2": 0.65078, "loss_yns_2": 0.14927, "loss_cls_3": 0.96195, "loss_box_3": 1.77659, "loss_cns_3": 0.65119, "loss_yns_3": 0.14995, "loss_cls_4": 0.97715, "loss_box_4": 1.79439, "loss_cns_4": 0.6514, "loss_yns_4": 0.14953, "loss_cls_5": 0.968, "loss_box_5": 1.79376, "loss_cns_5": 0.64921, "loss_yns_5": 0.15373, "loss_cls_dn_0": 0.22178, "loss_box_dn_0": 0.77894, "loss_cls_dn_1": 0.15482, "loss_box_dn_1": 0.7633, "loss_cls_dn_2": 0.15546, "loss_box_dn_2": 0.74334, "loss_cls_dn_3": 0.15657, "loss_box_dn_3": 0.7484, "loss_cls_dn_4": 0.1605, "loss_box_dn_4": 0.76237, "loss_cls_dn_5": 0.1675, "loss_box_dn_5": 0.77054, "loss_dense_depth": 0.75888, "loss": 27.49578, "grad_norm": 51.4416, "time": 1.49887} -{"mode": "train", "epoch": 1, "iter": 152, "lr": 0.00016, "memory": 49163, "data_time": 0.07582, "loss_cls_0": 0.87027, "loss_box_0": 1.70402, "loss_cns_0": 0.61498, "loss_yns_0": 0.1479, "loss_cls_1": 0.95728, "loss_box_1": 1.7988, "loss_cns_1": 0.64591, "loss_yns_1": 0.15033, "loss_cls_2": 0.95133, "loss_box_2": 1.75849, "loss_cns_2": 0.65298, "loss_yns_2": 0.15124, "loss_cls_3": 0.96966, "loss_box_3": 1.74309, "loss_cns_3": 0.65354, "loss_yns_3": 0.15237, "loss_cls_4": 0.98336, "loss_box_4": 1.75741, "loss_cns_4": 0.65428, "loss_yns_4": 0.15177, "loss_cls_5": 0.98944, "loss_box_5": 1.73977, "loss_cns_5": 0.6519, "loss_yns_5": 0.15177, "loss_cls_dn_0": 0.22107, "loss_box_dn_0": 0.77757, "loss_cls_dn_1": 0.15544, "loss_box_dn_1": 0.7687, "loss_cls_dn_2": 0.1513, "loss_box_dn_2": 0.74583, "loss_cls_dn_3": 0.15087, "loss_box_dn_3": 0.74471, "loss_cls_dn_4": 0.15635, "loss_box_dn_4": 0.75636, "loss_cls_dn_5": 0.16459, "loss_box_dn_5": 0.74936, "loss_dense_depth": 0.73183, "loss": 27.27587, "grad_norm": 38.62796, "time": 1.48886} -{"mode": "train", "epoch": 1, "iter": 153, "lr": 0.00016, "memory": 49163, "data_time": 0.07177, "loss_cls_0": 0.90068, "loss_box_0": 1.71861, "loss_cns_0": 0.60724, "loss_yns_0": 0.14716, "loss_cls_1": 0.9643, "loss_box_1": 1.77961, "loss_cns_1": 0.64709, "loss_yns_1": 0.15096, "loss_cls_2": 0.96354, "loss_box_2": 1.75875, "loss_cns_2": 0.65368, "loss_yns_2": 0.15288, "loss_cls_3": 0.98616, "loss_box_3": 1.74876, "loss_cns_3": 0.65417, "loss_yns_3": 0.15375, "loss_cls_4": 0.98078, "loss_box_4": 1.75625, "loss_cns_4": 0.65559, "loss_yns_4": 0.15324, "loss_cls_5": 0.98594, "loss_box_5": 1.74772, "loss_cns_5": 0.65393, "loss_yns_5": 0.15106, "loss_cls_dn_0": 0.23406, "loss_box_dn_0": 0.78529, "loss_cls_dn_1": 0.14815, "loss_box_dn_1": 0.74484, "loss_cls_dn_2": 0.14454, "loss_box_dn_2": 0.72981, "loss_cls_dn_3": 0.14983, "loss_box_dn_3": 0.73275, "loss_cls_dn_4": 0.15785, "loss_box_dn_4": 0.74242, "loss_cls_dn_5": 0.16853, "loss_box_dn_5": 0.73712, "loss_dense_depth": 0.76489, "loss": 27.31195, "grad_norm": 45.01832, "time": 1.48499} -{"mode": "train", "epoch": 1, "iter": 154, "lr": 0.00016, "memory": 49163, "data_time": 0.07377, "loss_cls_0": 0.857, "loss_box_0": 1.70504, "loss_cns_0": 0.61357, "loss_yns_0": 0.14983, "loss_cls_1": 0.91925, "loss_box_1": 1.75714, "loss_cns_1": 0.64378, "loss_yns_1": 0.14895, "loss_cls_2": 0.95506, "loss_box_2": 1.74523, "loss_cns_2": 0.65062, "loss_yns_2": 0.15169, "loss_cls_3": 0.95209, "loss_box_3": 1.73503, "loss_cns_3": 0.65224, "loss_yns_3": 0.15244, "loss_cls_4": 0.96262, "loss_box_4": 1.72689, "loss_cns_4": 0.65949, "loss_yns_4": 0.15152, "loss_cls_5": 0.95152, "loss_box_5": 1.74665, "loss_cns_5": 0.65653, "loss_yns_5": 0.15306, "loss_cls_dn_0": 0.22467, "loss_box_dn_0": 0.77467, "loss_cls_dn_1": 0.14342, "loss_box_dn_1": 0.73989, "loss_cls_dn_2": 0.14593, "loss_box_dn_2": 0.73082, "loss_cls_dn_3": 0.15685, "loss_box_dn_3": 0.73174, "loss_cls_dn_4": 0.15963, "loss_box_dn_4": 0.73803, "loss_cls_dn_5": 0.16856, "loss_box_dn_5": 0.74506, "loss_dense_depth": 0.72272, "loss": 26.97924, "grad_norm": 44.32511, "time": 1.48383} -{"mode": "train", "epoch": 1, "iter": 155, "lr": 0.00016, "memory": 49163, "data_time": 0.0801, "loss_cls_0": 0.84243, "loss_box_0": 1.71589, "loss_cns_0": 0.62018, "loss_yns_0": 0.15237, "loss_cls_1": 0.88146, "loss_box_1": 1.73855, "loss_cns_1": 0.64268, "loss_yns_1": 0.14927, "loss_cls_2": 0.93536, "loss_box_2": 1.71398, "loss_cns_2": 0.64935, "loss_yns_2": 0.15064, "loss_cls_3": 0.92555, "loss_box_3": 1.69666, "loss_cns_3": 0.65345, "loss_yns_3": 0.15184, "loss_cls_4": 0.94754, "loss_box_4": 1.69057, "loss_cns_4": 0.65947, "loss_yns_4": 0.15283, "loss_cls_5": 0.93679, "loss_box_5": 1.70638, "loss_cns_5": 0.6559, "loss_yns_5": 0.15144, "loss_cls_dn_0": 0.21918, "loss_box_dn_0": 0.78067, "loss_cls_dn_1": 0.1436, "loss_box_dn_1": 0.74011, "loss_cls_dn_2": 0.14649, "loss_box_dn_2": 0.73462, "loss_cls_dn_3": 0.15296, "loss_box_dn_3": 0.73165, "loss_cls_dn_4": 0.15563, "loss_box_dn_4": 0.73963, "loss_cls_dn_5": 0.16018, "loss_box_dn_5": 0.74725, "loss_dense_depth": 0.73047, "loss": 26.70304, "grad_norm": 53.47087, "time": 1.48686} -{"mode": "train", "epoch": 1, "iter": 156, "lr": 0.00016, "memory": 49163, "data_time": 0.0714, "loss_cls_0": 0.86327, "loss_box_0": 1.73292, "loss_cns_0": 0.61743, "loss_yns_0": 0.15128, "loss_cls_1": 0.89477, "loss_box_1": 1.77571, "loss_cns_1": 0.63791, "loss_yns_1": 0.15016, "loss_cls_2": 0.92442, "loss_box_2": 1.73733, "loss_cns_2": 0.64494, "loss_yns_2": 0.15297, "loss_cls_3": 0.94191, "loss_box_3": 1.72028, "loss_cns_3": 0.64905, "loss_yns_3": 0.15267, "loss_cls_4": 0.97534, "loss_box_4": 1.70167, "loss_cns_4": 0.6483, "loss_yns_4": 0.15106, "loss_cls_5": 0.96233, "loss_box_5": 1.71481, "loss_cns_5": 0.64998, "loss_yns_5": 0.15276, "loss_cls_dn_0": 0.22476, "loss_box_dn_0": 0.78037, "loss_cls_dn_1": 0.14532, "loss_box_dn_1": 0.76345, "loss_cls_dn_2": 0.14328, "loss_box_dn_2": 0.74808, "loss_cls_dn_3": 0.14724, "loss_box_dn_3": 0.74361, "loss_cls_dn_4": 0.15192, "loss_box_dn_4": 0.75141, "loss_cls_dn_5": 0.15709, "loss_box_dn_5": 0.75388, "loss_dense_depth": 0.7431, "loss": 26.95677, "grad_norm": 38.76596, "time": 1.48449} -{"mode": "train", "epoch": 1, "iter": 157, "lr": 0.00016, "memory": 49163, "data_time": 0.07172, "loss_cls_0": 0.85671, "loss_box_0": 1.71896, "loss_cns_0": 0.61993, "loss_yns_0": 0.15207, "loss_cls_1": 0.90609, "loss_box_1": 1.80927, "loss_cns_1": 0.63387, "loss_yns_1": 0.15098, "loss_cls_2": 0.96665, "loss_box_2": 1.73676, "loss_cns_2": 0.65007, "loss_yns_2": 0.14934, "loss_cls_3": 0.96551, "loss_box_3": 1.73797, "loss_cns_3": 0.65185, "loss_yns_3": 0.15117, "loss_cls_4": 0.98588, "loss_box_4": 1.73511, "loss_cns_4": 0.65197, "loss_yns_4": 0.15015, "loss_cls_5": 0.98969, "loss_box_5": 1.76125, "loss_cns_5": 0.64979, "loss_yns_5": 0.15406, "loss_cls_dn_0": 0.21992, "loss_box_dn_0": 0.77503, "loss_cls_dn_1": 0.14374, "loss_box_dn_1": 0.73282, "loss_cls_dn_2": 0.14177, "loss_box_dn_2": 0.71438, "loss_cls_dn_3": 0.14383, "loss_box_dn_3": 0.7154, "loss_cls_dn_4": 0.14954, "loss_box_dn_4": 0.72709, "loss_cls_dn_5": 0.16338, "loss_box_dn_5": 0.73568, "loss_dense_depth": 0.73927, "loss": 27.03696, "grad_norm": 41.85922, "time": 1.49087} -{"mode": "train", "epoch": 1, "iter": 158, "lr": 0.00016, "memory": 49163, "data_time": 0.07149, "loss_cls_0": 0.82554, "loss_box_0": 1.71998, "loss_cns_0": 0.61651, "loss_yns_0": 0.15257, "loss_cls_1": 0.88124, "loss_box_1": 1.77636, "loss_cns_1": 0.63718, "loss_yns_1": 0.15211, "loss_cls_2": 0.94494, "loss_box_2": 1.71925, "loss_cns_2": 0.65079, "loss_yns_2": 0.15291, "loss_cls_3": 0.93604, "loss_box_3": 1.70946, "loss_cns_3": 0.65109, "loss_yns_3": 0.15242, "loss_cls_4": 0.93976, "loss_box_4": 1.72116, "loss_cns_4": 0.65791, "loss_yns_4": 0.15263, "loss_cls_5": 0.95163, "loss_box_5": 1.7371, "loss_cns_5": 0.65125, "loss_yns_5": 0.15344, "loss_cls_dn_0": 0.21355, "loss_box_dn_0": 0.77768, "loss_cls_dn_1": 0.1454, "loss_box_dn_1": 0.74882, "loss_cls_dn_2": 0.14397, "loss_box_dn_2": 0.73217, "loss_cls_dn_3": 0.14496, "loss_box_dn_3": 0.734, "loss_cls_dn_4": 0.15265, "loss_box_dn_4": 0.74306, "loss_cls_dn_5": 0.16593, "loss_box_dn_5": 0.75661, "loss_dense_depth": 0.71343, "loss": 26.81552, "grad_norm": 49.8675, "time": 1.48337} -{"mode": "train", "epoch": 1, "iter": 159, "lr": 0.00016, "memory": 49163, "data_time": 0.07137, "loss_cls_0": 0.84024, "loss_box_0": 1.75111, "loss_cns_0": 0.61715, "loss_yns_0": 0.15431, "loss_cls_1": 0.8921, "loss_box_1": 1.77623, "loss_cns_1": 0.64302, "loss_yns_1": 0.15451, "loss_cls_2": 0.92254, "loss_box_2": 1.7193, "loss_cns_2": 0.65327, "loss_yns_2": 0.15632, "loss_cls_3": 0.94675, "loss_box_3": 1.71407, "loss_cns_3": 0.65288, "loss_yns_3": 0.15707, "loss_cls_4": 0.96293, "loss_box_4": 1.71188, "loss_cns_4": 0.65768, "loss_yns_4": 0.15638, "loss_cls_5": 0.93808, "loss_box_5": 1.72162, "loss_cns_5": 0.65266, "loss_yns_5": 0.15574, "loss_cls_dn_0": 0.21418, "loss_box_dn_0": 0.77898, "loss_cls_dn_1": 0.14177, "loss_box_dn_1": 0.74655, "loss_cls_dn_2": 0.14013, "loss_box_dn_2": 0.72846, "loss_cls_dn_3": 0.14305, "loss_box_dn_3": 0.73119, "loss_cls_dn_4": 0.14997, "loss_box_dn_4": 0.73262, "loss_cls_dn_5": 0.15693, "loss_box_dn_5": 0.74029, "loss_dense_depth": 0.73435, "loss": 26.84632, "grad_norm": 43.58723, "time": 1.49312} -{"mode": "train", "epoch": 1, "iter": 160, "lr": 0.00016, "memory": 49163, "data_time": 0.09011, "loss_cls_0": 0.83783, "loss_box_0": 1.74766, "loss_cns_0": 0.61693, "loss_yns_0": 0.15506, "loss_cls_1": 0.88117, "loss_box_1": 1.76825, "loss_cns_1": 0.64469, "loss_yns_1": 0.15474, "loss_cls_2": 0.92001, "loss_box_2": 1.70329, "loss_cns_2": 0.65248, "loss_yns_2": 0.15417, "loss_cls_3": 0.93311, "loss_box_3": 1.70599, "loss_cns_3": 0.65142, "loss_yns_3": 0.15549, "loss_cls_4": 0.95576, "loss_box_4": 1.70553, "loss_cns_4": 0.65118, "loss_yns_4": 0.15489, "loss_cls_5": 0.94144, "loss_box_5": 1.73457, "loss_cns_5": 0.64802, "loss_yns_5": 0.15462, "loss_cls_dn_0": 0.21441, "loss_box_dn_0": 0.7763, "loss_cls_dn_1": 0.13938, "loss_box_dn_1": 0.74093, "loss_cls_dn_2": 0.13731, "loss_box_dn_2": 0.71959, "loss_cls_dn_3": 0.13981, "loss_box_dn_3": 0.72316, "loss_cls_dn_4": 0.1465, "loss_box_dn_4": 0.72536, "loss_cls_dn_5": 0.15133, "loss_box_dn_5": 0.73995, "loss_dense_depth": 0.72281, "loss": 26.70512, "grad_norm": 39.85465, "time": 1.50188} -{"mode": "train", "epoch": 1, "iter": 161, "lr": 0.00016, "memory": 49163, "data_time": 0.07703, "loss_cls_0": 0.84481, "loss_box_0": 1.7329, "loss_cns_0": 0.61625, "loss_yns_0": 0.15472, "loss_cls_1": 0.88778, "loss_box_1": 1.75949, "loss_cns_1": 0.64537, "loss_yns_1": 0.15584, "loss_cls_2": 0.92432, "loss_box_2": 1.69697, "loss_cns_2": 0.65447, "loss_yns_2": 0.15471, "loss_cls_3": 0.93103, "loss_box_3": 1.68089, "loss_cns_3": 0.65606, "loss_yns_3": 0.15626, "loss_cls_4": 0.94102, "loss_box_4": 1.68229, "loss_cns_4": 0.65673, "loss_yns_4": 0.15719, "loss_cls_5": 0.93494, "loss_box_5": 1.70799, "loss_cns_5": 0.65468, "loss_yns_5": 0.15581, "loss_cls_dn_0": 0.21455, "loss_box_dn_0": 0.77745, "loss_cls_dn_1": 0.1392, "loss_box_dn_1": 0.71651, "loss_cls_dn_2": 0.13777, "loss_box_dn_2": 0.6914, "loss_cls_dn_3": 0.14235, "loss_box_dn_3": 0.69109, "loss_cls_dn_4": 0.14593, "loss_box_dn_4": 0.69792, "loss_cls_dn_5": 0.15075, "loss_box_dn_5": 0.71412, "loss_dense_depth": 0.7189, "loss": 26.48044, "grad_norm": 30.64851, "time": 1.5584} -{"mode": "train", "epoch": 1, "iter": 162, "lr": 0.00016, "memory": 49163, "data_time": 0.15366, "loss_cls_0": 0.84434, "loss_box_0": 1.73618, "loss_cns_0": 0.61533, "loss_yns_0": 0.15638, "loss_cls_1": 0.89761, "loss_box_1": 1.75768, "loss_cns_1": 0.64349, "loss_yns_1": 0.1574, "loss_cls_2": 0.92947, "loss_box_2": 1.69538, "loss_cns_2": 0.65337, "loss_yns_2": 0.15691, "loss_cls_3": 0.93515, "loss_box_3": 1.682, "loss_cns_3": 0.6546, "loss_yns_3": 0.15631, "loss_cls_4": 0.94473, "loss_box_4": 1.68424, "loss_cns_4": 0.65453, "loss_yns_4": 0.15717, "loss_cls_5": 0.94514, "loss_box_5": 1.68458, "loss_cns_5": 0.6521, "loss_yns_5": 0.15527, "loss_cls_dn_0": 0.21642, "loss_box_dn_0": 0.78088, "loss_cls_dn_1": 0.13523, "loss_box_dn_1": 0.72959, "loss_cls_dn_2": 0.13326, "loss_box_dn_2": 0.71014, "loss_cls_dn_3": 0.14069, "loss_box_dn_3": 0.71218, "loss_cls_dn_4": 0.14507, "loss_box_dn_4": 0.72226, "loss_cls_dn_5": 0.14877, "loss_box_dn_5": 0.72908, "loss_dense_depth": 0.73124, "loss": 26.58415, "grad_norm": 37.71714, "time": 1.56322} -{"mode": "train", "epoch": 1, "iter": 163, "lr": 0.00016, "memory": 49163, "data_time": 0.07623, "loss_cls_0": 0.83167, "loss_box_0": 1.75833, "loss_cns_0": 0.62083, "loss_yns_0": 0.15745, "loss_cls_1": 0.90348, "loss_box_1": 1.76262, "loss_cns_1": 0.6436, "loss_yns_1": 0.15845, "loss_cls_2": 0.94724, "loss_box_2": 1.70976, "loss_cns_2": 0.65255, "loss_yns_2": 0.15741, "loss_cls_3": 0.93548, "loss_box_3": 1.69996, "loss_cns_3": 0.65208, "loss_yns_3": 0.15716, "loss_cls_4": 0.94673, "loss_box_4": 1.6942, "loss_cns_4": 0.65529, "loss_yns_4": 0.15756, "loss_cls_5": 0.94317, "loss_box_5": 1.71351, "loss_cns_5": 0.65143, "loss_yns_5": 0.15765, "loss_cls_dn_0": 0.21292, "loss_box_dn_0": 0.76819, "loss_cls_dn_1": 0.1365, "loss_box_dn_1": 0.7182, "loss_cls_dn_2": 0.1355, "loss_box_dn_2": 0.70566, "loss_cls_dn_3": 0.1367, "loss_box_dn_3": 0.70809, "loss_cls_dn_4": 0.14532, "loss_box_dn_4": 0.71507, "loss_cls_dn_5": 0.1482, "loss_box_dn_5": 0.72671, "loss_dense_depth": 0.70309, "loss": 26.62775, "grad_norm": 30.13676, "time": 1.52673} -{"mode": "train", "epoch": 1, "iter": 164, "lr": 0.00017, "memory": 49163, "data_time": 0.07769, "loss_cls_0": 0.83787, "loss_box_0": 1.77606, "loss_cns_0": 0.6249, "loss_yns_0": 0.15998, "loss_cls_1": 0.90653, "loss_box_1": 1.76957, "loss_cns_1": 0.64973, "loss_yns_1": 0.15923, "loss_cls_2": 0.93344, "loss_box_2": 1.71299, "loss_cns_2": 0.65643, "loss_yns_2": 0.1585, "loss_cls_3": 0.93697, "loss_box_3": 1.69808, "loss_cns_3": 0.65846, "loss_yns_3": 0.15841, "loss_cls_4": 0.95096, "loss_box_4": 1.69658, "loss_cns_4": 0.65867, "loss_yns_4": 0.15794, "loss_cls_5": 0.94313, "loss_box_5": 1.72655, "loss_cns_5": 0.65623, "loss_yns_5": 0.15779, "loss_cls_dn_0": 0.21312, "loss_box_dn_0": 0.7732, "loss_cls_dn_1": 0.13611, "loss_box_dn_1": 0.74242, "loss_cls_dn_2": 0.1333, "loss_box_dn_2": 0.72265, "loss_cls_dn_3": 0.13477, "loss_box_dn_3": 0.71988, "loss_cls_dn_4": 0.14075, "loss_box_dn_4": 0.72329, "loss_cls_dn_5": 0.14616, "loss_box_dn_5": 0.74049, "loss_dense_depth": 0.72207, "loss": 26.79321, "grad_norm": 30.13081, "time": 1.56603} -{"mode": "train", "epoch": 1, "iter": 165, "lr": 0.00017, "memory": 49163, "data_time": 0.09093, "loss_cls_0": 0.84094, "loss_box_0": 1.78021, "loss_cns_0": 0.6178, "loss_yns_0": 0.16075, "loss_cls_1": 0.90099, "loss_box_1": 1.77302, "loss_cns_1": 0.64764, "loss_yns_1": 0.15962, "loss_cls_2": 0.93941, "loss_box_2": 1.71765, "loss_cns_2": 0.65456, "loss_yns_2": 0.16104, "loss_cls_3": 0.939, "loss_box_3": 1.69811, "loss_cns_3": 0.65728, "loss_yns_3": 0.16108, "loss_cls_4": 0.95482, "loss_box_4": 1.69224, "loss_cns_4": 0.65653, "loss_yns_4": 0.16072, "loss_cls_5": 0.94775, "loss_box_5": 1.71119, "loss_cns_5": 0.65611, "loss_yns_5": 0.16066, "loss_cls_dn_0": 0.21653, "loss_box_dn_0": 0.76922, "loss_cls_dn_1": 0.13447, "loss_box_dn_1": 0.74586, "loss_cls_dn_2": 0.13251, "loss_box_dn_2": 0.71746, "loss_cls_dn_3": 0.13538, "loss_box_dn_3": 0.71464, "loss_cls_dn_4": 0.13739, "loss_box_dn_4": 0.71231, "loss_cls_dn_5": 0.14495, "loss_box_dn_5": 0.72535, "loss_dense_depth": 0.75151, "loss": 26.78669, "grad_norm": 32.62266, "time": 1.50377} -{"mode": "train", "epoch": 1, "iter": 166, "lr": 0.00017, "memory": 49163, "data_time": 0.0739, "loss_cls_0": 0.82259, "loss_box_0": 1.78224, "loss_cns_0": 0.61308, "loss_yns_0": 0.15862, "loss_cls_1": 0.90227, "loss_box_1": 1.72296, "loss_cns_1": 0.65223, "loss_yns_1": 0.16016, "loss_cls_2": 0.92341, "loss_box_2": 1.68574, "loss_cns_2": 0.65646, "loss_yns_2": 0.1608, "loss_cls_3": 0.93777, "loss_box_3": 1.67362, "loss_cns_3": 0.65919, "loss_yns_3": 0.1615, "loss_cls_4": 0.945, "loss_box_4": 1.66722, "loss_cns_4": 0.65812, "loss_yns_4": 0.16255, "loss_cls_5": 0.93612, "loss_box_5": 1.68553, "loss_cns_5": 0.65783, "loss_yns_5": 0.16064, "loss_cls_dn_0": 0.21012, "loss_box_dn_0": 0.76533, "loss_cls_dn_1": 0.13252, "loss_box_dn_1": 0.72992, "loss_cls_dn_2": 0.13207, "loss_box_dn_2": 0.71241, "loss_cls_dn_3": 0.14093, "loss_box_dn_3": 0.71572, "loss_cls_dn_4": 0.13989, "loss_box_dn_4": 0.71288, "loss_cls_dn_5": 0.14661, "loss_box_dn_5": 0.72328, "loss_dense_depth": 0.73664, "loss": 26.54398, "grad_norm": 32.14774, "time": 1.5021} -{"mode": "train", "epoch": 1, "iter": 167, "lr": 0.00017, "memory": 49163, "data_time": 0.07596, "loss_cls_0": 0.86649, "loss_box_0": 1.77327, "loss_cns_0": 0.60993, "loss_yns_0": 0.16023, "loss_cls_1": 0.92982, "loss_box_1": 1.73242, "loss_cns_1": 0.64802, "loss_yns_1": 0.16169, "loss_cls_2": 0.94645, "loss_box_2": 1.6863, "loss_cns_2": 0.6536, "loss_yns_2": 0.16361, "loss_cls_3": 0.96382, "loss_box_3": 1.68261, "loss_cns_3": 0.65447, "loss_yns_3": 0.16424, "loss_cls_4": 0.97095, "loss_box_4": 1.68447, "loss_cns_4": 0.6529, "loss_yns_4": 0.16289, "loss_cls_5": 0.96259, "loss_box_5": 1.69081, "loss_cns_5": 0.6509, "loss_yns_5": 0.16103, "loss_cls_dn_0": 0.21936, "loss_box_dn_0": 0.77191, "loss_cls_dn_1": 0.13805, "loss_box_dn_1": 0.73886, "loss_cls_dn_2": 0.13963, "loss_box_dn_2": 0.72264, "loss_cls_dn_3": 0.14368, "loss_box_dn_3": 0.72965, "loss_cls_dn_4": 0.14663, "loss_box_dn_4": 0.73419, "loss_cls_dn_5": 0.15239, "loss_box_dn_5": 0.73985, "loss_dense_depth": 0.77691, "loss": 26.88726, "grad_norm": 34.9921, "time": 6.70366} -{"mode": "train", "epoch": 1, "iter": 168, "lr": 0.00017, "memory": 49163, "data_time": 0.07245, "loss_cls_0": 0.85409, "loss_box_0": 1.73282, "loss_cns_0": 0.61264, "loss_yns_0": 0.16045, "loss_cls_1": 0.9137, "loss_box_1": 1.71829, "loss_cns_1": 0.64932, "loss_yns_1": 0.16075, "loss_cls_2": 0.93924, "loss_box_2": 1.67442, "loss_cns_2": 0.65493, "loss_yns_2": 0.1629, "loss_cls_3": 0.95143, "loss_box_3": 1.65176, "loss_cns_3": 0.65594, "loss_yns_3": 0.16302, "loss_cls_4": 0.96357, "loss_box_4": 1.6652, "loss_cns_4": 0.65557, "loss_yns_4": 0.16183, "loss_cls_5": 0.95388, "loss_box_5": 1.66769, "loss_cns_5": 0.65439, "loss_yns_5": 0.16144, "loss_cls_dn_0": 0.21429, "loss_box_dn_0": 0.7635, "loss_cls_dn_1": 0.13768, "loss_box_dn_1": 0.7358, "loss_cls_dn_2": 0.13749, "loss_box_dn_2": 0.721, "loss_cls_dn_3": 0.13954, "loss_box_dn_3": 0.72469, "loss_cls_dn_4": 0.14614, "loss_box_dn_4": 0.73594, "loss_cls_dn_5": 0.15286, "loss_box_dn_5": 0.74707, "loss_dense_depth": 0.74913, "loss": 26.6444, "grad_norm": 28.86006, "time": 1.46593} -{"mode": "train", "epoch": 1, "iter": 169, "lr": 0.00017, "memory": 49163, "data_time": 0.07442, "loss_cls_0": 0.88681, "loss_box_0": 1.76461, "loss_cns_0": 0.61007, "loss_yns_0": 0.16311, "loss_cls_1": 0.94275, "loss_box_1": 1.72059, "loss_cns_1": 0.63719, "loss_yns_1": 0.16027, "loss_cls_2": 0.95898, "loss_box_2": 1.68803, "loss_cns_2": 0.64926, "loss_yns_2": 0.16257, "loss_cls_3": 0.98085, "loss_box_3": 1.66936, "loss_cns_3": 0.65169, "loss_yns_3": 0.16176, "loss_cls_4": 0.97499, "loss_box_4": 1.68491, "loss_cns_4": 0.65251, "loss_yns_4": 0.16421, "loss_cls_5": 0.97182, "loss_box_5": 1.69329, "loss_cns_5": 0.64956, "loss_yns_5": 0.1643, "loss_cls_dn_0": 0.22285, "loss_box_dn_0": 0.76994, "loss_cls_dn_1": 0.13792, "loss_box_dn_1": 0.75055, "loss_cls_dn_2": 0.13361, "loss_box_dn_2": 0.73789, "loss_cls_dn_3": 0.13672, "loss_box_dn_3": 0.74179, "loss_cls_dn_4": 0.14204, "loss_box_dn_4": 0.7562, "loss_cls_dn_5": 0.1493, "loss_box_dn_5": 0.77078, "loss_dense_depth": 0.81141, "loss": 27.02449, "grad_norm": 38.2, "time": 1.48044} -{"mode": "train", "epoch": 1, "iter": 170, "lr": 0.00017, "memory": 49163, "data_time": 0.07745, "loss_cls_0": 0.87023, "loss_box_0": 1.74711, "loss_cns_0": 0.62095, "loss_yns_0": 0.16218, "loss_cls_1": 0.9285, "loss_box_1": 1.69453, "loss_cns_1": 0.64461, "loss_yns_1": 0.15913, "loss_cls_2": 0.96358, "loss_box_2": 1.64901, "loss_cns_2": 0.65631, "loss_yns_2": 0.16156, "loss_cls_3": 0.96736, "loss_box_3": 1.63898, "loss_cns_3": 0.66001, "loss_yns_3": 0.16253, "loss_cls_4": 0.98658, "loss_box_4": 1.63796, "loss_cns_4": 0.65837, "loss_yns_4": 0.16248, "loss_cls_5": 0.97469, "loss_box_5": 1.64583, "loss_cns_5": 0.65657, "loss_yns_5": 0.16338, "loss_cls_dn_0": 0.222, "loss_box_dn_0": 0.76608, "loss_cls_dn_1": 0.13877, "loss_box_dn_1": 0.74069, "loss_cls_dn_2": 0.13478, "loss_box_dn_2": 0.72807, "loss_cls_dn_3": 0.1381, "loss_box_dn_3": 0.73247, "loss_cls_dn_4": 0.14511, "loss_box_dn_4": 0.7456, "loss_cls_dn_5": 0.15031, "loss_box_dn_5": 0.75965, "loss_dense_depth": 0.75515, "loss": 26.72924, "grad_norm": 34.04764, "time": 1.49639} -{"mode": "train", "epoch": 1, "iter": 171, "lr": 0.00017, "memory": 49163, "data_time": 0.07603, "loss_cls_0": 0.85893, "loss_box_0": 1.73489, "loss_cns_0": 0.61666, "loss_yns_0": 0.15933, "loss_cls_1": 0.92387, "loss_box_1": 1.7025, "loss_cns_1": 0.64641, "loss_yns_1": 0.15845, "loss_cls_2": 0.95997, "loss_box_2": 1.66399, "loss_cns_2": 0.65459, "loss_yns_2": 0.16027, "loss_cls_3": 0.96374, "loss_box_3": 1.66543, "loss_cns_3": 0.65336, "loss_yns_3": 0.16037, "loss_cls_4": 0.97114, "loss_box_4": 1.66456, "loss_cns_4": 0.65388, "loss_yns_4": 0.15997, "loss_cls_5": 0.96205, "loss_box_5": 1.66048, "loss_cns_5": 0.65335, "loss_yns_5": 0.16147, "loss_cls_dn_0": 0.2187, "loss_box_dn_0": 0.76708, "loss_cls_dn_1": 0.13683, "loss_box_dn_1": 0.74321, "loss_cls_dn_2": 0.13608, "loss_box_dn_2": 0.72846, "loss_cls_dn_3": 0.13784, "loss_box_dn_3": 0.73278, "loss_cls_dn_4": 0.14596, "loss_box_dn_4": 0.74536, "loss_cls_dn_5": 0.15284, "loss_box_dn_5": 0.75225, "loss_dense_depth": 0.72615, "loss": 26.69319, "grad_norm": 41.42186, "time": 1.48797} -{"mode": "train", "epoch": 1, "iter": 172, "lr": 0.00017, "memory": 49163, "data_time": 0.07525, "loss_cls_0": 0.87865, "loss_box_0": 1.76739, "loss_cns_0": 0.61268, "loss_yns_0": 0.16027, "loss_cls_1": 0.92219, "loss_box_1": 1.72877, "loss_cns_1": 0.64255, "loss_yns_1": 0.15945, "loss_cls_2": 0.96048, "loss_box_2": 1.68031, "loss_cns_2": 0.65105, "loss_yns_2": 0.16041, "loss_cls_3": 0.96596, "loss_box_3": 1.67648, "loss_cns_3": 0.65391, "loss_yns_3": 0.16154, "loss_cls_4": 0.98507, "loss_box_4": 1.6699, "loss_cns_4": 0.65014, "loss_yns_4": 0.15966, "loss_cls_5": 0.971, "loss_box_5": 1.6768, "loss_cns_5": 0.65083, "loss_yns_5": 0.16121, "loss_cls_dn_0": 0.22369, "loss_box_dn_0": 0.76144, "loss_cls_dn_1": 0.13532, "loss_box_dn_1": 0.74694, "loss_cls_dn_2": 0.13282, "loss_box_dn_2": 0.72751, "loss_cls_dn_3": 0.13352, "loss_box_dn_3": 0.72698, "loss_cls_dn_4": 0.14105, "loss_box_dn_4": 0.73577, "loss_cls_dn_5": 0.14977, "loss_box_dn_5": 0.7368, "loss_dense_depth": 0.74333, "loss": 26.80163, "grad_norm": 37.34507, "time": 1.48645} -{"mode": "train", "epoch": 1, "iter": 173, "lr": 0.00017, "memory": 49163, "data_time": 0.07942, "loss_cls_0": 0.86108, "loss_box_0": 1.75747, "loss_cns_0": 0.61986, "loss_yns_0": 0.16141, "loss_cls_1": 0.91508, "loss_box_1": 1.73669, "loss_cns_1": 0.64445, "loss_yns_1": 0.16056, "loss_cls_2": 0.95485, "loss_box_2": 1.68935, "loss_cns_2": 0.65376, "loss_yns_2": 0.1595, "loss_cls_3": 0.9592, "loss_box_3": 1.679, "loss_cns_3": 0.65471, "loss_yns_3": 0.1595, "loss_cls_4": 0.98413, "loss_box_4": 1.67061, "loss_cns_4": 0.65231, "loss_yns_4": 0.15914, "loss_cls_5": 0.9572, "loss_box_5": 1.70923, "loss_cns_5": 0.65237, "loss_yns_5": 0.15966, "loss_cls_dn_0": 0.22318, "loss_box_dn_0": 0.76326, "loss_cls_dn_1": 0.13666, "loss_box_dn_1": 0.72947, "loss_cls_dn_2": 0.13555, "loss_box_dn_2": 0.71224, "loss_cls_dn_3": 0.13414, "loss_box_dn_3": 0.7103, "loss_cls_dn_4": 0.14204, "loss_box_dn_4": 0.71158, "loss_cls_dn_5": 0.14579, "loss_box_dn_5": 0.72257, "loss_dense_depth": 0.74161, "loss": 26.71952, "grad_norm": 39.64397, "time": 1.48551} -{"mode": "train", "epoch": 1, "iter": 174, "lr": 0.00017, "memory": 49163, "data_time": 0.07592, "loss_cls_0": 0.85765, "loss_box_0": 1.77252, "loss_cns_0": 0.61508, "loss_yns_0": 0.1588, "loss_cls_1": 0.93128, "loss_box_1": 1.74101, "loss_cns_1": 0.64546, "loss_yns_1": 0.15896, "loss_cls_2": 0.93833, "loss_box_2": 1.70496, "loss_cns_2": 0.64924, "loss_yns_2": 0.15736, "loss_cls_3": 0.95638, "loss_box_3": 1.69928, "loss_cns_3": 0.65153, "loss_yns_3": 0.15782, "loss_cls_4": 0.95675, "loss_box_4": 1.68915, "loss_cns_4": 0.65022, "loss_yns_4": 0.15896, "loss_cls_5": 0.95562, "loss_box_5": 1.71468, "loss_cns_5": 0.64807, "loss_yns_5": 0.15978, "loss_cls_dn_0": 0.21846, "loss_box_dn_0": 0.77185, "loss_cls_dn_1": 0.1382, "loss_box_dn_1": 0.73151, "loss_cls_dn_2": 0.13803, "loss_box_dn_2": 0.71584, "loss_cls_dn_3": 0.13751, "loss_box_dn_3": 0.72055, "loss_cls_dn_4": 0.14546, "loss_box_dn_4": 0.72085, "loss_cls_dn_5": 0.15068, "loss_box_dn_5": 0.73451, "loss_dense_depth": 0.74817, "loss": 26.80052, "grad_norm": 39.02592, "time": 1.49135} -{"mode": "train", "epoch": 1, "iter": 175, "lr": 0.00017, "memory": 49163, "data_time": 0.08516, "loss_cls_0": 0.85582, "loss_box_0": 1.74495, "loss_cns_0": 0.61913, "loss_yns_0": 0.1594, "loss_cls_1": 0.9232, "loss_box_1": 1.71002, "loss_cns_1": 0.64807, "loss_yns_1": 0.15776, "loss_cls_2": 0.94623, "loss_box_2": 1.67929, "loss_cns_2": 0.6518, "loss_yns_2": 0.15815, "loss_cls_3": 0.96115, "loss_box_3": 1.67675, "loss_cns_3": 0.65361, "loss_yns_3": 0.15876, "loss_cls_4": 0.9603, "loss_box_4": 1.67453, "loss_cns_4": 0.65298, "loss_yns_4": 0.15771, "loss_cls_5": 0.95369, "loss_box_5": 1.69909, "loss_cns_5": 0.65162, "loss_yns_5": 0.16106, "loss_cls_dn_0": 0.22121, "loss_box_dn_0": 0.76369, "loss_cls_dn_1": 0.13606, "loss_box_dn_1": 0.73683, "loss_cls_dn_2": 0.13216, "loss_box_dn_2": 0.72692, "loss_cls_dn_3": 0.13219, "loss_box_dn_3": 0.73783, "loss_cls_dn_4": 0.14045, "loss_box_dn_4": 0.74878, "loss_cls_dn_5": 0.14908, "loss_box_dn_5": 0.7629, "loss_dense_depth": 0.75581, "loss": 26.759, "grad_norm": 46.05129, "time": 1.48582} -{"mode": "train", "epoch": 1, "iter": 176, "lr": 0.00017, "memory": 49163, "data_time": 0.07562, "loss_cls_0": 0.85295, "loss_box_0": 1.73073, "loss_cns_0": 0.61608, "loss_yns_0": 0.15424, "loss_cls_1": 0.93706, "loss_box_1": 1.71195, "loss_cns_1": 0.6458, "loss_yns_1": 0.15384, "loss_cls_2": 0.94858, "loss_box_2": 1.67657, "loss_cns_2": 0.65247, "loss_yns_2": 0.15248, "loss_cls_3": 0.95263, "loss_box_3": 1.67058, "loss_cns_3": 0.65236, "loss_yns_3": 0.15351, "loss_cls_4": 0.9609, "loss_box_4": 1.67629, "loss_cns_4": 0.65388, "loss_yns_4": 0.15328, "loss_cls_5": 0.95396, "loss_box_5": 1.68534, "loss_cns_5": 0.65112, "loss_yns_5": 0.15393, "loss_cls_dn_0": 0.22362, "loss_box_dn_0": 0.77183, "loss_cls_dn_1": 0.13923, "loss_box_dn_1": 0.73861, "loss_cls_dn_2": 0.13444, "loss_box_dn_2": 0.72995, "loss_cls_dn_3": 0.13414, "loss_box_dn_3": 0.73892, "loss_cls_dn_4": 0.14091, "loss_box_dn_4": 0.75104, "loss_cls_dn_5": 0.14963, "loss_box_dn_5": 0.76326, "loss_dense_depth": 0.78078, "loss": 26.7469, "grad_norm": 38.08336, "time": 1.49741} -{"mode": "train", "epoch": 1, "iter": 177, "lr": 0.00017, "memory": 49163, "data_time": 0.07493, "loss_cls_0": 0.84694, "loss_box_0": 1.71666, "loss_cns_0": 0.61536, "loss_yns_0": 0.15207, "loss_cls_1": 0.93423, "loss_box_1": 1.67435, "loss_cns_1": 0.64682, "loss_yns_1": 0.15255, "loss_cls_2": 0.94263, "loss_box_2": 1.65818, "loss_cns_2": 0.65457, "loss_yns_2": 0.15009, "loss_cls_3": 0.94907, "loss_box_3": 1.65759, "loss_cns_3": 0.65806, "loss_yns_3": 0.15129, "loss_cls_4": 0.95767, "loss_box_4": 1.65895, "loss_cns_4": 0.65533, "loss_yns_4": 0.15166, "loss_cls_5": 0.9453, "loss_box_5": 1.66676, "loss_cns_5": 0.65256, "loss_yns_5": 0.15116, "loss_cls_dn_0": 0.22349, "loss_box_dn_0": 0.76419, "loss_cls_dn_1": 0.13699, "loss_box_dn_1": 0.75113, "loss_cls_dn_2": 0.13493, "loss_box_dn_2": 0.73776, "loss_cls_dn_3": 0.13742, "loss_box_dn_3": 0.7448, "loss_cls_dn_4": 0.14052, "loss_box_dn_4": 0.75203, "loss_cls_dn_5": 0.14817, "loss_box_dn_5": 0.7622, "loss_dense_depth": 0.75659, "loss": 26.59009, "grad_norm": 39.23107, "time": 1.4914} -{"mode": "train", "epoch": 1, "iter": 178, "lr": 0.00017, "memory": 49163, "data_time": 0.07002, "loss_cls_0": 0.84056, "loss_box_0": 1.72977, "loss_cns_0": 0.6168, "loss_yns_0": 0.1521, "loss_cls_1": 0.90801, "loss_box_1": 1.70553, "loss_cns_1": 0.64693, "loss_yns_1": 0.15144, "loss_cls_2": 0.92816, "loss_box_2": 1.69647, "loss_cns_2": 0.65007, "loss_yns_2": 0.1521, "loss_cls_3": 0.94277, "loss_box_3": 1.691, "loss_cns_3": 0.65699, "loss_yns_3": 0.15203, "loss_cls_4": 0.96736, "loss_box_4": 1.68463, "loss_cns_4": 0.65445, "loss_yns_4": 0.15196, "loss_cls_5": 0.94172, "loss_box_5": 1.69325, "loss_cns_5": 0.6553, "loss_yns_5": 0.15212, "loss_cls_dn_0": 0.2209, "loss_box_dn_0": 0.7638, "loss_cls_dn_1": 0.13768, "loss_box_dn_1": 0.75142, "loss_cls_dn_2": 0.13667, "loss_box_dn_2": 0.73799, "loss_cls_dn_3": 0.13929, "loss_box_dn_3": 0.74396, "loss_cls_dn_4": 0.14257, "loss_box_dn_4": 0.74632, "loss_cls_dn_5": 0.14882, "loss_box_dn_5": 0.7548, "loss_dense_depth": 0.75834, "loss": 26.70405, "grad_norm": 43.59504, "time": 1.4729} -{"mode": "train", "epoch": 1, "iter": 179, "lr": 0.00017, "memory": 49163, "data_time": 0.07074, "loss_cls_0": 0.83207, "loss_box_0": 1.72603, "loss_cns_0": 0.61694, "loss_yns_0": 0.15247, "loss_cls_1": 0.90382, "loss_box_1": 1.71599, "loss_cns_1": 0.64647, "loss_yns_1": 0.1518, "loss_cls_2": 0.92149, "loss_box_2": 1.68119, "loss_cns_2": 0.65005, "loss_yns_2": 0.15153, "loss_cls_3": 0.93551, "loss_box_3": 1.67054, "loss_cns_3": 0.65644, "loss_yns_3": 0.15182, "loss_cls_4": 0.95478, "loss_box_4": 1.65999, "loss_cns_4": 0.65612, "loss_yns_4": 0.15226, "loss_cls_5": 0.94024, "loss_box_5": 1.66321, "loss_cns_5": 0.65292, "loss_yns_5": 0.15219, "loss_cls_dn_0": 0.22007, "loss_box_dn_0": 0.76879, "loss_cls_dn_1": 0.13729, "loss_box_dn_1": 0.73983, "loss_cls_dn_2": 0.13638, "loss_box_dn_2": 0.72143, "loss_cls_dn_3": 0.13769, "loss_box_dn_3": 0.719, "loss_cls_dn_4": 0.14009, "loss_box_dn_4": 0.71839, "loss_cls_dn_5": 0.14427, "loss_box_dn_5": 0.72168, "loss_dense_depth": 0.74654, "loss": 26.44732, "grad_norm": 27.24173, "time": 1.47536} -{"mode": "train", "epoch": 1, "iter": 180, "lr": 0.00017, "memory": 49163, "data_time": 0.08433, "loss_cls_0": 0.83273, "loss_box_0": 1.68584, "loss_cns_0": 0.61552, "loss_yns_0": 0.14918, "loss_cls_1": 0.88853, "loss_box_1": 1.68189, "loss_cns_1": 0.64648, "loss_yns_1": 0.15044, "loss_cls_2": 0.92911, "loss_box_2": 1.62403, "loss_cns_2": 0.64883, "loss_yns_2": 0.15001, "loss_cls_3": 0.92581, "loss_box_3": 1.62447, "loss_cns_3": 0.65489, "loss_yns_3": 0.15023, "loss_cls_4": 0.93621, "loss_box_4": 1.62711, "loss_cns_4": 0.65407, "loss_yns_4": 0.15056, "loss_cls_5": 0.93065, "loss_box_5": 1.63163, "loss_cns_5": 0.65311, "loss_yns_5": 0.15051, "loss_cls_dn_0": 0.22084, "loss_box_dn_0": 0.76962, "loss_cls_dn_1": 0.13414, "loss_box_dn_1": 0.73374, "loss_cls_dn_2": 0.13213, "loss_box_dn_2": 0.71199, "loss_cls_dn_3": 0.13301, "loss_box_dn_3": 0.70544, "loss_cls_dn_4": 0.13981, "loss_box_dn_4": 0.70833, "loss_cls_dn_5": 0.14271, "loss_box_dn_5": 0.71027, "loss_dense_depth": 0.76854, "loss": 26.10241, "grad_norm": 33.54756, "time": 1.49457} -{"mode": "train", "epoch": 1, "iter": 181, "lr": 0.00017, "memory": 49163, "data_time": 0.07161, "loss_cls_0": 0.819, "loss_box_0": 1.73519, "loss_cns_0": 0.61451, "loss_yns_0": 0.15121, "loss_cls_1": 0.87842, "loss_box_1": 1.70619, "loss_cns_1": 0.63883, "loss_yns_1": 0.14901, "loss_cls_2": 0.91904, "loss_box_2": 1.65273, "loss_cns_2": 0.64771, "loss_yns_2": 0.14959, "loss_cls_3": 0.92528, "loss_box_3": 1.64329, "loss_cns_3": 0.65374, "loss_yns_3": 0.14954, "loss_cls_4": 0.94012, "loss_box_4": 1.63185, "loss_cns_4": 0.65331, "loss_yns_4": 0.15024, "loss_cls_5": 0.92988, "loss_box_5": 1.62553, "loss_cns_5": 0.65116, "loss_yns_5": 0.14988, "loss_cls_dn_0": 0.22001, "loss_box_dn_0": 0.77378, "loss_cls_dn_1": 0.13193, "loss_box_dn_1": 0.71953, "loss_cls_dn_2": 0.13166, "loss_box_dn_2": 0.70177, "loss_cls_dn_3": 0.13378, "loss_box_dn_3": 0.70622, "loss_cls_dn_4": 0.14251, "loss_box_dn_4": 0.71112, "loss_cls_dn_5": 0.14506, "loss_box_dn_5": 0.71711, "loss_dense_depth": 0.79401, "loss": 26.19378, "grad_norm": 38.40791, "time": 1.5687} -{"mode": "train", "epoch": 1, "iter": 182, "lr": 0.00017, "memory": 49163, "data_time": 0.17743, "loss_cls_0": 0.82154, "loss_box_0": 1.71117, "loss_cns_0": 0.62099, "loss_yns_0": 0.14834, "loss_cls_1": 0.87567, "loss_box_1": 1.72164, "loss_cns_1": 0.64358, "loss_yns_1": 0.14826, "loss_cls_2": 0.91658, "loss_box_2": 1.66919, "loss_cns_2": 0.65006, "loss_yns_2": 0.15033, "loss_cls_3": 0.92961, "loss_box_3": 1.65163, "loss_cns_3": 0.65539, "loss_yns_3": 0.14882, "loss_cls_4": 0.92566, "loss_box_4": 1.64144, "loss_cns_4": 0.65383, "loss_yns_4": 0.14936, "loss_cls_5": 0.91061, "loss_box_5": 1.64823, "loss_cns_5": 0.65158, "loss_yns_5": 0.14901, "loss_cls_dn_0": 0.21711, "loss_box_dn_0": 0.77493, "loss_cls_dn_1": 0.13333, "loss_box_dn_1": 0.72915, "loss_cls_dn_2": 0.13395, "loss_box_dn_2": 0.72431, "loss_cls_dn_3": 0.13829, "loss_box_dn_3": 0.73486, "loss_cls_dn_4": 0.14147, "loss_box_dn_4": 0.74477, "loss_cls_dn_5": 0.1447, "loss_box_dn_5": 0.76438, "loss_dense_depth": 0.74449, "loss": 26.31825, "grad_norm": 42.43811, "time": 1.58587} diff --git a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20251117_141745.log b/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20251117_141745.log deleted file mode 100644 index 482129f51ee7511d34ab09a3e05ada066c3bfbde..0000000000000000000000000000000000000000 --- a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20251117_141745.log +++ /dev/null @@ -1,3644 +0,0 @@ -2025-11-17 14:17:45,747 - mmdet - INFO - Environment info: ------------------------------------------------------------- -sys.platform: linux -Python: 3.10.12 (main, May 27 2025, 17:12:29) [GCC 11.4.0] -CUDA available: True -GPU 0,1,2,3,4,5,6,7: BW200, UBB BW1000 -CUDA_HOME: /opt/dtk -NVCC: Not Available -GCC: x86_64-linux-gnu-gcc (Ubuntu 11.4.0-1ubuntu1~22.04) 11.4.0 -PyTorch: 2.4.1 -PyTorch compiling details: PyTorch built with: - - GCC 10.3 - - C++ Version: 201703 - - Intel(R) Math Kernel Library Version 2020.0.4 Product Build 20200917 for Intel(R) 64 architecture applications - - OpenMP 201511 (a.k.a. OpenMP 4.5) - - LAPACK is enabled (usually provided by MKL) - - NNPACK is enabled - - CPU capability usage: AVX512 - - HIP Runtime 6.3.25211 - - MIOpen 2.17.0 - - Magma 2.8.0 - - Build settings: BLAS_INFO=mkl, BUILD_TYPE=Release, CXX_COMPILER=/opt/rh/gcc-toolset-10/root/usr/bin/c++, CXX_FLAGS= -D_GLIBCXX_USE_CXX11_ABI=1 -fvisibility-inlines-hidden -DUSE_PTHREADPOOL -DNDEBUG -DUSE_KINETO -DLIBKINETO_NOCUPTI -DUSE_FBGEMM -DUSE_PYTORCH_QNNPACK -DUSE_XNNPACK -DSYMBOLICATE_MOBILE_DEBUG_HANDLE -O2 -fPIC -Wall -Wextra -Werror=return-type -Werror=non-virtual-dtor -Werror=bool-operation -Wnarrowing -Wno-missing-field-initializers -Wno-type-limits -Wno-array-bounds -Wno-unknown-pragmas -Wno-unused-parameter -Wno-unused-function -Wno-unused-result -Wno-strict-overflow -Wno-strict-aliasing -Wno-stringop-overflow -Wsuggest-override -Wno-psabi -Wno-error=pedantic -Wno-error=old-style-cast -Wno-missing-braces -fdiagnostics-color=always -faligned-new -Wno-unused-but-set-variable -Wno-maybe-uninitialized -fno-math-errno -fno-trapping-math -Werror=format -Wno-stringop-overflow, FORCE_FALLBACK_CUDA_MPI=1, LAPACK_INFO=mkl, PERF_WITH_AVX=1, PERF_WITH_AVX2=1, PERF_WITH_AVX512=1, TORCH_VERSION=2.4.1, USE_CUDA=0, USE_CUDNN=OFF, USE_CUSPARSELT=OFF, USE_EXCEPTION_PTR=1, USE_GFLAGS=1, USE_GLOG=1, USE_GLOO=1, USE_MKL=ON, USE_MKLDNN=0, USE_MPI=1, USE_NCCL=1, USE_NNPACK=ON, USE_OPENMP=1, USE_ROCM=ON, USE_ROCM_KERNEL_ASSERT=OFF, - -TorchVision: 0.19.1 -OpenCV: 4.11.0 -MMCV: 1.6.1 -MMCV Compiler: GCC 11.4 -MMCV CUDA Compiler: rocm not available -MMDetection: 2.26.0+c41df4b ------------------------------------------------------------- - -2025-11-17 14:17:46,714 - mmdet - INFO - Distributed training: True -2025-11-17 14:17:47,437 - mmdet - INFO - Config: -plugin = True -plugin_dir = 'projects/mmdet3d_plugin/' -dist_params = dict(backend='nccl') -log_level = 'INFO' -work_dir = './work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704' -total_batch_size = 160 -num_gpus = 8 -batch_size = 20 -num_iters_per_epoch = 175 -num_epochs = 100 -checkpoint_epoch_interval = 20 -checkpoint_config = dict(interval=3500) -log_config = dict( - interval=1, - hooks=[ - dict(type='TextLoggerHook', by_epoch=False), - dict(type='TensorboardLoggerHook') - ]) -load_from = None -resume_from = None -workflow = [('train', 1)] -fp16 = dict(loss_scale=32.0) -input_shape = (704, 256) -tracking_test = True -tracking_threshold = 0.2 -class_names = [ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier', - 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' -] -num_classes = 10 -embed_dims = 256 -num_groups = 8 -num_decoder = 6 -num_single_frame_decoder = 1 -use_deformable_func = True -strides = [4, 8, 16, 32] -num_levels = 4 -num_depth_layers = 3 -drop_out = 0.1 -temporal = True -decouple_attn = True -with_quality_estimation = True -model = dict( - type='Sparse4D', - use_grid_mask=True, - use_deformable_func=True, - img_backbone=dict( - type='ResNet', - depth=50, - num_stages=4, - frozen_stages=-1, - norm_eval=False, - style='pytorch', - with_cp=True, - out_indices=(0, 1, 2, 3), - norm_cfg=dict(type='BN', requires_grad=True), - pretrained='ckpt/resnet50-19c8e357.pth'), - img_neck=dict( - type='FPN', - num_outs=4, - start_level=0, - out_channels=256, - add_extra_convs='on_output', - relu_before_extra_convs=True, - in_channels=[256, 512, 1024, 2048]), - depth_branch=dict( - type='DenseDepthNet', - embed_dims=256, - num_depth_layers=3, - loss_weight=0.2), - head=dict( - type='Sparse4DHead', - cls_threshold_to_reg=0.05, - decouple_attn=True, - instance_bank=dict( - type='InstanceBank', - num_anchor=900, - embed_dims=256, - anchor='nuscenes_kmeans900.npy', - anchor_handler=dict(type='SparseBox3DKeyPointsGenerator'), - num_temp_instances=600, - confidence_decay=0.6, - feat_grad=False), - anchor_encoder=dict( - type='SparseBox3DEncoder', - vel_dims=3, - embed_dims=[128, 32, 32, 64], - mode='cat', - output_fc=False, - in_loops=1, - out_loops=4), - num_single_frame_decoder=1, - operation_order=[ - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine' - ], - temp_graph_model=dict( - type='MultiheadAttention', - embed_dims=512, - num_heads=8, - batch_first=True, - dropout=0.1), - graph_model=dict( - type='MultiheadAttention', - embed_dims=512, - num_heads=8, - batch_first=True, - dropout=0.1), - norm_layer=dict(type='LN', normalized_shape=256), - ffn=dict( - type='AsymmetricFFN', - in_channels=512, - pre_norm=dict(type='LN'), - embed_dims=256, - feedforward_channels=1024, - num_fcs=2, - ffn_drop=0.1, - act_cfg=dict(type='ReLU', inplace=True)), - deformable_model=dict( - type='DeformableFeatureAggregation', - embed_dims=256, - num_groups=8, - num_levels=4, - num_cams=6, - attn_drop=0.15, - use_deformable_func=True, - use_camera_embed=True, - residual_mode='cat', - kps_generator=dict( - type='SparseBox3DKeyPointsGenerator', - num_learnable_pts=6, - fix_scale=[[0, 0, 0], [0.45, 0, 0], [-0.45, 0, 0], - [0, 0.45, 0], [0, -0.45, 0], [0, 0, 0.45], - [0, 0, -0.45]])), - refine_layer=dict( - type='SparseBox3DRefinementModule', - embed_dims=256, - num_cls=10, - refine_yaw=True, - with_quality_estimation=True), - sampler=dict( - type='SparseBox3DTarget', - num_dn_groups=5, - num_temp_dn_groups=3, - dn_noise_scale=[2.0, 2.0, 2.0, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5], - max_dn_gt=32, - add_neg_dn=True, - cls_weight=2.0, - box_weight=0.25, - reg_weights=[2.0, 2.0, 2.0, 0.5, 0.5, 0.5, 0.0, 0.0, 0.0, 0.0], - cls_wise_reg_weights=dict( - {9: [2.0, 2.0, 2.0, 1.0, 1.0, 1.0, 0.0, 0.0, 1.0, 1.0]})), - loss_cls=dict( - type='FocalLoss', - use_sigmoid=True, - gamma=2.0, - alpha=0.25, - loss_weight=2.0), - loss_reg=dict( - type='SparseBox3DLoss', - loss_box=dict(type='L1Loss', loss_weight=0.25), - loss_centerness=dict(type='CrossEntropyLoss', use_sigmoid=True), - loss_yawness=dict(type='GaussianFocalLoss'), - cls_allow_reverse=[5]), - decoder=dict(type='SparseBox3DDecoder'), - reg_weights=[2.0, 2.0, 2.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0])) -dataset_type = 'NuScenes3DDetTrackDataset' -data_root = 'data/nuscenes/' -anno_root = 'data/nuscenes_anno_pkls/' -file_client_args = dict(backend='disk') -img_norm_cfg = dict( - mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375], to_rgb=True) -train_pipeline = [ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict( - type='LoadPointsFromFile', - coord_type='LIDAR', - load_dim=5, - use_dim=5, - file_client_args=dict(backend='disk')), - dict(type='ResizeCropFlipImage'), - dict(type='MultiScaleDepthMapGenerator', downsample=[4, 8, 16]), - dict(type='BBoxRotation'), - dict(type='PhotoMetricDistortionMultiViewImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict( - type='CircleObjectRangeFilter', - class_dist_thred=[55, 55, 55, 55, 55, 55, 55, 55, 55, 55]), - dict( - type='InstanceNameFilter', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ]), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=[ - 'img', 'timestamp', 'projection_mat', 'image_wh', 'gt_depth', - 'focal', 'gt_bboxes_3d', 'gt_labels_3d' - ], - meta_keys=['T_global', 'T_global_inv', 'timestamp', 'instance_id']) -] -test_pipeline = [ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='ResizeCropFlipImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=['img', 'timestamp', 'projection_mat', 'image_wh'], - meta_keys=['T_global', 'T_global_inv', 'timestamp']) -] -input_modality = dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False) -data_basic_config = dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier', - 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval') -data_aug_conf = dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]) -data = dict( - samples_per_gpu=20, - workers_per_gpu=20, - train=dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval', - ann_file='data/nuscenes_anno_pkls/nuscenes_infos_train.pkl', - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict( - type='LoadPointsFromFile', - coord_type='LIDAR', - load_dim=5, - use_dim=5, - file_client_args=dict(backend='disk')), - dict(type='ResizeCropFlipImage'), - dict(type='MultiScaleDepthMapGenerator', downsample=[4, 8, 16]), - dict(type='BBoxRotation'), - dict(type='PhotoMetricDistortionMultiViewImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict( - type='CircleObjectRangeFilter', - class_dist_thred=[55, 55, 55, 55, 55, 55, 55, 55, 55, 55]), - dict( - type='InstanceNameFilter', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', - 'traffic_cone' - ]), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=[ - 'img', 'timestamp', 'projection_mat', 'image_wh', - 'gt_depth', 'focal', 'gt_bboxes_3d', 'gt_labels_3d' - ], - meta_keys=[ - 'T_global', 'T_global_inv', 'timestamp', 'instance_id' - ]) - ], - test_mode=False, - data_aug_conf=dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]), - with_seq_flag=True, - sequences_split_num=2, - keep_consistent_seq_aug=True), - val=dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval', - ann_file='data/nuscenes_anno_pkls/nuscenes_infos_val.pkl', - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='ResizeCropFlipImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=['img', 'timestamp', 'projection_mat', 'image_wh'], - meta_keys=['T_global', 'T_global_inv', 'timestamp']) - ], - data_aug_conf=dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]), - test_mode=True, - tracking=True, - tracking_threshold=0.2), - test=dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval', - ann_file='data/nuscenes_anno_pkls/nuscenes_infos_val.pkl', - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='ResizeCropFlipImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=['img', 'timestamp', 'projection_mat', 'image_wh'], - meta_keys=['T_global', 'T_global_inv', 'timestamp']) - ], - data_aug_conf=dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]), - test_mode=True, - tracking=True, - tracking_threshold=0.2)) -optimizer = dict( - type='Miopen_AdamW', - lr=0.0006, - weight_decay=0.001, - paramwise_cfg=dict(custom_keys=dict(img_backbone=dict(lr_mult=0.5)))) -optimizer_config = dict(grad_clip=dict(max_norm=25, norm_type=2)) -lr_config = dict( - policy='CosineAnnealing', - warmup='linear', - warmup_iters=500, - warmup_ratio=0.3333333333333333, - min_lr_ratio=0.001) -runner = dict(type='IterBasedRunner', max_iters=17500) -vis_pipeline = [ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='Collect', keys=['img'], meta_keys=['timestamp', 'lidar2img']) -] -evaluation = dict( - interval=3500, - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict( - type='Collect', keys=['img'], meta_keys=['timestamp', 'lidar2img']) - ]) -gpu_ids = range(0, 8) - -2025-11-17 14:17:47,438 - mmdet - INFO - Set random seed to 0, deterministic: False -2025-11-17 14:17:47,736 - mmdet - INFO - initialize ResNet with init_cfg {'type': 'Pretrained', 'checkpoint': 'ckpt/resnet50-19c8e357.pth'} -2025-11-17 14:17:48,114 - mmdet - INFO - initialize FPN with init_cfg {'type': 'Xavier', 'layer': 'Conv2d', 'distribution': 'uniform'} -Name of parameter - Initialization information - -img_backbone.conv1.weight - torch.Size([64, 3, 7, 7]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.bn1.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.bn1.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.conv1.weight - torch.Size([64, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn1.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn1.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.conv2.weight - torch.Size([64, 64, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn2.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn2.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.conv3.weight - torch.Size([256, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn3.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn3.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.downsample.0.weight - torch.Size([256, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.downsample.1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.downsample.1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.conv1.weight - torch.Size([64, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn1.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn1.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.conv2.weight - torch.Size([64, 64, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn2.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn2.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.conv3.weight - torch.Size([256, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn3.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn3.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.conv1.weight - torch.Size([64, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn1.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn1.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.conv2.weight - torch.Size([64, 64, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn2.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn2.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.conv3.weight - torch.Size([256, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn3.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn3.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.conv1.weight - torch.Size([128, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn1.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn1.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.conv2.weight - torch.Size([128, 128, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn2.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn2.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.conv3.weight - torch.Size([512, 128, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn3.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn3.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.downsample.0.weight - torch.Size([512, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.downsample.1.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.downsample.1.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.conv1.weight - torch.Size([128, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn1.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn1.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.conv2.weight - torch.Size([128, 128, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn2.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn2.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.conv3.weight - torch.Size([512, 128, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn3.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn3.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.conv1.weight - torch.Size([128, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn1.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn1.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.conv2.weight - torch.Size([128, 128, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn2.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn2.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.conv3.weight - torch.Size([512, 128, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn3.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn3.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.conv1.weight - torch.Size([128, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn1.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn1.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.conv2.weight - torch.Size([128, 128, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn2.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn2.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.conv3.weight - torch.Size([512, 128, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn3.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn3.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.conv1.weight - torch.Size([256, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.downsample.0.weight - torch.Size([1024, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.downsample.1.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.downsample.1.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.conv1.weight - torch.Size([512, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn1.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn1.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.conv2.weight - torch.Size([512, 512, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn2.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn2.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.conv3.weight - torch.Size([2048, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn3.weight - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn3.bias - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.downsample.0.weight - torch.Size([2048, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.downsample.1.weight - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.downsample.1.bias - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.conv1.weight - torch.Size([512, 2048, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn1.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn1.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.conv2.weight - torch.Size([512, 512, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn2.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn2.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.conv3.weight - torch.Size([2048, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn3.weight - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn3.bias - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.conv1.weight - torch.Size([512, 2048, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn1.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn1.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.conv2.weight - torch.Size([512, 512, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn2.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn2.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.conv3.weight - torch.Size([2048, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn3.weight - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn3.bias - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_neck.lateral_convs.0.conv.weight - torch.Size([256, 256, 1, 1]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.lateral_convs.0.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.lateral_convs.1.conv.weight - torch.Size([256, 512, 1, 1]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.lateral_convs.1.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.lateral_convs.2.conv.weight - torch.Size([256, 1024, 1, 1]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.lateral_convs.2.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.lateral_convs.3.conv.weight - torch.Size([256, 2048, 1, 1]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.lateral_convs.3.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.fpn_convs.0.conv.weight - torch.Size([256, 256, 3, 3]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.fpn_convs.0.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.fpn_convs.1.conv.weight - torch.Size([256, 256, 3, 3]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.fpn_convs.1.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.fpn_convs.2.conv.weight - torch.Size([256, 256, 3, 3]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.fpn_convs.2.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.fpn_convs.3.conv.weight - torch.Size([256, 256, 3, 3]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.fpn_convs.3.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.instance_bank.anchor - torch.Size([900, 11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.instance_bank.instance_feature - torch.Size([900, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.instance_bank.anchor_handler.fix_scale - torch.Size([1, 3]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.0.weight - torch.Size([128, 3]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.0.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.2.weight - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.2.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.3.weight - torch.Size([128, 128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.3.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.5.weight - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.5.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.6.weight - torch.Size([128, 128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.6.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.8.weight - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.8.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.9.weight - torch.Size([128, 128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.9.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.11.weight - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.11.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.0.weight - torch.Size([32, 3]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.0.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.2.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.2.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.3.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.3.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.5.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.5.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.6.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.6.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.8.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.8.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.9.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.9.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.11.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.11.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.0.weight - torch.Size([32, 2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.0.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.2.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.2.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.3.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.3.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.5.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.5.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.6.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.6.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.8.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.8.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.9.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.9.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.11.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.11.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.0.weight - torch.Size([64, 3]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.0.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.2.weight - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.2.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.3.weight - torch.Size([64, 64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.3.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.5.weight - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.5.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.6.weight - torch.Size([64, 64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.6.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.8.weight - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.8.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.9.weight - torch.Size([64, 64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.9.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.11.weight - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.11.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.1.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.1.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.1.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.1.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.1.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.1.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.1.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.1.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.3.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.4.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.4.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.4.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.4.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.5.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.5.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.5.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.5.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.6.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.6.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.8.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.8.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.8.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.8.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.8.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.8.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.8.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.8.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.10.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.11.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.11.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.11.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.11.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.12.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.12.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.12.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.12.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.13.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.13.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.15.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.15.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.15.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.15.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.15.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.15.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.15.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.15.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.16.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.16.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.17.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.18.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.18.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.18.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.18.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.19.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.19.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.19.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.19.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.20.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.20.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.22.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.22.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.22.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.22.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.22.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.22.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.22.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.22.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.23.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.23.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.24.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.25.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.25.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.25.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.25.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.26.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.26.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.26.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.26.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.27.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.27.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.29.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.29.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.29.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.29.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.29.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.29.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.29.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.29.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.30.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.30.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.31.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.32.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.32.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.32.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.32.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.33.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.33.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.33.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.33.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.34.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.34.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.36.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.36.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.36.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.36.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.36.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.36.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.36.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.36.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.37.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.37.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.38.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.fc_before.weight - torch.Size([512, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.fc_after.weight - torch.Size([256, 512]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.0.weight - torch.Size([1, 256, 1, 1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.0.bias - torch.Size([1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.1.weight - torch.Size([1, 256, 1, 1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.1.bias - torch.Size([1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.2.weight - torch.Size([1, 256, 1, 1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.2.bias - torch.Size([1]): -The value is the same before and after calling `init_weights` of Sparse4D -2025-11-17 14:17:48,204 - mmdet - INFO - Model: -Sparse4D( - (img_backbone): ResNet( - (conv1): Conv2d(3, 64, kernel_size=(7, 7), stride=(2, 2), padding=(3, 3), bias=False) - (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (maxpool): MaxPool2d(kernel_size=3, stride=2, padding=1, dilation=1, ceil_mode=False) - (layer1): ResLayer( - (0): Bottleneck( - (conv1): Conv2d(64, 64, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (downsample): Sequential( - (0): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - ) - ) - (1): Bottleneck( - (conv1): Conv2d(256, 64, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (2): Bottleneck( - (conv1): Conv2d(256, 64, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - ) - (layer2): ResLayer( - (0): Bottleneck( - (conv1): Conv2d(256, 128, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (downsample): Sequential( - (0): Conv2d(256, 512, kernel_size=(1, 1), stride=(2, 2), bias=False) - (1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - ) - ) - (1): Bottleneck( - (conv1): Conv2d(512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (2): Bottleneck( - (conv1): Conv2d(512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (3): Bottleneck( - (conv1): Conv2d(512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - ) - (layer3): ResLayer( - (0): Bottleneck( - (conv1): Conv2d(512, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (downsample): Sequential( - (0): Conv2d(512, 1024, kernel_size=(1, 1), stride=(2, 2), bias=False) - (1): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - ) - ) - (1): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (2): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (3): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (4): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (5): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - ) - (layer4): ResLayer( - (0): Bottleneck( - (conv1): Conv2d(1024, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(512, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (downsample): Sequential( - (0): Conv2d(1024, 2048, kernel_size=(1, 1), stride=(2, 2), bias=False) - (1): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - ) - ) - (1): Bottleneck( - (conv1): Conv2d(2048, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(512, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (2): Bottleneck( - (conv1): Conv2d(2048, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(512, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - ) - ) - init_cfg={'type': 'Pretrained', 'checkpoint': 'ckpt/resnet50-19c8e357.pth'} - (img_neck): FPN( - (lateral_convs): ModuleList( - (0): ConvModule( - (conv): Conv2d(256, 256, kernel_size=(1, 1), stride=(1, 1)) - ) - (1): ConvModule( - (conv): Conv2d(512, 256, kernel_size=(1, 1), stride=(1, 1)) - ) - (2): ConvModule( - (conv): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1)) - ) - (3): ConvModule( - (conv): Conv2d(2048, 256, kernel_size=(1, 1), stride=(1, 1)) - ) - ) - (fpn_convs): ModuleList( - (0-3): 4 x ConvModule( - (conv): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1)) - ) - ) - ) - init_cfg={'type': 'Xavier', 'layer': 'Conv2d', 'distribution': 'uniform'} - (head): Sparse4DHead( - (instance_bank): InstanceBank( - (anchor_handler): SparseBox3DKeyPointsGenerator() - ) - (anchor_encoder): SparseBox3DEncoder( - (pos_fc): Sequential( - (0): Linear(in_features=3, out_features=128, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((128,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=128, out_features=128, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((128,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=128, out_features=128, bias=True) - (7): ReLU(inplace=True) - (8): LayerNorm((128,), eps=1e-05, elementwise_affine=True) - (9): Linear(in_features=128, out_features=128, bias=True) - (10): ReLU(inplace=True) - (11): LayerNorm((128,), eps=1e-05, elementwise_affine=True) - ) - (size_fc): Sequential( - (0): Linear(in_features=3, out_features=32, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=32, out_features=32, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=32, out_features=32, bias=True) - (7): ReLU(inplace=True) - (8): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (9): Linear(in_features=32, out_features=32, bias=True) - (10): ReLU(inplace=True) - (11): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - ) - (yaw_fc): Sequential( - (0): Linear(in_features=2, out_features=32, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=32, out_features=32, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=32, out_features=32, bias=True) - (7): ReLU(inplace=True) - (8): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (9): Linear(in_features=32, out_features=32, bias=True) - (10): ReLU(inplace=True) - (11): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - ) - (vel_fc): Sequential( - (0): Linear(in_features=3, out_features=64, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((64,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=64, out_features=64, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((64,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=64, out_features=64, bias=True) - (7): ReLU(inplace=True) - (8): LayerNorm((64,), eps=1e-05, elementwise_affine=True) - (9): Linear(in_features=64, out_features=64, bias=True) - (10): ReLU(inplace=True) - (11): LayerNorm((64,), eps=1e-05, elementwise_affine=True) - ) - ) - (loss_cls): FocalLoss() - (loss_reg): SparseBox3DLoss( - (loss_box): L1Loss() - (loss_cns): CrossEntropyLoss(avg_non_ignore=False) - (loss_yns): GaussianFocalLoss() - ) - (layers): ModuleList( - (0): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (1): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (4-5): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (6): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (7): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (8): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (11-12): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (13): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (14): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (15): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (16): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (17): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (18-19): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (20): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (21): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (22): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (23): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (24): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (25-26): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (27): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (28): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (29): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (30): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (31): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (32-33): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (34): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (35): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (36): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (37): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (38): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - ) - (fc_before): Linear(in_features=256, out_features=512, bias=False) - (fc_after): Linear(in_features=512, out_features=256, bias=False) - ) - (depth_branch): DenseDepthNet( - (depth_layers): ModuleList( - (0-2): 3 x Conv2d(256, 1, kernel_size=(1, 1), stride=(1, 1)) - ) - ) - (grid_mask): GridMask() -) -2025-11-17 14:18:00,903 - mmdet - INFO - Start running, host: root@VM-120-96-tencentos, work_dir: /cfs/auto/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704 -2025-11-17 14:18:00,904 - mmdet - INFO - Hooks will be executed in the following order: -before_run: -(VERY_HIGH ) CosineAnnealingLrUpdaterHook -(ABOVE_NORMAL) Fp16OptimizerHook -(NORMAL ) CheckpointHook -(NORMAL ) CustomDistEvalHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -before_train_epoch: -(VERY_HIGH ) CosineAnnealingLrUpdaterHook -(NORMAL ) CustomDistEvalHook -(LOW ) IterTimerHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -before_train_iter: -(VERY_HIGH ) CosineAnnealingLrUpdaterHook -(NORMAL ) CustomDistEvalHook -(LOW ) IterTimerHook - -------------------- -after_train_iter: -(ABOVE_NORMAL) Fp16OptimizerHook -(NORMAL ) CheckpointHook -(NORMAL ) CustomDistEvalHook -(LOW ) IterTimerHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -after_train_epoch: -(NORMAL ) CheckpointHook -(NORMAL ) CustomDistEvalHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -before_val_epoch: -(LOW ) IterTimerHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -before_val_iter: -(LOW ) IterTimerHook - -------------------- -after_val_iter: -(LOW ) IterTimerHook - -------------------- -after_val_epoch: -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -after_run: -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -2025-11-17 14:18:00,904 - mmdet - INFO - workflow: [('train', 1)], max: 17500 iters -2025-11-17 14:18:00,906 - mmdet - INFO - Checkpoints will be saved to /cfs/auto/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704 by HardDiskBackend. -2025-11-17 14:19:58,455 - mmdet - INFO - Iter [1/17500] lr: 1.000e-04, eta: 23 days, 14:41:35, time: 116.584, data_time: 10.839, memory: 49163, loss_cls_0: 2.3613, loss_box_0: 0.0138, loss_cns_0: 0.0027, loss_yns_0: 0.0008, loss_cls_1: 2.1544, loss_box_1: 0.1071, loss_cns_1: 0.0243, loss_yns_1: 0.0066, loss_cls_2: 2.3119, loss_box_2: 0.0050, loss_cns_2: 0.0006, loss_yns_2: 0.0003, loss_cls_3: 2.3899, loss_box_3: 0.0295, loss_cns_3: 0.0050, loss_yns_3: 0.0014, loss_cls_4: 2.0284, loss_box_4: 0.4122, loss_cns_4: 0.0530, loss_yns_4: 0.0252, loss_cls_5: 2.4250, loss_box_5: 0.0166, loss_cns_5: 0.0020, loss_yns_5: 0.0016, loss_cls_dn_0: 1.1980, loss_box_dn_0: 1.4603, loss_cls_dn_1: 1.1102, loss_box_dn_1: 1.7318, loss_cls_dn_2: 1.1741, loss_box_dn_2: 1.9719, loss_cls_dn_3: 1.1721, loss_box_dn_3: 2.2418, loss_cls_dn_4: 1.0528, loss_box_dn_4: 2.4268, loss_cls_dn_5: 1.2387, loss_box_dn_5: 2.6773, loss_dense_depth: 1.8643, loss: 35.6988, grad_norm: 268.2582 -2025-11-17 14:20:00,407 - mmdet - INFO - Iter [2/17500] lr: 1.004e-04, eta: 12 days, 0:05:54, time: 1.962, data_time: 0.085, memory: 49163, loss_cls_0: 2.0105, loss_box_0: 0.0132, loss_cns_0: 0.0032, loss_yns_0: 0.0011, loss_cls_1: 2.0082, loss_box_1: 0.1947, loss_cns_1: 0.0350, loss_yns_1: 0.0105, loss_cls_2: 2.0980, loss_box_2: 0.2609, loss_cns_2: 0.0249, loss_yns_2: 0.0111, loss_cls_3: 1.9533, loss_box_3: 0.3533, loss_cns_3: 0.0461, loss_yns_3: 0.0165, loss_cls_4: 1.7938, loss_box_4: 1.6376, loss_cns_4: 0.1655, loss_yns_4: 0.0573, loss_cls_5: 2.0562, loss_box_5: 0.5497, loss_cns_5: 0.0615, loss_yns_5: 0.0190, loss_cls_dn_0: 1.0101, loss_box_dn_0: 1.2554, loss_cls_dn_1: 0.9529, loss_box_dn_1: 2.4087, loss_cls_dn_2: 0.9686, loss_box_dn_2: 2.5309, loss_cls_dn_3: 0.9134, loss_box_dn_3: 2.6172, loss_cls_dn_4: 0.8396, loss_box_dn_4: 2.8875, loss_cls_dn_5: 0.9860, loss_box_dn_5: 3.1244, loss_dense_depth: 1.7113, loss: 37.5871, grad_norm: 66.5134 -2025-11-17 14:20:01,918 - mmdet - INFO - Iter [3/17500] lr: 1.008e-04, eta: 8 days, 2:30:07, time: 1.511, data_time: 0.081, memory: 49163, loss_cls_0: 1.3908, loss_box_0: 2.5596, loss_cns_0: 0.6161, loss_yns_0: 0.2168, loss_cls_1: 1.7263, loss_box_1: 2.3780, loss_cns_1: 0.3383, loss_yns_1: 0.1275, loss_cls_2: 1.7476, loss_box_2: 4.5396, loss_cns_2: 0.3706, loss_yns_2: 0.2069, loss_cls_3: 1.5965, loss_box_3: 5.3351, loss_cns_3: 0.4611, loss_yns_3: 0.2203, loss_cls_4: 1.5098, loss_box_4: 5.2848, loss_cns_4: 0.4358, loss_yns_4: 0.2093, loss_cls_5: 1.6223, loss_box_5: 4.2421, loss_cns_5: 0.2987, loss_yns_5: 0.1318, loss_cls_dn_0: 0.6478, loss_box_dn_0: 1.1885, loss_cls_dn_1: 0.8065, loss_box_dn_1: 2.3836, loss_cls_dn_2: 0.7665, loss_box_dn_2: 2.6115, loss_cls_dn_3: 0.6676, loss_box_dn_3: 2.8361, loss_cls_dn_4: 0.6823, loss_box_dn_4: 3.0838, loss_cls_dn_5: 0.7731, loss_box_dn_5: 3.3794, loss_dense_depth: 1.5865, loss: 58.9789, grad_norm: 116.2252 -2025-11-17 14:20:03,460 - mmdet - INFO - Iter [4/17500] lr: 1.012e-04, eta: 6 days, 3:44:28, time: 1.542, data_time: 0.085, memory: 49163, loss_cls_0: 1.4045, loss_box_0: 2.6805, loss_cns_0: 0.5211, loss_yns_0: 0.2467, loss_cls_1: 1.5605, loss_box_1: 3.6639, loss_cns_1: 0.4533, loss_yns_1: 0.2086, loss_cls_2: 1.7181, loss_box_2: 3.8058, loss_cns_2: 0.4427, loss_yns_2: 0.2100, loss_cls_3: 1.4650, loss_box_3: 4.3084, loss_cns_3: 0.4651, loss_yns_3: 0.2071, loss_cls_4: 1.4556, loss_box_4: 4.8089, loss_cns_4: 0.4200, loss_yns_4: 0.1936, loss_cls_5: 1.4064, loss_box_5: 5.3628, loss_cns_5: 0.4467, loss_yns_5: 0.2039, loss_cls_dn_0: 0.5189, loss_box_dn_0: 1.2625, loss_cls_dn_1: 0.6715, loss_box_dn_1: 2.6585, loss_cls_dn_2: 0.6441, loss_box_dn_2: 2.7647, loss_cls_dn_3: 0.5841, loss_box_dn_3: 3.0378, loss_cls_dn_4: 0.5511, loss_box_dn_4: 3.3054, loss_cls_dn_5: 0.6018, loss_box_dn_5: 3.5784, loss_dense_depth: 1.6279, loss: 59.4659, grad_norm: 120.1968 -2025-11-17 14:20:04,981 - mmdet - INFO - Iter [5/17500] lr: 1.016e-04, eta: 4 days, 23:39:58, time: 1.523, data_time: 0.086, memory: 49163, loss_cls_0: 1.3494, loss_box_0: 2.9085, loss_cns_0: 0.5072, loss_yns_0: 0.1987, loss_cls_1: 1.5452, loss_box_1: 4.2878, loss_cns_1: 0.4204, loss_yns_1: 0.2140, loss_cls_2: 1.4945, loss_box_2: 4.2617, loss_cns_2: 0.4239, loss_yns_2: 0.1980, loss_cls_3: 1.3788, loss_box_3: 4.4110, loss_cns_3: 0.3968, loss_yns_3: 0.2132, loss_cls_4: 1.3357, loss_box_4: 4.7240, loss_cns_4: 0.3724, loss_yns_4: 0.2049, loss_cls_5: 1.3545, loss_box_5: 4.9667, loss_cns_5: 0.3736, loss_yns_5: 0.2197, loss_cls_dn_0: 0.5687, loss_box_dn_0: 1.3163, loss_cls_dn_1: 0.6081, loss_box_dn_1: 2.5533, loss_cls_dn_2: 0.6368, loss_box_dn_2: 2.6859, loss_cls_dn_3: 0.5347, loss_box_dn_3: 2.7933, loss_cls_dn_4: 0.5410, loss_box_dn_4: 3.0513, loss_cls_dn_5: 0.5202, loss_box_dn_5: 3.1206, loss_dense_depth: 1.6734, loss: 58.3645, grad_norm: 128.4735 -2025-11-17 14:20:06,492 - mmdet - INFO - Iter [6/17500] lr: 1.020e-04, eta: 4 days, 4:56:20, time: 1.510, data_time: 0.083, memory: 49163, loss_cls_0: 1.2577, loss_box_0: 2.5537, loss_cns_0: 0.6168, loss_yns_0: 0.1854, loss_cls_1: 1.3715, loss_box_1: 3.8356, loss_cns_1: 0.4517, loss_yns_1: 0.1941, loss_cls_2: 1.3644, loss_box_2: 3.9268, loss_cns_2: 0.4480, loss_yns_2: 0.2097, loss_cls_3: 1.3430, loss_box_3: 3.8915, loss_cns_3: 0.4694, loss_yns_3: 0.1846, loss_cls_4: 1.3325, loss_box_4: 4.2588, loss_cns_4: 0.4057, loss_yns_4: 0.1890, loss_cls_5: 1.3641, loss_box_5: 4.5955, loss_cns_5: 0.3609, loss_yns_5: 0.2004, loss_cls_dn_0: 0.5607, loss_box_dn_0: 1.1894, loss_cls_dn_1: 0.5489, loss_box_dn_1: 2.6294, loss_cls_dn_2: 0.5703, loss_box_dn_2: 2.6519, loss_cls_dn_3: 0.4858, loss_box_dn_3: 2.7001, loss_cls_dn_4: 0.4721, loss_box_dn_4: 2.9607, loss_cls_dn_5: 0.4548, loss_box_dn_5: 3.1447, loss_dense_depth: 1.8552, loss: 55.2348, grad_norm: 124.7906 -2025-11-17 14:20:08,000 - mmdet - INFO - Iter [7/17500] lr: 1.024e-04, eta: 3 days, 15:33:39, time: 1.508, data_time: 0.081, memory: 49163, loss_cls_0: 1.2565, loss_box_0: 2.3369, loss_cns_0: 0.6502, loss_yns_0: 0.2017, loss_cls_1: 1.2864, loss_box_1: 3.6737, loss_cns_1: 0.5018, loss_yns_1: 0.1915, loss_cls_2: 1.3655, loss_box_2: 3.8502, loss_cns_2: 0.4676, loss_yns_2: 0.1935, loss_cls_3: 1.2915, loss_box_3: 3.8636, loss_cns_3: 0.4648, loss_yns_3: 0.1835, loss_cls_4: 1.3696, loss_box_4: 4.1796, loss_cns_4: 0.4460, loss_yns_4: 0.2256, loss_cls_5: 1.3489, loss_box_5: 4.4405, loss_cns_5: 0.4892, loss_yns_5: 0.1861, loss_cls_dn_0: 0.5200, loss_box_dn_0: 1.0891, loss_cls_dn_1: 0.4730, loss_box_dn_1: 2.4940, loss_cls_dn_2: 0.4865, loss_box_dn_2: 2.4961, loss_cls_dn_3: 0.4378, loss_box_dn_3: 2.5308, loss_cls_dn_4: 0.4076, loss_box_dn_4: 2.6838, loss_cls_dn_5: 0.3945, loss_box_dn_5: 2.7976, loss_dense_depth: 2.0104, loss: 53.2856, grad_norm: 119.4658 -2025-11-17 14:20:09,497 - mmdet - INFO - Iter [8/17500] lr: 1.028e-04, eta: 3 days, 5:31:14, time: 1.497, data_time: 0.086, memory: 49163, loss_cls_0: 1.2711, loss_box_0: 2.2341, loss_cns_0: 0.6183, loss_yns_0: 0.1741, loss_cls_1: 1.3011, loss_box_1: 3.6105, loss_cns_1: 0.5101, loss_yns_1: 0.1915, loss_cls_2: 1.3378, loss_box_2: 3.7665, loss_cns_2: 0.3934, loss_yns_2: 0.1898, loss_cls_3: 1.2908, loss_box_3: 3.8936, loss_cns_3: 0.3777, loss_yns_3: 0.1871, loss_cls_4: 1.2638, loss_box_4: 3.8291, loss_cns_4: 0.3730, loss_yns_4: 0.1997, loss_cls_5: 1.3408, loss_box_5: 3.8422, loss_cns_5: 0.3820, loss_yns_5: 0.2062, loss_cls_dn_0: 0.4824, loss_box_dn_0: 1.0396, loss_cls_dn_1: 0.4736, loss_box_dn_1: 1.6613, loss_cls_dn_2: 0.4925, loss_box_dn_2: 1.6855, loss_cls_dn_3: 0.4369, loss_box_dn_3: 1.8499, loss_cls_dn_4: 0.4647, loss_box_dn_4: 1.7980, loss_cls_dn_5: 0.4267, loss_box_dn_5: 1.8519, loss_dense_depth: 2.2805, loss: 47.7278, grad_norm: 105.5086 -2025-11-17 14:20:10,998 - mmdet - INFO - Iter [9/17500] lr: 1.032e-04, eta: 2 days, 21:42:49, time: 1.501, data_time: 0.096, memory: 49163, loss_cls_0: 1.2316, loss_box_0: 2.1793, loss_cns_0: 0.6285, loss_yns_0: 0.1858, loss_cls_1: 1.2885, loss_box_1: 3.0344, loss_cns_1: 0.5242, loss_yns_1: 0.1846, loss_cls_2: 1.2915, loss_box_2: 3.1646, loss_cns_2: 0.5081, loss_yns_2: 0.1999, loss_cls_3: 1.2697, loss_box_3: 3.2726, loss_cns_3: 0.5322, loss_yns_3: 0.2018, loss_cls_4: 1.2493, loss_box_4: 3.3253, loss_cns_4: 0.5076, loss_yns_4: 0.1865, loss_cls_5: 1.2857, loss_box_5: 3.4569, loss_cns_5: 0.5536, loss_yns_5: 0.1899, loss_cls_dn_0: 0.4869, loss_box_dn_0: 1.0411, loss_cls_dn_1: 0.4348, loss_box_dn_1: 1.5091, loss_cls_dn_2: 0.4822, loss_box_dn_2: 1.6898, loss_cls_dn_3: 0.4336, loss_box_dn_3: 1.9019, loss_cls_dn_4: 0.4689, loss_box_dn_4: 1.9699, loss_cls_dn_5: 0.4486, loss_box_dn_5: 2.1127, loss_dense_depth: 2.1887, loss: 45.6203, grad_norm: 84.3352 -2025-11-17 14:20:12,502 - mmdet - INFO - Iter [10/17500] lr: 1.036e-04, eta: 2 days, 15:28:11, time: 1.505, data_time: 0.086, memory: 49163, loss_cls_0: 1.2561, loss_box_0: 2.2066, loss_cns_0: 0.6261, loss_yns_0: 0.1802, loss_cls_1: 1.2838, loss_box_1: 3.0329, loss_cns_1: 0.4987, loss_yns_1: 0.1793, loss_cls_2: 1.2531, loss_box_2: 3.1547, loss_cns_2: 0.4782, loss_yns_2: 0.2035, loss_cls_3: 1.2392, loss_box_3: 3.3491, loss_cns_3: 0.4614, loss_yns_3: 0.1890, loss_cls_4: 1.2657, loss_box_4: 3.6408, loss_cns_4: 0.3894, loss_yns_4: 0.2073, loss_cls_5: 1.2528, loss_box_5: 3.9569, loss_cns_5: 0.3348, loss_yns_5: 0.1907, loss_cls_dn_0: 0.4721, loss_box_dn_0: 1.0354, loss_cls_dn_1: 0.4106, loss_box_dn_1: 1.7132, loss_cls_dn_2: 0.4387, loss_box_dn_2: 1.8566, loss_cls_dn_3: 0.4429, loss_box_dn_3: 1.9555, loss_cls_dn_4: 0.4110, loss_box_dn_4: 2.0810, loss_cls_dn_5: 0.4361, loss_box_dn_5: 2.2700, loss_dense_depth: 2.3033, loss: 46.6567, grad_norm: 104.0207 -2025-11-17 14:20:14,008 - mmdet - INFO - Iter [11/17500] lr: 1.040e-04, eta: 2 days, 10:21:40, time: 1.505, data_time: 0.083, memory: 49163, loss_cls_0: 1.2182, loss_box_0: 2.2782, loss_cns_0: 0.6162, loss_yns_0: 0.1709, loss_cls_1: 1.2500, loss_box_1: 3.0960, loss_cns_1: 0.5115, loss_yns_1: 0.1773, loss_cls_2: 1.2586, loss_box_2: 3.1874, loss_cns_2: 0.4804, loss_yns_2: 0.1984, loss_cls_3: 1.2858, loss_box_3: 3.3515, loss_cns_3: 0.4797, loss_yns_3: 0.1818, loss_cls_4: 1.2739, loss_box_4: 3.5440, loss_cns_4: 0.4542, loss_yns_4: 0.1896, loss_cls_5: 1.2482, loss_box_5: 3.8486, loss_cns_5: 0.4189, loss_yns_5: 0.1784, loss_cls_dn_0: 0.4616, loss_box_dn_0: 1.0305, loss_cls_dn_1: 0.4094, loss_box_dn_1: 1.8143, loss_cls_dn_2: 0.4081, loss_box_dn_2: 1.8781, loss_cls_dn_3: 0.4255, loss_box_dn_3: 1.8947, loss_cls_dn_4: 0.3887, loss_box_dn_4: 1.9996, loss_cls_dn_5: 0.4192, loss_box_dn_5: 2.1169, loss_dense_depth: 2.2229, loss: 46.3673, grad_norm: 105.3113 -2025-11-17 14:20:15,497 - mmdet - INFO - Iter [12/17500] lr: 1.044e-04, eta: 2 days, 6:05:53, time: 1.490, data_time: 0.083, memory: 49163, loss_cls_0: 1.2381, loss_box_0: 2.2675, loss_cns_0: 0.6144, loss_yns_0: 0.1746, loss_cls_1: 1.2261, loss_box_1: 3.2087, loss_cns_1: 0.5211, loss_yns_1: 0.1846, loss_cls_2: 1.2179, loss_box_2: 3.3033, loss_cns_2: 0.4733, loss_yns_2: 0.1913, loss_cls_3: 1.2691, loss_box_3: 3.3461, loss_cns_3: 0.4666, loss_yns_3: 0.1803, loss_cls_4: 1.2158, loss_box_4: 3.3948, loss_cns_4: 0.4730, loss_yns_4: 0.1837, loss_cls_5: 1.2432, loss_box_5: 3.5584, loss_cns_5: 0.4838, loss_yns_5: 0.1826, loss_cls_dn_0: 0.4584, loss_box_dn_0: 1.0348, loss_cls_dn_1: 0.4066, loss_box_dn_1: 1.9311, loss_cls_dn_2: 0.3985, loss_box_dn_2: 1.9494, loss_cls_dn_3: 0.3913, loss_box_dn_3: 1.9689, loss_cls_dn_4: 0.3751, loss_box_dn_4: 2.0891, loss_cls_dn_5: 0.3962, loss_box_dn_5: 2.2333, loss_dense_depth: 2.2580, loss: 46.5093, grad_norm: 71.7435 -2025-11-17 14:20:17,021 - mmdet - INFO - Iter [13/17500] lr: 1.048e-04, eta: 2 days, 2:30:10, time: 1.523, data_time: 0.085, memory: 49163, loss_cls_0: 1.1627, loss_box_0: 2.1916, loss_cns_0: 0.6140, loss_yns_0: 0.1710, loss_cls_1: 1.2166, loss_box_1: 2.9106, loss_cns_1: 0.5258, loss_yns_1: 0.1894, loss_cls_2: 1.2224, loss_box_2: 3.1524, loss_cns_2: 0.5182, loss_yns_2: 0.1777, loss_cls_3: 1.2910, loss_box_3: 3.1645, loss_cns_3: 0.4888, loss_yns_3: 0.1739, loss_cls_4: 1.2203, loss_box_4: 3.1445, loss_cns_4: 0.5017, loss_yns_4: 0.1856, loss_cls_5: 1.2670, loss_box_5: 3.3212, loss_cns_5: 0.4723, loss_yns_5: 0.1776, loss_cls_dn_0: 0.4930, loss_box_dn_0: 1.0416, loss_cls_dn_1: 0.4215, loss_box_dn_1: 1.5977, loss_cls_dn_2: 0.4347, loss_box_dn_2: 1.7031, loss_cls_dn_3: 0.4099, loss_box_dn_3: 1.8085, loss_cls_dn_4: 0.4170, loss_box_dn_4: 1.9613, loss_cls_dn_5: 0.4347, loss_box_dn_5: 2.1285, loss_dense_depth: 2.1866, loss: 44.4989, grad_norm: 76.3482 -2025-11-17 14:20:18,518 - mmdet - INFO - Iter [14/17500] lr: 1.052e-04, eta: 1 day, 23:24:45, time: 1.498, data_time: 0.083, memory: 49163, loss_cls_0: 1.1798, loss_box_0: 2.2392, loss_cns_0: 0.6144, loss_yns_0: 0.1716, loss_cls_1: 1.2858, loss_box_1: 2.6858, loss_cns_1: 0.6216, loss_yns_1: 0.1749, loss_cls_2: 1.2733, loss_box_2: 2.9058, loss_cns_2: 0.5726, loss_yns_2: 0.1823, loss_cls_3: 1.2986, loss_box_3: 2.9565, loss_cns_3: 0.5670, loss_yns_3: 0.1738, loss_cls_4: 1.2342, loss_box_4: 2.9558, loss_cns_4: 0.5825, loss_yns_4: 0.1841, loss_cls_5: 1.2525, loss_box_5: 3.2021, loss_cns_5: 0.5463, loss_yns_5: 0.1827, loss_cls_dn_0: 0.4856, loss_box_dn_0: 1.0461, loss_cls_dn_1: 0.4313, loss_box_dn_1: 1.4800, loss_cls_dn_2: 0.4458, loss_box_dn_2: 1.5864, loss_cls_dn_3: 0.4295, loss_box_dn_3: 1.7061, loss_cls_dn_4: 0.4493, loss_box_dn_4: 1.8211, loss_cls_dn_5: 0.4535, loss_box_dn_5: 1.9722, loss_dense_depth: 2.4326, loss: 43.7828, grad_norm: 95.3719 -2025-11-17 14:20:20,008 - mmdet - INFO - Iter [15/17500] lr: 1.056e-04, eta: 1 day, 20:43:52, time: 1.489, data_time: 0.084, memory: 49163, loss_cls_0: 1.2343, loss_box_0: 2.3121, loss_cns_0: 0.6007, loss_yns_0: 0.1712, loss_cls_1: 1.2411, loss_box_1: 2.7626, loss_cns_1: 0.5822, loss_yns_1: 0.1835, loss_cls_2: 1.2757, loss_box_2: 2.8265, loss_cns_2: 0.5578, loss_yns_2: 0.1795, loss_cls_3: 1.2540, loss_box_3: 2.9063, loss_cns_3: 0.5704, loss_yns_3: 0.1736, loss_cls_4: 1.2412, loss_box_4: 2.9429, loss_cns_4: 0.5529, loss_yns_4: 0.1821, loss_cls_5: 1.2673, loss_box_5: 3.1532, loss_cns_5: 0.5553, loss_yns_5: 0.1770, loss_cls_dn_0: 0.4443, loss_box_dn_0: 1.0474, loss_cls_dn_1: 0.4416, loss_box_dn_1: 1.5878, loss_cls_dn_2: 0.4429, loss_box_dn_2: 1.5892, loss_cls_dn_3: 0.4531, loss_box_dn_3: 1.6826, loss_cls_dn_4: 0.4579, loss_box_dn_4: 1.7170, loss_cls_dn_5: 0.4566, loss_box_dn_5: 1.8295, loss_dense_depth: 2.1862, loss: 43.2397, grad_norm: 76.3296 -2025-11-17 14:20:21,498 - mmdet - INFO - Iter [16/17500] lr: 1.060e-04, eta: 1 day, 18:23:07, time: 1.490, data_time: 0.081, memory: 49163, loss_cls_0: 1.1896, loss_box_0: 2.3379, loss_cns_0: 0.5897, loss_yns_0: 0.1706, loss_cls_1: 1.2175, loss_box_1: 2.8366, loss_cns_1: 0.5333, loss_yns_1: 0.1848, loss_cls_2: 1.2418, loss_box_2: 2.8543, loss_cns_2: 0.5374, loss_yns_2: 0.1866, loss_cls_3: 1.2583, loss_box_3: 2.8301, loss_cns_3: 0.5378, loss_yns_3: 0.1694, loss_cls_4: 1.2434, loss_box_4: 2.8040, loss_cns_4: 0.5489, loss_yns_4: 0.1755, loss_cls_5: 1.2688, loss_box_5: 2.9526, loss_cns_5: 0.5538, loss_yns_5: 0.1833, loss_cls_dn_0: 0.4622, loss_box_dn_0: 1.0383, loss_cls_dn_1: 0.4724, loss_box_dn_1: 1.5750, loss_cls_dn_2: 0.4526, loss_box_dn_2: 1.5089, loss_cls_dn_3: 0.4699, loss_box_dn_3: 1.5745, loss_cls_dn_4: 0.4490, loss_box_dn_4: 1.5449, loss_cls_dn_5: 0.4583, loss_box_dn_5: 1.6456, loss_dense_depth: 2.2377, loss: 42.2957, grad_norm: 61.7152 -2025-11-17 14:20:22,988 - mmdet - INFO - Iter [17/17500] lr: 1.064e-04, eta: 1 day, 16:18:56, time: 1.490, data_time: 0.082, memory: 49163, loss_cls_0: 1.1582, loss_box_0: 2.3217, loss_cns_0: 0.5924, loss_yns_0: 0.1700, loss_cls_1: 1.2139, loss_box_1: 3.0122, loss_cns_1: 0.5074, loss_yns_1: 0.1802, loss_cls_2: 1.2311, loss_box_2: 2.9399, loss_cns_2: 0.5289, loss_yns_2: 0.1871, loss_cls_3: 1.2488, loss_box_3: 2.8863, loss_cns_3: 0.5404, loss_yns_3: 0.1769, loss_cls_4: 1.2453, loss_box_4: 2.9150, loss_cns_4: 0.5395, loss_yns_4: 0.1767, loss_cls_5: 1.2656, loss_box_5: 2.9842, loss_cns_5: 0.5317, loss_yns_5: 0.1811, loss_cls_dn_0: 0.4811, loss_box_dn_0: 1.0517, loss_cls_dn_1: 0.4462, loss_box_dn_1: 1.7382, loss_cls_dn_2: 0.4374, loss_box_dn_2: 1.6769, loss_cls_dn_3: 0.4368, loss_box_dn_3: 1.7511, loss_cls_dn_4: 0.4197, loss_box_dn_4: 1.7575, loss_cls_dn_5: 0.4301, loss_box_dn_5: 1.8383, loss_dense_depth: 2.0764, loss: 43.2756, grad_norm: 51.3784 -2025-11-17 14:20:24,493 - mmdet - INFO - Iter [18/17500] lr: 1.068e-04, eta: 1 day, 14:28:45, time: 1.504, data_time: 0.086, memory: 49163, loss_cls_0: 1.1641, loss_box_0: 2.2745, loss_cns_0: 0.5948, loss_yns_0: 0.1685, loss_cls_1: 1.2297, loss_box_1: 2.8693, loss_cns_1: 0.5400, loss_yns_1: 0.1789, loss_cls_2: 1.2328, loss_box_2: 2.9507, loss_cns_2: 0.5244, loss_yns_2: 0.1843, loss_cls_3: 1.2641, loss_box_3: 3.0095, loss_cns_3: 0.5420, loss_yns_3: 0.1744, loss_cls_4: 1.2356, loss_box_4: 3.1344, loss_cns_4: 0.4966, loss_yns_4: 0.1819, loss_cls_5: 1.2401, loss_box_5: 3.1891, loss_cns_5: 0.4765, loss_yns_5: 0.1929, loss_cls_dn_0: 0.4730, loss_box_dn_0: 1.0362, loss_cls_dn_1: 0.4167, loss_box_dn_1: 1.6770, loss_cls_dn_2: 0.4226, loss_box_dn_2: 1.7337, loss_cls_dn_3: 0.4152, loss_box_dn_3: 1.8504, loss_cls_dn_4: 0.4139, loss_box_dn_4: 1.9221, loss_cls_dn_5: 0.4325, loss_box_dn_5: 2.0173, loss_dense_depth: 2.0652, loss: 43.9249, grad_norm: 79.8852 -2025-11-17 14:20:25,984 - mmdet - INFO - Iter [19/17500] lr: 1.072e-04, eta: 1 day, 12:50:00, time: 1.492, data_time: 0.085, memory: 49163, loss_cls_0: 1.1721, loss_box_0: 2.2920, loss_cns_0: 0.5879, loss_yns_0: 0.1680, loss_cls_1: 1.2215, loss_box_1: 2.8264, loss_cns_1: 0.5463, loss_yns_1: 0.1824, loss_cls_2: 1.2187, loss_box_2: 2.9184, loss_cns_2: 0.5324, loss_yns_2: 0.1802, loss_cls_3: 1.2675, loss_box_3: 2.9850, loss_cns_3: 0.5333, loss_yns_3: 0.1718, loss_cls_4: 1.2195, loss_box_4: 3.0956, loss_cns_4: 0.5029, loss_yns_4: 0.1864, loss_cls_5: 1.2291, loss_box_5: 3.1574, loss_cns_5: 0.5047, loss_yns_5: 0.1753, loss_cls_dn_0: 0.4767, loss_box_dn_0: 1.0405, loss_cls_dn_1: 0.4260, loss_box_dn_1: 1.3563, loss_cls_dn_2: 0.4399, loss_box_dn_2: 1.5270, loss_cls_dn_3: 0.4253, loss_box_dn_3: 1.6413, loss_cls_dn_4: 0.4217, loss_box_dn_4: 1.7846, loss_cls_dn_5: 0.4417, loss_box_dn_5: 1.9031, loss_dense_depth: 1.9217, loss: 42.6806, grad_norm: 90.3513 -2025-11-17 14:20:27,469 - mmdet - INFO - Iter [20/17500] lr: 1.076e-04, eta: 1 day, 11:21:00, time: 1.484, data_time: 0.078, memory: 49163, loss_cls_0: 1.1689, loss_box_0: 2.3032, loss_cns_0: 0.5849, loss_yns_0: 0.1689, loss_cls_1: 1.2297, loss_box_1: 2.7315, loss_cns_1: 0.5635, loss_yns_1: 0.1790, loss_cls_2: 1.2284, loss_box_2: 2.7397, loss_cns_2: 0.5668, loss_yns_2: 0.1776, loss_cls_3: 1.2682, loss_box_3: 2.7408, loss_cns_3: 0.5626, loss_yns_3: 0.1726, loss_cls_4: 1.2683, loss_box_4: 2.7683, loss_cns_4: 0.5868, loss_yns_4: 0.1808, loss_cls_5: 1.2883, loss_box_5: 2.8145, loss_cns_5: 0.6007, loss_yns_5: 0.1858, loss_cls_dn_0: 0.4629, loss_box_dn_0: 1.0239, loss_cls_dn_1: 0.4032, loss_box_dn_1: 1.4585, loss_cls_dn_2: 0.4161, loss_box_dn_2: 1.5146, loss_cls_dn_3: 0.3995, loss_box_dn_3: 1.5422, loss_cls_dn_4: 0.3850, loss_box_dn_4: 1.6045, loss_cls_dn_5: 0.3921, loss_box_dn_5: 1.6514, loss_dense_depth: 1.6732, loss: 41.0067, grad_norm: 52.5082 -2025-11-17 14:20:29,034 - mmdet - INFO - Iter [21/17500] lr: 1.080e-04, eta: 1 day, 10:01:35, time: 1.565, data_time: 0.138, memory: 49163, loss_cls_0: 1.1458, loss_box_0: 2.2657, loss_cns_0: 0.5942, loss_yns_0: 0.1675, loss_cls_1: 1.2294, loss_box_1: 2.7546, loss_cns_1: 0.5720, loss_yns_1: 0.1736, loss_cls_2: 1.2271, loss_box_2: 2.7729, loss_cns_2: 0.5496, loss_yns_2: 0.1783, loss_cls_3: 1.2424, loss_box_3: 2.8081, loss_cns_3: 0.5449, loss_yns_3: 0.1701, loss_cls_4: 1.2556, loss_box_4: 2.8652, loss_cns_4: 0.5333, loss_yns_4: 0.1797, loss_cls_5: 1.2698, loss_box_5: 2.9841, loss_cns_5: 0.4774, loss_yns_5: 0.1828, loss_cls_dn_0: 0.4633, loss_box_dn_0: 1.0330, loss_cls_dn_1: 0.4177, loss_box_dn_1: 1.2503, loss_cls_dn_2: 0.4337, loss_box_dn_2: 1.2617, loss_cls_dn_3: 0.4352, loss_box_dn_3: 1.2847, loss_cls_dn_4: 0.4237, loss_box_dn_4: 1.3800, loss_cls_dn_5: 0.4221, loss_box_dn_5: 1.3928, loss_dense_depth: 1.5476, loss: 39.8899, grad_norm: 82.4642 -2025-11-17 14:20:30,678 - mmdet - INFO - Iter [22/17500] lr: 1.084e-04, eta: 1 day, 8:50:28, time: 1.645, data_time: 0.118, memory: 49163, loss_cls_0: 1.1588, loss_box_0: 2.2314, loss_cns_0: 0.6023, loss_yns_0: 0.1675, loss_cls_1: 1.2458, loss_box_1: 2.8057, loss_cns_1: 0.5604, loss_yns_1: 0.1774, loss_cls_2: 1.2242, loss_box_2: 2.8530, loss_cns_2: 0.5363, loss_yns_2: 0.1877, loss_cls_3: 1.2429, loss_box_3: 2.9284, loss_cns_3: 0.5416, loss_yns_3: 0.1721, loss_cls_4: 1.2287, loss_box_4: 2.9732, loss_cns_4: 0.5228, loss_yns_4: 0.1802, loss_cls_5: 1.2564, loss_box_5: 3.0323, loss_cns_5: 0.4910, loss_yns_5: 0.1762, loss_cls_dn_0: 0.4565, loss_box_dn_0: 1.0272, loss_cls_dn_1: 0.4280, loss_box_dn_1: 1.2617, loss_cls_dn_2: 0.4462, loss_box_dn_2: 1.3297, loss_cls_dn_3: 0.4616, loss_box_dn_3: 1.4136, loss_cls_dn_4: 0.4685, loss_box_dn_4: 1.5581, loss_cls_dn_5: 0.4552, loss_box_dn_5: 1.5652, loss_dense_depth: 1.4744, loss: 40.8422, grad_norm: 95.1494 -2025-11-17 14:20:32,203 - mmdet - INFO - Iter [23/17500] lr: 1.088e-04, eta: 1 day, 7:43:59, time: 1.524, data_time: 0.080, memory: 49163, loss_cls_0: 1.1415, loss_box_0: 2.1790, loss_cns_0: 0.6094, loss_yns_0: 0.1695, loss_cls_1: 1.2357, loss_box_1: 2.7908, loss_cns_1: 0.5602, loss_yns_1: 0.1749, loss_cls_2: 1.2252, loss_box_2: 2.8175, loss_cns_2: 0.5543, loss_yns_2: 0.1889, loss_cls_3: 1.2539, loss_box_3: 2.9431, loss_cns_3: 0.5655, loss_yns_3: 0.1731, loss_cls_4: 1.2361, loss_box_4: 2.9896, loss_cns_4: 0.5548, loss_yns_4: 0.1723, loss_cls_5: 1.2579, loss_box_5: 2.9306, loss_cns_5: 0.5544, loss_yns_5: 0.1775, loss_cls_dn_0: 0.4513, loss_box_dn_0: 1.0006, loss_cls_dn_1: 0.4308, loss_box_dn_1: 1.3582, loss_cls_dn_2: 0.4494, loss_box_dn_2: 1.4112, loss_cls_dn_3: 0.4674, loss_box_dn_3: 1.5196, loss_cls_dn_4: 0.4773, loss_box_dn_4: 1.6488, loss_cls_dn_5: 0.4656, loss_box_dn_5: 1.6542, loss_dense_depth: 1.4966, loss: 41.2866, grad_norm: 90.6274 -2025-11-17 14:20:33,739 - mmdet - INFO - Iter [24/17500] lr: 1.092e-04, eta: 1 day, 6:43:10, time: 1.535, data_time: 0.084, memory: 49163, loss_cls_0: 1.1636, loss_box_0: 2.2063, loss_cns_0: 0.6099, loss_yns_0: 0.1693, loss_cls_1: 1.2471, loss_box_1: 2.8118, loss_cns_1: 0.5543, loss_yns_1: 0.1735, loss_cls_2: 1.2492, loss_box_2: 2.8069, loss_cns_2: 0.5525, loss_yns_2: 0.1766, loss_cls_3: 1.2698, loss_box_3: 2.9442, loss_cns_3: 0.5416, loss_yns_3: 0.1731, loss_cls_4: 1.2492, loss_box_4: 2.9645, loss_cns_4: 0.5389, loss_yns_4: 0.1793, loss_cls_5: 1.2734, loss_box_5: 2.8920, loss_cns_5: 0.5470, loss_yns_5: 0.1766, loss_cls_dn_0: 0.4681, loss_box_dn_0: 1.0182, loss_cls_dn_1: 0.4155, loss_box_dn_1: 1.5324, loss_cls_dn_2: 0.4300, loss_box_dn_2: 1.5442, loss_cls_dn_3: 0.4451, loss_box_dn_3: 1.6576, loss_cls_dn_4: 0.4444, loss_box_dn_4: 1.7586, loss_cls_dn_5: 0.4476, loss_box_dn_5: 1.7643, loss_dense_depth: 1.4236, loss: 41.8202, grad_norm: 87.7901 -2025-11-17 14:20:35,242 - mmdet - INFO - Iter [25/17500] lr: 1.096e-04, eta: 1 day, 5:46:53, time: 1.505, data_time: 0.088, memory: 49163, loss_cls_0: 1.1659, loss_box_0: 2.2248, loss_cns_0: 0.6091, loss_yns_0: 0.1682, loss_cls_1: 1.2466, loss_box_1: 2.7345, loss_cns_1: 0.5557, loss_yns_1: 0.1734, loss_cls_2: 1.2511, loss_box_2: 2.8376, loss_cns_2: 0.5480, loss_yns_2: 0.1822, loss_cls_3: 1.2595, loss_box_3: 2.9575, loss_cns_3: 0.5340, loss_yns_3: 0.1719, loss_cls_4: 1.2953, loss_box_4: 2.8326, loss_cns_4: 0.5550, loss_yns_4: 0.1829, loss_cls_5: 1.2670, loss_box_5: 2.7942, loss_cns_5: 0.5658, loss_yns_5: 0.1758, loss_cls_dn_0: 0.4699, loss_box_dn_0: 1.0029, loss_cls_dn_1: 0.3978, loss_box_dn_1: 1.5493, loss_cls_dn_2: 0.4073, loss_box_dn_2: 1.5850, loss_cls_dn_3: 0.4210, loss_box_dn_3: 1.6934, loss_cls_dn_4: 0.4019, loss_box_dn_4: 1.7104, loss_cls_dn_5: 0.4320, loss_box_dn_5: 1.7395, loss_dense_depth: 1.3832, loss: 41.4822, grad_norm: 75.9091 -2025-11-17 14:20:36,775 - mmdet - INFO - Iter [26/17500] lr: 1.100e-04, eta: 1 day, 4:55:13, time: 1.532, data_time: 0.089, memory: 49163, loss_cls_0: 1.1492, loss_box_0: 2.2319, loss_cns_0: 0.6083, loss_yns_0: 0.1674, loss_cls_1: 1.2547, loss_box_1: 2.6176, loss_cns_1: 0.5936, loss_yns_1: 0.1741, loss_cls_2: 1.2647, loss_box_2: 2.7425, loss_cns_2: 0.5969, loss_yns_2: 0.1776, loss_cls_3: 1.2595, loss_box_3: 2.8094, loss_cns_3: 0.5925, loss_yns_3: 0.1726, loss_cls_4: 1.2911, loss_box_4: 2.7187, loss_cns_4: 0.6031, loss_yns_4: 0.1733, loss_cls_5: 1.2610, loss_box_5: 2.7332, loss_cns_5: 0.6047, loss_yns_5: 0.1732, loss_cls_dn_0: 0.4483, loss_box_dn_0: 1.0088, loss_cls_dn_1: 0.4127, loss_box_dn_1: 1.2842, loss_cls_dn_2: 0.4175, loss_box_dn_2: 1.3894, loss_cls_dn_3: 0.4359, loss_box_dn_3: 1.4832, loss_cls_dn_4: 0.4203, loss_box_dn_4: 1.4679, loss_cls_dn_5: 0.4468, loss_box_dn_5: 1.5043, loss_dense_depth: 1.3396, loss: 40.0300, grad_norm: 73.3091 -2025-11-17 14:20:38,280 - mmdet - INFO - Iter [27/17500] lr: 1.104e-04, eta: 1 day, 4:07:04, time: 1.504, data_time: 0.097, memory: 49163, loss_cls_0: 1.1540, loss_box_0: 2.2470, loss_cns_0: 0.6039, loss_yns_0: 0.1710, loss_cls_1: 1.2205, loss_box_1: 2.5965, loss_cns_1: 0.5879, loss_yns_1: 0.1725, loss_cls_2: 1.2473, loss_box_2: 2.7375, loss_cns_2: 0.5790, loss_yns_2: 0.1780, loss_cls_3: 1.2518, loss_box_3: 2.7633, loss_cns_3: 0.5779, loss_yns_3: 0.1720, loss_cls_4: 1.2328, loss_box_4: 2.8824, loss_cns_4: 0.5520, loss_yns_4: 0.1870, loss_cls_5: 1.2456, loss_box_5: 3.0056, loss_cns_5: 0.5350, loss_yns_5: 0.1761, loss_cls_dn_0: 0.4377, loss_box_dn_0: 1.0037, loss_cls_dn_1: 0.4169, loss_box_dn_1: 1.1983, loss_cls_dn_2: 0.4098, loss_box_dn_2: 1.2850, loss_cls_dn_3: 0.4222, loss_box_dn_3: 1.3414, loss_cls_dn_4: 0.4438, loss_box_dn_4: 1.3548, loss_cls_dn_5: 0.4262, loss_box_dn_5: 1.4439, loss_dense_depth: 1.3297, loss: 39.5899, grad_norm: 85.9763 -2025-11-17 14:20:39,774 - mmdet - INFO - Iter [28/17500] lr: 1.108e-04, eta: 1 day, 3:22:17, time: 1.496, data_time: 0.081, memory: 49163, loss_cls_0: 1.1334, loss_box_0: 2.2194, loss_cns_0: 0.6076, loss_yns_0: 0.1679, loss_cls_1: 1.1770, loss_box_1: 2.5820, loss_cns_1: 0.5673, loss_yns_1: 0.1697, loss_cls_2: 1.2021, loss_box_2: 2.6324, loss_cns_2: 0.5635, loss_yns_2: 0.1746, loss_cls_3: 1.2502, loss_box_3: 2.6191, loss_cns_3: 0.5657, loss_yns_3: 0.1749, loss_cls_4: 1.2005, loss_box_4: 2.7061, loss_cns_4: 0.5571, loss_yns_4: 0.1869, loss_cls_5: 1.2977, loss_box_5: 2.7454, loss_cns_5: 0.5662, loss_yns_5: 0.1786, loss_cls_dn_0: 0.4306, loss_box_dn_0: 0.9969, loss_cls_dn_1: 0.4218, loss_box_dn_1: 1.2521, loss_cls_dn_2: 0.4096, loss_box_dn_2: 1.3037, loss_cls_dn_3: 0.4021, loss_box_dn_3: 1.3453, loss_cls_dn_4: 0.4564, loss_box_dn_4: 1.3648, loss_cls_dn_5: 0.3950, loss_box_dn_5: 1.4747, loss_dense_depth: 1.2559, loss: 38.7544, grad_norm: 84.4652 -2025-11-17 14:20:41,305 - mmdet - INFO - Iter [29/17500] lr: 1.112e-04, eta: 1 day, 2:40:56, time: 1.531, data_time: 0.098, memory: 49163, loss_cls_0: 1.1004, loss_box_0: 2.1533, loss_cns_0: 0.6119, loss_yns_0: 0.1694, loss_cls_1: 1.1695, loss_box_1: 2.5368, loss_cns_1: 0.5780, loss_yns_1: 0.1712, loss_cls_2: 1.1906, loss_box_2: 2.5085, loss_cns_2: 0.5818, loss_yns_2: 0.1761, loss_cls_3: 1.2479, loss_box_3: 2.5522, loss_cns_3: 0.5836, loss_yns_3: 0.1770, loss_cls_4: 1.1991, loss_box_4: 2.5693, loss_cns_4: 0.5877, loss_yns_4: 0.1817, loss_cls_5: 1.2987, loss_box_5: 2.6045, loss_cns_5: 0.5802, loss_yns_5: 0.1755, loss_cls_dn_0: 0.4364, loss_box_dn_0: 0.9913, loss_cls_dn_1: 0.4235, loss_box_dn_1: 1.1912, loss_cls_dn_2: 0.4182, loss_box_dn_2: 1.2044, loss_cls_dn_3: 0.4071, loss_box_dn_3: 1.2737, loss_cls_dn_4: 0.4650, loss_box_dn_4: 1.2896, loss_cls_dn_5: 0.4043, loss_box_dn_5: 1.4123, loss_dense_depth: 1.2357, loss: 37.8573, grad_norm: 64.1875 -2025-11-17 14:20:42,827 - mmdet - INFO - Iter [30/17500] lr: 1.116e-04, eta: 1 day, 2:02:15, time: 1.521, data_time: 0.082, memory: 49163, loss_cls_0: 1.0846, loss_box_0: 2.0597, loss_cns_0: 0.6161, loss_yns_0: 0.1696, loss_cls_1: 1.1504, loss_box_1: 2.5143, loss_cns_1: 0.5896, loss_yns_1: 0.1741, loss_cls_2: 1.1662, loss_box_2: 2.5340, loss_cns_2: 0.5920, loss_yns_2: 0.1790, loss_cls_3: 1.1997, loss_box_3: 2.5987, loss_cns_3: 0.5874, loss_yns_3: 0.1808, loss_cls_4: 1.1840, loss_box_4: 2.5906, loss_cns_4: 0.5785, loss_yns_4: 0.1776, loss_cls_5: 1.1954, loss_box_5: 2.6291, loss_cns_5: 0.5624, loss_yns_5: 0.1738, loss_cls_dn_0: 0.4360, loss_box_dn_0: 0.9923, loss_cls_dn_1: 0.4021, loss_box_dn_1: 1.2770, loss_cls_dn_2: 0.4136, loss_box_dn_2: 1.2941, loss_cls_dn_3: 0.4165, loss_box_dn_3: 1.3608, loss_cls_dn_4: 0.4307, loss_box_dn_4: 1.3719, loss_cls_dn_5: 0.4225, loss_box_dn_5: 1.4642, loss_dense_depth: 1.2329, loss: 38.0021, grad_norm: 75.9488 -2025-11-17 14:20:44,326 - mmdet - INFO - Iter [31/17500] lr: 1.120e-04, eta: 1 day, 1:25:51, time: 1.499, data_time: 0.078, memory: 49163, loss_cls_0: 1.0793, loss_box_0: 2.0822, loss_cns_0: 0.6146, loss_yns_0: 0.1718, loss_cls_1: 1.1505, loss_box_1: 2.5074, loss_cns_1: 0.5867, loss_yns_1: 0.1734, loss_cls_2: 1.1589, loss_box_2: 2.5081, loss_cns_2: 0.5928, loss_yns_2: 0.1779, loss_cls_3: 1.1867, loss_box_3: 2.5812, loss_cns_3: 0.5877, loss_yns_3: 0.1754, loss_cls_4: 1.1862, loss_box_4: 2.6511, loss_cns_4: 0.5730, loss_yns_4: 0.1815, loss_cls_5: 1.1889, loss_box_5: 2.7307, loss_cns_5: 0.5576, loss_yns_5: 0.1770, loss_cls_dn_0: 0.4454, loss_box_dn_0: 0.9882, loss_cls_dn_1: 0.3825, loss_box_dn_1: 1.2810, loss_cls_dn_2: 0.4052, loss_box_dn_2: 1.3103, loss_cls_dn_3: 0.4316, loss_box_dn_3: 1.3633, loss_cls_dn_4: 0.4033, loss_box_dn_4: 1.4084, loss_cls_dn_5: 0.4398, loss_box_dn_5: 1.4835, loss_dense_depth: 1.1987, loss: 38.1215, grad_norm: 74.6407 -2025-11-17 14:20:45,832 - mmdet - INFO - Iter [32/17500] lr: 1.124e-04, eta: 1 day, 0:51:47, time: 1.507, data_time: 0.076, memory: 49163, loss_cls_0: 1.0582, loss_box_0: 2.1124, loss_cns_0: 0.6122, loss_yns_0: 0.1693, loss_cls_1: 1.1373, loss_box_1: 2.4939, loss_cns_1: 0.5766, loss_yns_1: 0.1692, loss_cls_2: 1.1506, loss_box_2: 2.5083, loss_cns_2: 0.5822, loss_yns_2: 0.1749, loss_cls_3: 1.1611, loss_box_3: 2.5325, loss_cns_3: 0.5856, loss_yns_3: 0.1946, loss_cls_4: 1.1836, loss_box_4: 2.6705, loss_cns_4: 0.5756, loss_yns_4: 0.1870, loss_cls_5: 1.1761, loss_box_5: 2.7584, loss_cns_5: 0.5674, loss_yns_5: 0.1755, loss_cls_dn_0: 0.4419, loss_box_dn_0: 0.9939, loss_cls_dn_1: 0.3663, loss_box_dn_1: 1.2595, loss_cls_dn_2: 0.3848, loss_box_dn_2: 1.2911, loss_cls_dn_3: 0.4258, loss_box_dn_3: 1.3198, loss_cls_dn_4: 0.3771, loss_box_dn_4: 1.4081, loss_cls_dn_5: 0.4398, loss_box_dn_5: 1.4698, loss_dense_depth: 1.1363, loss: 37.8272, grad_norm: 70.5180 -2025-11-17 14:20:47,337 - mmdet - INFO - Iter [33/17500] lr: 1.128e-04, eta: 1 day, 0:19:46, time: 1.505, data_time: 0.075, memory: 49163, loss_cls_0: 1.0734, loss_box_0: 2.1094, loss_cns_0: 0.6134, loss_yns_0: 0.1687, loss_cls_1: 1.1358, loss_box_1: 2.4791, loss_cns_1: 0.5807, loss_yns_1: 0.1704, loss_cls_2: 1.1679, loss_box_2: 2.5127, loss_cns_2: 0.5833, loss_yns_2: 0.1757, loss_cls_3: 1.1568, loss_box_3: 2.4971, loss_cns_3: 0.5863, loss_yns_3: 0.1883, loss_cls_4: 1.2103, loss_box_4: 2.5568, loss_cns_4: 0.5859, loss_yns_4: 0.1797, loss_cls_5: 1.1719, loss_box_5: 2.5658, loss_cns_5: 0.5914, loss_yns_5: 0.1720, loss_cls_dn_0: 0.4366, loss_box_dn_0: 1.0033, loss_cls_dn_1: 0.3441, loss_box_dn_1: 1.3224, loss_cls_dn_2: 0.3577, loss_box_dn_2: 1.3491, loss_cls_dn_3: 0.3876, loss_box_dn_3: 1.3345, loss_cls_dn_4: 0.3508, loss_box_dn_4: 1.3857, loss_cls_dn_5: 0.4069, loss_box_dn_5: 1.4040, loss_dense_depth: 1.1109, loss: 37.4265, grad_norm: 57.2268 -2025-11-17 14:20:48,825 - mmdet - INFO - Iter [34/17500] lr: 1.132e-04, eta: 23:49:30, time: 1.488, data_time: 0.071, memory: 49163, loss_cls_0: 1.0727, loss_box_0: 2.0914, loss_cns_0: 0.6136, loss_yns_0: 0.1712, loss_cls_1: 1.1332, loss_box_1: 2.4307, loss_cns_1: 0.5876, loss_yns_1: 0.1729, loss_cls_2: 1.1594, loss_box_2: 2.4914, loss_cns_2: 0.5840, loss_yns_2: 0.1731, loss_cls_3: 1.1750, loss_box_3: 2.4422, loss_cns_3: 0.5875, loss_yns_3: 0.1698, loss_cls_4: 1.1854, loss_box_4: 2.4991, loss_cns_4: 0.5790, loss_yns_4: 0.1733, loss_cls_5: 1.1657, loss_box_5: 2.5519, loss_cns_5: 0.5907, loss_yns_5: 0.1736, loss_cls_dn_0: 0.4283, loss_box_dn_0: 1.0035, loss_cls_dn_1: 0.3395, loss_box_dn_1: 1.3410, loss_cls_dn_2: 0.3601, loss_box_dn_2: 1.3551, loss_cls_dn_3: 0.3686, loss_box_dn_3: 1.3187, loss_cls_dn_4: 0.3571, loss_box_dn_4: 1.3609, loss_cls_dn_5: 0.3927, loss_box_dn_5: 1.3756, loss_dense_depth: 1.1061, loss: 37.0817, grad_norm: 51.0880 -2025-11-17 14:20:50,310 - mmdet - INFO - Iter [35/17500] lr: 1.136e-04, eta: 23:20:56, time: 1.486, data_time: 0.071, memory: 49163, loss_cls_0: 1.0672, loss_box_0: 2.0466, loss_cns_0: 0.6195, loss_yns_0: 0.1712, loss_cls_1: 1.1294, loss_box_1: 2.5282, loss_cns_1: 0.5865, loss_yns_1: 0.1727, loss_cls_2: 1.1459, loss_box_2: 2.5882, loss_cns_2: 0.5842, loss_yns_2: 0.1739, loss_cls_3: 1.1650, loss_box_3: 2.5629, loss_cns_3: 0.5934, loss_yns_3: 0.1779, loss_cls_4: 1.1553, loss_box_4: 2.6354, loss_cns_4: 0.5887, loss_yns_4: 0.1728, loss_cls_5: 1.1618, loss_box_5: 2.7001, loss_cns_5: 0.6045, loss_yns_5: 0.1787, loss_cls_dn_0: 0.4141, loss_box_dn_0: 0.9917, loss_cls_dn_1: 0.3353, loss_box_dn_1: 1.3325, loss_cls_dn_2: 0.3686, loss_box_dn_2: 1.3565, loss_cls_dn_3: 0.3739, loss_box_dn_3: 1.3490, loss_cls_dn_4: 0.3753, loss_box_dn_4: 1.4037, loss_cls_dn_5: 0.3867, loss_box_dn_5: 1.4479, loss_dense_depth: 1.0784, loss: 37.7237, grad_norm: 72.6019 -2025-11-17 14:20:51,804 - mmdet - INFO - Iter [36/17500] lr: 1.140e-04, eta: 22:54:00, time: 1.493, data_time: 0.071, memory: 49163, loss_cls_0: 1.0648, loss_box_0: 2.0265, loss_cns_0: 0.6173, loss_yns_0: 0.1692, loss_cls_1: 1.1149, loss_box_1: 2.6210, loss_cns_1: 0.5783, loss_yns_1: 0.1710, loss_cls_2: 1.1248, loss_box_2: 2.6509, loss_cns_2: 0.5741, loss_yns_2: 0.1787, loss_cls_3: 1.1421, loss_box_3: 2.6870, loss_cns_3: 0.5781, loss_yns_3: 0.1809, loss_cls_4: 1.1405, loss_box_4: 2.7379, loss_cns_4: 0.5856, loss_yns_4: 0.1754, loss_cls_5: 1.1618, loss_box_5: 2.8142, loss_cns_5: 0.5881, loss_yns_5: 0.1755, loss_cls_dn_0: 0.4023, loss_box_dn_0: 0.9849, loss_cls_dn_1: 0.3495, loss_box_dn_1: 1.2031, loss_cls_dn_2: 0.3929, loss_box_dn_2: 1.2356, loss_cls_dn_3: 0.4112, loss_box_dn_3: 1.2849, loss_cls_dn_4: 0.4097, loss_box_dn_4: 1.3577, loss_cls_dn_5: 0.3995, loss_box_dn_5: 1.4399, loss_dense_depth: 1.0640, loss: 37.7940, grad_norm: 66.1207 -2025-11-17 14:20:53,296 - mmdet - INFO - Iter [37/17500] lr: 1.144e-04, eta: 22:28:32, time: 1.493, data_time: 0.080, memory: 49163, loss_cls_0: 1.0600, loss_box_0: 2.0122, loss_cns_0: 0.6176, loss_yns_0: 0.1724, loss_cls_1: 1.1172, loss_box_1: 2.6329, loss_cns_1: 0.5787, loss_yns_1: 0.1698, loss_cls_2: 1.1420, loss_box_2: 2.6783, loss_cns_2: 0.5724, loss_yns_2: 0.1764, loss_cls_3: 1.1408, loss_box_3: 2.7273, loss_cns_3: 0.5748, loss_yns_3: 0.1749, loss_cls_4: 1.1439, loss_box_4: 2.7748, loss_cns_4: 0.5746, loss_yns_4: 0.1750, loss_cls_5: 1.1951, loss_box_5: 2.8459, loss_cns_5: 0.5795, loss_yns_5: 0.1712, loss_cls_dn_0: 0.4066, loss_box_dn_0: 0.9956, loss_cls_dn_1: 0.3322, loss_box_dn_1: 1.3324, loss_cls_dn_2: 0.3731, loss_box_dn_2: 1.3520, loss_cls_dn_3: 0.3936, loss_box_dn_3: 1.4085, loss_cls_dn_4: 0.4008, loss_box_dn_4: 1.4690, loss_cls_dn_5: 0.3688, loss_box_dn_5: 1.5465, loss_dense_depth: 1.1057, loss: 38.4922, grad_norm: 67.7674 -2025-11-17 14:20:54,782 - mmdet - INFO - Iter [38/17500] lr: 1.148e-04, eta: 22:04:21, time: 1.485, data_time: 0.078, memory: 49163, loss_cls_0: 1.0383, loss_box_0: 2.0175, loss_cns_0: 0.6165, loss_yns_0: 0.1705, loss_cls_1: 1.0987, loss_box_1: 2.5200, loss_cns_1: 0.5817, loss_yns_1: 0.1706, loss_cls_2: 1.1277, loss_box_2: 2.5535, loss_cns_2: 0.5812, loss_yns_2: 0.1756, loss_cls_3: 1.1397, loss_box_3: 2.5727, loss_cns_3: 0.5859, loss_yns_3: 0.1699, loss_cls_4: 1.1339, loss_box_4: 2.5795, loss_cns_4: 0.5879, loss_yns_4: 0.1742, loss_cls_5: 1.2166, loss_box_5: 2.6001, loss_cns_5: 0.5935, loss_yns_5: 0.1710, loss_cls_dn_0: 0.4196, loss_box_dn_0: 1.0025, loss_cls_dn_1: 0.3163, loss_box_dn_1: 1.2787, loss_cls_dn_2: 0.3596, loss_box_dn_2: 1.2893, loss_cls_dn_3: 0.3715, loss_box_dn_3: 1.3347, loss_cls_dn_4: 0.3855, loss_box_dn_4: 1.3619, loss_cls_dn_5: 0.3525, loss_box_dn_5: 1.4244, loss_dense_depth: 1.0979, loss: 37.1709, grad_norm: 55.5430 -2025-11-17 14:20:56,266 - mmdet - INFO - Iter [39/17500] lr: 1.152e-04, eta: 21:41:23, time: 1.484, data_time: 0.079, memory: 49163, loss_cls_0: 1.0577, loss_box_0: 2.0140, loss_cns_0: 0.6172, loss_yns_0: 0.1670, loss_cls_1: 1.1121, loss_box_1: 2.3386, loss_cns_1: 0.6037, loss_yns_1: 0.1720, loss_cls_2: 1.1215, loss_box_2: 2.3584, loss_cns_2: 0.6039, loss_yns_2: 0.1724, loss_cls_3: 1.1745, loss_box_3: 2.4092, loss_cns_3: 0.6038, loss_yns_3: 0.1695, loss_cls_4: 1.1537, loss_box_4: 2.3729, loss_cns_4: 0.6115, loss_yns_4: 0.1710, loss_cls_5: 1.2325, loss_box_5: 2.4350, loss_cns_5: 0.5974, loss_yns_5: 0.1686, loss_cls_dn_0: 0.4231, loss_box_dn_0: 1.0028, loss_cls_dn_1: 0.3160, loss_box_dn_1: 1.1568, loss_cls_dn_2: 0.3716, loss_box_dn_2: 1.1595, loss_cls_dn_3: 0.3641, loss_box_dn_3: 1.2022, loss_cls_dn_4: 0.3765, loss_box_dn_4: 1.1952, loss_cls_dn_5: 0.3532, loss_box_dn_5: 1.2578, loss_dense_depth: 1.0541, loss: 35.6712, grad_norm: 51.8397 -2025-11-17 14:20:57,778 - mmdet - INFO - Iter [40/17500] lr: 1.156e-04, eta: 21:19:47, time: 1.512, data_time: 0.077, memory: 49163, loss_cls_0: 1.0744, loss_box_0: 2.0091, loss_cns_0: 0.6167, loss_yns_0: 0.1661, loss_cls_1: 1.1143, loss_box_1: 2.3195, loss_cns_1: 0.6017, loss_yns_1: 0.1745, loss_cls_2: 1.1227, loss_box_2: 2.3131, loss_cns_2: 0.6026, loss_yns_2: 0.1734, loss_cls_3: 1.1567, loss_box_3: 2.3595, loss_cns_3: 0.6047, loss_yns_3: 0.1714, loss_cls_4: 1.1582, loss_box_4: 2.3468, loss_cns_4: 0.6096, loss_yns_4: 0.1728, loss_cls_5: 1.1668, loss_box_5: 2.3990, loss_cns_5: 0.6017, loss_yns_5: 0.1698, loss_cls_dn_0: 0.4058, loss_box_dn_0: 1.0112, loss_cls_dn_1: 0.3130, loss_box_dn_1: 1.1485, loss_cls_dn_2: 0.3701, loss_box_dn_2: 1.1478, loss_cls_dn_3: 0.3612, loss_box_dn_3: 1.1902, loss_cls_dn_4: 0.3654, loss_box_dn_4: 1.1745, loss_cls_dn_5: 0.3739, loss_box_dn_5: 1.2377, loss_dense_depth: 1.0863, loss: 35.3906, grad_norm: 59.0124 -2025-11-17 14:20:59,305 - mmdet - INFO - Iter [41/17500] lr: 1.160e-04, eta: 20:59:20, time: 1.527, data_time: 0.106, memory: 49163, loss_cls_0: 1.0293, loss_box_0: 1.9911, loss_cns_0: 0.6100, loss_yns_0: 0.1648, loss_cls_1: 1.0972, loss_box_1: 2.3233, loss_cns_1: 0.6020, loss_yns_1: 0.1717, loss_cls_2: 1.1143, loss_box_2: 2.3148, loss_cns_2: 0.6025, loss_yns_2: 0.1696, loss_cls_3: 1.1439, loss_box_3: 2.3428, loss_cns_3: 0.6079, loss_yns_3: 0.1705, loss_cls_4: 1.1512, loss_box_4: 2.3650, loss_cns_4: 0.6064, loss_yns_4: 0.1706, loss_cls_5: 1.1455, loss_box_5: 2.3918, loss_cns_5: 0.6144, loss_yns_5: 0.1646, loss_cls_dn_0: 0.4181, loss_box_dn_0: 1.0030, loss_cls_dn_1: 0.2973, loss_box_dn_1: 1.1472, loss_cls_dn_2: 0.3516, loss_box_dn_2: 1.1484, loss_cls_dn_3: 0.3561, loss_box_dn_3: 1.1657, loss_cls_dn_4: 0.3574, loss_box_dn_4: 1.1524, loss_cls_dn_5: 0.3850, loss_box_dn_5: 1.1898, loss_dense_depth: 1.0460, loss: 35.0833, grad_norm: 56.6911 -2025-11-17 14:21:00,839 - mmdet - INFO - Iter [42/17500] lr: 1.164e-04, eta: 20:39:55, time: 1.535, data_time: 0.104, memory: 49163, loss_cls_0: 1.0227, loss_box_0: 1.9801, loss_cns_0: 0.6154, loss_yns_0: 0.1649, loss_cls_1: 1.0878, loss_box_1: 2.3123, loss_cns_1: 0.6079, loss_yns_1: 0.1706, loss_cls_2: 1.1011, loss_box_2: 2.3126, loss_cns_2: 0.6047, loss_yns_2: 0.1688, loss_cls_3: 1.1298, loss_box_3: 2.3373, loss_cns_3: 0.6090, loss_yns_3: 0.1705, loss_cls_4: 1.1299, loss_box_4: 2.3591, loss_cns_4: 0.6030, loss_yns_4: 0.1681, loss_cls_5: 1.1438, loss_box_5: 2.3798, loss_cns_5: 0.6107, loss_yns_5: 0.1659, loss_cls_dn_0: 0.4092, loss_box_dn_0: 1.0029, loss_cls_dn_1: 0.3023, loss_box_dn_1: 1.1061, loss_cls_dn_2: 0.3430, loss_box_dn_2: 1.1157, loss_cls_dn_3: 0.3687, loss_box_dn_3: 1.1445, loss_cls_dn_4: 0.3684, loss_box_dn_4: 1.1472, loss_cls_dn_5: 0.4113, loss_box_dn_5: 1.1712, loss_dense_depth: 1.0266, loss: 34.8729, grad_norm: 53.4017 -2025-11-17 14:21:02,369 - mmdet - INFO - Iter [43/17500] lr: 1.168e-04, eta: 20:21:21, time: 1.529, data_time: 0.077, memory: 49163, loss_cls_0: 1.0429, loss_box_0: 2.0001, loss_cns_0: 0.6126, loss_yns_0: 0.1666, loss_cls_1: 1.1092, loss_box_1: 2.2659, loss_cns_1: 0.6020, loss_yns_1: 0.1668, loss_cls_2: 1.1277, loss_box_2: 2.2775, loss_cns_2: 0.6059, loss_yns_2: 0.1666, loss_cls_3: 1.1425, loss_box_3: 2.3051, loss_cns_3: 0.6106, loss_yns_3: 0.1678, loss_cls_4: 1.1402, loss_box_4: 2.3046, loss_cns_4: 0.6046, loss_yns_4: 0.1679, loss_cls_5: 1.1623, loss_box_5: 2.3399, loss_cns_5: 0.6057, loss_yns_5: 0.1680, loss_cls_dn_0: 0.4097, loss_box_dn_0: 0.9899, loss_cls_dn_1: 0.3062, loss_box_dn_1: 1.0693, loss_cls_dn_2: 0.3410, loss_box_dn_2: 1.0900, loss_cls_dn_3: 0.3828, loss_box_dn_3: 1.1359, loss_cls_dn_4: 0.3923, loss_box_dn_4: 1.1448, loss_cls_dn_5: 0.4286, loss_box_dn_5: 1.1861, loss_dense_depth: 1.0555, loss: 34.7951, grad_norm: 57.3072 -2025-11-17 14:21:03,875 - mmdet - INFO - Iter [44/17500] lr: 1.172e-04, eta: 20:03:29, time: 1.507, data_time: 0.079, memory: 49163, loss_cls_0: 1.0530, loss_box_0: 2.0241, loss_cns_0: 0.6094, loss_yns_0: 0.1699, loss_cls_1: 1.0998, loss_box_1: 2.2798, loss_cns_1: 0.6022, loss_yns_1: 0.1666, loss_cls_2: 1.1232, loss_box_2: 2.2667, loss_cns_2: 0.6020, loss_yns_2: 0.1657, loss_cls_3: 1.1395, loss_box_3: 2.2896, loss_cns_3: 0.6025, loss_yns_3: 0.1650, loss_cls_4: 1.1375, loss_box_4: 2.2788, loss_cns_4: 0.5991, loss_yns_4: 0.1640, loss_cls_5: 1.1532, loss_box_5: 2.3247, loss_cns_5: 0.6010, loss_yns_5: 0.1751, loss_cls_dn_0: 0.4106, loss_box_dn_0: 0.9880, loss_cls_dn_1: 0.2974, loss_box_dn_1: 1.1207, loss_cls_dn_2: 0.3334, loss_box_dn_2: 1.1300, loss_cls_dn_3: 0.3840, loss_box_dn_3: 1.1632, loss_cls_dn_4: 0.4037, loss_box_dn_4: 1.1741, loss_cls_dn_5: 0.4203, loss_box_dn_5: 1.2459, loss_dense_depth: 1.0608, loss: 34.9246, grad_norm: 45.4589 -2025-11-17 14:21:05,402 - mmdet - INFO - Iter [45/17500] lr: 1.176e-04, eta: 19:46:30, time: 1.520, data_time: 0.083, memory: 49163, loss_cls_0: 1.0563, loss_box_0: 2.0340, loss_cns_0: 0.6087, loss_yns_0: 0.1682, loss_cls_1: 1.1226, loss_box_1: 2.3220, loss_cns_1: 0.5958, loss_yns_1: 0.1700, loss_cls_2: 1.1771, loss_box_2: 2.2838, loss_cns_2: 0.6032, loss_yns_2: 0.1682, loss_cls_3: 1.1565, loss_box_3: 2.2750, loss_cns_3: 0.6057, loss_yns_3: 0.1678, loss_cls_4: 1.1268, loss_box_4: 2.2947, loss_cns_4: 0.6073, loss_yns_4: 0.1684, loss_cls_5: 1.1527, loss_box_5: 2.3214, loss_cns_5: 0.6097, loss_yns_5: 0.1737, loss_cls_dn_0: 0.4103, loss_box_dn_0: 0.9825, loss_cls_dn_1: 0.2844, loss_box_dn_1: 1.1054, loss_cls_dn_2: 0.3225, loss_box_dn_2: 1.1002, loss_cls_dn_3: 0.3711, loss_box_dn_3: 1.1111, loss_cls_dn_4: 0.3989, loss_box_dn_4: 1.1428, loss_cls_dn_5: 0.4046, loss_box_dn_5: 1.2090, loss_dense_depth: 1.0605, loss: 34.8732, grad_norm: 44.9593 -2025-11-17 14:21:06,919 - mmdet - INFO - Iter [46/17500] lr: 1.180e-04, eta: 19:30:17, time: 1.523, data_time: 0.090, memory: 49163, loss_cls_0: 1.0349, loss_box_0: 2.0126, loss_cns_0: 0.6081, loss_yns_0: 0.1675, loss_cls_1: 1.1250, loss_box_1: 2.4014, loss_cns_1: 0.5941, loss_yns_1: 0.1671, loss_cls_2: 1.1639, loss_box_2: 2.3967, loss_cns_2: 0.6099, loss_yns_2: 0.1647, loss_cls_3: 1.1909, loss_box_3: 2.3749, loss_cns_3: 0.6110, loss_yns_3: 0.1651, loss_cls_4: 1.1197, loss_box_4: 2.4164, loss_cns_4: 0.6127, loss_yns_4: 0.1726, loss_cls_5: 1.1404, loss_box_5: 2.4207, loss_cns_5: 0.6191, loss_yns_5: 0.1765, loss_cls_dn_0: 0.4175, loss_box_dn_0: 0.9579, loss_cls_dn_1: 0.2733, loss_box_dn_1: 1.0622, loss_cls_dn_2: 0.3111, loss_box_dn_2: 1.0520, loss_cls_dn_3: 0.3414, loss_box_dn_3: 1.0372, loss_cls_dn_4: 0.3714, loss_box_dn_4: 1.0793, loss_cls_dn_5: 0.3818, loss_box_dn_5: 1.1159, loss_dense_depth: 1.0081, loss: 34.8748, grad_norm: 52.2720 -2025-11-17 14:21:08,419 - mmdet - INFO - Iter [47/17500] lr: 1.184e-04, eta: 19:14:35, time: 1.498, data_time: 0.107, memory: 49163, loss_cls_0: 1.0516, loss_box_0: 2.0138, loss_cns_0: 0.6092, loss_yns_0: 0.1647, loss_cls_1: 1.1124, loss_box_1: 2.4104, loss_cns_1: 0.5981, loss_yns_1: 0.1666, loss_cls_2: 1.1396, loss_box_2: 2.4423, loss_cns_2: 0.6099, loss_yns_2: 0.1665, loss_cls_3: 1.2158, loss_box_3: 2.3900, loss_cns_3: 0.6131, loss_yns_3: 0.1664, loss_cls_4: 1.1959, loss_box_4: 2.4249, loss_cns_4: 0.6123, loss_yns_4: 0.1703, loss_cls_5: 1.1796, loss_box_5: 2.4315, loss_cns_5: 0.6151, loss_yns_5: 0.1811, loss_cls_dn_0: 0.4223, loss_box_dn_0: 0.9556, loss_cls_dn_1: 0.2809, loss_box_dn_1: 1.0367, loss_cls_dn_2: 0.3374, loss_box_dn_2: 1.0259, loss_cls_dn_3: 0.3458, loss_box_dn_3: 0.9950, loss_cls_dn_4: 0.3488, loss_box_dn_4: 1.0341, loss_cls_dn_5: 0.3714, loss_box_dn_5: 1.0583, loss_dense_depth: 1.0006, loss: 34.8939, grad_norm: 43.0224 -2025-11-17 14:21:09,905 - mmdet - INFO - Iter [48/17500] lr: 1.188e-04, eta: 18:59:29, time: 1.488, data_time: 0.078, memory: 49163, loss_cls_0: 1.0275, loss_box_0: 2.0071, loss_cns_0: 0.6097, loss_yns_0: 0.1618, loss_cls_1: 1.0968, loss_box_1: 2.3415, loss_cns_1: 0.5898, loss_yns_1: 0.1629, loss_cls_2: 1.1561, loss_box_2: 2.3472, loss_cns_2: 0.6117, loss_yns_2: 0.1656, loss_cls_3: 1.1434, loss_box_3: 2.3408, loss_cns_3: 0.6137, loss_yns_3: 0.1650, loss_cls_4: 1.1582, loss_box_4: 2.4019, loss_cns_4: 0.6076, loss_yns_4: 0.1691, loss_cls_5: 1.1833, loss_box_5: 2.4226, loss_cns_5: 0.6069, loss_yns_5: 0.1751, loss_cls_dn_0: 0.4131, loss_box_dn_0: 0.9475, loss_cls_dn_1: 0.2700, loss_box_dn_1: 1.1023, loss_cls_dn_2: 0.3339, loss_box_dn_2: 1.0600, loss_cls_dn_3: 0.3636, loss_box_dn_3: 1.0468, loss_cls_dn_4: 0.3337, loss_box_dn_4: 1.0911, loss_cls_dn_5: 0.3510, loss_box_dn_5: 1.1151, loss_dense_depth: 1.0037, loss: 34.6971, grad_norm: 51.3990 -2025-11-17 14:21:11,426 - mmdet - INFO - Iter [49/17500] lr: 1.192e-04, eta: 18:45:12, time: 1.521, data_time: 0.094, memory: 49163, loss_cls_0: 1.0512, loss_box_0: 1.9891, loss_cns_0: 0.6089, loss_yns_0: 0.1632, loss_cls_1: 1.0671, loss_box_1: 2.3362, loss_cns_1: 0.5926, loss_yns_1: 0.1667, loss_cls_2: 1.1327, loss_box_2: 2.3004, loss_cns_2: 0.6133, loss_yns_2: 0.1676, loss_cls_3: 1.1689, loss_box_3: 2.2981, loss_cns_3: 0.6154, loss_yns_3: 0.1660, loss_cls_4: 1.1245, loss_box_4: 2.3207, loss_cns_4: 0.6130, loss_yns_4: 0.1705, loss_cls_5: 1.1587, loss_box_5: 2.3576, loss_cns_5: 0.6130, loss_yns_5: 0.1710, loss_cls_dn_0: 0.3959, loss_box_dn_0: 0.9404, loss_cls_dn_1: 0.2520, loss_box_dn_1: 1.1606, loss_cls_dn_2: 0.3117, loss_box_dn_2: 1.1083, loss_cls_dn_3: 0.3764, loss_box_dn_3: 1.1107, loss_cls_dn_4: 0.3323, loss_box_dn_4: 1.1329, loss_cls_dn_5: 0.3381, loss_box_dn_5: 1.1633, loss_dense_depth: 1.0169, loss: 34.6058, grad_norm: 41.7912 -2025-11-17 14:21:12,919 - mmdet - INFO - Iter [50/17500] lr: 1.196e-04, eta: 18:31:18, time: 1.492, data_time: 0.079, memory: 49163, loss_cls_0: 1.0425, loss_box_0: 1.9835, loss_cns_0: 0.6115, loss_yns_0: 0.1643, loss_cls_1: 1.1097, loss_box_1: 2.3735, loss_cns_1: 0.5990, loss_yns_1: 0.1668, loss_cls_2: 1.1101, loss_box_2: 2.3344, loss_cns_2: 0.6176, loss_yns_2: 0.1661, loss_cls_3: 1.1711, loss_box_3: 2.3373, loss_cns_3: 0.6173, loss_yns_3: 0.1676, loss_cls_4: 1.1185, loss_box_4: 2.3482, loss_cns_4: 0.6180, loss_yns_4: 0.1765, loss_cls_5: 1.1489, loss_box_5: 2.3900, loss_cns_5: 0.6102, loss_yns_5: 0.1709, loss_cls_dn_0: 0.3778, loss_box_dn_0: 0.9516, loss_cls_dn_1: 0.2459, loss_box_dn_1: 1.1210, loss_cls_dn_2: 0.2891, loss_box_dn_2: 1.0833, loss_cls_dn_3: 0.3753, loss_box_dn_3: 1.1047, loss_cls_dn_4: 0.3382, loss_box_dn_4: 1.1079, loss_cls_dn_5: 0.3427, loss_box_dn_5: 1.1490, loss_dense_depth: 0.9969, loss: 34.6372, grad_norm: 48.5743 -2025-11-17 14:21:14,431 - mmdet - INFO - Iter [51/17500] lr: 1.200e-04, eta: 18:18:05, time: 1.513, data_time: 0.083, memory: 49163, loss_cls_0: 1.0215, loss_box_0: 1.9585, loss_cns_0: 0.6096, loss_yns_0: 0.1631, loss_cls_1: 1.0483, loss_box_1: 2.3478, loss_cns_1: 0.5970, loss_yns_1: 0.1640, loss_cls_2: 1.0906, loss_box_2: 2.3832, loss_cns_2: 0.6168, loss_yns_2: 0.1647, loss_cls_3: 1.1502, loss_box_3: 2.3615, loss_cns_3: 0.6215, loss_yns_3: 0.1663, loss_cls_4: 1.1073, loss_box_4: 2.3670, loss_cns_4: 0.6209, loss_yns_4: 0.1729, loss_cls_5: 1.1446, loss_box_5: 2.4067, loss_cns_5: 0.6190, loss_yns_5: 0.1668, loss_cls_dn_0: 0.3839, loss_box_dn_0: 0.9388, loss_cls_dn_1: 0.2407, loss_box_dn_1: 1.1607, loss_cls_dn_2: 0.2784, loss_box_dn_2: 1.1472, loss_cls_dn_3: 0.3516, loss_box_dn_3: 1.1574, loss_cls_dn_4: 0.3352, loss_box_dn_4: 1.1533, loss_cls_dn_5: 0.3353, loss_box_dn_5: 1.1902, loss_dense_depth: 0.9768, loss: 34.7194, grad_norm: 44.1346 -2025-11-17 14:21:15,923 - mmdet - INFO - Iter [52/17500] lr: 1.204e-04, eta: 18:05:15, time: 1.492, data_time: 0.078, memory: 49163, loss_cls_0: 1.0351, loss_box_0: 1.9806, loss_cns_0: 0.6009, loss_yns_0: 0.1635, loss_cls_1: 1.0723, loss_box_1: 2.2916, loss_cns_1: 0.5813, loss_yns_1: 0.1625, loss_cls_2: 1.1073, loss_box_2: 2.3669, loss_cns_2: 0.6097, loss_yns_2: 0.1654, loss_cls_3: 1.1264, loss_box_3: 2.3366, loss_cns_3: 0.6211, loss_yns_3: 0.1669, loss_cls_4: 1.1100, loss_box_4: 2.3441, loss_cns_4: 0.6212, loss_yns_4: 0.1665, loss_cls_5: 1.1446, loss_box_5: 2.3751, loss_cns_5: 0.6249, loss_yns_5: 0.1667, loss_cls_dn_0: 0.3969, loss_box_dn_0: 0.9309, loss_cls_dn_1: 0.2449, loss_box_dn_1: 1.1356, loss_cls_dn_2: 0.2777, loss_box_dn_2: 1.1355, loss_cls_dn_3: 0.3330, loss_box_dn_3: 1.1383, loss_cls_dn_4: 0.3302, loss_box_dn_4: 1.1419, loss_cls_dn_5: 0.3413, loss_box_dn_5: 1.1768, loss_dense_depth: 0.9712, loss: 34.4957, grad_norm: 47.0603 -2025-11-17 14:21:17,419 - mmdet - INFO - Iter [53/17500] lr: 1.208e-04, eta: 17:52:55, time: 1.496, data_time: 0.074, memory: 49163, loss_cls_0: 1.0477, loss_box_0: 1.9753, loss_cns_0: 0.6011, loss_yns_0: 0.1661, loss_cls_1: 1.0789, loss_box_1: 2.4049, loss_cns_1: 0.5729, loss_yns_1: 0.1606, loss_cls_2: 1.1284, loss_box_2: 2.4596, loss_cns_2: 0.6008, loss_yns_2: 0.1658, loss_cls_3: 1.1694, loss_box_3: 2.4392, loss_cns_3: 0.6106, loss_yns_3: 0.1646, loss_cls_4: 1.1106, loss_box_4: 2.4462, loss_cns_4: 0.6098, loss_yns_4: 0.1631, loss_cls_5: 1.1401, loss_box_5: 2.4651, loss_cns_5: 0.6124, loss_yns_5: 0.1657, loss_cls_dn_0: 0.3995, loss_box_dn_0: 0.9140, loss_cls_dn_1: 0.2409, loss_box_dn_1: 1.0838, loss_cls_dn_2: 0.2731, loss_box_dn_2: 1.0676, loss_cls_dn_3: 0.3105, loss_box_dn_3: 1.0723, loss_cls_dn_4: 0.3174, loss_box_dn_4: 1.0752, loss_cls_dn_5: 0.3506, loss_box_dn_5: 1.1024, loss_dense_depth: 1.0516, loss: 34.7181, grad_norm: 40.4805 -2025-11-17 14:21:18,921 - mmdet - INFO - Iter [54/17500] lr: 1.212e-04, eta: 17:41:04, time: 1.502, data_time: 0.078, memory: 49163, loss_cls_0: 0.9993, loss_box_0: 1.9469, loss_cns_0: 0.6125, loss_yns_0: 0.1630, loss_cls_1: 1.0367, loss_box_1: 2.3117, loss_cns_1: 0.6081, loss_yns_1: 0.1620, loss_cls_2: 1.0606, loss_box_2: 2.2914, loss_cns_2: 0.6233, loss_yns_2: 0.1603, loss_cls_3: 1.1520, loss_box_3: 2.2983, loss_cns_3: 0.6273, loss_yns_3: 0.1596, loss_cls_4: 1.0894, loss_box_4: 2.2943, loss_cns_4: 0.6265, loss_yns_4: 0.1614, loss_cls_5: 1.1102, loss_box_5: 2.3063, loss_cns_5: 0.6277, loss_yns_5: 0.1651, loss_cls_dn_0: 0.3770, loss_box_dn_0: 0.9080, loss_cls_dn_1: 0.2296, loss_box_dn_1: 1.0790, loss_cls_dn_2: 0.2696, loss_box_dn_2: 1.0586, loss_cls_dn_3: 0.2978, loss_box_dn_3: 1.0627, loss_cls_dn_4: 0.3116, loss_box_dn_4: 1.0575, loss_cls_dn_5: 0.3494, loss_box_dn_5: 1.0762, loss_dense_depth: 0.9833, loss: 33.6543, grad_norm: 51.1489 -2025-11-17 14:21:20,416 - mmdet - INFO - Iter [55/17500] lr: 1.216e-04, eta: 17:29:38, time: 1.496, data_time: 0.088, memory: 49163, loss_cls_0: 1.0230, loss_box_0: 1.9839, loss_cns_0: 0.6101, loss_yns_0: 0.1634, loss_cls_1: 1.0683, loss_box_1: 2.2350, loss_cns_1: 0.6145, loss_yns_1: 0.1655, loss_cls_2: 1.0829, loss_box_2: 2.2108, loss_cns_2: 0.6263, loss_yns_2: 0.1646, loss_cls_3: 1.1106, loss_box_3: 2.1995, loss_cns_3: 0.6314, loss_yns_3: 0.1655, loss_cls_4: 1.0819, loss_box_4: 2.2135, loss_cns_4: 0.6286, loss_yns_4: 0.1750, loss_cls_5: 1.1046, loss_box_5: 2.2081, loss_cns_5: 0.6290, loss_yns_5: 0.1654, loss_cls_dn_0: 0.3778, loss_box_dn_0: 0.8988, loss_cls_dn_1: 0.2372, loss_box_dn_1: 0.9181, loss_cls_dn_2: 0.2800, loss_box_dn_2: 0.9059, loss_cls_dn_3: 0.3179, loss_box_dn_3: 0.9106, loss_cls_dn_4: 0.3387, loss_box_dn_4: 0.9155, loss_cls_dn_5: 0.3874, loss_box_dn_5: 0.9309, loss_dense_depth: 0.9857, loss: 32.6660, grad_norm: 58.2673 -2025-11-17 14:21:21,929 - mmdet - INFO - Iter [56/17500] lr: 1.220e-04, eta: 17:18:40, time: 1.510, data_time: 0.078, memory: 49163, loss_cls_0: 1.0162, loss_box_0: 1.9437, loss_cns_0: 0.6110, loss_yns_0: 0.1672, loss_cls_1: 1.0745, loss_box_1: 2.1673, loss_cns_1: 0.6133, loss_yns_1: 0.1653, loss_cls_2: 1.0976, loss_box_2: 2.1503, loss_cns_2: 0.6193, loss_yns_2: 0.1660, loss_cls_3: 1.1148, loss_box_3: 2.1329, loss_cns_3: 0.6256, loss_yns_3: 0.1647, loss_cls_4: 1.0743, loss_box_4: 2.1752, loss_cns_4: 0.6230, loss_yns_4: 0.1878, loss_cls_5: 1.0974, loss_box_5: 2.1799, loss_cns_5: 0.6229, loss_yns_5: 0.1648, loss_cls_dn_0: 0.3658, loss_box_dn_0: 0.8966, loss_cls_dn_1: 0.2343, loss_box_dn_1: 0.9149, loss_cls_dn_2: 0.2767, loss_box_dn_2: 0.8992, loss_cls_dn_3: 0.3142, loss_box_dn_3: 0.8956, loss_cls_dn_4: 0.3288, loss_box_dn_4: 0.9204, loss_cls_dn_5: 0.3863, loss_box_dn_5: 0.9389, loss_dense_depth: 0.9721, loss: 32.2990, grad_norm: 43.4109 -2025-11-17 14:21:23,427 - mmdet - INFO - Iter [57/17500] lr: 1.224e-04, eta: 17:08:02, time: 1.501, data_time: 0.082, memory: 49163, loss_cls_0: 1.0099, loss_box_0: 1.9180, loss_cns_0: 0.6141, loss_yns_0: 0.1669, loss_cls_1: 1.0507, loss_box_1: 2.1652, loss_cns_1: 0.6184, loss_yns_1: 0.1658, loss_cls_2: 1.0739, loss_box_2: 2.1325, loss_cns_2: 0.6215, loss_yns_2: 0.1664, loss_cls_3: 1.1060, loss_box_3: 2.1129, loss_cns_3: 0.6251, loss_yns_3: 0.1653, loss_cls_4: 1.1075, loss_box_4: 2.1505, loss_cns_4: 0.6239, loss_yns_4: 0.1809, loss_cls_5: 1.0881, loss_box_5: 2.1735, loss_cns_5: 0.6236, loss_yns_5: 0.1659, loss_cls_dn_0: 0.3456, loss_box_dn_0: 0.8917, loss_cls_dn_1: 0.2313, loss_box_dn_1: 0.9103, loss_cls_dn_2: 0.2568, loss_box_dn_2: 0.8868, loss_cls_dn_3: 0.3003, loss_box_dn_3: 0.8897, loss_cls_dn_4: 0.2984, loss_box_dn_4: 0.9204, loss_cls_dn_5: 0.3691, loss_box_dn_5: 0.9460, loss_dense_depth: 0.9768, loss: 32.0498, grad_norm: 46.6586 -2025-11-17 14:21:24,924 - mmdet - INFO - Iter [58/17500] lr: 1.228e-04, eta: 16:57:45, time: 1.495, data_time: 0.079, memory: 49163, loss_cls_0: 0.9931, loss_box_0: 1.9157, loss_cns_0: 0.6133, loss_yns_0: 0.1640, loss_cls_1: 1.0357, loss_box_1: 2.1517, loss_cns_1: 0.6126, loss_yns_1: 0.1655, loss_cls_2: 1.0730, loss_box_2: 2.0830, loss_cns_2: 0.6292, loss_yns_2: 0.1671, loss_cls_3: 1.0908, loss_box_3: 2.0527, loss_cns_3: 0.6300, loss_yns_3: 0.1666, loss_cls_4: 1.0954, loss_box_4: 2.0632, loss_cns_4: 0.6303, loss_yns_4: 0.1668, loss_cls_5: 1.0784, loss_box_5: 2.0710, loss_cns_5: 0.6335, loss_yns_5: 0.1667, loss_cls_dn_0: 0.3379, loss_box_dn_0: 0.8814, loss_cls_dn_1: 0.2304, loss_box_dn_1: 0.9289, loss_cls_dn_2: 0.2444, loss_box_dn_2: 0.8958, loss_cls_dn_3: 0.2797, loss_box_dn_3: 0.9008, loss_cls_dn_4: 0.2837, loss_box_dn_4: 0.9215, loss_cls_dn_5: 0.3460, loss_box_dn_5: 0.9428, loss_dense_depth: 0.9729, loss: 31.6155, grad_norm: 50.5309 -2025-11-17 14:21:26,424 - mmdet - INFO - Iter [59/17500] lr: 1.232e-04, eta: 16:47:50, time: 1.499, data_time: 0.083, memory: 49163, loss_cls_0: 0.9596, loss_box_0: 1.8974, loss_cns_0: 0.6122, loss_yns_0: 0.1650, loss_cls_1: 1.0483, loss_box_1: 2.1042, loss_cns_1: 0.6204, loss_yns_1: 0.1643, loss_cls_2: 1.0747, loss_box_2: 2.0692, loss_cns_2: 0.6327, loss_yns_2: 0.1671, loss_cls_3: 1.0872, loss_box_3: 2.0608, loss_cns_3: 0.6308, loss_yns_3: 0.1667, loss_cls_4: 1.0621, loss_box_4: 2.0932, loss_cns_4: 0.6261, loss_yns_4: 0.1704, loss_cls_5: 1.0873, loss_box_5: 2.0798, loss_cns_5: 0.6352, loss_yns_5: 0.1696, loss_cls_dn_0: 0.3592, loss_box_dn_0: 0.8784, loss_cls_dn_1: 0.2344, loss_box_dn_1: 0.9266, loss_cls_dn_2: 0.2482, loss_box_dn_2: 0.8972, loss_cls_dn_3: 0.2676, loss_box_dn_3: 0.8984, loss_cls_dn_4: 0.2899, loss_box_dn_4: 0.9203, loss_cls_dn_5: 0.3231, loss_box_dn_5: 0.9346, loss_dense_depth: 0.9056, loss: 31.4677, grad_norm: 52.3561 -2025-11-17 14:21:27,921 - mmdet - INFO - Iter [60/17500] lr: 1.236e-04, eta: 16:38:14, time: 1.498, data_time: 0.081, memory: 49163, loss_cls_0: 0.9774, loss_box_0: 1.8567, loss_cns_0: 0.6054, loss_yns_0: 0.1617, loss_cls_1: 1.0568, loss_box_1: 2.0156, loss_cns_1: 0.6244, loss_yns_1: 0.1632, loss_cls_2: 1.0772, loss_box_2: 1.9901, loss_cns_2: 0.6371, loss_yns_2: 0.1659, loss_cls_3: 1.0919, loss_box_3: 1.9870, loss_cns_3: 0.6371, loss_yns_3: 0.1642, loss_cls_4: 1.1088, loss_box_4: 2.0021, loss_cns_4: 0.6333, loss_yns_4: 0.1812, loss_cls_5: 1.1192, loss_box_5: 2.0042, loss_cns_5: 0.6397, loss_yns_5: 0.1666, loss_cls_dn_0: 0.3745, loss_box_dn_0: 0.8962, loss_cls_dn_1: 0.2293, loss_box_dn_1: 0.9099, loss_cls_dn_2: 0.2442, loss_box_dn_2: 0.8817, loss_cls_dn_3: 0.2610, loss_box_dn_3: 0.8738, loss_cls_dn_4: 0.2957, loss_box_dn_4: 0.8940, loss_cls_dn_5: 0.3044, loss_box_dn_5: 0.9099, loss_dense_depth: 0.9399, loss: 31.0812, grad_norm: 34.7147 -2025-11-17 14:21:29,453 - mmdet - INFO - Iter [61/17500] lr: 1.240e-04, eta: 16:29:07, time: 1.534, data_time: 0.103, memory: 49163, loss_cls_0: 0.9959, loss_box_0: 1.8812, loss_cns_0: 0.6049, loss_yns_0: 0.1599, loss_cls_1: 1.0383, loss_box_1: 2.0255, loss_cns_1: 0.6254, loss_yns_1: 0.1624, loss_cls_2: 1.0780, loss_box_2: 1.9861, loss_cns_2: 0.6402, loss_yns_2: 0.1633, loss_cls_3: 1.0797, loss_box_3: 1.9976, loss_cns_3: 0.6436, loss_yns_3: 0.1626, loss_cls_4: 1.0645, loss_box_4: 2.0059, loss_cns_4: 0.6392, loss_yns_4: 0.1825, loss_cls_5: 1.1368, loss_box_5: 2.0248, loss_cns_5: 0.6407, loss_yns_5: 0.1644, loss_cls_dn_0: 0.3363, loss_box_dn_0: 0.8989, loss_cls_dn_1: 0.2220, loss_box_dn_1: 0.9234, loss_cls_dn_2: 0.2387, loss_box_dn_2: 0.8900, loss_cls_dn_3: 0.2575, loss_box_dn_3: 0.8837, loss_cls_dn_4: 0.2677, loss_box_dn_4: 0.9032, loss_cls_dn_5: 0.2888, loss_box_dn_5: 0.9204, loss_dense_depth: 0.9935, loss: 31.1278, grad_norm: 53.2495 -2025-11-17 14:21:31,029 - mmdet - INFO - Iter [62/17500] lr: 1.244e-04, eta: 16:20:30, time: 1.575, data_time: 0.150, memory: 49163, loss_cls_0: 1.0168, loss_box_0: 1.8724, loss_cns_0: 0.6111, loss_yns_0: 0.1600, loss_cls_1: 1.0291, loss_box_1: 2.1226, loss_cns_1: 0.6236, loss_yns_1: 0.1613, loss_cls_2: 1.0878, loss_box_2: 2.0629, loss_cns_2: 0.6408, loss_yns_2: 0.1628, loss_cls_3: 1.0862, loss_box_3: 2.0784, loss_cns_3: 0.6437, loss_yns_3: 0.1616, loss_cls_4: 1.0911, loss_box_4: 2.0745, loss_cns_4: 0.6437, loss_yns_4: 0.1702, loss_cls_5: 1.1019, loss_box_5: 2.0924, loss_cns_5: 0.6438, loss_yns_5: 0.1655, loss_cls_dn_0: 0.3170, loss_box_dn_0: 0.8897, loss_cls_dn_1: 0.2193, loss_box_dn_1: 0.8913, loss_cls_dn_2: 0.2379, loss_box_dn_2: 0.8594, loss_cls_dn_3: 0.2585, loss_box_dn_3: 0.8568, loss_cls_dn_4: 0.2426, loss_box_dn_4: 0.8687, loss_cls_dn_5: 0.2934, loss_box_dn_5: 0.8828, loss_dense_depth: 0.9519, loss: 31.2732, grad_norm: 45.0261 -2025-11-17 14:21:32,553 - mmdet - INFO - Iter [63/17500] lr: 1.248e-04, eta: 16:11:54, time: 1.523, data_time: 0.075, memory: 49163, loss_cls_0: 0.9991, loss_box_0: 1.8649, loss_cns_0: 0.6129, loss_yns_0: 0.1623, loss_cls_1: 1.0362, loss_box_1: 2.1264, loss_cns_1: 0.6206, loss_yns_1: 0.1634, loss_cls_2: 1.0851, loss_box_2: 2.0600, loss_cns_2: 0.6396, loss_yns_2: 0.1661, loss_cls_3: 1.0820, loss_box_3: 2.0576, loss_cns_3: 0.6402, loss_yns_3: 0.1633, loss_cls_4: 1.1183, loss_box_4: 2.0693, loss_cns_4: 0.6410, loss_yns_4: 0.1657, loss_cls_5: 1.0928, loss_box_5: 2.0490, loss_cns_5: 0.6444, loss_yns_5: 0.1612, loss_cls_dn_0: 0.3323, loss_box_dn_0: 0.8803, loss_cls_dn_1: 0.2143, loss_box_dn_1: 0.8768, loss_cls_dn_2: 0.2335, loss_box_dn_2: 0.8444, loss_cls_dn_3: 0.2445, loss_box_dn_3: 0.8568, loss_cls_dn_4: 0.2296, loss_box_dn_4: 0.8645, loss_cls_dn_5: 0.2920, loss_box_dn_5: 0.8741, loss_dense_depth: 1.0398, loss: 31.2041, grad_norm: 51.9207 -2025-11-17 14:21:34,052 - mmdet - INFO - Iter [64/17500] lr: 1.252e-04, eta: 16:03:28, time: 1.499, data_time: 0.074, memory: 49163, loss_cls_0: 0.9751, loss_box_0: 1.8431, loss_cns_0: 0.6170, loss_yns_0: 0.1617, loss_cls_1: 1.0411, loss_box_1: 2.1223, loss_cns_1: 0.6194, loss_yns_1: 0.1652, loss_cls_2: 1.0659, loss_box_2: 2.0604, loss_cns_2: 0.6403, loss_yns_2: 0.1656, loss_cls_3: 1.0884, loss_box_3: 2.0573, loss_cns_3: 0.6406, loss_yns_3: 0.1655, loss_cls_4: 1.1027, loss_box_4: 2.0676, loss_cns_4: 0.6401, loss_yns_4: 0.1718, loss_cls_5: 1.1089, loss_box_5: 2.0553, loss_cns_5: 0.6430, loss_yns_5: 0.1634, loss_cls_dn_0: 0.3285, loss_box_dn_0: 0.8686, loss_cls_dn_1: 0.2124, loss_box_dn_1: 0.8954, loss_cls_dn_2: 0.2303, loss_box_dn_2: 0.8745, loss_cls_dn_3: 0.2339, loss_box_dn_3: 0.9079, loss_cls_dn_4: 0.2290, loss_box_dn_4: 0.9250, loss_cls_dn_5: 0.2925, loss_box_dn_5: 0.9471, loss_dense_depth: 0.9092, loss: 31.2358, grad_norm: 54.2021 -2025-11-17 14:21:35,547 - mmdet - INFO - Iter [65/17500] lr: 1.256e-04, eta: 15:55:16, time: 1.495, data_time: 0.079, memory: 49163, loss_cls_0: 0.9641, loss_box_0: 1.8582, loss_cns_0: 0.6184, loss_yns_0: 0.1617, loss_cls_1: 1.0011, loss_box_1: 2.0933, loss_cns_1: 0.6152, loss_yns_1: 0.1624, loss_cls_2: 1.0497, loss_box_2: 2.0488, loss_cns_2: 0.6423, loss_yns_2: 0.1627, loss_cls_3: 1.0785, loss_box_3: 2.0524, loss_cns_3: 0.6436, loss_yns_3: 0.1652, loss_cls_4: 1.0763, loss_box_4: 2.0396, loss_cns_4: 0.6392, loss_yns_4: 0.1761, loss_cls_5: 1.0667, loss_box_5: 2.0445, loss_cns_5: 0.6418, loss_yns_5: 0.1623, loss_cls_dn_0: 0.3168, loss_box_dn_0: 0.8740, loss_cls_dn_1: 0.2111, loss_box_dn_1: 0.9896, loss_cls_dn_2: 0.2277, loss_box_dn_2: 0.9827, loss_cls_dn_3: 0.2287, loss_box_dn_3: 1.0177, loss_cls_dn_4: 0.2202, loss_box_dn_4: 1.0369, loss_cls_dn_5: 0.2612, loss_box_dn_5: 1.0572, loss_dense_depth: 0.9566, loss: 31.5443, grad_norm: 53.2754 -2025-11-17 14:21:37,054 - mmdet - INFO - Iter [66/17500] lr: 1.260e-04, eta: 15:47:23, time: 1.508, data_time: 0.077, memory: 49163, loss_cls_0: 0.9794, loss_box_0: 1.8447, loss_cns_0: 0.6186, loss_yns_0: 0.1627, loss_cls_1: 1.0034, loss_box_1: 2.0935, loss_cns_1: 0.6079, loss_yns_1: 0.1634, loss_cls_2: 1.0806, loss_box_2: 2.0584, loss_cns_2: 0.6384, loss_yns_2: 0.1648, loss_cls_3: 1.0998, loss_box_3: 2.0411, loss_cns_3: 0.6425, loss_yns_3: 0.1652, loss_cls_4: 1.1393, loss_box_4: 2.0473, loss_cns_4: 0.6388, loss_yns_4: 0.1795, loss_cls_5: 1.1848, loss_box_5: 2.0395, loss_cns_5: 0.6401, loss_yns_5: 0.1638, loss_cls_dn_0: 0.2994, loss_box_dn_0: 0.8636, loss_cls_dn_1: 0.1985, loss_box_dn_1: 1.0845, loss_cls_dn_2: 0.2182, loss_box_dn_2: 1.0793, loss_cls_dn_3: 0.2265, loss_box_dn_3: 1.0980, loss_cls_dn_4: 0.2197, loss_box_dn_4: 1.1199, loss_cls_dn_5: 0.2367, loss_box_dn_5: 1.1280, loss_dense_depth: 0.9230, loss: 32.0930, grad_norm: 93.4729 -2025-11-17 14:21:38,553 - mmdet - INFO - Iter [67/17500] lr: 1.264e-04, eta: 15:39:41, time: 1.498, data_time: 0.105, memory: 49163, loss_cls_0: 0.9916, loss_box_0: 1.8488, loss_cns_0: 0.6168, loss_yns_0: 0.1631, loss_cls_1: 1.0530, loss_box_1: 2.0884, loss_cns_1: 0.6063, loss_yns_1: 0.1619, loss_cls_2: 1.1024, loss_box_2: 2.1128, loss_cns_2: 0.6362, loss_yns_2: 0.1615, loss_cls_3: 1.1110, loss_box_3: 2.0764, loss_cns_3: 0.6393, loss_yns_3: 0.1646, loss_cls_4: 1.1330, loss_box_4: 2.0774, loss_cns_4: 0.6375, loss_yns_4: 0.1641, loss_cls_5: 1.1093, loss_box_5: 2.0771, loss_cns_5: 0.6366, loss_yns_5: 0.1650, loss_cls_dn_0: 0.3023, loss_box_dn_0: 0.8576, loss_cls_dn_1: 0.2000, loss_box_dn_1: 1.0120, loss_cls_dn_2: 0.2209, loss_box_dn_2: 1.0069, loss_cls_dn_3: 0.2266, loss_box_dn_3: 1.0042, loss_cls_dn_4: 0.2228, loss_box_dn_4: 1.0174, loss_cls_dn_5: 0.2475, loss_box_dn_5: 1.0166, loss_dense_depth: 0.9667, loss: 31.8357, grad_norm: 51.5809 -2025-11-17 14:21:40,030 - mmdet - INFO - Iter [68/17500] lr: 1.268e-04, eta: 15:32:08, time: 1.478, data_time: 0.071, memory: 49163, loss_cls_0: 0.9697, loss_box_0: 1.8183, loss_cns_0: 0.6192, loss_yns_0: 0.1627, loss_cls_1: 1.0220, loss_box_1: 1.9710, loss_cns_1: 0.6191, loss_yns_1: 0.1616, loss_cls_2: 1.0720, loss_box_2: 1.9603, loss_cns_2: 0.6405, loss_yns_2: 0.1641, loss_cls_3: 1.0944, loss_box_3: 1.9486, loss_cns_3: 0.6452, loss_yns_3: 0.1631, loss_cls_4: 1.1221, loss_box_4: 1.9498, loss_cns_4: 0.6439, loss_yns_4: 0.1627, loss_cls_5: 1.0807, loss_box_5: 1.9448, loss_cns_5: 0.6421, loss_yns_5: 0.1627, loss_cls_dn_0: 0.3132, loss_box_dn_0: 0.8596, loss_cls_dn_1: 0.2071, loss_box_dn_1: 0.9210, loss_cls_dn_2: 0.2272, loss_box_dn_2: 0.9031, loss_cls_dn_3: 0.2336, loss_box_dn_3: 0.9090, loss_cls_dn_4: 0.2379, loss_box_dn_4: 0.9080, loss_cls_dn_5: 0.2648, loss_box_dn_5: 0.9073, loss_dense_depth: 0.9207, loss: 30.5531, grad_norm: 40.7411 -2025-11-17 14:21:41,525 - mmdet - INFO - Iter [69/17500] lr: 1.272e-04, eta: 15:24:52, time: 1.495, data_time: 0.080, memory: 49163, loss_cls_0: 0.9316, loss_box_0: 1.7871, loss_cns_0: 0.6210, loss_yns_0: 0.1586, loss_cls_1: 0.9849, loss_box_1: 1.9665, loss_cns_1: 0.6302, loss_yns_1: 0.1599, loss_cls_2: 1.0313, loss_box_2: 1.9052, loss_cns_2: 0.6459, loss_yns_2: 0.1621, loss_cls_3: 1.0516, loss_box_3: 1.9206, loss_cns_3: 0.6539, loss_yns_3: 0.1640, loss_cls_4: 1.0546, loss_box_4: 1.9121, loss_cns_4: 0.6512, loss_yns_4: 0.1618, loss_cls_5: 1.1036, loss_box_5: 1.9212, loss_cns_5: 0.6487, loss_yns_5: 0.1599, loss_cls_dn_0: 0.2955, loss_box_dn_0: 0.8494, loss_cls_dn_1: 0.2053, loss_box_dn_1: 0.9057, loss_cls_dn_2: 0.2227, loss_box_dn_2: 0.8831, loss_cls_dn_3: 0.2288, loss_box_dn_3: 0.8963, loss_cls_dn_4: 0.2313, loss_box_dn_4: 0.8842, loss_cls_dn_5: 0.2544, loss_box_dn_5: 0.8941, loss_dense_depth: 0.8851, loss: 30.0232, grad_norm: 57.8993 -2025-11-17 14:21:43,016 - mmdet - INFO - Iter [70/17500] lr: 1.276e-04, eta: 15:17:47, time: 1.490, data_time: 0.075, memory: 49163, loss_cls_0: 0.9471, loss_box_0: 1.7814, loss_cns_0: 0.6154, loss_yns_0: 0.1591, loss_cls_1: 0.9969, loss_box_1: 1.9409, loss_cns_1: 0.6296, loss_yns_1: 0.1601, loss_cls_2: 1.0308, loss_box_2: 1.8879, loss_cns_2: 0.6426, loss_yns_2: 0.1599, loss_cls_3: 1.0505, loss_box_3: 1.9024, loss_cns_3: 0.6478, loss_yns_3: 0.1630, loss_cls_4: 1.0600, loss_box_4: 1.9110, loss_cns_4: 0.6462, loss_yns_4: 0.1596, loss_cls_5: 1.0539, loss_box_5: 1.9330, loss_cns_5: 0.6455, loss_yns_5: 0.1632, loss_cls_dn_0: 0.2894, loss_box_dn_0: 0.8505, loss_cls_dn_1: 0.2095, loss_box_dn_1: 0.8147, loss_cls_dn_2: 0.2259, loss_box_dn_2: 0.8015, loss_cls_dn_3: 0.2268, loss_box_dn_3: 0.8224, loss_cls_dn_4: 0.2291, loss_box_dn_4: 0.8260, loss_cls_dn_5: 0.2658, loss_box_dn_5: 0.8554, loss_dense_depth: 0.9231, loss: 29.6276, grad_norm: 52.5066 -2025-11-17 14:21:44,613 - mmdet - INFO - Iter [71/17500] lr: 1.280e-04, eta: 15:11:20, time: 1.597, data_time: 0.075, memory: 49163, loss_cls_0: 0.9401, loss_box_0: 1.7913, loss_cns_0: 0.6151, loss_yns_0: 0.1580, loss_cls_1: 0.9866, loss_box_1: 2.0041, loss_cns_1: 0.6257, loss_yns_1: 0.1593, loss_cls_2: 1.0283, loss_box_2: 1.9571, loss_cns_2: 0.6423, loss_yns_2: 0.1594, loss_cls_3: 1.0450, loss_box_3: 1.9454, loss_cns_3: 0.6453, loss_yns_3: 0.1593, loss_cls_4: 1.0431, loss_box_4: 1.9473, loss_cns_4: 0.6478, loss_yns_4: 0.1571, loss_cls_5: 1.1012, loss_box_5: 1.9649, loss_cns_5: 0.6479, loss_yns_5: 0.1608, loss_cls_dn_0: 0.2810, loss_box_dn_0: 0.8561, loss_cls_dn_1: 0.2056, loss_box_dn_1: 0.8512, loss_cls_dn_2: 0.2206, loss_box_dn_2: 0.8377, loss_cls_dn_3: 0.2196, loss_box_dn_3: 0.8495, loss_cls_dn_4: 0.2311, loss_box_dn_4: 0.8703, loss_cls_dn_5: 0.2891, loss_box_dn_5: 0.8990, loss_dense_depth: 0.8750, loss: 30.0178, grad_norm: 53.5567 -2025-11-17 14:21:46,104 - mmdet - INFO - Iter [72/17500] lr: 1.284e-04, eta: 15:04:39, time: 1.492, data_time: 0.078, memory: 49163, loss_cls_0: 0.9405, loss_box_0: 1.7771, loss_cns_0: 0.6131, loss_yns_0: 0.1577, loss_cls_1: 0.9895, loss_box_1: 1.9986, loss_cns_1: 0.6253, loss_yns_1: 0.1603, loss_cls_2: 1.0325, loss_box_2: 1.9430, loss_cns_2: 0.6439, loss_yns_2: 0.1588, loss_cls_3: 1.0644, loss_box_3: 1.9425, loss_cns_3: 0.6451, loss_yns_3: 0.1587, loss_cls_4: 1.0574, loss_box_4: 1.9241, loss_cns_4: 0.6468, loss_yns_4: 0.1570, loss_cls_5: 1.1146, loss_box_5: 1.9484, loss_cns_5: 0.6454, loss_yns_5: 0.1575, loss_cls_dn_0: 0.2819, loss_box_dn_0: 0.8461, loss_cls_dn_1: 0.2116, loss_box_dn_1: 0.8582, loss_cls_dn_2: 0.2235, loss_box_dn_2: 0.8504, loss_cls_dn_3: 0.2261, loss_box_dn_3: 0.8630, loss_cls_dn_4: 0.2311, loss_box_dn_4: 0.8888, loss_cls_dn_5: 0.2877, loss_box_dn_5: 0.9148, loss_dense_depth: 0.9093, loss: 30.0948, grad_norm: 62.6343 -2025-11-17 14:21:47,592 - mmdet - INFO - Iter [73/17500] lr: 1.288e-04, eta: 14:58:07, time: 1.488, data_time: 0.079, memory: 49163, loss_cls_0: 0.9543, loss_box_0: 1.7947, loss_cns_0: 0.6054, loss_yns_0: 0.1552, loss_cls_1: 1.0017, loss_box_1: 2.0359, loss_cns_1: 0.6193, loss_yns_1: 0.1597, loss_cls_2: 1.0374, loss_box_2: 1.9681, loss_cns_2: 0.6392, loss_yns_2: 0.1604, loss_cls_3: 1.0572, loss_box_3: 1.9607, loss_cns_3: 0.6435, loss_yns_3: 0.1591, loss_cls_4: 1.0589, loss_box_4: 1.9614, loss_cns_4: 0.6418, loss_yns_4: 0.1586, loss_cls_5: 1.0936, loss_box_5: 1.9790, loss_cns_5: 0.6427, loss_yns_5: 0.1586, loss_cls_dn_0: 0.2813, loss_box_dn_0: 0.8451, loss_cls_dn_1: 0.2100, loss_box_dn_1: 0.8860, loss_cls_dn_2: 0.2201, loss_box_dn_2: 0.8638, loss_cls_dn_3: 0.2234, loss_box_dn_3: 0.8747, loss_cls_dn_4: 0.2215, loss_box_dn_4: 0.9057, loss_cls_dn_5: 0.2698, loss_box_dn_5: 0.9321, loss_dense_depth: 0.8836, loss: 30.2630, grad_norm: 44.4973 -2025-11-17 14:21:49,079 - mmdet - INFO - Iter [74/17500] lr: 1.292e-04, eta: 14:51:46, time: 1.486, data_time: 0.078, memory: 49163, loss_cls_0: 0.9490, loss_box_0: 1.8131, loss_cns_0: 0.6058, loss_yns_0: 0.1565, loss_cls_1: 0.9951, loss_box_1: 2.0975, loss_cns_1: 0.6167, loss_yns_1: 0.1609, loss_cls_2: 1.0375, loss_box_2: 2.0244, loss_cns_2: 0.6347, loss_yns_2: 0.1631, loss_cls_3: 1.0390, loss_box_3: 2.0000, loss_cns_3: 0.6452, loss_yns_3: 0.1614, loss_cls_4: 1.0622, loss_box_4: 2.0092, loss_cns_4: 0.6416, loss_yns_4: 0.1598, loss_cls_5: 1.0570, loss_box_5: 2.0257, loss_cns_5: 0.6435, loss_yns_5: 0.1604, loss_cls_dn_0: 0.2782, loss_box_dn_0: 0.8476, loss_cls_dn_1: 0.2033, loss_box_dn_1: 0.9339, loss_cls_dn_2: 0.2141, loss_box_dn_2: 0.9006, loss_cls_dn_3: 0.2193, loss_box_dn_3: 0.9063, loss_cls_dn_4: 0.2174, loss_box_dn_4: 0.9270, loss_cls_dn_5: 0.2411, loss_box_dn_5: 0.9499, loss_dense_depth: 0.8837, loss: 30.5817, grad_norm: 55.4082 -2025-11-17 14:21:50,609 - mmdet - INFO - Iter [75/17500] lr: 1.296e-04, eta: 14:45:45, time: 1.529, data_time: 0.079, memory: 49163, loss_cls_0: 0.9522, loss_box_0: 1.8366, loss_cns_0: 0.6052, loss_yns_0: 0.1596, loss_cls_1: 1.0119, loss_box_1: 2.0787, loss_cns_1: 0.6276, loss_yns_1: 0.1636, loss_cls_2: 1.0379, loss_box_2: 2.0522, loss_cns_2: 0.6365, loss_yns_2: 0.1630, loss_cls_3: 1.0523, loss_box_3: 2.0318, loss_cns_3: 0.6466, loss_yns_3: 0.1615, loss_cls_4: 1.0788, loss_box_4: 2.0448, loss_cns_4: 0.6456, loss_yns_4: 0.1607, loss_cls_5: 1.0961, loss_box_5: 2.0644, loss_cns_5: 0.6495, loss_yns_5: 0.1611, loss_cls_dn_0: 0.2732, loss_box_dn_0: 0.8375, loss_cls_dn_1: 0.1983, loss_box_dn_1: 0.9381, loss_cls_dn_2: 0.2146, loss_box_dn_2: 0.9061, loss_cls_dn_3: 0.2188, loss_box_dn_3: 0.9055, loss_cls_dn_4: 0.2200, loss_box_dn_4: 0.9186, loss_cls_dn_5: 0.2308, loss_box_dn_5: 0.9317, loss_dense_depth: 0.8619, loss: 30.7733, grad_norm: 69.9711 -2025-11-17 14:21:52,096 - mmdet - INFO - Iter [76/17500] lr: 1.300e-04, eta: 14:39:43, time: 1.487, data_time: 0.077, memory: 49163, loss_cls_0: 0.9514, loss_box_0: 1.8206, loss_cns_0: 0.6054, loss_yns_0: 0.1602, loss_cls_1: 1.0182, loss_box_1: 2.0396, loss_cns_1: 0.6369, loss_yns_1: 0.1646, loss_cls_2: 1.0338, loss_box_2: 2.0272, loss_cns_2: 0.6429, loss_yns_2: 0.1615, loss_cls_3: 1.0506, loss_box_3: 2.0180, loss_cns_3: 0.6475, loss_yns_3: 0.1610, loss_cls_4: 1.0534, loss_box_4: 2.0189, loss_cns_4: 0.6504, loss_yns_4: 0.1606, loss_cls_5: 1.0819, loss_box_5: 2.0306, loss_cns_5: 0.6521, loss_yns_5: 0.1593, loss_cls_dn_0: 0.2721, loss_box_dn_0: 0.8388, loss_cls_dn_1: 0.1996, loss_box_dn_1: 0.8775, loss_cls_dn_2: 0.2162, loss_box_dn_2: 0.8467, loss_cls_dn_3: 0.2145, loss_box_dn_3: 0.8416, loss_cls_dn_4: 0.2154, loss_box_dn_4: 0.8486, loss_cls_dn_5: 0.2245, loss_box_dn_5: 0.8522, loss_dense_depth: 0.8693, loss: 30.2636, grad_norm: 47.2352 -2025-11-17 14:21:53,584 - mmdet - INFO - Iter [77/17500] lr: 1.304e-04, eta: 14:33:52, time: 1.490, data_time: 0.079, memory: 49163, loss_cls_0: 0.9449, loss_box_0: 1.8764, loss_cns_0: 0.6019, loss_yns_0: 0.1591, loss_cls_1: 1.0218, loss_box_1: 2.0013, loss_cns_1: 0.6400, loss_yns_1: 0.1642, loss_cls_2: 1.0531, loss_box_2: 1.9643, loss_cns_2: 0.6464, loss_yns_2: 0.1620, loss_cls_3: 1.0633, loss_box_3: 1.9792, loss_cns_3: 0.6525, loss_yns_3: 0.1627, loss_cls_4: 1.0688, loss_box_4: 1.9551, loss_cns_4: 0.6558, loss_yns_4: 0.1631, loss_cls_5: 1.0822, loss_box_5: 1.9536, loss_cns_5: 0.6539, loss_yns_5: 0.1629, loss_cls_dn_0: 0.2695, loss_box_dn_0: 0.8357, loss_cls_dn_1: 0.1999, loss_box_dn_1: 0.7965, loss_cls_dn_2: 0.2189, loss_box_dn_2: 0.7705, loss_cls_dn_3: 0.2133, loss_box_dn_3: 0.7745, loss_cls_dn_4: 0.2136, loss_box_dn_4: 0.7791, loss_cls_dn_5: 0.2253, loss_box_dn_5: 0.7862, loss_dense_depth: 0.8576, loss: 29.7291, grad_norm: 29.7545 -2025-11-17 14:21:55,064 - mmdet - INFO - Iter [78/17500] lr: 1.308e-04, eta: 14:28:07, time: 1.479, data_time: 0.076, memory: 49163, loss_cls_0: 0.9557, loss_box_0: 1.8744, loss_cns_0: 0.6100, loss_yns_0: 0.1621, loss_cls_1: 1.0434, loss_box_1: 2.0761, loss_cns_1: 0.6347, loss_yns_1: 0.1643, loss_cls_2: 1.0659, loss_box_2: 2.0412, loss_cns_2: 0.6419, loss_yns_2: 0.1642, loss_cls_3: 1.0684, loss_box_3: 2.0584, loss_cns_3: 0.6491, loss_yns_3: 0.1640, loss_cls_4: 1.0894, loss_box_4: 2.0477, loss_cns_4: 0.6508, loss_yns_4: 0.1637, loss_cls_5: 1.0867, loss_box_5: 2.0461, loss_cns_5: 0.6510, loss_yns_5: 0.1635, loss_cls_dn_0: 0.2671, loss_box_dn_0: 0.8293, loss_cls_dn_1: 0.1983, loss_box_dn_1: 0.8092, loss_cls_dn_2: 0.2181, loss_box_dn_2: 0.7936, loss_cls_dn_3: 0.2159, loss_box_dn_3: 0.8126, loss_cls_dn_4: 0.2191, loss_box_dn_4: 0.8260, loss_cls_dn_5: 0.2208, loss_box_dn_5: 0.8458, loss_dense_depth: 0.8936, loss: 30.4222, grad_norm: 53.2245 -2025-11-17 14:21:56,556 - mmdet - INFO - Iter [79/17500] lr: 1.312e-04, eta: 14:22:34, time: 1.493, data_time: 0.080, memory: 49163, loss_cls_0: 0.9233, loss_box_0: 1.8700, loss_cns_0: 0.6209, loss_yns_0: 0.1620, loss_cls_1: 1.0162, loss_box_1: 2.0778, loss_cns_1: 0.6397, loss_yns_1: 0.1649, loss_cls_2: 1.0234, loss_box_2: 2.0441, loss_cns_2: 0.6440, loss_yns_2: 0.1627, loss_cls_3: 1.0487, loss_box_3: 2.0558, loss_cns_3: 0.6470, loss_yns_3: 0.1633, loss_cls_4: 1.0617, loss_box_4: 2.0632, loss_cns_4: 0.6486, loss_yns_4: 0.1626, loss_cls_5: 1.0692, loss_box_5: 2.0587, loss_cns_5: 0.6501, loss_yns_5: 0.1634, loss_cls_dn_0: 0.2567, loss_box_dn_0: 0.8326, loss_cls_dn_1: 0.1998, loss_box_dn_1: 0.8570, loss_cls_dn_2: 0.2148, loss_box_dn_2: 0.8478, loss_cls_dn_3: 0.2178, loss_box_dn_3: 0.8730, loss_cls_dn_4: 0.2232, loss_box_dn_4: 0.8908, loss_cls_dn_5: 0.2286, loss_box_dn_5: 0.9210, loss_dense_depth: 0.8274, loss: 30.5319, grad_norm: 56.6721 -2025-11-17 14:21:58,060 - mmdet - INFO - Iter [80/17500] lr: 1.316e-04, eta: 14:17:12, time: 1.504, data_time: 0.078, memory: 49163, loss_cls_0: 0.9294, loss_box_0: 1.8693, loss_cns_0: 0.6196, loss_yns_0: 0.1602, loss_cls_1: 1.0178, loss_box_1: 2.0641, loss_cns_1: 0.6381, loss_yns_1: 0.1662, loss_cls_2: 1.0238, loss_box_2: 2.0127, loss_cns_2: 0.6466, loss_yns_2: 0.1647, loss_cls_3: 1.0456, loss_box_3: 2.0172, loss_cns_3: 0.6479, loss_yns_3: 0.1601, loss_cls_4: 1.0377, loss_box_4: 2.0209, loss_cns_4: 0.6505, loss_yns_4: 0.1604, loss_cls_5: 1.0706, loss_box_5: 2.0145, loss_cns_5: 0.6536, loss_yns_5: 0.1655, loss_cls_dn_0: 0.2559, loss_box_dn_0: 0.8272, loss_cls_dn_1: 0.2016, loss_box_dn_1: 0.8906, loss_cls_dn_2: 0.2033, loss_box_dn_2: 0.8724, loss_cls_dn_3: 0.2064, loss_box_dn_3: 0.8889, loss_cls_dn_4: 0.2108, loss_box_dn_4: 0.9004, loss_cls_dn_5: 0.2278, loss_box_dn_5: 0.9253, loss_dense_depth: 0.8469, loss: 30.4145, grad_norm: 41.1353 -2025-11-17 14:21:59,587 - mmdet - INFO - Iter [81/17500] lr: 1.320e-04, eta: 14:12:02, time: 1.528, data_time: 0.105, memory: 49163, loss_cls_0: 0.9555, loss_box_0: 1.8949, loss_cns_0: 0.6149, loss_yns_0: 0.1612, loss_cls_1: 1.0259, loss_box_1: 2.0531, loss_cns_1: 0.6339, loss_yns_1: 0.1632, loss_cls_2: 1.0832, loss_box_2: 1.9825, loss_cns_2: 0.6478, loss_yns_2: 0.1623, loss_cls_3: 1.0851, loss_box_3: 1.9734, loss_cns_3: 0.6512, loss_yns_3: 0.1605, loss_cls_4: 1.0629, loss_box_4: 1.9871, loss_cns_4: 0.6522, loss_yns_4: 0.1596, loss_cls_5: 1.0985, loss_box_5: 1.9746, loss_cns_5: 0.6543, loss_yns_5: 0.1639, loss_cls_dn_0: 0.2630, loss_box_dn_0: 0.8339, loss_cls_dn_1: 0.1966, loss_box_dn_1: 0.9174, loss_cls_dn_2: 0.2006, loss_box_dn_2: 0.8879, loss_cls_dn_3: 0.2002, loss_box_dn_3: 0.8883, loss_cls_dn_4: 0.2016, loss_box_dn_4: 0.8919, loss_cls_dn_5: 0.2285, loss_box_dn_5: 0.8983, loss_dense_depth: 0.8546, loss: 30.4642, grad_norm: 45.8926 -2025-11-17 14:22:01,114 - mmdet - INFO - Iter [82/17500] lr: 1.324e-04, eta: 14:07:00, time: 1.526, data_time: 0.105, memory: 49163, loss_cls_0: 0.9558, loss_box_0: 1.8893, loss_cns_0: 0.6117, loss_yns_0: 0.1571, loss_cls_1: 1.0251, loss_box_1: 2.0527, loss_cns_1: 0.6322, loss_yns_1: 0.1596, loss_cls_2: 1.0627, loss_box_2: 1.9984, loss_cns_2: 0.6445, loss_yns_2: 0.1575, loss_cls_3: 1.0741, loss_box_3: 1.9840, loss_cns_3: 0.6493, loss_yns_3: 0.1579, loss_cls_4: 1.0729, loss_box_4: 1.9977, loss_cns_4: 0.6480, loss_yns_4: 0.1582, loss_cls_5: 1.0933, loss_box_5: 1.9817, loss_cns_5: 0.6506, loss_yns_5: 0.1599, loss_cls_dn_0: 0.2602, loss_box_dn_0: 0.8377, loss_cls_dn_1: 0.1836, loss_box_dn_1: 0.9121, loss_cls_dn_2: 0.1945, loss_box_dn_2: 0.8847, loss_cls_dn_3: 0.1988, loss_box_dn_3: 0.8787, loss_cls_dn_4: 0.1981, loss_box_dn_4: 0.8777, loss_cls_dn_5: 0.2200, loss_box_dn_5: 0.8770, loss_dense_depth: 0.8940, loss: 30.3912, grad_norm: 50.6532 -2025-11-17 14:22:02,642 - mmdet - INFO - Iter [83/17500] lr: 1.328e-04, eta: 14:02:05, time: 1.527, data_time: 0.081, memory: 49163, loss_cls_0: 0.9490, loss_box_0: 1.8452, loss_cns_0: 0.6135, loss_yns_0: 0.1570, loss_cls_1: 1.0853, loss_box_1: 1.9964, loss_cns_1: 0.6268, loss_yns_1: 0.1586, loss_cls_2: 1.0255, loss_box_2: 1.9250, loss_cns_2: 0.6417, loss_yns_2: 0.1577, loss_cls_3: 1.0464, loss_box_3: 1.9109, loss_cns_3: 0.6484, loss_yns_3: 0.1583, loss_cls_4: 1.0544, loss_box_4: 1.9049, loss_cns_4: 0.6475, loss_yns_4: 0.1582, loss_cls_5: 1.0938, loss_box_5: 1.9059, loss_cns_5: 0.6459, loss_yns_5: 0.1592, loss_cls_dn_0: 0.2571, loss_box_dn_0: 0.8348, loss_cls_dn_1: 0.1801, loss_box_dn_1: 0.8549, loss_cls_dn_2: 0.1913, loss_box_dn_2: 0.8295, loss_cls_dn_3: 0.1954, loss_box_dn_3: 0.8322, loss_cls_dn_4: 0.1951, loss_box_dn_4: 0.8323, loss_cls_dn_5: 0.2105, loss_box_dn_5: 0.8401, loss_dense_depth: 0.8591, loss: 29.6280, grad_norm: 42.7286 -2025-11-17 14:22:04,163 - mmdet - INFO - Iter [84/17500] lr: 1.332e-04, eta: 13:57:17, time: 1.522, data_time: 0.083, memory: 49163, loss_cls_0: 0.9302, loss_box_0: 1.8222, loss_cns_0: 0.6155, loss_yns_0: 0.1556, loss_cls_1: 1.0252, loss_box_1: 1.9588, loss_cns_1: 0.6421, loss_yns_1: 0.1581, loss_cls_2: 1.0542, loss_box_2: 1.9165, loss_cns_2: 0.6490, loss_yns_2: 0.1588, loss_cls_3: 1.0690, loss_box_3: 1.9342, loss_cns_3: 0.6513, loss_yns_3: 0.1586, loss_cls_4: 1.0430, loss_box_4: 1.9292, loss_cns_4: 0.6494, loss_yns_4: 0.1581, loss_cls_5: 1.0549, loss_box_5: 1.9331, loss_cns_5: 0.6475, loss_yns_5: 0.1591, loss_cls_dn_0: 0.2538, loss_box_dn_0: 0.8210, loss_cls_dn_1: 0.1919, loss_box_dn_1: 0.7893, loss_cls_dn_2: 0.1969, loss_box_dn_2: 0.7761, loss_cls_dn_3: 0.2074, loss_box_dn_3: 0.7948, loss_cls_dn_4: 0.2014, loss_box_dn_4: 0.8060, loss_cls_dn_5: 0.2194, loss_box_dn_5: 0.8253, loss_dense_depth: 0.8313, loss: 29.3881, grad_norm: 47.5335 -2025-11-17 14:22:05,675 - mmdet - INFO - Iter [85/17500] lr: 1.336e-04, eta: 13:52:32, time: 1.512, data_time: 0.089, memory: 49163, loss_cls_0: 0.9497, loss_box_0: 1.8374, loss_cns_0: 0.6192, loss_yns_0: 0.1604, loss_cls_1: 1.1007, loss_box_1: 1.9878, loss_cns_1: 0.6296, loss_yns_1: 0.1586, loss_cls_2: 1.0907, loss_box_2: 1.9855, loss_cns_2: 0.6447, loss_yns_2: 0.1631, loss_cls_3: 1.0910, loss_box_3: 1.9910, loss_cns_3: 0.6482, loss_yns_3: 0.1624, loss_cls_4: 1.0677, loss_box_4: 1.9919, loss_cns_4: 0.6467, loss_yns_4: 0.1628, loss_cls_5: 1.0716, loss_box_5: 1.9865, loss_cns_5: 0.6477, loss_yns_5: 0.1632, loss_cls_dn_0: 0.2591, loss_box_dn_0: 0.8336, loss_cls_dn_1: 0.1994, loss_box_dn_1: 0.8410, loss_cls_dn_2: 0.2027, loss_box_dn_2: 0.8343, loss_cls_dn_3: 0.2077, loss_box_dn_3: 0.8479, loss_cls_dn_4: 0.2055, loss_box_dn_4: 0.8662, loss_cls_dn_5: 0.2223, loss_box_dn_5: 0.8819, loss_dense_depth: 0.8493, loss: 30.2090, grad_norm: 50.9319 -2025-11-17 14:22:07,180 - mmdet - INFO - Iter [86/17500] lr: 1.340e-04, eta: 13:47:54, time: 1.506, data_time: 0.087, memory: 49163, loss_cls_0: 0.9400, loss_box_0: 1.8220, loss_cns_0: 0.6207, loss_yns_0: 0.1626, loss_cls_1: 1.0533, loss_box_1: 1.9374, loss_cns_1: 0.6313, loss_yns_1: 0.1640, loss_cls_2: 1.0662, loss_box_2: 1.9043, loss_cns_2: 0.6421, loss_yns_2: 0.1651, loss_cls_3: 1.0859, loss_box_3: 1.8944, loss_cns_3: 0.6451, loss_yns_3: 0.1677, loss_cls_4: 1.0379, loss_box_4: 1.8924, loss_cns_4: 0.6448, loss_yns_4: 0.1652, loss_cls_5: 1.0591, loss_box_5: 1.8909, loss_cns_5: 0.6497, loss_yns_5: 0.1636, loss_cls_dn_0: 0.2466, loss_box_dn_0: 0.8289, loss_cls_dn_1: 0.1932, loss_box_dn_1: 0.8431, loss_cls_dn_2: 0.1969, loss_box_dn_2: 0.8321, loss_cls_dn_3: 0.1997, loss_box_dn_3: 0.8340, loss_cls_dn_4: 0.2007, loss_box_dn_4: 0.8495, loss_cls_dn_5: 0.2117, loss_box_dn_5: 0.8649, loss_dense_depth: 0.8668, loss: 29.5736, grad_norm: 38.8204 -2025-11-17 14:22:08,699 - mmdet - INFO - Iter [87/17500] lr: 1.344e-04, eta: 13:43:24, time: 1.519, data_time: 0.112, memory: 49163, loss_cls_0: 0.9119, loss_box_0: 1.8012, loss_cns_0: 0.6204, loss_yns_0: 0.1608, loss_cls_1: 0.9949, loss_box_1: 1.9233, loss_cns_1: 0.6391, loss_yns_1: 0.1631, loss_cls_2: 1.0107, loss_box_2: 1.8862, loss_cns_2: 0.6480, loss_yns_2: 0.1638, loss_cls_3: 1.0417, loss_box_3: 1.8924, loss_cns_3: 0.6537, loss_yns_3: 0.1651, loss_cls_4: 1.0283, loss_box_4: 1.8979, loss_cns_4: 0.6533, loss_yns_4: 0.1663, loss_cls_5: 1.0544, loss_box_5: 1.8982, loss_cns_5: 0.6554, loss_yns_5: 0.1615, loss_cls_dn_0: 0.2454, loss_box_dn_0: 0.8180, loss_cls_dn_1: 0.1862, loss_box_dn_1: 0.8133, loss_cls_dn_2: 0.1930, loss_box_dn_2: 0.7977, loss_cls_dn_3: 0.1958, loss_box_dn_3: 0.7966, loss_cls_dn_4: 0.2015, loss_box_dn_4: 0.8040, loss_cls_dn_5: 0.2093, loss_box_dn_5: 0.8170, loss_dense_depth: 0.8336, loss: 29.1032, grad_norm: 42.2063 -2025-11-17 14:22:10,185 - mmdet - INFO - Iter [88/17500] lr: 1.348e-04, eta: 13:38:54, time: 1.485, data_time: 0.079, memory: 49163, loss_cls_0: 0.9401, loss_box_0: 1.8084, loss_cns_0: 0.6140, loss_yns_0: 0.1577, loss_cls_1: 1.0197, loss_box_1: 1.9568, loss_cns_1: 0.6394, loss_yns_1: 0.1602, loss_cls_2: 1.0305, loss_box_2: 1.9298, loss_cns_2: 0.6478, loss_yns_2: 0.1619, loss_cls_3: 1.0523, loss_box_3: 1.9169, loss_cns_3: 0.6550, loss_yns_3: 0.1629, loss_cls_4: 1.0384, loss_box_4: 1.9246, loss_cns_4: 0.6535, loss_yns_4: 0.1611, loss_cls_5: 1.0624, loss_box_5: 1.9102, loss_cns_5: 0.6551, loss_yns_5: 0.1585, loss_cls_dn_0: 0.2500, loss_box_dn_0: 0.8297, loss_cls_dn_1: 0.1831, loss_box_dn_1: 0.8078, loss_cls_dn_2: 0.1958, loss_box_dn_2: 0.7870, loss_cls_dn_3: 0.1991, loss_box_dn_3: 0.7813, loss_cls_dn_4: 0.1992, loss_box_dn_4: 0.7804, loss_cls_dn_5: 0.2109, loss_box_dn_5: 0.7829, loss_dense_depth: 0.8237, loss: 29.2479, grad_norm: 55.7770 -2025-11-17 14:22:11,691 - mmdet - INFO - Iter [89/17500] lr: 1.352e-04, eta: 13:34:34, time: 1.508, data_time: 0.093, memory: 49163, loss_cls_0: 0.9276, loss_box_0: 1.8168, loss_cns_0: 0.6143, loss_yns_0: 0.1568, loss_cls_1: 1.0327, loss_box_1: 1.9422, loss_cns_1: 0.6344, loss_yns_1: 0.1599, loss_cls_2: 1.0357, loss_box_2: 1.9090, loss_cns_2: 0.6471, loss_yns_2: 0.1603, loss_cls_3: 1.0471, loss_box_3: 1.8795, loss_cns_3: 0.6497, loss_yns_3: 0.1576, loss_cls_4: 1.0180, loss_box_4: 1.8898, loss_cns_4: 0.6488, loss_yns_4: 0.1577, loss_cls_5: 1.0438, loss_box_5: 1.8752, loss_cns_5: 0.6514, loss_yns_5: 0.1581, loss_cls_dn_0: 0.2502, loss_box_dn_0: 0.8289, loss_cls_dn_1: 0.1859, loss_box_dn_1: 0.7916, loss_cls_dn_2: 0.2014, loss_box_dn_2: 0.7624, loss_cls_dn_3: 0.2006, loss_box_dn_3: 0.7552, loss_cls_dn_4: 0.1972, loss_box_dn_4: 0.7547, loss_cls_dn_5: 0.2085, loss_box_dn_5: 0.7546, loss_dense_depth: 0.8238, loss: 28.9288, grad_norm: 49.2108 -2025-11-17 14:22:13,183 - mmdet - INFO - Iter [90/17500] lr: 1.356e-04, eta: 13:30:16, time: 1.491, data_time: 0.079, memory: 49163, loss_cls_0: 0.9420, loss_box_0: 1.8090, loss_cns_0: 0.6148, loss_yns_0: 0.1554, loss_cls_1: 1.0274, loss_box_1: 2.0011, loss_cns_1: 0.6346, loss_yns_1: 0.1579, loss_cls_2: 1.0446, loss_box_2: 1.9608, loss_cns_2: 0.6438, loss_yns_2: 0.1602, loss_cls_3: 1.0573, loss_box_3: 1.9216, loss_cns_3: 0.6478, loss_yns_3: 0.1586, loss_cls_4: 1.0310, loss_box_4: 1.9130, loss_cns_4: 0.6493, loss_yns_4: 0.1563, loss_cls_5: 1.0470, loss_box_5: 1.9225, loss_cns_5: 0.6528, loss_yns_5: 0.1603, loss_cls_dn_0: 0.2484, loss_box_dn_0: 0.8387, loss_cls_dn_1: 0.1860, loss_box_dn_1: 0.7865, loss_cls_dn_2: 0.1953, loss_box_dn_2: 0.7646, loss_cls_dn_3: 0.1967, loss_box_dn_3: 0.7562, loss_cls_dn_4: 0.1970, loss_box_dn_4: 0.7632, loss_cls_dn_5: 0.2055, loss_box_dn_5: 0.7777, loss_dense_depth: 0.8531, loss: 29.2378, grad_norm: 35.8318 -2025-11-17 14:22:14,717 - mmdet - INFO - Iter [91/17500] lr: 1.360e-04, eta: 13:26:12, time: 1.532, data_time: 0.083, memory: 49163, loss_cls_0: 0.9216, loss_box_0: 1.7573, loss_cns_0: 0.6099, loss_yns_0: 0.1491, loss_cls_1: 1.0052, loss_box_1: 1.9848, loss_cns_1: 0.6347, loss_yns_1: 0.1568, loss_cls_2: 1.0219, loss_box_2: 1.9287, loss_cns_2: 0.6448, loss_yns_2: 0.1567, loss_cls_3: 1.0416, loss_box_3: 1.9327, loss_cns_3: 0.6497, loss_yns_3: 0.1578, loss_cls_4: 1.0519, loss_box_4: 1.9312, loss_cns_4: 0.6494, loss_yns_4: 0.1543, loss_cls_5: 1.0828, loss_box_5: 1.9348, loss_cns_5: 0.6544, loss_yns_5: 0.1579, loss_cls_dn_0: 0.2427, loss_box_dn_0: 0.8173, loss_cls_dn_1: 0.1873, loss_box_dn_1: 0.8053, loss_cls_dn_2: 0.1926, loss_box_dn_2: 0.7872, loss_cls_dn_3: 0.2004, loss_box_dn_3: 0.7977, loss_cls_dn_4: 0.2068, loss_box_dn_4: 0.8200, loss_cls_dn_5: 0.2135, loss_box_dn_5: 0.8427, loss_dense_depth: 0.8101, loss: 29.2935, grad_norm: 46.8888 -2025-11-17 14:22:16,215 - mmdet - INFO - Iter [92/17500] lr: 1.364e-04, eta: 13:22:08, time: 1.500, data_time: 0.080, memory: 49163, loss_cls_0: 0.9266, loss_box_0: 1.7468, loss_cns_0: 0.6086, loss_yns_0: 0.1485, loss_cls_1: 0.9994, loss_box_1: 2.0019, loss_cns_1: 0.6365, loss_yns_1: 0.1555, loss_cls_2: 1.0127, loss_box_2: 1.9594, loss_cns_2: 0.6466, loss_yns_2: 0.1546, loss_cls_3: 1.0342, loss_box_3: 1.9865, loss_cns_3: 0.6534, loss_yns_3: 0.1580, loss_cls_4: 1.0389, loss_box_4: 1.9939, loss_cns_4: 0.6502, loss_yns_4: 0.1555, loss_cls_5: 1.0633, loss_box_5: 1.9892, loss_cns_5: 0.6516, loss_yns_5: 0.1555, loss_cls_dn_0: 0.2437, loss_box_dn_0: 0.8168, loss_cls_dn_1: 0.1884, loss_box_dn_1: 0.8115, loss_cls_dn_2: 0.1911, loss_box_dn_2: 0.8005, loss_cls_dn_3: 0.2011, loss_box_dn_3: 0.8217, loss_cls_dn_4: 0.2107, loss_box_dn_4: 0.8451, loss_cls_dn_5: 0.2200, loss_box_dn_5: 0.8659, loss_dense_depth: 0.8298, loss: 29.5737, grad_norm: 53.0519 -2025-11-17 14:22:17,709 - mmdet - INFO - Iter [93/17500] lr: 1.368e-04, eta: 13:18:07, time: 1.494, data_time: 0.080, memory: 49163, loss_cls_0: 0.9135, loss_box_0: 1.7710, loss_cns_0: 0.6205, loss_yns_0: 0.1516, loss_cls_1: 1.0163, loss_box_1: 1.9710, loss_cns_1: 0.6413, loss_yns_1: 0.1531, loss_cls_2: 1.0177, loss_box_2: 1.9204, loss_cns_2: 0.6512, loss_yns_2: 0.1545, loss_cls_3: 1.0406, loss_box_3: 1.9336, loss_cns_3: 0.6566, loss_yns_3: 0.1559, loss_cls_4: 1.0339, loss_box_4: 1.9424, loss_cns_4: 0.6538, loss_yns_4: 0.1539, loss_cls_5: 1.0424, loss_box_5: 1.9345, loss_cns_5: 0.6566, loss_yns_5: 0.1553, loss_cls_dn_0: 0.2410, loss_box_dn_0: 0.8129, loss_cls_dn_1: 0.1931, loss_box_dn_1: 0.8200, loss_cls_dn_2: 0.1957, loss_box_dn_2: 0.8059, loss_cls_dn_3: 0.2053, loss_box_dn_3: 0.8158, loss_cls_dn_4: 0.2124, loss_box_dn_4: 0.8286, loss_cls_dn_5: 0.2281, loss_box_dn_5: 0.8425, loss_dense_depth: 0.7966, loss: 29.3399, grad_norm: 46.3246 -2025-11-17 14:22:19,189 - mmdet - INFO - Iter [94/17500] lr: 1.372e-04, eta: 13:14:09, time: 1.480, data_time: 0.080, memory: 49163, loss_cls_0: 0.9280, loss_box_0: 1.8005, loss_cns_0: 0.6209, loss_yns_0: 0.1548, loss_cls_1: 1.0330, loss_box_1: 1.9798, loss_cns_1: 0.6336, loss_yns_1: 0.1551, loss_cls_2: 1.0181, loss_box_2: 1.9335, loss_cns_2: 0.6441, loss_yns_2: 0.1560, loss_cls_3: 1.0584, loss_box_3: 1.9118, loss_cns_3: 0.6537, loss_yns_3: 0.1569, loss_cls_4: 1.0423, loss_box_4: 1.9243, loss_cns_4: 0.6505, loss_yns_4: 0.1581, loss_cls_5: 1.0665, loss_box_5: 1.9065, loss_cns_5: 0.6529, loss_yns_5: 0.1575, loss_cls_dn_0: 0.2391, loss_box_dn_0: 0.8073, loss_cls_dn_1: 0.1929, loss_box_dn_1: 0.7990, loss_cls_dn_2: 0.1938, loss_box_dn_2: 0.7737, loss_cls_dn_3: 0.2004, loss_box_dn_3: 0.7682, loss_cls_dn_4: 0.1959, loss_box_dn_4: 0.7720, loss_cls_dn_5: 0.2181, loss_box_dn_5: 0.7777, loss_dense_depth: 0.7937, loss: 29.1286, grad_norm: 36.7754 -2025-11-17 14:22:20,738 - mmdet - INFO - Iter [95/17500] lr: 1.376e-04, eta: 13:10:28, time: 1.549, data_time: 0.087, memory: 49163, loss_cls_0: 0.9273, loss_box_0: 1.8188, loss_cns_0: 0.6189, loss_yns_0: 0.1555, loss_cls_1: 1.0191, loss_box_1: 1.9434, loss_cns_1: 0.6377, loss_yns_1: 0.1550, loss_cls_2: 1.0285, loss_box_2: 1.9020, loss_cns_2: 0.6507, loss_yns_2: 0.1563, loss_cls_3: 1.1042, loss_box_3: 1.8666, loss_cns_3: 0.6545, loss_yns_3: 0.1590, loss_cls_4: 1.0354, loss_box_4: 1.8851, loss_cns_4: 0.6552, loss_yns_4: 0.1574, loss_cls_5: 1.0667, loss_box_5: 1.8768, loss_cns_5: 0.6572, loss_yns_5: 0.1561, loss_cls_dn_0: 0.2384, loss_box_dn_0: 0.8057, loss_cls_dn_1: 0.1877, loss_box_dn_1: 0.7706, loss_cls_dn_2: 0.1916, loss_box_dn_2: 0.7438, loss_cls_dn_3: 0.1939, loss_box_dn_3: 0.7379, loss_cls_dn_4: 0.1920, loss_box_dn_4: 0.7411, loss_cls_dn_5: 0.2079, loss_box_dn_5: 0.7442, loss_dense_depth: 0.8398, loss: 28.8821, grad_norm: 45.0664 -2025-11-17 14:22:22,221 - mmdet - INFO - Iter [96/17500] lr: 1.380e-04, eta: 13:06:40, time: 1.482, data_time: 0.076, memory: 49163, loss_cls_0: 0.9449, loss_box_0: 1.8399, loss_cns_0: 0.6140, loss_yns_0: 0.1563, loss_cls_1: 1.0256, loss_box_1: 1.9553, loss_cns_1: 0.6315, loss_yns_1: 0.1562, loss_cls_2: 1.0092, loss_box_2: 1.9237, loss_cns_2: 0.6450, loss_yns_2: 0.1550, loss_cls_3: 1.0428, loss_box_3: 1.9243, loss_cns_3: 0.6472, loss_yns_3: 0.1609, loss_cls_4: 1.0471, loss_box_4: 1.9293, loss_cns_4: 0.6509, loss_yns_4: 0.1588, loss_cls_5: 1.0648, loss_box_5: 1.9234, loss_cns_5: 0.6498, loss_yns_5: 0.1575, loss_cls_dn_0: 0.2465, loss_box_dn_0: 0.8124, loss_cls_dn_1: 0.1864, loss_box_dn_1: 0.7841, loss_cls_dn_2: 0.1903, loss_box_dn_2: 0.7626, loss_cls_dn_3: 0.1910, loss_box_dn_3: 0.7697, loss_cls_dn_4: 0.1977, loss_box_dn_4: 0.7727, loss_cls_dn_5: 0.2058, loss_box_dn_5: 0.7728, loss_dense_depth: 0.8390, loss: 29.1445, grad_norm: 50.7653 -2025-11-17 14:22:23,706 - mmdet - INFO - Iter [97/17500] lr: 1.384e-04, eta: 13:02:58, time: 1.486, data_time: 0.078, memory: 49163, loss_cls_0: 0.9437, loss_box_0: 1.8242, loss_cns_0: 0.6189, loss_yns_0: 0.1558, loss_cls_1: 1.0042, loss_box_1: 1.9410, loss_cns_1: 0.6317, loss_yns_1: 0.1552, loss_cls_2: 1.0166, loss_box_2: 1.9052, loss_cns_2: 0.6399, loss_yns_2: 0.1543, loss_cls_3: 1.0337, loss_box_3: 1.9253, loss_cns_3: 0.6497, loss_yns_3: 0.1605, loss_cls_4: 1.0377, loss_box_4: 1.9255, loss_cns_4: 0.6502, loss_yns_4: 0.1568, loss_cls_5: 1.0554, loss_box_5: 1.8938, loss_cns_5: 0.6543, loss_yns_5: 0.1548, loss_cls_dn_0: 0.2482, loss_box_dn_0: 0.8131, loss_cls_dn_1: 0.1867, loss_box_dn_1: 0.7784, loss_cls_dn_2: 0.1928, loss_box_dn_2: 0.7622, loss_cls_dn_3: 0.1947, loss_box_dn_3: 0.7780, loss_cls_dn_4: 0.1952, loss_box_dn_4: 0.7835, loss_cls_dn_5: 0.1998, loss_box_dn_5: 0.7799, loss_dense_depth: 0.8324, loss: 29.0330, grad_norm: 48.9200 -2025-11-17 14:22:25,197 - mmdet - INFO - Iter [98/17500] lr: 1.388e-04, eta: 12:59:20, time: 1.491, data_time: 0.079, memory: 49163, loss_cls_0: 0.9226, loss_box_0: 1.8145, loss_cns_0: 0.6246, loss_yns_0: 0.1563, loss_cls_1: 0.9847, loss_box_1: 1.8879, loss_cns_1: 0.6377, loss_yns_1: 0.1549, loss_cls_2: 1.0127, loss_box_2: 1.8366, loss_cns_2: 0.6434, loss_yns_2: 0.1535, loss_cls_3: 1.0420, loss_box_3: 1.8629, loss_cns_3: 0.6538, loss_yns_3: 0.1567, loss_cls_4: 1.0227, loss_box_4: 1.8579, loss_cns_4: 0.6489, loss_yns_4: 0.1566, loss_cls_5: 1.0376, loss_box_5: 1.8409, loss_cns_5: 0.6515, loss_yns_5: 0.1570, loss_cls_dn_0: 0.2418, loss_box_dn_0: 0.8126, loss_cls_dn_1: 0.1838, loss_box_dn_1: 0.8016, loss_cls_dn_2: 0.1921, loss_box_dn_2: 0.7874, loss_cls_dn_3: 0.1970, loss_box_dn_3: 0.8072, loss_cls_dn_4: 0.1937, loss_box_dn_4: 0.8157, loss_cls_dn_5: 0.2076, loss_box_dn_5: 0.8201, loss_dense_depth: 0.8131, loss: 28.7917, grad_norm: 39.8320 -2025-11-17 14:22:26,686 - mmdet - INFO - Iter [99/17500] lr: 1.392e-04, eta: 12:55:47, time: 1.489, data_time: 0.079, memory: 49163, loss_cls_0: 0.9234, loss_box_0: 1.7955, loss_cns_0: 0.6222, loss_yns_0: 0.1548, loss_cls_1: 1.0629, loss_box_1: 1.9358, loss_cns_1: 0.6286, loss_yns_1: 0.1543, loss_cls_2: 1.0199, loss_box_2: 1.8808, loss_cns_2: 0.6405, loss_yns_2: 0.1545, loss_cls_3: 1.0573, loss_box_3: 1.8862, loss_cns_3: 0.6439, loss_yns_3: 0.1580, loss_cls_4: 1.0690, loss_box_4: 1.9104, loss_cns_4: 0.6421, loss_yns_4: 0.1560, loss_cls_5: 1.0536, loss_box_5: 1.8976, loss_cns_5: 0.6459, loss_yns_5: 0.1589, loss_cls_dn_0: 0.2402, loss_box_dn_0: 0.8063, loss_cls_dn_1: 0.1837, loss_box_dn_1: 0.8043, loss_cls_dn_2: 0.1851, loss_box_dn_2: 0.7910, loss_cls_dn_3: 0.1931, loss_box_dn_3: 0.8019, loss_cls_dn_4: 0.1903, loss_box_dn_4: 0.8253, loss_cls_dn_5: 0.2140, loss_box_dn_5: 0.8271, loss_dense_depth: 0.8492, loss: 29.1638, grad_norm: 54.8397 -2025-11-17 14:22:28,177 - mmdet - INFO - Iter [100/17500] lr: 1.396e-04, eta: 12:52:18, time: 1.492, data_time: 0.080, memory: 49163, loss_cls_0: 0.9180, loss_box_0: 1.8058, loss_cns_0: 0.6190, loss_yns_0: 0.1547, loss_cls_1: 1.0342, loss_box_1: 1.9609, loss_cns_1: 0.6265, loss_yns_1: 0.1539, loss_cls_2: 1.0072, loss_box_2: 1.8974, loss_cns_2: 0.6440, loss_yns_2: 0.1564, loss_cls_3: 1.0798, loss_box_3: 1.8785, loss_cns_3: 0.6479, loss_yns_3: 0.1625, loss_cls_4: 1.0279, loss_box_4: 1.9394, loss_cns_4: 0.6482, loss_yns_4: 0.1564, loss_cls_5: 1.0450, loss_box_5: 1.8993, loss_cns_5: 0.6478, loss_yns_5: 0.1562, loss_cls_dn_0: 0.2406, loss_box_dn_0: 0.8172, loss_cls_dn_1: 0.1843, loss_box_dn_1: 0.8162, loss_cls_dn_2: 0.1828, loss_box_dn_2: 0.8013, loss_cls_dn_3: 0.1930, loss_box_dn_3: 0.8045, loss_cls_dn_4: 0.1898, loss_box_dn_4: 0.8395, loss_cls_dn_5: 0.2078, loss_box_dn_5: 0.8273, loss_dense_depth: 0.8183, loss: 29.1895, grad_norm: 48.9197 -2025-11-17 14:22:29,705 - mmdet - INFO - Iter [101/17500] lr: 1.400e-04, eta: 12:49:00, time: 1.528, data_time: 0.105, memory: 49163, loss_cls_0: 0.9375, loss_box_0: 1.7774, loss_cns_0: 0.6141, loss_yns_0: 0.1561, loss_cls_1: 0.9846, loss_box_1: 1.9614, loss_cns_1: 0.6309, loss_yns_1: 0.1556, loss_cls_2: 1.0201, loss_box_2: 1.9014, loss_cns_2: 0.6465, loss_yns_2: 0.1575, loss_cls_3: 1.0860, loss_box_3: 1.8935, loss_cns_3: 0.6512, loss_yns_3: 0.1639, loss_cls_4: 1.0411, loss_box_4: 1.9393, loss_cns_4: 0.6522, loss_yns_4: 0.1586, loss_cls_5: 1.0497, loss_box_5: 1.8961, loss_cns_5: 0.6496, loss_yns_5: 0.1585, loss_cls_dn_0: 0.2423, loss_box_dn_0: 0.8094, loss_cls_dn_1: 0.1830, loss_box_dn_1: 0.8231, loss_cls_dn_2: 0.1837, loss_box_dn_2: 0.7989, loss_cls_dn_3: 0.1949, loss_box_dn_3: 0.8003, loss_cls_dn_4: 0.1942, loss_box_dn_4: 0.8228, loss_cls_dn_5: 0.2033, loss_box_dn_5: 0.8031, loss_dense_depth: 0.8462, loss: 29.1876, grad_norm: 56.3358 -2025-11-17 14:22:31,242 - mmdet - INFO - Iter [102/17500] lr: 1.404e-04, eta: 12:45:47, time: 1.537, data_time: 0.102, memory: 49163, loss_cls_0: 0.9185, loss_box_0: 1.7972, loss_cns_0: 0.6146, loss_yns_0: 0.1582, loss_cls_1: 0.9974, loss_box_1: 1.9682, loss_cns_1: 0.6275, loss_yns_1: 0.1576, loss_cls_2: 1.0171, loss_box_2: 1.9027, loss_cns_2: 0.6432, loss_yns_2: 0.1587, loss_cls_3: 1.0477, loss_box_3: 1.8890, loss_cns_3: 0.6463, loss_yns_3: 0.1606, loss_cls_4: 1.0368, loss_box_4: 1.9144, loss_cns_4: 0.6492, loss_yns_4: 0.1591, loss_cls_5: 1.0479, loss_box_5: 1.8932, loss_cns_5: 0.6491, loss_yns_5: 0.1591, loss_cls_dn_0: 0.2440, loss_box_dn_0: 0.8131, loss_cls_dn_1: 0.1835, loss_box_dn_1: 0.7801, loss_cls_dn_2: 0.1846, loss_box_dn_2: 0.7521, loss_cls_dn_3: 0.1923, loss_box_dn_3: 0.7530, loss_cls_dn_4: 0.1962, loss_box_dn_4: 0.7621, loss_cls_dn_5: 0.2017, loss_box_dn_5: 0.7541, loss_dense_depth: 0.8211, loss: 28.8513, grad_norm: 29.9505 -2025-11-17 14:22:32,801 - mmdet - INFO - Iter [103/17500] lr: 1.408e-04, eta: 12:42:42, time: 1.558, data_time: 0.078, memory: 49163, loss_cls_0: 0.9528, loss_box_0: 1.8117, loss_cns_0: 0.6212, loss_yns_0: 0.1608, loss_cls_1: 0.9783, loss_box_1: 1.9775, loss_cns_1: 0.6308, loss_yns_1: 0.1581, loss_cls_2: 1.0375, loss_box_2: 1.9304, loss_cns_2: 0.6445, loss_yns_2: 0.1599, loss_cls_3: 1.0823, loss_box_3: 1.8983, loss_cns_3: 0.6496, loss_yns_3: 0.1596, loss_cls_4: 1.0180, loss_box_4: 1.9043, loss_cns_4: 0.6453, loss_yns_4: 0.1594, loss_cls_5: 1.0320, loss_box_5: 1.9407, loss_cns_5: 0.6462, loss_yns_5: 0.1597, loss_cls_dn_0: 0.2380, loss_box_dn_0: 0.8035, loss_cls_dn_1: 0.1785, loss_box_dn_1: 0.7777, loss_cls_dn_2: 0.1813, loss_box_dn_2: 0.7476, loss_cls_dn_3: 0.1848, loss_box_dn_3: 0.7438, loss_cls_dn_4: 0.1855, loss_box_dn_4: 0.7451, loss_cls_dn_5: 0.1940, loss_box_dn_5: 0.7658, loss_dense_depth: 0.8899, loss: 28.9944, grad_norm: 59.5053 -2025-11-17 14:22:34,308 - mmdet - INFO - Iter [104/17500] lr: 1.412e-04, eta: 12:39:31, time: 1.508, data_time: 0.080, memory: 49163, loss_cls_0: 0.9154, loss_box_0: 1.8083, loss_cns_0: 0.6183, loss_yns_0: 0.1588, loss_cls_1: 0.9761, loss_box_1: 1.9848, loss_cns_1: 0.6344, loss_yns_1: 0.1556, loss_cls_2: 1.0188, loss_box_2: 1.9442, loss_cns_2: 0.6476, loss_yns_2: 0.1572, loss_cls_3: 1.0662, loss_box_3: 1.9074, loss_cns_3: 0.6550, loss_yns_3: 0.1593, loss_cls_4: 1.0092, loss_box_4: 1.8929, loss_cns_4: 0.6529, loss_yns_4: 0.1583, loss_cls_5: 1.0254, loss_box_5: 1.9468, loss_cns_5: 0.6528, loss_yns_5: 0.1585, loss_cls_dn_0: 0.2404, loss_box_dn_0: 0.8157, loss_cls_dn_1: 0.1761, loss_box_dn_1: 0.7787, loss_cls_dn_2: 0.1783, loss_box_dn_2: 0.7487, loss_cls_dn_3: 0.1832, loss_box_dn_3: 0.7450, loss_cls_dn_4: 0.1817, loss_box_dn_4: 0.7420, loss_cls_dn_5: 0.1919, loss_box_dn_5: 0.7733, loss_dense_depth: 0.8824, loss: 28.9415, grad_norm: 52.5898 -2025-11-17 14:22:35,842 - mmdet - INFO - Iter [105/17500] lr: 1.416e-04, eta: 12:36:29, time: 1.534, data_time: 0.085, memory: 49163, loss_cls_0: 0.8912, loss_box_0: 1.7547, loss_cns_0: 0.6207, loss_yns_0: 0.1581, loss_cls_1: 0.9865, loss_box_1: 1.9279, loss_cns_1: 0.6382, loss_yns_1: 0.1578, loss_cls_2: 0.9914, loss_box_2: 1.8944, loss_cns_2: 0.6490, loss_yns_2: 0.1563, loss_cls_3: 0.9999, loss_box_3: 1.8761, loss_cns_3: 0.6510, loss_yns_3: 0.1618, loss_cls_4: 0.9997, loss_box_4: 1.8693, loss_cns_4: 0.6510, loss_yns_4: 0.1571, loss_cls_5: 1.0128, loss_box_5: 1.9011, loss_cns_5: 0.6519, loss_yns_5: 0.1601, loss_cls_dn_0: 0.2372, loss_box_dn_0: 0.8124, loss_cls_dn_1: 0.1754, loss_box_dn_1: 0.8050, loss_cls_dn_2: 0.1783, loss_box_dn_2: 0.7834, loss_cls_dn_3: 0.1844, loss_box_dn_3: 0.7830, loss_cls_dn_4: 0.1865, loss_box_dn_4: 0.7830, loss_cls_dn_5: 0.1965, loss_box_dn_5: 0.8066, loss_dense_depth: 0.8099, loss: 28.6595, grad_norm: 61.6568 -2025-11-17 14:22:37,353 - mmdet - INFO - Iter [106/17500] lr: 1.420e-04, eta: 12:33:26, time: 1.510, data_time: 0.083, memory: 49163, loss_cls_0: 0.9147, loss_box_0: 1.7309, loss_cns_0: 0.6119, loss_yns_0: 0.1572, loss_cls_1: 0.9963, loss_box_1: 1.9035, loss_cns_1: 0.6399, loss_yns_1: 0.1580, loss_cls_2: 1.0159, loss_box_2: 1.8873, loss_cns_2: 0.6472, loss_yns_2: 0.1570, loss_cls_3: 1.0199, loss_box_3: 1.8672, loss_cns_3: 0.6505, loss_yns_3: 0.1616, loss_cls_4: 1.0025, loss_box_4: 1.8550, loss_cns_4: 0.6496, loss_yns_4: 0.1567, loss_cls_5: 1.0175, loss_box_5: 1.8690, loss_cns_5: 0.6509, loss_yns_5: 0.1587, loss_cls_dn_0: 0.2355, loss_box_dn_0: 0.7956, loss_cls_dn_1: 0.1765, loss_box_dn_1: 0.7878, loss_cls_dn_2: 0.1836, loss_box_dn_2: 0.7756, loss_cls_dn_3: 0.1904, loss_box_dn_3: 0.7767, loss_cls_dn_4: 0.1863, loss_box_dn_4: 0.7772, loss_cls_dn_5: 0.1942, loss_box_dn_5: 0.7966, loss_dense_depth: 0.8540, loss: 28.6088, grad_norm: 51.5143 -2025-11-17 14:22:38,868 - mmdet - INFO - Iter [107/17500] lr: 1.424e-04, eta: 12:30:27, time: 1.514, data_time: 0.109, memory: 49163, loss_cls_0: 0.8925, loss_box_0: 1.7642, loss_cns_0: 0.6129, loss_yns_0: 0.1574, loss_cls_1: 0.9868, loss_box_1: 1.8423, loss_cns_1: 0.6425, loss_yns_1: 0.1577, loss_cls_2: 1.0114, loss_box_2: 1.8144, loss_cns_2: 0.6509, loss_yns_2: 0.1572, loss_cls_3: 1.0272, loss_box_3: 1.8033, loss_cns_3: 0.6519, loss_yns_3: 0.1580, loss_cls_4: 1.0354, loss_box_4: 1.8061, loss_cns_4: 0.6532, loss_yns_4: 0.1570, loss_cls_5: 1.0528, loss_box_5: 1.8114, loss_cns_5: 0.6519, loss_yns_5: 0.1582, loss_cls_dn_0: 0.2371, loss_box_dn_0: 0.8223, loss_cls_dn_1: 0.1785, loss_box_dn_1: 0.7825, loss_cls_dn_2: 0.1829, loss_box_dn_2: 0.7768, loss_cls_dn_3: 0.1874, loss_box_dn_3: 0.7910, loss_cls_dn_4: 0.1849, loss_box_dn_4: 0.8041, loss_cls_dn_5: 0.1906, loss_box_dn_5: 0.8241, loss_dense_depth: 0.8380, loss: 28.4569, grad_norm: 49.7816 -2025-11-17 14:22:40,338 - mmdet - INFO - Iter [108/17500] lr: 1.428e-04, eta: 12:27:24, time: 1.471, data_time: 0.071, memory: 49163, loss_cls_0: 0.9156, loss_box_0: 1.7973, loss_cns_0: 0.6179, loss_yns_0: 0.1594, loss_cls_1: 0.9979, loss_box_1: 1.8859, loss_cns_1: 0.6416, loss_yns_1: 0.1584, loss_cls_2: 1.0046, loss_box_2: 1.8327, loss_cns_2: 0.6494, loss_yns_2: 0.1566, loss_cls_3: 1.0233, loss_box_3: 1.8314, loss_cns_3: 0.6480, loss_yns_3: 0.1582, loss_cls_4: 1.0391, loss_box_4: 1.8360, loss_cns_4: 0.6494, loss_yns_4: 0.1594, loss_cls_5: 1.0278, loss_box_5: 1.8516, loss_cns_5: 0.6467, loss_yns_5: 0.1587, loss_cls_dn_0: 0.2382, loss_box_dn_0: 0.8235, loss_cls_dn_1: 0.1818, loss_box_dn_1: 0.7893, loss_cls_dn_2: 0.1820, loss_box_dn_2: 0.7787, loss_cls_dn_3: 0.1864, loss_box_dn_3: 0.7960, loss_cls_dn_4: 0.1902, loss_box_dn_4: 0.8107, loss_cls_dn_5: 0.1923, loss_box_dn_5: 0.8284, loss_dense_depth: 0.8046, loss: 28.6491, grad_norm: 61.3648 -2025-11-17 14:22:41,825 - mmdet - INFO - Iter [109/17500] lr: 1.432e-04, eta: 12:24:28, time: 1.488, data_time: 0.086, memory: 49163, loss_cls_0: 0.9141, loss_box_0: 1.7900, loss_cns_0: 0.6121, loss_yns_0: 0.1581, loss_cls_1: 1.0045, loss_box_1: 1.9362, loss_cns_1: 0.6393, loss_yns_1: 0.1603, loss_cls_2: 1.0208, loss_box_2: 1.8768, loss_cns_2: 0.6478, loss_yns_2: 0.1616, loss_cls_3: 1.0321, loss_box_3: 1.8549, loss_cns_3: 0.6515, loss_yns_3: 0.1598, loss_cls_4: 1.0147, loss_box_4: 1.8530, loss_cns_4: 0.6484, loss_yns_4: 0.1597, loss_cls_5: 1.0280, loss_box_5: 1.8775, loss_cns_5: 0.6481, loss_yns_5: 0.1603, loss_cls_dn_0: 0.2394, loss_box_dn_0: 0.8150, loss_cls_dn_1: 0.1755, loss_box_dn_1: 0.8031, loss_cls_dn_2: 0.1797, loss_box_dn_2: 0.7820, loss_cls_dn_3: 0.1854, loss_box_dn_3: 0.7832, loss_cls_dn_4: 0.1846, loss_box_dn_4: 0.7867, loss_cls_dn_5: 0.1899, loss_box_dn_5: 0.7976, loss_dense_depth: 0.8578, loss: 28.7894, grad_norm: 43.4087 -2025-11-17 14:22:43,305 - mmdet - INFO - Iter [110/17500] lr: 1.436e-04, eta: 12:21:33, time: 1.479, data_time: 0.076, memory: 49163, loss_cls_0: 0.9188, loss_box_0: 1.7989, loss_cns_0: 0.6101, loss_yns_0: 0.1561, loss_cls_1: 0.9924, loss_box_1: 1.9659, loss_cns_1: 0.6309, loss_yns_1: 0.1583, loss_cls_2: 1.0217, loss_box_2: 1.8985, loss_cns_2: 0.6412, loss_yns_2: 0.1608, loss_cls_3: 1.0365, loss_box_3: 1.8682, loss_cns_3: 0.6486, loss_yns_3: 0.1587, loss_cls_4: 1.0351, loss_box_4: 1.8741, loss_cns_4: 0.6462, loss_yns_4: 0.1583, loss_cls_5: 1.0415, loss_box_5: 1.8956, loss_cns_5: 0.6437, loss_yns_5: 0.1584, loss_cls_dn_0: 0.2428, loss_box_dn_0: 0.8060, loss_cls_dn_1: 0.1725, loss_box_dn_1: 0.7858, loss_cls_dn_2: 0.1771, loss_box_dn_2: 0.7577, loss_cls_dn_3: 0.1835, loss_box_dn_3: 0.7478, loss_cls_dn_4: 0.1799, loss_box_dn_4: 0.7503, loss_cls_dn_5: 0.1885, loss_box_dn_5: 0.7570, loss_dense_depth: 0.8363, loss: 28.7036, grad_norm: 47.8934 -2025-11-17 14:22:44,816 - mmdet - INFO - Iter [111/17500] lr: 1.440e-04, eta: 12:18:46, time: 1.511, data_time: 0.073, memory: 49163, loss_cls_0: 0.9187, loss_box_0: 1.8047, loss_cns_0: 0.6108, loss_yns_0: 0.1552, loss_cls_1: 0.9833, loss_box_1: 1.9413, loss_cns_1: 0.6313, loss_yns_1: 0.1568, loss_cls_2: 1.0042, loss_box_2: 1.8718, loss_cns_2: 0.6453, loss_yns_2: 0.1593, loss_cls_3: 1.0158, loss_box_3: 1.8656, loss_cns_3: 0.6470, loss_yns_3: 0.1580, loss_cls_4: 1.0311, loss_box_4: 1.8648, loss_cns_4: 0.6474, loss_yns_4: 0.1573, loss_cls_5: 1.0257, loss_box_5: 1.8563, loss_cns_5: 0.6456, loss_yns_5: 0.1579, loss_cls_dn_0: 0.2425, loss_box_dn_0: 0.8046, loss_cls_dn_1: 0.1713, loss_box_dn_1: 0.7638, loss_cls_dn_2: 0.1752, loss_box_dn_2: 0.7319, loss_cls_dn_3: 0.1800, loss_box_dn_3: 0.7298, loss_cls_dn_4: 0.1842, loss_box_dn_4: 0.7329, loss_cls_dn_5: 0.1884, loss_box_dn_5: 0.7311, loss_dense_depth: 0.8794, loss: 28.4701, grad_norm: 29.3678 -2025-11-17 14:22:46,311 - mmdet - INFO - Iter [112/17500] lr: 1.444e-04, eta: 12:16:00, time: 1.495, data_time: 0.074, memory: 49163, loss_cls_0: 0.9389, loss_box_0: 1.8276, loss_cns_0: 0.6096, loss_yns_0: 0.1588, loss_cls_1: 1.0087, loss_box_1: 1.9187, loss_cns_1: 0.6311, loss_yns_1: 0.1553, loss_cls_2: 1.0264, loss_box_2: 1.8928, loss_cns_2: 0.6398, loss_yns_2: 0.1551, loss_cls_3: 1.0159, loss_box_3: 1.9090, loss_cns_3: 0.6412, loss_yns_3: 0.1563, loss_cls_4: 1.0445, loss_box_4: 1.9058, loss_cns_4: 0.6394, loss_yns_4: 0.1573, loss_cls_5: 1.0290, loss_box_5: 1.8926, loss_cns_5: 0.6396, loss_yns_5: 0.1576, loss_cls_dn_0: 0.2476, loss_box_dn_0: 0.8057, loss_cls_dn_1: 0.1721, loss_box_dn_1: 0.7512, loss_cls_dn_2: 0.1735, loss_box_dn_2: 0.7326, loss_cls_dn_3: 0.1787, loss_box_dn_3: 0.7395, loss_cls_dn_4: 0.1879, loss_box_dn_4: 0.7460, loss_cls_dn_5: 0.1883, loss_box_dn_5: 0.7473, loss_dense_depth: 0.9009, loss: 28.7226, grad_norm: 51.7774 -2025-11-17 14:22:47,801 - mmdet - INFO - Iter [113/17500] lr: 1.448e-04, eta: 12:13:16, time: 1.490, data_time: 0.078, memory: 49163, loss_cls_0: 0.8999, loss_box_0: 1.7996, loss_cns_0: 0.6172, loss_yns_0: 0.1567, loss_cls_1: 0.9602, loss_box_1: 1.8733, loss_cns_1: 0.6326, loss_yns_1: 0.1548, loss_cls_2: 0.9869, loss_box_2: 1.8446, loss_cns_2: 0.6406, loss_yns_2: 0.1568, loss_cls_3: 0.9881, loss_box_3: 1.8506, loss_cns_3: 0.6442, loss_yns_3: 0.1539, loss_cls_4: 0.9976, loss_box_4: 1.8352, loss_cns_4: 0.6426, loss_yns_4: 0.1550, loss_cls_5: 0.9963, loss_box_5: 1.8295, loss_cns_5: 0.6440, loss_yns_5: 0.1563, loss_cls_dn_0: 0.2344, loss_box_dn_0: 0.8045, loss_cls_dn_1: 0.1662, loss_box_dn_1: 0.7787, loss_cls_dn_2: 0.1664, loss_box_dn_2: 0.7642, loss_cls_dn_3: 0.1747, loss_box_dn_3: 0.7749, loss_cls_dn_4: 0.1827, loss_box_dn_4: 0.7833, loss_cls_dn_5: 0.1888, loss_box_dn_5: 0.7923, loss_dense_depth: 0.8980, loss: 28.3259, grad_norm: 42.0371 -2025-11-17 14:22:49,289 - mmdet - INFO - Iter [114/17500] lr: 1.452e-04, eta: 12:10:35, time: 1.487, data_time: 0.076, memory: 49163, loss_cls_0: 0.9038, loss_box_0: 1.8127, loss_cns_0: 0.6178, loss_yns_0: 0.1569, loss_cls_1: 0.9787, loss_box_1: 1.8548, loss_cns_1: 0.6451, loss_yns_1: 0.1583, loss_cls_2: 0.9875, loss_box_2: 1.8261, loss_cns_2: 0.6510, loss_yns_2: 0.1620, loss_cls_3: 1.0021, loss_box_3: 1.8333, loss_cns_3: 0.6474, loss_yns_3: 0.1562, loss_cls_4: 0.9970, loss_box_4: 1.8421, loss_cns_4: 0.6488, loss_yns_4: 0.1568, loss_cls_5: 1.0062, loss_box_5: 1.8505, loss_cns_5: 0.6484, loss_yns_5: 0.1585, loss_cls_dn_0: 0.2421, loss_box_dn_0: 0.8001, loss_cls_dn_1: 0.1662, loss_box_dn_1: 0.7852, loss_cls_dn_2: 0.1684, loss_box_dn_2: 0.7781, loss_cls_dn_3: 0.1742, loss_box_dn_3: 0.7928, loss_cls_dn_4: 0.1793, loss_box_dn_4: 0.8090, loss_cls_dn_5: 0.1898, loss_box_dn_5: 0.8211, loss_dense_depth: 0.8615, loss: 28.4699, grad_norm: 56.9018 -2025-11-17 14:22:50,832 - mmdet - INFO - Iter [115/17500] lr: 1.456e-04, eta: 12:08:04, time: 1.543, data_time: 0.086, memory: 49163, loss_cls_0: 0.8787, loss_box_0: 1.7724, loss_cns_0: 0.6189, loss_yns_0: 0.1555, loss_cls_1: 0.9767, loss_box_1: 1.8856, loss_cns_1: 0.6482, loss_yns_1: 0.1564, loss_cls_2: 0.9767, loss_box_2: 1.8717, loss_cns_2: 0.6470, loss_yns_2: 0.1588, loss_cls_3: 0.9985, loss_box_3: 1.8785, loss_cns_3: 0.6442, loss_yns_3: 0.1574, loss_cls_4: 0.9958, loss_box_4: 1.8967, loss_cns_4: 0.6470, loss_yns_4: 0.1580, loss_cls_5: 1.0007, loss_box_5: 1.8889, loss_cns_5: 0.6493, loss_yns_5: 0.1571, loss_cls_dn_0: 0.2347, loss_box_dn_0: 0.8007, loss_cls_dn_1: 0.1697, loss_box_dn_1: 0.7990, loss_cls_dn_2: 0.1707, loss_box_dn_2: 0.7881, loss_cls_dn_3: 0.1752, loss_box_dn_3: 0.8032, loss_cls_dn_4: 0.1783, loss_box_dn_4: 0.8209, loss_cls_dn_5: 0.1875, loss_box_dn_5: 0.8190, loss_dense_depth: 0.8634, loss: 28.6294, grad_norm: 55.5526 -2025-11-17 14:22:52,332 - mmdet - INFO - Iter [116/17500] lr: 1.460e-04, eta: 12:05:30, time: 1.499, data_time: 0.079, memory: 49163, loss_cls_0: 0.9050, loss_box_0: 1.7924, loss_cns_0: 0.6170, loss_yns_0: 0.1576, loss_cls_1: 0.9913, loss_box_1: 1.9181, loss_cns_1: 0.6378, loss_yns_1: 0.1563, loss_cls_2: 0.9948, loss_box_2: 1.8754, loss_cns_2: 0.6415, loss_yns_2: 0.1568, loss_cls_3: 1.0082, loss_box_3: 1.8679, loss_cns_3: 0.6446, loss_yns_3: 0.1582, loss_cls_4: 1.0247, loss_box_4: 1.8842, loss_cns_4: 0.6472, loss_yns_4: 0.1601, loss_cls_5: 1.0213, loss_box_5: 1.8883, loss_cns_5: 0.6455, loss_yns_5: 0.1587, loss_cls_dn_0: 0.2382, loss_box_dn_0: 0.7977, loss_cls_dn_1: 0.1689, loss_box_dn_1: 0.7861, loss_cls_dn_2: 0.1714, loss_box_dn_2: 0.7646, loss_cls_dn_3: 0.1764, loss_box_dn_3: 0.7652, loss_cls_dn_4: 0.1811, loss_box_dn_4: 0.7795, loss_cls_dn_5: 0.1943, loss_box_dn_5: 0.7790, loss_dense_depth: 0.8672, loss: 28.6225, grad_norm: 40.1478 -2025-11-17 14:22:53,814 - mmdet - INFO - Iter [117/17500] lr: 1.464e-04, eta: 12:02:55, time: 1.483, data_time: 0.078, memory: 49163, loss_cls_0: 0.8840, loss_box_0: 1.7660, loss_cns_0: 0.6227, loss_yns_0: 0.1550, loss_cls_1: 0.9533, loss_box_1: 1.9544, loss_cns_1: 0.6321, loss_yns_1: 0.1566, loss_cls_2: 0.9738, loss_box_2: 1.8850, loss_cns_2: 0.6413, loss_yns_2: 0.1552, loss_cls_3: 0.9884, loss_box_3: 1.8583, loss_cns_3: 0.6455, loss_yns_3: 0.1573, loss_cls_4: 1.0013, loss_box_4: 1.8544, loss_cns_4: 0.6474, loss_yns_4: 0.1559, loss_cls_5: 1.0000, loss_box_5: 1.8848, loss_cns_5: 0.6446, loss_yns_5: 0.1562, loss_cls_dn_0: 0.2326, loss_box_dn_0: 0.7967, loss_cls_dn_1: 0.1681, loss_box_dn_1: 0.7942, loss_cls_dn_2: 0.1718, loss_box_dn_2: 0.7683, loss_cls_dn_3: 0.1761, loss_box_dn_3: 0.7528, loss_cls_dn_4: 0.1814, loss_box_dn_4: 0.7557, loss_cls_dn_5: 0.1900, loss_box_dn_5: 0.7673, loss_dense_depth: 0.8340, loss: 28.3626, grad_norm: 47.6404 -2025-11-17 14:22:55,304 - mmdet - INFO - Iter [118/17500] lr: 1.468e-04, eta: 12:00:25, time: 1.489, data_time: 0.077, memory: 49163, loss_cls_0: 0.9030, loss_box_0: 1.7631, loss_cns_0: 0.6241, loss_yns_0: 0.1558, loss_cls_1: 0.9475, loss_box_1: 1.9024, loss_cns_1: 0.6344, loss_yns_1: 0.1553, loss_cls_2: 0.9743, loss_box_2: 1.8539, loss_cns_2: 0.6412, loss_yns_2: 0.1554, loss_cls_3: 0.9932, loss_box_3: 1.8210, loss_cns_3: 0.6471, loss_yns_3: 0.1579, loss_cls_4: 0.9876, loss_box_4: 1.8085, loss_cns_4: 0.6486, loss_yns_4: 0.1554, loss_cls_5: 0.9851, loss_box_5: 1.8412, loss_cns_5: 0.6509, loss_yns_5: 0.1566, loss_cls_dn_0: 0.2345, loss_box_dn_0: 0.7931, loss_cls_dn_1: 0.1672, loss_box_dn_1: 0.7807, loss_cls_dn_2: 0.1715, loss_box_dn_2: 0.7538, loss_cls_dn_3: 0.1772, loss_box_dn_3: 0.7413, loss_cls_dn_4: 0.1776, loss_box_dn_4: 0.7404, loss_cls_dn_5: 0.1815, loss_box_dn_5: 0.7569, loss_dense_depth: 0.8537, loss: 28.0928, grad_norm: 52.1213 -2025-11-17 14:22:56,783 - mmdet - INFO - Iter [119/17500] lr: 1.472e-04, eta: 11:57:55, time: 1.480, data_time: 0.079, memory: 49163, loss_cls_0: 0.8916, loss_box_0: 1.7645, loss_cns_0: 0.6213, loss_yns_0: 0.1539, loss_cls_1: 0.9591, loss_box_1: 1.8869, loss_cns_1: 0.6360, loss_yns_1: 0.1541, loss_cls_2: 0.9768, loss_box_2: 1.8493, loss_cns_2: 0.6401, loss_yns_2: 0.1576, loss_cls_3: 0.9869, loss_box_3: 1.8316, loss_cns_3: 0.6458, loss_yns_3: 0.1557, loss_cls_4: 0.9949, loss_box_4: 1.8132, loss_cns_4: 0.6467, loss_yns_4: 0.1568, loss_cls_5: 0.9931, loss_box_5: 1.8282, loss_cns_5: 0.6504, loss_yns_5: 0.1563, loss_cls_dn_0: 0.2307, loss_box_dn_0: 0.7842, loss_cls_dn_1: 0.1689, loss_box_dn_1: 0.7509, loss_cls_dn_2: 0.1739, loss_box_dn_2: 0.7280, loss_cls_dn_3: 0.1748, loss_box_dn_3: 0.7288, loss_cls_dn_4: 0.1778, loss_box_dn_4: 0.7285, loss_cls_dn_5: 0.1795, loss_box_dn_5: 0.7405, loss_dense_depth: 0.8346, loss: 27.9517, grad_norm: 41.4942 -2025-11-17 14:22:58,274 - mmdet - INFO - Iter [120/17500] lr: 1.476e-04, eta: 11:55:30, time: 1.492, data_time: 0.076, memory: 49163, loss_cls_0: 0.8810, loss_box_0: 1.7728, loss_cns_0: 0.6153, loss_yns_0: 0.1540, loss_cls_1: 0.9590, loss_box_1: 1.8864, loss_cns_1: 0.6377, loss_yns_1: 0.1550, loss_cls_2: 0.9758, loss_box_2: 1.8413, loss_cns_2: 0.6467, loss_yns_2: 0.1611, loss_cls_3: 0.9874, loss_box_3: 1.8356, loss_cns_3: 0.6465, loss_yns_3: 0.1563, loss_cls_4: 0.9954, loss_box_4: 1.8246, loss_cns_4: 0.6455, loss_yns_4: 0.1565, loss_cls_5: 0.9988, loss_box_5: 1.8210, loss_cns_5: 0.6490, loss_yns_5: 0.1578, loss_cls_dn_0: 0.2295, loss_box_dn_0: 0.7943, loss_cls_dn_1: 0.1661, loss_box_dn_1: 0.7587, loss_cls_dn_2: 0.1683, loss_box_dn_2: 0.7388, loss_cls_dn_3: 0.1687, loss_box_dn_3: 0.7559, loss_cls_dn_4: 0.1731, loss_box_dn_4: 0.7674, loss_cls_dn_5: 0.1835, loss_box_dn_5: 0.7776, loss_dense_depth: 0.8204, loss: 28.0630, grad_norm: 49.0828 -2025-11-17 14:22:59,829 - mmdet - INFO - Iter [121/17500] lr: 1.480e-04, eta: 11:53:16, time: 1.554, data_time: 0.107, memory: 49163, loss_cls_0: 0.9065, loss_box_0: 1.7604, loss_cns_0: 0.6163, loss_yns_0: 0.1558, loss_cls_1: 0.9521, loss_box_1: 1.9529, loss_cns_1: 0.6366, loss_yns_1: 0.1569, loss_cls_2: 0.9809, loss_box_2: 1.9044, loss_cns_2: 0.6457, loss_yns_2: 0.1595, loss_cls_3: 0.9983, loss_box_3: 1.8937, loss_cns_3: 0.6445, loss_yns_3: 0.1594, loss_cls_4: 1.0114, loss_box_4: 1.8786, loss_cns_4: 0.6451, loss_yns_4: 0.1565, loss_cls_5: 1.0158, loss_box_5: 1.8845, loss_cns_5: 0.6466, loss_yns_5: 0.1580, loss_cls_dn_0: 0.2315, loss_box_dn_0: 0.7992, loss_cls_dn_1: 0.1614, loss_box_dn_1: 0.7720, loss_cls_dn_2: 0.1648, loss_box_dn_2: 0.7588, loss_cls_dn_3: 0.1681, loss_box_dn_3: 0.7725, loss_cls_dn_4: 0.1741, loss_box_dn_4: 0.7792, loss_cls_dn_5: 0.1805, loss_box_dn_5: 0.7941, loss_dense_depth: 0.8227, loss: 28.4994, grad_norm: 61.8098 -2025-11-17 14:23:01,353 - mmdet - INFO - Iter [122/17500] lr: 1.484e-04, eta: 11:51:00, time: 1.524, data_time: 0.105, memory: 49163, loss_cls_0: 0.9047, loss_box_0: 1.7232, loss_cns_0: 0.6155, loss_yns_0: 0.1559, loss_cls_1: 0.9480, loss_box_1: 1.9096, loss_cns_1: 0.6341, loss_yns_1: 0.1582, loss_cls_2: 0.9992, loss_box_2: 1.8259, loss_cns_2: 0.6459, loss_yns_2: 0.1572, loss_cls_3: 1.0037, loss_box_3: 1.8156, loss_cns_3: 0.6473, loss_yns_3: 0.1634, loss_cls_4: 1.0105, loss_box_4: 1.8099, loss_cns_4: 0.6500, loss_yns_4: 0.1579, loss_cls_5: 1.0121, loss_box_5: 1.8139, loss_cns_5: 0.6489, loss_yns_5: 0.1585, loss_cls_dn_0: 0.2329, loss_box_dn_0: 0.7905, loss_cls_dn_1: 0.1612, loss_box_dn_1: 0.7751, loss_cls_dn_2: 0.1653, loss_box_dn_2: 0.7526, loss_cls_dn_3: 0.1695, loss_box_dn_3: 0.7552, loss_cls_dn_4: 0.1792, loss_box_dn_4: 0.7574, loss_cls_dn_5: 0.1771, loss_box_dn_5: 0.7710, loss_dense_depth: 0.7992, loss: 28.0553, grad_norm: 42.3386 -2025-11-17 14:23:02,881 - mmdet - INFO - Iter [123/17500] lr: 1.488e-04, eta: 11:48:46, time: 1.529, data_time: 0.082, memory: 49163, loss_cls_0: 0.8988, loss_box_0: 1.7377, loss_cns_0: 0.6186, loss_yns_0: 0.1582, loss_cls_1: 0.9486, loss_box_1: 1.8969, loss_cns_1: 0.6326, loss_yns_1: 0.1587, loss_cls_2: 0.9929, loss_box_2: 1.8173, loss_cns_2: 0.6484, loss_yns_2: 0.1576, loss_cls_3: 0.9957, loss_box_3: 1.8241, loss_cns_3: 0.6473, loss_yns_3: 0.1579, loss_cls_4: 0.9971, loss_box_4: 1.8423, loss_cns_4: 0.6476, loss_yns_4: 0.1578, loss_cls_5: 0.9996, loss_box_5: 1.8199, loss_cns_5: 0.6469, loss_yns_5: 0.1565, loss_cls_dn_0: 0.2269, loss_box_dn_0: 0.7872, loss_cls_dn_1: 0.1622, loss_box_dn_1: 0.7723, loss_cls_dn_2: 0.1693, loss_box_dn_2: 0.7417, loss_cls_dn_3: 0.1706, loss_box_dn_3: 0.7465, loss_cls_dn_4: 0.1742, loss_box_dn_4: 0.7577, loss_cls_dn_5: 0.1753, loss_box_dn_5: 0.7554, loss_dense_depth: 0.8149, loss: 28.0133, grad_norm: 45.0651 -2025-11-17 14:23:04,397 - mmdet - INFO - Iter [124/17500] lr: 1.492e-04, eta: 11:46:33, time: 1.516, data_time: 0.081, memory: 49163, loss_cls_0: 0.8981, loss_box_0: 1.7728, loss_cns_0: 0.6197, loss_yns_0: 0.1590, loss_cls_1: 0.9479, loss_box_1: 1.8891, loss_cns_1: 0.6379, loss_yns_1: 0.1587, loss_cls_2: 0.9911, loss_box_2: 1.8233, loss_cns_2: 0.6513, loss_yns_2: 0.1609, loss_cls_3: 1.0001, loss_box_3: 1.8268, loss_cns_3: 0.6528, loss_yns_3: 0.1544, loss_cls_4: 1.0132, loss_box_4: 1.8632, loss_cns_4: 0.6494, loss_yns_4: 0.1564, loss_cls_5: 1.0068, loss_box_5: 1.8298, loss_cns_5: 0.6538, loss_yns_5: 0.1571, loss_cls_dn_0: 0.2263, loss_box_dn_0: 0.7900, loss_cls_dn_1: 0.1582, loss_box_dn_1: 0.7592, loss_cls_dn_2: 0.1650, loss_box_dn_2: 0.7332, loss_cls_dn_3: 0.1683, loss_box_dn_3: 0.7372, loss_cls_dn_4: 0.1676, loss_box_dn_4: 0.7574, loss_cls_dn_5: 0.1767, loss_box_dn_5: 0.7489, loss_dense_depth: 0.7847, loss: 28.0464, grad_norm: 50.3967 -2025-11-17 14:23:05,945 - mmdet - INFO - Iter [125/17500] lr: 1.496e-04, eta: 11:44:27, time: 1.548, data_time: 0.091, memory: 49163, loss_cls_0: 0.8983, loss_box_0: 1.8062, loss_cns_0: 0.6130, loss_yns_0: 0.1584, loss_cls_1: 0.9627, loss_box_1: 1.8252, loss_cns_1: 0.6428, loss_yns_1: 0.1581, loss_cls_2: 0.9947, loss_box_2: 1.7756, loss_cns_2: 0.6506, loss_yns_2: 0.1655, loss_cls_3: 1.0107, loss_box_3: 1.7764, loss_cns_3: 0.6515, loss_yns_3: 0.1582, loss_cls_4: 1.0182, loss_box_4: 1.8020, loss_cns_4: 0.6489, loss_yns_4: 0.1586, loss_cls_5: 1.0218, loss_box_5: 1.7907, loss_cns_5: 0.6578, loss_yns_5: 0.1592, loss_cls_dn_0: 0.2328, loss_box_dn_0: 0.7889, loss_cls_dn_1: 0.1596, loss_box_dn_1: 0.7572, loss_cls_dn_2: 0.1639, loss_box_dn_2: 0.7347, loss_cls_dn_3: 0.1689, loss_box_dn_3: 0.7341, loss_cls_dn_4: 0.1677, loss_box_dn_4: 0.7489, loss_cls_dn_5: 0.1800, loss_box_dn_5: 0.7474, loss_dense_depth: 0.7799, loss: 27.8692, grad_norm: 45.9591 -2025-11-17 14:23:07,461 - mmdet - INFO - Iter [126/17500] lr: 1.500e-04, eta: 11:42:18, time: 1.515, data_time: 0.089, memory: 49163, loss_cls_0: 0.9138, loss_box_0: 1.7672, loss_cns_0: 0.6154, loss_yns_0: 0.1552, loss_cls_1: 0.9619, loss_box_1: 1.8371, loss_cns_1: 0.6392, loss_yns_1: 0.1548, loss_cls_2: 0.9949, loss_box_2: 1.7862, loss_cns_2: 0.6493, loss_yns_2: 0.1588, loss_cls_3: 1.0133, loss_box_3: 1.7798, loss_cns_3: 0.6495, loss_yns_3: 0.1581, loss_cls_4: 1.0081, loss_box_4: 1.7854, loss_cns_4: 0.6487, loss_yns_4: 0.1568, loss_cls_5: 1.0189, loss_box_5: 1.7941, loss_cns_5: 0.6537, loss_yns_5: 0.1576, loss_cls_dn_0: 0.2333, loss_box_dn_0: 0.7908, loss_cls_dn_1: 0.1620, loss_box_dn_1: 0.7556, loss_cls_dn_2: 0.1643, loss_box_dn_2: 0.7253, loss_cls_dn_3: 0.1687, loss_box_dn_3: 0.7230, loss_cls_dn_4: 0.1715, loss_box_dn_4: 0.7288, loss_cls_dn_5: 0.1828, loss_box_dn_5: 0.7401, loss_dense_depth: 0.7691, loss: 27.7728, grad_norm: 40.4605 -2025-11-17 14:23:08,977 - mmdet - INFO - Iter [127/17500] lr: 1.504e-04, eta: 11:40:11, time: 1.516, data_time: 0.113, memory: 49163, loss_cls_0: 0.9031, loss_box_0: 1.7515, loss_cns_0: 0.6163, loss_yns_0: 0.1551, loss_cls_1: 0.9691, loss_box_1: 1.7931, loss_cns_1: 0.6412, loss_yns_1: 0.1578, loss_cls_2: 0.9953, loss_box_2: 1.7573, loss_cns_2: 0.6506, loss_yns_2: 0.1569, loss_cls_3: 1.0098, loss_box_3: 1.7459, loss_cns_3: 0.6519, loss_yns_3: 0.1599, loss_cls_4: 1.0088, loss_box_4: 1.7450, loss_cns_4: 0.6508, loss_yns_4: 0.1550, loss_cls_5: 1.0075, loss_box_5: 1.7639, loss_cns_5: 0.6515, loss_yns_5: 0.1574, loss_cls_dn_0: 0.2312, loss_box_dn_0: 0.7856, loss_cls_dn_1: 0.1593, loss_box_dn_1: 0.7485, loss_cls_dn_2: 0.1619, loss_box_dn_2: 0.7267, loss_cls_dn_3: 0.1667, loss_box_dn_3: 0.7248, loss_cls_dn_4: 0.1715, loss_box_dn_4: 0.7310, loss_cls_dn_5: 0.1786, loss_box_dn_5: 0.7432, loss_dense_depth: 0.7591, loss: 27.5428, grad_norm: 49.8544 -2025-11-17 14:23:10,467 - mmdet - INFO - Iter [128/17500] lr: 1.508e-04, eta: 11:38:03, time: 1.489, data_time: 0.080, memory: 49163, loss_cls_0: 0.9175, loss_box_0: 1.7408, loss_cns_0: 0.6211, loss_yns_0: 0.1586, loss_cls_1: 0.9683, loss_box_1: 1.8004, loss_cns_1: 0.6389, loss_yns_1: 0.1643, loss_cls_2: 0.9874, loss_box_2: 1.7495, loss_cns_2: 0.6491, loss_yns_2: 0.1679, loss_cls_3: 1.0012, loss_box_3: 1.7474, loss_cns_3: 0.6506, loss_yns_3: 0.1601, loss_cls_4: 1.0018, loss_box_4: 1.7564, loss_cns_4: 0.6485, loss_yns_4: 0.1647, loss_cls_5: 1.0008, loss_box_5: 1.7709, loss_cns_5: 0.6534, loss_yns_5: 0.1620, loss_cls_dn_0: 0.2289, loss_box_dn_0: 0.7850, loss_cls_dn_1: 0.1569, loss_box_dn_1: 0.7329, loss_cls_dn_2: 0.1597, loss_box_dn_2: 0.7128, loss_cls_dn_3: 0.1622, loss_box_dn_3: 0.7123, loss_cls_dn_4: 0.1653, loss_box_dn_4: 0.7212, loss_cls_dn_5: 0.1724, loss_box_dn_5: 0.7310, loss_dense_depth: 0.7841, loss: 27.5064, grad_norm: 34.8556 -2025-11-17 14:23:11,957 - mmdet - INFO - Iter [129/17500] lr: 1.512e-04, eta: 11:35:57, time: 1.492, data_time: 0.089, memory: 49163, loss_cls_0: 0.8896, loss_box_0: 1.7228, loss_cns_0: 0.6213, loss_yns_0: 0.1576, loss_cls_1: 0.9539, loss_box_1: 1.7848, loss_cns_1: 0.6414, loss_yns_1: 0.1593, loss_cls_2: 0.9687, loss_box_2: 1.7361, loss_cns_2: 0.6500, loss_yns_2: 0.1667, loss_cls_3: 0.9872, loss_box_3: 1.7228, loss_cns_3: 0.6491, loss_yns_3: 0.1572, loss_cls_4: 0.9922, loss_box_4: 1.7296, loss_cns_4: 0.6484, loss_yns_4: 0.1629, loss_cls_5: 1.0107, loss_box_5: 1.7613, loss_cns_5: 0.6557, loss_yns_5: 0.1621, loss_cls_dn_0: 0.2245, loss_box_dn_0: 0.7816, loss_cls_dn_1: 0.1583, loss_box_dn_1: 0.7326, loss_cls_dn_2: 0.1601, loss_box_dn_2: 0.7132, loss_cls_dn_3: 0.1614, loss_box_dn_3: 0.7127, loss_cls_dn_4: 0.1647, loss_box_dn_4: 0.7174, loss_cls_dn_5: 0.1740, loss_box_dn_5: 0.7325, loss_dense_depth: 0.7595, loss: 27.2841, grad_norm: 44.4685 -2025-11-17 14:23:13,440 - mmdet - INFO - Iter [130/17500] lr: 1.516e-04, eta: 11:33:51, time: 1.482, data_time: 0.077, memory: 49163, loss_cls_0: 0.9019, loss_box_0: 1.7552, loss_cns_0: 0.6098, loss_yns_0: 0.1531, loss_cls_1: 0.9689, loss_box_1: 1.8139, loss_cns_1: 0.6428, loss_yns_1: 0.1559, loss_cls_2: 0.9872, loss_box_2: 1.7698, loss_cns_2: 0.6531, loss_yns_2: 0.1580, loss_cls_3: 0.9918, loss_box_3: 1.7442, loss_cns_3: 0.6525, loss_yns_3: 0.1547, loss_cls_4: 1.0017, loss_box_4: 1.7411, loss_cns_4: 0.6539, loss_yns_4: 0.1548, loss_cls_5: 1.0086, loss_box_5: 1.7759, loss_cns_5: 0.6606, loss_yns_5: 0.1566, loss_cls_dn_0: 0.2285, loss_box_dn_0: 0.7928, loss_cls_dn_1: 0.1580, loss_box_dn_1: 0.7509, loss_cls_dn_2: 0.1623, loss_box_dn_2: 0.7351, loss_cls_dn_3: 0.1622, loss_box_dn_3: 0.7317, loss_cls_dn_4: 0.1704, loss_box_dn_4: 0.7328, loss_cls_dn_5: 0.1775, loss_box_dn_5: 0.7472, loss_dense_depth: 0.7918, loss: 27.6076, grad_norm: 44.8923 -2025-11-17 14:23:14,958 - mmdet - INFO - Iter [131/17500] lr: 1.520e-04, eta: 11:31:52, time: 1.518, data_time: 0.080, memory: 49163, loss_cls_0: 0.9192, loss_box_0: 1.7490, loss_cns_0: 0.6140, loss_yns_0: 0.1534, loss_cls_1: 0.9691, loss_box_1: 1.8539, loss_cns_1: 0.6444, loss_yns_1: 0.1575, loss_cls_2: 0.9837, loss_box_2: 1.8132, loss_cns_2: 0.6517, loss_yns_2: 0.1538, loss_cls_3: 0.9935, loss_box_3: 1.7886, loss_cns_3: 0.6548, loss_yns_3: 0.1532, loss_cls_4: 0.9991, loss_box_4: 1.7811, loss_cns_4: 0.6531, loss_yns_4: 0.1557, loss_cls_5: 1.0088, loss_box_5: 1.7937, loss_cns_5: 0.6533, loss_yns_5: 0.1537, loss_cls_dn_0: 0.2314, loss_box_dn_0: 0.7995, loss_cls_dn_1: 0.1555, loss_box_dn_1: 0.7551, loss_cls_dn_2: 0.1592, loss_box_dn_2: 0.7371, loss_cls_dn_3: 0.1620, loss_box_dn_3: 0.7309, loss_cls_dn_4: 0.1693, loss_box_dn_4: 0.7264, loss_cls_dn_5: 0.1749, loss_box_dn_5: 0.7328, loss_dense_depth: 0.8132, loss: 27.7990, grad_norm: 33.0946 -2025-11-17 14:23:16,461 - mmdet - INFO - Iter [132/17500] lr: 1.524e-04, eta: 11:29:53, time: 1.504, data_time: 0.078, memory: 49163, loss_cls_0: 0.9001, loss_box_0: 1.7664, loss_cns_0: 0.6233, loss_yns_0: 0.1517, loss_cls_1: 0.9914, loss_box_1: 1.8305, loss_cns_1: 0.6441, loss_yns_1: 0.1589, loss_cls_2: 1.0081, loss_box_2: 1.7926, loss_cns_2: 0.6482, loss_yns_2: 0.1536, loss_cls_3: 1.0081, loss_box_3: 1.7838, loss_cns_3: 0.6511, loss_yns_3: 0.1507, loss_cls_4: 1.0296, loss_box_4: 1.7863, loss_cns_4: 0.6502, loss_yns_4: 0.1536, loss_cls_5: 1.0017, loss_box_5: 1.7760, loss_cns_5: 0.6541, loss_yns_5: 0.1527, loss_cls_dn_0: 0.2330, loss_box_dn_0: 0.7854, loss_cls_dn_1: 0.1555, loss_box_dn_1: 0.7372, loss_cls_dn_2: 0.1592, loss_box_dn_2: 0.7172, loss_cls_dn_3: 0.1656, loss_box_dn_3: 0.7167, loss_cls_dn_4: 0.1728, loss_box_dn_4: 0.7205, loss_cls_dn_5: 0.1788, loss_box_dn_5: 0.7217, loss_dense_depth: 0.8059, loss: 27.7364, grad_norm: 52.9101 -2025-11-17 14:23:17,957 - mmdet - INFO - Iter [133/17500] lr: 1.528e-04, eta: 11:27:55, time: 1.495, data_time: 0.079, memory: 49163, loss_cls_0: 0.9044, loss_box_0: 1.7505, loss_cns_0: 0.6198, loss_yns_0: 0.1527, loss_cls_1: 0.9782, loss_box_1: 1.8459, loss_cns_1: 0.6412, loss_yns_1: 0.1590, loss_cls_2: 0.9874, loss_box_2: 1.8107, loss_cns_2: 0.6446, loss_yns_2: 0.1566, loss_cls_3: 1.0102, loss_box_3: 1.8052, loss_cns_3: 0.6450, loss_yns_3: 0.1516, loss_cls_4: 1.0098, loss_box_4: 1.8166, loss_cns_4: 0.6463, loss_yns_4: 0.1557, loss_cls_5: 1.0296, loss_box_5: 1.8134, loss_cns_5: 0.6504, loss_yns_5: 0.1532, loss_cls_dn_0: 0.2314, loss_box_dn_0: 0.7879, loss_cls_dn_1: 0.1564, loss_box_dn_1: 0.7352, loss_cls_dn_2: 0.1586, loss_box_dn_2: 0.7122, loss_cls_dn_3: 0.1655, loss_box_dn_3: 0.7153, loss_cls_dn_4: 0.1720, loss_box_dn_4: 0.7266, loss_cls_dn_5: 0.1760, loss_box_dn_5: 0.7320, loss_dense_depth: 0.8076, loss: 27.8147, grad_norm: 47.9721 -2025-11-17 14:23:19,456 - mmdet - INFO - Iter [134/17500] lr: 1.532e-04, eta: 11:25:58, time: 1.498, data_time: 0.081, memory: 49163, loss_cls_0: 0.8810, loss_box_0: 1.7208, loss_cns_0: 0.6238, loss_yns_0: 0.1570, loss_cls_1: 0.9604, loss_box_1: 1.8245, loss_cns_1: 0.6387, loss_yns_1: 0.1556, loss_cls_2: 0.9840, loss_box_2: 1.7786, loss_cns_2: 0.6442, loss_yns_2: 0.1546, loss_cls_3: 1.0003, loss_box_3: 1.7631, loss_cns_3: 0.6449, loss_yns_3: 0.1516, loss_cls_4: 0.9806, loss_box_4: 1.7791, loss_cns_4: 0.6486, loss_yns_4: 0.1528, loss_cls_5: 0.9859, loss_box_5: 1.7822, loss_cns_5: 0.6492, loss_yns_5: 0.1515, loss_cls_dn_0: 0.2236, loss_box_dn_0: 0.7842, loss_cls_dn_1: 0.1571, loss_box_dn_1: 0.7238, loss_cls_dn_2: 0.1563, loss_box_dn_2: 0.7022, loss_cls_dn_3: 0.1604, loss_box_dn_3: 0.7035, loss_cls_dn_4: 0.1652, loss_box_dn_4: 0.7109, loss_cls_dn_5: 0.1669, loss_box_dn_5: 0.7180, loss_dense_depth: 0.7834, loss: 27.3686, grad_norm: 36.2402 -2025-11-17 14:23:21,019 - mmdet - INFO - Iter [135/17500] lr: 1.536e-04, eta: 11:24:12, time: 1.564, data_time: 0.084, memory: 49163, loss_cls_0: 0.9036, loss_box_0: 1.7745, loss_cns_0: 0.6230, loss_yns_0: 0.1610, loss_cls_1: 0.9780, loss_box_1: 1.8691, loss_cns_1: 0.6423, loss_yns_1: 0.1578, loss_cls_2: 1.0001, loss_box_2: 1.8291, loss_cns_2: 0.6469, loss_yns_2: 0.1529, loss_cls_3: 0.9962, loss_box_3: 1.8212, loss_cns_3: 0.6512, loss_yns_3: 0.1564, loss_cls_4: 0.9967, loss_box_4: 1.8270, loss_cns_4: 0.6489, loss_yns_4: 0.1541, loss_cls_5: 0.9908, loss_box_5: 1.8281, loss_cns_5: 0.6511, loss_yns_5: 0.1565, loss_cls_dn_0: 0.2367, loss_box_dn_0: 0.7886, loss_cls_dn_1: 0.1579, loss_box_dn_1: 0.7336, loss_cls_dn_2: 0.1599, loss_box_dn_2: 0.7171, loss_cls_dn_3: 0.1642, loss_box_dn_3: 0.7189, loss_cls_dn_4: 0.1705, loss_box_dn_4: 0.7234, loss_cls_dn_5: 0.1757, loss_box_dn_5: 0.7247, loss_dense_depth: 0.8255, loss: 27.9135, grad_norm: 35.7541 -2025-11-17 14:23:22,504 - mmdet - INFO - Iter [136/17500] lr: 1.540e-04, eta: 11:22:18, time: 1.484, data_time: 0.078, memory: 49163, loss_cls_0: 0.8616, loss_box_0: 1.7397, loss_cns_0: 0.6210, loss_yns_0: 0.1521, loss_cls_1: 0.9352, loss_box_1: 1.8541, loss_cns_1: 0.6427, loss_yns_1: 0.1572, loss_cls_2: 0.9583, loss_box_2: 1.7984, loss_cns_2: 0.6519, loss_yns_2: 0.1516, loss_cls_3: 0.9660, loss_box_3: 1.7957, loss_cns_3: 0.6518, loss_yns_3: 0.1527, loss_cls_4: 0.9644, loss_box_4: 1.8030, loss_cns_4: 0.6538, loss_yns_4: 0.1535, loss_cls_5: 0.9649, loss_box_5: 1.7983, loss_cns_5: 0.6539, loss_yns_5: 0.1532, loss_cls_dn_0: 0.2321, loss_box_dn_0: 0.7804, loss_cls_dn_1: 0.1527, loss_box_dn_1: 0.7235, loss_cls_dn_2: 0.1548, loss_box_dn_2: 0.7083, loss_cls_dn_3: 0.1643, loss_box_dn_3: 0.7112, loss_cls_dn_4: 0.1677, loss_box_dn_4: 0.7191, loss_cls_dn_5: 0.1739, loss_box_dn_5: 0.7182, loss_dense_depth: 0.7822, loss: 27.4231, grad_norm: 45.3887 -2025-11-17 14:23:24,077 - mmdet - INFO - Iter [137/17500] lr: 1.544e-04, eta: 11:20:36, time: 1.574, data_time: 0.078, memory: 49163, loss_cls_0: 0.8790, loss_box_0: 1.7268, loss_cns_0: 0.6157, loss_yns_0: 0.1518, loss_cls_1: 0.9298, loss_box_1: 1.8335, loss_cns_1: 0.6408, loss_yns_1: 0.1525, loss_cls_2: 0.9533, loss_box_2: 1.7811, loss_cns_2: 0.6488, loss_yns_2: 0.1512, loss_cls_3: 0.9621, loss_box_3: 1.7706, loss_cns_3: 0.6464, loss_yns_3: 0.1508, loss_cls_4: 0.9677, loss_box_4: 1.7755, loss_cns_4: 0.6500, loss_yns_4: 0.1513, loss_cls_5: 0.9648, loss_box_5: 1.7721, loss_cns_5: 0.6479, loss_yns_5: 0.1506, loss_cls_dn_0: 0.2384, loss_box_dn_0: 0.7859, loss_cls_dn_1: 0.1496, loss_box_dn_1: 0.7243, loss_cls_dn_2: 0.1521, loss_box_dn_2: 0.7059, loss_cls_dn_3: 0.1598, loss_box_dn_3: 0.7058, loss_cls_dn_4: 0.1656, loss_box_dn_4: 0.7106, loss_cls_dn_5: 0.1691, loss_box_dn_5: 0.7086, loss_dense_depth: 0.7772, loss: 27.2272, grad_norm: 37.3196 -2025-11-17 14:23:25,570 - mmdet - INFO - Iter [138/17500] lr: 1.548e-04, eta: 11:18:46, time: 1.492, data_time: 0.078, memory: 49163, loss_cls_0: 0.8848, loss_box_0: 1.7115, loss_cns_0: 0.6222, loss_yns_0: 0.1557, loss_cls_1: 0.9396, loss_box_1: 1.7922, loss_cns_1: 0.6427, loss_yns_1: 0.1488, loss_cls_2: 0.9681, loss_box_2: 1.7262, loss_cns_2: 0.6494, loss_yns_2: 0.1482, loss_cls_3: 0.9586, loss_box_3: 1.7152, loss_cns_3: 0.6493, loss_yns_3: 0.1491, loss_cls_4: 0.9681, loss_box_4: 1.7185, loss_cns_4: 0.6505, loss_yns_4: 0.1478, loss_cls_5: 0.9657, loss_box_5: 1.7221, loss_cns_5: 0.6512, loss_yns_5: 0.1499, loss_cls_dn_0: 0.2325, loss_box_dn_0: 0.7926, loss_cls_dn_1: 0.1462, loss_box_dn_1: 0.7347, loss_cls_dn_2: 0.1481, loss_box_dn_2: 0.7085, loss_cls_dn_3: 0.1488, loss_box_dn_3: 0.7066, loss_cls_dn_4: 0.1578, loss_box_dn_4: 0.7062, loss_cls_dn_5: 0.1607, loss_box_dn_5: 0.7081, loss_dense_depth: 0.8120, loss: 26.9983, grad_norm: 27.8423 -2025-11-17 14:23:27,061 - mmdet - INFO - Iter [139/17500] lr: 1.552e-04, eta: 11:16:57, time: 1.492, data_time: 0.078, memory: 49163, loss_cls_0: 0.8491, loss_box_0: 1.7055, loss_cns_0: 0.6255, loss_yns_0: 0.1541, loss_cls_1: 0.9325, loss_box_1: 1.7963, loss_cns_1: 0.6449, loss_yns_1: 0.1485, loss_cls_2: 0.9602, loss_box_2: 1.7460, loss_cns_2: 0.6503, loss_yns_2: 0.1497, loss_cls_3: 0.9414, loss_box_3: 1.7370, loss_cns_3: 0.6497, loss_yns_3: 0.1520, loss_cls_4: 0.9491, loss_box_4: 1.7355, loss_cns_4: 0.6501, loss_yns_4: 0.1495, loss_cls_5: 0.9411, loss_box_5: 1.7506, loss_cns_5: 0.6492, loss_yns_5: 0.1517, loss_cls_dn_0: 0.2238, loss_box_dn_0: 0.7877, loss_cls_dn_1: 0.1465, loss_box_dn_1: 0.7268, loss_cls_dn_2: 0.1502, loss_box_dn_2: 0.7015, loss_cls_dn_3: 0.1516, loss_box_dn_3: 0.6990, loss_cls_dn_4: 0.1603, loss_box_dn_4: 0.7008, loss_cls_dn_5: 0.1653, loss_box_dn_5: 0.7088, loss_dense_depth: 0.8426, loss: 26.9845, grad_norm: 39.5632 -2025-11-17 14:23:28,553 - mmdet - INFO - Iter [140/17500] lr: 1.556e-04, eta: 11:15:09, time: 1.491, data_time: 0.080, memory: 49163, loss_cls_0: 0.8347, loss_box_0: 1.7137, loss_cns_0: 0.6259, loss_yns_0: 0.1473, loss_cls_1: 0.9147, loss_box_1: 1.7588, loss_cns_1: 0.6480, loss_yns_1: 0.1478, loss_cls_2: 0.9400, loss_box_2: 1.7206, loss_cns_2: 0.6537, loss_yns_2: 0.1473, loss_cls_3: 0.9344, loss_box_3: 1.7121, loss_cns_3: 0.6536, loss_yns_3: 0.1485, loss_cls_4: 0.9368, loss_box_4: 1.7233, loss_cns_4: 0.6509, loss_yns_4: 0.1471, loss_cls_5: 0.9343, loss_box_5: 1.7304, loss_cns_5: 0.6497, loss_yns_5: 0.1484, loss_cls_dn_0: 0.2203, loss_box_dn_0: 0.7944, loss_cls_dn_1: 0.1429, loss_box_dn_1: 0.7254, loss_cls_dn_2: 0.1461, loss_box_dn_2: 0.7030, loss_cls_dn_3: 0.1492, loss_box_dn_3: 0.7002, loss_cls_dn_4: 0.1525, loss_box_dn_4: 0.7074, loss_cls_dn_5: 0.1575, loss_box_dn_5: 0.7141, loss_dense_depth: 0.7669, loss: 26.7019, grad_norm: 46.5057 -2025-11-17 14:23:30,098 - mmdet - INFO - Iter [141/17500] lr: 1.560e-04, eta: 11:13:30, time: 1.545, data_time: 0.107, memory: 49163, loss_cls_0: 0.8990, loss_box_0: 1.7687, loss_cns_0: 0.6274, loss_yns_0: 0.1512, loss_cls_1: 0.9342, loss_box_1: 1.8316, loss_cns_1: 0.6447, loss_yns_1: 0.1501, loss_cls_2: 0.9533, loss_box_2: 1.7700, loss_cns_2: 0.6533, loss_yns_2: 0.1499, loss_cls_3: 0.9603, loss_box_3: 1.7585, loss_cns_3: 0.6527, loss_yns_3: 0.1499, loss_cls_4: 0.9527, loss_box_4: 1.7675, loss_cns_4: 0.6525, loss_yns_4: 0.1497, loss_cls_5: 0.9637, loss_box_5: 1.7700, loss_cns_5: 0.6506, loss_yns_5: 0.1490, loss_cls_dn_0: 0.2339, loss_box_dn_0: 0.7916, loss_cls_dn_1: 0.1505, loss_box_dn_1: 0.7425, loss_cls_dn_2: 0.1516, loss_box_dn_2: 0.7155, loss_cls_dn_3: 0.1528, loss_box_dn_3: 0.7139, loss_cls_dn_4: 0.1562, loss_box_dn_4: 0.7196, loss_cls_dn_5: 0.1618, loss_box_dn_5: 0.7253, loss_dense_depth: 0.8518, loss: 27.3777, grad_norm: 34.0330 -2025-11-17 14:23:31,692 - mmdet - INFO - Iter [142/17500] lr: 1.564e-04, eta: 11:11:58, time: 1.594, data_time: 0.104, memory: 49163, loss_cls_0: 0.9179, loss_box_0: 1.7923, loss_cns_0: 0.6198, loss_yns_0: 0.1590, loss_cls_1: 0.9668, loss_box_1: 1.8656, loss_cns_1: 0.6387, loss_yns_1: 0.1506, loss_cls_2: 0.9874, loss_box_2: 1.8179, loss_cns_2: 0.6461, loss_yns_2: 0.1517, loss_cls_3: 0.9744, loss_box_3: 1.8126, loss_cns_3: 0.6459, loss_yns_3: 0.1513, loss_cls_4: 0.9674, loss_box_4: 1.8185, loss_cns_4: 0.6457, loss_yns_4: 0.1518, loss_cls_5: 0.9837, loss_box_5: 1.8224, loss_cns_5: 0.6445, loss_yns_5: 0.1529, loss_cls_dn_0: 0.2337, loss_box_dn_0: 0.7864, loss_cls_dn_1: 0.1491, loss_box_dn_1: 0.7387, loss_cls_dn_2: 0.1498, loss_box_dn_2: 0.7157, loss_cls_dn_3: 0.1541, loss_box_dn_3: 0.7199, loss_cls_dn_4: 0.1581, loss_box_dn_4: 0.7227, loss_cls_dn_5: 0.1651, loss_box_dn_5: 0.7303, loss_dense_depth: 0.8143, loss: 27.7229, grad_norm: 40.0733 -2025-11-17 14:23:33,222 - mmdet - INFO - Iter [143/17500] lr: 1.568e-04, eta: 11:10:19, time: 1.531, data_time: 0.076, memory: 49163, loss_cls_0: 0.8593, loss_box_0: 1.7740, loss_cns_0: 0.6200, loss_yns_0: 0.1554, loss_cls_1: 0.9452, loss_box_1: 1.8364, loss_cns_1: 0.6438, loss_yns_1: 0.1494, loss_cls_2: 0.9686, loss_box_2: 1.7981, loss_cns_2: 0.6480, loss_yns_2: 0.1505, loss_cls_3: 0.9600, loss_box_3: 1.7784, loss_cns_3: 0.6491, loss_yns_3: 0.1524, loss_cls_4: 0.9534, loss_box_4: 1.7836, loss_cns_4: 0.6485, loss_yns_4: 0.1503, loss_cls_5: 0.9649, loss_box_5: 1.7975, loss_cns_5: 0.6498, loss_yns_5: 0.1514, loss_cls_dn_0: 0.2231, loss_box_dn_0: 0.7826, loss_cls_dn_1: 0.1504, loss_box_dn_1: 0.7438, loss_cls_dn_2: 0.1490, loss_box_dn_2: 0.7243, loss_cls_dn_3: 0.1585, loss_box_dn_3: 0.7235, loss_cls_dn_4: 0.1612, loss_box_dn_4: 0.7260, loss_cls_dn_5: 0.1669, loss_box_dn_5: 0.7349, loss_dense_depth: 0.7924, loss: 27.4250, grad_norm: 43.8141 -2025-11-17 14:23:34,737 - mmdet - INFO - Iter [144/17500] lr: 1.572e-04, eta: 11:08:40, time: 1.515, data_time: 0.079, memory: 49163, loss_cls_0: 0.8752, loss_box_0: 1.7779, loss_cns_0: 0.6199, loss_yns_0: 0.1513, loss_cls_1: 0.9352, loss_box_1: 1.8581, loss_cns_1: 0.6404, loss_yns_1: 0.1486, loss_cls_2: 0.9691, loss_box_2: 1.8097, loss_cns_2: 0.6494, loss_yns_2: 0.1490, loss_cls_3: 0.9708, loss_box_3: 1.8006, loss_cns_3: 0.6492, loss_yns_3: 0.1509, loss_cls_4: 0.9646, loss_box_4: 1.7935, loss_cns_4: 0.6497, loss_yns_4: 0.1489, loss_cls_5: 0.9739, loss_box_5: 1.8161, loss_cns_5: 0.6479, loss_yns_5: 0.1486, loss_cls_dn_0: 0.2304, loss_box_dn_0: 0.7862, loss_cls_dn_1: 0.1562, loss_box_dn_1: 0.7557, loss_cls_dn_2: 0.1545, loss_box_dn_2: 0.7389, loss_cls_dn_3: 0.1627, loss_box_dn_3: 0.7365, loss_cls_dn_4: 0.1654, loss_box_dn_4: 0.7386, loss_cls_dn_5: 0.1688, loss_box_dn_5: 0.7482, loss_dense_depth: 0.7851, loss: 27.6256, grad_norm: 48.7689 -2025-11-17 14:23:36,246 - mmdet - INFO - Iter [145/17500] lr: 1.576e-04, eta: 11:07:02, time: 1.508, data_time: 0.087, memory: 49163, loss_cls_0: 0.8767, loss_box_0: 1.7825, loss_cns_0: 0.6215, loss_yns_0: 0.1500, loss_cls_1: 0.9232, loss_box_1: 1.8631, loss_cns_1: 0.6397, loss_yns_1: 0.1496, loss_cls_2: 0.9538, loss_box_2: 1.7975, loss_cns_2: 0.6510, loss_yns_2: 0.1500, loss_cls_3: 0.9554, loss_box_3: 1.8014, loss_cns_3: 0.6490, loss_yns_3: 0.1514, loss_cls_4: 0.9538, loss_box_4: 1.7815, loss_cns_4: 0.6509, loss_yns_4: 0.1498, loss_cls_5: 0.9589, loss_box_5: 1.7970, loss_cns_5: 0.6476, loss_yns_5: 0.1494, loss_cls_dn_0: 0.2259, loss_box_dn_0: 0.7818, loss_cls_dn_1: 0.1557, loss_box_dn_1: 0.7650, loss_cls_dn_2: 0.1532, loss_box_dn_2: 0.7359, loss_cls_dn_3: 0.1568, loss_box_dn_3: 0.7341, loss_cls_dn_4: 0.1613, loss_box_dn_4: 0.7335, loss_cls_dn_5: 0.1668, loss_box_dn_5: 0.7368, loss_dense_depth: 0.7960, loss: 27.5076, grad_norm: 46.0184 -2025-11-17 14:23:37,762 - mmdet - INFO - Iter [146/17500] lr: 1.580e-04, eta: 11:05:25, time: 1.508, data_time: 0.084, memory: 49163, loss_cls_0: 0.8661, loss_box_0: 1.7835, loss_cns_0: 0.6197, loss_yns_0: 0.1496, loss_cls_1: 0.9347, loss_box_1: 1.8431, loss_cns_1: 0.6410, loss_yns_1: 0.1479, loss_cls_2: 0.9519, loss_box_2: 1.7745, loss_cns_2: 0.6506, loss_yns_2: 0.1487, loss_cls_3: 0.9604, loss_box_3: 1.7696, loss_cns_3: 0.6578, loss_yns_3: 0.1480, loss_cls_4: 0.9539, loss_box_4: 1.7708, loss_cns_4: 0.6526, loss_yns_4: 0.1486, loss_cls_5: 0.9626, loss_box_5: 1.7846, loss_cns_5: 0.6521, loss_yns_5: 0.1491, loss_cls_dn_0: 0.2228, loss_box_dn_0: 0.7846, loss_cls_dn_1: 0.1493, loss_box_dn_1: 0.7454, loss_cls_dn_2: 0.1481, loss_box_dn_2: 0.7208, loss_cls_dn_3: 0.1505, loss_box_dn_3: 0.7221, loss_cls_dn_4: 0.1566, loss_box_dn_4: 0.7296, loss_cls_dn_5: 0.1656, loss_box_dn_5: 0.7375, loss_dense_depth: 0.7706, loss: 27.3251, grad_norm: 36.1391 -2025-11-17 14:23:39,261 - mmdet - INFO - Iter [147/17500] lr: 1.584e-04, eta: 11:03:49, time: 1.508, data_time: 0.110, memory: 49163, loss_cls_0: 0.8790, loss_box_0: 1.7819, loss_cns_0: 0.6191, loss_yns_0: 0.1540, loss_cls_1: 0.9467, loss_box_1: 1.8132, loss_cns_1: 0.6446, loss_yns_1: 0.1502, loss_cls_2: 0.9653, loss_box_2: 1.7697, loss_cns_2: 0.6508, loss_yns_2: 0.1502, loss_cls_3: 0.9701, loss_box_3: 1.7666, loss_cns_3: 0.6602, loss_yns_3: 0.1507, loss_cls_4: 0.9660, loss_box_4: 1.7787, loss_cns_4: 0.6526, loss_yns_4: 0.1506, loss_cls_5: 0.9763, loss_box_5: 1.7978, loss_cns_5: 0.6525, loss_yns_5: 0.1520, loss_cls_dn_0: 0.2248, loss_box_dn_0: 0.7798, loss_cls_dn_1: 0.1487, loss_box_dn_1: 0.7339, loss_cls_dn_2: 0.1505, loss_box_dn_2: 0.7218, loss_cls_dn_3: 0.1516, loss_box_dn_3: 0.7248, loss_cls_dn_4: 0.1578, loss_box_dn_4: 0.7348, loss_cls_dn_5: 0.1655, loss_box_dn_5: 0.7487, loss_dense_depth: 0.7990, loss: 27.4404, grad_norm: 44.2038 -2025-11-17 14:23:40,758 - mmdet - INFO - Iter [148/17500] lr: 1.588e-04, eta: 11:02:13, time: 1.497, data_time: 0.081, memory: 49163, loss_cls_0: 0.8523, loss_box_0: 1.7679, loss_cns_0: 0.6183, loss_yns_0: 0.1530, loss_cls_1: 0.9376, loss_box_1: 1.7849, loss_cns_1: 0.6478, loss_yns_1: 0.1470, loss_cls_2: 0.9590, loss_box_2: 1.7347, loss_cns_2: 0.6517, loss_yns_2: 0.1479, loss_cls_3: 0.9571, loss_box_3: 1.7335, loss_cns_3: 0.6551, loss_yns_3: 0.1480, loss_cls_4: 0.9452, loss_box_4: 1.7362, loss_cns_4: 0.6528, loss_yns_4: 0.1486, loss_cls_5: 0.9549, loss_box_5: 1.7344, loss_cns_5: 0.6534, loss_yns_5: 0.1497, loss_cls_dn_0: 0.2202, loss_box_dn_0: 0.7810, loss_cls_dn_1: 0.1470, loss_box_dn_1: 0.7362, loss_cls_dn_2: 0.1491, loss_box_dn_2: 0.7203, loss_cls_dn_3: 0.1500, loss_box_dn_3: 0.7240, loss_cls_dn_4: 0.1553, loss_box_dn_4: 0.7306, loss_cls_dn_5: 0.1618, loss_box_dn_5: 0.7378, loss_dense_depth: 0.7531, loss: 27.0374, grad_norm: 35.5597 -2025-11-17 14:23:42,275 - mmdet - INFO - Iter [149/17500] lr: 1.592e-04, eta: 11:00:40, time: 1.516, data_time: 0.093, memory: 49163, loss_cls_0: 0.8531, loss_box_0: 1.7639, loss_cns_0: 0.6192, loss_yns_0: 0.1498, loss_cls_1: 0.9268, loss_box_1: 1.7823, loss_cns_1: 0.6456, loss_yns_1: 0.1488, loss_cls_2: 0.9505, loss_box_2: 1.7293, loss_cns_2: 0.6492, loss_yns_2: 0.1480, loss_cls_3: 0.9549, loss_box_3: 1.7228, loss_cns_3: 0.6484, loss_yns_3: 0.1480, loss_cls_4: 0.9471, loss_box_4: 1.7324, loss_cns_4: 0.6479, loss_yns_4: 0.1478, loss_cls_5: 0.9603, loss_box_5: 1.7286, loss_cns_5: 0.6485, loss_yns_5: 0.1485, loss_cls_dn_0: 0.2192, loss_box_dn_0: 0.7771, loss_cls_dn_1: 0.1443, loss_box_dn_1: 0.7389, loss_cls_dn_2: 0.1462, loss_box_dn_2: 0.7153, loss_cls_dn_3: 0.1471, loss_box_dn_3: 0.7102, loss_cls_dn_4: 0.1510, loss_box_dn_4: 0.7173, loss_cls_dn_5: 0.1571, loss_box_dn_5: 0.7215, loss_dense_depth: 0.7560, loss: 26.9030, grad_norm: 36.3794 -2025-11-17 14:23:43,773 - mmdet - INFO - Iter [150/17500] lr: 1.596e-04, eta: 10:59:07, time: 1.496, data_time: 0.076, memory: 49163, loss_cls_0: 0.8738, loss_box_0: 1.7504, loss_cns_0: 0.6200, loss_yns_0: 0.1490, loss_cls_1: 0.9204, loss_box_1: 1.7898, loss_cns_1: 0.6469, loss_yns_1: 0.1474, loss_cls_2: 0.9589, loss_box_2: 1.7288, loss_cns_2: 0.6549, loss_yns_2: 0.1457, loss_cls_3: 0.9651, loss_box_3: 1.7251, loss_cns_3: 0.6541, loss_yns_3: 0.1463, loss_cls_4: 0.9582, loss_box_4: 1.7192, loss_cns_4: 0.6550, loss_yns_4: 0.1465, loss_cls_5: 0.9646, loss_box_5: 1.7304, loss_cns_5: 0.6560, loss_yns_5: 0.1465, loss_cls_dn_0: 0.2201, loss_box_dn_0: 0.7747, loss_cls_dn_1: 0.1450, loss_box_dn_1: 0.7165, loss_cls_dn_2: 0.1496, loss_box_dn_2: 0.6917, loss_cls_dn_3: 0.1507, loss_box_dn_3: 0.6906, loss_cls_dn_4: 0.1545, loss_box_dn_4: 0.6901, loss_cls_dn_5: 0.1594, loss_box_dn_5: 0.7008, loss_dense_depth: 0.7692, loss: 26.8659, grad_norm: 30.0119 -2025-11-17 14:23:45,270 - mmdet - INFO - Iter [151/17500] lr: 1.600e-04, eta: 10:57:35, time: 1.500, data_time: 0.079, memory: 49163, loss_cls_0: 0.8760, loss_box_0: 1.7447, loss_cns_0: 0.6218, loss_yns_0: 0.1498, loss_cls_1: 0.9358, loss_box_1: 1.7927, loss_cns_1: 0.6448, loss_yns_1: 0.1494, loss_cls_2: 0.9545, loss_box_2: 1.7442, loss_cns_2: 0.6511, loss_yns_2: 0.1474, loss_cls_3: 0.9615, loss_box_3: 1.7469, loss_cns_3: 0.6538, loss_yns_3: 0.1472, loss_cls_4: 0.9598, loss_box_4: 1.7377, loss_cns_4: 0.6533, loss_yns_4: 0.1478, loss_cls_5: 0.9678, loss_box_5: 1.7374, loss_cns_5: 0.6530, loss_yns_5: 0.1493, loss_cls_dn_0: 0.2191, loss_box_dn_0: 0.7786, loss_cls_dn_1: 0.1426, loss_box_dn_1: 0.7223, loss_cls_dn_2: 0.1453, loss_box_dn_2: 0.6984, loss_cls_dn_3: 0.1457, loss_box_dn_3: 0.7029, loss_cls_dn_4: 0.1493, loss_box_dn_4: 0.7016, loss_cls_dn_5: 0.1576, loss_box_dn_5: 0.7097, loss_dense_depth: 0.7729, loss: 26.9735, grad_norm: 34.4131 -2025-11-17 14:23:46,762 - mmdet - INFO - Iter [152/17500] lr: 1.604e-04, eta: 10:56:03, time: 1.491, data_time: 0.076, memory: 49163, loss_cls_0: 0.8625, loss_box_0: 1.7145, loss_cns_0: 0.6206, loss_yns_0: 0.1534, loss_cls_1: 0.9430, loss_box_1: 1.8124, loss_cns_1: 0.6440, loss_yns_1: 0.1505, loss_cls_2: 0.9556, loss_box_2: 1.7688, loss_cns_2: 0.6476, loss_yns_2: 0.1505, loss_cls_3: 0.9565, loss_box_3: 1.7685, loss_cns_3: 0.6565, loss_yns_3: 0.1500, loss_cls_4: 0.9695, loss_box_4: 1.7766, loss_cns_4: 0.6539, loss_yns_4: 0.1520, loss_cls_5: 0.9686, loss_box_5: 1.7880, loss_cns_5: 0.6535, loss_yns_5: 0.1523, loss_cls_dn_0: 0.2173, loss_box_dn_0: 0.7742, loss_cls_dn_1: 0.1438, loss_box_dn_1: 0.7302, loss_cls_dn_2: 0.1430, loss_box_dn_2: 0.7113, loss_cls_dn_3: 0.1444, loss_box_dn_3: 0.7164, loss_cls_dn_4: 0.1501, loss_box_dn_4: 0.7270, loss_cls_dn_5: 0.1598, loss_box_dn_5: 0.7397, loss_dense_depth: 0.7430, loss: 27.1699, grad_norm: 36.3343 -2025-11-17 14:23:48,257 - mmdet - INFO - Iter [153/17500] lr: 1.608e-04, eta: 10:54:33, time: 1.494, data_time: 0.076, memory: 49163, loss_cls_0: 0.8919, loss_box_0: 1.7194, loss_cns_0: 0.6150, loss_yns_0: 0.1500, loss_cls_1: 0.9318, loss_box_1: 1.8128, loss_cns_1: 0.6439, loss_yns_1: 0.1516, loss_cls_2: 0.9637, loss_box_2: 1.7608, loss_cns_2: 0.6508, loss_yns_2: 0.1524, loss_cls_3: 0.9642, loss_box_3: 1.7575, loss_cns_3: 0.6547, loss_yns_3: 0.1512, loss_cls_4: 0.9582, loss_box_4: 1.7618, loss_cns_4: 0.6507, loss_yns_4: 0.1519, loss_cls_5: 0.9796, loss_box_5: 1.7692, loss_cns_5: 0.6519, loss_yns_5: 0.1513, loss_cls_dn_0: 0.2272, loss_box_dn_0: 0.7827, loss_cls_dn_1: 0.1416, loss_box_dn_1: 0.7462, loss_cls_dn_2: 0.1425, loss_box_dn_2: 0.7294, loss_cls_dn_3: 0.1452, loss_box_dn_3: 0.7378, loss_cls_dn_4: 0.1507, loss_box_dn_4: 0.7523, loss_cls_dn_5: 0.1617, loss_box_dn_5: 0.7658, loss_dense_depth: 0.7646, loss: 27.2935, grad_norm: 39.8934 -2025-11-17 14:23:49,757 - mmdet - INFO - Iter [154/17500] lr: 1.612e-04, eta: 10:53:05, time: 1.499, data_time: 0.079, memory: 49163, loss_cls_0: 0.8592, loss_box_0: 1.7091, loss_cns_0: 0.6184, loss_yns_0: 0.1502, loss_cls_1: 0.9088, loss_box_1: 1.7453, loss_cns_1: 0.6448, loss_yns_1: 0.1498, loss_cls_2: 0.9318, loss_box_2: 1.6934, loss_cns_2: 0.6506, loss_yns_2: 0.1494, loss_cls_3: 0.9348, loss_box_3: 1.6995, loss_cns_3: 0.6523, loss_yns_3: 0.1511, loss_cls_4: 0.9487, loss_box_4: 1.6966, loss_cns_4: 0.6513, loss_yns_4: 0.1498, loss_cls_5: 0.9508, loss_box_5: 1.6923, loss_cns_5: 0.6533, loss_yns_5: 0.1505, loss_cls_dn_0: 0.2202, loss_box_dn_0: 0.7740, loss_cls_dn_1: 0.1394, loss_box_dn_1: 0.7496, loss_cls_dn_2: 0.1411, loss_box_dn_2: 0.7246, loss_cls_dn_3: 0.1429, loss_box_dn_3: 0.7358, loss_cls_dn_4: 0.1475, loss_box_dn_4: 0.7442, loss_cls_dn_5: 0.1530, loss_box_dn_5: 0.7500, loss_dense_depth: 0.7483, loss: 26.7126, grad_norm: 41.9203 -2025-11-17 14:23:51,323 - mmdet - INFO - Iter [155/17500] lr: 1.616e-04, eta: 10:51:45, time: 1.567, data_time: 0.084, memory: 49163, loss_cls_0: 0.8595, loss_box_0: 1.7136, loss_cns_0: 0.6230, loss_yns_0: 0.1532, loss_cls_1: 0.9066, loss_box_1: 1.6968, loss_cns_1: 0.6506, loss_yns_1: 0.1495, loss_cls_2: 0.9423, loss_box_2: 1.6665, loss_cns_2: 0.6544, loss_yns_2: 0.1488, loss_cls_3: 0.9390, loss_box_3: 1.6588, loss_cns_3: 0.6618, loss_yns_3: 0.1496, loss_cls_4: 0.9471, loss_box_4: 1.6498, loss_cns_4: 0.6632, loss_yns_4: 0.1486, loss_cls_5: 0.9267, loss_box_5: 1.6546, loss_cns_5: 0.6575, loss_yns_5: 0.1507, loss_cls_dn_0: 0.2153, loss_box_dn_0: 0.7788, loss_cls_dn_1: 0.1374, loss_box_dn_1: 0.7423, loss_cls_dn_2: 0.1402, loss_box_dn_2: 0.7220, loss_cls_dn_3: 0.1428, loss_box_dn_3: 0.7240, loss_cls_dn_4: 0.1460, loss_box_dn_4: 0.7249, loss_cls_dn_5: 0.1490, loss_box_dn_5: 0.7303, loss_dense_depth: 0.7367, loss: 26.4617, grad_norm: 35.8051 -2025-11-17 14:23:52,806 - mmdet - INFO - Iter [156/17500] lr: 1.620e-04, eta: 10:50:17, time: 1.483, data_time: 0.077, memory: 49163, loss_cls_0: 0.8428, loss_box_0: 1.7274, loss_cns_0: 0.6171, loss_yns_0: 0.1529, loss_cls_1: 0.9146, loss_box_1: 1.7057, loss_cns_1: 0.6473, loss_yns_1: 0.1511, loss_cls_2: 0.9278, loss_box_2: 1.6769, loss_cns_2: 0.6521, loss_yns_2: 0.1528, loss_cls_3: 0.9275, loss_box_3: 1.6673, loss_cns_3: 0.6603, loss_yns_3: 0.1519, loss_cls_4: 0.9454, loss_box_4: 1.6688, loss_cns_4: 0.6575, loss_yns_4: 0.1515, loss_cls_5: 0.9336, loss_box_5: 1.6728, loss_cns_5: 0.6550, loss_yns_5: 0.1509, loss_cls_dn_0: 0.2212, loss_box_dn_0: 0.7780, loss_cls_dn_1: 0.1385, loss_box_dn_1: 0.7538, loss_cls_dn_2: 0.1390, loss_box_dn_2: 0.7386, loss_cls_dn_3: 0.1433, loss_box_dn_3: 0.7355, loss_cls_dn_4: 0.1519, loss_box_dn_4: 0.7364, loss_cls_dn_5: 0.1530, loss_box_dn_5: 0.7445, loss_dense_depth: 0.7579, loss: 26.6027, grad_norm: 34.9708 -2025-11-17 14:23:54,346 - mmdet - INFO - Iter [157/17500] lr: 1.624e-04, eta: 10:48:56, time: 1.539, data_time: 0.076, memory: 49163, loss_cls_0: 0.8519, loss_box_0: 1.7148, loss_cns_0: 0.6187, loss_yns_0: 0.1533, loss_cls_1: 0.9153, loss_box_1: 1.7407, loss_cns_1: 0.6455, loss_yns_1: 0.1514, loss_cls_2: 0.9306, loss_box_2: 1.6960, loss_cns_2: 0.6492, loss_yns_2: 0.1526, loss_cls_3: 0.9293, loss_box_3: 1.6961, loss_cns_3: 0.6521, loss_yns_3: 0.1528, loss_cls_4: 0.9317, loss_box_4: 1.6948, loss_cns_4: 0.6492, loss_yns_4: 0.1519, loss_cls_5: 0.9378, loss_box_5: 1.6986, loss_cns_5: 0.6495, loss_yns_5: 0.1528, loss_cls_dn_0: 0.2152, loss_box_dn_0: 0.7730, loss_cls_dn_1: 0.1397, loss_box_dn_1: 0.7333, loss_cls_dn_2: 0.1386, loss_box_dn_2: 0.7166, loss_cls_dn_3: 0.1424, loss_box_dn_3: 0.7185, loss_cls_dn_4: 0.1476, loss_box_dn_4: 0.7208, loss_cls_dn_5: 0.1515, loss_box_dn_5: 0.7281, loss_dense_depth: 0.7382, loss: 26.5800, grad_norm: 37.5921 -2025-11-17 14:23:55,840 - mmdet - INFO - Iter [158/17500] lr: 1.628e-04, eta: 10:47:31, time: 1.492, data_time: 0.075, memory: 49163, loss_cls_0: 0.8353, loss_box_0: 1.7320, loss_cns_0: 0.6177, loss_yns_0: 0.1539, loss_cls_1: 0.9118, loss_box_1: 1.7455, loss_cns_1: 0.6472, loss_yns_1: 0.1542, loss_cls_2: 0.9348, loss_box_2: 1.6990, loss_cns_2: 0.6511, loss_yns_2: 0.1540, loss_cls_3: 0.9306, loss_box_3: 1.6903, loss_cns_3: 0.6524, loss_yns_3: 0.1540, loss_cls_4: 0.9388, loss_box_4: 1.6941, loss_cns_4: 0.6510, loss_yns_4: 0.1528, loss_cls_5: 0.9363, loss_box_5: 1.7024, loss_cns_5: 0.6522, loss_yns_5: 0.1539, loss_cls_dn_0: 0.2078, loss_box_dn_0: 0.7759, loss_cls_dn_1: 0.1392, loss_box_dn_1: 0.7419, loss_cls_dn_2: 0.1390, loss_box_dn_2: 0.7250, loss_cls_dn_3: 0.1390, loss_box_dn_3: 0.7295, loss_cls_dn_4: 0.1445, loss_box_dn_4: 0.7350, loss_cls_dn_5: 0.1482, loss_box_dn_5: 0.7469, loss_dense_depth: 0.7210, loss: 26.6382, grad_norm: 36.3201 -2025-11-17 14:23:57,335 - mmdet - INFO - Iter [159/17500] lr: 1.632e-04, eta: 10:46:08, time: 1.497, data_time: 0.080, memory: 49163, loss_cls_0: 0.8409, loss_box_0: 1.7427, loss_cns_0: 0.6186, loss_yns_0: 0.1545, loss_cls_1: 0.9132, loss_box_1: 1.7644, loss_cns_1: 0.6494, loss_yns_1: 0.1553, loss_cls_2: 0.9409, loss_box_2: 1.7154, loss_cns_2: 0.6541, loss_yns_2: 0.1539, loss_cls_3: 0.9398, loss_box_3: 1.7097, loss_cns_3: 0.6562, loss_yns_3: 0.1529, loss_cls_4: 0.9408, loss_box_4: 1.7090, loss_cns_4: 0.6550, loss_yns_4: 0.1530, loss_cls_5: 0.9347, loss_box_5: 1.7404, loss_cns_5: 0.6548, loss_yns_5: 0.1522, loss_cls_dn_0: 0.2103, loss_box_dn_0: 0.7757, loss_cls_dn_1: 0.1367, loss_box_dn_1: 0.7543, loss_cls_dn_2: 0.1393, loss_box_dn_2: 0.7355, loss_cls_dn_3: 0.1406, loss_box_dn_3: 0.7418, loss_cls_dn_4: 0.1467, loss_box_dn_4: 0.7505, loss_cls_dn_5: 0.1528, loss_box_dn_5: 0.7680, loss_dense_depth: 0.7353, loss: 26.8894, grad_norm: 45.0796 -2025-11-17 14:23:58,840 - mmdet - INFO - Iter [160/17500] lr: 1.636e-04, eta: 10:44:47, time: 1.506, data_time: 0.083, memory: 49163, loss_cls_0: 0.8336, loss_box_0: 1.7238, loss_cns_0: 0.6191, loss_yns_0: 0.1539, loss_cls_1: 0.8996, loss_box_1: 1.7665, loss_cns_1: 0.6471, loss_yns_1: 0.1541, loss_cls_2: 0.9229, loss_box_2: 1.7271, loss_cns_2: 0.6535, loss_yns_2: 0.1542, loss_cls_3: 0.9316, loss_box_3: 1.7093, loss_cns_3: 0.6570, loss_yns_3: 0.1540, loss_cls_4: 0.9330, loss_box_4: 1.7065, loss_cns_4: 0.6597, loss_yns_4: 0.1535, loss_cls_5: 0.9265, loss_box_5: 1.7203, loss_cns_5: 0.6560, loss_yns_5: 0.1543, loss_cls_dn_0: 0.2081, loss_box_dn_0: 0.7717, loss_cls_dn_1: 0.1380, loss_box_dn_1: 0.7622, loss_cls_dn_2: 0.1399, loss_box_dn_2: 0.7487, loss_cls_dn_3: 0.1409, loss_box_dn_3: 0.7523, loss_cls_dn_4: 0.1494, loss_box_dn_4: 0.7596, loss_cls_dn_5: 0.1514, loss_box_dn_5: 0.7661, loss_dense_depth: 0.7291, loss: 26.8343, grad_norm: 46.1102 -2025-11-17 14:24:00,371 - mmdet - INFO - Iter [161/17500] lr: 1.640e-04, eta: 10:43:29, time: 1.531, data_time: 0.114, memory: 49163, loss_cls_0: 0.8283, loss_box_0: 1.7354, loss_cns_0: 0.6209, loss_yns_0: 0.1554, loss_cls_1: 0.9037, loss_box_1: 1.7408, loss_cns_1: 0.6532, loss_yns_1: 0.1535, loss_cls_2: 0.9299, loss_box_2: 1.7073, loss_cns_2: 0.6574, loss_yns_2: 0.1545, loss_cls_3: 0.9266, loss_box_3: 1.7058, loss_cns_3: 0.6595, loss_yns_3: 0.1539, loss_cls_4: 0.9445, loss_box_4: 1.7019, loss_cns_4: 0.6621, loss_yns_4: 0.1548, loss_cls_5: 0.9320, loss_box_5: 1.6981, loss_cns_5: 0.6566, loss_yns_5: 0.1538, loss_cls_dn_0: 0.2076, loss_box_dn_0: 0.7706, loss_cls_dn_1: 0.1395, loss_box_dn_1: 0.7478, loss_cls_dn_2: 0.1402, loss_box_dn_2: 0.7360, loss_cls_dn_3: 0.1420, loss_box_dn_3: 0.7444, loss_cls_dn_4: 0.1486, loss_box_dn_4: 0.7514, loss_cls_dn_5: 0.1481, loss_box_dn_5: 0.7505, loss_dense_depth: 0.7255, loss: 26.7421, grad_norm: 46.0896 -2025-11-17 14:24:01,914 - mmdet - INFO - Iter [162/17500] lr: 1.644e-04, eta: 10:42:14, time: 1.541, data_time: 0.102, memory: 49163, loss_cls_0: 0.8377, loss_box_0: 1.7473, loss_cns_0: 0.6180, loss_yns_0: 0.1559, loss_cls_1: 0.9057, loss_box_1: 1.7426, loss_cns_1: 0.6475, loss_yns_1: 0.1562, loss_cls_2: 0.9264, loss_box_2: 1.6979, loss_cns_2: 0.6523, loss_yns_2: 0.1562, loss_cls_3: 0.9270, loss_box_3: 1.7043, loss_cns_3: 0.6549, loss_yns_3: 0.1570, loss_cls_4: 0.9326, loss_box_4: 1.6994, loss_cns_4: 0.6519, loss_yns_4: 0.1580, loss_cls_5: 0.9272, loss_box_5: 1.7070, loss_cns_5: 0.6512, loss_yns_5: 0.1569, loss_cls_dn_0: 0.2096, loss_box_dn_0: 0.7781, loss_cls_dn_1: 0.1316, loss_box_dn_1: 0.7293, loss_cls_dn_2: 0.1328, loss_box_dn_2: 0.7080, loss_cls_dn_3: 0.1332, loss_box_dn_3: 0.7147, loss_cls_dn_4: 0.1375, loss_box_dn_4: 0.7192, loss_cls_dn_5: 0.1415, loss_box_dn_5: 0.7253, loss_dense_depth: 0.7178, loss: 26.5497, grad_norm: 37.2206 -2025-11-17 14:24:03,437 - mmdet - INFO - Iter [163/17500] lr: 1.648e-04, eta: 10:40:57, time: 1.525, data_time: 0.081, memory: 49163, loss_cls_0: 0.8314, loss_box_0: 1.7660, loss_cns_0: 0.6211, loss_yns_0: 0.1569, loss_cls_1: 0.9107, loss_box_1: 1.7606, loss_cns_1: 0.6496, loss_yns_1: 0.1578, loss_cls_2: 0.9390, loss_box_2: 1.7163, loss_cns_2: 0.6535, loss_yns_2: 0.1568, loss_cls_3: 0.9240, loss_box_3: 1.7073, loss_cns_3: 0.6547, loss_yns_3: 0.1579, loss_cls_4: 0.9281, loss_box_4: 1.7026, loss_cns_4: 0.6556, loss_yns_4: 0.1580, loss_cls_5: 0.9244, loss_box_5: 1.7189, loss_cns_5: 0.6519, loss_yns_5: 0.1579, loss_cls_dn_0: 0.2090, loss_box_dn_0: 0.7706, loss_cls_dn_1: 0.1348, loss_box_dn_1: 0.7148, loss_cls_dn_2: 0.1378, loss_box_dn_2: 0.6937, loss_cls_dn_3: 0.1394, loss_box_dn_3: 0.6939, loss_cls_dn_4: 0.1421, loss_box_dn_4: 0.6979, loss_cls_dn_5: 0.1445, loss_box_dn_5: 0.7066, loss_dense_depth: 0.7175, loss: 26.5639, grad_norm: 39.8112 -2025-11-17 14:24:04,947 - mmdet - INFO - Iter [164/17500] lr: 1.652e-04, eta: 10:39:40, time: 1.511, data_time: 0.081, memory: 49163, loss_cls_0: 0.8352, loss_box_0: 1.7742, loss_cns_0: 0.6240, loss_yns_0: 0.1590, loss_cls_1: 0.9068, loss_box_1: 1.7570, loss_cns_1: 0.6523, loss_yns_1: 0.1587, loss_cls_2: 0.9292, loss_box_2: 1.7132, loss_cns_2: 0.6555, loss_yns_2: 0.1584, loss_cls_3: 0.9215, loss_box_3: 1.6994, loss_cns_3: 0.6571, loss_yns_3: 0.1593, loss_cls_4: 0.9338, loss_box_4: 1.6984, loss_cns_4: 0.6616, loss_yns_4: 0.1590, loss_cls_5: 0.9346, loss_box_5: 1.7036, loss_cns_5: 0.6578, loss_yns_5: 0.1595, loss_cls_dn_0: 0.2088, loss_box_dn_0: 0.7748, loss_cls_dn_1: 0.1318, loss_box_dn_1: 0.7058, loss_cls_dn_2: 0.1355, loss_box_dn_2: 0.6887, loss_cls_dn_3: 0.1371, loss_box_dn_3: 0.6880, loss_cls_dn_4: 0.1423, loss_box_dn_4: 0.6954, loss_cls_dn_5: 0.1441, loss_box_dn_5: 0.7014, loss_dense_depth: 0.7243, loss: 26.5470, grad_norm: 39.8943 -2025-11-17 14:24:06,465 - mmdet - INFO - Iter [165/17500] lr: 1.656e-04, eta: 10:38:25, time: 1.517, data_time: 0.084, memory: 49163, loss_cls_0: 0.8546, loss_box_0: 1.7881, loss_cns_0: 0.6176, loss_yns_0: 0.1605, loss_cls_1: 0.9051, loss_box_1: 1.7414, loss_cns_1: 0.6481, loss_yns_1: 0.1615, loss_cls_2: 0.9189, loss_box_2: 1.7148, loss_cns_2: 0.6536, loss_yns_2: 0.1593, loss_cls_3: 0.9262, loss_box_3: 1.7008, loss_cns_3: 0.6549, loss_yns_3: 0.1591, loss_cls_4: 0.9393, loss_box_4: 1.7012, loss_cns_4: 0.6574, loss_yns_4: 0.1598, loss_cls_5: 0.9369, loss_box_5: 1.7020, loss_cns_5: 0.6548, loss_yns_5: 0.1607, loss_cls_dn_0: 0.2089, loss_box_dn_0: 0.7703, loss_cls_dn_1: 0.1295, loss_box_dn_1: 0.7157, loss_cls_dn_2: 0.1318, loss_box_dn_2: 0.7046, loss_cls_dn_3: 0.1307, loss_box_dn_3: 0.7072, loss_cls_dn_4: 0.1367, loss_box_dn_4: 0.7207, loss_cls_dn_5: 0.1400, loss_box_dn_5: 0.7312, loss_dense_depth: 0.7408, loss: 26.6446, grad_norm: 37.2160 -2025-11-17 14:24:07,971 - mmdet - INFO - Iter [166/17500] lr: 1.660e-04, eta: 10:37:09, time: 1.506, data_time: 0.088, memory: 49163, loss_cls_0: 0.8273, loss_box_0: 1.8030, loss_cns_0: 0.6123, loss_yns_0: 0.1589, loss_cls_1: 0.9097, loss_box_1: 1.7094, loss_cns_1: 0.6531, loss_yns_1: 0.1614, loss_cls_2: 0.9397, loss_box_2: 1.6727, loss_cns_2: 0.6573, loss_yns_2: 0.1600, loss_cls_3: 0.9308, loss_box_3: 1.6642, loss_cns_3: 0.6593, loss_yns_3: 0.1602, loss_cls_4: 0.9252, loss_box_4: 1.6671, loss_cns_4: 0.6598, loss_yns_4: 0.1616, loss_cls_5: 0.9172, loss_box_5: 1.6776, loss_cns_5: 0.6570, loss_yns_5: 0.1612, loss_cls_dn_0: 0.2019, loss_box_dn_0: 0.7654, loss_cls_dn_1: 0.1296, loss_box_dn_1: 0.7360, loss_cls_dn_2: 0.1290, loss_box_dn_2: 0.7232, loss_cls_dn_3: 0.1314, loss_box_dn_3: 0.7308, loss_cls_dn_4: 0.1369, loss_box_dn_4: 0.7470, loss_cls_dn_5: 0.1410, loss_box_dn_5: 0.7633, loss_dense_depth: 0.7260, loss: 26.5675, grad_norm: 44.4173 -2025-11-17 14:24:14,905 - mmdet - INFO - Iter [167/17500] lr: 1.664e-04, eta: 10:45:18, time: 6.935, data_time: 0.112, memory: 49163, loss_cls_0: 0.8791, loss_box_0: 1.7874, loss_cns_0: 0.6065, loss_yns_0: 0.1600, loss_cls_1: 0.9399, loss_box_1: 1.7138, loss_cns_1: 0.6526, loss_yns_1: 0.1588, loss_cls_2: 0.9572, loss_box_2: 1.6934, loss_cns_2: 0.6558, loss_yns_2: 0.1593, loss_cls_3: 0.9486, loss_box_3: 1.6920, loss_cns_3: 0.6567, loss_yns_3: 0.1599, loss_cls_4: 0.9540, loss_box_4: 1.6916, loss_cns_4: 0.6550, loss_yns_4: 0.1631, loss_cls_5: 0.9495, loss_box_5: 1.7060, loss_cns_5: 0.6547, loss_yns_5: 0.1626, loss_cls_dn_0: 0.2129, loss_box_dn_0: 0.7732, loss_cls_dn_1: 0.1370, loss_box_dn_1: 0.7837, loss_cls_dn_2: 0.1359, loss_box_dn_2: 0.7732, loss_cls_dn_3: 0.1382, loss_box_dn_3: 0.7803, loss_cls_dn_4: 0.1427, loss_box_dn_4: 0.7919, loss_cls_dn_5: 0.1477, loss_box_dn_5: 0.8090, loss_dense_depth: 0.7733, loss: 27.1566, grad_norm: 47.4976 -2025-11-17 14:24:16,387 - mmdet - INFO - Iter [168/17500] lr: 1.668e-04, eta: 10:43:58, time: 1.480, data_time: 0.077, memory: 49163, loss_cls_0: 0.8631, loss_box_0: 1.7325, loss_cns_0: 0.6122, loss_yns_0: 0.1593, loss_cls_1: 0.9412, loss_box_1: 1.6952, loss_cns_1: 0.6529, loss_yns_1: 0.1609, loss_cls_2: 0.9527, loss_box_2: 1.6654, loss_cns_2: 0.6550, loss_yns_2: 0.1609, loss_cls_3: 0.9436, loss_box_3: 1.6623, loss_cns_3: 0.6565, loss_yns_3: 0.1614, loss_cls_4: 0.9463, loss_box_4: 1.6648, loss_cns_4: 0.6577, loss_yns_4: 0.1619, loss_cls_5: 0.9445, loss_box_5: 1.6760, loss_cns_5: 0.6581, loss_yns_5: 0.1626, loss_cls_dn_0: 0.2071, loss_box_dn_0: 0.7614, loss_cls_dn_1: 0.1389, loss_box_dn_1: 0.7891, loss_cls_dn_2: 0.1375, loss_box_dn_2: 0.7752, loss_cls_dn_3: 0.1385, loss_box_dn_3: 0.7813, loss_cls_dn_4: 0.1447, loss_box_dn_4: 0.7887, loss_cls_dn_5: 0.1494, loss_box_dn_5: 0.8019, loss_dense_depth: 0.7463, loss: 26.9067, grad_norm: 37.7368 -2025-11-17 14:24:17,895 - mmdet - INFO - Iter [169/17500] lr: 1.672e-04, eta: 10:42:42, time: 1.508, data_time: 0.092, memory: 49163, loss_cls_0: 0.8831, loss_box_0: 1.7649, loss_cns_0: 0.6117, loss_yns_0: 0.1624, loss_cls_1: 0.9465, loss_box_1: 1.7622, loss_cns_1: 0.6425, loss_yns_1: 0.1629, loss_cls_2: 0.9697, loss_box_2: 1.7001, loss_cns_2: 0.6472, loss_yns_2: 0.1622, loss_cls_3: 0.9639, loss_box_3: 1.7015, loss_cns_3: 0.6476, loss_yns_3: 0.1616, loss_cls_4: 0.9667, loss_box_4: 1.7002, loss_cns_4: 0.6535, loss_yns_4: 0.1614, loss_cls_5: 0.9594, loss_box_5: 1.7180, loss_cns_5: 0.6512, loss_yns_5: 0.1616, loss_cls_dn_0: 0.2157, loss_box_dn_0: 0.7673, loss_cls_dn_1: 0.1341, loss_box_dn_1: 0.7777, loss_cls_dn_2: 0.1352, loss_box_dn_2: 0.7531, loss_cls_dn_3: 0.1358, loss_box_dn_3: 0.7589, loss_cls_dn_4: 0.1426, loss_box_dn_4: 0.7653, loss_cls_dn_5: 0.1446, loss_box_dn_5: 0.7702, loss_dense_depth: 0.7984, loss: 27.1608, grad_norm: 49.3716 -2025-11-17 14:24:19,389 - mmdet - INFO - Iter [170/17500] lr: 1.676e-04, eta: 10:41:25, time: 1.494, data_time: 0.080, memory: 49163, loss_cls_0: 0.8746, loss_box_0: 1.7463, loss_cns_0: 0.6206, loss_yns_0: 0.1617, loss_cls_1: 0.9364, loss_box_1: 1.7134, loss_cns_1: 0.6478, loss_yns_1: 0.1613, loss_cls_2: 0.9706, loss_box_2: 1.6423, loss_cns_2: 0.6554, loss_yns_2: 0.1610, loss_cls_3: 0.9639, loss_box_3: 1.6292, loss_cns_3: 0.6567, loss_yns_3: 0.1606, loss_cls_4: 0.9589, loss_box_4: 1.6344, loss_cns_4: 0.6570, loss_yns_4: 0.1613, loss_cls_5: 0.9615, loss_box_5: 1.6501, loss_cns_5: 0.6549, loss_yns_5: 0.1625, loss_cls_dn_0: 0.2149, loss_box_dn_0: 0.7643, loss_cls_dn_1: 0.1364, loss_box_dn_1: 0.7379, loss_cls_dn_2: 0.1398, loss_box_dn_2: 0.7125, loss_cls_dn_3: 0.1402, loss_box_dn_3: 0.7164, loss_cls_dn_4: 0.1450, loss_box_dn_4: 0.7225, loss_cls_dn_5: 0.1461, loss_box_dn_5: 0.7277, loss_dense_depth: 0.7486, loss: 26.5949, grad_norm: 40.6960 -2025-11-17 14:24:20,876 - mmdet - INFO - Iter [171/17500] lr: 1.680e-04, eta: 10:40:08, time: 1.487, data_time: 0.077, memory: 49163, loss_cls_0: 0.8633, loss_box_0: 1.7376, loss_cns_0: 0.6164, loss_yns_0: 0.1591, loss_cls_1: 0.9340, loss_box_1: 1.7107, loss_cns_1: 0.6456, loss_yns_1: 0.1597, loss_cls_2: 0.9648, loss_box_2: 1.6812, loss_cns_2: 0.6523, loss_yns_2: 0.1597, loss_cls_3: 0.9596, loss_box_3: 1.6639, loss_cns_3: 0.6546, loss_yns_3: 0.1604, loss_cls_4: 0.9545, loss_box_4: 1.6696, loss_cns_4: 0.6519, loss_yns_4: 0.1602, loss_cls_5: 0.9528, loss_box_5: 1.6652, loss_cns_5: 0.6539, loss_yns_5: 0.1601, loss_cls_dn_0: 0.2117, loss_box_dn_0: 0.7662, loss_cls_dn_1: 0.1404, loss_box_dn_1: 0.7304, loss_cls_dn_2: 0.1429, loss_box_dn_2: 0.7229, loss_cls_dn_3: 0.1476, loss_box_dn_3: 0.7231, loss_cls_dn_4: 0.1501, loss_box_dn_4: 0.7319, loss_cls_dn_5: 0.1530, loss_box_dn_5: 0.7362, loss_dense_depth: 0.7161, loss: 26.6634, grad_norm: 46.3934 -2025-11-17 14:24:22,365 - mmdet - INFO - Iter [172/17500] lr: 1.684e-04, eta: 10:38:53, time: 1.489, data_time: 0.079, memory: 49163, loss_cls_0: 0.8661, loss_box_0: 1.7490, loss_cns_0: 0.6129, loss_yns_0: 0.1594, loss_cls_1: 0.9346, loss_box_1: 1.7353, loss_cns_1: 0.6467, loss_yns_1: 0.1606, loss_cls_2: 0.9516, loss_box_2: 1.7193, loss_cns_2: 0.6517, loss_yns_2: 0.1614, loss_cls_3: 0.9500, loss_box_3: 1.6985, loss_cns_3: 0.6550, loss_yns_3: 0.1618, loss_cls_4: 0.9543, loss_box_4: 1.6994, loss_cns_4: 0.6537, loss_yns_4: 0.1616, loss_cls_5: 0.9521, loss_box_5: 1.6837, loss_cns_5: 0.6577, loss_yns_5: 0.1614, loss_cls_dn_0: 0.2129, loss_box_dn_0: 0.7575, loss_cls_dn_1: 0.1348, loss_box_dn_1: 0.7304, loss_cls_dn_2: 0.1359, loss_box_dn_2: 0.7305, loss_cls_dn_3: 0.1394, loss_box_dn_3: 0.7316, loss_cls_dn_4: 0.1466, loss_box_dn_4: 0.7429, loss_cls_dn_5: 0.1550, loss_box_dn_5: 0.7443, loss_dense_depth: 0.7285, loss: 26.8283, grad_norm: 52.0756 -2025-11-17 14:24:23,863 - mmdet - INFO - Iter [173/17500] lr: 1.688e-04, eta: 10:37:39, time: 1.498, data_time: 0.080, memory: 49163, loss_cls_0: 0.8534, loss_box_0: 1.7197, loss_cns_0: 0.6175, loss_yns_0: 0.1572, loss_cls_1: 0.9230, loss_box_1: 1.7131, loss_cns_1: 0.6481, loss_yns_1: 0.1593, loss_cls_2: 0.9412, loss_box_2: 1.6985, loss_cns_2: 0.6522, loss_yns_2: 0.1581, loss_cls_3: 0.9554, loss_box_3: 1.6785, loss_cns_3: 0.6516, loss_yns_3: 0.1598, loss_cls_4: 0.9533, loss_box_4: 1.6782, loss_cns_4: 0.6512, loss_yns_4: 0.1582, loss_cls_5: 0.9506, loss_box_5: 1.6697, loss_cns_5: 0.6518, loss_yns_5: 0.1590, loss_cls_dn_0: 0.2129, loss_box_dn_0: 0.7648, loss_cls_dn_1: 0.1359, loss_box_dn_1: 0.7339, loss_cls_dn_2: 0.1381, loss_box_dn_2: 0.7297, loss_cls_dn_3: 0.1406, loss_box_dn_3: 0.7297, loss_cls_dn_4: 0.1446, loss_box_dn_4: 0.7402, loss_cls_dn_5: 0.1487, loss_box_dn_5: 0.7463, loss_dense_depth: 0.7293, loss: 26.6531, grad_norm: 42.5569 -2025-11-17 14:24:25,349 - mmdet - INFO - Iter [174/17500] lr: 1.692e-04, eta: 10:36:25, time: 1.486, data_time: 0.081, memory: 49163, loss_cls_0: 0.8628, loss_box_0: 1.7452, loss_cns_0: 0.6079, loss_yns_0: 0.1554, loss_cls_1: 0.9166, loss_box_1: 1.7161, loss_cns_1: 0.6388, loss_yns_1: 0.1575, loss_cls_2: 0.9373, loss_box_2: 1.6897, loss_cns_2: 0.6523, loss_yns_2: 0.1572, loss_cls_3: 0.9458, loss_box_3: 1.6791, loss_cns_3: 0.6492, loss_yns_3: 0.1564, loss_cls_4: 0.9396, loss_box_4: 1.6802, loss_cns_4: 0.6457, loss_yns_4: 0.1571, loss_cls_5: 0.9475, loss_box_5: 1.6793, loss_cns_5: 0.6477, loss_yns_5: 0.1575, loss_cls_dn_0: 0.2110, loss_box_dn_0: 0.7746, loss_cls_dn_1: 0.1367, loss_box_dn_1: 0.7430, loss_cls_dn_2: 0.1415, loss_box_dn_2: 0.7289, loss_cls_dn_3: 0.1472, loss_box_dn_3: 0.7311, loss_cls_dn_4: 0.1474, loss_box_dn_4: 0.7395, loss_cls_dn_5: 0.1477, loss_box_dn_5: 0.7494, loss_dense_depth: 0.7426, loss: 26.6627, grad_norm: 34.5213 -2025-11-17 14:24:26,888 - mmdet - INFO - Iter [175/17500] lr: 1.696e-04, eta: 10:35:17, time: 1.538, data_time: 0.084, memory: 49163, loss_cls_0: 0.8554, loss_box_0: 1.7361, loss_cns_0: 0.6178, loss_yns_0: 0.1576, loss_cls_1: 0.9231, loss_box_1: 1.6865, loss_cns_1: 0.6490, loss_yns_1: 0.1570, loss_cls_2: 0.9370, loss_box_2: 1.6766, loss_cns_2: 0.6536, loss_yns_2: 0.1564, loss_cls_3: 0.9395, loss_box_3: 1.6586, loss_cns_3: 0.6562, loss_yns_3: 0.1559, loss_cls_4: 0.9473, loss_box_4: 1.6584, loss_cns_4: 0.6557, loss_yns_4: 0.1566, loss_cls_5: 0.9417, loss_box_5: 1.6721, loss_cns_5: 0.6554, loss_yns_5: 0.1560, loss_cls_dn_0: 0.2133, loss_box_dn_0: 0.7568, loss_cls_dn_1: 0.1337, loss_box_dn_1: 0.7299, loss_cls_dn_2: 0.1397, loss_box_dn_2: 0.7200, loss_cls_dn_3: 0.1451, loss_box_dn_3: 0.7209, loss_cls_dn_4: 0.1422, loss_box_dn_4: 0.7312, loss_cls_dn_5: 0.1476, loss_box_dn_5: 0.7395, loss_dense_depth: 0.7403, loss: 26.5196, grad_norm: 45.0810 -2025-11-17 14:24:28,375 - mmdet - INFO - Iter [176/17500] lr: 1.700e-04, eta: 10:34:05, time: 1.488, data_time: 0.082, memory: 49163, loss_cls_0: 0.8639, loss_box_0: 1.7243, loss_cns_0: 0.6214, loss_yns_0: 0.1554, loss_cls_1: 0.9265, loss_box_1: 1.6759, loss_cns_1: 0.6496, loss_yns_1: 0.1536, loss_cls_2: 0.9337, loss_box_2: 1.6854, loss_cns_2: 0.6537, loss_yns_2: 0.1525, loss_cls_3: 0.9334, loss_box_3: 1.6591, loss_cns_3: 0.6574, loss_yns_3: 0.1532, loss_cls_4: 0.9430, loss_box_4: 1.6639, loss_cns_4: 0.6577, loss_yns_4: 0.1535, loss_cls_5: 0.9386, loss_box_5: 1.6658, loss_cns_5: 0.6555, loss_yns_5: 0.1525, loss_cls_dn_0: 0.2141, loss_box_dn_0: 0.7635, loss_cls_dn_1: 0.1379, loss_box_dn_1: 0.7328, loss_cls_dn_2: 0.1411, loss_box_dn_2: 0.7292, loss_cls_dn_3: 0.1387, loss_box_dn_3: 0.7245, loss_cls_dn_4: 0.1410, loss_box_dn_4: 0.7337, loss_cls_dn_5: 0.1455, loss_box_dn_5: 0.7393, loss_dense_depth: 0.7681, loss: 26.5388, grad_norm: 40.1366 -2025-11-17 14:24:29,917 - mmdet - INFO - Iter [177/17500] lr: 1.704e-04, eta: 10:32:58, time: 1.541, data_time: 0.080, memory: 49163, loss_cls_0: 0.8553, loss_box_0: 1.7309, loss_cns_0: 0.6214, loss_yns_0: 0.1522, loss_cls_1: 0.9172, loss_box_1: 1.6607, loss_cns_1: 0.6534, loss_yns_1: 0.1526, loss_cls_2: 0.9257, loss_box_2: 1.6532, loss_cns_2: 0.6598, loss_yns_2: 0.1516, loss_cls_3: 0.9268, loss_box_3: 1.6374, loss_cns_3: 0.6584, loss_yns_3: 0.1536, loss_cls_4: 0.9268, loss_box_4: 1.6391, loss_cns_4: 0.6571, loss_yns_4: 0.1514, loss_cls_5: 0.9311, loss_box_5: 1.6371, loss_cns_5: 0.6559, loss_yns_5: 0.1509, loss_cls_dn_0: 0.2120, loss_box_dn_0: 0.7600, loss_cls_dn_1: 0.1353, loss_box_dn_1: 0.7379, loss_cls_dn_2: 0.1348, loss_box_dn_2: 0.7267, loss_cls_dn_3: 0.1373, loss_box_dn_3: 0.7227, loss_cls_dn_4: 0.1475, loss_box_dn_4: 0.7289, loss_cls_dn_5: 0.1433, loss_box_dn_5: 0.7342, loss_dense_depth: 0.7697, loss: 26.3498, grad_norm: 37.3558 -2025-11-17 14:24:31,396 - mmdet - INFO - Iter [178/17500] lr: 1.708e-04, eta: 10:31:47, time: 1.480, data_time: 0.077, memory: 49163, loss_cls_0: 0.8363, loss_box_0: 1.7324, loss_cns_0: 0.6184, loss_yns_0: 0.1514, loss_cls_1: 0.9039, loss_box_1: 1.6952, loss_cns_1: 0.6467, loss_yns_1: 0.1522, loss_cls_2: 0.9231, loss_box_2: 1.6693, loss_cns_2: 0.6514, loss_yns_2: 0.1504, loss_cls_3: 0.9369, loss_box_3: 1.6525, loss_cns_3: 0.6501, loss_yns_3: 0.1507, loss_cls_4: 0.9259, loss_box_4: 1.6535, loss_cns_4: 0.6524, loss_yns_4: 0.1500, loss_cls_5: 0.9263, loss_box_5: 1.6552, loss_cns_5: 0.6495, loss_yns_5: 0.1505, loss_cls_dn_0: 0.2115, loss_box_dn_0: 0.7618, loss_cls_dn_1: 0.1340, loss_box_dn_1: 0.7186, loss_cls_dn_2: 0.1372, loss_box_dn_2: 0.7065, loss_cls_dn_3: 0.1469, loss_box_dn_3: 0.7069, loss_cls_dn_4: 0.1495, loss_box_dn_4: 0.7146, loss_cls_dn_5: 0.1506, loss_box_dn_5: 0.7209, loss_dense_depth: 0.7737, loss: 26.3170, grad_norm: 43.2892 -2025-11-17 14:24:32,889 - mmdet - INFO - Iter [179/17500] lr: 1.712e-04, eta: 10:30:37, time: 1.492, data_time: 0.076, memory: 49163, loss_cls_0: 0.8422, loss_box_0: 1.7193, loss_cns_0: 0.6156, loss_yns_0: 0.1529, loss_cls_1: 0.9037, loss_box_1: 1.6994, loss_cns_1: 0.6478, loss_yns_1: 0.1539, loss_cls_2: 0.9249, loss_box_2: 1.6711, loss_cns_2: 0.6517, loss_yns_2: 0.1509, loss_cls_3: 0.9347, loss_box_3: 1.6501, loss_cns_3: 0.6527, loss_yns_3: 0.1516, loss_cls_4: 0.9401, loss_box_4: 1.6512, loss_cns_4: 0.6542, loss_yns_4: 0.1510, loss_cls_5: 0.9377, loss_box_5: 1.6596, loss_cns_5: 0.6534, loss_yns_5: 0.1515, loss_cls_dn_0: 0.2136, loss_box_dn_0: 0.7726, loss_cls_dn_1: 0.1358, loss_box_dn_1: 0.7192, loss_cls_dn_2: 0.1422, loss_box_dn_2: 0.7078, loss_cls_dn_3: 0.1535, loss_box_dn_3: 0.7075, loss_cls_dn_4: 0.1450, loss_box_dn_4: 0.7119, loss_cls_dn_5: 0.1523, loss_box_dn_5: 0.7182, loss_dense_depth: 0.7434, loss: 26.3442, grad_norm: 40.8283 -2025-11-17 14:24:34,388 - mmdet - INFO - Iter [180/17500] lr: 1.716e-04, eta: 10:29:29, time: 1.500, data_time: 0.080, memory: 49163, loss_cls_0: 0.8310, loss_box_0: 1.6919, loss_cns_0: 0.6200, loss_yns_0: 0.1510, loss_cls_1: 0.9117, loss_box_1: 1.6454, loss_cns_1: 0.6536, loss_yns_1: 0.1522, loss_cls_2: 0.9204, loss_box_2: 1.6246, loss_cns_2: 0.6572, loss_yns_2: 0.1511, loss_cls_3: 0.9195, loss_box_3: 1.6071, loss_cns_3: 0.6597, loss_yns_3: 0.1500, loss_cls_4: 0.9313, loss_box_4: 1.6121, loss_cns_4: 0.6575, loss_yns_4: 0.1520, loss_cls_5: 0.9237, loss_box_5: 1.6081, loss_cns_5: 0.6569, loss_yns_5: 0.1508, loss_cls_dn_0: 0.2138, loss_box_dn_0: 0.7691, loss_cls_dn_1: 0.1327, loss_box_dn_1: 0.7200, loss_cls_dn_2: 0.1372, loss_box_dn_2: 0.7150, loss_cls_dn_3: 0.1454, loss_box_dn_3: 0.7150, loss_cls_dn_4: 0.1409, loss_box_dn_4: 0.7196, loss_cls_dn_5: 0.1503, loss_box_dn_5: 0.7260, loss_dense_depth: 0.8026, loss: 26.1260, grad_norm: 46.1899 -2025-11-17 14:24:35,930 - mmdet - INFO - Iter [181/17500] lr: 1.720e-04, eta: 10:28:26, time: 1.541, data_time: 0.109, memory: 49163, loss_cls_0: 0.8179, loss_box_0: 1.7296, loss_cns_0: 0.6151, loss_yns_0: 0.1496, loss_cls_1: 0.9017, loss_box_1: 1.6878, loss_cns_1: 0.6483, loss_yns_1: 0.1504, loss_cls_2: 0.9225, loss_box_2: 1.6641, loss_cns_2: 0.6563, loss_yns_2: 0.1503, loss_cls_3: 0.9177, loss_box_3: 1.6548, loss_cns_3: 0.6567, loss_yns_3: 0.1501, loss_cls_4: 0.9200, loss_box_4: 1.6530, loss_cns_4: 0.6526, loss_yns_4: 0.1510, loss_cls_5: 0.9237, loss_box_5: 1.6563, loss_cns_5: 0.6524, loss_yns_5: 0.1506, loss_cls_dn_0: 0.2109, loss_box_dn_0: 0.7688, loss_cls_dn_1: 0.1318, loss_box_dn_1: 0.7228, loss_cls_dn_2: 0.1361, loss_box_dn_2: 0.7129, loss_cls_dn_3: 0.1387, loss_box_dn_3: 0.7177, loss_cls_dn_4: 0.1440, loss_box_dn_4: 0.7244, loss_cls_dn_5: 0.1501, loss_box_dn_5: 0.7344, loss_dense_depth: 0.7847, loss: 26.3102, grad_norm: 39.7334 -2025-11-17 14:24:37,493 - mmdet - INFO - Iter [182/17500] lr: 1.724e-04, eta: 10:27:25, time: 1.563, data_time: 0.102, memory: 49163, loss_cls_0: 0.8104, loss_box_0: 1.6951, loss_cns_0: 0.6198, loss_yns_0: 0.1472, loss_cls_1: 0.8869, loss_box_1: 1.6831, loss_cns_1: 0.6485, loss_yns_1: 0.1469, loss_cls_2: 0.9015, loss_box_2: 1.6419, loss_cns_2: 0.6571, loss_yns_2: 0.1475, loss_cls_3: 0.9021, loss_box_3: 1.6426, loss_cns_3: 0.6577, loss_yns_3: 0.1485, loss_cls_4: 0.9159, loss_box_4: 1.6425, loss_cns_4: 0.6549, loss_yns_4: 0.1459, loss_cls_5: 0.9053, loss_box_5: 1.6484, loss_cns_5: 0.6565, loss_yns_5: 0.1465, loss_cls_dn_0: 0.2070, loss_box_dn_0: 0.7642, loss_cls_dn_1: 0.1320, loss_box_dn_1: 0.7311, loss_cls_dn_2: 0.1341, loss_box_dn_2: 0.7126, loss_cls_dn_3: 0.1355, loss_box_dn_3: 0.7195, loss_cls_dn_4: 0.1426, loss_box_dn_4: 0.7267, loss_cls_dn_5: 0.1462, loss_box_dn_5: 0.7349, loss_dense_depth: 0.7320, loss: 26.0711, grad_norm: 37.9191 -2025-11-17 14:24:39,030 - mmdet - INFO - Iter [183/17500] lr: 1.728e-04, eta: 10:26:23, time: 1.537, data_time: 0.078, memory: 49163, loss_cls_0: 0.8146, loss_box_0: 1.6660, loss_cns_0: 0.6199, loss_yns_0: 0.1452, loss_cls_1: 0.8820, loss_box_1: 1.6259, loss_cns_1: 0.6504, loss_yns_1: 0.1473, loss_cls_2: 0.9001, loss_box_2: 1.5935, loss_cns_2: 0.6590, loss_yns_2: 0.1484, loss_cls_3: 0.9077, loss_box_3: 1.5960, loss_cns_3: 0.6619, loss_yns_3: 0.1494, loss_cls_4: 0.9124, loss_box_4: 1.5982, loss_cns_4: 0.6591, loss_yns_4: 0.1474, loss_cls_5: 0.9040, loss_box_5: 1.6093, loss_cns_5: 0.6593, loss_yns_5: 0.1479, loss_cls_dn_0: 0.2081, loss_box_dn_0: 0.7542, loss_cls_dn_1: 0.1280, loss_box_dn_1: 0.7208, loss_cls_dn_2: 0.1291, loss_box_dn_2: 0.7026, loss_cls_dn_3: 0.1334, loss_box_dn_3: 0.7073, loss_cls_dn_4: 0.1413, loss_box_dn_4: 0.7158, loss_cls_dn_5: 0.1408, loss_box_dn_5: 0.7238, loss_dense_depth: 0.7743, loss: 25.7845, grad_norm: 45.7353 -2025-11-17 14:24:40,558 - mmdet - INFO - Iter [184/17500] lr: 1.732e-04, eta: 10:25:20, time: 1.529, data_time: 0.081, memory: 49163, loss_cls_0: 0.8288, loss_box_0: 1.7003, loss_cns_0: 0.6194, loss_yns_0: 0.1460, loss_cls_1: 0.8948, loss_box_1: 1.6717, loss_cns_1: 0.6510, loss_yns_1: 0.1468, loss_cls_2: 0.9157, loss_box_2: 1.6542, loss_cns_2: 0.6566, loss_yns_2: 0.1464, loss_cls_3: 0.9408, loss_box_3: 1.6489, loss_cns_3: 0.6586, loss_yns_3: 0.1477, loss_cls_4: 0.9239, loss_box_4: 1.6461, loss_cns_4: 0.6597, loss_yns_4: 0.1476, loss_cls_5: 0.9263, loss_box_5: 1.6549, loss_cns_5: 0.6603, loss_yns_5: 0.1466, loss_cls_dn_0: 0.2155, loss_box_dn_0: 0.7596, loss_cls_dn_1: 0.1344, loss_box_dn_1: 0.7226, loss_cls_dn_2: 0.1344, loss_box_dn_2: 0.7064, loss_cls_dn_3: 0.1458, loss_box_dn_3: 0.7041, loss_cls_dn_4: 0.1503, loss_box_dn_4: 0.7085, loss_cls_dn_5: 0.1489, loss_box_dn_5: 0.7174, loss_dense_depth: 0.7639, loss: 26.2046, grad_norm: 40.4894 -2025-11-17 14:24:42,077 - mmdet - INFO - Iter [185/17500] lr: 1.736e-04, eta: 10:24:17, time: 1.517, data_time: 0.086, memory: 49163, loss_cls_0: 0.8334, loss_box_0: 1.6851, loss_cns_0: 0.6145, loss_yns_0: 0.1456, loss_cls_1: 0.8980, loss_box_1: 1.6779, loss_cns_1: 0.6482, loss_yns_1: 0.1477, loss_cls_2: 0.9071, loss_box_2: 1.6494, loss_cns_2: 0.6504, loss_yns_2: 0.1470, loss_cls_3: 0.9110, loss_box_3: 1.6476, loss_cns_3: 0.6529, loss_yns_3: 0.1466, loss_cls_4: 0.9155, loss_box_4: 1.6402, loss_cns_4: 0.6511, loss_yns_4: 0.1471, loss_cls_5: 0.9185, loss_box_5: 1.6471, loss_cns_5: 0.6540, loss_yns_5: 0.1479, loss_cls_dn_0: 0.2134, loss_box_dn_0: 0.7659, loss_cls_dn_1: 0.1358, loss_box_dn_1: 0.7208, loss_cls_dn_2: 0.1358, loss_box_dn_2: 0.7001, loss_cls_dn_3: 0.1429, loss_box_dn_3: 0.6974, loss_cls_dn_4: 0.1416, loss_box_dn_4: 0.7015, loss_cls_dn_5: 0.1440, loss_box_dn_5: 0.7065, loss_dense_depth: 0.7761, loss: 26.0656, grad_norm: 33.7751 -2025-11-17 14:24:43,608 - mmdet - INFO - Iter [186/17500] lr: 1.740e-04, eta: 10:23:16, time: 1.532, data_time: 0.087, memory: 49163, loss_cls_0: 0.8329, loss_box_0: 1.6789, loss_cns_0: 0.6224, loss_yns_0: 0.1488, loss_cls_1: 0.8951, loss_box_1: 1.6645, loss_cns_1: 0.6480, loss_yns_1: 0.1482, loss_cls_2: 0.9161, loss_box_2: 1.6058, loss_cns_2: 0.6561, loss_yns_2: 0.1482, loss_cls_3: 0.9161, loss_box_3: 1.5999, loss_cns_3: 0.6539, loss_yns_3: 0.1472, loss_cls_4: 0.9302, loss_box_4: 1.6024, loss_cns_4: 0.6525, loss_yns_4: 0.1472, loss_cls_5: 0.9244, loss_box_5: 1.6108, loss_cns_5: 0.6543, loss_yns_5: 0.1500, loss_cls_dn_0: 0.2074, loss_box_dn_0: 0.7635, loss_cls_dn_1: 0.1361, loss_box_dn_1: 0.7219, loss_cls_dn_2: 0.1372, loss_box_dn_2: 0.7006, loss_cls_dn_3: 0.1389, loss_box_dn_3: 0.7050, loss_cls_dn_4: 0.1448, loss_box_dn_4: 0.7146, loss_cls_dn_5: 0.1460, loss_box_dn_5: 0.7198, loss_dense_depth: 0.7504, loss: 25.9403, grad_norm: 38.8310 -2025-11-17 14:24:45,142 - mmdet - INFO - Iter [187/17500] lr: 1.744e-04, eta: 10:22:16, time: 1.532, data_time: 0.118, memory: 49163, loss_cls_0: 0.8447, loss_box_0: 1.6874, loss_cns_0: 0.6237, loss_yns_0: 0.1493, loss_cls_1: 0.9184, loss_box_1: 1.6495, loss_cns_1: 0.6547, loss_yns_1: 0.1491, loss_cls_2: 0.9330, loss_box_2: 1.6129, loss_cns_2: 0.6596, loss_yns_2: 0.1494, loss_cls_3: 0.9316, loss_box_3: 1.6117, loss_cns_3: 0.6583, loss_yns_3: 0.1505, loss_cls_4: 0.9411, loss_box_4: 1.6131, loss_cns_4: 0.6618, loss_yns_4: 0.1509, loss_cls_5: 0.9304, loss_box_5: 1.6235, loss_cns_5: 0.6611, loss_yns_5: 0.1516, loss_cls_dn_0: 0.2123, loss_box_dn_0: 0.7588, loss_cls_dn_1: 0.1322, loss_box_dn_1: 0.7283, loss_cls_dn_2: 0.1339, loss_box_dn_2: 0.7184, loss_cls_dn_3: 0.1378, loss_box_dn_3: 0.7239, loss_cls_dn_4: 0.1464, loss_box_dn_4: 0.7375, loss_cls_dn_5: 0.1449, loss_box_dn_5: 0.7462, loss_dense_depth: 0.7561, loss: 26.1939, grad_norm: 44.9514 -2025-11-17 14:24:46,638 - mmdet - INFO - Iter [188/17500] lr: 1.748e-04, eta: 10:21:13, time: 1.497, data_time: 0.076, memory: 49163, loss_cls_0: 0.8404, loss_box_0: 1.7143, loss_cns_0: 0.6230, loss_yns_0: 0.1506, loss_cls_1: 0.9192, loss_box_1: 1.6799, loss_cns_1: 0.6490, loss_yns_1: 0.1499, loss_cls_2: 0.9412, loss_box_2: 1.6539, loss_cns_2: 0.6516, loss_yns_2: 0.1499, loss_cls_3: 0.9335, loss_box_3: 1.6364, loss_cns_3: 0.6516, loss_yns_3: 0.1511, loss_cls_4: 0.9312, loss_box_4: 1.6407, loss_cns_4: 0.6533, loss_yns_4: 0.1505, loss_cls_5: 0.9235, loss_box_5: 1.6711, loss_cns_5: 0.6543, loss_yns_5: 0.1506, loss_cls_dn_0: 0.2099, loss_box_dn_0: 0.7603, loss_cls_dn_1: 0.1345, loss_box_dn_1: 0.7548, loss_cls_dn_2: 0.1376, loss_box_dn_2: 0.7465, loss_cls_dn_3: 0.1393, loss_box_dn_3: 0.7493, loss_cls_dn_4: 0.1432, loss_box_dn_4: 0.7671, loss_cls_dn_5: 0.1424, loss_box_dn_5: 0.7838, loss_dense_depth: 0.7629, loss: 26.5021, grad_norm: 43.0530 -2025-11-17 14:24:48,221 - mmdet - INFO - Iter [189/17500] lr: 1.752e-04, eta: 10:20:19, time: 1.583, data_time: 0.175, memory: 49163, loss_cls_0: 0.8444, loss_box_0: 1.6876, loss_cns_0: 0.6285, loss_yns_0: 0.1523, loss_cls_1: 0.9190, loss_box_1: 1.6429, loss_cns_1: 0.6546, loss_yns_1: 0.1528, loss_cls_2: 0.9428, loss_box_2: 1.6114, loss_cns_2: 0.6588, loss_yns_2: 0.1521, loss_cls_3: 0.9303, loss_box_3: 1.5967, loss_cns_3: 0.6574, loss_yns_3: 0.1518, loss_cls_4: 0.9408, loss_box_4: 1.6048, loss_cns_4: 0.6552, loss_yns_4: 0.1515, loss_cls_5: 0.9362, loss_box_5: 1.6180, loss_cns_5: 0.6599, loss_yns_5: 0.1537, loss_cls_dn_0: 0.2087, loss_box_dn_0: 0.7602, loss_cls_dn_1: 0.1333, loss_box_dn_1: 0.7550, loss_cls_dn_2: 0.1355, loss_box_dn_2: 0.7437, loss_cls_dn_3: 0.1384, loss_box_dn_3: 0.7446, loss_cls_dn_4: 0.1459, loss_box_dn_4: 0.7617, loss_cls_dn_5: 0.1494, loss_box_dn_5: 0.7708, loss_dense_depth: 0.7738, loss: 26.3246, grad_norm: 44.8282 -2025-11-17 14:24:49,715 - mmdet - INFO - Iter [190/17500] lr: 1.755e-04, eta: 10:19:17, time: 1.495, data_time: 0.073, memory: 49163, loss_cls_0: 0.8568, loss_box_0: 1.7226, loss_cns_0: 0.6215, loss_yns_0: 0.1527, loss_cls_1: 0.9252, loss_box_1: 1.6708, loss_cns_1: 0.6519, loss_yns_1: 0.1521, loss_cls_2: 0.9414, loss_box_2: 1.6521, loss_cns_2: 0.6564, loss_yns_2: 0.1537, loss_cls_3: 0.9362, loss_box_3: 1.6652, loss_cns_3: 0.6536, loss_yns_3: 0.1531, loss_cls_4: 0.9453, loss_box_4: 1.6689, loss_cns_4: 0.6513, loss_yns_4: 0.1547, loss_cls_5: 0.9445, loss_box_5: 1.6538, loss_cns_5: 0.6552, loss_yns_5: 0.1556, loss_cls_dn_0: 0.2139, loss_box_dn_0: 0.7671, loss_cls_dn_1: 0.1342, loss_box_dn_1: 0.7485, loss_cls_dn_2: 0.1364, loss_box_dn_2: 0.7381, loss_cls_dn_3: 0.1381, loss_box_dn_3: 0.7442, loss_cls_dn_4: 0.1485, loss_box_dn_4: 0.7587, loss_cls_dn_5: 0.1532, loss_box_dn_5: 0.7584, loss_dense_depth: 0.7779, loss: 26.6118, grad_norm: 49.9452 -2025-11-17 14:24:51,193 - mmdet - INFO - Iter [191/17500] lr: 1.759e-04, eta: 10:18:14, time: 1.478, data_time: 0.075, memory: 49163, loss_cls_0: 0.8425, loss_box_0: 1.7238, loss_cns_0: 0.6199, loss_yns_0: 0.1527, loss_cls_1: 0.9282, loss_box_1: 1.6499, loss_cns_1: 0.6542, loss_yns_1: 0.1527, loss_cls_2: 0.9493, loss_box_2: 1.6318, loss_cns_2: 0.6565, loss_yns_2: 0.1526, loss_cls_3: 0.9588, loss_box_3: 1.6468, loss_cns_3: 0.6561, loss_yns_3: 0.1526, loss_cls_4: 0.9475, loss_box_4: 1.6385, loss_cns_4: 0.6544, loss_yns_4: 0.1531, loss_cls_5: 0.9390, loss_box_5: 1.6264, loss_cns_5: 0.6597, loss_yns_5: 0.1540, loss_cls_dn_0: 0.2134, loss_box_dn_0: 0.7521, loss_cls_dn_1: 0.1354, loss_box_dn_1: 0.7274, loss_cls_dn_2: 0.1358, loss_box_dn_2: 0.7154, loss_cls_dn_3: 0.1363, loss_box_dn_3: 0.7219, loss_cls_dn_4: 0.1435, loss_box_dn_4: 0.7301, loss_cls_dn_5: 0.1478, loss_box_dn_5: 0.7347, loss_dense_depth: 0.7558, loss: 26.3509, grad_norm: 36.9788 -2025-11-17 14:24:52,685 - mmdet - INFO - Iter [192/17500] lr: 1.763e-04, eta: 10:17:13, time: 1.491, data_time: 0.077, memory: 49163, loss_cls_0: 0.8399, loss_box_0: 1.7202, loss_cns_0: 0.6208, loss_yns_0: 0.1531, loss_cls_1: 0.9281, loss_box_1: 1.6717, loss_cns_1: 0.6521, loss_yns_1: 0.1526, loss_cls_2: 0.9375, loss_box_2: 1.6541, loss_cns_2: 0.6521, loss_yns_2: 0.1529, loss_cls_3: 0.9409, loss_box_3: 1.6673, loss_cns_3: 0.6544, loss_yns_3: 0.1534, loss_cls_4: 0.9373, loss_box_4: 1.6538, loss_cns_4: 0.6530, loss_yns_4: 0.1535, loss_cls_5: 0.9424, loss_box_5: 1.6619, loss_cns_5: 0.6557, loss_yns_5: 0.1528, loss_cls_dn_0: 0.2133, loss_box_dn_0: 0.7486, loss_cls_dn_1: 0.1393, loss_box_dn_1: 0.7213, loss_cls_dn_2: 0.1390, loss_box_dn_2: 0.7093, loss_cls_dn_3: 0.1386, loss_box_dn_3: 0.7147, loss_cls_dn_4: 0.1444, loss_box_dn_4: 0.7155, loss_cls_dn_5: 0.1458, loss_box_dn_5: 0.7292, loss_dense_depth: 0.7341, loss: 26.3545, grad_norm: 43.3742 -2025-11-17 14:24:54,180 - mmdet - INFO - Iter [193/17500] lr: 1.767e-04, eta: 10:16:13, time: 1.496, data_time: 0.079, memory: 49163, loss_cls_0: 0.8286, loss_box_0: 1.6986, loss_cns_0: 0.6261, loss_yns_0: 0.1505, loss_cls_1: 0.9189, loss_box_1: 1.6586, loss_cns_1: 0.6539, loss_yns_1: 0.1503, loss_cls_2: 0.9310, loss_box_2: 1.6346, loss_cns_2: 0.6551, loss_yns_2: 0.1508, loss_cls_3: 0.9342, loss_box_3: 1.6431, loss_cns_3: 0.6548, loss_yns_3: 0.1498, loss_cls_4: 0.9478, loss_box_4: 1.6207, loss_cns_4: 0.6522, loss_yns_4: 0.1484, loss_cls_5: 0.9343, loss_box_5: 1.6496, loss_cns_5: 0.6584, loss_yns_5: 0.1491, loss_cls_dn_0: 0.2087, loss_box_dn_0: 0.7630, loss_cls_dn_1: 0.1357, loss_box_dn_1: 0.7103, loss_cls_dn_2: 0.1356, loss_box_dn_2: 0.6945, loss_cls_dn_3: 0.1363, loss_box_dn_3: 0.6989, loss_cls_dn_4: 0.1399, loss_box_dn_4: 0.7035, loss_cls_dn_5: 0.1432, loss_box_dn_5: 0.7223, loss_dense_depth: 0.7399, loss: 26.1315, grad_norm: 33.3202 -2025-11-17 14:24:55,680 - mmdet - INFO - Iter [194/17500] lr: 1.771e-04, eta: 10:15:14, time: 1.499, data_time: 0.076, memory: 49163, loss_cls_0: 0.8470, loss_box_0: 1.7116, loss_cns_0: 0.6215, loss_yns_0: 0.1534, loss_cls_1: 0.9247, loss_box_1: 1.7215, loss_cns_1: 0.6458, loss_yns_1: 0.1532, loss_cls_2: 0.9453, loss_box_2: 1.6934, loss_cns_2: 0.6518, loss_yns_2: 0.1546, loss_cls_3: 0.9502, loss_box_3: 1.6815, loss_cns_3: 0.6488, loss_yns_3: 0.1533, loss_cls_4: 0.9504, loss_box_4: 1.6870, loss_cns_4: 0.6508, loss_yns_4: 0.1544, loss_cls_5: 0.9443, loss_box_5: 1.7050, loss_cns_5: 0.6703, loss_yns_5: 0.1549, loss_cls_dn_0: 0.2100, loss_box_dn_0: 0.7563, loss_cls_dn_1: 0.1327, loss_box_dn_1: 0.7219, loss_cls_dn_2: 0.1325, loss_box_dn_2: 0.7096, loss_cls_dn_3: 0.1355, loss_box_dn_3: 0.7158, loss_cls_dn_4: 0.1387, loss_box_dn_4: 0.7248, loss_cls_dn_5: 0.1427, loss_box_dn_5: 0.7412, loss_dense_depth: 0.7711, loss: 26.6073, grad_norm: 41.5152 -2025-11-17 14:24:57,228 - mmdet - INFO - Iter [195/17500] lr: 1.775e-04, eta: 10:14:20, time: 1.549, data_time: 0.079, memory: 49163, loss_cls_0: 0.8126, loss_box_0: 1.6860, loss_cns_0: 0.6323, loss_yns_0: 0.1509, loss_cls_1: 0.8869, loss_box_1: 1.6580, loss_cns_1: 0.6491, loss_yns_1: 0.1506, loss_cls_2: 0.9006, loss_box_2: 1.6261, loss_cns_2: 0.6557, loss_yns_2: 0.1515, loss_cls_3: 0.9049, loss_box_3: 1.6309, loss_cns_3: 0.6541, loss_yns_3: 0.1511, loss_cls_4: 0.9082, loss_box_4: 1.6228, loss_cns_4: 0.6571, loss_yns_4: 0.1538, loss_cls_5: 0.9030, loss_box_5: 1.6334, loss_cns_5: 0.6699, loss_yns_5: 0.1527, loss_cls_dn_0: 0.1974, loss_box_dn_0: 0.7533, loss_cls_dn_1: 0.1301, loss_box_dn_1: 0.7344, loss_cls_dn_2: 0.1304, loss_box_dn_2: 0.7258, loss_cls_dn_3: 0.1317, loss_box_dn_3: 0.7444, loss_cls_dn_4: 0.1402, loss_box_dn_4: 0.7472, loss_cls_dn_5: 0.1417, loss_box_dn_5: 0.7561, loss_dense_depth: 0.7307, loss: 26.0658, grad_norm: 36.3254 -2025-11-17 14:24:58,709 - mmdet - INFO - Iter [196/17500] lr: 1.779e-04, eta: 10:13:21, time: 1.480, data_time: 0.078, memory: 49163, loss_cls_0: 0.7941, loss_box_0: 1.6780, loss_cns_0: 0.6262, loss_yns_0: 0.1513, loss_cls_1: 0.8719, loss_box_1: 1.6525, loss_cns_1: 0.6498, loss_yns_1: 0.1509, loss_cls_2: 0.8906, loss_box_2: 1.6087, loss_cns_2: 0.6552, loss_yns_2: 0.1516, loss_cls_3: 0.8939, loss_box_3: 1.6374, loss_cns_3: 0.6568, loss_yns_3: 0.1521, loss_cls_4: 0.9004, loss_box_4: 1.6183, loss_cns_4: 0.6608, loss_yns_4: 0.1530, loss_cls_5: 0.8938, loss_box_5: 1.6294, loss_cns_5: 0.6599, loss_yns_5: 0.1520, loss_cls_dn_0: 0.1986, loss_box_dn_0: 0.7611, loss_cls_dn_1: 0.1319, loss_box_dn_1: 0.7621, loss_cls_dn_2: 0.1318, loss_box_dn_2: 0.7516, loss_cls_dn_3: 0.1345, loss_box_dn_3: 0.7723, loss_cls_dn_4: 0.1401, loss_box_dn_4: 0.7729, loss_cls_dn_5: 0.1414, loss_box_dn_5: 0.7780, loss_dense_depth: 0.7655, loss: 26.1304, grad_norm: 41.3495 -2025-11-17 14:25:00,249 - mmdet - INFO - Iter [197/17500] lr: 1.783e-04, eta: 10:12:27, time: 1.540, data_time: 0.079, memory: 49163, loss_cls_0: 0.8265, loss_box_0: 1.7475, loss_cns_0: 0.6239, loss_yns_0: 0.1512, loss_cls_1: 0.8888, loss_box_1: 1.7046, loss_cns_1: 0.6554, loss_yns_1: 0.1518, loss_cls_2: 0.9043, loss_box_2: 1.6714, loss_cns_2: 0.6580, loss_yns_2: 0.1526, loss_cls_3: 0.9106, loss_box_3: 1.6779, loss_cns_3: 0.6590, loss_yns_3: 0.1537, loss_cls_4: 0.9249, loss_box_4: 1.6675, loss_cns_4: 0.6598, loss_yns_4: 0.1538, loss_cls_5: 0.9019, loss_box_5: 1.6889, loss_cns_5: 0.6569, loss_yns_5: 0.1522, loss_cls_dn_0: 0.2050, loss_box_dn_0: 0.7541, loss_cls_dn_1: 0.1306, loss_box_dn_1: 0.7547, loss_cls_dn_2: 0.1326, loss_box_dn_2: 0.7378, loss_cls_dn_3: 0.1357, loss_box_dn_3: 0.7538, loss_cls_dn_4: 0.1388, loss_box_dn_4: 0.7530, loss_cls_dn_5: 0.1437, loss_box_dn_5: 0.7628, loss_dense_depth: 0.7676, loss: 26.5132, grad_norm: 32.3958 -2025-11-17 14:25:01,740 - mmdet - INFO - Iter [198/17500] lr: 1.787e-04, eta: 10:11:30, time: 1.492, data_time: 0.075, memory: 49163, loss_cls_0: 0.8437, loss_box_0: 1.7468, loss_cns_0: 0.6190, loss_yns_0: 0.1537, loss_cls_1: 0.9048, loss_box_1: 1.6923, loss_cns_1: 0.6489, loss_yns_1: 0.1546, loss_cls_2: 0.9209, loss_box_2: 1.6487, loss_cns_2: 0.6562, loss_yns_2: 0.1546, loss_cls_3: 0.9193, loss_box_3: 1.6468, loss_cns_3: 0.6548, loss_yns_3: 0.1550, loss_cls_4: 0.9260, loss_box_4: 1.6438, loss_cns_4: 0.6550, loss_yns_4: 0.1552, loss_cls_5: 0.9157, loss_box_5: 1.6495, loss_cns_5: 0.6571, loss_yns_5: 0.1544, loss_cls_dn_0: 0.2050, loss_box_dn_0: 0.7632, loss_cls_dn_1: 0.1323, loss_box_dn_1: 0.7670, loss_cls_dn_2: 0.1344, loss_box_dn_2: 0.7465, loss_cls_dn_3: 0.1378, loss_box_dn_3: 0.7503, loss_cls_dn_4: 0.1406, loss_box_dn_4: 0.7498, loss_cls_dn_5: 0.1467, loss_box_dn_5: 0.7555, loss_dense_depth: 0.7692, loss: 26.4750, grad_norm: 34.2294 -2025-11-17 14:25:03,252 - mmdet - INFO - Iter [199/17500] lr: 1.791e-04, eta: 10:10:35, time: 1.512, data_time: 0.084, memory: 49163, loss_cls_0: 0.8348, loss_box_0: 1.7472, loss_cns_0: 0.6176, loss_yns_0: 0.1546, loss_cls_1: 0.8974, loss_box_1: 1.7022, loss_cns_1: 0.6503, loss_yns_1: 0.1542, loss_cls_2: 0.9118, loss_box_2: 1.6606, loss_cns_2: 0.6546, loss_yns_2: 0.1542, loss_cls_3: 0.9058, loss_box_3: 1.6532, loss_cns_3: 0.6560, loss_yns_3: 0.1541, loss_cls_4: 0.9042, loss_box_4: 1.6526, loss_cns_4: 0.6566, loss_yns_4: 0.1544, loss_cls_5: 0.9108, loss_box_5: 1.6618, loss_cns_5: 0.6671, loss_yns_5: 0.1549, loss_cls_dn_0: 0.2009, loss_box_dn_0: 0.7519, loss_cls_dn_1: 0.1313, loss_box_dn_1: 0.7548, loss_cls_dn_2: 0.1309, loss_box_dn_2: 0.7295, loss_cls_dn_3: 0.1320, loss_box_dn_3: 0.7299, loss_cls_dn_4: 0.1371, loss_box_dn_4: 0.7333, loss_cls_dn_5: 0.1409, loss_box_dn_5: 0.7411, loss_dense_depth: 0.7984, loss: 26.3831, grad_norm: 27.5325 -2025-11-17 14:25:04,751 - mmdet - INFO - Iter [200/17500] lr: 1.795e-04, eta: 10:09:39, time: 1.500, data_time: 0.081, memory: 49163, loss_cls_0: 0.8254, loss_box_0: 1.7553, loss_cns_0: 0.6136, loss_yns_0: 0.1546, loss_cls_1: 0.9001, loss_box_1: 1.7097, loss_cns_1: 0.6529, loss_yns_1: 0.1541, loss_cls_2: 0.9094, loss_box_2: 1.6708, loss_cns_2: 0.6568, loss_yns_2: 0.1525, loss_cls_3: 0.9035, loss_box_3: 1.6635, loss_cns_3: 0.6585, loss_yns_3: 0.1524, loss_cls_4: 0.9130, loss_box_4: 1.6502, loss_cns_4: 0.6569, loss_yns_4: 0.1525, loss_cls_5: 0.8977, loss_box_5: 1.6805, loss_cns_5: 0.6648, loss_yns_5: 0.1534, loss_cls_dn_0: 0.2024, loss_box_dn_0: 0.7545, loss_cls_dn_1: 0.1320, loss_box_dn_1: 0.7469, loss_cls_dn_2: 0.1307, loss_box_dn_2: 0.7286, loss_cls_dn_3: 0.1340, loss_box_dn_3: 0.7289, loss_cls_dn_4: 0.1385, loss_box_dn_4: 0.7322, loss_cls_dn_5: 0.1412, loss_box_dn_5: 0.7436, loss_dense_depth: 0.8176, loss: 26.4332, grad_norm: 30.4820 -2025-11-17 14:25:06,309 - mmdet - INFO - Iter [201/17500] lr: 1.799e-04, eta: 10:08:49, time: 1.558, data_time: 0.080, memory: 49163, loss_cls_0: 0.8148, loss_box_0: 1.7054, loss_cns_0: 0.6172, loss_yns_0: 0.1544, loss_cls_1: 0.9037, loss_box_1: 1.6830, loss_cns_1: 0.6491, loss_yns_1: 0.1536, loss_cls_2: 0.9075, loss_box_2: 1.6711, loss_cns_2: 0.6536, loss_yns_2: 0.1533, loss_cls_3: 0.9026, loss_box_3: 1.6720, loss_cns_3: 0.6526, loss_yns_3: 0.1519, loss_cls_4: 0.9180, loss_box_4: 1.6586, loss_cns_4: 0.6497, loss_yns_4: 0.1518, loss_cls_5: 0.8963, loss_box_5: 1.6902, loss_cns_5: 0.6518, loss_yns_5: 0.1531, loss_cls_dn_0: 0.1964, loss_box_dn_0: 0.7533, loss_cls_dn_1: 0.1274, loss_box_dn_1: 0.7486, loss_cls_dn_2: 0.1264, loss_box_dn_2: 0.7405, loss_cls_dn_3: 0.1296, loss_box_dn_3: 0.7441, loss_cls_dn_4: 0.1343, loss_box_dn_4: 0.7487, loss_cls_dn_5: 0.1358, loss_box_dn_5: 0.7575, loss_dense_depth: 0.7628, loss: 26.3206, grad_norm: 37.2929 -2025-11-17 14:25:07,895 - mmdet - INFO - Iter [202/17500] lr: 1.803e-04, eta: 10:08:02, time: 1.585, data_time: 0.080, memory: 49163, loss_cls_0: 0.8097, loss_box_0: 1.6774, loss_cns_0: 0.6122, loss_yns_0: 0.1533, loss_cls_1: 0.9051, loss_box_1: 1.6644, loss_cns_1: 0.6440, loss_yns_1: 0.1536, loss_cls_2: 0.9081, loss_box_2: 1.6477, loss_cns_2: 0.6513, loss_yns_2: 0.1532, loss_cls_3: 0.9064, loss_box_3: 1.6477, loss_cns_3: 0.6488, loss_yns_3: 0.1525, loss_cls_4: 0.9109, loss_box_4: 1.6449, loss_cns_4: 0.6486, loss_yns_4: 0.1526, loss_cls_5: 0.9051, loss_box_5: 1.6691, loss_cns_5: 0.6498, loss_yns_5: 0.1543, loss_cls_dn_0: 0.1951, loss_box_dn_0: 0.7575, loss_cls_dn_1: 0.1280, loss_box_dn_1: 0.7535, loss_cls_dn_2: 0.1269, loss_box_dn_2: 0.7453, loss_cls_dn_3: 0.1288, loss_box_dn_3: 0.7554, loss_cls_dn_4: 0.1326, loss_box_dn_4: 0.7715, loss_cls_dn_5: 0.1371, loss_box_dn_5: 0.7837, loss_dense_depth: 0.7853, loss: 26.2717, grad_norm: 38.5668 -2025-11-17 14:25:09,408 - mmdet - INFO - Iter [203/17500] lr: 1.807e-04, eta: 10:07:09, time: 1.514, data_time: 0.076, memory: 49163, loss_cls_0: 0.8127, loss_box_0: 1.7012, loss_cns_0: 0.6115, loss_yns_0: 0.1500, loss_cls_1: 0.8999, loss_box_1: 1.7147, loss_cns_1: 0.6423, loss_yns_1: 0.1524, loss_cls_2: 0.9057, loss_box_2: 1.7026, loss_cns_2: 0.6455, loss_yns_2: 0.1515, loss_cls_3: 0.9026, loss_box_3: 1.7005, loss_cns_3: 0.6475, loss_yns_3: 0.1531, loss_cls_4: 0.9043, loss_box_4: 1.7084, loss_cns_4: 0.6467, loss_yns_4: 0.1544, loss_cls_5: 0.8963, loss_box_5: 1.7253, loss_cns_5: 0.6479, loss_yns_5: 0.1540, loss_cls_dn_0: 0.1976, loss_box_dn_0: 0.7561, loss_cls_dn_1: 0.1272, loss_box_dn_1: 0.7642, loss_cls_dn_2: 0.1269, loss_box_dn_2: 0.7560, loss_cls_dn_3: 0.1271, loss_box_dn_3: 0.7626, loss_cls_dn_4: 0.1304, loss_box_dn_4: 0.7776, loss_cls_dn_5: 0.1351, loss_box_dn_5: 0.7948, loss_dense_depth: 0.8073, loss: 26.5939, grad_norm: 34.7352 -2025-11-17 14:25:10,905 - mmdet - INFO - Iter [204/17500] lr: 1.811e-04, eta: 10:06:15, time: 1.496, data_time: 0.074, memory: 49163, loss_cls_0: 0.7998, loss_box_0: 1.7297, loss_cns_0: 0.6209, loss_yns_0: 0.1524, loss_cls_1: 0.8913, loss_box_1: 1.6880, loss_cns_1: 0.6521, loss_yns_1: 0.1520, loss_cls_2: 0.8950, loss_box_2: 1.6883, loss_cns_2: 0.6524, loss_yns_2: 0.1524, loss_cls_3: 0.8928, loss_box_3: 1.6745, loss_cns_3: 0.6540, loss_yns_3: 0.1531, loss_cls_4: 0.8956, loss_box_4: 1.6756, loss_cns_4: 0.6516, loss_yns_4: 0.1532, loss_cls_5: 0.8857, loss_box_5: 1.6926, loss_cns_5: 0.6550, loss_yns_5: 0.1521, loss_cls_dn_0: 0.1989, loss_box_dn_0: 0.7617, loss_cls_dn_1: 0.1200, loss_box_dn_1: 0.7213, loss_cls_dn_2: 0.1219, loss_box_dn_2: 0.7218, loss_cls_dn_3: 0.1245, loss_box_dn_3: 0.7278, loss_cls_dn_4: 0.1274, loss_box_dn_4: 0.7437, loss_cls_dn_5: 0.1352, loss_box_dn_5: 0.7661, loss_dense_depth: 0.7642, loss: 26.2442, grad_norm: 38.6440 -2025-11-17 14:25:12,424 - mmdet - INFO - Iter [205/17500] lr: 1.815e-04, eta: 10:05:24, time: 1.518, data_time: 0.080, memory: 49163, loss_cls_0: 0.8082, loss_box_0: 1.7469, loss_cns_0: 0.6165, loss_yns_0: 0.1536, loss_cls_1: 0.8991, loss_box_1: 1.6922, loss_cns_1: 0.6540, loss_yns_1: 0.1516, loss_cls_2: 0.9018, loss_box_2: 1.6778, loss_cns_2: 0.6553, loss_yns_2: 0.1501, loss_cls_3: 0.8976, loss_box_3: 1.6681, loss_cns_3: 0.6564, loss_yns_3: 0.1509, loss_cls_4: 0.9178, loss_box_4: 1.6747, loss_cns_4: 0.6539, loss_yns_4: 0.1524, loss_cls_5: 0.8956, loss_box_5: 1.6812, loss_cns_5: 0.6571, loss_yns_5: 0.1520, loss_cls_dn_0: 0.2038, loss_box_dn_0: 0.7655, loss_cls_dn_1: 0.1285, loss_box_dn_1: 0.7377, loss_cls_dn_2: 0.1308, loss_box_dn_2: 0.7311, loss_cls_dn_3: 0.1355, loss_box_dn_3: 0.7379, loss_cls_dn_4: 0.1362, loss_box_dn_4: 0.7534, loss_cls_dn_5: 0.1442, loss_box_dn_5: 0.7686, loss_dense_depth: 0.8076, loss: 26.4457, grad_norm: 41.5086 -2025-11-17 14:25:13,938 - mmdet - INFO - Iter [206/17500] lr: 1.819e-04, eta: 10:04:33, time: 1.516, data_time: 0.080, memory: 49163, loss_cls_0: 0.8160, loss_box_0: 1.7281, loss_cns_0: 0.6196, loss_yns_0: 0.1529, loss_cls_1: 0.8979, loss_box_1: 1.7014, loss_cns_1: 0.6533, loss_yns_1: 0.1531, loss_cls_2: 0.9067, loss_box_2: 1.6779, loss_cns_2: 0.6556, loss_yns_2: 0.1509, loss_cls_3: 0.9062, loss_box_3: 1.6668, loss_cns_3: 0.6576, loss_yns_3: 0.1525, loss_cls_4: 0.9057, loss_box_4: 1.6717, loss_cns_4: 0.6546, loss_yns_4: 0.1531, loss_cls_5: 0.9026, loss_box_5: 1.6639, loss_cns_5: 0.6546, loss_yns_5: 0.1538, loss_cls_dn_0: 0.2039, loss_box_dn_0: 0.7610, loss_cls_dn_1: 0.1285, loss_box_dn_1: 0.7384, loss_cls_dn_2: 0.1303, loss_box_dn_2: 0.7290, loss_cls_dn_3: 0.1347, loss_box_dn_3: 0.7289, loss_cls_dn_4: 0.1336, loss_box_dn_4: 0.7382, loss_cls_dn_5: 0.1377, loss_box_dn_5: 0.7454, loss_dense_depth: 0.7566, loss: 26.3228, grad_norm: 35.5129 -2025-11-17 14:25:15,443 - mmdet - INFO - Iter [207/17500] lr: 1.823e-04, eta: 10:03:41, time: 1.505, data_time: 0.088, memory: 49163, loss_cls_0: 0.8053, loss_box_0: 1.7258, loss_cns_0: 0.6189, loss_yns_0: 0.1524, loss_cls_1: 0.8931, loss_box_1: 1.7053, loss_cns_1: 0.6524, loss_yns_1: 0.1517, loss_cls_2: 0.9107, loss_box_2: 1.6774, loss_cns_2: 0.6547, loss_yns_2: 0.1525, loss_cls_3: 0.9016, loss_box_3: 1.6597, loss_cns_3: 0.6543, loss_yns_3: 0.1532, loss_cls_4: 0.9033, loss_box_4: 1.6695, loss_cns_4: 0.6571, loss_yns_4: 0.1538, loss_cls_5: 0.9016, loss_box_5: 1.6615, loss_cns_5: 0.6522, loss_yns_5: 0.1527, loss_cls_dn_0: 0.2012, loss_box_dn_0: 0.7576, loss_cls_dn_1: 0.1295, loss_box_dn_1: 0.7390, loss_cls_dn_2: 0.1315, loss_box_dn_2: 0.7207, loss_cls_dn_3: 0.1325, loss_box_dn_3: 0.7168, loss_cls_dn_4: 0.1352, loss_box_dn_4: 0.7236, loss_cls_dn_5: 0.1390, loss_box_dn_5: 0.7267, loss_dense_depth: 0.7703, loss: 26.2444, grad_norm: 35.9082 -2025-11-17 14:25:16,941 - mmdet - INFO - Iter [208/17500] lr: 1.827e-04, eta: 10:02:49, time: 1.498, data_time: 0.077, memory: 49163, loss_cls_0: 0.8166, loss_box_0: 1.7447, loss_cns_0: 0.6213, loss_yns_0: 0.1561, loss_cls_1: 0.8973, loss_box_1: 1.6957, loss_cns_1: 0.6508, loss_yns_1: 0.1572, loss_cls_2: 0.9063, loss_box_2: 1.6760, loss_cns_2: 0.6516, loss_yns_2: 0.1560, loss_cls_3: 0.9073, loss_box_3: 1.6538, loss_cns_3: 0.6514, loss_yns_3: 0.1557, loss_cls_4: 0.9042, loss_box_4: 1.6634, loss_cns_4: 0.6540, loss_yns_4: 0.1561, loss_cls_5: 0.9018, loss_box_5: 1.6682, loss_cns_5: 0.6517, loss_yns_5: 0.1561, loss_cls_dn_0: 0.2038, loss_box_dn_0: 0.7619, loss_cls_dn_1: 0.1266, loss_box_dn_1: 0.7102, loss_cls_dn_2: 0.1278, loss_box_dn_2: 0.6956, loss_cls_dn_3: 0.1284, loss_box_dn_3: 0.6949, loss_cls_dn_4: 0.1329, loss_box_dn_4: 0.7018, loss_cls_dn_5: 0.1425, loss_box_dn_5: 0.7103, loss_dense_depth: 0.7502, loss: 26.1400, grad_norm: 29.6161 -2025-11-17 14:25:18,478 - mmdet - INFO - Iter [209/17500] lr: 1.831e-04, eta: 10:02:01, time: 1.537, data_time: 0.124, memory: 49163, loss_cls_0: 0.8336, loss_box_0: 1.7632, loss_cns_0: 0.6147, loss_yns_0: 0.1528, loss_cls_1: 0.9143, loss_box_1: 1.7007, loss_cns_1: 0.6485, loss_yns_1: 0.1521, loss_cls_2: 0.9276, loss_box_2: 1.6977, loss_cns_2: 0.6505, loss_yns_2: 0.1528, loss_cls_3: 0.9216, loss_box_3: 1.6728, loss_cns_3: 0.6488, loss_yns_3: 0.1518, loss_cls_4: 0.9322, loss_box_4: 1.6754, loss_cns_4: 0.6480, loss_yns_4: 0.1527, loss_cls_5: 0.9198, loss_box_5: 1.6905, loss_cns_5: 0.6505, loss_yns_5: 0.1530, loss_cls_dn_0: 0.2033, loss_box_dn_0: 0.7570, loss_cls_dn_1: 0.1254, loss_box_dn_1: 0.7090, loss_cls_dn_2: 0.1267, loss_box_dn_2: 0.6992, loss_cls_dn_3: 0.1275, loss_box_dn_3: 0.7005, loss_cls_dn_4: 0.1313, loss_box_dn_4: 0.7066, loss_cls_dn_5: 0.1365, loss_box_dn_5: 0.7226, loss_dense_depth: 0.7828, loss: 26.3539, grad_norm: 41.2971 -2025-11-17 14:25:19,973 - mmdet - INFO - Iter [210/17500] lr: 1.835e-04, eta: 10:01:10, time: 1.495, data_time: 0.079, memory: 49163, loss_cls_0: 0.7987, loss_box_0: 1.7057, loss_cns_0: 0.6225, loss_yns_0: 0.1521, loss_cls_1: 0.8784, loss_box_1: 1.6767, loss_cns_1: 0.6474, loss_yns_1: 0.1511, loss_cls_2: 0.8950, loss_box_2: 1.6503, loss_cns_2: 0.6530, loss_yns_2: 0.1511, loss_cls_3: 0.8790, loss_box_3: 1.6314, loss_cns_3: 0.6532, loss_yns_3: 0.1503, loss_cls_4: 0.9022, loss_box_4: 1.6257, loss_cns_4: 0.6490, loss_yns_4: 0.1507, loss_cls_5: 0.8847, loss_box_5: 1.6371, loss_cns_5: 0.6535, loss_yns_5: 0.1531, loss_cls_dn_0: 0.1898, loss_box_dn_0: 0.7505, loss_cls_dn_1: 0.1212, loss_box_dn_1: 0.7153, loss_cls_dn_2: 0.1229, loss_box_dn_2: 0.7038, loss_cls_dn_3: 0.1248, loss_box_dn_3: 0.7091, loss_cls_dn_4: 0.1283, loss_box_dn_4: 0.7139, loss_cls_dn_5: 0.1308, loss_box_dn_5: 0.7283, loss_dense_depth: 0.7226, loss: 25.8130, grad_norm: 32.9599 -2025-11-17 14:25:21,467 - mmdet - INFO - Iter [211/17500] lr: 1.839e-04, eta: 10:00:20, time: 1.494, data_time: 0.082, memory: 49163, loss_cls_0: 0.8008, loss_box_0: 1.7158, loss_cns_0: 0.6209, loss_yns_0: 0.1515, loss_cls_1: 0.8655, loss_box_1: 1.6413, loss_cns_1: 0.6489, loss_yns_1: 0.1512, loss_cls_2: 0.8788, loss_box_2: 1.6121, loss_cns_2: 0.6540, loss_yns_2: 0.1520, loss_cls_3: 0.8832, loss_box_3: 1.6017, loss_cns_3: 0.6586, loss_yns_3: 0.1511, loss_cls_4: 0.8899, loss_box_4: 1.6038, loss_cns_4: 0.6534, loss_yns_4: 0.1510, loss_cls_5: 0.8915, loss_box_5: 1.6023, loss_cns_5: 0.6546, loss_yns_5: 0.1539, loss_cls_dn_0: 0.1879, loss_box_dn_0: 0.7552, loss_cls_dn_1: 0.1174, loss_box_dn_1: 0.7141, loss_cls_dn_2: 0.1156, loss_box_dn_2: 0.7008, loss_cls_dn_3: 0.1200, loss_box_dn_3: 0.7124, loss_cls_dn_4: 0.1243, loss_box_dn_4: 0.7197, loss_cls_dn_5: 0.1296, loss_box_dn_5: 0.7329, loss_dense_depth: 0.7708, loss: 25.6886, grad_norm: 49.8483 -2025-11-17 14:25:22,959 - mmdet - INFO - Iter [212/17500] lr: 1.843e-04, eta: 9:59:30, time: 1.492, data_time: 0.079, memory: 49163, loss_cls_0: 0.8063, loss_box_0: 1.7192, loss_cns_0: 0.6250, loss_yns_0: 0.1496, loss_cls_1: 0.8671, loss_box_1: 1.6916, loss_cns_1: 0.6441, loss_yns_1: 0.1496, loss_cls_2: 0.8824, loss_box_2: 1.6543, loss_cns_2: 0.6487, loss_yns_2: 0.1498, loss_cls_3: 0.8889, loss_box_3: 1.6428, loss_cns_3: 0.6518, loss_yns_3: 0.1499, loss_cls_4: 0.8882, loss_box_4: 1.6438, loss_cns_4: 0.6504, loss_yns_4: 0.1491, loss_cls_5: 0.8911, loss_box_5: 1.6589, loss_cns_5: 0.6502, loss_yns_5: 0.1496, loss_cls_dn_0: 0.1946, loss_box_dn_0: 0.7530, loss_cls_dn_1: 0.1238, loss_box_dn_1: 0.7300, loss_cls_dn_2: 0.1216, loss_box_dn_2: 0.7117, loss_cls_dn_3: 0.1241, loss_box_dn_3: 0.7209, loss_cls_dn_4: 0.1275, loss_box_dn_4: 0.7313, loss_cls_dn_5: 0.1305, loss_box_dn_5: 0.7473, loss_dense_depth: 0.7531, loss: 25.9717, grad_norm: 42.3349 -2025-11-17 14:25:24,447 - mmdet - INFO - Iter [213/17500] lr: 1.847e-04, eta: 9:58:39, time: 1.487, data_time: 0.080, memory: 49163, loss_cls_0: 0.8063, loss_box_0: 1.7243, loss_cns_0: 0.6262, loss_yns_0: 0.1529, loss_cls_1: 0.8787, loss_box_1: 1.6359, loss_cns_1: 0.6483, loss_yns_1: 0.1512, loss_cls_2: 0.9090, loss_box_2: 1.6158, loss_cns_2: 0.6498, loss_yns_2: 0.1497, loss_cls_3: 0.9024, loss_box_3: 1.6059, loss_cns_3: 0.6491, loss_yns_3: 0.1499, loss_cls_4: 0.8902, loss_box_4: 1.6249, loss_cns_4: 0.6568, loss_yns_4: 0.1514, loss_cls_5: 0.8920, loss_box_5: 1.6390, loss_cns_5: 0.6535, loss_yns_5: 0.1501, loss_cls_dn_0: 0.1987, loss_box_dn_0: 0.7509, loss_cls_dn_1: 0.1262, loss_box_dn_1: 0.7215, loss_cls_dn_2: 0.1257, loss_box_dn_2: 0.7145, loss_cls_dn_3: 0.1302, loss_box_dn_3: 0.7203, loss_cls_dn_4: 0.1311, loss_box_dn_4: 0.7329, loss_cls_dn_5: 0.1365, loss_box_dn_5: 0.7506, loss_dense_depth: 0.7852, loss: 25.9377, grad_norm: 40.9952 -2025-11-17 14:25:25,935 - mmdet - INFO - Iter [214/17500] lr: 1.851e-04, eta: 9:57:50, time: 1.488, data_time: 0.082, memory: 49163, loss_cls_0: 0.8145, loss_box_0: 1.7229, loss_cns_0: 0.6233, loss_yns_0: 0.1537, loss_cls_1: 0.8742, loss_box_1: 1.6864, loss_cns_1: 0.6471, loss_yns_1: 0.1516, loss_cls_2: 0.8930, loss_box_2: 1.6298, loss_cns_2: 0.6524, loss_yns_2: 0.1526, loss_cls_3: 0.8918, loss_box_3: 1.6158, loss_cns_3: 0.6509, loss_yns_3: 0.1520, loss_cls_4: 0.9144, loss_box_4: 1.6179, loss_cns_4: 0.6550, loss_yns_4: 0.1540, loss_cls_5: 0.9059, loss_box_5: 1.6203, loss_cns_5: 0.6537, loss_yns_5: 0.1523, loss_cls_dn_0: 0.1931, loss_box_dn_0: 0.7617, loss_cls_dn_1: 0.1223, loss_box_dn_1: 0.7366, loss_cls_dn_2: 0.1215, loss_box_dn_2: 0.7299, loss_cls_dn_3: 0.1273, loss_box_dn_3: 0.7308, loss_cls_dn_4: 0.1296, loss_box_dn_4: 0.7385, loss_cls_dn_5: 0.1354, loss_box_dn_5: 0.7473, loss_dense_depth: 0.7569, loss: 26.0164, grad_norm: 45.2492 -2025-11-17 14:25:27,483 - mmdet - INFO - Iter [215/17500] lr: 1.855e-04, eta: 9:57:05, time: 1.548, data_time: 0.079, memory: 49163, loss_cls_0: 0.8284, loss_box_0: 1.7229, loss_cns_0: 0.6229, loss_yns_0: 0.1524, loss_cls_1: 0.8888, loss_box_1: 1.6500, loss_cns_1: 0.6478, loss_yns_1: 0.1528, loss_cls_2: 0.9019, loss_box_2: 1.6230, loss_cns_2: 0.6497, loss_yns_2: 0.1542, loss_cls_3: 0.9069, loss_box_3: 1.6204, loss_cns_3: 0.6523, loss_yns_3: 0.1521, loss_cls_4: 0.9083, loss_box_4: 1.6008, loss_cns_4: 0.6536, loss_yns_4: 0.1526, loss_cls_5: 0.9016, loss_box_5: 1.6116, loss_cns_5: 0.6530, loss_yns_5: 0.1527, loss_cls_dn_0: 0.1999, loss_box_dn_0: 0.7500, loss_cls_dn_1: 0.1250, loss_box_dn_1: 0.7238, loss_cls_dn_2: 0.1255, loss_box_dn_2: 0.7086, loss_cls_dn_3: 0.1285, loss_box_dn_3: 0.7110, loss_cls_dn_4: 0.1314, loss_box_dn_4: 0.7124, loss_cls_dn_5: 0.1396, loss_box_dn_5: 0.7170, loss_dense_depth: 0.8085, loss: 25.9417, grad_norm: 34.9063 -2025-11-17 14:25:28,976 - mmdet - INFO - Iter [216/17500] lr: 1.859e-04, eta: 9:56:17, time: 1.493, data_time: 0.079, memory: 49163, loss_cls_0: 0.8349, loss_box_0: 1.7373, loss_cns_0: 0.6214, loss_yns_0: 0.1543, loss_cls_1: 0.9039, loss_box_1: 1.6411, loss_cns_1: 0.6520, loss_yns_1: 0.1537, loss_cls_2: 0.9200, loss_box_2: 1.6290, loss_cns_2: 0.6537, loss_yns_2: 0.1535, loss_cls_3: 0.9208, loss_box_3: 1.6259, loss_cns_3: 0.6511, loss_yns_3: 0.1538, loss_cls_4: 0.9160, loss_box_4: 1.6210, loss_cns_4: 0.6559, loss_yns_4: 0.1540, loss_cls_5: 0.9088, loss_box_5: 1.6292, loss_cns_5: 0.6517, loss_yns_5: 0.1542, loss_cls_dn_0: 0.2018, loss_box_dn_0: 0.7629, loss_cls_dn_1: 0.1303, loss_box_dn_1: 0.7300, loss_cls_dn_2: 0.1294, loss_box_dn_2: 0.7138, loss_cls_dn_3: 0.1301, loss_box_dn_3: 0.7121, loss_cls_dn_4: 0.1334, loss_box_dn_4: 0.7154, loss_cls_dn_5: 0.1381, loss_box_dn_5: 0.7193, loss_dense_depth: 0.7877, loss: 26.1015, grad_norm: 35.5707 -2025-11-17 14:25:30,513 - mmdet - INFO - Iter [217/17500] lr: 1.863e-04, eta: 9:55:32, time: 1.537, data_time: 0.078, memory: 49163, loss_cls_0: 0.8008, loss_box_0: 1.7008, loss_cns_0: 0.6239, loss_yns_0: 0.1572, loss_cls_1: 0.8749, loss_box_1: 1.6210, loss_cns_1: 0.6518, loss_yns_1: 0.1562, loss_cls_2: 0.8825, loss_box_2: 1.6078, loss_cns_2: 0.6515, loss_yns_2: 0.1556, loss_cls_3: 0.8748, loss_box_3: 1.5878, loss_cns_3: 0.6520, loss_yns_3: 0.1565, loss_cls_4: 0.8759, loss_box_4: 1.5955, loss_cns_4: 0.6549, loss_yns_4: 0.1557, loss_cls_5: 0.8819, loss_box_5: 1.5873, loss_cns_5: 0.6544, loss_yns_5: 0.1582, loss_cls_dn_0: 0.1932, loss_box_dn_0: 0.7467, loss_cls_dn_1: 0.1236, loss_box_dn_1: 0.7045, loss_cls_dn_2: 0.1224, loss_box_dn_2: 0.6944, loss_cls_dn_3: 0.1217, loss_box_dn_3: 0.6882, loss_cls_dn_4: 0.1271, loss_box_dn_4: 0.6968, loss_cls_dn_5: 0.1305, loss_box_dn_5: 0.6954, loss_dense_depth: 0.7176, loss: 25.4809, grad_norm: 33.2232 -2025-11-17 14:25:32,005 - mmdet - INFO - Iter [218/17500] lr: 1.867e-04, eta: 9:54:44, time: 1.492, data_time: 0.079, memory: 49163, loss_cls_0: 0.7954, loss_box_0: 1.6877, loss_cns_0: 0.6272, loss_yns_0: 0.1550, loss_cls_1: 0.8720, loss_box_1: 1.6252, loss_cns_1: 0.6520, loss_yns_1: 0.1549, loss_cls_2: 0.8818, loss_box_2: 1.6116, loss_cns_2: 0.6529, loss_yns_2: 0.1536, loss_cls_3: 0.8784, loss_box_3: 1.5978, loss_cns_3: 0.6524, loss_yns_3: 0.1542, loss_cls_4: 0.8829, loss_box_4: 1.6026, loss_cns_4: 0.6547, loss_yns_4: 0.1547, loss_cls_5: 0.8842, loss_box_5: 1.5961, loss_cns_5: 0.6555, loss_yns_5: 0.1546, loss_cls_dn_0: 0.1919, loss_box_dn_0: 0.7464, loss_cls_dn_1: 0.1257, loss_box_dn_1: 0.6779, loss_cls_dn_2: 0.1274, loss_box_dn_2: 0.6687, loss_cls_dn_3: 0.1256, loss_box_dn_3: 0.6688, loss_cls_dn_4: 0.1310, loss_box_dn_4: 0.6795, loss_cls_dn_5: 0.1368, loss_box_dn_5: 0.6820, loss_dense_depth: 0.7560, loss: 25.4552, grad_norm: 35.3641 -2025-11-17 14:25:33,493 - mmdet - INFO - Iter [219/17500] lr: 1.871e-04, eta: 9:53:57, time: 1.489, data_time: 0.081, memory: 49163, loss_cls_0: 0.8199, loss_box_0: 1.7238, loss_cns_0: 0.6214, loss_yns_0: 0.1540, loss_cls_1: 0.8967, loss_box_1: 1.6605, loss_cns_1: 0.6521, loss_yns_1: 0.1536, loss_cls_2: 0.9087, loss_box_2: 1.6352, loss_cns_2: 0.6553, loss_yns_2: 0.1533, loss_cls_3: 0.9114, loss_box_3: 1.6257, loss_cns_3: 0.6556, loss_yns_3: 0.1520, loss_cls_4: 0.9161, loss_box_4: 1.6353, loss_cns_4: 0.6594, loss_yns_4: 0.1546, loss_cls_5: 0.9169, loss_box_5: 1.6426, loss_cns_5: 0.6596, loss_yns_5: 0.1534, loss_cls_dn_0: 0.1982, loss_box_dn_0: 0.7491, loss_cls_dn_1: 0.1264, loss_box_dn_1: 0.6781, loss_cls_dn_2: 0.1259, loss_box_dn_2: 0.6705, loss_cls_dn_3: 0.1267, loss_box_dn_3: 0.6771, loss_cls_dn_4: 0.1306, loss_box_dn_4: 0.6919, loss_cls_dn_5: 0.1397, loss_box_dn_5: 0.7040, loss_dense_depth: 0.7404, loss: 25.8759, grad_norm: 31.4464 -2025-11-17 14:25:34,980 - mmdet - INFO - Iter [220/17500] lr: 1.875e-04, eta: 9:53:10, time: 1.486, data_time: 0.081, memory: 49163, loss_cls_0: 0.7902, loss_box_0: 1.7004, loss_cns_0: 0.6259, loss_yns_0: 0.1531, loss_cls_1: 0.8784, loss_box_1: 1.6563, loss_cns_1: 0.6552, loss_yns_1: 0.1543, loss_cls_2: 0.8870, loss_box_2: 1.6390, loss_cns_2: 0.6548, loss_yns_2: 0.1536, loss_cls_3: 0.8875, loss_box_3: 1.6229, loss_cns_3: 0.6552, loss_yns_3: 0.1524, loss_cls_4: 0.8794, loss_box_4: 1.6317, loss_cns_4: 0.6554, loss_yns_4: 0.1528, loss_cls_5: 0.8915, loss_box_5: 1.6415, loss_cns_5: 0.6540, loss_yns_5: 0.1529, loss_cls_dn_0: 0.1899, loss_box_dn_0: 0.7418, loss_cls_dn_1: 0.1207, loss_box_dn_1: 0.6927, loss_cls_dn_2: 0.1191, loss_box_dn_2: 0.6899, loss_cls_dn_3: 0.1196, loss_box_dn_3: 0.6936, loss_cls_dn_4: 0.1225, loss_box_dn_4: 0.7093, loss_cls_dn_5: 0.1302, loss_box_dn_5: 0.7220, loss_dense_depth: 0.7554, loss: 25.7321, grad_norm: 35.4549 -2025-11-17 14:25:36,510 - mmdet - INFO - Iter [221/17500] lr: 1.879e-04, eta: 9:52:26, time: 1.531, data_time: 0.073, memory: 49163, loss_cls_0: 0.7959, loss_box_0: 1.6899, loss_cns_0: 0.6190, loss_yns_0: 0.1528, loss_cls_1: 0.8824, loss_box_1: 1.6633, loss_cns_1: 0.6545, loss_yns_1: 0.1517, loss_cls_2: 0.8907, loss_box_2: 1.6506, loss_cns_2: 0.6533, loss_yns_2: 0.1523, loss_cls_3: 0.8868, loss_box_3: 1.6330, loss_cns_3: 0.6539, loss_yns_3: 0.1515, loss_cls_4: 0.8883, loss_box_4: 1.6268, loss_cns_4: 0.6566, loss_yns_4: 0.1510, loss_cls_5: 0.8914, loss_box_5: 1.6374, loss_cns_5: 0.6524, loss_yns_5: 0.1527, loss_cls_dn_0: 0.1926, loss_box_dn_0: 0.7455, loss_cls_dn_1: 0.1276, loss_box_dn_1: 0.7069, loss_cls_dn_2: 0.1234, loss_box_dn_2: 0.6992, loss_cls_dn_3: 0.1251, loss_box_dn_3: 0.7031, loss_cls_dn_4: 0.1294, loss_box_dn_4: 0.7124, loss_cls_dn_5: 0.1315, loss_box_dn_5: 0.7250, loss_dense_depth: 0.7582, loss: 25.8180, grad_norm: 32.1720 -2025-11-17 14:25:38,071 - mmdet - INFO - Iter [222/17500] lr: 1.883e-04, eta: 9:51:46, time: 1.562, data_time: 0.075, memory: 49163, loss_cls_0: 0.7920, loss_box_0: 1.6877, loss_cns_0: 0.6178, loss_yns_0: 0.1508, loss_cls_1: 0.8768, loss_box_1: 1.6176, loss_cns_1: 0.6572, loss_yns_1: 0.1501, loss_cls_2: 0.8894, loss_box_2: 1.5992, loss_cns_2: 0.6587, loss_yns_2: 0.1498, loss_cls_3: 0.8922, loss_box_3: 1.5903, loss_cns_3: 0.6589, loss_yns_3: 0.1496, loss_cls_4: 0.8939, loss_box_4: 1.5749, loss_cns_4: 0.6598, loss_yns_4: 0.1504, loss_cls_5: 0.8884, loss_box_5: 1.5945, loss_cns_5: 0.6583, loss_yns_5: 0.1502, loss_cls_dn_0: 0.1944, loss_box_dn_0: 0.7468, loss_cls_dn_1: 0.1249, loss_box_dn_1: 0.7074, loss_cls_dn_2: 0.1230, loss_box_dn_2: 0.6948, loss_cls_dn_3: 0.1261, loss_box_dn_3: 0.6984, loss_cls_dn_4: 0.1306, loss_box_dn_4: 0.7023, loss_cls_dn_5: 0.1338, loss_box_dn_5: 0.7116, loss_dense_depth: 0.7527, loss: 25.5555, grad_norm: 32.6403 -2025-11-17 14:25:39,599 - mmdet - INFO - Iter [223/17500] lr: 1.887e-04, eta: 9:51:03, time: 1.527, data_time: 0.075, memory: 49163, loss_cls_0: 0.8132, loss_box_0: 1.7328, loss_cns_0: 0.6204, loss_yns_0: 0.1528, loss_cls_1: 0.8836, loss_box_1: 1.6645, loss_cns_1: 0.6546, loss_yns_1: 0.1525, loss_cls_2: 0.8942, loss_box_2: 1.6306, loss_cns_2: 0.6557, loss_yns_2: 0.1520, loss_cls_3: 0.9015, loss_box_3: 1.6094, loss_cns_3: 0.6574, loss_yns_3: 0.1510, loss_cls_4: 0.9065, loss_box_4: 1.6134, loss_cns_4: 0.6576, loss_yns_4: 0.1527, loss_cls_5: 0.9056, loss_box_5: 1.6066, loss_cns_5: 0.6578, loss_yns_5: 0.1520, loss_cls_dn_0: 0.1980, loss_box_dn_0: 0.7519, loss_cls_dn_1: 0.1283, loss_box_dn_1: 0.7222, loss_cls_dn_2: 0.1279, loss_box_dn_2: 0.7105, loss_cls_dn_3: 0.1307, loss_box_dn_3: 0.7085, loss_cls_dn_4: 0.1336, loss_box_dn_4: 0.7129, loss_cls_dn_5: 0.1373, loss_box_dn_5: 0.7130, loss_dense_depth: 0.7771, loss: 25.9304, grad_norm: 29.0301 -2025-11-17 14:25:41,104 - mmdet - INFO - Iter [224/17500] lr: 1.891e-04, eta: 9:50:18, time: 1.505, data_time: 0.078, memory: 49163, loss_cls_0: 0.7844, loss_box_0: 1.6978, loss_cns_0: 0.6238, loss_yns_0: 0.1517, loss_cls_1: 0.8634, loss_box_1: 1.6669, loss_cns_1: 0.6536, loss_yns_1: 0.1529, loss_cls_2: 0.8812, loss_box_2: 1.6359, loss_cns_2: 0.6572, loss_yns_2: 0.1510, loss_cls_3: 0.8888, loss_box_3: 1.6231, loss_cns_3: 0.6580, loss_yns_3: 0.1510, loss_cls_4: 0.8877, loss_box_4: 1.6276, loss_cns_4: 0.6576, loss_yns_4: 0.1510, loss_cls_5: 0.8932, loss_box_5: 1.6287, loss_cns_5: 0.6570, loss_yns_5: 0.1517, loss_cls_dn_0: 0.1867, loss_box_dn_0: 0.7498, loss_cls_dn_1: 0.1225, loss_box_dn_1: 0.7042, loss_cls_dn_2: 0.1225, loss_box_dn_2: 0.6898, loss_cls_dn_3: 0.1228, loss_box_dn_3: 0.6890, loss_cls_dn_4: 0.1250, loss_box_dn_4: 0.6919, loss_cls_dn_5: 0.1321, loss_box_dn_5: 0.6936, loss_dense_depth: 0.7389, loss: 25.6640, grad_norm: 30.2348 -2025-11-17 14:25:42,617 - mmdet - INFO - Iter [225/17500] lr: 1.895e-04, eta: 9:49:35, time: 1.514, data_time: 0.080, memory: 49163, loss_cls_0: 0.7817, loss_box_0: 1.7096, loss_cns_0: 0.6225, loss_yns_0: 0.1523, loss_cls_1: 0.8690, loss_box_1: 1.6664, loss_cns_1: 0.6502, loss_yns_1: 0.1531, loss_cls_2: 0.8857, loss_box_2: 1.6208, loss_cns_2: 0.6564, loss_yns_2: 0.1517, loss_cls_3: 0.8894, loss_box_3: 1.6167, loss_cns_3: 0.6576, loss_yns_3: 0.1526, loss_cls_4: 0.8869, loss_box_4: 1.6139, loss_cns_4: 0.6579, loss_yns_4: 0.1533, loss_cls_5: 0.8913, loss_box_5: 1.6282, loss_cns_5: 0.6581, loss_yns_5: 0.1532, loss_cls_dn_0: 0.1839, loss_box_dn_0: 0.7452, loss_cls_dn_1: 0.1226, loss_box_dn_1: 0.6901, loss_cls_dn_2: 0.1216, loss_box_dn_2: 0.6744, loss_cls_dn_3: 0.1204, loss_box_dn_3: 0.6817, loss_cls_dn_4: 0.1253, loss_box_dn_4: 0.6865, loss_cls_dn_5: 0.1287, loss_box_dn_5: 0.6981, loss_dense_depth: 0.7465, loss: 25.6037, grad_norm: 32.5947 -2025-11-17 14:25:44,157 - mmdet - INFO - Iter [226/17500] lr: 1.899e-04, eta: 9:48:54, time: 1.538, data_time: 0.078, memory: 49163, loss_cls_0: 0.8030, loss_box_0: 1.6986, loss_cns_0: 0.6278, loss_yns_0: 0.1512, loss_cls_1: 0.8785, loss_box_1: 1.6682, loss_cns_1: 0.6509, loss_yns_1: 0.1517, loss_cls_2: 0.8974, loss_box_2: 1.6248, loss_cns_2: 0.6549, loss_yns_2: 0.1499, loss_cls_3: 0.8945, loss_box_3: 1.6071, loss_cns_3: 0.6559, loss_yns_3: 0.1495, loss_cls_4: 0.8999, loss_box_4: 1.6114, loss_cns_4: 0.6568, loss_yns_4: 0.1511, loss_cls_5: 0.8963, loss_box_5: 1.6164, loss_cns_5: 0.6550, loss_yns_5: 0.1501, loss_cls_dn_0: 0.1887, loss_box_dn_0: 0.7475, loss_cls_dn_1: 0.1253, loss_box_dn_1: 0.6975, loss_cls_dn_2: 0.1239, loss_box_dn_2: 0.6869, loss_cls_dn_3: 0.1233, loss_box_dn_3: 0.6956, loss_cls_dn_4: 0.1290, loss_box_dn_4: 0.7057, loss_cls_dn_5: 0.1302, loss_box_dn_5: 0.7195, loss_dense_depth: 0.7320, loss: 25.7059, grad_norm: 31.9934 -2025-11-17 14:25:45,673 - mmdet - INFO - Iter [227/17500] lr: 1.903e-04, eta: 9:48:11, time: 1.511, data_time: 0.087, memory: 49163, loss_cls_0: 0.8039, loss_box_0: 1.7314, loss_cns_0: 0.6196, loss_yns_0: 0.1504, loss_cls_1: 0.8589, loss_box_1: 1.7038, loss_cns_1: 0.6401, loss_yns_1: 0.1485, loss_cls_2: 0.8870, loss_box_2: 1.6612, loss_cns_2: 0.6480, loss_yns_2: 0.1472, loss_cls_3: 0.8875, loss_box_3: 1.6402, loss_cns_3: 0.6480, loss_yns_3: 0.1481, loss_cls_4: 0.8887, loss_box_4: 1.6561, loss_cns_4: 0.6495, loss_yns_4: 0.1495, loss_cls_5: 0.8883, loss_box_5: 1.6503, loss_cns_5: 0.6486, loss_yns_5: 0.1486, loss_cls_dn_0: 0.1890, loss_box_dn_0: 0.7457, loss_cls_dn_1: 0.1231, loss_box_dn_1: 0.7160, loss_cls_dn_2: 0.1208, loss_box_dn_2: 0.7090, loss_cls_dn_3: 0.1215, loss_box_dn_3: 0.7183, loss_cls_dn_4: 0.1271, loss_box_dn_4: 0.7356, loss_cls_dn_5: 0.1305, loss_box_dn_5: 0.7503, loss_dense_depth: 0.7557, loss: 25.9458, grad_norm: 38.0890 -2025-11-17 14:25:47,170 - mmdet - INFO - Iter [228/17500] lr: 1.907e-04, eta: 9:47:28, time: 1.504, data_time: 0.092, memory: 49163, loss_cls_0: 0.8166, loss_box_0: 1.7247, loss_cns_0: 0.6256, loss_yns_0: 0.1538, loss_cls_1: 0.8625, loss_box_1: 1.7052, loss_cns_1: 0.6453, loss_yns_1: 0.1517, loss_cls_2: 0.8720, loss_box_2: 1.6681, loss_cns_2: 0.6518, loss_yns_2: 0.1497, loss_cls_3: 0.8748, loss_box_3: 1.6606, loss_cns_3: 0.6500, loss_yns_3: 0.1495, loss_cls_4: 0.8762, loss_box_4: 1.6703, loss_cns_4: 0.6488, loss_yns_4: 0.1510, loss_cls_5: 0.8821, loss_box_5: 1.6760, loss_cns_5: 0.6509, loss_yns_5: 0.1509, loss_cls_dn_0: 0.1904, loss_box_dn_0: 0.7432, loss_cls_dn_1: 0.1216, loss_box_dn_1: 0.7418, loss_cls_dn_2: 0.1205, loss_box_dn_2: 0.7316, loss_cls_dn_3: 0.1221, loss_box_dn_3: 0.7437, loss_cls_dn_4: 0.1251, loss_box_dn_4: 0.7603, loss_cls_dn_5: 0.1281, loss_box_dn_5: 0.7724, loss_dense_depth: 0.7474, loss: 26.1165, grad_norm: 36.6641 -2025-11-17 14:25:48,755 - mmdet - INFO - Iter [229/17500] lr: 1.911e-04, eta: 9:46:52, time: 1.583, data_time: 0.178, memory: 49163, loss_cls_0: 0.8129, loss_box_0: 1.7257, loss_cns_0: 0.6173, loss_yns_0: 0.1491, loss_cls_1: 0.8604, loss_box_1: 1.6672, loss_cns_1: 0.6479, loss_yns_1: 0.1499, loss_cls_2: 0.8789, loss_box_2: 1.6399, loss_cns_2: 0.6480, loss_yns_2: 0.1467, loss_cls_3: 0.8757, loss_box_3: 1.6388, loss_cns_3: 0.6490, loss_yns_3: 0.1474, loss_cls_4: 0.8794, loss_box_4: 1.6369, loss_cns_4: 0.6492, loss_yns_4: 0.1473, loss_cls_5: 0.8897, loss_box_5: 1.6413, loss_cns_5: 0.6534, loss_yns_5: 0.1477, loss_cls_dn_0: 0.1842, loss_box_dn_0: 0.7396, loss_cls_dn_1: 0.1243, loss_box_dn_1: 0.7585, loss_cls_dn_2: 0.1231, loss_box_dn_2: 0.7429, loss_cls_dn_3: 0.1220, loss_box_dn_3: 0.7527, loss_cls_dn_4: 0.1274, loss_box_dn_4: 0.7688, loss_cls_dn_5: 0.1324, loss_box_dn_5: 0.7782, loss_dense_depth: 0.7623, loss: 26.0162, grad_norm: 37.8265 -2025-11-17 14:25:50,244 - mmdet - INFO - Iter [230/17500] lr: 1.915e-04, eta: 9:46:09, time: 1.490, data_time: 0.081, memory: 49163, loss_cls_0: 0.8135, loss_box_0: 1.7363, loss_cns_0: 0.6189, loss_yns_0: 0.1495, loss_cls_1: 0.8670, loss_box_1: 1.6557, loss_cns_1: 0.6491, loss_yns_1: 0.1498, loss_cls_2: 0.8813, loss_box_2: 1.6405, loss_cns_2: 0.6533, loss_yns_2: 0.1495, loss_cls_3: 0.8707, loss_box_3: 1.6332, loss_cns_3: 0.6537, loss_yns_3: 0.1496, loss_cls_4: 0.8724, loss_box_4: 1.6310, loss_cns_4: 0.6554, loss_yns_4: 0.1493, loss_cls_5: 0.8718, loss_box_5: 1.6470, loss_cns_5: 0.6574, loss_yns_5: 0.1514, loss_cls_dn_0: 0.1869, loss_box_dn_0: 0.7513, loss_cls_dn_1: 0.1220, loss_box_dn_1: 0.7497, loss_cls_dn_2: 0.1192, loss_box_dn_2: 0.7322, loss_cls_dn_3: 0.1186, loss_box_dn_3: 0.7375, loss_cls_dn_4: 0.1241, loss_box_dn_4: 0.7476, loss_cls_dn_5: 0.1265, loss_box_dn_5: 0.7590, loss_dense_depth: 0.7605, loss: 25.9422, grad_norm: 32.1963 -2025-11-17 14:25:51,732 - mmdet - INFO - Iter [231/17500] lr: 1.919e-04, eta: 9:45:26, time: 1.488, data_time: 0.078, memory: 49163, loss_cls_0: 0.7792, loss_box_0: 1.6985, loss_cns_0: 0.6244, loss_yns_0: 0.1503, loss_cls_1: 0.8644, loss_box_1: 1.6119, loss_cns_1: 0.6549, loss_yns_1: 0.1481, loss_cls_2: 0.8659, loss_box_2: 1.6008, loss_cns_2: 0.6584, loss_yns_2: 0.1471, loss_cls_3: 0.8560, loss_box_3: 1.5907, loss_cns_3: 0.6579, loss_yns_3: 0.1466, loss_cls_4: 0.8661, loss_box_4: 1.5918, loss_cns_4: 0.6578, loss_yns_4: 0.1473, loss_cls_5: 0.8607, loss_box_5: 1.6078, loss_cns_5: 0.6564, loss_yns_5: 0.1469, loss_cls_dn_0: 0.1794, loss_box_dn_0: 0.7437, loss_cls_dn_1: 0.1198, loss_box_dn_1: 0.7117, loss_cls_dn_2: 0.1161, loss_box_dn_2: 0.6976, loss_cls_dn_3: 0.1184, loss_box_dn_3: 0.6981, loss_cls_dn_4: 0.1241, loss_box_dn_4: 0.7061, loss_cls_dn_5: 0.1269, loss_box_dn_5: 0.7163, loss_dense_depth: 0.7273, loss: 25.3754, grad_norm: 36.2046 -2025-11-17 14:25:53,225 - mmdet - INFO - Iter [232/17500] lr: 1.923e-04, eta: 9:44:43, time: 1.493, data_time: 0.078, memory: 49163, loss_cls_0: 0.8028, loss_box_0: 1.7083, loss_cns_0: 0.6234, loss_yns_0: 0.1504, loss_cls_1: 0.8821, loss_box_1: 1.6241, loss_cns_1: 0.6569, loss_yns_1: 0.1474, loss_cls_2: 0.8836, loss_box_2: 1.6148, loss_cns_2: 0.6589, loss_yns_2: 0.1488, loss_cls_3: 0.8752, loss_box_3: 1.5961, loss_cns_3: 0.6581, loss_yns_3: 0.1472, loss_cls_4: 0.8786, loss_box_4: 1.5974, loss_cns_4: 0.6590, loss_yns_4: 0.1489, loss_cls_5: 0.8715, loss_box_5: 1.6140, loss_cns_5: 0.6585, loss_yns_5: 0.1484, loss_cls_dn_0: 0.1789, loss_box_dn_0: 0.7556, loss_cls_dn_1: 0.1217, loss_box_dn_1: 0.7046, loss_cls_dn_2: 0.1205, loss_box_dn_2: 0.6946, loss_cls_dn_3: 0.1221, loss_box_dn_3: 0.6918, loss_cls_dn_4: 0.1322, loss_box_dn_4: 0.7014, loss_cls_dn_5: 0.1344, loss_box_dn_5: 0.7116, loss_dense_depth: 0.7543, loss: 25.5782, grad_norm: 30.0314 -2025-11-17 14:25:54,706 - mmdet - INFO - Iter [233/17500] lr: 1.927e-04, eta: 9:44:00, time: 1.481, data_time: 0.078, memory: 49163, loss_cls_0: 0.8175, loss_box_0: 1.6949, loss_cns_0: 0.6242, loss_yns_0: 0.1514, loss_cls_1: 0.8883, loss_box_1: 1.5997, loss_cns_1: 0.6547, loss_yns_1: 0.1499, loss_cls_2: 0.8986, loss_box_2: 1.5827, loss_cns_2: 0.6563, loss_yns_2: 0.1491, loss_cls_3: 0.8961, loss_box_3: 1.5601, loss_cns_3: 0.6555, loss_yns_3: 0.1478, loss_cls_4: 0.8928, loss_box_4: 1.5667, loss_cns_4: 0.6562, loss_yns_4: 0.1482, loss_cls_5: 0.8948, loss_box_5: 1.5716, loss_cns_5: 0.6566, loss_yns_5: 0.1488, loss_cls_dn_0: 0.1762, loss_box_dn_0: 0.7577, loss_cls_dn_1: 0.1255, loss_box_dn_1: 0.7249, loss_cls_dn_2: 0.1241, loss_box_dn_2: 0.7099, loss_cls_dn_3: 0.1264, loss_box_dn_3: 0.7072, loss_cls_dn_4: 0.1307, loss_box_dn_4: 0.7173, loss_cls_dn_5: 0.1336, loss_box_dn_5: 0.7258, loss_dense_depth: 0.7611, loss: 25.5828, grad_norm: 34.6374 -2025-11-17 14:25:56,262 - mmdet - INFO - Iter [234/17500] lr: 1.931e-04, eta: 9:43:24, time: 1.556, data_time: 0.079, memory: 49163, loss_cls_0: 0.7806, loss_box_0: 1.6641, loss_cns_0: 0.6215, loss_yns_0: 0.1505, loss_cls_1: 0.8730, loss_box_1: 1.5690, loss_cns_1: 0.6538, loss_yns_1: 0.1506, loss_cls_2: 0.8811, loss_box_2: 1.5387, loss_cns_2: 0.6562, loss_yns_2: 0.1501, loss_cls_3: 0.8817, loss_box_3: 1.5271, loss_cns_3: 0.6560, loss_yns_3: 0.1493, loss_cls_4: 0.8787, loss_box_4: 1.5249, loss_cns_4: 0.6579, loss_yns_4: 0.1502, loss_cls_5: 0.8849, loss_box_5: 1.5365, loss_cns_5: 0.6567, loss_yns_5: 0.1503, loss_cls_dn_0: 0.1753, loss_box_dn_0: 0.7526, loss_cls_dn_1: 0.1167, loss_box_dn_1: 0.7173, loss_cls_dn_2: 0.1184, loss_box_dn_2: 0.7021, loss_cls_dn_3: 0.1218, loss_box_dn_3: 0.7019, loss_cls_dn_4: 0.1201, loss_box_dn_4: 0.7099, loss_cls_dn_5: 0.1270, loss_box_dn_5: 0.7206, loss_dense_depth: 0.7435, loss: 25.1706, grad_norm: 32.4673 -2025-11-17 14:25:57,758 - mmdet - INFO - Iter [235/17500] lr: 1.935e-04, eta: 9:42:42, time: 1.495, data_time: 0.076, memory: 49163, loss_cls_0: 0.8106, loss_box_0: 1.6782, loss_cns_0: 0.6207, loss_yns_0: 0.1509, loss_cls_1: 0.8809, loss_box_1: 1.5501, loss_cns_1: 0.6517, loss_yns_1: 0.1516, loss_cls_2: 0.8864, loss_box_2: 1.5517, loss_cns_2: 0.6533, loss_yns_2: 0.1514, loss_cls_3: 0.8803, loss_box_3: 1.5440, loss_cns_3: 0.6531, loss_yns_3: 0.1518, loss_cls_4: 0.8856, loss_box_4: 1.5443, loss_cns_4: 0.6565, loss_yns_4: 0.1516, loss_cls_5: 0.8806, loss_box_5: 1.5493, loss_cns_5: 0.6561, loss_yns_5: 0.1530, loss_cls_dn_0: 0.1786, loss_box_dn_0: 0.7425, loss_cls_dn_1: 0.1183, loss_box_dn_1: 0.7012, loss_cls_dn_2: 0.1195, loss_box_dn_2: 0.6917, loss_cls_dn_3: 0.1207, loss_box_dn_3: 0.6963, loss_cls_dn_4: 0.1213, loss_box_dn_4: 0.7076, loss_cls_dn_5: 0.1278, loss_box_dn_5: 0.7216, loss_dense_depth: 0.7697, loss: 25.2605, grad_norm: 29.2441 -2025-11-17 14:25:59,230 - mmdet - INFO - Iter [236/17500] lr: 1.939e-04, eta: 9:42:00, time: 1.473, data_time: 0.076, memory: 49163, loss_cls_0: 0.7745, loss_box_0: 1.6796, loss_cns_0: 0.6206, loss_yns_0: 0.1532, loss_cls_1: 0.8585, loss_box_1: 1.5633, loss_cns_1: 0.6529, loss_yns_1: 0.1539, loss_cls_2: 0.8738, loss_box_2: 1.5469, loss_cns_2: 0.6548, loss_yns_2: 0.1519, loss_cls_3: 0.8618, loss_box_3: 1.5429, loss_cns_3: 0.6558, loss_yns_3: 0.1529, loss_cls_4: 0.8656, loss_box_4: 1.5447, loss_cns_4: 0.6549, loss_yns_4: 0.1535, loss_cls_5: 0.8659, loss_box_5: 1.5414, loss_cns_5: 0.6549, loss_yns_5: 0.1537, loss_cls_dn_0: 0.1663, loss_box_dn_0: 0.7344, loss_cls_dn_1: 0.1219, loss_box_dn_1: 0.7037, loss_cls_dn_2: 0.1183, loss_box_dn_2: 0.6937, loss_cls_dn_3: 0.1165, loss_box_dn_3: 0.7014, loss_cls_dn_4: 0.1186, loss_box_dn_4: 0.7146, loss_cls_dn_5: 0.1233, loss_box_dn_5: 0.7276, loss_dense_depth: 0.7709, loss: 25.1430, grad_norm: 35.6523 -2025-11-17 14:26:00,780 - mmdet - INFO - Iter [237/17500] lr: 1.943e-04, eta: 9:41:23, time: 1.550, data_time: 0.077, memory: 49163, loss_cls_0: 0.7742, loss_box_0: 1.6567, loss_cns_0: 0.6230, loss_yns_0: 0.1535, loss_cls_1: 0.8548, loss_box_1: 1.5447, loss_cns_1: 0.6562, loss_yns_1: 0.1539, loss_cls_2: 0.8703, loss_box_2: 1.5118, loss_cns_2: 0.6591, loss_yns_2: 0.1538, loss_cls_3: 0.8724, loss_box_3: 1.5110, loss_cns_3: 0.6597, loss_yns_3: 0.1529, loss_cls_4: 0.8741, loss_box_4: 1.5054, loss_cns_4: 0.6590, loss_yns_4: 0.1540, loss_cls_5: 0.8799, loss_box_5: 1.5158, loss_cns_5: 0.6586, loss_yns_5: 0.1543, loss_cls_dn_0: 0.1755, loss_box_dn_0: 0.7440, loss_cls_dn_1: 0.1171, loss_box_dn_1: 0.7190, loss_cls_dn_2: 0.1141, loss_box_dn_2: 0.7038, loss_cls_dn_3: 0.1186, loss_box_dn_3: 0.7165, loss_cls_dn_4: 0.1212, loss_box_dn_4: 0.7239, loss_cls_dn_5: 0.1255, loss_box_dn_5: 0.7353, loss_dense_depth: 0.7316, loss: 25.0549, grad_norm: 34.8065 -2025-11-17 14:26:02,267 - mmdet - INFO - Iter [238/17500] lr: 1.947e-04, eta: 9:40:43, time: 1.488, data_time: 0.082, memory: 49163, loss_cls_0: 0.7740, loss_box_0: 1.6426, loss_cns_0: 0.6255, loss_yns_0: 0.1541, loss_cls_1: 0.8563, loss_box_1: 1.5416, loss_cns_1: 0.6592, loss_yns_1: 0.1547, loss_cls_2: 0.8708, loss_box_2: 1.5062, loss_cns_2: 0.6619, loss_yns_2: 0.1531, loss_cls_3: 0.8659, loss_box_3: 1.4994, loss_cns_3: 0.6626, loss_yns_3: 0.1537, loss_cls_4: 0.8689, loss_box_4: 1.4950, loss_cns_4: 0.6622, loss_yns_4: 0.1542, loss_cls_5: 0.8732, loss_box_5: 1.5030, loss_cns_5: 0.6615, loss_yns_5: 0.1551, loss_cls_dn_0: 0.1736, loss_box_dn_0: 0.7410, loss_cls_dn_1: 0.1155, loss_box_dn_1: 0.7381, loss_cls_dn_2: 0.1184, loss_box_dn_2: 0.7167, loss_cls_dn_3: 0.1206, loss_box_dn_3: 0.7248, loss_cls_dn_4: 0.1215, loss_box_dn_4: 0.7302, loss_cls_dn_5: 0.1275, loss_box_dn_5: 0.7360, loss_dense_depth: 0.7417, loss: 25.0601, grad_norm: 32.2392 -2025-11-17 14:26:03,775 - mmdet - INFO - Iter [239/17500] lr: 1.951e-04, eta: 9:40:04, time: 1.508, data_time: 0.081, memory: 49163, loss_cls_0: 0.7739, loss_box_0: 1.6184, loss_cns_0: 0.6291, loss_yns_0: 0.1539, loss_cls_1: 0.8548, loss_box_1: 1.5515, loss_cns_1: 0.6585, loss_yns_1: 0.1553, loss_cls_2: 0.8775, loss_box_2: 1.4993, loss_cns_2: 0.6627, loss_yns_2: 0.1540, loss_cls_3: 0.8792, loss_box_3: 1.4809, loss_cns_3: 0.6650, loss_yns_3: 0.1534, loss_cls_4: 0.8768, loss_box_4: 1.4803, loss_cns_4: 0.6647, loss_yns_4: 0.1544, loss_cls_5: 0.8663, loss_box_5: 1.4793, loss_cns_5: 0.6637, loss_yns_5: 0.1549, loss_cls_dn_0: 0.1687, loss_box_dn_0: 0.7322, loss_cls_dn_1: 0.1165, loss_box_dn_1: 0.7165, loss_cls_dn_2: 0.1170, loss_box_dn_2: 0.7024, loss_cls_dn_3: 0.1190, loss_box_dn_3: 0.7046, loss_cls_dn_4: 0.1199, loss_box_dn_4: 0.7096, loss_cls_dn_5: 0.1244, loss_box_dn_5: 0.7103, loss_dense_depth: 0.7282, loss: 24.8774, grad_norm: 29.6566 -2025-11-17 14:26:05,307 - mmdet - INFO - Iter [240/17500] lr: 1.955e-04, eta: 9:39:27, time: 1.533, data_time: 0.082, memory: 49163, loss_cls_0: 0.7741, loss_box_0: 1.6261, loss_cns_0: 0.6267, loss_yns_0: 0.1525, loss_cls_1: 0.8738, loss_box_1: 1.5264, loss_cns_1: 0.6636, loss_yns_1: 0.1539, loss_cls_2: 0.8872, loss_box_2: 1.4916, loss_cns_2: 0.6648, loss_yns_2: 0.1520, loss_cls_3: 0.8858, loss_box_3: 1.4729, loss_cns_3: 0.6662, loss_yns_3: 0.1519, loss_cls_4: 0.8859, loss_box_4: 1.4679, loss_cns_4: 0.6652, loss_yns_4: 0.1519, loss_cls_5: 0.8814, loss_box_5: 1.4760, loss_cns_5: 0.6669, loss_yns_5: 0.1541, loss_cls_dn_0: 0.1709, loss_box_dn_0: 0.7401, loss_cls_dn_1: 0.1179, loss_box_dn_1: 0.6812, loss_cls_dn_2: 0.1188, loss_box_dn_2: 0.6718, loss_cls_dn_3: 0.1213, loss_box_dn_3: 0.6756, loss_cls_dn_4: 0.1206, loss_box_dn_4: 0.6808, loss_cls_dn_5: 0.1245, loss_box_dn_5: 0.6863, loss_dense_depth: 0.6974, loss: 24.7262, grad_norm: 31.7308 -2025-11-17 14:26:06,850 - mmdet - INFO - Iter [241/17500] lr: 1.959e-04, eta: 9:38:51, time: 1.542, data_time: 0.079, memory: 49163, loss_cls_0: 0.7897, loss_box_0: 1.6406, loss_cns_0: 0.6243, loss_yns_0: 0.1553, loss_cls_1: 0.8753, loss_box_1: 1.5642, loss_cns_1: 0.6593, loss_yns_1: 0.1564, loss_cls_2: 0.8888, loss_box_2: 1.5200, loss_cns_2: 0.6642, loss_yns_2: 0.1555, loss_cls_3: 0.8922, loss_box_3: 1.5074, loss_cns_3: 0.6645, loss_yns_3: 0.1559, loss_cls_4: 0.8907, loss_box_4: 1.5017, loss_cns_4: 0.6635, loss_yns_4: 0.1554, loss_cls_5: 0.8975, loss_box_5: 1.5108, loss_cns_5: 0.6664, loss_yns_5: 0.1573, loss_cls_dn_0: 0.1707, loss_box_dn_0: 0.7424, loss_cls_dn_1: 0.1196, loss_box_dn_1: 0.6901, loss_cls_dn_2: 0.1198, loss_box_dn_2: 0.6813, loss_cls_dn_3: 0.1208, loss_box_dn_3: 0.6881, loss_cls_dn_4: 0.1203, loss_box_dn_4: 0.6975, loss_cls_dn_5: 0.1241, loss_box_dn_5: 0.7097, loss_dense_depth: 0.7697, loss: 25.1109, grad_norm: 30.4349 -2025-11-17 14:26:08,420 - mmdet - INFO - Iter [242/17500] lr: 1.963e-04, eta: 9:38:18, time: 1.571, data_time: 0.078, memory: 49163, loss_cls_0: 0.7932, loss_box_0: 1.6443, loss_cns_0: 0.6272, loss_yns_0: 0.1555, loss_cls_1: 0.8781, loss_box_1: 1.5599, loss_cns_1: 0.6587, loss_yns_1: 0.1567, loss_cls_2: 0.9053, loss_box_2: 1.5203, loss_cns_2: 0.6614, loss_yns_2: 0.1559, loss_cls_3: 0.8968, loss_box_3: 1.5124, loss_cns_3: 0.6618, loss_yns_3: 0.1554, loss_cls_4: 0.8966, loss_box_4: 1.5146, loss_cns_4: 0.6606, loss_yns_4: 0.1549, loss_cls_5: 0.9004, loss_box_5: 1.5236, loss_cns_5: 0.6607, loss_yns_5: 0.1559, loss_cls_dn_0: 0.1658, loss_box_dn_0: 0.7398, loss_cls_dn_1: 0.1180, loss_box_dn_1: 0.7021, loss_cls_dn_2: 0.1188, loss_box_dn_2: 0.6944, loss_cls_dn_3: 0.1182, loss_box_dn_3: 0.7032, loss_cls_dn_4: 0.1227, loss_box_dn_4: 0.7175, loss_cls_dn_5: 0.1263, loss_box_dn_5: 0.7303, loss_dense_depth: 0.7008, loss: 25.1683, grad_norm: 36.0097 -2025-11-17 14:26:09,924 - mmdet - INFO - Iter [243/17500] lr: 1.967e-04, eta: 9:37:40, time: 1.503, data_time: 0.077, memory: 49163, loss_cls_0: 0.7908, loss_box_0: 1.6252, loss_cns_0: 0.6242, loss_yns_0: 0.1535, loss_cls_1: 0.8744, loss_box_1: 1.5939, loss_cns_1: 0.6566, loss_yns_1: 0.1543, loss_cls_2: 0.8937, loss_box_2: 1.5658, loss_cns_2: 0.6587, loss_yns_2: 0.1538, loss_cls_3: 0.8789, loss_box_3: 1.5620, loss_cns_3: 0.6579, loss_yns_3: 0.1535, loss_cls_4: 0.8802, loss_box_4: 1.5688, loss_cns_4: 0.6564, loss_yns_4: 0.1552, loss_cls_5: 0.8823, loss_box_5: 1.5700, loss_cns_5: 0.6552, loss_yns_5: 0.1544, loss_cls_dn_0: 0.1706, loss_box_dn_0: 0.7344, loss_cls_dn_1: 0.1216, loss_box_dn_1: 0.7121, loss_cls_dn_2: 0.1218, loss_box_dn_2: 0.7061, loss_cls_dn_3: 0.1215, loss_box_dn_3: 0.7149, loss_cls_dn_4: 0.1279, loss_box_dn_4: 0.7345, loss_cls_dn_5: 0.1295, loss_box_dn_5: 0.7455, loss_dense_depth: 0.7034, loss: 25.3635, grad_norm: 37.2759 -2025-11-17 14:26:11,437 - mmdet - INFO - Iter [244/17500] lr: 1.971e-04, eta: 9:37:03, time: 1.513, data_time: 0.080, memory: 49163, loss_cls_0: 0.7958, loss_box_0: 1.6576, loss_cns_0: 0.6186, loss_yns_0: 0.1568, loss_cls_1: 0.8848, loss_box_1: 1.5928, loss_cns_1: 0.6488, loss_yns_1: 0.1548, loss_cls_2: 0.8959, loss_box_2: 1.5656, loss_cns_2: 0.6544, loss_yns_2: 0.1555, loss_cls_3: 0.8915, loss_box_3: 1.5429, loss_cns_3: 0.6516, loss_yns_3: 0.1544, loss_cls_4: 0.8920, loss_box_4: 1.5588, loss_cns_4: 0.6524, loss_yns_4: 0.1547, loss_cls_5: 0.9068, loss_box_5: 1.5629, loss_cns_5: 0.6523, loss_yns_5: 0.1544, loss_cls_dn_0: 0.1710, loss_box_dn_0: 0.7399, loss_cls_dn_1: 0.1209, loss_box_dn_1: 0.7324, loss_cls_dn_2: 0.1228, loss_box_dn_2: 0.7249, loss_cls_dn_3: 0.1233, loss_box_dn_3: 0.7287, loss_cls_dn_4: 0.1271, loss_box_dn_4: 0.7439, loss_cls_dn_5: 0.1287, loss_box_dn_5: 0.7533, loss_dense_depth: 0.7239, loss: 25.4970, grad_norm: 37.2123 -2025-11-17 14:26:12,935 - mmdet - INFO - Iter [245/17500] lr: 1.975e-04, eta: 9:36:25, time: 1.498, data_time: 0.080, memory: 49163, loss_cls_0: 0.7681, loss_box_0: 1.6729, loss_cns_0: 0.6216, loss_yns_0: 0.1554, loss_cls_1: 0.8556, loss_box_1: 1.5684, loss_cns_1: 0.6515, loss_yns_1: 0.1534, loss_cls_2: 0.8741, loss_box_2: 1.5306, loss_cns_2: 0.6545, loss_yns_2: 0.1530, loss_cls_3: 0.8666, loss_box_3: 1.5213, loss_cns_3: 0.6537, loss_yns_3: 0.1535, loss_cls_4: 0.8740, loss_box_4: 1.5252, loss_cns_4: 0.6547, loss_yns_4: 0.1537, loss_cls_5: 0.8737, loss_box_5: 1.5458, loss_cns_5: 0.6549, loss_yns_5: 0.1547, loss_cls_dn_0: 0.1709, loss_box_dn_0: 0.7383, loss_cls_dn_1: 0.1241, loss_box_dn_1: 0.7418, loss_cls_dn_2: 0.1254, loss_box_dn_2: 0.7310, loss_cls_dn_3: 0.1263, loss_box_dn_3: 0.7318, loss_cls_dn_4: 0.1289, loss_box_dn_4: 0.7378, loss_cls_dn_5: 0.1330, loss_box_dn_5: 0.7473, loss_dense_depth: 0.6946, loss: 25.2220, grad_norm: 34.6702 -2025-11-17 14:26:14,432 - mmdet - INFO - Iter [246/17500] lr: 1.979e-04, eta: 9:35:47, time: 1.496, data_time: 0.078, memory: 49163, loss_cls_0: 0.7640, loss_box_0: 1.6853, loss_cns_0: 0.6220, loss_yns_0: 0.1550, loss_cls_1: 0.8590, loss_box_1: 1.5626, loss_cns_1: 0.6573, loss_yns_1: 0.1565, loss_cls_2: 0.8641, loss_box_2: 1.5415, loss_cns_2: 0.6574, loss_yns_2: 0.1549, loss_cls_3: 0.8540, loss_box_3: 1.5304, loss_cns_3: 0.6575, loss_yns_3: 0.1542, loss_cls_4: 0.8653, loss_box_4: 1.5223, loss_cns_4: 0.6581, loss_yns_4: 0.1548, loss_cls_5: 0.8595, loss_box_5: 1.5434, loss_cns_5: 0.6568, loss_yns_5: 0.1548, loss_cls_dn_0: 0.1670, loss_box_dn_0: 0.7368, loss_cls_dn_1: 0.1214, loss_box_dn_1: 0.7181, loss_cls_dn_2: 0.1209, loss_box_dn_2: 0.7086, loss_cls_dn_3: 0.1216, loss_box_dn_3: 0.7045, loss_cls_dn_4: 0.1245, loss_box_dn_4: 0.7063, loss_cls_dn_5: 0.1281, loss_box_dn_5: 0.7134, loss_dense_depth: 0.7132, loss: 25.0750, grad_norm: 25.5331 -2025-11-17 14:26:15,954 - mmdet - INFO - Iter [247/17500] lr: 1.983e-04, eta: 9:35:12, time: 1.523, data_time: 0.083, memory: 49163, loss_cls_0: 0.8043, loss_box_0: 1.6915, loss_cns_0: 0.6192, loss_yns_0: 0.1568, loss_cls_1: 0.8689, loss_box_1: 1.5495, loss_cns_1: 0.6619, loss_yns_1: 0.1585, loss_cls_2: 0.8833, loss_box_2: 1.5345, loss_cns_2: 0.6623, loss_yns_2: 0.1571, loss_cls_3: 0.8775, loss_box_3: 1.5179, loss_cns_3: 0.6619, loss_yns_3: 0.1562, loss_cls_4: 0.8818, loss_box_4: 1.5245, loss_cns_4: 0.6615, loss_yns_4: 0.1571, loss_cls_5: 0.8912, loss_box_5: 1.5286, loss_cns_5: 0.6620, loss_yns_5: 0.1585, loss_cls_dn_0: 0.1669, loss_box_dn_0: 0.7461, loss_cls_dn_1: 0.1256, loss_box_dn_1: 0.7210, loss_cls_dn_2: 0.1234, loss_box_dn_2: 0.7141, loss_cls_dn_3: 0.1243, loss_box_dn_3: 0.7084, loss_cls_dn_4: 0.1315, loss_box_dn_4: 0.7172, loss_cls_dn_5: 0.1357, loss_box_dn_5: 0.7231, loss_dense_depth: 0.7000, loss: 25.2640, grad_norm: 39.7995 -2025-11-17 14:26:17,464 - mmdet - INFO - Iter [248/17500] lr: 1.987e-04, eta: 9:34:35, time: 1.508, data_time: 0.087, memory: 49163, loss_cls_0: 0.7976, loss_box_0: 1.6662, loss_cns_0: 0.6167, loss_yns_0: 0.1544, loss_cls_1: 0.8796, loss_box_1: 1.5658, loss_cns_1: 0.6615, loss_yns_1: 0.1567, loss_cls_2: 0.8860, loss_box_2: 1.5354, loss_cns_2: 0.6642, loss_yns_2: 0.1547, loss_cls_3: 0.8752, loss_box_3: 1.5239, loss_cns_3: 0.6664, loss_yns_3: 0.1557, loss_cls_4: 0.8840, loss_box_4: 1.5248, loss_cns_4: 0.6636, loss_yns_4: 0.1553, loss_cls_5: 0.8806, loss_box_5: 1.5327, loss_cns_5: 0.6666, loss_yns_5: 0.1567, loss_cls_dn_0: 0.1682, loss_box_dn_0: 0.7363, loss_cls_dn_1: 0.1251, loss_box_dn_1: 0.7162, loss_cls_dn_2: 0.1228, loss_box_dn_2: 0.7041, loss_cls_dn_3: 0.1232, loss_box_dn_3: 0.7055, loss_cls_dn_4: 0.1295, loss_box_dn_4: 0.7188, loss_cls_dn_5: 0.1296, loss_box_dn_5: 0.7309, loss_dense_depth: 0.7027, loss: 25.2373, grad_norm: 29.3831 -2025-11-17 14:26:19,060 - mmdet - INFO - Iter [249/17500] lr: 1.991e-04, eta: 9:34:06, time: 1.598, data_time: 0.179, memory: 49163, loss_cls_0: 0.7953, loss_box_0: 1.6834, loss_cns_0: 0.6257, loss_yns_0: 0.1571, loss_cls_1: 0.8939, loss_box_1: 1.5937, loss_cns_1: 0.6632, loss_yns_1: 0.1570, loss_cls_2: 0.9002, loss_box_2: 1.5818, loss_cns_2: 0.6609, loss_yns_2: 0.1565, loss_cls_3: 0.8976, loss_box_3: 1.5788, loss_cns_3: 0.6606, loss_yns_3: 0.1559, loss_cls_4: 0.8911, loss_box_4: 1.5790, loss_cns_4: 0.6614, loss_yns_4: 0.1559, loss_cls_5: 0.9066, loss_box_5: 1.5881, loss_cns_5: 0.6621, loss_yns_5: 0.1564, loss_cls_dn_0: 0.1709, loss_box_dn_0: 0.7398, loss_cls_dn_1: 0.1184, loss_box_dn_1: 0.7200, loss_cls_dn_2: 0.1170, loss_box_dn_2: 0.7062, loss_cls_dn_3: 0.1180, loss_box_dn_3: 0.7144, loss_cls_dn_4: 0.1212, loss_box_dn_4: 0.7273, loss_cls_dn_5: 0.1240, loss_box_dn_5: 0.7437, loss_dense_depth: 0.7089, loss: 25.5922, grad_norm: 39.9065 -2025-11-17 14:26:20,558 - mmdet - INFO - Iter [250/17500] lr: 1.995e-04, eta: 9:33:29, time: 1.498, data_time: 0.081, memory: 49163, loss_cls_0: 0.7555, loss_box_0: 1.6781, loss_cns_0: 0.6249, loss_yns_0: 0.1561, loss_cls_1: 0.8660, loss_box_1: 1.5773, loss_cns_1: 0.6638, loss_yns_1: 0.1598, loss_cls_2: 0.8628, loss_box_2: 1.5809, loss_cns_2: 0.6620, loss_yns_2: 0.1566, loss_cls_3: 0.8515, loss_box_3: 1.5735, loss_cns_3: 0.6618, loss_yns_3: 0.1567, loss_cls_4: 0.8586, loss_box_4: 1.5714, loss_cns_4: 0.6575, loss_yns_4: 0.1558, loss_cls_5: 0.8578, loss_box_5: 1.5907, loss_cns_5: 0.6587, loss_yns_5: 0.1567, loss_cls_dn_0: 0.1652, loss_box_dn_0: 0.7329, loss_cls_dn_1: 0.1165, loss_box_dn_1: 0.7240, loss_cls_dn_2: 0.1181, loss_box_dn_2: 0.7206, loss_cls_dn_3: 0.1187, loss_box_dn_3: 0.7373, loss_cls_dn_4: 0.1237, loss_box_dn_4: 0.7561, loss_cls_dn_5: 0.1259, loss_box_dn_5: 0.7782, loss_dense_depth: 0.7016, loss: 25.4132, grad_norm: 44.0130 -2025-11-17 14:26:22,128 - mmdet - INFO - Iter [251/17500] lr: 1.999e-04, eta: 9:32:58, time: 1.570, data_time: 0.079, memory: 49163, loss_cls_0: 0.7780, loss_box_0: 1.6672, loss_cns_0: 0.6231, loss_yns_0: 0.1518, loss_cls_1: 0.8778, loss_box_1: 1.5914, loss_cns_1: 0.6580, loss_yns_1: 0.1549, loss_cls_2: 0.8956, loss_box_2: 1.5640, loss_cns_2: 0.6578, loss_yns_2: 0.1537, loss_cls_3: 0.8868, loss_box_3: 1.5433, loss_cns_3: 0.6637, loss_yns_3: 0.1518, loss_cls_4: 0.8843, loss_box_4: 1.5368, loss_cns_4: 0.6546, loss_yns_4: 0.1502, loss_cls_5: 0.8613, loss_box_5: 1.5733, loss_cns_5: 0.6568, loss_yns_5: 0.1526, loss_cls_dn_0: 0.1724, loss_box_dn_0: 0.7418, loss_cls_dn_1: 0.1213, loss_box_dn_1: 0.7480, loss_cls_dn_2: 0.1239, loss_box_dn_2: 0.7508, loss_cls_dn_3: 0.1246, loss_box_dn_3: 0.7657, loss_cls_dn_4: 0.1241, loss_box_dn_4: 0.7824, loss_cls_dn_5: 0.1271, loss_box_dn_5: 0.8018, loss_dense_depth: 0.7331, loss: 25.6058, grad_norm: 47.3185 -2025-11-17 14:26:23,627 - mmdet - INFO - Iter [252/17500] lr: 2.003e-04, eta: 9:32:22, time: 1.498, data_time: 0.078, memory: 49163, loss_cls_0: 0.7687, loss_box_0: 1.6765, loss_cns_0: 0.6201, loss_yns_0: 0.1525, loss_cls_1: 0.8615, loss_box_1: 1.6193, loss_cns_1: 0.6518, loss_yns_1: 0.1550, loss_cls_2: 0.8730, loss_box_2: 1.5743, loss_cns_2: 0.6538, loss_yns_2: 0.1511, loss_cls_3: 0.8632, loss_box_3: 1.5603, loss_cns_3: 0.6640, loss_yns_3: 0.1519, loss_cls_4: 0.8463, loss_box_4: 1.5609, loss_cns_4: 0.6549, loss_yns_4: 0.1520, loss_cls_5: 0.8481, loss_box_5: 1.5821, loss_cns_5: 0.6579, loss_yns_5: 0.1543, loss_cls_dn_0: 0.1687, loss_box_dn_0: 0.7321, loss_cls_dn_1: 0.1164, loss_box_dn_1: 0.7141, loss_cls_dn_2: 0.1171, loss_box_dn_2: 0.7067, loss_cls_dn_3: 0.1165, loss_box_dn_3: 0.7121, loss_cls_dn_4: 0.1156, loss_box_dn_4: 0.7232, loss_cls_dn_5: 0.1196, loss_box_dn_5: 0.7332, loss_dense_depth: 0.7039, loss: 25.2328, grad_norm: 31.4549 -2025-11-17 14:26:25,127 - mmdet - INFO - Iter [253/17500] lr: 2.007e-04, eta: 9:31:47, time: 1.501, data_time: 0.075, memory: 49163, loss_cls_0: 0.7767, loss_box_0: 1.6810, loss_cns_0: 0.6272, loss_yns_0: 0.1496, loss_cls_1: 0.8657, loss_box_1: 1.5726, loss_cns_1: 0.6575, loss_yns_1: 0.1501, loss_cls_2: 0.8663, loss_box_2: 1.5657, loss_cns_2: 0.6595, loss_yns_2: 0.1502, loss_cls_3: 0.8665, loss_box_3: 1.5421, loss_cns_3: 0.6608, loss_yns_3: 0.1494, loss_cls_4: 0.8747, loss_box_4: 1.5479, loss_cns_4: 0.6596, loss_yns_4: 0.1502, loss_cls_5: 0.8602, loss_box_5: 1.5424, loss_cns_5: 0.6595, loss_yns_5: 0.1489, loss_cls_dn_0: 0.1708, loss_box_dn_0: 0.7341, loss_cls_dn_1: 0.1199, loss_box_dn_1: 0.7138, loss_cls_dn_2: 0.1172, loss_box_dn_2: 0.7039, loss_cls_dn_3: 0.1186, loss_box_dn_3: 0.7016, loss_cls_dn_4: 0.1218, loss_box_dn_4: 0.7079, loss_cls_dn_5: 0.1234, loss_box_dn_5: 0.7109, loss_dense_depth: 0.7426, loss: 25.1708, grad_norm: 38.5302 -2025-11-17 14:26:26,670 - mmdet - INFO - Iter [254/17500] lr: 2.011e-04, eta: 9:31:15, time: 1.543, data_time: 0.074, memory: 49163, loss_cls_0: 0.7546, loss_box_0: 1.6979, loss_cns_0: 0.6227, loss_yns_0: 0.1510, loss_cls_1: 0.8600, loss_box_1: 1.5873, loss_cns_1: 0.6553, loss_yns_1: 0.1501, loss_cls_2: 0.8643, loss_box_2: 1.5756, loss_cns_2: 0.6587, loss_yns_2: 0.1490, loss_cls_3: 0.8610, loss_box_3: 1.5497, loss_cns_3: 0.6612, loss_yns_3: 0.1491, loss_cls_4: 0.8530, loss_box_4: 1.5481, loss_cns_4: 0.6592, loss_yns_4: 0.1505, loss_cls_5: 0.8563, loss_box_5: 1.5351, loss_cns_5: 0.6562, loss_yns_5: 0.1489, loss_cls_dn_0: 0.1646, loss_box_dn_0: 0.7391, loss_cls_dn_1: 0.1183, loss_box_dn_1: 0.7056, loss_cls_dn_2: 0.1170, loss_box_dn_2: 0.6933, loss_cls_dn_3: 0.1190, loss_box_dn_3: 0.6861, loss_cls_dn_4: 0.1205, loss_box_dn_4: 0.6860, loss_cls_dn_5: 0.1268, loss_box_dn_5: 0.6895, loss_dense_depth: 0.7249, loss: 25.0455, grad_norm: 34.5553 -2025-11-17 14:26:28,154 - mmdet - INFO - Iter [255/17500] lr: 2.015e-04, eta: 9:30:39, time: 1.484, data_time: 0.075, memory: 49163, loss_cls_0: 0.7571, loss_box_0: 1.6887, loss_cns_0: 0.6245, loss_yns_0: 0.1508, loss_cls_1: 0.8449, loss_box_1: 1.5294, loss_cns_1: 0.6604, loss_yns_1: 0.1501, loss_cls_2: 0.8503, loss_box_2: 1.5160, loss_cns_2: 0.6621, loss_yns_2: 0.1514, loss_cls_3: 0.8412, loss_box_3: 1.5085, loss_cns_3: 0.6686, loss_yns_3: 0.1513, loss_cls_4: 0.8464, loss_box_4: 1.5043, loss_cns_4: 0.6622, loss_yns_4: 0.1512, loss_cls_5: 0.8792, loss_box_5: 1.4956, loss_cns_5: 0.6583, loss_yns_5: 0.1508, loss_cls_dn_0: 0.1621, loss_box_dn_0: 0.7406, loss_cls_dn_1: 0.1164, loss_box_dn_1: 0.7048, loss_cls_dn_2: 0.1169, loss_box_dn_2: 0.6876, loss_cls_dn_3: 0.1162, loss_box_dn_3: 0.6842, loss_cls_dn_4: 0.1194, loss_box_dn_4: 0.6845, loss_cls_dn_5: 0.1229, loss_box_dn_5: 0.6924, loss_dense_depth: 0.7229, loss: 24.7741, grad_norm: 32.7425 -2025-11-17 14:26:29,643 - mmdet - INFO - Iter [256/17500] lr: 2.019e-04, eta: 9:30:03, time: 1.487, data_time: 0.079, memory: 49163, loss_cls_0: 0.7698, loss_box_0: 1.6966, loss_cns_0: 0.6252, loss_yns_0: 0.1521, loss_cls_1: 0.8470, loss_box_1: 1.5579, loss_cns_1: 0.6566, loss_yns_1: 0.1508, loss_cls_2: 0.8659, loss_box_2: 1.5094, loss_cns_2: 0.6615, loss_yns_2: 0.1515, loss_cls_3: 0.8749, loss_box_3: 1.5077, loss_cns_3: 0.6689, loss_yns_3: 0.1518, loss_cls_4: 0.8746, loss_box_4: 1.5025, loss_cns_4: 0.6626, loss_yns_4: 0.1523, loss_cls_5: 0.8713, loss_box_5: 1.5043, loss_cns_5: 0.6608, loss_yns_5: 0.1525, loss_cls_dn_0: 0.1606, loss_box_dn_0: 0.7315, loss_cls_dn_1: 0.1110, loss_box_dn_1: 0.6725, loss_cls_dn_2: 0.1129, loss_box_dn_2: 0.6580, loss_cls_dn_3: 0.1108, loss_box_dn_3: 0.6653, loss_cls_dn_4: 0.1145, loss_box_dn_4: 0.6754, loss_cls_dn_5: 0.1161, loss_box_dn_5: 0.6880, loss_dense_depth: 0.7390, loss: 24.7843, grad_norm: 33.9327 -2025-11-17 14:26:31,170 - mmdet - INFO - Iter [257/17500] lr: 2.023e-04, eta: 9:29:30, time: 1.528, data_time: 0.074, memory: 49163, loss_cls_0: 0.7858, loss_box_0: 1.7297, loss_cns_0: 0.6228, loss_yns_0: 0.1518, loss_cls_1: 0.8572, loss_box_1: 1.6518, loss_cns_1: 0.6521, loss_yns_1: 0.1500, loss_cls_2: 0.8755, loss_box_2: 1.5900, loss_cns_2: 0.6561, loss_yns_2: 0.1517, loss_cls_3: 0.8771, loss_box_3: 1.5728, loss_cns_3: 0.6607, loss_yns_3: 0.1503, loss_cls_4: 0.8742, loss_box_4: 1.5783, loss_cns_4: 0.6583, loss_yns_4: 0.1522, loss_cls_5: 0.8875, loss_box_5: 1.5926, loss_cns_5: 0.6578, loss_yns_5: 0.1511, loss_cls_dn_0: 0.1662, loss_box_dn_0: 0.7341, loss_cls_dn_1: 0.1130, loss_box_dn_1: 0.6859, loss_cls_dn_2: 0.1151, loss_box_dn_2: 0.6753, loss_cls_dn_3: 0.1137, loss_box_dn_3: 0.6850, loss_cls_dn_4: 0.1150, loss_box_dn_4: 0.7012, loss_cls_dn_5: 0.1172, loss_box_dn_5: 0.7158, loss_dense_depth: 0.7363, loss: 25.3612, grad_norm: 37.3859 -2025-11-17 14:26:32,664 - mmdet - INFO - Iter [258/17500] lr: 2.027e-04, eta: 9:28:56, time: 1.494, data_time: 0.079, memory: 49163, loss_cls_0: 0.7743, loss_box_0: 1.7096, loss_cns_0: 0.6223, loss_yns_0: 0.1495, loss_cls_1: 0.8545, loss_box_1: 1.6287, loss_cns_1: 0.6559, loss_yns_1: 0.1496, loss_cls_2: 0.8661, loss_box_2: 1.5817, loss_cns_2: 0.6584, loss_yns_2: 0.1499, loss_cls_3: 0.8670, loss_box_3: 1.5675, loss_cns_3: 0.6599, loss_yns_3: 0.1496, loss_cls_4: 0.8696, loss_box_4: 1.5704, loss_cns_4: 0.6607, loss_yns_4: 0.1495, loss_cls_5: 0.8883, loss_box_5: 1.5823, loss_cns_5: 0.6591, loss_yns_5: 0.1500, loss_cls_dn_0: 0.1622, loss_box_dn_0: 0.7273, loss_cls_dn_1: 0.1162, loss_box_dn_1: 0.6894, loss_cls_dn_2: 0.1155, loss_box_dn_2: 0.6830, loss_cls_dn_3: 0.1157, loss_box_dn_3: 0.6888, loss_cls_dn_4: 0.1196, loss_box_dn_4: 0.7053, loss_cls_dn_5: 0.1191, loss_box_dn_5: 0.7188, loss_dense_depth: 0.7293, loss: 25.2646, grad_norm: 32.4308 -2025-11-17 14:26:34,164 - mmdet - INFO - Iter [259/17500] lr: 2.031e-04, eta: 9:28:22, time: 1.499, data_time: 0.077, memory: 49163, loss_cls_0: 0.8086, loss_box_0: 1.7221, loss_cns_0: 0.6186, loss_yns_0: 0.1494, loss_cls_1: 0.8796, loss_box_1: 1.6681, loss_cns_1: 0.6514, loss_yns_1: 0.1492, loss_cls_2: 0.9063, loss_box_2: 1.6429, loss_cns_2: 0.6540, loss_yns_2: 0.1477, loss_cls_3: 0.9114, loss_box_3: 1.6180, loss_cns_3: 0.6523, loss_yns_3: 0.1467, loss_cls_4: 0.9141, loss_box_4: 1.6204, loss_cns_4: 0.6546, loss_yns_4: 0.1483, loss_cls_5: 0.9090, loss_box_5: 1.6267, loss_cns_5: 0.6569, loss_yns_5: 0.1498, loss_cls_dn_0: 0.1648, loss_box_dn_0: 0.7402, loss_cls_dn_1: 0.1197, loss_box_dn_1: 0.7082, loss_cls_dn_2: 0.1195, loss_box_dn_2: 0.6974, loss_cls_dn_3: 0.1219, loss_box_dn_3: 0.6988, loss_cls_dn_4: 0.1281, loss_box_dn_4: 0.7108, loss_cls_dn_5: 0.1260, loss_box_dn_5: 0.7191, loss_dense_depth: 0.7798, loss: 25.8406, grad_norm: 37.3332 -2025-11-17 14:26:41,019 - mmdet - INFO - Iter [260/17500] lr: 2.035e-04, eta: 9:33:43, time: 6.857, data_time: 0.072, memory: 49163, loss_cls_0: 0.7926, loss_box_0: 1.7080, loss_cns_0: 0.6219, loss_yns_0: 0.1501, loss_cls_1: 0.8725, loss_box_1: 1.6599, loss_cns_1: 0.6531, loss_yns_1: 0.1523, loss_cls_2: 0.9048, loss_box_2: 1.6193, loss_cns_2: 0.6562, loss_yns_2: 0.1505, loss_cls_3: 0.9051, loss_box_3: 1.5944, loss_cns_3: 0.6556, loss_yns_3: 0.1495, loss_cls_4: 0.9098, loss_box_4: 1.5899, loss_cns_4: 0.6564, loss_yns_4: 0.1500, loss_cls_5: 0.9053, loss_box_5: 1.5947, loss_cns_5: 0.6609, loss_yns_5: 0.1495, loss_cls_dn_0: 0.1637, loss_box_dn_0: 0.7354, loss_cls_dn_1: 0.1234, loss_box_dn_1: 0.7204, loss_cls_dn_2: 0.1221, loss_box_dn_2: 0.7027, loss_cls_dn_3: 0.1220, loss_box_dn_3: 0.6983, loss_cls_dn_4: 0.1255, loss_box_dn_4: 0.7040, loss_cls_dn_5: 0.1252, loss_box_dn_5: 0.7098, loss_dense_depth: 0.7569, loss: 25.6718, grad_norm: 30.2957 -2025-11-17 14:26:42,519 - mmdet - INFO - Iter [261/17500] lr: 2.039e-04, eta: 9:33:09, time: 1.500, data_time: 0.072, memory: 49163, loss_cls_0: 0.7971, loss_box_0: 1.7310, loss_cns_0: 0.6187, loss_yns_0: 0.1485, loss_cls_1: 0.8848, loss_box_1: 1.6203, loss_cns_1: 0.6560, loss_yns_1: 0.1486, loss_cls_2: 0.9021, loss_box_2: 1.6061, loss_cns_2: 0.6587, loss_yns_2: 0.1494, loss_cls_3: 0.9024, loss_box_3: 1.5729, loss_cns_3: 0.6593, loss_yns_3: 0.1478, loss_cls_4: 0.9055, loss_box_4: 1.5737, loss_cns_4: 0.6596, loss_yns_4: 0.1485, loss_cls_5: 0.9067, loss_box_5: 1.5781, loss_cns_5: 0.6606, loss_yns_5: 0.1486, loss_cls_dn_0: 0.1666, loss_box_dn_0: 0.7429, loss_cls_dn_1: 0.1199, loss_box_dn_1: 0.6998, loss_cls_dn_2: 0.1189, loss_box_dn_2: 0.6852, loss_cls_dn_3: 0.1196, loss_box_dn_3: 0.6804, loss_cls_dn_4: 0.1219, loss_box_dn_4: 0.6788, loss_cls_dn_5: 0.1240, loss_box_dn_5: 0.6825, loss_dense_depth: 0.7796, loss: 25.5051, grad_norm: 31.4190 -2025-11-17 14:26:44,087 - mmdet - INFO - Iter [262/17500] lr: 2.043e-04, eta: 9:32:39, time: 1.568, data_time: 0.074, memory: 49163, loss_cls_0: 0.7906, loss_box_0: 1.7222, loss_cns_0: 0.6207, loss_yns_0: 0.1456, loss_cls_1: 0.8815, loss_box_1: 1.6393, loss_cns_1: 0.6554, loss_yns_1: 0.1462, loss_cls_2: 0.9076, loss_box_2: 1.6258, loss_cns_2: 0.6582, loss_yns_2: 0.1469, loss_cls_3: 0.9046, loss_box_3: 1.6034, loss_cns_3: 0.6598, loss_yns_3: 0.1455, loss_cls_4: 0.9140, loss_box_4: 1.5985, loss_cns_4: 0.6577, loss_yns_4: 0.1461, loss_cls_5: 0.9063, loss_box_5: 1.5954, loss_cns_5: 0.6570, loss_yns_5: 0.1457, loss_cls_dn_0: 0.1603, loss_box_dn_0: 0.7498, loss_cls_dn_1: 0.1185, loss_box_dn_1: 0.6932, loss_cls_dn_2: 0.1188, loss_box_dn_2: 0.6801, loss_cls_dn_3: 0.1192, loss_box_dn_3: 0.6758, loss_cls_dn_4: 0.1220, loss_box_dn_4: 0.6733, loss_cls_dn_5: 0.1250, loss_box_dn_5: 0.6765, loss_dense_depth: 0.7433, loss: 25.5296, grad_norm: 37.1912 -2025-11-17 14:26:45,584 - mmdet - INFO - Iter [263/17500] lr: 2.047e-04, eta: 9:32:04, time: 1.495, data_time: 0.077, memory: 49163, loss_cls_0: 0.7800, loss_box_0: 1.7146, loss_cns_0: 0.6132, loss_yns_0: 0.1449, loss_cls_1: 0.8654, loss_box_1: 1.6384, loss_cns_1: 0.6501, loss_yns_1: 0.1451, loss_cls_2: 0.8804, loss_box_2: 1.6135, loss_cns_2: 0.6554, loss_yns_2: 0.1451, loss_cls_3: 0.8848, loss_box_3: 1.5846, loss_cns_3: 0.6558, loss_yns_3: 0.1439, loss_cls_4: 0.8891, loss_box_4: 1.5832, loss_cns_4: 0.6548, loss_yns_4: 0.1434, loss_cls_5: 0.8927, loss_box_5: 1.5808, loss_cns_5: 0.6548, loss_yns_5: 0.1433, loss_cls_dn_0: 0.1615, loss_box_dn_0: 0.7587, loss_cls_dn_1: 0.1190, loss_box_dn_1: 0.6943, loss_cls_dn_2: 0.1173, loss_box_dn_2: 0.6774, loss_cls_dn_3: 0.1171, loss_box_dn_3: 0.6737, loss_cls_dn_4: 0.1192, loss_box_dn_4: 0.6746, loss_cls_dn_5: 0.1199, loss_box_dn_5: 0.6835, loss_dense_depth: 0.7539, loss: 25.3274, grad_norm: 27.4706 -2025-11-17 14:26:47,098 - mmdet - INFO - Iter [264/17500] lr: 2.051e-04, eta: 9:31:31, time: 1.515, data_time: 0.080, memory: 49163, loss_cls_0: 0.7740, loss_box_0: 1.7099, loss_cns_0: 0.6171, loss_yns_0: 0.1491, loss_cls_1: 0.8609, loss_box_1: 1.6335, loss_cns_1: 0.6521, loss_yns_1: 0.1467, loss_cls_2: 0.8740, loss_box_2: 1.6021, loss_cns_2: 0.6556, loss_yns_2: 0.1458, loss_cls_3: 0.8768, loss_box_3: 1.5869, loss_cns_3: 0.6574, loss_yns_3: 0.1446, loss_cls_4: 0.8830, loss_box_4: 1.5914, loss_cns_4: 0.6568, loss_yns_4: 0.1446, loss_cls_5: 0.8928, loss_box_5: 1.5912, loss_cns_5: 0.6570, loss_yns_5: 0.1447, loss_cls_dn_0: 0.1609, loss_box_dn_0: 0.7425, loss_cls_dn_1: 0.1208, loss_box_dn_1: 0.6969, loss_cls_dn_2: 0.1177, loss_box_dn_2: 0.6822, loss_cls_dn_3: 0.1197, loss_box_dn_3: 0.6903, loss_cls_dn_4: 0.1245, loss_box_dn_4: 0.7009, loss_cls_dn_5: 0.1241, loss_box_dn_5: 0.7155, loss_dense_depth: 0.7666, loss: 25.4103, grad_norm: 35.3441 -2025-11-17 14:26:48,590 - mmdet - INFO - Iter [265/17500] lr: 2.055e-04, eta: 9:30:56, time: 1.491, data_time: 0.077, memory: 49163, loss_cls_0: 0.7991, loss_box_0: 1.7454, loss_cns_0: 0.6192, loss_yns_0: 0.1475, loss_cls_1: 0.8649, loss_box_1: 1.6391, loss_cns_1: 0.6534, loss_yns_1: 0.1460, loss_cls_2: 0.8829, loss_box_2: 1.6088, loss_cns_2: 0.6585, loss_yns_2: 0.1463, loss_cls_3: 0.8871, loss_box_3: 1.6115, loss_cns_3: 0.6671, loss_yns_3: 0.1478, loss_cls_4: 0.8865, loss_box_4: 1.6022, loss_cns_4: 0.6604, loss_yns_4: 0.1493, loss_cls_5: 0.8910, loss_box_5: 1.6058, loss_cns_5: 0.6593, loss_yns_5: 0.1484, loss_cls_dn_0: 0.1661, loss_box_dn_0: 0.7491, loss_cls_dn_1: 0.1174, loss_box_dn_1: 0.7097, loss_cls_dn_2: 0.1161, loss_box_dn_2: 0.7012, loss_cls_dn_3: 0.1241, loss_box_dn_3: 0.7136, loss_cls_dn_4: 0.1278, loss_box_dn_4: 0.7201, loss_cls_dn_5: 0.1278, loss_box_dn_5: 0.7316, loss_dense_depth: 0.7647, loss: 25.6969, grad_norm: 42.1134 -2025-11-17 14:26:50,089 - mmdet - INFO - Iter [266/17500] lr: 2.059e-04, eta: 9:30:23, time: 1.500, data_time: 0.078, memory: 49163, loss_cls_0: 0.7675, loss_box_0: 1.7133, loss_cns_0: 0.6210, loss_yns_0: 0.1439, loss_cls_1: 0.8432, loss_box_1: 1.6069, loss_cns_1: 0.6553, loss_yns_1: 0.1449, loss_cls_2: 0.8523, loss_box_2: 1.5896, loss_cns_2: 0.6600, loss_yns_2: 0.1446, loss_cls_3: 0.8590, loss_box_3: 1.5868, loss_cns_3: 0.6681, loss_yns_3: 0.1465, loss_cls_4: 0.8552, loss_box_4: 1.5753, loss_cns_4: 0.6611, loss_yns_4: 0.1476, loss_cls_5: 0.8596, loss_box_5: 1.5921, loss_cns_5: 0.6606, loss_yns_5: 0.1469, loss_cls_dn_0: 0.1590, loss_box_dn_0: 0.7553, loss_cls_dn_1: 0.1171, loss_box_dn_1: 0.7245, loss_cls_dn_2: 0.1162, loss_box_dn_2: 0.7172, loss_cls_dn_3: 0.1229, loss_box_dn_3: 0.7258, loss_cls_dn_4: 0.1241, loss_box_dn_4: 0.7282, loss_cls_dn_5: 0.1251, loss_box_dn_5: 0.7414, loss_dense_depth: 0.7826, loss: 25.4408, grad_norm: 35.4792 -2025-11-17 14:26:51,589 - mmdet - INFO - Iter [267/17500] lr: 2.063e-04, eta: 9:29:50, time: 1.501, data_time: 0.083, memory: 49163, loss_cls_0: 0.7594, loss_box_0: 1.6842, loss_cns_0: 0.6233, loss_yns_0: 0.1439, loss_cls_1: 0.8371, loss_box_1: 1.5804, loss_cns_1: 0.6571, loss_yns_1: 0.1431, loss_cls_2: 0.8512, loss_box_2: 1.5588, loss_cns_2: 0.6587, loss_yns_2: 0.1426, loss_cls_3: 0.8489, loss_box_3: 1.5525, loss_cns_3: 0.6624, loss_yns_3: 0.1441, loss_cls_4: 0.8468, loss_box_4: 1.5562, loss_cns_4: 0.6596, loss_yns_4: 0.1435, loss_cls_5: 0.8476, loss_box_5: 1.5747, loss_cns_5: 0.6607, loss_yns_5: 0.1451, loss_cls_dn_0: 0.1565, loss_box_dn_0: 0.7466, loss_cls_dn_1: 0.1139, loss_box_dn_1: 0.7244, loss_cls_dn_2: 0.1126, loss_box_dn_2: 0.7114, loss_cls_dn_3: 0.1158, loss_box_dn_3: 0.7132, loss_cls_dn_4: 0.1175, loss_box_dn_4: 0.7156, loss_cls_dn_5: 0.1181, loss_box_dn_5: 0.7269, loss_dense_depth: 0.7476, loss: 25.1019, grad_norm: 36.0624 -2025-11-17 14:26:53,085 - mmdet - INFO - Iter [268/17500] lr: 2.067e-04, eta: 9:29:16, time: 1.494, data_time: 0.085, memory: 49163, loss_cls_0: 0.7601, loss_box_0: 1.6790, loss_cns_0: 0.6298, loss_yns_0: 0.1476, loss_cls_1: 0.8430, loss_box_1: 1.5742, loss_cns_1: 0.6645, loss_yns_1: 0.1446, loss_cls_2: 0.8545, loss_box_2: 1.5486, loss_cns_2: 0.6651, loss_yns_2: 0.1433, loss_cls_3: 0.8493, loss_box_3: 1.5256, loss_cns_3: 0.6640, loss_yns_3: 0.1437, loss_cls_4: 0.8483, loss_box_4: 1.5261, loss_cns_4: 0.6638, loss_yns_4: 0.1430, loss_cls_5: 0.8468, loss_box_5: 1.5356, loss_cns_5: 0.6654, loss_yns_5: 0.1446, loss_cls_dn_0: 0.1590, loss_box_dn_0: 0.7417, loss_cls_dn_1: 0.1184, loss_box_dn_1: 0.7255, loss_cls_dn_2: 0.1163, loss_box_dn_2: 0.7133, loss_cls_dn_3: 0.1179, loss_box_dn_3: 0.7077, loss_cls_dn_4: 0.1188, loss_box_dn_4: 0.7076, loss_cls_dn_5: 0.1213, loss_box_dn_5: 0.7117, loss_dense_depth: 0.7563, loss: 25.0261, grad_norm: 29.3242 -2025-11-17 14:26:54,656 - mmdet - INFO - Iter [269/17500] lr: 2.071e-04, eta: 9:28:48, time: 1.572, data_time: 0.165, memory: 49163, loss_cls_0: 0.7552, loss_box_0: 1.6859, loss_cns_0: 0.6293, loss_yns_0: 0.1490, loss_cls_1: 0.8369, loss_box_1: 1.5828, loss_cns_1: 0.6612, loss_yns_1: 0.1474, loss_cls_2: 0.8446, loss_box_2: 1.5710, loss_cns_2: 0.6614, loss_yns_2: 0.1474, loss_cls_3: 0.8530, loss_box_3: 1.5412, loss_cns_3: 0.6633, loss_yns_3: 0.1462, loss_cls_4: 0.8584, loss_box_4: 1.5368, loss_cns_4: 0.6615, loss_yns_4: 0.1446, loss_cls_5: 0.8558, loss_box_5: 1.5326, loss_cns_5: 0.6609, loss_yns_5: 0.1466, loss_cls_dn_0: 0.1540, loss_box_dn_0: 0.7374, loss_cls_dn_1: 0.1147, loss_box_dn_1: 0.7180, loss_cls_dn_2: 0.1123, loss_box_dn_2: 0.7093, loss_cls_dn_3: 0.1157, loss_box_dn_3: 0.6973, loss_cls_dn_4: 0.1173, loss_box_dn_4: 0.6954, loss_cls_dn_5: 0.1216, loss_box_dn_5: 0.6940, loss_dense_depth: 0.7594, loss: 25.0191, grad_norm: 36.1152 -2025-11-17 14:26:56,153 - mmdet - INFO - Iter [270/17500] lr: 2.075e-04, eta: 9:28:15, time: 1.495, data_time: 0.076, memory: 49163, loss_cls_0: 0.7979, loss_box_0: 1.7157, loss_cns_0: 0.6206, loss_yns_0: 0.1498, loss_cls_1: 0.8719, loss_box_1: 1.6122, loss_cns_1: 0.6550, loss_yns_1: 0.1480, loss_cls_2: 0.8910, loss_box_2: 1.5964, loss_cns_2: 0.6569, loss_yns_2: 0.1468, loss_cls_3: 0.8892, loss_box_3: 1.5708, loss_cns_3: 0.6582, loss_yns_3: 0.1473, loss_cls_4: 0.8838, loss_box_4: 1.5644, loss_cns_4: 0.6540, loss_yns_4: 0.1468, loss_cls_5: 0.8866, loss_box_5: 1.5678, loss_cns_5: 0.6535, loss_yns_5: 0.1478, loss_cls_dn_0: 0.1599, loss_box_dn_0: 0.7390, loss_cls_dn_1: 0.1136, loss_box_dn_1: 0.6870, loss_cls_dn_2: 0.1139, loss_box_dn_2: 0.6782, loss_cls_dn_3: 0.1174, loss_box_dn_3: 0.6735, loss_cls_dn_4: 0.1247, loss_box_dn_4: 0.6757, loss_cls_dn_5: 0.1328, loss_box_dn_5: 0.6803, loss_dense_depth: 0.7786, loss: 25.3069, grad_norm: 36.3115 -2025-11-17 14:26:57,653 - mmdet - INFO - Iter [271/17500] lr: 2.079e-04, eta: 9:27:42, time: 1.499, data_time: 0.076, memory: 49163, loss_cls_0: 0.7404, loss_box_0: 1.6737, loss_cns_0: 0.6282, loss_yns_0: 0.1519, loss_cls_1: 0.8554, loss_box_1: 1.5747, loss_cns_1: 0.6560, loss_yns_1: 0.1472, loss_cls_2: 0.8558, loss_box_2: 1.5484, loss_cns_2: 0.6579, loss_yns_2: 0.1453, loss_cls_3: 0.8521, loss_box_3: 1.5495, loss_cns_3: 0.6561, loss_yns_3: 0.1461, loss_cls_4: 0.8473, loss_box_4: 1.5385, loss_cns_4: 0.6563, loss_yns_4: 0.1460, loss_cls_5: 0.8531, loss_box_5: 1.5391, loss_cns_5: 0.6555, loss_yns_5: 0.1482, loss_cls_dn_0: 0.1488, loss_box_dn_0: 0.7335, loss_cls_dn_1: 0.1134, loss_box_dn_1: 0.6858, loss_cls_dn_2: 0.1133, loss_box_dn_2: 0.6724, loss_cls_dn_3: 0.1145, loss_box_dn_3: 0.6828, loss_cls_dn_4: 0.1198, loss_box_dn_4: 0.6854, loss_cls_dn_5: 0.1231, loss_box_dn_5: 0.6945, loss_dense_depth: 0.7240, loss: 24.8342, grad_norm: 35.5734 -2025-11-17 14:26:59,137 - mmdet - INFO - Iter [272/17500] lr: 2.083e-04, eta: 9:27:09, time: 1.486, data_time: 0.076, memory: 49163, loss_cls_0: 0.7487, loss_box_0: 1.6889, loss_cns_0: 0.6249, loss_yns_0: 0.1540, loss_cls_1: 0.8476, loss_box_1: 1.5625, loss_cns_1: 0.6545, loss_yns_1: 0.1496, loss_cls_2: 0.8517, loss_box_2: 1.5224, loss_cns_2: 0.6580, loss_yns_2: 0.1475, loss_cls_3: 0.8529, loss_box_3: 1.5532, loss_cns_3: 0.6562, loss_yns_3: 0.1495, loss_cls_4: 0.8453, loss_box_4: 1.5490, loss_cns_4: 0.6582, loss_yns_4: 0.1485, loss_cls_5: 0.8518, loss_box_5: 1.5451, loss_cns_5: 0.6542, loss_yns_5: 0.1497, loss_cls_dn_0: 0.1472, loss_box_dn_0: 0.7342, loss_cls_dn_1: 0.1104, loss_box_dn_1: 0.7054, loss_cls_dn_2: 0.1107, loss_box_dn_2: 0.6956, loss_cls_dn_3: 0.1108, loss_box_dn_3: 0.7207, loss_cls_dn_4: 0.1120, loss_box_dn_4: 0.7310, loss_cls_dn_5: 0.1142, loss_box_dn_5: 0.7446, loss_dense_depth: 0.7164, loss: 24.9768, grad_norm: 43.1626 -2025-11-17 14:27:00,642 - mmdet - INFO - Iter [273/17500] lr: 2.087e-04, eta: 9:26:38, time: 1.505, data_time: 0.075, memory: 49163, loss_cls_0: 0.7639, loss_box_0: 1.6752, loss_cns_0: 0.6239, loss_yns_0: 0.1489, loss_cls_1: 0.8421, loss_box_1: 1.5608, loss_cns_1: 0.6552, loss_yns_1: 0.1466, loss_cls_2: 0.8507, loss_box_2: 1.5282, loss_cns_2: 0.6570, loss_yns_2: 0.1461, loss_cls_3: 0.8573, loss_box_3: 1.5427, loss_cns_3: 0.6583, loss_yns_3: 0.1471, loss_cls_4: 0.8535, loss_box_4: 1.5396, loss_cns_4: 0.6594, loss_yns_4: 0.1468, loss_cls_5: 0.8637, loss_box_5: 1.5365, loss_cns_5: 0.6556, loss_yns_5: 0.1480, loss_cls_dn_0: 0.1558, loss_box_dn_0: 0.7387, loss_cls_dn_1: 0.1106, loss_box_dn_1: 0.7162, loss_cls_dn_2: 0.1123, loss_box_dn_2: 0.7077, loss_cls_dn_3: 0.1157, loss_box_dn_3: 0.7251, loss_cls_dn_4: 0.1179, loss_box_dn_4: 0.7340, loss_cls_dn_5: 0.1226, loss_box_dn_5: 0.7467, loss_dense_depth: 0.7300, loss: 25.0403, grad_norm: 45.9547 -2025-11-17 14:27:02,184 - mmdet - INFO - Iter [274/17500] lr: 2.091e-04, eta: 9:26:09, time: 1.541, data_time: 0.078, memory: 49163, loss_cls_0: 0.7543, loss_box_0: 1.6696, loss_cns_0: 0.6276, loss_yns_0: 0.1466, loss_cls_1: 0.8232, loss_box_1: 1.5497, loss_cns_1: 0.6563, loss_yns_1: 0.1456, loss_cls_2: 0.8410, loss_box_2: 1.5142, loss_cns_2: 0.6594, loss_yns_2: 0.1449, loss_cls_3: 0.8454, loss_box_3: 1.5113, loss_cns_3: 0.6633, loss_yns_3: 0.1451, loss_cls_4: 0.8376, loss_box_4: 1.5110, loss_cns_4: 0.6643, loss_yns_4: 0.1454, loss_cls_5: 0.8468, loss_box_5: 1.5079, loss_cns_5: 0.6598, loss_yns_5: 0.1458, loss_cls_dn_0: 0.1579, loss_box_dn_0: 0.7377, loss_cls_dn_1: 0.1090, loss_box_dn_1: 0.6908, loss_cls_dn_2: 0.1087, loss_box_dn_2: 0.6828, loss_cls_dn_3: 0.1115, loss_box_dn_3: 0.6886, loss_cls_dn_4: 0.1154, loss_box_dn_4: 0.6967, loss_cls_dn_5: 0.1182, loss_box_dn_5: 0.7059, loss_dense_depth: 0.7384, loss: 24.6777, grad_norm: 40.9384 -2025-11-17 14:27:03,686 - mmdet - INFO - Iter [275/17500] lr: 2.095e-04, eta: 9:25:37, time: 1.501, data_time: 0.077, memory: 49163, loss_cls_0: 0.7405, loss_box_0: 1.6374, loss_cns_0: 0.6285, loss_yns_0: 0.1462, loss_cls_1: 0.7948, loss_box_1: 1.5337, loss_cns_1: 0.6561, loss_yns_1: 0.1435, loss_cls_2: 0.8176, loss_box_2: 1.4854, loss_cns_2: 0.6598, loss_yns_2: 0.1419, loss_cls_3: 0.8177, loss_box_3: 1.4770, loss_cns_3: 0.6588, loss_yns_3: 0.1425, loss_cls_4: 0.8179, loss_box_4: 1.4771, loss_cns_4: 0.6585, loss_yns_4: 0.1415, loss_cls_5: 0.8267, loss_box_5: 1.4804, loss_cns_5: 0.6597, loss_yns_5: 0.1418, loss_cls_dn_0: 0.1580, loss_box_dn_0: 0.7303, loss_cls_dn_1: 0.1059, loss_box_dn_1: 0.6843, loss_cls_dn_2: 0.1091, loss_box_dn_2: 0.6688, loss_cls_dn_3: 0.1088, loss_box_dn_3: 0.6701, loss_cls_dn_4: 0.1113, loss_box_dn_4: 0.6787, loss_cls_dn_5: 0.1150, loss_box_dn_5: 0.6864, loss_dense_depth: 0.7100, loss: 24.2222, grad_norm: 32.0934 -2025-11-17 14:27:05,185 - mmdet - INFO - Iter [276/17500] lr: 2.099e-04, eta: 9:25:06, time: 1.500, data_time: 0.078, memory: 49163, loss_cls_0: 0.7514, loss_box_0: 1.6456, loss_cns_0: 0.6240, loss_yns_0: 0.1455, loss_cls_1: 0.8241, loss_box_1: 1.5238, loss_cns_1: 0.6599, loss_yns_1: 0.1437, loss_cls_2: 0.8434, loss_box_2: 1.4997, loss_cns_2: 0.6616, loss_yns_2: 0.1413, loss_cls_3: 0.8555, loss_box_3: 1.4858, loss_cns_3: 0.6599, loss_yns_3: 0.1408, loss_cls_4: 0.8564, loss_box_4: 1.4793, loss_cns_4: 0.6614, loss_yns_4: 0.1390, loss_cls_5: 0.8691, loss_box_5: 1.4816, loss_cns_5: 0.6624, loss_yns_5: 0.1406, loss_cls_dn_0: 0.1534, loss_box_dn_0: 0.7302, loss_cls_dn_1: 0.1079, loss_box_dn_1: 0.6706, loss_cls_dn_2: 0.1113, loss_box_dn_2: 0.6573, loss_cls_dn_3: 0.1106, loss_box_dn_3: 0.6521, loss_cls_dn_4: 0.1124, loss_box_dn_4: 0.6537, loss_cls_dn_5: 0.1156, loss_box_dn_5: 0.6550, loss_dense_depth: 0.7279, loss: 24.3540, grad_norm: 44.7045 -2025-11-17 14:27:06,733 - mmdet - INFO - Iter [277/17500] lr: 2.103e-04, eta: 9:24:38, time: 1.549, data_time: 0.082, memory: 49163, loss_cls_0: 0.7477, loss_box_0: 1.6721, loss_cns_0: 0.6259, loss_yns_0: 0.1446, loss_cls_1: 0.8237, loss_box_1: 1.5859, loss_cns_1: 0.6529, loss_yns_1: 0.1444, loss_cls_2: 0.8359, loss_box_2: 1.5598, loss_cns_2: 0.6560, loss_yns_2: 0.1429, loss_cls_3: 0.8518, loss_box_3: 1.5490, loss_cns_3: 0.6565, loss_yns_3: 0.1427, loss_cls_4: 0.8493, loss_box_4: 1.5300, loss_cns_4: 0.6573, loss_yns_4: 0.1433, loss_cls_5: 0.8572, loss_box_5: 1.5292, loss_cns_5: 0.6568, loss_yns_5: 0.1427, loss_cls_dn_0: 0.1483, loss_box_dn_0: 0.7282, loss_cls_dn_1: 0.1066, loss_box_dn_1: 0.6588, loss_cls_dn_2: 0.1068, loss_box_dn_2: 0.6426, loss_cls_dn_3: 0.1075, loss_box_dn_3: 0.6395, loss_cls_dn_4: 0.1107, loss_box_dn_4: 0.6357, loss_cls_dn_5: 0.1132, loss_box_dn_5: 0.6346, loss_dense_depth: 0.7336, loss: 24.5239, grad_norm: 40.7445 -2025-11-17 14:27:08,228 - mmdet - INFO - Iter [278/17500] lr: 2.107e-04, eta: 9:24:06, time: 1.494, data_time: 0.081, memory: 49163, loss_cls_0: 0.7688, loss_box_0: 1.6466, loss_cns_0: 0.6219, loss_yns_0: 0.1462, loss_cls_1: 0.8248, loss_box_1: 1.6092, loss_cns_1: 0.6528, loss_yns_1: 0.1475, loss_cls_2: 0.8339, loss_box_2: 1.5480, loss_cns_2: 0.6602, loss_yns_2: 0.1473, loss_cls_3: 0.8380, loss_box_3: 1.5533, loss_cns_3: 0.6581, loss_yns_3: 0.1459, loss_cls_4: 0.8492, loss_box_4: 1.5366, loss_cns_4: 0.6593, loss_yns_4: 0.1459, loss_cls_5: 0.8510, loss_box_5: 1.5379, loss_cns_5: 0.6588, loss_yns_5: 0.1447, loss_cls_dn_0: 0.1512, loss_box_dn_0: 0.7240, loss_cls_dn_1: 0.1070, loss_box_dn_1: 0.6637, loss_cls_dn_2: 0.1043, loss_box_dn_2: 0.6419, loss_cls_dn_3: 0.1042, loss_box_dn_3: 0.6501, loss_cls_dn_4: 0.1102, loss_box_dn_4: 0.6464, loss_cls_dn_5: 0.1119, loss_box_dn_5: 0.6503, loss_dense_depth: 0.6980, loss: 24.5492, grad_norm: 39.3576 -2025-11-17 14:27:09,732 - mmdet - INFO - Iter [279/17500] lr: 2.111e-04, eta: 9:23:36, time: 1.503, data_time: 0.079, memory: 49163, loss_cls_0: 0.7842, loss_box_0: 1.6326, loss_cns_0: 0.6233, loss_yns_0: 0.1477, loss_cls_1: 0.8491, loss_box_1: 1.5807, loss_cns_1: 0.6564, loss_yns_1: 0.1480, loss_cls_2: 0.8593, loss_box_2: 1.5342, loss_cns_2: 0.6616, loss_yns_2: 0.1484, loss_cls_3: 0.8538, loss_box_3: 1.5417, loss_cns_3: 0.6596, loss_yns_3: 0.1462, loss_cls_4: 0.8675, loss_box_4: 1.5310, loss_cns_4: 0.6602, loss_yns_4: 0.1451, loss_cls_5: 0.8705, loss_box_5: 1.5354, loss_cns_5: 0.6601, loss_yns_5: 0.1464, loss_cls_dn_0: 0.1593, loss_box_dn_0: 0.7266, loss_cls_dn_1: 0.1095, loss_box_dn_1: 0.6654, loss_cls_dn_2: 0.1078, loss_box_dn_2: 0.6481, loss_cls_dn_3: 0.1088, loss_box_dn_3: 0.6580, loss_cls_dn_4: 0.1151, loss_box_dn_4: 0.6597, loss_cls_dn_5: 0.1167, loss_box_dn_5: 0.6661, loss_dense_depth: 0.7321, loss: 24.7163, grad_norm: 38.9410 -2025-11-17 14:27:11,235 - mmdet - INFO - Iter [280/17500] lr: 2.115e-04, eta: 9:23:06, time: 1.504, data_time: 0.081, memory: 49163, loss_cls_0: 0.7641, loss_box_0: 1.6566, loss_cns_0: 0.6235, loss_yns_0: 0.1469, loss_cls_1: 0.8331, loss_box_1: 1.6090, loss_cns_1: 0.6521, loss_yns_1: 0.1472, loss_cls_2: 0.8458, loss_box_2: 1.5769, loss_cns_2: 0.6557, loss_yns_2: 0.1456, loss_cls_3: 0.8434, loss_box_3: 1.5718, loss_cns_3: 0.6561, loss_yns_3: 0.1453, loss_cls_4: 0.8515, loss_box_4: 1.5809, loss_cns_4: 0.6562, loss_yns_4: 0.1448, loss_cls_5: 0.8568, loss_box_5: 1.5860, loss_cns_5: 0.6553, loss_yns_5: 0.1450, loss_cls_dn_0: 0.1522, loss_box_dn_0: 0.7308, loss_cls_dn_1: 0.1084, loss_box_dn_1: 0.6923, loss_cls_dn_2: 0.1083, loss_box_dn_2: 0.6771, loss_cls_dn_3: 0.1094, loss_box_dn_3: 0.6826, loss_cls_dn_4: 0.1123, loss_box_dn_4: 0.6929, loss_cls_dn_5: 0.1134, loss_box_dn_5: 0.7001, loss_dense_depth: 0.7083, loss: 24.9377, grad_norm: 47.5183 -2025-11-17 14:27:12,775 - mmdet - INFO - Iter [281/17500] lr: 2.119e-04, eta: 9:22:38, time: 1.540, data_time: 0.081, memory: 49163, loss_cls_0: 0.7521, loss_box_0: 1.6471, loss_cns_0: 0.6271, loss_yns_0: 0.1462, loss_cls_1: 0.8209, loss_box_1: 1.5682, loss_cns_1: 0.6511, loss_yns_1: 0.1433, loss_cls_2: 0.8289, loss_box_2: 1.5282, loss_cns_2: 0.6548, loss_yns_2: 0.1417, loss_cls_3: 0.8292, loss_box_3: 1.5231, loss_cns_3: 0.6569, loss_yns_3: 0.1415, loss_cls_4: 0.8349, loss_box_4: 1.5325, loss_cns_4: 0.6572, loss_yns_4: 0.1426, loss_cls_5: 0.8369, loss_box_5: 1.5308, loss_cns_5: 0.6578, loss_yns_5: 0.1428, loss_cls_dn_0: 0.1526, loss_box_dn_0: 0.7303, loss_cls_dn_1: 0.1068, loss_box_dn_1: 0.6969, loss_cls_dn_2: 0.1077, loss_box_dn_2: 0.6845, loss_cls_dn_3: 0.1056, loss_box_dn_3: 0.6893, loss_cls_dn_4: 0.1073, loss_box_dn_4: 0.6998, loss_cls_dn_5: 0.1113, loss_box_dn_5: 0.7044, loss_dense_depth: 0.7239, loss: 24.6161, grad_norm: 34.6845 -2025-11-17 14:27:14,335 - mmdet - INFO - Iter [282/17500] lr: 2.123e-04, eta: 9:22:11, time: 1.559, data_time: 0.078, memory: 49163, loss_cls_0: 0.7775, loss_box_0: 1.6435, loss_cns_0: 0.6289, loss_yns_0: 0.1497, loss_cls_1: 0.8282, loss_box_1: 1.5575, loss_cns_1: 0.6559, loss_yns_1: 0.1456, loss_cls_2: 0.8414, loss_box_2: 1.5075, loss_cns_2: 0.6518, loss_yns_2: 0.1449, loss_cls_3: 0.8437, loss_box_3: 1.5092, loss_cns_3: 0.6544, loss_yns_3: 0.1442, loss_cls_4: 0.8419, loss_box_4: 1.5149, loss_cns_4: 0.6580, loss_yns_4: 0.1455, loss_cls_5: 0.8444, loss_box_5: 1.5208, loss_cns_5: 0.6602, loss_yns_5: 0.1458, loss_cls_dn_0: 0.1526, loss_box_dn_0: 0.7317, loss_cls_dn_1: 0.1085, loss_box_dn_1: 0.7089, loss_cls_dn_2: 0.1097, loss_box_dn_2: 0.6978, loss_cls_dn_3: 0.1101, loss_box_dn_3: 0.7038, loss_cls_dn_4: 0.1153, loss_box_dn_4: 0.7086, loss_cls_dn_5: 0.1231, loss_box_dn_5: 0.7143, loss_dense_depth: 0.7046, loss: 24.7045, grad_norm: 42.5482 -2025-11-17 14:27:15,853 - mmdet - INFO - Iter [283/17500] lr: 2.127e-04, eta: 9:21:43, time: 1.519, data_time: 0.082, memory: 49163, loss_cls_0: 0.7871, loss_box_0: 1.6518, loss_cns_0: 0.6267, loss_yns_0: 0.1492, loss_cls_1: 0.8324, loss_box_1: 1.6008, loss_cns_1: 0.6497, loss_yns_1: 0.1456, loss_cls_2: 0.8279, loss_box_2: 1.5553, loss_cns_2: 0.6516, loss_yns_2: 0.1444, loss_cls_3: 0.8324, loss_box_3: 1.5450, loss_cns_3: 0.6517, loss_yns_3: 0.1444, loss_cls_4: 0.8465, loss_box_4: 1.5479, loss_cns_4: 0.6536, loss_yns_4: 0.1453, loss_cls_5: 0.8400, loss_box_5: 1.5493, loss_cns_5: 0.6524, loss_yns_5: 0.1456, loss_cls_dn_0: 0.1514, loss_box_dn_0: 0.7260, loss_cls_dn_1: 0.1152, loss_box_dn_1: 0.6976, loss_cls_dn_2: 0.1162, loss_box_dn_2: 0.6802, loss_cls_dn_3: 0.1178, loss_box_dn_3: 0.6803, loss_cls_dn_4: 0.1246, loss_box_dn_4: 0.6797, loss_cls_dn_5: 0.1289, loss_box_dn_5: 0.6874, loss_dense_depth: 0.7081, loss: 24.7903, grad_norm: 41.3918 -2025-11-17 14:27:17,398 - mmdet - INFO - Iter [284/17500] lr: 2.131e-04, eta: 9:21:16, time: 1.546, data_time: 0.076, memory: 49163, loss_cls_0: 0.7618, loss_box_0: 1.6361, loss_cns_0: 0.6261, loss_yns_0: 0.1510, loss_cls_1: 0.8269, loss_box_1: 1.5555, loss_cns_1: 0.6517, loss_yns_1: 0.1485, loss_cls_2: 0.8349, loss_box_2: 1.5309, loss_cns_2: 0.6545, loss_yns_2: 0.1472, loss_cls_3: 0.8459, loss_box_3: 1.5097, loss_cns_3: 0.6516, loss_yns_3: 0.1466, loss_cls_4: 0.8525, loss_box_4: 1.5126, loss_cns_4: 0.6510, loss_yns_4: 0.1470, loss_cls_5: 0.8571, loss_box_5: 1.5038, loss_cns_5: 0.6526, loss_yns_5: 0.1460, loss_cls_dn_0: 0.1518, loss_box_dn_0: 0.7259, loss_cls_dn_1: 0.1167, loss_box_dn_1: 0.6848, loss_cls_dn_2: 0.1156, loss_box_dn_2: 0.6666, loss_cls_dn_3: 0.1147, loss_box_dn_3: 0.6626, loss_cls_dn_4: 0.1210, loss_box_dn_4: 0.6629, loss_cls_dn_5: 0.1190, loss_box_dn_5: 0.6665, loss_dense_depth: 0.6994, loss: 24.5092, grad_norm: 41.7580 -2025-11-17 14:27:18,895 - mmdet - INFO - Iter [285/17500] lr: 2.135e-04, eta: 9:20:46, time: 1.495, data_time: 0.074, memory: 49163, loss_cls_0: 0.7895, loss_box_0: 1.6315, loss_cns_0: 0.6225, loss_yns_0: 0.1513, loss_cls_1: 0.8561, loss_box_1: 1.5208, loss_cns_1: 0.6520, loss_yns_1: 0.1494, loss_cls_2: 0.8529, loss_box_2: 1.4977, loss_cns_2: 0.6556, loss_yns_2: 0.1505, loss_cls_3: 0.8589, loss_box_3: 1.4817, loss_cns_3: 0.6555, loss_yns_3: 0.1493, loss_cls_4: 0.8715, loss_box_4: 1.4737, loss_cns_4: 0.6537, loss_yns_4: 0.1495, loss_cls_5: 0.8861, loss_box_5: 1.4735, loss_cns_5: 0.6540, loss_yns_5: 0.1480, loss_cls_dn_0: 0.1576, loss_box_dn_0: 0.7348, loss_cls_dn_1: 0.1150, loss_box_dn_1: 0.6670, loss_cls_dn_2: 0.1127, loss_box_dn_2: 0.6501, loss_cls_dn_3: 0.1126, loss_box_dn_3: 0.6457, loss_cls_dn_4: 0.1164, loss_box_dn_4: 0.6479, loss_cls_dn_5: 0.1165, loss_box_dn_5: 0.6517, loss_dense_depth: 0.7336, loss: 24.4467, grad_norm: 25.2106 -2025-11-17 14:27:20,412 - mmdet - INFO - Iter [286/17500] lr: 2.139e-04, eta: 9:20:18, time: 1.519, data_time: 0.078, memory: 49163, loss_cls_0: 0.7612, loss_box_0: 1.6388, loss_cns_0: 0.6229, loss_yns_0: 0.1528, loss_cls_1: 0.8445, loss_box_1: 1.5435, loss_cns_1: 0.6527, loss_yns_1: 0.1510, loss_cls_2: 0.8524, loss_box_2: 1.5205, loss_cns_2: 0.6548, loss_yns_2: 0.1496, loss_cls_3: 0.8545, loss_box_3: 1.5059, loss_cns_3: 0.6545, loss_yns_3: 0.1495, loss_cls_4: 0.8578, loss_box_4: 1.4887, loss_cns_4: 0.6541, loss_yns_4: 0.1495, loss_cls_5: 0.8533, loss_box_5: 1.5056, loss_cns_5: 0.6546, loss_yns_5: 0.1489, loss_cls_dn_0: 0.1497, loss_box_dn_0: 0.7234, loss_cls_dn_1: 0.1095, loss_box_dn_1: 0.6611, loss_cls_dn_2: 0.1083, loss_box_dn_2: 0.6471, loss_cls_dn_3: 0.1092, loss_box_dn_3: 0.6466, loss_cls_dn_4: 0.1127, loss_box_dn_4: 0.6473, loss_cls_dn_5: 0.1154, loss_box_dn_5: 0.6545, loss_dense_depth: 0.7215, loss: 24.4280, grad_norm: 40.0130 -2025-11-17 14:27:21,942 - mmdet - INFO - Iter [287/17500] lr: 2.143e-04, eta: 9:19:50, time: 1.530, data_time: 0.085, memory: 49163, loss_cls_0: 0.7895, loss_box_0: 1.6395, loss_cns_0: 0.6239, loss_yns_0: 0.1544, loss_cls_1: 0.8616, loss_box_1: 1.6047, loss_cns_1: 0.6558, loss_yns_1: 0.1529, loss_cls_2: 0.8715, loss_box_2: 1.5667, loss_cns_2: 0.6605, loss_yns_2: 0.1535, loss_cls_3: 0.8852, loss_box_3: 1.5547, loss_cns_3: 0.6576, loss_yns_3: 0.1525, loss_cls_4: 0.8720, loss_box_4: 1.5459, loss_cns_4: 0.6554, loss_yns_4: 0.1526, loss_cls_5: 0.8891, loss_box_5: 1.5701, loss_cns_5: 0.6570, loss_yns_5: 0.1526, loss_cls_dn_0: 0.1539, loss_box_dn_0: 0.7307, loss_cls_dn_1: 0.1137, loss_box_dn_1: 0.6780, loss_cls_dn_2: 0.1176, loss_box_dn_2: 0.6622, loss_cls_dn_3: 0.1207, loss_box_dn_3: 0.6643, loss_cls_dn_4: 0.1286, loss_box_dn_4: 0.6696, loss_cls_dn_5: 0.1251, loss_box_dn_5: 0.6802, loss_dense_depth: 0.7274, loss: 25.0512, grad_norm: 43.7209 -2025-11-17 14:27:23,439 - mmdet - INFO - Iter [288/17500] lr: 2.147e-04, eta: 9:19:21, time: 1.497, data_time: 0.086, memory: 49163, loss_cls_0: 0.7920, loss_box_0: 1.6518, loss_cns_0: 0.6252, loss_yns_0: 0.1539, loss_cls_1: 0.8658, loss_box_1: 1.6207, loss_cns_1: 0.6561, loss_yns_1: 0.1525, loss_cls_2: 0.8671, loss_box_2: 1.5831, loss_cns_2: 0.6614, loss_yns_2: 0.1524, loss_cls_3: 0.8721, loss_box_3: 1.5680, loss_cns_3: 0.6563, loss_yns_3: 0.1528, loss_cls_4: 0.8703, loss_box_4: 1.5724, loss_cns_4: 0.6557, loss_yns_4: 0.1522, loss_cls_5: 0.8896, loss_box_5: 1.5892, loss_cns_5: 0.6563, loss_yns_5: 0.1515, loss_cls_dn_0: 0.1512, loss_box_dn_0: 0.7381, loss_cls_dn_1: 0.1126, loss_box_dn_1: 0.6836, loss_cls_dn_2: 0.1182, loss_box_dn_2: 0.6693, loss_cls_dn_3: 0.1193, loss_box_dn_3: 0.6696, loss_cls_dn_4: 0.1251, loss_box_dn_4: 0.6748, loss_cls_dn_5: 0.1197, loss_box_dn_5: 0.6852, loss_dense_depth: 0.7197, loss: 25.1549, grad_norm: 32.0054 -2025-11-17 14:27:25,004 - mmdet - INFO - Iter [289/17500] lr: 2.151e-04, eta: 9:18:56, time: 1.565, data_time: 0.168, memory: 49163, loss_cls_0: 0.7481, loss_box_0: 1.6489, loss_cns_0: 0.6251, loss_yns_0: 0.1516, loss_cls_1: 0.8442, loss_box_1: 1.5458, loss_cns_1: 0.6575, loss_yns_1: 0.1514, loss_cls_2: 0.8499, loss_box_2: 1.5281, loss_cns_2: 0.6622, loss_yns_2: 0.1514, loss_cls_3: 0.8432, loss_box_3: 1.5263, loss_cns_3: 0.6597, loss_yns_3: 0.1515, loss_cls_4: 0.8412, loss_box_4: 1.5156, loss_cns_4: 0.6588, loss_yns_4: 0.1510, loss_cls_5: 0.8417, loss_box_5: 1.5171, loss_cns_5: 0.6608, loss_yns_5: 0.1506, loss_cls_dn_0: 0.1431, loss_box_dn_0: 0.7321, loss_cls_dn_1: 0.1127, loss_box_dn_1: 0.6780, loss_cls_dn_2: 0.1142, loss_box_dn_2: 0.6652, loss_cls_dn_3: 0.1140, loss_box_dn_3: 0.6686, loss_cls_dn_4: 0.1136, loss_box_dn_4: 0.6671, loss_cls_dn_5: 0.1147, loss_box_dn_5: 0.6683, loss_dense_depth: 0.6861, loss: 24.5595, grad_norm: 34.2499 -2025-11-17 14:27:26,505 - mmdet - INFO - Iter [290/17500] lr: 2.155e-04, eta: 9:18:28, time: 1.500, data_time: 0.077, memory: 49163, loss_cls_0: 0.7826, loss_box_0: 1.6438, loss_cns_0: 0.6146, loss_yns_0: 0.1497, loss_cls_1: 0.8531, loss_box_1: 1.5120, loss_cns_1: 0.6551, loss_yns_1: 0.1495, loss_cls_2: 0.8597, loss_box_2: 1.4910, loss_cns_2: 0.6563, loss_yns_2: 0.1495, loss_cls_3: 0.8659, loss_box_3: 1.4895, loss_cns_3: 0.6553, loss_yns_3: 0.1495, loss_cls_4: 0.8573, loss_box_4: 1.4790, loss_cns_4: 0.6563, loss_yns_4: 0.1506, loss_cls_5: 0.8618, loss_box_5: 1.4807, loss_cns_5: 0.6622, loss_yns_5: 0.1515, loss_cls_dn_0: 0.1492, loss_box_dn_0: 0.7294, loss_cls_dn_1: 0.1117, loss_box_dn_1: 0.6703, loss_cls_dn_2: 0.1113, loss_box_dn_2: 0.6590, loss_cls_dn_3: 0.1116, loss_box_dn_3: 0.6626, loss_cls_dn_4: 0.1152, loss_box_dn_4: 0.6573, loss_cls_dn_5: 0.1164, loss_box_dn_5: 0.6563, loss_dense_depth: 0.7375, loss: 24.4644, grad_norm: 28.6480 -2025-11-17 14:27:28,016 - mmdet - INFO - Iter [291/17500] lr: 2.159e-04, eta: 9:18:00, time: 1.511, data_time: 0.077, memory: 49163, loss_cls_0: 0.7867, loss_box_0: 1.6549, loss_cns_0: 0.6250, loss_yns_0: 0.1521, loss_cls_1: 0.8545, loss_box_1: 1.5349, loss_cns_1: 0.6594, loss_yns_1: 0.1508, loss_cls_2: 0.8577, loss_box_2: 1.5060, loss_cns_2: 0.6658, loss_yns_2: 0.1518, loss_cls_3: 0.8606, loss_box_3: 1.4947, loss_cns_3: 0.6589, loss_yns_3: 0.1508, loss_cls_4: 0.8620, loss_box_4: 1.4890, loss_cns_4: 0.6588, loss_yns_4: 0.1501, loss_cls_5: 0.8703, loss_box_5: 1.4976, loss_cns_5: 0.6607, loss_yns_5: 0.1510, loss_cls_dn_0: 0.1526, loss_box_dn_0: 0.7273, loss_cls_dn_1: 0.1084, loss_box_dn_1: 0.6624, loss_cls_dn_2: 0.1102, loss_box_dn_2: 0.6492, loss_cls_dn_3: 0.1099, loss_box_dn_3: 0.6473, loss_cls_dn_4: 0.1109, loss_box_dn_4: 0.6433, loss_cls_dn_5: 0.1146, loss_box_dn_5: 0.6453, loss_dense_depth: 0.7493, loss: 24.5348, grad_norm: 25.5402 -2025-11-17 14:27:29,517 - mmdet - INFO - Iter [292/17500] lr: 2.163e-04, eta: 9:17:32, time: 1.502, data_time: 0.077, memory: 49163, loss_cls_0: 0.7492, loss_box_0: 1.6684, loss_cns_0: 0.6239, loss_yns_0: 0.1518, loss_cls_1: 0.8416, loss_box_1: 1.5399, loss_cns_1: 0.6582, loss_yns_1: 0.1481, loss_cls_2: 0.8406, loss_box_2: 1.5143, loss_cns_2: 0.6640, loss_yns_2: 0.1490, loss_cls_3: 0.8391, loss_box_3: 1.5075, loss_cns_3: 0.6610, loss_yns_3: 0.1488, loss_cls_4: 0.8451, loss_box_4: 1.5001, loss_cns_4: 0.6605, loss_yns_4: 0.1489, loss_cls_5: 0.8537, loss_box_5: 1.5025, loss_cns_5: 0.6604, loss_yns_5: 0.1482, loss_cls_dn_0: 0.1480, loss_box_dn_0: 0.7406, loss_cls_dn_1: 0.1133, loss_box_dn_1: 0.6556, loss_cls_dn_2: 0.1147, loss_box_dn_2: 0.6374, loss_cls_dn_3: 0.1131, loss_box_dn_3: 0.6344, loss_cls_dn_4: 0.1119, loss_box_dn_4: 0.6322, loss_cls_dn_5: 0.1153, loss_box_dn_5: 0.6314, loss_dense_depth: 0.7049, loss: 24.3776, grad_norm: 31.7307 -2025-11-17 14:27:31,025 - mmdet - INFO - Iter [293/17500] lr: 2.167e-04, eta: 9:17:05, time: 1.508, data_time: 0.071, memory: 49163, loss_cls_0: 0.7706, loss_box_0: 1.6870, loss_cns_0: 0.6205, loss_yns_0: 0.1476, loss_cls_1: 0.8438, loss_box_1: 1.5398, loss_cns_1: 0.6560, loss_yns_1: 0.1463, loss_cls_2: 0.8512, loss_box_2: 1.5084, loss_cns_2: 0.6602, loss_yns_2: 0.1456, loss_cls_3: 0.8454, loss_box_3: 1.5107, loss_cns_3: 0.6588, loss_yns_3: 0.1454, loss_cls_4: 0.8532, loss_box_4: 1.5013, loss_cns_4: 0.6564, loss_yns_4: 0.1453, loss_cls_5: 0.8478, loss_box_5: 1.5201, loss_cns_5: 0.6607, loss_yns_5: 0.1462, loss_cls_dn_0: 0.1531, loss_box_dn_0: 0.7331, loss_cls_dn_1: 0.1096, loss_box_dn_1: 0.6584, loss_cls_dn_2: 0.1107, loss_box_dn_2: 0.6403, loss_cls_dn_3: 0.1080, loss_box_dn_3: 0.6408, loss_cls_dn_4: 0.1104, loss_box_dn_4: 0.6415, loss_cls_dn_5: 0.1117, loss_box_dn_5: 0.6482, loss_dense_depth: 0.7350, loss: 24.4691, grad_norm: 25.7136 -2025-11-17 14:27:32,579 - mmdet - INFO - Iter [294/17500] lr: 2.170e-04, eta: 9:16:40, time: 1.553, data_time: 0.074, memory: 49163, loss_cls_0: 0.7840, loss_box_0: 1.6836, loss_cns_0: 0.6246, loss_yns_0: 0.1495, loss_cls_1: 0.8352, loss_box_1: 1.5659, loss_cns_1: 0.6546, loss_yns_1: 0.1461, loss_cls_2: 0.8458, loss_box_2: 1.5233, loss_cns_2: 0.6579, loss_yns_2: 0.1452, loss_cls_3: 0.8462, loss_box_3: 1.5294, loss_cns_3: 0.6583, loss_yns_3: 0.1454, loss_cls_4: 0.8431, loss_box_4: 1.5258, loss_cns_4: 0.6580, loss_yns_4: 0.1455, loss_cls_5: 0.8444, loss_box_5: 1.5339, loss_cns_5: 0.6581, loss_yns_5: 0.1463, loss_cls_dn_0: 0.1483, loss_box_dn_0: 0.7240, loss_cls_dn_1: 0.1065, loss_box_dn_1: 0.6578, loss_cls_dn_2: 0.1077, loss_box_dn_2: 0.6398, loss_cls_dn_3: 0.1068, loss_box_dn_3: 0.6429, loss_cls_dn_4: 0.1099, loss_box_dn_4: 0.6437, loss_cls_dn_5: 0.1115, loss_box_dn_5: 0.6513, loss_dense_depth: 0.7579, loss: 24.5583, grad_norm: 29.0115 -2025-11-17 14:27:34,082 - mmdet - INFO - Iter [295/17500] lr: 2.174e-04, eta: 9:16:12, time: 1.504, data_time: 0.070, memory: 49163, loss_cls_0: 0.7728, loss_box_0: 1.6704, loss_cns_0: 0.6271, loss_yns_0: 0.1490, loss_cls_1: 0.8458, loss_box_1: 1.5594, loss_cns_1: 0.6593, loss_yns_1: 0.1477, loss_cls_2: 0.8615, loss_box_2: 1.5205, loss_cns_2: 0.6680, loss_yns_2: 0.1487, loss_cls_3: 0.8583, loss_box_3: 1.5217, loss_cns_3: 0.6621, loss_yns_3: 0.1471, loss_cls_4: 0.8578, loss_box_4: 1.5148, loss_cns_4: 0.6653, loss_yns_4: 0.1479, loss_cls_5: 0.8587, loss_box_5: 1.5249, loss_cns_5: 0.6595, loss_yns_5: 0.1470, loss_cls_dn_0: 0.1457, loss_box_dn_0: 0.7283, loss_cls_dn_1: 0.1086, loss_box_dn_1: 0.6698, loss_cls_dn_2: 0.1122, loss_box_dn_2: 0.6532, loss_cls_dn_3: 0.1121, loss_box_dn_3: 0.6578, loss_cls_dn_4: 0.1148, loss_box_dn_4: 0.6579, loss_cls_dn_5: 0.1166, loss_box_dn_5: 0.6671, loss_dense_depth: 0.7098, loss: 24.6493, grad_norm: 30.3992 -2025-11-17 14:27:35,587 - mmdet - INFO - Iter [296/17500] lr: 2.178e-04, eta: 9:15:45, time: 1.505, data_time: 0.074, memory: 49163, loss_cls_0: 0.7739, loss_box_0: 1.6749, loss_cns_0: 0.6245, loss_yns_0: 0.1505, loss_cls_1: 0.8519, loss_box_1: 1.5413, loss_cns_1: 0.6599, loss_yns_1: 0.1475, loss_cls_2: 0.8677, loss_box_2: 1.5196, loss_cns_2: 0.6674, loss_yns_2: 0.1498, loss_cls_3: 0.8650, loss_box_3: 1.5107, loss_cns_3: 0.6607, loss_yns_3: 0.1483, loss_cls_4: 0.8614, loss_box_4: 1.5140, loss_cns_4: 0.6640, loss_yns_4: 0.1486, loss_cls_5: 0.8655, loss_box_5: 1.5165, loss_cns_5: 0.6604, loss_yns_5: 0.1481, loss_cls_dn_0: 0.1487, loss_box_dn_0: 0.7303, loss_cls_dn_1: 0.1120, loss_box_dn_1: 0.6642, loss_cls_dn_2: 0.1138, loss_box_dn_2: 0.6489, loss_cls_dn_3: 0.1137, loss_box_dn_3: 0.6499, loss_cls_dn_4: 0.1138, loss_box_dn_4: 0.6515, loss_cls_dn_5: 0.1170, loss_box_dn_5: 0.6580, loss_dense_depth: 0.7594, loss: 24.6736, grad_norm: 24.8756 -2025-11-17 14:27:37,108 - mmdet - INFO - Iter [297/17500] lr: 2.182e-04, eta: 9:15:19, time: 1.519, data_time: 0.068, memory: 49163, loss_cls_0: 0.7986, loss_box_0: 1.7285, loss_cns_0: 0.6232, loss_yns_0: 0.1516, loss_cls_1: 0.8554, loss_box_1: 1.5897, loss_cns_1: 0.6565, loss_yns_1: 0.1507, loss_cls_2: 0.8661, loss_box_2: 1.5657, loss_cns_2: 0.6601, loss_yns_2: 0.1505, loss_cls_3: 0.8743, loss_box_3: 1.5604, loss_cns_3: 0.6578, loss_yns_3: 0.1503, loss_cls_4: 0.8765, loss_box_4: 1.5605, loss_cns_4: 0.6592, loss_yns_4: 0.1504, loss_cls_5: 0.8768, loss_box_5: 1.5599, loss_cns_5: 0.6576, loss_yns_5: 0.1508, loss_cls_dn_0: 0.1489, loss_box_dn_0: 0.7347, loss_cls_dn_1: 0.1113, loss_box_dn_1: 0.6542, loss_cls_dn_2: 0.1101, loss_box_dn_2: 0.6401, loss_cls_dn_3: 0.1107, loss_box_dn_3: 0.6393, loss_cls_dn_4: 0.1133, loss_box_dn_4: 0.6409, loss_cls_dn_5: 0.1145, loss_box_dn_5: 0.6454, loss_dense_depth: 0.8211, loss: 25.0156, grad_norm: 27.6894 -2025-11-17 14:27:38,590 - mmdet - INFO - Iter [298/17500] lr: 2.186e-04, eta: 9:14:51, time: 1.483, data_time: 0.072, memory: 49163, loss_cls_0: 0.7851, loss_box_0: 1.6840, loss_cns_0: 0.6302, loss_yns_0: 0.1512, loss_cls_1: 0.8573, loss_box_1: 1.5894, loss_cns_1: 0.6571, loss_yns_1: 0.1520, loss_cls_2: 0.8601, loss_box_2: 1.5618, loss_cns_2: 0.6585, loss_yns_2: 0.1516, loss_cls_3: 0.8717, loss_box_3: 1.5587, loss_cns_3: 0.6589, loss_yns_3: 0.1514, loss_cls_4: 0.8716, loss_box_4: 1.5522, loss_cns_4: 0.6592, loss_yns_4: 0.1515, loss_cls_5: 0.8772, loss_box_5: 1.5507, loss_cns_5: 0.6584, loss_yns_5: 0.1514, loss_cls_dn_0: 0.1438, loss_box_dn_0: 0.7213, loss_cls_dn_1: 0.1092, loss_box_dn_1: 0.6474, loss_cls_dn_2: 0.1090, loss_box_dn_2: 0.6324, loss_cls_dn_3: 0.1096, loss_box_dn_3: 0.6290, loss_cls_dn_4: 0.1138, loss_box_dn_4: 0.6282, loss_cls_dn_5: 0.1131, loss_box_dn_5: 0.6294, loss_dense_depth: 0.7442, loss: 24.7816, grad_norm: 31.4279 -2025-11-17 14:27:40,080 - mmdet - INFO - Iter [299/17500] lr: 2.190e-04, eta: 9:14:23, time: 1.491, data_time: 0.070, memory: 49163, loss_cls_0: 0.8027, loss_box_0: 1.6892, loss_cns_0: 0.6247, loss_yns_0: 0.1527, loss_cls_1: 0.8678, loss_box_1: 1.6236, loss_cns_1: 0.6537, loss_yns_1: 0.1519, loss_cls_2: 0.8759, loss_box_2: 1.5861, loss_cns_2: 0.6558, loss_yns_2: 0.1524, loss_cls_3: 0.8818, loss_box_3: 1.5798, loss_cns_3: 0.6551, loss_yns_3: 0.1518, loss_cls_4: 0.8883, loss_box_4: 1.5772, loss_cns_4: 0.6611, loss_yns_4: 0.1511, loss_cls_5: 0.8925, loss_box_5: 1.5746, loss_cns_5: 0.6551, loss_yns_5: 0.1527, loss_cls_dn_0: 0.1417, loss_box_dn_0: 0.7331, loss_cls_dn_1: 0.1090, loss_box_dn_1: 0.6550, loss_cls_dn_2: 0.1085, loss_box_dn_2: 0.6411, loss_cls_dn_3: 0.1085, loss_box_dn_3: 0.6382, loss_cls_dn_4: 0.1119, loss_box_dn_4: 0.6380, loss_cls_dn_5: 0.1145, loss_box_dn_5: 0.6372, loss_dense_depth: 0.8363, loss: 25.1308, grad_norm: 26.0169 -2025-11-17 14:27:41,568 - mmdet - INFO - Iter [300/17500] lr: 2.194e-04, eta: 9:13:56, time: 1.488, data_time: 0.071, memory: 49163, loss_cls_0: 0.7823, loss_box_0: 1.6673, loss_cns_0: 0.6259, loss_yns_0: 0.1527, loss_cls_1: 0.8465, loss_box_1: 1.6094, loss_cns_1: 0.6546, loss_yns_1: 0.1513, loss_cls_2: 0.8608, loss_box_2: 1.5819, loss_cns_2: 0.6577, loss_yns_2: 0.1519, loss_cls_3: 0.8583, loss_box_3: 1.5734, loss_cns_3: 0.6566, loss_yns_3: 0.1518, loss_cls_4: 0.8634, loss_box_4: 1.5677, loss_cns_4: 0.6644, loss_yns_4: 0.1504, loss_cls_5: 0.8677, loss_box_5: 1.5714, loss_cns_5: 0.6562, loss_yns_5: 0.1504, loss_cls_dn_0: 0.1441, loss_box_dn_0: 0.7297, loss_cls_dn_1: 0.1108, loss_box_dn_1: 0.6594, loss_cls_dn_2: 0.1117, loss_box_dn_2: 0.6463, loss_cls_dn_3: 0.1108, loss_box_dn_3: 0.6427, loss_cls_dn_4: 0.1147, loss_box_dn_4: 0.6420, loss_cls_dn_5: 0.1170, loss_box_dn_5: 0.6441, loss_dense_depth: 0.7272, loss: 24.8746, grad_norm: 33.9512 -2025-11-17 14:27:43,099 - mmdet - INFO - Iter [301/17500] lr: 2.198e-04, eta: 9:13:31, time: 1.530, data_time: 0.068, memory: 49163, loss_cls_0: 0.7995, loss_box_0: 1.6731, loss_cns_0: 0.6225, loss_yns_0: 0.1503, loss_cls_1: 0.8550, loss_box_1: 1.6308, loss_cns_1: 0.6506, loss_yns_1: 0.1493, loss_cls_2: 0.8644, loss_box_2: 1.5917, loss_cns_2: 0.6562, loss_yns_2: 0.1496, loss_cls_3: 0.8677, loss_box_3: 1.5823, loss_cns_3: 0.6528, loss_yns_3: 0.1496, loss_cls_4: 0.8664, loss_box_4: 1.5807, loss_cns_4: 0.6555, loss_yns_4: 0.1495, loss_cls_5: 0.8712, loss_box_5: 1.5848, loss_cns_5: 0.6532, loss_yns_5: 0.1502, loss_cls_dn_0: 0.1417, loss_box_dn_0: 0.7354, loss_cls_dn_1: 0.1064, loss_box_dn_1: 0.6565, loss_cls_dn_2: 0.1087, loss_box_dn_2: 0.6445, loss_cls_dn_3: 0.1085, loss_box_dn_3: 0.6429, loss_cls_dn_4: 0.1089, loss_box_dn_4: 0.6466, loss_cls_dn_5: 0.1111, loss_box_dn_5: 0.6525, loss_dense_depth: 0.8061, loss: 25.0267, grad_norm: 23.8849 -2025-11-17 14:27:44,676 - mmdet - INFO - Iter [302/17500] lr: 2.202e-04, eta: 9:13:09, time: 1.579, data_time: 0.068, memory: 49163, loss_cls_0: 0.7900, loss_box_0: 1.6616, loss_cns_0: 0.6230, loss_yns_0: 0.1520, loss_cls_1: 0.8482, loss_box_1: 1.5873, loss_cns_1: 0.6478, loss_yns_1: 0.1510, loss_cls_2: 0.8592, loss_box_2: 1.5485, loss_cns_2: 0.6497, loss_yns_2: 0.1504, loss_cls_3: 0.8576, loss_box_3: 1.5490, loss_cns_3: 0.6517, loss_yns_3: 0.1517, loss_cls_4: 0.8615, loss_box_4: 1.5530, loss_cns_4: 0.6557, loss_yns_4: 0.1523, loss_cls_5: 0.8623, loss_box_5: 1.5516, loss_cns_5: 0.6524, loss_yns_5: 0.1536, loss_cls_dn_0: 0.1400, loss_box_dn_0: 0.7237, loss_cls_dn_1: 0.1059, loss_box_dn_1: 0.6554, loss_cls_dn_2: 0.1067, loss_box_dn_2: 0.6438, loss_cls_dn_3: 0.1063, loss_box_dn_3: 0.6450, loss_cls_dn_4: 0.1082, loss_box_dn_4: 0.6498, loss_cls_dn_5: 0.1094, loss_box_dn_5: 0.6571, loss_dense_depth: 0.7488, loss: 24.7207, grad_norm: 25.8651 -2025-11-17 14:27:46,158 - mmdet - INFO - Iter [303/17500] lr: 2.206e-04, eta: 9:12:41, time: 1.481, data_time: 0.070, memory: 49163, loss_cls_0: 0.7910, loss_box_0: 1.6805, loss_cns_0: 0.6243, loss_yns_0: 0.1557, loss_cls_1: 0.8589, loss_box_1: 1.6071, loss_cns_1: 0.6496, loss_yns_1: 0.1539, loss_cls_2: 0.8819, loss_box_2: 1.5628, loss_cns_2: 0.6528, loss_yns_2: 0.1530, loss_cls_3: 0.8693, loss_box_3: 1.5615, loss_cns_3: 0.6547, loss_yns_3: 0.1535, loss_cls_4: 0.8689, loss_box_4: 1.5664, loss_cns_4: 0.6610, loss_yns_4: 0.1530, loss_cls_5: 0.8690, loss_box_5: 1.5706, loss_cns_5: 0.6551, loss_yns_5: 0.1540, loss_cls_dn_0: 0.1413, loss_box_dn_0: 0.7213, loss_cls_dn_1: 0.1063, loss_box_dn_1: 0.6677, loss_cls_dn_2: 0.1062, loss_box_dn_2: 0.6541, loss_cls_dn_3: 0.1051, loss_box_dn_3: 0.6531, loss_cls_dn_4: 0.1097, loss_box_dn_4: 0.6554, loss_cls_dn_5: 0.1089, loss_box_dn_5: 0.6601, loss_dense_depth: 0.7135, loss: 24.9108, grad_norm: 23.6894 -2025-11-17 14:27:47,704 - mmdet - INFO - Iter [304/17500] lr: 2.210e-04, eta: 9:12:18, time: 1.545, data_time: 0.069, memory: 49163, loss_cls_0: 0.7765, loss_box_0: 1.6935, loss_cns_0: 0.6244, loss_yns_0: 0.1525, loss_cls_1: 0.8370, loss_box_1: 1.5773, loss_cns_1: 0.6532, loss_yns_1: 0.1526, loss_cls_2: 0.8505, loss_box_2: 1.5445, loss_cns_2: 0.6590, loss_yns_2: 0.1536, loss_cls_3: 0.8504, loss_box_3: 1.5345, loss_cns_3: 0.6571, loss_yns_3: 0.1524, loss_cls_4: 0.8531, loss_box_4: 1.5346, loss_cns_4: 0.6598, loss_yns_4: 0.1540, loss_cls_5: 0.8496, loss_box_5: 1.5363, loss_cns_5: 0.6564, loss_yns_5: 0.1526, loss_cls_dn_0: 0.1400, loss_box_dn_0: 0.7230, loss_cls_dn_1: 0.1072, loss_box_dn_1: 0.6671, loss_cls_dn_2: 0.1055, loss_box_dn_2: 0.6492, loss_cls_dn_3: 0.1058, loss_box_dn_3: 0.6458, loss_cls_dn_4: 0.1078, loss_box_dn_4: 0.6455, loss_cls_dn_5: 0.1090, loss_box_dn_5: 0.6467, loss_dense_depth: 0.7359, loss: 24.6541, grad_norm: 21.0630 -2025-11-17 14:27:49,195 - mmdet - INFO - Iter [305/17500] lr: 2.214e-04, eta: 9:11:51, time: 1.492, data_time: 0.073, memory: 49163, loss_cls_0: 0.7558, loss_box_0: 1.6886, loss_cns_0: 0.6289, loss_yns_0: 0.1534, loss_cls_1: 0.8335, loss_box_1: 1.5380, loss_cns_1: 0.6593, loss_yns_1: 0.1522, loss_cls_2: 0.8443, loss_box_2: 1.5130, loss_cns_2: 0.6610, loss_yns_2: 0.1521, loss_cls_3: 0.8363, loss_box_3: 1.5022, loss_cns_3: 0.6591, loss_yns_3: 0.1511, loss_cls_4: 0.8376, loss_box_4: 1.5011, loss_cns_4: 0.6606, loss_yns_4: 0.1504, loss_cls_5: 0.8347, loss_box_5: 1.5083, loss_cns_5: 0.6598, loss_yns_5: 0.1515, loss_cls_dn_0: 0.1390, loss_box_dn_0: 0.7241, loss_cls_dn_1: 0.1074, loss_box_dn_1: 0.6662, loss_cls_dn_2: 0.1052, loss_box_dn_2: 0.6515, loss_cls_dn_3: 0.1044, loss_box_dn_3: 0.6495, loss_cls_dn_4: 0.1053, loss_box_dn_4: 0.6494, loss_cls_dn_5: 0.1082, loss_box_dn_5: 0.6529, loss_dense_depth: 0.7017, loss: 24.3975, grad_norm: 27.0627 -2025-11-17 14:27:50,701 - mmdet - INFO - Iter [306/17500] lr: 2.218e-04, eta: 9:11:26, time: 1.505, data_time: 0.075, memory: 49163, loss_cls_0: 0.7699, loss_box_0: 1.6654, loss_cns_0: 0.6243, loss_yns_0: 0.1532, loss_cls_1: 0.8406, loss_box_1: 1.5379, loss_cns_1: 0.6591, loss_yns_1: 0.1506, loss_cls_2: 0.8529, loss_box_2: 1.5054, loss_cns_2: 0.6583, loss_yns_2: 0.1505, loss_cls_3: 0.8462, loss_box_3: 1.4853, loss_cns_3: 0.6556, loss_yns_3: 0.1511, loss_cls_4: 0.8427, loss_box_4: 1.4983, loss_cns_4: 0.6568, loss_yns_4: 0.1503, loss_cls_5: 0.8484, loss_box_5: 1.5027, loss_cns_5: 0.6560, loss_yns_5: 0.1505, loss_cls_dn_0: 0.1432, loss_box_dn_0: 0.7261, loss_cls_dn_1: 0.1096, loss_box_dn_1: 0.6620, loss_cls_dn_2: 0.1081, loss_box_dn_2: 0.6476, loss_cls_dn_3: 0.1078, loss_box_dn_3: 0.6467, loss_cls_dn_4: 0.1092, loss_box_dn_4: 0.6484, loss_cls_dn_5: 0.1115, loss_box_dn_5: 0.6525, loss_dense_depth: 0.7240, loss: 24.4085, grad_norm: 29.4622 -2025-11-17 14:27:52,209 - mmdet - INFO - Iter [307/17500] lr: 2.222e-04, eta: 9:11:01, time: 1.508, data_time: 0.077, memory: 49163, loss_cls_0: 0.7624, loss_box_0: 1.6563, loss_cns_0: 0.6245, loss_yns_0: 0.1533, loss_cls_1: 0.8258, loss_box_1: 1.5277, loss_cns_1: 0.6563, loss_yns_1: 0.1531, loss_cls_2: 0.8383, loss_box_2: 1.5039, loss_cns_2: 0.6594, loss_yns_2: 0.1529, loss_cls_3: 0.8327, loss_box_3: 1.4884, loss_cns_3: 0.6553, loss_yns_3: 0.1525, loss_cls_4: 0.8338, loss_box_4: 1.4838, loss_cns_4: 0.6570, loss_yns_4: 0.1519, loss_cls_5: 0.8298, loss_box_5: 1.4903, loss_cns_5: 0.6574, loss_yns_5: 0.1512, loss_cls_dn_0: 0.1413, loss_box_dn_0: 0.7218, loss_cls_dn_1: 0.1075, loss_box_dn_1: 0.6607, loss_cls_dn_2: 0.1065, loss_box_dn_2: 0.6460, loss_cls_dn_3: 0.1063, loss_box_dn_3: 0.6479, loss_cls_dn_4: 0.1087, loss_box_dn_4: 0.6522, loss_cls_dn_5: 0.1097, loss_box_dn_5: 0.6556, loss_dense_depth: 0.7045, loss: 24.2666, grad_norm: 25.0490 -2025-11-17 14:27:55,296 - mmdet - INFO - Iter [308/17500] lr: 2.226e-04, eta: 9:10:37, time: 1.534, data_time: 0.080, memory: 49163, loss_cls_0: 0.7815, loss_box_0: 1.6682, loss_cns_0: 0.6272, loss_yns_0: 0.1505, loss_cls_1: 0.8374, loss_box_1: 1.5527, loss_cns_1: 0.6557, loss_yns_1: 0.1484, loss_cls_2: 0.8488, loss_box_2: 1.5186, loss_cns_2: 0.6576, loss_yns_2: 0.1492, loss_cls_3: 0.8630, loss_box_3: 1.5084, loss_cns_3: 0.6588, loss_yns_3: 0.1499, loss_cls_4: 0.8611, loss_box_4: 1.5088, loss_cns_4: 0.6587, loss_yns_4: 0.1502, loss_cls_5: 0.8537, loss_box_5: 1.5161, loss_cns_5: 0.6589, loss_yns_5: 0.1506, loss_cls_dn_0: 0.1468, loss_box_dn_0: 0.7289, loss_cls_dn_1: 0.1089, loss_box_dn_1: 0.6769, loss_cls_dn_2: 0.1102, loss_box_dn_2: 0.6632, loss_cls_dn_3: 0.1097, loss_box_dn_3: 0.6660, loss_cls_dn_4: 0.1107, loss_box_dn_4: 0.6722, loss_cls_dn_5: 0.1130, loss_box_dn_5: 0.6792, loss_dense_depth: 0.6945, loss: 24.6141, grad_norm: 31.6591 -2025-11-17 14:27:56,850 - mmdet - INFO - Iter [309/17500] lr: 2.230e-04, eta: 9:11:41, time: 3.107, data_time: 1.699, memory: 49163, loss_cls_0: 0.7837, loss_box_0: 1.6802, loss_cns_0: 0.6222, loss_yns_0: 0.1514, loss_cls_1: 0.8413, loss_box_1: 1.5554, loss_cns_1: 0.6577, loss_yns_1: 0.1501, loss_cls_2: 0.8526, loss_box_2: 1.5255, loss_cns_2: 0.6597, loss_yns_2: 0.1509, loss_cls_3: 0.8492, loss_box_3: 1.5257, loss_cns_3: 0.6606, loss_yns_3: 0.1502, loss_cls_4: 0.8537, loss_box_4: 1.5283, loss_cns_4: 0.6604, loss_yns_4: 0.1508, loss_cls_5: 0.8531, loss_box_5: 1.5284, loss_cns_5: 0.6601, loss_yns_5: 0.1507, loss_cls_dn_0: 0.1438, loss_box_dn_0: 0.7270, loss_cls_dn_1: 0.1091, loss_box_dn_1: 0.6896, loss_cls_dn_2: 0.1144, loss_box_dn_2: 0.6796, loss_cls_dn_3: 0.1131, loss_box_dn_3: 0.6861, loss_cls_dn_4: 0.1137, loss_box_dn_4: 0.6960, loss_cls_dn_5: 0.1184, loss_box_dn_5: 0.7035, loss_dense_depth: 0.7149, loss: 24.8113, grad_norm: 34.6583 -2025-11-17 14:27:58,344 - mmdet - INFO - Iter [310/17500] lr: 2.234e-04, eta: 9:11:15, time: 1.494, data_time: 0.079, memory: 49163, loss_cls_0: 0.7508, loss_box_0: 1.6458, loss_cns_0: 0.6264, loss_yns_0: 0.1501, loss_cls_1: 0.8321, loss_box_1: 1.5125, loss_cns_1: 0.6558, loss_yns_1: 0.1481, loss_cls_2: 0.8451, loss_box_2: 1.4899, loss_cns_2: 0.6565, loss_yns_2: 0.1472, loss_cls_3: 0.8383, loss_box_3: 1.4902, loss_cns_3: 0.6593, loss_yns_3: 0.1476, loss_cls_4: 0.8480, loss_box_4: 1.4920, loss_cns_4: 0.6586, loss_yns_4: 0.1472, loss_cls_5: 0.8441, loss_box_5: 1.5032, loss_cns_5: 0.6619, loss_yns_5: 0.1481, loss_cls_dn_0: 0.1418, loss_box_dn_0: 0.7285, loss_cls_dn_1: 0.1091, loss_box_dn_1: 0.6859, loss_cls_dn_2: 0.1131, loss_box_dn_2: 0.6775, loss_cls_dn_3: 0.1103, loss_box_dn_3: 0.6812, loss_cls_dn_4: 0.1112, loss_box_dn_4: 0.6885, loss_cls_dn_5: 0.1157, loss_box_dn_5: 0.6941, loss_dense_depth: 0.6827, loss: 24.4384, grad_norm: 31.6873 -2025-11-17 14:27:59,835 - mmdet - INFO - Iter [311/17500] lr: 2.238e-04, eta: 9:10:49, time: 1.490, data_time: 0.076, memory: 49163, loss_cls_0: 0.7690, loss_box_0: 1.6527, loss_cns_0: 0.6260, loss_yns_0: 0.1522, loss_cls_1: 0.8228, loss_box_1: 1.5402, loss_cns_1: 0.6550, loss_yns_1: 0.1490, loss_cls_2: 0.8347, loss_box_2: 1.5139, loss_cns_2: 0.6537, loss_yns_2: 0.1466, loss_cls_3: 0.8390, loss_box_3: 1.5041, loss_cns_3: 0.6553, loss_yns_3: 0.1478, loss_cls_4: 0.8412, loss_box_4: 1.5107, loss_cns_4: 0.6574, loss_yns_4: 0.1477, loss_cls_5: 0.8442, loss_box_5: 1.5256, loss_cns_5: 0.6583, loss_yns_5: 0.1477, loss_cls_dn_0: 0.1447, loss_box_dn_0: 0.7301, loss_cls_dn_1: 0.1097, loss_box_dn_1: 0.6866, loss_cls_dn_2: 0.1100, loss_box_dn_2: 0.6751, loss_cls_dn_3: 0.1098, loss_box_dn_3: 0.6721, loss_cls_dn_4: 0.1112, loss_box_dn_4: 0.6746, loss_cls_dn_5: 0.1129, loss_box_dn_5: 0.6781, loss_dense_depth: 0.7189, loss: 24.5284, grad_norm: 31.4746 -2025-11-17 14:28:01,329 - mmdet - INFO - Iter [312/17500] lr: 2.242e-04, eta: 9:10:24, time: 1.494, data_time: 0.078, memory: 49163, loss_cls_0: 0.7763, loss_box_0: 1.6713, loss_cns_0: 0.6251, loss_yns_0: 0.1523, loss_cls_1: 0.8428, loss_box_1: 1.5559, loss_cns_1: 0.6551, loss_yns_1: 0.1494, loss_cls_2: 0.8409, loss_box_2: 1.5176, loss_cns_2: 0.6542, loss_yns_2: 0.1471, loss_cls_3: 0.8463, loss_box_3: 1.5113, loss_cns_3: 0.6567, loss_yns_3: 0.1473, loss_cls_4: 0.8532, loss_box_4: 1.5078, loss_cns_4: 0.6568, loss_yns_4: 0.1481, loss_cls_5: 0.8683, loss_box_5: 1.5076, loss_cns_5: 0.6569, loss_yns_5: 0.1476, loss_cls_dn_0: 0.1470, loss_box_dn_0: 0.7212, loss_cls_dn_1: 0.1121, loss_box_dn_1: 0.6654, loss_cls_dn_2: 0.1107, loss_box_dn_2: 0.6478, loss_cls_dn_3: 0.1097, loss_box_dn_3: 0.6440, loss_cls_dn_4: 0.1126, loss_box_dn_4: 0.6423, loss_cls_dn_5: 0.1124, loss_box_dn_5: 0.6413, loss_dense_depth: 0.7112, loss: 24.4737, grad_norm: 26.8096 -2025-11-17 14:28:02,843 - mmdet - INFO - Iter [313/17500] lr: 2.246e-04, eta: 9:09:59, time: 1.511, data_time: 0.081, memory: 49163, loss_cls_0: 0.7657, loss_box_0: 1.6210, loss_cns_0: 0.6223, loss_yns_0: 0.1516, loss_cls_1: 0.8231, loss_box_1: 1.5183, loss_cns_1: 0.6536, loss_yns_1: 0.1498, loss_cls_2: 0.8271, loss_box_2: 1.4845, loss_cns_2: 0.6575, loss_yns_2: 0.1499, loss_cls_3: 0.8287, loss_box_3: 1.4797, loss_cns_3: 0.6557, loss_yns_3: 0.1491, loss_cls_4: 0.8374, loss_box_4: 1.4750, loss_cns_4: 0.6595, loss_yns_4: 0.1495, loss_cls_5: 0.8411, loss_box_5: 1.4777, loss_cns_5: 0.6572, loss_yns_5: 0.1485, loss_cls_dn_0: 0.1386, loss_box_dn_0: 0.7259, loss_cls_dn_1: 0.1072, loss_box_dn_1: 0.6523, loss_cls_dn_2: 0.1049, loss_box_dn_2: 0.6368, loss_cls_dn_3: 0.1042, loss_box_dn_3: 0.6388, loss_cls_dn_4: 0.1064, loss_box_dn_4: 0.6388, loss_cls_dn_5: 0.1078, loss_box_dn_5: 0.6445, loss_dense_depth: 0.7027, loss: 24.0924, grad_norm: 36.4531 -2025-11-17 14:28:04,373 - mmdet - INFO - Iter [314/17500] lr: 2.250e-04, eta: 9:09:36, time: 1.531, data_time: 0.080, memory: 49163, loss_cls_0: 0.7957, loss_box_0: 1.6339, loss_cns_0: 0.6250, loss_yns_0: 0.1493, loss_cls_1: 0.8383, loss_box_1: 1.5246, loss_cns_1: 0.6534, loss_yns_1: 0.1476, loss_cls_2: 0.8439, loss_box_2: 1.4954, loss_cns_2: 0.6585, loss_yns_2: 0.1489, loss_cls_3: 0.8511, loss_box_3: 1.4863, loss_cns_3: 0.6567, loss_yns_3: 0.1488, loss_cls_4: 0.8605, loss_box_4: 1.4809, loss_cns_4: 0.6607, loss_yns_4: 0.1497, loss_cls_5: 0.8618, loss_box_5: 1.4811, loss_cns_5: 0.6577, loss_yns_5: 0.1494, loss_cls_dn_0: 0.1452, loss_box_dn_0: 0.7261, loss_cls_dn_1: 0.1128, loss_box_dn_1: 0.6524, loss_cls_dn_2: 0.1112, loss_box_dn_2: 0.6376, loss_cls_dn_3: 0.1111, loss_box_dn_3: 0.6405, loss_cls_dn_4: 0.1121, loss_box_dn_4: 0.6460, loss_cls_dn_5: 0.1155, loss_box_dn_5: 0.6572, loss_dense_depth: 0.7033, loss: 24.3304, grad_norm: 32.0367 -2025-11-17 14:28:05,877 - mmdet - INFO - Iter [315/17500] lr: 2.254e-04, eta: 9:09:12, time: 1.504, data_time: 0.077, memory: 49163, loss_cls_0: 0.7719, loss_box_0: 1.6439, loss_cns_0: 0.6236, loss_yns_0: 0.1496, loss_cls_1: 0.8403, loss_box_1: 1.5288, loss_cns_1: 0.6490, loss_yns_1: 0.1493, loss_cls_2: 0.8537, loss_box_2: 1.5031, loss_cns_2: 0.6525, loss_yns_2: 0.1479, loss_cls_3: 0.8501, loss_box_3: 1.4968, loss_cns_3: 0.6508, loss_yns_3: 0.1491, loss_cls_4: 0.8588, loss_box_4: 1.5007, loss_cns_4: 0.6510, loss_yns_4: 0.1492, loss_cls_5: 0.8628, loss_box_5: 1.4998, loss_cns_5: 0.6498, loss_yns_5: 0.1485, loss_cls_dn_0: 0.1400, loss_box_dn_0: 0.7347, loss_cls_dn_1: 0.1105, loss_box_dn_1: 0.6729, loss_cls_dn_2: 0.1074, loss_box_dn_2: 0.6620, loss_cls_dn_3: 0.1069, loss_box_dn_3: 0.6688, loss_cls_dn_4: 0.1095, loss_box_dn_4: 0.6817, loss_cls_dn_5: 0.1118, loss_box_dn_5: 0.6939, loss_dense_depth: 0.6852, loss: 24.4660, grad_norm: 35.6628 -2025-11-17 14:28:07,372 - mmdet - INFO - Iter [316/17500] lr: 2.258e-04, eta: 9:08:47, time: 1.495, data_time: 0.081, memory: 49163, loss_cls_0: 0.7684, loss_box_0: 1.6149, loss_cns_0: 0.6269, loss_yns_0: 0.1474, loss_cls_1: 0.8445, loss_box_1: 1.5134, loss_cns_1: 0.6509, loss_yns_1: 0.1460, loss_cls_2: 0.8485, loss_box_2: 1.4938, loss_cns_2: 0.6535, loss_yns_2: 0.1450, loss_cls_3: 0.8552, loss_box_3: 1.4969, loss_cns_3: 0.6548, loss_yns_3: 0.1468, loss_cls_4: 0.8630, loss_box_4: 1.4922, loss_cns_4: 0.6555, loss_yns_4: 0.1455, loss_cls_5: 0.8616, loss_box_5: 1.4917, loss_cns_5: 0.6534, loss_yns_5: 0.1450, loss_cls_dn_0: 0.1379, loss_box_dn_0: 0.7242, loss_cls_dn_1: 0.1122, loss_box_dn_1: 0.6925, loss_cls_dn_2: 0.1098, loss_box_dn_2: 0.6864, loss_cls_dn_3: 0.1101, loss_box_dn_3: 0.6966, loss_cls_dn_4: 0.1115, loss_box_dn_4: 0.7114, loss_cls_dn_5: 0.1107, loss_box_dn_5: 0.7243, loss_dense_depth: 0.6803, loss: 24.5229, grad_norm: 38.3826 -2025-11-17 14:28:08,893 - mmdet - INFO - Iter [317/17500] lr: 2.262e-04, eta: 9:08:23, time: 1.522, data_time: 0.079, memory: 49163, loss_cls_0: 0.7463, loss_box_0: 1.6223, loss_cns_0: 0.6295, loss_yns_0: 0.1498, loss_cls_1: 0.8312, loss_box_1: 1.4878, loss_cns_1: 0.6519, loss_yns_1: 0.1471, loss_cls_2: 0.8346, loss_box_2: 1.4733, loss_cns_2: 0.6543, loss_yns_2: 0.1453, loss_cls_3: 0.8333, loss_box_3: 1.4679, loss_cns_3: 0.6554, loss_yns_3: 0.1459, loss_cls_4: 0.8417, loss_box_4: 1.4659, loss_cns_4: 0.6561, loss_yns_4: 0.1451, loss_cls_5: 0.8467, loss_box_5: 1.4726, loss_cns_5: 0.6543, loss_yns_5: 0.1464, loss_cls_dn_0: 0.1368, loss_box_dn_0: 0.7284, loss_cls_dn_1: 0.1101, loss_box_dn_1: 0.7295, loss_cls_dn_2: 0.1098, loss_box_dn_2: 0.7250, loss_cls_dn_3: 0.1101, loss_box_dn_3: 0.7288, loss_cls_dn_4: 0.1093, loss_box_dn_4: 0.7408, loss_cls_dn_5: 0.1092, loss_box_dn_5: 0.7529, loss_dense_depth: 0.6732, loss: 24.4686, grad_norm: 31.3503 -2025-11-17 14:28:10,395 - mmdet - INFO - Iter [318/17500] lr: 2.266e-04, eta: 9:07:59, time: 1.500, data_time: 0.082, memory: 49163, loss_cls_0: 0.7827, loss_box_0: 1.6356, loss_cns_0: 0.6289, loss_yns_0: 0.1507, loss_cls_1: 0.8294, loss_box_1: 1.5282, loss_cns_1: 0.6537, loss_yns_1: 0.1476, loss_cls_2: 0.8569, loss_box_2: 1.4988, loss_cns_2: 0.6510, loss_yns_2: 0.1458, loss_cls_3: 0.8536, loss_box_3: 1.4781, loss_cns_3: 0.6500, loss_yns_3: 0.1465, loss_cls_4: 0.8579, loss_box_4: 1.4818, loss_cns_4: 0.6534, loss_yns_4: 0.1469, loss_cls_5: 0.8604, loss_box_5: 1.4857, loss_cns_5: 0.6530, loss_yns_5: 0.1482, loss_cls_dn_0: 0.1388, loss_box_dn_0: 0.7224, loss_cls_dn_1: 0.1053, loss_box_dn_1: 0.6970, loss_cls_dn_2: 0.1085, loss_box_dn_2: 0.6848, loss_cls_dn_3: 0.1070, loss_box_dn_3: 0.6813, loss_cls_dn_4: 0.1054, loss_box_dn_4: 0.6854, loss_cls_dn_5: 0.1081, loss_box_dn_5: 0.6901, loss_dense_depth: 0.7086, loss: 24.4675, grad_norm: 37.1478 -2025-11-17 14:28:13,774 - mmdet - INFO - Iter [319/17500] lr: 2.270e-04, eta: 9:09:16, time: 3.381, data_time: 0.081, memory: 49163, loss_cls_0: 0.7699, loss_box_0: 1.6132, loss_cns_0: 0.6303, loss_yns_0: 0.1508, loss_cls_1: 0.8262, loss_box_1: 1.5171, loss_cns_1: 0.6563, loss_yns_1: 0.1499, loss_cls_2: 0.8283, loss_box_2: 1.4730, loss_cns_2: 0.6580, loss_yns_2: 0.1488, loss_cls_3: 0.8326, loss_box_3: 1.4690, loss_cns_3: 0.6590, loss_yns_3: 0.1495, loss_cls_4: 0.8426, loss_box_4: 1.4666, loss_cns_4: 0.6604, loss_yns_4: 0.1490, loss_cls_5: 0.8447, loss_box_5: 1.4690, loss_cns_5: 0.6574, loss_yns_5: 0.1499, loss_cls_dn_0: 0.1341, loss_box_dn_0: 0.7241, loss_cls_dn_1: 0.1016, loss_box_dn_1: 0.6660, loss_cls_dn_2: 0.1025, loss_box_dn_2: 0.6453, loss_cls_dn_3: 0.1024, loss_box_dn_3: 0.6415, loss_cls_dn_4: 0.1036, loss_box_dn_4: 0.6409, loss_cls_dn_5: 0.1073, loss_box_dn_5: 0.6440, loss_dense_depth: 0.6794, loss: 24.0639, grad_norm: 25.3727 -2025-11-17 14:28:15,258 - mmdet - INFO - Iter [320/17500] lr: 2.274e-04, eta: 9:08:51, time: 1.483, data_time: 0.080, memory: 49163, loss_cls_0: 0.7586, loss_box_0: 1.6162, loss_cns_0: 0.6272, loss_yns_0: 0.1483, loss_cls_1: 0.8396, loss_box_1: 1.5160, loss_cns_1: 0.6603, loss_yns_1: 0.1491, loss_cls_2: 0.8485, loss_box_2: 1.4826, loss_cns_2: 0.6620, loss_yns_2: 0.1480, loss_cls_3: 0.8399, loss_box_3: 1.4740, loss_cns_3: 0.6625, loss_yns_3: 0.1482, loss_cls_4: 0.8454, loss_box_4: 1.4715, loss_cns_4: 0.6634, loss_yns_4: 0.1481, loss_cls_5: 0.8451, loss_box_5: 1.4732, loss_cns_5: 0.6619, loss_yns_5: 0.1473, loss_cls_dn_0: 0.1315, loss_box_dn_0: 0.7181, loss_cls_dn_1: 0.1050, loss_box_dn_1: 0.6479, loss_cls_dn_2: 0.1029, loss_box_dn_2: 0.6294, loss_cls_dn_3: 0.1025, loss_box_dn_3: 0.6262, loss_cls_dn_4: 0.1064, loss_box_dn_4: 0.6269, loss_cls_dn_5: 0.1069, loss_box_dn_5: 0.6289, loss_dense_depth: 0.6703, loss: 24.0401, grad_norm: 26.8707 -2025-11-17 14:28:16,786 - mmdet - INFO - Iter [321/17500] lr: 2.278e-04, eta: 9:08:28, time: 1.528, data_time: 0.070, memory: 49163, loss_cls_0: 0.7441, loss_box_0: 1.5881, loss_cns_0: 0.6275, loss_yns_0: 0.1489, loss_cls_1: 0.8028, loss_box_1: 1.4995, loss_cns_1: 0.6595, loss_yns_1: 0.1498, loss_cls_2: 0.8130, loss_box_2: 1.4679, loss_cns_2: 0.6607, loss_yns_2: 0.1498, loss_cls_3: 0.8098, loss_box_3: 1.4616, loss_cns_3: 0.6621, loss_yns_3: 0.1493, loss_cls_4: 0.8126, loss_box_4: 1.4556, loss_cns_4: 0.6615, loss_yns_4: 0.1496, loss_cls_5: 0.8188, loss_box_5: 1.4509, loss_cns_5: 0.6606, loss_yns_5: 0.1498, loss_cls_dn_0: 0.1296, loss_box_dn_0: 0.7168, loss_cls_dn_1: 0.1043, loss_box_dn_1: 0.6424, loss_cls_dn_2: 0.1016, loss_box_dn_2: 0.6298, loss_cls_dn_3: 0.1015, loss_box_dn_3: 0.6280, loss_cls_dn_4: 0.1044, loss_box_dn_4: 0.6290, loss_cls_dn_5: 0.1060, loss_box_dn_5: 0.6323, loss_dense_depth: 0.6403, loss: 23.7199, grad_norm: 26.9048 -2025-11-17 14:28:20,213 - mmdet - INFO - Iter [322/17500] lr: 2.282e-04, eta: 9:09:47, time: 3.428, data_time: 0.079, memory: 49163, loss_cls_0: 0.7546, loss_box_0: 1.6113, loss_cns_0: 0.6284, loss_yns_0: 0.1496, loss_cls_1: 0.8169, loss_box_1: 1.4794, loss_cns_1: 0.6593, loss_yns_1: 0.1483, loss_cls_2: 0.8255, loss_box_2: 1.4503, loss_cns_2: 0.6604, loss_yns_2: 0.1492, loss_cls_3: 0.8340, loss_box_3: 1.4454, loss_cns_3: 0.6605, loss_yns_3: 0.1497, loss_cls_4: 0.8370, loss_box_4: 1.4305, loss_cns_4: 0.6569, loss_yns_4: 0.1485, loss_cls_5: 0.8339, loss_box_5: 1.4358, loss_cns_5: 0.6598, loss_yns_5: 0.1490, loss_cls_dn_0: 0.1300, loss_box_dn_0: 0.7213, loss_cls_dn_1: 0.1037, loss_box_dn_1: 0.6579, loss_cls_dn_2: 0.1008, loss_box_dn_2: 0.6459, loss_cls_dn_3: 0.1004, loss_box_dn_3: 0.6489, loss_cls_dn_4: 0.1020, loss_box_dn_4: 0.6548, loss_cls_dn_5: 0.1031, loss_box_dn_5: 0.6630, loss_dense_depth: 0.6584, loss: 23.8642, grad_norm: 29.9801 -2025-11-17 14:28:23,523 - mmdet - INFO - Iter [323/17500] lr: 2.286e-04, eta: 9:10:59, time: 3.310, data_time: 0.080, memory: 49163, loss_cls_0: 0.7533, loss_box_0: 1.6221, loss_cns_0: 0.6330, loss_yns_0: 0.1473, loss_cls_1: 0.8100, loss_box_1: 1.5306, loss_cns_1: 0.6571, loss_yns_1: 0.1470, loss_cls_2: 0.8340, loss_box_2: 1.5074, loss_cns_2: 0.6562, loss_yns_2: 0.1482, loss_cls_3: 0.8204, loss_box_3: 1.5050, loss_cns_3: 0.6583, loss_yns_3: 0.1473, loss_cls_4: 0.8296, loss_box_4: 1.4957, loss_cns_4: 0.6538, loss_yns_4: 0.1477, loss_cls_5: 0.8304, loss_box_5: 1.5108, loss_cns_5: 0.6562, loss_yns_5: 0.1484, loss_cls_dn_0: 0.1308, loss_box_dn_0: 0.7227, loss_cls_dn_1: 0.1015, loss_box_dn_1: 0.6820, loss_cls_dn_2: 0.1015, loss_box_dn_2: 0.6735, loss_cls_dn_3: 0.1005, loss_box_dn_3: 0.6834, loss_cls_dn_4: 0.1050, loss_box_dn_4: 0.6969, loss_cls_dn_5: 0.1041, loss_box_dn_5: 0.7126, loss_dense_depth: 0.6503, loss: 24.3148, grad_norm: 44.9689 -2025-11-17 14:28:25,062 - mmdet - INFO - Iter [324/17500] lr: 2.290e-04, eta: 9:10:37, time: 1.540, data_time: 0.076, memory: 49163, loss_cls_0: 0.7613, loss_box_0: 1.6396, loss_cns_0: 0.6326, loss_yns_0: 0.1464, loss_cls_1: 0.8094, loss_box_1: 1.5441, loss_cns_1: 0.6569, loss_yns_1: 0.1460, loss_cls_2: 0.8312, loss_box_2: 1.5074, loss_cns_2: 0.6575, loss_yns_2: 0.1458, loss_cls_3: 0.8293, loss_box_3: 1.4995, loss_cns_3: 0.6615, loss_yns_3: 0.1462, loss_cls_4: 0.8362, loss_box_4: 1.4957, loss_cns_4: 0.6585, loss_yns_4: 0.1462, loss_cls_5: 0.8369, loss_box_5: 1.5071, loss_cns_5: 0.6608, loss_yns_5: 0.1479, loss_cls_dn_0: 0.1281, loss_box_dn_0: 0.7187, loss_cls_dn_1: 0.1009, loss_box_dn_1: 0.6894, loss_cls_dn_2: 0.1012, loss_box_dn_2: 0.6785, loss_cls_dn_3: 0.1008, loss_box_dn_3: 0.6866, loss_cls_dn_4: 0.1047, loss_box_dn_4: 0.7006, loss_cls_dn_5: 0.1047, loss_box_dn_5: 0.7158, loss_dense_depth: 0.6611, loss: 24.3952, grad_norm: 38.5532 -2025-11-17 14:28:26,550 - mmdet - INFO - Iter [325/17500] lr: 2.294e-04, eta: 9:10:12, time: 1.488, data_time: 0.074, memory: 49163, loss_cls_0: 0.7235, loss_box_0: 1.6346, loss_cns_0: 0.6321, loss_yns_0: 0.1463, loss_cls_1: 0.7994, loss_box_1: 1.5289, loss_cns_1: 0.6575, loss_yns_1: 0.1455, loss_cls_2: 0.8263, loss_box_2: 1.4972, loss_cns_2: 0.6565, loss_yns_2: 0.1461, loss_cls_3: 0.8098, loss_box_3: 1.5076, loss_cns_3: 0.6615, loss_yns_3: 0.1455, loss_cls_4: 0.8150, loss_box_4: 1.5022, loss_cns_4: 0.6551, loss_yns_4: 0.1447, loss_cls_5: 0.8117, loss_box_5: 1.5114, loss_cns_5: 0.6578, loss_yns_5: 0.1460, loss_cls_dn_0: 0.1270, loss_box_dn_0: 0.7203, loss_cls_dn_1: 0.1027, loss_box_dn_1: 0.6882, loss_cls_dn_2: 0.1024, loss_box_dn_2: 0.6864, loss_cls_dn_3: 0.1017, loss_box_dn_3: 0.6937, loss_cls_dn_4: 0.1038, loss_box_dn_4: 0.7040, loss_cls_dn_5: 0.1044, loss_box_dn_5: 0.7132, loss_dense_depth: 0.6719, loss: 24.2820, grad_norm: 43.4070 -2025-11-17 14:28:28,055 - mmdet - INFO - Iter [326/17500] lr: 2.298e-04, eta: 9:09:48, time: 1.503, data_time: 0.081, memory: 49163, loss_cls_0: 0.7224, loss_box_0: 1.6350, loss_cns_0: 0.6317, loss_yns_0: 0.1449, loss_cls_1: 0.7996, loss_box_1: 1.5305, loss_cns_1: 0.6594, loss_yns_1: 0.1453, loss_cls_2: 0.8259, loss_box_2: 1.4897, loss_cns_2: 0.6600, loss_yns_2: 0.1453, loss_cls_3: 0.8060, loss_box_3: 1.4979, loss_cns_3: 0.6633, loss_yns_3: 0.1458, loss_cls_4: 0.8024, loss_box_4: 1.4979, loss_cns_4: 0.6616, loss_yns_4: 0.1462, loss_cls_5: 0.8006, loss_box_5: 1.4972, loss_cns_5: 0.6603, loss_yns_5: 0.1455, loss_cls_dn_0: 0.1237, loss_box_dn_0: 0.7146, loss_cls_dn_1: 0.0997, loss_box_dn_1: 0.6743, loss_cls_dn_2: 0.1015, loss_box_dn_2: 0.6622, loss_cls_dn_3: 0.0987, loss_box_dn_3: 0.6624, loss_cls_dn_4: 0.1008, loss_box_dn_4: 0.6653, loss_cls_dn_5: 0.1010, loss_box_dn_5: 0.6690, loss_dense_depth: 0.6618, loss: 24.0494, grad_norm: 39.6590 -2025-11-17 14:28:29,565 - mmdet - INFO - Iter [327/17500] lr: 2.302e-04, eta: 9:09:24, time: 1.512, data_time: 0.084, memory: 49163, loss_cls_0: 0.7314, loss_box_0: 1.6516, loss_cns_0: 0.6319, loss_yns_0: 0.1433, loss_cls_1: 0.8023, loss_box_1: 1.5389, loss_cns_1: 0.6591, loss_yns_1: 0.1426, loss_cls_2: 0.8132, loss_box_2: 1.5044, loss_cns_2: 0.6592, loss_yns_2: 0.1420, loss_cls_3: 0.8157, loss_box_3: 1.5002, loss_cns_3: 0.6600, loss_yns_3: 0.1418, loss_cls_4: 0.8179, loss_box_4: 1.5022, loss_cns_4: 0.6605, loss_yns_4: 0.1419, loss_cls_5: 0.8201, loss_box_5: 1.5069, loss_cns_5: 0.6620, loss_yns_5: 0.1421, loss_cls_dn_0: 0.1296, loss_box_dn_0: 0.7142, loss_cls_dn_1: 0.1022, loss_box_dn_1: 0.6483, loss_cls_dn_2: 0.1046, loss_box_dn_2: 0.6304, loss_cls_dn_3: 0.1011, loss_box_dn_3: 0.6275, loss_cls_dn_4: 0.1040, loss_box_dn_4: 0.6292, loss_cls_dn_5: 0.1049, loss_box_dn_5: 0.6313, loss_dense_depth: 0.6721, loss: 23.9908, grad_norm: 40.1181 -2025-11-17 14:28:31,059 - mmdet - INFO - Iter [328/17500] lr: 2.306e-04, eta: 9:09:00, time: 1.493, data_time: 0.085, memory: 49163, loss_cls_0: 0.7442, loss_box_0: 1.6602, loss_cns_0: 0.6369, loss_yns_0: 0.1460, loss_cls_1: 0.8179, loss_box_1: 1.5351, loss_cns_1: 0.6618, loss_yns_1: 0.1439, loss_cls_2: 0.8214, loss_box_2: 1.5083, loss_cns_2: 0.6612, loss_yns_2: 0.1444, loss_cls_3: 0.8256, loss_box_3: 1.4995, loss_cns_3: 0.6613, loss_yns_3: 0.1448, loss_cls_4: 0.8243, loss_box_4: 1.5003, loss_cns_4: 0.6617, loss_yns_4: 0.1450, loss_cls_5: 0.8200, loss_box_5: 1.4988, loss_cns_5: 0.6610, loss_yns_5: 0.1458, loss_cls_dn_0: 0.1267, loss_box_dn_0: 0.7212, loss_cls_dn_1: 0.1005, loss_box_dn_1: 0.6375, loss_cls_dn_2: 0.1019, loss_box_dn_2: 0.6232, loss_cls_dn_3: 0.0994, loss_box_dn_3: 0.6215, loss_cls_dn_4: 0.1024, loss_box_dn_4: 0.6235, loss_cls_dn_5: 0.1028, loss_box_dn_5: 0.6265, loss_dense_depth: 0.6803, loss: 24.0372, grad_norm: 35.5364 -2025-11-17 14:28:32,625 - mmdet - INFO - Iter [329/17500] lr: 2.310e-04, eta: 9:08:40, time: 1.566, data_time: 0.174, memory: 49163, loss_cls_0: 0.7400, loss_box_0: 1.6620, loss_cns_0: 0.6331, loss_yns_0: 0.1450, loss_cls_1: 0.8063, loss_box_1: 1.5786, loss_cns_1: 0.6595, loss_yns_1: 0.1434, loss_cls_2: 0.8122, loss_box_2: 1.5386, loss_cns_2: 0.6595, loss_yns_2: 0.1436, loss_cls_3: 0.8168, loss_box_3: 1.5250, loss_cns_3: 0.6599, loss_yns_3: 0.1429, loss_cls_4: 0.8222, loss_box_4: 1.5220, loss_cns_4: 0.6603, loss_yns_4: 0.1432, loss_cls_5: 0.8202, loss_box_5: 1.5267, loss_cns_5: 0.6589, loss_yns_5: 0.1437, loss_cls_dn_0: 0.1315, loss_box_dn_0: 0.7234, loss_cls_dn_1: 0.1033, loss_box_dn_1: 0.6452, loss_cls_dn_2: 0.1027, loss_box_dn_2: 0.6305, loss_cls_dn_3: 0.1029, loss_box_dn_3: 0.6266, loss_cls_dn_4: 0.1058, loss_box_dn_4: 0.6282, loss_cls_dn_5: 0.1057, loss_box_dn_5: 0.6347, loss_dense_depth: 0.6990, loss: 24.2031, grad_norm: 39.8850 -2025-11-17 14:28:34,117 - mmdet - INFO - Iter [330/17500] lr: 2.314e-04, eta: 9:08:16, time: 1.493, data_time: 0.080, memory: 49163, loss_cls_0: 0.7290, loss_box_0: 1.6299, loss_cns_0: 0.6340, loss_yns_0: 0.1477, loss_cls_1: 0.7932, loss_box_1: 1.5382, loss_cns_1: 0.6622, loss_yns_1: 0.1443, loss_cls_2: 0.8073, loss_box_2: 1.4955, loss_cns_2: 0.6617, loss_yns_2: 0.1447, loss_cls_3: 0.8095, loss_box_3: 1.4859, loss_cns_3: 0.6633, loss_yns_3: 0.1442, loss_cls_4: 0.8176, loss_box_4: 1.4846, loss_cns_4: 0.6632, loss_yns_4: 0.1445, loss_cls_5: 0.8141, loss_box_5: 1.4871, loss_cns_5: 0.6622, loss_yns_5: 0.1441, loss_cls_dn_0: 0.1335, loss_box_dn_0: 0.7269, loss_cls_dn_1: 0.1018, loss_box_dn_1: 0.6545, loss_cls_dn_2: 0.0995, loss_box_dn_2: 0.6377, loss_cls_dn_3: 0.1003, loss_box_dn_3: 0.6376, loss_cls_dn_4: 0.1032, loss_box_dn_4: 0.6426, loss_cls_dn_5: 0.1024, loss_box_dn_5: 0.6495, loss_dense_depth: 0.6835, loss: 23.9810, grad_norm: 34.4776 -2025-11-17 14:28:35,628 - mmdet - INFO - Iter [331/17500] lr: 2.318e-04, eta: 9:07:53, time: 1.511, data_time: 0.073, memory: 49163, loss_cls_0: 0.7346, loss_box_0: 1.6309, loss_cns_0: 0.6260, loss_yns_0: 0.1439, loss_cls_1: 0.7984, loss_box_1: 1.5296, loss_cns_1: 0.6570, loss_yns_1: 0.1437, loss_cls_2: 0.8104, loss_box_2: 1.5065, loss_cns_2: 0.6588, loss_yns_2: 0.1431, loss_cls_3: 0.8089, loss_box_3: 1.5086, loss_cns_3: 0.6600, loss_yns_3: 0.1430, loss_cls_4: 0.8156, loss_box_4: 1.5112, loss_cns_4: 0.6607, loss_yns_4: 0.1437, loss_cls_5: 0.8115, loss_box_5: 1.5104, loss_cns_5: 0.6594, loss_yns_5: 0.1426, loss_cls_dn_0: 0.1314, loss_box_dn_0: 0.7174, loss_cls_dn_1: 0.1029, loss_box_dn_1: 0.6526, loss_cls_dn_2: 0.1000, loss_box_dn_2: 0.6399, loss_cls_dn_3: 0.1012, loss_box_dn_3: 0.6459, loss_cls_dn_4: 0.1035, loss_box_dn_4: 0.6538, loss_cls_dn_5: 0.1046, loss_box_dn_5: 0.6583, loss_dense_depth: 0.6891, loss: 24.0592, grad_norm: 35.3012 -2025-11-17 14:28:37,127 - mmdet - INFO - Iter [332/17500] lr: 2.322e-04, eta: 9:07:29, time: 1.498, data_time: 0.081, memory: 49163, loss_cls_0: 0.7282, loss_box_0: 1.6313, loss_cns_0: 0.6328, loss_yns_0: 0.1467, loss_cls_1: 0.7933, loss_box_1: 1.5091, loss_cns_1: 0.6607, loss_yns_1: 0.1438, loss_cls_2: 0.8080, loss_box_2: 1.4782, loss_cns_2: 0.6637, loss_yns_2: 0.1440, loss_cls_3: 0.8080, loss_box_3: 1.4802, loss_cns_3: 0.6640, loss_yns_3: 0.1442, loss_cls_4: 0.8095, loss_box_4: 1.4834, loss_cns_4: 0.6643, loss_yns_4: 0.1444, loss_cls_5: 0.8161, loss_box_5: 1.4826, loss_cns_5: 0.6621, loss_yns_5: 0.1443, loss_cls_dn_0: 0.1312, loss_box_dn_0: 0.7193, loss_cls_dn_1: 0.1047, loss_box_dn_1: 0.6642, loss_cls_dn_2: 0.1014, loss_box_dn_2: 0.6517, loss_cls_dn_3: 0.1019, loss_box_dn_3: 0.6547, loss_cls_dn_4: 0.1035, loss_box_dn_4: 0.6618, loss_cls_dn_5: 0.1061, loss_box_dn_5: 0.6684, loss_dense_depth: 0.6750, loss: 23.9866, grad_norm: 38.7434 -2025-11-17 14:28:38,615 - mmdet - INFO - Iter [333/17500] lr: 2.326e-04, eta: 9:07:05, time: 1.488, data_time: 0.078, memory: 49163, loss_cls_0: 0.7580, loss_box_0: 1.6501, loss_cns_0: 0.6293, loss_yns_0: 0.1447, loss_cls_1: 0.8138, loss_box_1: 1.5022, loss_cns_1: 0.6573, loss_yns_1: 0.1426, loss_cls_2: 0.8167, loss_box_2: 1.4747, loss_cns_2: 0.6593, loss_yns_2: 0.1414, loss_cls_3: 0.8236, loss_box_3: 1.4542, loss_cns_3: 0.6600, loss_yns_3: 0.1422, loss_cls_4: 0.8302, loss_box_4: 1.4640, loss_cns_4: 0.6637, loss_yns_4: 0.1413, loss_cls_5: 0.8411, loss_box_5: 1.4568, loss_cns_5: 0.6604, loss_yns_5: 0.1411, loss_cls_dn_0: 0.1388, loss_box_dn_0: 0.7271, loss_cls_dn_1: 0.1030, loss_box_dn_1: 0.6683, loss_cls_dn_2: 0.1002, loss_box_dn_2: 0.6559, loss_cls_dn_3: 0.1006, loss_box_dn_3: 0.6520, loss_cls_dn_4: 0.1029, loss_box_dn_4: 0.6537, loss_cls_dn_5: 0.1053, loss_box_dn_5: 0.6593, loss_dense_depth: 0.7223, loss: 24.0581, grad_norm: 31.4260 -2025-11-17 14:28:40,167 - mmdet - INFO - Iter [334/17500] lr: 2.330e-04, eta: 9:06:45, time: 1.552, data_time: 0.079, memory: 49163, loss_cls_0: 0.7631, loss_box_0: 1.6385, loss_cns_0: 0.6286, loss_yns_0: 0.1457, loss_cls_1: 0.8243, loss_box_1: 1.5160, loss_cns_1: 0.6580, loss_yns_1: 0.1435, loss_cls_2: 0.8346, loss_box_2: 1.4852, loss_cns_2: 0.6564, loss_yns_2: 0.1426, loss_cls_3: 0.8344, loss_box_3: 1.4748, loss_cns_3: 0.6580, loss_yns_3: 0.1422, loss_cls_4: 0.8396, loss_box_4: 1.4746, loss_cns_4: 0.6584, loss_yns_4: 0.1415, loss_cls_5: 0.8400, loss_box_5: 1.4737, loss_cns_5: 0.6581, loss_yns_5: 0.1426, loss_cls_dn_0: 0.1347, loss_box_dn_0: 0.7197, loss_cls_dn_1: 0.1050, loss_box_dn_1: 0.6715, loss_cls_dn_2: 0.1043, loss_box_dn_2: 0.6580, loss_cls_dn_3: 0.1050, loss_box_dn_3: 0.6544, loss_cls_dn_4: 0.1058, loss_box_dn_4: 0.6531, loss_cls_dn_5: 0.1083, loss_box_dn_5: 0.6547, loss_dense_depth: 0.7044, loss: 24.1533, grad_norm: 32.8841 -2025-11-17 14:28:41,677 - mmdet - INFO - Iter [335/17500] lr: 2.334e-04, eta: 9:06:23, time: 1.511, data_time: 0.081, memory: 49163, loss_cls_0: 0.7591, loss_box_0: 1.6378, loss_cns_0: 0.6303, loss_yns_0: 0.1478, loss_cls_1: 0.8144, loss_box_1: 1.5327, loss_cns_1: 0.6537, loss_yns_1: 0.1429, loss_cls_2: 0.8229, loss_box_2: 1.4959, loss_cns_2: 0.6537, loss_yns_2: 0.1427, loss_cls_3: 0.8164, loss_box_3: 1.4941, loss_cns_3: 0.6572, loss_yns_3: 0.1428, loss_cls_4: 0.8243, loss_box_4: 1.4962, loss_cns_4: 0.6573, loss_yns_4: 0.1430, loss_cls_5: 0.8263, loss_box_5: 1.5022, loss_cns_5: 0.6571, loss_yns_5: 0.1432, loss_cls_dn_0: 0.1362, loss_box_dn_0: 0.7203, loss_cls_dn_1: 0.1044, loss_box_dn_1: 0.6681, loss_cls_dn_2: 0.1042, loss_box_dn_2: 0.6510, loss_cls_dn_3: 0.1026, loss_box_dn_3: 0.6455, loss_cls_dn_4: 0.1034, loss_box_dn_4: 0.6447, loss_cls_dn_5: 0.1046, loss_box_dn_5: 0.6455, loss_dense_depth: 0.6970, loss: 24.1214, grad_norm: 28.0009 -2025-11-17 14:28:43,163 - mmdet - INFO - Iter [336/17500] lr: 2.338e-04, eta: 9:05:59, time: 1.485, data_time: 0.078, memory: 49163, loss_cls_0: 0.7576, loss_box_0: 1.6365, loss_cns_0: 0.6259, loss_yns_0: 0.1456, loss_cls_1: 0.8232, loss_box_1: 1.5681, loss_cns_1: 0.6535, loss_yns_1: 0.1425, loss_cls_2: 0.8228, loss_box_2: 1.5402, loss_cns_2: 0.6541, loss_yns_2: 0.1431, loss_cls_3: 0.8236, loss_box_3: 1.5332, loss_cns_3: 0.6567, loss_yns_3: 0.1414, loss_cls_4: 0.8265, loss_box_4: 1.5364, loss_cns_4: 0.6574, loss_yns_4: 0.1422, loss_cls_5: 0.8365, loss_box_5: 1.5369, loss_cns_5: 0.6556, loss_yns_5: 0.1430, loss_cls_dn_0: 0.1376, loss_box_dn_0: 0.7188, loss_cls_dn_1: 0.1026, loss_box_dn_1: 0.6637, loss_cls_dn_2: 0.1017, loss_box_dn_2: 0.6478, loss_cls_dn_3: 0.1013, loss_box_dn_3: 0.6435, loss_cls_dn_4: 0.1024, loss_box_dn_4: 0.6441, loss_cls_dn_5: 0.1038, loss_box_dn_5: 0.6467, loss_dense_depth: 0.6865, loss: 24.3028, grad_norm: 29.0790 -2025-11-17 14:28:44,709 - mmdet - INFO - Iter [337/17500] lr: 2.342e-04, eta: 9:05:39, time: 1.545, data_time: 0.077, memory: 49163, loss_cls_0: 0.7553, loss_box_0: 1.6358, loss_cns_0: 0.6279, loss_yns_0: 0.1450, loss_cls_1: 0.8247, loss_box_1: 1.5489, loss_cns_1: 0.6581, loss_yns_1: 0.1426, loss_cls_2: 0.8242, loss_box_2: 1.5251, loss_cns_2: 0.6546, loss_yns_2: 0.1415, loss_cls_3: 0.8332, loss_box_3: 1.5165, loss_cns_3: 0.6551, loss_yns_3: 0.1412, loss_cls_4: 0.8305, loss_box_4: 1.5178, loss_cns_4: 0.6574, loss_yns_4: 0.1408, loss_cls_5: 0.8361, loss_box_5: 1.5261, loss_cns_5: 0.6567, loss_yns_5: 0.1403, loss_cls_dn_0: 0.1349, loss_box_dn_0: 0.7151, loss_cls_dn_1: 0.1063, loss_box_dn_1: 0.6497, loss_cls_dn_2: 0.1055, loss_box_dn_2: 0.6375, loss_cls_dn_3: 0.1058, loss_box_dn_3: 0.6374, loss_cls_dn_4: 0.1072, loss_box_dn_4: 0.6414, loss_cls_dn_5: 0.1091, loss_box_dn_5: 0.6523, loss_dense_depth: 0.6893, loss: 24.2268, grad_norm: 36.3720 -2025-11-17 14:28:46,190 - mmdet - INFO - Iter [338/17500] lr: 2.346e-04, eta: 9:05:15, time: 1.482, data_time: 0.077, memory: 49163, loss_cls_0: 0.7523, loss_box_0: 1.6224, loss_cns_0: 0.6344, loss_yns_0: 0.1429, loss_cls_1: 0.8231, loss_box_1: 1.5160, loss_cns_1: 0.6593, loss_yns_1: 0.1397, loss_cls_2: 0.8278, loss_box_2: 1.5087, loss_cns_2: 0.6595, loss_yns_2: 0.1399, loss_cls_3: 0.8297, loss_box_3: 1.4934, loss_cns_3: 0.6601, loss_yns_3: 0.1377, loss_cls_4: 0.8299, loss_box_4: 1.4918, loss_cns_4: 0.6601, loss_yns_4: 0.1385, loss_cls_5: 0.8321, loss_box_5: 1.4937, loss_cns_5: 0.6595, loss_yns_5: 0.1390, loss_cls_dn_0: 0.1331, loss_box_dn_0: 0.7201, loss_cls_dn_1: 0.1067, loss_box_dn_1: 0.6510, loss_cls_dn_2: 0.1059, loss_box_dn_2: 0.6435, loss_cls_dn_3: 0.1052, loss_box_dn_3: 0.6444, loss_cls_dn_4: 0.1065, loss_box_dn_4: 0.6478, loss_cls_dn_5: 0.1085, loss_box_dn_5: 0.6585, loss_dense_depth: 0.6905, loss: 24.1134, grad_norm: 31.8635 -2025-11-17 14:28:47,665 - mmdet - INFO - Iter [339/17500] lr: 2.350e-04, eta: 9:04:51, time: 1.476, data_time: 0.072, memory: 49163, loss_cls_0: 0.7431, loss_box_0: 1.6408, loss_cns_0: 0.6300, loss_yns_0: 0.1441, loss_cls_1: 0.8307, loss_box_1: 1.5125, loss_cns_1: 0.6607, loss_yns_1: 0.1419, loss_cls_2: 0.8347, loss_box_2: 1.5008, loss_cns_2: 0.6614, loss_yns_2: 0.1404, loss_cls_3: 0.8319, loss_box_3: 1.4764, loss_cns_3: 0.6603, loss_yns_3: 0.1377, loss_cls_4: 0.8369, loss_box_4: 1.4787, loss_cns_4: 0.6590, loss_yns_4: 0.1384, loss_cls_5: 0.8384, loss_box_5: 1.4789, loss_cns_5: 0.6595, loss_yns_5: 0.1379, loss_cls_dn_0: 0.1373, loss_box_dn_0: 0.7190, loss_cls_dn_1: 0.1059, loss_box_dn_1: 0.6572, loss_cls_dn_2: 0.1055, loss_box_dn_2: 0.6484, loss_cls_dn_3: 0.1037, loss_box_dn_3: 0.6452, loss_cls_dn_4: 0.1058, loss_box_dn_4: 0.6455, loss_cls_dn_5: 0.1073, loss_box_dn_5: 0.6491, loss_dense_depth: 0.7009, loss: 24.1060, grad_norm: 26.6232 -2025-11-17 14:28:49,158 - mmdet - INFO - Iter [340/17500] lr: 2.354e-04, eta: 9:04:29, time: 1.493, data_time: 0.079, memory: 49163, loss_cls_0: 0.7622, loss_box_0: 1.6437, loss_cns_0: 0.6320, loss_yns_0: 0.1470, loss_cls_1: 0.8248, loss_box_1: 1.5282, loss_cns_1: 0.6597, loss_yns_1: 0.1432, loss_cls_2: 0.8316, loss_box_2: 1.4938, loss_cns_2: 0.6631, loss_yns_2: 0.1443, loss_cls_3: 0.8374, loss_box_3: 1.4819, loss_cns_3: 0.6638, loss_yns_3: 0.1428, loss_cls_4: 0.8415, loss_box_4: 1.4774, loss_cns_4: 0.6646, loss_yns_4: 0.1428, loss_cls_5: 0.8439, loss_box_5: 1.4831, loss_cns_5: 0.6645, loss_yns_5: 0.1419, loss_cls_dn_0: 0.1369, loss_box_dn_0: 0.7216, loss_cls_dn_1: 0.1078, loss_box_dn_1: 0.6696, loss_cls_dn_2: 0.1073, loss_box_dn_2: 0.6545, loss_cls_dn_3: 0.1081, loss_box_dn_3: 0.6510, loss_cls_dn_4: 0.1085, loss_box_dn_4: 0.6471, loss_cls_dn_5: 0.1097, loss_box_dn_5: 0.6495, loss_dense_depth: 0.7042, loss: 24.2352, grad_norm: 31.9209 -2025-11-17 14:28:50,682 - mmdet - INFO - Iter [341/17500] lr: 2.358e-04, eta: 9:04:08, time: 1.522, data_time: 0.078, memory: 49163, loss_cls_0: 0.7756, loss_box_0: 1.6487, loss_cns_0: 0.6323, loss_yns_0: 0.1474, loss_cls_1: 0.8255, loss_box_1: 1.5904, loss_cns_1: 0.6550, loss_yns_1: 0.1449, loss_cls_2: 0.8300, loss_box_2: 1.5571, loss_cns_2: 0.6559, loss_yns_2: 0.1438, loss_cls_3: 0.8356, loss_box_3: 1.5483, loss_cns_3: 0.6548, loss_yns_3: 0.1432, loss_cls_4: 0.8364, loss_box_4: 1.5411, loss_cns_4: 0.6555, loss_yns_4: 0.1431, loss_cls_5: 0.8458, loss_box_5: 1.5440, loss_cns_5: 0.6566, loss_yns_5: 0.1438, loss_cls_dn_0: 0.1307, loss_box_dn_0: 0.7193, loss_cls_dn_1: 0.1046, loss_box_dn_1: 0.6604, loss_cls_dn_2: 0.1045, loss_box_dn_2: 0.6460, loss_cls_dn_3: 0.1050, loss_box_dn_3: 0.6441, loss_cls_dn_4: 0.1059, loss_box_dn_4: 0.6414, loss_cls_dn_5: 0.1084, loss_box_dn_5: 0.6431, loss_dense_depth: 0.7187, loss: 24.4871, grad_norm: 30.1464 -2025-11-17 14:28:52,258 - mmdet - INFO - Iter [342/17500] lr: 2.362e-04, eta: 9:03:49, time: 1.577, data_time: 0.075, memory: 49163, loss_cls_0: 0.7464, loss_box_0: 1.6207, loss_cns_0: 0.6283, loss_yns_0: 0.1456, loss_cls_1: 0.8086, loss_box_1: 1.5445, loss_cns_1: 0.6525, loss_yns_1: 0.1448, loss_cls_2: 0.8192, loss_box_2: 1.5225, loss_cns_2: 0.6524, loss_yns_2: 0.1431, loss_cls_3: 0.8177, loss_box_3: 1.5261, loss_cns_3: 0.6594, loss_yns_3: 0.1428, loss_cls_4: 0.8208, loss_box_4: 1.5246, loss_cns_4: 0.6543, loss_yns_4: 0.1424, loss_cls_5: 0.8261, loss_box_5: 1.5216, loss_cns_5: 0.6567, loss_yns_5: 0.1429, loss_cls_dn_0: 0.1267, loss_box_dn_0: 0.7207, loss_cls_dn_1: 0.1031, loss_box_dn_1: 0.6448, loss_cls_dn_2: 0.1015, loss_box_dn_2: 0.6339, loss_cls_dn_3: 0.1005, loss_box_dn_3: 0.6310, loss_cls_dn_4: 0.1010, loss_box_dn_4: 0.6310, loss_cls_dn_5: 0.1037, loss_box_dn_5: 0.6297, loss_dense_depth: 0.7068, loss: 24.0982, grad_norm: 30.6895 -2025-11-17 14:28:53,753 - mmdet - INFO - Iter [343/17500] lr: 2.366e-04, eta: 9:03:27, time: 1.495, data_time: 0.078, memory: 49163, loss_cls_0: 0.7620, loss_box_0: 1.6403, loss_cns_0: 0.6256, loss_yns_0: 0.1462, loss_cls_1: 0.8225, loss_box_1: 1.5848, loss_cns_1: 0.6506, loss_yns_1: 0.1440, loss_cls_2: 0.8375, loss_box_2: 1.5501, loss_cns_2: 0.6509, loss_yns_2: 0.1418, loss_cls_3: 0.8352, loss_box_3: 1.5508, loss_cns_3: 0.6561, loss_yns_3: 0.1427, loss_cls_4: 0.8381, loss_box_4: 1.5532, loss_cns_4: 0.6542, loss_yns_4: 0.1426, loss_cls_5: 0.8461, loss_box_5: 1.5477, loss_cns_5: 0.6544, loss_yns_5: 0.1429, loss_cls_dn_0: 0.1292, loss_box_dn_0: 0.7108, loss_cls_dn_1: 0.1013, loss_box_dn_1: 0.6444, loss_cls_dn_2: 0.1003, loss_box_dn_2: 0.6338, loss_cls_dn_3: 0.1002, loss_box_dn_3: 0.6315, loss_cls_dn_4: 0.1019, loss_box_dn_4: 0.6332, loss_cls_dn_5: 0.1053, loss_box_dn_5: 0.6335, loss_dense_depth: 0.6996, loss: 24.3454, grad_norm: 32.7682 -2025-11-17 14:28:55,302 - mmdet - INFO - Iter [344/17500] lr: 2.370e-04, eta: 9:03:08, time: 1.549, data_time: 0.081, memory: 49163, loss_cls_0: 0.7757, loss_box_0: 1.6471, loss_cns_0: 0.6233, loss_yns_0: 0.1447, loss_cls_1: 0.8221, loss_box_1: 1.5657, loss_cns_1: 0.6550, loss_yns_1: 0.1419, loss_cls_2: 0.8359, loss_box_2: 1.5272, loss_cns_2: 0.6553, loss_yns_2: 0.1407, loss_cls_3: 0.8384, loss_box_3: 1.5269, loss_cns_3: 0.6559, loss_yns_3: 0.1396, loss_cls_4: 0.8413, loss_box_4: 1.5298, loss_cns_4: 0.6557, loss_yns_4: 0.1406, loss_cls_5: 0.8459, loss_box_5: 1.5232, loss_cns_5: 0.6566, loss_yns_5: 0.1393, loss_cls_dn_0: 0.1298, loss_box_dn_0: 0.7175, loss_cls_dn_1: 0.1019, loss_box_dn_1: 0.6437, loss_cls_dn_2: 0.1027, loss_box_dn_2: 0.6342, loss_cls_dn_3: 0.1041, loss_box_dn_3: 0.6356, loss_cls_dn_4: 0.1038, loss_box_dn_4: 0.6398, loss_cls_dn_5: 0.1059, loss_box_dn_5: 0.6460, loss_dense_depth: 0.7297, loss: 24.3223, grad_norm: 32.5147 -2025-11-17 14:28:56,809 - mmdet - INFO - Iter [345/17500] lr: 2.374e-04, eta: 9:02:46, time: 1.508, data_time: 0.078, memory: 49163, loss_cls_0: 0.7416, loss_box_0: 1.6545, loss_cns_0: 0.6301, loss_yns_0: 0.1423, loss_cls_1: 0.8010, loss_box_1: 1.5667, loss_cns_1: 0.6525, loss_yns_1: 0.1408, loss_cls_2: 0.8095, loss_box_2: 1.5433, loss_cns_2: 0.6559, loss_yns_2: 0.1397, loss_cls_3: 0.8164, loss_box_3: 1.5337, loss_cns_3: 0.6545, loss_yns_3: 0.1383, loss_cls_4: 0.8156, loss_box_4: 1.5287, loss_cns_4: 0.6539, loss_yns_4: 0.1381, loss_cls_5: 0.8200, loss_box_5: 1.5356, loss_cns_5: 0.6565, loss_yns_5: 0.1379, loss_cls_dn_0: 0.1234, loss_box_dn_0: 0.7095, loss_cls_dn_1: 0.1022, loss_box_dn_1: 0.6463, loss_cls_dn_2: 0.1012, loss_box_dn_2: 0.6380, loss_cls_dn_3: 0.1012, loss_box_dn_3: 0.6369, loss_cls_dn_4: 0.1008, loss_box_dn_4: 0.6421, loss_cls_dn_5: 0.1019, loss_box_dn_5: 0.6498, loss_dense_depth: 0.6888, loss: 24.1495, grad_norm: 39.1470 -2025-11-17 14:28:58,314 - mmdet - INFO - Iter [346/17500] lr: 2.378e-04, eta: 9:02:25, time: 1.505, data_time: 0.077, memory: 49163, loss_cls_0: 0.7591, loss_box_0: 1.6521, loss_cns_0: 0.6310, loss_yns_0: 0.1425, loss_cls_1: 0.8008, loss_box_1: 1.5218, loss_cns_1: 0.6575, loss_yns_1: 0.1384, loss_cls_2: 0.8143, loss_box_2: 1.4917, loss_cns_2: 0.6614, loss_yns_2: 0.1384, loss_cls_3: 0.8062, loss_box_3: 1.4892, loss_cns_3: 0.6626, loss_yns_3: 0.1386, loss_cls_4: 0.8087, loss_box_4: 1.4956, loss_cns_4: 0.6610, loss_yns_4: 0.1387, loss_cls_5: 0.8129, loss_box_5: 1.4915, loss_cns_5: 0.6607, loss_yns_5: 0.1386, loss_cls_dn_0: 0.1272, loss_box_dn_0: 0.7195, loss_cls_dn_1: 0.1026, loss_box_dn_1: 0.6536, loss_cls_dn_2: 0.1014, loss_box_dn_2: 0.6437, loss_cls_dn_3: 0.1010, loss_box_dn_3: 0.6462, loss_cls_dn_4: 0.1022, loss_box_dn_4: 0.6535, loss_cls_dn_5: 0.1035, loss_box_dn_5: 0.6594, loss_dense_depth: 0.7058, loss: 24.0333, grad_norm: 31.7939 -2025-11-17 14:28:59,858 - mmdet - INFO - Iter [347/17500] lr: 2.382e-04, eta: 9:02:05, time: 1.542, data_time: 0.083, memory: 49163, loss_cls_0: 0.7250, loss_box_0: 1.6204, loss_cns_0: 0.6336, loss_yns_0: 0.1432, loss_cls_1: 0.7899, loss_box_1: 1.4877, loss_cns_1: 0.6610, loss_yns_1: 0.1400, loss_cls_2: 0.8054, loss_box_2: 1.4770, loss_cns_2: 0.6627, loss_yns_2: 0.1402, loss_cls_3: 0.8064, loss_box_3: 1.4736, loss_cns_3: 0.6725, loss_yns_3: 0.1406, loss_cls_4: 0.8157, loss_box_4: 1.4679, loss_cns_4: 0.6708, loss_yns_4: 0.1388, loss_cls_5: 0.8089, loss_box_5: 1.4634, loss_cns_5: 0.6677, loss_yns_5: 0.1395, loss_cls_dn_0: 0.1235, loss_box_dn_0: 0.7092, loss_cls_dn_1: 0.1006, loss_box_dn_1: 0.6541, loss_cls_dn_2: 0.0983, loss_box_dn_2: 0.6483, loss_cls_dn_3: 0.0977, loss_box_dn_3: 0.6503, loss_cls_dn_4: 0.1019, loss_box_dn_4: 0.6528, loss_cls_dn_5: 0.1016, loss_box_dn_5: 0.6522, loss_dense_depth: 0.6853, loss: 23.8277, grad_norm: 47.9029 -2025-11-17 14:29:01,345 - mmdet - INFO - Iter [348/17500] lr: 2.386e-04, eta: 9:01:43, time: 1.488, data_time: 0.083, memory: 49163, loss_cls_0: 0.7519, loss_box_0: 1.6134, loss_cns_0: 0.6328, loss_yns_0: 0.1444, loss_cls_1: 0.8150, loss_box_1: 1.4828, loss_cns_1: 0.6625, loss_yns_1: 0.1409, loss_cls_2: 0.8257, loss_box_2: 1.4518, loss_cns_2: 0.6651, loss_yns_2: 0.1407, loss_cls_3: 0.8284, loss_box_3: 1.4428, loss_cns_3: 0.6716, loss_yns_3: 0.1414, loss_cls_4: 0.8319, loss_box_4: 1.4414, loss_cns_4: 0.6687, loss_yns_4: 0.1401, loss_cls_5: 0.8247, loss_box_5: 1.4584, loss_cns_5: 0.6666, loss_yns_5: 0.1402, loss_cls_dn_0: 0.1240, loss_box_dn_0: 0.7120, loss_cls_dn_1: 0.1009, loss_box_dn_1: 0.6597, loss_cls_dn_2: 0.0990, loss_box_dn_2: 0.6442, loss_cls_dn_3: 0.0993, loss_box_dn_3: 0.6418, loss_cls_dn_4: 0.1020, loss_box_dn_4: 0.6427, loss_cls_dn_5: 0.1011, loss_box_dn_5: 0.6477, loss_dense_depth: 0.7132, loss: 23.8704, grad_norm: 24.7358 -2025-11-17 14:29:02,924 - mmdet - INFO - Iter [349/17500] lr: 2.390e-04, eta: 9:01:26, time: 1.579, data_time: 0.173, memory: 49163, loss_cls_0: 0.7351, loss_box_0: 1.6314, loss_cns_0: 0.6316, loss_yns_0: 0.1440, loss_cls_1: 0.8057, loss_box_1: 1.4805, loss_cns_1: 0.6603, loss_yns_1: 0.1408, loss_cls_2: 0.8046, loss_box_2: 1.4566, loss_cns_2: 0.6628, loss_yns_2: 0.1410, loss_cls_3: 0.8073, loss_box_3: 1.4433, loss_cns_3: 0.6620, loss_yns_3: 0.1388, loss_cls_4: 0.8173, loss_box_4: 1.4448, loss_cns_4: 0.6601, loss_yns_4: 0.1393, loss_cls_5: 0.8195, loss_box_5: 1.4707, loss_cns_5: 0.6602, loss_yns_5: 0.1390, loss_cls_dn_0: 0.1234, loss_box_dn_0: 0.7110, loss_cls_dn_1: 0.0990, loss_box_dn_1: 0.6593, loss_cls_dn_2: 0.0972, loss_box_dn_2: 0.6407, loss_cls_dn_3: 0.0970, loss_box_dn_3: 0.6353, loss_cls_dn_4: 0.0986, loss_box_dn_4: 0.6357, loss_cls_dn_5: 0.0985, loss_box_dn_5: 0.6456, loss_dense_depth: 0.7198, loss: 23.7582, grad_norm: 51.3004 -2025-11-17 14:29:04,438 - mmdet - INFO - Iter [350/17500] lr: 2.394e-04, eta: 9:01:06, time: 1.514, data_time: 0.078, memory: 49163, loss_cls_0: 0.7140, loss_box_0: 1.6013, loss_cns_0: 0.6365, loss_yns_0: 0.1440, loss_cls_1: 0.7901, loss_box_1: 1.4530, loss_cns_1: 0.6591, loss_yns_1: 0.1410, loss_cls_2: 0.7914, loss_box_2: 1.4173, loss_cns_2: 0.6585, loss_yns_2: 0.1400, loss_cls_3: 0.7954, loss_box_3: 1.4155, loss_cns_3: 0.6623, loss_yns_3: 0.1397, loss_cls_4: 0.8071, loss_box_4: 1.4026, loss_cns_4: 0.6636, loss_yns_4: 0.1397, loss_cls_5: 0.8047, loss_box_5: 1.4126, loss_cns_5: 0.6608, loss_yns_5: 0.1402, loss_cls_dn_0: 0.1206, loss_box_dn_0: 0.7113, loss_cls_dn_1: 0.0998, loss_box_dn_1: 0.6447, loss_cls_dn_2: 0.0997, loss_box_dn_2: 0.6289, loss_cls_dn_3: 0.0982, loss_box_dn_3: 0.6254, loss_cls_dn_4: 0.0998, loss_box_dn_4: 0.6219, loss_cls_dn_5: 0.0993, loss_box_dn_5: 0.6259, loss_dense_depth: 0.6965, loss: 23.3626, grad_norm: 32.2100 -2025-11-17 14:29:05,941 - mmdet - INFO - Iter [351/17500] lr: 2.398e-04, eta: 9:00:45, time: 1.504, data_time: 0.078, memory: 49163, loss_cls_0: 0.7372, loss_box_0: 1.6486, loss_cns_0: 0.6317, loss_yns_0: 0.1485, loss_cls_1: 0.7949, loss_box_1: 1.5030, loss_cns_1: 0.6558, loss_yns_1: 0.1446, loss_cls_2: 0.8022, loss_box_2: 1.4510, loss_cns_2: 0.6580, loss_yns_2: 0.1462, loss_cls_3: 0.8193, loss_box_3: 1.4416, loss_cns_3: 0.6616, loss_yns_3: 0.1439, loss_cls_4: 0.8269, loss_box_4: 1.4472, loss_cns_4: 0.6635, loss_yns_4: 0.1448, loss_cls_5: 0.8294, loss_box_5: 1.4491, loss_cns_5: 0.6603, loss_yns_5: 0.1445, loss_cls_dn_0: 0.1242, loss_box_dn_0: 0.7189, loss_cls_dn_1: 0.1003, loss_box_dn_1: 0.6418, loss_cls_dn_2: 0.1012, loss_box_dn_2: 0.6311, loss_cls_dn_3: 0.0995, loss_box_dn_3: 0.6274, loss_cls_dn_4: 0.1017, loss_box_dn_4: 0.6290, loss_cls_dn_5: 0.1015, loss_box_dn_5: 0.6327, loss_dense_depth: 0.7472, loss: 23.8103, grad_norm: 59.0076 -2025-11-17 14:29:07,440 - mmdet - INFO - Iter [352/17500] lr: 2.402e-04, eta: 9:00:24, time: 1.498, data_time: 0.078, memory: 49163, loss_cls_0: 0.7479, loss_box_0: 1.6334, loss_cns_0: 0.6319, loss_yns_0: 0.1493, loss_cls_1: 0.8027, loss_box_1: 1.5439, loss_cns_1: 0.6534, loss_yns_1: 0.1455, loss_cls_2: 0.8109, loss_box_2: 1.4732, loss_cns_2: 0.6553, loss_yns_2: 0.1465, loss_cls_3: 0.8202, loss_box_3: 1.4479, loss_cns_3: 0.6606, loss_yns_3: 0.1459, loss_cls_4: 0.8307, loss_box_4: 1.4528, loss_cns_4: 0.6608, loss_yns_4: 0.1451, loss_cls_5: 0.8238, loss_box_5: 1.4542, loss_cns_5: 0.6574, loss_yns_5: 0.1452, loss_cls_dn_0: 0.1245, loss_box_dn_0: 0.7153, loss_cls_dn_1: 0.0995, loss_box_dn_1: 0.6470, loss_cls_dn_2: 0.1002, loss_box_dn_2: 0.6365, loss_cls_dn_3: 0.0992, loss_box_dn_3: 0.6288, loss_cls_dn_4: 0.1027, loss_box_dn_4: 0.6314, loss_cls_dn_5: 0.1024, loss_box_dn_5: 0.6364, loss_dense_depth: 0.7486, loss: 23.9110, grad_norm: 50.9620 -2025-11-17 14:29:08,937 - mmdet - INFO - Iter [353/17500] lr: 2.406e-04, eta: 9:00:03, time: 1.497, data_time: 0.079, memory: 49163, loss_cls_0: 0.7390, loss_box_0: 1.6322, loss_cns_0: 0.6308, loss_yns_0: 0.1488, loss_cls_1: 0.8167, loss_box_1: 1.4539, loss_cns_1: 0.6609, loss_yns_1: 0.1443, loss_cls_2: 0.8140, loss_box_2: 1.4251, loss_cns_2: 0.6583, loss_yns_2: 0.1446, loss_cls_3: 0.8172, loss_box_3: 1.4162, loss_cns_3: 0.6627, loss_yns_3: 0.1445, loss_cls_4: 0.8372, loss_box_4: 1.4262, loss_cns_4: 0.6641, loss_yns_4: 0.1460, loss_cls_5: 0.8238, loss_box_5: 1.4274, loss_cns_5: 0.6616, loss_yns_5: 0.1459, loss_cls_dn_0: 0.1245, loss_box_dn_0: 0.7280, loss_cls_dn_1: 0.0990, loss_box_dn_1: 0.6450, loss_cls_dn_2: 0.0995, loss_box_dn_2: 0.6367, loss_cls_dn_3: 0.0992, loss_box_dn_3: 0.6330, loss_cls_dn_4: 0.1029, loss_box_dn_4: 0.6384, loss_cls_dn_5: 0.1021, loss_box_dn_5: 0.6439, loss_dense_depth: 0.7030, loss: 23.6964, grad_norm: 47.4328 -2025-11-17 14:29:10,480 - mmdet - INFO - Iter [354/17500] lr: 2.410e-04, eta: 8:59:44, time: 1.543, data_time: 0.078, memory: 49163, loss_cls_0: 0.7451, loss_box_0: 1.6369, loss_cns_0: 0.6281, loss_yns_0: 0.1464, loss_cls_1: 0.8156, loss_box_1: 1.4940, loss_cns_1: 0.6582, loss_yns_1: 0.1446, loss_cls_2: 0.8188, loss_box_2: 1.4668, loss_cns_2: 0.6581, loss_yns_2: 0.1430, loss_cls_3: 0.8252, loss_box_3: 1.4602, loss_cns_3: 0.6605, loss_yns_3: 0.1427, loss_cls_4: 0.8335, loss_box_4: 1.4680, loss_cns_4: 0.6591, loss_yns_4: 0.1431, loss_cls_5: 0.8309, loss_box_5: 1.4698, loss_cns_5: 0.6579, loss_yns_5: 0.1435, loss_cls_dn_0: 0.1248, loss_box_dn_0: 0.7256, loss_cls_dn_1: 0.0978, loss_box_dn_1: 0.6560, loss_cls_dn_2: 0.0983, loss_box_dn_2: 0.6436, loss_cls_dn_3: 0.0973, loss_box_dn_3: 0.6424, loss_cls_dn_4: 0.0993, loss_box_dn_4: 0.6491, loss_cls_dn_5: 0.1001, loss_box_dn_5: 0.6536, loss_dense_depth: 0.7322, loss: 23.9701, grad_norm: 56.7515 -2025-11-17 14:29:11,990 - mmdet - INFO - Iter [355/17500] lr: 2.414e-04, eta: 8:59:24, time: 1.509, data_time: 0.078, memory: 49163, loss_cls_0: 0.7524, loss_box_0: 1.6202, loss_cns_0: 0.6313, loss_yns_0: 0.1461, loss_cls_1: 0.8215, loss_box_1: 1.4968, loss_cns_1: 0.6576, loss_yns_1: 0.1452, loss_cls_2: 0.8250, loss_box_2: 1.4670, loss_cns_2: 0.6582, loss_yns_2: 0.1452, loss_cls_3: 0.8264, loss_box_3: 1.4522, loss_cns_3: 0.6577, loss_yns_3: 0.1437, loss_cls_4: 0.8301, loss_box_4: 1.4536, loss_cns_4: 0.6591, loss_yns_4: 0.1443, loss_cls_5: 0.8311, loss_box_5: 1.4585, loss_cns_5: 0.6581, loss_yns_5: 0.1445, loss_cls_dn_0: 0.1276, loss_box_dn_0: 0.7165, loss_cls_dn_1: 0.0999, loss_box_dn_1: 0.6460, loss_cls_dn_2: 0.1000, loss_box_dn_2: 0.6321, loss_cls_dn_3: 0.0980, loss_box_dn_3: 0.6303, loss_cls_dn_4: 0.0993, loss_box_dn_4: 0.6337, loss_cls_dn_5: 0.1006, loss_box_dn_5: 0.6365, loss_dense_depth: 0.7047, loss: 23.8514, grad_norm: 28.5592 -2025-11-17 14:29:13,483 - mmdet - INFO - Iter [356/17500] lr: 2.418e-04, eta: 8:59:03, time: 1.493, data_time: 0.079, memory: 49163, loss_cls_0: 0.7587, loss_box_0: 1.6253, loss_cns_0: 0.6325, loss_yns_0: 0.1505, loss_cls_1: 0.8359, loss_box_1: 1.5059, loss_cns_1: 0.6563, loss_yns_1: 0.1474, loss_cls_2: 0.8441, loss_box_2: 1.4609, loss_cns_2: 0.6565, loss_yns_2: 0.1479, loss_cls_3: 0.8531, loss_box_3: 1.4516, loss_cns_3: 0.6580, loss_yns_3: 0.1455, loss_cls_4: 0.8622, loss_box_4: 1.4530, loss_cns_4: 0.6614, loss_yns_4: 0.1467, loss_cls_5: 0.8601, loss_box_5: 1.4738, loss_cns_5: 0.6569, loss_yns_5: 0.1466, loss_cls_dn_0: 0.1273, loss_box_dn_0: 0.7269, loss_cls_dn_1: 0.1018, loss_box_dn_1: 0.6378, loss_cls_dn_2: 0.1000, loss_box_dn_2: 0.6199, loss_cls_dn_3: 0.0988, loss_box_dn_3: 0.6153, loss_cls_dn_4: 0.1001, loss_box_dn_4: 0.6187, loss_cls_dn_5: 0.1014, loss_box_dn_5: 0.6249, loss_dense_depth: 0.7547, loss: 24.0184, grad_norm: 49.0821 -2025-11-17 14:29:15,023 - mmdet - INFO - Iter [357/17500] lr: 2.422e-04, eta: 8:58:44, time: 1.540, data_time: 0.079, memory: 49163, loss_cls_0: 0.7512, loss_box_0: 1.6079, loss_cns_0: 0.6304, loss_yns_0: 0.1470, loss_cls_1: 0.8010, loss_box_1: 1.5580, loss_cns_1: 0.6525, loss_yns_1: 0.1456, loss_cls_2: 0.8123, loss_box_2: 1.4934, loss_cns_2: 0.6560, loss_yns_2: 0.1468, loss_cls_3: 0.8137, loss_box_3: 1.4839, loss_cns_3: 0.6568, loss_yns_3: 0.1454, loss_cls_4: 0.8324, loss_box_4: 1.4831, loss_cns_4: 0.6594, loss_yns_4: 0.1455, loss_cls_5: 0.8188, loss_box_5: 1.4946, loss_cns_5: 0.6548, loss_yns_5: 0.1456, loss_cls_dn_0: 0.1239, loss_box_dn_0: 0.7194, loss_cls_dn_1: 0.0986, loss_box_dn_1: 0.6558, loss_cls_dn_2: 0.0977, loss_box_dn_2: 0.6269, loss_cls_dn_3: 0.0967, loss_box_dn_3: 0.6210, loss_cls_dn_4: 0.1006, loss_box_dn_4: 0.6243, loss_cls_dn_5: 0.1016, loss_box_dn_5: 0.6279, loss_dense_depth: 0.7195, loss: 23.9498, grad_norm: 43.6080 -2025-11-17 14:29:16,506 - mmdet - INFO - Iter [358/17500] lr: 2.426e-04, eta: 8:58:23, time: 1.484, data_time: 0.079, memory: 49163, loss_cls_0: 0.7662, loss_box_0: 1.5902, loss_cns_0: 0.6314, loss_yns_0: 0.1444, loss_cls_1: 0.7930, loss_box_1: 1.4674, loss_cns_1: 0.6558, loss_yns_1: 0.1424, loss_cls_2: 0.7995, loss_box_2: 1.4474, loss_cns_2: 0.6595, loss_yns_2: 0.1426, loss_cls_3: 0.8031, loss_box_3: 1.4373, loss_cns_3: 0.6567, loss_yns_3: 0.1436, loss_cls_4: 0.8153, loss_box_4: 1.4279, loss_cns_4: 0.6555, loss_yns_4: 0.1430, loss_cls_5: 0.8204, loss_box_5: 1.4356, loss_cns_5: 0.6567, loss_yns_5: 0.1430, loss_cls_dn_0: 0.1251, loss_box_dn_0: 0.7128, loss_cls_dn_1: 0.1021, loss_box_dn_1: 0.6443, loss_cls_dn_2: 0.1031, loss_box_dn_2: 0.6245, loss_cls_dn_3: 0.1016, loss_box_dn_3: 0.6213, loss_cls_dn_4: 0.1034, loss_box_dn_4: 0.6217, loss_cls_dn_5: 0.1039, loss_box_dn_5: 0.6199, loss_dense_depth: 0.7042, loss: 23.5659, grad_norm: 48.2275 -2025-11-17 14:29:17,990 - mmdet - INFO - Iter [359/17500] lr: 2.429e-04, eta: 8:58:02, time: 1.484, data_time: 0.077, memory: 49163, loss_cls_0: 0.7227, loss_box_0: 1.5601, loss_cns_0: 0.6320, loss_yns_0: 0.1400, loss_cls_1: 0.7761, loss_box_1: 1.4654, loss_cns_1: 0.6579, loss_yns_1: 0.1395, loss_cls_2: 0.7954, loss_box_2: 1.4110, loss_cns_2: 0.6604, loss_yns_2: 0.1380, loss_cls_3: 0.7966, loss_box_3: 1.4057, loss_cns_3: 0.6607, loss_yns_3: 0.1386, loss_cls_4: 0.7979, loss_box_4: 1.3974, loss_cns_4: 0.6617, loss_yns_4: 0.1393, loss_cls_5: 0.8187, loss_box_5: 1.3975, loss_cns_5: 0.6605, loss_yns_5: 0.1386, loss_cls_dn_0: 0.1265, loss_box_dn_0: 0.7141, loss_cls_dn_1: 0.1037, loss_box_dn_1: 0.6366, loss_cls_dn_2: 0.1058, loss_box_dn_2: 0.6198, loss_cls_dn_3: 0.1057, loss_box_dn_3: 0.6181, loss_cls_dn_4: 0.1037, loss_box_dn_4: 0.6169, loss_cls_dn_5: 0.1065, loss_box_dn_5: 0.6172, loss_dense_depth: 0.7239, loss: 23.3100, grad_norm: 43.0829 -2025-11-17 14:29:19,496 - mmdet - INFO - Iter [360/17500] lr: 2.433e-04, eta: 8:57:42, time: 1.505, data_time: 0.083, memory: 49163, loss_cls_0: 0.7418, loss_box_0: 1.5845, loss_cns_0: 0.6292, loss_yns_0: 0.1430, loss_cls_1: 0.7916, loss_box_1: 1.4873, loss_cns_1: 0.6579, loss_yns_1: 0.1432, loss_cls_2: 0.8127, loss_box_2: 1.4391, loss_cns_2: 0.6589, loss_yns_2: 0.1425, loss_cls_3: 0.8252, loss_box_3: 1.4341, loss_cns_3: 0.6639, loss_yns_3: 0.1420, loss_cls_4: 0.8293, loss_box_4: 1.4262, loss_cns_4: 0.6633, loss_yns_4: 0.1427, loss_cls_5: 0.8279, loss_box_5: 1.4365, loss_cns_5: 0.6641, loss_yns_5: 0.1430, loss_cls_dn_0: 0.1261, loss_box_dn_0: 0.7116, loss_cls_dn_1: 0.1029, loss_box_dn_1: 0.6275, loss_cls_dn_2: 0.1027, loss_box_dn_2: 0.6132, loss_cls_dn_3: 0.1027, loss_box_dn_3: 0.6141, loss_cls_dn_4: 0.1033, loss_box_dn_4: 0.6142, loss_cls_dn_5: 0.1057, loss_box_dn_5: 0.6199, loss_dense_depth: 0.6968, loss: 23.5703, grad_norm: 37.4627 -2025-11-17 14:29:21,018 - mmdet - INFO - Iter [361/17500] lr: 2.437e-04, eta: 8:57:23, time: 1.523, data_time: 0.079, memory: 49163, loss_cls_0: 0.7582, loss_box_0: 1.5709, loss_cns_0: 0.6279, loss_yns_0: 0.1446, loss_cls_1: 0.8044, loss_box_1: 1.4469, loss_cns_1: 0.6621, loss_yns_1: 0.1443, loss_cls_2: 0.8271, loss_box_2: 1.4360, loss_cns_2: 0.6640, loss_yns_2: 0.1441, loss_cls_3: 0.8422, loss_box_3: 1.4255, loss_cns_3: 0.6676, loss_yns_3: 0.1445, loss_cls_4: 0.8477, loss_box_4: 1.4162, loss_cns_4: 0.6662, loss_yns_4: 0.1462, loss_cls_5: 0.8472, loss_box_5: 1.4245, loss_cns_5: 0.6664, loss_yns_5: 0.1453, loss_cls_dn_0: 0.1236, loss_box_dn_0: 0.7111, loss_cls_dn_1: 0.1020, loss_box_dn_1: 0.6332, loss_cls_dn_2: 0.1025, loss_box_dn_2: 0.6292, loss_cls_dn_3: 0.1023, loss_box_dn_3: 0.6292, loss_cls_dn_4: 0.1036, loss_box_dn_4: 0.6316, loss_cls_dn_5: 0.1044, loss_box_dn_5: 0.6380, loss_dense_depth: 0.7231, loss: 23.7037, grad_norm: 48.4849 -2025-11-17 14:29:22,596 - mmdet - INFO - Iter [362/17500] lr: 2.441e-04, eta: 8:57:07, time: 1.577, data_time: 0.081, memory: 49163, loss_cls_0: 0.7621, loss_box_0: 1.5938, loss_cns_0: 0.6263, loss_yns_0: 0.1462, loss_cls_1: 0.8098, loss_box_1: 1.4714, loss_cns_1: 0.6609, loss_yns_1: 0.1448, loss_cls_2: 0.8263, loss_box_2: 1.4584, loss_cns_2: 0.6610, loss_yns_2: 0.1454, loss_cls_3: 0.8316, loss_box_3: 1.4393, loss_cns_3: 0.6618, loss_yns_3: 0.1445, loss_cls_4: 0.8346, loss_box_4: 1.4402, loss_cns_4: 0.6616, loss_yns_4: 0.1457, loss_cls_5: 0.8365, loss_box_5: 1.4392, loss_cns_5: 0.6621, loss_yns_5: 0.1457, loss_cls_dn_0: 0.1290, loss_box_dn_0: 0.7112, loss_cls_dn_1: 0.1017, loss_box_dn_1: 0.6366, loss_cls_dn_2: 0.1018, loss_box_dn_2: 0.6349, loss_cls_dn_3: 0.1017, loss_box_dn_3: 0.6315, loss_cls_dn_4: 0.1030, loss_box_dn_4: 0.6321, loss_cls_dn_5: 0.1048, loss_box_dn_5: 0.6355, loss_dense_depth: 0.7420, loss: 23.8148, grad_norm: 34.6633 -2025-11-17 14:29:26,021 - mmdet - INFO - Iter [363/17500] lr: 2.445e-04, eta: 8:58:18, time: 3.427, data_time: 0.079, memory: 49163, loss_cls_0: 0.7462, loss_box_0: 1.5925, loss_cns_0: 0.6327, loss_yns_0: 0.1420, loss_cls_1: 0.8039, loss_box_1: 1.4744, loss_cns_1: 0.6599, loss_yns_1: 0.1399, loss_cls_2: 0.8274, loss_box_2: 1.4628, loss_cns_2: 0.6605, loss_yns_2: 0.1394, loss_cls_3: 0.8318, loss_box_3: 1.4363, loss_cns_3: 0.6607, loss_yns_3: 0.1393, loss_cls_4: 0.8313, loss_box_4: 1.4446, loss_cns_4: 0.6637, loss_yns_4: 0.1390, loss_cls_5: 0.8375, loss_box_5: 1.4443, loss_cns_5: 0.6618, loss_yns_5: 0.1398, loss_cls_dn_0: 0.1217, loss_box_dn_0: 0.7214, loss_cls_dn_1: 0.0968, loss_box_dn_1: 0.6424, loss_cls_dn_2: 0.0975, loss_box_dn_2: 0.6391, loss_cls_dn_3: 0.0978, loss_box_dn_3: 0.6289, loss_cls_dn_4: 0.0980, loss_box_dn_4: 0.6303, loss_cls_dn_5: 0.1014, loss_box_dn_5: 0.6318, loss_dense_depth: 0.7223, loss: 23.7409, grad_norm: 47.0331 -2025-11-17 14:29:27,551 - mmdet - INFO - Iter [364/17500] lr: 2.449e-04, eta: 8:57:59, time: 1.531, data_time: 0.077, memory: 49163, loss_cls_0: 0.7381, loss_box_0: 1.5829, loss_cns_0: 0.6334, loss_yns_0: 0.1409, loss_cls_1: 0.7896, loss_box_1: 1.4901, loss_cns_1: 0.6594, loss_yns_1: 0.1391, loss_cls_2: 0.8136, loss_box_2: 1.4629, loss_cns_2: 0.6593, loss_yns_2: 0.1391, loss_cls_3: 0.8154, loss_box_3: 1.4436, loss_cns_3: 0.6610, loss_yns_3: 0.1377, loss_cls_4: 0.8144, loss_box_4: 1.4442, loss_cns_4: 0.6628, loss_yns_4: 0.1389, loss_cls_5: 0.8199, loss_box_5: 1.4430, loss_cns_5: 0.6608, loss_yns_5: 0.1397, loss_cls_dn_0: 0.1210, loss_box_dn_0: 0.7164, loss_cls_dn_1: 0.0953, loss_box_dn_1: 0.6302, loss_cls_dn_2: 0.0963, loss_box_dn_2: 0.6206, loss_cls_dn_3: 0.0974, loss_box_dn_3: 0.6113, loss_cls_dn_4: 0.0967, loss_box_dn_4: 0.6111, loss_cls_dn_5: 0.0994, loss_box_dn_5: 0.6092, loss_dense_depth: 0.7391, loss: 23.5738, grad_norm: 33.8155 -2025-11-17 14:29:29,051 - mmdet - INFO - Iter [365/17500] lr: 2.453e-04, eta: 8:57:39, time: 1.498, data_time: 0.078, memory: 49163, loss_cls_0: 0.7349, loss_box_0: 1.6202, loss_cns_0: 0.6348, loss_yns_0: 0.1417, loss_cls_1: 0.7872, loss_box_1: 1.5133, loss_cns_1: 0.6563, loss_yns_1: 0.1403, loss_cls_2: 0.8058, loss_box_2: 1.4626, loss_cns_2: 0.6588, loss_yns_2: 0.1398, loss_cls_3: 0.8103, loss_box_3: 1.4627, loss_cns_3: 0.6589, loss_yns_3: 0.1391, loss_cls_4: 0.8183, loss_box_4: 1.4570, loss_cns_4: 0.6588, loss_yns_4: 0.1390, loss_cls_5: 0.8200, loss_box_5: 1.4639, loss_cns_5: 0.6597, loss_yns_5: 0.1402, loss_cls_dn_0: 0.1208, loss_box_dn_0: 0.7249, loss_cls_dn_1: 0.0985, loss_box_dn_1: 0.6265, loss_cls_dn_2: 0.0996, loss_box_dn_2: 0.6065, loss_cls_dn_3: 0.0994, loss_box_dn_3: 0.6037, loss_cls_dn_4: 0.1001, loss_box_dn_4: 0.6025, loss_cls_dn_5: 0.1004, loss_box_dn_5: 0.6037, loss_dense_depth: 0.6977, loss: 23.6078, grad_norm: 43.8579 -2025-11-17 14:29:30,576 - mmdet - INFO - Iter [366/17500] lr: 2.457e-04, eta: 8:57:21, time: 1.524, data_time: 0.082, memory: 49163, loss_cls_0: 0.7108, loss_box_0: 1.6044, loss_cns_0: 0.6349, loss_yns_0: 0.1390, loss_cls_1: 0.7682, loss_box_1: 1.5167, loss_cns_1: 0.6550, loss_yns_1: 0.1370, loss_cls_2: 0.7951, loss_box_2: 1.4653, loss_cns_2: 0.6574, loss_yns_2: 0.1369, loss_cls_3: 0.7975, loss_box_3: 1.4619, loss_cns_3: 0.6576, loss_yns_3: 0.1367, loss_cls_4: 0.8027, loss_box_4: 1.4531, loss_cns_4: 0.6582, loss_yns_4: 0.1361, loss_cls_5: 0.7995, loss_box_5: 1.4596, loss_cns_5: 0.6592, loss_yns_5: 0.1369, loss_cls_dn_0: 0.1161, loss_box_dn_0: 0.7220, loss_cls_dn_1: 0.0952, loss_box_dn_1: 0.6280, loss_cls_dn_2: 0.0942, loss_box_dn_2: 0.6116, loss_cls_dn_3: 0.0934, loss_box_dn_3: 0.6086, loss_cls_dn_4: 0.0950, loss_box_dn_4: 0.6076, loss_cls_dn_5: 0.0952, loss_box_dn_5: 0.6085, loss_dense_depth: 0.7082, loss: 23.4635, grad_norm: 43.5177 -2025-11-17 14:29:32,101 - mmdet - INFO - Iter [367/17500] lr: 2.461e-04, eta: 8:57:02, time: 1.528, data_time: 0.087, memory: 49163, loss_cls_0: 0.7090, loss_box_0: 1.6169, loss_cns_0: 0.6337, loss_yns_0: 0.1361, loss_cls_1: 0.7745, loss_box_1: 1.4632, loss_cns_1: 0.6595, loss_yns_1: 0.1337, loss_cls_2: 0.7940, loss_box_2: 1.4278, loss_cns_2: 0.6594, loss_yns_2: 0.1356, loss_cls_3: 0.7961, loss_box_3: 1.4338, loss_cns_3: 0.6610, loss_yns_3: 0.1355, loss_cls_4: 0.8009, loss_box_4: 1.4302, loss_cns_4: 0.6595, loss_yns_4: 0.1358, loss_cls_5: 0.8043, loss_box_5: 1.4376, loss_cns_5: 0.6606, loss_yns_5: 0.1362, loss_cls_dn_0: 0.1181, loss_box_dn_0: 0.7213, loss_cls_dn_1: 0.0956, loss_box_dn_1: 0.6321, loss_cls_dn_2: 0.0950, loss_box_dn_2: 0.6228, loss_cls_dn_3: 0.0949, loss_box_dn_3: 0.6184, loss_cls_dn_4: 0.0963, loss_box_dn_4: 0.6217, loss_cls_dn_5: 0.0975, loss_box_dn_5: 0.6226, loss_dense_depth: 0.6869, loss: 23.3579, grad_norm: 36.1068 -2025-11-17 14:29:33,612 - mmdet - INFO - Iter [368/17500] lr: 2.465e-04, eta: 8:56:43, time: 1.509, data_time: 0.085, memory: 49163, loss_cls_0: 0.7198, loss_box_0: 1.6145, loss_cns_0: 0.6294, loss_yns_0: 0.1363, loss_cls_1: 0.7818, loss_box_1: 1.4646, loss_cns_1: 0.6582, loss_yns_1: 0.1337, loss_cls_2: 0.8024, loss_box_2: 1.4345, loss_cns_2: 0.6602, loss_yns_2: 0.1338, loss_cls_3: 0.8030, loss_box_3: 1.4342, loss_cns_3: 0.6580, loss_yns_3: 0.1338, loss_cls_4: 0.8081, loss_box_4: 1.4342, loss_cns_4: 0.6565, loss_yns_4: 0.1324, loss_cls_5: 0.8103, loss_box_5: 1.4350, loss_cns_5: 0.6584, loss_yns_5: 0.1338, loss_cls_dn_0: 0.1178, loss_box_dn_0: 0.7185, loss_cls_dn_1: 0.0951, loss_box_dn_1: 0.6311, loss_cls_dn_2: 0.0947, loss_box_dn_2: 0.6247, loss_cls_dn_3: 0.0952, loss_box_dn_3: 0.6206, loss_cls_dn_4: 0.0962, loss_box_dn_4: 0.6262, loss_cls_dn_5: 0.0976, loss_box_dn_5: 0.6289, loss_dense_depth: 0.7143, loss: 23.4279, grad_norm: 43.5680 -2025-11-17 14:29:35,223 - mmdet - INFO - Iter [369/17500] lr: 2.469e-04, eta: 8:56:29, time: 1.611, data_time: 0.213, memory: 49163, loss_cls_0: 0.7133, loss_box_0: 1.5850, loss_cns_0: 0.6374, loss_yns_0: 0.1364, loss_cls_1: 0.7773, loss_box_1: 1.4567, loss_cns_1: 0.6623, loss_yns_1: 0.1321, loss_cls_2: 0.7877, loss_box_2: 1.4143, loss_cns_2: 0.6618, loss_yns_2: 0.1326, loss_cls_3: 0.7908, loss_box_3: 1.4163, loss_cns_3: 0.6616, loss_yns_3: 0.1323, loss_cls_4: 0.7943, loss_box_4: 1.4144, loss_cns_4: 0.6618, loss_yns_4: 0.1323, loss_cls_5: 0.7988, loss_box_5: 1.4164, loss_cns_5: 0.6629, loss_yns_5: 0.1336, loss_cls_dn_0: 0.1166, loss_box_dn_0: 0.7098, loss_cls_dn_1: 0.0974, loss_box_dn_1: 0.6339, loss_cls_dn_2: 0.0963, loss_box_dn_2: 0.6237, loss_cls_dn_3: 0.0973, loss_box_dn_3: 0.6198, loss_cls_dn_4: 0.0978, loss_box_dn_4: 0.6217, loss_cls_dn_5: 0.0976, loss_box_dn_5: 0.6279, loss_dense_depth: 0.6895, loss: 23.2417, grad_norm: 34.7248 -2025-11-17 14:29:36,728 - mmdet - INFO - Iter [370/17500] lr: 2.473e-04, eta: 8:56:10, time: 1.505, data_time: 0.080, memory: 49163, loss_cls_0: 0.7225, loss_box_0: 1.5913, loss_cns_0: 0.6408, loss_yns_0: 0.1378, loss_cls_1: 0.7868, loss_box_1: 1.4440, loss_cns_1: 0.6639, loss_yns_1: 0.1373, loss_cls_2: 0.7903, loss_box_2: 1.4340, loss_cns_2: 0.6651, loss_yns_2: 0.1374, loss_cls_3: 0.7961, loss_box_3: 1.4187, loss_cns_3: 0.6658, loss_yns_3: 0.1376, loss_cls_4: 0.7982, loss_box_4: 1.4204, loss_cns_4: 0.6665, loss_yns_4: 0.1386, loss_cls_5: 0.7980, loss_box_5: 1.4254, loss_cns_5: 0.6652, loss_yns_5: 0.1397, loss_cls_dn_0: 0.1158, loss_box_dn_0: 0.7155, loss_cls_dn_1: 0.0968, loss_box_dn_1: 0.6364, loss_cls_dn_2: 0.0954, loss_box_dn_2: 0.6287, loss_cls_dn_3: 0.0951, loss_box_dn_3: 0.6221, loss_cls_dn_4: 0.0963, loss_box_dn_4: 0.6209, loss_cls_dn_5: 0.0956, loss_box_dn_5: 0.6254, loss_dense_depth: 0.7154, loss: 23.3806, grad_norm: 35.8661 -2025-11-17 14:29:38,229 - mmdet - INFO - Iter [371/17500] lr: 2.477e-04, eta: 8:55:50, time: 1.500, data_time: 0.079, memory: 49163, loss_cls_0: 0.7102, loss_box_0: 1.5859, loss_cns_0: 0.6376, loss_yns_0: 0.1393, loss_cls_1: 0.7763, loss_box_1: 1.4418, loss_cns_1: 0.6601, loss_yns_1: 0.1382, loss_cls_2: 0.7878, loss_box_2: 1.4176, loss_cns_2: 0.6605, loss_yns_2: 0.1390, loss_cls_3: 0.7798, loss_box_3: 1.4120, loss_cns_3: 0.6591, loss_yns_3: 0.1383, loss_cls_4: 0.7834, loss_box_4: 1.4139, loss_cns_4: 0.6587, loss_yns_4: 0.1394, loss_cls_5: 0.7854, loss_box_5: 1.4141, loss_cns_5: 0.6600, loss_yns_5: 0.1402, loss_cls_dn_0: 0.1086, loss_box_dn_0: 0.7125, loss_cls_dn_1: 0.0918, loss_box_dn_1: 0.6223, loss_cls_dn_2: 0.0907, loss_box_dn_2: 0.6104, loss_cls_dn_3: 0.0898, loss_box_dn_3: 0.6066, loss_cls_dn_4: 0.0919, loss_box_dn_4: 0.6065, loss_cls_dn_5: 0.0913, loss_box_dn_5: 0.6076, loss_dense_depth: 0.6943, loss: 23.1027, grad_norm: 27.0350 -2025-11-17 14:29:39,730 - mmdet - INFO - Iter [372/17500] lr: 2.481e-04, eta: 8:55:31, time: 1.500, data_time: 0.082, memory: 49163, loss_cls_0: 0.7261, loss_box_0: 1.5945, loss_cns_0: 0.6361, loss_yns_0: 0.1406, loss_cls_1: 0.7831, loss_box_1: 1.4832, loss_cns_1: 0.6578, loss_yns_1: 0.1395, loss_cls_2: 0.7885, loss_box_2: 1.4505, loss_cns_2: 0.6593, loss_yns_2: 0.1404, loss_cls_3: 0.7966, loss_box_3: 1.4434, loss_cns_3: 0.6572, loss_yns_3: 0.1394, loss_cls_4: 0.7939, loss_box_4: 1.4533, loss_cns_4: 0.6583, loss_yns_4: 0.1408, loss_cls_5: 0.8058, loss_box_5: 1.4560, loss_cns_5: 0.6579, loss_yns_5: 0.1421, loss_cls_dn_0: 0.1151, loss_box_dn_0: 0.7137, loss_cls_dn_1: 0.0931, loss_box_dn_1: 0.6184, loss_cls_dn_2: 0.0915, loss_box_dn_2: 0.6025, loss_cls_dn_3: 0.0916, loss_box_dn_3: 0.6009, loss_cls_dn_4: 0.0925, loss_box_dn_4: 0.6022, loss_cls_dn_5: 0.0934, loss_box_dn_5: 0.6047, loss_dense_depth: 0.7273, loss: 23.3910, grad_norm: 33.8834 -2025-11-17 14:29:41,241 - mmdet - INFO - Iter [373/17500] lr: 2.485e-04, eta: 8:55:13, time: 1.512, data_time: 0.080, memory: 49163, loss_cls_0: 0.7530, loss_box_0: 1.6324, loss_cns_0: 0.6330, loss_yns_0: 0.1414, loss_cls_1: 0.8164, loss_box_1: 1.5013, loss_cns_1: 0.6599, loss_yns_1: 0.1399, loss_cls_2: 0.8181, loss_box_2: 1.4798, loss_cns_2: 0.6630, loss_yns_2: 0.1414, loss_cls_3: 0.8286, loss_box_3: 1.4638, loss_cns_3: 0.6604, loss_yns_3: 0.1410, loss_cls_4: 0.8281, loss_box_4: 1.4676, loss_cns_4: 0.6613, loss_yns_4: 0.1415, loss_cls_5: 0.8320, loss_box_5: 1.4680, loss_cns_5: 0.6617, loss_yns_5: 0.1435, loss_cls_dn_0: 0.1157, loss_box_dn_0: 0.7028, loss_cls_dn_1: 0.0927, loss_box_dn_1: 0.6261, loss_cls_dn_2: 0.0915, loss_box_dn_2: 0.6111, loss_cls_dn_3: 0.0910, loss_box_dn_3: 0.6065, loss_cls_dn_4: 0.0909, loss_box_dn_4: 0.6091, loss_cls_dn_5: 0.0921, loss_box_dn_5: 0.6110, loss_dense_depth: 0.7320, loss: 23.7496, grad_norm: 31.9682 -2025-11-17 14:29:42,786 - mmdet - INFO - Iter [374/17500] lr: 2.489e-04, eta: 8:54:56, time: 1.546, data_time: 0.077, memory: 49163, loss_cls_0: 0.7311, loss_box_0: 1.6053, loss_cns_0: 0.6304, loss_yns_0: 0.1424, loss_cls_1: 0.8017, loss_box_1: 1.4906, loss_cns_1: 0.6574, loss_yns_1: 0.1414, loss_cls_2: 0.8077, loss_box_2: 1.4540, loss_cns_2: 0.6576, loss_yns_2: 0.1410, loss_cls_3: 0.8092, loss_box_3: 1.4370, loss_cns_3: 0.6566, loss_yns_3: 0.1413, loss_cls_4: 0.8146, loss_box_4: 1.4355, loss_cns_4: 0.6567, loss_yns_4: 0.1418, loss_cls_5: 0.8133, loss_box_5: 1.4348, loss_cns_5: 0.6584, loss_yns_5: 0.1416, loss_cls_dn_0: 0.1131, loss_box_dn_0: 0.7084, loss_cls_dn_1: 0.0926, loss_box_dn_1: 0.6219, loss_cls_dn_2: 0.0922, loss_box_dn_2: 0.6053, loss_cls_dn_3: 0.0909, loss_box_dn_3: 0.6010, loss_cls_dn_4: 0.0917, loss_box_dn_4: 0.6021, loss_cls_dn_5: 0.0929, loss_box_dn_5: 0.6045, loss_dense_depth: 0.7063, loss: 23.4243, grad_norm: 22.3878 -2025-11-17 14:29:44,282 - mmdet - INFO - Iter [375/17500] lr: 2.493e-04, eta: 8:54:36, time: 1.493, data_time: 0.077, memory: 49163, loss_cls_0: 0.7286, loss_box_0: 1.6048, loss_cns_0: 0.6325, loss_yns_0: 0.1447, loss_cls_1: 0.7881, loss_box_1: 1.4371, loss_cns_1: 0.6598, loss_yns_1: 0.1427, loss_cls_2: 0.7989, loss_box_2: 1.4196, loss_cns_2: 0.6591, loss_yns_2: 0.1420, loss_cls_3: 0.8006, loss_box_3: 1.4163, loss_cns_3: 0.6578, loss_yns_3: 0.1423, loss_cls_4: 0.8026, loss_box_4: 1.4199, loss_cns_4: 0.6595, loss_yns_4: 0.1443, loss_cls_5: 0.8027, loss_box_5: 1.4246, loss_cns_5: 0.6590, loss_yns_5: 0.1428, loss_cls_dn_0: 0.1126, loss_box_dn_0: 0.7147, loss_cls_dn_1: 0.0925, loss_box_dn_1: 0.6149, loss_cls_dn_2: 0.0925, loss_box_dn_2: 0.6089, loss_cls_dn_3: 0.0910, loss_box_dn_3: 0.6121, loss_cls_dn_4: 0.0925, loss_box_dn_4: 0.6141, loss_cls_dn_5: 0.0930, loss_box_dn_5: 0.6197, loss_dense_depth: 0.7031, loss: 23.2918, grad_norm: 35.0325 -2025-11-17 14:29:45,775 - mmdet - INFO - Iter [376/17500] lr: 2.497e-04, eta: 8:54:17, time: 1.495, data_time: 0.084, memory: 49163, loss_cls_0: 0.7546, loss_box_0: 1.6062, loss_cns_0: 0.6343, loss_yns_0: 0.1447, loss_cls_1: 0.8109, loss_box_1: 1.4604, loss_cns_1: 0.6573, loss_yns_1: 0.1444, loss_cls_2: 0.8175, loss_box_2: 1.4378, loss_cns_2: 0.6568, loss_yns_2: 0.1433, loss_cls_3: 0.8288, loss_box_3: 1.4232, loss_cns_3: 0.6524, loss_yns_3: 0.1426, loss_cls_4: 0.8321, loss_box_4: 1.4450, loss_cns_4: 0.6579, loss_yns_4: 0.1438, loss_cls_5: 0.8325, loss_box_5: 1.4556, loss_cns_5: 0.6572, loss_yns_5: 0.1452, loss_cls_dn_0: 0.1181, loss_box_dn_0: 0.7037, loss_cls_dn_1: 0.0962, loss_box_dn_1: 0.6528, loss_cls_dn_2: 0.0960, loss_box_dn_2: 0.6448, loss_cls_dn_3: 0.0945, loss_box_dn_3: 0.6509, loss_cls_dn_4: 0.0951, loss_box_dn_4: 0.6570, loss_cls_dn_5: 0.0955, loss_box_dn_5: 0.6656, loss_dense_depth: 0.7306, loss: 23.7857, grad_norm: 33.6987 -2025-11-17 14:29:47,294 - mmdet - INFO - Iter [377/17500] lr: 2.501e-04, eta: 8:53:59, time: 1.520, data_time: 0.078, memory: 49163, loss_cls_0: 0.7502, loss_box_0: 1.5777, loss_cns_0: 0.6322, loss_yns_0: 0.1448, loss_cls_1: 0.8108, loss_box_1: 1.4522, loss_cns_1: 0.6567, loss_yns_1: 0.1440, loss_cls_2: 0.8179, loss_box_2: 1.4359, loss_cns_2: 0.6552, loss_yns_2: 0.1432, loss_cls_3: 0.8306, loss_box_3: 1.4199, loss_cns_3: 0.6545, loss_yns_3: 0.1420, loss_cls_4: 0.8357, loss_box_4: 1.4289, loss_cns_4: 0.6598, loss_yns_4: 0.1445, loss_cls_5: 0.8282, loss_box_5: 1.4264, loss_cns_5: 0.6577, loss_yns_5: 0.1435, loss_cls_dn_0: 0.1118, loss_box_dn_0: 0.7160, loss_cls_dn_1: 0.0952, loss_box_dn_1: 0.6697, loss_cls_dn_2: 0.0943, loss_box_dn_2: 0.6583, loss_cls_dn_3: 0.0932, loss_box_dn_3: 0.6581, loss_cls_dn_4: 0.0939, loss_box_dn_4: 0.6622, loss_cls_dn_5: 0.0940, loss_box_dn_5: 0.6656, loss_dense_depth: 0.7243, loss: 23.7293, grad_norm: 40.1313 -2025-11-17 14:29:48,789 - mmdet - INFO - Iter [378/17500] lr: 2.505e-04, eta: 8:53:40, time: 1.494, data_time: 0.079, memory: 49163, loss_cls_0: 0.7417, loss_box_0: 1.5647, loss_cns_0: 0.6304, loss_yns_0: 0.1451, loss_cls_1: 0.7956, loss_box_1: 1.4659, loss_cns_1: 0.6562, loss_yns_1: 0.1448, loss_cls_2: 0.8009, loss_box_2: 1.4509, loss_cns_2: 0.6566, loss_yns_2: 0.1451, loss_cls_3: 0.8039, loss_box_3: 1.4303, loss_cns_3: 0.6558, loss_yns_3: 0.1436, loss_cls_4: 0.8108, loss_box_4: 1.4316, loss_cns_4: 0.6564, loss_yns_4: 0.1469, loss_cls_5: 0.8050, loss_box_5: 1.4316, loss_cns_5: 0.6563, loss_yns_5: 0.1425, loss_cls_dn_0: 0.1110, loss_box_dn_0: 0.7090, loss_cls_dn_1: 0.0910, loss_box_dn_1: 0.6573, loss_cls_dn_2: 0.0897, loss_box_dn_2: 0.6413, loss_cls_dn_3: 0.0892, loss_box_dn_3: 0.6347, loss_cls_dn_4: 0.0899, loss_box_dn_4: 0.6351, loss_cls_dn_5: 0.0901, loss_box_dn_5: 0.6369, loss_dense_depth: 0.7228, loss: 23.5105, grad_norm: 28.4398 -2025-11-17 14:29:50,300 - mmdet - INFO - Iter [379/17500] lr: 2.509e-04, eta: 8:53:22, time: 1.510, data_time: 0.083, memory: 49163, loss_cls_0: 0.7281, loss_box_0: 1.5857, loss_cns_0: 0.6328, loss_yns_0: 0.1444, loss_cls_1: 0.8028, loss_box_1: 1.4811, loss_cns_1: 0.6572, loss_yns_1: 0.1430, loss_cls_2: 0.8167, loss_box_2: 1.4586, loss_cns_2: 0.6611, loss_yns_2: 0.1433, loss_cls_3: 0.8183, loss_box_3: 1.4380, loss_cns_3: 0.6597, loss_yns_3: 0.1420, loss_cls_4: 0.8197, loss_box_4: 1.4455, loss_cns_4: 0.6590, loss_yns_4: 0.1448, loss_cls_5: 0.8279, loss_box_5: 1.4513, loss_cns_5: 0.6599, loss_yns_5: 0.1429, loss_cls_dn_0: 0.1096, loss_box_dn_0: 0.7021, loss_cls_dn_1: 0.0900, loss_box_dn_1: 0.6450, loss_cls_dn_2: 0.0889, loss_box_dn_2: 0.6242, loss_cls_dn_3: 0.0890, loss_box_dn_3: 0.6181, loss_cls_dn_4: 0.0903, loss_box_dn_4: 0.6229, loss_cls_dn_5: 0.0910, loss_box_dn_5: 0.6256, loss_dense_depth: 0.7364, loss: 23.5969, grad_norm: 41.6038 -2025-11-17 14:29:51,805 - mmdet - INFO - Iter [380/17500] lr: 2.513e-04, eta: 8:53:04, time: 1.506, data_time: 0.083, memory: 49163, loss_cls_0: 0.7472, loss_box_0: 1.5741, loss_cns_0: 0.6378, loss_yns_0: 0.1436, loss_cls_1: 0.8129, loss_box_1: 1.4368, loss_cns_1: 0.6597, loss_yns_1: 0.1423, loss_cls_2: 0.8306, loss_box_2: 1.3997, loss_cns_2: 0.6623, loss_yns_2: 0.1428, loss_cls_3: 0.8349, loss_box_3: 1.3883, loss_cns_3: 0.6613, loss_yns_3: 0.1413, loss_cls_4: 0.8350, loss_box_4: 1.3947, loss_cns_4: 0.6620, loss_yns_4: 0.1423, loss_cls_5: 0.8451, loss_box_5: 1.3924, loss_cns_5: 0.6620, loss_yns_5: 0.1432, loss_cls_dn_0: 0.1154, loss_box_dn_0: 0.7111, loss_cls_dn_1: 0.0920, loss_box_dn_1: 0.6404, loss_cls_dn_2: 0.0918, loss_box_dn_2: 0.6162, loss_cls_dn_3: 0.0914, loss_box_dn_3: 0.6137, loss_cls_dn_4: 0.0929, loss_box_dn_4: 0.6164, loss_cls_dn_5: 0.0942, loss_box_dn_5: 0.6170, loss_dense_depth: 0.7593, loss: 23.4443, grad_norm: 34.1570 -2025-11-17 14:29:53,343 - mmdet - INFO - Iter [381/17500] lr: 2.517e-04, eta: 8:52:47, time: 1.539, data_time: 0.078, memory: 49163, loss_cls_0: 0.7166, loss_box_0: 1.5559, loss_cns_0: 0.6361, loss_yns_0: 0.1402, loss_cls_1: 0.7957, loss_box_1: 1.4564, loss_cns_1: 0.6598, loss_yns_1: 0.1397, loss_cls_2: 0.8108, loss_box_2: 1.4240, loss_cns_2: 0.6607, loss_yns_2: 0.1400, loss_cls_3: 0.8241, loss_box_3: 1.4089, loss_cns_3: 0.6614, loss_yns_3: 0.1398, loss_cls_4: 0.8301, loss_box_4: 1.4051, loss_cns_4: 0.6618, loss_yns_4: 0.1406, loss_cls_5: 0.8252, loss_box_5: 1.4148, loss_cns_5: 0.6618, loss_yns_5: 0.1413, loss_cls_dn_0: 0.1114, loss_box_dn_0: 0.7189, loss_cls_dn_1: 0.0928, loss_box_dn_1: 0.6309, loss_cls_dn_2: 0.0929, loss_box_dn_2: 0.6187, loss_cls_dn_3: 0.0928, loss_box_dn_3: 0.6156, loss_cls_dn_4: 0.0936, loss_box_dn_4: 0.6157, loss_cls_dn_5: 0.0949, loss_box_dn_5: 0.6226, loss_dense_depth: 0.6991, loss: 23.3506, grad_norm: 46.5753 -2025-11-17 14:29:54,899 - mmdet - INFO - Iter [382/17500] lr: 2.521e-04, eta: 8:52:32, time: 1.556, data_time: 0.077, memory: 49163, loss_cls_0: 0.7155, loss_box_0: 1.5617, loss_cns_0: 0.6381, loss_yns_0: 0.1410, loss_cls_1: 0.7978, loss_box_1: 1.4822, loss_cns_1: 0.6641, loss_yns_1: 0.1405, loss_cls_2: 0.8129, loss_box_2: 1.4253, loss_cns_2: 0.6661, loss_yns_2: 0.1405, loss_cls_3: 0.8212, loss_box_3: 1.4129, loss_cns_3: 0.6686, loss_yns_3: 0.1404, loss_cls_4: 0.8117, loss_box_4: 1.4145, loss_cns_4: 0.6663, loss_yns_4: 0.1433, loss_cls_5: 0.8150, loss_box_5: 1.4175, loss_cns_5: 0.6678, loss_yns_5: 0.1414, loss_cls_dn_0: 0.1112, loss_box_dn_0: 0.7116, loss_cls_dn_1: 0.0936, loss_box_dn_1: 0.6310, loss_cls_dn_2: 0.0950, loss_box_dn_2: 0.6157, loss_cls_dn_3: 0.0943, loss_box_dn_3: 0.6111, loss_cls_dn_4: 0.0936, loss_box_dn_4: 0.6131, loss_cls_dn_5: 0.0943, loss_box_dn_5: 0.6206, loss_dense_depth: 0.7315, loss: 23.4228, grad_norm: 36.0050 -2025-11-17 14:29:56,404 - mmdet - INFO - Iter [383/17500] lr: 2.525e-04, eta: 8:52:14, time: 1.505, data_time: 0.079, memory: 49163, loss_cls_0: 0.7357, loss_box_0: 1.6111, loss_cns_0: 0.6363, loss_yns_0: 0.1433, loss_cls_1: 0.7923, loss_box_1: 1.4587, loss_cns_1: 0.6641, loss_yns_1: 0.1413, loss_cls_2: 0.8078, loss_box_2: 1.4429, loss_cns_2: 0.6632, loss_yns_2: 0.1416, loss_cls_3: 0.8203, loss_box_3: 1.4450, loss_cns_3: 0.6636, loss_yns_3: 0.1402, loss_cls_4: 0.8213, loss_box_4: 1.4497, loss_cns_4: 0.6637, loss_yns_4: 0.1416, loss_cls_5: 0.8169, loss_box_5: 1.4420, loss_cns_5: 0.6628, loss_yns_5: 0.1389, loss_cls_dn_0: 0.1101, loss_box_dn_0: 0.7204, loss_cls_dn_1: 0.0935, loss_box_dn_1: 0.6298, loss_cls_dn_2: 0.0935, loss_box_dn_2: 0.6229, loss_cls_dn_3: 0.0932, loss_box_dn_3: 0.6221, loss_cls_dn_4: 0.0933, loss_box_dn_4: 0.6269, loss_cls_dn_5: 0.0936, loss_box_dn_5: 0.6288, loss_dense_depth: 0.6936, loss: 23.5663, grad_norm: 50.5972 -2025-11-17 14:29:57,952 - mmdet - INFO - Iter [384/17500] lr: 2.529e-04, eta: 8:51:58, time: 1.549, data_time: 0.076, memory: 49163, loss_cls_0: 0.7630, loss_box_0: 1.6246, loss_cns_0: 0.6323, loss_yns_0: 0.1427, loss_cls_1: 0.8022, loss_box_1: 1.4813, loss_cns_1: 0.6622, loss_yns_1: 0.1419, loss_cls_2: 0.8142, loss_box_2: 1.4734, loss_cns_2: 0.6602, loss_yns_2: 0.1415, loss_cls_3: 0.8291, loss_box_3: 1.4751, loss_cns_3: 0.6622, loss_yns_3: 0.1417, loss_cls_4: 0.8418, loss_box_4: 1.4805, loss_cns_4: 0.6628, loss_yns_4: 0.1439, loss_cls_5: 0.8372, loss_box_5: 1.4756, loss_cns_5: 0.6623, loss_yns_5: 0.1411, loss_cls_dn_0: 0.1130, loss_box_dn_0: 0.7303, loss_cls_dn_1: 0.0914, loss_box_dn_1: 0.6402, loss_cls_dn_2: 0.0906, loss_box_dn_2: 0.6337, loss_cls_dn_3: 0.0905, loss_box_dn_3: 0.6335, loss_cls_dn_4: 0.0917, loss_box_dn_4: 0.6369, loss_cls_dn_5: 0.0917, loss_box_dn_5: 0.6377, loss_dense_depth: 0.7458, loss: 23.9198, grad_norm: 52.0119 -2025-11-17 14:29:59,462 - mmdet - INFO - Iter [385/17500] lr: 2.533e-04, eta: 8:51:40, time: 1.509, data_time: 0.075, memory: 49163, loss_cls_0: 0.7290, loss_box_0: 1.5881, loss_cns_0: 0.6402, loss_yns_0: 0.1430, loss_cls_1: 0.7974, loss_box_1: 1.4898, loss_cns_1: 0.6627, loss_yns_1: 0.1430, loss_cls_2: 0.7975, loss_box_2: 1.4573, loss_cns_2: 0.6634, loss_yns_2: 0.1430, loss_cls_3: 0.8168, loss_box_3: 1.4382, loss_cns_3: 0.6636, loss_yns_3: 0.1437, loss_cls_4: 0.8264, loss_box_4: 1.4443, loss_cns_4: 0.6635, loss_yns_4: 0.1432, loss_cls_5: 0.8278, loss_box_5: 1.4575, loss_cns_5: 0.6641, loss_yns_5: 0.1434, loss_cls_dn_0: 0.1123, loss_box_dn_0: 0.7099, loss_cls_dn_1: 0.0894, loss_box_dn_1: 0.6370, loss_cls_dn_2: 0.0885, loss_box_dn_2: 0.6230, loss_cls_dn_3: 0.0875, loss_box_dn_3: 0.6187, loss_cls_dn_4: 0.0899, loss_box_dn_4: 0.6217, loss_cls_dn_5: 0.0911, loss_box_dn_5: 0.6278, loss_dense_depth: 0.7172, loss: 23.6007, grad_norm: 32.0292 -2025-11-17 14:30:00,975 - mmdet - INFO - Iter [386/17500] lr: 2.537e-04, eta: 8:51:22, time: 1.513, data_time: 0.077, memory: 49163, loss_cls_0: 0.7485, loss_box_0: 1.5794, loss_cns_0: 0.6342, loss_yns_0: 0.1370, loss_cls_1: 0.8090, loss_box_1: 1.4952, loss_cns_1: 0.6590, loss_yns_1: 0.1389, loss_cls_2: 0.8181, loss_box_2: 1.4812, loss_cns_2: 0.6574, loss_yns_2: 0.1364, loss_cls_3: 0.8236, loss_box_3: 1.4691, loss_cns_3: 0.6585, loss_yns_3: 0.1364, loss_cls_4: 0.8273, loss_box_4: 1.4739, loss_cns_4: 0.6591, loss_yns_4: 0.1371, loss_cls_5: 0.8358, loss_box_5: 1.4829, loss_cns_5: 0.6590, loss_yns_5: 0.1370, loss_cls_dn_0: 0.1142, loss_box_dn_0: 0.7109, loss_cls_dn_1: 0.0913, loss_box_dn_1: 0.6314, loss_cls_dn_2: 0.0900, loss_box_dn_2: 0.6203, loss_cls_dn_3: 0.0906, loss_box_dn_3: 0.6127, loss_cls_dn_4: 0.0922, loss_box_dn_4: 0.6147, loss_cls_dn_5: 0.0945, loss_box_dn_5: 0.6195, loss_dense_depth: 0.7538, loss: 23.7300, grad_norm: 44.3518 -2025-11-17 14:30:02,509 - mmdet - INFO - Iter [387/17500] lr: 2.541e-04, eta: 8:51:06, time: 1.534, data_time: 0.087, memory: 49163, loss_cls_0: 0.7312, loss_box_0: 1.5671, loss_cns_0: 0.6304, loss_yns_0: 0.1370, loss_cls_1: 0.7776, loss_box_1: 1.4972, loss_cns_1: 0.6561, loss_yns_1: 0.1354, loss_cls_2: 0.7985, loss_box_2: 1.4673, loss_cns_2: 0.6525, loss_yns_2: 0.1351, loss_cls_3: 0.7967, loss_box_3: 1.4653, loss_cns_3: 0.6525, loss_yns_3: 0.1344, loss_cls_4: 0.8123, loss_box_4: 1.4487, loss_cns_4: 0.6508, loss_yns_4: 0.1365, loss_cls_5: 0.8118, loss_box_5: 1.4508, loss_cns_5: 0.6525, loss_yns_5: 0.1341, loss_cls_dn_0: 0.1119, loss_box_dn_0: 0.7032, loss_cls_dn_1: 0.0914, loss_box_dn_1: 0.6274, loss_cls_dn_2: 0.0910, loss_box_dn_2: 0.6167, loss_cls_dn_3: 0.0909, loss_box_dn_3: 0.6118, loss_cls_dn_4: 0.0927, loss_box_dn_4: 0.6121, loss_cls_dn_5: 0.0951, loss_box_dn_5: 0.6109, loss_dense_depth: 0.7472, loss: 23.4342, grad_norm: 29.5179 -2025-11-17 14:30:04,004 - mmdet - INFO - Iter [388/17500] lr: 2.545e-04, eta: 8:50:48, time: 1.495, data_time: 0.085, memory: 49163, loss_cls_0: 0.7110, loss_box_0: 1.5763, loss_cns_0: 0.6334, loss_yns_0: 0.1387, loss_cls_1: 0.7599, loss_box_1: 1.5153, loss_cns_1: 0.6541, loss_yns_1: 0.1360, loss_cls_2: 0.7815, loss_box_2: 1.4708, loss_cns_2: 0.6549, loss_yns_2: 0.1350, loss_cls_3: 0.7849, loss_box_3: 1.4762, loss_cns_3: 0.6530, loss_yns_3: 0.1345, loss_cls_4: 0.7935, loss_box_4: 1.4735, loss_cns_4: 0.6501, loss_yns_4: 0.1366, loss_cls_5: 0.7971, loss_box_5: 1.4774, loss_cns_5: 0.6522, loss_yns_5: 0.1348, loss_cls_dn_0: 0.1110, loss_box_dn_0: 0.7082, loss_cls_dn_1: 0.0898, loss_box_dn_1: 0.6274, loss_cls_dn_2: 0.0890, loss_box_dn_2: 0.6146, loss_cls_dn_3: 0.0898, loss_box_dn_3: 0.6162, loss_cls_dn_4: 0.0914, loss_box_dn_4: 0.6195, loss_cls_dn_5: 0.0927, loss_box_dn_5: 0.6237, loss_dense_depth: 0.7638, loss: 23.4681, grad_norm: 48.6972 -2025-11-17 14:30:09,026 - mmdet - INFO - Iter [389/17500] lr: 2.549e-04, eta: 8:53:05, time: 5.023, data_time: 0.173, memory: 49163, loss_cls_0: 0.7231, loss_box_0: 1.5812, loss_cns_0: 0.6395, loss_yns_0: 0.1417, loss_cls_1: 0.7697, loss_box_1: 1.5023, loss_cns_1: 0.6553, loss_yns_1: 0.1394, loss_cls_2: 0.7912, loss_box_2: 1.4642, loss_cns_2: 0.6584, loss_yns_2: 0.1391, loss_cls_3: 0.7880, loss_box_3: 1.4507, loss_cns_3: 0.6554, loss_yns_3: 0.1386, loss_cls_4: 0.7913, loss_box_4: 1.4587, loss_cns_4: 0.6557, loss_yns_4: 0.1394, loss_cls_5: 0.7997, loss_box_5: 1.4729, loss_cns_5: 0.6558, loss_yns_5: 0.1402, loss_cls_dn_0: 0.1079, loss_box_dn_0: 0.7038, loss_cls_dn_1: 0.0893, loss_box_dn_1: 0.6376, loss_cls_dn_2: 0.0889, loss_box_dn_2: 0.6247, loss_cls_dn_3: 0.0887, loss_box_dn_3: 0.6237, loss_cls_dn_4: 0.0900, loss_box_dn_4: 0.6274, loss_cls_dn_5: 0.0905, loss_box_dn_5: 0.6348, loss_dense_depth: 0.8447, loss: 23.6037, grad_norm: 45.9348 -2025-11-17 14:30:10,504 - mmdet - INFO - Iter [390/17500] lr: 2.553e-04, eta: 8:52:46, time: 1.479, data_time: 0.073, memory: 49163, loss_cls_0: 0.7227, loss_box_0: 1.6153, loss_cns_0: 0.6321, loss_yns_0: 0.1462, loss_cls_1: 0.7676, loss_box_1: 1.4891, loss_cns_1: 0.6584, loss_yns_1: 0.1433, loss_cls_2: 0.7958, loss_box_2: 1.4500, loss_cns_2: 0.6594, loss_yns_2: 0.1421, loss_cls_3: 0.7857, loss_box_3: 1.4262, loss_cns_3: 0.6597, loss_yns_3: 0.1426, loss_cls_4: 0.7929, loss_box_4: 1.4210, loss_cns_4: 0.6609, loss_yns_4: 0.1430, loss_cls_5: 0.8016, loss_box_5: 1.4221, loss_cns_5: 0.6610, loss_yns_5: 0.1426, loss_cls_dn_0: 0.1084, loss_box_dn_0: 0.7178, loss_cls_dn_1: 0.0918, loss_box_dn_1: 0.6287, loss_cls_dn_2: 0.0904, loss_box_dn_2: 0.6131, loss_cls_dn_3: 0.0895, loss_box_dn_3: 0.6078, loss_cls_dn_4: 0.0902, loss_box_dn_4: 0.6077, loss_cls_dn_5: 0.0913, loss_box_dn_5: 0.6101, loss_dense_depth: 0.7602, loss: 23.3883, grad_norm: 28.2437 -2025-11-17 14:30:11,987 - mmdet - INFO - Iter [391/17500] lr: 2.557e-04, eta: 8:52:27, time: 1.481, data_time: 0.078, memory: 49163, loss_cls_0: 0.7095, loss_box_0: 1.5818, loss_cns_0: 0.6351, loss_yns_0: 0.1405, loss_cls_1: 0.7724, loss_box_1: 1.4604, loss_cns_1: 0.6623, loss_yns_1: 0.1377, loss_cls_2: 0.7885, loss_box_2: 1.4349, loss_cns_2: 0.6612, loss_yns_2: 0.1384, loss_cls_3: 0.7911, loss_box_3: 1.4246, loss_cns_3: 0.6625, loss_yns_3: 0.1394, loss_cls_4: 0.7977, loss_box_4: 1.4203, loss_cns_4: 0.6638, loss_yns_4: 0.1389, loss_cls_5: 0.8038, loss_box_5: 1.4197, loss_cns_5: 0.6647, loss_yns_5: 0.1390, loss_cls_dn_0: 0.1103, loss_box_dn_0: 0.7059, loss_cls_dn_1: 0.0950, loss_box_dn_1: 0.6363, loss_cls_dn_2: 0.0932, loss_box_dn_2: 0.6238, loss_cls_dn_3: 0.0942, loss_box_dn_3: 0.6215, loss_cls_dn_4: 0.0960, loss_box_dn_4: 0.6214, loss_cls_dn_5: 0.0972, loss_box_dn_5: 0.6250, loss_dense_depth: 0.7977, loss: 23.4057, grad_norm: 39.0105 -2025-11-17 14:30:13,512 - mmdet - INFO - Iter [392/17500] lr: 2.561e-04, eta: 8:52:11, time: 1.524, data_time: 0.076, memory: 49163, loss_cls_0: 0.7404, loss_box_0: 1.5909, loss_cns_0: 0.6332, loss_yns_0: 0.1435, loss_cls_1: 0.8049, loss_box_1: 1.4800, loss_cns_1: 0.6652, loss_yns_1: 0.1419, loss_cls_2: 0.8175, loss_box_2: 1.4411, loss_cns_2: 0.6646, loss_yns_2: 0.1414, loss_cls_3: 0.8080, loss_box_3: 1.4411, loss_cns_3: 0.6654, loss_yns_3: 0.1415, loss_cls_4: 0.8165, loss_box_4: 1.4334, loss_cns_4: 0.6654, loss_yns_4: 0.1420, loss_cls_5: 0.8157, loss_box_5: 1.4364, loss_cns_5: 0.6664, loss_yns_5: 0.1420, loss_cls_dn_0: 0.1151, loss_box_dn_0: 0.7155, loss_cls_dn_1: 0.0923, loss_box_dn_1: 0.6390, loss_cls_dn_2: 0.0923, loss_box_dn_2: 0.6207, loss_cls_dn_3: 0.0932, loss_box_dn_3: 0.6189, loss_cls_dn_4: 0.0947, loss_box_dn_4: 0.6164, loss_cls_dn_5: 0.0953, loss_box_dn_5: 0.6202, loss_dense_depth: 0.7910, loss: 23.6428, grad_norm: 29.2127 -2025-11-17 14:30:15,035 - mmdet - INFO - Iter [393/17500] lr: 2.565e-04, eta: 8:51:54, time: 1.524, data_time: 0.078, memory: 49163, loss_cls_0: 0.7474, loss_box_0: 1.6174, loss_cns_0: 0.6318, loss_yns_0: 0.1449, loss_cls_1: 0.8098, loss_box_1: 1.4813, loss_cns_1: 0.6614, loss_yns_1: 0.1422, loss_cls_2: 0.8223, loss_box_2: 1.4373, loss_cns_2: 0.6626, loss_yns_2: 0.1417, loss_cls_3: 0.8188, loss_box_3: 1.4388, loss_cns_3: 0.6603, loss_yns_3: 0.1409, loss_cls_4: 0.8235, loss_box_4: 1.4314, loss_cns_4: 0.6608, loss_yns_4: 0.1417, loss_cls_5: 0.8264, loss_box_5: 1.4320, loss_cns_5: 0.6605, loss_yns_5: 0.1410, loss_cls_dn_0: 0.1161, loss_box_dn_0: 0.7347, loss_cls_dn_1: 0.0922, loss_box_dn_1: 0.6564, loss_cls_dn_2: 0.0920, loss_box_dn_2: 0.6305, loss_cls_dn_3: 0.0915, loss_box_dn_3: 0.6270, loss_cls_dn_4: 0.0918, loss_box_dn_4: 0.6242, loss_cls_dn_5: 0.0929, loss_box_dn_5: 0.6252, loss_dense_depth: 0.7898, loss: 23.7405, grad_norm: 36.0218 -2025-11-17 14:30:16,580 - mmdet - INFO - Iter [394/17500] lr: 2.569e-04, eta: 8:51:38, time: 1.546, data_time: 0.073, memory: 49163, loss_cls_0: 0.7167, loss_box_0: 1.5505, loss_cns_0: 0.6360, loss_yns_0: 0.1447, loss_cls_1: 0.7939, loss_box_1: 1.4342, loss_cns_1: 0.6614, loss_yns_1: 0.1420, loss_cls_2: 0.8020, loss_box_2: 1.4037, loss_cns_2: 0.6595, loss_yns_2: 0.1436, loss_cls_3: 0.8071, loss_box_3: 1.3884, loss_cns_3: 0.6576, loss_yns_3: 0.1422, loss_cls_4: 0.8041, loss_box_4: 1.3902, loss_cns_4: 0.6589, loss_yns_4: 0.1440, loss_cls_5: 0.8061, loss_box_5: 1.3882, loss_cns_5: 0.6588, loss_yns_5: 0.1424, loss_cls_dn_0: 0.1162, loss_box_dn_0: 0.7092, loss_cls_dn_1: 0.0952, loss_box_dn_1: 0.6323, loss_cls_dn_2: 0.0936, loss_box_dn_2: 0.6116, loss_cls_dn_3: 0.0928, loss_box_dn_3: 0.6052, loss_cls_dn_4: 0.0928, loss_box_dn_4: 0.6033, loss_cls_dn_5: 0.0935, loss_box_dn_5: 0.6033, loss_dense_depth: 0.8743, loss: 23.2995, grad_norm: 31.2336 -2025-11-17 14:30:18,068 - mmdet - INFO - Iter [395/17500] lr: 2.573e-04, eta: 8:51:20, time: 1.488, data_time: 0.071, memory: 49163, loss_cls_0: 0.7255, loss_box_0: 1.5696, loss_cns_0: 0.6371, loss_yns_0: 0.1481, loss_cls_1: 0.7861, loss_box_1: 1.4132, loss_cns_1: 0.6596, loss_yns_1: 0.1452, loss_cls_2: 0.7910, loss_box_2: 1.3941, loss_cns_2: 0.6562, loss_yns_2: 0.1438, loss_cls_3: 0.7890, loss_box_3: 1.3807, loss_cns_3: 0.6563, loss_yns_3: 0.1443, loss_cls_4: 0.7945, loss_box_4: 1.3808, loss_cns_4: 0.6580, loss_yns_4: 0.1438, loss_cls_5: 0.7989, loss_box_5: 1.3741, loss_cns_5: 0.6565, loss_yns_5: 0.1419, loss_cls_dn_0: 0.1135, loss_box_dn_0: 0.7153, loss_cls_dn_1: 0.0936, loss_box_dn_1: 0.6232, loss_cls_dn_2: 0.0901, loss_box_dn_2: 0.6146, loss_cls_dn_3: 0.0893, loss_box_dn_3: 0.6097, loss_cls_dn_4: 0.0905, loss_box_dn_4: 0.6067, loss_cls_dn_5: 0.0906, loss_box_dn_5: 0.6076, loss_dense_depth: 0.7888, loss: 23.1220, grad_norm: 35.3314 -2025-11-17 14:30:19,564 - mmdet - INFO - Iter [396/17500] lr: 2.577e-04, eta: 8:51:02, time: 1.496, data_time: 0.072, memory: 49163, loss_cls_0: 0.7235, loss_box_0: 1.5477, loss_cns_0: 0.6385, loss_yns_0: 0.1452, loss_cls_1: 0.7848, loss_box_1: 1.4065, loss_cns_1: 0.6598, loss_yns_1: 0.1427, loss_cls_2: 0.7941, loss_box_2: 1.3633, loss_cns_2: 0.6609, loss_yns_2: 0.1418, loss_cls_3: 0.8003, loss_box_3: 1.3472, loss_cns_3: 0.6578, loss_yns_3: 0.1416, loss_cls_4: 0.8123, loss_box_4: 1.3449, loss_cns_4: 0.6604, loss_yns_4: 0.1417, loss_cls_5: 0.8135, loss_box_5: 1.3430, loss_cns_5: 0.6599, loss_yns_5: 0.1404, loss_cls_dn_0: 0.1090, loss_box_dn_0: 0.7241, loss_cls_dn_1: 0.0915, loss_box_dn_1: 0.6292, loss_cls_dn_2: 0.0899, loss_box_dn_2: 0.6092, loss_cls_dn_3: 0.0896, loss_box_dn_3: 0.6076, loss_cls_dn_4: 0.0917, loss_box_dn_4: 0.6028, loss_cls_dn_5: 0.0923, loss_box_dn_5: 0.6046, loss_dense_depth: 0.8202, loss: 23.0336, grad_norm: 37.1008 -2025-11-17 14:30:21,100 - mmdet - INFO - Iter [397/17500] lr: 2.581e-04, eta: 8:50:46, time: 1.533, data_time: 0.069, memory: 49163, loss_cls_0: 0.7297, loss_box_0: 1.5707, loss_cns_0: 0.6360, loss_yns_0: 0.1463, loss_cls_1: 0.7825, loss_box_1: 1.4167, loss_cns_1: 0.6593, loss_yns_1: 0.1442, loss_cls_2: 0.7899, loss_box_2: 1.3935, loss_cns_2: 0.6595, loss_yns_2: 0.1426, loss_cls_3: 0.7945, loss_box_3: 1.3986, loss_cns_3: 0.6616, loss_yns_3: 0.1435, loss_cls_4: 0.8021, loss_box_4: 1.3951, loss_cns_4: 0.6596, loss_yns_4: 0.1438, loss_cls_5: 0.8072, loss_box_5: 1.4046, loss_cns_5: 0.6610, loss_yns_5: 0.1435, loss_cls_dn_0: 0.1104, loss_box_dn_0: 0.7030, loss_cls_dn_1: 0.0921, loss_box_dn_1: 0.6233, loss_cls_dn_2: 0.0917, loss_box_dn_2: 0.6065, loss_cls_dn_3: 0.0906, loss_box_dn_3: 0.6061, loss_cls_dn_4: 0.0922, loss_box_dn_4: 0.6045, loss_cls_dn_5: 0.0942, loss_box_dn_5: 0.6102, loss_dense_depth: 0.7843, loss: 23.1950, grad_norm: 37.5291 -2025-11-17 14:30:22,594 - mmdet - INFO - Iter [398/17500] lr: 2.585e-04, eta: 8:50:28, time: 1.495, data_time: 0.072, memory: 49163, loss_cls_0: 0.7528, loss_box_0: 1.5949, loss_cns_0: 0.6330, loss_yns_0: 0.1474, loss_cls_1: 0.8018, loss_box_1: 1.4409, loss_cns_1: 0.6592, loss_yns_1: 0.1455, loss_cls_2: 0.8172, loss_box_2: 1.4370, loss_cns_2: 0.6605, loss_yns_2: 0.1439, loss_cls_3: 0.8281, loss_box_3: 1.4198, loss_cns_3: 0.6588, loss_yns_3: 0.1446, loss_cls_4: 0.8261, loss_box_4: 1.4286, loss_cns_4: 0.6587, loss_yns_4: 0.1450, loss_cls_5: 0.8321, loss_box_5: 1.4306, loss_cns_5: 0.6591, loss_yns_5: 0.1439, loss_cls_dn_0: 0.1174, loss_box_dn_0: 0.7062, loss_cls_dn_1: 0.0921, loss_box_dn_1: 0.6201, loss_cls_dn_2: 0.0921, loss_box_dn_2: 0.6094, loss_cls_dn_3: 0.0902, loss_box_dn_3: 0.6052, loss_cls_dn_4: 0.0913, loss_box_dn_4: 0.6077, loss_cls_dn_5: 0.0933, loss_box_dn_5: 0.6125, loss_dense_depth: 0.7931, loss: 23.5400, grad_norm: 37.2931 -2025-11-17 14:30:24,086 - mmdet - INFO - Iter [399/17500] lr: 2.589e-04, eta: 8:50:11, time: 1.493, data_time: 0.076, memory: 49163, loss_cls_0: 0.7291, loss_box_0: 1.5617, loss_cns_0: 0.6335, loss_yns_0: 0.1447, loss_cls_1: 0.7962, loss_box_1: 1.4382, loss_cns_1: 0.6633, loss_yns_1: 0.1444, loss_cls_2: 0.8082, loss_box_2: 1.4180, loss_cns_2: 0.6637, loss_yns_2: 0.1443, loss_cls_3: 0.8159, loss_box_3: 1.3985, loss_cns_3: 0.6633, loss_yns_3: 0.1445, loss_cls_4: 0.8166, loss_box_4: 1.4002, loss_cns_4: 0.6651, loss_yns_4: 0.1457, loss_cls_5: 0.8217, loss_box_5: 1.4015, loss_cns_5: 0.6673, loss_yns_5: 0.1437, loss_cls_dn_0: 0.1193, loss_box_dn_0: 0.7060, loss_cls_dn_1: 0.0947, loss_box_dn_1: 0.6166, loss_cls_dn_2: 0.0945, loss_box_dn_2: 0.6069, loss_cls_dn_3: 0.0927, loss_box_dn_3: 0.6006, loss_cls_dn_4: 0.0937, loss_box_dn_4: 0.6025, loss_cls_dn_5: 0.0939, loss_box_dn_5: 0.6066, loss_dense_depth: 0.7928, loss: 23.3501, grad_norm: 32.9680 -2025-11-17 14:30:25,568 - mmdet - INFO - Iter [400/17500] lr: 2.593e-04, eta: 8:49:53, time: 1.482, data_time: 0.073, memory: 49163, loss_cls_0: 0.7382, loss_box_0: 1.5541, loss_cns_0: 0.6315, loss_yns_0: 0.1429, loss_cls_1: 0.8035, loss_box_1: 1.4203, loss_cns_1: 0.6635, loss_yns_1: 0.1425, loss_cls_2: 0.8100, loss_box_2: 1.4024, loss_cns_2: 0.6625, loss_yns_2: 0.1414, loss_cls_3: 0.8204, loss_box_3: 1.3929, loss_cns_3: 0.6631, loss_yns_3: 0.1418, loss_cls_4: 0.8234, loss_box_4: 1.3958, loss_cns_4: 0.6649, loss_yns_4: 0.1415, loss_cls_5: 0.8352, loss_box_5: 1.3986, loss_cns_5: 0.6666, loss_yns_5: 0.1415, loss_cls_dn_0: 0.1164, loss_box_dn_0: 0.7077, loss_cls_dn_1: 0.0914, loss_box_dn_1: 0.6259, loss_cls_dn_2: 0.0910, loss_box_dn_2: 0.6198, loss_cls_dn_3: 0.0921, loss_box_dn_3: 0.6135, loss_cls_dn_4: 0.0912, loss_box_dn_4: 0.6156, loss_cls_dn_5: 0.0927, loss_box_dn_5: 0.6177, loss_dense_depth: 0.7566, loss: 23.3303, grad_norm: 37.5038 -2025-11-17 14:30:27,105 - mmdet - INFO - Iter [401/17500] lr: 2.597e-04, eta: 8:49:37, time: 1.538, data_time: 0.135, memory: 49163, loss_cls_0: 0.7498, loss_box_0: 1.5676, loss_cns_0: 0.6304, loss_yns_0: 0.1442, loss_cls_1: 0.8183, loss_box_1: 1.4423, loss_cns_1: 0.6588, loss_yns_1: 0.1448, loss_cls_2: 0.8362, loss_box_2: 1.4114, loss_cns_2: 0.6586, loss_yns_2: 0.1434, loss_cls_3: 0.8357, loss_box_3: 1.3920, loss_cns_3: 0.6595, loss_yns_3: 0.1429, loss_cls_4: 0.8266, loss_box_4: 1.4020, loss_cns_4: 0.6603, loss_yns_4: 0.1441, loss_cls_5: 0.8308, loss_box_5: 1.4060, loss_cns_5: 0.6607, loss_yns_5: 0.1433, loss_cls_dn_0: 0.1159, loss_box_dn_0: 0.7109, loss_cls_dn_1: 0.0944, loss_box_dn_1: 0.6113, loss_cls_dn_2: 0.0930, loss_box_dn_2: 0.6006, loss_cls_dn_3: 0.0939, loss_box_dn_3: 0.5944, loss_cls_dn_4: 0.0928, loss_box_dn_4: 0.5954, loss_cls_dn_5: 0.0947, loss_box_dn_5: 0.5966, loss_dense_depth: 0.7779, loss: 23.3814, grad_norm: 30.6159 -2025-11-17 14:30:28,643 - mmdet - INFO - Iter [402/17500] lr: 2.601e-04, eta: 8:49:22, time: 1.538, data_time: 0.069, memory: 49163, loss_cls_0: 0.7258, loss_box_0: 1.5698, loss_cns_0: 0.6351, loss_yns_0: 0.1451, loss_cls_1: 0.7877, loss_box_1: 1.4495, loss_cns_1: 0.6596, loss_yns_1: 0.1432, loss_cls_2: 0.8089, loss_box_2: 1.4133, loss_cns_2: 0.6598, loss_yns_2: 0.1422, loss_cls_3: 0.8100, loss_box_3: 1.3981, loss_cns_3: 0.6605, loss_yns_3: 0.1416, loss_cls_4: 0.8116, loss_box_4: 1.4048, loss_cns_4: 0.6615, loss_yns_4: 0.1429, loss_cls_5: 0.8159, loss_box_5: 1.4127, loss_cns_5: 0.6619, loss_yns_5: 0.1420, loss_cls_dn_0: 0.1106, loss_box_dn_0: 0.7076, loss_cls_dn_1: 0.0906, loss_box_dn_1: 0.6226, loss_cls_dn_2: 0.0895, loss_box_dn_2: 0.6073, loss_cls_dn_3: 0.0893, loss_box_dn_3: 0.6032, loss_cls_dn_4: 0.0901, loss_box_dn_4: 0.6035, loss_cls_dn_5: 0.0907, loss_box_dn_5: 0.6070, loss_dense_depth: 0.7318, loss: 23.2474, grad_norm: 32.8478 -2025-11-17 14:30:30,136 - mmdet - INFO - Iter [403/17500] lr: 2.605e-04, eta: 8:49:04, time: 1.493, data_time: 0.076, memory: 49163, loss_cls_0: 0.7359, loss_box_0: 1.5852, loss_cns_0: 0.6369, loss_yns_0: 0.1414, loss_cls_1: 0.7951, loss_box_1: 1.4527, loss_cns_1: 0.6600, loss_yns_1: 0.1405, loss_cls_2: 0.8081, loss_box_2: 1.4120, loss_cns_2: 0.6613, loss_yns_2: 0.1400, loss_cls_3: 0.8170, loss_box_3: 1.4071, loss_cns_3: 0.6601, loss_yns_3: 0.1395, loss_cls_4: 0.8229, loss_box_4: 1.4063, loss_cns_4: 0.6600, loss_yns_4: 0.1391, loss_cls_5: 0.8282, loss_box_5: 1.4045, loss_cns_5: 0.6603, loss_yns_5: 0.1376, loss_cls_dn_0: 0.1122, loss_box_dn_0: 0.7191, loss_cls_dn_1: 0.0909, loss_box_dn_1: 0.6296, loss_cls_dn_2: 0.0896, loss_box_dn_2: 0.6100, loss_cls_dn_3: 0.0895, loss_box_dn_3: 0.6076, loss_cls_dn_4: 0.0905, loss_box_dn_4: 0.6061, loss_cls_dn_5: 0.0908, loss_box_dn_5: 0.6076, loss_dense_depth: 0.7451, loss: 23.3405, grad_norm: 23.0301 -2025-11-17 14:30:31,691 - mmdet - INFO - Iter [404/17500] lr: 2.609e-04, eta: 8:48:50, time: 1.554, data_time: 0.077, memory: 49163, loss_cls_0: 0.7202, loss_box_0: 1.5965, loss_cns_0: 0.6344, loss_yns_0: 0.1417, loss_cls_1: 0.7945, loss_box_1: 1.4526, loss_cns_1: 0.6591, loss_yns_1: 0.1389, loss_cls_2: 0.8170, loss_box_2: 1.4273, loss_cns_2: 0.6605, loss_yns_2: 0.1384, loss_cls_3: 0.8198, loss_box_3: 1.4176, loss_cns_3: 0.6615, loss_yns_3: 0.1383, loss_cls_4: 0.8125, loss_box_4: 1.4125, loss_cns_4: 0.6602, loss_yns_4: 0.1387, loss_cls_5: 0.8167, loss_box_5: 1.4114, loss_cns_5: 0.6605, loss_yns_5: 0.1384, loss_cls_dn_0: 0.1143, loss_box_dn_0: 0.7192, loss_cls_dn_1: 0.0914, loss_box_dn_1: 0.6137, loss_cls_dn_2: 0.0918, loss_box_dn_2: 0.6002, loss_cls_dn_3: 0.0929, loss_box_dn_3: 0.5978, loss_cls_dn_4: 0.0929, loss_box_dn_4: 0.5978, loss_cls_dn_5: 0.0946, loss_box_dn_5: 0.5979, loss_dense_depth: 0.7211, loss: 23.2947, grad_norm: 31.8344 -2025-11-17 14:30:33,199 - mmdet - INFO - Iter [405/17500] lr: 2.613e-04, eta: 8:48:33, time: 1.509, data_time: 0.080, memory: 49163, loss_cls_0: 0.7431, loss_box_0: 1.6077, loss_cns_0: 0.6332, loss_yns_0: 0.1406, loss_cls_1: 0.7978, loss_box_1: 1.4684, loss_cns_1: 0.6586, loss_yns_1: 0.1381, loss_cls_2: 0.8169, loss_box_2: 1.4472, loss_cns_2: 0.6604, loss_yns_2: 0.1399, loss_cls_3: 0.8196, loss_box_3: 1.4364, loss_cns_3: 0.6602, loss_yns_3: 0.1403, loss_cls_4: 0.8275, loss_box_4: 1.4300, loss_cns_4: 0.6573, loss_yns_4: 0.1396, loss_cls_5: 0.8303, loss_box_5: 1.4314, loss_cns_5: 0.6577, loss_yns_5: 0.1392, loss_cls_dn_0: 0.1153, loss_box_dn_0: 0.7194, loss_cls_dn_1: 0.0938, loss_box_dn_1: 0.6232, loss_cls_dn_2: 0.0940, loss_box_dn_2: 0.6067, loss_cls_dn_3: 0.0940, loss_box_dn_3: 0.6027, loss_cls_dn_4: 0.0949, loss_box_dn_4: 0.6025, loss_cls_dn_5: 0.0953, loss_box_dn_5: 0.6038, loss_dense_depth: 0.7437, loss: 23.5104, grad_norm: 31.4703 -2025-11-17 14:30:34,726 - mmdet - INFO - Iter [406/17500] lr: 2.617e-04, eta: 8:48:17, time: 1.526, data_time: 0.089, memory: 49163, loss_cls_0: 0.7302, loss_box_0: 1.5863, loss_cns_0: 0.6348, loss_yns_0: 0.1379, loss_cls_1: 0.7820, loss_box_1: 1.4728, loss_cns_1: 0.6592, loss_yns_1: 0.1367, loss_cls_2: 0.7945, loss_box_2: 1.4486, loss_cns_2: 0.6605, loss_yns_2: 0.1375, loss_cls_3: 0.7951, loss_box_3: 1.4453, loss_cns_3: 0.6598, loss_yns_3: 0.1367, loss_cls_4: 0.8046, loss_box_4: 1.4309, loss_cns_4: 0.6591, loss_yns_4: 0.1360, loss_cls_5: 0.8101, loss_box_5: 1.4299, loss_cns_5: 0.6591, loss_yns_5: 0.1352, loss_cls_dn_0: 0.1139, loss_box_dn_0: 0.7149, loss_cls_dn_1: 0.0940, loss_box_dn_1: 0.6265, loss_cls_dn_2: 0.0935, loss_box_dn_2: 0.6037, loss_cls_dn_3: 0.0927, loss_box_dn_3: 0.6018, loss_cls_dn_4: 0.0936, loss_box_dn_4: 0.6005, loss_cls_dn_5: 0.0947, loss_box_dn_5: 0.6004, loss_dense_depth: 0.7094, loss: 23.3224, grad_norm: 25.5421 -2025-11-17 14:30:36,245 - mmdet - INFO - Iter [407/17500] lr: 2.621e-04, eta: 8:48:02, time: 1.519, data_time: 0.079, memory: 49163, loss_cls_0: 0.7361, loss_box_0: 1.5929, loss_cns_0: 0.6380, loss_yns_0: 0.1400, loss_cls_1: 0.7832, loss_box_1: 1.4558, loss_cns_1: 0.6639, loss_yns_1: 0.1367, loss_cls_2: 0.7966, loss_box_2: 1.4321, loss_cns_2: 0.6621, loss_yns_2: 0.1358, loss_cls_3: 0.8064, loss_box_3: 1.4326, loss_cns_3: 0.6614, loss_yns_3: 0.1360, loss_cls_4: 0.7998, loss_box_4: 1.4284, loss_cns_4: 0.6612, loss_yns_4: 0.1356, loss_cls_5: 0.8022, loss_box_5: 1.4182, loss_cns_5: 0.6604, loss_yns_5: 0.1356, loss_cls_dn_0: 0.1141, loss_box_dn_0: 0.7111, loss_cls_dn_1: 0.0926, loss_box_dn_1: 0.6324, loss_cls_dn_2: 0.0934, loss_box_dn_2: 0.6186, loss_cls_dn_3: 0.0932, loss_box_dn_3: 0.6204, loss_cls_dn_4: 0.0951, loss_box_dn_4: 0.6201, loss_cls_dn_5: 0.0971, loss_box_dn_5: 0.6216, loss_dense_depth: 0.7270, loss: 23.3876, grad_norm: 39.4182 -2025-11-17 14:30:37,738 - mmdet - INFO - Iter [408/17500] lr: 2.624e-04, eta: 8:47:45, time: 1.494, data_time: 0.078, memory: 49163, loss_cls_0: 0.7417, loss_box_0: 1.5991, loss_cns_0: 0.6382, loss_yns_0: 0.1437, loss_cls_1: 0.7766, loss_box_1: 1.4552, loss_cns_1: 0.6651, loss_yns_1: 0.1401, loss_cls_2: 0.7842, loss_box_2: 1.4356, loss_cns_2: 0.6663, loss_yns_2: 0.1397, loss_cls_3: 0.7933, loss_box_3: 1.4305, loss_cns_3: 0.6658, loss_yns_3: 0.1400, loss_cls_4: 0.7945, loss_box_4: 1.4226, loss_cns_4: 0.6663, loss_yns_4: 0.1415, loss_cls_5: 0.7949, loss_box_5: 1.4220, loss_cns_5: 0.6657, loss_yns_5: 0.1381, loss_cls_dn_0: 0.1167, loss_box_dn_0: 0.7152, loss_cls_dn_1: 0.0960, loss_box_dn_1: 0.6442, loss_cls_dn_2: 0.0976, loss_box_dn_2: 0.6336, loss_cls_dn_3: 0.0967, loss_box_dn_3: 0.6335, loss_cls_dn_4: 0.0975, loss_box_dn_4: 0.6301, loss_cls_dn_5: 0.0981, loss_box_dn_5: 0.6328, loss_dense_depth: 0.7433, loss: 23.4962, grad_norm: 37.2831 -2025-11-17 14:30:39,260 - mmdet - INFO - Iter [409/17500] lr: 2.628e-04, eta: 8:47:29, time: 1.522, data_time: 0.113, memory: 49163, loss_cls_0: 0.7292, loss_box_0: 1.5807, loss_cns_0: 0.6425, loss_yns_0: 0.1456, loss_cls_1: 0.7787, loss_box_1: 1.4815, loss_cns_1: 0.6653, loss_yns_1: 0.1419, loss_cls_2: 0.7832, loss_box_2: 1.4557, loss_cns_2: 0.6636, loss_yns_2: 0.1409, loss_cls_3: 0.7906, loss_box_3: 1.4389, loss_cns_3: 0.6613, loss_yns_3: 0.1404, loss_cls_4: 0.7939, loss_box_4: 1.4502, loss_cns_4: 0.6628, loss_yns_4: 0.1437, loss_cls_5: 0.7948, loss_box_5: 1.4412, loss_cns_5: 0.6642, loss_yns_5: 0.1396, loss_cls_dn_0: 0.1148, loss_box_dn_0: 0.7107, loss_cls_dn_1: 0.0990, loss_box_dn_1: 0.6369, loss_cls_dn_2: 0.0990, loss_box_dn_2: 0.6166, loss_cls_dn_3: 0.0970, loss_box_dn_3: 0.6113, loss_cls_dn_4: 0.0993, loss_box_dn_4: 0.6117, loss_cls_dn_5: 0.0987, loss_box_dn_5: 0.6105, loss_dense_depth: 0.6900, loss: 23.4258, grad_norm: 31.9445 -2025-11-17 14:30:40,751 - mmdet - INFO - Iter [410/17500] lr: 2.632e-04, eta: 8:47:12, time: 1.490, data_time: 0.079, memory: 49163, loss_cls_0: 0.7284, loss_box_0: 1.5528, loss_cns_0: 0.6401, loss_yns_0: 0.1425, loss_cls_1: 0.7816, loss_box_1: 1.4261, loss_cns_1: 0.6629, loss_yns_1: 0.1392, loss_cls_2: 0.7796, loss_box_2: 1.4011, loss_cns_2: 0.6625, loss_yns_2: 0.1403, loss_cls_3: 0.7865, loss_box_3: 1.3942, loss_cns_3: 0.6627, loss_yns_3: 0.1393, loss_cls_4: 0.7945, loss_box_4: 1.3984, loss_cns_4: 0.6636, loss_yns_4: 0.1424, loss_cls_5: 0.7987, loss_box_5: 1.3912, loss_cns_5: 0.6665, loss_yns_5: 0.1412, loss_cls_dn_0: 0.1104, loss_box_dn_0: 0.7085, loss_cls_dn_1: 0.0962, loss_box_dn_1: 0.6224, loss_cls_dn_2: 0.0949, loss_box_dn_2: 0.6003, loss_cls_dn_3: 0.0947, loss_box_dn_3: 0.5974, loss_cls_dn_4: 0.0972, loss_box_dn_4: 0.5988, loss_cls_dn_5: 0.0988, loss_box_dn_5: 0.5972, loss_dense_depth: 0.6934, loss: 23.0463, grad_norm: 34.5043 -2025-11-17 14:30:42,252 - mmdet - INFO - Iter [411/17500] lr: 2.636e-04, eta: 8:46:56, time: 1.502, data_time: 0.079, memory: 49163, loss_cls_0: 0.7310, loss_box_0: 1.5755, loss_cns_0: 0.6386, loss_yns_0: 0.1416, loss_cls_1: 0.7883, loss_box_1: 1.4319, loss_cns_1: 0.6624, loss_yns_1: 0.1401, loss_cls_2: 0.7890, loss_box_2: 1.4181, loss_cns_2: 0.6596, loss_yns_2: 0.1388, loss_cls_3: 0.7971, loss_box_3: 1.4076, loss_cns_3: 0.6587, loss_yns_3: 0.1392, loss_cls_4: 0.8137, loss_box_4: 1.4047, loss_cns_4: 0.6622, loss_yns_4: 0.1389, loss_cls_5: 0.8132, loss_box_5: 1.4027, loss_cns_5: 0.6648, loss_yns_5: 0.1402, loss_cls_dn_0: 0.1125, loss_box_dn_0: 0.7072, loss_cls_dn_1: 0.0933, loss_box_dn_1: 0.6369, loss_cls_dn_2: 0.0935, loss_box_dn_2: 0.6241, loss_cls_dn_3: 0.0941, loss_box_dn_3: 0.6226, loss_cls_dn_4: 0.0968, loss_box_dn_4: 0.6208, loss_cls_dn_5: 0.0977, loss_box_dn_5: 0.6178, loss_dense_depth: 0.7154, loss: 23.2904, grad_norm: 36.7079 -2025-11-17 14:30:43,764 - mmdet - INFO - Iter [412/17500] lr: 2.640e-04, eta: 8:46:40, time: 1.512, data_time: 0.080, memory: 49163, loss_cls_0: 0.7089, loss_box_0: 1.5762, loss_cns_0: 0.6414, loss_yns_0: 0.1404, loss_cls_1: 0.7766, loss_box_1: 1.4517, loss_cns_1: 0.6606, loss_yns_1: 0.1394, loss_cls_2: 0.7760, loss_box_2: 1.4202, loss_cns_2: 0.6569, loss_yns_2: 0.1370, loss_cls_3: 0.7796, loss_box_3: 1.4077, loss_cns_3: 0.6562, loss_yns_3: 0.1361, loss_cls_4: 0.7930, loss_box_4: 1.4187, loss_cns_4: 0.6603, loss_yns_4: 0.1388, loss_cls_5: 0.7928, loss_box_5: 1.4219, loss_cns_5: 0.6628, loss_yns_5: 0.1386, loss_cls_dn_0: 0.1136, loss_box_dn_0: 0.7128, loss_cls_dn_1: 0.0931, loss_box_dn_1: 0.6331, loss_cls_dn_2: 0.0929, loss_box_dn_2: 0.6184, loss_cls_dn_3: 0.0919, loss_box_dn_3: 0.6124, loss_cls_dn_4: 0.0914, loss_box_dn_4: 0.6130, loss_cls_dn_5: 0.0928, loss_box_dn_5: 0.6112, loss_dense_depth: 0.6706, loss: 23.1393, grad_norm: 26.1167 -2025-11-17 14:30:45,284 - mmdet - INFO - Iter [413/17500] lr: 2.644e-04, eta: 8:46:24, time: 1.520, data_time: 0.077, memory: 49163, loss_cls_0: 0.7165, loss_box_0: 1.5778, loss_cns_0: 0.6399, loss_yns_0: 0.1432, loss_cls_1: 0.7758, loss_box_1: 1.4564, loss_cns_1: 0.6583, loss_yns_1: 0.1404, loss_cls_2: 0.7872, loss_box_2: 1.4093, loss_cns_2: 0.6498, loss_yns_2: 0.1383, loss_cls_3: 0.8007, loss_box_3: 1.3842, loss_cns_3: 0.6484, loss_yns_3: 0.1372, loss_cls_4: 0.8136, loss_box_4: 1.4014, loss_cns_4: 0.6527, loss_yns_4: 0.1388, loss_cls_5: 0.8120, loss_box_5: 1.4140, loss_cns_5: 0.6551, loss_yns_5: 0.1391, loss_cls_dn_0: 0.1116, loss_box_dn_0: 0.7120, loss_cls_dn_1: 0.0904, loss_box_dn_1: 0.6295, loss_cls_dn_2: 0.0902, loss_box_dn_2: 0.6126, loss_cls_dn_3: 0.0901, loss_box_dn_3: 0.6057, loss_cls_dn_4: 0.0940, loss_box_dn_4: 0.6100, loss_cls_dn_5: 0.0983, loss_box_dn_5: 0.6124, loss_dense_depth: 0.6875, loss: 23.1349, grad_norm: 34.1199 -2025-11-17 14:30:46,828 - mmdet - INFO - Iter [414/17500] lr: 2.648e-04, eta: 8:46:10, time: 1.543, data_time: 0.072, memory: 49163, loss_cls_0: 0.7262, loss_box_0: 1.6096, loss_cns_0: 0.6412, loss_yns_0: 0.1431, loss_cls_1: 0.7724, loss_box_1: 1.4680, loss_cns_1: 0.6561, loss_yns_1: 0.1382, loss_cls_2: 0.7928, loss_box_2: 1.4295, loss_cns_2: 0.6517, loss_yns_2: 0.1385, loss_cls_3: 0.8020, loss_box_3: 1.4088, loss_cns_3: 0.6490, loss_yns_3: 0.1375, loss_cls_4: 0.8180, loss_box_4: 1.4150, loss_cns_4: 0.6514, loss_yns_4: 0.1381, loss_cls_5: 0.8076, loss_box_5: 1.4192, loss_cns_5: 0.6554, loss_yns_5: 0.1389, loss_cls_dn_0: 0.1136, loss_box_dn_0: 0.7119, loss_cls_dn_1: 0.0953, loss_box_dn_1: 0.6370, loss_cls_dn_2: 0.0948, loss_box_dn_2: 0.6188, loss_cls_dn_3: 0.0932, loss_box_dn_3: 0.6134, loss_cls_dn_4: 0.0961, loss_box_dn_4: 0.6175, loss_cls_dn_5: 0.0986, loss_box_dn_5: 0.6194, loss_dense_depth: 0.7037, loss: 23.3215, grad_norm: 33.6008 -2025-11-17 14:30:48,336 - mmdet - INFO - Iter [415/17500] lr: 2.652e-04, eta: 8:45:54, time: 1.509, data_time: 0.076, memory: 49163, loss_cls_0: 0.7370, loss_box_0: 1.5838, loss_cns_0: 0.6433, loss_yns_0: 0.1439, loss_cls_1: 0.7737, loss_box_1: 1.4535, loss_cns_1: 0.6616, loss_yns_1: 0.1394, loss_cls_2: 0.7857, loss_box_2: 1.4219, loss_cns_2: 0.6623, loss_yns_2: 0.1408, loss_cls_3: 0.7883, loss_box_3: 1.4102, loss_cns_3: 0.6612, loss_yns_3: 0.1401, loss_cls_4: 0.7945, loss_box_4: 1.4089, loss_cns_4: 0.6607, loss_yns_4: 0.1401, loss_cls_5: 0.7890, loss_box_5: 1.4105, loss_cns_5: 0.6633, loss_yns_5: 0.1395, loss_cls_dn_0: 0.1157, loss_box_dn_0: 0.7038, loss_cls_dn_1: 0.0955, loss_box_dn_1: 0.6309, loss_cls_dn_2: 0.0955, loss_box_dn_2: 0.6129, loss_cls_dn_3: 0.0926, loss_box_dn_3: 0.6099, loss_cls_dn_4: 0.0944, loss_box_dn_4: 0.6117, loss_cls_dn_5: 0.0965, loss_box_dn_5: 0.6136, loss_dense_depth: 0.6852, loss: 23.2114, grad_norm: 28.5379 -2025-11-17 14:30:49,831 - mmdet - INFO - Iter [416/17500] lr: 2.656e-04, eta: 8:45:38, time: 1.494, data_time: 0.075, memory: 49163, loss_cls_0: 0.7141, loss_box_0: 1.5875, loss_cns_0: 0.6404, loss_yns_0: 0.1437, loss_cls_1: 0.7701, loss_box_1: 1.4295, loss_cns_1: 0.6642, loss_yns_1: 0.1390, loss_cls_2: 0.7906, loss_box_2: 1.4125, loss_cns_2: 0.6646, loss_yns_2: 0.1399, loss_cls_3: 0.7935, loss_box_3: 1.4004, loss_cns_3: 0.6640, loss_yns_3: 0.1388, loss_cls_4: 0.7903, loss_box_4: 1.4051, loss_cns_4: 0.6646, loss_yns_4: 0.1402, loss_cls_5: 0.7836, loss_box_5: 1.4098, loss_cns_5: 0.6655, loss_yns_5: 0.1394, loss_cls_dn_0: 0.1170, loss_box_dn_0: 0.7143, loss_cls_dn_1: 0.0971, loss_box_dn_1: 0.6261, loss_cls_dn_2: 0.0960, loss_box_dn_2: 0.6107, loss_cls_dn_3: 0.0945, loss_box_dn_3: 0.6050, loss_cls_dn_4: 0.0980, loss_box_dn_4: 0.6064, loss_cls_dn_5: 0.1013, loss_box_dn_5: 0.6085, loss_dense_depth: 0.6831, loss: 23.1491, grad_norm: 36.1625 -2025-11-17 14:30:51,364 - mmdet - INFO - Iter [417/17500] lr: 2.660e-04, eta: 8:45:23, time: 1.535, data_time: 0.077, memory: 49163, loss_cls_0: 0.7402, loss_box_0: 1.5727, loss_cns_0: 0.6330, loss_yns_0: 0.1431, loss_cls_1: 0.7700, loss_box_1: 1.4428, loss_cns_1: 0.6612, loss_yns_1: 0.1408, loss_cls_2: 0.7879, loss_box_2: 1.4284, loss_cns_2: 0.6629, loss_yns_2: 0.1413, loss_cls_3: 0.7883, loss_box_3: 1.4141, loss_cns_3: 0.6619, loss_yns_3: 0.1417, loss_cls_4: 0.7901, loss_box_4: 1.4165, loss_cns_4: 0.6618, loss_yns_4: 0.1425, loss_cls_5: 0.7937, loss_box_5: 1.4180, loss_cns_5: 0.6623, loss_yns_5: 0.1440, loss_cls_dn_0: 0.1171, loss_box_dn_0: 0.7196, loss_cls_dn_1: 0.1001, loss_box_dn_1: 0.6339, loss_cls_dn_2: 0.0984, loss_box_dn_2: 0.6201, loss_cls_dn_3: 0.0975, loss_box_dn_3: 0.6140, loss_cls_dn_4: 0.1003, loss_box_dn_4: 0.6170, loss_cls_dn_5: 0.1030, loss_box_dn_5: 0.6183, loss_dense_depth: 0.7130, loss: 23.3118, grad_norm: 34.8120 -2025-11-17 14:30:52,853 - mmdet - INFO - Iter [418/17500] lr: 2.664e-04, eta: 8:45:07, time: 1.488, data_time: 0.077, memory: 49163, loss_cls_0: 0.7334, loss_box_0: 1.5730, loss_cns_0: 0.6321, loss_yns_0: 0.1421, loss_cls_1: 0.7673, loss_box_1: 1.4304, loss_cns_1: 0.6571, loss_yns_1: 0.1424, loss_cls_2: 0.7732, loss_box_2: 1.4190, loss_cns_2: 0.6613, loss_yns_2: 0.1435, loss_cls_3: 0.7751, loss_box_3: 1.4023, loss_cns_3: 0.6591, loss_yns_3: 0.1427, loss_cls_4: 0.7789, loss_box_4: 1.3997, loss_cns_4: 0.6587, loss_yns_4: 0.1436, loss_cls_5: 0.7773, loss_box_5: 1.4021, loss_cns_5: 0.6598, loss_yns_5: 0.1446, loss_cls_dn_0: 0.1108, loss_box_dn_0: 0.7024, loss_cls_dn_1: 0.0983, loss_box_dn_1: 0.6321, loss_cls_dn_2: 0.0963, loss_box_dn_2: 0.6183, loss_cls_dn_3: 0.0965, loss_box_dn_3: 0.6126, loss_cls_dn_4: 0.0962, loss_box_dn_4: 0.6131, loss_cls_dn_5: 0.0974, loss_box_dn_5: 0.6144, loss_dense_depth: 0.6878, loss: 23.0949, grad_norm: 28.9360 -2025-11-17 14:30:54,353 - mmdet - INFO - Iter [419/17500] lr: 2.668e-04, eta: 8:44:51, time: 1.499, data_time: 0.077, memory: 49163, loss_cls_0: 0.7736, loss_box_0: 1.6458, loss_cns_0: 0.6294, loss_yns_0: 0.1434, loss_cls_1: 0.8015, loss_box_1: 1.4756, loss_cns_1: 0.6596, loss_yns_1: 0.1415, loss_cls_2: 0.8089, loss_box_2: 1.4482, loss_cns_2: 0.6632, loss_yns_2: 0.1427, loss_cls_3: 0.8162, loss_box_3: 1.4446, loss_cns_3: 0.6631, loss_yns_3: 0.1418, loss_cls_4: 0.8168, loss_box_4: 1.4407, loss_cns_4: 0.6643, loss_yns_4: 0.1433, loss_cls_5: 0.8209, loss_box_5: 1.4457, loss_cns_5: 0.6661, loss_yns_5: 0.1422, loss_cls_dn_0: 0.1164, loss_box_dn_0: 0.7112, loss_cls_dn_1: 0.0963, loss_box_dn_1: 0.6386, loss_cls_dn_2: 0.0961, loss_box_dn_2: 0.6201, loss_cls_dn_3: 0.0970, loss_box_dn_3: 0.6170, loss_cls_dn_4: 0.0952, loss_box_dn_4: 0.6147, loss_cls_dn_5: 0.0952, loss_box_dn_5: 0.6178, loss_dense_depth: 0.7239, loss: 23.6786, grad_norm: 27.5950 -2025-11-17 14:30:55,863 - mmdet - INFO - Iter [420/17500] lr: 2.672e-04, eta: 8:44:35, time: 1.511, data_time: 0.082, memory: 49163, loss_cls_0: 0.7393, loss_box_0: 1.6413, loss_cns_0: 0.6355, loss_yns_0: 0.1431, loss_cls_1: 0.7858, loss_box_1: 1.4779, loss_cns_1: 0.6598, loss_yns_1: 0.1405, loss_cls_2: 0.8009, loss_box_2: 1.4478, loss_cns_2: 0.6613, loss_yns_2: 0.1405, loss_cls_3: 0.8110, loss_box_3: 1.4445, loss_cns_3: 0.6616, loss_yns_3: 0.1409, loss_cls_4: 0.8123, loss_box_4: 1.4369, loss_cns_4: 0.6644, loss_yns_4: 0.1406, loss_cls_5: 0.8090, loss_box_5: 1.4297, loss_cns_5: 0.6645, loss_yns_5: 0.1410, loss_cls_dn_0: 0.1164, loss_box_dn_0: 0.7118, loss_cls_dn_1: 0.0955, loss_box_dn_1: 0.6413, loss_cls_dn_2: 0.0962, loss_box_dn_2: 0.6227, loss_cls_dn_3: 0.0957, loss_box_dn_3: 0.6203, loss_cls_dn_4: 0.0943, loss_box_dn_4: 0.6176, loss_cls_dn_5: 0.0953, loss_box_dn_5: 0.6179, loss_dense_depth: 0.7074, loss: 23.5624, grad_norm: 32.2672 -2025-11-17 14:30:57,418 - mmdet - INFO - Iter [421/17500] lr: 2.676e-04, eta: 8:44:22, time: 1.554, data_time: 0.139, memory: 49163, loss_cls_0: 0.7595, loss_box_0: 1.6733, loss_cns_0: 0.6364, loss_yns_0: 0.1402, loss_cls_1: 0.7909, loss_box_1: 1.4880, loss_cns_1: 0.6605, loss_yns_1: 0.1383, loss_cls_2: 0.7961, loss_box_2: 1.4646, loss_cns_2: 0.6609, loss_yns_2: 0.1391, loss_cls_3: 0.8006, loss_box_3: 1.4567, loss_cns_3: 0.6618, loss_yns_3: 0.1395, loss_cls_4: 0.8099, loss_box_4: 1.4470, loss_cns_4: 0.6624, loss_yns_4: 0.1379, loss_cls_5: 0.8143, loss_box_5: 1.4480, loss_cns_5: 0.6620, loss_yns_5: 0.1385, loss_cls_dn_0: 0.1240, loss_box_dn_0: 0.7080, loss_cls_dn_1: 0.0927, loss_box_dn_1: 0.6428, loss_cls_dn_2: 0.0926, loss_box_dn_2: 0.6309, loss_cls_dn_3: 0.0905, loss_box_dn_3: 0.6281, loss_cls_dn_4: 0.0900, loss_box_dn_4: 0.6250, loss_cls_dn_5: 0.0923, loss_box_dn_5: 0.6251, loss_dense_depth: 0.6922, loss: 23.6607, grad_norm: 22.9408 -2025-11-17 14:30:58,969 - mmdet - INFO - Iter [422/17500] lr: 2.680e-04, eta: 8:44:08, time: 1.551, data_time: 0.079, memory: 49163, loss_cls_0: 0.7183, loss_box_0: 1.6511, loss_cns_0: 0.6373, loss_yns_0: 0.1414, loss_cls_1: 0.7732, loss_box_1: 1.4780, loss_cns_1: 0.6612, loss_yns_1: 0.1410, loss_cls_2: 0.7776, loss_box_2: 1.4424, loss_cns_2: 0.6615, loss_yns_2: 0.1406, loss_cls_3: 0.7856, loss_box_3: 1.4382, loss_cns_3: 0.6613, loss_yns_3: 0.1399, loss_cls_4: 0.7879, loss_box_4: 1.4285, loss_cns_4: 0.6624, loss_yns_4: 0.1414, loss_cls_5: 0.8030, loss_box_5: 1.4253, loss_cns_5: 0.6616, loss_yns_5: 0.1400, loss_cls_dn_0: 0.1160, loss_box_dn_0: 0.7147, loss_cls_dn_1: 0.0933, loss_box_dn_1: 0.6302, loss_cls_dn_2: 0.0916, loss_box_dn_2: 0.6093, loss_cls_dn_3: 0.0902, loss_box_dn_3: 0.6068, loss_cls_dn_4: 0.0904, loss_box_dn_4: 0.6043, loss_cls_dn_5: 0.0925, loss_box_dn_5: 0.6049, loss_dense_depth: 0.6752, loss: 23.3180, grad_norm: 27.5766 -2025-11-17 14:31:00,466 - mmdet - INFO - Iter [423/17500] lr: 2.684e-04, eta: 8:43:52, time: 1.498, data_time: 0.082, memory: 49163, loss_cls_0: 0.7112, loss_box_0: 1.6049, loss_cns_0: 0.6394, loss_yns_0: 0.1403, loss_cls_1: 0.7701, loss_box_1: 1.4320, loss_cns_1: 0.6641, loss_yns_1: 0.1399, loss_cls_2: 0.7738, loss_box_2: 1.4044, loss_cns_2: 0.6635, loss_yns_2: 0.1404, loss_cls_3: 0.7753, loss_box_3: 1.3896, loss_cns_3: 0.6617, loss_yns_3: 0.1392, loss_cls_4: 0.7724, loss_box_4: 1.3868, loss_cns_4: 0.6630, loss_yns_4: 0.1428, loss_cls_5: 0.7799, loss_box_5: 1.3789, loss_cns_5: 0.6620, loss_yns_5: 0.1390, loss_cls_dn_0: 0.1118, loss_box_dn_0: 0.7155, loss_cls_dn_1: 0.0929, loss_box_dn_1: 0.6131, loss_cls_dn_2: 0.0928, loss_box_dn_2: 0.5954, loss_cls_dn_3: 0.0919, loss_box_dn_3: 0.5920, loss_cls_dn_4: 0.0912, loss_box_dn_4: 0.5890, loss_cls_dn_5: 0.0910, loss_box_dn_5: 0.5898, loss_dense_depth: 0.6669, loss: 22.9078, grad_norm: 25.8180 -2025-11-17 14:31:02,007 - mmdet - INFO - Iter [424/17500] lr: 2.688e-04, eta: 8:43:39, time: 1.540, data_time: 0.077, memory: 49163, loss_cls_0: 0.7567, loss_box_0: 1.6241, loss_cns_0: 0.6346, loss_yns_0: 0.1421, loss_cls_1: 0.7852, loss_box_1: 1.4575, loss_cns_1: 0.6615, loss_yns_1: 0.1396, loss_cls_2: 0.7838, loss_box_2: 1.4328, loss_cns_2: 0.6609, loss_yns_2: 0.1402, loss_cls_3: 0.7845, loss_box_3: 1.4242, loss_cns_3: 0.6607, loss_yns_3: 0.1379, loss_cls_4: 0.7950, loss_box_4: 1.4212, loss_cns_4: 0.6616, loss_yns_4: 0.1388, loss_cls_5: 0.7973, loss_box_5: 1.4244, loss_cns_5: 0.6629, loss_yns_5: 0.1378, loss_cls_dn_0: 0.1259, loss_box_dn_0: 0.7094, loss_cls_dn_1: 0.0947, loss_box_dn_1: 0.6232, loss_cls_dn_2: 0.0946, loss_box_dn_2: 0.6046, loss_cls_dn_3: 0.0951, loss_box_dn_3: 0.6004, loss_cls_dn_4: 0.0944, loss_box_dn_4: 0.5996, loss_cls_dn_5: 0.0946, loss_box_dn_5: 0.6019, loss_dense_depth: 0.6993, loss: 23.3032, grad_norm: 27.6886 diff --git a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20251117_141745.log.json b/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20251117_141745.log.json deleted file mode 100644 index 78289a8c44b49bfe4805c2341839bda762dfe174..0000000000000000000000000000000000000000 --- a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20251117_141745.log.json +++ /dev/null @@ -1,425 +0,0 @@ -{"env_info": "sys.platform: linux\nPython: 3.10.12 (main, May 27 2025, 17:12:29) [GCC 11.4.0]\nCUDA available: True\nGPU 0,1,2,3,4,5,6,7: BW200, UBB BW1000\nCUDA_HOME: /opt/dtk\nNVCC: Not Available\nGCC: x86_64-linux-gnu-gcc (Ubuntu 11.4.0-1ubuntu1~22.04) 11.4.0\nPyTorch: 2.4.1\nPyTorch compiling details: PyTorch built with:\n - GCC 10.3\n - C++ Version: 201703\n - Intel(R) Math Kernel Library Version 2020.0.4 Product Build 20200917 for Intel(R) 64 architecture applications\n - OpenMP 201511 (a.k.a. OpenMP 4.5)\n - LAPACK is enabled (usually provided by MKL)\n - NNPACK is enabled\n - CPU capability usage: AVX512\n - HIP Runtime 6.3.25211\n - MIOpen 2.17.0\n - Magma 2.8.0\n - Build settings: BLAS_INFO=mkl, BUILD_TYPE=Release, CXX_COMPILER=/opt/rh/gcc-toolset-10/root/usr/bin/c++, CXX_FLAGS= -D_GLIBCXX_USE_CXX11_ABI=1 -fvisibility-inlines-hidden -DUSE_PTHREADPOOL -DNDEBUG -DUSE_KINETO -DLIBKINETO_NOCUPTI -DUSE_FBGEMM -DUSE_PYTORCH_QNNPACK -DUSE_XNNPACK -DSYMBOLICATE_MOBILE_DEBUG_HANDLE -O2 -fPIC -Wall -Wextra -Werror=return-type -Werror=non-virtual-dtor -Werror=bool-operation -Wnarrowing -Wno-missing-field-initializers -Wno-type-limits -Wno-array-bounds -Wno-unknown-pragmas -Wno-unused-parameter -Wno-unused-function -Wno-unused-result -Wno-strict-overflow -Wno-strict-aliasing -Wno-stringop-overflow -Wsuggest-override -Wno-psabi -Wno-error=pedantic -Wno-error=old-style-cast -Wno-missing-braces -fdiagnostics-color=always -faligned-new -Wno-unused-but-set-variable -Wno-maybe-uninitialized -fno-math-errno -fno-trapping-math -Werror=format -Wno-stringop-overflow, FORCE_FALLBACK_CUDA_MPI=1, LAPACK_INFO=mkl, PERF_WITH_AVX=1, PERF_WITH_AVX2=1, PERF_WITH_AVX512=1, TORCH_VERSION=2.4.1, USE_CUDA=0, USE_CUDNN=OFF, USE_CUSPARSELT=OFF, USE_EXCEPTION_PTR=1, USE_GFLAGS=1, USE_GLOG=1, USE_GLOO=1, USE_MKL=ON, USE_MKLDNN=0, USE_MPI=1, USE_NCCL=1, USE_NNPACK=ON, USE_OPENMP=1, USE_ROCM=ON, USE_ROCM_KERNEL_ASSERT=OFF, \n\nTorchVision: 0.19.1\nOpenCV: 4.11.0\nMMCV: 1.6.1\nMMCV Compiler: GCC 11.4\nMMCV CUDA Compiler: rocm not available\nMMDetection: 2.26.0+c41df4b", "config": "plugin = True\nplugin_dir = 'projects/mmdet3d_plugin/'\ndist_params = dict(backend='nccl')\nlog_level = 'INFO'\nwork_dir = './work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704'\ntotal_batch_size = 160\nnum_gpus = 8\nbatch_size = 20\nnum_iters_per_epoch = 175\nnum_epochs = 100\ncheckpoint_epoch_interval = 20\ncheckpoint_config = dict(interval=3500)\nlog_config = dict(\n interval=1,\n hooks=[\n dict(type='TextLoggerHook', by_epoch=False),\n dict(type='TensorboardLoggerHook')\n ])\nload_from = None\nresume_from = None\nworkflow = [('train', 1)]\nfp16 = dict(loss_scale=32.0)\ninput_shape = (704, 256)\ntracking_test = True\ntracking_threshold = 0.2\nclass_names = [\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier',\n 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n]\nnum_classes = 10\nembed_dims = 256\nnum_groups = 8\nnum_decoder = 6\nnum_single_frame_decoder = 1\nuse_deformable_func = True\nstrides = [4, 8, 16, 32]\nnum_levels = 4\nnum_depth_layers = 3\ndrop_out = 0.1\ntemporal = True\ndecouple_attn = True\nwith_quality_estimation = True\nmodel = dict(\n type='Sparse4D',\n use_grid_mask=True,\n use_deformable_func=True,\n img_backbone=dict(\n type='ResNet',\n depth=50,\n num_stages=4,\n frozen_stages=-1,\n norm_eval=False,\n style='pytorch',\n with_cp=True,\n out_indices=(0, 1, 2, 3),\n norm_cfg=dict(type='BN', requires_grad=True),\n pretrained='ckpt/resnet50-19c8e357.pth'),\n img_neck=dict(\n type='FPN',\n num_outs=4,\n start_level=0,\n out_channels=256,\n add_extra_convs='on_output',\n relu_before_extra_convs=True,\n in_channels=[256, 512, 1024, 2048]),\n depth_branch=dict(\n type='DenseDepthNet',\n embed_dims=256,\n num_depth_layers=3,\n loss_weight=0.2),\n head=dict(\n type='Sparse4DHead',\n cls_threshold_to_reg=0.05,\n decouple_attn=True,\n instance_bank=dict(\n type='InstanceBank',\n num_anchor=900,\n embed_dims=256,\n anchor='nuscenes_kmeans900.npy',\n anchor_handler=dict(type='SparseBox3DKeyPointsGenerator'),\n num_temp_instances=600,\n confidence_decay=0.6,\n feat_grad=False),\n anchor_encoder=dict(\n type='SparseBox3DEncoder',\n vel_dims=3,\n embed_dims=[128, 32, 32, 64],\n mode='cat',\n output_fc=False,\n in_loops=1,\n out_loops=4),\n num_single_frame_decoder=1,\n operation_order=[\n 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm',\n 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm',\n 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm',\n 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm',\n 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm',\n 'deformable', 'ffn', 'norm', 'refine'\n ],\n temp_graph_model=dict(\n type='MultiheadAttention',\n embed_dims=512,\n num_heads=8,\n batch_first=True,\n dropout=0.1),\n graph_model=dict(\n type='MultiheadAttention',\n embed_dims=512,\n num_heads=8,\n batch_first=True,\n dropout=0.1),\n norm_layer=dict(type='LN', normalized_shape=256),\n ffn=dict(\n type='AsymmetricFFN',\n in_channels=512,\n pre_norm=dict(type='LN'),\n embed_dims=256,\n feedforward_channels=1024,\n num_fcs=2,\n ffn_drop=0.1,\n act_cfg=dict(type='ReLU', inplace=True)),\n deformable_model=dict(\n type='DeformableFeatureAggregation',\n embed_dims=256,\n num_groups=8,\n num_levels=4,\n num_cams=6,\n attn_drop=0.15,\n use_deformable_func=True,\n use_camera_embed=True,\n residual_mode='cat',\n kps_generator=dict(\n type='SparseBox3DKeyPointsGenerator',\n num_learnable_pts=6,\n fix_scale=[[0, 0, 0], [0.45, 0, 0], [-0.45, 0, 0],\n [0, 0.45, 0], [0, -0.45, 0], [0, 0, 0.45],\n [0, 0, -0.45]])),\n refine_layer=dict(\n type='SparseBox3DRefinementModule',\n embed_dims=256,\n num_cls=10,\n refine_yaw=True,\n with_quality_estimation=True),\n sampler=dict(\n type='SparseBox3DTarget',\n num_dn_groups=5,\n num_temp_dn_groups=3,\n dn_noise_scale=[2.0, 2.0, 2.0, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5],\n max_dn_gt=32,\n add_neg_dn=True,\n cls_weight=2.0,\n box_weight=0.25,\n reg_weights=[2.0, 2.0, 2.0, 0.5, 0.5, 0.5, 0.0, 0.0, 0.0, 0.0],\n cls_wise_reg_weights=dict(\n {9: [2.0, 2.0, 2.0, 1.0, 1.0, 1.0, 0.0, 0.0, 1.0, 1.0]})),\n loss_cls=dict(\n type='FocalLoss',\n use_sigmoid=True,\n gamma=2.0,\n alpha=0.25,\n loss_weight=2.0),\n loss_reg=dict(\n type='SparseBox3DLoss',\n loss_box=dict(type='L1Loss', loss_weight=0.25),\n loss_centerness=dict(type='CrossEntropyLoss', use_sigmoid=True),\n loss_yawness=dict(type='GaussianFocalLoss'),\n cls_allow_reverse=[5]),\n decoder=dict(type='SparseBox3DDecoder'),\n reg_weights=[2.0, 2.0, 2.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0]))\ndataset_type = 'NuScenes3DDetTrackDataset'\ndata_root = 'data/nuscenes/'\nanno_root = 'data/nuscenes_anno_pkls/'\nfile_client_args = dict(backend='disk')\nimg_norm_cfg = dict(\n mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375], to_rgb=True)\ntrain_pipeline = [\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(\n type='LoadPointsFromFile',\n coord_type='LIDAR',\n load_dim=5,\n use_dim=5,\n file_client_args=dict(backend='disk')),\n dict(type='ResizeCropFlipImage'),\n dict(type='MultiScaleDepthMapGenerator', downsample=[4, 8, 16]),\n dict(type='BBoxRotation'),\n dict(type='PhotoMetricDistortionMultiViewImage'),\n dict(\n type='NormalizeMultiviewImage',\n mean=[123.675, 116.28, 103.53],\n std=[58.395, 57.12, 57.375],\n to_rgb=True),\n dict(\n type='CircleObjectRangeFilter',\n class_dist_thred=[55, 55, 55, 55, 55, 55, 55, 55, 55, 55]),\n dict(\n type='InstanceNameFilter',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ]),\n dict(type='NuScenesSparse4DAdaptor'),\n dict(\n type='Collect',\n keys=[\n 'img', 'timestamp', 'projection_mat', 'image_wh', 'gt_depth',\n 'focal', 'gt_bboxes_3d', 'gt_labels_3d'\n ],\n meta_keys=['T_global', 'T_global_inv', 'timestamp', 'instance_id'])\n]\ntest_pipeline = [\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(type='ResizeCropFlipImage'),\n dict(\n type='NormalizeMultiviewImage',\n mean=[123.675, 116.28, 103.53],\n std=[58.395, 57.12, 57.375],\n to_rgb=True),\n dict(type='NuScenesSparse4DAdaptor'),\n dict(\n type='Collect',\n keys=['img', 'timestamp', 'projection_mat', 'image_wh'],\n meta_keys=['T_global', 'T_global_inv', 'timestamp'])\n]\ninput_modality = dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False)\ndata_basic_config = dict(\n type='NuScenes3DDetTrackDataset',\n data_root='data/nuscenes/',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier',\n 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ],\n modality=dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False),\n version='v1.0-trainval')\ndata_aug_conf = dict(\n resize_lim=(0.4, 0.47),\n final_dim=(256, 704),\n bot_pct_lim=(0.0, 0.0),\n rot_lim=(-5.4, 5.4),\n H=900,\n W=1600,\n rand_flip=True,\n rot3d_range=[-0.3925, 0.3925])\ndata = dict(\n samples_per_gpu=20,\n workers_per_gpu=20,\n train=dict(\n type='NuScenes3DDetTrackDataset',\n data_root='data/nuscenes/',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ],\n modality=dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False),\n version='v1.0-trainval',\n ann_file='data/nuscenes_anno_pkls/nuscenes_infos_train.pkl',\n pipeline=[\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(\n type='LoadPointsFromFile',\n coord_type='LIDAR',\n load_dim=5,\n use_dim=5,\n file_client_args=dict(backend='disk')),\n dict(type='ResizeCropFlipImage'),\n dict(type='MultiScaleDepthMapGenerator', downsample=[4, 8, 16]),\n dict(type='BBoxRotation'),\n dict(type='PhotoMetricDistortionMultiViewImage'),\n dict(\n type='NormalizeMultiviewImage',\n mean=[123.675, 116.28, 103.53],\n std=[58.395, 57.12, 57.375],\n to_rgb=True),\n dict(\n type='CircleObjectRangeFilter',\n class_dist_thred=[55, 55, 55, 55, 55, 55, 55, 55, 55, 55]),\n dict(\n type='InstanceNameFilter',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian',\n 'traffic_cone'\n ]),\n dict(type='NuScenesSparse4DAdaptor'),\n dict(\n type='Collect',\n keys=[\n 'img', 'timestamp', 'projection_mat', 'image_wh',\n 'gt_depth', 'focal', 'gt_bboxes_3d', 'gt_labels_3d'\n ],\n meta_keys=[\n 'T_global', 'T_global_inv', 'timestamp', 'instance_id'\n ])\n ],\n test_mode=False,\n data_aug_conf=dict(\n resize_lim=(0.4, 0.47),\n final_dim=(256, 704),\n bot_pct_lim=(0.0, 0.0),\n rot_lim=(-5.4, 5.4),\n H=900,\n W=1600,\n rand_flip=True,\n rot3d_range=[-0.3925, 0.3925]),\n with_seq_flag=True,\n sequences_split_num=2,\n keep_consistent_seq_aug=True),\n val=dict(\n type='NuScenes3DDetTrackDataset',\n data_root='data/nuscenes/',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ],\n modality=dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False),\n version='v1.0-trainval',\n ann_file='data/nuscenes_anno_pkls/nuscenes_infos_val.pkl',\n pipeline=[\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(type='ResizeCropFlipImage'),\n dict(\n type='NormalizeMultiviewImage',\n mean=[123.675, 116.28, 103.53],\n std=[58.395, 57.12, 57.375],\n to_rgb=True),\n dict(type='NuScenesSparse4DAdaptor'),\n dict(\n type='Collect',\n keys=['img', 'timestamp', 'projection_mat', 'image_wh'],\n meta_keys=['T_global', 'T_global_inv', 'timestamp'])\n ],\n data_aug_conf=dict(\n resize_lim=(0.4, 0.47),\n final_dim=(256, 704),\n bot_pct_lim=(0.0, 0.0),\n rot_lim=(-5.4, 5.4),\n H=900,\n W=1600,\n rand_flip=True,\n rot3d_range=[-0.3925, 0.3925]),\n test_mode=True,\n tracking=True,\n tracking_threshold=0.2),\n test=dict(\n type='NuScenes3DDetTrackDataset',\n data_root='data/nuscenes/',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ],\n modality=dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False),\n version='v1.0-trainval',\n ann_file='data/nuscenes_anno_pkls/nuscenes_infos_val.pkl',\n pipeline=[\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(type='ResizeCropFlipImage'),\n dict(\n type='NormalizeMultiviewImage',\n mean=[123.675, 116.28, 103.53],\n std=[58.395, 57.12, 57.375],\n to_rgb=True),\n dict(type='NuScenesSparse4DAdaptor'),\n dict(\n type='Collect',\n keys=['img', 'timestamp', 'projection_mat', 'image_wh'],\n meta_keys=['T_global', 'T_global_inv', 'timestamp'])\n ],\n data_aug_conf=dict(\n resize_lim=(0.4, 0.47),\n final_dim=(256, 704),\n bot_pct_lim=(0.0, 0.0),\n rot_lim=(-5.4, 5.4),\n H=900,\n W=1600,\n rand_flip=True,\n rot3d_range=[-0.3925, 0.3925]),\n test_mode=True,\n tracking=True,\n tracking_threshold=0.2))\noptimizer = dict(\n type='Miopen_AdamW',\n lr=0.0006,\n weight_decay=0.001,\n paramwise_cfg=dict(custom_keys=dict(img_backbone=dict(lr_mult=0.5))))\noptimizer_config = dict(grad_clip=dict(max_norm=25, norm_type=2))\nlr_config = dict(\n policy='CosineAnnealing',\n warmup='linear',\n warmup_iters=500,\n warmup_ratio=0.3333333333333333,\n min_lr_ratio=0.001)\nrunner = dict(type='IterBasedRunner', max_iters=17500)\nvis_pipeline = [\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(type='Collect', keys=['img'], meta_keys=['timestamp', 'lidar2img'])\n]\nevaluation = dict(\n interval=3500,\n pipeline=[\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(\n type='Collect', keys=['img'], meta_keys=['timestamp', 'lidar2img'])\n ])\ngpu_ids = range(0, 8)\n", "seed": 0, "exp_name": "sparse4dv3_temporal_r50_1x8_bs6_256x704.py"} -{"mode": "train", "epoch": 1, "iter": 1, "lr": 0.0001, "memory": 49163, "data_time": 10.83884, "loss_cls_0": 2.36133, "loss_box_0": 0.01384, "loss_cns_0": 0.0027, "loss_yns_0": 0.00078, "loss_cls_1": 2.15436, "loss_box_1": 0.10706, "loss_cns_1": 0.02434, "loss_yns_1": 0.00662, "loss_cls_2": 2.31186, "loss_box_2": 0.00504, "loss_cns_2": 0.00059, "loss_yns_2": 0.00029, "loss_cls_3": 2.38993, "loss_box_3": 0.02945, "loss_cns_3": 0.00504, "loss_yns_3": 0.00144, "loss_cls_4": 2.02836, "loss_box_4": 0.4122, "loss_cns_4": 0.05299, "loss_yns_4": 0.02517, "loss_cls_5": 2.42501, "loss_box_5": 0.01665, "loss_cns_5": 0.00202, "loss_yns_5": 0.00157, "loss_cls_dn_0": 1.19804, "loss_box_dn_0": 1.4603, "loss_cls_dn_1": 1.11019, "loss_box_dn_1": 1.73179, "loss_cls_dn_2": 1.17413, "loss_box_dn_2": 1.97187, "loss_cls_dn_3": 1.17206, "loss_box_dn_3": 2.24183, "loss_cls_dn_4": 1.05279, "loss_box_dn_4": 2.42679, "loss_cls_dn_5": 1.23869, "loss_box_dn_5": 2.6773, "loss_dense_depth": 1.86432, "loss": 35.69876, "grad_norm": 268.25821, "time": 116.58354} -{"mode": "train", "epoch": 1, "iter": 2, "lr": 0.0001, "memory": 49163, "data_time": 0.08491, "loss_cls_0": 2.01055, "loss_box_0": 0.01322, "loss_cns_0": 0.00324, "loss_yns_0": 0.00113, "loss_cls_1": 2.00818, "loss_box_1": 0.19475, "loss_cns_1": 0.03496, "loss_yns_1": 0.01049, "loss_cls_2": 2.09795, "loss_box_2": 0.26089, "loss_cns_2": 0.0249, "loss_yns_2": 0.01111, "loss_cls_3": 1.95332, "loss_box_3": 0.35328, "loss_cns_3": 0.04605, "loss_yns_3": 0.01649, "loss_cls_4": 1.79378, "loss_box_4": 1.63758, "loss_cns_4": 0.16546, "loss_yns_4": 0.05726, "loss_cls_5": 2.05616, "loss_box_5": 0.54973, "loss_cns_5": 0.06148, "loss_yns_5": 0.01902, "loss_cls_dn_0": 1.0101, "loss_box_dn_0": 1.25542, "loss_cls_dn_1": 0.95291, "loss_box_dn_1": 2.40866, "loss_cls_dn_2": 0.9686, "loss_box_dn_2": 2.53092, "loss_cls_dn_3": 0.91339, "loss_box_dn_3": 2.61725, "loss_cls_dn_4": 0.8396, "loss_box_dn_4": 2.8875, "loss_cls_dn_5": 0.98604, "loss_box_dn_5": 3.12443, "loss_dense_depth": 1.71131, "loss": 37.58712, "grad_norm": 66.5134, "time": 1.96199} -{"mode": "train", "epoch": 1, "iter": 3, "lr": 0.0001, "memory": 49163, "data_time": 0.08101, "loss_cls_0": 1.39078, "loss_box_0": 2.55963, "loss_cns_0": 0.61611, "loss_yns_0": 0.21675, "loss_cls_1": 1.72634, "loss_box_1": 2.37796, "loss_cns_1": 0.33834, "loss_yns_1": 0.12753, "loss_cls_2": 1.74758, "loss_box_2": 4.53964, "loss_cns_2": 0.3706, "loss_yns_2": 0.2069, "loss_cls_3": 1.59646, "loss_box_3": 5.33508, "loss_cns_3": 0.46109, "loss_yns_3": 0.2203, "loss_cls_4": 1.50979, "loss_box_4": 5.28485, "loss_cns_4": 0.43583, "loss_yns_4": 0.20932, "loss_cls_5": 1.62233, "loss_box_5": 4.24209, "loss_cns_5": 0.29869, "loss_yns_5": 0.13181, "loss_cls_dn_0": 0.64785, "loss_box_dn_0": 1.18846, "loss_cls_dn_1": 0.80652, "loss_box_dn_1": 2.38361, "loss_cls_dn_2": 0.76648, "loss_box_dn_2": 2.6115, "loss_cls_dn_3": 0.66761, "loss_box_dn_3": 2.83612, "loss_cls_dn_4": 0.68228, "loss_box_dn_4": 3.08375, "loss_cls_dn_5": 0.77307, "loss_box_dn_5": 3.37938, "loss_dense_depth": 1.58651, "loss": 58.97893, "grad_norm": 116.22523, "time": 1.51061} -{"mode": "train", "epoch": 1, "iter": 4, "lr": 0.0001, "memory": 49163, "data_time": 0.08536, "loss_cls_0": 1.4045, "loss_box_0": 2.6805, "loss_cns_0": 0.52106, "loss_yns_0": 0.24665, "loss_cls_1": 1.56046, "loss_box_1": 3.66386, "loss_cns_1": 0.45329, "loss_yns_1": 0.20865, "loss_cls_2": 1.71811, "loss_box_2": 3.8058, "loss_cns_2": 0.44266, "loss_yns_2": 0.21002, "loss_cls_3": 1.46503, "loss_box_3": 4.30839, "loss_cns_3": 0.46507, "loss_yns_3": 0.20708, "loss_cls_4": 1.45564, "loss_box_4": 4.80889, "loss_cns_4": 0.42004, "loss_yns_4": 0.1936, "loss_cls_5": 1.40641, "loss_box_5": 5.36285, "loss_cns_5": 0.44666, "loss_yns_5": 0.20392, "loss_cls_dn_0": 0.5189, "loss_box_dn_0": 1.26253, "loss_cls_dn_1": 0.67151, "loss_box_dn_1": 2.65854, "loss_cls_dn_2": 0.64412, "loss_box_dn_2": 2.76469, "loss_cls_dn_3": 0.58408, "loss_box_dn_3": 3.03781, "loss_cls_dn_4": 0.5511, "loss_box_dn_4": 3.30541, "loss_cls_dn_5": 0.60184, "loss_box_dn_5": 3.57842, "loss_dense_depth": 1.62789, "loss": 59.46594, "grad_norm": 120.19677, "time": 1.54155} -{"mode": "train", "epoch": 1, "iter": 5, "lr": 0.0001, "memory": 49163, "data_time": 0.08558, "loss_cls_0": 1.34939, "loss_box_0": 2.90852, "loss_cns_0": 0.50722, "loss_yns_0": 0.19873, "loss_cls_1": 1.54519, "loss_box_1": 4.28782, "loss_cns_1": 0.42036, "loss_yns_1": 0.21395, "loss_cls_2": 1.4945, "loss_box_2": 4.26174, "loss_cns_2": 0.42392, "loss_yns_2": 0.19805, "loss_cls_3": 1.37876, "loss_box_3": 4.41104, "loss_cns_3": 0.39683, "loss_yns_3": 0.2132, "loss_cls_4": 1.33573, "loss_box_4": 4.72401, "loss_cns_4": 0.37237, "loss_yns_4": 0.20489, "loss_cls_5": 1.35455, "loss_box_5": 4.96669, "loss_cns_5": 0.37359, "loss_yns_5": 0.21968, "loss_cls_dn_0": 0.56869, "loss_box_dn_0": 1.31635, "loss_cls_dn_1": 0.60808, "loss_box_dn_1": 2.5533, "loss_cls_dn_2": 0.63683, "loss_box_dn_2": 2.68591, "loss_cls_dn_3": 0.53469, "loss_box_dn_3": 2.79333, "loss_cls_dn_4": 0.54099, "loss_box_dn_4": 3.05133, "loss_cls_dn_5": 0.52022, "loss_box_dn_5": 3.12062, "loss_dense_depth": 1.6734, "loss": 58.36448, "grad_norm": 128.47354, "time": 1.52282} -{"mode": "train", "epoch": 1, "iter": 6, "lr": 0.0001, "memory": 49163, "data_time": 0.08266, "loss_cls_0": 1.25766, "loss_box_0": 2.55374, "loss_cns_0": 0.61684, "loss_yns_0": 0.18543, "loss_cls_1": 1.37146, "loss_box_1": 3.83558, "loss_cns_1": 0.45169, "loss_yns_1": 0.19412, "loss_cls_2": 1.3644, "loss_box_2": 3.92678, "loss_cns_2": 0.44801, "loss_yns_2": 0.20973, "loss_cls_3": 1.34295, "loss_box_3": 3.89151, "loss_cns_3": 0.46936, "loss_yns_3": 0.18465, "loss_cls_4": 1.33246, "loss_box_4": 4.25878, "loss_cns_4": 0.40573, "loss_yns_4": 0.18897, "loss_cls_5": 1.36414, "loss_box_5": 4.59553, "loss_cns_5": 0.36085, "loss_yns_5": 0.20044, "loss_cls_dn_0": 0.56069, "loss_box_dn_0": 1.18938, "loss_cls_dn_1": 0.54891, "loss_box_dn_1": 2.62944, "loss_cls_dn_2": 0.57034, "loss_box_dn_2": 2.65191, "loss_cls_dn_3": 0.48582, "loss_box_dn_3": 2.70011, "loss_cls_dn_4": 0.47205, "loss_box_dn_4": 2.96069, "loss_cls_dn_5": 0.45476, "loss_box_dn_5": 3.14474, "loss_dense_depth": 1.85516, "loss": 55.23483, "grad_norm": 124.7906, "time": 1.50964} -{"mode": "train", "epoch": 1, "iter": 7, "lr": 0.0001, "memory": 49163, "data_time": 0.08134, "loss_cls_0": 1.25648, "loss_box_0": 2.33686, "loss_cns_0": 0.65019, "loss_yns_0": 0.20174, "loss_cls_1": 1.28638, "loss_box_1": 3.67366, "loss_cns_1": 0.5018, "loss_yns_1": 0.19147, "loss_cls_2": 1.36555, "loss_box_2": 3.85018, "loss_cns_2": 0.46764, "loss_yns_2": 0.19346, "loss_cls_3": 1.29152, "loss_box_3": 3.86356, "loss_cns_3": 0.46484, "loss_yns_3": 0.18349, "loss_cls_4": 1.36963, "loss_box_4": 4.17957, "loss_cns_4": 0.44598, "loss_yns_4": 0.22562, "loss_cls_5": 1.34892, "loss_box_5": 4.44053, "loss_cns_5": 0.4892, "loss_yns_5": 0.18608, "loss_cls_dn_0": 0.52, "loss_box_dn_0": 1.08909, "loss_cls_dn_1": 0.473, "loss_box_dn_1": 2.49404, "loss_cls_dn_2": 0.48651, "loss_box_dn_2": 2.49612, "loss_cls_dn_3": 0.43785, "loss_box_dn_3": 2.53085, "loss_cls_dn_4": 0.40756, "loss_box_dn_4": 2.68381, "loss_cls_dn_5": 0.39447, "loss_box_dn_5": 2.79758, "loss_dense_depth": 2.01038, "loss": 53.2856, "grad_norm": 119.46581, "time": 1.50807} -{"mode": "train", "epoch": 1, "iter": 8, "lr": 0.0001, "memory": 49163, "data_time": 0.08562, "loss_cls_0": 1.27107, "loss_box_0": 2.23411, "loss_cns_0": 0.61829, "loss_yns_0": 0.17406, "loss_cls_1": 1.30109, "loss_box_1": 3.6105, "loss_cns_1": 0.51007, "loss_yns_1": 0.19154, "loss_cls_2": 1.33784, "loss_box_2": 3.76649, "loss_cns_2": 0.39339, "loss_yns_2": 0.18981, "loss_cls_3": 1.29081, "loss_box_3": 3.89365, "loss_cns_3": 0.37769, "loss_yns_3": 0.18714, "loss_cls_4": 1.26377, "loss_box_4": 3.82913, "loss_cns_4": 0.37299, "loss_yns_4": 0.1997, "loss_cls_5": 1.34081, "loss_box_5": 3.84222, "loss_cns_5": 0.38199, "loss_yns_5": 0.20623, "loss_cls_dn_0": 0.48244, "loss_box_dn_0": 1.03957, "loss_cls_dn_1": 0.47355, "loss_box_dn_1": 1.66126, "loss_cls_dn_2": 0.49246, "loss_box_dn_2": 1.6855, "loss_cls_dn_3": 0.43693, "loss_box_dn_3": 1.84993, "loss_cls_dn_4": 0.4647, "loss_box_dn_4": 1.79798, "loss_cls_dn_5": 0.42669, "loss_box_dn_5": 1.85192, "loss_dense_depth": 2.28053, "loss": 47.72784, "grad_norm": 105.50864, "time": 1.49709} -{"mode": "train", "epoch": 1, "iter": 9, "lr": 0.0001, "memory": 49163, "data_time": 0.09593, "loss_cls_0": 1.23164, "loss_box_0": 2.17926, "loss_cns_0": 0.62848, "loss_yns_0": 0.1858, "loss_cls_1": 1.28854, "loss_box_1": 3.03438, "loss_cns_1": 0.52418, "loss_yns_1": 0.18465, "loss_cls_2": 1.29145, "loss_box_2": 3.16461, "loss_cns_2": 0.50812, "loss_yns_2": 0.19988, "loss_cls_3": 1.26967, "loss_box_3": 3.27257, "loss_cns_3": 0.53221, "loss_yns_3": 0.20177, "loss_cls_4": 1.24935, "loss_box_4": 3.32534, "loss_cns_4": 0.50761, "loss_yns_4": 0.18653, "loss_cls_5": 1.28566, "loss_box_5": 3.45695, "loss_cns_5": 0.55356, "loss_yns_5": 0.18993, "loss_cls_dn_0": 0.48686, "loss_box_dn_0": 1.04108, "loss_cls_dn_1": 0.43484, "loss_box_dn_1": 1.50907, "loss_cls_dn_2": 0.48218, "loss_box_dn_2": 1.68982, "loss_cls_dn_3": 0.43362, "loss_box_dn_3": 1.90194, "loss_cls_dn_4": 0.4689, "loss_box_dn_4": 1.9699, "loss_cls_dn_5": 0.44857, "loss_box_dn_5": 2.11273, "loss_dense_depth": 2.18868, "loss": 45.62035, "grad_norm": 84.3352, "time": 1.5009} -{"mode": "train", "epoch": 1, "iter": 10, "lr": 0.0001, "memory": 49163, "data_time": 0.08627, "loss_cls_0": 1.25609, "loss_box_0": 2.20659, "loss_cns_0": 0.62609, "loss_yns_0": 0.18018, "loss_cls_1": 1.28377, "loss_box_1": 3.03292, "loss_cns_1": 0.49866, "loss_yns_1": 0.17933, "loss_cls_2": 1.2531, "loss_box_2": 3.15469, "loss_cns_2": 0.47825, "loss_yns_2": 0.20348, "loss_cls_3": 1.23919, "loss_box_3": 3.34913, "loss_cns_3": 0.46142, "loss_yns_3": 0.18903, "loss_cls_4": 1.26571, "loss_box_4": 3.64084, "loss_cns_4": 0.38943, "loss_yns_4": 0.20731, "loss_cls_5": 1.2528, "loss_box_5": 3.95689, "loss_cns_5": 0.33478, "loss_yns_5": 0.19072, "loss_cls_dn_0": 0.47211, "loss_box_dn_0": 1.03541, "loss_cls_dn_1": 0.41055, "loss_box_dn_1": 1.71324, "loss_cls_dn_2": 0.43871, "loss_box_dn_2": 1.85661, "loss_cls_dn_3": 0.44289, "loss_box_dn_3": 1.95547, "loss_cls_dn_4": 0.41101, "loss_box_dn_4": 2.08097, "loss_cls_dn_5": 0.43609, "loss_box_dn_5": 2.26996, "loss_dense_depth": 2.30326, "loss": 46.65668, "grad_norm": 104.02073, "time": 1.50478} -{"mode": "train", "epoch": 1, "iter": 11, "lr": 0.0001, "memory": 49163, "data_time": 0.08257, "loss_cls_0": 1.21822, "loss_box_0": 2.27816, "loss_cns_0": 0.61619, "loss_yns_0": 0.1709, "loss_cls_1": 1.24995, "loss_box_1": 3.09604, "loss_cns_1": 0.51153, "loss_yns_1": 0.17734, "loss_cls_2": 1.25859, "loss_box_2": 3.18744, "loss_cns_2": 0.48037, "loss_yns_2": 0.19841, "loss_cls_3": 1.28584, "loss_box_3": 3.35151, "loss_cns_3": 0.47967, "loss_yns_3": 0.18184, "loss_cls_4": 1.27394, "loss_box_4": 3.54398, "loss_cns_4": 0.45425, "loss_yns_4": 0.18964, "loss_cls_5": 1.24816, "loss_box_5": 3.84857, "loss_cns_5": 0.4189, "loss_yns_5": 0.17841, "loss_cls_dn_0": 0.46163, "loss_box_dn_0": 1.03049, "loss_cls_dn_1": 0.40944, "loss_box_dn_1": 1.81435, "loss_cls_dn_2": 0.40805, "loss_box_dn_2": 1.87809, "loss_cls_dn_3": 0.42551, "loss_box_dn_3": 1.89467, "loss_cls_dn_4": 0.38872, "loss_box_dn_4": 1.99957, "loss_cls_dn_5": 0.41919, "loss_box_dn_5": 2.11688, "loss_dense_depth": 2.22288, "loss": 46.36732, "grad_norm": 105.3113, "time": 1.50514} -{"mode": "train", "epoch": 1, "iter": 12, "lr": 0.0001, "memory": 49163, "data_time": 0.08311, "loss_cls_0": 1.23812, "loss_box_0": 2.26746, "loss_cns_0": 0.61444, "loss_yns_0": 0.17462, "loss_cls_1": 1.22611, "loss_box_1": 3.20869, "loss_cns_1": 0.52111, "loss_yns_1": 0.18457, "loss_cls_2": 1.21792, "loss_box_2": 3.3033, "loss_cns_2": 0.47335, "loss_yns_2": 0.19125, "loss_cls_3": 1.26913, "loss_box_3": 3.34612, "loss_cns_3": 0.46655, "loss_yns_3": 0.18031, "loss_cls_4": 1.21584, "loss_box_4": 3.39482, "loss_cns_4": 0.47302, "loss_yns_4": 0.1837, "loss_cls_5": 1.24318, "loss_box_5": 3.55844, "loss_cns_5": 0.48385, "loss_yns_5": 0.18257, "loss_cls_dn_0": 0.4584, "loss_box_dn_0": 1.0348, "loss_cls_dn_1": 0.40659, "loss_box_dn_1": 1.93112, "loss_cls_dn_2": 0.3985, "loss_box_dn_2": 1.9494, "loss_cls_dn_3": 0.39132, "loss_box_dn_3": 1.96894, "loss_cls_dn_4": 0.37515, "loss_box_dn_4": 2.0891, "loss_cls_dn_5": 0.39621, "loss_box_dn_5": 2.23329, "loss_dense_depth": 2.25796, "loss": 46.50927, "grad_norm": 71.74351, "time": 1.49048} -{"mode": "train", "epoch": 1, "iter": 13, "lr": 0.0001, "memory": 49163, "data_time": 0.08464, "loss_cls_0": 1.16266, "loss_box_0": 2.19164, "loss_cns_0": 0.61396, "loss_yns_0": 0.17096, "loss_cls_1": 1.21662, "loss_box_1": 2.91055, "loss_cns_1": 0.52577, "loss_yns_1": 0.18943, "loss_cls_2": 1.22242, "loss_box_2": 3.15237, "loss_cns_2": 0.51822, "loss_yns_2": 0.17774, "loss_cls_3": 1.29101, "loss_box_3": 3.1645, "loss_cns_3": 0.48884, "loss_yns_3": 0.17393, "loss_cls_4": 1.22027, "loss_box_4": 3.14455, "loss_cns_4": 0.50173, "loss_yns_4": 0.18556, "loss_cls_5": 1.26697, "loss_box_5": 3.32118, "loss_cns_5": 0.47226, "loss_yns_5": 0.17757, "loss_cls_dn_0": 0.49301, "loss_box_dn_0": 1.0416, "loss_cls_dn_1": 0.42151, "loss_box_dn_1": 1.5977, "loss_cls_dn_2": 0.43475, "loss_box_dn_2": 1.70311, "loss_cls_dn_3": 0.40989, "loss_box_dn_3": 1.8085, "loss_cls_dn_4": 0.41698, "loss_box_dn_4": 1.96126, "loss_cls_dn_5": 0.43475, "loss_box_dn_5": 2.12849, "loss_dense_depth": 2.1866, "loss": 44.49887, "grad_norm": 76.34818, "time": 1.5228} -{"mode": "train", "epoch": 1, "iter": 14, "lr": 0.00011, "memory": 49163, "data_time": 0.08314, "loss_cls_0": 1.17978, "loss_box_0": 2.23921, "loss_cns_0": 0.61442, "loss_yns_0": 0.17161, "loss_cls_1": 1.28581, "loss_box_1": 2.68575, "loss_cns_1": 0.62164, "loss_yns_1": 0.17488, "loss_cls_2": 1.27332, "loss_box_2": 2.90579, "loss_cns_2": 0.57262, "loss_yns_2": 0.18231, "loss_cls_3": 1.29865, "loss_box_3": 2.95647, "loss_cns_3": 0.56696, "loss_yns_3": 0.17377, "loss_cls_4": 1.23419, "loss_box_4": 2.95584, "loss_cns_4": 0.58251, "loss_yns_4": 0.18406, "loss_cls_5": 1.25248, "loss_box_5": 3.20207, "loss_cns_5": 0.54632, "loss_yns_5": 0.18271, "loss_cls_dn_0": 0.48556, "loss_box_dn_0": 1.04608, "loss_cls_dn_1": 0.43132, "loss_box_dn_1": 1.48005, "loss_cls_dn_2": 0.44579, "loss_box_dn_2": 1.58638, "loss_cls_dn_3": 0.42953, "loss_box_dn_3": 1.70612, "loss_cls_dn_4": 0.44933, "loss_box_dn_4": 1.82113, "loss_cls_dn_5": 0.4535, "loss_box_dn_5": 1.97221, "loss_dense_depth": 2.43265, "loss": 43.78281, "grad_norm": 95.37186, "time": 1.49803} -{"mode": "train", "epoch": 1, "iter": 15, "lr": 0.00011, "memory": 49163, "data_time": 0.08364, "loss_cls_0": 1.23432, "loss_box_0": 2.31211, "loss_cns_0": 0.60068, "loss_yns_0": 0.17122, "loss_cls_1": 1.24107, "loss_box_1": 2.76262, "loss_cns_1": 0.58218, "loss_yns_1": 0.18347, "loss_cls_2": 1.27569, "loss_box_2": 2.82654, "loss_cns_2": 0.55775, "loss_yns_2": 0.17949, "loss_cls_3": 1.25397, "loss_box_3": 2.90629, "loss_cns_3": 0.57038, "loss_yns_3": 0.17359, "loss_cls_4": 1.24124, "loss_box_4": 2.94292, "loss_cns_4": 0.55293, "loss_yns_4": 0.18213, "loss_cls_5": 1.26731, "loss_box_5": 3.15318, "loss_cns_5": 0.55535, "loss_yns_5": 0.17695, "loss_cls_dn_0": 0.44433, "loss_box_dn_0": 1.04745, "loss_cls_dn_1": 0.44161, "loss_box_dn_1": 1.58784, "loss_cls_dn_2": 0.44287, "loss_box_dn_2": 1.5892, "loss_cls_dn_3": 0.45307, "loss_box_dn_3": 1.68261, "loss_cls_dn_4": 0.45789, "loss_box_dn_4": 1.71704, "loss_cls_dn_5": 0.45661, "loss_box_dn_5": 1.82953, "loss_dense_depth": 2.1862, "loss": 43.23967, "grad_norm": 76.32958, "time": 1.48889} -{"mode": "train", "epoch": 1, "iter": 16, "lr": 0.00011, "memory": 49163, "data_time": 0.08086, "loss_cls_0": 1.18963, "loss_box_0": 2.33791, "loss_cns_0": 0.58966, "loss_yns_0": 0.17063, "loss_cls_1": 1.21755, "loss_box_1": 2.83659, "loss_cns_1": 0.53329, "loss_yns_1": 0.18482, "loss_cls_2": 1.24184, "loss_box_2": 2.85432, "loss_cns_2": 0.53739, "loss_yns_2": 0.1866, "loss_cls_3": 1.25831, "loss_box_3": 2.8301, "loss_cns_3": 0.53784, "loss_yns_3": 0.16944, "loss_cls_4": 1.24341, "loss_box_4": 2.80395, "loss_cns_4": 0.54891, "loss_yns_4": 0.17546, "loss_cls_5": 1.26885, "loss_box_5": 2.95263, "loss_cns_5": 0.55375, "loss_yns_5": 0.18326, "loss_cls_dn_0": 0.46225, "loss_box_dn_0": 1.03833, "loss_cls_dn_1": 0.47242, "loss_box_dn_1": 1.57499, "loss_cls_dn_2": 0.45263, "loss_box_dn_2": 1.50892, "loss_cls_dn_3": 0.4699, "loss_box_dn_3": 1.57452, "loss_cls_dn_4": 0.44896, "loss_box_dn_4": 1.54493, "loss_cls_dn_5": 0.45832, "loss_box_dn_5": 1.64564, "loss_dense_depth": 2.23769, "loss": 42.29567, "grad_norm": 61.71518, "time": 1.4897} -{"mode": "train", "epoch": 1, "iter": 17, "lr": 0.00011, "memory": 49163, "data_time": 0.08211, "loss_cls_0": 1.15823, "loss_box_0": 2.32167, "loss_cns_0": 0.5924, "loss_yns_0": 0.17003, "loss_cls_1": 1.21391, "loss_box_1": 3.01216, "loss_cns_1": 0.50739, "loss_yns_1": 0.18017, "loss_cls_2": 1.23108, "loss_box_2": 2.93987, "loss_cns_2": 0.52887, "loss_yns_2": 0.18711, "loss_cls_3": 1.24878, "loss_box_3": 2.88631, "loss_cns_3": 0.54042, "loss_yns_3": 0.17694, "loss_cls_4": 1.24532, "loss_box_4": 2.91495, "loss_cns_4": 0.53953, "loss_yns_4": 0.17669, "loss_cls_5": 1.26564, "loss_box_5": 2.98421, "loss_cns_5": 0.5317, "loss_yns_5": 0.18106, "loss_cls_dn_0": 0.48106, "loss_box_dn_0": 1.05169, "loss_cls_dn_1": 0.44616, "loss_box_dn_1": 1.73821, "loss_cls_dn_2": 0.43742, "loss_box_dn_2": 1.67686, "loss_cls_dn_3": 0.4368, "loss_box_dn_3": 1.75106, "loss_cls_dn_4": 0.41967, "loss_box_dn_4": 1.75747, "loss_cls_dn_5": 0.43006, "loss_box_dn_5": 1.8383, "loss_dense_depth": 2.07638, "loss": 43.27557, "grad_norm": 51.37837, "time": 1.49044} -{"mode": "train", "epoch": 1, "iter": 18, "lr": 0.00011, "memory": 49163, "data_time": 0.08606, "loss_cls_0": 1.16412, "loss_box_0": 2.27451, "loss_cns_0": 0.59476, "loss_yns_0": 0.16847, "loss_cls_1": 1.22969, "loss_box_1": 2.86927, "loss_cns_1": 0.53996, "loss_yns_1": 0.17895, "loss_cls_2": 1.23278, "loss_box_2": 2.95068, "loss_cns_2": 0.52438, "loss_yns_2": 0.18434, "loss_cls_3": 1.26414, "loss_box_3": 3.00947, "loss_cns_3": 0.542, "loss_yns_3": 0.17438, "loss_cls_4": 1.23561, "loss_box_4": 3.13439, "loss_cns_4": 0.49659, "loss_yns_4": 0.18192, "loss_cls_5": 1.24006, "loss_box_5": 3.18911, "loss_cns_5": 0.47647, "loss_yns_5": 0.1929, "loss_cls_dn_0": 0.47302, "loss_box_dn_0": 1.03623, "loss_cls_dn_1": 0.4167, "loss_box_dn_1": 1.67701, "loss_cls_dn_2": 0.4226, "loss_box_dn_2": 1.73373, "loss_cls_dn_3": 0.4152, "loss_box_dn_3": 1.85044, "loss_cls_dn_4": 0.41386, "loss_box_dn_4": 1.92206, "loss_cls_dn_5": 0.43255, "loss_box_dn_5": 2.01735, "loss_dense_depth": 2.06521, "loss": 43.92488, "grad_norm": 79.8852, "time": 1.50399} -{"mode": "train", "epoch": 1, "iter": 19, "lr": 0.00011, "memory": 49163, "data_time": 0.08494, "loss_cls_0": 1.17213, "loss_box_0": 2.29201, "loss_cns_0": 0.58794, "loss_yns_0": 0.16798, "loss_cls_1": 1.2215, "loss_box_1": 2.82642, "loss_cns_1": 0.54627, "loss_yns_1": 0.1824, "loss_cls_2": 1.21866, "loss_box_2": 2.91845, "loss_cns_2": 0.53237, "loss_yns_2": 0.18022, "loss_cls_3": 1.26752, "loss_box_3": 2.98502, "loss_cns_3": 0.5333, "loss_yns_3": 0.17181, "loss_cls_4": 1.21948, "loss_box_4": 3.09562, "loss_cns_4": 0.50286, "loss_yns_4": 0.18644, "loss_cls_5": 1.22906, "loss_box_5": 3.15737, "loss_cns_5": 0.50471, "loss_yns_5": 0.17532, "loss_cls_dn_0": 0.4767, "loss_box_dn_0": 1.04047, "loss_cls_dn_1": 0.42601, "loss_box_dn_1": 1.35632, "loss_cls_dn_2": 0.43991, "loss_box_dn_2": 1.52698, "loss_cls_dn_3": 0.42532, "loss_box_dn_3": 1.64132, "loss_cls_dn_4": 0.42168, "loss_box_dn_4": 1.78458, "loss_cls_dn_5": 0.4417, "loss_box_dn_5": 1.90312, "loss_dense_depth": 1.92166, "loss": 42.6806, "grad_norm": 90.35128, "time": 1.49208} -{"mode": "train", "epoch": 1, "iter": 20, "lr": 0.00011, "memory": 49163, "data_time": 0.07757, "loss_cls_0": 1.16885, "loss_box_0": 2.30315, "loss_cns_0": 0.58492, "loss_yns_0": 0.16887, "loss_cls_1": 1.22965, "loss_box_1": 2.73153, "loss_cns_1": 0.56354, "loss_yns_1": 0.17901, "loss_cls_2": 1.22837, "loss_box_2": 2.73966, "loss_cns_2": 0.56677, "loss_yns_2": 0.17758, "loss_cls_3": 1.26823, "loss_box_3": 2.74078, "loss_cns_3": 0.56263, "loss_yns_3": 0.17255, "loss_cls_4": 1.26825, "loss_box_4": 2.76826, "loss_cns_4": 0.5868, "loss_yns_4": 0.18078, "loss_cls_5": 1.28831, "loss_box_5": 2.81453, "loss_cns_5": 0.60067, "loss_yns_5": 0.18578, "loss_cls_dn_0": 0.46286, "loss_box_dn_0": 1.02394, "loss_cls_dn_1": 0.40322, "loss_box_dn_1": 1.45852, "loss_cls_dn_2": 0.41612, "loss_box_dn_2": 1.51463, "loss_cls_dn_3": 0.3995, "loss_box_dn_3": 1.54223, "loss_cls_dn_4": 0.38498, "loss_box_dn_4": 1.60454, "loss_cls_dn_5": 0.39213, "loss_box_dn_5": 1.65137, "loss_dense_depth": 1.67318, "loss": 41.00668, "grad_norm": 52.50822, "time": 1.48438} -{"mode": "train", "epoch": 1, "iter": 21, "lr": 0.00011, "memory": 49163, "data_time": 0.1384, "loss_cls_0": 1.1458, "loss_box_0": 2.26572, "loss_cns_0": 0.59418, "loss_yns_0": 0.16752, "loss_cls_1": 1.22942, "loss_box_1": 2.75458, "loss_cns_1": 0.57201, "loss_yns_1": 0.1736, "loss_cls_2": 1.22708, "loss_box_2": 2.77294, "loss_cns_2": 0.54959, "loss_yns_2": 0.17829, "loss_cls_3": 1.24242, "loss_box_3": 2.80808, "loss_cns_3": 0.54486, "loss_yns_3": 0.1701, "loss_cls_4": 1.25564, "loss_box_4": 2.86522, "loss_cns_4": 0.53329, "loss_yns_4": 0.17967, "loss_cls_5": 1.26979, "loss_box_5": 2.98413, "loss_cns_5": 0.47736, "loss_yns_5": 0.18278, "loss_cls_dn_0": 0.46328, "loss_box_dn_0": 1.03301, "loss_cls_dn_1": 0.41768, "loss_box_dn_1": 1.2503, "loss_cls_dn_2": 0.4337, "loss_box_dn_2": 1.26175, "loss_cls_dn_3": 0.43524, "loss_box_dn_3": 1.28469, "loss_cls_dn_4": 0.42372, "loss_box_dn_4": 1.37999, "loss_cls_dn_5": 0.42205, "loss_box_dn_5": 1.39277, "loss_dense_depth": 1.54763, "loss": 39.8899, "grad_norm": 82.46423, "time": 1.56479} -{"mode": "train", "epoch": 1, "iter": 22, "lr": 0.00011, "memory": 49163, "data_time": 0.1176, "loss_cls_0": 1.15875, "loss_box_0": 2.23142, "loss_cns_0": 0.60234, "loss_yns_0": 0.16754, "loss_cls_1": 1.24579, "loss_box_1": 2.80567, "loss_cns_1": 0.56037, "loss_yns_1": 0.17742, "loss_cls_2": 1.22416, "loss_box_2": 2.85297, "loss_cns_2": 0.53635, "loss_yns_2": 0.18769, "loss_cls_3": 1.24293, "loss_box_3": 2.9284, "loss_cns_3": 0.54155, "loss_yns_3": 0.17213, "loss_cls_4": 1.22874, "loss_box_4": 2.97319, "loss_cns_4": 0.52276, "loss_yns_4": 0.18017, "loss_cls_5": 1.25643, "loss_box_5": 3.03234, "loss_cns_5": 0.49098, "loss_yns_5": 0.17618, "loss_cls_dn_0": 0.4565, "loss_box_dn_0": 1.02722, "loss_cls_dn_1": 0.42801, "loss_box_dn_1": 1.26172, "loss_cls_dn_2": 0.4462, "loss_box_dn_2": 1.32973, "loss_cls_dn_3": 0.46164, "loss_box_dn_3": 1.41356, "loss_cls_dn_4": 0.46849, "loss_box_dn_4": 1.55814, "loss_cls_dn_5": 0.45515, "loss_box_dn_5": 1.5652, "loss_dense_depth": 1.47438, "loss": 40.84222, "grad_norm": 95.14938, "time": 1.64513} -{"mode": "train", "epoch": 1, "iter": 23, "lr": 0.00011, "memory": 49163, "data_time": 0.0798, "loss_cls_0": 1.14155, "loss_box_0": 2.17897, "loss_cns_0": 0.60936, "loss_yns_0": 0.1695, "loss_cls_1": 1.23567, "loss_box_1": 2.79082, "loss_cns_1": 0.56018, "loss_yns_1": 0.17489, "loss_cls_2": 1.22522, "loss_box_2": 2.81748, "loss_cns_2": 0.55431, "loss_yns_2": 0.18886, "loss_cls_3": 1.25385, "loss_box_3": 2.9431, "loss_cns_3": 0.56549, "loss_yns_3": 0.17312, "loss_cls_4": 1.23606, "loss_box_4": 2.98958, "loss_cns_4": 0.55475, "loss_yns_4": 0.17233, "loss_cls_5": 1.25793, "loss_box_5": 2.9306, "loss_cns_5": 0.55443, "loss_yns_5": 0.17755, "loss_cls_dn_0": 0.4513, "loss_box_dn_0": 1.0006, "loss_cls_dn_1": 0.43085, "loss_box_dn_1": 1.35817, "loss_cls_dn_2": 0.44937, "loss_box_dn_2": 1.41116, "loss_cls_dn_3": 0.46745, "loss_box_dn_3": 1.51958, "loss_cls_dn_4": 0.47727, "loss_box_dn_4": 1.64885, "loss_cls_dn_5": 0.46555, "loss_box_dn_5": 1.65422, "loss_dense_depth": 1.49662, "loss": 41.28659, "grad_norm": 90.62737, "time": 1.52358} -{"mode": "train", "epoch": 1, "iter": 24, "lr": 0.00011, "memory": 49163, "data_time": 0.08389, "loss_cls_0": 1.16358, "loss_box_0": 2.20625, "loss_cns_0": 0.60987, "loss_yns_0": 0.16935, "loss_cls_1": 1.24714, "loss_box_1": 2.8118, "loss_cns_1": 0.55432, "loss_yns_1": 0.17354, "loss_cls_2": 1.24918, "loss_box_2": 2.8069, "loss_cns_2": 0.55254, "loss_yns_2": 0.17662, "loss_cls_3": 1.26983, "loss_box_3": 2.94417, "loss_cns_3": 0.5416, "loss_yns_3": 0.17309, "loss_cls_4": 1.24919, "loss_box_4": 2.96449, "loss_cns_4": 0.5389, "loss_yns_4": 0.17928, "loss_cls_5": 1.27336, "loss_box_5": 2.89199, "loss_cns_5": 0.54702, "loss_yns_5": 0.17655, "loss_cls_dn_0": 0.46807, "loss_box_dn_0": 1.01824, "loss_cls_dn_1": 0.41548, "loss_box_dn_1": 1.53238, "loss_cls_dn_2": 0.42998, "loss_box_dn_2": 1.54425, "loss_cls_dn_3": 0.44505, "loss_box_dn_3": 1.65763, "loss_cls_dn_4": 0.44444, "loss_box_dn_4": 1.75863, "loss_cls_dn_5": 0.44756, "loss_box_dn_5": 1.76434, "loss_dense_depth": 1.42359, "loss": 41.8202, "grad_norm": 87.79007, "time": 1.53505} -{"mode": "train", "epoch": 1, "iter": 25, "lr": 0.00011, "memory": 49163, "data_time": 0.08811, "loss_cls_0": 1.16595, "loss_box_0": 2.2248, "loss_cns_0": 0.6091, "loss_yns_0": 0.16824, "loss_cls_1": 1.24657, "loss_box_1": 2.73453, "loss_cns_1": 0.55568, "loss_yns_1": 0.17344, "loss_cls_2": 1.25107, "loss_box_2": 2.83765, "loss_cns_2": 0.54795, "loss_yns_2": 0.18222, "loss_cls_3": 1.25951, "loss_box_3": 2.95753, "loss_cns_3": 0.53404, "loss_yns_3": 0.17188, "loss_cls_4": 1.29533, "loss_box_4": 2.83258, "loss_cns_4": 0.55498, "loss_yns_4": 0.18287, "loss_cls_5": 1.26704, "loss_box_5": 2.79423, "loss_cns_5": 0.56582, "loss_yns_5": 0.17578, "loss_cls_dn_0": 0.46992, "loss_box_dn_0": 1.0029, "loss_cls_dn_1": 0.39776, "loss_box_dn_1": 1.54925, "loss_cls_dn_2": 0.40734, "loss_box_dn_2": 1.585, "loss_cls_dn_3": 0.42096, "loss_box_dn_3": 1.6934, "loss_cls_dn_4": 0.40189, "loss_box_dn_4": 1.71043, "loss_cls_dn_5": 0.43195, "loss_box_dn_5": 1.73947, "loss_dense_depth": 1.38317, "loss": 41.48225, "grad_norm": 75.9091, "time": 1.5054} -{"mode": "train", "epoch": 1, "iter": 26, "lr": 0.00011, "memory": 49163, "data_time": 0.08904, "loss_cls_0": 1.14916, "loss_box_0": 2.23189, "loss_cns_0": 0.60834, "loss_yns_0": 0.16736, "loss_cls_1": 1.25474, "loss_box_1": 2.61759, "loss_cns_1": 0.59364, "loss_yns_1": 0.17405, "loss_cls_2": 1.26475, "loss_box_2": 2.74254, "loss_cns_2": 0.59691, "loss_yns_2": 0.17756, "loss_cls_3": 1.2595, "loss_box_3": 2.80944, "loss_cns_3": 0.59255, "loss_yns_3": 0.17259, "loss_cls_4": 1.29106, "loss_box_4": 2.71874, "loss_cns_4": 0.60309, "loss_yns_4": 0.17332, "loss_cls_5": 1.26101, "loss_box_5": 2.73324, "loss_cns_5": 0.60473, "loss_yns_5": 0.17322, "loss_cls_dn_0": 0.44835, "loss_box_dn_0": 1.00882, "loss_cls_dn_1": 0.41268, "loss_box_dn_1": 1.28419, "loss_cls_dn_2": 0.41752, "loss_box_dn_2": 1.38937, "loss_cls_dn_3": 0.43588, "loss_box_dn_3": 1.48324, "loss_cls_dn_4": 0.42033, "loss_box_dn_4": 1.46792, "loss_cls_dn_5": 0.44681, "loss_box_dn_5": 1.5043, "loss_dense_depth": 1.33957, "loss": 40.03001, "grad_norm": 73.30914, "time": 1.53174} -{"mode": "train", "epoch": 1, "iter": 27, "lr": 0.00011, "memory": 49163, "data_time": 0.09735, "loss_cls_0": 1.15402, "loss_box_0": 2.24701, "loss_cns_0": 0.60389, "loss_yns_0": 0.17102, "loss_cls_1": 1.22054, "loss_box_1": 2.59646, "loss_cns_1": 0.58792, "loss_yns_1": 0.17252, "loss_cls_2": 1.24728, "loss_box_2": 2.7375, "loss_cns_2": 0.57903, "loss_yns_2": 0.17804, "loss_cls_3": 1.25179, "loss_box_3": 2.76326, "loss_cns_3": 0.57792, "loss_yns_3": 0.17195, "loss_cls_4": 1.23279, "loss_box_4": 2.8824, "loss_cns_4": 0.55198, "loss_yns_4": 0.18699, "loss_cls_5": 1.24558, "loss_box_5": 3.00561, "loss_cns_5": 0.53495, "loss_yns_5": 0.17612, "loss_cls_dn_0": 0.43772, "loss_box_dn_0": 1.00371, "loss_cls_dn_1": 0.41691, "loss_box_dn_1": 1.19828, "loss_cls_dn_2": 0.40979, "loss_box_dn_2": 1.28502, "loss_cls_dn_3": 0.42223, "loss_box_dn_3": 1.34136, "loss_cls_dn_4": 0.44377, "loss_box_dn_4": 1.35478, "loss_cls_dn_5": 0.42617, "loss_box_dn_5": 1.44392, "loss_dense_depth": 1.32965, "loss": 39.5899, "grad_norm": 85.97633, "time": 1.50432} -{"mode": "train", "epoch": 1, "iter": 28, "lr": 0.00011, "memory": 49163, "data_time": 0.08053, "loss_cls_0": 1.1334, "loss_box_0": 2.21939, "loss_cns_0": 0.60764, "loss_yns_0": 0.16792, "loss_cls_1": 1.17701, "loss_box_1": 2.58204, "loss_cns_1": 0.56728, "loss_yns_1": 0.16975, "loss_cls_2": 1.20211, "loss_box_2": 2.63235, "loss_cns_2": 0.56354, "loss_yns_2": 0.17463, "loss_cls_3": 1.25017, "loss_box_3": 2.61914, "loss_cns_3": 0.56574, "loss_yns_3": 0.17493, "loss_cls_4": 1.20046, "loss_box_4": 2.7061, "loss_cns_4": 0.55712, "loss_yns_4": 0.18686, "loss_cls_5": 1.29772, "loss_box_5": 2.74538, "loss_cns_5": 0.56622, "loss_yns_5": 0.17863, "loss_cls_dn_0": 0.43056, "loss_box_dn_0": 0.99688, "loss_cls_dn_1": 0.42184, "loss_box_dn_1": 1.25211, "loss_cls_dn_2": 0.40963, "loss_box_dn_2": 1.30366, "loss_cls_dn_3": 0.4021, "loss_box_dn_3": 1.34531, "loss_cls_dn_4": 0.4564, "loss_box_dn_4": 1.3648, "loss_cls_dn_5": 0.39505, "loss_box_dn_5": 1.47467, "loss_dense_depth": 1.25588, "loss": 38.75442, "grad_norm": 84.46518, "time": 1.49595} -{"mode": "train", "epoch": 1, "iter": 29, "lr": 0.00011, "memory": 49163, "data_time": 0.09761, "loss_cls_0": 1.1004, "loss_box_0": 2.15331, "loss_cns_0": 0.61192, "loss_yns_0": 0.16937, "loss_cls_1": 1.1695, "loss_box_1": 2.53677, "loss_cns_1": 0.578, "loss_yns_1": 0.17118, "loss_cls_2": 1.19059, "loss_box_2": 2.50846, "loss_cns_2": 0.58176, "loss_yns_2": 0.17612, "loss_cls_3": 1.24785, "loss_box_3": 2.55221, "loss_cns_3": 0.5836, "loss_yns_3": 0.17696, "loss_cls_4": 1.19908, "loss_box_4": 2.56928, "loss_cns_4": 0.58769, "loss_yns_4": 0.1817, "loss_cls_5": 1.29874, "loss_box_5": 2.60454, "loss_cns_5": 0.58023, "loss_yns_5": 0.1755, "loss_cls_dn_0": 0.43637, "loss_box_dn_0": 0.99127, "loss_cls_dn_1": 0.42355, "loss_box_dn_1": 1.19118, "loss_cls_dn_2": 0.41822, "loss_box_dn_2": 1.20436, "loss_cls_dn_3": 0.40708, "loss_box_dn_3": 1.27369, "loss_cls_dn_4": 0.46497, "loss_box_dn_4": 1.28958, "loss_cls_dn_5": 0.40431, "loss_box_dn_5": 1.41227, "loss_dense_depth": 1.23574, "loss": 37.85735, "grad_norm": 64.18747, "time": 1.53143} -{"mode": "train", "epoch": 1, "iter": 30, "lr": 0.00011, "memory": 49163, "data_time": 0.08225, "loss_cls_0": 1.08463, "loss_box_0": 2.05971, "loss_cns_0": 0.61609, "loss_yns_0": 0.16956, "loss_cls_1": 1.15036, "loss_box_1": 2.51428, "loss_cns_1": 0.58962, "loss_yns_1": 0.17415, "loss_cls_2": 1.16616, "loss_box_2": 2.53398, "loss_cns_2": 0.59195, "loss_yns_2": 0.17904, "loss_cls_3": 1.19965, "loss_box_3": 2.59869, "loss_cns_3": 0.58742, "loss_yns_3": 0.18081, "loss_cls_4": 1.18401, "loss_box_4": 2.59058, "loss_cns_4": 0.57847, "loss_yns_4": 0.17756, "loss_cls_5": 1.1954, "loss_box_5": 2.62912, "loss_cns_5": 0.56244, "loss_yns_5": 0.17381, "loss_cls_dn_0": 0.43602, "loss_box_dn_0": 0.99228, "loss_cls_dn_1": 0.40212, "loss_box_dn_1": 1.27701, "loss_cls_dn_2": 0.41359, "loss_box_dn_2": 1.29407, "loss_cls_dn_3": 0.41653, "loss_box_dn_3": 1.36079, "loss_cls_dn_4": 0.43069, "loss_box_dn_4": 1.3719, "loss_cls_dn_5": 0.42247, "loss_box_dn_5": 1.46418, "loss_dense_depth": 1.2329, "loss": 38.00207, "grad_norm": 75.94876, "time": 1.52061} -{"mode": "train", "epoch": 1, "iter": 31, "lr": 0.00011, "memory": 49163, "data_time": 0.07843, "loss_cls_0": 1.07927, "loss_box_0": 2.08224, "loss_cns_0": 0.61457, "loss_yns_0": 0.17177, "loss_cls_1": 1.15049, "loss_box_1": 2.50736, "loss_cns_1": 0.58668, "loss_yns_1": 0.17343, "loss_cls_2": 1.15885, "loss_box_2": 2.50809, "loss_cns_2": 0.5928, "loss_yns_2": 0.17789, "loss_cls_3": 1.18673, "loss_box_3": 2.58121, "loss_cns_3": 0.58768, "loss_yns_3": 0.17544, "loss_cls_4": 1.18624, "loss_box_4": 2.6511, "loss_cns_4": 0.57299, "loss_yns_4": 0.18151, "loss_cls_5": 1.18889, "loss_box_5": 2.73067, "loss_cns_5": 0.55762, "loss_yns_5": 0.17696, "loss_cls_dn_0": 0.44539, "loss_box_dn_0": 0.98817, "loss_cls_dn_1": 0.38247, "loss_box_dn_1": 1.28102, "loss_cls_dn_2": 0.40516, "loss_box_dn_2": 1.31025, "loss_cls_dn_3": 0.43162, "loss_box_dn_3": 1.36332, "loss_cls_dn_4": 0.4033, "loss_box_dn_4": 1.40841, "loss_cls_dn_5": 0.4398, "loss_box_dn_5": 1.48348, "loss_dense_depth": 1.19866, "loss": 38.12153, "grad_norm": 74.6407, "time": 1.49913} -{"mode": "train", "epoch": 1, "iter": 32, "lr": 0.00011, "memory": 49163, "data_time": 0.0764, "loss_cls_0": 1.05822, "loss_box_0": 2.11244, "loss_cns_0": 0.61218, "loss_yns_0": 0.16926, "loss_cls_1": 1.13726, "loss_box_1": 2.49393, "loss_cns_1": 0.57656, "loss_yns_1": 0.16925, "loss_cls_2": 1.1506, "loss_box_2": 2.50826, "loss_cns_2": 0.58218, "loss_yns_2": 0.17493, "loss_cls_3": 1.16111, "loss_box_3": 2.53248, "loss_cns_3": 0.58559, "loss_yns_3": 0.19462, "loss_cls_4": 1.18363, "loss_box_4": 2.67054, "loss_cns_4": 0.57563, "loss_yns_4": 0.18696, "loss_cls_5": 1.1761, "loss_box_5": 2.75841, "loss_cns_5": 0.56738, "loss_yns_5": 0.17555, "loss_cls_dn_0": 0.44187, "loss_box_dn_0": 0.99388, "loss_cls_dn_1": 0.36633, "loss_box_dn_1": 1.25947, "loss_cls_dn_2": 0.38476, "loss_box_dn_2": 1.29114, "loss_cls_dn_3": 0.42582, "loss_box_dn_3": 1.31983, "loss_cls_dn_4": 0.3771, "loss_box_dn_4": 1.40811, "loss_cls_dn_5": 0.43982, "loss_box_dn_5": 1.46975, "loss_dense_depth": 1.13628, "loss": 37.8272, "grad_norm": 70.51804, "time": 1.50676} -{"mode": "train", "epoch": 1, "iter": 33, "lr": 0.00011, "memory": 49163, "data_time": 0.07454, "loss_cls_0": 1.07341, "loss_box_0": 2.1094, "loss_cns_0": 0.61344, "loss_yns_0": 0.1687, "loss_cls_1": 1.13584, "loss_box_1": 2.47912, "loss_cns_1": 0.58065, "loss_yns_1": 0.17043, "loss_cls_2": 1.16791, "loss_box_2": 2.51265, "loss_cns_2": 0.58326, "loss_yns_2": 0.17574, "loss_cls_3": 1.15681, "loss_box_3": 2.49709, "loss_cns_3": 0.58633, "loss_yns_3": 0.18826, "loss_cls_4": 1.21031, "loss_box_4": 2.55685, "loss_cns_4": 0.58587, "loss_yns_4": 0.17972, "loss_cls_5": 1.17192, "loss_box_5": 2.5658, "loss_cns_5": 0.59143, "loss_yns_5": 0.17199, "loss_cls_dn_0": 0.43661, "loss_box_dn_0": 1.00328, "loss_cls_dn_1": 0.34414, "loss_box_dn_1": 1.32241, "loss_cls_dn_2": 0.35772, "loss_box_dn_2": 1.3491, "loss_cls_dn_3": 0.3876, "loss_box_dn_3": 1.3345, "loss_cls_dn_4": 0.35076, "loss_box_dn_4": 1.38566, "loss_cls_dn_5": 0.40687, "loss_box_dn_5": 1.40397, "loss_dense_depth": 1.11093, "loss": 37.42646, "grad_norm": 57.22676, "time": 1.50503} -{"mode": "train", "epoch": 1, "iter": 34, "lr": 0.00011, "memory": 49163, "data_time": 0.07098, "loss_cls_0": 1.07265, "loss_box_0": 2.09138, "loss_cns_0": 0.61363, "loss_yns_0": 0.17119, "loss_cls_1": 1.13324, "loss_box_1": 2.43069, "loss_cns_1": 0.5876, "loss_yns_1": 0.17294, "loss_cls_2": 1.15944, "loss_box_2": 2.49141, "loss_cns_2": 0.58398, "loss_yns_2": 0.17308, "loss_cls_3": 1.17497, "loss_box_3": 2.44221, "loss_cns_3": 0.58748, "loss_yns_3": 0.16977, "loss_cls_4": 1.18544, "loss_box_4": 2.49911, "loss_cns_4": 0.57898, "loss_yns_4": 0.17331, "loss_cls_5": 1.16572, "loss_box_5": 2.55195, "loss_cns_5": 0.59067, "loss_yns_5": 0.17358, "loss_cls_dn_0": 0.4283, "loss_box_dn_0": 1.00346, "loss_cls_dn_1": 0.33952, "loss_box_dn_1": 1.34103, "loss_cls_dn_2": 0.36007, "loss_box_dn_2": 1.35509, "loss_cls_dn_3": 0.36861, "loss_box_dn_3": 1.31871, "loss_cls_dn_4": 0.3571, "loss_box_dn_4": 1.36092, "loss_cls_dn_5": 0.39266, "loss_box_dn_5": 1.37564, "loss_dense_depth": 1.10611, "loss": 37.08167, "grad_norm": 51.088, "time": 1.48779} -{"mode": "train", "epoch": 1, "iter": 35, "lr": 0.00011, "memory": 49163, "data_time": 0.071, "loss_cls_0": 1.06717, "loss_box_0": 2.04662, "loss_cns_0": 0.6195, "loss_yns_0": 0.1712, "loss_cls_1": 1.12941, "loss_box_1": 2.52824, "loss_cns_1": 0.58646, "loss_yns_1": 0.17266, "loss_cls_2": 1.1459, "loss_box_2": 2.58822, "loss_cns_2": 0.58419, "loss_yns_2": 0.17395, "loss_cls_3": 1.16504, "loss_box_3": 2.56291, "loss_cns_3": 0.59343, "loss_yns_3": 0.17789, "loss_cls_4": 1.15525, "loss_box_4": 2.63542, "loss_cns_4": 0.5887, "loss_yns_4": 0.17281, "loss_cls_5": 1.16183, "loss_box_5": 2.70011, "loss_cns_5": 0.60449, "loss_yns_5": 0.1787, "loss_cls_dn_0": 0.41412, "loss_box_dn_0": 0.99166, "loss_cls_dn_1": 0.33534, "loss_box_dn_1": 1.33252, "loss_cls_dn_2": 0.36861, "loss_box_dn_2": 1.35652, "loss_cls_dn_3": 0.37388, "loss_box_dn_3": 1.34898, "loss_cls_dn_4": 0.3753, "loss_box_dn_4": 1.40371, "loss_cls_dn_5": 0.38666, "loss_box_dn_5": 1.44788, "loss_dense_depth": 1.07841, "loss": 37.7237, "grad_norm": 72.6019, "time": 1.48568} -{"mode": "train", "epoch": 1, "iter": 36, "lr": 0.00011, "memory": 49163, "data_time": 0.07127, "loss_cls_0": 1.06482, "loss_box_0": 2.02648, "loss_cns_0": 0.61729, "loss_yns_0": 0.16917, "loss_cls_1": 1.11486, "loss_box_1": 2.62099, "loss_cns_1": 0.57828, "loss_yns_1": 0.17102, "loss_cls_2": 1.12476, "loss_box_2": 2.65092, "loss_cns_2": 0.57407, "loss_yns_2": 0.17874, "loss_cls_3": 1.14207, "loss_box_3": 2.68705, "loss_cns_3": 0.57807, "loss_yns_3": 0.1809, "loss_cls_4": 1.1405, "loss_box_4": 2.73792, "loss_cns_4": 0.5856, "loss_yns_4": 0.17536, "loss_cls_5": 1.16175, "loss_box_5": 2.81424, "loss_cns_5": 0.58814, "loss_yns_5": 0.17551, "loss_cls_dn_0": 0.40234, "loss_box_dn_0": 0.98495, "loss_cls_dn_1": 0.34951, "loss_box_dn_1": 1.20314, "loss_cls_dn_2": 0.39291, "loss_box_dn_2": 1.23555, "loss_cls_dn_3": 0.41122, "loss_box_dn_3": 1.28493, "loss_cls_dn_4": 0.40974, "loss_box_dn_4": 1.35773, "loss_cls_dn_5": 0.39953, "loss_box_dn_5": 1.43987, "loss_dense_depth": 1.06404, "loss": 37.79397, "grad_norm": 66.1207, "time": 1.49283} -{"mode": "train", "epoch": 1, "iter": 37, "lr": 0.00011, "memory": 49163, "data_time": 0.08036, "loss_cls_0": 1.05999, "loss_box_0": 2.01219, "loss_cns_0": 0.61756, "loss_yns_0": 0.17241, "loss_cls_1": 1.11718, "loss_box_1": 2.63287, "loss_cns_1": 0.57867, "loss_yns_1": 0.16978, "loss_cls_2": 1.14195, "loss_box_2": 2.67831, "loss_cns_2": 0.5724, "loss_yns_2": 0.1764, "loss_cls_3": 1.1408, "loss_box_3": 2.72734, "loss_cns_3": 0.57477, "loss_yns_3": 0.17486, "loss_cls_4": 1.14385, "loss_box_4": 2.77484, "loss_cns_4": 0.57463, "loss_yns_4": 0.17497, "loss_cls_5": 1.19505, "loss_box_5": 2.8459, "loss_cns_5": 0.57949, "loss_yns_5": 0.17117, "loss_cls_dn_0": 0.40661, "loss_box_dn_0": 0.99555, "loss_cls_dn_1": 0.33216, "loss_box_dn_1": 1.33243, "loss_cls_dn_2": 0.37308, "loss_box_dn_2": 1.35205, "loss_cls_dn_3": 0.39355, "loss_box_dn_3": 1.4085, "loss_cls_dn_4": 0.40081, "loss_box_dn_4": 1.46895, "loss_cls_dn_5": 0.36883, "loss_box_dn_5": 1.54651, "loss_dense_depth": 1.10573, "loss": 38.49216, "grad_norm": 67.76736, "time": 1.49305} -{"mode": "train", "epoch": 1, "iter": 38, "lr": 0.00011, "memory": 49163, "data_time": 0.07823, "loss_cls_0": 1.03833, "loss_box_0": 2.01749, "loss_cns_0": 0.61649, "loss_yns_0": 0.17046, "loss_cls_1": 1.09871, "loss_box_1": 2.51997, "loss_cns_1": 0.58175, "loss_yns_1": 0.17055, "loss_cls_2": 1.12772, "loss_box_2": 2.55351, "loss_cns_2": 0.58118, "loss_yns_2": 0.17564, "loss_cls_3": 1.13974, "loss_box_3": 2.57274, "loss_cns_3": 0.58586, "loss_yns_3": 0.16994, "loss_cls_4": 1.13385, "loss_box_4": 2.57952, "loss_cns_4": 0.5879, "loss_yns_4": 0.17415, "loss_cls_5": 1.21665, "loss_box_5": 2.60012, "loss_cns_5": 0.59345, "loss_yns_5": 0.17099, "loss_cls_dn_0": 0.41961, "loss_box_dn_0": 1.00246, "loss_cls_dn_1": 0.31626, "loss_box_dn_1": 1.27873, "loss_cls_dn_2": 0.35957, "loss_box_dn_2": 1.28935, "loss_cls_dn_3": 0.37151, "loss_box_dn_3": 1.33466, "loss_cls_dn_4": 0.38549, "loss_box_dn_4": 1.36186, "loss_cls_dn_5": 0.35249, "loss_box_dn_5": 1.42438, "loss_dense_depth": 1.09786, "loss": 37.17093, "grad_norm": 55.54302, "time": 1.48461} -{"mode": "train", "epoch": 1, "iter": 39, "lr": 0.00012, "memory": 49163, "data_time": 0.07857, "loss_cls_0": 1.05769, "loss_box_0": 2.01398, "loss_cns_0": 0.61718, "loss_yns_0": 0.16701, "loss_cls_1": 1.11213, "loss_box_1": 2.33863, "loss_cns_1": 0.60367, "loss_yns_1": 0.17204, "loss_cls_2": 1.12146, "loss_box_2": 2.35843, "loss_cns_2": 0.60393, "loss_yns_2": 0.1724, "loss_cls_3": 1.17452, "loss_box_3": 2.40922, "loss_cns_3": 0.60382, "loss_yns_3": 0.16953, "loss_cls_4": 1.15369, "loss_box_4": 2.37291, "loss_cns_4": 0.61152, "loss_yns_4": 0.171, "loss_cls_5": 1.23252, "loss_box_5": 2.43498, "loss_cns_5": 0.59741, "loss_yns_5": 0.16859, "loss_cls_dn_0": 0.4231, "loss_box_dn_0": 1.00275, "loss_cls_dn_1": 0.31602, "loss_box_dn_1": 1.15683, "loss_cls_dn_2": 0.37159, "loss_box_dn_2": 1.1595, "loss_cls_dn_3": 0.36405, "loss_box_dn_3": 1.20225, "loss_cls_dn_4": 0.37649, "loss_box_dn_4": 1.19517, "loss_cls_dn_5": 0.35324, "loss_box_dn_5": 1.25779, "loss_dense_depth": 1.05414, "loss": 35.6712, "grad_norm": 51.8397, "time": 1.48389} -{"mode": "train", "epoch": 1, "iter": 40, "lr": 0.00012, "memory": 49163, "data_time": 0.07719, "loss_cls_0": 1.07438, "loss_box_0": 2.00908, "loss_cns_0": 0.61668, "loss_yns_0": 0.16615, "loss_cls_1": 1.11425, "loss_box_1": 2.3195, "loss_cns_1": 0.60171, "loss_yns_1": 0.17445, "loss_cls_2": 1.12272, "loss_box_2": 2.31305, "loss_cns_2": 0.60262, "loss_yns_2": 0.17343, "loss_cls_3": 1.15668, "loss_box_3": 2.35946, "loss_cns_3": 0.60473, "loss_yns_3": 0.1714, "loss_cls_4": 1.15821, "loss_box_4": 2.3468, "loss_cns_4": 0.60965, "loss_yns_4": 0.17284, "loss_cls_5": 1.16676, "loss_box_5": 2.39905, "loss_cns_5": 0.60173, "loss_yns_5": 0.16978, "loss_cls_dn_0": 0.40579, "loss_box_dn_0": 1.01119, "loss_cls_dn_1": 0.31299, "loss_box_dn_1": 1.14847, "loss_cls_dn_2": 0.37009, "loss_box_dn_2": 1.14782, "loss_cls_dn_3": 0.36116, "loss_box_dn_3": 1.19016, "loss_cls_dn_4": 0.36545, "loss_box_dn_4": 1.17447, "loss_cls_dn_5": 0.37392, "loss_box_dn_5": 1.23765, "loss_dense_depth": 1.0863, "loss": 35.39055, "grad_norm": 59.01241, "time": 1.51214} -{"mode": "train", "epoch": 1, "iter": 41, "lr": 0.00012, "memory": 49163, "data_time": 0.10583, "loss_cls_0": 1.02933, "loss_box_0": 1.99108, "loss_cns_0": 0.60997, "loss_yns_0": 0.16478, "loss_cls_1": 1.09717, "loss_box_1": 2.32332, "loss_cns_1": 0.60197, "loss_yns_1": 0.17172, "loss_cls_2": 1.1143, "loss_box_2": 2.31479, "loss_cns_2": 0.60249, "loss_yns_2": 0.16964, "loss_cls_3": 1.14389, "loss_box_3": 2.3428, "loss_cns_3": 0.60793, "loss_yns_3": 0.17046, "loss_cls_4": 1.15118, "loss_box_4": 2.36497, "loss_cns_4": 0.60637, "loss_yns_4": 0.1706, "loss_cls_5": 1.14554, "loss_box_5": 2.3918, "loss_cns_5": 0.6144, "loss_yns_5": 0.16461, "loss_cls_dn_0": 0.41812, "loss_box_dn_0": 1.00305, "loss_cls_dn_1": 0.2973, "loss_box_dn_1": 1.14718, "loss_cls_dn_2": 0.35161, "loss_box_dn_2": 1.14845, "loss_cls_dn_3": 0.35605, "loss_box_dn_3": 1.16572, "loss_cls_dn_4": 0.35738, "loss_box_dn_4": 1.15243, "loss_cls_dn_5": 0.38501, "loss_box_dn_5": 1.18985, "loss_dense_depth": 1.04604, "loss": 35.0833, "grad_norm": 56.69111, "time": 1.52733} -{"mode": "train", "epoch": 1, "iter": 42, "lr": 0.00012, "memory": 49163, "data_time": 0.10402, "loss_cls_0": 1.0227, "loss_box_0": 1.9801, "loss_cns_0": 0.61541, "loss_yns_0": 0.16488, "loss_cls_1": 1.08784, "loss_box_1": 2.31227, "loss_cns_1": 0.60793, "loss_yns_1": 0.17065, "loss_cls_2": 1.10113, "loss_box_2": 2.31257, "loss_cns_2": 0.60475, "loss_yns_2": 0.16878, "loss_cls_3": 1.12979, "loss_box_3": 2.33726, "loss_cns_3": 0.60895, "loss_yns_3": 0.17055, "loss_cls_4": 1.12991, "loss_box_4": 2.35909, "loss_cns_4": 0.60298, "loss_yns_4": 0.16813, "loss_cls_5": 1.14381, "loss_box_5": 2.37981, "loss_cns_5": 0.6107, "loss_yns_5": 0.16587, "loss_cls_dn_0": 0.40921, "loss_box_dn_0": 1.00289, "loss_cls_dn_1": 0.30232, "loss_box_dn_1": 1.10611, "loss_cls_dn_2": 0.34298, "loss_box_dn_2": 1.11569, "loss_cls_dn_3": 0.36872, "loss_box_dn_3": 1.14455, "loss_cls_dn_4": 0.36836, "loss_box_dn_4": 1.14718, "loss_cls_dn_5": 0.41129, "loss_box_dn_5": 1.17115, "loss_dense_depth": 1.02663, "loss": 34.87293, "grad_norm": 53.40171, "time": 1.5347} -{"mode": "train", "epoch": 1, "iter": 43, "lr": 0.00012, "memory": 49163, "data_time": 0.07669, "loss_cls_0": 1.04294, "loss_box_0": 2.00007, "loss_cns_0": 0.61263, "loss_yns_0": 0.16658, "loss_cls_1": 1.10921, "loss_box_1": 2.2659, "loss_cns_1": 0.60202, "loss_yns_1": 0.16679, "loss_cls_2": 1.12766, "loss_box_2": 2.27753, "loss_cns_2": 0.60586, "loss_yns_2": 0.16661, "loss_cls_3": 1.14246, "loss_box_3": 2.3051, "loss_cns_3": 0.61062, "loss_yns_3": 0.16777, "loss_cls_4": 1.1402, "loss_box_4": 2.3046, "loss_cns_4": 0.6046, "loss_yns_4": 0.16786, "loss_cls_5": 1.16231, "loss_box_5": 2.33989, "loss_cns_5": 0.60574, "loss_yns_5": 0.16801, "loss_cls_dn_0": 0.40967, "loss_box_dn_0": 0.98986, "loss_cls_dn_1": 0.30621, "loss_box_dn_1": 1.06933, "loss_cls_dn_2": 0.34101, "loss_box_dn_2": 1.09, "loss_cls_dn_3": 0.38277, "loss_box_dn_3": 1.13593, "loss_cls_dn_4": 0.3923, "loss_box_dn_4": 1.14484, "loss_cls_dn_5": 0.42861, "loss_box_dn_5": 1.18612, "loss_dense_depth": 1.05546, "loss": 34.79507, "grad_norm": 57.30717, "time": 1.52897} -{"mode": "train", "epoch": 1, "iter": 44, "lr": 0.00012, "memory": 49163, "data_time": 0.0787, "loss_cls_0": 1.05301, "loss_box_0": 2.02411, "loss_cns_0": 0.60945, "loss_yns_0": 0.16991, "loss_cls_1": 1.09984, "loss_box_1": 2.27976, "loss_cns_1": 0.60218, "loss_yns_1": 0.16661, "loss_cls_2": 1.12319, "loss_box_2": 2.26667, "loss_cns_2": 0.60201, "loss_yns_2": 0.16566, "loss_cls_3": 1.13952, "loss_box_3": 2.28961, "loss_cns_3": 0.60247, "loss_yns_3": 0.16501, "loss_cls_4": 1.13752, "loss_box_4": 2.27878, "loss_cns_4": 0.5991, "loss_yns_4": 0.16404, "loss_cls_5": 1.15322, "loss_box_5": 2.32474, "loss_cns_5": 0.60103, "loss_yns_5": 0.17512, "loss_cls_dn_0": 0.41059, "loss_box_dn_0": 0.98799, "loss_cls_dn_1": 0.29739, "loss_box_dn_1": 1.12072, "loss_cls_dn_2": 0.33345, "loss_box_dn_2": 1.13004, "loss_cls_dn_3": 0.38398, "loss_box_dn_3": 1.16317, "loss_cls_dn_4": 0.40373, "loss_box_dn_4": 1.17409, "loss_cls_dn_5": 0.42027, "loss_box_dn_5": 1.24586, "loss_dense_depth": 1.06079, "loss": 34.92463, "grad_norm": 45.45889, "time": 1.50692} -{"mode": "train", "epoch": 1, "iter": 45, "lr": 0.00012, "memory": 49163, "data_time": 0.08269, "loss_cls_0": 1.05631, "loss_box_0": 2.03401, "loss_cns_0": 0.60874, "loss_yns_0": 0.16821, "loss_cls_1": 1.1226, "loss_box_1": 2.32201, "loss_cns_1": 0.5958, "loss_yns_1": 0.17003, "loss_cls_2": 1.17713, "loss_box_2": 2.2838, "loss_cns_2": 0.60319, "loss_yns_2": 0.16823, "loss_cls_3": 1.15646, "loss_box_3": 2.27495, "loss_cns_3": 0.60574, "loss_yns_3": 0.16777, "loss_cls_4": 1.1268, "loss_box_4": 2.29471, "loss_cns_4": 0.60734, "loss_yns_4": 0.16843, "loss_cls_5": 1.15274, "loss_box_5": 2.32138, "loss_cns_5": 0.60972, "loss_yns_5": 0.17369, "loss_cls_dn_0": 0.41031, "loss_box_dn_0": 0.98249, "loss_cls_dn_1": 0.2844, "loss_box_dn_1": 1.10538, "loss_cls_dn_2": 0.32254, "loss_box_dn_2": 1.10022, "loss_cls_dn_3": 0.37108, "loss_box_dn_3": 1.11113, "loss_cls_dn_4": 0.39888, "loss_box_dn_4": 1.14277, "loss_cls_dn_5": 0.40461, "loss_box_dn_5": 1.20903, "loss_dense_depth": 1.06054, "loss": 34.87318, "grad_norm": 44.95927, "time": 1.52032} -{"mode": "train", "epoch": 1, "iter": 46, "lr": 0.00012, "memory": 49163, "data_time": 0.08956, "loss_cls_0": 1.03487, "loss_box_0": 2.01255, "loss_cns_0": 0.60812, "loss_yns_0": 0.16752, "loss_cls_1": 1.12499, "loss_box_1": 2.40141, "loss_cns_1": 0.59409, "loss_yns_1": 0.16705, "loss_cls_2": 1.16388, "loss_box_2": 2.39666, "loss_cns_2": 0.60993, "loss_yns_2": 0.16468, "loss_cls_3": 1.19094, "loss_box_3": 2.37494, "loss_cns_3": 0.61098, "loss_yns_3": 0.16509, "loss_cls_4": 1.11973, "loss_box_4": 2.41636, "loss_cns_4": 0.61274, "loss_yns_4": 0.17264, "loss_cls_5": 1.14036, "loss_box_5": 2.42068, "loss_cns_5": 0.61906, "loss_yns_5": 0.17655, "loss_cls_dn_0": 0.41748, "loss_box_dn_0": 0.95787, "loss_cls_dn_1": 0.27327, "loss_box_dn_1": 1.06223, "loss_cls_dn_2": 0.31109, "loss_box_dn_2": 1.05202, "loss_cls_dn_3": 0.34142, "loss_box_dn_3": 1.03716, "loss_cls_dn_4": 0.37138, "loss_box_dn_4": 1.07929, "loss_cls_dn_5": 0.38178, "loss_box_dn_5": 1.11592, "loss_dense_depth": 1.00806, "loss": 34.87479, "grad_norm": 52.27202, "time": 1.52314} -{"mode": "train", "epoch": 1, "iter": 47, "lr": 0.00012, "memory": 49163, "data_time": 0.10707, "loss_cls_0": 1.05163, "loss_box_0": 2.01382, "loss_cns_0": 0.60916, "loss_yns_0": 0.16471, "loss_cls_1": 1.11242, "loss_box_1": 2.41039, "loss_cns_1": 0.59806, "loss_yns_1": 0.16659, "loss_cls_2": 1.13962, "loss_box_2": 2.44228, "loss_cns_2": 0.60991, "loss_yns_2": 0.16649, "loss_cls_3": 1.21577, "loss_box_3": 2.39001, "loss_cns_3": 0.61308, "loss_yns_3": 0.16636, "loss_cls_4": 1.19586, "loss_box_4": 2.42491, "loss_cns_4": 0.61228, "loss_yns_4": 0.17031, "loss_cls_5": 1.17961, "loss_box_5": 2.43152, "loss_cns_5": 0.61508, "loss_yns_5": 0.18113, "loss_cls_dn_0": 0.42228, "loss_box_dn_0": 0.9556, "loss_cls_dn_1": 0.28095, "loss_box_dn_1": 1.03672, "loss_cls_dn_2": 0.33744, "loss_box_dn_2": 1.02594, "loss_cls_dn_3": 0.34582, "loss_box_dn_3": 0.995, "loss_cls_dn_4": 0.34882, "loss_box_dn_4": 1.03413, "loss_cls_dn_5": 0.37138, "loss_box_dn_5": 1.05828, "loss_dense_depth": 1.00059, "loss": 34.89394, "grad_norm": 43.02238, "time": 1.49834} -{"mode": "train", "epoch": 1, "iter": 48, "lr": 0.00012, "memory": 49163, "data_time": 0.07824, "loss_cls_0": 1.02746, "loss_box_0": 2.00712, "loss_cns_0": 0.60972, "loss_yns_0": 0.16178, "loss_cls_1": 1.09677, "loss_box_1": 2.34147, "loss_cns_1": 0.58983, "loss_yns_1": 0.16291, "loss_cls_2": 1.15613, "loss_box_2": 2.34721, "loss_cns_2": 0.61172, "loss_yns_2": 0.16561, "loss_cls_3": 1.14341, "loss_box_3": 2.34081, "loss_cns_3": 0.61367, "loss_yns_3": 0.16504, "loss_cls_4": 1.15821, "loss_box_4": 2.40193, "loss_cns_4": 0.60761, "loss_yns_4": 0.16907, "loss_cls_5": 1.18329, "loss_box_5": 2.42259, "loss_cns_5": 0.6069, "loss_yns_5": 0.17509, "loss_cls_dn_0": 0.41309, "loss_box_dn_0": 0.94746, "loss_cls_dn_1": 0.26995, "loss_box_dn_1": 1.10229, "loss_cls_dn_2": 0.33392, "loss_box_dn_2": 1.06002, "loss_cls_dn_3": 0.36356, "loss_box_dn_3": 1.0468, "loss_cls_dn_4": 0.33366, "loss_box_dn_4": 1.0911, "loss_cls_dn_5": 0.35102, "loss_box_dn_5": 1.11514, "loss_dense_depth": 1.00369, "loss": 34.69706, "grad_norm": 51.39897, "time": 1.4878} -{"mode": "train", "epoch": 1, "iter": 49, "lr": 0.00012, "memory": 49163, "data_time": 0.09426, "loss_cls_0": 1.05122, "loss_box_0": 1.98915, "loss_cns_0": 0.60891, "loss_yns_0": 0.16317, "loss_cls_1": 1.0671, "loss_box_1": 2.33621, "loss_cns_1": 0.59257, "loss_yns_1": 0.16669, "loss_cls_2": 1.13266, "loss_box_2": 2.30042, "loss_cns_2": 0.61329, "loss_yns_2": 0.16756, "loss_cls_3": 1.16888, "loss_box_3": 2.29812, "loss_cns_3": 0.61535, "loss_yns_3": 0.16599, "loss_cls_4": 1.12448, "loss_box_4": 2.32065, "loss_cns_4": 0.61304, "loss_yns_4": 0.17052, "loss_cls_5": 1.15869, "loss_box_5": 2.35762, "loss_cns_5": 0.61298, "loss_yns_5": 0.17096, "loss_cls_dn_0": 0.39588, "loss_box_dn_0": 0.94043, "loss_cls_dn_1": 0.252, "loss_box_dn_1": 1.16062, "loss_cls_dn_2": 0.3117, "loss_box_dn_2": 1.10833, "loss_cls_dn_3": 0.37636, "loss_box_dn_3": 1.11066, "loss_cls_dn_4": 0.33229, "loss_box_dn_4": 1.13291, "loss_cls_dn_5": 0.33813, "loss_box_dn_5": 1.16333, "loss_dense_depth": 1.01693, "loss": 34.6058, "grad_norm": 41.79116, "time": 1.5211} -{"mode": "train", "epoch": 1, "iter": 50, "lr": 0.00012, "memory": 49163, "data_time": 0.07928, "loss_cls_0": 1.04245, "loss_box_0": 1.98351, "loss_cns_0": 0.61146, "loss_yns_0": 0.16433, "loss_cls_1": 1.1097, "loss_box_1": 2.37348, "loss_cns_1": 0.59904, "loss_yns_1": 0.16682, "loss_cls_2": 1.11014, "loss_box_2": 2.33442, "loss_cns_2": 0.61763, "loss_yns_2": 0.16608, "loss_cls_3": 1.17111, "loss_box_3": 2.33726, "loss_cns_3": 0.61735, "loss_yns_3": 0.16765, "loss_cls_4": 1.11851, "loss_box_4": 2.34816, "loss_cns_4": 0.61801, "loss_yns_4": 0.17647, "loss_cls_5": 1.14893, "loss_box_5": 2.39003, "loss_cns_5": 0.61022, "loss_yns_5": 0.17088, "loss_cls_dn_0": 0.37778, "loss_box_dn_0": 0.95163, "loss_cls_dn_1": 0.24593, "loss_box_dn_1": 1.12104, "loss_cls_dn_2": 0.28913, "loss_box_dn_2": 1.08328, "loss_cls_dn_3": 0.37535, "loss_box_dn_3": 1.10472, "loss_cls_dn_4": 0.33822, "loss_box_dn_4": 1.10786, "loss_cls_dn_5": 0.34272, "loss_box_dn_5": 1.14904, "loss_dense_depth": 0.9969, "loss": 34.63723, "grad_norm": 48.57431, "time": 1.49177} -{"mode": "train", "epoch": 1, "iter": 51, "lr": 0.00012, "memory": 49163, "data_time": 0.0834, "loss_cls_0": 1.02146, "loss_box_0": 1.95849, "loss_cns_0": 0.6096, "loss_yns_0": 0.16313, "loss_cls_1": 1.04834, "loss_box_1": 2.34779, "loss_cns_1": 0.59696, "loss_yns_1": 0.16405, "loss_cls_2": 1.09061, "loss_box_2": 2.38322, "loss_cns_2": 0.61684, "loss_yns_2": 0.16472, "loss_cls_3": 1.15019, "loss_box_3": 2.36153, "loss_cns_3": 0.6215, "loss_yns_3": 0.16625, "loss_cls_4": 1.10726, "loss_box_4": 2.36695, "loss_cns_4": 0.62086, "loss_yns_4": 0.17294, "loss_cls_5": 1.14465, "loss_box_5": 2.40671, "loss_cns_5": 0.61898, "loss_yns_5": 0.16685, "loss_cls_dn_0": 0.38391, "loss_box_dn_0": 0.93877, "loss_cls_dn_1": 0.24067, "loss_box_dn_1": 1.16066, "loss_cls_dn_2": 0.27839, "loss_box_dn_2": 1.14717, "loss_cls_dn_3": 0.35163, "loss_box_dn_3": 1.15744, "loss_cls_dn_4": 0.33522, "loss_box_dn_4": 1.15333, "loss_cls_dn_5": 0.33532, "loss_box_dn_5": 1.19025, "loss_dense_depth": 0.97678, "loss": 34.7194, "grad_norm": 44.13463, "time": 1.51302} -{"mode": "train", "epoch": 1, "iter": 52, "lr": 0.00012, "memory": 49163, "data_time": 0.07793, "loss_cls_0": 1.03515, "loss_box_0": 1.98063, "loss_cns_0": 0.60086, "loss_yns_0": 0.16348, "loss_cls_1": 1.07234, "loss_box_1": 2.29161, "loss_cns_1": 0.5813, "loss_yns_1": 0.1625, "loss_cls_2": 1.10725, "loss_box_2": 2.36687, "loss_cns_2": 0.60966, "loss_yns_2": 0.16538, "loss_cls_3": 1.1264, "loss_box_3": 2.33665, "loss_cns_3": 0.62114, "loss_yns_3": 0.16689, "loss_cls_4": 1.10998, "loss_box_4": 2.34406, "loss_cns_4": 0.62121, "loss_yns_4": 0.16655, "loss_cls_5": 1.14465, "loss_box_5": 2.37507, "loss_cns_5": 0.62491, "loss_yns_5": 0.16674, "loss_cls_dn_0": 0.39687, "loss_box_dn_0": 0.93093, "loss_cls_dn_1": 0.24489, "loss_box_dn_1": 1.13559, "loss_cls_dn_2": 0.27771, "loss_box_dn_2": 1.13554, "loss_cls_dn_3": 0.33301, "loss_box_dn_3": 1.13832, "loss_cls_dn_4": 0.33024, "loss_box_dn_4": 1.14192, "loss_cls_dn_5": 0.34131, "loss_box_dn_5": 1.17683, "loss_dense_depth": 0.97123, "loss": 34.49567, "grad_norm": 47.06027, "time": 1.49209} -{"mode": "train", "epoch": 1, "iter": 53, "lr": 0.00012, "memory": 49163, "data_time": 0.07371, "loss_cls_0": 1.04767, "loss_box_0": 1.97534, "loss_cns_0": 0.6011, "loss_yns_0": 0.16613, "loss_cls_1": 1.0789, "loss_box_1": 2.40494, "loss_cns_1": 0.57288, "loss_yns_1": 0.16062, "loss_cls_2": 1.12842, "loss_box_2": 2.45961, "loss_cns_2": 0.60082, "loss_yns_2": 0.16578, "loss_cls_3": 1.16944, "loss_box_3": 2.43921, "loss_cns_3": 0.61062, "loss_yns_3": 0.16462, "loss_cls_4": 1.11056, "loss_box_4": 2.44623, "loss_cns_4": 0.6098, "loss_yns_4": 0.16309, "loss_cls_5": 1.14008, "loss_box_5": 2.4651, "loss_cns_5": 0.61242, "loss_yns_5": 0.16572, "loss_cls_dn_0": 0.39953, "loss_box_dn_0": 0.91404, "loss_cls_dn_1": 0.24088, "loss_box_dn_1": 1.0838, "loss_cls_dn_2": 0.27309, "loss_box_dn_2": 1.06765, "loss_cls_dn_3": 0.3105, "loss_box_dn_3": 1.07231, "loss_cls_dn_4": 0.31739, "loss_box_dn_4": 1.07517, "loss_cls_dn_5": 0.35061, "loss_box_dn_5": 1.10236, "loss_dense_depth": 1.05164, "loss": 34.71806, "grad_norm": 40.48055, "time": 1.49642} -{"mode": "train", "epoch": 1, "iter": 54, "lr": 0.00012, "memory": 49163, "data_time": 0.07768, "loss_cls_0": 0.99933, "loss_box_0": 1.94688, "loss_cns_0": 0.61251, "loss_yns_0": 0.16295, "loss_cls_1": 1.03674, "loss_box_1": 2.3117, "loss_cns_1": 0.60809, "loss_yns_1": 0.16198, "loss_cls_2": 1.06057, "loss_box_2": 2.2914, "loss_cns_2": 0.62335, "loss_yns_2": 0.16032, "loss_cls_3": 1.15204, "loss_box_3": 2.29826, "loss_cns_3": 0.62728, "loss_yns_3": 0.15959, "loss_cls_4": 1.08937, "loss_box_4": 2.2943, "loss_cns_4": 0.62648, "loss_yns_4": 0.16142, "loss_cls_5": 1.11022, "loss_box_5": 2.30631, "loss_cns_5": 0.62774, "loss_yns_5": 0.16505, "loss_cls_dn_0": 0.37702, "loss_box_dn_0": 0.90804, "loss_cls_dn_1": 0.22957, "loss_box_dn_1": 1.07904, "loss_cls_dn_2": 0.26957, "loss_box_dn_2": 1.05862, "loss_cls_dn_3": 0.29776, "loss_box_dn_3": 1.06272, "loss_cls_dn_4": 0.31163, "loss_box_dn_4": 1.05754, "loss_cls_dn_5": 0.34943, "loss_box_dn_5": 1.07618, "loss_dense_depth": 0.98328, "loss": 33.65429, "grad_norm": 51.14889, "time": 1.50214} -{"mode": "train", "epoch": 1, "iter": 55, "lr": 0.00012, "memory": 49163, "data_time": 0.08799, "loss_cls_0": 1.02301, "loss_box_0": 1.98386, "loss_cns_0": 0.61011, "loss_yns_0": 0.16343, "loss_cls_1": 1.06833, "loss_box_1": 2.23501, "loss_cns_1": 0.6145, "loss_yns_1": 0.16549, "loss_cls_2": 1.0829, "loss_box_2": 2.21077, "loss_cns_2": 0.62627, "loss_yns_2": 0.1646, "loss_cls_3": 1.1106, "loss_box_3": 2.19948, "loss_cns_3": 0.63144, "loss_yns_3": 0.16554, "loss_cls_4": 1.08191, "loss_box_4": 2.2135, "loss_cns_4": 0.62857, "loss_yns_4": 0.17497, "loss_cls_5": 1.10465, "loss_box_5": 2.2081, "loss_cns_5": 0.62896, "loss_yns_5": 0.16545, "loss_cls_dn_0": 0.37779, "loss_box_dn_0": 0.89881, "loss_cls_dn_1": 0.23724, "loss_box_dn_1": 0.91806, "loss_cls_dn_2": 0.27998, "loss_box_dn_2": 0.90587, "loss_cls_dn_3": 0.31792, "loss_box_dn_3": 0.91059, "loss_cls_dn_4": 0.33871, "loss_box_dn_4": 0.91554, "loss_cls_dn_5": 0.38737, "loss_box_dn_5": 0.93095, "loss_dense_depth": 0.98572, "loss": 32.666, "grad_norm": 58.26726, "time": 1.49581} -{"mode": "train", "epoch": 1, "iter": 56, "lr": 0.00012, "memory": 49163, "data_time": 0.07831, "loss_cls_0": 1.01623, "loss_box_0": 1.94365, "loss_cns_0": 0.61096, "loss_yns_0": 0.16722, "loss_cls_1": 1.0745, "loss_box_1": 2.16729, "loss_cns_1": 0.61334, "loss_yns_1": 0.16534, "loss_cls_2": 1.09764, "loss_box_2": 2.1503, "loss_cns_2": 0.61929, "loss_yns_2": 0.16604, "loss_cls_3": 1.11476, "loss_box_3": 2.13294, "loss_cns_3": 0.62563, "loss_yns_3": 0.16469, "loss_cls_4": 1.07427, "loss_box_4": 2.17516, "loss_cns_4": 0.62301, "loss_yns_4": 0.18783, "loss_cls_5": 1.09738, "loss_box_5": 2.17993, "loss_cns_5": 0.62292, "loss_yns_5": 0.16482, "loss_cls_dn_0": 0.36576, "loss_box_dn_0": 0.89658, "loss_cls_dn_1": 0.23433, "loss_box_dn_1": 0.9149, "loss_cls_dn_2": 0.27668, "loss_box_dn_2": 0.89925, "loss_cls_dn_3": 0.31419, "loss_box_dn_3": 0.89562, "loss_cls_dn_4": 0.32881, "loss_box_dn_4": 0.92041, "loss_cls_dn_5": 0.38634, "loss_box_dn_5": 0.93886, "loss_dense_depth": 0.97215, "loss": 32.29902, "grad_norm": 43.41092, "time": 1.50995} -{"mode": "train", "epoch": 1, "iter": 57, "lr": 0.00012, "memory": 49163, "data_time": 0.08223, "loss_cls_0": 1.00991, "loss_box_0": 1.91802, "loss_cns_0": 0.61412, "loss_yns_0": 0.16688, "loss_cls_1": 1.05066, "loss_box_1": 2.16524, "loss_cns_1": 0.61841, "loss_yns_1": 0.16576, "loss_cls_2": 1.07392, "loss_box_2": 2.13254, "loss_cns_2": 0.62151, "loss_yns_2": 0.16638, "loss_cls_3": 1.10601, "loss_box_3": 2.11289, "loss_cns_3": 0.62514, "loss_yns_3": 0.16531, "loss_cls_4": 1.1075, "loss_box_4": 2.15049, "loss_cns_4": 0.6239, "loss_yns_4": 0.18094, "loss_cls_5": 1.08812, "loss_box_5": 2.17348, "loss_cns_5": 0.6236, "loss_yns_5": 0.1659, "loss_cls_dn_0": 0.34557, "loss_box_dn_0": 0.89171, "loss_cls_dn_1": 0.23135, "loss_box_dn_1": 0.9103, "loss_cls_dn_2": 0.25679, "loss_box_dn_2": 0.88683, "loss_cls_dn_3": 0.30028, "loss_box_dn_3": 0.88972, "loss_cls_dn_4": 0.29841, "loss_box_dn_4": 0.92035, "loss_cls_dn_5": 0.36908, "loss_box_dn_5": 0.94599, "loss_dense_depth": 0.97679, "loss": 32.04979, "grad_norm": 46.65861, "time": 1.50059} -{"mode": "train", "epoch": 1, "iter": 58, "lr": 0.00012, "memory": 49163, "data_time": 0.07899, "loss_cls_0": 0.99313, "loss_box_0": 1.91573, "loss_cns_0": 0.61332, "loss_yns_0": 0.16396, "loss_cls_1": 1.03571, "loss_box_1": 2.15175, "loss_cns_1": 0.61261, "loss_yns_1": 0.16549, "loss_cls_2": 1.07303, "loss_box_2": 2.08304, "loss_cns_2": 0.62918, "loss_yns_2": 0.16708, "loss_cls_3": 1.09084, "loss_box_3": 2.05275, "loss_cns_3": 0.62996, "loss_yns_3": 0.16661, "loss_cls_4": 1.09535, "loss_box_4": 2.06317, "loss_cns_4": 0.63032, "loss_yns_4": 0.1668, "loss_cls_5": 1.07842, "loss_box_5": 2.07095, "loss_cns_5": 0.63347, "loss_yns_5": 0.16667, "loss_cls_dn_0": 0.33785, "loss_box_dn_0": 0.8814, "loss_cls_dn_1": 0.23038, "loss_box_dn_1": 0.92886, "loss_cls_dn_2": 0.24444, "loss_box_dn_2": 0.89581, "loss_cls_dn_3": 0.27974, "loss_box_dn_3": 0.90078, "loss_cls_dn_4": 0.28368, "loss_box_dn_4": 0.92151, "loss_cls_dn_5": 0.34601, "loss_box_dn_5": 0.94279, "loss_dense_depth": 0.97293, "loss": 31.61549, "grad_norm": 50.53089, "time": 1.49545} -{"mode": "train", "epoch": 1, "iter": 59, "lr": 0.00012, "memory": 49163, "data_time": 0.0831, "loss_cls_0": 0.95964, "loss_box_0": 1.89741, "loss_cns_0": 0.61215, "loss_yns_0": 0.16503, "loss_cls_1": 1.04827, "loss_box_1": 2.10416, "loss_cns_1": 0.6204, "loss_yns_1": 0.16433, "loss_cls_2": 1.07473, "loss_box_2": 2.06915, "loss_cns_2": 0.63266, "loss_yns_2": 0.16709, "loss_cls_3": 1.08717, "loss_box_3": 2.06079, "loss_cns_3": 0.63083, "loss_yns_3": 0.16672, "loss_cls_4": 1.06209, "loss_box_4": 2.09315, "loss_cns_4": 0.62608, "loss_yns_4": 0.17042, "loss_cls_5": 1.08728, "loss_box_5": 2.07982, "loss_cns_5": 0.63517, "loss_yns_5": 0.16964, "loss_cls_dn_0": 0.35922, "loss_box_dn_0": 0.87841, "loss_cls_dn_1": 0.23441, "loss_box_dn_1": 0.92664, "loss_cls_dn_2": 0.2482, "loss_box_dn_2": 0.89722, "loss_cls_dn_3": 0.26757, "loss_box_dn_3": 0.89842, "loss_cls_dn_4": 0.28994, "loss_box_dn_4": 0.92028, "loss_cls_dn_5": 0.32308, "loss_box_dn_5": 0.9346, "loss_dense_depth": 0.90557, "loss": 31.46774, "grad_norm": 52.35609, "time": 1.49943} -{"mode": "train", "epoch": 1, "iter": 60, "lr": 0.00012, "memory": 49163, "data_time": 0.08057, "loss_cls_0": 0.97736, "loss_box_0": 1.85672, "loss_cns_0": 0.60541, "loss_yns_0": 0.16166, "loss_cls_1": 1.05683, "loss_box_1": 2.01564, "loss_cns_1": 0.62437, "loss_yns_1": 0.16321, "loss_cls_2": 1.07724, "loss_box_2": 1.99005, "loss_cns_2": 0.63709, "loss_yns_2": 0.16589, "loss_cls_3": 1.09191, "loss_box_3": 1.98695, "loss_cns_3": 0.63706, "loss_yns_3": 0.16424, "loss_cls_4": 1.10878, "loss_box_4": 2.00213, "loss_cns_4": 0.6333, "loss_yns_4": 0.1812, "loss_cls_5": 1.11923, "loss_box_5": 2.00422, "loss_cns_5": 0.63972, "loss_yns_5": 0.16661, "loss_cls_dn_0": 0.37446, "loss_box_dn_0": 0.89617, "loss_cls_dn_1": 0.22934, "loss_box_dn_1": 0.90986, "loss_cls_dn_2": 0.24418, "loss_box_dn_2": 0.88174, "loss_cls_dn_3": 0.26103, "loss_box_dn_3": 0.87383, "loss_cls_dn_4": 0.29569, "loss_box_dn_4": 0.89399, "loss_cls_dn_5": 0.30439, "loss_box_dn_5": 0.90988, "loss_dense_depth": 0.93986, "loss": 31.08125, "grad_norm": 34.71469, "time": 1.49801} -{"mode": "train", "epoch": 1, "iter": 61, "lr": 0.00012, "memory": 49163, "data_time": 0.10265, "loss_cls_0": 0.99592, "loss_box_0": 1.88119, "loss_cns_0": 0.60488, "loss_yns_0": 0.15991, "loss_cls_1": 1.03831, "loss_box_1": 2.02547, "loss_cns_1": 0.62536, "loss_yns_1": 0.1624, "loss_cls_2": 1.07796, "loss_box_2": 1.98615, "loss_cns_2": 0.64016, "loss_yns_2": 0.1633, "loss_cls_3": 1.07975, "loss_box_3": 1.99761, "loss_cns_3": 0.64362, "loss_yns_3": 0.16264, "loss_cls_4": 1.06451, "loss_box_4": 2.00594, "loss_cns_4": 0.63919, "loss_yns_4": 0.18251, "loss_cls_5": 1.13685, "loss_box_5": 2.0248, "loss_cns_5": 0.64074, "loss_yns_5": 0.16438, "loss_cls_dn_0": 0.33632, "loss_box_dn_0": 0.89894, "loss_cls_dn_1": 0.22205, "loss_box_dn_1": 0.92336, "loss_cls_dn_2": 0.23872, "loss_box_dn_2": 0.89, "loss_cls_dn_3": 0.2575, "loss_box_dn_3": 0.88374, "loss_cls_dn_4": 0.26767, "loss_box_dn_4": 0.90322, "loss_cls_dn_5": 0.28881, "loss_box_dn_5": 0.9204, "loss_dense_depth": 0.99354, "loss": 31.12777, "grad_norm": 53.24954, "time": 1.53358} -{"mode": "train", "epoch": 1, "iter": 62, "lr": 0.00012, "memory": 49163, "data_time": 0.14962, "loss_cls_0": 1.01677, "loss_box_0": 1.87236, "loss_cns_0": 0.61106, "loss_yns_0": 0.16001, "loss_cls_1": 1.02908, "loss_box_1": 2.12259, "loss_cns_1": 0.62357, "loss_yns_1": 0.1613, "loss_cls_2": 1.08776, "loss_box_2": 2.06289, "loss_cns_2": 0.64078, "loss_yns_2": 0.16279, "loss_cls_3": 1.08617, "loss_box_3": 2.07837, "loss_cns_3": 0.64374, "loss_yns_3": 0.16157, "loss_cls_4": 1.09113, "loss_box_4": 2.07446, "loss_cns_4": 0.64369, "loss_yns_4": 0.1702, "loss_cls_5": 1.10192, "loss_box_5": 2.09244, "loss_cns_5": 0.6438, "loss_yns_5": 0.16547, "loss_cls_dn_0": 0.31697, "loss_box_dn_0": 0.88974, "loss_cls_dn_1": 0.21929, "loss_box_dn_1": 0.8913, "loss_cls_dn_2": 0.2379, "loss_box_dn_2": 0.85939, "loss_cls_dn_3": 0.25846, "loss_box_dn_3": 0.85684, "loss_cls_dn_4": 0.24264, "loss_box_dn_4": 0.86871, "loss_cls_dn_5": 0.29336, "loss_box_dn_5": 0.88277, "loss_dense_depth": 0.95191, "loss": 31.27317, "grad_norm": 45.02608, "time": 1.57547} -{"mode": "train", "epoch": 1, "iter": 63, "lr": 0.00012, "memory": 49163, "data_time": 0.07493, "loss_cls_0": 0.99908, "loss_box_0": 1.86485, "loss_cns_0": 0.61295, "loss_yns_0": 0.16226, "loss_cls_1": 1.03617, "loss_box_1": 2.12644, "loss_cns_1": 0.62055, "loss_yns_1": 0.16338, "loss_cls_2": 1.08513, "loss_box_2": 2.06001, "loss_cns_2": 0.6396, "loss_yns_2": 0.16614, "loss_cls_3": 1.08196, "loss_box_3": 2.05759, "loss_cns_3": 0.64016, "loss_yns_3": 0.16334, "loss_cls_4": 1.11832, "loss_box_4": 2.06931, "loss_cns_4": 0.64097, "loss_yns_4": 0.16568, "loss_cls_5": 1.09275, "loss_box_5": 2.04899, "loss_cns_5": 0.64436, "loss_yns_5": 0.16116, "loss_cls_dn_0": 0.33234, "loss_box_dn_0": 0.88026, "loss_cls_dn_1": 0.21426, "loss_box_dn_1": 0.87676, "loss_cls_dn_2": 0.2335, "loss_box_dn_2": 0.84441, "loss_cls_dn_3": 0.24454, "loss_box_dn_3": 0.85685, "loss_cls_dn_4": 0.22964, "loss_box_dn_4": 0.86454, "loss_cls_dn_5": 0.29199, "loss_box_dn_5": 0.87409, "loss_dense_depth": 1.03982, "loss": 31.20413, "grad_norm": 51.92075, "time": 1.52281} -{"mode": "train", "epoch": 1, "iter": 64, "lr": 0.00013, "memory": 49163, "data_time": 0.07391, "loss_cls_0": 0.9751, "loss_box_0": 1.84311, "loss_cns_0": 0.61695, "loss_yns_0": 0.16167, "loss_cls_1": 1.04114, "loss_box_1": 2.12229, "loss_cns_1": 0.61942, "loss_yns_1": 0.16524, "loss_cls_2": 1.06586, "loss_box_2": 2.06038, "loss_cns_2": 0.64035, "loss_yns_2": 0.16556, "loss_cls_3": 1.08838, "loss_box_3": 2.05729, "loss_cns_3": 0.6406, "loss_yns_3": 0.16548, "loss_cls_4": 1.10269, "loss_box_4": 2.06757, "loss_cns_4": 0.64006, "loss_yns_4": 0.17185, "loss_cls_5": 1.10895, "loss_box_5": 2.05526, "loss_cns_5": 0.64301, "loss_yns_5": 0.16337, "loss_cls_dn_0": 0.32847, "loss_box_dn_0": 0.86864, "loss_cls_dn_1": 0.21236, "loss_box_dn_1": 0.89536, "loss_cls_dn_2": 0.23032, "loss_box_dn_2": 0.87448, "loss_cls_dn_3": 0.23389, "loss_box_dn_3": 0.90788, "loss_cls_dn_4": 0.22902, "loss_box_dn_4": 0.92498, "loss_cls_dn_5": 0.2925, "loss_box_dn_5": 0.94708, "loss_dense_depth": 0.90921, "loss": 31.23577, "grad_norm": 54.20214, "time": 1.49889} -{"mode": "train", "epoch": 1, "iter": 65, "lr": 0.00013, "memory": 49163, "data_time": 0.07915, "loss_cls_0": 0.96412, "loss_box_0": 1.85815, "loss_cns_0": 0.61841, "loss_yns_0": 0.16166, "loss_cls_1": 1.00106, "loss_box_1": 2.09332, "loss_cns_1": 0.61518, "loss_yns_1": 0.16236, "loss_cls_2": 1.04974, "loss_box_2": 2.04876, "loss_cns_2": 0.64232, "loss_yns_2": 0.16266, "loss_cls_3": 1.07849, "loss_box_3": 2.05235, "loss_cns_3": 0.64357, "loss_yns_3": 0.16524, "loss_cls_4": 1.07629, "loss_box_4": 2.03961, "loss_cns_4": 0.63916, "loss_yns_4": 0.17608, "loss_cls_5": 1.06669, "loss_box_5": 2.04454, "loss_cns_5": 0.6418, "loss_yns_5": 0.16234, "loss_cls_dn_0": 0.3168, "loss_box_dn_0": 0.87396, "loss_cls_dn_1": 0.21108, "loss_box_dn_1": 0.98957, "loss_cls_dn_2": 0.22774, "loss_box_dn_2": 0.98273, "loss_cls_dn_3": 0.2287, "loss_box_dn_3": 1.01767, "loss_cls_dn_4": 0.22024, "loss_box_dn_4": 1.0369, "loss_cls_dn_5": 0.26125, "loss_box_dn_5": 1.05717, "loss_dense_depth": 0.95663, "loss": 31.54433, "grad_norm": 53.27539, "time": 1.49549} -{"mode": "train", "epoch": 1, "iter": 66, "lr": 0.00013, "memory": 49163, "data_time": 0.07711, "loss_cls_0": 0.97936, "loss_box_0": 1.84469, "loss_cns_0": 0.61864, "loss_yns_0": 0.16268, "loss_cls_1": 1.00337, "loss_box_1": 2.09355, "loss_cns_1": 0.6079, "loss_yns_1": 0.16339, "loss_cls_2": 1.08062, "loss_box_2": 2.05845, "loss_cns_2": 0.63842, "loss_yns_2": 0.16484, "loss_cls_3": 1.0998, "loss_box_3": 2.04107, "loss_cns_3": 0.64254, "loss_yns_3": 0.16522, "loss_cls_4": 1.13934, "loss_box_4": 2.04732, "loss_cns_4": 0.63883, "loss_yns_4": 0.17948, "loss_cls_5": 1.18478, "loss_box_5": 2.03951, "loss_cns_5": 0.64012, "loss_yns_5": 0.16376, "loss_cls_dn_0": 0.29938, "loss_box_dn_0": 0.86358, "loss_cls_dn_1": 0.19847, "loss_box_dn_1": 1.08453, "loss_cls_dn_2": 0.21817, "loss_box_dn_2": 1.07927, "loss_cls_dn_3": 0.22652, "loss_box_dn_3": 1.09804, "loss_cls_dn_4": 0.21974, "loss_box_dn_4": 1.11991, "loss_cls_dn_5": 0.23668, "loss_box_dn_5": 1.12797, "loss_dense_depth": 0.92304, "loss": 32.09299, "grad_norm": 93.47295, "time": 1.50813} -{"mode": "train", "epoch": 1, "iter": 67, "lr": 0.00013, "memory": 49163, "data_time": 0.10496, "loss_cls_0": 0.99161, "loss_box_0": 1.84878, "loss_cns_0": 0.61682, "loss_yns_0": 0.16305, "loss_cls_1": 1.05298, "loss_box_1": 2.08836, "loss_cns_1": 0.60632, "loss_yns_1": 0.16193, "loss_cls_2": 1.10245, "loss_box_2": 2.11279, "loss_cns_2": 0.63622, "loss_yns_2": 0.16154, "loss_cls_3": 1.11096, "loss_box_3": 2.07643, "loss_cns_3": 0.63929, "loss_yns_3": 0.16462, "loss_cls_4": 1.13297, "loss_box_4": 2.07744, "loss_cns_4": 0.63753, "loss_yns_4": 0.16412, "loss_cls_5": 1.10928, "loss_box_5": 2.07712, "loss_cns_5": 0.63665, "loss_yns_5": 0.16503, "loss_cls_dn_0": 0.30226, "loss_box_dn_0": 0.85762, "loss_cls_dn_1": 0.20003, "loss_box_dn_1": 1.01198, "loss_cls_dn_2": 0.22088, "loss_box_dn_2": 1.00686, "loss_cls_dn_3": 0.22663, "loss_box_dn_3": 1.00419, "loss_cls_dn_4": 0.22278, "loss_box_dn_4": 1.01737, "loss_cls_dn_5": 0.24751, "loss_box_dn_5": 1.01662, "loss_dense_depth": 0.96669, "loss": 31.8357, "grad_norm": 51.58087, "time": 1.49812} -{"mode": "train", "epoch": 1, "iter": 68, "lr": 0.00013, "memory": 49163, "data_time": 0.07078, "loss_cls_0": 0.9697, "loss_box_0": 1.81832, "loss_cns_0": 0.61918, "loss_yns_0": 0.16272, "loss_cls_1": 1.02203, "loss_box_1": 1.97103, "loss_cns_1": 0.61914, "loss_yns_1": 0.16157, "loss_cls_2": 1.07198, "loss_box_2": 1.96032, "loss_cns_2": 0.64047, "loss_yns_2": 0.1641, "loss_cls_3": 1.09442, "loss_box_3": 1.94859, "loss_cns_3": 0.64524, "loss_yns_3": 0.16307, "loss_cls_4": 1.12214, "loss_box_4": 1.94983, "loss_cns_4": 0.64389, "loss_yns_4": 0.16272, "loss_cls_5": 1.0807, "loss_box_5": 1.94477, "loss_cns_5": 0.64213, "loss_yns_5": 0.16273, "loss_cls_dn_0": 0.31319, "loss_box_dn_0": 0.8596, "loss_cls_dn_1": 0.20713, "loss_box_dn_1": 0.921, "loss_cls_dn_2": 0.22715, "loss_box_dn_2": 0.9031, "loss_cls_dn_3": 0.23358, "loss_box_dn_3": 0.90896, "loss_cls_dn_4": 0.23786, "loss_box_dn_4": 0.90799, "loss_cls_dn_5": 0.26481, "loss_box_dn_5": 0.90726, "loss_dense_depth": 0.92066, "loss": 30.55307, "grad_norm": 40.7411, "time": 1.47772} -{"mode": "train", "epoch": 1, "iter": 69, "lr": 0.00013, "memory": 49163, "data_time": 0.08026, "loss_cls_0": 0.93163, "loss_box_0": 1.78714, "loss_cns_0": 0.62095, "loss_yns_0": 0.15857, "loss_cls_1": 0.98491, "loss_box_1": 1.96648, "loss_cns_1": 0.6302, "loss_yns_1": 0.15988, "loss_cls_2": 1.03134, "loss_box_2": 1.90521, "loss_cns_2": 0.64585, "loss_yns_2": 0.16212, "loss_cls_3": 1.05155, "loss_box_3": 1.92059, "loss_cns_3": 0.65387, "loss_yns_3": 0.16401, "loss_cls_4": 1.05463, "loss_box_4": 1.9121, "loss_cns_4": 0.65116, "loss_yns_4": 0.1618, "loss_cls_5": 1.10357, "loss_box_5": 1.92117, "loss_cns_5": 0.64872, "loss_yns_5": 0.15988, "loss_cls_dn_0": 0.29551, "loss_box_dn_0": 0.84936, "loss_cls_dn_1": 0.20533, "loss_box_dn_1": 0.90567, "loss_cls_dn_2": 0.22272, "loss_box_dn_2": 0.88305, "loss_cls_dn_3": 0.22884, "loss_box_dn_3": 0.89634, "loss_cls_dn_4": 0.23127, "loss_box_dn_4": 0.88416, "loss_cls_dn_5": 0.25436, "loss_box_dn_5": 0.8941, "loss_dense_depth": 0.88515, "loss": 30.02321, "grad_norm": 57.89933, "time": 1.49457} -{"mode": "train", "epoch": 1, "iter": 70, "lr": 0.00013, "memory": 49163, "data_time": 0.07522, "loss_cls_0": 0.9471, "loss_box_0": 1.78138, "loss_cns_0": 0.61539, "loss_yns_0": 0.15908, "loss_cls_1": 0.99693, "loss_box_1": 1.94094, "loss_cns_1": 0.62959, "loss_yns_1": 0.16009, "loss_cls_2": 1.03077, "loss_box_2": 1.88787, "loss_cns_2": 0.64255, "loss_yns_2": 0.15991, "loss_cls_3": 1.05047, "loss_box_3": 1.90236, "loss_cns_3": 0.64777, "loss_yns_3": 0.16298, "loss_cls_4": 1.05996, "loss_box_4": 1.91104, "loss_cns_4": 0.64617, "loss_yns_4": 0.15958, "loss_cls_5": 1.05389, "loss_box_5": 1.93303, "loss_cns_5": 0.64555, "loss_yns_5": 0.16317, "loss_cls_dn_0": 0.28937, "loss_box_dn_0": 0.85053, "loss_cls_dn_1": 0.20949, "loss_box_dn_1": 0.81473, "loss_cls_dn_2": 0.22589, "loss_box_dn_2": 0.80153, "loss_cls_dn_3": 0.22678, "loss_box_dn_3": 0.82238, "loss_cls_dn_4": 0.22911, "loss_box_dn_4": 0.82597, "loss_cls_dn_5": 0.26577, "loss_box_dn_5": 0.85541, "loss_dense_depth": 0.92305, "loss": 29.62758, "grad_norm": 52.50658, "time": 1.49032} -{"mode": "train", "epoch": 1, "iter": 71, "lr": 0.00013, "memory": 49163, "data_time": 0.07454, "loss_cls_0": 0.94014, "loss_box_0": 1.79126, "loss_cns_0": 0.61506, "loss_yns_0": 0.15801, "loss_cls_1": 0.98659, "loss_box_1": 2.00406, "loss_cns_1": 0.62566, "loss_yns_1": 0.15927, "loss_cls_2": 1.02825, "loss_box_2": 1.95709, "loss_cns_2": 0.64227, "loss_yns_2": 0.15936, "loss_cls_3": 1.04497, "loss_box_3": 1.94537, "loss_cns_3": 0.64532, "loss_yns_3": 0.15927, "loss_cls_4": 1.04308, "loss_box_4": 1.94729, "loss_cns_4": 0.64776, "loss_yns_4": 0.15711, "loss_cls_5": 1.10121, "loss_box_5": 1.96487, "loss_cns_5": 0.64792, "loss_yns_5": 0.16084, "loss_cls_dn_0": 0.28101, "loss_box_dn_0": 0.85612, "loss_cls_dn_1": 0.20556, "loss_box_dn_1": 0.85119, "loss_cls_dn_2": 0.2206, "loss_box_dn_2": 0.83767, "loss_cls_dn_3": 0.21959, "loss_box_dn_3": 0.84953, "loss_cls_dn_4": 0.23109, "loss_box_dn_4": 0.87027, "loss_cls_dn_5": 0.28912, "loss_box_dn_5": 0.89898, "loss_dense_depth": 0.87502, "loss": 30.01779, "grad_norm": 53.55674, "time": 1.5972} -{"mode": "train", "epoch": 1, "iter": 72, "lr": 0.00013, "memory": 49163, "data_time": 0.07819, "loss_cls_0": 0.94047, "loss_box_0": 1.77711, "loss_cns_0": 0.61312, "loss_yns_0": 0.15773, "loss_cls_1": 0.98947, "loss_box_1": 1.99855, "loss_cns_1": 0.62534, "loss_yns_1": 0.16025, "loss_cls_2": 1.03245, "loss_box_2": 1.94298, "loss_cns_2": 0.64386, "loss_yns_2": 0.15883, "loss_cls_3": 1.06444, "loss_box_3": 1.94247, "loss_cns_3": 0.64515, "loss_yns_3": 0.1587, "loss_cls_4": 1.05739, "loss_box_4": 1.92413, "loss_cns_4": 0.64682, "loss_yns_4": 0.15696, "loss_cls_5": 1.11464, "loss_box_5": 1.94838, "loss_cns_5": 0.64539, "loss_yns_5": 0.15754, "loss_cls_dn_0": 0.28195, "loss_box_dn_0": 0.84609, "loss_cls_dn_1": 0.21164, "loss_box_dn_1": 0.8582, "loss_cls_dn_2": 0.22347, "loss_box_dn_2": 0.85044, "loss_cls_dn_3": 0.22611, "loss_box_dn_3": 0.86305, "loss_cls_dn_4": 0.23106, "loss_box_dn_4": 0.88877, "loss_cls_dn_5": 0.28767, "loss_box_dn_5": 0.91484, "loss_dense_depth": 0.90931, "loss": 30.09476, "grad_norm": 62.63432, "time": 1.49164} -{"mode": "train", "epoch": 1, "iter": 73, "lr": 0.00013, "memory": 49163, "data_time": 0.07858, "loss_cls_0": 0.95431, "loss_box_0": 1.79473, "loss_cns_0": 0.60539, "loss_yns_0": 0.15524, "loss_cls_1": 1.00171, "loss_box_1": 2.03586, "loss_cns_1": 0.61927, "loss_yns_1": 0.15967, "loss_cls_2": 1.03736, "loss_box_2": 1.96815, "loss_cns_2": 0.63917, "loss_yns_2": 0.16041, "loss_cls_3": 1.05725, "loss_box_3": 1.96065, "loss_cns_3": 0.64345, "loss_yns_3": 0.15911, "loss_cls_4": 1.05886, "loss_box_4": 1.96139, "loss_cns_4": 0.64176, "loss_yns_4": 0.15857, "loss_cls_5": 1.09358, "loss_box_5": 1.97896, "loss_cns_5": 0.64269, "loss_yns_5": 0.15857, "loss_cls_dn_0": 0.28134, "loss_box_dn_0": 0.84507, "loss_cls_dn_1": 0.20998, "loss_box_dn_1": 0.88596, "loss_cls_dn_2": 0.22009, "loss_box_dn_2": 0.86379, "loss_cls_dn_3": 0.2234, "loss_box_dn_3": 0.87469, "loss_cls_dn_4": 0.22151, "loss_box_dn_4": 0.90565, "loss_cls_dn_5": 0.26976, "loss_box_dn_5": 0.93212, "loss_dense_depth": 0.88357, "loss": 30.26304, "grad_norm": 44.49735, "time": 1.4884} -{"mode": "train", "epoch": 1, "iter": 74, "lr": 0.00013, "memory": 49163, "data_time": 0.0775, "loss_cls_0": 0.94902, "loss_box_0": 1.8131, "loss_cns_0": 0.60576, "loss_yns_0": 0.15651, "loss_cls_1": 0.99508, "loss_box_1": 2.09751, "loss_cns_1": 0.61668, "loss_yns_1": 0.16088, "loss_cls_2": 1.03752, "loss_box_2": 2.02443, "loss_cns_2": 0.63473, "loss_yns_2": 0.16314, "loss_cls_3": 1.03903, "loss_box_3": 2.00004, "loss_cns_3": 0.64516, "loss_yns_3": 0.1614, "loss_cls_4": 1.06218, "loss_box_4": 2.00921, "loss_cns_4": 0.64161, "loss_yns_4": 0.15984, "loss_cls_5": 1.05703, "loss_box_5": 2.02571, "loss_cns_5": 0.64352, "loss_yns_5": 0.16039, "loss_cls_dn_0": 0.27821, "loss_box_dn_0": 0.84756, "loss_cls_dn_1": 0.20333, "loss_box_dn_1": 0.93385, "loss_cls_dn_2": 0.21409, "loss_box_dn_2": 0.90065, "loss_cls_dn_3": 0.21932, "loss_box_dn_3": 0.90626, "loss_cls_dn_4": 0.21739, "loss_box_dn_4": 0.92696, "loss_cls_dn_5": 0.2411, "loss_box_dn_5": 0.94988, "loss_dense_depth": 0.88366, "loss": 30.58174, "grad_norm": 55.4082, "time": 1.48554} -{"mode": "train", "epoch": 1, "iter": 75, "lr": 0.00013, "memory": 49163, "data_time": 0.0789, "loss_cls_0": 0.95219, "loss_box_0": 1.83664, "loss_cns_0": 0.60522, "loss_yns_0": 0.15955, "loss_cls_1": 1.0119, "loss_box_1": 2.07871, "loss_cns_1": 0.62763, "loss_yns_1": 0.16362, "loss_cls_2": 1.03785, "loss_box_2": 2.05216, "loss_cns_2": 0.63651, "loss_yns_2": 0.16299, "loss_cls_3": 1.05228, "loss_box_3": 2.03176, "loss_cns_3": 0.64664, "loss_yns_3": 0.16155, "loss_cls_4": 1.07876, "loss_box_4": 2.04484, "loss_cns_4": 0.64565, "loss_yns_4": 0.16071, "loss_cls_5": 1.09609, "loss_box_5": 2.06442, "loss_cns_5": 0.64954, "loss_yns_5": 0.16107, "loss_cls_dn_0": 0.27323, "loss_box_dn_0": 0.83748, "loss_cls_dn_1": 0.1983, "loss_box_dn_1": 0.93807, "loss_cls_dn_2": 0.21462, "loss_box_dn_2": 0.90613, "loss_cls_dn_3": 0.21881, "loss_box_dn_3": 0.90546, "loss_cls_dn_4": 0.21996, "loss_box_dn_4": 0.91859, "loss_cls_dn_5": 0.23084, "loss_box_dn_5": 0.93166, "loss_dense_depth": 0.86187, "loss": 30.77328, "grad_norm": 69.97112, "time": 1.52912} -{"mode": "train", "epoch": 1, "iter": 76, "lr": 0.00013, "memory": 49163, "data_time": 0.07721, "loss_cls_0": 0.95136, "loss_box_0": 1.82059, "loss_cns_0": 0.60538, "loss_yns_0": 0.16022, "loss_cls_1": 1.01822, "loss_box_1": 2.03961, "loss_cns_1": 0.63687, "loss_yns_1": 0.16461, "loss_cls_2": 1.0338, "loss_box_2": 2.02719, "loss_cns_2": 0.64286, "loss_yns_2": 0.16154, "loss_cls_3": 1.05062, "loss_box_3": 2.01796, "loss_cns_3": 0.6475, "loss_yns_3": 0.16102, "loss_cls_4": 1.05337, "loss_box_4": 2.01891, "loss_cns_4": 0.65044, "loss_yns_4": 0.16056, "loss_cls_5": 1.08192, "loss_box_5": 2.03065, "loss_cns_5": 0.65207, "loss_yns_5": 0.15935, "loss_cls_dn_0": 0.27212, "loss_box_dn_0": 0.83884, "loss_cls_dn_1": 0.19955, "loss_box_dn_1": 0.87748, "loss_cls_dn_2": 0.21617, "loss_box_dn_2": 0.84669, "loss_cls_dn_3": 0.21452, "loss_box_dn_3": 0.84162, "loss_cls_dn_4": 0.2154, "loss_box_dn_4": 0.84862, "loss_cls_dn_5": 0.22448, "loss_box_dn_5": 0.85221, "loss_dense_depth": 0.86934, "loss": 30.26364, "grad_norm": 47.23517, "time": 1.48687} -{"mode": "train", "epoch": 1, "iter": 77, "lr": 0.00013, "memory": 49163, "data_time": 0.07947, "loss_cls_0": 0.94492, "loss_box_0": 1.87641, "loss_cns_0": 0.6019, "loss_yns_0": 0.15914, "loss_cls_1": 1.0218, "loss_box_1": 2.00132, "loss_cns_1": 0.63997, "loss_yns_1": 0.16422, "loss_cls_2": 1.05311, "loss_box_2": 1.9643, "loss_cns_2": 0.64642, "loss_yns_2": 0.16201, "loss_cls_3": 1.0633, "loss_box_3": 1.97915, "loss_cns_3": 0.65254, "loss_yns_3": 0.1627, "loss_cls_4": 1.06883, "loss_box_4": 1.95507, "loss_cns_4": 0.6558, "loss_yns_4": 0.16312, "loss_cls_5": 1.08218, "loss_box_5": 1.95359, "loss_cns_5": 0.65393, "loss_yns_5": 0.16287, "loss_cls_dn_0": 0.2695, "loss_box_dn_0": 0.83569, "loss_cls_dn_1": 0.19991, "loss_box_dn_1": 0.79654, "loss_cls_dn_2": 0.21887, "loss_box_dn_2": 0.77046, "loss_cls_dn_3": 0.2133, "loss_box_dn_3": 0.77449, "loss_cls_dn_4": 0.21361, "loss_box_dn_4": 0.77906, "loss_cls_dn_5": 0.22531, "loss_box_dn_5": 0.78622, "loss_dense_depth": 0.85756, "loss": 29.72907, "grad_norm": 29.75451, "time": 1.48968} -{"mode": "train", "epoch": 1, "iter": 78, "lr": 0.00013, "memory": 49163, "data_time": 0.07642, "loss_cls_0": 0.95574, "loss_box_0": 1.87445, "loss_cns_0": 0.61004, "loss_yns_0": 0.16213, "loss_cls_1": 1.04342, "loss_box_1": 2.07609, "loss_cns_1": 0.63471, "loss_yns_1": 0.16425, "loss_cls_2": 1.06594, "loss_box_2": 2.04117, "loss_cns_2": 0.64187, "loss_yns_2": 0.16419, "loss_cls_3": 1.06844, "loss_box_3": 2.05842, "loss_cns_3": 0.64912, "loss_yns_3": 0.16396, "loss_cls_4": 1.08936, "loss_box_4": 2.04768, "loss_cns_4": 0.65085, "loss_yns_4": 0.16372, "loss_cls_5": 1.08672, "loss_box_5": 2.04615, "loss_cns_5": 0.65099, "loss_yns_5": 0.16352, "loss_cls_dn_0": 0.26706, "loss_box_dn_0": 0.82929, "loss_cls_dn_1": 0.19831, "loss_box_dn_1": 0.8092, "loss_cls_dn_2": 0.21811, "loss_box_dn_2": 0.79356, "loss_cls_dn_3": 0.21592, "loss_box_dn_3": 0.8126, "loss_cls_dn_4": 0.21907, "loss_box_dn_4": 0.82599, "loss_cls_dn_5": 0.22076, "loss_box_dn_5": 0.84582, "loss_dense_depth": 0.89362, "loss": 30.42224, "grad_norm": 53.22449, "time": 1.47944} -{"mode": "train", "epoch": 1, "iter": 79, "lr": 0.00013, "memory": 49163, "data_time": 0.08043, "loss_cls_0": 0.92326, "loss_box_0": 1.87, "loss_cns_0": 0.6209, "loss_yns_0": 0.16204, "loss_cls_1": 1.01621, "loss_box_1": 2.07778, "loss_cns_1": 0.6397, "loss_yns_1": 0.16493, "loss_cls_2": 1.02343, "loss_box_2": 2.04408, "loss_cns_2": 0.644, "loss_yns_2": 0.16272, "loss_cls_3": 1.04871, "loss_box_3": 2.05578, "loss_cns_3": 0.64702, "loss_yns_3": 0.16329, "loss_cls_4": 1.06174, "loss_box_4": 2.06322, "loss_cns_4": 0.64859, "loss_yns_4": 0.1626, "loss_cls_5": 1.06916, "loss_box_5": 2.05866, "loss_cns_5": 0.65007, "loss_yns_5": 0.16342, "loss_cls_dn_0": 0.25669, "loss_box_dn_0": 0.83264, "loss_cls_dn_1": 0.19977, "loss_box_dn_1": 0.85698, "loss_cls_dn_2": 0.21484, "loss_box_dn_2": 0.84776, "loss_cls_dn_3": 0.21781, "loss_box_dn_3": 0.87303, "loss_cls_dn_4": 0.22318, "loss_box_dn_4": 0.89082, "loss_cls_dn_5": 0.2286, "loss_box_dn_5": 0.92101, "loss_dense_depth": 0.82744, "loss": 30.53192, "grad_norm": 56.6721, "time": 1.49254} -{"mode": "train", "epoch": 1, "iter": 80, "lr": 0.00013, "memory": 49163, "data_time": 0.07769, "loss_cls_0": 0.92941, "loss_box_0": 1.86925, "loss_cns_0": 0.61961, "loss_yns_0": 0.16019, "loss_cls_1": 1.01785, "loss_box_1": 2.0641, "loss_cns_1": 0.63805, "loss_yns_1": 0.16622, "loss_cls_2": 1.02382, "loss_box_2": 2.0127, "loss_cns_2": 0.64656, "loss_yns_2": 0.16469, "loss_cls_3": 1.04561, "loss_box_3": 2.01718, "loss_cns_3": 0.64789, "loss_yns_3": 0.1601, "loss_cls_4": 1.03766, "loss_box_4": 2.02095, "loss_cns_4": 0.65053, "loss_yns_4": 0.16043, "loss_cls_5": 1.0706, "loss_box_5": 2.01454, "loss_cns_5": 0.65365, "loss_yns_5": 0.1655, "loss_cls_dn_0": 0.25585, "loss_box_dn_0": 0.82723, "loss_cls_dn_1": 0.20155, "loss_box_dn_1": 0.89064, "loss_cls_dn_2": 0.20329, "loss_box_dn_2": 0.87242, "loss_cls_dn_3": 0.20645, "loss_box_dn_3": 0.88891, "loss_cls_dn_4": 0.21078, "loss_box_dn_4": 0.90035, "loss_cls_dn_5": 0.22781, "loss_box_dn_5": 0.92527, "loss_dense_depth": 0.84691, "loss": 30.41454, "grad_norm": 41.13528, "time": 1.50373} -{"mode": "train", "epoch": 1, "iter": 81, "lr": 0.00013, "memory": 49163, "data_time": 0.10497, "loss_cls_0": 0.95547, "loss_box_0": 1.89488, "loss_cns_0": 0.61494, "loss_yns_0": 0.16122, "loss_cls_1": 1.02588, "loss_box_1": 2.0531, "loss_cns_1": 0.6339, "loss_yns_1": 0.16318, "loss_cls_2": 1.08315, "loss_box_2": 1.98247, "loss_cns_2": 0.64775, "loss_yns_2": 0.16234, "loss_cls_3": 1.08511, "loss_box_3": 1.97336, "loss_cns_3": 0.65116, "loss_yns_3": 0.16048, "loss_cls_4": 1.06289, "loss_box_4": 1.98706, "loss_cns_4": 0.65218, "loss_yns_4": 0.15963, "loss_cls_5": 1.09848, "loss_box_5": 1.9746, "loss_cns_5": 0.65434, "loss_yns_5": 0.16386, "loss_cls_dn_0": 0.26295, "loss_box_dn_0": 0.83391, "loss_cls_dn_1": 0.19664, "loss_box_dn_1": 0.91742, "loss_cls_dn_2": 0.20061, "loss_box_dn_2": 0.88786, "loss_cls_dn_3": 0.20024, "loss_box_dn_3": 0.88828, "loss_cls_dn_4": 0.20157, "loss_box_dn_4": 0.89192, "loss_cls_dn_5": 0.22853, "loss_box_dn_5": 0.89832, "loss_dense_depth": 0.85458, "loss": 30.46424, "grad_norm": 45.8926, "time": 1.52786} -{"mode": "train", "epoch": 1, "iter": 82, "lr": 0.00013, "memory": 49163, "data_time": 0.10475, "loss_cls_0": 0.95575, "loss_box_0": 1.8893, "loss_cns_0": 0.61174, "loss_yns_0": 0.15706, "loss_cls_1": 1.02512, "loss_box_1": 2.05268, "loss_cns_1": 0.63218, "loss_yns_1": 0.15963, "loss_cls_2": 1.06273, "loss_box_2": 1.99839, "loss_cns_2": 0.64447, "loss_yns_2": 0.15752, "loss_cls_3": 1.07406, "loss_box_3": 1.98397, "loss_cns_3": 0.64928, "loss_yns_3": 0.15794, "loss_cls_4": 1.07286, "loss_box_4": 1.99775, "loss_cns_4": 0.648, "loss_yns_4": 0.15825, "loss_cls_5": 1.09334, "loss_box_5": 1.98169, "loss_cns_5": 0.65056, "loss_yns_5": 0.15992, "loss_cls_dn_0": 0.26019, "loss_box_dn_0": 0.83775, "loss_cls_dn_1": 0.18357, "loss_box_dn_1": 0.91215, "loss_cls_dn_2": 0.19445, "loss_box_dn_2": 0.88469, "loss_cls_dn_3": 0.19878, "loss_box_dn_3": 0.87866, "loss_cls_dn_4": 0.19814, "loss_box_dn_4": 0.87767, "loss_cls_dn_5": 0.22005, "loss_box_dn_5": 0.87697, "loss_dense_depth": 0.89398, "loss": 30.39123, "grad_norm": 50.65323, "time": 1.52557} -{"mode": "train", "epoch": 1, "iter": 83, "lr": 0.00013, "memory": 49163, "data_time": 0.08077, "loss_cls_0": 0.94897, "loss_box_0": 1.84515, "loss_cns_0": 0.61351, "loss_yns_0": 0.15698, "loss_cls_1": 1.08531, "loss_box_1": 1.99645, "loss_cns_1": 0.62679, "loss_yns_1": 0.15862, "loss_cls_2": 1.02548, "loss_box_2": 1.92504, "loss_cns_2": 0.64169, "loss_yns_2": 0.15766, "loss_cls_3": 1.04642, "loss_box_3": 1.91091, "loss_cns_3": 0.64844, "loss_yns_3": 0.15833, "loss_cls_4": 1.05443, "loss_box_4": 1.9049, "loss_cns_4": 0.64749, "loss_yns_4": 0.15822, "loss_cls_5": 1.09382, "loss_box_5": 1.90595, "loss_cns_5": 0.64589, "loss_yns_5": 0.1592, "loss_cls_dn_0": 0.25711, "loss_box_dn_0": 0.83479, "loss_cls_dn_1": 0.18014, "loss_box_dn_1": 0.85493, "loss_cls_dn_2": 0.19126, "loss_box_dn_2": 0.82947, "loss_cls_dn_3": 0.19537, "loss_box_dn_3": 0.83216, "loss_cls_dn_4": 0.19514, "loss_box_dn_4": 0.83228, "loss_cls_dn_5": 0.21053, "loss_box_dn_5": 0.84008, "loss_dense_depth": 0.85915, "loss": 29.62804, "grad_norm": 42.72862, "time": 1.52703} -{"mode": "train", "epoch": 1, "iter": 84, "lr": 0.00013, "memory": 49163, "data_time": 0.08287, "loss_cls_0": 0.93018, "loss_box_0": 1.82219, "loss_cns_0": 0.61545, "loss_yns_0": 0.1556, "loss_cls_1": 1.02521, "loss_box_1": 1.95881, "loss_cns_1": 0.64207, "loss_yns_1": 0.15809, "loss_cls_2": 1.0542, "loss_box_2": 1.9165, "loss_cns_2": 0.649, "loss_yns_2": 0.15883, "loss_cls_3": 1.06897, "loss_box_3": 1.93424, "loss_cns_3": 0.65125, "loss_yns_3": 0.15856, "loss_cls_4": 1.04299, "loss_box_4": 1.92924, "loss_cns_4": 0.64941, "loss_yns_4": 0.15805, "loss_cls_5": 1.05491, "loss_box_5": 1.93307, "loss_cns_5": 0.64751, "loss_yns_5": 0.15915, "loss_cls_dn_0": 0.25379, "loss_box_dn_0": 0.82096, "loss_cls_dn_1": 0.19186, "loss_box_dn_1": 0.7893, "loss_cls_dn_2": 0.19693, "loss_box_dn_2": 0.7761, "loss_cls_dn_3": 0.20736, "loss_box_dn_3": 0.79478, "loss_cls_dn_4": 0.20142, "loss_box_dn_4": 0.80604, "loss_cls_dn_5": 0.21944, "loss_box_dn_5": 0.8253, "loss_dense_depth": 0.83129, "loss": 29.38807, "grad_norm": 47.53352, "time": 1.5222} -{"mode": "train", "epoch": 1, "iter": 85, "lr": 0.00013, "memory": 49163, "data_time": 0.08947, "loss_cls_0": 0.94973, "loss_box_0": 1.83744, "loss_cns_0": 0.61915, "loss_yns_0": 0.16041, "loss_cls_1": 1.10068, "loss_box_1": 1.98777, "loss_cns_1": 0.62962, "loss_yns_1": 0.15859, "loss_cls_2": 1.09069, "loss_box_2": 1.98547, "loss_cns_2": 0.64468, "loss_yns_2": 0.16307, "loss_cls_3": 1.09103, "loss_box_3": 1.99098, "loss_cns_3": 0.64823, "loss_yns_3": 0.16236, "loss_cls_4": 1.06774, "loss_box_4": 1.99195, "loss_cns_4": 0.64669, "loss_yns_4": 0.16283, "loss_cls_5": 1.07164, "loss_box_5": 1.9865, "loss_cns_5": 0.64772, "loss_yns_5": 0.16316, "loss_cls_dn_0": 0.25914, "loss_box_dn_0": 0.83357, "loss_cls_dn_1": 0.1994, "loss_box_dn_1": 0.84103, "loss_cls_dn_2": 0.2027, "loss_box_dn_2": 0.83431, "loss_cls_dn_3": 0.20774, "loss_box_dn_3": 0.84791, "loss_cls_dn_4": 0.20547, "loss_box_dn_4": 0.86621, "loss_cls_dn_5": 0.22227, "loss_box_dn_5": 0.88187, "loss_dense_depth": 0.84927, "loss": 30.20903, "grad_norm": 50.9319, "time": 1.51202} -{"mode": "train", "epoch": 1, "iter": 86, "lr": 0.00013, "memory": 49163, "data_time": 0.08663, "loss_cls_0": 0.94003, "loss_box_0": 1.82204, "loss_cns_0": 0.62073, "loss_yns_0": 0.16263, "loss_cls_1": 1.0533, "loss_box_1": 1.9374, "loss_cns_1": 0.63128, "loss_yns_1": 0.164, "loss_cls_2": 1.06625, "loss_box_2": 1.90434, "loss_cns_2": 0.64207, "loss_yns_2": 0.16506, "loss_cls_3": 1.08588, "loss_box_3": 1.89438, "loss_cns_3": 0.64512, "loss_yns_3": 0.16766, "loss_cls_4": 1.03787, "loss_box_4": 1.8924, "loss_cns_4": 0.64482, "loss_yns_4": 0.16516, "loss_cls_5": 1.05907, "loss_box_5": 1.8909, "loss_cns_5": 0.64968, "loss_yns_5": 0.16359, "loss_cls_dn_0": 0.24659, "loss_box_dn_0": 0.82893, "loss_cls_dn_1": 0.19316, "loss_box_dn_1": 0.84308, "loss_cls_dn_2": 0.19689, "loss_box_dn_2": 0.83207, "loss_cls_dn_3": 0.19967, "loss_box_dn_3": 0.83396, "loss_cls_dn_4": 0.20075, "loss_box_dn_4": 0.84954, "loss_cls_dn_5": 0.21169, "loss_box_dn_5": 0.86485, "loss_dense_depth": 0.8668, "loss": 29.57362, "grad_norm": 38.82041, "time": 1.50569} -{"mode": "train", "epoch": 1, "iter": 87, "lr": 0.00013, "memory": 49163, "data_time": 0.11241, "loss_cls_0": 0.91194, "loss_box_0": 1.80124, "loss_cns_0": 0.62039, "loss_yns_0": 0.16077, "loss_cls_1": 0.99488, "loss_box_1": 1.92331, "loss_cns_1": 0.63907, "loss_yns_1": 0.16314, "loss_cls_2": 1.01066, "loss_box_2": 1.88616, "loss_cns_2": 0.64797, "loss_yns_2": 0.16382, "loss_cls_3": 1.04168, "loss_box_3": 1.89242, "loss_cns_3": 0.6537, "loss_yns_3": 0.16515, "loss_cls_4": 1.02833, "loss_box_4": 1.89788, "loss_cns_4": 0.65335, "loss_yns_4": 0.16635, "loss_cls_5": 1.05438, "loss_box_5": 1.8982, "loss_cns_5": 0.65545, "loss_yns_5": 0.16155, "loss_cls_dn_0": 0.24542, "loss_box_dn_0": 0.81802, "loss_cls_dn_1": 0.18625, "loss_box_dn_1": 0.81329, "loss_cls_dn_2": 0.19301, "loss_box_dn_2": 0.79773, "loss_cls_dn_3": 0.19575, "loss_box_dn_3": 0.79656, "loss_cls_dn_4": 0.20151, "loss_box_dn_4": 0.80399, "loss_cls_dn_5": 0.20934, "loss_box_dn_5": 0.81699, "loss_dense_depth": 0.83358, "loss": 29.1032, "grad_norm": 42.20626, "time": 1.51915} -{"mode": "train", "epoch": 1, "iter": 88, "lr": 0.00013, "memory": 49163, "data_time": 0.07863, "loss_cls_0": 0.94006, "loss_box_0": 1.80844, "loss_cns_0": 0.61397, "loss_yns_0": 0.15766, "loss_cls_1": 1.01966, "loss_box_1": 1.95675, "loss_cns_1": 0.6394, "loss_yns_1": 0.16018, "loss_cls_2": 1.03045, "loss_box_2": 1.92978, "loss_cns_2": 0.64778, "loss_yns_2": 0.1619, "loss_cls_3": 1.05233, "loss_box_3": 1.91693, "loss_cns_3": 0.65497, "loss_yns_3": 0.16286, "loss_cls_4": 1.03842, "loss_box_4": 1.92456, "loss_cns_4": 0.6535, "loss_yns_4": 0.16106, "loss_cls_5": 1.06239, "loss_box_5": 1.91021, "loss_cns_5": 0.6551, "loss_yns_5": 0.15853, "loss_cls_dn_0": 0.24995, "loss_box_dn_0": 0.8297, "loss_cls_dn_1": 0.18308, "loss_box_dn_1": 0.80785, "loss_cls_dn_2": 0.19582, "loss_box_dn_2": 0.78704, "loss_cls_dn_3": 0.19908, "loss_box_dn_3": 0.78135, "loss_cls_dn_4": 0.19921, "loss_box_dn_4": 0.78042, "loss_cls_dn_5": 0.21088, "loss_box_dn_5": 0.78289, "loss_dense_depth": 0.82373, "loss": 29.24788, "grad_norm": 55.77703, "time": 1.48501} -{"mode": "train", "epoch": 1, "iter": 89, "lr": 0.00014, "memory": 49163, "data_time": 0.09336, "loss_cls_0": 0.92764, "loss_box_0": 1.81684, "loss_cns_0": 0.61426, "loss_yns_0": 0.1568, "loss_cls_1": 1.03265, "loss_box_1": 1.94225, "loss_cns_1": 0.63439, "loss_yns_1": 0.15988, "loss_cls_2": 1.03573, "loss_box_2": 1.90902, "loss_cns_2": 0.64707, "loss_yns_2": 0.16031, "loss_cls_3": 1.04707, "loss_box_3": 1.87954, "loss_cns_3": 0.64967, "loss_yns_3": 0.15761, "loss_cls_4": 1.01796, "loss_box_4": 1.88981, "loss_cns_4": 0.64883, "loss_yns_4": 0.15774, "loss_cls_5": 1.04381, "loss_box_5": 1.87525, "loss_cns_5": 0.65145, "loss_yns_5": 0.15812, "loss_cls_dn_0": 0.2502, "loss_box_dn_0": 0.82892, "loss_cls_dn_1": 0.18593, "loss_box_dn_1": 0.7916, "loss_cls_dn_2": 0.20144, "loss_box_dn_2": 0.76238, "loss_cls_dn_3": 0.20058, "loss_box_dn_3": 0.75519, "loss_cls_dn_4": 0.19724, "loss_box_dn_4": 0.75472, "loss_cls_dn_5": 0.20852, "loss_box_dn_5": 0.75458, "loss_dense_depth": 0.82376, "loss": 28.92876, "grad_norm": 49.21077, "time": 1.50752} -{"mode": "train", "epoch": 1, "iter": 90, "lr": 0.00014, "memory": 49163, "data_time": 0.07914, "loss_cls_0": 0.94205, "loss_box_0": 1.80902, "loss_cns_0": 0.61481, "loss_yns_0": 0.15544, "loss_cls_1": 1.02737, "loss_box_1": 2.00108, "loss_cns_1": 0.63456, "loss_yns_1": 0.15792, "loss_cls_2": 1.0446, "loss_box_2": 1.96079, "loss_cns_2": 0.64379, "loss_yns_2": 0.16019, "loss_cls_3": 1.05729, "loss_box_3": 1.92158, "loss_cns_3": 0.64783, "loss_yns_3": 0.15863, "loss_cls_4": 1.03101, "loss_box_4": 1.91295, "loss_cns_4": 0.64932, "loss_yns_4": 0.15626, "loss_cls_5": 1.04698, "loss_box_5": 1.92248, "loss_cns_5": 0.65276, "loss_yns_5": 0.16027, "loss_cls_dn_0": 0.24837, "loss_box_dn_0": 0.83868, "loss_cls_dn_1": 0.18598, "loss_box_dn_1": 0.78647, "loss_cls_dn_2": 0.19534, "loss_box_dn_2": 0.76458, "loss_cls_dn_3": 0.19674, "loss_box_dn_3": 0.7562, "loss_cls_dn_4": 0.19701, "loss_box_dn_4": 0.76317, "loss_cls_dn_5": 0.20554, "loss_box_dn_5": 0.77767, "loss_dense_depth": 0.85307, "loss": 29.23778, "grad_norm": 35.8318, "time": 1.4908} -{"mode": "train", "epoch": 1, "iter": 91, "lr": 0.00014, "memory": 49163, "data_time": 0.0829, "loss_cls_0": 0.92161, "loss_box_0": 1.75732, "loss_cns_0": 0.60989, "loss_yns_0": 0.14908, "loss_cls_1": 1.00519, "loss_box_1": 1.98479, "loss_cns_1": 0.63471, "loss_yns_1": 0.15677, "loss_cls_2": 1.02186, "loss_box_2": 1.9287, "loss_cns_2": 0.64483, "loss_yns_2": 0.15674, "loss_cls_3": 1.04155, "loss_box_3": 1.93275, "loss_cns_3": 0.64975, "loss_yns_3": 0.15783, "loss_cls_4": 1.05193, "loss_box_4": 1.93123, "loss_cns_4": 0.64942, "loss_yns_4": 0.15429, "loss_cls_5": 1.08282, "loss_box_5": 1.93477, "loss_cns_5": 0.65439, "loss_yns_5": 0.15788, "loss_cls_dn_0": 0.24272, "loss_box_dn_0": 0.81727, "loss_cls_dn_1": 0.18729, "loss_box_dn_1": 0.80531, "loss_cls_dn_2": 0.19256, "loss_box_dn_2": 0.78721, "loss_cls_dn_3": 0.20037, "loss_box_dn_3": 0.79768, "loss_cls_dn_4": 0.2068, "loss_box_dn_4": 0.82, "loss_cls_dn_5": 0.2135, "loss_box_dn_5": 0.84266, "loss_dense_depth": 0.81005, "loss": 29.29351, "grad_norm": 46.88882, "time": 1.5325} -{"mode": "train", "epoch": 1, "iter": 92, "lr": 0.00014, "memory": 49163, "data_time": 0.07982, "loss_cls_0": 0.9266, "loss_box_0": 1.74684, "loss_cns_0": 0.60864, "loss_yns_0": 0.14848, "loss_cls_1": 0.9994, "loss_box_1": 2.00187, "loss_cns_1": 0.63649, "loss_yns_1": 0.15548, "loss_cls_2": 1.01268, "loss_box_2": 1.95941, "loss_cns_2": 0.64665, "loss_yns_2": 0.15464, "loss_cls_3": 1.03416, "loss_box_3": 1.9865, "loss_cns_3": 0.6534, "loss_yns_3": 0.15804, "loss_cls_4": 1.03892, "loss_box_4": 1.99388, "loss_cns_4": 0.65019, "loss_yns_4": 0.15553, "loss_cls_5": 1.06334, "loss_box_5": 1.98918, "loss_cns_5": 0.65162, "loss_yns_5": 0.1555, "loss_cls_dn_0": 0.2437, "loss_box_dn_0": 0.81676, "loss_cls_dn_1": 0.18842, "loss_box_dn_1": 0.81155, "loss_cls_dn_2": 0.19107, "loss_box_dn_2": 0.80046, "loss_cls_dn_3": 0.20114, "loss_box_dn_3": 0.82165, "loss_cls_dn_4": 0.21068, "loss_box_dn_4": 0.84511, "loss_cls_dn_5": 0.21998, "loss_box_dn_5": 0.86592, "loss_dense_depth": 0.82978, "loss": 29.57368, "grad_norm": 53.05187, "time": 1.49976} -{"mode": "train", "epoch": 1, "iter": 93, "lr": 0.00014, "memory": 49163, "data_time": 0.07963, "loss_cls_0": 0.91354, "loss_box_0": 1.771, "loss_cns_0": 0.62053, "loss_yns_0": 0.15156, "loss_cls_1": 1.01635, "loss_box_1": 1.97101, "loss_cns_1": 0.64135, "loss_yns_1": 0.15307, "loss_cls_2": 1.01774, "loss_box_2": 1.92044, "loss_cns_2": 0.65122, "loss_yns_2": 0.15452, "loss_cls_3": 1.04063, "loss_box_3": 1.93361, "loss_cns_3": 0.65665, "loss_yns_3": 0.1559, "loss_cls_4": 1.03395, "loss_box_4": 1.94241, "loss_cns_4": 0.6538, "loss_yns_4": 0.15391, "loss_cls_5": 1.04237, "loss_box_5": 1.93453, "loss_cns_5": 0.65658, "loss_yns_5": 0.15525, "loss_cls_dn_0": 0.24098, "loss_box_dn_0": 0.81286, "loss_cls_dn_1": 0.19313, "loss_box_dn_1": 0.82003, "loss_cls_dn_2": 0.19572, "loss_box_dn_2": 0.80588, "loss_cls_dn_3": 0.20527, "loss_box_dn_3": 0.81584, "loss_cls_dn_4": 0.21239, "loss_box_dn_4": 0.82862, "loss_cls_dn_5": 0.22812, "loss_box_dn_5": 0.84247, "loss_dense_depth": 0.79663, "loss": 29.33986, "grad_norm": 46.32459, "time": 1.49378} -{"mode": "train", "epoch": 1, "iter": 94, "lr": 0.00014, "memory": 49163, "data_time": 0.08049, "loss_cls_0": 0.92802, "loss_box_0": 1.80047, "loss_cns_0": 0.62088, "loss_yns_0": 0.15483, "loss_cls_1": 1.03303, "loss_box_1": 1.97978, "loss_cns_1": 0.63357, "loss_yns_1": 0.15506, "loss_cls_2": 1.01805, "loss_box_2": 1.93349, "loss_cns_2": 0.64409, "loss_yns_2": 0.15599, "loss_cls_3": 1.05835, "loss_box_3": 1.91182, "loss_cns_3": 0.65373, "loss_yns_3": 0.15689, "loss_cls_4": 1.04232, "loss_box_4": 1.92426, "loss_cns_4": 0.65045, "loss_yns_4": 0.15814, "loss_cls_5": 1.06647, "loss_box_5": 1.90654, "loss_cns_5": 0.6529, "loss_yns_5": 0.15746, "loss_cls_dn_0": 0.23913, "loss_box_dn_0": 0.80728, "loss_cls_dn_1": 0.19292, "loss_box_dn_1": 0.79896, "loss_cls_dn_2": 0.19384, "loss_box_dn_2": 0.7737, "loss_cls_dn_3": 0.20039, "loss_box_dn_3": 0.76824, "loss_cls_dn_4": 0.19591, "loss_box_dn_4": 0.77202, "loss_cls_dn_5": 0.21812, "loss_box_dn_5": 0.77772, "loss_dense_depth": 0.79374, "loss": 29.12858, "grad_norm": 36.77536, "time": 1.47988} -{"mode": "train", "epoch": 1, "iter": 95, "lr": 0.00014, "memory": 49163, "data_time": 0.08741, "loss_cls_0": 0.92726, "loss_box_0": 1.81879, "loss_cns_0": 0.61891, "loss_yns_0": 0.15555, "loss_cls_1": 1.01909, "loss_box_1": 1.94342, "loss_cns_1": 0.63774, "loss_yns_1": 0.155, "loss_cls_2": 1.02848, "loss_box_2": 1.90195, "loss_cns_2": 0.65068, "loss_yns_2": 0.15627, "loss_cls_3": 1.10422, "loss_box_3": 1.86656, "loss_cns_3": 0.6545, "loss_yns_3": 0.15897, "loss_cls_4": 1.03542, "loss_box_4": 1.88509, "loss_cns_4": 0.65525, "loss_yns_4": 0.15737, "loss_cls_5": 1.06669, "loss_box_5": 1.8768, "loss_cns_5": 0.65724, "loss_yns_5": 0.15612, "loss_cls_dn_0": 0.23839, "loss_box_dn_0": 0.80569, "loss_cls_dn_1": 0.1877, "loss_box_dn_1": 0.7706, "loss_cls_dn_2": 0.19162, "loss_box_dn_2": 0.74376, "loss_cls_dn_3": 0.19394, "loss_box_dn_3": 0.73795, "loss_cls_dn_4": 0.19201, "loss_box_dn_4": 0.74112, "loss_cls_dn_5": 0.20791, "loss_box_dn_5": 0.74421, "loss_dense_depth": 0.83984, "loss": 28.88213, "grad_norm": 45.06637, "time": 1.5494} -{"mode": "train", "epoch": 1, "iter": 96, "lr": 0.00014, "memory": 49163, "data_time": 0.07648, "loss_cls_0": 0.94494, "loss_box_0": 1.83991, "loss_cns_0": 0.61398, "loss_yns_0": 0.15629, "loss_cls_1": 1.02557, "loss_box_1": 1.9553, "loss_cns_1": 0.63149, "loss_yns_1": 0.15618, "loss_cls_2": 1.00924, "loss_box_2": 1.92373, "loss_cns_2": 0.64497, "loss_yns_2": 0.15503, "loss_cls_3": 1.04281, "loss_box_3": 1.92427, "loss_cns_3": 0.64723, "loss_yns_3": 0.16089, "loss_cls_4": 1.04707, "loss_box_4": 1.92932, "loss_cns_4": 0.65086, "loss_yns_4": 0.15882, "loss_cls_5": 1.06484, "loss_box_5": 1.92345, "loss_cns_5": 0.64982, "loss_yns_5": 0.15749, "loss_cls_dn_0": 0.24655, "loss_box_dn_0": 0.81237, "loss_cls_dn_1": 0.18642, "loss_box_dn_1": 0.78407, "loss_cls_dn_2": 0.19028, "loss_box_dn_2": 0.7626, "loss_cls_dn_3": 0.19102, "loss_box_dn_3": 0.76975, "loss_cls_dn_4": 0.19771, "loss_box_dn_4": 0.77271, "loss_cls_dn_5": 0.2058, "loss_box_dn_5": 0.77279, "loss_dense_depth": 0.839, "loss": 29.14454, "grad_norm": 50.76527, "time": 1.48162} -{"mode": "train", "epoch": 1, "iter": 97, "lr": 0.00014, "memory": 49163, "data_time": 0.07819, "loss_cls_0": 0.94371, "loss_box_0": 1.82424, "loss_cns_0": 0.61885, "loss_yns_0": 0.1558, "loss_cls_1": 1.00425, "loss_box_1": 1.94098, "loss_cns_1": 0.6317, "loss_yns_1": 0.15518, "loss_cls_2": 1.01655, "loss_box_2": 1.90515, "loss_cns_2": 0.63991, "loss_yns_2": 0.15429, "loss_cls_3": 1.03366, "loss_box_3": 1.92529, "loss_cns_3": 0.64973, "loss_yns_3": 0.16052, "loss_cls_4": 1.03768, "loss_box_4": 1.92546, "loss_cns_4": 0.65023, "loss_yns_4": 0.15678, "loss_cls_5": 1.05537, "loss_box_5": 1.89377, "loss_cns_5": 0.65434, "loss_yns_5": 0.15475, "loss_cls_dn_0": 0.24817, "loss_box_dn_0": 0.81305, "loss_cls_dn_1": 0.18668, "loss_box_dn_1": 0.77838, "loss_cls_dn_2": 0.19277, "loss_box_dn_2": 0.76219, "loss_cls_dn_3": 0.19469, "loss_box_dn_3": 0.77797, "loss_cls_dn_4": 0.19525, "loss_box_dn_4": 0.78353, "loss_cls_dn_5": 0.19981, "loss_box_dn_5": 0.77989, "loss_dense_depth": 0.83244, "loss": 29.03303, "grad_norm": 48.91999, "time": 1.48578} -{"mode": "train", "epoch": 1, "iter": 98, "lr": 0.00014, "memory": 49163, "data_time": 0.0793, "loss_cls_0": 0.92258, "loss_box_0": 1.8145, "loss_cns_0": 0.62456, "loss_yns_0": 0.15627, "loss_cls_1": 0.98469, "loss_box_1": 1.88785, "loss_cns_1": 0.63766, "loss_yns_1": 0.15487, "loss_cls_2": 1.01269, "loss_box_2": 1.83663, "loss_cns_2": 0.64344, "loss_yns_2": 0.15348, "loss_cls_3": 1.04203, "loss_box_3": 1.86292, "loss_cns_3": 0.65377, "loss_yns_3": 0.15674, "loss_cls_4": 1.02274, "loss_box_4": 1.85794, "loss_cns_4": 0.64895, "loss_yns_4": 0.15661, "loss_cls_5": 1.03762, "loss_box_5": 1.84087, "loss_cns_5": 0.65145, "loss_yns_5": 0.15704, "loss_cls_dn_0": 0.24184, "loss_box_dn_0": 0.81256, "loss_cls_dn_1": 0.18376, "loss_box_dn_1": 0.80163, "loss_cls_dn_2": 0.19212, "loss_box_dn_2": 0.78742, "loss_cls_dn_3": 0.19698, "loss_box_dn_3": 0.80721, "loss_cls_dn_4": 0.19369, "loss_box_dn_4": 0.81574, "loss_cls_dn_5": 0.20765, "loss_box_dn_5": 0.82007, "loss_dense_depth": 0.81311, "loss": 28.79172, "grad_norm": 39.83196, "time": 1.49092} -{"mode": "train", "epoch": 1, "iter": 99, "lr": 0.00014, "memory": 49163, "data_time": 0.07869, "loss_cls_0": 0.92343, "loss_box_0": 1.79555, "loss_cns_0": 0.62221, "loss_yns_0": 0.15484, "loss_cls_1": 1.06287, "loss_box_1": 1.93584, "loss_cns_1": 0.62858, "loss_yns_1": 0.15432, "loss_cls_2": 1.01994, "loss_box_2": 1.88079, "loss_cns_2": 0.64048, "loss_yns_2": 0.15448, "loss_cls_3": 1.05735, "loss_box_3": 1.88619, "loss_cns_3": 0.64392, "loss_yns_3": 0.15796, "loss_cls_4": 1.06896, "loss_box_4": 1.91039, "loss_cns_4": 0.64209, "loss_yns_4": 0.15596, "loss_cls_5": 1.05363, "loss_box_5": 1.89765, "loss_cns_5": 0.64595, "loss_yns_5": 0.15891, "loss_cls_dn_0": 0.24016, "loss_box_dn_0": 0.80631, "loss_cls_dn_1": 0.1837, "loss_box_dn_1": 0.80428, "loss_cls_dn_2": 0.18514, "loss_box_dn_2": 0.79101, "loss_cls_dn_3": 0.1931, "loss_box_dn_3": 0.80193, "loss_cls_dn_4": 0.19025, "loss_box_dn_4": 0.82527, "loss_cls_dn_5": 0.21404, "loss_box_dn_5": 0.82712, "loss_dense_depth": 0.84923, "loss": 29.16381, "grad_norm": 54.83968, "time": 1.48904} -{"mode": "train", "epoch": 1, "iter": 100, "lr": 0.00014, "memory": 49163, "data_time": 0.0795, "loss_cls_0": 0.91801, "loss_box_0": 1.80576, "loss_cns_0": 0.619, "loss_yns_0": 0.15474, "loss_cls_1": 1.03423, "loss_box_1": 1.96091, "loss_cns_1": 0.62651, "loss_yns_1": 0.15393, "loss_cls_2": 1.00715, "loss_box_2": 1.89737, "loss_cns_2": 0.644, "loss_yns_2": 0.15642, "loss_cls_3": 1.07981, "loss_box_3": 1.87852, "loss_cns_3": 0.64789, "loss_yns_3": 0.16245, "loss_cls_4": 1.02787, "loss_box_4": 1.93945, "loss_cns_4": 0.64816, "loss_yns_4": 0.15638, "loss_cls_5": 1.045, "loss_box_5": 1.8993, "loss_cns_5": 0.64776, "loss_yns_5": 0.15618, "loss_cls_dn_0": 0.24064, "loss_box_dn_0": 0.8172, "loss_cls_dn_1": 0.18432, "loss_box_dn_1": 0.81624, "loss_cls_dn_2": 0.1828, "loss_box_dn_2": 0.80133, "loss_cls_dn_3": 0.19302, "loss_box_dn_3": 0.80451, "loss_cls_dn_4": 0.18978, "loss_box_dn_4": 0.83946, "loss_cls_dn_5": 0.20783, "loss_box_dn_5": 0.82727, "loss_dense_depth": 0.81835, "loss": 29.18953, "grad_norm": 48.91971, "time": 1.49176} -{"mode": "train", "epoch": 1, "iter": 101, "lr": 0.00014, "memory": 49163, "data_time": 0.10538, "loss_cls_0": 0.9375, "loss_box_0": 1.77738, "loss_cns_0": 0.61413, "loss_yns_0": 0.15613, "loss_cls_1": 0.98458, "loss_box_1": 1.96138, "loss_cns_1": 0.63094, "loss_yns_1": 0.15557, "loss_cls_2": 1.0201, "loss_box_2": 1.90136, "loss_cns_2": 0.64648, "loss_yns_2": 0.15746, "loss_cls_3": 1.08604, "loss_box_3": 1.89346, "loss_cns_3": 0.65116, "loss_yns_3": 0.16388, "loss_cls_4": 1.04113, "loss_box_4": 1.93932, "loss_cns_4": 0.65216, "loss_yns_4": 0.15855, "loss_cls_5": 1.04967, "loss_box_5": 1.89608, "loss_cns_5": 0.64961, "loss_yns_5": 0.15847, "loss_cls_dn_0": 0.24233, "loss_box_dn_0": 0.80945, "loss_cls_dn_1": 0.18295, "loss_box_dn_1": 0.82312, "loss_cls_dn_2": 0.1837, "loss_box_dn_2": 0.79885, "loss_cls_dn_3": 0.19488, "loss_box_dn_3": 0.8003, "loss_cls_dn_4": 0.19416, "loss_box_dn_4": 0.82279, "loss_cls_dn_5": 0.20325, "loss_box_dn_5": 0.80314, "loss_dense_depth": 0.84615, "loss": 29.18762, "grad_norm": 56.3358, "time": 1.52752} -{"mode": "train", "epoch": 1, "iter": 102, "lr": 0.00014, "memory": 49163, "data_time": 0.10227, "loss_cls_0": 0.91845, "loss_box_0": 1.79724, "loss_cns_0": 0.61457, "loss_yns_0": 0.15824, "loss_cls_1": 0.99744, "loss_box_1": 1.9682, "loss_cns_1": 0.62753, "loss_yns_1": 0.15761, "loss_cls_2": 1.01706, "loss_box_2": 1.90268, "loss_cns_2": 0.64324, "loss_yns_2": 0.15871, "loss_cls_3": 1.04774, "loss_box_3": 1.88898, "loss_cns_3": 0.64631, "loss_yns_3": 0.16063, "loss_cls_4": 1.03682, "loss_box_4": 1.91435, "loss_cns_4": 0.64924, "loss_yns_4": 0.15907, "loss_cls_5": 1.0479, "loss_box_5": 1.89322, "loss_cns_5": 0.64915, "loss_yns_5": 0.15911, "loss_cls_dn_0": 0.244, "loss_box_dn_0": 0.81313, "loss_cls_dn_1": 0.18348, "loss_box_dn_1": 0.78008, "loss_cls_dn_2": 0.18457, "loss_box_dn_2": 0.75207, "loss_cls_dn_3": 0.19233, "loss_box_dn_3": 0.75297, "loss_cls_dn_4": 0.19625, "loss_box_dn_4": 0.76214, "loss_cls_dn_5": 0.20168, "loss_box_dn_5": 0.75407, "loss_dense_depth": 0.82109, "loss": 28.85135, "grad_norm": 29.95048, "time": 1.53721} -{"mode": "train", "epoch": 1, "iter": 103, "lr": 0.00014, "memory": 49163, "data_time": 0.0779, "loss_cls_0": 0.95278, "loss_box_0": 1.81173, "loss_cns_0": 0.62115, "loss_yns_0": 0.16081, "loss_cls_1": 0.97831, "loss_box_1": 1.97748, "loss_cns_1": 0.6308, "loss_yns_1": 0.15815, "loss_cls_2": 1.03746, "loss_box_2": 1.93035, "loss_cns_2": 0.64447, "loss_yns_2": 0.15986, "loss_cls_3": 1.08232, "loss_box_3": 1.89828, "loss_cns_3": 0.64957, "loss_yns_3": 0.15958, "loss_cls_4": 1.01802, "loss_box_4": 1.90434, "loss_cns_4": 0.64529, "loss_yns_4": 0.15937, "loss_cls_5": 1.03205, "loss_box_5": 1.94074, "loss_cns_5": 0.64618, "loss_yns_5": 0.15972, "loss_cls_dn_0": 0.23796, "loss_box_dn_0": 0.80353, "loss_cls_dn_1": 0.17845, "loss_box_dn_1": 0.7777, "loss_cls_dn_2": 0.18135, "loss_box_dn_2": 0.74759, "loss_cls_dn_3": 0.18484, "loss_box_dn_3": 0.74381, "loss_cls_dn_4": 0.18555, "loss_box_dn_4": 0.74513, "loss_cls_dn_5": 0.19402, "loss_box_dn_5": 0.76578, "loss_dense_depth": 0.88994, "loss": 28.99444, "grad_norm": 59.50528, "time": 1.55806} -{"mode": "train", "epoch": 1, "iter": 104, "lr": 0.00014, "memory": 49163, "data_time": 0.08001, "loss_cls_0": 0.91541, "loss_box_0": 1.80834, "loss_cns_0": 0.61831, "loss_yns_0": 0.15885, "loss_cls_1": 0.97612, "loss_box_1": 1.98477, "loss_cns_1": 0.63437, "loss_yns_1": 0.15562, "loss_cls_2": 1.01879, "loss_box_2": 1.94419, "loss_cns_2": 0.6476, "loss_yns_2": 0.15719, "loss_cls_3": 1.06618, "loss_box_3": 1.90735, "loss_cns_3": 0.65498, "loss_yns_3": 0.15926, "loss_cls_4": 1.00917, "loss_box_4": 1.89295, "loss_cns_4": 0.65292, "loss_yns_4": 0.15825, "loss_cls_5": 1.02542, "loss_box_5": 1.94683, "loss_cns_5": 0.65276, "loss_yns_5": 0.1585, "loss_cls_dn_0": 0.24043, "loss_box_dn_0": 0.81565, "loss_cls_dn_1": 0.17606, "loss_box_dn_1": 0.77868, "loss_cls_dn_2": 0.17832, "loss_box_dn_2": 0.74867, "loss_cls_dn_3": 0.18323, "loss_box_dn_3": 0.74501, "loss_cls_dn_4": 0.18166, "loss_box_dn_4": 0.74202, "loss_cls_dn_5": 0.19192, "loss_box_dn_5": 0.77333, "loss_dense_depth": 0.88244, "loss": 28.94153, "grad_norm": 52.5898, "time": 1.50764} -{"mode": "train", "epoch": 1, "iter": 105, "lr": 0.00014, "memory": 49163, "data_time": 0.08547, "loss_cls_0": 0.89117, "loss_box_0": 1.75469, "loss_cns_0": 0.62065, "loss_yns_0": 0.15814, "loss_cls_1": 0.98653, "loss_box_1": 1.92788, "loss_cns_1": 0.63821, "loss_yns_1": 0.15782, "loss_cls_2": 0.99137, "loss_box_2": 1.89442, "loss_cns_2": 0.64901, "loss_yns_2": 0.15634, "loss_cls_3": 0.9999, "loss_box_3": 1.87615, "loss_cns_3": 0.65103, "loss_yns_3": 0.16176, "loss_cls_4": 0.99974, "loss_box_4": 1.86929, "loss_cns_4": 0.65095, "loss_yns_4": 0.1571, "loss_cls_5": 1.01276, "loss_box_5": 1.90112, "loss_cns_5": 0.65186, "loss_yns_5": 0.16006, "loss_cls_dn_0": 0.23719, "loss_box_dn_0": 0.81236, "loss_cls_dn_1": 0.17545, "loss_box_dn_1": 0.80499, "loss_cls_dn_2": 0.17829, "loss_box_dn_2": 0.78338, "loss_cls_dn_3": 0.18436, "loss_box_dn_3": 0.78302, "loss_cls_dn_4": 0.18651, "loss_box_dn_4": 0.783, "loss_cls_dn_5": 0.19652, "loss_box_dn_5": 0.80659, "loss_dense_depth": 0.80988, "loss": 28.65947, "grad_norm": 61.65676, "time": 1.5344} -{"mode": "train", "epoch": 1, "iter": 106, "lr": 0.00014, "memory": 49163, "data_time": 0.08332, "loss_cls_0": 0.91469, "loss_box_0": 1.73089, "loss_cns_0": 0.61195, "loss_yns_0": 0.15719, "loss_cls_1": 0.99626, "loss_box_1": 1.90349, "loss_cns_1": 0.63986, "loss_yns_1": 0.15798, "loss_cls_2": 1.01591, "loss_box_2": 1.88734, "loss_cns_2": 0.64718, "loss_yns_2": 0.15696, "loss_cls_3": 1.01988, "loss_box_3": 1.86716, "loss_cns_3": 0.65045, "loss_yns_3": 0.16162, "loss_cls_4": 1.00254, "loss_box_4": 1.85495, "loss_cns_4": 0.64962, "loss_yns_4": 0.15673, "loss_cls_5": 1.01751, "loss_box_5": 1.86905, "loss_cns_5": 0.65091, "loss_yns_5": 0.15872, "loss_cls_dn_0": 0.23554, "loss_box_dn_0": 0.79561, "loss_cls_dn_1": 0.17652, "loss_box_dn_1": 0.78775, "loss_cls_dn_2": 0.1836, "loss_box_dn_2": 0.77562, "loss_cls_dn_3": 0.19036, "loss_box_dn_3": 0.77668, "loss_cls_dn_4": 0.18628, "loss_box_dn_4": 0.77721, "loss_cls_dn_5": 0.19417, "loss_box_dn_5": 0.79657, "loss_dense_depth": 0.85399, "loss": 28.60876, "grad_norm": 51.5143, "time": 1.51044} -{"mode": "train", "epoch": 1, "iter": 107, "lr": 0.00014, "memory": 49163, "data_time": 0.10869, "loss_cls_0": 0.8925, "loss_box_0": 1.76421, "loss_cns_0": 0.61288, "loss_yns_0": 0.15741, "loss_cls_1": 0.98678, "loss_box_1": 1.84228, "loss_cns_1": 0.64252, "loss_yns_1": 0.15771, "loss_cls_2": 1.01139, "loss_box_2": 1.81439, "loss_cns_2": 0.6509, "loss_yns_2": 0.15719, "loss_cls_3": 1.02717, "loss_box_3": 1.80332, "loss_cns_3": 0.65192, "loss_yns_3": 0.158, "loss_cls_4": 1.03537, "loss_box_4": 1.80609, "loss_cns_4": 0.65322, "loss_yns_4": 0.15703, "loss_cls_5": 1.05275, "loss_box_5": 1.81143, "loss_cns_5": 0.65188, "loss_yns_5": 0.15815, "loss_cls_dn_0": 0.23714, "loss_box_dn_0": 0.82235, "loss_cls_dn_1": 0.17851, "loss_box_dn_1": 0.78253, "loss_cls_dn_2": 0.18288, "loss_box_dn_2": 0.77679, "loss_cls_dn_3": 0.18743, "loss_box_dn_3": 0.79098, "loss_cls_dn_4": 0.18495, "loss_box_dn_4": 0.80409, "loss_cls_dn_5": 0.19064, "loss_box_dn_5": 0.82413, "loss_dense_depth": 0.83803, "loss": 28.45694, "grad_norm": 49.78157, "time": 1.51426} -{"mode": "train", "epoch": 1, "iter": 108, "lr": 0.00014, "memory": 49163, "data_time": 0.07125, "loss_cls_0": 0.91557, "loss_box_0": 1.79726, "loss_cns_0": 0.61791, "loss_yns_0": 0.15942, "loss_cls_1": 0.99791, "loss_box_1": 1.88594, "loss_cns_1": 0.6416, "loss_yns_1": 0.15844, "loss_cls_2": 1.00463, "loss_box_2": 1.83265, "loss_cns_2": 0.64944, "loss_yns_2": 0.15662, "loss_cls_3": 1.02326, "loss_box_3": 1.83141, "loss_cns_3": 0.64802, "loss_yns_3": 0.15817, "loss_cls_4": 1.03912, "loss_box_4": 1.83602, "loss_cns_4": 0.6494, "loss_yns_4": 0.15938, "loss_cls_5": 1.02784, "loss_box_5": 1.85164, "loss_cns_5": 0.64673, "loss_yns_5": 0.15869, "loss_cls_dn_0": 0.23821, "loss_box_dn_0": 0.82351, "loss_cls_dn_1": 0.18182, "loss_box_dn_1": 0.78927, "loss_cls_dn_2": 0.18205, "loss_box_dn_2": 0.77865, "loss_cls_dn_3": 0.18636, "loss_box_dn_3": 0.79603, "loss_cls_dn_4": 0.1902, "loss_box_dn_4": 0.81067, "loss_cls_dn_5": 0.19231, "loss_box_dn_5": 0.82839, "loss_dense_depth": 0.80455, "loss": 28.64907, "grad_norm": 61.36477, "time": 1.47079} -{"mode": "train", "epoch": 1, "iter": 109, "lr": 0.00014, "memory": 49163, "data_time": 0.08552, "loss_cls_0": 0.91408, "loss_box_0": 1.78997, "loss_cns_0": 0.61211, "loss_yns_0": 0.15805, "loss_cls_1": 1.00446, "loss_box_1": 1.93617, "loss_cns_1": 0.63927, "loss_yns_1": 0.16026, "loss_cls_2": 1.02076, "loss_box_2": 1.87681, "loss_cns_2": 0.64784, "loss_yns_2": 0.16157, "loss_cls_3": 1.03206, "loss_box_3": 1.8549, "loss_cns_3": 0.65151, "loss_yns_3": 0.15978, "loss_cls_4": 1.01473, "loss_box_4": 1.85296, "loss_cns_4": 0.64843, "loss_yns_4": 0.15972, "loss_cls_5": 1.02802, "loss_box_5": 1.87755, "loss_cns_5": 0.6481, "loss_yns_5": 0.16029, "loss_cls_dn_0": 0.2394, "loss_box_dn_0": 0.81503, "loss_cls_dn_1": 0.17553, "loss_box_dn_1": 0.80311, "loss_cls_dn_2": 0.17972, "loss_box_dn_2": 0.78202, "loss_cls_dn_3": 0.18536, "loss_box_dn_3": 0.78324, "loss_cls_dn_4": 0.18461, "loss_box_dn_4": 0.78673, "loss_cls_dn_5": 0.1899, "loss_box_dn_5": 0.79755, "loss_dense_depth": 0.85779, "loss": 28.78936, "grad_norm": 43.40875, "time": 1.4876} -{"mode": "train", "epoch": 1, "iter": 110, "lr": 0.00014, "memory": 49163, "data_time": 0.07562, "loss_cls_0": 0.91884, "loss_box_0": 1.79891, "loss_cns_0": 0.61014, "loss_yns_0": 0.15607, "loss_cls_1": 0.99236, "loss_box_1": 1.9659, "loss_cns_1": 0.63088, "loss_yns_1": 0.15826, "loss_cls_2": 1.02172, "loss_box_2": 1.89848, "loss_cns_2": 0.64122, "loss_yns_2": 0.16081, "loss_cls_3": 1.03647, "loss_box_3": 1.86818, "loss_cns_3": 0.64856, "loss_yns_3": 0.15867, "loss_cls_4": 1.03507, "loss_box_4": 1.87412, "loss_cns_4": 0.64622, "loss_yns_4": 0.15827, "loss_cls_5": 1.04151, "loss_box_5": 1.89563, "loss_cns_5": 0.64367, "loss_yns_5": 0.15838, "loss_cls_dn_0": 0.24278, "loss_box_dn_0": 0.80595, "loss_cls_dn_1": 0.17245, "loss_box_dn_1": 0.7858, "loss_cls_dn_2": 0.17708, "loss_box_dn_2": 0.75772, "loss_cls_dn_3": 0.18352, "loss_box_dn_3": 0.7478, "loss_cls_dn_4": 0.17991, "loss_box_dn_4": 0.75033, "loss_cls_dn_5": 0.18853, "loss_box_dn_5": 0.75703, "loss_dense_depth": 0.83634, "loss": 28.70359, "grad_norm": 47.89336, "time": 1.47918} -{"mode": "train", "epoch": 1, "iter": 111, "lr": 0.00014, "memory": 49163, "data_time": 0.07308, "loss_cls_0": 0.91873, "loss_box_0": 1.80466, "loss_cns_0": 0.61077, "loss_yns_0": 0.15522, "loss_cls_1": 0.98332, "loss_box_1": 1.94129, "loss_cns_1": 0.63128, "loss_yns_1": 0.15676, "loss_cls_2": 1.00422, "loss_box_2": 1.87184, "loss_cns_2": 0.64535, "loss_yns_2": 0.15929, "loss_cls_3": 1.0158, "loss_box_3": 1.86564, "loss_cns_3": 0.64702, "loss_yns_3": 0.15804, "loss_cls_4": 1.03109, "loss_box_4": 1.8648, "loss_cns_4": 0.64738, "loss_yns_4": 0.15728, "loss_cls_5": 1.02568, "loss_box_5": 1.85626, "loss_cns_5": 0.64564, "loss_yns_5": 0.15788, "loss_cls_dn_0": 0.24246, "loss_box_dn_0": 0.80457, "loss_cls_dn_1": 0.17127, "loss_box_dn_1": 0.76377, "loss_cls_dn_2": 0.17517, "loss_box_dn_2": 0.73194, "loss_cls_dn_3": 0.17998, "loss_box_dn_3": 0.72979, "loss_cls_dn_4": 0.18418, "loss_box_dn_4": 0.73291, "loss_cls_dn_5": 0.18842, "loss_box_dn_5": 0.73107, "loss_dense_depth": 0.87939, "loss": 28.47013, "grad_norm": 29.36781, "time": 1.51112} -{"mode": "train", "epoch": 1, "iter": 112, "lr": 0.00014, "memory": 49163, "data_time": 0.07413, "loss_cls_0": 0.93886, "loss_box_0": 1.82757, "loss_cns_0": 0.60958, "loss_yns_0": 0.15885, "loss_cls_1": 1.00874, "loss_box_1": 1.91875, "loss_cns_1": 0.63106, "loss_yns_1": 0.15531, "loss_cls_2": 1.02639, "loss_box_2": 1.89285, "loss_cns_2": 0.63981, "loss_yns_2": 0.15509, "loss_cls_3": 1.01595, "loss_box_3": 1.90903, "loss_cns_3": 0.64123, "loss_yns_3": 0.1563, "loss_cls_4": 1.04455, "loss_box_4": 1.90576, "loss_cns_4": 0.63941, "loss_yns_4": 0.15734, "loss_cls_5": 1.02901, "loss_box_5": 1.89258, "loss_cns_5": 0.63964, "loss_yns_5": 0.1576, "loss_cls_dn_0": 0.24762, "loss_box_dn_0": 0.80568, "loss_cls_dn_1": 0.17212, "loss_box_dn_1": 0.75125, "loss_cls_dn_2": 0.17349, "loss_box_dn_2": 0.73257, "loss_cls_dn_3": 0.17873, "loss_box_dn_3": 0.73951, "loss_cls_dn_4": 0.18791, "loss_box_dn_4": 0.74597, "loss_cls_dn_5": 0.18827, "loss_box_dn_5": 0.74732, "loss_dense_depth": 0.90085, "loss": 28.72256, "grad_norm": 51.77737, "time": 1.49521} -{"mode": "train", "epoch": 1, "iter": 113, "lr": 0.00014, "memory": 49163, "data_time": 0.07765, "loss_cls_0": 0.89988, "loss_box_0": 1.7996, "loss_cns_0": 0.6172, "loss_yns_0": 0.15671, "loss_cls_1": 0.9602, "loss_box_1": 1.87334, "loss_cns_1": 0.63261, "loss_yns_1": 0.15484, "loss_cls_2": 0.98695, "loss_box_2": 1.84455, "loss_cns_2": 0.64061, "loss_yns_2": 0.15678, "loss_cls_3": 0.98807, "loss_box_3": 1.85064, "loss_cns_3": 0.64417, "loss_yns_3": 0.15387, "loss_cls_4": 0.99764, "loss_box_4": 1.83521, "loss_cns_4": 0.64263, "loss_yns_4": 0.15501, "loss_cls_5": 0.99632, "loss_box_5": 1.82945, "loss_cns_5": 0.64403, "loss_yns_5": 0.15634, "loss_cls_dn_0": 0.2344, "loss_box_dn_0": 0.80452, "loss_cls_dn_1": 0.16623, "loss_box_dn_1": 0.77873, "loss_cls_dn_2": 0.16642, "loss_box_dn_2": 0.76417, "loss_cls_dn_3": 0.17468, "loss_box_dn_3": 0.77494, "loss_cls_dn_4": 0.18271, "loss_box_dn_4": 0.78332, "loss_cls_dn_5": 0.18884, "loss_box_dn_5": 0.7923, "loss_dense_depth": 0.898, "loss": 28.32591, "grad_norm": 42.03709, "time": 1.48991} -{"mode": "train", "epoch": 1, "iter": 114, "lr": 0.00015, "memory": 49163, "data_time": 0.076, "loss_cls_0": 0.90384, "loss_box_0": 1.81267, "loss_cns_0": 0.61781, "loss_yns_0": 0.15694, "loss_cls_1": 0.97869, "loss_box_1": 1.8548, "loss_cns_1": 0.64511, "loss_yns_1": 0.15825, "loss_cls_2": 0.98754, "loss_box_2": 1.82605, "loss_cns_2": 0.65097, "loss_yns_2": 0.162, "loss_cls_3": 1.00213, "loss_box_3": 1.83332, "loss_cns_3": 0.6474, "loss_yns_3": 0.15618, "loss_cls_4": 0.99695, "loss_box_4": 1.84215, "loss_cns_4": 0.64875, "loss_yns_4": 0.15679, "loss_cls_5": 1.00624, "loss_box_5": 1.8505, "loss_cns_5": 0.64844, "loss_yns_5": 0.15846, "loss_cls_dn_0": 0.24208, "loss_box_dn_0": 0.80011, "loss_cls_dn_1": 0.16623, "loss_box_dn_1": 0.78524, "loss_cls_dn_2": 0.16836, "loss_box_dn_2": 0.77807, "loss_cls_dn_3": 0.17423, "loss_box_dn_3": 0.79279, "loss_cls_dn_4": 0.17929, "loss_box_dn_4": 0.80904, "loss_cls_dn_5": 0.18984, "loss_box_dn_5": 0.82109, "loss_dense_depth": 0.86153, "loss": 28.46988, "grad_norm": 56.90181, "time": 1.48743} -{"mode": "train", "epoch": 1, "iter": 115, "lr": 0.00015, "memory": 49163, "data_time": 0.08622, "loss_cls_0": 0.87873, "loss_box_0": 1.77235, "loss_cns_0": 0.61894, "loss_yns_0": 0.15549, "loss_cls_1": 0.97675, "loss_box_1": 1.8856, "loss_cns_1": 0.64824, "loss_yns_1": 0.15638, "loss_cls_2": 0.97668, "loss_box_2": 1.8717, "loss_cns_2": 0.64695, "loss_yns_2": 0.1588, "loss_cls_3": 0.9985, "loss_box_3": 1.87854, "loss_cns_3": 0.64422, "loss_yns_3": 0.15743, "loss_cls_4": 0.99576, "loss_box_4": 1.89673, "loss_cns_4": 0.64704, "loss_yns_4": 0.15799, "loss_cls_5": 1.00073, "loss_box_5": 1.88894, "loss_cns_5": 0.64934, "loss_yns_5": 0.15711, "loss_cls_dn_0": 0.23469, "loss_box_dn_0": 0.80066, "loss_cls_dn_1": 0.16966, "loss_box_dn_1": 0.79904, "loss_cls_dn_2": 0.17071, "loss_box_dn_2": 0.78812, "loss_cls_dn_3": 0.1752, "loss_box_dn_3": 0.80322, "loss_cls_dn_4": 0.17834, "loss_box_dn_4": 0.82094, "loss_cls_dn_5": 0.1875, "loss_box_dn_5": 0.81897, "loss_dense_depth": 0.86338, "loss": 28.62938, "grad_norm": 55.55263, "time": 1.54281} -{"mode": "train", "epoch": 1, "iter": 116, "lr": 0.00015, "memory": 49163, "data_time": 0.07901, "loss_cls_0": 0.90499, "loss_box_0": 1.79237, "loss_cns_0": 0.617, "loss_yns_0": 0.15763, "loss_cls_1": 0.99132, "loss_box_1": 1.91812, "loss_cns_1": 0.63779, "loss_yns_1": 0.15635, "loss_cls_2": 0.99479, "loss_box_2": 1.87545, "loss_cns_2": 0.64145, "loss_yns_2": 0.15676, "loss_cls_3": 1.00824, "loss_box_3": 1.86788, "loss_cns_3": 0.64463, "loss_yns_3": 0.15816, "loss_cls_4": 1.02466, "loss_box_4": 1.88416, "loss_cns_4": 0.64722, "loss_yns_4": 0.16006, "loss_cls_5": 1.02129, "loss_box_5": 1.88828, "loss_cns_5": 0.64547, "loss_yns_5": 0.15873, "loss_cls_dn_0": 0.23823, "loss_box_dn_0": 0.79769, "loss_cls_dn_1": 0.16885, "loss_box_dn_1": 0.78611, "loss_cls_dn_2": 0.1714, "loss_box_dn_2": 0.76464, "loss_cls_dn_3": 0.17643, "loss_box_dn_3": 0.76523, "loss_cls_dn_4": 0.18109, "loss_box_dn_4": 0.77948, "loss_cls_dn_5": 0.19432, "loss_box_dn_5": 0.77897, "loss_dense_depth": 0.86723, "loss": 28.62247, "grad_norm": 40.14781, "time": 1.49925} -{"mode": "train", "epoch": 1, "iter": 117, "lr": 0.00015, "memory": 49163, "data_time": 0.07765, "loss_cls_0": 0.88399, "loss_box_0": 1.76601, "loss_cns_0": 0.62266, "loss_yns_0": 0.15504, "loss_cls_1": 0.95327, "loss_box_1": 1.95436, "loss_cns_1": 0.63211, "loss_yns_1": 0.15661, "loss_cls_2": 0.97377, "loss_box_2": 1.88498, "loss_cns_2": 0.64134, "loss_yns_2": 0.15518, "loss_cls_3": 0.98836, "loss_box_3": 1.85826, "loss_cns_3": 0.6455, "loss_yns_3": 0.15726, "loss_cls_4": 1.00131, "loss_box_4": 1.85444, "loss_cns_4": 0.64743, "loss_yns_4": 0.15593, "loss_cls_5": 1.00005, "loss_box_5": 1.88483, "loss_cns_5": 0.64456, "loss_yns_5": 0.15624, "loss_cls_dn_0": 0.23263, "loss_box_dn_0": 0.79669, "loss_cls_dn_1": 0.16807, "loss_box_dn_1": 0.79421, "loss_cls_dn_2": 0.17184, "loss_box_dn_2": 0.76834, "loss_cls_dn_3": 0.17608, "loss_box_dn_3": 0.75277, "loss_cls_dn_4": 0.1814, "loss_box_dn_4": 0.75573, "loss_cls_dn_5": 0.19001, "loss_box_dn_5": 0.7673, "loss_dense_depth": 0.83403, "loss": 28.36257, "grad_norm": 47.64043, "time": 1.48269} -{"mode": "train", "epoch": 1, "iter": 118, "lr": 0.00015, "memory": 49163, "data_time": 0.07703, "loss_cls_0": 0.90299, "loss_box_0": 1.76306, "loss_cns_0": 0.62408, "loss_yns_0": 0.15582, "loss_cls_1": 0.94751, "loss_box_1": 1.90236, "loss_cns_1": 0.63441, "loss_yns_1": 0.15527, "loss_cls_2": 0.97432, "loss_box_2": 1.85386, "loss_cns_2": 0.64115, "loss_yns_2": 0.1554, "loss_cls_3": 0.99316, "loss_box_3": 1.82099, "loss_cns_3": 0.64714, "loss_yns_3": 0.15793, "loss_cls_4": 0.98761, "loss_box_4": 1.80846, "loss_cns_4": 0.64861, "loss_yns_4": 0.15541, "loss_cls_5": 0.98514, "loss_box_5": 1.84124, "loss_cns_5": 0.65093, "loss_yns_5": 0.15662, "loss_cls_dn_0": 0.23446, "loss_box_dn_0": 0.79307, "loss_cls_dn_1": 0.1672, "loss_box_dn_1": 0.78066, "loss_cls_dn_2": 0.17151, "loss_box_dn_2": 0.75379, "loss_cls_dn_3": 0.17723, "loss_box_dn_3": 0.74129, "loss_cls_dn_4": 0.17764, "loss_box_dn_4": 0.74038, "loss_cls_dn_5": 0.18155, "loss_box_dn_5": 0.75689, "loss_dense_depth": 0.85368, "loss": 28.09283, "grad_norm": 52.12132, "time": 1.48901} -{"mode": "train", "epoch": 1, "iter": 119, "lr": 0.00015, "memory": 49163, "data_time": 0.07922, "loss_cls_0": 0.89162, "loss_box_0": 1.7645, "loss_cns_0": 0.62129, "loss_yns_0": 0.15388, "loss_cls_1": 0.95908, "loss_box_1": 1.88688, "loss_cns_1": 0.63596, "loss_yns_1": 0.15411, "loss_cls_2": 0.9768, "loss_box_2": 1.84934, "loss_cns_2": 0.64012, "loss_yns_2": 0.15759, "loss_cls_3": 0.98689, "loss_box_3": 1.83161, "loss_cns_3": 0.64578, "loss_yns_3": 0.15569, "loss_cls_4": 0.99494, "loss_box_4": 1.81318, "loss_cns_4": 0.64667, "loss_yns_4": 0.15676, "loss_cls_5": 0.9931, "loss_box_5": 1.82822, "loss_cns_5": 0.65035, "loss_yns_5": 0.15628, "loss_cls_dn_0": 0.2307, "loss_box_dn_0": 0.78421, "loss_cls_dn_1": 0.16888, "loss_box_dn_1": 0.75095, "loss_cls_dn_2": 0.17386, "loss_box_dn_2": 0.72796, "loss_cls_dn_3": 0.17482, "loss_box_dn_3": 0.72881, "loss_cls_dn_4": 0.17781, "loss_box_dn_4": 0.72852, "loss_cls_dn_5": 0.17946, "loss_box_dn_5": 0.74046, "loss_dense_depth": 0.83462, "loss": 27.9517, "grad_norm": 41.49418, "time": 1.48032} -{"mode": "train", "epoch": 1, "iter": 120, "lr": 0.00015, "memory": 49163, "data_time": 0.07636, "loss_cls_0": 0.88099, "loss_box_0": 1.77282, "loss_cns_0": 0.6153, "loss_yns_0": 0.15403, "loss_cls_1": 0.95898, "loss_box_1": 1.88644, "loss_cns_1": 0.63772, "loss_yns_1": 0.15498, "loss_cls_2": 0.97579, "loss_box_2": 1.84133, "loss_cns_2": 0.64671, "loss_yns_2": 0.16106, "loss_cls_3": 0.98741, "loss_box_3": 1.8356, "loss_cns_3": 0.64646, "loss_yns_3": 0.15632, "loss_cls_4": 0.99543, "loss_box_4": 1.82464, "loss_cns_4": 0.64547, "loss_yns_4": 0.15649, "loss_cls_5": 0.99881, "loss_box_5": 1.82105, "loss_cns_5": 0.64896, "loss_yns_5": 0.15783, "loss_cls_dn_0": 0.22948, "loss_box_dn_0": 0.79425, "loss_cls_dn_1": 0.16614, "loss_box_dn_1": 0.75874, "loss_cls_dn_2": 0.16827, "loss_box_dn_2": 0.73882, "loss_cls_dn_3": 0.16871, "loss_box_dn_3": 0.75589, "loss_cls_dn_4": 0.17311, "loss_box_dn_4": 0.76739, "loss_cls_dn_5": 0.18353, "loss_box_dn_5": 0.77765, "loss_dense_depth": 0.82037, "loss": 28.06296, "grad_norm": 49.08284, "time": 1.49218} -{"mode": "train", "epoch": 1, "iter": 121, "lr": 0.00015, "memory": 49163, "data_time": 0.10719, "loss_cls_0": 0.90649, "loss_box_0": 1.76038, "loss_cns_0": 0.61635, "loss_yns_0": 0.15583, "loss_cls_1": 0.95212, "loss_box_1": 1.95289, "loss_cns_1": 0.63662, "loss_yns_1": 0.1569, "loss_cls_2": 0.98094, "loss_box_2": 1.90437, "loss_cns_2": 0.64567, "loss_yns_2": 0.15946, "loss_cls_3": 0.99831, "loss_box_3": 1.89375, "loss_cns_3": 0.64445, "loss_yns_3": 0.15944, "loss_cls_4": 1.01138, "loss_box_4": 1.87856, "loss_cns_4": 0.64509, "loss_yns_4": 0.15653, "loss_cls_5": 1.01583, "loss_box_5": 1.88446, "loss_cns_5": 0.64659, "loss_yns_5": 0.15797, "loss_cls_dn_0": 0.23147, "loss_box_dn_0": 0.7992, "loss_cls_dn_1": 0.16142, "loss_box_dn_1": 0.77204, "loss_cls_dn_2": 0.16483, "loss_box_dn_2": 0.75878, "loss_cls_dn_3": 0.16812, "loss_box_dn_3": 0.77247, "loss_cls_dn_4": 0.1741, "loss_box_dn_4": 0.77925, "loss_cls_dn_5": 0.18054, "loss_box_dn_5": 0.79408, "loss_dense_depth": 0.82269, "loss": 28.49936, "grad_norm": 61.80978, "time": 1.55368} -{"mode": "train", "epoch": 1, "iter": 122, "lr": 0.00015, "memory": 49163, "data_time": 0.10513, "loss_cls_0": 0.90466, "loss_box_0": 1.72321, "loss_cns_0": 0.61553, "loss_yns_0": 0.15595, "loss_cls_1": 0.94795, "loss_box_1": 1.9096, "loss_cns_1": 0.63412, "loss_yns_1": 0.15817, "loss_cls_2": 0.99922, "loss_box_2": 1.82592, "loss_cns_2": 0.6459, "loss_yns_2": 0.15722, "loss_cls_3": 1.0037, "loss_box_3": 1.81558, "loss_cns_3": 0.64728, "loss_yns_3": 0.16337, "loss_cls_4": 1.01051, "loss_box_4": 1.80988, "loss_cns_4": 0.65002, "loss_yns_4": 0.15787, "loss_cls_5": 1.01205, "loss_box_5": 1.81394, "loss_cns_5": 0.64893, "loss_yns_5": 0.15852, "loss_cls_dn_0": 0.23295, "loss_box_dn_0": 0.79047, "loss_cls_dn_1": 0.16117, "loss_box_dn_1": 0.77508, "loss_cls_dn_2": 0.16525, "loss_box_dn_2": 0.75263, "loss_cls_dn_3": 0.1695, "loss_box_dn_3": 0.75525, "loss_cls_dn_4": 0.17916, "loss_box_dn_4": 0.75739, "loss_cls_dn_5": 0.17707, "loss_box_dn_5": 0.77105, "loss_dense_depth": 0.79918, "loss": 28.05526, "grad_norm": 42.33861, "time": 1.52405} -{"mode": "train", "epoch": 1, "iter": 123, "lr": 0.00015, "memory": 49163, "data_time": 0.08233, "loss_cls_0": 0.89878, "loss_box_0": 1.73768, "loss_cns_0": 0.6186, "loss_yns_0": 0.15822, "loss_cls_1": 0.94862, "loss_box_1": 1.89687, "loss_cns_1": 0.6326, "loss_yns_1": 0.1587, "loss_cls_2": 0.99287, "loss_box_2": 1.81731, "loss_cns_2": 0.64841, "loss_yns_2": 0.15764, "loss_cls_3": 0.99573, "loss_box_3": 1.82405, "loss_cns_3": 0.64731, "loss_yns_3": 0.15789, "loss_cls_4": 0.99714, "loss_box_4": 1.8423, "loss_cns_4": 0.64762, "loss_yns_4": 0.15775, "loss_cls_5": 0.99965, "loss_box_5": 1.81994, "loss_cns_5": 0.64686, "loss_yns_5": 0.15652, "loss_cls_dn_0": 0.22692, "loss_box_dn_0": 0.7872, "loss_cls_dn_1": 0.16222, "loss_box_dn_1": 0.77235, "loss_cls_dn_2": 0.16926, "loss_box_dn_2": 0.7417, "loss_cls_dn_3": 0.17059, "loss_box_dn_3": 0.7465, "loss_cls_dn_4": 0.17424, "loss_box_dn_4": 0.75772, "loss_cls_dn_5": 0.17531, "loss_box_dn_5": 0.75536, "loss_dense_depth": 0.81486, "loss": 28.01329, "grad_norm": 45.06505, "time": 1.52915} -{"mode": "train", "epoch": 1, "iter": 124, "lr": 0.00015, "memory": 49163, "data_time": 0.08133, "loss_cls_0": 0.89813, "loss_box_0": 1.77281, "loss_cns_0": 0.61973, "loss_yns_0": 0.15901, "loss_cls_1": 0.94786, "loss_box_1": 1.88913, "loss_cns_1": 0.63794, "loss_yns_1": 0.15869, "loss_cls_2": 0.99109, "loss_box_2": 1.82326, "loss_cns_2": 0.6513, "loss_yns_2": 0.16092, "loss_cls_3": 1.00009, "loss_box_3": 1.82676, "loss_cns_3": 0.65278, "loss_yns_3": 0.15443, "loss_cls_4": 1.01324, "loss_box_4": 1.86323, "loss_cns_4": 0.64942, "loss_yns_4": 0.15638, "loss_cls_5": 1.00684, "loss_box_5": 1.82981, "loss_cns_5": 0.65382, "loss_yns_5": 0.15705, "loss_cls_dn_0": 0.22625, "loss_box_dn_0": 0.78998, "loss_cls_dn_1": 0.1582, "loss_box_dn_1": 0.75924, "loss_cls_dn_2": 0.16498, "loss_box_dn_2": 0.73322, "loss_cls_dn_3": 0.16828, "loss_box_dn_3": 0.73722, "loss_cls_dn_4": 0.16755, "loss_box_dn_4": 0.75745, "loss_cls_dn_5": 0.17666, "loss_box_dn_5": 0.74886, "loss_dense_depth": 0.78474, "loss": 28.04636, "grad_norm": 50.3967, "time": 1.51572} -{"mode": "train", "epoch": 1, "iter": 125, "lr": 0.00015, "memory": 49163, "data_time": 0.09063, "loss_cls_0": 0.89828, "loss_box_0": 1.80615, "loss_cns_0": 0.61298, "loss_yns_0": 0.15841, "loss_cls_1": 0.96265, "loss_box_1": 1.82522, "loss_cns_1": 0.64278, "loss_yns_1": 0.15807, "loss_cls_2": 0.99471, "loss_box_2": 1.77556, "loss_cns_2": 0.6506, "loss_yns_2": 0.16547, "loss_cls_3": 1.01065, "loss_box_3": 1.77644, "loss_cns_3": 0.65155, "loss_yns_3": 0.15816, "loss_cls_4": 1.0182, "loss_box_4": 1.80202, "loss_cns_4": 0.64885, "loss_yns_4": 0.15862, "loss_cls_5": 1.02184, "loss_box_5": 1.79073, "loss_cns_5": 0.65784, "loss_yns_5": 0.15919, "loss_cls_dn_0": 0.23284, "loss_box_dn_0": 0.78893, "loss_cls_dn_1": 0.15963, "loss_box_dn_1": 0.75725, "loss_cls_dn_2": 0.16386, "loss_box_dn_2": 0.73466, "loss_cls_dn_3": 0.16895, "loss_box_dn_3": 0.73407, "loss_cls_dn_4": 0.1677, "loss_box_dn_4": 0.74894, "loss_cls_dn_5": 0.18, "loss_box_dn_5": 0.74744, "loss_dense_depth": 0.77991, "loss": 27.86918, "grad_norm": 45.95914, "time": 1.54794} -{"mode": "train", "epoch": 1, "iter": 126, "lr": 0.00015, "memory": 49163, "data_time": 0.089, "loss_cls_0": 0.91379, "loss_box_0": 1.76722, "loss_cns_0": 0.61535, "loss_yns_0": 0.15525, "loss_cls_1": 0.96191, "loss_box_1": 1.83713, "loss_cns_1": 0.63916, "loss_yns_1": 0.15477, "loss_cls_2": 0.9949, "loss_box_2": 1.78619, "loss_cns_2": 0.6493, "loss_yns_2": 0.15879, "loss_cls_3": 1.01327, "loss_box_3": 1.77984, "loss_cns_3": 0.64949, "loss_yns_3": 0.15813, "loss_cls_4": 1.00811, "loss_box_4": 1.78539, "loss_cns_4": 0.6487, "loss_yns_4": 0.15677, "loss_cls_5": 1.01888, "loss_box_5": 1.7941, "loss_cns_5": 0.65368, "loss_yns_5": 0.15755, "loss_cls_dn_0": 0.23329, "loss_box_dn_0": 0.79079, "loss_cls_dn_1": 0.16198, "loss_box_dn_1": 0.7556, "loss_cls_dn_2": 0.16426, "loss_box_dn_2": 0.72535, "loss_cls_dn_3": 0.16866, "loss_box_dn_3": 0.72296, "loss_cls_dn_4": 0.17148, "loss_box_dn_4": 0.72878, "loss_cls_dn_5": 0.18283, "loss_box_dn_5": 0.7401, "loss_dense_depth": 0.76907, "loss": 27.77283, "grad_norm": 40.46047, "time": 1.51489} -{"mode": "train", "epoch": 1, "iter": 127, "lr": 0.00015, "memory": 49163, "data_time": 0.11283, "loss_cls_0": 0.90312, "loss_box_0": 1.75149, "loss_cns_0": 0.61634, "loss_yns_0": 0.15511, "loss_cls_1": 0.96906, "loss_box_1": 1.79313, "loss_cns_1": 0.6412, "loss_yns_1": 0.15781, "loss_cls_2": 0.9953, "loss_box_2": 1.75729, "loss_cns_2": 0.6506, "loss_yns_2": 0.1569, "loss_cls_3": 1.00979, "loss_box_3": 1.74589, "loss_cns_3": 0.65191, "loss_yns_3": 0.15992, "loss_cls_4": 1.00878, "loss_box_4": 1.74503, "loss_cns_4": 0.65076, "loss_yns_4": 0.15496, "loss_cls_5": 1.00752, "loss_box_5": 1.76386, "loss_cns_5": 0.65148, "loss_yns_5": 0.15743, "loss_cls_dn_0": 0.23121, "loss_box_dn_0": 0.78557, "loss_cls_dn_1": 0.15932, "loss_box_dn_1": 0.74848, "loss_cls_dn_2": 0.16188, "loss_box_dn_2": 0.72672, "loss_cls_dn_3": 0.1667, "loss_box_dn_3": 0.72479, "loss_cls_dn_4": 0.17154, "loss_box_dn_4": 0.731, "loss_cls_dn_5": 0.17857, "loss_box_dn_5": 0.74324, "loss_dense_depth": 0.75913, "loss": 27.54284, "grad_norm": 49.85443, "time": 1.51636} -{"mode": "train", "epoch": 1, "iter": 128, "lr": 0.00015, "memory": 49163, "data_time": 0.08038, "loss_cls_0": 0.9175, "loss_box_0": 1.74078, "loss_cns_0": 0.62113, "loss_yns_0": 0.15864, "loss_cls_1": 0.9683, "loss_box_1": 1.80044, "loss_cns_1": 0.6389, "loss_yns_1": 0.16434, "loss_cls_2": 0.98743, "loss_box_2": 1.74948, "loss_cns_2": 0.64913, "loss_yns_2": 0.16791, "loss_cls_3": 1.00121, "loss_box_3": 1.74744, "loss_cns_3": 0.65062, "loss_yns_3": 0.16005, "loss_cls_4": 1.00178, "loss_box_4": 1.7564, "loss_cns_4": 0.64846, "loss_yns_4": 0.16474, "loss_cls_5": 1.00078, "loss_box_5": 1.77087, "loss_cns_5": 0.65341, "loss_yns_5": 0.16204, "loss_cls_dn_0": 0.22887, "loss_box_dn_0": 0.78498, "loss_cls_dn_1": 0.15688, "loss_box_dn_1": 0.73291, "loss_cls_dn_2": 0.1597, "loss_box_dn_2": 0.71284, "loss_cls_dn_3": 0.16216, "loss_box_dn_3": 0.71229, "loss_cls_dn_4": 0.16527, "loss_box_dn_4": 0.72121, "loss_cls_dn_5": 0.17245, "loss_box_dn_5": 0.73098, "loss_dense_depth": 0.7841, "loss": 27.50639, "grad_norm": 34.85564, "time": 1.48909} -{"mode": "train", "epoch": 1, "iter": 129, "lr": 0.00015, "memory": 49163, "data_time": 0.0894, "loss_cls_0": 0.88959, "loss_box_0": 1.72279, "loss_cns_0": 0.62125, "loss_yns_0": 0.15755, "loss_cls_1": 0.9539, "loss_box_1": 1.78484, "loss_cns_1": 0.64144, "loss_yns_1": 0.15931, "loss_cls_2": 0.96872, "loss_box_2": 1.73607, "loss_cns_2": 0.65005, "loss_yns_2": 0.16672, "loss_cls_3": 0.98719, "loss_box_3": 1.72277, "loss_cns_3": 0.64908, "loss_yns_3": 0.15717, "loss_cls_4": 0.99219, "loss_box_4": 1.72965, "loss_cns_4": 0.64844, "loss_yns_4": 0.1629, "loss_cls_5": 1.01073, "loss_box_5": 1.76135, "loss_cns_5": 0.65574, "loss_yns_5": 0.16208, "loss_cls_dn_0": 0.22454, "loss_box_dn_0": 0.78162, "loss_cls_dn_1": 0.15832, "loss_box_dn_1": 0.73263, "loss_cls_dn_2": 0.16012, "loss_box_dn_2": 0.71317, "loss_cls_dn_3": 0.16145, "loss_box_dn_3": 0.71271, "loss_cls_dn_4": 0.16467, "loss_box_dn_4": 0.71736, "loss_cls_dn_5": 0.17397, "loss_box_dn_5": 0.73251, "loss_dense_depth": 0.7595, "loss": 27.28406, "grad_norm": 44.46855, "time": 1.49151} -{"mode": "train", "epoch": 1, "iter": 130, "lr": 0.00015, "memory": 49163, "data_time": 0.07749, "loss_cls_0": 0.90195, "loss_box_0": 1.75524, "loss_cns_0": 0.60982, "loss_yns_0": 0.15311, "loss_cls_1": 0.96893, "loss_box_1": 1.81394, "loss_cns_1": 0.64277, "loss_yns_1": 0.15589, "loss_cls_2": 0.98718, "loss_box_2": 1.76982, "loss_cns_2": 0.65312, "loss_yns_2": 0.158, "loss_cls_3": 0.99178, "loss_box_3": 1.74424, "loss_cns_3": 0.65246, "loss_yns_3": 0.15474, "loss_cls_4": 1.00172, "loss_box_4": 1.74111, "loss_cns_4": 0.65388, "loss_yns_4": 0.15482, "loss_cls_5": 1.00864, "loss_box_5": 1.77595, "loss_cns_5": 0.66058, "loss_yns_5": 0.15656, "loss_cls_dn_0": 0.22853, "loss_box_dn_0": 0.79285, "loss_cls_dn_1": 0.15804, "loss_box_dn_1": 0.75091, "loss_cls_dn_2": 0.16233, "loss_box_dn_2": 0.73513, "loss_cls_dn_3": 0.16217, "loss_box_dn_3": 0.73167, "loss_cls_dn_4": 0.17042, "loss_box_dn_4": 0.73284, "loss_cls_dn_5": 0.17748, "loss_box_dn_5": 0.74722, "loss_dense_depth": 0.79178, "loss": 27.6076, "grad_norm": 44.89235, "time": 1.48162} -{"mode": "train", "epoch": 1, "iter": 131, "lr": 0.00015, "memory": 49163, "data_time": 0.08032, "loss_cls_0": 0.91921, "loss_box_0": 1.74904, "loss_cns_0": 0.61395, "loss_yns_0": 0.1534, "loss_cls_1": 0.96914, "loss_box_1": 1.8539, "loss_cns_1": 0.64438, "loss_yns_1": 0.15745, "loss_cls_2": 0.98373, "loss_box_2": 1.81325, "loss_cns_2": 0.65168, "loss_yns_2": 0.15378, "loss_cls_3": 0.9935, "loss_box_3": 1.78861, "loss_cns_3": 0.65481, "loss_yns_3": 0.15324, "loss_cls_4": 0.99913, "loss_box_4": 1.78106, "loss_cns_4": 0.65314, "loss_yns_4": 0.15571, "loss_cls_5": 1.00884, "loss_box_5": 1.79374, "loss_cns_5": 0.65329, "loss_yns_5": 0.15368, "loss_cls_dn_0": 0.23138, "loss_box_dn_0": 0.79955, "loss_cls_dn_1": 0.1555, "loss_box_dn_1": 0.75507, "loss_cls_dn_2": 0.15924, "loss_box_dn_2": 0.73709, "loss_cls_dn_3": 0.162, "loss_box_dn_3": 0.73089, "loss_cls_dn_4": 0.16934, "loss_box_dn_4": 0.72642, "loss_cls_dn_5": 0.17492, "loss_box_dn_5": 0.73279, "loss_dense_depth": 0.81321, "loss": 27.79904, "grad_norm": 33.09463, "time": 1.51822} -{"mode": "train", "epoch": 1, "iter": 132, "lr": 0.00015, "memory": 49163, "data_time": 0.07813, "loss_cls_0": 0.90014, "loss_box_0": 1.76635, "loss_cns_0": 0.62333, "loss_yns_0": 0.15168, "loss_cls_1": 0.99144, "loss_box_1": 1.83054, "loss_cns_1": 0.64407, "loss_yns_1": 0.15888, "loss_cls_2": 1.00807, "loss_box_2": 1.79264, "loss_cns_2": 0.64823, "loss_yns_2": 0.15358, "loss_cls_3": 1.00813, "loss_box_3": 1.78378, "loss_cns_3": 0.65115, "loss_yns_3": 0.15068, "loss_cls_4": 1.0296, "loss_box_4": 1.7863, "loss_cns_4": 0.65024, "loss_yns_4": 0.1536, "loss_cls_5": 1.00168, "loss_box_5": 1.77603, "loss_cns_5": 0.65406, "loss_yns_5": 0.15269, "loss_cls_dn_0": 0.23303, "loss_box_dn_0": 0.78541, "loss_cls_dn_1": 0.15547, "loss_box_dn_1": 0.73721, "loss_cls_dn_2": 0.15917, "loss_box_dn_2": 0.71724, "loss_cls_dn_3": 0.16561, "loss_box_dn_3": 0.71669, "loss_cls_dn_4": 0.17278, "loss_box_dn_4": 0.72048, "loss_cls_dn_5": 0.17883, "loss_box_dn_5": 0.72166, "loss_dense_depth": 0.8059, "loss": 27.73637, "grad_norm": 52.91009, "time": 1.5036} -{"mode": "train", "epoch": 1, "iter": 133, "lr": 0.00015, "memory": 49163, "data_time": 0.079, "loss_cls_0": 0.90441, "loss_box_0": 1.7505, "loss_cns_0": 0.61977, "loss_yns_0": 0.15272, "loss_cls_1": 0.97824, "loss_box_1": 1.84593, "loss_cns_1": 0.64123, "loss_yns_1": 0.15897, "loss_cls_2": 0.98743, "loss_box_2": 1.81072, "loss_cns_2": 0.64458, "loss_yns_2": 0.15665, "loss_cls_3": 1.01016, "loss_box_3": 1.80523, "loss_cns_3": 0.64504, "loss_yns_3": 0.15162, "loss_cls_4": 1.00981, "loss_box_4": 1.81661, "loss_cns_4": 0.64625, "loss_yns_4": 0.15566, "loss_cls_5": 1.02962, "loss_box_5": 1.81335, "loss_cns_5": 0.65045, "loss_yns_5": 0.15323, "loss_cls_dn_0": 0.23139, "loss_box_dn_0": 0.78791, "loss_cls_dn_1": 0.15638, "loss_box_dn_1": 0.73516, "loss_cls_dn_2": 0.15863, "loss_box_dn_2": 0.71222, "loss_cls_dn_3": 0.16546, "loss_box_dn_3": 0.71526, "loss_cls_dn_4": 0.17196, "loss_box_dn_4": 0.72662, "loss_cls_dn_5": 0.17603, "loss_box_dn_5": 0.73197, "loss_dense_depth": 0.80757, "loss": 27.81473, "grad_norm": 47.97208, "time": 1.4954} -{"mode": "train", "epoch": 1, "iter": 134, "lr": 0.00015, "memory": 49163, "data_time": 0.08085, "loss_cls_0": 0.88101, "loss_box_0": 1.72081, "loss_cns_0": 0.62382, "loss_yns_0": 0.15703, "loss_cls_1": 0.96041, "loss_box_1": 1.82454, "loss_cns_1": 0.63874, "loss_yns_1": 0.1556, "loss_cls_2": 0.98398, "loss_box_2": 1.77859, "loss_cns_2": 0.6442, "loss_yns_2": 0.15458, "loss_cls_3": 1.00033, "loss_box_3": 1.76309, "loss_cns_3": 0.64487, "loss_yns_3": 0.15157, "loss_cls_4": 0.98059, "loss_box_4": 1.77911, "loss_cns_4": 0.64858, "loss_yns_4": 0.15277, "loss_cls_5": 0.98594, "loss_box_5": 1.78217, "loss_cns_5": 0.64922, "loss_yns_5": 0.15146, "loss_cls_dn_0": 0.22362, "loss_box_dn_0": 0.78423, "loss_cls_dn_1": 0.15715, "loss_box_dn_1": 0.72382, "loss_cls_dn_2": 0.15634, "loss_box_dn_2": 0.70216, "loss_cls_dn_3": 0.16044, "loss_box_dn_3": 0.70351, "loss_cls_dn_4": 0.16519, "loss_box_dn_4": 0.71087, "loss_cls_dn_5": 0.16687, "loss_box_dn_5": 0.71797, "loss_dense_depth": 0.78336, "loss": 27.36856, "grad_norm": 36.24024, "time": 1.49798} -{"mode": "train", "epoch": 1, "iter": 135, "lr": 0.00015, "memory": 49163, "data_time": 0.08436, "loss_cls_0": 0.90364, "loss_box_0": 1.77453, "loss_cns_0": 0.62303, "loss_yns_0": 0.16101, "loss_cls_1": 0.97802, "loss_box_1": 1.86906, "loss_cns_1": 0.6423, "loss_yns_1": 0.15776, "loss_cls_2": 1.00014, "loss_box_2": 1.82915, "loss_cns_2": 0.64691, "loss_yns_2": 0.15293, "loss_cls_3": 0.99618, "loss_box_3": 1.82123, "loss_cns_3": 0.65117, "loss_yns_3": 0.15645, "loss_cls_4": 0.9967, "loss_box_4": 1.82698, "loss_cns_4": 0.64895, "loss_yns_4": 0.15408, "loss_cls_5": 0.9908, "loss_box_5": 1.82814, "loss_cns_5": 0.65114, "loss_yns_5": 0.15647, "loss_cls_dn_0": 0.23667, "loss_box_dn_0": 0.78858, "loss_cls_dn_1": 0.15794, "loss_box_dn_1": 0.73358, "loss_cls_dn_2": 0.15989, "loss_box_dn_2": 0.71706, "loss_cls_dn_3": 0.16419, "loss_box_dn_3": 0.71892, "loss_cls_dn_4": 0.17054, "loss_box_dn_4": 0.72336, "loss_cls_dn_5": 0.17569, "loss_box_dn_5": 0.72473, "loss_dense_depth": 0.82552, "loss": 27.91347, "grad_norm": 35.75413, "time": 1.5643} -{"mode": "train", "epoch": 1, "iter": 136, "lr": 0.00015, "memory": 49163, "data_time": 0.07801, "loss_cls_0": 0.86157, "loss_box_0": 1.73972, "loss_cns_0": 0.62099, "loss_yns_0": 0.15214, "loss_cls_1": 0.93516, "loss_box_1": 1.85413, "loss_cns_1": 0.64269, "loss_yns_1": 0.15718, "loss_cls_2": 0.95831, "loss_box_2": 1.79842, "loss_cns_2": 0.65194, "loss_yns_2": 0.15156, "loss_cls_3": 0.966, "loss_box_3": 1.79575, "loss_cns_3": 0.6518, "loss_yns_3": 0.15269, "loss_cls_4": 0.96436, "loss_box_4": 1.80296, "loss_cns_4": 0.65375, "loss_yns_4": 0.15349, "loss_cls_5": 0.96486, "loss_box_5": 1.79828, "loss_cns_5": 0.65392, "loss_yns_5": 0.15322, "loss_cls_dn_0": 0.23214, "loss_box_dn_0": 0.78042, "loss_cls_dn_1": 0.1527, "loss_box_dn_1": 0.72349, "loss_cls_dn_2": 0.1548, "loss_box_dn_2": 0.7083, "loss_cls_dn_3": 0.1643, "loss_box_dn_3": 0.71116, "loss_cls_dn_4": 0.16765, "loss_box_dn_4": 0.71905, "loss_cls_dn_5": 0.17387, "loss_box_dn_5": 0.71817, "loss_dense_depth": 0.78216, "loss": 27.4231, "grad_norm": 45.38873, "time": 1.48401} -{"mode": "train", "epoch": 1, "iter": 137, "lr": 0.00015, "memory": 49163, "data_time": 0.07757, "loss_cls_0": 0.87896, "loss_box_0": 1.7268, "loss_cns_0": 0.61572, "loss_yns_0": 0.15177, "loss_cls_1": 0.9298, "loss_box_1": 1.8335, "loss_cns_1": 0.64081, "loss_yns_1": 0.15249, "loss_cls_2": 0.95333, "loss_box_2": 1.78112, "loss_cns_2": 0.64881, "loss_yns_2": 0.15124, "loss_cls_3": 0.96213, "loss_box_3": 1.77059, "loss_cns_3": 0.64643, "loss_yns_3": 0.15075, "loss_cls_4": 0.96773, "loss_box_4": 1.77549, "loss_cns_4": 0.65004, "loss_yns_4": 0.15133, "loss_cls_5": 0.96482, "loss_box_5": 1.77207, "loss_cns_5": 0.64795, "loss_yns_5": 0.15061, "loss_cls_dn_0": 0.23838, "loss_box_dn_0": 0.78589, "loss_cls_dn_1": 0.14962, "loss_box_dn_1": 0.72431, "loss_cls_dn_2": 0.15208, "loss_box_dn_2": 0.70593, "loss_cls_dn_3": 0.15978, "loss_box_dn_3": 0.70584, "loss_cls_dn_4": 0.16561, "loss_box_dn_4": 0.71057, "loss_cls_dn_5": 0.16908, "loss_box_dn_5": 0.70856, "loss_dense_depth": 0.77725, "loss": 27.22721, "grad_norm": 37.3196, "time": 1.57403} -{"mode": "train", "epoch": 1, "iter": 138, "lr": 0.00015, "memory": 49163, "data_time": 0.07802, "loss_cls_0": 0.88484, "loss_box_0": 1.71152, "loss_cns_0": 0.62224, "loss_yns_0": 0.15568, "loss_cls_1": 0.93962, "loss_box_1": 1.79215, "loss_cns_1": 0.64272, "loss_yns_1": 0.14878, "loss_cls_2": 0.96807, "loss_box_2": 1.72617, "loss_cns_2": 0.64939, "loss_yns_2": 0.14822, "loss_cls_3": 0.95858, "loss_box_3": 1.71522, "loss_cns_3": 0.64928, "loss_yns_3": 0.14908, "loss_cls_4": 0.9681, "loss_box_4": 1.71855, "loss_cns_4": 0.65054, "loss_yns_4": 0.14775, "loss_cls_5": 0.96571, "loss_box_5": 1.72212, "loss_cns_5": 0.65124, "loss_yns_5": 0.14988, "loss_cls_dn_0": 0.23252, "loss_box_dn_0": 0.79261, "loss_cls_dn_1": 0.14622, "loss_box_dn_1": 0.73472, "loss_cls_dn_2": 0.14813, "loss_box_dn_2": 0.70848, "loss_cls_dn_3": 0.14883, "loss_box_dn_3": 0.70662, "loss_cls_dn_4": 0.15776, "loss_box_dn_4": 0.70618, "loss_cls_dn_5": 0.16068, "loss_box_dn_5": 0.70807, "loss_dense_depth": 0.81202, "loss": 26.99832, "grad_norm": 27.84233, "time": 1.49231} -{"mode": "train", "epoch": 1, "iter": 139, "lr": 0.00016, "memory": 49163, "data_time": 0.07809, "loss_cls_0": 0.84907, "loss_box_0": 1.70552, "loss_cns_0": 0.62549, "loss_yns_0": 0.1541, "loss_cls_1": 0.93246, "loss_box_1": 1.79631, "loss_cns_1": 0.64493, "loss_yns_1": 0.14851, "loss_cls_2": 0.96022, "loss_box_2": 1.74604, "loss_cns_2": 0.65027, "loss_yns_2": 0.14971, "loss_cls_3": 0.94142, "loss_box_3": 1.737, "loss_cns_3": 0.64968, "loss_yns_3": 0.152, "loss_cls_4": 0.94909, "loss_box_4": 1.73546, "loss_cns_4": 0.65005, "loss_yns_4": 0.1495, "loss_cls_5": 0.94113, "loss_box_5": 1.7506, "loss_cns_5": 0.64919, "loss_yns_5": 0.15174, "loss_cls_dn_0": 0.22381, "loss_box_dn_0": 0.78773, "loss_cls_dn_1": 0.14649, "loss_box_dn_1": 0.72677, "loss_cls_dn_2": 0.15025, "loss_box_dn_2": 0.70146, "loss_cls_dn_3": 0.15163, "loss_box_dn_3": 0.69903, "loss_cls_dn_4": 0.16028, "loss_box_dn_4": 0.70082, "loss_cls_dn_5": 0.1653, "loss_box_dn_5": 0.70884, "loss_dense_depth": 0.84256, "loss": 26.98446, "grad_norm": 39.56317, "time": 1.49237} -{"mode": "train", "epoch": 1, "iter": 140, "lr": 0.00016, "memory": 49163, "data_time": 0.07972, "loss_cls_0": 0.83473, "loss_box_0": 1.71373, "loss_cns_0": 0.62586, "loss_yns_0": 0.14731, "loss_cls_1": 0.91465, "loss_box_1": 1.75877, "loss_cns_1": 0.64797, "loss_yns_1": 0.14776, "loss_cls_2": 0.94002, "loss_box_2": 1.7206, "loss_cns_2": 0.65374, "loss_yns_2": 0.14733, "loss_cls_3": 0.9344, "loss_box_3": 1.7121, "loss_cns_3": 0.65357, "loss_yns_3": 0.14849, "loss_cls_4": 0.93685, "loss_box_4": 1.72334, "loss_cns_4": 0.65087, "loss_yns_4": 0.14708, "loss_cls_5": 0.93425, "loss_box_5": 1.73041, "loss_cns_5": 0.64972, "loss_yns_5": 0.14842, "loss_cls_dn_0": 0.2203, "loss_box_dn_0": 0.7944, "loss_cls_dn_1": 0.14293, "loss_box_dn_1": 0.72539, "loss_cls_dn_2": 0.14613, "loss_box_dn_2": 0.70299, "loss_cls_dn_3": 0.14924, "loss_box_dn_3": 0.70022, "loss_cls_dn_4": 0.15247, "loss_box_dn_4": 0.70742, "loss_cls_dn_5": 0.15752, "loss_box_dn_5": 0.71408, "loss_dense_depth": 0.76688, "loss": 26.70195, "grad_norm": 46.50574, "time": 1.49127} -{"mode": "train", "epoch": 1, "iter": 141, "lr": 0.00016, "memory": 49163, "data_time": 0.10702, "loss_cls_0": 0.89903, "loss_box_0": 1.76872, "loss_cns_0": 0.62737, "loss_yns_0": 0.15118, "loss_cls_1": 0.93423, "loss_box_1": 1.83156, "loss_cns_1": 0.64475, "loss_yns_1": 0.15006, "loss_cls_2": 0.95328, "loss_box_2": 1.77001, "loss_cns_2": 0.65327, "loss_yns_2": 0.14989, "loss_cls_3": 0.96033, "loss_box_3": 1.75854, "loss_cns_3": 0.65271, "loss_yns_3": 0.14993, "loss_cls_4": 0.95268, "loss_box_4": 1.76752, "loss_cns_4": 0.65253, "loss_yns_4": 0.14969, "loss_cls_5": 0.96374, "loss_box_5": 1.77002, "loss_cns_5": 0.6506, "loss_yns_5": 0.14899, "loss_cls_dn_0": 0.23394, "loss_box_dn_0": 0.7916, "loss_cls_dn_1": 0.15046, "loss_box_dn_1": 0.74247, "loss_cls_dn_2": 0.15163, "loss_box_dn_2": 0.71552, "loss_cls_dn_3": 0.15282, "loss_box_dn_3": 0.71386, "loss_cls_dn_4": 0.15619, "loss_box_dn_4": 0.71962, "loss_cls_dn_5": 0.16185, "loss_box_dn_5": 0.72532, "loss_dense_depth": 0.85177, "loss": 27.37766, "grad_norm": 34.03296, "time": 1.54501} -{"mode": "train", "epoch": 1, "iter": 142, "lr": 0.00016, "memory": 49163, "data_time": 0.10353, "loss_cls_0": 0.91791, "loss_box_0": 1.79225, "loss_cns_0": 0.61981, "loss_yns_0": 0.15904, "loss_cls_1": 0.96682, "loss_box_1": 1.86565, "loss_cns_1": 0.63872, "loss_yns_1": 0.15055, "loss_cls_2": 0.9874, "loss_box_2": 1.81786, "loss_cns_2": 0.64609, "loss_yns_2": 0.15169, "loss_cls_3": 0.97439, "loss_box_3": 1.81257, "loss_cns_3": 0.64594, "loss_yns_3": 0.15134, "loss_cls_4": 0.96742, "loss_box_4": 1.81855, "loss_cns_4": 0.64572, "loss_yns_4": 0.15181, "loss_cls_5": 0.98372, "loss_box_5": 1.82244, "loss_cns_5": 0.64445, "loss_yns_5": 0.15287, "loss_cls_dn_0": 0.23365, "loss_box_dn_0": 0.78639, "loss_cls_dn_1": 0.14915, "loss_box_dn_1": 0.7387, "loss_cls_dn_2": 0.14979, "loss_box_dn_2": 0.71566, "loss_cls_dn_3": 0.15411, "loss_box_dn_3": 0.7199, "loss_cls_dn_4": 0.15814, "loss_box_dn_4": 0.72268, "loss_cls_dn_5": 0.16506, "loss_box_dn_5": 0.73032, "loss_dense_depth": 0.81431, "loss": 27.72285, "grad_norm": 40.07326, "time": 1.59397} -{"mode": "train", "epoch": 1, "iter": 143, "lr": 0.00016, "memory": 49163, "data_time": 0.07579, "loss_cls_0": 0.85929, "loss_box_0": 1.77402, "loss_cns_0": 0.62005, "loss_yns_0": 0.1554, "loss_cls_1": 0.94525, "loss_box_1": 1.8364, "loss_cns_1": 0.64381, "loss_yns_1": 0.14937, "loss_cls_2": 0.96863, "loss_box_2": 1.79814, "loss_cns_2": 0.64799, "loss_yns_2": 0.15052, "loss_cls_3": 0.95999, "loss_box_3": 1.77845, "loss_cns_3": 0.64914, "loss_yns_3": 0.15242, "loss_cls_4": 0.95343, "loss_box_4": 1.78365, "loss_cns_4": 0.64847, "loss_yns_4": 0.15035, "loss_cls_5": 0.96491, "loss_box_5": 1.79754, "loss_cns_5": 0.64979, "loss_yns_5": 0.15142, "loss_cls_dn_0": 0.2231, "loss_box_dn_0": 0.78259, "loss_cls_dn_1": 0.15039, "loss_box_dn_1": 0.74376, "loss_cls_dn_2": 0.14903, "loss_box_dn_2": 0.72428, "loss_cls_dn_3": 0.15847, "loss_box_dn_3": 0.72354, "loss_cls_dn_4": 0.16124, "loss_box_dn_4": 0.72604, "loss_cls_dn_5": 0.16687, "loss_box_dn_5": 0.7349, "loss_dense_depth": 0.79244, "loss": 27.42504, "grad_norm": 43.8141, "time": 1.5307} -{"mode": "train", "epoch": 1, "iter": 144, "lr": 0.00016, "memory": 49163, "data_time": 0.07941, "loss_cls_0": 0.87524, "loss_box_0": 1.77789, "loss_cns_0": 0.61992, "loss_yns_0": 0.15133, "loss_cls_1": 0.93523, "loss_box_1": 1.85809, "loss_cns_1": 0.64037, "loss_yns_1": 0.14857, "loss_cls_2": 0.96909, "loss_box_2": 1.80974, "loss_cns_2": 0.64938, "loss_yns_2": 0.14897, "loss_cls_3": 0.97082, "loss_box_3": 1.80059, "loss_cns_3": 0.64921, "loss_yns_3": 0.1509, "loss_cls_4": 0.96461, "loss_box_4": 1.79349, "loss_cns_4": 0.64967, "loss_yns_4": 0.14887, "loss_cls_5": 0.97386, "loss_box_5": 1.81606, "loss_cns_5": 0.6479, "loss_yns_5": 0.14859, "loss_cls_dn_0": 0.23038, "loss_box_dn_0": 0.78624, "loss_cls_dn_1": 0.15622, "loss_box_dn_1": 0.75565, "loss_cls_dn_2": 0.15446, "loss_box_dn_2": 0.73893, "loss_cls_dn_3": 0.16265, "loss_box_dn_3": 0.73653, "loss_cls_dn_4": 0.16542, "loss_box_dn_4": 0.7386, "loss_cls_dn_5": 0.16875, "loss_box_dn_5": 0.74822, "loss_dense_depth": 0.78512, "loss": 27.62558, "grad_norm": 48.76894, "time": 1.51469} -{"mode": "train", "epoch": 1, "iter": 145, "lr": 0.00016, "memory": 49163, "data_time": 0.08654, "loss_cls_0": 0.87667, "loss_box_0": 1.78248, "loss_cns_0": 0.62154, "loss_yns_0": 0.14996, "loss_cls_1": 0.92322, "loss_box_1": 1.86306, "loss_cns_1": 0.63971, "loss_yns_1": 0.14958, "loss_cls_2": 0.95383, "loss_box_2": 1.79746, "loss_cns_2": 0.651, "loss_yns_2": 0.14999, "loss_cls_3": 0.95536, "loss_box_3": 1.80136, "loss_cns_3": 0.64897, "loss_yns_3": 0.15145, "loss_cls_4": 0.9538, "loss_box_4": 1.78149, "loss_cns_4": 0.65094, "loss_yns_4": 0.14979, "loss_cls_5": 0.95888, "loss_box_5": 1.79704, "loss_cns_5": 0.64764, "loss_yns_5": 0.14944, "loss_cls_dn_0": 0.22592, "loss_box_dn_0": 0.78185, "loss_cls_dn_1": 0.15572, "loss_box_dn_1": 0.76496, "loss_cls_dn_2": 0.15317, "loss_box_dn_2": 0.73588, "loss_cls_dn_3": 0.15682, "loss_box_dn_3": 0.73409, "loss_cls_dn_4": 0.16133, "loss_box_dn_4": 0.73353, "loss_cls_dn_5": 0.16684, "loss_box_dn_5": 0.7368, "loss_dense_depth": 0.796, "loss": 27.50757, "grad_norm": 46.01843, "time": 1.50808} -{"mode": "train", "epoch": 1, "iter": 146, "lr": 0.00016, "memory": 49163, "data_time": 0.08446, "loss_cls_0": 0.86606, "loss_box_0": 1.78346, "loss_cns_0": 0.61972, "loss_yns_0": 0.14965, "loss_cls_1": 0.93466, "loss_box_1": 1.8431, "loss_cns_1": 0.64096, "loss_yns_1": 0.14787, "loss_cls_2": 0.95195, "loss_box_2": 1.77449, "loss_cns_2": 0.65064, "loss_yns_2": 0.14869, "loss_cls_3": 0.96042, "loss_box_3": 1.76963, "loss_cns_3": 0.65775, "loss_yns_3": 0.148, "loss_cls_4": 0.95387, "loss_box_4": 1.77084, "loss_cns_4": 0.65257, "loss_yns_4": 0.14857, "loss_cls_5": 0.96258, "loss_box_5": 1.78463, "loss_cns_5": 0.65209, "loss_yns_5": 0.14915, "loss_cls_dn_0": 0.22284, "loss_box_dn_0": 0.78458, "loss_cls_dn_1": 0.14932, "loss_box_dn_1": 0.74544, "loss_cls_dn_2": 0.14806, "loss_box_dn_2": 0.72077, "loss_cls_dn_3": 0.15052, "loss_box_dn_3": 0.72213, "loss_cls_dn_4": 0.15664, "loss_box_dn_4": 0.72964, "loss_cls_dn_5": 0.1656, "loss_box_dn_5": 0.73755, "loss_dense_depth": 0.77063, "loss": 27.32505, "grad_norm": 36.13913, "time": 1.50843} -{"mode": "train", "epoch": 1, "iter": 147, "lr": 0.00016, "memory": 49163, "data_time": 0.10965, "loss_cls_0": 0.87903, "loss_box_0": 1.78193, "loss_cns_0": 0.61913, "loss_yns_0": 0.15398, "loss_cls_1": 0.94671, "loss_box_1": 1.81316, "loss_cns_1": 0.64464, "loss_yns_1": 0.15019, "loss_cls_2": 0.96533, "loss_box_2": 1.76971, "loss_cns_2": 0.65076, "loss_yns_2": 0.15022, "loss_cls_3": 0.97006, "loss_box_3": 1.76655, "loss_cns_3": 0.66024, "loss_yns_3": 0.15071, "loss_cls_4": 0.96595, "loss_box_4": 1.77867, "loss_cns_4": 0.65263, "loss_yns_4": 0.15063, "loss_cls_5": 0.97626, "loss_box_5": 1.79775, "loss_cns_5": 0.65248, "loss_yns_5": 0.15199, "loss_cls_dn_0": 0.22476, "loss_box_dn_0": 0.77982, "loss_cls_dn_1": 0.14871, "loss_box_dn_1": 0.73391, "loss_cls_dn_2": 0.15051, "loss_box_dn_2": 0.72176, "loss_cls_dn_3": 0.15162, "loss_box_dn_3": 0.72481, "loss_cls_dn_4": 0.15777, "loss_box_dn_4": 0.73484, "loss_cls_dn_5": 0.16547, "loss_box_dn_5": 0.7487, "loss_dense_depth": 0.79903, "loss": 27.44044, "grad_norm": 44.20379, "time": 1.50759} -{"mode": "train", "epoch": 1, "iter": 148, "lr": 0.00016, "memory": 49163, "data_time": 0.08137, "loss_cls_0": 0.85229, "loss_box_0": 1.76791, "loss_cns_0": 0.61833, "loss_yns_0": 0.15299, "loss_cls_1": 0.93756, "loss_box_1": 1.78489, "loss_cns_1": 0.64777, "loss_yns_1": 0.14696, "loss_cls_2": 0.95903, "loss_box_2": 1.73467, "loss_cns_2": 0.65169, "loss_yns_2": 0.1479, "loss_cls_3": 0.9571, "loss_box_3": 1.73349, "loss_cns_3": 0.65513, "loss_yns_3": 0.14803, "loss_cls_4": 0.94521, "loss_box_4": 1.73621, "loss_cns_4": 0.65282, "loss_yns_4": 0.14864, "loss_cls_5": 0.95489, "loss_box_5": 1.73444, "loss_cns_5": 0.6534, "loss_yns_5": 0.1497, "loss_cls_dn_0": 0.22015, "loss_box_dn_0": 0.78101, "loss_cls_dn_1": 0.14702, "loss_box_dn_1": 0.73623, "loss_cls_dn_2": 0.14907, "loss_box_dn_2": 0.7203, "loss_cls_dn_3": 0.14996, "loss_box_dn_3": 0.72395, "loss_cls_dn_4": 0.15527, "loss_box_dn_4": 0.73064, "loss_cls_dn_5": 0.1618, "loss_box_dn_5": 0.73781, "loss_dense_depth": 0.75315, "loss": 27.03743, "grad_norm": 35.55965, "time": 1.49697} -{"mode": "train", "epoch": 1, "iter": 149, "lr": 0.00016, "memory": 49163, "data_time": 0.09322, "loss_cls_0": 0.85314, "loss_box_0": 1.76386, "loss_cns_0": 0.61923, "loss_yns_0": 0.14982, "loss_cls_1": 0.92682, "loss_box_1": 1.78228, "loss_cns_1": 0.64556, "loss_yns_1": 0.14881, "loss_cls_2": 0.95054, "loss_box_2": 1.72929, "loss_cns_2": 0.64922, "loss_yns_2": 0.14796, "loss_cls_3": 0.95493, "loss_box_3": 1.72285, "loss_cns_3": 0.64842, "loss_yns_3": 0.14803, "loss_cls_4": 0.94713, "loss_box_4": 1.73243, "loss_cns_4": 0.64786, "loss_yns_4": 0.14784, "loss_cls_5": 0.96026, "loss_box_5": 1.72859, "loss_cns_5": 0.64851, "loss_yns_5": 0.1485, "loss_cls_dn_0": 0.21925, "loss_box_dn_0": 0.77705, "loss_cls_dn_1": 0.14427, "loss_box_dn_1": 0.73893, "loss_cls_dn_2": 0.14618, "loss_box_dn_2": 0.71528, "loss_cls_dn_3": 0.14705, "loss_box_dn_3": 0.71019, "loss_cls_dn_4": 0.15102, "loss_box_dn_4": 0.71732, "loss_cls_dn_5": 0.15708, "loss_box_dn_5": 0.72153, "loss_dense_depth": 0.75597, "loss": 26.90301, "grad_norm": 36.37938, "time": 1.5163} -{"mode": "train", "epoch": 1, "iter": 150, "lr": 0.00016, "memory": 49163, "data_time": 0.076, "loss_cls_0": 0.8738, "loss_box_0": 1.75039, "loss_cns_0": 0.62004, "loss_yns_0": 0.14901, "loss_cls_1": 0.92044, "loss_box_1": 1.78977, "loss_cns_1": 0.64686, "loss_yns_1": 0.14741, "loss_cls_2": 0.95892, "loss_box_2": 1.72879, "loss_cns_2": 0.65489, "loss_yns_2": 0.14571, "loss_cls_3": 0.96508, "loss_box_3": 1.72507, "loss_cns_3": 0.65411, "loss_yns_3": 0.1463, "loss_cls_4": 0.95825, "loss_box_4": 1.71924, "loss_cns_4": 0.65499, "loss_yns_4": 0.14648, "loss_cls_5": 0.96459, "loss_box_5": 1.73036, "loss_cns_5": 0.65605, "loss_yns_5": 0.14647, "loss_cls_dn_0": 0.22007, "loss_box_dn_0": 0.7747, "loss_cls_dn_1": 0.14496, "loss_box_dn_1": 0.7165, "loss_cls_dn_2": 0.14957, "loss_box_dn_2": 0.69171, "loss_cls_dn_3": 0.15068, "loss_box_dn_3": 0.6906, "loss_cls_dn_4": 0.15454, "loss_box_dn_4": 0.69008, "loss_cls_dn_5": 0.15938, "loss_box_dn_5": 0.70083, "loss_dense_depth": 0.76921, "loss": 26.86588, "grad_norm": 30.0119, "time": 1.49552} -{"mode": "train", "epoch": 1, "iter": 151, "lr": 0.00016, "memory": 49163, "data_time": 0.07938, "loss_cls_0": 0.87596, "loss_box_0": 1.74467, "loss_cns_0": 0.62176, "loss_yns_0": 0.14982, "loss_cls_1": 0.93583, "loss_box_1": 1.79271, "loss_cns_1": 0.64475, "loss_yns_1": 0.14937, "loss_cls_2": 0.95454, "loss_box_2": 1.74422, "loss_cns_2": 0.65107, "loss_yns_2": 0.14737, "loss_cls_3": 0.96146, "loss_box_3": 1.74688, "loss_cns_3": 0.65376, "loss_yns_3": 0.1472, "loss_cls_4": 0.95981, "loss_box_4": 1.73769, "loss_cns_4": 0.65332, "loss_yns_4": 0.1478, "loss_cls_5": 0.96783, "loss_box_5": 1.73738, "loss_cns_5": 0.653, "loss_yns_5": 0.14929, "loss_cls_dn_0": 0.21911, "loss_box_dn_0": 0.77863, "loss_cls_dn_1": 0.14264, "loss_box_dn_1": 0.72227, "loss_cls_dn_2": 0.1453, "loss_box_dn_2": 0.69842, "loss_cls_dn_3": 0.1457, "loss_box_dn_3": 0.70286, "loss_cls_dn_4": 0.14929, "loss_box_dn_4": 0.70163, "loss_cls_dn_5": 0.1576, "loss_box_dn_5": 0.70971, "loss_dense_depth": 0.77287, "loss": 26.97349, "grad_norm": 34.41307, "time": 1.50044} -{"mode": "train", "epoch": 1, "iter": 152, "lr": 0.00016, "memory": 49163, "data_time": 0.07627, "loss_cls_0": 0.86251, "loss_box_0": 1.71454, "loss_cns_0": 0.62059, "loss_yns_0": 0.15342, "loss_cls_1": 0.94298, "loss_box_1": 1.81243, "loss_cns_1": 0.64404, "loss_yns_1": 0.15053, "loss_cls_2": 0.95557, "loss_box_2": 1.76877, "loss_cns_2": 0.64761, "loss_yns_2": 0.15049, "loss_cls_3": 0.95653, "loss_box_3": 1.76852, "loss_cns_3": 0.65654, "loss_yns_3": 0.14999, "loss_cls_4": 0.96947, "loss_box_4": 1.77655, "loss_cns_4": 0.65385, "loss_yns_4": 0.15203, "loss_cls_5": 0.96862, "loss_box_5": 1.78801, "loss_cns_5": 0.65354, "loss_yns_5": 0.15229, "loss_cls_dn_0": 0.21735, "loss_box_dn_0": 0.77421, "loss_cls_dn_1": 0.14382, "loss_box_dn_1": 0.73019, "loss_cls_dn_2": 0.14303, "loss_box_dn_2": 0.71132, "loss_cls_dn_3": 0.14444, "loss_box_dn_3": 0.71645, "loss_cls_dn_4": 0.15007, "loss_box_dn_4": 0.72705, "loss_cls_dn_5": 0.15981, "loss_box_dn_5": 0.73972, "loss_dense_depth": 0.74301, "loss": 27.1699, "grad_norm": 36.33428, "time": 1.49146} -{"mode": "train", "epoch": 1, "iter": 153, "lr": 0.00016, "memory": 49163, "data_time": 0.07598, "loss_cls_0": 0.89187, "loss_box_0": 1.71937, "loss_cns_0": 0.61502, "loss_yns_0": 0.14995, "loss_cls_1": 0.93178, "loss_box_1": 1.81281, "loss_cns_1": 0.64386, "loss_yns_1": 0.15163, "loss_cls_2": 0.9637, "loss_box_2": 1.76077, "loss_cns_2": 0.65076, "loss_yns_2": 0.15238, "loss_cls_3": 0.96422, "loss_box_3": 1.75746, "loss_cns_3": 0.65468, "loss_yns_3": 0.15118, "loss_cls_4": 0.95824, "loss_box_4": 1.76182, "loss_cns_4": 0.65067, "loss_yns_4": 0.15192, "loss_cls_5": 0.97959, "loss_box_5": 1.76917, "loss_cns_5": 0.65189, "loss_yns_5": 0.15126, "loss_cls_dn_0": 0.22724, "loss_box_dn_0": 0.78267, "loss_cls_dn_1": 0.14156, "loss_box_dn_1": 0.74621, "loss_cls_dn_2": 0.14251, "loss_box_dn_2": 0.72938, "loss_cls_dn_3": 0.14516, "loss_box_dn_3": 0.73777, "loss_cls_dn_4": 0.15066, "loss_box_dn_4": 0.75227, "loss_cls_dn_5": 0.16167, "loss_box_dn_5": 0.76583, "loss_dense_depth": 0.76457, "loss": 27.29347, "grad_norm": 39.89338, "time": 1.49406} -{"mode": "train", "epoch": 1, "iter": 154, "lr": 0.00016, "memory": 49163, "data_time": 0.07875, "loss_cls_0": 0.85918, "loss_box_0": 1.70912, "loss_cns_0": 0.61842, "loss_yns_0": 0.15017, "loss_cls_1": 0.90879, "loss_box_1": 1.74535, "loss_cns_1": 0.64479, "loss_yns_1": 0.14979, "loss_cls_2": 0.93185, "loss_box_2": 1.69343, "loss_cns_2": 0.65058, "loss_yns_2": 0.14943, "loss_cls_3": 0.93483, "loss_box_3": 1.69949, "loss_cns_3": 0.65228, "loss_yns_3": 0.15109, "loss_cls_4": 0.94872, "loss_box_4": 1.69662, "loss_cns_4": 0.6513, "loss_yns_4": 0.14979, "loss_cls_5": 0.95085, "loss_box_5": 1.69234, "loss_cns_5": 0.65329, "loss_yns_5": 0.15052, "loss_cls_dn_0": 0.22024, "loss_box_dn_0": 0.77404, "loss_cls_dn_1": 0.13936, "loss_box_dn_1": 0.74957, "loss_cls_dn_2": 0.14114, "loss_box_dn_2": 0.72457, "loss_cls_dn_3": 0.14291, "loss_box_dn_3": 0.73577, "loss_cls_dn_4": 0.14747, "loss_box_dn_4": 0.74422, "loss_cls_dn_5": 0.15299, "loss_box_dn_5": 0.74996, "loss_dense_depth": 0.74834, "loss": 26.7126, "grad_norm": 41.9203, "time": 1.49884} -{"mode": "train", "epoch": 1, "iter": 155, "lr": 0.00016, "memory": 49163, "data_time": 0.08379, "loss_cls_0": 0.85951, "loss_box_0": 1.71358, "loss_cns_0": 0.62297, "loss_yns_0": 0.15318, "loss_cls_1": 0.90663, "loss_box_1": 1.69678, "loss_cns_1": 0.65063, "loss_yns_1": 0.14949, "loss_cls_2": 0.94228, "loss_box_2": 1.66652, "loss_cns_2": 0.65444, "loss_yns_2": 0.14875, "loss_cls_3": 0.93899, "loss_box_3": 1.65883, "loss_cns_3": 0.6618, "loss_yns_3": 0.14958, "loss_cls_4": 0.94708, "loss_box_4": 1.64977, "loss_cns_4": 0.66317, "loss_yns_4": 0.14858, "loss_cls_5": 0.92665, "loss_box_5": 1.65464, "loss_cns_5": 0.65754, "loss_yns_5": 0.15066, "loss_cls_dn_0": 0.21529, "loss_box_dn_0": 0.7788, "loss_cls_dn_1": 0.13738, "loss_box_dn_1": 0.74231, "loss_cls_dn_2": 0.1402, "loss_box_dn_2": 0.722, "loss_cls_dn_3": 0.14277, "loss_box_dn_3": 0.72405, "loss_cls_dn_4": 0.14601, "loss_box_dn_4": 0.7249, "loss_cls_dn_5": 0.14901, "loss_box_dn_5": 0.73027, "loss_dense_depth": 0.73668, "loss": 26.46168, "grad_norm": 35.80515, "time": 1.56738} -{"mode": "train", "epoch": 1, "iter": 156, "lr": 0.00016, "memory": 49163, "data_time": 0.07697, "loss_cls_0": 0.84278, "loss_box_0": 1.72741, "loss_cns_0": 0.6171, "loss_yns_0": 0.15289, "loss_cls_1": 0.91455, "loss_box_1": 1.70565, "loss_cns_1": 0.64734, "loss_yns_1": 0.15107, "loss_cls_2": 0.92785, "loss_box_2": 1.67686, "loss_cns_2": 0.65209, "loss_yns_2": 0.15283, "loss_cls_3": 0.92753, "loss_box_3": 1.6673, "loss_cns_3": 0.66034, "loss_yns_3": 0.1519, "loss_cls_4": 0.94541, "loss_box_4": 1.66884, "loss_cns_4": 0.65748, "loss_yns_4": 0.15154, "loss_cls_5": 0.93359, "loss_box_5": 1.67277, "loss_cns_5": 0.65503, "loss_yns_5": 0.15089, "loss_cls_dn_0": 0.22117, "loss_box_dn_0": 0.77801, "loss_cls_dn_1": 0.13847, "loss_box_dn_1": 0.7538, "loss_cls_dn_2": 0.13904, "loss_box_dn_2": 0.7386, "loss_cls_dn_3": 0.14332, "loss_box_dn_3": 0.73548, "loss_cls_dn_4": 0.15187, "loss_box_dn_4": 0.73645, "loss_cls_dn_5": 0.15304, "loss_box_dn_5": 0.74452, "loss_dense_depth": 0.75785, "loss": 26.60266, "grad_norm": 34.97077, "time": 1.48269} -{"mode": "train", "epoch": 1, "iter": 157, "lr": 0.00016, "memory": 49163, "data_time": 0.07566, "loss_cls_0": 0.85193, "loss_box_0": 1.71475, "loss_cns_0": 0.61867, "loss_yns_0": 0.15334, "loss_cls_1": 0.91525, "loss_box_1": 1.74067, "loss_cns_1": 0.64549, "loss_yns_1": 0.15136, "loss_cls_2": 0.93056, "loss_box_2": 1.69601, "loss_cns_2": 0.64924, "loss_yns_2": 0.15257, "loss_cls_3": 0.92931, "loss_box_3": 1.69608, "loss_cns_3": 0.65207, "loss_yns_3": 0.15275, "loss_cls_4": 0.9317, "loss_box_4": 1.69481, "loss_cns_4": 0.64916, "loss_yns_4": 0.15195, "loss_cls_5": 0.93784, "loss_box_5": 1.69861, "loss_cns_5": 0.64948, "loss_yns_5": 0.15278, "loss_cls_dn_0": 0.21521, "loss_box_dn_0": 0.77305, "loss_cls_dn_1": 0.13971, "loss_box_dn_1": 0.73334, "loss_cls_dn_2": 0.13864, "loss_box_dn_2": 0.71656, "loss_cls_dn_3": 0.14243, "loss_box_dn_3": 0.71852, "loss_cls_dn_4": 0.14756, "loss_box_dn_4": 0.72083, "loss_cls_dn_5": 0.15145, "loss_box_dn_5": 0.72808, "loss_dense_depth": 0.73825, "loss": 26.58001, "grad_norm": 37.5921, "time": 1.53882} -{"mode": "train", "epoch": 1, "iter": 158, "lr": 0.00016, "memory": 49163, "data_time": 0.07479, "loss_cls_0": 0.83529, "loss_box_0": 1.73198, "loss_cns_0": 0.61772, "loss_yns_0": 0.15388, "loss_cls_1": 0.91179, "loss_box_1": 1.74548, "loss_cns_1": 0.64718, "loss_yns_1": 0.15422, "loss_cls_2": 0.93481, "loss_box_2": 1.69898, "loss_cns_2": 0.65111, "loss_yns_2": 0.15404, "loss_cls_3": 0.93063, "loss_box_3": 1.69034, "loss_cns_3": 0.6524, "loss_yns_3": 0.15397, "loss_cls_4": 0.93883, "loss_box_4": 1.69405, "loss_cns_4": 0.65101, "loss_yns_4": 0.15279, "loss_cls_5": 0.93626, "loss_box_5": 1.7024, "loss_cns_5": 0.65222, "loss_yns_5": 0.15389, "loss_cls_dn_0": 0.20782, "loss_box_dn_0": 0.7759, "loss_cls_dn_1": 0.13923, "loss_box_dn_1": 0.74186, "loss_cls_dn_2": 0.13902, "loss_box_dn_2": 0.72498, "loss_cls_dn_3": 0.13897, "loss_box_dn_3": 0.72953, "loss_cls_dn_4": 0.14451, "loss_box_dn_4": 0.73504, "loss_cls_dn_5": 0.14817, "loss_box_dn_5": 0.74689, "loss_dense_depth": 0.72105, "loss": 26.63824, "grad_norm": 36.32011, "time": 1.49238} -{"mode": "train", "epoch": 1, "iter": 159, "lr": 0.00016, "memory": 49163, "data_time": 0.08004, "loss_cls_0": 0.84093, "loss_box_0": 1.74268, "loss_cns_0": 0.61859, "loss_yns_0": 0.15447, "loss_cls_1": 0.91317, "loss_box_1": 1.76444, "loss_cns_1": 0.64942, "loss_yns_1": 0.1553, "loss_cls_2": 0.94093, "loss_box_2": 1.71543, "loss_cns_2": 0.65415, "loss_yns_2": 0.15389, "loss_cls_3": 0.93979, "loss_box_3": 1.70969, "loss_cns_3": 0.65624, "loss_yns_3": 0.15292, "loss_cls_4": 0.94084, "loss_box_4": 1.70899, "loss_cns_4": 0.65504, "loss_yns_4": 0.15298, "loss_cls_5": 0.93471, "loss_box_5": 1.74038, "loss_cns_5": 0.65477, "loss_yns_5": 0.15219, "loss_cls_dn_0": 0.21034, "loss_box_dn_0": 0.77574, "loss_cls_dn_1": 0.13672, "loss_box_dn_1": 0.75434, "loss_cls_dn_2": 0.13928, "loss_box_dn_2": 0.73546, "loss_cls_dn_3": 0.14057, "loss_box_dn_3": 0.74176, "loss_cls_dn_4": 0.14668, "loss_box_dn_4": 0.75055, "loss_cls_dn_5": 0.15277, "loss_box_dn_5": 0.76798, "loss_dense_depth": 0.73525, "loss": 26.88937, "grad_norm": 45.07962, "time": 1.49724} -{"mode": "train", "epoch": 1, "iter": 160, "lr": 0.00016, "memory": 49163, "data_time": 0.08268, "loss_cls_0": 0.83358, "loss_box_0": 1.72383, "loss_cns_0": 0.61908, "loss_yns_0": 0.15389, "loss_cls_1": 0.89955, "loss_box_1": 1.76648, "loss_cns_1": 0.64706, "loss_yns_1": 0.15408, "loss_cls_2": 0.92289, "loss_box_2": 1.72714, "loss_cns_2": 0.65355, "loss_yns_2": 0.15416, "loss_cls_3": 0.93159, "loss_box_3": 1.70928, "loss_cns_3": 0.657, "loss_yns_3": 0.15403, "loss_cls_4": 0.93298, "loss_box_4": 1.70654, "loss_cns_4": 0.65968, "loss_yns_4": 0.15349, "loss_cls_5": 0.92654, "loss_box_5": 1.72027, "loss_cns_5": 0.65597, "loss_yns_5": 0.15428, "loss_cls_dn_0": 0.20808, "loss_box_dn_0": 0.77169, "loss_cls_dn_1": 0.13801, "loss_box_dn_1": 0.76224, "loss_cls_dn_2": 0.13988, "loss_box_dn_2": 0.74866, "loss_cls_dn_3": 0.14094, "loss_box_dn_3": 0.75226, "loss_cls_dn_4": 0.1494, "loss_box_dn_4": 0.75963, "loss_cls_dn_5": 0.15137, "loss_box_dn_5": 0.76608, "loss_dense_depth": 0.72909, "loss": 26.83428, "grad_norm": 46.11024, "time": 1.50576} -{"mode": "train", "epoch": 1, "iter": 161, "lr": 0.00016, "memory": 49163, "data_time": 0.11395, "loss_cls_0": 0.82831, "loss_box_0": 1.7354, "loss_cns_0": 0.62092, "loss_yns_0": 0.15544, "loss_cls_1": 0.90369, "loss_box_1": 1.74082, "loss_cns_1": 0.65316, "loss_yns_1": 0.15352, "loss_cls_2": 0.92993, "loss_box_2": 1.70725, "loss_cns_2": 0.65742, "loss_yns_2": 0.15446, "loss_cls_3": 0.92663, "loss_box_3": 1.70576, "loss_cns_3": 0.65952, "loss_yns_3": 0.1539, "loss_cls_4": 0.94452, "loss_box_4": 1.7019, "loss_cns_4": 0.66213, "loss_yns_4": 0.15478, "loss_cls_5": 0.93199, "loss_box_5": 1.69815, "loss_cns_5": 0.65665, "loss_yns_5": 0.15376, "loss_cls_dn_0": 0.20757, "loss_box_dn_0": 0.77058, "loss_cls_dn_1": 0.13952, "loss_box_dn_1": 0.74776, "loss_cls_dn_2": 0.14018, "loss_box_dn_2": 0.73599, "loss_cls_dn_3": 0.14203, "loss_box_dn_3": 0.74442, "loss_cls_dn_4": 0.14857, "loss_box_dn_4": 0.75142, "loss_cls_dn_5": 0.14806, "loss_box_dn_5": 0.75053, "loss_dense_depth": 0.72547, "loss": 26.7421, "grad_norm": 46.08959, "time": 1.53145} -{"mode": "train", "epoch": 1, "iter": 162, "lr": 0.00016, "memory": 49163, "data_time": 0.10229, "loss_cls_0": 0.83773, "loss_box_0": 1.74728, "loss_cns_0": 0.618, "loss_yns_0": 0.15594, "loss_cls_1": 0.9057, "loss_box_1": 1.74263, "loss_cns_1": 0.64748, "loss_yns_1": 0.15617, "loss_cls_2": 0.92644, "loss_box_2": 1.69787, "loss_cns_2": 0.65229, "loss_yns_2": 0.15622, "loss_cls_3": 0.92703, "loss_box_3": 1.70429, "loss_cns_3": 0.65489, "loss_yns_3": 0.15698, "loss_cls_4": 0.93259, "loss_box_4": 1.69938, "loss_cns_4": 0.6519, "loss_yns_4": 0.15805, "loss_cls_5": 0.9272, "loss_box_5": 1.70702, "loss_cns_5": 0.65118, "loss_yns_5": 0.15688, "loss_cls_dn_0": 0.20956, "loss_box_dn_0": 0.77809, "loss_cls_dn_1": 0.13164, "loss_box_dn_1": 0.72928, "loss_cls_dn_2": 0.13278, "loss_box_dn_2": 0.70796, "loss_cls_dn_3": 0.13321, "loss_box_dn_3": 0.71473, "loss_cls_dn_4": 0.13748, "loss_box_dn_4": 0.71923, "loss_cls_dn_5": 0.1415, "loss_box_dn_5": 0.72528, "loss_dense_depth": 0.71777, "loss": 26.54967, "grad_norm": 37.22063, "time": 1.54139} -{"mode": "train", "epoch": 1, "iter": 163, "lr": 0.00016, "memory": 49163, "data_time": 0.08082, "loss_cls_0": 0.83136, "loss_box_0": 1.76603, "loss_cns_0": 0.62114, "loss_yns_0": 0.15694, "loss_cls_1": 0.91075, "loss_box_1": 1.76057, "loss_cns_1": 0.64964, "loss_yns_1": 0.15778, "loss_cls_2": 0.939, "loss_box_2": 1.71627, "loss_cns_2": 0.65353, "loss_yns_2": 0.15679, "loss_cls_3": 0.92402, "loss_box_3": 1.70727, "loss_cns_3": 0.65473, "loss_yns_3": 0.15786, "loss_cls_4": 0.92813, "loss_box_4": 1.70258, "loss_cns_4": 0.6556, "loss_yns_4": 0.15798, "loss_cls_5": 0.92445, "loss_box_5": 1.71888, "loss_cns_5": 0.65191, "loss_yns_5": 0.15791, "loss_cls_dn_0": 0.20902, "loss_box_dn_0": 0.77058, "loss_cls_dn_1": 0.13485, "loss_box_dn_1": 0.7148, "loss_cls_dn_2": 0.13779, "loss_box_dn_2": 0.69367, "loss_cls_dn_3": 0.13939, "loss_box_dn_3": 0.69393, "loss_cls_dn_4": 0.14212, "loss_box_dn_4": 0.69795, "loss_cls_dn_5": 0.14446, "loss_box_dn_5": 0.70664, "loss_dense_depth": 0.71754, "loss": 26.56388, "grad_norm": 39.81119, "time": 1.52472} -{"mode": "train", "epoch": 1, "iter": 164, "lr": 0.00017, "memory": 49163, "data_time": 0.08146, "loss_cls_0": 0.83521, "loss_box_0": 1.77419, "loss_cns_0": 0.624, "loss_yns_0": 0.15904, "loss_cls_1": 0.9068, "loss_box_1": 1.75704, "loss_cns_1": 0.65226, "loss_yns_1": 0.15866, "loss_cls_2": 0.92916, "loss_box_2": 1.71319, "loss_cns_2": 0.65546, "loss_yns_2": 0.15836, "loss_cls_3": 0.92152, "loss_box_3": 1.69943, "loss_cns_3": 0.65714, "loss_yns_3": 0.15935, "loss_cls_4": 0.93376, "loss_box_4": 1.69836, "loss_cns_4": 0.66163, "loss_yns_4": 0.159, "loss_cls_5": 0.93456, "loss_box_5": 1.70363, "loss_cns_5": 0.65777, "loss_yns_5": 0.15952, "loss_cls_dn_0": 0.20876, "loss_box_dn_0": 0.77482, "loss_cls_dn_1": 0.1318, "loss_box_dn_1": 0.7058, "loss_cls_dn_2": 0.13549, "loss_box_dn_2": 0.68872, "loss_cls_dn_3": 0.13714, "loss_box_dn_3": 0.68802, "loss_cls_dn_4": 0.14225, "loss_box_dn_4": 0.69539, "loss_cls_dn_5": 0.14409, "loss_box_dn_5": 0.70138, "loss_dense_depth": 0.7243, "loss": 26.54699, "grad_norm": 39.89431, "time": 1.51051} -{"mode": "train", "epoch": 1, "iter": 165, "lr": 0.00017, "memory": 49163, "data_time": 0.08422, "loss_cls_0": 0.85459, "loss_box_0": 1.78812, "loss_cns_0": 0.61758, "loss_yns_0": 0.1605, "loss_cls_1": 0.90515, "loss_box_1": 1.7414, "loss_cns_1": 0.64813, "loss_yns_1": 0.16146, "loss_cls_2": 0.91895, "loss_box_2": 1.71478, "loss_cns_2": 0.6536, "loss_yns_2": 0.1593, "loss_cls_3": 0.92615, "loss_box_3": 1.70077, "loss_cns_3": 0.65487, "loss_yns_3": 0.15908, "loss_cls_4": 0.93933, "loss_box_4": 1.70123, "loss_cns_4": 0.65735, "loss_yns_4": 0.15981, "loss_cls_5": 0.93693, "loss_box_5": 1.70197, "loss_cns_5": 0.65478, "loss_yns_5": 0.16074, "loss_cls_dn_0": 0.20885, "loss_box_dn_0": 0.77033, "loss_cls_dn_1": 0.12953, "loss_box_dn_1": 0.71565, "loss_cls_dn_2": 0.13178, "loss_box_dn_2": 0.70462, "loss_cls_dn_3": 0.13072, "loss_box_dn_3": 0.70718, "loss_cls_dn_4": 0.1367, "loss_box_dn_4": 0.72074, "loss_cls_dn_5": 0.14002, "loss_box_dn_5": 0.73117, "loss_dense_depth": 0.74077, "loss": 26.64462, "grad_norm": 37.21601, "time": 1.51657} -{"mode": "train", "epoch": 1, "iter": 166, "lr": 0.00017, "memory": 49163, "data_time": 0.08788, "loss_cls_0": 0.82729, "loss_box_0": 1.80302, "loss_cns_0": 0.61232, "loss_yns_0": 0.15891, "loss_cls_1": 0.90973, "loss_box_1": 1.70937, "loss_cns_1": 0.65308, "loss_yns_1": 0.16143, "loss_cls_2": 0.93974, "loss_box_2": 1.67268, "loss_cns_2": 0.65726, "loss_yns_2": 0.16002, "loss_cls_3": 0.93075, "loss_box_3": 1.66422, "loss_cns_3": 0.65932, "loss_yns_3": 0.16024, "loss_cls_4": 0.92519, "loss_box_4": 1.6671, "loss_cns_4": 0.65979, "loss_yns_4": 0.16157, "loss_cls_5": 0.91716, "loss_box_5": 1.67762, "loss_cns_5": 0.65699, "loss_yns_5": 0.16122, "loss_cls_dn_0": 0.20187, "loss_box_dn_0": 0.76536, "loss_cls_dn_1": 0.12959, "loss_box_dn_1": 0.73597, "loss_cls_dn_2": 0.129, "loss_box_dn_2": 0.72321, "loss_cls_dn_3": 0.13137, "loss_box_dn_3": 0.73075, "loss_cls_dn_4": 0.13692, "loss_box_dn_4": 0.74703, "loss_cls_dn_5": 0.14102, "loss_box_dn_5": 0.7633, "loss_dense_depth": 0.72603, "loss": 26.56745, "grad_norm": 44.41732, "time": 1.50558} -{"mode": "train", "epoch": 1, "iter": 167, "lr": 0.00017, "memory": 49163, "data_time": 0.11228, "loss_cls_0": 0.87913, "loss_box_0": 1.78738, "loss_cns_0": 0.60654, "loss_yns_0": 0.15999, "loss_cls_1": 0.93992, "loss_box_1": 1.71385, "loss_cns_1": 0.65255, "loss_yns_1": 0.15878, "loss_cls_2": 0.95716, "loss_box_2": 1.69341, "loss_cns_2": 0.65576, "loss_yns_2": 0.15933, "loss_cls_3": 0.94857, "loss_box_3": 1.692, "loss_cns_3": 0.65667, "loss_yns_3": 0.15993, "loss_cls_4": 0.95399, "loss_box_4": 1.6916, "loss_cns_4": 0.655, "loss_yns_4": 0.16312, "loss_cls_5": 0.94947, "loss_box_5": 1.70601, "loss_cns_5": 0.65474, "loss_yns_5": 0.16265, "loss_cls_dn_0": 0.21292, "loss_box_dn_0": 0.77315, "loss_cls_dn_1": 0.137, "loss_box_dn_1": 0.78374, "loss_cls_dn_2": 0.13591, "loss_box_dn_2": 0.77318, "loss_cls_dn_3": 0.13822, "loss_box_dn_3": 0.78035, "loss_cls_dn_4": 0.14273, "loss_box_dn_4": 0.79188, "loss_cls_dn_5": 0.1477, "loss_box_dn_5": 0.80895, "loss_dense_depth": 0.77333, "loss": 27.15663, "grad_norm": 47.49758, "time": 6.93536} -{"mode": "train", "epoch": 1, "iter": 168, "lr": 0.00017, "memory": 49163, "data_time": 0.07706, "loss_cls_0": 0.86306, "loss_box_0": 1.73247, "loss_cns_0": 0.61223, "loss_yns_0": 0.15929, "loss_cls_1": 0.94119, "loss_box_1": 1.69518, "loss_cns_1": 0.6529, "loss_yns_1": 0.16088, "loss_cls_2": 0.95275, "loss_box_2": 1.66535, "loss_cns_2": 0.655, "loss_yns_2": 0.16089, "loss_cls_3": 0.94355, "loss_box_3": 1.66227, "loss_cns_3": 0.65651, "loss_yns_3": 0.16141, "loss_cls_4": 0.94626, "loss_box_4": 1.66483, "loss_cns_4": 0.65771, "loss_yns_4": 0.16185, "loss_cls_5": 0.94448, "loss_box_5": 1.67599, "loss_cns_5": 0.65806, "loss_yns_5": 0.16258, "loss_cls_dn_0": 0.20715, "loss_box_dn_0": 0.76139, "loss_cls_dn_1": 0.13892, "loss_box_dn_1": 0.78909, "loss_cls_dn_2": 0.13751, "loss_box_dn_2": 0.77521, "loss_cls_dn_3": 0.13849, "loss_box_dn_3": 0.78134, "loss_cls_dn_4": 0.14473, "loss_box_dn_4": 0.7887, "loss_cls_dn_5": 0.14937, "loss_box_dn_5": 0.80185, "loss_dense_depth": 0.74627, "loss": 26.90671, "grad_norm": 37.73682, "time": 1.48039} -{"mode": "train", "epoch": 1, "iter": 169, "lr": 0.00017, "memory": 49163, "data_time": 0.09231, "loss_cls_0": 0.88314, "loss_box_0": 1.7649, "loss_cns_0": 0.61172, "loss_yns_0": 0.16244, "loss_cls_1": 0.94652, "loss_box_1": 1.76218, "loss_cns_1": 0.64253, "loss_yns_1": 0.16294, "loss_cls_2": 0.96968, "loss_box_2": 1.70014, "loss_cns_2": 0.64718, "loss_yns_2": 0.16224, "loss_cls_3": 0.96385, "loss_box_3": 1.70146, "loss_cns_3": 0.64757, "loss_yns_3": 0.16157, "loss_cls_4": 0.96674, "loss_box_4": 1.70024, "loss_cns_4": 0.65346, "loss_yns_4": 0.16136, "loss_cls_5": 0.95937, "loss_box_5": 1.71801, "loss_cns_5": 0.65115, "loss_yns_5": 0.16159, "loss_cls_dn_0": 0.21573, "loss_box_dn_0": 0.76729, "loss_cls_dn_1": 0.13411, "loss_box_dn_1": 0.77766, "loss_cls_dn_2": 0.13519, "loss_box_dn_2": 0.75306, "loss_cls_dn_3": 0.13575, "loss_box_dn_3": 0.75894, "loss_cls_dn_4": 0.14257, "loss_box_dn_4": 0.76533, "loss_cls_dn_5": 0.1446, "loss_box_dn_5": 0.77015, "loss_dense_depth": 0.79841, "loss": 27.16079, "grad_norm": 49.37163, "time": 1.50812} -{"mode": "train", "epoch": 1, "iter": 170, "lr": 0.00017, "memory": 49163, "data_time": 0.07993, "loss_cls_0": 0.87464, "loss_box_0": 1.74629, "loss_cns_0": 0.62065, "loss_yns_0": 0.16169, "loss_cls_1": 0.93644, "loss_box_1": 1.71339, "loss_cns_1": 0.6478, "loss_yns_1": 0.16132, "loss_cls_2": 0.9706, "loss_box_2": 1.64233, "loss_cns_2": 0.65544, "loss_yns_2": 0.16099, "loss_cls_3": 0.96393, "loss_box_3": 1.62916, "loss_cns_3": 0.65675, "loss_yns_3": 0.16061, "loss_cls_4": 0.9589, "loss_box_4": 1.63441, "loss_cns_4": 0.65698, "loss_yns_4": 0.16125, "loss_cls_5": 0.96151, "loss_box_5": 1.65012, "loss_cns_5": 0.65491, "loss_yns_5": 0.16251, "loss_cls_dn_0": 0.2149, "loss_box_dn_0": 0.76432, "loss_cls_dn_1": 0.13636, "loss_box_dn_1": 0.73793, "loss_cls_dn_2": 0.13981, "loss_box_dn_2": 0.71252, "loss_cls_dn_3": 0.14015, "loss_box_dn_3": 0.71645, "loss_cls_dn_4": 0.14499, "loss_box_dn_4": 0.72247, "loss_cls_dn_5": 0.14606, "loss_box_dn_5": 0.7277, "loss_dense_depth": 0.7486, "loss": 26.59488, "grad_norm": 40.69604, "time": 1.49396} -{"mode": "train", "epoch": 1, "iter": 171, "lr": 0.00017, "memory": 49163, "data_time": 0.07717, "loss_cls_0": 0.86331, "loss_box_0": 1.73761, "loss_cns_0": 0.61642, "loss_yns_0": 0.15906, "loss_cls_1": 0.93403, "loss_box_1": 1.71068, "loss_cns_1": 0.64563, "loss_yns_1": 0.1597, "loss_cls_2": 0.96483, "loss_box_2": 1.68119, "loss_cns_2": 0.65227, "loss_yns_2": 0.15972, "loss_cls_3": 0.95962, "loss_box_3": 1.66392, "loss_cns_3": 0.65457, "loss_yns_3": 0.16038, "loss_cls_4": 0.9545, "loss_box_4": 1.66962, "loss_cns_4": 0.65192, "loss_yns_4": 0.16018, "loss_cls_5": 0.95284, "loss_box_5": 1.66518, "loss_cns_5": 0.65388, "loss_yns_5": 0.16009, "loss_cls_dn_0": 0.21166, "loss_box_dn_0": 0.76619, "loss_cls_dn_1": 0.14036, "loss_box_dn_1": 0.73038, "loss_cls_dn_2": 0.14289, "loss_box_dn_2": 0.72291, "loss_cls_dn_3": 0.14757, "loss_box_dn_3": 0.72305, "loss_cls_dn_4": 0.15009, "loss_box_dn_4": 0.73186, "loss_cls_dn_5": 0.15302, "loss_box_dn_5": 0.73618, "loss_dense_depth": 0.71608, "loss": 26.66338, "grad_norm": 46.39335, "time": 1.48748} -{"mode": "train", "epoch": 1, "iter": 172, "lr": 0.00017, "memory": 49163, "data_time": 0.0788, "loss_cls_0": 0.86614, "loss_box_0": 1.74899, "loss_cns_0": 0.61294, "loss_yns_0": 0.15944, "loss_cls_1": 0.93459, "loss_box_1": 1.73535, "loss_cns_1": 0.64671, "loss_yns_1": 0.16059, "loss_cls_2": 0.95161, "loss_box_2": 1.71933, "loss_cns_2": 0.65168, "loss_yns_2": 0.16137, "loss_cls_3": 0.95002, "loss_box_3": 1.69853, "loss_cns_3": 0.655, "loss_yns_3": 0.16179, "loss_cls_4": 0.95427, "loss_box_4": 1.69942, "loss_cns_4": 0.65372, "loss_yns_4": 0.1616, "loss_cls_5": 0.95207, "loss_box_5": 1.68371, "loss_cns_5": 0.65771, "loss_yns_5": 0.16145, "loss_cls_dn_0": 0.21287, "loss_box_dn_0": 0.75747, "loss_cls_dn_1": 0.13479, "loss_box_dn_1": 0.7304, "loss_cls_dn_2": 0.13591, "loss_box_dn_2": 0.73045, "loss_cls_dn_3": 0.13941, "loss_box_dn_3": 0.7316, "loss_cls_dn_4": 0.14663, "loss_box_dn_4": 0.74294, "loss_cls_dn_5": 0.15497, "loss_box_dn_5": 0.74428, "loss_dense_depth": 0.72854, "loss": 26.8283, "grad_norm": 52.07562, "time": 1.48928} -{"mode": "train", "epoch": 1, "iter": 173, "lr": 0.00017, "memory": 49163, "data_time": 0.08014, "loss_cls_0": 0.85339, "loss_box_0": 1.71969, "loss_cns_0": 0.61754, "loss_yns_0": 0.15717, "loss_cls_1": 0.92302, "loss_box_1": 1.71305, "loss_cns_1": 0.64811, "loss_yns_1": 0.15925, "loss_cls_2": 0.94124, "loss_box_2": 1.69848, "loss_cns_2": 0.65216, "loss_yns_2": 0.15806, "loss_cls_3": 0.95542, "loss_box_3": 1.6785, "loss_cns_3": 0.6516, "loss_yns_3": 0.15977, "loss_cls_4": 0.95334, "loss_box_4": 1.67821, "loss_cns_4": 0.65118, "loss_yns_4": 0.15816, "loss_cls_5": 0.95057, "loss_box_5": 1.66975, "loss_cns_5": 0.65176, "loss_yns_5": 0.15903, "loss_cls_dn_0": 0.21289, "loss_box_dn_0": 0.76479, "loss_cls_dn_1": 0.13586, "loss_box_dn_1": 0.73386, "loss_cls_dn_2": 0.13814, "loss_box_dn_2": 0.72967, "loss_cls_dn_3": 0.14062, "loss_box_dn_3": 0.72972, "loss_cls_dn_4": 0.14463, "loss_box_dn_4": 0.74022, "loss_cls_dn_5": 0.1487, "loss_box_dn_5": 0.74626, "loss_dense_depth": 0.7293, "loss": 26.65308, "grad_norm": 42.55689, "time": 1.49752} -{"mode": "train", "epoch": 1, "iter": 174, "lr": 0.00017, "memory": 49163, "data_time": 0.08081, "loss_cls_0": 0.86281, "loss_box_0": 1.74517, "loss_cns_0": 0.60795, "loss_yns_0": 0.15539, "loss_cls_1": 0.91661, "loss_box_1": 1.71611, "loss_cns_1": 0.63884, "loss_yns_1": 0.15755, "loss_cls_2": 0.93727, "loss_box_2": 1.68969, "loss_cns_2": 0.65235, "loss_yns_2": 0.1572, "loss_cls_3": 0.94575, "loss_box_3": 1.67914, "loss_cns_3": 0.64922, "loss_yns_3": 0.15638, "loss_cls_4": 0.93957, "loss_box_4": 1.68018, "loss_cns_4": 0.64567, "loss_yns_4": 0.15712, "loss_cls_5": 0.94755, "loss_box_5": 1.67931, "loss_cns_5": 0.64768, "loss_yns_5": 0.15754, "loss_cls_dn_0": 0.21095, "loss_box_dn_0": 0.77456, "loss_cls_dn_1": 0.13671, "loss_box_dn_1": 0.743, "loss_cls_dn_2": 0.14153, "loss_box_dn_2": 0.72889, "loss_cls_dn_3": 0.14719, "loss_box_dn_3": 0.7311, "loss_cls_dn_4": 0.14742, "loss_box_dn_4": 0.73954, "loss_cls_dn_5": 0.14772, "loss_box_dn_5": 0.74943, "loss_dense_depth": 0.7426, "loss": 26.66272, "grad_norm": 34.52126, "time": 1.48609} -{"mode": "train", "epoch": 1, "iter": 175, "lr": 0.00017, "memory": 49163, "data_time": 0.08362, "loss_cls_0": 0.85538, "loss_box_0": 1.73614, "loss_cns_0": 0.61783, "loss_yns_0": 0.15756, "loss_cls_1": 0.92308, "loss_box_1": 1.68649, "loss_cns_1": 0.649, "loss_yns_1": 0.15697, "loss_cls_2": 0.93697, "loss_box_2": 1.67655, "loss_cns_2": 0.65364, "loss_yns_2": 0.15639, "loss_cls_3": 0.93948, "loss_box_3": 1.65865, "loss_cns_3": 0.65617, "loss_yns_3": 0.15592, "loss_cls_4": 0.94726, "loss_box_4": 1.65844, "loss_cns_4": 0.65566, "loss_yns_4": 0.15664, "loss_cls_5": 0.9417, "loss_box_5": 1.67215, "loss_cns_5": 0.65536, "loss_yns_5": 0.15598, "loss_cls_dn_0": 0.21332, "loss_box_dn_0": 0.75676, "loss_cls_dn_1": 0.13375, "loss_box_dn_1": 0.72985, "loss_cls_dn_2": 0.13969, "loss_box_dn_2": 0.72, "loss_cls_dn_3": 0.14511, "loss_box_dn_3": 0.72089, "loss_cls_dn_4": 0.14221, "loss_box_dn_4": 0.7312, "loss_cls_dn_5": 0.14758, "loss_box_dn_5": 0.73948, "loss_dense_depth": 0.74034, "loss": 26.5196, "grad_norm": 45.08095, "time": 1.5383} -{"mode": "train", "epoch": 1, "iter": 176, "lr": 0.00017, "memory": 49163, "data_time": 0.08155, "loss_cls_0": 0.86386, "loss_box_0": 1.72428, "loss_cns_0": 0.6214, "loss_yns_0": 0.15537, "loss_cls_1": 0.92654, "loss_box_1": 1.67593, "loss_cns_1": 0.64959, "loss_yns_1": 0.1536, "loss_cls_2": 0.93368, "loss_box_2": 1.68536, "loss_cns_2": 0.65371, "loss_yns_2": 0.15252, "loss_cls_3": 0.93336, "loss_box_3": 1.6591, "loss_cns_3": 0.6574, "loss_yns_3": 0.15323, "loss_cls_4": 0.94296, "loss_box_4": 1.6639, "loss_cns_4": 0.65767, "loss_yns_4": 0.15348, "loss_cls_5": 0.93856, "loss_box_5": 1.66577, "loss_cns_5": 0.6555, "loss_yns_5": 0.15249, "loss_cls_dn_0": 0.21405, "loss_box_dn_0": 0.76346, "loss_cls_dn_1": 0.13795, "loss_box_dn_1": 0.73284, "loss_cls_dn_2": 0.14109, "loss_box_dn_2": 0.72922, "loss_cls_dn_3": 0.13875, "loss_box_dn_3": 0.72451, "loss_cls_dn_4": 0.141, "loss_box_dn_4": 0.73369, "loss_cls_dn_5": 0.14555, "loss_box_dn_5": 0.7393, "loss_dense_depth": 0.76814, "loss": 26.53881, "grad_norm": 40.13657, "time": 1.48837} -{"mode": "train", "epoch": 1, "iter": 177, "lr": 0.00017, "memory": 49163, "data_time": 0.08018, "loss_cls_0": 0.85526, "loss_box_0": 1.73089, "loss_cns_0": 0.62138, "loss_yns_0": 0.15221, "loss_cls_1": 0.91722, "loss_box_1": 1.66073, "loss_cns_1": 0.65335, "loss_yns_1": 0.15262, "loss_cls_2": 0.92575, "loss_box_2": 1.65318, "loss_cns_2": 0.65983, "loss_yns_2": 0.15164, "loss_cls_3": 0.92676, "loss_box_3": 1.63743, "loss_cns_3": 0.6584, "loss_yns_3": 0.15357, "loss_cls_4": 0.9268, "loss_box_4": 1.63914, "loss_cns_4": 0.65709, "loss_yns_4": 0.15138, "loss_cls_5": 0.93112, "loss_box_5": 1.63715, "loss_cns_5": 0.65587, "loss_yns_5": 0.15085, "loss_cls_dn_0": 0.21205, "loss_box_dn_0": 0.75998, "loss_cls_dn_1": 0.13528, "loss_box_dn_1": 0.7379, "loss_cls_dn_2": 0.1348, "loss_box_dn_2": 0.72671, "loss_cls_dn_3": 0.13725, "loss_box_dn_3": 0.72267, "loss_cls_dn_4": 0.1475, "loss_box_dn_4": 0.72886, "loss_cls_dn_5": 0.14332, "loss_box_dn_5": 0.73417, "loss_dense_depth": 0.76967, "loss": 26.34978, "grad_norm": 37.35575, "time": 1.54072} -{"mode": "train", "epoch": 1, "iter": 178, "lr": 0.00017, "memory": 49163, "data_time": 0.07672, "loss_cls_0": 0.8363, "loss_box_0": 1.73243, "loss_cns_0": 0.61841, "loss_yns_0": 0.15138, "loss_cls_1": 0.90387, "loss_box_1": 1.69516, "loss_cns_1": 0.64669, "loss_yns_1": 0.15218, "loss_cls_2": 0.92308, "loss_box_2": 1.66928, "loss_cns_2": 0.65145, "loss_yns_2": 0.15037, "loss_cls_3": 0.93689, "loss_box_3": 1.65251, "loss_cns_3": 0.65008, "loss_yns_3": 0.15071, "loss_cls_4": 0.92594, "loss_box_4": 1.65351, "loss_cns_4": 0.65243, "loss_yns_4": 0.15002, "loss_cls_5": 0.92632, "loss_box_5": 1.65521, "loss_cns_5": 0.64953, "loss_yns_5": 0.15054, "loss_cls_dn_0": 0.21151, "loss_box_dn_0": 0.7618, "loss_cls_dn_1": 0.13399, "loss_box_dn_1": 0.71859, "loss_cls_dn_2": 0.13718, "loss_box_dn_2": 0.70655, "loss_cls_dn_3": 0.14688, "loss_box_dn_3": 0.70687, "loss_cls_dn_4": 0.14952, "loss_box_dn_4": 0.7146, "loss_cls_dn_5": 0.15056, "loss_box_dn_5": 0.72086, "loss_dense_depth": 0.77373, "loss": 26.31696, "grad_norm": 43.28925, "time": 1.48048} -{"mode": "train", "epoch": 1, "iter": 179, "lr": 0.00017, "memory": 49163, "data_time": 0.07556, "loss_cls_0": 0.84219, "loss_box_0": 1.71932, "loss_cns_0": 0.61561, "loss_yns_0": 0.15293, "loss_cls_1": 0.90368, "loss_box_1": 1.69939, "loss_cns_1": 0.64784, "loss_yns_1": 0.15388, "loss_cls_2": 0.92494, "loss_box_2": 1.6711, "loss_cns_2": 0.65167, "loss_yns_2": 0.1509, "loss_cls_3": 0.93473, "loss_box_3": 1.65012, "loss_cns_3": 0.65275, "loss_yns_3": 0.15162, "loss_cls_4": 0.94012, "loss_box_4": 1.65116, "loss_cns_4": 0.65416, "loss_yns_4": 0.15102, "loss_cls_5": 0.93773, "loss_box_5": 1.65958, "loss_cns_5": 0.65342, "loss_yns_5": 0.15152, "loss_cls_dn_0": 0.21357, "loss_box_dn_0": 0.77258, "loss_cls_dn_1": 0.13577, "loss_box_dn_1": 0.71918, "loss_cls_dn_2": 0.14218, "loss_box_dn_2": 0.70779, "loss_cls_dn_3": 0.1535, "loss_box_dn_3": 0.70749, "loss_cls_dn_4": 0.14499, "loss_box_dn_4": 0.71186, "loss_cls_dn_5": 0.15233, "loss_box_dn_5": 0.7182, "loss_dense_depth": 0.74344, "loss": 26.34422, "grad_norm": 40.82825, "time": 1.49195} -{"mode": "train", "epoch": 1, "iter": 180, "lr": 0.00017, "memory": 49163, "data_time": 0.08017, "loss_cls_0": 0.83097, "loss_box_0": 1.69192, "loss_cns_0": 0.61998, "loss_yns_0": 0.15095, "loss_cls_1": 0.91168, "loss_box_1": 1.64537, "loss_cns_1": 0.65358, "loss_yns_1": 0.15222, "loss_cls_2": 0.92044, "loss_box_2": 1.62464, "loss_cns_2": 0.65717, "loss_yns_2": 0.15111, "loss_cls_3": 0.91948, "loss_box_3": 1.60706, "loss_cns_3": 0.65972, "loss_yns_3": 0.14996, "loss_cls_4": 0.93133, "loss_box_4": 1.61208, "loss_cns_4": 0.6575, "loss_yns_4": 0.15198, "loss_cls_5": 0.92365, "loss_box_5": 1.6081, "loss_cns_5": 0.65688, "loss_yns_5": 0.15081, "loss_cls_dn_0": 0.21377, "loss_box_dn_0": 0.76915, "loss_cls_dn_1": 0.13265, "loss_box_dn_1": 0.71997, "loss_cls_dn_2": 0.13721, "loss_box_dn_2": 0.71498, "loss_cls_dn_3": 0.14541, "loss_box_dn_3": 0.71499, "loss_cls_dn_4": 0.1409, "loss_box_dn_4": 0.71956, "loss_cls_dn_5": 0.15026, "loss_box_dn_5": 0.72601, "loss_dense_depth": 0.80256, "loss": 26.12602, "grad_norm": 46.18991, "time": 1.4997} -{"mode": "train", "epoch": 1, "iter": 181, "lr": 0.00017, "memory": 49163, "data_time": 0.10864, "loss_cls_0": 0.81791, "loss_box_0": 1.72963, "loss_cns_0": 0.6151, "loss_yns_0": 0.14956, "loss_cls_1": 0.90173, "loss_box_1": 1.68784, "loss_cns_1": 0.64828, "loss_yns_1": 0.15037, "loss_cls_2": 0.92255, "loss_box_2": 1.66415, "loss_cns_2": 0.65632, "loss_yns_2": 0.15029, "loss_cls_3": 0.91774, "loss_box_3": 1.65483, "loss_cns_3": 0.65672, "loss_yns_3": 0.15008, "loss_cls_4": 0.91996, "loss_box_4": 1.65304, "loss_cns_4": 0.65264, "loss_yns_4": 0.15096, "loss_cls_5": 0.92368, "loss_box_5": 1.65634, "loss_cns_5": 0.65244, "loss_yns_5": 0.15061, "loss_cls_dn_0": 0.21088, "loss_box_dn_0": 0.76876, "loss_cls_dn_1": 0.13181, "loss_box_dn_1": 0.7228, "loss_cls_dn_2": 0.13613, "loss_box_dn_2": 0.7129, "loss_cls_dn_3": 0.13865, "loss_box_dn_3": 0.71766, "loss_cls_dn_4": 0.14404, "loss_box_dn_4": 0.72445, "loss_cls_dn_5": 0.15011, "loss_box_dn_5": 0.73445, "loss_dense_depth": 0.78474, "loss": 26.31015, "grad_norm": 39.73343, "time": 1.54128} -{"mode": "train", "epoch": 1, "iter": 182, "lr": 0.00017, "memory": 49163, "data_time": 0.10219, "loss_cls_0": 0.81037, "loss_box_0": 1.69511, "loss_cns_0": 0.61985, "loss_yns_0": 0.14724, "loss_cls_1": 0.88689, "loss_box_1": 1.68311, "loss_cns_1": 0.64851, "loss_yns_1": 0.14692, "loss_cls_2": 0.90155, "loss_box_2": 1.64189, "loss_cns_2": 0.65706, "loss_yns_2": 0.14753, "loss_cls_3": 0.90209, "loss_box_3": 1.64256, "loss_cns_3": 0.65767, "loss_yns_3": 0.1485, "loss_cls_4": 0.91586, "loss_box_4": 1.64247, "loss_cns_4": 0.65488, "loss_yns_4": 0.14594, "loss_cls_5": 0.90529, "loss_box_5": 1.64844, "loss_cns_5": 0.65651, "loss_yns_5": 0.14653, "loss_cls_dn_0": 0.20704, "loss_box_dn_0": 0.76415, "loss_cls_dn_1": 0.13202, "loss_box_dn_1": 0.73106, "loss_cls_dn_2": 0.13409, "loss_box_dn_2": 0.71256, "loss_cls_dn_3": 0.1355, "loss_box_dn_3": 0.7195, "loss_cls_dn_4": 0.14256, "loss_box_dn_4": 0.72666, "loss_cls_dn_5": 0.14623, "loss_box_dn_5": 0.73493, "loss_dense_depth": 0.73202, "loss": 26.0711, "grad_norm": 37.91912, "time": 1.56287} -{"mode": "train", "epoch": 1, "iter": 183, "lr": 0.00017, "memory": 49163, "data_time": 0.07848, "loss_cls_0": 0.81464, "loss_box_0": 1.66596, "loss_cns_0": 0.61994, "loss_yns_0": 0.14523, "loss_cls_1": 0.88201, "loss_box_1": 1.62594, "loss_cns_1": 0.65042, "loss_yns_1": 0.14732, "loss_cls_2": 0.9001, "loss_box_2": 1.59355, "loss_cns_2": 0.65902, "loss_yns_2": 0.14837, "loss_cls_3": 0.90772, "loss_box_3": 1.59603, "loss_cns_3": 0.66186, "loss_yns_3": 0.14938, "loss_cls_4": 0.91241, "loss_box_4": 1.59816, "loss_cns_4": 0.65909, "loss_yns_4": 0.14745, "loss_cls_5": 0.90396, "loss_box_5": 1.60933, "loss_cns_5": 0.65927, "loss_yns_5": 0.14791, "loss_cls_dn_0": 0.20808, "loss_box_dn_0": 0.75416, "loss_cls_dn_1": 0.12805, "loss_box_dn_1": 0.72077, "loss_cls_dn_2": 0.12906, "loss_box_dn_2": 0.70258, "loss_cls_dn_3": 0.13345, "loss_box_dn_3": 0.7073, "loss_cls_dn_4": 0.14127, "loss_box_dn_4": 0.71578, "loss_cls_dn_5": 0.14078, "loss_box_dn_5": 0.72383, "loss_dense_depth": 0.77429, "loss": 25.78448, "grad_norm": 45.73533, "time": 1.53734} -{"mode": "train", "epoch": 1, "iter": 184, "lr": 0.00017, "memory": 49163, "data_time": 0.0815, "loss_cls_0": 0.82875, "loss_box_0": 1.70028, "loss_cns_0": 0.61942, "loss_yns_0": 0.14595, "loss_cls_1": 0.89483, "loss_box_1": 1.67172, "loss_cns_1": 0.65098, "loss_yns_1": 0.14679, "loss_cls_2": 0.91571, "loss_box_2": 1.65417, "loss_cns_2": 0.65656, "loss_yns_2": 0.14637, "loss_cls_3": 0.94081, "loss_box_3": 1.64889, "loss_cns_3": 0.65856, "loss_yns_3": 0.14766, "loss_cls_4": 0.92392, "loss_box_4": 1.6461, "loss_cns_4": 0.65967, "loss_yns_4": 0.14763, "loss_cls_5": 0.92625, "loss_box_5": 1.65492, "loss_cns_5": 0.66034, "loss_yns_5": 0.14662, "loss_cls_dn_0": 0.2155, "loss_box_dn_0": 0.75963, "loss_cls_dn_1": 0.13438, "loss_box_dn_1": 0.7226, "loss_cls_dn_2": 0.13439, "loss_box_dn_2": 0.70641, "loss_cls_dn_3": 0.14582, "loss_box_dn_3": 0.70407, "loss_cls_dn_4": 0.15027, "loss_box_dn_4": 0.70854, "loss_cls_dn_5": 0.14886, "loss_box_dn_5": 0.7174, "loss_dense_depth": 0.7639, "loss": 26.20465, "grad_norm": 40.48936, "time": 1.52887} -{"mode": "train", "epoch": 1, "iter": 185, "lr": 0.00017, "memory": 49163, "data_time": 0.08573, "loss_cls_0": 0.83343, "loss_box_0": 1.68509, "loss_cns_0": 0.61453, "loss_yns_0": 0.14561, "loss_cls_1": 0.89796, "loss_box_1": 1.67788, "loss_cns_1": 0.64816, "loss_yns_1": 0.1477, "loss_cls_2": 0.90709, "loss_box_2": 1.6494, "loss_cns_2": 0.65035, "loss_yns_2": 0.14704, "loss_cls_3": 0.91101, "loss_box_3": 1.64763, "loss_cns_3": 0.65292, "loss_yns_3": 0.14656, "loss_cls_4": 0.91549, "loss_box_4": 1.64019, "loss_cns_4": 0.65109, "loss_yns_4": 0.1471, "loss_cls_5": 0.91852, "loss_box_5": 1.64705, "loss_cns_5": 0.65404, "loss_yns_5": 0.14793, "loss_cls_dn_0": 0.21338, "loss_box_dn_0": 0.76592, "loss_cls_dn_1": 0.13578, "loss_box_dn_1": 0.72076, "loss_cls_dn_2": 0.13575, "loss_box_dn_2": 0.70011, "loss_cls_dn_3": 0.14289, "loss_box_dn_3": 0.6974, "loss_cls_dn_4": 0.14165, "loss_box_dn_4": 0.70152, "loss_cls_dn_5": 0.14403, "loss_box_dn_5": 0.7065, "loss_dense_depth": 0.77614, "loss": 26.06563, "grad_norm": 33.77507, "time": 1.5172} -{"mode": "train", "epoch": 1, "iter": 186, "lr": 0.00017, "memory": 49163, "data_time": 0.0872, "loss_cls_0": 0.83293, "loss_box_0": 1.67889, "loss_cns_0": 0.62236, "loss_yns_0": 0.14883, "loss_cls_1": 0.89507, "loss_box_1": 1.66454, "loss_cns_1": 0.64802, "loss_yns_1": 0.14825, "loss_cls_2": 0.91612, "loss_box_2": 1.60584, "loss_cns_2": 0.65613, "loss_yns_2": 0.14821, "loss_cls_3": 0.91606, "loss_box_3": 1.59991, "loss_cns_3": 0.65387, "loss_yns_3": 0.1472, "loss_cls_4": 0.93024, "loss_box_4": 1.60243, "loss_cns_4": 0.65251, "loss_yns_4": 0.14715, "loss_cls_5": 0.92435, "loss_box_5": 1.61082, "loss_cns_5": 0.65431, "loss_yns_5": 0.15003, "loss_cls_dn_0": 0.2074, "loss_box_dn_0": 0.76349, "loss_cls_dn_1": 0.13611, "loss_box_dn_1": 0.72189, "loss_cls_dn_2": 0.13718, "loss_box_dn_2": 0.70059, "loss_cls_dn_3": 0.13891, "loss_box_dn_3": 0.70502, "loss_cls_dn_4": 0.14476, "loss_box_dn_4": 0.7146, "loss_cls_dn_5": 0.14596, "loss_box_dn_5": 0.71984, "loss_dense_depth": 0.75043, "loss": 25.94027, "grad_norm": 38.83096, "time": 1.53249} -{"mode": "train", "epoch": 1, "iter": 187, "lr": 0.00017, "memory": 49163, "data_time": 0.11832, "loss_cls_0": 0.84475, "loss_box_0": 1.68744, "loss_cns_0": 0.62368, "loss_yns_0": 0.14927, "loss_cls_1": 0.91845, "loss_box_1": 1.64954, "loss_cns_1": 0.65475, "loss_yns_1": 0.14913, "loss_cls_2": 0.933, "loss_box_2": 1.61291, "loss_cns_2": 0.65963, "loss_yns_2": 0.14937, "loss_cls_3": 0.93155, "loss_box_3": 1.6117, "loss_cns_3": 0.65833, "loss_yns_3": 0.15046, "loss_cls_4": 0.94111, "loss_box_4": 1.61308, "loss_cns_4": 0.66178, "loss_yns_4": 0.15087, "loss_cls_5": 0.93036, "loss_box_5": 1.62346, "loss_cns_5": 0.66111, "loss_yns_5": 0.15156, "loss_cls_dn_0": 0.21228, "loss_box_dn_0": 0.75878, "loss_cls_dn_1": 0.13216, "loss_box_dn_1": 0.72827, "loss_cls_dn_2": 0.13388, "loss_box_dn_2": 0.71844, "loss_cls_dn_3": 0.13784, "loss_box_dn_3": 0.72388, "loss_cls_dn_4": 0.14638, "loss_box_dn_4": 0.73753, "loss_cls_dn_5": 0.14493, "loss_box_dn_5": 0.74616, "loss_dense_depth": 0.75609, "loss": 26.1939, "grad_norm": 44.95137, "time": 1.53245} -{"mode": "train", "epoch": 1, "iter": 188, "lr": 0.00017, "memory": 49163, "data_time": 0.07631, "loss_cls_0": 0.84038, "loss_box_0": 1.71429, "loss_cns_0": 0.62297, "loss_yns_0": 0.15056, "loss_cls_1": 0.91921, "loss_box_1": 1.67989, "loss_cns_1": 0.64904, "loss_yns_1": 0.14987, "loss_cls_2": 0.94119, "loss_box_2": 1.65393, "loss_cns_2": 0.65156, "loss_yns_2": 0.14994, "loss_cls_3": 0.93346, "loss_box_3": 1.63638, "loss_cns_3": 0.65159, "loss_yns_3": 0.15108, "loss_cls_4": 0.93125, "loss_box_4": 1.6407, "loss_cns_4": 0.65334, "loss_yns_4": 0.15051, "loss_cls_5": 0.92352, "loss_box_5": 1.67107, "loss_cns_5": 0.65434, "loss_yns_5": 0.15063, "loss_cls_dn_0": 0.20989, "loss_box_dn_0": 0.76029, "loss_cls_dn_1": 0.13446, "loss_box_dn_1": 0.75481, "loss_cls_dn_2": 0.13757, "loss_box_dn_2": 0.74647, "loss_cls_dn_3": 0.13928, "loss_box_dn_3": 0.74925, "loss_cls_dn_4": 0.14323, "loss_box_dn_4": 0.76712, "loss_cls_dn_5": 0.14242, "loss_box_dn_5": 0.78378, "loss_dense_depth": 0.76285, "loss": 26.50212, "grad_norm": 43.05302, "time": 1.49715} -{"mode": "train", "epoch": 1, "iter": 189, "lr": 0.00018, "memory": 49163, "data_time": 0.17476, "loss_cls_0": 0.84443, "loss_box_0": 1.68756, "loss_cns_0": 0.62849, "loss_yns_0": 0.15232, "loss_cls_1": 0.91898, "loss_box_1": 1.64293, "loss_cns_1": 0.65461, "loss_yns_1": 0.15276, "loss_cls_2": 0.94282, "loss_box_2": 1.61141, "loss_cns_2": 0.65876, "loss_yns_2": 0.15206, "loss_cls_3": 0.93027, "loss_box_3": 1.59674, "loss_cns_3": 0.65739, "loss_yns_3": 0.15177, "loss_cls_4": 0.94081, "loss_box_4": 1.60481, "loss_cns_4": 0.65518, "loss_yns_4": 0.15145, "loss_cls_5": 0.93623, "loss_box_5": 1.61804, "loss_cns_5": 0.65993, "loss_yns_5": 0.15366, "loss_cls_dn_0": 0.20867, "loss_box_dn_0": 0.76023, "loss_cls_dn_1": 0.13325, "loss_box_dn_1": 0.75505, "loss_cls_dn_2": 0.13553, "loss_box_dn_2": 0.74372, "loss_cls_dn_3": 0.13838, "loss_box_dn_3": 0.74463, "loss_cls_dn_4": 0.14593, "loss_box_dn_4": 0.76171, "loss_cls_dn_5": 0.14942, "loss_box_dn_5": 0.77082, "loss_dense_depth": 0.7738, "loss": 26.32455, "grad_norm": 44.82819, "time": 1.58262} -{"mode": "train", "epoch": 1, "iter": 190, "lr": 0.00018, "memory": 49163, "data_time": 0.07347, "loss_cls_0": 0.85685, "loss_box_0": 1.7226, "loss_cns_0": 0.6215, "loss_yns_0": 0.15266, "loss_cls_1": 0.9252, "loss_box_1": 1.67076, "loss_cns_1": 0.65189, "loss_yns_1": 0.15212, "loss_cls_2": 0.94137, "loss_box_2": 1.65209, "loss_cns_2": 0.65636, "loss_yns_2": 0.15371, "loss_cls_3": 0.93617, "loss_box_3": 1.66525, "loss_cns_3": 0.6536, "loss_yns_3": 0.15312, "loss_cls_4": 0.94533, "loss_box_4": 1.66886, "loss_cns_4": 0.65126, "loss_yns_4": 0.15469, "loss_cls_5": 0.9445, "loss_box_5": 1.65378, "loss_cns_5": 0.65515, "loss_yns_5": 0.15561, "loss_cls_dn_0": 0.21393, "loss_box_dn_0": 0.76711, "loss_cls_dn_1": 0.1342, "loss_box_dn_1": 0.74855, "loss_cls_dn_2": 0.13636, "loss_box_dn_2": 0.73813, "loss_cls_dn_3": 0.13811, "loss_box_dn_3": 0.74417, "loss_cls_dn_4": 0.14853, "loss_box_dn_4": 0.75874, "loss_cls_dn_5": 0.15322, "loss_box_dn_5": 0.75842, "loss_dense_depth": 0.77789, "loss": 26.61179, "grad_norm": 49.94524, "time": 1.49509} -{"mode": "train", "epoch": 1, "iter": 191, "lr": 0.00018, "memory": 49163, "data_time": 0.07536, "loss_cls_0": 0.84254, "loss_box_0": 1.72382, "loss_cns_0": 0.61989, "loss_yns_0": 0.15266, "loss_cls_1": 0.92817, "loss_box_1": 1.64992, "loss_cns_1": 0.65423, "loss_yns_1": 0.15273, "loss_cls_2": 0.94931, "loss_box_2": 1.63184, "loss_cns_2": 0.65654, "loss_yns_2": 0.15262, "loss_cls_3": 0.95879, "loss_box_3": 1.64678, "loss_cns_3": 0.65612, "loss_yns_3": 0.15257, "loss_cls_4": 0.94745, "loss_box_4": 1.63852, "loss_cns_4": 0.65437, "loss_yns_4": 0.15315, "loss_cls_5": 0.93899, "loss_box_5": 1.62642, "loss_cns_5": 0.65974, "loss_yns_5": 0.154, "loss_cls_dn_0": 0.21342, "loss_box_dn_0": 0.75212, "loss_cls_dn_1": 0.13543, "loss_box_dn_1": 0.72744, "loss_cls_dn_2": 0.13581, "loss_box_dn_2": 0.71541, "loss_cls_dn_3": 0.13629, "loss_box_dn_3": 0.72192, "loss_cls_dn_4": 0.14346, "loss_box_dn_4": 0.73012, "loss_cls_dn_5": 0.14776, "loss_box_dn_5": 0.73473, "loss_dense_depth": 0.75582, "loss": 26.3509, "grad_norm": 36.97875, "time": 1.47775} -{"mode": "train", "epoch": 1, "iter": 192, "lr": 0.00018, "memory": 49163, "data_time": 0.07669, "loss_cls_0": 0.83989, "loss_box_0": 1.72017, "loss_cns_0": 0.62081, "loss_yns_0": 0.15309, "loss_cls_1": 0.9281, "loss_box_1": 1.67166, "loss_cns_1": 0.65214, "loss_yns_1": 0.15255, "loss_cls_2": 0.9375, "loss_box_2": 1.65408, "loss_cns_2": 0.65211, "loss_yns_2": 0.1529, "loss_cls_3": 0.9409, "loss_box_3": 1.66726, "loss_cns_3": 0.65445, "loss_yns_3": 0.1534, "loss_cls_4": 0.93732, "loss_box_4": 1.65385, "loss_cns_4": 0.653, "loss_yns_4": 0.15352, "loss_cls_5": 0.9424, "loss_box_5": 1.66191, "loss_cns_5": 0.65566, "loss_yns_5": 0.15281, "loss_cls_dn_0": 0.21333, "loss_box_dn_0": 0.74857, "loss_cls_dn_1": 0.1393, "loss_box_dn_1": 0.72132, "loss_cls_dn_2": 0.13902, "loss_box_dn_2": 0.70928, "loss_cls_dn_3": 0.1386, "loss_box_dn_3": 0.71466, "loss_cls_dn_4": 0.14441, "loss_box_dn_4": 0.71549, "loss_cls_dn_5": 0.14581, "loss_box_dn_5": 0.72916, "loss_dense_depth": 0.73407, "loss": 26.35447, "grad_norm": 43.3742, "time": 1.4906} -{"mode": "train", "epoch": 1, "iter": 193, "lr": 0.00018, "memory": 49163, "data_time": 0.07874, "loss_cls_0": 0.82865, "loss_box_0": 1.69863, "loss_cns_0": 0.62612, "loss_yns_0": 0.15053, "loss_cls_1": 0.91892, "loss_box_1": 1.65862, "loss_cns_1": 0.65388, "loss_yns_1": 0.15032, "loss_cls_2": 0.93102, "loss_box_2": 1.63459, "loss_cns_2": 0.65513, "loss_yns_2": 0.15084, "loss_cls_3": 0.93421, "loss_box_3": 1.64312, "loss_cns_3": 0.65481, "loss_yns_3": 0.14977, "loss_cls_4": 0.94783, "loss_box_4": 1.62069, "loss_cns_4": 0.65218, "loss_yns_4": 0.14837, "loss_cls_5": 0.9343, "loss_box_5": 1.64961, "loss_cns_5": 0.65836, "loss_yns_5": 0.14915, "loss_cls_dn_0": 0.20872, "loss_box_dn_0": 0.76303, "loss_cls_dn_1": 0.13573, "loss_box_dn_1": 0.71034, "loss_cls_dn_2": 0.1356, "loss_box_dn_2": 0.69451, "loss_cls_dn_3": 0.13631, "loss_box_dn_3": 0.69888, "loss_cls_dn_4": 0.13991, "loss_box_dn_4": 0.70347, "loss_cls_dn_5": 0.14317, "loss_box_dn_5": 0.72231, "loss_dense_depth": 0.73993, "loss": 26.13155, "grad_norm": 33.32022, "time": 1.4962} -{"mode": "train", "epoch": 1, "iter": 194, "lr": 0.00018, "memory": 49163, "data_time": 0.0755, "loss_cls_0": 0.84699, "loss_box_0": 1.71162, "loss_cns_0": 0.62148, "loss_yns_0": 0.15336, "loss_cls_1": 0.92465, "loss_box_1": 1.72152, "loss_cns_1": 0.64581, "loss_yns_1": 0.15317, "loss_cls_2": 0.94526, "loss_box_2": 1.69337, "loss_cns_2": 0.65177, "loss_yns_2": 0.15455, "loss_cls_3": 0.95024, "loss_box_3": 1.68151, "loss_cns_3": 0.64875, "loss_yns_3": 0.15334, "loss_cls_4": 0.95042, "loss_box_4": 1.68695, "loss_cns_4": 0.6508, "loss_yns_4": 0.15444, "loss_cls_5": 0.94434, "loss_box_5": 1.70502, "loss_cns_5": 0.67027, "loss_yns_5": 0.15493, "loss_cls_dn_0": 0.20997, "loss_box_dn_0": 0.75625, "loss_cls_dn_1": 0.1327, "loss_box_dn_1": 0.7219, "loss_cls_dn_2": 0.13246, "loss_box_dn_2": 0.70955, "loss_cls_dn_3": 0.13551, "loss_box_dn_3": 0.71576, "loss_cls_dn_4": 0.13874, "loss_box_dn_4": 0.7248, "loss_cls_dn_5": 0.14269, "loss_box_dn_5": 0.74122, "loss_dense_depth": 0.77112, "loss": 26.60728, "grad_norm": 41.51515, "time": 1.49946} -{"mode": "train", "epoch": 1, "iter": 195, "lr": 0.00018, "memory": 49163, "data_time": 0.07851, "loss_cls_0": 0.81262, "loss_box_0": 1.68602, "loss_cns_0": 0.63231, "loss_yns_0": 0.15091, "loss_cls_1": 0.88685, "loss_box_1": 1.65799, "loss_cns_1": 0.64909, "loss_yns_1": 0.15065, "loss_cls_2": 0.90061, "loss_box_2": 1.62609, "loss_cns_2": 0.65575, "loss_yns_2": 0.15149, "loss_cls_3": 0.90491, "loss_box_3": 1.63095, "loss_cns_3": 0.65408, "loss_yns_3": 0.15113, "loss_cls_4": 0.9082, "loss_box_4": 1.62283, "loss_cns_4": 0.65714, "loss_yns_4": 0.15377, "loss_cls_5": 0.90297, "loss_box_5": 1.63343, "loss_cns_5": 0.66987, "loss_yns_5": 0.15274, "loss_cls_dn_0": 0.19738, "loss_box_dn_0": 0.75331, "loss_cls_dn_1": 0.13011, "loss_box_dn_1": 0.73438, "loss_cls_dn_2": 0.13038, "loss_box_dn_2": 0.7258, "loss_cls_dn_3": 0.13174, "loss_box_dn_3": 0.74442, "loss_cls_dn_4": 0.14021, "loss_box_dn_4": 0.74716, "loss_cls_dn_5": 0.14174, "loss_box_dn_5": 0.75611, "loss_dense_depth": 0.7307, "loss": 26.06584, "grad_norm": 36.32542, "time": 1.54911} -{"mode": "train", "epoch": 1, "iter": 196, "lr": 0.00018, "memory": 49163, "data_time": 0.07788, "loss_cls_0": 0.79406, "loss_box_0": 1.67799, "loss_cns_0": 0.62622, "loss_yns_0": 0.15129, "loss_cls_1": 0.87188, "loss_box_1": 1.65248, "loss_cns_1": 0.6498, "loss_yns_1": 0.15095, "loss_cls_2": 0.89059, "loss_box_2": 1.60868, "loss_cns_2": 0.65522, "loss_yns_2": 0.15156, "loss_cls_3": 0.89394, "loss_box_3": 1.6374, "loss_cns_3": 0.65676, "loss_yns_3": 0.15211, "loss_cls_4": 0.90038, "loss_box_4": 1.61827, "loss_cns_4": 0.66078, "loss_yns_4": 0.15298, "loss_cls_5": 0.89384, "loss_box_5": 1.6294, "loss_cns_5": 0.65987, "loss_yns_5": 0.15203, "loss_cls_dn_0": 0.19858, "loss_box_dn_0": 0.76108, "loss_cls_dn_1": 0.13191, "loss_box_dn_1": 0.76208, "loss_cls_dn_2": 0.13184, "loss_box_dn_2": 0.75164, "loss_cls_dn_3": 0.13455, "loss_box_dn_3": 0.77233, "loss_cls_dn_4": 0.14008, "loss_box_dn_4": 0.77293, "loss_cls_dn_5": 0.14136, "loss_box_dn_5": 0.778, "loss_dense_depth": 0.76555, "loss": 26.13044, "grad_norm": 41.34949, "time": 1.47963} -{"mode": "train", "epoch": 1, "iter": 197, "lr": 0.00018, "memory": 49163, "data_time": 0.07923, "loss_cls_0": 0.82653, "loss_box_0": 1.74747, "loss_cns_0": 0.62386, "loss_yns_0": 0.15117, "loss_cls_1": 0.88876, "loss_box_1": 1.70465, "loss_cns_1": 0.65536, "loss_yns_1": 0.15178, "loss_cls_2": 0.90432, "loss_box_2": 1.67144, "loss_cns_2": 0.65796, "loss_yns_2": 0.15264, "loss_cls_3": 0.91058, "loss_box_3": 1.67792, "loss_cns_3": 0.659, "loss_yns_3": 0.15371, "loss_cls_4": 0.9249, "loss_box_4": 1.66755, "loss_cns_4": 0.65978, "loss_yns_4": 0.15381, "loss_cls_5": 0.90188, "loss_box_5": 1.68887, "loss_cns_5": 0.65688, "loss_yns_5": 0.15225, "loss_cls_dn_0": 0.20501, "loss_box_dn_0": 0.75408, "loss_cls_dn_1": 0.13056, "loss_box_dn_1": 0.75465, "loss_cls_dn_2": 0.13261, "loss_box_dn_2": 0.73783, "loss_cls_dn_3": 0.1357, "loss_box_dn_3": 0.75378, "loss_cls_dn_4": 0.13884, "loss_box_dn_4": 0.75299, "loss_cls_dn_5": 0.14368, "loss_box_dn_5": 0.76279, "loss_dense_depth": 0.76756, "loss": 26.51318, "grad_norm": 32.3958, "time": 1.53954} -{"mode": "train", "epoch": 1, "iter": 198, "lr": 0.00018, "memory": 49163, "data_time": 0.07481, "loss_cls_0": 0.84375, "loss_box_0": 1.74677, "loss_cns_0": 0.61903, "loss_yns_0": 0.15366, "loss_cls_1": 0.90482, "loss_box_1": 1.69232, "loss_cns_1": 0.64885, "loss_yns_1": 0.15461, "loss_cls_2": 0.92089, "loss_box_2": 1.64873, "loss_cns_2": 0.6562, "loss_yns_2": 0.15455, "loss_cls_3": 0.91927, "loss_box_3": 1.64676, "loss_cns_3": 0.65476, "loss_yns_3": 0.15495, "loss_cls_4": 0.92602, "loss_box_4": 1.64382, "loss_cns_4": 0.65497, "loss_yns_4": 0.15521, "loss_cls_5": 0.91573, "loss_box_5": 1.6495, "loss_cns_5": 0.6571, "loss_yns_5": 0.15442, "loss_cls_dn_0": 0.20501, "loss_box_dn_0": 0.76321, "loss_cls_dn_1": 0.13231, "loss_box_dn_1": 0.76697, "loss_cls_dn_2": 0.13443, "loss_box_dn_2": 0.74654, "loss_cls_dn_3": 0.13776, "loss_box_dn_3": 0.75033, "loss_cls_dn_4": 0.14064, "loss_box_dn_4": 0.74976, "loss_cls_dn_5": 0.14665, "loss_box_dn_5": 0.75552, "loss_dense_depth": 0.7692, "loss": 26.475, "grad_norm": 34.22936, "time": 1.49199} -{"mode": "train", "epoch": 1, "iter": 199, "lr": 0.00018, "memory": 49163, "data_time": 0.08365, "loss_cls_0": 0.83482, "loss_box_0": 1.74722, "loss_cns_0": 0.61756, "loss_yns_0": 0.15462, "loss_cls_1": 0.89742, "loss_box_1": 1.70222, "loss_cns_1": 0.65031, "loss_yns_1": 0.15419, "loss_cls_2": 0.91183, "loss_box_2": 1.66062, "loss_cns_2": 0.65462, "loss_yns_2": 0.15423, "loss_cls_3": 0.90578, "loss_box_3": 1.65316, "loss_cns_3": 0.65597, "loss_yns_3": 0.15409, "loss_cls_4": 0.90418, "loss_box_4": 1.65264, "loss_cns_4": 0.65662, "loss_yns_4": 0.15442, "loss_cls_5": 0.91078, "loss_box_5": 1.66179, "loss_cns_5": 0.66714, "loss_yns_5": 0.15491, "loss_cls_dn_0": 0.20088, "loss_box_dn_0": 0.75186, "loss_cls_dn_1": 0.13132, "loss_box_dn_1": 0.75481, "loss_cls_dn_2": 0.13091, "loss_box_dn_2": 0.72948, "loss_cls_dn_3": 0.13199, "loss_box_dn_3": 0.7299, "loss_cls_dn_4": 0.13712, "loss_box_dn_4": 0.73334, "loss_cls_dn_5": 0.14086, "loss_box_dn_5": 0.74113, "loss_dense_depth": 0.79838, "loss": 26.38312, "grad_norm": 27.53248, "time": 1.51153} -{"mode": "train", "epoch": 1, "iter": 200, "lr": 0.00018, "memory": 49163, "data_time": 0.08057, "loss_cls_0": 0.82544, "loss_box_0": 1.75532, "loss_cns_0": 0.61359, "loss_yns_0": 0.15457, "loss_cls_1": 0.90011, "loss_box_1": 1.7097, "loss_cns_1": 0.65289, "loss_yns_1": 0.15415, "loss_cls_2": 0.90936, "loss_box_2": 1.67083, "loss_cns_2": 0.65681, "loss_yns_2": 0.15252, "loss_cls_3": 0.90347, "loss_box_3": 1.66349, "loss_cns_3": 0.65849, "loss_yns_3": 0.15244, "loss_cls_4": 0.91302, "loss_box_4": 1.65021, "loss_cns_4": 0.65691, "loss_yns_4": 0.15246, "loss_cls_5": 0.89767, "loss_box_5": 1.68046, "loss_cns_5": 0.66483, "loss_yns_5": 0.15342, "loss_cls_dn_0": 0.20245, "loss_box_dn_0": 0.75454, "loss_cls_dn_1": 0.13196, "loss_box_dn_1": 0.74689, "loss_cls_dn_2": 0.1307, "loss_box_dn_2": 0.72864, "loss_cls_dn_3": 0.13398, "loss_box_dn_3": 0.72891, "loss_cls_dn_4": 0.13849, "loss_box_dn_4": 0.73217, "loss_cls_dn_5": 0.14115, "loss_box_dn_5": 0.74358, "loss_dense_depth": 0.81759, "loss": 26.43318, "grad_norm": 30.482, "time": 1.49989} -{"mode": "train", "epoch": 1, "iter": 201, "lr": 0.00018, "memory": 49163, "data_time": 0.08031, "loss_cls_0": 0.81479, "loss_box_0": 1.70536, "loss_cns_0": 0.61716, "loss_yns_0": 0.15435, "loss_cls_1": 0.90368, "loss_box_1": 1.68296, "loss_cns_1": 0.64912, "loss_yns_1": 0.15359, "loss_cls_2": 0.90753, "loss_box_2": 1.67107, "loss_cns_2": 0.6536, "loss_yns_2": 0.15329, "loss_cls_3": 0.90263, "loss_box_3": 1.67202, "loss_cns_3": 0.65265, "loss_yns_3": 0.15191, "loss_cls_4": 0.91798, "loss_box_4": 1.65864, "loss_cns_4": 0.64968, "loss_yns_4": 0.15183, "loss_cls_5": 0.89626, "loss_box_5": 1.69022, "loss_cns_5": 0.65183, "loss_yns_5": 0.15307, "loss_cls_dn_0": 0.19641, "loss_box_dn_0": 0.75327, "loss_cls_dn_1": 0.12744, "loss_box_dn_1": 0.74859, "loss_cls_dn_2": 0.12636, "loss_box_dn_2": 0.74054, "loss_cls_dn_3": 0.12961, "loss_box_dn_3": 0.7441, "loss_cls_dn_4": 0.1343, "loss_box_dn_4": 0.74867, "loss_cls_dn_5": 0.13581, "loss_box_dn_5": 0.75745, "loss_dense_depth": 0.7628, "loss": 26.32056, "grad_norm": 37.29288, "time": 1.55751} -{"mode": "train", "epoch": 1, "iter": 202, "lr": 0.00018, "memory": 49163, "data_time": 0.08045, "loss_cls_0": 0.80967, "loss_box_0": 1.67743, "loss_cns_0": 0.61219, "loss_yns_0": 0.15333, "loss_cls_1": 0.90507, "loss_box_1": 1.66436, "loss_cns_1": 0.64402, "loss_yns_1": 0.15365, "loss_cls_2": 0.90808, "loss_box_2": 1.64768, "loss_cns_2": 0.65129, "loss_yns_2": 0.15323, "loss_cls_3": 0.9064, "loss_box_3": 1.64768, "loss_cns_3": 0.64884, "loss_yns_3": 0.15251, "loss_cls_4": 0.91093, "loss_box_4": 1.64489, "loss_cns_4": 0.6486, "loss_yns_4": 0.1526, "loss_cls_5": 0.90512, "loss_box_5": 1.66913, "loss_cns_5": 0.64982, "loss_yns_5": 0.15432, "loss_cls_dn_0": 0.19513, "loss_box_dn_0": 0.75755, "loss_cls_dn_1": 0.12798, "loss_box_dn_1": 0.7535, "loss_cls_dn_2": 0.12693, "loss_box_dn_2": 0.74531, "loss_cls_dn_3": 0.1288, "loss_box_dn_3": 0.75543, "loss_cls_dn_4": 0.13264, "loss_box_dn_4": 0.77147, "loss_cls_dn_5": 0.1371, "loss_box_dn_5": 0.78367, "loss_dense_depth": 0.78533, "loss": 26.27166, "grad_norm": 38.56682, "time": 1.58463} -{"mode": "train", "epoch": 1, "iter": 203, "lr": 0.00018, "memory": 49163, "data_time": 0.07573, "loss_cls_0": 0.81266, "loss_box_0": 1.70122, "loss_cns_0": 0.61154, "loss_yns_0": 0.14997, "loss_cls_1": 0.89988, "loss_box_1": 1.71471, "loss_cns_1": 0.64227, "loss_yns_1": 0.15241, "loss_cls_2": 0.90574, "loss_box_2": 1.70255, "loss_cns_2": 0.64548, "loss_yns_2": 0.1515, "loss_cls_3": 0.90259, "loss_box_3": 1.70049, "loss_cns_3": 0.64747, "loss_yns_3": 0.15315, "loss_cls_4": 0.90431, "loss_box_4": 1.70842, "loss_cns_4": 0.6467, "loss_yns_4": 0.1544, "loss_cls_5": 0.89632, "loss_box_5": 1.72531, "loss_cns_5": 0.64789, "loss_yns_5": 0.15405, "loss_cls_dn_0": 0.1976, "loss_box_dn_0": 0.75612, "loss_cls_dn_1": 0.12722, "loss_box_dn_1": 0.76417, "loss_cls_dn_2": 0.12691, "loss_box_dn_2": 0.756, "loss_cls_dn_3": 0.12707, "loss_box_dn_3": 0.76258, "loss_cls_dn_4": 0.13037, "loss_box_dn_4": 0.77758, "loss_cls_dn_5": 0.13514, "loss_box_dn_5": 0.79484, "loss_dense_depth": 0.80731, "loss": 26.59393, "grad_norm": 34.73523, "time": 1.51449} -{"mode": "train", "epoch": 1, "iter": 204, "lr": 0.00018, "memory": 49163, "data_time": 0.07398, "loss_cls_0": 0.79979, "loss_box_0": 1.72969, "loss_cns_0": 0.62094, "loss_yns_0": 0.15242, "loss_cls_1": 0.89126, "loss_box_1": 1.68799, "loss_cns_1": 0.65209, "loss_yns_1": 0.15199, "loss_cls_2": 0.89497, "loss_box_2": 1.68829, "loss_cns_2": 0.6524, "loss_yns_2": 0.15238, "loss_cls_3": 0.89285, "loss_box_3": 1.67451, "loss_cns_3": 0.65397, "loss_yns_3": 0.15306, "loss_cls_4": 0.8956, "loss_box_4": 1.67558, "loss_cns_4": 0.65156, "loss_yns_4": 0.1532, "loss_cls_5": 0.88568, "loss_box_5": 1.69262, "loss_cns_5": 0.65498, "loss_yns_5": 0.1521, "loss_cls_dn_0": 0.19887, "loss_box_dn_0": 0.76174, "loss_cls_dn_1": 0.11998, "loss_box_dn_1": 0.72127, "loss_cls_dn_2": 0.12186, "loss_box_dn_2": 0.72177, "loss_cls_dn_3": 0.12448, "loss_box_dn_3": 0.72776, "loss_cls_dn_4": 0.1274, "loss_box_dn_4": 0.74371, "loss_cls_dn_5": 0.13524, "loss_box_dn_5": 0.76605, "loss_dense_depth": 0.76421, "loss": 26.24424, "grad_norm": 38.64396, "time": 1.49613} -{"mode": "train", "epoch": 1, "iter": 205, "lr": 0.00018, "memory": 49163, "data_time": 0.08012, "loss_cls_0": 0.80825, "loss_box_0": 1.74694, "loss_cns_0": 0.61648, "loss_yns_0": 0.15364, "loss_cls_1": 0.89908, "loss_box_1": 1.69224, "loss_cns_1": 0.65402, "loss_yns_1": 0.15157, "loss_cls_2": 0.90182, "loss_box_2": 1.67779, "loss_cns_2": 0.65526, "loss_yns_2": 0.1501, "loss_cls_3": 0.89756, "loss_box_3": 1.66806, "loss_cns_3": 0.65643, "loss_yns_3": 0.15088, "loss_cls_4": 0.91783, "loss_box_4": 1.67469, "loss_cns_4": 0.65392, "loss_yns_4": 0.15245, "loss_cls_5": 0.89556, "loss_box_5": 1.68121, "loss_cns_5": 0.65709, "loss_yns_5": 0.15202, "loss_cls_dn_0": 0.20376, "loss_box_dn_0": 0.7655, "loss_cls_dn_1": 0.12846, "loss_box_dn_1": 0.73768, "loss_cls_dn_2": 0.13082, "loss_box_dn_2": 0.73109, "loss_cls_dn_3": 0.13549, "loss_box_dn_3": 0.73786, "loss_cls_dn_4": 0.1362, "loss_box_dn_4": 0.75345, "loss_cls_dn_5": 0.1442, "loss_box_dn_5": 0.76864, "loss_dense_depth": 0.80763, "loss": 26.44567, "grad_norm": 41.50859, "time": 1.5181} -{"mode": "train", "epoch": 1, "iter": 206, "lr": 0.00018, "memory": 49163, "data_time": 0.07963, "loss_cls_0": 0.81602, "loss_box_0": 1.72814, "loss_cns_0": 0.61956, "loss_yns_0": 0.15292, "loss_cls_1": 0.89793, "loss_box_1": 1.70137, "loss_cns_1": 0.65331, "loss_yns_1": 0.15313, "loss_cls_2": 0.90671, "loss_box_2": 1.67791, "loss_cns_2": 0.65558, "loss_yns_2": 0.15087, "loss_cls_3": 0.90624, "loss_box_3": 1.6668, "loss_cns_3": 0.6576, "loss_yns_3": 0.15248, "loss_cls_4": 0.90572, "loss_box_4": 1.6717, "loss_cns_4": 0.65458, "loss_yns_4": 0.15311, "loss_cls_5": 0.90256, "loss_box_5": 1.66387, "loss_cns_5": 0.6546, "loss_yns_5": 0.15381, "loss_cls_dn_0": 0.20387, "loss_box_dn_0": 0.761, "loss_cls_dn_1": 0.12853, "loss_box_dn_1": 0.7384, "loss_cls_dn_2": 0.13029, "loss_box_dn_2": 0.72899, "loss_cls_dn_3": 0.13474, "loss_box_dn_3": 0.72888, "loss_cls_dn_4": 0.1336, "loss_box_dn_4": 0.73821, "loss_cls_dn_5": 0.13774, "loss_box_dn_5": 0.74537, "loss_dense_depth": 0.75664, "loss": 26.32278, "grad_norm": 35.51291, "time": 1.51556} -{"mode": "train", "epoch": 1, "iter": 207, "lr": 0.00018, "memory": 49163, "data_time": 0.08817, "loss_cls_0": 0.80526, "loss_box_0": 1.72579, "loss_cns_0": 0.6189, "loss_yns_0": 0.15243, "loss_cls_1": 0.89315, "loss_box_1": 1.70526, "loss_cns_1": 0.65244, "loss_yns_1": 0.15175, "loss_cls_2": 0.91072, "loss_box_2": 1.67737, "loss_cns_2": 0.6547, "loss_yns_2": 0.1525, "loss_cls_3": 0.90159, "loss_box_3": 1.65969, "loss_cns_3": 0.65428, "loss_yns_3": 0.15323, "loss_cls_4": 0.90334, "loss_box_4": 1.66952, "loss_cns_4": 0.65707, "loss_yns_4": 0.15379, "loss_cls_5": 0.90165, "loss_box_5": 1.66151, "loss_cns_5": 0.65218, "loss_yns_5": 0.1527, "loss_cls_dn_0": 0.20118, "loss_box_dn_0": 0.75758, "loss_cls_dn_1": 0.12952, "loss_box_dn_1": 0.73896, "loss_cls_dn_2": 0.13145, "loss_box_dn_2": 0.72068, "loss_cls_dn_3": 0.13255, "loss_box_dn_3": 0.71682, "loss_cls_dn_4": 0.13522, "loss_box_dn_4": 0.72359, "loss_cls_dn_5": 0.13901, "loss_box_dn_5": 0.72669, "loss_dense_depth": 0.77035, "loss": 26.24439, "grad_norm": 35.90817, "time": 1.50475} -{"mode": "train", "epoch": 1, "iter": 208, "lr": 0.00018, "memory": 49163, "data_time": 0.07708, "loss_cls_0": 0.81662, "loss_box_0": 1.74468, "loss_cns_0": 0.62127, "loss_yns_0": 0.15606, "loss_cls_1": 0.8973, "loss_box_1": 1.6957, "loss_cns_1": 0.65083, "loss_yns_1": 0.15718, "loss_cls_2": 0.90632, "loss_box_2": 1.67601, "loss_cns_2": 0.65162, "loss_yns_2": 0.15595, "loss_cls_3": 0.90727, "loss_box_3": 1.65383, "loss_cns_3": 0.65136, "loss_yns_3": 0.15571, "loss_cls_4": 0.90424, "loss_box_4": 1.66338, "loss_cns_4": 0.65402, "loss_yns_4": 0.15614, "loss_cls_5": 0.90176, "loss_box_5": 1.66819, "loss_cns_5": 0.65167, "loss_yns_5": 0.15609, "loss_cls_dn_0": 0.20381, "loss_box_dn_0": 0.76185, "loss_cls_dn_1": 0.12657, "loss_box_dn_1": 0.71018, "loss_cls_dn_2": 0.12781, "loss_box_dn_2": 0.69559, "loss_cls_dn_3": 0.12841, "loss_box_dn_3": 0.6949, "loss_cls_dn_4": 0.13286, "loss_box_dn_4": 0.7018, "loss_cls_dn_5": 0.14251, "loss_box_dn_5": 0.71031, "loss_dense_depth": 0.7502, "loss": 26.13998, "grad_norm": 29.61605, "time": 1.49832} -{"mode": "train", "epoch": 1, "iter": 209, "lr": 0.00018, "memory": 49163, "data_time": 0.12412, "loss_cls_0": 0.83365, "loss_box_0": 1.76321, "loss_cns_0": 0.61469, "loss_yns_0": 0.15276, "loss_cls_1": 0.91432, "loss_box_1": 1.70068, "loss_cns_1": 0.6485, "loss_yns_1": 0.15205, "loss_cls_2": 0.92762, "loss_box_2": 1.69771, "loss_cns_2": 0.65045, "loss_yns_2": 0.15279, "loss_cls_3": 0.92162, "loss_box_3": 1.67279, "loss_cns_3": 0.64883, "loss_yns_3": 0.15182, "loss_cls_4": 0.9322, "loss_box_4": 1.67543, "loss_cns_4": 0.64803, "loss_yns_4": 0.15267, "loss_cls_5": 0.9198, "loss_box_5": 1.69045, "loss_cns_5": 0.65047, "loss_yns_5": 0.15303, "loss_cls_dn_0": 0.20325, "loss_box_dn_0": 0.75703, "loss_cls_dn_1": 0.12537, "loss_box_dn_1": 0.70897, "loss_cls_dn_2": 0.12671, "loss_box_dn_2": 0.69922, "loss_cls_dn_3": 0.12746, "loss_box_dn_3": 0.70051, "loss_cls_dn_4": 0.13126, "loss_box_dn_4": 0.70665, "loss_cls_dn_5": 0.13653, "loss_box_dn_5": 0.72264, "loss_dense_depth": 0.78276, "loss": 26.35395, "grad_norm": 41.29715, "time": 1.53697} -{"mode": "train", "epoch": 1, "iter": 210, "lr": 0.00018, "memory": 49163, "data_time": 0.07929, "loss_cls_0": 0.79868, "loss_box_0": 1.70567, "loss_cns_0": 0.62247, "loss_yns_0": 0.15211, "loss_cls_1": 0.87837, "loss_box_1": 1.67673, "loss_cns_1": 0.64738, "loss_yns_1": 0.15108, "loss_cls_2": 0.89496, "loss_box_2": 1.65034, "loss_cns_2": 0.65301, "loss_yns_2": 0.15109, "loss_cls_3": 0.87895, "loss_box_3": 1.63136, "loss_cns_3": 0.65317, "loss_yns_3": 0.15034, "loss_cls_4": 0.90222, "loss_box_4": 1.62573, "loss_cns_4": 0.64903, "loss_yns_4": 0.15065, "loss_cls_5": 0.88469, "loss_box_5": 1.6371, "loss_cns_5": 0.65346, "loss_yns_5": 0.1531, "loss_cls_dn_0": 0.18985, "loss_box_dn_0": 0.75051, "loss_cls_dn_1": 0.12122, "loss_box_dn_1": 0.71529, "loss_cls_dn_2": 0.12289, "loss_box_dn_2": 0.70376, "loss_cls_dn_3": 0.12478, "loss_box_dn_3": 0.7091, "loss_cls_dn_4": 0.12829, "loss_box_dn_4": 0.71387, "loss_cls_dn_5": 0.13079, "loss_box_dn_5": 0.72833, "loss_dense_depth": 0.72265, "loss": 25.81302, "grad_norm": 32.95994, "time": 1.49487} -{"mode": "train", "epoch": 1, "iter": 211, "lr": 0.00018, "memory": 49163, "data_time": 0.0815, "loss_cls_0": 0.80078, "loss_box_0": 1.71576, "loss_cns_0": 0.6209, "loss_yns_0": 0.15152, "loss_cls_1": 0.8655, "loss_box_1": 1.64133, "loss_cns_1": 0.64887, "loss_yns_1": 0.15121, "loss_cls_2": 0.87883, "loss_box_2": 1.61208, "loss_cns_2": 0.65401, "loss_yns_2": 0.15196, "loss_cls_3": 0.88317, "loss_box_3": 1.60173, "loss_cns_3": 0.65857, "loss_yns_3": 0.15106, "loss_cls_4": 0.8899, "loss_box_4": 1.60378, "loss_cns_4": 0.65343, "loss_yns_4": 0.15103, "loss_cls_5": 0.89151, "loss_box_5": 1.60233, "loss_cns_5": 0.65465, "loss_yns_5": 0.15385, "loss_cls_dn_0": 0.18785, "loss_box_dn_0": 0.75517, "loss_cls_dn_1": 0.11739, "loss_box_dn_1": 0.71412, "loss_cls_dn_2": 0.11564, "loss_box_dn_2": 0.70083, "loss_cls_dn_3": 0.12004, "loss_box_dn_3": 0.71238, "loss_cls_dn_4": 0.12434, "loss_box_dn_4": 0.71969, "loss_cls_dn_5": 0.12962, "loss_box_dn_5": 0.7329, "loss_dense_depth": 0.77084, "loss": 25.68858, "grad_norm": 49.84832, "time": 1.49389} -{"mode": "train", "epoch": 1, "iter": 212, "lr": 0.00018, "memory": 49163, "data_time": 0.0788, "loss_cls_0": 0.80633, "loss_box_0": 1.71922, "loss_cns_0": 0.62502, "loss_yns_0": 0.14963, "loss_cls_1": 0.86712, "loss_box_1": 1.69159, "loss_cns_1": 0.64407, "loss_yns_1": 0.1496, "loss_cls_2": 0.88241, "loss_box_2": 1.65434, "loss_cns_2": 0.6487, "loss_yns_2": 0.14983, "loss_cls_3": 0.88887, "loss_box_3": 1.64275, "loss_cns_3": 0.65176, "loss_yns_3": 0.14988, "loss_cls_4": 0.88815, "loss_box_4": 1.64376, "loss_cns_4": 0.6504, "loss_yns_4": 0.14908, "loss_cls_5": 0.89113, "loss_box_5": 1.65892, "loss_cns_5": 0.65018, "loss_yns_5": 0.14957, "loss_cls_dn_0": 0.19455, "loss_box_dn_0": 0.75302, "loss_cls_dn_1": 0.12377, "loss_box_dn_1": 0.73003, "loss_cls_dn_2": 0.12161, "loss_box_dn_2": 0.7117, "loss_cls_dn_3": 0.12415, "loss_box_dn_3": 0.7209, "loss_cls_dn_4": 0.12754, "loss_box_dn_4": 0.73129, "loss_cls_dn_5": 0.13051, "loss_box_dn_5": 0.74725, "loss_dense_depth": 0.75309, "loss": 25.97172, "grad_norm": 42.33492, "time": 1.49223} -{"mode": "train", "epoch": 1, "iter": 213, "lr": 0.00018, "memory": 49163, "data_time": 0.07952, "loss_cls_0": 0.80632, "loss_box_0": 1.72428, "loss_cns_0": 0.62622, "loss_yns_0": 0.15287, "loss_cls_1": 0.87874, "loss_box_1": 1.63586, "loss_cns_1": 0.64831, "loss_yns_1": 0.15125, "loss_cls_2": 0.909, "loss_box_2": 1.61577, "loss_cns_2": 0.64977, "loss_yns_2": 0.14974, "loss_cls_3": 0.90243, "loss_box_3": 1.60594, "loss_cns_3": 0.64907, "loss_yns_3": 0.14992, "loss_cls_4": 0.89017, "loss_box_4": 1.62494, "loss_cns_4": 0.65681, "loss_yns_4": 0.15137, "loss_cls_5": 0.89198, "loss_box_5": 1.63904, "loss_cns_5": 0.6535, "loss_yns_5": 0.15013, "loss_cls_dn_0": 0.19868, "loss_box_dn_0": 0.75088, "loss_cls_dn_1": 0.12616, "loss_box_dn_1": 0.7215, "loss_cls_dn_2": 0.12567, "loss_box_dn_2": 0.71454, "loss_cls_dn_3": 0.13019, "loss_box_dn_3": 0.7203, "loss_cls_dn_4": 0.13106, "loss_box_dn_4": 0.73293, "loss_cls_dn_5": 0.13647, "loss_box_dn_5": 0.75065, "loss_dense_depth": 0.78521, "loss": 25.93767, "grad_norm": 40.99518, "time": 1.48711} -{"mode": "train", "epoch": 1, "iter": 214, "lr": 0.00019, "memory": 49163, "data_time": 0.08231, "loss_cls_0": 0.81448, "loss_box_0": 1.72288, "loss_cns_0": 0.62326, "loss_yns_0": 0.15369, "loss_cls_1": 0.87416, "loss_box_1": 1.6864, "loss_cns_1": 0.64707, "loss_yns_1": 0.15157, "loss_cls_2": 0.89298, "loss_box_2": 1.62978, "loss_cns_2": 0.65242, "loss_yns_2": 0.15259, "loss_cls_3": 0.89179, "loss_box_3": 1.61585, "loss_cns_3": 0.65094, "loss_yns_3": 0.152, "loss_cls_4": 0.91441, "loss_box_4": 1.61791, "loss_cns_4": 0.655, "loss_yns_4": 0.15404, "loss_cls_5": 0.90591, "loss_box_5": 1.6203, "loss_cns_5": 0.65371, "loss_yns_5": 0.15225, "loss_cls_dn_0": 0.19314, "loss_box_dn_0": 0.76173, "loss_cls_dn_1": 0.12226, "loss_box_dn_1": 0.73663, "loss_cls_dn_2": 0.12151, "loss_box_dn_2": 0.72993, "loss_cls_dn_3": 0.12731, "loss_box_dn_3": 0.73082, "loss_cls_dn_4": 0.12957, "loss_box_dn_4": 0.73853, "loss_cls_dn_5": 0.13541, "loss_box_dn_5": 0.74734, "loss_dense_depth": 0.7569, "loss": 26.01645, "grad_norm": 45.24918, "time": 1.48839} -{"mode": "train", "epoch": 1, "iter": 215, "lr": 0.00019, "memory": 49163, "data_time": 0.0795, "loss_cls_0": 0.82837, "loss_box_0": 1.7229, "loss_cns_0": 0.62286, "loss_yns_0": 0.15237, "loss_cls_1": 0.88882, "loss_box_1": 1.65002, "loss_cns_1": 0.64782, "loss_yns_1": 0.15277, "loss_cls_2": 0.90187, "loss_box_2": 1.62297, "loss_cns_2": 0.64966, "loss_yns_2": 0.15418, "loss_cls_3": 0.90694, "loss_box_3": 1.62042, "loss_cns_3": 0.65226, "loss_yns_3": 0.15215, "loss_cls_4": 0.90828, "loss_box_4": 1.60076, "loss_cns_4": 0.65363, "loss_yns_4": 0.15263, "loss_cls_5": 0.90157, "loss_box_5": 1.61158, "loss_cns_5": 0.653, "loss_yns_5": 0.15266, "loss_cls_dn_0": 0.19993, "loss_box_dn_0": 0.74996, "loss_cls_dn_1": 0.12497, "loss_box_dn_1": 0.72378, "loss_cls_dn_2": 0.12549, "loss_box_dn_2": 0.70863, "loss_cls_dn_3": 0.1285, "loss_box_dn_3": 0.71101, "loss_cls_dn_4": 0.13143, "loss_box_dn_4": 0.71236, "loss_cls_dn_5": 0.13963, "loss_box_dn_5": 0.71704, "loss_dense_depth": 0.80848, "loss": 25.94171, "grad_norm": 34.90634, "time": 1.54777} -{"mode": "train", "epoch": 1, "iter": 216, "lr": 0.00019, "memory": 49163, "data_time": 0.07943, "loss_cls_0": 0.8349, "loss_box_0": 1.73733, "loss_cns_0": 0.62143, "loss_yns_0": 0.15429, "loss_cls_1": 0.90386, "loss_box_1": 1.64107, "loss_cns_1": 0.652, "loss_yns_1": 0.15372, "loss_cls_2": 0.92004, "loss_box_2": 1.62901, "loss_cns_2": 0.65373, "loss_yns_2": 0.15353, "loss_cls_3": 0.92084, "loss_box_3": 1.62588, "loss_cns_3": 0.65112, "loss_yns_3": 0.15377, "loss_cls_4": 0.91602, "loss_box_4": 1.62102, "loss_cns_4": 0.65588, "loss_yns_4": 0.15398, "loss_cls_5": 0.90883, "loss_box_5": 1.62923, "loss_cns_5": 0.65166, "loss_yns_5": 0.15424, "loss_cls_dn_0": 0.20177, "loss_box_dn_0": 0.76291, "loss_cls_dn_1": 0.13028, "loss_box_dn_1": 0.73, "loss_cls_dn_2": 0.12936, "loss_box_dn_2": 0.71376, "loss_cls_dn_3": 0.13005, "loss_box_dn_3": 0.71212, "loss_cls_dn_4": 0.1334, "loss_box_dn_4": 0.71539, "loss_cls_dn_5": 0.13812, "loss_box_dn_5": 0.71932, "loss_dense_depth": 0.78766, "loss": 26.1015, "grad_norm": 35.57066, "time": 1.49268} -{"mode": "train", "epoch": 1, "iter": 217, "lr": 0.00019, "memory": 49163, "data_time": 0.07827, "loss_cls_0": 0.80082, "loss_box_0": 1.70077, "loss_cns_0": 0.62394, "loss_yns_0": 0.15715, "loss_cls_1": 0.87486, "loss_box_1": 1.62104, "loss_cns_1": 0.65184, "loss_yns_1": 0.15619, "loss_cls_2": 0.88246, "loss_box_2": 1.60778, "loss_cns_2": 0.65151, "loss_yns_2": 0.15557, "loss_cls_3": 0.87482, "loss_box_3": 1.58775, "loss_cns_3": 0.65198, "loss_yns_3": 0.15649, "loss_cls_4": 0.87595, "loss_box_4": 1.59555, "loss_cns_4": 0.65487, "loss_yns_4": 0.15567, "loss_cls_5": 0.88187, "loss_box_5": 1.58727, "loss_cns_5": 0.6544, "loss_yns_5": 0.15815, "loss_cls_dn_0": 0.19322, "loss_box_dn_0": 0.74672, "loss_cls_dn_1": 0.12361, "loss_box_dn_1": 0.70453, "loss_cls_dn_2": 0.12242, "loss_box_dn_2": 0.6944, "loss_cls_dn_3": 0.12165, "loss_box_dn_3": 0.68823, "loss_cls_dn_4": 0.1271, "loss_box_dn_4": 0.69678, "loss_cls_dn_5": 0.13055, "loss_box_dn_5": 0.69541, "loss_dense_depth": 0.71761, "loss": 25.48094, "grad_norm": 33.2232, "time": 1.53669} -{"mode": "train", "epoch": 1, "iter": 218, "lr": 0.00019, "memory": 49163, "data_time": 0.07909, "loss_cls_0": 0.79545, "loss_box_0": 1.68771, "loss_cns_0": 0.62723, "loss_yns_0": 0.15496, "loss_cls_1": 0.87204, "loss_box_1": 1.62523, "loss_cns_1": 0.65205, "loss_yns_1": 0.15486, "loss_cls_2": 0.88179, "loss_box_2": 1.6116, "loss_cns_2": 0.65291, "loss_yns_2": 0.15361, "loss_cls_3": 0.87841, "loss_box_3": 1.59784, "loss_cns_3": 0.65242, "loss_yns_3": 0.15423, "loss_cls_4": 0.88292, "loss_box_4": 1.60257, "loss_cns_4": 0.65468, "loss_yns_4": 0.15466, "loss_cls_5": 0.88422, "loss_box_5": 1.59612, "loss_cns_5": 0.65547, "loss_yns_5": 0.15457, "loss_cls_dn_0": 0.19193, "loss_box_dn_0": 0.74642, "loss_cls_dn_1": 0.12574, "loss_box_dn_1": 0.67789, "loss_cls_dn_2": 0.12738, "loss_box_dn_2": 0.66874, "loss_cls_dn_3": 0.12556, "loss_box_dn_3": 0.66878, "loss_cls_dn_4": 0.13102, "loss_box_dn_4": 0.67952, "loss_cls_dn_5": 0.13677, "loss_box_dn_5": 0.68196, "loss_dense_depth": 0.75597, "loss": 25.45523, "grad_norm": 35.36411, "time": 1.49219} -{"mode": "train", "epoch": 1, "iter": 219, "lr": 0.00019, "memory": 49163, "data_time": 0.08101, "loss_cls_0": 0.81991, "loss_box_0": 1.72383, "loss_cns_0": 0.62145, "loss_yns_0": 0.15403, "loss_cls_1": 0.89668, "loss_box_1": 1.66055, "loss_cns_1": 0.65209, "loss_yns_1": 0.15357, "loss_cls_2": 0.9087, "loss_box_2": 1.63517, "loss_cns_2": 0.65526, "loss_yns_2": 0.1533, "loss_cls_3": 0.91145, "loss_box_3": 1.6257, "loss_cns_3": 0.65558, "loss_yns_3": 0.15204, "loss_cls_4": 0.91606, "loss_box_4": 1.63531, "loss_cns_4": 0.65937, "loss_yns_4": 0.15463, "loss_cls_5": 0.91695, "loss_box_5": 1.64258, "loss_cns_5": 0.65957, "loss_yns_5": 0.1534, "loss_cls_dn_0": 0.19821, "loss_box_dn_0": 0.7491, "loss_cls_dn_1": 0.12637, "loss_box_dn_1": 0.67813, "loss_cls_dn_2": 0.12589, "loss_box_dn_2": 0.67054, "loss_cls_dn_3": 0.12674, "loss_box_dn_3": 0.67707, "loss_cls_dn_4": 0.13064, "loss_box_dn_4": 0.69191, "loss_cls_dn_5": 0.13971, "loss_box_dn_5": 0.70402, "loss_dense_depth": 0.74039, "loss": 25.87589, "grad_norm": 31.44639, "time": 1.48856} -{"mode": "train", "epoch": 1, "iter": 220, "lr": 0.00019, "memory": 49163, "data_time": 0.08096, "loss_cls_0": 0.79022, "loss_box_0": 1.70036, "loss_cns_0": 0.62592, "loss_yns_0": 0.15314, "loss_cls_1": 0.87837, "loss_box_1": 1.65629, "loss_cns_1": 0.65521, "loss_yns_1": 0.15428, "loss_cls_2": 0.88704, "loss_box_2": 1.639, "loss_cns_2": 0.65481, "loss_yns_2": 0.15357, "loss_cls_3": 0.88749, "loss_box_3": 1.62287, "loss_cns_3": 0.65523, "loss_yns_3": 0.15245, "loss_cls_4": 0.87943, "loss_box_4": 1.63174, "loss_cns_4": 0.65539, "loss_yns_4": 0.15283, "loss_cls_5": 0.89149, "loss_box_5": 1.64146, "loss_cns_5": 0.65397, "loss_yns_5": 0.15289, "loss_cls_dn_0": 0.18987, "loss_box_dn_0": 0.74177, "loss_cls_dn_1": 0.12075, "loss_box_dn_1": 0.69268, "loss_cls_dn_2": 0.11912, "loss_box_dn_2": 0.68992, "loss_cls_dn_3": 0.11961, "loss_box_dn_3": 0.6936, "loss_cls_dn_4": 0.12246, "loss_box_dn_4": 0.70926, "loss_cls_dn_5": 0.13021, "loss_box_dn_5": 0.72201, "loss_dense_depth": 0.75541, "loss": 25.73212, "grad_norm": 35.45487, "time": 1.48637} -{"mode": "train", "epoch": 1, "iter": 221, "lr": 0.00019, "memory": 49163, "data_time": 0.07303, "loss_cls_0": 0.79593, "loss_box_0": 1.68992, "loss_cns_0": 0.61896, "loss_yns_0": 0.1528, "loss_cls_1": 0.88242, "loss_box_1": 1.66332, "loss_cns_1": 0.65454, "loss_yns_1": 0.15167, "loss_cls_2": 0.89065, "loss_box_2": 1.65059, "loss_cns_2": 0.65334, "loss_yns_2": 0.15229, "loss_cls_3": 0.88678, "loss_box_3": 1.63302, "loss_cns_3": 0.65387, "loss_yns_3": 0.15152, "loss_cls_4": 0.88826, "loss_box_4": 1.62684, "loss_cns_4": 0.65663, "loss_yns_4": 0.15105, "loss_cls_5": 0.89139, "loss_box_5": 1.63739, "loss_cns_5": 0.65237, "loss_yns_5": 0.15267, "loss_cls_dn_0": 0.19257, "loss_box_dn_0": 0.74545, "loss_cls_dn_1": 0.12756, "loss_box_dn_1": 0.70693, "loss_cls_dn_2": 0.12342, "loss_box_dn_2": 0.69919, "loss_cls_dn_3": 0.12508, "loss_box_dn_3": 0.70313, "loss_cls_dn_4": 0.12942, "loss_box_dn_4": 0.71237, "loss_cls_dn_5": 0.13152, "loss_box_dn_5": 0.72502, "loss_dense_depth": 0.75816, "loss": 25.81801, "grad_norm": 32.172, "time": 1.53059} -{"mode": "train", "epoch": 1, "iter": 222, "lr": 0.00019, "memory": 49163, "data_time": 0.07521, "loss_cls_0": 0.79197, "loss_box_0": 1.68767, "loss_cns_0": 0.61785, "loss_yns_0": 0.15083, "loss_cls_1": 0.87679, "loss_box_1": 1.6176, "loss_cns_1": 0.65722, "loss_yns_1": 0.15012, "loss_cls_2": 0.88941, "loss_box_2": 1.59923, "loss_cns_2": 0.65868, "loss_yns_2": 0.14977, "loss_cls_3": 0.89222, "loss_box_3": 1.59034, "loss_cns_3": 0.65885, "loss_yns_3": 0.14957, "loss_cls_4": 0.89391, "loss_box_4": 1.57495, "loss_cns_4": 0.65977, "loss_yns_4": 0.15041, "loss_cls_5": 0.88841, "loss_box_5": 1.59447, "loss_cns_5": 0.6583, "loss_yns_5": 0.15016, "loss_cls_dn_0": 0.19443, "loss_box_dn_0": 0.74679, "loss_cls_dn_1": 0.12485, "loss_box_dn_1": 0.70741, "loss_cls_dn_2": 0.12305, "loss_box_dn_2": 0.6948, "loss_cls_dn_3": 0.12613, "loss_box_dn_3": 0.6984, "loss_cls_dn_4": 0.1306, "loss_box_dn_4": 0.70234, "loss_cls_dn_5": 0.1338, "loss_box_dn_5": 0.71164, "loss_dense_depth": 0.75272, "loss": 25.55546, "grad_norm": 32.64032, "time": 1.56172} -{"mode": "train", "epoch": 1, "iter": 223, "lr": 0.00019, "memory": 49163, "data_time": 0.07533, "loss_cls_0": 0.81324, "loss_box_0": 1.73278, "loss_cns_0": 0.62038, "loss_yns_0": 0.15281, "loss_cls_1": 0.88364, "loss_box_1": 1.66448, "loss_cns_1": 0.65457, "loss_yns_1": 0.15252, "loss_cls_2": 0.89419, "loss_box_2": 1.63064, "loss_cns_2": 0.65565, "loss_yns_2": 0.15202, "loss_cls_3": 0.90153, "loss_box_3": 1.60937, "loss_cns_3": 0.65741, "loss_yns_3": 0.15103, "loss_cls_4": 0.90645, "loss_box_4": 1.61343, "loss_cns_4": 0.6576, "loss_yns_4": 0.15267, "loss_cls_5": 0.90559, "loss_box_5": 1.60658, "loss_cns_5": 0.65776, "loss_yns_5": 0.15201, "loss_cls_dn_0": 0.19805, "loss_box_dn_0": 0.75192, "loss_cls_dn_1": 0.12835, "loss_box_dn_1": 0.72217, "loss_cls_dn_2": 0.12791, "loss_box_dn_2": 0.71053, "loss_cls_dn_3": 0.13075, "loss_box_dn_3": 0.70855, "loss_cls_dn_4": 0.13356, "loss_box_dn_4": 0.71288, "loss_cls_dn_5": 0.1373, "loss_box_dn_5": 0.713, "loss_dense_depth": 0.77709, "loss": 25.93042, "grad_norm": 29.03014, "time": 1.52728} -{"mode": "train", "epoch": 1, "iter": 224, "lr": 0.00019, "memory": 49163, "data_time": 0.07845, "loss_cls_0": 0.78437, "loss_box_0": 1.69779, "loss_cns_0": 0.62381, "loss_yns_0": 0.15175, "loss_cls_1": 0.86342, "loss_box_1": 1.66689, "loss_cns_1": 0.65363, "loss_yns_1": 0.15291, "loss_cls_2": 0.88122, "loss_box_2": 1.63591, "loss_cns_2": 0.65717, "loss_yns_2": 0.151, "loss_cls_3": 0.88876, "loss_box_3": 1.62307, "loss_cns_3": 0.65801, "loss_yns_3": 0.15097, "loss_cls_4": 0.88769, "loss_box_4": 1.6276, "loss_cns_4": 0.65762, "loss_yns_4": 0.15101, "loss_cls_5": 0.89315, "loss_box_5": 1.62874, "loss_cns_5": 0.657, "loss_yns_5": 0.15167, "loss_cls_dn_0": 0.18672, "loss_box_dn_0": 0.74981, "loss_cls_dn_1": 0.12252, "loss_box_dn_1": 0.70419, "loss_cls_dn_2": 0.12255, "loss_box_dn_2": 0.68976, "loss_cls_dn_3": 0.12284, "loss_box_dn_3": 0.68903, "loss_cls_dn_4": 0.12499, "loss_box_dn_4": 0.69187, "loss_cls_dn_5": 0.13211, "loss_box_dn_5": 0.69362, "loss_dense_depth": 0.73887, "loss": 25.66404, "grad_norm": 30.23482, "time": 1.50473} -{"mode": "train", "epoch": 1, "iter": 225, "lr": 0.00019, "memory": 49163, "data_time": 0.08008, "loss_cls_0": 0.78173, "loss_box_0": 1.70961, "loss_cns_0": 0.62252, "loss_yns_0": 0.15235, "loss_cls_1": 0.86902, "loss_box_1": 1.66638, "loss_cns_1": 0.65019, "loss_yns_1": 0.15314, "loss_cls_2": 0.88571, "loss_box_2": 1.62084, "loss_cns_2": 0.6564, "loss_yns_2": 0.15169, "loss_cls_3": 0.88939, "loss_box_3": 1.61673, "loss_cns_3": 0.65757, "loss_yns_3": 0.15263, "loss_cls_4": 0.88692, "loss_box_4": 1.61393, "loss_cns_4": 0.65788, "loss_yns_4": 0.15327, "loss_cls_5": 0.8913, "loss_box_5": 1.62816, "loss_cns_5": 0.65811, "loss_yns_5": 0.15324, "loss_cls_dn_0": 0.18393, "loss_box_dn_0": 0.74517, "loss_cls_dn_1": 0.12262, "loss_box_dn_1": 0.6901, "loss_cls_dn_2": 0.12165, "loss_box_dn_2": 0.67445, "loss_cls_dn_3": 0.12044, "loss_box_dn_3": 0.68167, "loss_cls_dn_4": 0.12528, "loss_box_dn_4": 0.68647, "loss_cls_dn_5": 0.1287, "loss_box_dn_5": 0.69809, "loss_dense_depth": 0.74645, "loss": 25.60374, "grad_norm": 32.59467, "time": 1.51382} -{"mode": "train", "epoch": 1, "iter": 226, "lr": 0.00019, "memory": 49163, "data_time": 0.07835, "loss_cls_0": 0.80303, "loss_box_0": 1.69864, "loss_cns_0": 0.62777, "loss_yns_0": 0.15116, "loss_cls_1": 0.87851, "loss_box_1": 1.66816, "loss_cns_1": 0.65094, "loss_yns_1": 0.15166, "loss_cls_2": 0.89743, "loss_box_2": 1.62481, "loss_cns_2": 0.65494, "loss_yns_2": 0.14985, "loss_cls_3": 0.89448, "loss_box_3": 1.60708, "loss_cns_3": 0.65586, "loss_yns_3": 0.14949, "loss_cls_4": 0.89993, "loss_box_4": 1.61138, "loss_cns_4": 0.65679, "loss_yns_4": 0.15114, "loss_cls_5": 0.8963, "loss_box_5": 1.61636, "loss_cns_5": 0.655, "loss_yns_5": 0.15006, "loss_cls_dn_0": 0.18866, "loss_box_dn_0": 0.74754, "loss_cls_dn_1": 0.12531, "loss_box_dn_1": 0.6975, "loss_cls_dn_2": 0.12393, "loss_box_dn_2": 0.68691, "loss_cls_dn_3": 0.12332, "loss_box_dn_3": 0.6956, "loss_cls_dn_4": 0.12896, "loss_box_dn_4": 0.70569, "loss_cls_dn_5": 0.13018, "loss_box_dn_5": 0.71953, "loss_dense_depth": 0.73197, "loss": 25.70587, "grad_norm": 31.99336, "time": 1.53801} -{"mode": "train", "epoch": 1, "iter": 227, "lr": 0.00019, "memory": 49163, "data_time": 0.0875, "loss_cls_0": 0.8039, "loss_box_0": 1.7314, "loss_cns_0": 0.61955, "loss_yns_0": 0.15036, "loss_cls_1": 0.85893, "loss_box_1": 1.70379, "loss_cns_1": 0.6401, "loss_yns_1": 0.14847, "loss_cls_2": 0.887, "loss_box_2": 1.6612, "loss_cns_2": 0.64798, "loss_yns_2": 0.14721, "loss_cls_3": 0.88753, "loss_box_3": 1.64017, "loss_cns_3": 0.64801, "loss_yns_3": 0.14815, "loss_cls_4": 0.88869, "loss_box_4": 1.65606, "loss_cns_4": 0.64949, "loss_yns_4": 0.14948, "loss_cls_5": 0.88827, "loss_box_5": 1.65033, "loss_cns_5": 0.64856, "loss_yns_5": 0.14857, "loss_cls_dn_0": 0.18896, "loss_box_dn_0": 0.74569, "loss_cls_dn_1": 0.12305, "loss_box_dn_1": 0.71601, "loss_cls_dn_2": 0.12082, "loss_box_dn_2": 0.709, "loss_cls_dn_3": 0.12154, "loss_box_dn_3": 0.71829, "loss_cls_dn_4": 0.1271, "loss_box_dn_4": 0.73556, "loss_cls_dn_5": 0.13055, "loss_box_dn_5": 0.75031, "loss_dense_depth": 0.75572, "loss": 25.94578, "grad_norm": 38.08898, "time": 1.5113} -{"mode": "train", "epoch": 1, "iter": 228, "lr": 0.00019, "memory": 49163, "data_time": 0.09182, "loss_cls_0": 0.81658, "loss_box_0": 1.72468, "loss_cns_0": 0.62561, "loss_yns_0": 0.15379, "loss_cls_1": 0.8625, "loss_box_1": 1.70515, "loss_cns_1": 0.64531, "loss_yns_1": 0.15165, "loss_cls_2": 0.87201, "loss_box_2": 1.66806, "loss_cns_2": 0.65184, "loss_yns_2": 0.14973, "loss_cls_3": 0.87478, "loss_box_3": 1.66059, "loss_cns_3": 0.65003, "loss_yns_3": 0.1495, "loss_cls_4": 0.87625, "loss_box_4": 1.67034, "loss_cns_4": 0.64885, "loss_yns_4": 0.15104, "loss_cls_5": 0.8821, "loss_box_5": 1.67597, "loss_cns_5": 0.65086, "loss_yns_5": 0.15094, "loss_cls_dn_0": 0.19037, "loss_box_dn_0": 0.74316, "loss_cls_dn_1": 0.12162, "loss_box_dn_1": 0.7418, "loss_cls_dn_2": 0.12055, "loss_box_dn_2": 0.73163, "loss_cls_dn_3": 0.12213, "loss_box_dn_3": 0.74369, "loss_cls_dn_4": 0.12511, "loss_box_dn_4": 0.76031, "loss_cls_dn_5": 0.12811, "loss_box_dn_5": 0.77243, "loss_dense_depth": 0.74739, "loss": 26.11646, "grad_norm": 36.66411, "time": 1.50402} -{"mode": "train", "epoch": 1, "iter": 229, "lr": 0.00019, "memory": 49163, "data_time": 0.17826, "loss_cls_0": 0.81291, "loss_box_0": 1.72572, "loss_cns_0": 0.61733, "loss_yns_0": 0.1491, "loss_cls_1": 0.86039, "loss_box_1": 1.6672, "loss_cns_1": 0.64786, "loss_yns_1": 0.14988, "loss_cls_2": 0.87888, "loss_box_2": 1.63994, "loss_cns_2": 0.64798, "loss_yns_2": 0.14665, "loss_cls_3": 0.87574, "loss_box_3": 1.63883, "loss_cns_3": 0.64903, "loss_yns_3": 0.14741, "loss_cls_4": 0.87939, "loss_box_4": 1.63687, "loss_cns_4": 0.64924, "loss_yns_4": 0.14727, "loss_cls_5": 0.88967, "loss_box_5": 1.64133, "loss_cns_5": 0.65339, "loss_yns_5": 0.14772, "loss_cls_dn_0": 0.18423, "loss_box_dn_0": 0.73963, "loss_cls_dn_1": 0.12434, "loss_box_dn_1": 0.75845, "loss_cls_dn_2": 0.12311, "loss_box_dn_2": 0.74288, "loss_cls_dn_3": 0.12196, "loss_box_dn_3": 0.75266, "loss_cls_dn_4": 0.12744, "loss_box_dn_4": 0.76878, "loss_cls_dn_5": 0.13241, "loss_box_dn_5": 0.77821, "loss_dense_depth": 0.76235, "loss": 26.01617, "grad_norm": 37.82653, "time": 1.58339} -{"mode": "train", "epoch": 1, "iter": 230, "lr": 0.00019, "memory": 49163, "data_time": 0.0811, "loss_cls_0": 0.81354, "loss_box_0": 1.73625, "loss_cns_0": 0.61892, "loss_yns_0": 0.14949, "loss_cls_1": 0.86699, "loss_box_1": 1.6557, "loss_cns_1": 0.64906, "loss_yns_1": 0.14979, "loss_cls_2": 0.88129, "loss_box_2": 1.64048, "loss_cns_2": 0.65329, "loss_yns_2": 0.14953, "loss_cls_3": 0.87068, "loss_box_3": 1.63324, "loss_cns_3": 0.65371, "loss_yns_3": 0.1496, "loss_cls_4": 0.87235, "loss_box_4": 1.63095, "loss_cns_4": 0.65536, "loss_yns_4": 0.14926, "loss_cls_5": 0.87181, "loss_box_5": 1.64698, "loss_cns_5": 0.65743, "loss_yns_5": 0.15137, "loss_cls_dn_0": 0.18689, "loss_box_dn_0": 0.75134, "loss_cls_dn_1": 0.12202, "loss_box_dn_1": 0.74967, "loss_cls_dn_2": 0.11916, "loss_box_dn_2": 0.73224, "loss_cls_dn_3": 0.11863, "loss_box_dn_3": 0.73746, "loss_cls_dn_4": 0.12413, "loss_box_dn_4": 0.7476, "loss_cls_dn_5": 0.12653, "loss_box_dn_5": 0.75901, "loss_dense_depth": 0.76047, "loss": 25.94223, "grad_norm": 32.19631, "time": 1.48952} -{"mode": "train", "epoch": 1, "iter": 231, "lr": 0.00019, "memory": 49163, "data_time": 0.07784, "loss_cls_0": 0.7792, "loss_box_0": 1.69845, "loss_cns_0": 0.62444, "loss_yns_0": 0.15032, "loss_cls_1": 0.86439, "loss_box_1": 1.61194, "loss_cns_1": 0.65489, "loss_yns_1": 0.14808, "loss_cls_2": 0.86592, "loss_box_2": 1.60077, "loss_cns_2": 0.6584, "loss_yns_2": 0.14711, "loss_cls_3": 0.85601, "loss_box_3": 1.59075, "loss_cns_3": 0.65791, "loss_yns_3": 0.14662, "loss_cls_4": 0.86605, "loss_box_4": 1.59175, "loss_cns_4": 0.65781, "loss_yns_4": 0.14729, "loss_cls_5": 0.8607, "loss_box_5": 1.60781, "loss_cns_5": 0.65644, "loss_yns_5": 0.14688, "loss_cls_dn_0": 0.17936, "loss_box_dn_0": 0.7437, "loss_cls_dn_1": 0.11979, "loss_box_dn_1": 0.71171, "loss_cls_dn_2": 0.11605, "loss_box_dn_2": 0.69758, "loss_cls_dn_3": 0.11839, "loss_box_dn_3": 0.69813, "loss_cls_dn_4": 0.12406, "loss_box_dn_4": 0.70613, "loss_cls_dn_5": 0.12695, "loss_box_dn_5": 0.71627, "loss_dense_depth": 0.72735, "loss": 25.37542, "grad_norm": 36.20464, "time": 1.48796} -{"mode": "train", "epoch": 1, "iter": 232, "lr": 0.00019, "memory": 49163, "data_time": 0.07796, "loss_cls_0": 0.80279, "loss_box_0": 1.70831, "loss_cns_0": 0.62342, "loss_yns_0": 0.1504, "loss_cls_1": 0.8821, "loss_box_1": 1.62415, "loss_cns_1": 0.65693, "loss_yns_1": 0.14739, "loss_cls_2": 0.88357, "loss_box_2": 1.61484, "loss_cns_2": 0.65895, "loss_yns_2": 0.14876, "loss_cls_3": 0.87516, "loss_box_3": 1.5961, "loss_cns_3": 0.65809, "loss_yns_3": 0.14724, "loss_cls_4": 0.8786, "loss_box_4": 1.5974, "loss_cns_4": 0.65897, "loss_yns_4": 0.14893, "loss_cls_5": 0.87152, "loss_box_5": 1.61397, "loss_cns_5": 0.65845, "loss_yns_5": 0.14841, "loss_cls_dn_0": 0.1789, "loss_box_dn_0": 0.7556, "loss_cls_dn_1": 0.12175, "loss_box_dn_1": 0.7046, "loss_cls_dn_2": 0.12046, "loss_box_dn_2": 0.69463, "loss_cls_dn_3": 0.12212, "loss_box_dn_3": 0.69176, "loss_cls_dn_4": 0.1322, "loss_box_dn_4": 0.70138, "loss_cls_dn_5": 0.13442, "loss_box_dn_5": 0.71158, "loss_dense_depth": 0.75434, "loss": 25.57817, "grad_norm": 30.03135, "time": 1.4927} -{"mode": "train", "epoch": 1, "iter": 233, "lr": 0.00019, "memory": 49163, "data_time": 0.07818, "loss_cls_0": 0.81745, "loss_box_0": 1.69489, "loss_cns_0": 0.62417, "loss_yns_0": 0.15144, "loss_cls_1": 0.88827, "loss_box_1": 1.59968, "loss_cns_1": 0.65468, "loss_yns_1": 0.14987, "loss_cls_2": 0.8986, "loss_box_2": 1.58267, "loss_cns_2": 0.65627, "loss_yns_2": 0.14906, "loss_cls_3": 0.89605, "loss_box_3": 1.56015, "loss_cns_3": 0.65548, "loss_yns_3": 0.14777, "loss_cls_4": 0.89279, "loss_box_4": 1.56671, "loss_cns_4": 0.65615, "loss_yns_4": 0.14822, "loss_cls_5": 0.89483, "loss_box_5": 1.57162, "loss_cns_5": 0.6566, "loss_yns_5": 0.14882, "loss_cls_dn_0": 0.17617, "loss_box_dn_0": 0.75766, "loss_cls_dn_1": 0.12552, "loss_box_dn_1": 0.72494, "loss_cls_dn_2": 0.12414, "loss_box_dn_2": 0.70987, "loss_cls_dn_3": 0.12645, "loss_box_dn_3": 0.7072, "loss_cls_dn_4": 0.13075, "loss_box_dn_4": 0.71728, "loss_cls_dn_5": 0.13361, "loss_box_dn_5": 0.72584, "loss_dense_depth": 0.76112, "loss": 25.58278, "grad_norm": 34.63741, "time": 1.48096} -{"mode": "train", "epoch": 1, "iter": 234, "lr": 0.00019, "memory": 49163, "data_time": 0.07878, "loss_cls_0": 0.78057, "loss_box_0": 1.66414, "loss_cns_0": 0.62149, "loss_yns_0": 0.15048, "loss_cls_1": 0.87296, "loss_box_1": 1.56904, "loss_cns_1": 0.65376, "loss_yns_1": 0.1506, "loss_cls_2": 0.88112, "loss_box_2": 1.5387, "loss_cns_2": 0.65621, "loss_yns_2": 0.15009, "loss_cls_3": 0.88173, "loss_box_3": 1.52713, "loss_cns_3": 0.65595, "loss_yns_3": 0.14932, "loss_cls_4": 0.87866, "loss_box_4": 1.52488, "loss_cns_4": 0.65794, "loss_yns_4": 0.15024, "loss_cls_5": 0.88488, "loss_box_5": 1.53652, "loss_cns_5": 0.65671, "loss_yns_5": 0.15025, "loss_cls_dn_0": 0.17532, "loss_box_dn_0": 0.75257, "loss_cls_dn_1": 0.11669, "loss_box_dn_1": 0.71733, "loss_cls_dn_2": 0.11845, "loss_box_dn_2": 0.70206, "loss_cls_dn_3": 0.12178, "loss_box_dn_3": 0.70186, "loss_cls_dn_4": 0.12012, "loss_box_dn_4": 0.70988, "loss_cls_dn_5": 0.12704, "loss_box_dn_5": 0.72058, "loss_dense_depth": 0.74355, "loss": 25.1706, "grad_norm": 32.46733, "time": 1.55613} -{"mode": "train", "epoch": 1, "iter": 235, "lr": 0.00019, "memory": 49163, "data_time": 0.07642, "loss_cls_0": 0.8106, "loss_box_0": 1.67821, "loss_cns_0": 0.62066, "loss_yns_0": 0.15095, "loss_cls_1": 0.88086, "loss_box_1": 1.55014, "loss_cns_1": 0.65168, "loss_yns_1": 0.15162, "loss_cls_2": 0.88639, "loss_box_2": 1.55175, "loss_cns_2": 0.65326, "loss_yns_2": 0.15138, "loss_cls_3": 0.88032, "loss_box_3": 1.54402, "loss_cns_3": 0.65313, "loss_yns_3": 0.15182, "loss_cls_4": 0.8856, "loss_box_4": 1.54431, "loss_cns_4": 0.65651, "loss_yns_4": 0.15155, "loss_cls_5": 0.88059, "loss_box_5": 1.54929, "loss_cns_5": 0.6561, "loss_yns_5": 0.15296, "loss_cls_dn_0": 0.17859, "loss_box_dn_0": 0.74253, "loss_cls_dn_1": 0.11826, "loss_box_dn_1": 0.70124, "loss_cls_dn_2": 0.11953, "loss_box_dn_2": 0.69167, "loss_cls_dn_3": 0.12068, "loss_box_dn_3": 0.69626, "loss_cls_dn_4": 0.1213, "loss_box_dn_4": 0.7076, "loss_cls_dn_5": 0.12785, "loss_box_dn_5": 0.72165, "loss_dense_depth": 0.76968, "loss": 25.26054, "grad_norm": 29.2441, "time": 1.49498} -{"mode": "train", "epoch": 1, "iter": 236, "lr": 0.00019, "memory": 49163, "data_time": 0.07593, "loss_cls_0": 0.77448, "loss_box_0": 1.67959, "loss_cns_0": 0.62058, "loss_yns_0": 0.15322, "loss_cls_1": 0.8585, "loss_box_1": 1.56327, "loss_cns_1": 0.65295, "loss_yns_1": 0.1539, "loss_cls_2": 0.87378, "loss_box_2": 1.54686, "loss_cns_2": 0.65482, "loss_yns_2": 0.1519, "loss_cls_3": 0.86181, "loss_box_3": 1.54287, "loss_cns_3": 0.65577, "loss_yns_3": 0.15292, "loss_cls_4": 0.86565, "loss_box_4": 1.5447, "loss_cns_4": 0.65485, "loss_yns_4": 0.15355, "loss_cls_5": 0.86586, "loss_box_5": 1.54144, "loss_cns_5": 0.65489, "loss_yns_5": 0.15374, "loss_cls_dn_0": 0.16633, "loss_box_dn_0": 0.73437, "loss_cls_dn_1": 0.12187, "loss_box_dn_1": 0.70367, "loss_cls_dn_2": 0.11833, "loss_box_dn_2": 0.69366, "loss_cls_dn_3": 0.11645, "loss_box_dn_3": 0.70145, "loss_cls_dn_4": 0.11855, "loss_box_dn_4": 0.71465, "loss_cls_dn_5": 0.12332, "loss_box_dn_5": 0.7276, "loss_dense_depth": 0.77085, "loss": 25.14299, "grad_norm": 35.65234, "time": 1.47289} -{"mode": "train", "epoch": 1, "iter": 237, "lr": 0.00019, "memory": 49163, "data_time": 0.07661, "loss_cls_0": 0.77422, "loss_box_0": 1.65666, "loss_cns_0": 0.62299, "loss_yns_0": 0.15349, "loss_cls_1": 0.8548, "loss_box_1": 1.54468, "loss_cns_1": 0.6562, "loss_yns_1": 0.1539, "loss_cls_2": 0.87033, "loss_box_2": 1.51178, "loss_cns_2": 0.65909, "loss_yns_2": 0.15382, "loss_cls_3": 0.87236, "loss_box_3": 1.51097, "loss_cns_3": 0.65973, "loss_yns_3": 0.15286, "loss_cls_4": 0.87411, "loss_box_4": 1.50537, "loss_cns_4": 0.65904, "loss_yns_4": 0.15396, "loss_cls_5": 0.87993, "loss_box_5": 1.51577, "loss_cns_5": 0.6586, "loss_yns_5": 0.15427, "loss_cls_dn_0": 0.17545, "loss_box_dn_0": 0.74395, "loss_cls_dn_1": 0.1171, "loss_box_dn_1": 0.71901, "loss_cls_dn_2": 0.11411, "loss_box_dn_2": 0.70383, "loss_cls_dn_3": 0.11855, "loss_box_dn_3": 0.71649, "loss_cls_dn_4": 0.12116, "loss_box_dn_4": 0.7239, "loss_cls_dn_5": 0.12552, "loss_box_dn_5": 0.73531, "loss_dense_depth": 0.73158, "loss": 25.0549, "grad_norm": 34.80653, "time": 1.54972} -{"mode": "train", "epoch": 1, "iter": 238, "lr": 0.00019, "memory": 49163, "data_time": 0.08187, "loss_cls_0": 0.77396, "loss_box_0": 1.64256, "loss_cns_0": 0.62551, "loss_yns_0": 0.15411, "loss_cls_1": 0.8563, "loss_box_1": 1.54156, "loss_cns_1": 0.65917, "loss_yns_1": 0.15474, "loss_cls_2": 0.87084, "loss_box_2": 1.50619, "loss_cns_2": 0.6619, "loss_yns_2": 0.15309, "loss_cls_3": 0.86586, "loss_box_3": 1.49942, "loss_cns_3": 0.66262, "loss_yns_3": 0.15374, "loss_cls_4": 0.86891, "loss_box_4": 1.49499, "loss_cns_4": 0.66216, "loss_yns_4": 0.15422, "loss_cls_5": 0.87319, "loss_box_5": 1.50305, "loss_cns_5": 0.66152, "loss_yns_5": 0.15513, "loss_cls_dn_0": 0.17358, "loss_box_dn_0": 0.74102, "loss_cls_dn_1": 0.11547, "loss_box_dn_1": 0.73806, "loss_cls_dn_2": 0.11837, "loss_box_dn_2": 0.71669, "loss_cls_dn_3": 0.12059, "loss_box_dn_3": 0.7248, "loss_cls_dn_4": 0.12147, "loss_box_dn_4": 0.73017, "loss_cls_dn_5": 0.12751, "loss_box_dn_5": 0.73601, "loss_dense_depth": 0.74166, "loss": 25.06013, "grad_norm": 32.2392, "time": 1.48804} -{"mode": "train", "epoch": 1, "iter": 239, "lr": 0.0002, "memory": 49163, "data_time": 0.08126, "loss_cls_0": 0.77385, "loss_box_0": 1.61841, "loss_cns_0": 0.62914, "loss_yns_0": 0.15387, "loss_cls_1": 0.85482, "loss_box_1": 1.55153, "loss_cns_1": 0.65855, "loss_yns_1": 0.15533, "loss_cls_2": 0.87751, "loss_box_2": 1.49934, "loss_cns_2": 0.66268, "loss_yns_2": 0.15403, "loss_cls_3": 0.87921, "loss_box_3": 1.48089, "loss_cns_3": 0.66496, "loss_yns_3": 0.15342, "loss_cls_4": 0.87683, "loss_box_4": 1.48034, "loss_cns_4": 0.6647, "loss_yns_4": 0.15443, "loss_cls_5": 0.86633, "loss_box_5": 1.47935, "loss_cns_5": 0.66367, "loss_yns_5": 0.15485, "loss_cls_dn_0": 0.16872, "loss_box_dn_0": 0.73217, "loss_cls_dn_1": 0.11654, "loss_box_dn_1": 0.71651, "loss_cls_dn_2": 0.117, "loss_box_dn_2": 0.70245, "loss_cls_dn_3": 0.11898, "loss_box_dn_3": 0.7046, "loss_cls_dn_4": 0.11987, "loss_box_dn_4": 0.70958, "loss_cls_dn_5": 0.1244, "loss_box_dn_5": 0.71031, "loss_dense_depth": 0.72821, "loss": 24.87738, "grad_norm": 29.65655, "time": 1.50778} -{"mode": "train", "epoch": 1, "iter": 240, "lr": 0.0002, "memory": 49163, "data_time": 0.08155, "loss_cls_0": 0.77413, "loss_box_0": 1.62607, "loss_cns_0": 0.62666, "loss_yns_0": 0.15253, "loss_cls_1": 0.87377, "loss_box_1": 1.52641, "loss_cns_1": 0.66358, "loss_yns_1": 0.15392, "loss_cls_2": 0.88725, "loss_box_2": 1.49162, "loss_cns_2": 0.66481, "loss_yns_2": 0.15201, "loss_cls_3": 0.8858, "loss_box_3": 1.47285, "loss_cns_3": 0.66624, "loss_yns_3": 0.15194, "loss_cls_4": 0.88587, "loss_box_4": 1.46794, "loss_cns_4": 0.66519, "loss_yns_4": 0.15189, "loss_cls_5": 0.88144, "loss_box_5": 1.47599, "loss_cns_5": 0.66686, "loss_yns_5": 0.15413, "loss_cls_dn_0": 0.17093, "loss_box_dn_0": 0.74013, "loss_cls_dn_1": 0.11793, "loss_box_dn_1": 0.68122, "loss_cls_dn_2": 0.1188, "loss_box_dn_2": 0.67177, "loss_cls_dn_3": 0.12129, "loss_box_dn_3": 0.67563, "loss_cls_dn_4": 0.12059, "loss_box_dn_4": 0.68081, "loss_cls_dn_5": 0.1245, "loss_box_dn_5": 0.68634, "loss_dense_depth": 0.69744, "loss": 24.72625, "grad_norm": 31.73083, "time": 1.53252} -{"mode": "train", "epoch": 1, "iter": 241, "lr": 0.0002, "memory": 49163, "data_time": 0.07898, "loss_cls_0": 0.78975, "loss_box_0": 1.64062, "loss_cns_0": 0.62425, "loss_yns_0": 0.15531, "loss_cls_1": 0.87528, "loss_box_1": 1.56423, "loss_cns_1": 0.65926, "loss_yns_1": 0.15635, "loss_cls_2": 0.88879, "loss_box_2": 1.52003, "loss_cns_2": 0.66418, "loss_yns_2": 0.15551, "loss_cls_3": 0.89224, "loss_box_3": 1.50742, "loss_cns_3": 0.6645, "loss_yns_3": 0.15585, "loss_cls_4": 0.89065, "loss_box_4": 1.50166, "loss_cns_4": 0.66353, "loss_yns_4": 0.1554, "loss_cls_5": 0.89754, "loss_box_5": 1.51077, "loss_cns_5": 0.66643, "loss_yns_5": 0.15725, "loss_cls_dn_0": 0.17069, "loss_box_dn_0": 0.74241, "loss_cls_dn_1": 0.11963, "loss_box_dn_1": 0.69008, "loss_cls_dn_2": 0.11976, "loss_box_dn_2": 0.68131, "loss_cls_dn_3": 0.12081, "loss_box_dn_3": 0.68806, "loss_cls_dn_4": 0.12034, "loss_box_dn_4": 0.69748, "loss_cls_dn_5": 0.1241, "loss_box_dn_5": 0.70974, "loss_dense_depth": 0.76971, "loss": 25.11093, "grad_norm": 30.43493, "time": 1.54156} -{"mode": "train", "epoch": 1, "iter": 242, "lr": 0.0002, "memory": 49163, "data_time": 0.078, "loss_cls_0": 0.79323, "loss_box_0": 1.64433, "loss_cns_0": 0.6272, "loss_yns_0": 0.15553, "loss_cls_1": 0.87807, "loss_box_1": 1.55992, "loss_cns_1": 0.65872, "loss_yns_1": 0.15672, "loss_cls_2": 0.90531, "loss_box_2": 1.52027, "loss_cns_2": 0.66141, "loss_yns_2": 0.15594, "loss_cls_3": 0.89683, "loss_box_3": 1.51244, "loss_cns_3": 0.66179, "loss_yns_3": 0.15539, "loss_cls_4": 0.89659, "loss_box_4": 1.51456, "loss_cns_4": 0.66064, "loss_yns_4": 0.1549, "loss_cls_5": 0.90037, "loss_box_5": 1.52362, "loss_cns_5": 0.66067, "loss_yns_5": 0.15595, "loss_cls_dn_0": 0.16577, "loss_box_dn_0": 0.73983, "loss_cls_dn_1": 0.11798, "loss_box_dn_1": 0.70208, "loss_cls_dn_2": 0.11878, "loss_box_dn_2": 0.69439, "loss_cls_dn_3": 0.11824, "loss_box_dn_3": 0.70322, "loss_cls_dn_4": 0.12267, "loss_box_dn_4": 0.71752, "loss_cls_dn_5": 0.12634, "loss_box_dn_5": 0.73028, "loss_dense_depth": 0.70079, "loss": 25.1683, "grad_norm": 36.00968, "time": 1.57071} -{"mode": "train", "epoch": 1, "iter": 243, "lr": 0.0002, "memory": 49163, "data_time": 0.07665, "loss_cls_0": 0.79082, "loss_box_0": 1.62518, "loss_cns_0": 0.62417, "loss_yns_0": 0.15349, "loss_cls_1": 0.87441, "loss_box_1": 1.59387, "loss_cns_1": 0.65658, "loss_yns_1": 0.15429, "loss_cls_2": 0.89371, "loss_box_2": 1.5658, "loss_cns_2": 0.65875, "loss_yns_2": 0.15381, "loss_cls_3": 0.87894, "loss_box_3": 1.56204, "loss_cns_3": 0.65785, "loss_yns_3": 0.15353, "loss_cls_4": 0.8802, "loss_box_4": 1.56884, "loss_cns_4": 0.65639, "loss_yns_4": 0.15524, "loss_cls_5": 0.88232, "loss_box_5": 1.56998, "loss_cns_5": 0.65522, "loss_yns_5": 0.15445, "loss_cls_dn_0": 0.17055, "loss_box_dn_0": 0.73439, "loss_cls_dn_1": 0.12161, "loss_box_dn_1": 0.71208, "loss_cls_dn_2": 0.12185, "loss_box_dn_2": 0.70605, "loss_cls_dn_3": 0.1215, "loss_box_dn_3": 0.71487, "loss_cls_dn_4": 0.12791, "loss_box_dn_4": 0.73452, "loss_cls_dn_5": 0.12946, "loss_box_dn_5": 0.74551, "loss_dense_depth": 0.70335, "loss": 25.36353, "grad_norm": 37.27586, "time": 1.50348} -{"mode": "train", "epoch": 1, "iter": 244, "lr": 0.0002, "memory": 49163, "data_time": 0.07999, "loss_cls_0": 0.79576, "loss_box_0": 1.65756, "loss_cns_0": 0.61858, "loss_yns_0": 0.15678, "loss_cls_1": 0.88479, "loss_box_1": 1.59284, "loss_cns_1": 0.64876, "loss_yns_1": 0.15475, "loss_cls_2": 0.89589, "loss_box_2": 1.56562, "loss_cns_2": 0.6544, "loss_yns_2": 0.15551, "loss_cls_3": 0.8915, "loss_box_3": 1.54287, "loss_cns_3": 0.65162, "loss_yns_3": 0.15442, "loss_cls_4": 0.89198, "loss_box_4": 1.55881, "loss_cns_4": 0.65239, "loss_yns_4": 0.15468, "loss_cls_5": 0.90682, "loss_box_5": 1.56293, "loss_cns_5": 0.65229, "loss_yns_5": 0.15442, "loss_cls_dn_0": 0.17096, "loss_box_dn_0": 0.73992, "loss_cls_dn_1": 0.12091, "loss_box_dn_1": 0.73237, "loss_cls_dn_2": 0.12283, "loss_box_dn_2": 0.72492, "loss_cls_dn_3": 0.12335, "loss_box_dn_3": 0.72869, "loss_cls_dn_4": 0.12712, "loss_box_dn_4": 0.74395, "loss_cls_dn_5": 0.12874, "loss_box_dn_5": 0.75332, "loss_dense_depth": 0.72394, "loss": 25.49699, "grad_norm": 37.21225, "time": 1.51347} -{"mode": "train", "epoch": 1, "iter": 245, "lr": 0.0002, "memory": 49163, "data_time": 0.07951, "loss_cls_0": 0.76809, "loss_box_0": 1.67285, "loss_cns_0": 0.62156, "loss_yns_0": 0.15545, "loss_cls_1": 0.8556, "loss_box_1": 1.56843, "loss_cns_1": 0.65154, "loss_yns_1": 0.1534, "loss_cls_2": 0.87407, "loss_box_2": 1.53057, "loss_cns_2": 0.6545, "loss_yns_2": 0.15296, "loss_cls_3": 0.86663, "loss_box_3": 1.5213, "loss_cns_3": 0.65365, "loss_yns_3": 0.15351, "loss_cls_4": 0.87397, "loss_box_4": 1.52518, "loss_cns_4": 0.6547, "loss_yns_4": 0.15371, "loss_cls_5": 0.87373, "loss_box_5": 1.54577, "loss_cns_5": 0.65486, "loss_yns_5": 0.15474, "loss_cls_dn_0": 0.17093, "loss_box_dn_0": 0.73826, "loss_cls_dn_1": 0.12409, "loss_box_dn_1": 0.74182, "loss_cls_dn_2": 0.1254, "loss_box_dn_2": 0.731, "loss_cls_dn_3": 0.12629, "loss_box_dn_3": 0.73181, "loss_cls_dn_4": 0.12888, "loss_box_dn_4": 0.7378, "loss_cls_dn_5": 0.13303, "loss_box_dn_5": 0.74732, "loss_dense_depth": 0.69456, "loss": 25.22196, "grad_norm": 34.67018, "time": 1.49821} -{"mode": "train", "epoch": 1, "iter": 246, "lr": 0.0002, "memory": 49163, "data_time": 0.07779, "loss_cls_0": 0.76398, "loss_box_0": 1.68531, "loss_cns_0": 0.62201, "loss_yns_0": 0.15496, "loss_cls_1": 0.85898, "loss_box_1": 1.56257, "loss_cns_1": 0.65734, "loss_yns_1": 0.15647, "loss_cls_2": 0.86414, "loss_box_2": 1.54146, "loss_cns_2": 0.65738, "loss_yns_2": 0.15493, "loss_cls_3": 0.85401, "loss_box_3": 1.53037, "loss_cns_3": 0.65754, "loss_yns_3": 0.15423, "loss_cls_4": 0.86529, "loss_box_4": 1.52229, "loss_cns_4": 0.65806, "loss_yns_4": 0.15478, "loss_cls_5": 0.8595, "loss_box_5": 1.54345, "loss_cns_5": 0.65682, "loss_yns_5": 0.15479, "loss_cls_dn_0": 0.16698, "loss_box_dn_0": 0.73677, "loss_cls_dn_1": 0.12136, "loss_box_dn_1": 0.7181, "loss_cls_dn_2": 0.1209, "loss_box_dn_2": 0.70857, "loss_cls_dn_3": 0.12162, "loss_box_dn_3": 0.70451, "loss_cls_dn_4": 0.12449, "loss_box_dn_4": 0.70631, "loss_cls_dn_5": 0.12807, "loss_box_dn_5": 0.7134, "loss_dense_depth": 0.71323, "loss": 25.07498, "grad_norm": 25.5331, "time": 1.4962} -{"mode": "train", "epoch": 1, "iter": 247, "lr": 0.0002, "memory": 49163, "data_time": 0.08334, "loss_cls_0": 0.80426, "loss_box_0": 1.69151, "loss_cns_0": 0.61922, "loss_yns_0": 0.15683, "loss_cls_1": 0.8689, "loss_box_1": 1.54954, "loss_cns_1": 0.6619, "loss_yns_1": 0.1585, "loss_cls_2": 0.88333, "loss_box_2": 1.53447, "loss_cns_2": 0.6623, "loss_yns_2": 0.15708, "loss_cls_3": 0.87749, "loss_box_3": 1.5179, "loss_cns_3": 0.6619, "loss_yns_3": 0.15619, "loss_cls_4": 0.88183, "loss_box_4": 1.52453, "loss_cns_4": 0.66152, "loss_yns_4": 0.15713, "loss_cls_5": 0.89117, "loss_box_5": 1.52856, "loss_cns_5": 0.66204, "loss_yns_5": 0.15853, "loss_cls_dn_0": 0.16693, "loss_box_dn_0": 0.74608, "loss_cls_dn_1": 0.1256, "loss_box_dn_1": 0.72097, "loss_cls_dn_2": 0.12344, "loss_box_dn_2": 0.71408, "loss_cls_dn_3": 0.12428, "loss_box_dn_3": 0.70842, "loss_cls_dn_4": 0.13153, "loss_box_dn_4": 0.71724, "loss_cls_dn_5": 0.13571, "loss_box_dn_5": 0.72309, "loss_dense_depth": 0.70004, "loss": 25.26404, "grad_norm": 39.79951, "time": 1.52251} -{"mode": "train", "epoch": 1, "iter": 248, "lr": 0.0002, "memory": 49163, "data_time": 0.08749, "loss_cls_0": 0.79758, "loss_box_0": 1.66617, "loss_cns_0": 0.61675, "loss_yns_0": 0.1544, "loss_cls_1": 0.87964, "loss_box_1": 1.56583, "loss_cns_1": 0.66149, "loss_yns_1": 0.15665, "loss_cls_2": 0.88596, "loss_box_2": 1.53538, "loss_cns_2": 0.66424, "loss_yns_2": 0.15465, "loss_cls_3": 0.87521, "loss_box_3": 1.5239, "loss_cns_3": 0.66645, "loss_yns_3": 0.15566, "loss_cls_4": 0.88404, "loss_box_4": 1.52477, "loss_cns_4": 0.66364, "loss_yns_4": 0.15534, "loss_cls_5": 0.88062, "loss_box_5": 1.53268, "loss_cns_5": 0.66662, "loss_yns_5": 0.15673, "loss_cls_dn_0": 0.16817, "loss_box_dn_0": 0.73631, "loss_cls_dn_1": 0.12509, "loss_box_dn_1": 0.71623, "loss_cls_dn_2": 0.12283, "loss_box_dn_2": 0.70411, "loss_cls_dn_3": 0.12316, "loss_box_dn_3": 0.70545, "loss_cls_dn_4": 0.12946, "loss_box_dn_4": 0.71879, "loss_cls_dn_5": 0.1296, "loss_box_dn_5": 0.73093, "loss_dense_depth": 0.70271, "loss": 25.23725, "grad_norm": 29.3831, "time": 1.5085} -{"mode": "train", "epoch": 1, "iter": 249, "lr": 0.0002, "memory": 49163, "data_time": 0.17932, "loss_cls_0": 0.79534, "loss_box_0": 1.68339, "loss_cns_0": 0.62572, "loss_yns_0": 0.15707, "loss_cls_1": 0.89391, "loss_box_1": 1.59368, "loss_cns_1": 0.66322, "loss_yns_1": 0.15699, "loss_cls_2": 0.90016, "loss_box_2": 1.58183, "loss_cns_2": 0.66086, "loss_yns_2": 0.15647, "loss_cls_3": 0.89762, "loss_box_3": 1.57884, "loss_cns_3": 0.66064, "loss_yns_3": 0.1559, "loss_cls_4": 0.89111, "loss_box_4": 1.57897, "loss_cns_4": 0.66135, "loss_yns_4": 0.15593, "loss_cls_5": 0.90663, "loss_box_5": 1.58814, "loss_cns_5": 0.66211, "loss_yns_5": 0.15642, "loss_cls_dn_0": 0.1709, "loss_box_dn_0": 0.7398, "loss_cls_dn_1": 0.11843, "loss_box_dn_1": 0.71997, "loss_cls_dn_2": 0.11702, "loss_box_dn_2": 0.70618, "loss_cls_dn_3": 0.11805, "loss_box_dn_3": 0.71443, "loss_cls_dn_4": 0.12123, "loss_box_dn_4": 0.7273, "loss_cls_dn_5": 0.12396, "loss_box_dn_5": 0.74371, "loss_dense_depth": 0.70891, "loss": 25.59221, "grad_norm": 39.90646, "time": 1.59798} -{"mode": "train", "epoch": 1, "iter": 250, "lr": 0.0002, "memory": 49163, "data_time": 0.08101, "loss_cls_0": 0.75554, "loss_box_0": 1.67814, "loss_cns_0": 0.62489, "loss_yns_0": 0.15608, "loss_cls_1": 0.86598, "loss_box_1": 1.57728, "loss_cns_1": 0.66379, "loss_yns_1": 0.15982, "loss_cls_2": 0.86283, "loss_box_2": 1.58085, "loss_cns_2": 0.66197, "loss_yns_2": 0.15656, "loss_cls_3": 0.85153, "loss_box_3": 1.57348, "loss_cns_3": 0.66183, "loss_yns_3": 0.1567, "loss_cls_4": 0.85856, "loss_box_4": 1.5714, "loss_cns_4": 0.65755, "loss_yns_4": 0.15579, "loss_cls_5": 0.85779, "loss_box_5": 1.59075, "loss_cns_5": 0.65874, "loss_yns_5": 0.15665, "loss_cls_dn_0": 0.16518, "loss_box_dn_0": 0.73286, "loss_cls_dn_1": 0.11648, "loss_box_dn_1": 0.72402, "loss_cls_dn_2": 0.11806, "loss_box_dn_2": 0.72059, "loss_cls_dn_3": 0.11868, "loss_box_dn_3": 0.73734, "loss_cls_dn_4": 0.12375, "loss_box_dn_4": 0.75614, "loss_cls_dn_5": 0.12592, "loss_box_dn_5": 0.77815, "loss_dense_depth": 0.70155, "loss": 25.41323, "grad_norm": 44.01296, "time": 1.49779} -{"mode": "train", "epoch": 1, "iter": 251, "lr": 0.0002, "memory": 49163, "data_time": 0.07868, "loss_cls_0": 0.77798, "loss_box_0": 1.66722, "loss_cns_0": 0.62308, "loss_yns_0": 0.15184, "loss_cls_1": 0.87783, "loss_box_1": 1.59139, "loss_cns_1": 0.65796, "loss_yns_1": 0.15495, "loss_cls_2": 0.89559, "loss_box_2": 1.56398, "loss_cns_2": 0.65782, "loss_yns_2": 0.15366, "loss_cls_3": 0.88683, "loss_box_3": 1.5433, "loss_cns_3": 0.66374, "loss_yns_3": 0.15184, "loss_cls_4": 0.88431, "loss_box_4": 1.53682, "loss_cns_4": 0.65456, "loss_yns_4": 0.1502, "loss_cls_5": 0.8613, "loss_box_5": 1.57327, "loss_cns_5": 0.65681, "loss_yns_5": 0.15261, "loss_cls_dn_0": 0.17242, "loss_box_dn_0": 0.74181, "loss_cls_dn_1": 0.12134, "loss_box_dn_1": 0.74797, "loss_cls_dn_2": 0.1239, "loss_box_dn_2": 0.75076, "loss_cls_dn_3": 0.12456, "loss_box_dn_3": 0.76573, "loss_cls_dn_4": 0.12406, "loss_box_dn_4": 0.78239, "loss_cls_dn_5": 0.12708, "loss_box_dn_5": 0.80178, "loss_dense_depth": 0.7331, "loss": 25.60575, "grad_norm": 47.31849, "time": 1.56993} -{"mode": "train", "epoch": 1, "iter": 252, "lr": 0.0002, "memory": 49163, "data_time": 0.07834, "loss_cls_0": 0.76874, "loss_box_0": 1.67655, "loss_cns_0": 0.62011, "loss_yns_0": 0.15247, "loss_cls_1": 0.86147, "loss_box_1": 1.6193, "loss_cns_1": 0.65181, "loss_yns_1": 0.155, "loss_cls_2": 0.87305, "loss_box_2": 1.5743, "loss_cns_2": 0.65382, "loss_yns_2": 0.15105, "loss_cls_3": 0.86315, "loss_box_3": 1.5603, "loss_cns_3": 0.66396, "loss_yns_3": 0.15189, "loss_cls_4": 0.84628, "loss_box_4": 1.56088, "loss_cns_4": 0.65494, "loss_yns_4": 0.15203, "loss_cls_5": 0.84808, "loss_box_5": 1.58215, "loss_cns_5": 0.65794, "loss_yns_5": 0.15432, "loss_cls_dn_0": 0.16866, "loss_box_dn_0": 0.73214, "loss_cls_dn_1": 0.1164, "loss_box_dn_1": 0.71412, "loss_cls_dn_2": 0.11706, "loss_box_dn_2": 0.70673, "loss_cls_dn_3": 0.11654, "loss_box_dn_3": 0.7121, "loss_cls_dn_4": 0.11556, "loss_box_dn_4": 0.72325, "loss_cls_dn_5": 0.11956, "loss_box_dn_5": 0.73323, "loss_dense_depth": 0.70388, "loss": 25.23281, "grad_norm": 31.45489, "time": 1.49783} -{"mode": "train", "epoch": 1, "iter": 253, "lr": 0.0002, "memory": 49163, "data_time": 0.07492, "loss_cls_0": 0.77665, "loss_box_0": 1.68097, "loss_cns_0": 0.62718, "loss_yns_0": 0.14964, "loss_cls_1": 0.86568, "loss_box_1": 1.57261, "loss_cns_1": 0.65751, "loss_yns_1": 0.15012, "loss_cls_2": 0.86626, "loss_box_2": 1.5657, "loss_cns_2": 0.65955, "loss_yns_2": 0.15023, "loss_cls_3": 0.86651, "loss_box_3": 1.5421, "loss_cns_3": 0.66079, "loss_yns_3": 0.14941, "loss_cls_4": 0.87467, "loss_box_4": 1.54795, "loss_cns_4": 0.65959, "loss_yns_4": 0.15022, "loss_cls_5": 0.86021, "loss_box_5": 1.54236, "loss_cns_5": 0.65947, "loss_yns_5": 0.14891, "loss_cls_dn_0": 0.17077, "loss_box_dn_0": 0.73406, "loss_cls_dn_1": 0.11993, "loss_box_dn_1": 0.71381, "loss_cls_dn_2": 0.11723, "loss_box_dn_2": 0.70388, "loss_cls_dn_3": 0.11857, "loss_box_dn_3": 0.70156, "loss_cls_dn_4": 0.12183, "loss_box_dn_4": 0.70793, "loss_cls_dn_5": 0.12343, "loss_box_dn_5": 0.71088, "loss_dense_depth": 0.74263, "loss": 25.1708, "grad_norm": 38.53019, "time": 1.50126} -{"mode": "train", "epoch": 1, "iter": 254, "lr": 0.0002, "memory": 49163, "data_time": 0.07406, "loss_cls_0": 0.75458, "loss_box_0": 1.69791, "loss_cns_0": 0.62274, "loss_yns_0": 0.15095, "loss_cls_1": 0.86002, "loss_box_1": 1.58731, "loss_cns_1": 0.65532, "loss_yns_1": 0.1501, "loss_cls_2": 0.86434, "loss_box_2": 1.57557, "loss_cns_2": 0.65872, "loss_yns_2": 0.14904, "loss_cls_3": 0.86102, "loss_box_3": 1.54973, "loss_cns_3": 0.66118, "loss_yns_3": 0.14907, "loss_cls_4": 0.85299, "loss_box_4": 1.54814, "loss_cns_4": 0.65924, "loss_yns_4": 0.15047, "loss_cls_5": 0.85627, "loss_box_5": 1.53505, "loss_cns_5": 0.65623, "loss_yns_5": 0.14893, "loss_cls_dn_0": 0.16461, "loss_box_dn_0": 0.73911, "loss_cls_dn_1": 0.11826, "loss_box_dn_1": 0.70558, "loss_cls_dn_2": 0.11702, "loss_box_dn_2": 0.69325, "loss_cls_dn_3": 0.11895, "loss_box_dn_3": 0.68606, "loss_cls_dn_4": 0.12049, "loss_box_dn_4": 0.68601, "loss_cls_dn_5": 0.12683, "loss_box_dn_5": 0.68947, "loss_dense_depth": 0.72489, "loss": 25.04548, "grad_norm": 34.55534, "time": 1.54269} -{"mode": "train", "epoch": 1, "iter": 255, "lr": 0.0002, "memory": 49163, "data_time": 0.07459, "loss_cls_0": 0.75706, "loss_box_0": 1.68869, "loss_cns_0": 0.62448, "loss_yns_0": 0.15083, "loss_cls_1": 0.84493, "loss_box_1": 1.52942, "loss_cns_1": 0.66039, "loss_yns_1": 0.15014, "loss_cls_2": 0.85027, "loss_box_2": 1.516, "loss_cns_2": 0.6621, "loss_yns_2": 0.15135, "loss_cls_3": 0.8412, "loss_box_3": 1.50847, "loss_cns_3": 0.6686, "loss_yns_3": 0.15128, "loss_cls_4": 0.8464, "loss_box_4": 1.50433, "loss_cns_4": 0.66217, "loss_yns_4": 0.15116, "loss_cls_5": 0.87919, "loss_box_5": 1.4956, "loss_cns_5": 0.6583, "loss_yns_5": 0.1508, "loss_cls_dn_0": 0.16212, "loss_box_dn_0": 0.74062, "loss_cls_dn_1": 0.11639, "loss_box_dn_1": 0.70483, "loss_cls_dn_2": 0.11689, "loss_box_dn_2": 0.68761, "loss_cls_dn_3": 0.11619, "loss_box_dn_3": 0.68423, "loss_cls_dn_4": 0.11943, "loss_box_dn_4": 0.68446, "loss_cls_dn_5": 0.12287, "loss_box_dn_5": 0.69238, "loss_dense_depth": 0.7229, "loss": 24.7741, "grad_norm": 32.74247, "time": 1.48424} -{"mode": "train", "epoch": 1, "iter": 256, "lr": 0.0002, "memory": 49163, "data_time": 0.07908, "loss_cls_0": 0.76981, "loss_box_0": 1.69661, "loss_cns_0": 0.62521, "loss_yns_0": 0.15213, "loss_cls_1": 0.84699, "loss_box_1": 1.55792, "loss_cns_1": 0.6566, "loss_yns_1": 0.15081, "loss_cls_2": 0.86591, "loss_box_2": 1.50943, "loss_cns_2": 0.66152, "loss_yns_2": 0.1515, "loss_cls_3": 0.87491, "loss_box_3": 1.50769, "loss_cns_3": 0.66891, "loss_yns_3": 0.1518, "loss_cls_4": 0.87455, "loss_box_4": 1.50247, "loss_cns_4": 0.66264, "loss_yns_4": 0.15234, "loss_cls_5": 0.87133, "loss_box_5": 1.50429, "loss_cns_5": 0.66078, "loss_yns_5": 0.1525, "loss_cls_dn_0": 0.16059, "loss_box_dn_0": 0.73149, "loss_cls_dn_1": 0.11103, "loss_box_dn_1": 0.67246, "loss_cls_dn_2": 0.11293, "loss_box_dn_2": 0.65804, "loss_cls_dn_3": 0.11084, "loss_box_dn_3": 0.66528, "loss_cls_dn_4": 0.11446, "loss_box_dn_4": 0.67545, "loss_cls_dn_5": 0.1161, "loss_box_dn_5": 0.68801, "loss_dense_depth": 0.73899, "loss": 24.7843, "grad_norm": 33.93274, "time": 1.48722} -{"mode": "train", "epoch": 1, "iter": 257, "lr": 0.0002, "memory": 49163, "data_time": 0.0741, "loss_cls_0": 0.78584, "loss_box_0": 1.72975, "loss_cns_0": 0.62276, "loss_yns_0": 0.15181, "loss_cls_1": 0.85724, "loss_box_1": 1.6518, "loss_cns_1": 0.65207, "loss_yns_1": 0.15003, "loss_cls_2": 0.8755, "loss_box_2": 1.59004, "loss_cns_2": 0.6561, "loss_yns_2": 0.15168, "loss_cls_3": 0.87709, "loss_box_3": 1.57276, "loss_cns_3": 0.66066, "loss_yns_3": 0.15034, "loss_cls_4": 0.8742, "loss_box_4": 1.57829, "loss_cns_4": 0.65829, "loss_yns_4": 0.15216, "loss_cls_5": 0.88755, "loss_box_5": 1.59258, "loss_cns_5": 0.6578, "loss_yns_5": 0.15113, "loss_cls_dn_0": 0.16617, "loss_box_dn_0": 0.73411, "loss_cls_dn_1": 0.11296, "loss_box_dn_1": 0.6859, "loss_cls_dn_2": 0.11511, "loss_box_dn_2": 0.67531, "loss_cls_dn_3": 0.11368, "loss_box_dn_3": 0.68503, "loss_cls_dn_4": 0.11498, "loss_box_dn_4": 0.70119, "loss_cls_dn_5": 0.11724, "loss_box_dn_5": 0.71576, "loss_dense_depth": 0.73631, "loss": 25.3612, "grad_norm": 37.38593, "time": 1.52794} -{"mode": "train", "epoch": 1, "iter": 258, "lr": 0.0002, "memory": 49163, "data_time": 0.0791, "loss_cls_0": 0.7743, "loss_box_0": 1.70956, "loss_cns_0": 0.62227, "loss_yns_0": 0.14955, "loss_cls_1": 0.8545, "loss_box_1": 1.62871, "loss_cns_1": 0.65587, "loss_yns_1": 0.14961, "loss_cls_2": 0.86606, "loss_box_2": 1.58174, "loss_cns_2": 0.65843, "loss_yns_2": 0.14992, "loss_cls_3": 0.86697, "loss_box_3": 1.56749, "loss_cns_3": 0.65985, "loss_yns_3": 0.14956, "loss_cls_4": 0.86965, "loss_box_4": 1.57038, "loss_cns_4": 0.66074, "loss_yns_4": 0.14947, "loss_cls_5": 0.8883, "loss_box_5": 1.5823, "loss_cns_5": 0.6591, "loss_yns_5": 0.15002, "loss_cls_dn_0": 0.1622, "loss_box_dn_0": 0.72732, "loss_cls_dn_1": 0.11619, "loss_box_dn_1": 0.68938, "loss_cls_dn_2": 0.11555, "loss_box_dn_2": 0.68297, "loss_cls_dn_3": 0.1157, "loss_box_dn_3": 0.68882, "loss_cls_dn_4": 0.11961, "loss_box_dn_4": 0.70532, "loss_cls_dn_5": 0.11909, "loss_box_dn_5": 0.71875, "loss_dense_depth": 0.72932, "loss": 25.26459, "grad_norm": 32.43076, "time": 1.49442} -{"mode": "train", "epoch": 1, "iter": 259, "lr": 0.0002, "memory": 49163, "data_time": 0.07747, "loss_cls_0": 0.80865, "loss_box_0": 1.72214, "loss_cns_0": 0.61863, "loss_yns_0": 0.14941, "loss_cls_1": 0.87958, "loss_box_1": 1.66812, "loss_cns_1": 0.65139, "loss_yns_1": 0.14918, "loss_cls_2": 0.90633, "loss_box_2": 1.64286, "loss_cns_2": 0.65397, "loss_yns_2": 0.1477, "loss_cls_3": 0.91144, "loss_box_3": 1.61803, "loss_cns_3": 0.65232, "loss_yns_3": 0.14667, "loss_cls_4": 0.91405, "loss_box_4": 1.62043, "loss_cns_4": 0.65462, "loss_yns_4": 0.14835, "loss_cls_5": 0.90904, "loss_box_5": 1.62673, "loss_cns_5": 0.65686, "loss_yns_5": 0.14984, "loss_cls_dn_0": 0.16478, "loss_box_dn_0": 0.7402, "loss_cls_dn_1": 0.11971, "loss_box_dn_1": 0.70821, "loss_cls_dn_2": 0.1195, "loss_box_dn_2": 0.69739, "loss_cls_dn_3": 0.12185, "loss_box_dn_3": 0.69879, "loss_cls_dn_4": 0.12812, "loss_box_dn_4": 0.71076, "loss_cls_dn_5": 0.12602, "loss_box_dn_5": 0.71914, "loss_dense_depth": 0.77975, "loss": 25.84056, "grad_norm": 37.33321, "time": 1.49927} -{"mode": "train", "epoch": 1, "iter": 260, "lr": 0.0002, "memory": 49163, "data_time": 0.07152, "loss_cls_0": 0.79261, "loss_box_0": 1.708, "loss_cns_0": 0.62187, "loss_yns_0": 0.15006, "loss_cls_1": 0.87245, "loss_box_1": 1.65994, "loss_cns_1": 0.65313, "loss_yns_1": 0.15233, "loss_cls_2": 0.90482, "loss_box_2": 1.61933, "loss_cns_2": 0.65618, "loss_yns_2": 0.15051, "loss_cls_3": 0.90515, "loss_box_3": 1.59442, "loss_cns_3": 0.65557, "loss_yns_3": 0.14953, "loss_cls_4": 0.90979, "loss_box_4": 1.58985, "loss_cns_4": 0.65638, "loss_yns_4": 0.15001, "loss_cls_5": 0.90533, "loss_box_5": 1.5947, "loss_cns_5": 0.66094, "loss_yns_5": 0.1495, "loss_cls_dn_0": 0.16367, "loss_box_dn_0": 0.73545, "loss_cls_dn_1": 0.12339, "loss_box_dn_1": 0.72039, "loss_cls_dn_2": 0.12214, "loss_box_dn_2": 0.7027, "loss_cls_dn_3": 0.12205, "loss_box_dn_3": 0.69829, "loss_cls_dn_4": 0.12546, "loss_box_dn_4": 0.70396, "loss_cls_dn_5": 0.1252, "loss_box_dn_5": 0.70977, "loss_dense_depth": 0.7569, "loss": 25.67177, "grad_norm": 30.29574, "time": 6.85666} -{"mode": "train", "epoch": 1, "iter": 261, "lr": 0.0002, "memory": 49163, "data_time": 0.07233, "loss_cls_0": 0.79709, "loss_box_0": 1.73102, "loss_cns_0": 0.61869, "loss_yns_0": 0.14855, "loss_cls_1": 0.88478, "loss_box_1": 1.62034, "loss_cns_1": 0.656, "loss_yns_1": 0.14855, "loss_cls_2": 0.90207, "loss_box_2": 1.60613, "loss_cns_2": 0.65867, "loss_yns_2": 0.14935, "loss_cls_3": 0.90237, "loss_box_3": 1.57292, "loss_cns_3": 0.65925, "loss_yns_3": 0.14775, "loss_cls_4": 0.9055, "loss_box_4": 1.57367, "loss_cns_4": 0.65962, "loss_yns_4": 0.14851, "loss_cls_5": 0.90668, "loss_box_5": 1.57812, "loss_cns_5": 0.66057, "loss_yns_5": 0.14863, "loss_cls_dn_0": 0.1666, "loss_box_dn_0": 0.74292, "loss_cls_dn_1": 0.11993, "loss_box_dn_1": 0.69984, "loss_cls_dn_2": 0.11886, "loss_box_dn_2": 0.6852, "loss_cls_dn_3": 0.11955, "loss_box_dn_3": 0.68042, "loss_cls_dn_4": 0.12192, "loss_box_dn_4": 0.67879, "loss_cls_dn_5": 0.12405, "loss_box_dn_5": 0.68254, "loss_dense_depth": 0.77963, "loss": 25.50509, "grad_norm": 31.419, "time": 1.5} -{"mode": "train", "epoch": 1, "iter": 262, "lr": 0.0002, "memory": 49163, "data_time": 0.07438, "loss_cls_0": 0.7906, "loss_box_0": 1.72221, "loss_cns_0": 0.62073, "loss_yns_0": 0.14558, "loss_cls_1": 0.88151, "loss_box_1": 1.6393, "loss_cns_1": 0.6554, "loss_yns_1": 0.14618, "loss_cls_2": 0.90755, "loss_box_2": 1.62579, "loss_cns_2": 0.65821, "loss_yns_2": 0.14692, "loss_cls_3": 0.90458, "loss_box_3": 1.6034, "loss_cns_3": 0.65976, "loss_yns_3": 0.1455, "loss_cls_4": 0.91404, "loss_box_4": 1.59853, "loss_cns_4": 0.65773, "loss_yns_4": 0.14606, "loss_cls_5": 0.90625, "loss_box_5": 1.59537, "loss_cns_5": 0.65704, "loss_yns_5": 0.14567, "loss_cls_dn_0": 0.16032, "loss_box_dn_0": 0.74977, "loss_cls_dn_1": 0.11849, "loss_box_dn_1": 0.69322, "loss_cls_dn_2": 0.11882, "loss_box_dn_2": 0.68011, "loss_cls_dn_3": 0.11919, "loss_box_dn_3": 0.67581, "loss_cls_dn_4": 0.12198, "loss_box_dn_4": 0.67326, "loss_cls_dn_5": 0.12502, "loss_box_dn_5": 0.67649, "loss_dense_depth": 0.74327, "loss": 25.52964, "grad_norm": 37.19118, "time": 1.56759} -{"mode": "train", "epoch": 1, "iter": 263, "lr": 0.0002, "memory": 49163, "data_time": 0.07666, "loss_cls_0": 0.78002, "loss_box_0": 1.71461, "loss_cns_0": 0.61316, "loss_yns_0": 0.14488, "loss_cls_1": 0.86539, "loss_box_1": 1.63836, "loss_cns_1": 0.65013, "loss_yns_1": 0.14508, "loss_cls_2": 0.88043, "loss_box_2": 1.61352, "loss_cns_2": 0.65538, "loss_yns_2": 0.14509, "loss_cls_3": 0.8848, "loss_box_3": 1.58461, "loss_cns_3": 0.65575, "loss_yns_3": 0.14391, "loss_cls_4": 0.88913, "loss_box_4": 1.58322, "loss_cns_4": 0.65475, "loss_yns_4": 0.14336, "loss_cls_5": 0.89274, "loss_box_5": 1.58082, "loss_cns_5": 0.6548, "loss_yns_5": 0.14331, "loss_cls_dn_0": 0.16149, "loss_box_dn_0": 0.75869, "loss_cls_dn_1": 0.11901, "loss_box_dn_1": 0.69427, "loss_cls_dn_2": 0.11734, "loss_box_dn_2": 0.67739, "loss_cls_dn_3": 0.11709, "loss_box_dn_3": 0.67368, "loss_cls_dn_4": 0.11922, "loss_box_dn_4": 0.67464, "loss_cls_dn_5": 0.11994, "loss_box_dn_5": 0.68352, "loss_dense_depth": 0.75392, "loss": 25.32744, "grad_norm": 27.47056, "time": 1.49528} -{"mode": "train", "epoch": 1, "iter": 264, "lr": 0.00021, "memory": 49163, "data_time": 0.08, "loss_cls_0": 0.77397, "loss_box_0": 1.70992, "loss_cns_0": 0.61713, "loss_yns_0": 0.14914, "loss_cls_1": 0.86093, "loss_box_1": 1.63346, "loss_cns_1": 0.65214, "loss_yns_1": 0.1467, "loss_cls_2": 0.87398, "loss_box_2": 1.60207, "loss_cns_2": 0.65565, "loss_yns_2": 0.14578, "loss_cls_3": 0.87677, "loss_box_3": 1.58693, "loss_cns_3": 0.65738, "loss_yns_3": 0.14462, "loss_cls_4": 0.88295, "loss_box_4": 1.5914, "loss_cns_4": 0.6568, "loss_yns_4": 0.1446, "loss_cls_5": 0.89281, "loss_box_5": 1.59116, "loss_cns_5": 0.65698, "loss_yns_5": 0.14468, "loss_cls_dn_0": 0.16087, "loss_box_dn_0": 0.74248, "loss_cls_dn_1": 0.12079, "loss_box_dn_1": 0.69688, "loss_cls_dn_2": 0.11766, "loss_box_dn_2": 0.68224, "loss_cls_dn_3": 0.11969, "loss_box_dn_3": 0.69026, "loss_cls_dn_4": 0.12451, "loss_box_dn_4": 0.70086, "loss_cls_dn_5": 0.12408, "loss_box_dn_5": 0.71546, "loss_dense_depth": 0.76658, "loss": 25.41034, "grad_norm": 35.34405, "time": 1.51517} -{"mode": "train", "epoch": 1, "iter": 265, "lr": 0.00021, "memory": 49163, "data_time": 0.07716, "loss_cls_0": 0.79905, "loss_box_0": 1.74542, "loss_cns_0": 0.61919, "loss_yns_0": 0.14754, "loss_cls_1": 0.86495, "loss_box_1": 1.63908, "loss_cns_1": 0.65339, "loss_yns_1": 0.14604, "loss_cls_2": 0.88292, "loss_box_2": 1.60876, "loss_cns_2": 0.65853, "loss_yns_2": 0.14633, "loss_cls_3": 0.88709, "loss_box_3": 1.61146, "loss_cns_3": 0.66714, "loss_yns_3": 0.14777, "loss_cls_4": 0.88647, "loss_box_4": 1.60217, "loss_cns_4": 0.66041, "loss_yns_4": 0.14929, "loss_cls_5": 0.89098, "loss_box_5": 1.60577, "loss_cns_5": 0.65932, "loss_yns_5": 0.14844, "loss_cls_dn_0": 0.16612, "loss_box_dn_0": 0.74912, "loss_cls_dn_1": 0.11738, "loss_box_dn_1": 0.70966, "loss_cls_dn_2": 0.11613, "loss_box_dn_2": 0.70117, "loss_cls_dn_3": 0.12407, "loss_box_dn_3": 0.71365, "loss_cls_dn_4": 0.12784, "loss_box_dn_4": 0.72007, "loss_cls_dn_5": 0.12782, "loss_box_dn_5": 0.73158, "loss_dense_depth": 0.76475, "loss": 25.69687, "grad_norm": 42.11339, "time": 1.49129} -{"mode": "train", "epoch": 1, "iter": 266, "lr": 0.00021, "memory": 49163, "data_time": 0.07816, "loss_cls_0": 0.7675, "loss_box_0": 1.71328, "loss_cns_0": 0.62105, "loss_yns_0": 0.14389, "loss_cls_1": 0.84318, "loss_box_1": 1.60688, "loss_cns_1": 0.65534, "loss_yns_1": 0.14488, "loss_cls_2": 0.85235, "loss_box_2": 1.58959, "loss_cns_2": 0.66, "loss_yns_2": 0.14463, "loss_cls_3": 0.85899, "loss_box_3": 1.58683, "loss_cns_3": 0.66814, "loss_yns_3": 0.14645, "loss_cls_4": 0.85517, "loss_box_4": 1.5753, "loss_cns_4": 0.66107, "loss_yns_4": 0.14765, "loss_cls_5": 0.85964, "loss_box_5": 1.59206, "loss_cns_5": 0.66063, "loss_yns_5": 0.14689, "loss_cls_dn_0": 0.15903, "loss_box_dn_0": 0.75529, "loss_cls_dn_1": 0.11712, "loss_box_dn_1": 0.72453, "loss_cls_dn_2": 0.11617, "loss_box_dn_2": 0.71719, "loss_cls_dn_3": 0.12295, "loss_box_dn_3": 0.72578, "loss_cls_dn_4": 0.12408, "loss_box_dn_4": 0.72816, "loss_cls_dn_5": 0.12513, "loss_box_dn_5": 0.74138, "loss_dense_depth": 0.78256, "loss": 25.44077, "grad_norm": 35.47916, "time": 1.50007} -{"mode": "train", "epoch": 1, "iter": 267, "lr": 0.00021, "memory": 49163, "data_time": 0.08263, "loss_cls_0": 0.75939, "loss_box_0": 1.68418, "loss_cns_0": 0.6233, "loss_yns_0": 0.14385, "loss_cls_1": 0.83712, "loss_box_1": 1.58041, "loss_cns_1": 0.65714, "loss_yns_1": 0.14307, "loss_cls_2": 0.85117, "loss_box_2": 1.55884, "loss_cns_2": 0.65869, "loss_yns_2": 0.14263, "loss_cls_3": 0.84888, "loss_box_3": 1.5525, "loss_cns_3": 0.66235, "loss_yns_3": 0.14413, "loss_cls_4": 0.84678, "loss_box_4": 1.55624, "loss_cns_4": 0.65959, "loss_yns_4": 0.14346, "loss_cls_5": 0.8476, "loss_box_5": 1.57471, "loss_cns_5": 0.66065, "loss_yns_5": 0.14515, "loss_cls_dn_0": 0.15651, "loss_box_dn_0": 0.74661, "loss_cls_dn_1": 0.11392, "loss_box_dn_1": 0.72445, "loss_cls_dn_2": 0.11259, "loss_box_dn_2": 0.71141, "loss_cls_dn_3": 0.11579, "loss_box_dn_3": 0.71323, "loss_cls_dn_4": 0.11745, "loss_box_dn_4": 0.71557, "loss_cls_dn_5": 0.11806, "loss_box_dn_5": 0.72688, "loss_dense_depth": 0.74761, "loss": 25.10189, "grad_norm": 36.06237, "time": 1.50071} -{"mode": "train", "epoch": 1, "iter": 268, "lr": 0.00021, "memory": 49163, "data_time": 0.08469, "loss_cls_0": 0.76008, "loss_box_0": 1.67901, "loss_cns_0": 0.62982, "loss_yns_0": 0.14764, "loss_cls_1": 0.84298, "loss_box_1": 1.57416, "loss_cns_1": 0.66452, "loss_yns_1": 0.14458, "loss_cls_2": 0.85446, "loss_box_2": 1.54865, "loss_cns_2": 0.66508, "loss_yns_2": 0.14332, "loss_cls_3": 0.84927, "loss_box_3": 1.52561, "loss_cns_3": 0.66404, "loss_yns_3": 0.14374, "loss_cls_4": 0.84834, "loss_box_4": 1.5261, "loss_cns_4": 0.66377, "loss_yns_4": 0.14298, "loss_cls_5": 0.84676, "loss_box_5": 1.53562, "loss_cns_5": 0.66544, "loss_yns_5": 0.14457, "loss_cls_dn_0": 0.15905, "loss_box_dn_0": 0.74171, "loss_cls_dn_1": 0.11842, "loss_box_dn_1": 0.72546, "loss_cls_dn_2": 0.11629, "loss_box_dn_2": 0.71334, "loss_cls_dn_3": 0.11792, "loss_box_dn_3": 0.70765, "loss_cls_dn_4": 0.11877, "loss_box_dn_4": 0.70763, "loss_cls_dn_5": 0.12127, "loss_box_dn_5": 0.71174, "loss_dense_depth": 0.75628, "loss": 25.02608, "grad_norm": 29.32419, "time": 1.49446} -{"mode": "train", "epoch": 1, "iter": 269, "lr": 0.00021, "memory": 49163, "data_time": 0.16549, "loss_cls_0": 0.75517, "loss_box_0": 1.68594, "loss_cns_0": 0.62931, "loss_yns_0": 0.14903, "loss_cls_1": 0.83687, "loss_box_1": 1.58278, "loss_cns_1": 0.66118, "loss_yns_1": 0.14739, "loss_cls_2": 0.84456, "loss_box_2": 1.57099, "loss_cns_2": 0.66138, "loss_yns_2": 0.14736, "loss_cls_3": 0.85302, "loss_box_3": 1.54117, "loss_cns_3": 0.66326, "loss_yns_3": 0.1462, "loss_cls_4": 0.85836, "loss_box_4": 1.5368, "loss_cns_4": 0.66153, "loss_yns_4": 0.14463, "loss_cls_5": 0.85581, "loss_box_5": 1.53258, "loss_cns_5": 0.6609, "loss_yns_5": 0.14662, "loss_cls_dn_0": 0.15397, "loss_box_dn_0": 0.73745, "loss_cls_dn_1": 0.11474, "loss_box_dn_1": 0.71802, "loss_cls_dn_2": 0.11226, "loss_box_dn_2": 0.70925, "loss_cls_dn_3": 0.11566, "loss_box_dn_3": 0.6973, "loss_cls_dn_4": 0.11725, "loss_box_dn_4": 0.69538, "loss_cls_dn_5": 0.12157, "loss_box_dn_5": 0.69405, "loss_dense_depth": 0.75939, "loss": 25.01911, "grad_norm": 36.11523, "time": 1.57204} -{"mode": "train", "epoch": 1, "iter": 270, "lr": 0.00021, "memory": 49163, "data_time": 0.07566, "loss_cls_0": 0.79788, "loss_box_0": 1.71567, "loss_cns_0": 0.62065, "loss_yns_0": 0.14984, "loss_cls_1": 0.87188, "loss_box_1": 1.61219, "loss_cns_1": 0.65495, "loss_yns_1": 0.14801, "loss_cls_2": 0.89101, "loss_box_2": 1.59638, "loss_cns_2": 0.65694, "loss_yns_2": 0.14676, "loss_cls_3": 0.88924, "loss_box_3": 1.57079, "loss_cns_3": 0.65816, "loss_yns_3": 0.1473, "loss_cls_4": 0.88376, "loss_box_4": 1.56441, "loss_cns_4": 0.654, "loss_yns_4": 0.14678, "loss_cls_5": 0.88663, "loss_box_5": 1.5678, "loss_cns_5": 0.65352, "loss_yns_5": 0.14782, "loss_cls_dn_0": 0.15986, "loss_box_dn_0": 0.73904, "loss_cls_dn_1": 0.11363, "loss_box_dn_1": 0.68698, "loss_cls_dn_2": 0.11394, "loss_box_dn_2": 0.67817, "loss_cls_dn_3": 0.11744, "loss_box_dn_3": 0.67346, "loss_cls_dn_4": 0.12465, "loss_box_dn_4": 0.67565, "loss_cls_dn_5": 0.1328, "loss_box_dn_5": 0.68027, "loss_dense_depth": 0.77861, "loss": 25.30691, "grad_norm": 36.31151, "time": 1.49544} -{"mode": "train", "epoch": 1, "iter": 271, "lr": 0.00021, "memory": 49163, "data_time": 0.07637, "loss_cls_0": 0.7404, "loss_box_0": 1.67374, "loss_cns_0": 0.62824, "loss_yns_0": 0.15186, "loss_cls_1": 0.85543, "loss_box_1": 1.57475, "loss_cns_1": 0.65605, "loss_yns_1": 0.14722, "loss_cls_2": 0.85578, "loss_box_2": 1.54835, "loss_cns_2": 0.65789, "loss_yns_2": 0.14529, "loss_cls_3": 0.85206, "loss_box_3": 1.54946, "loss_cns_3": 0.65605, "loss_yns_3": 0.14605, "loss_cls_4": 0.84728, "loss_box_4": 1.53853, "loss_cns_4": 0.65626, "loss_yns_4": 0.14605, "loss_cls_5": 0.85307, "loss_box_5": 1.53914, "loss_cns_5": 0.65553, "loss_yns_5": 0.14824, "loss_cls_dn_0": 0.14883, "loss_box_dn_0": 0.73352, "loss_cls_dn_1": 0.11339, "loss_box_dn_1": 0.68583, "loss_cls_dn_2": 0.11327, "loss_box_dn_2": 0.67241, "loss_cls_dn_3": 0.11454, "loss_box_dn_3": 0.68285, "loss_cls_dn_4": 0.11981, "loss_box_dn_4": 0.68535, "loss_cls_dn_5": 0.12313, "loss_box_dn_5": 0.69453, "loss_dense_depth": 0.72399, "loss": 24.83417, "grad_norm": 35.57342, "time": 1.49948} -{"mode": "train", "epoch": 1, "iter": 272, "lr": 0.00021, "memory": 49163, "data_time": 0.07631, "loss_cls_0": 0.74866, "loss_box_0": 1.68887, "loss_cns_0": 0.62486, "loss_yns_0": 0.15396, "loss_cls_1": 0.84757, "loss_box_1": 1.56246, "loss_cns_1": 0.65452, "loss_yns_1": 0.14957, "loss_cls_2": 0.85175, "loss_box_2": 1.52241, "loss_cns_2": 0.65796, "loss_yns_2": 0.14749, "loss_cls_3": 0.85293, "loss_box_3": 1.55317, "loss_cns_3": 0.65617, "loss_yns_3": 0.1495, "loss_cls_4": 0.84527, "loss_box_4": 1.54895, "loss_cns_4": 0.65819, "loss_yns_4": 0.14852, "loss_cls_5": 0.85177, "loss_box_5": 1.5451, "loss_cns_5": 0.65423, "loss_yns_5": 0.14974, "loss_cls_dn_0": 0.14719, "loss_box_dn_0": 0.73417, "loss_cls_dn_1": 0.11042, "loss_box_dn_1": 0.70542, "loss_cls_dn_2": 0.11065, "loss_box_dn_2": 0.69561, "loss_cls_dn_3": 0.11081, "loss_box_dn_3": 0.72068, "loss_cls_dn_4": 0.11203, "loss_box_dn_4": 0.73102, "loss_cls_dn_5": 0.11418, "loss_box_dn_5": 0.74459, "loss_dense_depth": 0.71641, "loss": 24.97679, "grad_norm": 43.16264, "time": 1.48629} -{"mode": "train", "epoch": 1, "iter": 273, "lr": 0.00021, "memory": 49163, "data_time": 0.07543, "loss_cls_0": 0.76387, "loss_box_0": 1.67517, "loss_cns_0": 0.62393, "loss_yns_0": 0.14893, "loss_cls_1": 0.84213, "loss_box_1": 1.56083, "loss_cns_1": 0.65516, "loss_yns_1": 0.14656, "loss_cls_2": 0.85073, "loss_box_2": 1.52823, "loss_cns_2": 0.65703, "loss_yns_2": 0.1461, "loss_cls_3": 0.85731, "loss_box_3": 1.54266, "loss_cns_3": 0.65833, "loss_yns_3": 0.14708, "loss_cls_4": 0.85351, "loss_box_4": 1.53958, "loss_cns_4": 0.65944, "loss_yns_4": 0.14679, "loss_cls_5": 0.86366, "loss_box_5": 1.53648, "loss_cns_5": 0.65561, "loss_yns_5": 0.14798, "loss_cls_dn_0": 0.15584, "loss_box_dn_0": 0.73871, "loss_cls_dn_1": 0.11058, "loss_box_dn_1": 0.71617, "loss_cls_dn_2": 0.11229, "loss_box_dn_2": 0.70765, "loss_cls_dn_3": 0.11568, "loss_box_dn_3": 0.72514, "loss_cls_dn_4": 0.11786, "loss_box_dn_4": 0.73396, "loss_cls_dn_5": 0.12259, "loss_box_dn_5": 0.74672, "loss_dense_depth": 0.72998, "loss": 25.04025, "grad_norm": 45.9547, "time": 1.5046} -{"mode": "train", "epoch": 1, "iter": 274, "lr": 0.00021, "memory": 49163, "data_time": 0.07765, "loss_cls_0": 0.75425, "loss_box_0": 1.66958, "loss_cns_0": 0.62755, "loss_yns_0": 0.14659, "loss_cls_1": 0.82319, "loss_box_1": 1.54971, "loss_cns_1": 0.65635, "loss_yns_1": 0.14565, "loss_cls_2": 0.841, "loss_box_2": 1.51416, "loss_cns_2": 0.65938, "loss_yns_2": 0.14486, "loss_cls_3": 0.84539, "loss_box_3": 1.51133, "loss_cns_3": 0.66325, "loss_yns_3": 0.14514, "loss_cls_4": 0.83759, "loss_box_4": 1.511, "loss_cns_4": 0.66434, "loss_yns_4": 0.14544, "loss_cls_5": 0.8468, "loss_box_5": 1.50793, "loss_cns_5": 0.65981, "loss_yns_5": 0.14575, "loss_cls_dn_0": 0.15791, "loss_box_dn_0": 0.73773, "loss_cls_dn_1": 0.10896, "loss_box_dn_1": 0.69084, "loss_cls_dn_2": 0.10875, "loss_box_dn_2": 0.68277, "loss_cls_dn_3": 0.11152, "loss_box_dn_3": 0.68856, "loss_cls_dn_4": 0.11543, "loss_box_dn_4": 0.69672, "loss_cls_dn_5": 0.11818, "loss_box_dn_5": 0.7059, "loss_dense_depth": 0.73839, "loss": 24.67767, "grad_norm": 40.93837, "time": 1.54136} -{"mode": "train", "epoch": 1, "iter": 275, "lr": 0.00021, "memory": 49163, "data_time": 0.077, "loss_cls_0": 0.7405, "loss_box_0": 1.63742, "loss_cns_0": 0.62855, "loss_yns_0": 0.14622, "loss_cls_1": 0.79482, "loss_box_1": 1.53371, "loss_cns_1": 0.6561, "loss_yns_1": 0.14353, "loss_cls_2": 0.81759, "loss_box_2": 1.48545, "loss_cns_2": 0.65981, "loss_yns_2": 0.14186, "loss_cls_3": 0.81773, "loss_box_3": 1.47696, "loss_cns_3": 0.65885, "loss_yns_3": 0.14251, "loss_cls_4": 0.81791, "loss_box_4": 1.47711, "loss_cns_4": 0.65852, "loss_yns_4": 0.14154, "loss_cls_5": 0.8267, "loss_box_5": 1.48038, "loss_cns_5": 0.65975, "loss_yns_5": 0.14183, "loss_cls_dn_0": 0.15805, "loss_box_dn_0": 0.73032, "loss_cls_dn_1": 0.10589, "loss_box_dn_1": 0.68434, "loss_cls_dn_2": 0.10912, "loss_box_dn_2": 0.66879, "loss_cls_dn_3": 0.10882, "loss_box_dn_3": 0.6701, "loss_cls_dn_4": 0.11132, "loss_box_dn_4": 0.67871, "loss_cls_dn_5": 0.11501, "loss_box_dn_5": 0.68639, "loss_dense_depth": 0.70999, "loss": 24.22219, "grad_norm": 32.0934, "time": 1.5006} -{"mode": "train", "epoch": 1, "iter": 276, "lr": 0.00021, "memory": 49163, "data_time": 0.07787, "loss_cls_0": 0.75136, "loss_box_0": 1.64559, "loss_cns_0": 0.62398, "loss_yns_0": 0.14549, "loss_cls_1": 0.82414, "loss_box_1": 1.52378, "loss_cns_1": 0.65989, "loss_yns_1": 0.14371, "loss_cls_2": 0.84341, "loss_box_2": 1.49972, "loss_cns_2": 0.66156, "loss_yns_2": 0.14126, "loss_cls_3": 0.85553, "loss_box_3": 1.48583, "loss_cns_3": 0.65995, "loss_yns_3": 0.14079, "loss_cls_4": 0.85644, "loss_box_4": 1.47932, "loss_cns_4": 0.66143, "loss_yns_4": 0.13902, "loss_cls_5": 0.86906, "loss_box_5": 1.48161, "loss_cns_5": 0.66238, "loss_yns_5": 0.1406, "loss_cls_dn_0": 0.15338, "loss_box_dn_0": 0.73024, "loss_cls_dn_1": 0.10788, "loss_box_dn_1": 0.67063, "loss_cls_dn_2": 0.11133, "loss_box_dn_2": 0.65734, "loss_cls_dn_3": 0.11055, "loss_box_dn_3": 0.65213, "loss_cls_dn_4": 0.11243, "loss_box_dn_4": 0.65373, "loss_cls_dn_5": 0.11564, "loss_box_dn_5": 0.65497, "loss_dense_depth": 0.7279, "loss": 24.35398, "grad_norm": 44.70445, "time": 1.50008} -{"mode": "train", "epoch": 1, "iter": 277, "lr": 0.00021, "memory": 49163, "data_time": 0.08158, "loss_cls_0": 0.74769, "loss_box_0": 1.67213, "loss_cns_0": 0.62589, "loss_yns_0": 0.14458, "loss_cls_1": 0.82372, "loss_box_1": 1.58594, "loss_cns_1": 0.6529, "loss_yns_1": 0.14438, "loss_cls_2": 0.83591, "loss_box_2": 1.55984, "loss_cns_2": 0.65596, "loss_yns_2": 0.14288, "loss_cls_3": 0.85181, "loss_box_3": 1.54903, "loss_cns_3": 0.65655, "loss_yns_3": 0.14274, "loss_cls_4": 0.84934, "loss_box_4": 1.53001, "loss_cns_4": 0.65735, "loss_yns_4": 0.14327, "loss_cls_5": 0.85723, "loss_box_5": 1.52916, "loss_cns_5": 0.6568, "loss_yns_5": 0.14268, "loss_cls_dn_0": 0.14832, "loss_box_dn_0": 0.72824, "loss_cls_dn_1": 0.10658, "loss_box_dn_1": 0.65881, "loss_cls_dn_2": 0.10677, "loss_box_dn_2": 0.64256, "loss_cls_dn_3": 0.10749, "loss_box_dn_3": 0.63955, "loss_cls_dn_4": 0.11067, "loss_box_dn_4": 0.6357, "loss_cls_dn_5": 0.11317, "loss_box_dn_5": 0.63463, "loss_dense_depth": 0.73361, "loss": 24.52386, "grad_norm": 40.74448, "time": 1.54899} -{"mode": "train", "epoch": 1, "iter": 278, "lr": 0.00021, "memory": 49163, "data_time": 0.08056, "loss_cls_0": 0.76877, "loss_box_0": 1.64655, "loss_cns_0": 0.62193, "loss_yns_0": 0.14621, "loss_cls_1": 0.82483, "loss_box_1": 1.60921, "loss_cns_1": 0.6528, "loss_yns_1": 0.14752, "loss_cls_2": 0.83385, "loss_box_2": 1.54799, "loss_cns_2": 0.66025, "loss_yns_2": 0.14734, "loss_cls_3": 0.83796, "loss_box_3": 1.55328, "loss_cns_3": 0.65813, "loss_yns_3": 0.14589, "loss_cls_4": 0.84922, "loss_box_4": 1.53664, "loss_cns_4": 0.65926, "loss_yns_4": 0.14586, "loss_cls_5": 0.851, "loss_box_5": 1.53791, "loss_cns_5": 0.65875, "loss_yns_5": 0.14474, "loss_cls_dn_0": 0.15125, "loss_box_dn_0": 0.72405, "loss_cls_dn_1": 0.10697, "loss_box_dn_1": 0.66368, "loss_cls_dn_2": 0.10427, "loss_box_dn_2": 0.64191, "loss_cls_dn_3": 0.10423, "loss_box_dn_3": 0.6501, "loss_cls_dn_4": 0.1102, "loss_box_dn_4": 0.6464, "loss_cls_dn_5": 0.11191, "loss_box_dn_5": 0.65032, "loss_dense_depth": 0.698, "loss": 24.54918, "grad_norm": 39.35755, "time": 1.49375} -{"mode": "train", "epoch": 1, "iter": 279, "lr": 0.00021, "memory": 49163, "data_time": 0.07889, "loss_cls_0": 0.78423, "loss_box_0": 1.63265, "loss_cns_0": 0.62334, "loss_yns_0": 0.14774, "loss_cls_1": 0.84906, "loss_box_1": 1.58074, "loss_cns_1": 0.65636, "loss_yns_1": 0.14798, "loss_cls_2": 0.85927, "loss_box_2": 1.53418, "loss_cns_2": 0.66162, "loss_yns_2": 0.14844, "loss_cls_3": 0.85378, "loss_box_3": 1.5417, "loss_cns_3": 0.65956, "loss_yns_3": 0.14616, "loss_cls_4": 0.86754, "loss_box_4": 1.53102, "loss_cns_4": 0.66017, "loss_yns_4": 0.14515, "loss_cls_5": 0.87046, "loss_box_5": 1.53535, "loss_cns_5": 0.66014, "loss_yns_5": 0.14638, "loss_cls_dn_0": 0.15934, "loss_box_dn_0": 0.72662, "loss_cls_dn_1": 0.10954, "loss_box_dn_1": 0.66541, "loss_cls_dn_2": 0.10784, "loss_box_dn_2": 0.64807, "loss_cls_dn_3": 0.10877, "loss_box_dn_3": 0.65803, "loss_cls_dn_4": 0.11507, "loss_box_dn_4": 0.65973, "loss_cls_dn_5": 0.11666, "loss_box_dn_5": 0.66612, "loss_dense_depth": 0.73213, "loss": 24.71635, "grad_norm": 38.94103, "time": 1.50328} -{"mode": "train", "epoch": 1, "iter": 280, "lr": 0.00021, "memory": 49163, "data_time": 0.0815, "loss_cls_0": 0.76413, "loss_box_0": 1.65658, "loss_cns_0": 0.62347, "loss_yns_0": 0.14685, "loss_cls_1": 0.83308, "loss_box_1": 1.60895, "loss_cns_1": 0.65215, "loss_yns_1": 0.14723, "loss_cls_2": 0.84582, "loss_box_2": 1.57694, "loss_cns_2": 0.65566, "loss_yns_2": 0.14561, "loss_cls_3": 0.84341, "loss_box_3": 1.57182, "loss_cns_3": 0.65609, "loss_yns_3": 0.14529, "loss_cls_4": 0.85149, "loss_box_4": 1.58086, "loss_cns_4": 0.65619, "loss_yns_4": 0.14476, "loss_cls_5": 0.85683, "loss_box_5": 1.58601, "loss_cns_5": 0.65532, "loss_yns_5": 0.14499, "loss_cls_dn_0": 0.15223, "loss_box_dn_0": 0.73084, "loss_cls_dn_1": 0.10845, "loss_box_dn_1": 0.69229, "loss_cls_dn_2": 0.10828, "loss_box_dn_2": 0.67712, "loss_cls_dn_3": 0.10935, "loss_box_dn_3": 0.6826, "loss_cls_dn_4": 0.11228, "loss_box_dn_4": 0.69289, "loss_cls_dn_5": 0.1134, "loss_box_dn_5": 0.70014, "loss_dense_depth": 0.7083, "loss": 24.93769, "grad_norm": 47.51827, "time": 1.50441} -{"mode": "train", "epoch": 1, "iter": 281, "lr": 0.00021, "memory": 49163, "data_time": 0.08115, "loss_cls_0": 0.75206, "loss_box_0": 1.64712, "loss_cns_0": 0.62709, "loss_yns_0": 0.1462, "loss_cls_1": 0.82095, "loss_box_1": 1.56821, "loss_cns_1": 0.65106, "loss_yns_1": 0.14331, "loss_cls_2": 0.82886, "loss_box_2": 1.52824, "loss_cns_2": 0.65479, "loss_yns_2": 0.14167, "loss_cls_3": 0.82921, "loss_box_3": 1.52308, "loss_cns_3": 0.65692, "loss_yns_3": 0.1415, "loss_cls_4": 0.8349, "loss_box_4": 1.53246, "loss_cns_4": 0.65724, "loss_yns_4": 0.14258, "loss_cls_5": 0.83692, "loss_box_5": 1.53075, "loss_cns_5": 0.65782, "loss_yns_5": 0.14277, "loss_cls_dn_0": 0.15257, "loss_box_dn_0": 0.73032, "loss_cls_dn_1": 0.10679, "loss_box_dn_1": 0.69689, "loss_cls_dn_2": 0.10772, "loss_box_dn_2": 0.68449, "loss_cls_dn_3": 0.10564, "loss_box_dn_3": 0.68933, "loss_cls_dn_4": 0.10731, "loss_box_dn_4": 0.69979, "loss_cls_dn_5": 0.11127, "loss_box_dn_5": 0.70441, "loss_dense_depth": 0.72386, "loss": 24.61609, "grad_norm": 34.68447, "time": 1.53974} -{"mode": "train", "epoch": 1, "iter": 282, "lr": 0.00021, "memory": 49163, "data_time": 0.07796, "loss_cls_0": 0.7775, "loss_box_0": 1.64345, "loss_cns_0": 0.62886, "loss_yns_0": 0.14974, "loss_cls_1": 0.8282, "loss_box_1": 1.55746, "loss_cns_1": 0.65588, "loss_yns_1": 0.14564, "loss_cls_2": 0.84144, "loss_box_2": 1.50747, "loss_cns_2": 0.65183, "loss_yns_2": 0.14489, "loss_cls_3": 0.84374, "loss_box_3": 1.50916, "loss_cns_3": 0.65442, "loss_yns_3": 0.14421, "loss_cls_4": 0.84188, "loss_box_4": 1.51493, "loss_cns_4": 0.65797, "loss_yns_4": 0.14553, "loss_cls_5": 0.8444, "loss_box_5": 1.52079, "loss_cns_5": 0.66019, "loss_yns_5": 0.14581, "loss_cls_dn_0": 0.15255, "loss_box_dn_0": 0.73167, "loss_cls_dn_1": 0.10855, "loss_box_dn_1": 0.7089, "loss_cls_dn_2": 0.10971, "loss_box_dn_2": 0.69784, "loss_cls_dn_3": 0.11014, "loss_box_dn_3": 0.70375, "loss_cls_dn_4": 0.11533, "loss_box_dn_4": 0.70858, "loss_cls_dn_5": 0.12311, "loss_box_dn_5": 0.71433, "loss_dense_depth": 0.70464, "loss": 24.7045, "grad_norm": 42.54821, "time": 1.5594} -{"mode": "train", "epoch": 1, "iter": 283, "lr": 0.00021, "memory": 49163, "data_time": 0.08206, "loss_cls_0": 0.78713, "loss_box_0": 1.65184, "loss_cns_0": 0.62675, "loss_yns_0": 0.14923, "loss_cls_1": 0.83243, "loss_box_1": 1.60084, "loss_cns_1": 0.64973, "loss_yns_1": 0.14564, "loss_cls_2": 0.82789, "loss_box_2": 1.55526, "loss_cns_2": 0.65156, "loss_yns_2": 0.14438, "loss_cls_3": 0.83241, "loss_box_3": 1.54505, "loss_cns_3": 0.6517, "loss_yns_3": 0.14444, "loss_cls_4": 0.8465, "loss_box_4": 1.54795, "loss_cns_4": 0.65364, "loss_yns_4": 0.14531, "loss_cls_5": 0.83996, "loss_box_5": 1.5493, "loss_cns_5": 0.65235, "loss_yns_5": 0.14558, "loss_cls_dn_0": 0.1514, "loss_box_dn_0": 0.72604, "loss_cls_dn_1": 0.11516, "loss_box_dn_1": 0.69757, "loss_cls_dn_2": 0.11622, "loss_box_dn_2": 0.68016, "loss_cls_dn_3": 0.11775, "loss_box_dn_3": 0.68034, "loss_cls_dn_4": 0.12463, "loss_box_dn_4": 0.67975, "loss_cls_dn_5": 0.12886, "loss_box_dn_5": 0.68743, "loss_dense_depth": 0.70814, "loss": 24.79032, "grad_norm": 41.39184, "time": 1.51893} -{"mode": "train", "epoch": 1, "iter": 284, "lr": 0.00021, "memory": 49163, "data_time": 0.07612, "loss_cls_0": 0.76178, "loss_box_0": 1.63613, "loss_cns_0": 0.62614, "loss_yns_0": 0.15097, "loss_cls_1": 0.8269, "loss_box_1": 1.55545, "loss_cns_1": 0.65172, "loss_yns_1": 0.14849, "loss_cls_2": 0.83488, "loss_box_2": 1.53085, "loss_cns_2": 0.65449, "loss_yns_2": 0.1472, "loss_cls_3": 0.84592, "loss_box_3": 1.50973, "loss_cns_3": 0.65164, "loss_yns_3": 0.14657, "loss_cls_4": 0.85249, "loss_box_4": 1.51264, "loss_cns_4": 0.65104, "loss_yns_4": 0.14702, "loss_cls_5": 0.85707, "loss_box_5": 1.5038, "loss_cns_5": 0.65262, "loss_yns_5": 0.14604, "loss_cls_dn_0": 0.15184, "loss_box_dn_0": 0.72595, "loss_cls_dn_1": 0.11668, "loss_box_dn_1": 0.68483, "loss_cls_dn_2": 0.11563, "loss_box_dn_2": 0.66663, "loss_cls_dn_3": 0.11474, "loss_box_dn_3": 0.66256, "loss_cls_dn_4": 0.12102, "loss_box_dn_4": 0.66289, "loss_cls_dn_5": 0.119, "loss_box_dn_5": 0.66652, "loss_dense_depth": 0.69935, "loss": 24.50924, "grad_norm": 41.75801, "time": 1.54577} -{"mode": "train", "epoch": 1, "iter": 285, "lr": 0.00021, "memory": 49163, "data_time": 0.07427, "loss_cls_0": 0.78952, "loss_box_0": 1.63146, "loss_cns_0": 0.62247, "loss_yns_0": 0.15131, "loss_cls_1": 0.85612, "loss_box_1": 1.52083, "loss_cns_1": 0.652, "loss_yns_1": 0.14936, "loss_cls_2": 0.85293, "loss_box_2": 1.49772, "loss_cns_2": 0.65556, "loss_yns_2": 0.15049, "loss_cls_3": 0.85886, "loss_box_3": 1.48168, "loss_cns_3": 0.65548, "loss_yns_3": 0.14928, "loss_cls_4": 0.87152, "loss_box_4": 1.47373, "loss_cns_4": 0.65373, "loss_yns_4": 0.14946, "loss_cls_5": 0.8861, "loss_box_5": 1.47353, "loss_cns_5": 0.65402, "loss_yns_5": 0.14802, "loss_cls_dn_0": 0.15765, "loss_box_dn_0": 0.73478, "loss_cls_dn_1": 0.11498, "loss_box_dn_1": 0.66701, "loss_cls_dn_2": 0.11268, "loss_box_dn_2": 0.65013, "loss_cls_dn_3": 0.11261, "loss_box_dn_3": 0.64566, "loss_cls_dn_4": 0.11637, "loss_box_dn_4": 0.64786, "loss_cls_dn_5": 0.11654, "loss_box_dn_5": 0.65165, "loss_dense_depth": 0.73358, "loss": 24.44666, "grad_norm": 25.21062, "time": 1.49458} -{"mode": "train", "epoch": 1, "iter": 286, "lr": 0.00021, "memory": 49163, "data_time": 0.07829, "loss_cls_0": 0.76122, "loss_box_0": 1.63879, "loss_cns_0": 0.62287, "loss_yns_0": 0.1528, "loss_cls_1": 0.84451, "loss_box_1": 1.5435, "loss_cns_1": 0.65272, "loss_yns_1": 0.15101, "loss_cls_2": 0.85243, "loss_box_2": 1.52048, "loss_cns_2": 0.65484, "loss_yns_2": 0.14958, "loss_cls_3": 0.85449, "loss_box_3": 1.50589, "loss_cns_3": 0.65448, "loss_yns_3": 0.14953, "loss_cls_4": 0.85779, "loss_box_4": 1.48866, "loss_cns_4": 0.65406, "loss_yns_4": 0.14953, "loss_cls_5": 0.85326, "loss_box_5": 1.50565, "loss_cns_5": 0.65455, "loss_yns_5": 0.14885, "loss_cls_dn_0": 0.14975, "loss_box_dn_0": 0.72339, "loss_cls_dn_1": 0.1095, "loss_box_dn_1": 0.66111, "loss_cls_dn_2": 0.10834, "loss_box_dn_2": 0.64714, "loss_cls_dn_3": 0.10925, "loss_box_dn_3": 0.64661, "loss_cls_dn_4": 0.11266, "loss_box_dn_4": 0.6473, "loss_cls_dn_5": 0.11543, "loss_box_dn_5": 0.65448, "loss_dense_depth": 0.72153, "loss": 24.42797, "grad_norm": 40.01304, "time": 1.51944} -{"mode": "train", "epoch": 1, "iter": 287, "lr": 0.00021, "memory": 49163, "data_time": 0.08518, "loss_cls_0": 0.78951, "loss_box_0": 1.63951, "loss_cns_0": 0.62394, "loss_yns_0": 0.15442, "loss_cls_1": 0.8616, "loss_box_1": 1.60472, "loss_cns_1": 0.6558, "loss_yns_1": 0.15285, "loss_cls_2": 0.87154, "loss_box_2": 1.56672, "loss_cns_2": 0.66054, "loss_yns_2": 0.15345, "loss_cls_3": 0.8852, "loss_box_3": 1.55472, "loss_cns_3": 0.65756, "loss_yns_3": 0.15246, "loss_cls_4": 0.87199, "loss_box_4": 1.54591, "loss_cns_4": 0.65538, "loss_yns_4": 0.15262, "loss_cls_5": 0.88907, "loss_box_5": 1.57009, "loss_cns_5": 0.65704, "loss_yns_5": 0.15263, "loss_cls_dn_0": 0.15389, "loss_box_dn_0": 0.73065, "loss_cls_dn_1": 0.11372, "loss_box_dn_1": 0.67796, "loss_cls_dn_2": 0.1176, "loss_box_dn_2": 0.66219, "loss_cls_dn_3": 0.12071, "loss_box_dn_3": 0.66431, "loss_cls_dn_4": 0.12863, "loss_box_dn_4": 0.66956, "loss_cls_dn_5": 0.12515, "loss_box_dn_5": 0.68019, "loss_dense_depth": 0.72741, "loss": 25.05124, "grad_norm": 43.72089, "time": 1.52956} -{"mode": "train", "epoch": 1, "iter": 288, "lr": 0.00021, "memory": 49163, "data_time": 0.0857, "loss_cls_0": 0.79205, "loss_box_0": 1.65181, "loss_cns_0": 0.62524, "loss_yns_0": 0.15393, "loss_cls_1": 0.86577, "loss_box_1": 1.62072, "loss_cns_1": 0.65607, "loss_yns_1": 0.15253, "loss_cls_2": 0.86712, "loss_box_2": 1.58315, "loss_cns_2": 0.66136, "loss_yns_2": 0.15235, "loss_cls_3": 0.87213, "loss_box_3": 1.56804, "loss_cns_3": 0.65629, "loss_yns_3": 0.15282, "loss_cls_4": 0.87031, "loss_box_4": 1.57239, "loss_cns_4": 0.6557, "loss_yns_4": 0.15221, "loss_cls_5": 0.88959, "loss_box_5": 1.58919, "loss_cns_5": 0.65625, "loss_yns_5": 0.15149, "loss_cls_dn_0": 0.15124, "loss_box_dn_0": 0.73807, "loss_cls_dn_1": 0.11261, "loss_box_dn_1": 0.68363, "loss_cls_dn_2": 0.11818, "loss_box_dn_2": 0.6693, "loss_cls_dn_3": 0.11931, "loss_box_dn_3": 0.66955, "loss_cls_dn_4": 0.12509, "loss_box_dn_4": 0.6748, "loss_cls_dn_5": 0.11974, "loss_box_dn_5": 0.68518, "loss_dense_depth": 0.71968, "loss": 25.15489, "grad_norm": 32.00537, "time": 1.49668} -{"mode": "train", "epoch": 1, "iter": 289, "lr": 0.00022, "memory": 49163, "data_time": 0.16807, "loss_cls_0": 0.74811, "loss_box_0": 1.6489, "loss_cns_0": 0.62514, "loss_yns_0": 0.15158, "loss_cls_1": 0.84425, "loss_box_1": 1.54582, "loss_cns_1": 0.65752, "loss_yns_1": 0.15137, "loss_cls_2": 0.84991, "loss_box_2": 1.52809, "loss_cns_2": 0.66221, "loss_yns_2": 0.15136, "loss_cls_3": 0.84322, "loss_box_3": 1.52626, "loss_cns_3": 0.65967, "loss_yns_3": 0.15147, "loss_cls_4": 0.84121, "loss_box_4": 1.51555, "loss_cns_4": 0.6588, "loss_yns_4": 0.15097, "loss_cls_5": 0.84169, "loss_box_5": 1.51711, "loss_cns_5": 0.66082, "loss_yns_5": 0.1506, "loss_cls_dn_0": 0.14313, "loss_box_dn_0": 0.73212, "loss_cls_dn_1": 0.11268, "loss_box_dn_1": 0.678, "loss_cls_dn_2": 0.11422, "loss_box_dn_2": 0.66524, "loss_cls_dn_3": 0.11398, "loss_box_dn_3": 0.66863, "loss_cls_dn_4": 0.1136, "loss_box_dn_4": 0.66715, "loss_cls_dn_5": 0.11466, "loss_box_dn_5": 0.66832, "loss_dense_depth": 0.68613, "loss": 24.55949, "grad_norm": 34.2499, "time": 1.56543} -{"mode": "train", "epoch": 1, "iter": 290, "lr": 0.00022, "memory": 49163, "data_time": 0.07664, "loss_cls_0": 0.78261, "loss_box_0": 1.64385, "loss_cns_0": 0.61456, "loss_yns_0": 0.14969, "loss_cls_1": 0.85313, "loss_box_1": 1.51197, "loss_cns_1": 0.65514, "loss_yns_1": 0.14949, "loss_cls_2": 0.85969, "loss_box_2": 1.49104, "loss_cns_2": 0.65635, "loss_yns_2": 0.14952, "loss_cls_3": 0.86589, "loss_box_3": 1.48945, "loss_cns_3": 0.65527, "loss_yns_3": 0.14947, "loss_cls_4": 0.85733, "loss_box_4": 1.47904, "loss_cns_4": 0.65635, "loss_yns_4": 0.15058, "loss_cls_5": 0.86176, "loss_box_5": 1.48067, "loss_cns_5": 0.66222, "loss_yns_5": 0.15146, "loss_cls_dn_0": 0.14922, "loss_box_dn_0": 0.72944, "loss_cls_dn_1": 0.11171, "loss_box_dn_1": 0.67028, "loss_cls_dn_2": 0.11133, "loss_box_dn_2": 0.65902, "loss_cls_dn_3": 0.11162, "loss_box_dn_3": 0.66261, "loss_cls_dn_4": 0.11516, "loss_box_dn_4": 0.65728, "loss_cls_dn_5": 0.11639, "loss_box_dn_5": 0.65633, "loss_dense_depth": 0.73751, "loss": 24.46441, "grad_norm": 28.648, "time": 1.49979} -{"mode": "train", "epoch": 1, "iter": 291, "lr": 0.00022, "memory": 49163, "data_time": 0.0773, "loss_cls_0": 0.78674, "loss_box_0": 1.6549, "loss_cns_0": 0.625, "loss_yns_0": 0.15206, "loss_cls_1": 0.85448, "loss_box_1": 1.53492, "loss_cns_1": 0.65939, "loss_yns_1": 0.15077, "loss_cls_2": 0.85766, "loss_box_2": 1.50597, "loss_cns_2": 0.6658, "loss_yns_2": 0.15178, "loss_cls_3": 0.86061, "loss_box_3": 1.49475, "loss_cns_3": 0.65888, "loss_yns_3": 0.15082, "loss_cls_4": 0.86205, "loss_box_4": 1.48905, "loss_cns_4": 0.65883, "loss_yns_4": 0.15008, "loss_cls_5": 0.87029, "loss_box_5": 1.49761, "loss_cns_5": 0.66073, "loss_yns_5": 0.15101, "loss_cls_dn_0": 0.15255, "loss_box_dn_0": 0.72726, "loss_cls_dn_1": 0.10842, "loss_box_dn_1": 0.66236, "loss_cls_dn_2": 0.1102, "loss_box_dn_2": 0.64915, "loss_cls_dn_3": 0.10992, "loss_box_dn_3": 0.64731, "loss_cls_dn_4": 0.11089, "loss_box_dn_4": 0.64327, "loss_cls_dn_5": 0.11463, "loss_box_dn_5": 0.64534, "loss_dense_depth": 0.7493, "loss": 24.53477, "grad_norm": 25.54021, "time": 1.5113} -{"mode": "train", "epoch": 1, "iter": 292, "lr": 0.00022, "memory": 49163, "data_time": 0.07689, "loss_cls_0": 0.74924, "loss_box_0": 1.66844, "loss_cns_0": 0.62388, "loss_yns_0": 0.15178, "loss_cls_1": 0.84162, "loss_box_1": 1.53989, "loss_cns_1": 0.65818, "loss_yns_1": 0.14807, "loss_cls_2": 0.84059, "loss_box_2": 1.5143, "loss_cns_2": 0.66398, "loss_yns_2": 0.14904, "loss_cls_3": 0.83913, "loss_box_3": 1.50747, "loss_cns_3": 0.66099, "loss_yns_3": 0.14877, "loss_cls_4": 0.84509, "loss_box_4": 1.50014, "loss_cns_4": 0.66048, "loss_yns_4": 0.14888, "loss_cls_5": 0.85368, "loss_box_5": 1.50249, "loss_cns_5": 0.66038, "loss_yns_5": 0.14819, "loss_cls_dn_0": 0.14803, "loss_box_dn_0": 0.74062, "loss_cls_dn_1": 0.11325, "loss_box_dn_1": 0.65562, "loss_cls_dn_2": 0.11472, "loss_box_dn_2": 0.63739, "loss_cls_dn_3": 0.1131, "loss_box_dn_3": 0.63444, "loss_cls_dn_4": 0.11188, "loss_box_dn_4": 0.63223, "loss_cls_dn_5": 0.11526, "loss_box_dn_5": 0.63141, "loss_dense_depth": 0.70491, "loss": 24.37756, "grad_norm": 31.73069, "time": 1.50189} -{"mode": "train", "epoch": 1, "iter": 293, "lr": 0.00022, "memory": 49163, "data_time": 0.07126, "loss_cls_0": 0.77064, "loss_box_0": 1.68704, "loss_cns_0": 0.62053, "loss_yns_0": 0.14758, "loss_cls_1": 0.84382, "loss_box_1": 1.53981, "loss_cns_1": 0.656, "loss_yns_1": 0.14625, "loss_cls_2": 0.85121, "loss_box_2": 1.5084, "loss_cns_2": 0.6602, "loss_yns_2": 0.14556, "loss_cls_3": 0.8454, "loss_box_3": 1.51075, "loss_cns_3": 0.65879, "loss_yns_3": 0.14537, "loss_cls_4": 0.85322, "loss_box_4": 1.50132, "loss_cns_4": 0.65637, "loss_yns_4": 0.14532, "loss_cls_5": 0.84783, "loss_box_5": 1.52011, "loss_cns_5": 0.66066, "loss_yns_5": 0.14616, "loss_cls_dn_0": 0.15308, "loss_box_dn_0": 0.73314, "loss_cls_dn_1": 0.10963, "loss_box_dn_1": 0.65843, "loss_cls_dn_2": 0.1107, "loss_box_dn_2": 0.64028, "loss_cls_dn_3": 0.10805, "loss_box_dn_3": 0.64077, "loss_cls_dn_4": 0.11036, "loss_box_dn_4": 0.64146, "loss_cls_dn_5": 0.11167, "loss_box_dn_5": 0.64821, "loss_dense_depth": 0.73502, "loss": 24.46912, "grad_norm": 25.71359, "time": 1.50757} -{"mode": "train", "epoch": 1, "iter": 294, "lr": 0.00022, "memory": 49163, "data_time": 0.07412, "loss_cls_0": 0.784, "loss_box_0": 1.68359, "loss_cns_0": 0.62456, "loss_yns_0": 0.14948, "loss_cls_1": 0.83518, "loss_box_1": 1.5659, "loss_cns_1": 0.65456, "loss_yns_1": 0.14607, "loss_cls_2": 0.84581, "loss_box_2": 1.52335, "loss_cns_2": 0.65791, "loss_yns_2": 0.14523, "loss_cls_3": 0.84623, "loss_box_3": 1.52945, "loss_cns_3": 0.65835, "loss_yns_3": 0.14542, "loss_cls_4": 0.84308, "loss_box_4": 1.52581, "loss_cns_4": 0.65802, "loss_yns_4": 0.14548, "loss_cls_5": 0.84436, "loss_box_5": 1.53388, "loss_cns_5": 0.65813, "loss_yns_5": 0.14629, "loss_cls_dn_0": 0.14832, "loss_box_dn_0": 0.724, "loss_cls_dn_1": 0.10654, "loss_box_dn_1": 0.6578, "loss_cls_dn_2": 0.10771, "loss_box_dn_2": 0.63979, "loss_cls_dn_3": 0.10684, "loss_box_dn_3": 0.64292, "loss_cls_dn_4": 0.10985, "loss_box_dn_4": 0.64366, "loss_cls_dn_5": 0.11153, "loss_box_dn_5": 0.65127, "loss_dense_depth": 0.7579, "loss": 24.55828, "grad_norm": 29.01145, "time": 1.55325} -{"mode": "train", "epoch": 1, "iter": 295, "lr": 0.00022, "memory": 49163, "data_time": 0.07037, "loss_cls_0": 0.7728, "loss_box_0": 1.67037, "loss_cns_0": 0.62707, "loss_yns_0": 0.14904, "loss_cls_1": 0.84584, "loss_box_1": 1.55938, "loss_cns_1": 0.65929, "loss_yns_1": 0.14768, "loss_cls_2": 0.86152, "loss_box_2": 1.52051, "loss_cns_2": 0.66802, "loss_yns_2": 0.14869, "loss_cls_3": 0.85825, "loss_box_3": 1.52168, "loss_cns_3": 0.66213, "loss_yns_3": 0.14706, "loss_cls_4": 0.85784, "loss_box_4": 1.51483, "loss_cns_4": 0.66532, "loss_yns_4": 0.14794, "loss_cls_5": 0.85872, "loss_box_5": 1.52493, "loss_cns_5": 0.65951, "loss_yns_5": 0.14701, "loss_cls_dn_0": 0.14568, "loss_box_dn_0": 0.72832, "loss_cls_dn_1": 0.10865, "loss_box_dn_1": 0.6698, "loss_cls_dn_2": 0.11217, "loss_box_dn_2": 0.65316, "loss_cls_dn_3": 0.11205, "loss_box_dn_3": 0.65781, "loss_cls_dn_4": 0.11482, "loss_box_dn_4": 0.65792, "loss_cls_dn_5": 0.11661, "loss_box_dn_5": 0.66705, "loss_dense_depth": 0.70982, "loss": 24.6493, "grad_norm": 30.39915, "time": 1.5035} -{"mode": "train", "epoch": 1, "iter": 296, "lr": 0.00022, "memory": 49163, "data_time": 0.07357, "loss_cls_0": 0.77392, "loss_box_0": 1.67495, "loss_cns_0": 0.62454, "loss_yns_0": 0.15047, "loss_cls_1": 0.85186, "loss_box_1": 1.54131, "loss_cns_1": 0.6599, "loss_yns_1": 0.14752, "loss_cls_2": 0.86771, "loss_box_2": 1.51956, "loss_cns_2": 0.66743, "loss_yns_2": 0.14977, "loss_cls_3": 0.86503, "loss_box_3": 1.51071, "loss_cns_3": 0.66065, "loss_yns_3": 0.1483, "loss_cls_4": 0.86141, "loss_box_4": 1.51397, "loss_cns_4": 0.66397, "loss_yns_4": 0.14863, "loss_cls_5": 0.86552, "loss_box_5": 1.51653, "loss_cns_5": 0.6604, "loss_yns_5": 0.14815, "loss_cls_dn_0": 0.14869, "loss_box_dn_0": 0.73026, "loss_cls_dn_1": 0.112, "loss_box_dn_1": 0.66424, "loss_cls_dn_2": 0.11376, "loss_box_dn_2": 0.64895, "loss_cls_dn_3": 0.11371, "loss_box_dn_3": 0.64991, "loss_cls_dn_4": 0.11383, "loss_box_dn_4": 0.65155, "loss_cls_dn_5": 0.11703, "loss_box_dn_5": 0.65804, "loss_dense_depth": 0.75944, "loss": 24.67361, "grad_norm": 24.87562, "time": 1.50508} -{"mode": "train", "epoch": 1, "iter": 297, "lr": 0.00022, "memory": 49163, "data_time": 0.06826, "loss_cls_0": 0.79864, "loss_box_0": 1.72851, "loss_cns_0": 0.62319, "loss_yns_0": 0.15158, "loss_cls_1": 0.85543, "loss_box_1": 1.58966, "loss_cns_1": 0.65653, "loss_yns_1": 0.15068, "loss_cls_2": 0.86612, "loss_box_2": 1.56567, "loss_cns_2": 0.66012, "loss_yns_2": 0.15046, "loss_cls_3": 0.87433, "loss_box_3": 1.5604, "loss_cns_3": 0.6578, "loss_yns_3": 0.1503, "loss_cls_4": 0.87651, "loss_box_4": 1.5605, "loss_cns_4": 0.65924, "loss_yns_4": 0.15035, "loss_cls_5": 0.87679, "loss_box_5": 1.55991, "loss_cns_5": 0.65756, "loss_yns_5": 0.1508, "loss_cls_dn_0": 0.14888, "loss_box_dn_0": 0.73467, "loss_cls_dn_1": 0.11133, "loss_box_dn_1": 0.65418, "loss_cls_dn_2": 0.11007, "loss_box_dn_2": 0.6401, "loss_cls_dn_3": 0.11069, "loss_box_dn_3": 0.63932, "loss_cls_dn_4": 0.11326, "loss_box_dn_4": 0.64091, "loss_cls_dn_5": 0.11453, "loss_box_dn_5": 0.64544, "loss_dense_depth": 0.82109, "loss": 25.01557, "grad_norm": 27.68944, "time": 1.5188} -{"mode": "train", "epoch": 1, "iter": 298, "lr": 0.00022, "memory": 49163, "data_time": 0.07165, "loss_cls_0": 0.7851, "loss_box_0": 1.68396, "loss_cns_0": 0.63021, "loss_yns_0": 0.15116, "loss_cls_1": 0.85732, "loss_box_1": 1.58944, "loss_cns_1": 0.65708, "loss_yns_1": 0.15201, "loss_cls_2": 0.86009, "loss_box_2": 1.56184, "loss_cns_2": 0.65854, "loss_yns_2": 0.15161, "loss_cls_3": 0.87173, "loss_box_3": 1.55867, "loss_cns_3": 0.65886, "loss_yns_3": 0.1514, "loss_cls_4": 0.8716, "loss_box_4": 1.55223, "loss_cns_4": 0.65919, "loss_yns_4": 0.15146, "loss_cls_5": 0.8772, "loss_box_5": 1.55066, "loss_cns_5": 0.6584, "loss_yns_5": 0.15143, "loss_cls_dn_0": 0.14381, "loss_box_dn_0": 0.72128, "loss_cls_dn_1": 0.10916, "loss_box_dn_1": 0.64742, "loss_cls_dn_2": 0.10895, "loss_box_dn_2": 0.63242, "loss_cls_dn_3": 0.10956, "loss_box_dn_3": 0.62904, "loss_cls_dn_4": 0.11384, "loss_box_dn_4": 0.62817, "loss_cls_dn_5": 0.11313, "loss_box_dn_5": 0.62939, "loss_dense_depth": 0.74423, "loss": 24.78159, "grad_norm": 31.42793, "time": 1.48282} -{"mode": "train", "epoch": 1, "iter": 299, "lr": 0.00022, "memory": 49163, "data_time": 0.06988, "loss_cls_0": 0.80268, "loss_box_0": 1.68922, "loss_cns_0": 0.62468, "loss_yns_0": 0.15266, "loss_cls_1": 0.86782, "loss_box_1": 1.6236, "loss_cns_1": 0.65371, "loss_yns_1": 0.15195, "loss_cls_2": 0.87588, "loss_box_2": 1.58608, "loss_cns_2": 0.65582, "loss_yns_2": 0.15245, "loss_cls_3": 0.88176, "loss_box_3": 1.57981, "loss_cns_3": 0.65515, "loss_yns_3": 0.15181, "loss_cls_4": 0.88834, "loss_box_4": 1.57723, "loss_cns_4": 0.66114, "loss_yns_4": 0.15113, "loss_cls_5": 0.89248, "loss_box_5": 1.57459, "loss_cns_5": 0.65508, "loss_yns_5": 0.15267, "loss_cls_dn_0": 0.14169, "loss_box_dn_0": 0.73314, "loss_cls_dn_1": 0.10902, "loss_box_dn_1": 0.65497, "loss_cls_dn_2": 0.10855, "loss_box_dn_2": 0.64106, "loss_cls_dn_3": 0.10852, "loss_box_dn_3": 0.63815, "loss_cls_dn_4": 0.11194, "loss_box_dn_4": 0.63799, "loss_cls_dn_5": 0.11449, "loss_box_dn_5": 0.63719, "loss_dense_depth": 0.83632, "loss": 25.13079, "grad_norm": 26.01691, "time": 1.49133} -{"mode": "train", "epoch": 1, "iter": 300, "lr": 0.00022, "memory": 49163, "data_time": 0.07125, "loss_cls_0": 0.78229, "loss_box_0": 1.66734, "loss_cns_0": 0.6259, "loss_yns_0": 0.15269, "loss_cls_1": 0.84652, "loss_box_1": 1.60945, "loss_cns_1": 0.65459, "loss_yns_1": 0.15134, "loss_cls_2": 0.86078, "loss_box_2": 1.58192, "loss_cns_2": 0.65774, "loss_yns_2": 0.15186, "loss_cls_3": 0.85828, "loss_box_3": 1.57341, "loss_cns_3": 0.65664, "loss_yns_3": 0.15176, "loss_cls_4": 0.86337, "loss_box_4": 1.56769, "loss_cns_4": 0.66439, "loss_yns_4": 0.1504, "loss_cls_5": 0.86775, "loss_box_5": 1.57144, "loss_cns_5": 0.65618, "loss_yns_5": 0.15037, "loss_cls_dn_0": 0.14412, "loss_box_dn_0": 0.72968, "loss_cls_dn_1": 0.11079, "loss_box_dn_1": 0.65935, "loss_cls_dn_2": 0.11172, "loss_box_dn_2": 0.64631, "loss_cls_dn_3": 0.11079, "loss_box_dn_3": 0.6427, "loss_cls_dn_4": 0.11467, "loss_box_dn_4": 0.64195, "loss_cls_dn_5": 0.11703, "loss_box_dn_5": 0.6441, "loss_dense_depth": 0.72725, "loss": 24.87457, "grad_norm": 33.95119, "time": 1.48817} -{"mode": "train", "epoch": 1, "iter": 301, "lr": 0.00022, "memory": 49163, "data_time": 0.068, "loss_cls_0": 0.79953, "loss_box_0": 1.67308, "loss_cns_0": 0.62247, "loss_yns_0": 0.15031, "loss_cls_1": 0.855, "loss_box_1": 1.6308, "loss_cns_1": 0.65056, "loss_yns_1": 0.14933, "loss_cls_2": 0.86437, "loss_box_2": 1.59171, "loss_cns_2": 0.65624, "loss_yns_2": 0.14958, "loss_cls_3": 0.86766, "loss_box_3": 1.58235, "loss_cns_3": 0.65278, "loss_yns_3": 0.14961, "loss_cls_4": 0.86643, "loss_box_4": 1.58066, "loss_cns_4": 0.65545, "loss_yns_4": 0.14949, "loss_cls_5": 0.87117, "loss_box_5": 1.58485, "loss_cns_5": 0.65322, "loss_yns_5": 0.15025, "loss_cls_dn_0": 0.14172, "loss_box_dn_0": 0.7354, "loss_cls_dn_1": 0.10644, "loss_box_dn_1": 0.65652, "loss_cls_dn_2": 0.10874, "loss_box_dn_2": 0.64447, "loss_cls_dn_3": 0.10849, "loss_box_dn_3": 0.64287, "loss_cls_dn_4": 0.10892, "loss_box_dn_4": 0.64658, "loss_cls_dn_5": 0.11107, "loss_box_dn_5": 0.65246, "loss_dense_depth": 0.80611, "loss": 25.02667, "grad_norm": 23.88492, "time": 1.53001} -{"mode": "train", "epoch": 1, "iter": 302, "lr": 0.00022, "memory": 49163, "data_time": 0.06839, "loss_cls_0": 0.79001, "loss_box_0": 1.6616, "loss_cns_0": 0.62303, "loss_yns_0": 0.15199, "loss_cls_1": 0.84816, "loss_box_1": 1.58725, "loss_cns_1": 0.64781, "loss_yns_1": 0.15103, "loss_cls_2": 0.85916, "loss_box_2": 1.54845, "loss_cns_2": 0.64965, "loss_yns_2": 0.15035, "loss_cls_3": 0.85757, "loss_box_3": 1.54898, "loss_cns_3": 0.65168, "loss_yns_3": 0.15166, "loss_cls_4": 0.86147, "loss_box_4": 1.55297, "loss_cns_4": 0.65567, "loss_yns_4": 0.15226, "loss_cls_5": 0.86227, "loss_box_5": 1.55164, "loss_cns_5": 0.65236, "loss_yns_5": 0.15359, "loss_cls_dn_0": 0.13997, "loss_box_dn_0": 0.72372, "loss_cls_dn_1": 0.10589, "loss_box_dn_1": 0.65542, "loss_cls_dn_2": 0.10666, "loss_box_dn_2": 0.64383, "loss_cls_dn_3": 0.10627, "loss_box_dn_3": 0.64498, "loss_cls_dn_4": 0.10821, "loss_box_dn_4": 0.64984, "loss_cls_dn_5": 0.1094, "loss_box_dn_5": 0.65713, "loss_dense_depth": 0.74878, "loss": 24.72073, "grad_norm": 25.8651, "time": 1.5786} -{"mode": "train", "epoch": 1, "iter": 303, "lr": 0.00022, "memory": 49163, "data_time": 0.06954, "loss_cls_0": 0.79096, "loss_box_0": 1.68045, "loss_cns_0": 0.62425, "loss_yns_0": 0.15567, "loss_cls_1": 0.85891, "loss_box_1": 1.60708, "loss_cns_1": 0.64962, "loss_yns_1": 0.15393, "loss_cls_2": 0.88186, "loss_box_2": 1.56279, "loss_cns_2": 0.65279, "loss_yns_2": 0.15299, "loss_cls_3": 0.86931, "loss_box_3": 1.56153, "loss_cns_3": 0.65469, "loss_yns_3": 0.15353, "loss_cls_4": 0.86887, "loss_box_4": 1.5664, "loss_cns_4": 0.66098, "loss_yns_4": 0.153, "loss_cls_5": 0.86902, "loss_box_5": 1.57058, "loss_cns_5": 0.65509, "loss_yns_5": 0.15404, "loss_cls_dn_0": 0.14125, "loss_box_dn_0": 0.72128, "loss_cls_dn_1": 0.10626, "loss_box_dn_1": 0.66771, "loss_cls_dn_2": 0.1062, "loss_box_dn_2": 0.65408, "loss_cls_dn_3": 0.10509, "loss_box_dn_3": 0.65307, "loss_cls_dn_4": 0.10971, "loss_box_dn_4": 0.65537, "loss_cls_dn_5": 0.10886, "loss_box_dn_5": 0.66009, "loss_dense_depth": 0.71345, "loss": 24.9108, "grad_norm": 23.68938, "time": 1.48131} -{"mode": "train", "epoch": 1, "iter": 304, "lr": 0.00022, "memory": 49163, "data_time": 0.06883, "loss_cls_0": 0.77654, "loss_box_0": 1.6935, "loss_cns_0": 0.62444, "loss_yns_0": 0.15255, "loss_cls_1": 0.83695, "loss_box_1": 1.57732, "loss_cns_1": 0.6532, "loss_yns_1": 0.15263, "loss_cls_2": 0.85055, "loss_box_2": 1.54453, "loss_cns_2": 0.659, "loss_yns_2": 0.15356, "loss_cls_3": 0.85042, "loss_box_3": 1.53448, "loss_cns_3": 0.65706, "loss_yns_3": 0.15237, "loss_cls_4": 0.85312, "loss_box_4": 1.53465, "loss_cns_4": 0.65984, "loss_yns_4": 0.15403, "loss_cls_5": 0.84964, "loss_box_5": 1.53625, "loss_cns_5": 0.65635, "loss_yns_5": 0.15257, "loss_cls_dn_0": 0.14, "loss_box_dn_0": 0.72299, "loss_cls_dn_1": 0.10725, "loss_box_dn_1": 0.66712, "loss_cls_dn_2": 0.10547, "loss_box_dn_2": 0.64921, "loss_cls_dn_3": 0.10579, "loss_box_dn_3": 0.64576, "loss_cls_dn_4": 0.10783, "loss_box_dn_4": 0.64554, "loss_cls_dn_5": 0.109, "loss_box_dn_5": 0.6467, "loss_dense_depth": 0.73585, "loss": 24.65406, "grad_norm": 21.06301, "time": 1.54524} -{"mode": "train", "epoch": 1, "iter": 305, "lr": 0.00022, "memory": 49163, "data_time": 0.07349, "loss_cls_0": 0.75577, "loss_box_0": 1.68857, "loss_cns_0": 0.62891, "loss_yns_0": 0.15343, "loss_cls_1": 0.83355, "loss_box_1": 1.53796, "loss_cns_1": 0.65928, "loss_yns_1": 0.15223, "loss_cls_2": 0.84427, "loss_box_2": 1.513, "loss_cns_2": 0.66104, "loss_yns_2": 0.15205, "loss_cls_3": 0.83632, "loss_box_3": 1.50218, "loss_cns_3": 0.65913, "loss_yns_3": 0.15113, "loss_cls_4": 0.83761, "loss_box_4": 1.50108, "loss_cns_4": 0.66059, "loss_yns_4": 0.15036, "loss_cls_5": 0.83472, "loss_box_5": 1.5083, "loss_cns_5": 0.6598, "loss_yns_5": 0.15151, "loss_cls_dn_0": 0.13902, "loss_box_dn_0": 0.72412, "loss_cls_dn_1": 0.10737, "loss_box_dn_1": 0.66622, "loss_cls_dn_2": 0.10517, "loss_box_dn_2": 0.65148, "loss_cls_dn_3": 0.10442, "loss_box_dn_3": 0.64945, "loss_cls_dn_4": 0.10529, "loss_box_dn_4": 0.64944, "loss_cls_dn_5": 0.10816, "loss_box_dn_5": 0.65291, "loss_dense_depth": 0.70172, "loss": 24.39754, "grad_norm": 27.06274, "time": 1.4922} -{"mode": "train", "epoch": 1, "iter": 306, "lr": 0.00022, "memory": 49163, "data_time": 0.07518, "loss_cls_0": 0.76991, "loss_box_0": 1.66539, "loss_cns_0": 0.62427, "loss_yns_0": 0.15316, "loss_cls_1": 0.84064, "loss_box_1": 1.53785, "loss_cns_1": 0.65911, "loss_yns_1": 0.15057, "loss_cls_2": 0.85291, "loss_box_2": 1.50542, "loss_cns_2": 0.65827, "loss_yns_2": 0.15049, "loss_cls_3": 0.84624, "loss_box_3": 1.48529, "loss_cns_3": 0.65559, "loss_yns_3": 0.15105, "loss_cls_4": 0.84269, "loss_box_4": 1.49829, "loss_cns_4": 0.65677, "loss_yns_4": 0.15026, "loss_cls_5": 0.84839, "loss_box_5": 1.5027, "loss_cns_5": 0.65595, "loss_yns_5": 0.1505, "loss_cls_dn_0": 0.14324, "loss_box_dn_0": 0.72609, "loss_cls_dn_1": 0.10965, "loss_box_dn_1": 0.66204, "loss_cls_dn_2": 0.10811, "loss_box_dn_2": 0.64761, "loss_cls_dn_3": 0.10784, "loss_box_dn_3": 0.64669, "loss_cls_dn_4": 0.10922, "loss_box_dn_4": 0.64836, "loss_cls_dn_5": 0.11146, "loss_box_dn_5": 0.65249, "loss_dense_depth": 0.72402, "loss": 24.40852, "grad_norm": 29.46219, "time": 1.50523} -{"mode": "train", "epoch": 1, "iter": 307, "lr": 0.00022, "memory": 49163, "data_time": 0.07723, "loss_cls_0": 0.76244, "loss_box_0": 1.6563, "loss_cns_0": 0.62448, "loss_yns_0": 0.1533, "loss_cls_1": 0.82584, "loss_box_1": 1.52768, "loss_cns_1": 0.65625, "loss_yns_1": 0.15314, "loss_cls_2": 0.83831, "loss_box_2": 1.50392, "loss_cns_2": 0.65937, "loss_yns_2": 0.15285, "loss_cls_3": 0.83269, "loss_box_3": 1.48838, "loss_cns_3": 0.65532, "loss_yns_3": 0.15253, "loss_cls_4": 0.83377, "loss_box_4": 1.48384, "loss_cns_4": 0.65699, "loss_yns_4": 0.15186, "loss_cls_5": 0.82979, "loss_box_5": 1.49026, "loss_cns_5": 0.65739, "loss_yns_5": 0.15117, "loss_cls_dn_0": 0.14132, "loss_box_dn_0": 0.72181, "loss_cls_dn_1": 0.10753, "loss_box_dn_1": 0.66071, "loss_cls_dn_2": 0.10645, "loss_box_dn_2": 0.64595, "loss_cls_dn_3": 0.10631, "loss_box_dn_3": 0.64788, "loss_cls_dn_4": 0.10871, "loss_box_dn_4": 0.65223, "loss_cls_dn_5": 0.10974, "loss_box_dn_5": 0.65559, "loss_dense_depth": 0.70446, "loss": 24.26656, "grad_norm": 25.04902, "time": 1.5082} -{"mode": "train", "epoch": 1, "iter": 308, "lr": 0.00022, "memory": 49163, "data_time": 0.07999, "loss_cls_0": 0.78145, "loss_box_0": 1.66822, "loss_cns_0": 0.62724, "loss_yns_0": 0.15048, "loss_cls_1": 0.83745, "loss_box_1": 1.55271, "loss_cns_1": 0.65573, "loss_yns_1": 0.14837, "loss_cls_2": 0.84881, "loss_box_2": 1.5186, "loss_cns_2": 0.65761, "loss_yns_2": 0.14924, "loss_cls_3": 0.86301, "loss_box_3": 1.50838, "loss_cns_3": 0.65879, "loss_yns_3": 0.14988, "loss_cls_4": 0.86111, "loss_box_4": 1.50881, "loss_cns_4": 0.65867, "loss_yns_4": 0.15024, "loss_cls_5": 0.85374, "loss_box_5": 1.5161, "loss_cns_5": 0.65887, "loss_yns_5": 0.1506, "loss_cls_dn_0": 0.14678, "loss_box_dn_0": 0.72888, "loss_cls_dn_1": 0.10885, "loss_box_dn_1": 0.6769, "loss_cls_dn_2": 0.11015, "loss_box_dn_2": 0.66319, "loss_cls_dn_3": 0.10968, "loss_box_dn_3": 0.66599, "loss_cls_dn_4": 0.11069, "loss_box_dn_4": 0.67217, "loss_cls_dn_5": 0.113, "loss_box_dn_5": 0.67919, "loss_dense_depth": 0.69447, "loss": 24.61406, "grad_norm": 31.65915, "time": 1.53417} -{"mode": "train", "epoch": 1, "iter": 309, "lr": 0.00022, "memory": 49163, "data_time": 1.69939, "loss_cls_0": 0.78368, "loss_box_0": 1.68016, "loss_cns_0": 0.62224, "loss_yns_0": 0.15137, "loss_cls_1": 0.84134, "loss_box_1": 1.55544, "loss_cns_1": 0.65766, "loss_yns_1": 0.15008, "loss_cls_2": 0.85263, "loss_box_2": 1.52553, "loss_cns_2": 0.65974, "loss_yns_2": 0.15089, "loss_cls_3": 0.84916, "loss_box_3": 1.52571, "loss_cns_3": 0.66057, "loss_yns_3": 0.15018, "loss_cls_4": 0.8537, "loss_box_4": 1.52832, "loss_cns_4": 0.66041, "loss_yns_4": 0.15078, "loss_cls_5": 0.85314, "loss_box_5": 1.52836, "loss_cns_5": 0.66013, "loss_yns_5": 0.15067, "loss_cls_dn_0": 0.14381, "loss_box_dn_0": 0.72704, "loss_cls_dn_1": 0.10909, "loss_box_dn_1": 0.68959, "loss_cls_dn_2": 0.11435, "loss_box_dn_2": 0.67965, "loss_cls_dn_3": 0.11311, "loss_box_dn_3": 0.6861, "loss_cls_dn_4": 0.11372, "loss_box_dn_4": 0.69601, "loss_cls_dn_5": 0.11844, "loss_box_dn_5": 0.70353, "loss_dense_depth": 0.71494, "loss": 24.81127, "grad_norm": 34.65828, "time": 3.10723} -{"mode": "train", "epoch": 1, "iter": 310, "lr": 0.00022, "memory": 49163, "data_time": 0.07906, "loss_cls_0": 0.75076, "loss_box_0": 1.64581, "loss_cns_0": 0.6264, "loss_yns_0": 0.15008, "loss_cls_1": 0.83206, "loss_box_1": 1.51248, "loss_cns_1": 0.65583, "loss_yns_1": 0.14811, "loss_cls_2": 0.84511, "loss_box_2": 1.48988, "loss_cns_2": 0.65652, "loss_yns_2": 0.14715, "loss_cls_3": 0.83834, "loss_box_3": 1.49019, "loss_cns_3": 0.65927, "loss_yns_3": 0.14763, "loss_cls_4": 0.84801, "loss_box_4": 1.49203, "loss_cns_4": 0.65864, "loss_yns_4": 0.14716, "loss_cls_5": 0.84407, "loss_box_5": 1.50319, "loss_cns_5": 0.6619, "loss_yns_5": 0.14806, "loss_cls_dn_0": 0.1418, "loss_box_dn_0": 0.72855, "loss_cls_dn_1": 0.10913, "loss_box_dn_1": 0.68589, "loss_cls_dn_2": 0.1131, "loss_box_dn_2": 0.67752, "loss_cls_dn_3": 0.11031, "loss_box_dn_3": 0.68121, "loss_cls_dn_4": 0.11122, "loss_box_dn_4": 0.68854, "loss_cls_dn_5": 0.11567, "loss_box_dn_5": 0.69412, "loss_dense_depth": 0.68267, "loss": 24.43839, "grad_norm": 31.68727, "time": 1.49366} -{"mode": "train", "epoch": 1, "iter": 311, "lr": 0.00022, "memory": 49163, "data_time": 0.07576, "loss_cls_0": 0.76899, "loss_box_0": 1.65267, "loss_cns_0": 0.62599, "loss_yns_0": 0.15216, "loss_cls_1": 0.82284, "loss_box_1": 1.54021, "loss_cns_1": 0.65497, "loss_yns_1": 0.14897, "loss_cls_2": 0.83466, "loss_box_2": 1.51391, "loss_cns_2": 0.65365, "loss_yns_2": 0.14663, "loss_cls_3": 0.83902, "loss_box_3": 1.50409, "loss_cns_3": 0.65533, "loss_yns_3": 0.14779, "loss_cls_4": 0.84116, "loss_box_4": 1.51068, "loss_cns_4": 0.65737, "loss_yns_4": 0.14771, "loss_cls_5": 0.84418, "loss_box_5": 1.52556, "loss_cns_5": 0.65832, "loss_yns_5": 0.14765, "loss_cls_dn_0": 0.14472, "loss_box_dn_0": 0.73012, "loss_cls_dn_1": 0.10972, "loss_box_dn_1": 0.68663, "loss_cls_dn_2": 0.10997, "loss_box_dn_2": 0.67507, "loss_cls_dn_3": 0.10985, "loss_box_dn_3": 0.67209, "loss_cls_dn_4": 0.11119, "loss_box_dn_4": 0.67465, "loss_cls_dn_5": 0.11285, "loss_box_dn_5": 0.67813, "loss_dense_depth": 0.71895, "loss": 24.52844, "grad_norm": 31.47464, "time": 1.49031} -{"mode": "train", "epoch": 1, "iter": 312, "lr": 0.00022, "memory": 49163, "data_time": 0.07781, "loss_cls_0": 0.77631, "loss_box_0": 1.67126, "loss_cns_0": 0.62512, "loss_yns_0": 0.15233, "loss_cls_1": 0.84281, "loss_box_1": 1.55585, "loss_cns_1": 0.65512, "loss_yns_1": 0.14942, "loss_cls_2": 0.84089, "loss_box_2": 1.51764, "loss_cns_2": 0.6542, "loss_yns_2": 0.14714, "loss_cls_3": 0.84629, "loss_box_3": 1.51126, "loss_cns_3": 0.65668, "loss_yns_3": 0.1473, "loss_cls_4": 0.85322, "loss_box_4": 1.50783, "loss_cns_4": 0.65677, "loss_yns_4": 0.1481, "loss_cls_5": 0.86829, "loss_box_5": 1.50764, "loss_cns_5": 0.65692, "loss_yns_5": 0.14755, "loss_cls_dn_0": 0.14705, "loss_box_dn_0": 0.72125, "loss_cls_dn_1": 0.11212, "loss_box_dn_1": 0.6654, "loss_cls_dn_2": 0.11071, "loss_box_dn_2": 0.64777, "loss_cls_dn_3": 0.10965, "loss_box_dn_3": 0.64399, "loss_cls_dn_4": 0.11257, "loss_box_dn_4": 0.64229, "loss_cls_dn_5": 0.11243, "loss_box_dn_5": 0.64129, "loss_dense_depth": 0.71122, "loss": 24.4737, "grad_norm": 26.80961, "time": 1.49444} -{"mode": "train", "epoch": 1, "iter": 313, "lr": 0.00022, "memory": 49163, "data_time": 0.08072, "loss_cls_0": 0.76571, "loss_box_0": 1.62095, "loss_cns_0": 0.62233, "loss_yns_0": 0.15159, "loss_cls_1": 0.82309, "loss_box_1": 1.51834, "loss_cns_1": 0.65355, "loss_yns_1": 0.14981, "loss_cls_2": 0.82706, "loss_box_2": 1.48452, "loss_cns_2": 0.65752, "loss_yns_2": 0.14986, "loss_cls_3": 0.82874, "loss_box_3": 1.47967, "loss_cns_3": 0.65569, "loss_yns_3": 0.14912, "loss_cls_4": 0.83741, "loss_box_4": 1.47498, "loss_cns_4": 0.65945, "loss_yns_4": 0.14954, "loss_cls_5": 0.84108, "loss_box_5": 1.47774, "loss_cns_5": 0.65721, "loss_yns_5": 0.1485, "loss_cls_dn_0": 0.13865, "loss_box_dn_0": 0.72591, "loss_cls_dn_1": 0.10722, "loss_box_dn_1": 0.65226, "loss_cls_dn_2": 0.1049, "loss_box_dn_2": 0.63682, "loss_cls_dn_3": 0.10422, "loss_box_dn_3": 0.63883, "loss_cls_dn_4": 0.10643, "loss_box_dn_4": 0.63878, "loss_cls_dn_5": 0.10777, "loss_box_dn_5": 0.64446, "loss_dense_depth": 0.70271, "loss": 24.09244, "grad_norm": 36.45312, "time": 1.5108} -{"mode": "train", "epoch": 1, "iter": 314, "lr": 0.00023, "memory": 49163, "data_time": 0.07979, "loss_cls_0": 0.79574, "loss_box_0": 1.63387, "loss_cns_0": 0.62502, "loss_yns_0": 0.14928, "loss_cls_1": 0.83829, "loss_box_1": 1.52457, "loss_cns_1": 0.65339, "loss_yns_1": 0.14762, "loss_cls_2": 0.84391, "loss_box_2": 1.49543, "loss_cns_2": 0.65851, "loss_yns_2": 0.14892, "loss_cls_3": 0.85106, "loss_box_3": 1.48635, "loss_cns_3": 0.65667, "loss_yns_3": 0.14879, "loss_cls_4": 0.8605, "loss_box_4": 1.48091, "loss_cns_4": 0.66068, "loss_yns_4": 0.14967, "loss_cls_5": 0.86184, "loss_box_5": 1.48113, "loss_cns_5": 0.6577, "loss_yns_5": 0.14943, "loss_cls_dn_0": 0.14523, "loss_box_dn_0": 0.72608, "loss_cls_dn_1": 0.11283, "loss_box_dn_1": 0.6524, "loss_cls_dn_2": 0.11119, "loss_box_dn_2": 0.63761, "loss_cls_dn_3": 0.11111, "loss_box_dn_3": 0.64049, "loss_cls_dn_4": 0.11213, "loss_box_dn_4": 0.646, "loss_cls_dn_5": 0.11552, "loss_box_dn_5": 0.65723, "loss_dense_depth": 0.70328, "loss": 24.33039, "grad_norm": 32.03669, "time": 1.53143} -{"mode": "train", "epoch": 1, "iter": 315, "lr": 0.00023, "memory": 49163, "data_time": 0.07702, "loss_cls_0": 0.77187, "loss_box_0": 1.64385, "loss_cns_0": 0.62356, "loss_yns_0": 0.14965, "loss_cls_1": 0.84028, "loss_box_1": 1.52875, "loss_cns_1": 0.64901, "loss_yns_1": 0.14928, "loss_cls_2": 0.85365, "loss_box_2": 1.50309, "loss_cns_2": 0.65249, "loss_yns_2": 0.14793, "loss_cls_3": 0.85012, "loss_box_3": 1.49677, "loss_cns_3": 0.6508, "loss_yns_3": 0.14907, "loss_cls_4": 0.85877, "loss_box_4": 1.50072, "loss_cns_4": 0.65099, "loss_yns_4": 0.14921, "loss_cls_5": 0.86275, "loss_box_5": 1.49977, "loss_cns_5": 0.64975, "loss_yns_5": 0.14852, "loss_cls_dn_0": 0.14004, "loss_box_dn_0": 0.73471, "loss_cls_dn_1": 0.11048, "loss_box_dn_1": 0.67286, "loss_cls_dn_2": 0.1074, "loss_box_dn_2": 0.66201, "loss_cls_dn_3": 0.10687, "loss_box_dn_3": 0.6688, "loss_cls_dn_4": 0.10952, "loss_box_dn_4": 0.68173, "loss_cls_dn_5": 0.11177, "loss_box_dn_5": 0.69391, "loss_dense_depth": 0.68519, "loss": 24.46595, "grad_norm": 35.66285, "time": 1.5039} -{"mode": "train", "epoch": 1, "iter": 316, "lr": 0.00023, "memory": 49163, "data_time": 0.08128, "loss_cls_0": 0.76842, "loss_box_0": 1.61492, "loss_cns_0": 0.62689, "loss_yns_0": 0.14742, "loss_cls_1": 0.84447, "loss_box_1": 1.51336, "loss_cns_1": 0.65093, "loss_yns_1": 0.14603, "loss_cls_2": 0.84854, "loss_box_2": 1.49379, "loss_cns_2": 0.65347, "loss_yns_2": 0.14496, "loss_cls_3": 0.85521, "loss_box_3": 1.49687, "loss_cns_3": 0.65485, "loss_yns_3": 0.14682, "loss_cls_4": 0.863, "loss_box_4": 1.49222, "loss_cns_4": 0.65549, "loss_yns_4": 0.14548, "loss_cls_5": 0.86164, "loss_box_5": 1.49175, "loss_cns_5": 0.65339, "loss_yns_5": 0.14501, "loss_cls_dn_0": 0.13788, "loss_box_dn_0": 0.7242, "loss_cls_dn_1": 0.11225, "loss_box_dn_1": 0.69254, "loss_cls_dn_2": 0.10984, "loss_box_dn_2": 0.68636, "loss_cls_dn_3": 0.11014, "loss_box_dn_3": 0.69659, "loss_cls_dn_4": 0.11149, "loss_box_dn_4": 0.71136, "loss_cls_dn_5": 0.11074, "loss_box_dn_5": 0.72427, "loss_dense_depth": 0.68029, "loss": 24.52291, "grad_norm": 38.38264, "time": 1.49521} -{"mode": "train", "epoch": 1, "iter": 317, "lr": 0.00023, "memory": 49163, "data_time": 0.07926, "loss_cls_0": 0.74634, "loss_box_0": 1.62225, "loss_cns_0": 0.62947, "loss_yns_0": 0.14984, "loss_cls_1": 0.83122, "loss_box_1": 1.48775, "loss_cns_1": 0.65188, "loss_yns_1": 0.14706, "loss_cls_2": 0.83463, "loss_box_2": 1.47327, "loss_cns_2": 0.65425, "loss_yns_2": 0.14534, "loss_cls_3": 0.83328, "loss_box_3": 1.4679, "loss_cns_3": 0.65536, "loss_yns_3": 0.14591, "loss_cls_4": 0.8417, "loss_box_4": 1.46588, "loss_cns_4": 0.65615, "loss_yns_4": 0.14513, "loss_cls_5": 0.8467, "loss_box_5": 1.47261, "loss_cns_5": 0.65425, "loss_yns_5": 0.14638, "loss_cls_dn_0": 0.13683, "loss_box_dn_0": 0.72838, "loss_cls_dn_1": 0.11011, "loss_box_dn_1": 0.72949, "loss_cls_dn_2": 0.10983, "loss_box_dn_2": 0.72499, "loss_cls_dn_3": 0.11007, "loss_box_dn_3": 0.72883, "loss_cls_dn_4": 0.10929, "loss_box_dn_4": 0.74083, "loss_cls_dn_5": 0.10922, "loss_box_dn_5": 0.75292, "loss_dense_depth": 0.67321, "loss": 24.46856, "grad_norm": 31.35027, "time": 1.52235} -{"mode": "train", "epoch": 1, "iter": 318, "lr": 0.00023, "memory": 49163, "data_time": 0.08158, "loss_cls_0": 0.78272, "loss_box_0": 1.63562, "loss_cns_0": 0.62888, "loss_yns_0": 0.15069, "loss_cls_1": 0.82938, "loss_box_1": 1.52819, "loss_cns_1": 0.65372, "loss_yns_1": 0.1476, "loss_cls_2": 0.85691, "loss_box_2": 1.49885, "loss_cns_2": 0.65099, "loss_yns_2": 0.14583, "loss_cls_3": 0.85358, "loss_box_3": 1.47809, "loss_cns_3": 0.65003, "loss_yns_3": 0.14646, "loss_cls_4": 0.85794, "loss_box_4": 1.48179, "loss_cns_4": 0.65337, "loss_yns_4": 0.14686, "loss_cls_5": 0.86038, "loss_box_5": 1.48572, "loss_cns_5": 0.65299, "loss_yns_5": 0.14825, "loss_cls_dn_0": 0.13878, "loss_box_dn_0": 0.72239, "loss_cls_dn_1": 0.10528, "loss_box_dn_1": 0.69704, "loss_cls_dn_2": 0.10848, "loss_box_dn_2": 0.68475, "loss_cls_dn_3": 0.10697, "loss_box_dn_3": 0.68134, "loss_cls_dn_4": 0.10539, "loss_box_dn_4": 0.6854, "loss_cls_dn_5": 0.10809, "loss_box_dn_5": 0.69015, "loss_dense_depth": 0.70862, "loss": 24.46752, "grad_norm": 37.14784, "time": 1.50032} -{"mode": "train", "epoch": 1, "iter": 319, "lr": 0.00023, "memory": 49163, "data_time": 0.08095, "loss_cls_0": 0.76987, "loss_box_0": 1.61315, "loss_cns_0": 0.63033, "loss_yns_0": 0.1508, "loss_cls_1": 0.82622, "loss_box_1": 1.51707, "loss_cns_1": 0.6563, "loss_yns_1": 0.14991, "loss_cls_2": 0.8283, "loss_box_2": 1.47296, "loss_cns_2": 0.65803, "loss_yns_2": 0.14883, "loss_cls_3": 0.83257, "loss_box_3": 1.46896, "loss_cns_3": 0.65899, "loss_yns_3": 0.1495, "loss_cls_4": 0.84258, "loss_box_4": 1.46662, "loss_cns_4": 0.6604, "loss_yns_4": 0.14905, "loss_cls_5": 0.84469, "loss_box_5": 1.46902, "loss_cns_5": 0.65735, "loss_yns_5": 0.14985, "loss_cls_dn_0": 0.1341, "loss_box_dn_0": 0.72413, "loss_cls_dn_1": 0.10159, "loss_box_dn_1": 0.66597, "loss_cls_dn_2": 0.10245, "loss_box_dn_2": 0.6453, "loss_cls_dn_3": 0.10238, "loss_box_dn_3": 0.64149, "loss_cls_dn_4": 0.10361, "loss_box_dn_4": 0.64085, "loss_cls_dn_5": 0.10728, "loss_box_dn_5": 0.64398, "loss_dense_depth": 0.67944, "loss": 24.06392, "grad_norm": 25.37266, "time": 3.38142} -{"mode": "train", "epoch": 1, "iter": 320, "lr": 0.00023, "memory": 49163, "data_time": 0.08006, "loss_cls_0": 0.75859, "loss_box_0": 1.61616, "loss_cns_0": 0.62721, "loss_yns_0": 0.14834, "loss_cls_1": 0.83965, "loss_box_1": 1.51603, "loss_cns_1": 0.66029, "loss_yns_1": 0.14908, "loss_cls_2": 0.84854, "loss_box_2": 1.48261, "loss_cns_2": 0.66201, "loss_yns_2": 0.14801, "loss_cls_3": 0.83985, "loss_box_3": 1.47404, "loss_cns_3": 0.66245, "loss_yns_3": 0.14823, "loss_cls_4": 0.8454, "loss_box_4": 1.47152, "loss_cns_4": 0.66339, "loss_yns_4": 0.14809, "loss_cls_5": 0.84513, "loss_box_5": 1.4732, "loss_cns_5": 0.66193, "loss_yns_5": 0.14729, "loss_cls_dn_0": 0.13154, "loss_box_dn_0": 0.71807, "loss_cls_dn_1": 0.10502, "loss_box_dn_1": 0.64788, "loss_cls_dn_2": 0.10293, "loss_box_dn_2": 0.62941, "loss_cls_dn_3": 0.10254, "loss_box_dn_3": 0.62616, "loss_cls_dn_4": 0.10645, "loss_box_dn_4": 0.62694, "loss_cls_dn_5": 0.10692, "loss_box_dn_5": 0.62888, "loss_dense_depth": 0.67033, "loss": 24.04011, "grad_norm": 26.87072, "time": 1.48284} -{"mode": "train", "epoch": 1, "iter": 321, "lr": 0.00023, "memory": 49163, "data_time": 0.07001, "loss_cls_0": 0.74412, "loss_box_0": 1.58814, "loss_cns_0": 0.62754, "loss_yns_0": 0.14886, "loss_cls_1": 0.80276, "loss_box_1": 1.49947, "loss_cns_1": 0.65946, "loss_yns_1": 0.14984, "loss_cls_2": 0.81298, "loss_box_2": 1.46795, "loss_cns_2": 0.66073, "loss_yns_2": 0.14981, "loss_cls_3": 0.80981, "loss_box_3": 1.46163, "loss_cns_3": 0.6621, "loss_yns_3": 0.14929, "loss_cls_4": 0.8126, "loss_box_4": 1.45556, "loss_cns_4": 0.66155, "loss_yns_4": 0.14965, "loss_cls_5": 0.81885, "loss_box_5": 1.45091, "loss_cns_5": 0.66057, "loss_yns_5": 0.14981, "loss_cls_dn_0": 0.12963, "loss_box_dn_0": 0.71678, "loss_cls_dn_1": 0.1043, "loss_box_dn_1": 0.6424, "loss_cls_dn_2": 0.10159, "loss_box_dn_2": 0.62985, "loss_cls_dn_3": 0.10152, "loss_box_dn_3": 0.62797, "loss_cls_dn_4": 0.1044, "loss_box_dn_4": 0.62902, "loss_cls_dn_5": 0.10598, "loss_box_dn_5": 0.63225, "loss_dense_depth": 0.64026, "loss": 23.71993, "grad_norm": 26.90475, "time": 1.52829} -{"mode": "train", "epoch": 1, "iter": 322, "lr": 0.00023, "memory": 49163, "data_time": 0.07869, "loss_cls_0": 0.75459, "loss_box_0": 1.61132, "loss_cns_0": 0.62844, "loss_yns_0": 0.14955, "loss_cls_1": 0.8169, "loss_box_1": 1.47937, "loss_cns_1": 0.65927, "loss_yns_1": 0.14834, "loss_cls_2": 0.8255, "loss_box_2": 1.45029, "loss_cns_2": 0.66039, "loss_yns_2": 0.14924, "loss_cls_3": 0.83402, "loss_box_3": 1.44537, "loss_cns_3": 0.66047, "loss_yns_3": 0.14966, "loss_cls_4": 0.83697, "loss_box_4": 1.4305, "loss_cns_4": 0.65689, "loss_yns_4": 0.14845, "loss_cls_5": 0.83387, "loss_box_5": 1.43584, "loss_cns_5": 0.65976, "loss_yns_5": 0.14896, "loss_cls_dn_0": 0.12997, "loss_box_dn_0": 0.72134, "loss_cls_dn_1": 0.10374, "loss_box_dn_1": 0.65786, "loss_cls_dn_2": 0.10079, "loss_box_dn_2": 0.64592, "loss_cls_dn_3": 0.10037, "loss_box_dn_3": 0.64889, "loss_cls_dn_4": 0.102, "loss_box_dn_4": 0.65479, "loss_cls_dn_5": 0.10313, "loss_box_dn_5": 0.66305, "loss_dense_depth": 0.65835, "loss": 23.86416, "grad_norm": 29.98009, "time": 3.42802} -{"mode": "train", "epoch": 1, "iter": 323, "lr": 0.00023, "memory": 49163, "data_time": 0.07966, "loss_cls_0": 0.75327, "loss_box_0": 1.62212, "loss_cns_0": 0.63295, "loss_yns_0": 0.14726, "loss_cls_1": 0.80998, "loss_box_1": 1.53055, "loss_cns_1": 0.65713, "loss_yns_1": 0.14704, "loss_cls_2": 0.83401, "loss_box_2": 1.5074, "loss_cns_2": 0.6562, "loss_yns_2": 0.14816, "loss_cls_3": 0.82044, "loss_box_3": 1.50503, "loss_cns_3": 0.65828, "loss_yns_3": 0.14729, "loss_cls_4": 0.82958, "loss_box_4": 1.4957, "loss_cns_4": 0.65382, "loss_yns_4": 0.1477, "loss_cls_5": 0.83037, "loss_box_5": 1.51084, "loss_cns_5": 0.65623, "loss_yns_5": 0.14841, "loss_cls_dn_0": 0.13076, "loss_box_dn_0": 0.72271, "loss_cls_dn_1": 0.10151, "loss_box_dn_1": 0.68202, "loss_cls_dn_2": 0.10155, "loss_box_dn_2": 0.67351, "loss_cls_dn_3": 0.10053, "loss_box_dn_3": 0.68342, "loss_cls_dn_4": 0.10504, "loss_box_dn_4": 0.69695, "loss_cls_dn_5": 0.10414, "loss_box_dn_5": 0.71263, "loss_dense_depth": 0.65031, "loss": 24.31483, "grad_norm": 44.96894, "time": 3.30973} -{"mode": "train", "epoch": 1, "iter": 324, "lr": 0.00023, "memory": 49163, "data_time": 0.07644, "loss_cls_0": 0.76134, "loss_box_0": 1.63965, "loss_cns_0": 0.63256, "loss_yns_0": 0.14645, "loss_cls_1": 0.80937, "loss_box_1": 1.54414, "loss_cns_1": 0.6569, "loss_yns_1": 0.14599, "loss_cls_2": 0.8312, "loss_box_2": 1.50739, "loss_cns_2": 0.6575, "loss_yns_2": 0.14578, "loss_cls_3": 0.8293, "loss_box_3": 1.49951, "loss_cns_3": 0.66154, "loss_yns_3": 0.14625, "loss_cls_4": 0.83619, "loss_box_4": 1.49567, "loss_cns_4": 0.65846, "loss_yns_4": 0.14625, "loss_cls_5": 0.83691, "loss_box_5": 1.50709, "loss_cns_5": 0.6608, "loss_yns_5": 0.14794, "loss_cls_dn_0": 0.12811, "loss_box_dn_0": 0.71866, "loss_cls_dn_1": 0.10088, "loss_box_dn_1": 0.6894, "loss_cls_dn_2": 0.10123, "loss_box_dn_2": 0.67848, "loss_cls_dn_3": 0.10083, "loss_box_dn_3": 0.68657, "loss_cls_dn_4": 0.10473, "loss_box_dn_4": 0.70062, "loss_cls_dn_5": 0.10466, "loss_box_dn_5": 0.71575, "loss_dense_depth": 0.6611, "loss": 24.3952, "grad_norm": 38.55319, "time": 1.53994} -{"mode": "train", "epoch": 1, "iter": 325, "lr": 0.00023, "memory": 49163, "data_time": 0.07382, "loss_cls_0": 0.72345, "loss_box_0": 1.63465, "loss_cns_0": 0.6321, "loss_yns_0": 0.14634, "loss_cls_1": 0.79941, "loss_box_1": 1.52889, "loss_cns_1": 0.6575, "loss_yns_1": 0.14551, "loss_cls_2": 0.82631, "loss_box_2": 1.49725, "loss_cns_2": 0.65651, "loss_yns_2": 0.14613, "loss_cls_3": 0.80983, "loss_box_3": 1.5076, "loss_cns_3": 0.66148, "loss_yns_3": 0.14548, "loss_cls_4": 0.81497, "loss_box_4": 1.50222, "loss_cns_4": 0.65507, "loss_yns_4": 0.14468, "loss_cls_5": 0.81174, "loss_box_5": 1.51139, "loss_cns_5": 0.65776, "loss_yns_5": 0.14603, "loss_cls_dn_0": 0.12697, "loss_box_dn_0": 0.7203, "loss_cls_dn_1": 0.10269, "loss_box_dn_1": 0.68823, "loss_cls_dn_2": 0.1024, "loss_box_dn_2": 0.68642, "loss_cls_dn_3": 0.10165, "loss_box_dn_3": 0.69368, "loss_cls_dn_4": 0.10385, "loss_box_dn_4": 0.70402, "loss_cls_dn_5": 0.10445, "loss_box_dn_5": 0.71321, "loss_dense_depth": 0.67186, "loss": 24.28202, "grad_norm": 43.40701, "time": 1.48777} -{"mode": "train", "epoch": 1, "iter": 326, "lr": 0.00023, "memory": 49163, "data_time": 0.08143, "loss_cls_0": 0.7224, "loss_box_0": 1.63499, "loss_cns_0": 0.63175, "loss_yns_0": 0.14487, "loss_cls_1": 0.79963, "loss_box_1": 1.53055, "loss_cns_1": 0.65939, "loss_yns_1": 0.14527, "loss_cls_2": 0.8259, "loss_box_2": 1.48973, "loss_cns_2": 0.65996, "loss_yns_2": 0.14529, "loss_cls_3": 0.80597, "loss_box_3": 1.49795, "loss_cns_3": 0.66329, "loss_yns_3": 0.14581, "loss_cls_4": 0.8024, "loss_box_4": 1.49793, "loss_cns_4": 0.66159, "loss_yns_4": 0.14619, "loss_cls_5": 0.80056, "loss_box_5": 1.49721, "loss_cns_5": 0.66034, "loss_yns_5": 0.14547, "loss_cls_dn_0": 0.12373, "loss_box_dn_0": 0.7146, "loss_cls_dn_1": 0.09974, "loss_box_dn_1": 0.67431, "loss_cls_dn_2": 0.10149, "loss_box_dn_2": 0.66224, "loss_cls_dn_3": 0.09865, "loss_box_dn_3": 0.66242, "loss_cls_dn_4": 0.10075, "loss_box_dn_4": 0.66532, "loss_cls_dn_5": 0.101, "loss_box_dn_5": 0.669, "loss_dense_depth": 0.66175, "loss": 24.04942, "grad_norm": 39.65895, "time": 1.50284} -{"mode": "train", "epoch": 1, "iter": 327, "lr": 0.00023, "memory": 49163, "data_time": 0.08438, "loss_cls_0": 0.73143, "loss_box_0": 1.65161, "loss_cns_0": 0.63193, "loss_yns_0": 0.14329, "loss_cls_1": 0.80227, "loss_box_1": 1.53894, "loss_cns_1": 0.65905, "loss_yns_1": 0.14257, "loss_cls_2": 0.81315, "loss_box_2": 1.50444, "loss_cns_2": 0.65921, "loss_yns_2": 0.14204, "loss_cls_3": 0.81566, "loss_box_3": 1.50024, "loss_cns_3": 0.66003, "loss_yns_3": 0.14183, "loss_cls_4": 0.81791, "loss_box_4": 1.5022, "loss_cns_4": 0.6605, "loss_yns_4": 0.14193, "loss_cls_5": 0.82013, "loss_box_5": 1.50686, "loss_cns_5": 0.66198, "loss_yns_5": 0.14211, "loss_cls_dn_0": 0.12964, "loss_box_dn_0": 0.71417, "loss_cls_dn_1": 0.10222, "loss_box_dn_1": 0.64831, "loss_cls_dn_2": 0.10456, "loss_box_dn_2": 0.63041, "loss_cls_dn_3": 0.10112, "loss_box_dn_3": 0.6275, "loss_cls_dn_4": 0.10398, "loss_box_dn_4": 0.62924, "loss_cls_dn_5": 0.10492, "loss_box_dn_5": 0.63126, "loss_dense_depth": 0.67214, "loss": 23.99077, "grad_norm": 40.11806, "time": 1.51159} -{"mode": "train", "epoch": 1, "iter": 328, "lr": 0.00023, "memory": 49163, "data_time": 0.08452, "loss_cls_0": 0.74419, "loss_box_0": 1.66017, "loss_cns_0": 0.63694, "loss_yns_0": 0.14601, "loss_cls_1": 0.81786, "loss_box_1": 1.53512, "loss_cns_1": 0.66181, "loss_yns_1": 0.1439, "loss_cls_2": 0.82142, "loss_box_2": 1.50829, "loss_cns_2": 0.66123, "loss_yns_2": 0.14444, "loss_cls_3": 0.82563, "loss_box_3": 1.49951, "loss_cns_3": 0.6613, "loss_yns_3": 0.14484, "loss_cls_4": 0.82433, "loss_box_4": 1.50025, "loss_cns_4": 0.66174, "loss_yns_4": 0.14504, "loss_cls_5": 0.82004, "loss_box_5": 1.49885, "loss_cns_5": 0.66099, "loss_yns_5": 0.14583, "loss_cls_dn_0": 0.12668, "loss_box_dn_0": 0.72124, "loss_cls_dn_1": 0.10053, "loss_box_dn_1": 0.63753, "loss_cls_dn_2": 0.1019, "loss_box_dn_2": 0.62317, "loss_cls_dn_3": 0.09935, "loss_box_dn_3": 0.62152, "loss_cls_dn_4": 0.10236, "loss_box_dn_4": 0.62351, "loss_cls_dn_5": 0.10285, "loss_box_dn_5": 0.62654, "loss_dense_depth": 0.68028, "loss": 24.03719, "grad_norm": 35.53638, "time": 1.49285} -{"mode": "train", "epoch": 1, "iter": 329, "lr": 0.00023, "memory": 49163, "data_time": 0.17388, "loss_cls_0": 0.73996, "loss_box_0": 1.66201, "loss_cns_0": 0.63309, "loss_yns_0": 0.14497, "loss_cls_1": 0.80632, "loss_box_1": 1.57862, "loss_cns_1": 0.65948, "loss_yns_1": 0.14344, "loss_cls_2": 0.81215, "loss_box_2": 1.53858, "loss_cns_2": 0.6595, "loss_yns_2": 0.1436, "loss_cls_3": 0.81676, "loss_box_3": 1.52497, "loss_cns_3": 0.65991, "loss_yns_3": 0.14289, "loss_cls_4": 0.82218, "loss_box_4": 1.52202, "loss_cns_4": 0.66033, "loss_yns_4": 0.14324, "loss_cls_5": 0.8202, "loss_box_5": 1.5267, "loss_cns_5": 0.65891, "loss_yns_5": 0.14367, "loss_cls_dn_0": 0.13147, "loss_box_dn_0": 0.72344, "loss_cls_dn_1": 0.1033, "loss_box_dn_1": 0.64523, "loss_cls_dn_2": 0.10274, "loss_box_dn_2": 0.63047, "loss_cls_dn_3": 0.10289, "loss_box_dn_3": 0.6266, "loss_cls_dn_4": 0.10583, "loss_box_dn_4": 0.62824, "loss_cls_dn_5": 0.10572, "loss_box_dn_5": 0.63471, "loss_dense_depth": 0.69903, "loss": 24.20314, "grad_norm": 39.88502, "time": 1.56573} -{"mode": "train", "epoch": 1, "iter": 330, "lr": 0.00023, "memory": 49163, "data_time": 0.07963, "loss_cls_0": 0.72903, "loss_box_0": 1.62994, "loss_cns_0": 0.63397, "loss_yns_0": 0.14771, "loss_cls_1": 0.79317, "loss_box_1": 1.5382, "loss_cns_1": 0.66221, "loss_yns_1": 0.14432, "loss_cls_2": 0.80728, "loss_box_2": 1.4955, "loss_cns_2": 0.66167, "loss_yns_2": 0.14472, "loss_cls_3": 0.80947, "loss_box_3": 1.48591, "loss_cns_3": 0.66329, "loss_yns_3": 0.14418, "loss_cls_4": 0.81758, "loss_box_4": 1.48462, "loss_cns_4": 0.66318, "loss_yns_4": 0.14454, "loss_cls_5": 0.81413, "loss_box_5": 1.48709, "loss_cns_5": 0.66222, "loss_yns_5": 0.14408, "loss_cls_dn_0": 0.13354, "loss_box_dn_0": 0.72694, "loss_cls_dn_1": 0.10184, "loss_box_dn_1": 0.65446, "loss_cls_dn_2": 0.09946, "loss_box_dn_2": 0.6377, "loss_cls_dn_3": 0.10027, "loss_box_dn_3": 0.63764, "loss_cls_dn_4": 0.10317, "loss_box_dn_4": 0.64256, "loss_cls_dn_5": 0.1024, "loss_box_dn_5": 0.64947, "loss_dense_depth": 0.68352, "loss": 23.98098, "grad_norm": 34.47762, "time": 1.4932} -{"mode": "train", "epoch": 1, "iter": 331, "lr": 0.00023, "memory": 49163, "data_time": 0.0735, "loss_cls_0": 0.73456, "loss_box_0": 1.63094, "loss_cns_0": 0.62598, "loss_yns_0": 0.14394, "loss_cls_1": 0.79838, "loss_box_1": 1.52962, "loss_cns_1": 0.65697, "loss_yns_1": 0.14373, "loss_cls_2": 0.81043, "loss_box_2": 1.50653, "loss_cns_2": 0.65885, "loss_yns_2": 0.14312, "loss_cls_3": 0.80888, "loss_box_3": 1.50856, "loss_cns_3": 0.66003, "loss_yns_3": 0.14299, "loss_cls_4": 0.81563, "loss_box_4": 1.51116, "loss_cns_4": 0.66071, "loss_yns_4": 0.14373, "loss_cls_5": 0.81148, "loss_box_5": 1.51037, "loss_cns_5": 0.6594, "loss_yns_5": 0.1426, "loss_cls_dn_0": 0.1314, "loss_box_dn_0": 0.71741, "loss_cls_dn_1": 0.10287, "loss_box_dn_1": 0.65261, "loss_cls_dn_2": 0.10004, "loss_box_dn_2": 0.63991, "loss_cls_dn_3": 0.10118, "loss_box_dn_3": 0.64586, "loss_cls_dn_4": 0.10349, "loss_box_dn_4": 0.65385, "loss_cls_dn_5": 0.10457, "loss_box_dn_5": 0.65833, "loss_dense_depth": 0.68907, "loss": 24.05917, "grad_norm": 35.30116, "time": 1.51065} -{"mode": "train", "epoch": 1, "iter": 332, "lr": 0.00023, "memory": 49163, "data_time": 0.08091, "loss_cls_0": 0.72817, "loss_box_0": 1.6313, "loss_cns_0": 0.63278, "loss_yns_0": 0.14666, "loss_cls_1": 0.79331, "loss_box_1": 1.50908, "loss_cns_1": 0.66075, "loss_yns_1": 0.14376, "loss_cls_2": 0.80797, "loss_box_2": 1.47823, "loss_cns_2": 0.66366, "loss_yns_2": 0.14398, "loss_cls_3": 0.80796, "loss_box_3": 1.48021, "loss_cns_3": 0.66397, "loss_yns_3": 0.14422, "loss_cls_4": 0.80952, "loss_box_4": 1.48345, "loss_cns_4": 0.66432, "loss_yns_4": 0.14439, "loss_cls_5": 0.81612, "loss_box_5": 1.4826, "loss_cns_5": 0.66209, "loss_yns_5": 0.14427, "loss_cls_dn_0": 0.13117, "loss_box_dn_0": 0.71933, "loss_cls_dn_1": 0.10467, "loss_box_dn_1": 0.66417, "loss_cls_dn_2": 0.1014, "loss_box_dn_2": 0.65167, "loss_cls_dn_3": 0.10192, "loss_box_dn_3": 0.65469, "loss_cls_dn_4": 0.1035, "loss_box_dn_4": 0.66176, "loss_cls_dn_5": 0.10611, "loss_box_dn_5": 0.66843, "loss_dense_depth": 0.67502, "loss": 23.98661, "grad_norm": 38.74337, "time": 1.49836} -{"mode": "train", "epoch": 1, "iter": 333, "lr": 0.00023, "memory": 49163, "data_time": 0.07843, "loss_cls_0": 0.75804, "loss_box_0": 1.6501, "loss_cns_0": 0.62929, "loss_yns_0": 0.14468, "loss_cls_1": 0.81379, "loss_box_1": 1.50221, "loss_cns_1": 0.65728, "loss_yns_1": 0.14263, "loss_cls_2": 0.81674, "loss_box_2": 1.47468, "loss_cns_2": 0.65926, "loss_yns_2": 0.14138, "loss_cls_3": 0.82364, "loss_box_3": 1.45423, "loss_cns_3": 0.65996, "loss_yns_3": 0.14219, "loss_cls_4": 0.83016, "loss_box_4": 1.46397, "loss_cns_4": 0.66374, "loss_yns_4": 0.14134, "loss_cls_5": 0.84111, "loss_box_5": 1.45683, "loss_cns_5": 0.66035, "loss_yns_5": 0.1411, "loss_cls_dn_0": 0.13878, "loss_box_dn_0": 0.72705, "loss_cls_dn_1": 0.10298, "loss_box_dn_1": 0.66834, "loss_cls_dn_2": 0.10017, "loss_box_dn_2": 0.65593, "loss_cls_dn_3": 0.10061, "loss_box_dn_3": 0.65204, "loss_cls_dn_4": 0.10288, "loss_box_dn_4": 0.65374, "loss_cls_dn_5": 0.10529, "loss_box_dn_5": 0.65925, "loss_dense_depth": 0.72229, "loss": 24.05806, "grad_norm": 31.42604, "time": 1.48766} -{"mode": "train", "epoch": 1, "iter": 334, "lr": 0.00023, "memory": 49163, "data_time": 0.07942, "loss_cls_0": 0.76314, "loss_box_0": 1.6385, "loss_cns_0": 0.6286, "loss_yns_0": 0.14571, "loss_cls_1": 0.82426, "loss_box_1": 1.51598, "loss_cns_1": 0.658, "loss_yns_1": 0.14353, "loss_cls_2": 0.83464, "loss_box_2": 1.48518, "loss_cns_2": 0.65641, "loss_yns_2": 0.14256, "loss_cls_3": 0.8344, "loss_box_3": 1.47484, "loss_cns_3": 0.65802, "loss_yns_3": 0.14225, "loss_cls_4": 0.83964, "loss_box_4": 1.47462, "loss_cns_4": 0.65838, "loss_yns_4": 0.14148, "loss_cls_5": 0.83996, "loss_box_5": 1.47365, "loss_cns_5": 0.65806, "loss_yns_5": 0.14259, "loss_cls_dn_0": 0.13471, "loss_box_dn_0": 0.71968, "loss_cls_dn_1": 0.10497, "loss_box_dn_1": 0.67149, "loss_cls_dn_2": 0.1043, "loss_box_dn_2": 0.65801, "loss_cls_dn_3": 0.10496, "loss_box_dn_3": 0.6544, "loss_cls_dn_4": 0.10583, "loss_box_dn_4": 0.65311, "loss_cls_dn_5": 0.10827, "loss_box_dn_5": 0.65475, "loss_dense_depth": 0.70445, "loss": 24.15331, "grad_norm": 32.88414, "time": 1.55208} -{"mode": "train", "epoch": 1, "iter": 335, "lr": 0.00023, "memory": 49163, "data_time": 0.08127, "loss_cls_0": 0.75911, "loss_box_0": 1.63775, "loss_cns_0": 0.63027, "loss_yns_0": 0.14777, "loss_cls_1": 0.81441, "loss_box_1": 1.53268, "loss_cns_1": 0.65374, "loss_yns_1": 0.14285, "loss_cls_2": 0.8229, "loss_box_2": 1.49593, "loss_cns_2": 0.65365, "loss_yns_2": 0.14267, "loss_cls_3": 0.8164, "loss_box_3": 1.49408, "loss_cns_3": 0.65721, "loss_yns_3": 0.1428, "loss_cls_4": 0.82434, "loss_box_4": 1.49624, "loss_cns_4": 0.65728, "loss_yns_4": 0.14297, "loss_cls_5": 0.82628, "loss_box_5": 1.5022, "loss_cns_5": 0.6571, "loss_yns_5": 0.14324, "loss_cls_dn_0": 0.13625, "loss_box_dn_0": 0.72026, "loss_cls_dn_1": 0.10443, "loss_box_dn_1": 0.66815, "loss_cls_dn_2": 0.10418, "loss_box_dn_2": 0.65105, "loss_cls_dn_3": 0.1026, "loss_box_dn_3": 0.64547, "loss_cls_dn_4": 0.10341, "loss_box_dn_4": 0.64468, "loss_cls_dn_5": 0.10459, "loss_box_dn_5": 0.64551, "loss_dense_depth": 0.69699, "loss": 24.12144, "grad_norm": 28.00086, "time": 1.51115} -{"mode": "train", "epoch": 1, "iter": 336, "lr": 0.00023, "memory": 49163, "data_time": 0.07836, "loss_cls_0": 0.75765, "loss_box_0": 1.63645, "loss_cns_0": 0.62591, "loss_yns_0": 0.14562, "loss_cls_1": 0.8232, "loss_box_1": 1.5681, "loss_cns_1": 0.65347, "loss_yns_1": 0.14246, "loss_cls_2": 0.8228, "loss_box_2": 1.5402, "loss_cns_2": 0.6541, "loss_yns_2": 0.14305, "loss_cls_3": 0.82361, "loss_box_3": 1.53324, "loss_cns_3": 0.65674, "loss_yns_3": 0.14139, "loss_cls_4": 0.82646, "loss_box_4": 1.53635, "loss_cns_4": 0.65735, "loss_yns_4": 0.1422, "loss_cls_5": 0.83646, "loss_box_5": 1.53692, "loss_cns_5": 0.65557, "loss_yns_5": 0.14295, "loss_cls_dn_0": 0.13756, "loss_box_dn_0": 0.71881, "loss_cls_dn_1": 0.10257, "loss_box_dn_1": 0.66373, "loss_cls_dn_2": 0.10174, "loss_box_dn_2": 0.64775, "loss_cls_dn_3": 0.1013, "loss_box_dn_3": 0.64354, "loss_cls_dn_4": 0.10244, "loss_box_dn_4": 0.6441, "loss_cls_dn_5": 0.10378, "loss_box_dn_5": 0.64666, "loss_dense_depth": 0.68651, "loss": 24.30275, "grad_norm": 29.07896, "time": 1.4854} -{"mode": "train", "epoch": 1, "iter": 337, "lr": 0.00023, "memory": 49163, "data_time": 0.07706, "loss_cls_0": 0.75533, "loss_box_0": 1.63583, "loss_cns_0": 0.62791, "loss_yns_0": 0.14498, "loss_cls_1": 0.82466, "loss_box_1": 1.54888, "loss_cns_1": 0.65814, "loss_yns_1": 0.14256, "loss_cls_2": 0.82419, "loss_box_2": 1.52508, "loss_cns_2": 0.65459, "loss_yns_2": 0.14147, "loss_cls_3": 0.83317, "loss_box_3": 1.51648, "loss_cns_3": 0.65513, "loss_yns_3": 0.14116, "loss_cls_4": 0.83049, "loss_box_4": 1.51782, "loss_cns_4": 0.65739, "loss_yns_4": 0.14079, "loss_cls_5": 0.8361, "loss_box_5": 1.5261, "loss_cns_5": 0.65674, "loss_yns_5": 0.14027, "loss_cls_dn_0": 0.13487, "loss_box_dn_0": 0.71508, "loss_cls_dn_1": 0.10632, "loss_box_dn_1": 0.64974, "loss_cls_dn_2": 0.10554, "loss_box_dn_2": 0.63746, "loss_cls_dn_3": 0.10577, "loss_box_dn_3": 0.63735, "loss_cls_dn_4": 0.10724, "loss_box_dn_4": 0.64144, "loss_cls_dn_5": 0.1091, "loss_box_dn_5": 0.65231, "loss_dense_depth": 0.68933, "loss": 24.22682, "grad_norm": 36.37205, "time": 1.54501} -{"mode": "train", "epoch": 1, "iter": 338, "lr": 0.00023, "memory": 49163, "data_time": 0.07716, "loss_cls_0": 0.7523, "loss_box_0": 1.62243, "loss_cns_0": 0.63443, "loss_yns_0": 0.14288, "loss_cls_1": 0.82309, "loss_box_1": 1.51598, "loss_cns_1": 0.65931, "loss_yns_1": 0.13974, "loss_cls_2": 0.8278, "loss_box_2": 1.50869, "loss_cns_2": 0.65949, "loss_yns_2": 0.13993, "loss_cls_3": 0.82973, "loss_box_3": 1.49341, "loss_cns_3": 0.66012, "loss_yns_3": 0.13772, "loss_cls_4": 0.82989, "loss_box_4": 1.49175, "loss_cns_4": 0.66009, "loss_yns_4": 0.13854, "loss_cls_5": 0.83215, "loss_box_5": 1.49365, "loss_cns_5": 0.6595, "loss_yns_5": 0.13902, "loss_cls_dn_0": 0.13312, "loss_box_dn_0": 0.72013, "loss_cls_dn_1": 0.10671, "loss_box_dn_1": 0.65097, "loss_cls_dn_2": 0.10588, "loss_box_dn_2": 0.64353, "loss_cls_dn_3": 0.10523, "loss_box_dn_3": 0.64438, "loss_cls_dn_4": 0.10654, "loss_box_dn_4": 0.64782, "loss_cls_dn_5": 0.10847, "loss_box_dn_5": 0.65851, "loss_dense_depth": 0.69049, "loss": 24.11343, "grad_norm": 31.86345, "time": 1.48156} -{"mode": "train", "epoch": 1, "iter": 339, "lr": 0.00023, "memory": 49163, "data_time": 0.07157, "loss_cls_0": 0.74311, "loss_box_0": 1.64076, "loss_cns_0": 0.62998, "loss_yns_0": 0.14414, "loss_cls_1": 0.83067, "loss_box_1": 1.51253, "loss_cns_1": 0.66071, "loss_yns_1": 0.14186, "loss_cls_2": 0.83474, "loss_box_2": 1.50077, "loss_cns_2": 0.66136, "loss_yns_2": 0.1404, "loss_cls_3": 0.83191, "loss_box_3": 1.47636, "loss_cns_3": 0.66027, "loss_yns_3": 0.13767, "loss_cls_4": 0.83695, "loss_box_4": 1.47873, "loss_cns_4": 0.65895, "loss_yns_4": 0.13839, "loss_cls_5": 0.83842, "loss_box_5": 1.47895, "loss_cns_5": 0.65953, "loss_yns_5": 0.13794, "loss_cls_dn_0": 0.13729, "loss_box_dn_0": 0.71902, "loss_cls_dn_1": 0.10589, "loss_box_dn_1": 0.65724, "loss_cls_dn_2": 0.10548, "loss_box_dn_2": 0.64845, "loss_cls_dn_3": 0.10374, "loss_box_dn_3": 0.64522, "loss_cls_dn_4": 0.10576, "loss_box_dn_4": 0.64555, "loss_cls_dn_5": 0.10732, "loss_box_dn_5": 0.64907, "loss_dense_depth": 0.70089, "loss": 24.10602, "grad_norm": 26.6232, "time": 1.47641} -{"mode": "train", "epoch": 1, "iter": 340, "lr": 0.00024, "memory": 49163, "data_time": 0.07896, "loss_cls_0": 0.76218, "loss_box_0": 1.64368, "loss_cns_0": 0.63196, "loss_yns_0": 0.14699, "loss_cls_1": 0.82483, "loss_box_1": 1.52825, "loss_cns_1": 0.65974, "loss_yns_1": 0.14316, "loss_cls_2": 0.83162, "loss_box_2": 1.49383, "loss_cns_2": 0.66311, "loss_yns_2": 0.14429, "loss_cls_3": 0.83744, "loss_box_3": 1.48195, "loss_cns_3": 0.66379, "loss_yns_3": 0.14278, "loss_cls_4": 0.84148, "loss_box_4": 1.47742, "loss_cns_4": 0.66464, "loss_yns_4": 0.14282, "loss_cls_5": 0.84385, "loss_box_5": 1.48312, "loss_cns_5": 0.66453, "loss_yns_5": 0.14185, "loss_cls_dn_0": 0.13689, "loss_box_dn_0": 0.72155, "loss_cls_dn_1": 0.10778, "loss_box_dn_1": 0.66959, "loss_cls_dn_2": 0.10728, "loss_box_dn_2": 0.65448, "loss_cls_dn_3": 0.10814, "loss_box_dn_3": 0.65104, "loss_cls_dn_4": 0.1085, "loss_box_dn_4": 0.64715, "loss_cls_dn_5": 0.10973, "loss_box_dn_5": 0.64954, "loss_dense_depth": 0.70421, "loss": 24.23519, "grad_norm": 31.92095, "time": 1.49266} -{"mode": "train", "epoch": 1, "iter": 341, "lr": 0.00024, "memory": 49163, "data_time": 0.07803, "loss_cls_0": 0.77556, "loss_box_0": 1.64868, "loss_cns_0": 0.63233, "loss_yns_0": 0.1474, "loss_cls_1": 0.82549, "loss_box_1": 1.59039, "loss_cns_1": 0.65501, "loss_yns_1": 0.14487, "loss_cls_2": 0.83005, "loss_box_2": 1.55714, "loss_cns_2": 0.65587, "loss_yns_2": 0.14381, "loss_cls_3": 0.83563, "loss_box_3": 1.54826, "loss_cns_3": 0.65484, "loss_yns_3": 0.14319, "loss_cls_4": 0.83642, "loss_box_4": 1.54111, "loss_cns_4": 0.65552, "loss_yns_4": 0.14311, "loss_cls_5": 0.84576, "loss_box_5": 1.54403, "loss_cns_5": 0.65659, "loss_yns_5": 0.14383, "loss_cls_dn_0": 0.13067, "loss_box_dn_0": 0.71932, "loss_cls_dn_1": 0.10462, "loss_box_dn_1": 0.6604, "loss_cls_dn_2": 0.10451, "loss_box_dn_2": 0.646, "loss_cls_dn_3": 0.105, "loss_box_dn_3": 0.64409, "loss_cls_dn_4": 0.10594, "loss_box_dn_4": 0.64138, "loss_cls_dn_5": 0.10845, "loss_box_dn_5": 0.6431, "loss_dense_depth": 0.71871, "loss": 24.48709, "grad_norm": 30.14641, "time": 1.52236} -{"mode": "train", "epoch": 1, "iter": 342, "lr": 0.00024, "memory": 49163, "data_time": 0.07541, "loss_cls_0": 0.74636, "loss_box_0": 1.62068, "loss_cns_0": 0.62828, "loss_yns_0": 0.14563, "loss_cls_1": 0.80859, "loss_box_1": 1.54455, "loss_cns_1": 0.65249, "loss_yns_1": 0.14476, "loss_cls_2": 0.81924, "loss_box_2": 1.52248, "loss_cns_2": 0.65239, "loss_yns_2": 0.14311, "loss_cls_3": 0.81765, "loss_box_3": 1.52609, "loss_cns_3": 0.65942, "loss_yns_3": 0.1428, "loss_cls_4": 0.82075, "loss_box_4": 1.52462, "loss_cns_4": 0.65426, "loss_yns_4": 0.14245, "loss_cls_5": 0.82608, "loss_box_5": 1.52159, "loss_cns_5": 0.65667, "loss_yns_5": 0.14295, "loss_cls_dn_0": 0.1267, "loss_box_dn_0": 0.72065, "loss_cls_dn_1": 0.10309, "loss_box_dn_1": 0.64481, "loss_cls_dn_2": 0.10147, "loss_box_dn_2": 0.63386, "loss_cls_dn_3": 0.10049, "loss_box_dn_3": 0.63102, "loss_cls_dn_4": 0.10097, "loss_box_dn_4": 0.63101, "loss_cls_dn_5": 0.10368, "loss_box_dn_5": 0.62973, "loss_dense_depth": 0.70679, "loss": 24.09816, "grad_norm": 30.68948, "time": 1.57735} -{"mode": "train", "epoch": 1, "iter": 343, "lr": 0.00024, "memory": 49163, "data_time": 0.07836, "loss_cls_0": 0.76201, "loss_box_0": 1.6403, "loss_cns_0": 0.62562, "loss_yns_0": 0.14618, "loss_cls_1": 0.82248, "loss_box_1": 1.58479, "loss_cns_1": 0.65056, "loss_yns_1": 0.14398, "loss_cls_2": 0.83755, "loss_box_2": 1.55011, "loss_cns_2": 0.65092, "loss_yns_2": 0.14176, "loss_cls_3": 0.83523, "loss_box_3": 1.55083, "loss_cns_3": 0.6561, "loss_yns_3": 0.14269, "loss_cls_4": 0.83807, "loss_box_4": 1.55318, "loss_cns_4": 0.65424, "loss_yns_4": 0.14255, "loss_cls_5": 0.84606, "loss_box_5": 1.54773, "loss_cns_5": 0.65444, "loss_yns_5": 0.14292, "loss_cls_dn_0": 0.12922, "loss_box_dn_0": 0.7108, "loss_cls_dn_1": 0.10127, "loss_box_dn_1": 0.64438, "loss_cls_dn_2": 0.10031, "loss_box_dn_2": 0.63385, "loss_cls_dn_3": 0.10016, "loss_box_dn_3": 0.63148, "loss_cls_dn_4": 0.10194, "loss_box_dn_4": 0.63324, "loss_cls_dn_5": 0.10534, "loss_box_dn_5": 0.63347, "loss_dense_depth": 0.69962, "loss": 24.34541, "grad_norm": 32.76817, "time": 1.49458} -{"mode": "train", "epoch": 1, "iter": 344, "lr": 0.00024, "memory": 49163, "data_time": 0.08059, "loss_cls_0": 0.7757, "loss_box_0": 1.64706, "loss_cns_0": 0.62325, "loss_yns_0": 0.14473, "loss_cls_1": 0.82209, "loss_box_1": 1.56565, "loss_cns_1": 0.65499, "loss_yns_1": 0.1419, "loss_cls_2": 0.8359, "loss_box_2": 1.52723, "loss_cns_2": 0.65528, "loss_yns_2": 0.14068, "loss_cls_3": 0.83835, "loss_box_3": 1.52694, "loss_cns_3": 0.6559, "loss_yns_3": 0.13958, "loss_cls_4": 0.8413, "loss_box_4": 1.5298, "loss_cns_4": 0.65565, "loss_yns_4": 0.14064, "loss_cls_5": 0.84587, "loss_box_5": 1.52322, "loss_cns_5": 0.65657, "loss_yns_5": 0.13932, "loss_cls_dn_0": 0.1298, "loss_box_dn_0": 0.71754, "loss_cls_dn_1": 0.10194, "loss_box_dn_1": 0.64368, "loss_cls_dn_2": 0.10272, "loss_box_dn_2": 0.63416, "loss_cls_dn_3": 0.10406, "loss_box_dn_3": 0.6356, "loss_cls_dn_4": 0.10381, "loss_box_dn_4": 0.63976, "loss_cls_dn_5": 0.10594, "loss_box_dn_5": 0.64603, "loss_dense_depth": 0.72968, "loss": 24.32232, "grad_norm": 32.51468, "time": 1.54894} -{"mode": "train", "epoch": 1, "iter": 345, "lr": 0.00024, "memory": 49163, "data_time": 0.07766, "loss_cls_0": 0.74156, "loss_box_0": 1.65446, "loss_cns_0": 0.63011, "loss_yns_0": 0.14232, "loss_cls_1": 0.80105, "loss_box_1": 1.56673, "loss_cns_1": 0.65246, "loss_yns_1": 0.14075, "loss_cls_2": 0.80954, "loss_box_2": 1.54329, "loss_cns_2": 0.6559, "loss_yns_2": 0.13972, "loss_cls_3": 0.81644, "loss_box_3": 1.53366, "loss_cns_3": 0.65452, "loss_yns_3": 0.13834, "loss_cls_4": 0.81563, "loss_box_4": 1.52869, "loss_cns_4": 0.6539, "loss_yns_4": 0.13807, "loss_cls_5": 0.82004, "loss_box_5": 1.53561, "loss_cns_5": 0.65653, "loss_yns_5": 0.13795, "loss_cls_dn_0": 0.12344, "loss_box_dn_0": 0.70946, "loss_cls_dn_1": 0.10225, "loss_box_dn_1": 0.64634, "loss_cls_dn_2": 0.10121, "loss_box_dn_2": 0.638, "loss_cls_dn_3": 0.10117, "loss_box_dn_3": 0.63692, "loss_cls_dn_4": 0.10077, "loss_box_dn_4": 0.64214, "loss_cls_dn_5": 0.10194, "loss_box_dn_5": 0.64981, "loss_dense_depth": 0.68882, "loss": 24.14954, "grad_norm": 39.14698, "time": 1.5081} -{"mode": "train", "epoch": 1, "iter": 346, "lr": 0.00024, "memory": 49163, "data_time": 0.07704, "loss_cls_0": 0.75912, "loss_box_0": 1.65214, "loss_cns_0": 0.63102, "loss_yns_0": 0.14249, "loss_cls_1": 0.8008, "loss_box_1": 1.52183, "loss_cns_1": 0.65752, "loss_yns_1": 0.13841, "loss_cls_2": 0.81431, "loss_box_2": 1.49173, "loss_cns_2": 0.66139, "loss_yns_2": 0.13844, "loss_cls_3": 0.80625, "loss_box_3": 1.48917, "loss_cns_3": 0.66265, "loss_yns_3": 0.13864, "loss_cls_4": 0.80873, "loss_box_4": 1.49562, "loss_cns_4": 0.66096, "loss_yns_4": 0.13865, "loss_cls_5": 0.81291, "loss_box_5": 1.49153, "loss_cns_5": 0.66071, "loss_yns_5": 0.13859, "loss_cls_dn_0": 0.12724, "loss_box_dn_0": 0.71949, "loss_cls_dn_1": 0.10261, "loss_box_dn_1": 0.65359, "loss_cls_dn_2": 0.10139, "loss_box_dn_2": 0.64366, "loss_cls_dn_3": 0.10104, "loss_box_dn_3": 0.64616, "loss_cls_dn_4": 0.10224, "loss_box_dn_4": 0.65351, "loss_cls_dn_5": 0.10353, "loss_box_dn_5": 0.65937, "loss_dense_depth": 0.70584, "loss": 24.03328, "grad_norm": 31.79389, "time": 1.50507} -{"mode": "train", "epoch": 1, "iter": 347, "lr": 0.00024, "memory": 49163, "data_time": 0.08347, "loss_cls_0": 0.72495, "loss_box_0": 1.62036, "loss_cns_0": 0.63358, "loss_yns_0": 0.14322, "loss_cls_1": 0.78992, "loss_box_1": 1.48766, "loss_cns_1": 0.66097, "loss_yns_1": 0.14004, "loss_cls_2": 0.80539, "loss_box_2": 1.47698, "loss_cns_2": 0.66267, "loss_yns_2": 0.14021, "loss_cls_3": 0.8064, "loss_box_3": 1.47364, "loss_cns_3": 0.67252, "loss_yns_3": 0.14062, "loss_cls_4": 0.81573, "loss_box_4": 1.46794, "loss_cns_4": 0.67078, "loss_yns_4": 0.13883, "loss_cls_5": 0.80893, "loss_box_5": 1.46337, "loss_cns_5": 0.66767, "loss_yns_5": 0.13954, "loss_cls_dn_0": 0.12352, "loss_box_dn_0": 0.70925, "loss_cls_dn_1": 0.10063, "loss_box_dn_1": 0.65408, "loss_cls_dn_2": 0.09828, "loss_box_dn_2": 0.64827, "loss_cls_dn_3": 0.0977, "loss_box_dn_3": 0.65028, "loss_cls_dn_4": 0.10191, "loss_box_dn_4": 0.65276, "loss_cls_dn_5": 0.10158, "loss_box_dn_5": 0.65215, "loss_dense_depth": 0.68535, "loss": 23.82769, "grad_norm": 47.90293, "time": 1.54218} -{"mode": "train", "epoch": 1, "iter": 348, "lr": 0.00024, "memory": 49163, "data_time": 0.08298, "loss_cls_0": 0.75186, "loss_box_0": 1.61344, "loss_cns_0": 0.63279, "loss_yns_0": 0.14438, "loss_cls_1": 0.81501, "loss_box_1": 1.48276, "loss_cns_1": 0.66249, "loss_yns_1": 0.14093, "loss_cls_2": 0.82566, "loss_box_2": 1.45181, "loss_cns_2": 0.66507, "loss_yns_2": 0.1407, "loss_cls_3": 0.82838, "loss_box_3": 1.44278, "loss_cns_3": 0.67156, "loss_yns_3": 0.14136, "loss_cls_4": 0.83189, "loss_box_4": 1.44143, "loss_cns_4": 0.66869, "loss_yns_4": 0.14011, "loss_cls_5": 0.82465, "loss_box_5": 1.45836, "loss_cns_5": 0.66655, "loss_yns_5": 0.14017, "loss_cls_dn_0": 0.12396, "loss_box_dn_0": 0.71204, "loss_cls_dn_1": 0.10088, "loss_box_dn_1": 0.65966, "loss_cls_dn_2": 0.09904, "loss_box_dn_2": 0.64417, "loss_cls_dn_3": 0.09928, "loss_box_dn_3": 0.64179, "loss_cls_dn_4": 0.10203, "loss_box_dn_4": 0.64273, "loss_cls_dn_5": 0.10111, "loss_box_dn_5": 0.64769, "loss_dense_depth": 0.71322, "loss": 23.87041, "grad_norm": 24.73577, "time": 1.48836} -{"mode": "train", "epoch": 1, "iter": 349, "lr": 0.00024, "memory": 49163, "data_time": 0.17333, "loss_cls_0": 0.7351, "loss_box_0": 1.63143, "loss_cns_0": 0.63163, "loss_yns_0": 0.14399, "loss_cls_1": 0.80568, "loss_box_1": 1.4805, "loss_cns_1": 0.66032, "loss_yns_1": 0.14077, "loss_cls_2": 0.80462, "loss_box_2": 1.45665, "loss_cns_2": 0.66275, "loss_yns_2": 0.14103, "loss_cls_3": 0.80733, "loss_box_3": 1.44333, "loss_cns_3": 0.66195, "loss_yns_3": 0.13884, "loss_cls_4": 0.81728, "loss_box_4": 1.44475, "loss_cns_4": 0.6601, "loss_yns_4": 0.13934, "loss_cls_5": 0.81953, "loss_box_5": 1.47066, "loss_cns_5": 0.66016, "loss_yns_5": 0.13905, "loss_cls_dn_0": 0.12338, "loss_box_dn_0": 0.71098, "loss_cls_dn_1": 0.09904, "loss_box_dn_1": 0.65935, "loss_cls_dn_2": 0.09721, "loss_box_dn_2": 0.64072, "loss_cls_dn_3": 0.09699, "loss_box_dn_3": 0.63533, "loss_cls_dn_4": 0.09864, "loss_box_dn_4": 0.63575, "loss_cls_dn_5": 0.09852, "loss_box_dn_5": 0.64563, "loss_dense_depth": 0.71981, "loss": 23.75816, "grad_norm": 51.30045, "time": 1.57899} -{"mode": "train", "epoch": 1, "iter": 350, "lr": 0.00024, "memory": 49163, "data_time": 0.07848, "loss_cls_0": 0.71395, "loss_box_0": 1.60132, "loss_cns_0": 0.63652, "loss_yns_0": 0.14396, "loss_cls_1": 0.79013, "loss_box_1": 1.45301, "loss_cns_1": 0.65912, "loss_yns_1": 0.14104, "loss_cls_2": 0.79138, "loss_box_2": 1.41729, "loss_cns_2": 0.65846, "loss_yns_2": 0.14003, "loss_cls_3": 0.79544, "loss_box_3": 1.4155, "loss_cns_3": 0.66228, "loss_yns_3": 0.13971, "loss_cls_4": 0.80713, "loss_box_4": 1.40255, "loss_cns_4": 0.66355, "loss_yns_4": 0.13975, "loss_cls_5": 0.80469, "loss_box_5": 1.41259, "loss_cns_5": 0.66084, "loss_yns_5": 0.14022, "loss_cls_dn_0": 0.12057, "loss_box_dn_0": 0.71126, "loss_cls_dn_1": 0.09984, "loss_box_dn_1": 0.64469, "loss_cls_dn_2": 0.09971, "loss_box_dn_2": 0.62886, "loss_cls_dn_3": 0.09822, "loss_box_dn_3": 0.62543, "loss_cls_dn_4": 0.09985, "loss_box_dn_4": 0.62192, "loss_cls_dn_5": 0.09932, "loss_box_dn_5": 0.62592, "loss_dense_depth": 0.69652, "loss": 23.36258, "grad_norm": 32.21, "time": 1.51378} -{"mode": "train", "epoch": 1, "iter": 351, "lr": 0.00024, "memory": 49163, "data_time": 0.07773, "loss_cls_0": 0.73716, "loss_box_0": 1.64861, "loss_cns_0": 0.63172, "loss_yns_0": 0.14853, "loss_cls_1": 0.79495, "loss_box_1": 1.503, "loss_cns_1": 0.65577, "loss_yns_1": 0.14464, "loss_cls_2": 0.80225, "loss_box_2": 1.45099, "loss_cns_2": 0.65796, "loss_yns_2": 0.14619, "loss_cls_3": 0.81928, "loss_box_3": 1.44157, "loss_cns_3": 0.66159, "loss_yns_3": 0.14387, "loss_cls_4": 0.82687, "loss_box_4": 1.44725, "loss_cns_4": 0.66345, "loss_yns_4": 0.14476, "loss_cls_5": 0.82942, "loss_box_5": 1.44906, "loss_cns_5": 0.6603, "loss_yns_5": 0.14452, "loss_cls_dn_0": 0.12418, "loss_box_dn_0": 0.71886, "loss_cls_dn_1": 0.10035, "loss_box_dn_1": 0.64182, "loss_cls_dn_2": 0.10118, "loss_box_dn_2": 0.63108, "loss_cls_dn_3": 0.09954, "loss_box_dn_3": 0.62739, "loss_cls_dn_4": 0.10174, "loss_box_dn_4": 0.62901, "loss_cls_dn_5": 0.10152, "loss_box_dn_5": 0.6327, "loss_dense_depth": 0.74718, "loss": 23.81026, "grad_norm": 59.00757, "time": 1.50376} -{"mode": "train", "epoch": 1, "iter": 352, "lr": 0.00024, "memory": 49163, "data_time": 0.07812, "loss_cls_0": 0.74793, "loss_box_0": 1.63336, "loss_cns_0": 0.63192, "loss_yns_0": 0.14926, "loss_cls_1": 0.80271, "loss_box_1": 1.5439, "loss_cns_1": 0.65342, "loss_yns_1": 0.14553, "loss_cls_2": 0.81089, "loss_box_2": 1.47321, "loss_cns_2": 0.65528, "loss_yns_2": 0.14647, "loss_cls_3": 0.82017, "loss_box_3": 1.44788, "loss_cns_3": 0.66058, "loss_yns_3": 0.14592, "loss_cls_4": 0.83065, "loss_box_4": 1.45284, "loss_cns_4": 0.66083, "loss_yns_4": 0.14515, "loss_cls_5": 0.82378, "loss_box_5": 1.45424, "loss_cns_5": 0.65743, "loss_yns_5": 0.14516, "loss_cls_dn_0": 0.12447, "loss_box_dn_0": 0.71529, "loss_cls_dn_1": 0.09951, "loss_box_dn_1": 0.64701, "loss_cls_dn_2": 0.10015, "loss_box_dn_2": 0.63652, "loss_cls_dn_3": 0.09925, "loss_box_dn_3": 0.62881, "loss_cls_dn_4": 0.10269, "loss_box_dn_4": 0.63139, "loss_cls_dn_5": 0.10236, "loss_box_dn_5": 0.63644, "loss_dense_depth": 0.74864, "loss": 23.91104, "grad_norm": 50.96195, "time": 1.49805} -{"mode": "train", "epoch": 1, "iter": 353, "lr": 0.00024, "memory": 49163, "data_time": 0.07948, "loss_cls_0": 0.73895, "loss_box_0": 1.63218, "loss_cns_0": 0.63078, "loss_yns_0": 0.14882, "loss_cls_1": 0.8167, "loss_box_1": 1.45391, "loss_cns_1": 0.66089, "loss_yns_1": 0.14427, "loss_cls_2": 0.814, "loss_box_2": 1.42505, "loss_cns_2": 0.65827, "loss_yns_2": 0.14457, "loss_cls_3": 0.81719, "loss_box_3": 1.41619, "loss_cns_3": 0.66265, "loss_yns_3": 0.14449, "loss_cls_4": 0.83718, "loss_box_4": 1.42623, "loss_cns_4": 0.66407, "loss_yns_4": 0.14598, "loss_cls_5": 0.82383, "loss_box_5": 1.42736, "loss_cns_5": 0.66161, "loss_yns_5": 0.14591, "loss_cls_dn_0": 0.12453, "loss_box_dn_0": 0.72797, "loss_cls_dn_1": 0.09903, "loss_box_dn_1": 0.64504, "loss_cls_dn_2": 0.09949, "loss_box_dn_2": 0.63675, "loss_cls_dn_3": 0.09924, "loss_box_dn_3": 0.63296, "loss_cls_dn_4": 0.1029, "loss_box_dn_4": 0.63842, "loss_cls_dn_5": 0.10211, "loss_box_dn_5": 0.64391, "loss_dense_depth": 0.70295, "loss": 23.6964, "grad_norm": 47.43281, "time": 1.49735} -{"mode": "train", "epoch": 1, "iter": 354, "lr": 0.00024, "memory": 49163, "data_time": 0.07797, "loss_cls_0": 0.74507, "loss_box_0": 1.63691, "loss_cns_0": 0.6281, "loss_yns_0": 0.14642, "loss_cls_1": 0.81561, "loss_box_1": 1.49401, "loss_cns_1": 0.65821, "loss_yns_1": 0.14464, "loss_cls_2": 0.81881, "loss_box_2": 1.4668, "loss_cns_2": 0.65811, "loss_yns_2": 0.14303, "loss_cls_3": 0.82521, "loss_box_3": 1.46017, "loss_cns_3": 0.66047, "loss_yns_3": 0.14269, "loss_cls_4": 0.83346, "loss_box_4": 1.46796, "loss_cns_4": 0.65912, "loss_yns_4": 0.1431, "loss_cls_5": 0.83086, "loss_box_5": 1.46984, "loss_cns_5": 0.65794, "loss_yns_5": 0.14346, "loss_cls_dn_0": 0.12475, "loss_box_dn_0": 0.72559, "loss_cls_dn_1": 0.09776, "loss_box_dn_1": 0.65605, "loss_cls_dn_2": 0.09827, "loss_box_dn_2": 0.64362, "loss_cls_dn_3": 0.09728, "loss_box_dn_3": 0.64243, "loss_cls_dn_4": 0.09934, "loss_box_dn_4": 0.64909, "loss_cls_dn_5": 0.10008, "loss_box_dn_5": 0.65356, "loss_dense_depth": 0.73223, "loss": 23.97006, "grad_norm": 56.75154, "time": 1.54269} -{"mode": "train", "epoch": 1, "iter": 355, "lr": 0.00024, "memory": 49163, "data_time": 0.07806, "loss_cls_0": 0.75237, "loss_box_0": 1.62018, "loss_cns_0": 0.63132, "loss_yns_0": 0.14612, "loss_cls_1": 0.82154, "loss_box_1": 1.49679, "loss_cns_1": 0.65762, "loss_yns_1": 0.14522, "loss_cls_2": 0.825, "loss_box_2": 1.46704, "loss_cns_2": 0.65824, "loss_yns_2": 0.14523, "loss_cls_3": 0.82643, "loss_box_3": 1.45221, "loss_cns_3": 0.65775, "loss_yns_3": 0.14373, "loss_cls_4": 0.83015, "loss_box_4": 1.45357, "loss_cns_4": 0.65906, "loss_yns_4": 0.1443, "loss_cls_5": 0.83113, "loss_box_5": 1.45851, "loss_cns_5": 0.65806, "loss_yns_5": 0.14452, "loss_cls_dn_0": 0.12764, "loss_box_dn_0": 0.71653, "loss_cls_dn_1": 0.09986, "loss_box_dn_1": 0.64603, "loss_cls_dn_2": 0.10001, "loss_box_dn_2": 0.6321, "loss_cls_dn_3": 0.09803, "loss_box_dn_3": 0.63027, "loss_cls_dn_4": 0.09934, "loss_box_dn_4": 0.63375, "loss_cls_dn_5": 0.10056, "loss_box_dn_5": 0.63651, "loss_dense_depth": 0.70467, "loss": 23.85138, "grad_norm": 28.55916, "time": 1.5094} -{"mode": "train", "epoch": 1, "iter": 356, "lr": 0.00024, "memory": 49163, "data_time": 0.07935, "loss_cls_0": 0.7587, "loss_box_0": 1.62525, "loss_cns_0": 0.63252, "loss_yns_0": 0.15048, "loss_cls_1": 0.8359, "loss_box_1": 1.5059, "loss_cns_1": 0.65625, "loss_yns_1": 0.14745, "loss_cls_2": 0.84406, "loss_box_2": 1.4609, "loss_cns_2": 0.65653, "loss_yns_2": 0.1479, "loss_cls_3": 0.85306, "loss_box_3": 1.45161, "loss_cns_3": 0.65804, "loss_yns_3": 0.1455, "loss_cls_4": 0.86222, "loss_box_4": 1.453, "loss_cns_4": 0.66141, "loss_yns_4": 0.1467, "loss_cls_5": 0.86011, "loss_box_5": 1.47381, "loss_cns_5": 0.65689, "loss_yns_5": 0.1466, "loss_cls_dn_0": 0.12725, "loss_box_dn_0": 0.72685, "loss_cls_dn_1": 0.10183, "loss_box_dn_1": 0.63781, "loss_cls_dn_2": 0.10002, "loss_box_dn_2": 0.61992, "loss_cls_dn_3": 0.09882, "loss_box_dn_3": 0.61531, "loss_cls_dn_4": 0.10009, "loss_box_dn_4": 0.61872, "loss_cls_dn_5": 0.10137, "loss_box_dn_5": 0.62493, "loss_dense_depth": 0.7547, "loss": 24.0184, "grad_norm": 49.08212, "time": 1.49281} -{"mode": "train", "epoch": 1, "iter": 357, "lr": 0.00024, "memory": 49163, "data_time": 0.07902, "loss_cls_0": 0.75115, "loss_box_0": 1.60792, "loss_cns_0": 0.63039, "loss_yns_0": 0.14696, "loss_cls_1": 0.80101, "loss_box_1": 1.55797, "loss_cns_1": 0.6525, "loss_yns_1": 0.14564, "loss_cls_2": 0.8123, "loss_box_2": 1.49336, "loss_cns_2": 0.65605, "loss_yns_2": 0.14684, "loss_cls_3": 0.81374, "loss_box_3": 1.48387, "loss_cns_3": 0.65683, "loss_yns_3": 0.14537, "loss_cls_4": 0.83238, "loss_box_4": 1.48306, "loss_cns_4": 0.65938, "loss_yns_4": 0.14548, "loss_cls_5": 0.81879, "loss_box_5": 1.49455, "loss_cns_5": 0.65476, "loss_yns_5": 0.14559, "loss_cls_dn_0": 0.12387, "loss_box_dn_0": 0.71941, "loss_cls_dn_1": 0.09863, "loss_box_dn_1": 0.6558, "loss_cls_dn_2": 0.09775, "loss_box_dn_2": 0.62686, "loss_cls_dn_3": 0.09669, "loss_box_dn_3": 0.62099, "loss_cls_dn_4": 0.10061, "loss_box_dn_4": 0.62434, "loss_cls_dn_5": 0.10155, "loss_box_dn_5": 0.62786, "loss_dense_depth": 0.71952, "loss": 23.94979, "grad_norm": 43.60805, "time": 1.54017} -{"mode": "train", "epoch": 1, "iter": 358, "lr": 0.00024, "memory": 49163, "data_time": 0.07904, "loss_cls_0": 0.7662, "loss_box_0": 1.59018, "loss_cns_0": 0.6314, "loss_yns_0": 0.14445, "loss_cls_1": 0.79296, "loss_box_1": 1.46738, "loss_cns_1": 0.65575, "loss_yns_1": 0.1424, "loss_cls_2": 0.79952, "loss_box_2": 1.44739, "loss_cns_2": 0.6595, "loss_yns_2": 0.14265, "loss_cls_3": 0.80312, "loss_box_3": 1.43726, "loss_cns_3": 0.65672, "loss_yns_3": 0.14357, "loss_cls_4": 0.81531, "loss_box_4": 1.42788, "loss_cns_4": 0.65551, "loss_yns_4": 0.14296, "loss_cls_5": 0.82045, "loss_box_5": 1.43556, "loss_cns_5": 0.65674, "loss_yns_5": 0.14297, "loss_cls_dn_0": 0.1251, "loss_box_dn_0": 0.71282, "loss_cls_dn_1": 0.10212, "loss_box_dn_1": 0.64434, "loss_cls_dn_2": 0.10306, "loss_box_dn_2": 0.62454, "loss_cls_dn_3": 0.10158, "loss_box_dn_3": 0.62128, "loss_cls_dn_4": 0.10341, "loss_box_dn_4": 0.62175, "loss_cls_dn_5": 0.10391, "loss_box_dn_5": 0.61995, "loss_dense_depth": 0.70423, "loss": 23.56589, "grad_norm": 48.22754, "time": 1.48434} -{"mode": "train", "epoch": 1, "iter": 359, "lr": 0.00024, "memory": 49163, "data_time": 0.07741, "loss_cls_0": 0.72265, "loss_box_0": 1.56012, "loss_cns_0": 0.63196, "loss_yns_0": 0.13997, "loss_cls_1": 0.77613, "loss_box_1": 1.46538, "loss_cns_1": 0.65795, "loss_yns_1": 0.13954, "loss_cls_2": 0.79537, "loss_box_2": 1.41096, "loss_cns_2": 0.66042, "loss_yns_2": 0.13795, "loss_cls_3": 0.79655, "loss_box_3": 1.40573, "loss_cns_3": 0.66073, "loss_yns_3": 0.13858, "loss_cls_4": 0.79789, "loss_box_4": 1.39745, "loss_cns_4": 0.66165, "loss_yns_4": 0.13927, "loss_cls_5": 0.81869, "loss_box_5": 1.39746, "loss_cns_5": 0.66049, "loss_yns_5": 0.13856, "loss_cls_dn_0": 0.12652, "loss_box_dn_0": 0.71411, "loss_cls_dn_1": 0.10373, "loss_box_dn_1": 0.63661, "loss_cls_dn_2": 0.10575, "loss_box_dn_2": 0.61977, "loss_cls_dn_3": 0.10566, "loss_box_dn_3": 0.61807, "loss_cls_dn_4": 0.10375, "loss_box_dn_4": 0.61689, "loss_cls_dn_5": 0.10647, "loss_box_dn_5": 0.61721, "loss_dense_depth": 0.72394, "loss": 23.30996, "grad_norm": 43.08294, "time": 1.48357} -{"mode": "train", "epoch": 1, "iter": 360, "lr": 0.00024, "memory": 49163, "data_time": 0.08276, "loss_cls_0": 0.74176, "loss_box_0": 1.58447, "loss_cns_0": 0.62919, "loss_yns_0": 0.14302, "loss_cls_1": 0.7916, "loss_box_1": 1.48726, "loss_cns_1": 0.65786, "loss_yns_1": 0.14319, "loss_cls_2": 0.81268, "loss_box_2": 1.43907, "loss_cns_2": 0.65893, "loss_yns_2": 0.14246, "loss_cls_3": 0.82521, "loss_box_3": 1.43413, "loss_cns_3": 0.6639, "loss_yns_3": 0.14195, "loss_cls_4": 0.82927, "loss_box_4": 1.42618, "loss_cns_4": 0.6633, "loss_yns_4": 0.14275, "loss_cls_5": 0.82787, "loss_box_5": 1.43648, "loss_cns_5": 0.66412, "loss_yns_5": 0.143, "loss_cls_dn_0": 0.12611, "loss_box_dn_0": 0.71163, "loss_cls_dn_1": 0.10287, "loss_box_dn_1": 0.62751, "loss_cls_dn_2": 0.10274, "loss_box_dn_2": 0.61317, "loss_cls_dn_3": 0.10268, "loss_box_dn_3": 0.61407, "loss_cls_dn_4": 0.10332, "loss_box_dn_4": 0.61423, "loss_cls_dn_5": 0.10567, "loss_box_dn_5": 0.6199, "loss_dense_depth": 0.6968, "loss": 23.5703, "grad_norm": 37.46267, "time": 1.50491} -{"mode": "train", "epoch": 1, "iter": 361, "lr": 0.00024, "memory": 49163, "data_time": 0.07931, "loss_cls_0": 0.75821, "loss_box_0": 1.5709, "loss_cns_0": 0.62794, "loss_yns_0": 0.14457, "loss_cls_1": 0.80445, "loss_box_1": 1.4469, "loss_cns_1": 0.66209, "loss_yns_1": 0.14435, "loss_cls_2": 0.82705, "loss_box_2": 1.43603, "loss_cns_2": 0.66396, "loss_yns_2": 0.14406, "loss_cls_3": 0.84223, "loss_box_3": 1.42548, "loss_cns_3": 0.66756, "loss_yns_3": 0.14451, "loss_cls_4": 0.84773, "loss_box_4": 1.41618, "loss_cns_4": 0.66618, "loss_yns_4": 0.14615, "loss_cls_5": 0.84719, "loss_box_5": 1.42447, "loss_cns_5": 0.66644, "loss_yns_5": 0.14529, "loss_cls_dn_0": 0.12363, "loss_box_dn_0": 0.71112, "loss_cls_dn_1": 0.10199, "loss_box_dn_1": 0.63319, "loss_cls_dn_2": 0.1025, "loss_box_dn_2": 0.62919, "loss_cls_dn_3": 0.10232, "loss_box_dn_3": 0.62917, "loss_cls_dn_4": 0.10358, "loss_box_dn_4": 0.63164, "loss_cls_dn_5": 0.10439, "loss_box_dn_5": 0.63804, "loss_dense_depth": 0.72307, "loss": 23.70371, "grad_norm": 48.48491, "time": 1.52258} -{"mode": "train", "epoch": 1, "iter": 362, "lr": 0.00024, "memory": 49163, "data_time": 0.08083, "loss_cls_0": 0.76208, "loss_box_0": 1.59381, "loss_cns_0": 0.62629, "loss_yns_0": 0.14621, "loss_cls_1": 0.80976, "loss_box_1": 1.47136, "loss_cns_1": 0.66092, "loss_yns_1": 0.14482, "loss_cls_2": 0.82631, "loss_box_2": 1.45838, "loss_cns_2": 0.66102, "loss_yns_2": 0.14539, "loss_cls_3": 0.83157, "loss_box_3": 1.43933, "loss_cns_3": 0.66179, "loss_yns_3": 0.14447, "loss_cls_4": 0.83463, "loss_box_4": 1.44017, "loss_cns_4": 0.6616, "loss_yns_4": 0.14572, "loss_cls_5": 0.83647, "loss_box_5": 1.43919, "loss_cns_5": 0.66206, "loss_yns_5": 0.14566, "loss_cls_dn_0": 0.12899, "loss_box_dn_0": 0.71123, "loss_cls_dn_1": 0.10171, "loss_box_dn_1": 0.63659, "loss_cls_dn_2": 0.10183, "loss_box_dn_2": 0.63489, "loss_cls_dn_3": 0.10166, "loss_box_dn_3": 0.63148, "loss_cls_dn_4": 0.10304, "loss_box_dn_4": 0.63215, "loss_cls_dn_5": 0.10478, "loss_box_dn_5": 0.63551, "loss_dense_depth": 0.74197, "loss": 23.81484, "grad_norm": 34.66333, "time": 1.57689} -{"mode": "train", "epoch": 1, "iter": 363, "lr": 0.00024, "memory": 49163, "data_time": 0.07901, "loss_cls_0": 0.74623, "loss_box_0": 1.59249, "loss_cns_0": 0.63267, "loss_yns_0": 0.14203, "loss_cls_1": 0.80388, "loss_box_1": 1.47439, "loss_cns_1": 0.65993, "loss_yns_1": 0.13987, "loss_cls_2": 0.82738, "loss_box_2": 1.4628, "loss_cns_2": 0.66046, "loss_yns_2": 0.13943, "loss_cls_3": 0.83183, "loss_box_3": 1.4363, "loss_cns_3": 0.66069, "loss_yns_3": 0.13932, "loss_cls_4": 0.83127, "loss_box_4": 1.4446, "loss_cns_4": 0.66373, "loss_yns_4": 0.13896, "loss_cls_5": 0.83747, "loss_box_5": 1.44432, "loss_cns_5": 0.66177, "loss_yns_5": 0.13982, "loss_cls_dn_0": 0.12174, "loss_box_dn_0": 0.72138, "loss_cls_dn_1": 0.0968, "loss_box_dn_1": 0.64239, "loss_cls_dn_2": 0.09748, "loss_box_dn_2": 0.63907, "loss_cls_dn_3": 0.09778, "loss_box_dn_3": 0.6289, "loss_cls_dn_4": 0.09795, "loss_box_dn_4": 0.63026, "loss_cls_dn_5": 0.10144, "loss_box_dn_5": 0.63179, "loss_dense_depth": 0.72232, "loss": 23.74095, "grad_norm": 47.0331, "time": 3.42655} -{"mode": "train", "epoch": 1, "iter": 364, "lr": 0.00024, "memory": 49163, "data_time": 0.0771, "loss_cls_0": 0.73806, "loss_box_0": 1.5829, "loss_cns_0": 0.63343, "loss_yns_0": 0.14088, "loss_cls_1": 0.78961, "loss_box_1": 1.49011, "loss_cns_1": 0.65939, "loss_yns_1": 0.13908, "loss_cls_2": 0.81365, "loss_box_2": 1.46285, "loss_cns_2": 0.65927, "loss_yns_2": 0.1391, "loss_cls_3": 0.81544, "loss_box_3": 1.44362, "loss_cns_3": 0.66104, "loss_yns_3": 0.13772, "loss_cls_4": 0.81435, "loss_box_4": 1.44415, "loss_cns_4": 0.66276, "loss_yns_4": 0.1389, "loss_cls_5": 0.81991, "loss_box_5": 1.44304, "loss_cns_5": 0.66076, "loss_yns_5": 0.13973, "loss_cls_dn_0": 0.12098, "loss_box_dn_0": 0.71643, "loss_cls_dn_1": 0.09531, "loss_box_dn_1": 0.63025, "loss_cls_dn_2": 0.09626, "loss_box_dn_2": 0.6206, "loss_cls_dn_3": 0.09744, "loss_box_dn_3": 0.61129, "loss_cls_dn_4": 0.09675, "loss_box_dn_4": 0.61107, "loss_cls_dn_5": 0.09939, "loss_box_dn_5": 0.60921, "loss_dense_depth": 0.73912, "loss": 23.57384, "grad_norm": 33.81545, "time": 1.53052} -{"mode": "train", "epoch": 1, "iter": 365, "lr": 0.00025, "memory": 49163, "data_time": 0.07791, "loss_cls_0": 0.73493, "loss_box_0": 1.6202, "loss_cns_0": 0.6348, "loss_yns_0": 0.14172, "loss_cls_1": 0.78718, "loss_box_1": 1.51331, "loss_cns_1": 0.65634, "loss_yns_1": 0.1403, "loss_cls_2": 0.80579, "loss_box_2": 1.46255, "loss_cns_2": 0.65881, "loss_yns_2": 0.13983, "loss_cls_3": 0.81032, "loss_box_3": 1.46273, "loss_cns_3": 0.65892, "loss_yns_3": 0.13908, "loss_cls_4": 0.81831, "loss_box_4": 1.45701, "loss_cns_4": 0.65877, "loss_yns_4": 0.13903, "loss_cls_5": 0.82, "loss_box_5": 1.46387, "loss_cns_5": 0.65966, "loss_yns_5": 0.14018, "loss_cls_dn_0": 0.1208, "loss_box_dn_0": 0.7249, "loss_cls_dn_1": 0.09848, "loss_box_dn_1": 0.62646, "loss_cls_dn_2": 0.09964, "loss_box_dn_2": 0.60649, "loss_cls_dn_3": 0.0994, "loss_box_dn_3": 0.60367, "loss_cls_dn_4": 0.10015, "loss_box_dn_4": 0.60248, "loss_cls_dn_5": 0.10035, "loss_box_dn_5": 0.60373, "loss_dense_depth": 0.69765, "loss": 23.60782, "grad_norm": 43.85792, "time": 1.49841} -{"mode": "train", "epoch": 1, "iter": 366, "lr": 0.00025, "memory": 49163, "data_time": 0.08213, "loss_cls_0": 0.71084, "loss_box_0": 1.6044, "loss_cns_0": 0.63492, "loss_yns_0": 0.139, "loss_cls_1": 0.7682, "loss_box_1": 1.51671, "loss_cns_1": 0.65499, "loss_yns_1": 0.13703, "loss_cls_2": 0.79514, "loss_box_2": 1.46533, "loss_cns_2": 0.65745, "loss_yns_2": 0.13693, "loss_cls_3": 0.79748, "loss_box_3": 1.46187, "loss_cns_3": 0.65757, "loss_yns_3": 0.13674, "loss_cls_4": 0.80274, "loss_box_4": 1.45309, "loss_cns_4": 0.65822, "loss_yns_4": 0.13609, "loss_cls_5": 0.79949, "loss_box_5": 1.45956, "loss_cns_5": 0.65919, "loss_yns_5": 0.13693, "loss_cls_dn_0": 0.11611, "loss_box_dn_0": 0.72202, "loss_cls_dn_1": 0.09518, "loss_box_dn_1": 0.62802, "loss_cls_dn_2": 0.09423, "loss_box_dn_2": 0.61162, "loss_cls_dn_3": 0.0934, "loss_box_dn_3": 0.60856, "loss_cls_dn_4": 0.09497, "loss_box_dn_4": 0.6076, "loss_cls_dn_5": 0.09517, "loss_box_dn_5": 0.60851, "loss_dense_depth": 0.70822, "loss": 23.46353, "grad_norm": 43.51766, "time": 1.5236} -{"mode": "train", "epoch": 1, "iter": 367, "lr": 0.00025, "memory": 49163, "data_time": 0.08664, "loss_cls_0": 0.70897, "loss_box_0": 1.6169, "loss_cns_0": 0.63366, "loss_yns_0": 0.13614, "loss_cls_1": 0.77446, "loss_box_1": 1.4632, "loss_cns_1": 0.65955, "loss_yns_1": 0.13372, "loss_cls_2": 0.79398, "loss_box_2": 1.42777, "loss_cns_2": 0.65941, "loss_yns_2": 0.13561, "loss_cls_3": 0.79606, "loss_box_3": 1.4338, "loss_cns_3": 0.66095, "loss_yns_3": 0.13555, "loss_cls_4": 0.80092, "loss_box_4": 1.43018, "loss_cns_4": 0.65945, "loss_yns_4": 0.13585, "loss_cls_5": 0.80428, "loss_box_5": 1.43761, "loss_cns_5": 0.66061, "loss_yns_5": 0.13616, "loss_cls_dn_0": 0.11811, "loss_box_dn_0": 0.7213, "loss_cls_dn_1": 0.0956, "loss_box_dn_1": 0.63215, "loss_cls_dn_2": 0.09496, "loss_box_dn_2": 0.62276, "loss_cls_dn_3": 0.09489, "loss_box_dn_3": 0.61837, "loss_cls_dn_4": 0.09631, "loss_box_dn_4": 0.62175, "loss_cls_dn_5": 0.09746, "loss_box_dn_5": 0.62259, "loss_dense_depth": 0.68693, "loss": 23.35793, "grad_norm": 36.10677, "time": 1.52833} -{"mode": "train", "epoch": 1, "iter": 368, "lr": 0.00025, "memory": 49163, "data_time": 0.08477, "loss_cls_0": 0.71975, "loss_box_0": 1.61452, "loss_cns_0": 0.62943, "loss_yns_0": 0.13629, "loss_cls_1": 0.78182, "loss_box_1": 1.46458, "loss_cns_1": 0.65818, "loss_yns_1": 0.13373, "loss_cls_2": 0.80238, "loss_box_2": 1.43453, "loss_cns_2": 0.66016, "loss_yns_2": 0.13376, "loss_cls_3": 0.80303, "loss_box_3": 1.43422, "loss_cns_3": 0.65796, "loss_yns_3": 0.13384, "loss_cls_4": 0.80808, "loss_box_4": 1.4342, "loss_cns_4": 0.65646, "loss_yns_4": 0.13237, "loss_cls_5": 0.81034, "loss_box_5": 1.435, "loss_cns_5": 0.65838, "loss_yns_5": 0.13383, "loss_cls_dn_0": 0.11777, "loss_box_dn_0": 0.71852, "loss_cls_dn_1": 0.09514, "loss_box_dn_1": 0.63112, "loss_cls_dn_2": 0.09465, "loss_box_dn_2": 0.62467, "loss_cls_dn_3": 0.09518, "loss_box_dn_3": 0.62063, "loss_cls_dn_4": 0.09625, "loss_box_dn_4": 0.62623, "loss_cls_dn_5": 0.09763, "loss_box_dn_5": 0.62892, "loss_dense_depth": 0.7143, "loss": 23.42786, "grad_norm": 43.56801, "time": 1.50928} -{"mode": "train", "epoch": 1, "iter": 369, "lr": 0.00025, "memory": 49163, "data_time": 0.21257, "loss_cls_0": 0.71327, "loss_box_0": 1.585, "loss_cns_0": 0.63736, "loss_yns_0": 0.13638, "loss_cls_1": 0.77734, "loss_box_1": 1.45671, "loss_cns_1": 0.66227, "loss_yns_1": 0.13206, "loss_cls_2": 0.78773, "loss_box_2": 1.41425, "loss_cns_2": 0.66176, "loss_yns_2": 0.13258, "loss_cls_3": 0.79084, "loss_box_3": 1.41634, "loss_cns_3": 0.6616, "loss_yns_3": 0.13232, "loss_cls_4": 0.79434, "loss_box_4": 1.4144, "loss_cns_4": 0.6618, "loss_yns_4": 0.13234, "loss_cls_5": 0.79877, "loss_box_5": 1.41635, "loss_cns_5": 0.66285, "loss_yns_5": 0.13362, "loss_cls_dn_0": 0.1166, "loss_box_dn_0": 0.70977, "loss_cls_dn_1": 0.09743, "loss_box_dn_1": 0.63388, "loss_cls_dn_2": 0.0963, "loss_box_dn_2": 0.62372, "loss_cls_dn_3": 0.09734, "loss_box_dn_3": 0.61978, "loss_cls_dn_4": 0.09784, "loss_box_dn_4": 0.62172, "loss_cls_dn_5": 0.09762, "loss_box_dn_5": 0.62795, "loss_dense_depth": 0.68949, "loss": 23.24172, "grad_norm": 34.72483, "time": 1.61137} -{"mode": "train", "epoch": 1, "iter": 370, "lr": 0.00025, "memory": 49163, "data_time": 0.08021, "loss_cls_0": 0.72251, "loss_box_0": 1.59132, "loss_cns_0": 0.6408, "loss_yns_0": 0.13782, "loss_cls_1": 0.78676, "loss_box_1": 1.44395, "loss_cns_1": 0.66387, "loss_yns_1": 0.13729, "loss_cls_2": 0.79033, "loss_box_2": 1.43399, "loss_cns_2": 0.66513, "loss_yns_2": 0.13736, "loss_cls_3": 0.7961, "loss_box_3": 1.41867, "loss_cns_3": 0.66575, "loss_yns_3": 0.13764, "loss_cls_4": 0.79818, "loss_box_4": 1.42039, "loss_cns_4": 0.66645, "loss_yns_4": 0.13859, "loss_cls_5": 0.79801, "loss_box_5": 1.42541, "loss_cns_5": 0.66523, "loss_yns_5": 0.13972, "loss_cls_dn_0": 0.1158, "loss_box_dn_0": 0.71549, "loss_cls_dn_1": 0.09683, "loss_box_dn_1": 0.63639, "loss_cls_dn_2": 0.09541, "loss_box_dn_2": 0.62867, "loss_cls_dn_3": 0.09508, "loss_box_dn_3": 0.62207, "loss_cls_dn_4": 0.09633, "loss_box_dn_4": 0.62088, "loss_cls_dn_5": 0.09565, "loss_box_dn_5": 0.62536, "loss_dense_depth": 0.71541, "loss": 23.38062, "grad_norm": 35.86614, "time": 1.50472} -{"mode": "train", "epoch": 1, "iter": 371, "lr": 0.00025, "memory": 49163, "data_time": 0.07876, "loss_cls_0": 0.71016, "loss_box_0": 1.58586, "loss_cns_0": 0.63761, "loss_yns_0": 0.1393, "loss_cls_1": 0.77629, "loss_box_1": 1.44179, "loss_cns_1": 0.66011, "loss_yns_1": 0.13821, "loss_cls_2": 0.78775, "loss_box_2": 1.41756, "loss_cns_2": 0.66048, "loss_yns_2": 0.13902, "loss_cls_3": 0.77976, "loss_box_3": 1.41203, "loss_cns_3": 0.65913, "loss_yns_3": 0.13828, "loss_cls_4": 0.78339, "loss_box_4": 1.41393, "loss_cns_4": 0.65873, "loss_yns_4": 0.13939, "loss_cls_5": 0.78543, "loss_box_5": 1.41408, "loss_cns_5": 0.65996, "loss_yns_5": 0.14018, "loss_cls_dn_0": 0.10859, "loss_box_dn_0": 0.71254, "loss_cls_dn_1": 0.09179, "loss_box_dn_1": 0.62226, "loss_cls_dn_2": 0.09074, "loss_box_dn_2": 0.61038, "loss_cls_dn_3": 0.08977, "loss_box_dn_3": 0.60663, "loss_cls_dn_4": 0.09191, "loss_box_dn_4": 0.60655, "loss_cls_dn_5": 0.09126, "loss_box_dn_5": 0.60762, "loss_dense_depth": 0.69428, "loss": 23.10274, "grad_norm": 27.03503, "time": 1.50043} -{"mode": "train", "epoch": 1, "iter": 372, "lr": 0.00025, "memory": 49163, "data_time": 0.08179, "loss_cls_0": 0.72614, "loss_box_0": 1.59454, "loss_cns_0": 0.63606, "loss_yns_0": 0.14057, "loss_cls_1": 0.78308, "loss_box_1": 1.48315, "loss_cns_1": 0.65783, "loss_yns_1": 0.13949, "loss_cls_2": 0.78855, "loss_box_2": 1.45047, "loss_cns_2": 0.65926, "loss_yns_2": 0.14043, "loss_cls_3": 0.79655, "loss_box_3": 1.44339, "loss_cns_3": 0.65722, "loss_yns_3": 0.13941, "loss_cls_4": 0.79387, "loss_box_4": 1.45329, "loss_cns_4": 0.65829, "loss_yns_4": 0.1408, "loss_cls_5": 0.8058, "loss_box_5": 1.45596, "loss_cns_5": 0.65786, "loss_yns_5": 0.14214, "loss_cls_dn_0": 0.11512, "loss_box_dn_0": 0.71366, "loss_cls_dn_1": 0.09314, "loss_box_dn_1": 0.61842, "loss_cls_dn_2": 0.09148, "loss_box_dn_2": 0.60254, "loss_cls_dn_3": 0.09156, "loss_box_dn_3": 0.60093, "loss_cls_dn_4": 0.09251, "loss_box_dn_4": 0.6022, "loss_cls_dn_5": 0.09338, "loss_box_dn_5": 0.60466, "loss_dense_depth": 0.72729, "loss": 23.39104, "grad_norm": 33.88337, "time": 1.50035} -{"mode": "train", "epoch": 1, "iter": 373, "lr": 0.00025, "memory": 49163, "data_time": 0.07979, "loss_cls_0": 0.75304, "loss_box_0": 1.63235, "loss_cns_0": 0.63301, "loss_yns_0": 0.14143, "loss_cls_1": 0.81645, "loss_box_1": 1.50134, "loss_cns_1": 0.65988, "loss_yns_1": 0.13991, "loss_cls_2": 0.81815, "loss_box_2": 1.47982, "loss_cns_2": 0.66299, "loss_yns_2": 0.14135, "loss_cls_3": 0.82858, "loss_box_3": 1.46378, "loss_cns_3": 0.66038, "loss_yns_3": 0.14099, "loss_cls_4": 0.82812, "loss_box_4": 1.46762, "loss_cns_4": 0.66126, "loss_yns_4": 0.14149, "loss_cls_5": 0.83204, "loss_box_5": 1.46798, "loss_cns_5": 0.66174, "loss_yns_5": 0.14355, "loss_cls_dn_0": 0.11575, "loss_box_dn_0": 0.70276, "loss_cls_dn_1": 0.09266, "loss_box_dn_1": 0.62606, "loss_cls_dn_2": 0.09152, "loss_box_dn_2": 0.61107, "loss_cls_dn_3": 0.09104, "loss_box_dn_3": 0.60646, "loss_cls_dn_4": 0.0909, "loss_box_dn_4": 0.60908, "loss_cls_dn_5": 0.09214, "loss_box_dn_5": 0.61099, "loss_dense_depth": 0.73198, "loss": 23.74964, "grad_norm": 31.96816, "time": 1.51169} -{"mode": "train", "epoch": 1, "iter": 374, "lr": 0.00025, "memory": 49163, "data_time": 0.07742, "loss_cls_0": 0.7311, "loss_box_0": 1.60531, "loss_cns_0": 0.63044, "loss_yns_0": 0.14238, "loss_cls_1": 0.80168, "loss_box_1": 1.49065, "loss_cns_1": 0.65744, "loss_yns_1": 0.14141, "loss_cls_2": 0.8077, "loss_box_2": 1.45396, "loss_cns_2": 0.65762, "loss_yns_2": 0.14103, "loss_cls_3": 0.80915, "loss_box_3": 1.437, "loss_cns_3": 0.65657, "loss_yns_3": 0.1413, "loss_cls_4": 0.81463, "loss_box_4": 1.43553, "loss_cns_4": 0.65666, "loss_yns_4": 0.14183, "loss_cls_5": 0.81334, "loss_box_5": 1.43476, "loss_cns_5": 0.65843, "loss_yns_5": 0.14155, "loss_cls_dn_0": 0.1131, "loss_box_dn_0": 0.70843, "loss_cls_dn_1": 0.09257, "loss_box_dn_1": 0.62185, "loss_cls_dn_2": 0.09219, "loss_box_dn_2": 0.60527, "loss_cls_dn_3": 0.09091, "loss_box_dn_3": 0.60098, "loss_cls_dn_4": 0.09168, "loss_box_dn_4": 0.60213, "loss_cls_dn_5": 0.09288, "loss_box_dn_5": 0.60454, "loss_dense_depth": 0.70627, "loss": 23.42426, "grad_norm": 22.38781, "time": 1.54552} -{"mode": "train", "epoch": 1, "iter": 375, "lr": 0.00025, "memory": 49163, "data_time": 0.07698, "loss_cls_0": 0.7286, "loss_box_0": 1.60479, "loss_cns_0": 0.63253, "loss_yns_0": 0.1447, "loss_cls_1": 0.78805, "loss_box_1": 1.43711, "loss_cns_1": 0.65977, "loss_yns_1": 0.14268, "loss_cls_2": 0.79894, "loss_box_2": 1.4196, "loss_cns_2": 0.65906, "loss_yns_2": 0.14198, "loss_cls_3": 0.80064, "loss_box_3": 1.41634, "loss_cns_3": 0.65777, "loss_yns_3": 0.14229, "loss_cls_4": 0.80263, "loss_box_4": 1.41993, "loss_cns_4": 0.65952, "loss_yns_4": 0.14433, "loss_cls_5": 0.80265, "loss_box_5": 1.4246, "loss_cns_5": 0.65898, "loss_yns_5": 0.14279, "loss_cls_dn_0": 0.11255, "loss_box_dn_0": 0.71468, "loss_cls_dn_1": 0.09254, "loss_box_dn_1": 0.61486, "loss_cls_dn_2": 0.09248, "loss_box_dn_2": 0.60892, "loss_cls_dn_3": 0.09097, "loss_box_dn_3": 0.61212, "loss_cls_dn_4": 0.09249, "loss_box_dn_4": 0.61414, "loss_cls_dn_5": 0.09303, "loss_box_dn_5": 0.61967, "loss_dense_depth": 0.70308, "loss": 23.29183, "grad_norm": 35.03255, "time": 1.49316} -{"mode": "train", "epoch": 1, "iter": 376, "lr": 0.00025, "memory": 49163, "data_time": 0.08372, "loss_cls_0": 0.75464, "loss_box_0": 1.6062, "loss_cns_0": 0.63428, "loss_yns_0": 0.14471, "loss_cls_1": 0.81094, "loss_box_1": 1.46038, "loss_cns_1": 0.65732, "loss_yns_1": 0.14436, "loss_cls_2": 0.81753, "loss_box_2": 1.43781, "loss_cns_2": 0.65683, "loss_yns_2": 0.14335, "loss_cls_3": 0.82883, "loss_box_3": 1.4232, "loss_cns_3": 0.6524, "loss_yns_3": 0.14262, "loss_cls_4": 0.83213, "loss_box_4": 1.44504, "loss_cns_4": 0.65795, "loss_yns_4": 0.14377, "loss_cls_5": 0.83249, "loss_box_5": 1.45564, "loss_cns_5": 0.65723, "loss_yns_5": 0.14524, "loss_cls_dn_0": 0.11806, "loss_box_dn_0": 0.70374, "loss_cls_dn_1": 0.09619, "loss_box_dn_1": 0.65282, "loss_cls_dn_2": 0.09601, "loss_box_dn_2": 0.64484, "loss_cls_dn_3": 0.09446, "loss_box_dn_3": 0.65086, "loss_cls_dn_4": 0.0951, "loss_box_dn_4": 0.65704, "loss_cls_dn_5": 0.09552, "loss_box_dn_5": 0.66558, "loss_dense_depth": 0.7306, "loss": 23.78571, "grad_norm": 33.69868, "time": 1.4946} -{"mode": "train", "epoch": 1, "iter": 377, "lr": 0.00025, "memory": 49163, "data_time": 0.07763, "loss_cls_0": 0.75021, "loss_box_0": 1.57774, "loss_cns_0": 0.6322, "loss_yns_0": 0.14482, "loss_cls_1": 0.81076, "loss_box_1": 1.45216, "loss_cns_1": 0.65668, "loss_yns_1": 0.14398, "loss_cls_2": 0.81785, "loss_box_2": 1.43595, "loss_cns_2": 0.65525, "loss_yns_2": 0.14322, "loss_cls_3": 0.8306, "loss_box_3": 1.41988, "loss_cns_3": 0.6545, "loss_yns_3": 0.142, "loss_cls_4": 0.83573, "loss_box_4": 1.42893, "loss_cns_4": 0.65985, "loss_yns_4": 0.14449, "loss_cls_5": 0.82817, "loss_box_5": 1.42643, "loss_cns_5": 0.65773, "loss_yns_5": 0.1435, "loss_cls_dn_0": 0.11185, "loss_box_dn_0": 0.71602, "loss_cls_dn_1": 0.09522, "loss_box_dn_1": 0.66971, "loss_cls_dn_2": 0.09429, "loss_box_dn_2": 0.65835, "loss_cls_dn_3": 0.09324, "loss_box_dn_3": 0.65805, "loss_cls_dn_4": 0.09388, "loss_box_dn_4": 0.66222, "loss_cls_dn_5": 0.09401, "loss_box_dn_5": 0.66556, "loss_dense_depth": 0.72427, "loss": 23.72927, "grad_norm": 40.13134, "time": 1.52039} -{"mode": "train", "epoch": 1, "iter": 378, "lr": 0.00025, "memory": 49163, "data_time": 0.07935, "loss_cls_0": 0.7417, "loss_box_0": 1.56467, "loss_cns_0": 0.63039, "loss_yns_0": 0.14513, "loss_cls_1": 0.79556, "loss_box_1": 1.46588, "loss_cns_1": 0.65618, "loss_yns_1": 0.1448, "loss_cls_2": 0.80091, "loss_box_2": 1.45087, "loss_cns_2": 0.65661, "loss_yns_2": 0.14511, "loss_cls_3": 0.80388, "loss_box_3": 1.4303, "loss_cns_3": 0.65581, "loss_yns_3": 0.14362, "loss_cls_4": 0.81084, "loss_box_4": 1.43157, "loss_cns_4": 0.6564, "loss_yns_4": 0.14691, "loss_cls_5": 0.80497, "loss_box_5": 1.4316, "loss_cns_5": 0.65626, "loss_yns_5": 0.14248, "loss_cls_dn_0": 0.11104, "loss_box_dn_0": 0.70897, "loss_cls_dn_1": 0.09102, "loss_box_dn_1": 0.65727, "loss_cls_dn_2": 0.08974, "loss_box_dn_2": 0.64127, "loss_cls_dn_3": 0.08925, "loss_box_dn_3": 0.63469, "loss_cls_dn_4": 0.08995, "loss_box_dn_4": 0.6351, "loss_cls_dn_5": 0.0901, "loss_box_dn_5": 0.63689, "loss_dense_depth": 0.72275, "loss": 23.51049, "grad_norm": 28.43977, "time": 1.4944} -{"mode": "train", "epoch": 1, "iter": 379, "lr": 0.00025, "memory": 49163, "data_time": 0.08321, "loss_cls_0": 0.72807, "loss_box_0": 1.58571, "loss_cns_0": 0.6328, "loss_yns_0": 0.14444, "loss_cls_1": 0.80275, "loss_box_1": 1.4811, "loss_cns_1": 0.65721, "loss_yns_1": 0.14297, "loss_cls_2": 0.8167, "loss_box_2": 1.45857, "loss_cns_2": 0.66107, "loss_yns_2": 0.14326, "loss_cls_3": 0.81832, "loss_box_3": 1.43804, "loss_cns_3": 0.65971, "loss_yns_3": 0.142, "loss_cls_4": 0.81972, "loss_box_4": 1.44551, "loss_cns_4": 0.65903, "loss_yns_4": 0.14476, "loss_cls_5": 0.82793, "loss_box_5": 1.45134, "loss_cns_5": 0.65988, "loss_yns_5": 0.14285, "loss_cls_dn_0": 0.10963, "loss_box_dn_0": 0.70208, "loss_cls_dn_1": 0.08996, "loss_box_dn_1": 0.64502, "loss_cls_dn_2": 0.08893, "loss_box_dn_2": 0.6242, "loss_cls_dn_3": 0.08905, "loss_box_dn_3": 0.6181, "loss_cls_dn_4": 0.0903, "loss_box_dn_4": 0.62291, "loss_cls_dn_5": 0.091, "loss_box_dn_5": 0.62558, "loss_dense_depth": 0.73643, "loss": 23.59692, "grad_norm": 41.60384, "time": 1.51004} -{"mode": "train", "epoch": 1, "iter": 380, "lr": 0.00025, "memory": 49163, "data_time": 0.08252, "loss_cls_0": 0.74718, "loss_box_0": 1.5741, "loss_cns_0": 0.63781, "loss_yns_0": 0.14355, "loss_cls_1": 0.81289, "loss_box_1": 1.4368, "loss_cns_1": 0.65966, "loss_yns_1": 0.14229, "loss_cls_2": 0.83064, "loss_box_2": 1.39972, "loss_cns_2": 0.66229, "loss_yns_2": 0.1428, "loss_cls_3": 0.83491, "loss_box_3": 1.38829, "loss_cns_3": 0.66127, "loss_yns_3": 0.14129, "loss_cls_4": 0.83504, "loss_box_4": 1.39474, "loss_cns_4": 0.66204, "loss_yns_4": 0.14227, "loss_cls_5": 0.8451, "loss_box_5": 1.39235, "loss_cns_5": 0.66203, "loss_yns_5": 0.14323, "loss_cls_dn_0": 0.11544, "loss_box_dn_0": 0.71111, "loss_cls_dn_1": 0.09201, "loss_box_dn_1": 0.64039, "loss_cls_dn_2": 0.09185, "loss_box_dn_2": 0.6162, "loss_cls_dn_3": 0.09145, "loss_box_dn_3": 0.61369, "loss_cls_dn_4": 0.09288, "loss_box_dn_4": 0.61644, "loss_cls_dn_5": 0.09422, "loss_box_dn_5": 0.61697, "loss_dense_depth": 0.7593, "loss": 23.44425, "grad_norm": 34.157, "time": 1.50581} -{"mode": "train", "epoch": 1, "iter": 381, "lr": 0.00025, "memory": 49163, "data_time": 0.07781, "loss_cls_0": 0.71659, "loss_box_0": 1.55588, "loss_cns_0": 0.63609, "loss_yns_0": 0.1402, "loss_cls_1": 0.79568, "loss_box_1": 1.45642, "loss_cns_1": 0.65976, "loss_yns_1": 0.13968, "loss_cls_2": 0.8108, "loss_box_2": 1.42403, "loss_cns_2": 0.66065, "loss_yns_2": 0.14001, "loss_cls_3": 0.82412, "loss_box_3": 1.40889, "loss_cns_3": 0.66137, "loss_yns_3": 0.13984, "loss_cls_4": 0.83014, "loss_box_4": 1.40513, "loss_cns_4": 0.66176, "loss_yns_4": 0.14062, "loss_cls_5": 0.82524, "loss_box_5": 1.41477, "loss_cns_5": 0.66178, "loss_yns_5": 0.14125, "loss_cls_dn_0": 0.11142, "loss_box_dn_0": 0.71888, "loss_cls_dn_1": 0.09278, "loss_box_dn_1": 0.63088, "loss_cls_dn_2": 0.09293, "loss_box_dn_2": 0.61867, "loss_cls_dn_3": 0.09282, "loss_box_dn_3": 0.61564, "loss_cls_dn_4": 0.09358, "loss_box_dn_4": 0.61575, "loss_cls_dn_5": 0.09486, "loss_box_dn_5": 0.6226, "loss_dense_depth": 0.69907, "loss": 23.35056, "grad_norm": 46.57533, "time": 1.53916} -{"mode": "train", "epoch": 1, "iter": 382, "lr": 0.00025, "memory": 49163, "data_time": 0.07704, "loss_cls_0": 0.71549, "loss_box_0": 1.56172, "loss_cns_0": 0.63813, "loss_yns_0": 0.14096, "loss_cls_1": 0.79784, "loss_box_1": 1.48217, "loss_cns_1": 0.66405, "loss_yns_1": 0.14053, "loss_cls_2": 0.81292, "loss_box_2": 1.42534, "loss_cns_2": 0.66605, "loss_yns_2": 0.14051, "loss_cls_3": 0.82118, "loss_box_3": 1.41285, "loss_cns_3": 0.66858, "loss_yns_3": 0.14035, "loss_cls_4": 0.81174, "loss_box_4": 1.4145, "loss_cns_4": 0.66634, "loss_yns_4": 0.14331, "loss_cls_5": 0.81496, "loss_box_5": 1.41753, "loss_cns_5": 0.6678, "loss_yns_5": 0.14139, "loss_cls_dn_0": 0.11122, "loss_box_dn_0": 0.71161, "loss_cls_dn_1": 0.09356, "loss_box_dn_1": 0.63101, "loss_cls_dn_2": 0.09498, "loss_box_dn_2": 0.61575, "loss_cls_dn_3": 0.09426, "loss_box_dn_3": 0.61111, "loss_cls_dn_4": 0.09359, "loss_box_dn_4": 0.61314, "loss_cls_dn_5": 0.09426, "loss_box_dn_5": 0.62061, "loss_dense_depth": 0.73145, "loss": 23.42277, "grad_norm": 36.00499, "time": 1.55558} -{"mode": "train", "epoch": 1, "iter": 383, "lr": 0.00025, "memory": 49163, "data_time": 0.07917, "loss_cls_0": 0.73574, "loss_box_0": 1.61113, "loss_cns_0": 0.63635, "loss_yns_0": 0.14331, "loss_cls_1": 0.79231, "loss_box_1": 1.45868, "loss_cns_1": 0.66409, "loss_yns_1": 0.14134, "loss_cls_2": 0.80784, "loss_box_2": 1.44294, "loss_cns_2": 0.66318, "loss_yns_2": 0.14165, "loss_cls_3": 0.82034, "loss_box_3": 1.44498, "loss_cns_3": 0.66356, "loss_yns_3": 0.14021, "loss_cls_4": 0.82129, "loss_box_4": 1.44968, "loss_cns_4": 0.66372, "loss_yns_4": 0.14164, "loss_cls_5": 0.81691, "loss_box_5": 1.44202, "loss_cns_5": 0.66282, "loss_yns_5": 0.13891, "loss_cls_dn_0": 0.11011, "loss_box_dn_0": 0.72039, "loss_cls_dn_1": 0.09349, "loss_box_dn_1": 0.62979, "loss_cls_dn_2": 0.09355, "loss_box_dn_2": 0.62288, "loss_cls_dn_3": 0.09321, "loss_box_dn_3": 0.62209, "loss_cls_dn_4": 0.0933, "loss_box_dn_4": 0.62689, "loss_cls_dn_5": 0.09359, "loss_box_dn_5": 0.62878, "loss_dense_depth": 0.69358, "loss": 23.56629, "grad_norm": 50.59716, "time": 1.50483} -{"mode": "train", "epoch": 1, "iter": 384, "lr": 0.00025, "memory": 49163, "data_time": 0.07597, "loss_cls_0": 0.76298, "loss_box_0": 1.62461, "loss_cns_0": 0.63233, "loss_yns_0": 0.14267, "loss_cls_1": 0.80225, "loss_box_1": 1.48126, "loss_cns_1": 0.66221, "loss_yns_1": 0.14192, "loss_cls_2": 0.8142, "loss_box_2": 1.47339, "loss_cns_2": 0.66022, "loss_yns_2": 0.14145, "loss_cls_3": 0.82914, "loss_box_3": 1.4751, "loss_cns_3": 0.66219, "loss_yns_3": 0.14173, "loss_cls_4": 0.84182, "loss_box_4": 1.48054, "loss_cns_4": 0.66278, "loss_yns_4": 0.14393, "loss_cls_5": 0.83723, "loss_box_5": 1.47561, "loss_cns_5": 0.66226, "loss_yns_5": 0.14106, "loss_cls_dn_0": 0.11295, "loss_box_dn_0": 0.73033, "loss_cls_dn_1": 0.09139, "loss_box_dn_1": 0.64025, "loss_cls_dn_2": 0.09056, "loss_box_dn_2": 0.63365, "loss_cls_dn_3": 0.09048, "loss_box_dn_3": 0.63351, "loss_cls_dn_4": 0.09166, "loss_box_dn_4": 0.63692, "loss_cls_dn_5": 0.09174, "loss_box_dn_5": 0.63772, "loss_dense_depth": 0.74579, "loss": 23.9198, "grad_norm": 52.01192, "time": 1.54871} -{"mode": "train", "epoch": 1, "iter": 385, "lr": 0.00025, "memory": 49163, "data_time": 0.07538, "loss_cls_0": 0.72904, "loss_box_0": 1.58813, "loss_cns_0": 0.6402, "loss_yns_0": 0.14296, "loss_cls_1": 0.79743, "loss_box_1": 1.48976, "loss_cns_1": 0.66268, "loss_yns_1": 0.14301, "loss_cls_2": 0.79746, "loss_box_2": 1.45727, "loss_cns_2": 0.66341, "loss_yns_2": 0.14295, "loss_cls_3": 0.81682, "loss_box_3": 1.43821, "loss_cns_3": 0.66361, "loss_yns_3": 0.14372, "loss_cls_4": 0.82639, "loss_box_4": 1.44433, "loss_cns_4": 0.66348, "loss_yns_4": 0.14324, "loss_cls_5": 0.82778, "loss_box_5": 1.45752, "loss_cns_5": 0.66409, "loss_yns_5": 0.1434, "loss_cls_dn_0": 0.11228, "loss_box_dn_0": 0.70993, "loss_cls_dn_1": 0.08935, "loss_box_dn_1": 0.63703, "loss_cls_dn_2": 0.08847, "loss_box_dn_2": 0.62298, "loss_cls_dn_3": 0.08745, "loss_box_dn_3": 0.61874, "loss_cls_dn_4": 0.08986, "loss_box_dn_4": 0.62169, "loss_cls_dn_5": 0.09109, "loss_box_dn_5": 0.62779, "loss_dense_depth": 0.71718, "loss": 23.60074, "grad_norm": 32.02924, "time": 1.50939} -{"mode": "train", "epoch": 1, "iter": 386, "lr": 0.00025, "memory": 49163, "data_time": 0.0774, "loss_cls_0": 0.74852, "loss_box_0": 1.57939, "loss_cns_0": 0.63415, "loss_yns_0": 0.13703, "loss_cls_1": 0.80897, "loss_box_1": 1.49519, "loss_cns_1": 0.65902, "loss_yns_1": 0.13891, "loss_cls_2": 0.81814, "loss_box_2": 1.48125, "loss_cns_2": 0.65736, "loss_yns_2": 0.13643, "loss_cls_3": 0.82358, "loss_box_3": 1.46908, "loss_cns_3": 0.65853, "loss_yns_3": 0.13645, "loss_cls_4": 0.82732, "loss_box_4": 1.47387, "loss_cns_4": 0.65908, "loss_yns_4": 0.13708, "loss_cls_5": 0.83576, "loss_box_5": 1.48286, "loss_cns_5": 0.65896, "loss_yns_5": 0.13695, "loss_cls_dn_0": 0.1142, "loss_box_dn_0": 0.71089, "loss_cls_dn_1": 0.0913, "loss_box_dn_1": 0.63143, "loss_cls_dn_2": 0.09002, "loss_box_dn_2": 0.62034, "loss_cls_dn_3": 0.09056, "loss_box_dn_3": 0.6127, "loss_cls_dn_4": 0.09216, "loss_box_dn_4": 0.61474, "loss_cls_dn_5": 0.0945, "loss_box_dn_5": 0.61948, "loss_dense_depth": 0.75378, "loss": 23.73, "grad_norm": 44.35184, "time": 1.51272} -{"mode": "train", "epoch": 1, "iter": 387, "lr": 0.00025, "memory": 49163, "data_time": 0.08718, "loss_cls_0": 0.73119, "loss_box_0": 1.56708, "loss_cns_0": 0.63035, "loss_yns_0": 0.137, "loss_cls_1": 0.77765, "loss_box_1": 1.49721, "loss_cns_1": 0.65612, "loss_yns_1": 0.13538, "loss_cls_2": 0.7985, "loss_box_2": 1.46732, "loss_cns_2": 0.65251, "loss_yns_2": 0.13511, "loss_cls_3": 0.79667, "loss_box_3": 1.46531, "loss_cns_3": 0.65247, "loss_yns_3": 0.13444, "loss_cls_4": 0.8123, "loss_box_4": 1.44871, "loss_cns_4": 0.65082, "loss_yns_4": 0.13651, "loss_cls_5": 0.81178, "loss_box_5": 1.45081, "loss_cns_5": 0.65254, "loss_yns_5": 0.13414, "loss_cls_dn_0": 0.11186, "loss_box_dn_0": 0.7032, "loss_cls_dn_1": 0.09139, "loss_box_dn_1": 0.62743, "loss_cls_dn_2": 0.09102, "loss_box_dn_2": 0.61671, "loss_cls_dn_3": 0.09087, "loss_box_dn_3": 0.61176, "loss_cls_dn_4": 0.09272, "loss_box_dn_4": 0.61212, "loss_cls_dn_5": 0.09514, "loss_box_dn_5": 0.61091, "loss_dense_depth": 0.74721, "loss": 23.43424, "grad_norm": 29.51794, "time": 1.53376} -{"mode": "train", "epoch": 1, "iter": 388, "lr": 0.00025, "memory": 49163, "data_time": 0.08477, "loss_cls_0": 0.71098, "loss_box_0": 1.57634, "loss_cns_0": 0.63345, "loss_yns_0": 0.1387, "loss_cls_1": 0.75991, "loss_box_1": 1.51531, "loss_cns_1": 0.65409, "loss_yns_1": 0.13603, "loss_cls_2": 0.78154, "loss_box_2": 1.47081, "loss_cns_2": 0.65488, "loss_yns_2": 0.13497, "loss_cls_3": 0.78494, "loss_box_3": 1.47617, "loss_cns_3": 0.65296, "loss_yns_3": 0.13453, "loss_cls_4": 0.79345, "loss_box_4": 1.47353, "loss_cns_4": 0.65013, "loss_yns_4": 0.13661, "loss_cls_5": 0.79713, "loss_box_5": 1.47745, "loss_cns_5": 0.65217, "loss_yns_5": 0.13479, "loss_cls_dn_0": 0.11103, "loss_box_dn_0": 0.7082, "loss_cls_dn_1": 0.08984, "loss_box_dn_1": 0.62736, "loss_cls_dn_2": 0.08905, "loss_box_dn_2": 0.61457, "loss_cls_dn_3": 0.08985, "loss_box_dn_3": 0.61616, "loss_cls_dn_4": 0.09143, "loss_box_dn_4": 0.6195, "loss_cls_dn_5": 0.09269, "loss_box_dn_5": 0.62373, "loss_dense_depth": 0.76381, "loss": 23.46806, "grad_norm": 48.69717, "time": 1.49495} -{"mode": "train", "epoch": 1, "iter": 389, "lr": 0.00025, "memory": 49163, "data_time": 0.1727, "loss_cls_0": 0.72313, "loss_box_0": 1.5812, "loss_cns_0": 0.63947, "loss_yns_0": 0.14173, "loss_cls_1": 0.76968, "loss_box_1": 1.50232, "loss_cns_1": 0.65531, "loss_yns_1": 0.13944, "loss_cls_2": 0.79117, "loss_box_2": 1.46421, "loss_cns_2": 0.65844, "loss_yns_2": 0.13909, "loss_cls_3": 0.78796, "loss_box_3": 1.45072, "loss_cns_3": 0.65543, "loss_yns_3": 0.1386, "loss_cls_4": 0.79129, "loss_box_4": 1.45871, "loss_cns_4": 0.65572, "loss_yns_4": 0.1394, "loss_cls_5": 0.79973, "loss_box_5": 1.47286, "loss_cns_5": 0.65577, "loss_yns_5": 0.14022, "loss_cls_dn_0": 0.10792, "loss_box_dn_0": 0.70382, "loss_cls_dn_1": 0.08932, "loss_box_dn_1": 0.63761, "loss_cls_dn_2": 0.08887, "loss_box_dn_2": 0.62475, "loss_cls_dn_3": 0.08869, "loss_box_dn_3": 0.62374, "loss_cls_dn_4": 0.08996, "loss_box_dn_4": 0.62742, "loss_cls_dn_5": 0.09046, "loss_box_dn_5": 0.63484, "loss_dense_depth": 0.84475, "loss": 23.60373, "grad_norm": 45.93485, "time": 5.02309} -{"mode": "train", "epoch": 1, "iter": 390, "lr": 0.00026, "memory": 49163, "data_time": 0.07269, "loss_cls_0": 0.72266, "loss_box_0": 1.61534, "loss_cns_0": 0.63209, "loss_yns_0": 0.14616, "loss_cls_1": 0.76755, "loss_box_1": 1.4891, "loss_cns_1": 0.65843, "loss_yns_1": 0.14334, "loss_cls_2": 0.79578, "loss_box_2": 1.45004, "loss_cns_2": 0.6594, "loss_yns_2": 0.14211, "loss_cls_3": 0.78566, "loss_box_3": 1.4262, "loss_cns_3": 0.65965, "loss_yns_3": 0.14257, "loss_cls_4": 0.7929, "loss_box_4": 1.421, "loss_cns_4": 0.66093, "loss_yns_4": 0.14303, "loss_cls_5": 0.80156, "loss_box_5": 1.42212, "loss_cns_5": 0.66096, "loss_yns_5": 0.14264, "loss_cls_dn_0": 0.10836, "loss_box_dn_0": 0.71778, "loss_cls_dn_1": 0.09176, "loss_box_dn_1": 0.62869, "loss_cls_dn_2": 0.09038, "loss_box_dn_2": 0.61313, "loss_cls_dn_3": 0.08955, "loss_box_dn_3": 0.60777, "loss_cls_dn_4": 0.09025, "loss_box_dn_4": 0.60773, "loss_cls_dn_5": 0.09131, "loss_box_dn_5": 0.61011, "loss_dense_depth": 0.76024, "loss": 23.3883, "grad_norm": 28.24373, "time": 1.47882} -{"mode": "train", "epoch": 1, "iter": 391, "lr": 0.00026, "memory": 49163, "data_time": 0.07779, "loss_cls_0": 0.70954, "loss_box_0": 1.58181, "loss_cns_0": 0.63514, "loss_yns_0": 0.14045, "loss_cls_1": 0.77239, "loss_box_1": 1.46045, "loss_cns_1": 0.66225, "loss_yns_1": 0.13774, "loss_cls_2": 0.78854, "loss_box_2": 1.43488, "loss_cns_2": 0.66118, "loss_yns_2": 0.13839, "loss_cls_3": 0.79108, "loss_box_3": 1.42461, "loss_cns_3": 0.66246, "loss_yns_3": 0.13936, "loss_cls_4": 0.79769, "loss_box_4": 1.42031, "loss_cns_4": 0.66378, "loss_yns_4": 0.13886, "loss_cls_5": 0.80378, "loss_box_5": 1.41975, "loss_cns_5": 0.66469, "loss_yns_5": 0.13902, "loss_cls_dn_0": 0.11026, "loss_box_dn_0": 0.70591, "loss_cls_dn_1": 0.09499, "loss_box_dn_1": 0.63632, "loss_cls_dn_2": 0.09315, "loss_box_dn_2": 0.62378, "loss_cls_dn_3": 0.09424, "loss_box_dn_3": 0.62151, "loss_cls_dn_4": 0.09604, "loss_box_dn_4": 0.62139, "loss_cls_dn_5": 0.09718, "loss_box_dn_5": 0.62504, "loss_dense_depth": 0.79774, "loss": 23.40572, "grad_norm": 39.01048, "time": 1.48114} -{"mode": "train", "epoch": 1, "iter": 392, "lr": 0.00026, "memory": 49163, "data_time": 0.07625, "loss_cls_0": 0.74039, "loss_box_0": 1.5909, "loss_cns_0": 0.63322, "loss_yns_0": 0.14354, "loss_cls_1": 0.80488, "loss_box_1": 1.48, "loss_cns_1": 0.66517, "loss_yns_1": 0.14191, "loss_cls_2": 0.81749, "loss_box_2": 1.44106, "loss_cns_2": 0.66457, "loss_yns_2": 0.14137, "loss_cls_3": 0.808, "loss_box_3": 1.44108, "loss_cns_3": 0.66539, "loss_yns_3": 0.14152, "loss_cls_4": 0.81652, "loss_box_4": 1.43336, "loss_cns_4": 0.66537, "loss_yns_4": 0.14196, "loss_cls_5": 0.81569, "loss_box_5": 1.43643, "loss_cns_5": 0.66641, "loss_yns_5": 0.142, "loss_cls_dn_0": 0.11514, "loss_box_dn_0": 0.71549, "loss_cls_dn_1": 0.09231, "loss_box_dn_1": 0.63898, "loss_cls_dn_2": 0.09234, "loss_box_dn_2": 0.62067, "loss_cls_dn_3": 0.09316, "loss_box_dn_3": 0.61889, "loss_cls_dn_4": 0.09473, "loss_box_dn_4": 0.61641, "loss_cls_dn_5": 0.09528, "loss_box_dn_5": 0.62016, "loss_dense_depth": 0.79098, "loss": 23.64277, "grad_norm": 29.21275, "time": 1.52442} -{"mode": "train", "epoch": 1, "iter": 393, "lr": 0.00026, "memory": 49163, "data_time": 0.07781, "loss_cls_0": 0.74741, "loss_box_0": 1.61739, "loss_cns_0": 0.63177, "loss_yns_0": 0.14493, "loss_cls_1": 0.8098, "loss_box_1": 1.48133, "loss_cns_1": 0.66141, "loss_yns_1": 0.14222, "loss_cls_2": 0.82225, "loss_box_2": 1.43728, "loss_cns_2": 0.66258, "loss_yns_2": 0.14173, "loss_cls_3": 0.81883, "loss_box_3": 1.43875, "loss_cns_3": 0.66028, "loss_yns_3": 0.14088, "loss_cls_4": 0.82347, "loss_box_4": 1.43138, "loss_cns_4": 0.66078, "loss_yns_4": 0.14169, "loss_cls_5": 0.82641, "loss_box_5": 1.43202, "loss_cns_5": 0.6605, "loss_yns_5": 0.141, "loss_cls_dn_0": 0.11614, "loss_box_dn_0": 0.73475, "loss_cls_dn_1": 0.0922, "loss_box_dn_1": 0.65639, "loss_cls_dn_2": 0.09204, "loss_box_dn_2": 0.63045, "loss_cls_dn_3": 0.09153, "loss_box_dn_3": 0.62699, "loss_cls_dn_4": 0.09181, "loss_box_dn_4": 0.62418, "loss_cls_dn_5": 0.09295, "loss_box_dn_5": 0.62517, "loss_dense_depth": 0.78981, "loss": 23.7405, "grad_norm": 36.02184, "time": 1.52412} -{"mode": "train", "epoch": 1, "iter": 394, "lr": 0.00026, "memory": 49163, "data_time": 0.073, "loss_cls_0": 0.71672, "loss_box_0": 1.55053, "loss_cns_0": 0.63595, "loss_yns_0": 0.14469, "loss_cls_1": 0.79385, "loss_box_1": 1.4342, "loss_cns_1": 0.66141, "loss_yns_1": 0.14204, "loss_cls_2": 0.80201, "loss_box_2": 1.40366, "loss_cns_2": 0.65952, "loss_yns_2": 0.14361, "loss_cls_3": 0.80708, "loss_box_3": 1.38844, "loss_cns_3": 0.65763, "loss_yns_3": 0.14216, "loss_cls_4": 0.8041, "loss_box_4": 1.39021, "loss_cns_4": 0.65894, "loss_yns_4": 0.14404, "loss_cls_5": 0.80606, "loss_box_5": 1.38823, "loss_cns_5": 0.65883, "loss_yns_5": 0.14243, "loss_cls_dn_0": 0.11617, "loss_box_dn_0": 0.70917, "loss_cls_dn_1": 0.09519, "loss_box_dn_1": 0.63226, "loss_cls_dn_2": 0.09356, "loss_box_dn_2": 0.61158, "loss_cls_dn_3": 0.09276, "loss_box_dn_3": 0.60524, "loss_cls_dn_4": 0.09276, "loss_box_dn_4": 0.60328, "loss_cls_dn_5": 0.09353, "loss_box_dn_5": 0.60332, "loss_dense_depth": 0.87435, "loss": 23.29951, "grad_norm": 31.23363, "time": 1.54554} -{"mode": "train", "epoch": 1, "iter": 395, "lr": 0.00026, "memory": 49163, "data_time": 0.07083, "loss_cls_0": 0.72549, "loss_box_0": 1.56957, "loss_cns_0": 0.63707, "loss_yns_0": 0.14811, "loss_cls_1": 0.78609, "loss_box_1": 1.41318, "loss_cns_1": 0.65961, "loss_yns_1": 0.14523, "loss_cls_2": 0.79102, "loss_box_2": 1.39412, "loss_cns_2": 0.65624, "loss_yns_2": 0.14385, "loss_cls_3": 0.78896, "loss_box_3": 1.38065, "loss_cns_3": 0.65629, "loss_yns_3": 0.14428, "loss_cls_4": 0.79453, "loss_box_4": 1.38082, "loss_cns_4": 0.65798, "loss_yns_4": 0.14379, "loss_cls_5": 0.79893, "loss_box_5": 1.37406, "loss_cns_5": 0.65646, "loss_yns_5": 0.14192, "loss_cls_dn_0": 0.11354, "loss_box_dn_0": 0.71528, "loss_cls_dn_1": 0.09365, "loss_box_dn_1": 0.62323, "loss_cls_dn_2": 0.09015, "loss_box_dn_2": 0.61459, "loss_cls_dn_3": 0.08934, "loss_box_dn_3": 0.60974, "loss_cls_dn_4": 0.09051, "loss_box_dn_4": 0.60671, "loss_cls_dn_5": 0.09065, "loss_box_dn_5": 0.60763, "loss_dense_depth": 0.7888, "loss": 23.12205, "grad_norm": 35.33138, "time": 1.4876} -{"mode": "train", "epoch": 1, "iter": 396, "lr": 0.00026, "memory": 49163, "data_time": 0.07177, "loss_cls_0": 0.72352, "loss_box_0": 1.54766, "loss_cns_0": 0.63852, "loss_yns_0": 0.14525, "loss_cls_1": 0.78479, "loss_box_1": 1.40649, "loss_cns_1": 0.65982, "loss_yns_1": 0.14267, "loss_cls_2": 0.7941, "loss_box_2": 1.36329, "loss_cns_2": 0.66095, "loss_yns_2": 0.14178, "loss_cls_3": 0.80034, "loss_box_3": 1.34723, "loss_cns_3": 0.65782, "loss_yns_3": 0.14158, "loss_cls_4": 0.81232, "loss_box_4": 1.34493, "loss_cns_4": 0.66037, "loss_yns_4": 0.14168, "loss_cls_5": 0.81354, "loss_box_5": 1.34302, "loss_cns_5": 0.65995, "loss_yns_5": 0.14037, "loss_cls_dn_0": 0.10903, "loss_box_dn_0": 0.72406, "loss_cls_dn_1": 0.09146, "loss_box_dn_1": 0.62922, "loss_cls_dn_2": 0.08995, "loss_box_dn_2": 0.60915, "loss_cls_dn_3": 0.08963, "loss_box_dn_3": 0.60762, "loss_cls_dn_4": 0.09167, "loss_box_dn_4": 0.60276, "loss_cls_dn_5": 0.09228, "loss_box_dn_5": 0.60462, "loss_dense_depth": 0.82016, "loss": 23.03358, "grad_norm": 37.10076, "time": 1.4961} -{"mode": "train", "epoch": 1, "iter": 397, "lr": 0.00026, "memory": 49163, "data_time": 0.06857, "loss_cls_0": 0.72971, "loss_box_0": 1.57068, "loss_cns_0": 0.63599, "loss_yns_0": 0.14625, "loss_cls_1": 0.7825, "loss_box_1": 1.41673, "loss_cns_1": 0.65929, "loss_yns_1": 0.14419, "loss_cls_2": 0.78985, "loss_box_2": 1.39351, "loss_cns_2": 0.65953, "loss_yns_2": 0.14259, "loss_cls_3": 0.79452, "loss_box_3": 1.39865, "loss_cns_3": 0.66161, "loss_yns_3": 0.14348, "loss_cls_4": 0.80206, "loss_box_4": 1.39513, "loss_cns_4": 0.65962, "loss_yns_4": 0.1438, "loss_cls_5": 0.80717, "loss_box_5": 1.40456, "loss_cns_5": 0.66101, "loss_yns_5": 0.14349, "loss_cls_dn_0": 0.11042, "loss_box_dn_0": 0.70304, "loss_cls_dn_1": 0.09209, "loss_box_dn_1": 0.62333, "loss_cls_dn_2": 0.09168, "loss_box_dn_2": 0.60649, "loss_cls_dn_3": 0.09063, "loss_box_dn_3": 0.60607, "loss_cls_dn_4": 0.09216, "loss_box_dn_4": 0.60452, "loss_cls_dn_5": 0.09419, "loss_box_dn_5": 0.61021, "loss_dense_depth": 0.78426, "loss": 23.19501, "grad_norm": 37.52914, "time": 1.53317} -{"mode": "train", "epoch": 1, "iter": 398, "lr": 0.00026, "memory": 49163, "data_time": 0.07157, "loss_cls_0": 0.75276, "loss_box_0": 1.59486, "loss_cns_0": 0.63301, "loss_yns_0": 0.14737, "loss_cls_1": 0.80175, "loss_box_1": 1.44089, "loss_cns_1": 0.65915, "loss_yns_1": 0.14552, "loss_cls_2": 0.8172, "loss_box_2": 1.43704, "loss_cns_2": 0.66051, "loss_yns_2": 0.14391, "loss_cls_3": 0.82813, "loss_box_3": 1.41981, "loss_cns_3": 0.65877, "loss_yns_3": 0.14455, "loss_cls_4": 0.82607, "loss_box_4": 1.42863, "loss_cns_4": 0.65874, "loss_yns_4": 0.145, "loss_cls_5": 0.83209, "loss_box_5": 1.43058, "loss_cns_5": 0.65912, "loss_yns_5": 0.14386, "loss_cls_dn_0": 0.11742, "loss_box_dn_0": 0.70619, "loss_cls_dn_1": 0.09212, "loss_box_dn_1": 0.6201, "loss_cls_dn_2": 0.09205, "loss_box_dn_2": 0.60944, "loss_cls_dn_3": 0.09024, "loss_box_dn_3": 0.60515, "loss_cls_dn_4": 0.09134, "loss_box_dn_4": 0.60773, "loss_cls_dn_5": 0.09326, "loss_box_dn_5": 0.61248, "loss_dense_depth": 0.79311, "loss": 23.53998, "grad_norm": 37.29315, "time": 1.49548} -{"mode": "train", "epoch": 1, "iter": 399, "lr": 0.00026, "memory": 49163, "data_time": 0.07562, "loss_cls_0": 0.72914, "loss_box_0": 1.56171, "loss_cns_0": 0.63346, "loss_yns_0": 0.1447, "loss_cls_1": 0.79618, "loss_box_1": 1.43819, "loss_cns_1": 0.66329, "loss_yns_1": 0.14439, "loss_cls_2": 0.80821, "loss_box_2": 1.41803, "loss_cns_2": 0.66372, "loss_yns_2": 0.14433, "loss_cls_3": 0.81588, "loss_box_3": 1.39849, "loss_cns_3": 0.66331, "loss_yns_3": 0.14446, "loss_cls_4": 0.8166, "loss_box_4": 1.40022, "loss_cns_4": 0.66506, "loss_yns_4": 0.14567, "loss_cls_5": 0.82171, "loss_box_5": 1.40155, "loss_cns_5": 0.66725, "loss_yns_5": 0.14369, "loss_cls_dn_0": 0.1193, "loss_box_dn_0": 0.70598, "loss_cls_dn_1": 0.0947, "loss_box_dn_1": 0.61665, "loss_cls_dn_2": 0.09453, "loss_box_dn_2": 0.60693, "loss_cls_dn_3": 0.09266, "loss_box_dn_3": 0.60063, "loss_cls_dn_4": 0.09366, "loss_box_dn_4": 0.60254, "loss_cls_dn_5": 0.09388, "loss_box_dn_5": 0.60659, "loss_dense_depth": 0.7928, "loss": 23.35008, "grad_norm": 32.96799, "time": 1.49329} -{"mode": "train", "epoch": 1, "iter": 400, "lr": 0.00026, "memory": 49163, "data_time": 0.07298, "loss_cls_0": 0.73825, "loss_box_0": 1.55408, "loss_cns_0": 0.63146, "loss_yns_0": 0.14294, "loss_cls_1": 0.8035, "loss_box_1": 1.42028, "loss_cns_1": 0.66354, "loss_yns_1": 0.14253, "loss_cls_2": 0.80996, "loss_box_2": 1.40245, "loss_cns_2": 0.66253, "loss_yns_2": 0.14139, "loss_cls_3": 0.82041, "loss_box_3": 1.39294, "loss_cns_3": 0.66315, "loss_yns_3": 0.14182, "loss_cls_4": 0.82339, "loss_box_4": 1.39577, "loss_cns_4": 0.6649, "loss_yns_4": 0.1415, "loss_cls_5": 0.83524, "loss_box_5": 1.39859, "loss_cns_5": 0.66663, "loss_yns_5": 0.14148, "loss_cls_dn_0": 0.11639, "loss_box_dn_0": 0.70774, "loss_cls_dn_1": 0.09142, "loss_box_dn_1": 0.62586, "loss_cls_dn_2": 0.09096, "loss_box_dn_2": 0.61978, "loss_cls_dn_3": 0.09209, "loss_box_dn_3": 0.61353, "loss_cls_dn_4": 0.09123, "loss_box_dn_4": 0.61556, "loss_cls_dn_5": 0.09268, "loss_box_dn_5": 0.61772, "loss_dense_depth": 0.75658, "loss": 23.33027, "grad_norm": 37.5038, "time": 1.48158} -{"mode": "train", "epoch": 1, "iter": 401, "lr": 0.00026, "memory": 49163, "data_time": 0.13526, "loss_cls_0": 0.74978, "loss_box_0": 1.56761, "loss_cns_0": 0.63039, "loss_yns_0": 0.1442, "loss_cls_1": 0.81832, "loss_box_1": 1.44225, "loss_cns_1": 0.65876, "loss_yns_1": 0.14475, "loss_cls_2": 0.83621, "loss_box_2": 1.41143, "loss_cns_2": 0.65858, "loss_yns_2": 0.14338, "loss_cls_3": 0.83572, "loss_box_3": 1.39199, "loss_cns_3": 0.65948, "loss_yns_3": 0.1429, "loss_cls_4": 0.82661, "loss_box_4": 1.402, "loss_cns_4": 0.66025, "loss_yns_4": 0.1441, "loss_cls_5": 0.83081, "loss_box_5": 1.40598, "loss_cns_5": 0.66069, "loss_yns_5": 0.14332, "loss_cls_dn_0": 0.11594, "loss_box_dn_0": 0.71092, "loss_cls_dn_1": 0.09437, "loss_box_dn_1": 0.61131, "loss_cls_dn_2": 0.09304, "loss_box_dn_2": 0.6006, "loss_cls_dn_3": 0.0939, "loss_box_dn_3": 0.59444, "loss_cls_dn_4": 0.09278, "loss_box_dn_4": 0.59537, "loss_cls_dn_5": 0.09471, "loss_box_dn_5": 0.59658, "loss_dense_depth": 0.77793, "loss": 23.38141, "grad_norm": 30.61593, "time": 1.538} -{"mode": "train", "epoch": 1, "iter": 402, "lr": 0.00026, "memory": 49163, "data_time": 0.06866, "loss_cls_0": 0.72581, "loss_box_0": 1.56978, "loss_cns_0": 0.63514, "loss_yns_0": 0.14514, "loss_cls_1": 0.78769, "loss_box_1": 1.44953, "loss_cns_1": 0.65961, "loss_yns_1": 0.14323, "loss_cls_2": 0.80893, "loss_box_2": 1.41332, "loss_cns_2": 0.65981, "loss_yns_2": 0.1422, "loss_cls_3": 0.80997, "loss_box_3": 1.39807, "loss_cns_3": 0.66047, "loss_yns_3": 0.14165, "loss_cls_4": 0.8116, "loss_box_4": 1.4048, "loss_cns_4": 0.66146, "loss_yns_4": 0.14287, "loss_cls_5": 0.81587, "loss_box_5": 1.4127, "loss_cns_5": 0.66194, "loss_yns_5": 0.14204, "loss_cls_dn_0": 0.1106, "loss_box_dn_0": 0.70759, "loss_cls_dn_1": 0.09056, "loss_box_dn_1": 0.62263, "loss_cls_dn_2": 0.08954, "loss_box_dn_2": 0.60733, "loss_cls_dn_3": 0.08934, "loss_box_dn_3": 0.60317, "loss_cls_dn_4": 0.09006, "loss_box_dn_4": 0.60349, "loss_cls_dn_5": 0.09068, "loss_box_dn_5": 0.60704, "loss_dense_depth": 0.73177, "loss": 23.24741, "grad_norm": 32.84783, "time": 1.5378} -{"mode": "train", "epoch": 1, "iter": 403, "lr": 0.00026, "memory": 49163, "data_time": 0.07632, "loss_cls_0": 0.73586, "loss_box_0": 1.58523, "loss_cns_0": 0.63686, "loss_yns_0": 0.14141, "loss_cls_1": 0.79511, "loss_box_1": 1.45271, "loss_cns_1": 0.66005, "loss_yns_1": 0.14048, "loss_cls_2": 0.80814, "loss_box_2": 1.41199, "loss_cns_2": 0.66128, "loss_yns_2": 0.14001, "loss_cls_3": 0.817, "loss_box_3": 1.4071, "loss_cns_3": 0.66014, "loss_yns_3": 0.13947, "loss_cls_4": 0.82287, "loss_box_4": 1.4063, "loss_cns_4": 0.65999, "loss_yns_4": 0.13913, "loss_cls_5": 0.82823, "loss_box_5": 1.40453, "loss_cns_5": 0.66028, "loss_yns_5": 0.13762, "loss_cls_dn_0": 0.11224, "loss_box_dn_0": 0.71907, "loss_cls_dn_1": 0.09094, "loss_box_dn_1": 0.62964, "loss_cls_dn_2": 0.08958, "loss_box_dn_2": 0.61005, "loss_cls_dn_3": 0.08949, "loss_box_dn_3": 0.60765, "loss_cls_dn_4": 0.0905, "loss_box_dn_4": 0.60611, "loss_cls_dn_5": 0.09077, "loss_box_dn_5": 0.60757, "loss_dense_depth": 0.74509, "loss": 23.34051, "grad_norm": 23.03015, "time": 1.49253} -{"mode": "train", "epoch": 1, "iter": 404, "lr": 0.00026, "memory": 49163, "data_time": 0.07734, "loss_cls_0": 0.72019, "loss_box_0": 1.5965, "loss_cns_0": 0.63438, "loss_yns_0": 0.14174, "loss_cls_1": 0.79449, "loss_box_1": 1.45264, "loss_cns_1": 0.65911, "loss_yns_1": 0.13886, "loss_cls_2": 0.81703, "loss_box_2": 1.42734, "loss_cns_2": 0.66054, "loss_yns_2": 0.13837, "loss_cls_3": 0.81983, "loss_box_3": 1.41755, "loss_cns_3": 0.6615, "loss_yns_3": 0.13832, "loss_cls_4": 0.81255, "loss_box_4": 1.41245, "loss_cns_4": 0.66019, "loss_yns_4": 0.13865, "loss_cls_5": 0.81673, "loss_box_5": 1.41137, "loss_cns_5": 0.66048, "loss_yns_5": 0.13841, "loss_cls_dn_0": 0.11427, "loss_box_dn_0": 0.71924, "loss_cls_dn_1": 0.09138, "loss_box_dn_1": 0.61369, "loss_cls_dn_2": 0.09182, "loss_box_dn_2": 0.60017, "loss_cls_dn_3": 0.09288, "loss_box_dn_3": 0.59777, "loss_cls_dn_4": 0.09291, "loss_box_dn_4": 0.59775, "loss_cls_dn_5": 0.09464, "loss_box_dn_5": 0.59786, "loss_dense_depth": 0.72112, "loss": 23.29473, "grad_norm": 31.83437, "time": 1.55387} -{"mode": "train", "epoch": 1, "iter": 405, "lr": 0.00026, "memory": 49163, "data_time": 0.08024, "loss_cls_0": 0.74307, "loss_box_0": 1.60767, "loss_cns_0": 0.63317, "loss_yns_0": 0.14057, "loss_cls_1": 0.79778, "loss_box_1": 1.4684, "loss_cns_1": 0.65857, "loss_yns_1": 0.13805, "loss_cls_2": 0.81688, "loss_box_2": 1.44717, "loss_cns_2": 0.66037, "loss_yns_2": 0.13992, "loss_cls_3": 0.81962, "loss_box_3": 1.43637, "loss_cns_3": 0.66017, "loss_yns_3": 0.14029, "loss_cls_4": 0.82752, "loss_box_4": 1.43, "loss_cns_4": 0.65725, "loss_yns_4": 0.13964, "loss_cls_5": 0.83025, "loss_box_5": 1.43135, "loss_cns_5": 0.65771, "loss_yns_5": 0.1392, "loss_cls_dn_0": 0.11532, "loss_box_dn_0": 0.71937, "loss_cls_dn_1": 0.09377, "loss_box_dn_1": 0.62324, "loss_cls_dn_2": 0.09402, "loss_box_dn_2": 0.60674, "loss_cls_dn_3": 0.094, "loss_box_dn_3": 0.60273, "loss_cls_dn_4": 0.09494, "loss_box_dn_4": 0.60248, "loss_cls_dn_5": 0.09534, "loss_box_dn_5": 0.60379, "loss_dense_depth": 0.74372, "loss": 23.51044, "grad_norm": 31.47028, "time": 1.50895} -{"mode": "train", "epoch": 1, "iter": 406, "lr": 0.00026, "memory": 49163, "data_time": 0.08946, "loss_cls_0": 0.7302, "loss_box_0": 1.58631, "loss_cns_0": 0.63478, "loss_yns_0": 0.13792, "loss_cls_1": 0.78196, "loss_box_1": 1.47282, "loss_cns_1": 0.65922, "loss_yns_1": 0.13672, "loss_cls_2": 0.79454, "loss_box_2": 1.4486, "loss_cns_2": 0.66048, "loss_yns_2": 0.13751, "loss_cls_3": 0.79506, "loss_box_3": 1.44527, "loss_cns_3": 0.6598, "loss_yns_3": 0.13672, "loss_cls_4": 0.80462, "loss_box_4": 1.43091, "loss_cns_4": 0.65908, "loss_yns_4": 0.13595, "loss_cls_5": 0.81007, "loss_box_5": 1.42989, "loss_cns_5": 0.65911, "loss_yns_5": 0.13522, "loss_cls_dn_0": 0.11392, "loss_box_dn_0": 0.71486, "loss_cls_dn_1": 0.09405, "loss_box_dn_1": 0.62647, "loss_cls_dn_2": 0.09348, "loss_box_dn_2": 0.6037, "loss_cls_dn_3": 0.09269, "loss_box_dn_3": 0.6018, "loss_cls_dn_4": 0.09364, "loss_box_dn_4": 0.60052, "loss_cls_dn_5": 0.09468, "loss_box_dn_5": 0.60044, "loss_dense_depth": 0.70941, "loss": 23.32242, "grad_norm": 25.54209, "time": 1.52616} -{"mode": "train", "epoch": 1, "iter": 407, "lr": 0.00026, "memory": 49163, "data_time": 0.07897, "loss_cls_0": 0.73607, "loss_box_0": 1.59289, "loss_cns_0": 0.63803, "loss_yns_0": 0.14004, "loss_cls_1": 0.78319, "loss_box_1": 1.45575, "loss_cns_1": 0.66387, "loss_yns_1": 0.13671, "loss_cls_2": 0.79665, "loss_box_2": 1.43207, "loss_cns_2": 0.66208, "loss_yns_2": 0.13581, "loss_cls_3": 0.80644, "loss_box_3": 1.43265, "loss_cns_3": 0.66143, "loss_yns_3": 0.13595, "loss_cls_4": 0.79975, "loss_box_4": 1.4284, "loss_cns_4": 0.66118, "loss_yns_4": 0.13561, "loss_cls_5": 0.80223, "loss_box_5": 1.41822, "loss_cns_5": 0.66038, "loss_yns_5": 0.13558, "loss_cls_dn_0": 0.11412, "loss_box_dn_0": 0.71109, "loss_cls_dn_1": 0.09259, "loss_box_dn_1": 0.63243, "loss_cls_dn_2": 0.09337, "loss_box_dn_2": 0.61859, "loss_cls_dn_3": 0.09315, "loss_box_dn_3": 0.62044, "loss_cls_dn_4": 0.09511, "loss_box_dn_4": 0.6201, "loss_cls_dn_5": 0.09706, "loss_box_dn_5": 0.62164, "loss_dense_depth": 0.72695, "loss": 23.38761, "grad_norm": 39.41816, "time": 1.51916} -{"mode": "train", "epoch": 1, "iter": 408, "lr": 0.00026, "memory": 49163, "data_time": 0.07812, "loss_cls_0": 0.74171, "loss_box_0": 1.59914, "loss_cns_0": 0.63824, "loss_yns_0": 0.14365, "loss_cls_1": 0.77655, "loss_box_1": 1.45522, "loss_cns_1": 0.66513, "loss_yns_1": 0.14011, "loss_cls_2": 0.78422, "loss_box_2": 1.43563, "loss_cns_2": 0.66629, "loss_yns_2": 0.13967, "loss_cls_3": 0.7933, "loss_box_3": 1.43052, "loss_cns_3": 0.66581, "loss_yns_3": 0.14002, "loss_cls_4": 0.79453, "loss_box_4": 1.42258, "loss_cns_4": 0.66634, "loss_yns_4": 0.14153, "loss_cls_5": 0.79493, "loss_box_5": 1.42198, "loss_cns_5": 0.66566, "loss_yns_5": 0.13806, "loss_cls_dn_0": 0.11673, "loss_box_dn_0": 0.71521, "loss_cls_dn_1": 0.09595, "loss_box_dn_1": 0.64422, "loss_cls_dn_2": 0.09763, "loss_box_dn_2": 0.6336, "loss_cls_dn_3": 0.09674, "loss_box_dn_3": 0.63349, "loss_cls_dn_4": 0.09754, "loss_box_dn_4": 0.63007, "loss_cls_dn_5": 0.09813, "loss_box_dn_5": 0.63277, "loss_dense_depth": 0.74326, "loss": 23.49616, "grad_norm": 37.28313, "time": 1.49369} -{"mode": "train", "epoch": 1, "iter": 409, "lr": 0.00026, "memory": 49163, "data_time": 0.11264, "loss_cls_0": 0.72924, "loss_box_0": 1.58067, "loss_cns_0": 0.64249, "loss_yns_0": 0.14556, "loss_cls_1": 0.77865, "loss_box_1": 1.48149, "loss_cns_1": 0.66526, "loss_yns_1": 0.14192, "loss_cls_2": 0.78324, "loss_box_2": 1.45573, "loss_cns_2": 0.66362, "loss_yns_2": 0.14092, "loss_cls_3": 0.79057, "loss_box_3": 1.43895, "loss_cns_3": 0.6613, "loss_yns_3": 0.14038, "loss_cls_4": 0.79392, "loss_box_4": 1.4502, "loss_cns_4": 0.66279, "loss_yns_4": 0.14367, "loss_cls_5": 0.79484, "loss_box_5": 1.44124, "loss_cns_5": 0.66418, "loss_yns_5": 0.13958, "loss_cls_dn_0": 0.11475, "loss_box_dn_0": 0.71067, "loss_cls_dn_1": 0.099, "loss_box_dn_1": 0.63691, "loss_cls_dn_2": 0.09902, "loss_box_dn_2": 0.61663, "loss_cls_dn_3": 0.09699, "loss_box_dn_3": 0.6113, "loss_cls_dn_4": 0.0993, "loss_box_dn_4": 0.61166, "loss_cls_dn_5": 0.09868, "loss_box_dn_5": 0.61052, "loss_dense_depth": 0.68995, "loss": 23.42577, "grad_norm": 31.94447, "time": 1.52184} -{"mode": "train", "epoch": 1, "iter": 410, "lr": 0.00026, "memory": 49163, "data_time": 0.0791, "loss_cls_0": 0.72845, "loss_box_0": 1.55276, "loss_cns_0": 0.64012, "loss_yns_0": 0.1425, "loss_cls_1": 0.78157, "loss_box_1": 1.42607, "loss_cns_1": 0.66292, "loss_yns_1": 0.1392, "loss_cls_2": 0.77956, "loss_box_2": 1.40107, "loss_cns_2": 0.66247, "loss_yns_2": 0.14032, "loss_cls_3": 0.7865, "loss_box_3": 1.39419, "loss_cns_3": 0.6627, "loss_yns_3": 0.13935, "loss_cls_4": 0.79453, "loss_box_4": 1.3984, "loss_cns_4": 0.66361, "loss_yns_4": 0.14242, "loss_cls_5": 0.79867, "loss_box_5": 1.39118, "loss_cns_5": 0.66647, "loss_yns_5": 0.14123, "loss_cls_dn_0": 0.11043, "loss_box_dn_0": 0.7085, "loss_cls_dn_1": 0.09618, "loss_box_dn_1": 0.62236, "loss_cls_dn_2": 0.09491, "loss_box_dn_2": 0.60029, "loss_cls_dn_3": 0.09467, "loss_box_dn_3": 0.59738, "loss_cls_dn_4": 0.09717, "loss_box_dn_4": 0.59878, "loss_cls_dn_5": 0.09877, "loss_box_dn_5": 0.59723, "loss_dense_depth": 0.6934, "loss": 23.04631, "grad_norm": 34.50429, "time": 1.49001} -{"mode": "train", "epoch": 1, "iter": 411, "lr": 0.00026, "memory": 49163, "data_time": 0.07855, "loss_cls_0": 0.73099, "loss_box_0": 1.5755, "loss_cns_0": 0.6386, "loss_yns_0": 0.14156, "loss_cls_1": 0.78828, "loss_box_1": 1.4319, "loss_cns_1": 0.66239, "loss_yns_1": 0.14006, "loss_cls_2": 0.78896, "loss_box_2": 1.41814, "loss_cns_2": 0.65964, "loss_yns_2": 0.1388, "loss_cls_3": 0.79709, "loss_box_3": 1.40757, "loss_cns_3": 0.65867, "loss_yns_3": 0.13919, "loss_cls_4": 0.81371, "loss_box_4": 1.40467, "loss_cns_4": 0.66225, "loss_yns_4": 0.13892, "loss_cls_5": 0.81319, "loss_box_5": 1.40266, "loss_cns_5": 0.66476, "loss_yns_5": 0.14023, "loss_cls_dn_0": 0.11251, "loss_box_dn_0": 0.70717, "loss_cls_dn_1": 0.09333, "loss_box_dn_1": 0.63686, "loss_cls_dn_2": 0.09349, "loss_box_dn_2": 0.62412, "loss_cls_dn_3": 0.09411, "loss_box_dn_3": 0.62261, "loss_cls_dn_4": 0.09676, "loss_box_dn_4": 0.62081, "loss_cls_dn_5": 0.09772, "loss_box_dn_5": 0.61777, "loss_dense_depth": 0.71541, "loss": 23.29041, "grad_norm": 36.70786, "time": 1.50164} -{"mode": "train", "epoch": 1, "iter": 412, "lr": 0.00026, "memory": 49163, "data_time": 0.07952, "loss_cls_0": 0.70888, "loss_box_0": 1.57619, "loss_cns_0": 0.64136, "loss_yns_0": 0.14042, "loss_cls_1": 0.77662, "loss_box_1": 1.45169, "loss_cns_1": 0.6606, "loss_yns_1": 0.13943, "loss_cls_2": 0.77604, "loss_box_2": 1.42021, "loss_cns_2": 0.65688, "loss_yns_2": 0.13703, "loss_cls_3": 0.77965, "loss_box_3": 1.40771, "loss_cns_3": 0.6562, "loss_yns_3": 0.13614, "loss_cls_4": 0.79298, "loss_box_4": 1.41869, "loss_cns_4": 0.6603, "loss_yns_4": 0.13885, "loss_cls_5": 0.79283, "loss_box_5": 1.42188, "loss_cns_5": 0.66283, "loss_yns_5": 0.13861, "loss_cls_dn_0": 0.1136, "loss_box_dn_0": 0.71277, "loss_cls_dn_1": 0.09309, "loss_box_dn_1": 0.63314, "loss_cls_dn_2": 0.09294, "loss_box_dn_2": 0.6184, "loss_cls_dn_3": 0.09193, "loss_box_dn_3": 0.61242, "loss_cls_dn_4": 0.09138, "loss_box_dn_4": 0.61304, "loss_cls_dn_5": 0.09278, "loss_box_dn_5": 0.61116, "loss_dense_depth": 0.67058, "loss": 23.13925, "grad_norm": 26.11674, "time": 1.51237} -{"mode": "train", "epoch": 1, "iter": 413, "lr": 0.00026, "memory": 49163, "data_time": 0.07745, "loss_cls_0": 0.71652, "loss_box_0": 1.57781, "loss_cns_0": 0.63994, "loss_yns_0": 0.14323, "loss_cls_1": 0.77585, "loss_box_1": 1.45644, "loss_cns_1": 0.65833, "loss_yns_1": 0.14042, "loss_cls_2": 0.78723, "loss_box_2": 1.40935, "loss_cns_2": 0.64979, "loss_yns_2": 0.13835, "loss_cls_3": 0.80067, "loss_box_3": 1.38417, "loss_cns_3": 0.64843, "loss_yns_3": 0.13725, "loss_cls_4": 0.81362, "loss_box_4": 1.40145, "loss_cns_4": 0.65274, "loss_yns_4": 0.13881, "loss_cls_5": 0.81199, "loss_box_5": 1.41398, "loss_cns_5": 0.65506, "loss_yns_5": 0.13906, "loss_cls_dn_0": 0.11156, "loss_box_dn_0": 0.71198, "loss_cls_dn_1": 0.09042, "loss_box_dn_1": 0.62946, "loss_cls_dn_2": 0.0902, "loss_box_dn_2": 0.61263, "loss_cls_dn_3": 0.09011, "loss_box_dn_3": 0.60572, "loss_cls_dn_4": 0.09403, "loss_box_dn_4": 0.60999, "loss_cls_dn_5": 0.09833, "loss_box_dn_5": 0.61245, "loss_dense_depth": 0.68751, "loss": 23.13489, "grad_norm": 34.11992, "time": 1.52023} -{"mode": "train", "epoch": 1, "iter": 414, "lr": 0.00026, "memory": 49163, "data_time": 0.07161, "loss_cls_0": 0.72618, "loss_box_0": 1.60956, "loss_cns_0": 0.64118, "loss_yns_0": 0.14306, "loss_cls_1": 0.7724, "loss_box_1": 1.46802, "loss_cns_1": 0.65615, "loss_yns_1": 0.13822, "loss_cls_2": 0.7928, "loss_box_2": 1.42946, "loss_cns_2": 0.65173, "loss_yns_2": 0.13853, "loss_cls_3": 0.80204, "loss_box_3": 1.40882, "loss_cns_3": 0.64898, "loss_yns_3": 0.13751, "loss_cls_4": 0.81796, "loss_box_4": 1.41499, "loss_cns_4": 0.65137, "loss_yns_4": 0.13809, "loss_cls_5": 0.8076, "loss_box_5": 1.41925, "loss_cns_5": 0.65541, "loss_yns_5": 0.1389, "loss_cls_dn_0": 0.11357, "loss_box_dn_0": 0.71191, "loss_cls_dn_1": 0.0953, "loss_box_dn_1": 0.63699, "loss_cls_dn_2": 0.09482, "loss_box_dn_2": 0.61882, "loss_cls_dn_3": 0.09317, "loss_box_dn_3": 0.61338, "loss_cls_dn_4": 0.0961, "loss_box_dn_4": 0.61752, "loss_cls_dn_5": 0.09862, "loss_box_dn_5": 0.61941, "loss_dense_depth": 0.70372, "loss": 23.32153, "grad_norm": 33.60084, "time": 1.54268} -{"mode": "train", "epoch": 1, "iter": 415, "lr": 0.00027, "memory": 49163, "data_time": 0.07646, "loss_cls_0": 0.73699, "loss_box_0": 1.58383, "loss_cns_0": 0.64332, "loss_yns_0": 0.1439, "loss_cls_1": 0.7737, "loss_box_1": 1.45347, "loss_cns_1": 0.66159, "loss_yns_1": 0.13937, "loss_cls_2": 0.78568, "loss_box_2": 1.42194, "loss_cns_2": 0.66227, "loss_yns_2": 0.14081, "loss_cls_3": 0.78827, "loss_box_3": 1.41024, "loss_cns_3": 0.66118, "loss_yns_3": 0.14015, "loss_cls_4": 0.79452, "loss_box_4": 1.40887, "loss_cns_4": 0.66075, "loss_yns_4": 0.14007, "loss_cls_5": 0.78897, "loss_box_5": 1.41054, "loss_cns_5": 0.66325, "loss_yns_5": 0.13952, "loss_cls_dn_0": 0.11565, "loss_box_dn_0": 0.70378, "loss_cls_dn_1": 0.09549, "loss_box_dn_1": 0.63093, "loss_cls_dn_2": 0.09547, "loss_box_dn_2": 0.61292, "loss_cls_dn_3": 0.09259, "loss_box_dn_3": 0.60991, "loss_cls_dn_4": 0.09439, "loss_box_dn_4": 0.61174, "loss_cls_dn_5": 0.09647, "loss_box_dn_5": 0.61357, "loss_dense_depth": 0.68524, "loss": 23.21136, "grad_norm": 28.53786, "time": 1.50922} -{"mode": "train", "epoch": 1, "iter": 416, "lr": 0.00027, "memory": 49163, "data_time": 0.07473, "loss_cls_0": 0.71406, "loss_box_0": 1.58753, "loss_cns_0": 0.64035, "loss_yns_0": 0.14374, "loss_cls_1": 0.77009, "loss_box_1": 1.42951, "loss_cns_1": 0.66419, "loss_yns_1": 0.13896, "loss_cls_2": 0.79057, "loss_box_2": 1.41253, "loss_cns_2": 0.66462, "loss_yns_2": 0.13985, "loss_cls_3": 0.79349, "loss_box_3": 1.40037, "loss_cns_3": 0.66399, "loss_yns_3": 0.1388, "loss_cls_4": 0.79027, "loss_box_4": 1.40512, "loss_cns_4": 0.66463, "loss_yns_4": 0.14025, "loss_cls_5": 0.78357, "loss_box_5": 1.40977, "loss_cns_5": 0.66551, "loss_yns_5": 0.13938, "loss_cls_dn_0": 0.11698, "loss_box_dn_0": 0.71426, "loss_cls_dn_1": 0.09706, "loss_box_dn_1": 0.62612, "loss_cls_dn_2": 0.09595, "loss_box_dn_2": 0.6107, "loss_cls_dn_3": 0.0945, "loss_box_dn_3": 0.60504, "loss_cls_dn_4": 0.09803, "loss_box_dn_4": 0.60637, "loss_cls_dn_5": 0.10134, "loss_box_dn_5": 0.60848, "loss_dense_depth": 0.68315, "loss": 23.14913, "grad_norm": 36.16254, "time": 1.49351} -{"mode": "train", "epoch": 1, "iter": 417, "lr": 0.00027, "memory": 49163, "data_time": 0.07658, "loss_cls_0": 0.74019, "loss_box_0": 1.5727, "loss_cns_0": 0.63302, "loss_yns_0": 0.14315, "loss_cls_1": 0.77, "loss_box_1": 1.44275, "loss_cns_1": 0.66117, "loss_yns_1": 0.14084, "loss_cls_2": 0.78795, "loss_box_2": 1.42835, "loss_cns_2": 0.66293, "loss_yns_2": 0.14134, "loss_cls_3": 0.78833, "loss_box_3": 1.41408, "loss_cns_3": 0.66187, "loss_yns_3": 0.14173, "loss_cls_4": 0.79012, "loss_box_4": 1.41645, "loss_cns_4": 0.66179, "loss_yns_4": 0.1425, "loss_cls_5": 0.79374, "loss_box_5": 1.41801, "loss_cns_5": 0.66233, "loss_yns_5": 0.14397, "loss_cls_dn_0": 0.11712, "loss_box_dn_0": 0.71959, "loss_cls_dn_1": 0.10013, "loss_box_dn_1": 0.63391, "loss_cls_dn_2": 0.09837, "loss_box_dn_2": 0.62013, "loss_cls_dn_3": 0.09751, "loss_box_dn_3": 0.61404, "loss_cls_dn_4": 0.10034, "loss_box_dn_4": 0.61698, "loss_cls_dn_5": 0.10301, "loss_box_dn_5": 0.61829, "loss_dense_depth": 0.71302, "loss": 23.31178, "grad_norm": 34.81203, "time": 1.53541} -{"mode": "train", "epoch": 1, "iter": 418, "lr": 0.00027, "memory": 49163, "data_time": 0.07664, "loss_cls_0": 0.73336, "loss_box_0": 1.573, "loss_cns_0": 0.6321, "loss_yns_0": 0.14209, "loss_cls_1": 0.76726, "loss_box_1": 1.43043, "loss_cns_1": 0.65709, "loss_yns_1": 0.14241, "loss_cls_2": 0.77324, "loss_box_2": 1.41896, "loss_cns_2": 0.66127, "loss_yns_2": 0.14353, "loss_cls_3": 0.77508, "loss_box_3": 1.40231, "loss_cns_3": 0.65913, "loss_yns_3": 0.14274, "loss_cls_4": 0.77888, "loss_box_4": 1.39967, "loss_cns_4": 0.65874, "loss_yns_4": 0.14357, "loss_cls_5": 0.77726, "loss_box_5": 1.40209, "loss_cns_5": 0.65981, "loss_yns_5": 0.14464, "loss_cls_dn_0": 0.11076, "loss_box_dn_0": 0.70236, "loss_cls_dn_1": 0.09826, "loss_box_dn_1": 0.63213, "loss_cls_dn_2": 0.09634, "loss_box_dn_2": 0.61834, "loss_cls_dn_3": 0.09654, "loss_box_dn_3": 0.61261, "loss_cls_dn_4": 0.09616, "loss_box_dn_4": 0.6131, "loss_cls_dn_5": 0.09742, "loss_box_dn_5": 0.61438, "loss_dense_depth": 0.68782, "loss": 23.09486, "grad_norm": 28.936, "time": 1.48814} -{"mode": "train", "epoch": 1, "iter": 419, "lr": 0.00027, "memory": 49163, "data_time": 0.07705, "loss_cls_0": 0.77357, "loss_box_0": 1.64576, "loss_cns_0": 0.62943, "loss_yns_0": 0.14341, "loss_cls_1": 0.80148, "loss_box_1": 1.47565, "loss_cns_1": 0.65959, "loss_yns_1": 0.14151, "loss_cls_2": 0.8089, "loss_box_2": 1.44817, "loss_cns_2": 0.66318, "loss_yns_2": 0.14274, "loss_cls_3": 0.81615, "loss_box_3": 1.4446, "loss_cns_3": 0.66312, "loss_yns_3": 0.14182, "loss_cls_4": 0.81677, "loss_box_4": 1.44066, "loss_cns_4": 0.66432, "loss_yns_4": 0.14327, "loss_cls_5": 0.82087, "loss_box_5": 1.44565, "loss_cns_5": 0.6661, "loss_yns_5": 0.14218, "loss_cls_dn_0": 0.11637, "loss_box_dn_0": 0.71122, "loss_cls_dn_1": 0.09634, "loss_box_dn_1": 0.63862, "loss_cls_dn_2": 0.09615, "loss_box_dn_2": 0.62013, "loss_cls_dn_3": 0.097, "loss_box_dn_3": 0.61704, "loss_cls_dn_4": 0.09524, "loss_box_dn_4": 0.61468, "loss_cls_dn_5": 0.0952, "loss_box_dn_5": 0.61782, "loss_dense_depth": 0.72389, "loss": 23.67859, "grad_norm": 27.59499, "time": 1.49858} -{"mode": "train", "epoch": 1, "iter": 420, "lr": 0.00027, "memory": 49163, "data_time": 0.08154, "loss_cls_0": 0.73929, "loss_box_0": 1.64127, "loss_cns_0": 0.63548, "loss_yns_0": 0.14307, "loss_cls_1": 0.78583, "loss_box_1": 1.47793, "loss_cns_1": 0.6598, "loss_yns_1": 0.14051, "loss_cls_2": 0.80091, "loss_box_2": 1.44784, "loss_cns_2": 0.66127, "loss_yns_2": 0.14051, "loss_cls_3": 0.81099, "loss_box_3": 1.44448, "loss_cns_3": 0.66163, "loss_yns_3": 0.14094, "loss_cls_4": 0.81225, "loss_box_4": 1.43687, "loss_cns_4": 0.66438, "loss_yns_4": 0.14064, "loss_cls_5": 0.80897, "loss_box_5": 1.42966, "loss_cns_5": 0.66447, "loss_yns_5": 0.14097, "loss_cls_dn_0": 0.11645, "loss_box_dn_0": 0.71181, "loss_cls_dn_1": 0.0955, "loss_box_dn_1": 0.64135, "loss_cls_dn_2": 0.0962, "loss_box_dn_2": 0.62271, "loss_cls_dn_3": 0.09566, "loss_box_dn_3": 0.62026, "loss_cls_dn_4": 0.09433, "loss_box_dn_4": 0.61757, "loss_cls_dn_5": 0.09529, "loss_box_dn_5": 0.6179, "loss_dense_depth": 0.70738, "loss": 23.56237, "grad_norm": 32.26723, "time": 1.51143} -{"mode": "train", "epoch": 1, "iter": 421, "lr": 0.00027, "memory": 49163, "data_time": 0.13903, "loss_cls_0": 0.75947, "loss_box_0": 1.67326, "loss_cns_0": 0.63641, "loss_yns_0": 0.1402, "loss_cls_1": 0.79093, "loss_box_1": 1.48805, "loss_cns_1": 0.66054, "loss_yns_1": 0.13828, "loss_cls_2": 0.79613, "loss_box_2": 1.46458, "loss_cns_2": 0.66088, "loss_yns_2": 0.13909, "loss_cls_3": 0.80055, "loss_box_3": 1.45669, "loss_cns_3": 0.66176, "loss_yns_3": 0.13947, "loss_cls_4": 0.80992, "loss_box_4": 1.44703, "loss_cns_4": 0.66237, "loss_yns_4": 0.13791, "loss_cls_5": 0.81429, "loss_box_5": 1.448, "loss_cns_5": 0.66202, "loss_yns_5": 0.13854, "loss_cls_dn_0": 0.12401, "loss_box_dn_0": 0.70802, "loss_cls_dn_1": 0.09269, "loss_box_dn_1": 0.6428, "loss_cls_dn_2": 0.09259, "loss_box_dn_2": 0.63089, "loss_cls_dn_3": 0.09053, "loss_box_dn_3": 0.62813, "loss_cls_dn_4": 0.09002, "loss_box_dn_4": 0.625, "loss_cls_dn_5": 0.09235, "loss_box_dn_5": 0.62512, "loss_dense_depth": 0.6922, "loss": 23.66072, "grad_norm": 22.94078, "time": 1.55408} -{"mode": "train", "epoch": 1, "iter": 422, "lr": 0.00027, "memory": 49163, "data_time": 0.07865, "loss_cls_0": 0.71825, "loss_box_0": 1.65113, "loss_cns_0": 0.63731, "loss_yns_0": 0.14141, "loss_cls_1": 0.77317, "loss_box_1": 1.47804, "loss_cns_1": 0.66116, "loss_yns_1": 0.141, "loss_cls_2": 0.77764, "loss_box_2": 1.44238, "loss_cns_2": 0.66151, "loss_yns_2": 0.14062, "loss_cls_3": 0.7856, "loss_box_3": 1.4382, "loss_cns_3": 0.66129, "loss_yns_3": 0.13993, "loss_cls_4": 0.78787, "loss_box_4": 1.42849, "loss_cns_4": 0.66238, "loss_yns_4": 0.14143, "loss_cls_5": 0.80303, "loss_box_5": 1.42529, "loss_cns_5": 0.66165, "loss_yns_5": 0.14002, "loss_cls_dn_0": 0.11595, "loss_box_dn_0": 0.71467, "loss_cls_dn_1": 0.09331, "loss_box_dn_1": 0.63015, "loss_cls_dn_2": 0.09158, "loss_box_dn_2": 0.60933, "loss_cls_dn_3": 0.09018, "loss_box_dn_3": 0.60678, "loss_cls_dn_4": 0.09043, "loss_box_dn_4": 0.60433, "loss_cls_dn_5": 0.09249, "loss_box_dn_5": 0.60485, "loss_dense_depth": 0.67519, "loss": 23.31805, "grad_norm": 27.57657, "time": 1.55147} -{"mode": "train", "epoch": 1, "iter": 423, "lr": 0.00027, "memory": 49163, "data_time": 0.08165, "loss_cls_0": 0.71123, "loss_box_0": 1.60492, "loss_cns_0": 0.63938, "loss_yns_0": 0.14035, "loss_cls_1": 0.77009, "loss_box_1": 1.43196, "loss_cns_1": 0.66412, "loss_yns_1": 0.13988, "loss_cls_2": 0.77377, "loss_box_2": 1.40438, "loss_cns_2": 0.66349, "loss_yns_2": 0.14043, "loss_cls_3": 0.77527, "loss_box_3": 1.38956, "loss_cns_3": 0.66166, "loss_yns_3": 0.13922, "loss_cls_4": 0.77244, "loss_box_4": 1.38679, "loss_cns_4": 0.66299, "loss_yns_4": 0.14278, "loss_cls_5": 0.77989, "loss_box_5": 1.37888, "loss_cns_5": 0.66203, "loss_yns_5": 0.13898, "loss_cls_dn_0": 0.11185, "loss_box_dn_0": 0.71554, "loss_cls_dn_1": 0.09287, "loss_box_dn_1": 0.61309, "loss_cls_dn_2": 0.09282, "loss_box_dn_2": 0.59539, "loss_cls_dn_3": 0.09185, "loss_box_dn_3": 0.59196, "loss_cls_dn_4": 0.09117, "loss_box_dn_4": 0.58903, "loss_cls_dn_5": 0.09102, "loss_box_dn_5": 0.58985, "loss_dense_depth": 0.66686, "loss": 22.9078, "grad_norm": 25.81797, "time": 1.49772} -{"mode": "train", "epoch": 1, "iter": 424, "lr": 0.00027, "memory": 49163, "data_time": 0.0775, "loss_cls_0": 0.75667, "loss_box_0": 1.62411, "loss_cns_0": 0.63462, "loss_yns_0": 0.14207, "loss_cls_1": 0.78522, "loss_box_1": 1.45747, "loss_cns_1": 0.66148, "loss_yns_1": 0.13958, "loss_cls_2": 0.78379, "loss_box_2": 1.43276, "loss_cns_2": 0.66093, "loss_yns_2": 0.14021, "loss_cls_3": 0.78451, "loss_box_3": 1.42423, "loss_cns_3": 0.66066, "loss_yns_3": 0.13787, "loss_cls_4": 0.795, "loss_box_4": 1.42121, "loss_cns_4": 0.66164, "loss_yns_4": 0.13882, "loss_cls_5": 0.79733, "loss_box_5": 1.42445, "loss_cns_5": 0.66294, "loss_yns_5": 0.13782, "loss_cls_dn_0": 0.12594, "loss_box_dn_0": 0.70942, "loss_cls_dn_1": 0.09468, "loss_box_dn_1": 0.62316, "loss_cls_dn_2": 0.09455, "loss_box_dn_2": 0.60462, "loss_cls_dn_3": 0.09514, "loss_box_dn_3": 0.60039, "loss_cls_dn_4": 0.09441, "loss_box_dn_4": 0.59965, "loss_cls_dn_5": 0.09462, "loss_box_dn_5": 0.60195, "loss_dense_depth": 0.69929, "loss": 23.30321, "grad_norm": 27.68864, "time": 1.54013} diff --git a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20260114_183056.log b/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20260114_183056.log deleted file mode 100644 index 4f3a4a05ce0e05e86fffed97c39a492a8388fc91..0000000000000000000000000000000000000000 --- a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20260114_183056.log +++ /dev/null @@ -1,465 +0,0 @@ -2026-01-14 18:30:56,586 - mmdet - INFO - Environment info: ------------------------------------------------------------- -sys.platform: linux -Python: 3.10.12 (main, Aug 15 2025, 14:32:43) [GCC 11.4.0] -CUDA available: True -GPU 0,1,2,3,4,5,6,7: BW151 -CUDA_HOME: /opt/dtk -NVCC: Not Available -GCC: x86_64-linux-gnu-gcc (Ubuntu 11.4.0-1ubuntu1~22.04) 11.4.0 -PyTorch: 2.5.1 -PyTorch compiling details: PyTorch built with: - - GCC 10.3 - - C++ Version: 201703 - - Intel(R) Math Kernel Library Version 2020.0.4 Product Build 20200917 for Intel(R) 64 architecture applications - - OpenMP 201511 (a.k.a. OpenMP 4.5) - - LAPACK is enabled (usually provided by MKL) - - NNPACK is enabled - - CPU capability usage: AVX512 - - HIP Runtime 6.3.25521 - - MIOpen 2.18.0 - - Magma 2.8.0 - - Build settings: BLAS_INFO=mkl, BUILD_TYPE=Release, CXX_COMPILER=/opt/rh/gcc-toolset-10/root/usr/bin/c++, CXX_FLAGS= -D_GLIBCXX_USE_CXX11_ABI=1 -fvisibility-inlines-hidden -DUSE_PTHREADPOOL -DNDEBUG -DUSE_KINETO -DLIBKINETO_NOCUPTI -DLIBKINETO_NOXPUPTI=ON -DUSE_FBGEMM -DUSE_PYTORCH_QNNPACK -DUSE_XNNPACK -DSYMBOLICATE_MOBILE_DEBUG_HANDLE -O2 -fPIC -Wall -Wextra -Werror=return-type -Werror=non-virtual-dtor -Werror=bool-operation -Wnarrowing -Wno-missing-field-initializers -Wno-type-limits -Wno-array-bounds -Wno-unknown-pragmas -Wno-unused-parameter -Wno-strict-overflow -Wno-strict-aliasing -Wno-stringop-overflow -Wsuggest-override -Wno-psabi -Wno-error=old-style-cast -Wno-missing-braces -fdiagnostics-color=always -faligned-new -Wno-unused-but-set-variable -Wno-maybe-uninitialized -fno-math-errno -fno-trapping-math -Werror=format -Wno-stringop-overflow, FORCE_FALLBACK_CUDA_MPI=1, LAPACK_INFO=mkl, PERF_WITH_AVX=1, PERF_WITH_AVX2=1, TORCH_VERSION=2.5.1, USE_CUDA=0, USE_CUDNN=OFF, USE_CUSPARSELT=OFF, USE_EXCEPTION_PTR=1, USE_GFLAGS=1, USE_GLOG=1, USE_GLOO=1, USE_MKL=ON, USE_MKLDNN=0, USE_MPI=1, USE_NCCL=1, USE_NNPACK=ON, USE_OPENMP=1, USE_ROCM=ON, USE_ROCM_KERNEL_ASSERT=OFF, - -TorchVision: 0.20.1 -OpenCV: 4.12.0 -MMCV: 1.6.1 -MMCV Compiler: GCC 10.3 -MMCV CUDA Compiler: rocm not available -MMDetection: 2.25.1+c41df4b ------------------------------------------------------------- - -2026-01-14 18:30:57,335 - mmdet - INFO - Distributed training: True -2026-01-14 18:30:58,066 - mmdet - INFO - Config: -plugin = True -plugin_dir = 'projects/mmdet3d_plugin/' -dist_params = dict(backend='nccl') -log_level = 'INFO' -work_dir = './work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704' -total_batch_size = 160 -num_gpus = 8 -batch_size = 20 -num_iters_per_epoch = 175 -num_epochs = 100 -checkpoint_epoch_interval = 20 -checkpoint_config = dict(interval=3500) -log_config = dict( - interval=1, - hooks=[ - dict(type='TextLoggerHook', by_epoch=False), - dict(type='TensorboardLoggerHook') - ]) -load_from = None -resume_from = None -workflow = [('train', 1)] -fp16 = dict(loss_scale=32.0) -input_shape = (704, 256) -tracking_test = True -tracking_threshold = 0.2 -class_names = [ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier', - 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' -] -num_classes = 10 -embed_dims = 256 -num_groups = 8 -num_decoder = 6 -num_single_frame_decoder = 1 -use_deformable_func = True -strides = [4, 8, 16, 32] -num_levels = 4 -num_depth_layers = 3 -drop_out = 0.1 -temporal = True -decouple_attn = True -with_quality_estimation = True -model = dict( - type='Sparse4D', - use_grid_mask=True, - use_deformable_func=True, - img_backbone=dict( - type='ResNet', - depth=50, - num_stages=4, - frozen_stages=-1, - norm_eval=False, - style='pytorch', - with_cp=True, - out_indices=(0, 1, 2, 3), - norm_cfg=dict(type='BN', requires_grad=True), - pretrained='ckpt/resnet50-19c8e357.pth'), - img_neck=dict( - type='FPN', - num_outs=4, - start_level=0, - out_channels=256, - add_extra_convs='on_output', - relu_before_extra_convs=True, - in_channels=[256, 512, 1024, 2048]), - depth_branch=dict( - type='DenseDepthNet', - embed_dims=256, - num_depth_layers=3, - loss_weight=0.2), - head=dict( - type='Sparse4DHead', - cls_threshold_to_reg=0.05, - decouple_attn=True, - instance_bank=dict( - type='InstanceBank', - num_anchor=900, - embed_dims=256, - anchor='nuscenes_kmeans900.npy', - anchor_handler=dict(type='SparseBox3DKeyPointsGenerator'), - num_temp_instances=600, - confidence_decay=0.6, - feat_grad=False), - anchor_encoder=dict( - type='SparseBox3DEncoder', - vel_dims=3, - embed_dims=[128, 32, 32, 64], - mode='cat', - output_fc=False, - in_loops=1, - out_loops=4), - num_single_frame_decoder=1, - operation_order=[ - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine' - ], - temp_graph_model=dict( - type='MultiheadAttention', - embed_dims=512, - num_heads=8, - batch_first=True, - dropout=0.1), - graph_model=dict( - type='MultiheadAttention', - embed_dims=512, - num_heads=8, - batch_first=True, - dropout=0.1), - norm_layer=dict(type='LN', normalized_shape=256), - ffn=dict( - type='AsymmetricFFN', - in_channels=512, - pre_norm=dict(type='LN'), - embed_dims=256, - feedforward_channels=1024, - num_fcs=2, - ffn_drop=0.1, - act_cfg=dict(type='ReLU', inplace=True)), - deformable_model=dict( - type='DeformableFeatureAggregation', - embed_dims=256, - num_groups=8, - num_levels=4, - num_cams=6, - attn_drop=0.15, - use_deformable_func=True, - use_camera_embed=True, - residual_mode='cat', - kps_generator=dict( - type='SparseBox3DKeyPointsGenerator', - num_learnable_pts=6, - fix_scale=[[0, 0, 0], [0.45, 0, 0], [-0.45, 0, 0], - [0, 0.45, 0], [0, -0.45, 0], [0, 0, 0.45], - [0, 0, -0.45]])), - refine_layer=dict( - type='SparseBox3DRefinementModule', - embed_dims=256, - num_cls=10, - refine_yaw=True, - with_quality_estimation=True), - sampler=dict( - type='SparseBox3DTarget', - num_dn_groups=5, - num_temp_dn_groups=3, - dn_noise_scale=[2.0, 2.0, 2.0, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5], - max_dn_gt=32, - add_neg_dn=True, - cls_weight=2.0, - box_weight=0.25, - reg_weights=[2.0, 2.0, 2.0, 0.5, 0.5, 0.5, 0.0, 0.0, 0.0, 0.0], - cls_wise_reg_weights=dict( - {9: [2.0, 2.0, 2.0, 1.0, 1.0, 1.0, 0.0, 0.0, 1.0, 1.0]})), - loss_cls=dict( - type='FocalLoss', - use_sigmoid=True, - gamma=2.0, - alpha=0.25, - loss_weight=2.0), - loss_reg=dict( - type='SparseBox3DLoss', - loss_box=dict(type='L1Loss', loss_weight=0.25), - loss_centerness=dict(type='CrossEntropyLoss', use_sigmoid=True), - loss_yawness=dict(type='GaussianFocalLoss'), - cls_allow_reverse=[5]), - decoder=dict(type='SparseBox3DDecoder'), - reg_weights=[2.0, 2.0, 2.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0])) -dataset_type = 'NuScenes3DDetTrackDataset' -data_root = 'data/nuscenes/' -anno_root = 'data/nuscenes_anno_pkls/' -file_client_args = dict(backend='disk') -img_norm_cfg = dict( - mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375], to_rgb=True) -train_pipeline = [ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict( - type='LoadPointsFromFile', - coord_type='LIDAR', - load_dim=5, - use_dim=5, - file_client_args=dict(backend='disk')), - dict(type='ResizeCropFlipImage'), - dict(type='MultiScaleDepthMapGenerator', downsample=[4, 8, 16]), - dict(type='BBoxRotation'), - dict(type='PhotoMetricDistortionMultiViewImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict( - type='CircleObjectRangeFilter', - class_dist_thred=[55, 55, 55, 55, 55, 55, 55, 55, 55, 55]), - dict( - type='InstanceNameFilter', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ]), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=[ - 'img', 'timestamp', 'projection_mat', 'image_wh', 'gt_depth', - 'focal', 'gt_bboxes_3d', 'gt_labels_3d' - ], - meta_keys=['T_global', 'T_global_inv', 'timestamp', 'instance_id']) -] -test_pipeline = [ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='ResizeCropFlipImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=['img', 'timestamp', 'projection_mat', 'image_wh'], - meta_keys=['T_global', 'T_global_inv', 'timestamp']) -] -input_modality = dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False) -data_basic_config = dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier', - 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval') -data_aug_conf = dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]) -data = dict( - samples_per_gpu=20, - workers_per_gpu=20, - train=dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval', - ann_file='data/nuscenes_anno_pkls/nuscenes_infos_train.pkl', - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict( - type='LoadPointsFromFile', - coord_type='LIDAR', - load_dim=5, - use_dim=5, - file_client_args=dict(backend='disk')), - dict(type='ResizeCropFlipImage'), - dict(type='MultiScaleDepthMapGenerator', downsample=[4, 8, 16]), - dict(type='BBoxRotation'), - dict(type='PhotoMetricDistortionMultiViewImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict( - type='CircleObjectRangeFilter', - class_dist_thred=[55, 55, 55, 55, 55, 55, 55, 55, 55, 55]), - dict( - type='InstanceNameFilter', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', - 'traffic_cone' - ]), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=[ - 'img', 'timestamp', 'projection_mat', 'image_wh', - 'gt_depth', 'focal', 'gt_bboxes_3d', 'gt_labels_3d' - ], - meta_keys=[ - 'T_global', 'T_global_inv', 'timestamp', 'instance_id' - ]) - ], - test_mode=False, - data_aug_conf=dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]), - with_seq_flag=True, - sequences_split_num=2, - keep_consistent_seq_aug=True), - val=dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval', - ann_file='data/nuscenes_anno_pkls/nuscenes_infos_val.pkl', - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='ResizeCropFlipImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=['img', 'timestamp', 'projection_mat', 'image_wh'], - meta_keys=['T_global', 'T_global_inv', 'timestamp']) - ], - data_aug_conf=dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]), - test_mode=True, - tracking=True, - tracking_threshold=0.2), - test=dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval', - ann_file='data/nuscenes_anno_pkls/nuscenes_infos_val.pkl', - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='ResizeCropFlipImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=['img', 'timestamp', 'projection_mat', 'image_wh'], - meta_keys=['T_global', 'T_global_inv', 'timestamp']) - ], - data_aug_conf=dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]), - test_mode=True, - tracking=True, - tracking_threshold=0.2)) -optimizer = dict( - type='Miopen_AdamW', - lr=0.0006, - weight_decay=0.001, - paramwise_cfg=dict(custom_keys=dict(img_backbone=dict(lr_mult=0.5)))) -optimizer_config = dict(grad_clip=dict(max_norm=25, norm_type=2)) -lr_config = dict( - policy='CosineAnnealing', - warmup='linear', - warmup_iters=500, - warmup_ratio=0.3333333333333333, - min_lr_ratio=0.001) -runner = dict(type='IterBasedRunner', max_iters=17500) -vis_pipeline = [ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='Collect', keys=['img'], meta_keys=['timestamp', 'lidar2img']) -] -evaluation = dict( - interval=3500, - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict( - type='Collect', keys=['img'], meta_keys=['timestamp', 'lidar2img']) - ]) -gpu_ids = range(0, 8) - -2026-01-14 18:30:58,067 - mmdet - INFO - Set random seed to 0, deterministic: False diff --git a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20260114_201232.log b/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20260114_201232.log deleted file mode 100644 index 6190161dbc055d05144557954ea7452f10f4a088..0000000000000000000000000000000000000000 --- a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20260114_201232.log +++ /dev/null @@ -1,469 +0,0 @@ -2026-01-14 20:12:32,746 - mmdet - INFO - Environment info: ------------------------------------------------------------- -sys.platform: linux -Python: 3.10.12 (main, Aug 15 2025, 14:32:43) [GCC 11.4.0] -CUDA available: True -GPU 0,1,2,3,4,5,6,7: BW151 -CUDA_HOME: /opt/dtk/cuda/cuda -NVCC: Cuda compilation tools, release 12.6, V12.6.77 -clang version 17.0.0 -Target: x86_64-unknown-linux-gnu -Thread model: posix -InstalledDir: /opt/dtk-25.04.4/llvm/bi -GCC: x86_64-linux-gnu-gcc (Ubuntu 11.4.0-1ubuntu1~22.04) 11.4.0 -PyTorch: 2.5.1 -PyTorch compiling details: PyTorch built with: - - GCC 10.3 - - C++ Version: 201703 - - Intel(R) Math Kernel Library Version 2020.0.4 Product Build 20200917 for Intel(R) 64 architecture applications - - OpenMP 201511 (a.k.a. OpenMP 4.5) - - LAPACK is enabled (usually provided by MKL) - - NNPACK is enabled - - CPU capability usage: AVX512 - - HIP Runtime 6.3.25521 - - MIOpen 2.18.0 - - Magma 2.8.0 - - Build settings: BLAS_INFO=mkl, BUILD_TYPE=Release, CXX_COMPILER=/opt/rh/gcc-toolset-10/root/usr/bin/c++, CXX_FLAGS= -D_GLIBCXX_USE_CXX11_ABI=1 -fvisibility-inlines-hidden -DUSE_PTHREADPOOL -DNDEBUG -DUSE_KINETO -DLIBKINETO_NOCUPTI -DLIBKINETO_NOXPUPTI=ON -DUSE_FBGEMM -DUSE_PYTORCH_QNNPACK -DUSE_XNNPACK -DSYMBOLICATE_MOBILE_DEBUG_HANDLE -O2 -fPIC -Wall -Wextra -Werror=return-type -Werror=non-virtual-dtor -Werror=bool-operation -Wnarrowing -Wno-missing-field-initializers -Wno-type-limits -Wno-array-bounds -Wno-unknown-pragmas -Wno-unused-parameter -Wno-strict-overflow -Wno-strict-aliasing -Wno-stringop-overflow -Wsuggest-override -Wno-psabi -Wno-error=old-style-cast -Wno-missing-braces -fdiagnostics-color=always -faligned-new -Wno-unused-but-set-variable -Wno-maybe-uninitialized -fno-math-errno -fno-trapping-math -Werror=format -Wno-stringop-overflow, FORCE_FALLBACK_CUDA_MPI=1, LAPACK_INFO=mkl, PERF_WITH_AVX=1, PERF_WITH_AVX2=1, TORCH_VERSION=2.5.1, USE_CUDA=0, USE_CUDNN=OFF, USE_CUSPARSELT=OFF, USE_EXCEPTION_PTR=1, USE_GFLAGS=1, USE_GLOG=1, USE_GLOO=1, USE_MKL=ON, USE_MKLDNN=0, USE_MPI=1, USE_NCCL=1, USE_NNPACK=ON, USE_OPENMP=1, USE_ROCM=ON, USE_ROCM_KERNEL_ASSERT=OFF, - -TorchVision: 0.20.1 -OpenCV: 4.12.0 -MMCV: 1.6.1 -MMCV Compiler: GCC 10.3 -MMCV CUDA Compiler: rocm not available -MMDetection: 2.25.1+c41df4b ------------------------------------------------------------- - -2026-01-14 20:12:33,691 - mmdet - INFO - Distributed training: True -2026-01-14 20:12:34,421 - mmdet - INFO - Config: -plugin = True -plugin_dir = 'projects/mmdet3d_plugin/' -dist_params = dict(backend='nccl') -log_level = 'INFO' -work_dir = './work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704' -total_batch_size = 160 -num_gpus = 8 -batch_size = 20 -num_iters_per_epoch = 175 -num_epochs = 100 -checkpoint_epoch_interval = 20 -checkpoint_config = dict(interval=3500) -log_config = dict( - interval=1, - hooks=[ - dict(type='TextLoggerHook', by_epoch=False), - dict(type='TensorboardLoggerHook') - ]) -load_from = None -resume_from = None -workflow = [('train', 1)] -fp16 = dict(loss_scale=32.0) -input_shape = (704, 256) -tracking_test = True -tracking_threshold = 0.2 -class_names = [ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier', - 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' -] -num_classes = 10 -embed_dims = 256 -num_groups = 8 -num_decoder = 6 -num_single_frame_decoder = 1 -use_deformable_func = True -strides = [4, 8, 16, 32] -num_levels = 4 -num_depth_layers = 3 -drop_out = 0.1 -temporal = True -decouple_attn = True -with_quality_estimation = True -model = dict( - type='Sparse4D', - use_grid_mask=True, - use_deformable_func=True, - img_backbone=dict( - type='ResNet', - depth=50, - num_stages=4, - frozen_stages=-1, - norm_eval=False, - style='pytorch', - with_cp=True, - out_indices=(0, 1, 2, 3), - norm_cfg=dict(type='BN', requires_grad=True), - pretrained='ckpt/resnet50-19c8e357.pth'), - img_neck=dict( - type='FPN', - num_outs=4, - start_level=0, - out_channels=256, - add_extra_convs='on_output', - relu_before_extra_convs=True, - in_channels=[256, 512, 1024, 2048]), - depth_branch=dict( - type='DenseDepthNet', - embed_dims=256, - num_depth_layers=3, - loss_weight=0.2), - head=dict( - type='Sparse4DHead', - cls_threshold_to_reg=0.05, - decouple_attn=True, - instance_bank=dict( - type='InstanceBank', - num_anchor=900, - embed_dims=256, - anchor='nuscenes_kmeans900.npy', - anchor_handler=dict(type='SparseBox3DKeyPointsGenerator'), - num_temp_instances=600, - confidence_decay=0.6, - feat_grad=False), - anchor_encoder=dict( - type='SparseBox3DEncoder', - vel_dims=3, - embed_dims=[128, 32, 32, 64], - mode='cat', - output_fc=False, - in_loops=1, - out_loops=4), - num_single_frame_decoder=1, - operation_order=[ - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine' - ], - temp_graph_model=dict( - type='MultiheadAttention', - embed_dims=512, - num_heads=8, - batch_first=True, - dropout=0.1), - graph_model=dict( - type='MultiheadAttention', - embed_dims=512, - num_heads=8, - batch_first=True, - dropout=0.1), - norm_layer=dict(type='LN', normalized_shape=256), - ffn=dict( - type='AsymmetricFFN', - in_channels=512, - pre_norm=dict(type='LN'), - embed_dims=256, - feedforward_channels=1024, - num_fcs=2, - ffn_drop=0.1, - act_cfg=dict(type='ReLU', inplace=True)), - deformable_model=dict( - type='DeformableFeatureAggregation', - embed_dims=256, - num_groups=8, - num_levels=4, - num_cams=6, - attn_drop=0.15, - use_deformable_func=True, - use_camera_embed=True, - residual_mode='cat', - kps_generator=dict( - type='SparseBox3DKeyPointsGenerator', - num_learnable_pts=6, - fix_scale=[[0, 0, 0], [0.45, 0, 0], [-0.45, 0, 0], - [0, 0.45, 0], [0, -0.45, 0], [0, 0, 0.45], - [0, 0, -0.45]])), - refine_layer=dict( - type='SparseBox3DRefinementModule', - embed_dims=256, - num_cls=10, - refine_yaw=True, - with_quality_estimation=True), - sampler=dict( - type='SparseBox3DTarget', - num_dn_groups=5, - num_temp_dn_groups=3, - dn_noise_scale=[2.0, 2.0, 2.0, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5], - max_dn_gt=32, - add_neg_dn=True, - cls_weight=2.0, - box_weight=0.25, - reg_weights=[2.0, 2.0, 2.0, 0.5, 0.5, 0.5, 0.0, 0.0, 0.0, 0.0], - cls_wise_reg_weights=dict( - {9: [2.0, 2.0, 2.0, 1.0, 1.0, 1.0, 0.0, 0.0, 1.0, 1.0]})), - loss_cls=dict( - type='FocalLoss', - use_sigmoid=True, - gamma=2.0, - alpha=0.25, - loss_weight=2.0), - loss_reg=dict( - type='SparseBox3DLoss', - loss_box=dict(type='L1Loss', loss_weight=0.25), - loss_centerness=dict(type='CrossEntropyLoss', use_sigmoid=True), - loss_yawness=dict(type='GaussianFocalLoss'), - cls_allow_reverse=[5]), - decoder=dict(type='SparseBox3DDecoder'), - reg_weights=[2.0, 2.0, 2.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0])) -dataset_type = 'NuScenes3DDetTrackDataset' -data_root = 'data/nuscenes/' -anno_root = 'data/nuscenes_anno_pkls/' -file_client_args = dict(backend='disk') -img_norm_cfg = dict( - mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375], to_rgb=True) -train_pipeline = [ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict( - type='LoadPointsFromFile', - coord_type='LIDAR', - load_dim=5, - use_dim=5, - file_client_args=dict(backend='disk')), - dict(type='ResizeCropFlipImage'), - dict(type='MultiScaleDepthMapGenerator', downsample=[4, 8, 16]), - dict(type='BBoxRotation'), - dict(type='PhotoMetricDistortionMultiViewImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict( - type='CircleObjectRangeFilter', - class_dist_thred=[55, 55, 55, 55, 55, 55, 55, 55, 55, 55]), - dict( - type='InstanceNameFilter', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ]), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=[ - 'img', 'timestamp', 'projection_mat', 'image_wh', 'gt_depth', - 'focal', 'gt_bboxes_3d', 'gt_labels_3d' - ], - meta_keys=['T_global', 'T_global_inv', 'timestamp', 'instance_id']) -] -test_pipeline = [ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='ResizeCropFlipImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=['img', 'timestamp', 'projection_mat', 'image_wh'], - meta_keys=['T_global', 'T_global_inv', 'timestamp']) -] -input_modality = dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False) -data_basic_config = dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier', - 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval') -data_aug_conf = dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]) -data = dict( - samples_per_gpu=20, - workers_per_gpu=20, - train=dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval', - ann_file='data/nuscenes_anno_pkls/nuscenes_infos_train.pkl', - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict( - type='LoadPointsFromFile', - coord_type='LIDAR', - load_dim=5, - use_dim=5, - file_client_args=dict(backend='disk')), - dict(type='ResizeCropFlipImage'), - dict(type='MultiScaleDepthMapGenerator', downsample=[4, 8, 16]), - dict(type='BBoxRotation'), - dict(type='PhotoMetricDistortionMultiViewImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict( - type='CircleObjectRangeFilter', - class_dist_thred=[55, 55, 55, 55, 55, 55, 55, 55, 55, 55]), - dict( - type='InstanceNameFilter', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', - 'traffic_cone' - ]), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=[ - 'img', 'timestamp', 'projection_mat', 'image_wh', - 'gt_depth', 'focal', 'gt_bboxes_3d', 'gt_labels_3d' - ], - meta_keys=[ - 'T_global', 'T_global_inv', 'timestamp', 'instance_id' - ]) - ], - test_mode=False, - data_aug_conf=dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]), - with_seq_flag=True, - sequences_split_num=2, - keep_consistent_seq_aug=True), - val=dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval', - ann_file='data/nuscenes_anno_pkls/nuscenes_infos_val.pkl', - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='ResizeCropFlipImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=['img', 'timestamp', 'projection_mat', 'image_wh'], - meta_keys=['T_global', 'T_global_inv', 'timestamp']) - ], - data_aug_conf=dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]), - test_mode=True, - tracking=True, - tracking_threshold=0.2), - test=dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval', - ann_file='data/nuscenes_anno_pkls/nuscenes_infos_val.pkl', - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='ResizeCropFlipImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=['img', 'timestamp', 'projection_mat', 'image_wh'], - meta_keys=['T_global', 'T_global_inv', 'timestamp']) - ], - data_aug_conf=dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]), - test_mode=True, - tracking=True, - tracking_threshold=0.2)) -optimizer = dict( - type='Miopen_AdamW', - lr=0.0006, - weight_decay=0.001, - paramwise_cfg=dict(custom_keys=dict(img_backbone=dict(lr_mult=0.5)))) -optimizer_config = dict(grad_clip=dict(max_norm=25, norm_type=2)) -lr_config = dict( - policy='CosineAnnealing', - warmup='linear', - warmup_iters=500, - warmup_ratio=0.3333333333333333, - min_lr_ratio=0.001) -runner = dict(type='IterBasedRunner', max_iters=17500) -vis_pipeline = [ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='Collect', keys=['img'], meta_keys=['timestamp', 'lidar2img']) -] -evaluation = dict( - interval=3500, - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict( - type='Collect', keys=['img'], meta_keys=['timestamp', 'lidar2img']) - ]) -gpu_ids = range(0, 8) - -2026-01-14 20:12:34,421 - mmdet - INFO - Set random seed to 0, deterministic: False diff --git a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20260114_201948.log b/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20260114_201948.log deleted file mode 100644 index 30094d8aefb740a82f6c29cfbeb29b1dfb2be746..0000000000000000000000000000000000000000 --- a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20260114_201948.log +++ /dev/null @@ -1,465 +0,0 @@ -2026-01-14 20:19:48,736 - mmdet - INFO - Environment info: ------------------------------------------------------------- -sys.platform: linux -Python: 3.10.12 (main, Aug 15 2025, 14:32:43) [GCC 11.4.0] -CUDA available: True -GPU 0,1,2,3,4,5,6,7: BW151 -CUDA_HOME: /opt/dtk -NVCC: Not Available -GCC: x86_64-linux-gnu-gcc (Ubuntu 11.4.0-1ubuntu1~22.04) 11.4.0 -PyTorch: 2.5.1 -PyTorch compiling details: PyTorch built with: - - GCC 10.3 - - C++ Version: 201703 - - Intel(R) Math Kernel Library Version 2020.0.4 Product Build 20200917 for Intel(R) 64 architecture applications - - OpenMP 201511 (a.k.a. OpenMP 4.5) - - LAPACK is enabled (usually provided by MKL) - - NNPACK is enabled - - CPU capability usage: AVX512 - - HIP Runtime 6.3.25521 - - MIOpen 2.18.0 - - Magma 2.8.0 - - Build settings: BLAS_INFO=mkl, BUILD_TYPE=Release, CXX_COMPILER=/opt/rh/gcc-toolset-10/root/usr/bin/c++, CXX_FLAGS= -D_GLIBCXX_USE_CXX11_ABI=1 -fvisibility-inlines-hidden -DUSE_PTHREADPOOL -DNDEBUG -DUSE_KINETO -DLIBKINETO_NOCUPTI -DLIBKINETO_NOXPUPTI=ON -DUSE_FBGEMM -DUSE_PYTORCH_QNNPACK -DUSE_XNNPACK -DSYMBOLICATE_MOBILE_DEBUG_HANDLE -O2 -fPIC -Wall -Wextra -Werror=return-type -Werror=non-virtual-dtor -Werror=bool-operation -Wnarrowing -Wno-missing-field-initializers -Wno-type-limits -Wno-array-bounds -Wno-unknown-pragmas -Wno-unused-parameter -Wno-strict-overflow -Wno-strict-aliasing -Wno-stringop-overflow -Wsuggest-override -Wno-psabi -Wno-error=old-style-cast -Wno-missing-braces -fdiagnostics-color=always -faligned-new -Wno-unused-but-set-variable -Wno-maybe-uninitialized -fno-math-errno -fno-trapping-math -Werror=format -Wno-stringop-overflow, FORCE_FALLBACK_CUDA_MPI=1, LAPACK_INFO=mkl, PERF_WITH_AVX=1, PERF_WITH_AVX2=1, TORCH_VERSION=2.5.1, USE_CUDA=0, USE_CUDNN=OFF, USE_CUSPARSELT=OFF, USE_EXCEPTION_PTR=1, USE_GFLAGS=1, USE_GLOG=1, USE_GLOO=1, USE_MKL=ON, USE_MKLDNN=0, USE_MPI=1, USE_NCCL=1, USE_NNPACK=ON, USE_OPENMP=1, USE_ROCM=ON, USE_ROCM_KERNEL_ASSERT=OFF, - -TorchVision: 0.20.1 -OpenCV: 4.12.0 -MMCV: 1.6.1 -MMCV Compiler: GCC 10.3 -MMCV CUDA Compiler: rocm not available -MMDetection: 2.25.1+c41df4b ------------------------------------------------------------- - -2026-01-14 20:19:49,479 - mmdet - INFO - Distributed training: True -2026-01-14 20:19:50,395 - mmdet - INFO - Config: -plugin = True -plugin_dir = 'projects/mmdet3d_plugin/' -dist_params = dict(backend='nccl') -log_level = 'INFO' -work_dir = './work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704' -total_batch_size = 160 -num_gpus = 8 -batch_size = 20 -num_iters_per_epoch = 175 -num_epochs = 100 -checkpoint_epoch_interval = 20 -checkpoint_config = dict(interval=3500) -log_config = dict( - interval=1, - hooks=[ - dict(type='TextLoggerHook', by_epoch=False), - dict(type='TensorboardLoggerHook') - ]) -load_from = None -resume_from = None -workflow = [('train', 1)] -fp16 = dict(loss_scale=32.0) -input_shape = (704, 256) -tracking_test = True -tracking_threshold = 0.2 -class_names = [ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier', - 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' -] -num_classes = 10 -embed_dims = 256 -num_groups = 8 -num_decoder = 6 -num_single_frame_decoder = 1 -use_deformable_func = True -strides = [4, 8, 16, 32] -num_levels = 4 -num_depth_layers = 3 -drop_out = 0.1 -temporal = True -decouple_attn = True -with_quality_estimation = True -model = dict( - type='Sparse4D', - use_grid_mask=True, - use_deformable_func=True, - img_backbone=dict( - type='ResNet', - depth=50, - num_stages=4, - frozen_stages=-1, - norm_eval=False, - style='pytorch', - with_cp=True, - out_indices=(0, 1, 2, 3), - norm_cfg=dict(type='BN', requires_grad=True), - pretrained='ckpt/resnet50-19c8e357.pth'), - img_neck=dict( - type='FPN', - num_outs=4, - start_level=0, - out_channels=256, - add_extra_convs='on_output', - relu_before_extra_convs=True, - in_channels=[256, 512, 1024, 2048]), - depth_branch=dict( - type='DenseDepthNet', - embed_dims=256, - num_depth_layers=3, - loss_weight=0.2), - head=dict( - type='Sparse4DHead', - cls_threshold_to_reg=0.05, - decouple_attn=True, - instance_bank=dict( - type='InstanceBank', - num_anchor=900, - embed_dims=256, - anchor='nuscenes_kmeans900.npy', - anchor_handler=dict(type='SparseBox3DKeyPointsGenerator'), - num_temp_instances=600, - confidence_decay=0.6, - feat_grad=False), - anchor_encoder=dict( - type='SparseBox3DEncoder', - vel_dims=3, - embed_dims=[128, 32, 32, 64], - mode='cat', - output_fc=False, - in_loops=1, - out_loops=4), - num_single_frame_decoder=1, - operation_order=[ - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine' - ], - temp_graph_model=dict( - type='MultiheadAttention', - embed_dims=512, - num_heads=8, - batch_first=True, - dropout=0.1), - graph_model=dict( - type='MultiheadAttention', - embed_dims=512, - num_heads=8, - batch_first=True, - dropout=0.1), - norm_layer=dict(type='LN', normalized_shape=256), - ffn=dict( - type='AsymmetricFFN', - in_channels=512, - pre_norm=dict(type='LN'), - embed_dims=256, - feedforward_channels=1024, - num_fcs=2, - ffn_drop=0.1, - act_cfg=dict(type='ReLU', inplace=True)), - deformable_model=dict( - type='DeformableFeatureAggregation', - embed_dims=256, - num_groups=8, - num_levels=4, - num_cams=6, - attn_drop=0.15, - use_deformable_func=True, - use_camera_embed=True, - residual_mode='cat', - kps_generator=dict( - type='SparseBox3DKeyPointsGenerator', - num_learnable_pts=6, - fix_scale=[[0, 0, 0], [0.45, 0, 0], [-0.45, 0, 0], - [0, 0.45, 0], [0, -0.45, 0], [0, 0, 0.45], - [0, 0, -0.45]])), - refine_layer=dict( - type='SparseBox3DRefinementModule', - embed_dims=256, - num_cls=10, - refine_yaw=True, - with_quality_estimation=True), - sampler=dict( - type='SparseBox3DTarget', - num_dn_groups=5, - num_temp_dn_groups=3, - dn_noise_scale=[2.0, 2.0, 2.0, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5], - max_dn_gt=32, - add_neg_dn=True, - cls_weight=2.0, - box_weight=0.25, - reg_weights=[2.0, 2.0, 2.0, 0.5, 0.5, 0.5, 0.0, 0.0, 0.0, 0.0], - cls_wise_reg_weights=dict( - {9: [2.0, 2.0, 2.0, 1.0, 1.0, 1.0, 0.0, 0.0, 1.0, 1.0]})), - loss_cls=dict( - type='FocalLoss', - use_sigmoid=True, - gamma=2.0, - alpha=0.25, - loss_weight=2.0), - loss_reg=dict( - type='SparseBox3DLoss', - loss_box=dict(type='L1Loss', loss_weight=0.25), - loss_centerness=dict(type='CrossEntropyLoss', use_sigmoid=True), - loss_yawness=dict(type='GaussianFocalLoss'), - cls_allow_reverse=[5]), - decoder=dict(type='SparseBox3DDecoder'), - reg_weights=[2.0, 2.0, 2.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0])) -dataset_type = 'NuScenes3DDetTrackDataset' -data_root = 'data/nuscenes/' -anno_root = 'data/nuscenes_anno_pkls/' -file_client_args = dict(backend='disk') -img_norm_cfg = dict( - mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375], to_rgb=True) -train_pipeline = [ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict( - type='LoadPointsFromFile', - coord_type='LIDAR', - load_dim=5, - use_dim=5, - file_client_args=dict(backend='disk')), - dict(type='ResizeCropFlipImage'), - dict(type='MultiScaleDepthMapGenerator', downsample=[4, 8, 16]), - dict(type='BBoxRotation'), - dict(type='PhotoMetricDistortionMultiViewImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict( - type='CircleObjectRangeFilter', - class_dist_thred=[55, 55, 55, 55, 55, 55, 55, 55, 55, 55]), - dict( - type='InstanceNameFilter', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ]), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=[ - 'img', 'timestamp', 'projection_mat', 'image_wh', 'gt_depth', - 'focal', 'gt_bboxes_3d', 'gt_labels_3d' - ], - meta_keys=['T_global', 'T_global_inv', 'timestamp', 'instance_id']) -] -test_pipeline = [ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='ResizeCropFlipImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=['img', 'timestamp', 'projection_mat', 'image_wh'], - meta_keys=['T_global', 'T_global_inv', 'timestamp']) -] -input_modality = dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False) -data_basic_config = dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier', - 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval') -data_aug_conf = dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]) -data = dict( - samples_per_gpu=20, - workers_per_gpu=20, - train=dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval', - ann_file='data/nuscenes_anno_pkls/nuscenes_infos_train.pkl', - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict( - type='LoadPointsFromFile', - coord_type='LIDAR', - load_dim=5, - use_dim=5, - file_client_args=dict(backend='disk')), - dict(type='ResizeCropFlipImage'), - dict(type='MultiScaleDepthMapGenerator', downsample=[4, 8, 16]), - dict(type='BBoxRotation'), - dict(type='PhotoMetricDistortionMultiViewImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict( - type='CircleObjectRangeFilter', - class_dist_thred=[55, 55, 55, 55, 55, 55, 55, 55, 55, 55]), - dict( - type='InstanceNameFilter', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', - 'traffic_cone' - ]), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=[ - 'img', 'timestamp', 'projection_mat', 'image_wh', - 'gt_depth', 'focal', 'gt_bboxes_3d', 'gt_labels_3d' - ], - meta_keys=[ - 'T_global', 'T_global_inv', 'timestamp', 'instance_id' - ]) - ], - test_mode=False, - data_aug_conf=dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]), - with_seq_flag=True, - sequences_split_num=2, - keep_consistent_seq_aug=True), - val=dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval', - ann_file='data/nuscenes_anno_pkls/nuscenes_infos_val.pkl', - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='ResizeCropFlipImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=['img', 'timestamp', 'projection_mat', 'image_wh'], - meta_keys=['T_global', 'T_global_inv', 'timestamp']) - ], - data_aug_conf=dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]), - test_mode=True, - tracking=True, - tracking_threshold=0.2), - test=dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval', - ann_file='data/nuscenes_anno_pkls/nuscenes_infos_val.pkl', - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='ResizeCropFlipImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=['img', 'timestamp', 'projection_mat', 'image_wh'], - meta_keys=['T_global', 'T_global_inv', 'timestamp']) - ], - data_aug_conf=dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]), - test_mode=True, - tracking=True, - tracking_threshold=0.2)) -optimizer = dict( - type='Miopen_AdamW', - lr=0.0006, - weight_decay=0.001, - paramwise_cfg=dict(custom_keys=dict(img_backbone=dict(lr_mult=0.5)))) -optimizer_config = dict(grad_clip=dict(max_norm=25, norm_type=2)) -lr_config = dict( - policy='CosineAnnealing', - warmup='linear', - warmup_iters=500, - warmup_ratio=0.3333333333333333, - min_lr_ratio=0.001) -runner = dict(type='IterBasedRunner', max_iters=17500) -vis_pipeline = [ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='Collect', keys=['img'], meta_keys=['timestamp', 'lidar2img']) -] -evaluation = dict( - interval=3500, - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict( - type='Collect', keys=['img'], meta_keys=['timestamp', 'lidar2img']) - ]) -gpu_ids = range(0, 8) - -2026-01-14 20:19:50,396 - mmdet - INFO - Set random seed to 0, deterministic: False diff --git a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20260114_202426.log b/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20260114_202426.log deleted file mode 100644 index f0232094d0b4a8663752faa064827cd4c0668f79..0000000000000000000000000000000000000000 --- a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20260114_202426.log +++ /dev/null @@ -1,465 +0,0 @@ -2026-01-14 20:24:27,005 - mmdet - INFO - Environment info: ------------------------------------------------------------- -sys.platform: linux -Python: 3.10.12 (main, Aug 15 2025, 14:32:43) [GCC 11.4.0] -CUDA available: True -GPU 0,1,2,3,4,5,6,7: BW151 -CUDA_HOME: /opt/dtk -NVCC: Not Available -GCC: x86_64-linux-gnu-gcc (Ubuntu 11.4.0-1ubuntu1~22.04) 11.4.0 -PyTorch: 2.5.1 -PyTorch compiling details: PyTorch built with: - - GCC 10.3 - - C++ Version: 201703 - - Intel(R) Math Kernel Library Version 2020.0.4 Product Build 20200917 for Intel(R) 64 architecture applications - - OpenMP 201511 (a.k.a. OpenMP 4.5) - - LAPACK is enabled (usually provided by MKL) - - NNPACK is enabled - - CPU capability usage: AVX512 - - HIP Runtime 6.3.25521 - - MIOpen 2.18.0 - - Magma 2.8.0 - - Build settings: BLAS_INFO=mkl, BUILD_TYPE=Release, CXX_COMPILER=/opt/rh/gcc-toolset-10/root/usr/bin/c++, CXX_FLAGS= -D_GLIBCXX_USE_CXX11_ABI=1 -fvisibility-inlines-hidden -DUSE_PTHREADPOOL -DNDEBUG -DUSE_KINETO -DLIBKINETO_NOCUPTI -DLIBKINETO_NOXPUPTI=ON -DUSE_FBGEMM -DUSE_PYTORCH_QNNPACK -DUSE_XNNPACK -DSYMBOLICATE_MOBILE_DEBUG_HANDLE -O2 -fPIC -Wall -Wextra -Werror=return-type -Werror=non-virtual-dtor -Werror=bool-operation -Wnarrowing -Wno-missing-field-initializers -Wno-type-limits -Wno-array-bounds -Wno-unknown-pragmas -Wno-unused-parameter -Wno-strict-overflow -Wno-strict-aliasing -Wno-stringop-overflow -Wsuggest-override -Wno-psabi -Wno-error=old-style-cast -Wno-missing-braces -fdiagnostics-color=always -faligned-new -Wno-unused-but-set-variable -Wno-maybe-uninitialized -fno-math-errno -fno-trapping-math -Werror=format -Wno-stringop-overflow, FORCE_FALLBACK_CUDA_MPI=1, LAPACK_INFO=mkl, PERF_WITH_AVX=1, PERF_WITH_AVX2=1, TORCH_VERSION=2.5.1, USE_CUDA=0, USE_CUDNN=OFF, USE_CUSPARSELT=OFF, USE_EXCEPTION_PTR=1, USE_GFLAGS=1, USE_GLOG=1, USE_GLOO=1, USE_MKL=ON, USE_MKLDNN=0, USE_MPI=1, USE_NCCL=1, USE_NNPACK=ON, USE_OPENMP=1, USE_ROCM=ON, USE_ROCM_KERNEL_ASSERT=OFF, - -TorchVision: 0.20.1 -OpenCV: 4.12.0 -MMCV: 1.6.1 -MMCV Compiler: GCC 10.3 -MMCV CUDA Compiler: rocm not available -MMDetection: 2.25.1+c41df4b ------------------------------------------------------------- - -2026-01-14 20:24:27,748 - mmdet - INFO - Distributed training: True -2026-01-14 20:24:28,653 - mmdet - INFO - Config: -plugin = True -plugin_dir = 'projects/mmdet3d_plugin/' -dist_params = dict(backend='nccl') -log_level = 'INFO' -work_dir = './work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704' -total_batch_size = 160 -num_gpus = 8 -batch_size = 20 -num_iters_per_epoch = 175 -num_epochs = 100 -checkpoint_epoch_interval = 20 -checkpoint_config = dict(interval=3500) -log_config = dict( - interval=1, - hooks=[ - dict(type='TextLoggerHook', by_epoch=False), - dict(type='TensorboardLoggerHook') - ]) -load_from = None -resume_from = None -workflow = [('train', 1)] -fp16 = dict(loss_scale=32.0) -input_shape = (704, 256) -tracking_test = True -tracking_threshold = 0.2 -class_names = [ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier', - 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' -] -num_classes = 10 -embed_dims = 256 -num_groups = 8 -num_decoder = 6 -num_single_frame_decoder = 1 -use_deformable_func = True -strides = [4, 8, 16, 32] -num_levels = 4 -num_depth_layers = 3 -drop_out = 0.1 -temporal = True -decouple_attn = True -with_quality_estimation = True -model = dict( - type='Sparse4D', - use_grid_mask=True, - use_deformable_func=True, - img_backbone=dict( - type='ResNet', - depth=50, - num_stages=4, - frozen_stages=-1, - norm_eval=False, - style='pytorch', - with_cp=True, - out_indices=(0, 1, 2, 3), - norm_cfg=dict(type='BN', requires_grad=True), - pretrained='ckpt/resnet50-19c8e357.pth'), - img_neck=dict( - type='FPN', - num_outs=4, - start_level=0, - out_channels=256, - add_extra_convs='on_output', - relu_before_extra_convs=True, - in_channels=[256, 512, 1024, 2048]), - depth_branch=dict( - type='DenseDepthNet', - embed_dims=256, - num_depth_layers=3, - loss_weight=0.2), - head=dict( - type='Sparse4DHead', - cls_threshold_to_reg=0.05, - decouple_attn=True, - instance_bank=dict( - type='InstanceBank', - num_anchor=900, - embed_dims=256, - anchor='nuscenes_kmeans900.npy', - anchor_handler=dict(type='SparseBox3DKeyPointsGenerator'), - num_temp_instances=600, - confidence_decay=0.6, - feat_grad=False), - anchor_encoder=dict( - type='SparseBox3DEncoder', - vel_dims=3, - embed_dims=[128, 32, 32, 64], - mode='cat', - output_fc=False, - in_loops=1, - out_loops=4), - num_single_frame_decoder=1, - operation_order=[ - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine' - ], - temp_graph_model=dict( - type='MultiheadAttention', - embed_dims=512, - num_heads=8, - batch_first=True, - dropout=0.1), - graph_model=dict( - type='MultiheadAttention', - embed_dims=512, - num_heads=8, - batch_first=True, - dropout=0.1), - norm_layer=dict(type='LN', normalized_shape=256), - ffn=dict( - type='AsymmetricFFN', - in_channels=512, - pre_norm=dict(type='LN'), - embed_dims=256, - feedforward_channels=1024, - num_fcs=2, - ffn_drop=0.1, - act_cfg=dict(type='ReLU', inplace=True)), - deformable_model=dict( - type='DeformableFeatureAggregation', - embed_dims=256, - num_groups=8, - num_levels=4, - num_cams=6, - attn_drop=0.15, - use_deformable_func=True, - use_camera_embed=True, - residual_mode='cat', - kps_generator=dict( - type='SparseBox3DKeyPointsGenerator', - num_learnable_pts=6, - fix_scale=[[0, 0, 0], [0.45, 0, 0], [-0.45, 0, 0], - [0, 0.45, 0], [0, -0.45, 0], [0, 0, 0.45], - [0, 0, -0.45]])), - refine_layer=dict( - type='SparseBox3DRefinementModule', - embed_dims=256, - num_cls=10, - refine_yaw=True, - with_quality_estimation=True), - sampler=dict( - type='SparseBox3DTarget', - num_dn_groups=5, - num_temp_dn_groups=3, - dn_noise_scale=[2.0, 2.0, 2.0, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5], - max_dn_gt=32, - add_neg_dn=True, - cls_weight=2.0, - box_weight=0.25, - reg_weights=[2.0, 2.0, 2.0, 0.5, 0.5, 0.5, 0.0, 0.0, 0.0, 0.0], - cls_wise_reg_weights=dict( - {9: [2.0, 2.0, 2.0, 1.0, 1.0, 1.0, 0.0, 0.0, 1.0, 1.0]})), - loss_cls=dict( - type='FocalLoss', - use_sigmoid=True, - gamma=2.0, - alpha=0.25, - loss_weight=2.0), - loss_reg=dict( - type='SparseBox3DLoss', - loss_box=dict(type='L1Loss', loss_weight=0.25), - loss_centerness=dict(type='CrossEntropyLoss', use_sigmoid=True), - loss_yawness=dict(type='GaussianFocalLoss'), - cls_allow_reverse=[5]), - decoder=dict(type='SparseBox3DDecoder'), - reg_weights=[2.0, 2.0, 2.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0])) -dataset_type = 'NuScenes3DDetTrackDataset' -data_root = 'data/nuscenes/' -anno_root = 'data/nuscenes_anno_pkls/' -file_client_args = dict(backend='disk') -img_norm_cfg = dict( - mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375], to_rgb=True) -train_pipeline = [ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict( - type='LoadPointsFromFile', - coord_type='LIDAR', - load_dim=5, - use_dim=5, - file_client_args=dict(backend='disk')), - dict(type='ResizeCropFlipImage'), - dict(type='MultiScaleDepthMapGenerator', downsample=[4, 8, 16]), - dict(type='BBoxRotation'), - dict(type='PhotoMetricDistortionMultiViewImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict( - type='CircleObjectRangeFilter', - class_dist_thred=[55, 55, 55, 55, 55, 55, 55, 55, 55, 55]), - dict( - type='InstanceNameFilter', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ]), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=[ - 'img', 'timestamp', 'projection_mat', 'image_wh', 'gt_depth', - 'focal', 'gt_bboxes_3d', 'gt_labels_3d' - ], - meta_keys=['T_global', 'T_global_inv', 'timestamp', 'instance_id']) -] -test_pipeline = [ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='ResizeCropFlipImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=['img', 'timestamp', 'projection_mat', 'image_wh'], - meta_keys=['T_global', 'T_global_inv', 'timestamp']) -] -input_modality = dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False) -data_basic_config = dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier', - 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval') -data_aug_conf = dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]) -data = dict( - samples_per_gpu=20, - workers_per_gpu=20, - train=dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval', - ann_file='data/nuscenes_anno_pkls/nuscenes_infos_train.pkl', - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict( - type='LoadPointsFromFile', - coord_type='LIDAR', - load_dim=5, - use_dim=5, - file_client_args=dict(backend='disk')), - dict(type='ResizeCropFlipImage'), - dict(type='MultiScaleDepthMapGenerator', downsample=[4, 8, 16]), - dict(type='BBoxRotation'), - dict(type='PhotoMetricDistortionMultiViewImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict( - type='CircleObjectRangeFilter', - class_dist_thred=[55, 55, 55, 55, 55, 55, 55, 55, 55, 55]), - dict( - type='InstanceNameFilter', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', - 'traffic_cone' - ]), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=[ - 'img', 'timestamp', 'projection_mat', 'image_wh', - 'gt_depth', 'focal', 'gt_bboxes_3d', 'gt_labels_3d' - ], - meta_keys=[ - 'T_global', 'T_global_inv', 'timestamp', 'instance_id' - ]) - ], - test_mode=False, - data_aug_conf=dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]), - with_seq_flag=True, - sequences_split_num=2, - keep_consistent_seq_aug=True), - val=dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval', - ann_file='data/nuscenes_anno_pkls/nuscenes_infos_val.pkl', - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='ResizeCropFlipImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=['img', 'timestamp', 'projection_mat', 'image_wh'], - meta_keys=['T_global', 'T_global_inv', 'timestamp']) - ], - data_aug_conf=dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]), - test_mode=True, - tracking=True, - tracking_threshold=0.2), - test=dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval', - ann_file='data/nuscenes_anno_pkls/nuscenes_infos_val.pkl', - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='ResizeCropFlipImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=['img', 'timestamp', 'projection_mat', 'image_wh'], - meta_keys=['T_global', 'T_global_inv', 'timestamp']) - ], - data_aug_conf=dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]), - test_mode=True, - tracking=True, - tracking_threshold=0.2)) -optimizer = dict( - type='Miopen_AdamW', - lr=0.0006, - weight_decay=0.001, - paramwise_cfg=dict(custom_keys=dict(img_backbone=dict(lr_mult=0.5)))) -optimizer_config = dict(grad_clip=dict(max_norm=25, norm_type=2)) -lr_config = dict( - policy='CosineAnnealing', - warmup='linear', - warmup_iters=500, - warmup_ratio=0.3333333333333333, - min_lr_ratio=0.001) -runner = dict(type='IterBasedRunner', max_iters=17500) -vis_pipeline = [ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='Collect', keys=['img'], meta_keys=['timestamp', 'lidar2img']) -] -evaluation = dict( - interval=3500, - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict( - type='Collect', keys=['img'], meta_keys=['timestamp', 'lidar2img']) - ]) -gpu_ids = range(0, 8) - -2026-01-14 20:24:28,653 - mmdet - INFO - Set random seed to 0, deterministic: False diff --git a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20260114_202527.log b/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20260114_202527.log deleted file mode 100644 index eaad6d254bf5c5e516a351204a09b026954e0530..0000000000000000000000000000000000000000 --- a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20260114_202527.log +++ /dev/null @@ -1,465 +0,0 @@ -2026-01-14 20:25:27,213 - mmdet - INFO - Environment info: ------------------------------------------------------------- -sys.platform: linux -Python: 3.10.12 (main, Aug 15 2025, 14:32:43) [GCC 11.4.0] -CUDA available: True -GPU 0,1,2,3,4,5,6,7: BW151 -CUDA_HOME: /opt/dtk -NVCC: Not Available -GCC: x86_64-linux-gnu-gcc (Ubuntu 11.4.0-1ubuntu1~22.04) 11.4.0 -PyTorch: 2.5.1 -PyTorch compiling details: PyTorch built with: - - GCC 10.3 - - C++ Version: 201703 - - Intel(R) Math Kernel Library Version 2020.0.4 Product Build 20200917 for Intel(R) 64 architecture applications - - OpenMP 201511 (a.k.a. OpenMP 4.5) - - LAPACK is enabled (usually provided by MKL) - - NNPACK is enabled - - CPU capability usage: AVX512 - - HIP Runtime 6.3.25521 - - MIOpen 2.18.0 - - Magma 2.8.0 - - Build settings: BLAS_INFO=mkl, BUILD_TYPE=Release, CXX_COMPILER=/opt/rh/gcc-toolset-10/root/usr/bin/c++, CXX_FLAGS= -D_GLIBCXX_USE_CXX11_ABI=1 -fvisibility-inlines-hidden -DUSE_PTHREADPOOL -DNDEBUG -DUSE_KINETO -DLIBKINETO_NOCUPTI -DLIBKINETO_NOXPUPTI=ON -DUSE_FBGEMM -DUSE_PYTORCH_QNNPACK -DUSE_XNNPACK -DSYMBOLICATE_MOBILE_DEBUG_HANDLE -O2 -fPIC -Wall -Wextra -Werror=return-type -Werror=non-virtual-dtor -Werror=bool-operation -Wnarrowing -Wno-missing-field-initializers -Wno-type-limits -Wno-array-bounds -Wno-unknown-pragmas -Wno-unused-parameter -Wno-strict-overflow -Wno-strict-aliasing -Wno-stringop-overflow -Wsuggest-override -Wno-psabi -Wno-error=old-style-cast -Wno-missing-braces -fdiagnostics-color=always -faligned-new -Wno-unused-but-set-variable -Wno-maybe-uninitialized -fno-math-errno -fno-trapping-math -Werror=format -Wno-stringop-overflow, FORCE_FALLBACK_CUDA_MPI=1, LAPACK_INFO=mkl, PERF_WITH_AVX=1, PERF_WITH_AVX2=1, TORCH_VERSION=2.5.1, USE_CUDA=0, USE_CUDNN=OFF, USE_CUSPARSELT=OFF, USE_EXCEPTION_PTR=1, USE_GFLAGS=1, USE_GLOG=1, USE_GLOO=1, USE_MKL=ON, USE_MKLDNN=0, USE_MPI=1, USE_NCCL=1, USE_NNPACK=ON, USE_OPENMP=1, USE_ROCM=ON, USE_ROCM_KERNEL_ASSERT=OFF, - -TorchVision: 0.20.1 -OpenCV: 4.12.0 -MMCV: 1.6.1 -MMCV Compiler: GCC 10.3 -MMCV CUDA Compiler: rocm not available -MMDetection: 2.25.1+c41df4b ------------------------------------------------------------- - -2026-01-14 20:25:27,953 - mmdet - INFO - Distributed training: True -2026-01-14 20:25:28,874 - mmdet - INFO - Config: -plugin = True -plugin_dir = 'projects/mmdet3d_plugin/' -dist_params = dict(backend='nccl') -log_level = 'INFO' -work_dir = './work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704' -total_batch_size = 160 -num_gpus = 8 -batch_size = 20 -num_iters_per_epoch = 175 -num_epochs = 100 -checkpoint_epoch_interval = 20 -checkpoint_config = dict(interval=3500) -log_config = dict( - interval=1, - hooks=[ - dict(type='TextLoggerHook', by_epoch=False), - dict(type='TensorboardLoggerHook') - ]) -load_from = None -resume_from = None -workflow = [('train', 1)] -fp16 = dict(loss_scale=32.0) -input_shape = (704, 256) -tracking_test = True -tracking_threshold = 0.2 -class_names = [ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier', - 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' -] -num_classes = 10 -embed_dims = 256 -num_groups = 8 -num_decoder = 6 -num_single_frame_decoder = 1 -use_deformable_func = True -strides = [4, 8, 16, 32] -num_levels = 4 -num_depth_layers = 3 -drop_out = 0.1 -temporal = True -decouple_attn = True -with_quality_estimation = True -model = dict( - type='Sparse4D', - use_grid_mask=True, - use_deformable_func=True, - img_backbone=dict( - type='ResNet', - depth=50, - num_stages=4, - frozen_stages=-1, - norm_eval=False, - style='pytorch', - with_cp=True, - out_indices=(0, 1, 2, 3), - norm_cfg=dict(type='BN', requires_grad=True), - pretrained='ckpt/resnet50-19c8e357.pth'), - img_neck=dict( - type='FPN', - num_outs=4, - start_level=0, - out_channels=256, - add_extra_convs='on_output', - relu_before_extra_convs=True, - in_channels=[256, 512, 1024, 2048]), - depth_branch=dict( - type='DenseDepthNet', - embed_dims=256, - num_depth_layers=3, - loss_weight=0.2), - head=dict( - type='Sparse4DHead', - cls_threshold_to_reg=0.05, - decouple_attn=True, - instance_bank=dict( - type='InstanceBank', - num_anchor=900, - embed_dims=256, - anchor='nuscenes_kmeans900.npy', - anchor_handler=dict(type='SparseBox3DKeyPointsGenerator'), - num_temp_instances=600, - confidence_decay=0.6, - feat_grad=False), - anchor_encoder=dict( - type='SparseBox3DEncoder', - vel_dims=3, - embed_dims=[128, 32, 32, 64], - mode='cat', - output_fc=False, - in_loops=1, - out_loops=4), - num_single_frame_decoder=1, - operation_order=[ - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine' - ], - temp_graph_model=dict( - type='MultiheadAttention', - embed_dims=512, - num_heads=8, - batch_first=True, - dropout=0.1), - graph_model=dict( - type='MultiheadAttention', - embed_dims=512, - num_heads=8, - batch_first=True, - dropout=0.1), - norm_layer=dict(type='LN', normalized_shape=256), - ffn=dict( - type='AsymmetricFFN', - in_channels=512, - pre_norm=dict(type='LN'), - embed_dims=256, - feedforward_channels=1024, - num_fcs=2, - ffn_drop=0.1, - act_cfg=dict(type='ReLU', inplace=True)), - deformable_model=dict( - type='DeformableFeatureAggregation', - embed_dims=256, - num_groups=8, - num_levels=4, - num_cams=6, - attn_drop=0.15, - use_deformable_func=True, - use_camera_embed=True, - residual_mode='cat', - kps_generator=dict( - type='SparseBox3DKeyPointsGenerator', - num_learnable_pts=6, - fix_scale=[[0, 0, 0], [0.45, 0, 0], [-0.45, 0, 0], - [0, 0.45, 0], [0, -0.45, 0], [0, 0, 0.45], - [0, 0, -0.45]])), - refine_layer=dict( - type='SparseBox3DRefinementModule', - embed_dims=256, - num_cls=10, - refine_yaw=True, - with_quality_estimation=True), - sampler=dict( - type='SparseBox3DTarget', - num_dn_groups=5, - num_temp_dn_groups=3, - dn_noise_scale=[2.0, 2.0, 2.0, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5], - max_dn_gt=32, - add_neg_dn=True, - cls_weight=2.0, - box_weight=0.25, - reg_weights=[2.0, 2.0, 2.0, 0.5, 0.5, 0.5, 0.0, 0.0, 0.0, 0.0], - cls_wise_reg_weights=dict( - {9: [2.0, 2.0, 2.0, 1.0, 1.0, 1.0, 0.0, 0.0, 1.0, 1.0]})), - loss_cls=dict( - type='FocalLoss', - use_sigmoid=True, - gamma=2.0, - alpha=0.25, - loss_weight=2.0), - loss_reg=dict( - type='SparseBox3DLoss', - loss_box=dict(type='L1Loss', loss_weight=0.25), - loss_centerness=dict(type='CrossEntropyLoss', use_sigmoid=True), - loss_yawness=dict(type='GaussianFocalLoss'), - cls_allow_reverse=[5]), - decoder=dict(type='SparseBox3DDecoder'), - reg_weights=[2.0, 2.0, 2.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0])) -dataset_type = 'NuScenes3DDetTrackDataset' -data_root = 'data/nuscenes/' -anno_root = 'data/nuscenes_anno_pkls/' -file_client_args = dict(backend='disk') -img_norm_cfg = dict( - mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375], to_rgb=True) -train_pipeline = [ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict( - type='LoadPointsFromFile', - coord_type='LIDAR', - load_dim=5, - use_dim=5, - file_client_args=dict(backend='disk')), - dict(type='ResizeCropFlipImage'), - dict(type='MultiScaleDepthMapGenerator', downsample=[4, 8, 16]), - dict(type='BBoxRotation'), - dict(type='PhotoMetricDistortionMultiViewImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict( - type='CircleObjectRangeFilter', - class_dist_thred=[55, 55, 55, 55, 55, 55, 55, 55, 55, 55]), - dict( - type='InstanceNameFilter', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ]), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=[ - 'img', 'timestamp', 'projection_mat', 'image_wh', 'gt_depth', - 'focal', 'gt_bboxes_3d', 'gt_labels_3d' - ], - meta_keys=['T_global', 'T_global_inv', 'timestamp', 'instance_id']) -] -test_pipeline = [ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='ResizeCropFlipImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=['img', 'timestamp', 'projection_mat', 'image_wh'], - meta_keys=['T_global', 'T_global_inv', 'timestamp']) -] -input_modality = dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False) -data_basic_config = dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier', - 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval') -data_aug_conf = dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]) -data = dict( - samples_per_gpu=20, - workers_per_gpu=20, - train=dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval', - ann_file='data/nuscenes_anno_pkls/nuscenes_infos_train.pkl', - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict( - type='LoadPointsFromFile', - coord_type='LIDAR', - load_dim=5, - use_dim=5, - file_client_args=dict(backend='disk')), - dict(type='ResizeCropFlipImage'), - dict(type='MultiScaleDepthMapGenerator', downsample=[4, 8, 16]), - dict(type='BBoxRotation'), - dict(type='PhotoMetricDistortionMultiViewImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict( - type='CircleObjectRangeFilter', - class_dist_thred=[55, 55, 55, 55, 55, 55, 55, 55, 55, 55]), - dict( - type='InstanceNameFilter', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', - 'traffic_cone' - ]), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=[ - 'img', 'timestamp', 'projection_mat', 'image_wh', - 'gt_depth', 'focal', 'gt_bboxes_3d', 'gt_labels_3d' - ], - meta_keys=[ - 'T_global', 'T_global_inv', 'timestamp', 'instance_id' - ]) - ], - test_mode=False, - data_aug_conf=dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]), - with_seq_flag=True, - sequences_split_num=2, - keep_consistent_seq_aug=True), - val=dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval', - ann_file='data/nuscenes_anno_pkls/nuscenes_infos_val.pkl', - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='ResizeCropFlipImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=['img', 'timestamp', 'projection_mat', 'image_wh'], - meta_keys=['T_global', 'T_global_inv', 'timestamp']) - ], - data_aug_conf=dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]), - test_mode=True, - tracking=True, - tracking_threshold=0.2), - test=dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval', - ann_file='data/nuscenes_anno_pkls/nuscenes_infos_val.pkl', - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='ResizeCropFlipImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=['img', 'timestamp', 'projection_mat', 'image_wh'], - meta_keys=['T_global', 'T_global_inv', 'timestamp']) - ], - data_aug_conf=dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]), - test_mode=True, - tracking=True, - tracking_threshold=0.2)) -optimizer = dict( - type='Miopen_AdamW', - lr=0.0006, - weight_decay=0.001, - paramwise_cfg=dict(custom_keys=dict(img_backbone=dict(lr_mult=0.5)))) -optimizer_config = dict(grad_clip=dict(max_norm=25, norm_type=2)) -lr_config = dict( - policy='CosineAnnealing', - warmup='linear', - warmup_iters=500, - warmup_ratio=0.3333333333333333, - min_lr_ratio=0.001) -runner = dict(type='IterBasedRunner', max_iters=17500) -vis_pipeline = [ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='Collect', keys=['img'], meta_keys=['timestamp', 'lidar2img']) -] -evaluation = dict( - interval=3500, - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict( - type='Collect', keys=['img'], meta_keys=['timestamp', 'lidar2img']) - ]) -gpu_ids = range(0, 8) - -2026-01-14 20:25:28,875 - mmdet - INFO - Set random seed to 0, deterministic: False diff --git a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20260114_202744.log b/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20260114_202744.log deleted file mode 100644 index 8453940b930177cff15e0b4d636b5eb322751bb9..0000000000000000000000000000000000000000 --- a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20260114_202744.log +++ /dev/null @@ -1,469 +0,0 @@ -2026-01-14 20:27:44,264 - mmdet - INFO - Environment info: ------------------------------------------------------------- -sys.platform: linux -Python: 3.10.12 (main, Aug 15 2025, 14:32:43) [GCC 11.4.0] -CUDA available: True -GPU 0,1,2,3,4,5,6,7: BW151 -CUDA_HOME: /opt/dtk/cuda/cuda -NVCC: Cuda compilation tools, release 12.6, V12.6.77 -clang version 17.0.0 -Target: x86_64-unknown-linux-gnu -Thread model: posix -InstalledDir: /opt/dtk-25.04.4/llvm/bi -GCC: x86_64-linux-gnu-gcc (Ubuntu 11.4.0-1ubuntu1~22.04) 11.4.0 -PyTorch: 2.5.1 -PyTorch compiling details: PyTorch built with: - - GCC 10.3 - - C++ Version: 201703 - - Intel(R) Math Kernel Library Version 2020.0.4 Product Build 20200917 for Intel(R) 64 architecture applications - - OpenMP 201511 (a.k.a. OpenMP 4.5) - - LAPACK is enabled (usually provided by MKL) - - NNPACK is enabled - - CPU capability usage: AVX512 - - HIP Runtime 6.3.25521 - - MIOpen 2.18.0 - - Magma 2.8.0 - - Build settings: BLAS_INFO=mkl, BUILD_TYPE=Release, CXX_COMPILER=/opt/rh/gcc-toolset-10/root/usr/bin/c++, CXX_FLAGS= -D_GLIBCXX_USE_CXX11_ABI=1 -fvisibility-inlines-hidden -DUSE_PTHREADPOOL -DNDEBUG -DUSE_KINETO -DLIBKINETO_NOCUPTI -DLIBKINETO_NOXPUPTI=ON -DUSE_FBGEMM -DUSE_PYTORCH_QNNPACK -DUSE_XNNPACK -DSYMBOLICATE_MOBILE_DEBUG_HANDLE -O2 -fPIC -Wall -Wextra -Werror=return-type -Werror=non-virtual-dtor -Werror=bool-operation -Wnarrowing -Wno-missing-field-initializers -Wno-type-limits -Wno-array-bounds -Wno-unknown-pragmas -Wno-unused-parameter -Wno-strict-overflow -Wno-strict-aliasing -Wno-stringop-overflow -Wsuggest-override -Wno-psabi -Wno-error=old-style-cast -Wno-missing-braces -fdiagnostics-color=always -faligned-new -Wno-unused-but-set-variable -Wno-maybe-uninitialized -fno-math-errno -fno-trapping-math -Werror=format -Wno-stringop-overflow, FORCE_FALLBACK_CUDA_MPI=1, LAPACK_INFO=mkl, PERF_WITH_AVX=1, PERF_WITH_AVX2=1, TORCH_VERSION=2.5.1, USE_CUDA=0, USE_CUDNN=OFF, USE_CUSPARSELT=OFF, USE_EXCEPTION_PTR=1, USE_GFLAGS=1, USE_GLOG=1, USE_GLOO=1, USE_MKL=ON, USE_MKLDNN=0, USE_MPI=1, USE_NCCL=1, USE_NNPACK=ON, USE_OPENMP=1, USE_ROCM=ON, USE_ROCM_KERNEL_ASSERT=OFF, - -TorchVision: 0.20.1 -OpenCV: 4.12.0 -MMCV: 1.6.1 -MMCV Compiler: GCC 10.3 -MMCV CUDA Compiler: rocm not available -MMDetection: 2.25.1+c41df4b ------------------------------------------------------------- - -2026-01-14 20:27:45,020 - mmdet - INFO - Distributed training: True -2026-01-14 20:27:45,956 - mmdet - INFO - Config: -plugin = True -plugin_dir = 'projects/mmdet3d_plugin/' -dist_params = dict(backend='nccl') -log_level = 'INFO' -work_dir = './work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704' -total_batch_size = 160 -num_gpus = 8 -batch_size = 20 -num_iters_per_epoch = 175 -num_epochs = 100 -checkpoint_epoch_interval = 20 -checkpoint_config = dict(interval=3500) -log_config = dict( - interval=1, - hooks=[ - dict(type='TextLoggerHook', by_epoch=False), - dict(type='TensorboardLoggerHook') - ]) -load_from = None -resume_from = None -workflow = [('train', 1)] -fp16 = dict(loss_scale=32.0) -input_shape = (704, 256) -tracking_test = True -tracking_threshold = 0.2 -class_names = [ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier', - 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' -] -num_classes = 10 -embed_dims = 256 -num_groups = 8 -num_decoder = 6 -num_single_frame_decoder = 1 -use_deformable_func = True -strides = [4, 8, 16, 32] -num_levels = 4 -num_depth_layers = 3 -drop_out = 0.1 -temporal = True -decouple_attn = True -with_quality_estimation = True -model = dict( - type='Sparse4D', - use_grid_mask=True, - use_deformable_func=True, - img_backbone=dict( - type='ResNet', - depth=50, - num_stages=4, - frozen_stages=-1, - norm_eval=False, - style='pytorch', - with_cp=True, - out_indices=(0, 1, 2, 3), - norm_cfg=dict(type='BN', requires_grad=True), - pretrained='ckpt/resnet50-19c8e357.pth'), - img_neck=dict( - type='FPN', - num_outs=4, - start_level=0, - out_channels=256, - add_extra_convs='on_output', - relu_before_extra_convs=True, - in_channels=[256, 512, 1024, 2048]), - depth_branch=dict( - type='DenseDepthNet', - embed_dims=256, - num_depth_layers=3, - loss_weight=0.2), - head=dict( - type='Sparse4DHead', - cls_threshold_to_reg=0.05, - decouple_attn=True, - instance_bank=dict( - type='InstanceBank', - num_anchor=900, - embed_dims=256, - anchor='nuscenes_kmeans900.npy', - anchor_handler=dict(type='SparseBox3DKeyPointsGenerator'), - num_temp_instances=600, - confidence_decay=0.6, - feat_grad=False), - anchor_encoder=dict( - type='SparseBox3DEncoder', - vel_dims=3, - embed_dims=[128, 32, 32, 64], - mode='cat', - output_fc=False, - in_loops=1, - out_loops=4), - num_single_frame_decoder=1, - operation_order=[ - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine' - ], - temp_graph_model=dict( - type='MultiheadAttention', - embed_dims=512, - num_heads=8, - batch_first=True, - dropout=0.1), - graph_model=dict( - type='MultiheadAttention', - embed_dims=512, - num_heads=8, - batch_first=True, - dropout=0.1), - norm_layer=dict(type='LN', normalized_shape=256), - ffn=dict( - type='AsymmetricFFN', - in_channels=512, - pre_norm=dict(type='LN'), - embed_dims=256, - feedforward_channels=1024, - num_fcs=2, - ffn_drop=0.1, - act_cfg=dict(type='ReLU', inplace=True)), - deformable_model=dict( - type='DeformableFeatureAggregation', - embed_dims=256, - num_groups=8, - num_levels=4, - num_cams=6, - attn_drop=0.15, - use_deformable_func=True, - use_camera_embed=True, - residual_mode='cat', - kps_generator=dict( - type='SparseBox3DKeyPointsGenerator', - num_learnable_pts=6, - fix_scale=[[0, 0, 0], [0.45, 0, 0], [-0.45, 0, 0], - [0, 0.45, 0], [0, -0.45, 0], [0, 0, 0.45], - [0, 0, -0.45]])), - refine_layer=dict( - type='SparseBox3DRefinementModule', - embed_dims=256, - num_cls=10, - refine_yaw=True, - with_quality_estimation=True), - sampler=dict( - type='SparseBox3DTarget', - num_dn_groups=5, - num_temp_dn_groups=3, - dn_noise_scale=[2.0, 2.0, 2.0, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5], - max_dn_gt=32, - add_neg_dn=True, - cls_weight=2.0, - box_weight=0.25, - reg_weights=[2.0, 2.0, 2.0, 0.5, 0.5, 0.5, 0.0, 0.0, 0.0, 0.0], - cls_wise_reg_weights=dict( - {9: [2.0, 2.0, 2.0, 1.0, 1.0, 1.0, 0.0, 0.0, 1.0, 1.0]})), - loss_cls=dict( - type='FocalLoss', - use_sigmoid=True, - gamma=2.0, - alpha=0.25, - loss_weight=2.0), - loss_reg=dict( - type='SparseBox3DLoss', - loss_box=dict(type='L1Loss', loss_weight=0.25), - loss_centerness=dict(type='CrossEntropyLoss', use_sigmoid=True), - loss_yawness=dict(type='GaussianFocalLoss'), - cls_allow_reverse=[5]), - decoder=dict(type='SparseBox3DDecoder'), - reg_weights=[2.0, 2.0, 2.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0])) -dataset_type = 'NuScenes3DDetTrackDataset' -data_root = 'data/nuscenes/' -anno_root = 'data/nuscenes_anno_pkls/' -file_client_args = dict(backend='disk') -img_norm_cfg = dict( - mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375], to_rgb=True) -train_pipeline = [ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict( - type='LoadPointsFromFile', - coord_type='LIDAR', - load_dim=5, - use_dim=5, - file_client_args=dict(backend='disk')), - dict(type='ResizeCropFlipImage'), - dict(type='MultiScaleDepthMapGenerator', downsample=[4, 8, 16]), - dict(type='BBoxRotation'), - dict(type='PhotoMetricDistortionMultiViewImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict( - type='CircleObjectRangeFilter', - class_dist_thred=[55, 55, 55, 55, 55, 55, 55, 55, 55, 55]), - dict( - type='InstanceNameFilter', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ]), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=[ - 'img', 'timestamp', 'projection_mat', 'image_wh', 'gt_depth', - 'focal', 'gt_bboxes_3d', 'gt_labels_3d' - ], - meta_keys=['T_global', 'T_global_inv', 'timestamp', 'instance_id']) -] -test_pipeline = [ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='ResizeCropFlipImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=['img', 'timestamp', 'projection_mat', 'image_wh'], - meta_keys=['T_global', 'T_global_inv', 'timestamp']) -] -input_modality = dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False) -data_basic_config = dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier', - 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval') -data_aug_conf = dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]) -data = dict( - samples_per_gpu=20, - workers_per_gpu=20, - train=dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval', - ann_file='data/nuscenes_anno_pkls/nuscenes_infos_train.pkl', - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict( - type='LoadPointsFromFile', - coord_type='LIDAR', - load_dim=5, - use_dim=5, - file_client_args=dict(backend='disk')), - dict(type='ResizeCropFlipImage'), - dict(type='MultiScaleDepthMapGenerator', downsample=[4, 8, 16]), - dict(type='BBoxRotation'), - dict(type='PhotoMetricDistortionMultiViewImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict( - type='CircleObjectRangeFilter', - class_dist_thred=[55, 55, 55, 55, 55, 55, 55, 55, 55, 55]), - dict( - type='InstanceNameFilter', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', - 'traffic_cone' - ]), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=[ - 'img', 'timestamp', 'projection_mat', 'image_wh', - 'gt_depth', 'focal', 'gt_bboxes_3d', 'gt_labels_3d' - ], - meta_keys=[ - 'T_global', 'T_global_inv', 'timestamp', 'instance_id' - ]) - ], - test_mode=False, - data_aug_conf=dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]), - with_seq_flag=True, - sequences_split_num=2, - keep_consistent_seq_aug=True), - val=dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval', - ann_file='data/nuscenes_anno_pkls/nuscenes_infos_val.pkl', - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='ResizeCropFlipImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=['img', 'timestamp', 'projection_mat', 'image_wh'], - meta_keys=['T_global', 'T_global_inv', 'timestamp']) - ], - data_aug_conf=dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]), - test_mode=True, - tracking=True, - tracking_threshold=0.2), - test=dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval', - ann_file='data/nuscenes_anno_pkls/nuscenes_infos_val.pkl', - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='ResizeCropFlipImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=['img', 'timestamp', 'projection_mat', 'image_wh'], - meta_keys=['T_global', 'T_global_inv', 'timestamp']) - ], - data_aug_conf=dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]), - test_mode=True, - tracking=True, - tracking_threshold=0.2)) -optimizer = dict( - type='Miopen_AdamW', - lr=0.0006, - weight_decay=0.001, - paramwise_cfg=dict(custom_keys=dict(img_backbone=dict(lr_mult=0.5)))) -optimizer_config = dict(grad_clip=dict(max_norm=25, norm_type=2)) -lr_config = dict( - policy='CosineAnnealing', - warmup='linear', - warmup_iters=500, - warmup_ratio=0.3333333333333333, - min_lr_ratio=0.001) -runner = dict(type='IterBasedRunner', max_iters=17500) -vis_pipeline = [ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='Collect', keys=['img'], meta_keys=['timestamp', 'lidar2img']) -] -evaluation = dict( - interval=3500, - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict( - type='Collect', keys=['img'], meta_keys=['timestamp', 'lidar2img']) - ]) -gpu_ids = range(0, 8) - -2026-01-14 20:27:45,956 - mmdet - INFO - Set random seed to 0, deterministic: False diff --git a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20260114_204317.log b/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20260114_204317.log deleted file mode 100644 index 21dc612e748ef492bc2caa483e10b0329166bba8..0000000000000000000000000000000000000000 --- a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20260114_204317.log +++ /dev/null @@ -1,469 +0,0 @@ -2026-01-14 20:43:17,446 - mmdet - INFO - Environment info: ------------------------------------------------------------- -sys.platform: linux -Python: 3.10.12 (main, Aug 15 2025, 14:32:43) [GCC 11.4.0] -CUDA available: True -GPU 0,1,2,3,4,5,6,7: BW151 -CUDA_HOME: /opt/dtk/cuda/cuda -NVCC: Cuda compilation tools, release 12.6, V12.6.77 -clang version 17.0.0 -Target: x86_64-unknown-linux-gnu -Thread model: posix -InstalledDir: /opt/dtk-25.04.4/llvm/bi -GCC: x86_64-linux-gnu-gcc (Ubuntu 11.4.0-1ubuntu1~22.04) 11.4.0 -PyTorch: 2.5.1 -PyTorch compiling details: PyTorch built with: - - GCC 10.3 - - C++ Version: 201703 - - Intel(R) Math Kernel Library Version 2020.0.4 Product Build 20200917 for Intel(R) 64 architecture applications - - OpenMP 201511 (a.k.a. OpenMP 4.5) - - LAPACK is enabled (usually provided by MKL) - - NNPACK is enabled - - CPU capability usage: AVX512 - - HIP Runtime 6.3.25521 - - MIOpen 2.18.0 - - Magma 2.8.0 - - Build settings: BLAS_INFO=mkl, BUILD_TYPE=Release, CXX_COMPILER=/opt/rh/gcc-toolset-10/root/usr/bin/c++, CXX_FLAGS= -D_GLIBCXX_USE_CXX11_ABI=1 -fvisibility-inlines-hidden -DUSE_PTHREADPOOL -DNDEBUG -DUSE_KINETO -DLIBKINETO_NOCUPTI -DLIBKINETO_NOXPUPTI=ON -DUSE_FBGEMM -DUSE_PYTORCH_QNNPACK -DUSE_XNNPACK -DSYMBOLICATE_MOBILE_DEBUG_HANDLE -O2 -fPIC -Wall -Wextra -Werror=return-type -Werror=non-virtual-dtor -Werror=bool-operation -Wnarrowing -Wno-missing-field-initializers -Wno-type-limits -Wno-array-bounds -Wno-unknown-pragmas -Wno-unused-parameter -Wno-strict-overflow -Wno-strict-aliasing -Wno-stringop-overflow -Wsuggest-override -Wno-psabi -Wno-error=old-style-cast -Wno-missing-braces -fdiagnostics-color=always -faligned-new -Wno-unused-but-set-variable -Wno-maybe-uninitialized -fno-math-errno -fno-trapping-math -Werror=format -Wno-stringop-overflow, FORCE_FALLBACK_CUDA_MPI=1, LAPACK_INFO=mkl, PERF_WITH_AVX=1, PERF_WITH_AVX2=1, TORCH_VERSION=2.5.1, USE_CUDA=0, USE_CUDNN=OFF, USE_CUSPARSELT=OFF, USE_EXCEPTION_PTR=1, USE_GFLAGS=1, USE_GLOG=1, USE_GLOO=1, USE_MKL=ON, USE_MKLDNN=0, USE_MPI=1, USE_NCCL=1, USE_NNPACK=ON, USE_OPENMP=1, USE_ROCM=ON, USE_ROCM_KERNEL_ASSERT=OFF, - -TorchVision: 0.20.1 -OpenCV: 4.12.0 -MMCV: 1.6.1 -MMCV Compiler: GCC 10.3 -MMCV CUDA Compiler: rocm not available -MMDetection: 2.25.1+c41df4b ------------------------------------------------------------- - -2026-01-14 20:43:18,200 - mmdet - INFO - Distributed training: True -2026-01-14 20:43:19,132 - mmdet - INFO - Config: -plugin = True -plugin_dir = 'projects/mmdet3d_plugin/' -dist_params = dict(backend='nccl') -log_level = 'INFO' -work_dir = './work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704' -total_batch_size = 160 -num_gpus = 8 -batch_size = 20 -num_iters_per_epoch = 175 -num_epochs = 100 -checkpoint_epoch_interval = 20 -checkpoint_config = dict(interval=3500) -log_config = dict( - interval=1, - hooks=[ - dict(type='TextLoggerHook', by_epoch=False), - dict(type='TensorboardLoggerHook') - ]) -load_from = None -resume_from = None -workflow = [('train', 1)] -fp16 = dict(loss_scale=32.0) -input_shape = (704, 256) -tracking_test = True -tracking_threshold = 0.2 -class_names = [ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier', - 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' -] -num_classes = 10 -embed_dims = 256 -num_groups = 8 -num_decoder = 6 -num_single_frame_decoder = 1 -use_deformable_func = True -strides = [4, 8, 16, 32] -num_levels = 4 -num_depth_layers = 3 -drop_out = 0.1 -temporal = True -decouple_attn = True -with_quality_estimation = True -model = dict( - type='Sparse4D', - use_grid_mask=True, - use_deformable_func=True, - img_backbone=dict( - type='ResNet', - depth=50, - num_stages=4, - frozen_stages=-1, - norm_eval=False, - style='pytorch', - with_cp=True, - out_indices=(0, 1, 2, 3), - norm_cfg=dict(type='BN', requires_grad=True), - pretrained='ckpt/resnet50-19c8e357.pth'), - img_neck=dict( - type='FPN', - num_outs=4, - start_level=0, - out_channels=256, - add_extra_convs='on_output', - relu_before_extra_convs=True, - in_channels=[256, 512, 1024, 2048]), - depth_branch=dict( - type='DenseDepthNet', - embed_dims=256, - num_depth_layers=3, - loss_weight=0.2), - head=dict( - type='Sparse4DHead', - cls_threshold_to_reg=0.05, - decouple_attn=True, - instance_bank=dict( - type='InstanceBank', - num_anchor=900, - embed_dims=256, - anchor='nuscenes_kmeans900.npy', - anchor_handler=dict(type='SparseBox3DKeyPointsGenerator'), - num_temp_instances=600, - confidence_decay=0.6, - feat_grad=False), - anchor_encoder=dict( - type='SparseBox3DEncoder', - vel_dims=3, - embed_dims=[128, 32, 32, 64], - mode='cat', - output_fc=False, - in_loops=1, - out_loops=4), - num_single_frame_decoder=1, - operation_order=[ - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine' - ], - temp_graph_model=dict( - type='MultiheadAttention', - embed_dims=512, - num_heads=8, - batch_first=True, - dropout=0.1), - graph_model=dict( - type='MultiheadAttention', - embed_dims=512, - num_heads=8, - batch_first=True, - dropout=0.1), - norm_layer=dict(type='LN', normalized_shape=256), - ffn=dict( - type='AsymmetricFFN', - in_channels=512, - pre_norm=dict(type='LN'), - embed_dims=256, - feedforward_channels=1024, - num_fcs=2, - ffn_drop=0.1, - act_cfg=dict(type='ReLU', inplace=True)), - deformable_model=dict( - type='DeformableFeatureAggregation', - embed_dims=256, - num_groups=8, - num_levels=4, - num_cams=6, - attn_drop=0.15, - use_deformable_func=True, - use_camera_embed=True, - residual_mode='cat', - kps_generator=dict( - type='SparseBox3DKeyPointsGenerator', - num_learnable_pts=6, - fix_scale=[[0, 0, 0], [0.45, 0, 0], [-0.45, 0, 0], - [0, 0.45, 0], [0, -0.45, 0], [0, 0, 0.45], - [0, 0, -0.45]])), - refine_layer=dict( - type='SparseBox3DRefinementModule', - embed_dims=256, - num_cls=10, - refine_yaw=True, - with_quality_estimation=True), - sampler=dict( - type='SparseBox3DTarget', - num_dn_groups=5, - num_temp_dn_groups=3, - dn_noise_scale=[2.0, 2.0, 2.0, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5], - max_dn_gt=32, - add_neg_dn=True, - cls_weight=2.0, - box_weight=0.25, - reg_weights=[2.0, 2.0, 2.0, 0.5, 0.5, 0.5, 0.0, 0.0, 0.0, 0.0], - cls_wise_reg_weights=dict( - {9: [2.0, 2.0, 2.0, 1.0, 1.0, 1.0, 0.0, 0.0, 1.0, 1.0]})), - loss_cls=dict( - type='FocalLoss', - use_sigmoid=True, - gamma=2.0, - alpha=0.25, - loss_weight=2.0), - loss_reg=dict( - type='SparseBox3DLoss', - loss_box=dict(type='L1Loss', loss_weight=0.25), - loss_centerness=dict(type='CrossEntropyLoss', use_sigmoid=True), - loss_yawness=dict(type='GaussianFocalLoss'), - cls_allow_reverse=[5]), - decoder=dict(type='SparseBox3DDecoder'), - reg_weights=[2.0, 2.0, 2.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0])) -dataset_type = 'NuScenes3DDetTrackDataset' -data_root = 'data/nuscenes/' -anno_root = 'data/nuscenes_anno_pkls/' -file_client_args = dict(backend='disk') -img_norm_cfg = dict( - mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375], to_rgb=True) -train_pipeline = [ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict( - type='LoadPointsFromFile', - coord_type='LIDAR', - load_dim=5, - use_dim=5, - file_client_args=dict(backend='disk')), - dict(type='ResizeCropFlipImage'), - dict(type='MultiScaleDepthMapGenerator', downsample=[4, 8, 16]), - dict(type='BBoxRotation'), - dict(type='PhotoMetricDistortionMultiViewImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict( - type='CircleObjectRangeFilter', - class_dist_thred=[55, 55, 55, 55, 55, 55, 55, 55, 55, 55]), - dict( - type='InstanceNameFilter', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ]), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=[ - 'img', 'timestamp', 'projection_mat', 'image_wh', 'gt_depth', - 'focal', 'gt_bboxes_3d', 'gt_labels_3d' - ], - meta_keys=['T_global', 'T_global_inv', 'timestamp', 'instance_id']) -] -test_pipeline = [ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='ResizeCropFlipImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=['img', 'timestamp', 'projection_mat', 'image_wh'], - meta_keys=['T_global', 'T_global_inv', 'timestamp']) -] -input_modality = dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False) -data_basic_config = dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier', - 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval') -data_aug_conf = dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]) -data = dict( - samples_per_gpu=20, - workers_per_gpu=20, - train=dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval', - ann_file='data/nuscenes_anno_pkls/nuscenes_infos_train.pkl', - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict( - type='LoadPointsFromFile', - coord_type='LIDAR', - load_dim=5, - use_dim=5, - file_client_args=dict(backend='disk')), - dict(type='ResizeCropFlipImage'), - dict(type='MultiScaleDepthMapGenerator', downsample=[4, 8, 16]), - dict(type='BBoxRotation'), - dict(type='PhotoMetricDistortionMultiViewImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict( - type='CircleObjectRangeFilter', - class_dist_thred=[55, 55, 55, 55, 55, 55, 55, 55, 55, 55]), - dict( - type='InstanceNameFilter', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', - 'traffic_cone' - ]), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=[ - 'img', 'timestamp', 'projection_mat', 'image_wh', - 'gt_depth', 'focal', 'gt_bboxes_3d', 'gt_labels_3d' - ], - meta_keys=[ - 'T_global', 'T_global_inv', 'timestamp', 'instance_id' - ]) - ], - test_mode=False, - data_aug_conf=dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]), - with_seq_flag=True, - sequences_split_num=2, - keep_consistent_seq_aug=True), - val=dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval', - ann_file='data/nuscenes_anno_pkls/nuscenes_infos_val.pkl', - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='ResizeCropFlipImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=['img', 'timestamp', 'projection_mat', 'image_wh'], - meta_keys=['T_global', 'T_global_inv', 'timestamp']) - ], - data_aug_conf=dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]), - test_mode=True, - tracking=True, - tracking_threshold=0.2), - test=dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval', - ann_file='data/nuscenes_anno_pkls/nuscenes_infos_val.pkl', - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='ResizeCropFlipImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=['img', 'timestamp', 'projection_mat', 'image_wh'], - meta_keys=['T_global', 'T_global_inv', 'timestamp']) - ], - data_aug_conf=dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]), - test_mode=True, - tracking=True, - tracking_threshold=0.2)) -optimizer = dict( - type='Miopen_AdamW', - lr=0.0006, - weight_decay=0.001, - paramwise_cfg=dict(custom_keys=dict(img_backbone=dict(lr_mult=0.5)))) -optimizer_config = dict(grad_clip=dict(max_norm=25, norm_type=2)) -lr_config = dict( - policy='CosineAnnealing', - warmup='linear', - warmup_iters=500, - warmup_ratio=0.3333333333333333, - min_lr_ratio=0.001) -runner = dict(type='IterBasedRunner', max_iters=17500) -vis_pipeline = [ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='Collect', keys=['img'], meta_keys=['timestamp', 'lidar2img']) -] -evaluation = dict( - interval=3500, - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict( - type='Collect', keys=['img'], meta_keys=['timestamp', 'lidar2img']) - ]) -gpu_ids = range(0, 8) - -2026-01-14 20:43:19,133 - mmdet - INFO - Set random seed to 0, deterministic: False diff --git a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20260114_205231.log b/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20260114_205231.log deleted file mode 100644 index 90c7f54162baa2509d17f225c6b433060d908545..0000000000000000000000000000000000000000 --- a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20260114_205231.log +++ /dev/null @@ -1,465 +0,0 @@ -2026-01-14 20:52:31,502 - mmdet - INFO - Environment info: ------------------------------------------------------------- -sys.platform: linux -Python: 3.10.12 (main, Aug 15 2025, 14:32:43) [GCC 11.4.0] -CUDA available: True -GPU 0,1,2,3,4,5,6,7: BW151 -CUDA_HOME: /opt/dtk -NVCC: Not Available -GCC: x86_64-linux-gnu-gcc (Ubuntu 11.4.0-1ubuntu1~22.04) 11.4.0 -PyTorch: 2.5.1 -PyTorch compiling details: PyTorch built with: - - GCC 10.3 - - C++ Version: 201703 - - Intel(R) Math Kernel Library Version 2020.0.4 Product Build 20200917 for Intel(R) 64 architecture applications - - OpenMP 201511 (a.k.a. OpenMP 4.5) - - LAPACK is enabled (usually provided by MKL) - - NNPACK is enabled - - CPU capability usage: AVX512 - - HIP Runtime 6.3.25521 - - MIOpen 2.18.0 - - Magma 2.8.0 - - Build settings: BLAS_INFO=mkl, BUILD_TYPE=Release, CXX_COMPILER=/opt/rh/gcc-toolset-10/root/usr/bin/c++, CXX_FLAGS= -D_GLIBCXX_USE_CXX11_ABI=1 -fvisibility-inlines-hidden -DUSE_PTHREADPOOL -DNDEBUG -DUSE_KINETO -DLIBKINETO_NOCUPTI -DLIBKINETO_NOXPUPTI=ON -DUSE_FBGEMM -DUSE_PYTORCH_QNNPACK -DUSE_XNNPACK -DSYMBOLICATE_MOBILE_DEBUG_HANDLE -O2 -fPIC -Wall -Wextra -Werror=return-type -Werror=non-virtual-dtor -Werror=bool-operation -Wnarrowing -Wno-missing-field-initializers -Wno-type-limits -Wno-array-bounds -Wno-unknown-pragmas -Wno-unused-parameter -Wno-strict-overflow -Wno-strict-aliasing -Wno-stringop-overflow -Wsuggest-override -Wno-psabi -Wno-error=old-style-cast -Wno-missing-braces -fdiagnostics-color=always -faligned-new -Wno-unused-but-set-variable -Wno-maybe-uninitialized -fno-math-errno -fno-trapping-math -Werror=format -Wno-stringop-overflow, FORCE_FALLBACK_CUDA_MPI=1, LAPACK_INFO=mkl, PERF_WITH_AVX=1, PERF_WITH_AVX2=1, TORCH_VERSION=2.5.1, USE_CUDA=0, USE_CUDNN=OFF, USE_CUSPARSELT=OFF, USE_EXCEPTION_PTR=1, USE_GFLAGS=1, USE_GLOG=1, USE_GLOO=1, USE_MKL=ON, USE_MKLDNN=0, USE_MPI=1, USE_NCCL=1, USE_NNPACK=ON, USE_OPENMP=1, USE_ROCM=ON, USE_ROCM_KERNEL_ASSERT=OFF, - -TorchVision: 0.20.1 -OpenCV: 4.12.0 -MMCV: 1.6.1 -MMCV Compiler: GCC 10.3 -MMCV CUDA Compiler: rocm not available -MMDetection: 2.25.1+c41df4b ------------------------------------------------------------- - -2026-01-14 20:52:32,248 - mmdet - INFO - Distributed training: True -2026-01-14 20:52:33,165 - mmdet - INFO - Config: -plugin = True -plugin_dir = 'projects/mmdet3d_plugin/' -dist_params = dict(backend='nccl') -log_level = 'INFO' -work_dir = './work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704' -total_batch_size = 160 -num_gpus = 8 -batch_size = 20 -num_iters_per_epoch = 175 -num_epochs = 100 -checkpoint_epoch_interval = 20 -checkpoint_config = dict(interval=3500) -log_config = dict( - interval=1, - hooks=[ - dict(type='TextLoggerHook', by_epoch=False), - dict(type='TensorboardLoggerHook') - ]) -load_from = None -resume_from = None -workflow = [('train', 1)] -fp16 = dict(loss_scale=32.0) -input_shape = (704, 256) -tracking_test = True -tracking_threshold = 0.2 -class_names = [ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier', - 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' -] -num_classes = 10 -embed_dims = 256 -num_groups = 8 -num_decoder = 6 -num_single_frame_decoder = 1 -use_deformable_func = True -strides = [4, 8, 16, 32] -num_levels = 4 -num_depth_layers = 3 -drop_out = 0.1 -temporal = True -decouple_attn = True -with_quality_estimation = True -model = dict( - type='Sparse4D', - use_grid_mask=True, - use_deformable_func=True, - img_backbone=dict( - type='ResNet', - depth=50, - num_stages=4, - frozen_stages=-1, - norm_eval=False, - style='pytorch', - with_cp=True, - out_indices=(0, 1, 2, 3), - norm_cfg=dict(type='BN', requires_grad=True), - pretrained='ckpt/resnet50-19c8e357.pth'), - img_neck=dict( - type='FPN', - num_outs=4, - start_level=0, - out_channels=256, - add_extra_convs='on_output', - relu_before_extra_convs=True, - in_channels=[256, 512, 1024, 2048]), - depth_branch=dict( - type='DenseDepthNet', - embed_dims=256, - num_depth_layers=3, - loss_weight=0.2), - head=dict( - type='Sparse4DHead', - cls_threshold_to_reg=0.05, - decouple_attn=True, - instance_bank=dict( - type='InstanceBank', - num_anchor=900, - embed_dims=256, - anchor='nuscenes_kmeans900.npy', - anchor_handler=dict(type='SparseBox3DKeyPointsGenerator'), - num_temp_instances=600, - confidence_decay=0.6, - feat_grad=False), - anchor_encoder=dict( - type='SparseBox3DEncoder', - vel_dims=3, - embed_dims=[128, 32, 32, 64], - mode='cat', - output_fc=False, - in_loops=1, - out_loops=4), - num_single_frame_decoder=1, - operation_order=[ - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine' - ], - temp_graph_model=dict( - type='MultiheadAttention', - embed_dims=512, - num_heads=8, - batch_first=True, - dropout=0.1), - graph_model=dict( - type='MultiheadAttention', - embed_dims=512, - num_heads=8, - batch_first=True, - dropout=0.1), - norm_layer=dict(type='LN', normalized_shape=256), - ffn=dict( - type='AsymmetricFFN', - in_channels=512, - pre_norm=dict(type='LN'), - embed_dims=256, - feedforward_channels=1024, - num_fcs=2, - ffn_drop=0.1, - act_cfg=dict(type='ReLU', inplace=True)), - deformable_model=dict( - type='DeformableFeatureAggregation', - embed_dims=256, - num_groups=8, - num_levels=4, - num_cams=6, - attn_drop=0.15, - use_deformable_func=True, - use_camera_embed=True, - residual_mode='cat', - kps_generator=dict( - type='SparseBox3DKeyPointsGenerator', - num_learnable_pts=6, - fix_scale=[[0, 0, 0], [0.45, 0, 0], [-0.45, 0, 0], - [0, 0.45, 0], [0, -0.45, 0], [0, 0, 0.45], - [0, 0, -0.45]])), - refine_layer=dict( - type='SparseBox3DRefinementModule', - embed_dims=256, - num_cls=10, - refine_yaw=True, - with_quality_estimation=True), - sampler=dict( - type='SparseBox3DTarget', - num_dn_groups=5, - num_temp_dn_groups=3, - dn_noise_scale=[2.0, 2.0, 2.0, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5], - max_dn_gt=32, - add_neg_dn=True, - cls_weight=2.0, - box_weight=0.25, - reg_weights=[2.0, 2.0, 2.0, 0.5, 0.5, 0.5, 0.0, 0.0, 0.0, 0.0], - cls_wise_reg_weights=dict( - {9: [2.0, 2.0, 2.0, 1.0, 1.0, 1.0, 0.0, 0.0, 1.0, 1.0]})), - loss_cls=dict( - type='FocalLoss', - use_sigmoid=True, - gamma=2.0, - alpha=0.25, - loss_weight=2.0), - loss_reg=dict( - type='SparseBox3DLoss', - loss_box=dict(type='L1Loss', loss_weight=0.25), - loss_centerness=dict(type='CrossEntropyLoss', use_sigmoid=True), - loss_yawness=dict(type='GaussianFocalLoss'), - cls_allow_reverse=[5]), - decoder=dict(type='SparseBox3DDecoder'), - reg_weights=[2.0, 2.0, 2.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0])) -dataset_type = 'NuScenes3DDetTrackDataset' -data_root = 'data/nuscenes/' -anno_root = 'data/nuscenes_anno_pkls/' -file_client_args = dict(backend='disk') -img_norm_cfg = dict( - mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375], to_rgb=True) -train_pipeline = [ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict( - type='LoadPointsFromFile', - coord_type='LIDAR', - load_dim=5, - use_dim=5, - file_client_args=dict(backend='disk')), - dict(type='ResizeCropFlipImage'), - dict(type='MultiScaleDepthMapGenerator', downsample=[4, 8, 16]), - dict(type='BBoxRotation'), - dict(type='PhotoMetricDistortionMultiViewImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict( - type='CircleObjectRangeFilter', - class_dist_thred=[55, 55, 55, 55, 55, 55, 55, 55, 55, 55]), - dict( - type='InstanceNameFilter', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ]), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=[ - 'img', 'timestamp', 'projection_mat', 'image_wh', 'gt_depth', - 'focal', 'gt_bboxes_3d', 'gt_labels_3d' - ], - meta_keys=['T_global', 'T_global_inv', 'timestamp', 'instance_id']) -] -test_pipeline = [ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='ResizeCropFlipImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=['img', 'timestamp', 'projection_mat', 'image_wh'], - meta_keys=['T_global', 'T_global_inv', 'timestamp']) -] -input_modality = dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False) -data_basic_config = dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier', - 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval') -data_aug_conf = dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]) -data = dict( - samples_per_gpu=20, - workers_per_gpu=20, - train=dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval', - ann_file='data/nuscenes_anno_pkls/nuscenes_infos_train.pkl', - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict( - type='LoadPointsFromFile', - coord_type='LIDAR', - load_dim=5, - use_dim=5, - file_client_args=dict(backend='disk')), - dict(type='ResizeCropFlipImage'), - dict(type='MultiScaleDepthMapGenerator', downsample=[4, 8, 16]), - dict(type='BBoxRotation'), - dict(type='PhotoMetricDistortionMultiViewImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict( - type='CircleObjectRangeFilter', - class_dist_thred=[55, 55, 55, 55, 55, 55, 55, 55, 55, 55]), - dict( - type='InstanceNameFilter', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', - 'traffic_cone' - ]), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=[ - 'img', 'timestamp', 'projection_mat', 'image_wh', - 'gt_depth', 'focal', 'gt_bboxes_3d', 'gt_labels_3d' - ], - meta_keys=[ - 'T_global', 'T_global_inv', 'timestamp', 'instance_id' - ]) - ], - test_mode=False, - data_aug_conf=dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]), - with_seq_flag=True, - sequences_split_num=2, - keep_consistent_seq_aug=True), - val=dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval', - ann_file='data/nuscenes_anno_pkls/nuscenes_infos_val.pkl', - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='ResizeCropFlipImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=['img', 'timestamp', 'projection_mat', 'image_wh'], - meta_keys=['T_global', 'T_global_inv', 'timestamp']) - ], - data_aug_conf=dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]), - test_mode=True, - tracking=True, - tracking_threshold=0.2), - test=dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval', - ann_file='data/nuscenes_anno_pkls/nuscenes_infos_val.pkl', - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='ResizeCropFlipImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=['img', 'timestamp', 'projection_mat', 'image_wh'], - meta_keys=['T_global', 'T_global_inv', 'timestamp']) - ], - data_aug_conf=dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]), - test_mode=True, - tracking=True, - tracking_threshold=0.2)) -optimizer = dict( - type='Miopen_AdamW', - lr=0.0006, - weight_decay=0.001, - paramwise_cfg=dict(custom_keys=dict(img_backbone=dict(lr_mult=0.5)))) -optimizer_config = dict(grad_clip=dict(max_norm=25, norm_type=2)) -lr_config = dict( - policy='CosineAnnealing', - warmup='linear', - warmup_iters=500, - warmup_ratio=0.3333333333333333, - min_lr_ratio=0.001) -runner = dict(type='IterBasedRunner', max_iters=17500) -vis_pipeline = [ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='Collect', keys=['img'], meta_keys=['timestamp', 'lidar2img']) -] -evaluation = dict( - interval=3500, - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict( - type='Collect', keys=['img'], meta_keys=['timestamp', 'lidar2img']) - ]) -gpu_ids = range(0, 8) - -2026-01-14 20:52:33,166 - mmdet - INFO - Set random seed to 0, deterministic: False diff --git a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20260114_205352.log b/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20260114_205352.log deleted file mode 100644 index 7bcdf5fbed2ce29c0e7133f7f3a2e439066665dc..0000000000000000000000000000000000000000 --- a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20260114_205352.log +++ /dev/null @@ -1,465 +0,0 @@ -2026-01-14 20:53:52,399 - mmdet - INFO - Environment info: ------------------------------------------------------------- -sys.platform: linux -Python: 3.10.12 (main, Aug 15 2025, 14:32:43) [GCC 11.4.0] -CUDA available: True -GPU 0,1,2,3,4,5,6,7: BW151 -CUDA_HOME: /opt/dtk -NVCC: Not Available -GCC: x86_64-linux-gnu-gcc (Ubuntu 11.4.0-1ubuntu1~22.04) 11.4.0 -PyTorch: 2.5.1 -PyTorch compiling details: PyTorch built with: - - GCC 10.3 - - C++ Version: 201703 - - Intel(R) Math Kernel Library Version 2020.0.4 Product Build 20200917 for Intel(R) 64 architecture applications - - OpenMP 201511 (a.k.a. OpenMP 4.5) - - LAPACK is enabled (usually provided by MKL) - - NNPACK is enabled - - CPU capability usage: AVX512 - - HIP Runtime 6.3.25521 - - MIOpen 2.18.0 - - Magma 2.8.0 - - Build settings: BLAS_INFO=mkl, BUILD_TYPE=Release, CXX_COMPILER=/opt/rh/gcc-toolset-10/root/usr/bin/c++, CXX_FLAGS= -D_GLIBCXX_USE_CXX11_ABI=1 -fvisibility-inlines-hidden -DUSE_PTHREADPOOL -DNDEBUG -DUSE_KINETO -DLIBKINETO_NOCUPTI -DLIBKINETO_NOXPUPTI=ON -DUSE_FBGEMM -DUSE_PYTORCH_QNNPACK -DUSE_XNNPACK -DSYMBOLICATE_MOBILE_DEBUG_HANDLE -O2 -fPIC -Wall -Wextra -Werror=return-type -Werror=non-virtual-dtor -Werror=bool-operation -Wnarrowing -Wno-missing-field-initializers -Wno-type-limits -Wno-array-bounds -Wno-unknown-pragmas -Wno-unused-parameter -Wno-strict-overflow -Wno-strict-aliasing -Wno-stringop-overflow -Wsuggest-override -Wno-psabi -Wno-error=old-style-cast -Wno-missing-braces -fdiagnostics-color=always -faligned-new -Wno-unused-but-set-variable -Wno-maybe-uninitialized -fno-math-errno -fno-trapping-math -Werror=format -Wno-stringop-overflow, FORCE_FALLBACK_CUDA_MPI=1, LAPACK_INFO=mkl, PERF_WITH_AVX=1, PERF_WITH_AVX2=1, TORCH_VERSION=2.5.1, USE_CUDA=0, USE_CUDNN=OFF, USE_CUSPARSELT=OFF, USE_EXCEPTION_PTR=1, USE_GFLAGS=1, USE_GLOG=1, USE_GLOO=1, USE_MKL=ON, USE_MKLDNN=0, USE_MPI=1, USE_NCCL=1, USE_NNPACK=ON, USE_OPENMP=1, USE_ROCM=ON, USE_ROCM_KERNEL_ASSERT=OFF, - -TorchVision: 0.20.1 -OpenCV: 4.12.0 -MMCV: 1.6.1 -MMCV Compiler: GCC 10.3 -MMCV CUDA Compiler: rocm not available -MMDetection: 2.25.1+c41df4b ------------------------------------------------------------- - -2026-01-14 20:53:53,141 - mmdet - INFO - Distributed training: True -2026-01-14 20:53:54,061 - mmdet - INFO - Config: -plugin = True -plugin_dir = 'projects/mmdet3d_plugin/' -dist_params = dict(backend='nccl') -log_level = 'INFO' -work_dir = './work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704' -total_batch_size = 160 -num_gpus = 8 -batch_size = 20 -num_iters_per_epoch = 175 -num_epochs = 100 -checkpoint_epoch_interval = 20 -checkpoint_config = dict(interval=3500) -log_config = dict( - interval=1, - hooks=[ - dict(type='TextLoggerHook', by_epoch=False), - dict(type='TensorboardLoggerHook') - ]) -load_from = None -resume_from = None -workflow = [('train', 1)] -fp16 = dict(loss_scale=32.0) -input_shape = (704, 256) -tracking_test = True -tracking_threshold = 0.2 -class_names = [ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier', - 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' -] -num_classes = 10 -embed_dims = 256 -num_groups = 8 -num_decoder = 6 -num_single_frame_decoder = 1 -use_deformable_func = True -strides = [4, 8, 16, 32] -num_levels = 4 -num_depth_layers = 3 -drop_out = 0.1 -temporal = True -decouple_attn = True -with_quality_estimation = True -model = dict( - type='Sparse4D', - use_grid_mask=True, - use_deformable_func=True, - img_backbone=dict( - type='ResNet', - depth=50, - num_stages=4, - frozen_stages=-1, - norm_eval=False, - style='pytorch', - with_cp=True, - out_indices=(0, 1, 2, 3), - norm_cfg=dict(type='BN', requires_grad=True), - pretrained='ckpt/resnet50-19c8e357.pth'), - img_neck=dict( - type='FPN', - num_outs=4, - start_level=0, - out_channels=256, - add_extra_convs='on_output', - relu_before_extra_convs=True, - in_channels=[256, 512, 1024, 2048]), - depth_branch=dict( - type='DenseDepthNet', - embed_dims=256, - num_depth_layers=3, - loss_weight=0.2), - head=dict( - type='Sparse4DHead', - cls_threshold_to_reg=0.05, - decouple_attn=True, - instance_bank=dict( - type='InstanceBank', - num_anchor=900, - embed_dims=256, - anchor='nuscenes_kmeans900.npy', - anchor_handler=dict(type='SparseBox3DKeyPointsGenerator'), - num_temp_instances=600, - confidence_decay=0.6, - feat_grad=False), - anchor_encoder=dict( - type='SparseBox3DEncoder', - vel_dims=3, - embed_dims=[128, 32, 32, 64], - mode='cat', - output_fc=False, - in_loops=1, - out_loops=4), - num_single_frame_decoder=1, - operation_order=[ - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine' - ], - temp_graph_model=dict( - type='MultiheadAttention', - embed_dims=512, - num_heads=8, - batch_first=True, - dropout=0.1), - graph_model=dict( - type='MultiheadAttention', - embed_dims=512, - num_heads=8, - batch_first=True, - dropout=0.1), - norm_layer=dict(type='LN', normalized_shape=256), - ffn=dict( - type='AsymmetricFFN', - in_channels=512, - pre_norm=dict(type='LN'), - embed_dims=256, - feedforward_channels=1024, - num_fcs=2, - ffn_drop=0.1, - act_cfg=dict(type='ReLU', inplace=True)), - deformable_model=dict( - type='DeformableFeatureAggregation', - embed_dims=256, - num_groups=8, - num_levels=4, - num_cams=6, - attn_drop=0.15, - use_deformable_func=True, - use_camera_embed=True, - residual_mode='cat', - kps_generator=dict( - type='SparseBox3DKeyPointsGenerator', - num_learnable_pts=6, - fix_scale=[[0, 0, 0], [0.45, 0, 0], [-0.45, 0, 0], - [0, 0.45, 0], [0, -0.45, 0], [0, 0, 0.45], - [0, 0, -0.45]])), - refine_layer=dict( - type='SparseBox3DRefinementModule', - embed_dims=256, - num_cls=10, - refine_yaw=True, - with_quality_estimation=True), - sampler=dict( - type='SparseBox3DTarget', - num_dn_groups=5, - num_temp_dn_groups=3, - dn_noise_scale=[2.0, 2.0, 2.0, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5], - max_dn_gt=32, - add_neg_dn=True, - cls_weight=2.0, - box_weight=0.25, - reg_weights=[2.0, 2.0, 2.0, 0.5, 0.5, 0.5, 0.0, 0.0, 0.0, 0.0], - cls_wise_reg_weights=dict( - {9: [2.0, 2.0, 2.0, 1.0, 1.0, 1.0, 0.0, 0.0, 1.0, 1.0]})), - loss_cls=dict( - type='FocalLoss', - use_sigmoid=True, - gamma=2.0, - alpha=0.25, - loss_weight=2.0), - loss_reg=dict( - type='SparseBox3DLoss', - loss_box=dict(type='L1Loss', loss_weight=0.25), - loss_centerness=dict(type='CrossEntropyLoss', use_sigmoid=True), - loss_yawness=dict(type='GaussianFocalLoss'), - cls_allow_reverse=[5]), - decoder=dict(type='SparseBox3DDecoder'), - reg_weights=[2.0, 2.0, 2.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0])) -dataset_type = 'NuScenes3DDetTrackDataset' -data_root = 'data/nuscenes/' -anno_root = 'data/nuscenes_anno_pkls/' -file_client_args = dict(backend='disk') -img_norm_cfg = dict( - mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375], to_rgb=True) -train_pipeline = [ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict( - type='LoadPointsFromFile', - coord_type='LIDAR', - load_dim=5, - use_dim=5, - file_client_args=dict(backend='disk')), - dict(type='ResizeCropFlipImage'), - dict(type='MultiScaleDepthMapGenerator', downsample=[4, 8, 16]), - dict(type='BBoxRotation'), - dict(type='PhotoMetricDistortionMultiViewImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict( - type='CircleObjectRangeFilter', - class_dist_thred=[55, 55, 55, 55, 55, 55, 55, 55, 55, 55]), - dict( - type='InstanceNameFilter', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ]), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=[ - 'img', 'timestamp', 'projection_mat', 'image_wh', 'gt_depth', - 'focal', 'gt_bboxes_3d', 'gt_labels_3d' - ], - meta_keys=['T_global', 'T_global_inv', 'timestamp', 'instance_id']) -] -test_pipeline = [ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='ResizeCropFlipImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=['img', 'timestamp', 'projection_mat', 'image_wh'], - meta_keys=['T_global', 'T_global_inv', 'timestamp']) -] -input_modality = dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False) -data_basic_config = dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier', - 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval') -data_aug_conf = dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]) -data = dict( - samples_per_gpu=20, - workers_per_gpu=20, - train=dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval', - ann_file='data/nuscenes_anno_pkls/nuscenes_infos_train.pkl', - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict( - type='LoadPointsFromFile', - coord_type='LIDAR', - load_dim=5, - use_dim=5, - file_client_args=dict(backend='disk')), - dict(type='ResizeCropFlipImage'), - dict(type='MultiScaleDepthMapGenerator', downsample=[4, 8, 16]), - dict(type='BBoxRotation'), - dict(type='PhotoMetricDistortionMultiViewImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict( - type='CircleObjectRangeFilter', - class_dist_thred=[55, 55, 55, 55, 55, 55, 55, 55, 55, 55]), - dict( - type='InstanceNameFilter', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', - 'traffic_cone' - ]), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=[ - 'img', 'timestamp', 'projection_mat', 'image_wh', - 'gt_depth', 'focal', 'gt_bboxes_3d', 'gt_labels_3d' - ], - meta_keys=[ - 'T_global', 'T_global_inv', 'timestamp', 'instance_id' - ]) - ], - test_mode=False, - data_aug_conf=dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]), - with_seq_flag=True, - sequences_split_num=2, - keep_consistent_seq_aug=True), - val=dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval', - ann_file='data/nuscenes_anno_pkls/nuscenes_infos_val.pkl', - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='ResizeCropFlipImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=['img', 'timestamp', 'projection_mat', 'image_wh'], - meta_keys=['T_global', 'T_global_inv', 'timestamp']) - ], - data_aug_conf=dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]), - test_mode=True, - tracking=True, - tracking_threshold=0.2), - test=dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval', - ann_file='data/nuscenes_anno_pkls/nuscenes_infos_val.pkl', - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='ResizeCropFlipImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=['img', 'timestamp', 'projection_mat', 'image_wh'], - meta_keys=['T_global', 'T_global_inv', 'timestamp']) - ], - data_aug_conf=dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]), - test_mode=True, - tracking=True, - tracking_threshold=0.2)) -optimizer = dict( - type='Miopen_AdamW', - lr=0.0006, - weight_decay=0.001, - paramwise_cfg=dict(custom_keys=dict(img_backbone=dict(lr_mult=0.5)))) -optimizer_config = dict(grad_clip=dict(max_norm=25, norm_type=2)) -lr_config = dict( - policy='CosineAnnealing', - warmup='linear', - warmup_iters=500, - warmup_ratio=0.3333333333333333, - min_lr_ratio=0.001) -runner = dict(type='IterBasedRunner', max_iters=17500) -vis_pipeline = [ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='Collect', keys=['img'], meta_keys=['timestamp', 'lidar2img']) -] -evaluation = dict( - interval=3500, - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict( - type='Collect', keys=['img'], meta_keys=['timestamp', 'lidar2img']) - ]) -gpu_ids = range(0, 8) - -2026-01-14 20:53:54,061 - mmdet - INFO - Set random seed to 0, deterministic: False diff --git a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20260114_205716.log b/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20260114_205716.log deleted file mode 100644 index 19e770359ca90fc36a32e64f755ba152d9867417..0000000000000000000000000000000000000000 --- a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20260114_205716.log +++ /dev/null @@ -1,465 +0,0 @@ -2026-01-14 20:57:17,008 - mmdet - INFO - Environment info: ------------------------------------------------------------- -sys.platform: linux -Python: 3.10.12 (main, Aug 15 2025, 14:32:43) [GCC 11.4.0] -CUDA available: True -GPU 0,1,2,3,4,5,6,7: BW151 -CUDA_HOME: /opt/dtk -NVCC: Not Available -GCC: x86_64-linux-gnu-gcc (Ubuntu 11.4.0-1ubuntu1~22.04) 11.4.0 -PyTorch: 2.5.1 -PyTorch compiling details: PyTorch built with: - - GCC 10.3 - - C++ Version: 201703 - - Intel(R) Math Kernel Library Version 2020.0.4 Product Build 20200917 for Intel(R) 64 architecture applications - - OpenMP 201511 (a.k.a. OpenMP 4.5) - - LAPACK is enabled (usually provided by MKL) - - NNPACK is enabled - - CPU capability usage: AVX512 - - HIP Runtime 6.3.25521 - - MIOpen 2.18.0 - - Magma 2.8.0 - - Build settings: BLAS_INFO=mkl, BUILD_TYPE=Release, CXX_COMPILER=/opt/rh/gcc-toolset-10/root/usr/bin/c++, CXX_FLAGS= -D_GLIBCXX_USE_CXX11_ABI=1 -fvisibility-inlines-hidden -DUSE_PTHREADPOOL -DNDEBUG -DUSE_KINETO -DLIBKINETO_NOCUPTI -DLIBKINETO_NOXPUPTI=ON -DUSE_FBGEMM -DUSE_PYTORCH_QNNPACK -DUSE_XNNPACK -DSYMBOLICATE_MOBILE_DEBUG_HANDLE -O2 -fPIC -Wall -Wextra -Werror=return-type -Werror=non-virtual-dtor -Werror=bool-operation -Wnarrowing -Wno-missing-field-initializers -Wno-type-limits -Wno-array-bounds -Wno-unknown-pragmas -Wno-unused-parameter -Wno-strict-overflow -Wno-strict-aliasing -Wno-stringop-overflow -Wsuggest-override -Wno-psabi -Wno-error=old-style-cast -Wno-missing-braces -fdiagnostics-color=always -faligned-new -Wno-unused-but-set-variable -Wno-maybe-uninitialized -fno-math-errno -fno-trapping-math -Werror=format -Wno-stringop-overflow, FORCE_FALLBACK_CUDA_MPI=1, LAPACK_INFO=mkl, PERF_WITH_AVX=1, PERF_WITH_AVX2=1, TORCH_VERSION=2.5.1, USE_CUDA=0, USE_CUDNN=OFF, USE_CUSPARSELT=OFF, USE_EXCEPTION_PTR=1, USE_GFLAGS=1, USE_GLOG=1, USE_GLOO=1, USE_MKL=ON, USE_MKLDNN=0, USE_MPI=1, USE_NCCL=1, USE_NNPACK=ON, USE_OPENMP=1, USE_ROCM=ON, USE_ROCM_KERNEL_ASSERT=OFF, - -TorchVision: 0.20.1 -OpenCV: 4.12.0 -MMCV: 1.6.1 -MMCV Compiler: GCC 10.3 -MMCV CUDA Compiler: rocm not available -MMDetection: 2.25.1+c41df4b ------------------------------------------------------------- - -2026-01-14 20:57:17,768 - mmdet - INFO - Distributed training: True -2026-01-14 20:57:18,696 - mmdet - INFO - Config: -plugin = True -plugin_dir = 'projects/mmdet3d_plugin/' -dist_params = dict(backend='nccl') -log_level = 'INFO' -work_dir = './work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704' -total_batch_size = 160 -num_gpus = 8 -batch_size = 20 -num_iters_per_epoch = 175 -num_epochs = 100 -checkpoint_epoch_interval = 20 -checkpoint_config = dict(interval=3500) -log_config = dict( - interval=1, - hooks=[ - dict(type='TextLoggerHook', by_epoch=False), - dict(type='TensorboardLoggerHook') - ]) -load_from = None -resume_from = None -workflow = [('train', 1)] -fp16 = dict(loss_scale=32.0) -input_shape = (704, 256) -tracking_test = True -tracking_threshold = 0.2 -class_names = [ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier', - 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' -] -num_classes = 10 -embed_dims = 256 -num_groups = 8 -num_decoder = 6 -num_single_frame_decoder = 1 -use_deformable_func = True -strides = [4, 8, 16, 32] -num_levels = 4 -num_depth_layers = 3 -drop_out = 0.1 -temporal = True -decouple_attn = True -with_quality_estimation = True -model = dict( - type='Sparse4D', - use_grid_mask=True, - use_deformable_func=True, - img_backbone=dict( - type='ResNet', - depth=50, - num_stages=4, - frozen_stages=-1, - norm_eval=False, - style='pytorch', - with_cp=True, - out_indices=(0, 1, 2, 3), - norm_cfg=dict(type='BN', requires_grad=True), - pretrained='ckpt/resnet50-19c8e357.pth'), - img_neck=dict( - type='FPN', - num_outs=4, - start_level=0, - out_channels=256, - add_extra_convs='on_output', - relu_before_extra_convs=True, - in_channels=[256, 512, 1024, 2048]), - depth_branch=dict( - type='DenseDepthNet', - embed_dims=256, - num_depth_layers=3, - loss_weight=0.2), - head=dict( - type='Sparse4DHead', - cls_threshold_to_reg=0.05, - decouple_attn=True, - instance_bank=dict( - type='InstanceBank', - num_anchor=900, - embed_dims=256, - anchor='nuscenes_kmeans900.npy', - anchor_handler=dict(type='SparseBox3DKeyPointsGenerator'), - num_temp_instances=600, - confidence_decay=0.6, - feat_grad=False), - anchor_encoder=dict( - type='SparseBox3DEncoder', - vel_dims=3, - embed_dims=[128, 32, 32, 64], - mode='cat', - output_fc=False, - in_loops=1, - out_loops=4), - num_single_frame_decoder=1, - operation_order=[ - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine' - ], - temp_graph_model=dict( - type='MultiheadAttention', - embed_dims=512, - num_heads=8, - batch_first=True, - dropout=0.1), - graph_model=dict( - type='MultiheadAttention', - embed_dims=512, - num_heads=8, - batch_first=True, - dropout=0.1), - norm_layer=dict(type='LN', normalized_shape=256), - ffn=dict( - type='AsymmetricFFN', - in_channels=512, - pre_norm=dict(type='LN'), - embed_dims=256, - feedforward_channels=1024, - num_fcs=2, - ffn_drop=0.1, - act_cfg=dict(type='ReLU', inplace=True)), - deformable_model=dict( - type='DeformableFeatureAggregation', - embed_dims=256, - num_groups=8, - num_levels=4, - num_cams=6, - attn_drop=0.15, - use_deformable_func=True, - use_camera_embed=True, - residual_mode='cat', - kps_generator=dict( - type='SparseBox3DKeyPointsGenerator', - num_learnable_pts=6, - fix_scale=[[0, 0, 0], [0.45, 0, 0], [-0.45, 0, 0], - [0, 0.45, 0], [0, -0.45, 0], [0, 0, 0.45], - [0, 0, -0.45]])), - refine_layer=dict( - type='SparseBox3DRefinementModule', - embed_dims=256, - num_cls=10, - refine_yaw=True, - with_quality_estimation=True), - sampler=dict( - type='SparseBox3DTarget', - num_dn_groups=5, - num_temp_dn_groups=3, - dn_noise_scale=[2.0, 2.0, 2.0, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5], - max_dn_gt=32, - add_neg_dn=True, - cls_weight=2.0, - box_weight=0.25, - reg_weights=[2.0, 2.0, 2.0, 0.5, 0.5, 0.5, 0.0, 0.0, 0.0, 0.0], - cls_wise_reg_weights=dict( - {9: [2.0, 2.0, 2.0, 1.0, 1.0, 1.0, 0.0, 0.0, 1.0, 1.0]})), - loss_cls=dict( - type='FocalLoss', - use_sigmoid=True, - gamma=2.0, - alpha=0.25, - loss_weight=2.0), - loss_reg=dict( - type='SparseBox3DLoss', - loss_box=dict(type='L1Loss', loss_weight=0.25), - loss_centerness=dict(type='CrossEntropyLoss', use_sigmoid=True), - loss_yawness=dict(type='GaussianFocalLoss'), - cls_allow_reverse=[5]), - decoder=dict(type='SparseBox3DDecoder'), - reg_weights=[2.0, 2.0, 2.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0])) -dataset_type = 'NuScenes3DDetTrackDataset' -data_root = 'data/nuscenes/' -anno_root = 'data/nuscenes_anno_pkls/' -file_client_args = dict(backend='disk') -img_norm_cfg = dict( - mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375], to_rgb=True) -train_pipeline = [ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict( - type='LoadPointsFromFile', - coord_type='LIDAR', - load_dim=5, - use_dim=5, - file_client_args=dict(backend='disk')), - dict(type='ResizeCropFlipImage'), - dict(type='MultiScaleDepthMapGenerator', downsample=[4, 8, 16]), - dict(type='BBoxRotation'), - dict(type='PhotoMetricDistortionMultiViewImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict( - type='CircleObjectRangeFilter', - class_dist_thred=[55, 55, 55, 55, 55, 55, 55, 55, 55, 55]), - dict( - type='InstanceNameFilter', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ]), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=[ - 'img', 'timestamp', 'projection_mat', 'image_wh', 'gt_depth', - 'focal', 'gt_bboxes_3d', 'gt_labels_3d' - ], - meta_keys=['T_global', 'T_global_inv', 'timestamp', 'instance_id']) -] -test_pipeline = [ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='ResizeCropFlipImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=['img', 'timestamp', 'projection_mat', 'image_wh'], - meta_keys=['T_global', 'T_global_inv', 'timestamp']) -] -input_modality = dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False) -data_basic_config = dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier', - 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval') -data_aug_conf = dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]) -data = dict( - samples_per_gpu=20, - workers_per_gpu=20, - train=dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval', - ann_file='data/nuscenes_anno_pkls/nuscenes_infos_train.pkl', - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict( - type='LoadPointsFromFile', - coord_type='LIDAR', - load_dim=5, - use_dim=5, - file_client_args=dict(backend='disk')), - dict(type='ResizeCropFlipImage'), - dict(type='MultiScaleDepthMapGenerator', downsample=[4, 8, 16]), - dict(type='BBoxRotation'), - dict(type='PhotoMetricDistortionMultiViewImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict( - type='CircleObjectRangeFilter', - class_dist_thred=[55, 55, 55, 55, 55, 55, 55, 55, 55, 55]), - dict( - type='InstanceNameFilter', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', - 'traffic_cone' - ]), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=[ - 'img', 'timestamp', 'projection_mat', 'image_wh', - 'gt_depth', 'focal', 'gt_bboxes_3d', 'gt_labels_3d' - ], - meta_keys=[ - 'T_global', 'T_global_inv', 'timestamp', 'instance_id' - ]) - ], - test_mode=False, - data_aug_conf=dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]), - with_seq_flag=True, - sequences_split_num=2, - keep_consistent_seq_aug=True), - val=dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval', - ann_file='data/nuscenes_anno_pkls/nuscenes_infos_val.pkl', - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='ResizeCropFlipImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=['img', 'timestamp', 'projection_mat', 'image_wh'], - meta_keys=['T_global', 'T_global_inv', 'timestamp']) - ], - data_aug_conf=dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]), - test_mode=True, - tracking=True, - tracking_threshold=0.2), - test=dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval', - ann_file='data/nuscenes_anno_pkls/nuscenes_infos_val.pkl', - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='ResizeCropFlipImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=['img', 'timestamp', 'projection_mat', 'image_wh'], - meta_keys=['T_global', 'T_global_inv', 'timestamp']) - ], - data_aug_conf=dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]), - test_mode=True, - tracking=True, - tracking_threshold=0.2)) -optimizer = dict( - type='Miopen_AdamW', - lr=0.0006, - weight_decay=0.001, - paramwise_cfg=dict(custom_keys=dict(img_backbone=dict(lr_mult=0.5)))) -optimizer_config = dict(grad_clip=dict(max_norm=25, norm_type=2)) -lr_config = dict( - policy='CosineAnnealing', - warmup='linear', - warmup_iters=500, - warmup_ratio=0.3333333333333333, - min_lr_ratio=0.001) -runner = dict(type='IterBasedRunner', max_iters=17500) -vis_pipeline = [ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='Collect', keys=['img'], meta_keys=['timestamp', 'lidar2img']) -] -evaluation = dict( - interval=3500, - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict( - type='Collect', keys=['img'], meta_keys=['timestamp', 'lidar2img']) - ]) -gpu_ids = range(0, 8) - -2026-01-14 20:57:18,696 - mmdet - INFO - Set random seed to 0, deterministic: False diff --git a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20260114_205952.log b/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20260114_205952.log deleted file mode 100644 index cb99ae0fe0f2269b6b428e17ff1c3f9c351a8655..0000000000000000000000000000000000000000 --- a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20260114_205952.log +++ /dev/null @@ -1,465 +0,0 @@ -2026-01-14 20:59:52,122 - mmdet - INFO - Environment info: ------------------------------------------------------------- -sys.platform: linux -Python: 3.10.12 (main, Aug 15 2025, 14:32:43) [GCC 11.4.0] -CUDA available: True -GPU 0,1,2,3,4,5,6,7: BW151 -CUDA_HOME: /opt/dtk -NVCC: Not Available -GCC: x86_64-linux-gnu-gcc (Ubuntu 11.4.0-1ubuntu1~22.04) 11.4.0 -PyTorch: 2.5.1 -PyTorch compiling details: PyTorch built with: - - GCC 10.3 - - C++ Version: 201703 - - Intel(R) Math Kernel Library Version 2020.0.4 Product Build 20200917 for Intel(R) 64 architecture applications - - OpenMP 201511 (a.k.a. OpenMP 4.5) - - LAPACK is enabled (usually provided by MKL) - - NNPACK is enabled - - CPU capability usage: AVX512 - - HIP Runtime 6.3.25521 - - MIOpen 2.18.0 - - Magma 2.8.0 - - Build settings: BLAS_INFO=mkl, BUILD_TYPE=Release, CXX_COMPILER=/opt/rh/gcc-toolset-10/root/usr/bin/c++, CXX_FLAGS= -D_GLIBCXX_USE_CXX11_ABI=1 -fvisibility-inlines-hidden -DUSE_PTHREADPOOL -DNDEBUG -DUSE_KINETO -DLIBKINETO_NOCUPTI -DLIBKINETO_NOXPUPTI=ON -DUSE_FBGEMM -DUSE_PYTORCH_QNNPACK -DUSE_XNNPACK -DSYMBOLICATE_MOBILE_DEBUG_HANDLE -O2 -fPIC -Wall -Wextra -Werror=return-type -Werror=non-virtual-dtor -Werror=bool-operation -Wnarrowing -Wno-missing-field-initializers -Wno-type-limits -Wno-array-bounds -Wno-unknown-pragmas -Wno-unused-parameter -Wno-strict-overflow -Wno-strict-aliasing -Wno-stringop-overflow -Wsuggest-override -Wno-psabi -Wno-error=old-style-cast -Wno-missing-braces -fdiagnostics-color=always -faligned-new -Wno-unused-but-set-variable -Wno-maybe-uninitialized -fno-math-errno -fno-trapping-math -Werror=format -Wno-stringop-overflow, FORCE_FALLBACK_CUDA_MPI=1, LAPACK_INFO=mkl, PERF_WITH_AVX=1, PERF_WITH_AVX2=1, TORCH_VERSION=2.5.1, USE_CUDA=0, USE_CUDNN=OFF, USE_CUSPARSELT=OFF, USE_EXCEPTION_PTR=1, USE_GFLAGS=1, USE_GLOG=1, USE_GLOO=1, USE_MKL=ON, USE_MKLDNN=0, USE_MPI=1, USE_NCCL=1, USE_NNPACK=ON, USE_OPENMP=1, USE_ROCM=ON, USE_ROCM_KERNEL_ASSERT=OFF, - -TorchVision: 0.20.1 -OpenCV: 4.12.0 -MMCV: 1.6.1 -MMCV Compiler: GCC 10.3 -MMCV CUDA Compiler: rocm not available -MMDetection: 2.25.1+c41df4b ------------------------------------------------------------- - -2026-01-14 20:59:52,868 - mmdet - INFO - Distributed training: True -2026-01-14 20:59:53,799 - mmdet - INFO - Config: -plugin = True -plugin_dir = 'projects/mmdet3d_plugin/' -dist_params = dict(backend='nccl') -log_level = 'INFO' -work_dir = './work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704' -total_batch_size = 160 -num_gpus = 8 -batch_size = 20 -num_iters_per_epoch = 175 -num_epochs = 100 -checkpoint_epoch_interval = 20 -checkpoint_config = dict(interval=3500) -log_config = dict( - interval=1, - hooks=[ - dict(type='TextLoggerHook', by_epoch=False), - dict(type='TensorboardLoggerHook') - ]) -load_from = None -resume_from = None -workflow = [('train', 1)] -fp16 = dict(loss_scale=32.0) -input_shape = (704, 256) -tracking_test = True -tracking_threshold = 0.2 -class_names = [ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier', - 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' -] -num_classes = 10 -embed_dims = 256 -num_groups = 8 -num_decoder = 6 -num_single_frame_decoder = 1 -use_deformable_func = True -strides = [4, 8, 16, 32] -num_levels = 4 -num_depth_layers = 3 -drop_out = 0.1 -temporal = True -decouple_attn = True -with_quality_estimation = True -model = dict( - type='Sparse4D', - use_grid_mask=True, - use_deformable_func=True, - img_backbone=dict( - type='ResNet', - depth=50, - num_stages=4, - frozen_stages=-1, - norm_eval=False, - style='pytorch', - with_cp=True, - out_indices=(0, 1, 2, 3), - norm_cfg=dict(type='BN', requires_grad=True), - pretrained='ckpt/resnet50-19c8e357.pth'), - img_neck=dict( - type='FPN', - num_outs=4, - start_level=0, - out_channels=256, - add_extra_convs='on_output', - relu_before_extra_convs=True, - in_channels=[256, 512, 1024, 2048]), - depth_branch=dict( - type='DenseDepthNet', - embed_dims=256, - num_depth_layers=3, - loss_weight=0.2), - head=dict( - type='Sparse4DHead', - cls_threshold_to_reg=0.05, - decouple_attn=True, - instance_bank=dict( - type='InstanceBank', - num_anchor=900, - embed_dims=256, - anchor='nuscenes_kmeans900.npy', - anchor_handler=dict(type='SparseBox3DKeyPointsGenerator'), - num_temp_instances=600, - confidence_decay=0.6, - feat_grad=False), - anchor_encoder=dict( - type='SparseBox3DEncoder', - vel_dims=3, - embed_dims=[128, 32, 32, 64], - mode='cat', - output_fc=False, - in_loops=1, - out_loops=4), - num_single_frame_decoder=1, - operation_order=[ - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine' - ], - temp_graph_model=dict( - type='MultiheadAttention', - embed_dims=512, - num_heads=8, - batch_first=True, - dropout=0.1), - graph_model=dict( - type='MultiheadAttention', - embed_dims=512, - num_heads=8, - batch_first=True, - dropout=0.1), - norm_layer=dict(type='LN', normalized_shape=256), - ffn=dict( - type='AsymmetricFFN', - in_channels=512, - pre_norm=dict(type='LN'), - embed_dims=256, - feedforward_channels=1024, - num_fcs=2, - ffn_drop=0.1, - act_cfg=dict(type='ReLU', inplace=True)), - deformable_model=dict( - type='DeformableFeatureAggregation', - embed_dims=256, - num_groups=8, - num_levels=4, - num_cams=6, - attn_drop=0.15, - use_deformable_func=True, - use_camera_embed=True, - residual_mode='cat', - kps_generator=dict( - type='SparseBox3DKeyPointsGenerator', - num_learnable_pts=6, - fix_scale=[[0, 0, 0], [0.45, 0, 0], [-0.45, 0, 0], - [0, 0.45, 0], [0, -0.45, 0], [0, 0, 0.45], - [0, 0, -0.45]])), - refine_layer=dict( - type='SparseBox3DRefinementModule', - embed_dims=256, - num_cls=10, - refine_yaw=True, - with_quality_estimation=True), - sampler=dict( - type='SparseBox3DTarget', - num_dn_groups=5, - num_temp_dn_groups=3, - dn_noise_scale=[2.0, 2.0, 2.0, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5], - max_dn_gt=32, - add_neg_dn=True, - cls_weight=2.0, - box_weight=0.25, - reg_weights=[2.0, 2.0, 2.0, 0.5, 0.5, 0.5, 0.0, 0.0, 0.0, 0.0], - cls_wise_reg_weights=dict( - {9: [2.0, 2.0, 2.0, 1.0, 1.0, 1.0, 0.0, 0.0, 1.0, 1.0]})), - loss_cls=dict( - type='FocalLoss', - use_sigmoid=True, - gamma=2.0, - alpha=0.25, - loss_weight=2.0), - loss_reg=dict( - type='SparseBox3DLoss', - loss_box=dict(type='L1Loss', loss_weight=0.25), - loss_centerness=dict(type='CrossEntropyLoss', use_sigmoid=True), - loss_yawness=dict(type='GaussianFocalLoss'), - cls_allow_reverse=[5]), - decoder=dict(type='SparseBox3DDecoder'), - reg_weights=[2.0, 2.0, 2.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0])) -dataset_type = 'NuScenes3DDetTrackDataset' -data_root = 'data/nuscenes/' -anno_root = 'data/nuscenes_anno_pkls/' -file_client_args = dict(backend='disk') -img_norm_cfg = dict( - mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375], to_rgb=True) -train_pipeline = [ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict( - type='LoadPointsFromFile', - coord_type='LIDAR', - load_dim=5, - use_dim=5, - file_client_args=dict(backend='disk')), - dict(type='ResizeCropFlipImage'), - dict(type='MultiScaleDepthMapGenerator', downsample=[4, 8, 16]), - dict(type='BBoxRotation'), - dict(type='PhotoMetricDistortionMultiViewImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict( - type='CircleObjectRangeFilter', - class_dist_thred=[55, 55, 55, 55, 55, 55, 55, 55, 55, 55]), - dict( - type='InstanceNameFilter', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ]), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=[ - 'img', 'timestamp', 'projection_mat', 'image_wh', 'gt_depth', - 'focal', 'gt_bboxes_3d', 'gt_labels_3d' - ], - meta_keys=['T_global', 'T_global_inv', 'timestamp', 'instance_id']) -] -test_pipeline = [ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='ResizeCropFlipImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=['img', 'timestamp', 'projection_mat', 'image_wh'], - meta_keys=['T_global', 'T_global_inv', 'timestamp']) -] -input_modality = dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False) -data_basic_config = dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier', - 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval') -data_aug_conf = dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]) -data = dict( - samples_per_gpu=20, - workers_per_gpu=20, - train=dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval', - ann_file='data/nuscenes_anno_pkls/nuscenes_infos_train.pkl', - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict( - type='LoadPointsFromFile', - coord_type='LIDAR', - load_dim=5, - use_dim=5, - file_client_args=dict(backend='disk')), - dict(type='ResizeCropFlipImage'), - dict(type='MultiScaleDepthMapGenerator', downsample=[4, 8, 16]), - dict(type='BBoxRotation'), - dict(type='PhotoMetricDistortionMultiViewImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict( - type='CircleObjectRangeFilter', - class_dist_thred=[55, 55, 55, 55, 55, 55, 55, 55, 55, 55]), - dict( - type='InstanceNameFilter', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', - 'traffic_cone' - ]), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=[ - 'img', 'timestamp', 'projection_mat', 'image_wh', - 'gt_depth', 'focal', 'gt_bboxes_3d', 'gt_labels_3d' - ], - meta_keys=[ - 'T_global', 'T_global_inv', 'timestamp', 'instance_id' - ]) - ], - test_mode=False, - data_aug_conf=dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]), - with_seq_flag=True, - sequences_split_num=2, - keep_consistent_seq_aug=True), - val=dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval', - ann_file='data/nuscenes_anno_pkls/nuscenes_infos_val.pkl', - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='ResizeCropFlipImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=['img', 'timestamp', 'projection_mat', 'image_wh'], - meta_keys=['T_global', 'T_global_inv', 'timestamp']) - ], - data_aug_conf=dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]), - test_mode=True, - tracking=True, - tracking_threshold=0.2), - test=dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval', - ann_file='data/nuscenes_anno_pkls/nuscenes_infos_val.pkl', - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='ResizeCropFlipImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=['img', 'timestamp', 'projection_mat', 'image_wh'], - meta_keys=['T_global', 'T_global_inv', 'timestamp']) - ], - data_aug_conf=dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]), - test_mode=True, - tracking=True, - tracking_threshold=0.2)) -optimizer = dict( - type='Miopen_AdamW', - lr=0.0006, - weight_decay=0.001, - paramwise_cfg=dict(custom_keys=dict(img_backbone=dict(lr_mult=0.5)))) -optimizer_config = dict(grad_clip=dict(max_norm=25, norm_type=2)) -lr_config = dict( - policy='CosineAnnealing', - warmup='linear', - warmup_iters=500, - warmup_ratio=0.3333333333333333, - min_lr_ratio=0.001) -runner = dict(type='IterBasedRunner', max_iters=17500) -vis_pipeline = [ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='Collect', keys=['img'], meta_keys=['timestamp', 'lidar2img']) -] -evaluation = dict( - interval=3500, - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict( - type='Collect', keys=['img'], meta_keys=['timestamp', 'lidar2img']) - ]) -gpu_ids = range(0, 8) - -2026-01-14 20:59:53,799 - mmdet - INFO - Set random seed to 0, deterministic: False diff --git a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20260114_210630.log b/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20260114_210630.log deleted file mode 100644 index 6bd489f31c14d9c4f0086c079fb1175fc9663183..0000000000000000000000000000000000000000 --- a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20260114_210630.log +++ /dev/null @@ -1,465 +0,0 @@ -2026-01-14 21:06:30,590 - mmdet - INFO - Environment info: ------------------------------------------------------------- -sys.platform: linux -Python: 3.10.12 (main, Aug 15 2025, 14:32:43) [GCC 11.4.0] -CUDA available: True -GPU 0,1,2,3,4,5,6,7: BW151 -CUDA_HOME: /opt/dtk -NVCC: Not Available -GCC: x86_64-linux-gnu-gcc (Ubuntu 11.4.0-1ubuntu1~22.04) 11.4.0 -PyTorch: 2.5.1 -PyTorch compiling details: PyTorch built with: - - GCC 10.3 - - C++ Version: 201703 - - Intel(R) Math Kernel Library Version 2020.0.4 Product Build 20200917 for Intel(R) 64 architecture applications - - OpenMP 201511 (a.k.a. OpenMP 4.5) - - LAPACK is enabled (usually provided by MKL) - - NNPACK is enabled - - CPU capability usage: AVX512 - - HIP Runtime 6.3.25521 - - MIOpen 2.18.0 - - Magma 2.8.0 - - Build settings: BLAS_INFO=mkl, BUILD_TYPE=Release, CXX_COMPILER=/opt/rh/gcc-toolset-10/root/usr/bin/c++, CXX_FLAGS= -D_GLIBCXX_USE_CXX11_ABI=1 -fvisibility-inlines-hidden -DUSE_PTHREADPOOL -DNDEBUG -DUSE_KINETO -DLIBKINETO_NOCUPTI -DLIBKINETO_NOXPUPTI=ON -DUSE_FBGEMM -DUSE_PYTORCH_QNNPACK -DUSE_XNNPACK -DSYMBOLICATE_MOBILE_DEBUG_HANDLE -O2 -fPIC -Wall -Wextra -Werror=return-type -Werror=non-virtual-dtor -Werror=bool-operation -Wnarrowing -Wno-missing-field-initializers -Wno-type-limits -Wno-array-bounds -Wno-unknown-pragmas -Wno-unused-parameter -Wno-strict-overflow -Wno-strict-aliasing -Wno-stringop-overflow -Wsuggest-override -Wno-psabi -Wno-error=old-style-cast -Wno-missing-braces -fdiagnostics-color=always -faligned-new -Wno-unused-but-set-variable -Wno-maybe-uninitialized -fno-math-errno -fno-trapping-math -Werror=format -Wno-stringop-overflow, FORCE_FALLBACK_CUDA_MPI=1, LAPACK_INFO=mkl, PERF_WITH_AVX=1, PERF_WITH_AVX2=1, TORCH_VERSION=2.5.1, USE_CUDA=0, USE_CUDNN=OFF, USE_CUSPARSELT=OFF, USE_EXCEPTION_PTR=1, USE_GFLAGS=1, USE_GLOG=1, USE_GLOO=1, USE_MKL=ON, USE_MKLDNN=0, USE_MPI=1, USE_NCCL=1, USE_NNPACK=ON, USE_OPENMP=1, USE_ROCM=ON, USE_ROCM_KERNEL_ASSERT=OFF, - -TorchVision: 0.20.1 -OpenCV: 4.12.0 -MMCV: 1.6.1 -MMCV Compiler: GCC 10.3 -MMCV CUDA Compiler: rocm not available -MMDetection: 2.25.1+c41df4b ------------------------------------------------------------- - -2026-01-14 21:06:31,403 - mmdet - INFO - Distributed training: True -2026-01-14 21:06:32,283 - mmdet - INFO - Config: -plugin = True -plugin_dir = 'projects/mmdet3d_plugin/' -dist_params = dict(backend='nccl') -log_level = 'INFO' -work_dir = './work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704' -total_batch_size = 160 -num_gpus = 8 -batch_size = 20 -num_iters_per_epoch = 175 -num_epochs = 100 -checkpoint_epoch_interval = 20 -checkpoint_config = dict(interval=3500) -log_config = dict( - interval=1, - hooks=[ - dict(type='TextLoggerHook', by_epoch=False), - dict(type='TensorboardLoggerHook') - ]) -load_from = None -resume_from = None -workflow = [('train', 1)] -fp16 = dict(loss_scale=32.0) -input_shape = (704, 256) -tracking_test = True -tracking_threshold = 0.2 -class_names = [ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier', - 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' -] -num_classes = 10 -embed_dims = 256 -num_groups = 8 -num_decoder = 6 -num_single_frame_decoder = 1 -use_deformable_func = True -strides = [4, 8, 16, 32] -num_levels = 4 -num_depth_layers = 3 -drop_out = 0.1 -temporal = True -decouple_attn = True -with_quality_estimation = True -model = dict( - type='Sparse4D', - use_grid_mask=True, - use_deformable_func=True, - img_backbone=dict( - type='ResNet', - depth=50, - num_stages=4, - frozen_stages=-1, - norm_eval=False, - style='pytorch', - with_cp=True, - out_indices=(0, 1, 2, 3), - norm_cfg=dict(type='BN', requires_grad=True), - pretrained='ckpt/resnet50-19c8e357.pth'), - img_neck=dict( - type='FPN', - num_outs=4, - start_level=0, - out_channels=256, - add_extra_convs='on_output', - relu_before_extra_convs=True, - in_channels=[256, 512, 1024, 2048]), - depth_branch=dict( - type='DenseDepthNet', - embed_dims=256, - num_depth_layers=3, - loss_weight=0.2), - head=dict( - type='Sparse4DHead', - cls_threshold_to_reg=0.05, - decouple_attn=True, - instance_bank=dict( - type='InstanceBank', - num_anchor=900, - embed_dims=256, - anchor='nuscenes_kmeans900.npy', - anchor_handler=dict(type='SparseBox3DKeyPointsGenerator'), - num_temp_instances=600, - confidence_decay=0.6, - feat_grad=False), - anchor_encoder=dict( - type='SparseBox3DEncoder', - vel_dims=3, - embed_dims=[128, 32, 32, 64], - mode='cat', - output_fc=False, - in_loops=1, - out_loops=4), - num_single_frame_decoder=1, - operation_order=[ - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine' - ], - temp_graph_model=dict( - type='MultiheadAttention', - embed_dims=512, - num_heads=8, - batch_first=True, - dropout=0.1), - graph_model=dict( - type='MultiheadAttention', - embed_dims=512, - num_heads=8, - batch_first=True, - dropout=0.1), - norm_layer=dict(type='LN', normalized_shape=256), - ffn=dict( - type='AsymmetricFFN', - in_channels=512, - pre_norm=dict(type='LN'), - embed_dims=256, - feedforward_channels=1024, - num_fcs=2, - ffn_drop=0.1, - act_cfg=dict(type='ReLU', inplace=True)), - deformable_model=dict( - type='DeformableFeatureAggregation', - embed_dims=256, - num_groups=8, - num_levels=4, - num_cams=6, - attn_drop=0.15, - use_deformable_func=True, - use_camera_embed=True, - residual_mode='cat', - kps_generator=dict( - type='SparseBox3DKeyPointsGenerator', - num_learnable_pts=6, - fix_scale=[[0, 0, 0], [0.45, 0, 0], [-0.45, 0, 0], - [0, 0.45, 0], [0, -0.45, 0], [0, 0, 0.45], - [0, 0, -0.45]])), - refine_layer=dict( - type='SparseBox3DRefinementModule', - embed_dims=256, - num_cls=10, - refine_yaw=True, - with_quality_estimation=True), - sampler=dict( - type='SparseBox3DTarget', - num_dn_groups=5, - num_temp_dn_groups=3, - dn_noise_scale=[2.0, 2.0, 2.0, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5], - max_dn_gt=32, - add_neg_dn=True, - cls_weight=2.0, - box_weight=0.25, - reg_weights=[2.0, 2.0, 2.0, 0.5, 0.5, 0.5, 0.0, 0.0, 0.0, 0.0], - cls_wise_reg_weights=dict( - {9: [2.0, 2.0, 2.0, 1.0, 1.0, 1.0, 0.0, 0.0, 1.0, 1.0]})), - loss_cls=dict( - type='FocalLoss', - use_sigmoid=True, - gamma=2.0, - alpha=0.25, - loss_weight=2.0), - loss_reg=dict( - type='SparseBox3DLoss', - loss_box=dict(type='L1Loss', loss_weight=0.25), - loss_centerness=dict(type='CrossEntropyLoss', use_sigmoid=True), - loss_yawness=dict(type='GaussianFocalLoss'), - cls_allow_reverse=[5]), - decoder=dict(type='SparseBox3DDecoder'), - reg_weights=[2.0, 2.0, 2.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0])) -dataset_type = 'NuScenes3DDetTrackDataset' -data_root = 'data/nuscenes/' -anno_root = 'data/nuscenes_anno_pkls/' -file_client_args = dict(backend='disk') -img_norm_cfg = dict( - mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375], to_rgb=True) -train_pipeline = [ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict( - type='LoadPointsFromFile', - coord_type='LIDAR', - load_dim=5, - use_dim=5, - file_client_args=dict(backend='disk')), - dict(type='ResizeCropFlipImage'), - dict(type='MultiScaleDepthMapGenerator', downsample=[4, 8, 16]), - dict(type='BBoxRotation'), - dict(type='PhotoMetricDistortionMultiViewImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict( - type='CircleObjectRangeFilter', - class_dist_thred=[55, 55, 55, 55, 55, 55, 55, 55, 55, 55]), - dict( - type='InstanceNameFilter', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ]), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=[ - 'img', 'timestamp', 'projection_mat', 'image_wh', 'gt_depth', - 'focal', 'gt_bboxes_3d', 'gt_labels_3d' - ], - meta_keys=['T_global', 'T_global_inv', 'timestamp', 'instance_id']) -] -test_pipeline = [ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='ResizeCropFlipImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=['img', 'timestamp', 'projection_mat', 'image_wh'], - meta_keys=['T_global', 'T_global_inv', 'timestamp']) -] -input_modality = dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False) -data_basic_config = dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier', - 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval') -data_aug_conf = dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]) -data = dict( - samples_per_gpu=20, - workers_per_gpu=20, - train=dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval', - ann_file='data/nuscenes_anno_pkls/nuscenes_infos_train.pkl', - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict( - type='LoadPointsFromFile', - coord_type='LIDAR', - load_dim=5, - use_dim=5, - file_client_args=dict(backend='disk')), - dict(type='ResizeCropFlipImage'), - dict(type='MultiScaleDepthMapGenerator', downsample=[4, 8, 16]), - dict(type='BBoxRotation'), - dict(type='PhotoMetricDistortionMultiViewImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict( - type='CircleObjectRangeFilter', - class_dist_thred=[55, 55, 55, 55, 55, 55, 55, 55, 55, 55]), - dict( - type='InstanceNameFilter', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', - 'traffic_cone' - ]), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=[ - 'img', 'timestamp', 'projection_mat', 'image_wh', - 'gt_depth', 'focal', 'gt_bboxes_3d', 'gt_labels_3d' - ], - meta_keys=[ - 'T_global', 'T_global_inv', 'timestamp', 'instance_id' - ]) - ], - test_mode=False, - data_aug_conf=dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]), - with_seq_flag=True, - sequences_split_num=2, - keep_consistent_seq_aug=True), - val=dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval', - ann_file='data/nuscenes_anno_pkls/nuscenes_infos_val.pkl', - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='ResizeCropFlipImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=['img', 'timestamp', 'projection_mat', 'image_wh'], - meta_keys=['T_global', 'T_global_inv', 'timestamp']) - ], - data_aug_conf=dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]), - test_mode=True, - tracking=True, - tracking_threshold=0.2), - test=dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval', - ann_file='data/nuscenes_anno_pkls/nuscenes_infos_val.pkl', - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='ResizeCropFlipImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=['img', 'timestamp', 'projection_mat', 'image_wh'], - meta_keys=['T_global', 'T_global_inv', 'timestamp']) - ], - data_aug_conf=dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]), - test_mode=True, - tracking=True, - tracking_threshold=0.2)) -optimizer = dict( - type='Miopen_AdamW', - lr=0.0006, - weight_decay=0.001, - paramwise_cfg=dict(custom_keys=dict(img_backbone=dict(lr_mult=0.5)))) -optimizer_config = dict(grad_clip=dict(max_norm=25, norm_type=2)) -lr_config = dict( - policy='CosineAnnealing', - warmup='linear', - warmup_iters=500, - warmup_ratio=0.3333333333333333, - min_lr_ratio=0.001) -runner = dict(type='IterBasedRunner', max_iters=17500) -vis_pipeline = [ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='Collect', keys=['img'], meta_keys=['timestamp', 'lidar2img']) -] -evaluation = dict( - interval=3500, - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict( - type='Collect', keys=['img'], meta_keys=['timestamp', 'lidar2img']) - ]) -gpu_ids = range(0, 8) - -2026-01-14 21:06:32,283 - mmdet - INFO - Set random seed to 0, deterministic: False diff --git a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20260114_211219.log b/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20260114_211219.log deleted file mode 100644 index b8f45bbc6f351cdfc50beb812f95061fb0431fe0..0000000000000000000000000000000000000000 --- a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20260114_211219.log +++ /dev/null @@ -1,3163 +0,0 @@ -2026-01-14 21:12:19,633 - mmdet - INFO - Environment info: ------------------------------------------------------------- -sys.platform: linux -Python: 3.10.12 (main, Aug 15 2025, 14:32:43) [GCC 11.4.0] -CUDA available: True -GPU 0,1,2,3,4,5,6,7: BW151 -CUDA_HOME: /opt/dtk -NVCC: Not Available -GCC: x86_64-linux-gnu-gcc (Ubuntu 11.4.0-1ubuntu1~22.04) 11.4.0 -PyTorch: 2.5.1 -PyTorch compiling details: PyTorch built with: - - GCC 10.3 - - C++ Version: 201703 - - Intel(R) Math Kernel Library Version 2020.0.4 Product Build 20200917 for Intel(R) 64 architecture applications - - OpenMP 201511 (a.k.a. OpenMP 4.5) - - LAPACK is enabled (usually provided by MKL) - - NNPACK is enabled - - CPU capability usage: AVX512 - - HIP Runtime 6.3.25521 - - MIOpen 2.18.0 - - Magma 2.8.0 - - Build settings: BLAS_INFO=mkl, BUILD_TYPE=Release, CXX_COMPILER=/opt/rh/gcc-toolset-10/root/usr/bin/c++, CXX_FLAGS= -D_GLIBCXX_USE_CXX11_ABI=1 -fvisibility-inlines-hidden -DUSE_PTHREADPOOL -DNDEBUG -DUSE_KINETO -DLIBKINETO_NOCUPTI -DLIBKINETO_NOXPUPTI=ON -DUSE_FBGEMM -DUSE_PYTORCH_QNNPACK -DUSE_XNNPACK -DSYMBOLICATE_MOBILE_DEBUG_HANDLE -O2 -fPIC -Wall -Wextra -Werror=return-type -Werror=non-virtual-dtor -Werror=bool-operation -Wnarrowing -Wno-missing-field-initializers -Wno-type-limits -Wno-array-bounds -Wno-unknown-pragmas -Wno-unused-parameter -Wno-strict-overflow -Wno-strict-aliasing -Wno-stringop-overflow -Wsuggest-override -Wno-psabi -Wno-error=old-style-cast -Wno-missing-braces -fdiagnostics-color=always -faligned-new -Wno-unused-but-set-variable -Wno-maybe-uninitialized -fno-math-errno -fno-trapping-math -Werror=format -Wno-stringop-overflow, FORCE_FALLBACK_CUDA_MPI=1, LAPACK_INFO=mkl, PERF_WITH_AVX=1, PERF_WITH_AVX2=1, TORCH_VERSION=2.5.1, USE_CUDA=0, USE_CUDNN=OFF, USE_CUSPARSELT=OFF, USE_EXCEPTION_PTR=1, USE_GFLAGS=1, USE_GLOG=1, USE_GLOO=1, USE_MKL=ON, USE_MKLDNN=0, USE_MPI=1, USE_NCCL=1, USE_NNPACK=ON, USE_OPENMP=1, USE_ROCM=ON, USE_ROCM_KERNEL_ASSERT=OFF, - -TorchVision: 0.20.1 -OpenCV: 4.12.0 -MMCV: 1.6.1 -MMCV Compiler: GCC 10.3 -MMCV CUDA Compiler: rocm not available -MMDetection: 2.25.1+c41df4b ------------------------------------------------------------- - -2026-01-14 21:12:20,382 - mmdet - INFO - Distributed training: True -2026-01-14 21:12:21,299 - mmdet - INFO - Config: -plugin = True -plugin_dir = 'projects/mmdet3d_plugin/' -dist_params = dict(backend='nccl') -log_level = 'INFO' -work_dir = './work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704' -total_batch_size = 160 -num_gpus = 8 -batch_size = 20 -num_iters_per_epoch = 175 -num_epochs = 100 -checkpoint_epoch_interval = 20 -checkpoint_config = dict(interval=3500) -log_config = dict( - interval=1, - hooks=[ - dict(type='TextLoggerHook', by_epoch=False), - dict(type='TensorboardLoggerHook') - ]) -load_from = None -resume_from = None -workflow = [('train', 1)] -fp16 = dict(loss_scale=32.0) -input_shape = (704, 256) -tracking_test = True -tracking_threshold = 0.2 -class_names = [ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier', - 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' -] -num_classes = 10 -embed_dims = 256 -num_groups = 8 -num_decoder = 6 -num_single_frame_decoder = 1 -use_deformable_func = True -strides = [4, 8, 16, 32] -num_levels = 4 -num_depth_layers = 3 -drop_out = 0.1 -temporal = True -decouple_attn = True -with_quality_estimation = True -model = dict( - type='Sparse4D', - use_grid_mask=True, - use_deformable_func=True, - img_backbone=dict( - type='ResNet', - depth=50, - num_stages=4, - frozen_stages=-1, - norm_eval=False, - style='pytorch', - with_cp=True, - out_indices=(0, 1, 2, 3), - norm_cfg=dict(type='BN', requires_grad=True), - pretrained='ckpt/resnet50-19c8e357.pth'), - img_neck=dict( - type='FPN', - num_outs=4, - start_level=0, - out_channels=256, - add_extra_convs='on_output', - relu_before_extra_convs=True, - in_channels=[256, 512, 1024, 2048]), - depth_branch=dict( - type='DenseDepthNet', - embed_dims=256, - num_depth_layers=3, - loss_weight=0.2), - head=dict( - type='Sparse4DHead', - cls_threshold_to_reg=0.05, - decouple_attn=True, - instance_bank=dict( - type='InstanceBank', - num_anchor=900, - embed_dims=256, - anchor='nuscenes_kmeans900.npy', - anchor_handler=dict(type='SparseBox3DKeyPointsGenerator'), - num_temp_instances=600, - confidence_decay=0.6, - feat_grad=False), - anchor_encoder=dict( - type='SparseBox3DEncoder', - vel_dims=3, - embed_dims=[128, 32, 32, 64], - mode='cat', - output_fc=False, - in_loops=1, - out_loops=4), - num_single_frame_decoder=1, - operation_order=[ - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine' - ], - temp_graph_model=dict( - type='MultiheadAttention', - embed_dims=512, - num_heads=8, - batch_first=True, - dropout=0.1), - graph_model=dict( - type='MultiheadAttention', - embed_dims=512, - num_heads=8, - batch_first=True, - dropout=0.1), - norm_layer=dict(type='LN', normalized_shape=256), - ffn=dict( - type='AsymmetricFFN', - in_channels=512, - pre_norm=dict(type='LN'), - embed_dims=256, - feedforward_channels=1024, - num_fcs=2, - ffn_drop=0.1, - act_cfg=dict(type='ReLU', inplace=True)), - deformable_model=dict( - type='DeformableFeatureAggregation', - embed_dims=256, - num_groups=8, - num_levels=4, - num_cams=6, - attn_drop=0.15, - use_deformable_func=True, - use_camera_embed=True, - residual_mode='cat', - kps_generator=dict( - type='SparseBox3DKeyPointsGenerator', - num_learnable_pts=6, - fix_scale=[[0, 0, 0], [0.45, 0, 0], [-0.45, 0, 0], - [0, 0.45, 0], [0, -0.45, 0], [0, 0, 0.45], - [0, 0, -0.45]])), - refine_layer=dict( - type='SparseBox3DRefinementModule', - embed_dims=256, - num_cls=10, - refine_yaw=True, - with_quality_estimation=True), - sampler=dict( - type='SparseBox3DTarget', - num_dn_groups=5, - num_temp_dn_groups=3, - dn_noise_scale=[2.0, 2.0, 2.0, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5], - max_dn_gt=32, - add_neg_dn=True, - cls_weight=2.0, - box_weight=0.25, - reg_weights=[2.0, 2.0, 2.0, 0.5, 0.5, 0.5, 0.0, 0.0, 0.0, 0.0], - cls_wise_reg_weights=dict( - {9: [2.0, 2.0, 2.0, 1.0, 1.0, 1.0, 0.0, 0.0, 1.0, 1.0]})), - loss_cls=dict( - type='FocalLoss', - use_sigmoid=True, - gamma=2.0, - alpha=0.25, - loss_weight=2.0), - loss_reg=dict( - type='SparseBox3DLoss', - loss_box=dict(type='L1Loss', loss_weight=0.25), - loss_centerness=dict(type='CrossEntropyLoss', use_sigmoid=True), - loss_yawness=dict(type='GaussianFocalLoss'), - cls_allow_reverse=[5]), - decoder=dict(type='SparseBox3DDecoder'), - reg_weights=[2.0, 2.0, 2.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0])) -dataset_type = 'NuScenes3DDetTrackDataset' -data_root = 'data/nuscenes/' -anno_root = 'data/nuscenes_anno_pkls/' -file_client_args = dict(backend='disk') -img_norm_cfg = dict( - mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375], to_rgb=True) -train_pipeline = [ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict( - type='LoadPointsFromFile', - coord_type='LIDAR', - load_dim=5, - use_dim=5, - file_client_args=dict(backend='disk')), - dict(type='ResizeCropFlipImage'), - dict(type='MultiScaleDepthMapGenerator', downsample=[4, 8, 16]), - dict(type='BBoxRotation'), - dict(type='PhotoMetricDistortionMultiViewImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict( - type='CircleObjectRangeFilter', - class_dist_thred=[55, 55, 55, 55, 55, 55, 55, 55, 55, 55]), - dict( - type='InstanceNameFilter', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ]), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=[ - 'img', 'timestamp', 'projection_mat', 'image_wh', 'gt_depth', - 'focal', 'gt_bboxes_3d', 'gt_labels_3d' - ], - meta_keys=['T_global', 'T_global_inv', 'timestamp', 'instance_id']) -] -test_pipeline = [ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='ResizeCropFlipImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=['img', 'timestamp', 'projection_mat', 'image_wh'], - meta_keys=['T_global', 'T_global_inv', 'timestamp']) -] -input_modality = dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False) -data_basic_config = dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier', - 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval') -data_aug_conf = dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]) -data = dict( - samples_per_gpu=20, - workers_per_gpu=20, - train=dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval', - ann_file='data/nuscenes_anno_pkls/nuscenes_infos_train.pkl', - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict( - type='LoadPointsFromFile', - coord_type='LIDAR', - load_dim=5, - use_dim=5, - file_client_args=dict(backend='disk')), - dict(type='ResizeCropFlipImage'), - dict(type='MultiScaleDepthMapGenerator', downsample=[4, 8, 16]), - dict(type='BBoxRotation'), - dict(type='PhotoMetricDistortionMultiViewImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict( - type='CircleObjectRangeFilter', - class_dist_thred=[55, 55, 55, 55, 55, 55, 55, 55, 55, 55]), - dict( - type='InstanceNameFilter', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', - 'traffic_cone' - ]), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=[ - 'img', 'timestamp', 'projection_mat', 'image_wh', - 'gt_depth', 'focal', 'gt_bboxes_3d', 'gt_labels_3d' - ], - meta_keys=[ - 'T_global', 'T_global_inv', 'timestamp', 'instance_id' - ]) - ], - test_mode=False, - data_aug_conf=dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]), - with_seq_flag=True, - sequences_split_num=2, - keep_consistent_seq_aug=True), - val=dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval', - ann_file='data/nuscenes_anno_pkls/nuscenes_infos_val.pkl', - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='ResizeCropFlipImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=['img', 'timestamp', 'projection_mat', 'image_wh'], - meta_keys=['T_global', 'T_global_inv', 'timestamp']) - ], - data_aug_conf=dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]), - test_mode=True, - tracking=True, - tracking_threshold=0.2), - test=dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval', - ann_file='data/nuscenes_anno_pkls/nuscenes_infos_val.pkl', - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='ResizeCropFlipImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=['img', 'timestamp', 'projection_mat', 'image_wh'], - meta_keys=['T_global', 'T_global_inv', 'timestamp']) - ], - data_aug_conf=dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]), - test_mode=True, - tracking=True, - tracking_threshold=0.2)) -optimizer = dict( - type='Miopen_AdamW', - lr=0.0006, - weight_decay=0.001, - paramwise_cfg=dict(custom_keys=dict(img_backbone=dict(lr_mult=0.5)))) -optimizer_config = dict(grad_clip=dict(max_norm=25, norm_type=2)) -lr_config = dict( - policy='CosineAnnealing', - warmup='linear', - warmup_iters=500, - warmup_ratio=0.3333333333333333, - min_lr_ratio=0.001) -runner = dict(type='IterBasedRunner', max_iters=17500) -vis_pipeline = [ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='Collect', keys=['img'], meta_keys=['timestamp', 'lidar2img']) -] -evaluation = dict( - interval=3500, - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict( - type='Collect', keys=['img'], meta_keys=['timestamp', 'lidar2img']) - ]) -gpu_ids = range(0, 8) - -2026-01-14 21:12:21,299 - mmdet - INFO - Set random seed to 0, deterministic: False -2026-01-14 21:12:21,634 - mmdet - INFO - initialize ResNet with init_cfg {'type': 'Pretrained', 'checkpoint': 'ckpt/resnet50-19c8e357.pth'} -2026-01-14 21:12:21,873 - mmdet - INFO - initialize FPN with init_cfg {'type': 'Xavier', 'layer': 'Conv2d', 'distribution': 'uniform'} -Name of parameter - Initialization information - -img_backbone.conv1.weight - torch.Size([64, 3, 7, 7]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.bn1.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.bn1.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.conv1.weight - torch.Size([64, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn1.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn1.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.conv2.weight - torch.Size([64, 64, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn2.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn2.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.conv3.weight - torch.Size([256, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn3.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn3.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.downsample.0.weight - torch.Size([256, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.downsample.1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.downsample.1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.conv1.weight - torch.Size([64, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn1.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn1.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.conv2.weight - torch.Size([64, 64, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn2.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn2.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.conv3.weight - torch.Size([256, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn3.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn3.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.conv1.weight - torch.Size([64, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn1.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn1.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.conv2.weight - torch.Size([64, 64, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn2.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn2.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.conv3.weight - torch.Size([256, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn3.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn3.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.conv1.weight - torch.Size([128, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn1.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn1.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.conv2.weight - torch.Size([128, 128, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn2.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn2.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.conv3.weight - torch.Size([512, 128, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn3.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn3.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.downsample.0.weight - torch.Size([512, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.downsample.1.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.downsample.1.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.conv1.weight - torch.Size([128, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn1.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn1.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.conv2.weight - torch.Size([128, 128, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn2.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn2.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.conv3.weight - torch.Size([512, 128, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn3.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn3.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.conv1.weight - torch.Size([128, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn1.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn1.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.conv2.weight - torch.Size([128, 128, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn2.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn2.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.conv3.weight - torch.Size([512, 128, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn3.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn3.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.conv1.weight - torch.Size([128, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn1.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn1.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.conv2.weight - torch.Size([128, 128, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn2.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn2.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.conv3.weight - torch.Size([512, 128, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn3.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn3.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.conv1.weight - torch.Size([256, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.downsample.0.weight - torch.Size([1024, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.downsample.1.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.downsample.1.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.conv1.weight - torch.Size([512, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn1.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn1.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.conv2.weight - torch.Size([512, 512, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn2.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn2.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.conv3.weight - torch.Size([2048, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn3.weight - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn3.bias - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.downsample.0.weight - torch.Size([2048, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.downsample.1.weight - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.downsample.1.bias - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.conv1.weight - torch.Size([512, 2048, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn1.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn1.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.conv2.weight - torch.Size([512, 512, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn2.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn2.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.conv3.weight - torch.Size([2048, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn3.weight - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn3.bias - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.conv1.weight - torch.Size([512, 2048, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn1.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn1.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.conv2.weight - torch.Size([512, 512, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn2.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn2.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.conv3.weight - torch.Size([2048, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn3.weight - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn3.bias - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_neck.lateral_convs.0.conv.weight - torch.Size([256, 256, 1, 1]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.lateral_convs.0.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.lateral_convs.1.conv.weight - torch.Size([256, 512, 1, 1]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.lateral_convs.1.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.lateral_convs.2.conv.weight - torch.Size([256, 1024, 1, 1]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.lateral_convs.2.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.lateral_convs.3.conv.weight - torch.Size([256, 2048, 1, 1]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.lateral_convs.3.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.fpn_convs.0.conv.weight - torch.Size([256, 256, 3, 3]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.fpn_convs.0.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.fpn_convs.1.conv.weight - torch.Size([256, 256, 3, 3]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.fpn_convs.1.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.fpn_convs.2.conv.weight - torch.Size([256, 256, 3, 3]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.fpn_convs.2.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.fpn_convs.3.conv.weight - torch.Size([256, 256, 3, 3]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.fpn_convs.3.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.instance_bank.anchor - torch.Size([900, 11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.instance_bank.instance_feature - torch.Size([900, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.instance_bank.anchor_handler.fix_scale - torch.Size([1, 3]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.0.weight - torch.Size([128, 3]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.0.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.2.weight - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.2.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.3.weight - torch.Size([128, 128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.3.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.5.weight - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.5.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.6.weight - torch.Size([128, 128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.6.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.8.weight - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.8.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.9.weight - torch.Size([128, 128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.9.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.11.weight - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.11.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.0.weight - torch.Size([32, 3]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.0.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.2.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.2.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.3.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.3.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.5.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.5.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.6.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.6.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.8.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.8.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.9.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.9.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.11.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.11.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.0.weight - torch.Size([32, 2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.0.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.2.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.2.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.3.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.3.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.5.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.5.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.6.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.6.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.8.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.8.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.9.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.9.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.11.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.11.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.0.weight - torch.Size([64, 3]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.0.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.2.weight - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.2.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.3.weight - torch.Size([64, 64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.3.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.5.weight - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.5.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.6.weight - torch.Size([64, 64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.6.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.8.weight - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.8.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.9.weight - torch.Size([64, 64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.9.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.11.weight - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.11.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.1.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.1.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.1.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.1.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.1.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.1.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.1.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.1.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.3.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.4.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.4.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.4.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.4.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.5.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.5.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.5.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.5.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.6.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.6.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.8.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.8.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.8.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.8.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.8.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.8.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.8.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.8.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.10.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.11.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.11.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.11.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.11.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.12.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.12.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.12.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.12.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.13.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.13.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.15.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.15.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.15.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.15.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.15.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.15.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.15.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.15.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.16.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.16.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.17.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.18.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.18.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.18.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.18.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.19.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.19.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.19.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.19.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.20.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.20.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.22.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.22.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.22.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.22.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.22.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.22.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.22.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.22.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.23.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.23.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.24.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.25.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.25.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.25.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.25.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.26.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.26.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.26.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.26.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.27.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.27.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.29.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.29.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.29.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.29.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.29.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.29.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.29.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.29.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.30.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.30.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.31.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.32.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.32.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.32.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.32.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.33.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.33.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.33.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.33.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.34.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.34.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.36.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.36.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.36.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.36.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.36.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.36.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.36.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.36.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.37.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.37.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.38.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.fc_before.weight - torch.Size([512, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.fc_after.weight - torch.Size([256, 512]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.0.weight - torch.Size([1, 256, 1, 1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.0.bias - torch.Size([1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.1.weight - torch.Size([1, 256, 1, 1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.1.bias - torch.Size([1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.2.weight - torch.Size([1, 256, 1, 1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.2.bias - torch.Size([1]): -The value is the same before and after calling `init_weights` of Sparse4D -2026-01-14 21:12:21,976 - mmdet - INFO - Model: -Sparse4D( - (img_backbone): ResNet( - (conv1): Conv2d(3, 64, kernel_size=(7, 7), stride=(2, 2), padding=(3, 3), bias=False) - (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (maxpool): MaxPool2d(kernel_size=3, stride=2, padding=1, dilation=1, ceil_mode=False) - (layer1): ResLayer( - (0): Bottleneck( - (conv1): Conv2d(64, 64, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (downsample): Sequential( - (0): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - ) - ) - (1): Bottleneck( - (conv1): Conv2d(256, 64, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (2): Bottleneck( - (conv1): Conv2d(256, 64, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - ) - (layer2): ResLayer( - (0): Bottleneck( - (conv1): Conv2d(256, 128, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (downsample): Sequential( - (0): Conv2d(256, 512, kernel_size=(1, 1), stride=(2, 2), bias=False) - (1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - ) - ) - (1): Bottleneck( - (conv1): Conv2d(512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (2): Bottleneck( - (conv1): Conv2d(512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (3): Bottleneck( - (conv1): Conv2d(512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - ) - (layer3): ResLayer( - (0): Bottleneck( - (conv1): Conv2d(512, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (downsample): Sequential( - (0): Conv2d(512, 1024, kernel_size=(1, 1), stride=(2, 2), bias=False) - (1): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - ) - ) - (1): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (2): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (3): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (4): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (5): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - ) - (layer4): ResLayer( - (0): Bottleneck( - (conv1): Conv2d(1024, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(512, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (downsample): Sequential( - (0): Conv2d(1024, 2048, kernel_size=(1, 1), stride=(2, 2), bias=False) - (1): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - ) - ) - (1): Bottleneck( - (conv1): Conv2d(2048, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(512, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (2): Bottleneck( - (conv1): Conv2d(2048, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(512, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - ) - ) - init_cfg={'type': 'Pretrained', 'checkpoint': 'ckpt/resnet50-19c8e357.pth'} - (img_neck): FPN( - (lateral_convs): ModuleList( - (0): ConvModule( - (conv): Conv2d(256, 256, kernel_size=(1, 1), stride=(1, 1)) - ) - (1): ConvModule( - (conv): Conv2d(512, 256, kernel_size=(1, 1), stride=(1, 1)) - ) - (2): ConvModule( - (conv): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1)) - ) - (3): ConvModule( - (conv): Conv2d(2048, 256, kernel_size=(1, 1), stride=(1, 1)) - ) - ) - (fpn_convs): ModuleList( - (0-3): 4 x ConvModule( - (conv): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1)) - ) - ) - ) - init_cfg={'type': 'Xavier', 'layer': 'Conv2d', 'distribution': 'uniform'} - (head): Sparse4DHead( - (instance_bank): InstanceBank( - (anchor_handler): SparseBox3DKeyPointsGenerator() - ) - (anchor_encoder): SparseBox3DEncoder( - (pos_fc): Sequential( - (0): Linear(in_features=3, out_features=128, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((128,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=128, out_features=128, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((128,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=128, out_features=128, bias=True) - (7): ReLU(inplace=True) - (8): LayerNorm((128,), eps=1e-05, elementwise_affine=True) - (9): Linear(in_features=128, out_features=128, bias=True) - (10): ReLU(inplace=True) - (11): LayerNorm((128,), eps=1e-05, elementwise_affine=True) - ) - (size_fc): Sequential( - (0): Linear(in_features=3, out_features=32, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=32, out_features=32, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=32, out_features=32, bias=True) - (7): ReLU(inplace=True) - (8): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (9): Linear(in_features=32, out_features=32, bias=True) - (10): ReLU(inplace=True) - (11): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - ) - (yaw_fc): Sequential( - (0): Linear(in_features=2, out_features=32, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=32, out_features=32, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=32, out_features=32, bias=True) - (7): ReLU(inplace=True) - (8): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (9): Linear(in_features=32, out_features=32, bias=True) - (10): ReLU(inplace=True) - (11): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - ) - (vel_fc): Sequential( - (0): Linear(in_features=3, out_features=64, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((64,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=64, out_features=64, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((64,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=64, out_features=64, bias=True) - (7): ReLU(inplace=True) - (8): LayerNorm((64,), eps=1e-05, elementwise_affine=True) - (9): Linear(in_features=64, out_features=64, bias=True) - (10): ReLU(inplace=True) - (11): LayerNorm((64,), eps=1e-05, elementwise_affine=True) - ) - ) - (loss_cls): FocalLoss() - (loss_reg): SparseBox3DLoss( - (loss_box): L1Loss() - (loss_cns): CrossEntropyLoss(avg_non_ignore=False) - (loss_yns): GaussianFocalLoss() - ) - (layers): ModuleList( - (0): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (1): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (4-5): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (6): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (7): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (8): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (11-12): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (13): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (14): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (15): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (16): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (17): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (18-19): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (20): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (21): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (22): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (23): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (24): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (25-26): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (27): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (28): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (29): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (30): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (31): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (32-33): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (34): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (35): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (36): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (37): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (38): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - ) - (fc_before): Linear(in_features=256, out_features=512, bias=False) - (fc_after): Linear(in_features=512, out_features=256, bias=False) - ) - (depth_branch): DenseDepthNet( - (depth_layers): ModuleList( - (0-2): 3 x Conv2d(256, 1, kernel_size=(1, 1), stride=(1, 1)) - ) - ) - (grid_mask): GridMask() -) diff --git a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20260114_212059.log b/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20260114_212059.log deleted file mode 100644 index 1231818062464248224c30fb560be7ecfd983bef..0000000000000000000000000000000000000000 --- a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20260114_212059.log +++ /dev/null @@ -1,3163 +0,0 @@ -2026-01-14 21:20:59,561 - mmdet - INFO - Environment info: ------------------------------------------------------------- -sys.platform: linux -Python: 3.10.12 (main, Aug 15 2025, 14:32:43) [GCC 11.4.0] -CUDA available: True -GPU 0,1,2,3,4,5,6,7: BW151 -CUDA_HOME: /opt/dtk -NVCC: Not Available -GCC: x86_64-linux-gnu-gcc (Ubuntu 11.4.0-1ubuntu1~22.04) 11.4.0 -PyTorch: 2.5.1 -PyTorch compiling details: PyTorch built with: - - GCC 10.3 - - C++ Version: 201703 - - Intel(R) Math Kernel Library Version 2020.0.4 Product Build 20200917 for Intel(R) 64 architecture applications - - OpenMP 201511 (a.k.a. OpenMP 4.5) - - LAPACK is enabled (usually provided by MKL) - - NNPACK is enabled - - CPU capability usage: AVX512 - - HIP Runtime 6.3.25521 - - MIOpen 2.18.0 - - Magma 2.8.0 - - Build settings: BLAS_INFO=mkl, BUILD_TYPE=Release, CXX_COMPILER=/opt/rh/gcc-toolset-10/root/usr/bin/c++, CXX_FLAGS= -D_GLIBCXX_USE_CXX11_ABI=1 -fvisibility-inlines-hidden -DUSE_PTHREADPOOL -DNDEBUG -DUSE_KINETO -DLIBKINETO_NOCUPTI -DLIBKINETO_NOXPUPTI=ON -DUSE_FBGEMM -DUSE_PYTORCH_QNNPACK -DUSE_XNNPACK -DSYMBOLICATE_MOBILE_DEBUG_HANDLE -O2 -fPIC -Wall -Wextra -Werror=return-type -Werror=non-virtual-dtor -Werror=bool-operation -Wnarrowing -Wno-missing-field-initializers -Wno-type-limits -Wno-array-bounds -Wno-unknown-pragmas -Wno-unused-parameter -Wno-strict-overflow -Wno-strict-aliasing -Wno-stringop-overflow -Wsuggest-override -Wno-psabi -Wno-error=old-style-cast -Wno-missing-braces -fdiagnostics-color=always -faligned-new -Wno-unused-but-set-variable -Wno-maybe-uninitialized -fno-math-errno -fno-trapping-math -Werror=format -Wno-stringop-overflow, FORCE_FALLBACK_CUDA_MPI=1, LAPACK_INFO=mkl, PERF_WITH_AVX=1, PERF_WITH_AVX2=1, TORCH_VERSION=2.5.1, USE_CUDA=0, USE_CUDNN=OFF, USE_CUSPARSELT=OFF, USE_EXCEPTION_PTR=1, USE_GFLAGS=1, USE_GLOG=1, USE_GLOO=1, USE_MKL=ON, USE_MKLDNN=0, USE_MPI=1, USE_NCCL=1, USE_NNPACK=ON, USE_OPENMP=1, USE_ROCM=ON, USE_ROCM_KERNEL_ASSERT=OFF, - -TorchVision: 0.20.1 -OpenCV: 4.12.0 -MMCV: 1.6.1 -MMCV Compiler: GCC 10.3 -MMCV CUDA Compiler: rocm not available -MMDetection: 2.25.1+c41df4b ------------------------------------------------------------- - -2026-01-14 21:21:00,303 - mmdet - INFO - Distributed training: True -2026-01-14 21:21:01,205 - mmdet - INFO - Config: -plugin = True -plugin_dir = 'projects/mmdet3d_plugin/' -dist_params = dict(backend='nccl') -log_level = 'INFO' -work_dir = './work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704' -total_batch_size = 160 -num_gpus = 8 -batch_size = 20 -num_iters_per_epoch = 175 -num_epochs = 100 -checkpoint_epoch_interval = 20 -checkpoint_config = dict(interval=3500) -log_config = dict( - interval=1, - hooks=[ - dict(type='TextLoggerHook', by_epoch=False), - dict(type='TensorboardLoggerHook') - ]) -load_from = None -resume_from = None -workflow = [('train', 1)] -fp16 = dict(loss_scale=32.0) -input_shape = (704, 256) -tracking_test = True -tracking_threshold = 0.2 -class_names = [ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier', - 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' -] -num_classes = 10 -embed_dims = 256 -num_groups = 8 -num_decoder = 6 -num_single_frame_decoder = 1 -use_deformable_func = True -strides = [4, 8, 16, 32] -num_levels = 4 -num_depth_layers = 3 -drop_out = 0.1 -temporal = True -decouple_attn = True -with_quality_estimation = True -model = dict( - type='Sparse4D', - use_grid_mask=True, - use_deformable_func=True, - img_backbone=dict( - type='ResNet', - depth=50, - num_stages=4, - frozen_stages=-1, - norm_eval=False, - style='pytorch', - with_cp=True, - out_indices=(0, 1, 2, 3), - norm_cfg=dict(type='BN', requires_grad=True), - pretrained='ckpt/resnet50-19c8e357.pth'), - img_neck=dict( - type='FPN', - num_outs=4, - start_level=0, - out_channels=256, - add_extra_convs='on_output', - relu_before_extra_convs=True, - in_channels=[256, 512, 1024, 2048]), - depth_branch=dict( - type='DenseDepthNet', - embed_dims=256, - num_depth_layers=3, - loss_weight=0.2), - head=dict( - type='Sparse4DHead', - cls_threshold_to_reg=0.05, - decouple_attn=True, - instance_bank=dict( - type='InstanceBank', - num_anchor=900, - embed_dims=256, - anchor='nuscenes_kmeans900.npy', - anchor_handler=dict(type='SparseBox3DKeyPointsGenerator'), - num_temp_instances=600, - confidence_decay=0.6, - feat_grad=False), - anchor_encoder=dict( - type='SparseBox3DEncoder', - vel_dims=3, - embed_dims=[128, 32, 32, 64], - mode='cat', - output_fc=False, - in_loops=1, - out_loops=4), - num_single_frame_decoder=1, - operation_order=[ - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine' - ], - temp_graph_model=dict( - type='MultiheadAttention', - embed_dims=512, - num_heads=8, - batch_first=True, - dropout=0.1), - graph_model=dict( - type='MultiheadAttention', - embed_dims=512, - num_heads=8, - batch_first=True, - dropout=0.1), - norm_layer=dict(type='LN', normalized_shape=256), - ffn=dict( - type='AsymmetricFFN', - in_channels=512, - pre_norm=dict(type='LN'), - embed_dims=256, - feedforward_channels=1024, - num_fcs=2, - ffn_drop=0.1, - act_cfg=dict(type='ReLU', inplace=True)), - deformable_model=dict( - type='DeformableFeatureAggregation', - embed_dims=256, - num_groups=8, - num_levels=4, - num_cams=6, - attn_drop=0.15, - use_deformable_func=True, - use_camera_embed=True, - residual_mode='cat', - kps_generator=dict( - type='SparseBox3DKeyPointsGenerator', - num_learnable_pts=6, - fix_scale=[[0, 0, 0], [0.45, 0, 0], [-0.45, 0, 0], - [0, 0.45, 0], [0, -0.45, 0], [0, 0, 0.45], - [0, 0, -0.45]])), - refine_layer=dict( - type='SparseBox3DRefinementModule', - embed_dims=256, - num_cls=10, - refine_yaw=True, - with_quality_estimation=True), - sampler=dict( - type='SparseBox3DTarget', - num_dn_groups=5, - num_temp_dn_groups=3, - dn_noise_scale=[2.0, 2.0, 2.0, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5], - max_dn_gt=32, - add_neg_dn=True, - cls_weight=2.0, - box_weight=0.25, - reg_weights=[2.0, 2.0, 2.0, 0.5, 0.5, 0.5, 0.0, 0.0, 0.0, 0.0], - cls_wise_reg_weights=dict( - {9: [2.0, 2.0, 2.0, 1.0, 1.0, 1.0, 0.0, 0.0, 1.0, 1.0]})), - loss_cls=dict( - type='FocalLoss', - use_sigmoid=True, - gamma=2.0, - alpha=0.25, - loss_weight=2.0), - loss_reg=dict( - type='SparseBox3DLoss', - loss_box=dict(type='L1Loss', loss_weight=0.25), - loss_centerness=dict(type='CrossEntropyLoss', use_sigmoid=True), - loss_yawness=dict(type='GaussianFocalLoss'), - cls_allow_reverse=[5]), - decoder=dict(type='SparseBox3DDecoder'), - reg_weights=[2.0, 2.0, 2.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0])) -dataset_type = 'NuScenes3DDetTrackDataset' -data_root = 'data/nuscenes/' -anno_root = 'data/nuscenes_anno_pkls/' -file_client_args = dict(backend='disk') -img_norm_cfg = dict( - mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375], to_rgb=True) -train_pipeline = [ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict( - type='LoadPointsFromFile', - coord_type='LIDAR', - load_dim=5, - use_dim=5, - file_client_args=dict(backend='disk')), - dict(type='ResizeCropFlipImage'), - dict(type='MultiScaleDepthMapGenerator', downsample=[4, 8, 16]), - dict(type='BBoxRotation'), - dict(type='PhotoMetricDistortionMultiViewImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict( - type='CircleObjectRangeFilter', - class_dist_thred=[55, 55, 55, 55, 55, 55, 55, 55, 55, 55]), - dict( - type='InstanceNameFilter', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ]), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=[ - 'img', 'timestamp', 'projection_mat', 'image_wh', 'gt_depth', - 'focal', 'gt_bboxes_3d', 'gt_labels_3d' - ], - meta_keys=['T_global', 'T_global_inv', 'timestamp', 'instance_id']) -] -test_pipeline = [ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='ResizeCropFlipImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=['img', 'timestamp', 'projection_mat', 'image_wh'], - meta_keys=['T_global', 'T_global_inv', 'timestamp']) -] -input_modality = dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False) -data_basic_config = dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier', - 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval') -data_aug_conf = dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]) -data = dict( - samples_per_gpu=20, - workers_per_gpu=20, - train=dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval', - ann_file='data/nuscenes_anno_pkls/nuscenes_infos_train.pkl', - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict( - type='LoadPointsFromFile', - coord_type='LIDAR', - load_dim=5, - use_dim=5, - file_client_args=dict(backend='disk')), - dict(type='ResizeCropFlipImage'), - dict(type='MultiScaleDepthMapGenerator', downsample=[4, 8, 16]), - dict(type='BBoxRotation'), - dict(type='PhotoMetricDistortionMultiViewImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict( - type='CircleObjectRangeFilter', - class_dist_thred=[55, 55, 55, 55, 55, 55, 55, 55, 55, 55]), - dict( - type='InstanceNameFilter', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', - 'traffic_cone' - ]), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=[ - 'img', 'timestamp', 'projection_mat', 'image_wh', - 'gt_depth', 'focal', 'gt_bboxes_3d', 'gt_labels_3d' - ], - meta_keys=[ - 'T_global', 'T_global_inv', 'timestamp', 'instance_id' - ]) - ], - test_mode=False, - data_aug_conf=dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]), - with_seq_flag=True, - sequences_split_num=2, - keep_consistent_seq_aug=True), - val=dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval', - ann_file='data/nuscenes_anno_pkls/nuscenes_infos_val.pkl', - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='ResizeCropFlipImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=['img', 'timestamp', 'projection_mat', 'image_wh'], - meta_keys=['T_global', 'T_global_inv', 'timestamp']) - ], - data_aug_conf=dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]), - test_mode=True, - tracking=True, - tracking_threshold=0.2), - test=dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval', - ann_file='data/nuscenes_anno_pkls/nuscenes_infos_val.pkl', - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='ResizeCropFlipImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=['img', 'timestamp', 'projection_mat', 'image_wh'], - meta_keys=['T_global', 'T_global_inv', 'timestamp']) - ], - data_aug_conf=dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]), - test_mode=True, - tracking=True, - tracking_threshold=0.2)) -optimizer = dict( - type='Miopen_AdamW', - lr=0.0006, - weight_decay=0.001, - paramwise_cfg=dict(custom_keys=dict(img_backbone=dict(lr_mult=0.5)))) -optimizer_config = dict(grad_clip=dict(max_norm=25, norm_type=2)) -lr_config = dict( - policy='CosineAnnealing', - warmup='linear', - warmup_iters=500, - warmup_ratio=0.3333333333333333, - min_lr_ratio=0.001) -runner = dict(type='IterBasedRunner', max_iters=17500) -vis_pipeline = [ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='Collect', keys=['img'], meta_keys=['timestamp', 'lidar2img']) -] -evaluation = dict( - interval=3500, - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict( - type='Collect', keys=['img'], meta_keys=['timestamp', 'lidar2img']) - ]) -gpu_ids = range(0, 8) - -2026-01-14 21:21:01,206 - mmdet - INFO - Set random seed to 0, deterministic: False -2026-01-14 21:21:01,538 - mmdet - INFO - initialize ResNet with init_cfg {'type': 'Pretrained', 'checkpoint': 'ckpt/resnet50-19c8e357.pth'} -2026-01-14 21:21:01,764 - mmdet - INFO - initialize FPN with init_cfg {'type': 'Xavier', 'layer': 'Conv2d', 'distribution': 'uniform'} -Name of parameter - Initialization information - -img_backbone.conv1.weight - torch.Size([64, 3, 7, 7]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.bn1.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.bn1.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.conv1.weight - torch.Size([64, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn1.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn1.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.conv2.weight - torch.Size([64, 64, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn2.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn2.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.conv3.weight - torch.Size([256, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn3.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn3.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.downsample.0.weight - torch.Size([256, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.downsample.1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.downsample.1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.conv1.weight - torch.Size([64, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn1.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn1.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.conv2.weight - torch.Size([64, 64, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn2.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn2.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.conv3.weight - torch.Size([256, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn3.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn3.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.conv1.weight - torch.Size([64, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn1.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn1.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.conv2.weight - torch.Size([64, 64, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn2.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn2.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.conv3.weight - torch.Size([256, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn3.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn3.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.conv1.weight - torch.Size([128, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn1.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn1.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.conv2.weight - torch.Size([128, 128, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn2.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn2.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.conv3.weight - torch.Size([512, 128, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn3.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn3.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.downsample.0.weight - torch.Size([512, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.downsample.1.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.downsample.1.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.conv1.weight - torch.Size([128, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn1.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn1.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.conv2.weight - torch.Size([128, 128, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn2.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn2.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.conv3.weight - torch.Size([512, 128, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn3.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn3.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.conv1.weight - torch.Size([128, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn1.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn1.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.conv2.weight - torch.Size([128, 128, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn2.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn2.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.conv3.weight - torch.Size([512, 128, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn3.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn3.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.conv1.weight - torch.Size([128, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn1.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn1.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.conv2.weight - torch.Size([128, 128, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn2.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn2.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.conv3.weight - torch.Size([512, 128, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn3.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn3.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.conv1.weight - torch.Size([256, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.downsample.0.weight - torch.Size([1024, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.downsample.1.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.downsample.1.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.conv1.weight - torch.Size([512, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn1.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn1.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.conv2.weight - torch.Size([512, 512, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn2.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn2.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.conv3.weight - torch.Size([2048, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn3.weight - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn3.bias - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.downsample.0.weight - torch.Size([2048, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.downsample.1.weight - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.downsample.1.bias - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.conv1.weight - torch.Size([512, 2048, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn1.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn1.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.conv2.weight - torch.Size([512, 512, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn2.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn2.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.conv3.weight - torch.Size([2048, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn3.weight - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn3.bias - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.conv1.weight - torch.Size([512, 2048, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn1.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn1.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.conv2.weight - torch.Size([512, 512, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn2.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn2.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.conv3.weight - torch.Size([2048, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn3.weight - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn3.bias - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_neck.lateral_convs.0.conv.weight - torch.Size([256, 256, 1, 1]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.lateral_convs.0.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.lateral_convs.1.conv.weight - torch.Size([256, 512, 1, 1]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.lateral_convs.1.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.lateral_convs.2.conv.weight - torch.Size([256, 1024, 1, 1]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.lateral_convs.2.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.lateral_convs.3.conv.weight - torch.Size([256, 2048, 1, 1]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.lateral_convs.3.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.fpn_convs.0.conv.weight - torch.Size([256, 256, 3, 3]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.fpn_convs.0.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.fpn_convs.1.conv.weight - torch.Size([256, 256, 3, 3]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.fpn_convs.1.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.fpn_convs.2.conv.weight - torch.Size([256, 256, 3, 3]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.fpn_convs.2.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.fpn_convs.3.conv.weight - torch.Size([256, 256, 3, 3]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.fpn_convs.3.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.instance_bank.anchor - torch.Size([900, 11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.instance_bank.instance_feature - torch.Size([900, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.instance_bank.anchor_handler.fix_scale - torch.Size([1, 3]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.0.weight - torch.Size([128, 3]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.0.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.2.weight - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.2.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.3.weight - torch.Size([128, 128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.3.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.5.weight - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.5.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.6.weight - torch.Size([128, 128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.6.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.8.weight - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.8.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.9.weight - torch.Size([128, 128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.9.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.11.weight - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.11.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.0.weight - torch.Size([32, 3]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.0.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.2.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.2.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.3.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.3.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.5.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.5.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.6.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.6.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.8.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.8.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.9.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.9.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.11.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.11.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.0.weight - torch.Size([32, 2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.0.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.2.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.2.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.3.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.3.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.5.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.5.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.6.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.6.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.8.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.8.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.9.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.9.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.11.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.11.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.0.weight - torch.Size([64, 3]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.0.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.2.weight - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.2.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.3.weight - torch.Size([64, 64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.3.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.5.weight - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.5.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.6.weight - torch.Size([64, 64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.6.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.8.weight - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.8.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.9.weight - torch.Size([64, 64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.9.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.11.weight - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.11.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.1.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.1.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.1.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.1.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.1.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.1.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.1.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.1.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.3.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.4.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.4.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.4.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.4.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.5.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.5.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.5.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.5.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.6.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.6.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.8.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.8.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.8.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.8.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.8.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.8.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.8.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.8.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.10.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.11.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.11.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.11.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.11.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.12.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.12.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.12.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.12.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.13.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.13.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.15.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.15.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.15.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.15.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.15.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.15.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.15.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.15.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.16.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.16.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.17.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.18.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.18.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.18.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.18.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.19.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.19.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.19.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.19.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.20.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.20.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.22.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.22.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.22.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.22.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.22.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.22.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.22.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.22.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.23.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.23.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.24.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.25.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.25.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.25.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.25.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.26.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.26.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.26.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.26.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.27.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.27.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.29.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.29.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.29.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.29.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.29.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.29.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.29.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.29.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.30.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.30.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.31.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.32.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.32.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.32.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.32.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.33.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.33.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.33.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.33.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.34.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.34.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.36.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.36.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.36.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.36.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.36.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.36.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.36.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.36.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.37.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.37.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.38.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.fc_before.weight - torch.Size([512, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.fc_after.weight - torch.Size([256, 512]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.0.weight - torch.Size([1, 256, 1, 1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.0.bias - torch.Size([1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.1.weight - torch.Size([1, 256, 1, 1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.1.bias - torch.Size([1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.2.weight - torch.Size([1, 256, 1, 1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.2.bias - torch.Size([1]): -The value is the same before and after calling `init_weights` of Sparse4D -2026-01-14 21:21:01,867 - mmdet - INFO - Model: -Sparse4D( - (img_backbone): ResNet( - (conv1): Conv2d(3, 64, kernel_size=(7, 7), stride=(2, 2), padding=(3, 3), bias=False) - (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (maxpool): MaxPool2d(kernel_size=3, stride=2, padding=1, dilation=1, ceil_mode=False) - (layer1): ResLayer( - (0): Bottleneck( - (conv1): Conv2d(64, 64, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (downsample): Sequential( - (0): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - ) - ) - (1): Bottleneck( - (conv1): Conv2d(256, 64, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (2): Bottleneck( - (conv1): Conv2d(256, 64, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - ) - (layer2): ResLayer( - (0): Bottleneck( - (conv1): Conv2d(256, 128, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (downsample): Sequential( - (0): Conv2d(256, 512, kernel_size=(1, 1), stride=(2, 2), bias=False) - (1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - ) - ) - (1): Bottleneck( - (conv1): Conv2d(512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (2): Bottleneck( - (conv1): Conv2d(512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (3): Bottleneck( - (conv1): Conv2d(512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - ) - (layer3): ResLayer( - (0): Bottleneck( - (conv1): Conv2d(512, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (downsample): Sequential( - (0): Conv2d(512, 1024, kernel_size=(1, 1), stride=(2, 2), bias=False) - (1): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - ) - ) - (1): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (2): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (3): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (4): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (5): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - ) - (layer4): ResLayer( - (0): Bottleneck( - (conv1): Conv2d(1024, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(512, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (downsample): Sequential( - (0): Conv2d(1024, 2048, kernel_size=(1, 1), stride=(2, 2), bias=False) - (1): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - ) - ) - (1): Bottleneck( - (conv1): Conv2d(2048, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(512, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (2): Bottleneck( - (conv1): Conv2d(2048, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(512, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - ) - ) - init_cfg={'type': 'Pretrained', 'checkpoint': 'ckpt/resnet50-19c8e357.pth'} - (img_neck): FPN( - (lateral_convs): ModuleList( - (0): ConvModule( - (conv): Conv2d(256, 256, kernel_size=(1, 1), stride=(1, 1)) - ) - (1): ConvModule( - (conv): Conv2d(512, 256, kernel_size=(1, 1), stride=(1, 1)) - ) - (2): ConvModule( - (conv): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1)) - ) - (3): ConvModule( - (conv): Conv2d(2048, 256, kernel_size=(1, 1), stride=(1, 1)) - ) - ) - (fpn_convs): ModuleList( - (0-3): 4 x ConvModule( - (conv): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1)) - ) - ) - ) - init_cfg={'type': 'Xavier', 'layer': 'Conv2d', 'distribution': 'uniform'} - (head): Sparse4DHead( - (instance_bank): InstanceBank( - (anchor_handler): SparseBox3DKeyPointsGenerator() - ) - (anchor_encoder): SparseBox3DEncoder( - (pos_fc): Sequential( - (0): Linear(in_features=3, out_features=128, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((128,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=128, out_features=128, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((128,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=128, out_features=128, bias=True) - (7): ReLU(inplace=True) - (8): LayerNorm((128,), eps=1e-05, elementwise_affine=True) - (9): Linear(in_features=128, out_features=128, bias=True) - (10): ReLU(inplace=True) - (11): LayerNorm((128,), eps=1e-05, elementwise_affine=True) - ) - (size_fc): Sequential( - (0): Linear(in_features=3, out_features=32, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=32, out_features=32, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=32, out_features=32, bias=True) - (7): ReLU(inplace=True) - (8): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (9): Linear(in_features=32, out_features=32, bias=True) - (10): ReLU(inplace=True) - (11): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - ) - (yaw_fc): Sequential( - (0): Linear(in_features=2, out_features=32, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=32, out_features=32, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=32, out_features=32, bias=True) - (7): ReLU(inplace=True) - (8): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (9): Linear(in_features=32, out_features=32, bias=True) - (10): ReLU(inplace=True) - (11): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - ) - (vel_fc): Sequential( - (0): Linear(in_features=3, out_features=64, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((64,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=64, out_features=64, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((64,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=64, out_features=64, bias=True) - (7): ReLU(inplace=True) - (8): LayerNorm((64,), eps=1e-05, elementwise_affine=True) - (9): Linear(in_features=64, out_features=64, bias=True) - (10): ReLU(inplace=True) - (11): LayerNorm((64,), eps=1e-05, elementwise_affine=True) - ) - ) - (loss_cls): FocalLoss() - (loss_reg): SparseBox3DLoss( - (loss_box): L1Loss() - (loss_cns): CrossEntropyLoss(avg_non_ignore=False) - (loss_yns): GaussianFocalLoss() - ) - (layers): ModuleList( - (0): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (1): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (4-5): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (6): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (7): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (8): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (11-12): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (13): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (14): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (15): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (16): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (17): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (18-19): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (20): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (21): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (22): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (23): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (24): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (25-26): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (27): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (28): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (29): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (30): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (31): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (32-33): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (34): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (35): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (36): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (37): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (38): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - ) - (fc_before): Linear(in_features=256, out_features=512, bias=False) - (fc_after): Linear(in_features=512, out_features=256, bias=False) - ) - (depth_branch): DenseDepthNet( - (depth_layers): ModuleList( - (0-2): 3 x Conv2d(256, 1, kernel_size=(1, 1), stride=(1, 1)) - ) - ) - (grid_mask): GridMask() -) diff --git a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20260114_212152.log b/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20260114_212152.log deleted file mode 100644 index 5bbec80cbd3712ecd234d0d11143489dac1102fd..0000000000000000000000000000000000000000 --- a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20260114_212152.log +++ /dev/null @@ -1,3220 +0,0 @@ -2026-01-14 21:21:52,505 - mmdet - INFO - Environment info: ------------------------------------------------------------- -sys.platform: linux -Python: 3.10.12 (main, Aug 15 2025, 14:32:43) [GCC 11.4.0] -CUDA available: True -GPU 0,1,2,3,4,5,6,7: BW151 -CUDA_HOME: /opt/dtk -NVCC: Not Available -GCC: x86_64-linux-gnu-gcc (Ubuntu 11.4.0-1ubuntu1~22.04) 11.4.0 -PyTorch: 2.5.1 -PyTorch compiling details: PyTorch built with: - - GCC 10.3 - - C++ Version: 201703 - - Intel(R) Math Kernel Library Version 2020.0.4 Product Build 20200917 for Intel(R) 64 architecture applications - - OpenMP 201511 (a.k.a. OpenMP 4.5) - - LAPACK is enabled (usually provided by MKL) - - NNPACK is enabled - - CPU capability usage: AVX512 - - HIP Runtime 6.3.25521 - - MIOpen 2.18.0 - - Magma 2.8.0 - - Build settings: BLAS_INFO=mkl, BUILD_TYPE=Release, CXX_COMPILER=/opt/rh/gcc-toolset-10/root/usr/bin/c++, CXX_FLAGS= -D_GLIBCXX_USE_CXX11_ABI=1 -fvisibility-inlines-hidden -DUSE_PTHREADPOOL -DNDEBUG -DUSE_KINETO -DLIBKINETO_NOCUPTI -DLIBKINETO_NOXPUPTI=ON -DUSE_FBGEMM -DUSE_PYTORCH_QNNPACK -DUSE_XNNPACK -DSYMBOLICATE_MOBILE_DEBUG_HANDLE -O2 -fPIC -Wall -Wextra -Werror=return-type -Werror=non-virtual-dtor -Werror=bool-operation -Wnarrowing -Wno-missing-field-initializers -Wno-type-limits -Wno-array-bounds -Wno-unknown-pragmas -Wno-unused-parameter -Wno-strict-overflow -Wno-strict-aliasing -Wno-stringop-overflow -Wsuggest-override -Wno-psabi -Wno-error=old-style-cast -Wno-missing-braces -fdiagnostics-color=always -faligned-new -Wno-unused-but-set-variable -Wno-maybe-uninitialized -fno-math-errno -fno-trapping-math -Werror=format -Wno-stringop-overflow, FORCE_FALLBACK_CUDA_MPI=1, LAPACK_INFO=mkl, PERF_WITH_AVX=1, PERF_WITH_AVX2=1, TORCH_VERSION=2.5.1, USE_CUDA=0, USE_CUDNN=OFF, USE_CUSPARSELT=OFF, USE_EXCEPTION_PTR=1, USE_GFLAGS=1, USE_GLOG=1, USE_GLOO=1, USE_MKL=ON, USE_MKLDNN=0, USE_MPI=1, USE_NCCL=1, USE_NNPACK=ON, USE_OPENMP=1, USE_ROCM=ON, USE_ROCM_KERNEL_ASSERT=OFF, - -TorchVision: 0.20.1 -OpenCV: 4.12.0 -MMCV: 1.6.1 -MMCV Compiler: GCC 10.3 -MMCV CUDA Compiler: rocm not available -MMDetection: 2.25.1+c41df4b ------------------------------------------------------------- - -2026-01-14 21:21:53,247 - mmdet - INFO - Distributed training: True -2026-01-14 21:21:54,152 - mmdet - INFO - Config: -plugin = True -plugin_dir = 'projects/mmdet3d_plugin/' -dist_params = dict(backend='nccl') -log_level = 'INFO' -work_dir = './work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704' -total_batch_size = 160 -num_gpus = 8 -batch_size = 20 -num_iters_per_epoch = 175 -num_epochs = 100 -checkpoint_epoch_interval = 20 -checkpoint_config = dict(interval=3500) -log_config = dict( - interval=1, - hooks=[ - dict(type='TextLoggerHook', by_epoch=False), - dict(type='TensorboardLoggerHook') - ]) -load_from = None -resume_from = None -workflow = [('train', 1)] -fp16 = dict(loss_scale=32.0) -input_shape = (704, 256) -tracking_test = True -tracking_threshold = 0.2 -class_names = [ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier', - 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' -] -num_classes = 10 -embed_dims = 256 -num_groups = 8 -num_decoder = 6 -num_single_frame_decoder = 1 -use_deformable_func = True -strides = [4, 8, 16, 32] -num_levels = 4 -num_depth_layers = 3 -drop_out = 0.1 -temporal = True -decouple_attn = True -with_quality_estimation = True -model = dict( - type='Sparse4D', - use_grid_mask=True, - use_deformable_func=True, - img_backbone=dict( - type='ResNet', - depth=50, - num_stages=4, - frozen_stages=-1, - norm_eval=False, - style='pytorch', - with_cp=True, - out_indices=(0, 1, 2, 3), - norm_cfg=dict(type='BN', requires_grad=True), - pretrained='ckpt/resnet50-19c8e357.pth'), - img_neck=dict( - type='FPN', - num_outs=4, - start_level=0, - out_channels=256, - add_extra_convs='on_output', - relu_before_extra_convs=True, - in_channels=[256, 512, 1024, 2048]), - depth_branch=dict( - type='DenseDepthNet', - embed_dims=256, - num_depth_layers=3, - loss_weight=0.2), - head=dict( - type='Sparse4DHead', - cls_threshold_to_reg=0.05, - decouple_attn=True, - instance_bank=dict( - type='InstanceBank', - num_anchor=900, - embed_dims=256, - anchor='nuscenes_kmeans900.npy', - anchor_handler=dict(type='SparseBox3DKeyPointsGenerator'), - num_temp_instances=600, - confidence_decay=0.6, - feat_grad=False), - anchor_encoder=dict( - type='SparseBox3DEncoder', - vel_dims=3, - embed_dims=[128, 32, 32, 64], - mode='cat', - output_fc=False, - in_loops=1, - out_loops=4), - num_single_frame_decoder=1, - operation_order=[ - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine' - ], - temp_graph_model=dict( - type='MultiheadAttention', - embed_dims=512, - num_heads=8, - batch_first=True, - dropout=0.1), - graph_model=dict( - type='MultiheadAttention', - embed_dims=512, - num_heads=8, - batch_first=True, - dropout=0.1), - norm_layer=dict(type='LN', normalized_shape=256), - ffn=dict( - type='AsymmetricFFN', - in_channels=512, - pre_norm=dict(type='LN'), - embed_dims=256, - feedforward_channels=1024, - num_fcs=2, - ffn_drop=0.1, - act_cfg=dict(type='ReLU', inplace=True)), - deformable_model=dict( - type='DeformableFeatureAggregation', - embed_dims=256, - num_groups=8, - num_levels=4, - num_cams=6, - attn_drop=0.15, - use_deformable_func=True, - use_camera_embed=True, - residual_mode='cat', - kps_generator=dict( - type='SparseBox3DKeyPointsGenerator', - num_learnable_pts=6, - fix_scale=[[0, 0, 0], [0.45, 0, 0], [-0.45, 0, 0], - [0, 0.45, 0], [0, -0.45, 0], [0, 0, 0.45], - [0, 0, -0.45]])), - refine_layer=dict( - type='SparseBox3DRefinementModule', - embed_dims=256, - num_cls=10, - refine_yaw=True, - with_quality_estimation=True), - sampler=dict( - type='SparseBox3DTarget', - num_dn_groups=5, - num_temp_dn_groups=3, - dn_noise_scale=[2.0, 2.0, 2.0, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5], - max_dn_gt=32, - add_neg_dn=True, - cls_weight=2.0, - box_weight=0.25, - reg_weights=[2.0, 2.0, 2.0, 0.5, 0.5, 0.5, 0.0, 0.0, 0.0, 0.0], - cls_wise_reg_weights=dict( - {9: [2.0, 2.0, 2.0, 1.0, 1.0, 1.0, 0.0, 0.0, 1.0, 1.0]})), - loss_cls=dict( - type='FocalLoss', - use_sigmoid=True, - gamma=2.0, - alpha=0.25, - loss_weight=2.0), - loss_reg=dict( - type='SparseBox3DLoss', - loss_box=dict(type='L1Loss', loss_weight=0.25), - loss_centerness=dict(type='CrossEntropyLoss', use_sigmoid=True), - loss_yawness=dict(type='GaussianFocalLoss'), - cls_allow_reverse=[5]), - decoder=dict(type='SparseBox3DDecoder'), - reg_weights=[2.0, 2.0, 2.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0])) -dataset_type = 'NuScenes3DDetTrackDataset' -data_root = 'data/nuscenes/' -anno_root = 'data/nuscenes_anno_pkls/' -file_client_args = dict(backend='disk') -img_norm_cfg = dict( - mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375], to_rgb=True) -train_pipeline = [ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict( - type='LoadPointsFromFile', - coord_type='LIDAR', - load_dim=5, - use_dim=5, - file_client_args=dict(backend='disk')), - dict(type='ResizeCropFlipImage'), - dict(type='MultiScaleDepthMapGenerator', downsample=[4, 8, 16]), - dict(type='BBoxRotation'), - dict(type='PhotoMetricDistortionMultiViewImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict( - type='CircleObjectRangeFilter', - class_dist_thred=[55, 55, 55, 55, 55, 55, 55, 55, 55, 55]), - dict( - type='InstanceNameFilter', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ]), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=[ - 'img', 'timestamp', 'projection_mat', 'image_wh', 'gt_depth', - 'focal', 'gt_bboxes_3d', 'gt_labels_3d' - ], - meta_keys=['T_global', 'T_global_inv', 'timestamp', 'instance_id']) -] -test_pipeline = [ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='ResizeCropFlipImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=['img', 'timestamp', 'projection_mat', 'image_wh'], - meta_keys=['T_global', 'T_global_inv', 'timestamp']) -] -input_modality = dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False) -data_basic_config = dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier', - 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval') -data_aug_conf = dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]) -data = dict( - samples_per_gpu=20, - workers_per_gpu=20, - train=dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval', - ann_file='data/nuscenes_anno_pkls/nuscenes_infos_train.pkl', - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict( - type='LoadPointsFromFile', - coord_type='LIDAR', - load_dim=5, - use_dim=5, - file_client_args=dict(backend='disk')), - dict(type='ResizeCropFlipImage'), - dict(type='MultiScaleDepthMapGenerator', downsample=[4, 8, 16]), - dict(type='BBoxRotation'), - dict(type='PhotoMetricDistortionMultiViewImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict( - type='CircleObjectRangeFilter', - class_dist_thred=[55, 55, 55, 55, 55, 55, 55, 55, 55, 55]), - dict( - type='InstanceNameFilter', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', - 'traffic_cone' - ]), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=[ - 'img', 'timestamp', 'projection_mat', 'image_wh', - 'gt_depth', 'focal', 'gt_bboxes_3d', 'gt_labels_3d' - ], - meta_keys=[ - 'T_global', 'T_global_inv', 'timestamp', 'instance_id' - ]) - ], - test_mode=False, - data_aug_conf=dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]), - with_seq_flag=True, - sequences_split_num=2, - keep_consistent_seq_aug=True), - val=dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval', - ann_file='data/nuscenes_anno_pkls/nuscenes_infos_val.pkl', - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='ResizeCropFlipImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=['img', 'timestamp', 'projection_mat', 'image_wh'], - meta_keys=['T_global', 'T_global_inv', 'timestamp']) - ], - data_aug_conf=dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]), - test_mode=True, - tracking=True, - tracking_threshold=0.2), - test=dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval', - ann_file='data/nuscenes_anno_pkls/nuscenes_infos_val.pkl', - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='ResizeCropFlipImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=['img', 'timestamp', 'projection_mat', 'image_wh'], - meta_keys=['T_global', 'T_global_inv', 'timestamp']) - ], - data_aug_conf=dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]), - test_mode=True, - tracking=True, - tracking_threshold=0.2)) -optimizer = dict( - type='AdamW', - lr=0.0006, - weight_decay=0.001, - paramwise_cfg=dict(custom_keys=dict(img_backbone=dict(lr_mult=0.5)))) -optimizer_config = dict(grad_clip=dict(max_norm=25, norm_type=2)) -lr_config = dict( - policy='CosineAnnealing', - warmup='linear', - warmup_iters=500, - warmup_ratio=0.3333333333333333, - min_lr_ratio=0.001) -runner = dict(type='IterBasedRunner', max_iters=17500) -vis_pipeline = [ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='Collect', keys=['img'], meta_keys=['timestamp', 'lidar2img']) -] -evaluation = dict( - interval=3500, - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict( - type='Collect', keys=['img'], meta_keys=['timestamp', 'lidar2img']) - ]) -gpu_ids = range(0, 8) - -2026-01-14 21:21:54,152 - mmdet - INFO - Set random seed to 0, deterministic: False -2026-01-14 21:21:54,487 - mmdet - INFO - initialize ResNet with init_cfg {'type': 'Pretrained', 'checkpoint': 'ckpt/resnet50-19c8e357.pth'} -2026-01-14 21:21:54,695 - mmdet - INFO - initialize FPN with init_cfg {'type': 'Xavier', 'layer': 'Conv2d', 'distribution': 'uniform'} -Name of parameter - Initialization information - -img_backbone.conv1.weight - torch.Size([64, 3, 7, 7]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.bn1.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.bn1.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.conv1.weight - torch.Size([64, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn1.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn1.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.conv2.weight - torch.Size([64, 64, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn2.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn2.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.conv3.weight - torch.Size([256, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn3.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn3.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.downsample.0.weight - torch.Size([256, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.downsample.1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.downsample.1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.conv1.weight - torch.Size([64, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn1.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn1.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.conv2.weight - torch.Size([64, 64, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn2.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn2.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.conv3.weight - torch.Size([256, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn3.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn3.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.conv1.weight - torch.Size([64, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn1.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn1.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.conv2.weight - torch.Size([64, 64, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn2.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn2.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.conv3.weight - torch.Size([256, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn3.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn3.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.conv1.weight - torch.Size([128, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn1.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn1.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.conv2.weight - torch.Size([128, 128, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn2.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn2.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.conv3.weight - torch.Size([512, 128, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn3.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn3.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.downsample.0.weight - torch.Size([512, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.downsample.1.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.downsample.1.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.conv1.weight - torch.Size([128, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn1.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn1.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.conv2.weight - torch.Size([128, 128, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn2.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn2.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.conv3.weight - torch.Size([512, 128, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn3.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn3.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.conv1.weight - torch.Size([128, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn1.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn1.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.conv2.weight - torch.Size([128, 128, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn2.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn2.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.conv3.weight - torch.Size([512, 128, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn3.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn3.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.conv1.weight - torch.Size([128, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn1.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn1.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.conv2.weight - torch.Size([128, 128, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn2.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn2.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.conv3.weight - torch.Size([512, 128, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn3.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn3.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.conv1.weight - torch.Size([256, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.downsample.0.weight - torch.Size([1024, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.downsample.1.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.downsample.1.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.conv1.weight - torch.Size([512, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn1.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn1.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.conv2.weight - torch.Size([512, 512, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn2.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn2.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.conv3.weight - torch.Size([2048, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn3.weight - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn3.bias - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.downsample.0.weight - torch.Size([2048, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.downsample.1.weight - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.downsample.1.bias - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.conv1.weight - torch.Size([512, 2048, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn1.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn1.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.conv2.weight - torch.Size([512, 512, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn2.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn2.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.conv3.weight - torch.Size([2048, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn3.weight - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn3.bias - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.conv1.weight - torch.Size([512, 2048, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn1.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn1.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.conv2.weight - torch.Size([512, 512, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn2.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn2.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.conv3.weight - torch.Size([2048, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn3.weight - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn3.bias - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_neck.lateral_convs.0.conv.weight - torch.Size([256, 256, 1, 1]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.lateral_convs.0.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.lateral_convs.1.conv.weight - torch.Size([256, 512, 1, 1]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.lateral_convs.1.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.lateral_convs.2.conv.weight - torch.Size([256, 1024, 1, 1]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.lateral_convs.2.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.lateral_convs.3.conv.weight - torch.Size([256, 2048, 1, 1]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.lateral_convs.3.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.fpn_convs.0.conv.weight - torch.Size([256, 256, 3, 3]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.fpn_convs.0.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.fpn_convs.1.conv.weight - torch.Size([256, 256, 3, 3]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.fpn_convs.1.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.fpn_convs.2.conv.weight - torch.Size([256, 256, 3, 3]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.fpn_convs.2.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.fpn_convs.3.conv.weight - torch.Size([256, 256, 3, 3]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.fpn_convs.3.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.instance_bank.anchor - torch.Size([900, 11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.instance_bank.instance_feature - torch.Size([900, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.instance_bank.anchor_handler.fix_scale - torch.Size([1, 3]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.0.weight - torch.Size([128, 3]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.0.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.2.weight - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.2.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.3.weight - torch.Size([128, 128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.3.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.5.weight - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.5.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.6.weight - torch.Size([128, 128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.6.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.8.weight - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.8.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.9.weight - torch.Size([128, 128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.9.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.11.weight - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.11.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.0.weight - torch.Size([32, 3]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.0.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.2.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.2.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.3.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.3.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.5.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.5.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.6.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.6.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.8.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.8.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.9.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.9.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.11.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.11.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.0.weight - torch.Size([32, 2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.0.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.2.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.2.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.3.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.3.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.5.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.5.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.6.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.6.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.8.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.8.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.9.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.9.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.11.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.11.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.0.weight - torch.Size([64, 3]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.0.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.2.weight - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.2.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.3.weight - torch.Size([64, 64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.3.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.5.weight - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.5.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.6.weight - torch.Size([64, 64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.6.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.8.weight - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.8.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.9.weight - torch.Size([64, 64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.9.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.11.weight - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.11.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.1.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.1.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.1.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.1.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.1.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.1.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.1.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.1.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.3.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.4.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.4.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.4.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.4.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.5.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.5.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.5.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.5.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.6.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.6.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.8.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.8.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.8.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.8.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.8.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.8.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.8.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.8.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.10.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.11.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.11.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.11.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.11.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.12.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.12.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.12.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.12.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.13.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.13.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.15.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.15.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.15.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.15.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.15.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.15.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.15.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.15.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.16.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.16.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.17.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.18.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.18.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.18.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.18.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.19.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.19.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.19.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.19.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.20.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.20.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.22.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.22.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.22.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.22.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.22.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.22.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.22.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.22.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.23.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.23.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.24.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.25.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.25.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.25.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.25.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.26.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.26.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.26.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.26.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.27.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.27.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.29.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.29.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.29.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.29.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.29.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.29.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.29.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.29.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.30.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.30.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.31.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.32.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.32.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.32.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.32.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.33.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.33.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.33.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.33.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.34.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.34.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.36.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.36.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.36.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.36.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.36.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.36.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.36.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.36.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.37.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.37.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.38.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.fc_before.weight - torch.Size([512, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.fc_after.weight - torch.Size([256, 512]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.0.weight - torch.Size([1, 256, 1, 1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.0.bias - torch.Size([1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.1.weight - torch.Size([1, 256, 1, 1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.1.bias - torch.Size([1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.2.weight - torch.Size([1, 256, 1, 1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.2.bias - torch.Size([1]): -The value is the same before and after calling `init_weights` of Sparse4D -2026-01-14 21:21:54,800 - mmdet - INFO - Model: -Sparse4D( - (img_backbone): ResNet( - (conv1): Conv2d(3, 64, kernel_size=(7, 7), stride=(2, 2), padding=(3, 3), bias=False) - (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (maxpool): MaxPool2d(kernel_size=3, stride=2, padding=1, dilation=1, ceil_mode=False) - (layer1): ResLayer( - (0): Bottleneck( - (conv1): Conv2d(64, 64, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (downsample): Sequential( - (0): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - ) - ) - (1): Bottleneck( - (conv1): Conv2d(256, 64, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (2): Bottleneck( - (conv1): Conv2d(256, 64, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - ) - (layer2): ResLayer( - (0): Bottleneck( - (conv1): Conv2d(256, 128, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (downsample): Sequential( - (0): Conv2d(256, 512, kernel_size=(1, 1), stride=(2, 2), bias=False) - (1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - ) - ) - (1): Bottleneck( - (conv1): Conv2d(512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (2): Bottleneck( - (conv1): Conv2d(512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (3): Bottleneck( - (conv1): Conv2d(512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - ) - (layer3): ResLayer( - (0): Bottleneck( - (conv1): Conv2d(512, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (downsample): Sequential( - (0): Conv2d(512, 1024, kernel_size=(1, 1), stride=(2, 2), bias=False) - (1): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - ) - ) - (1): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (2): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (3): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (4): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (5): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - ) - (layer4): ResLayer( - (0): Bottleneck( - (conv1): Conv2d(1024, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(512, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (downsample): Sequential( - (0): Conv2d(1024, 2048, kernel_size=(1, 1), stride=(2, 2), bias=False) - (1): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - ) - ) - (1): Bottleneck( - (conv1): Conv2d(2048, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(512, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (2): Bottleneck( - (conv1): Conv2d(2048, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(512, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - ) - ) - init_cfg={'type': 'Pretrained', 'checkpoint': 'ckpt/resnet50-19c8e357.pth'} - (img_neck): FPN( - (lateral_convs): ModuleList( - (0): ConvModule( - (conv): Conv2d(256, 256, kernel_size=(1, 1), stride=(1, 1)) - ) - (1): ConvModule( - (conv): Conv2d(512, 256, kernel_size=(1, 1), stride=(1, 1)) - ) - (2): ConvModule( - (conv): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1)) - ) - (3): ConvModule( - (conv): Conv2d(2048, 256, kernel_size=(1, 1), stride=(1, 1)) - ) - ) - (fpn_convs): ModuleList( - (0-3): 4 x ConvModule( - (conv): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1)) - ) - ) - ) - init_cfg={'type': 'Xavier', 'layer': 'Conv2d', 'distribution': 'uniform'} - (head): Sparse4DHead( - (instance_bank): InstanceBank( - (anchor_handler): SparseBox3DKeyPointsGenerator() - ) - (anchor_encoder): SparseBox3DEncoder( - (pos_fc): Sequential( - (0): Linear(in_features=3, out_features=128, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((128,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=128, out_features=128, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((128,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=128, out_features=128, bias=True) - (7): ReLU(inplace=True) - (8): LayerNorm((128,), eps=1e-05, elementwise_affine=True) - (9): Linear(in_features=128, out_features=128, bias=True) - (10): ReLU(inplace=True) - (11): LayerNorm((128,), eps=1e-05, elementwise_affine=True) - ) - (size_fc): Sequential( - (0): Linear(in_features=3, out_features=32, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=32, out_features=32, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=32, out_features=32, bias=True) - (7): ReLU(inplace=True) - (8): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (9): Linear(in_features=32, out_features=32, bias=True) - (10): ReLU(inplace=True) - (11): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - ) - (yaw_fc): Sequential( - (0): Linear(in_features=2, out_features=32, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=32, out_features=32, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=32, out_features=32, bias=True) - (7): ReLU(inplace=True) - (8): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (9): Linear(in_features=32, out_features=32, bias=True) - (10): ReLU(inplace=True) - (11): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - ) - (vel_fc): Sequential( - (0): Linear(in_features=3, out_features=64, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((64,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=64, out_features=64, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((64,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=64, out_features=64, bias=True) - (7): ReLU(inplace=True) - (8): LayerNorm((64,), eps=1e-05, elementwise_affine=True) - (9): Linear(in_features=64, out_features=64, bias=True) - (10): ReLU(inplace=True) - (11): LayerNorm((64,), eps=1e-05, elementwise_affine=True) - ) - ) - (loss_cls): FocalLoss() - (loss_reg): SparseBox3DLoss( - (loss_box): L1Loss() - (loss_cns): CrossEntropyLoss(avg_non_ignore=False) - (loss_yns): GaussianFocalLoss() - ) - (layers): ModuleList( - (0): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (1): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (4-5): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (6): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (7): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (8): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (11-12): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (13): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (14): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (15): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (16): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (17): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (18-19): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (20): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (21): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (22): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (23): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (24): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (25-26): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (27): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (28): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (29): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (30): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (31): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (32-33): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (34): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (35): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (36): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (37): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (38): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - ) - (fc_before): Linear(in_features=256, out_features=512, bias=False) - (fc_after): Linear(in_features=512, out_features=256, bias=False) - ) - (depth_branch): DenseDepthNet( - (depth_layers): ModuleList( - (0-2): 3 x Conv2d(256, 1, kernel_size=(1, 1), stride=(1, 1)) - ) - ) - (grid_mask): GridMask() -) -2026-01-14 21:22:06,388 - mmdet - INFO - Start running, host: root@bw150-1, work_dir: /workspace/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704 -2026-01-14 21:22:06,389 - mmdet - INFO - Hooks will be executed in the following order: -before_run: -(VERY_HIGH ) CosineAnnealingLrUpdaterHook -(ABOVE_NORMAL) Fp16OptimizerHook -(NORMAL ) CheckpointHook -(NORMAL ) CustomDistEvalHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -before_train_epoch: -(VERY_HIGH ) CosineAnnealingLrUpdaterHook -(NORMAL ) CustomDistEvalHook -(LOW ) IterTimerHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -before_train_iter: -(VERY_HIGH ) CosineAnnealingLrUpdaterHook -(NORMAL ) CustomDistEvalHook -(LOW ) IterTimerHook - -------------------- -after_train_iter: -(ABOVE_NORMAL) Fp16OptimizerHook -(NORMAL ) CheckpointHook -(NORMAL ) CustomDistEvalHook -(LOW ) IterTimerHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -after_train_epoch: -(NORMAL ) CheckpointHook -(NORMAL ) CustomDistEvalHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -before_val_epoch: -(LOW ) IterTimerHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -before_val_iter: -(LOW ) IterTimerHook - -------------------- -after_val_iter: -(LOW ) IterTimerHook - -------------------- -after_val_epoch: -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -after_run: -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -2026-01-14 21:22:06,389 - mmdet - INFO - workflow: [('train', 1)], max: 17500 iters -2026-01-14 21:22:06,391 - mmdet - INFO - Checkpoints will be saved to /workspace/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704 by HardDiskBackend. diff --git a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20260114_212152.log.json b/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20260114_212152.log.json deleted file mode 100644 index ea2a21ab5f5569092898c906a8d97e06b8c494d9..0000000000000000000000000000000000000000 --- a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20260114_212152.log.json +++ /dev/null @@ -1 +0,0 @@ -{"env_info": "sys.platform: linux\nPython: 3.10.12 (main, Aug 15 2025, 14:32:43) [GCC 11.4.0]\nCUDA available: True\nGPU 0,1,2,3,4,5,6,7: BW151\nCUDA_HOME: /opt/dtk\nNVCC: Not Available\nGCC: x86_64-linux-gnu-gcc (Ubuntu 11.4.0-1ubuntu1~22.04) 11.4.0\nPyTorch: 2.5.1\nPyTorch compiling details: PyTorch built with:\n - GCC 10.3\n - C++ Version: 201703\n - Intel(R) Math Kernel Library Version 2020.0.4 Product Build 20200917 for Intel(R) 64 architecture applications\n - OpenMP 201511 (a.k.a. OpenMP 4.5)\n - LAPACK is enabled (usually provided by MKL)\n - NNPACK is enabled\n - CPU capability usage: AVX512\n - HIP Runtime 6.3.25521\n - MIOpen 2.18.0\n - Magma 2.8.0\n - Build settings: BLAS_INFO=mkl, BUILD_TYPE=Release, CXX_COMPILER=/opt/rh/gcc-toolset-10/root/usr/bin/c++, CXX_FLAGS= -D_GLIBCXX_USE_CXX11_ABI=1 -fvisibility-inlines-hidden -DUSE_PTHREADPOOL -DNDEBUG -DUSE_KINETO -DLIBKINETO_NOCUPTI -DLIBKINETO_NOXPUPTI=ON -DUSE_FBGEMM -DUSE_PYTORCH_QNNPACK -DUSE_XNNPACK -DSYMBOLICATE_MOBILE_DEBUG_HANDLE -O2 -fPIC -Wall -Wextra -Werror=return-type -Werror=non-virtual-dtor -Werror=bool-operation -Wnarrowing -Wno-missing-field-initializers -Wno-type-limits -Wno-array-bounds -Wno-unknown-pragmas -Wno-unused-parameter -Wno-strict-overflow -Wno-strict-aliasing -Wno-stringop-overflow -Wsuggest-override -Wno-psabi -Wno-error=old-style-cast -Wno-missing-braces -fdiagnostics-color=always -faligned-new -Wno-unused-but-set-variable -Wno-maybe-uninitialized -fno-math-errno -fno-trapping-math -Werror=format -Wno-stringop-overflow, FORCE_FALLBACK_CUDA_MPI=1, LAPACK_INFO=mkl, PERF_WITH_AVX=1, PERF_WITH_AVX2=1, TORCH_VERSION=2.5.1, USE_CUDA=0, USE_CUDNN=OFF, USE_CUSPARSELT=OFF, USE_EXCEPTION_PTR=1, USE_GFLAGS=1, USE_GLOG=1, USE_GLOO=1, USE_MKL=ON, USE_MKLDNN=0, USE_MPI=1, USE_NCCL=1, USE_NNPACK=ON, USE_OPENMP=1, USE_ROCM=ON, USE_ROCM_KERNEL_ASSERT=OFF, \n\nTorchVision: 0.20.1\nOpenCV: 4.12.0\nMMCV: 1.6.1\nMMCV Compiler: GCC 10.3\nMMCV CUDA Compiler: rocm not available\nMMDetection: 2.25.1+c41df4b", "config": "plugin = True\nplugin_dir = 'projects/mmdet3d_plugin/'\ndist_params = dict(backend='nccl')\nlog_level = 'INFO'\nwork_dir = './work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704'\ntotal_batch_size = 160\nnum_gpus = 8\nbatch_size = 20\nnum_iters_per_epoch = 175\nnum_epochs = 100\ncheckpoint_epoch_interval = 20\ncheckpoint_config = dict(interval=3500)\nlog_config = dict(\n interval=1,\n hooks=[\n dict(type='TextLoggerHook', by_epoch=False),\n dict(type='TensorboardLoggerHook')\n ])\nload_from = None\nresume_from = None\nworkflow = [('train', 1)]\nfp16 = dict(loss_scale=32.0)\ninput_shape = (704, 256)\ntracking_test = True\ntracking_threshold = 0.2\nclass_names = [\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier',\n 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n]\nnum_classes = 10\nembed_dims = 256\nnum_groups = 8\nnum_decoder = 6\nnum_single_frame_decoder = 1\nuse_deformable_func = True\nstrides = [4, 8, 16, 32]\nnum_levels = 4\nnum_depth_layers = 3\ndrop_out = 0.1\ntemporal = True\ndecouple_attn = True\nwith_quality_estimation = True\nmodel = dict(\n type='Sparse4D',\n use_grid_mask=True,\n use_deformable_func=True,\n img_backbone=dict(\n type='ResNet',\n depth=50,\n num_stages=4,\n frozen_stages=-1,\n norm_eval=False,\n style='pytorch',\n with_cp=True,\n out_indices=(0, 1, 2, 3),\n norm_cfg=dict(type='BN', requires_grad=True),\n pretrained='ckpt/resnet50-19c8e357.pth'),\n img_neck=dict(\n type='FPN',\n num_outs=4,\n start_level=0,\n out_channels=256,\n add_extra_convs='on_output',\n relu_before_extra_convs=True,\n in_channels=[256, 512, 1024, 2048]),\n depth_branch=dict(\n type='DenseDepthNet',\n embed_dims=256,\n num_depth_layers=3,\n loss_weight=0.2),\n head=dict(\n type='Sparse4DHead',\n cls_threshold_to_reg=0.05,\n decouple_attn=True,\n instance_bank=dict(\n type='InstanceBank',\n num_anchor=900,\n embed_dims=256,\n anchor='nuscenes_kmeans900.npy',\n anchor_handler=dict(type='SparseBox3DKeyPointsGenerator'),\n num_temp_instances=600,\n confidence_decay=0.6,\n feat_grad=False),\n anchor_encoder=dict(\n type='SparseBox3DEncoder',\n vel_dims=3,\n embed_dims=[128, 32, 32, 64],\n mode='cat',\n output_fc=False,\n in_loops=1,\n out_loops=4),\n num_single_frame_decoder=1,\n operation_order=[\n 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm',\n 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm',\n 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm',\n 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm',\n 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm',\n 'deformable', 'ffn', 'norm', 'refine'\n ],\n temp_graph_model=dict(\n type='MultiheadAttention',\n embed_dims=512,\n num_heads=8,\n batch_first=True,\n dropout=0.1),\n graph_model=dict(\n type='MultiheadAttention',\n embed_dims=512,\n num_heads=8,\n batch_first=True,\n dropout=0.1),\n norm_layer=dict(type='LN', normalized_shape=256),\n ffn=dict(\n type='AsymmetricFFN',\n in_channels=512,\n pre_norm=dict(type='LN'),\n embed_dims=256,\n feedforward_channels=1024,\n num_fcs=2,\n ffn_drop=0.1,\n act_cfg=dict(type='ReLU', inplace=True)),\n deformable_model=dict(\n type='DeformableFeatureAggregation',\n embed_dims=256,\n num_groups=8,\n num_levels=4,\n num_cams=6,\n attn_drop=0.15,\n use_deformable_func=True,\n use_camera_embed=True,\n residual_mode='cat',\n kps_generator=dict(\n type='SparseBox3DKeyPointsGenerator',\n num_learnable_pts=6,\n fix_scale=[[0, 0, 0], [0.45, 0, 0], [-0.45, 0, 0],\n [0, 0.45, 0], [0, -0.45, 0], [0, 0, 0.45],\n [0, 0, -0.45]])),\n refine_layer=dict(\n type='SparseBox3DRefinementModule',\n embed_dims=256,\n num_cls=10,\n refine_yaw=True,\n with_quality_estimation=True),\n sampler=dict(\n type='SparseBox3DTarget',\n num_dn_groups=5,\n num_temp_dn_groups=3,\n dn_noise_scale=[2.0, 2.0, 2.0, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5],\n max_dn_gt=32,\n add_neg_dn=True,\n cls_weight=2.0,\n box_weight=0.25,\n reg_weights=[2.0, 2.0, 2.0, 0.5, 0.5, 0.5, 0.0, 0.0, 0.0, 0.0],\n cls_wise_reg_weights=dict(\n {9: [2.0, 2.0, 2.0, 1.0, 1.0, 1.0, 0.0, 0.0, 1.0, 1.0]})),\n loss_cls=dict(\n type='FocalLoss',\n use_sigmoid=True,\n gamma=2.0,\n alpha=0.25,\n loss_weight=2.0),\n loss_reg=dict(\n type='SparseBox3DLoss',\n loss_box=dict(type='L1Loss', loss_weight=0.25),\n loss_centerness=dict(type='CrossEntropyLoss', use_sigmoid=True),\n loss_yawness=dict(type='GaussianFocalLoss'),\n cls_allow_reverse=[5]),\n decoder=dict(type='SparseBox3DDecoder'),\n reg_weights=[2.0, 2.0, 2.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0]))\ndataset_type = 'NuScenes3DDetTrackDataset'\ndata_root = 'data/nuscenes/'\nanno_root = 'data/nuscenes_anno_pkls/'\nfile_client_args = dict(backend='disk')\nimg_norm_cfg = dict(\n mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375], to_rgb=True)\ntrain_pipeline = [\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(\n type='LoadPointsFromFile',\n coord_type='LIDAR',\n load_dim=5,\n use_dim=5,\n file_client_args=dict(backend='disk')),\n dict(type='ResizeCropFlipImage'),\n dict(type='MultiScaleDepthMapGenerator', downsample=[4, 8, 16]),\n dict(type='BBoxRotation'),\n dict(type='PhotoMetricDistortionMultiViewImage'),\n dict(\n type='NormalizeMultiviewImage',\n mean=[123.675, 116.28, 103.53],\n std=[58.395, 57.12, 57.375],\n to_rgb=True),\n dict(\n type='CircleObjectRangeFilter',\n class_dist_thred=[55, 55, 55, 55, 55, 55, 55, 55, 55, 55]),\n dict(\n type='InstanceNameFilter',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ]),\n dict(type='NuScenesSparse4DAdaptor'),\n dict(\n type='Collect',\n keys=[\n 'img', 'timestamp', 'projection_mat', 'image_wh', 'gt_depth',\n 'focal', 'gt_bboxes_3d', 'gt_labels_3d'\n ],\n meta_keys=['T_global', 'T_global_inv', 'timestamp', 'instance_id'])\n]\ntest_pipeline = [\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(type='ResizeCropFlipImage'),\n dict(\n type='NormalizeMultiviewImage',\n mean=[123.675, 116.28, 103.53],\n std=[58.395, 57.12, 57.375],\n to_rgb=True),\n dict(type='NuScenesSparse4DAdaptor'),\n dict(\n type='Collect',\n keys=['img', 'timestamp', 'projection_mat', 'image_wh'],\n meta_keys=['T_global', 'T_global_inv', 'timestamp'])\n]\ninput_modality = dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False)\ndata_basic_config = dict(\n type='NuScenes3DDetTrackDataset',\n data_root='data/nuscenes/',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier',\n 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ],\n modality=dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False),\n version='v1.0-trainval')\ndata_aug_conf = dict(\n resize_lim=(0.4, 0.47),\n final_dim=(256, 704),\n bot_pct_lim=(0.0, 0.0),\n rot_lim=(-5.4, 5.4),\n H=900,\n W=1600,\n rand_flip=True,\n rot3d_range=[-0.3925, 0.3925])\ndata = dict(\n samples_per_gpu=20,\n workers_per_gpu=20,\n train=dict(\n type='NuScenes3DDetTrackDataset',\n data_root='data/nuscenes/',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ],\n modality=dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False),\n version='v1.0-trainval',\n ann_file='data/nuscenes_anno_pkls/nuscenes_infos_train.pkl',\n pipeline=[\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(\n type='LoadPointsFromFile',\n coord_type='LIDAR',\n load_dim=5,\n use_dim=5,\n file_client_args=dict(backend='disk')),\n dict(type='ResizeCropFlipImage'),\n dict(type='MultiScaleDepthMapGenerator', downsample=[4, 8, 16]),\n dict(type='BBoxRotation'),\n dict(type='PhotoMetricDistortionMultiViewImage'),\n dict(\n type='NormalizeMultiviewImage',\n mean=[123.675, 116.28, 103.53],\n std=[58.395, 57.12, 57.375],\n to_rgb=True),\n dict(\n type='CircleObjectRangeFilter',\n class_dist_thred=[55, 55, 55, 55, 55, 55, 55, 55, 55, 55]),\n dict(\n type='InstanceNameFilter',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian',\n 'traffic_cone'\n ]),\n dict(type='NuScenesSparse4DAdaptor'),\n dict(\n type='Collect',\n keys=[\n 'img', 'timestamp', 'projection_mat', 'image_wh',\n 'gt_depth', 'focal', 'gt_bboxes_3d', 'gt_labels_3d'\n ],\n meta_keys=[\n 'T_global', 'T_global_inv', 'timestamp', 'instance_id'\n ])\n ],\n test_mode=False,\n data_aug_conf=dict(\n resize_lim=(0.4, 0.47),\n final_dim=(256, 704),\n bot_pct_lim=(0.0, 0.0),\n rot_lim=(-5.4, 5.4),\n H=900,\n W=1600,\n rand_flip=True,\n rot3d_range=[-0.3925, 0.3925]),\n with_seq_flag=True,\n sequences_split_num=2,\n keep_consistent_seq_aug=True),\n val=dict(\n type='NuScenes3DDetTrackDataset',\n data_root='data/nuscenes/',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ],\n modality=dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False),\n version='v1.0-trainval',\n ann_file='data/nuscenes_anno_pkls/nuscenes_infos_val.pkl',\n pipeline=[\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(type='ResizeCropFlipImage'),\n dict(\n type='NormalizeMultiviewImage',\n mean=[123.675, 116.28, 103.53],\n std=[58.395, 57.12, 57.375],\n to_rgb=True),\n dict(type='NuScenesSparse4DAdaptor'),\n dict(\n type='Collect',\n keys=['img', 'timestamp', 'projection_mat', 'image_wh'],\n meta_keys=['T_global', 'T_global_inv', 'timestamp'])\n ],\n data_aug_conf=dict(\n resize_lim=(0.4, 0.47),\n final_dim=(256, 704),\n bot_pct_lim=(0.0, 0.0),\n rot_lim=(-5.4, 5.4),\n H=900,\n W=1600,\n rand_flip=True,\n rot3d_range=[-0.3925, 0.3925]),\n test_mode=True,\n tracking=True,\n tracking_threshold=0.2),\n test=dict(\n type='NuScenes3DDetTrackDataset',\n data_root='data/nuscenes/',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ],\n modality=dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False),\n version='v1.0-trainval',\n ann_file='data/nuscenes_anno_pkls/nuscenes_infos_val.pkl',\n pipeline=[\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(type='ResizeCropFlipImage'),\n dict(\n type='NormalizeMultiviewImage',\n mean=[123.675, 116.28, 103.53],\n std=[58.395, 57.12, 57.375],\n to_rgb=True),\n dict(type='NuScenesSparse4DAdaptor'),\n dict(\n type='Collect',\n keys=['img', 'timestamp', 'projection_mat', 'image_wh'],\n meta_keys=['T_global', 'T_global_inv', 'timestamp'])\n ],\n data_aug_conf=dict(\n resize_lim=(0.4, 0.47),\n final_dim=(256, 704),\n bot_pct_lim=(0.0, 0.0),\n rot_lim=(-5.4, 5.4),\n H=900,\n W=1600,\n rand_flip=True,\n rot3d_range=[-0.3925, 0.3925]),\n test_mode=True,\n tracking=True,\n tracking_threshold=0.2))\noptimizer = dict(\n type='AdamW',\n lr=0.0006,\n weight_decay=0.001,\n paramwise_cfg=dict(custom_keys=dict(img_backbone=dict(lr_mult=0.5))))\noptimizer_config = dict(grad_clip=dict(max_norm=25, norm_type=2))\nlr_config = dict(\n policy='CosineAnnealing',\n warmup='linear',\n warmup_iters=500,\n warmup_ratio=0.3333333333333333,\n min_lr_ratio=0.001)\nrunner = dict(type='IterBasedRunner', max_iters=17500)\nvis_pipeline = [\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(type='Collect', keys=['img'], meta_keys=['timestamp', 'lidar2img'])\n]\nevaluation = dict(\n interval=3500,\n pipeline=[\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(\n type='Collect', keys=['img'], meta_keys=['timestamp', 'lidar2img'])\n ])\ngpu_ids = range(0, 8)\n", "seed": 0, "exp_name": "sparse4dv3_temporal_r50_1x8_bs6_256x704.py"} diff --git a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20260114_212343.log b/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20260114_212343.log deleted file mode 100644 index bb283184b2abbb83e31d76b4915f741d1d70f60a..0000000000000000000000000000000000000000 --- a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20260114_212343.log +++ /dev/null @@ -1,3220 +0,0 @@ -2026-01-14 21:23:43,379 - mmdet - INFO - Environment info: ------------------------------------------------------------- -sys.platform: linux -Python: 3.10.12 (main, Aug 15 2025, 14:32:43) [GCC 11.4.0] -CUDA available: True -GPU 0,1,2,3,4,5,6,7: BW151 -CUDA_HOME: /opt/dtk -NVCC: Not Available -GCC: x86_64-linux-gnu-gcc (Ubuntu 11.4.0-1ubuntu1~22.04) 11.4.0 -PyTorch: 2.5.1 -PyTorch compiling details: PyTorch built with: - - GCC 10.3 - - C++ Version: 201703 - - Intel(R) Math Kernel Library Version 2020.0.4 Product Build 20200917 for Intel(R) 64 architecture applications - - OpenMP 201511 (a.k.a. OpenMP 4.5) - - LAPACK is enabled (usually provided by MKL) - - NNPACK is enabled - - CPU capability usage: AVX512 - - HIP Runtime 6.3.25521 - - MIOpen 2.18.0 - - Magma 2.8.0 - - Build settings: BLAS_INFO=mkl, BUILD_TYPE=Release, CXX_COMPILER=/opt/rh/gcc-toolset-10/root/usr/bin/c++, CXX_FLAGS= -D_GLIBCXX_USE_CXX11_ABI=1 -fvisibility-inlines-hidden -DUSE_PTHREADPOOL -DNDEBUG -DUSE_KINETO -DLIBKINETO_NOCUPTI -DLIBKINETO_NOXPUPTI=ON -DUSE_FBGEMM -DUSE_PYTORCH_QNNPACK -DUSE_XNNPACK -DSYMBOLICATE_MOBILE_DEBUG_HANDLE -O2 -fPIC -Wall -Wextra -Werror=return-type -Werror=non-virtual-dtor -Werror=bool-operation -Wnarrowing -Wno-missing-field-initializers -Wno-type-limits -Wno-array-bounds -Wno-unknown-pragmas -Wno-unused-parameter -Wno-strict-overflow -Wno-strict-aliasing -Wno-stringop-overflow -Wsuggest-override -Wno-psabi -Wno-error=old-style-cast -Wno-missing-braces -fdiagnostics-color=always -faligned-new -Wno-unused-but-set-variable -Wno-maybe-uninitialized -fno-math-errno -fno-trapping-math -Werror=format -Wno-stringop-overflow, FORCE_FALLBACK_CUDA_MPI=1, LAPACK_INFO=mkl, PERF_WITH_AVX=1, PERF_WITH_AVX2=1, TORCH_VERSION=2.5.1, USE_CUDA=0, USE_CUDNN=OFF, USE_CUSPARSELT=OFF, USE_EXCEPTION_PTR=1, USE_GFLAGS=1, USE_GLOG=1, USE_GLOO=1, USE_MKL=ON, USE_MKLDNN=0, USE_MPI=1, USE_NCCL=1, USE_NNPACK=ON, USE_OPENMP=1, USE_ROCM=ON, USE_ROCM_KERNEL_ASSERT=OFF, - -TorchVision: 0.20.1 -OpenCV: 4.12.0 -MMCV: 1.6.1 -MMCV Compiler: GCC 10.3 -MMCV CUDA Compiler: rocm not available -MMDetection: 2.25.1+c41df4b ------------------------------------------------------------- - -2026-01-14 21:23:44,115 - mmdet - INFO - Distributed training: True -2026-01-14 21:23:44,999 - mmdet - INFO - Config: -plugin = True -plugin_dir = 'projects/mmdet3d_plugin/' -dist_params = dict(backend='nccl') -log_level = 'INFO' -work_dir = './work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704' -total_batch_size = 160 -num_gpus = 8 -batch_size = 20 -num_iters_per_epoch = 175 -num_epochs = 100 -checkpoint_epoch_interval = 20 -checkpoint_config = dict(interval=3500) -log_config = dict( - interval=1, - hooks=[ - dict(type='TextLoggerHook', by_epoch=False), - dict(type='TensorboardLoggerHook') - ]) -load_from = None -resume_from = None -workflow = [('train', 1)] -fp16 = dict(loss_scale=32.0) -input_shape = (704, 256) -tracking_test = True -tracking_threshold = 0.2 -class_names = [ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier', - 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' -] -num_classes = 10 -embed_dims = 256 -num_groups = 8 -num_decoder = 6 -num_single_frame_decoder = 1 -use_deformable_func = True -strides = [4, 8, 16, 32] -num_levels = 4 -num_depth_layers = 3 -drop_out = 0.1 -temporal = True -decouple_attn = True -with_quality_estimation = True -model = dict( - type='Sparse4D', - use_grid_mask=True, - use_deformable_func=True, - img_backbone=dict( - type='ResNet', - depth=50, - num_stages=4, - frozen_stages=-1, - norm_eval=False, - style='pytorch', - with_cp=True, - out_indices=(0, 1, 2, 3), - norm_cfg=dict(type='BN', requires_grad=True), - pretrained='ckpt/resnet50-19c8e357.pth'), - img_neck=dict( - type='FPN', - num_outs=4, - start_level=0, - out_channels=256, - add_extra_convs='on_output', - relu_before_extra_convs=True, - in_channels=[256, 512, 1024, 2048]), - depth_branch=dict( - type='DenseDepthNet', - embed_dims=256, - num_depth_layers=3, - loss_weight=0.2), - head=dict( - type='Sparse4DHead', - cls_threshold_to_reg=0.05, - decouple_attn=True, - instance_bank=dict( - type='InstanceBank', - num_anchor=900, - embed_dims=256, - anchor='nuscenes_kmeans900.npy', - anchor_handler=dict(type='SparseBox3DKeyPointsGenerator'), - num_temp_instances=600, - confidence_decay=0.6, - feat_grad=False), - anchor_encoder=dict( - type='SparseBox3DEncoder', - vel_dims=3, - embed_dims=[128, 32, 32, 64], - mode='cat', - output_fc=False, - in_loops=1, - out_loops=4), - num_single_frame_decoder=1, - operation_order=[ - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine' - ], - temp_graph_model=dict( - type='MultiheadAttention', - embed_dims=512, - num_heads=8, - batch_first=True, - dropout=0.1), - graph_model=dict( - type='MultiheadAttention', - embed_dims=512, - num_heads=8, - batch_first=True, - dropout=0.1), - norm_layer=dict(type='LN', normalized_shape=256), - ffn=dict( - type='AsymmetricFFN', - in_channels=512, - pre_norm=dict(type='LN'), - embed_dims=256, - feedforward_channels=1024, - num_fcs=2, - ffn_drop=0.1, - act_cfg=dict(type='ReLU', inplace=True)), - deformable_model=dict( - type='DeformableFeatureAggregation', - embed_dims=256, - num_groups=8, - num_levels=4, - num_cams=6, - attn_drop=0.15, - use_deformable_func=True, - use_camera_embed=True, - residual_mode='cat', - kps_generator=dict( - type='SparseBox3DKeyPointsGenerator', - num_learnable_pts=6, - fix_scale=[[0, 0, 0], [0.45, 0, 0], [-0.45, 0, 0], - [0, 0.45, 0], [0, -0.45, 0], [0, 0, 0.45], - [0, 0, -0.45]])), - refine_layer=dict( - type='SparseBox3DRefinementModule', - embed_dims=256, - num_cls=10, - refine_yaw=True, - with_quality_estimation=True), - sampler=dict( - type='SparseBox3DTarget', - num_dn_groups=5, - num_temp_dn_groups=3, - dn_noise_scale=[2.0, 2.0, 2.0, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5], - max_dn_gt=32, - add_neg_dn=True, - cls_weight=2.0, - box_weight=0.25, - reg_weights=[2.0, 2.0, 2.0, 0.5, 0.5, 0.5, 0.0, 0.0, 0.0, 0.0], - cls_wise_reg_weights=dict( - {9: [2.0, 2.0, 2.0, 1.0, 1.0, 1.0, 0.0, 0.0, 1.0, 1.0]})), - loss_cls=dict( - type='FocalLoss', - use_sigmoid=True, - gamma=2.0, - alpha=0.25, - loss_weight=2.0), - loss_reg=dict( - type='SparseBox3DLoss', - loss_box=dict(type='L1Loss', loss_weight=0.25), - loss_centerness=dict(type='CrossEntropyLoss', use_sigmoid=True), - loss_yawness=dict(type='GaussianFocalLoss'), - cls_allow_reverse=[5]), - decoder=dict(type='SparseBox3DDecoder'), - reg_weights=[2.0, 2.0, 2.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0])) -dataset_type = 'NuScenes3DDetTrackDataset' -data_root = 'data/nuscenes/' -anno_root = 'data/nuscenes_anno_pkls/' -file_client_args = dict(backend='disk') -img_norm_cfg = dict( - mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375], to_rgb=True) -train_pipeline = [ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict( - type='LoadPointsFromFile', - coord_type='LIDAR', - load_dim=5, - use_dim=5, - file_client_args=dict(backend='disk')), - dict(type='ResizeCropFlipImage'), - dict(type='MultiScaleDepthMapGenerator', downsample=[4, 8, 16]), - dict(type='BBoxRotation'), - dict(type='PhotoMetricDistortionMultiViewImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict( - type='CircleObjectRangeFilter', - class_dist_thred=[55, 55, 55, 55, 55, 55, 55, 55, 55, 55]), - dict( - type='InstanceNameFilter', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ]), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=[ - 'img', 'timestamp', 'projection_mat', 'image_wh', 'gt_depth', - 'focal', 'gt_bboxes_3d', 'gt_labels_3d' - ], - meta_keys=['T_global', 'T_global_inv', 'timestamp', 'instance_id']) -] -test_pipeline = [ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='ResizeCropFlipImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=['img', 'timestamp', 'projection_mat', 'image_wh'], - meta_keys=['T_global', 'T_global_inv', 'timestamp']) -] -input_modality = dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False) -data_basic_config = dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier', - 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval') -data_aug_conf = dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]) -data = dict( - samples_per_gpu=20, - workers_per_gpu=20, - train=dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval', - ann_file='data/nuscenes_anno_pkls/nuscenes_infos_train.pkl', - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict( - type='LoadPointsFromFile', - coord_type='LIDAR', - load_dim=5, - use_dim=5, - file_client_args=dict(backend='disk')), - dict(type='ResizeCropFlipImage'), - dict(type='MultiScaleDepthMapGenerator', downsample=[4, 8, 16]), - dict(type='BBoxRotation'), - dict(type='PhotoMetricDistortionMultiViewImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict( - type='CircleObjectRangeFilter', - class_dist_thred=[55, 55, 55, 55, 55, 55, 55, 55, 55, 55]), - dict( - type='InstanceNameFilter', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', - 'traffic_cone' - ]), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=[ - 'img', 'timestamp', 'projection_mat', 'image_wh', - 'gt_depth', 'focal', 'gt_bboxes_3d', 'gt_labels_3d' - ], - meta_keys=[ - 'T_global', 'T_global_inv', 'timestamp', 'instance_id' - ]) - ], - test_mode=False, - data_aug_conf=dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]), - with_seq_flag=True, - sequences_split_num=2, - keep_consistent_seq_aug=True), - val=dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval', - ann_file='data/nuscenes_anno_pkls/nuscenes_infos_val.pkl', - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='ResizeCropFlipImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=['img', 'timestamp', 'projection_mat', 'image_wh'], - meta_keys=['T_global', 'T_global_inv', 'timestamp']) - ], - data_aug_conf=dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]), - test_mode=True, - tracking=True, - tracking_threshold=0.2), - test=dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval', - ann_file='data/nuscenes_anno_pkls/nuscenes_infos_val.pkl', - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='ResizeCropFlipImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=['img', 'timestamp', 'projection_mat', 'image_wh'], - meta_keys=['T_global', 'T_global_inv', 'timestamp']) - ], - data_aug_conf=dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]), - test_mode=True, - tracking=True, - tracking_threshold=0.2)) -optimizer = dict( - type='AdamW', - lr=0.0006, - weight_decay=0.001, - paramwise_cfg=dict(custom_keys=dict(img_backbone=dict(lr_mult=0.5)))) -optimizer_config = dict(grad_clip=dict(max_norm=25, norm_type=2)) -lr_config = dict( - policy='CosineAnnealing', - warmup='linear', - warmup_iters=500, - warmup_ratio=0.3333333333333333, - min_lr_ratio=0.001) -runner = dict(type='IterBasedRunner', max_iters=17500) -vis_pipeline = [ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='Collect', keys=['img'], meta_keys=['timestamp', 'lidar2img']) -] -evaluation = dict( - interval=3500, - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict( - type='Collect', keys=['img'], meta_keys=['timestamp', 'lidar2img']) - ]) -gpu_ids = range(0, 8) - -2026-01-14 21:23:44,999 - mmdet - INFO - Set random seed to 0, deterministic: False -2026-01-14 21:23:45,326 - mmdet - INFO - initialize ResNet with init_cfg {'type': 'Pretrained', 'checkpoint': 'ckpt/resnet50-19c8e357.pth'} -2026-01-14 21:23:45,553 - mmdet - INFO - initialize FPN with init_cfg {'type': 'Xavier', 'layer': 'Conv2d', 'distribution': 'uniform'} -Name of parameter - Initialization information - -img_backbone.conv1.weight - torch.Size([64, 3, 7, 7]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.bn1.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.bn1.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.conv1.weight - torch.Size([64, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn1.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn1.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.conv2.weight - torch.Size([64, 64, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn2.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn2.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.conv3.weight - torch.Size([256, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn3.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn3.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.downsample.0.weight - torch.Size([256, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.downsample.1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.downsample.1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.conv1.weight - torch.Size([64, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn1.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn1.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.conv2.weight - torch.Size([64, 64, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn2.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn2.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.conv3.weight - torch.Size([256, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn3.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn3.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.conv1.weight - torch.Size([64, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn1.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn1.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.conv2.weight - torch.Size([64, 64, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn2.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn2.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.conv3.weight - torch.Size([256, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn3.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn3.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.conv1.weight - torch.Size([128, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn1.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn1.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.conv2.weight - torch.Size([128, 128, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn2.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn2.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.conv3.weight - torch.Size([512, 128, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn3.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn3.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.downsample.0.weight - torch.Size([512, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.downsample.1.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.downsample.1.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.conv1.weight - torch.Size([128, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn1.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn1.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.conv2.weight - torch.Size([128, 128, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn2.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn2.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.conv3.weight - torch.Size([512, 128, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn3.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn3.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.conv1.weight - torch.Size([128, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn1.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn1.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.conv2.weight - torch.Size([128, 128, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn2.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn2.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.conv3.weight - torch.Size([512, 128, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn3.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn3.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.conv1.weight - torch.Size([128, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn1.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn1.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.conv2.weight - torch.Size([128, 128, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn2.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn2.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.conv3.weight - torch.Size([512, 128, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn3.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn3.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.conv1.weight - torch.Size([256, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.downsample.0.weight - torch.Size([1024, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.downsample.1.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.downsample.1.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.conv1.weight - torch.Size([512, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn1.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn1.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.conv2.weight - torch.Size([512, 512, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn2.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn2.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.conv3.weight - torch.Size([2048, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn3.weight - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn3.bias - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.downsample.0.weight - torch.Size([2048, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.downsample.1.weight - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.downsample.1.bias - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.conv1.weight - torch.Size([512, 2048, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn1.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn1.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.conv2.weight - torch.Size([512, 512, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn2.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn2.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.conv3.weight - torch.Size([2048, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn3.weight - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn3.bias - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.conv1.weight - torch.Size([512, 2048, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn1.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn1.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.conv2.weight - torch.Size([512, 512, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn2.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn2.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.conv3.weight - torch.Size([2048, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn3.weight - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn3.bias - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_neck.lateral_convs.0.conv.weight - torch.Size([256, 256, 1, 1]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.lateral_convs.0.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.lateral_convs.1.conv.weight - torch.Size([256, 512, 1, 1]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.lateral_convs.1.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.lateral_convs.2.conv.weight - torch.Size([256, 1024, 1, 1]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.lateral_convs.2.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.lateral_convs.3.conv.weight - torch.Size([256, 2048, 1, 1]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.lateral_convs.3.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.fpn_convs.0.conv.weight - torch.Size([256, 256, 3, 3]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.fpn_convs.0.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.fpn_convs.1.conv.weight - torch.Size([256, 256, 3, 3]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.fpn_convs.1.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.fpn_convs.2.conv.weight - torch.Size([256, 256, 3, 3]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.fpn_convs.2.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.fpn_convs.3.conv.weight - torch.Size([256, 256, 3, 3]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.fpn_convs.3.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.instance_bank.anchor - torch.Size([900, 11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.instance_bank.instance_feature - torch.Size([900, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.instance_bank.anchor_handler.fix_scale - torch.Size([1, 3]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.0.weight - torch.Size([128, 3]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.0.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.2.weight - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.2.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.3.weight - torch.Size([128, 128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.3.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.5.weight - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.5.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.6.weight - torch.Size([128, 128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.6.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.8.weight - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.8.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.9.weight - torch.Size([128, 128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.9.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.11.weight - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.11.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.0.weight - torch.Size([32, 3]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.0.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.2.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.2.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.3.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.3.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.5.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.5.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.6.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.6.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.8.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.8.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.9.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.9.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.11.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.11.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.0.weight - torch.Size([32, 2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.0.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.2.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.2.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.3.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.3.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.5.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.5.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.6.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.6.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.8.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.8.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.9.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.9.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.11.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.11.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.0.weight - torch.Size([64, 3]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.0.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.2.weight - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.2.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.3.weight - torch.Size([64, 64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.3.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.5.weight - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.5.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.6.weight - torch.Size([64, 64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.6.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.8.weight - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.8.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.9.weight - torch.Size([64, 64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.9.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.11.weight - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.11.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.1.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.1.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.1.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.1.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.1.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.1.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.1.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.1.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.3.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.4.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.4.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.4.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.4.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.5.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.5.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.5.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.5.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.6.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.6.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.8.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.8.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.8.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.8.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.8.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.8.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.8.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.8.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.10.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.11.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.11.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.11.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.11.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.12.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.12.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.12.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.12.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.13.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.13.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.15.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.15.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.15.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.15.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.15.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.15.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.15.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.15.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.16.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.16.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.17.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.18.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.18.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.18.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.18.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.19.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.19.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.19.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.19.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.20.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.20.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.22.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.22.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.22.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.22.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.22.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.22.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.22.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.22.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.23.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.23.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.24.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.25.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.25.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.25.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.25.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.26.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.26.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.26.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.26.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.27.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.27.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.29.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.29.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.29.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.29.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.29.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.29.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.29.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.29.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.30.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.30.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.31.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.32.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.32.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.32.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.32.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.33.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.33.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.33.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.33.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.34.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.34.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.36.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.36.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.36.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.36.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.36.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.36.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.36.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.36.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.37.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.37.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.38.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.fc_before.weight - torch.Size([512, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.fc_after.weight - torch.Size([256, 512]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.0.weight - torch.Size([1, 256, 1, 1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.0.bias - torch.Size([1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.1.weight - torch.Size([1, 256, 1, 1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.1.bias - torch.Size([1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.2.weight - torch.Size([1, 256, 1, 1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.2.bias - torch.Size([1]): -The value is the same before and after calling `init_weights` of Sparse4D -2026-01-14 21:23:45,657 - mmdet - INFO - Model: -Sparse4D( - (img_backbone): ResNet( - (conv1): Conv2d(3, 64, kernel_size=(7, 7), stride=(2, 2), padding=(3, 3), bias=False) - (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (maxpool): MaxPool2d(kernel_size=3, stride=2, padding=1, dilation=1, ceil_mode=False) - (layer1): ResLayer( - (0): Bottleneck( - (conv1): Conv2d(64, 64, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (downsample): Sequential( - (0): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - ) - ) - (1): Bottleneck( - (conv1): Conv2d(256, 64, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (2): Bottleneck( - (conv1): Conv2d(256, 64, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - ) - (layer2): ResLayer( - (0): Bottleneck( - (conv1): Conv2d(256, 128, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (downsample): Sequential( - (0): Conv2d(256, 512, kernel_size=(1, 1), stride=(2, 2), bias=False) - (1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - ) - ) - (1): Bottleneck( - (conv1): Conv2d(512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (2): Bottleneck( - (conv1): Conv2d(512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (3): Bottleneck( - (conv1): Conv2d(512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - ) - (layer3): ResLayer( - (0): Bottleneck( - (conv1): Conv2d(512, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (downsample): Sequential( - (0): Conv2d(512, 1024, kernel_size=(1, 1), stride=(2, 2), bias=False) - (1): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - ) - ) - (1): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (2): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (3): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (4): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (5): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - ) - (layer4): ResLayer( - (0): Bottleneck( - (conv1): Conv2d(1024, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(512, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (downsample): Sequential( - (0): Conv2d(1024, 2048, kernel_size=(1, 1), stride=(2, 2), bias=False) - (1): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - ) - ) - (1): Bottleneck( - (conv1): Conv2d(2048, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(512, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (2): Bottleneck( - (conv1): Conv2d(2048, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(512, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - ) - ) - init_cfg={'type': 'Pretrained', 'checkpoint': 'ckpt/resnet50-19c8e357.pth'} - (img_neck): FPN( - (lateral_convs): ModuleList( - (0): ConvModule( - (conv): Conv2d(256, 256, kernel_size=(1, 1), stride=(1, 1)) - ) - (1): ConvModule( - (conv): Conv2d(512, 256, kernel_size=(1, 1), stride=(1, 1)) - ) - (2): ConvModule( - (conv): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1)) - ) - (3): ConvModule( - (conv): Conv2d(2048, 256, kernel_size=(1, 1), stride=(1, 1)) - ) - ) - (fpn_convs): ModuleList( - (0-3): 4 x ConvModule( - (conv): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1)) - ) - ) - ) - init_cfg={'type': 'Xavier', 'layer': 'Conv2d', 'distribution': 'uniform'} - (head): Sparse4DHead( - (instance_bank): InstanceBank( - (anchor_handler): SparseBox3DKeyPointsGenerator() - ) - (anchor_encoder): SparseBox3DEncoder( - (pos_fc): Sequential( - (0): Linear(in_features=3, out_features=128, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((128,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=128, out_features=128, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((128,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=128, out_features=128, bias=True) - (7): ReLU(inplace=True) - (8): LayerNorm((128,), eps=1e-05, elementwise_affine=True) - (9): Linear(in_features=128, out_features=128, bias=True) - (10): ReLU(inplace=True) - (11): LayerNorm((128,), eps=1e-05, elementwise_affine=True) - ) - (size_fc): Sequential( - (0): Linear(in_features=3, out_features=32, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=32, out_features=32, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=32, out_features=32, bias=True) - (7): ReLU(inplace=True) - (8): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (9): Linear(in_features=32, out_features=32, bias=True) - (10): ReLU(inplace=True) - (11): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - ) - (yaw_fc): Sequential( - (0): Linear(in_features=2, out_features=32, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=32, out_features=32, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=32, out_features=32, bias=True) - (7): ReLU(inplace=True) - (8): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (9): Linear(in_features=32, out_features=32, bias=True) - (10): ReLU(inplace=True) - (11): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - ) - (vel_fc): Sequential( - (0): Linear(in_features=3, out_features=64, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((64,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=64, out_features=64, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((64,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=64, out_features=64, bias=True) - (7): ReLU(inplace=True) - (8): LayerNorm((64,), eps=1e-05, elementwise_affine=True) - (9): Linear(in_features=64, out_features=64, bias=True) - (10): ReLU(inplace=True) - (11): LayerNorm((64,), eps=1e-05, elementwise_affine=True) - ) - ) - (loss_cls): FocalLoss() - (loss_reg): SparseBox3DLoss( - (loss_box): L1Loss() - (loss_cns): CrossEntropyLoss(avg_non_ignore=False) - (loss_yns): GaussianFocalLoss() - ) - (layers): ModuleList( - (0): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (1): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (4-5): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (6): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (7): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (8): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (11-12): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (13): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (14): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (15): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (16): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (17): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (18-19): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (20): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (21): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (22): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (23): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (24): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (25-26): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (27): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (28): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (29): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (30): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (31): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (32-33): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (34): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (35): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (36): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (37): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (38): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - ) - (fc_before): Linear(in_features=256, out_features=512, bias=False) - (fc_after): Linear(in_features=512, out_features=256, bias=False) - ) - (depth_branch): DenseDepthNet( - (depth_layers): ModuleList( - (0-2): 3 x Conv2d(256, 1, kernel_size=(1, 1), stride=(1, 1)) - ) - ) - (grid_mask): GridMask() -) -2026-01-14 21:23:56,818 - mmdet - INFO - Start running, host: root@bw150-1, work_dir: /workspace/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704 -2026-01-14 21:23:56,818 - mmdet - INFO - Hooks will be executed in the following order: -before_run: -(VERY_HIGH ) CosineAnnealingLrUpdaterHook -(ABOVE_NORMAL) Fp16OptimizerHook -(NORMAL ) CheckpointHook -(NORMAL ) CustomDistEvalHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -before_train_epoch: -(VERY_HIGH ) CosineAnnealingLrUpdaterHook -(NORMAL ) CustomDistEvalHook -(LOW ) IterTimerHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -before_train_iter: -(VERY_HIGH ) CosineAnnealingLrUpdaterHook -(NORMAL ) CustomDistEvalHook -(LOW ) IterTimerHook - -------------------- -after_train_iter: -(ABOVE_NORMAL) Fp16OptimizerHook -(NORMAL ) CheckpointHook -(NORMAL ) CustomDistEvalHook -(LOW ) IterTimerHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -after_train_epoch: -(NORMAL ) CheckpointHook -(NORMAL ) CustomDistEvalHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -before_val_epoch: -(LOW ) IterTimerHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -before_val_iter: -(LOW ) IterTimerHook - -------------------- -after_val_iter: -(LOW ) IterTimerHook - -------------------- -after_val_epoch: -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -after_run: -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -2026-01-14 21:23:56,818 - mmdet - INFO - workflow: [('train', 1)], max: 17500 iters -2026-01-14 21:23:56,820 - mmdet - INFO - Checkpoints will be saved to /workspace/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704 by HardDiskBackend. diff --git a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20260114_212343.log.json b/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20260114_212343.log.json deleted file mode 100644 index ea2a21ab5f5569092898c906a8d97e06b8c494d9..0000000000000000000000000000000000000000 --- a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20260114_212343.log.json +++ /dev/null @@ -1 +0,0 @@ -{"env_info": "sys.platform: linux\nPython: 3.10.12 (main, Aug 15 2025, 14:32:43) [GCC 11.4.0]\nCUDA available: True\nGPU 0,1,2,3,4,5,6,7: BW151\nCUDA_HOME: /opt/dtk\nNVCC: Not Available\nGCC: x86_64-linux-gnu-gcc (Ubuntu 11.4.0-1ubuntu1~22.04) 11.4.0\nPyTorch: 2.5.1\nPyTorch compiling details: PyTorch built with:\n - GCC 10.3\n - C++ Version: 201703\n - Intel(R) Math Kernel Library Version 2020.0.4 Product Build 20200917 for Intel(R) 64 architecture applications\n - OpenMP 201511 (a.k.a. OpenMP 4.5)\n - LAPACK is enabled (usually provided by MKL)\n - NNPACK is enabled\n - CPU capability usage: AVX512\n - HIP Runtime 6.3.25521\n - MIOpen 2.18.0\n - Magma 2.8.0\n - Build settings: BLAS_INFO=mkl, BUILD_TYPE=Release, CXX_COMPILER=/opt/rh/gcc-toolset-10/root/usr/bin/c++, CXX_FLAGS= -D_GLIBCXX_USE_CXX11_ABI=1 -fvisibility-inlines-hidden -DUSE_PTHREADPOOL -DNDEBUG -DUSE_KINETO -DLIBKINETO_NOCUPTI -DLIBKINETO_NOXPUPTI=ON -DUSE_FBGEMM -DUSE_PYTORCH_QNNPACK -DUSE_XNNPACK -DSYMBOLICATE_MOBILE_DEBUG_HANDLE -O2 -fPIC -Wall -Wextra -Werror=return-type -Werror=non-virtual-dtor -Werror=bool-operation -Wnarrowing -Wno-missing-field-initializers -Wno-type-limits -Wno-array-bounds -Wno-unknown-pragmas -Wno-unused-parameter -Wno-strict-overflow -Wno-strict-aliasing -Wno-stringop-overflow -Wsuggest-override -Wno-psabi -Wno-error=old-style-cast -Wno-missing-braces -fdiagnostics-color=always -faligned-new -Wno-unused-but-set-variable -Wno-maybe-uninitialized -fno-math-errno -fno-trapping-math -Werror=format -Wno-stringop-overflow, FORCE_FALLBACK_CUDA_MPI=1, LAPACK_INFO=mkl, PERF_WITH_AVX=1, PERF_WITH_AVX2=1, TORCH_VERSION=2.5.1, USE_CUDA=0, USE_CUDNN=OFF, USE_CUSPARSELT=OFF, USE_EXCEPTION_PTR=1, USE_GFLAGS=1, USE_GLOG=1, USE_GLOO=1, USE_MKL=ON, USE_MKLDNN=0, USE_MPI=1, USE_NCCL=1, USE_NNPACK=ON, USE_OPENMP=1, USE_ROCM=ON, USE_ROCM_KERNEL_ASSERT=OFF, \n\nTorchVision: 0.20.1\nOpenCV: 4.12.0\nMMCV: 1.6.1\nMMCV Compiler: GCC 10.3\nMMCV CUDA Compiler: rocm not available\nMMDetection: 2.25.1+c41df4b", "config": "plugin = True\nplugin_dir = 'projects/mmdet3d_plugin/'\ndist_params = dict(backend='nccl')\nlog_level = 'INFO'\nwork_dir = './work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704'\ntotal_batch_size = 160\nnum_gpus = 8\nbatch_size = 20\nnum_iters_per_epoch = 175\nnum_epochs = 100\ncheckpoint_epoch_interval = 20\ncheckpoint_config = dict(interval=3500)\nlog_config = dict(\n interval=1,\n hooks=[\n dict(type='TextLoggerHook', by_epoch=False),\n dict(type='TensorboardLoggerHook')\n ])\nload_from = None\nresume_from = None\nworkflow = [('train', 1)]\nfp16 = dict(loss_scale=32.0)\ninput_shape = (704, 256)\ntracking_test = True\ntracking_threshold = 0.2\nclass_names = [\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier',\n 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n]\nnum_classes = 10\nembed_dims = 256\nnum_groups = 8\nnum_decoder = 6\nnum_single_frame_decoder = 1\nuse_deformable_func = True\nstrides = [4, 8, 16, 32]\nnum_levels = 4\nnum_depth_layers = 3\ndrop_out = 0.1\ntemporal = True\ndecouple_attn = True\nwith_quality_estimation = True\nmodel = dict(\n type='Sparse4D',\n use_grid_mask=True,\n use_deformable_func=True,\n img_backbone=dict(\n type='ResNet',\n depth=50,\n num_stages=4,\n frozen_stages=-1,\n norm_eval=False,\n style='pytorch',\n with_cp=True,\n out_indices=(0, 1, 2, 3),\n norm_cfg=dict(type='BN', requires_grad=True),\n pretrained='ckpt/resnet50-19c8e357.pth'),\n img_neck=dict(\n type='FPN',\n num_outs=4,\n start_level=0,\n out_channels=256,\n add_extra_convs='on_output',\n relu_before_extra_convs=True,\n in_channels=[256, 512, 1024, 2048]),\n depth_branch=dict(\n type='DenseDepthNet',\n embed_dims=256,\n num_depth_layers=3,\n loss_weight=0.2),\n head=dict(\n type='Sparse4DHead',\n cls_threshold_to_reg=0.05,\n decouple_attn=True,\n instance_bank=dict(\n type='InstanceBank',\n num_anchor=900,\n embed_dims=256,\n anchor='nuscenes_kmeans900.npy',\n anchor_handler=dict(type='SparseBox3DKeyPointsGenerator'),\n num_temp_instances=600,\n confidence_decay=0.6,\n feat_grad=False),\n anchor_encoder=dict(\n type='SparseBox3DEncoder',\n vel_dims=3,\n embed_dims=[128, 32, 32, 64],\n mode='cat',\n output_fc=False,\n in_loops=1,\n out_loops=4),\n num_single_frame_decoder=1,\n operation_order=[\n 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm',\n 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm',\n 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm',\n 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm',\n 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm',\n 'deformable', 'ffn', 'norm', 'refine'\n ],\n temp_graph_model=dict(\n type='MultiheadAttention',\n embed_dims=512,\n num_heads=8,\n batch_first=True,\n dropout=0.1),\n graph_model=dict(\n type='MultiheadAttention',\n embed_dims=512,\n num_heads=8,\n batch_first=True,\n dropout=0.1),\n norm_layer=dict(type='LN', normalized_shape=256),\n ffn=dict(\n type='AsymmetricFFN',\n in_channels=512,\n pre_norm=dict(type='LN'),\n embed_dims=256,\n feedforward_channels=1024,\n num_fcs=2,\n ffn_drop=0.1,\n act_cfg=dict(type='ReLU', inplace=True)),\n deformable_model=dict(\n type='DeformableFeatureAggregation',\n embed_dims=256,\n num_groups=8,\n num_levels=4,\n num_cams=6,\n attn_drop=0.15,\n use_deformable_func=True,\n use_camera_embed=True,\n residual_mode='cat',\n kps_generator=dict(\n type='SparseBox3DKeyPointsGenerator',\n num_learnable_pts=6,\n fix_scale=[[0, 0, 0], [0.45, 0, 0], [-0.45, 0, 0],\n [0, 0.45, 0], [0, -0.45, 0], [0, 0, 0.45],\n [0, 0, -0.45]])),\n refine_layer=dict(\n type='SparseBox3DRefinementModule',\n embed_dims=256,\n num_cls=10,\n refine_yaw=True,\n with_quality_estimation=True),\n sampler=dict(\n type='SparseBox3DTarget',\n num_dn_groups=5,\n num_temp_dn_groups=3,\n dn_noise_scale=[2.0, 2.0, 2.0, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5],\n max_dn_gt=32,\n add_neg_dn=True,\n cls_weight=2.0,\n box_weight=0.25,\n reg_weights=[2.0, 2.0, 2.0, 0.5, 0.5, 0.5, 0.0, 0.0, 0.0, 0.0],\n cls_wise_reg_weights=dict(\n {9: [2.0, 2.0, 2.0, 1.0, 1.0, 1.0, 0.0, 0.0, 1.0, 1.0]})),\n loss_cls=dict(\n type='FocalLoss',\n use_sigmoid=True,\n gamma=2.0,\n alpha=0.25,\n loss_weight=2.0),\n loss_reg=dict(\n type='SparseBox3DLoss',\n loss_box=dict(type='L1Loss', loss_weight=0.25),\n loss_centerness=dict(type='CrossEntropyLoss', use_sigmoid=True),\n loss_yawness=dict(type='GaussianFocalLoss'),\n cls_allow_reverse=[5]),\n decoder=dict(type='SparseBox3DDecoder'),\n reg_weights=[2.0, 2.0, 2.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0]))\ndataset_type = 'NuScenes3DDetTrackDataset'\ndata_root = 'data/nuscenes/'\nanno_root = 'data/nuscenes_anno_pkls/'\nfile_client_args = dict(backend='disk')\nimg_norm_cfg = dict(\n mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375], to_rgb=True)\ntrain_pipeline = [\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(\n type='LoadPointsFromFile',\n coord_type='LIDAR',\n load_dim=5,\n use_dim=5,\n file_client_args=dict(backend='disk')),\n dict(type='ResizeCropFlipImage'),\n dict(type='MultiScaleDepthMapGenerator', downsample=[4, 8, 16]),\n dict(type='BBoxRotation'),\n dict(type='PhotoMetricDistortionMultiViewImage'),\n dict(\n type='NormalizeMultiviewImage',\n mean=[123.675, 116.28, 103.53],\n std=[58.395, 57.12, 57.375],\n to_rgb=True),\n dict(\n type='CircleObjectRangeFilter',\n class_dist_thred=[55, 55, 55, 55, 55, 55, 55, 55, 55, 55]),\n dict(\n type='InstanceNameFilter',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ]),\n dict(type='NuScenesSparse4DAdaptor'),\n dict(\n type='Collect',\n keys=[\n 'img', 'timestamp', 'projection_mat', 'image_wh', 'gt_depth',\n 'focal', 'gt_bboxes_3d', 'gt_labels_3d'\n ],\n meta_keys=['T_global', 'T_global_inv', 'timestamp', 'instance_id'])\n]\ntest_pipeline = [\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(type='ResizeCropFlipImage'),\n dict(\n type='NormalizeMultiviewImage',\n mean=[123.675, 116.28, 103.53],\n std=[58.395, 57.12, 57.375],\n to_rgb=True),\n dict(type='NuScenesSparse4DAdaptor'),\n dict(\n type='Collect',\n keys=['img', 'timestamp', 'projection_mat', 'image_wh'],\n meta_keys=['T_global', 'T_global_inv', 'timestamp'])\n]\ninput_modality = dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False)\ndata_basic_config = dict(\n type='NuScenes3DDetTrackDataset',\n data_root='data/nuscenes/',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier',\n 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ],\n modality=dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False),\n version='v1.0-trainval')\ndata_aug_conf = dict(\n resize_lim=(0.4, 0.47),\n final_dim=(256, 704),\n bot_pct_lim=(0.0, 0.0),\n rot_lim=(-5.4, 5.4),\n H=900,\n W=1600,\n rand_flip=True,\n rot3d_range=[-0.3925, 0.3925])\ndata = dict(\n samples_per_gpu=20,\n workers_per_gpu=20,\n train=dict(\n type='NuScenes3DDetTrackDataset',\n data_root='data/nuscenes/',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ],\n modality=dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False),\n version='v1.0-trainval',\n ann_file='data/nuscenes_anno_pkls/nuscenes_infos_train.pkl',\n pipeline=[\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(\n type='LoadPointsFromFile',\n coord_type='LIDAR',\n load_dim=5,\n use_dim=5,\n file_client_args=dict(backend='disk')),\n dict(type='ResizeCropFlipImage'),\n dict(type='MultiScaleDepthMapGenerator', downsample=[4, 8, 16]),\n dict(type='BBoxRotation'),\n dict(type='PhotoMetricDistortionMultiViewImage'),\n dict(\n type='NormalizeMultiviewImage',\n mean=[123.675, 116.28, 103.53],\n std=[58.395, 57.12, 57.375],\n to_rgb=True),\n dict(\n type='CircleObjectRangeFilter',\n class_dist_thred=[55, 55, 55, 55, 55, 55, 55, 55, 55, 55]),\n dict(\n type='InstanceNameFilter',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian',\n 'traffic_cone'\n ]),\n dict(type='NuScenesSparse4DAdaptor'),\n dict(\n type='Collect',\n keys=[\n 'img', 'timestamp', 'projection_mat', 'image_wh',\n 'gt_depth', 'focal', 'gt_bboxes_3d', 'gt_labels_3d'\n ],\n meta_keys=[\n 'T_global', 'T_global_inv', 'timestamp', 'instance_id'\n ])\n ],\n test_mode=False,\n data_aug_conf=dict(\n resize_lim=(0.4, 0.47),\n final_dim=(256, 704),\n bot_pct_lim=(0.0, 0.0),\n rot_lim=(-5.4, 5.4),\n H=900,\n W=1600,\n rand_flip=True,\n rot3d_range=[-0.3925, 0.3925]),\n with_seq_flag=True,\n sequences_split_num=2,\n keep_consistent_seq_aug=True),\n val=dict(\n type='NuScenes3DDetTrackDataset',\n data_root='data/nuscenes/',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ],\n modality=dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False),\n version='v1.0-trainval',\n ann_file='data/nuscenes_anno_pkls/nuscenes_infos_val.pkl',\n pipeline=[\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(type='ResizeCropFlipImage'),\n dict(\n type='NormalizeMultiviewImage',\n mean=[123.675, 116.28, 103.53],\n std=[58.395, 57.12, 57.375],\n to_rgb=True),\n dict(type='NuScenesSparse4DAdaptor'),\n dict(\n type='Collect',\n keys=['img', 'timestamp', 'projection_mat', 'image_wh'],\n meta_keys=['T_global', 'T_global_inv', 'timestamp'])\n ],\n data_aug_conf=dict(\n resize_lim=(0.4, 0.47),\n final_dim=(256, 704),\n bot_pct_lim=(0.0, 0.0),\n rot_lim=(-5.4, 5.4),\n H=900,\n W=1600,\n rand_flip=True,\n rot3d_range=[-0.3925, 0.3925]),\n test_mode=True,\n tracking=True,\n tracking_threshold=0.2),\n test=dict(\n type='NuScenes3DDetTrackDataset',\n data_root='data/nuscenes/',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ],\n modality=dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False),\n version='v1.0-trainval',\n ann_file='data/nuscenes_anno_pkls/nuscenes_infos_val.pkl',\n pipeline=[\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(type='ResizeCropFlipImage'),\n dict(\n type='NormalizeMultiviewImage',\n mean=[123.675, 116.28, 103.53],\n std=[58.395, 57.12, 57.375],\n to_rgb=True),\n dict(type='NuScenesSparse4DAdaptor'),\n dict(\n type='Collect',\n keys=['img', 'timestamp', 'projection_mat', 'image_wh'],\n meta_keys=['T_global', 'T_global_inv', 'timestamp'])\n ],\n data_aug_conf=dict(\n resize_lim=(0.4, 0.47),\n final_dim=(256, 704),\n bot_pct_lim=(0.0, 0.0),\n rot_lim=(-5.4, 5.4),\n H=900,\n W=1600,\n rand_flip=True,\n rot3d_range=[-0.3925, 0.3925]),\n test_mode=True,\n tracking=True,\n tracking_threshold=0.2))\noptimizer = dict(\n type='AdamW',\n lr=0.0006,\n weight_decay=0.001,\n paramwise_cfg=dict(custom_keys=dict(img_backbone=dict(lr_mult=0.5))))\noptimizer_config = dict(grad_clip=dict(max_norm=25, norm_type=2))\nlr_config = dict(\n policy='CosineAnnealing',\n warmup='linear',\n warmup_iters=500,\n warmup_ratio=0.3333333333333333,\n min_lr_ratio=0.001)\nrunner = dict(type='IterBasedRunner', max_iters=17500)\nvis_pipeline = [\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(type='Collect', keys=['img'], meta_keys=['timestamp', 'lidar2img'])\n]\nevaluation = dict(\n interval=3500,\n pipeline=[\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(\n type='Collect', keys=['img'], meta_keys=['timestamp', 'lidar2img'])\n ])\ngpu_ids = range(0, 8)\n", "seed": 0, "exp_name": "sparse4dv3_temporal_r50_1x8_bs6_256x704.py"} diff --git a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20260114_212907.log b/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20260114_212907.log deleted file mode 100644 index e9b92fd971ba174b659e85bccf3aa7a5eeca87ed..0000000000000000000000000000000000000000 --- a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20260114_212907.log +++ /dev/null @@ -1,3729 +0,0 @@ -2026-01-14 21:29:07,380 - mmdet - INFO - Environment info: ------------------------------------------------------------- -sys.platform: linux -Python: 3.10.12 (main, Aug 15 2025, 14:32:43) [GCC 11.4.0] -CUDA available: True -GPU 0,1,2,3,4,5,6,7: BW151 -CUDA_HOME: /opt/dtk -NVCC: Not Available -GCC: x86_64-linux-gnu-gcc (Ubuntu 11.4.0-1ubuntu1~22.04) 11.4.0 -PyTorch: 2.5.1 -PyTorch compiling details: PyTorch built with: - - GCC 10.3 - - C++ Version: 201703 - - Intel(R) Math Kernel Library Version 2020.0.4 Product Build 20200917 for Intel(R) 64 architecture applications - - OpenMP 201511 (a.k.a. OpenMP 4.5) - - LAPACK is enabled (usually provided by MKL) - - NNPACK is enabled - - CPU capability usage: AVX512 - - HIP Runtime 6.3.25521 - - MIOpen 2.18.0 - - Magma 2.8.0 - - Build settings: BLAS_INFO=mkl, BUILD_TYPE=Release, CXX_COMPILER=/opt/rh/gcc-toolset-10/root/usr/bin/c++, CXX_FLAGS= -D_GLIBCXX_USE_CXX11_ABI=1 -fvisibility-inlines-hidden -DUSE_PTHREADPOOL -DNDEBUG -DUSE_KINETO -DLIBKINETO_NOCUPTI -DLIBKINETO_NOXPUPTI=ON -DUSE_FBGEMM -DUSE_PYTORCH_QNNPACK -DUSE_XNNPACK -DSYMBOLICATE_MOBILE_DEBUG_HANDLE -O2 -fPIC -Wall -Wextra -Werror=return-type -Werror=non-virtual-dtor -Werror=bool-operation -Wnarrowing -Wno-missing-field-initializers -Wno-type-limits -Wno-array-bounds -Wno-unknown-pragmas -Wno-unused-parameter -Wno-strict-overflow -Wno-strict-aliasing -Wno-stringop-overflow -Wsuggest-override -Wno-psabi -Wno-error=old-style-cast -Wno-missing-braces -fdiagnostics-color=always -faligned-new -Wno-unused-but-set-variable -Wno-maybe-uninitialized -fno-math-errno -fno-trapping-math -Werror=format -Wno-stringop-overflow, FORCE_FALLBACK_CUDA_MPI=1, LAPACK_INFO=mkl, PERF_WITH_AVX=1, PERF_WITH_AVX2=1, TORCH_VERSION=2.5.1, USE_CUDA=0, USE_CUDNN=OFF, USE_CUSPARSELT=OFF, USE_EXCEPTION_PTR=1, USE_GFLAGS=1, USE_GLOG=1, USE_GLOO=1, USE_MKL=ON, USE_MKLDNN=0, USE_MPI=1, USE_NCCL=1, USE_NNPACK=ON, USE_OPENMP=1, USE_ROCM=ON, USE_ROCM_KERNEL_ASSERT=OFF, - -TorchVision: 0.20.1 -OpenCV: 4.12.0 -MMCV: 1.6.1 -MMCV Compiler: GCC 10.3 -MMCV CUDA Compiler: rocm not available -MMDetection: 2.25.1+c41df4b ------------------------------------------------------------- - -2026-01-14 21:29:08,125 - mmdet - INFO - Distributed training: True -2026-01-14 21:29:09,031 - mmdet - INFO - Config: -plugin = True -plugin_dir = 'projects/mmdet3d_plugin/' -dist_params = dict(backend='nccl') -log_level = 'INFO' -work_dir = './work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704' -total_batch_size = 160 -num_gpus = 8 -batch_size = 20 -num_iters_per_epoch = 175 -num_epochs = 100 -checkpoint_epoch_interval = 20 -checkpoint_config = dict(interval=3500) -log_config = dict( - interval=1, - hooks=[ - dict(type='TextLoggerHook', by_epoch=False), - dict(type='TensorboardLoggerHook') - ]) -load_from = None -resume_from = None -workflow = [('train', 1)] -fp16 = dict(loss_scale=32.0) -input_shape = (704, 256) -tracking_test = True -tracking_threshold = 0.2 -class_names = [ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier', - 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' -] -num_classes = 10 -embed_dims = 256 -num_groups = 8 -num_decoder = 6 -num_single_frame_decoder = 1 -use_deformable_func = True -strides = [4, 8, 16, 32] -num_levels = 4 -num_depth_layers = 3 -drop_out = 0.1 -temporal = True -decouple_attn = True -with_quality_estimation = True -model = dict( - type='Sparse4D', - use_grid_mask=True, - use_deformable_func=True, - img_backbone=dict( - type='ResNet', - depth=50, - num_stages=4, - frozen_stages=-1, - norm_eval=False, - style='pytorch', - with_cp=True, - out_indices=(0, 1, 2, 3), - norm_cfg=dict(type='BN', requires_grad=True), - pretrained='ckpt/resnet50-19c8e357.pth'), - img_neck=dict( - type='FPN', - num_outs=4, - start_level=0, - out_channels=256, - add_extra_convs='on_output', - relu_before_extra_convs=True, - in_channels=[256, 512, 1024, 2048]), - depth_branch=dict( - type='DenseDepthNet', - embed_dims=256, - num_depth_layers=3, - loss_weight=0.2), - head=dict( - type='Sparse4DHead', - cls_threshold_to_reg=0.05, - decouple_attn=True, - instance_bank=dict( - type='InstanceBank', - num_anchor=900, - embed_dims=256, - anchor='nuscenes_kmeans900.npy', - anchor_handler=dict(type='SparseBox3DKeyPointsGenerator'), - num_temp_instances=600, - confidence_decay=0.6, - feat_grad=False), - anchor_encoder=dict( - type='SparseBox3DEncoder', - vel_dims=3, - embed_dims=[128, 32, 32, 64], - mode='cat', - output_fc=False, - in_loops=1, - out_loops=4), - num_single_frame_decoder=1, - operation_order=[ - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine' - ], - temp_graph_model=dict( - type='MultiheadAttention', - embed_dims=512, - num_heads=8, - batch_first=True, - dropout=0.1), - graph_model=dict( - type='MultiheadAttention', - embed_dims=512, - num_heads=8, - batch_first=True, - dropout=0.1), - norm_layer=dict(type='LN', normalized_shape=256), - ffn=dict( - type='AsymmetricFFN', - in_channels=512, - pre_norm=dict(type='LN'), - embed_dims=256, - feedforward_channels=1024, - num_fcs=2, - ffn_drop=0.1, - act_cfg=dict(type='ReLU', inplace=True)), - deformable_model=dict( - type='DeformableFeatureAggregation', - embed_dims=256, - num_groups=8, - num_levels=4, - num_cams=6, - attn_drop=0.15, - use_deformable_func=True, - use_camera_embed=True, - residual_mode='cat', - kps_generator=dict( - type='SparseBox3DKeyPointsGenerator', - num_learnable_pts=6, - fix_scale=[[0, 0, 0], [0.45, 0, 0], [-0.45, 0, 0], - [0, 0.45, 0], [0, -0.45, 0], [0, 0, 0.45], - [0, 0, -0.45]])), - refine_layer=dict( - type='SparseBox3DRefinementModule', - embed_dims=256, - num_cls=10, - refine_yaw=True, - with_quality_estimation=True), - sampler=dict( - type='SparseBox3DTarget', - num_dn_groups=5, - num_temp_dn_groups=3, - dn_noise_scale=[2.0, 2.0, 2.0, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5], - max_dn_gt=32, - add_neg_dn=True, - cls_weight=2.0, - box_weight=0.25, - reg_weights=[2.0, 2.0, 2.0, 0.5, 0.5, 0.5, 0.0, 0.0, 0.0, 0.0], - cls_wise_reg_weights=dict( - {9: [2.0, 2.0, 2.0, 1.0, 1.0, 1.0, 0.0, 0.0, 1.0, 1.0]})), - loss_cls=dict( - type='FocalLoss', - use_sigmoid=True, - gamma=2.0, - alpha=0.25, - loss_weight=2.0), - loss_reg=dict( - type='SparseBox3DLoss', - loss_box=dict(type='L1Loss', loss_weight=0.25), - loss_centerness=dict(type='CrossEntropyLoss', use_sigmoid=True), - loss_yawness=dict(type='GaussianFocalLoss'), - cls_allow_reverse=[5]), - decoder=dict(type='SparseBox3DDecoder'), - reg_weights=[2.0, 2.0, 2.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0])) -dataset_type = 'NuScenes3DDetTrackDataset' -data_root = 'data/nuscenes/' -anno_root = 'data/nuscenes_anno_pkls/' -file_client_args = dict(backend='disk') -img_norm_cfg = dict( - mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375], to_rgb=True) -train_pipeline = [ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict( - type='LoadPointsFromFile', - coord_type='LIDAR', - load_dim=5, - use_dim=5, - file_client_args=dict(backend='disk')), - dict(type='ResizeCropFlipImage'), - dict(type='MultiScaleDepthMapGenerator', downsample=[4, 8, 16]), - dict(type='BBoxRotation'), - dict(type='PhotoMetricDistortionMultiViewImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict( - type='CircleObjectRangeFilter', - class_dist_thred=[55, 55, 55, 55, 55, 55, 55, 55, 55, 55]), - dict( - type='InstanceNameFilter', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ]), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=[ - 'img', 'timestamp', 'projection_mat', 'image_wh', 'gt_depth', - 'focal', 'gt_bboxes_3d', 'gt_labels_3d' - ], - meta_keys=['T_global', 'T_global_inv', 'timestamp', 'instance_id']) -] -test_pipeline = [ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='ResizeCropFlipImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=['img', 'timestamp', 'projection_mat', 'image_wh'], - meta_keys=['T_global', 'T_global_inv', 'timestamp']) -] -input_modality = dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False) -data_basic_config = dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier', - 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval') -data_aug_conf = dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]) -data = dict( - samples_per_gpu=20, - workers_per_gpu=20, - train=dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval', - ann_file='data/nuscenes_anno_pkls/nuscenes_infos_train.pkl', - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict( - type='LoadPointsFromFile', - coord_type='LIDAR', - load_dim=5, - use_dim=5, - file_client_args=dict(backend='disk')), - dict(type='ResizeCropFlipImage'), - dict(type='MultiScaleDepthMapGenerator', downsample=[4, 8, 16]), - dict(type='BBoxRotation'), - dict(type='PhotoMetricDistortionMultiViewImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict( - type='CircleObjectRangeFilter', - class_dist_thred=[55, 55, 55, 55, 55, 55, 55, 55, 55, 55]), - dict( - type='InstanceNameFilter', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', - 'traffic_cone' - ]), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=[ - 'img', 'timestamp', 'projection_mat', 'image_wh', - 'gt_depth', 'focal', 'gt_bboxes_3d', 'gt_labels_3d' - ], - meta_keys=[ - 'T_global', 'T_global_inv', 'timestamp', 'instance_id' - ]) - ], - test_mode=False, - data_aug_conf=dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]), - with_seq_flag=True, - sequences_split_num=2, - keep_consistent_seq_aug=True), - val=dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval', - ann_file='data/nuscenes_anno_pkls/nuscenes_infos_val.pkl', - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='ResizeCropFlipImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=['img', 'timestamp', 'projection_mat', 'image_wh'], - meta_keys=['T_global', 'T_global_inv', 'timestamp']) - ], - data_aug_conf=dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]), - test_mode=True, - tracking=True, - tracking_threshold=0.2), - test=dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval', - ann_file='data/nuscenes_anno_pkls/nuscenes_infos_val.pkl', - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='ResizeCropFlipImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=['img', 'timestamp', 'projection_mat', 'image_wh'], - meta_keys=['T_global', 'T_global_inv', 'timestamp']) - ], - data_aug_conf=dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]), - test_mode=True, - tracking=True, - tracking_threshold=0.2)) -optimizer = dict( - type='AdamW', - lr=0.0006, - weight_decay=0.001, - paramwise_cfg=dict(custom_keys=dict(img_backbone=dict(lr_mult=0.5)))) -optimizer_config = dict(grad_clip=dict(max_norm=25, norm_type=2)) -lr_config = dict( - policy='CosineAnnealing', - warmup='linear', - warmup_iters=500, - warmup_ratio=0.3333333333333333, - min_lr_ratio=0.001) -runner = dict(type='IterBasedRunner', max_iters=17500) -vis_pipeline = [ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='Collect', keys=['img'], meta_keys=['timestamp', 'lidar2img']) -] -evaluation = dict( - interval=3500, - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict( - type='Collect', keys=['img'], meta_keys=['timestamp', 'lidar2img']) - ]) -gpu_ids = range(0, 8) - -2026-01-14 21:29:09,031 - mmdet - INFO - Set random seed to 0, deterministic: False -2026-01-14 21:29:09,361 - mmdet - INFO - initialize ResNet with init_cfg {'type': 'Pretrained', 'checkpoint': 'ckpt/resnet50-19c8e357.pth'} -2026-01-14 21:29:09,582 - mmdet - INFO - initialize FPN with init_cfg {'type': 'Xavier', 'layer': 'Conv2d', 'distribution': 'uniform'} -Name of parameter - Initialization information - -img_backbone.conv1.weight - torch.Size([64, 3, 7, 7]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.bn1.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.bn1.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.conv1.weight - torch.Size([64, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn1.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn1.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.conv2.weight - torch.Size([64, 64, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn2.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn2.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.conv3.weight - torch.Size([256, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn3.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn3.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.downsample.0.weight - torch.Size([256, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.downsample.1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.downsample.1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.conv1.weight - torch.Size([64, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn1.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn1.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.conv2.weight - torch.Size([64, 64, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn2.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn2.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.conv3.weight - torch.Size([256, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn3.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn3.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.conv1.weight - torch.Size([64, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn1.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn1.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.conv2.weight - torch.Size([64, 64, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn2.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn2.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.conv3.weight - torch.Size([256, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn3.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn3.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.conv1.weight - torch.Size([128, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn1.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn1.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.conv2.weight - torch.Size([128, 128, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn2.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn2.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.conv3.weight - torch.Size([512, 128, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn3.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn3.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.downsample.0.weight - torch.Size([512, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.downsample.1.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.downsample.1.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.conv1.weight - torch.Size([128, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn1.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn1.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.conv2.weight - torch.Size([128, 128, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn2.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn2.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.conv3.weight - torch.Size([512, 128, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn3.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn3.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.conv1.weight - torch.Size([128, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn1.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn1.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.conv2.weight - torch.Size([128, 128, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn2.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn2.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.conv3.weight - torch.Size([512, 128, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn3.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn3.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.conv1.weight - torch.Size([128, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn1.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn1.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.conv2.weight - torch.Size([128, 128, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn2.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn2.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.conv3.weight - torch.Size([512, 128, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn3.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn3.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.conv1.weight - torch.Size([256, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.downsample.0.weight - torch.Size([1024, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.downsample.1.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.downsample.1.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.conv1.weight - torch.Size([512, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn1.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn1.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.conv2.weight - torch.Size([512, 512, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn2.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn2.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.conv3.weight - torch.Size([2048, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn3.weight - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn3.bias - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.downsample.0.weight - torch.Size([2048, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.downsample.1.weight - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.downsample.1.bias - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.conv1.weight - torch.Size([512, 2048, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn1.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn1.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.conv2.weight - torch.Size([512, 512, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn2.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn2.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.conv3.weight - torch.Size([2048, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn3.weight - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn3.bias - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.conv1.weight - torch.Size([512, 2048, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn1.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn1.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.conv2.weight - torch.Size([512, 512, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn2.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn2.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.conv3.weight - torch.Size([2048, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn3.weight - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn3.bias - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_neck.lateral_convs.0.conv.weight - torch.Size([256, 256, 1, 1]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.lateral_convs.0.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.lateral_convs.1.conv.weight - torch.Size([256, 512, 1, 1]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.lateral_convs.1.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.lateral_convs.2.conv.weight - torch.Size([256, 1024, 1, 1]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.lateral_convs.2.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.lateral_convs.3.conv.weight - torch.Size([256, 2048, 1, 1]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.lateral_convs.3.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.fpn_convs.0.conv.weight - torch.Size([256, 256, 3, 3]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.fpn_convs.0.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.fpn_convs.1.conv.weight - torch.Size([256, 256, 3, 3]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.fpn_convs.1.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.fpn_convs.2.conv.weight - torch.Size([256, 256, 3, 3]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.fpn_convs.2.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.fpn_convs.3.conv.weight - torch.Size([256, 256, 3, 3]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.fpn_convs.3.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.instance_bank.anchor - torch.Size([900, 11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.instance_bank.instance_feature - torch.Size([900, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.instance_bank.anchor_handler.fix_scale - torch.Size([1, 3]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.0.weight - torch.Size([128, 3]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.0.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.2.weight - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.2.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.3.weight - torch.Size([128, 128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.3.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.5.weight - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.5.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.6.weight - torch.Size([128, 128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.6.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.8.weight - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.8.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.9.weight - torch.Size([128, 128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.9.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.11.weight - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.11.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.0.weight - torch.Size([32, 3]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.0.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.2.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.2.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.3.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.3.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.5.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.5.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.6.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.6.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.8.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.8.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.9.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.9.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.11.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.11.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.0.weight - torch.Size([32, 2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.0.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.2.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.2.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.3.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.3.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.5.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.5.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.6.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.6.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.8.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.8.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.9.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.9.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.11.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.11.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.0.weight - torch.Size([64, 3]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.0.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.2.weight - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.2.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.3.weight - torch.Size([64, 64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.3.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.5.weight - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.5.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.6.weight - torch.Size([64, 64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.6.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.8.weight - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.8.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.9.weight - torch.Size([64, 64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.9.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.11.weight - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.11.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.1.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.1.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.1.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.1.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.1.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.1.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.1.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.1.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.3.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.4.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.4.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.4.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.4.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.5.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.5.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.5.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.5.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.6.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.6.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.8.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.8.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.8.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.8.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.8.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.8.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.8.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.8.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.10.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.11.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.11.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.11.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.11.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.12.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.12.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.12.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.12.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.13.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.13.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.15.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.15.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.15.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.15.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.15.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.15.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.15.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.15.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.16.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.16.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.17.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.18.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.18.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.18.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.18.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.19.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.19.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.19.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.19.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.20.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.20.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.22.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.22.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.22.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.22.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.22.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.22.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.22.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.22.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.23.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.23.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.24.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.25.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.25.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.25.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.25.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.26.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.26.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.26.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.26.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.27.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.27.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.29.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.29.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.29.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.29.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.29.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.29.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.29.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.29.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.30.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.30.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.31.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.32.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.32.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.32.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.32.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.33.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.33.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.33.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.33.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.34.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.34.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.36.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.36.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.36.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.36.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.36.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.36.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.36.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.36.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.37.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.37.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.38.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.fc_before.weight - torch.Size([512, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.fc_after.weight - torch.Size([256, 512]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.0.weight - torch.Size([1, 256, 1, 1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.0.bias - torch.Size([1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.1.weight - torch.Size([1, 256, 1, 1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.1.bias - torch.Size([1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.2.weight - torch.Size([1, 256, 1, 1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.2.bias - torch.Size([1]): -The value is the same before and after calling `init_weights` of Sparse4D -2026-01-14 21:29:09,685 - mmdet - INFO - Model: -Sparse4D( - (img_backbone): ResNet( - (conv1): Conv2d(3, 64, kernel_size=(7, 7), stride=(2, 2), padding=(3, 3), bias=False) - (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (maxpool): MaxPool2d(kernel_size=3, stride=2, padding=1, dilation=1, ceil_mode=False) - (layer1): ResLayer( - (0): Bottleneck( - (conv1): Conv2d(64, 64, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (downsample): Sequential( - (0): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - ) - ) - (1): Bottleneck( - (conv1): Conv2d(256, 64, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (2): Bottleneck( - (conv1): Conv2d(256, 64, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - ) - (layer2): ResLayer( - (0): Bottleneck( - (conv1): Conv2d(256, 128, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (downsample): Sequential( - (0): Conv2d(256, 512, kernel_size=(1, 1), stride=(2, 2), bias=False) - (1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - ) - ) - (1): Bottleneck( - (conv1): Conv2d(512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (2): Bottleneck( - (conv1): Conv2d(512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (3): Bottleneck( - (conv1): Conv2d(512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - ) - (layer3): ResLayer( - (0): Bottleneck( - (conv1): Conv2d(512, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (downsample): Sequential( - (0): Conv2d(512, 1024, kernel_size=(1, 1), stride=(2, 2), bias=False) - (1): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - ) - ) - (1): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (2): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (3): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (4): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (5): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - ) - (layer4): ResLayer( - (0): Bottleneck( - (conv1): Conv2d(1024, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(512, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (downsample): Sequential( - (0): Conv2d(1024, 2048, kernel_size=(1, 1), stride=(2, 2), bias=False) - (1): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - ) - ) - (1): Bottleneck( - (conv1): Conv2d(2048, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(512, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (2): Bottleneck( - (conv1): Conv2d(2048, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(512, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - ) - ) - init_cfg={'type': 'Pretrained', 'checkpoint': 'ckpt/resnet50-19c8e357.pth'} - (img_neck): FPN( - (lateral_convs): ModuleList( - (0): ConvModule( - (conv): Conv2d(256, 256, kernel_size=(1, 1), stride=(1, 1)) - ) - (1): ConvModule( - (conv): Conv2d(512, 256, kernel_size=(1, 1), stride=(1, 1)) - ) - (2): ConvModule( - (conv): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1)) - ) - (3): ConvModule( - (conv): Conv2d(2048, 256, kernel_size=(1, 1), stride=(1, 1)) - ) - ) - (fpn_convs): ModuleList( - (0-3): 4 x ConvModule( - (conv): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1)) - ) - ) - ) - init_cfg={'type': 'Xavier', 'layer': 'Conv2d', 'distribution': 'uniform'} - (head): Sparse4DHead( - (instance_bank): InstanceBank( - (anchor_handler): SparseBox3DKeyPointsGenerator() - ) - (anchor_encoder): SparseBox3DEncoder( - (pos_fc): Sequential( - (0): Linear(in_features=3, out_features=128, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((128,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=128, out_features=128, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((128,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=128, out_features=128, bias=True) - (7): ReLU(inplace=True) - (8): LayerNorm((128,), eps=1e-05, elementwise_affine=True) - (9): Linear(in_features=128, out_features=128, bias=True) - (10): ReLU(inplace=True) - (11): LayerNorm((128,), eps=1e-05, elementwise_affine=True) - ) - (size_fc): Sequential( - (0): Linear(in_features=3, out_features=32, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=32, out_features=32, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=32, out_features=32, bias=True) - (7): ReLU(inplace=True) - (8): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (9): Linear(in_features=32, out_features=32, bias=True) - (10): ReLU(inplace=True) - (11): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - ) - (yaw_fc): Sequential( - (0): Linear(in_features=2, out_features=32, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=32, out_features=32, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=32, out_features=32, bias=True) - (7): ReLU(inplace=True) - (8): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (9): Linear(in_features=32, out_features=32, bias=True) - (10): ReLU(inplace=True) - (11): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - ) - (vel_fc): Sequential( - (0): Linear(in_features=3, out_features=64, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((64,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=64, out_features=64, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((64,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=64, out_features=64, bias=True) - (7): ReLU(inplace=True) - (8): LayerNorm((64,), eps=1e-05, elementwise_affine=True) - (9): Linear(in_features=64, out_features=64, bias=True) - (10): ReLU(inplace=True) - (11): LayerNorm((64,), eps=1e-05, elementwise_affine=True) - ) - ) - (loss_cls): FocalLoss() - (loss_reg): SparseBox3DLoss( - (loss_box): L1Loss() - (loss_cns): CrossEntropyLoss(avg_non_ignore=False) - (loss_yns): GaussianFocalLoss() - ) - (layers): ModuleList( - (0): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (1): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (4-5): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (6): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (7): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (8): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (11-12): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (13): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (14): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (15): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (16): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (17): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (18-19): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (20): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (21): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (22): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (23): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (24): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (25-26): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (27): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (28): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (29): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (30): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (31): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (32-33): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (34): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (35): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (36): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (37): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (38): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - ) - (fc_before): Linear(in_features=256, out_features=512, bias=False) - (fc_after): Linear(in_features=512, out_features=256, bias=False) - ) - (depth_branch): DenseDepthNet( - (depth_layers): ModuleList( - (0-2): 3 x Conv2d(256, 1, kernel_size=(1, 1), stride=(1, 1)) - ) - ) - (grid_mask): GridMask() -) -2026-01-14 21:29:20,805 - mmdet - INFO - Start running, host: root@bw150-1, work_dir: /workspace/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704 -2026-01-14 21:29:20,806 - mmdet - INFO - Hooks will be executed in the following order: -before_run: -(VERY_HIGH ) CosineAnnealingLrUpdaterHook -(ABOVE_NORMAL) Fp16OptimizerHook -(NORMAL ) CheckpointHook -(NORMAL ) CustomDistEvalHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -before_train_epoch: -(VERY_HIGH ) CosineAnnealingLrUpdaterHook -(NORMAL ) CustomDistEvalHook -(LOW ) IterTimerHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -before_train_iter: -(VERY_HIGH ) CosineAnnealingLrUpdaterHook -(NORMAL ) CustomDistEvalHook -(LOW ) IterTimerHook - -------------------- -after_train_iter: -(ABOVE_NORMAL) Fp16OptimizerHook -(NORMAL ) CheckpointHook -(NORMAL ) CustomDistEvalHook -(LOW ) IterTimerHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -after_train_epoch: -(NORMAL ) CheckpointHook -(NORMAL ) CustomDistEvalHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -before_val_epoch: -(LOW ) IterTimerHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -before_val_iter: -(LOW ) IterTimerHook - -------------------- -after_val_iter: -(LOW ) IterTimerHook - -------------------- -after_val_epoch: -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -after_run: -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -2026-01-14 21:29:20,806 - mmdet - INFO - workflow: [('train', 1)], max: 17500 iters -2026-01-14 21:29:20,808 - mmdet - INFO - Checkpoints will be saved to /workspace/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704 by HardDiskBackend. -2026-01-14 21:31:32,339 - mmdet - INFO - Iter [1/17500] lr: 1.000e-04, eta: 26 days, 9:22:47, time: 130.303, data_time: 17.201, memory: 49164, loss_cls_0: 2.3613, loss_box_0: 0.0138, loss_cns_0: 0.0027, loss_yns_0: 0.0008, loss_cls_1: 2.1544, loss_box_1: 0.1081, loss_cns_1: 0.0245, loss_yns_1: 0.0067, loss_cls_2: 2.3121, loss_box_2: 0.0050, loss_cns_2: 0.0006, loss_yns_2: 0.0003, loss_cls_3: 2.3902, loss_box_3: 0.0295, loss_cns_3: 0.0050, loss_yns_3: 0.0014, loss_cls_4: 2.0279, loss_box_4: 0.4149, loss_cns_4: 0.0532, loss_yns_4: 0.0250, loss_cls_5: 2.4248, loss_box_5: 0.0180, loss_cns_5: 0.0022, loss_yns_5: 0.0016, loss_cls_dn_0: 1.1980, loss_box_dn_0: 1.4603, loss_cls_dn_1: 1.1102, loss_box_dn_1: 1.7318, loss_cls_dn_2: 1.1741, loss_box_dn_2: 1.9719, loss_cls_dn_3: 1.1721, loss_box_dn_3: 2.2418, loss_cls_dn_4: 1.0528, loss_box_dn_4: 2.4268, loss_cls_dn_5: 1.2387, loss_box_dn_5: 2.6774, loss_dense_depth: 1.8643, loss: 35.7043, grad_norm: 269.8448 -2026-01-14 21:31:34,892 - mmdet - INFO - Iter [2/17500] lr: 1.004e-04, eta: 13 days, 10:52:15, time: 2.551, data_time: 0.157, memory: 49164, loss_cls_0: 2.0434, loss_box_0: 0.0137, loss_cns_0: 0.0038, loss_yns_0: 0.0015, loss_cls_1: 2.0095, loss_box_1: 0.1520, loss_cns_1: 0.0254, loss_yns_1: 0.0079, loss_cls_2: 2.0965, loss_box_2: 0.2799, loss_cns_2: 0.0263, loss_yns_2: 0.0108, loss_cls_3: 1.9406, loss_box_3: 0.3995, loss_cns_3: 0.0531, loss_yns_3: 0.0157, loss_cls_4: 1.8024, loss_box_4: 1.6764, loss_cns_4: 0.1705, loss_yns_4: 0.0606, loss_cls_5: 2.0554, loss_box_5: 0.5607, loss_cns_5: 0.0622, loss_yns_5: 0.0194, loss_cls_dn_0: 1.0256, loss_box_dn_0: 1.2558, loss_cls_dn_1: 0.9520, loss_box_dn_1: 2.4169, loss_cls_dn_2: 0.9672, loss_box_dn_2: 2.5330, loss_cls_dn_3: 0.9083, loss_box_dn_3: 2.6175, loss_cls_dn_4: 0.8406, loss_box_dn_4: 2.8744, loss_cls_dn_5: 0.9870, loss_box_dn_5: 3.1114, loss_dense_depth: 1.7092, loss: 37.6860, grad_norm: 66.9191 -2026-01-14 21:31:36,494 - mmdet - INFO - Iter [3/17500] lr: 1.008e-04, eta: 9 days, 1:46:18, time: 1.566, data_time: 0.074, memory: 49164, loss_cls_0: 1.4738, loss_box_0: 2.6170, loss_cns_0: 0.6129, loss_yns_0: 0.2247, loss_cls_1: 1.7629, loss_box_1: 2.0263, loss_cns_1: 0.3107, loss_yns_1: 0.1152, loss_cls_2: 1.7844, loss_box_2: 4.0564, loss_cns_2: 0.3559, loss_yns_2: 0.1958, loss_cls_3: 1.6199, loss_box_3: 4.8150, loss_cns_3: 0.4399, loss_yns_3: 0.2040, loss_cls_4: 1.5776, loss_box_4: 4.0089, loss_cns_4: 0.3702, loss_yns_4: 0.1578, loss_cls_5: 1.6956, loss_box_5: 2.7738, loss_cns_5: 0.2160, loss_yns_5: 0.0920, loss_cls_dn_0: 0.7142, loss_box_dn_0: 1.2281, loss_cls_dn_1: 0.8291, loss_box_dn_1: 2.4144, loss_cls_dn_2: 0.8039, loss_box_dn_2: 2.6250, loss_cls_dn_3: 0.7124, loss_box_dn_3: 2.8220, loss_cls_dn_4: 0.7262, loss_box_dn_4: 3.0874, loss_cls_dn_5: 0.8089, loss_box_dn_5: 3.3226, loss_dense_depth: 1.6839, loss: 55.2847, grad_norm: 102.5957 -2026-01-14 21:31:38,110 - mmdet - INFO - Iter [4/17500] lr: 1.012e-04, eta: 6 days, 21:19:43, time: 1.654, data_time: 0.096, memory: 49164, loss_cls_0: 1.3805, loss_box_0: 2.4607, loss_cns_0: 0.5760, loss_yns_0: 0.1829, loss_cls_1: 1.5947, loss_box_1: 3.0691, loss_cns_1: 0.4490, loss_yns_1: 0.1804, loss_cls_2: 1.7143, loss_box_2: 3.7528, loss_cns_2: 0.4341, loss_yns_2: 0.1900, loss_cls_3: 1.5198, loss_box_3: 4.3099, loss_cns_3: 0.4388, loss_yns_3: 0.2114, loss_cls_4: 1.4713, loss_box_4: 4.8596, loss_cns_4: 0.3549, loss_yns_4: 0.1987, loss_cls_5: 1.5028, loss_box_5: 4.9886, loss_cns_5: 0.4388, loss_yns_5: 0.1901, loss_cls_dn_0: 0.5797, loss_box_dn_0: 1.1674, loss_cls_dn_1: 0.7201, loss_box_dn_1: 2.5675, loss_cls_dn_2: 0.6949, loss_box_dn_2: 2.6395, loss_cls_dn_3: 0.6160, loss_box_dn_3: 2.8463, loss_cls_dn_4: 0.5992, loss_box_dn_4: 3.0473, loss_cls_dn_5: 0.6737, loss_box_dn_5: 3.2112, loss_dense_depth: 1.6088, loss: 57.4410, grad_norm: 119.5187 -2026-01-14 21:31:39,854 - mmdet - INFO - Iter [5/17500] lr: 1.016e-04, eta: 5 days, 13:56:10, time: 1.729, data_time: 0.072, memory: 49164, loss_cls_0: 1.3459, loss_box_0: 2.7535, loss_cns_0: 0.5052, loss_yns_0: 0.1956, loss_cls_1: 1.5438, loss_box_1: 3.8496, loss_cns_1: 0.4023, loss_yns_1: 0.2026, loss_cls_2: 1.6107, loss_box_2: 3.9277, loss_cns_2: 0.3868, loss_yns_2: 0.1949, loss_cls_3: 1.4622, loss_box_3: 4.0055, loss_cns_3: 0.4000, loss_yns_3: 0.1941, loss_cls_4: 1.3978, loss_box_4: 4.1975, loss_cns_4: 0.3837, loss_yns_4: 0.1941, loss_cls_5: 1.3932, loss_box_5: 4.4300, loss_cns_5: 0.4093, loss_yns_5: 0.1989, loss_cls_dn_0: 0.5304, loss_box_dn_0: 1.2164, loss_cls_dn_1: 0.6522, loss_box_dn_1: 2.1946, loss_cls_dn_2: 0.6518, loss_box_dn_2: 2.2967, loss_cls_dn_3: 0.5613, loss_box_dn_3: 2.4180, loss_cls_dn_4: 0.5564, loss_box_dn_4: 2.5781, loss_cls_dn_5: 0.5867, loss_box_dn_5: 2.6584, loss_dense_depth: 1.5811, loss: 54.0670, grad_norm: 109.1646 -2026-01-14 21:31:41,442 - mmdet - INFO - Iter [6/17500] lr: 1.020e-04, eta: 4 days, 16:52:46, time: 1.571, data_time: 0.090, memory: 49164, loss_cls_0: 1.3014, loss_box_0: 2.5530, loss_cns_0: 0.5628, loss_yns_0: 0.1811, loss_cls_1: 1.4716, loss_box_1: 3.8208, loss_cns_1: 0.3747, loss_yns_1: 0.1906, loss_cls_2: 1.4804, loss_box_2: 4.0100, loss_cns_2: 0.3508, loss_yns_2: 0.1900, loss_cls_3: 1.3515, loss_box_3: 4.0537, loss_cns_3: 0.3322, loss_yns_3: 0.1937, loss_cls_4: 1.3223, loss_box_4: 4.3573, loss_cns_4: 0.2843, loss_yns_4: 0.1952, loss_cls_5: 1.3349, loss_box_5: 4.5102, loss_cns_5: 0.2830, loss_yns_5: 0.1987, loss_cls_dn_0: 0.5330, loss_box_dn_0: 1.1604, loss_cls_dn_1: 0.5864, loss_box_dn_1: 2.3485, loss_cls_dn_2: 0.5843, loss_box_dn_2: 2.4036, loss_cls_dn_3: 0.5176, loss_box_dn_3: 2.4588, loss_cls_dn_4: 0.4923, loss_box_dn_4: 2.6789, loss_cls_dn_5: 0.4914, loss_box_dn_5: 2.7404, loss_dense_depth: 1.4970, loss: 53.3966, grad_norm: 115.9684 -2026-01-14 21:31:43,068 - mmdet - INFO - Iter [7/17500] lr: 1.024e-04, eta: 4 days, 1:52:56, time: 1.634, data_time: 0.108, memory: 49164, loss_cls_0: 1.2787, loss_box_0: 2.3793, loss_cns_0: 0.6467, loss_yns_0: 0.1772, loss_cls_1: 1.3528, loss_box_1: 3.5613, loss_cns_1: 0.4446, loss_yns_1: 0.1861, loss_cls_2: 1.3881, loss_box_2: 3.6843, loss_cns_2: 0.4398, loss_yns_2: 0.1812, loss_cls_3: 1.2989, loss_box_3: 3.5803, loss_cns_3: 0.4559, loss_yns_3: 0.1827, loss_cls_4: 1.2902, loss_box_4: 3.9173, loss_cns_4: 0.4022, loss_yns_4: 0.1868, loss_cls_5: 1.3259, loss_box_5: 4.2060, loss_cns_5: 0.3741, loss_yns_5: 0.1906, loss_cls_dn_0: 0.5431, loss_box_dn_0: 1.1006, loss_cls_dn_1: 0.5241, loss_box_dn_1: 2.3997, loss_cls_dn_2: 0.5252, loss_box_dn_2: 2.4008, loss_cls_dn_3: 0.4761, loss_box_dn_3: 2.4197, loss_cls_dn_4: 0.4512, loss_box_dn_4: 2.6279, loss_cls_dn_5: 0.4360, loss_box_dn_5: 2.7304, loss_dense_depth: 1.4353, loss: 51.2009, grad_norm: 105.6673 -2026-01-14 21:31:44,610 - mmdet - INFO - Iter [8/17500] lr: 1.028e-04, eta: 3 days, 14:35:38, time: 1.567, data_time: 0.088, memory: 49164, loss_cls_0: 1.2394, loss_box_0: 2.2865, loss_cns_0: 0.6458, loss_yns_0: 0.1811, loss_cls_1: 1.2702, loss_box_1: 3.5117, loss_cns_1: 0.4862, loss_yns_1: 0.1849, loss_cls_2: 1.3568, loss_box_2: 3.6242, loss_cns_2: 0.4601, loss_yns_2: 0.1837, loss_cls_3: 1.2898, loss_box_3: 3.5973, loss_cns_3: 0.4856, loss_yns_3: 0.1853, loss_cls_4: 1.2967, loss_box_4: 3.6355, loss_cns_4: 0.4791, loss_yns_4: 0.1840, loss_cls_5: 1.3312, loss_box_5: 3.7179, loss_cns_5: 0.4927, loss_yns_5: 0.1940, loss_cls_dn_0: 0.5305, loss_box_dn_0: 1.0383, loss_cls_dn_1: 0.5337, loss_box_dn_1: 1.8193, loss_cls_dn_2: 0.5395, loss_box_dn_2: 1.8051, loss_cls_dn_3: 0.4935, loss_box_dn_3: 1.8005, loss_cls_dn_4: 0.4635, loss_box_dn_4: 1.9139, loss_cls_dn_5: 0.4366, loss_box_dn_5: 1.9466, loss_dense_depth: 1.4736, loss: 47.1144, grad_norm: 85.4461 -2026-01-14 21:31:46,185 - mmdet - INFO - Iter [9/17500] lr: 1.032e-04, eta: 3 days, 5:49:03, time: 1.574, data_time: 0.075, memory: 49164, loss_cls_0: 1.2450, loss_box_0: 2.2252, loss_cns_0: 0.6184, loss_yns_0: 0.1778, loss_cls_1: 1.2557, loss_box_1: 3.2779, loss_cns_1: 0.5134, loss_yns_1: 0.1823, loss_cls_2: 1.3495, loss_box_2: 3.3757, loss_cns_2: 0.4688, loss_yns_2: 0.1824, loss_cls_3: 1.2671, loss_box_3: 3.4249, loss_cns_3: 0.4890, loss_yns_3: 0.1889, loss_cls_4: 1.2723, loss_box_4: 3.3837, loss_cns_4: 0.4913, loss_yns_4: 0.1917, loss_cls_5: 1.3184, loss_box_5: 3.4723, loss_cns_5: 0.5095, loss_yns_5: 0.1897, loss_cls_dn_0: 0.4948, loss_box_dn_0: 1.0199, loss_cls_dn_1: 0.4942, loss_box_dn_1: 1.5049, loss_cls_dn_2: 0.5175, loss_box_dn_2: 1.5232, loss_cls_dn_3: 0.4702, loss_box_dn_3: 1.5701, loss_cls_dn_4: 0.4429, loss_box_dn_4: 1.5546, loss_cls_dn_5: 0.4236, loss_box_dn_5: 1.6427, loss_dense_depth: 1.3543, loss: 44.0840, grad_norm: 60.7525 -2026-01-14 21:31:47,795 - mmdet - INFO - Iter [10/17500] lr: 1.036e-04, eta: 2 days, 22:48:52, time: 1.611, data_time: 0.077, memory: 49164, loss_cls_0: 1.2230, loss_box_0: 2.1935, loss_cns_0: 0.6145, loss_yns_0: 0.1727, loss_cls_1: 1.2580, loss_box_1: 3.1051, loss_cns_1: 0.5368, loss_yns_1: 0.1758, loss_cls_2: 1.2772, loss_box_2: 3.1722, loss_cns_2: 0.5286, loss_yns_2: 0.1784, loss_cls_3: 1.2574, loss_box_3: 3.3351, loss_cns_3: 0.5439, loss_yns_3: 0.1874, loss_cls_4: 1.2526, loss_box_4: 3.2750, loss_cns_4: 0.5593, loss_yns_4: 0.1854, loss_cls_5: 1.2725, loss_box_5: 3.4016, loss_cns_5: 0.5715, loss_yns_5: 0.1817, loss_cls_dn_0: 0.4769, loss_box_dn_0: 1.0215, loss_cls_dn_1: 0.4526, loss_box_dn_1: 1.6170, loss_cls_dn_2: 0.4752, loss_box_dn_2: 1.6690, loss_cls_dn_3: 0.4349, loss_box_dn_3: 1.7659, loss_cls_dn_4: 0.4237, loss_box_dn_4: 1.7685, loss_cls_dn_5: 0.4288, loss_box_dn_5: 1.9183, loss_dense_depth: 1.3866, loss: 44.2982, grad_norm: 74.1258 -2026-01-14 21:31:49,360 - mmdet - INFO - Iter [11/17500] lr: 1.040e-04, eta: 2 days, 17:03:49, time: 1.564, data_time: 0.075, memory: 49164, loss_cls_0: 1.2206, loss_box_0: 2.2267, loss_cns_0: 0.6231, loss_yns_0: 0.1757, loss_cls_1: 1.2541, loss_box_1: 3.0219, loss_cns_1: 0.5245, loss_yns_1: 0.1778, loss_cls_2: 1.2724, loss_box_2: 2.9821, loss_cns_2: 0.5223, loss_yns_2: 0.1842, loss_cls_3: 1.2568, loss_box_3: 3.0471, loss_cns_3: 0.5772, loss_yns_3: 0.1794, loss_cls_4: 1.2398, loss_box_4: 2.9787, loss_cns_4: 0.5843, loss_yns_4: 0.1801, loss_cls_5: 1.2650, loss_box_5: 3.2118, loss_cns_5: 0.5630, loss_yns_5: 0.1846, loss_cls_dn_0: 0.4729, loss_box_dn_0: 1.0310, loss_cls_dn_1: 0.4213, loss_box_dn_1: 1.9276, loss_cls_dn_2: 0.4396, loss_box_dn_2: 1.9554, loss_cls_dn_3: 0.4137, loss_box_dn_3: 2.0303, loss_cls_dn_4: 0.4049, loss_box_dn_4: 2.0385, loss_cls_dn_5: 0.4182, loss_box_dn_5: 2.2392, loss_dense_depth: 1.2493, loss: 44.4950, grad_norm: 68.6185 -2026-01-14 21:31:50,945 - mmdet - INFO - Iter [12/17500] lr: 1.044e-04, eta: 2 days, 12:16:48, time: 1.585, data_time: 0.078, memory: 49164, loss_cls_0: 1.2147, loss_box_0: 2.2313, loss_cns_0: 0.6261, loss_yns_0: 0.1745, loss_cls_1: 1.2673, loss_box_1: 2.9318, loss_cns_1: 0.5035, loss_yns_1: 0.1741, loss_cls_2: 1.2920, loss_box_2: 2.9054, loss_cns_2: 0.5052, loss_yns_2: 0.1811, loss_cls_3: 1.2454, loss_box_3: 2.8804, loss_cns_3: 0.5407, loss_yns_3: 0.1781, loss_cls_4: 1.2460, loss_box_4: 2.9252, loss_cns_4: 0.5742, loss_yns_4: 0.1772, loss_cls_5: 1.2739, loss_box_5: 3.1744, loss_cns_5: 0.5499, loss_yns_5: 0.1859, loss_cls_dn_0: 0.4722, loss_box_dn_0: 1.0414, loss_cls_dn_1: 0.3900, loss_box_dn_1: 2.2405, loss_cls_dn_2: 0.4147, loss_box_dn_2: 2.2243, loss_cls_dn_3: 0.3989, loss_box_dn_3: 2.2510, loss_cls_dn_4: 0.3844, loss_box_dn_4: 2.2824, loss_cls_dn_5: 0.3977, loss_box_dn_5: 2.4183, loss_dense_depth: 1.3262, loss: 45.2007, grad_norm: 68.7727 -2026-01-14 21:31:52,522 - mmdet - INFO - Iter [13/17500] lr: 1.048e-04, eta: 2 days, 8:13:45, time: 1.577, data_time: 0.073, memory: 49164, loss_cls_0: 1.1994, loss_box_0: 2.2506, loss_cns_0: 0.6087, loss_yns_0: 0.1712, loss_cls_1: 1.2589, loss_box_1: 2.8059, loss_cns_1: 0.5313, loss_yns_1: 0.1738, loss_cls_2: 1.2750, loss_box_2: 2.8779, loss_cns_2: 0.5354, loss_yns_2: 0.1821, loss_cls_3: 1.2379, loss_box_3: 2.8407, loss_cns_3: 0.5413, loss_yns_3: 0.1777, loss_cls_4: 1.2639, loss_box_4: 2.8735, loss_cns_4: 0.5450, loss_yns_4: 0.1770, loss_cls_5: 1.2780, loss_box_5: 2.9383, loss_cns_5: 0.5362, loss_yns_5: 0.1804, loss_cls_dn_0: 0.4621, loss_box_dn_0: 1.0461, loss_cls_dn_1: 0.4212, loss_box_dn_1: 1.6229, loss_cls_dn_2: 0.4491, loss_box_dn_2: 1.6271, loss_cls_dn_3: 0.4485, loss_box_dn_3: 1.6788, loss_cls_dn_4: 0.4304, loss_box_dn_4: 1.7873, loss_cls_dn_5: 0.4449, loss_box_dn_5: 1.7949, loss_dense_depth: 1.1948, loss: 41.8683, grad_norm: 60.0293 -2026-01-14 21:31:54,107 - mmdet - INFO - Iter [14/17500] lr: 1.052e-04, eta: 2 days, 4:45:35, time: 1.585, data_time: 0.073, memory: 49164, loss_cls_0: 1.2196, loss_box_0: 2.2518, loss_cns_0: 0.5939, loss_yns_0: 0.1720, loss_cls_1: 1.2733, loss_box_1: 2.7187, loss_cns_1: 0.5600, loss_yns_1: 0.1720, loss_cls_2: 1.2693, loss_box_2: 2.8299, loss_cns_2: 0.5601, loss_yns_2: 0.1842, loss_cls_3: 1.2660, loss_box_3: 2.8802, loss_cns_3: 0.5578, loss_yns_3: 0.1730, loss_cls_4: 1.2889, loss_box_4: 2.8898, loss_cns_4: 0.5413, loss_yns_4: 0.1741, loss_cls_5: 1.2860, loss_box_5: 2.8781, loss_cns_5: 0.5531, loss_yns_5: 0.1784, loss_cls_dn_0: 0.4549, loss_box_dn_0: 1.0333, loss_cls_dn_1: 0.4436, loss_box_dn_1: 1.2540, loss_cls_dn_2: 0.4712, loss_box_dn_2: 1.2770, loss_cls_dn_3: 0.4625, loss_box_dn_3: 1.3672, loss_cls_dn_4: 0.4518, loss_box_dn_4: 1.4869, loss_cls_dn_5: 0.4686, loss_box_dn_5: 1.4336, loss_dense_depth: 1.1977, loss: 40.2736, grad_norm: 78.8375 -2026-01-14 21:31:55,712 - mmdet - INFO - Iter [15/17500] lr: 1.056e-04, eta: 2 days, 1:44:58, time: 1.574, data_time: 0.075, memory: 49164, loss_cls_0: 1.2198, loss_box_0: 2.2881, loss_cns_0: 0.5869, loss_yns_0: 0.1722, loss_cls_1: 1.3002, loss_box_1: 2.5592, loss_cns_1: 0.5783, loss_yns_1: 0.1724, loss_cls_2: 1.2633, loss_box_2: 2.5971, loss_cns_2: 0.5826, loss_yns_2: 0.1795, loss_cls_3: 1.2938, loss_box_3: 2.6801, loss_cns_3: 0.5922, loss_yns_3: 0.1744, loss_cls_4: 1.2922, loss_box_4: 2.7479, loss_cns_4: 0.5810, loss_yns_4: 0.1805, loss_cls_5: 1.2839, loss_box_5: 2.8429, loss_cns_5: 0.5883, loss_yns_5: 0.1808, loss_cls_dn_0: 0.4546, loss_box_dn_0: 1.0278, loss_cls_dn_1: 0.4498, loss_box_dn_1: 1.3121, loss_cls_dn_2: 0.4630, loss_box_dn_2: 1.3109, loss_cls_dn_3: 0.4407, loss_box_dn_3: 1.3910, loss_cls_dn_4: 0.4463, loss_box_dn_4: 1.4890, loss_cls_dn_5: 0.4674, loss_box_dn_5: 1.4966, loss_dense_depth: 1.2037, loss: 39.8909, grad_norm: 79.1593 -2026-01-14 21:31:57,282 - mmdet - INFO - Iter [16/17500] lr: 1.060e-04, eta: 1 day, 23:07:23, time: 1.600, data_time: 0.111, memory: 49164, loss_cls_0: 1.1952, loss_box_0: 2.3048, loss_cns_0: 0.5841, loss_yns_0: 0.1740, loss_cls_1: 1.2984, loss_box_1: 2.6386, loss_cns_1: 0.5664, loss_yns_1: 0.1731, loss_cls_2: 1.2709, loss_box_2: 2.6762, loss_cns_2: 0.5675, loss_yns_2: 0.1798, loss_cls_3: 1.3093, loss_box_3: 2.6666, loss_cns_3: 0.5840, loss_yns_3: 0.1774, loss_cls_4: 1.2590, loss_box_4: 2.7403, loss_cns_4: 0.5802, loss_yns_4: 0.1805, loss_cls_5: 1.2683, loss_box_5: 2.8509, loss_cns_5: 0.5983, loss_yns_5: 0.1771, loss_cls_dn_0: 0.4775, loss_box_dn_0: 1.0110, loss_cls_dn_1: 0.4651, loss_box_dn_1: 1.3091, loss_cls_dn_2: 0.4706, loss_box_dn_2: 1.2832, loss_cls_dn_3: 0.4349, loss_box_dn_3: 1.3375, loss_cls_dn_4: 0.4583, loss_box_dn_4: 1.3993, loss_cls_dn_5: 0.4768, loss_box_dn_5: 1.4410, loss_dense_depth: 1.1534, loss: 39.7387, grad_norm: 69.1777 -2026-01-14 21:31:58,966 - mmdet - INFO - Iter [17/17500] lr: 1.064e-04, eta: 1 day, 20:48:57, time: 1.636, data_time: 0.074, memory: 49164, loss_cls_0: 1.1752, loss_box_0: 2.2955, loss_cns_0: 0.5940, loss_yns_0: 0.1713, loss_cls_1: 1.2752, loss_box_1: 2.8497, loss_cns_1: 0.5493, loss_yns_1: 0.1760, loss_cls_2: 1.2688, loss_box_2: 2.9123, loss_cns_2: 0.5470, loss_yns_2: 0.1747, loss_cls_3: 1.2924, loss_box_3: 2.8936, loss_cns_3: 0.5664, loss_yns_3: 0.1756, loss_cls_4: 1.2504, loss_box_4: 2.9000, loss_cns_4: 0.5746, loss_yns_4: 0.1774, loss_cls_5: 1.2671, loss_box_5: 2.8510, loss_cns_5: 0.5913, loss_yns_5: 0.1741, loss_cls_dn_0: 0.4818, loss_box_dn_0: 1.0088, loss_cls_dn_1: 0.4579, loss_box_dn_1: 1.4078, loss_cls_dn_2: 0.4554, loss_box_dn_2: 1.4043, loss_cls_dn_3: 0.4212, loss_box_dn_3: 1.4465, loss_cls_dn_4: 0.4436, loss_box_dn_4: 1.4730, loss_cls_dn_5: 0.4574, loss_box_dn_5: 1.5048, loss_dense_depth: 1.1667, loss: 40.8319, grad_norm: 67.3141 -2026-01-14 21:32:00,536 - mmdet - INFO - Iter [18/17500] lr: 1.068e-04, eta: 1 day, 18:45:36, time: 1.617, data_time: 0.107, memory: 49164, loss_cls_0: 1.1750, loss_box_0: 2.2402, loss_cns_0: 0.6145, loss_yns_0: 0.1710, loss_cls_1: 1.2495, loss_box_1: 2.8498, loss_cns_1: 0.5500, loss_yns_1: 0.1802, loss_cls_2: 1.2606, loss_box_2: 2.8987, loss_cns_2: 0.5515, loss_yns_2: 0.1741, loss_cls_3: 1.2555, loss_box_3: 2.9022, loss_cns_3: 0.5578, loss_yns_3: 0.1751, loss_cls_4: 1.2489, loss_box_4: 2.9032, loss_cns_4: 0.5749, loss_yns_4: 0.1724, loss_cls_5: 1.2631, loss_box_5: 2.9313, loss_cns_5: 0.5688, loss_yns_5: 0.1752, loss_cls_dn_0: 0.4788, loss_box_dn_0: 1.0008, loss_cls_dn_1: 0.4487, loss_box_dn_1: 1.4158, loss_cls_dn_2: 0.4478, loss_box_dn_2: 1.4579, loss_cls_dn_3: 0.4235, loss_box_dn_3: 1.5123, loss_cls_dn_4: 0.4336, loss_box_dn_4: 1.5200, loss_cls_dn_5: 0.4488, loss_box_dn_5: 1.6226, loss_dense_depth: 1.0556, loss: 40.9100, grad_norm: 69.5493 -2026-01-14 21:32:02,119 - mmdet - INFO - Iter [19/17500] lr: 1.072e-04, eta: 1 day, 16:54:42, time: 1.584, data_time: 0.077, memory: 49164, loss_cls_0: 1.1868, loss_box_0: 2.2478, loss_cns_0: 0.6113, loss_yns_0: 0.1715, loss_cls_1: 1.2400, loss_box_1: 2.8318, loss_cns_1: 0.5556, loss_yns_1: 0.1769, loss_cls_2: 1.2610, loss_box_2: 2.8624, loss_cns_2: 0.5512, loss_yns_2: 0.1767, loss_cls_3: 1.2510, loss_box_3: 2.8680, loss_cns_3: 0.5458, loss_yns_3: 0.1737, loss_cls_4: 1.2616, loss_box_4: 2.8639, loss_cns_4: 0.5732, loss_yns_4: 0.1737, loss_cls_5: 1.2650, loss_box_5: 2.9286, loss_cns_5: 0.5466, loss_yns_5: 0.1749, loss_cls_dn_0: 0.4772, loss_box_dn_0: 1.0175, loss_cls_dn_1: 0.4583, loss_box_dn_1: 1.2667, loss_cls_dn_2: 0.4580, loss_box_dn_2: 1.3743, loss_cls_dn_3: 0.4435, loss_box_dn_3: 1.4629, loss_cls_dn_4: 0.4371, loss_box_dn_4: 1.4698, loss_cls_dn_5: 0.4558, loss_box_dn_5: 1.6229, loss_dense_depth: 1.1386, loss: 40.5818, grad_norm: 79.3254 -2026-01-14 21:32:03,720 - mmdet - INFO - Iter [20/17500] lr: 1.076e-04, eta: 1 day, 15:14:43, time: 1.570, data_time: 0.077, memory: 49164, loss_cls_0: 1.1909, loss_box_0: 2.2219, loss_cns_0: 0.6105, loss_yns_0: 0.1707, loss_cls_1: 1.2470, loss_box_1: 2.8400, loss_cns_1: 0.5596, loss_yns_1: 0.1765, loss_cls_2: 1.2552, loss_box_2: 2.8255, loss_cns_2: 0.5596, loss_yns_2: 0.1772, loss_cls_3: 1.2413, loss_box_3: 2.8197, loss_cns_3: 0.5537, loss_yns_3: 0.1723, loss_cls_4: 1.2598, loss_box_4: 2.8132, loss_cns_4: 0.5720, loss_yns_4: 0.1740, loss_cls_5: 1.2667, loss_box_5: 2.8605, loss_cns_5: 0.5513, loss_yns_5: 0.1755, loss_cls_dn_0: 0.4668, loss_box_dn_0: 0.9976, loss_cls_dn_1: 0.4207, loss_box_dn_1: 1.4990, loss_cls_dn_2: 0.4219, loss_box_dn_2: 1.5606, loss_cls_dn_3: 0.4163, loss_box_dn_3: 1.5959, loss_cls_dn_4: 0.4099, loss_box_dn_4: 1.5770, loss_cls_dn_5: 0.4177, loss_box_dn_5: 1.6745, loss_dense_depth: 1.0637, loss: 40.8161, grad_norm: 58.7348 -2026-01-14 21:32:05,436 - mmdet - INFO - Iter [21/17500] lr: 1.080e-04, eta: 1 day, 13:46:39, time: 1.744, data_time: 0.143, memory: 49164, loss_cls_0: 1.1509, loss_box_0: 2.1333, loss_cns_0: 0.6154, loss_yns_0: 0.1682, loss_cls_1: 1.2491, loss_box_1: 2.8597, loss_cns_1: 0.5606, loss_yns_1: 0.1745, loss_cls_2: 1.2477, loss_box_2: 2.8527, loss_cns_2: 0.5586, loss_yns_2: 0.1739, loss_cls_3: 1.2442, loss_box_3: 2.9111, loss_cns_3: 0.5557, loss_yns_3: 0.1717, loss_cls_4: 1.2524, loss_box_4: 2.8816, loss_cns_4: 0.5556, loss_yns_4: 0.1725, loss_cls_5: 1.2673, loss_box_5: 2.9341, loss_cns_5: 0.5523, loss_yns_5: 0.1749, loss_cls_dn_0: 0.4803, loss_box_dn_0: 1.0060, loss_cls_dn_1: 0.4353, loss_box_dn_1: 1.2389, loss_cls_dn_2: 0.4294, loss_box_dn_2: 1.2651, loss_cls_dn_3: 0.4337, loss_box_dn_3: 1.3095, loss_cls_dn_4: 0.4335, loss_box_dn_4: 1.2841, loss_cls_dn_5: 0.4367, loss_box_dn_5: 1.3419, loss_dense_depth: 1.0619, loss: 39.5742, grad_norm: 63.4659 -2026-01-14 21:32:07,116 - mmdet - INFO - Iter [22/17500] lr: 1.084e-04, eta: 1 day, 12:25:33, time: 1.666, data_time: 0.167, memory: 49164, loss_cls_0: 1.1595, loss_box_0: 2.0904, loss_cns_0: 0.6171, loss_yns_0: 0.1684, loss_cls_1: 1.2529, loss_box_1: 2.8486, loss_cns_1: 0.5453, loss_yns_1: 0.1749, loss_cls_2: 1.2598, loss_box_2: 2.8014, loss_cns_2: 0.5508, loss_yns_2: 0.1742, loss_cls_3: 1.2613, loss_box_3: 2.8285, loss_cns_3: 0.5579, loss_yns_3: 0.1723, loss_cls_4: 1.2522, loss_box_4: 2.8141, loss_cns_4: 0.5548, loss_yns_4: 0.1713, loss_cls_5: 1.2639, loss_box_5: 2.8634, loss_cns_5: 0.5649, loss_yns_5: 0.1746, loss_cls_dn_0: 0.4815, loss_box_dn_0: 1.0037, loss_cls_dn_1: 0.4424, loss_box_dn_1: 1.1120, loss_cls_dn_2: 0.4284, loss_box_dn_2: 1.0996, loss_cls_dn_3: 0.4367, loss_box_dn_3: 1.1483, loss_cls_dn_4: 0.4410, loss_box_dn_4: 1.1775, loss_cls_dn_5: 0.4473, loss_box_dn_5: 1.2094, loss_dense_depth: 1.0432, loss: 38.5933, grad_norm: 60.9630 -2026-01-14 21:32:08,703 - mmdet - INFO - Iter [23/17500] lr: 1.088e-04, eta: 1 day, 11:10:20, time: 1.573, data_time: 0.086, memory: 49164, loss_cls_0: 1.1501, loss_box_0: 2.0869, loss_cns_0: 0.6202, loss_yns_0: 0.1694, loss_cls_1: 1.2240, loss_box_1: 2.8681, loss_cns_1: 0.5370, loss_yns_1: 0.1734, loss_cls_2: 1.2474, loss_box_2: 2.7761, loss_cns_2: 0.5494, loss_yns_2: 0.1768, loss_cls_3: 1.2455, loss_box_3: 2.7871, loss_cns_3: 0.5630, loss_yns_3: 0.1752, loss_cls_4: 1.2404, loss_box_4: 2.8463, loss_cns_4: 0.5682, loss_yns_4: 0.1707, loss_cls_5: 1.2386, loss_box_5: 2.8670, loss_cns_5: 0.5729, loss_yns_5: 0.1734, loss_cls_dn_0: 0.4745, loss_box_dn_0: 0.9861, loss_cls_dn_1: 0.4415, loss_box_dn_1: 1.1820, loss_cls_dn_2: 0.4238, loss_box_dn_2: 1.1482, loss_cls_dn_3: 0.4345, loss_box_dn_3: 1.2084, loss_cls_dn_4: 0.4279, loss_box_dn_4: 1.3274, loss_cls_dn_5: 0.4489, loss_box_dn_5: 1.3487, loss_dense_depth: 1.0587, loss: 38.9373, grad_norm: 59.0734 -2026-01-14 21:32:10,277 - mmdet - INFO - Iter [24/17500] lr: 1.092e-04, eta: 1 day, 10:01:29, time: 1.582, data_time: 0.094, memory: 49164, loss_cls_0: 1.1691, loss_box_0: 2.1280, loss_cns_0: 0.6247, loss_yns_0: 0.1693, loss_cls_1: 1.2366, loss_box_1: 2.8881, loss_cns_1: 0.5516, loss_yns_1: 0.1724, loss_cls_2: 1.2574, loss_box_2: 2.8389, loss_cns_2: 0.5598, loss_yns_2: 0.1736, loss_cls_3: 1.2459, loss_box_3: 2.8771, loss_cns_3: 0.5683, loss_yns_3: 0.1714, loss_cls_4: 1.2511, loss_box_4: 2.9413, loss_cns_4: 0.5755, loss_yns_4: 0.1709, loss_cls_5: 1.2428, loss_box_5: 2.9272, loss_cns_5: 0.5686, loss_yns_5: 0.1685, loss_cls_dn_0: 0.4726, loss_box_dn_0: 0.9998, loss_cls_dn_1: 0.4306, loss_box_dn_1: 1.2815, loss_cls_dn_2: 0.4189, loss_box_dn_2: 1.2728, loss_cls_dn_3: 0.4277, loss_box_dn_3: 1.3640, loss_cls_dn_4: 0.4145, loss_box_dn_4: 1.5127, loss_cls_dn_5: 0.4394, loss_box_dn_5: 1.5125, loss_dense_depth: 1.0894, loss: 40.1144, grad_norm: 63.7911 -2026-01-14 21:32:11,898 - mmdet - INFO - Iter [25/17500] lr: 1.096e-04, eta: 1 day, 8:58:51, time: 1.643, data_time: 0.086, memory: 49164, loss_cls_0: 1.1792, loss_box_0: 2.1445, loss_cns_0: 0.6263, loss_yns_0: 0.1688, loss_cls_1: 1.2277, loss_box_1: 2.7953, loss_cns_1: 0.5544, loss_yns_1: 0.1718, loss_cls_2: 1.2611, loss_box_2: 2.7907, loss_cns_2: 0.5652, loss_yns_2: 0.1721, loss_cls_3: 1.2404, loss_box_3: 2.7966, loss_cns_3: 0.5730, loss_yns_3: 0.1719, loss_cls_4: 1.2380, loss_box_4: 2.8459, loss_cns_4: 0.5843, loss_yns_4: 0.1735, loss_cls_5: 1.2450, loss_box_5: 2.8029, loss_cns_5: 0.5851, loss_yns_5: 0.1672, loss_cls_dn_0: 0.4615, loss_box_dn_0: 0.9938, loss_cls_dn_1: 0.4202, loss_box_dn_1: 1.3082, loss_cls_dn_2: 0.4145, loss_box_dn_2: 1.3329, loss_cls_dn_3: 0.4244, loss_box_dn_3: 1.4242, loss_cls_dn_4: 0.4133, loss_box_dn_4: 1.5415, loss_cls_dn_5: 0.4327, loss_box_dn_5: 1.5147, loss_dense_depth: 1.0634, loss: 39.8262, grad_norm: 64.0369 -2026-01-14 21:32:13,524 - mmdet - INFO - Iter [26/17500] lr: 1.100e-04, eta: 1 day, 8:00:31, time: 1.597, data_time: 0.077, memory: 49164, loss_cls_0: 1.1595, loss_box_0: 2.1207, loss_cns_0: 0.6247, loss_yns_0: 0.1692, loss_cls_1: 1.2180, loss_box_1: 2.6627, loss_cns_1: 0.5732, loss_yns_1: 0.1698, loss_cls_2: 1.2470, loss_box_2: 2.6871, loss_cns_2: 0.5891, loss_yns_2: 0.1750, loss_cls_3: 1.2312, loss_box_3: 2.6915, loss_cns_3: 0.5890, loss_yns_3: 0.1692, loss_cls_4: 1.2343, loss_box_4: 2.7146, loss_cns_4: 0.6084, loss_yns_4: 0.1698, loss_cls_5: 1.2450, loss_box_5: 2.7305, loss_cns_5: 0.6136, loss_yns_5: 0.1724, loss_cls_dn_0: 0.4486, loss_box_dn_0: 0.9987, loss_cls_dn_1: 0.4375, loss_box_dn_1: 1.1948, loss_cls_dn_2: 0.4298, loss_box_dn_2: 1.2289, loss_cls_dn_3: 0.4423, loss_box_dn_3: 1.3056, loss_cls_dn_4: 0.4357, loss_box_dn_4: 1.3798, loss_cls_dn_5: 0.4530, loss_box_dn_5: 1.3693, loss_dense_depth: 1.0340, loss: 38.7237, grad_norm: 69.6041 -2026-01-14 21:32:15,088 - mmdet - INFO - Iter [27/17500] lr: 1.104e-04, eta: 1 day, 7:06:28, time: 1.593, data_time: 0.093, memory: 49164, loss_cls_0: 1.1458, loss_box_0: 2.1210, loss_cns_0: 0.6186, loss_yns_0: 0.1710, loss_cls_1: 1.2155, loss_box_1: 2.5933, loss_cns_1: 0.5779, loss_yns_1: 0.1728, loss_cls_2: 1.2377, loss_box_2: 2.6488, loss_cns_2: 0.5888, loss_yns_2: 0.1778, loss_cls_3: 1.2214, loss_box_3: 2.6383, loss_cns_3: 0.5739, loss_yns_3: 0.1710, loss_cls_4: 1.2362, loss_box_4: 2.6659, loss_cns_4: 0.5775, loss_yns_4: 0.1718, loss_cls_5: 1.2283, loss_box_5: 2.7127, loss_cns_5: 0.5674, loss_yns_5: 0.1730, loss_cls_dn_0: 0.4475, loss_box_dn_0: 0.9950, loss_cls_dn_1: 0.4186, loss_box_dn_1: 1.1897, loss_cls_dn_2: 0.4173, loss_box_dn_2: 1.1750, loss_cls_dn_3: 0.4277, loss_box_dn_3: 1.1842, loss_cls_dn_4: 0.4226, loss_box_dn_4: 1.2124, loss_cls_dn_5: 0.4357, loss_box_dn_5: 1.2279, loss_dense_depth: 1.0256, loss: 37.7854, grad_norm: 63.1164 -2026-01-14 21:32:16,668 - mmdet - INFO - Iter [28/17500] lr: 1.108e-04, eta: 1 day, 6:16:09, time: 1.583, data_time: 0.077, memory: 49164, loss_cls_0: 1.1302, loss_box_0: 2.1000, loss_cns_0: 0.6166, loss_yns_0: 0.1705, loss_cls_1: 1.2090, loss_box_1: 2.4726, loss_cns_1: 0.5849, loss_yns_1: 0.1724, loss_cls_2: 1.2254, loss_box_2: 2.4564, loss_cns_2: 0.5930, loss_yns_2: 0.1769, loss_cls_3: 1.2306, loss_box_3: 2.4514, loss_cns_3: 0.5905, loss_yns_3: 0.1729, loss_cls_4: 1.2433, loss_box_4: 2.5100, loss_cns_4: 0.6010, loss_yns_4: 0.1729, loss_cls_5: 1.2556, loss_box_5: 2.4614, loss_cns_5: 0.5923, loss_yns_5: 0.1754, loss_cls_dn_0: 0.4406, loss_box_dn_0: 0.9880, loss_cls_dn_1: 0.4026, loss_box_dn_1: 1.1689, loss_cls_dn_2: 0.4088, loss_box_dn_2: 1.1253, loss_cls_dn_3: 0.4050, loss_box_dn_3: 1.1270, loss_cls_dn_4: 0.4025, loss_box_dn_4: 1.1580, loss_cls_dn_5: 0.4044, loss_box_dn_5: 1.1747, loss_dense_depth: 1.0114, loss: 36.5825, grad_norm: 52.4632 -2026-01-14 21:32:18,226 - mmdet - INFO - Iter [29/17500] lr: 1.112e-04, eta: 1 day, 5:29:04, time: 1.558, data_time: 0.075, memory: 49164, loss_cls_0: 1.1056, loss_box_0: 2.0648, loss_cns_0: 0.6183, loss_yns_0: 0.1686, loss_cls_1: 1.2001, loss_box_1: 2.4410, loss_cns_1: 0.6000, loss_yns_1: 0.1726, loss_cls_2: 1.1963, loss_box_2: 2.4001, loss_cns_2: 0.6081, loss_yns_2: 0.1764, loss_cls_3: 1.2078, loss_box_3: 2.4384, loss_cns_3: 0.6084, loss_yns_3: 0.1732, loss_cls_4: 1.2222, loss_box_4: 2.5302, loss_cns_4: 0.6050, loss_yns_4: 0.1727, loss_cls_5: 1.2500, loss_box_5: 2.5374, loss_cns_5: 0.5900, loss_yns_5: 0.1717, loss_cls_dn_0: 0.4380, loss_box_dn_0: 0.9845, loss_cls_dn_1: 0.4056, loss_box_dn_1: 1.0807, loss_cls_dn_2: 0.4303, loss_box_dn_2: 1.0768, loss_cls_dn_3: 0.4158, loss_box_dn_3: 1.1177, loss_cls_dn_4: 0.4124, loss_box_dn_4: 1.1768, loss_cls_dn_5: 0.4067, loss_box_dn_5: 1.2692, loss_dense_depth: 1.0695, loss: 36.5428, grad_norm: 68.6206 -2026-01-14 21:32:19,854 - mmdet - INFO - Iter [30/17500] lr: 1.116e-04, eta: 1 day, 4:45:49, time: 1.628, data_time: 0.074, memory: 49164, loss_cls_0: 1.0849, loss_box_0: 2.0109, loss_cns_0: 0.6223, loss_yns_0: 0.1706, loss_cls_1: 1.1642, loss_box_1: 2.4477, loss_cns_1: 0.5985, loss_yns_1: 0.1736, loss_cls_2: 1.1821, loss_box_2: 2.4487, loss_cns_2: 0.6099, loss_yns_2: 0.1760, loss_cls_3: 1.1821, loss_box_3: 2.4922, loss_cns_3: 0.6049, loss_yns_3: 0.1740, loss_cls_4: 1.1804, loss_box_4: 2.5653, loss_cns_4: 0.5933, loss_yns_4: 0.1712, loss_cls_5: 1.1919, loss_box_5: 2.6155, loss_cns_5: 0.5871, loss_yns_5: 0.1707, loss_cls_dn_0: 0.4389, loss_box_dn_0: 0.9766, loss_cls_dn_1: 0.3977, loss_box_dn_1: 1.1436, loss_cls_dn_2: 0.4348, loss_box_dn_2: 1.1979, loss_cls_dn_3: 0.4169, loss_box_dn_3: 1.2516, loss_cls_dn_4: 0.4129, loss_box_dn_4: 1.3069, loss_cls_dn_5: 0.4023, loss_box_dn_5: 1.4308, loss_dense_depth: 1.0274, loss: 37.0566, grad_norm: 78.7138 -2026-01-14 21:32:21,462 - mmdet - INFO - Iter [31/17500] lr: 1.120e-04, eta: 1 day, 4:05:09, time: 1.608, data_time: 0.073, memory: 49164, loss_cls_0: 1.0735, loss_box_0: 2.0440, loss_cns_0: 0.6230, loss_yns_0: 0.1737, loss_cls_1: 1.1534, loss_box_1: 2.4715, loss_cns_1: 0.5976, loss_yns_1: 0.1714, loss_cls_2: 1.1737, loss_box_2: 2.4847, loss_cns_2: 0.6043, loss_yns_2: 0.1741, loss_cls_3: 1.1771, loss_box_3: 2.4960, loss_cns_3: 0.6039, loss_yns_3: 0.1737, loss_cls_4: 1.1660, loss_box_4: 2.5392, loss_cns_4: 0.6012, loss_yns_4: 0.1713, loss_cls_5: 1.1714, loss_box_5: 2.5661, loss_cns_5: 0.6044, loss_yns_5: 0.1713, loss_cls_dn_0: 0.4453, loss_box_dn_0: 0.9670, loss_cls_dn_1: 0.3925, loss_box_dn_1: 1.2124, loss_cls_dn_2: 0.4314, loss_box_dn_2: 1.2647, loss_cls_dn_3: 0.4155, loss_box_dn_3: 1.3087, loss_cls_dn_4: 0.4130, loss_box_dn_4: 1.3430, loss_cls_dn_5: 0.4081, loss_box_dn_5: 1.4530, loss_dense_depth: 1.0325, loss: 37.2736, grad_norm: 57.5837 -2026-01-14 21:32:23,094 - mmdet - INFO - Iter [32/17500] lr: 1.124e-04, eta: 1 day, 3:27:13, time: 1.629, data_time: 0.060, memory: 49164, loss_cls_0: 1.0564, loss_box_0: 2.0863, loss_cns_0: 0.6200, loss_yns_0: 0.1714, loss_cls_1: 1.1264, loss_box_1: 2.5104, loss_cns_1: 0.5783, loss_yns_1: 0.1699, loss_cls_2: 1.1370, loss_box_2: 2.5202, loss_cns_2: 0.5783, loss_yns_2: 0.1698, loss_cls_3: 1.1558, loss_box_3: 2.5057, loss_cns_3: 0.5894, loss_yns_3: 0.1745, loss_cls_4: 1.1518, loss_box_4: 2.5299, loss_cns_4: 0.5812, loss_yns_4: 0.1706, loss_cls_5: 1.1632, loss_box_5: 2.5854, loss_cns_5: 0.5879, loss_yns_5: 0.1716, loss_cls_dn_0: 0.4273, loss_box_dn_0: 0.9776, loss_cls_dn_1: 0.3789, loss_box_dn_1: 1.1862, loss_cls_dn_2: 0.4076, loss_box_dn_2: 1.2299, loss_cls_dn_3: 0.3937, loss_box_dn_3: 1.2550, loss_cls_dn_4: 0.3956, loss_box_dn_4: 1.2719, loss_cls_dn_5: 0.4037, loss_box_dn_5: 1.3691, loss_dense_depth: 1.0687, loss: 36.8565, grad_norm: 65.8603 -2026-01-14 21:32:24,786 - mmdet - INFO - Iter [33/17500] lr: 1.128e-04, eta: 1 day, 2:52:09, time: 1.694, data_time: 0.077, memory: 49164, loss_cls_0: 1.0939, loss_box_0: 2.0658, loss_cns_0: 0.6233, loss_yns_0: 0.1702, loss_cls_1: 1.1278, loss_box_1: 2.5535, loss_cns_1: 0.5622, loss_yns_1: 0.1694, loss_cls_2: 1.1345, loss_box_2: 2.4945, loss_cns_2: 0.5834, loss_yns_2: 0.1716, loss_cls_3: 1.1683, loss_box_3: 2.4553, loss_cns_3: 0.5916, loss_yns_3: 0.1758, loss_cls_4: 1.1592, loss_box_4: 2.4355, loss_cns_4: 0.5960, loss_yns_4: 0.1709, loss_cls_5: 1.1587, loss_box_5: 2.4582, loss_cns_5: 0.5991, loss_yns_5: 0.1720, loss_cls_dn_0: 0.4162, loss_box_dn_0: 0.9833, loss_cls_dn_1: 0.3448, loss_box_dn_1: 1.3275, loss_cls_dn_2: 0.3653, loss_box_dn_2: 1.3467, loss_cls_dn_3: 0.3505, loss_box_dn_3: 1.3429, loss_cls_dn_4: 0.3573, loss_box_dn_4: 1.3339, loss_cls_dn_5: 0.3718, loss_box_dn_5: 1.3854, loss_dense_depth: 0.9990, loss: 36.8151, grad_norm: 52.4361 -2026-01-14 21:32:26,357 - mmdet - INFO - Iter [34/17500] lr: 1.132e-04, eta: 1 day, 2:18:05, time: 1.570, data_time: 0.083, memory: 49164, loss_cls_0: 1.1016, loss_box_0: 2.0252, loss_cns_0: 0.6245, loss_yns_0: 0.1721, loss_cls_1: 1.1505, loss_box_1: 2.4148, loss_cns_1: 0.5847, loss_yns_1: 0.1708, loss_cls_2: 1.1836, loss_box_2: 2.4071, loss_cns_2: 0.5994, loss_yns_2: 0.1723, loss_cls_3: 1.1821, loss_box_3: 2.4301, loss_cns_3: 0.5949, loss_yns_3: 0.1743, loss_cls_4: 1.1826, loss_box_4: 2.4193, loss_cns_4: 0.6033, loss_yns_4: 0.1706, loss_cls_5: 1.1541, loss_box_5: 2.3966, loss_cns_5: 0.6064, loss_yns_5: 0.1737, loss_cls_dn_0: 0.4151, loss_box_dn_0: 0.9690, loss_cls_dn_1: 0.3265, loss_box_dn_1: 1.2821, loss_cls_dn_2: 0.3344, loss_box_dn_2: 1.2855, loss_cls_dn_3: 0.3336, loss_box_dn_3: 1.2861, loss_cls_dn_4: 0.3352, loss_box_dn_4: 1.2786, loss_cls_dn_5: 0.3549, loss_box_dn_5: 1.2751, loss_dense_depth: 1.0706, loss: 36.2411, grad_norm: 65.4751 -2026-01-14 21:32:27,946 - mmdet - INFO - Iter [35/17500] lr: 1.136e-04, eta: 1 day, 1:46:07, time: 1.588, data_time: 0.072, memory: 49164, loss_cls_0: 1.0742, loss_box_0: 1.9736, loss_cns_0: 0.6247, loss_yns_0: 0.1720, loss_cls_1: 1.1774, loss_box_1: 2.3595, loss_cns_1: 0.5899, loss_yns_1: 0.1696, loss_cls_2: 1.2310, loss_box_2: 2.3585, loss_cns_2: 0.5947, loss_yns_2: 0.1731, loss_cls_3: 1.1747, loss_box_3: 2.4321, loss_cns_3: 0.5806, loss_yns_3: 0.1732, loss_cls_4: 1.1993, loss_box_4: 2.4644, loss_cns_4: 0.5926, loss_yns_4: 0.1703, loss_cls_5: 1.1780, loss_box_5: 2.4115, loss_cns_5: 0.5989, loss_yns_5: 0.1726, loss_cls_dn_0: 0.4234, loss_box_dn_0: 0.9441, loss_cls_dn_1: 0.3215, loss_box_dn_1: 1.1799, loss_cls_dn_2: 0.3252, loss_box_dn_2: 1.1746, loss_cls_dn_3: 0.3388, loss_box_dn_3: 1.2051, loss_cls_dn_4: 0.3356, loss_box_dn_4: 1.2475, loss_cls_dn_5: 0.3522, loss_box_dn_5: 1.2254, loss_dense_depth: 1.0579, loss: 35.7777, grad_norm: 68.9308 -2026-01-14 21:32:29,542 - mmdet - INFO - Iter [36/17500] lr: 1.140e-04, eta: 1 day, 1:16:00, time: 1.598, data_time: 0.073, memory: 49164, loss_cls_0: 1.0477, loss_box_0: 1.9699, loss_cns_0: 0.6210, loss_yns_0: 0.1719, loss_cls_1: 1.1386, loss_box_1: 2.3962, loss_cns_1: 0.5958, loss_yns_1: 0.1717, loss_cls_2: 1.1691, loss_box_2: 2.3649, loss_cns_2: 0.6070, loss_yns_2: 0.1710, loss_cls_3: 1.1466, loss_box_3: 2.3957, loss_cns_3: 0.6098, loss_yns_3: 0.1712, loss_cls_4: 1.1415, loss_box_4: 2.4257, loss_cns_4: 0.6116, loss_yns_4: 0.1695, loss_cls_5: 1.1681, loss_box_5: 2.3987, loss_cns_5: 0.6140, loss_yns_5: 0.1712, loss_cls_dn_0: 0.4338, loss_box_dn_0: 0.9318, loss_cls_dn_1: 0.3489, loss_box_dn_1: 1.0402, loss_cls_dn_2: 0.3581, loss_box_dn_2: 1.0247, loss_cls_dn_3: 0.3786, loss_box_dn_3: 1.0900, loss_cls_dn_4: 0.3816, loss_box_dn_4: 1.2095, loss_cls_dn_5: 0.3888, loss_box_dn_5: 1.2162, loss_dense_depth: 1.0422, loss: 35.2932, grad_norm: 67.1271 -2026-01-14 21:32:31,171 - mmdet - INFO - Iter [37/17500] lr: 1.144e-04, eta: 1 day, 0:47:46, time: 1.629, data_time: 0.073, memory: 49164, loss_cls_0: 1.0528, loss_box_0: 1.9953, loss_cns_0: 0.6212, loss_yns_0: 0.1736, loss_cls_1: 1.1170, loss_box_1: 2.4959, loss_cns_1: 0.6026, loss_yns_1: 0.1717, loss_cls_2: 1.1416, loss_box_2: 2.5549, loss_cns_2: 0.5957, loss_yns_2: 0.1718, loss_cls_3: 1.1412, loss_box_3: 2.5693, loss_cns_3: 0.5973, loss_yns_3: 0.1705, loss_cls_4: 1.1363, loss_box_4: 2.6103, loss_cns_4: 0.5977, loss_yns_4: 0.1694, loss_cls_5: 1.1481, loss_box_5: 2.6524, loss_cns_5: 0.5782, loss_yns_5: 0.1703, loss_cls_dn_0: 0.4439, loss_box_dn_0: 0.9332, loss_cls_dn_1: 0.3523, loss_box_dn_1: 1.1269, loss_cls_dn_2: 0.3684, loss_box_dn_2: 1.1329, loss_cls_dn_3: 0.3823, loss_box_dn_3: 1.1964, loss_cls_dn_4: 0.3979, loss_box_dn_4: 1.3239, loss_cls_dn_5: 0.3909, loss_box_dn_5: 1.3645, loss_dense_depth: 0.9774, loss: 36.6262, grad_norm: 92.6782 -2026-01-14 21:32:32,743 - mmdet - INFO - Iter [38/17500] lr: 1.148e-04, eta: 1 day, 0:20:34, time: 1.572, data_time: 0.074, memory: 49164, loss_cls_0: 1.0365, loss_box_0: 2.0063, loss_cns_0: 0.6234, loss_yns_0: 0.1698, loss_cls_1: 1.0892, loss_box_1: 2.5125, loss_cns_1: 0.5813, loss_yns_1: 0.1702, loss_cls_2: 1.1366, loss_box_2: 2.5527, loss_cns_2: 0.5814, loss_yns_2: 0.1727, loss_cls_3: 1.1389, loss_box_3: 2.5750, loss_cns_3: 0.5813, loss_yns_3: 0.1684, loss_cls_4: 1.1386, loss_box_4: 2.6156, loss_cns_4: 0.5805, loss_yns_4: 0.1701, loss_cls_5: 1.1359, loss_box_5: 2.6431, loss_cns_5: 0.5600, loss_yns_5: 0.1672, loss_cls_dn_0: 0.4372, loss_box_dn_0: 0.9352, loss_cls_dn_1: 0.3699, loss_box_dn_1: 1.1827, loss_cls_dn_2: 0.3896, loss_box_dn_2: 1.1621, loss_cls_dn_3: 0.3930, loss_box_dn_3: 1.2182, loss_cls_dn_4: 0.4162, loss_box_dn_4: 1.3309, loss_cls_dn_5: 0.4066, loss_box_dn_5: 1.3639, loss_dense_depth: 1.0764, loss: 36.7893, grad_norm: 86.8822 -2026-01-14 21:32:34,322 - mmdet - INFO - Iter [39/17500] lr: 1.152e-04, eta: 23:54:49, time: 1.578, data_time: 0.087, memory: 49164, loss_cls_0: 1.0575, loss_box_0: 1.9693, loss_cns_0: 0.6261, loss_yns_0: 0.1698, loss_cls_1: 1.1193, loss_box_1: 2.3866, loss_cns_1: 0.5883, loss_yns_1: 0.1698, loss_cls_2: 1.1621, loss_box_2: 2.3515, loss_cns_2: 0.6053, loss_yns_2: 0.1709, loss_cls_3: 1.1511, loss_box_3: 2.3962, loss_cns_3: 0.6152, loss_yns_3: 0.1716, loss_cls_4: 1.1471, loss_box_4: 2.4247, loss_cns_4: 0.6098, loss_yns_4: 0.1679, loss_cls_5: 1.1531, loss_box_5: 2.4413, loss_cns_5: 0.6251, loss_yns_5: 0.1697, loss_cls_dn_0: 0.4211, loss_box_dn_0: 0.9193, loss_cls_dn_1: 0.3832, loss_box_dn_1: 1.1370, loss_cls_dn_2: 0.4065, loss_box_dn_2: 1.0846, loss_cls_dn_3: 0.3961, loss_box_dn_3: 1.1336, loss_cls_dn_4: 0.4192, loss_box_dn_4: 1.2221, loss_cls_dn_5: 0.4114, loss_box_dn_5: 1.2424, loss_dense_depth: 1.0037, loss: 35.6294, grad_norm: 69.6466 -2026-01-14 21:32:35,899 - mmdet - INFO - Iter [40/17500] lr: 1.156e-04, eta: 23:30:20, time: 1.578, data_time: 0.072, memory: 49164, loss_cls_0: 1.0791, loss_box_0: 1.9559, loss_cns_0: 0.6278, loss_yns_0: 0.1729, loss_cls_1: 1.1064, loss_box_1: 2.3912, loss_cns_1: 0.5913, loss_yns_1: 0.1718, loss_cls_2: 1.1533, loss_box_2: 2.3855, loss_cns_2: 0.6036, loss_yns_2: 0.1700, loss_cls_3: 1.1443, loss_box_3: 2.4468, loss_cns_3: 0.6083, loss_yns_3: 0.1704, loss_cls_4: 1.1378, loss_box_4: 2.4847, loss_cns_4: 0.6032, loss_yns_4: 0.1682, loss_cls_5: 1.1554, loss_box_5: 2.5386, loss_cns_5: 0.6159, loss_yns_5: 0.1697, loss_cls_dn_0: 0.3963, loss_box_dn_0: 0.9141, loss_cls_dn_1: 0.3853, loss_box_dn_1: 1.1202, loss_cls_dn_2: 0.4023, loss_box_dn_2: 1.0853, loss_cls_dn_3: 0.3823, loss_box_dn_3: 1.1170, loss_cls_dn_4: 0.3936, loss_box_dn_4: 1.1682, loss_cls_dn_5: 0.3943, loss_box_dn_5: 1.1844, loss_dense_depth: 1.0060, loss: 35.6013, grad_norm: 86.2683 -2026-01-14 21:32:37,578 - mmdet - INFO - Iter [41/17500] lr: 1.160e-04, eta: 23:07:46, time: 1.677, data_time: 0.112, memory: 49164, loss_cls_0: 1.0657, loss_box_0: 1.9483, loss_cns_0: 0.6253, loss_yns_0: 0.1741, loss_cls_1: 1.0871, loss_box_1: 2.3743, loss_cns_1: 0.5856, loss_yns_1: 0.1668, loss_cls_2: 1.1082, loss_box_2: 2.3693, loss_cns_2: 0.5992, loss_yns_2: 0.1706, loss_cls_3: 1.1291, loss_box_3: 2.3776, loss_cns_3: 0.6008, loss_yns_3: 0.1691, loss_cls_4: 1.1307, loss_box_4: 2.3924, loss_cns_4: 0.6027, loss_yns_4: 0.1706, loss_cls_5: 1.1392, loss_box_5: 2.4622, loss_cns_5: 0.6146, loss_yns_5: 0.1664, loss_cls_dn_0: 0.3827, loss_box_dn_0: 0.9108, loss_cls_dn_1: 0.3568, loss_box_dn_1: 1.1020, loss_cls_dn_2: 0.3669, loss_box_dn_2: 1.0786, loss_cls_dn_3: 0.3467, loss_box_dn_3: 1.0725, loss_cls_dn_4: 0.3446, loss_box_dn_4: 1.0915, loss_cls_dn_5: 0.3554, loss_box_dn_5: 1.1080, loss_dense_depth: 1.0186, loss: 34.7649, grad_norm: 69.3061 -2026-01-14 21:32:39,260 - mmdet - INFO - Iter [42/17500] lr: 1.164e-04, eta: 22:46:06, time: 1.654, data_time: 0.174, memory: 49164, loss_cls_0: 1.0461, loss_box_0: 1.9308, loss_cns_0: 0.6268, loss_yns_0: 0.1671, loss_cls_1: 1.1082, loss_box_1: 2.3377, loss_cns_1: 0.6032, loss_yns_1: 0.1660, loss_cls_2: 1.1271, loss_box_2: 2.3027, loss_cns_2: 0.6247, loss_yns_2: 0.1690, loss_cls_3: 1.1579, loss_box_3: 2.2637, loss_cns_3: 0.6289, loss_yns_3: 0.1703, loss_cls_4: 1.1716, loss_box_4: 2.2589, loss_cns_4: 0.6431, loss_yns_4: 0.1685, loss_cls_5: 1.1500, loss_box_5: 2.2736, loss_cns_5: 0.6516, loss_yns_5: 0.1658, loss_cls_dn_0: 0.3913, loss_box_dn_0: 0.9144, loss_cls_dn_1: 0.3471, loss_box_dn_1: 1.0205, loss_cls_dn_2: 0.3523, loss_box_dn_2: 1.0008, loss_cls_dn_3: 0.3465, loss_box_dn_3: 0.9815, loss_cls_dn_4: 0.3300, loss_box_dn_4: 0.9823, loss_cls_dn_5: 0.3527, loss_box_dn_5: 1.0040, loss_dense_depth: 0.9509, loss: 33.8875, grad_norm: 58.8759 -2026-01-14 21:32:40,814 - mmdet - INFO - Iter [43/17500] lr: 1.168e-04, eta: 22:24:58, time: 1.582, data_time: 0.093, memory: 49164, loss_cls_0: 1.0291, loss_box_0: 1.9793, loss_cns_0: 0.6164, loss_yns_0: 0.1653, loss_cls_1: 1.1432, loss_box_1: 2.2960, loss_cns_1: 0.6086, loss_yns_1: 0.1642, loss_cls_2: 1.1906, loss_box_2: 2.2873, loss_cns_2: 0.6212, loss_yns_2: 0.1664, loss_cls_3: 1.1577, loss_box_3: 2.2978, loss_cns_3: 0.6207, loss_yns_3: 0.1654, loss_cls_4: 1.1899, loss_box_4: 2.3601, loss_cns_4: 0.6190, loss_yns_4: 0.1649, loss_cls_5: 1.1767, loss_box_5: 2.3720, loss_cns_5: 0.6117, loss_yns_5: 0.1643, loss_cls_dn_0: 0.4238, loss_box_dn_0: 0.9231, loss_cls_dn_1: 0.3523, loss_box_dn_1: 0.9824, loss_cls_dn_2: 0.3587, loss_box_dn_2: 0.9991, loss_cls_dn_3: 0.3704, loss_box_dn_3: 1.0204, loss_cls_dn_4: 0.3477, loss_box_dn_4: 1.0528, loss_cls_dn_5: 0.3746, loss_box_dn_5: 1.1152, loss_dense_depth: 0.9609, loss: 34.4490, grad_norm: 67.4939 -2026-01-14 21:32:42,418 - mmdet - INFO - Iter [44/17500] lr: 1.172e-04, eta: 22:04:45, time: 1.576, data_time: 0.073, memory: 49164, loss_cls_0: 1.0281, loss_box_0: 2.0072, loss_cns_0: 0.6108, loss_yns_0: 0.1644, loss_cls_1: 1.1057, loss_box_1: 2.2880, loss_cns_1: 0.6030, loss_yns_1: 0.1622, loss_cls_2: 1.1504, loss_box_2: 2.3116, loss_cns_2: 0.6030, loss_yns_2: 0.1654, loss_cls_3: 1.1177, loss_box_3: 2.3792, loss_cns_3: 0.6037, loss_yns_3: 0.1648, loss_cls_4: 1.1334, loss_box_4: 2.4377, loss_cns_4: 0.5950, loss_yns_4: 0.1649, loss_cls_5: 1.1421, loss_box_5: 2.4753, loss_cns_5: 0.5917, loss_yns_5: 0.1646, loss_cls_dn_0: 0.4447, loss_box_dn_0: 0.9198, loss_cls_dn_1: 0.3589, loss_box_dn_1: 1.0668, loss_cls_dn_2: 0.3702, loss_box_dn_2: 1.1030, loss_cls_dn_3: 0.3879, loss_box_dn_3: 1.1512, loss_cls_dn_4: 0.3681, loss_box_dn_4: 1.1891, loss_cls_dn_5: 0.3917, loss_box_dn_5: 1.2786, loss_dense_depth: 1.0408, loss: 35.2407, grad_norm: 78.5318 -2026-01-14 21:32:44,082 - mmdet - INFO - Iter [45/17500] lr: 1.176e-04, eta: 21:45:53, time: 1.649, data_time: 0.097, memory: 49164, loss_cls_0: 1.0562, loss_box_0: 1.9842, loss_cns_0: 0.6150, loss_yns_0: 0.1654, loss_cls_1: 1.1080, loss_box_1: 2.3631, loss_cns_1: 0.5965, loss_yns_1: 0.1660, loss_cls_2: 1.1329, loss_box_2: 2.3520, loss_cns_2: 0.6064, loss_yns_2: 0.1676, loss_cls_3: 1.1397, loss_box_3: 2.3836, loss_cns_3: 0.6115, loss_yns_3: 0.1669, loss_cls_4: 1.1351, loss_box_4: 2.3965, loss_cns_4: 0.6144, loss_yns_4: 0.1653, loss_cls_5: 1.1425, loss_box_5: 2.4250, loss_cns_5: 0.6111, loss_yns_5: 0.1659, loss_cls_dn_0: 0.4443, loss_box_dn_0: 0.9157, loss_cls_dn_1: 0.3663, loss_box_dn_1: 1.1298, loss_cls_dn_2: 0.3812, loss_box_dn_2: 1.1745, loss_cls_dn_3: 0.3996, loss_box_dn_3: 1.2108, loss_cls_dn_4: 0.3853, loss_box_dn_4: 1.2313, loss_cls_dn_5: 0.4046, loss_box_dn_5: 1.3156, loss_dense_depth: 0.9793, loss: 35.6094, grad_norm: 74.2983 -2026-01-14 21:32:45,687 - mmdet - INFO - Iter [46/17500] lr: 1.180e-04, eta: 21:27:41, time: 1.623, data_time: 0.102, memory: 49164, loss_cls_0: 1.0515, loss_box_0: 1.9837, loss_cns_0: 0.6180, loss_yns_0: 0.1643, loss_cls_1: 1.1167, loss_box_1: 2.5306, loss_cns_1: 0.5909, loss_yns_1: 0.1622, loss_cls_2: 1.1163, loss_box_2: 2.4541, loss_cns_2: 0.6101, loss_yns_2: 0.1646, loss_cls_3: 1.1453, loss_box_3: 2.4355, loss_cns_3: 0.6153, loss_yns_3: 0.1640, loss_cls_4: 1.1225, loss_box_4: 2.4175, loss_cns_4: 0.6185, loss_yns_4: 0.1666, loss_cls_5: 1.1273, loss_box_5: 2.4650, loss_cns_5: 0.6100, loss_yns_5: 0.1622, loss_cls_dn_0: 0.4267, loss_box_dn_0: 0.9102, loss_cls_dn_1: 0.3582, loss_box_dn_1: 1.1350, loss_cls_dn_2: 0.3768, loss_box_dn_2: 1.1724, loss_cls_dn_3: 0.3876, loss_box_dn_3: 1.1870, loss_cls_dn_4: 0.3801, loss_box_dn_4: 1.1862, loss_cls_dn_5: 0.3930, loss_box_dn_5: 1.2575, loss_dense_depth: 0.9488, loss: 35.7323, grad_norm: 62.7892 -2026-01-14 21:32:47,272 - mmdet - INFO - Iter [47/17500] lr: 1.184e-04, eta: 21:10:11, time: 1.611, data_time: 0.101, memory: 49164, loss_cls_0: 1.0342, loss_box_0: 2.0008, loss_cns_0: 0.6196, loss_yns_0: 0.1636, loss_cls_1: 1.1303, loss_box_1: 2.4626, loss_cns_1: 0.6052, loss_yns_1: 0.1618, loss_cls_2: 1.1305, loss_box_2: 2.4274, loss_cns_2: 0.6131, loss_yns_2: 0.1625, loss_cls_3: 1.1470, loss_box_3: 2.4463, loss_cns_3: 0.6129, loss_yns_3: 0.1617, loss_cls_4: 1.1292, loss_box_4: 2.4441, loss_cns_4: 0.6071, loss_yns_4: 0.1689, loss_cls_5: 1.1380, loss_box_5: 2.4593, loss_cns_5: 0.6033, loss_yns_5: 0.1616, loss_cls_dn_0: 0.4159, loss_box_dn_0: 0.9169, loss_cls_dn_1: 0.3515, loss_box_dn_1: 1.0932, loss_cls_dn_2: 0.3744, loss_box_dn_2: 1.1281, loss_cls_dn_3: 0.3788, loss_box_dn_3: 1.1459, loss_cls_dn_4: 0.3797, loss_box_dn_4: 1.1432, loss_cls_dn_5: 0.3872, loss_box_dn_5: 1.1904, loss_dense_depth: 0.9511, loss: 35.4474, grad_norm: 62.5237 -2026-01-14 21:32:48,855 - mmdet - INFO - Iter [48/17500] lr: 1.188e-04, eta: 20:53:15, time: 1.583, data_time: 0.074, memory: 49164, loss_cls_0: 1.0313, loss_box_0: 1.9915, loss_cns_0: 0.6179, loss_yns_0: 0.1632, loss_cls_1: 1.1036, loss_box_1: 2.4043, loss_cns_1: 0.6041, loss_yns_1: 0.1651, loss_cls_2: 1.1234, loss_box_2: 2.3494, loss_cns_2: 0.6223, loss_yns_2: 0.1637, loss_cls_3: 1.1350, loss_box_3: 2.3811, loss_cns_3: 0.6254, loss_yns_3: 0.1640, loss_cls_4: 1.1206, loss_box_4: 2.3859, loss_cns_4: 0.6275, loss_yns_4: 0.1668, loss_cls_5: 1.1314, loss_box_5: 2.3782, loss_cns_5: 0.6280, loss_yns_5: 0.1647, loss_cls_dn_0: 0.4031, loss_box_dn_0: 0.9034, loss_cls_dn_1: 0.3326, loss_box_dn_1: 1.1342, loss_cls_dn_2: 0.3634, loss_box_dn_2: 1.1264, loss_cls_dn_3: 0.3596, loss_box_dn_3: 1.1350, loss_cls_dn_4: 0.3655, loss_box_dn_4: 1.1279, loss_cls_dn_5: 0.3735, loss_box_dn_5: 1.1450, loss_dense_depth: 0.9283, loss: 34.9462, grad_norm: 64.9400 -2026-01-14 21:32:50,462 - mmdet - INFO - Iter [49/17500] lr: 1.192e-04, eta: 20:37:09, time: 1.608, data_time: 0.074, memory: 49164, loss_cls_0: 1.0778, loss_box_0: 1.9880, loss_cns_0: 0.6177, loss_yns_0: 0.1646, loss_cls_1: 1.1186, loss_box_1: 2.4294, loss_cns_1: 0.6007, loss_yns_1: 0.1657, loss_cls_2: 1.1162, loss_box_2: 2.3950, loss_cns_2: 0.6177, loss_yns_2: 0.1620, loss_cls_3: 1.1524, loss_box_3: 2.4170, loss_cns_3: 0.6228, loss_yns_3: 0.1634, loss_cls_4: 1.1302, loss_box_4: 2.4177, loss_cns_4: 0.6243, loss_yns_4: 0.1686, loss_cls_5: 1.1303, loss_box_5: 2.4258, loss_cns_5: 0.6249, loss_yns_5: 0.1655, loss_cls_dn_0: 0.3972, loss_box_dn_0: 0.8938, loss_cls_dn_1: 0.3221, loss_box_dn_1: 1.1597, loss_cls_dn_2: 0.3535, loss_box_dn_2: 1.1384, loss_cls_dn_3: 0.3491, loss_box_dn_3: 1.1318, loss_cls_dn_4: 0.3510, loss_box_dn_4: 1.1242, loss_cls_dn_5: 0.3666, loss_box_dn_5: 1.1350, loss_dense_depth: 0.9484, loss: 35.1672, grad_norm: 68.3342 -2026-01-14 21:32:52,036 - mmdet - INFO - Iter [50/17500] lr: 1.196e-04, eta: 20:21:29, time: 1.574, data_time: 0.081, memory: 49164, loss_cls_0: 1.0575, loss_box_0: 1.9696, loss_cns_0: 0.6196, loss_yns_0: 0.1639, loss_cls_1: 1.1330, loss_box_1: 2.4855, loss_cns_1: 0.5961, loss_yns_1: 0.1663, loss_cls_2: 1.1054, loss_box_2: 2.4272, loss_cns_2: 0.6106, loss_yns_2: 0.1622, loss_cls_3: 1.1528, loss_box_3: 2.4273, loss_cns_3: 0.6174, loss_yns_3: 0.1637, loss_cls_4: 1.1229, loss_box_4: 2.4372, loss_cns_4: 0.6171, loss_yns_4: 0.1689, loss_cls_5: 1.1181, loss_box_5: 2.4668, loss_cns_5: 0.6176, loss_yns_5: 0.1652, loss_cls_dn_0: 0.3847, loss_box_dn_0: 0.8989, loss_cls_dn_1: 0.3264, loss_box_dn_1: 1.0737, loss_cls_dn_2: 0.3546, loss_box_dn_2: 1.0278, loss_cls_dn_3: 0.3490, loss_box_dn_3: 1.0188, loss_cls_dn_4: 0.3458, loss_box_dn_4: 1.0307, loss_cls_dn_5: 0.3674, loss_box_dn_5: 1.0553, loss_dense_depth: 0.9108, loss: 34.7156, grad_norm: 53.8646 -2026-01-14 21:32:53,603 - mmdet - INFO - Iter [51/17500] lr: 1.200e-04, eta: 20:06:24, time: 1.567, data_time: 0.074, memory: 49164, loss_cls_0: 1.0091, loss_box_0: 1.9540, loss_cns_0: 0.6197, loss_yns_0: 0.1625, loss_cls_1: 1.0838, loss_box_1: 2.4869, loss_cns_1: 0.6005, loss_yns_1: 0.1641, loss_cls_2: 1.0884, loss_box_2: 2.4211, loss_cns_2: 0.6204, loss_yns_2: 0.1634, loss_cls_3: 1.1225, loss_box_3: 2.4204, loss_cns_3: 0.6277, loss_yns_3: 0.1662, loss_cls_4: 1.1170, loss_box_4: 2.4544, loss_cns_4: 0.6294, loss_yns_4: 0.1677, loss_cls_5: 1.1169, loss_box_5: 2.4604, loss_cns_5: 0.6298, loss_yns_5: 0.1658, loss_cls_dn_0: 0.3891, loss_box_dn_0: 0.8857, loss_cls_dn_1: 0.3181, loss_box_dn_1: 1.1117, loss_cls_dn_2: 0.3450, loss_box_dn_2: 1.0427, loss_cls_dn_3: 0.3359, loss_box_dn_3: 1.0465, loss_cls_dn_4: 0.3295, loss_box_dn_4: 1.0794, loss_cls_dn_5: 0.3466, loss_box_dn_5: 1.1048, loss_dense_depth: 0.8967, loss: 34.6839, grad_norm: 52.2908 -2026-01-14 21:32:55,188 - mmdet - INFO - Iter [52/17500] lr: 1.204e-04, eta: 19:52:00, time: 1.585, data_time: 0.073, memory: 49164, loss_cls_0: 1.0140, loss_box_0: 1.9736, loss_cns_0: 0.6132, loss_yns_0: 0.1650, loss_cls_1: 1.0927, loss_box_1: 2.3586, loss_cns_1: 0.6164, loss_yns_1: 0.1681, loss_cls_2: 1.0996, loss_box_2: 2.3515, loss_cns_2: 0.6233, loss_yns_2: 0.1648, loss_cls_3: 1.1227, loss_box_3: 2.3892, loss_cns_3: 0.6206, loss_yns_3: 0.1678, loss_cls_4: 1.1561, loss_box_4: 2.4260, loss_cns_4: 0.6212, loss_yns_4: 0.1671, loss_cls_5: 1.1587, loss_box_5: 2.4131, loss_cns_5: 0.6198, loss_yns_5: 0.1653, loss_cls_dn_0: 0.4071, loss_box_dn_0: 0.8913, loss_cls_dn_1: 0.3209, loss_box_dn_1: 1.1049, loss_cls_dn_2: 0.3405, loss_box_dn_2: 1.0788, loss_cls_dn_3: 0.3435, loss_box_dn_3: 1.1262, loss_cls_dn_4: 0.3331, loss_box_dn_4: 1.1776, loss_cls_dn_5: 0.3459, loss_box_dn_5: 1.2122, loss_dense_depth: 0.9663, loss: 34.9163, grad_norm: 70.6968 -2026-01-14 21:32:56,765 - mmdet - INFO - Iter [53/17500] lr: 1.208e-04, eta: 19:38:05, time: 1.574, data_time: 0.065, memory: 49164, loss_cls_0: 1.0498, loss_box_0: 1.9941, loss_cns_0: 0.6159, loss_yns_0: 0.1655, loss_cls_1: 1.1387, loss_box_1: 2.3709, loss_cns_1: 0.6123, loss_yns_1: 0.1693, loss_cls_2: 1.1498, loss_box_2: 2.3927, loss_cns_2: 0.6130, loss_yns_2: 0.1701, loss_cls_3: 1.1607, loss_box_3: 2.4357, loss_cns_3: 0.6098, loss_yns_3: 0.1681, loss_cls_4: 1.1754, loss_box_4: 2.4400, loss_cns_4: 0.6152, loss_yns_4: 0.1659, loss_cls_5: 1.1816, loss_box_5: 2.4235, loss_cns_5: 0.6128, loss_yns_5: 0.1648, loss_cls_dn_0: 0.4243, loss_box_dn_0: 0.8830, loss_cls_dn_1: 0.3260, loss_box_dn_1: 1.0767, loss_cls_dn_2: 0.3377, loss_box_dn_2: 1.0684, loss_cls_dn_3: 0.3529, loss_box_dn_3: 1.1291, loss_cls_dn_4: 0.3415, loss_box_dn_4: 1.1750, loss_cls_dn_5: 0.3551, loss_box_dn_5: 1.2152, loss_dense_depth: 0.9250, loss: 35.2055, grad_norm: 67.0576 -2026-01-14 21:32:58,340 - mmdet - INFO - Iter [54/17500] lr: 1.212e-04, eta: 19:24:41, time: 1.578, data_time: 0.072, memory: 49164, loss_cls_0: 1.0220, loss_box_0: 1.9238, loss_cns_0: 0.6231, loss_yns_0: 0.1633, loss_cls_1: 1.1358, loss_box_1: 2.2806, loss_cns_1: 0.6233, loss_yns_1: 0.1617, loss_cls_2: 1.1559, loss_box_2: 2.2555, loss_cns_2: 0.6304, loss_yns_2: 0.1621, loss_cls_3: 1.1470, loss_box_3: 2.2761, loss_cns_3: 0.6335, loss_yns_3: 0.1630, loss_cls_4: 1.1164, loss_box_4: 2.2882, loss_cns_4: 0.6284, loss_yns_4: 0.1619, loss_cls_5: 1.1229, loss_box_5: 2.2953, loss_cns_5: 0.6315, loss_yns_5: 0.1652, loss_cls_dn_0: 0.4115, loss_box_dn_0: 0.8813, loss_cls_dn_1: 0.3186, loss_box_dn_1: 1.0946, loss_cls_dn_2: 0.3219, loss_box_dn_2: 1.0767, loss_cls_dn_3: 0.3486, loss_box_dn_3: 1.1274, loss_cls_dn_4: 0.3423, loss_box_dn_4: 1.1672, loss_cls_dn_5: 0.3576, loss_box_dn_5: 1.2101, loss_dense_depth: 0.9461, loss: 34.3708, grad_norm: 46.9150 -2026-01-14 21:32:59,970 - mmdet - INFO - Iter [55/17500] lr: 1.216e-04, eta: 19:12:04, time: 1.631, data_time: 0.072, memory: 49164, loss_cls_0: 1.0086, loss_box_0: 1.9543, loss_cns_0: 0.6128, loss_yns_0: 0.1646, loss_cls_1: 1.0686, loss_box_1: 2.2325, loss_cns_1: 0.6163, loss_yns_1: 0.1644, loss_cls_2: 1.1011, loss_box_2: 2.1924, loss_cns_2: 0.6284, loss_yns_2: 0.1666, loss_cls_3: 1.1056, loss_box_3: 2.2217, loss_cns_3: 0.6348, loss_yns_3: 0.1641, loss_cls_4: 1.1305, loss_box_4: 2.2705, loss_cns_4: 0.6291, loss_yns_4: 0.1625, loss_cls_5: 1.1241, loss_box_5: 2.3028, loss_cns_5: 0.6348, loss_yns_5: 0.1652, loss_cls_dn_0: 0.4128, loss_box_dn_0: 0.8748, loss_cls_dn_1: 0.3402, loss_box_dn_1: 0.9707, loss_cls_dn_2: 0.3427, loss_box_dn_2: 0.9455, loss_cls_dn_3: 0.3737, loss_box_dn_3: 0.9899, loss_cls_dn_4: 0.3823, loss_box_dn_4: 1.0488, loss_cls_dn_5: 0.4036, loss_box_dn_5: 1.0974, loss_dense_depth: 1.0600, loss: 33.6986, grad_norm: 57.8834 -2026-01-14 21:33:01,575 - mmdet - INFO - Iter [56/17500] lr: 1.220e-04, eta: 18:59:35, time: 1.570, data_time: 0.072, memory: 49164, loss_cls_0: 0.9834, loss_box_0: 1.9167, loss_cns_0: 0.6137, loss_yns_0: 0.1627, loss_cls_1: 1.0718, loss_box_1: 2.1377, loss_cns_1: 0.6170, loss_yns_1: 0.1643, loss_cls_2: 1.1380, loss_box_2: 2.1008, loss_cns_2: 0.6318, loss_yns_2: 0.1705, loss_cls_3: 1.1041, loss_box_3: 2.1154, loss_cns_3: 0.6383, loss_yns_3: 0.1670, loss_cls_4: 1.1546, loss_box_4: 2.1588, loss_cns_4: 0.6343, loss_yns_4: 0.1617, loss_cls_5: 1.1341, loss_box_5: 2.1637, loss_cns_5: 0.6432, loss_yns_5: 0.1634, loss_cls_dn_0: 0.3860, loss_box_dn_0: 0.8743, loss_cls_dn_1: 0.3256, loss_box_dn_1: 0.9867, loss_cls_dn_2: 0.3427, loss_box_dn_2: 0.9596, loss_cls_dn_3: 0.3534, loss_box_dn_3: 0.9770, loss_cls_dn_4: 0.3740, loss_box_dn_4: 1.0143, loss_cls_dn_5: 0.3890, loss_box_dn_5: 1.0322, loss_dense_depth: 0.9501, loss: 32.9117, grad_norm: 65.9675 -2026-01-14 21:33:03,176 - mmdet - INFO - Iter [57/17500] lr: 1.224e-04, eta: 18:47:52, time: 1.634, data_time: 0.096, memory: 49164, loss_cls_0: 1.0195, loss_box_0: 1.8748, loss_cns_0: 0.6196, loss_yns_0: 0.1636, loss_cls_1: 1.1369, loss_box_1: 2.2006, loss_cns_1: 0.6142, loss_yns_1: 0.1711, loss_cls_2: 1.1665, loss_box_2: 2.1787, loss_cns_2: 0.6269, loss_yns_2: 0.1700, loss_cls_3: 1.1748, loss_box_3: 2.1753, loss_cns_3: 0.6312, loss_yns_3: 0.1701, loss_cls_4: 1.1561, loss_box_4: 2.1718, loss_cns_4: 0.6326, loss_yns_4: 0.1641, loss_cls_5: 1.1501, loss_box_5: 2.1767, loss_cns_5: 0.6329, loss_yns_5: 0.1690, loss_cls_dn_0: 0.3679, loss_box_dn_0: 0.8709, loss_cls_dn_1: 0.3090, loss_box_dn_1: 0.9412, loss_cls_dn_2: 0.3345, loss_box_dn_2: 0.9171, loss_cls_dn_3: 0.3277, loss_box_dn_3: 0.9163, loss_cls_dn_4: 0.3544, loss_box_dn_4: 0.9232, loss_cls_dn_5: 0.3593, loss_box_dn_5: 0.9385, loss_dense_depth: 1.0247, loss: 32.9317, grad_norm: 58.9154 -2026-01-14 21:33:04,789 - mmdet - INFO - Iter [58/17500] lr: 1.228e-04, eta: 18:36:12, time: 1.565, data_time: 0.075, memory: 49164, loss_cls_0: 1.0134, loss_box_0: 1.8652, loss_cns_0: 0.6202, loss_yns_0: 0.1639, loss_cls_1: 1.0776, loss_box_1: 2.1662, loss_cns_1: 0.6181, loss_yns_1: 0.1697, loss_cls_2: 1.0945, loss_box_2: 2.1252, loss_cns_2: 0.6319, loss_yns_2: 0.1702, loss_cls_3: 1.1105, loss_box_3: 2.1295, loss_cns_3: 0.6341, loss_yns_3: 0.1679, loss_cls_4: 1.0818, loss_box_4: 2.1375, loss_cns_4: 0.6359, loss_yns_4: 0.1673, loss_cls_5: 1.1083, loss_box_5: 2.1722, loss_cns_5: 0.6354, loss_yns_5: 0.1705, loss_cls_dn_0: 0.3610, loss_box_dn_0: 0.8669, loss_cls_dn_1: 0.2962, loss_box_dn_1: 0.9182, loss_cls_dn_2: 0.3354, loss_box_dn_2: 0.8937, loss_cls_dn_3: 0.3228, loss_box_dn_3: 0.8958, loss_cls_dn_4: 0.3473, loss_box_dn_4: 0.9068, loss_cls_dn_5: 0.3427, loss_box_dn_5: 0.9439, loss_dense_depth: 0.9330, loss: 32.2304, grad_norm: 41.6459 -2026-01-14 21:33:06,378 - mmdet - INFO - Iter [59/17500] lr: 1.232e-04, eta: 18:25:17, time: 1.638, data_time: 0.123, memory: 49164, loss_cls_0: 0.9767, loss_box_0: 1.8916, loss_cns_0: 0.6189, loss_yns_0: 0.1611, loss_cls_1: 1.0564, loss_box_1: 2.1363, loss_cns_1: 0.6300, loss_yns_1: 0.1667, loss_cls_2: 1.0815, loss_box_2: 2.1249, loss_cns_2: 0.6405, loss_yns_2: 0.1720, loss_cls_3: 1.0923, loss_box_3: 2.1479, loss_cns_3: 0.6410, loss_yns_3: 0.1671, loss_cls_4: 1.0953, loss_box_4: 2.1804, loss_cns_4: 0.6397, loss_yns_4: 0.1667, loss_cls_5: 1.1263, loss_box_5: 2.2033, loss_cns_5: 0.6415, loss_yns_5: 0.1671, loss_cls_dn_0: 0.3730, loss_box_dn_0: 0.8706, loss_cls_dn_1: 0.3020, loss_box_dn_1: 0.9213, loss_cls_dn_2: 0.3531, loss_box_dn_2: 0.9177, loss_cls_dn_3: 0.3493, loss_box_dn_3: 0.9330, loss_cls_dn_4: 0.3607, loss_box_dn_4: 0.9623, loss_cls_dn_5: 0.3596, loss_box_dn_5: 1.0045, loss_dense_depth: 0.9983, loss: 32.6305, grad_norm: 63.3599 -2026-01-14 21:33:08,041 - mmdet - INFO - Iter [60/17500] lr: 1.236e-04, eta: 18:14:26, time: 1.577, data_time: 0.081, memory: 49164, loss_cls_0: 0.9674, loss_box_0: 1.8622, loss_cns_0: 0.6211, loss_yns_0: 0.1584, loss_cls_1: 1.1133, loss_box_1: 2.0481, loss_cns_1: 0.6380, loss_yns_1: 0.1685, loss_cls_2: 1.1326, loss_box_2: 2.0684, loss_cns_2: 0.6425, loss_yns_2: 0.1692, loss_cls_3: 1.1190, loss_box_3: 2.0618, loss_cns_3: 0.6445, loss_yns_3: 0.1653, loss_cls_4: 1.1492, loss_box_4: 2.0691, loss_cns_4: 0.6467, loss_yns_4: 0.1626, loss_cls_5: 1.1306, loss_box_5: 2.0868, loss_cns_5: 0.6442, loss_yns_5: 0.1629, loss_cls_dn_0: 0.3881, loss_box_dn_0: 0.8809, loss_cls_dn_1: 0.3083, loss_box_dn_1: 0.9330, loss_cls_dn_2: 0.3633, loss_box_dn_2: 0.9541, loss_cls_dn_3: 0.3663, loss_box_dn_3: 0.9594, loss_cls_dn_4: 0.3641, loss_box_dn_4: 0.9823, loss_cls_dn_5: 0.3747, loss_box_dn_5: 1.0235, loss_dense_depth: 0.9091, loss: 32.4394, grad_norm: 63.2280 -2026-01-14 21:33:09,741 - mmdet - INFO - Iter [61/17500] lr: 1.240e-04, eta: 18:04:55, time: 1.780, data_time: 0.201, memory: 49164, loss_cls_0: 0.9745, loss_box_0: 1.8700, loss_cns_0: 0.6153, loss_yns_0: 0.1579, loss_cls_1: 1.1111, loss_box_1: 2.1090, loss_cns_1: 0.6261, loss_yns_1: 0.1661, loss_cls_2: 1.1181, loss_box_2: 2.1068, loss_cns_2: 0.6363, loss_yns_2: 0.1650, loss_cls_3: 1.0974, loss_box_3: 2.0770, loss_cns_3: 0.6411, loss_yns_3: 0.1661, loss_cls_4: 1.0925, loss_box_4: 2.1149, loss_cns_4: 0.6401, loss_yns_4: 0.1614, loss_cls_5: 1.0917, loss_box_5: 2.1281, loss_cns_5: 0.6367, loss_yns_5: 0.1653, loss_cls_dn_0: 0.3952, loss_box_dn_0: 0.8752, loss_cls_dn_1: 0.3153, loss_box_dn_1: 0.9839, loss_cls_dn_2: 0.3647, loss_box_dn_2: 1.0010, loss_cls_dn_3: 0.3657, loss_box_dn_3: 0.9925, loss_cls_dn_4: 0.3503, loss_box_dn_4: 1.0190, loss_cls_dn_5: 0.3761, loss_box_dn_5: 1.0498, loss_dense_depth: 1.0314, loss: 32.7888, grad_norm: 46.6971 -2026-01-14 21:33:11,397 - mmdet - INFO - Iter [62/17500] lr: 1.244e-04, eta: 17:55:09, time: 1.661, data_time: 0.174, memory: 49164, loss_cls_0: 0.9889, loss_box_0: 1.8848, loss_cns_0: 0.6171, loss_yns_0: 0.1581, loss_cls_1: 1.0765, loss_box_1: 2.1790, loss_cns_1: 0.6290, loss_yns_1: 0.1629, loss_cls_2: 1.0901, loss_box_2: 2.1791, loss_cns_2: 0.6393, loss_yns_2: 0.1621, loss_cls_3: 1.0850, loss_box_3: 2.1560, loss_cns_3: 0.6445, loss_yns_3: 0.1642, loss_cls_4: 1.0940, loss_box_4: 2.2489, loss_cns_4: 0.6389, loss_yns_4: 0.1615, loss_cls_5: 1.1033, loss_box_5: 2.2400, loss_cns_5: 0.6371, loss_yns_5: 0.1643, loss_cls_dn_0: 0.3880, loss_box_dn_0: 0.8727, loss_cls_dn_1: 0.3268, loss_box_dn_1: 0.9568, loss_cls_dn_2: 0.3722, loss_box_dn_2: 0.9711, loss_cls_dn_3: 0.3669, loss_box_dn_3: 0.9621, loss_cls_dn_4: 0.3453, loss_box_dn_4: 1.0093, loss_cls_dn_5: 0.3809, loss_box_dn_5: 1.0231, loss_dense_depth: 0.9335, loss: 33.0136, grad_norm: 64.5209 -2026-01-14 21:33:12,941 - mmdet - INFO - Iter [63/17500] lr: 1.248e-04, eta: 17:45:09, time: 1.544, data_time: 0.070, memory: 49164, loss_cls_0: 1.0016, loss_box_0: 1.8936, loss_cns_0: 0.6162, loss_yns_0: 0.1611, loss_cls_1: 1.0582, loss_box_1: 2.2150, loss_cns_1: 0.6287, loss_yns_1: 0.1624, loss_cls_2: 1.0765, loss_box_2: 2.2001, loss_cns_2: 0.6399, loss_yns_2: 0.1622, loss_cls_3: 1.1198, loss_box_3: 2.1814, loss_cns_3: 0.6470, loss_yns_3: 0.1627, loss_cls_4: 1.1616, loss_box_4: 2.2464, loss_cns_4: 0.6437, loss_yns_4: 0.1617, loss_cls_5: 1.1528, loss_box_5: 2.2132, loss_cns_5: 0.6464, loss_yns_5: 0.1635, loss_cls_dn_0: 0.3726, loss_box_dn_0: 0.8733, loss_cls_dn_1: 0.3086, loss_box_dn_1: 0.9610, loss_cls_dn_2: 0.3413, loss_box_dn_2: 0.9511, loss_cls_dn_3: 0.3374, loss_box_dn_3: 0.9421, loss_cls_dn_4: 0.3196, loss_box_dn_4: 0.9755, loss_cls_dn_5: 0.3472, loss_box_dn_5: 0.9757, loss_dense_depth: 0.9912, loss: 33.0122, grad_norm: 63.9347 -2026-01-14 21:33:14,535 - mmdet - INFO - Iter [64/17500] lr: 1.252e-04, eta: 17:35:32, time: 1.564, data_time: 0.077, memory: 49164, loss_cls_0: 0.9943, loss_box_0: 1.9151, loss_cns_0: 0.6154, loss_yns_0: 0.1606, loss_cls_1: 1.0713, loss_box_1: 2.2594, loss_cns_1: 0.6188, loss_yns_1: 0.1631, loss_cls_2: 1.0945, loss_box_2: 2.2055, loss_cns_2: 0.6350, loss_yns_2: 0.1646, loss_cls_3: 1.0992, loss_box_3: 2.2000, loss_cns_3: 0.6439, loss_yns_3: 0.1618, loss_cls_4: 1.1285, loss_box_4: 2.2008, loss_cns_4: 0.6414, loss_yns_4: 0.1621, loss_cls_5: 1.1349, loss_box_5: 2.2202, loss_cns_5: 0.6440, loss_yns_5: 0.1620, loss_cls_dn_0: 0.3549, loss_box_dn_0: 0.8806, loss_cls_dn_1: 0.2815, loss_box_dn_1: 0.9657, loss_cls_dn_2: 0.3015, loss_box_dn_2: 0.9213, loss_cls_dn_3: 0.3145, loss_box_dn_3: 0.9168, loss_cls_dn_4: 0.3028, loss_box_dn_4: 0.9255, loss_cls_dn_5: 0.3184, loss_box_dn_5: 0.9378, loss_dense_depth: 0.9504, loss: 32.6684, grad_norm: 44.4620 -2026-01-14 21:33:16,142 - mmdet - INFO - Iter [65/17500] lr: 1.256e-04, eta: 17:26:34, time: 1.637, data_time: 0.109, memory: 49164, loss_cls_0: 0.9771, loss_box_0: 1.8935, loss_cns_0: 0.6190, loss_yns_0: 0.1615, loss_cls_1: 1.0477, loss_box_1: 2.2064, loss_cns_1: 0.6189, loss_yns_1: 0.1652, loss_cls_2: 1.1187, loss_box_2: 2.1688, loss_cns_2: 0.6336, loss_yns_2: 0.1672, loss_cls_3: 1.0758, loss_box_3: 2.1801, loss_cns_3: 0.6409, loss_yns_3: 0.1655, loss_cls_4: 1.0778, loss_box_4: 2.1600, loss_cns_4: 0.6419, loss_yns_4: 0.1651, loss_cls_5: 1.0856, loss_box_5: 2.1899, loss_cns_5: 0.6430, loss_yns_5: 0.1624, loss_cls_dn_0: 0.3615, loss_box_dn_0: 0.8740, loss_cls_dn_1: 0.2625, loss_box_dn_1: 0.9810, loss_cls_dn_2: 0.2790, loss_box_dn_2: 0.9344, loss_cls_dn_3: 0.3069, loss_box_dn_3: 0.9389, loss_cls_dn_4: 0.3043, loss_box_dn_4: 0.9424, loss_cls_dn_5: 0.3170, loss_box_dn_5: 0.9599, loss_dense_depth: 0.9033, loss: 32.3305, grad_norm: 57.9181 -2026-01-14 21:33:17,722 - mmdet - INFO - Iter [66/17500] lr: 1.260e-04, eta: 17:17:36, time: 1.581, data_time: 0.072, memory: 49164, loss_cls_0: 0.9558, loss_box_0: 1.8225, loss_cns_0: 0.6237, loss_yns_0: 0.1617, loss_cls_1: 1.0362, loss_box_1: 2.1218, loss_cns_1: 0.6289, loss_yns_1: 0.1658, loss_cls_2: 1.1373, loss_box_2: 2.1330, loss_cns_2: 0.6364, loss_yns_2: 0.1641, loss_cls_3: 1.1015, loss_box_3: 2.1440, loss_cns_3: 0.6440, loss_yns_3: 0.1669, loss_cls_4: 1.0817, loss_box_4: 2.1399, loss_cns_4: 0.6478, loss_yns_4: 0.1671, loss_cls_5: 1.0943, loss_box_5: 2.1471, loss_cns_5: 0.6476, loss_yns_5: 0.1656, loss_cls_dn_0: 0.3635, loss_box_dn_0: 0.8460, loss_cls_dn_1: 0.2458, loss_box_dn_1: 0.9829, loss_cls_dn_2: 0.2656, loss_box_dn_2: 0.9562, loss_cls_dn_3: 0.2892, loss_box_dn_3: 0.9627, loss_cls_dn_4: 0.3058, loss_box_dn_4: 0.9753, loss_cls_dn_5: 0.3195, loss_box_dn_5: 0.9874, loss_dense_depth: 0.9475, loss: 32.1819, grad_norm: 57.8413 -2026-01-14 21:33:19,308 - mmdet - INFO - Iter [67/17500] lr: 1.264e-04, eta: 17:08:56, time: 1.586, data_time: 0.072, memory: 49164, loss_cls_0: 0.9633, loss_box_0: 1.8537, loss_cns_0: 0.6175, loss_yns_0: 0.1610, loss_cls_1: 1.0409, loss_box_1: 2.1578, loss_cns_1: 0.6269, loss_yns_1: 0.1645, loss_cls_2: 1.0733, loss_box_2: 2.1516, loss_cns_2: 0.6349, loss_yns_2: 0.1630, loss_cls_3: 1.1096, loss_box_3: 2.1597, loss_cns_3: 0.6418, loss_yns_3: 0.1671, loss_cls_4: 1.0823, loss_box_4: 2.1495, loss_cns_4: 0.6465, loss_yns_4: 0.1653, loss_cls_5: 1.0938, loss_box_5: 2.1791, loss_cns_5: 0.6460, loss_yns_5: 0.1653, loss_cls_dn_0: 0.3807, loss_box_dn_0: 0.8492, loss_cls_dn_1: 0.2543, loss_box_dn_1: 0.9646, loss_cls_dn_2: 0.2851, loss_box_dn_2: 0.9518, loss_cls_dn_3: 0.2934, loss_box_dn_3: 0.9667, loss_cls_dn_4: 0.3262, loss_box_dn_4: 0.9851, loss_cls_dn_5: 0.3444, loss_box_dn_5: 1.0142, loss_dense_depth: 0.8947, loss: 32.3247, grad_norm: 58.9397 -2026-01-14 21:33:20,907 - mmdet - INFO - Iter [68/17500] lr: 1.268e-04, eta: 17:00:35, time: 1.599, data_time: 0.074, memory: 49164, loss_cls_0: 0.9606, loss_box_0: 1.8843, loss_cns_0: 0.6198, loss_yns_0: 0.1610, loss_cls_1: 1.0442, loss_box_1: 2.1291, loss_cns_1: 0.6252, loss_yns_1: 0.1636, loss_cls_2: 1.0633, loss_box_2: 2.0632, loss_cns_2: 0.6416, loss_yns_2: 0.1644, loss_cls_3: 1.0854, loss_box_3: 2.0844, loss_cns_3: 0.6465, loss_yns_3: 0.1653, loss_cls_4: 1.0839, loss_box_4: 2.0672, loss_cns_4: 0.6471, loss_yns_4: 0.1643, loss_cls_5: 1.0823, loss_box_5: 2.1118, loss_cns_5: 0.6442, loss_yns_5: 0.1659, loss_cls_dn_0: 0.3859, loss_box_dn_0: 0.8607, loss_cls_dn_1: 0.2698, loss_box_dn_1: 0.9514, loss_cls_dn_2: 0.3208, loss_box_dn_2: 0.9352, loss_cls_dn_3: 0.3157, loss_box_dn_3: 0.9648, loss_cls_dn_4: 0.3546, loss_box_dn_4: 0.9860, loss_cls_dn_5: 0.3698, loss_box_dn_5: 1.0298, loss_dense_depth: 0.9195, loss: 32.1326, grad_norm: 53.9893 -2026-01-14 21:33:22,462 - mmdet - INFO - Iter [69/17500] lr: 1.272e-04, eta: 16:52:16, time: 1.554, data_time: 0.074, memory: 49164, loss_cls_0: 0.9351, loss_box_0: 1.8721, loss_cns_0: 0.6265, loss_yns_0: 0.1588, loss_cls_1: 1.0306, loss_box_1: 2.1013, loss_cns_1: 0.6307, loss_yns_1: 0.1595, loss_cls_2: 1.0705, loss_box_2: 2.0456, loss_cns_2: 0.6475, loss_yns_2: 0.1679, loss_cls_3: 1.0798, loss_box_3: 2.0543, loss_cns_3: 0.6514, loss_yns_3: 0.1617, loss_cls_4: 1.0786, loss_box_4: 2.0767, loss_cns_4: 0.6486, loss_yns_4: 0.1616, loss_cls_5: 1.0692, loss_box_5: 2.0737, loss_cns_5: 0.6469, loss_yns_5: 0.1623, loss_cls_dn_0: 0.3641, loss_box_dn_0: 0.8529, loss_cls_dn_1: 0.2675, loss_box_dn_1: 0.9574, loss_cls_dn_2: 0.3344, loss_box_dn_2: 0.9447, loss_cls_dn_3: 0.3326, loss_box_dn_3: 0.9788, loss_cls_dn_4: 0.3555, loss_box_dn_4: 1.0158, loss_cls_dn_5: 0.3652, loss_box_dn_5: 1.0423, loss_dense_depth: 0.8416, loss: 31.9635, grad_norm: 58.4570 -2026-01-14 21:33:24,042 - mmdet - INFO - Iter [70/17500] lr: 1.276e-04, eta: 16:44:19, time: 1.581, data_time: 0.076, memory: 49164, loss_cls_0: 0.9505, loss_box_0: 1.8767, loss_cns_0: 0.6243, loss_yns_0: 0.1592, loss_cls_1: 1.0134, loss_box_1: 2.1066, loss_cns_1: 0.6235, loss_yns_1: 0.1594, loss_cls_2: 1.0630, loss_box_2: 2.0404, loss_cns_2: 0.6420, loss_yns_2: 0.1639, loss_cls_3: 1.0777, loss_box_3: 2.0073, loss_cns_3: 0.6482, loss_yns_3: 0.1601, loss_cls_4: 1.0712, loss_box_4: 2.0475, loss_cns_4: 0.6448, loss_yns_4: 0.1588, loss_cls_5: 1.0838, loss_box_5: 2.0091, loss_cns_5: 0.6481, loss_yns_5: 0.1598, loss_cls_dn_0: 0.3580, loss_box_dn_0: 0.8536, loss_cls_dn_1: 0.2746, loss_box_dn_1: 0.9601, loss_cls_dn_2: 0.3440, loss_box_dn_2: 0.9473, loss_cls_dn_3: 0.3460, loss_box_dn_3: 0.9648, loss_cls_dn_4: 0.3449, loss_box_dn_4: 1.0081, loss_cls_dn_5: 0.3567, loss_box_dn_5: 1.0162, loss_dense_depth: 0.9001, loss: 31.8137, grad_norm: 70.1516 -2026-01-14 21:33:25,611 - mmdet - INFO - Iter [71/17500] lr: 1.280e-04, eta: 16:36:32, time: 1.568, data_time: 0.073, memory: 49164, loss_cls_0: 0.9604, loss_box_0: 1.8468, loss_cns_0: 0.6243, loss_yns_0: 0.1578, loss_cls_1: 1.0070, loss_box_1: 2.1297, loss_cns_1: 0.6189, loss_yns_1: 0.1599, loss_cls_2: 1.0413, loss_box_2: 2.0274, loss_cns_2: 0.6421, loss_yns_2: 0.1590, loss_cls_3: 1.0737, loss_box_3: 2.0098, loss_cns_3: 0.6488, loss_yns_3: 0.1605, loss_cls_4: 1.1037, loss_box_4: 2.0068, loss_cns_4: 0.6485, loss_yns_4: 0.1563, loss_cls_5: 1.1413, loss_box_5: 2.0018, loss_cns_5: 0.6484, loss_yns_5: 0.1597, loss_cls_dn_0: 0.3376, loss_box_dn_0: 0.8535, loss_cls_dn_1: 0.2593, loss_box_dn_1: 0.9735, loss_cls_dn_2: 0.3167, loss_box_dn_2: 0.9355, loss_cls_dn_3: 0.3150, loss_box_dn_3: 0.9372, loss_cls_dn_4: 0.3004, loss_box_dn_4: 0.9518, loss_cls_dn_5: 0.3163, loss_box_dn_5: 0.9613, loss_dense_depth: 0.8732, loss: 31.4653, grad_norm: 46.7049 -2026-01-14 21:33:27,220 - mmdet - INFO - Iter [72/17500] lr: 1.284e-04, eta: 16:29:07, time: 1.610, data_time: 0.073, memory: 49164, loss_cls_0: 0.9623, loss_box_0: 1.8160, loss_cns_0: 0.6225, loss_yns_0: 0.1602, loss_cls_1: 1.0561, loss_box_1: 2.0295, loss_cns_1: 0.6264, loss_yns_1: 0.1602, loss_cls_2: 1.0739, loss_box_2: 1.9577, loss_cns_2: 0.6430, loss_yns_2: 0.1618, loss_cls_3: 1.0663, loss_box_3: 1.9852, loss_cns_3: 0.6507, loss_yns_3: 0.1624, loss_cls_4: 1.1134, loss_box_4: 1.9647, loss_cns_4: 0.6525, loss_yns_4: 0.1578, loss_cls_5: 1.0964, loss_box_5: 1.9803, loss_cns_5: 0.6477, loss_yns_5: 0.1628, loss_cls_dn_0: 0.3471, loss_box_dn_0: 0.8508, loss_cls_dn_1: 0.2574, loss_box_dn_1: 0.8568, loss_cls_dn_2: 0.3003, loss_box_dn_2: 0.8180, loss_cls_dn_3: 0.3091, loss_box_dn_3: 0.8306, loss_cls_dn_4: 0.2979, loss_box_dn_4: 0.8290, loss_cls_dn_5: 0.3258, loss_box_dn_5: 0.8391, loss_dense_depth: 0.9046, loss: 30.6765, grad_norm: 52.2715 -2026-01-14 21:33:28,779 - mmdet - INFO - Iter [73/17500] lr: 1.288e-04, eta: 16:21:43, time: 1.559, data_time: 0.071, memory: 49164, loss_cls_0: 0.9583, loss_box_0: 1.8396, loss_cns_0: 0.6157, loss_yns_0: 0.1588, loss_cls_1: 1.0516, loss_box_1: 2.0500, loss_cns_1: 0.6244, loss_yns_1: 0.1605, loss_cls_2: 1.1107, loss_box_2: 2.0033, loss_cns_2: 0.6419, loss_yns_2: 0.1617, loss_cls_3: 1.0845, loss_box_3: 2.0158, loss_cns_3: 0.6471, loss_yns_3: 0.1627, loss_cls_4: 1.0797, loss_box_4: 2.0131, loss_cns_4: 0.6490, loss_yns_4: 0.1614, loss_cls_5: 1.0760, loss_box_5: 2.0253, loss_cns_5: 0.6439, loss_yns_5: 0.1674, loss_cls_dn_0: 0.3669, loss_box_dn_0: 0.8607, loss_cls_dn_1: 0.2533, loss_box_dn_1: 0.8569, loss_cls_dn_2: 0.2897, loss_box_dn_2: 0.8278, loss_cls_dn_3: 0.3147, loss_box_dn_3: 0.8351, loss_cls_dn_4: 0.3153, loss_box_dn_4: 0.8384, loss_cls_dn_5: 0.3436, loss_box_dn_5: 0.8531, loss_dense_depth: 0.8698, loss: 30.9280, grad_norm: 59.1842 -2026-01-14 21:33:30,382 - mmdet - INFO - Iter [74/17500] lr: 1.292e-04, eta: 16:14:35, time: 1.578, data_time: 0.078, memory: 49164, loss_cls_0: 0.9446, loss_box_0: 1.8249, loss_cns_0: 0.6175, loss_yns_0: 0.1596, loss_cls_1: 1.0061, loss_box_1: 2.0737, loss_cns_1: 0.6237, loss_yns_1: 0.1624, loss_cls_2: 1.0654, loss_box_2: 2.0235, loss_cns_2: 0.6441, loss_yns_2: 0.1674, loss_cls_3: 1.0791, loss_box_3: 2.0175, loss_cns_3: 0.6501, loss_yns_3: 0.1651, loss_cls_4: 1.0646, loss_box_4: 2.0030, loss_cns_4: 0.6501, loss_yns_4: 0.1634, loss_cls_5: 1.0780, loss_box_5: 2.0470, loss_cns_5: 0.6469, loss_yns_5: 0.1660, loss_cls_dn_0: 0.3777, loss_box_dn_0: 0.8545, loss_cls_dn_1: 0.2488, loss_box_dn_1: 0.8719, loss_cls_dn_2: 0.2826, loss_box_dn_2: 0.8436, loss_cls_dn_3: 0.3206, loss_box_dn_3: 0.8484, loss_cls_dn_4: 0.3274, loss_box_dn_4: 0.8535, loss_cls_dn_5: 0.3545, loss_box_dn_5: 0.8877, loss_dense_depth: 0.8872, loss: 31.0020, grad_norm: 50.5578 -2026-01-14 21:33:31,959 - mmdet - INFO - Iter [75/17500] lr: 1.296e-04, eta: 16:07:44, time: 1.601, data_time: 0.092, memory: 49164, loss_cls_0: 0.9451, loss_box_0: 1.8463, loss_cns_0: 0.6153, loss_yns_0: 0.1618, loss_cls_1: 1.0311, loss_box_1: 2.0740, loss_cns_1: 0.6268, loss_yns_1: 0.1655, loss_cls_2: 1.0936, loss_box_2: 2.0217, loss_cns_2: 0.6422, loss_yns_2: 0.1682, loss_cls_3: 1.0521, loss_box_3: 2.0220, loss_cns_3: 0.6501, loss_yns_3: 0.1676, loss_cls_4: 1.0654, loss_box_4: 2.0204, loss_cns_4: 0.6481, loss_yns_4: 0.1637, loss_cls_5: 1.0688, loss_box_5: 2.0662, loss_cns_5: 0.6492, loss_yns_5: 0.1633, loss_cls_dn_0: 0.3681, loss_box_dn_0: 0.8432, loss_cls_dn_1: 0.2391, loss_box_dn_1: 0.8744, loss_cls_dn_2: 0.2762, loss_box_dn_2: 0.8634, loss_cls_dn_3: 0.3006, loss_box_dn_3: 0.8739, loss_cls_dn_4: 0.3107, loss_box_dn_4: 0.8929, loss_cls_dn_5: 0.3422, loss_box_dn_5: 0.9314, loss_dense_depth: 0.8533, loss: 31.0978, grad_norm: 39.2093 -2026-01-14 21:33:33,540 - mmdet - INFO - Iter [76/17500] lr: 1.300e-04, eta: 16:01:00, time: 1.582, data_time: 0.075, memory: 49164, loss_cls_0: 0.9711, loss_box_0: 1.8466, loss_cns_0: 0.6208, loss_yns_0: 0.1633, loss_cls_1: 1.0857, loss_box_1: 2.1502, loss_cns_1: 0.6274, loss_yns_1: 0.1644, loss_cls_2: 1.1024, loss_box_2: 2.1126, loss_cns_2: 0.6429, loss_yns_2: 0.1648, loss_cls_3: 1.0957, loss_box_3: 2.0863, loss_cns_3: 0.6533, loss_yns_3: 0.1722, loss_cls_4: 1.1113, loss_box_4: 2.1241, loss_cns_4: 0.6522, loss_yns_4: 0.1655, loss_cls_5: 1.0844, loss_box_5: 2.1081, loss_cns_5: 0.6507, loss_yns_5: 0.1709, loss_cls_dn_0: 0.3475, loss_box_dn_0: 0.8455, loss_cls_dn_1: 0.2412, loss_box_dn_1: 0.9231, loss_cls_dn_2: 0.2825, loss_box_dn_2: 0.9244, loss_cls_dn_3: 0.2842, loss_box_dn_3: 0.9279, loss_cls_dn_4: 0.3004, loss_box_dn_4: 0.9727, loss_cls_dn_5: 0.3240, loss_box_dn_5: 0.9942, loss_dense_depth: 0.9084, loss: 32.0027, grad_norm: 60.3147 -2026-01-14 21:33:35,122 - mmdet - INFO - Iter [77/17500] lr: 1.304e-04, eta: 15:54:26, time: 1.582, data_time: 0.073, memory: 49164, loss_cls_0: 0.9757, loss_box_0: 1.8742, loss_cns_0: 0.6176, loss_yns_0: 0.1616, loss_cls_1: 1.0754, loss_box_1: 2.1739, loss_cns_1: 0.6329, loss_yns_1: 0.1616, loss_cls_2: 1.0888, loss_box_2: 2.1373, loss_cns_2: 0.6415, loss_yns_2: 0.1639, loss_cls_3: 1.0568, loss_box_3: 2.0992, loss_cns_3: 0.6508, loss_yns_3: 0.1683, loss_cls_4: 1.0640, loss_box_4: 2.1391, loss_cns_4: 0.6514, loss_yns_4: 0.1648, loss_cls_5: 1.0697, loss_box_5: 2.1224, loss_cns_5: 0.6472, loss_yns_5: 0.1728, loss_cls_dn_0: 0.3384, loss_box_dn_0: 0.8327, loss_cls_dn_1: 0.2450, loss_box_dn_1: 0.9442, loss_cls_dn_2: 0.2956, loss_box_dn_2: 0.9362, loss_cls_dn_3: 0.2859, loss_box_dn_3: 0.9247, loss_cls_dn_4: 0.3170, loss_box_dn_4: 0.9694, loss_cls_dn_5: 0.3226, loss_box_dn_5: 0.9939, loss_dense_depth: 0.8857, loss: 32.0022, grad_norm: 48.4579 -2026-01-14 21:33:36,737 - mmdet - INFO - Iter [78/17500] lr: 1.308e-04, eta: 15:48:09, time: 1.615, data_time: 0.074, memory: 49164, loss_cls_0: 0.9699, loss_box_0: 1.9007, loss_cns_0: 0.6197, loss_yns_0: 0.1619, loss_cls_1: 1.0464, loss_box_1: 2.1898, loss_cns_1: 0.6236, loss_yns_1: 0.1627, loss_cls_2: 1.1092, loss_box_2: 2.1087, loss_cns_2: 0.6352, loss_yns_2: 0.1752, loss_cls_3: 1.0585, loss_box_3: 2.1112, loss_cns_3: 0.6437, loss_yns_3: 0.1640, loss_cls_4: 1.0708, loss_box_4: 2.0696, loss_cns_4: 0.6486, loss_yns_4: 0.1640, loss_cls_5: 1.0598, loss_box_5: 2.0629, loss_cns_5: 0.6491, loss_yns_5: 0.1658, loss_cls_dn_0: 0.3491, loss_box_dn_0: 0.8330, loss_cls_dn_1: 0.2338, loss_box_dn_1: 0.9677, loss_cls_dn_2: 0.2988, loss_box_dn_2: 0.9253, loss_cls_dn_3: 0.2860, loss_box_dn_3: 0.9095, loss_cls_dn_4: 0.3196, loss_box_dn_4: 0.9155, loss_cls_dn_5: 0.3121, loss_box_dn_5: 0.9369, loss_dense_depth: 0.9109, loss: 31.7693, grad_norm: 38.9887 -2026-01-14 21:33:38,303 - mmdet - INFO - Iter [79/17500] lr: 1.312e-04, eta: 15:41:51, time: 1.566, data_time: 0.074, memory: 49164, loss_cls_0: 0.9420, loss_box_0: 1.8707, loss_cns_0: 0.6240, loss_yns_0: 0.1599, loss_cls_1: 1.0087, loss_box_1: 2.0961, loss_cns_1: 0.6342, loss_yns_1: 0.1640, loss_cls_2: 1.0561, loss_box_2: 2.0539, loss_cns_2: 0.6454, loss_yns_2: 0.1732, loss_cls_3: 1.0638, loss_box_3: 2.1108, loss_cns_3: 0.6475, loss_yns_3: 0.1675, loss_cls_4: 1.0847, loss_box_4: 2.0622, loss_cns_4: 0.6529, loss_yns_4: 0.1663, loss_cls_5: 1.1050, loss_box_5: 2.0234, loss_cns_5: 0.6538, loss_yns_5: 0.1618, loss_cls_dn_0: 0.3423, loss_box_dn_0: 0.8390, loss_cls_dn_1: 0.2176, loss_box_dn_1: 0.8929, loss_cls_dn_2: 0.2760, loss_box_dn_2: 0.8674, loss_cls_dn_3: 0.2643, loss_box_dn_3: 0.8718, loss_cls_dn_4: 0.2832, loss_box_dn_4: 0.8650, loss_cls_dn_5: 0.2908, loss_box_dn_5: 0.8669, loss_dense_depth: 0.8218, loss: 31.0266, grad_norm: 53.3586 -2026-01-14 21:33:39,893 - mmdet - INFO - Iter [80/17500] lr: 1.316e-04, eta: 15:35:48, time: 1.589, data_time: 0.083, memory: 49164, loss_cls_0: 0.9433, loss_box_0: 1.8546, loss_cns_0: 0.6245, loss_yns_0: 0.1589, loss_cls_1: 1.0452, loss_box_1: 2.0860, loss_cns_1: 0.6318, loss_yns_1: 0.1627, loss_cls_2: 1.0423, loss_box_2: 2.0310, loss_cns_2: 0.6458, loss_yns_2: 0.1653, loss_cls_3: 1.0880, loss_box_3: 2.0606, loss_cns_3: 0.6521, loss_yns_3: 0.1715, loss_cls_4: 1.1257, loss_box_4: 2.0261, loss_cns_4: 0.6555, loss_yns_4: 0.1636, loss_cls_5: 1.0886, loss_box_5: 2.0014, loss_cns_5: 0.6519, loss_yns_5: 0.1629, loss_cls_dn_0: 0.3490, loss_box_dn_0: 0.8314, loss_cls_dn_1: 0.2136, loss_box_dn_1: 0.8735, loss_cls_dn_2: 0.2526, loss_box_dn_2: 0.8461, loss_cls_dn_3: 0.2466, loss_box_dn_3: 0.8502, loss_cls_dn_4: 0.2587, loss_box_dn_4: 0.8504, loss_cls_dn_5: 0.2854, loss_box_dn_5: 0.8476, loss_dense_depth: 0.8589, loss: 30.8033, grad_norm: 44.8805 -2026-01-14 21:33:41,587 - mmdet - INFO - Iter [81/17500] lr: 1.320e-04, eta: 15:30:16, time: 1.694, data_time: 0.114, memory: 49164, loss_cls_0: 0.9589, loss_box_0: 1.8687, loss_cns_0: 0.6161, loss_yns_0: 0.1602, loss_cls_1: 1.0194, loss_box_1: 2.0471, loss_cns_1: 0.6248, loss_yns_1: 0.1589, loss_cls_2: 1.0734, loss_box_2: 1.9899, loss_cns_2: 0.6456, loss_yns_2: 0.1629, loss_cls_3: 1.0533, loss_box_3: 1.9893, loss_cns_3: 0.6514, loss_yns_3: 0.1647, loss_cls_4: 1.0483, loss_box_4: 1.9674, loss_cns_4: 0.6523, loss_yns_4: 0.1599, loss_cls_5: 1.0567, loss_box_5: 1.9964, loss_cns_5: 0.6510, loss_yns_5: 0.1615, loss_cls_dn_0: 0.3607, loss_box_dn_0: 0.8350, loss_cls_dn_1: 0.2116, loss_box_dn_1: 0.8785, loss_cls_dn_2: 0.2390, loss_box_dn_2: 0.8554, loss_cls_dn_3: 0.2476, loss_box_dn_3: 0.8593, loss_cls_dn_4: 0.2699, loss_box_dn_4: 0.8674, loss_cls_dn_5: 0.3165, loss_box_dn_5: 0.8833, loss_dense_depth: 0.8533, loss: 30.5556, grad_norm: 40.4857 -2026-01-14 21:33:43,267 - mmdet - INFO - Iter [82/17500] lr: 1.324e-04, eta: 15:24:43, time: 1.652, data_time: 0.170, memory: 49164, loss_cls_0: 0.9598, loss_box_0: 1.8685, loss_cns_0: 0.6226, loss_yns_0: 0.1562, loss_cls_1: 1.0162, loss_box_1: 2.0110, loss_cns_1: 0.6021, loss_yns_1: 0.1522, loss_cls_2: 1.0632, loss_box_2: 2.0464, loss_cns_2: 0.6462, loss_yns_2: 0.1662, loss_cls_3: 1.0696, loss_box_3: 2.0407, loss_cns_3: 0.6486, loss_yns_3: 0.1602, loss_cls_4: 1.0709, loss_box_4: 2.0460, loss_cns_4: 0.6463, loss_yns_4: 0.1600, loss_cls_5: 1.0911, loss_box_5: 2.0735, loss_cns_5: 0.6455, loss_yns_5: 0.1581, loss_cls_dn_0: 0.3545, loss_box_dn_0: 0.8309, loss_cls_dn_1: 0.2107, loss_box_dn_1: 0.8826, loss_cls_dn_2: 0.2356, loss_box_dn_2: 0.8711, loss_cls_dn_3: 0.2579, loss_box_dn_3: 0.8787, loss_cls_dn_4: 0.2711, loss_box_dn_4: 0.8943, loss_cls_dn_5: 0.3246, loss_box_dn_5: 0.9151, loss_dense_depth: 0.8930, loss: 30.9409, grad_norm: 47.4329 -2026-01-14 21:33:44,810 - mmdet - INFO - Iter [83/17500] lr: 1.328e-04, eta: 15:19:00, time: 1.570, data_time: 0.093, memory: 49164, loss_cls_0: 0.9630, loss_box_0: 1.8250, loss_cns_0: 0.6281, loss_yns_0: 0.1589, loss_cls_1: 1.0092, loss_box_1: 1.9328, loss_cns_1: 0.5949, loss_yns_1: 0.1574, loss_cls_2: 1.0535, loss_box_2: 1.9900, loss_cns_2: 0.6379, loss_yns_2: 0.1673, loss_cls_3: 1.0700, loss_box_3: 1.9844, loss_cns_3: 0.6434, loss_yns_3: 0.1624, loss_cls_4: 1.0971, loss_box_4: 1.9848, loss_cns_4: 0.6419, loss_yns_4: 0.1619, loss_cls_5: 1.0818, loss_box_5: 1.9932, loss_cns_5: 0.6399, loss_yns_5: 0.1610, loss_cls_dn_0: 0.3368, loss_box_dn_0: 0.8250, loss_cls_dn_1: 0.2144, loss_box_dn_1: 0.8750, loss_cls_dn_2: 0.2371, loss_box_dn_2: 0.8671, loss_cls_dn_3: 0.2527, loss_box_dn_3: 0.8776, loss_cls_dn_4: 0.2530, loss_box_dn_4: 0.8877, loss_cls_dn_5: 0.2952, loss_box_dn_5: 0.9050, loss_dense_depth: 0.8698, loss: 30.4359, grad_norm: 47.8320 -2026-01-14 21:33:46,408 - mmdet - INFO - Iter [84/17500] lr: 1.332e-04, eta: 15:13:32, time: 1.598, data_time: 0.075, memory: 49164, loss_cls_0: 0.9250, loss_box_0: 1.8193, loss_cns_0: 0.6286, loss_yns_0: 0.1573, loss_cls_1: 0.9622, loss_box_1: 1.9785, loss_cns_1: 0.6150, loss_yns_1: 0.1579, loss_cls_2: 1.0124, loss_box_2: 1.9505, loss_cns_2: 0.6454, loss_yns_2: 0.1611, loss_cls_3: 1.0266, loss_box_3: 1.9565, loss_cns_3: 0.6511, loss_yns_3: 0.1610, loss_cls_4: 1.0497, loss_box_4: 1.9524, loss_cns_4: 0.6516, loss_yns_4: 0.1593, loss_cls_5: 1.0772, loss_box_5: 1.9553, loss_cns_5: 0.6496, loss_yns_5: 0.1586, loss_cls_dn_0: 0.3181, loss_box_dn_0: 0.8185, loss_cls_dn_1: 0.2142, loss_box_dn_1: 0.8474, loss_cls_dn_2: 0.2369, loss_box_dn_2: 0.8235, loss_cls_dn_3: 0.2483, loss_box_dn_3: 0.8432, loss_cls_dn_4: 0.2445, loss_box_dn_4: 0.8543, loss_cls_dn_5: 0.2719, loss_box_dn_5: 0.8656, loss_dense_depth: 0.8025, loss: 29.8514, grad_norm: 36.9244 -2026-01-14 21:33:48,070 - mmdet - INFO - Iter [85/17500] lr: 1.336e-04, eta: 15:08:25, time: 1.663, data_time: 0.074, memory: 49164, loss_cls_0: 0.9504, loss_box_0: 1.8387, loss_cns_0: 0.6238, loss_yns_0: 0.1581, loss_cls_1: 0.9920, loss_box_1: 1.9890, loss_cns_1: 0.6155, loss_yns_1: 0.1591, loss_cls_2: 1.0382, loss_box_2: 2.0033, loss_cns_2: 0.6406, loss_yns_2: 0.1606, loss_cls_3: 1.0416, loss_box_3: 2.0091, loss_cns_3: 0.6426, loss_yns_3: 0.1628, loss_cls_4: 1.0575, loss_box_4: 2.0199, loss_cns_4: 0.6422, loss_yns_4: 0.1612, loss_cls_5: 1.0537, loss_box_5: 2.0043, loss_cns_5: 0.6425, loss_yns_5: 0.1621, loss_cls_dn_0: 0.3486, loss_box_dn_0: 0.8356, loss_cls_dn_1: 0.2135, loss_box_dn_1: 0.8503, loss_cls_dn_2: 0.2442, loss_box_dn_2: 0.8266, loss_cls_dn_3: 0.2485, loss_box_dn_3: 0.8450, loss_cls_dn_4: 0.2521, loss_box_dn_4: 0.8640, loss_cls_dn_5: 0.2721, loss_box_dn_5: 0.8623, loss_dense_depth: 0.8646, loss: 30.2963, grad_norm: 56.8787 -2026-01-14 21:33:49,685 - mmdet - INFO - Iter [86/17500] lr: 1.340e-04, eta: 15:03:15, time: 1.615, data_time: 0.076, memory: 49164, loss_cls_0: 0.9245, loss_box_0: 1.8076, loss_cns_0: 0.6222, loss_yns_0: 0.1606, loss_cls_1: 0.9871, loss_box_1: 1.9558, loss_cns_1: 0.6239, loss_yns_1: 0.1625, loss_cls_2: 1.0291, loss_box_2: 1.9087, loss_cns_2: 0.6428, loss_yns_2: 0.1617, loss_cls_3: 1.0289, loss_box_3: 1.9029, loss_cns_3: 0.6475, loss_yns_3: 0.1614, loss_cls_4: 1.0749, loss_box_4: 1.9023, loss_cns_4: 0.6474, loss_yns_4: 0.1630, loss_cls_5: 1.0620, loss_box_5: 1.8778, loss_cns_5: 0.6475, loss_yns_5: 0.1642, loss_cls_dn_0: 0.3405, loss_box_dn_0: 0.8273, loss_cls_dn_1: 0.2080, loss_box_dn_1: 0.8273, loss_cls_dn_2: 0.2363, loss_box_dn_2: 0.8015, loss_cls_dn_3: 0.2377, loss_box_dn_3: 0.8015, loss_cls_dn_4: 0.2483, loss_box_dn_4: 0.8174, loss_cls_dn_5: 0.2646, loss_box_dn_5: 0.8125, loss_dense_depth: 0.8509, loss: 29.5399, grad_norm: 53.4501 -2026-01-14 21:33:51,296 - mmdet - INFO - Iter [87/17500] lr: 1.344e-04, eta: 14:58:06, time: 1.583, data_time: 0.084, memory: 49164, loss_cls_0: 0.9007, loss_box_0: 1.8033, loss_cns_0: 0.6174, loss_yns_0: 0.1584, loss_cls_1: 0.9866, loss_box_1: 2.0012, loss_cns_1: 0.6235, loss_yns_1: 0.1667, loss_cls_2: 1.0051, loss_box_2: 1.9297, loss_cns_2: 0.6436, loss_yns_2: 0.1681, loss_cls_3: 1.0392, loss_box_3: 1.9100, loss_cns_3: 0.6498, loss_yns_3: 0.1614, loss_cls_4: 1.0418, loss_box_4: 1.8915, loss_cns_4: 0.6506, loss_yns_4: 0.1648, loss_cls_5: 1.0373, loss_box_5: 1.9031, loss_cns_5: 0.6486, loss_yns_5: 0.1629, loss_cls_dn_0: 0.3265, loss_box_dn_0: 0.8205, loss_cls_dn_1: 0.2060, loss_box_dn_1: 0.8195, loss_cls_dn_2: 0.2303, loss_box_dn_2: 0.7973, loss_cls_dn_3: 0.2302, loss_box_dn_3: 0.7862, loss_cls_dn_4: 0.2373, loss_box_dn_4: 0.7966, loss_cls_dn_5: 0.2607, loss_box_dn_5: 0.8174, loss_dense_depth: 0.8398, loss: 29.4337, grad_norm: 45.7670 -2026-01-14 21:33:52,848 - mmdet - INFO - Iter [88/17500] lr: 1.348e-04, eta: 14:53:03, time: 1.580, data_time: 0.094, memory: 49164, loss_cls_0: 0.9270, loss_box_0: 1.7888, loss_cns_0: 0.6130, loss_yns_0: 0.1565, loss_cls_1: 0.9819, loss_box_1: 1.9093, loss_cns_1: 0.6320, loss_yns_1: 0.1639, loss_cls_2: 1.0365, loss_box_2: 1.8772, loss_cns_2: 0.6453, loss_yns_2: 0.1587, loss_cls_3: 1.0396, loss_box_3: 1.8700, loss_cns_3: 0.6514, loss_yns_3: 0.1608, loss_cls_4: 1.0490, loss_box_4: 1.8770, loss_cns_4: 0.6515, loss_yns_4: 0.1656, loss_cls_5: 1.0543, loss_box_5: 1.8850, loss_cns_5: 0.6517, loss_yns_5: 0.1593, loss_cls_dn_0: 0.3194, loss_box_dn_0: 0.8192, loss_cls_dn_1: 0.2089, loss_box_dn_1: 0.8008, loss_cls_dn_2: 0.2335, loss_box_dn_2: 0.8011, loss_cls_dn_3: 0.2306, loss_box_dn_3: 0.7986, loss_cls_dn_4: 0.2362, loss_box_dn_4: 0.8214, loss_cls_dn_5: 0.2671, loss_box_dn_5: 0.8476, loss_dense_depth: 0.8176, loss: 29.3072, grad_norm: 54.1470 -2026-01-14 21:33:54,459 - mmdet - INFO - Iter [89/17500] lr: 1.352e-04, eta: 14:48:04, time: 1.565, data_time: 0.075, memory: 49164, loss_cls_0: 0.9328, loss_box_0: 1.8107, loss_cns_0: 0.6124, loss_yns_0: 0.1553, loss_cls_1: 1.0046, loss_box_1: 1.9146, loss_cns_1: 0.6312, loss_yns_1: 0.1579, loss_cls_2: 1.0533, loss_box_2: 1.8980, loss_cns_2: 0.6447, loss_yns_2: 0.1564, loss_cls_3: 1.0428, loss_box_3: 1.8994, loss_cns_3: 0.6480, loss_yns_3: 0.1583, loss_cls_4: 1.0392, loss_box_4: 1.9209, loss_cns_4: 0.6474, loss_yns_4: 0.1615, loss_cls_5: 1.0406, loss_box_5: 1.9222, loss_cns_5: 0.6484, loss_yns_5: 0.1588, loss_cls_dn_0: 0.3086, loss_box_dn_0: 0.8195, loss_cls_dn_1: 0.2158, loss_box_dn_1: 0.8334, loss_cls_dn_2: 0.2377, loss_box_dn_2: 0.8390, loss_cls_dn_3: 0.2307, loss_box_dn_3: 0.8365, loss_cls_dn_4: 0.2322, loss_box_dn_4: 0.8661, loss_cls_dn_5: 0.2531, loss_box_dn_5: 0.8788, loss_dense_depth: 0.8017, loss: 29.6125, grad_norm: 62.6129 -2026-01-14 21:33:56,025 - mmdet - INFO - Iter [90/17500] lr: 1.356e-04, eta: 14:43:21, time: 1.612, data_time: 0.108, memory: 49164, loss_cls_0: 0.9350, loss_box_0: 1.8059, loss_cns_0: 0.6164, loss_yns_0: 0.1550, loss_cls_1: 1.0119, loss_box_1: 1.9620, loss_cns_1: 0.6396, loss_yns_1: 0.1600, loss_cls_2: 1.0435, loss_box_2: 1.9240, loss_cns_2: 0.6520, loss_yns_2: 0.1573, loss_cls_3: 1.0321, loss_box_3: 1.9146, loss_cns_3: 0.6527, loss_yns_3: 0.1584, loss_cls_4: 1.0592, loss_box_4: 1.9256, loss_cns_4: 0.6535, loss_yns_4: 0.1587, loss_cls_5: 1.0745, loss_box_5: 1.9410, loss_cns_5: 0.6527, loss_yns_5: 0.1576, loss_cls_dn_0: 0.3138, loss_box_dn_0: 0.8301, loss_cls_dn_1: 0.2199, loss_box_dn_1: 0.8394, loss_cls_dn_2: 0.2379, loss_box_dn_2: 0.8328, loss_cls_dn_3: 0.2287, loss_box_dn_3: 0.8174, loss_cls_dn_4: 0.2296, loss_box_dn_4: 0.8371, loss_cls_dn_5: 0.2438, loss_box_dn_5: 0.8451, loss_dense_depth: 0.8442, loss: 29.7629, grad_norm: 53.8847 -2026-01-14 21:33:57,687 - mmdet - INFO - Iter [91/17500] lr: 1.360e-04, eta: 14:38:43, time: 1.609, data_time: 0.075, memory: 49164, loss_cls_0: 0.9249, loss_box_0: 1.7975, loss_cns_0: 0.6173, loss_yns_0: 0.1532, loss_cls_1: 1.0317, loss_box_1: 1.9125, loss_cns_1: 0.6464, loss_yns_1: 0.1578, loss_cls_2: 1.0218, loss_box_2: 1.8774, loss_cns_2: 0.6560, loss_yns_2: 0.1571, loss_cls_3: 1.0475, loss_box_3: 1.8881, loss_cns_3: 0.6548, loss_yns_3: 0.1571, loss_cls_4: 1.0309, loss_box_4: 1.8757, loss_cns_4: 0.6559, loss_yns_4: 0.1611, loss_cls_5: 1.0395, loss_box_5: 1.9019, loss_cns_5: 0.6533, loss_yns_5: 0.1557, loss_cls_dn_0: 0.3102, loss_box_dn_0: 0.8200, loss_cls_dn_1: 0.2182, loss_box_dn_1: 0.8083, loss_cls_dn_2: 0.2310, loss_box_dn_2: 0.7984, loss_cls_dn_3: 0.2342, loss_box_dn_3: 0.7907, loss_cls_dn_4: 0.2270, loss_box_dn_4: 0.7945, loss_cls_dn_5: 0.2438, loss_box_dn_5: 0.8139, loss_dense_depth: 0.8043, loss: 29.2694, grad_norm: 40.8866 -2026-01-14 21:33:59,286 - mmdet - INFO - Iter [92/17500] lr: 1.364e-04, eta: 14:34:20, time: 1.652, data_time: 0.146, memory: 49164, loss_cls_0: 0.9133, loss_box_0: 1.7877, loss_cns_0: 0.6188, loss_yns_0: 0.1524, loss_cls_1: 1.0135, loss_box_1: 1.9552, loss_cns_1: 0.6377, loss_yns_1: 0.1554, loss_cls_2: 1.0070, loss_box_2: 1.9233, loss_cns_2: 0.6466, loss_yns_2: 0.1560, loss_cls_3: 1.0452, loss_box_3: 1.9381, loss_cns_3: 0.6528, loss_yns_3: 0.1581, loss_cls_4: 1.0253, loss_box_4: 1.9256, loss_cns_4: 0.6532, loss_yns_4: 0.1598, loss_cls_5: 1.0440, loss_box_5: 1.9327, loss_cns_5: 0.6511, loss_yns_5: 0.1564, loss_cls_dn_0: 0.3048, loss_box_dn_0: 0.8156, loss_cls_dn_1: 0.2116, loss_box_dn_1: 0.7880, loss_cls_dn_2: 0.2182, loss_box_dn_2: 0.7781, loss_cls_dn_3: 0.2300, loss_box_dn_3: 0.7811, loss_cls_dn_4: 0.2241, loss_box_dn_4: 0.7808, loss_cls_dn_5: 0.2489, loss_box_dn_5: 0.7988, loss_dense_depth: 0.8292, loss: 29.3184, grad_norm: 51.3590 -2026-01-14 21:34:00,893 - mmdet - INFO - Iter [93/17500] lr: 1.368e-04, eta: 14:29:44, time: 1.558, data_time: 0.073, memory: 49164, loss_cls_0: 0.9192, loss_box_0: 1.7721, loss_cns_0: 0.6228, loss_yns_0: 0.1522, loss_cls_1: 1.0009, loss_box_1: 1.9009, loss_cns_1: 0.6409, loss_yns_1: 0.1541, loss_cls_2: 1.0476, loss_box_2: 1.8816, loss_cns_2: 0.6481, loss_yns_2: 0.1544, loss_cls_3: 1.0373, loss_box_3: 1.8697, loss_cns_3: 0.6532, loss_yns_3: 0.1574, loss_cls_4: 1.0189, loss_box_4: 1.8640, loss_cns_4: 0.6528, loss_yns_4: 0.1564, loss_cls_5: 1.0271, loss_box_5: 1.8834, loss_cns_5: 0.6488, loss_yns_5: 0.1545, loss_cls_dn_0: 0.2945, loss_box_dn_0: 0.8114, loss_cls_dn_1: 0.2064, loss_box_dn_1: 0.7906, loss_cls_dn_2: 0.2134, loss_box_dn_2: 0.7848, loss_cls_dn_3: 0.2162, loss_box_dn_3: 0.7814, loss_cls_dn_4: 0.2169, loss_box_dn_4: 0.7872, loss_cls_dn_5: 0.2377, loss_box_dn_5: 0.8090, loss_dense_depth: 0.7906, loss: 28.9584, grad_norm: 53.2069 -2026-01-14 21:34:02,549 - mmdet - INFO - Iter [94/17500] lr: 1.372e-04, eta: 14:25:37, time: 1.677, data_time: 0.108, memory: 49164, loss_cls_0: 0.9329, loss_box_0: 1.8174, loss_cns_0: 0.6187, loss_yns_0: 0.1524, loss_cls_1: 1.0260, loss_box_1: 1.9353, loss_cns_1: 0.6365, loss_yns_1: 0.1555, loss_cls_2: 1.0241, loss_box_2: 1.8775, loss_cns_2: 0.6463, loss_yns_2: 0.1544, loss_cls_3: 1.0242, loss_box_3: 1.8714, loss_cns_3: 0.6542, loss_yns_3: 0.1566, loss_cls_4: 1.0281, loss_box_4: 1.8516, loss_cns_4: 0.6546, loss_yns_4: 0.1554, loss_cls_5: 1.0335, loss_box_5: 1.8487, loss_cns_5: 0.6525, loss_yns_5: 0.1541, loss_cls_dn_0: 0.3015, loss_box_dn_0: 0.8122, loss_cls_dn_1: 0.2115, loss_box_dn_1: 0.8006, loss_cls_dn_2: 0.2190, loss_box_dn_2: 0.7813, loss_cls_dn_3: 0.2152, loss_box_dn_3: 0.7857, loss_cls_dn_4: 0.2168, loss_box_dn_4: 0.7908, loss_cls_dn_5: 0.2318, loss_box_dn_5: 0.8081, loss_dense_depth: 0.7913, loss: 29.0275, grad_norm: 35.2644 -2026-01-14 21:34:04,106 - mmdet - INFO - Iter [95/17500] lr: 1.376e-04, eta: 14:21:18, time: 1.585, data_time: 0.095, memory: 49164, loss_cls_0: 0.9217, loss_box_0: 1.8214, loss_cns_0: 0.6201, loss_yns_0: 0.1531, loss_cls_1: 1.0437, loss_box_1: 1.9458, loss_cns_1: 0.6380, loss_yns_1: 0.1552, loss_cls_2: 1.0341, loss_box_2: 1.8623, loss_cns_2: 0.6517, loss_yns_2: 0.1578, loss_cls_3: 1.0488, loss_box_3: 1.8790, loss_cns_3: 0.6561, loss_yns_3: 0.1585, loss_cls_4: 1.0481, loss_box_4: 1.8617, loss_cns_4: 0.6569, loss_yns_4: 0.1600, loss_cls_5: 1.0639, loss_box_5: 1.8505, loss_cns_5: 0.6581, loss_yns_5: 0.1568, loss_cls_dn_0: 0.3048, loss_box_dn_0: 0.8110, loss_cls_dn_1: 0.2089, loss_box_dn_1: 0.7951, loss_cls_dn_2: 0.2151, loss_box_dn_2: 0.7706, loss_cls_dn_3: 0.2110, loss_box_dn_3: 0.7922, loss_cls_dn_4: 0.2124, loss_box_dn_4: 0.7955, loss_cls_dn_5: 0.2253, loss_box_dn_5: 0.8070, loss_dense_depth: 0.8721, loss: 29.2242, grad_norm: 51.9529 -2026-01-14 21:34:05,778 - mmdet - INFO - Iter [96/17500] lr: 1.380e-04, eta: 14:17:13, time: 1.635, data_time: 0.081, memory: 49164, loss_cls_0: 0.9447, loss_box_0: 1.8466, loss_cns_0: 0.6203, loss_yns_0: 0.1560, loss_cls_1: 1.0516, loss_box_1: 1.9835, loss_cns_1: 0.6318, loss_yns_1: 0.1577, loss_cls_2: 1.0523, loss_box_2: 1.9306, loss_cns_2: 0.6443, loss_yns_2: 0.1593, loss_cls_3: 1.0465, loss_box_3: 1.9331, loss_cns_3: 0.6473, loss_yns_3: 0.1606, loss_cls_4: 1.0546, loss_box_4: 1.9185, loss_cns_4: 0.6494, loss_yns_4: 0.1651, loss_cls_5: 1.0675, loss_box_5: 1.9094, loss_cns_5: 0.6491, loss_yns_5: 0.1592, loss_cls_dn_0: 0.3008, loss_box_dn_0: 0.8109, loss_cls_dn_1: 0.2134, loss_box_dn_1: 0.8054, loss_cls_dn_2: 0.2199, loss_box_dn_2: 0.7839, loss_cls_dn_3: 0.2130, loss_box_dn_3: 0.8031, loss_cls_dn_4: 0.2147, loss_box_dn_4: 0.8071, loss_cls_dn_5: 0.2292, loss_box_dn_5: 0.8138, loss_dense_depth: 0.8079, loss: 29.5621, grad_norm: 49.1669 -2026-01-14 21:34:07,406 - mmdet - INFO - Iter [97/17500] lr: 1.384e-04, eta: 14:13:18, time: 1.665, data_time: 0.100, memory: 49164, loss_cls_0: 0.9799, loss_box_0: 1.8533, loss_cns_0: 0.6183, loss_yns_0: 0.1562, loss_cls_1: 1.0313, loss_box_1: 1.9495, loss_cns_1: 0.6338, loss_yns_1: 0.1587, loss_cls_2: 1.0530, loss_box_2: 1.9123, loss_cns_2: 0.6462, loss_yns_2: 0.1568, loss_cls_3: 1.0907, loss_box_3: 1.9080, loss_cns_3: 0.6483, loss_yns_3: 0.1577, loss_cls_4: 1.0827, loss_box_4: 1.9021, loss_cns_4: 0.6502, loss_yns_4: 0.1646, loss_cls_5: 1.0848, loss_box_5: 1.8966, loss_cns_5: 0.6496, loss_yns_5: 0.1577, loss_cls_dn_0: 0.2879, loss_box_dn_0: 0.8189, loss_cls_dn_1: 0.2081, loss_box_dn_1: 0.7943, loss_cls_dn_2: 0.2153, loss_box_dn_2: 0.7836, loss_cls_dn_3: 0.2119, loss_box_dn_3: 0.7927, loss_cls_dn_4: 0.2119, loss_box_dn_4: 0.8078, loss_cls_dn_5: 0.2272, loss_box_dn_5: 0.8112, loss_dense_depth: 0.8508, loss: 29.5639, grad_norm: 66.6381 -2026-01-14 21:34:09,000 - mmdet - INFO - Iter [98/17500] lr: 1.388e-04, eta: 14:09:10, time: 1.558, data_time: 0.074, memory: 49164, loss_cls_0: 0.9526, loss_box_0: 1.8270, loss_cns_0: 0.6205, loss_yns_0: 0.1553, loss_cls_1: 1.0257, loss_box_1: 1.9135, loss_cns_1: 0.6397, loss_yns_1: 0.1578, loss_cls_2: 1.0659, loss_box_2: 1.8933, loss_cns_2: 0.6544, loss_yns_2: 0.1572, loss_cls_3: 1.0519, loss_box_3: 1.9081, loss_cns_3: 0.6539, loss_yns_3: 0.1575, loss_cls_4: 1.0494, loss_box_4: 1.8970, loss_cns_4: 0.6566, loss_yns_4: 0.1605, loss_cls_5: 1.0653, loss_box_5: 1.8900, loss_cns_5: 0.6525, loss_yns_5: 0.1586, loss_cls_dn_0: 0.2844, loss_box_dn_0: 0.8221, loss_cls_dn_1: 0.2113, loss_box_dn_1: 0.8070, loss_cls_dn_2: 0.2221, loss_box_dn_2: 0.8021, loss_cls_dn_3: 0.2135, loss_box_dn_3: 0.8155, loss_cls_dn_4: 0.2176, loss_box_dn_4: 0.8331, loss_cls_dn_5: 0.2338, loss_box_dn_5: 0.8446, loss_dense_depth: 0.8285, loss: 29.5000, grad_norm: 53.6799 -2026-01-14 21:34:10,563 - mmdet - INFO - Iter [99/17500] lr: 1.392e-04, eta: 14:05:13, time: 1.598, data_time: 0.120, memory: 49164, loss_cls_0: 0.9332, loss_box_0: 1.7938, loss_cns_0: 0.6167, loss_yns_0: 0.1529, loss_cls_1: 1.0039, loss_box_1: 1.9346, loss_cns_1: 0.6324, loss_yns_1: 0.1583, loss_cls_2: 1.0332, loss_box_2: 1.8885, loss_cns_2: 0.6456, loss_yns_2: 0.1592, loss_cls_3: 1.0553, loss_box_3: 1.8896, loss_cns_3: 0.6472, loss_yns_3: 0.1596, loss_cls_4: 1.0481, loss_box_4: 1.8794, loss_cns_4: 0.6487, loss_yns_4: 0.1583, loss_cls_5: 1.0619, loss_box_5: 1.8718, loss_cns_5: 0.6472, loss_yns_5: 0.1572, loss_cls_dn_0: 0.2849, loss_box_dn_0: 0.8113, loss_cls_dn_1: 0.2037, loss_box_dn_1: 0.8255, loss_cls_dn_2: 0.2160, loss_box_dn_2: 0.8133, loss_cls_dn_3: 0.2169, loss_box_dn_3: 0.8184, loss_cls_dn_4: 0.2258, loss_box_dn_4: 0.8300, loss_cls_dn_5: 0.2489, loss_box_dn_5: 0.8524, loss_dense_depth: 0.8629, loss: 29.3865, grad_norm: 53.4640 -2026-01-14 21:34:12,164 - mmdet - INFO - Iter [100/17500] lr: 1.396e-04, eta: 14:01:21, time: 1.599, data_time: 0.084, memory: 49164, loss_cls_0: 0.9207, loss_box_0: 1.8037, loss_cns_0: 0.6163, loss_yns_0: 0.1541, loss_cls_1: 0.9850, loss_box_1: 1.9429, loss_cns_1: 0.6308, loss_yns_1: 0.1569, loss_cls_2: 1.0228, loss_box_2: 1.8939, loss_cns_2: 0.6418, loss_yns_2: 0.1590, loss_cls_3: 1.0196, loss_box_3: 1.8618, loss_cns_3: 0.6468, loss_yns_3: 0.1610, loss_cls_4: 1.0198, loss_box_4: 1.8547, loss_cns_4: 0.6497, loss_yns_4: 0.1619, loss_cls_5: 1.0332, loss_box_5: 1.8551, loss_cns_5: 0.6493, loss_yns_5: 0.1597, loss_cls_dn_0: 0.2760, loss_box_dn_0: 0.8177, loss_cls_dn_1: 0.1949, loss_box_dn_1: 0.8434, loss_cls_dn_2: 0.2071, loss_box_dn_2: 0.8221, loss_cls_dn_3: 0.2094, loss_box_dn_3: 0.8184, loss_cls_dn_4: 0.2169, loss_box_dn_4: 0.8319, loss_cls_dn_5: 0.2346, loss_box_dn_5: 0.8570, loss_dense_depth: 0.8176, loss: 29.1478, grad_norm: 50.7003 -2026-01-14 21:34:13,829 - mmdet - INFO - Iter [101/17500] lr: 1.400e-04, eta: 13:57:46, time: 1.667, data_time: 0.122, memory: 49164, loss_cls_0: 0.9195, loss_box_0: 1.7688, loss_cns_0: 0.6163, loss_yns_0: 0.1559, loss_cls_1: 1.0238, loss_box_1: 1.9405, loss_cns_1: 0.6362, loss_yns_1: 0.1594, loss_cls_2: 1.1259, loss_box_2: 1.8770, loss_cns_2: 0.6486, loss_yns_2: 0.1589, loss_cls_3: 1.0698, loss_box_3: 1.8584, loss_cns_3: 0.6536, loss_yns_3: 0.1604, loss_cls_4: 1.1004, loss_box_4: 1.8552, loss_cns_4: 0.6535, loss_yns_4: 0.1609, loss_cls_5: 1.0845, loss_box_5: 1.8594, loss_cns_5: 0.6539, loss_yns_5: 0.1615, loss_cls_dn_0: 0.2800, loss_box_dn_0: 0.8076, loss_cls_dn_1: 0.1957, loss_box_dn_1: 0.8425, loss_cls_dn_2: 0.2086, loss_box_dn_2: 0.8148, loss_cls_dn_3: 0.2072, loss_box_dn_3: 0.8123, loss_cls_dn_4: 0.2092, loss_box_dn_4: 0.8217, loss_cls_dn_5: 0.2205, loss_box_dn_5: 0.8341, loss_dense_depth: 0.8457, loss: 29.4023, grad_norm: 68.7398 -2026-01-14 21:34:15,497 - mmdet - INFO - Iter [102/17500] lr: 1.404e-04, eta: 13:54:15, time: 1.669, data_time: 0.164, memory: 49164, loss_cls_0: 0.9379, loss_box_0: 1.7862, loss_cns_0: 0.6103, loss_yns_0: 0.1573, loss_cls_1: 0.9902, loss_box_1: 1.8932, loss_cns_1: 0.6410, loss_yns_1: 0.1591, loss_cls_2: 1.0308, loss_box_2: 1.8469, loss_cns_2: 0.6508, loss_yns_2: 0.1588, loss_cls_3: 1.0325, loss_box_3: 1.8645, loss_cns_3: 0.6539, loss_yns_3: 0.1602, loss_cls_4: 1.0268, loss_box_4: 1.8635, loss_cns_4: 0.6518, loss_yns_4: 0.1606, loss_cls_5: 1.0370, loss_box_5: 1.8655, loss_cns_5: 0.6522, loss_yns_5: 0.1589, loss_cls_dn_0: 0.2834, loss_box_dn_0: 0.8106, loss_cls_dn_1: 0.1969, loss_box_dn_1: 0.7930, loss_cls_dn_2: 0.2070, loss_box_dn_2: 0.7766, loss_cls_dn_3: 0.2072, loss_box_dn_3: 0.7849, loss_cls_dn_4: 0.2097, loss_box_dn_4: 0.7977, loss_cls_dn_5: 0.2213, loss_box_dn_5: 0.8036, loss_dense_depth: 0.8299, loss: 28.9118, grad_norm: 44.7789 -2026-01-14 21:34:17,085 - mmdet - INFO - Iter [103/17500] lr: 1.408e-04, eta: 13:50:34, time: 1.586, data_time: 0.074, memory: 49164, loss_cls_0: 0.9126, loss_box_0: 1.7924, loss_cns_0: 0.6176, loss_yns_0: 0.1586, loss_cls_1: 0.9786, loss_box_1: 1.9238, loss_cns_1: 0.6394, loss_yns_1: 0.1615, loss_cls_2: 1.0413, loss_box_2: 1.8640, loss_cns_2: 0.6528, loss_yns_2: 0.1621, loss_cls_3: 1.0246, loss_box_3: 1.8651, loss_cns_3: 0.6569, loss_yns_3: 0.1628, loss_cls_4: 1.0384, loss_box_4: 1.8473, loss_cns_4: 0.6553, loss_yns_4: 0.1613, loss_cls_5: 1.0318, loss_box_5: 1.8449, loss_cns_5: 0.6597, loss_yns_5: 0.1646, loss_cls_dn_0: 0.2644, loss_box_dn_0: 0.8070, loss_cls_dn_1: 0.1879, loss_box_dn_1: 0.7859, loss_cls_dn_2: 0.2012, loss_box_dn_2: 0.7777, loss_cls_dn_3: 0.1972, loss_box_dn_3: 0.7774, loss_cls_dn_4: 0.2043, loss_box_dn_4: 0.7851, loss_cls_dn_5: 0.2136, loss_box_dn_5: 0.7893, loss_dense_depth: 0.8504, loss: 28.8586, grad_norm: 44.6700 -2026-01-14 21:34:18,648 - mmdet - INFO - Iter [104/17500] lr: 1.412e-04, eta: 13:46:53, time: 1.563, data_time: 0.078, memory: 49164, loss_cls_0: 0.9316, loss_box_0: 1.8288, loss_cns_0: 0.6157, loss_yns_0: 0.1575, loss_cls_1: 0.9733, loss_box_1: 1.9677, loss_cns_1: 0.6355, loss_yns_1: 0.1586, loss_cls_2: 1.0510, loss_box_2: 1.9037, loss_cns_2: 0.6502, loss_yns_2: 0.1597, loss_cls_3: 1.0251, loss_box_3: 1.8846, loss_cns_3: 0.6547, loss_yns_3: 0.1601, loss_cls_4: 1.0254, loss_box_4: 1.8764, loss_cns_4: 0.6548, loss_yns_4: 0.1598, loss_cls_5: 1.0326, loss_box_5: 1.8884, loss_cns_5: 0.6526, loss_yns_5: 0.1615, loss_cls_dn_0: 0.2587, loss_box_dn_0: 0.8240, loss_cls_dn_1: 0.1823, loss_box_dn_1: 0.7800, loss_cls_dn_2: 0.1994, loss_box_dn_2: 0.7844, loss_cls_dn_3: 0.1911, loss_box_dn_3: 0.7832, loss_cls_dn_4: 0.1984, loss_box_dn_4: 0.7945, loss_cls_dn_5: 0.2053, loss_box_dn_5: 0.8130, loss_dense_depth: 0.8079, loss: 29.0317, grad_norm: 46.9978 -2026-01-14 21:34:20,334 - mmdet - INFO - Iter [105/17500] lr: 1.416e-04, eta: 13:43:32, time: 1.658, data_time: 0.077, memory: 49164, loss_cls_0: 0.8820, loss_box_0: 1.7471, loss_cns_0: 0.6197, loss_yns_0: 0.1555, loss_cls_1: 0.9583, loss_box_1: 1.9016, loss_cns_1: 0.6427, loss_yns_1: 0.1591, loss_cls_2: 1.0195, loss_box_2: 1.8489, loss_cns_2: 0.6569, loss_yns_2: 0.1599, loss_cls_3: 1.0129, loss_box_3: 1.8560, loss_cns_3: 0.6580, loss_yns_3: 0.1588, loss_cls_4: 1.0261, loss_box_4: 1.8486, loss_cns_4: 0.6587, loss_yns_4: 0.1603, loss_cls_5: 1.0402, loss_box_5: 1.8593, loss_cns_5: 0.6575, loss_yns_5: 0.1571, loss_cls_dn_0: 0.2564, loss_box_dn_0: 0.8142, loss_cls_dn_1: 0.1833, loss_box_dn_1: 0.8146, loss_cls_dn_2: 0.1993, loss_box_dn_2: 0.8175, loss_cls_dn_3: 0.1915, loss_box_dn_3: 0.8312, loss_cls_dn_4: 0.1962, loss_box_dn_4: 0.8451, loss_cls_dn_5: 0.2026, loss_box_dn_5: 0.8681, loss_dense_depth: 0.8041, loss: 28.8686, grad_norm: 51.9232 -2026-01-14 21:34:21,934 - mmdet - INFO - Iter [106/17500] lr: 1.420e-04, eta: 13:40:03, time: 1.584, data_time: 0.094, memory: 49164, loss_cls_0: 0.9081, loss_box_0: 1.6829, loss_cns_0: 0.6032, loss_yns_0: 0.1498, loss_cls_1: 0.9644, loss_box_1: 1.9040, loss_cns_1: 0.6416, loss_yns_1: 0.1571, loss_cls_2: 1.0172, loss_box_2: 1.8772, loss_cns_2: 0.6534, loss_yns_2: 0.1582, loss_cls_3: 1.0099, loss_box_3: 1.9093, loss_cns_3: 0.6573, loss_yns_3: 0.1584, loss_cls_4: 1.0171, loss_box_4: 1.8838, loss_cns_4: 0.6567, loss_yns_4: 0.1595, loss_cls_5: 1.0220, loss_box_5: 1.8745, loss_cns_5: 0.6557, loss_yns_5: 0.1617, loss_cls_dn_0: 0.2591, loss_box_dn_0: 0.7936, loss_cls_dn_1: 0.1838, loss_box_dn_1: 0.8220, loss_cls_dn_2: 0.1960, loss_box_dn_2: 0.8242, loss_cls_dn_3: 0.1902, loss_box_dn_3: 0.8487, loss_cls_dn_4: 0.1919, loss_box_dn_4: 0.8509, loss_cls_dn_5: 0.1996, loss_box_dn_5: 0.8649, loss_dense_depth: 0.7938, loss: 28.9015, grad_norm: 53.7210 -2026-01-14 21:34:23,473 - mmdet - INFO - Iter [107/17500] lr: 1.424e-04, eta: 13:36:38, time: 1.583, data_time: 0.106, memory: 49164, loss_cls_0: 0.8826, loss_box_0: 1.7181, loss_cns_0: 0.6105, loss_yns_0: 0.1529, loss_cls_1: 0.9556, loss_box_1: 1.8833, loss_cns_1: 0.6426, loss_yns_1: 0.1569, loss_cls_2: 1.0088, loss_box_2: 1.8318, loss_cns_2: 0.6553, loss_yns_2: 0.1576, loss_cls_3: 1.0034, loss_box_3: 1.8625, loss_cns_3: 0.6603, loss_yns_3: 0.1573, loss_cls_4: 1.0162, loss_box_4: 1.8253, loss_cns_4: 0.6604, loss_yns_4: 0.1583, loss_cls_5: 1.0240, loss_box_5: 1.8022, loss_cns_5: 0.6579, loss_yns_5: 0.1635, loss_cls_dn_0: 0.2538, loss_box_dn_0: 0.8136, loss_cls_dn_1: 0.1831, loss_box_dn_1: 0.8016, loss_cls_dn_2: 0.1940, loss_box_dn_2: 0.7951, loss_cls_dn_3: 0.1883, loss_box_dn_3: 0.8143, loss_cls_dn_4: 0.1944, loss_box_dn_4: 0.8065, loss_cls_dn_5: 0.2062, loss_box_dn_5: 0.8049, loss_dense_depth: 0.8094, loss: 28.5125, grad_norm: 44.9346 -2026-01-14 21:34:25,051 - mmdet - INFO - Iter [108/17500] lr: 1.428e-04, eta: 13:33:15, time: 1.577, data_time: 0.075, memory: 49164, loss_cls_0: 0.9042, loss_box_0: 1.7765, loss_cns_0: 0.6223, loss_yns_0: 0.1564, loss_cls_1: 0.9732, loss_box_1: 1.8774, loss_cns_1: 0.6419, loss_yns_1: 0.1590, loss_cls_2: 1.0345, loss_box_2: 1.8159, loss_cns_2: 0.6544, loss_yns_2: 0.1602, loss_cls_3: 1.0715, loss_box_3: 1.8298, loss_cns_3: 0.6574, loss_yns_3: 0.1578, loss_cls_4: 1.0340, loss_box_4: 1.8413, loss_cns_4: 0.6565, loss_yns_4: 0.1596, loss_cls_5: 1.0485, loss_box_5: 1.8269, loss_cns_5: 0.6558, loss_yns_5: 0.1587, loss_cls_dn_0: 0.2505, loss_box_dn_0: 0.8129, loss_cls_dn_1: 0.1862, loss_box_dn_1: 0.7577, loss_cls_dn_2: 0.1937, loss_box_dn_2: 0.7472, loss_cls_dn_3: 0.1928, loss_box_dn_3: 0.7523, loss_cls_dn_4: 0.1970, loss_box_dn_4: 0.7606, loss_cls_dn_5: 0.2085, loss_box_dn_5: 0.7615, loss_dense_depth: 0.7797, loss: 28.4744, grad_norm: 54.7869 -2026-01-14 21:34:26,613 - mmdet - INFO - Iter [109/17500] lr: 1.432e-04, eta: 13:29:54, time: 1.562, data_time: 0.075, memory: 49164, loss_cls_0: 0.9001, loss_box_0: 1.8117, loss_cns_0: 0.6117, loss_yns_0: 0.1561, loss_cls_1: 0.9662, loss_box_1: 1.8916, loss_cns_1: 0.6354, loss_yns_1: 0.1589, loss_cls_2: 1.0174, loss_box_2: 1.8347, loss_cns_2: 0.6484, loss_yns_2: 0.1618, loss_cls_3: 1.0252, loss_box_3: 1.8326, loss_cns_3: 0.6512, loss_yns_3: 0.1593, loss_cls_4: 1.0398, loss_box_4: 1.8709, loss_cns_4: 0.6510, loss_yns_4: 0.1619, loss_cls_5: 1.0544, loss_box_5: 1.8740, loss_cns_5: 0.6502, loss_yns_5: 0.1610, loss_cls_dn_0: 0.2564, loss_box_dn_0: 0.8223, loss_cls_dn_1: 0.1845, loss_box_dn_1: 0.7816, loss_cls_dn_2: 0.1917, loss_box_dn_2: 0.7646, loss_cls_dn_3: 0.1887, loss_box_dn_3: 0.7706, loss_cls_dn_4: 0.1962, loss_box_dn_4: 0.7987, loss_cls_dn_5: 0.2086, loss_box_dn_5: 0.8142, loss_dense_depth: 0.7912, loss: 28.6947, grad_norm: 43.9923 -2026-01-14 21:34:28,183 - mmdet - INFO - Iter [110/17500] lr: 1.436e-04, eta: 13:26:38, time: 1.571, data_time: 0.075, memory: 49164, loss_cls_0: 0.9131, loss_box_0: 1.8110, loss_cns_0: 0.6158, loss_yns_0: 0.1554, loss_cls_1: 0.9833, loss_box_1: 1.9195, loss_cns_1: 0.6273, loss_yns_1: 0.1570, loss_cls_2: 1.0125, loss_box_2: 1.8624, loss_cns_2: 0.6456, loss_yns_2: 0.1567, loss_cls_3: 1.0437, loss_box_3: 1.8427, loss_cns_3: 0.6473, loss_yns_3: 0.1575, loss_cls_4: 1.0300, loss_box_4: 1.8697, loss_cns_4: 0.6477, loss_yns_4: 0.1596, loss_cls_5: 1.0405, loss_box_5: 1.8959, loss_cns_5: 0.6412, loss_yns_5: 0.1621, loss_cls_dn_0: 0.2576, loss_box_dn_0: 0.8076, loss_cls_dn_1: 0.1896, loss_box_dn_1: 0.8346, loss_cls_dn_2: 0.1944, loss_box_dn_2: 0.8230, loss_cls_dn_3: 0.1923, loss_box_dn_3: 0.8402, loss_cls_dn_4: 0.1994, loss_box_dn_4: 0.8853, loss_cls_dn_5: 0.2118, loss_box_dn_5: 0.9157, loss_dense_depth: 0.8165, loss: 29.1655, grad_norm: 47.2830 -2026-01-14 21:34:29,771 - mmdet - INFO - Iter [111/17500] lr: 1.440e-04, eta: 13:23:28, time: 1.589, data_time: 0.077, memory: 49164, loss_cls_0: 0.9177, loss_box_0: 1.8186, loss_cns_0: 0.6209, loss_yns_0: 0.1574, loss_cls_1: 0.9675, loss_box_1: 1.9156, loss_cns_1: 0.6275, loss_yns_1: 0.1563, loss_cls_2: 1.0208, loss_box_2: 1.8601, loss_cns_2: 0.6427, loss_yns_2: 0.1563, loss_cls_3: 1.0371, loss_box_3: 1.8585, loss_cns_3: 0.6467, loss_yns_3: 0.1567, loss_cls_4: 1.0260, loss_box_4: 1.8752, loss_cns_4: 0.6465, loss_yns_4: 0.1573, loss_cls_5: 1.0378, loss_box_5: 1.8962, loss_cns_5: 0.6420, loss_yns_5: 0.1575, loss_cls_dn_0: 0.2505, loss_box_dn_0: 0.8032, loss_cls_dn_1: 0.1875, loss_box_dn_1: 0.8679, loss_cls_dn_2: 0.1979, loss_box_dn_2: 0.8585, loss_cls_dn_3: 0.1966, loss_box_dn_3: 0.8755, loss_cls_dn_4: 0.1982, loss_box_dn_4: 0.9148, loss_cls_dn_5: 0.2076, loss_box_dn_5: 0.9387, loss_dense_depth: 0.8206, loss: 29.3163, grad_norm: 52.2755 -2026-01-14 21:34:31,342 - mmdet - INFO - Iter [112/17500] lr: 1.444e-04, eta: 13:20:19, time: 1.571, data_time: 0.069, memory: 49164, loss_cls_0: 0.9308, loss_box_0: 1.8498, loss_cns_0: 0.6170, loss_yns_0: 0.1598, loss_cls_1: 0.9688, loss_box_1: 1.9456, loss_cns_1: 0.6219, loss_yns_1: 0.1562, loss_cls_2: 1.0278, loss_box_2: 1.8852, loss_cns_2: 0.6418, loss_yns_2: 0.1559, loss_cls_3: 1.0269, loss_box_3: 1.8852, loss_cns_3: 0.6478, loss_yns_3: 0.1580, loss_cls_4: 1.0255, loss_box_4: 1.8745, loss_cns_4: 0.6456, loss_yns_4: 0.1616, loss_cls_5: 1.0367, loss_box_5: 1.8837, loss_cns_5: 0.6503, loss_yns_5: 0.1573, loss_cls_dn_0: 0.2582, loss_box_dn_0: 0.8066, loss_cls_dn_1: 0.1875, loss_box_dn_1: 0.9050, loss_cls_dn_2: 0.2000, loss_box_dn_2: 0.8853, loss_cls_dn_3: 0.1988, loss_box_dn_3: 0.8942, loss_cls_dn_4: 0.1965, loss_box_dn_4: 0.9104, loss_cls_dn_5: 0.2049, loss_box_dn_5: 0.9319, loss_dense_depth: 0.8570, loss: 29.5502, grad_norm: 49.9503 -2026-01-14 21:34:32,926 - mmdet - INFO - Iter [113/17500] lr: 1.448e-04, eta: 13:17:15, time: 1.583, data_time: 0.074, memory: 49164, loss_cls_0: 0.8831, loss_box_0: 1.8159, loss_cns_0: 0.6188, loss_yns_0: 0.1538, loss_cls_1: 0.9345, loss_box_1: 1.8779, loss_cns_1: 0.6328, loss_yns_1: 0.1522, loss_cls_2: 1.0018, loss_box_2: 1.8164, loss_cns_2: 0.6493, loss_yns_2: 0.1534, loss_cls_3: 0.9891, loss_box_3: 1.7955, loss_cns_3: 0.6569, loss_yns_3: 0.1546, loss_cls_4: 0.9961, loss_box_4: 1.8130, loss_cns_4: 0.6580, loss_yns_4: 0.1575, loss_cls_5: 1.0050, loss_box_5: 1.8191, loss_cns_5: 0.6574, loss_yns_5: 0.1539, loss_cls_dn_0: 0.2505, loss_box_dn_0: 0.8020, loss_cls_dn_1: 0.1837, loss_box_dn_1: 0.8547, loss_cls_dn_2: 0.1958, loss_box_dn_2: 0.8262, loss_cls_dn_3: 0.1972, loss_box_dn_3: 0.8191, loss_cls_dn_4: 0.1959, loss_box_dn_4: 0.8365, loss_cls_dn_5: 0.2052, loss_box_dn_5: 0.8504, loss_dense_depth: 0.7902, loss: 28.5537, grad_norm: 39.9600 -2026-01-14 21:34:34,584 - mmdet - INFO - Iter [114/17500] lr: 1.452e-04, eta: 13:14:21, time: 1.631, data_time: 0.077, memory: 49164, loss_cls_0: 0.8983, loss_box_0: 1.8379, loss_cns_0: 0.6118, loss_yns_0: 0.1553, loss_cls_1: 0.9704, loss_box_1: 1.8889, loss_cns_1: 0.6376, loss_yns_1: 0.1544, loss_cls_2: 1.0008, loss_box_2: 1.8152, loss_cns_2: 0.6495, loss_yns_2: 0.1555, loss_cls_3: 1.0122, loss_box_3: 1.7958, loss_cns_3: 0.6525, loss_yns_3: 0.1560, loss_cls_4: 1.0084, loss_box_4: 1.8297, loss_cns_4: 0.6539, loss_yns_4: 0.1557, loss_cls_5: 1.0130, loss_box_5: 1.8198, loss_cns_5: 0.6499, loss_yns_5: 0.1565, loss_cls_dn_0: 0.2590, loss_box_dn_0: 0.8067, loss_cls_dn_1: 0.1783, loss_box_dn_1: 0.8071, loss_cls_dn_2: 0.1885, loss_box_dn_2: 0.7872, loss_cls_dn_3: 0.1882, loss_box_dn_3: 0.7847, loss_cls_dn_4: 0.1940, loss_box_dn_4: 0.8094, loss_cls_dn_5: 0.2046, loss_box_dn_5: 0.8142, loss_dense_depth: 0.8025, loss: 28.5034, grad_norm: 46.8218 -2026-01-14 21:34:36,139 - mmdet - INFO - Iter [115/17500] lr: 1.456e-04, eta: 13:11:23, time: 1.584, data_time: 0.094, memory: 49164, loss_cls_0: 0.8602, loss_box_0: 1.7923, loss_cns_0: 0.6159, loss_yns_0: 0.1531, loss_cls_1: 0.9695, loss_box_1: 1.8747, loss_cns_1: 0.6418, loss_yns_1: 0.1539, loss_cls_2: 1.0132, loss_box_2: 1.8256, loss_cns_2: 0.6523, loss_yns_2: 0.1528, loss_cls_3: 1.0714, loss_box_3: 1.8051, loss_cns_3: 0.6554, loss_yns_3: 0.1528, loss_cls_4: 0.9987, loss_box_4: 1.8036, loss_cns_4: 0.6555, loss_yns_4: 0.1537, loss_cls_5: 1.0103, loss_box_5: 1.7938, loss_cns_5: 0.6530, loss_yns_5: 0.1578, loss_cls_dn_0: 0.2486, loss_box_dn_0: 0.8069, loss_cls_dn_1: 0.1780, loss_box_dn_1: 0.7967, loss_cls_dn_2: 0.1860, loss_box_dn_2: 0.7879, loss_cls_dn_3: 0.1874, loss_box_dn_3: 0.7942, loss_cls_dn_4: 0.1902, loss_box_dn_4: 0.8080, loss_cls_dn_5: 0.1999, loss_box_dn_5: 0.8178, loss_dense_depth: 0.7902, loss: 28.4084, grad_norm: 51.5741 -2026-01-14 21:34:37,761 - mmdet - INFO - Iter [116/17500] lr: 1.460e-04, eta: 13:08:34, time: 1.621, data_time: 0.075, memory: 49164, loss_cls_0: 0.8790, loss_box_0: 1.7911, loss_cns_0: 0.6144, loss_yns_0: 0.1533, loss_cls_1: 0.9664, loss_box_1: 1.8915, loss_cns_1: 0.6444, loss_yns_1: 0.1566, loss_cls_2: 1.0030, loss_box_2: 1.8686, loss_cns_2: 0.6525, loss_yns_2: 0.1553, loss_cls_3: 1.0143, loss_box_3: 1.8497, loss_cns_3: 0.6540, loss_yns_3: 0.1554, loss_cls_4: 1.0069, loss_box_4: 1.8309, loss_cns_4: 0.6543, loss_yns_4: 0.1614, loss_cls_5: 1.0059, loss_box_5: 1.8915, loss_cns_5: 0.6484, loss_yns_5: 0.1596, loss_cls_dn_0: 0.2525, loss_box_dn_0: 0.8019, loss_cls_dn_1: 0.1807, loss_box_dn_1: 0.8028, loss_cls_dn_2: 0.1912, loss_box_dn_2: 0.8145, loss_cls_dn_3: 0.1944, loss_box_dn_3: 0.8349, loss_cls_dn_4: 0.2032, loss_box_dn_4: 0.8573, loss_cls_dn_5: 0.2106, loss_box_dn_5: 0.9056, loss_dense_depth: 0.8036, loss: 28.8616, grad_norm: 52.0596 -2026-01-14 21:34:39,395 - mmdet - INFO - Iter [117/17500] lr: 1.464e-04, eta: 13:05:50, time: 1.633, data_time: 0.077, memory: 49164, loss_cls_0: 0.8642, loss_box_0: 1.7522, loss_cns_0: 0.6179, loss_yns_0: 0.1503, loss_cls_1: 0.9582, loss_box_1: 1.8547, loss_cns_1: 0.6477, loss_yns_1: 0.1576, loss_cls_2: 0.9796, loss_box_2: 1.8138, loss_cns_2: 0.6552, loss_yns_2: 0.1591, loss_cls_3: 0.9913, loss_box_3: 1.7989, loss_cns_3: 0.6563, loss_yns_3: 0.1550, loss_cls_4: 1.0088, loss_box_4: 1.7965, loss_cns_4: 0.6536, loss_yns_4: 0.1604, loss_cls_5: 1.0167, loss_box_5: 1.8289, loss_cns_5: 0.6523, loss_yns_5: 0.1572, loss_cls_dn_0: 0.2434, loss_box_dn_0: 0.7990, loss_cls_dn_1: 0.1807, loss_box_dn_1: 0.8475, loss_cls_dn_2: 0.1877, loss_box_dn_2: 0.8585, loss_cls_dn_3: 0.1917, loss_box_dn_3: 0.8807, loss_cls_dn_4: 0.2002, loss_box_dn_4: 0.9114, loss_cls_dn_5: 0.2139, loss_box_dn_5: 0.9523, loss_dense_depth: 0.8039, loss: 28.7574, grad_norm: 45.0115 -2026-01-14 21:34:40,988 - mmdet - INFO - Iter [118/17500] lr: 1.468e-04, eta: 13:02:58, time: 1.566, data_time: 0.076, memory: 49164, loss_cls_0: 0.8954, loss_box_0: 1.7376, loss_cns_0: 0.6209, loss_yns_0: 0.1541, loss_cls_1: 0.9672, loss_box_1: 1.8328, loss_cns_1: 0.6436, loss_yns_1: 0.1557, loss_cls_2: 0.9855, loss_box_2: 1.8036, loss_cns_2: 0.6497, loss_yns_2: 0.1564, loss_cls_3: 0.9959, loss_box_3: 1.8124, loss_cns_3: 0.6525, loss_yns_3: 0.1537, loss_cls_4: 0.9911, loss_box_4: 1.8155, loss_cns_4: 0.6494, loss_yns_4: 0.1548, loss_cls_5: 1.0012, loss_box_5: 1.7979, loss_cns_5: 0.6477, loss_yns_5: 0.1541, loss_cls_dn_0: 0.2386, loss_box_dn_0: 0.7942, loss_cls_dn_1: 0.1784, loss_box_dn_1: 0.8717, loss_cls_dn_2: 0.1852, loss_box_dn_2: 0.8771, loss_cls_dn_3: 0.1880, loss_box_dn_3: 0.9008, loss_cls_dn_4: 0.1880, loss_box_dn_4: 0.9242, loss_cls_dn_5: 0.2012, loss_box_dn_5: 0.9433, loss_dense_depth: 0.8001, loss: 28.7200, grad_norm: 47.6191 -2026-01-14 21:34:42,601 - mmdet - INFO - Iter [119/17500] lr: 1.472e-04, eta: 13:00:20, time: 1.640, data_time: 0.113, memory: 49164, loss_cls_0: 0.8752, loss_box_0: 1.7384, loss_cns_0: 0.6117, loss_yns_0: 0.1534, loss_cls_1: 0.9739, loss_box_1: 1.8496, loss_cns_1: 0.6464, loss_yns_1: 0.1545, loss_cls_2: 0.9962, loss_box_2: 1.7854, loss_cns_2: 0.6501, loss_yns_2: 0.1552, loss_cls_3: 1.0074, loss_box_3: 1.7843, loss_cns_3: 0.6534, loss_yns_3: 0.1528, loss_cls_4: 1.0155, loss_box_4: 1.7703, loss_cns_4: 0.6527, loss_yns_4: 0.1523, loss_cls_5: 1.0057, loss_box_5: 1.7608, loss_cns_5: 0.6498, loss_yns_5: 0.1552, loss_cls_dn_0: 0.2349, loss_box_dn_0: 0.7912, loss_cls_dn_1: 0.1756, loss_box_dn_1: 0.8895, loss_cls_dn_2: 0.1809, loss_box_dn_2: 0.8790, loss_cls_dn_3: 0.1850, loss_box_dn_3: 0.8850, loss_cls_dn_4: 0.1846, loss_box_dn_4: 0.8866, loss_cls_dn_5: 0.1921, loss_box_dn_5: 0.8962, loss_dense_depth: 0.7995, loss: 28.5305, grad_norm: 37.3264 -2026-01-14 21:34:44,170 - mmdet - INFO - Iter [120/17500] lr: 1.476e-04, eta: 12:57:35, time: 1.571, data_time: 0.086, memory: 49164, loss_cls_0: 0.8675, loss_box_0: 1.7465, loss_cns_0: 0.6072, loss_yns_0: 0.1518, loss_cls_1: 0.9658, loss_box_1: 1.8122, loss_cns_1: 0.6510, loss_yns_1: 0.1557, loss_cls_2: 0.9898, loss_box_2: 1.7731, loss_cns_2: 0.6571, loss_yns_2: 0.1541, loss_cls_3: 1.0100, loss_box_3: 1.7836, loss_cns_3: 0.6576, loss_yns_3: 0.1535, loss_cls_4: 0.9839, loss_box_4: 1.7769, loss_cns_4: 0.6562, loss_yns_4: 0.1541, loss_cls_5: 0.9888, loss_box_5: 1.7848, loss_cns_5: 0.6542, loss_yns_5: 0.1594, loss_cls_dn_0: 0.2385, loss_box_dn_0: 0.8082, loss_cls_dn_1: 0.1747, loss_box_dn_1: 0.8053, loss_cls_dn_2: 0.1782, loss_box_dn_2: 0.7823, loss_cls_dn_3: 0.1863, loss_box_dn_3: 0.7877, loss_cls_dn_4: 0.1805, loss_box_dn_4: 0.7846, loss_cls_dn_5: 0.1907, loss_box_dn_5: 0.7906, loss_dense_depth: 0.8235, loss: 28.0257, grad_norm: 45.2440 -2026-01-14 21:34:45,841 - mmdet - INFO - Iter [121/17500] lr: 1.480e-04, eta: 12:55:06, time: 1.669, data_time: 0.116, memory: 49164, loss_cls_0: 0.8793, loss_box_0: 1.7747, loss_cns_0: 0.6199, loss_yns_0: 0.1535, loss_cls_1: 0.9776, loss_box_1: 1.8589, loss_cns_1: 0.6487, loss_yns_1: 0.1551, loss_cls_2: 0.9963, loss_box_2: 1.8282, loss_cns_2: 0.6542, loss_yns_2: 0.1558, loss_cls_3: 1.0094, loss_box_3: 1.8401, loss_cns_3: 0.6570, loss_yns_3: 0.1581, loss_cls_4: 1.0021, loss_box_4: 1.8203, loss_cns_4: 0.6568, loss_yns_4: 0.1612, loss_cls_5: 1.0080, loss_box_5: 1.8341, loss_cns_5: 0.6555, loss_yns_5: 0.1593, loss_cls_dn_0: 0.2372, loss_box_dn_0: 0.8063, loss_cls_dn_1: 0.1705, loss_box_dn_1: 0.8013, loss_cls_dn_2: 0.1775, loss_box_dn_2: 0.7856, loss_cls_dn_3: 0.1789, loss_box_dn_3: 0.8007, loss_cls_dn_4: 0.1791, loss_box_dn_4: 0.8041, loss_cls_dn_5: 0.1892, loss_box_dn_5: 0.8208, loss_dense_depth: 0.7964, loss: 28.4117, grad_norm: 49.8754 -2026-01-14 21:34:47,516 - mmdet - INFO - Iter [122/17500] lr: 1.484e-04, eta: 12:52:41, time: 1.677, data_time: 0.176, memory: 49164, loss_cls_0: 0.9023, loss_box_0: 1.7893, loss_cns_0: 0.6207, loss_yns_0: 0.1584, loss_cls_1: 0.9931, loss_box_1: 1.8520, loss_cns_1: 0.6454, loss_yns_1: 0.1584, loss_cls_2: 1.0112, loss_box_2: 1.8037, loss_cns_2: 0.6510, loss_yns_2: 0.1602, loss_cls_3: 1.0078, loss_box_3: 1.8059, loss_cns_3: 0.6506, loss_yns_3: 0.1621, loss_cls_4: 1.0141, loss_box_4: 1.8006, loss_cns_4: 0.6517, loss_yns_4: 0.1626, loss_cls_5: 1.0068, loss_box_5: 1.8502, loss_cns_5: 0.6452, loss_yns_5: 0.1627, loss_cls_dn_0: 0.2363, loss_box_dn_0: 0.8050, loss_cls_dn_1: 0.1696, loss_box_dn_1: 0.8047, loss_cls_dn_2: 0.1779, loss_box_dn_2: 0.7939, loss_cls_dn_3: 0.1738, loss_box_dn_3: 0.8160, loss_cls_dn_4: 0.1794, loss_box_dn_4: 0.8365, loss_cls_dn_5: 0.1865, loss_box_dn_5: 0.8743, loss_dense_depth: 0.7743, loss: 28.4939, grad_norm: 48.7589 -2026-01-14 21:34:49,129 - mmdet - INFO - Iter [123/17500] lr: 1.488e-04, eta: 12:50:10, time: 1.614, data_time: 0.075, memory: 49164, loss_cls_0: 0.8728, loss_box_0: 1.7539, loss_cns_0: 0.6224, loss_yns_0: 0.1561, loss_cls_1: 0.9639, loss_box_1: 1.8421, loss_cns_1: 0.6455, loss_yns_1: 0.1562, loss_cls_2: 0.9895, loss_box_2: 1.8030, loss_cns_2: 0.6513, loss_yns_2: 0.1577, loss_cls_3: 0.9994, loss_box_3: 1.8042, loss_cns_3: 0.6522, loss_yns_3: 0.1626, loss_cls_4: 1.0013, loss_box_4: 1.8357, loss_cns_4: 0.6516, loss_yns_4: 0.1588, loss_cls_5: 0.9980, loss_box_5: 1.8556, loss_cns_5: 0.6487, loss_yns_5: 0.1631, loss_cls_dn_0: 0.2316, loss_box_dn_0: 0.7918, loss_cls_dn_1: 0.1689, loss_box_dn_1: 0.8622, loss_cls_dn_2: 0.1791, loss_box_dn_2: 0.8635, loss_cls_dn_3: 0.1751, loss_box_dn_3: 0.8970, loss_cls_dn_4: 0.1827, loss_box_dn_4: 0.9438, loss_cls_dn_5: 0.1890, loss_box_dn_5: 0.9833, loss_dense_depth: 0.8102, loss: 28.8239, grad_norm: 50.9192 -2026-01-14 21:34:50,677 - mmdet - INFO - Iter [124/17500] lr: 1.492e-04, eta: 12:47:31, time: 1.548, data_time: 0.073, memory: 49164, loss_cls_0: 0.8751, loss_box_0: 1.7644, loss_cns_0: 0.6275, loss_yns_0: 0.1564, loss_cls_1: 0.9654, loss_box_1: 1.8061, loss_cns_1: 0.6575, loss_yns_1: 0.1572, loss_cls_2: 1.0018, loss_box_2: 1.7927, loss_cns_2: 0.6556, loss_yns_2: 0.1565, loss_cls_3: 0.9989, loss_box_3: 1.8178, loss_cns_3: 0.6549, loss_yns_3: 0.1600, loss_cls_4: 1.0040, loss_box_4: 1.8437, loss_cns_4: 0.6510, loss_yns_4: 0.1583, loss_cls_5: 1.0090, loss_box_5: 1.8418, loss_cns_5: 0.6556, loss_yns_5: 0.1609, loss_cls_dn_0: 0.2331, loss_box_dn_0: 0.7929, loss_cls_dn_1: 0.1665, loss_box_dn_1: 0.8763, loss_cls_dn_2: 0.1762, loss_box_dn_2: 0.8908, loss_cls_dn_3: 0.1724, loss_box_dn_3: 0.9232, loss_cls_dn_4: 0.1793, loss_box_dn_4: 0.9591, loss_cls_dn_5: 0.1930, loss_box_dn_5: 0.9873, loss_dense_depth: 0.7707, loss: 28.8930, grad_norm: 57.4208 -2026-01-14 21:34:52,278 - mmdet - INFO - Iter [125/17500] lr: 1.496e-04, eta: 12:45:03, time: 1.600, data_time: 0.072, memory: 49164, loss_cls_0: 0.8787, loss_box_0: 1.7912, loss_cns_0: 0.6254, loss_yns_0: 0.1576, loss_cls_1: 0.9687, loss_box_1: 1.7868, loss_cns_1: 0.6564, loss_yns_1: 0.1580, loss_cls_2: 1.0164, loss_box_2: 1.7559, loss_cns_2: 0.6554, loss_yns_2: 0.1579, loss_cls_3: 1.0055, loss_box_3: 1.7766, loss_cns_3: 0.6570, loss_yns_3: 0.1576, loss_cls_4: 1.0150, loss_box_4: 1.7701, loss_cns_4: 0.6557, loss_yns_4: 0.1584, loss_cls_5: 1.0187, loss_box_5: 1.7856, loss_cns_5: 0.6567, loss_yns_5: 0.1594, loss_cls_dn_0: 0.2348, loss_box_dn_0: 0.7858, loss_cls_dn_1: 0.1697, loss_box_dn_1: 0.8959, loss_cls_dn_2: 0.1799, loss_box_dn_2: 0.8984, loss_cls_dn_3: 0.1752, loss_box_dn_3: 0.9148, loss_cls_dn_4: 0.1795, loss_box_dn_4: 0.9298, loss_cls_dn_5: 0.1963, loss_box_dn_5: 0.9515, loss_dense_depth: 0.7722, loss: 28.7083, grad_norm: 42.7264 -2026-01-14 21:34:53,871 - mmdet - INFO - Iter [126/17500] lr: 1.500e-04, eta: 12:42:36, time: 1.594, data_time: 0.078, memory: 49164, loss_cls_0: 0.9115, loss_box_0: 1.7780, loss_cns_0: 0.6241, loss_yns_0: 0.1560, loss_cls_1: 0.9632, loss_box_1: 1.8538, loss_cns_1: 0.6469, loss_yns_1: 0.1582, loss_cls_2: 0.9966, loss_box_2: 1.8107, loss_cns_2: 0.6523, loss_yns_2: 0.1587, loss_cls_3: 1.0061, loss_box_3: 1.8089, loss_cns_3: 0.6528, loss_yns_3: 0.1606, loss_cls_4: 1.0081, loss_box_4: 1.7933, loss_cns_4: 0.6526, loss_yns_4: 0.1596, loss_cls_5: 1.0122, loss_box_5: 1.8304, loss_cns_5: 0.6497, loss_yns_5: 0.1591, loss_cls_dn_0: 0.2321, loss_box_dn_0: 0.7869, loss_cls_dn_1: 0.1719, loss_box_dn_1: 0.8493, loss_cls_dn_2: 0.1796, loss_box_dn_2: 0.8299, loss_cls_dn_3: 0.1776, loss_box_dn_3: 0.8293, loss_cls_dn_4: 0.1799, loss_box_dn_4: 0.8354, loss_cls_dn_5: 0.1916, loss_box_dn_5: 0.8487, loss_dense_depth: 0.7717, loss: 28.4873, grad_norm: 46.7077 -2026-01-14 21:34:55,494 - mmdet - INFO - Iter [127/17500] lr: 1.504e-04, eta: 12:40:08, time: 1.576, data_time: 0.071, memory: 49164, loss_cls_0: 0.8746, loss_box_0: 1.7224, loss_cns_0: 0.6180, loss_yns_0: 0.1544, loss_cls_1: 0.9512, loss_box_1: 1.8544, loss_cns_1: 0.6457, loss_yns_1: 0.1537, loss_cls_2: 0.9882, loss_box_2: 1.7691, loss_cns_2: 0.6530, loss_yns_2: 0.1562, loss_cls_3: 0.9992, loss_box_3: 1.7615, loss_cns_3: 0.6543, loss_yns_3: 0.1612, loss_cls_4: 1.0162, loss_box_4: 1.7648, loss_cns_4: 0.6542, loss_yns_4: 0.1630, loss_cls_5: 1.0172, loss_box_5: 1.7714, loss_cns_5: 0.6557, loss_yns_5: 0.1580, loss_cls_dn_0: 0.2344, loss_box_dn_0: 0.7840, loss_cls_dn_1: 0.1688, loss_box_dn_1: 0.7854, loss_cls_dn_2: 0.1754, loss_box_dn_2: 0.7518, loss_cls_dn_3: 0.1728, loss_box_dn_3: 0.7448, loss_cls_dn_4: 0.1800, loss_box_dn_4: 0.7536, loss_cls_dn_5: 0.1893, loss_box_dn_5: 0.7623, loss_dense_depth: 0.7734, loss: 27.7935, grad_norm: 39.6782 -2026-01-14 21:34:57,080 - mmdet - INFO - Iter [128/17500] lr: 1.508e-04, eta: 12:37:51, time: 1.632, data_time: 0.122, memory: 49164, loss_cls_0: 0.9140, loss_box_0: 1.6910, loss_cns_0: 0.6109, loss_yns_0: 0.1507, loss_cls_1: 0.9507, loss_box_1: 1.8059, loss_cns_1: 0.6474, loss_yns_1: 0.1554, loss_cls_2: 0.9870, loss_box_2: 1.7583, loss_cns_2: 0.6494, loss_yns_2: 0.1575, loss_cls_3: 0.9985, loss_box_3: 1.7378, loss_cns_3: 0.6512, loss_yns_3: 0.1594, loss_cls_4: 1.0162, loss_box_4: 1.7399, loss_cns_4: 0.6498, loss_yns_4: 0.1605, loss_cls_5: 1.0029, loss_box_5: 1.7585, loss_cns_5: 0.6540, loss_yns_5: 0.1584, loss_cls_dn_0: 0.2380, loss_box_dn_0: 0.7929, loss_cls_dn_1: 0.1678, loss_box_dn_1: 0.7528, loss_cls_dn_2: 0.1729, loss_box_dn_2: 0.7468, loss_cls_dn_3: 0.1723, loss_box_dn_3: 0.7574, loss_cls_dn_4: 0.1800, loss_box_dn_4: 0.7774, loss_cls_dn_5: 0.1877, loss_box_dn_5: 0.8055, loss_dense_depth: 0.8167, loss: 27.7337, grad_norm: 43.1923 -2026-01-14 21:34:58,657 - mmdet - INFO - Iter [129/17500] lr: 1.512e-04, eta: 12:35:28, time: 1.576, data_time: 0.071, memory: 49164, loss_cls_0: 0.8551, loss_box_0: 1.7484, loss_cns_0: 0.6154, loss_yns_0: 0.1528, loss_cls_1: 0.9517, loss_box_1: 1.7548, loss_cns_1: 0.6459, loss_yns_1: 0.1558, loss_cls_2: 0.9873, loss_box_2: 1.7293, loss_cns_2: 0.6518, loss_yns_2: 0.1577, loss_cls_3: 0.9867, loss_box_3: 1.7001, loss_cns_3: 0.6533, loss_yns_3: 0.1582, loss_cls_4: 0.9934, loss_box_4: 1.7158, loss_cns_4: 0.6517, loss_yns_4: 0.1583, loss_cls_5: 0.9965, loss_box_5: 1.7155, loss_cns_5: 0.6528, loss_yns_5: 0.1594, loss_cls_dn_0: 0.2308, loss_box_dn_0: 0.7979, loss_cls_dn_1: 0.1652, loss_box_dn_1: 0.7775, loss_cls_dn_2: 0.1720, loss_box_dn_2: 0.7894, loss_cls_dn_3: 0.1732, loss_box_dn_3: 0.8114, loss_cls_dn_4: 0.1776, loss_box_dn_4: 0.8455, loss_cls_dn_5: 0.1877, loss_box_dn_5: 0.8743, loss_dense_depth: 0.7908, loss: 27.7412, grad_norm: 48.1232 -2026-01-14 21:35:00,271 - mmdet - INFO - Iter [130/17500] lr: 1.516e-04, eta: 12:33:12, time: 1.614, data_time: 0.075, memory: 49164, loss_cls_0: 0.8990, loss_box_0: 1.7874, loss_cns_0: 0.6182, loss_yns_0: 0.1520, loss_cls_1: 0.9588, loss_box_1: 1.7801, loss_cns_1: 0.6459, loss_yns_1: 0.1542, loss_cls_2: 0.9863, loss_box_2: 1.7657, loss_cns_2: 0.6521, loss_yns_2: 0.1533, loss_cls_3: 1.0001, loss_box_3: 1.7281, loss_cns_3: 0.6520, loss_yns_3: 0.1541, loss_cls_4: 1.0077, loss_box_4: 1.7342, loss_cns_4: 0.6486, loss_yns_4: 0.1555, loss_cls_5: 1.0150, loss_box_5: 1.7390, loss_cns_5: 0.6543, loss_yns_5: 0.1541, loss_cls_dn_0: 0.2307, loss_box_dn_0: 0.8048, loss_cls_dn_1: 0.1670, loss_box_dn_1: 0.8290, loss_cls_dn_2: 0.1723, loss_box_dn_2: 0.8487, loss_cls_dn_3: 0.1745, loss_box_dn_3: 0.8628, loss_cls_dn_4: 0.1770, loss_box_dn_4: 0.8895, loss_cls_dn_5: 0.1878, loss_box_dn_5: 0.9132, loss_dense_depth: 0.7964, loss: 28.2494, grad_norm: 46.7404 -2026-01-14 21:35:01,886 - mmdet - INFO - Iter [131/17500] lr: 1.520e-04, eta: 12:30:59, time: 1.616, data_time: 0.082, memory: 49164, loss_cls_0: 0.8943, loss_box_0: 1.7706, loss_cns_0: 0.6228, loss_yns_0: 0.1525, loss_cls_1: 0.9511, loss_box_1: 1.7892, loss_cns_1: 0.6490, loss_yns_1: 0.1543, loss_cls_2: 0.9838, loss_box_2: 1.7627, loss_cns_2: 0.6556, loss_yns_2: 0.1548, loss_cls_3: 0.9985, loss_box_3: 1.7301, loss_cns_3: 0.6584, loss_yns_3: 0.1570, loss_cls_4: 1.0281, loss_box_4: 1.6991, loss_cns_4: 0.6562, loss_yns_4: 0.1613, loss_cls_5: 0.9986, loss_box_5: 1.7153, loss_cns_5: 0.6607, loss_yns_5: 0.1557, loss_cls_dn_0: 0.2322, loss_box_dn_0: 0.7990, loss_cls_dn_1: 0.1653, loss_box_dn_1: 0.8172, loss_cls_dn_2: 0.1739, loss_box_dn_2: 0.8254, loss_cls_dn_3: 0.1731, loss_box_dn_3: 0.8293, loss_cls_dn_4: 0.1806, loss_box_dn_4: 0.8299, loss_cls_dn_5: 0.1869, loss_box_dn_5: 0.8508, loss_dense_depth: 0.8167, loss: 28.0402, grad_norm: 38.9896 -2026-01-14 21:35:03,480 - mmdet - INFO - Iter [132/17500] lr: 1.524e-04, eta: 12:28:41, time: 1.562, data_time: 0.076, memory: 49164, loss_cls_0: 0.8700, loss_box_0: 1.7428, loss_cns_0: 0.6182, loss_yns_0: 0.1491, loss_cls_1: 0.9371, loss_box_1: 1.8113, loss_cns_1: 0.6459, loss_yns_1: 0.1505, loss_cls_2: 0.9731, loss_box_2: 1.7742, loss_cns_2: 0.6554, loss_yns_2: 0.1505, loss_cls_3: 0.9817, loss_box_3: 1.7720, loss_cns_3: 0.6550, loss_yns_3: 0.1531, loss_cls_4: 0.9859, loss_box_4: 1.7342, loss_cns_4: 0.6577, loss_yns_4: 0.1562, loss_cls_5: 0.9934, loss_box_5: 1.7538, loss_cns_5: 0.6568, loss_yns_5: 0.1537, loss_cls_dn_0: 0.2356, loss_box_dn_0: 0.7861, loss_cls_dn_1: 0.1666, loss_box_dn_1: 0.7721, loss_cls_dn_2: 0.1779, loss_box_dn_2: 0.7674, loss_cls_dn_3: 0.1741, loss_box_dn_3: 0.7742, loss_cls_dn_4: 0.1829, loss_box_dn_4: 0.7644, loss_cls_dn_5: 0.1887, loss_box_dn_5: 0.7802, loss_dense_depth: 0.8235, loss: 27.7255, grad_norm: 46.6325 -2026-01-14 21:35:05,038 - mmdet - INFO - Iter [133/17500] lr: 1.528e-04, eta: 12:26:28, time: 1.590, data_time: 0.103, memory: 49164, loss_cls_0: 0.9038, loss_box_0: 1.7157, loss_cns_0: 0.6086, loss_yns_0: 0.1466, loss_cls_1: 0.9427, loss_box_1: 1.8325, loss_cns_1: 0.6427, loss_yns_1: 0.1499, loss_cls_2: 0.9820, loss_box_2: 1.7500, loss_cns_2: 0.6464, loss_yns_2: 0.1511, loss_cls_3: 1.0069, loss_box_3: 1.7572, loss_cns_3: 0.6522, loss_yns_3: 0.1518, loss_cls_4: 0.9951, loss_box_4: 1.7247, loss_cns_4: 0.6528, loss_yns_4: 0.1514, loss_cls_5: 0.9959, loss_box_5: 1.7260, loss_cns_5: 0.6544, loss_yns_5: 0.1519, loss_cls_dn_0: 0.2409, loss_box_dn_0: 0.7894, loss_cls_dn_1: 0.1692, loss_box_dn_1: 0.7460, loss_cls_dn_2: 0.1771, loss_box_dn_2: 0.7262, loss_cls_dn_3: 0.1725, loss_box_dn_3: 0.7310, loss_cls_dn_4: 0.1777, loss_box_dn_4: 0.7185, loss_cls_dn_5: 0.1878, loss_box_dn_5: 0.7181, loss_dense_depth: 0.7902, loss: 27.4370, grad_norm: 35.0868 -2026-01-14 21:35:06,605 - mmdet - INFO - Iter [134/17500] lr: 1.532e-04, eta: 12:24:14, time: 1.567, data_time: 0.074, memory: 49164, loss_cls_0: 0.8487, loss_box_0: 1.7052, loss_cns_0: 0.6210, loss_yns_0: 0.1488, loss_cls_1: 0.9147, loss_box_1: 1.8304, loss_cns_1: 0.6425, loss_yns_1: 0.1493, loss_cls_2: 0.9554, loss_box_2: 1.7562, loss_cns_2: 0.6506, loss_yns_2: 0.1504, loss_cls_3: 0.9622, loss_box_3: 1.7269, loss_cns_3: 0.6547, loss_yns_3: 0.1504, loss_cls_4: 0.9834, loss_box_4: 1.7211, loss_cns_4: 0.6548, loss_yns_4: 0.1544, loss_cls_5: 0.9795, loss_box_5: 1.7147, loss_cns_5: 0.6555, loss_yns_5: 0.1504, loss_cls_dn_0: 0.2299, loss_box_dn_0: 0.7884, loss_cls_dn_1: 0.1653, loss_box_dn_1: 0.7213, loss_cls_dn_2: 0.1701, loss_box_dn_2: 0.7007, loss_cls_dn_3: 0.1681, loss_box_dn_3: 0.6969, loss_cls_dn_4: 0.1692, loss_box_dn_4: 0.6972, loss_cls_dn_5: 0.1841, loss_box_dn_5: 0.6980, loss_dense_depth: 0.8444, loss: 27.1148, grad_norm: 39.1717 -2026-01-14 21:35:08,198 - mmdet - INFO - Iter [135/17500] lr: 1.536e-04, eta: 12:22:06, time: 1.594, data_time: 0.072, memory: 49164, loss_cls_0: 0.8815, loss_box_0: 1.7656, loss_cns_0: 0.6253, loss_yns_0: 0.1541, loss_cls_1: 0.9322, loss_box_1: 1.8195, loss_cns_1: 0.6439, loss_yns_1: 0.1520, loss_cls_2: 0.9595, loss_box_2: 1.7876, loss_cns_2: 0.6518, loss_yns_2: 0.1520, loss_cls_3: 0.9892, loss_box_3: 1.7787, loss_cns_3: 0.6512, loss_yns_3: 0.1554, loss_cls_4: 0.9947, loss_box_4: 1.8042, loss_cns_4: 0.6527, loss_yns_4: 0.1566, loss_cls_5: 0.9859, loss_box_5: 1.8049, loss_cns_5: 0.6528, loss_yns_5: 0.1539, loss_cls_dn_0: 0.2353, loss_box_dn_0: 0.7914, loss_cls_dn_1: 0.1634, loss_box_dn_1: 0.7385, loss_cls_dn_2: 0.1687, loss_box_dn_2: 0.7284, loss_cls_dn_3: 0.1716, loss_box_dn_3: 0.7379, loss_cls_dn_4: 0.1723, loss_box_dn_4: 0.7596, loss_cls_dn_5: 0.1809, loss_box_dn_5: 0.7748, loss_dense_depth: 0.8768, loss: 27.8045, grad_norm: 50.3198 -2026-01-14 21:35:09,804 - mmdet - INFO - Iter [136/17500] lr: 1.540e-04, eta: 12:20:01, time: 1.606, data_time: 0.074, memory: 49164, loss_cls_0: 0.8944, loss_box_0: 1.7147, loss_cns_0: 0.6291, loss_yns_0: 0.1496, loss_cls_1: 0.9096, loss_box_1: 1.8024, loss_cns_1: 0.6485, loss_yns_1: 0.1480, loss_cls_2: 0.9419, loss_box_2: 1.7559, loss_cns_2: 0.6586, loss_yns_2: 0.1519, loss_cls_3: 0.9618, loss_box_3: 1.7571, loss_cns_3: 0.6534, loss_yns_3: 0.1552, loss_cls_4: 0.9724, loss_box_4: 1.7523, loss_cns_4: 0.6550, loss_yns_4: 0.1535, loss_cls_5: 0.9486, loss_box_5: 1.7432, loss_cns_5: 0.6560, loss_yns_5: 0.1539, loss_cls_dn_0: 0.2267, loss_box_dn_0: 0.7793, loss_cls_dn_1: 0.1593, loss_box_dn_1: 0.7737, loss_cls_dn_2: 0.1635, loss_box_dn_2: 0.7734, loss_cls_dn_3: 0.1639, loss_box_dn_3: 0.7907, loss_cls_dn_4: 0.1682, loss_box_dn_4: 0.8127, loss_cls_dn_5: 0.1722, loss_box_dn_5: 0.8311, loss_dense_depth: 0.8490, loss: 27.6305, grad_norm: 46.1321 -2026-01-14 21:35:11,368 - mmdet - INFO - Iter [137/17500] lr: 1.544e-04, eta: 12:17:53, time: 1.564, data_time: 0.073, memory: 49164, loss_cls_0: 0.8375, loss_box_0: 1.7400, loss_cns_0: 0.6232, loss_yns_0: 0.1504, loss_cls_1: 0.9178, loss_box_1: 1.8046, loss_cns_1: 0.6477, loss_yns_1: 0.1483, loss_cls_2: 0.9391, loss_box_2: 1.7455, loss_cns_2: 0.6513, loss_yns_2: 0.1521, loss_cls_3: 0.9593, loss_box_3: 1.7392, loss_cns_3: 0.6519, loss_yns_3: 0.1518, loss_cls_4: 0.9519, loss_box_4: 1.7520, loss_cns_4: 0.6512, loss_yns_4: 0.1517, loss_cls_5: 0.9490, loss_box_5: 1.7956, loss_cns_5: 0.6442, loss_yns_5: 0.1513, loss_cls_dn_0: 0.2388, loss_box_dn_0: 0.7848, loss_cls_dn_1: 0.1564, loss_box_dn_1: 0.8329, loss_cls_dn_2: 0.1603, loss_box_dn_2: 0.8301, loss_cls_dn_3: 0.1637, loss_box_dn_3: 0.8432, loss_cls_dn_4: 0.1619, loss_box_dn_4: 0.8663, loss_cls_dn_5: 0.1692, loss_box_dn_5: 0.8929, loss_dense_depth: 0.8766, loss: 27.8836, grad_norm: 50.7997 -2026-01-14 21:35:13,055 - mmdet - INFO - Iter [138/17500] lr: 1.548e-04, eta: 12:15:56, time: 1.640, data_time: 0.076, memory: 49164, loss_cls_0: 0.8497, loss_box_0: 1.7076, loss_cns_0: 0.6262, loss_yns_0: 0.1478, loss_cls_1: 0.9177, loss_box_1: 1.8076, loss_cns_1: 0.6464, loss_yns_1: 0.1469, loss_cls_2: 0.9377, loss_box_2: 1.7438, loss_cns_2: 0.6503, loss_yns_2: 0.1501, loss_cls_3: 0.9460, loss_box_3: 1.7388, loss_cns_3: 0.6568, loss_yns_3: 0.1497, loss_cls_4: 0.9596, loss_box_4: 1.7472, loss_cns_4: 0.6581, loss_yns_4: 0.1491, loss_cls_5: 0.9484, loss_box_5: 1.7663, loss_cns_5: 0.6512, loss_yns_5: 0.1500, loss_cls_dn_0: 0.2394, loss_box_dn_0: 0.7865, loss_cls_dn_1: 0.1532, loss_box_dn_1: 0.8475, loss_cls_dn_2: 0.1580, loss_box_dn_2: 0.8331, loss_cls_dn_3: 0.1610, loss_box_dn_3: 0.8383, loss_cls_dn_4: 0.1622, loss_box_dn_4: 0.8551, loss_cls_dn_5: 0.1719, loss_box_dn_5: 0.8680, loss_dense_depth: 0.8544, loss: 27.7817, grad_norm: 37.8857 -2026-01-14 21:35:14,631 - mmdet - INFO - Iter [139/17500] lr: 1.552e-04, eta: 12:13:58, time: 1.622, data_time: 0.114, memory: 49164, loss_cls_0: 0.8313, loss_box_0: 1.6824, loss_cns_0: 0.6287, loss_yns_0: 0.1460, loss_cls_1: 0.8974, loss_box_1: 1.8236, loss_cns_1: 0.6447, loss_yns_1: 0.1469, loss_cls_2: 0.9309, loss_box_2: 1.7801, loss_cns_2: 0.6487, loss_yns_2: 0.1476, loss_cls_3: 0.9588, loss_box_3: 1.7735, loss_cns_3: 0.6488, loss_yns_3: 0.1487, loss_cls_4: 0.9348, loss_box_4: 1.7624, loss_cns_4: 0.6533, loss_yns_4: 0.1478, loss_cls_5: 0.9811, loss_box_5: 1.7357, loss_cns_5: 0.6570, loss_yns_5: 0.1482, loss_cls_dn_0: 0.2284, loss_box_dn_0: 0.7822, loss_cls_dn_1: 0.1526, loss_box_dn_1: 0.8110, loss_cls_dn_2: 0.1583, loss_box_dn_2: 0.8009, loss_cls_dn_3: 0.1591, loss_box_dn_3: 0.8007, loss_cls_dn_4: 0.1641, loss_box_dn_4: 0.8019, loss_cls_dn_5: 0.1818, loss_box_dn_5: 0.7932, loss_dense_depth: 0.8533, loss: 27.5460, grad_norm: 56.1621 -2026-01-14 21:35:16,239 - mmdet - INFO - Iter [140/17500] lr: 1.556e-04, eta: 12:12:00, time: 1.608, data_time: 0.087, memory: 49164, loss_cls_0: 0.8226, loss_box_0: 1.6785, loss_cns_0: 0.6293, loss_yns_0: 0.1449, loss_cls_1: 0.9014, loss_box_1: 1.7633, loss_cns_1: 0.6527, loss_yns_1: 0.1474, loss_cls_2: 0.9267, loss_box_2: 1.7128, loss_cns_2: 0.6610, loss_yns_2: 0.1473, loss_cls_3: 0.9454, loss_box_3: 1.6936, loss_cns_3: 0.6573, loss_yns_3: 0.1483, loss_cls_4: 0.9341, loss_box_4: 1.6948, loss_cns_4: 0.6584, loss_yns_4: 0.1478, loss_cls_5: 0.9371, loss_box_5: 1.6854, loss_cns_5: 0.6615, loss_yns_5: 0.1476, loss_cls_dn_0: 0.2238, loss_box_dn_0: 0.7848, loss_cls_dn_1: 0.1460, loss_box_dn_1: 0.7819, loss_cls_dn_2: 0.1532, loss_box_dn_2: 0.7662, loss_cls_dn_3: 0.1548, loss_box_dn_3: 0.7591, loss_cls_dn_4: 0.1595, loss_box_dn_4: 0.7562, loss_cls_dn_5: 0.1705, loss_box_dn_5: 0.7565, loss_dense_depth: 0.8056, loss: 26.9173, grad_norm: 39.4557 -2026-01-14 21:35:17,952 - mmdet - INFO - Iter [141/17500] lr: 1.560e-04, eta: 12:10:17, time: 1.710, data_time: 0.112, memory: 49164, loss_cls_0: 0.8544, loss_box_0: 1.7725, loss_cns_0: 0.6256, loss_yns_0: 0.1484, loss_cls_1: 0.9256, loss_box_1: 1.7806, loss_cns_1: 0.6517, loss_yns_1: 0.1460, loss_cls_2: 0.9550, loss_box_2: 1.7640, loss_cns_2: 0.6493, loss_yns_2: 0.1476, loss_cls_3: 0.9538, loss_box_3: 1.7772, loss_cns_3: 0.6544, loss_yns_3: 0.1493, loss_cls_4: 0.9662, loss_box_4: 1.7731, loss_cns_4: 0.6520, loss_yns_4: 0.1492, loss_cls_5: 0.9662, loss_box_5: 1.7953, loss_cns_5: 0.6473, loss_yns_5: 0.1505, loss_cls_dn_0: 0.2373, loss_box_dn_0: 0.7997, loss_cls_dn_1: 0.1515, loss_box_dn_1: 0.7424, loss_cls_dn_2: 0.1581, loss_box_dn_2: 0.7395, loss_cls_dn_3: 0.1619, loss_box_dn_3: 0.7507, loss_cls_dn_4: 0.1664, loss_box_dn_4: 0.7582, loss_cls_dn_5: 0.1735, loss_box_dn_5: 0.7828, loss_dense_depth: 0.8357, loss: 27.5132, grad_norm: 61.2260 -2026-01-14 21:35:19,710 - mmdet - INFO - Iter [142/17500] lr: 1.564e-04, eta: 12:08:41, time: 1.761, data_time: 0.171, memory: 49164, loss_cls_0: 0.8681, loss_box_0: 1.7707, loss_cns_0: 0.6217, loss_yns_0: 0.1507, loss_cls_1: 0.9387, loss_box_1: 1.8291, loss_cns_1: 0.6416, loss_yns_1: 0.1477, loss_cls_2: 0.9683, loss_box_2: 1.7680, loss_cns_2: 0.6463, loss_yns_2: 0.1497, loss_cls_3: 0.9701, loss_box_3: 1.7728, loss_cns_3: 0.6541, loss_yns_3: 0.1524, loss_cls_4: 0.9744, loss_box_4: 1.7764, loss_cns_4: 0.6530, loss_yns_4: 0.1528, loss_cls_5: 0.9905, loss_box_5: 1.7998, loss_cns_5: 0.6504, loss_yns_5: 0.1538, loss_cls_dn_0: 0.2340, loss_box_dn_0: 0.7906, loss_cls_dn_1: 0.1535, loss_box_dn_1: 0.7510, loss_cls_dn_2: 0.1585, loss_box_dn_2: 0.7491, loss_cls_dn_3: 0.1639, loss_box_dn_3: 0.7674, loss_cls_dn_4: 0.1695, loss_box_dn_4: 0.7875, loss_cls_dn_5: 0.1749, loss_box_dn_5: 0.8227, loss_dense_depth: 0.8269, loss: 27.7504, grad_norm: 54.5313 -2026-01-14 21:35:21,260 - mmdet - INFO - Iter [143/17500] lr: 1.568e-04, eta: 12:06:41, time: 1.548, data_time: 0.065, memory: 49164, loss_cls_0: 0.8443, loss_box_0: 1.7718, loss_cns_0: 0.6212, loss_yns_0: 0.1476, loss_cls_1: 0.9196, loss_box_1: 1.8263, loss_cns_1: 0.6456, loss_yns_1: 0.1498, loss_cls_2: 0.9455, loss_box_2: 1.8057, loss_cns_2: 0.6483, loss_yns_2: 0.1500, loss_cls_3: 0.9552, loss_box_3: 1.8088, loss_cns_3: 0.6489, loss_yns_3: 0.1521, loss_cls_4: 0.9544, loss_box_4: 1.8329, loss_cns_4: 0.6488, loss_yns_4: 0.1528, loss_cls_5: 0.9859, loss_box_5: 1.8834, loss_cns_5: 0.6517, loss_yns_5: 0.1515, loss_cls_dn_0: 0.2258, loss_box_dn_0: 0.7898, loss_cls_dn_1: 0.1570, loss_box_dn_1: 0.8119, loss_cls_dn_2: 0.1605, loss_box_dn_2: 0.8216, loss_cls_dn_3: 0.1616, loss_box_dn_3: 0.8446, loss_cls_dn_4: 0.1676, loss_box_dn_4: 0.8777, loss_cls_dn_5: 0.1762, loss_box_dn_5: 0.9275, loss_dense_depth: 0.8092, loss: 28.2332, grad_norm: 63.7160 -2026-01-14 21:35:22,874 - mmdet - INFO - Iter [144/17500] lr: 1.572e-04, eta: 12:04:44, time: 1.568, data_time: 0.076, memory: 49164, loss_cls_0: 0.8447, loss_box_0: 1.7996, loss_cns_0: 0.6126, loss_yns_0: 0.1464, loss_cls_1: 0.9140, loss_box_1: 1.8537, loss_cns_1: 0.6417, loss_yns_1: 0.1507, loss_cls_2: 0.9375, loss_box_2: 1.8311, loss_cns_2: 0.6495, loss_yns_2: 0.1518, loss_cls_3: 0.9587, loss_box_3: 1.8271, loss_cns_3: 0.6487, loss_yns_3: 0.1558, loss_cls_4: 0.9560, loss_box_4: 1.8366, loss_cns_4: 0.6469, loss_yns_4: 0.1558, loss_cls_5: 0.9751, loss_box_5: 1.8721, loss_cns_5: 0.6476, loss_yns_5: 0.1516, loss_cls_dn_0: 0.2340, loss_box_dn_0: 0.7984, loss_cls_dn_1: 0.1611, loss_box_dn_1: 0.8623, loss_cls_dn_2: 0.1619, loss_box_dn_2: 0.8649, loss_cls_dn_3: 0.1651, loss_box_dn_3: 0.8893, loss_cls_dn_4: 0.1657, loss_box_dn_4: 0.9165, loss_cls_dn_5: 0.1782, loss_box_dn_5: 0.9547, loss_dense_depth: 0.7941, loss: 28.5114, grad_norm: 63.9243 -2026-01-14 21:35:24,492 - mmdet - INFO - Iter [145/17500] lr: 1.576e-04, eta: 12:03:01, time: 1.666, data_time: 0.110, memory: 49164, loss_cls_0: 0.8350, loss_box_0: 1.7754, loss_cns_0: 0.6198, loss_yns_0: 0.1507, loss_cls_1: 0.9096, loss_box_1: 1.8094, loss_cns_1: 0.6475, loss_yns_1: 0.1522, loss_cls_2: 0.9438, loss_box_2: 1.7754, loss_cns_2: 0.6516, loss_yns_2: 0.1520, loss_cls_3: 0.9438, loss_box_3: 1.7843, loss_cns_3: 0.6542, loss_yns_3: 0.1554, loss_cls_4: 0.9598, loss_box_4: 1.7635, loss_cns_4: 0.6519, loss_yns_4: 0.1532, loss_cls_5: 0.9474, loss_box_5: 1.7559, loss_cns_5: 0.6544, loss_yns_5: 0.1515, loss_cls_dn_0: 0.2282, loss_box_dn_0: 0.7824, loss_cls_dn_1: 0.1583, loss_box_dn_1: 0.8945, loss_cls_dn_2: 0.1602, loss_box_dn_2: 0.8881, loss_cls_dn_3: 0.1699, loss_box_dn_3: 0.9073, loss_cls_dn_4: 0.1653, loss_box_dn_4: 0.9132, loss_cls_dn_5: 0.1730, loss_box_dn_5: 0.9283, loss_dense_depth: 0.8165, loss: 28.1831, grad_norm: 42.6806 -2026-01-14 21:35:26,069 - mmdet - INFO - Iter [146/17500] lr: 1.580e-04, eta: 12:01:09, time: 1.579, data_time: 0.077, memory: 49164, loss_cls_0: 0.8329, loss_box_0: 1.7608, loss_cns_0: 0.6187, loss_yns_0: 0.1478, loss_cls_1: 0.9150, loss_box_1: 1.7967, loss_cns_1: 0.6481, loss_yns_1: 0.1486, loss_cls_2: 0.9601, loss_box_2: 1.7994, loss_cns_2: 0.6463, loss_yns_2: 0.1502, loss_cls_3: 0.9468, loss_box_3: 1.8184, loss_cns_3: 0.6447, loss_yns_3: 0.1501, loss_cls_4: 0.9602, loss_box_4: 1.8151, loss_cns_4: 0.6461, loss_yns_4: 0.1505, loss_cls_5: 0.9776, loss_box_5: 1.7629, loss_cns_5: 0.6505, loss_yns_5: 0.1507, loss_cls_dn_0: 0.2245, loss_box_dn_0: 0.7806, loss_cls_dn_1: 0.1515, loss_box_dn_1: 0.8184, loss_cls_dn_2: 0.1565, loss_box_dn_2: 0.8167, loss_cls_dn_3: 0.1622, loss_box_dn_3: 0.8287, loss_cls_dn_4: 0.1634, loss_box_dn_4: 0.8275, loss_cls_dn_5: 0.1682, loss_box_dn_5: 0.8182, loss_dense_depth: 0.7806, loss: 27.7953, grad_norm: 62.9242 -2026-01-14 21:35:27,696 - mmdet - INFO - Iter [147/17500] lr: 1.584e-04, eta: 11:59:25, time: 1.626, data_time: 0.074, memory: 49164, loss_cls_0: 0.8332, loss_box_0: 1.7569, loss_cns_0: 0.6208, loss_yns_0: 0.1477, loss_cls_1: 0.9082, loss_box_1: 1.7619, loss_cns_1: 0.6512, loss_yns_1: 0.1487, loss_cls_2: 0.9593, loss_box_2: 1.7330, loss_cns_2: 0.6594, loss_yns_2: 0.1508, loss_cls_3: 0.9401, loss_box_3: 1.7427, loss_cns_3: 0.6588, loss_yns_3: 0.1494, loss_cls_4: 0.9546, loss_box_4: 1.7490, loss_cns_4: 0.6600, loss_yns_4: 0.1489, loss_cls_5: 0.9976, loss_box_5: 1.7039, loss_cns_5: 0.6634, loss_yns_5: 0.1505, loss_cls_dn_0: 0.2238, loss_box_dn_0: 0.7810, loss_cls_dn_1: 0.1510, loss_box_dn_1: 0.7857, loss_cls_dn_2: 0.1563, loss_box_dn_2: 0.7751, loss_cls_dn_3: 0.1559, loss_box_dn_3: 0.7799, loss_cls_dn_4: 0.1607, loss_box_dn_4: 0.7773, loss_cls_dn_5: 0.1638, loss_box_dn_5: 0.7692, loss_dense_depth: 0.7989, loss: 27.3286, grad_norm: 45.5875 -2026-01-14 21:35:29,276 - mmdet - INFO - Iter [148/17500] lr: 1.588e-04, eta: 11:57:35, time: 1.578, data_time: 0.090, memory: 49164, loss_cls_0: 0.8190, loss_box_0: 1.7467, loss_cns_0: 0.6214, loss_yns_0: 0.1463, loss_cls_1: 0.9065, loss_box_1: 1.7399, loss_cns_1: 0.6524, loss_yns_1: 0.1461, loss_cls_2: 0.9361, loss_box_2: 1.6848, loss_cns_2: 0.6666, loss_yns_2: 0.1483, loss_cls_3: 0.9409, loss_box_3: 1.6838, loss_cns_3: 0.6619, loss_yns_3: 0.1490, loss_cls_4: 0.9601, loss_box_4: 1.7036, loss_cns_4: 0.6652, loss_yns_4: 0.1477, loss_cls_5: 0.9512, loss_box_5: 1.7218, loss_cns_5: 0.6576, loss_yns_5: 0.1486, loss_cls_dn_0: 0.2201, loss_box_dn_0: 0.7922, loss_cls_dn_1: 0.1491, loss_box_dn_1: 0.7352, loss_cls_dn_2: 0.1535, loss_box_dn_2: 0.7144, loss_cls_dn_3: 0.1560, loss_box_dn_3: 0.7200, loss_cls_dn_4: 0.1635, loss_box_dn_4: 0.7251, loss_cls_dn_5: 0.1676, loss_box_dn_5: 0.7429, loss_dense_depth: 0.7655, loss: 26.8105, grad_norm: 48.1424 -2026-01-14 21:35:30,857 - mmdet - INFO - Iter [149/17500] lr: 1.592e-04, eta: 11:55:48, time: 1.583, data_time: 0.088, memory: 49164, loss_cls_0: 0.8214, loss_box_0: 1.7510, loss_cns_0: 0.6209, loss_yns_0: 0.1469, loss_cls_1: 0.9131, loss_box_1: 1.8113, loss_cns_1: 0.6456, loss_yns_1: 0.1467, loss_cls_2: 0.9514, loss_box_2: 1.7721, loss_cns_2: 0.6522, loss_yns_2: 0.1467, loss_cls_3: 0.9473, loss_box_3: 1.7742, loss_cns_3: 0.6496, loss_yns_3: 0.1511, loss_cls_4: 0.9658, loss_box_4: 1.7675, loss_cns_4: 0.6489, loss_yns_4: 0.1500, loss_cls_5: 0.9594, loss_box_5: 1.7878, loss_cns_5: 0.6471, loss_yns_5: 0.1482, loss_cls_dn_0: 0.2174, loss_box_dn_0: 0.7811, loss_cls_dn_1: 0.1501, loss_box_dn_1: 0.7727, loss_cls_dn_2: 0.1559, loss_box_dn_2: 0.7613, loss_cls_dn_3: 0.1585, loss_box_dn_3: 0.7778, loss_cls_dn_4: 0.1664, loss_box_dn_4: 0.7943, loss_cls_dn_5: 0.1714, loss_box_dn_5: 0.8216, loss_dense_depth: 0.7512, loss: 27.4558, grad_norm: 58.6477 -2026-01-14 21:35:32,456 - mmdet - INFO - Iter [150/17500] lr: 1.596e-04, eta: 11:54:04, time: 1.598, data_time: 0.078, memory: 49164, loss_cls_0: 0.8503, loss_box_0: 1.7539, loss_cns_0: 0.6181, loss_yns_0: 0.1458, loss_cls_1: 0.9162, loss_box_1: 1.7732, loss_cns_1: 0.6467, loss_yns_1: 0.1461, loss_cls_2: 0.9631, loss_box_2: 1.7479, loss_cns_2: 0.6525, loss_yns_2: 0.1453, loss_cls_3: 0.9675, loss_box_3: 1.7490, loss_cns_3: 0.6592, loss_yns_3: 0.1518, loss_cls_4: 0.9610, loss_box_4: 1.7344, loss_cns_4: 0.6540, loss_yns_4: 0.1571, loss_cls_5: 0.9685, loss_box_5: 1.7363, loss_cns_5: 0.6525, loss_yns_5: 0.1486, loss_cls_dn_0: 0.2186, loss_box_dn_0: 0.7759, loss_cls_dn_1: 0.1521, loss_box_dn_1: 0.8075, loss_cls_dn_2: 0.1581, loss_box_dn_2: 0.8200, loss_cls_dn_3: 0.1626, loss_box_dn_3: 0.8422, loss_cls_dn_4: 0.1679, loss_box_dn_4: 0.8683, loss_cls_dn_5: 0.1748, loss_box_dn_5: 0.8950, loss_dense_depth: 0.7752, loss: 27.7171, grad_norm: 52.1710 -2026-01-14 21:35:34,020 - mmdet - INFO - Iter [151/17500] lr: 1.600e-04, eta: 11:52:18, time: 1.565, data_time: 0.069, memory: 49164, loss_cls_0: 0.8440, loss_box_0: 1.7586, loss_cns_0: 0.6185, loss_yns_0: 0.1475, loss_cls_1: 0.9050, loss_box_1: 1.8080, loss_cns_1: 0.6423, loss_yns_1: 0.1480, loss_cls_2: 0.9430, loss_box_2: 1.7380, loss_cns_2: 0.6498, loss_yns_2: 0.1484, loss_cls_3: 0.9557, loss_box_3: 1.7478, loss_cns_3: 0.6564, loss_yns_3: 0.1511, loss_cls_4: 0.9430, loss_box_4: 1.7429, loss_cns_4: 0.6546, loss_yns_4: 0.1524, loss_cls_5: 0.9534, loss_box_5: 1.7626, loss_cns_5: 0.6523, loss_yns_5: 0.1500, loss_cls_dn_0: 0.2151, loss_box_dn_0: 0.7848, loss_cls_dn_1: 0.1481, loss_box_dn_1: 0.8398, loss_cls_dn_2: 0.1516, loss_box_dn_2: 0.8334, loss_cls_dn_3: 0.1546, loss_box_dn_3: 0.8489, loss_cls_dn_4: 0.1572, loss_box_dn_4: 0.8716, loss_cls_dn_5: 0.1649, loss_box_dn_5: 0.8980, loss_dense_depth: 0.7823, loss: 27.7236, grad_norm: 54.8539 -2026-01-14 21:35:35,624 - mmdet - INFO - Iter [152/17500] lr: 1.604e-04, eta: 11:50:34, time: 1.576, data_time: 0.075, memory: 49164, loss_cls_0: 0.8275, loss_box_0: 1.7221, loss_cns_0: 0.6209, loss_yns_0: 0.1501, loss_cls_1: 0.9111, loss_box_1: 1.7891, loss_cns_1: 0.6448, loss_yns_1: 0.1500, loss_cls_2: 0.9331, loss_box_2: 1.7372, loss_cns_2: 0.6541, loss_yns_2: 0.1508, loss_cls_3: 0.9384, loss_box_3: 1.7234, loss_cns_3: 0.6544, loss_yns_3: 0.1499, loss_cls_4: 0.9479, loss_box_4: 1.7095, loss_cns_4: 0.6565, loss_yns_4: 0.1503, loss_cls_5: 0.9470, loss_box_5: 1.7147, loss_cns_5: 0.6541, loss_yns_5: 0.1502, loss_cls_dn_0: 0.2161, loss_box_dn_0: 0.7743, loss_cls_dn_1: 0.1497, loss_box_dn_1: 0.8507, loss_cls_dn_2: 0.1525, loss_box_dn_2: 0.8367, loss_cls_dn_3: 0.1535, loss_box_dn_3: 0.8376, loss_cls_dn_4: 0.1605, loss_box_dn_4: 0.8485, loss_cls_dn_5: 0.1655, loss_box_dn_5: 0.8633, loss_dense_depth: 0.7542, loss: 27.4501, grad_norm: 47.2588 -2026-01-14 21:35:37,193 - mmdet - INFO - Iter [153/17500] lr: 1.608e-04, eta: 11:48:54, time: 1.597, data_time: 0.110, memory: 49164, loss_cls_0: 0.8539, loss_box_0: 1.7357, loss_cns_0: 0.6185, loss_yns_0: 0.1487, loss_cls_1: 0.9171, loss_box_1: 1.7364, loss_cns_1: 0.6488, loss_yns_1: 0.1502, loss_cls_2: 0.9479, loss_box_2: 1.7512, loss_cns_2: 0.6516, loss_yns_2: 0.1498, loss_cls_3: 0.9820, loss_box_3: 1.7195, loss_cns_3: 0.6519, loss_yns_3: 0.1543, loss_cls_4: 0.9618, loss_box_4: 1.7022, loss_cns_4: 0.6560, loss_yns_4: 0.1552, loss_cls_5: 0.9623, loss_box_5: 1.6837, loss_cns_5: 0.6534, loss_yns_5: 0.1485, loss_cls_dn_0: 0.2282, loss_box_dn_0: 0.7831, loss_cls_dn_1: 0.1488, loss_box_dn_1: 0.7793, loss_cls_dn_2: 0.1506, loss_box_dn_2: 0.7816, loss_cls_dn_3: 0.1557, loss_box_dn_3: 0.7784, loss_cls_dn_4: 0.1623, loss_box_dn_4: 0.7778, loss_cls_dn_5: 0.1661, loss_box_dn_5: 0.7781, loss_dense_depth: 0.7843, loss: 27.2151, grad_norm: 46.2675 -2026-01-14 21:35:38,853 - mmdet - INFO - Iter [154/17500] lr: 1.612e-04, eta: 11:47:22, time: 1.657, data_time: 0.080, memory: 49164, loss_cls_0: 0.8295, loss_box_0: 1.7019, loss_cns_0: 0.6192, loss_yns_0: 0.1469, loss_cls_1: 0.9006, loss_box_1: 1.6997, loss_cns_1: 0.6452, loss_yns_1: 0.1481, loss_cls_2: 0.9467, loss_box_2: 1.6673, loss_cns_2: 0.6521, loss_yns_2: 0.1477, loss_cls_3: 0.9724, loss_box_3: 1.6527, loss_cns_3: 0.6522, loss_yns_3: 0.1530, loss_cls_4: 0.9566, loss_box_4: 1.6590, loss_cns_4: 0.6542, loss_yns_4: 0.1513, loss_cls_5: 0.9637, loss_box_5: 1.6623, loss_cns_5: 0.6524, loss_yns_5: 0.1490, loss_cls_dn_0: 0.2197, loss_box_dn_0: 0.7697, loss_cls_dn_1: 0.1472, loss_box_dn_1: 0.7557, loss_cls_dn_2: 0.1511, loss_box_dn_2: 0.7434, loss_cls_dn_3: 0.1561, loss_box_dn_3: 0.7484, loss_cls_dn_4: 0.1576, loss_box_dn_4: 0.7493, loss_cls_dn_5: 0.1603, loss_box_dn_5: 0.7559, loss_dense_depth: 0.7301, loss: 26.6279, grad_norm: 40.5230 -2026-01-14 21:35:40,451 - mmdet - INFO - Iter [155/17500] lr: 1.616e-04, eta: 11:45:42, time: 1.569, data_time: 0.077, memory: 49164, loss_cls_0: 0.8044, loss_box_0: 1.6887, loss_cns_0: 0.6234, loss_yns_0: 0.1487, loss_cls_1: 0.8873, loss_box_1: 1.7003, loss_cns_1: 0.6462, loss_yns_1: 0.1478, loss_cls_2: 0.9269, loss_box_2: 1.6603, loss_cns_2: 0.6548, loss_yns_2: 0.1497, loss_cls_3: 0.9422, loss_box_3: 1.6494, loss_cns_3: 0.6528, loss_yns_3: 0.1500, loss_cls_4: 0.9329, loss_box_4: 1.6821, loss_cns_4: 0.6514, loss_yns_4: 0.1483, loss_cls_5: 0.9437, loss_box_5: 1.6989, loss_cns_5: 0.6496, loss_yns_5: 0.1530, loss_cls_dn_0: 0.2135, loss_box_dn_0: 0.7750, loss_cls_dn_1: 0.1456, loss_box_dn_1: 0.7477, loss_cls_dn_2: 0.1469, loss_box_dn_2: 0.7351, loss_cls_dn_3: 0.1544, loss_box_dn_3: 0.7425, loss_cls_dn_4: 0.1512, loss_box_dn_4: 0.7605, loss_cls_dn_5: 0.1636, loss_box_dn_5: 0.7720, loss_dense_depth: 0.7946, loss: 26.5954, grad_norm: 52.7315 -2026-01-14 21:35:42,031 - mmdet - INFO - Iter [156/17500] lr: 1.620e-04, eta: 11:44:06, time: 1.606, data_time: 0.110, memory: 49164, loss_cls_0: 0.8253, loss_box_0: 1.7132, loss_cns_0: 0.6219, loss_yns_0: 0.1501, loss_cls_1: 0.8856, loss_box_1: 1.6922, loss_cns_1: 0.6464, loss_yns_1: 0.1492, loss_cls_2: 0.9199, loss_box_2: 1.6593, loss_cns_2: 0.6597, loss_yns_2: 0.1528, loss_cls_3: 0.9309, loss_box_3: 1.6394, loss_cns_3: 0.6566, loss_yns_3: 0.1502, loss_cls_4: 0.9383, loss_box_4: 1.6414, loss_cns_4: 0.6554, loss_yns_4: 0.1502, loss_cls_5: 0.9415, loss_box_5: 1.6569, loss_cns_5: 0.6587, loss_yns_5: 0.1523, loss_cls_dn_0: 0.2181, loss_box_dn_0: 0.7796, loss_cls_dn_1: 0.1472, loss_box_dn_1: 0.7701, loss_cls_dn_2: 0.1483, loss_box_dn_2: 0.7637, loss_cls_dn_3: 0.1532, loss_box_dn_3: 0.7692, loss_cls_dn_4: 0.1524, loss_box_dn_4: 0.7850, loss_cls_dn_5: 0.1645, loss_box_dn_5: 0.8000, loss_dense_depth: 0.7578, loss: 26.6568, grad_norm: 44.4687 -2026-01-14 21:35:43,599 - mmdet - INFO - Iter [157/17500] lr: 1.624e-04, eta: 11:42:29, time: 1.573, data_time: 0.076, memory: 49164, loss_cls_0: 0.8213, loss_box_0: 1.7223, loss_cns_0: 0.6214, loss_yns_0: 0.1522, loss_cls_1: 0.8839, loss_box_1: 1.7274, loss_cns_1: 0.6438, loss_yns_1: 0.1491, loss_cls_2: 0.9289, loss_box_2: 1.6920, loss_cns_2: 0.6631, loss_yns_2: 0.1525, loss_cls_3: 0.9398, loss_box_3: 1.6930, loss_cns_3: 0.6552, loss_yns_3: 0.1522, loss_cls_4: 0.9531, loss_box_4: 1.6895, loss_cns_4: 0.6526, loss_yns_4: 0.1515, loss_cls_5: 0.9381, loss_box_5: 1.7294, loss_cns_5: 0.6534, loss_yns_5: 0.1534, loss_cls_dn_0: 0.2134, loss_box_dn_0: 0.7762, loss_cls_dn_1: 0.1445, loss_box_dn_1: 0.7588, loss_cls_dn_2: 0.1476, loss_box_dn_2: 0.7639, loss_cls_dn_3: 0.1493, loss_box_dn_3: 0.7823, loss_cls_dn_4: 0.1527, loss_box_dn_4: 0.7987, loss_cls_dn_5: 0.1574, loss_box_dn_5: 0.8221, loss_dense_depth: 0.7595, loss: 26.9455, grad_norm: 53.0739 -2026-01-14 21:35:45,228 - mmdet - INFO - Iter [158/17500] lr: 1.628e-04, eta: 11:40:58, time: 1.628, data_time: 0.078, memory: 49164, loss_cls_0: 0.7926, loss_box_0: 1.7117, loss_cns_0: 0.6191, loss_yns_0: 0.1523, loss_cls_1: 0.8709, loss_box_1: 1.7139, loss_cns_1: 0.6481, loss_yns_1: 0.1512, loss_cls_2: 0.9212, loss_box_2: 1.6634, loss_cns_2: 0.6631, loss_yns_2: 0.1519, loss_cls_3: 0.9454, loss_box_3: 1.6666, loss_cns_3: 0.6592, loss_yns_3: 0.1528, loss_cls_4: 0.9215, loss_box_4: 1.6713, loss_cns_4: 0.6581, loss_yns_4: 0.1520, loss_cls_5: 0.9216, loss_box_5: 1.6883, loss_cns_5: 0.6584, loss_yns_5: 0.1587, loss_cls_dn_0: 0.2074, loss_box_dn_0: 0.7739, loss_cls_dn_1: 0.1390, loss_box_dn_1: 0.7811, loss_cls_dn_2: 0.1466, loss_box_dn_2: 0.7801, loss_cls_dn_3: 0.1482, loss_box_dn_3: 0.8012, loss_cls_dn_4: 0.1525, loss_box_dn_4: 0.8128, loss_cls_dn_5: 0.1604, loss_box_dn_5: 0.8301, loss_dense_depth: 0.7418, loss: 26.7889, grad_norm: 49.8478 -2026-01-14 21:35:46,814 - mmdet - INFO - Iter [159/17500] lr: 1.632e-04, eta: 11:39:24, time: 1.583, data_time: 0.085, memory: 49164, loss_cls_0: 0.7925, loss_box_0: 1.7302, loss_cns_0: 0.6174, loss_yns_0: 0.1529, loss_cls_1: 0.8796, loss_box_1: 1.7153, loss_cns_1: 0.6535, loss_yns_1: 0.1526, loss_cls_2: 0.9134, loss_box_2: 1.6947, loss_cns_2: 0.6568, loss_yns_2: 0.1522, loss_cls_3: 0.9408, loss_box_3: 1.6926, loss_cns_3: 0.6575, loss_yns_3: 0.1526, loss_cls_4: 0.9281, loss_box_4: 1.6807, loss_cns_4: 0.6573, loss_yns_4: 0.1546, loss_cls_5: 0.9416, loss_box_5: 1.6599, loss_cns_5: 0.6595, loss_yns_5: 0.1547, loss_cls_dn_0: 0.2086, loss_box_dn_0: 0.7780, loss_cls_dn_1: 0.1407, loss_box_dn_1: 0.7883, loss_cls_dn_2: 0.1453, loss_box_dn_2: 0.7941, loss_cls_dn_3: 0.1487, loss_box_dn_3: 0.8118, loss_cls_dn_4: 0.1573, loss_box_dn_4: 0.8180, loss_cls_dn_5: 0.1648, loss_box_dn_5: 0.8224, loss_dense_depth: 0.7920, loss: 26.9609, grad_norm: 59.1067 -2026-01-14 21:35:48,402 - mmdet - INFO - Iter [160/17500] lr: 1.636e-04, eta: 11:37:52, time: 1.591, data_time: 0.090, memory: 49164, loss_cls_0: 0.7929, loss_box_0: 1.7138, loss_cns_0: 0.6154, loss_yns_0: 0.1535, loss_cls_1: 0.8804, loss_box_1: 1.6991, loss_cns_1: 0.6526, loss_yns_1: 0.1525, loss_cls_2: 0.9124, loss_box_2: 1.6882, loss_cns_2: 0.6538, loss_yns_2: 0.1525, loss_cls_3: 0.9245, loss_box_3: 1.6810, loss_cns_3: 0.6558, loss_yns_3: 0.1536, loss_cls_4: 0.9169, loss_box_4: 1.6683, loss_cns_4: 0.6553, loss_yns_4: 0.1536, loss_cls_5: 0.9344, loss_box_5: 1.6508, loss_cns_5: 0.6571, loss_yns_5: 0.1539, loss_cls_dn_0: 0.2096, loss_box_dn_0: 0.7726, loss_cls_dn_1: 0.1408, loss_box_dn_1: 0.7712, loss_cls_dn_2: 0.1413, loss_box_dn_2: 0.7833, loss_cls_dn_3: 0.1448, loss_box_dn_3: 0.7908, loss_cls_dn_4: 0.1514, loss_box_dn_4: 0.7982, loss_cls_dn_5: 0.1575, loss_box_dn_5: 0.8032, loss_dense_depth: 0.7524, loss: 26.6896, grad_norm: 45.6469 -2026-01-14 21:35:50,091 - mmdet - INFO - Iter [161/17500] lr: 1.640e-04, eta: 11:36:31, time: 1.687, data_time: 0.114, memory: 49164, loss_cls_0: 0.7841, loss_box_0: 1.7225, loss_cns_0: 0.6183, loss_yns_0: 0.1534, loss_cls_1: 0.8774, loss_box_1: 1.7318, loss_cns_1: 0.6507, loss_yns_1: 0.1554, loss_cls_2: 0.9044, loss_box_2: 1.6600, loss_cns_2: 0.6581, loss_yns_2: 0.1548, loss_cls_3: 0.9325, loss_box_3: 1.6220, loss_cns_3: 0.6580, loss_yns_3: 0.1568, loss_cls_4: 0.9225, loss_box_4: 1.6301, loss_cns_4: 0.6586, loss_yns_4: 0.1555, loss_cls_5: 0.9329, loss_box_5: 1.6283, loss_cns_5: 0.6612, loss_yns_5: 0.1553, loss_cls_dn_0: 0.2056, loss_box_dn_0: 0.7775, loss_cls_dn_1: 0.1400, loss_box_dn_1: 0.7377, loss_cls_dn_2: 0.1407, loss_box_dn_2: 0.7276, loss_cls_dn_3: 0.1433, loss_box_dn_3: 0.7241, loss_cls_dn_4: 0.1476, loss_box_dn_4: 0.7374, loss_cls_dn_5: 0.1521, loss_box_dn_5: 0.7500, loss_dense_depth: 0.7971, loss: 26.3651, grad_norm: 39.2234 -2026-01-14 21:35:51,768 - mmdet - INFO - Iter [162/17500] lr: 1.644e-04, eta: 11:35:10, time: 1.676, data_time: 0.168, memory: 49164, loss_cls_0: 0.8124, loss_box_0: 1.7505, loss_cns_0: 0.6162, loss_yns_0: 0.1549, loss_cls_1: 0.8781, loss_box_1: 1.7528, loss_cns_1: 0.6455, loss_yns_1: 0.1560, loss_cls_2: 0.9170, loss_box_2: 1.6780, loss_cns_2: 0.6535, loss_yns_2: 0.1574, loss_cls_3: 0.9352, loss_box_3: 1.6602, loss_cns_3: 0.6564, loss_yns_3: 0.1579, loss_cls_4: 0.9243, loss_box_4: 1.6631, loss_cns_4: 0.6560, loss_yns_4: 0.1574, loss_cls_5: 0.9299, loss_box_5: 1.6569, loss_cns_5: 0.6590, loss_yns_5: 0.1584, loss_cls_dn_0: 0.2092, loss_box_dn_0: 0.7892, loss_cls_dn_1: 0.1354, loss_box_dn_1: 0.7504, loss_cls_dn_2: 0.1408, loss_box_dn_2: 0.7287, loss_cls_dn_3: 0.1449, loss_box_dn_3: 0.7310, loss_cls_dn_4: 0.1460, loss_box_dn_4: 0.7407, loss_cls_dn_5: 0.1516, loss_box_dn_5: 0.7508, loss_dense_depth: 0.8461, loss: 26.6518, grad_norm: 49.9927 -2026-01-14 21:35:53,329 - mmdet - INFO - Iter [163/17500] lr: 1.648e-04, eta: 11:33:38, time: 1.562, data_time: 0.074, memory: 49164, loss_cls_0: 0.7981, loss_box_0: 1.7417, loss_cns_0: 0.6209, loss_yns_0: 0.1564, loss_cls_1: 0.8904, loss_box_1: 1.6990, loss_cns_1: 0.6504, loss_yns_1: 0.1560, loss_cls_2: 0.9171, loss_box_2: 1.6601, loss_cns_2: 0.6549, loss_yns_2: 0.1571, loss_cls_3: 0.9133, loss_box_3: 1.6360, loss_cns_3: 0.6580, loss_yns_3: 0.1573, loss_cls_4: 0.9173, loss_box_4: 1.6332, loss_cns_4: 0.6575, loss_yns_4: 0.1577, loss_cls_5: 0.9246, loss_box_5: 1.6366, loss_cns_5: 0.6562, loss_yns_5: 0.1570, loss_cls_dn_0: 0.2076, loss_box_dn_0: 0.7677, loss_cls_dn_1: 0.1411, loss_box_dn_1: 0.7351, loss_cls_dn_2: 0.1431, loss_box_dn_2: 0.7212, loss_cls_dn_3: 0.1451, loss_box_dn_3: 0.7199, loss_cls_dn_4: 0.1530, loss_box_dn_4: 0.7287, loss_cls_dn_5: 0.1541, loss_box_dn_5: 0.7390, loss_dense_depth: 0.7317, loss: 26.2941, grad_norm: 31.2308 -2026-01-14 21:35:54,932 - mmdet - INFO - Iter [164/17500] lr: 1.652e-04, eta: 11:32:08, time: 1.574, data_time: 0.073, memory: 49164, loss_cls_0: 0.8099, loss_box_0: 1.7419, loss_cns_0: 0.6208, loss_yns_0: 0.1570, loss_cls_1: 0.8901, loss_box_1: 1.6778, loss_cns_1: 0.6568, loss_yns_1: 0.1564, loss_cls_2: 0.9103, loss_box_2: 1.6732, loss_cns_2: 0.6584, loss_yns_2: 0.1555, loss_cls_3: 0.9162, loss_box_3: 1.6329, loss_cns_3: 0.6607, loss_yns_3: 0.1582, loss_cls_4: 0.9143, loss_box_4: 1.6368, loss_cns_4: 0.6597, loss_yns_4: 0.1558, loss_cls_5: 0.9264, loss_box_5: 1.6360, loss_cns_5: 0.6600, loss_yns_5: 0.1577, loss_cls_dn_0: 0.2081, loss_box_dn_0: 0.7706, loss_cls_dn_1: 0.1359, loss_box_dn_1: 0.7346, loss_cls_dn_2: 0.1387, loss_box_dn_2: 0.7295, loss_cls_dn_3: 0.1414, loss_box_dn_3: 0.7205, loss_cls_dn_4: 0.1492, loss_box_dn_4: 0.7313, loss_cls_dn_5: 0.1518, loss_box_dn_5: 0.7396, loss_dense_depth: 0.8483, loss: 26.4223, grad_norm: 46.8690 -2026-01-14 21:35:56,554 - mmdet - INFO - Iter [165/17500] lr: 1.656e-04, eta: 11:30:47, time: 1.650, data_time: 0.099, memory: 49164, loss_cls_0: 0.8283, loss_box_0: 1.7556, loss_cns_0: 0.6132, loss_yns_0: 0.1559, loss_cls_1: 0.8884, loss_box_1: 1.6821, loss_cns_1: 0.6544, loss_yns_1: 0.1582, loss_cls_2: 0.9131, loss_box_2: 1.6653, loss_cns_2: 0.6590, loss_yns_2: 0.1590, loss_cls_3: 0.9347, loss_box_3: 1.6320, loss_cns_3: 0.6613, loss_yns_3: 0.1598, loss_cls_4: 0.9200, loss_box_4: 1.6450, loss_cns_4: 0.6607, loss_yns_4: 0.1577, loss_cls_5: 0.9303, loss_box_5: 1.6432, loss_cns_5: 0.6647, loss_yns_5: 0.1599, loss_cls_dn_0: 0.2108, loss_box_dn_0: 0.7705, loss_cls_dn_1: 0.1352, loss_box_dn_1: 0.7337, loss_cls_dn_2: 0.1377, loss_box_dn_2: 0.7268, loss_cls_dn_3: 0.1407, loss_box_dn_3: 0.7272, loss_cls_dn_4: 0.1416, loss_box_dn_4: 0.7445, loss_cls_dn_5: 0.1528, loss_box_dn_5: 0.7532, loss_dense_depth: 0.7514, loss: 26.4280, grad_norm: 45.0069 -2026-01-14 21:35:58,170 - mmdet - INFO - Iter [166/17500] lr: 1.660e-04, eta: 11:29:21, time: 1.589, data_time: 0.074, memory: 49164, loss_cls_0: 0.7969, loss_box_0: 1.7605, loss_cns_0: 0.6117, loss_yns_0: 0.1559, loss_cls_1: 0.8969, loss_box_1: 1.6636, loss_cns_1: 0.6565, loss_yns_1: 0.1581, loss_cls_2: 0.9102, loss_box_2: 1.6470, loss_cns_2: 0.6596, loss_yns_2: 0.1594, loss_cls_3: 0.9240, loss_box_3: 1.6357, loss_cns_3: 0.6615, loss_yns_3: 0.1577, loss_cls_4: 0.9228, loss_box_4: 1.6324, loss_cns_4: 0.6619, loss_yns_4: 0.1595, loss_cls_5: 0.9274, loss_box_5: 1.6239, loss_cns_5: 0.6646, loss_yns_5: 0.1609, loss_cls_dn_0: 0.2032, loss_box_dn_0: 0.7631, loss_cls_dn_1: 0.1354, loss_box_dn_1: 0.7424, loss_cls_dn_2: 0.1369, loss_box_dn_2: 0.7388, loss_cls_dn_3: 0.1395, loss_box_dn_3: 0.7495, loss_cls_dn_4: 0.1416, loss_box_dn_4: 0.7637, loss_cls_dn_5: 0.1494, loss_box_dn_5: 0.7659, loss_dense_depth: 0.8129, loss: 26.4509, grad_norm: 46.7545 -2026-01-14 21:36:05,136 - mmdet - INFO - Iter [167/17500] lr: 1.664e-04, eta: 11:37:17, time: 6.995, data_time: 0.102, memory: 49164, loss_cls_0: 0.8470, loss_box_0: 1.7780, loss_cns_0: 0.6147, loss_yns_0: 0.1618, loss_cls_1: 0.9154, loss_box_1: 1.6828, loss_cns_1: 0.6508, loss_yns_1: 0.1594, loss_cls_2: 0.9367, loss_box_2: 1.6517, loss_cns_2: 0.6532, loss_yns_2: 0.1580, loss_cls_3: 0.9504, loss_box_3: 1.6430, loss_cns_3: 0.6575, loss_yns_3: 0.1620, loss_cls_4: 0.9473, loss_box_4: 1.6367, loss_cns_4: 0.6586, loss_yns_4: 0.1591, loss_cls_5: 0.9510, loss_box_5: 1.6337, loss_cns_5: 0.6584, loss_yns_5: 0.1603, loss_cls_dn_0: 0.2125, loss_box_dn_0: 0.7719, loss_cls_dn_1: 0.1404, loss_box_dn_1: 0.7676, loss_cls_dn_2: 0.1422, loss_box_dn_2: 0.7682, loss_cls_dn_3: 0.1442, loss_box_dn_3: 0.7743, loss_cls_dn_4: 0.1461, loss_box_dn_4: 0.7801, loss_cls_dn_5: 0.1496, loss_box_dn_5: 0.7852, loss_dense_depth: 0.8030, loss: 26.8130, grad_norm: 38.8631 -2026-01-14 21:36:06,663 - mmdet - INFO - Iter [168/17500] lr: 1.668e-04, eta: 11:35:43, time: 1.526, data_time: 0.070, memory: 49164, loss_cls_0: 0.8282, loss_box_0: 1.7406, loss_cns_0: 0.6216, loss_yns_0: 0.1621, loss_cls_1: 0.8984, loss_box_1: 1.6768, loss_cns_1: 0.6532, loss_yns_1: 0.1604, loss_cls_2: 0.9254, loss_box_2: 1.6340, loss_cns_2: 0.6592, loss_yns_2: 0.1618, loss_cls_3: 0.9371, loss_box_3: 1.6351, loss_cns_3: 0.6597, loss_yns_3: 0.1663, loss_cls_4: 0.9467, loss_box_4: 1.6292, loss_cns_4: 0.6613, loss_yns_4: 0.1673, loss_cls_5: 0.9488, loss_box_5: 1.6255, loss_cns_5: 0.6596, loss_yns_5: 0.1609, loss_cls_dn_0: 0.2074, loss_box_dn_0: 0.7616, loss_cls_dn_1: 0.1398, loss_box_dn_1: 0.7593, loss_cls_dn_2: 0.1434, loss_box_dn_2: 0.7600, loss_cls_dn_3: 0.1440, loss_box_dn_3: 0.7719, loss_cls_dn_4: 0.1499, loss_box_dn_4: 0.7783, loss_cls_dn_5: 0.1564, loss_box_dn_5: 0.7915, loss_dense_depth: 0.8219, loss: 26.7046, grad_norm: 47.2439 -2026-01-14 21:36:08,206 - mmdet - INFO - Iter [169/17500] lr: 1.672e-04, eta: 11:34:12, time: 1.542, data_time: 0.070, memory: 49164, loss_cls_0: 0.8689, loss_box_0: 1.7536, loss_cns_0: 0.6144, loss_yns_0: 0.1591, loss_cls_1: 0.9211, loss_box_1: 1.7047, loss_cns_1: 0.6486, loss_yns_1: 0.1630, loss_cls_2: 0.9426, loss_box_2: 1.6626, loss_cns_2: 0.6558, loss_yns_2: 0.1630, loss_cls_3: 0.9556, loss_box_3: 1.6776, loss_cns_3: 0.6574, loss_yns_3: 0.1645, loss_cls_4: 0.9537, loss_box_4: 1.6658, loss_cns_4: 0.6577, loss_yns_4: 0.1652, loss_cls_5: 0.9592, loss_box_5: 1.6558, loss_cns_5: 0.6577, loss_yns_5: 0.1604, loss_cls_dn_0: 0.2175, loss_box_dn_0: 0.7670, loss_cls_dn_1: 0.1385, loss_box_dn_1: 0.7566, loss_cls_dn_2: 0.1411, loss_box_dn_2: 0.7546, loss_cls_dn_3: 0.1414, loss_box_dn_3: 0.7678, loss_cls_dn_4: 0.1466, loss_box_dn_4: 0.7718, loss_cls_dn_5: 0.1548, loss_box_dn_5: 0.7837, loss_dense_depth: 0.8190, loss: 26.9484, grad_norm: 42.1674 -2026-01-14 21:36:09,754 - mmdet - INFO - Iter [170/17500] lr: 1.676e-04, eta: 11:32:42, time: 1.548, data_time: 0.074, memory: 49164, loss_cls_0: 0.8473, loss_box_0: 1.7417, loss_cns_0: 0.6204, loss_yns_0: 0.1604, loss_cls_1: 0.9230, loss_box_1: 1.6599, loss_cns_1: 0.6548, loss_yns_1: 0.1605, loss_cls_2: 0.9403, loss_box_2: 1.6183, loss_cns_2: 0.6599, loss_yns_2: 0.1590, loss_cls_3: 0.9492, loss_box_3: 1.5893, loss_cns_3: 0.6600, loss_yns_3: 0.1594, loss_cls_4: 0.9576, loss_box_4: 1.5874, loss_cns_4: 0.6604, loss_yns_4: 0.1580, loss_cls_5: 0.9544, loss_box_5: 1.5849, loss_cns_5: 0.6628, loss_yns_5: 0.1585, loss_cls_dn_0: 0.2140, loss_box_dn_0: 0.7631, loss_cls_dn_1: 0.1382, loss_box_dn_1: 0.7206, loss_cls_dn_2: 0.1413, loss_box_dn_2: 0.7195, loss_cls_dn_3: 0.1438, loss_box_dn_3: 0.7127, loss_cls_dn_4: 0.1482, loss_box_dn_4: 0.7179, loss_cls_dn_5: 0.1536, loss_box_dn_5: 0.7255, loss_dense_depth: 0.7759, loss: 26.3016, grad_norm: 32.5987 -2026-01-14 21:36:11,350 - mmdet - INFO - Iter [171/17500] lr: 1.680e-04, eta: 11:31:15, time: 1.558, data_time: 0.076, memory: 49164, loss_cls_0: 0.8373, loss_box_0: 1.7168, loss_cns_0: 0.6208, loss_yns_0: 0.1594, loss_cls_1: 0.9073, loss_box_1: 1.6753, loss_cns_1: 0.6508, loss_yns_1: 0.1583, loss_cls_2: 0.9511, loss_box_2: 1.6431, loss_cns_2: 0.6573, loss_yns_2: 0.1578, loss_cls_3: 0.9387, loss_box_3: 1.6150, loss_cns_3: 0.6590, loss_yns_3: 0.1613, loss_cls_4: 0.9440, loss_box_4: 1.6162, loss_cns_4: 0.6580, loss_yns_4: 0.1625, loss_cls_5: 0.9468, loss_box_5: 1.6088, loss_cns_5: 0.6602, loss_yns_5: 0.1562, loss_cls_dn_0: 0.2092, loss_box_dn_0: 0.7595, loss_cls_dn_1: 0.1392, loss_box_dn_1: 0.7081, loss_cls_dn_2: 0.1427, loss_box_dn_2: 0.7088, loss_cls_dn_3: 0.1431, loss_box_dn_3: 0.7069, loss_cls_dn_4: 0.1475, loss_box_dn_4: 0.7158, loss_cls_dn_5: 0.1520, loss_box_dn_5: 0.7144, loss_dense_depth: 0.7879, loss: 26.2973, grad_norm: 43.7901 -2026-01-14 21:36:12,946 - mmdet - INFO - Iter [172/17500] lr: 1.684e-04, eta: 11:29:56, time: 1.633, data_time: 0.102, memory: 49164, loss_cls_0: 0.8352, loss_box_0: 1.7455, loss_cns_0: 0.6174, loss_yns_0: 0.1592, loss_cls_1: 0.9231, loss_box_1: 1.6745, loss_cns_1: 0.6475, loss_yns_1: 0.1561, loss_cls_2: 0.9455, loss_box_2: 1.6549, loss_cns_2: 0.6561, loss_yns_2: 0.1603, loss_cls_3: 0.9562, loss_box_3: 1.6505, loss_cns_3: 0.6579, loss_yns_3: 0.1623, loss_cls_4: 0.9622, loss_box_4: 1.6489, loss_cns_4: 0.6571, loss_yns_4: 0.1639, loss_cls_5: 0.9568, loss_box_5: 1.6523, loss_cns_5: 0.6573, loss_yns_5: 0.1599, loss_cls_dn_0: 0.2120, loss_box_dn_0: 0.7592, loss_cls_dn_1: 0.1354, loss_box_dn_1: 0.7266, loss_cls_dn_2: 0.1369, loss_box_dn_2: 0.7246, loss_cls_dn_3: 0.1504, loss_box_dn_3: 0.7358, loss_cls_dn_4: 0.1470, loss_box_dn_4: 0.7525, loss_cls_dn_5: 0.1552, loss_box_dn_5: 0.7617, loss_dense_depth: 0.7481, loss: 26.6059, grad_norm: 44.5574 -2026-01-14 21:36:14,527 - mmdet - INFO - Iter [173/17500] lr: 1.688e-04, eta: 11:28:33, time: 1.581, data_time: 0.077, memory: 49164, loss_cls_0: 0.8221, loss_box_0: 1.7164, loss_cns_0: 0.6232, loss_yns_0: 0.1567, loss_cls_1: 0.9088, loss_box_1: 1.6791, loss_cns_1: 0.6520, loss_yns_1: 0.1557, loss_cls_2: 0.9350, loss_box_2: 1.6394, loss_cns_2: 0.6579, loss_yns_2: 0.1560, loss_cls_3: 0.9478, loss_box_3: 1.6498, loss_cns_3: 0.6572, loss_yns_3: 0.1570, loss_cls_4: 0.9593, loss_box_4: 1.6534, loss_cns_4: 0.6580, loss_yns_4: 0.1573, loss_cls_5: 0.9576, loss_box_5: 1.6651, loss_cns_5: 0.6566, loss_yns_5: 0.1561, loss_cls_dn_0: 0.2116, loss_box_dn_0: 0.7584, loss_cls_dn_1: 0.1367, loss_box_dn_1: 0.7409, loss_cls_dn_2: 0.1378, loss_box_dn_2: 0.7346, loss_cls_dn_3: 0.1434, loss_box_dn_3: 0.7496, loss_cls_dn_4: 0.1431, loss_box_dn_4: 0.7606, loss_cls_dn_5: 0.1468, loss_box_dn_5: 0.7801, loss_dense_depth: 0.7921, loss: 26.6133, grad_norm: 53.3395 -2026-01-14 21:36:16,092 - mmdet - INFO - Iter [174/17500] lr: 1.692e-04, eta: 11:27:09, time: 1.565, data_time: 0.076, memory: 49164, loss_cls_0: 0.8321, loss_box_0: 1.7419, loss_cns_0: 0.6136, loss_yns_0: 0.1571, loss_cls_1: 0.9190, loss_box_1: 1.6673, loss_cns_1: 0.6484, loss_yns_1: 0.1575, loss_cls_2: 0.9484, loss_box_2: 1.6681, loss_cns_2: 0.6543, loss_yns_2: 0.1556, loss_cls_3: 0.9327, loss_box_3: 1.6577, loss_cns_3: 0.6543, loss_yns_3: 0.1549, loss_cls_4: 0.9383, loss_box_4: 1.6484, loss_cns_4: 0.6573, loss_yns_4: 0.1553, loss_cls_5: 0.9627, loss_box_5: 1.6532, loss_cns_5: 0.6598, loss_yns_5: 0.1568, loss_cls_dn_0: 0.2111, loss_box_dn_0: 0.7673, loss_cls_dn_1: 0.1370, loss_box_dn_1: 0.7506, loss_cls_dn_2: 0.1402, loss_box_dn_2: 0.7524, loss_cls_dn_3: 0.1462, loss_box_dn_3: 0.7599, loss_cls_dn_4: 0.1453, loss_box_dn_4: 0.7710, loss_cls_dn_5: 0.1619, loss_box_dn_5: 0.7890, loss_dense_depth: 0.7774, loss: 26.7039, grad_norm: 44.2203 -2026-01-14 21:36:17,659 - mmdet - INFO - Iter [175/17500] lr: 1.696e-04, eta: 11:25:46, time: 1.568, data_time: 0.072, memory: 49164, loss_cls_0: 0.8201, loss_box_0: 1.7211, loss_cns_0: 0.6185, loss_yns_0: 0.1592, loss_cls_1: 0.9005, loss_box_1: 1.6661, loss_cns_1: 0.6522, loss_yns_1: 0.1600, loss_cls_2: 0.9274, loss_box_2: 1.6719, loss_cns_2: 0.6536, loss_yns_2: 0.1589, loss_cls_3: 0.9385, loss_box_3: 1.6548, loss_cns_3: 0.6573, loss_yns_3: 0.1635, loss_cls_4: 0.9416, loss_box_4: 1.6566, loss_cns_4: 0.6569, loss_yns_4: 0.1611, loss_cls_5: 0.9424, loss_box_5: 1.6624, loss_cns_5: 0.6580, loss_yns_5: 0.1570, loss_cls_dn_0: 0.2128, loss_box_dn_0: 0.7534, loss_cls_dn_1: 0.1336, loss_box_dn_1: 0.7460, loss_cls_dn_2: 0.1347, loss_box_dn_2: 0.7565, loss_cls_dn_3: 0.1352, loss_box_dn_3: 0.7614, loss_cls_dn_4: 0.1374, loss_box_dn_4: 0.7804, loss_cls_dn_5: 0.1449, loss_box_dn_5: 0.7971, loss_dense_depth: 0.7862, loss: 26.6395, grad_norm: 50.5833 -2026-01-14 21:36:19,272 - mmdet - INFO - Iter [176/17500] lr: 1.700e-04, eta: 11:24:28, time: 1.612, data_time: 0.075, memory: 49164, loss_cls_0: 0.8148, loss_box_0: 1.7040, loss_cns_0: 0.6206, loss_yns_0: 0.1545, loss_cls_1: 0.8969, loss_box_1: 1.6573, loss_cns_1: 0.6504, loss_yns_1: 0.1536, loss_cls_2: 0.9252, loss_box_2: 1.6363, loss_cns_2: 0.6566, loss_yns_2: 0.1555, loss_cls_3: 0.9288, loss_box_3: 1.6251, loss_cns_3: 0.6618, loss_yns_3: 0.1630, loss_cls_4: 0.9438, loss_box_4: 1.6243, loss_cns_4: 0.6606, loss_yns_4: 0.1600, loss_cls_5: 0.9426, loss_box_5: 1.6336, loss_cns_5: 0.6582, loss_yns_5: 0.1531, loss_cls_dn_0: 0.2132, loss_box_dn_0: 0.7654, loss_cls_dn_1: 0.1373, loss_box_dn_1: 0.7584, loss_cls_dn_2: 0.1367, loss_box_dn_2: 0.7538, loss_cls_dn_3: 0.1384, loss_box_dn_3: 0.7575, loss_cls_dn_4: 0.1399, loss_box_dn_4: 0.7688, loss_cls_dn_5: 0.1438, loss_box_dn_5: 0.7818, loss_dense_depth: 0.8002, loss: 26.4757, grad_norm: 36.5375 -2026-01-14 21:36:20,842 - mmdet - INFO - Iter [177/17500] lr: 1.704e-04, eta: 11:23:08, time: 1.569, data_time: 0.071, memory: 49164, loss_cls_0: 0.8050, loss_box_0: 1.6990, loss_cns_0: 0.6162, loss_yns_0: 0.1499, loss_cls_1: 0.8842, loss_box_1: 1.6485, loss_cns_1: 0.6490, loss_yns_1: 0.1492, loss_cls_2: 0.9229, loss_box_2: 1.6165, loss_cns_2: 0.6559, loss_yns_2: 0.1477, loss_cls_3: 0.9444, loss_box_3: 1.6103, loss_cns_3: 0.6558, loss_yns_3: 0.1515, loss_cls_4: 0.9449, loss_box_4: 1.6131, loss_cns_4: 0.6566, loss_yns_4: 0.1496, loss_cls_5: 0.9303, loss_box_5: 1.6013, loss_cns_5: 0.6581, loss_yns_5: 0.1508, loss_cls_dn_0: 0.2123, loss_box_dn_0: 0.7737, loss_cls_dn_1: 0.1361, loss_box_dn_1: 0.7662, loss_cls_dn_2: 0.1359, loss_box_dn_2: 0.7490, loss_cls_dn_3: 0.1397, loss_box_dn_3: 0.7473, loss_cls_dn_4: 0.1445, loss_box_dn_4: 0.7533, loss_cls_dn_5: 0.1462, loss_box_dn_5: 0.7497, loss_dense_depth: 0.7865, loss: 26.2512, grad_norm: 44.0711 -2026-01-14 21:36:22,407 - mmdet - INFO - Iter [178/17500] lr: 1.708e-04, eta: 11:21:47, time: 1.566, data_time: 0.075, memory: 49164, loss_cls_0: 0.7981, loss_box_0: 1.7012, loss_cns_0: 0.6172, loss_yns_0: 0.1506, loss_cls_1: 0.8816, loss_box_1: 1.6812, loss_cns_1: 0.6465, loss_yns_1: 0.1493, loss_cls_2: 0.9186, loss_box_2: 1.6332, loss_cns_2: 0.6541, loss_yns_2: 0.1487, loss_cls_3: 0.9200, loss_box_3: 1.6038, loss_cns_3: 0.6544, loss_yns_3: 0.1493, loss_cls_4: 0.9257, loss_box_4: 1.5936, loss_cns_4: 0.6533, loss_yns_4: 0.1500, loss_cls_5: 0.9286, loss_box_5: 1.6211, loss_cns_5: 0.6579, loss_yns_5: 0.1500, loss_cls_dn_0: 0.2094, loss_box_dn_0: 0.7679, loss_cls_dn_1: 0.1374, loss_box_dn_1: 0.7246, loss_cls_dn_2: 0.1420, loss_box_dn_2: 0.7127, loss_cls_dn_3: 0.1485, loss_box_dn_3: 0.7056, loss_cls_dn_4: 0.1444, loss_box_dn_4: 0.7095, loss_cls_dn_5: 0.1575, loss_box_dn_5: 0.7164, loss_dense_depth: 0.7814, loss: 26.0453, grad_norm: 33.8181 -2026-01-14 21:36:24,005 - mmdet - INFO - Iter [179/17500] lr: 1.712e-04, eta: 11:20:31, time: 1.598, data_time: 0.081, memory: 49164, loss_cls_0: 0.7983, loss_box_0: 1.7157, loss_cns_0: 0.6140, loss_yns_0: 0.1507, loss_cls_1: 0.8939, loss_box_1: 1.6553, loss_cns_1: 0.6440, loss_yns_1: 0.1526, loss_cls_2: 0.9121, loss_box_2: 1.6097, loss_cns_2: 0.6535, loss_yns_2: 0.1526, loss_cls_3: 0.9437, loss_box_3: 1.5929, loss_cns_3: 0.6540, loss_yns_3: 0.1526, loss_cls_4: 0.9668, loss_box_4: 1.5819, loss_cns_4: 0.6499, loss_yns_4: 0.1534, loss_cls_5: 0.9471, loss_box_5: 1.6114, loss_cns_5: 0.6540, loss_yns_5: 0.1521, loss_cls_dn_0: 0.2110, loss_box_dn_0: 0.7761, loss_cls_dn_1: 0.1374, loss_box_dn_1: 0.7222, loss_cls_dn_2: 0.1382, loss_box_dn_2: 0.7137, loss_cls_dn_3: 0.1385, loss_box_dn_3: 0.7173, loss_cls_dn_4: 0.1447, loss_box_dn_4: 0.7239, loss_cls_dn_5: 0.1460, loss_box_dn_5: 0.7434, loss_dense_depth: 0.8071, loss: 26.1319, grad_norm: 40.2895 -2026-01-14 21:36:25,579 - mmdet - INFO - Iter [180/17500] lr: 1.716e-04, eta: 11:19:13, time: 1.574, data_time: 0.079, memory: 49164, loss_cls_0: 0.7934, loss_box_0: 1.6824, loss_cns_0: 0.6253, loss_yns_0: 0.1495, loss_cls_1: 0.8791, loss_box_1: 1.6318, loss_cns_1: 0.6499, loss_yns_1: 0.1481, loss_cls_2: 0.9075, loss_box_2: 1.6125, loss_cns_2: 0.6582, loss_yns_2: 0.1512, loss_cls_3: 0.9296, loss_box_3: 1.5859, loss_cns_3: 0.6663, loss_yns_3: 0.1557, loss_cls_4: 0.9284, loss_box_4: 1.5828, loss_cns_4: 0.6616, loss_yns_4: 0.1543, loss_cls_5: 0.9252, loss_box_5: 1.5860, loss_cns_5: 0.6596, loss_yns_5: 0.1515, loss_cls_dn_0: 0.2099, loss_box_dn_0: 0.7655, loss_cls_dn_1: 0.1352, loss_box_dn_1: 0.7286, loss_cls_dn_2: 0.1353, loss_box_dn_2: 0.7255, loss_cls_dn_3: 0.1360, loss_box_dn_3: 0.7338, loss_cls_dn_4: 0.1392, loss_box_dn_4: 0.7449, loss_cls_dn_5: 0.1447, loss_box_dn_5: 0.7670, loss_dense_depth: 0.7756, loss: 26.0175, grad_norm: 36.0766 -2026-01-14 21:36:27,264 - mmdet - INFO - Iter [181/17500] lr: 1.720e-04, eta: 11:18:07, time: 1.685, data_time: 0.110, memory: 49164, loss_cls_0: 0.7920, loss_box_0: 1.7139, loss_cns_0: 0.6188, loss_yns_0: 0.1502, loss_cls_1: 0.8834, loss_box_1: 1.6423, loss_cns_1: 0.6523, loss_yns_1: 0.1500, loss_cls_2: 0.9113, loss_box_2: 1.6127, loss_cns_2: 0.6547, loss_yns_2: 0.1500, loss_cls_3: 0.9680, loss_box_3: 1.5942, loss_cns_3: 0.6564, loss_yns_3: 0.1519, loss_cls_4: 1.0152, loss_box_4: 1.5900, loss_cns_4: 0.6559, loss_yns_4: 0.1507, loss_cls_5: 0.9314, loss_box_5: 1.6062, loss_cns_5: 0.6545, loss_yns_5: 0.1503, loss_cls_dn_0: 0.2094, loss_box_dn_0: 0.7668, loss_cls_dn_1: 0.1340, loss_box_dn_1: 0.7449, loss_cls_dn_2: 0.1341, loss_box_dn_2: 0.7455, loss_cls_dn_3: 0.1347, loss_box_dn_3: 0.7568, loss_cls_dn_4: 0.1379, loss_box_dn_4: 0.7673, loss_cls_dn_5: 0.1405, loss_box_dn_5: 0.7945, loss_dense_depth: 0.8314, loss: 26.3541, grad_norm: 50.2957 -2026-01-14 21:36:28,907 - mmdet - INFO - Iter [182/17500] lr: 1.724e-04, eta: 11:16:57, time: 1.641, data_time: 0.165, memory: 49164, loss_cls_0: 0.7818, loss_box_0: 1.6742, loss_cns_0: 0.6230, loss_yns_0: 0.1499, loss_cls_1: 0.8824, loss_box_1: 1.6485, loss_cns_1: 0.6546, loss_yns_1: 0.1532, loss_cls_2: 0.9017, loss_box_2: 1.6221, loss_cns_2: 0.6554, loss_yns_2: 0.1479, loss_cls_3: 0.9058, loss_box_3: 1.6147, loss_cns_3: 0.6575, loss_yns_3: 0.1491, loss_cls_4: 0.9121, loss_box_4: 1.6134, loss_cns_4: 0.6556, loss_yns_4: 0.1489, loss_cls_5: 0.9104, loss_box_5: 1.6257, loss_cns_5: 0.6556, loss_yns_5: 0.1475, loss_cls_dn_0: 0.2068, loss_box_dn_0: 0.7666, loss_cls_dn_1: 0.1342, loss_box_dn_1: 0.7774, loss_cls_dn_2: 0.1380, loss_box_dn_2: 0.7780, loss_cls_dn_3: 0.1471, loss_box_dn_3: 0.7845, loss_cls_dn_4: 0.1463, loss_box_dn_4: 0.7951, loss_cls_dn_5: 0.1557, loss_box_dn_5: 0.8162, loss_dense_depth: 0.7612, loss: 26.2979, grad_norm: 36.0028 -2026-01-14 21:36:30,573 - mmdet - INFO - Iter [183/17500] lr: 1.728e-04, eta: 11:15:47, time: 1.622, data_time: 0.075, memory: 49164, loss_cls_0: 0.7760, loss_box_0: 1.6576, loss_cns_0: 0.6230, loss_yns_0: 0.1464, loss_cls_1: 0.8781, loss_box_1: 1.5801, loss_cns_1: 0.6568, loss_yns_1: 0.1469, loss_cls_2: 0.9102, loss_box_2: 1.5378, loss_cns_2: 0.6576, loss_yns_2: 0.1453, loss_cls_3: 0.9135, loss_box_3: 1.5148, loss_cns_3: 0.6642, loss_yns_3: 0.1463, loss_cls_4: 0.9180, loss_box_4: 1.5177, loss_cns_4: 0.6631, loss_yns_4: 0.1464, loss_cls_5: 0.9038, loss_box_5: 1.5156, loss_cns_5: 0.6622, loss_yns_5: 0.1462, loss_cls_dn_0: 0.2051, loss_box_dn_0: 0.7546, loss_cls_dn_1: 0.1309, loss_box_dn_1: 0.7390, loss_cls_dn_2: 0.1387, loss_box_dn_2: 0.7350, loss_cls_dn_3: 0.1440, loss_box_dn_3: 0.7297, loss_cls_dn_4: 0.1435, loss_box_dn_4: 0.7352, loss_cls_dn_5: 0.1477, loss_box_dn_5: 0.7394, loss_dense_depth: 0.8011, loss: 25.5715, grad_norm: 37.6812 -2026-01-14 21:36:32,146 - mmdet - INFO - Iter [184/17500] lr: 1.732e-04, eta: 11:14:36, time: 1.619, data_time: 0.119, memory: 49164, loss_cls_0: 0.8045, loss_box_0: 1.6775, loss_cns_0: 0.6259, loss_yns_0: 0.1439, loss_cls_1: 0.8839, loss_box_1: 1.6155, loss_cns_1: 0.6581, loss_yns_1: 0.1434, loss_cls_2: 0.9006, loss_box_2: 1.5893, loss_cns_2: 0.6608, loss_yns_2: 0.1448, loss_cls_3: 0.9116, loss_box_3: 1.5664, loss_cns_3: 0.6654, loss_yns_3: 0.1452, loss_cls_4: 0.9207, loss_box_4: 1.5559, loss_cns_4: 0.6635, loss_yns_4: 0.1453, loss_cls_5: 0.9280, loss_box_5: 1.5545, loss_cns_5: 0.6631, loss_yns_5: 0.1465, loss_cls_dn_0: 0.2100, loss_box_dn_0: 0.7635, loss_cls_dn_1: 0.1310, loss_box_dn_1: 0.7107, loss_cls_dn_2: 0.1352, loss_box_dn_2: 0.7073, loss_cls_dn_3: 0.1355, loss_box_dn_3: 0.7075, loss_cls_dn_4: 0.1407, loss_box_dn_4: 0.7087, loss_cls_dn_5: 0.1412, loss_box_dn_5: 0.7125, loss_dense_depth: 0.8063, loss: 25.7244, grad_norm: 35.0791 -2026-01-14 21:36:33,767 - mmdet - INFO - Iter [185/17500] lr: 1.736e-04, eta: 11:13:27, time: 1.620, data_time: 0.069, memory: 49164, loss_cls_0: 0.7935, loss_box_0: 1.6701, loss_cns_0: 0.6214, loss_yns_0: 0.1475, loss_cls_1: 0.8913, loss_box_1: 1.5961, loss_cns_1: 0.6581, loss_yns_1: 0.1451, loss_cls_2: 0.9041, loss_box_2: 1.5594, loss_cns_2: 0.6609, loss_yns_2: 0.1446, loss_cls_3: 0.9052, loss_box_3: 1.5505, loss_cns_3: 0.6604, loss_yns_3: 0.1452, loss_cls_4: 0.9140, loss_box_4: 1.5507, loss_cns_4: 0.6612, loss_yns_4: 0.1445, loss_cls_5: 0.9140, loss_box_5: 1.5401, loss_cns_5: 0.6608, loss_yns_5: 0.1453, loss_cls_dn_0: 0.2115, loss_box_dn_0: 0.7633, loss_cls_dn_1: 0.1337, loss_box_dn_1: 0.7123, loss_cls_dn_2: 0.1351, loss_box_dn_2: 0.7040, loss_cls_dn_3: 0.1363, loss_box_dn_3: 0.7107, loss_cls_dn_4: 0.1427, loss_box_dn_4: 0.7177, loss_cls_dn_5: 0.1450, loss_box_dn_5: 0.7203, loss_dense_depth: 0.8233, loss: 25.6402, grad_norm: 30.5972 -2026-01-14 21:36:35,353 - mmdet - INFO - Iter [186/17500] lr: 1.740e-04, eta: 11:12:15, time: 1.587, data_time: 0.074, memory: 49164, loss_cls_0: 0.8070, loss_box_0: 1.6468, loss_cns_0: 0.6157, loss_yns_0: 0.1464, loss_cls_1: 0.8969, loss_box_1: 1.5889, loss_cns_1: 0.6602, loss_yns_1: 0.1477, loss_cls_2: 0.9097, loss_box_2: 1.5468, loss_cns_2: 0.6630, loss_yns_2: 0.1462, loss_cls_3: 0.9210, loss_box_3: 1.5408, loss_cns_3: 0.6631, loss_yns_3: 0.1465, loss_cls_4: 0.9294, loss_box_4: 1.5732, loss_cns_4: 0.6623, loss_yns_4: 0.1469, loss_cls_5: 0.9251, loss_box_5: 1.5668, loss_cns_5: 0.6610, loss_yns_5: 0.1466, loss_cls_dn_0: 0.2064, loss_box_dn_0: 0.7674, loss_cls_dn_1: 0.1332, loss_box_dn_1: 0.7194, loss_cls_dn_2: 0.1342, loss_box_dn_2: 0.7142, loss_cls_dn_3: 0.1390, loss_box_dn_3: 0.7284, loss_cls_dn_4: 0.1468, loss_box_dn_4: 0.7521, loss_cls_dn_5: 0.1485, loss_box_dn_5: 0.7617, loss_dense_depth: 0.8222, loss: 25.8317, grad_norm: 41.3447 -2026-01-14 21:36:36,931 - mmdet - INFO - Iter [187/17500] lr: 1.744e-04, eta: 11:11:03, time: 1.578, data_time: 0.075, memory: 49164, loss_cls_0: 0.8103, loss_box_0: 1.6626, loss_cns_0: 0.6203, loss_yns_0: 0.1455, loss_cls_1: 0.9117, loss_box_1: 1.6053, loss_cns_1: 0.6565, loss_yns_1: 0.1483, loss_cls_2: 0.9193, loss_box_2: 1.5907, loss_cns_2: 0.6597, loss_yns_2: 0.1489, loss_cls_3: 0.9354, loss_box_3: 1.5742, loss_cns_3: 0.6598, loss_yns_3: 0.1501, loss_cls_4: 0.9407, loss_box_4: 1.5789, loss_cns_4: 0.6593, loss_yns_4: 0.1499, loss_cls_5: 0.9365, loss_box_5: 1.5861, loss_cns_5: 0.6583, loss_yns_5: 0.1487, loss_cls_dn_0: 0.2088, loss_box_dn_0: 0.7627, loss_cls_dn_1: 0.1306, loss_box_dn_1: 0.7427, loss_cls_dn_2: 0.1302, loss_box_dn_2: 0.7428, loss_cls_dn_3: 0.1343, loss_box_dn_3: 0.7525, loss_cls_dn_4: 0.1402, loss_box_dn_4: 0.7726, loss_cls_dn_5: 0.1455, loss_box_dn_5: 0.7880, loss_dense_depth: 0.8246, loss: 26.1321, grad_norm: 40.1872 -2026-01-14 21:36:38,544 - mmdet - INFO - Iter [188/17500] lr: 1.748e-04, eta: 11:09:55, time: 1.612, data_time: 0.073, memory: 49164, loss_cls_0: 0.8100, loss_box_0: 1.7130, loss_cns_0: 0.6208, loss_yns_0: 0.1456, loss_cls_1: 0.8971, loss_box_1: 1.6262, loss_cns_1: 0.6557, loss_yns_1: 0.1446, loss_cls_2: 0.9169, loss_box_2: 1.6117, loss_cns_2: 0.6593, loss_yns_2: 0.1450, loss_cls_3: 0.9166, loss_box_3: 1.5812, loss_cns_3: 0.6575, loss_yns_3: 0.1462, loss_cls_4: 0.9258, loss_box_4: 1.5837, loss_cns_4: 0.6572, loss_yns_4: 0.1451, loss_cls_5: 0.9195, loss_box_5: 1.5895, loss_cns_5: 0.6573, loss_yns_5: 0.1445, loss_cls_dn_0: 0.2049, loss_box_dn_0: 0.7611, loss_cls_dn_1: 0.1331, loss_box_dn_1: 0.7638, loss_cls_dn_2: 0.1341, loss_box_dn_2: 0.7649, loss_cls_dn_3: 0.1357, loss_box_dn_3: 0.7613, loss_cls_dn_4: 0.1415, loss_box_dn_4: 0.7743, loss_cls_dn_5: 0.1452, loss_box_dn_5: 0.7832, loss_dense_depth: 0.8142, loss: 26.1876, grad_norm: 41.5670 -2026-01-14 21:36:40,237 - mmdet - INFO - Iter [189/17500] lr: 1.752e-04, eta: 11:08:51, time: 1.646, data_time: 0.073, memory: 49164, loss_cls_0: 0.8181, loss_box_0: 1.6678, loss_cns_0: 0.6311, loss_yns_0: 0.1524, loss_cls_1: 0.8923, loss_box_1: 1.6190, loss_cns_1: 0.6597, loss_yns_1: 0.1493, loss_cls_2: 0.9208, loss_box_2: 1.6128, loss_cns_2: 0.6581, loss_yns_2: 0.1485, loss_cls_3: 0.9406, loss_box_3: 1.5898, loss_cns_3: 0.6589, loss_yns_3: 0.1508, loss_cls_4: 0.9362, loss_box_4: 1.5654, loss_cns_4: 0.6608, loss_yns_4: 0.1517, loss_cls_5: 0.9354, loss_box_5: 1.5668, loss_cns_5: 0.6601, loss_yns_5: 0.1502, loss_cls_dn_0: 0.2073, loss_box_dn_0: 0.7571, loss_cls_dn_1: 0.1363, loss_box_dn_1: 0.7505, loss_cls_dn_2: 0.1372, loss_box_dn_2: 0.7451, loss_cls_dn_3: 0.1390, loss_box_dn_3: 0.7385, loss_cls_dn_4: 0.1498, loss_box_dn_4: 0.7336, loss_cls_dn_5: 0.1551, loss_box_dn_5: 0.7328, loss_dense_depth: 0.8205, loss: 26.0993, grad_norm: 51.6228 -2026-01-14 21:36:41,788 - mmdet - INFO - Iter [190/17500] lr: 1.755e-04, eta: 11:07:43, time: 1.598, data_time: 0.118, memory: 49164, loss_cls_0: 0.8182, loss_box_0: 1.7046, loss_cns_0: 0.6238, loss_yns_0: 0.1533, loss_cls_1: 0.8964, loss_box_1: 1.6628, loss_cns_1: 0.6529, loss_yns_1: 0.1500, loss_cls_2: 0.9333, loss_box_2: 1.6556, loss_cns_2: 0.6556, loss_yns_2: 0.1511, loss_cls_3: 0.9518, loss_box_3: 1.6325, loss_cns_3: 0.6560, loss_yns_3: 0.1542, loss_cls_4: 0.9440, loss_box_4: 1.6140, loss_cns_4: 0.6573, loss_yns_4: 0.1550, loss_cls_5: 0.9363, loss_box_5: 1.6228, loss_cns_5: 0.6574, loss_yns_5: 0.1531, loss_cls_dn_0: 0.2140, loss_box_dn_0: 0.7635, loss_cls_dn_1: 0.1373, loss_box_dn_1: 0.7180, loss_cls_dn_2: 0.1418, loss_box_dn_2: 0.7115, loss_cls_dn_3: 0.1421, loss_box_dn_3: 0.7110, loss_cls_dn_4: 0.1480, loss_box_dn_4: 0.7092, loss_cls_dn_5: 0.1532, loss_box_dn_5: 0.7176, loss_dense_depth: 0.8262, loss: 26.2857, grad_norm: 46.3242 -2026-01-14 21:36:43,343 - mmdet - INFO - Iter [191/17500] lr: 1.759e-04, eta: 11:06:32, time: 1.555, data_time: 0.073, memory: 49164, loss_cls_0: 0.8166, loss_box_0: 1.7005, loss_cns_0: 0.6167, loss_yns_0: 0.1504, loss_cls_1: 0.8969, loss_box_1: 1.6473, loss_cns_1: 0.6525, loss_yns_1: 0.1523, loss_cls_2: 0.9370, loss_box_2: 1.6257, loss_cns_2: 0.6549, loss_yns_2: 0.1517, loss_cls_3: 0.9315, loss_box_3: 1.6210, loss_cns_3: 0.6569, loss_yns_3: 0.1544, loss_cls_4: 0.9397, loss_box_4: 1.6434, loss_cns_4: 0.6528, loss_yns_4: 0.1552, loss_cls_5: 0.9361, loss_box_5: 1.6501, loss_cns_5: 0.6522, loss_yns_5: 0.1525, loss_cls_dn_0: 0.2148, loss_box_dn_0: 0.7523, loss_cls_dn_1: 0.1342, loss_box_dn_1: 0.7129, loss_cls_dn_2: 0.1379, loss_box_dn_2: 0.7155, loss_cls_dn_3: 0.1379, loss_box_dn_3: 0.7344, loss_cls_dn_4: 0.1394, loss_box_dn_4: 0.7540, loss_cls_dn_5: 0.1487, loss_box_dn_5: 0.7757, loss_dense_depth: 0.7883, loss: 26.2943, grad_norm: 52.9777 -2026-01-14 21:36:44,910 - mmdet - INFO - Iter [192/17500] lr: 1.763e-04, eta: 11:05:22, time: 1.568, data_time: 0.079, memory: 49164, loss_cls_0: 0.8161, loss_box_0: 1.6934, loss_cns_0: 0.6260, loss_yns_0: 0.1511, loss_cls_1: 0.8971, loss_box_1: 1.6869, loss_cns_1: 0.6530, loss_yns_1: 0.1504, loss_cls_2: 0.9265, loss_box_2: 1.6319, loss_cns_2: 0.6558, loss_yns_2: 0.1502, loss_cls_3: 0.9370, loss_box_3: 1.6430, loss_cns_3: 0.6569, loss_yns_3: 0.1515, loss_cls_4: 0.9324, loss_box_4: 1.6447, loss_cns_4: 0.6554, loss_yns_4: 0.1527, loss_cls_5: 0.9381, loss_box_5: 1.6327, loss_cns_5: 0.6549, loss_yns_5: 0.1536, loss_cls_dn_0: 0.2133, loss_box_dn_0: 0.7512, loss_cls_dn_1: 0.1345, loss_box_dn_1: 0.7530, loss_cls_dn_2: 0.1373, loss_box_dn_2: 0.7495, loss_cls_dn_3: 0.1418, loss_box_dn_3: 0.7796, loss_cls_dn_4: 0.1454, loss_box_dn_4: 0.7920, loss_cls_dn_5: 0.1494, loss_box_dn_5: 0.8100, loss_dense_depth: 0.7702, loss: 26.5184, grad_norm: 46.7557 -2026-01-14 21:36:46,557 - mmdet - INFO - Iter [193/17500] lr: 1.767e-04, eta: 11:04:21, time: 1.646, data_time: 0.076, memory: 49164, loss_cls_0: 0.7927, loss_box_0: 1.7158, loss_cns_0: 0.6267, loss_yns_0: 0.1490, loss_cls_1: 0.8846, loss_box_1: 1.6757, loss_cns_1: 0.6565, loss_yns_1: 0.1483, loss_cls_2: 0.9072, loss_box_2: 1.6181, loss_cns_2: 0.6565, loss_yns_2: 0.1476, loss_cls_3: 0.9357, loss_box_3: 1.6170, loss_cns_3: 0.6543, loss_yns_3: 0.1485, loss_cls_4: 0.9261, loss_box_4: 1.6198, loss_cns_4: 0.6583, loss_yns_4: 0.1491, loss_cls_5: 0.9316, loss_box_5: 1.6337, loss_cns_5: 0.6560, loss_yns_5: 0.1508, loss_cls_dn_0: 0.2081, loss_box_dn_0: 0.7721, loss_cls_dn_1: 0.1344, loss_box_dn_1: 0.8006, loss_cls_dn_2: 0.1353, loss_box_dn_2: 0.7966, loss_cls_dn_3: 0.1400, loss_box_dn_3: 0.8251, loss_cls_dn_4: 0.1433, loss_box_dn_4: 0.8391, loss_cls_dn_5: 0.1459, loss_box_dn_5: 0.8606, loss_dense_depth: 0.7858, loss: 26.6466, grad_norm: 54.7663 -2026-01-14 21:36:48,139 - mmdet - INFO - Iter [194/17500] lr: 1.771e-04, eta: 11:03:12, time: 1.561, data_time: 0.075, memory: 49164, loss_cls_0: 0.8048, loss_box_0: 1.7344, loss_cns_0: 0.6254, loss_yns_0: 0.1524, loss_cls_1: 0.9154, loss_box_1: 1.6992, loss_cns_1: 0.6530, loss_yns_1: 0.1511, loss_cls_2: 0.9556, loss_box_2: 1.6670, loss_cns_2: 0.6538, loss_yns_2: 0.1515, loss_cls_3: 0.9308, loss_box_3: 1.6437, loss_cns_3: 0.6536, loss_yns_3: 0.1509, loss_cls_4: 0.9166, loss_box_4: 1.6665, loss_cns_4: 0.6594, loss_yns_4: 0.1538, loss_cls_5: 0.9242, loss_box_5: 1.6553, loss_cns_5: 0.6527, loss_yns_5: 0.1512, loss_cls_dn_0: 0.2082, loss_box_dn_0: 0.7621, loss_cls_dn_1: 0.1308, loss_box_dn_1: 0.8128, loss_cls_dn_2: 0.1343, loss_box_dn_2: 0.8111, loss_cls_dn_3: 0.1338, loss_box_dn_3: 0.8234, loss_cls_dn_4: 0.1359, loss_box_dn_4: 0.8446, loss_cls_dn_5: 0.1420, loss_box_dn_5: 0.8524, loss_dense_depth: 0.7731, loss: 26.8868, grad_norm: 46.2413 -2026-01-14 21:36:49,745 - mmdet - INFO - Iter [195/17500] lr: 1.775e-04, eta: 11:02:11, time: 1.628, data_time: 0.091, memory: 49164, loss_cls_0: 0.7682, loss_box_0: 1.6963, loss_cns_0: 0.6313, loss_yns_0: 0.1513, loss_cls_1: 0.8713, loss_box_1: 1.6243, loss_cns_1: 0.6544, loss_yns_1: 0.1466, loss_cls_2: 0.8841, loss_box_2: 1.6164, loss_cns_2: 0.6576, loss_yns_2: 0.1461, loss_cls_3: 0.8939, loss_box_3: 1.6143, loss_cns_3: 0.6560, loss_yns_3: 0.1480, loss_cls_4: 0.8977, loss_box_4: 1.6182, loss_cns_4: 0.6586, loss_yns_4: 0.1454, loss_cls_5: 0.9117, loss_box_5: 1.5829, loss_cns_5: 0.6574, loss_yns_5: 0.1465, loss_cls_dn_0: 0.1975, loss_box_dn_0: 0.7556, loss_cls_dn_1: 0.1262, loss_box_dn_1: 0.7779, loss_cls_dn_2: 0.1291, loss_box_dn_2: 0.7807, loss_cls_dn_3: 0.1328, loss_box_dn_3: 0.7836, loss_cls_dn_4: 0.1346, loss_box_dn_4: 0.7983, loss_cls_dn_5: 0.1389, loss_box_dn_5: 0.7863, loss_dense_depth: 0.7891, loss: 26.1089, grad_norm: 61.2778 -2026-01-14 21:36:51,313 - mmdet - INFO - Iter [196/17500] lr: 1.779e-04, eta: 11:01:04, time: 1.566, data_time: 0.073, memory: 49164, loss_cls_0: 0.7574, loss_box_0: 1.6759, loss_cns_0: 0.6270, loss_yns_0: 0.1490, loss_cls_1: 0.9159, loss_box_1: 1.5887, loss_cns_1: 0.6528, loss_yns_1: 0.1494, loss_cls_2: 0.8949, loss_box_2: 1.5663, loss_cns_2: 0.6597, loss_yns_2: 0.1490, loss_cls_3: 0.8764, loss_box_3: 1.5546, loss_cns_3: 0.6634, loss_yns_3: 0.1493, loss_cls_4: 0.8855, loss_box_4: 1.5502, loss_cns_4: 0.6603, loss_yns_4: 0.1477, loss_cls_5: 0.9097, loss_box_5: 1.5365, loss_cns_5: 0.6611, loss_yns_5: 0.1475, loss_cls_dn_0: 0.1969, loss_box_dn_0: 0.7577, loss_cls_dn_1: 0.1304, loss_box_dn_1: 0.7790, loss_cls_dn_2: 0.1315, loss_box_dn_2: 0.7685, loss_cls_dn_3: 0.1393, loss_box_dn_3: 0.7606, loss_cls_dn_4: 0.1411, loss_box_dn_4: 0.7612, loss_cls_dn_5: 0.1426, loss_box_dn_5: 0.7539, loss_dense_depth: 0.7569, loss: 25.7478, grad_norm: 42.3661 -2026-01-14 21:36:52,957 - mmdet - INFO - Iter [197/17500] lr: 1.783e-04, eta: 11:00:05, time: 1.644, data_time: 0.080, memory: 49164, loss_cls_0: 0.7968, loss_box_0: 1.7354, loss_cns_0: 0.6204, loss_yns_0: 0.1492, loss_cls_1: 0.9046, loss_box_1: 1.6381, loss_cns_1: 0.6509, loss_yns_1: 0.1508, loss_cls_2: 0.9026, loss_box_2: 1.6577, loss_cns_2: 0.6552, loss_yns_2: 0.1505, loss_cls_3: 0.9024, loss_box_3: 1.6321, loss_cns_3: 0.6596, loss_yns_3: 0.1506, loss_cls_4: 0.8964, loss_box_4: 1.6243, loss_cns_4: 0.6576, loss_yns_4: 0.1510, loss_cls_5: 0.8979, loss_box_5: 1.6491, loss_cns_5: 0.6541, loss_yns_5: 0.1502, loss_cls_dn_0: 0.2027, loss_box_dn_0: 0.7495, loss_cls_dn_1: 0.1321, loss_box_dn_1: 0.7090, loss_cls_dn_2: 0.1308, loss_box_dn_2: 0.7105, loss_cls_dn_3: 0.1360, loss_box_dn_3: 0.7065, loss_cls_dn_4: 0.1401, loss_box_dn_4: 0.7032, loss_cls_dn_5: 0.1450, loss_box_dn_5: 0.7172, loss_dense_depth: 0.8013, loss: 26.0213, grad_norm: 49.6155 -2026-01-14 21:36:54,532 - mmdet - INFO - Iter [198/17500] lr: 1.787e-04, eta: 10:59:00, time: 1.576, data_time: 0.075, memory: 49164, loss_cls_0: 0.8098, loss_box_0: 1.7508, loss_cns_0: 0.6191, loss_yns_0: 0.1552, loss_cls_1: 0.8818, loss_box_1: 1.6438, loss_cns_1: 0.6531, loss_yns_1: 0.1539, loss_cls_2: 0.9023, loss_box_2: 1.6382, loss_cns_2: 0.6528, loss_yns_2: 0.1527, loss_cls_3: 0.9076, loss_box_3: 1.6338, loss_cns_3: 0.6545, loss_yns_3: 0.1524, loss_cls_4: 0.9081, loss_box_4: 1.6271, loss_cns_4: 0.6557, loss_yns_4: 0.1526, loss_cls_5: 0.9088, loss_box_5: 1.6247, loss_cns_5: 0.6510, loss_yns_5: 0.1549, loss_cls_dn_0: 0.2016, loss_box_dn_0: 0.7631, loss_cls_dn_1: 0.1332, loss_box_dn_1: 0.7120, loss_cls_dn_2: 0.1341, loss_box_dn_2: 0.7187, loss_cls_dn_3: 0.1358, loss_box_dn_3: 0.7325, loss_cls_dn_4: 0.1417, loss_box_dn_4: 0.7378, loss_cls_dn_5: 0.1453, loss_box_dn_5: 0.7491, loss_dense_depth: 0.7991, loss: 26.1487, grad_norm: 47.4839 -2026-01-14 21:36:56,097 - mmdet - INFO - Iter [199/17500] lr: 1.791e-04, eta: 10:57:55, time: 1.565, data_time: 0.084, memory: 49164, loss_cls_0: 0.7922, loss_box_0: 1.7438, loss_cns_0: 0.6113, loss_yns_0: 0.1548, loss_cls_1: 0.8945, loss_box_1: 1.6732, loss_cns_1: 0.6544, loss_yns_1: 0.1549, loss_cls_2: 0.9024, loss_box_2: 1.6542, loss_cns_2: 0.6561, loss_yns_2: 0.1553, loss_cls_3: 0.9014, loss_box_3: 1.6703, loss_cns_3: 0.6523, loss_yns_3: 0.1547, loss_cls_4: 0.9104, loss_box_4: 1.6485, loss_cns_4: 0.6495, loss_yns_4: 0.1533, loss_cls_5: 0.8998, loss_box_5: 1.6633, loss_cns_5: 0.6499, loss_yns_5: 0.1554, loss_cls_dn_0: 0.1981, loss_box_dn_0: 0.7623, loss_cls_dn_1: 0.1293, loss_box_dn_1: 0.7440, loss_cls_dn_2: 0.1330, loss_box_dn_2: 0.7465, loss_cls_dn_3: 0.1423, loss_box_dn_3: 0.7712, loss_cls_dn_4: 0.1441, loss_box_dn_4: 0.7827, loss_cls_dn_5: 0.1439, loss_box_dn_5: 0.8050, loss_dense_depth: 0.7811, loss: 26.4394, grad_norm: 50.1729 -2026-01-14 21:36:57,694 - mmdet - INFO - Iter [200/17500] lr: 1.795e-04, eta: 10:56:54, time: 1.596, data_time: 0.083, memory: 49164, loss_cls_0: 0.7987, loss_box_0: 1.7291, loss_cns_0: 0.6086, loss_yns_0: 0.1503, loss_cls_1: 0.9202, loss_box_1: 1.6988, loss_cns_1: 0.6565, loss_yns_1: 0.1520, loss_cls_2: 0.9052, loss_box_2: 1.6583, loss_cns_2: 0.6546, loss_yns_2: 0.1514, loss_cls_3: 0.8907, loss_box_3: 1.6562, loss_cns_3: 0.6542, loss_yns_3: 0.1501, loss_cls_4: 0.8948, loss_box_4: 1.6452, loss_cns_4: 0.6512, loss_yns_4: 0.1503, loss_cls_5: 0.9096, loss_box_5: 1.6550, loss_cns_5: 0.6529, loss_yns_5: 0.1497, loss_cls_dn_0: 0.2018, loss_box_dn_0: 0.7549, loss_cls_dn_1: 0.1290, loss_box_dn_1: 0.7713, loss_cls_dn_2: 0.1292, loss_box_dn_2: 0.7731, loss_cls_dn_3: 0.1380, loss_box_dn_3: 0.7810, loss_cls_dn_4: 0.1354, loss_box_dn_4: 0.7885, loss_cls_dn_5: 0.1372, loss_box_dn_5: 0.8091, loss_dense_depth: 0.7823, loss: 26.4747, grad_norm: 40.3889 -2026-01-14 21:36:59,370 - mmdet - INFO - Iter [201/17500] lr: 1.799e-04, eta: 10:56:00, time: 1.676, data_time: 0.113, memory: 49164, loss_cls_0: 0.8030, loss_box_0: 1.7041, loss_cns_0: 0.6143, loss_yns_0: 0.1485, loss_cls_1: 0.8883, loss_box_1: 1.7247, loss_cns_1: 0.6532, loss_yns_1: 0.1504, loss_cls_2: 0.9004, loss_box_2: 1.6889, loss_cns_2: 0.6502, loss_yns_2: 0.1487, loss_cls_3: 0.9044, loss_box_3: 1.6661, loss_cns_3: 0.6506, loss_yns_3: 0.1488, loss_cls_4: 0.9064, loss_box_4: 1.6817, loss_cns_4: 0.6499, loss_yns_4: 0.1482, loss_cls_5: 0.9301, loss_box_5: 1.6587, loss_cns_5: 0.6530, loss_yns_5: 0.1506, loss_cls_dn_0: 0.1995, loss_box_dn_0: 0.7579, loss_cls_dn_1: 0.1239, loss_box_dn_1: 0.7947, loss_cls_dn_2: 0.1254, loss_box_dn_2: 0.7956, loss_cls_dn_3: 0.1340, loss_box_dn_3: 0.7889, loss_cls_dn_4: 0.1376, loss_box_dn_4: 0.8018, loss_cls_dn_5: 0.1467, loss_box_dn_5: 0.8028, loss_dense_depth: 0.7490, loss: 26.5811, grad_norm: 51.6741 -2026-01-14 21:37:01,073 - mmdet - INFO - Iter [202/17500] lr: 1.803e-04, eta: 10:55:08, time: 1.702, data_time: 0.171, memory: 49164, loss_cls_0: 0.7737, loss_box_0: 1.6839, loss_cns_0: 0.6181, loss_yns_0: 0.1512, loss_cls_1: 0.8794, loss_box_1: 1.6520, loss_cns_1: 0.6480, loss_yns_1: 0.1506, loss_cls_2: 0.8844, loss_box_2: 1.6203, loss_cns_2: 0.6560, loss_yns_2: 0.1511, loss_cls_3: 0.8926, loss_box_3: 1.6092, loss_cns_3: 0.6528, loss_yns_3: 0.1505, loss_cls_4: 0.8984, loss_box_4: 1.6032, loss_cns_4: 0.6536, loss_yns_4: 0.1510, loss_cls_5: 0.9110, loss_box_5: 1.5880, loss_cns_5: 0.6558, loss_yns_5: 0.1514, loss_cls_dn_0: 0.1925, loss_box_dn_0: 0.7635, loss_cls_dn_1: 0.1259, loss_box_dn_1: 0.7763, loss_cls_dn_2: 0.1272, loss_box_dn_2: 0.7636, loss_cls_dn_3: 0.1284, loss_box_dn_3: 0.7613, loss_cls_dn_4: 0.1368, loss_box_dn_4: 0.7669, loss_cls_dn_5: 0.1484, loss_box_dn_5: 0.7661, loss_dense_depth: 0.7850, loss: 26.0284, grad_norm: 36.0611 -2026-01-14 21:37:02,674 - mmdet - INFO - Iter [203/17500] lr: 1.807e-04, eta: 10:54:06, time: 1.572, data_time: 0.075, memory: 49164, loss_cls_0: 0.8037, loss_box_0: 1.7133, loss_cns_0: 0.6175, loss_yns_0: 0.1502, loss_cls_1: 0.8661, loss_box_1: 1.7154, loss_cns_1: 0.6418, loss_yns_1: 0.1483, loss_cls_2: 0.9116, loss_box_2: 1.6990, loss_cns_2: 0.6437, loss_yns_2: 0.1501, loss_cls_3: 0.9265, loss_box_3: 1.6903, loss_cns_3: 0.6472, loss_yns_3: 0.1533, loss_cls_4: 0.9091, loss_box_4: 1.6854, loss_cns_4: 0.6483, loss_yns_4: 0.1533, loss_cls_5: 0.9203, loss_box_5: 1.6967, loss_cns_5: 0.6477, loss_yns_5: 0.1523, loss_cls_dn_0: 0.1929, loss_box_dn_0: 0.7569, loss_cls_dn_1: 0.1239, loss_box_dn_1: 0.7740, loss_cls_dn_2: 0.1277, loss_box_dn_2: 0.7577, loss_cls_dn_3: 0.1288, loss_box_dn_3: 0.7536, loss_cls_dn_4: 0.1298, loss_box_dn_4: 0.7524, loss_cls_dn_5: 0.1354, loss_box_dn_5: 0.7606, loss_dense_depth: 0.8010, loss: 26.4862, grad_norm: 54.7307 -2026-01-14 21:37:04,242 - mmdet - INFO - Iter [204/17500] lr: 1.811e-04, eta: 10:53:07, time: 1.596, data_time: 0.091, memory: 49164, loss_cls_0: 0.7893, loss_box_0: 1.7092, loss_cns_0: 0.6248, loss_yns_0: 0.1488, loss_cls_1: 0.8837, loss_box_1: 1.6863, loss_cns_1: 0.6514, loss_yns_1: 0.1462, loss_cls_2: 0.9017, loss_box_2: 1.6518, loss_cns_2: 0.6520, loss_yns_2: 0.1488, loss_cls_3: 0.9108, loss_box_3: 1.6232, loss_cns_3: 0.6501, loss_yns_3: 0.1504, loss_cls_4: 0.8973, loss_box_4: 1.6452, loss_cns_4: 0.6559, loss_yns_4: 0.1499, loss_cls_5: 0.9199, loss_box_5: 1.6467, loss_cns_5: 0.6567, loss_yns_5: 0.1512, loss_cls_dn_0: 0.1971, loss_box_dn_0: 0.7496, loss_cls_dn_1: 0.1211, loss_box_dn_1: 0.6967, loss_cls_dn_2: 0.1220, loss_box_dn_2: 0.6816, loss_cls_dn_3: 0.1245, loss_box_dn_3: 0.6842, loss_cls_dn_4: 0.1256, loss_box_dn_4: 0.6965, loss_cls_dn_5: 0.1355, loss_box_dn_5: 0.7101, loss_dense_depth: 0.8064, loss: 25.9020, grad_norm: 44.1873 -2026-01-14 21:37:05,932 - mmdet - INFO - Iter [205/17500] lr: 1.815e-04, eta: 10:52:16, time: 1.693, data_time: 0.078, memory: 49164, loss_cls_0: 0.7862, loss_box_0: 1.7522, loss_cns_0: 0.6179, loss_yns_0: 0.1514, loss_cls_1: 0.8947, loss_box_1: 1.6491, loss_cns_1: 0.6430, loss_yns_1: 0.1475, loss_cls_2: 0.9217, loss_box_2: 1.6110, loss_cns_2: 0.6480, loss_yns_2: 0.1483, loss_cls_3: 0.9310, loss_box_3: 1.6178, loss_cns_3: 0.6451, loss_yns_3: 0.1482, loss_cls_4: 0.9162, loss_box_4: 1.6602, loss_cns_4: 0.6519, loss_yns_4: 0.1500, loss_cls_5: 0.9247, loss_box_5: 1.6629, loss_cns_5: 0.6502, loss_yns_5: 0.1508, loss_cls_dn_0: 0.2037, loss_box_dn_0: 0.7735, loss_cls_dn_1: 0.1324, loss_box_dn_1: 0.7174, loss_cls_dn_2: 0.1318, loss_box_dn_2: 0.7147, loss_cls_dn_3: 0.1321, loss_box_dn_3: 0.7384, loss_cls_dn_4: 0.1323, loss_box_dn_4: 0.7632, loss_cls_dn_5: 0.1407, loss_box_dn_5: 0.7905, loss_dense_depth: 0.7916, loss: 26.2424, grad_norm: 60.0342 -2026-01-14 21:37:07,538 - mmdet - INFO - Iter [206/17500] lr: 1.819e-04, eta: 10:51:16, time: 1.577, data_time: 0.075, memory: 49164, loss_cls_0: 0.7978, loss_box_0: 1.7219, loss_cns_0: 0.6256, loss_yns_0: 0.1521, loss_cls_1: 0.8805, loss_box_1: 1.6982, loss_cns_1: 0.6505, loss_yns_1: 0.1499, loss_cls_2: 0.9087, loss_box_2: 1.6457, loss_cns_2: 0.6568, loss_yns_2: 0.1518, loss_cls_3: 0.9130, loss_box_3: 1.6523, loss_cns_3: 0.6536, loss_yns_3: 0.1511, loss_cls_4: 0.9140, loss_box_4: 1.6628, loss_cns_4: 0.6551, loss_yns_4: 0.1520, loss_cls_5: 0.9264, loss_box_5: 1.6852, loss_cns_5: 0.6618, loss_yns_5: 0.1520, loss_cls_dn_0: 0.2038, loss_box_dn_0: 0.7635, loss_cls_dn_1: 0.1270, loss_box_dn_1: 0.7511, loss_cls_dn_2: 0.1297, loss_box_dn_2: 0.7482, loss_cls_dn_3: 0.1305, loss_box_dn_3: 0.7695, loss_cls_dn_4: 0.1321, loss_box_dn_4: 0.7925, loss_cls_dn_5: 0.1441, loss_box_dn_5: 0.8285, loss_dense_depth: 0.8084, loss: 26.5477, grad_norm: 48.6443 -2026-01-14 21:37:09,091 - mmdet - INFO - Iter [207/17500] lr: 1.823e-04, eta: 10:50:18, time: 1.582, data_time: 0.091, memory: 49164, loss_cls_0: 0.7873, loss_box_0: 1.7278, loss_cns_0: 0.6240, loss_yns_0: 0.1537, loss_cls_1: 0.9061, loss_box_1: 1.6998, loss_cns_1: 0.6548, loss_yns_1: 0.1507, loss_cls_2: 0.9193, loss_box_2: 1.6699, loss_cns_2: 0.6588, loss_yns_2: 0.1527, loss_cls_3: 0.9207, loss_box_3: 1.6559, loss_cns_3: 0.6549, loss_yns_3: 0.1528, loss_cls_4: 0.9108, loss_box_4: 1.6679, loss_cns_4: 0.6553, loss_yns_4: 0.1542, loss_cls_5: 0.9321, loss_box_5: 1.6652, loss_cns_5: 0.6621, loss_yns_5: 0.1520, loss_cls_dn_0: 0.2041, loss_box_dn_0: 0.7534, loss_cls_dn_1: 0.1337, loss_box_dn_1: 0.7727, loss_cls_dn_2: 0.1430, loss_box_dn_2: 0.7706, loss_cls_dn_3: 0.1367, loss_box_dn_3: 0.7835, loss_cls_dn_4: 0.1360, loss_box_dn_4: 0.8069, loss_cls_dn_5: 0.1456, loss_box_dn_5: 0.8288, loss_dense_depth: 0.7957, loss: 26.6996, grad_norm: 57.2848 -2026-01-14 21:37:10,735 - mmdet - INFO - Iter [208/17500] lr: 1.827e-04, eta: 10:49:20, time: 1.596, data_time: 0.074, memory: 49164, loss_cls_0: 0.7914, loss_box_0: 1.7312, loss_cns_0: 0.6226, loss_yns_0: 0.1547, loss_cls_1: 0.9014, loss_box_1: 1.7325, loss_cns_1: 0.6532, loss_yns_1: 0.1535, loss_cls_2: 0.9074, loss_box_2: 1.6721, loss_cns_2: 0.6527, loss_yns_2: 0.1547, loss_cls_3: 0.9125, loss_box_3: 1.6575, loss_cns_3: 0.6539, loss_yns_3: 0.1554, loss_cls_4: 0.9068, loss_box_4: 1.6759, loss_cns_4: 0.6529, loss_yns_4: 0.1557, loss_cls_5: 0.9264, loss_box_5: 1.6609, loss_cns_5: 0.6531, loss_yns_5: 0.1538, loss_cls_dn_0: 0.2021, loss_box_dn_0: 0.7633, loss_cls_dn_1: 0.1297, loss_box_dn_1: 0.7679, loss_cls_dn_2: 0.1361, loss_box_dn_2: 0.7533, loss_cls_dn_3: 0.1325, loss_box_dn_3: 0.7567, loss_cls_dn_4: 0.1339, loss_box_dn_4: 0.7756, loss_cls_dn_5: 0.1406, loss_box_dn_5: 0.7834, loss_dense_depth: 0.7673, loss: 26.5344, grad_norm: 56.3326 -2026-01-14 21:37:12,293 - mmdet - INFO - Iter [209/17500] lr: 1.831e-04, eta: 10:48:25, time: 1.605, data_time: 0.108, memory: 49164, loss_cls_0: 0.8245, loss_box_0: 1.7514, loss_cns_0: 0.6160, loss_yns_0: 0.1520, loss_cls_1: 0.8976, loss_box_1: 1.6984, loss_cns_1: 0.6517, loss_yns_1: 0.1505, loss_cls_2: 0.9083, loss_box_2: 1.6591, loss_cns_2: 0.6527, loss_yns_2: 0.1500, loss_cls_3: 0.9218, loss_box_3: 1.6509, loss_cns_3: 0.6548, loss_yns_3: 0.1511, loss_cls_4: 0.9238, loss_box_4: 1.6334, loss_cns_4: 0.6577, loss_yns_4: 0.1499, loss_cls_5: 0.9287, loss_box_5: 1.6488, loss_cns_5: 0.6565, loss_yns_5: 0.1509, loss_cls_dn_0: 0.2070, loss_box_dn_0: 0.7599, loss_cls_dn_1: 0.1278, loss_box_dn_1: 0.7539, loss_cls_dn_2: 0.1284, loss_box_dn_2: 0.7469, loss_cls_dn_3: 0.1323, loss_box_dn_3: 0.7467, loss_cls_dn_4: 0.1396, loss_box_dn_4: 0.7486, loss_cls_dn_5: 0.1463, loss_box_dn_5: 0.7590, loss_dense_depth: 0.8220, loss: 26.4589, grad_norm: 48.3193 -2026-01-14 21:37:13,891 - mmdet - INFO - Iter [210/17500] lr: 1.835e-04, eta: 10:47:29, time: 1.598, data_time: 0.073, memory: 49164, loss_cls_0: 0.7822, loss_box_0: 1.7180, loss_cns_0: 0.6193, loss_yns_0: 0.1513, loss_cls_1: 0.8784, loss_box_1: 1.6501, loss_cns_1: 0.6521, loss_yns_1: 0.1480, loss_cls_2: 0.9013, loss_box_2: 1.6279, loss_cns_2: 0.6536, loss_yns_2: 0.1500, loss_cls_3: 0.9151, loss_box_3: 1.6133, loss_cns_3: 0.6543, loss_yns_3: 0.1496, loss_cls_4: 0.9131, loss_box_4: 1.5895, loss_cns_4: 0.6588, loss_yns_4: 0.1501, loss_cls_5: 0.9010, loss_box_5: 1.6113, loss_cns_5: 0.6593, loss_yns_5: 0.1494, loss_cls_dn_0: 0.1910, loss_box_dn_0: 0.7538, loss_cls_dn_1: 0.1269, loss_box_dn_1: 0.7327, loss_cls_dn_2: 0.1257, loss_box_dn_2: 0.7253, loss_cls_dn_3: 0.1282, loss_box_dn_3: 0.7172, loss_cls_dn_4: 0.1352, loss_box_dn_4: 0.7106, loss_cls_dn_5: 0.1358, loss_box_dn_5: 0.7187, loss_dense_depth: 0.7590, loss: 25.8572, grad_norm: 46.2945 -2026-01-14 21:37:15,500 - mmdet - INFO - Iter [211/17500] lr: 1.839e-04, eta: 10:46:30, time: 1.562, data_time: 0.073, memory: 49164, loss_cls_0: 0.7834, loss_box_0: 1.7031, loss_cns_0: 0.6236, loss_yns_0: 0.1497, loss_cls_1: 0.8859, loss_box_1: 1.6099, loss_cns_1: 0.6551, loss_yns_1: 0.1489, loss_cls_2: 0.9078, loss_box_2: 1.5720, loss_cns_2: 0.6571, loss_yns_2: 0.1494, loss_cls_3: 0.9202, loss_box_3: 1.5767, loss_cns_3: 0.6589, loss_yns_3: 0.1490, loss_cls_4: 0.9129, loss_box_4: 1.5799, loss_cns_4: 0.6587, loss_yns_4: 0.1496, loss_cls_5: 0.9118, loss_box_5: 1.5823, loss_cns_5: 0.6569, loss_yns_5: 0.1487, loss_cls_dn_0: 0.1893, loss_box_dn_0: 0.7515, loss_cls_dn_1: 0.1242, loss_box_dn_1: 0.6990, loss_cls_dn_2: 0.1217, loss_box_dn_2: 0.6885, loss_cls_dn_3: 0.1229, loss_box_dn_3: 0.6984, loss_cls_dn_4: 0.1298, loss_box_dn_4: 0.7108, loss_cls_dn_5: 0.1320, loss_box_dn_5: 0.7187, loss_dense_depth: 0.7699, loss: 25.6082, grad_norm: 49.2623 -2026-01-14 21:37:17,133 - mmdet - INFO - Iter [212/17500] lr: 1.843e-04, eta: 10:45:42, time: 1.680, data_time: 0.127, memory: 49164, loss_cls_0: 0.7840, loss_box_0: 1.6992, loss_cns_0: 0.6278, loss_yns_0: 0.1475, loss_cls_1: 0.8815, loss_box_1: 1.6513, loss_cns_1: 0.6503, loss_yns_1: 0.1461, loss_cls_2: 0.8900, loss_box_2: 1.6314, loss_cns_2: 0.6532, loss_yns_2: 0.1460, loss_cls_3: 0.8928, loss_box_3: 1.6467, loss_cns_3: 0.6574, loss_yns_3: 0.1478, loss_cls_4: 0.8972, loss_box_4: 1.6499, loss_cns_4: 0.6601, loss_yns_4: 0.1476, loss_cls_5: 0.9061, loss_box_5: 1.6572, loss_cns_5: 0.6529, loss_yns_5: 0.1471, loss_cls_dn_0: 0.1967, loss_box_dn_0: 0.7536, loss_cls_dn_1: 0.1289, loss_box_dn_1: 0.7152, loss_cls_dn_2: 0.1273, loss_box_dn_2: 0.7216, loss_cls_dn_3: 0.1266, loss_box_dn_3: 0.7439, loss_cls_dn_4: 0.1298, loss_box_dn_4: 0.7629, loss_cls_dn_5: 0.1330, loss_box_dn_5: 0.7796, loss_dense_depth: 0.7631, loss: 26.0534, grad_norm: 59.3440 -2026-01-14 21:37:18,820 - mmdet - INFO - Iter [213/17500] lr: 1.847e-04, eta: 10:44:51, time: 1.641, data_time: 0.079, memory: 49164, loss_cls_0: 0.7904, loss_box_0: 1.6884, loss_cns_0: 0.6289, loss_yns_0: 0.1531, loss_cls_1: 0.8591, loss_box_1: 1.6391, loss_cns_1: 0.6500, loss_yns_1: 0.1494, loss_cls_2: 0.8753, loss_box_2: 1.6130, loss_cns_2: 0.6564, loss_yns_2: 0.1505, loss_cls_3: 0.8862, loss_box_3: 1.5899, loss_cns_3: 0.6561, loss_yns_3: 0.1492, loss_cls_4: 0.8813, loss_box_4: 1.6010, loss_cns_4: 0.6573, loss_yns_4: 0.1483, loss_cls_5: 0.8795, loss_box_5: 1.6105, loss_cns_5: 0.6562, loss_yns_5: 0.1486, loss_cls_dn_0: 0.1982, loss_box_dn_0: 0.7479, loss_cls_dn_1: 0.1262, loss_box_dn_1: 0.7589, loss_cls_dn_2: 0.1276, loss_box_dn_2: 0.7719, loss_cls_dn_3: 0.1260, loss_box_dn_3: 0.7940, loss_cls_dn_4: 0.1323, loss_box_dn_4: 0.8173, loss_cls_dn_5: 0.1409, loss_box_dn_5: 0.8394, loss_dense_depth: 0.7914, loss: 26.0899, grad_norm: 36.2895 -2026-01-14 21:37:20,461 - mmdet - INFO - Iter [214/17500] lr: 1.851e-04, eta: 10:44:02, time: 1.660, data_time: 0.106, memory: 49164, loss_cls_0: 0.7826, loss_box_0: 1.7086, loss_cns_0: 0.6256, loss_yns_0: 0.1530, loss_cls_1: 0.8640, loss_box_1: 1.6507, loss_cns_1: 0.6497, loss_yns_1: 0.1509, loss_cls_2: 0.8939, loss_box_2: 1.6360, loss_cns_2: 0.6533, loss_yns_2: 0.1510, loss_cls_3: 0.9151, loss_box_3: 1.6102, loss_cns_3: 0.6550, loss_yns_3: 0.1509, loss_cls_4: 0.8964, loss_box_4: 1.6408, loss_cns_4: 0.6525, loss_yns_4: 0.1503, loss_cls_5: 0.9044, loss_box_5: 1.6436, loss_cns_5: 0.6519, loss_yns_5: 0.1514, loss_cls_dn_0: 0.1962, loss_box_dn_0: 0.7590, loss_cls_dn_1: 0.1252, loss_box_dn_1: 0.7878, loss_cls_dn_2: 0.1262, loss_box_dn_2: 0.7987, loss_cls_dn_3: 0.1291, loss_box_dn_3: 0.8098, loss_cls_dn_4: 0.1396, loss_box_dn_4: 0.8288, loss_cls_dn_5: 0.1474, loss_box_dn_5: 0.8439, loss_dense_depth: 0.8079, loss: 26.4416, grad_norm: 54.9015 -2026-01-14 21:37:22,020 - mmdet - INFO - Iter [215/17500] lr: 1.855e-04, eta: 10:43:07, time: 1.584, data_time: 0.090, memory: 49164, loss_cls_0: 0.8126, loss_box_0: 1.7058, loss_cns_0: 0.6213, loss_yns_0: 0.1508, loss_cls_1: 0.8896, loss_box_1: 1.6567, loss_cns_1: 0.6514, loss_yns_1: 0.1512, loss_cls_2: 0.8978, loss_box_2: 1.6399, loss_cns_2: 0.6538, loss_yns_2: 0.1507, loss_cls_3: 0.9209, loss_box_3: 1.5975, loss_cns_3: 0.6578, loss_yns_3: 0.1516, loss_cls_4: 0.9082, loss_box_4: 1.6055, loss_cns_4: 0.6564, loss_yns_4: 0.1504, loss_cls_5: 0.9205, loss_box_5: 1.6047, loss_cns_5: 0.6562, loss_yns_5: 0.1506, loss_cls_dn_0: 0.2050, loss_box_dn_0: 0.7496, loss_cls_dn_1: 0.1330, loss_box_dn_1: 0.7674, loss_cls_dn_2: 0.1306, loss_box_dn_2: 0.7682, loss_cls_dn_3: 0.1341, loss_box_dn_3: 0.7613, loss_cls_dn_4: 0.1386, loss_box_dn_4: 0.7662, loss_cls_dn_5: 0.1416, loss_box_dn_5: 0.7722, loss_dense_depth: 0.8080, loss: 26.2379, grad_norm: 42.9469 -2026-01-14 21:37:23,584 - mmdet - INFO - Iter [216/17500] lr: 1.859e-04, eta: 10:42:12, time: 1.566, data_time: 0.077, memory: 49164, loss_cls_0: 0.8189, loss_box_0: 1.7395, loss_cns_0: 0.6193, loss_yns_0: 0.1494, loss_cls_1: 0.9215, loss_box_1: 1.6689, loss_cns_1: 0.6545, loss_yns_1: 0.1548, loss_cls_2: 0.9088, loss_box_2: 1.6167, loss_cns_2: 0.6558, loss_yns_2: 0.1515, loss_cls_3: 0.9138, loss_box_3: 1.5924, loss_cns_3: 0.6553, loss_yns_3: 0.1523, loss_cls_4: 0.9099, loss_box_4: 1.5918, loss_cns_4: 0.6570, loss_yns_4: 0.1526, loss_cls_5: 0.9219, loss_box_5: 1.5965, loss_cns_5: 0.6573, loss_yns_5: 0.1526, loss_cls_dn_0: 0.2078, loss_box_dn_0: 0.7675, loss_cls_dn_1: 0.1393, loss_box_dn_1: 0.7733, loss_cls_dn_2: 0.1346, loss_box_dn_2: 0.7580, loss_cls_dn_3: 0.1395, loss_box_dn_3: 0.7447, loss_cls_dn_4: 0.1447, loss_box_dn_4: 0.7431, loss_cls_dn_5: 0.1465, loss_box_dn_5: 0.7430, loss_dense_depth: 0.7935, loss: 26.2483, grad_norm: 41.3554 -2026-01-14 21:37:25,177 - mmdet - INFO - Iter [217/17500] lr: 1.863e-04, eta: 10:41:16, time: 1.561, data_time: 0.075, memory: 49164, loss_cls_0: 0.7918, loss_box_0: 1.7074, loss_cns_0: 0.6222, loss_yns_0: 0.1577, loss_cls_1: 0.8599, loss_box_1: 1.6515, loss_cns_1: 0.6509, loss_yns_1: 0.1561, loss_cls_2: 0.8796, loss_box_2: 1.5997, loss_cns_2: 0.6498, loss_yns_2: 0.1546, loss_cls_3: 0.8852, loss_box_3: 1.5850, loss_cns_3: 0.6600, loss_yns_3: 0.1546, loss_cls_4: 0.8897, loss_box_4: 1.5776, loss_cns_4: 0.6527, loss_yns_4: 0.1541, loss_cls_5: 0.8931, loss_box_5: 1.5877, loss_cns_5: 0.6556, loss_yns_5: 0.1550, loss_cls_dn_0: 0.1923, loss_box_dn_0: 0.7559, loss_cls_dn_1: 0.1282, loss_box_dn_1: 0.7310, loss_cls_dn_2: 0.1328, loss_box_dn_2: 0.7136, loss_cls_dn_3: 0.1358, loss_box_dn_3: 0.7112, loss_cls_dn_4: 0.1436, loss_box_dn_4: 0.7137, loss_cls_dn_5: 0.1501, loss_box_dn_5: 0.7199, loss_dense_depth: 0.7379, loss: 25.6976, grad_norm: 53.9939 -2026-01-14 21:37:26,730 - mmdet - INFO - Iter [218/17500] lr: 1.867e-04, eta: 10:40:23, time: 1.586, data_time: 0.092, memory: 49164, loss_cls_0: 0.7852, loss_box_0: 1.6800, loss_cns_0: 0.6256, loss_yns_0: 0.1537, loss_cls_1: 0.8789, loss_box_1: 1.6436, loss_cns_1: 0.6515, loss_yns_1: 0.1507, loss_cls_2: 0.8880, loss_box_2: 1.6065, loss_cns_2: 0.6483, loss_yns_2: 0.1509, loss_cls_3: 0.8862, loss_box_3: 1.6036, loss_cns_3: 0.6613, loss_yns_3: 0.1531, loss_cls_4: 0.8785, loss_box_4: 1.6042, loss_cns_4: 0.6565, loss_yns_4: 0.1537, loss_cls_5: 0.8877, loss_box_5: 1.6060, loss_cns_5: 0.6575, loss_yns_5: 0.1548, loss_cls_dn_0: 0.1923, loss_box_dn_0: 0.7612, loss_cls_dn_1: 0.1365, loss_box_dn_1: 0.6975, loss_cls_dn_2: 0.1431, loss_box_dn_2: 0.6924, loss_cls_dn_3: 0.1429, loss_box_dn_3: 0.7034, loss_cls_dn_4: 0.1443, loss_box_dn_4: 0.7180, loss_cls_dn_5: 0.1503, loss_box_dn_5: 0.7351, loss_dense_depth: 0.7524, loss: 25.7354, grad_norm: 46.5809 -2026-01-14 21:37:28,342 - mmdet - INFO - Iter [219/17500] lr: 1.871e-04, eta: 10:39:31, time: 1.586, data_time: 0.084, memory: 49164, loss_cls_0: 0.7952, loss_box_0: 1.7185, loss_cns_0: 0.6210, loss_yns_0: 0.1542, loss_cls_1: 0.8891, loss_box_1: 1.6796, loss_cns_1: 0.6471, loss_yns_1: 0.1515, loss_cls_2: 0.8921, loss_box_2: 1.6392, loss_cns_2: 0.6515, loss_yns_2: 0.1508, loss_cls_3: 0.9056, loss_box_3: 1.6199, loss_cns_3: 0.6548, loss_yns_3: 0.1529, loss_cls_4: 0.9103, loss_box_4: 1.6171, loss_cns_4: 0.6539, loss_yns_4: 0.1534, loss_cls_5: 0.9153, loss_box_5: 1.6322, loss_cns_5: 0.6554, loss_yns_5: 0.1542, loss_cls_dn_0: 0.2025, loss_box_dn_0: 0.7608, loss_cls_dn_1: 0.1376, loss_box_dn_1: 0.7351, loss_cls_dn_2: 0.1380, loss_box_dn_2: 0.7415, loss_cls_dn_3: 0.1363, loss_box_dn_3: 0.7623, loss_cls_dn_4: 0.1388, loss_box_dn_4: 0.7865, loss_cls_dn_5: 0.1461, loss_box_dn_5: 0.8183, loss_dense_depth: 0.7317, loss: 26.2500, grad_norm: 50.1926 -2026-01-14 21:37:29,930 - mmdet - INFO - Iter [220/17500] lr: 1.875e-04, eta: 10:38:41, time: 1.615, data_time: 0.091, memory: 49164, loss_cls_0: 0.7805, loss_box_0: 1.7038, loss_cns_0: 0.6280, loss_yns_0: 0.1522, loss_cls_1: 0.8583, loss_box_1: 1.6607, loss_cns_1: 0.6530, loss_yns_1: 0.1509, loss_cls_2: 0.8685, loss_box_2: 1.6486, loss_cns_2: 0.6547, loss_yns_2: 0.1505, loss_cls_3: 0.9035, loss_box_3: 1.6199, loss_cns_3: 0.6605, loss_yns_3: 0.1530, loss_cls_4: 0.9200, loss_box_4: 1.6240, loss_cns_4: 0.6566, loss_yns_4: 0.1533, loss_cls_5: 0.9048, loss_box_5: 1.6321, loss_cns_5: 0.6617, loss_yns_5: 0.1528, loss_cls_dn_0: 0.1982, loss_box_dn_0: 0.7488, loss_cls_dn_1: 0.1337, loss_box_dn_1: 0.7445, loss_cls_dn_2: 0.1322, loss_box_dn_2: 0.7499, loss_cls_dn_3: 0.1333, loss_box_dn_3: 0.7691, loss_cls_dn_4: 0.1414, loss_box_dn_4: 0.7923, loss_cls_dn_5: 0.1438, loss_box_dn_5: 0.8258, loss_dense_depth: 0.7500, loss: 26.2147, grad_norm: 57.4172 -2026-01-14 21:37:31,607 - mmdet - INFO - Iter [221/17500] lr: 1.879e-04, eta: 10:37:57, time: 1.675, data_time: 0.118, memory: 49164, loss_cls_0: 0.7823, loss_box_0: 1.7147, loss_cns_0: 0.6244, loss_yns_0: 0.1516, loss_cls_1: 0.8602, loss_box_1: 1.6697, loss_cns_1: 0.6544, loss_yns_1: 0.1516, loss_cls_2: 0.8827, loss_box_2: 1.6838, loss_cns_2: 0.6536, loss_yns_2: 0.1521, loss_cls_3: 0.8950, loss_box_3: 1.6535, loss_cns_3: 0.6621, loss_yns_3: 0.1549, loss_cls_4: 0.8876, loss_box_4: 1.6566, loss_cns_4: 0.6533, loss_yns_4: 0.1534, loss_cls_5: 0.8833, loss_box_5: 1.6464, loss_cns_5: 0.6625, loss_yns_5: 0.1522, loss_cls_dn_0: 0.1963, loss_box_dn_0: 0.7604, loss_cls_dn_1: 0.1372, loss_box_dn_1: 0.7680, loss_cls_dn_2: 0.1352, loss_box_dn_2: 0.7709, loss_cls_dn_3: 0.1362, loss_box_dn_3: 0.7806, loss_cls_dn_4: 0.1389, loss_box_dn_4: 0.7973, loss_cls_dn_5: 0.1398, loss_box_dn_5: 0.8197, loss_dense_depth: 0.7627, loss: 26.3849, grad_norm: 54.0606 -2026-01-14 21:37:33,260 - mmdet - INFO - Iter [222/17500] lr: 1.883e-04, eta: 10:37:10, time: 1.648, data_time: 0.169, memory: 49164, loss_cls_0: 0.7805, loss_box_0: 1.6946, loss_cns_0: 0.6238, loss_yns_0: 0.1509, loss_cls_1: 0.8639, loss_box_1: 1.6135, loss_cns_1: 0.6574, loss_yns_1: 0.1488, loss_cls_2: 0.8870, loss_box_2: 1.6022, loss_cns_2: 0.6592, loss_yns_2: 0.1499, loss_cls_3: 0.8794, loss_box_3: 1.5799, loss_cns_3: 0.6655, loss_yns_3: 0.1516, loss_cls_4: 0.8885, loss_box_4: 1.5665, loss_cns_4: 0.6618, loss_yns_4: 0.1497, loss_cls_5: 0.8832, loss_box_5: 1.5661, loss_cns_5: 0.6637, loss_yns_5: 0.1494, loss_cls_dn_0: 0.1919, loss_box_dn_0: 0.7535, loss_cls_dn_1: 0.1255, loss_box_dn_1: 0.7465, loss_cls_dn_2: 0.1334, loss_box_dn_2: 0.7420, loss_cls_dn_3: 0.1306, loss_box_dn_3: 0.7394, loss_cls_dn_4: 0.1320, loss_box_dn_4: 0.7429, loss_cls_dn_5: 0.1379, loss_box_dn_5: 0.7578, loss_dense_depth: 0.7428, loss: 25.7134, grad_norm: 40.1061 -2026-01-14 21:37:34,816 - mmdet - INFO - Iter [223/17500] lr: 1.887e-04, eta: 10:36:18, time: 1.563, data_time: 0.075, memory: 49164, loss_cls_0: 0.8141, loss_box_0: 1.7000, loss_cns_0: 0.6218, loss_yns_0: 0.1527, loss_cls_1: 0.8860, loss_box_1: 1.6372, loss_cns_1: 0.6564, loss_yns_1: 0.1494, loss_cls_2: 0.8874, loss_box_2: 1.6029, loss_cns_2: 0.6570, loss_yns_2: 0.1511, loss_cls_3: 0.9260, loss_box_3: 1.5701, loss_cns_3: 0.6560, loss_yns_3: 0.1505, loss_cls_4: 0.9291, loss_box_4: 1.5599, loss_cns_4: 0.6587, loss_yns_4: 0.1530, loss_cls_5: 0.9157, loss_box_5: 1.5794, loss_cns_5: 0.6572, loss_yns_5: 0.1517, loss_cls_dn_0: 0.2048, loss_box_dn_0: 0.7506, loss_cls_dn_1: 0.1329, loss_box_dn_1: 0.7505, loss_cls_dn_2: 0.1378, loss_box_dn_2: 0.7356, loss_cls_dn_3: 0.1374, loss_box_dn_3: 0.7234, loss_cls_dn_4: 0.1421, loss_box_dn_4: 0.7181, loss_cls_dn_5: 0.1450, loss_box_dn_5: 0.7294, loss_dense_depth: 0.7611, loss: 25.8922, grad_norm: 46.0773 -2026-01-14 21:37:36,379 - mmdet - INFO - Iter [224/17500] lr: 1.891e-04, eta: 10:35:25, time: 1.563, data_time: 0.072, memory: 49164, loss_cls_0: 0.7839, loss_box_0: 1.6738, loss_cns_0: 0.6229, loss_yns_0: 0.1497, loss_cls_1: 0.8615, loss_box_1: 1.6671, loss_cns_1: 0.6576, loss_yns_1: 0.1494, loss_cls_2: 0.8710, loss_box_2: 1.6061, loss_cns_2: 0.6561, loss_yns_2: 0.1505, loss_cls_3: 0.8897, loss_box_3: 1.5832, loss_cns_3: 0.6560, loss_yns_3: 0.1531, loss_cls_4: 0.8904, loss_box_4: 1.5830, loss_cns_4: 0.6599, loss_yns_4: 0.1529, loss_cls_5: 0.8726, loss_box_5: 1.5952, loss_cns_5: 0.6628, loss_yns_5: 0.1498, loss_cls_dn_0: 0.1981, loss_box_dn_0: 0.7515, loss_cls_dn_1: 0.1293, loss_box_dn_1: 0.7175, loss_cls_dn_2: 0.1287, loss_box_dn_2: 0.6987, loss_cls_dn_3: 0.1290, loss_box_dn_3: 0.6966, loss_cls_dn_4: 0.1333, loss_box_dn_4: 0.6996, loss_cls_dn_5: 0.1361, loss_box_dn_5: 0.7093, loss_dense_depth: 0.7319, loss: 25.5579, grad_norm: 37.2516 -2026-01-14 21:37:38,023 - mmdet - INFO - Iter [225/17500] lr: 1.895e-04, eta: 10:34:39, time: 1.636, data_time: 0.071, memory: 49164, loss_cls_0: 0.7716, loss_box_0: 1.7103, loss_cns_0: 0.6231, loss_yns_0: 0.1522, loss_cls_1: 0.8749, loss_box_1: 1.6929, loss_cns_1: 0.6555, loss_yns_1: 0.1512, loss_cls_2: 0.8761, loss_box_2: 1.6575, loss_cns_2: 0.6582, loss_yns_2: 0.1548, loss_cls_3: 0.9080, loss_box_3: 1.6600, loss_cns_3: 0.6592, loss_yns_3: 0.1531, loss_cls_4: 0.9078, loss_box_4: 1.6775, loss_cns_4: 0.6614, loss_yns_4: 0.1530, loss_cls_5: 0.9506, loss_box_5: 1.6913, loss_cns_5: 0.6654, loss_yns_5: 0.1553, loss_cls_dn_0: 0.1974, loss_box_dn_0: 0.7500, loss_cls_dn_1: 0.1276, loss_box_dn_1: 0.7278, loss_cls_dn_2: 0.1272, loss_box_dn_2: 0.7300, loss_cls_dn_3: 0.1290, loss_box_dn_3: 0.7516, loss_cls_dn_4: 0.1297, loss_box_dn_4: 0.7758, loss_cls_dn_5: 0.1356, loss_box_dn_5: 0.7938, loss_dense_depth: 0.7257, loss: 26.3219, grad_norm: 54.1442 -2026-01-14 21:37:39,612 - mmdet - INFO - Iter [226/17500] lr: 1.899e-04, eta: 10:33:51, time: 1.595, data_time: 0.080, memory: 49164, loss_cls_0: 0.7795, loss_box_0: 1.6759, loss_cns_0: 0.6284, loss_yns_0: 0.1517, loss_cls_1: 0.9009, loss_box_1: 1.6435, loss_cns_1: 0.6591, loss_yns_1: 0.1524, loss_cls_2: 0.9062, loss_box_2: 1.6405, loss_cns_2: 0.6594, loss_yns_2: 0.1523, loss_cls_3: 0.9347, loss_box_3: 1.6468, loss_cns_3: 0.6613, loss_yns_3: 0.1507, loss_cls_4: 0.9198, loss_box_4: 1.6539, loss_cns_4: 0.6593, loss_yns_4: 0.1511, loss_cls_5: 0.9103, loss_box_5: 1.6532, loss_cns_5: 0.6594, loss_yns_5: 0.1537, loss_cls_dn_0: 0.1950, loss_box_dn_0: 0.7515, loss_cls_dn_1: 0.1333, loss_box_dn_1: 0.7789, loss_cls_dn_2: 0.1302, loss_box_dn_2: 0.7954, loss_cls_dn_3: 0.1327, loss_box_dn_3: 0.8161, loss_cls_dn_4: 0.1325, loss_box_dn_4: 0.8393, loss_cls_dn_5: 0.1360, loss_box_dn_5: 0.8602, loss_dense_depth: 0.7233, loss: 26.5282, grad_norm: 49.2058 -2026-01-14 21:37:41,212 - mmdet - INFO - Iter [227/17500] lr: 1.903e-04, eta: 10:33:03, time: 1.601, data_time: 0.074, memory: 49164, loss_cls_0: 0.7705, loss_box_0: 1.6859, loss_cns_0: 0.6188, loss_yns_0: 0.1469, loss_cls_1: 0.8695, loss_box_1: 1.6606, loss_cns_1: 0.6518, loss_yns_1: 0.1466, loss_cls_2: 0.9088, loss_box_2: 1.6721, loss_cns_2: 0.6529, loss_yns_2: 0.1475, loss_cls_3: 0.8930, loss_box_3: 1.6471, loss_cns_3: 0.6574, loss_yns_3: 0.1476, loss_cls_4: 0.8945, loss_box_4: 1.6533, loss_cns_4: 0.6547, loss_yns_4: 0.1489, loss_cls_5: 0.9181, loss_box_5: 1.6552, loss_cns_5: 0.6520, loss_yns_5: 0.1470, loss_cls_dn_0: 0.1951, loss_box_dn_0: 0.7586, loss_cls_dn_1: 0.1254, loss_box_dn_1: 0.7889, loss_cls_dn_2: 0.1301, loss_box_dn_2: 0.8135, loss_cls_dn_3: 0.1349, loss_box_dn_3: 0.8131, loss_cls_dn_4: 0.1301, loss_box_dn_4: 0.8308, loss_cls_dn_5: 0.1371, loss_box_dn_5: 0.8539, loss_dense_depth: 0.7316, loss: 26.4440, grad_norm: 51.4652 -2026-01-14 21:37:42,764 - mmdet - INFO - Iter [228/17500] lr: 1.907e-04, eta: 10:32:12, time: 1.554, data_time: 0.073, memory: 49164, loss_cls_0: 0.7752, loss_box_0: 1.6976, loss_cns_0: 0.6263, loss_yns_0: 0.1487, loss_cls_1: 0.9030, loss_box_1: 1.6328, loss_cns_1: 0.6495, loss_yns_1: 0.1472, loss_cls_2: 0.9150, loss_box_2: 1.6383, loss_cns_2: 0.6535, loss_yns_2: 0.1486, loss_cls_3: 0.8913, loss_box_3: 1.6124, loss_cns_3: 0.6602, loss_yns_3: 0.1492, loss_cls_4: 0.8917, loss_box_4: 1.6183, loss_cns_4: 0.6572, loss_yns_4: 0.1505, loss_cls_5: 0.9338, loss_box_5: 1.6193, loss_cns_5: 0.6515, loss_yns_5: 0.1485, loss_cls_dn_0: 0.1967, loss_box_dn_0: 0.7484, loss_cls_dn_1: 0.1269, loss_box_dn_1: 0.8138, loss_cls_dn_2: 0.1293, loss_box_dn_2: 0.8274, loss_cls_dn_3: 0.1332, loss_box_dn_3: 0.8220, loss_cls_dn_4: 0.1307, loss_box_dn_4: 0.8365, loss_cls_dn_5: 0.1380, loss_box_dn_5: 0.8605, loss_dense_depth: 0.7358, loss: 26.4191, grad_norm: 44.9475 -2026-01-14 21:37:44,323 - mmdet - INFO - Iter [229/17500] lr: 1.911e-04, eta: 10:31:21, time: 1.557, data_time: 0.074, memory: 49164, loss_cls_0: 0.8096, loss_box_0: 1.7398, loss_cns_0: 0.6244, loss_yns_0: 0.1485, loss_cls_1: 0.8668, loss_box_1: 1.6733, loss_cns_1: 0.6502, loss_yns_1: 0.1461, loss_cls_2: 0.8703, loss_box_2: 1.6404, loss_cns_2: 0.6590, loss_yns_2: 0.1476, loss_cls_3: 0.8962, loss_box_3: 1.6219, loss_cns_3: 0.6624, loss_yns_3: 0.1477, loss_cls_4: 0.8842, loss_box_4: 1.6287, loss_cns_4: 0.6610, loss_yns_4: 0.1500, loss_cls_5: 0.8823, loss_box_5: 1.6297, loss_cns_5: 0.6604, loss_yns_5: 0.1490, loss_cls_dn_0: 0.1883, loss_box_dn_0: 0.7475, loss_cls_dn_1: 0.1284, loss_box_dn_1: 0.7891, loss_cls_dn_2: 0.1272, loss_box_dn_2: 0.7796, loss_cls_dn_3: 0.1304, loss_box_dn_3: 0.7728, loss_cls_dn_4: 0.1352, loss_box_dn_4: 0.7814, loss_cls_dn_5: 0.1374, loss_box_dn_5: 0.7882, loss_dense_depth: 0.7374, loss: 26.1923, grad_norm: 44.6769 -2026-01-14 21:37:45,890 - mmdet - INFO - Iter [230/17500] lr: 1.915e-04, eta: 10:30:32, time: 1.566, data_time: 0.074, memory: 49164, loss_cls_0: 0.7854, loss_box_0: 1.7352, loss_cns_0: 0.6259, loss_yns_0: 0.1485, loss_cls_1: 0.8669, loss_box_1: 1.6769, loss_cns_1: 0.6511, loss_yns_1: 0.1481, loss_cls_2: 0.9114, loss_box_2: 1.6259, loss_cns_2: 0.6588, loss_yns_2: 0.1484, loss_cls_3: 0.9108, loss_box_3: 1.6036, loss_cns_3: 0.6614, loss_yns_3: 0.1474, loss_cls_4: 0.8950, loss_box_4: 1.6103, loss_cns_4: 0.6637, loss_yns_4: 0.1473, loss_cls_5: 0.9039, loss_box_5: 1.6041, loss_cns_5: 0.6652, loss_yns_5: 0.1499, loss_cls_dn_0: 0.1906, loss_box_dn_0: 0.7565, loss_cls_dn_1: 0.1228, loss_box_dn_1: 0.7438, loss_cls_dn_2: 0.1206, loss_box_dn_2: 0.7228, loss_cls_dn_3: 0.1235, loss_box_dn_3: 0.7170, loss_cls_dn_4: 0.1252, loss_box_dn_4: 0.7208, loss_cls_dn_5: 0.1291, loss_box_dn_5: 0.7235, loss_dense_depth: 0.7315, loss: 25.8728, grad_norm: 33.5947 -2026-01-14 21:37:47,516 - mmdet - INFO - Iter [231/17500] lr: 1.919e-04, eta: 10:29:44, time: 1.577, data_time: 0.076, memory: 49164, loss_cls_0: 0.7855, loss_box_0: 1.6510, loss_cns_0: 0.6194, loss_yns_0: 0.1441, loss_cls_1: 0.8383, loss_box_1: 1.5587, loss_cns_1: 0.6539, loss_yns_1: 0.1454, loss_cls_2: 0.8699, loss_box_2: 1.5776, loss_cns_2: 0.6603, loss_yns_2: 0.1470, loss_cls_3: 0.9238, loss_box_3: 1.5088, loss_cns_3: 0.6647, loss_yns_3: 0.1454, loss_cls_4: 0.9158, loss_box_4: 1.5256, loss_cns_4: 0.6650, loss_yns_4: 0.1475, loss_cls_5: 0.9165, loss_box_5: 1.5532, loss_cns_5: 0.6716, loss_yns_5: 0.1505, loss_cls_dn_0: 0.1870, loss_box_dn_0: 0.7544, loss_cls_dn_1: 0.1196, loss_box_dn_1: 0.6944, loss_cls_dn_2: 0.1221, loss_box_dn_2: 0.6979, loss_cls_dn_3: 0.1338, loss_box_dn_3: 0.6981, loss_cls_dn_4: 0.1343, loss_box_dn_4: 0.7142, loss_cls_dn_5: 0.1416, loss_box_dn_5: 0.7328, loss_dense_depth: 0.7199, loss: 25.2895, grad_norm: 51.7444 -2026-01-14 21:37:49,093 - mmdet - INFO - Iter [232/17500] lr: 1.923e-04, eta: 10:28:59, time: 1.615, data_time: 0.111, memory: 49164, loss_cls_0: 0.8297, loss_box_0: 1.6267, loss_cns_0: 0.6070, loss_yns_0: 0.1393, loss_cls_1: 0.8505, loss_box_1: 1.6137, loss_cns_1: 0.6511, loss_yns_1: 0.1452, loss_cls_2: 0.8676, loss_box_2: 1.6090, loss_cns_2: 0.6537, loss_yns_2: 0.1457, loss_cls_3: 0.9386, loss_box_3: 1.4948, loss_cns_3: 0.6401, loss_yns_3: 0.1408, loss_cls_4: 0.9425, loss_box_4: 1.5652, loss_cns_4: 0.6514, loss_yns_4: 0.1461, loss_cls_5: 0.8938, loss_box_5: 1.6330, loss_cns_5: 0.6574, loss_yns_5: 0.1483, loss_cls_dn_0: 0.1868, loss_box_dn_0: 0.7683, loss_cls_dn_1: 0.1262, loss_box_dn_1: 0.7310, loss_cls_dn_2: 0.1240, loss_box_dn_2: 0.7441, loss_cls_dn_3: 0.1311, loss_box_dn_3: 0.7529, loss_cls_dn_4: 0.1419, loss_box_dn_4: 0.7783, loss_cls_dn_5: 0.1424, loss_box_dn_5: 0.8104, loss_dense_depth: 0.7249, loss: 25.7534, grad_norm: 53.1739 -2026-01-14 21:37:50,651 - mmdet - INFO - Iter [233/17500] lr: 1.927e-04, eta: 10:28:11, time: 1.570, data_time: 0.087, memory: 49164, loss_cls_0: 0.8165, loss_box_0: 1.6395, loss_cns_0: 0.6168, loss_yns_0: 0.1457, loss_cls_1: 0.8825, loss_box_1: 1.6084, loss_cns_1: 0.6513, loss_yns_1: 0.1482, loss_cls_2: 0.8795, loss_box_2: 1.6059, loss_cns_2: 0.6553, loss_yns_2: 0.1471, loss_cls_3: 0.9019, loss_box_3: 1.5461, loss_cns_3: 0.6540, loss_yns_3: 0.1476, loss_cls_4: 0.9137, loss_box_4: 1.5637, loss_cns_4: 0.6576, loss_yns_4: 0.1474, loss_cls_5: 0.9362, loss_box_5: 1.6017, loss_cns_5: 0.6580, loss_yns_5: 0.1481, loss_cls_dn_0: 0.1847, loss_box_dn_0: 0.7541, loss_cls_dn_1: 0.1301, loss_box_dn_1: 0.7631, loss_cls_dn_2: 0.1313, loss_box_dn_2: 0.7751, loss_cls_dn_3: 0.1315, loss_box_dn_3: 0.7886, loss_cls_dn_4: 0.1351, loss_box_dn_4: 0.8081, loss_cls_dn_5: 0.1433, loss_box_dn_5: 0.8379, loss_dense_depth: 0.7267, loss: 25.9825, grad_norm: 48.6338 -2026-01-14 21:37:52,235 - mmdet - INFO - Iter [234/17500] lr: 1.931e-04, eta: 10:27:25, time: 1.585, data_time: 0.073, memory: 49164, loss_cls_0: 0.7689, loss_box_0: 1.6613, loss_cns_0: 0.6282, loss_yns_0: 0.1482, loss_cls_1: 0.9158, loss_box_1: 1.6246, loss_cns_1: 0.6540, loss_yns_1: 0.1516, loss_cls_2: 0.9028, loss_box_2: 1.5785, loss_cns_2: 0.6580, loss_yns_2: 0.1519, loss_cls_3: 0.9268, loss_box_3: 1.5441, loss_cns_3: 0.6652, loss_yns_3: 0.1532, loss_cls_4: 0.9303, loss_box_4: 1.5393, loss_cns_4: 0.6623, loss_yns_4: 0.1511, loss_cls_5: 0.9778, loss_box_5: 1.5553, loss_cns_5: 0.6658, loss_yns_5: 0.1526, loss_cls_dn_0: 0.1737, loss_box_dn_0: 0.7489, loss_cls_dn_1: 0.1213, loss_box_dn_1: 0.7520, loss_cls_dn_2: 0.1233, loss_box_dn_2: 0.7469, loss_cls_dn_3: 0.1256, loss_box_dn_3: 0.7506, loss_cls_dn_4: 0.1272, loss_box_dn_4: 0.7617, loss_cls_dn_5: 0.1406, loss_box_dn_5: 0.7825, loss_dense_depth: 0.7274, loss: 25.8491, grad_norm: 39.5159 -2026-01-14 21:37:53,821 - mmdet - INFO - Iter [235/17500] lr: 1.935e-04, eta: 10:26:37, time: 1.559, data_time: 0.073, memory: 49164, loss_cls_0: 0.7836, loss_box_0: 1.6741, loss_cns_0: 0.6330, loss_yns_0: 0.1499, loss_cls_1: 0.8665, loss_box_1: 1.5924, loss_cns_1: 0.6574, loss_yns_1: 0.1483, loss_cls_2: 0.8977, loss_box_2: 1.5758, loss_cns_2: 0.6594, loss_yns_2: 0.1493, loss_cls_3: 0.9025, loss_box_3: 1.5450, loss_cns_3: 0.6596, loss_yns_3: 0.1496, loss_cls_4: 0.9053, loss_box_4: 1.5490, loss_cns_4: 0.6623, loss_yns_4: 0.1512, loss_cls_5: 0.9231, loss_box_5: 1.5342, loss_cns_5: 0.6614, loss_yns_5: 0.1497, loss_cls_dn_0: 0.1770, loss_box_dn_0: 0.7393, loss_cls_dn_1: 0.1203, loss_box_dn_1: 0.7161, loss_cls_dn_2: 0.1238, loss_box_dn_2: 0.7119, loss_cls_dn_3: 0.1258, loss_box_dn_3: 0.7066, loss_cls_dn_4: 0.1283, loss_box_dn_4: 0.7119, loss_cls_dn_5: 0.1357, loss_box_dn_5: 0.7196, loss_dense_depth: 0.7636, loss: 25.4603, grad_norm: 44.6177 -2026-01-14 21:37:55,484 - mmdet - INFO - Iter [236/17500] lr: 1.939e-04, eta: 10:25:56, time: 1.641, data_time: 0.091, memory: 49164, loss_cls_0: 0.7628, loss_box_0: 1.6883, loss_cns_0: 0.6259, loss_yns_0: 0.1515, loss_cls_1: 0.9029, loss_box_1: 1.5735, loss_cns_1: 0.6467, loss_yns_1: 0.1485, loss_cls_2: 0.9154, loss_box_2: 1.5446, loss_cns_2: 0.6438, loss_yns_2: 0.1475, loss_cls_3: 0.8881, loss_box_3: 1.5390, loss_cns_3: 0.6545, loss_yns_3: 0.1513, loss_cls_4: 0.8953, loss_box_4: 1.5414, loss_cns_4: 0.6594, loss_yns_4: 0.1525, loss_cls_5: 0.9359, loss_box_5: 1.5012, loss_cns_5: 0.6492, loss_yns_5: 0.1498, loss_cls_dn_0: 0.1719, loss_box_dn_0: 0.7406, loss_cls_dn_1: 0.1235, loss_box_dn_1: 0.6909, loss_cls_dn_2: 0.1257, loss_box_dn_2: 0.6818, loss_cls_dn_3: 0.1267, loss_box_dn_3: 0.6713, loss_cls_dn_4: 0.1281, loss_box_dn_4: 0.6747, loss_cls_dn_5: 0.1437, loss_box_dn_5: 0.6837, loss_dense_depth: 0.7398, loss: 25.1717, grad_norm: 42.4786 -2026-01-14 21:37:57,048 - mmdet - INFO - Iter [237/17500] lr: 1.943e-04, eta: 10:25:12, time: 1.611, data_time: 0.118, memory: 49164, loss_cls_0: 0.7726, loss_box_0: 1.6804, loss_cns_0: 0.6289, loss_yns_0: 0.1506, loss_cls_1: 0.8768, loss_box_1: 1.5504, loss_cns_1: 0.6542, loss_yns_1: 0.1505, loss_cls_2: 0.8756, loss_box_2: 1.5236, loss_cns_2: 0.6548, loss_yns_2: 0.1504, loss_cls_3: 0.8730, loss_box_3: 1.5244, loss_cns_3: 0.6642, loss_yns_3: 0.1513, loss_cls_4: 0.8780, loss_box_4: 1.4970, loss_cns_4: 0.6651, loss_yns_4: 0.1520, loss_cls_5: 0.9166, loss_box_5: 1.4941, loss_cns_5: 0.6590, loss_yns_5: 0.1505, loss_cls_dn_0: 0.1782, loss_box_dn_0: 0.7584, loss_cls_dn_1: 0.1216, loss_box_dn_1: 0.6873, loss_cls_dn_2: 0.1231, loss_box_dn_2: 0.6790, loss_cls_dn_3: 0.1227, loss_box_dn_3: 0.6804, loss_cls_dn_4: 0.1220, loss_box_dn_4: 0.6787, loss_cls_dn_5: 0.1285, loss_box_dn_5: 0.6942, loss_dense_depth: 0.7002, loss: 24.9685, grad_norm: 36.4896 -2026-01-14 21:37:58,610 - mmdet - INFO - Iter [238/17500] lr: 1.947e-04, eta: 10:24:26, time: 1.564, data_time: 0.076, memory: 49164, loss_cls_0: 0.7704, loss_box_0: 1.6434, loss_cns_0: 0.6340, loss_yns_0: 0.1518, loss_cls_1: 0.8761, loss_box_1: 1.5570, loss_cns_1: 0.6609, loss_yns_1: 0.1523, loss_cls_2: 0.8874, loss_box_2: 1.5238, loss_cns_2: 0.6627, loss_yns_2: 0.1519, loss_cls_3: 0.9267, loss_box_3: 1.5240, loss_cns_3: 0.6704, loss_yns_3: 0.1526, loss_cls_4: 0.9096, loss_box_4: 1.5116, loss_cns_4: 0.6655, loss_yns_4: 0.1529, loss_cls_5: 0.8942, loss_box_5: 1.5159, loss_cns_5: 0.6636, loss_yns_5: 0.1527, loss_cls_dn_0: 0.1740, loss_box_dn_0: 0.7476, loss_cls_dn_1: 0.1198, loss_box_dn_1: 0.7019, loss_cls_dn_2: 0.1233, loss_box_dn_2: 0.7010, loss_cls_dn_3: 0.1285, loss_box_dn_3: 0.7120, loss_cls_dn_4: 0.1276, loss_box_dn_4: 0.7162, loss_cls_dn_5: 0.1297, loss_box_dn_5: 0.7305, loss_dense_depth: 0.6667, loss: 25.1903, grad_norm: 39.1444 -2026-01-14 21:38:00,182 - mmdet - INFO - Iter [239/17500] lr: 1.951e-04, eta: 10:23:41, time: 1.573, data_time: 0.083, memory: 49164, loss_cls_0: 0.7636, loss_box_0: 1.5974, loss_cns_0: 0.6280, loss_yns_0: 0.1504, loss_cls_1: 0.8780, loss_box_1: 1.5106, loss_cns_1: 0.6637, loss_yns_1: 0.1520, loss_cls_2: 0.9094, loss_box_2: 1.4838, loss_cns_2: 0.6641, loss_yns_2: 0.1520, loss_cls_3: 0.8931, loss_box_3: 1.4860, loss_cns_3: 0.6680, loss_yns_3: 0.1524, loss_cls_4: 0.8839, loss_box_4: 1.5130, loss_cns_4: 0.6621, loss_yns_4: 0.1510, loss_cls_5: 0.8899, loss_box_5: 1.5208, loss_cns_5: 0.6633, loss_yns_5: 0.1519, loss_cls_dn_0: 0.1735, loss_box_dn_0: 0.7488, loss_cls_dn_1: 0.1232, loss_box_dn_1: 0.7105, loss_cls_dn_2: 0.1244, loss_box_dn_2: 0.7117, loss_cls_dn_3: 0.1261, loss_box_dn_3: 0.7304, loss_cls_dn_4: 0.1288, loss_box_dn_4: 0.7480, loss_cls_dn_5: 0.1343, loss_box_dn_5: 0.7664, loss_dense_depth: 0.7013, loss: 25.1156, grad_norm: 37.9519 -2026-01-14 21:38:01,749 - mmdet - INFO - Iter [240/17500] lr: 1.955e-04, eta: 10:22:55, time: 1.567, data_time: 0.079, memory: 49164, loss_cls_0: 0.7750, loss_box_0: 1.6013, loss_cns_0: 0.6197, loss_yns_0: 0.1482, loss_cls_1: 0.8849, loss_box_1: 1.5019, loss_cns_1: 0.6663, loss_yns_1: 0.1510, loss_cls_2: 0.8935, loss_box_2: 1.4827, loss_cns_2: 0.6679, loss_yns_2: 0.1511, loss_cls_3: 0.9046, loss_box_3: 1.4799, loss_cns_3: 0.6665, loss_yns_3: 0.1518, loss_cls_4: 0.9011, loss_box_4: 1.4940, loss_cns_4: 0.6656, loss_yns_4: 0.1494, loss_cls_5: 0.8959, loss_box_5: 1.5049, loss_cns_5: 0.6645, loss_yns_5: 0.1506, loss_cls_dn_0: 0.1782, loss_box_dn_0: 0.7564, loss_cls_dn_1: 0.1241, loss_box_dn_1: 0.7372, loss_cls_dn_2: 0.1229, loss_box_dn_2: 0.7432, loss_cls_dn_3: 0.1286, loss_box_dn_3: 0.7628, loss_cls_dn_4: 0.1326, loss_box_dn_4: 0.7835, loss_cls_dn_5: 0.1343, loss_box_dn_5: 0.8036, loss_dense_depth: 0.7198, loss: 25.2994, grad_norm: 43.1804 -2026-01-14 21:38:03,418 - mmdet - INFO - Iter [241/17500] lr: 1.959e-04, eta: 10:22:17, time: 1.667, data_time: 0.109, memory: 49164, loss_cls_0: 0.7971, loss_box_0: 1.5879, loss_cns_0: 0.6135, loss_yns_0: 0.1503, loss_cls_1: 0.9021, loss_box_1: 1.5531, loss_cns_1: 0.6631, loss_yns_1: 0.1545, loss_cls_2: 0.9100, loss_box_2: 1.5269, loss_cns_2: 0.6636, loss_yns_2: 0.1530, loss_cls_3: 0.9428, loss_box_3: 1.5222, loss_cns_3: 0.6649, loss_yns_3: 0.1530, loss_cls_4: 0.9238, loss_box_4: 1.5095, loss_cns_4: 0.6631, loss_yns_4: 0.1518, loss_cls_5: 0.9150, loss_box_5: 1.5257, loss_cns_5: 0.6607, loss_yns_5: 0.1519, loss_cls_dn_0: 0.1832, loss_box_dn_0: 0.7471, loss_cls_dn_1: 0.1256, loss_box_dn_1: 0.7557, loss_cls_dn_2: 0.1256, loss_box_dn_2: 0.7570, loss_cls_dn_3: 0.1346, loss_box_dn_3: 0.7651, loss_cls_dn_4: 0.1319, loss_box_dn_4: 0.7740, loss_cls_dn_5: 0.1337, loss_box_dn_5: 0.7899, loss_dense_depth: 0.7470, loss: 25.6303, grad_norm: 38.3599 -2026-01-14 21:38:05,062 - mmdet - INFO - Iter [242/17500] lr: 1.963e-04, eta: 10:21:38, time: 1.646, data_time: 0.171, memory: 49164, loss_cls_0: 0.7737, loss_box_0: 1.6014, loss_cns_0: 0.6236, loss_yns_0: 0.1519, loss_cls_1: 0.8904, loss_box_1: 1.5718, loss_cns_1: 0.6625, loss_yns_1: 0.1544, loss_cls_2: 0.9113, loss_box_2: 1.5486, loss_cns_2: 0.6634, loss_yns_2: 0.1533, loss_cls_3: 0.9370, loss_box_3: 1.5164, loss_cns_3: 0.6692, loss_yns_3: 0.1537, loss_cls_4: 0.9052, loss_box_4: 1.5057, loss_cns_4: 0.6650, loss_yns_4: 0.1529, loss_cls_5: 0.9207, loss_box_5: 1.5210, loss_cns_5: 0.6632, loss_yns_5: 0.1529, loss_cls_dn_0: 0.1733, loss_box_dn_0: 0.7394, loss_cls_dn_1: 0.1258, loss_box_dn_1: 0.7411, loss_cls_dn_2: 0.1257, loss_box_dn_2: 0.7408, loss_cls_dn_3: 0.1322, loss_box_dn_3: 0.7274, loss_cls_dn_4: 0.1283, loss_box_dn_4: 0.7318, loss_cls_dn_5: 0.1342, loss_box_dn_5: 0.7352, loss_dense_depth: 0.6906, loss: 25.3951, grad_norm: 39.8746 -2026-01-14 21:38:06,703 - mmdet - INFO - Iter [243/17500] lr: 1.967e-04, eta: 10:20:56, time: 1.593, data_time: 0.071, memory: 49164, loss_cls_0: 0.7677, loss_box_0: 1.6287, loss_cns_0: 0.6287, loss_yns_0: 0.1540, loss_cls_1: 0.8710, loss_box_1: 1.5841, loss_cns_1: 0.6588, loss_yns_1: 0.1525, loss_cls_2: 0.9029, loss_box_2: 1.5386, loss_cns_2: 0.6648, loss_yns_2: 0.1517, loss_cls_3: 0.9049, loss_box_3: 1.5198, loss_cns_3: 0.6668, loss_yns_3: 0.1530, loss_cls_4: 0.8912, loss_box_4: 1.5404, loss_cns_4: 0.6628, loss_yns_4: 0.1524, loss_cls_5: 0.9057, loss_box_5: 1.5327, loss_cns_5: 0.6625, loss_yns_5: 0.1534, loss_cls_dn_0: 0.1749, loss_box_dn_0: 0.7410, loss_cls_dn_1: 0.1250, loss_box_dn_1: 0.6909, loss_cls_dn_2: 0.1250, loss_box_dn_2: 0.6868, loss_cls_dn_3: 0.1306, loss_box_dn_3: 0.6780, loss_cls_dn_4: 0.1310, loss_box_dn_4: 0.6912, loss_cls_dn_5: 0.1385, loss_box_dn_5: 0.6915, loss_dense_depth: 0.6985, loss: 25.1520, grad_norm: 41.5103 -2026-01-14 21:38:08,302 - mmdet - INFO - Iter [244/17500] lr: 1.971e-04, eta: 10:20:15, time: 1.615, data_time: 0.122, memory: 49164, loss_cls_0: 0.7785, loss_box_0: 1.6584, loss_cns_0: 0.6285, loss_yns_0: 0.1547, loss_cls_1: 0.8627, loss_box_1: 1.5845, loss_cns_1: 0.6575, loss_yns_1: 0.1544, loss_cls_2: 0.8925, loss_box_2: 1.5540, loss_cns_2: 0.6616, loss_yns_2: 0.1546, loss_cls_3: 0.8911, loss_box_3: 1.5622, loss_cns_3: 0.6649, loss_yns_3: 0.1548, loss_cls_4: 0.8889, loss_box_4: 1.5563, loss_cns_4: 0.6615, loss_yns_4: 0.1548, loss_cls_5: 0.8946, loss_box_5: 1.5627, loss_cns_5: 0.6625, loss_yns_5: 0.1567, loss_cls_dn_0: 0.1794, loss_box_dn_0: 0.7458, loss_cls_dn_1: 0.1248, loss_box_dn_1: 0.6896, loss_cls_dn_2: 0.1239, loss_box_dn_2: 0.6812, loss_cls_dn_3: 0.1277, loss_box_dn_3: 0.6908, loss_cls_dn_4: 0.1324, loss_box_dn_4: 0.6996, loss_cls_dn_5: 0.1352, loss_box_dn_5: 0.7161, loss_dense_depth: 0.7722, loss: 25.3716, grad_norm: 34.4308 -2026-01-14 21:38:09,886 - mmdet - INFO - Iter [245/17500] lr: 1.975e-04, eta: 10:19:35, time: 1.616, data_time: 0.095, memory: 49164, loss_cls_0: 0.7480, loss_box_0: 1.6430, loss_cns_0: 0.6256, loss_yns_0: 0.1524, loss_cls_1: 0.8485, loss_box_1: 1.5755, loss_cns_1: 0.6473, loss_yns_1: 0.1523, loss_cls_2: 0.8571, loss_box_2: 1.5338, loss_cns_2: 0.6573, loss_yns_2: 0.1520, loss_cls_3: 0.8760, loss_box_3: 1.5354, loss_cns_3: 0.6562, loss_yns_3: 0.1530, loss_cls_4: 0.8670, loss_box_4: 1.5184, loss_cns_4: 0.6573, loss_yns_4: 0.1527, loss_cls_5: 0.8866, loss_box_5: 1.5257, loss_cns_5: 0.6569, loss_yns_5: 0.1523, loss_cls_dn_0: 0.1743, loss_box_dn_0: 0.7349, loss_cls_dn_1: 0.1280, loss_box_dn_1: 0.6945, loss_cls_dn_2: 0.1290, loss_box_dn_2: 0.6879, loss_cls_dn_3: 0.1354, loss_box_dn_3: 0.7040, loss_cls_dn_4: 0.1343, loss_box_dn_4: 0.7137, loss_cls_dn_5: 0.1395, loss_box_dn_5: 0.7391, loss_dense_depth: 0.7093, loss: 25.0540, grad_norm: 41.0922 -2026-01-14 21:38:11,578 - mmdet - INFO - Iter [246/17500] lr: 1.979e-04, eta: 10:19:00, time: 1.693, data_time: 0.069, memory: 49164, loss_cls_0: 0.7452, loss_box_0: 1.6584, loss_cns_0: 0.6257, loss_yns_0: 0.1522, loss_cls_1: 0.8163, loss_box_1: 1.6070, loss_cns_1: 0.6462, loss_yns_1: 0.1511, loss_cls_2: 0.8374, loss_box_2: 1.5507, loss_cns_2: 0.6566, loss_yns_2: 0.1517, loss_cls_3: 0.8554, loss_box_3: 1.5388, loss_cns_3: 0.6596, loss_yns_3: 0.1523, loss_cls_4: 0.8506, loss_box_4: 1.5187, loss_cns_4: 0.6590, loss_yns_4: 0.1513, loss_cls_5: 0.8925, loss_box_5: 1.4995, loss_cns_5: 0.6528, loss_yns_5: 0.1512, loss_cls_dn_0: 0.1690, loss_box_dn_0: 0.7445, loss_cls_dn_1: 0.1256, loss_box_dn_1: 0.7208, loss_cls_dn_2: 0.1276, loss_box_dn_2: 0.7130, loss_cls_dn_3: 0.1322, loss_box_dn_3: 0.7263, loss_cls_dn_4: 0.1345, loss_box_dn_4: 0.7339, loss_cls_dn_5: 0.1440, loss_box_dn_5: 0.7520, loss_dense_depth: 0.7177, loss: 25.1213, grad_norm: 39.3766 -2026-01-14 21:38:13,148 - mmdet - INFO - Iter [247/17500] lr: 1.983e-04, eta: 10:18:17, time: 1.568, data_time: 0.072, memory: 49164, loss_cls_0: 0.7813, loss_box_0: 1.6742, loss_cns_0: 0.6243, loss_yns_0: 0.1558, loss_cls_1: 0.8256, loss_box_1: 1.5714, loss_cns_1: 0.6444, loss_yns_1: 0.1534, loss_cls_2: 0.8591, loss_box_2: 1.5361, loss_cns_2: 0.6568, loss_yns_2: 0.1544, loss_cls_3: 0.8661, loss_box_3: 1.5118, loss_cns_3: 0.6660, loss_yns_3: 0.1552, loss_cls_4: 0.8695, loss_box_4: 1.5092, loss_cns_4: 0.6646, loss_yns_4: 0.1537, loss_cls_5: 0.8887, loss_box_5: 1.5070, loss_cns_5: 0.6600, loss_yns_5: 0.1542, loss_cls_dn_0: 0.1697, loss_box_dn_0: 0.7529, loss_cls_dn_1: 0.1263, loss_box_dn_1: 0.7285, loss_cls_dn_2: 0.1269, loss_box_dn_2: 0.7222, loss_cls_dn_3: 0.1289, loss_box_dn_3: 0.7284, loss_cls_dn_4: 0.1341, loss_box_dn_4: 0.7400, loss_cls_dn_5: 0.1397, loss_box_dn_5: 0.7536, loss_dense_depth: 0.7336, loss: 25.2275, grad_norm: 31.1068 -2026-01-14 21:38:14,745 - mmdet - INFO - Iter [248/17500] lr: 1.987e-04, eta: 10:17:34, time: 1.561, data_time: 0.076, memory: 49164, loss_cls_0: 0.7668, loss_box_0: 1.6568, loss_cns_0: 0.6262, loss_yns_0: 0.1517, loss_cls_1: 0.8307, loss_box_1: 1.5889, loss_cns_1: 0.6504, loss_yns_1: 0.1512, loss_cls_2: 0.8533, loss_box_2: 1.5416, loss_cns_2: 0.6581, loss_yns_2: 0.1522, loss_cls_3: 0.8566, loss_box_3: 1.5354, loss_cns_3: 0.6652, loss_yns_3: 0.1519, loss_cls_4: 0.8683, loss_box_4: 1.5242, loss_cns_4: 0.6675, loss_yns_4: 0.1520, loss_cls_5: 0.8822, loss_box_5: 1.5388, loss_cns_5: 0.6683, loss_yns_5: 0.1543, loss_cls_dn_0: 0.1710, loss_box_dn_0: 0.7398, loss_cls_dn_1: 0.1258, loss_box_dn_1: 0.7179, loss_cls_dn_2: 0.1256, loss_box_dn_2: 0.7131, loss_cls_dn_3: 0.1274, loss_box_dn_3: 0.7217, loss_cls_dn_4: 0.1300, loss_box_dn_4: 0.7315, loss_cls_dn_5: 0.1345, loss_box_dn_5: 0.7438, loss_dense_depth: 0.7248, loss: 25.1997, grad_norm: 41.8169 -2026-01-14 21:38:16,301 - mmdet - INFO - Iter [249/17500] lr: 1.991e-04, eta: 10:16:54, time: 1.593, data_time: 0.097, memory: 49164, loss_cls_0: 0.7709, loss_box_0: 1.6661, loss_cns_0: 0.6313, loss_yns_0: 0.1548, loss_cls_1: 0.8451, loss_box_1: 1.5513, loss_cns_1: 0.6640, loss_yns_1: 0.1539, loss_cls_2: 0.8548, loss_box_2: 1.5349, loss_cns_2: 0.6659, loss_yns_2: 0.1519, loss_cls_3: 0.8676, loss_box_3: 1.5404, loss_cns_3: 0.6658, loss_yns_3: 0.1529, loss_cls_4: 0.8738, loss_box_4: 1.5296, loss_cns_4: 0.6678, loss_yns_4: 0.1523, loss_cls_5: 0.8722, loss_box_5: 1.5310, loss_cns_5: 0.6675, loss_yns_5: 0.1556, loss_cls_dn_0: 0.1732, loss_box_dn_0: 0.7500, loss_cls_dn_1: 0.1230, loss_box_dn_1: 0.7030, loss_cls_dn_2: 0.1217, loss_box_dn_2: 0.6966, loss_cls_dn_3: 0.1218, loss_box_dn_3: 0.7048, loss_cls_dn_4: 0.1244, loss_box_dn_4: 0.7086, loss_cls_dn_5: 0.1296, loss_box_dn_5: 0.7134, loss_dense_depth: 0.6917, loss: 25.0834, grad_norm: 33.8697 -2026-01-14 21:38:17,860 - mmdet - INFO - Iter [250/17500] lr: 1.995e-04, eta: 10:16:11, time: 1.559, data_time: 0.073, memory: 49164, loss_cls_0: 0.7347, loss_box_0: 1.6651, loss_cns_0: 0.6304, loss_yns_0: 0.1522, loss_cls_1: 0.8317, loss_box_1: 1.5428, loss_cns_1: 0.6629, loss_yns_1: 0.1523, loss_cls_2: 0.8415, loss_box_2: 1.5330, loss_cns_2: 0.6617, loss_yns_2: 0.1521, loss_cls_3: 0.8370, loss_box_3: 1.5101, loss_cns_3: 0.6620, loss_yns_3: 0.1523, loss_cls_4: 0.8380, loss_box_4: 1.5311, loss_cns_4: 0.6665, loss_yns_4: 0.1498, loss_cls_5: 0.8413, loss_box_5: 1.5362, loss_cns_5: 0.6653, loss_yns_5: 0.1522, loss_cls_dn_0: 0.1685, loss_box_dn_0: 0.7398, loss_cls_dn_1: 0.1223, loss_box_dn_1: 0.6872, loss_cls_dn_2: 0.1216, loss_box_dn_2: 0.6900, loss_cls_dn_3: 0.1210, loss_box_dn_3: 0.6905, loss_cls_dn_4: 0.1245, loss_box_dn_4: 0.7043, loss_cls_dn_5: 0.1280, loss_box_dn_5: 0.7174, loss_dense_depth: 0.6876, loss: 24.8050, grad_norm: 40.8395 -2026-01-14 21:38:19,472 - mmdet - INFO - Iter [251/17500] lr: 1.999e-04, eta: 10:15:32, time: 1.612, data_time: 0.072, memory: 49164, loss_cls_0: 0.7549, loss_box_0: 1.6398, loss_cns_0: 0.6304, loss_yns_0: 0.1483, loss_cls_1: 0.8455, loss_box_1: 1.5490, loss_cns_1: 0.6602, loss_yns_1: 0.1498, loss_cls_2: 0.8615, loss_box_2: 1.5167, loss_cns_2: 0.6616, loss_yns_2: 0.1503, loss_cls_3: 0.8614, loss_box_3: 1.5135, loss_cns_3: 0.6621, loss_yns_3: 0.1492, loss_cls_4: 0.8678, loss_box_4: 1.5249, loss_cns_4: 0.6683, loss_yns_4: 0.1489, loss_cls_5: 0.8620, loss_box_5: 1.5324, loss_cns_5: 0.6647, loss_yns_5: 0.1500, loss_cls_dn_0: 0.1698, loss_box_dn_0: 0.7450, loss_cls_dn_1: 0.1185, loss_box_dn_1: 0.7080, loss_cls_dn_2: 0.1203, loss_box_dn_2: 0.7035, loss_cls_dn_3: 0.1220, loss_box_dn_3: 0.7138, loss_cls_dn_4: 0.1328, loss_box_dn_4: 0.7245, loss_cls_dn_5: 0.1317, loss_box_dn_5: 0.7421, loss_dense_depth: 0.7203, loss: 25.0258, grad_norm: 43.2766 -2026-01-14 21:38:21,064 - mmdet - INFO - Iter [252/17500] lr: 2.003e-04, eta: 10:14:53, time: 1.593, data_time: 0.071, memory: 49164, loss_cls_0: 0.7337, loss_box_0: 1.6749, loss_cns_0: 0.6269, loss_yns_0: 0.1502, loss_cls_1: 0.8327, loss_box_1: 1.5415, loss_cns_1: 0.6613, loss_yns_1: 0.1501, loss_cls_2: 0.8396, loss_box_2: 1.5232, loss_cns_2: 0.6636, loss_yns_2: 0.1502, loss_cls_3: 0.8533, loss_box_3: 1.5252, loss_cns_3: 0.6669, loss_yns_3: 0.1501, loss_cls_4: 0.8471, loss_box_4: 1.5148, loss_cns_4: 0.6712, loss_yns_4: 0.1485, loss_cls_5: 0.8471, loss_box_5: 1.5291, loss_cns_5: 0.6662, loss_yns_5: 0.1498, loss_cls_dn_0: 0.1653, loss_box_dn_0: 0.7416, loss_cls_dn_1: 0.1154, loss_box_dn_1: 0.7142, loss_cls_dn_2: 0.1173, loss_box_dn_2: 0.7082, loss_cls_dn_3: 0.1164, loss_box_dn_3: 0.7158, loss_cls_dn_4: 0.1235, loss_box_dn_4: 0.7220, loss_cls_dn_5: 0.1245, loss_box_dn_5: 0.7439, loss_dense_depth: 0.6911, loss: 24.9164, grad_norm: 34.3669 -2026-01-14 21:38:22,642 - mmdet - INFO - Iter [253/17500] lr: 2.007e-04, eta: 10:14:12, time: 1.578, data_time: 0.071, memory: 49164, loss_cls_0: 0.7501, loss_box_0: 1.6731, loss_cns_0: 0.6285, loss_yns_0: 0.1524, loss_cls_1: 0.8467, loss_box_1: 1.5223, loss_cns_1: 0.6637, loss_yns_1: 0.1475, loss_cls_2: 0.8418, loss_box_2: 1.5063, loss_cns_2: 0.6634, loss_yns_2: 0.1471, loss_cls_3: 0.8444, loss_box_3: 1.4862, loss_cns_3: 0.6661, loss_yns_3: 0.1472, loss_cls_4: 0.8449, loss_box_4: 1.4962, loss_cns_4: 0.6662, loss_yns_4: 0.1457, loss_cls_5: 0.8595, loss_box_5: 1.5041, loss_cns_5: 0.6679, loss_yns_5: 0.1472, loss_cls_dn_0: 0.1679, loss_box_dn_0: 0.7470, loss_cls_dn_1: 0.1206, loss_box_dn_1: 0.7241, loss_cls_dn_2: 0.1192, loss_box_dn_2: 0.7284, loss_cls_dn_3: 0.1207, loss_box_dn_3: 0.7276, loss_cls_dn_4: 0.1232, loss_box_dn_4: 0.7415, loss_cls_dn_5: 0.1322, loss_box_dn_5: 0.7597, loss_dense_depth: 0.7671, loss: 24.9978, grad_norm: 45.0728 -2026-01-14 21:38:24,255 - mmdet - INFO - Iter [254/17500] lr: 2.011e-04, eta: 10:13:33, time: 1.585, data_time: 0.071, memory: 49164, loss_cls_0: 0.7445, loss_box_0: 1.6799, loss_cns_0: 0.6268, loss_yns_0: 0.1506, loss_cls_1: 0.8434, loss_box_1: 1.5776, loss_cns_1: 0.6642, loss_yns_1: 0.1492, loss_cls_2: 0.8442, loss_box_2: 1.5404, loss_cns_2: 0.6621, loss_yns_2: 0.1506, loss_cls_3: 0.8448, loss_box_3: 1.5176, loss_cns_3: 0.6632, loss_yns_3: 0.1483, loss_cls_4: 0.8480, loss_box_4: 1.5316, loss_cns_4: 0.6641, loss_yns_4: 0.1491, loss_cls_5: 0.8521, loss_box_5: 1.5460, loss_cns_5: 0.6648, loss_yns_5: 0.1511, loss_cls_dn_0: 0.1658, loss_box_dn_0: 0.7452, loss_cls_dn_1: 0.1250, loss_box_dn_1: 0.7320, loss_cls_dn_2: 0.1227, loss_box_dn_2: 0.7302, loss_cls_dn_3: 0.1236, loss_box_dn_3: 0.7215, loss_cls_dn_4: 0.1269, loss_box_dn_4: 0.7261, loss_cls_dn_5: 0.1344, loss_box_dn_5: 0.7425, loss_dense_depth: 0.7235, loss: 25.1336, grad_norm: 43.3312 -2026-01-14 21:38:25,824 - mmdet - INFO - Iter [255/17500] lr: 2.015e-04, eta: 10:12:54, time: 1.597, data_time: 0.087, memory: 49164, loss_cls_0: 0.7358, loss_box_0: 1.6586, loss_cns_0: 0.6304, loss_yns_0: 0.1506, loss_cls_1: 0.8288, loss_box_1: 1.5487, loss_cns_1: 0.6666, loss_yns_1: 0.1510, loss_cls_2: 0.8343, loss_box_2: 1.5279, loss_cns_2: 0.6641, loss_yns_2: 0.1511, loss_cls_3: 0.8385, loss_box_3: 1.5102, loss_cns_3: 0.6674, loss_yns_3: 0.1503, loss_cls_4: 0.8400, loss_box_4: 1.5064, loss_cns_4: 0.6697, loss_yns_4: 0.1495, loss_cls_5: 0.8422, loss_box_5: 1.5243, loss_cns_5: 0.6657, loss_yns_5: 0.1502, loss_cls_dn_0: 0.1641, loss_box_dn_0: 0.7404, loss_cls_dn_1: 0.1233, loss_box_dn_1: 0.7171, loss_cls_dn_2: 0.1224, loss_box_dn_2: 0.7141, loss_cls_dn_3: 0.1225, loss_box_dn_3: 0.7052, loss_cls_dn_4: 0.1289, loss_box_dn_4: 0.7005, loss_cls_dn_5: 0.1265, loss_box_dn_5: 0.7132, loss_dense_depth: 0.7261, loss: 24.8666, grad_norm: 38.6993 -2026-01-14 21:38:27,394 - mmdet - INFO - Iter [256/17500] lr: 2.019e-04, eta: 10:12:14, time: 1.567, data_time: 0.074, memory: 49164, loss_cls_0: 0.7454, loss_box_0: 1.6602, loss_cns_0: 0.6215, loss_yns_0: 0.1494, loss_cls_1: 0.8300, loss_box_1: 1.5258, loss_cns_1: 0.6657, loss_yns_1: 0.1480, loss_cls_2: 0.8410, loss_box_2: 1.4965, loss_cns_2: 0.6680, loss_yns_2: 0.1501, loss_cls_3: 0.8436, loss_box_3: 1.5117, loss_cns_3: 0.6672, loss_yns_3: 0.1502, loss_cls_4: 0.8539, loss_box_4: 1.4736, loss_cns_4: 0.6684, loss_yns_4: 0.1489, loss_cls_5: 0.8530, loss_box_5: 1.4832, loss_cns_5: 0.6654, loss_yns_5: 0.1503, loss_cls_dn_0: 0.1607, loss_box_dn_0: 0.7361, loss_cls_dn_1: 0.1140, loss_box_dn_1: 0.6801, loss_cls_dn_2: 0.1147, loss_box_dn_2: 0.6734, loss_cls_dn_3: 0.1191, loss_box_dn_3: 0.6831, loss_cls_dn_4: 0.1252, loss_box_dn_4: 0.6809, loss_cls_dn_5: 0.1193, loss_box_dn_5: 0.6879, loss_dense_depth: 0.7267, loss: 24.5926, grad_norm: 44.3602 -2026-01-14 21:38:28,973 - mmdet - INFO - Iter [257/17500] lr: 2.023e-04, eta: 10:11:35, time: 1.581, data_time: 0.077, memory: 49164, loss_cls_0: 0.7637, loss_box_0: 1.6807, loss_cns_0: 0.6243, loss_yns_0: 0.1512, loss_cls_1: 0.8378, loss_box_1: 1.5667, loss_cns_1: 0.6607, loss_yns_1: 0.1503, loss_cls_2: 0.8562, loss_box_2: 1.5468, loss_cns_2: 0.6626, loss_yns_2: 0.1523, loss_cls_3: 0.8503, loss_box_3: 1.5420, loss_cns_3: 0.6660, loss_yns_3: 0.1491, loss_cls_4: 0.8527, loss_box_4: 1.5197, loss_cns_4: 0.6653, loss_yns_4: 0.1489, loss_cls_5: 0.8571, loss_box_5: 1.5291, loss_cns_5: 0.6637, loss_yns_5: 0.1479, loss_cls_dn_0: 0.1609, loss_box_dn_0: 0.7388, loss_cls_dn_1: 0.1140, loss_box_dn_1: 0.6736, loss_cls_dn_2: 0.1165, loss_box_dn_2: 0.6727, loss_cls_dn_3: 0.1212, loss_box_dn_3: 0.6821, loss_cls_dn_4: 0.1218, loss_box_dn_4: 0.6813, loss_cls_dn_5: 0.1231, loss_box_dn_5: 0.6993, loss_dense_depth: 0.7561, loss: 24.9065, grad_norm: 34.5981 -2026-01-14 21:38:30,563 - mmdet - INFO - Iter [258/17500] lr: 2.027e-04, eta: 10:10:57, time: 1.591, data_time: 0.073, memory: 49164, loss_cls_0: 0.7522, loss_box_0: 1.7020, loss_cns_0: 0.6163, loss_yns_0: 0.1484, loss_cls_1: 0.8458, loss_box_1: 1.5883, loss_cns_1: 0.6561, loss_yns_1: 0.1483, loss_cls_2: 0.8537, loss_box_2: 1.5539, loss_cns_2: 0.6643, loss_yns_2: 0.1488, loss_cls_3: 0.8570, loss_box_3: 1.5322, loss_cns_3: 0.6662, loss_yns_3: 0.1466, loss_cls_4: 0.8529, loss_box_4: 1.5208, loss_cns_4: 0.6638, loss_yns_4: 0.1466, loss_cls_5: 0.8561, loss_box_5: 1.5132, loss_cns_5: 0.6640, loss_yns_5: 0.1483, loss_cls_dn_0: 0.1602, loss_box_dn_0: 0.7357, loss_cls_dn_1: 0.1176, loss_box_dn_1: 0.6691, loss_cls_dn_2: 0.1176, loss_box_dn_2: 0.6661, loss_cls_dn_3: 0.1170, loss_box_dn_3: 0.6699, loss_cls_dn_4: 0.1182, loss_box_dn_4: 0.6731, loss_cls_dn_5: 0.1217, loss_box_dn_5: 0.6866, loss_dense_depth: 0.7633, loss: 24.8617, grad_norm: 36.2046 -2026-01-14 21:38:32,139 - mmdet - INFO - Iter [259/17500] lr: 2.031e-04, eta: 10:10:18, time: 1.576, data_time: 0.085, memory: 49164, loss_cls_0: 0.7943, loss_box_0: 1.7042, loss_cns_0: 0.6220, loss_yns_0: 0.1472, loss_cls_1: 0.8595, loss_box_1: 1.6480, loss_cns_1: 0.6532, loss_yns_1: 0.1450, loss_cls_2: 0.8829, loss_box_2: 1.6025, loss_cns_2: 0.6550, loss_yns_2: 0.1447, loss_cls_3: 0.8847, loss_box_3: 1.6160, loss_cns_3: 0.6627, loss_yns_3: 0.1452, loss_cls_4: 0.8892, loss_box_4: 1.6147, loss_cns_4: 0.6675, loss_yns_4: 0.1459, loss_cls_5: 0.8902, loss_box_5: 1.5863, loss_cns_5: 0.6550, loss_yns_5: 0.1461, loss_cls_dn_0: 0.1658, loss_box_dn_0: 0.7416, loss_cls_dn_1: 0.1179, loss_box_dn_1: 0.6760, loss_cls_dn_2: 0.1177, loss_box_dn_2: 0.6694, loss_cls_dn_3: 0.1196, loss_box_dn_3: 0.6793, loss_cls_dn_4: 0.1226, loss_box_dn_4: 0.6865, loss_cls_dn_5: 0.1249, loss_box_dn_5: 0.6945, loss_dense_depth: 0.8234, loss: 25.5008, grad_norm: 41.0489 -2026-01-14 21:38:40,686 - mmdet - INFO - Iter [260/17500] lr: 2.035e-04, eta: 10:17:22, time: 8.547, data_time: 0.082, memory: 49164, loss_cls_0: 0.7829, loss_box_0: 1.6791, loss_cns_0: 0.6296, loss_yns_0: 0.1506, loss_cls_1: 0.8367, loss_box_1: 1.6190, loss_cns_1: 0.6546, loss_yns_1: 0.1455, loss_cls_2: 0.8633, loss_box_2: 1.5777, loss_cns_2: 0.6569, loss_yns_2: 0.1466, loss_cls_3: 0.8751, loss_box_3: 1.5671, loss_cns_3: 0.6629, loss_yns_3: 0.1449, loss_cls_4: 0.8739, loss_box_4: 1.5664, loss_cns_4: 0.6668, loss_yns_4: 0.1456, loss_cls_5: 0.8727, loss_box_5: 1.5634, loss_cns_5: 0.6601, loss_yns_5: 0.1459, loss_cls_dn_0: 0.1591, loss_box_dn_0: 0.7411, loss_cls_dn_1: 0.1198, loss_box_dn_1: 0.6867, loss_cls_dn_2: 0.1204, loss_box_dn_2: 0.6786, loss_cls_dn_3: 0.1216, loss_box_dn_3: 0.6777, loss_cls_dn_4: 0.1230, loss_box_dn_4: 0.6810, loss_cls_dn_5: 0.1269, loss_box_dn_5: 0.6905, loss_dense_depth: 0.7526, loss: 25.1663, grad_norm: 26.7044 -2026-01-14 21:38:42,310 - mmdet - INFO - Iter [261/17500] lr: 2.039e-04, eta: 10:16:45, time: 1.624, data_time: 0.100, memory: 49164, loss_cls_0: 0.7809, loss_box_0: 1.6880, loss_cns_0: 0.6219, loss_yns_0: 0.1484, loss_cls_1: 0.8396, loss_box_1: 1.5880, loss_cns_1: 0.6530, loss_yns_1: 0.1455, loss_cls_2: 0.8657, loss_box_2: 1.5377, loss_cns_2: 0.6567, loss_yns_2: 0.1442, loss_cls_3: 0.8635, loss_box_3: 1.5364, loss_cns_3: 0.6568, loss_yns_3: 0.1435, loss_cls_4: 0.8856, loss_box_4: 1.5336, loss_cns_4: 0.6586, loss_yns_4: 0.1440, loss_cls_5: 0.8875, loss_box_5: 1.5302, loss_cns_5: 0.6580, loss_yns_5: 0.1423, loss_cls_dn_0: 0.1612, loss_box_dn_0: 0.7420, loss_cls_dn_1: 0.1221, loss_box_dn_1: 0.6867, loss_cls_dn_2: 0.1226, loss_box_dn_2: 0.6770, loss_cls_dn_3: 0.1234, loss_box_dn_3: 0.6763, loss_cls_dn_4: 0.1310, loss_box_dn_4: 0.6786, loss_cls_dn_5: 0.1345, loss_box_dn_5: 0.6811, loss_dense_depth: 0.9340, loss: 25.1801, grad_norm: 38.7666 -2026-01-14 21:38:43,934 - mmdet - INFO - Iter [262/17500] lr: 2.043e-04, eta: 10:16:09, time: 1.624, data_time: 0.160, memory: 49164, loss_cls_0: 0.7878, loss_box_0: 1.7108, loss_cns_0: 0.6188, loss_yns_0: 0.1457, loss_cls_1: 0.8447, loss_box_1: 1.5613, loss_cns_1: 0.6579, loss_yns_1: 0.1440, loss_cls_2: 0.8641, loss_box_2: 1.5315, loss_cns_2: 0.6572, loss_yns_2: 0.1433, loss_cls_3: 0.8702, loss_box_3: 1.5165, loss_cns_3: 0.6648, loss_yns_3: 0.1446, loss_cls_4: 0.8827, loss_box_4: 1.5138, loss_cns_4: 0.6654, loss_yns_4: 0.1439, loss_cls_5: 0.8963, loss_box_5: 1.4955, loss_cns_5: 0.6617, loss_yns_5: 0.1440, loss_cls_dn_0: 0.1624, loss_box_dn_0: 0.7451, loss_cls_dn_1: 0.1183, loss_box_dn_1: 0.6747, loss_cls_dn_2: 0.1188, loss_box_dn_2: 0.6644, loss_cls_dn_3: 0.1199, loss_box_dn_3: 0.6608, loss_cls_dn_4: 0.1269, loss_box_dn_4: 0.6601, loss_cls_dn_5: 0.1288, loss_box_dn_5: 0.6581, loss_dense_depth: 0.9096, loss: 25.0144, grad_norm: 28.3236 -2026-01-14 21:38:45,511 - mmdet - INFO - Iter [263/17500] lr: 2.047e-04, eta: 10:15:28, time: 1.563, data_time: 0.070, memory: 49164, loss_cls_0: 0.7742, loss_box_0: 1.6847, loss_cns_0: 0.6259, loss_yns_0: 0.1431, loss_cls_1: 0.8337, loss_box_1: 1.5848, loss_cns_1: 0.6553, loss_yns_1: 0.1406, loss_cls_2: 0.8537, loss_box_2: 1.5421, loss_cns_2: 0.6579, loss_yns_2: 0.1418, loss_cls_3: 0.8613, loss_box_3: 1.5363, loss_cns_3: 0.6604, loss_yns_3: 0.1395, loss_cls_4: 0.8592, loss_box_4: 1.5263, loss_cns_4: 0.6626, loss_yns_4: 0.1394, loss_cls_5: 0.8865, loss_box_5: 1.5279, loss_cns_5: 0.6584, loss_yns_5: 0.1413, loss_cls_dn_0: 0.1623, loss_box_dn_0: 0.7454, loss_cls_dn_1: 0.1131, loss_box_dn_1: 0.6687, loss_cls_dn_2: 0.1147, loss_box_dn_2: 0.6557, loss_cls_dn_3: 0.1165, loss_box_dn_3: 0.6566, loss_cls_dn_4: 0.1180, loss_box_dn_4: 0.6562, loss_cls_dn_5: 0.1221, loss_box_dn_5: 0.6633, loss_dense_depth: 0.8294, loss: 24.8588, grad_norm: 20.2960 -2026-01-14 21:38:47,101 - mmdet - INFO - Iter [264/17500] lr: 2.051e-04, eta: 10:14:51, time: 1.603, data_time: 0.078, memory: 49164, loss_cls_0: 0.7633, loss_box_0: 1.7111, loss_cns_0: 0.6263, loss_yns_0: 0.1445, loss_cls_1: 0.8543, loss_box_1: 1.5997, loss_cns_1: 0.6605, loss_yns_1: 0.1417, loss_cls_2: 0.8714, loss_box_2: 1.5504, loss_cns_2: 0.6636, loss_yns_2: 0.1419, loss_cls_3: 0.8620, loss_box_3: 1.5487, loss_cns_3: 0.6622, loss_yns_3: 0.1401, loss_cls_4: 0.8756, loss_box_4: 1.5362, loss_cns_4: 0.6644, loss_yns_4: 0.1408, loss_cls_5: 0.8743, loss_box_5: 1.5342, loss_cns_5: 0.6631, loss_yns_5: 0.1413, loss_cls_dn_0: 0.1569, loss_box_dn_0: 0.7422, loss_cls_dn_1: 0.1157, loss_box_dn_1: 0.6860, loss_cls_dn_2: 0.1188, loss_box_dn_2: 0.6704, loss_cls_dn_3: 0.1164, loss_box_dn_3: 0.6794, loss_cls_dn_4: 0.1217, loss_box_dn_4: 0.6849, loss_cls_dn_5: 0.1238, loss_box_dn_5: 0.6945, loss_dense_depth: 0.9977, loss: 25.2799, grad_norm: 38.2798 -2026-01-14 21:38:48,713 - mmdet - INFO - Iter [265/17500] lr: 2.055e-04, eta: 10:14:15, time: 1.611, data_time: 0.075, memory: 49164, loss_cls_0: 0.7869, loss_box_0: 1.7134, loss_cns_0: 0.6305, loss_yns_0: 0.1456, loss_cls_1: 0.8616, loss_box_1: 1.5850, loss_cns_1: 0.6602, loss_yns_1: 0.1437, loss_cls_2: 0.8669, loss_box_2: 1.5736, loss_cns_2: 0.6633, loss_yns_2: 0.1411, loss_cls_3: 0.8714, loss_box_3: 1.5563, loss_cns_3: 0.6608, loss_yns_3: 0.1414, loss_cls_4: 0.8764, loss_box_4: 1.5595, loss_cns_4: 0.6618, loss_yns_4: 0.1426, loss_cls_5: 0.8821, loss_box_5: 1.5548, loss_cns_5: 0.6608, loss_yns_5: 0.1424, loss_cls_dn_0: 0.1677, loss_box_dn_0: 0.7391, loss_cls_dn_1: 0.1158, loss_box_dn_1: 0.6969, loss_cls_dn_2: 0.1160, loss_box_dn_2: 0.6927, loss_cls_dn_3: 0.1171, loss_box_dn_3: 0.7004, loss_cls_dn_4: 0.1228, loss_box_dn_4: 0.7136, loss_cls_dn_5: 0.1222, loss_box_dn_5: 0.7255, loss_dense_depth: 0.8190, loss: 25.3312, grad_norm: 33.5996 -2026-01-14 21:38:50,336 - mmdet - INFO - Iter [266/17500] lr: 2.059e-04, eta: 10:13:37, time: 1.594, data_time: 0.073, memory: 49164, loss_cls_0: 0.7685, loss_box_0: 1.6657, loss_cns_0: 0.6309, loss_yns_0: 0.1441, loss_cls_1: 0.8349, loss_box_1: 1.5526, loss_cns_1: 0.6606, loss_yns_1: 0.1405, loss_cls_2: 0.8396, loss_box_2: 1.5346, loss_cns_2: 0.6625, loss_yns_2: 0.1408, loss_cls_3: 0.8376, loss_box_3: 1.5238, loss_cns_3: 0.6651, loss_yns_3: 0.1421, loss_cls_4: 0.8416, loss_box_4: 1.5226, loss_cns_4: 0.6646, loss_yns_4: 0.1430, loss_cls_5: 0.8559, loss_box_5: 1.5129, loss_cns_5: 0.6668, loss_yns_5: 0.1428, loss_cls_dn_0: 0.1698, loss_box_dn_0: 0.7427, loss_cls_dn_1: 0.1169, loss_box_dn_1: 0.7185, loss_cls_dn_2: 0.1166, loss_box_dn_2: 0.7172, loss_cls_dn_3: 0.1191, loss_box_dn_3: 0.7187, loss_cls_dn_4: 0.1220, loss_box_dn_4: 0.7298, loss_cls_dn_5: 0.1224, loss_box_dn_5: 0.7370, loss_dense_depth: 0.9147, loss: 25.1396, grad_norm: 33.5614 -2026-01-14 21:38:51,927 - mmdet - INFO - Iter [267/17500] lr: 2.063e-04, eta: 10:13:02, time: 1.620, data_time: 0.105, memory: 49164, loss_cls_0: 0.7462, loss_box_0: 1.6544, loss_cns_0: 0.6269, loss_yns_0: 0.1416, loss_cls_1: 0.8267, loss_box_1: 1.5432, loss_cns_1: 0.6596, loss_yns_1: 0.1386, loss_cls_2: 0.8387, loss_box_2: 1.5065, loss_cns_2: 0.6596, loss_yns_2: 0.1396, loss_cls_3: 0.8356, loss_box_3: 1.5087, loss_cns_3: 0.6641, loss_yns_3: 0.1434, loss_cls_4: 0.8437, loss_box_4: 1.5081, loss_cns_4: 0.6642, loss_yns_4: 0.1445, loss_cls_5: 0.8592, loss_box_5: 1.5031, loss_cns_5: 0.6655, loss_yns_5: 0.1440, loss_cls_dn_0: 0.1612, loss_box_dn_0: 0.7392, loss_cls_dn_1: 0.1140, loss_box_dn_1: 0.7162, loss_cls_dn_2: 0.1149, loss_box_dn_2: 0.7068, loss_cls_dn_3: 0.1172, loss_box_dn_3: 0.7029, loss_cls_dn_4: 0.1169, loss_box_dn_4: 0.7051, loss_cls_dn_5: 0.1184, loss_box_dn_5: 0.7101, loss_dense_depth: 0.8442, loss: 24.8329, grad_norm: 40.4333 -2026-01-14 21:38:53,511 - mmdet - INFO - Iter [268/17500] lr: 2.067e-04, eta: 10:12:24, time: 1.586, data_time: 0.077, memory: 49164, loss_cls_0: 0.7498, loss_box_0: 1.6359, loss_cns_0: 0.6351, loss_yns_0: 0.1453, loss_cls_1: 0.8223, loss_box_1: 1.5046, loss_cns_1: 0.6674, loss_yns_1: 0.1450, loss_cls_2: 0.8350, loss_box_2: 1.4630, loss_cns_2: 0.6738, loss_yns_2: 0.1443, loss_cls_3: 0.8411, loss_box_3: 1.4649, loss_cns_3: 0.6682, loss_yns_3: 0.1453, loss_cls_4: 0.8393, loss_box_4: 1.4582, loss_cns_4: 0.6679, loss_yns_4: 0.1454, loss_cls_5: 0.8484, loss_box_5: 1.4496, loss_cns_5: 0.6692, loss_yns_5: 0.1458, loss_cls_dn_0: 0.1601, loss_box_dn_0: 0.7356, loss_cls_dn_1: 0.1190, loss_box_dn_1: 0.6972, loss_cls_dn_2: 0.1196, loss_box_dn_2: 0.6852, loss_cls_dn_3: 0.1208, loss_box_dn_3: 0.6818, loss_cls_dn_4: 0.1203, loss_box_dn_4: 0.6792, loss_cls_dn_5: 0.1210, loss_box_dn_5: 0.6783, loss_dense_depth: 0.8334, loss: 24.5162, grad_norm: 25.8280 -2026-01-14 21:38:55,137 - mmdet - INFO - Iter [269/17500] lr: 2.071e-04, eta: 10:11:49, time: 1.623, data_time: 0.074, memory: 49164, loss_cls_0: 0.7578, loss_box_0: 1.6365, loss_cns_0: 0.6316, loss_yns_0: 0.1451, loss_cls_1: 0.8261, loss_box_1: 1.5180, loss_cns_1: 0.6610, loss_yns_1: 0.1437, loss_cls_2: 0.8398, loss_box_2: 1.4895, loss_cns_2: 0.6626, loss_yns_2: 0.1447, loss_cls_3: 0.8441, loss_box_3: 1.5049, loss_cns_3: 0.6596, loss_yns_3: 0.1459, loss_cls_4: 0.8611, loss_box_4: 1.4878, loss_cns_4: 0.6586, loss_yns_4: 0.1455, loss_cls_5: 0.8488, loss_box_5: 1.4974, loss_cns_5: 0.6608, loss_yns_5: 0.1449, loss_cls_dn_0: 0.1589, loss_box_dn_0: 0.7406, loss_cls_dn_1: 0.1141, loss_box_dn_1: 0.6868, loss_cls_dn_2: 0.1129, loss_box_dn_2: 0.6777, loss_cls_dn_3: 0.1133, loss_box_dn_3: 0.6790, loss_cls_dn_4: 0.1214, loss_box_dn_4: 0.6771, loss_cls_dn_5: 0.1196, loss_box_dn_5: 0.6783, loss_dense_depth: 0.8659, loss: 24.6615, grad_norm: 52.5067 -2026-01-14 21:38:56,750 - mmdet - INFO - Iter [270/17500] lr: 2.075e-04, eta: 10:11:15, time: 1.617, data_time: 0.080, memory: 49164, loss_cls_0: 0.7887, loss_box_0: 1.6973, loss_cns_0: 0.6212, loss_yns_0: 0.1476, loss_cls_1: 0.8466, loss_box_1: 1.5540, loss_cns_1: 0.6589, loss_yns_1: 0.1458, loss_cls_2: 0.8708, loss_box_2: 1.5164, loss_cns_2: 0.6597, loss_yns_2: 0.1463, loss_cls_3: 0.8807, loss_box_3: 1.5032, loss_cns_3: 0.6571, loss_yns_3: 0.1470, loss_cls_4: 0.8842, loss_box_4: 1.4886, loss_cns_4: 0.6536, loss_yns_4: 0.1434, loss_cls_5: 0.8802, loss_box_5: 1.5213, loss_cns_5: 0.6600, loss_yns_5: 0.1447, loss_cls_dn_0: 0.1668, loss_box_dn_0: 0.7531, loss_cls_dn_1: 0.1120, loss_box_dn_1: 0.6822, loss_cls_dn_2: 0.1144, loss_box_dn_2: 0.6718, loss_cls_dn_3: 0.1200, loss_box_dn_3: 0.6762, loss_cls_dn_4: 0.1236, loss_box_dn_4: 0.6870, loss_cls_dn_5: 0.1198, loss_box_dn_5: 0.7000, loss_dense_depth: 0.9960, loss: 25.1404, grad_norm: 36.1958 -2026-01-14 21:38:58,426 - mmdet - INFO - Iter [271/17500] lr: 2.079e-04, eta: 10:10:40, time: 1.625, data_time: 0.077, memory: 49164, loss_cls_0: 0.7278, loss_box_0: 1.6707, loss_cns_0: 0.6293, loss_yns_0: 0.1476, loss_cls_1: 0.8237, loss_box_1: 1.5465, loss_cns_1: 0.6591, loss_yns_1: 0.1450, loss_cls_2: 0.8543, loss_box_2: 1.5359, loss_cns_2: 0.6635, loss_yns_2: 0.1452, loss_cls_3: 0.8640, loss_box_3: 1.5051, loss_cns_3: 0.6613, loss_yns_3: 0.1452, loss_cls_4: 0.8699, loss_box_4: 1.5006, loss_cns_4: 0.6621, loss_yns_4: 0.1434, loss_cls_5: 0.8794, loss_box_5: 1.5002, loss_cns_5: 0.6625, loss_yns_5: 0.1448, loss_cls_dn_0: 0.1568, loss_box_dn_0: 0.7510, loss_cls_dn_1: 0.1108, loss_box_dn_1: 0.6943, loss_cls_dn_2: 0.1121, loss_box_dn_2: 0.7021, loss_cls_dn_3: 0.1174, loss_box_dn_3: 0.7002, loss_cls_dn_4: 0.1162, loss_box_dn_4: 0.7116, loss_cls_dn_5: 0.1163, loss_box_dn_5: 0.7287, loss_dense_depth: 0.8177, loss: 24.9224, grad_norm: 59.6384 -2026-01-14 21:39:00,005 - mmdet - INFO - Iter [272/17500] lr: 2.083e-04, eta: 10:10:07, time: 1.627, data_time: 0.109, memory: 49164, loss_cls_0: 0.7532, loss_box_0: 1.6248, loss_cns_0: 0.6279, loss_yns_0: 0.1461, loss_cls_1: 0.8410, loss_box_1: 1.5431, loss_cns_1: 0.6618, loss_yns_1: 0.1458, loss_cls_2: 0.8550, loss_box_2: 1.5247, loss_cns_2: 0.6652, loss_yns_2: 0.1454, loss_cls_3: 0.8642, loss_box_3: 1.4928, loss_cns_3: 0.6633, loss_yns_3: 0.1452, loss_cls_4: 0.8656, loss_box_4: 1.4867, loss_cns_4: 0.6653, loss_yns_4: 0.1460, loss_cls_5: 0.8587, loss_box_5: 1.4848, loss_cns_5: 0.6640, loss_yns_5: 0.1438, loss_cls_dn_0: 0.1580, loss_box_dn_0: 0.7331, loss_cls_dn_1: 0.1101, loss_box_dn_1: 0.7132, loss_cls_dn_2: 0.1109, loss_box_dn_2: 0.7203, loss_cls_dn_3: 0.1149, loss_box_dn_3: 0.7177, loss_cls_dn_4: 0.1161, loss_box_dn_4: 0.7295, loss_cls_dn_5: 0.1164, loss_box_dn_5: 0.7466, loss_dense_depth: 0.8516, loss: 24.9528, grad_norm: 46.1294 -2026-01-14 21:39:03,061 - mmdet - INFO - Iter [273/17500] lr: 2.087e-04, eta: 10:09:29, time: 1.560, data_time: 0.075, memory: 49164, loss_cls_0: 0.7704, loss_box_0: 1.6521, loss_cns_0: 0.6112, loss_yns_0: 0.1422, loss_cls_1: 0.8275, loss_box_1: 1.5394, loss_cns_1: 0.6584, loss_yns_1: 0.1428, loss_cls_2: 0.8356, loss_box_2: 1.5263, loss_cns_2: 0.6628, loss_yns_2: 0.1439, loss_cls_3: 0.8388, loss_box_3: 1.5117, loss_cns_3: 0.6577, loss_yns_3: 0.1424, loss_cls_4: 0.8467, loss_box_4: 1.5225, loss_cns_4: 0.6558, loss_yns_4: 0.1427, loss_cls_5: 0.8806, loss_box_5: 1.4976, loss_cns_5: 0.6518, loss_yns_5: 0.1405, loss_cls_dn_0: 0.1623, loss_box_dn_0: 0.7780, loss_cls_dn_1: 0.1136, loss_box_dn_1: 0.7188, loss_cls_dn_2: 0.1117, loss_box_dn_2: 0.7141, loss_cls_dn_3: 0.1130, loss_box_dn_3: 0.7151, loss_cls_dn_4: 0.1152, loss_box_dn_4: 0.7295, loss_cls_dn_5: 0.1167, loss_box_dn_5: 0.7342, loss_dense_depth: 0.7902, loss: 24.9137, grad_norm: 56.3077 -2026-01-14 21:39:04,640 - mmdet - INFO - Iter [274/17500] lr: 2.091e-04, eta: 10:10:27, time: 3.077, data_time: 1.552, memory: 49164, loss_cls_0: 0.7620, loss_box_0: 1.6380, loss_cns_0: 0.6226, loss_yns_0: 0.1434, loss_cls_1: 0.8124, loss_box_1: 1.5005, loss_cns_1: 0.6577, loss_yns_1: 0.1413, loss_cls_2: 0.8173, loss_box_2: 1.4607, loss_cns_2: 0.6596, loss_yns_2: 0.1413, loss_cls_3: 0.8237, loss_box_3: 1.4398, loss_cns_3: 0.6583, loss_yns_3: 0.1412, loss_cls_4: 0.8392, loss_box_4: 1.4402, loss_cns_4: 0.6588, loss_yns_4: 0.1407, loss_cls_5: 0.8440, loss_box_5: 1.4231, loss_cns_5: 0.6608, loss_yns_5: 0.1405, loss_cls_dn_0: 0.1617, loss_box_dn_0: 0.7744, loss_cls_dn_1: 0.1078, loss_box_dn_1: 0.6804, loss_cls_dn_2: 0.1057, loss_box_dn_2: 0.6612, loss_cls_dn_3: 0.1063, loss_box_dn_3: 0.6522, loss_cls_dn_4: 0.1143, loss_box_dn_4: 0.6564, loss_cls_dn_5: 0.1143, loss_box_dn_5: 0.6565, loss_dense_depth: 0.8716, loss: 24.2301, grad_norm: 37.7547 -2026-01-14 21:39:06,224 - mmdet - INFO - Iter [275/17500] lr: 2.095e-04, eta: 10:09:51, time: 1.584, data_time: 0.071, memory: 49164, loss_cls_0: 0.7385, loss_box_0: 1.6527, loss_cns_0: 0.6281, loss_yns_0: 0.1435, loss_cls_1: 0.7911, loss_box_1: 1.5221, loss_cns_1: 0.6555, loss_yns_1: 0.1401, loss_cls_2: 0.8075, loss_box_2: 1.5190, loss_cns_2: 0.6545, loss_yns_2: 0.1411, loss_cls_3: 0.8299, loss_box_3: 1.5109, loss_cns_3: 0.6564, loss_yns_3: 0.1405, loss_cls_4: 0.8567, loss_box_4: 1.5014, loss_cns_4: 0.6559, loss_yns_4: 0.1425, loss_cls_5: 0.8474, loss_box_5: 1.5141, loss_cns_5: 0.6545, loss_yns_5: 0.1401, loss_cls_dn_0: 0.1584, loss_box_dn_0: 0.7462, loss_cls_dn_1: 0.1049, loss_box_dn_1: 0.6548, loss_cls_dn_2: 0.1062, loss_box_dn_2: 0.6524, loss_cls_dn_3: 0.1109, loss_box_dn_3: 0.6494, loss_cls_dn_4: 0.1182, loss_box_dn_4: 0.6470, loss_cls_dn_5: 0.1173, loss_box_dn_5: 0.6594, loss_dense_depth: 0.8467, loss: 24.4159, grad_norm: 61.4061 -2026-01-14 21:39:07,774 - mmdet - INFO - Iter [276/17500] lr: 2.099e-04, eta: 10:09:13, time: 1.550, data_time: 0.071, memory: 49164, loss_cls_0: 0.7508, loss_box_0: 1.6683, loss_cns_0: 0.6267, loss_yns_0: 0.1429, loss_cls_1: 0.8127, loss_box_1: 1.5380, loss_cns_1: 0.6538, loss_yns_1: 0.1416, loss_cls_2: 0.8224, loss_box_2: 1.5073, loss_cns_2: 0.6571, loss_yns_2: 0.1416, loss_cls_3: 0.8453, loss_box_3: 1.5107, loss_cns_3: 0.6561, loss_yns_3: 0.1398, loss_cls_4: 0.8600, loss_box_4: 1.4922, loss_cns_4: 0.6590, loss_yns_4: 0.1415, loss_cls_5: 0.8578, loss_box_5: 1.5031, loss_cns_5: 0.6605, loss_yns_5: 0.1417, loss_cls_dn_0: 0.1600, loss_box_dn_0: 0.7488, loss_cls_dn_1: 0.1086, loss_box_dn_1: 0.6559, loss_cls_dn_2: 0.1073, loss_box_dn_2: 0.6458, loss_cls_dn_3: 0.1113, loss_box_dn_3: 0.6517, loss_cls_dn_4: 0.1161, loss_box_dn_4: 0.6495, loss_cls_dn_5: 0.1154, loss_box_dn_5: 0.6619, loss_dense_depth: 0.8264, loss: 24.4896, grad_norm: 54.2076 -2026-01-14 21:39:09,360 - mmdet - INFO - Iter [277/17500] lr: 2.103e-04, eta: 10:08:37, time: 1.587, data_time: 0.073, memory: 49164, loss_cls_0: 0.7475, loss_box_0: 1.6651, loss_cns_0: 0.6295, loss_yns_0: 0.1447, loss_cls_1: 0.8158, loss_box_1: 1.5376, loss_cns_1: 0.6574, loss_yns_1: 0.1435, loss_cls_2: 0.8180, loss_box_2: 1.5253, loss_cns_2: 0.6628, loss_yns_2: 0.1420, loss_cls_3: 0.8246, loss_box_3: 1.5328, loss_cns_3: 0.6570, loss_yns_3: 0.1409, loss_cls_4: 0.8352, loss_box_4: 1.5291, loss_cns_4: 0.6622, loss_yns_4: 0.1410, loss_cls_5: 0.8389, loss_box_5: 1.5169, loss_cns_5: 0.6602, loss_yns_5: 0.1418, loss_cls_dn_0: 0.1558, loss_box_dn_0: 0.7387, loss_cls_dn_1: 0.1053, loss_box_dn_1: 0.6614, loss_cls_dn_2: 0.1051, loss_box_dn_2: 0.6568, loss_cls_dn_3: 0.1079, loss_box_dn_3: 0.6703, loss_cls_dn_4: 0.1171, loss_box_dn_4: 0.6802, loss_cls_dn_5: 0.1183, loss_box_dn_5: 0.6869, loss_dense_depth: 0.8360, loss: 24.6095, grad_norm: 57.4768 -2026-01-14 21:39:10,928 - mmdet - INFO - Iter [278/17500] lr: 2.107e-04, eta: 10:08:01, time: 1.566, data_time: 0.071, memory: 49164, loss_cls_0: 0.7523, loss_box_0: 1.6303, loss_cns_0: 0.6279, loss_yns_0: 0.1434, loss_cls_1: 0.8304, loss_box_1: 1.5411, loss_cns_1: 0.6565, loss_yns_1: 0.1422, loss_cls_2: 0.8387, loss_box_2: 1.5639, loss_cns_2: 0.6601, loss_yns_2: 0.1410, loss_cls_3: 0.8556, loss_box_3: 1.5506, loss_cns_3: 0.6534, loss_yns_3: 0.1402, loss_cls_4: 0.8582, loss_box_4: 1.5490, loss_cns_4: 0.6567, loss_yns_4: 0.1392, loss_cls_5: 0.8677, loss_box_5: 1.5358, loss_cns_5: 0.6552, loss_yns_5: 0.1401, loss_cls_dn_0: 0.1524, loss_box_dn_0: 0.7301, loss_cls_dn_1: 0.1055, loss_box_dn_1: 0.6942, loss_cls_dn_2: 0.1086, loss_box_dn_2: 0.6984, loss_cls_dn_3: 0.1122, loss_box_dn_3: 0.7040, loss_cls_dn_4: 0.1163, loss_box_dn_4: 0.7122, loss_cls_dn_5: 0.1176, loss_box_dn_5: 0.7215, loss_dense_depth: 0.7552, loss: 24.8578, grad_norm: 60.7384 -2026-01-14 21:39:12,533 - mmdet - INFO - Iter [279/17500] lr: 2.111e-04, eta: 10:07:25, time: 1.577, data_time: 0.081, memory: 49164, loss_cls_0: 0.7528, loss_box_0: 1.6167, loss_cns_0: 0.6291, loss_yns_0: 0.1446, loss_cls_1: 0.8391, loss_box_1: 1.5132, loss_cns_1: 0.6552, loss_yns_1: 0.1439, loss_cls_2: 0.8454, loss_box_2: 1.5001, loss_cns_2: 0.6643, loss_yns_2: 0.1429, loss_cls_3: 0.8442, loss_box_3: 1.4796, loss_cns_3: 0.6649, loss_yns_3: 0.1437, loss_cls_4: 0.8496, loss_box_4: 1.4815, loss_cns_4: 0.6647, loss_yns_4: 0.1410, loss_cls_5: 0.8632, loss_box_5: 1.4834, loss_cns_5: 0.6652, loss_yns_5: 0.1416, loss_cls_dn_0: 0.1559, loss_box_dn_0: 0.7349, loss_cls_dn_1: 0.1102, loss_box_dn_1: 0.7001, loss_cls_dn_2: 0.1140, loss_box_dn_2: 0.6945, loss_cls_dn_3: 0.1134, loss_box_dn_3: 0.6961, loss_cls_dn_4: 0.1165, loss_box_dn_4: 0.6987, loss_cls_dn_5: 0.1201, loss_box_dn_5: 0.7085, loss_dense_depth: 0.8310, loss: 24.6641, grad_norm: 40.2545 -2026-01-14 21:39:14,102 - mmdet - INFO - Iter [280/17500] lr: 2.115e-04, eta: 10:06:51, time: 1.600, data_time: 0.096, memory: 49164, loss_cls_0: 0.7462, loss_box_0: 1.6326, loss_cns_0: 0.6299, loss_yns_0: 0.1449, loss_cls_1: 0.8225, loss_box_1: 1.5143, loss_cns_1: 0.6589, loss_yns_1: 0.1429, loss_cls_2: 0.8356, loss_box_2: 1.5370, loss_cns_2: 0.6585, loss_yns_2: 0.1428, loss_cls_3: 0.8327, loss_box_3: 1.5185, loss_cns_3: 0.6682, loss_yns_3: 0.1425, loss_cls_4: 0.8475, loss_box_4: 1.5199, loss_cns_4: 0.6652, loss_yns_4: 0.1419, loss_cls_5: 0.8446, loss_box_5: 1.5176, loss_cns_5: 0.6695, loss_yns_5: 0.1423, loss_cls_dn_0: 0.1503, loss_box_dn_0: 0.7400, loss_cls_dn_1: 0.1097, loss_box_dn_1: 0.7070, loss_cls_dn_2: 0.1101, loss_box_dn_2: 0.7107, loss_cls_dn_3: 0.1113, loss_box_dn_3: 0.7094, loss_cls_dn_4: 0.1137, loss_box_dn_4: 0.7144, loss_cls_dn_5: 0.1176, loss_box_dn_5: 0.7212, loss_dense_depth: 0.7919, loss: 24.7839, grad_norm: 50.5571 -2026-01-14 21:39:15,832 - mmdet - INFO - Iter [281/17500] lr: 2.119e-04, eta: 10:06:23, time: 1.680, data_time: 0.109, memory: 49164, loss_cls_0: 0.7410, loss_box_0: 1.6133, loss_cns_0: 0.6358, loss_yns_0: 0.1437, loss_cls_1: 0.8331, loss_box_1: 1.4775, loss_cns_1: 0.6641, loss_yns_1: 0.1395, loss_cls_2: 0.8527, loss_box_2: 1.4869, loss_cns_2: 0.6621, loss_yns_2: 0.1393, loss_cls_3: 0.8479, loss_box_3: 1.4681, loss_cns_3: 0.6713, loss_yns_3: 0.1402, loss_cls_4: 0.8578, loss_box_4: 1.4507, loss_cns_4: 0.6701, loss_yns_4: 0.1412, loss_cls_5: 0.8502, loss_box_5: 1.4488, loss_cns_5: 0.6704, loss_yns_5: 0.1405, loss_cls_dn_0: 0.1533, loss_box_dn_0: 0.7328, loss_cls_dn_1: 0.1086, loss_box_dn_1: 0.6820, loss_cls_dn_2: 0.1098, loss_box_dn_2: 0.6783, loss_cls_dn_3: 0.1153, loss_box_dn_3: 0.6703, loss_cls_dn_4: 0.1127, loss_box_dn_4: 0.6670, loss_cls_dn_5: 0.1161, loss_box_dn_5: 0.6706, loss_dense_depth: 0.8201, loss: 24.3831, grad_norm: 35.4708 -2026-01-14 21:39:17,484 - mmdet - INFO - Iter [282/17500] lr: 2.123e-04, eta: 10:05:55, time: 1.702, data_time: 0.202, memory: 49164, loss_cls_0: 0.7458, loss_box_0: 1.6477, loss_cns_0: 0.6273, loss_yns_0: 0.1471, loss_cls_1: 0.8375, loss_box_1: 1.4995, loss_cns_1: 0.6601, loss_yns_1: 0.1451, loss_cls_2: 0.8496, loss_box_2: 1.4637, loss_cns_2: 0.6642, loss_yns_2: 0.1453, loss_cls_3: 0.8408, loss_box_3: 1.4787, loss_cns_3: 0.6603, loss_yns_3: 0.1447, loss_cls_4: 0.8540, loss_box_4: 1.4643, loss_cns_4: 0.6597, loss_yns_4: 0.1460, loss_cls_5: 0.8550, loss_box_5: 1.4697, loss_cns_5: 0.6581, loss_yns_5: 0.1461, loss_cls_dn_0: 0.1647, loss_box_dn_0: 0.7542, loss_cls_dn_1: 0.1130, loss_box_dn_1: 0.6816, loss_cls_dn_2: 0.1151, loss_box_dn_2: 0.6694, loss_cls_dn_3: 0.1192, loss_box_dn_3: 0.6714, loss_cls_dn_4: 0.1170, loss_box_dn_4: 0.6754, loss_cls_dn_5: 0.1177, loss_box_dn_5: 0.6854, loss_dense_depth: 0.7387, loss: 24.4332, grad_norm: 45.3409 -2026-01-14 21:39:19,039 - mmdet - INFO - Iter [283/17500] lr: 2.127e-04, eta: 10:05:19, time: 1.553, data_time: 0.062, memory: 49164, loss_cls_0: 0.7537, loss_box_0: 1.6360, loss_cns_0: 0.6273, loss_yns_0: 0.1469, loss_cls_1: 0.8236, loss_box_1: 1.4887, loss_cns_1: 0.6571, loss_yns_1: 0.1469, loss_cls_2: 0.8397, loss_box_2: 1.4594, loss_cns_2: 0.6588, loss_yns_2: 0.1445, loss_cls_3: 0.8421, loss_box_3: 1.4736, loss_cns_3: 0.6651, loss_yns_3: 0.1450, loss_cls_4: 0.8559, loss_box_4: 1.4503, loss_cns_4: 0.6603, loss_yns_4: 0.1450, loss_cls_5: 0.8689, loss_box_5: 1.4474, loss_cns_5: 0.6641, loss_yns_5: 0.1450, loss_cls_dn_0: 0.1600, loss_box_dn_0: 0.7401, loss_cls_dn_1: 0.1130, loss_box_dn_1: 0.6747, loss_cls_dn_2: 0.1108, loss_box_dn_2: 0.6624, loss_cls_dn_3: 0.1119, loss_box_dn_3: 0.6683, loss_cls_dn_4: 0.1138, loss_box_dn_4: 0.6752, loss_cls_dn_5: 0.1164, loss_box_dn_5: 0.6795, loss_dense_depth: 0.7774, loss: 24.3488, grad_norm: 41.2192 -2026-01-14 21:39:20,653 - mmdet - INFO - Iter [284/17500] lr: 2.131e-04, eta: 10:04:46, time: 1.588, data_time: 0.075, memory: 49164, loss_cls_0: 0.7482, loss_box_0: 1.6125, loss_cns_0: 0.6300, loss_yns_0: 0.1485, loss_cls_1: 0.8167, loss_box_1: 1.5430, loss_cns_1: 0.6526, loss_yns_1: 0.1483, loss_cls_2: 0.8399, loss_box_2: 1.4792, loss_cns_2: 0.6543, loss_yns_2: 0.1470, loss_cls_3: 0.8337, loss_box_3: 1.4826, loss_cns_3: 0.6650, loss_yns_3: 0.1475, loss_cls_4: 0.8372, loss_box_4: 1.4897, loss_cns_4: 0.6576, loss_yns_4: 0.1471, loss_cls_5: 0.8626, loss_box_5: 1.4669, loss_cns_5: 0.6605, loss_yns_5: 0.1465, loss_cls_dn_0: 0.1553, loss_box_dn_0: 0.7369, loss_cls_dn_1: 0.1107, loss_box_dn_1: 0.6890, loss_cls_dn_2: 0.1122, loss_box_dn_2: 0.6717, loss_cls_dn_3: 0.1137, loss_box_dn_3: 0.6813, loss_cls_dn_4: 0.1145, loss_box_dn_4: 0.6871, loss_cls_dn_5: 0.1187, loss_box_dn_5: 0.6906, loss_dense_depth: 0.7625, loss: 24.4610, grad_norm: 44.4423 -2026-01-14 21:39:22,352 - mmdet - INFO - Iter [285/17500] lr: 2.135e-04, eta: 10:04:19, time: 1.708, data_time: 0.104, memory: 49164, loss_cls_0: 0.7762, loss_box_0: 1.6456, loss_cns_0: 0.6314, loss_yns_0: 0.1510, loss_cls_1: 0.8291, loss_box_1: 1.5462, loss_cns_1: 0.6523, loss_yns_1: 0.1486, loss_cls_2: 0.8455, loss_box_2: 1.5094, loss_cns_2: 0.6579, loss_yns_2: 0.1493, loss_cls_3: 0.8489, loss_box_3: 1.5038, loss_cns_3: 0.6595, loss_yns_3: 0.1487, loss_cls_4: 0.8591, loss_box_4: 1.4826, loss_cns_4: 0.6572, loss_yns_4: 0.1480, loss_cls_5: 0.8635, loss_box_5: 1.4778, loss_cns_5: 0.6555, loss_yns_5: 0.1485, loss_cls_dn_0: 0.1680, loss_box_dn_0: 0.7450, loss_cls_dn_1: 0.1107, loss_box_dn_1: 0.6868, loss_cls_dn_2: 0.1117, loss_box_dn_2: 0.6766, loss_cls_dn_3: 0.1156, loss_box_dn_3: 0.6871, loss_cls_dn_4: 0.1155, loss_box_dn_4: 0.6865, loss_cls_dn_5: 0.1177, loss_box_dn_5: 0.6978, loss_dense_depth: 0.8152, loss: 24.7299, grad_norm: 43.2106 -2026-01-14 21:39:24,037 - mmdet - INFO - Iter [286/17500] lr: 2.139e-04, eta: 10:03:53, time: 1.702, data_time: 0.116, memory: 49164, loss_cls_0: 0.7521, loss_box_0: 1.5981, loss_cns_0: 0.6267, loss_yns_0: 0.1494, loss_cls_1: 0.8389, loss_box_1: 1.5005, loss_cns_1: 0.6584, loss_yns_1: 0.1497, loss_cls_2: 0.8513, loss_box_2: 1.4852, loss_cns_2: 0.6616, loss_yns_2: 0.1486, loss_cls_3: 0.8543, loss_box_3: 1.4857, loss_cns_3: 0.6562, loss_yns_3: 0.1480, loss_cls_4: 0.8737, loss_box_4: 1.4820, loss_cns_4: 0.6558, loss_yns_4: 0.1478, loss_cls_5: 0.8775, loss_box_5: 1.4850, loss_cns_5: 0.6558, loss_yns_5: 0.1478, loss_cls_dn_0: 0.1626, loss_box_dn_0: 0.7297, loss_cls_dn_1: 0.1097, loss_box_dn_1: 0.6775, loss_cls_dn_2: 0.1071, loss_box_dn_2: 0.6763, loss_cls_dn_3: 0.1084, loss_box_dn_3: 0.6819, loss_cls_dn_4: 0.1101, loss_box_dn_4: 0.6861, loss_cls_dn_5: 0.1129, loss_box_dn_5: 0.6950, loss_dense_depth: 0.7538, loss: 24.5014, grad_norm: 44.1693 -2026-01-14 21:39:25,594 - mmdet - INFO - Iter [287/17500] lr: 2.143e-04, eta: 10:03:18, time: 1.553, data_time: 0.077, memory: 49164, loss_cls_0: 0.7597, loss_box_0: 1.5852, loss_cns_0: 0.6238, loss_yns_0: 0.1491, loss_cls_1: 0.8504, loss_box_1: 1.5309, loss_cns_1: 0.6599, loss_yns_1: 0.1505, loss_cls_2: 0.8635, loss_box_2: 1.5066, loss_cns_2: 0.6637, loss_yns_2: 0.1496, loss_cls_3: 0.8668, loss_box_3: 1.4936, loss_cns_3: 0.6625, loss_yns_3: 0.1498, loss_cls_4: 0.8696, loss_box_4: 1.4994, loss_cns_4: 0.6567, loss_yns_4: 0.1494, loss_cls_5: 0.8839, loss_box_5: 1.4963, loss_cns_5: 0.6643, loss_yns_5: 0.1508, loss_cls_dn_0: 0.1547, loss_box_dn_0: 0.7424, loss_cls_dn_1: 0.1129, loss_box_dn_1: 0.6833, loss_cls_dn_2: 0.1117, loss_box_dn_2: 0.6800, loss_cls_dn_3: 0.1143, loss_box_dn_3: 0.6755, loss_cls_dn_4: 0.1174, loss_box_dn_4: 0.6834, loss_cls_dn_5: 0.1185, loss_box_dn_5: 0.6835, loss_dense_depth: 0.7501, loss: 24.6637, grad_norm: 37.7777 -2026-01-14 21:39:27,155 - mmdet - INFO - Iter [288/17500] lr: 2.147e-04, eta: 10:02:43, time: 1.564, data_time: 0.075, memory: 49164, loss_cls_0: 0.7833, loss_box_0: 1.5713, loss_cns_0: 0.6166, loss_yns_0: 0.1461, loss_cls_1: 0.8408, loss_box_1: 1.5337, loss_cns_1: 0.6574, loss_yns_1: 0.1497, loss_cls_2: 0.8404, loss_box_2: 1.5259, loss_cns_2: 0.6568, loss_yns_2: 0.1495, loss_cls_3: 0.8616, loss_box_3: 1.5062, loss_cns_3: 0.6609, loss_yns_3: 0.1488, loss_cls_4: 0.8787, loss_box_4: 1.4765, loss_cns_4: 0.6572, loss_yns_4: 0.1487, loss_cls_5: 0.8706, loss_box_5: 1.4988, loss_cns_5: 0.6668, loss_yns_5: 0.1513, loss_cls_dn_0: 0.1570, loss_box_dn_0: 0.7424, loss_cls_dn_1: 0.1103, loss_box_dn_1: 0.6789, loss_cls_dn_2: 0.1099, loss_box_dn_2: 0.6760, loss_cls_dn_3: 0.1150, loss_box_dn_3: 0.6651, loss_cls_dn_4: 0.1143, loss_box_dn_4: 0.6585, loss_cls_dn_5: 0.1131, loss_box_dn_5: 0.6618, loss_dense_depth: 0.7559, loss: 24.5557, grad_norm: 45.2596 -2026-01-14 21:39:28,742 - mmdet - INFO - Iter [289/17500] lr: 2.151e-04, eta: 10:02:11, time: 1.588, data_time: 0.075, memory: 49164, loss_cls_0: 0.7364, loss_box_0: 1.5784, loss_cns_0: 0.6291, loss_yns_0: 0.1497, loss_cls_1: 0.8034, loss_box_1: 1.5027, loss_cns_1: 0.6588, loss_yns_1: 0.1514, loss_cls_2: 0.8183, loss_box_2: 1.4986, loss_cns_2: 0.6599, loss_yns_2: 0.1506, loss_cls_3: 0.8292, loss_box_3: 1.4766, loss_cns_3: 0.6587, loss_yns_3: 0.1483, loss_cls_4: 0.8495, loss_box_4: 1.4373, loss_cns_4: 0.6574, loss_yns_4: 0.1481, loss_cls_5: 0.8383, loss_box_5: 1.4721, loss_cns_5: 0.6606, loss_yns_5: 0.1479, loss_cls_dn_0: 0.1573, loss_box_dn_0: 0.7251, loss_cls_dn_1: 0.1084, loss_box_dn_1: 0.6606, loss_cls_dn_2: 0.1083, loss_box_dn_2: 0.6549, loss_cls_dn_3: 0.1088, loss_box_dn_3: 0.6521, loss_cls_dn_4: 0.1083, loss_box_dn_4: 0.6473, loss_cls_dn_5: 0.1097, loss_box_dn_5: 0.6534, loss_dense_depth: 0.7268, loss: 24.0823, grad_norm: 38.8030 -2026-01-14 21:39:30,362 - mmdet - INFO - Iter [290/17500] lr: 2.155e-04, eta: 10:01:40, time: 1.621, data_time: 0.073, memory: 49164, loss_cls_0: 0.7648, loss_box_0: 1.6437, loss_cns_0: 0.6268, loss_yns_0: 0.1503, loss_cls_1: 0.8305, loss_box_1: 1.5195, loss_cns_1: 0.6572, loss_yns_1: 0.1500, loss_cls_2: 0.8490, loss_box_2: 1.4854, loss_cns_2: 0.6616, loss_yns_2: 0.1500, loss_cls_3: 0.8535, loss_box_3: 1.4648, loss_cns_3: 0.6557, loss_yns_3: 0.1494, loss_cls_4: 0.8595, loss_box_4: 1.4912, loss_cns_4: 0.6568, loss_yns_4: 0.1489, loss_cls_5: 0.8618, loss_box_5: 1.4560, loss_cns_5: 0.6560, loss_yns_5: 0.1478, loss_cls_dn_0: 0.1548, loss_box_dn_0: 0.7388, loss_cls_dn_1: 0.1110, loss_box_dn_1: 0.6733, loss_cls_dn_2: 0.1106, loss_box_dn_2: 0.6582, loss_cls_dn_3: 0.1130, loss_box_dn_3: 0.6562, loss_cls_dn_4: 0.1133, loss_box_dn_4: 0.6766, loss_cls_dn_5: 0.1179, loss_box_dn_5: 0.6687, loss_dense_depth: 0.7579, loss: 24.4406, grad_norm: 43.6522 -2026-01-14 21:39:31,947 - mmdet - INFO - Iter [291/17500] lr: 2.159e-04, eta: 10:01:08, time: 1.582, data_time: 0.077, memory: 49164, loss_cls_0: 0.7896, loss_box_0: 1.6542, loss_cns_0: 0.6342, loss_yns_0: 0.1546, loss_cls_1: 0.8365, loss_box_1: 1.5320, loss_cns_1: 0.6588, loss_yns_1: 0.1521, loss_cls_2: 0.8486, loss_box_2: 1.5012, loss_cns_2: 0.6681, loss_yns_2: 0.1522, loss_cls_3: 0.8611, loss_box_3: 1.4815, loss_cns_3: 0.6641, loss_yns_3: 0.1528, loss_cls_4: 0.8707, loss_box_4: 1.5032, loss_cns_4: 0.6641, loss_yns_4: 0.1518, loss_cls_5: 0.8774, loss_box_5: 1.4608, loss_cns_5: 0.6631, loss_yns_5: 0.1517, loss_cls_dn_0: 0.1518, loss_box_dn_0: 0.7371, loss_cls_dn_1: 0.1050, loss_box_dn_1: 0.6798, loss_cls_dn_2: 0.1076, loss_box_dn_2: 0.6682, loss_cls_dn_3: 0.1097, loss_box_dn_3: 0.6675, loss_cls_dn_4: 0.1110, loss_box_dn_4: 0.6840, loss_cls_dn_5: 0.1188, loss_box_dn_5: 0.6789, loss_dense_depth: 0.7678, loss: 24.6716, grad_norm: 41.9273 -2026-01-14 21:39:33,579 - mmdet - INFO - Iter [292/17500] lr: 2.163e-04, eta: 10:00:35, time: 1.578, data_time: 0.078, memory: 49164, loss_cls_0: 0.7408, loss_box_0: 1.6394, loss_cns_0: 0.6323, loss_yns_0: 0.1500, loss_cls_1: 0.8081, loss_box_1: 1.5394, loss_cns_1: 0.6558, loss_yns_1: 0.1488, loss_cls_2: 0.8241, loss_box_2: 1.5099, loss_cns_2: 0.6638, loss_yns_2: 0.1504, loss_cls_3: 0.8331, loss_box_3: 1.4971, loss_cns_3: 0.6657, loss_yns_3: 0.1511, loss_cls_4: 0.8432, loss_box_4: 1.4867, loss_cns_4: 0.6670, loss_yns_4: 0.1512, loss_cls_5: 0.8500, loss_box_5: 1.4792, loss_cns_5: 0.6678, loss_yns_5: 0.1493, loss_cls_dn_0: 0.1512, loss_box_dn_0: 0.7355, loss_cls_dn_1: 0.1086, loss_box_dn_1: 0.6688, loss_cls_dn_2: 0.1090, loss_box_dn_2: 0.6631, loss_cls_dn_3: 0.1096, loss_box_dn_3: 0.6640, loss_cls_dn_4: 0.1132, loss_box_dn_4: 0.6664, loss_cls_dn_5: 0.1148, loss_box_dn_5: 0.6700, loss_dense_depth: 0.7235, loss: 24.4020, grad_norm: 38.7562 -2026-01-14 21:39:35,162 - mmdet - INFO - Iter [293/17500] lr: 2.167e-04, eta: 10:00:06, time: 1.640, data_time: 0.140, memory: 49164, loss_cls_0: 0.7540, loss_box_0: 1.6739, loss_cns_0: 0.6297, loss_yns_0: 0.1477, loss_cls_1: 0.8298, loss_box_1: 1.5101, loss_cns_1: 0.6612, loss_yns_1: 0.1438, loss_cls_2: 0.8314, loss_box_2: 1.4833, loss_cns_2: 0.6614, loss_yns_2: 0.1448, loss_cls_3: 0.8426, loss_box_3: 1.4631, loss_cns_3: 0.6642, loss_yns_3: 0.1452, loss_cls_4: 0.8564, loss_box_4: 1.4741, loss_cns_4: 0.6628, loss_yns_4: 0.1443, loss_cls_5: 0.8761, loss_box_5: 1.4578, loss_cns_5: 0.6643, loss_yns_5: 0.1442, loss_cls_dn_0: 0.1516, loss_box_dn_0: 0.7320, loss_cls_dn_1: 0.1071, loss_box_dn_1: 0.6718, loss_cls_dn_2: 0.1053, loss_box_dn_2: 0.6628, loss_cls_dn_3: 0.1061, loss_box_dn_3: 0.6591, loss_cls_dn_4: 0.1112, loss_box_dn_4: 0.6677, loss_cls_dn_5: 0.1116, loss_box_dn_5: 0.6675, loss_dense_depth: 0.7612, loss: 24.3814, grad_norm: 29.8016 -2026-01-14 21:39:36,729 - mmdet - INFO - Iter [294/17500] lr: 2.170e-04, eta: 9:59:33, time: 1.566, data_time: 0.069, memory: 49164, loss_cls_0: 0.7641, loss_box_0: 1.6560, loss_cns_0: 0.6268, loss_yns_0: 0.1478, loss_cls_1: 0.8206, loss_box_1: 1.5007, loss_cns_1: 0.6640, loss_yns_1: 0.1443, loss_cls_2: 0.8374, loss_box_2: 1.4785, loss_cns_2: 0.6691, loss_yns_2: 0.1435, loss_cls_3: 0.8510, loss_box_3: 1.4593, loss_cns_3: 0.6622, loss_yns_3: 0.1438, loss_cls_4: 0.8600, loss_box_4: 1.4625, loss_cns_4: 0.6636, loss_yns_4: 0.1445, loss_cls_5: 0.8750, loss_box_5: 1.4536, loss_cns_5: 0.6638, loss_yns_5: 0.1445, loss_cls_dn_0: 0.1468, loss_box_dn_0: 0.7328, loss_cls_dn_1: 0.1076, loss_box_dn_1: 0.6665, loss_cls_dn_2: 0.1075, loss_box_dn_2: 0.6565, loss_cls_dn_3: 0.1093, loss_box_dn_3: 0.6524, loss_cls_dn_4: 0.1086, loss_box_dn_4: 0.6554, loss_cls_dn_5: 0.1138, loss_box_dn_5: 0.6548, loss_dense_depth: 0.7445, loss: 24.2934, grad_norm: 32.1420 -2026-01-14 21:39:38,309 - mmdet - INFO - Iter [295/17500] lr: 2.174e-04, eta: 9:59:02, time: 1.581, data_time: 0.072, memory: 49164, loss_cls_0: 0.7710, loss_box_0: 1.6288, loss_cns_0: 0.6272, loss_yns_0: 0.1491, loss_cls_1: 0.8478, loss_box_1: 1.4914, loss_cns_1: 0.6620, loss_yns_1: 0.1468, loss_cls_2: 0.8575, loss_box_2: 1.4637, loss_cns_2: 0.6785, loss_yns_2: 0.1458, loss_cls_3: 0.8585, loss_box_3: 1.4490, loss_cns_3: 0.6653, loss_yns_3: 0.1471, loss_cls_4: 0.8746, loss_box_4: 1.4456, loss_cns_4: 0.6628, loss_yns_4: 0.1465, loss_cls_5: 0.8760, loss_box_5: 1.4382, loss_cns_5: 0.6668, loss_yns_5: 0.1477, loss_cls_dn_0: 0.1460, loss_box_dn_0: 0.7305, loss_cls_dn_1: 0.1077, loss_box_dn_1: 0.6570, loss_cls_dn_2: 0.1090, loss_box_dn_2: 0.6401, loss_cls_dn_3: 0.1115, loss_box_dn_3: 0.6368, loss_cls_dn_4: 0.1170, loss_box_dn_4: 0.6390, loss_cls_dn_5: 0.1172, loss_box_dn_5: 0.6353, loss_dense_depth: 0.7513, loss: 24.2459, grad_norm: 22.1258 -2026-01-14 21:39:39,925 - mmdet - INFO - Iter [296/17500] lr: 2.178e-04, eta: 9:58:32, time: 1.616, data_time: 0.075, memory: 49164, loss_cls_0: 0.7573, loss_box_0: 1.6465, loss_cns_0: 0.6281, loss_yns_0: 0.1479, loss_cls_1: 0.8415, loss_box_1: 1.5109, loss_cns_1: 0.6575, loss_yns_1: 0.1477, loss_cls_2: 0.8484, loss_box_2: 1.4855, loss_cns_2: 0.6599, loss_yns_2: 0.1475, loss_cls_3: 0.8531, loss_box_3: 1.4763, loss_cns_3: 0.6636, loss_yns_3: 0.1479, loss_cls_4: 0.8751, loss_box_4: 1.4629, loss_cns_4: 0.6608, loss_yns_4: 0.1475, loss_cls_5: 0.8619, loss_box_5: 1.4618, loss_cns_5: 0.6615, loss_yns_5: 0.1471, loss_cls_dn_0: 0.1512, loss_box_dn_0: 0.7305, loss_cls_dn_1: 0.1048, loss_box_dn_1: 0.6537, loss_cls_dn_2: 0.1077, loss_box_dn_2: 0.6401, loss_cls_dn_3: 0.1102, loss_box_dn_3: 0.6417, loss_cls_dn_4: 0.1185, loss_box_dn_4: 0.6448, loss_cls_dn_5: 0.1111, loss_box_dn_5: 0.6466, loss_dense_depth: 0.7502, loss: 24.3089, grad_norm: 29.7443 -2026-01-14 21:39:41,487 - mmdet - INFO - Iter [297/17500] lr: 2.182e-04, eta: 9:57:59, time: 1.559, data_time: 0.073, memory: 49164, loss_cls_0: 0.7978, loss_box_0: 1.6953, loss_cns_0: 0.6284, loss_yns_0: 0.1496, loss_cls_1: 0.8440, loss_box_1: 1.5748, loss_cns_1: 0.6563, loss_yns_1: 0.1501, loss_cls_2: 0.8582, loss_box_2: 1.5390, loss_cns_2: 0.6612, loss_yns_2: 0.1488, loss_cls_3: 0.8686, loss_box_3: 1.5413, loss_cns_3: 0.6569, loss_yns_3: 0.1484, loss_cls_4: 0.8777, loss_box_4: 1.5381, loss_cns_4: 0.6579, loss_yns_4: 0.1490, loss_cls_5: 0.8746, loss_box_5: 1.5319, loss_cns_5: 0.6564, loss_yns_5: 0.1484, loss_cls_dn_0: 0.1569, loss_box_dn_0: 0.7367, loss_cls_dn_1: 0.1070, loss_box_dn_1: 0.6553, loss_cls_dn_2: 0.1099, loss_box_dn_2: 0.6479, loss_cls_dn_3: 0.1107, loss_box_dn_3: 0.6521, loss_cls_dn_4: 0.1129, loss_box_dn_4: 0.6593, loss_cls_dn_5: 0.1148, loss_box_dn_5: 0.6716, loss_dense_depth: 0.7428, loss: 24.8307, grad_norm: 41.4682 -2026-01-14 21:39:43,089 - mmdet - INFO - Iter [298/17500] lr: 2.186e-04, eta: 9:57:29, time: 1.604, data_time: 0.076, memory: 49164, loss_cls_0: 0.7933, loss_box_0: 1.6450, loss_cns_0: 0.6339, loss_yns_0: 0.1514, loss_cls_1: 0.8513, loss_box_1: 1.5694, loss_cns_1: 0.6551, loss_yns_1: 0.1477, loss_cls_2: 0.8667, loss_box_2: 1.5363, loss_cns_2: 0.6677, loss_yns_2: 0.1484, loss_cls_3: 0.8902, loss_box_3: 1.5389, loss_cns_3: 0.6594, loss_yns_3: 0.1481, loss_cls_4: 0.8854, loss_box_4: 1.5240, loss_cns_4: 0.6591, loss_yns_4: 0.1484, loss_cls_5: 0.8767, loss_box_5: 1.5230, loss_cns_5: 0.6595, loss_yns_5: 0.1487, loss_cls_dn_0: 0.1487, loss_box_dn_0: 0.7276, loss_cls_dn_1: 0.1097, loss_box_dn_1: 0.6632, loss_cls_dn_2: 0.1122, loss_box_dn_2: 0.6574, loss_cls_dn_3: 0.1112, loss_box_dn_3: 0.6630, loss_cls_dn_4: 0.1157, loss_box_dn_4: 0.6694, loss_cls_dn_5: 0.1176, loss_box_dn_5: 0.6843, loss_dense_depth: 0.7491, loss: 24.8570, grad_norm: 27.4959 -2026-01-14 21:39:44,671 - mmdet - INFO - Iter [299/17500] lr: 2.190e-04, eta: 9:56:59, time: 1.583, data_time: 0.090, memory: 49164, loss_cls_0: 0.7885, loss_box_0: 1.6559, loss_cns_0: 0.6249, loss_yns_0: 0.1530, loss_cls_1: 0.8791, loss_box_1: 1.5536, loss_cns_1: 0.6465, loss_yns_1: 0.1468, loss_cls_2: 0.8964, loss_box_2: 1.5056, loss_cns_2: 0.6563, loss_yns_2: 0.1479, loss_cls_3: 0.8796, loss_box_3: 1.5051, loss_cns_3: 0.6566, loss_yns_3: 0.1483, loss_cls_4: 0.8689, loss_box_4: 1.5131, loss_cns_4: 0.6599, loss_yns_4: 0.1511, loss_cls_5: 0.8965, loss_box_5: 1.5173, loss_cns_5: 0.6603, loss_yns_5: 0.1505, loss_cls_dn_0: 0.1564, loss_box_dn_0: 0.7368, loss_cls_dn_1: 0.1074, loss_box_dn_1: 0.6702, loss_cls_dn_2: 0.1076, loss_box_dn_2: 0.6643, loss_cls_dn_3: 0.1077, loss_box_dn_3: 0.6655, loss_cls_dn_4: 0.1125, loss_box_dn_4: 0.6742, loss_cls_dn_5: 0.1131, loss_box_dn_5: 0.6869, loss_dense_depth: 0.7673, loss: 24.8315, grad_norm: 48.1803 -2026-01-14 21:39:46,270 - mmdet - INFO - Iter [300/17500] lr: 2.194e-04, eta: 9:56:29, time: 1.599, data_time: 0.080, memory: 49164, loss_cls_0: 0.8091, loss_box_0: 1.6373, loss_cns_0: 0.6255, loss_yns_0: 0.1513, loss_cls_1: 0.8730, loss_box_1: 1.5336, loss_cns_1: 0.6449, loss_yns_1: 0.1449, loss_cls_2: 0.8842, loss_box_2: 1.4991, loss_cns_2: 0.6503, loss_yns_2: 0.1467, loss_cls_3: 0.8812, loss_box_3: 1.4795, loss_cns_3: 0.6536, loss_yns_3: 0.1464, loss_cls_4: 0.8626, loss_box_4: 1.4971, loss_cns_4: 0.6584, loss_yns_4: 0.1469, loss_cls_5: 0.8692, loss_box_5: 1.4989, loss_cns_5: 0.6604, loss_yns_5: 0.1485, loss_cls_dn_0: 0.1757, loss_box_dn_0: 0.7370, loss_cls_dn_1: 0.1099, loss_box_dn_1: 0.6766, loss_cls_dn_2: 0.1104, loss_box_dn_2: 0.6640, loss_cls_dn_3: 0.1123, loss_box_dn_3: 0.6572, loss_cls_dn_4: 0.1155, loss_box_dn_4: 0.6573, loss_cls_dn_5: 0.1149, loss_box_dn_5: 0.6600, loss_dense_depth: 0.7816, loss: 24.6751, grad_norm: 32.5147 -2026-01-14 21:39:47,927 - mmdet - INFO - Iter [301/17500] lr: 2.198e-04, eta: 9:56:02, time: 1.656, data_time: 0.107, memory: 49164, loss_cls_0: 0.7991, loss_box_0: 1.6375, loss_cns_0: 0.6289, loss_yns_0: 0.1470, loss_cls_1: 0.8676, loss_box_1: 1.5840, loss_cns_1: 0.6546, loss_yns_1: 0.1460, loss_cls_2: 0.8816, loss_box_2: 1.5537, loss_cns_2: 0.6553, loss_yns_2: 0.1480, loss_cls_3: 0.8846, loss_box_3: 1.5495, loss_cns_3: 0.6570, loss_yns_3: 0.1463, loss_cls_4: 0.8778, loss_box_4: 1.5628, loss_cns_4: 0.6555, loss_yns_4: 0.1477, loss_cls_5: 0.8836, loss_box_5: 1.5429, loss_cns_5: 0.6544, loss_yns_5: 0.1476, loss_cls_dn_0: 0.1521, loss_box_dn_0: 0.7326, loss_cls_dn_1: 0.1093, loss_box_dn_1: 0.6593, loss_cls_dn_2: 0.1106, loss_box_dn_2: 0.6521, loss_cls_dn_3: 0.1091, loss_box_dn_3: 0.6501, loss_cls_dn_4: 0.1112, loss_box_dn_4: 0.6523, loss_cls_dn_5: 0.1130, loss_box_dn_5: 0.6464, loss_dense_depth: 0.7828, loss: 24.8937, grad_norm: 45.0953 -2026-01-14 21:39:49,607 - mmdet - INFO - Iter [302/17500] lr: 2.202e-04, eta: 9:55:38, time: 1.682, data_time: 0.165, memory: 49164, loss_cls_0: 0.7821, loss_box_0: 1.6452, loss_cns_0: 0.6302, loss_yns_0: 0.1512, loss_cls_1: 0.8626, loss_box_1: 1.5658, loss_cns_1: 0.6557, loss_yns_1: 0.1508, loss_cls_2: 0.8848, loss_box_2: 1.5280, loss_cns_2: 0.6581, loss_yns_2: 0.1511, loss_cls_3: 0.8732, loss_box_3: 1.5193, loss_cns_3: 0.6601, loss_yns_3: 0.1502, loss_cls_4: 0.8721, loss_box_4: 1.5296, loss_cns_4: 0.6607, loss_yns_4: 0.1508, loss_cls_5: 0.8948, loss_box_5: 1.5210, loss_cns_5: 0.6604, loss_yns_5: 0.1515, loss_cls_dn_0: 0.1532, loss_box_dn_0: 0.7230, loss_cls_dn_1: 0.1105, loss_box_dn_1: 0.6519, loss_cls_dn_2: 0.1130, loss_box_dn_2: 0.6428, loss_cls_dn_3: 0.1107, loss_box_dn_3: 0.6418, loss_cls_dn_4: 0.1125, loss_box_dn_4: 0.6484, loss_cls_dn_5: 0.1159, loss_box_dn_5: 0.6446, loss_dense_depth: 0.8192, loss: 24.7968, grad_norm: 46.9765 -2026-01-14 21:39:51,160 - mmdet - INFO - Iter [303/17500] lr: 2.206e-04, eta: 9:55:06, time: 1.552, data_time: 0.071, memory: 49164, loss_cls_0: 0.8182, loss_box_0: 1.7016, loss_cns_0: 0.6257, loss_yns_0: 0.1563, loss_cls_1: 0.9091, loss_box_1: 1.5791, loss_cns_1: 0.6612, loss_yns_1: 0.1547, loss_cls_2: 0.9056, loss_box_2: 1.5650, loss_cns_2: 0.6662, loss_yns_2: 0.1542, loss_cls_3: 0.8997, loss_box_3: 1.5686, loss_cns_3: 0.6622, loss_yns_3: 0.1555, loss_cls_4: 0.9141, loss_box_4: 1.5617, loss_cns_4: 0.6825, loss_yns_4: 0.1562, loss_cls_5: 0.9093, loss_box_5: 1.5638, loss_cns_5: 0.6661, loss_yns_5: 0.1555, loss_cls_dn_0: 0.1867, loss_box_dn_0: 0.7459, loss_cls_dn_1: 0.1110, loss_box_dn_1: 0.6621, loss_cls_dn_2: 0.1114, loss_box_dn_2: 0.6557, loss_cls_dn_3: 0.1141, loss_box_dn_3: 0.6580, loss_cls_dn_4: 0.1122, loss_box_dn_4: 0.6619, loss_cls_dn_5: 0.1121, loss_box_dn_5: 0.6650, loss_dense_depth: 0.7749, loss: 25.3630, grad_norm: 46.2159 -2026-01-14 21:39:52,774 - mmdet - INFO - Iter [304/17500] lr: 2.210e-04, eta: 9:54:37, time: 1.614, data_time: 0.074, memory: 49164, loss_cls_0: 0.7732, loss_box_0: 1.6623, loss_cns_0: 0.6288, loss_yns_0: 0.1533, loss_cls_1: 0.8673, loss_box_1: 1.5165, loss_cns_1: 0.6622, loss_yns_1: 0.1520, loss_cls_2: 0.8631, loss_box_2: 1.5176, loss_cns_2: 0.6697, loss_yns_2: 0.1519, loss_cls_3: 0.8610, loss_box_3: 1.5308, loss_cns_3: 0.6614, loss_yns_3: 0.1503, loss_cls_4: 0.8687, loss_box_4: 1.5276, loss_cns_4: 0.6671, loss_yns_4: 0.1524, loss_cls_5: 0.8554, loss_box_5: 1.5278, loss_cns_5: 0.6611, loss_yns_5: 0.1530, loss_cls_dn_0: 0.1607, loss_box_dn_0: 0.7336, loss_cls_dn_1: 0.1096, loss_box_dn_1: 0.6588, loss_cls_dn_2: 0.1097, loss_box_dn_2: 0.6590, loss_cls_dn_3: 0.1143, loss_box_dn_3: 0.6691, loss_cls_dn_4: 0.1132, loss_box_dn_4: 0.6786, loss_cls_dn_5: 0.1140, loss_box_dn_5: 0.6878, loss_dense_depth: 0.7854, loss: 24.8279, grad_norm: 43.1314 -2026-01-14 21:39:54,405 - mmdet - INFO - Iter [305/17500] lr: 2.214e-04, eta: 9:54:10, time: 1.629, data_time: 0.073, memory: 49164, loss_cls_0: 0.7437, loss_box_0: 1.6704, loss_cns_0: 0.6269, loss_yns_0: 0.1516, loss_cls_1: 0.8259, loss_box_1: 1.5316, loss_cns_1: 0.6607, loss_yns_1: 0.1526, loss_cls_2: 0.8345, loss_box_2: 1.5185, loss_cns_2: 0.6689, loss_yns_2: 0.1504, loss_cls_3: 0.8472, loss_box_3: 1.4759, loss_cns_3: 0.6591, loss_yns_3: 0.1489, loss_cls_4: 0.8531, loss_box_4: 1.4545, loss_cns_4: 0.6563, loss_yns_4: 0.1492, loss_cls_5: 0.8450, loss_box_5: 1.4881, loss_cns_5: 0.6577, loss_yns_5: 0.1520, loss_cls_dn_0: 0.1447, loss_box_dn_0: 0.7359, loss_cls_dn_1: 0.1085, loss_box_dn_1: 0.6819, loss_cls_dn_2: 0.1067, loss_box_dn_2: 0.6841, loss_cls_dn_3: 0.1086, loss_box_dn_3: 0.6909, loss_cls_dn_4: 0.1133, loss_box_dn_4: 0.6977, loss_cls_dn_5: 0.1133, loss_box_dn_5: 0.7117, loss_dense_depth: 0.7256, loss: 24.5455, grad_norm: 50.8426 -2026-01-14 21:39:56,065 - mmdet - INFO - Iter [306/17500] lr: 2.218e-04, eta: 9:53:45, time: 1.660, data_time: 0.079, memory: 49164, loss_cls_0: 0.7701, loss_box_0: 1.6509, loss_cns_0: 0.6225, loss_yns_0: 0.1487, loss_cls_1: 0.8461, loss_box_1: 1.5138, loss_cns_1: 0.6565, loss_yns_1: 0.1518, loss_cls_2: 0.8519, loss_box_2: 1.4891, loss_cns_2: 0.6603, loss_yns_2: 0.1485, loss_cls_3: 0.8579, loss_box_3: 1.4603, loss_cns_3: 0.6556, loss_yns_3: 0.1478, loss_cls_4: 0.8572, loss_box_4: 1.4390, loss_cns_4: 0.6588, loss_yns_4: 0.1504, loss_cls_5: 0.8644, loss_box_5: 1.4398, loss_cns_5: 0.6515, loss_yns_5: 0.1486, loss_cls_dn_0: 0.1602, loss_box_dn_0: 0.7334, loss_cls_dn_1: 0.1147, loss_box_dn_1: 0.6899, loss_cls_dn_2: 0.1127, loss_box_dn_2: 0.6828, loss_cls_dn_3: 0.1125, loss_box_dn_3: 0.6857, loss_cls_dn_4: 0.1184, loss_box_dn_4: 0.6860, loss_cls_dn_5: 0.1186, loss_box_dn_5: 0.6966, loss_dense_depth: 0.7923, loss: 24.5454, grad_norm: 38.1346 -2026-01-14 21:39:57,702 - mmdet - INFO - Iter [307/17500] lr: 2.222e-04, eta: 9:53:16, time: 1.587, data_time: 0.078, memory: 49164, loss_cls_0: 0.7542, loss_box_0: 1.6392, loss_cns_0: 0.6289, loss_yns_0: 0.1487, loss_cls_1: 0.8379, loss_box_1: 1.5099, loss_cns_1: 0.6564, loss_yns_1: 0.1510, loss_cls_2: 0.8350, loss_box_2: 1.5075, loss_cns_2: 0.6589, loss_yns_2: 0.1516, loss_cls_3: 0.8416, loss_box_3: 1.4963, loss_cns_3: 0.6628, loss_yns_3: 0.1506, loss_cls_4: 0.8468, loss_box_4: 1.4934, loss_cns_4: 0.6618, loss_yns_4: 0.1524, loss_cls_5: 0.8438, loss_box_5: 1.4779, loss_cns_5: 0.6583, loss_yns_5: 0.1504, loss_cls_dn_0: 0.1650, loss_box_dn_0: 0.7293, loss_cls_dn_1: 0.1118, loss_box_dn_1: 0.6762, loss_cls_dn_2: 0.1091, loss_box_dn_2: 0.6701, loss_cls_dn_3: 0.1092, loss_box_dn_3: 0.6656, loss_cls_dn_4: 0.1120, loss_box_dn_4: 0.6703, loss_cls_dn_5: 0.1130, loss_box_dn_5: 0.6767, loss_dense_depth: 0.7274, loss: 24.4511, grad_norm: 42.0309 -2026-01-14 21:39:59,244 - mmdet - INFO - Iter [308/17500] lr: 2.226e-04, eta: 9:52:47, time: 1.594, data_time: 0.112, memory: 49164, loss_cls_0: 0.7696, loss_box_0: 1.6206, loss_cns_0: 0.6310, loss_yns_0: 0.1481, loss_cls_1: 0.8403, loss_box_1: 1.5337, loss_cns_1: 0.6526, loss_yns_1: 0.1468, loss_cls_2: 0.8363, loss_box_2: 1.5095, loss_cns_2: 0.6621, loss_yns_2: 0.1478, loss_cls_3: 0.8565, loss_box_3: 1.5309, loss_cns_3: 0.6661, loss_yns_3: 0.1490, loss_cls_4: 0.8603, loss_box_4: 1.5104, loss_cns_4: 0.6630, loss_yns_4: 0.1487, loss_cls_5: 0.8505, loss_box_5: 1.5000, loss_cns_5: 0.6612, loss_yns_5: 0.1488, loss_cls_dn_0: 0.1604, loss_box_dn_0: 0.7355, loss_cls_dn_1: 0.1134, loss_box_dn_1: 0.6762, loss_cls_dn_2: 0.1116, loss_box_dn_2: 0.6653, loss_cls_dn_3: 0.1146, loss_box_dn_3: 0.6691, loss_cls_dn_4: 0.1185, loss_box_dn_4: 0.6653, loss_cls_dn_5: 0.1188, loss_box_dn_5: 0.6686, loss_dense_depth: 0.7904, loss: 24.6515, grad_norm: 40.1557 -2026-01-14 21:40:00,864 - mmdet - INFO - Iter [309/17500] lr: 2.230e-04, eta: 9:52:18, time: 1.577, data_time: 0.073, memory: 49164, loss_cls_0: 0.7682, loss_box_0: 1.6271, loss_cns_0: 0.6276, loss_yns_0: 0.1499, loss_cls_1: 0.8279, loss_box_1: 1.5446, loss_cns_1: 0.6592, loss_yns_1: 0.1474, loss_cls_2: 0.8365, loss_box_2: 1.5336, loss_cns_2: 0.6688, loss_yns_2: 0.1485, loss_cls_3: 0.8559, loss_box_3: 1.5525, loss_cns_3: 0.6612, loss_yns_3: 0.1478, loss_cls_4: 0.8601, loss_box_4: 1.5108, loss_cns_4: 0.6619, loss_yns_4: 0.1480, loss_cls_5: 0.8616, loss_box_5: 1.4944, loss_cns_5: 0.6587, loss_yns_5: 0.1488, loss_cls_dn_0: 0.1522, loss_box_dn_0: 0.7280, loss_cls_dn_1: 0.1149, loss_box_dn_1: 0.6587, loss_cls_dn_2: 0.1159, loss_box_dn_2: 0.6532, loss_cls_dn_3: 0.1200, loss_box_dn_3: 0.6675, loss_cls_dn_4: 0.1245, loss_box_dn_4: 0.6566, loss_cls_dn_5: 0.1216, loss_box_dn_5: 0.6566, loss_dense_depth: 0.7221, loss: 24.5929, grad_norm: 43.0101 -2026-01-14 21:40:02,495 - mmdet - INFO - Iter [310/17500] lr: 2.234e-04, eta: 9:51:52, time: 1.638, data_time: 0.106, memory: 49164, loss_cls_0: 0.7485, loss_box_0: 1.6244, loss_cns_0: 0.6341, loss_yns_0: 0.1513, loss_cls_1: 0.8140, loss_box_1: 1.5210, loss_cns_1: 0.6627, loss_yns_1: 0.1477, loss_cls_2: 0.8365, loss_box_2: 1.4914, loss_cns_2: 0.6667, loss_yns_2: 0.1470, loss_cls_3: 0.8505, loss_box_3: 1.5011, loss_cns_3: 0.6632, loss_yns_3: 0.1468, loss_cls_4: 0.8553, loss_box_4: 1.4920, loss_cns_4: 0.6652, loss_yns_4: 0.1468, loss_cls_5: 0.8643, loss_box_5: 1.4722, loss_cns_5: 0.6665, loss_yns_5: 0.1467, loss_cls_dn_0: 0.1488, loss_box_dn_0: 0.7313, loss_cls_dn_1: 0.1130, loss_box_dn_1: 0.6536, loss_cls_dn_2: 0.1131, loss_box_dn_2: 0.6435, loss_cls_dn_3: 0.1139, loss_box_dn_3: 0.6566, loss_cls_dn_4: 0.1174, loss_box_dn_4: 0.6632, loss_cls_dn_5: 0.1159, loss_box_dn_5: 0.6660, loss_dense_depth: 0.7754, loss: 24.4276, grad_norm: 36.6520 -2026-01-14 21:40:04,080 - mmdet - INFO - Iter [311/17500] lr: 2.238e-04, eta: 9:51:23, time: 1.590, data_time: 0.099, memory: 49164, loss_cls_0: 0.7595, loss_box_0: 1.6295, loss_cns_0: 0.6297, loss_yns_0: 0.1509, loss_cls_1: 0.8224, loss_box_1: 1.5370, loss_cns_1: 0.6596, loss_yns_1: 0.1456, loss_cls_2: 0.8491, loss_box_2: 1.5049, loss_cns_2: 0.6596, loss_yns_2: 0.1459, loss_cls_3: 0.8748, loss_box_3: 1.4951, loss_cns_3: 0.6617, loss_yns_3: 0.1475, loss_cls_4: 0.8695, loss_box_4: 1.5132, loss_cns_4: 0.6616, loss_yns_4: 0.1461, loss_cls_5: 0.8692, loss_box_5: 1.5372, loss_cns_5: 0.6611, loss_yns_5: 0.1474, loss_cls_dn_0: 0.1555, loss_box_dn_0: 0.7360, loss_cls_dn_1: 0.1165, loss_box_dn_1: 0.6628, loss_cls_dn_2: 0.1135, loss_box_dn_2: 0.6567, loss_cls_dn_3: 0.1150, loss_box_dn_3: 0.6594, loss_cls_dn_4: 0.1170, loss_box_dn_4: 0.6780, loss_cls_dn_5: 0.1196, loss_box_dn_5: 0.6963, loss_dense_depth: 0.7461, loss: 24.6504, grad_norm: 45.3305 -2026-01-14 21:40:05,643 - mmdet - INFO - Iter [312/17500] lr: 2.242e-04, eta: 9:50:55, time: 1.595, data_time: 0.103, memory: 49164, loss_cls_0: 0.7792, loss_box_0: 1.6369, loss_cns_0: 0.6276, loss_yns_0: 0.1488, loss_cls_1: 0.8328, loss_box_1: 1.5585, loss_cns_1: 0.6565, loss_yns_1: 0.1469, loss_cls_2: 0.8493, loss_box_2: 1.5218, loss_cns_2: 0.6588, loss_yns_2: 0.1470, loss_cls_3: 0.8596, loss_box_3: 1.5093, loss_cns_3: 0.6602, loss_yns_3: 0.1470, loss_cls_4: 0.8689, loss_box_4: 1.5072, loss_cns_4: 0.6641, loss_yns_4: 0.1466, loss_cls_5: 0.8666, loss_box_5: 1.5133, loss_cns_5: 0.6596, loss_yns_5: 0.1473, loss_cls_dn_0: 0.1583, loss_box_dn_0: 0.7274, loss_cls_dn_1: 0.1165, loss_box_dn_1: 0.6651, loss_cls_dn_2: 0.1177, loss_box_dn_2: 0.6566, loss_cls_dn_3: 0.1221, loss_box_dn_3: 0.6586, loss_cls_dn_4: 0.1226, loss_box_dn_4: 0.6644, loss_cls_dn_5: 0.1227, loss_box_dn_5: 0.6780, loss_dense_depth: 0.7762, loss: 24.7001, grad_norm: 42.8414 -2026-01-14 21:40:07,242 - mmdet - INFO - Iter [313/17500] lr: 2.246e-04, eta: 9:50:27, time: 1.577, data_time: 0.074, memory: 49164, loss_cls_0: 0.7674, loss_box_0: 1.5809, loss_cns_0: 0.6309, loss_yns_0: 0.1505, loss_cls_1: 0.8286, loss_box_1: 1.5013, loss_cns_1: 0.6591, loss_yns_1: 0.1508, loss_cls_2: 0.8394, loss_box_2: 1.4536, loss_cns_2: 0.6615, loss_yns_2: 0.1475, loss_cls_3: 0.8464, loss_box_3: 1.4559, loss_cns_3: 0.6601, loss_yns_3: 0.1479, loss_cls_4: 0.8604, loss_box_4: 1.4380, loss_cns_4: 0.6629, loss_yns_4: 0.1473, loss_cls_5: 0.8490, loss_box_5: 1.4294, loss_cns_5: 0.6594, loss_yns_5: 0.1478, loss_cls_dn_0: 0.1528, loss_box_dn_0: 0.7250, loss_cls_dn_1: 0.1125, loss_box_dn_1: 0.6749, loss_cls_dn_2: 0.1116, loss_box_dn_2: 0.6580, loss_cls_dn_3: 0.1164, loss_box_dn_3: 0.6617, loss_cls_dn_4: 0.1168, loss_box_dn_4: 0.6578, loss_cls_dn_5: 0.1144, loss_box_dn_5: 0.6599, loss_dense_depth: 0.7079, loss: 24.1456, grad_norm: 29.7356 -2026-01-14 21:40:08,872 - mmdet - INFO - Iter [314/17500] lr: 2.250e-04, eta: 9:50:00, time: 1.611, data_time: 0.099, memory: 49164, loss_cls_0: 0.7760, loss_box_0: 1.6112, loss_cns_0: 0.6277, loss_yns_0: 0.1497, loss_cls_1: 0.8328, loss_box_1: 1.4860, loss_cns_1: 0.6606, loss_yns_1: 0.1479, loss_cls_2: 0.8480, loss_box_2: 1.4566, loss_cns_2: 0.6614, loss_yns_2: 0.1460, loss_cls_3: 0.8529, loss_box_3: 1.4686, loss_cns_3: 0.6593, loss_yns_3: 0.1466, loss_cls_4: 0.8597, loss_box_4: 1.4388, loss_cns_4: 0.6612, loss_yns_4: 0.1465, loss_cls_5: 0.8621, loss_box_5: 1.4467, loss_cns_5: 0.6605, loss_yns_5: 0.1477, loss_cls_dn_0: 0.1524, loss_box_dn_0: 0.7251, loss_cls_dn_1: 0.1166, loss_box_dn_1: 0.6645, loss_cls_dn_2: 0.1217, loss_box_dn_2: 0.6602, loss_cls_dn_3: 0.1253, loss_box_dn_3: 0.6634, loss_cls_dn_4: 0.1261, loss_box_dn_4: 0.6491, loss_cls_dn_5: 0.1241, loss_box_dn_5: 0.6529, loss_dense_depth: 0.7802, loss: 24.3162, grad_norm: 40.0630 -2026-01-14 21:40:10,450 - mmdet - INFO - Iter [315/17500] lr: 2.254e-04, eta: 9:49:32, time: 1.590, data_time: 0.102, memory: 49164, loss_cls_0: 0.7643, loss_box_0: 1.6299, loss_cns_0: 0.6278, loss_yns_0: 0.1498, loss_cls_1: 0.8330, loss_box_1: 1.4829, loss_cns_1: 0.6552, loss_yns_1: 0.1485, loss_cls_2: 0.8425, loss_box_2: 1.4601, loss_cns_2: 0.6578, loss_yns_2: 0.1473, loss_cls_3: 0.8458, loss_box_3: 1.4627, loss_cns_3: 0.6563, loss_yns_3: 0.1480, loss_cls_4: 0.8513, loss_box_4: 1.4500, loss_cns_4: 0.6590, loss_yns_4: 0.1497, loss_cls_5: 0.8553, loss_box_5: 1.4507, loss_cns_5: 0.6571, loss_yns_5: 0.1496, loss_cls_dn_0: 0.1471, loss_box_dn_0: 0.7354, loss_cls_dn_1: 0.1118, loss_box_dn_1: 0.6661, loss_cls_dn_2: 0.1106, loss_box_dn_2: 0.6584, loss_cls_dn_3: 0.1140, loss_box_dn_3: 0.6593, loss_cls_dn_4: 0.1134, loss_box_dn_4: 0.6505, loss_cls_dn_5: 0.1142, loss_box_dn_5: 0.6528, loss_dense_depth: 0.7423, loss: 24.2107, grad_norm: 28.0557 -2026-01-14 21:40:12,022 - mmdet - INFO - Iter [316/17500] lr: 2.258e-04, eta: 9:49:05, time: 1.600, data_time: 0.109, memory: 49164, loss_cls_0: 0.7598, loss_box_0: 1.5712, loss_cns_0: 0.6310, loss_yns_0: 0.1466, loss_cls_1: 0.8328, loss_box_1: 1.4579, loss_cns_1: 0.6566, loss_yns_1: 0.1455, loss_cls_2: 0.8436, loss_box_2: 1.4319, loss_cns_2: 0.6597, loss_yns_2: 0.1464, loss_cls_3: 0.8419, loss_box_3: 1.4153, loss_cns_3: 0.6558, loss_yns_3: 0.1451, loss_cls_4: 0.8434, loss_box_4: 1.4287, loss_cns_4: 0.6572, loss_yns_4: 0.1460, loss_cls_5: 0.8484, loss_box_5: 1.4319, loss_cns_5: 0.6568, loss_yns_5: 0.1460, loss_cls_dn_0: 0.1474, loss_box_dn_0: 0.7260, loss_cls_dn_1: 0.1142, loss_box_dn_1: 0.6469, loss_cls_dn_2: 0.1183, loss_box_dn_2: 0.6349, loss_cls_dn_3: 0.1212, loss_box_dn_3: 0.6345, loss_cls_dn_4: 0.1194, loss_box_dn_4: 0.6379, loss_cls_dn_5: 0.1187, loss_box_dn_5: 0.6438, loss_dense_depth: 0.7513, loss: 23.9140, grad_norm: 37.8716 -2026-01-14 21:40:13,610 - mmdet - INFO - Iter [317/17500] lr: 2.262e-04, eta: 9:48:38, time: 1.589, data_time: 0.073, memory: 49164, loss_cls_0: 0.7335, loss_box_0: 1.5819, loss_cns_0: 0.6329, loss_yns_0: 0.1452, loss_cls_1: 0.8190, loss_box_1: 1.4857, loss_cns_1: 0.6587, loss_yns_1: 0.1465, loss_cls_2: 0.8222, loss_box_2: 1.4545, loss_cns_2: 0.6603, loss_yns_2: 0.1476, loss_cls_3: 0.8193, loss_box_3: 1.4474, loss_cns_3: 0.6603, loss_yns_3: 0.1451, loss_cls_4: 0.8255, loss_box_4: 1.4524, loss_cns_4: 0.6601, loss_yns_4: 0.1458, loss_cls_5: 0.8287, loss_box_5: 1.4501, loss_cns_5: 0.6597, loss_yns_5: 0.1458, loss_cls_dn_0: 0.1434, loss_box_dn_0: 0.7359, loss_cls_dn_1: 0.1141, loss_box_dn_1: 0.6571, loss_cls_dn_2: 0.1176, loss_box_dn_2: 0.6459, loss_cls_dn_3: 0.1177, loss_box_dn_3: 0.6521, loss_cls_dn_4: 0.1149, loss_box_dn_4: 0.6575, loss_cls_dn_5: 0.1166, loss_box_dn_5: 0.6644, loss_dense_depth: 0.7074, loss: 23.9724, grad_norm: 30.4168 -2026-01-14 21:40:15,256 - mmdet - INFO - Iter [318/17500] lr: 2.266e-04, eta: 9:48:12, time: 1.615, data_time: 0.074, memory: 49164, loss_cls_0: 0.7749, loss_box_0: 1.5886, loss_cns_0: 0.6320, loss_yns_0: 0.1493, loss_cls_1: 0.8369, loss_box_1: 1.4883, loss_cns_1: 0.6626, loss_yns_1: 0.1469, loss_cls_2: 0.8421, loss_box_2: 1.4727, loss_cns_2: 0.6625, loss_yns_2: 0.1479, loss_cls_3: 0.8578, loss_box_3: 1.4712, loss_cns_3: 0.6615, loss_yns_3: 0.1470, loss_cls_4: 0.8523, loss_box_4: 1.4849, loss_cns_4: 0.6598, loss_yns_4: 0.1459, loss_cls_5: 0.8567, loss_box_5: 1.4826, loss_cns_5: 0.6598, loss_yns_5: 0.1460, loss_cls_dn_0: 0.1447, loss_box_dn_0: 0.7247, loss_cls_dn_1: 0.1042, loss_box_dn_1: 0.6677, loss_cls_dn_2: 0.1031, loss_box_dn_2: 0.6666, loss_cls_dn_3: 0.1122, loss_box_dn_3: 0.6743, loss_cls_dn_4: 0.1081, loss_box_dn_4: 0.6856, loss_cls_dn_5: 0.1114, loss_box_dn_5: 0.6911, loss_dense_depth: 0.7621, loss: 24.3864, grad_norm: 41.0228 -2026-01-14 21:40:16,822 - mmdet - INFO - Iter [319/17500] lr: 2.270e-04, eta: 9:47:45, time: 1.595, data_time: 0.106, memory: 49164, loss_cls_0: 0.7512, loss_box_0: 1.5991, loss_cns_0: 0.6353, loss_yns_0: 0.1484, loss_cls_1: 0.8228, loss_box_1: 1.4810, loss_cns_1: 0.6647, loss_yns_1: 0.1462, loss_cls_2: 0.8315, loss_box_2: 1.4498, loss_cns_2: 0.6633, loss_yns_2: 0.1456, loss_cls_3: 0.8376, loss_box_3: 1.4458, loss_cns_3: 0.6631, loss_yns_3: 0.1459, loss_cls_4: 0.8316, loss_box_4: 1.4404, loss_cns_4: 0.6616, loss_yns_4: 0.1446, loss_cls_5: 0.8376, loss_box_5: 1.4347, loss_cns_5: 0.6618, loss_yns_5: 0.1446, loss_cls_dn_0: 0.1388, loss_box_dn_0: 0.7311, loss_cls_dn_1: 0.1036, loss_box_dn_1: 0.6721, loss_cls_dn_2: 0.1047, loss_box_dn_2: 0.6721, loss_cls_dn_3: 0.1149, loss_box_dn_3: 0.6703, loss_cls_dn_4: 0.1057, loss_box_dn_4: 0.6698, loss_cls_dn_5: 0.1074, loss_box_dn_5: 0.6739, loss_dense_depth: 0.7290, loss: 24.0815, grad_norm: 33.6708 -2026-01-14 21:40:18,426 - mmdet - INFO - Iter [320/17500] lr: 2.274e-04, eta: 9:47:19, time: 1.605, data_time: 0.083, memory: 49164, loss_cls_0: 0.7369, loss_box_0: 1.6080, loss_cns_0: 0.6358, loss_yns_0: 0.1475, loss_cls_1: 0.8195, loss_box_1: 1.4749, loss_cns_1: 0.6643, loss_yns_1: 0.1452, loss_cls_2: 0.8232, loss_box_2: 1.4456, loss_cns_2: 0.6620, loss_yns_2: 0.1462, loss_cls_3: 0.8148, loss_box_3: 1.4258, loss_cns_3: 0.6601, loss_yns_3: 0.1438, loss_cls_4: 0.8206, loss_box_4: 1.4341, loss_cns_4: 0.6603, loss_yns_4: 0.1440, loss_cls_5: 0.8303, loss_box_5: 1.4326, loss_cns_5: 0.6621, loss_yns_5: 0.1436, loss_cls_dn_0: 0.1360, loss_box_dn_0: 0.7197, loss_cls_dn_1: 0.1041, loss_box_dn_1: 0.6640, loss_cls_dn_2: 0.1093, loss_box_dn_2: 0.6642, loss_cls_dn_3: 0.1144, loss_box_dn_3: 0.6522, loss_cls_dn_4: 0.1100, loss_box_dn_4: 0.6528, loss_cls_dn_5: 0.1102, loss_box_dn_5: 0.6545, loss_dense_depth: 0.7022, loss: 23.8745, grad_norm: 34.1449 -2026-01-14 21:40:20,103 - mmdet - INFO - Iter [321/17500] lr: 2.278e-04, eta: 9:46:56, time: 1.656, data_time: 0.110, memory: 49164, loss_cls_0: 0.7072, loss_box_0: 1.5804, loss_cns_0: 0.6352, loss_yns_0: 0.1455, loss_cls_1: 0.7944, loss_box_1: 1.4618, loss_cns_1: 0.6622, loss_yns_1: 0.1451, loss_cls_2: 0.7974, loss_box_2: 1.4385, loss_cns_2: 0.6546, loss_yns_2: 0.1448, loss_cls_3: 0.7956, loss_box_3: 1.4290, loss_cns_3: 0.6576, loss_yns_3: 0.1430, loss_cls_4: 0.7970, loss_box_4: 1.4325, loss_cns_4: 0.6628, loss_yns_4: 0.1432, loss_cls_5: 0.8030, loss_box_5: 1.4291, loss_cns_5: 0.6621, loss_yns_5: 0.1445, loss_cls_dn_0: 0.1355, loss_box_dn_0: 0.7248, loss_cls_dn_1: 0.1058, loss_box_dn_1: 0.6628, loss_cls_dn_2: 0.1110, loss_box_dn_2: 0.6508, loss_cls_dn_3: 0.1117, loss_box_dn_3: 0.6417, loss_cls_dn_4: 0.1104, loss_box_dn_4: 0.6371, loss_cls_dn_5: 0.1104, loss_box_dn_5: 0.6395, loss_dense_depth: 0.6837, loss: 23.5914, grad_norm: 38.2558 -2026-01-14 21:40:21,750 - mmdet - INFO - Iter [322/17500] lr: 2.282e-04, eta: 9:46:33, time: 1.668, data_time: 0.189, memory: 49164, loss_cls_0: 0.7331, loss_box_0: 1.5766, loss_cns_0: 0.6332, loss_yns_0: 0.1475, loss_cls_1: 0.8056, loss_box_1: 1.4227, loss_cns_1: 0.6637, loss_yns_1: 0.1451, loss_cls_2: 0.8182, loss_box_2: 1.3870, loss_cns_2: 0.6570, loss_yns_2: 0.1447, loss_cls_3: 0.8303, loss_box_3: 1.3863, loss_cns_3: 0.6625, loss_yns_3: 0.1444, loss_cls_4: 0.8239, loss_box_4: 1.3954, loss_cns_4: 0.6627, loss_yns_4: 0.1440, loss_cls_5: 0.8496, loss_box_5: 1.4079, loss_cns_5: 0.6639, loss_yns_5: 0.1465, loss_cls_dn_0: 0.1390, loss_box_dn_0: 0.7229, loss_cls_dn_1: 0.1030, loss_box_dn_1: 0.6595, loss_cls_dn_2: 0.1041, loss_box_dn_2: 0.6488, loss_cls_dn_3: 0.1055, loss_box_dn_3: 0.6456, loss_cls_dn_4: 0.1051, loss_box_dn_4: 0.6531, loss_cls_dn_5: 0.1088, loss_box_dn_5: 0.6596, loss_dense_depth: 0.6816, loss: 23.5881, grad_norm: 29.3308 -2026-01-14 21:40:23,330 - mmdet - INFO - Iter [323/17500] lr: 2.286e-04, eta: 9:46:06, time: 1.577, data_time: 0.076, memory: 49164, loss_cls_0: 0.7265, loss_box_0: 1.5694, loss_cns_0: 0.6341, loss_yns_0: 0.1418, loss_cls_1: 0.7905, loss_box_1: 1.4723, loss_cns_1: 0.6637, loss_yns_1: 0.1425, loss_cls_2: 0.8040, loss_box_2: 1.4449, loss_cns_2: 0.6636, loss_yns_2: 0.1437, loss_cls_3: 0.8051, loss_box_3: 1.4363, loss_cns_3: 0.6625, loss_yns_3: 0.1439, loss_cls_4: 0.8159, loss_box_4: 1.4403, loss_cns_4: 0.6625, loss_yns_4: 0.1453, loss_cls_5: 0.8183, loss_box_5: 1.4571, loss_cns_5: 0.6659, loss_yns_5: 0.1460, loss_cls_dn_0: 0.1400, loss_box_dn_0: 0.7230, loss_cls_dn_1: 0.0986, loss_box_dn_1: 0.6633, loss_cls_dn_2: 0.0997, loss_box_dn_2: 0.6559, loss_cls_dn_3: 0.1023, loss_box_dn_3: 0.6577, loss_cls_dn_4: 0.1052, loss_box_dn_4: 0.6716, loss_cls_dn_5: 0.1053, loss_box_dn_5: 0.6876, loss_dense_depth: 0.6769, loss: 23.7832, grad_norm: 36.3234 -2026-01-14 21:40:24,944 - mmdet - INFO - Iter [324/17500] lr: 2.290e-04, eta: 9:45:42, time: 1.619, data_time: 0.077, memory: 49164, loss_cls_0: 0.7192, loss_box_0: 1.6145, loss_cns_0: 0.6335, loss_yns_0: 0.1433, loss_cls_1: 0.7953, loss_box_1: 1.4883, loss_cns_1: 0.6626, loss_yns_1: 0.1439, loss_cls_2: 0.8032, loss_box_2: 1.4739, loss_cns_2: 0.6620, loss_yns_2: 0.1443, loss_cls_3: 0.8075, loss_box_3: 1.4509, loss_cns_3: 0.6639, loss_yns_3: 0.1431, loss_cls_4: 0.8162, loss_box_4: 1.4606, loss_cns_4: 0.6647, loss_yns_4: 0.1456, loss_cls_5: 0.8140, loss_box_5: 1.4646, loss_cns_5: 0.6704, loss_yns_5: 0.1469, loss_cls_dn_0: 0.1346, loss_box_dn_0: 0.7186, loss_cls_dn_1: 0.1003, loss_box_dn_1: 0.6621, loss_cls_dn_2: 0.1011, loss_box_dn_2: 0.6517, loss_cls_dn_3: 0.1021, loss_box_dn_3: 0.6439, loss_cls_dn_4: 0.1037, loss_box_dn_4: 0.6559, loss_cls_dn_5: 0.1044, loss_box_dn_5: 0.6648, loss_dense_depth: 0.6902, loss: 23.8658, grad_norm: 30.1478 -2026-01-14 21:40:26,585 - mmdet - INFO - Iter [325/17500] lr: 2.294e-04, eta: 9:45:18, time: 1.641, data_time: 0.074, memory: 49164, loss_cls_0: 0.7100, loss_box_0: 1.6212, loss_cns_0: 0.6407, loss_yns_0: 0.1454, loss_cls_1: 0.7886, loss_box_1: 1.4725, loss_cns_1: 0.6667, loss_yns_1: 0.1437, loss_cls_2: 0.8003, loss_box_2: 1.4629, loss_cns_2: 0.6670, loss_yns_2: 0.1452, loss_cls_3: 0.7997, loss_box_3: 1.4397, loss_cns_3: 0.6653, loss_yns_3: 0.1438, loss_cls_4: 0.8120, loss_box_4: 1.4359, loss_cns_4: 0.6670, loss_yns_4: 0.1439, loss_cls_5: 0.8125, loss_box_5: 1.4345, loss_cns_5: 0.6673, loss_yns_5: 0.1444, loss_cls_dn_0: 0.1341, loss_box_dn_0: 0.7209, loss_cls_dn_1: 0.1021, loss_box_dn_1: 0.6509, loss_cls_dn_2: 0.1035, loss_box_dn_2: 0.6433, loss_cls_dn_3: 0.1037, loss_box_dn_3: 0.6353, loss_cls_dn_4: 0.1059, loss_box_dn_4: 0.6358, loss_cls_dn_5: 0.1102, loss_box_dn_5: 0.6411, loss_dense_depth: 0.6942, loss: 23.7111, grad_norm: 36.9478 -2026-01-14 21:40:28,344 - mmdet - INFO - Iter [326/17500] lr: 2.298e-04, eta: 9:44:59, time: 1.712, data_time: 0.073, memory: 49164, loss_cls_0: 0.7091, loss_box_0: 1.5976, loss_cns_0: 0.6404, loss_yns_0: 0.1458, loss_cls_1: 0.7747, loss_box_1: 1.4883, loss_cns_1: 0.6628, loss_yns_1: 0.1415, loss_cls_2: 0.7861, loss_box_2: 1.4532, loss_cns_2: 0.6641, loss_yns_2: 0.1443, loss_cls_3: 0.7932, loss_box_3: 1.4350, loss_cns_3: 0.6639, loss_yns_3: 0.1432, loss_cls_4: 0.7990, loss_box_4: 1.4280, loss_cns_4: 0.6653, loss_yns_4: 0.1430, loss_cls_5: 0.8001, loss_box_5: 1.4373, loss_cns_5: 0.6644, loss_yns_5: 0.1419, loss_cls_dn_0: 0.1280, loss_box_dn_0: 0.7129, loss_cls_dn_1: 0.1018, loss_box_dn_1: 0.6332, loss_cls_dn_2: 0.1006, loss_box_dn_2: 0.6203, loss_cls_dn_3: 0.1004, loss_box_dn_3: 0.6150, loss_cls_dn_4: 0.1037, loss_box_dn_4: 0.6135, loss_cls_dn_5: 0.1051, loss_box_dn_5: 0.6231, loss_dense_depth: 0.6801, loss: 23.4600, grad_norm: 33.6995 -2026-01-14 21:40:29,928 - mmdet - INFO - Iter [327/17500] lr: 2.302e-04, eta: 9:44:33, time: 1.602, data_time: 0.104, memory: 49164, loss_cls_0: 0.7179, loss_box_0: 1.6011, loss_cns_0: 0.6351, loss_yns_0: 0.1412, loss_cls_1: 0.7882, loss_box_1: 1.4674, loss_cns_1: 0.6615, loss_yns_1: 0.1395, loss_cls_2: 0.8075, loss_box_2: 1.4418, loss_cns_2: 0.6582, loss_yns_2: 0.1406, loss_cls_3: 0.8099, loss_box_3: 1.4492, loss_cns_3: 0.6596, loss_yns_3: 0.1402, loss_cls_4: 0.8040, loss_box_4: 1.4509, loss_cns_4: 0.6652, loss_yns_4: 0.1425, loss_cls_5: 0.8025, loss_box_5: 1.4374, loss_cns_5: 0.6625, loss_yns_5: 0.1407, loss_cls_dn_0: 0.1352, loss_box_dn_0: 0.7176, loss_cls_dn_1: 0.1027, loss_box_dn_1: 0.6313, loss_cls_dn_2: 0.1022, loss_box_dn_2: 0.6193, loss_cls_dn_3: 0.1026, loss_box_dn_3: 0.6240, loss_cls_dn_4: 0.1047, loss_box_dn_4: 0.6213, loss_cls_dn_5: 0.1045, loss_box_dn_5: 0.6240, loss_dense_depth: 0.6752, loss: 23.5291, grad_norm: 31.4884 -2026-01-14 21:40:31,486 - mmdet - INFO - Iter [328/17500] lr: 2.306e-04, eta: 9:44:07, time: 1.586, data_time: 0.104, memory: 49164, loss_cls_0: 0.7164, loss_box_0: 1.6083, loss_cns_0: 0.6356, loss_yns_0: 0.1433, loss_cls_1: 0.7932, loss_box_1: 1.4932, loss_cns_1: 0.6630, loss_yns_1: 0.1409, loss_cls_2: 0.7952, loss_box_2: 1.4419, loss_cns_2: 0.6622, loss_yns_2: 0.1401, loss_cls_3: 0.8026, loss_box_3: 1.4568, loss_cns_3: 0.6649, loss_yns_3: 0.1414, loss_cls_4: 0.8269, loss_box_4: 1.4509, loss_cns_4: 0.6661, loss_yns_4: 0.1426, loss_cls_5: 0.8095, loss_box_5: 1.4302, loss_cns_5: 0.6647, loss_yns_5: 0.1415, loss_cls_dn_0: 0.1338, loss_box_dn_0: 0.7254, loss_cls_dn_1: 0.1026, loss_box_dn_1: 0.6480, loss_cls_dn_2: 0.1028, loss_box_dn_2: 0.6371, loss_cls_dn_3: 0.1032, loss_box_dn_3: 0.6464, loss_cls_dn_4: 0.1054, loss_box_dn_4: 0.6444, loss_cls_dn_5: 0.1058, loss_box_dn_5: 0.6476, loss_dense_depth: 0.7055, loss: 23.7393, grad_norm: 42.9348 -2026-01-14 21:40:33,084 - mmdet - INFO - Iter [329/17500] lr: 2.310e-04, eta: 9:43:42, time: 1.598, data_time: 0.075, memory: 49164, loss_cls_0: 0.7170, loss_box_0: 1.6187, loss_cns_0: 0.6380, loss_yns_0: 0.1426, loss_cls_1: 0.7952, loss_box_1: 1.5004, loss_cns_1: 0.6662, loss_yns_1: 0.1412, loss_cls_2: 0.8003, loss_box_2: 1.4731, loss_cns_2: 0.6656, loss_yns_2: 0.1404, loss_cls_3: 0.8104, loss_box_3: 1.4622, loss_cns_3: 0.6660, loss_yns_3: 0.1398, loss_cls_4: 0.8190, loss_box_4: 1.4547, loss_cns_4: 0.6687, loss_yns_4: 0.1409, loss_cls_5: 0.8185, loss_box_5: 1.4492, loss_cns_5: 0.6665, loss_yns_5: 0.1395, loss_cls_dn_0: 0.1362, loss_box_dn_0: 0.7187, loss_cls_dn_1: 0.1045, loss_box_dn_1: 0.6566, loss_cls_dn_2: 0.1029, loss_box_dn_2: 0.6526, loss_cls_dn_3: 0.1022, loss_box_dn_3: 0.6553, loss_cls_dn_4: 0.1062, loss_box_dn_4: 0.6505, loss_cls_dn_5: 0.1071, loss_box_dn_5: 0.6568, loss_dense_depth: 0.7050, loss: 23.8887, grad_norm: 31.8651 -2026-01-14 21:40:34,683 - mmdet - INFO - Iter [330/17500] lr: 2.314e-04, eta: 9:43:16, time: 1.568, data_time: 0.084, memory: 49164, loss_cls_0: 0.7074, loss_box_0: 1.6124, loss_cns_0: 0.6398, loss_yns_0: 0.1424, loss_cls_1: 0.7893, loss_box_1: 1.4865, loss_cns_1: 0.6661, loss_yns_1: 0.1406, loss_cls_2: 0.7861, loss_box_2: 1.4659, loss_cns_2: 0.6674, loss_yns_2: 0.1408, loss_cls_3: 0.8085, loss_box_3: 1.4471, loss_cns_3: 0.6662, loss_yns_3: 0.1381, loss_cls_4: 0.8164, loss_box_4: 1.4547, loss_cns_4: 0.6685, loss_yns_4: 0.1396, loss_cls_5: 0.8004, loss_box_5: 1.4402, loss_cns_5: 0.6678, loss_yns_5: 0.1389, loss_cls_dn_0: 0.1392, loss_box_dn_0: 0.7272, loss_cls_dn_1: 0.1026, loss_box_dn_1: 0.6615, loss_cls_dn_2: 0.1014, loss_box_dn_2: 0.6547, loss_cls_dn_3: 0.1027, loss_box_dn_3: 0.6490, loss_cls_dn_4: 0.1050, loss_box_dn_4: 0.6494, loss_cls_dn_5: 0.1058, loss_box_dn_5: 0.6495, loss_dense_depth: 0.6882, loss: 23.7673, grad_norm: 47.5990 -2026-01-14 21:40:36,316 - mmdet - INFO - Iter [331/17500] lr: 2.318e-04, eta: 9:42:54, time: 1.662, data_time: 0.093, memory: 49164, loss_cls_0: 0.7200, loss_box_0: 1.6161, loss_cns_0: 0.6339, loss_yns_0: 0.1440, loss_cls_1: 0.7956, loss_box_1: 1.4982, loss_cns_1: 0.6606, loss_yns_1: 0.1414, loss_cls_2: 0.7997, loss_box_2: 1.4675, loss_cns_2: 0.6643, loss_yns_2: 0.1431, loss_cls_3: 0.8049, loss_box_3: 1.4566, loss_cns_3: 0.6641, loss_yns_3: 0.1410, loss_cls_4: 0.8236, loss_box_4: 1.4619, loss_cns_4: 0.6623, loss_yns_4: 0.1403, loss_cls_5: 0.8194, loss_box_5: 1.4469, loss_cns_5: 0.6622, loss_yns_5: 0.1409, loss_cls_dn_0: 0.1415, loss_box_dn_0: 0.7220, loss_cls_dn_1: 0.1047, loss_box_dn_1: 0.6491, loss_cls_dn_2: 0.1053, loss_box_dn_2: 0.6305, loss_cls_dn_3: 0.1070, loss_box_dn_3: 0.6232, loss_cls_dn_4: 0.1086, loss_box_dn_4: 0.6287, loss_cls_dn_5: 0.1108, loss_box_dn_5: 0.6260, loss_dense_depth: 0.6942, loss: 23.7601, grad_norm: 43.0546 -2026-01-14 21:40:37,890 - mmdet - INFO - Iter [332/17500] lr: 2.322e-04, eta: 9:42:28, time: 1.575, data_time: 0.080, memory: 49164, loss_cls_0: 0.7237, loss_box_0: 1.6109, loss_cns_0: 0.6378, loss_yns_0: 0.1452, loss_cls_1: 0.7832, loss_box_1: 1.4620, loss_cns_1: 0.6656, loss_yns_1: 0.1423, loss_cls_2: 0.8073, loss_box_2: 1.4434, loss_cns_2: 0.6661, loss_yns_2: 0.1420, loss_cls_3: 0.8202, loss_box_3: 1.4567, loss_cns_3: 0.6680, loss_yns_3: 0.1428, loss_cls_4: 0.8130, loss_box_4: 1.4434, loss_cns_4: 0.6703, loss_yns_4: 0.1434, loss_cls_5: 0.8059, loss_box_5: 1.4191, loss_cns_5: 0.6678, loss_yns_5: 0.1419, loss_cls_dn_0: 0.1376, loss_box_dn_0: 0.7340, loss_cls_dn_1: 0.1043, loss_box_dn_1: 0.6450, loss_cls_dn_2: 0.1026, loss_box_dn_2: 0.6314, loss_cls_dn_3: 0.1045, loss_box_dn_3: 0.6378, loss_cls_dn_4: 0.1067, loss_box_dn_4: 0.6342, loss_cls_dn_5: 0.1072, loss_box_dn_5: 0.6315, loss_dense_depth: 0.6826, loss: 23.6812, grad_norm: 48.9920 -2026-01-14 21:40:39,460 - mmdet - INFO - Iter [333/17500] lr: 2.326e-04, eta: 9:42:02, time: 1.571, data_time: 0.078, memory: 49164, loss_cls_0: 0.7367, loss_box_0: 1.6054, loss_cns_0: 0.6371, loss_yns_0: 0.1438, loss_cls_1: 0.8061, loss_box_1: 1.4798, loss_cns_1: 0.6645, loss_yns_1: 0.1416, loss_cls_2: 0.8240, loss_box_2: 1.4605, loss_cns_2: 0.6655, loss_yns_2: 0.1418, loss_cls_3: 0.8394, loss_box_3: 1.4726, loss_cns_3: 0.6662, loss_yns_3: 0.1412, loss_cls_4: 0.8414, loss_box_4: 1.4706, loss_cns_4: 0.6653, loss_yns_4: 0.1433, loss_cls_5: 0.8273, loss_box_5: 1.4481, loss_cns_5: 0.6624, loss_yns_5: 0.1409, loss_cls_dn_0: 0.1447, loss_box_dn_0: 0.7246, loss_cls_dn_1: 0.1041, loss_box_dn_1: 0.6515, loss_cls_dn_2: 0.1045, loss_box_dn_2: 0.6465, loss_cls_dn_3: 0.1105, loss_box_dn_3: 0.6575, loss_cls_dn_4: 0.1088, loss_box_dn_4: 0.6590, loss_cls_dn_5: 0.1073, loss_box_dn_5: 0.6650, loss_dense_depth: 0.7301, loss: 24.0393, grad_norm: 55.2659 -2026-01-14 21:40:41,032 - mmdet - INFO - Iter [334/17500] lr: 2.330e-04, eta: 9:41:36, time: 1.572, data_time: 0.077, memory: 49164, loss_cls_0: 0.7439, loss_box_0: 1.6256, loss_cns_0: 0.6262, loss_yns_0: 0.1444, loss_cls_1: 0.8298, loss_box_1: 1.4877, loss_cns_1: 0.6596, loss_yns_1: 0.1415, loss_cls_2: 0.8289, loss_box_2: 1.4988, loss_cns_2: 0.6637, loss_yns_2: 0.1446, loss_cls_3: 0.8362, loss_box_3: 1.4917, loss_cns_3: 0.6638, loss_yns_3: 0.1433, loss_cls_4: 0.8421, loss_box_4: 1.5100, loss_cns_4: 0.6677, loss_yns_4: 0.1450, loss_cls_5: 0.8594, loss_box_5: 1.4985, loss_cns_5: 0.6625, loss_yns_5: 0.1437, loss_cls_dn_0: 0.1404, loss_box_dn_0: 0.7299, loss_cls_dn_1: 0.1120, loss_box_dn_1: 0.6765, loss_cls_dn_2: 0.1109, loss_box_dn_2: 0.6740, loss_cls_dn_3: 0.1153, loss_box_dn_3: 0.6772, loss_cls_dn_4: 0.1115, loss_box_dn_4: 0.6911, loss_cls_dn_5: 0.1144, loss_box_dn_5: 0.6982, loss_dense_depth: 0.7241, loss: 24.4342, grad_norm: 61.3368 -2026-01-14 21:40:42,595 - mmdet - INFO - Iter [335/17500] lr: 2.334e-04, eta: 9:41:10, time: 1.563, data_time: 0.074, memory: 49164, loss_cls_0: 0.7336, loss_box_0: 1.6214, loss_cns_0: 0.6330, loss_yns_0: 0.1447, loss_cls_1: 0.8091, loss_box_1: 1.4872, loss_cns_1: 0.6601, loss_yns_1: 0.1425, loss_cls_2: 0.8147, loss_box_2: 1.4892, loss_cns_2: 0.6642, loss_yns_2: 0.1431, loss_cls_3: 0.8124, loss_box_3: 1.4778, loss_cns_3: 0.6647, loss_yns_3: 0.1425, loss_cls_4: 0.8136, loss_box_4: 1.4995, loss_cns_4: 0.6731, loss_yns_4: 0.1420, loss_cls_5: 0.8234, loss_box_5: 1.4805, loss_cns_5: 0.6681, loss_yns_5: 0.1418, loss_cls_dn_0: 0.1394, loss_box_dn_0: 0.7317, loss_cls_dn_1: 0.1087, loss_box_dn_1: 0.6824, loss_cls_dn_2: 0.1046, loss_box_dn_2: 0.6754, loss_cls_dn_3: 0.1056, loss_box_dn_3: 0.6748, loss_cls_dn_4: 0.1048, loss_box_dn_4: 0.6909, loss_cls_dn_5: 0.1100, loss_box_dn_5: 0.6901, loss_dense_depth: 0.6929, loss: 24.1936, grad_norm: 46.7175 -2026-01-14 21:40:44,205 - mmdet - INFO - Iter [336/17500] lr: 2.338e-04, eta: 9:40:44, time: 1.560, data_time: 0.073, memory: 49164, loss_cls_0: 0.7238, loss_box_0: 1.6256, loss_cns_0: 0.6314, loss_yns_0: 0.1425, loss_cls_1: 0.8009, loss_box_1: 1.5546, loss_cns_1: 0.6586, loss_yns_1: 0.1404, loss_cls_2: 0.8270, loss_box_2: 1.5038, loss_cns_2: 0.6636, loss_yns_2: 0.1413, loss_cls_3: 0.8342, loss_box_3: 1.5212, loss_cns_3: 0.6644, loss_yns_3: 0.1414, loss_cls_4: 0.8316, loss_box_4: 1.5309, loss_cns_4: 0.6665, loss_yns_4: 0.1423, loss_cls_5: 0.8235, loss_box_5: 1.5273, loss_cns_5: 0.6633, loss_yns_5: 0.1408, loss_cls_dn_0: 0.1370, loss_box_dn_0: 0.7268, loss_cls_dn_1: 0.1047, loss_box_dn_1: 0.6796, loss_cls_dn_2: 0.1041, loss_box_dn_2: 0.6710, loss_cls_dn_3: 0.1054, loss_box_dn_3: 0.6785, loss_cls_dn_4: 0.1081, loss_box_dn_4: 0.6800, loss_cls_dn_5: 0.1109, loss_box_dn_5: 0.6880, loss_dense_depth: 0.7010, loss: 24.3958, grad_norm: 57.2196 -2026-01-14 21:40:45,778 - mmdet - INFO - Iter [337/17500] lr: 2.342e-04, eta: 9:40:21, time: 1.623, data_time: 0.106, memory: 49164, loss_cls_0: 0.7273, loss_box_0: 1.6604, loss_cns_0: 0.6286, loss_yns_0: 0.1420, loss_cls_1: 0.8074, loss_box_1: 1.6186, loss_cns_1: 0.6577, loss_yns_1: 0.1409, loss_cls_2: 0.8285, loss_box_2: 1.5335, loss_cns_2: 0.6634, loss_yns_2: 0.1396, loss_cls_3: 0.8349, loss_box_3: 1.5363, loss_cns_3: 0.6618, loss_yns_3: 0.1404, loss_cls_4: 0.8292, loss_box_4: 1.5343, loss_cns_4: 0.6592, loss_yns_4: 0.1436, loss_cls_5: 0.8355, loss_box_5: 1.5443, loss_cns_5: 0.6605, loss_yns_5: 0.1419, loss_cls_dn_0: 0.1424, loss_box_dn_0: 0.7253, loss_cls_dn_1: 0.1089, loss_box_dn_1: 0.6801, loss_cls_dn_2: 0.1131, loss_box_dn_2: 0.6622, loss_cls_dn_3: 0.1158, loss_box_dn_3: 0.6662, loss_cls_dn_4: 0.1201, loss_box_dn_4: 0.6619, loss_cls_dn_5: 0.1199, loss_box_dn_5: 0.6698, loss_dense_depth: 0.7010, loss: 24.5564, grad_norm: 67.2095 -2026-01-14 21:40:47,342 - mmdet - INFO - Iter [338/17500] lr: 2.346e-04, eta: 9:39:56, time: 1.562, data_time: 0.074, memory: 49164, loss_cls_0: 0.7281, loss_box_0: 1.5997, loss_cns_0: 0.6374, loss_yns_0: 0.1415, loss_cls_1: 0.8090, loss_box_1: 1.5387, loss_cns_1: 0.6595, loss_yns_1: 0.1396, loss_cls_2: 0.8140, loss_box_2: 1.5105, loss_cns_2: 0.6608, loss_yns_2: 0.1397, loss_cls_3: 0.8097, loss_box_3: 1.5010, loss_cns_3: 0.6585, loss_yns_3: 0.1409, loss_cls_4: 0.8255, loss_box_4: 1.5042, loss_cns_4: 0.6584, loss_yns_4: 0.1436, loss_cls_5: 0.8293, loss_box_5: 1.4947, loss_cns_5: 0.6602, loss_yns_5: 0.1421, loss_cls_dn_0: 0.1404, loss_box_dn_0: 0.7243, loss_cls_dn_1: 0.1124, loss_box_dn_1: 0.6596, loss_cls_dn_2: 0.1147, loss_box_dn_2: 0.6526, loss_cls_dn_3: 0.1180, loss_box_dn_3: 0.6562, loss_cls_dn_4: 0.1229, loss_box_dn_4: 0.6568, loss_cls_dn_5: 0.1193, loss_box_dn_5: 0.6519, loss_dense_depth: 0.7109, loss: 24.1865, grad_norm: 56.8673 -2026-01-14 21:40:48,921 - mmdet - INFO - Iter [339/17500] lr: 2.350e-04, eta: 9:39:31, time: 1.581, data_time: 0.086, memory: 49164, loss_cls_0: 0.7214, loss_box_0: 1.6075, loss_cns_0: 0.6319, loss_yns_0: 0.1401, loss_cls_1: 0.8066, loss_box_1: 1.4978, loss_cns_1: 0.6630, loss_yns_1: 0.1387, loss_cls_2: 0.8123, loss_box_2: 1.4909, loss_cns_2: 0.6630, loss_yns_2: 0.1386, loss_cls_3: 0.8144, loss_box_3: 1.4822, loss_cns_3: 0.6610, loss_yns_3: 0.1403, loss_cls_4: 0.8197, loss_box_4: 1.5024, loss_cns_4: 0.6625, loss_yns_4: 0.1430, loss_cls_5: 0.8304, loss_box_5: 1.4727, loss_cns_5: 0.6641, loss_yns_5: 0.1417, loss_cls_dn_0: 0.1400, loss_box_dn_0: 0.7268, loss_cls_dn_1: 0.1090, loss_box_dn_1: 0.6571, loss_cls_dn_2: 0.1101, loss_box_dn_2: 0.6522, loss_cls_dn_3: 0.1102, loss_box_dn_3: 0.6548, loss_cls_dn_4: 0.1114, loss_box_dn_4: 0.6660, loss_cls_dn_5: 0.1108, loss_box_dn_5: 0.6604, loss_dense_depth: 0.7004, loss: 24.0549, grad_norm: 44.1787 -2026-01-14 21:40:50,494 - mmdet - INFO - Iter [340/17500] lr: 2.354e-04, eta: 9:39:06, time: 1.572, data_time: 0.081, memory: 49164, loss_cls_0: 0.7521, loss_box_0: 1.6020, loss_cns_0: 0.6311, loss_yns_0: 0.1415, loss_cls_1: 0.8115, loss_box_1: 1.5056, loss_cns_1: 0.6626, loss_yns_1: 0.1391, loss_cls_2: 0.8287, loss_box_2: 1.4772, loss_cns_2: 0.6675, loss_yns_2: 0.1404, loss_cls_3: 0.8371, loss_box_3: 1.4686, loss_cns_3: 0.6672, loss_yns_3: 0.1406, loss_cls_4: 0.8345, loss_box_4: 1.4886, loss_cns_4: 0.6741, loss_yns_4: 0.1424, loss_cls_5: 0.8530, loss_box_5: 1.4880, loss_cns_5: 0.6771, loss_yns_5: 0.1432, loss_cls_dn_0: 0.1368, loss_box_dn_0: 0.7285, loss_cls_dn_1: 0.1109, loss_box_dn_1: 0.6687, loss_cls_dn_2: 0.1147, loss_box_dn_2: 0.6580, loss_cls_dn_3: 0.1145, loss_box_dn_3: 0.6619, loss_cls_dn_4: 0.1144, loss_box_dn_4: 0.6668, loss_cls_dn_5: 0.1223, loss_box_dn_5: 0.6698, loss_dense_depth: 0.7079, loss: 24.2491, grad_norm: 51.0514 -2026-01-14 21:40:52,195 - mmdet - INFO - Iter [341/17500] lr: 2.358e-04, eta: 9:38:48, time: 1.700, data_time: 0.116, memory: 49164, loss_cls_0: 0.7454, loss_box_0: 1.5934, loss_cns_0: 0.6336, loss_yns_0: 0.1409, loss_cls_1: 0.8019, loss_box_1: 1.5303, loss_cns_1: 0.6586, loss_yns_1: 0.1401, loss_cls_2: 0.8131, loss_box_2: 1.4979, loss_cns_2: 0.6557, loss_yns_2: 0.1386, loss_cls_3: 0.8178, loss_box_3: 1.4935, loss_cns_3: 0.6591, loss_yns_3: 0.1382, loss_cls_4: 0.8227, loss_box_4: 1.4910, loss_cns_4: 0.6635, loss_yns_4: 0.1392, loss_cls_5: 0.8367, loss_box_5: 1.5021, loss_cns_5: 0.6622, loss_yns_5: 0.1422, loss_cls_dn_0: 0.1388, loss_box_dn_0: 0.7238, loss_cls_dn_1: 0.1081, loss_box_dn_1: 0.6663, loss_cls_dn_2: 0.1096, loss_box_dn_2: 0.6575, loss_cls_dn_3: 0.1100, loss_box_dn_3: 0.6550, loss_cls_dn_4: 0.1113, loss_box_dn_4: 0.6536, loss_cls_dn_5: 0.1149, loss_box_dn_5: 0.6583, loss_dense_depth: 0.7275, loss: 24.1524, grad_norm: 40.7114 -2026-01-14 21:40:53,855 - mmdet - INFO - Iter [342/17500] lr: 2.362e-04, eta: 9:38:28, time: 1.662, data_time: 0.170, memory: 49164, loss_cls_0: 0.7359, loss_box_0: 1.5978, loss_cns_0: 0.6335, loss_yns_0: 0.1435, loss_cls_1: 0.8119, loss_box_1: 1.5539, loss_cns_1: 0.6608, loss_yns_1: 0.1451, loss_cls_2: 0.8240, loss_box_2: 1.5181, loss_cns_2: 0.6582, loss_yns_2: 0.1427, loss_cls_3: 0.8289, loss_box_3: 1.5195, loss_cns_3: 0.6591, loss_yns_3: 0.1429, loss_cls_4: 0.8439, loss_box_4: 1.5179, loss_cns_4: 0.6585, loss_yns_4: 0.1494, loss_cls_5: 0.8388, loss_box_5: 1.5092, loss_cns_5: 0.6631, loss_yns_5: 0.1457, loss_cls_dn_0: 0.1333, loss_box_dn_0: 0.7212, loss_cls_dn_1: 0.1082, loss_box_dn_1: 0.6553, loss_cls_dn_2: 0.1057, loss_box_dn_2: 0.6391, loss_cls_dn_3: 0.1090, loss_box_dn_3: 0.6406, loss_cls_dn_4: 0.1068, loss_box_dn_4: 0.6458, loss_cls_dn_5: 0.1064, loss_box_dn_5: 0.6431, loss_dense_depth: 0.7153, loss: 24.2321, grad_norm: 43.8716 -2026-01-14 21:40:55,470 - mmdet - INFO - Iter [343/17500] lr: 2.366e-04, eta: 9:38:03, time: 1.564, data_time: 0.071, memory: 49164, loss_cls_0: 0.7454, loss_box_0: 1.6349, loss_cns_0: 0.6338, loss_yns_0: 0.1424, loss_cls_1: 0.8108, loss_box_1: 1.5859, loss_cns_1: 0.6581, loss_yns_1: 0.1401, loss_cls_2: 0.8193, loss_box_2: 1.5230, loss_cns_2: 0.6614, loss_yns_2: 0.1388, loss_cls_3: 0.8183, loss_box_3: 1.5250, loss_cns_3: 0.6614, loss_yns_3: 0.1393, loss_cls_4: 0.8345, loss_box_4: 1.5368, loss_cns_4: 0.6618, loss_yns_4: 0.1451, loss_cls_5: 0.8316, loss_box_5: 1.5288, loss_cns_5: 0.6664, loss_yns_5: 0.1414, loss_cls_dn_0: 0.1330, loss_box_dn_0: 0.7189, loss_cls_dn_1: 0.1060, loss_box_dn_1: 0.6682, loss_cls_dn_2: 0.1044, loss_box_dn_2: 0.6418, loss_cls_dn_3: 0.1057, loss_box_dn_3: 0.6461, loss_cls_dn_4: 0.1059, loss_box_dn_4: 0.6571, loss_cls_dn_5: 0.1061, loss_box_dn_5: 0.6538, loss_dense_depth: 0.7078, loss: 24.3390, grad_norm: 38.1685 -2026-01-14 21:40:57,029 - mmdet - INFO - Iter [344/17500] lr: 2.370e-04, eta: 9:37:40, time: 1.609, data_time: 0.109, memory: 49164, loss_cls_0: 0.7681, loss_box_0: 1.6574, loss_cns_0: 0.6294, loss_yns_0: 0.1423, loss_cls_1: 0.8326, loss_box_1: 1.5426, loss_cns_1: 0.6564, loss_yns_1: 0.1370, loss_cls_2: 0.8250, loss_box_2: 1.5302, loss_cns_2: 0.6588, loss_yns_2: 0.1371, loss_cls_3: 0.8330, loss_box_3: 1.5050, loss_cns_3: 0.6584, loss_yns_3: 0.1363, loss_cls_4: 0.8446, loss_box_4: 1.5261, loss_cns_4: 0.6619, loss_yns_4: 0.1384, loss_cls_5: 0.8473, loss_box_5: 1.4972, loss_cns_5: 0.6604, loss_yns_5: 0.1382, loss_cls_dn_0: 0.1377, loss_box_dn_0: 0.7292, loss_cls_dn_1: 0.1028, loss_box_dn_1: 0.6557, loss_cls_dn_2: 0.1039, loss_box_dn_2: 0.6525, loss_cls_dn_3: 0.1048, loss_box_dn_3: 0.6440, loss_cls_dn_4: 0.1062, loss_box_dn_4: 0.6499, loss_cls_dn_5: 0.1075, loss_box_dn_5: 0.6398, loss_dense_depth: 0.7273, loss: 24.3252, grad_norm: 42.3604 -2026-01-14 21:40:58,671 - mmdet - INFO - Iter [345/17500] lr: 2.374e-04, eta: 9:37:19, time: 1.639, data_time: 0.074, memory: 49164, loss_cls_0: 0.7250, loss_box_0: 1.6302, loss_cns_0: 0.6351, loss_yns_0: 0.1388, loss_cls_1: 0.7967, loss_box_1: 1.5186, loss_cns_1: 0.6582, loss_yns_1: 0.1375, loss_cls_2: 0.7992, loss_box_2: 1.5236, loss_cns_2: 0.6584, loss_yns_2: 0.1380, loss_cls_3: 0.8046, loss_box_3: 1.4964, loss_cns_3: 0.6632, loss_yns_3: 0.1372, loss_cls_4: 0.8229, loss_box_4: 1.4916, loss_cns_4: 0.6617, loss_yns_4: 0.1371, loss_cls_5: 0.8216, loss_box_5: 1.4796, loss_cns_5: 0.6612, loss_yns_5: 0.1374, loss_cls_dn_0: 0.1362, loss_box_dn_0: 0.7134, loss_cls_dn_1: 0.1025, loss_box_dn_1: 0.6392, loss_cls_dn_2: 0.1043, loss_box_dn_2: 0.6426, loss_cls_dn_3: 0.1058, loss_box_dn_3: 0.6354, loss_cls_dn_4: 0.1063, loss_box_dn_4: 0.6342, loss_cls_dn_5: 0.1076, loss_box_dn_5: 0.6353, loss_dense_depth: 0.7009, loss: 23.9376, grad_norm: 40.4354 -2026-01-14 21:41:00,366 - mmdet - INFO - Iter [346/17500] lr: 2.378e-04, eta: 9:37:01, time: 1.696, data_time: 0.077, memory: 49164, loss_cls_0: 0.7408, loss_box_0: 1.6294, loss_cns_0: 0.6291, loss_yns_0: 0.1388, loss_cls_1: 0.8093, loss_box_1: 1.4852, loss_cns_1: 0.6672, loss_yns_1: 0.1359, loss_cls_2: 0.8245, loss_box_2: 1.4701, loss_cns_2: 0.6682, loss_yns_2: 0.1368, loss_cls_3: 0.8333, loss_box_3: 1.4644, loss_cns_3: 0.6684, loss_yns_3: 0.1356, loss_cls_4: 0.8452, loss_box_4: 1.4676, loss_cns_4: 0.6631, loss_yns_4: 0.1359, loss_cls_5: 0.8430, loss_box_5: 1.4918, loss_cns_5: 0.6600, loss_yns_5: 0.1352, loss_cls_dn_0: 0.1359, loss_box_dn_0: 0.7227, loss_cls_dn_1: 0.1052, loss_box_dn_1: 0.6491, loss_cls_dn_2: 0.1061, loss_box_dn_2: 0.6409, loss_cls_dn_3: 0.1086, loss_box_dn_3: 0.6408, loss_cls_dn_4: 0.1085, loss_box_dn_4: 0.6485, loss_cls_dn_5: 0.1121, loss_box_dn_5: 0.6618, loss_dense_depth: 0.7052, loss: 24.0242, grad_norm: 40.3172 -2026-01-14 21:41:01,935 - mmdet - INFO - Iter [347/17500] lr: 2.382e-04, eta: 9:36:36, time: 1.562, data_time: 0.075, memory: 49164, loss_cls_0: 0.7130, loss_box_0: 1.5889, loss_cns_0: 0.6339, loss_yns_0: 0.1403, loss_cls_1: 0.7890, loss_box_1: 1.4504, loss_cns_1: 0.6660, loss_yns_1: 0.1379, loss_cls_2: 0.7995, loss_box_2: 1.4326, loss_cns_2: 0.6731, loss_yns_2: 0.1373, loss_cls_3: 0.8141, loss_box_3: 1.4221, loss_cns_3: 0.6677, loss_yns_3: 0.1379, loss_cls_4: 0.8249, loss_box_4: 1.4145, loss_cns_4: 0.6650, loss_yns_4: 0.1373, loss_cls_5: 0.8234, loss_box_5: 1.4124, loss_cns_5: 0.6640, loss_yns_5: 0.1369, loss_cls_dn_0: 0.1294, loss_box_dn_0: 0.7083, loss_cls_dn_1: 0.0996, loss_box_dn_1: 0.6535, loss_cls_dn_2: 0.0982, loss_box_dn_2: 0.6482, loss_cls_dn_3: 0.1027, loss_box_dn_3: 0.6453, loss_cls_dn_4: 0.1039, loss_box_dn_4: 0.6495, loss_cls_dn_5: 0.1045, loss_box_dn_5: 0.6561, loss_dense_depth: 0.7030, loss: 23.5840, grad_norm: 31.3635 -2026-01-14 21:41:03,597 - mmdet - INFO - Iter [348/17500] lr: 2.386e-04, eta: 9:36:16, time: 1.642, data_time: 0.078, memory: 49164, loss_cls_0: 0.7320, loss_box_0: 1.5951, loss_cns_0: 0.6322, loss_yns_0: 0.1418, loss_cls_1: 0.7878, loss_box_1: 1.4751, loss_cns_1: 0.6596, loss_yns_1: 0.1399, loss_cls_2: 0.8116, loss_box_2: 1.4293, loss_cns_2: 0.6677, loss_yns_2: 0.1374, loss_cls_3: 0.8265, loss_box_3: 1.4172, loss_cns_3: 0.6634, loss_yns_3: 0.1375, loss_cls_4: 0.8395, loss_box_4: 1.3995, loss_cns_4: 0.6609, loss_yns_4: 0.1393, loss_cls_5: 0.8289, loss_box_5: 1.4030, loss_cns_5: 0.6606, loss_yns_5: 0.1381, loss_cls_dn_0: 0.1335, loss_box_dn_0: 0.7166, loss_cls_dn_1: 0.1007, loss_box_dn_1: 0.6605, loss_cls_dn_2: 0.1007, loss_box_dn_2: 0.6552, loss_cls_dn_3: 0.1058, loss_box_dn_3: 0.6537, loss_cls_dn_4: 0.1060, loss_box_dn_4: 0.6524, loss_cls_dn_5: 0.1031, loss_box_dn_5: 0.6580, loss_dense_depth: 0.7311, loss: 23.7013, grad_norm: 42.0543 -2026-01-14 21:41:05,176 - mmdet - INFO - Iter [349/17500] lr: 2.390e-04, eta: 9:35:54, time: 1.605, data_time: 0.095, memory: 49164, loss_cls_0: 0.7190, loss_box_0: 1.5964, loss_cns_0: 0.6316, loss_yns_0: 0.1409, loss_cls_1: 0.7710, loss_box_1: 1.4971, loss_cns_1: 0.6616, loss_yns_1: 0.1383, loss_cls_2: 0.7969, loss_box_2: 1.4242, loss_cns_2: 0.6652, loss_yns_2: 0.1376, loss_cls_3: 0.8066, loss_box_3: 1.4056, loss_cns_3: 0.6687, loss_yns_3: 0.1382, loss_cls_4: 0.8175, loss_box_4: 1.4018, loss_cns_4: 0.6669, loss_yns_4: 0.1380, loss_cls_5: 0.8259, loss_box_5: 1.3977, loss_cns_5: 0.6670, loss_yns_5: 0.1381, loss_cls_dn_0: 0.1374, loss_box_dn_0: 0.7124, loss_cls_dn_1: 0.0997, loss_box_dn_1: 0.6563, loss_cls_dn_2: 0.0997, loss_box_dn_2: 0.6402, loss_cls_dn_3: 0.1001, loss_box_dn_3: 0.6372, loss_cls_dn_4: 0.1039, loss_box_dn_4: 0.6355, loss_cls_dn_5: 0.1063, loss_box_dn_5: 0.6358, loss_dense_depth: 0.7120, loss: 23.5285, grad_norm: 31.0871 -2026-01-14 21:41:06,778 - mmdet - INFO - Iter [350/17500] lr: 2.394e-04, eta: 9:35:31, time: 1.596, data_time: 0.079, memory: 49164, loss_cls_0: 0.7093, loss_box_0: 1.5756, loss_cns_0: 0.6363, loss_yns_0: 0.1404, loss_cls_1: 0.7659, loss_box_1: 1.4315, loss_cns_1: 0.6661, loss_yns_1: 0.1381, loss_cls_2: 0.7984, loss_box_2: 1.4266, loss_cns_2: 0.6723, loss_yns_2: 0.1378, loss_cls_3: 0.8127, loss_box_3: 1.4233, loss_cns_3: 0.6778, loss_yns_3: 0.1382, loss_cls_4: 0.8203, loss_box_4: 1.4245, loss_cns_4: 0.6719, loss_yns_4: 0.1376, loss_cls_5: 0.8143, loss_box_5: 1.3957, loss_cns_5: 0.6681, loss_yns_5: 0.1378, loss_cls_dn_0: 0.1282, loss_box_dn_0: 0.7174, loss_cls_dn_1: 0.0991, loss_box_dn_1: 0.6424, loss_cls_dn_2: 0.1018, loss_box_dn_2: 0.6376, loss_cls_dn_3: 0.1005, loss_box_dn_3: 0.6357, loss_cls_dn_4: 0.1046, loss_box_dn_4: 0.6435, loss_cls_dn_5: 0.1050, loss_box_dn_5: 0.6361, loss_dense_depth: 0.7046, loss: 23.4770, grad_norm: 46.1965 -2026-01-14 21:41:08,332 - mmdet - INFO - Iter [351/17500] lr: 2.398e-04, eta: 9:35:07, time: 1.559, data_time: 0.081, memory: 49164, loss_cls_0: 0.7461, loss_box_0: 1.5917, loss_cns_0: 0.6324, loss_yns_0: 0.1433, loss_cls_1: 0.8019, loss_box_1: 1.4354, loss_cns_1: 0.6668, loss_yns_1: 0.1425, loss_cls_2: 0.8343, loss_box_2: 1.4341, loss_cns_2: 0.6719, loss_yns_2: 0.1424, loss_cls_3: 0.8556, loss_box_3: 1.4147, loss_cns_3: 0.6714, loss_yns_3: 0.1427, loss_cls_4: 0.8550, loss_box_4: 1.4143, loss_cns_4: 0.6670, loss_yns_4: 0.1432, loss_cls_5: 0.8446, loss_box_5: 1.3947, loss_cns_5: 0.6659, loss_yns_5: 0.1433, loss_cls_dn_0: 0.1292, loss_box_dn_0: 0.7089, loss_cls_dn_1: 0.1014, loss_box_dn_1: 0.6331, loss_cls_dn_2: 0.1029, loss_box_dn_2: 0.6316, loss_cls_dn_3: 0.1048, loss_box_dn_3: 0.6243, loss_cls_dn_4: 0.1081, loss_box_dn_4: 0.6315, loss_cls_dn_5: 0.1077, loss_box_dn_5: 0.6305, loss_dense_depth: 0.7229, loss: 23.6922, grad_norm: 41.6411 -2026-01-14 21:41:09,892 - mmdet - INFO - Iter [352/17500] lr: 2.402e-04, eta: 9:34:43, time: 1.563, data_time: 0.078, memory: 49164, loss_cls_0: 0.7378, loss_box_0: 1.5992, loss_cns_0: 0.6319, loss_yns_0: 0.1461, loss_cls_1: 0.8142, loss_box_1: 1.4448, loss_cns_1: 0.6607, loss_yns_1: 0.1450, loss_cls_2: 0.8289, loss_box_2: 1.4289, loss_cns_2: 0.6632, loss_yns_2: 0.1440, loss_cls_3: 0.8368, loss_box_3: 1.3974, loss_cns_3: 0.6620, loss_yns_3: 0.1437, loss_cls_4: 0.8418, loss_box_4: 1.3992, loss_cns_4: 0.6612, loss_yns_4: 0.1442, loss_cls_5: 0.8429, loss_box_5: 1.4126, loss_cns_5: 0.6615, loss_yns_5: 0.1441, loss_cls_dn_0: 0.1321, loss_box_dn_0: 0.7233, loss_cls_dn_1: 0.1020, loss_box_dn_1: 0.6451, loss_cls_dn_2: 0.1033, loss_box_dn_2: 0.6417, loss_cls_dn_3: 0.1044, loss_box_dn_3: 0.6311, loss_cls_dn_4: 0.1054, loss_box_dn_4: 0.6369, loss_cls_dn_5: 0.1075, loss_box_dn_5: 0.6453, loss_dense_depth: 0.7317, loss: 23.7020, grad_norm: 35.3407 -2026-01-14 21:41:11,499 - mmdet - INFO - Iter [353/17500] lr: 2.406e-04, eta: 9:34:20, time: 1.572, data_time: 0.074, memory: 49164, loss_cls_0: 0.7270, loss_box_0: 1.5779, loss_cns_0: 0.6367, loss_yns_0: 0.1454, loss_cls_1: 0.7948, loss_box_1: 1.3984, loss_cns_1: 0.6635, loss_yns_1: 0.1432, loss_cls_2: 0.8160, loss_box_2: 1.3839, loss_cns_2: 0.6642, loss_yns_2: 0.1429, loss_cls_3: 0.8210, loss_box_3: 1.3677, loss_cns_3: 0.6631, loss_yns_3: 0.1436, loss_cls_4: 0.8222, loss_box_4: 1.3773, loss_cns_4: 0.6649, loss_yns_4: 0.1427, loss_cls_5: 0.8273, loss_box_5: 1.3588, loss_cns_5: 0.6638, loss_yns_5: 0.1429, loss_cls_dn_0: 0.1356, loss_box_dn_0: 0.7182, loss_cls_dn_1: 0.1017, loss_box_dn_1: 0.6317, loss_cls_dn_2: 0.1051, loss_box_dn_2: 0.6227, loss_cls_dn_3: 0.1019, loss_box_dn_3: 0.6208, loss_cls_dn_4: 0.1049, loss_box_dn_4: 0.6274, loss_cls_dn_5: 0.1104, loss_box_dn_5: 0.6224, loss_dense_depth: 0.6880, loss: 23.2800, grad_norm: 32.3866 -2026-01-14 21:41:13,073 - mmdet - INFO - Iter [354/17500] lr: 2.410e-04, eta: 9:33:58, time: 1.608, data_time: 0.114, memory: 49164, loss_cls_0: 0.7571, loss_box_0: 1.6025, loss_cns_0: 0.6334, loss_yns_0: 0.1437, loss_cls_1: 0.8115, loss_box_1: 1.4422, loss_cns_1: 0.6578, loss_yns_1: 0.1415, loss_cls_2: 0.8171, loss_box_2: 1.3871, loss_cns_2: 0.6587, loss_yns_2: 0.1392, loss_cls_3: 0.8236, loss_box_3: 1.3791, loss_cns_3: 0.6578, loss_yns_3: 0.1392, loss_cls_4: 0.8299, loss_box_4: 1.3812, loss_cns_4: 0.6618, loss_yns_4: 0.1403, loss_cls_5: 0.8292, loss_box_5: 1.3745, loss_cns_5: 0.6594, loss_yns_5: 0.1408, loss_cls_dn_0: 0.1300, loss_box_dn_0: 0.7235, loss_cls_dn_1: 0.0990, loss_box_dn_1: 0.6347, loss_cls_dn_2: 0.0999, loss_box_dn_2: 0.6140, loss_cls_dn_3: 0.0977, loss_box_dn_3: 0.6136, loss_cls_dn_4: 0.0988, loss_box_dn_4: 0.6160, loss_cls_dn_5: 0.1017, loss_box_dn_5: 0.6173, loss_dense_depth: 0.7080, loss: 23.3626, grad_norm: 36.3437 -2026-01-14 21:41:14,694 - mmdet - INFO - Iter [355/17500] lr: 2.414e-04, eta: 9:33:36, time: 1.592, data_time: 0.073, memory: 49164, loss_cls_0: 0.7318, loss_box_0: 1.5766, loss_cns_0: 0.6324, loss_yns_0: 0.1415, loss_cls_1: 0.8218, loss_box_1: 1.4385, loss_cns_1: 0.6590, loss_yns_1: 0.1414, loss_cls_2: 0.8215, loss_box_2: 1.3971, loss_cns_2: 0.6588, loss_yns_2: 0.1408, loss_cls_3: 0.8244, loss_box_3: 1.4031, loss_cns_3: 0.6613, loss_yns_3: 0.1410, loss_cls_4: 0.8295, loss_box_4: 1.3966, loss_cns_4: 0.6615, loss_yns_4: 0.1411, loss_cls_5: 0.8234, loss_box_5: 1.3891, loss_cns_5: 0.6579, loss_yns_5: 0.1405, loss_cls_dn_0: 0.1338, loss_box_dn_0: 0.7166, loss_cls_dn_1: 0.0991, loss_box_dn_1: 0.6331, loss_cls_dn_2: 0.0972, loss_box_dn_2: 0.6163, loss_cls_dn_3: 0.0975, loss_box_dn_3: 0.6177, loss_cls_dn_4: 0.1000, loss_box_dn_4: 0.6189, loss_cls_dn_5: 0.1021, loss_box_dn_5: 0.6175, loss_dense_depth: 0.6918, loss: 23.3725, grad_norm: 30.7440 -2026-01-14 21:41:16,245 - mmdet - INFO - Iter [356/17500] lr: 2.418e-04, eta: 9:33:14, time: 1.579, data_time: 0.092, memory: 49164, loss_cls_0: 0.7515, loss_box_0: 1.5637, loss_cns_0: 0.6367, loss_yns_0: 0.1456, loss_cls_1: 0.8248, loss_box_1: 1.4449, loss_cns_1: 0.6646, loss_yns_1: 0.1446, loss_cls_2: 0.8278, loss_box_2: 1.4171, loss_cns_2: 0.6625, loss_yns_2: 0.1451, loss_cls_3: 0.8376, loss_box_3: 1.4041, loss_cns_3: 0.6654, loss_yns_3: 0.1446, loss_cls_4: 0.8438, loss_box_4: 1.4114, loss_cns_4: 0.6638, loss_yns_4: 0.1450, loss_cls_5: 0.8567, loss_box_5: 1.3878, loss_cns_5: 0.6609, loss_yns_5: 0.1458, loss_cls_dn_0: 0.1362, loss_box_dn_0: 0.7236, loss_cls_dn_1: 0.1007, loss_box_dn_1: 0.6168, loss_cls_dn_2: 0.0989, loss_box_dn_2: 0.6093, loss_cls_dn_3: 0.1014, loss_box_dn_3: 0.6080, loss_cls_dn_4: 0.1011, loss_box_dn_4: 0.6100, loss_cls_dn_5: 0.1070, loss_box_dn_5: 0.6049, loss_dense_depth: 0.7233, loss: 23.5374, grad_norm: 44.3098 -2026-01-14 21:41:17,813 - mmdet - INFO - Iter [357/17500] lr: 2.422e-04, eta: 9:32:51, time: 1.571, data_time: 0.079, memory: 49164, loss_cls_0: 0.7634, loss_box_0: 1.5724, loss_cns_0: 0.6300, loss_yns_0: 0.1422, loss_cls_1: 0.8331, loss_box_1: 1.5071, loss_cns_1: 0.6578, loss_yns_1: 0.1429, loss_cls_2: 0.8080, loss_box_2: 1.4652, loss_cns_2: 0.6599, loss_yns_2: 0.1421, loss_cls_3: 0.8268, loss_box_3: 1.4309, loss_cns_3: 0.6583, loss_yns_3: 0.1421, loss_cls_4: 0.8311, loss_box_4: 1.4282, loss_cns_4: 0.6589, loss_yns_4: 0.1417, loss_cls_5: 0.8404, loss_box_5: 1.4276, loss_cns_5: 0.6609, loss_yns_5: 0.1423, loss_cls_dn_0: 0.1361, loss_box_dn_0: 0.7215, loss_cls_dn_1: 0.0998, loss_box_dn_1: 0.6395, loss_cls_dn_2: 0.0974, loss_box_dn_2: 0.6261, loss_cls_dn_3: 0.1018, loss_box_dn_3: 0.6159, loss_cls_dn_4: 0.1013, loss_box_dn_4: 0.6114, loss_cls_dn_5: 0.1037, loss_box_dn_5: 0.6132, loss_dense_depth: 0.6871, loss: 23.6682, grad_norm: 35.2448 -2026-01-14 21:41:19,412 - mmdet - INFO - Iter [358/17500] lr: 2.426e-04, eta: 9:32:28, time: 1.577, data_time: 0.076, memory: 49164, loss_cls_0: 0.7373, loss_box_0: 1.5566, loss_cns_0: 0.6339, loss_yns_0: 0.1444, loss_cls_1: 0.7982, loss_box_1: 1.4807, loss_cns_1: 0.6551, loss_yns_1: 0.1413, loss_cls_2: 0.7940, loss_box_2: 1.4797, loss_cns_2: 0.6571, loss_yns_2: 0.1411, loss_cls_3: 0.8005, loss_box_3: 1.4643, loss_cns_3: 0.6557, loss_yns_3: 0.1416, loss_cls_4: 0.8076, loss_box_4: 1.4766, loss_cns_4: 0.6614, loss_yns_4: 0.1408, loss_cls_5: 0.8111, loss_box_5: 1.4967, loss_cns_5: 0.6582, loss_yns_5: 0.1406, loss_cls_dn_0: 0.1355, loss_box_dn_0: 0.7173, loss_cls_dn_1: 0.1022, loss_box_dn_1: 0.6275, loss_cls_dn_2: 0.0993, loss_box_dn_2: 0.6263, loss_cls_dn_3: 0.1009, loss_box_dn_3: 0.6225, loss_cls_dn_4: 0.1016, loss_box_dn_4: 0.6277, loss_cls_dn_5: 0.1046, loss_box_dn_5: 0.6392, loss_dense_depth: 0.6893, loss: 23.6687, grad_norm: 45.1597 -2026-01-14 21:41:21,120 - mmdet - INFO - Iter [359/17500] lr: 2.429e-04, eta: 9:32:11, time: 1.692, data_time: 0.221, memory: 49164, loss_cls_0: 0.7075, loss_box_0: 1.5397, loss_cns_0: 0.6416, loss_yns_0: 0.1379, loss_cls_1: 0.7692, loss_box_1: 1.4253, loss_cns_1: 0.6622, loss_yns_1: 0.1387, loss_cls_2: 0.7922, loss_box_2: 1.4177, loss_cns_2: 0.6634, loss_yns_2: 0.1380, loss_cls_3: 0.7875, loss_box_3: 1.3981, loss_cns_3: 0.6679, loss_yns_3: 0.1380, loss_cls_4: 0.7882, loss_box_4: 1.4030, loss_cns_4: 0.6669, loss_yns_4: 0.1375, loss_cls_5: 0.7927, loss_box_5: 1.4126, loss_cns_5: 0.6640, loss_yns_5: 0.1376, loss_cls_dn_0: 0.1340, loss_box_dn_0: 0.7191, loss_cls_dn_1: 0.1032, loss_box_dn_1: 0.6310, loss_cls_dn_2: 0.1034, loss_box_dn_2: 0.6270, loss_cls_dn_3: 0.1014, loss_box_dn_3: 0.6228, loss_cls_dn_4: 0.1039, loss_box_dn_4: 0.6283, loss_cls_dn_5: 0.1095, loss_box_dn_5: 0.6350, loss_dense_depth: 0.6719, loss: 23.2181, grad_norm: 41.1673 -2026-01-14 21:41:22,677 - mmdet - INFO - Iter [360/17500] lr: 2.433e-04, eta: 9:31:50, time: 1.592, data_time: 0.098, memory: 49164, loss_cls_0: 0.7402, loss_box_0: 1.5647, loss_cns_0: 0.6396, loss_yns_0: 0.1408, loss_cls_1: 0.7918, loss_box_1: 1.4628, loss_cns_1: 0.6636, loss_yns_1: 0.1400, loss_cls_2: 0.7984, loss_box_2: 1.4352, loss_cns_2: 0.6647, loss_yns_2: 0.1399, loss_cls_3: 0.7968, loss_box_3: 1.4271, loss_cns_3: 0.6689, loss_yns_3: 0.1393, loss_cls_4: 0.8016, loss_box_4: 1.4419, loss_cns_4: 0.6671, loss_yns_4: 0.1385, loss_cls_5: 0.8115, loss_box_5: 1.4550, loss_cns_5: 0.6669, loss_yns_5: 0.1389, loss_cls_dn_0: 0.1326, loss_box_dn_0: 0.7179, loss_cls_dn_1: 0.1029, loss_box_dn_1: 0.6356, loss_cls_dn_2: 0.1041, loss_box_dn_2: 0.6157, loss_cls_dn_3: 0.1021, loss_box_dn_3: 0.6164, loss_cls_dn_4: 0.1049, loss_box_dn_4: 0.6223, loss_cls_dn_5: 0.1069, loss_box_dn_5: 0.6306, loss_dense_depth: 0.6886, loss: 23.5156, grad_norm: 38.0310 -2026-01-14 21:41:24,324 - mmdet - INFO - Iter [361/17500] lr: 2.437e-04, eta: 9:31:31, time: 1.649, data_time: 0.082, memory: 49164, loss_cls_0: 0.7386, loss_box_0: 1.5537, loss_cns_0: 0.6406, loss_yns_0: 0.1446, loss_cls_1: 0.8071, loss_box_1: 1.4451, loss_cns_1: 0.6635, loss_yns_1: 0.1425, loss_cls_2: 0.7993, loss_box_2: 1.4202, loss_cns_2: 0.6649, loss_yns_2: 0.1422, loss_cls_3: 0.8099, loss_box_3: 1.4103, loss_cns_3: 0.6653, loss_yns_3: 0.1415, loss_cls_4: 0.8115, loss_box_4: 1.4430, loss_cns_4: 0.6661, loss_yns_4: 0.1406, loss_cls_5: 0.8229, loss_box_5: 1.4657, loss_cns_5: 0.6676, loss_yns_5: 0.1413, loss_cls_dn_0: 0.1306, loss_box_dn_0: 0.7130, loss_cls_dn_1: 0.1027, loss_box_dn_1: 0.6276, loss_cls_dn_2: 0.1026, loss_box_dn_2: 0.6103, loss_cls_dn_3: 0.1010, loss_box_dn_3: 0.6108, loss_cls_dn_4: 0.1025, loss_box_dn_4: 0.6205, loss_cls_dn_5: 0.1047, loss_box_dn_5: 0.6333, loss_dense_depth: 0.6896, loss: 23.4972, grad_norm: 33.8172 -2026-01-14 21:41:25,997 - mmdet - INFO - Iter [362/17500] lr: 2.441e-04, eta: 9:31:12, time: 1.654, data_time: 0.101, memory: 49164, loss_cls_0: 0.7462, loss_box_0: 1.5538, loss_cns_0: 0.6336, loss_yns_0: 0.1406, loss_cls_1: 0.8113, loss_box_1: 1.4182, loss_cns_1: 0.6574, loss_yns_1: 0.1410, loss_cls_2: 0.8206, loss_box_2: 1.3890, loss_cns_2: 0.6608, loss_yns_2: 0.1404, loss_cls_3: 0.8242, loss_box_3: 1.3875, loss_cns_3: 0.6602, loss_yns_3: 0.1407, loss_cls_4: 0.8244, loss_box_4: 1.3902, loss_cns_4: 0.6594, loss_yns_4: 0.1408, loss_cls_5: 0.8302, loss_box_5: 1.4021, loss_cns_5: 0.6602, loss_yns_5: 0.1406, loss_cls_dn_0: 0.1360, loss_box_dn_0: 0.7173, loss_cls_dn_1: 0.1015, loss_box_dn_1: 0.6205, loss_cls_dn_2: 0.1014, loss_box_dn_2: 0.6120, loss_cls_dn_3: 0.0984, loss_box_dn_3: 0.6113, loss_cls_dn_4: 0.0984, loss_box_dn_4: 0.6135, loss_cls_dn_5: 0.1018, loss_box_dn_5: 0.6211, loss_dense_depth: 0.7020, loss: 23.3089, grad_norm: 34.2122 -2026-01-14 21:41:27,582 - mmdet - INFO - Iter [363/17500] lr: 2.445e-04, eta: 9:30:52, time: 1.605, data_time: 0.086, memory: 49164, loss_cls_0: 0.7388, loss_box_0: 1.5605, loss_cns_0: 0.6386, loss_yns_0: 0.1391, loss_cls_1: 0.7992, loss_box_1: 1.4443, loss_cns_1: 0.6595, loss_yns_1: 0.1404, loss_cls_2: 0.8132, loss_box_2: 1.4012, loss_cns_2: 0.6623, loss_yns_2: 0.1403, loss_cls_3: 0.8242, loss_box_3: 1.3850, loss_cns_3: 0.6646, loss_yns_3: 0.1406, loss_cls_4: 0.8135, loss_box_4: 1.4010, loss_cns_4: 0.6641, loss_yns_4: 0.1392, loss_cls_5: 0.8203, loss_box_5: 1.3967, loss_cns_5: 0.6620, loss_yns_5: 0.1403, loss_cls_dn_0: 0.1285, loss_box_dn_0: 0.7253, loss_cls_dn_1: 0.0961, loss_box_dn_1: 0.6383, loss_cls_dn_2: 0.0962, loss_box_dn_2: 0.6229, loss_cls_dn_3: 0.0950, loss_box_dn_3: 0.6236, loss_cls_dn_4: 0.0933, loss_box_dn_4: 0.6297, loss_cls_dn_5: 0.0957, loss_box_dn_5: 0.6338, loss_dense_depth: 0.6950, loss: 23.3620, grad_norm: 31.9364 -2026-01-14 21:41:29,202 - mmdet - INFO - Iter [364/17500] lr: 2.449e-04, eta: 9:30:32, time: 1.620, data_time: 0.079, memory: 49164, loss_cls_0: 0.7374, loss_box_0: 1.5745, loss_cns_0: 0.6407, loss_yns_0: 0.1379, loss_cls_1: 0.7969, loss_box_1: 1.4665, loss_cns_1: 0.6597, loss_yns_1: 0.1403, loss_cls_2: 0.7998, loss_box_2: 1.4574, loss_cns_2: 0.6626, loss_yns_2: 0.1391, loss_cls_3: 0.8248, loss_box_3: 1.4140, loss_cns_3: 0.6611, loss_yns_3: 0.1373, loss_cls_4: 0.8122, loss_box_4: 1.4128, loss_cns_4: 0.6613, loss_yns_4: 0.1372, loss_cls_5: 0.8178, loss_box_5: 1.4165, loss_cns_5: 0.6611, loss_yns_5: 0.1384, loss_cls_dn_0: 0.1299, loss_box_dn_0: 0.7261, loss_cls_dn_1: 0.0963, loss_box_dn_1: 0.6447, loss_cls_dn_2: 0.0971, loss_box_dn_2: 0.6357, loss_cls_dn_3: 0.0985, loss_box_dn_3: 0.6260, loss_cls_dn_4: 0.0962, loss_box_dn_4: 0.6256, loss_cls_dn_5: 0.0982, loss_box_dn_5: 0.6299, loss_dense_depth: 0.6804, loss: 23.4923, grad_norm: 36.3013 -2026-01-14 21:41:30,842 - mmdet - INFO - Iter [365/17500] lr: 2.453e-04, eta: 9:30:13, time: 1.638, data_time: 0.079, memory: 49164, loss_cls_0: 0.7328, loss_box_0: 1.5863, loss_cns_0: 0.6411, loss_yns_0: 0.1375, loss_cls_1: 0.8065, loss_box_1: 1.4685, loss_cns_1: 0.6617, loss_yns_1: 0.1374, loss_cls_2: 0.8134, loss_box_2: 1.4425, loss_cns_2: 0.6660, loss_yns_2: 0.1375, loss_cls_3: 0.8209, loss_box_3: 1.4222, loss_cns_3: 0.6617, loss_yns_3: 0.1370, loss_cls_4: 0.8086, loss_box_4: 1.4191, loss_cns_4: 0.6623, loss_yns_4: 0.1381, loss_cls_5: 0.8166, loss_box_5: 1.4348, loss_cns_5: 0.6638, loss_yns_5: 0.1383, loss_cls_dn_0: 0.1270, loss_box_dn_0: 0.7150, loss_cls_dn_1: 0.0992, loss_box_dn_1: 0.6341, loss_cls_dn_2: 0.0979, loss_box_dn_2: 0.6192, loss_cls_dn_3: 0.0992, loss_box_dn_3: 0.6104, loss_cls_dn_4: 0.0983, loss_box_dn_4: 0.6082, loss_cls_dn_5: 0.1003, loss_box_dn_5: 0.6176, loss_dense_depth: 0.6763, loss: 23.4571, grad_norm: 24.1420 -2026-01-14 21:41:32,524 - mmdet - INFO - Iter [366/17500] lr: 2.457e-04, eta: 9:29:55, time: 1.648, data_time: 0.079, memory: 49164, loss_cls_0: 0.6946, loss_box_0: 1.5592, loss_cns_0: 0.6393, loss_yns_0: 0.1361, loss_cls_1: 0.7627, loss_box_1: 1.4329, loss_cns_1: 0.6624, loss_yns_1: 0.1341, loss_cls_2: 0.7683, loss_box_2: 1.3971, loss_cns_2: 0.6606, loss_yns_2: 0.1349, loss_cls_3: 0.7653, loss_box_3: 1.3872, loss_cns_3: 0.6588, loss_yns_3: 0.1354, loss_cls_4: 0.7656, loss_box_4: 1.3963, loss_cns_4: 0.6596, loss_yns_4: 0.1365, loss_cls_5: 0.7767, loss_box_5: 1.3945, loss_cns_5: 0.6623, loss_yns_5: 0.1380, loss_cls_dn_0: 0.1240, loss_box_dn_0: 0.7148, loss_cls_dn_1: 0.0945, loss_box_dn_1: 0.6298, loss_cls_dn_2: 0.0936, loss_box_dn_2: 0.6122, loss_cls_dn_3: 0.0929, loss_box_dn_3: 0.6058, loss_cls_dn_4: 0.0934, loss_box_dn_4: 0.6087, loss_cls_dn_5: 0.0950, loss_box_dn_5: 0.6104, loss_dense_depth: 0.6798, loss: 22.9133, grad_norm: 29.1623 -2026-01-14 21:41:34,077 - mmdet - INFO - Iter [367/17500] lr: 2.461e-04, eta: 9:29:34, time: 1.589, data_time: 0.101, memory: 49164, loss_cls_0: 0.7171, loss_box_0: 1.5672, loss_cns_0: 0.6378, loss_yns_0: 0.1377, loss_cls_1: 0.7783, loss_box_1: 1.4408, loss_cns_1: 0.6651, loss_yns_1: 0.1382, loss_cls_2: 0.7887, loss_box_2: 1.3964, loss_cns_2: 0.6617, loss_yns_2: 0.1363, loss_cls_3: 0.7914, loss_box_3: 1.4070, loss_cns_3: 0.6638, loss_yns_3: 0.1374, loss_cls_4: 0.8012, loss_box_4: 1.4091, loss_cns_4: 0.6628, loss_yns_4: 0.1361, loss_cls_5: 0.8071, loss_box_5: 1.4228, loss_cns_5: 0.6661, loss_yns_5: 0.1380, loss_cls_dn_0: 0.1311, loss_box_dn_0: 0.7143, loss_cls_dn_1: 0.0962, loss_box_dn_1: 0.6280, loss_cls_dn_2: 0.0963, loss_box_dn_2: 0.6133, loss_cls_dn_3: 0.0961, loss_box_dn_3: 0.6128, loss_cls_dn_4: 0.1001, loss_box_dn_4: 0.6153, loss_cls_dn_5: 0.0987, loss_box_dn_5: 0.6235, loss_dense_depth: 0.7144, loss: 23.2479, grad_norm: 34.1798 -2026-01-14 21:41:35,662 - mmdet - INFO - Iter [368/17500] lr: 2.465e-04, eta: 9:29:13, time: 1.586, data_time: 0.074, memory: 49164, loss_cls_0: 0.7058, loss_box_0: 1.5856, loss_cns_0: 0.6353, loss_yns_0: 0.1360, loss_cls_1: 0.7708, loss_box_1: 1.4440, loss_cns_1: 0.6635, loss_yns_1: 0.1370, loss_cls_2: 0.7867, loss_box_2: 1.3964, loss_cns_2: 0.6603, loss_yns_2: 0.1326, loss_cls_3: 0.7852, loss_box_3: 1.3841, loss_cns_3: 0.6617, loss_yns_3: 0.1327, loss_cls_4: 0.7881, loss_box_4: 1.3900, loss_cns_4: 0.6642, loss_yns_4: 0.1337, loss_cls_5: 0.7928, loss_box_5: 1.3995, loss_cns_5: 0.6664, loss_yns_5: 0.1338, loss_cls_dn_0: 0.1269, loss_box_dn_0: 0.7150, loss_cls_dn_1: 0.0983, loss_box_dn_1: 0.6275, loss_cls_dn_2: 0.0969, loss_box_dn_2: 0.6163, loss_cls_dn_3: 0.0976, loss_box_dn_3: 0.6115, loss_cls_dn_4: 0.0990, loss_box_dn_4: 0.6141, loss_cls_dn_5: 0.1002, loss_box_dn_5: 0.6211, loss_dense_depth: 0.7490, loss: 23.1595, grad_norm: 27.8902 -2026-01-14 21:41:37,232 - mmdet - INFO - Iter [369/17500] lr: 2.469e-04, eta: 9:28:51, time: 1.568, data_time: 0.076, memory: 49164, loss_cls_0: 0.6961, loss_box_0: 1.5567, loss_cns_0: 0.6422, loss_yns_0: 0.1342, loss_cls_1: 0.7644, loss_box_1: 1.4250, loss_cns_1: 0.6674, loss_yns_1: 0.1299, loss_cls_2: 0.7837, loss_box_2: 1.3909, loss_cns_2: 0.6698, loss_yns_2: 0.1301, loss_cls_3: 0.7736, loss_box_3: 1.3673, loss_cns_3: 0.6675, loss_yns_3: 0.1316, loss_cls_4: 0.7884, loss_box_4: 1.3748, loss_cns_4: 0.6670, loss_yns_4: 0.1335, loss_cls_5: 0.7933, loss_box_5: 1.3750, loss_cns_5: 0.6677, loss_yns_5: 0.1329, loss_cls_dn_0: 0.1199, loss_box_dn_0: 0.7049, loss_cls_dn_1: 0.0968, loss_box_dn_1: 0.6281, loss_cls_dn_2: 0.0976, loss_box_dn_2: 0.6153, loss_cls_dn_3: 0.0998, loss_box_dn_3: 0.6071, loss_cls_dn_4: 0.0979, loss_box_dn_4: 0.6122, loss_cls_dn_5: 0.0997, loss_box_dn_5: 0.6178, loss_dense_depth: 0.8481, loss: 23.1081, grad_norm: 33.0709 -2026-01-14 21:41:38,798 - mmdet - INFO - Iter [370/17500] lr: 2.473e-04, eta: 9:28:29, time: 1.567, data_time: 0.075, memory: 49164, loss_cls_0: 0.7038, loss_box_0: 1.5951, loss_cns_0: 0.6392, loss_yns_0: 0.1349, loss_cls_1: 0.7797, loss_box_1: 1.4139, loss_cns_1: 0.6684, loss_yns_1: 0.1351, loss_cls_2: 0.7993, loss_box_2: 1.3701, loss_cns_2: 0.6675, loss_yns_2: 0.1333, loss_cls_3: 0.7941, loss_box_3: 1.3611, loss_cns_3: 0.6660, loss_yns_3: 0.1342, loss_cls_4: 0.7972, loss_box_4: 1.3771, loss_cns_4: 0.6673, loss_yns_4: 0.1338, loss_cls_5: 0.8048, loss_box_5: 1.3649, loss_cns_5: 0.6666, loss_yns_5: 0.1337, loss_cls_dn_0: 0.1317, loss_box_dn_0: 0.7275, loss_cls_dn_1: 0.0994, loss_box_dn_1: 0.6336, loss_cls_dn_2: 0.0987, loss_box_dn_2: 0.6156, loss_cls_dn_3: 0.0987, loss_box_dn_3: 0.6115, loss_cls_dn_4: 0.0985, loss_box_dn_4: 0.6142, loss_cls_dn_5: 0.0994, loss_box_dn_5: 0.6107, loss_dense_depth: 0.7894, loss: 23.1699, grad_norm: 28.1343 -2026-01-14 21:41:40,386 - mmdet - INFO - Iter [371/17500] lr: 2.477e-04, eta: 9:28:07, time: 1.560, data_time: 0.075, memory: 49164, loss_cls_0: 0.6868, loss_box_0: 1.5402, loss_cns_0: 0.6413, loss_yns_0: 0.1348, loss_cls_1: 0.7631, loss_box_1: 1.4101, loss_cns_1: 0.6649, loss_yns_1: 0.1368, loss_cls_2: 0.7798, loss_box_2: 1.3694, loss_cns_2: 0.6637, loss_yns_2: 0.1369, loss_cls_3: 0.7799, loss_box_3: 1.3504, loss_cns_3: 0.6611, loss_yns_3: 0.1366, loss_cls_4: 0.7782, loss_box_4: 1.3627, loss_cns_4: 0.6629, loss_yns_4: 0.1361, loss_cls_5: 0.7849, loss_box_5: 1.3518, loss_cns_5: 0.6600, loss_yns_5: 0.1353, loss_cls_dn_0: 0.1188, loss_box_dn_0: 0.7149, loss_cls_dn_1: 0.0959, loss_box_dn_1: 0.6223, loss_cls_dn_2: 0.0942, loss_box_dn_2: 0.6063, loss_cls_dn_3: 0.0947, loss_box_dn_3: 0.5986, loss_cls_dn_4: 0.0944, loss_box_dn_4: 0.6000, loss_cls_dn_5: 0.0960, loss_box_dn_5: 0.5965, loss_dense_depth: 0.7168, loss: 22.7771, grad_norm: 26.1610 -2026-01-14 21:41:41,957 - mmdet - INFO - Iter [372/17500] lr: 2.481e-04, eta: 9:27:47, time: 1.595, data_time: 0.090, memory: 49164, loss_cls_0: 0.7153, loss_box_0: 1.5714, loss_cns_0: 0.6304, loss_yns_0: 0.1377, loss_cls_1: 0.7777, loss_box_1: 1.4537, loss_cns_1: 0.6594, loss_yns_1: 0.1371, loss_cls_2: 0.7961, loss_box_2: 1.4183, loss_cns_2: 0.6585, loss_yns_2: 0.1368, loss_cls_3: 0.7901, loss_box_3: 1.4144, loss_cns_3: 0.6615, loss_yns_3: 0.1365, loss_cls_4: 0.7914, loss_box_4: 1.4285, loss_cns_4: 0.6583, loss_yns_4: 0.1382, loss_cls_5: 0.8014, loss_box_5: 1.4205, loss_cns_5: 0.6563, loss_yns_5: 0.1372, loss_cls_dn_0: 0.1219, loss_box_dn_0: 0.7346, loss_cls_dn_1: 0.0958, loss_box_dn_1: 0.6212, loss_cls_dn_2: 0.0969, loss_box_dn_2: 0.6050, loss_cls_dn_3: 0.0971, loss_box_dn_3: 0.5987, loss_cls_dn_4: 0.0952, loss_box_dn_4: 0.6030, loss_cls_dn_5: 0.0968, loss_box_dn_5: 0.6054, loss_dense_depth: 0.7427, loss: 23.2409, grad_norm: 36.6618 -2026-01-14 21:41:43,543 - mmdet - INFO - Iter [373/17500] lr: 2.485e-04, eta: 9:27:27, time: 1.590, data_time: 0.081, memory: 49164, loss_cls_0: 0.7503, loss_box_0: 1.5996, loss_cns_0: 0.6347, loss_yns_0: 0.1421, loss_cls_1: 0.8009, loss_box_1: 1.4575, loss_cns_1: 0.6563, loss_yns_1: 0.1376, loss_cls_2: 0.8211, loss_box_2: 1.4366, loss_cns_2: 0.6631, loss_yns_2: 0.1371, loss_cls_3: 0.8256, loss_box_3: 1.4081, loss_cns_3: 0.6601, loss_yns_3: 0.1390, loss_cls_4: 0.8095, loss_box_4: 1.4163, loss_cns_4: 0.6600, loss_yns_4: 0.1392, loss_cls_5: 0.8173, loss_box_5: 1.4220, loss_cns_5: 0.6597, loss_yns_5: 0.1403, loss_cls_dn_0: 0.1261, loss_box_dn_0: 0.7223, loss_cls_dn_1: 0.0977, loss_box_dn_1: 0.6309, loss_cls_dn_2: 0.0986, loss_box_dn_2: 0.6189, loss_cls_dn_3: 0.1009, loss_box_dn_3: 0.6083, loss_cls_dn_4: 0.0957, loss_box_dn_4: 0.6085, loss_cls_dn_5: 0.0944, loss_box_dn_5: 0.6114, loss_dense_depth: 0.7491, loss: 23.4970, grad_norm: 30.5693 -2026-01-14 21:41:45,127 - mmdet - INFO - Iter [374/17500] lr: 2.489e-04, eta: 9:27:06, time: 1.578, data_time: 0.077, memory: 49164, loss_cls_0: 0.7269, loss_box_0: 1.5841, loss_cns_0: 0.6284, loss_yns_0: 0.1414, loss_cls_1: 0.7930, loss_box_1: 1.4278, loss_cns_1: 0.6583, loss_yns_1: 0.1377, loss_cls_2: 0.8020, loss_box_2: 1.4059, loss_cns_2: 0.6644, loss_yns_2: 0.1383, loss_cls_3: 0.8019, loss_box_3: 1.3898, loss_cns_3: 0.6602, loss_yns_3: 0.1396, loss_cls_4: 0.8024, loss_box_4: 1.3981, loss_cns_4: 0.6583, loss_yns_4: 0.1393, loss_cls_5: 0.8097, loss_box_5: 1.4039, loss_cns_5: 0.6591, loss_yns_5: 0.1399, loss_cls_dn_0: 0.1222, loss_box_dn_0: 0.7193, loss_cls_dn_1: 0.0947, loss_box_dn_1: 0.6213, loss_cls_dn_2: 0.0946, loss_box_dn_2: 0.6081, loss_cls_dn_3: 0.0965, loss_box_dn_3: 0.6032, loss_cls_dn_4: 0.0935, loss_box_dn_4: 0.6095, loss_cls_dn_5: 0.0929, loss_box_dn_5: 0.6129, loss_dense_depth: 0.7259, loss: 23.2051, grad_norm: 24.9330 -2026-01-14 21:41:46,783 - mmdet - INFO - Iter [375/17500] lr: 2.493e-04, eta: 9:26:48, time: 1.630, data_time: 0.096, memory: 49164, loss_cls_0: 0.7309, loss_box_0: 1.5555, loss_cns_0: 0.6380, loss_yns_0: 0.1410, loss_cls_1: 0.7991, loss_box_1: 1.4162, loss_cns_1: 0.6634, loss_yns_1: 0.1384, loss_cls_2: 0.8086, loss_box_2: 1.4048, loss_cns_2: 0.6651, loss_yns_2: 0.1394, loss_cls_3: 0.8088, loss_box_3: 1.4016, loss_cns_3: 0.6623, loss_yns_3: 0.1379, loss_cls_4: 0.8099, loss_box_4: 1.3870, loss_cns_4: 0.6620, loss_yns_4: 0.1373, loss_cls_5: 0.8159, loss_box_5: 1.3894, loss_cns_5: 0.6642, loss_yns_5: 0.1376, loss_cls_dn_0: 0.1194, loss_box_dn_0: 0.7087, loss_cls_dn_1: 0.0923, loss_box_dn_1: 0.6132, loss_cls_dn_2: 0.0901, loss_box_dn_2: 0.6028, loss_cls_dn_3: 0.0932, loss_box_dn_3: 0.6025, loss_cls_dn_4: 0.0908, loss_box_dn_4: 0.5997, loss_cls_dn_5: 0.0918, loss_box_dn_5: 0.6016, loss_dense_depth: 0.7868, loss: 23.2074, grad_norm: 32.1077 -2026-01-14 21:41:48,405 - mmdet - INFO - Iter [376/17500] lr: 2.497e-04, eta: 9:26:31, time: 1.651, data_time: 0.095, memory: 49164, loss_cls_0: 0.7577, loss_box_0: 1.5841, loss_cns_0: 0.6352, loss_yns_0: 0.1422, loss_cls_1: 0.8148, loss_box_1: 1.4266, loss_cns_1: 0.6630, loss_yns_1: 0.1391, loss_cls_2: 0.8177, loss_box_2: 1.4054, loss_cns_2: 0.6621, loss_yns_2: 0.1394, loss_cls_3: 0.8301, loss_box_3: 1.3926, loss_cns_3: 0.6629, loss_yns_3: 0.1384, loss_cls_4: 0.8211, loss_box_4: 1.3915, loss_cns_4: 0.6616, loss_yns_4: 0.1383, loss_cls_5: 0.8269, loss_box_5: 1.4017, loss_cns_5: 0.6618, loss_yns_5: 0.1389, loss_cls_dn_0: 0.1267, loss_box_dn_0: 0.7216, loss_cls_dn_1: 0.0948, loss_box_dn_1: 0.6384, loss_cls_dn_2: 0.0925, loss_box_dn_2: 0.6239, loss_cls_dn_3: 0.0939, loss_box_dn_3: 0.6203, loss_cls_dn_4: 0.0917, loss_box_dn_4: 0.6194, loss_cls_dn_5: 0.0928, loss_box_dn_5: 0.6260, loss_dense_depth: 0.7933, loss: 23.4883, grad_norm: 29.5119 -2026-01-14 21:41:50,028 - mmdet - INFO - Iter [377/17500] lr: 2.501e-04, eta: 9:26:11, time: 1.587, data_time: 0.078, memory: 49164, loss_cls_0: 0.7426, loss_box_0: 1.5383, loss_cns_0: 0.6383, loss_yns_0: 0.1426, loss_cls_1: 0.7924, loss_box_1: 1.3989, loss_cns_1: 0.6629, loss_yns_1: 0.1388, loss_cls_2: 0.7944, loss_box_2: 1.3803, loss_cns_2: 0.6592, loss_yns_2: 0.1373, loss_cls_3: 0.8022, loss_box_3: 1.3684, loss_cns_3: 0.6646, loss_yns_3: 0.1381, loss_cls_4: 0.7987, loss_box_4: 1.3744, loss_cns_4: 0.6633, loss_yns_4: 0.1388, loss_cls_5: 0.8066, loss_box_5: 1.3669, loss_cns_5: 0.6643, loss_yns_5: 0.1384, loss_cls_dn_0: 0.1197, loss_box_dn_0: 0.7283, loss_cls_dn_1: 0.0941, loss_box_dn_1: 0.6453, loss_cls_dn_2: 0.0933, loss_box_dn_2: 0.6263, loss_cls_dn_3: 0.0962, loss_box_dn_3: 0.6183, loss_cls_dn_4: 0.0941, loss_box_dn_4: 0.6201, loss_cls_dn_5: 0.0933, loss_box_dn_5: 0.6211, loss_dense_depth: 0.7685, loss: 23.1694, grad_norm: 23.7924 -2026-01-14 21:41:51,589 - mmdet - INFO - Iter [378/17500] lr: 2.505e-04, eta: 9:25:51, time: 1.599, data_time: 0.113, memory: 49164, loss_cls_0: 0.7319, loss_box_0: 1.5472, loss_cns_0: 0.6390, loss_yns_0: 0.1437, loss_cls_1: 0.7832, loss_box_1: 1.4026, loss_cns_1: 0.6622, loss_yns_1: 0.1404, loss_cls_2: 0.7820, loss_box_2: 1.3793, loss_cns_2: 0.6618, loss_yns_2: 0.1396, loss_cls_3: 0.7952, loss_box_3: 1.3667, loss_cns_3: 0.6619, loss_yns_3: 0.1397, loss_cls_4: 0.7976, loss_box_4: 1.3721, loss_cns_4: 0.6632, loss_yns_4: 0.1405, loss_cls_5: 0.8059, loss_box_5: 1.3708, loss_cns_5: 0.6637, loss_yns_5: 0.1406, loss_cls_dn_0: 0.1211, loss_box_dn_0: 0.7204, loss_cls_dn_1: 0.0944, loss_box_dn_1: 0.6372, loss_cls_dn_2: 0.0939, loss_box_dn_2: 0.6222, loss_cls_dn_3: 0.0998, loss_box_dn_3: 0.6203, loss_cls_dn_4: 0.0925, loss_box_dn_4: 0.6225, loss_cls_dn_5: 0.0923, loss_box_dn_5: 0.6257, loss_dense_depth: 0.7802, loss: 23.1535, grad_norm: 31.9219 -2026-01-14 21:41:53,247 - mmdet - INFO - Iter [379/17500] lr: 2.509e-04, eta: 9:25:34, time: 1.656, data_time: 0.190, memory: 49164, loss_cls_0: 0.7281, loss_box_0: 1.5635, loss_cns_0: 0.6375, loss_yns_0: 0.1444, loss_cls_1: 0.8032, loss_box_1: 1.4252, loss_cns_1: 0.6602, loss_yns_1: 0.1390, loss_cls_2: 0.8085, loss_box_2: 1.3866, loss_cns_2: 0.6642, loss_yns_2: 0.1390, loss_cls_3: 0.8100, loss_box_3: 1.3898, loss_cns_3: 0.6594, loss_yns_3: 0.1395, loss_cls_4: 0.8308, loss_box_4: 1.3800, loss_cns_4: 0.6613, loss_yns_4: 0.1384, loss_cls_5: 0.8302, loss_box_5: 1.3841, loss_cns_5: 0.6594, loss_yns_5: 0.1387, loss_cls_dn_0: 0.1168, loss_box_dn_0: 0.7106, loss_cls_dn_1: 0.0902, loss_box_dn_1: 0.6328, loss_cls_dn_2: 0.0894, loss_box_dn_2: 0.6193, loss_cls_dn_3: 0.0898, loss_box_dn_3: 0.6207, loss_cls_dn_4: 0.0933, loss_box_dn_4: 0.6233, loss_cls_dn_5: 0.0938, loss_box_dn_5: 0.6322, loss_dense_depth: 0.7760, loss: 23.3093, grad_norm: 39.5412 -2026-01-14 21:41:54,838 - mmdet - INFO - Iter [380/17500] lr: 2.513e-04, eta: 9:25:15, time: 1.586, data_time: 0.084, memory: 49164, loss_cls_0: 0.7461, loss_box_0: 1.5521, loss_cns_0: 0.6404, loss_yns_0: 0.1401, loss_cls_1: 0.8030, loss_box_1: 1.4072, loss_cns_1: 0.6591, loss_yns_1: 0.1367, loss_cls_2: 0.8132, loss_box_2: 1.3720, loss_cns_2: 0.6577, loss_yns_2: 0.1359, loss_cls_3: 0.8213, loss_box_3: 1.3378, loss_cns_3: 0.6565, loss_yns_3: 0.1356, loss_cls_4: 0.8281, loss_box_4: 1.3263, loss_cns_4: 0.6554, loss_yns_4: 0.1360, loss_cls_5: 0.8287, loss_box_5: 1.3325, loss_cns_5: 0.6520, loss_yns_5: 0.1353, loss_cls_dn_0: 0.1205, loss_box_dn_0: 0.7147, loss_cls_dn_1: 0.0927, loss_box_dn_1: 0.6353, loss_cls_dn_2: 0.0926, loss_box_dn_2: 0.6279, loss_cls_dn_3: 0.0932, loss_box_dn_3: 0.6142, loss_cls_dn_4: 0.1007, loss_box_dn_4: 0.6149, loss_cls_dn_5: 0.1006, loss_box_dn_5: 0.6225, loss_dense_depth: 0.7664, loss: 23.1051, grad_norm: 30.4676 -2026-01-14 21:41:56,534 - mmdet - INFO - Iter [381/17500] lr: 2.517e-04, eta: 9:25:00, time: 1.702, data_time: 0.081, memory: 49164, loss_cls_0: 0.7271, loss_box_0: 1.5102, loss_cns_0: 0.6399, loss_yns_0: 0.1393, loss_cls_1: 0.7720, loss_box_1: 1.4007, loss_cns_1: 0.6593, loss_yns_1: 0.1347, loss_cls_2: 0.7787, loss_box_2: 1.3894, loss_cns_2: 0.6585, loss_yns_2: 0.1342, loss_cls_3: 0.7919, loss_box_3: 1.3590, loss_cns_3: 0.6601, loss_yns_3: 0.1340, loss_cls_4: 0.7950, loss_box_4: 1.3718, loss_cns_4: 0.6590, loss_yns_4: 0.1352, loss_cls_5: 0.8080, loss_box_5: 1.3676, loss_cns_5: 0.6567, loss_yns_5: 0.1354, loss_cls_dn_0: 0.1213, loss_box_dn_0: 0.7128, loss_cls_dn_1: 0.0917, loss_box_dn_1: 0.6210, loss_cls_dn_2: 0.0909, loss_box_dn_2: 0.6174, loss_cls_dn_3: 0.0908, loss_box_dn_3: 0.6040, loss_cls_dn_4: 0.0913, loss_box_dn_4: 0.6088, loss_cls_dn_5: 0.0955, loss_box_dn_5: 0.6140, loss_dense_depth: 0.7136, loss: 22.8906, grad_norm: 35.2820 -2026-01-14 21:41:58,229 - mmdet - INFO - Iter [382/17500] lr: 2.521e-04, eta: 9:24:45, time: 1.696, data_time: 0.085, memory: 49164, loss_cls_0: 0.7057, loss_box_0: 1.5356, loss_cns_0: 0.6378, loss_yns_0: 0.1386, loss_cls_1: 0.7780, loss_box_1: 1.4087, loss_cns_1: 0.6681, loss_yns_1: 0.1362, loss_cls_2: 0.7947, loss_box_2: 1.3983, loss_cns_2: 0.6656, loss_yns_2: 0.1343, loss_cls_3: 0.8047, loss_box_3: 1.3765, loss_cns_3: 0.6670, loss_yns_3: 0.1361, loss_cls_4: 0.8158, loss_box_4: 1.3756, loss_cns_4: 0.6671, loss_yns_4: 0.1363, loss_cls_5: 0.8201, loss_box_5: 1.3916, loss_cns_5: 0.6662, loss_yns_5: 0.1362, loss_cls_dn_0: 0.1190, loss_box_dn_0: 0.7176, loss_cls_dn_1: 0.0921, loss_box_dn_1: 0.6205, loss_cls_dn_2: 0.0907, loss_box_dn_2: 0.6070, loss_cls_dn_3: 0.0908, loss_box_dn_3: 0.6004, loss_cls_dn_4: 0.0908, loss_box_dn_4: 0.6047, loss_cls_dn_5: 0.0952, loss_box_dn_5: 0.6122, loss_dense_depth: 0.7207, loss: 23.0569, grad_norm: 32.3662 -2026-01-14 21:42:01,524 - mmdet - INFO - Iter [383/17500] lr: 2.525e-04, eta: 9:25:42, time: 3.293, data_time: 0.076, memory: 49164, loss_cls_0: 0.7449, loss_box_0: 1.5096, loss_cns_0: 0.6293, loss_yns_0: 0.1361, loss_cls_1: 0.8032, loss_box_1: 1.4021, loss_cns_1: 0.6701, loss_yns_1: 0.1353, loss_cls_2: 0.8037, loss_box_2: 1.3742, loss_cns_2: 0.6677, loss_yns_2: 0.1351, loss_cls_3: 0.8176, loss_box_3: 1.3722, loss_cns_3: 0.6670, loss_yns_3: 0.1349, loss_cls_4: 0.8235, loss_box_4: 1.3577, loss_cns_4: 0.6658, loss_yns_4: 0.1352, loss_cls_5: 0.8292, loss_box_5: 1.3734, loss_cns_5: 0.6664, loss_yns_5: 0.1361, loss_cls_dn_0: 0.1166, loss_box_dn_0: 0.7122, loss_cls_dn_1: 0.0950, loss_box_dn_1: 0.6214, loss_cls_dn_2: 0.0926, loss_box_dn_2: 0.6035, loss_cls_dn_3: 0.0970, loss_box_dn_3: 0.6049, loss_cls_dn_4: 0.0950, loss_box_dn_4: 0.6029, loss_cls_dn_5: 0.0955, loss_box_dn_5: 0.6096, loss_dense_depth: 0.6979, loss: 23.0345, grad_norm: 29.4554 -2026-01-14 21:42:03,163 - mmdet - INFO - Iter [384/17500] lr: 2.529e-04, eta: 9:25:25, time: 1.640, data_time: 0.080, memory: 49164, loss_cls_0: 0.7485, loss_box_0: 1.5220, loss_cns_0: 0.6303, loss_yns_0: 0.1364, loss_cls_1: 0.8029, loss_box_1: 1.4385, loss_cns_1: 0.6639, loss_yns_1: 0.1363, loss_cls_2: 0.8117, loss_box_2: 1.4172, loss_cns_2: 0.6633, loss_yns_2: 0.1357, loss_cls_3: 0.8265, loss_box_3: 1.4031, loss_cns_3: 0.6610, loss_yns_3: 0.1355, loss_cls_4: 0.8378, loss_box_4: 1.4087, loss_cns_4: 0.6637, loss_yns_4: 0.1349, loss_cls_5: 0.8295, loss_box_5: 1.4122, loss_cns_5: 0.6620, loss_yns_5: 0.1338, loss_cls_dn_0: 0.1174, loss_box_dn_0: 0.7069, loss_cls_dn_1: 0.0904, loss_box_dn_1: 0.6216, loss_cls_dn_2: 0.0905, loss_box_dn_2: 0.6127, loss_cls_dn_3: 0.0917, loss_box_dn_3: 0.6090, loss_cls_dn_4: 0.0908, loss_box_dn_4: 0.6121, loss_cls_dn_5: 0.0904, loss_box_dn_5: 0.6128, loss_dense_depth: 0.7191, loss: 23.2806, grad_norm: 28.2625 -2026-01-14 21:42:04,799 - mmdet - INFO - Iter [385/17500] lr: 2.533e-04, eta: 9:25:07, time: 1.634, data_time: 0.075, memory: 49164, loss_cls_0: 0.7327, loss_box_0: 1.5514, loss_cns_0: 0.6491, loss_yns_0: 0.1402, loss_cls_1: 0.7867, loss_box_1: 1.4195, loss_cns_1: 0.6676, loss_yns_1: 0.1394, loss_cls_2: 0.7918, loss_box_2: 1.4118, loss_cns_2: 0.6693, loss_yns_2: 0.1352, loss_cls_3: 0.8041, loss_box_3: 1.3940, loss_cns_3: 0.6652, loss_yns_3: 0.1347, loss_cls_4: 0.8112, loss_box_4: 1.4018, loss_cns_4: 0.6659, loss_yns_4: 0.1346, loss_cls_5: 0.8083, loss_box_5: 1.4081, loss_cns_5: 0.6672, loss_yns_5: 0.1351, loss_cls_dn_0: 0.1164, loss_box_dn_0: 0.7055, loss_cls_dn_1: 0.0871, loss_box_dn_1: 0.6195, loss_cls_dn_2: 0.0851, loss_box_dn_2: 0.6134, loss_cls_dn_3: 0.0851, loss_box_dn_3: 0.6093, loss_cls_dn_4: 0.0884, loss_box_dn_4: 0.6153, loss_cls_dn_5: 0.0911, loss_box_dn_5: 0.6181, loss_dense_depth: 0.6866, loss: 23.1461, grad_norm: 32.4552 -2026-01-14 21:42:07,954 - mmdet - INFO - Iter [386/17500] lr: 2.537e-04, eta: 9:24:51, time: 1.656, data_time: 0.075, memory: 49164, loss_cls_0: 0.7459, loss_box_0: 1.5578, loss_cns_0: 0.6445, loss_yns_0: 0.1358, loss_cls_1: 0.7919, loss_box_1: 1.4572, loss_cns_1: 0.6632, loss_yns_1: 0.1357, loss_cls_2: 0.7926, loss_box_2: 1.4642, loss_cns_2: 0.6637, loss_yns_2: 0.1330, loss_cls_3: 0.8074, loss_box_3: 1.4554, loss_cns_3: 0.6629, loss_yns_3: 0.1317, loss_cls_4: 0.8049, loss_box_4: 1.4529, loss_cns_4: 0.6648, loss_yns_4: 0.1323, loss_cls_5: 0.8160, loss_box_5: 1.4681, loss_cns_5: 0.6659, loss_yns_5: 0.1330, loss_cls_dn_0: 0.1186, loss_box_dn_0: 0.7115, loss_cls_dn_1: 0.0890, loss_box_dn_1: 0.6149, loss_cls_dn_2: 0.0868, loss_box_dn_2: 0.6117, loss_cls_dn_3: 0.0881, loss_box_dn_3: 0.6118, loss_cls_dn_4: 0.0908, loss_box_dn_4: 0.6106, loss_cls_dn_5: 0.0935, loss_box_dn_5: 0.6198, loss_dense_depth: 0.7253, loss: 23.4535, grad_norm: 34.6369 -2026-01-14 21:42:09,808 - mmdet - INFO - Iter [387/17500] lr: 2.541e-04, eta: 9:25:50, time: 3.354, data_time: 1.552, memory: 49164, loss_cls_0: 0.6999, loss_box_0: 1.5624, loss_cns_0: 0.6412, loss_yns_0: 0.1356, loss_cls_1: 0.7638, loss_box_1: 1.4295, loss_cns_1: 0.6645, loss_yns_1: 0.1344, loss_cls_2: 0.7602, loss_box_2: 1.3986, loss_cns_2: 0.6598, loss_yns_2: 0.1310, loss_cls_3: 0.7672, loss_box_3: 1.4074, loss_cns_3: 0.6616, loss_yns_3: 0.1305, loss_cls_4: 0.7795, loss_box_4: 1.3804, loss_cns_4: 0.6617, loss_yns_4: 0.1298, loss_cls_5: 0.7826, loss_box_5: 1.3976, loss_cns_5: 0.6646, loss_yns_5: 0.1317, loss_cls_dn_0: 0.1153, loss_box_dn_0: 0.7044, loss_cls_dn_1: 0.0896, loss_box_dn_1: 0.6301, loss_cls_dn_2: 0.0864, loss_box_dn_2: 0.6191, loss_cls_dn_3: 0.0871, loss_box_dn_3: 0.6206, loss_cls_dn_4: 0.0886, loss_box_dn_4: 0.6140, loss_cls_dn_5: 0.0905, loss_box_dn_5: 0.6192, loss_dense_depth: 0.7141, loss: 22.9545, grad_norm: 30.6386 -2026-01-14 21:42:11,361 - mmdet - INFO - Iter [388/17500] lr: 2.545e-04, eta: 9:25:29, time: 1.553, data_time: 0.075, memory: 49164, loss_cls_0: 0.7016, loss_box_0: 1.5404, loss_cns_0: 0.6404, loss_yns_0: 0.1364, loss_cls_1: 0.7572, loss_box_1: 1.4057, loss_cns_1: 0.6623, loss_yns_1: 0.1339, loss_cls_2: 0.7565, loss_box_2: 1.3911, loss_cns_2: 0.6603, loss_yns_2: 0.1316, loss_cls_3: 0.7618, loss_box_3: 1.3983, loss_cns_3: 0.6633, loss_yns_3: 0.1318, loss_cls_4: 0.7796, loss_box_4: 1.3764, loss_cns_4: 0.6623, loss_yns_4: 0.1310, loss_cls_5: 0.7801, loss_box_5: 1.3828, loss_cns_5: 0.6654, loss_yns_5: 0.1321, loss_cls_dn_0: 0.1102, loss_box_dn_0: 0.7135, loss_cls_dn_1: 0.0900, loss_box_dn_1: 0.6300, loss_cls_dn_2: 0.0869, loss_box_dn_2: 0.6193, loss_cls_dn_3: 0.0886, loss_box_dn_3: 0.6213, loss_cls_dn_4: 0.0890, loss_box_dn_4: 0.6158, loss_cls_dn_5: 0.0894, loss_box_dn_5: 0.6210, loss_dense_depth: 0.7032, loss: 22.8606, grad_norm: 31.6768 -2026-01-14 21:42:19,863 - mmdet - INFO - Iter [389/17500] lr: 2.549e-04, eta: 9:30:14, time: 8.503, data_time: 0.071, memory: 49164, loss_cls_0: 0.7031, loss_box_0: 1.5178, loss_cns_0: 0.6458, loss_yns_0: 0.1384, loss_cls_1: 0.7565, loss_box_1: 1.4158, loss_cns_1: 0.6643, loss_yns_1: 0.1374, loss_cls_2: 0.7623, loss_box_2: 1.3867, loss_cns_2: 0.6651, loss_yns_2: 0.1346, loss_cls_3: 0.7699, loss_box_3: 1.3701, loss_cns_3: 0.6620, loss_yns_3: 0.1343, loss_cls_4: 0.7754, loss_box_4: 1.3785, loss_cns_4: 0.6626, loss_yns_4: 0.1343, loss_cls_5: 0.7767, loss_box_5: 1.3723, loss_cns_5: 0.6635, loss_yns_5: 0.1358, loss_cls_dn_0: 0.1112, loss_box_dn_0: 0.7077, loss_cls_dn_1: 0.0885, loss_box_dn_1: 0.6389, loss_cls_dn_2: 0.0879, loss_box_dn_2: 0.6234, loss_cls_dn_3: 0.0868, loss_box_dn_3: 0.6149, loss_cls_dn_4: 0.0862, loss_box_dn_4: 0.6199, loss_cls_dn_5: 0.0888, loss_box_dn_5: 0.6238, loss_dense_depth: 0.7154, loss: 22.8565, grad_norm: 27.7653 -2026-01-14 21:42:21,405 - mmdet - INFO - Iter [390/17500] lr: 2.553e-04, eta: 9:29:52, time: 1.540, data_time: 0.068, memory: 49164, loss_cls_0: 0.7096, loss_box_0: 1.5505, loss_cns_0: 0.6395, loss_yns_0: 0.1419, loss_cls_1: 0.7685, loss_box_1: 1.4129, loss_cns_1: 0.6665, loss_yns_1: 0.1411, loss_cls_2: 0.7726, loss_box_2: 1.3817, loss_cns_2: 0.6675, loss_yns_2: 0.1384, loss_cls_3: 0.7715, loss_box_3: 1.3720, loss_cns_3: 0.6638, loss_yns_3: 0.1382, loss_cls_4: 0.7818, loss_box_4: 1.3825, loss_cns_4: 0.6638, loss_yns_4: 0.1396, loss_cls_5: 0.7870, loss_box_5: 1.3699, loss_cns_5: 0.6645, loss_yns_5: 0.1390, loss_cls_dn_0: 0.1132, loss_box_dn_0: 0.7126, loss_cls_dn_1: 0.0908, loss_box_dn_1: 0.6286, loss_cls_dn_2: 0.0911, loss_box_dn_2: 0.6092, loss_cls_dn_3: 0.0888, loss_box_dn_3: 0.6057, loss_cls_dn_4: 0.0910, loss_box_dn_4: 0.6083, loss_cls_dn_5: 0.0943, loss_box_dn_5: 0.6082, loss_dense_depth: 0.6888, loss: 22.8951, grad_norm: 30.0688 -2026-01-14 21:42:22,959 - mmdet - INFO - Iter [391/17500] lr: 2.557e-04, eta: 9:29:30, time: 1.552, data_time: 0.071, memory: 49164, loss_cls_0: 0.6906, loss_box_0: 1.5280, loss_cns_0: 0.6386, loss_yns_0: 0.1380, loss_cls_1: 0.7603, loss_box_1: 1.3830, loss_cns_1: 0.6666, loss_yns_1: 0.1364, loss_cls_2: 0.7545, loss_box_2: 1.3580, loss_cns_2: 0.6671, loss_yns_2: 0.1339, loss_cls_3: 0.7625, loss_box_3: 1.3677, loss_cns_3: 0.6640, loss_yns_3: 0.1334, loss_cls_4: 0.7651, loss_box_4: 1.3473, loss_cns_4: 0.6660, loss_yns_4: 0.1351, loss_cls_5: 0.7771, loss_box_5: 1.3410, loss_cns_5: 0.6655, loss_yns_5: 0.1349, loss_cls_dn_0: 0.1145, loss_box_dn_0: 0.7072, loss_cls_dn_1: 0.0917, loss_box_dn_1: 0.6322, loss_cls_dn_2: 0.0900, loss_box_dn_2: 0.6139, loss_cls_dn_3: 0.0905, loss_box_dn_3: 0.6214, loss_cls_dn_4: 0.0922, loss_box_dn_4: 0.6142, loss_cls_dn_5: 0.0937, loss_box_dn_5: 0.6151, loss_dense_depth: 0.6990, loss: 22.6902, grad_norm: 25.3831 -2026-01-14 21:42:24,579 - mmdet - INFO - Iter [392/17500] lr: 2.561e-04, eta: 9:29:10, time: 1.591, data_time: 0.074, memory: 49164, loss_cls_0: 0.7131, loss_box_0: 1.5470, loss_cns_0: 0.6376, loss_yns_0: 0.1409, loss_cls_1: 0.7786, loss_box_1: 1.4094, loss_cns_1: 0.6657, loss_yns_1: 0.1392, loss_cls_2: 0.7839, loss_box_2: 1.3940, loss_cns_2: 0.6634, loss_yns_2: 0.1403, loss_cls_3: 0.7931, loss_box_3: 1.3851, loss_cns_3: 0.6631, loss_yns_3: 0.1390, loss_cls_4: 0.7902, loss_box_4: 1.3895, loss_cns_4: 0.6641, loss_yns_4: 0.1399, loss_cls_5: 0.8077, loss_box_5: 1.3989, loss_cns_5: 0.6629, loss_yns_5: 0.1414, loss_cls_dn_0: 0.1194, loss_box_dn_0: 0.7078, loss_cls_dn_1: 0.0934, loss_box_dn_1: 0.6258, loss_cls_dn_2: 0.0903, loss_box_dn_2: 0.6146, loss_cls_dn_3: 0.0928, loss_box_dn_3: 0.6161, loss_cls_dn_4: 0.0924, loss_box_dn_4: 0.6193, loss_cls_dn_5: 0.0935, loss_box_dn_5: 0.6257, loss_dense_depth: 0.7069, loss: 23.0858, grad_norm: 29.9671 -2026-01-14 21:42:26,157 - mmdet - INFO - Iter [393/17500] lr: 2.565e-04, eta: 9:28:51, time: 1.609, data_time: 0.113, memory: 49164, loss_cls_0: 0.7221, loss_box_0: 1.5421, loss_cns_0: 0.6367, loss_yns_0: 0.1412, loss_cls_1: 0.7856, loss_box_1: 1.3903, loss_cns_1: 0.6643, loss_yns_1: 0.1408, loss_cls_2: 0.7863, loss_box_2: 1.3736, loss_cns_2: 0.6645, loss_yns_2: 0.1408, loss_cls_3: 0.7955, loss_box_3: 1.3536, loss_cns_3: 0.6622, loss_yns_3: 0.1381, loss_cls_4: 0.7980, loss_box_4: 1.3649, loss_cns_4: 0.6633, loss_yns_4: 0.1370, loss_cls_5: 0.8003, loss_box_5: 1.3607, loss_cns_5: 0.6618, loss_yns_5: 0.1372, loss_cls_dn_0: 0.1146, loss_box_dn_0: 0.7177, loss_cls_dn_1: 0.0970, loss_box_dn_1: 0.6303, loss_cls_dn_2: 0.0938, loss_box_dn_2: 0.6197, loss_cls_dn_3: 0.0929, loss_box_dn_3: 0.6132, loss_cls_dn_4: 0.0912, loss_box_dn_4: 0.6180, loss_cls_dn_5: 0.0912, loss_box_dn_5: 0.6145, loss_dense_depth: 0.7143, loss: 22.9695, grad_norm: 23.8596 -2026-01-14 21:42:27,720 - mmdet - INFO - Iter [394/17500] lr: 2.569e-04, eta: 9:28:31, time: 1.562, data_time: 0.074, memory: 49164, loss_cls_0: 0.7034, loss_box_0: 1.4978, loss_cns_0: 0.6413, loss_yns_0: 0.1421, loss_cls_1: 0.7782, loss_box_1: 1.4027, loss_cns_1: 0.6630, loss_yns_1: 0.1398, loss_cls_2: 0.7813, loss_box_2: 1.3657, loss_cns_2: 0.6626, loss_yns_2: 0.1412, loss_cls_3: 0.7895, loss_box_3: 1.3475, loss_cns_3: 0.6597, loss_yns_3: 0.1392, loss_cls_4: 0.7862, loss_box_4: 1.3400, loss_cns_4: 0.6619, loss_yns_4: 0.1397, loss_cls_5: 0.7868, loss_box_5: 1.3532, loss_cns_5: 0.6589, loss_yns_5: 0.1389, loss_cls_dn_0: 0.1172, loss_box_dn_0: 0.7060, loss_cls_dn_1: 0.0924, loss_box_dn_1: 0.6222, loss_cls_dn_2: 0.0898, loss_box_dn_2: 0.6059, loss_cls_dn_3: 0.0911, loss_box_dn_3: 0.6024, loss_cls_dn_4: 0.0927, loss_box_dn_4: 0.5979, loss_cls_dn_5: 0.0957, loss_box_dn_5: 0.6061, loss_dense_depth: 0.7185, loss: 22.7588, grad_norm: 34.4135 -2026-01-14 21:42:29,286 - mmdet - INFO - Iter [395/17500] lr: 2.573e-04, eta: 9:28:10, time: 1.568, data_time: 0.079, memory: 49164, loss_cls_0: 0.7031, loss_box_0: 1.5111, loss_cns_0: 0.6373, loss_yns_0: 0.1418, loss_cls_1: 0.7794, loss_box_1: 1.3755, loss_cns_1: 0.6633, loss_yns_1: 0.1405, loss_cls_2: 0.7846, loss_box_2: 1.3243, loss_cns_2: 0.6606, loss_yns_2: 0.1411, loss_cls_3: 0.7941, loss_box_3: 1.3228, loss_cns_3: 0.6580, loss_yns_3: 0.1406, loss_cls_4: 0.7898, loss_box_4: 1.3324, loss_cns_4: 0.6604, loss_yns_4: 0.1394, loss_cls_5: 0.7911, loss_box_5: 1.3341, loss_cns_5: 0.6546, loss_yns_5: 0.1403, loss_cls_dn_0: 0.1180, loss_box_dn_0: 0.7045, loss_cls_dn_1: 0.0904, loss_box_dn_1: 0.6318, loss_cls_dn_2: 0.0901, loss_box_dn_2: 0.6056, loss_cls_dn_3: 0.0935, loss_box_dn_3: 0.6080, loss_cls_dn_4: 0.0930, loss_box_dn_4: 0.6092, loss_cls_dn_5: 0.0921, loss_box_dn_5: 0.6173, loss_dense_depth: 0.7308, loss: 22.7045, grad_norm: 36.3796 -2026-01-14 21:42:30,910 - mmdet - INFO - Iter [396/17500] lr: 2.577e-04, eta: 9:27:52, time: 1.622, data_time: 0.079, memory: 49164, loss_cls_0: 0.6989, loss_box_0: 1.4980, loss_cns_0: 0.6376, loss_yns_0: 0.1423, loss_cls_1: 0.7827, loss_box_1: 1.3583, loss_cns_1: 0.6665, loss_yns_1: 0.1409, loss_cls_2: 0.7980, loss_box_2: 1.3231, loss_cns_2: 0.6631, loss_yns_2: 0.1424, loss_cls_3: 0.8028, loss_box_3: 1.3134, loss_cns_3: 0.6616, loss_yns_3: 0.1401, loss_cls_4: 0.7985, loss_box_4: 1.3192, loss_cns_4: 0.6644, loss_yns_4: 0.1395, loss_cls_5: 0.8030, loss_box_5: 1.3229, loss_cns_5: 0.6626, loss_yns_5: 0.1420, loss_cls_dn_0: 0.1125, loss_box_dn_0: 0.7116, loss_cls_dn_1: 0.0906, loss_box_dn_1: 0.6256, loss_cls_dn_2: 0.0908, loss_box_dn_2: 0.6118, loss_cls_dn_3: 0.0930, loss_box_dn_3: 0.6106, loss_cls_dn_4: 0.0912, loss_box_dn_4: 0.6131, loss_cls_dn_5: 0.0910, loss_box_dn_5: 0.6222, loss_dense_depth: 0.7031, loss: 22.6889, grad_norm: 31.9050 -2026-01-14 21:42:32,532 - mmdet - INFO - Iter [397/17500] lr: 2.581e-04, eta: 9:27:32, time: 1.576, data_time: 0.083, memory: 49164, loss_cls_0: 0.7218, loss_box_0: 1.5247, loss_cns_0: 0.6389, loss_yns_0: 0.1420, loss_cls_1: 0.7869, loss_box_1: 1.3497, loss_cns_1: 0.6646, loss_yns_1: 0.1419, loss_cls_2: 0.7956, loss_box_2: 1.3382, loss_cns_2: 0.6645, loss_yns_2: 0.1447, loss_cls_3: 0.8055, loss_box_3: 1.3250, loss_cns_3: 0.6635, loss_yns_3: 0.1410, loss_cls_4: 0.8105, loss_box_4: 1.3258, loss_cns_4: 0.6655, loss_yns_4: 0.1410, loss_cls_5: 0.8050, loss_box_5: 1.3288, loss_cns_5: 0.6648, loss_yns_5: 0.1419, loss_cls_dn_0: 0.1145, loss_box_dn_0: 0.6986, loss_cls_dn_1: 0.0925, loss_box_dn_1: 0.6216, loss_cls_dn_2: 0.0911, loss_box_dn_2: 0.6120, loss_cls_dn_3: 0.0922, loss_box_dn_3: 0.6109, loss_cls_dn_4: 0.0927, loss_box_dn_4: 0.6103, loss_cls_dn_5: 0.0953, loss_box_dn_5: 0.6167, loss_dense_depth: 0.6931, loss: 22.7731, grad_norm: 27.5271 -2026-01-14 21:42:34,096 - mmdet - INFO - Iter [398/17500] lr: 2.585e-04, eta: 9:27:14, time: 1.611, data_time: 0.113, memory: 49164, loss_cls_0: 0.7256, loss_box_0: 1.5417, loss_cns_0: 0.6341, loss_yns_0: 0.1418, loss_cls_1: 0.7922, loss_box_1: 1.3939, loss_cns_1: 0.6627, loss_yns_1: 0.1422, loss_cls_2: 0.7956, loss_box_2: 1.3580, loss_cns_2: 0.6621, loss_yns_2: 0.1405, loss_cls_3: 0.8046, loss_box_3: 1.3464, loss_cns_3: 0.6596, loss_yns_3: 0.1394, loss_cls_4: 0.8071, loss_box_4: 1.3441, loss_cns_4: 0.6604, loss_yns_4: 0.1397, loss_cls_5: 0.8183, loss_box_5: 1.3430, loss_cns_5: 0.6619, loss_yns_5: 0.1396, loss_cls_dn_0: 0.1143, loss_box_dn_0: 0.7040, loss_cls_dn_1: 0.0890, loss_box_dn_1: 0.6173, loss_cls_dn_2: 0.0870, loss_box_dn_2: 0.6013, loss_cls_dn_3: 0.0874, loss_box_dn_3: 0.5992, loss_cls_dn_4: 0.0886, loss_box_dn_4: 0.5974, loss_cls_dn_5: 0.0916, loss_box_dn_5: 0.5971, loss_dense_depth: 0.7661, loss: 22.8947, grad_norm: 28.3676 -2026-01-14 21:42:35,761 - mmdet - INFO - Iter [399/17500] lr: 2.589e-04, eta: 9:26:58, time: 1.663, data_time: 0.187, memory: 49164, loss_cls_0: 0.7145, loss_box_0: 1.5199, loss_cns_0: 0.6367, loss_yns_0: 0.1428, loss_cls_1: 0.7738, loss_box_1: 1.4117, loss_cns_1: 0.6622, loss_yns_1: 0.1404, loss_cls_2: 0.7866, loss_box_2: 1.3646, loss_cns_2: 0.6629, loss_yns_2: 0.1433, loss_cls_3: 0.7989, loss_box_3: 1.3574, loss_cns_3: 0.6632, loss_yns_3: 0.1411, loss_cls_4: 0.8031, loss_box_4: 1.3709, loss_cns_4: 0.6652, loss_yns_4: 0.1434, loss_cls_5: 0.8062, loss_box_5: 1.3591, loss_cns_5: 0.6651, loss_yns_5: 0.1451, loss_cls_dn_0: 0.1153, loss_box_dn_0: 0.7102, loss_cls_dn_1: 0.0904, loss_box_dn_1: 0.6252, loss_cls_dn_2: 0.0897, loss_box_dn_2: 0.5976, loss_cls_dn_3: 0.0886, loss_box_dn_3: 0.5959, loss_cls_dn_4: 0.0887, loss_box_dn_4: 0.6030, loss_cls_dn_5: 0.0891, loss_box_dn_5: 0.6024, loss_dense_depth: 0.7886, loss: 22.9628, grad_norm: 37.1123 -2026-01-14 21:42:39,077 - mmdet - INFO - Iter [400/17500] lr: 2.593e-04, eta: 9:27:53, time: 3.318, data_time: 0.076, memory: 49164, loss_cls_0: 0.7308, loss_box_0: 1.5196, loss_cns_0: 0.6407, loss_yns_0: 0.1440, loss_cls_1: 0.7880, loss_box_1: 1.3805, loss_cns_1: 0.6691, loss_yns_1: 0.1417, loss_cls_2: 0.7912, loss_box_2: 1.3411, loss_cns_2: 0.6659, loss_yns_2: 0.1464, loss_cls_3: 0.8008, loss_box_3: 1.3365, loss_cns_3: 0.6651, loss_yns_3: 0.1440, loss_cls_4: 0.8253, loss_box_4: 1.3365, loss_cns_4: 0.6625, loss_yns_4: 0.1437, loss_cls_5: 0.8173, loss_box_5: 1.3489, loss_cns_5: 0.6681, loss_yns_5: 0.1469, loss_cls_dn_0: 0.1145, loss_box_dn_0: 0.7144, loss_cls_dn_1: 0.0929, loss_box_dn_1: 0.6422, loss_cls_dn_2: 0.0921, loss_box_dn_2: 0.6331, loss_cls_dn_3: 0.0908, loss_box_dn_3: 0.6371, loss_cls_dn_4: 0.0926, loss_box_dn_4: 0.6498, loss_cls_dn_5: 0.0940, loss_box_dn_5: 0.6616, loss_dense_depth: 0.7182, loss: 23.0879, grad_norm: 37.1855 -2026-01-14 21:42:40,733 - mmdet - INFO - Iter [401/17500] lr: 2.597e-04, eta: 9:27:36, time: 1.657, data_time: 0.072, memory: 49164, loss_cls_0: 0.7383, loss_box_0: 1.5449, loss_cns_0: 0.6379, loss_yns_0: 0.1407, loss_cls_1: 0.7986, loss_box_1: 1.4092, loss_cns_1: 0.6648, loss_yns_1: 0.1407, loss_cls_2: 0.7957, loss_box_2: 1.3885, loss_cns_2: 0.6651, loss_yns_2: 0.1430, loss_cls_3: 0.7962, loss_box_3: 1.3703, loss_cns_3: 0.6642, loss_yns_3: 0.1396, loss_cls_4: 0.8211, loss_box_4: 1.3655, loss_cns_4: 0.6617, loss_yns_4: 0.1378, loss_cls_5: 0.8138, loss_box_5: 1.3864, loss_cns_5: 0.6650, loss_yns_5: 0.1395, loss_cls_dn_0: 0.1154, loss_box_dn_0: 0.7096, loss_cls_dn_1: 0.0963, loss_box_dn_1: 0.6544, loss_cls_dn_2: 0.0943, loss_box_dn_2: 0.6555, loss_cls_dn_3: 0.0934, loss_box_dn_3: 0.6581, loss_cls_dn_4: 0.0967, loss_box_dn_4: 0.6695, loss_cls_dn_5: 0.0953, loss_box_dn_5: 0.6842, loss_dense_depth: 0.8067, loss: 23.4578, grad_norm: 36.4028 -2026-01-14 21:42:42,379 - mmdet - INFO - Iter [402/17500] lr: 2.601e-04, eta: 9:27:19, time: 1.620, data_time: 0.084, memory: 49164, loss_cls_0: 0.7080, loss_box_0: 1.5428, loss_cns_0: 0.6419, loss_yns_0: 0.1411, loss_cls_1: 0.7684, loss_box_1: 1.4173, loss_cns_1: 0.6651, loss_yns_1: 0.1382, loss_cls_2: 0.7818, loss_box_2: 1.3979, loss_cns_2: 0.6685, loss_yns_2: 0.1390, loss_cls_3: 0.7970, loss_box_3: 1.3911, loss_cns_3: 0.6659, loss_yns_3: 0.1377, loss_cls_4: 0.7920, loss_box_4: 1.3980, loss_cns_4: 0.6660, loss_yns_4: 0.1398, loss_cls_5: 0.7884, loss_box_5: 1.4074, loss_cns_5: 0.6656, loss_yns_5: 0.1385, loss_cls_dn_0: 0.1149, loss_box_dn_0: 0.7091, loss_cls_dn_1: 0.0907, loss_box_dn_1: 0.6715, loss_cls_dn_2: 0.0898, loss_box_dn_2: 0.6676, loss_cls_dn_3: 0.0908, loss_box_dn_3: 0.6706, loss_cls_dn_4: 0.0923, loss_box_dn_4: 0.6781, loss_cls_dn_5: 0.0924, loss_box_dn_5: 0.6852, loss_dense_depth: 0.7326, loss: 23.3833, grad_norm: 40.9510 -2026-01-14 21:42:43,936 - mmdet - INFO - Iter [403/17500] lr: 2.605e-04, eta: 9:26:59, time: 1.582, data_time: 0.092, memory: 49164, loss_cls_0: 0.7227, loss_box_0: 1.5728, loss_cns_0: 0.6401, loss_yns_0: 0.1402, loss_cls_1: 0.7767, loss_box_1: 1.4259, loss_cns_1: 0.6672, loss_yns_1: 0.1367, loss_cls_2: 0.7978, loss_box_2: 1.4005, loss_cns_2: 0.6710, loss_yns_2: 0.1370, loss_cls_3: 0.8039, loss_box_3: 1.3839, loss_cns_3: 0.6666, loss_yns_3: 0.1361, loss_cls_4: 0.7938, loss_box_4: 1.3816, loss_cns_4: 0.6679, loss_yns_4: 0.1385, loss_cls_5: 0.7978, loss_box_5: 1.3856, loss_cns_5: 0.6676, loss_yns_5: 0.1371, loss_cls_dn_0: 0.1178, loss_box_dn_0: 0.7234, loss_cls_dn_1: 0.0899, loss_box_dn_1: 0.6515, loss_cls_dn_2: 0.0912, loss_box_dn_2: 0.6434, loss_cls_dn_3: 0.0926, loss_box_dn_3: 0.6398, loss_cls_dn_4: 0.0907, loss_box_dn_4: 0.6417, loss_cls_dn_5: 0.0922, loss_box_dn_5: 0.6457, loss_dense_depth: 0.7560, loss: 23.3251, grad_norm: 32.2789 -2026-01-14 21:42:45,536 - mmdet - INFO - Iter [404/17500] lr: 2.609e-04, eta: 9:26:41, time: 1.600, data_time: 0.074, memory: 49164, loss_cls_0: 0.7203, loss_box_0: 1.5571, loss_cns_0: 0.6386, loss_yns_0: 0.1389, loss_cls_1: 0.7683, loss_box_1: 1.4352, loss_cns_1: 0.6643, loss_yns_1: 0.1350, loss_cls_2: 0.7841, loss_box_2: 1.4001, loss_cns_2: 0.6633, loss_yns_2: 0.1347, loss_cls_3: 0.7851, loss_box_3: 1.3876, loss_cns_3: 0.6624, loss_yns_3: 0.1348, loss_cls_4: 0.7910, loss_box_4: 1.3913, loss_cns_4: 0.6616, loss_yns_4: 0.1362, loss_cls_5: 0.7975, loss_box_5: 1.3810, loss_cns_5: 0.6605, loss_yns_5: 0.1373, loss_cls_dn_0: 0.1116, loss_box_dn_0: 0.7116, loss_cls_dn_1: 0.0887, loss_box_dn_1: 0.6172, loss_cls_dn_2: 0.0883, loss_box_dn_2: 0.6034, loss_cls_dn_3: 0.0886, loss_box_dn_3: 0.5991, loss_cls_dn_4: 0.0889, loss_box_dn_4: 0.6015, loss_cls_dn_5: 0.0879, loss_box_dn_5: 0.6020, loss_dense_depth: 0.7833, loss: 23.0384, grad_norm: 44.4670 -2026-01-14 21:42:47,162 - mmdet - INFO - Iter [405/17500] lr: 2.613e-04, eta: 9:26:23, time: 1.625, data_time: 0.076, memory: 49164, loss_cls_0: 0.7336, loss_box_0: 1.5730, loss_cns_0: 0.6394, loss_yns_0: 0.1382, loss_cls_1: 0.7714, loss_box_1: 1.4314, loss_cns_1: 0.6599, loss_yns_1: 0.1353, loss_cls_2: 0.7914, loss_box_2: 1.4173, loss_cns_2: 0.6617, loss_yns_2: 0.1343, loss_cls_3: 0.8095, loss_box_3: 1.4004, loss_cns_3: 0.6590, loss_yns_3: 0.1345, loss_cls_4: 0.8011, loss_box_4: 1.3983, loss_cns_4: 0.6609, loss_yns_4: 0.1353, loss_cls_5: 0.8138, loss_box_5: 1.3833, loss_cns_5: 0.6552, loss_yns_5: 0.1352, loss_cls_dn_0: 0.1213, loss_box_dn_0: 0.7181, loss_cls_dn_1: 0.0875, loss_box_dn_1: 0.6084, loss_cls_dn_2: 0.0879, loss_box_dn_2: 0.6015, loss_cls_dn_3: 0.0893, loss_box_dn_3: 0.6032, loss_cls_dn_4: 0.0894, loss_box_dn_4: 0.6004, loss_cls_dn_5: 0.0896, loss_box_dn_5: 0.6044, loss_dense_depth: 0.7164, loss: 23.0906, grad_norm: 32.6955 -2026-01-14 21:42:48,841 - mmdet - INFO - Iter [406/17500] lr: 2.617e-04, eta: 9:26:07, time: 1.651, data_time: 0.078, memory: 49164, loss_cls_0: 0.7468, loss_box_0: 1.5557, loss_cns_0: 0.6414, loss_yns_0: 0.1372, loss_cls_1: 0.7786, loss_box_1: 1.4151, loss_cns_1: 0.6629, loss_yns_1: 0.1339, loss_cls_2: 0.7951, loss_box_2: 1.4124, loss_cns_2: 0.6691, loss_yns_2: 0.1356, loss_cls_3: 0.8078, loss_box_3: 1.3921, loss_cns_3: 0.6650, loss_yns_3: 0.1351, loss_cls_4: 0.8117, loss_box_4: 1.3885, loss_cns_4: 0.6666, loss_yns_4: 0.1341, loss_cls_5: 0.7930, loss_box_5: 1.4057, loss_cns_5: 0.6627, loss_yns_5: 0.1337, loss_cls_dn_0: 0.1206, loss_box_dn_0: 0.7133, loss_cls_dn_1: 0.0908, loss_box_dn_1: 0.6121, loss_cls_dn_2: 0.0916, loss_box_dn_2: 0.6064, loss_cls_dn_3: 0.0918, loss_box_dn_3: 0.6067, loss_cls_dn_4: 0.0922, loss_box_dn_4: 0.6123, loss_cls_dn_5: 0.0909, loss_box_dn_5: 0.6259, loss_dense_depth: 0.7518, loss: 23.1863, grad_norm: 41.3607 -2026-01-14 21:42:50,409 - mmdet - INFO - Iter [407/17500] lr: 2.621e-04, eta: 9:25:49, time: 1.599, data_time: 0.101, memory: 49164, loss_cls_0: 0.7397, loss_box_0: 1.5698, loss_cns_0: 0.6388, loss_yns_0: 0.1336, loss_cls_1: 0.7900, loss_box_1: 1.4007, loss_cns_1: 0.6658, loss_yns_1: 0.1315, loss_cls_2: 0.7966, loss_box_2: 1.3778, loss_cns_2: 0.6694, loss_yns_2: 0.1319, loss_cls_3: 0.8106, loss_box_3: 1.3640, loss_cns_3: 0.6665, loss_yns_3: 0.1321, loss_cls_4: 0.8180, loss_box_4: 1.3772, loss_cns_4: 0.6686, loss_yns_4: 0.1325, loss_cls_5: 0.8139, loss_box_5: 1.3685, loss_cns_5: 0.6674, loss_yns_5: 0.1321, loss_cls_dn_0: 0.1189, loss_box_dn_0: 0.7190, loss_cls_dn_1: 0.0921, loss_box_dn_1: 0.6359, loss_cls_dn_2: 0.0912, loss_box_dn_2: 0.6232, loss_cls_dn_3: 0.0914, loss_box_dn_3: 0.6258, loss_cls_dn_4: 0.0920, loss_box_dn_4: 0.6404, loss_cls_dn_5: 0.0910, loss_box_dn_5: 0.6483, loss_dense_depth: 0.7321, loss: 23.1983, grad_norm: 31.4799 -2026-01-14 21:42:51,973 - mmdet - INFO - Iter [408/17500] lr: 2.624e-04, eta: 9:25:29, time: 1.564, data_time: 0.076, memory: 49164, loss_cls_0: 0.7256, loss_box_0: 1.5420, loss_cns_0: 0.6356, loss_yns_0: 0.1380, loss_cls_1: 0.7821, loss_box_1: 1.3777, loss_cns_1: 0.6669, loss_yns_1: 0.1350, loss_cls_2: 0.7947, loss_box_2: 1.3529, loss_cns_2: 0.6670, loss_yns_2: 0.1346, loss_cls_3: 0.8058, loss_box_3: 1.3367, loss_cns_3: 0.6708, loss_yns_3: 0.1348, loss_cls_4: 0.8041, loss_box_4: 1.3424, loss_cns_4: 0.6704, loss_yns_4: 0.1362, loss_cls_5: 0.7967, loss_box_5: 1.3384, loss_cns_5: 0.6697, loss_yns_5: 0.1376, loss_cls_dn_0: 0.1167, loss_box_dn_0: 0.7183, loss_cls_dn_1: 0.0894, loss_box_dn_1: 0.6435, loss_cls_dn_2: 0.0886, loss_box_dn_2: 0.6308, loss_cls_dn_3: 0.0886, loss_box_dn_3: 0.6277, loss_cls_dn_4: 0.0915, loss_box_dn_4: 0.6334, loss_cls_dn_5: 0.0895, loss_box_dn_5: 0.6399, loss_dense_depth: 0.7298, loss: 22.9833, grad_norm: 27.6146 -2026-01-14 21:42:53,574 - mmdet - INFO - Iter [409/17500] lr: 2.628e-04, eta: 9:25:10, time: 1.570, data_time: 0.076, memory: 49164, loss_cls_0: 0.7134, loss_box_0: 1.5499, loss_cns_0: 0.6393, loss_yns_0: 0.1389, loss_cls_1: 0.7860, loss_box_1: 1.4180, loss_cns_1: 0.6657, loss_yns_1: 0.1367, loss_cls_2: 0.7888, loss_box_2: 1.3861, loss_cns_2: 0.6639, loss_yns_2: 0.1364, loss_cls_3: 0.8010, loss_box_3: 1.3704, loss_cns_3: 0.6633, loss_yns_3: 0.1352, loss_cls_4: 0.8014, loss_box_4: 1.3833, loss_cns_4: 0.6640, loss_yns_4: 0.1361, loss_cls_5: 0.7945, loss_box_5: 1.3975, loss_cns_5: 0.6654, loss_yns_5: 0.1369, loss_cls_dn_0: 0.1150, loss_box_dn_0: 0.7077, loss_cls_dn_1: 0.0910, loss_box_dn_1: 0.6304, loss_cls_dn_2: 0.0908, loss_box_dn_2: 0.6143, loss_cls_dn_3: 0.0910, loss_box_dn_3: 0.6112, loss_cls_dn_4: 0.0926, loss_box_dn_4: 0.6157, loss_cls_dn_5: 0.0931, loss_box_dn_5: 0.6197, loss_dense_depth: 0.7197, loss: 23.0645, grad_norm: 35.8735 -2026-01-14 21:42:55,178 - mmdet - INFO - Iter [410/17500] lr: 2.632e-04, eta: 9:24:52, time: 1.606, data_time: 0.101, memory: 49164, loss_cls_0: 0.7164, loss_box_0: 1.5369, loss_cns_0: 0.6452, loss_yns_0: 0.1410, loss_cls_1: 0.7803, loss_box_1: 1.3794, loss_cns_1: 0.6663, loss_yns_1: 0.1379, loss_cls_2: 0.7771, loss_box_2: 1.3689, loss_cns_2: 0.6674, loss_yns_2: 0.1367, loss_cls_3: 0.7929, loss_box_3: 1.3505, loss_cns_3: 0.6666, loss_yns_3: 0.1372, loss_cls_4: 0.7865, loss_box_4: 1.3551, loss_cns_4: 0.6663, loss_yns_4: 0.1377, loss_cls_5: 0.8005, loss_box_5: 1.3635, loss_cns_5: 0.6688, loss_yns_5: 0.1383, loss_cls_dn_0: 0.1113, loss_box_dn_0: 0.7073, loss_cls_dn_1: 0.0909, loss_box_dn_1: 0.6101, loss_cls_dn_2: 0.0901, loss_box_dn_2: 0.5996, loss_cls_dn_3: 0.0905, loss_box_dn_3: 0.5917, loss_cls_dn_4: 0.0904, loss_box_dn_4: 0.5921, loss_cls_dn_5: 0.0930, loss_box_dn_5: 0.5926, loss_dense_depth: 0.7339, loss: 22.8107, grad_norm: 27.4775 -2026-01-14 21:42:56,776 - mmdet - INFO - Iter [411/17500] lr: 2.636e-04, eta: 9:24:34, time: 1.599, data_time: 0.099, memory: 49164, loss_cls_0: 0.7240, loss_box_0: 1.5446, loss_cns_0: 0.6399, loss_yns_0: 0.1379, loss_cls_1: 0.7823, loss_box_1: 1.3796, loss_cns_1: 0.6665, loss_yns_1: 0.1359, loss_cls_2: 0.7939, loss_box_2: 1.3835, loss_cns_2: 0.6693, loss_yns_2: 0.1351, loss_cls_3: 0.8052, loss_box_3: 1.3553, loss_cns_3: 0.6661, loss_yns_3: 0.1352, loss_cls_4: 0.8032, loss_box_4: 1.3591, loss_cns_4: 0.6665, loss_yns_4: 0.1359, loss_cls_5: 0.8169, loss_box_5: 1.3729, loss_cns_5: 0.6673, loss_yns_5: 0.1359, loss_cls_dn_0: 0.1144, loss_box_dn_0: 0.7151, loss_cls_dn_1: 0.0893, loss_box_dn_1: 0.6240, loss_cls_dn_2: 0.0878, loss_box_dn_2: 0.6195, loss_cls_dn_3: 0.0892, loss_box_dn_3: 0.6101, loss_cls_dn_4: 0.0891, loss_box_dn_4: 0.6125, loss_cls_dn_5: 0.0904, loss_box_dn_5: 0.6228, loss_dense_depth: 0.7193, loss: 22.9955, grad_norm: 31.4195 -2026-01-14 21:42:58,382 - mmdet - INFO - Iter [412/17500] lr: 2.640e-04, eta: 9:24:17, time: 1.605, data_time: 0.098, memory: 49164, loss_cls_0: 0.7091, loss_box_0: 1.5206, loss_cns_0: 0.6427, loss_yns_0: 0.1362, loss_cls_1: 0.7809, loss_box_1: 1.3801, loss_cns_1: 0.6675, loss_yns_1: 0.1343, loss_cls_2: 0.7805, loss_box_2: 1.3721, loss_cns_2: 0.6697, loss_yns_2: 0.1357, loss_cls_3: 0.7881, loss_box_3: 1.3479, loss_cns_3: 0.6658, loss_yns_3: 0.1352, loss_cls_4: 0.7922, loss_box_4: 1.3506, loss_cns_4: 0.6658, loss_yns_4: 0.1361, loss_cls_5: 0.7961, loss_box_5: 1.3646, loss_cns_5: 0.6674, loss_yns_5: 0.1357, loss_cls_dn_0: 0.1139, loss_box_dn_0: 0.7212, loss_cls_dn_1: 0.0904, loss_box_dn_1: 0.6323, loss_cls_dn_2: 0.0886, loss_box_dn_2: 0.6219, loss_cls_dn_3: 0.0902, loss_box_dn_3: 0.6157, loss_cls_dn_4: 0.0898, loss_box_dn_4: 0.6213, loss_cls_dn_5: 0.0905, loss_box_dn_5: 0.6289, loss_dense_depth: 0.6906, loss: 22.8704, grad_norm: 32.1781 -2026-01-14 21:42:59,953 - mmdet - INFO - Iter [413/17500] lr: 2.644e-04, eta: 9:23:59, time: 1.601, data_time: 0.104, memory: 49164, loss_cls_0: 0.7258, loss_box_0: 1.5256, loss_cns_0: 0.6374, loss_yns_0: 0.1381, loss_cls_1: 0.7928, loss_box_1: 1.3761, loss_cns_1: 0.6598, loss_yns_1: 0.1368, loss_cls_2: 0.7787, loss_box_2: 1.3642, loss_cns_2: 0.6616, loss_yns_2: 0.1365, loss_cls_3: 0.7869, loss_box_3: 1.3525, loss_cns_3: 0.6626, loss_yns_3: 0.1369, loss_cls_4: 0.7911, loss_box_4: 1.3411, loss_cns_4: 0.6586, loss_yns_4: 0.1363, loss_cls_5: 0.8187, loss_box_5: 1.3299, loss_cns_5: 0.6583, loss_yns_5: 0.1352, loss_cls_dn_0: 0.1092, loss_box_dn_0: 0.7089, loss_cls_dn_1: 0.0903, loss_box_dn_1: 0.6356, loss_cls_dn_2: 0.0876, loss_box_dn_2: 0.6213, loss_cls_dn_3: 0.0869, loss_box_dn_3: 0.6187, loss_cls_dn_4: 0.0879, loss_box_dn_4: 0.6195, loss_cls_dn_5: 0.0897, loss_box_dn_5: 0.6190, loss_dense_depth: 0.7265, loss: 22.8427, grad_norm: 31.1645 -2026-01-14 21:43:01,554 - mmdet - INFO - Iter [414/17500] lr: 2.648e-04, eta: 9:23:40, time: 1.570, data_time: 0.071, memory: 49164, loss_cls_0: 0.7134, loss_box_0: 1.5649, loss_cns_0: 0.6463, loss_yns_0: 0.1422, loss_cls_1: 0.7682, loss_box_1: 1.3974, loss_cns_1: 0.6654, loss_yns_1: 0.1384, loss_cls_2: 0.7667, loss_box_2: 1.3751, loss_cns_2: 0.6639, loss_yns_2: 0.1381, loss_cls_3: 0.7754, loss_box_3: 1.3648, loss_cns_3: 0.6648, loss_yns_3: 0.1375, loss_cls_4: 0.7810, loss_box_4: 1.3570, loss_cns_4: 0.6631, loss_yns_4: 0.1392, loss_cls_5: 0.7899, loss_box_5: 1.3675, loss_cns_5: 0.6634, loss_yns_5: 0.1376, loss_cls_dn_0: 0.1139, loss_box_dn_0: 0.7089, loss_cls_dn_1: 0.0912, loss_box_dn_1: 0.6285, loss_cls_dn_2: 0.0893, loss_box_dn_2: 0.6123, loss_cls_dn_3: 0.0896, loss_box_dn_3: 0.6071, loss_cls_dn_4: 0.0899, loss_box_dn_4: 0.6062, loss_cls_dn_5: 0.0909, loss_box_dn_5: 0.6141, loss_dense_depth: 0.7254, loss: 22.8887, grad_norm: 29.4553 -2026-01-14 21:43:03,186 - mmdet - INFO - Iter [415/17500] lr: 2.652e-04, eta: 9:23:23, time: 1.616, data_time: 0.089, memory: 49164, loss_cls_0: 0.7297, loss_box_0: 1.5368, loss_cns_0: 0.6488, loss_yns_0: 0.1420, loss_cls_1: 0.7797, loss_box_1: 1.3831, loss_cns_1: 0.6657, loss_yns_1: 0.1369, loss_cls_2: 0.7796, loss_box_2: 1.3643, loss_cns_2: 0.6641, loss_yns_2: 0.1359, loss_cls_3: 0.7909, loss_box_3: 1.3679, loss_cns_3: 0.6636, loss_yns_3: 0.1369, loss_cls_4: 0.7991, loss_box_4: 1.3485, loss_cns_4: 0.6627, loss_yns_4: 0.1357, loss_cls_5: 0.8105, loss_box_5: 1.3443, loss_cns_5: 0.6630, loss_yns_5: 0.1356, loss_cls_dn_0: 0.1152, loss_box_dn_0: 0.7079, loss_cls_dn_1: 0.0912, loss_box_dn_1: 0.6200, loss_cls_dn_2: 0.0895, loss_box_dn_2: 0.6010, loss_cls_dn_3: 0.0909, loss_box_dn_3: 0.6008, loss_cls_dn_4: 0.0898, loss_box_dn_4: 0.5959, loss_cls_dn_5: 0.0922, loss_box_dn_5: 0.5986, loss_dense_depth: 0.6975, loss: 22.8161, grad_norm: 40.1182 -2026-01-14 21:43:04,775 - mmdet - INFO - Iter [416/17500] lr: 2.656e-04, eta: 9:23:06, time: 1.605, data_time: 0.112, memory: 49164, loss_cls_0: 0.7142, loss_box_0: 1.5574, loss_cns_0: 0.6457, loss_yns_0: 0.1432, loss_cls_1: 0.7851, loss_box_1: 1.3821, loss_cns_1: 0.6656, loss_yns_1: 0.1377, loss_cls_2: 0.7774, loss_box_2: 1.3711, loss_cns_2: 0.6643, loss_yns_2: 0.1379, loss_cls_3: 0.7850, loss_box_3: 1.3622, loss_cns_3: 0.6631, loss_yns_3: 0.1390, loss_cls_4: 0.7982, loss_box_4: 1.3569, loss_cns_4: 0.6638, loss_yns_4: 0.1395, loss_cls_5: 0.7993, loss_box_5: 1.3588, loss_cns_5: 0.6643, loss_yns_5: 0.1391, loss_cls_dn_0: 0.1131, loss_box_dn_0: 0.7068, loss_cls_dn_1: 0.0939, loss_box_dn_1: 0.6107, loss_cls_dn_2: 0.0916, loss_box_dn_2: 0.5978, loss_cls_dn_3: 0.0931, loss_box_dn_3: 0.5984, loss_cls_dn_4: 0.0948, loss_box_dn_4: 0.6009, loss_cls_dn_5: 0.0948, loss_box_dn_5: 0.6031, loss_dense_depth: 0.7129, loss: 22.8624, grad_norm: 32.5011 -2026-01-14 21:43:06,338 - mmdet - INFO - Iter [417/17500] lr: 2.660e-04, eta: 9:22:48, time: 1.592, data_time: 0.098, memory: 49164, loss_cls_0: 0.7231, loss_box_0: 1.5645, loss_cns_0: 0.6411, loss_yns_0: 0.1423, loss_cls_1: 0.7915, loss_box_1: 1.3805, loss_cns_1: 0.6660, loss_yns_1: 0.1392, loss_cls_2: 0.7976, loss_box_2: 1.3731, loss_cns_2: 0.6646, loss_yns_2: 0.1381, loss_cls_3: 0.8134, loss_box_3: 1.3708, loss_cns_3: 0.6666, loss_yns_3: 0.1383, loss_cls_4: 0.8207, loss_box_4: 1.3697, loss_cns_4: 0.6662, loss_yns_4: 0.1387, loss_cls_5: 0.8006, loss_box_5: 1.3710, loss_cns_5: 0.6665, loss_yns_5: 0.1391, loss_cls_dn_0: 0.1148, loss_box_dn_0: 0.7074, loss_cls_dn_1: 0.0965, loss_box_dn_1: 0.6162, loss_cls_dn_2: 0.0943, loss_box_dn_2: 0.6070, loss_cls_dn_3: 0.0965, loss_box_dn_3: 0.6099, loss_cls_dn_4: 0.0978, loss_box_dn_4: 0.6123, loss_cls_dn_5: 0.0985, loss_box_dn_5: 0.6154, loss_dense_depth: 0.7113, loss: 23.0610, grad_norm: 37.2775 -2026-01-14 21:43:07,905 - mmdet - INFO - Iter [418/17500] lr: 2.664e-04, eta: 9:22:29, time: 1.567, data_time: 0.080, memory: 49164, loss_cls_0: 0.7005, loss_box_0: 1.5646, loss_cns_0: 0.6403, loss_yns_0: 0.1438, loss_cls_1: 0.7699, loss_box_1: 1.4017, loss_cns_1: 0.6620, loss_yns_1: 0.1422, loss_cls_2: 0.7669, loss_box_2: 1.3902, loss_cns_2: 0.6657, loss_yns_2: 0.1410, loss_cls_3: 0.7770, loss_box_3: 1.3699, loss_cns_3: 0.6624, loss_yns_3: 0.1411, loss_cls_4: 0.7809, loss_box_4: 1.3721, loss_cns_4: 0.6647, loss_yns_4: 0.1436, loss_cls_5: 0.7819, loss_box_5: 1.3775, loss_cns_5: 0.6636, loss_yns_5: 0.1429, loss_cls_dn_0: 0.1082, loss_box_dn_0: 0.7016, loss_cls_dn_1: 0.0917, loss_box_dn_1: 0.6244, loss_cls_dn_2: 0.0904, loss_box_dn_2: 0.6118, loss_cls_dn_3: 0.0898, loss_box_dn_3: 0.6115, loss_cls_dn_4: 0.0904, loss_box_dn_4: 0.6094, loss_cls_dn_5: 0.0942, loss_box_dn_5: 0.6150, loss_dense_depth: 0.7022, loss: 22.9070, grad_norm: 28.8308 -2026-01-14 21:43:09,595 - mmdet - INFO - Iter [419/17500] lr: 2.668e-04, eta: 9:22:14, time: 1.651, data_time: 0.181, memory: 49164, loss_cls_0: 0.7538, loss_box_0: 1.6111, loss_cns_0: 0.6361, loss_yns_0: 0.1416, loss_cls_1: 0.7924, loss_box_1: 1.4553, loss_cns_1: 0.6649, loss_yns_1: 0.1417, loss_cls_2: 0.8085, loss_box_2: 1.4287, loss_cns_2: 0.6661, loss_yns_2: 0.1396, loss_cls_3: 0.8254, loss_box_3: 1.4081, loss_cns_3: 0.6658, loss_yns_3: 0.1393, loss_cls_4: 0.8332, loss_box_4: 1.4038, loss_cns_4: 0.6660, loss_yns_4: 0.1397, loss_cls_5: 0.8360, loss_box_5: 1.4022, loss_cns_5: 0.6654, loss_yns_5: 0.1401, loss_cls_dn_0: 0.1152, loss_box_dn_0: 0.7034, loss_cls_dn_1: 0.0913, loss_box_dn_1: 0.6312, loss_cls_dn_2: 0.0907, loss_box_dn_2: 0.6144, loss_cls_dn_3: 0.0893, loss_box_dn_3: 0.6123, loss_cls_dn_4: 0.0908, loss_box_dn_4: 0.6109, loss_cls_dn_5: 0.0913, loss_box_dn_5: 0.6081, loss_dense_depth: 0.7417, loss: 23.4554, grad_norm: 28.0538 -2026-01-14 21:43:11,174 - mmdet - INFO - Iter [420/17500] lr: 2.672e-04, eta: 9:21:57, time: 1.613, data_time: 0.123, memory: 49164, loss_cls_0: 0.7144, loss_box_0: 1.5786, loss_cns_0: 0.6378, loss_yns_0: 0.1405, loss_cls_1: 0.7948, loss_box_1: 1.4044, loss_cns_1: 0.6651, loss_yns_1: 0.1381, loss_cls_2: 0.8039, loss_box_2: 1.3753, loss_cns_2: 0.6654, loss_yns_2: 0.1375, loss_cls_3: 0.8110, loss_box_3: 1.3683, loss_cns_3: 0.6653, loss_yns_3: 0.1367, loss_cls_4: 0.8167, loss_box_4: 1.3787, loss_cns_4: 0.6658, loss_yns_4: 0.1362, loss_cls_5: 0.8039, loss_box_5: 1.3712, loss_cns_5: 0.6643, loss_yns_5: 0.1379, loss_cls_dn_0: 0.1105, loss_box_dn_0: 0.7047, loss_cls_dn_1: 0.0931, loss_box_dn_1: 0.6247, loss_cls_dn_2: 0.0946, loss_box_dn_2: 0.6077, loss_cls_dn_3: 0.0954, loss_box_dn_3: 0.6063, loss_cls_dn_4: 0.0965, loss_box_dn_4: 0.6108, loss_cls_dn_5: 0.0943, loss_box_dn_5: 0.6094, loss_dense_depth: 0.7321, loss: 23.0918, grad_norm: 25.8670 -2026-01-14 21:43:12,851 - mmdet - INFO - Iter [421/17500] lr: 2.676e-04, eta: 9:21:43, time: 1.666, data_time: 0.076, memory: 49164, loss_cls_0: 0.7431, loss_box_0: 1.5927, loss_cns_0: 0.6286, loss_yns_0: 0.1350, loss_cls_1: 0.7868, loss_box_1: 1.4167, loss_cns_1: 0.6588, loss_yns_1: 0.1362, loss_cls_2: 0.8000, loss_box_2: 1.3801, loss_cns_2: 0.6598, loss_yns_2: 0.1361, loss_cls_3: 0.8121, loss_box_3: 1.3838, loss_cns_3: 0.6594, loss_yns_3: 0.1357, loss_cls_4: 0.8108, loss_box_4: 1.3797, loss_cns_4: 0.6623, loss_yns_4: 0.1355, loss_cls_5: 0.8272, loss_box_5: 1.3665, loss_cns_5: 0.6593, loss_yns_5: 0.1360, loss_cls_dn_0: 0.1114, loss_box_dn_0: 0.7022, loss_cls_dn_1: 0.0903, loss_box_dn_1: 0.6392, loss_cls_dn_2: 0.0912, loss_box_dn_2: 0.6170, loss_cls_dn_3: 0.0942, loss_box_dn_3: 0.6185, loss_cls_dn_4: 0.0905, loss_box_dn_4: 0.6211, loss_cls_dn_5: 0.0889, loss_box_dn_5: 0.6263, loss_dense_depth: 0.7180, loss: 23.1510, grad_norm: 29.7100 -2026-01-14 21:43:14,488 - mmdet - INFO - Iter [422/17500] lr: 2.680e-04, eta: 9:21:28, time: 1.653, data_time: 0.098, memory: 49164, loss_cls_0: 0.7065, loss_box_0: 1.5884, loss_cns_0: 0.6413, loss_yns_0: 0.1379, loss_cls_1: 0.7759, loss_box_1: 1.4018, loss_cns_1: 0.6630, loss_yns_1: 0.1328, loss_cls_2: 0.7892, loss_box_2: 1.3697, loss_cns_2: 0.6630, loss_yns_2: 0.1339, loss_cls_3: 0.7983, loss_box_3: 1.3767, loss_cns_3: 0.6625, loss_yns_3: 0.1331, loss_cls_4: 0.8030, loss_box_4: 1.3815, loss_cns_4: 0.6618, loss_yns_4: 0.1353, loss_cls_5: 0.7968, loss_box_5: 1.3902, loss_cns_5: 0.6616, loss_yns_5: 0.1342, loss_cls_dn_0: 0.1139, loss_box_dn_0: 0.7124, loss_cls_dn_1: 0.0899, loss_box_dn_1: 0.6313, loss_cls_dn_2: 0.0892, loss_box_dn_2: 0.6095, loss_cls_dn_3: 0.0909, loss_box_dn_3: 0.6109, loss_cls_dn_4: 0.0892, loss_box_dn_4: 0.6199, loss_cls_dn_5: 0.0872, loss_box_dn_5: 0.6293, loss_dense_depth: 0.7049, loss: 23.0169, grad_norm: 39.6796 -2026-01-14 21:43:16,060 - mmdet - INFO - Iter [423/17500] lr: 2.684e-04, eta: 9:21:10, time: 1.571, data_time: 0.074, memory: 49164, loss_cls_0: 0.7115, loss_box_0: 1.5748, loss_cns_0: 0.6440, loss_yns_0: 0.1377, loss_cls_1: 0.7744, loss_box_1: 1.3885, loss_cns_1: 0.6683, loss_yns_1: 0.1363, loss_cls_2: 0.7777, loss_box_2: 1.3581, loss_cns_2: 0.6662, loss_yns_2: 0.1349, loss_cls_3: 0.7839, loss_box_3: 1.3439, loss_cns_3: 0.6662, loss_yns_3: 0.1354, loss_cls_4: 0.7828, loss_box_4: 1.3564, loss_cns_4: 0.6655, loss_yns_4: 0.1395, loss_cls_5: 0.7875, loss_box_5: 1.3623, loss_cns_5: 0.6644, loss_yns_5: 0.1390, loss_cls_dn_0: 0.1164, loss_box_dn_0: 0.7077, loss_cls_dn_1: 0.0917, loss_box_dn_1: 0.6272, loss_cls_dn_2: 0.0924, loss_box_dn_2: 0.6107, loss_cls_dn_3: 0.0927, loss_box_dn_3: 0.6072, loss_cls_dn_4: 0.0951, loss_box_dn_4: 0.6158, loss_cls_dn_5: 0.0928, loss_box_dn_5: 0.6172, loss_dense_depth: 0.7002, loss: 22.8663, grad_norm: 31.4051 -2026-01-14 21:43:17,762 - mmdet - INFO - Iter [424/17500] lr: 2.688e-04, eta: 9:20:56, time: 1.669, data_time: 0.073, memory: 49164, loss_cls_0: 0.7414, loss_box_0: 1.6203, loss_cns_0: 0.6422, loss_yns_0: 0.1384, loss_cls_1: 0.7807, loss_box_1: 1.4498, loss_cns_1: 0.6655, loss_yns_1: 0.1341, loss_cls_2: 0.7841, loss_box_2: 1.4298, loss_cns_2: 0.6657, loss_yns_2: 0.1344, loss_cls_3: 0.7877, loss_box_3: 1.4051, loss_cns_3: 0.6648, loss_yns_3: 0.1331, loss_cls_4: 0.8043, loss_box_4: 1.3979, loss_cns_4: 0.6618, loss_yns_4: 0.1320, loss_cls_5: 0.8022, loss_box_5: 1.3974, loss_cns_5: 0.6615, loss_yns_5: 0.1332, loss_cls_dn_0: 0.1178, loss_box_dn_0: 0.7075, loss_cls_dn_1: 0.0956, loss_box_dn_1: 0.6391, loss_cls_dn_2: 0.0946, loss_box_dn_2: 0.6278, loss_cls_dn_3: 0.0957, loss_box_dn_3: 0.6184, loss_cls_dn_4: 0.1021, loss_box_dn_4: 0.6171, loss_cls_dn_5: 0.0976, loss_box_dn_5: 0.6166, loss_dense_depth: 0.7284, loss: 23.3256, grad_norm: 43.0168 -2026-01-14 21:43:20,884 - mmdet - INFO - Iter [425/17500] lr: 2.692e-04, eta: 9:20:41, time: 1.666, data_time: 0.094, memory: 49164, loss_cls_0: 0.7094, loss_box_0: 1.5942, loss_cns_0: 0.6452, loss_yns_0: 0.1363, loss_cls_1: 0.7691, loss_box_1: 1.3820, loss_cns_1: 0.6652, loss_yns_1: 0.1302, loss_cls_2: 0.7831, loss_box_2: 1.3867, loss_cns_2: 0.6682, loss_yns_2: 0.1292, loss_cls_3: 0.7891, loss_box_3: 1.3736, loss_cns_3: 0.6667, loss_yns_3: 0.1295, loss_cls_4: 0.8023, loss_box_4: 1.3692, loss_cns_4: 0.6686, loss_yns_4: 0.1291, loss_cls_5: 0.7973, loss_box_5: 1.3828, loss_cns_5: 0.6703, loss_yns_5: 0.1302, loss_cls_dn_0: 0.1155, loss_box_dn_0: 0.7119, loss_cls_dn_1: 0.0955, loss_box_dn_1: 0.6316, loss_cls_dn_2: 0.0932, loss_box_dn_2: 0.6234, loss_cls_dn_3: 0.0937, loss_box_dn_3: 0.6236, loss_cls_dn_4: 0.0962, loss_box_dn_4: 0.6254, loss_cls_dn_5: 0.0939, loss_box_dn_5: 0.6332, loss_dense_depth: 0.6987, loss: 23.0433, grad_norm: 39.4834 -2026-01-14 21:43:22,560 - mmdet - INFO - Iter [426/17500] lr: 2.696e-04, eta: 9:21:27, time: 3.165, data_time: 1.541, memory: 49164, loss_cls_0: 0.7126, loss_box_0: 1.5549, loss_cns_0: 0.6409, loss_yns_0: 0.1394, loss_cls_1: 0.7709, loss_box_1: 1.4167, loss_cns_1: 0.6664, loss_yns_1: 0.1393, loss_cls_2: 0.7897, loss_box_2: 1.3957, loss_cns_2: 0.6674, loss_yns_2: 0.1382, loss_cls_3: 0.7993, loss_box_3: 1.3944, loss_cns_3: 0.6672, loss_yns_3: 0.1362, loss_cls_4: 0.7974, loss_box_4: 1.3768, loss_cns_4: 0.6676, loss_yns_4: 0.1357, loss_cls_5: 0.8008, loss_box_5: 1.3769, loss_cns_5: 0.6676, loss_yns_5: 0.1346, loss_cls_dn_0: 0.1170, loss_box_dn_0: 0.7105, loss_cls_dn_1: 0.0909, loss_box_dn_1: 0.6331, loss_cls_dn_2: 0.0898, loss_box_dn_2: 0.6222, loss_cls_dn_3: 0.0898, loss_box_dn_3: 0.6329, loss_cls_dn_4: 0.0932, loss_box_dn_4: 0.6332, loss_cls_dn_5: 0.0937, loss_box_dn_5: 0.6388, loss_dense_depth: 0.6948, loss: 23.1266, grad_norm: 41.0770 -2026-01-14 21:43:24,104 - mmdet - INFO - Iter [427/17500] lr: 2.700e-04, eta: 9:21:08, time: 1.544, data_time: 0.071, memory: 49164, loss_cls_0: 0.7109, loss_box_0: 1.5629, loss_cns_0: 0.6381, loss_yns_0: 0.1380, loss_cls_1: 0.7824, loss_box_1: 1.3972, loss_cns_1: 0.6649, loss_yns_1: 0.1361, loss_cls_2: 0.7872, loss_box_2: 1.3881, loss_cns_2: 0.6657, loss_yns_2: 0.1356, loss_cls_3: 0.8067, loss_box_3: 1.3807, loss_cns_3: 0.6655, loss_yns_3: 0.1338, loss_cls_4: 0.8119, loss_box_4: 1.3678, loss_cns_4: 0.6631, loss_yns_4: 0.1343, loss_cls_5: 0.8081, loss_box_5: 1.3780, loss_cns_5: 0.6667, loss_yns_5: 0.1337, loss_cls_dn_0: 0.1197, loss_box_dn_0: 0.7202, loss_cls_dn_1: 0.0930, loss_box_dn_1: 0.6392, loss_cls_dn_2: 0.0920, loss_box_dn_2: 0.6366, loss_cls_dn_3: 0.0954, loss_box_dn_3: 0.6368, loss_cls_dn_4: 0.0952, loss_box_dn_4: 0.6415, loss_cls_dn_5: 0.0936, loss_box_dn_5: 0.6472, loss_dense_depth: 0.6974, loss: 23.1651, grad_norm: 40.7147 -2026-01-14 21:43:25,668 - mmdet - INFO - Iter [428/17500] lr: 2.704e-04, eta: 9:20:49, time: 1.553, data_time: 0.071, memory: 49164, loss_cls_0: 0.7092, loss_box_0: 1.5703, loss_cns_0: 0.6411, loss_yns_0: 0.1432, loss_cls_1: 0.7620, loss_box_1: 1.4192, loss_cns_1: 0.6638, loss_yns_1: 0.1382, loss_cls_2: 0.7745, loss_box_2: 1.3974, loss_cns_2: 0.6648, loss_yns_2: 0.1381, loss_cls_3: 0.7860, loss_box_3: 1.3677, loss_cns_3: 0.6646, loss_yns_3: 0.1380, loss_cls_4: 0.8019, loss_box_4: 1.3542, loss_cns_4: 0.6636, loss_yns_4: 0.1383, loss_cls_5: 0.7987, loss_box_5: 1.3684, loss_cns_5: 0.6665, loss_yns_5: 0.1382, loss_cls_dn_0: 0.1182, loss_box_dn_0: 0.7146, loss_cls_dn_1: 0.0946, loss_box_dn_1: 0.6486, loss_cls_dn_2: 0.0939, loss_box_dn_2: 0.6447, loss_cls_dn_3: 0.0981, loss_box_dn_3: 0.6341, loss_cls_dn_4: 0.0966, loss_box_dn_4: 0.6339, loss_cls_dn_5: 0.0966, loss_box_dn_5: 0.6408, loss_dense_depth: 0.7043, loss: 23.1271, grad_norm: 30.5194 -2026-01-14 21:43:27,251 - mmdet - INFO - Iter [429/17500] lr: 2.708e-04, eta: 9:20:31, time: 1.557, data_time: 0.077, memory: 49164, loss_cls_0: 0.6994, loss_box_0: 1.5744, loss_cns_0: 0.6403, loss_yns_0: 0.1405, loss_cls_1: 0.7474, loss_box_1: 1.4100, loss_cns_1: 0.6606, loss_yns_1: 0.1369, loss_cls_2: 0.7684, loss_box_2: 1.3920, loss_cns_2: 0.6607, loss_yns_2: 0.1344, loss_cls_3: 0.7788, loss_box_3: 1.3629, loss_cns_3: 0.6603, loss_yns_3: 0.1344, loss_cls_4: 0.7916, loss_box_4: 1.3732, loss_cns_4: 0.6623, loss_yns_4: 0.1363, loss_cls_5: 0.7934, loss_box_5: 1.3641, loss_cns_5: 0.6627, loss_yns_5: 0.1389, loss_cls_dn_0: 0.1091, loss_box_dn_0: 0.7120, loss_cls_dn_1: 0.0914, loss_box_dn_1: 0.6278, loss_cls_dn_2: 0.0907, loss_box_dn_2: 0.6264, loss_cls_dn_3: 0.0930, loss_box_dn_3: 0.6133, loss_cls_dn_4: 0.0930, loss_box_dn_4: 0.6146, loss_cls_dn_5: 0.0941, loss_box_dn_5: 0.6087, loss_dense_depth: 0.7299, loss: 22.9280, grad_norm: 39.7986 -2026-01-14 21:43:28,823 - mmdet - INFO - Iter [430/17500] lr: 2.712e-04, eta: 9:20:15, time: 1.608, data_time: 0.102, memory: 49164, loss_cls_0: 0.7011, loss_box_0: 1.5542, loss_cns_0: 0.6441, loss_yns_0: 0.1372, loss_cls_1: 0.7631, loss_box_1: 1.3966, loss_cns_1: 0.6633, loss_yns_1: 0.1352, loss_cls_2: 0.7700, loss_box_2: 1.3681, loss_cns_2: 0.6607, loss_yns_2: 0.1345, loss_cls_3: 0.7754, loss_box_3: 1.3459, loss_cns_3: 0.6588, loss_yns_3: 0.1320, loss_cls_4: 0.7848, loss_box_4: 1.3682, loss_cns_4: 0.6626, loss_yns_4: 0.1349, loss_cls_5: 0.7913, loss_box_5: 1.3671, loss_cns_5: 0.6642, loss_yns_5: 0.1362, loss_cls_dn_0: 0.1133, loss_box_dn_0: 0.7085, loss_cls_dn_1: 0.0955, loss_box_dn_1: 0.6094, loss_cls_dn_2: 0.0963, loss_box_dn_2: 0.5915, loss_cls_dn_3: 0.0968, loss_box_dn_3: 0.5834, loss_cls_dn_4: 0.0988, loss_box_dn_4: 0.5868, loss_cls_dn_5: 0.0989, loss_box_dn_5: 0.5862, loss_dense_depth: 0.6919, loss: 22.7066, grad_norm: 28.9265 -2026-01-14 21:43:30,395 - mmdet - INFO - Iter [431/17500] lr: 2.716e-04, eta: 9:19:57, time: 1.574, data_time: 0.076, memory: 49164, loss_cls_0: 0.7087, loss_box_0: 1.5585, loss_cns_0: 0.6409, loss_yns_0: 0.1386, loss_cls_1: 0.7766, loss_box_1: 1.3975, loss_cns_1: 0.6638, loss_yns_1: 0.1349, loss_cls_2: 0.7821, loss_box_2: 1.3596, loss_cns_2: 0.6604, loss_yns_2: 0.1349, loss_cls_3: 0.8006, loss_box_3: 1.3283, loss_cns_3: 0.6555, loss_yns_3: 0.1316, loss_cls_4: 0.8001, loss_box_4: 1.3536, loss_cns_4: 0.6639, loss_yns_4: 0.1343, loss_cls_5: 0.8071, loss_box_5: 1.3717, loss_cns_5: 0.6670, loss_yns_5: 0.1346, loss_cls_dn_0: 0.1181, loss_box_dn_0: 0.7089, loss_cls_dn_1: 0.0969, loss_box_dn_1: 0.6282, loss_cls_dn_2: 0.0990, loss_box_dn_2: 0.6083, loss_cls_dn_3: 0.0994, loss_box_dn_3: 0.6067, loss_cls_dn_4: 0.1019, loss_box_dn_4: 0.6067, loss_cls_dn_5: 0.1020, loss_box_dn_5: 0.6179, loss_dense_depth: 0.7216, loss: 22.9205, grad_norm: 31.2431 -2026-01-14 21:43:31,970 - mmdet - INFO - Iter [432/17500] lr: 2.720e-04, eta: 9:19:40, time: 1.574, data_time: 0.075, memory: 49164, loss_cls_0: 0.7187, loss_box_0: 1.5690, loss_cns_0: 0.6396, loss_yns_0: 0.1401, loss_cls_1: 0.7783, loss_box_1: 1.4084, loss_cns_1: 0.6635, loss_yns_1: 0.1367, loss_cls_2: 0.7844, loss_box_2: 1.3932, loss_cns_2: 0.6623, loss_yns_2: 0.1375, loss_cls_3: 0.8072, loss_box_3: 1.3629, loss_cns_3: 0.6595, loss_yns_3: 0.1361, loss_cls_4: 0.8014, loss_box_4: 1.3841, loss_cns_4: 0.6646, loss_yns_4: 0.1364, loss_cls_5: 0.8027, loss_box_5: 1.3924, loss_cns_5: 0.6649, loss_yns_5: 0.1381, loss_cls_dn_0: 0.1191, loss_box_dn_0: 0.7162, loss_cls_dn_1: 0.1004, loss_box_dn_1: 0.6404, loss_cls_dn_2: 0.1024, loss_box_dn_2: 0.6335, loss_cls_dn_3: 0.1038, loss_box_dn_3: 0.6323, loss_cls_dn_4: 0.1052, loss_box_dn_4: 0.6376, loss_cls_dn_5: 0.1034, loss_box_dn_5: 0.6461, loss_dense_depth: 0.7137, loss: 23.2359, grad_norm: 32.8171 -2026-01-14 21:43:33,595 - mmdet - INFO - Iter [433/17500] lr: 2.724e-04, eta: 9:19:24, time: 1.625, data_time: 0.089, memory: 49164, loss_cls_0: 0.7001, loss_box_0: 1.5875, loss_cns_0: 0.6408, loss_yns_0: 0.1398, loss_cls_1: 0.7711, loss_box_1: 1.4271, loss_cns_1: 0.6642, loss_yns_1: 0.1356, loss_cls_2: 0.7794, loss_box_2: 1.4101, loss_cns_2: 0.6637, loss_yns_2: 0.1338, loss_cls_3: 0.7843, loss_box_3: 1.3863, loss_cns_3: 0.6638, loss_yns_3: 0.1339, loss_cls_4: 0.7846, loss_box_4: 1.3882, loss_cns_4: 0.6651, loss_yns_4: 0.1341, loss_cls_5: 0.7846, loss_box_5: 1.3833, loss_cns_5: 0.6638, loss_yns_5: 0.1344, loss_cls_dn_0: 0.1130, loss_box_dn_0: 0.7114, loss_cls_dn_1: 0.1038, loss_box_dn_1: 0.6515, loss_cls_dn_2: 0.1022, loss_box_dn_2: 0.6439, loss_cls_dn_3: 0.1011, loss_box_dn_3: 0.6405, loss_cls_dn_4: 0.1022, loss_box_dn_4: 0.6429, loss_cls_dn_5: 0.1022, loss_box_dn_5: 0.6461, loss_dense_depth: 0.6912, loss: 23.2112, grad_norm: 33.3689 -2026-01-14 21:43:35,198 - mmdet - INFO - Iter [434/17500] lr: 2.728e-04, eta: 9:19:07, time: 1.575, data_time: 0.086, memory: 49164, loss_cls_0: 0.7210, loss_box_0: 1.5901, loss_cns_0: 0.6397, loss_yns_0: 0.1405, loss_cls_1: 0.7714, loss_box_1: 1.4486, loss_cns_1: 0.6669, loss_yns_1: 0.1377, loss_cls_2: 0.7784, loss_box_2: 1.4221, loss_cns_2: 0.6613, loss_yns_2: 0.1372, loss_cls_3: 0.7852, loss_box_3: 1.4009, loss_cns_3: 0.6641, loss_yns_3: 0.1380, loss_cls_4: 0.7800, loss_box_4: 1.3961, loss_cns_4: 0.6621, loss_yns_4: 0.1377, loss_cls_5: 0.7900, loss_box_5: 1.3918, loss_cns_5: 0.6622, loss_yns_5: 0.1371, loss_cls_dn_0: 0.1128, loss_box_dn_0: 0.7107, loss_cls_dn_1: 0.0963, loss_box_dn_1: 0.6542, loss_cls_dn_2: 0.0957, loss_box_dn_2: 0.6370, loss_cls_dn_3: 0.0959, loss_box_dn_3: 0.6279, loss_cls_dn_4: 0.1000, loss_box_dn_4: 0.6248, loss_cls_dn_5: 0.0990, loss_box_dn_5: 0.6233, loss_dense_depth: 0.7010, loss: 23.2389, grad_norm: 25.9174 -2026-01-14 21:43:36,818 - mmdet - INFO - Iter [435/17500] lr: 2.732e-04, eta: 9:18:51, time: 1.617, data_time: 0.098, memory: 49164, loss_cls_0: 0.7300, loss_box_0: 1.6207, loss_cns_0: 0.6362, loss_yns_0: 0.1405, loss_cls_1: 0.7844, loss_box_1: 1.4486, loss_cns_1: 0.6636, loss_yns_1: 0.1371, loss_cls_2: 0.7892, loss_box_2: 1.4286, loss_cns_2: 0.6611, loss_yns_2: 0.1369, loss_cls_3: 0.8027, loss_box_3: 1.4166, loss_cns_3: 0.6609, loss_yns_3: 0.1373, loss_cls_4: 0.7986, loss_box_4: 1.4103, loss_cns_4: 0.6631, loss_yns_4: 0.1366, loss_cls_5: 0.8069, loss_box_5: 1.4037, loss_cns_5: 0.6605, loss_yns_5: 0.1363, loss_cls_dn_0: 0.1147, loss_box_dn_0: 0.7088, loss_cls_dn_1: 0.0946, loss_box_dn_1: 0.6344, loss_cls_dn_2: 0.0934, loss_box_dn_2: 0.6175, loss_cls_dn_3: 0.0969, loss_box_dn_3: 0.6123, loss_cls_dn_4: 0.0979, loss_box_dn_4: 0.6110, loss_cls_dn_5: 0.0956, loss_box_dn_5: 0.6150, loss_dense_depth: 0.7008, loss: 23.3032, grad_norm: 35.0061 -2026-01-14 21:43:38,385 - mmdet - INFO - Iter [436/17500] lr: 2.736e-04, eta: 9:18:35, time: 1.597, data_time: 0.104, memory: 49164, loss_cls_0: 0.7157, loss_box_0: 1.5954, loss_cns_0: 0.6361, loss_yns_0: 0.1408, loss_cls_1: 0.7989, loss_box_1: 1.4569, loss_cns_1: 0.6649, loss_yns_1: 0.1373, loss_cls_2: 0.8001, loss_box_2: 1.4222, loss_cns_2: 0.6662, loss_yns_2: 0.1380, loss_cls_3: 0.7994, loss_box_3: 1.4283, loss_cns_3: 0.6627, loss_yns_3: 0.1373, loss_cls_4: 0.8000, loss_box_4: 1.4221, loss_cns_4: 0.6693, loss_yns_4: 0.1362, loss_cls_5: 0.8042, loss_box_5: 1.4296, loss_cns_5: 0.6629, loss_yns_5: 0.1364, loss_cls_dn_0: 0.1113, loss_box_dn_0: 0.7142, loss_cls_dn_1: 0.0914, loss_box_dn_1: 0.6288, loss_cls_dn_2: 0.0888, loss_box_dn_2: 0.6137, loss_cls_dn_3: 0.0913, loss_box_dn_3: 0.6184, loss_cls_dn_4: 0.0944, loss_box_dn_4: 0.6163, loss_cls_dn_5: 0.0914, loss_box_dn_5: 0.6261, loss_dense_depth: 0.6921, loss: 23.3391, grad_norm: 30.8960 -2026-01-14 21:43:39,960 - mmdet - INFO - Iter [437/17500] lr: 2.740e-04, eta: 9:18:18, time: 1.576, data_time: 0.084, memory: 49164, loss_cls_0: 0.6938, loss_box_0: 1.5809, loss_cns_0: 0.6393, loss_yns_0: 0.1413, loss_cls_1: 0.7894, loss_box_1: 1.4492, loss_cns_1: 0.6652, loss_yns_1: 0.1368, loss_cls_2: 0.7898, loss_box_2: 1.4243, loss_cns_2: 0.6662, loss_yns_2: 0.1368, loss_cls_3: 0.7864, loss_box_3: 1.4212, loss_cns_3: 0.6619, loss_yns_3: 0.1369, loss_cls_4: 0.7862, loss_box_4: 1.4192, loss_cns_4: 0.6643, loss_yns_4: 0.1366, loss_cls_5: 0.7941, loss_box_5: 1.4212, loss_cns_5: 0.6635, loss_yns_5: 0.1373, loss_cls_dn_0: 0.1081, loss_box_dn_0: 0.7090, loss_cls_dn_1: 0.0937, loss_box_dn_1: 0.6294, loss_cls_dn_2: 0.0918, loss_box_dn_2: 0.6143, loss_cls_dn_3: 0.0917, loss_box_dn_3: 0.6162, loss_cls_dn_4: 0.0943, loss_box_dn_4: 0.6172, loss_cls_dn_5: 0.0945, loss_box_dn_5: 0.6237, loss_dense_depth: 0.6762, loss: 23.2018, grad_norm: 27.6645 -2026-01-14 21:43:41,552 - mmdet - INFO - Iter [438/17500] lr: 2.744e-04, eta: 9:18:01, time: 1.593, data_time: 0.084, memory: 49164, loss_cls_0: 0.7317, loss_box_0: 1.6203, loss_cns_0: 0.6329, loss_yns_0: 0.1398, loss_cls_1: 0.7943, loss_box_1: 1.4206, loss_cns_1: 0.6632, loss_yns_1: 0.1386, loss_cls_2: 0.7999, loss_box_2: 1.4048, loss_cns_2: 0.6607, loss_yns_2: 0.1380, loss_cls_3: 0.8003, loss_box_3: 1.4260, loss_cns_3: 0.6615, loss_yns_3: 0.1375, loss_cls_4: 0.8099, loss_box_4: 1.4153, loss_cns_4: 0.6624, loss_yns_4: 0.1380, loss_cls_5: 0.8184, loss_box_5: 1.4072, loss_cns_5: 0.6616, loss_yns_5: 0.1383, loss_cls_dn_0: 0.1158, loss_box_dn_0: 0.7203, loss_cls_dn_1: 0.0972, loss_box_dn_1: 0.6229, loss_cls_dn_2: 0.0961, loss_box_dn_2: 0.6091, loss_cls_dn_3: 0.0948, loss_box_dn_3: 0.6158, loss_cls_dn_4: 0.0950, loss_box_dn_4: 0.6163, loss_cls_dn_5: 0.0970, loss_box_dn_5: 0.6185, loss_dense_depth: 0.6918, loss: 23.3121, grad_norm: 32.5377 -2026-01-14 21:43:43,205 - mmdet - INFO - Iter [439/17500] lr: 2.748e-04, eta: 9:17:47, time: 1.653, data_time: 0.183, memory: 49164, loss_cls_0: 0.7178, loss_box_0: 1.5843, loss_cns_0: 0.6382, loss_yns_0: 0.1384, loss_cls_1: 0.7727, loss_box_1: 1.4065, loss_cns_1: 0.6591, loss_yns_1: 0.1364, loss_cls_2: 0.7744, loss_box_2: 1.4080, loss_cns_2: 0.6625, loss_yns_2: 0.1380, loss_cls_3: 0.7886, loss_box_3: 1.3910, loss_cns_3: 0.6589, loss_yns_3: 0.1358, loss_cls_4: 0.7869, loss_box_4: 1.3847, loss_cns_4: 0.6640, loss_yns_4: 0.1357, loss_cls_5: 0.7853, loss_box_5: 1.3842, loss_cns_5: 0.6618, loss_yns_5: 0.1359, loss_cls_dn_0: 0.1084, loss_box_dn_0: 0.7068, loss_cls_dn_1: 0.0947, loss_box_dn_1: 0.6186, loss_cls_dn_2: 0.0923, loss_box_dn_2: 0.6055, loss_cls_dn_3: 0.0911, loss_box_dn_3: 0.6025, loss_cls_dn_4: 0.0922, loss_box_dn_4: 0.6048, loss_cls_dn_5: 0.0919, loss_box_dn_5: 0.6049, loss_dense_depth: 0.7037, loss: 22.9667, grad_norm: 33.1294 -2026-01-14 21:43:44,822 - mmdet - INFO - Iter [440/17500] lr: 2.752e-04, eta: 9:17:32, time: 1.616, data_time: 0.074, memory: 49164, loss_cls_0: 0.7352, loss_box_0: 1.5923, loss_cns_0: 0.6396, loss_yns_0: 0.1399, loss_cls_1: 0.7775, loss_box_1: 1.4123, loss_cns_1: 0.6629, loss_yns_1: 0.1354, loss_cls_2: 0.7789, loss_box_2: 1.4163, loss_cns_2: 0.6653, loss_yns_2: 0.1358, loss_cls_3: 0.7831, loss_box_3: 1.3915, loss_cns_3: 0.6620, loss_yns_3: 0.1367, loss_cls_4: 0.7846, loss_box_4: 1.3944, loss_cns_4: 0.6672, loss_yns_4: 0.1372, loss_cls_5: 0.7946, loss_box_5: 1.4002, loss_cns_5: 0.6641, loss_yns_5: 0.1378, loss_cls_dn_0: 0.1108, loss_box_dn_0: 0.7089, loss_cls_dn_1: 0.0928, loss_box_dn_1: 0.6119, loss_cls_dn_2: 0.0901, loss_box_dn_2: 0.6010, loss_cls_dn_3: 0.0931, loss_box_dn_3: 0.5935, loss_cls_dn_4: 0.0942, loss_box_dn_4: 0.5984, loss_cls_dn_5: 0.0916, loss_box_dn_5: 0.5977, loss_dense_depth: 0.7044, loss: 23.0329, grad_norm: 26.2216 -2026-01-14 21:43:46,498 - mmdet - INFO - Iter [441/17500] lr: 2.756e-04, eta: 9:17:19, time: 1.675, data_time: 0.077, memory: 49164, loss_cls_0: 0.7178, loss_box_0: 1.5800, loss_cns_0: 0.6357, loss_yns_0: 0.1395, loss_cls_1: 0.7736, loss_box_1: 1.4435, loss_cns_1: 0.6629, loss_yns_1: 0.1371, loss_cls_2: 0.7864, loss_box_2: 1.4202, loss_cns_2: 0.6623, loss_yns_2: 0.1367, loss_cls_3: 0.7904, loss_box_3: 1.4031, loss_cns_3: 0.6611, loss_yns_3: 0.1360, loss_cls_4: 0.7888, loss_box_4: 1.4030, loss_cns_4: 0.6636, loss_yns_4: 0.1368, loss_cls_5: 0.7936, loss_box_5: 1.3995, loss_cns_5: 0.6625, loss_yns_5: 0.1368, loss_cls_dn_0: 0.1071, loss_box_dn_0: 0.7011, loss_cls_dn_1: 0.0917, loss_box_dn_1: 0.6150, loss_cls_dn_2: 0.0903, loss_box_dn_2: 0.5989, loss_cls_dn_3: 0.0940, loss_box_dn_3: 0.5979, loss_cls_dn_4: 0.0952, loss_box_dn_4: 0.5998, loss_cls_dn_5: 0.0950, loss_box_dn_5: 0.6006, loss_dense_depth: 0.7325, loss: 23.0898, grad_norm: 30.6812 -2026-01-14 21:43:48,148 - mmdet - INFO - Iter [442/17500] lr: 2.760e-04, eta: 9:17:05, time: 1.652, data_time: 0.089, memory: 49164, loss_cls_0: 0.7267, loss_box_0: 1.5568, loss_cns_0: 0.6393, loss_yns_0: 0.1386, loss_cls_1: 0.7811, loss_box_1: 1.4424, loss_cns_1: 0.6647, loss_yns_1: 0.1371, loss_cls_2: 0.7883, loss_box_2: 1.4148, loss_cns_2: 0.6613, loss_yns_2: 0.1363, loss_cls_3: 0.7880, loss_box_3: 1.4259, loss_cns_3: 0.6603, loss_yns_3: 0.1359, loss_cls_4: 0.7941, loss_box_4: 1.4419, loss_cns_4: 0.6623, loss_yns_4: 0.1371, loss_cls_5: 0.7978, loss_box_5: 1.4299, loss_cns_5: 0.6618, loss_yns_5: 0.1373, loss_cls_dn_0: 0.1072, loss_box_dn_0: 0.7083, loss_cls_dn_1: 0.0894, loss_box_dn_1: 0.6279, loss_cls_dn_2: 0.0890, loss_box_dn_2: 0.6128, loss_cls_dn_3: 0.0910, loss_box_dn_3: 0.6202, loss_cls_dn_4: 0.0909, loss_box_dn_4: 0.6285, loss_cls_dn_5: 0.0943, loss_box_dn_5: 0.6285, loss_dense_depth: 0.7274, loss: 23.2753, grad_norm: 42.1038 -2026-01-14 21:43:49,795 - mmdet - INFO - Iter [443/17500] lr: 2.764e-04, eta: 9:16:49, time: 1.592, data_time: 0.074, memory: 49164, loss_cls_0: 0.7174, loss_box_0: 1.5369, loss_cns_0: 0.6416, loss_yns_0: 0.1402, loss_cls_1: 0.7780, loss_box_1: 1.4386, loss_cns_1: 0.6679, loss_yns_1: 0.1373, loss_cls_2: 0.7826, loss_box_2: 1.4153, loss_cns_2: 0.6660, loss_yns_2: 0.1380, loss_cls_3: 0.7855, loss_box_3: 1.4069, loss_cns_3: 0.6638, loss_yns_3: 0.1365, loss_cls_4: 0.7977, loss_box_4: 1.4050, loss_cns_4: 0.6655, loss_yns_4: 0.1374, loss_cls_5: 0.7910, loss_box_5: 1.4101, loss_cns_5: 0.6653, loss_yns_5: 0.1375, loss_cls_dn_0: 0.1091, loss_box_dn_0: 0.7082, loss_cls_dn_1: 0.0903, loss_box_dn_1: 0.6220, loss_cls_dn_2: 0.0910, loss_box_dn_2: 0.6056, loss_cls_dn_3: 0.0903, loss_box_dn_3: 0.6026, loss_cls_dn_4: 0.0902, loss_box_dn_4: 0.6041, loss_cls_dn_5: 0.0926, loss_box_dn_5: 0.6076, loss_dense_depth: 0.7063, loss: 23.0820, grad_norm: 26.8994 -2026-01-14 21:43:51,355 - mmdet - INFO - Iter [444/17500] lr: 2.768e-04, eta: 9:16:34, time: 1.615, data_time: 0.120, memory: 49164, loss_cls_0: 0.7196, loss_box_0: 1.5722, loss_cns_0: 0.6427, loss_yns_0: 0.1401, loss_cls_1: 0.7764, loss_box_1: 1.4354, loss_cns_1: 0.6651, loss_yns_1: 0.1365, loss_cls_2: 0.7783, loss_box_2: 1.4354, loss_cns_2: 0.6642, loss_yns_2: 0.1371, loss_cls_3: 0.7832, loss_box_3: 1.3974, loss_cns_3: 0.6608, loss_yns_3: 0.1367, loss_cls_4: 0.7899, loss_box_4: 1.4104, loss_cns_4: 0.6639, loss_yns_4: 0.1364, loss_cls_5: 0.7917, loss_box_5: 1.4197, loss_cns_5: 0.6641, loss_yns_5: 0.1364, loss_cls_dn_0: 0.1052, loss_box_dn_0: 0.7143, loss_cls_dn_1: 0.0883, loss_box_dn_1: 0.6223, loss_cls_dn_2: 0.0886, loss_box_dn_2: 0.6143, loss_cls_dn_3: 0.0886, loss_box_dn_3: 0.6007, loss_cls_dn_4: 0.0895, loss_box_dn_4: 0.6041, loss_cls_dn_5: 0.0895, loss_box_dn_5: 0.6075, loss_dense_depth: 0.8071, loss: 23.2136, grad_norm: 38.0753 -2026-01-14 21:43:53,027 - mmdet - INFO - Iter [445/17500] lr: 2.772e-04, eta: 9:16:19, time: 1.638, data_time: 0.071, memory: 49164, loss_cls_0: 0.7485, loss_box_0: 1.5944, loss_cns_0: 0.6421, loss_yns_0: 0.1422, loss_cls_1: 0.7867, loss_box_1: 1.4349, loss_cns_1: 0.6602, loss_yns_1: 0.1393, loss_cls_2: 0.7971, loss_box_2: 1.4329, loss_cns_2: 0.6620, loss_yns_2: 0.1397, loss_cls_3: 0.8029, loss_box_3: 1.4034, loss_cns_3: 0.6605, loss_yns_3: 0.1384, loss_cls_4: 0.8078, loss_box_4: 1.4029, loss_cns_4: 0.6613, loss_yns_4: 0.1390, loss_cls_5: 0.8138, loss_box_5: 1.3918, loss_cns_5: 0.6581, loss_yns_5: 0.1379, loss_cls_dn_0: 0.1083, loss_box_dn_0: 0.7144, loss_cls_dn_1: 0.0895, loss_box_dn_1: 0.6156, loss_cls_dn_2: 0.0899, loss_box_dn_2: 0.6095, loss_cls_dn_3: 0.0936, loss_box_dn_3: 0.6022, loss_cls_dn_4: 0.0912, loss_box_dn_4: 0.6014, loss_cls_dn_5: 0.0935, loss_box_dn_5: 0.5988, loss_dense_depth: 0.7033, loss: 23.2087, grad_norm: 33.2911 -2026-01-14 21:43:54,739 - mmdet - INFO - Iter [446/17500] lr: 2.776e-04, eta: 9:16:08, time: 1.708, data_time: 0.107, memory: 49164, loss_cls_0: 0.7378, loss_box_0: 1.5537, loss_cns_0: 0.6412, loss_yns_0: 0.1416, loss_cls_1: 0.7843, loss_box_1: 1.4009, loss_cns_1: 0.6611, loss_yns_1: 0.1360, loss_cls_2: 0.7985, loss_box_2: 1.3807, loss_cns_2: 0.6628, loss_yns_2: 0.1364, loss_cls_3: 0.7986, loss_box_3: 1.3646, loss_cns_3: 0.6619, loss_yns_3: 0.1356, loss_cls_4: 0.7974, loss_box_4: 1.3638, loss_cns_4: 0.6619, loss_yns_4: 0.1363, loss_cls_5: 0.7954, loss_box_5: 1.3594, loss_cns_5: 0.6609, loss_yns_5: 0.1363, loss_cls_dn_0: 0.1064, loss_box_dn_0: 0.7107, loss_cls_dn_1: 0.0888, loss_box_dn_1: 0.6149, loss_cls_dn_2: 0.0894, loss_box_dn_2: 0.5986, loss_cls_dn_3: 0.0923, loss_box_dn_3: 0.5932, loss_cls_dn_4: 0.0911, loss_box_dn_4: 0.5972, loss_cls_dn_5: 0.0936, loss_box_dn_5: 0.5964, loss_dense_depth: 0.7161, loss: 22.8956, grad_norm: 27.9482 -2026-01-14 21:43:56,288 - mmdet - INFO - Iter [447/17500] lr: 2.780e-04, eta: 9:15:52, time: 1.587, data_time: 0.109, memory: 49164, loss_cls_0: 0.7273, loss_box_0: 1.5442, loss_cns_0: 0.6433, loss_yns_0: 0.1420, loss_cls_1: 0.7880, loss_box_1: 1.4080, loss_cns_1: 0.6654, loss_yns_1: 0.1395, loss_cls_2: 0.7940, loss_box_2: 1.3817, loss_cns_2: 0.6658, loss_yns_2: 0.1383, loss_cls_3: 0.7995, loss_box_3: 1.3883, loss_cns_3: 0.6633, loss_yns_3: 0.1389, loss_cls_4: 0.8039, loss_box_4: 1.3842, loss_cns_4: 0.6625, loss_yns_4: 0.1397, loss_cls_5: 0.8036, loss_box_5: 1.3889, loss_cns_5: 0.6609, loss_yns_5: 0.1387, loss_cls_dn_0: 0.1059, loss_box_dn_0: 0.7074, loss_cls_dn_1: 0.0898, loss_box_dn_1: 0.6247, loss_cls_dn_2: 0.0888, loss_box_dn_2: 0.6061, loss_cls_dn_3: 0.0897, loss_box_dn_3: 0.6080, loss_cls_dn_4: 0.0888, loss_box_dn_4: 0.6111, loss_cls_dn_5: 0.0886, loss_box_dn_5: 0.6175, loss_dense_depth: 0.7470, loss: 23.0832, grad_norm: 30.6577 -2026-01-14 21:43:57,839 - mmdet - INFO - Iter [448/17500] lr: 2.784e-04, eta: 9:15:35, time: 1.552, data_time: 0.072, memory: 49164, loss_cls_0: 0.7138, loss_box_0: 1.5563, loss_cns_0: 0.6449, loss_yns_0: 0.1404, loss_cls_1: 0.7848, loss_box_1: 1.4128, loss_cns_1: 0.6686, loss_yns_1: 0.1399, loss_cls_2: 0.7894, loss_box_2: 1.4000, loss_cns_2: 0.6724, loss_yns_2: 0.1390, loss_cls_3: 0.7993, loss_box_3: 1.3941, loss_cns_3: 0.6673, loss_yns_3: 0.1373, loss_cls_4: 0.8042, loss_box_4: 1.4022, loss_cns_4: 0.6654, loss_yns_4: 0.1377, loss_cls_5: 0.8063, loss_box_5: 1.4051, loss_cns_5: 0.6659, loss_yns_5: 0.1381, loss_cls_dn_0: 0.1095, loss_box_dn_0: 0.7015, loss_cls_dn_1: 0.0909, loss_box_dn_1: 0.6184, loss_cls_dn_2: 0.0886, loss_box_dn_2: 0.6102, loss_cls_dn_3: 0.0885, loss_box_dn_3: 0.6093, loss_cls_dn_4: 0.0871, loss_box_dn_4: 0.6150, loss_cls_dn_5: 0.0877, loss_box_dn_5: 0.6194, loss_dense_depth: 0.6775, loss: 23.0888, grad_norm: 33.1468 -2026-01-14 21:43:59,453 - mmdet - INFO - Iter [449/17500] lr: 2.787e-04, eta: 9:15:20, time: 1.613, data_time: 0.071, memory: 49164, loss_cls_0: 0.7100, loss_box_0: 1.5317, loss_cns_0: 0.6394, loss_yns_0: 0.1373, loss_cls_1: 0.7850, loss_box_1: 1.4247, loss_cns_1: 0.6654, loss_yns_1: 0.1376, loss_cls_2: 0.7943, loss_box_2: 1.4023, loss_cns_2: 0.6690, loss_yns_2: 0.1399, loss_cls_3: 0.7989, loss_box_3: 1.3900, loss_cns_3: 0.6674, loss_yns_3: 0.1380, loss_cls_4: 0.8101, loss_box_4: 1.3894, loss_cns_4: 0.6672, loss_yns_4: 0.1400, loss_cls_5: 0.8049, loss_box_5: 1.3822, loss_cns_5: 0.6669, loss_yns_5: 0.1374, loss_cls_dn_0: 0.1095, loss_box_dn_0: 0.7006, loss_cls_dn_1: 0.0916, loss_box_dn_1: 0.6183, loss_cls_dn_2: 0.0900, loss_box_dn_2: 0.6083, loss_cls_dn_3: 0.0900, loss_box_dn_3: 0.6022, loss_cls_dn_4: 0.0901, loss_box_dn_4: 0.6016, loss_cls_dn_5: 0.0899, loss_box_dn_5: 0.5995, loss_dense_depth: 0.7007, loss: 23.0211, grad_norm: 31.9611 -2026-01-14 21:44:01,009 - mmdet - INFO - Iter [450/17500] lr: 2.791e-04, eta: 9:15:03, time: 1.556, data_time: 0.072, memory: 49164, loss_cls_0: 0.7378, loss_box_0: 1.5661, loss_cns_0: 0.6294, loss_yns_0: 0.1409, loss_cls_1: 0.7852, loss_box_1: 1.4403, loss_cns_1: 0.6652, loss_yns_1: 0.1382, loss_cls_2: 0.7993, loss_box_2: 1.4101, loss_cns_2: 0.6646, loss_yns_2: 0.1403, loss_cls_3: 0.8129, loss_box_3: 1.3955, loss_cns_3: 0.6621, loss_yns_3: 0.1380, loss_cls_4: 0.8107, loss_box_4: 1.3858, loss_cns_4: 0.6636, loss_yns_4: 0.1383, loss_cls_5: 0.8179, loss_box_5: 1.3933, loss_cns_5: 0.6645, loss_yns_5: 0.1371, loss_cls_dn_0: 0.1081, loss_box_dn_0: 0.7042, loss_cls_dn_1: 0.0899, loss_box_dn_1: 0.6167, loss_cls_dn_2: 0.0914, loss_box_dn_2: 0.5961, loss_cls_dn_3: 0.0919, loss_box_dn_3: 0.5909, loss_cls_dn_4: 0.0906, loss_box_dn_4: 0.5839, loss_cls_dn_5: 0.0929, loss_box_dn_5: 0.5906, loss_dense_depth: 0.7114, loss: 23.0955, grad_norm: 25.8613 -2026-01-14 21:44:02,607 - mmdet - INFO - Iter [451/17500] lr: 2.795e-04, eta: 9:14:47, time: 1.598, data_time: 0.074, memory: 49164, loss_cls_0: 0.7343, loss_box_0: 1.5747, loss_cns_0: 0.6397, loss_yns_0: 0.1392, loss_cls_1: 0.7713, loss_box_1: 1.4495, loss_cns_1: 0.6630, loss_yns_1: 0.1362, loss_cls_2: 0.7778, loss_box_2: 1.4189, loss_cns_2: 0.6650, loss_yns_2: 0.1362, loss_cls_3: 0.7809, loss_box_3: 1.4067, loss_cns_3: 0.6635, loss_yns_3: 0.1363, loss_cls_4: 0.7911, loss_box_4: 1.4116, loss_cns_4: 0.6630, loss_yns_4: 0.1365, loss_cls_5: 0.7982, loss_box_5: 1.4259, loss_cns_5: 0.6650, loss_yns_5: 0.1367, loss_cls_dn_0: 0.1078, loss_box_dn_0: 0.7090, loss_cls_dn_1: 0.0897, loss_box_dn_1: 0.6147, loss_cls_dn_2: 0.0890, loss_box_dn_2: 0.6005, loss_cls_dn_3: 0.0876, loss_box_dn_3: 0.6004, loss_cls_dn_4: 0.0878, loss_box_dn_4: 0.6069, loss_cls_dn_5: 0.0895, loss_box_dn_5: 0.6173, loss_dense_depth: 0.7020, loss: 23.1234, grad_norm: 29.7472 -2026-01-14 21:44:04,174 - mmdet - INFO - Iter [452/17500] lr: 2.799e-04, eta: 9:14:31, time: 1.567, data_time: 0.074, memory: 49164, loss_cls_0: 0.7149, loss_box_0: 1.5754, loss_cns_0: 0.6365, loss_yns_0: 0.1380, loss_cls_1: 0.7636, loss_box_1: 1.4556, loss_cns_1: 0.6598, loss_yns_1: 0.1353, loss_cls_2: 0.7712, loss_box_2: 1.4218, loss_cns_2: 0.6603, loss_yns_2: 0.1338, loss_cls_3: 0.7844, loss_box_3: 1.4142, loss_cns_3: 0.6598, loss_yns_3: 0.1344, loss_cls_4: 0.7884, loss_box_4: 1.4020, loss_cns_4: 0.6539, loss_yns_4: 0.1332, loss_cls_5: 0.7863, loss_box_5: 1.4227, loss_cns_5: 0.6619, loss_yns_5: 0.1351, loss_cls_dn_0: 0.1037, loss_box_dn_0: 0.7024, loss_cls_dn_1: 0.0887, loss_box_dn_1: 0.6192, loss_cls_dn_2: 0.0871, loss_box_dn_2: 0.6074, loss_cls_dn_3: 0.0882, loss_box_dn_3: 0.6101, loss_cls_dn_4: 0.0892, loss_box_dn_4: 0.6197, loss_cls_dn_5: 0.0902, loss_box_dn_5: 0.6238, loss_dense_depth: 0.6869, loss: 23.0588, grad_norm: 34.6764 -2026-01-14 21:44:05,784 - mmdet - INFO - Iter [453/17500] lr: 2.803e-04, eta: 9:14:15, time: 1.580, data_time: 0.075, memory: 49164, loss_cls_0: 0.6999, loss_box_0: 1.5428, loss_cns_0: 0.6376, loss_yns_0: 0.1385, loss_cls_1: 0.7673, loss_box_1: 1.4266, loss_cns_1: 0.6532, loss_yns_1: 0.1360, loss_cls_2: 0.7739, loss_box_2: 1.3805, loss_cns_2: 0.6557, loss_yns_2: 0.1360, loss_cls_3: 0.7874, loss_box_3: 1.3568, loss_cns_3: 0.6519, loss_yns_3: 0.1355, loss_cls_4: 0.7888, loss_box_4: 1.3149, loss_cns_4: 0.6450, loss_yns_4: 0.1330, loss_cls_5: 0.7782, loss_box_5: 1.3733, loss_cns_5: 0.6551, loss_yns_5: 0.1373, loss_cls_dn_0: 0.1023, loss_box_dn_0: 0.7065, loss_cls_dn_1: 0.0876, loss_box_dn_1: 0.6440, loss_cls_dn_2: 0.0880, loss_box_dn_2: 0.6217, loss_cls_dn_3: 0.0879, loss_box_dn_3: 0.6190, loss_cls_dn_4: 0.0892, loss_box_dn_4: 0.6188, loss_cls_dn_5: 0.0892, loss_box_dn_5: 0.6268, loss_dense_depth: 0.6762, loss: 22.7626, grad_norm: 32.8559 -2026-01-14 21:44:07,428 - mmdet - INFO - Iter [454/17500] lr: 2.807e-04, eta: 9:14:02, time: 1.651, data_time: 0.095, memory: 49164, loss_cls_0: 0.7235, loss_box_0: 1.5846, loss_cns_0: 0.6368, loss_yns_0: 0.1394, loss_cls_1: 0.7650, loss_box_1: 1.4293, loss_cns_1: 0.6553, loss_yns_1: 0.1365, loss_cls_2: 0.7802, loss_box_2: 1.3919, loss_cns_2: 0.6578, loss_yns_2: 0.1358, loss_cls_3: 0.7899, loss_box_3: 1.3866, loss_cns_3: 0.6571, loss_yns_3: 0.1346, loss_cls_4: 0.7934, loss_box_4: 1.3698, loss_cns_4: 0.6522, loss_yns_4: 0.1332, loss_cls_5: 0.7958, loss_box_5: 1.4010, loss_cns_5: 0.6586, loss_yns_5: 0.1360, loss_cls_dn_0: 0.1093, loss_box_dn_0: 0.7104, loss_cls_dn_1: 0.0931, loss_box_dn_1: 0.6280, loss_cls_dn_2: 0.0915, loss_box_dn_2: 0.6061, loss_cls_dn_3: 0.0909, loss_box_dn_3: 0.6049, loss_cls_dn_4: 0.0932, loss_box_dn_4: 0.6010, loss_cls_dn_5: 0.0948, loss_box_dn_5: 0.6062, loss_dense_depth: 0.6929, loss: 22.9667, grad_norm: 26.5057 -2026-01-14 21:44:08,984 - mmdet - INFO - Iter [455/17500] lr: 2.811e-04, eta: 9:13:46, time: 1.577, data_time: 0.087, memory: 49164, loss_cls_0: 0.7172, loss_box_0: 1.5673, loss_cns_0: 0.6449, loss_yns_0: 0.1395, loss_cls_1: 0.7644, loss_box_1: 1.4221, loss_cns_1: 0.6666, loss_yns_1: 0.1377, loss_cls_2: 0.7761, loss_box_2: 1.3905, loss_cns_2: 0.6708, loss_yns_2: 0.1371, loss_cls_3: 0.7924, loss_box_3: 1.3924, loss_cns_3: 0.6690, loss_yns_3: 0.1372, loss_cls_4: 0.7925, loss_box_4: 1.3909, loss_cns_4: 0.6684, loss_yns_4: 0.1368, loss_cls_5: 0.8040, loss_box_5: 1.3983, loss_cns_5: 0.6675, loss_yns_5: 0.1378, loss_cls_dn_0: 0.1058, loss_box_dn_0: 0.6989, loss_cls_dn_1: 0.0920, loss_box_dn_1: 0.6103, loss_cls_dn_2: 0.0895, loss_box_dn_2: 0.5927, loss_cls_dn_3: 0.0921, loss_box_dn_3: 0.5962, loss_cls_dn_4: 0.0951, loss_box_dn_4: 0.5979, loss_cls_dn_5: 0.0964, loss_box_dn_5: 0.6011, loss_dense_depth: 0.6823, loss: 22.9716, grad_norm: 32.6155 -2026-01-14 21:44:10,548 - mmdet - INFO - Iter [456/17500] lr: 2.815e-04, eta: 9:13:29, time: 1.566, data_time: 0.082, memory: 49164, loss_cls_0: 0.7285, loss_box_0: 1.5450, loss_cns_0: 0.6432, loss_yns_0: 0.1417, loss_cls_1: 0.7610, loss_box_1: 1.4345, loss_cns_1: 0.6635, loss_yns_1: 0.1387, loss_cls_2: 0.7749, loss_box_2: 1.4056, loss_cns_2: 0.6630, loss_yns_2: 0.1386, loss_cls_3: 0.7891, loss_box_3: 1.4016, loss_cns_3: 0.6610, loss_yns_3: 0.1411, loss_cls_4: 0.7972, loss_box_4: 1.4039, loss_cns_4: 0.6623, loss_yns_4: 0.1419, loss_cls_5: 0.7928, loss_box_5: 1.4002, loss_cns_5: 0.6633, loss_yns_5: 0.1391, loss_cls_dn_0: 0.1077, loss_box_dn_0: 0.7021, loss_cls_dn_1: 0.0871, loss_box_dn_1: 0.6213, loss_cls_dn_2: 0.0868, loss_box_dn_2: 0.6055, loss_cls_dn_3: 0.0918, loss_box_dn_3: 0.6088, loss_cls_dn_4: 0.0975, loss_box_dn_4: 0.6136, loss_cls_dn_5: 0.0936, loss_box_dn_5: 0.6183, loss_dense_depth: 0.6748, loss: 23.0405, grad_norm: 32.8690 -2026-01-14 21:44:12,168 - mmdet - INFO - Iter [457/17500] lr: 2.819e-04, eta: 9:13:15, time: 1.619, data_time: 0.079, memory: 49164, loss_cls_0: 0.7450, loss_box_0: 1.5449, loss_cns_0: 0.6355, loss_yns_0: 0.1399, loss_cls_1: 0.7820, loss_box_1: 1.4618, loss_cns_1: 0.6672, loss_yns_1: 0.1398, loss_cls_2: 0.7922, loss_box_2: 1.4224, loss_cns_2: 0.6623, loss_yns_2: 0.1395, loss_cls_3: 0.7925, loss_box_3: 1.4211, loss_cns_3: 0.6646, loss_yns_3: 0.1412, loss_cls_4: 0.8072, loss_box_4: 1.4195, loss_cns_4: 0.6652, loss_yns_4: 0.1423, loss_cls_5: 0.8122, loss_box_5: 1.4153, loss_cns_5: 0.6654, loss_yns_5: 0.1403, loss_cls_dn_0: 0.1160, loss_box_dn_0: 0.7137, loss_cls_dn_1: 0.0916, loss_box_dn_1: 0.6242, loss_cls_dn_2: 0.0907, loss_box_dn_2: 0.6036, loss_cls_dn_3: 0.0921, loss_box_dn_3: 0.6033, loss_cls_dn_4: 0.0935, loss_box_dn_4: 0.6026, loss_cls_dn_5: 0.0927, loss_box_dn_5: 0.6061, loss_dense_depth: 0.7098, loss: 23.2591, grad_norm: 28.9618 -2026-01-14 21:44:13,774 - mmdet - INFO - Iter [458/17500] lr: 2.823e-04, eta: 9:12:59, time: 1.576, data_time: 0.080, memory: 49164, loss_cls_0: 0.7148, loss_box_0: 1.5707, loss_cns_0: 0.6337, loss_yns_0: 0.1428, loss_cls_1: 0.7773, loss_box_1: 1.4619, loss_cns_1: 0.6660, loss_yns_1: 0.1452, loss_cls_2: 0.7818, loss_box_2: 1.4188, loss_cns_2: 0.6622, loss_yns_2: 0.1436, loss_cls_3: 0.7957, loss_box_3: 1.4034, loss_cns_3: 0.6632, loss_yns_3: 0.1430, loss_cls_4: 0.8148, loss_box_4: 1.4249, loss_cns_4: 0.6664, loss_yns_4: 0.1430, loss_cls_5: 0.8108, loss_box_5: 1.4179, loss_cns_5: 0.6664, loss_yns_5: 0.1440, loss_cls_dn_0: 0.1099, loss_box_dn_0: 0.7081, loss_cls_dn_1: 0.0967, loss_box_dn_1: 0.6334, loss_cls_dn_2: 0.0929, loss_box_dn_2: 0.6122, loss_cls_dn_3: 0.0934, loss_box_dn_3: 0.6097, loss_cls_dn_4: 0.0952, loss_box_dn_4: 0.6134, loss_cls_dn_5: 0.0969, loss_box_dn_5: 0.6179, loss_dense_depth: 0.6993, loss: 23.2916, grad_norm: 30.5739 -2026-01-14 21:44:15,439 - mmdet - INFO - Iter [459/17500] lr: 2.827e-04, eta: 9:12:48, time: 1.695, data_time: 0.209, memory: 49164, loss_cls_0: 0.7182, loss_box_0: 1.5625, loss_cns_0: 0.6398, loss_yns_0: 0.1407, loss_cls_1: 0.7685, loss_box_1: 1.4304, loss_cns_1: 0.6620, loss_yns_1: 0.1394, loss_cls_2: 0.7751, loss_box_2: 1.3967, loss_cns_2: 0.6624, loss_yns_2: 0.1378, loss_cls_3: 0.7908, loss_box_3: 1.3748, loss_cns_3: 0.6576, loss_yns_3: 0.1383, loss_cls_4: 0.7896, loss_box_4: 1.3911, loss_cns_4: 0.6610, loss_yns_4: 0.1391, loss_cls_5: 0.8033, loss_box_5: 1.3785, loss_cns_5: 0.6622, loss_yns_5: 0.1390, loss_cls_dn_0: 0.1062, loss_box_dn_0: 0.7064, loss_cls_dn_1: 0.0972, loss_box_dn_1: 0.6298, loss_cls_dn_2: 0.0948, loss_box_dn_2: 0.6064, loss_cls_dn_3: 0.0968, loss_box_dn_3: 0.6071, loss_cls_dn_4: 0.0971, loss_box_dn_4: 0.6111, loss_cls_dn_5: 0.0986, loss_box_dn_5: 0.6067, loss_dense_depth: 0.7132, loss: 23.0302, grad_norm: 31.8771 -2026-01-14 21:44:17,071 - mmdet - INFO - Iter [460/17500] lr: 2.831e-04, eta: 9:12:34, time: 1.631, data_time: 0.076, memory: 49164, loss_cls_0: 0.7654, loss_box_0: 1.5660, loss_cns_0: 0.6394, loss_yns_0: 0.1443, loss_cls_1: 0.7834, loss_box_1: 1.3901, loss_cns_1: 0.6605, loss_yns_1: 0.1390, loss_cls_2: 0.7874, loss_box_2: 1.3530, loss_cns_2: 0.6644, loss_yns_2: 0.1371, loss_cls_3: 0.7973, loss_box_3: 1.3346, loss_cns_3: 0.6599, loss_yns_3: 0.1424, loss_cls_4: 0.8077, loss_box_4: 1.3394, loss_cns_4: 0.6628, loss_yns_4: 0.1420, loss_cls_5: 0.8204, loss_box_5: 1.3426, loss_cns_5: 0.6631, loss_yns_5: 0.1391, loss_cls_dn_0: 0.1143, loss_box_dn_0: 0.7097, loss_cls_dn_1: 0.0975, loss_box_dn_1: 0.6337, loss_cls_dn_2: 0.0950, loss_box_dn_2: 0.6052, loss_cls_dn_3: 0.0957, loss_box_dn_3: 0.5982, loss_cls_dn_4: 0.0957, loss_box_dn_4: 0.5975, loss_cls_dn_5: 0.0967, loss_box_dn_5: 0.6011, loss_dense_depth: 0.6957, loss: 22.9173, grad_norm: 23.7987 -2026-01-14 21:44:18,737 - mmdet - INFO - Iter [461/17500] lr: 2.835e-04, eta: 9:12:22, time: 1.667, data_time: 0.077, memory: 49164, loss_cls_0: 0.7380, loss_box_0: 1.5746, loss_cns_0: 0.6347, loss_yns_0: 0.1416, loss_cls_1: 0.7884, loss_box_1: 1.4053, loss_cns_1: 0.6609, loss_yns_1: 0.1386, loss_cls_2: 0.7925, loss_box_2: 1.3847, loss_cns_2: 0.6628, loss_yns_2: 0.1382, loss_cls_3: 0.8069, loss_box_3: 1.3735, loss_cns_3: 0.6630, loss_yns_3: 0.1394, loss_cls_4: 0.8070, loss_box_4: 1.3704, loss_cns_4: 0.6622, loss_yns_4: 0.1380, loss_cls_5: 0.8119, loss_box_5: 1.3790, loss_cns_5: 0.6631, loss_yns_5: 0.1383, loss_cls_dn_0: 0.1088, loss_box_dn_0: 0.7061, loss_cls_dn_1: 0.0937, loss_box_dn_1: 0.6101, loss_cls_dn_2: 0.0926, loss_box_dn_2: 0.5933, loss_cls_dn_3: 0.0922, loss_box_dn_3: 0.5916, loss_cls_dn_4: 0.0944, loss_box_dn_4: 0.5954, loss_cls_dn_5: 0.0940, loss_box_dn_5: 0.6066, loss_dense_depth: 0.6894, loss: 22.9813, grad_norm: 35.4826 -2026-01-14 21:44:20,361 - mmdet - INFO - Iter [462/17500] lr: 2.839e-04, eta: 9:12:08, time: 1.624, data_time: 0.087, memory: 49164, loss_cls_0: 0.7459, loss_box_0: 1.5602, loss_cns_0: 0.6377, loss_yns_0: 0.1422, loss_cls_1: 0.7908, loss_box_1: 1.4053, loss_cns_1: 0.6632, loss_yns_1: 0.1391, loss_cls_2: 0.7944, loss_box_2: 1.3860, loss_cns_2: 0.6613, loss_yns_2: 0.1403, loss_cls_3: 0.8073, loss_box_3: 1.3784, loss_cns_3: 0.6641, loss_yns_3: 0.1395, loss_cls_4: 0.8303, loss_box_4: 1.3743, loss_cns_4: 0.6674, loss_yns_4: 0.1400, loss_cls_5: 0.8179, loss_box_5: 1.3921, loss_cns_5: 0.6673, loss_yns_5: 0.1387, loss_cls_dn_0: 0.1106, loss_box_dn_0: 0.7129, loss_cls_dn_1: 0.0944, loss_box_dn_1: 0.6323, loss_cls_dn_2: 0.0948, loss_box_dn_2: 0.6121, loss_cls_dn_3: 0.0948, loss_box_dn_3: 0.6151, loss_cls_dn_4: 0.0975, loss_box_dn_4: 0.6153, loss_cls_dn_5: 0.0953, loss_box_dn_5: 0.6250, loss_dense_depth: 0.6892, loss: 23.1731, grad_norm: 41.8964 -2026-01-14 21:44:21,932 - mmdet - INFO - Iter [463/17500] lr: 2.843e-04, eta: 9:11:53, time: 1.571, data_time: 0.072, memory: 49164, loss_cls_0: 0.7348, loss_box_0: 1.5265, loss_cns_0: 0.6485, loss_yns_0: 0.1409, loss_cls_1: 0.7743, loss_box_1: 1.4018, loss_cns_1: 0.6686, loss_yns_1: 0.1387, loss_cls_2: 0.7781, loss_box_2: 1.3791, loss_cns_2: 0.6671, loss_yns_2: 0.1415, loss_cls_3: 0.7961, loss_box_3: 1.3538, loss_cns_3: 0.6691, loss_yns_3: 0.1403, loss_cls_4: 0.7992, loss_box_4: 1.3514, loss_cns_4: 0.6707, loss_yns_4: 0.1414, loss_cls_5: 0.8152, loss_box_5: 1.3735, loss_cns_5: 0.6711, loss_yns_5: 0.1400, loss_cls_dn_0: 0.1068, loss_box_dn_0: 0.7111, loss_cls_dn_1: 0.0929, loss_box_dn_1: 0.6248, loss_cls_dn_2: 0.0930, loss_box_dn_2: 0.5986, loss_cls_dn_3: 0.0924, loss_box_dn_3: 0.5907, loss_cls_dn_4: 0.0929, loss_box_dn_4: 0.5858, loss_cls_dn_5: 0.0920, loss_box_dn_5: 0.5964, loss_dense_depth: 0.6638, loss: 22.8629, grad_norm: 32.8530 -2026-01-14 21:44:23,551 - mmdet - INFO - Iter [464/17500] lr: 2.847e-04, eta: 9:11:39, time: 1.619, data_time: 0.073, memory: 49164, loss_cls_0: 0.7215, loss_box_0: 1.5403, loss_cns_0: 0.6475, loss_yns_0: 0.1416, loss_cls_1: 0.7711, loss_box_1: 1.3869, loss_cns_1: 0.6670, loss_yns_1: 0.1376, loss_cls_2: 0.7729, loss_box_2: 1.3617, loss_cns_2: 0.6656, loss_yns_2: 0.1413, loss_cls_3: 0.7803, loss_box_3: 1.3529, loss_cns_3: 0.6662, loss_yns_3: 0.1415, loss_cls_4: 0.7873, loss_box_4: 1.3595, loss_cns_4: 0.6680, loss_yns_4: 0.1455, loss_cls_5: 0.7969, loss_box_5: 1.3506, loss_cns_5: 0.6678, loss_yns_5: 0.1408, loss_cls_dn_0: 0.1062, loss_box_dn_0: 0.7159, loss_cls_dn_1: 0.0909, loss_box_dn_1: 0.6232, loss_cls_dn_2: 0.0898, loss_box_dn_2: 0.5938, loss_cls_dn_3: 0.0881, loss_box_dn_3: 0.5903, loss_cls_dn_4: 0.0884, loss_box_dn_4: 0.5938, loss_cls_dn_5: 0.0884, loss_box_dn_5: 0.5933, loss_dense_depth: 0.6689, loss: 22.7434, grad_norm: 32.1688 -2026-01-14 21:44:25,190 - mmdet - INFO - Iter [465/17500] lr: 2.851e-04, eta: 9:11:26, time: 1.638, data_time: 0.072, memory: 49164, loss_cls_0: 0.7217, loss_box_0: 1.5675, loss_cns_0: 0.6494, loss_yns_0: 0.1407, loss_cls_1: 0.7743, loss_box_1: 1.3804, loss_cns_1: 0.6685, loss_yns_1: 0.1348, loss_cls_2: 0.7818, loss_box_2: 1.3618, loss_cns_2: 0.6648, loss_yns_2: 0.1373, loss_cls_3: 0.7849, loss_box_3: 1.3622, loss_cns_3: 0.6683, loss_yns_3: 0.1389, loss_cls_4: 0.7936, loss_box_4: 1.3715, loss_cns_4: 0.6684, loss_yns_4: 0.1372, loss_cls_5: 0.8080, loss_box_5: 1.3567, loss_cns_5: 0.6687, loss_yns_5: 0.1362, loss_cls_dn_0: 0.1078, loss_box_dn_0: 0.7056, loss_cls_dn_1: 0.0918, loss_box_dn_1: 0.6054, loss_cls_dn_2: 0.0913, loss_box_dn_2: 0.5875, loss_cls_dn_3: 0.0910, loss_box_dn_3: 0.5872, loss_cls_dn_4: 0.0903, loss_box_dn_4: 0.5950, loss_cls_dn_5: 0.0929, loss_box_dn_5: 0.5943, loss_dense_depth: 0.6782, loss: 22.7959, grad_norm: 33.2706 -2026-01-14 21:44:26,838 - mmdet - INFO - Iter [466/17500] lr: 2.855e-04, eta: 9:11:13, time: 1.648, data_time: 0.073, memory: 49164, loss_cls_0: 0.7278, loss_box_0: 1.5481, loss_cns_0: 0.6435, loss_yns_0: 0.1404, loss_cls_1: 0.7849, loss_box_1: 1.3957, loss_cns_1: 0.6679, loss_yns_1: 0.1378, loss_cls_2: 0.7912, loss_box_2: 1.3831, loss_cns_2: 0.6653, loss_yns_2: 0.1383, loss_cls_3: 0.7944, loss_box_3: 1.3665, loss_cns_3: 0.6666, loss_yns_3: 0.1376, loss_cls_4: 0.8083, loss_box_4: 1.3849, loss_cns_4: 0.6679, loss_yns_4: 0.1378, loss_cls_5: 0.8216, loss_box_5: 1.3681, loss_cns_5: 0.6662, loss_yns_5: 0.1366, loss_cls_dn_0: 0.1131, loss_box_dn_0: 0.7179, loss_cls_dn_1: 0.0916, loss_box_dn_1: 0.6036, loss_cls_dn_2: 0.0926, loss_box_dn_2: 0.5940, loss_cls_dn_3: 0.0925, loss_box_dn_3: 0.5885, loss_cls_dn_4: 0.0908, loss_box_dn_4: 0.5965, loss_cls_dn_5: 0.0947, loss_box_dn_5: 0.5963, loss_dense_depth: 0.6882, loss: 22.9410, grad_norm: 40.4992 -2026-01-14 21:44:28,392 - mmdet - INFO - Iter [467/17500] lr: 2.859e-04, eta: 9:10:57, time: 1.555, data_time: 0.073, memory: 49164, loss_cls_0: 0.7224, loss_box_0: 1.5427, loss_cns_0: 0.6443, loss_yns_0: 0.1387, loss_cls_1: 0.7804, loss_box_1: 1.4162, loss_cns_1: 0.6679, loss_yns_1: 0.1357, loss_cls_2: 0.7912, loss_box_2: 1.3929, loss_cns_2: 0.6668, loss_yns_2: 0.1389, loss_cls_3: 0.7995, loss_box_3: 1.3880, loss_cns_3: 0.6682, loss_yns_3: 0.1369, loss_cls_4: 0.8169, loss_box_4: 1.3890, loss_cns_4: 0.6712, loss_yns_4: 0.1389, loss_cls_5: 0.8148, loss_box_5: 1.3746, loss_cns_5: 0.6667, loss_yns_5: 0.1361, loss_cls_dn_0: 0.1095, loss_box_dn_0: 0.7093, loss_cls_dn_1: 0.0927, loss_box_dn_1: 0.6182, loss_cls_dn_2: 0.0918, loss_box_dn_2: 0.6032, loss_cls_dn_3: 0.0904, loss_box_dn_3: 0.6014, loss_cls_dn_4: 0.0921, loss_box_dn_4: 0.6026, loss_cls_dn_5: 0.0920, loss_box_dn_5: 0.6034, loss_dense_depth: 0.6787, loss: 23.0241, grad_norm: 35.2532 -2026-01-14 21:44:35,329 - mmdet - INFO - Iter [468/17500] lr: 2.863e-04, eta: 9:13:57, time: 6.937, data_time: 0.074, memory: 49164, loss_cls_0: 0.7319, loss_box_0: 1.5500, loss_cns_0: 0.6381, loss_yns_0: 0.1390, loss_cls_1: 0.8008, loss_box_1: 1.4452, loss_cns_1: 0.6654, loss_yns_1: 0.1355, loss_cls_2: 0.8029, loss_box_2: 1.4294, loss_cns_2: 0.6649, loss_yns_2: 0.1400, loss_cls_3: 0.8182, loss_box_3: 1.4322, loss_cns_3: 0.6668, loss_yns_3: 0.1368, loss_cls_4: 0.8178, loss_box_4: 1.4204, loss_cns_4: 0.6706, loss_yns_4: 0.1395, loss_cls_5: 0.8259, loss_box_5: 1.4300, loss_cns_5: 0.6686, loss_yns_5: 0.1353, loss_cls_dn_0: 0.1100, loss_box_dn_0: 0.7078, loss_cls_dn_1: 0.0944, loss_box_dn_1: 0.6190, loss_cls_dn_2: 0.0935, loss_box_dn_2: 0.6038, loss_cls_dn_3: 0.0932, loss_box_dn_3: 0.6042, loss_cls_dn_4: 0.0938, loss_box_dn_4: 0.6029, loss_cls_dn_5: 0.0930, loss_box_dn_5: 0.6134, loss_dense_depth: 0.6780, loss: 23.3124, grad_norm: 34.7525 -2026-01-14 21:44:36,862 - mmdet - INFO - Iter [469/17500] lr: 2.867e-04, eta: 9:13:40, time: 1.533, data_time: 0.070, memory: 49164, loss_cls_0: 0.7424, loss_box_0: 1.5594, loss_cns_0: 0.6354, loss_yns_0: 0.1427, loss_cls_1: 0.8152, loss_box_1: 1.4216, loss_cns_1: 0.6662, loss_yns_1: 0.1392, loss_cls_2: 0.8204, loss_box_2: 1.3886, loss_cns_2: 0.6653, loss_yns_2: 0.1415, loss_cls_3: 0.8233, loss_box_3: 1.3847, loss_cns_3: 0.6672, loss_yns_3: 0.1406, loss_cls_4: 0.8367, loss_box_4: 1.3791, loss_cns_4: 0.6669, loss_yns_4: 0.1415, loss_cls_5: 0.8495, loss_box_5: 1.3925, loss_cns_5: 0.6648, loss_yns_5: 0.1402, loss_cls_dn_0: 0.1121, loss_box_dn_0: 0.7146, loss_cls_dn_1: 0.0951, loss_box_dn_1: 0.6235, loss_cls_dn_2: 0.0945, loss_box_dn_2: 0.6058, loss_cls_dn_3: 0.0931, loss_box_dn_3: 0.6062, loss_cls_dn_4: 0.0932, loss_box_dn_4: 0.6106, loss_cls_dn_5: 0.0945, loss_box_dn_5: 0.6202, loss_dense_depth: 0.6885, loss: 23.2766, grad_norm: 42.8984 -2026-01-14 21:44:38,398 - mmdet - INFO - Iter [470/17500] lr: 2.871e-04, eta: 9:13:23, time: 1.534, data_time: 0.071, memory: 49164, loss_cls_0: 0.7398, loss_box_0: 1.5664, loss_cns_0: 0.6393, loss_yns_0: 0.1436, loss_cls_1: 0.8164, loss_box_1: 1.4086, loss_cns_1: 0.6632, loss_yns_1: 0.1411, loss_cls_2: 0.8284, loss_box_2: 1.3901, loss_cns_2: 0.6607, loss_yns_2: 0.1406, loss_cls_3: 0.8340, loss_box_3: 1.3862, loss_cns_3: 0.6618, loss_yns_3: 0.1411, loss_cls_4: 0.8414, loss_box_4: 1.3715, loss_cns_4: 0.6598, loss_yns_4: 0.1399, loss_cls_5: 0.8461, loss_box_5: 1.3752, loss_cns_5: 0.6623, loss_yns_5: 0.1402, loss_cls_dn_0: 0.1138, loss_box_dn_0: 0.7083, loss_cls_dn_1: 0.0937, loss_box_dn_1: 0.6099, loss_cls_dn_2: 0.0918, loss_box_dn_2: 0.5920, loss_cls_dn_3: 0.0918, loss_box_dn_3: 0.5901, loss_cls_dn_4: 0.0920, loss_box_dn_4: 0.5888, loss_cls_dn_5: 0.0921, loss_box_dn_5: 0.5894, loss_dense_depth: 0.6735, loss: 23.1246, grad_norm: 30.0202 -2026-01-14 21:44:39,940 - mmdet - INFO - Iter [471/17500] lr: 2.875e-04, eta: 9:13:06, time: 1.544, data_time: 0.073, memory: 49164, loss_cls_0: 0.7443, loss_box_0: 1.5834, loss_cns_0: 0.6356, loss_yns_0: 0.1404, loss_cls_1: 0.8009, loss_box_1: 1.4242, loss_cns_1: 0.6642, loss_yns_1: 0.1396, loss_cls_2: 0.8152, loss_box_2: 1.4186, loss_cns_2: 0.6604, loss_yns_2: 0.1405, loss_cls_3: 0.8172, loss_box_3: 1.3991, loss_cns_3: 0.6599, loss_yns_3: 0.1400, loss_cls_4: 0.8140, loss_box_4: 1.3873, loss_cns_4: 0.6620, loss_yns_4: 0.1394, loss_cls_5: 0.8182, loss_box_5: 1.3933, loss_cns_5: 0.6655, loss_yns_5: 0.1387, loss_cls_dn_0: 0.1135, loss_box_dn_0: 0.7168, loss_cls_dn_1: 0.0920, loss_box_dn_1: 0.6076, loss_cls_dn_2: 0.0902, loss_box_dn_2: 0.5956, loss_cls_dn_3: 0.0899, loss_box_dn_3: 0.5891, loss_cls_dn_4: 0.0913, loss_box_dn_4: 0.5842, loss_cls_dn_5: 0.0915, loss_box_dn_5: 0.5907, loss_dense_depth: 0.7366, loss: 23.1908, grad_norm: 43.1009 -2026-01-14 21:44:41,621 - mmdet - INFO - Iter [472/17500] lr: 2.879e-04, eta: 9:12:54, time: 1.681, data_time: 0.076, memory: 49164, loss_cls_0: 0.7138, loss_box_0: 1.5167, loss_cns_0: 0.6427, loss_yns_0: 0.1420, loss_cls_1: 0.7920, loss_box_1: 1.3699, loss_cns_1: 0.6708, loss_yns_1: 0.1415, loss_cls_2: 0.8001, loss_box_2: 1.3410, loss_cns_2: 0.6641, loss_yns_2: 0.1426, loss_cls_3: 0.8018, loss_box_3: 1.3543, loss_cns_3: 0.6679, loss_yns_3: 0.1414, loss_cls_4: 0.8045, loss_box_4: 1.3419, loss_cns_4: 0.6723, loss_yns_4: 0.1424, loss_cls_5: 0.8159, loss_box_5: 1.3530, loss_cns_5: 0.6695, loss_yns_5: 0.1448, loss_cls_dn_0: 0.1088, loss_box_dn_0: 0.6957, loss_cls_dn_1: 0.0904, loss_box_dn_1: 0.6095, loss_cls_dn_2: 0.0889, loss_box_dn_2: 0.5999, loss_cls_dn_3: 0.0889, loss_box_dn_3: 0.6068, loss_cls_dn_4: 0.0918, loss_box_dn_4: 0.6052, loss_cls_dn_5: 0.0908, loss_box_dn_5: 0.6162, loss_dense_depth: 0.7036, loss: 22.8435, grad_norm: 40.6806 -2026-01-14 21:44:43,174 - mmdet - INFO - Iter [473/17500] lr: 2.883e-04, eta: 9:12:38, time: 1.553, data_time: 0.069, memory: 49164, loss_cls_0: 0.7135, loss_box_0: 1.5213, loss_cns_0: 0.6393, loss_yns_0: 0.1425, loss_cls_1: 0.7862, loss_box_1: 1.3867, loss_cns_1: 0.6695, loss_yns_1: 0.1396, loss_cls_2: 0.7928, loss_box_2: 1.3511, loss_cns_2: 0.6652, loss_yns_2: 0.1406, loss_cls_3: 0.7825, loss_box_3: 1.3659, loss_cns_3: 0.6675, loss_yns_3: 0.1393, loss_cls_4: 0.7897, loss_box_4: 1.3510, loss_cns_4: 0.6692, loss_yns_4: 0.1384, loss_cls_5: 0.7880, loss_box_5: 1.3472, loss_cns_5: 0.6672, loss_yns_5: 0.1392, loss_cls_dn_0: 0.1088, loss_box_dn_0: 0.7019, loss_cls_dn_1: 0.0931, loss_box_dn_1: 0.6263, loss_cls_dn_2: 0.0904, loss_box_dn_2: 0.6145, loss_cls_dn_3: 0.0903, loss_box_dn_3: 0.6202, loss_cls_dn_4: 0.0908, loss_box_dn_4: 0.6192, loss_cls_dn_5: 0.0900, loss_box_dn_5: 0.6194, loss_dense_depth: 0.7595, loss: 22.9177, grad_norm: 42.1944 -2026-01-14 21:44:44,807 - mmdet - INFO - Iter [474/17500] lr: 2.887e-04, eta: 9:12:24, time: 1.604, data_time: 0.075, memory: 49164, loss_cls_0: 0.7172, loss_box_0: 1.5406, loss_cns_0: 0.6351, loss_yns_0: 0.1409, loss_cls_1: 0.7809, loss_box_1: 1.3875, loss_cns_1: 0.6573, loss_yns_1: 0.1380, loss_cls_2: 0.7825, loss_box_2: 1.3586, loss_cns_2: 0.6596, loss_yns_2: 0.1402, loss_cls_3: 0.7795, loss_box_3: 1.3581, loss_cns_3: 0.6586, loss_yns_3: 0.1394, loss_cls_4: 0.7866, loss_box_4: 1.3631, loss_cns_4: 0.6607, loss_yns_4: 0.1435, loss_cls_5: 0.7783, loss_box_5: 1.3382, loss_cns_5: 0.6601, loss_yns_5: 0.1398, loss_cls_dn_0: 0.1083, loss_box_dn_0: 0.6971, loss_cls_dn_1: 0.0939, loss_box_dn_1: 0.6278, loss_cls_dn_2: 0.0923, loss_box_dn_2: 0.6113, loss_cls_dn_3: 0.0937, loss_box_dn_3: 0.6090, loss_cls_dn_4: 0.0952, loss_box_dn_4: 0.6088, loss_cls_dn_5: 0.0962, loss_box_dn_5: 0.5985, loss_dense_depth: 0.6969, loss: 22.7732, grad_norm: 42.6707 -2026-01-14 21:44:46,416 - mmdet - INFO - Iter [475/17500] lr: 2.891e-04, eta: 9:12:11, time: 1.636, data_time: 0.103, memory: 49164, loss_cls_0: 0.7011, loss_box_0: 1.5209, loss_cns_0: 0.6351, loss_yns_0: 0.1412, loss_cls_1: 0.7543, loss_box_1: 1.3604, loss_cns_1: 0.6565, loss_yns_1: 0.1371, loss_cls_2: 0.7601, loss_box_2: 1.3331, loss_cns_2: 0.6577, loss_yns_2: 0.1367, loss_cls_3: 0.7589, loss_box_3: 1.3546, loss_cns_3: 0.6577, loss_yns_3: 0.1386, loss_cls_4: 0.7590, loss_box_4: 1.3382, loss_cns_4: 0.6583, loss_yns_4: 0.1417, loss_cls_5: 0.7613, loss_box_5: 1.3251, loss_cns_5: 0.6617, loss_yns_5: 0.1373, loss_cls_dn_0: 0.1050, loss_box_dn_0: 0.7046, loss_cls_dn_1: 0.0928, loss_box_dn_1: 0.6063, loss_cls_dn_2: 0.0911, loss_box_dn_2: 0.5862, loss_cls_dn_3: 0.0902, loss_box_dn_3: 0.5869, loss_cls_dn_4: 0.0911, loss_box_dn_4: 0.5789, loss_cls_dn_5: 0.0940, loss_box_dn_5: 0.5767, loss_dense_depth: 0.6992, loss: 22.3896, grad_norm: 40.4125 -2026-01-14 21:44:47,996 - mmdet - INFO - Iter [476/17500] lr: 2.895e-04, eta: 9:11:56, time: 1.582, data_time: 0.091, memory: 49164, loss_cls_0: 0.7096, loss_box_0: 1.5261, loss_cns_0: 0.6394, loss_yns_0: 0.1416, loss_cls_1: 0.7471, loss_box_1: 1.3794, loss_cns_1: 0.6647, loss_yns_1: 0.1378, loss_cls_2: 0.7594, loss_box_2: 1.3370, loss_cns_2: 0.6653, loss_yns_2: 0.1375, loss_cls_3: 0.7721, loss_box_3: 1.3582, loss_cns_3: 0.6648, loss_yns_3: 0.1378, loss_cls_4: 0.7740, loss_box_4: 1.3637, loss_cns_4: 0.6669, loss_yns_4: 0.1379, loss_cls_5: 0.7525, loss_box_5: 1.3876, loss_cns_5: 0.6690, loss_yns_5: 0.1380, loss_cls_dn_0: 0.1030, loss_box_dn_0: 0.7126, loss_cls_dn_1: 0.0884, loss_box_dn_1: 0.5983, loss_cls_dn_2: 0.0860, loss_box_dn_2: 0.5761, loss_cls_dn_3: 0.0849, loss_box_dn_3: 0.5840, loss_cls_dn_4: 0.0846, loss_box_dn_4: 0.5894, loss_cls_dn_5: 0.0893, loss_box_dn_5: 0.6001, loss_dense_depth: 0.6852, loss: 22.5495, grad_norm: 39.2656 -2026-01-14 21:44:49,572 - mmdet - INFO - Iter [477/17500] lr: 2.899e-04, eta: 9:11:41, time: 1.576, data_time: 0.083, memory: 49164, loss_cls_0: 0.7212, loss_box_0: 1.5376, loss_cns_0: 0.6392, loss_yns_0: 0.1410, loss_cls_1: 0.7777, loss_box_1: 1.3828, loss_cns_1: 0.6716, loss_yns_1: 0.1377, loss_cls_2: 0.7734, loss_box_2: 1.3816, loss_cns_2: 0.6654, loss_yns_2: 0.1370, loss_cls_3: 0.7811, loss_box_3: 1.3855, loss_cns_3: 0.6640, loss_yns_3: 0.1368, loss_cls_4: 0.7824, loss_box_4: 1.3796, loss_cns_4: 0.6662, loss_yns_4: 0.1437, loss_cls_5: 0.7802, loss_box_5: 1.3857, loss_cns_5: 0.6661, loss_yns_5: 0.1376, loss_cls_dn_0: 0.1053, loss_box_dn_0: 0.7072, loss_cls_dn_1: 0.0914, loss_box_dn_1: 0.5940, loss_cls_dn_2: 0.0913, loss_box_dn_2: 0.5863, loss_cls_dn_3: 0.0941, loss_box_dn_3: 0.5906, loss_cls_dn_4: 0.0939, loss_box_dn_4: 0.5916, loss_cls_dn_5: 0.0965, loss_box_dn_5: 0.5948, loss_dense_depth: 0.6980, loss: 22.8101, grad_norm: 39.1262 -2026-01-14 21:44:51,144 - mmdet - INFO - Iter [478/17500] lr: 2.903e-04, eta: 9:11:26, time: 1.572, data_time: 0.082, memory: 49164, loss_cls_0: 0.7026, loss_box_0: 1.5374, loss_cns_0: 0.6364, loss_yns_0: 0.1402, loss_cls_1: 0.7695, loss_box_1: 1.3938, loss_cns_1: 0.6645, loss_yns_1: 0.1356, loss_cls_2: 0.7621, loss_box_2: 1.3773, loss_cns_2: 0.6641, loss_yns_2: 0.1377, loss_cls_3: 0.7633, loss_box_3: 1.3515, loss_cns_3: 0.6615, loss_yns_3: 0.1378, loss_cls_4: 0.7702, loss_box_4: 1.3584, loss_cns_4: 0.6651, loss_yns_4: 0.1464, loss_cls_5: 0.7722, loss_box_5: 1.3606, loss_cns_5: 0.6640, loss_yns_5: 0.1381, loss_cls_dn_0: 0.1049, loss_box_dn_0: 0.7088, loss_cls_dn_1: 0.0903, loss_box_dn_1: 0.6023, loss_cls_dn_2: 0.0896, loss_box_dn_2: 0.5846, loss_cls_dn_3: 0.0902, loss_box_dn_3: 0.5750, loss_cls_dn_4: 0.0911, loss_box_dn_4: 0.5751, loss_cls_dn_5: 0.0901, loss_box_dn_5: 0.5771, loss_dense_depth: 0.6798, loss: 22.5692, grad_norm: 28.9485 -2026-01-14 21:44:52,817 - mmdet - INFO - Iter [479/17500] lr: 2.907e-04, eta: 9:11:14, time: 1.672, data_time: 0.191, memory: 49164, loss_cls_0: 0.6916, loss_box_0: 1.5014, loss_cns_0: 0.6354, loss_yns_0: 0.1389, loss_cls_1: 0.7636, loss_box_1: 1.3560, loss_cns_1: 0.6664, loss_yns_1: 0.1371, loss_cls_2: 0.7575, loss_box_2: 1.3509, loss_cns_2: 0.6667, loss_yns_2: 0.1362, loss_cls_3: 0.7638, loss_box_3: 1.3456, loss_cns_3: 0.6640, loss_yns_3: 0.1361, loss_cls_4: 0.7690, loss_box_4: 1.3556, loss_cns_4: 0.6674, loss_yns_4: 0.1392, loss_cls_5: 0.7627, loss_box_5: 1.3743, loss_cns_5: 0.6677, loss_yns_5: 0.1366, loss_cls_dn_0: 0.0984, loss_box_dn_0: 0.7064, loss_cls_dn_1: 0.0901, loss_box_dn_1: 0.5908, loss_cls_dn_2: 0.0888, loss_box_dn_2: 0.5782, loss_cls_dn_3: 0.0864, loss_box_dn_3: 0.5753, loss_cls_dn_4: 0.0889, loss_box_dn_4: 0.5815, loss_cls_dn_5: 0.0921, loss_box_dn_5: 0.5908, loss_dense_depth: 0.6810, loss: 22.4324, grad_norm: 44.8219 -2026-01-14 21:44:54,392 - mmdet - INFO - Iter [480/17500] lr: 2.911e-04, eta: 9:10:59, time: 1.576, data_time: 0.075, memory: 49164, loss_cls_0: 0.7329, loss_box_0: 1.5057, loss_cns_0: 0.6360, loss_yns_0: 0.1385, loss_cls_1: 0.7746, loss_box_1: 1.3734, loss_cns_1: 0.6651, loss_yns_1: 0.1347, loss_cls_2: 0.7697, loss_box_2: 1.3414, loss_cns_2: 0.6628, loss_yns_2: 0.1327, loss_cls_3: 0.7716, loss_box_3: 1.3470, loss_cns_3: 0.6620, loss_yns_3: 0.1351, loss_cls_4: 0.7729, loss_box_4: 1.3636, loss_cns_4: 0.6637, loss_yns_4: 0.1339, loss_cls_5: 0.7736, loss_box_5: 1.3915, loss_cns_5: 0.6658, loss_yns_5: 0.1348, loss_cls_dn_0: 0.1037, loss_box_dn_0: 0.7083, loss_cls_dn_1: 0.0943, loss_box_dn_1: 0.5939, loss_cls_dn_2: 0.0913, loss_box_dn_2: 0.5757, loss_cls_dn_3: 0.0881, loss_box_dn_3: 0.5759, loss_cls_dn_4: 0.0916, loss_box_dn_4: 0.5838, loss_cls_dn_5: 0.0945, loss_box_dn_5: 0.5946, loss_dense_depth: 0.6923, loss: 22.5709, grad_norm: 36.1463 -2026-01-14 21:44:56,098 - mmdet - INFO - Iter [481/17500] lr: 2.915e-04, eta: 9:10:49, time: 1.707, data_time: 0.073, memory: 49164, loss_cls_0: 0.7043, loss_box_0: 1.4668, loss_cns_0: 0.6425, loss_yns_0: 0.1351, loss_cls_1: 0.7446, loss_box_1: 1.3632, loss_cns_1: 0.6664, loss_yns_1: 0.1314, loss_cls_2: 0.7465, loss_box_2: 1.3172, loss_cns_2: 0.6631, loss_yns_2: 0.1301, loss_cls_3: 0.7559, loss_box_3: 1.3222, loss_cns_3: 0.6631, loss_yns_3: 0.1315, loss_cls_4: 0.7532, loss_box_4: 1.3321, loss_cns_4: 0.6638, loss_yns_4: 0.1361, loss_cls_5: 0.7564, loss_box_5: 1.3617, loss_cns_5: 0.6657, loss_yns_5: 0.1318, loss_cls_dn_0: 0.1016, loss_box_dn_0: 0.7043, loss_cls_dn_1: 0.0914, loss_box_dn_1: 0.5940, loss_cls_dn_2: 0.0878, loss_box_dn_2: 0.5698, loss_cls_dn_3: 0.0850, loss_box_dn_3: 0.5706, loss_cls_dn_4: 0.0858, loss_box_dn_4: 0.5766, loss_cls_dn_5: 0.0881, loss_box_dn_5: 0.5879, loss_dense_depth: 0.6713, loss: 22.1987, grad_norm: 44.7404 -2026-01-14 21:44:57,741 - mmdet - INFO - Iter [482/17500] lr: 2.919e-04, eta: 9:10:36, time: 1.642, data_time: 0.089, memory: 49164, loss_cls_0: 0.7147, loss_box_0: 1.4786, loss_cns_0: 0.6453, loss_yns_0: 0.1369, loss_cls_1: 0.7512, loss_box_1: 1.3473, loss_cns_1: 0.6643, loss_yns_1: 0.1321, loss_cls_2: 0.7557, loss_box_2: 1.3202, loss_cns_2: 0.6622, loss_yns_2: 0.1297, loss_cls_3: 0.7636, loss_box_3: 1.3190, loss_cns_3: 0.6617, loss_yns_3: 0.1302, loss_cls_4: 0.7672, loss_box_4: 1.3270, loss_cns_4: 0.6638, loss_yns_4: 0.1356, loss_cls_5: 0.7658, loss_box_5: 1.3290, loss_cns_5: 0.6640, loss_yns_5: 0.1312, loss_cls_dn_0: 0.1007, loss_box_dn_0: 0.7001, loss_cls_dn_1: 0.0874, loss_box_dn_1: 0.5744, loss_cls_dn_2: 0.0863, loss_box_dn_2: 0.5578, loss_cls_dn_3: 0.0859, loss_box_dn_3: 0.5583, loss_cls_dn_4: 0.0861, loss_box_dn_4: 0.5602, loss_cls_dn_5: 0.0865, loss_box_dn_5: 0.5610, loss_dense_depth: 0.6901, loss: 22.1311, grad_norm: 32.8694 -2026-01-14 21:44:59,349 - mmdet - INFO - Iter [483/17500] lr: 2.923e-04, eta: 9:10:21, time: 1.570, data_time: 0.074, memory: 49164, loss_cls_0: 0.6889, loss_box_0: 1.4969, loss_cns_0: 0.6458, loss_yns_0: 0.1328, loss_cls_1: 0.7343, loss_box_1: 1.4077, loss_cns_1: 0.6626, loss_yns_1: 0.1296, loss_cls_2: 0.7415, loss_box_2: 1.3961, loss_cns_2: 0.6626, loss_yns_2: 0.1299, loss_cls_3: 0.7505, loss_box_3: 1.3750, loss_cns_3: 0.6630, loss_yns_3: 0.1304, loss_cls_4: 0.7559, loss_box_4: 1.3707, loss_cns_4: 0.6634, loss_yns_4: 0.1345, loss_cls_5: 0.7544, loss_box_5: 1.3741, loss_cns_5: 0.6640, loss_yns_5: 0.1304, loss_cls_dn_0: 0.0995, loss_box_dn_0: 0.6994, loss_cls_dn_1: 0.0838, loss_box_dn_1: 0.5768, loss_cls_dn_2: 0.0834, loss_box_dn_2: 0.5610, loss_cls_dn_3: 0.0831, loss_box_dn_3: 0.5561, loss_cls_dn_4: 0.0849, loss_box_dn_4: 0.5547, loss_cls_dn_5: 0.0836, loss_box_dn_5: 0.5557, loss_dense_depth: 0.6609, loss: 22.2778, grad_norm: 45.9988 -2026-01-14 21:45:00,973 - mmdet - INFO - Iter [484/17500] lr: 2.926e-04, eta: 9:10:09, time: 1.662, data_time: 0.114, memory: 49164, loss_cls_0: 0.6953, loss_box_0: 1.5101, loss_cns_0: 0.6432, loss_yns_0: 0.1356, loss_cls_1: 0.7472, loss_box_1: 1.4226, loss_cns_1: 0.6636, loss_yns_1: 0.1321, loss_cls_2: 0.7565, loss_box_2: 1.3832, loss_cns_2: 0.6647, loss_yns_2: 0.1314, loss_cls_3: 0.7605, loss_box_3: 1.3705, loss_cns_3: 0.6631, loss_yns_3: 0.1308, loss_cls_4: 0.7553, loss_box_4: 1.3716, loss_cns_4: 0.6634, loss_yns_4: 0.1342, loss_cls_5: 0.7579, loss_box_5: 1.3735, loss_cns_5: 0.6632, loss_yns_5: 0.1321, loss_cls_dn_0: 0.0997, loss_box_dn_0: 0.6972, loss_cls_dn_1: 0.0897, loss_box_dn_1: 0.6100, loss_cls_dn_2: 0.0901, loss_box_dn_2: 0.5851, loss_cls_dn_3: 0.0893, loss_box_dn_3: 0.5824, loss_cls_dn_4: 0.0893, loss_box_dn_4: 0.5859, loss_cls_dn_5: 0.0887, loss_box_dn_5: 0.5888, loss_dense_depth: 0.6801, loss: 22.5380, grad_norm: 41.8846 -2026-01-14 21:45:02,628 - mmdet - INFO - Iter [485/17500] lr: 2.930e-04, eta: 9:09:58, time: 1.656, data_time: 0.075, memory: 49164, loss_cls_0: 0.6720, loss_box_0: 1.4830, loss_cns_0: 0.6437, loss_yns_0: 0.1302, loss_cls_1: 0.7353, loss_box_1: 1.3727, loss_cns_1: 0.6630, loss_yns_1: 0.1283, loss_cls_2: 0.7361, loss_box_2: 1.3449, loss_cns_2: 0.6629, loss_yns_2: 0.1276, loss_cls_3: 0.7470, loss_box_3: 1.3364, loss_cns_3: 0.6640, loss_yns_3: 0.1280, loss_cls_4: 0.7375, loss_box_4: 1.3318, loss_cns_4: 0.6649, loss_yns_4: 0.1287, loss_cls_5: 0.7366, loss_box_5: 1.3196, loss_cns_5: 0.6616, loss_yns_5: 0.1288, loss_cls_dn_0: 0.0992, loss_box_dn_0: 0.7038, loss_cls_dn_1: 0.0907, loss_box_dn_1: 0.6118, loss_cls_dn_2: 0.0902, loss_box_dn_2: 0.5927, loss_cls_dn_3: 0.0893, loss_box_dn_3: 0.5857, loss_cls_dn_4: 0.0889, loss_box_dn_4: 0.5864, loss_cls_dn_5: 0.0887, loss_box_dn_5: 0.5840, loss_dense_depth: 0.6354, loss: 22.1315, grad_norm: 47.1188 -2026-01-14 21:45:04,291 - mmdet - INFO - Iter [486/17500] lr: 2.934e-04, eta: 9:09:46, time: 1.662, data_time: 0.075, memory: 49164, loss_cls_0: 0.6745, loss_box_0: 1.4697, loss_cns_0: 0.6442, loss_yns_0: 0.1288, loss_cls_1: 0.7222, loss_box_1: 1.3817, loss_cns_1: 0.6532, loss_yns_1: 0.1258, loss_cls_2: 0.7231, loss_box_2: 1.3688, loss_cns_2: 0.6518, loss_yns_2: 0.1248, loss_cls_3: 0.7283, loss_box_3: 1.3487, loss_cns_3: 0.6568, loss_yns_3: 0.1254, loss_cls_4: 0.7364, loss_box_4: 1.3133, loss_cns_4: 0.6530, loss_yns_4: 0.1236, loss_cls_5: 0.7457, loss_box_5: 1.3264, loss_cns_5: 0.6577, loss_yns_5: 0.1252, loss_cls_dn_0: 0.0961, loss_box_dn_0: 0.7009, loss_cls_dn_1: 0.0888, loss_box_dn_1: 0.6027, loss_cls_dn_2: 0.0875, loss_box_dn_2: 0.5843, loss_cls_dn_3: 0.0877, loss_box_dn_3: 0.5732, loss_cls_dn_4: 0.0869, loss_box_dn_4: 0.5648, loss_cls_dn_5: 0.0866, loss_box_dn_5: 0.5661, loss_dense_depth: 0.6696, loss: 22.0046, grad_norm: 33.9098 -2026-01-14 21:45:05,896 - mmdet - INFO - Iter [487/17500] lr: 2.938e-04, eta: 9:09:31, time: 1.562, data_time: 0.076, memory: 49164, loss_cls_0: 0.6708, loss_box_0: 1.4984, loss_cns_0: 0.6461, loss_yns_0: 0.1321, loss_cls_1: 0.7179, loss_box_1: 1.4050, loss_cns_1: 0.6587, loss_yns_1: 0.1291, loss_cls_2: 0.7169, loss_box_2: 1.3831, loss_cns_2: 0.6568, loss_yns_2: 0.1270, loss_cls_3: 0.7316, loss_box_3: 1.3620, loss_cns_3: 0.6597, loss_yns_3: 0.1273, loss_cls_4: 0.7375, loss_box_4: 1.3708, loss_cns_4: 0.6590, loss_yns_4: 0.1299, loss_cls_5: 0.7446, loss_box_5: 1.3907, loss_cns_5: 0.6611, loss_yns_5: 0.1279, loss_cls_dn_0: 0.0990, loss_box_dn_0: 0.7272, loss_cls_dn_1: 0.0870, loss_box_dn_1: 0.6114, loss_cls_dn_2: 0.0864, loss_box_dn_2: 0.5905, loss_cls_dn_3: 0.0850, loss_box_dn_3: 0.5834, loss_cls_dn_4: 0.0858, loss_box_dn_4: 0.5864, loss_cls_dn_5: 0.0861, loss_box_dn_5: 0.5969, loss_dense_depth: 0.6551, loss: 22.3243, grad_norm: 48.3403 -2026-01-14 21:45:07,515 - mmdet - INFO - Iter [488/17500] lr: 2.942e-04, eta: 9:09:19, time: 1.662, data_time: 0.101, memory: 49164, loss_cls_0: 0.6839, loss_box_0: 1.4891, loss_cns_0: 0.6437, loss_yns_0: 0.1342, loss_cls_1: 0.7404, loss_box_1: 1.3898, loss_cns_1: 0.6635, loss_yns_1: 0.1297, loss_cls_2: 0.7383, loss_box_2: 1.3454, loss_cns_2: 0.6627, loss_yns_2: 0.1278, loss_cls_3: 0.7397, loss_box_3: 1.3314, loss_cns_3: 0.6637, loss_yns_3: 0.1287, loss_cls_4: 0.7419, loss_box_4: 1.3337, loss_cns_4: 0.6707, loss_yns_4: 0.1340, loss_cls_5: 0.7448, loss_box_5: 1.3526, loss_cns_5: 0.6648, loss_yns_5: 0.1299, loss_cls_dn_0: 0.0985, loss_box_dn_0: 0.7043, loss_cls_dn_1: 0.0858, loss_box_dn_1: 0.6016, loss_cls_dn_2: 0.0880, loss_box_dn_2: 0.5757, loss_cls_dn_3: 0.0857, loss_box_dn_3: 0.5732, loss_cls_dn_4: 0.0863, loss_box_dn_4: 0.5740, loss_cls_dn_5: 0.0867, loss_box_dn_5: 0.5840, loss_dense_depth: 0.6866, loss: 22.2147, grad_norm: 37.1387 -2026-01-14 21:45:09,101 - mmdet - INFO - Iter [489/17500] lr: 2.946e-04, eta: 9:09:05, time: 1.587, data_time: 0.075, memory: 49164, loss_cls_0: 0.6981, loss_box_0: 1.5312, loss_cns_0: 0.6359, loss_yns_0: 0.1367, loss_cls_1: 0.7528, loss_box_1: 1.4125, loss_cns_1: 0.6636, loss_yns_1: 0.1342, loss_cls_2: 0.7638, loss_box_2: 1.4031, loss_cns_2: 0.6609, loss_yns_2: 0.1329, loss_cls_3: 0.7740, loss_box_3: 1.4072, loss_cns_3: 0.6594, loss_yns_3: 0.1342, loss_cls_4: 0.7654, loss_box_4: 1.3962, loss_cns_4: 0.6622, loss_yns_4: 0.1379, loss_cls_5: 0.7721, loss_box_5: 1.3966, loss_cns_5: 0.6584, loss_yns_5: 0.1352, loss_cls_dn_0: 0.0966, loss_box_dn_0: 0.7142, loss_cls_dn_1: 0.0869, loss_box_dn_1: 0.6055, loss_cls_dn_2: 0.0873, loss_box_dn_2: 0.5914, loss_cls_dn_3: 0.0861, loss_box_dn_3: 0.5917, loss_cls_dn_4: 0.0861, loss_box_dn_4: 0.5852, loss_cls_dn_5: 0.0868, loss_box_dn_5: 0.5881, loss_dense_depth: 0.6674, loss: 22.6979, grad_norm: 47.4752 -2026-01-14 21:45:10,658 - mmdet - INFO - Iter [490/17500] lr: 2.950e-04, eta: 9:08:50, time: 1.555, data_time: 0.070, memory: 49164, loss_cls_0: 0.7116, loss_box_0: 1.5548, loss_cns_0: 0.6334, loss_yns_0: 0.1393, loss_cls_1: 0.7665, loss_box_1: 1.4231, loss_cns_1: 0.6624, loss_yns_1: 0.1371, loss_cls_2: 0.7700, loss_box_2: 1.3996, loss_cns_2: 0.6598, loss_yns_2: 0.1360, loss_cls_3: 0.7840, loss_box_3: 1.4015, loss_cns_3: 0.6611, loss_yns_3: 0.1373, loss_cls_4: 0.7823, loss_box_4: 1.3991, loss_cns_4: 0.6624, loss_yns_4: 0.1388, loss_cls_5: 0.7922, loss_box_5: 1.3963, loss_cns_5: 0.6611, loss_yns_5: 0.1373, loss_cls_dn_0: 0.1012, loss_box_dn_0: 0.7307, loss_cls_dn_1: 0.0916, loss_box_dn_1: 0.6186, loss_cls_dn_2: 0.0906, loss_box_dn_2: 0.5996, loss_cls_dn_3: 0.0899, loss_box_dn_3: 0.5996, loss_cls_dn_4: 0.0912, loss_box_dn_4: 0.5958, loss_cls_dn_5: 0.0917, loss_box_dn_5: 0.5985, loss_dense_depth: 0.7109, loss: 22.9569, grad_norm: 41.0561 -2026-01-14 21:45:12,227 - mmdet - INFO - Iter [491/17500] lr: 2.954e-04, eta: 9:08:35, time: 1.571, data_time: 0.075, memory: 49164, loss_cls_0: 0.7174, loss_box_0: 1.5361, loss_cns_0: 0.6389, loss_yns_0: 0.1394, loss_cls_1: 0.7775, loss_box_1: 1.4191, loss_cns_1: 0.6658, loss_yns_1: 0.1363, loss_cls_2: 0.7805, loss_box_2: 1.3854, loss_cns_2: 0.6620, loss_yns_2: 0.1363, loss_cls_3: 0.7892, loss_box_3: 1.3861, loss_cns_3: 0.6633, loss_yns_3: 0.1374, loss_cls_4: 0.7919, loss_box_4: 1.3847, loss_cns_4: 0.6643, loss_yns_4: 0.1372, loss_cls_5: 0.7911, loss_box_5: 1.3859, loss_cns_5: 0.6637, loss_yns_5: 0.1369, loss_cls_dn_0: 0.1020, loss_box_dn_0: 0.7059, loss_cls_dn_1: 0.0903, loss_box_dn_1: 0.6226, loss_cls_dn_2: 0.0904, loss_box_dn_2: 0.6060, loss_cls_dn_3: 0.0901, loss_box_dn_3: 0.6061, loss_cls_dn_4: 0.0927, loss_box_dn_4: 0.6047, loss_cls_dn_5: 0.0916, loss_box_dn_5: 0.6077, loss_dense_depth: 0.6732, loss: 22.9096, grad_norm: 40.0508 -2026-01-14 21:45:13,815 - mmdet - INFO - Iter [492/17500] lr: 2.958e-04, eta: 9:08:21, time: 1.585, data_time: 0.073, memory: 49164, loss_cls_0: 0.6782, loss_box_0: 1.5461, loss_cns_0: 0.6407, loss_yns_0: 0.1393, loss_cls_1: 0.7696, loss_box_1: 1.3797, loss_cns_1: 0.6662, loss_yns_1: 0.1369, loss_cls_2: 0.7676, loss_box_2: 1.3643, loss_cns_2: 0.6637, loss_yns_2: 0.1361, loss_cls_3: 0.7898, loss_box_3: 1.3591, loss_cns_3: 0.6635, loss_yns_3: 0.1354, loss_cls_4: 0.7821, loss_box_4: 1.3564, loss_cns_4: 0.6634, loss_yns_4: 0.1354, loss_cls_5: 0.7725, loss_box_5: 1.3552, loss_cns_5: 0.6625, loss_yns_5: 0.1363, loss_cls_dn_0: 0.1008, loss_box_dn_0: 0.7035, loss_cls_dn_1: 0.0862, loss_box_dn_1: 0.6134, loss_cls_dn_2: 0.0854, loss_box_dn_2: 0.6017, loss_cls_dn_3: 0.0865, loss_box_dn_3: 0.5975, loss_cls_dn_4: 0.0882, loss_box_dn_4: 0.5945, loss_cls_dn_5: 0.0851, loss_box_dn_5: 0.5956, loss_dense_depth: 0.6756, loss: 22.6138, grad_norm: 36.5047 -2026-01-14 21:45:15,388 - mmdet - INFO - Iter [493/17500] lr: 2.962e-04, eta: 9:08:07, time: 1.575, data_time: 0.077, memory: 49164, loss_cls_0: 0.7067, loss_box_0: 1.5520, loss_cns_0: 0.6358, loss_yns_0: 0.1398, loss_cls_1: 0.7722, loss_box_1: 1.4147, loss_cns_1: 0.6638, loss_yns_1: 0.1398, loss_cls_2: 0.7784, loss_box_2: 1.3608, loss_cns_2: 0.6605, loss_yns_2: 0.1381, loss_cls_3: 0.7843, loss_box_3: 1.3585, loss_cns_3: 0.6594, loss_yns_3: 0.1374, loss_cls_4: 0.7908, loss_box_4: 1.3624, loss_cns_4: 0.6589, loss_yns_4: 0.1394, loss_cls_5: 0.7970, loss_box_5: 1.3679, loss_cns_5: 0.6576, loss_yns_5: 0.1384, loss_cls_dn_0: 0.1051, loss_box_dn_0: 0.7035, loss_cls_dn_1: 0.0912, loss_box_dn_1: 0.6140, loss_cls_dn_2: 0.0913, loss_box_dn_2: 0.5837, loss_cls_dn_3: 0.0922, loss_box_dn_3: 0.5807, loss_cls_dn_4: 0.0949, loss_box_dn_4: 0.5820, loss_cls_dn_5: 0.0905, loss_box_dn_5: 0.5863, loss_dense_depth: 0.6741, loss: 22.7041, grad_norm: 33.0921 -2026-01-14 21:45:16,963 - mmdet - INFO - Iter [494/17500] lr: 2.966e-04, eta: 9:07:53, time: 1.573, data_time: 0.074, memory: 49164, loss_cls_0: 0.6974, loss_box_0: 1.5116, loss_cns_0: 0.6417, loss_yns_0: 0.1371, loss_cls_1: 0.7677, loss_box_1: 1.4438, loss_cns_1: 0.6586, loss_yns_1: 0.1362, loss_cls_2: 0.7744, loss_box_2: 1.3826, loss_cns_2: 0.6561, loss_yns_2: 0.1370, loss_cls_3: 0.7750, loss_box_3: 1.3747, loss_cns_3: 0.6561, loss_yns_3: 0.1361, loss_cls_4: 0.7821, loss_box_4: 1.3822, loss_cns_4: 0.6608, loss_yns_4: 0.1403, loss_cls_5: 0.7930, loss_box_5: 1.3888, loss_cns_5: 0.6562, loss_yns_5: 0.1356, loss_cls_dn_0: 0.1017, loss_box_dn_0: 0.6953, loss_cls_dn_1: 0.0884, loss_box_dn_1: 0.6195, loss_cls_dn_2: 0.0899, loss_box_dn_2: 0.5867, loss_cls_dn_3: 0.0888, loss_box_dn_3: 0.5852, loss_cls_dn_4: 0.0905, loss_box_dn_4: 0.5881, loss_cls_dn_5: 0.0892, loss_box_dn_5: 0.5959, loss_dense_depth: 0.6849, loss: 22.7293, grad_norm: 37.0959 -2026-01-14 21:45:18,604 - mmdet - INFO - Iter [495/17500] lr: 2.970e-04, eta: 9:07:41, time: 1.643, data_time: 0.075, memory: 49164, loss_cls_0: 0.7154, loss_box_0: 1.5352, loss_cns_0: 0.6466, loss_yns_0: 0.1395, loss_cls_1: 0.7443, loss_box_1: 1.4059, loss_cns_1: 0.6590, loss_yns_1: 0.1342, loss_cls_2: 0.7587, loss_box_2: 1.3712, loss_cns_2: 0.6567, loss_yns_2: 0.1327, loss_cls_3: 0.7612, loss_box_3: 1.3828, loss_cns_3: 0.6586, loss_yns_3: 0.1337, loss_cls_4: 0.7619, loss_box_4: 1.3954, loss_cns_4: 0.6631, loss_yns_4: 0.1340, loss_cls_5: 0.7687, loss_box_5: 1.3909, loss_cns_5: 0.6560, loss_yns_5: 0.1337, loss_cls_dn_0: 0.1062, loss_box_dn_0: 0.7096, loss_cls_dn_1: 0.0875, loss_box_dn_1: 0.6098, loss_cls_dn_2: 0.0873, loss_box_dn_2: 0.5886, loss_cls_dn_3: 0.0857, loss_box_dn_3: 0.5873, loss_cls_dn_4: 0.0861, loss_box_dn_4: 0.5910, loss_cls_dn_5: 0.0875, loss_box_dn_5: 0.5932, loss_dense_depth: 0.6809, loss: 22.6401, grad_norm: 32.9936 -2026-01-14 21:45:20,184 - mmdet - INFO - Iter [496/17500] lr: 2.974e-04, eta: 9:07:27, time: 1.577, data_time: 0.082, memory: 49164, loss_cls_0: 0.7185, loss_box_0: 1.5293, loss_cns_0: 0.6464, loss_yns_0: 0.1384, loss_cls_1: 0.7519, loss_box_1: 1.4239, loss_cns_1: 0.6590, loss_yns_1: 0.1346, loss_cls_2: 0.7600, loss_box_2: 1.3919, loss_cns_2: 0.6586, loss_yns_2: 0.1341, loss_cls_3: 0.7607, loss_box_3: 1.3782, loss_cns_3: 0.6580, loss_yns_3: 0.1343, loss_cls_4: 0.7650, loss_box_4: 1.3874, loss_cns_4: 0.6585, loss_yns_4: 0.1340, loss_cls_5: 0.7745, loss_box_5: 1.3946, loss_cns_5: 0.6591, loss_yns_5: 0.1345, loss_cls_dn_0: 0.1044, loss_box_dn_0: 0.6986, loss_cls_dn_1: 0.0857, loss_box_dn_1: 0.6118, loss_cls_dn_2: 0.0849, loss_box_dn_2: 0.5924, loss_cls_dn_3: 0.0838, loss_box_dn_3: 0.5847, loss_cls_dn_4: 0.0875, loss_box_dn_4: 0.5855, loss_cls_dn_5: 0.0859, loss_box_dn_5: 0.5895, loss_dense_depth: 0.6712, loss: 22.6509, grad_norm: 29.6501 -2026-01-14 21:45:21,763 - mmdet - INFO - Iter [497/17500] lr: 2.978e-04, eta: 9:07:13, time: 1.582, data_time: 0.084, memory: 49164, loss_cls_0: 0.6928, loss_box_0: 1.5418, loss_cns_0: 0.6408, loss_yns_0: 0.1366, loss_cls_1: 0.7452, loss_box_1: 1.3681, loss_cns_1: 0.6627, loss_yns_1: 0.1355, loss_cls_2: 0.7528, loss_box_2: 1.3767, loss_cns_2: 0.6634, loss_yns_2: 0.1348, loss_cls_3: 0.7648, loss_box_3: 1.3539, loss_cns_3: 0.6632, loss_yns_3: 0.1351, loss_cls_4: 0.7617, loss_box_4: 1.3647, loss_cns_4: 0.6645, loss_yns_4: 0.1352, loss_cls_5: 0.7621, loss_box_5: 1.3612, loss_cns_5: 0.6647, loss_yns_5: 0.1356, loss_cls_dn_0: 0.0938, loss_box_dn_0: 0.7070, loss_cls_dn_1: 0.0834, loss_box_dn_1: 0.5910, loss_cls_dn_2: 0.0828, loss_box_dn_2: 0.5932, loss_cls_dn_3: 0.0826, loss_box_dn_3: 0.5884, loss_cls_dn_4: 0.0869, loss_box_dn_4: 0.5896, loss_cls_dn_5: 0.0858, loss_box_dn_5: 0.5891, loss_dense_depth: 0.6778, loss: 22.4694, grad_norm: 42.3956 -2026-01-14 21:45:23,388 - mmdet - INFO - Iter [498/17500] lr: 2.982e-04, eta: 9:06:59, time: 1.578, data_time: 0.081, memory: 49164, loss_cls_0: 0.7183, loss_box_0: 1.5326, loss_cns_0: 0.6355, loss_yns_0: 0.1352, loss_cls_1: 0.7424, loss_box_1: 1.3585, loss_cns_1: 0.6673, loss_yns_1: 0.1343, loss_cls_2: 0.7574, loss_box_2: 1.3481, loss_cns_2: 0.6667, loss_yns_2: 0.1334, loss_cls_3: 0.7587, loss_box_3: 1.3423, loss_cns_3: 0.6672, loss_yns_3: 0.1340, loss_cls_4: 0.7619, loss_box_4: 1.3555, loss_cns_4: 0.6674, loss_yns_4: 0.1350, loss_cls_5: 0.7673, loss_box_5: 1.3585, loss_cns_5: 0.6691, loss_yns_5: 0.1347, loss_cls_dn_0: 0.0991, loss_box_dn_0: 0.6988, loss_cls_dn_1: 0.0846, loss_box_dn_1: 0.6181, loss_cls_dn_2: 0.0838, loss_box_dn_2: 0.6152, loss_cls_dn_3: 0.0837, loss_box_dn_3: 0.6165, loss_cls_dn_4: 0.0877, loss_box_dn_4: 0.6247, loss_cls_dn_5: 0.0885, loss_box_dn_5: 0.6274, loss_dense_depth: 0.6960, loss: 22.6051, grad_norm: 39.6708 -2026-01-14 21:45:25,038 - mmdet - INFO - Iter [499/17500] lr: 2.986e-04, eta: 9:06:49, time: 1.696, data_time: 0.216, memory: 49164, loss_cls_0: 0.7506, loss_box_0: 1.5123, loss_cns_0: 0.6408, loss_yns_0: 0.1387, loss_cls_1: 0.7706, loss_box_1: 1.4025, loss_cns_1: 0.6677, loss_yns_1: 0.1378, loss_cls_2: 0.7816, loss_box_2: 1.3747, loss_cns_2: 0.6687, loss_yns_2: 0.1367, loss_cls_3: 0.7818, loss_box_3: 1.3694, loss_cns_3: 0.6689, loss_yns_3: 0.1376, loss_cls_4: 0.7824, loss_box_4: 1.3630, loss_cns_4: 0.6727, loss_yns_4: 0.1388, loss_cls_5: 0.7923, loss_box_5: 1.3544, loss_cns_5: 0.6681, loss_yns_5: 0.1372, loss_cls_dn_0: 0.1131, loss_box_dn_0: 0.6927, loss_cls_dn_1: 0.0872, loss_box_dn_1: 0.6304, loss_cls_dn_2: 0.0871, loss_box_dn_2: 0.6104, loss_cls_dn_3: 0.0876, loss_box_dn_3: 0.6072, loss_cls_dn_4: 0.0879, loss_box_dn_4: 0.6050, loss_cls_dn_5: 0.0900, loss_box_dn_5: 0.6027, loss_dense_depth: 0.6966, loss: 22.8470, grad_norm: 42.0250 -2026-01-14 21:45:26,598 - mmdet - INFO - Iter [500/17500] lr: 2.990e-04, eta: 9:06:34, time: 1.561, data_time: 0.075, memory: 49164, loss_cls_0: 0.7029, loss_box_0: 1.5107, loss_cns_0: 0.6482, loss_yns_0: 0.1383, loss_cls_1: 0.7567, loss_box_1: 1.3648, loss_cns_1: 0.6710, loss_yns_1: 0.1357, loss_cls_2: 0.7690, loss_box_2: 1.3487, loss_cns_2: 0.6695, loss_yns_2: 0.1336, loss_cls_3: 0.7637, loss_box_3: 1.3282, loss_cns_3: 0.6679, loss_yns_3: 0.1341, loss_cls_4: 0.7714, loss_box_4: 1.3191, loss_cns_4: 0.6724, loss_yns_4: 0.1353, loss_cls_5: 0.7765, loss_box_5: 1.3130, loss_cns_5: 0.6683, loss_yns_5: 0.1343, loss_cls_dn_0: 0.1042, loss_box_dn_0: 0.7009, loss_cls_dn_1: 0.0850, loss_box_dn_1: 0.6263, loss_cls_dn_2: 0.0847, loss_box_dn_2: 0.6087, loss_cls_dn_3: 0.0838, loss_box_dn_3: 0.5984, loss_cls_dn_4: 0.0833, loss_box_dn_4: 0.5925, loss_cls_dn_5: 0.0846, loss_box_dn_5: 0.5902, loss_dense_depth: 0.6874, loss: 22.4634, grad_norm: 41.8834 -2026-01-14 21:45:28,304 - mmdet - INFO - Iter [501/17500] lr: 2.994e-04, eta: 9:06:25, time: 1.706, data_time: 0.074, memory: 49164, loss_cls_0: 0.7001, loss_box_0: 1.4954, loss_cns_0: 0.6416, loss_yns_0: 0.1384, loss_cls_1: 0.7674, loss_box_1: 1.3408, loss_cns_1: 0.6675, loss_yns_1: 0.1347, loss_cls_2: 0.7735, loss_box_2: 1.3357, loss_cns_2: 0.6659, loss_yns_2: 0.1336, loss_cls_3: 0.7639, loss_box_3: 1.3210, loss_cns_3: 0.6652, loss_yns_3: 0.1328, loss_cls_4: 0.7800, loss_box_4: 1.3266, loss_cns_4: 0.6658, loss_yns_4: 0.1346, loss_cls_5: 0.7807, loss_box_5: 1.3259, loss_cns_5: 0.6663, loss_yns_5: 0.1341, loss_cls_dn_0: 0.0915, loss_box_dn_0: 0.6980, loss_cls_dn_1: 0.0819, loss_box_dn_1: 0.5977, loss_cls_dn_2: 0.0817, loss_box_dn_2: 0.5847, loss_cls_dn_3: 0.0805, loss_box_dn_3: 0.5762, loss_cls_dn_4: 0.0812, loss_box_dn_4: 0.5818, loss_cls_dn_5: 0.0816, loss_box_dn_5: 0.5862, loss_dense_depth: 0.6788, loss: 22.2933, grad_norm: 35.0270 -2026-01-14 21:45:29,980 - mmdet - INFO - Iter [502/17500] lr: 2.994e-04, eta: 9:06:14, time: 1.675, data_time: 0.089, memory: 49164, loss_cls_0: 0.7266, loss_box_0: 1.5140, loss_cns_0: 0.6370, loss_yns_0: 0.1369, loss_cls_1: 0.7742, loss_box_1: 1.4008, loss_cns_1: 0.6644, loss_yns_1: 0.1338, loss_cls_2: 0.7816, loss_box_2: 1.3594, loss_cns_2: 0.6639, loss_yns_2: 0.1312, loss_cls_3: 0.7786, loss_box_3: 1.3313, loss_cns_3: 0.6626, loss_yns_3: 0.1311, loss_cls_4: 0.7853, loss_box_4: 1.3505, loss_cns_4: 0.6640, loss_yns_4: 0.1327, loss_cls_5: 0.7943, loss_box_5: 1.3379, loss_cns_5: 0.6666, loss_yns_5: 0.1332, loss_cls_dn_0: 0.1037, loss_box_dn_0: 0.6951, loss_cls_dn_1: 0.0853, loss_box_dn_1: 0.6000, loss_cls_dn_2: 0.0846, loss_box_dn_2: 0.5788, loss_cls_dn_3: 0.0837, loss_box_dn_3: 0.5718, loss_cls_dn_4: 0.0835, loss_box_dn_4: 0.5781, loss_cls_dn_5: 0.0853, loss_box_dn_5: 0.5810, loss_dense_depth: 0.6740, loss: 22.4970, grad_norm: 35.2914 -2026-01-14 21:45:31,558 - mmdet - INFO - Iter [503/17500] lr: 2.994e-04, eta: 9:06:01, time: 1.580, data_time: 0.072, memory: 49164, loss_cls_0: 0.6916, loss_box_0: 1.5064, loss_cns_0: 0.6397, loss_yns_0: 0.1381, loss_cls_1: 0.7477, loss_box_1: 1.3599, loss_cns_1: 0.6601, loss_yns_1: 0.1351, loss_cls_2: 0.7737, loss_box_2: 1.3107, loss_cns_2: 0.6556, loss_yns_2: 0.1336, loss_cls_3: 0.7669, loss_box_3: 1.3153, loss_cns_3: 0.6568, loss_yns_3: 0.1341, loss_cls_4: 0.7712, loss_box_4: 1.3086, loss_cns_4: 0.6566, loss_yns_4: 0.1339, loss_cls_5: 0.7703, loss_box_5: 1.3122, loss_cns_5: 0.6596, loss_yns_5: 0.1344, loss_cls_dn_0: 0.0968, loss_box_dn_0: 0.7051, loss_cls_dn_1: 0.0790, loss_box_dn_1: 0.5808, loss_cls_dn_2: 0.0790, loss_box_dn_2: 0.5615, loss_cls_dn_3: 0.0777, loss_box_dn_3: 0.5563, loss_cls_dn_4: 0.0778, loss_box_dn_4: 0.5561, loss_cls_dn_5: 0.0781, loss_box_dn_5: 0.5603, loss_dense_depth: 0.6383, loss: 22.0186, grad_norm: 30.7008 -2026-01-14 21:45:33,191 - mmdet - INFO - Iter [504/17500] lr: 2.994e-04, eta: 9:05:48, time: 1.606, data_time: 0.072, memory: 49164, loss_cls_0: 0.7290, loss_box_0: 1.4929, loss_cns_0: 0.6328, loss_yns_0: 0.1367, loss_cls_1: 0.7602, loss_box_1: 1.3371, loss_cns_1: 0.6580, loss_yns_1: 0.1348, loss_cls_2: 0.7765, loss_box_2: 1.3041, loss_cns_2: 0.6550, loss_yns_2: 0.1327, loss_cls_3: 0.7806, loss_box_3: 1.3101, loss_cns_3: 0.6582, loss_yns_3: 0.1333, loss_cls_4: 0.7825, loss_box_4: 1.3052, loss_cns_4: 0.6581, loss_yns_4: 0.1345, loss_cls_5: 0.7875, loss_box_5: 1.3164, loss_cns_5: 0.6596, loss_yns_5: 0.1343, loss_cls_dn_0: 0.0952, loss_box_dn_0: 0.6948, loss_cls_dn_1: 0.0810, loss_box_dn_1: 0.5682, loss_cls_dn_2: 0.0806, loss_box_dn_2: 0.5524, loss_cls_dn_3: 0.0808, loss_box_dn_3: 0.5465, loss_cls_dn_4: 0.0805, loss_box_dn_4: 0.5454, loss_cls_dn_5: 0.0811, loss_box_dn_5: 0.5510, loss_dense_depth: 0.7095, loss: 22.0772, grad_norm: 23.9594 -2026-01-14 21:45:34,855 - mmdet - INFO - Iter [505/17500] lr: 2.994e-04, eta: 9:05:38, time: 1.687, data_time: 0.090, memory: 49164, loss_cls_0: 0.7105, loss_box_0: 1.5013, loss_cns_0: 0.6402, loss_yns_0: 0.1376, loss_cls_1: 0.7779, loss_box_1: 1.3393, loss_cns_1: 0.6602, loss_yns_1: 0.1346, loss_cls_2: 0.7874, loss_box_2: 1.3193, loss_cns_2: 0.6608, loss_yns_2: 0.1334, loss_cls_3: 0.7880, loss_box_3: 1.3254, loss_cns_3: 0.6611, loss_yns_3: 0.1332, loss_cls_4: 0.7886, loss_box_4: 1.3324, loss_cns_4: 0.6598, loss_yns_4: 0.1364, loss_cls_5: 0.7891, loss_box_5: 1.3277, loss_cns_5: 0.6609, loss_yns_5: 0.1358, loss_cls_dn_0: 0.0985, loss_box_dn_0: 0.7127, loss_cls_dn_1: 0.0833, loss_box_dn_1: 0.5827, loss_cls_dn_2: 0.0824, loss_box_dn_2: 0.5638, loss_cls_dn_3: 0.0813, loss_box_dn_3: 0.5655, loss_cls_dn_4: 0.0818, loss_box_dn_4: 0.5700, loss_cls_dn_5: 0.0835, loss_box_dn_5: 0.5701, loss_dense_depth: 0.6962, loss: 22.3127, grad_norm: 32.7639 -2026-01-14 21:45:36,510 - mmdet - INFO - Iter [506/17500] lr: 2.994e-04, eta: 9:05:27, time: 1.656, data_time: 0.077, memory: 49164, loss_cls_0: 0.6947, loss_box_0: 1.4971, loss_cns_0: 0.6365, loss_yns_0: 0.1366, loss_cls_1: 0.7673, loss_box_1: 1.3431, loss_cns_1: 0.6617, loss_yns_1: 0.1319, loss_cls_2: 0.7902, loss_box_2: 1.3123, loss_cns_2: 0.6612, loss_yns_2: 0.1314, loss_cls_3: 0.7844, loss_box_3: 1.3188, loss_cns_3: 0.6615, loss_yns_3: 0.1321, loss_cls_4: 0.7812, loss_box_4: 1.3159, loss_cns_4: 0.6619, loss_yns_4: 0.1326, loss_cls_5: 0.7888, loss_box_5: 1.3270, loss_cns_5: 0.6631, loss_yns_5: 0.1321, loss_cls_dn_0: 0.1004, loss_box_dn_0: 0.6995, loss_cls_dn_1: 0.0843, loss_box_dn_1: 0.5946, loss_cls_dn_2: 0.0840, loss_box_dn_2: 0.5745, loss_cls_dn_3: 0.0830, loss_box_dn_3: 0.5742, loss_cls_dn_4: 0.0838, loss_box_dn_4: 0.5727, loss_cls_dn_5: 0.0868, loss_box_dn_5: 0.5806, loss_dense_depth: 0.6671, loss: 22.2491, grad_norm: 26.8797 -2026-01-14 21:45:38,062 - mmdet - INFO - Iter [507/17500] lr: 2.994e-04, eta: 9:05:12, time: 1.553, data_time: 0.075, memory: 49164, loss_cls_0: 0.7222, loss_box_0: 1.4865, loss_cns_0: 0.6309, loss_yns_0: 0.1384, loss_cls_1: 0.7712, loss_box_1: 1.3243, loss_cns_1: 0.6645, loss_yns_1: 0.1352, loss_cls_2: 0.7917, loss_box_2: 1.2882, loss_cns_2: 0.6638, loss_yns_2: 0.1344, loss_cls_3: 0.7933, loss_box_3: 1.2848, loss_cns_3: 0.6644, loss_yns_3: 0.1342, loss_cls_4: 0.7898, loss_box_4: 1.2998, loss_cns_4: 0.6655, loss_yns_4: 0.1353, loss_cls_5: 0.7925, loss_box_5: 1.2976, loss_cns_5: 0.6667, loss_yns_5: 0.1353, loss_cls_dn_0: 0.0968, loss_box_dn_0: 0.7033, loss_cls_dn_1: 0.0857, loss_box_dn_1: 0.5904, loss_cls_dn_2: 0.0845, loss_box_dn_2: 0.5688, loss_cls_dn_3: 0.0864, loss_box_dn_3: 0.5626, loss_cls_dn_4: 0.0843, loss_box_dn_4: 0.5693, loss_cls_dn_5: 0.0852, loss_box_dn_5: 0.5688, loss_dense_depth: 0.7039, loss: 22.2005, grad_norm: 35.7749 -2026-01-14 21:45:39,614 - mmdet - INFO - Iter [508/17500] lr: 2.994e-04, eta: 9:04:58, time: 1.552, data_time: 0.074, memory: 49164, loss_cls_0: 0.6810, loss_box_0: 1.4561, loss_cns_0: 0.6441, loss_yns_0: 0.1368, loss_cls_1: 0.7558, loss_box_1: 1.2829, loss_cns_1: 0.6678, loss_yns_1: 0.1337, loss_cls_2: 0.7735, loss_box_2: 1.2552, loss_cns_2: 0.6666, loss_yns_2: 0.1334, loss_cls_3: 0.7641, loss_box_3: 1.2547, loss_cns_3: 0.6654, loss_yns_3: 0.1338, loss_cls_4: 0.7591, loss_box_4: 1.2547, loss_cns_4: 0.6664, loss_yns_4: 0.1342, loss_cls_5: 0.7572, loss_box_5: 1.2565, loss_cns_5: 0.6643, loss_yns_5: 0.1342, loss_cls_dn_0: 0.0951, loss_box_dn_0: 0.7001, loss_cls_dn_1: 0.0858, loss_box_dn_1: 0.5727, loss_cls_dn_2: 0.0869, loss_box_dn_2: 0.5593, loss_cls_dn_3: 0.0839, loss_box_dn_3: 0.5559, loss_cls_dn_4: 0.0830, loss_box_dn_4: 0.5566, loss_cls_dn_5: 0.0845, loss_box_dn_5: 0.5589, loss_dense_depth: 0.6514, loss: 21.7056, grad_norm: 27.2036 -2026-01-14 21:45:41,232 - mmdet - INFO - Iter [509/17500] lr: 2.994e-04, eta: 9:04:45, time: 1.591, data_time: 0.071, memory: 49164, loss_cls_0: 0.7298, loss_box_0: 1.5006, loss_cns_0: 0.6352, loss_yns_0: 0.1420, loss_cls_1: 0.7585, loss_box_1: 1.3716, loss_cns_1: 0.6615, loss_yns_1: 0.1394, loss_cls_2: 0.7829, loss_box_2: 1.3504, loss_cns_2: 0.6660, loss_yns_2: 0.1414, loss_cls_3: 0.7723, loss_box_3: 1.3217, loss_cns_3: 0.6644, loss_yns_3: 0.1381, loss_cls_4: 0.7713, loss_box_4: 1.3183, loss_cns_4: 0.6632, loss_yns_4: 0.1382, loss_cls_5: 0.7772, loss_box_5: 1.3131, loss_cns_5: 0.6644, loss_yns_5: 0.1384, loss_cls_dn_0: 0.0982, loss_box_dn_0: 0.6963, loss_cls_dn_1: 0.0851, loss_box_dn_1: 0.5947, loss_cls_dn_2: 0.0907, loss_box_dn_2: 0.5787, loss_cls_dn_3: 0.0840, loss_box_dn_3: 0.5703, loss_cls_dn_4: 0.0863, loss_box_dn_4: 0.5664, loss_cls_dn_5: 0.0875, loss_box_dn_5: 0.5677, loss_dense_depth: 0.7104, loss: 22.3762, grad_norm: 43.7046 diff --git a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20260114_212907.log.json b/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20260114_212907.log.json deleted file mode 100644 index 7263834050990a0c1085e4fdc3266bb8f680f945..0000000000000000000000000000000000000000 --- a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20260114_212907.log.json +++ /dev/null @@ -1,510 +0,0 @@ -{"env_info": "sys.platform: linux\nPython: 3.10.12 (main, Aug 15 2025, 14:32:43) [GCC 11.4.0]\nCUDA available: True\nGPU 0,1,2,3,4,5,6,7: BW151\nCUDA_HOME: /opt/dtk\nNVCC: Not Available\nGCC: x86_64-linux-gnu-gcc (Ubuntu 11.4.0-1ubuntu1~22.04) 11.4.0\nPyTorch: 2.5.1\nPyTorch compiling details: PyTorch built with:\n - GCC 10.3\n - C++ Version: 201703\n - Intel(R) Math Kernel Library Version 2020.0.4 Product Build 20200917 for Intel(R) 64 architecture applications\n - OpenMP 201511 (a.k.a. OpenMP 4.5)\n - LAPACK is enabled (usually provided by MKL)\n - NNPACK is enabled\n - CPU capability usage: AVX512\n - HIP Runtime 6.3.25521\n - MIOpen 2.18.0\n - Magma 2.8.0\n - Build settings: BLAS_INFO=mkl, BUILD_TYPE=Release, CXX_COMPILER=/opt/rh/gcc-toolset-10/root/usr/bin/c++, CXX_FLAGS= -D_GLIBCXX_USE_CXX11_ABI=1 -fvisibility-inlines-hidden -DUSE_PTHREADPOOL -DNDEBUG -DUSE_KINETO -DLIBKINETO_NOCUPTI -DLIBKINETO_NOXPUPTI=ON -DUSE_FBGEMM -DUSE_PYTORCH_QNNPACK -DUSE_XNNPACK -DSYMBOLICATE_MOBILE_DEBUG_HANDLE -O2 -fPIC -Wall -Wextra -Werror=return-type -Werror=non-virtual-dtor -Werror=bool-operation -Wnarrowing -Wno-missing-field-initializers -Wno-type-limits -Wno-array-bounds -Wno-unknown-pragmas -Wno-unused-parameter -Wno-strict-overflow -Wno-strict-aliasing -Wno-stringop-overflow -Wsuggest-override -Wno-psabi -Wno-error=old-style-cast -Wno-missing-braces -fdiagnostics-color=always -faligned-new -Wno-unused-but-set-variable -Wno-maybe-uninitialized -fno-math-errno -fno-trapping-math -Werror=format -Wno-stringop-overflow, FORCE_FALLBACK_CUDA_MPI=1, LAPACK_INFO=mkl, PERF_WITH_AVX=1, PERF_WITH_AVX2=1, TORCH_VERSION=2.5.1, USE_CUDA=0, USE_CUDNN=OFF, USE_CUSPARSELT=OFF, USE_EXCEPTION_PTR=1, USE_GFLAGS=1, USE_GLOG=1, USE_GLOO=1, USE_MKL=ON, USE_MKLDNN=0, USE_MPI=1, USE_NCCL=1, USE_NNPACK=ON, USE_OPENMP=1, USE_ROCM=ON, USE_ROCM_KERNEL_ASSERT=OFF, \n\nTorchVision: 0.20.1\nOpenCV: 4.12.0\nMMCV: 1.6.1\nMMCV Compiler: GCC 10.3\nMMCV CUDA Compiler: rocm not available\nMMDetection: 2.25.1+c41df4b", "config": "plugin = True\nplugin_dir = 'projects/mmdet3d_plugin/'\ndist_params = dict(backend='nccl')\nlog_level = 'INFO'\nwork_dir = './work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704'\ntotal_batch_size = 160\nnum_gpus = 8\nbatch_size = 20\nnum_iters_per_epoch = 175\nnum_epochs = 100\ncheckpoint_epoch_interval = 20\ncheckpoint_config = dict(interval=3500)\nlog_config = dict(\n interval=1,\n hooks=[\n dict(type='TextLoggerHook', by_epoch=False),\n dict(type='TensorboardLoggerHook')\n ])\nload_from = None\nresume_from = None\nworkflow = [('train', 1)]\nfp16 = dict(loss_scale=32.0)\ninput_shape = (704, 256)\ntracking_test = True\ntracking_threshold = 0.2\nclass_names = [\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier',\n 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n]\nnum_classes = 10\nembed_dims = 256\nnum_groups = 8\nnum_decoder = 6\nnum_single_frame_decoder = 1\nuse_deformable_func = True\nstrides = [4, 8, 16, 32]\nnum_levels = 4\nnum_depth_layers = 3\ndrop_out = 0.1\ntemporal = True\ndecouple_attn = True\nwith_quality_estimation = True\nmodel = dict(\n type='Sparse4D',\n use_grid_mask=True,\n use_deformable_func=True,\n img_backbone=dict(\n type='ResNet',\n depth=50,\n num_stages=4,\n frozen_stages=-1,\n norm_eval=False,\n style='pytorch',\n with_cp=True,\n out_indices=(0, 1, 2, 3),\n norm_cfg=dict(type='BN', requires_grad=True),\n pretrained='ckpt/resnet50-19c8e357.pth'),\n img_neck=dict(\n type='FPN',\n num_outs=4,\n start_level=0,\n out_channels=256,\n add_extra_convs='on_output',\n relu_before_extra_convs=True,\n in_channels=[256, 512, 1024, 2048]),\n depth_branch=dict(\n type='DenseDepthNet',\n embed_dims=256,\n num_depth_layers=3,\n loss_weight=0.2),\n head=dict(\n type='Sparse4DHead',\n cls_threshold_to_reg=0.05,\n decouple_attn=True,\n instance_bank=dict(\n type='InstanceBank',\n num_anchor=900,\n embed_dims=256,\n anchor='nuscenes_kmeans900.npy',\n anchor_handler=dict(type='SparseBox3DKeyPointsGenerator'),\n num_temp_instances=600,\n confidence_decay=0.6,\n feat_grad=False),\n anchor_encoder=dict(\n type='SparseBox3DEncoder',\n vel_dims=3,\n embed_dims=[128, 32, 32, 64],\n mode='cat',\n output_fc=False,\n in_loops=1,\n out_loops=4),\n num_single_frame_decoder=1,\n operation_order=[\n 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm',\n 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm',\n 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm',\n 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm',\n 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm',\n 'deformable', 'ffn', 'norm', 'refine'\n ],\n temp_graph_model=dict(\n type='MultiheadAttention',\n embed_dims=512,\n num_heads=8,\n batch_first=True,\n dropout=0.1),\n graph_model=dict(\n type='MultiheadAttention',\n embed_dims=512,\n num_heads=8,\n batch_first=True,\n dropout=0.1),\n norm_layer=dict(type='LN', normalized_shape=256),\n ffn=dict(\n type='AsymmetricFFN',\n in_channels=512,\n pre_norm=dict(type='LN'),\n embed_dims=256,\n feedforward_channels=1024,\n num_fcs=2,\n ffn_drop=0.1,\n act_cfg=dict(type='ReLU', inplace=True)),\n deformable_model=dict(\n type='DeformableFeatureAggregation',\n embed_dims=256,\n num_groups=8,\n num_levels=4,\n num_cams=6,\n attn_drop=0.15,\n use_deformable_func=True,\n use_camera_embed=True,\n residual_mode='cat',\n kps_generator=dict(\n type='SparseBox3DKeyPointsGenerator',\n num_learnable_pts=6,\n fix_scale=[[0, 0, 0], [0.45, 0, 0], [-0.45, 0, 0],\n [0, 0.45, 0], [0, -0.45, 0], [0, 0, 0.45],\n [0, 0, -0.45]])),\n refine_layer=dict(\n type='SparseBox3DRefinementModule',\n embed_dims=256,\n num_cls=10,\n refine_yaw=True,\n with_quality_estimation=True),\n sampler=dict(\n type='SparseBox3DTarget',\n num_dn_groups=5,\n num_temp_dn_groups=3,\n dn_noise_scale=[2.0, 2.0, 2.0, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5],\n max_dn_gt=32,\n add_neg_dn=True,\n cls_weight=2.0,\n box_weight=0.25,\n reg_weights=[2.0, 2.0, 2.0, 0.5, 0.5, 0.5, 0.0, 0.0, 0.0, 0.0],\n cls_wise_reg_weights=dict(\n {9: [2.0, 2.0, 2.0, 1.0, 1.0, 1.0, 0.0, 0.0, 1.0, 1.0]})),\n loss_cls=dict(\n type='FocalLoss',\n use_sigmoid=True,\n gamma=2.0,\n alpha=0.25,\n loss_weight=2.0),\n loss_reg=dict(\n type='SparseBox3DLoss',\n loss_box=dict(type='L1Loss', loss_weight=0.25),\n loss_centerness=dict(type='CrossEntropyLoss', use_sigmoid=True),\n loss_yawness=dict(type='GaussianFocalLoss'),\n cls_allow_reverse=[5]),\n decoder=dict(type='SparseBox3DDecoder'),\n reg_weights=[2.0, 2.0, 2.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0]))\ndataset_type = 'NuScenes3DDetTrackDataset'\ndata_root = 'data/nuscenes/'\nanno_root = 'data/nuscenes_anno_pkls/'\nfile_client_args = dict(backend='disk')\nimg_norm_cfg = dict(\n mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375], to_rgb=True)\ntrain_pipeline = [\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(\n type='LoadPointsFromFile',\n coord_type='LIDAR',\n load_dim=5,\n use_dim=5,\n file_client_args=dict(backend='disk')),\n dict(type='ResizeCropFlipImage'),\n dict(type='MultiScaleDepthMapGenerator', downsample=[4, 8, 16]),\n dict(type='BBoxRotation'),\n dict(type='PhotoMetricDistortionMultiViewImage'),\n dict(\n type='NormalizeMultiviewImage',\n mean=[123.675, 116.28, 103.53],\n std=[58.395, 57.12, 57.375],\n to_rgb=True),\n dict(\n type='CircleObjectRangeFilter',\n class_dist_thred=[55, 55, 55, 55, 55, 55, 55, 55, 55, 55]),\n dict(\n type='InstanceNameFilter',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ]),\n dict(type='NuScenesSparse4DAdaptor'),\n dict(\n type='Collect',\n keys=[\n 'img', 'timestamp', 'projection_mat', 'image_wh', 'gt_depth',\n 'focal', 'gt_bboxes_3d', 'gt_labels_3d'\n ],\n meta_keys=['T_global', 'T_global_inv', 'timestamp', 'instance_id'])\n]\ntest_pipeline = [\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(type='ResizeCropFlipImage'),\n dict(\n type='NormalizeMultiviewImage',\n mean=[123.675, 116.28, 103.53],\n std=[58.395, 57.12, 57.375],\n to_rgb=True),\n dict(type='NuScenesSparse4DAdaptor'),\n dict(\n type='Collect',\n keys=['img', 'timestamp', 'projection_mat', 'image_wh'],\n meta_keys=['T_global', 'T_global_inv', 'timestamp'])\n]\ninput_modality = dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False)\ndata_basic_config = dict(\n type='NuScenes3DDetTrackDataset',\n data_root='data/nuscenes/',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier',\n 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ],\n modality=dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False),\n version='v1.0-trainval')\ndata_aug_conf = dict(\n resize_lim=(0.4, 0.47),\n final_dim=(256, 704),\n bot_pct_lim=(0.0, 0.0),\n rot_lim=(-5.4, 5.4),\n H=900,\n W=1600,\n rand_flip=True,\n rot3d_range=[-0.3925, 0.3925])\ndata = dict(\n samples_per_gpu=20,\n workers_per_gpu=20,\n train=dict(\n type='NuScenes3DDetTrackDataset',\n data_root='data/nuscenes/',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ],\n modality=dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False),\n version='v1.0-trainval',\n ann_file='data/nuscenes_anno_pkls/nuscenes_infos_train.pkl',\n pipeline=[\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(\n type='LoadPointsFromFile',\n coord_type='LIDAR',\n load_dim=5,\n use_dim=5,\n file_client_args=dict(backend='disk')),\n dict(type='ResizeCropFlipImage'),\n dict(type='MultiScaleDepthMapGenerator', downsample=[4, 8, 16]),\n dict(type='BBoxRotation'),\n dict(type='PhotoMetricDistortionMultiViewImage'),\n dict(\n type='NormalizeMultiviewImage',\n mean=[123.675, 116.28, 103.53],\n std=[58.395, 57.12, 57.375],\n to_rgb=True),\n dict(\n type='CircleObjectRangeFilter',\n class_dist_thred=[55, 55, 55, 55, 55, 55, 55, 55, 55, 55]),\n dict(\n type='InstanceNameFilter',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian',\n 'traffic_cone'\n ]),\n dict(type='NuScenesSparse4DAdaptor'),\n dict(\n type='Collect',\n keys=[\n 'img', 'timestamp', 'projection_mat', 'image_wh',\n 'gt_depth', 'focal', 'gt_bboxes_3d', 'gt_labels_3d'\n ],\n meta_keys=[\n 'T_global', 'T_global_inv', 'timestamp', 'instance_id'\n ])\n ],\n test_mode=False,\n data_aug_conf=dict(\n resize_lim=(0.4, 0.47),\n final_dim=(256, 704),\n bot_pct_lim=(0.0, 0.0),\n rot_lim=(-5.4, 5.4),\n H=900,\n W=1600,\n rand_flip=True,\n rot3d_range=[-0.3925, 0.3925]),\n with_seq_flag=True,\n sequences_split_num=2,\n keep_consistent_seq_aug=True),\n val=dict(\n type='NuScenes3DDetTrackDataset',\n data_root='data/nuscenes/',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ],\n modality=dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False),\n version='v1.0-trainval',\n ann_file='data/nuscenes_anno_pkls/nuscenes_infos_val.pkl',\n pipeline=[\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(type='ResizeCropFlipImage'),\n dict(\n type='NormalizeMultiviewImage',\n mean=[123.675, 116.28, 103.53],\n std=[58.395, 57.12, 57.375],\n to_rgb=True),\n dict(type='NuScenesSparse4DAdaptor'),\n dict(\n type='Collect',\n keys=['img', 'timestamp', 'projection_mat', 'image_wh'],\n meta_keys=['T_global', 'T_global_inv', 'timestamp'])\n ],\n data_aug_conf=dict(\n resize_lim=(0.4, 0.47),\n final_dim=(256, 704),\n bot_pct_lim=(0.0, 0.0),\n rot_lim=(-5.4, 5.4),\n H=900,\n W=1600,\n rand_flip=True,\n rot3d_range=[-0.3925, 0.3925]),\n test_mode=True,\n tracking=True,\n tracking_threshold=0.2),\n test=dict(\n type='NuScenes3DDetTrackDataset',\n data_root='data/nuscenes/',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ],\n modality=dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False),\n version='v1.0-trainval',\n ann_file='data/nuscenes_anno_pkls/nuscenes_infos_val.pkl',\n pipeline=[\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(type='ResizeCropFlipImage'),\n dict(\n type='NormalizeMultiviewImage',\n mean=[123.675, 116.28, 103.53],\n std=[58.395, 57.12, 57.375],\n to_rgb=True),\n dict(type='NuScenesSparse4DAdaptor'),\n dict(\n type='Collect',\n keys=['img', 'timestamp', 'projection_mat', 'image_wh'],\n meta_keys=['T_global', 'T_global_inv', 'timestamp'])\n ],\n data_aug_conf=dict(\n resize_lim=(0.4, 0.47),\n final_dim=(256, 704),\n bot_pct_lim=(0.0, 0.0),\n rot_lim=(-5.4, 5.4),\n H=900,\n W=1600,\n rand_flip=True,\n rot3d_range=[-0.3925, 0.3925]),\n test_mode=True,\n tracking=True,\n tracking_threshold=0.2))\noptimizer = dict(\n type='AdamW',\n lr=0.0006,\n weight_decay=0.001,\n paramwise_cfg=dict(custom_keys=dict(img_backbone=dict(lr_mult=0.5))))\noptimizer_config = dict(grad_clip=dict(max_norm=25, norm_type=2))\nlr_config = dict(\n policy='CosineAnnealing',\n warmup='linear',\n warmup_iters=500,\n warmup_ratio=0.3333333333333333,\n min_lr_ratio=0.001)\nrunner = dict(type='IterBasedRunner', max_iters=17500)\nvis_pipeline = [\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(type='Collect', keys=['img'], meta_keys=['timestamp', 'lidar2img'])\n]\nevaluation = dict(\n interval=3500,\n pipeline=[\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(\n type='Collect', keys=['img'], meta_keys=['timestamp', 'lidar2img'])\n ])\ngpu_ids = range(0, 8)\n", "seed": 0, "exp_name": "sparse4dv3_temporal_r50_1x8_bs6_256x704.py"} -{"mode": "train", "epoch": 1, "iter": 1, "lr": 0.0001, "memory": 49164, "data_time": 17.20136, "loss_cls_0": 2.36133, "loss_box_0": 0.01384, "loss_cns_0": 0.0027, "loss_yns_0": 0.00079, "loss_cls_1": 2.15443, "loss_box_1": 0.10813, "loss_cns_1": 0.02455, "loss_yns_1": 0.00669, "loss_cls_2": 2.31207, "loss_box_2": 0.00503, "loss_cns_2": 0.00059, "loss_yns_2": 0.00029, "loss_cls_3": 2.39024, "loss_box_3": 0.02945, "loss_cns_3": 0.00504, "loss_yns_3": 0.00144, "loss_cls_4": 2.02793, "loss_box_4": 0.41487, "loss_cns_4": 0.05322, "loss_yns_4": 0.02495, "loss_cls_5": 2.42477, "loss_box_5": 0.01805, "loss_cns_5": 0.00217, "loss_yns_5": 0.0016, "loss_cls_dn_0": 1.19802, "loss_box_dn_0": 1.46031, "loss_cls_dn_1": 1.11018, "loss_box_dn_1": 1.73179, "loss_cls_dn_2": 1.17414, "loss_box_dn_2": 1.97188, "loss_cls_dn_3": 1.17205, "loss_box_dn_3": 2.24182, "loss_cls_dn_4": 1.05275, "loss_box_dn_4": 2.42684, "loss_cls_dn_5": 1.23866, "loss_box_dn_5": 2.67737, "loss_dense_depth": 1.86433, "loss": 35.70432, "grad_norm": 269.84476, "time": 130.3027} -{"mode": "train", "epoch": 1, "iter": 2, "lr": 0.0001, "memory": 49164, "data_time": 0.15706, "loss_cls_0": 2.04337, "loss_box_0": 0.01374, "loss_cns_0": 0.00385, "loss_yns_0": 0.00147, "loss_cls_1": 2.00953, "loss_box_1": 0.152, "loss_cns_1": 0.02542, "loss_yns_1": 0.00787, "loss_cls_2": 2.09647, "loss_box_2": 0.2799, "loss_cns_2": 0.02631, "loss_yns_2": 0.01078, "loss_cls_3": 1.94058, "loss_box_3": 0.39954, "loss_cns_3": 0.05314, "loss_yns_3": 0.01572, "loss_cls_4": 1.80239, "loss_box_4": 1.67637, "loss_cns_4": 0.17048, "loss_yns_4": 0.06058, "loss_cls_5": 2.05539, "loss_box_5": 0.56073, "loss_cns_5": 0.06218, "loss_yns_5": 0.01938, "loss_cls_dn_0": 1.02558, "loss_box_dn_0": 1.25576, "loss_cls_dn_1": 0.95205, "loss_box_dn_1": 2.41686, "loss_cls_dn_2": 0.96717, "loss_box_dn_2": 2.533, "loss_cls_dn_3": 0.90833, "loss_box_dn_3": 2.61748, "loss_cls_dn_4": 0.8406, "loss_box_dn_4": 2.87439, "loss_cls_dn_5": 0.98697, "loss_box_dn_5": 3.11138, "loss_dense_depth": 1.70921, "loss": 37.68597, "grad_norm": 66.91905, "time": 2.55081} -{"mode": "train", "epoch": 1, "iter": 3, "lr": 0.0001, "memory": 49164, "data_time": 0.07358, "loss_cls_0": 1.47378, "loss_box_0": 2.61702, "loss_cns_0": 0.61286, "loss_yns_0": 0.22472, "loss_cls_1": 1.76293, "loss_box_1": 2.02634, "loss_cns_1": 0.31065, "loss_yns_1": 0.11521, "loss_cls_2": 1.78436, "loss_box_2": 4.05639, "loss_cns_2": 0.35587, "loss_yns_2": 0.19582, "loss_cls_3": 1.6199, "loss_box_3": 4.81505, "loss_cns_3": 0.43995, "loss_yns_3": 0.20405, "loss_cls_4": 1.57762, "loss_box_4": 4.00888, "loss_cns_4": 0.37023, "loss_yns_4": 0.15781, "loss_cls_5": 1.69558, "loss_box_5": 2.77383, "loss_cns_5": 0.216, "loss_yns_5": 0.09196, "loss_cls_dn_0": 0.71422, "loss_box_dn_0": 1.22807, "loss_cls_dn_1": 0.82908, "loss_box_dn_1": 2.41439, "loss_cls_dn_2": 0.80388, "loss_box_dn_2": 2.62496, "loss_cls_dn_3": 0.71237, "loss_box_dn_3": 2.82196, "loss_cls_dn_4": 0.72616, "loss_box_dn_4": 3.08739, "loss_cls_dn_5": 0.80891, "loss_box_dn_5": 3.32263, "loss_dense_depth": 1.68392, "loss": 55.28474, "grad_norm": 102.59567, "time": 1.5659} -{"mode": "train", "epoch": 1, "iter": 4, "lr": 0.0001, "memory": 49164, "data_time": 0.09574, "loss_cls_0": 1.38055, "loss_box_0": 2.46066, "loss_cns_0": 0.57603, "loss_yns_0": 0.1829, "loss_cls_1": 1.59475, "loss_box_1": 3.06909, "loss_cns_1": 0.44903, "loss_yns_1": 0.18038, "loss_cls_2": 1.7143, "loss_box_2": 3.75279, "loss_cns_2": 0.43406, "loss_yns_2": 0.18996, "loss_cls_3": 1.51981, "loss_box_3": 4.30994, "loss_cns_3": 0.43883, "loss_yns_3": 0.21144, "loss_cls_4": 1.47135, "loss_box_4": 4.85956, "loss_cns_4": 0.35492, "loss_yns_4": 0.19872, "loss_cls_5": 1.50276, "loss_box_5": 4.98858, "loss_cns_5": 0.4388, "loss_yns_5": 0.19012, "loss_cls_dn_0": 0.57975, "loss_box_dn_0": 1.16738, "loss_cls_dn_1": 0.72013, "loss_box_dn_1": 2.5675, "loss_cls_dn_2": 0.69489, "loss_box_dn_2": 2.63951, "loss_cls_dn_3": 0.61602, "loss_box_dn_3": 2.84627, "loss_cls_dn_4": 0.59921, "loss_box_dn_4": 3.04733, "loss_cls_dn_5": 0.6737, "loss_box_dn_5": 3.21123, "loss_dense_depth": 1.60879, "loss": 57.44099, "grad_norm": 119.51874, "time": 1.65372} -{"mode": "train", "epoch": 1, "iter": 5, "lr": 0.0001, "memory": 49164, "data_time": 0.0716, "loss_cls_0": 1.34592, "loss_box_0": 2.75348, "loss_cns_0": 0.50523, "loss_yns_0": 0.1956, "loss_cls_1": 1.54383, "loss_box_1": 3.8496, "loss_cns_1": 0.40229, "loss_yns_1": 0.20256, "loss_cls_2": 1.61065, "loss_box_2": 3.92773, "loss_cns_2": 0.38684, "loss_yns_2": 0.19487, "loss_cls_3": 1.46221, "loss_box_3": 4.00547, "loss_cns_3": 0.40005, "loss_yns_3": 0.19409, "loss_cls_4": 1.39779, "loss_box_4": 4.19753, "loss_cns_4": 0.3837, "loss_yns_4": 0.19411, "loss_cls_5": 1.39322, "loss_box_5": 4.42998, "loss_cns_5": 0.40929, "loss_yns_5": 0.19888, "loss_cls_dn_0": 0.53038, "loss_box_dn_0": 1.21644, "loss_cls_dn_1": 0.65218, "loss_box_dn_1": 2.19462, "loss_cls_dn_2": 0.65184, "loss_box_dn_2": 2.2967, "loss_cls_dn_3": 0.56125, "loss_box_dn_3": 2.41797, "loss_cls_dn_4": 0.55643, "loss_box_dn_4": 2.5781, "loss_cls_dn_5": 0.58666, "loss_box_dn_5": 2.65841, "loss_dense_depth": 1.58112, "loss": 54.06702, "grad_norm": 109.16457, "time": 1.72938} -{"mode": "train", "epoch": 1, "iter": 6, "lr": 0.0001, "memory": 49164, "data_time": 0.09017, "loss_cls_0": 1.30139, "loss_box_0": 2.55299, "loss_cns_0": 0.56279, "loss_yns_0": 0.18108, "loss_cls_1": 1.47161, "loss_box_1": 3.82076, "loss_cns_1": 0.37468, "loss_yns_1": 0.19062, "loss_cls_2": 1.48044, "loss_box_2": 4.01001, "loss_cns_2": 0.35077, "loss_yns_2": 0.19, "loss_cls_3": 1.35152, "loss_box_3": 4.05365, "loss_cns_3": 0.33217, "loss_yns_3": 0.1937, "loss_cls_4": 1.32228, "loss_box_4": 4.35726, "loss_cns_4": 0.28425, "loss_yns_4": 0.19515, "loss_cls_5": 1.33488, "loss_box_5": 4.51018, "loss_cns_5": 0.28302, "loss_yns_5": 0.19871, "loss_cls_dn_0": 0.53304, "loss_box_dn_0": 1.16037, "loss_cls_dn_1": 0.58639, "loss_box_dn_1": 2.34852, "loss_cls_dn_2": 0.58431, "loss_box_dn_2": 2.40356, "loss_cls_dn_3": 0.5176, "loss_box_dn_3": 2.45883, "loss_cls_dn_4": 0.49228, "loss_box_dn_4": 2.67891, "loss_cls_dn_5": 0.49144, "loss_box_dn_5": 2.74044, "loss_dense_depth": 1.49701, "loss": 53.39661, "grad_norm": 115.96841, "time": 1.57092} -{"mode": "train", "epoch": 1, "iter": 7, "lr": 0.0001, "memory": 49164, "data_time": 0.10771, "loss_cls_0": 1.27865, "loss_box_0": 2.3793, "loss_cns_0": 0.64666, "loss_yns_0": 0.17724, "loss_cls_1": 1.35279, "loss_box_1": 3.56132, "loss_cns_1": 0.44456, "loss_yns_1": 0.1861, "loss_cls_2": 1.38808, "loss_box_2": 3.68434, "loss_cns_2": 0.43982, "loss_yns_2": 0.18121, "loss_cls_3": 1.29887, "loss_box_3": 3.58025, "loss_cns_3": 0.45586, "loss_yns_3": 0.18271, "loss_cls_4": 1.29016, "loss_box_4": 3.91729, "loss_cns_4": 0.40221, "loss_yns_4": 0.18681, "loss_cls_5": 1.32592, "loss_box_5": 4.20602, "loss_cns_5": 0.37408, "loss_yns_5": 0.19061, "loss_cls_dn_0": 0.54307, "loss_box_dn_0": 1.10064, "loss_cls_dn_1": 0.52413, "loss_box_dn_1": 2.39966, "loss_cls_dn_2": 0.52524, "loss_box_dn_2": 2.40077, "loss_cls_dn_3": 0.47607, "loss_box_dn_3": 2.41966, "loss_cls_dn_4": 0.4512, "loss_box_dn_4": 2.62786, "loss_cls_dn_5": 0.43598, "loss_box_dn_5": 2.73039, "loss_dense_depth": 1.43534, "loss": 51.20089, "grad_norm": 105.66734, "time": 1.63374} -{"mode": "train", "epoch": 1, "iter": 8, "lr": 0.0001, "memory": 49164, "data_time": 0.08808, "loss_cls_0": 1.23939, "loss_box_0": 2.2865, "loss_cns_0": 0.64578, "loss_yns_0": 0.18114, "loss_cls_1": 1.27019, "loss_box_1": 3.51166, "loss_cns_1": 0.48624, "loss_yns_1": 0.18493, "loss_cls_2": 1.35684, "loss_box_2": 3.62423, "loss_cns_2": 0.46014, "loss_yns_2": 0.18374, "loss_cls_3": 1.28979, "loss_box_3": 3.5973, "loss_cns_3": 0.48563, "loss_yns_3": 0.18527, "loss_cls_4": 1.29675, "loss_box_4": 3.63554, "loss_cns_4": 0.47908, "loss_yns_4": 0.18401, "loss_cls_5": 1.3312, "loss_box_5": 3.71786, "loss_cns_5": 0.49274, "loss_yns_5": 0.19399, "loss_cls_dn_0": 0.5305, "loss_box_dn_0": 1.03831, "loss_cls_dn_1": 0.53372, "loss_box_dn_1": 1.81931, "loss_cls_dn_2": 0.53946, "loss_box_dn_2": 1.80511, "loss_cls_dn_3": 0.49348, "loss_box_dn_3": 1.80045, "loss_cls_dn_4": 0.46346, "loss_box_dn_4": 1.91388, "loss_cls_dn_5": 0.43663, "loss_box_dn_5": 1.94657, "loss_dense_depth": 1.47358, "loss": 47.11439, "grad_norm": 85.44607, "time": 1.56713} -{"mode": "train", "epoch": 1, "iter": 9, "lr": 0.0001, "memory": 49164, "data_time": 0.07479, "loss_cls_0": 1.24504, "loss_box_0": 2.22516, "loss_cns_0": 0.61838, "loss_yns_0": 0.17782, "loss_cls_1": 1.25569, "loss_box_1": 3.27789, "loss_cns_1": 0.51344, "loss_yns_1": 0.18233, "loss_cls_2": 1.34953, "loss_box_2": 3.37566, "loss_cns_2": 0.46885, "loss_yns_2": 0.18243, "loss_cls_3": 1.26708, "loss_box_3": 3.42487, "loss_cns_3": 0.48901, "loss_yns_3": 0.18893, "loss_cls_4": 1.27229, "loss_box_4": 3.38371, "loss_cns_4": 0.49131, "loss_yns_4": 0.19165, "loss_cls_5": 1.31845, "loss_box_5": 3.47228, "loss_cns_5": 0.50954, "loss_yns_5": 0.18973, "loss_cls_dn_0": 0.49476, "loss_box_dn_0": 1.01989, "loss_cls_dn_1": 0.49423, "loss_box_dn_1": 1.50494, "loss_cls_dn_2": 0.5175, "loss_box_dn_2": 1.52316, "loss_cls_dn_3": 0.47021, "loss_box_dn_3": 1.57006, "loss_cls_dn_4": 0.44289, "loss_box_dn_4": 1.5546, "loss_cls_dn_5": 0.42363, "loss_box_dn_5": 1.6427, "loss_dense_depth": 1.35432, "loss": 44.08396, "grad_norm": 60.75248, "time": 1.57354} -{"mode": "train", "epoch": 1, "iter": 10, "lr": 0.0001, "memory": 49164, "data_time": 0.0766, "loss_cls_0": 1.22298, "loss_box_0": 2.19352, "loss_cns_0": 0.61455, "loss_yns_0": 0.17266, "loss_cls_1": 1.25802, "loss_box_1": 3.10506, "loss_cns_1": 0.5368, "loss_yns_1": 0.17577, "loss_cls_2": 1.27716, "loss_box_2": 3.17223, "loss_cns_2": 0.52855, "loss_yns_2": 0.17844, "loss_cls_3": 1.25736, "loss_box_3": 3.3351, "loss_cns_3": 0.54393, "loss_yns_3": 0.1874, "loss_cls_4": 1.25259, "loss_box_4": 3.27495, "loss_cns_4": 0.55934, "loss_yns_4": 0.18544, "loss_cls_5": 1.27251, "loss_box_5": 3.40161, "loss_cns_5": 0.57154, "loss_yns_5": 0.1817, "loss_cls_dn_0": 0.47689, "loss_box_dn_0": 1.0215, "loss_cls_dn_1": 0.4526, "loss_box_dn_1": 1.61701, "loss_cls_dn_2": 0.47522, "loss_box_dn_2": 1.66904, "loss_cls_dn_3": 0.43485, "loss_box_dn_3": 1.76592, "loss_cls_dn_4": 0.42373, "loss_box_dn_4": 1.76854, "loss_cls_dn_5": 0.4288, "loss_box_dn_5": 1.91826, "loss_dense_depth": 1.38658, "loss": 44.29816, "grad_norm": 74.12585, "time": 1.61124} -{"mode": "train", "epoch": 1, "iter": 11, "lr": 0.0001, "memory": 49164, "data_time": 0.07455, "loss_cls_0": 1.22061, "loss_box_0": 2.2267, "loss_cns_0": 0.62307, "loss_yns_0": 0.17565, "loss_cls_1": 1.25411, "loss_box_1": 3.0219, "loss_cns_1": 0.52454, "loss_yns_1": 0.17777, "loss_cls_2": 1.27239, "loss_box_2": 2.98206, "loss_cns_2": 0.52229, "loss_yns_2": 0.1842, "loss_cls_3": 1.25679, "loss_box_3": 3.04709, "loss_cns_3": 0.5772, "loss_yns_3": 0.17944, "loss_cls_4": 1.23976, "loss_box_4": 2.97869, "loss_cns_4": 0.5843, "loss_yns_4": 0.1801, "loss_cls_5": 1.26497, "loss_box_5": 3.21178, "loss_cns_5": 0.56297, "loss_yns_5": 0.18463, "loss_cls_dn_0": 0.47292, "loss_box_dn_0": 1.03101, "loss_cls_dn_1": 0.4213, "loss_box_dn_1": 1.92761, "loss_cls_dn_2": 0.43963, "loss_box_dn_2": 1.95541, "loss_cls_dn_3": 0.41368, "loss_box_dn_3": 2.03025, "loss_cls_dn_4": 0.40488, "loss_box_dn_4": 2.03854, "loss_cls_dn_5": 0.41825, "loss_box_dn_5": 2.23921, "loss_dense_depth": 1.24933, "loss": 44.49503, "grad_norm": 68.61853, "time": 1.56375} -{"mode": "train", "epoch": 1, "iter": 12, "lr": 0.0001, "memory": 49164, "data_time": 0.0779, "loss_cls_0": 1.21474, "loss_box_0": 2.23133, "loss_cns_0": 0.62606, "loss_yns_0": 0.17446, "loss_cls_1": 1.26727, "loss_box_1": 2.93182, "loss_cns_1": 0.5035, "loss_yns_1": 0.1741, "loss_cls_2": 1.29204, "loss_box_2": 2.90538, "loss_cns_2": 0.50523, "loss_yns_2": 0.18106, "loss_cls_3": 1.24545, "loss_box_3": 2.88037, "loss_cns_3": 0.5407, "loss_yns_3": 0.17812, "loss_cls_4": 1.24604, "loss_box_4": 2.92517, "loss_cns_4": 0.57425, "loss_yns_4": 0.17719, "loss_cls_5": 1.27395, "loss_box_5": 3.17441, "loss_cns_5": 0.54995, "loss_yns_5": 0.18595, "loss_cls_dn_0": 0.47223, "loss_box_dn_0": 1.04141, "loss_cls_dn_1": 0.39, "loss_box_dn_1": 2.24053, "loss_cls_dn_2": 0.41474, "loss_box_dn_2": 2.22435, "loss_cls_dn_3": 0.3989, "loss_box_dn_3": 2.25099, "loss_cls_dn_4": 0.38439, "loss_box_dn_4": 2.28242, "loss_cls_dn_5": 0.39769, "loss_box_dn_5": 2.4183, "loss_dense_depth": 1.32621, "loss": 45.20071, "grad_norm": 68.7727, "time": 1.58528} -{"mode": "train", "epoch": 1, "iter": 13, "lr": 0.0001, "memory": 49164, "data_time": 0.07273, "loss_cls_0": 1.1994, "loss_box_0": 2.25056, "loss_cns_0": 0.60868, "loss_yns_0": 0.17118, "loss_cls_1": 1.25894, "loss_box_1": 2.80589, "loss_cns_1": 0.53132, "loss_yns_1": 0.1738, "loss_cls_2": 1.275, "loss_box_2": 2.87791, "loss_cns_2": 0.53542, "loss_yns_2": 0.18211, "loss_cls_3": 1.2379, "loss_box_3": 2.84072, "loss_cns_3": 0.54125, "loss_yns_3": 0.17774, "loss_cls_4": 1.26389, "loss_box_4": 2.87354, "loss_cns_4": 0.54502, "loss_yns_4": 0.17704, "loss_cls_5": 1.27796, "loss_box_5": 2.93827, "loss_cns_5": 0.53623, "loss_yns_5": 0.18041, "loss_cls_dn_0": 0.46209, "loss_box_dn_0": 1.04612, "loss_cls_dn_1": 0.4212, "loss_box_dn_1": 1.62294, "loss_cls_dn_2": 0.44906, "loss_box_dn_2": 1.62711, "loss_cls_dn_3": 0.44854, "loss_box_dn_3": 1.67883, "loss_cls_dn_4": 0.43036, "loss_box_dn_4": 1.78735, "loss_cls_dn_5": 0.44489, "loss_box_dn_5": 1.79487, "loss_dense_depth": 1.19481, "loss": 41.86834, "grad_norm": 60.0293, "time": 1.57692} -{"mode": "train", "epoch": 1, "iter": 14, "lr": 0.00011, "memory": 49164, "data_time": 0.07341, "loss_cls_0": 1.2196, "loss_box_0": 2.25181, "loss_cns_0": 0.59385, "loss_yns_0": 0.17195, "loss_cls_1": 1.27328, "loss_box_1": 2.71875, "loss_cns_1": 0.56001, "loss_yns_1": 0.17204, "loss_cls_2": 1.26928, "loss_box_2": 2.82994, "loss_cns_2": 0.56008, "loss_yns_2": 0.18423, "loss_cls_3": 1.26599, "loss_box_3": 2.88017, "loss_cns_3": 0.55778, "loss_yns_3": 0.17302, "loss_cls_4": 1.28892, "loss_box_4": 2.88976, "loss_cns_4": 0.54126, "loss_yns_4": 0.17413, "loss_cls_5": 1.28597, "loss_box_5": 2.8781, "loss_cns_5": 0.55309, "loss_yns_5": 0.17839, "loss_cls_dn_0": 0.45488, "loss_box_dn_0": 1.03334, "loss_cls_dn_1": 0.4436, "loss_box_dn_1": 1.25401, "loss_cls_dn_2": 0.47116, "loss_box_dn_2": 1.277, "loss_cls_dn_3": 0.46248, "loss_box_dn_3": 1.36715, "loss_cls_dn_4": 0.45176, "loss_box_dn_4": 1.48688, "loss_cls_dn_5": 0.46859, "loss_box_dn_5": 1.43364, "loss_dense_depth": 1.19767, "loss": 40.27356, "grad_norm": 78.83748, "time": 1.58531} -{"mode": "train", "epoch": 1, "iter": 15, "lr": 0.00011, "memory": 49164, "data_time": 0.07475, "loss_cls_0": 1.21982, "loss_box_0": 2.28812, "loss_cns_0": 0.58686, "loss_yns_0": 0.17224, "loss_cls_1": 1.30018, "loss_box_1": 2.55921, "loss_cns_1": 0.57829, "loss_yns_1": 0.17241, "loss_cls_2": 1.26333, "loss_box_2": 2.59714, "loss_cns_2": 0.58263, "loss_yns_2": 0.17954, "loss_cls_3": 1.29385, "loss_box_3": 2.68011, "loss_cns_3": 0.59225, "loss_yns_3": 0.17441, "loss_cls_4": 1.29218, "loss_box_4": 2.74791, "loss_cns_4": 0.58101, "loss_yns_4": 0.18052, "loss_cls_5": 1.28395, "loss_box_5": 2.8429, "loss_cns_5": 0.58832, "loss_yns_5": 0.18077, "loss_cls_dn_0": 0.45457, "loss_box_dn_0": 1.02783, "loss_cls_dn_1": 0.44984, "loss_box_dn_1": 1.31209, "loss_cls_dn_2": 0.46303, "loss_box_dn_2": 1.3109, "loss_cls_dn_3": 0.44072, "loss_box_dn_3": 1.391, "loss_cls_dn_4": 0.44633, "loss_box_dn_4": 1.48902, "loss_cls_dn_5": 0.46744, "loss_box_dn_5": 1.49656, "loss_dense_depth": 1.20366, "loss": 39.89094, "grad_norm": 79.15927, "time": 1.57395} -{"mode": "train", "epoch": 1, "iter": 16, "lr": 0.00011, "memory": 49164, "data_time": 0.11106, "loss_cls_0": 1.1952, "loss_box_0": 2.30485, "loss_cns_0": 0.58407, "loss_yns_0": 0.17402, "loss_cls_1": 1.29839, "loss_box_1": 2.63863, "loss_cns_1": 0.56644, "loss_yns_1": 0.17311, "loss_cls_2": 1.27086, "loss_box_2": 2.67624, "loss_cns_2": 0.56746, "loss_yns_2": 0.17976, "loss_cls_3": 1.30932, "loss_box_3": 2.66663, "loss_cns_3": 0.58402, "loss_yns_3": 0.17741, "loss_cls_4": 1.25897, "loss_box_4": 2.74029, "loss_cns_4": 0.58018, "loss_yns_4": 0.18049, "loss_cls_5": 1.26832, "loss_box_5": 2.85089, "loss_cns_5": 0.59832, "loss_yns_5": 0.17709, "loss_cls_dn_0": 0.47747, "loss_box_dn_0": 1.01098, "loss_cls_dn_1": 0.46509, "loss_box_dn_1": 1.30908, "loss_cls_dn_2": 0.47065, "loss_box_dn_2": 1.28319, "loss_cls_dn_3": 0.43494, "loss_box_dn_3": 1.33747, "loss_cls_dn_4": 0.45834, "loss_box_dn_4": 1.39933, "loss_cls_dn_5": 0.47679, "loss_box_dn_5": 1.44099, "loss_dense_depth": 1.15341, "loss": 39.73867, "grad_norm": 69.17771, "time": 1.60044} -{"mode": "train", "epoch": 1, "iter": 17, "lr": 0.00011, "memory": 49164, "data_time": 0.07376, "loss_cls_0": 1.17519, "loss_box_0": 2.29551, "loss_cns_0": 0.594, "loss_yns_0": 0.17127, "loss_cls_1": 1.27521, "loss_box_1": 2.84967, "loss_cns_1": 0.54928, "loss_yns_1": 0.17597, "loss_cls_2": 1.26878, "loss_box_2": 2.91232, "loss_cns_2": 0.54701, "loss_yns_2": 0.17471, "loss_cls_3": 1.29238, "loss_box_3": 2.89361, "loss_cns_3": 0.56641, "loss_yns_3": 0.17557, "loss_cls_4": 1.25036, "loss_box_4": 2.89998, "loss_cns_4": 0.57456, "loss_yns_4": 0.17736, "loss_cls_5": 1.26708, "loss_box_5": 2.85104, "loss_cns_5": 0.59126, "loss_yns_5": 0.17412, "loss_cls_dn_0": 0.48181, "loss_box_dn_0": 1.00883, "loss_cls_dn_1": 0.45793, "loss_box_dn_1": 1.40781, "loss_cls_dn_2": 0.45539, "loss_box_dn_2": 1.40427, "loss_cls_dn_3": 0.42121, "loss_box_dn_3": 1.44652, "loss_cls_dn_4": 0.44364, "loss_box_dn_4": 1.47298, "loss_cls_dn_5": 0.45743, "loss_box_dn_5": 1.50477, "loss_dense_depth": 1.16668, "loss": 40.83192, "grad_norm": 67.31412, "time": 1.63585} -{"mode": "train", "epoch": 1, "iter": 18, "lr": 0.00011, "memory": 49164, "data_time": 0.10692, "loss_cls_0": 1.17496, "loss_box_0": 2.24021, "loss_cns_0": 0.61449, "loss_yns_0": 0.17102, "loss_cls_1": 1.24952, "loss_box_1": 2.84983, "loss_cns_1": 0.55003, "loss_yns_1": 0.18022, "loss_cls_2": 1.26057, "loss_box_2": 2.89873, "loss_cns_2": 0.5515, "loss_yns_2": 0.17412, "loss_cls_3": 1.25555, "loss_box_3": 2.9022, "loss_cns_3": 0.55779, "loss_yns_3": 0.17509, "loss_cls_4": 1.24885, "loss_box_4": 2.90317, "loss_cns_4": 0.57489, "loss_yns_4": 0.17244, "loss_cls_5": 1.26306, "loss_box_5": 2.9313, "loss_cns_5": 0.56884, "loss_yns_5": 0.1752, "loss_cls_dn_0": 0.47878, "loss_box_dn_0": 1.00077, "loss_cls_dn_1": 0.44871, "loss_box_dn_1": 1.41584, "loss_cls_dn_2": 0.44781, "loss_box_dn_2": 1.45794, "loss_cls_dn_3": 0.42354, "loss_box_dn_3": 1.51234, "loss_cls_dn_4": 0.43364, "loss_box_dn_4": 1.52004, "loss_cls_dn_5": 0.44882, "loss_box_dn_5": 1.62264, "loss_dense_depth": 1.05561, "loss": 40.91002, "grad_norm": 69.54926, "time": 1.61671} -{"mode": "train", "epoch": 1, "iter": 19, "lr": 0.00011, "memory": 49164, "data_time": 0.07691, "loss_cls_0": 1.18684, "loss_box_0": 2.24779, "loss_cns_0": 0.61131, "loss_yns_0": 0.17152, "loss_cls_1": 1.23998, "loss_box_1": 2.83184, "loss_cns_1": 0.55555, "loss_yns_1": 0.17694, "loss_cls_2": 1.26104, "loss_box_2": 2.86238, "loss_cns_2": 0.55122, "loss_yns_2": 0.1767, "loss_cls_3": 1.25098, "loss_box_3": 2.86796, "loss_cns_3": 0.54581, "loss_yns_3": 0.17373, "loss_cls_4": 1.2616, "loss_box_4": 2.86387, "loss_cns_4": 0.57322, "loss_yns_4": 0.17372, "loss_cls_5": 1.26501, "loss_box_5": 2.92865, "loss_cns_5": 0.54658, "loss_yns_5": 0.17491, "loss_cls_dn_0": 0.47716, "loss_box_dn_0": 1.01753, "loss_cls_dn_1": 0.45826, "loss_box_dn_1": 1.26671, "loss_cls_dn_2": 0.45801, "loss_box_dn_2": 1.37433, "loss_cls_dn_3": 0.44347, "loss_box_dn_3": 1.46293, "loss_cls_dn_4": 0.43705, "loss_box_dn_4": 1.46984, "loss_cls_dn_5": 0.45581, "loss_box_dn_5": 1.62295, "loss_dense_depth": 1.13858, "loss": 40.58182, "grad_norm": 79.32542, "time": 1.58377} -{"mode": "train", "epoch": 1, "iter": 20, "lr": 0.00011, "memory": 49164, "data_time": 0.07674, "loss_cls_0": 1.19092, "loss_box_0": 2.22193, "loss_cns_0": 0.6105, "loss_yns_0": 0.1707, "loss_cls_1": 1.24699, "loss_box_1": 2.84003, "loss_cns_1": 0.55961, "loss_yns_1": 0.17652, "loss_cls_2": 1.2552, "loss_box_2": 2.8255, "loss_cns_2": 0.55963, "loss_yns_2": 0.1772, "loss_cls_3": 1.24134, "loss_box_3": 2.8197, "loss_cns_3": 0.55366, "loss_yns_3": 0.1723, "loss_cls_4": 1.25983, "loss_box_4": 2.81315, "loss_cns_4": 0.57196, "loss_yns_4": 0.174, "loss_cls_5": 1.26668, "loss_box_5": 2.86047, "loss_cns_5": 0.5513, "loss_yns_5": 0.17554, "loss_cls_dn_0": 0.46678, "loss_box_dn_0": 0.9976, "loss_cls_dn_1": 0.42073, "loss_box_dn_1": 1.49896, "loss_cls_dn_2": 0.42192, "loss_box_dn_2": 1.56058, "loss_cls_dn_3": 0.41628, "loss_box_dn_3": 1.59588, "loss_cls_dn_4": 0.40992, "loss_box_dn_4": 1.57696, "loss_cls_dn_5": 0.41769, "loss_box_dn_5": 1.67448, "loss_dense_depth": 1.06369, "loss": 40.81611, "grad_norm": 58.73482, "time": 1.57014} -{"mode": "train", "epoch": 1, "iter": 21, "lr": 0.00011, "memory": 49164, "data_time": 0.14258, "loss_cls_0": 1.15085, "loss_box_0": 2.13326, "loss_cns_0": 0.61544, "loss_yns_0": 0.16818, "loss_cls_1": 1.24913, "loss_box_1": 2.85972, "loss_cns_1": 0.56064, "loss_yns_1": 0.1745, "loss_cls_2": 1.24772, "loss_box_2": 2.85268, "loss_cns_2": 0.55865, "loss_yns_2": 0.17385, "loss_cls_3": 1.24417, "loss_box_3": 2.91105, "loss_cns_3": 0.55575, "loss_yns_3": 0.17175, "loss_cls_4": 1.25236, "loss_box_4": 2.88163, "loss_cns_4": 0.55565, "loss_yns_4": 0.1725, "loss_cls_5": 1.26726, "loss_box_5": 2.93405, "loss_cns_5": 0.5523, "loss_yns_5": 0.17494, "loss_cls_dn_0": 0.48028, "loss_box_dn_0": 1.00597, "loss_cls_dn_1": 0.43532, "loss_box_dn_1": 1.23886, "loss_cls_dn_2": 0.42942, "loss_box_dn_2": 1.26511, "loss_cls_dn_3": 0.43366, "loss_box_dn_3": 1.30947, "loss_cls_dn_4": 0.43354, "loss_box_dn_4": 1.28407, "loss_cls_dn_5": 0.43667, "loss_box_dn_5": 1.34194, "loss_dense_depth": 1.06189, "loss": 39.57423, "grad_norm": 63.46587, "time": 1.74384} -{"mode": "train", "epoch": 1, "iter": 22, "lr": 0.00011, "memory": 49164, "data_time": 0.16721, "loss_cls_0": 1.15949, "loss_box_0": 2.09044, "loss_cns_0": 0.61709, "loss_yns_0": 0.16836, "loss_cls_1": 1.25286, "loss_box_1": 2.84863, "loss_cns_1": 0.54532, "loss_yns_1": 0.17491, "loss_cls_2": 1.25976, "loss_box_2": 2.80138, "loss_cns_2": 0.55083, "loss_yns_2": 0.17424, "loss_cls_3": 1.26129, "loss_box_3": 2.82855, "loss_cns_3": 0.55795, "loss_yns_3": 0.17233, "loss_cls_4": 1.25221, "loss_box_4": 2.81406, "loss_cns_4": 0.55481, "loss_yns_4": 0.17126, "loss_cls_5": 1.26385, "loss_box_5": 2.86335, "loss_cns_5": 0.56486, "loss_yns_5": 0.17462, "loss_cls_dn_0": 0.48149, "loss_box_dn_0": 1.00368, "loss_cls_dn_1": 0.44241, "loss_box_dn_1": 1.11199, "loss_cls_dn_2": 0.42838, "loss_box_dn_2": 1.09956, "loss_cls_dn_3": 0.43669, "loss_box_dn_3": 1.14828, "loss_cls_dn_4": 0.44105, "loss_box_dn_4": 1.17745, "loss_cls_dn_5": 0.44728, "loss_box_dn_5": 1.20936, "loss_dense_depth": 1.04322, "loss": 38.59327, "grad_norm": 60.96305, "time": 1.66644} -{"mode": "train", "epoch": 1, "iter": 23, "lr": 0.00011, "memory": 49164, "data_time": 0.0861, "loss_cls_0": 1.15007, "loss_box_0": 2.08691, "loss_cns_0": 0.62022, "loss_yns_0": 0.16943, "loss_cls_1": 1.224, "loss_box_1": 2.86808, "loss_cns_1": 0.537, "loss_yns_1": 0.17341, "loss_cls_2": 1.24738, "loss_box_2": 2.77609, "loss_cns_2": 0.54939, "loss_yns_2": 0.17679, "loss_cls_3": 1.24548, "loss_box_3": 2.78711, "loss_cns_3": 0.56296, "loss_yns_3": 0.17518, "loss_cls_4": 1.24038, "loss_box_4": 2.84626, "loss_cns_4": 0.56825, "loss_yns_4": 0.1707, "loss_cls_5": 1.2386, "loss_box_5": 2.86698, "loss_cns_5": 0.57287, "loss_yns_5": 0.17337, "loss_cls_dn_0": 0.47452, "loss_box_dn_0": 0.98607, "loss_cls_dn_1": 0.44147, "loss_box_dn_1": 1.18198, "loss_cls_dn_2": 0.42376, "loss_box_dn_2": 1.14817, "loss_cls_dn_3": 0.43452, "loss_box_dn_3": 1.20845, "loss_cls_dn_4": 0.42785, "loss_box_dn_4": 1.32736, "loss_cls_dn_5": 0.44887, "loss_box_dn_5": 1.34866, "loss_dense_depth": 1.05869, "loss": 38.93728, "grad_norm": 59.07341, "time": 1.57287} -{"mode": "train", "epoch": 1, "iter": 24, "lr": 0.00011, "memory": 49164, "data_time": 0.09437, "loss_cls_0": 1.16909, "loss_box_0": 2.12804, "loss_cns_0": 0.62469, "loss_yns_0": 0.16927, "loss_cls_1": 1.23662, "loss_box_1": 2.88809, "loss_cns_1": 0.55161, "loss_yns_1": 0.17242, "loss_cls_2": 1.25738, "loss_box_2": 2.83893, "loss_cns_2": 0.55979, "loss_yns_2": 0.17362, "loss_cls_3": 1.24591, "loss_box_3": 2.87709, "loss_cns_3": 0.56831, "loss_yns_3": 0.17142, "loss_cls_4": 1.25107, "loss_box_4": 2.94127, "loss_cns_4": 0.57552, "loss_yns_4": 0.17094, "loss_cls_5": 1.24278, "loss_box_5": 2.92724, "loss_cns_5": 0.56856, "loss_yns_5": 0.16845, "loss_cls_dn_0": 0.47261, "loss_box_dn_0": 0.99976, "loss_cls_dn_1": 0.43062, "loss_box_dn_1": 1.28146, "loss_cls_dn_2": 0.41893, "loss_box_dn_2": 1.27277, "loss_cls_dn_3": 0.42767, "loss_box_dn_3": 1.36404, "loss_cls_dn_4": 0.41452, "loss_box_dn_4": 1.51269, "loss_cls_dn_5": 0.43938, "loss_box_dn_5": 1.51247, "loss_dense_depth": 1.0894, "loss": 40.11442, "grad_norm": 63.79114, "time": 1.58187} -{"mode": "train", "epoch": 1, "iter": 25, "lr": 0.00011, "memory": 49164, "data_time": 0.08649, "loss_cls_0": 1.1792, "loss_box_0": 2.14445, "loss_cns_0": 0.62631, "loss_yns_0": 0.16879, "loss_cls_1": 1.22767, "loss_box_1": 2.79526, "loss_cns_1": 0.55436, "loss_yns_1": 0.17184, "loss_cls_2": 1.26113, "loss_box_2": 2.79071, "loss_cns_2": 0.56519, "loss_yns_2": 0.17208, "loss_cls_3": 1.24039, "loss_box_3": 2.79659, "loss_cns_3": 0.57297, "loss_yns_3": 0.1719, "loss_cls_4": 1.23799, "loss_box_4": 2.84593, "loss_cns_4": 0.58433, "loss_yns_4": 0.17349, "loss_cls_5": 1.24499, "loss_box_5": 2.80293, "loss_cns_5": 0.58506, "loss_yns_5": 0.16719, "loss_cls_dn_0": 0.46152, "loss_box_dn_0": 0.99376, "loss_cls_dn_1": 0.4202, "loss_box_dn_1": 1.30825, "loss_cls_dn_2": 0.41451, "loss_box_dn_2": 1.3329, "loss_cls_dn_3": 0.42436, "loss_box_dn_3": 1.42423, "loss_cls_dn_4": 0.41334, "loss_box_dn_4": 1.54152, "loss_cls_dn_5": 0.43272, "loss_box_dn_5": 1.51473, "loss_dense_depth": 1.06337, "loss": 39.82616, "grad_norm": 64.03695, "time": 1.64284} -{"mode": "train", "epoch": 1, "iter": 26, "lr": 0.00011, "memory": 49164, "data_time": 0.07663, "loss_cls_0": 1.15951, "loss_box_0": 2.12072, "loss_cns_0": 0.62465, "loss_yns_0": 0.16919, "loss_cls_1": 1.21804, "loss_box_1": 2.66267, "loss_cns_1": 0.57321, "loss_yns_1": 0.16983, "loss_cls_2": 1.24697, "loss_box_2": 2.68713, "loss_cns_2": 0.5891, "loss_yns_2": 0.17496, "loss_cls_3": 1.23118, "loss_box_3": 2.69147, "loss_cns_3": 0.58903, "loss_yns_3": 0.16917, "loss_cls_4": 1.23431, "loss_box_4": 2.71463, "loss_cns_4": 0.60839, "loss_yns_4": 0.16978, "loss_cls_5": 1.24498, "loss_box_5": 2.73054, "loss_cns_5": 0.61361, "loss_yns_5": 0.17238, "loss_cls_dn_0": 0.44864, "loss_box_dn_0": 0.99872, "loss_cls_dn_1": 0.43752, "loss_box_dn_1": 1.1948, "loss_cls_dn_2": 0.42981, "loss_box_dn_2": 1.22894, "loss_cls_dn_3": 0.44233, "loss_box_dn_3": 1.30561, "loss_cls_dn_4": 0.43573, "loss_box_dn_4": 1.37985, "loss_cls_dn_5": 0.45296, "loss_box_dn_5": 1.36935, "loss_dense_depth": 1.03402, "loss": 38.72373, "grad_norm": 69.60414, "time": 1.59679} -{"mode": "train", "epoch": 1, "iter": 27, "lr": 0.00011, "memory": 49164, "data_time": 0.09322, "loss_cls_0": 1.14584, "loss_box_0": 2.12099, "loss_cns_0": 0.61862, "loss_yns_0": 0.17099, "loss_cls_1": 1.21548, "loss_box_1": 2.59328, "loss_cns_1": 0.57794, "loss_yns_1": 0.17276, "loss_cls_2": 1.23772, "loss_box_2": 2.64882, "loss_cns_2": 0.58881, "loss_yns_2": 0.17776, "loss_cls_3": 1.22137, "loss_box_3": 2.63832, "loss_cns_3": 0.57389, "loss_yns_3": 0.17103, "loss_cls_4": 1.2362, "loss_box_4": 2.66585, "loss_cns_4": 0.57751, "loss_yns_4": 0.17178, "loss_cls_5": 1.22829, "loss_box_5": 2.7127, "loss_cns_5": 0.56743, "loss_yns_5": 0.17297, "loss_cls_dn_0": 0.4475, "loss_box_dn_0": 0.99502, "loss_cls_dn_1": 0.41857, "loss_box_dn_1": 1.18969, "loss_cls_dn_2": 0.41732, "loss_box_dn_2": 1.17498, "loss_cls_dn_3": 0.4277, "loss_box_dn_3": 1.18417, "loss_cls_dn_4": 0.42265, "loss_box_dn_4": 1.21235, "loss_cls_dn_5": 0.43568, "loss_box_dn_5": 1.22786, "loss_dense_depth": 1.0256, "loss": 37.78542, "grad_norm": 63.11638, "time": 1.59271} -{"mode": "train", "epoch": 1, "iter": 28, "lr": 0.00011, "memory": 49164, "data_time": 0.07726, "loss_cls_0": 1.13018, "loss_box_0": 2.10003, "loss_cns_0": 0.61665, "loss_yns_0": 0.17051, "loss_cls_1": 1.20904, "loss_box_1": 2.47256, "loss_cns_1": 0.58493, "loss_yns_1": 0.17238, "loss_cls_2": 1.22543, "loss_box_2": 2.45637, "loss_cns_2": 0.59298, "loss_yns_2": 0.17691, "loss_cls_3": 1.23063, "loss_box_3": 2.45143, "loss_cns_3": 0.59048, "loss_yns_3": 0.17291, "loss_cls_4": 1.2433, "loss_box_4": 2.50999, "loss_cns_4": 0.60096, "loss_yns_4": 0.17289, "loss_cls_5": 1.2556, "loss_box_5": 2.46145, "loss_cns_5": 0.59227, "loss_yns_5": 0.17545, "loss_cls_dn_0": 0.44059, "loss_box_dn_0": 0.98801, "loss_cls_dn_1": 0.40259, "loss_box_dn_1": 1.16886, "loss_cls_dn_2": 0.40876, "loss_box_dn_2": 1.12533, "loss_cls_dn_3": 0.40502, "loss_box_dn_3": 1.127, "loss_cls_dn_4": 0.40251, "loss_box_dn_4": 1.15803, "loss_cls_dn_5": 0.40441, "loss_box_dn_5": 1.17466, "loss_dense_depth": 1.01141, "loss": 36.58248, "grad_norm": 52.46316, "time": 1.58253} -{"mode": "train", "epoch": 1, "iter": 29, "lr": 0.00011, "memory": 49164, "data_time": 0.0745, "loss_cls_0": 1.10557, "loss_box_0": 2.06476, "loss_cns_0": 0.61835, "loss_yns_0": 0.1686, "loss_cls_1": 1.20007, "loss_box_1": 2.441, "loss_cns_1": 0.60001, "loss_yns_1": 0.17264, "loss_cls_2": 1.1963, "loss_box_2": 2.40011, "loss_cns_2": 0.60809, "loss_yns_2": 0.17643, "loss_cls_3": 1.20781, "loss_box_3": 2.43845, "loss_cns_3": 0.60838, "loss_yns_3": 0.17318, "loss_cls_4": 1.22222, "loss_box_4": 2.53021, "loss_cns_4": 0.60502, "loss_yns_4": 0.17271, "loss_cls_5": 1.24996, "loss_box_5": 2.53743, "loss_cns_5": 0.58999, "loss_yns_5": 0.17171, "loss_cls_dn_0": 0.43801, "loss_box_dn_0": 0.98455, "loss_cls_dn_1": 0.40557, "loss_box_dn_1": 1.08069, "loss_cls_dn_2": 0.43031, "loss_box_dn_2": 1.07681, "loss_cls_dn_3": 0.41577, "loss_box_dn_3": 1.11765, "loss_cls_dn_4": 0.41241, "loss_box_dn_4": 1.17676, "loss_cls_dn_5": 0.40667, "loss_box_dn_5": 1.2692, "loss_dense_depth": 1.06947, "loss": 36.54283, "grad_norm": 68.62055, "time": 1.55809} -{"mode": "train", "epoch": 1, "iter": 30, "lr": 0.00011, "memory": 49164, "data_time": 0.07368, "loss_cls_0": 1.08492, "loss_box_0": 2.0109, "loss_cns_0": 0.62233, "loss_yns_0": 0.17056, "loss_cls_1": 1.16418, "loss_box_1": 2.4477, "loss_cns_1": 0.59848, "loss_yns_1": 0.17355, "loss_cls_2": 1.18211, "loss_box_2": 2.4487, "loss_cns_2": 0.60995, "loss_yns_2": 0.17604, "loss_cls_3": 1.18212, "loss_box_3": 2.49224, "loss_cns_3": 0.60485, "loss_yns_3": 0.17396, "loss_cls_4": 1.18045, "loss_box_4": 2.56527, "loss_cns_4": 0.59325, "loss_yns_4": 0.17125, "loss_cls_5": 1.19195, "loss_box_5": 2.61555, "loss_cns_5": 0.58711, "loss_yns_5": 0.17069, "loss_cls_dn_0": 0.43895, "loss_box_dn_0": 0.97655, "loss_cls_dn_1": 0.39767, "loss_box_dn_1": 1.14362, "loss_cls_dn_2": 0.43484, "loss_box_dn_2": 1.19787, "loss_cls_dn_3": 0.41695, "loss_box_dn_3": 1.25161, "loss_cls_dn_4": 0.41294, "loss_box_dn_4": 1.30692, "loss_cls_dn_5": 0.40227, "loss_box_dn_5": 1.43083, "loss_dense_depth": 1.02744, "loss": 37.05656, "grad_norm": 78.71377, "time": 1.62838} -{"mode": "train", "epoch": 1, "iter": 31, "lr": 0.00011, "memory": 49164, "data_time": 0.07328, "loss_cls_0": 1.07353, "loss_box_0": 2.04395, "loss_cns_0": 0.62305, "loss_yns_0": 0.17375, "loss_cls_1": 1.15343, "loss_box_1": 2.47146, "loss_cns_1": 0.59759, "loss_yns_1": 0.17145, "loss_cls_2": 1.17373, "loss_box_2": 2.48469, "loss_cns_2": 0.60435, "loss_yns_2": 0.17414, "loss_cls_3": 1.17711, "loss_box_3": 2.49598, "loss_cns_3": 0.60386, "loss_yns_3": 0.17367, "loss_cls_4": 1.16598, "loss_box_4": 2.53917, "loss_cns_4": 0.60118, "loss_yns_4": 0.17127, "loss_cls_5": 1.17139, "loss_box_5": 2.56605, "loss_cns_5": 0.60442, "loss_yns_5": 0.17127, "loss_cls_dn_0": 0.44534, "loss_box_dn_0": 0.96696, "loss_cls_dn_1": 0.39252, "loss_box_dn_1": 1.21242, "loss_cls_dn_2": 0.4314, "loss_box_dn_2": 1.26472, "loss_cls_dn_3": 0.41547, "loss_box_dn_3": 1.30871, "loss_cls_dn_4": 0.41296, "loss_box_dn_4": 1.34302, "loss_cls_dn_5": 0.4081, "loss_box_dn_5": 1.45298, "loss_dense_depth": 1.0325, "loss": 37.27356, "grad_norm": 57.5837, "time": 1.60783} -{"mode": "train", "epoch": 1, "iter": 32, "lr": 0.00011, "memory": 49164, "data_time": 0.05969, "loss_cls_0": 1.05642, "loss_box_0": 2.08635, "loss_cns_0": 0.61998, "loss_yns_0": 0.17136, "loss_cls_1": 1.12643, "loss_box_1": 2.51041, "loss_cns_1": 0.57827, "loss_yns_1": 0.16988, "loss_cls_2": 1.13697, "loss_box_2": 2.52017, "loss_cns_2": 0.57828, "loss_yns_2": 0.16977, "loss_cls_3": 1.15585, "loss_box_3": 2.50575, "loss_cns_3": 0.58936, "loss_yns_3": 0.17454, "loss_cls_4": 1.15177, "loss_box_4": 2.52994, "loss_cns_4": 0.5812, "loss_yns_4": 0.17058, "loss_cls_5": 1.16322, "loss_box_5": 2.58537, "loss_cns_5": 0.58786, "loss_yns_5": 0.17162, "loss_cls_dn_0": 0.42733, "loss_box_dn_0": 0.97759, "loss_cls_dn_1": 0.37887, "loss_box_dn_1": 1.18625, "loss_cls_dn_2": 0.40763, "loss_box_dn_2": 1.22992, "loss_cls_dn_3": 0.39373, "loss_box_dn_3": 1.25499, "loss_cls_dn_4": 0.39555, "loss_box_dn_4": 1.27187, "loss_cls_dn_5": 0.40367, "loss_box_dn_5": 1.36906, "loss_dense_depth": 1.06874, "loss": 36.85654, "grad_norm": 65.86033, "time": 1.62943} -{"mode": "train", "epoch": 1, "iter": 33, "lr": 0.00011, "memory": 49164, "data_time": 0.07703, "loss_cls_0": 1.09392, "loss_box_0": 2.06583, "loss_cns_0": 0.62326, "loss_yns_0": 0.17023, "loss_cls_1": 1.12778, "loss_box_1": 2.55348, "loss_cns_1": 0.56221, "loss_yns_1": 0.16938, "loss_cls_2": 1.13449, "loss_box_2": 2.49454, "loss_cns_2": 0.58338, "loss_yns_2": 0.1716, "loss_cls_3": 1.16832, "loss_box_3": 2.45526, "loss_cns_3": 0.59157, "loss_yns_3": 0.17576, "loss_cls_4": 1.15922, "loss_box_4": 2.43555, "loss_cns_4": 0.59599, "loss_yns_4": 0.17094, "loss_cls_5": 1.15871, "loss_box_5": 2.45819, "loss_cns_5": 0.59909, "loss_yns_5": 0.17196, "loss_cls_dn_0": 0.41618, "loss_box_dn_0": 0.98327, "loss_cls_dn_1": 0.34476, "loss_box_dn_1": 1.32753, "loss_cls_dn_2": 0.36527, "loss_box_dn_2": 1.34672, "loss_cls_dn_3": 0.35049, "loss_box_dn_3": 1.34288, "loss_cls_dn_4": 0.35729, "loss_box_dn_4": 1.33389, "loss_cls_dn_5": 0.37179, "loss_box_dn_5": 1.38537, "loss_dense_depth": 0.99904, "loss": 36.81514, "grad_norm": 52.43612, "time": 1.69428} -{"mode": "train", "epoch": 1, "iter": 34, "lr": 0.00011, "memory": 49164, "data_time": 0.08271, "loss_cls_0": 1.10158, "loss_box_0": 2.02518, "loss_cns_0": 0.62451, "loss_yns_0": 0.17205, "loss_cls_1": 1.15049, "loss_box_1": 2.41483, "loss_cns_1": 0.58468, "loss_yns_1": 0.17077, "loss_cls_2": 1.18362, "loss_box_2": 2.40714, "loss_cns_2": 0.59937, "loss_yns_2": 0.17226, "loss_cls_3": 1.18208, "loss_box_3": 2.4301, "loss_cns_3": 0.59488, "loss_yns_3": 0.17428, "loss_cls_4": 1.18255, "loss_box_4": 2.4193, "loss_cns_4": 0.60325, "loss_yns_4": 0.17063, "loss_cls_5": 1.15406, "loss_box_5": 2.3966, "loss_cns_5": 0.60635, "loss_yns_5": 0.1737, "loss_cls_dn_0": 0.41512, "loss_box_dn_0": 0.96905, "loss_cls_dn_1": 0.32652, "loss_box_dn_1": 1.2821, "loss_cls_dn_2": 0.33444, "loss_box_dn_2": 1.28546, "loss_cls_dn_3": 0.33364, "loss_box_dn_3": 1.28613, "loss_cls_dn_4": 0.33517, "loss_box_dn_4": 1.27864, "loss_cls_dn_5": 0.35491, "loss_box_dn_5": 1.27511, "loss_dense_depth": 1.07057, "loss": 36.24113, "grad_norm": 65.47511, "time": 1.57026} -{"mode": "train", "epoch": 1, "iter": 35, "lr": 0.00011, "memory": 49164, "data_time": 0.07197, "loss_cls_0": 1.07423, "loss_box_0": 1.97364, "loss_cns_0": 0.6247, "loss_yns_0": 0.17203, "loss_cls_1": 1.17742, "loss_box_1": 2.35948, "loss_cns_1": 0.58989, "loss_yns_1": 0.16957, "loss_cls_2": 1.231, "loss_box_2": 2.3585, "loss_cns_2": 0.59472, "loss_yns_2": 0.17311, "loss_cls_3": 1.17467, "loss_box_3": 2.43211, "loss_cns_3": 0.58063, "loss_yns_3": 0.17321, "loss_cls_4": 1.19929, "loss_box_4": 2.46441, "loss_cns_4": 0.59264, "loss_yns_4": 0.17025, "loss_cls_5": 1.17797, "loss_box_5": 2.4115, "loss_cns_5": 0.59886, "loss_yns_5": 0.17264, "loss_cls_dn_0": 0.42341, "loss_box_dn_0": 0.94412, "loss_cls_dn_1": 0.32151, "loss_box_dn_1": 1.17995, "loss_cls_dn_2": 0.32516, "loss_box_dn_2": 1.17462, "loss_cls_dn_3": 0.33876, "loss_box_dn_3": 1.20506, "loss_cls_dn_4": 0.33564, "loss_box_dn_4": 1.24745, "loss_cls_dn_5": 0.35224, "loss_box_dn_5": 1.22544, "loss_dense_depth": 1.05787, "loss": 35.77768, "grad_norm": 68.93083, "time": 1.58788} -{"mode": "train", "epoch": 1, "iter": 36, "lr": 0.00011, "memory": 49164, "data_time": 0.0732, "loss_cls_0": 1.04767, "loss_box_0": 1.96991, "loss_cns_0": 0.621, "loss_yns_0": 0.17194, "loss_cls_1": 1.13861, "loss_box_1": 2.39618, "loss_cns_1": 0.59584, "loss_yns_1": 0.17171, "loss_cls_2": 1.16911, "loss_box_2": 2.36493, "loss_cns_2": 0.60698, "loss_yns_2": 0.17097, "loss_cls_3": 1.14663, "loss_box_3": 2.39574, "loss_cns_3": 0.60984, "loss_yns_3": 0.17122, "loss_cls_4": 1.14152, "loss_box_4": 2.4257, "loss_cns_4": 0.61159, "loss_yns_4": 0.16952, "loss_cls_5": 1.16813, "loss_box_5": 2.39866, "loss_cns_5": 0.61401, "loss_yns_5": 0.17124, "loss_cls_dn_0": 0.43382, "loss_box_dn_0": 0.93179, "loss_cls_dn_1": 0.34891, "loss_box_dn_1": 1.04016, "loss_cls_dn_2": 0.35811, "loss_box_dn_2": 1.02474, "loss_cls_dn_3": 0.37863, "loss_box_dn_3": 1.09, "loss_cls_dn_4": 0.38164, "loss_box_dn_4": 1.20948, "loss_cls_dn_5": 0.38884, "loss_box_dn_5": 1.2162, "loss_dense_depth": 1.04222, "loss": 35.29316, "grad_norm": 67.12705, "time": 1.59792} -{"mode": "train", "epoch": 1, "iter": 37, "lr": 0.00011, "memory": 49164, "data_time": 0.07286, "loss_cls_0": 1.05277, "loss_box_0": 1.99533, "loss_cns_0": 0.62123, "loss_yns_0": 0.17364, "loss_cls_1": 1.11703, "loss_box_1": 2.49594, "loss_cns_1": 0.60256, "loss_yns_1": 0.17174, "loss_cls_2": 1.14164, "loss_box_2": 2.55489, "loss_cns_2": 0.59568, "loss_yns_2": 0.17181, "loss_cls_3": 1.14123, "loss_box_3": 2.56934, "loss_cns_3": 0.59726, "loss_yns_3": 0.17053, "loss_cls_4": 1.13631, "loss_box_4": 2.61028, "loss_cns_4": 0.59772, "loss_yns_4": 0.16938, "loss_cls_5": 1.14811, "loss_box_5": 2.65243, "loss_cns_5": 0.57816, "loss_yns_5": 0.17025, "loss_cls_dn_0": 0.44391, "loss_box_dn_0": 0.93321, "loss_cls_dn_1": 0.3523, "loss_box_dn_1": 1.12689, "loss_cls_dn_2": 0.36841, "loss_box_dn_2": 1.13289, "loss_cls_dn_3": 0.38231, "loss_box_dn_3": 1.19636, "loss_cls_dn_4": 0.39787, "loss_box_dn_4": 1.32393, "loss_cls_dn_5": 0.39094, "loss_box_dn_5": 1.36449, "loss_dense_depth": 0.97742, "loss": 36.6262, "grad_norm": 92.67823, "time": 1.62917} -{"mode": "train", "epoch": 1, "iter": 38, "lr": 0.00011, "memory": 49164, "data_time": 0.0739, "loss_cls_0": 1.03648, "loss_box_0": 2.00631, "loss_cns_0": 0.62337, "loss_yns_0": 0.16982, "loss_cls_1": 1.08924, "loss_box_1": 2.51246, "loss_cns_1": 0.58132, "loss_yns_1": 0.17025, "loss_cls_2": 1.13657, "loss_box_2": 2.55271, "loss_cns_2": 0.58139, "loss_yns_2": 0.17267, "loss_cls_3": 1.13888, "loss_box_3": 2.57501, "loss_cns_3": 0.58134, "loss_yns_3": 0.16836, "loss_cls_4": 1.1386, "loss_box_4": 2.61559, "loss_cns_4": 0.58046, "loss_yns_4": 0.17011, "loss_cls_5": 1.13588, "loss_box_5": 2.64312, "loss_cns_5": 0.56003, "loss_yns_5": 0.16722, "loss_cls_dn_0": 0.43724, "loss_box_dn_0": 0.9352, "loss_cls_dn_1": 0.36994, "loss_box_dn_1": 1.18271, "loss_cls_dn_2": 0.38963, "loss_box_dn_2": 1.16214, "loss_cls_dn_3": 0.39302, "loss_box_dn_3": 1.21823, "loss_cls_dn_4": 0.41624, "loss_box_dn_4": 1.33093, "loss_cls_dn_5": 0.40655, "loss_box_dn_5": 1.36387, "loss_dense_depth": 1.07641, "loss": 36.78927, "grad_norm": 86.88221, "time": 1.57246} -{"mode": "train", "epoch": 1, "iter": 39, "lr": 0.00012, "memory": 49164, "data_time": 0.08691, "loss_cls_0": 1.05745, "loss_box_0": 1.96928, "loss_cns_0": 0.62613, "loss_yns_0": 0.16978, "loss_cls_1": 1.11926, "loss_box_1": 2.38657, "loss_cns_1": 0.5883, "loss_yns_1": 0.16975, "loss_cls_2": 1.16207, "loss_box_2": 2.35147, "loss_cns_2": 0.60526, "loss_yns_2": 0.17087, "loss_cls_3": 1.15109, "loss_box_3": 2.39622, "loss_cns_3": 0.61517, "loss_yns_3": 0.17155, "loss_cls_4": 1.14714, "loss_box_4": 2.42473, "loss_cns_4": 0.60983, "loss_yns_4": 0.16791, "loss_cls_5": 1.15308, "loss_box_5": 2.44131, "loss_cns_5": 0.62514, "loss_yns_5": 0.16972, "loss_cls_dn_0": 0.42109, "loss_box_dn_0": 0.91935, "loss_cls_dn_1": 0.38319, "loss_box_dn_1": 1.13704, "loss_cls_dn_2": 0.40648, "loss_box_dn_2": 1.0846, "loss_cls_dn_3": 0.39614, "loss_box_dn_3": 1.13364, "loss_cls_dn_4": 0.41917, "loss_box_dn_4": 1.22214, "loss_cls_dn_5": 0.41142, "loss_box_dn_5": 1.24237, "loss_dense_depth": 1.0037, "loss": 35.6294, "grad_norm": 69.64661, "time": 1.57771} -{"mode": "train", "epoch": 1, "iter": 40, "lr": 0.00012, "memory": 49164, "data_time": 0.07242, "loss_cls_0": 1.07906, "loss_box_0": 1.95586, "loss_cns_0": 0.62783, "loss_yns_0": 0.17288, "loss_cls_1": 1.10639, "loss_box_1": 2.39116, "loss_cns_1": 0.59128, "loss_yns_1": 0.17179, "loss_cls_2": 1.15326, "loss_box_2": 2.38549, "loss_cns_2": 0.60363, "loss_yns_2": 0.17004, "loss_cls_3": 1.14426, "loss_box_3": 2.4468, "loss_cns_3": 0.60834, "loss_yns_3": 0.1704, "loss_cls_4": 1.13778, "loss_box_4": 2.48468, "loss_cns_4": 0.60323, "loss_yns_4": 0.16821, "loss_cls_5": 1.15542, "loss_box_5": 2.53862, "loss_cns_5": 0.61591, "loss_yns_5": 0.16968, "loss_cls_dn_0": 0.39626, "loss_box_dn_0": 0.91413, "loss_cls_dn_1": 0.38532, "loss_box_dn_1": 1.12022, "loss_cls_dn_2": 0.40232, "loss_box_dn_2": 1.08525, "loss_cls_dn_3": 0.38227, "loss_box_dn_3": 1.11702, "loss_cls_dn_4": 0.39363, "loss_box_dn_4": 1.16819, "loss_cls_dn_5": 0.39433, "loss_box_dn_5": 1.18443, "loss_dense_depth": 1.00596, "loss": 35.60134, "grad_norm": 86.2683, "time": 1.57761} -{"mode": "train", "epoch": 1, "iter": 41, "lr": 0.00012, "memory": 49164, "data_time": 0.11152, "loss_cls_0": 1.06566, "loss_box_0": 1.94834, "loss_cns_0": 0.62532, "loss_yns_0": 0.17409, "loss_cls_1": 1.08707, "loss_box_1": 2.37426, "loss_cns_1": 0.58561, "loss_yns_1": 0.16675, "loss_cls_2": 1.10819, "loss_box_2": 2.36927, "loss_cns_2": 0.59917, "loss_yns_2": 0.17065, "loss_cls_3": 1.12908, "loss_box_3": 2.37756, "loss_cns_3": 0.6008, "loss_yns_3": 0.16909, "loss_cls_4": 1.13073, "loss_box_4": 2.39237, "loss_cns_4": 0.60268, "loss_yns_4": 0.1706, "loss_cls_5": 1.13923, "loss_box_5": 2.46216, "loss_cns_5": 0.61463, "loss_yns_5": 0.16645, "loss_cls_dn_0": 0.38272, "loss_box_dn_0": 0.91082, "loss_cls_dn_1": 0.35683, "loss_box_dn_1": 1.10202, "loss_cls_dn_2": 0.3669, "loss_box_dn_2": 1.07862, "loss_cls_dn_3": 0.34673, "loss_box_dn_3": 1.07248, "loss_cls_dn_4": 0.34456, "loss_box_dn_4": 1.09147, "loss_cls_dn_5": 0.35541, "loss_box_dn_5": 1.10803, "loss_dense_depth": 1.01858, "loss": 34.76492, "grad_norm": 69.30612, "time": 1.67739} -{"mode": "train", "epoch": 1, "iter": 42, "lr": 0.00012, "memory": 49164, "data_time": 0.17373, "loss_cls_0": 1.04609, "loss_box_0": 1.93075, "loss_cns_0": 0.62681, "loss_yns_0": 0.16708, "loss_cls_1": 1.1082, "loss_box_1": 2.33771, "loss_cns_1": 0.60322, "loss_yns_1": 0.16602, "loss_cls_2": 1.12713, "loss_box_2": 2.30269, "loss_cns_2": 0.62467, "loss_yns_2": 0.169, "loss_cls_3": 1.15794, "loss_box_3": 2.26369, "loss_cns_3": 0.62887, "loss_yns_3": 0.17026, "loss_cls_4": 1.1716, "loss_box_4": 2.25894, "loss_cns_4": 0.64311, "loss_yns_4": 0.1685, "loss_cls_5": 1.15, "loss_box_5": 2.27363, "loss_cns_5": 0.65165, "loss_yns_5": 0.16585, "loss_cls_dn_0": 0.39133, "loss_box_dn_0": 0.91438, "loss_cls_dn_1": 0.34708, "loss_box_dn_1": 1.02046, "loss_cls_dn_2": 0.35228, "loss_box_dn_2": 1.0008, "loss_cls_dn_3": 0.34647, "loss_box_dn_3": 0.98145, "loss_cls_dn_4": 0.33004, "loss_box_dn_4": 0.98226, "loss_cls_dn_5": 0.35267, "loss_box_dn_5": 1.00397, "loss_dense_depth": 0.95087, "loss": 33.88746, "grad_norm": 58.87595, "time": 1.65419} -{"mode": "train", "epoch": 1, "iter": 43, "lr": 0.00012, "memory": 49164, "data_time": 0.09293, "loss_cls_0": 1.02905, "loss_box_0": 1.97933, "loss_cns_0": 0.61644, "loss_yns_0": 0.16526, "loss_cls_1": 1.14316, "loss_box_1": 2.29595, "loss_cns_1": 0.60862, "loss_yns_1": 0.16418, "loss_cls_2": 1.19059, "loss_box_2": 2.28725, "loss_cns_2": 0.62118, "loss_yns_2": 0.16641, "loss_cls_3": 1.1577, "loss_box_3": 2.29778, "loss_cns_3": 0.62066, "loss_yns_3": 0.16539, "loss_cls_4": 1.18994, "loss_box_4": 2.36006, "loss_cns_4": 0.61898, "loss_yns_4": 0.16493, "loss_cls_5": 1.17674, "loss_box_5": 2.37196, "loss_cns_5": 0.61171, "loss_yns_5": 0.16429, "loss_cls_dn_0": 0.42376, "loss_box_dn_0": 0.92308, "loss_cls_dn_1": 0.35232, "loss_box_dn_1": 0.98241, "loss_cls_dn_2": 0.35872, "loss_box_dn_2": 0.9991, "loss_cls_dn_3": 0.37038, "loss_box_dn_3": 1.0204, "loss_cls_dn_4": 0.34768, "loss_box_dn_4": 1.05282, "loss_cls_dn_5": 0.37464, "loss_box_dn_5": 1.11519, "loss_dense_depth": 0.96095, "loss": 34.44901, "grad_norm": 67.49393, "time": 1.58247} -{"mode": "train", "epoch": 1, "iter": 44, "lr": 0.00012, "memory": 49164, "data_time": 0.0731, "loss_cls_0": 1.02805, "loss_box_0": 2.00723, "loss_cns_0": 0.61081, "loss_yns_0": 0.16439, "loss_cls_1": 1.10574, "loss_box_1": 2.28805, "loss_cns_1": 0.60305, "loss_yns_1": 0.16216, "loss_cls_2": 1.15041, "loss_box_2": 2.31165, "loss_cns_2": 0.60299, "loss_yns_2": 0.16544, "loss_cls_3": 1.11772, "loss_box_3": 2.37922, "loss_cns_3": 0.60368, "loss_yns_3": 0.16477, "loss_cls_4": 1.1334, "loss_box_4": 2.43775, "loss_cns_4": 0.59498, "loss_yns_4": 0.16486, "loss_cls_5": 1.14206, "loss_box_5": 2.47526, "loss_cns_5": 0.59166, "loss_yns_5": 0.1646, "loss_cls_dn_0": 0.44473, "loss_box_dn_0": 0.91981, "loss_cls_dn_1": 0.35888, "loss_box_dn_1": 1.06678, "loss_cls_dn_2": 0.37021, "loss_box_dn_2": 1.10298, "loss_cls_dn_3": 0.38795, "loss_box_dn_3": 1.15118, "loss_cls_dn_4": 0.36805, "loss_box_dn_4": 1.1891, "loss_cls_dn_5": 0.39174, "loss_box_dn_5": 1.27861, "loss_dense_depth": 1.0408, "loss": 35.24075, "grad_norm": 78.53178, "time": 1.57566} -{"mode": "train", "epoch": 1, "iter": 45, "lr": 0.00012, "memory": 49164, "data_time": 0.09682, "loss_cls_0": 1.05615, "loss_box_0": 1.98424, "loss_cns_0": 0.61499, "loss_yns_0": 0.16543, "loss_cls_1": 1.10802, "loss_box_1": 2.36315, "loss_cns_1": 0.5965, "loss_yns_1": 0.16597, "loss_cls_2": 1.13285, "loss_box_2": 2.35202, "loss_cns_2": 0.60645, "loss_yns_2": 0.16757, "loss_cls_3": 1.13973, "loss_box_3": 2.38365, "loss_cns_3": 0.61155, "loss_yns_3": 0.16687, "loss_cls_4": 1.13505, "loss_box_4": 2.39655, "loss_cns_4": 0.61442, "loss_yns_4": 0.16532, "loss_cls_5": 1.14252, "loss_box_5": 2.42503, "loss_cns_5": 0.61114, "loss_yns_5": 0.16593, "loss_cls_dn_0": 0.44427, "loss_box_dn_0": 0.91573, "loss_cls_dn_1": 0.36625, "loss_box_dn_1": 1.12978, "loss_cls_dn_2": 0.3812, "loss_box_dn_2": 1.1745, "loss_cls_dn_3": 0.39964, "loss_box_dn_3": 1.21083, "loss_cls_dn_4": 0.38526, "loss_box_dn_4": 1.23135, "loss_cls_dn_5": 0.40459, "loss_box_dn_5": 1.31559, "loss_dense_depth": 0.97934, "loss": 35.60944, "grad_norm": 74.29835, "time": 1.64875} -{"mode": "train", "epoch": 1, "iter": 46, "lr": 0.00012, "memory": 49164, "data_time": 0.10239, "loss_cls_0": 1.05145, "loss_box_0": 1.98373, "loss_cns_0": 0.61805, "loss_yns_0": 0.16429, "loss_cls_1": 1.11667, "loss_box_1": 2.5306, "loss_cns_1": 0.59087, "loss_yns_1": 0.1622, "loss_cls_2": 1.11633, "loss_box_2": 2.45411, "loss_cns_2": 0.61013, "loss_yns_2": 0.16463, "loss_cls_3": 1.14531, "loss_box_3": 2.43549, "loss_cns_3": 0.61529, "loss_yns_3": 0.16396, "loss_cls_4": 1.12253, "loss_box_4": 2.41746, "loss_cns_4": 0.61849, "loss_yns_4": 0.16658, "loss_cls_5": 1.1273, "loss_box_5": 2.46497, "loss_cns_5": 0.61001, "loss_yns_5": 0.16224, "loss_cls_dn_0": 0.42675, "loss_box_dn_0": 0.91021, "loss_cls_dn_1": 0.35822, "loss_box_dn_1": 1.13498, "loss_cls_dn_2": 0.37678, "loss_box_dn_2": 1.17244, "loss_cls_dn_3": 0.38759, "loss_box_dn_3": 1.18704, "loss_cls_dn_4": 0.38009, "loss_box_dn_4": 1.18624, "loss_cls_dn_5": 0.39298, "loss_box_dn_5": 1.25752, "loss_dense_depth": 0.9488, "loss": 35.73232, "grad_norm": 62.78919, "time": 1.62287} -{"mode": "train", "epoch": 1, "iter": 47, "lr": 0.00012, "memory": 49164, "data_time": 0.10087, "loss_cls_0": 1.03421, "loss_box_0": 2.00077, "loss_cns_0": 0.61964, "loss_yns_0": 0.16358, "loss_cls_1": 1.13033, "loss_box_1": 2.46262, "loss_cns_1": 0.60522, "loss_yns_1": 0.16185, "loss_cls_2": 1.13047, "loss_box_2": 2.4274, "loss_cns_2": 0.6131, "loss_yns_2": 0.16254, "loss_cls_3": 1.14698, "loss_box_3": 2.4463, "loss_cns_3": 0.6129, "loss_yns_3": 0.16174, "loss_cls_4": 1.12921, "loss_box_4": 2.44409, "loss_cns_4": 0.60715, "loss_yns_4": 0.16891, "loss_cls_5": 1.13799, "loss_box_5": 2.45932, "loss_cns_5": 0.6033, "loss_yns_5": 0.16158, "loss_cls_dn_0": 0.41594, "loss_box_dn_0": 0.91694, "loss_cls_dn_1": 0.35147, "loss_box_dn_1": 1.09319, "loss_cls_dn_2": 0.37437, "loss_box_dn_2": 1.12812, "loss_cls_dn_3": 0.37875, "loss_box_dn_3": 1.1459, "loss_cls_dn_4": 0.37969, "loss_box_dn_4": 1.14319, "loss_cls_dn_5": 0.38715, "loss_box_dn_5": 1.19039, "loss_dense_depth": 0.9511, "loss": 35.44739, "grad_norm": 62.52365, "time": 1.61076} -{"mode": "train", "epoch": 1, "iter": 48, "lr": 0.00012, "memory": 49164, "data_time": 0.07376, "loss_cls_0": 1.0313, "loss_box_0": 1.99154, "loss_cns_0": 0.61785, "loss_yns_0": 0.16317, "loss_cls_1": 1.10357, "loss_box_1": 2.40429, "loss_cns_1": 0.60406, "loss_yns_1": 0.16508, "loss_cls_2": 1.12336, "loss_box_2": 2.34943, "loss_cns_2": 0.62234, "loss_yns_2": 0.16367, "loss_cls_3": 1.13498, "loss_box_3": 2.38113, "loss_cns_3": 0.62541, "loss_yns_3": 0.16396, "loss_cls_4": 1.1206, "loss_box_4": 2.38587, "loss_cns_4": 0.62748, "loss_yns_4": 0.16677, "loss_cls_5": 1.1314, "loss_box_5": 2.37818, "loss_cns_5": 0.62803, "loss_yns_5": 0.16473, "loss_cls_dn_0": 0.40312, "loss_box_dn_0": 0.90341, "loss_cls_dn_1": 0.33263, "loss_box_dn_1": 1.13425, "loss_cls_dn_2": 0.3634, "loss_box_dn_2": 1.1264, "loss_cls_dn_3": 0.35959, "loss_box_dn_3": 1.13503, "loss_cls_dn_4": 0.36552, "loss_box_dn_4": 1.12788, "loss_cls_dn_5": 0.37354, "loss_box_dn_5": 1.14497, "loss_dense_depth": 0.92826, "loss": 34.94621, "grad_norm": 64.93996, "time": 1.58341} -{"mode": "train", "epoch": 1, "iter": 49, "lr": 0.00012, "memory": 49164, "data_time": 0.07393, "loss_cls_0": 1.0778, "loss_box_0": 1.98803, "loss_cns_0": 0.61773, "loss_yns_0": 0.16456, "loss_cls_1": 1.11864, "loss_box_1": 2.42943, "loss_cns_1": 0.60071, "loss_yns_1": 0.16572, "loss_cls_2": 1.11619, "loss_box_2": 2.39496, "loss_cns_2": 0.61773, "loss_yns_2": 0.16197, "loss_cls_3": 1.15235, "loss_box_3": 2.41698, "loss_cns_3": 0.62282, "loss_yns_3": 0.16336, "loss_cls_4": 1.13017, "loss_box_4": 2.41772, "loss_cns_4": 0.62434, "loss_yns_4": 0.1686, "loss_cls_5": 1.13028, "loss_box_5": 2.42585, "loss_cns_5": 0.6249, "loss_yns_5": 0.16554, "loss_cls_dn_0": 0.39719, "loss_box_dn_0": 0.89381, "loss_cls_dn_1": 0.32214, "loss_box_dn_1": 1.15969, "loss_cls_dn_2": 0.35349, "loss_box_dn_2": 1.13836, "loss_cls_dn_3": 0.34915, "loss_box_dn_3": 1.13182, "loss_cls_dn_4": 0.351, "loss_box_dn_4": 1.12417, "loss_cls_dn_5": 0.36659, "loss_box_dn_5": 1.13497, "loss_dense_depth": 0.94844, "loss": 35.16717, "grad_norm": 68.33418, "time": 1.60774} -{"mode": "train", "epoch": 1, "iter": 50, "lr": 0.00012, "memory": 49164, "data_time": 0.08063, "loss_cls_0": 1.05746, "loss_box_0": 1.96958, "loss_cns_0": 0.61956, "loss_yns_0": 0.16395, "loss_cls_1": 1.13298, "loss_box_1": 2.48555, "loss_cns_1": 0.5961, "loss_yns_1": 0.16626, "loss_cls_2": 1.10543, "loss_box_2": 2.42724, "loss_cns_2": 0.61057, "loss_yns_2": 0.16215, "loss_cls_3": 1.15278, "loss_box_3": 2.42727, "loss_cns_3": 0.61742, "loss_yns_3": 0.16367, "loss_cls_4": 1.1229, "loss_box_4": 2.43716, "loss_cns_4": 0.61711, "loss_yns_4": 0.16891, "loss_cls_5": 1.11814, "loss_box_5": 2.46678, "loss_cns_5": 0.61765, "loss_yns_5": 0.16525, "loss_cls_dn_0": 0.38467, "loss_box_dn_0": 0.89891, "loss_cls_dn_1": 0.32635, "loss_box_dn_1": 1.0737, "loss_cls_dn_2": 0.35456, "loss_box_dn_2": 1.02782, "loss_cls_dn_3": 0.34904, "loss_box_dn_3": 1.01876, "loss_cls_dn_4": 0.34577, "loss_box_dn_4": 1.03065, "loss_cls_dn_5": 0.36735, "loss_box_dn_5": 1.05533, "loss_dense_depth": 0.9108, "loss": 34.71561, "grad_norm": 53.86457, "time": 1.57397} -{"mode": "train", "epoch": 1, "iter": 51, "lr": 0.00012, "memory": 49164, "data_time": 0.07352, "loss_cls_0": 1.00913, "loss_box_0": 1.95404, "loss_cns_0": 0.6197, "loss_yns_0": 0.16251, "loss_cls_1": 1.08377, "loss_box_1": 2.48689, "loss_cns_1": 0.60054, "loss_yns_1": 0.16407, "loss_cls_2": 1.08835, "loss_box_2": 2.42111, "loss_cns_2": 0.62035, "loss_yns_2": 0.16343, "loss_cls_3": 1.12249, "loss_box_3": 2.42045, "loss_cns_3": 0.62775, "loss_yns_3": 0.16621, "loss_cls_4": 1.11703, "loss_box_4": 2.45439, "loss_cns_4": 0.62935, "loss_yns_4": 0.16773, "loss_cls_5": 1.11694, "loss_box_5": 2.46039, "loss_cns_5": 0.62978, "loss_yns_5": 0.16578, "loss_cls_dn_0": 0.38908, "loss_box_dn_0": 0.88567, "loss_cls_dn_1": 0.31812, "loss_box_dn_1": 1.11167, "loss_cls_dn_2": 0.34499, "loss_box_dn_2": 1.04271, "loss_cls_dn_3": 0.33593, "loss_box_dn_3": 1.04651, "loss_cls_dn_4": 0.32947, "loss_box_dn_4": 1.0794, "loss_cls_dn_5": 0.34665, "loss_box_dn_5": 1.10482, "loss_dense_depth": 0.89673, "loss": 34.68391, "grad_norm": 52.29075, "time": 1.5666} -{"mode": "train", "epoch": 1, "iter": 52, "lr": 0.00012, "memory": 49164, "data_time": 0.07326, "loss_cls_0": 1.01396, "loss_box_0": 1.97355, "loss_cns_0": 0.61323, "loss_yns_0": 0.16496, "loss_cls_1": 1.09266, "loss_box_1": 2.35857, "loss_cns_1": 0.61636, "loss_yns_1": 0.1681, "loss_cls_2": 1.09962, "loss_box_2": 2.35153, "loss_cns_2": 0.62327, "loss_yns_2": 0.16476, "loss_cls_3": 1.12271, "loss_box_3": 2.38919, "loss_cns_3": 0.62058, "loss_yns_3": 0.16784, "loss_cls_4": 1.15608, "loss_box_4": 2.42601, "loss_cns_4": 0.62119, "loss_yns_4": 0.16707, "loss_cls_5": 1.15871, "loss_box_5": 2.41307, "loss_cns_5": 0.61976, "loss_yns_5": 0.1653, "loss_cls_dn_0": 0.40712, "loss_box_dn_0": 0.89134, "loss_cls_dn_1": 0.32088, "loss_box_dn_1": 1.1049, "loss_cls_dn_2": 0.34048, "loss_box_dn_2": 1.07876, "loss_cls_dn_3": 0.34347, "loss_box_dn_3": 1.12615, "loss_cls_dn_4": 0.33306, "loss_box_dn_4": 1.17763, "loss_cls_dn_5": 0.34595, "loss_box_dn_5": 1.21221, "loss_dense_depth": 0.9663, "loss": 34.91629, "grad_norm": 70.69684, "time": 1.58478} -{"mode": "train", "epoch": 1, "iter": 53, "lr": 0.00012, "memory": 49164, "data_time": 0.06487, "loss_cls_0": 1.04982, "loss_box_0": 1.99406, "loss_cns_0": 0.61592, "loss_yns_0": 0.16546, "loss_cls_1": 1.13865, "loss_box_1": 2.37095, "loss_cns_1": 0.61233, "loss_yns_1": 0.16926, "loss_cls_2": 1.14976, "loss_box_2": 2.39269, "loss_cns_2": 0.61301, "loss_yns_2": 0.17008, "loss_cls_3": 1.16074, "loss_box_3": 2.43574, "loss_cns_3": 0.60985, "loss_yns_3": 0.16811, "loss_cls_4": 1.17544, "loss_box_4": 2.43996, "loss_cns_4": 0.61517, "loss_yns_4": 0.16593, "loss_cls_5": 1.18163, "loss_box_5": 2.42346, "loss_cns_5": 0.61275, "loss_yns_5": 0.16482, "loss_cls_dn_0": 0.42434, "loss_box_dn_0": 0.88304, "loss_cls_dn_1": 0.32595, "loss_box_dn_1": 1.07669, "loss_cls_dn_2": 0.3377, "loss_box_dn_2": 1.0684, "loss_cls_dn_3": 0.35293, "loss_box_dn_3": 1.12908, "loss_cls_dn_4": 0.34152, "loss_box_dn_4": 1.17501, "loss_cls_dn_5": 0.35512, "loss_box_dn_5": 1.21516, "loss_dense_depth": 0.92497, "loss": 35.20549, "grad_norm": 67.05763, "time": 1.57418} -{"mode": "train", "epoch": 1, "iter": 54, "lr": 0.00012, "memory": 49164, "data_time": 0.07221, "loss_cls_0": 1.02204, "loss_box_0": 1.92379, "loss_cns_0": 0.62308, "loss_yns_0": 0.16335, "loss_cls_1": 1.13582, "loss_box_1": 2.28058, "loss_cns_1": 0.62335, "loss_yns_1": 0.16166, "loss_cls_2": 1.15593, "loss_box_2": 2.25552, "loss_cns_2": 0.6304, "loss_yns_2": 0.16207, "loss_cls_3": 1.14696, "loss_box_3": 2.27614, "loss_cns_3": 0.63349, "loss_yns_3": 0.16297, "loss_cls_4": 1.11639, "loss_box_4": 2.28825, "loss_cns_4": 0.62837, "loss_yns_4": 0.16187, "loss_cls_5": 1.12288, "loss_box_5": 2.29525, "loss_cns_5": 0.63152, "loss_yns_5": 0.16517, "loss_cls_dn_0": 0.41149, "loss_box_dn_0": 0.88127, "loss_cls_dn_1": 0.31863, "loss_box_dn_1": 1.09458, "loss_cls_dn_2": 0.32189, "loss_box_dn_2": 1.07669, "loss_cls_dn_3": 0.34864, "loss_box_dn_3": 1.12744, "loss_cls_dn_4": 0.34226, "loss_box_dn_4": 1.16718, "loss_cls_dn_5": 0.35761, "loss_box_dn_5": 1.21015, "loss_dense_depth": 0.94611, "loss": 34.37077, "grad_norm": 46.915, "time": 1.57789} -{"mode": "train", "epoch": 1, "iter": 55, "lr": 0.00012, "memory": 49164, "data_time": 0.07221, "loss_cls_0": 1.00861, "loss_box_0": 1.95431, "loss_cns_0": 0.61282, "loss_yns_0": 0.16463, "loss_cls_1": 1.06861, "loss_box_1": 2.23247, "loss_cns_1": 0.61627, "loss_yns_1": 0.16443, "loss_cls_2": 1.10114, "loss_box_2": 2.19238, "loss_cns_2": 0.62839, "loss_yns_2": 0.16658, "loss_cls_3": 1.10565, "loss_box_3": 2.22169, "loss_cns_3": 0.63485, "loss_yns_3": 0.16409, "loss_cls_4": 1.13046, "loss_box_4": 2.27045, "loss_cns_4": 0.62914, "loss_yns_4": 0.1625, "loss_cls_5": 1.12406, "loss_box_5": 2.30275, "loss_cns_5": 0.63484, "loss_yns_5": 0.1652, "loss_cls_dn_0": 0.41284, "loss_box_dn_0": 0.87476, "loss_cls_dn_1": 0.34017, "loss_box_dn_1": 0.97066, "loss_cls_dn_2": 0.3427, "loss_box_dn_2": 0.94545, "loss_cls_dn_3": 0.3737, "loss_box_dn_3": 0.98986, "loss_cls_dn_4": 0.3823, "loss_box_dn_4": 1.04881, "loss_cls_dn_5": 0.40361, "loss_box_dn_5": 1.0974, "loss_dense_depth": 1.06003, "loss": 33.69862, "grad_norm": 57.88338, "time": 1.63088} -{"mode": "train", "epoch": 1, "iter": 56, "lr": 0.00012, "memory": 49164, "data_time": 0.07152, "loss_cls_0": 0.98342, "loss_box_0": 1.91667, "loss_cns_0": 0.61371, "loss_yns_0": 0.16268, "loss_cls_1": 1.07184, "loss_box_1": 2.13766, "loss_cns_1": 0.61696, "loss_yns_1": 0.16429, "loss_cls_2": 1.13803, "loss_box_2": 2.10076, "loss_cns_2": 0.63177, "loss_yns_2": 0.17045, "loss_cls_3": 1.10409, "loss_box_3": 2.1154, "loss_cns_3": 0.6383, "loss_yns_3": 0.16702, "loss_cls_4": 1.15465, "loss_box_4": 2.15883, "loss_cns_4": 0.63426, "loss_yns_4": 0.16174, "loss_cls_5": 1.13411, "loss_box_5": 2.16366, "loss_cns_5": 0.64321, "loss_yns_5": 0.16339, "loss_cls_dn_0": 0.38595, "loss_box_dn_0": 0.87428, "loss_cls_dn_1": 0.32556, "loss_box_dn_1": 0.98673, "loss_cls_dn_2": 0.34272, "loss_box_dn_2": 0.95962, "loss_cls_dn_3": 0.3534, "loss_box_dn_3": 0.97702, "loss_cls_dn_4": 0.37404, "loss_box_dn_4": 1.01425, "loss_cls_dn_5": 0.38895, "loss_box_dn_5": 1.03219, "loss_dense_depth": 0.95009, "loss": 32.91169, "grad_norm": 65.96749, "time": 1.57007} -{"mode": "train", "epoch": 1, "iter": 57, "lr": 0.00012, "memory": 49164, "data_time": 0.09585, "loss_cls_0": 1.01946, "loss_box_0": 1.87478, "loss_cns_0": 0.61958, "loss_yns_0": 0.16356, "loss_cls_1": 1.13692, "loss_box_1": 2.20062, "loss_cns_1": 0.6142, "loss_yns_1": 0.17109, "loss_cls_2": 1.16648, "loss_box_2": 2.17873, "loss_cns_2": 0.62688, "loss_yns_2": 0.16997, "loss_cls_3": 1.1748, "loss_box_3": 2.17527, "loss_cns_3": 0.63118, "loss_yns_3": 0.17008, "loss_cls_4": 1.15614, "loss_box_4": 2.17185, "loss_cns_4": 0.63256, "loss_yns_4": 0.16415, "loss_cls_5": 1.15006, "loss_box_5": 2.17671, "loss_cns_5": 0.63288, "loss_yns_5": 0.16904, "loss_cls_dn_0": 0.36787, "loss_box_dn_0": 0.87086, "loss_cls_dn_1": 0.30903, "loss_box_dn_1": 0.94122, "loss_cls_dn_2": 0.3345, "loss_box_dn_2": 0.91708, "loss_cls_dn_3": 0.32772, "loss_box_dn_3": 0.91634, "loss_cls_dn_4": 0.35444, "loss_box_dn_4": 0.92321, "loss_cls_dn_5": 0.35928, "loss_box_dn_5": 0.9385, "loss_dense_depth": 1.02471, "loss": 32.93174, "grad_norm": 58.91538, "time": 1.63433} -{"mode": "train", "epoch": 1, "iter": 58, "lr": 0.00012, "memory": 49164, "data_time": 0.07472, "loss_cls_0": 1.01342, "loss_box_0": 1.86519, "loss_cns_0": 0.62017, "loss_yns_0": 0.16388, "loss_cls_1": 1.0776, "loss_box_1": 2.16625, "loss_cns_1": 0.61808, "loss_yns_1": 0.16969, "loss_cls_2": 1.0945, "loss_box_2": 2.1252, "loss_cns_2": 0.63185, "loss_yns_2": 0.17019, "loss_cls_3": 1.11047, "loss_box_3": 2.12947, "loss_cns_3": 0.63409, "loss_yns_3": 0.16792, "loss_cls_4": 1.08184, "loss_box_4": 2.13753, "loss_cns_4": 0.63591, "loss_yns_4": 0.16729, "loss_cls_5": 1.10833, "loss_box_5": 2.17217, "loss_cns_5": 0.63536, "loss_yns_5": 0.17046, "loss_cls_dn_0": 0.36098, "loss_box_dn_0": 0.86692, "loss_cls_dn_1": 0.29618, "loss_box_dn_1": 0.91816, "loss_cls_dn_2": 0.3354, "loss_box_dn_2": 0.89367, "loss_cls_dn_3": 0.32277, "loss_box_dn_3": 0.89582, "loss_cls_dn_4": 0.34735, "loss_box_dn_4": 0.90679, "loss_cls_dn_5": 0.34266, "loss_box_dn_5": 0.94387, "loss_dense_depth": 0.933, "loss": 32.23042, "grad_norm": 41.64589, "time": 1.56499} -{"mode": "train", "epoch": 1, "iter": 59, "lr": 0.00012, "memory": 49164, "data_time": 0.1228, "loss_cls_0": 0.97666, "loss_box_0": 1.89159, "loss_cns_0": 0.61895, "loss_yns_0": 0.16113, "loss_cls_1": 1.05635, "loss_box_1": 2.13633, "loss_cns_1": 0.63001, "loss_yns_1": 0.16666, "loss_cls_2": 1.08153, "loss_box_2": 2.12494, "loss_cns_2": 0.64046, "loss_yns_2": 0.17196, "loss_cls_3": 1.09226, "loss_box_3": 2.14788, "loss_cns_3": 0.64104, "loss_yns_3": 0.16715, "loss_cls_4": 1.09534, "loss_box_4": 2.18035, "loss_cns_4": 0.63968, "loss_yns_4": 0.16668, "loss_cls_5": 1.12633, "loss_box_5": 2.20326, "loss_cns_5": 0.6415, "loss_yns_5": 0.16707, "loss_cls_dn_0": 0.373, "loss_box_dn_0": 0.87056, "loss_cls_dn_1": 0.30196, "loss_box_dn_1": 0.92133, "loss_cls_dn_2": 0.35312, "loss_box_dn_2": 0.91769, "loss_cls_dn_3": 0.34928, "loss_box_dn_3": 0.93303, "loss_cls_dn_4": 0.36073, "loss_box_dn_4": 0.96227, "loss_cls_dn_5": 0.35963, "loss_box_dn_5": 1.0045, "loss_dense_depth": 0.99832, "loss": 32.63053, "grad_norm": 63.35992, "time": 1.63784} -{"mode": "train", "epoch": 1, "iter": 60, "lr": 0.00012, "memory": 49164, "data_time": 0.08123, "loss_cls_0": 0.9674, "loss_box_0": 1.86219, "loss_cns_0": 0.62113, "loss_yns_0": 0.15842, "loss_cls_1": 1.11332, "loss_box_1": 2.04806, "loss_cns_1": 0.63803, "loss_yns_1": 0.16848, "loss_cls_2": 1.13259, "loss_box_2": 2.0684, "loss_cns_2": 0.64247, "loss_yns_2": 0.16918, "loss_cls_3": 1.11901, "loss_box_3": 2.06177, "loss_cns_3": 0.64448, "loss_yns_3": 0.16528, "loss_cls_4": 1.14925, "loss_box_4": 2.0691, "loss_cns_4": 0.64668, "loss_yns_4": 0.16259, "loss_cls_5": 1.13063, "loss_box_5": 2.08679, "loss_cns_5": 0.64423, "loss_yns_5": 0.1629, "loss_cls_dn_0": 0.38805, "loss_box_dn_0": 0.8809, "loss_cls_dn_1": 0.30832, "loss_box_dn_1": 0.93305, "loss_cls_dn_2": 0.36326, "loss_box_dn_2": 0.9541, "loss_cls_dn_3": 0.36633, "loss_box_dn_3": 0.95935, "loss_cls_dn_4": 0.36413, "loss_box_dn_4": 0.98226, "loss_cls_dn_5": 0.37473, "loss_box_dn_5": 1.0235, "loss_dense_depth": 0.90909, "loss": 32.43944, "grad_norm": 63.22801, "time": 1.57746} -{"mode": "train", "epoch": 1, "iter": 61, "lr": 0.00012, "memory": 49164, "data_time": 0.20106, "loss_cls_0": 0.9745, "loss_box_0": 1.87004, "loss_cns_0": 0.6153, "loss_yns_0": 0.15786, "loss_cls_1": 1.11114, "loss_box_1": 2.10896, "loss_cns_1": 0.62613, "loss_yns_1": 0.16613, "loss_cls_2": 1.11809, "loss_box_2": 2.10684, "loss_cns_2": 0.63633, "loss_yns_2": 0.16497, "loss_cls_3": 1.09741, "loss_box_3": 2.07698, "loss_cns_3": 0.64105, "loss_yns_3": 0.1661, "loss_cls_4": 1.09254, "loss_box_4": 2.11493, "loss_cns_4": 0.64011, "loss_yns_4": 0.16143, "loss_cls_5": 1.09169, "loss_box_5": 2.12813, "loss_cns_5": 0.63669, "loss_yns_5": 0.16531, "loss_cls_dn_0": 0.39519, "loss_box_dn_0": 0.87517, "loss_cls_dn_1": 0.31531, "loss_box_dn_1": 0.98387, "loss_cls_dn_2": 0.36474, "loss_box_dn_2": 1.00101, "loss_cls_dn_3": 0.3657, "loss_box_dn_3": 0.99249, "loss_cls_dn_4": 0.35033, "loss_box_dn_4": 1.01895, "loss_cls_dn_5": 0.37613, "loss_box_dn_5": 1.04981, "loss_dense_depth": 1.03144, "loss": 32.7888, "grad_norm": 46.69709, "time": 1.78043} -{"mode": "train", "epoch": 1, "iter": 62, "lr": 0.00012, "memory": 49164, "data_time": 0.17378, "loss_cls_0": 0.98889, "loss_box_0": 1.88478, "loss_cns_0": 0.61713, "loss_yns_0": 0.15809, "loss_cls_1": 1.07648, "loss_box_1": 2.17903, "loss_cns_1": 0.62904, "loss_yns_1": 0.16291, "loss_cls_2": 1.09014, "loss_box_2": 2.17915, "loss_cns_2": 0.63927, "loss_yns_2": 0.16207, "loss_cls_3": 1.08502, "loss_box_3": 2.15602, "loss_cns_3": 0.64452, "loss_yns_3": 0.16421, "loss_cls_4": 1.09405, "loss_box_4": 2.24892, "loss_cns_4": 0.63893, "loss_yns_4": 0.16154, "loss_cls_5": 1.10325, "loss_box_5": 2.23997, "loss_cns_5": 0.63706, "loss_yns_5": 0.16432, "loss_cls_dn_0": 0.388, "loss_box_dn_0": 0.87271, "loss_cls_dn_1": 0.32685, "loss_box_dn_1": 0.9568, "loss_cls_dn_2": 0.37224, "loss_box_dn_2": 0.97112, "loss_cls_dn_3": 0.36686, "loss_box_dn_3": 0.96212, "loss_cls_dn_4": 0.34532, "loss_box_dn_4": 1.00929, "loss_cls_dn_5": 0.38094, "loss_box_dn_5": 1.02309, "loss_dense_depth": 0.93349, "loss": 33.01361, "grad_norm": 64.52093, "time": 1.66126} -{"mode": "train", "epoch": 1, "iter": 63, "lr": 0.00012, "memory": 49164, "data_time": 0.07004, "loss_cls_0": 1.00162, "loss_box_0": 1.89363, "loss_cns_0": 0.61618, "loss_yns_0": 0.16111, "loss_cls_1": 1.05817, "loss_box_1": 2.21499, "loss_cns_1": 0.62869, "loss_yns_1": 0.16236, "loss_cls_2": 1.07654, "loss_box_2": 2.20005, "loss_cns_2": 0.63989, "loss_yns_2": 0.16221, "loss_cls_3": 1.11977, "loss_box_3": 2.18143, "loss_cns_3": 0.64695, "loss_yns_3": 0.16265, "loss_cls_4": 1.16156, "loss_box_4": 2.24641, "loss_cns_4": 0.64374, "loss_yns_4": 0.16172, "loss_cls_5": 1.15284, "loss_box_5": 2.21318, "loss_cns_5": 0.64643, "loss_yns_5": 0.16351, "loss_cls_dn_0": 0.37258, "loss_box_dn_0": 0.87332, "loss_cls_dn_1": 0.30857, "loss_box_dn_1": 0.96098, "loss_cls_dn_2": 0.34126, "loss_box_dn_2": 0.95114, "loss_cls_dn_3": 0.33738, "loss_box_dn_3": 0.94207, "loss_cls_dn_4": 0.31963, "loss_box_dn_4": 0.97545, "loss_cls_dn_5": 0.34721, "loss_box_dn_5": 0.97572, "loss_dense_depth": 0.99121, "loss": 33.01216, "grad_norm": 63.93467, "time": 1.54376} -{"mode": "train", "epoch": 1, "iter": 64, "lr": 0.00013, "memory": 49164, "data_time": 0.07702, "loss_cls_0": 0.99435, "loss_box_0": 1.91514, "loss_cns_0": 0.61544, "loss_yns_0": 0.16061, "loss_cls_1": 1.07125, "loss_box_1": 2.25945, "loss_cns_1": 0.61885, "loss_yns_1": 0.16313, "loss_cls_2": 1.09455, "loss_box_2": 2.20551, "loss_cns_2": 0.63499, "loss_yns_2": 0.16464, "loss_cls_3": 1.09925, "loss_box_3": 2.20003, "loss_cns_3": 0.64385, "loss_yns_3": 0.16183, "loss_cls_4": 1.12847, "loss_box_4": 2.20083, "loss_cns_4": 0.64139, "loss_yns_4": 0.16206, "loss_cls_5": 1.13494, "loss_box_5": 2.2202, "loss_cns_5": 0.64402, "loss_yns_5": 0.16201, "loss_cls_dn_0": 0.35487, "loss_box_dn_0": 0.88056, "loss_cls_dn_1": 0.28147, "loss_box_dn_1": 0.96574, "loss_cls_dn_2": 0.30147, "loss_box_dn_2": 0.92135, "loss_cls_dn_3": 0.31445, "loss_box_dn_3": 0.91675, "loss_cls_dn_4": 0.3028, "loss_box_dn_4": 0.92554, "loss_cls_dn_5": 0.31842, "loss_box_dn_5": 0.93779, "loss_dense_depth": 0.95039, "loss": 32.66839, "grad_norm": 44.462, "time": 1.56388} -{"mode": "train", "epoch": 1, "iter": 65, "lr": 0.00013, "memory": 49164, "data_time": 0.10934, "loss_cls_0": 0.97711, "loss_box_0": 1.89347, "loss_cns_0": 0.61904, "loss_yns_0": 0.16153, "loss_cls_1": 1.04773, "loss_box_1": 2.20641, "loss_cns_1": 0.61889, "loss_yns_1": 0.16519, "loss_cls_2": 1.11867, "loss_box_2": 2.16882, "loss_cns_2": 0.63358, "loss_yns_2": 0.16721, "loss_cls_3": 1.07579, "loss_box_3": 2.18015, "loss_cns_3": 0.64087, "loss_yns_3": 0.16553, "loss_cls_4": 1.0778, "loss_box_4": 2.16002, "loss_cns_4": 0.64187, "loss_yns_4": 0.16512, "loss_cls_5": 1.08556, "loss_box_5": 2.18992, "loss_cns_5": 0.64297, "loss_yns_5": 0.16235, "loss_cls_dn_0": 0.36147, "loss_box_dn_0": 0.874, "loss_cls_dn_1": 0.26246, "loss_box_dn_1": 0.98099, "loss_cls_dn_2": 0.27904, "loss_box_dn_2": 0.9344, "loss_cls_dn_3": 0.30693, "loss_box_dn_3": 0.9389, "loss_cls_dn_4": 0.30426, "loss_box_dn_4": 0.94236, "loss_cls_dn_5": 0.31697, "loss_box_dn_5": 0.95988, "loss_dense_depth": 0.90331, "loss": 32.33054, "grad_norm": 57.91806, "time": 1.63713} -{"mode": "train", "epoch": 1, "iter": 66, "lr": 0.00013, "memory": 49164, "data_time": 0.07227, "loss_cls_0": 0.9558, "loss_box_0": 1.82245, "loss_cns_0": 0.62365, "loss_yns_0": 0.16166, "loss_cls_1": 1.03619, "loss_box_1": 2.12178, "loss_cns_1": 0.6289, "loss_yns_1": 0.16581, "loss_cls_2": 1.13731, "loss_box_2": 2.13299, "loss_cns_2": 0.63642, "loss_yns_2": 0.16415, "loss_cls_3": 1.10153, "loss_box_3": 2.14397, "loss_cns_3": 0.64398, "loss_yns_3": 0.16688, "loss_cls_4": 1.08171, "loss_box_4": 2.13994, "loss_cns_4": 0.64783, "loss_yns_4": 0.16707, "loss_cls_5": 1.09429, "loss_box_5": 2.14712, "loss_cns_5": 0.64759, "loss_yns_5": 0.16563, "loss_cls_dn_0": 0.36349, "loss_box_dn_0": 0.84598, "loss_cls_dn_1": 0.24581, "loss_box_dn_1": 0.98289, "loss_cls_dn_2": 0.26557, "loss_box_dn_2": 0.95619, "loss_cls_dn_3": 0.28916, "loss_box_dn_3": 0.96269, "loss_cls_dn_4": 0.30577, "loss_box_dn_4": 0.97533, "loss_cls_dn_5": 0.31946, "loss_box_dn_5": 0.9874, "loss_dense_depth": 0.94748, "loss": 32.18185, "grad_norm": 57.84132, "time": 1.58062} -{"mode": "train", "epoch": 1, "iter": 67, "lr": 0.00013, "memory": 49164, "data_time": 0.07165, "loss_cls_0": 0.96334, "loss_box_0": 1.85375, "loss_cns_0": 0.6175, "loss_yns_0": 0.16098, "loss_cls_1": 1.04087, "loss_box_1": 2.15784, "loss_cns_1": 0.62689, "loss_yns_1": 0.16452, "loss_cls_2": 1.07334, "loss_box_2": 2.15161, "loss_cns_2": 0.63488, "loss_yns_2": 0.16295, "loss_cls_3": 1.10959, "loss_box_3": 2.15972, "loss_cns_3": 0.64178, "loss_yns_3": 0.16709, "loss_cls_4": 1.08226, "loss_box_4": 2.14951, "loss_cns_4": 0.64653, "loss_yns_4": 0.16532, "loss_cls_5": 1.09385, "loss_box_5": 2.17907, "loss_cns_5": 0.64602, "loss_yns_5": 0.16525, "loss_cls_dn_0": 0.38066, "loss_box_dn_0": 0.84922, "loss_cls_dn_1": 0.25428, "loss_box_dn_1": 0.9646, "loss_cls_dn_2": 0.2851, "loss_box_dn_2": 0.95176, "loss_cls_dn_3": 0.29336, "loss_box_dn_3": 0.96669, "loss_cls_dn_4": 0.32625, "loss_box_dn_4": 0.98514, "loss_cls_dn_5": 0.3444, "loss_box_dn_5": 1.01416, "loss_dense_depth": 0.89468, "loss": 32.32473, "grad_norm": 58.93967, "time": 1.5857} -{"mode": "train", "epoch": 1, "iter": 68, "lr": 0.00013, "memory": 49164, "data_time": 0.07352, "loss_cls_0": 0.96058, "loss_box_0": 1.88429, "loss_cns_0": 0.61977, "loss_yns_0": 0.16097, "loss_cls_1": 1.04423, "loss_box_1": 2.12915, "loss_cns_1": 0.6252, "loss_yns_1": 0.16361, "loss_cls_2": 1.06333, "loss_box_2": 2.06322, "loss_cns_2": 0.64156, "loss_yns_2": 0.16444, "loss_cls_3": 1.08535, "loss_box_3": 2.08441, "loss_cns_3": 0.64648, "loss_yns_3": 0.16532, "loss_cls_4": 1.08389, "loss_box_4": 2.06716, "loss_cns_4": 0.6471, "loss_yns_4": 0.16426, "loss_cls_5": 1.0823, "loss_box_5": 2.11182, "loss_cns_5": 0.64425, "loss_yns_5": 0.16585, "loss_cls_dn_0": 0.38593, "loss_box_dn_0": 0.86074, "loss_cls_dn_1": 0.26978, "loss_box_dn_1": 0.95138, "loss_cls_dn_2": 0.32078, "loss_box_dn_2": 0.93522, "loss_cls_dn_3": 0.31566, "loss_box_dn_3": 0.96482, "loss_cls_dn_4": 0.35463, "loss_box_dn_4": 0.98603, "loss_cls_dn_5": 0.36976, "loss_box_dn_5": 1.02984, "loss_dense_depth": 0.91945, "loss": 32.13257, "grad_norm": 53.98926, "time": 1.59946} -{"mode": "train", "epoch": 1, "iter": 69, "lr": 0.00013, "memory": 49164, "data_time": 0.07356, "loss_cls_0": 0.93506, "loss_box_0": 1.87212, "loss_cns_0": 0.62645, "loss_yns_0": 0.15879, "loss_cls_1": 1.03058, "loss_box_1": 2.10125, "loss_cns_1": 0.63073, "loss_yns_1": 0.15948, "loss_cls_2": 1.07053, "loss_box_2": 2.04559, "loss_cns_2": 0.64751, "loss_yns_2": 0.16787, "loss_cls_3": 1.07975, "loss_box_3": 2.05434, "loss_cns_3": 0.6514, "loss_yns_3": 0.16168, "loss_cls_4": 1.0786, "loss_box_4": 2.0767, "loss_cns_4": 0.64862, "loss_yns_4": 0.16161, "loss_cls_5": 1.06923, "loss_box_5": 2.0737, "loss_cns_5": 0.64691, "loss_yns_5": 0.16225, "loss_cls_dn_0": 0.36409, "loss_box_dn_0": 0.8529, "loss_cls_dn_1": 0.26751, "loss_box_dn_1": 0.95737, "loss_cls_dn_2": 0.33436, "loss_box_dn_2": 0.9447, "loss_cls_dn_3": 0.33263, "loss_box_dn_3": 0.9788, "loss_cls_dn_4": 0.35549, "loss_box_dn_4": 1.01578, "loss_cls_dn_5": 0.36517, "loss_box_dn_5": 1.04233, "loss_dense_depth": 0.84161, "loss": 31.9635, "grad_norm": 58.45697, "time": 1.55414} -{"mode": "train", "epoch": 1, "iter": 70, "lr": 0.00013, "memory": 49164, "data_time": 0.07647, "loss_cls_0": 0.95054, "loss_box_0": 1.87669, "loss_cns_0": 0.62429, "loss_yns_0": 0.15922, "loss_cls_1": 1.01341, "loss_box_1": 2.10661, "loss_cns_1": 0.62349, "loss_yns_1": 0.15938, "loss_cls_2": 1.06296, "loss_box_2": 2.04041, "loss_cns_2": 0.64203, "loss_yns_2": 0.16388, "loss_cls_3": 1.07774, "loss_box_3": 2.00734, "loss_cns_3": 0.64824, "loss_yns_3": 0.16011, "loss_cls_4": 1.0712, "loss_box_4": 2.04751, "loss_cns_4": 0.64481, "loss_yns_4": 0.1588, "loss_cls_5": 1.08385, "loss_box_5": 2.00905, "loss_cns_5": 0.64809, "loss_yns_5": 0.15976, "loss_cls_dn_0": 0.35796, "loss_box_dn_0": 0.85359, "loss_cls_dn_1": 0.27459, "loss_box_dn_1": 0.96011, "loss_cls_dn_2": 0.34396, "loss_box_dn_2": 0.94732, "loss_cls_dn_3": 0.34602, "loss_box_dn_3": 0.96475, "loss_cls_dn_4": 0.34491, "loss_box_dn_4": 1.0081, "loss_cls_dn_5": 0.35669, "loss_box_dn_5": 1.01623, "loss_dense_depth": 0.90008, "loss": 31.81372, "grad_norm": 70.15157, "time": 1.58086} -{"mode": "train", "epoch": 1, "iter": 71, "lr": 0.00013, "memory": 49164, "data_time": 0.0728, "loss_cls_0": 0.9604, "loss_box_0": 1.84682, "loss_cns_0": 0.62434, "loss_yns_0": 0.15777, "loss_cls_1": 1.00699, "loss_box_1": 2.12966, "loss_cns_1": 0.61887, "loss_yns_1": 0.15986, "loss_cls_2": 1.04133, "loss_box_2": 2.0274, "loss_cns_2": 0.64211, "loss_yns_2": 0.15903, "loss_cls_3": 1.07374, "loss_box_3": 2.00983, "loss_cns_3": 0.64881, "loss_yns_3": 0.16048, "loss_cls_4": 1.10368, "loss_box_4": 2.00681, "loss_cns_4": 0.64853, "loss_yns_4": 0.15631, "loss_cls_5": 1.14127, "loss_box_5": 2.00179, "loss_cns_5": 0.64844, "loss_yns_5": 0.15974, "loss_cls_dn_0": 0.33756, "loss_box_dn_0": 0.85347, "loss_cls_dn_1": 0.25931, "loss_box_dn_1": 0.9735, "loss_cls_dn_2": 0.31666, "loss_box_dn_2": 0.93546, "loss_cls_dn_3": 0.31498, "loss_box_dn_3": 0.93724, "loss_cls_dn_4": 0.30042, "loss_box_dn_4": 0.9518, "loss_cls_dn_5": 0.31632, "loss_box_dn_5": 0.96133, "loss_dense_depth": 0.87322, "loss": 31.46529, "grad_norm": 46.70491, "time": 1.56769} -{"mode": "train", "epoch": 1, "iter": 72, "lr": 0.00013, "memory": 49164, "data_time": 0.07346, "loss_cls_0": 0.96233, "loss_box_0": 1.81597, "loss_cns_0": 0.62252, "loss_yns_0": 0.16024, "loss_cls_1": 1.05614, "loss_box_1": 2.02946, "loss_cns_1": 0.62638, "loss_yns_1": 0.16019, "loss_cls_2": 1.07387, "loss_box_2": 1.95775, "loss_cns_2": 0.64301, "loss_yns_2": 0.16178, "loss_cls_3": 1.06627, "loss_box_3": 1.98524, "loss_cns_3": 0.65074, "loss_yns_3": 0.1624, "loss_cls_4": 1.11339, "loss_box_4": 1.96472, "loss_cns_4": 0.65252, "loss_yns_4": 0.1578, "loss_cls_5": 1.09643, "loss_box_5": 1.98033, "loss_cns_5": 0.64771, "loss_yns_5": 0.16285, "loss_cls_dn_0": 0.34706, "loss_box_dn_0": 0.85081, "loss_cls_dn_1": 0.25739, "loss_box_dn_1": 0.85683, "loss_cls_dn_2": 0.30034, "loss_box_dn_2": 0.81797, "loss_cls_dn_3": 0.30912, "loss_box_dn_3": 0.8306, "loss_cls_dn_4": 0.29789, "loss_box_dn_4": 0.82899, "loss_cls_dn_5": 0.32581, "loss_box_dn_5": 0.83908, "loss_dense_depth": 0.90457, "loss": 30.6765, "grad_norm": 52.2715, "time": 1.60961} -{"mode": "train", "epoch": 1, "iter": 73, "lr": 0.00013, "memory": 49164, "data_time": 0.07116, "loss_cls_0": 0.95831, "loss_box_0": 1.83963, "loss_cns_0": 0.61571, "loss_yns_0": 0.15883, "loss_cls_1": 1.05165, "loss_box_1": 2.05, "loss_cns_1": 0.62436, "loss_yns_1": 0.16051, "loss_cls_2": 1.11067, "loss_box_2": 2.00328, "loss_cns_2": 0.64185, "loss_yns_2": 0.16172, "loss_cls_3": 1.08445, "loss_box_3": 2.01584, "loss_cns_3": 0.64714, "loss_yns_3": 0.16268, "loss_cls_4": 1.07967, "loss_box_4": 2.01308, "loss_cns_4": 0.64905, "loss_yns_4": 0.16144, "loss_cls_5": 1.07602, "loss_box_5": 2.02534, "loss_cns_5": 0.6439, "loss_yns_5": 0.16736, "loss_cls_dn_0": 0.36694, "loss_box_dn_0": 0.86074, "loss_cls_dn_1": 0.25327, "loss_box_dn_1": 0.85691, "loss_cls_dn_2": 0.28974, "loss_box_dn_2": 0.82783, "loss_cls_dn_3": 0.31466, "loss_box_dn_3": 0.83508, "loss_cls_dn_4": 0.31529, "loss_box_dn_4": 0.83843, "loss_cls_dn_5": 0.34362, "loss_box_dn_5": 0.85314, "loss_dense_depth": 0.86983, "loss": 30.92796, "grad_norm": 59.18422, "time": 1.5587} -{"mode": "train", "epoch": 1, "iter": 74, "lr": 0.00013, "memory": 49164, "data_time": 0.07758, "loss_cls_0": 0.94463, "loss_box_0": 1.82495, "loss_cns_0": 0.61755, "loss_yns_0": 0.15959, "loss_cls_1": 1.00605, "loss_box_1": 2.07375, "loss_cns_1": 0.6237, "loss_yns_1": 0.16237, "loss_cls_2": 1.06543, "loss_box_2": 2.02354, "loss_cns_2": 0.64409, "loss_yns_2": 0.1674, "loss_cls_3": 1.07911, "loss_box_3": 2.01748, "loss_cns_3": 0.65009, "loss_yns_3": 0.1651, "loss_cls_4": 1.06459, "loss_box_4": 2.00298, "loss_cns_4": 0.65008, "loss_yns_4": 0.16336, "loss_cls_5": 1.07799, "loss_box_5": 2.04699, "loss_cns_5": 0.64688, "loss_yns_5": 0.166, "loss_cls_dn_0": 0.37767, "loss_box_dn_0": 0.85445, "loss_cls_dn_1": 0.24879, "loss_box_dn_1": 0.87193, "loss_cls_dn_2": 0.28261, "loss_box_dn_2": 0.8436, "loss_cls_dn_3": 0.3206, "loss_box_dn_3": 0.84836, "loss_cls_dn_4": 0.3274, "loss_box_dn_4": 0.85351, "loss_cls_dn_5": 0.35448, "loss_box_dn_5": 0.88767, "loss_dense_depth": 0.88722, "loss": 31.00199, "grad_norm": 50.5578, "time": 1.57773} -{"mode": "train", "epoch": 1, "iter": 75, "lr": 0.00013, "memory": 49164, "data_time": 0.0925, "loss_cls_0": 0.94512, "loss_box_0": 1.84629, "loss_cns_0": 0.61532, "loss_yns_0": 0.16176, "loss_cls_1": 1.03113, "loss_box_1": 2.07403, "loss_cns_1": 0.62682, "loss_yns_1": 0.16549, "loss_cls_2": 1.09357, "loss_box_2": 2.02166, "loss_cns_2": 0.64221, "loss_yns_2": 0.16817, "loss_cls_3": 1.05209, "loss_box_3": 2.02201, "loss_cns_3": 0.65007, "loss_yns_3": 0.16764, "loss_cls_4": 1.06536, "loss_box_4": 2.02043, "loss_cns_4": 0.64808, "loss_yns_4": 0.16366, "loss_cls_5": 1.06875, "loss_box_5": 2.06618, "loss_cns_5": 0.64916, "loss_yns_5": 0.16335, "loss_cls_dn_0": 0.36811, "loss_box_dn_0": 0.84321, "loss_cls_dn_1": 0.23915, "loss_box_dn_1": 0.87441, "loss_cls_dn_2": 0.27623, "loss_box_dn_2": 0.86337, "loss_cls_dn_3": 0.30062, "loss_box_dn_3": 0.8739, "loss_cls_dn_4": 0.31072, "loss_box_dn_4": 0.89287, "loss_cls_dn_5": 0.34225, "loss_box_dn_5": 0.93138, "loss_dense_depth": 0.85328, "loss": 31.09783, "grad_norm": 39.20927, "time": 1.6013} -{"mode": "train", "epoch": 1, "iter": 76, "lr": 0.00013, "memory": 49164, "data_time": 0.07543, "loss_cls_0": 0.97107, "loss_box_0": 1.84661, "loss_cns_0": 0.62082, "loss_yns_0": 0.16329, "loss_cls_1": 1.08566, "loss_box_1": 2.1502, "loss_cns_1": 0.62736, "loss_yns_1": 0.16441, "loss_cls_2": 1.10239, "loss_box_2": 2.11258, "loss_cns_2": 0.64288, "loss_yns_2": 0.16484, "loss_cls_3": 1.09567, "loss_box_3": 2.08634, "loss_cns_3": 0.65334, "loss_yns_3": 0.17218, "loss_cls_4": 1.11128, "loss_box_4": 2.12414, "loss_cns_4": 0.65219, "loss_yns_4": 0.16547, "loss_cls_5": 1.08442, "loss_box_5": 2.10807, "loss_cns_5": 0.65074, "loss_yns_5": 0.17093, "loss_cls_dn_0": 0.34752, "loss_box_dn_0": 0.84546, "loss_cls_dn_1": 0.24121, "loss_box_dn_1": 0.92305, "loss_cls_dn_2": 0.28251, "loss_box_dn_2": 0.92439, "loss_cls_dn_3": 0.28418, "loss_box_dn_3": 0.92792, "loss_cls_dn_4": 0.30042, "loss_box_dn_4": 0.97265, "loss_cls_dn_5": 0.32397, "loss_box_dn_5": 0.99425, "loss_dense_depth": 0.90838, "loss": 32.00274, "grad_norm": 60.31474, "time": 1.58226} -{"mode": "train", "epoch": 1, "iter": 77, "lr": 0.00013, "memory": 49164, "data_time": 0.07296, "loss_cls_0": 0.9757, "loss_box_0": 1.87425, "loss_cns_0": 0.61756, "loss_yns_0": 0.16164, "loss_cls_1": 1.07535, "loss_box_1": 2.17387, "loss_cns_1": 0.63286, "loss_yns_1": 0.16159, "loss_cls_2": 1.08884, "loss_box_2": 2.13727, "loss_cns_2": 0.6415, "loss_yns_2": 0.16386, "loss_cls_3": 1.05683, "loss_box_3": 2.09923, "loss_cns_3": 0.6508, "loss_yns_3": 0.16828, "loss_cls_4": 1.06402, "loss_box_4": 2.13914, "loss_cns_4": 0.65138, "loss_yns_4": 0.1648, "loss_cls_5": 1.06969, "loss_box_5": 2.12236, "loss_cns_5": 0.64718, "loss_yns_5": 0.17282, "loss_cls_dn_0": 0.33843, "loss_box_dn_0": 0.83273, "loss_cls_dn_1": 0.24502, "loss_box_dn_1": 0.94421, "loss_cls_dn_2": 0.29557, "loss_box_dn_2": 0.93622, "loss_cls_dn_3": 0.28593, "loss_box_dn_3": 0.92465, "loss_cls_dn_4": 0.31705, "loss_box_dn_4": 0.96939, "loss_cls_dn_5": 0.3226, "loss_box_dn_5": 0.99386, "loss_dense_depth": 0.88573, "loss": 32.00224, "grad_norm": 48.45794, "time": 1.58241} -{"mode": "train", "epoch": 1, "iter": 78, "lr": 0.00013, "memory": 49164, "data_time": 0.07391, "loss_cls_0": 0.96995, "loss_box_0": 1.90073, "loss_cns_0": 0.61973, "loss_yns_0": 0.16193, "loss_cls_1": 1.04643, "loss_box_1": 2.18975, "loss_cns_1": 0.62364, "loss_yns_1": 0.16271, "loss_cls_2": 1.10917, "loss_box_2": 2.10868, "loss_cns_2": 0.63515, "loss_yns_2": 0.1752, "loss_cls_3": 1.05853, "loss_box_3": 2.11122, "loss_cns_3": 0.64369, "loss_yns_3": 0.16395, "loss_cls_4": 1.07084, "loss_box_4": 2.06963, "loss_cns_4": 0.64862, "loss_yns_4": 0.16397, "loss_cls_5": 1.05984, "loss_box_5": 2.06295, "loss_cns_5": 0.64908, "loss_yns_5": 0.16581, "loss_cls_dn_0": 0.34905, "loss_box_dn_0": 0.83302, "loss_cls_dn_1": 0.23376, "loss_box_dn_1": 0.96771, "loss_cls_dn_2": 0.29879, "loss_box_dn_2": 0.92529, "loss_cls_dn_3": 0.28602, "loss_box_dn_3": 0.90953, "loss_cls_dn_4": 0.31961, "loss_box_dn_4": 0.91546, "loss_cls_dn_5": 0.31208, "loss_box_dn_5": 0.93691, "loss_dense_depth": 0.91086, "loss": 31.76929, "grad_norm": 38.98866, "time": 1.61467} -{"mode": "train", "epoch": 1, "iter": 79, "lr": 0.00013, "memory": 49164, "data_time": 0.07435, "loss_cls_0": 0.94198, "loss_box_0": 1.87068, "loss_cns_0": 0.62399, "loss_yns_0": 0.1599, "loss_cls_1": 1.00869, "loss_box_1": 2.09606, "loss_cns_1": 0.63419, "loss_yns_1": 0.16401, "loss_cls_2": 1.05606, "loss_box_2": 2.0539, "loss_cns_2": 0.64543, "loss_yns_2": 0.1732, "loss_cls_3": 1.06381, "loss_box_3": 2.1108, "loss_cns_3": 0.64748, "loss_yns_3": 0.16749, "loss_cls_4": 1.08467, "loss_box_4": 2.06216, "loss_cns_4": 0.6529, "loss_yns_4": 0.16627, "loss_cls_5": 1.10502, "loss_box_5": 2.02336, "loss_cns_5": 0.65376, "loss_yns_5": 0.16178, "loss_cls_dn_0": 0.34232, "loss_box_dn_0": 0.83899, "loss_cls_dn_1": 0.21758, "loss_box_dn_1": 0.89291, "loss_cls_dn_2": 0.27596, "loss_box_dn_2": 0.86737, "loss_cls_dn_3": 0.2643, "loss_box_dn_3": 0.8718, "loss_cls_dn_4": 0.28318, "loss_box_dn_4": 0.86499, "loss_cls_dn_5": 0.29084, "loss_box_dn_5": 0.86692, "loss_dense_depth": 0.82181, "loss": 31.02655, "grad_norm": 53.3586, "time": 1.56646} -{"mode": "train", "epoch": 1, "iter": 80, "lr": 0.00013, "memory": 49164, "data_time": 0.08298, "loss_cls_0": 0.94325, "loss_box_0": 1.85456, "loss_cns_0": 0.62454, "loss_yns_0": 0.1589, "loss_cls_1": 1.04522, "loss_box_1": 2.08605, "loss_cns_1": 0.63182, "loss_yns_1": 0.16272, "loss_cls_2": 1.04231, "loss_box_2": 2.03105, "loss_cns_2": 0.6458, "loss_yns_2": 0.16527, "loss_cls_3": 1.08802, "loss_box_3": 2.06063, "loss_cns_3": 0.65208, "loss_yns_3": 0.17146, "loss_cls_4": 1.12573, "loss_box_4": 2.02613, "loss_cns_4": 0.65548, "loss_yns_4": 0.16361, "loss_cls_5": 1.08857, "loss_box_5": 2.00137, "loss_cns_5": 0.65192, "loss_yns_5": 0.16288, "loss_cls_dn_0": 0.34899, "loss_box_dn_0": 0.83142, "loss_cls_dn_1": 0.2136, "loss_box_dn_1": 0.87349, "loss_cls_dn_2": 0.25258, "loss_box_dn_2": 0.84606, "loss_cls_dn_3": 0.24657, "loss_box_dn_3": 0.85021, "loss_cls_dn_4": 0.25873, "loss_box_dn_4": 0.85037, "loss_cls_dn_5": 0.28542, "loss_box_dn_5": 0.8476, "loss_dense_depth": 0.85889, "loss": 30.8033, "grad_norm": 44.88047, "time": 1.58941} -{"mode": "train", "epoch": 1, "iter": 81, "lr": 0.00013, "memory": 49164, "data_time": 0.11373, "loss_cls_0": 0.95891, "loss_box_0": 1.86867, "loss_cns_0": 0.61613, "loss_yns_0": 0.16023, "loss_cls_1": 1.01935, "loss_box_1": 2.04714, "loss_cns_1": 0.62476, "loss_yns_1": 0.1589, "loss_cls_2": 1.07337, "loss_box_2": 1.98993, "loss_cns_2": 0.6456, "loss_yns_2": 0.16288, "loss_cls_3": 1.05333, "loss_box_3": 1.9893, "loss_cns_3": 0.6514, "loss_yns_3": 0.1647, "loss_cls_4": 1.04832, "loss_box_4": 1.96741, "loss_cns_4": 0.6523, "loss_yns_4": 0.15985, "loss_cls_5": 1.05667, "loss_box_5": 1.9964, "loss_cns_5": 0.65101, "loss_yns_5": 0.16152, "loss_cls_dn_0": 0.36065, "loss_box_dn_0": 0.83504, "loss_cls_dn_1": 0.2116, "loss_box_dn_1": 0.87854, "loss_cls_dn_2": 0.23902, "loss_box_dn_2": 0.85538, "loss_cls_dn_3": 0.24759, "loss_box_dn_3": 0.85932, "loss_cls_dn_4": 0.26987, "loss_box_dn_4": 0.86739, "loss_cls_dn_5": 0.3165, "loss_box_dn_5": 0.88329, "loss_dense_depth": 0.85333, "loss": 30.55563, "grad_norm": 40.48567, "time": 1.69449} -{"mode": "train", "epoch": 1, "iter": 82, "lr": 0.00013, "memory": 49164, "data_time": 0.16974, "loss_cls_0": 0.95981, "loss_box_0": 1.86851, "loss_cns_0": 0.6226, "loss_yns_0": 0.15619, "loss_cls_1": 1.01618, "loss_box_1": 2.01097, "loss_cns_1": 0.60206, "loss_yns_1": 0.15224, "loss_cls_2": 1.06319, "loss_box_2": 2.04636, "loss_cns_2": 0.64621, "loss_yns_2": 0.16615, "loss_cls_3": 1.06958, "loss_box_3": 2.04069, "loss_cns_3": 0.64856, "loss_yns_3": 0.16021, "loss_cls_4": 1.0709, "loss_box_4": 2.046, "loss_cns_4": 0.64629, "loss_yns_4": 0.15996, "loss_cls_5": 1.0911, "loss_box_5": 2.07346, "loss_cns_5": 0.64549, "loss_yns_5": 0.15811, "loss_cls_dn_0": 0.35455, "loss_box_dn_0": 0.83092, "loss_cls_dn_1": 0.21065, "loss_box_dn_1": 0.88261, "loss_cls_dn_2": 0.23556, "loss_box_dn_2": 0.87107, "loss_cls_dn_3": 0.25787, "loss_box_dn_3": 0.87865, "loss_cls_dn_4": 0.27112, "loss_box_dn_4": 0.8943, "loss_cls_dn_5": 0.32463, "loss_box_dn_5": 0.91508, "loss_dense_depth": 0.89302, "loss": 30.94087, "grad_norm": 47.43293, "time": 1.65158} -{"mode": "train", "epoch": 1, "iter": 83, "lr": 0.00013, "memory": 49164, "data_time": 0.0927, "loss_cls_0": 0.96303, "loss_box_0": 1.82496, "loss_cns_0": 0.62813, "loss_yns_0": 0.15891, "loss_cls_1": 1.0092, "loss_box_1": 1.93279, "loss_cns_1": 0.59488, "loss_yns_1": 0.15745, "loss_cls_2": 1.05348, "loss_box_2": 1.98996, "loss_cns_2": 0.63786, "loss_yns_2": 0.1673, "loss_cls_3": 1.06998, "loss_box_3": 1.98442, "loss_cns_3": 0.64336, "loss_yns_3": 0.16235, "loss_cls_4": 1.09714, "loss_box_4": 1.98478, "loss_cns_4": 0.64195, "loss_yns_4": 0.16187, "loss_cls_5": 1.08178, "loss_box_5": 1.99324, "loss_cns_5": 0.63987, "loss_yns_5": 0.16097, "loss_cls_dn_0": 0.33684, "loss_box_dn_0": 0.82499, "loss_cls_dn_1": 0.21439, "loss_box_dn_1": 0.87499, "loss_cls_dn_2": 0.23706, "loss_box_dn_2": 0.8671, "loss_cls_dn_3": 0.25266, "loss_box_dn_3": 0.87757, "loss_cls_dn_4": 0.253, "loss_box_dn_4": 0.88766, "loss_cls_dn_5": 0.2952, "loss_box_dn_5": 0.90505, "loss_dense_depth": 0.86975, "loss": 30.43589, "grad_norm": 47.83202, "time": 1.57015} -{"mode": "train", "epoch": 1, "iter": 84, "lr": 0.00013, "memory": 49164, "data_time": 0.0748, "loss_cls_0": 0.92498, "loss_box_0": 1.81925, "loss_cns_0": 0.62859, "loss_yns_0": 0.15731, "loss_cls_1": 0.96217, "loss_box_1": 1.97851, "loss_cns_1": 0.61503, "loss_yns_1": 0.15792, "loss_cls_2": 1.01245, "loss_box_2": 1.95053, "loss_cns_2": 0.64544, "loss_yns_2": 0.16112, "loss_cls_3": 1.02659, "loss_box_3": 1.95648, "loss_cns_3": 0.65109, "loss_yns_3": 0.16101, "loss_cls_4": 1.04967, "loss_box_4": 1.9524, "loss_cns_4": 0.65164, "loss_yns_4": 0.15934, "loss_cls_5": 1.07721, "loss_box_5": 1.95533, "loss_cns_5": 0.64957, "loss_yns_5": 0.1586, "loss_cls_dn_0": 0.31814, "loss_box_dn_0": 0.81852, "loss_cls_dn_1": 0.21419, "loss_box_dn_1": 0.84744, "loss_cls_dn_2": 0.23692, "loss_box_dn_2": 0.82355, "loss_cls_dn_3": 0.24834, "loss_box_dn_3": 0.84316, "loss_cls_dn_4": 0.24449, "loss_box_dn_4": 0.85435, "loss_cls_dn_5": 0.27193, "loss_box_dn_5": 0.86565, "loss_dense_depth": 0.80248, "loss": 29.85139, "grad_norm": 36.92444, "time": 1.59825} -{"mode": "train", "epoch": 1, "iter": 85, "lr": 0.00013, "memory": 49164, "data_time": 0.07373, "loss_cls_0": 0.95036, "loss_box_0": 1.83868, "loss_cns_0": 0.62384, "loss_yns_0": 0.15813, "loss_cls_1": 0.99202, "loss_box_1": 1.98904, "loss_cns_1": 0.61554, "loss_yns_1": 0.15905, "loss_cls_2": 1.03821, "loss_box_2": 2.00328, "loss_cns_2": 0.64062, "loss_yns_2": 0.16055, "loss_cls_3": 1.04164, "loss_box_3": 2.00911, "loss_cns_3": 0.64264, "loss_yns_3": 0.16282, "loss_cls_4": 1.05746, "loss_box_4": 2.0199, "loss_cns_4": 0.64217, "loss_yns_4": 0.16122, "loss_cls_5": 1.05366, "loss_box_5": 2.00433, "loss_cns_5": 0.64245, "loss_yns_5": 0.16206, "loss_cls_dn_0": 0.34859, "loss_box_dn_0": 0.83564, "loss_cls_dn_1": 0.2135, "loss_box_dn_1": 0.85033, "loss_cls_dn_2": 0.24421, "loss_box_dn_2": 0.82664, "loss_cls_dn_3": 0.24849, "loss_box_dn_3": 0.84496, "loss_cls_dn_4": 0.25211, "loss_box_dn_4": 0.86399, "loss_cls_dn_5": 0.27213, "loss_box_dn_5": 0.8623, "loss_dense_depth": 0.8646, "loss": 30.29627, "grad_norm": 56.87868, "time": 1.66254} -{"mode": "train", "epoch": 1, "iter": 86, "lr": 0.00013, "memory": 49164, "data_time": 0.0762, "loss_cls_0": 0.92447, "loss_box_0": 1.80759, "loss_cns_0": 0.62225, "loss_yns_0": 0.16055, "loss_cls_1": 0.98714, "loss_box_1": 1.95583, "loss_cns_1": 0.6239, "loss_yns_1": 0.16247, "loss_cls_2": 1.02911, "loss_box_2": 1.9087, "loss_cns_2": 0.64277, "loss_yns_2": 0.16175, "loss_cls_3": 1.02886, "loss_box_3": 1.90289, "loss_cns_3": 0.64752, "loss_yns_3": 0.16135, "loss_cls_4": 1.07488, "loss_box_4": 1.90232, "loss_cns_4": 0.64739, "loss_yns_4": 0.16298, "loss_cls_5": 1.06201, "loss_box_5": 1.87775, "loss_cns_5": 0.6475, "loss_yns_5": 0.16418, "loss_cls_dn_0": 0.34051, "loss_box_dn_0": 0.82725, "loss_cls_dn_1": 0.20797, "loss_box_dn_1": 0.82732, "loss_cls_dn_2": 0.23627, "loss_box_dn_2": 0.80155, "loss_cls_dn_3": 0.23775, "loss_box_dn_3": 0.80148, "loss_cls_dn_4": 0.24835, "loss_box_dn_4": 0.81739, "loss_cls_dn_5": 0.26457, "loss_box_dn_5": 0.81249, "loss_dense_depth": 0.85086, "loss": 29.53992, "grad_norm": 53.45008, "time": 1.61493} -{"mode": "train", "epoch": 1, "iter": 87, "lr": 0.00013, "memory": 49164, "data_time": 0.08432, "loss_cls_0": 0.90067, "loss_box_0": 1.80331, "loss_cns_0": 0.61742, "loss_yns_0": 0.15837, "loss_cls_1": 0.98657, "loss_box_1": 2.00123, "loss_cns_1": 0.62352, "loss_yns_1": 0.16669, "loss_cls_2": 1.00505, "loss_box_2": 1.92971, "loss_cns_2": 0.64363, "loss_yns_2": 0.16815, "loss_cls_3": 1.03917, "loss_box_3": 1.91003, "loss_cns_3": 0.64977, "loss_yns_3": 0.16143, "loss_cls_4": 1.04179, "loss_box_4": 1.89155, "loss_cns_4": 0.65057, "loss_yns_4": 0.16484, "loss_cls_5": 1.03726, "loss_box_5": 1.90314, "loss_cns_5": 0.64864, "loss_yns_5": 0.16291, "loss_cls_dn_0": 0.32649, "loss_box_dn_0": 0.82048, "loss_cls_dn_1": 0.20595, "loss_box_dn_1": 0.81954, "loss_cls_dn_2": 0.23032, "loss_box_dn_2": 0.79729, "loss_cls_dn_3": 0.23024, "loss_box_dn_3": 0.78617, "loss_cls_dn_4": 0.23728, "loss_box_dn_4": 0.79664, "loss_cls_dn_5": 0.26074, "loss_box_dn_5": 0.8174, "loss_dense_depth": 0.8398, "loss": 29.43373, "grad_norm": 45.76698, "time": 1.58286} -{"mode": "train", "epoch": 1, "iter": 88, "lr": 0.00013, "memory": 49164, "data_time": 0.09368, "loss_cls_0": 0.92699, "loss_box_0": 1.78877, "loss_cns_0": 0.61297, "loss_yns_0": 0.15646, "loss_cls_1": 0.98185, "loss_box_1": 1.90934, "loss_cns_1": 0.63195, "loss_yns_1": 0.16389, "loss_cls_2": 1.03648, "loss_box_2": 1.87717, "loss_cns_2": 0.64528, "loss_yns_2": 0.15868, "loss_cls_3": 1.0396, "loss_box_3": 1.87001, "loss_cns_3": 0.65136, "loss_yns_3": 0.16083, "loss_cls_4": 1.04905, "loss_box_4": 1.87702, "loss_cns_4": 0.65153, "loss_yns_4": 0.16559, "loss_cls_5": 1.05431, "loss_box_5": 1.88501, "loss_cns_5": 0.65173, "loss_yns_5": 0.15929, "loss_cls_dn_0": 0.31942, "loss_box_dn_0": 0.8192, "loss_cls_dn_1": 0.20895, "loss_box_dn_1": 0.80083, "loss_cls_dn_2": 0.23348, "loss_box_dn_2": 0.80113, "loss_cls_dn_3": 0.23056, "loss_box_dn_3": 0.79856, "loss_cls_dn_4": 0.23618, "loss_box_dn_4": 0.82139, "loss_cls_dn_5": 0.26708, "loss_box_dn_5": 0.84762, "loss_dense_depth": 0.81761, "loss": 29.30717, "grad_norm": 54.14699, "time": 1.58039} -{"mode": "train", "epoch": 1, "iter": 89, "lr": 0.00014, "memory": 49164, "data_time": 0.07481, "loss_cls_0": 0.93276, "loss_box_0": 1.81072, "loss_cns_0": 0.61239, "loss_yns_0": 0.1553, "loss_cls_1": 1.00456, "loss_box_1": 1.91465, "loss_cns_1": 0.63124, "loss_yns_1": 0.15789, "loss_cls_2": 1.05329, "loss_box_2": 1.89803, "loss_cns_2": 0.64466, "loss_yns_2": 0.15638, "loss_cls_3": 1.04275, "loss_box_3": 1.89935, "loss_cns_3": 0.64803, "loss_yns_3": 0.15828, "loss_cls_4": 1.03922, "loss_box_4": 1.9209, "loss_cns_4": 0.64738, "loss_yns_4": 0.16147, "loss_cls_5": 1.04063, "loss_box_5": 1.92219, "loss_cns_5": 0.64843, "loss_yns_5": 0.15882, "loss_cls_dn_0": 0.30857, "loss_box_dn_0": 0.81953, "loss_cls_dn_1": 0.21579, "loss_box_dn_1": 0.83335, "loss_cls_dn_2": 0.23773, "loss_box_dn_2": 0.83899, "loss_cls_dn_3": 0.23071, "loss_box_dn_3": 0.83649, "loss_cls_dn_4": 0.23222, "loss_box_dn_4": 0.86612, "loss_cls_dn_5": 0.25314, "loss_box_dn_5": 0.8788, "loss_dense_depth": 0.8017, "loss": 29.61246, "grad_norm": 62.61287, "time": 1.56508} -{"mode": "train", "epoch": 1, "iter": 90, "lr": 0.00014, "memory": 49164, "data_time": 0.10785, "loss_cls_0": 0.93504, "loss_box_0": 1.80587, "loss_cns_0": 0.61644, "loss_yns_0": 0.15496, "loss_cls_1": 1.01187, "loss_box_1": 1.96201, "loss_cns_1": 0.63962, "loss_yns_1": 0.16005, "loss_cls_2": 1.04353, "loss_box_2": 1.92405, "loss_cns_2": 0.65202, "loss_yns_2": 0.15729, "loss_cls_3": 1.03207, "loss_box_3": 1.91458, "loss_cns_3": 0.65269, "loss_yns_3": 0.15837, "loss_cls_4": 1.05922, "loss_box_4": 1.92563, "loss_cns_4": 0.65345, "loss_yns_4": 0.15868, "loss_cls_5": 1.07453, "loss_box_5": 1.94099, "loss_cns_5": 0.65273, "loss_yns_5": 0.15759, "loss_cls_dn_0": 0.31382, "loss_box_dn_0": 0.83013, "loss_cls_dn_1": 0.21986, "loss_box_dn_1": 0.83945, "loss_cls_dn_2": 0.23786, "loss_box_dn_2": 0.83279, "loss_cls_dn_3": 0.22867, "loss_box_dn_3": 0.8174, "loss_cls_dn_4": 0.22955, "loss_box_dn_4": 0.83707, "loss_cls_dn_5": 0.24377, "loss_box_dn_5": 0.84506, "loss_dense_depth": 0.84424, "loss": 29.76295, "grad_norm": 53.88475, "time": 1.6123} -{"mode": "train", "epoch": 1, "iter": 91, "lr": 0.00014, "memory": 49164, "data_time": 0.07475, "loss_cls_0": 0.92486, "loss_box_0": 1.79748, "loss_cns_0": 0.61727, "loss_yns_0": 0.15325, "loss_cls_1": 1.0317, "loss_box_1": 1.91251, "loss_cns_1": 0.64645, "loss_yns_1": 0.15776, "loss_cls_2": 1.02181, "loss_box_2": 1.87738, "loss_cns_2": 0.65602, "loss_yns_2": 0.15706, "loss_cls_3": 1.04748, "loss_box_3": 1.88806, "loss_cns_3": 0.65479, "loss_yns_3": 0.15712, "loss_cls_4": 1.03092, "loss_box_4": 1.87568, "loss_cns_4": 0.65586, "loss_yns_4": 0.16113, "loss_cls_5": 1.03954, "loss_box_5": 1.90187, "loss_cns_5": 0.65328, "loss_yns_5": 0.15568, "loss_cls_dn_0": 0.31021, "loss_box_dn_0": 0.82, "loss_cls_dn_1": 0.21824, "loss_box_dn_1": 0.80826, "loss_cls_dn_2": 0.23095, "loss_box_dn_2": 0.79837, "loss_cls_dn_3": 0.23415, "loss_box_dn_3": 0.79069, "loss_cls_dn_4": 0.22703, "loss_box_dn_4": 0.79454, "loss_cls_dn_5": 0.24384, "loss_box_dn_5": 0.81394, "loss_dense_depth": 0.80426, "loss": 29.26944, "grad_norm": 40.88663, "time": 1.60913} -{"mode": "train", "epoch": 1, "iter": 92, "lr": 0.00014, "memory": 49164, "data_time": 0.14613, "loss_cls_0": 0.9133, "loss_box_0": 1.78772, "loss_cns_0": 0.61885, "loss_yns_0": 0.15238, "loss_cls_1": 1.01353, "loss_box_1": 1.95517, "loss_cns_1": 0.63769, "loss_yns_1": 0.15536, "loss_cls_2": 1.00698, "loss_box_2": 1.92329, "loss_cns_2": 0.64665, "loss_yns_2": 0.15602, "loss_cls_3": 1.04524, "loss_box_3": 1.93814, "loss_cns_3": 0.65275, "loss_yns_3": 0.15807, "loss_cls_4": 1.02526, "loss_box_4": 1.92559, "loss_cns_4": 0.65317, "loss_yns_4": 0.15982, "loss_cls_5": 1.04401, "loss_box_5": 1.93271, "loss_cns_5": 0.65112, "loss_yns_5": 0.15644, "loss_cls_dn_0": 0.30477, "loss_box_dn_0": 0.81557, "loss_cls_dn_1": 0.21158, "loss_box_dn_1": 0.78802, "loss_cls_dn_2": 0.21817, "loss_box_dn_2": 0.77815, "loss_cls_dn_3": 0.23001, "loss_box_dn_3": 0.78108, "loss_cls_dn_4": 0.22414, "loss_box_dn_4": 0.78079, "loss_cls_dn_5": 0.24889, "loss_box_dn_5": 0.7988, "loss_dense_depth": 0.82921, "loss": 29.31845, "grad_norm": 51.35901, "time": 1.65165} -{"mode": "train", "epoch": 1, "iter": 93, "lr": 0.00014, "memory": 49164, "data_time": 0.07303, "loss_cls_0": 0.91919, "loss_box_0": 1.77211, "loss_cns_0": 0.62279, "loss_yns_0": 0.15219, "loss_cls_1": 1.00093, "loss_box_1": 1.90091, "loss_cns_1": 0.64086, "loss_yns_1": 0.15409, "loss_cls_2": 1.04758, "loss_box_2": 1.88158, "loss_cns_2": 0.64807, "loss_yns_2": 0.15435, "loss_cls_3": 1.03733, "loss_box_3": 1.86974, "loss_cns_3": 0.65324, "loss_yns_3": 0.15737, "loss_cls_4": 1.01886, "loss_box_4": 1.86405, "loss_cns_4": 0.65281, "loss_yns_4": 0.15638, "loss_cls_5": 1.02712, "loss_box_5": 1.8834, "loss_cns_5": 0.64877, "loss_yns_5": 0.15451, "loss_cls_dn_0": 0.29448, "loss_box_dn_0": 0.81135, "loss_cls_dn_1": 0.20645, "loss_box_dn_1": 0.7906, "loss_cls_dn_2": 0.21344, "loss_box_dn_2": 0.78484, "loss_cls_dn_3": 0.21621, "loss_box_dn_3": 0.78143, "loss_cls_dn_4": 0.21691, "loss_box_dn_4": 0.78721, "loss_cls_dn_5": 0.23771, "loss_box_dn_5": 0.80897, "loss_dense_depth": 0.79059, "loss": 28.95842, "grad_norm": 53.20693, "time": 1.55766} -{"mode": "train", "epoch": 1, "iter": 94, "lr": 0.00014, "memory": 49164, "data_time": 0.10766, "loss_cls_0": 0.93287, "loss_box_0": 1.81742, "loss_cns_0": 0.61868, "loss_yns_0": 0.15237, "loss_cls_1": 1.02596, "loss_box_1": 1.93532, "loss_cns_1": 0.63647, "loss_yns_1": 0.1555, "loss_cls_2": 1.02408, "loss_box_2": 1.8775, "loss_cns_2": 0.6463, "loss_yns_2": 0.15439, "loss_cls_3": 1.02422, "loss_box_3": 1.8714, "loss_cns_3": 0.65424, "loss_yns_3": 0.15659, "loss_cls_4": 1.02809, "loss_box_4": 1.85156, "loss_cns_4": 0.65457, "loss_yns_4": 0.15537, "loss_cls_5": 1.03354, "loss_box_5": 1.8487, "loss_cns_5": 0.65247, "loss_yns_5": 0.15409, "loss_cls_dn_0": 0.30153, "loss_box_dn_0": 0.81221, "loss_cls_dn_1": 0.21154, "loss_box_dn_1": 0.80056, "loss_cls_dn_2": 0.21899, "loss_box_dn_2": 0.78127, "loss_cls_dn_3": 0.21522, "loss_box_dn_3": 0.7857, "loss_cls_dn_4": 0.21682, "loss_box_dn_4": 0.79079, "loss_cls_dn_5": 0.23176, "loss_box_dn_5": 0.80808, "loss_dense_depth": 0.79128, "loss": 29.02746, "grad_norm": 35.26435, "time": 1.67748} -{"mode": "train", "epoch": 1, "iter": 95, "lr": 0.00014, "memory": 49164, "data_time": 0.09452, "loss_cls_0": 0.92166, "loss_box_0": 1.82136, "loss_cns_0": 0.62011, "loss_yns_0": 0.1531, "loss_cls_1": 1.0437, "loss_box_1": 1.94577, "loss_cns_1": 0.63799, "loss_yns_1": 0.15518, "loss_cls_2": 1.03409, "loss_box_2": 1.86233, "loss_cns_2": 0.65169, "loss_yns_2": 0.1578, "loss_cls_3": 1.04875, "loss_box_3": 1.87902, "loss_cns_3": 0.65611, "loss_yns_3": 0.15853, "loss_cls_4": 1.04812, "loss_box_4": 1.86174, "loss_cns_4": 0.65689, "loss_yns_4": 0.16005, "loss_cls_5": 1.06394, "loss_box_5": 1.85051, "loss_cns_5": 0.65807, "loss_yns_5": 0.15681, "loss_cls_dn_0": 0.30483, "loss_box_dn_0": 0.81096, "loss_cls_dn_1": 0.20891, "loss_box_dn_1": 0.79514, "loss_cls_dn_2": 0.21513, "loss_box_dn_2": 0.77062, "loss_cls_dn_3": 0.21096, "loss_box_dn_3": 0.79215, "loss_cls_dn_4": 0.21236, "loss_box_dn_4": 0.79546, "loss_cls_dn_5": 0.22535, "loss_box_dn_5": 0.80699, "loss_dense_depth": 0.87206, "loss": 29.2242, "grad_norm": 51.95293, "time": 1.58501} -{"mode": "train", "epoch": 1, "iter": 96, "lr": 0.00014, "memory": 49164, "data_time": 0.08078, "loss_cls_0": 0.94471, "loss_box_0": 1.84657, "loss_cns_0": 0.6203, "loss_yns_0": 0.15599, "loss_cls_1": 1.05163, "loss_box_1": 1.98354, "loss_cns_1": 0.63181, "loss_yns_1": 0.15765, "loss_cls_2": 1.05232, "loss_box_2": 1.9306, "loss_cns_2": 0.64434, "loss_yns_2": 0.15935, "loss_cls_3": 1.04654, "loss_box_3": 1.93311, "loss_cns_3": 0.64728, "loss_yns_3": 0.16059, "loss_cls_4": 1.0546, "loss_box_4": 1.91852, "loss_cns_4": 0.64937, "loss_yns_4": 0.16509, "loss_cls_5": 1.06749, "loss_box_5": 1.9094, "loss_cns_5": 0.64911, "loss_yns_5": 0.15918, "loss_cls_dn_0": 0.30077, "loss_box_dn_0": 0.81094, "loss_cls_dn_1": 0.21335, "loss_box_dn_1": 0.80538, "loss_cls_dn_2": 0.21992, "loss_box_dn_2": 0.78388, "loss_cls_dn_3": 0.21301, "loss_box_dn_3": 0.80308, "loss_cls_dn_4": 0.21471, "loss_box_dn_4": 0.80714, "loss_cls_dn_5": 0.22916, "loss_box_dn_5": 0.81383, "loss_dense_depth": 0.8079, "loss": 29.56215, "grad_norm": 49.16695, "time": 1.63461} -{"mode": "train", "epoch": 1, "iter": 97, "lr": 0.00014, "memory": 49164, "data_time": 0.09965, "loss_cls_0": 0.97991, "loss_box_0": 1.85327, "loss_cns_0": 0.61827, "loss_yns_0": 0.15618, "loss_cls_1": 1.03131, "loss_box_1": 1.94952, "loss_cns_1": 0.63383, "loss_yns_1": 0.15872, "loss_cls_2": 1.05299, "loss_box_2": 1.91233, "loss_cns_2": 0.64617, "loss_yns_2": 0.15684, "loss_cls_3": 1.09074, "loss_box_3": 1.90801, "loss_cns_3": 0.64833, "loss_yns_3": 0.15772, "loss_cls_4": 1.08265, "loss_box_4": 1.90214, "loss_cns_4": 0.65024, "loss_yns_4": 0.1646, "loss_cls_5": 1.08479, "loss_box_5": 1.89659, "loss_cns_5": 0.64956, "loss_yns_5": 0.15773, "loss_cls_dn_0": 0.28791, "loss_box_dn_0": 0.8189, "loss_cls_dn_1": 0.20806, "loss_box_dn_1": 0.79428, "loss_cls_dn_2": 0.21525, "loss_box_dn_2": 0.78356, "loss_cls_dn_3": 0.21188, "loss_box_dn_3": 0.79271, "loss_cls_dn_4": 0.21185, "loss_box_dn_4": 0.80778, "loss_cls_dn_5": 0.22718, "loss_box_dn_5": 0.81122, "loss_dense_depth": 0.85082, "loss": 29.56387, "grad_norm": 66.63815, "time": 1.66527} -{"mode": "train", "epoch": 1, "iter": 98, "lr": 0.00014, "memory": 49164, "data_time": 0.0744, "loss_cls_0": 0.95259, "loss_box_0": 1.82703, "loss_cns_0": 0.62047, "loss_yns_0": 0.15531, "loss_cls_1": 1.02568, "loss_box_1": 1.91346, "loss_cns_1": 0.63973, "loss_yns_1": 0.15784, "loss_cls_2": 1.06594, "loss_box_2": 1.89327, "loss_cns_2": 0.65438, "loss_yns_2": 0.15715, "loss_cls_3": 1.05189, "loss_box_3": 1.90815, "loss_cns_3": 0.65394, "loss_yns_3": 0.15746, "loss_cls_4": 1.0494, "loss_box_4": 1.89705, "loss_cns_4": 0.6566, "loss_yns_4": 0.1605, "loss_cls_5": 1.06534, "loss_box_5": 1.89002, "loss_cns_5": 0.65254, "loss_yns_5": 0.15861, "loss_cls_dn_0": 0.28444, "loss_box_dn_0": 0.82207, "loss_cls_dn_1": 0.21127, "loss_box_dn_1": 0.807, "loss_cls_dn_2": 0.22212, "loss_box_dn_2": 0.80213, "loss_cls_dn_3": 0.21346, "loss_box_dn_3": 0.81555, "loss_cls_dn_4": 0.2176, "loss_box_dn_4": 0.8331, "loss_cls_dn_5": 0.23376, "loss_box_dn_5": 0.84463, "loss_dense_depth": 0.82854, "loss": 29.50001, "grad_norm": 53.67993, "time": 1.55777} -{"mode": "train", "epoch": 1, "iter": 99, "lr": 0.00014, "memory": 49164, "data_time": 0.12013, "loss_cls_0": 0.93319, "loss_box_0": 1.79381, "loss_cns_0": 0.61672, "loss_yns_0": 0.15286, "loss_cls_1": 1.00392, "loss_box_1": 1.9346, "loss_cns_1": 0.63241, "loss_yns_1": 0.15834, "loss_cls_2": 1.03323, "loss_box_2": 1.88851, "loss_cns_2": 0.64556, "loss_yns_2": 0.15921, "loss_cls_3": 1.05528, "loss_box_3": 1.88959, "loss_cns_3": 0.64718, "loss_yns_3": 0.15965, "loss_cls_4": 1.04812, "loss_box_4": 1.87936, "loss_cns_4": 0.64866, "loss_yns_4": 0.15833, "loss_cls_5": 1.06189, "loss_box_5": 1.87177, "loss_cns_5": 0.64722, "loss_yns_5": 0.15718, "loss_cls_dn_0": 0.28489, "loss_box_dn_0": 0.81126, "loss_cls_dn_1": 0.2037, "loss_box_dn_1": 0.82551, "loss_cls_dn_2": 0.21602, "loss_box_dn_2": 0.81328, "loss_cls_dn_3": 0.21689, "loss_box_dn_3": 0.81837, "loss_cls_dn_4": 0.22579, "loss_box_dn_4": 0.82999, "loss_cls_dn_5": 0.2489, "loss_box_dn_5": 0.85239, "loss_dense_depth": 0.8629, "loss": 29.38649, "grad_norm": 53.46401, "time": 1.59766} -{"mode": "train", "epoch": 1, "iter": 100, "lr": 0.00014, "memory": 49164, "data_time": 0.08375, "loss_cls_0": 0.92074, "loss_box_0": 1.80372, "loss_cns_0": 0.6163, "loss_yns_0": 0.15414, "loss_cls_1": 0.98498, "loss_box_1": 1.94288, "loss_cns_1": 0.63083, "loss_yns_1": 0.15689, "loss_cls_2": 1.02275, "loss_box_2": 1.89389, "loss_cns_2": 0.64178, "loss_yns_2": 0.15905, "loss_cls_3": 1.01962, "loss_box_3": 1.86183, "loss_cns_3": 0.64677, "loss_yns_3": 0.16099, "loss_cls_4": 1.01984, "loss_box_4": 1.85467, "loss_cns_4": 0.6497, "loss_yns_4": 0.16186, "loss_cls_5": 1.03322, "loss_box_5": 1.85515, "loss_cns_5": 0.64932, "loss_yns_5": 0.15974, "loss_cls_dn_0": 0.27602, "loss_box_dn_0": 0.81773, "loss_cls_dn_1": 0.19494, "loss_box_dn_1": 0.84344, "loss_cls_dn_2": 0.20709, "loss_box_dn_2": 0.82206, "loss_cls_dn_3": 0.20941, "loss_box_dn_3": 0.81836, "loss_cls_dn_4": 0.21694, "loss_box_dn_4": 0.83186, "loss_cls_dn_5": 0.23463, "loss_box_dn_5": 0.85704, "loss_dense_depth": 0.81764, "loss": 29.1478, "grad_norm": 50.70029, "time": 1.59912} -{"mode": "train", "epoch": 1, "iter": 101, "lr": 0.00014, "memory": 49164, "data_time": 0.1221, "loss_cls_0": 0.91949, "loss_box_0": 1.76875, "loss_cns_0": 0.61632, "loss_yns_0": 0.15591, "loss_cls_1": 1.02382, "loss_box_1": 1.94054, "loss_cns_1": 0.63617, "loss_yns_1": 0.15938, "loss_cls_2": 1.1259, "loss_box_2": 1.87704, "loss_cns_2": 0.64859, "loss_yns_2": 0.15888, "loss_cls_3": 1.06977, "loss_box_3": 1.85836, "loss_cns_3": 0.65362, "loss_yns_3": 0.16045, "loss_cls_4": 1.10044, "loss_box_4": 1.85519, "loss_cns_4": 0.65348, "loss_yns_4": 0.16091, "loss_cls_5": 1.08447, "loss_box_5": 1.85942, "loss_cns_5": 0.65389, "loss_yns_5": 0.16146, "loss_cls_dn_0": 0.27997, "loss_box_dn_0": 0.80763, "loss_cls_dn_1": 0.19571, "loss_box_dn_1": 0.84252, "loss_cls_dn_2": 0.2086, "loss_box_dn_2": 0.81484, "loss_cls_dn_3": 0.20721, "loss_box_dn_3": 0.8123, "loss_cls_dn_4": 0.20921, "loss_box_dn_4": 0.82172, "loss_cls_dn_5": 0.22053, "loss_box_dn_5": 0.83406, "loss_dense_depth": 0.84571, "loss": 29.40225, "grad_norm": 68.73978, "time": 1.66689} -{"mode": "train", "epoch": 1, "iter": 102, "lr": 0.00014, "memory": 49164, "data_time": 0.16412, "loss_cls_0": 0.93788, "loss_box_0": 1.78619, "loss_cns_0": 0.61033, "loss_yns_0": 0.15732, "loss_cls_1": 0.99015, "loss_box_1": 1.89315, "loss_cns_1": 0.64104, "loss_yns_1": 0.15914, "loss_cls_2": 1.03083, "loss_box_2": 1.84688, "loss_cns_2": 0.65084, "loss_yns_2": 0.15882, "loss_cls_3": 1.03247, "loss_box_3": 1.86446, "loss_cns_3": 0.6539, "loss_yns_3": 0.16024, "loss_cls_4": 1.02677, "loss_box_4": 1.86353, "loss_cns_4": 0.65179, "loss_yns_4": 0.16062, "loss_cls_5": 1.03697, "loss_box_5": 1.86546, "loss_cns_5": 0.65218, "loss_yns_5": 0.1589, "loss_cls_dn_0": 0.28344, "loss_box_dn_0": 0.81058, "loss_cls_dn_1": 0.19686, "loss_box_dn_1": 0.79302, "loss_cls_dn_2": 0.20704, "loss_box_dn_2": 0.77657, "loss_cls_dn_3": 0.2072, "loss_box_dn_3": 0.78495, "loss_cls_dn_4": 0.20974, "loss_box_dn_4": 0.79771, "loss_cls_dn_5": 0.2213, "loss_box_dn_5": 0.8036, "loss_dense_depth": 0.82989, "loss": 28.91177, "grad_norm": 44.77891, "time": 1.66937} -{"mode": "train", "epoch": 1, "iter": 103, "lr": 0.00014, "memory": 49164, "data_time": 0.07402, "loss_cls_0": 0.9126, "loss_box_0": 1.79241, "loss_cns_0": 0.61759, "loss_yns_0": 0.15861, "loss_cls_1": 0.9786, "loss_box_1": 1.92376, "loss_cns_1": 0.63937, "loss_yns_1": 0.16155, "loss_cls_2": 1.04134, "loss_box_2": 1.86402, "loss_cns_2": 0.65285, "loss_yns_2": 0.16206, "loss_cls_3": 1.0246, "loss_box_3": 1.86512, "loss_cns_3": 0.65687, "loss_yns_3": 0.16276, "loss_cls_4": 1.03835, "loss_box_4": 1.84726, "loss_cns_4": 0.6553, "loss_yns_4": 0.1613, "loss_cls_5": 1.03176, "loss_box_5": 1.84488, "loss_cns_5": 0.65974, "loss_yns_5": 0.16458, "loss_cls_dn_0": 0.26441, "loss_box_dn_0": 0.807, "loss_cls_dn_1": 0.18795, "loss_box_dn_1": 0.78587, "loss_cls_dn_2": 0.20116, "loss_box_dn_2": 0.77767, "loss_cls_dn_3": 0.19715, "loss_box_dn_3": 0.77741, "loss_cls_dn_4": 0.20431, "loss_box_dn_4": 0.7851, "loss_cls_dn_5": 0.21364, "loss_box_dn_5": 0.78928, "loss_dense_depth": 0.8504, "loss": 28.85864, "grad_norm": 44.66996, "time": 1.58562} -{"mode": "train", "epoch": 1, "iter": 104, "lr": 0.00014, "memory": 49164, "data_time": 0.07848, "loss_cls_0": 0.93157, "loss_box_0": 1.82879, "loss_cns_0": 0.61574, "loss_yns_0": 0.15754, "loss_cls_1": 0.97326, "loss_box_1": 1.96775, "loss_cns_1": 0.63551, "loss_yns_1": 0.15861, "loss_cls_2": 1.05099, "loss_box_2": 1.90368, "loss_cns_2": 0.65019, "loss_yns_2": 0.15971, "loss_cls_3": 1.02513, "loss_box_3": 1.88461, "loss_cns_3": 0.65466, "loss_yns_3": 0.16013, "loss_cls_4": 1.02544, "loss_box_4": 1.87645, "loss_cns_4": 0.65484, "loss_yns_4": 0.15979, "loss_cls_5": 1.03256, "loss_box_5": 1.88838, "loss_cns_5": 0.65256, "loss_yns_5": 0.16153, "loss_cls_dn_0": 0.25871, "loss_box_dn_0": 0.82402, "loss_cls_dn_1": 0.18234, "loss_box_dn_1": 0.78003, "loss_cls_dn_2": 0.1994, "loss_box_dn_2": 0.78436, "loss_cls_dn_3": 0.19114, "loss_box_dn_3": 0.78323, "loss_cls_dn_4": 0.19839, "loss_box_dn_4": 0.79446, "loss_cls_dn_5": 0.20528, "loss_box_dn_5": 0.81304, "loss_dense_depth": 0.80795, "loss": 29.03174, "grad_norm": 46.99785, "time": 1.56345} -{"mode": "train", "epoch": 1, "iter": 105, "lr": 0.00014, "memory": 49164, "data_time": 0.07663, "loss_cls_0": 0.88197, "loss_box_0": 1.74711, "loss_cns_0": 0.61966, "loss_yns_0": 0.15548, "loss_cls_1": 0.95827, "loss_box_1": 1.90159, "loss_cns_1": 0.64271, "loss_yns_1": 0.15912, "loss_cls_2": 1.01946, "loss_box_2": 1.84892, "loss_cns_2": 0.65688, "loss_yns_2": 0.15989, "loss_cls_3": 1.01288, "loss_box_3": 1.85598, "loss_cns_3": 0.65801, "loss_yns_3": 0.15883, "loss_cls_4": 1.02608, "loss_box_4": 1.84858, "loss_cns_4": 0.65866, "loss_yns_4": 0.16034, "loss_cls_5": 1.04024, "loss_box_5": 1.85931, "loss_cns_5": 0.65749, "loss_yns_5": 0.15712, "loss_cls_dn_0": 0.25636, "loss_box_dn_0": 0.81425, "loss_cls_dn_1": 0.18327, "loss_box_dn_1": 0.81463, "loss_cls_dn_2": 0.19928, "loss_box_dn_2": 0.81753, "loss_cls_dn_3": 0.19145, "loss_box_dn_3": 0.83119, "loss_cls_dn_4": 0.19623, "loss_box_dn_4": 0.84513, "loss_cls_dn_5": 0.20255, "loss_box_dn_5": 0.86809, "loss_dense_depth": 0.8041, "loss": 28.86862, "grad_norm": 51.92323, "time": 1.65829} -{"mode": "train", "epoch": 1, "iter": 106, "lr": 0.00014, "memory": 49164, "data_time": 0.09371, "loss_cls_0": 0.90811, "loss_box_0": 1.68293, "loss_cns_0": 0.60321, "loss_yns_0": 0.14979, "loss_cls_1": 0.96437, "loss_box_1": 1.90398, "loss_cns_1": 0.64161, "loss_yns_1": 0.15709, "loss_cls_2": 1.01717, "loss_box_2": 1.8772, "loss_cns_2": 0.65343, "loss_yns_2": 0.1582, "loss_cls_3": 1.0099, "loss_box_3": 1.90926, "loss_cns_3": 0.65732, "loss_yns_3": 0.15836, "loss_cls_4": 1.01707, "loss_box_4": 1.88377, "loss_cns_4": 0.65673, "loss_yns_4": 0.15955, "loss_cls_5": 1.02197, "loss_box_5": 1.87445, "loss_cns_5": 0.6557, "loss_yns_5": 0.1617, "loss_cls_dn_0": 0.25906, "loss_box_dn_0": 0.79361, "loss_cls_dn_1": 0.18375, "loss_box_dn_1": 0.82199, "loss_cls_dn_2": 0.19595, "loss_box_dn_2": 0.82419, "loss_cls_dn_3": 0.19021, "loss_box_dn_3": 0.84872, "loss_cls_dn_4": 0.1919, "loss_box_dn_4": 0.85092, "loss_cls_dn_5": 0.19964, "loss_box_dn_5": 0.86492, "loss_dense_depth": 0.79375, "loss": 28.90147, "grad_norm": 53.721, "time": 1.58398} -{"mode": "train", "epoch": 1, "iter": 107, "lr": 0.00014, "memory": 49164, "data_time": 0.10633, "loss_cls_0": 0.88258, "loss_box_0": 1.71812, "loss_cns_0": 0.61048, "loss_yns_0": 0.15286, "loss_cls_1": 0.95562, "loss_box_1": 1.88332, "loss_cns_1": 0.64259, "loss_yns_1": 0.15693, "loss_cls_2": 1.00881, "loss_box_2": 1.83177, "loss_cns_2": 0.65526, "loss_yns_2": 0.15763, "loss_cls_3": 1.00342, "loss_box_3": 1.86254, "loss_cns_3": 0.66032, "loss_yns_3": 0.15727, "loss_cls_4": 1.01618, "loss_box_4": 1.82531, "loss_cns_4": 0.66042, "loss_yns_4": 0.15829, "loss_cls_5": 1.02399, "loss_box_5": 1.80217, "loss_cns_5": 0.65792, "loss_yns_5": 0.16346, "loss_cls_dn_0": 0.25383, "loss_box_dn_0": 0.81356, "loss_cls_dn_1": 0.18314, "loss_box_dn_1": 0.80156, "loss_cls_dn_2": 0.19397, "loss_box_dn_2": 0.79508, "loss_cls_dn_3": 0.18831, "loss_box_dn_3": 0.81426, "loss_cls_dn_4": 0.1944, "loss_box_dn_4": 0.80652, "loss_cls_dn_5": 0.20625, "loss_box_dn_5": 0.80493, "loss_dense_depth": 0.80941, "loss": 28.51248, "grad_norm": 44.93464, "time": 1.58254} -{"mode": "train", "epoch": 1, "iter": 108, "lr": 0.00014, "memory": 49164, "data_time": 0.07477, "loss_cls_0": 0.90422, "loss_box_0": 1.77651, "loss_cns_0": 0.62225, "loss_yns_0": 0.15641, "loss_cls_1": 0.97325, "loss_box_1": 1.87744, "loss_cns_1": 0.64193, "loss_yns_1": 0.15895, "loss_cls_2": 1.03452, "loss_box_2": 1.81588, "loss_cns_2": 0.6544, "loss_yns_2": 0.16024, "loss_cls_3": 1.07154, "loss_box_3": 1.82979, "loss_cns_3": 0.65738, "loss_yns_3": 0.15778, "loss_cls_4": 1.03401, "loss_box_4": 1.84128, "loss_cns_4": 0.6565, "loss_yns_4": 0.15964, "loss_cls_5": 1.04855, "loss_box_5": 1.82694, "loss_cns_5": 0.65584, "loss_yns_5": 0.15875, "loss_cls_dn_0": 0.25046, "loss_box_dn_0": 0.81292, "loss_cls_dn_1": 0.18616, "loss_box_dn_1": 0.7577, "loss_cls_dn_2": 0.19366, "loss_box_dn_2": 0.74716, "loss_cls_dn_3": 0.19278, "loss_box_dn_3": 0.7523, "loss_cls_dn_4": 0.19704, "loss_box_dn_4": 0.7606, "loss_cls_dn_5": 0.2085, "loss_box_dn_5": 0.76147, "loss_dense_depth": 0.77967, "loss": 28.47442, "grad_norm": 54.78695, "time": 1.57704} -{"mode": "train", "epoch": 1, "iter": 109, "lr": 0.00014, "memory": 49164, "data_time": 0.07546, "loss_cls_0": 0.90009, "loss_box_0": 1.81172, "loss_cns_0": 0.61172, "loss_yns_0": 0.15605, "loss_cls_1": 0.96618, "loss_box_1": 1.89163, "loss_cns_1": 0.63536, "loss_yns_1": 0.15893, "loss_cls_2": 1.01743, "loss_box_2": 1.83466, "loss_cns_2": 0.64841, "loss_yns_2": 0.16176, "loss_cls_3": 1.02522, "loss_box_3": 1.83262, "loss_cns_3": 0.65119, "loss_yns_3": 0.15931, "loss_cls_4": 1.03982, "loss_box_4": 1.87085, "loss_cns_4": 0.65103, "loss_yns_4": 0.16185, "loss_cls_5": 1.05445, "loss_box_5": 1.87398, "loss_cns_5": 0.6502, "loss_yns_5": 0.16096, "loss_cls_dn_0": 0.25641, "loss_box_dn_0": 0.82229, "loss_cls_dn_1": 0.18454, "loss_box_dn_1": 0.7816, "loss_cls_dn_2": 0.19168, "loss_box_dn_2": 0.76459, "loss_cls_dn_3": 0.18871, "loss_box_dn_3": 0.77058, "loss_cls_dn_4": 0.19621, "loss_box_dn_4": 0.79867, "loss_cls_dn_5": 0.20864, "loss_box_dn_5": 0.81419, "loss_dense_depth": 0.79118, "loss": 28.69469, "grad_norm": 43.99228, "time": 1.56235} -{"mode": "train", "epoch": 1, "iter": 110, "lr": 0.00014, "memory": 49164, "data_time": 0.07517, "loss_cls_0": 0.91311, "loss_box_0": 1.81103, "loss_cns_0": 0.61583, "loss_yns_0": 0.15539, "loss_cls_1": 0.98329, "loss_box_1": 1.91954, "loss_cns_1": 0.62726, "loss_yns_1": 0.15699, "loss_cls_2": 1.01249, "loss_box_2": 1.86236, "loss_cns_2": 0.64557, "loss_yns_2": 0.15669, "loss_cls_3": 1.04372, "loss_box_3": 1.84267, "loss_cns_3": 0.64727, "loss_yns_3": 0.15747, "loss_cls_4": 1.03002, "loss_box_4": 1.86974, "loss_cns_4": 0.64771, "loss_yns_4": 0.15959, "loss_cls_5": 1.04045, "loss_box_5": 1.89588, "loss_cns_5": 0.6412, "loss_yns_5": 0.16213, "loss_cls_dn_0": 0.25759, "loss_box_dn_0": 0.80763, "loss_cls_dn_1": 0.18961, "loss_box_dn_1": 0.83465, "loss_cls_dn_2": 0.19438, "loss_box_dn_2": 0.82298, "loss_cls_dn_3": 0.19232, "loss_box_dn_3": 0.84017, "loss_cls_dn_4": 0.19936, "loss_box_dn_4": 0.88528, "loss_cls_dn_5": 0.21185, "loss_box_dn_5": 0.91573, "loss_dense_depth": 0.81654, "loss": 29.16549, "grad_norm": 47.28304, "time": 1.57071} -{"mode": "train", "epoch": 1, "iter": 111, "lr": 0.00014, "memory": 49164, "data_time": 0.07663, "loss_cls_0": 0.91772, "loss_box_0": 1.81859, "loss_cns_0": 0.62088, "loss_yns_0": 0.15738, "loss_cls_1": 0.96753, "loss_box_1": 1.91561, "loss_cns_1": 0.62749, "loss_yns_1": 0.15634, "loss_cls_2": 1.02078, "loss_box_2": 1.86014, "loss_cns_2": 0.64266, "loss_yns_2": 0.15625, "loss_cls_3": 1.03706, "loss_box_3": 1.85846, "loss_cns_3": 0.64667, "loss_yns_3": 0.15675, "loss_cls_4": 1.02605, "loss_box_4": 1.87516, "loss_cns_4": 0.64654, "loss_yns_4": 0.15729, "loss_cls_5": 1.03778, "loss_box_5": 1.89624, "loss_cns_5": 0.64203, "loss_yns_5": 0.15746, "loss_cls_dn_0": 0.2505, "loss_box_dn_0": 0.80322, "loss_cls_dn_1": 0.18745, "loss_box_dn_1": 0.86788, "loss_cls_dn_2": 0.19789, "loss_box_dn_2": 0.85846, "loss_cls_dn_3": 0.1966, "loss_box_dn_3": 0.87554, "loss_cls_dn_4": 0.19821, "loss_box_dn_4": 0.91483, "loss_cls_dn_5": 0.20758, "loss_box_dn_5": 0.93867, "loss_dense_depth": 0.82058, "loss": 29.31629, "grad_norm": 52.27549, "time": 1.58877} -{"mode": "train", "epoch": 1, "iter": 112, "lr": 0.00014, "memory": 49164, "data_time": 0.06901, "loss_cls_0": 0.93084, "loss_box_0": 1.84981, "loss_cns_0": 0.61704, "loss_yns_0": 0.15983, "loss_cls_1": 0.96885, "loss_box_1": 1.94561, "loss_cns_1": 0.62188, "loss_yns_1": 0.15618, "loss_cls_2": 1.0278, "loss_box_2": 1.88518, "loss_cns_2": 0.64181, "loss_yns_2": 0.15592, "loss_cls_3": 1.02687, "loss_box_3": 1.88517, "loss_cns_3": 0.64777, "loss_yns_3": 0.15803, "loss_cls_4": 1.02546, "loss_box_4": 1.87447, "loss_cns_4": 0.64565, "loss_yns_4": 0.16163, "loss_cls_5": 1.0367, "loss_box_5": 1.88373, "loss_cns_5": 0.65026, "loss_yns_5": 0.15733, "loss_cls_dn_0": 0.25825, "loss_box_dn_0": 0.80663, "loss_cls_dn_1": 0.18748, "loss_box_dn_1": 0.90498, "loss_cls_dn_2": 0.20004, "loss_box_dn_2": 0.88529, "loss_cls_dn_3": 0.19879, "loss_box_dn_3": 0.89417, "loss_cls_dn_4": 0.19655, "loss_box_dn_4": 0.9104, "loss_cls_dn_5": 0.20495, "loss_box_dn_5": 0.93187, "loss_dense_depth": 0.85697, "loss": 29.55018, "grad_norm": 49.95032, "time": 1.57124} -{"mode": "train", "epoch": 1, "iter": 113, "lr": 0.00014, "memory": 49164, "data_time": 0.07407, "loss_cls_0": 0.88311, "loss_box_0": 1.81585, "loss_cns_0": 0.61879, "loss_yns_0": 0.15384, "loss_cls_1": 0.93455, "loss_box_1": 1.87791, "loss_cns_1": 0.63277, "loss_yns_1": 0.15216, "loss_cls_2": 1.00184, "loss_box_2": 1.81644, "loss_cns_2": 0.64932, "loss_yns_2": 0.15343, "loss_cls_3": 0.98911, "loss_box_3": 1.79546, "loss_cns_3": 0.65694, "loss_yns_3": 0.15458, "loss_cls_4": 0.99606, "loss_box_4": 1.81304, "loss_cns_4": 0.65804, "loss_yns_4": 0.15749, "loss_cls_5": 1.00504, "loss_box_5": 1.8191, "loss_cns_5": 0.65742, "loss_yns_5": 0.15395, "loss_cls_dn_0": 0.25052, "loss_box_dn_0": 0.80197, "loss_cls_dn_1": 0.18372, "loss_box_dn_1": 0.85475, "loss_cls_dn_2": 0.19582, "loss_box_dn_2": 0.82617, "loss_cls_dn_3": 0.19718, "loss_box_dn_3": 0.8191, "loss_cls_dn_4": 0.19586, "loss_box_dn_4": 0.83652, "loss_cls_dn_5": 0.2052, "loss_box_dn_5": 0.85039, "loss_dense_depth": 0.79024, "loss": 28.55367, "grad_norm": 39.96003, "time": 1.58287} -{"mode": "train", "epoch": 1, "iter": 114, "lr": 0.00015, "memory": 49164, "data_time": 0.07745, "loss_cls_0": 0.8983, "loss_box_0": 1.83794, "loss_cns_0": 0.61183, "loss_yns_0": 0.15531, "loss_cls_1": 0.97039, "loss_box_1": 1.88887, "loss_cns_1": 0.6376, "loss_yns_1": 0.1544, "loss_cls_2": 1.00076, "loss_box_2": 1.81516, "loss_cns_2": 0.64955, "loss_yns_2": 0.15552, "loss_cls_3": 1.01218, "loss_box_3": 1.79584, "loss_cns_3": 0.65247, "loss_yns_3": 0.15598, "loss_cls_4": 1.00838, "loss_box_4": 1.82973, "loss_cns_4": 0.65387, "loss_yns_4": 0.15565, "loss_cls_5": 1.01303, "loss_box_5": 1.81976, "loss_cns_5": 0.64992, "loss_yns_5": 0.15653, "loss_cls_dn_0": 0.25897, "loss_box_dn_0": 0.80667, "loss_cls_dn_1": 0.17828, "loss_box_dn_1": 0.80714, "loss_cls_dn_2": 0.18851, "loss_box_dn_2": 0.78724, "loss_cls_dn_3": 0.18817, "loss_box_dn_3": 0.78475, "loss_cls_dn_4": 0.19404, "loss_box_dn_4": 0.80937, "loss_cls_dn_5": 0.20462, "loss_box_dn_5": 0.81416, "loss_dense_depth": 0.80254, "loss": 28.50344, "grad_norm": 46.82184, "time": 1.63063} -{"mode": "train", "epoch": 1, "iter": 115, "lr": 0.00015, "memory": 49164, "data_time": 0.09407, "loss_cls_0": 0.8602, "loss_box_0": 1.79234, "loss_cns_0": 0.61591, "loss_yns_0": 0.15313, "loss_cls_1": 0.96947, "loss_box_1": 1.87474, "loss_cns_1": 0.64177, "loss_yns_1": 0.15394, "loss_cls_2": 1.01325, "loss_box_2": 1.82562, "loss_cns_2": 0.65228, "loss_yns_2": 0.15284, "loss_cls_3": 1.07144, "loss_box_3": 1.80513, "loss_cns_3": 0.65543, "loss_yns_3": 0.15284, "loss_cls_4": 0.99865, "loss_box_4": 1.80363, "loss_cns_4": 0.65552, "loss_yns_4": 0.15371, "loss_cls_5": 1.01028, "loss_box_5": 1.79381, "loss_cns_5": 0.65298, "loss_yns_5": 0.15783, "loss_cls_dn_0": 0.24864, "loss_box_dn_0": 0.8069, "loss_cls_dn_1": 0.17796, "loss_box_dn_1": 0.7967, "loss_cls_dn_2": 0.18598, "loss_box_dn_2": 0.78793, "loss_cls_dn_3": 0.18735, "loss_box_dn_3": 0.79419, "loss_cls_dn_4": 0.19017, "loss_box_dn_4": 0.80796, "loss_cls_dn_5": 0.19989, "loss_box_dn_5": 0.81778, "loss_dense_depth": 0.79019, "loss": 28.40838, "grad_norm": 51.57408, "time": 1.58427} -{"mode": "train", "epoch": 1, "iter": 116, "lr": 0.00015, "memory": 49164, "data_time": 0.07488, "loss_cls_0": 0.87901, "loss_box_0": 1.7911, "loss_cns_0": 0.61443, "loss_yns_0": 0.15326, "loss_cls_1": 0.96639, "loss_box_1": 1.89151, "loss_cns_1": 0.64438, "loss_yns_1": 0.15657, "loss_cls_2": 1.00301, "loss_box_2": 1.86858, "loss_cns_2": 0.65255, "loss_yns_2": 0.1553, "loss_cls_3": 1.01432, "loss_box_3": 1.84975, "loss_cns_3": 0.65398, "loss_yns_3": 0.15538, "loss_cls_4": 1.00686, "loss_box_4": 1.83088, "loss_cns_4": 0.65435, "loss_yns_4": 0.16144, "loss_cls_5": 1.00591, "loss_box_5": 1.89147, "loss_cns_5": 0.64839, "loss_yns_5": 0.15961, "loss_cls_dn_0": 0.25247, "loss_box_dn_0": 0.80188, "loss_cls_dn_1": 0.1807, "loss_box_dn_1": 0.80282, "loss_cls_dn_2": 0.19125, "loss_box_dn_2": 0.81445, "loss_cls_dn_3": 0.19439, "loss_box_dn_3": 0.83493, "loss_cls_dn_4": 0.20317, "loss_box_dn_4": 0.85726, "loss_cls_dn_5": 0.21056, "loss_box_dn_5": 0.90564, "loss_dense_depth": 0.80364, "loss": 28.86158, "grad_norm": 52.05959, "time": 1.62051} -{"mode": "train", "epoch": 1, "iter": 117, "lr": 0.00015, "memory": 49164, "data_time": 0.07714, "loss_cls_0": 0.86424, "loss_box_0": 1.75217, "loss_cns_0": 0.61788, "loss_yns_0": 0.1503, "loss_cls_1": 0.95816, "loss_box_1": 1.85466, "loss_cns_1": 0.64773, "loss_yns_1": 0.15762, "loss_cls_2": 0.97958, "loss_box_2": 1.81377, "loss_cns_2": 0.65525, "loss_yns_2": 0.15906, "loss_cls_3": 0.99133, "loss_box_3": 1.79887, "loss_cns_3": 0.65631, "loss_yns_3": 0.15498, "loss_cls_4": 1.00881, "loss_box_4": 1.79654, "loss_cns_4": 0.65365, "loss_yns_4": 0.16044, "loss_cls_5": 1.01673, "loss_box_5": 1.82888, "loss_cns_5": 0.65227, "loss_yns_5": 0.15725, "loss_cls_dn_0": 0.24338, "loss_box_dn_0": 0.79896, "loss_cls_dn_1": 0.18066, "loss_box_dn_1": 0.84753, "loss_cls_dn_2": 0.18773, "loss_box_dn_2": 0.85851, "loss_cls_dn_3": 0.19167, "loss_box_dn_3": 0.88074, "loss_cls_dn_4": 0.20021, "loss_box_dn_4": 0.91143, "loss_cls_dn_5": 0.21391, "loss_box_dn_5": 0.95231, "loss_dense_depth": 0.80395, "loss": 28.75743, "grad_norm": 45.01146, "time": 1.63268} -{"mode": "train", "epoch": 1, "iter": 118, "lr": 0.00015, "memory": 49164, "data_time": 0.07646, "loss_cls_0": 0.89543, "loss_box_0": 1.7376, "loss_cns_0": 0.62094, "loss_yns_0": 0.15415, "loss_cls_1": 0.96718, "loss_box_1": 1.83285, "loss_cns_1": 0.64364, "loss_yns_1": 0.15572, "loss_cls_2": 0.98549, "loss_box_2": 1.80359, "loss_cns_2": 0.64975, "loss_yns_2": 0.15642, "loss_cls_3": 0.99595, "loss_box_3": 1.8124, "loss_cns_3": 0.65251, "loss_yns_3": 0.15368, "loss_cls_4": 0.99108, "loss_box_4": 1.81555, "loss_cns_4": 0.64941, "loss_yns_4": 0.15478, "loss_cls_5": 1.00124, "loss_box_5": 1.79793, "loss_cns_5": 0.6477, "loss_yns_5": 0.15413, "loss_cls_dn_0": 0.23865, "loss_box_dn_0": 0.79418, "loss_cls_dn_1": 0.17836, "loss_box_dn_1": 0.87173, "loss_cls_dn_2": 0.1852, "loss_box_dn_2": 0.87707, "loss_cls_dn_3": 0.18798, "loss_box_dn_3": 0.90079, "loss_cls_dn_4": 0.18805, "loss_box_dn_4": 0.92424, "loss_cls_dn_5": 0.20119, "loss_box_dn_5": 0.94335, "loss_dense_depth": 0.80007, "loss": 28.71997, "grad_norm": 47.61914, "time": 1.56582} -{"mode": "train", "epoch": 1, "iter": 119, "lr": 0.00015, "memory": 49164, "data_time": 0.11313, "loss_cls_0": 0.87516, "loss_box_0": 1.73845, "loss_cns_0": 0.61173, "loss_yns_0": 0.15339, "loss_cls_1": 0.97387, "loss_box_1": 1.84965, "loss_cns_1": 0.64638, "loss_yns_1": 0.15453, "loss_cls_2": 0.99625, "loss_box_2": 1.78537, "loss_cns_2": 0.65015, "loss_yns_2": 0.15524, "loss_cls_3": 1.00741, "loss_box_3": 1.7843, "loss_cns_3": 0.65337, "loss_yns_3": 0.15277, "loss_cls_4": 1.01549, "loss_box_4": 1.77028, "loss_cns_4": 0.65271, "loss_yns_4": 0.15225, "loss_cls_5": 1.00568, "loss_box_5": 1.76079, "loss_cns_5": 0.64979, "loss_yns_5": 0.15519, "loss_cls_dn_0": 0.23492, "loss_box_dn_0": 0.79121, "loss_cls_dn_1": 0.17564, "loss_box_dn_1": 0.88955, "loss_cls_dn_2": 0.18095, "loss_box_dn_2": 0.87898, "loss_cls_dn_3": 0.18499, "loss_box_dn_3": 0.88498, "loss_cls_dn_4": 0.18465, "loss_box_dn_4": 0.88665, "loss_cls_dn_5": 0.19206, "loss_box_dn_5": 0.89624, "loss_dense_depth": 0.79949, "loss": 28.5305, "grad_norm": 37.32643, "time": 1.64046} -{"mode": "train", "epoch": 1, "iter": 120, "lr": 0.00015, "memory": 49164, "data_time": 0.08643, "loss_cls_0": 0.86745, "loss_box_0": 1.74652, "loss_cns_0": 0.60718, "loss_yns_0": 0.1518, "loss_cls_1": 0.96576, "loss_box_1": 1.8122, "loss_cns_1": 0.65097, "loss_yns_1": 0.15574, "loss_cls_2": 0.98985, "loss_box_2": 1.77305, "loss_cns_2": 0.65714, "loss_yns_2": 0.15411, "loss_cls_3": 1.00996, "loss_box_3": 1.78365, "loss_cns_3": 0.65759, "loss_yns_3": 0.15347, "loss_cls_4": 0.98391, "loss_box_4": 1.77691, "loss_cns_4": 0.65623, "loss_yns_4": 0.15411, "loss_cls_5": 0.98877, "loss_box_5": 1.78484, "loss_cns_5": 0.65424, "loss_yns_5": 0.15938, "loss_cls_dn_0": 0.23845, "loss_box_dn_0": 0.80815, "loss_cls_dn_1": 0.1747, "loss_box_dn_1": 0.80533, "loss_cls_dn_2": 0.17817, "loss_box_dn_2": 0.78228, "loss_cls_dn_3": 0.18627, "loss_box_dn_3": 0.78766, "loss_cls_dn_4": 0.18054, "loss_box_dn_4": 0.78458, "loss_cls_dn_5": 0.19072, "loss_box_dn_5": 0.79057, "loss_dense_depth": 0.82346, "loss": 28.0257, "grad_norm": 45.24397, "time": 1.57057} -{"mode": "train", "epoch": 1, "iter": 121, "lr": 0.00015, "memory": 49164, "data_time": 0.1157, "loss_cls_0": 0.87928, "loss_box_0": 1.77471, "loss_cns_0": 0.61992, "loss_yns_0": 0.1535, "loss_cls_1": 0.97756, "loss_box_1": 1.85893, "loss_cns_1": 0.64867, "loss_yns_1": 0.15507, "loss_cls_2": 0.99626, "loss_box_2": 1.82824, "loss_cns_2": 0.65422, "loss_yns_2": 0.15575, "loss_cls_3": 1.00937, "loss_box_3": 1.84007, "loss_cns_3": 0.65701, "loss_yns_3": 0.15813, "loss_cls_4": 1.00214, "loss_box_4": 1.8203, "loss_cns_4": 0.6568, "loss_yns_4": 0.16116, "loss_cls_5": 1.00801, "loss_box_5": 1.83412, "loss_cns_5": 0.65553, "loss_yns_5": 0.1593, "loss_cls_dn_0": 0.23723, "loss_box_dn_0": 0.80635, "loss_cls_dn_1": 0.17046, "loss_box_dn_1": 0.80129, "loss_cls_dn_2": 0.17751, "loss_box_dn_2": 0.78558, "loss_cls_dn_3": 0.1789, "loss_box_dn_3": 0.80074, "loss_cls_dn_4": 0.17905, "loss_box_dn_4": 0.80409, "loss_cls_dn_5": 0.18924, "loss_box_dn_5": 0.82077, "loss_dense_depth": 0.79644, "loss": 28.41169, "grad_norm": 49.87539, "time": 1.6693} -{"mode": "train", "epoch": 1, "iter": 122, "lr": 0.00015, "memory": 49164, "data_time": 0.17617, "loss_cls_0": 0.90235, "loss_box_0": 1.7893, "loss_cns_0": 0.62067, "loss_yns_0": 0.15838, "loss_cls_1": 0.99314, "loss_box_1": 1.85195, "loss_cns_1": 0.64535, "loss_yns_1": 0.15836, "loss_cls_2": 1.01118, "loss_box_2": 1.80367, "loss_cns_2": 0.65103, "loss_yns_2": 0.16015, "loss_cls_3": 1.0078, "loss_box_3": 1.80589, "loss_cns_3": 0.6506, "loss_yns_3": 0.16211, "loss_cls_4": 1.01409, "loss_box_4": 1.80056, "loss_cns_4": 0.65167, "loss_yns_4": 0.16256, "loss_cls_5": 1.00681, "loss_box_5": 1.85019, "loss_cns_5": 0.64518, "loss_yns_5": 0.16274, "loss_cls_dn_0": 0.2363, "loss_box_dn_0": 0.80503, "loss_cls_dn_1": 0.16961, "loss_box_dn_1": 0.80472, "loss_cls_dn_2": 0.17791, "loss_box_dn_2": 0.79386, "loss_cls_dn_3": 0.17381, "loss_box_dn_3": 0.81601, "loss_cls_dn_4": 0.17936, "loss_box_dn_4": 0.83649, "loss_cls_dn_5": 0.18651, "loss_box_dn_5": 0.87429, "loss_dense_depth": 0.77428, "loss": 28.49392, "grad_norm": 48.75888, "time": 1.67678} -{"mode": "train", "epoch": 1, "iter": 123, "lr": 0.00015, "memory": 49164, "data_time": 0.07547, "loss_cls_0": 0.87282, "loss_box_0": 1.75386, "loss_cns_0": 0.62244, "loss_yns_0": 0.15614, "loss_cls_1": 0.96388, "loss_box_1": 1.84208, "loss_cns_1": 0.64551, "loss_yns_1": 0.15623, "loss_cls_2": 0.98955, "loss_box_2": 1.80296, "loss_cns_2": 0.65131, "loss_yns_2": 0.15772, "loss_cls_3": 0.99939, "loss_box_3": 1.80425, "loss_cns_3": 0.65217, "loss_yns_3": 0.1626, "loss_cls_4": 1.00132, "loss_box_4": 1.83572, "loss_cns_4": 0.65157, "loss_yns_4": 0.15883, "loss_cls_5": 0.99795, "loss_box_5": 1.85561, "loss_cns_5": 0.64873, "loss_yns_5": 0.16307, "loss_cls_dn_0": 0.23164, "loss_box_dn_0": 0.79182, "loss_cls_dn_1": 0.16888, "loss_box_dn_1": 0.86216, "loss_cls_dn_2": 0.17911, "loss_box_dn_2": 0.8635, "loss_cls_dn_3": 0.17511, "loss_box_dn_3": 0.89699, "loss_cls_dn_4": 0.18272, "loss_box_dn_4": 0.94375, "loss_cls_dn_5": 0.189, "loss_box_dn_5": 0.98333, "loss_dense_depth": 0.81017, "loss": 28.82388, "grad_norm": 50.91918, "time": 1.61362} -{"mode": "train", "epoch": 1, "iter": 124, "lr": 0.00015, "memory": 49164, "data_time": 0.07303, "loss_cls_0": 0.87512, "loss_box_0": 1.76444, "loss_cns_0": 0.62746, "loss_yns_0": 0.15638, "loss_cls_1": 0.96536, "loss_box_1": 1.80614, "loss_cns_1": 0.65746, "loss_yns_1": 0.15722, "loss_cls_2": 1.00181, "loss_box_2": 1.79268, "loss_cns_2": 0.65562, "loss_yns_2": 0.15646, "loss_cls_3": 0.99888, "loss_box_3": 1.81783, "loss_cns_3": 0.65493, "loss_yns_3": 0.16002, "loss_cls_4": 1.00399, "loss_box_4": 1.84368, "loss_cns_4": 0.65101, "loss_yns_4": 0.15828, "loss_cls_5": 1.00898, "loss_box_5": 1.84183, "loss_cns_5": 0.65562, "loss_yns_5": 0.16095, "loss_cls_dn_0": 0.23312, "loss_box_dn_0": 0.79285, "loss_cls_dn_1": 0.16655, "loss_box_dn_1": 0.87626, "loss_cls_dn_2": 0.17625, "loss_box_dn_2": 0.89084, "loss_cls_dn_3": 0.17242, "loss_box_dn_3": 0.92325, "loss_cls_dn_4": 0.17926, "loss_box_dn_4": 0.95908, "loss_cls_dn_5": 0.19302, "loss_box_dn_5": 0.9873, "loss_dense_depth": 0.77069, "loss": 28.89303, "grad_norm": 57.42082, "time": 1.54798} -{"mode": "train", "epoch": 1, "iter": 125, "lr": 0.00015, "memory": 49164, "data_time": 0.07244, "loss_cls_0": 0.87869, "loss_box_0": 1.79115, "loss_cns_0": 0.62544, "loss_yns_0": 0.15756, "loss_cls_1": 0.96869, "loss_box_1": 1.78681, "loss_cns_1": 0.65637, "loss_yns_1": 0.15796, "loss_cls_2": 1.01636, "loss_box_2": 1.75593, "loss_cns_2": 0.65544, "loss_yns_2": 0.15786, "loss_cls_3": 1.00547, "loss_box_3": 1.77664, "loss_cns_3": 0.65699, "loss_yns_3": 0.15759, "loss_cls_4": 1.015, "loss_box_4": 1.77014, "loss_cns_4": 0.65566, "loss_yns_4": 0.15843, "loss_cls_5": 1.01874, "loss_box_5": 1.78557, "loss_cns_5": 0.65674, "loss_yns_5": 0.15936, "loss_cls_dn_0": 0.2348, "loss_box_dn_0": 0.78584, "loss_cls_dn_1": 0.1697, "loss_box_dn_1": 0.89588, "loss_cls_dn_2": 0.17988, "loss_box_dn_2": 0.89836, "loss_cls_dn_3": 0.17515, "loss_box_dn_3": 0.9148, "loss_cls_dn_4": 0.17955, "loss_box_dn_4": 0.92982, "loss_cls_dn_5": 0.1963, "loss_box_dn_5": 0.95146, "loss_dense_depth": 0.77215, "loss": 28.70827, "grad_norm": 42.72639, "time": 1.60034} -{"mode": "train", "epoch": 1, "iter": 126, "lr": 0.00015, "memory": 49164, "data_time": 0.07815, "loss_cls_0": 0.91146, "loss_box_0": 1.778, "loss_cns_0": 0.62413, "loss_yns_0": 0.15598, "loss_cls_1": 0.96319, "loss_box_1": 1.85379, "loss_cns_1": 0.64686, "loss_yns_1": 0.15815, "loss_cls_2": 0.99658, "loss_box_2": 1.81074, "loss_cns_2": 0.65233, "loss_yns_2": 0.15874, "loss_cls_3": 1.00611, "loss_box_3": 1.80887, "loss_cns_3": 0.6528, "loss_yns_3": 0.16064, "loss_cls_4": 1.0081, "loss_box_4": 1.79329, "loss_cns_4": 0.65261, "loss_yns_4": 0.15959, "loss_cls_5": 1.01218, "loss_box_5": 1.83038, "loss_cns_5": 0.64972, "loss_yns_5": 0.15905, "loss_cls_dn_0": 0.23211, "loss_box_dn_0": 0.78694, "loss_cls_dn_1": 0.17187, "loss_box_dn_1": 0.84928, "loss_cls_dn_2": 0.17965, "loss_box_dn_2": 0.82993, "loss_cls_dn_3": 0.17758, "loss_box_dn_3": 0.82931, "loss_cls_dn_4": 0.17987, "loss_box_dn_4": 0.83544, "loss_cls_dn_5": 0.19157, "loss_box_dn_5": 0.84872, "loss_dense_depth": 0.77175, "loss": 28.48732, "grad_norm": 46.70768, "time": 1.59351} -{"mode": "train", "epoch": 1, "iter": 127, "lr": 0.00015, "memory": 49164, "data_time": 0.07091, "loss_cls_0": 0.87464, "loss_box_0": 1.72238, "loss_cns_0": 0.618, "loss_yns_0": 0.15442, "loss_cls_1": 0.95116, "loss_box_1": 1.85442, "loss_cns_1": 0.64575, "loss_yns_1": 0.15368, "loss_cls_2": 0.9882, "loss_box_2": 1.76909, "loss_cns_2": 0.65296, "loss_yns_2": 0.15622, "loss_cls_3": 0.9992, "loss_box_3": 1.7615, "loss_cns_3": 0.6543, "loss_yns_3": 0.16121, "loss_cls_4": 1.01623, "loss_box_4": 1.7648, "loss_cns_4": 0.65424, "loss_yns_4": 0.16296, "loss_cls_5": 1.01718, "loss_box_5": 1.77137, "loss_cns_5": 0.65573, "loss_yns_5": 0.15802, "loss_cls_dn_0": 0.23436, "loss_box_dn_0": 0.78401, "loss_cls_dn_1": 0.16884, "loss_box_dn_1": 0.78543, "loss_cls_dn_2": 0.17542, "loss_box_dn_2": 0.75175, "loss_cls_dn_3": 0.17276, "loss_box_dn_3": 0.7448, "loss_cls_dn_4": 0.17999, "loss_box_dn_4": 0.75356, "loss_cls_dn_5": 0.18928, "loss_box_dn_5": 0.76232, "loss_dense_depth": 0.77336, "loss": 27.79351, "grad_norm": 39.67825, "time": 1.57641} -{"mode": "train", "epoch": 1, "iter": 128, "lr": 0.00015, "memory": 49164, "data_time": 0.12186, "loss_cls_0": 0.91396, "loss_box_0": 1.69099, "loss_cns_0": 0.61086, "loss_yns_0": 0.15072, "loss_cls_1": 0.9507, "loss_box_1": 1.80593, "loss_cns_1": 0.64745, "loss_yns_1": 0.15538, "loss_cls_2": 0.98697, "loss_box_2": 1.75834, "loss_cns_2": 0.64943, "loss_yns_2": 0.15754, "loss_cls_3": 0.99855, "loss_box_3": 1.73776, "loss_cns_3": 0.65119, "loss_yns_3": 0.15944, "loss_cls_4": 1.01625, "loss_box_4": 1.73986, "loss_cns_4": 0.64979, "loss_yns_4": 0.16048, "loss_cls_5": 1.00286, "loss_box_5": 1.75849, "loss_cns_5": 0.654, "loss_yns_5": 0.15838, "loss_cls_dn_0": 0.23803, "loss_box_dn_0": 0.7929, "loss_cls_dn_1": 0.16778, "loss_box_dn_1": 0.75277, "loss_cls_dn_2": 0.1729, "loss_box_dn_2": 0.74677, "loss_cls_dn_3": 0.17233, "loss_box_dn_3": 0.75745, "loss_cls_dn_4": 0.18003, "loss_box_dn_4": 0.77744, "loss_cls_dn_5": 0.18774, "loss_box_dn_5": 0.80555, "loss_dense_depth": 0.81666, "loss": 27.73368, "grad_norm": 43.19228, "time": 1.63171} -{"mode": "train", "epoch": 1, "iter": 129, "lr": 0.00015, "memory": 49164, "data_time": 0.07073, "loss_cls_0": 0.85509, "loss_box_0": 1.74839, "loss_cns_0": 0.61544, "loss_yns_0": 0.15284, "loss_cls_1": 0.95173, "loss_box_1": 1.75476, "loss_cns_1": 0.64592, "loss_yns_1": 0.15581, "loss_cls_2": 0.98734, "loss_box_2": 1.72928, "loss_cns_2": 0.65176, "loss_yns_2": 0.15768, "loss_cls_3": 0.98671, "loss_box_3": 1.70014, "loss_cns_3": 0.65331, "loss_yns_3": 0.15817, "loss_cls_4": 0.99343, "loss_box_4": 1.71582, "loss_cns_4": 0.65166, "loss_yns_4": 0.15831, "loss_cls_5": 0.99654, "loss_box_5": 1.71553, "loss_cns_5": 0.65285, "loss_yns_5": 0.15939, "loss_cls_dn_0": 0.23083, "loss_box_dn_0": 0.79786, "loss_cls_dn_1": 0.16524, "loss_box_dn_1": 0.77749, "loss_cls_dn_2": 0.17196, "loss_box_dn_2": 0.78939, "loss_cls_dn_3": 0.1732, "loss_box_dn_3": 0.81143, "loss_cls_dn_4": 0.17762, "loss_box_dn_4": 0.84546, "loss_cls_dn_5": 0.18769, "loss_box_dn_5": 0.8743, "loss_dense_depth": 0.79081, "loss": 27.74118, "grad_norm": 48.1232, "time": 1.57597} -{"mode": "train", "epoch": 1, "iter": 130, "lr": 0.00015, "memory": 49164, "data_time": 0.07465, "loss_cls_0": 0.89901, "loss_box_0": 1.78739, "loss_cns_0": 0.61818, "loss_yns_0": 0.15203, "loss_cls_1": 0.95883, "loss_box_1": 1.78015, "loss_cns_1": 0.64593, "loss_yns_1": 0.15417, "loss_cls_2": 0.98634, "loss_box_2": 1.76567, "loss_cns_2": 0.65211, "loss_yns_2": 0.15332, "loss_cls_3": 1.00013, "loss_box_3": 1.7281, "loss_cns_3": 0.65196, "loss_yns_3": 0.15409, "loss_cls_4": 1.00766, "loss_box_4": 1.73424, "loss_cns_4": 0.6486, "loss_yns_4": 0.1555, "loss_cls_5": 1.01501, "loss_box_5": 1.73896, "loss_cns_5": 0.65427, "loss_yns_5": 0.15414, "loss_cls_dn_0": 0.23071, "loss_box_dn_0": 0.80476, "loss_cls_dn_1": 0.16701, "loss_box_dn_1": 0.829, "loss_cls_dn_2": 0.17228, "loss_box_dn_2": 0.84867, "loss_cls_dn_3": 0.17453, "loss_box_dn_3": 0.86283, "loss_cls_dn_4": 0.17697, "loss_box_dn_4": 0.88946, "loss_cls_dn_5": 0.18784, "loss_box_dn_5": 0.91321, "loss_dense_depth": 0.79637, "loss": 28.24942, "grad_norm": 46.74036, "time": 1.61423} -{"mode": "train", "epoch": 1, "iter": 131, "lr": 0.00015, "memory": 49164, "data_time": 0.08168, "loss_cls_0": 0.89426, "loss_box_0": 1.77059, "loss_cns_0": 0.62284, "loss_yns_0": 0.15248, "loss_cls_1": 0.9511, "loss_box_1": 1.78925, "loss_cns_1": 0.64903, "loss_yns_1": 0.15433, "loss_cls_2": 0.98384, "loss_box_2": 1.7627, "loss_cns_2": 0.6556, "loss_yns_2": 0.15477, "loss_cls_3": 0.99855, "loss_box_3": 1.73013, "loss_cns_3": 0.6584, "loss_yns_3": 0.15704, "loss_cls_4": 1.0281, "loss_box_4": 1.69914, "loss_cns_4": 0.65622, "loss_yns_4": 0.16127, "loss_cls_5": 0.99863, "loss_box_5": 1.71532, "loss_cns_5": 0.66072, "loss_yns_5": 0.1557, "loss_cls_dn_0": 0.23217, "loss_box_dn_0": 0.79895, "loss_cls_dn_1": 0.16535, "loss_box_dn_1": 0.81717, "loss_cls_dn_2": 0.17394, "loss_box_dn_2": 0.8254, "loss_cls_dn_3": 0.17308, "loss_box_dn_3": 0.82929, "loss_cls_dn_4": 0.18059, "loss_box_dn_4": 0.82988, "loss_cls_dn_5": 0.18687, "loss_box_dn_5": 0.85078, "loss_dense_depth": 0.81674, "loss": 28.04021, "grad_norm": 38.98964, "time": 1.61642} -{"mode": "train", "epoch": 1, "iter": 132, "lr": 0.00015, "memory": 49164, "data_time": 0.07646, "loss_cls_0": 0.86999, "loss_box_0": 1.74278, "loss_cns_0": 0.61818, "loss_yns_0": 0.14911, "loss_cls_1": 0.93713, "loss_box_1": 1.81129, "loss_cns_1": 0.6459, "loss_yns_1": 0.15045, "loss_cls_2": 0.9731, "loss_box_2": 1.77419, "loss_cns_2": 0.65537, "loss_yns_2": 0.15053, "loss_cls_3": 0.98174, "loss_box_3": 1.77201, "loss_cns_3": 0.65497, "loss_yns_3": 0.15315, "loss_cls_4": 0.98594, "loss_box_4": 1.73423, "loss_cns_4": 0.65774, "loss_yns_4": 0.15616, "loss_cls_5": 0.9934, "loss_box_5": 1.75381, "loss_cns_5": 0.6568, "loss_yns_5": 0.15366, "loss_cls_dn_0": 0.23559, "loss_box_dn_0": 0.78614, "loss_cls_dn_1": 0.16665, "loss_box_dn_1": 0.77205, "loss_cls_dn_2": 0.17793, "loss_box_dn_2": 0.76737, "loss_cls_dn_3": 0.1741, "loss_box_dn_3": 0.77423, "loss_cls_dn_4": 0.18295, "loss_box_dn_4": 0.76441, "loss_cls_dn_5": 0.18868, "loss_box_dn_5": 0.78025, "loss_dense_depth": 0.82351, "loss": 27.72549, "grad_norm": 46.63246, "time": 1.56221} -{"mode": "train", "epoch": 1, "iter": 133, "lr": 0.00015, "memory": 49164, "data_time": 0.10275, "loss_cls_0": 0.90376, "loss_box_0": 1.71569, "loss_cns_0": 0.60859, "loss_yns_0": 0.14664, "loss_cls_1": 0.94267, "loss_box_1": 1.83246, "loss_cns_1": 0.64271, "loss_yns_1": 0.14995, "loss_cls_2": 0.98203, "loss_box_2": 1.75002, "loss_cns_2": 0.64643, "loss_yns_2": 0.15108, "loss_cls_3": 1.00695, "loss_box_3": 1.75718, "loss_cns_3": 0.65215, "loss_yns_3": 0.15178, "loss_cls_4": 0.99514, "loss_box_4": 1.72475, "loss_cns_4": 0.65284, "loss_yns_4": 0.15138, "loss_cls_5": 0.99585, "loss_box_5": 1.72598, "loss_cns_5": 0.65438, "loss_yns_5": 0.15192, "loss_cls_dn_0": 0.24095, "loss_box_dn_0": 0.78942, "loss_cls_dn_1": 0.16924, "loss_box_dn_1": 0.74597, "loss_cls_dn_2": 0.17706, "loss_box_dn_2": 0.72622, "loss_cls_dn_3": 0.17255, "loss_box_dn_3": 0.73097, "loss_cls_dn_4": 0.1777, "loss_box_dn_4": 0.71847, "loss_cls_dn_5": 0.18784, "loss_box_dn_5": 0.7181, "loss_dense_depth": 0.79022, "loss": 27.43703, "grad_norm": 35.08681, "time": 1.58958} -{"mode": "train", "epoch": 1, "iter": 134, "lr": 0.00015, "memory": 49164, "data_time": 0.07363, "loss_cls_0": 0.84869, "loss_box_0": 1.70519, "loss_cns_0": 0.62099, "loss_yns_0": 0.14878, "loss_cls_1": 0.91472, "loss_box_1": 1.83043, "loss_cns_1": 0.64248, "loss_yns_1": 0.14925, "loss_cls_2": 0.9554, "loss_box_2": 1.75619, "loss_cns_2": 0.65064, "loss_yns_2": 0.15038, "loss_cls_3": 0.96218, "loss_box_3": 1.72694, "loss_cns_3": 0.65471, "loss_yns_3": 0.1504, "loss_cls_4": 0.98335, "loss_box_4": 1.72109, "loss_cns_4": 0.65482, "loss_yns_4": 0.15435, "loss_cls_5": 0.97953, "loss_box_5": 1.71467, "loss_cns_5": 0.65554, "loss_yns_5": 0.15038, "loss_cls_dn_0": 0.22992, "loss_box_dn_0": 0.78841, "loss_cls_dn_1": 0.16535, "loss_box_dn_1": 0.72131, "loss_cls_dn_2": 0.17015, "loss_box_dn_2": 0.70066, "loss_cls_dn_3": 0.16809, "loss_box_dn_3": 0.69693, "loss_cls_dn_4": 0.16916, "loss_box_dn_4": 0.69718, "loss_cls_dn_5": 0.18413, "loss_box_dn_5": 0.69803, "loss_dense_depth": 0.8444, "loss": 27.11483, "grad_norm": 39.17166, "time": 1.56666} -{"mode": "train", "epoch": 1, "iter": 135, "lr": 0.00015, "memory": 49164, "data_time": 0.07216, "loss_cls_0": 0.88151, "loss_box_0": 1.76556, "loss_cns_0": 0.62527, "loss_yns_0": 0.15406, "loss_cls_1": 0.93222, "loss_box_1": 1.81945, "loss_cns_1": 0.64385, "loss_yns_1": 0.15199, "loss_cls_2": 0.95952, "loss_box_2": 1.7876, "loss_cns_2": 0.65181, "loss_yns_2": 0.15205, "loss_cls_3": 0.98917, "loss_box_3": 1.77866, "loss_cns_3": 0.6512, "loss_yns_3": 0.15541, "loss_cls_4": 0.99469, "loss_box_4": 1.80419, "loss_cns_4": 0.65269, "loss_yns_4": 0.15657, "loss_cls_5": 0.98586, "loss_box_5": 1.80487, "loss_cns_5": 0.65284, "loss_yns_5": 0.15393, "loss_cls_dn_0": 0.23532, "loss_box_dn_0": 0.79143, "loss_cls_dn_1": 0.16339, "loss_box_dn_1": 0.73849, "loss_cls_dn_2": 0.16867, "loss_box_dn_2": 0.72836, "loss_cls_dn_3": 0.17162, "loss_box_dn_3": 0.73786, "loss_cls_dn_4": 0.17233, "loss_box_dn_4": 0.7596, "loss_cls_dn_5": 0.18088, "loss_box_dn_5": 0.77481, "loss_dense_depth": 0.87677, "loss": 27.80448, "grad_norm": 50.31979, "time": 1.59391} -{"mode": "train", "epoch": 1, "iter": 136, "lr": 0.00015, "memory": 49164, "data_time": 0.07408, "loss_cls_0": 0.89437, "loss_box_0": 1.71465, "loss_cns_0": 0.62909, "loss_yns_0": 0.14957, "loss_cls_1": 0.90956, "loss_box_1": 1.80237, "loss_cns_1": 0.6485, "loss_yns_1": 0.14797, "loss_cls_2": 0.94187, "loss_box_2": 1.75586, "loss_cns_2": 0.65856, "loss_yns_2": 0.15192, "loss_cls_3": 0.96176, "loss_box_3": 1.75714, "loss_cns_3": 0.65342, "loss_yns_3": 0.15515, "loss_cls_4": 0.97241, "loss_box_4": 1.75234, "loss_cns_4": 0.655, "loss_yns_4": 0.1535, "loss_cls_5": 0.94859, "loss_box_5": 1.74318, "loss_cns_5": 0.65604, "loss_yns_5": 0.15394, "loss_cls_dn_0": 0.22668, "loss_box_dn_0": 0.77928, "loss_cls_dn_1": 0.1593, "loss_box_dn_1": 0.77375, "loss_cls_dn_2": 0.16352, "loss_box_dn_2": 0.77341, "loss_cls_dn_3": 0.16392, "loss_box_dn_3": 0.79067, "loss_cls_dn_4": 0.1682, "loss_box_dn_4": 0.81269, "loss_cls_dn_5": 0.17225, "loss_box_dn_5": 0.83113, "loss_dense_depth": 0.849, "loss": 27.63053, "grad_norm": 46.13214, "time": 1.60624} -{"mode": "train", "epoch": 1, "iter": 137, "lr": 0.00015, "memory": 49164, "data_time": 0.0727, "loss_cls_0": 0.83755, "loss_box_0": 1.73996, "loss_cns_0": 0.62316, "loss_yns_0": 0.15041, "loss_cls_1": 0.91784, "loss_box_1": 1.80456, "loss_cns_1": 0.64768, "loss_yns_1": 0.14828, "loss_cls_2": 0.93909, "loss_box_2": 1.74552, "loss_cns_2": 0.6513, "loss_yns_2": 0.15207, "loss_cls_3": 0.95929, "loss_box_3": 1.73917, "loss_cns_3": 0.65195, "loss_yns_3": 0.1518, "loss_cls_4": 0.95192, "loss_box_4": 1.75203, "loss_cns_4": 0.65121, "loss_yns_4": 0.15167, "loss_cls_5": 0.94905, "loss_box_5": 1.79557, "loss_cns_5": 0.64416, "loss_yns_5": 0.15135, "loss_cls_dn_0": 0.23881, "loss_box_dn_0": 0.78478, "loss_cls_dn_1": 0.15636, "loss_box_dn_1": 0.83288, "loss_cls_dn_2": 0.16029, "loss_box_dn_2": 0.83009, "loss_cls_dn_3": 0.16367, "loss_box_dn_3": 0.84325, "loss_cls_dn_4": 0.16194, "loss_box_dn_4": 0.86628, "loss_cls_dn_5": 0.1692, "loss_box_dn_5": 0.89289, "loss_dense_depth": 0.87656, "loss": 27.88357, "grad_norm": 50.79967, "time": 1.56365} -{"mode": "train", "epoch": 1, "iter": 138, "lr": 0.00015, "memory": 49164, "data_time": 0.07615, "loss_cls_0": 0.84973, "loss_box_0": 1.70763, "loss_cns_0": 0.6262, "loss_yns_0": 0.14777, "loss_cls_1": 0.91767, "loss_box_1": 1.80761, "loss_cns_1": 0.64642, "loss_yns_1": 0.14694, "loss_cls_2": 0.93766, "loss_box_2": 1.74382, "loss_cns_2": 0.65025, "loss_yns_2": 0.15011, "loss_cls_3": 0.94604, "loss_box_3": 1.73884, "loss_cns_3": 0.65677, "loss_yns_3": 0.14973, "loss_cls_4": 0.95962, "loss_box_4": 1.74722, "loss_cns_4": 0.65812, "loss_yns_4": 0.14911, "loss_cls_5": 0.94843, "loss_box_5": 1.76629, "loss_cns_5": 0.65118, "loss_yns_5": 0.15004, "loss_cls_dn_0": 0.23942, "loss_box_dn_0": 0.78647, "loss_cls_dn_1": 0.15317, "loss_box_dn_1": 0.84753, "loss_cls_dn_2": 0.15801, "loss_box_dn_2": 0.83309, "loss_cls_dn_3": 0.161, "loss_box_dn_3": 0.83828, "loss_cls_dn_4": 0.16223, "loss_box_dn_4": 0.85506, "loss_cls_dn_5": 0.17189, "loss_box_dn_5": 0.868, "loss_dense_depth": 0.85439, "loss": 27.78174, "grad_norm": 37.88572, "time": 1.64017} -{"mode": "train", "epoch": 1, "iter": 139, "lr": 0.00016, "memory": 49164, "data_time": 0.11372, "loss_cls_0": 0.83131, "loss_box_0": 1.68236, "loss_cns_0": 0.62866, "loss_yns_0": 0.146, "loss_cls_1": 0.89737, "loss_box_1": 1.82363, "loss_cns_1": 0.6447, "loss_yns_1": 0.14691, "loss_cls_2": 0.93094, "loss_box_2": 1.7801, "loss_cns_2": 0.64871, "loss_yns_2": 0.1476, "loss_cls_3": 0.95881, "loss_box_3": 1.77347, "loss_cns_3": 0.64877, "loss_yns_3": 0.14869, "loss_cls_4": 0.93478, "loss_box_4": 1.76242, "loss_cns_4": 0.65333, "loss_yns_4": 0.14782, "loss_cls_5": 0.98114, "loss_box_5": 1.73569, "loss_cns_5": 0.65696, "loss_yns_5": 0.14822, "loss_cls_dn_0": 0.22844, "loss_box_dn_0": 0.78222, "loss_cls_dn_1": 0.15258, "loss_box_dn_1": 0.81099, "loss_cls_dn_2": 0.15835, "loss_box_dn_2": 0.80087, "loss_cls_dn_3": 0.15913, "loss_box_dn_3": 0.80075, "loss_cls_dn_4": 0.1641, "loss_box_dn_4": 0.80188, "loss_cls_dn_5": 0.18184, "loss_box_dn_5": 0.79321, "loss_dense_depth": 0.85329, "loss": 27.54601, "grad_norm": 56.16213, "time": 1.6221} -{"mode": "train", "epoch": 1, "iter": 140, "lr": 0.00016, "memory": 49164, "data_time": 0.08652, "loss_cls_0": 0.82262, "loss_box_0": 1.67849, "loss_cns_0": 0.6293, "loss_yns_0": 0.14487, "loss_cls_1": 0.90142, "loss_box_1": 1.76325, "loss_cns_1": 0.65271, "loss_yns_1": 0.14744, "loss_cls_2": 0.92673, "loss_box_2": 1.71276, "loss_cns_2": 0.66101, "loss_yns_2": 0.14731, "loss_cls_3": 0.94539, "loss_box_3": 1.69358, "loss_cns_3": 0.65729, "loss_yns_3": 0.14826, "loss_cls_4": 0.93409, "loss_box_4": 1.69484, "loss_cns_4": 0.65837, "loss_yns_4": 0.14781, "loss_cls_5": 0.93707, "loss_box_5": 1.68539, "loss_cns_5": 0.66152, "loss_yns_5": 0.14762, "loss_cls_dn_0": 0.22384, "loss_box_dn_0": 0.78478, "loss_cls_dn_1": 0.14597, "loss_box_dn_1": 0.78188, "loss_cls_dn_2": 0.1532, "loss_box_dn_2": 0.76618, "loss_cls_dn_3": 0.15477, "loss_box_dn_3": 0.7591, "loss_cls_dn_4": 0.15951, "loss_box_dn_4": 0.75618, "loss_cls_dn_5": 0.17052, "loss_box_dn_5": 0.75652, "loss_dense_depth": 0.80564, "loss": 26.91726, "grad_norm": 39.45575, "time": 1.60835} -{"mode": "train", "epoch": 1, "iter": 141, "lr": 0.00016, "memory": 49164, "data_time": 0.11237, "loss_cls_0": 0.85436, "loss_box_0": 1.77247, "loss_cns_0": 0.6256, "loss_yns_0": 0.14842, "loss_cls_1": 0.92564, "loss_box_1": 1.78057, "loss_cns_1": 0.65172, "loss_yns_1": 0.14597, "loss_cls_2": 0.95497, "loss_box_2": 1.76397, "loss_cns_2": 0.64932, "loss_yns_2": 0.14763, "loss_cls_3": 0.95383, "loss_box_3": 1.77723, "loss_cns_3": 0.65442, "loss_yns_3": 0.14928, "loss_cls_4": 0.96622, "loss_box_4": 1.77314, "loss_cns_4": 0.65204, "loss_yns_4": 0.14924, "loss_cls_5": 0.96623, "loss_box_5": 1.79531, "loss_cns_5": 0.64728, "loss_yns_5": 0.1505, "loss_cls_dn_0": 0.23734, "loss_box_dn_0": 0.79972, "loss_cls_dn_1": 0.15148, "loss_box_dn_1": 0.74244, "loss_cls_dn_2": 0.15806, "loss_box_dn_2": 0.73951, "loss_cls_dn_3": 0.16193, "loss_box_dn_3": 0.75075, "loss_cls_dn_4": 0.16642, "loss_box_dn_4": 0.75824, "loss_cls_dn_5": 0.17345, "loss_box_dn_5": 0.78276, "loss_dense_depth": 0.8357, "loss": 27.51316, "grad_norm": 61.22596, "time": 1.70998} -{"mode": "train", "epoch": 1, "iter": 142, "lr": 0.00016, "memory": 49164, "data_time": 0.17052, "loss_cls_0": 0.86809, "loss_box_0": 1.77068, "loss_cns_0": 0.62169, "loss_yns_0": 0.15068, "loss_cls_1": 0.93866, "loss_box_1": 1.82913, "loss_cns_1": 0.64156, "loss_yns_1": 0.14765, "loss_cls_2": 0.96825, "loss_box_2": 1.76797, "loss_cns_2": 0.64633, "loss_yns_2": 0.14969, "loss_cls_3": 0.97014, "loss_box_3": 1.77276, "loss_cns_3": 0.65414, "loss_yns_3": 0.15244, "loss_cls_4": 0.97436, "loss_box_4": 1.77636, "loss_cns_4": 0.65304, "loss_yns_4": 0.15282, "loss_cls_5": 0.99053, "loss_box_5": 1.79978, "loss_cns_5": 0.6504, "loss_yns_5": 0.15383, "loss_cls_dn_0": 0.23398, "loss_box_dn_0": 0.79058, "loss_cls_dn_1": 0.15345, "loss_box_dn_1": 0.75096, "loss_cls_dn_2": 0.15851, "loss_box_dn_2": 0.74905, "loss_cls_dn_3": 0.16393, "loss_box_dn_3": 0.76743, "loss_cls_dn_4": 0.16949, "loss_box_dn_4": 0.78748, "loss_cls_dn_5": 0.17488, "loss_box_dn_5": 0.82271, "loss_dense_depth": 0.82694, "loss": 27.75038, "grad_norm": 54.53135, "time": 1.76083} -{"mode": "train", "epoch": 1, "iter": 143, "lr": 0.00016, "memory": 49164, "data_time": 0.06453, "loss_cls_0": 0.84433, "loss_box_0": 1.77182, "loss_cns_0": 0.62121, "loss_yns_0": 0.14761, "loss_cls_1": 0.91958, "loss_box_1": 1.82626, "loss_cns_1": 0.64558, "loss_yns_1": 0.14976, "loss_cls_2": 0.94545, "loss_box_2": 1.80572, "loss_cns_2": 0.64831, "loss_yns_2": 0.15004, "loss_cls_3": 0.95522, "loss_box_3": 1.80884, "loss_cns_3": 0.64886, "loss_yns_3": 0.15213, "loss_cls_4": 0.95438, "loss_box_4": 1.83291, "loss_cns_4": 0.64884, "loss_yns_4": 0.15285, "loss_cls_5": 0.98588, "loss_box_5": 1.88341, "loss_cns_5": 0.65175, "loss_yns_5": 0.15145, "loss_cls_dn_0": 0.2258, "loss_box_dn_0": 0.78977, "loss_cls_dn_1": 0.15702, "loss_box_dn_1": 0.81191, "loss_cls_dn_2": 0.16046, "loss_box_dn_2": 0.82158, "loss_cls_dn_3": 0.16165, "loss_box_dn_3": 0.84459, "loss_cls_dn_4": 0.16765, "loss_box_dn_4": 0.87765, "loss_cls_dn_5": 0.17617, "loss_box_dn_5": 0.92748, "loss_dense_depth": 0.80925, "loss": 28.23318, "grad_norm": 63.71604, "time": 1.54754} -{"mode": "train", "epoch": 1, "iter": 144, "lr": 0.00016, "memory": 49164, "data_time": 0.07603, "loss_cls_0": 0.84467, "loss_box_0": 1.79963, "loss_cns_0": 0.61262, "loss_yns_0": 0.14638, "loss_cls_1": 0.914, "loss_box_1": 1.85371, "loss_cns_1": 0.64167, "loss_yns_1": 0.15066, "loss_cls_2": 0.93755, "loss_box_2": 1.83111, "loss_cns_2": 0.64946, "loss_yns_2": 0.15179, "loss_cls_3": 0.95865, "loss_box_3": 1.82706, "loss_cns_3": 0.64866, "loss_yns_3": 0.15584, "loss_cls_4": 0.95598, "loss_box_4": 1.83661, "loss_cns_4": 0.64693, "loss_yns_4": 0.1558, "loss_cls_5": 0.97508, "loss_box_5": 1.87205, "loss_cns_5": 0.64761, "loss_yns_5": 0.15161, "loss_cls_dn_0": 0.23403, "loss_box_dn_0": 0.79841, "loss_cls_dn_1": 0.16111, "loss_box_dn_1": 0.86232, "loss_cls_dn_2": 0.1619, "loss_box_dn_2": 0.86492, "loss_cls_dn_3": 0.16507, "loss_box_dn_3": 0.88928, "loss_cls_dn_4": 0.16572, "loss_box_dn_4": 0.91645, "loss_cls_dn_5": 0.17821, "loss_box_dn_5": 0.95474, "loss_dense_depth": 0.79408, "loss": 28.51136, "grad_norm": 63.92429, "time": 1.56829} -{"mode": "train", "epoch": 1, "iter": 145, "lr": 0.00016, "memory": 49164, "data_time": 0.10963, "loss_cls_0": 0.83501, "loss_box_0": 1.77544, "loss_cns_0": 0.61983, "loss_yns_0": 0.15069, "loss_cls_1": 0.90964, "loss_box_1": 1.80936, "loss_cns_1": 0.64754, "loss_yns_1": 0.15217, "loss_cls_2": 0.94377, "loss_box_2": 1.77543, "loss_cns_2": 0.65158, "loss_yns_2": 0.15199, "loss_cls_3": 0.94382, "loss_box_3": 1.78426, "loss_cns_3": 0.65423, "loss_yns_3": 0.15542, "loss_cls_4": 0.9598, "loss_box_4": 1.76349, "loss_cns_4": 0.65191, "loss_yns_4": 0.15321, "loss_cls_5": 0.94742, "loss_box_5": 1.75591, "loss_cns_5": 0.65438, "loss_yns_5": 0.15149, "loss_cls_dn_0": 0.22819, "loss_box_dn_0": 0.78242, "loss_cls_dn_1": 0.15835, "loss_box_dn_1": 0.89453, "loss_cls_dn_2": 0.16022, "loss_box_dn_2": 0.88813, "loss_cls_dn_3": 0.16986, "loss_box_dn_3": 0.9073, "loss_cls_dn_4": 0.16533, "loss_box_dn_4": 0.91318, "loss_cls_dn_5": 0.17302, "loss_box_dn_5": 0.92831, "loss_dense_depth": 0.81651, "loss": 28.18313, "grad_norm": 42.68056, "time": 1.66551} -{"mode": "train", "epoch": 1, "iter": 146, "lr": 0.00016, "memory": 49164, "data_time": 0.07687, "loss_cls_0": 0.83287, "loss_box_0": 1.76085, "loss_cns_0": 0.61875, "loss_yns_0": 0.1478, "loss_cls_1": 0.91503, "loss_box_1": 1.79668, "loss_cns_1": 0.64811, "loss_yns_1": 0.14864, "loss_cls_2": 0.96006, "loss_box_2": 1.79942, "loss_cns_2": 0.64633, "loss_yns_2": 0.15024, "loss_cls_3": 0.9468, "loss_box_3": 1.81836, "loss_cns_3": 0.64471, "loss_yns_3": 0.15007, "loss_cls_4": 0.96023, "loss_box_4": 1.8151, "loss_cns_4": 0.64612, "loss_yns_4": 0.15047, "loss_cls_5": 0.97755, "loss_box_5": 1.76295, "loss_cns_5": 0.65051, "loss_yns_5": 0.1507, "loss_cls_dn_0": 0.22446, "loss_box_dn_0": 0.78057, "loss_cls_dn_1": 0.15146, "loss_box_dn_1": 0.81836, "loss_cls_dn_2": 0.15653, "loss_box_dn_2": 0.81667, "loss_cls_dn_3": 0.16217, "loss_box_dn_3": 0.82871, "loss_cls_dn_4": 0.16344, "loss_box_dn_4": 0.82749, "loss_cls_dn_5": 0.16822, "loss_box_dn_5": 0.81822, "loss_dense_depth": 0.78063, "loss": 27.79526, "grad_norm": 62.92423, "time": 1.57851} -{"mode": "train", "epoch": 1, "iter": 147, "lr": 0.00016, "memory": 49164, "data_time": 0.07367, "loss_cls_0": 0.83323, "loss_box_0": 1.7569, "loss_cns_0": 0.62083, "loss_yns_0": 0.14767, "loss_cls_1": 0.90824, "loss_box_1": 1.76189, "loss_cns_1": 0.65118, "loss_yns_1": 0.14873, "loss_cls_2": 0.95925, "loss_box_2": 1.73304, "loss_cns_2": 0.65938, "loss_yns_2": 0.15082, "loss_cls_3": 0.94008, "loss_box_3": 1.74274, "loss_cns_3": 0.6588, "loss_yns_3": 0.14937, "loss_cls_4": 0.95459, "loss_box_4": 1.74901, "loss_cns_4": 0.65995, "loss_yns_4": 0.14891, "loss_cls_5": 0.99763, "loss_box_5": 1.70393, "loss_cns_5": 0.66342, "loss_yns_5": 0.15052, "loss_cls_dn_0": 0.22379, "loss_box_dn_0": 0.78099, "loss_cls_dn_1": 0.15102, "loss_box_dn_1": 0.78568, "loss_cls_dn_2": 0.1563, "loss_box_dn_2": 0.77506, "loss_cls_dn_3": 0.15592, "loss_box_dn_3": 0.77988, "loss_cls_dn_4": 0.16067, "loss_box_dn_4": 0.77732, "loss_cls_dn_5": 0.16381, "loss_box_dn_5": 0.76919, "loss_dense_depth": 0.79891, "loss": 27.32863, "grad_norm": 45.58746, "time": 1.6262} -{"mode": "train", "epoch": 1, "iter": 148, "lr": 0.00016, "memory": 49164, "data_time": 0.09016, "loss_cls_0": 0.81896, "loss_box_0": 1.74669, "loss_cns_0": 0.62144, "loss_yns_0": 0.14632, "loss_cls_1": 0.90651, "loss_box_1": 1.73991, "loss_cns_1": 0.65239, "loss_yns_1": 0.14609, "loss_cls_2": 0.93609, "loss_box_2": 1.68476, "loss_cns_2": 0.66655, "loss_yns_2": 0.14832, "loss_cls_3": 0.94089, "loss_box_3": 1.68382, "loss_cns_3": 0.66193, "loss_yns_3": 0.14899, "loss_cls_4": 0.96009, "loss_box_4": 1.70358, "loss_cns_4": 0.66515, "loss_yns_4": 0.1477, "loss_cls_5": 0.95125, "loss_box_5": 1.72175, "loss_cns_5": 0.65763, "loss_yns_5": 0.14864, "loss_cls_dn_0": 0.2201, "loss_box_dn_0": 0.79217, "loss_cls_dn_1": 0.14906, "loss_box_dn_1": 0.73522, "loss_cls_dn_2": 0.15348, "loss_box_dn_2": 0.71443, "loss_cls_dn_3": 0.15596, "loss_box_dn_3": 0.71999, "loss_cls_dn_4": 0.16346, "loss_box_dn_4": 0.72514, "loss_cls_dn_5": 0.16765, "loss_box_dn_5": 0.74289, "loss_dense_depth": 0.76549, "loss": 26.81048, "grad_norm": 48.14239, "time": 1.57834} -{"mode": "train", "epoch": 1, "iter": 149, "lr": 0.00016, "memory": 49164, "data_time": 0.08767, "loss_cls_0": 0.82139, "loss_box_0": 1.75105, "loss_cns_0": 0.62087, "loss_yns_0": 0.14686, "loss_cls_1": 0.91312, "loss_box_1": 1.81128, "loss_cns_1": 0.64565, "loss_yns_1": 0.14668, "loss_cls_2": 0.95139, "loss_box_2": 1.77207, "loss_cns_2": 0.65221, "loss_yns_2": 0.14667, "loss_cls_3": 0.94731, "loss_box_3": 1.77422, "loss_cns_3": 0.64956, "loss_yns_3": 0.15106, "loss_cls_4": 0.96582, "loss_box_4": 1.76751, "loss_cns_4": 0.64895, "loss_yns_4": 0.14999, "loss_cls_5": 0.95936, "loss_box_5": 1.78783, "loss_cns_5": 0.64705, "loss_yns_5": 0.14816, "loss_cls_dn_0": 0.21737, "loss_box_dn_0": 0.78113, "loss_cls_dn_1": 0.15008, "loss_box_dn_1": 0.77266, "loss_cls_dn_2": 0.15593, "loss_box_dn_2": 0.76134, "loss_cls_dn_3": 0.15851, "loss_box_dn_3": 0.77779, "loss_cls_dn_4": 0.1664, "loss_box_dn_4": 0.79427, "loss_cls_dn_5": 0.17143, "loss_box_dn_5": 0.82157, "loss_dense_depth": 0.75121, "loss": 27.45576, "grad_norm": 58.6477, "time": 1.58302} -{"mode": "train", "epoch": 1, "iter": 150, "lr": 0.00016, "memory": 49164, "data_time": 0.07818, "loss_cls_0": 0.85032, "loss_box_0": 1.75388, "loss_cns_0": 0.61808, "loss_yns_0": 0.14583, "loss_cls_1": 0.91618, "loss_box_1": 1.77319, "loss_cns_1": 0.6467, "loss_yns_1": 0.14611, "loss_cls_2": 0.96307, "loss_box_2": 1.74789, "loss_cns_2": 0.65253, "loss_yns_2": 0.1453, "loss_cls_3": 0.96749, "loss_box_3": 1.74901, "loss_cns_3": 0.65917, "loss_yns_3": 0.15183, "loss_cls_4": 0.961, "loss_box_4": 1.73441, "loss_cns_4": 0.65396, "loss_yns_4": 0.15713, "loss_cls_5": 0.96854, "loss_box_5": 1.73628, "loss_cns_5": 0.65253, "loss_yns_5": 0.14857, "loss_cls_dn_0": 0.21862, "loss_box_dn_0": 0.77591, "loss_cls_dn_1": 0.15215, "loss_box_dn_1": 0.80746, "loss_cls_dn_2": 0.15808, "loss_box_dn_2": 0.82004, "loss_cls_dn_3": 0.16262, "loss_box_dn_3": 0.84218, "loss_cls_dn_4": 0.16788, "loss_box_dn_4": 0.86827, "loss_cls_dn_5": 0.17478, "loss_box_dn_5": 0.89497, "loss_dense_depth": 0.77515, "loss": 27.71712, "grad_norm": 52.17097, "time": 1.59768} -{"mode": "train", "epoch": 1, "iter": 151, "lr": 0.00016, "memory": 49164, "data_time": 0.06875, "loss_cls_0": 0.84399, "loss_box_0": 1.75861, "loss_cns_0": 0.6185, "loss_yns_0": 0.14746, "loss_cls_1": 0.90502, "loss_box_1": 1.80796, "loss_cns_1": 0.6423, "loss_yns_1": 0.14802, "loss_cls_2": 0.94296, "loss_box_2": 1.73799, "loss_cns_2": 0.64977, "loss_yns_2": 0.1484, "loss_cls_3": 0.95574, "loss_box_3": 1.7478, "loss_cns_3": 0.65636, "loss_yns_3": 0.15111, "loss_cls_4": 0.94299, "loss_box_4": 1.74294, "loss_cns_4": 0.65463, "loss_yns_4": 0.15237, "loss_cls_5": 0.95335, "loss_box_5": 1.76264, "loss_cns_5": 0.65227, "loss_yns_5": 0.15003, "loss_cls_dn_0": 0.21511, "loss_box_dn_0": 0.78484, "loss_cls_dn_1": 0.14806, "loss_box_dn_1": 0.83984, "loss_cls_dn_2": 0.15156, "loss_box_dn_2": 0.83339, "loss_cls_dn_3": 0.1546, "loss_box_dn_3": 0.84893, "loss_cls_dn_4": 0.15722, "loss_box_dn_4": 0.87159, "loss_cls_dn_5": 0.16494, "loss_box_dn_5": 0.89801, "loss_dense_depth": 0.7823, "loss": 27.7236, "grad_norm": 54.85386, "time": 1.56486} -{"mode": "train", "epoch": 1, "iter": 152, "lr": 0.00016, "memory": 49164, "data_time": 0.07524, "loss_cls_0": 0.8275, "loss_box_0": 1.7221, "loss_cns_0": 0.62094, "loss_yns_0": 0.15007, "loss_cls_1": 0.91114, "loss_box_1": 1.78908, "loss_cns_1": 0.64476, "loss_yns_1": 0.14995, "loss_cls_2": 0.93311, "loss_box_2": 1.73718, "loss_cns_2": 0.65412, "loss_yns_2": 0.15076, "loss_cls_3": 0.93836, "loss_box_3": 1.72339, "loss_cns_3": 0.65443, "loss_yns_3": 0.14987, "loss_cls_4": 0.94794, "loss_box_4": 1.70953, "loss_cns_4": 0.65655, "loss_yns_4": 0.15025, "loss_cls_5": 0.94701, "loss_box_5": 1.71474, "loss_cns_5": 0.65408, "loss_yns_5": 0.15018, "loss_cls_dn_0": 0.21613, "loss_box_dn_0": 0.7743, "loss_cls_dn_1": 0.1497, "loss_box_dn_1": 0.85067, "loss_cls_dn_2": 0.15245, "loss_box_dn_2": 0.83673, "loss_cls_dn_3": 0.15353, "loss_box_dn_3": 0.83761, "loss_cls_dn_4": 0.16051, "loss_box_dn_4": 0.84848, "loss_cls_dn_5": 0.16554, "loss_box_dn_5": 0.86326, "loss_dense_depth": 0.75419, "loss": 27.45009, "grad_norm": 47.25884, "time": 1.57552} -{"mode": "train", "epoch": 1, "iter": 153, "lr": 0.00016, "memory": 49164, "data_time": 0.10951, "loss_cls_0": 0.85389, "loss_box_0": 1.73574, "loss_cns_0": 0.61848, "loss_yns_0": 0.14872, "loss_cls_1": 0.91713, "loss_box_1": 1.73641, "loss_cns_1": 0.64879, "loss_yns_1": 0.15025, "loss_cls_2": 0.94788, "loss_box_2": 1.75121, "loss_cns_2": 0.6516, "loss_yns_2": 0.14982, "loss_cls_3": 0.98203, "loss_box_3": 1.71955, "loss_cns_3": 0.65186, "loss_yns_3": 0.1543, "loss_cls_4": 0.96183, "loss_box_4": 1.70216, "loss_cns_4": 0.656, "loss_yns_4": 0.15523, "loss_cls_5": 0.96234, "loss_box_5": 1.68372, "loss_cns_5": 0.65342, "loss_yns_5": 0.14848, "loss_cls_dn_0": 0.22821, "loss_box_dn_0": 0.78314, "loss_cls_dn_1": 0.1488, "loss_box_dn_1": 0.77926, "loss_cls_dn_2": 0.15057, "loss_box_dn_2": 0.78164, "loss_cls_dn_3": 0.15571, "loss_box_dn_3": 0.77838, "loss_cls_dn_4": 0.16229, "loss_box_dn_4": 0.77781, "loss_cls_dn_5": 0.16612, "loss_box_dn_5": 0.7781, "loss_dense_depth": 0.78428, "loss": 27.21513, "grad_norm": 46.26753, "time": 1.59723} -{"mode": "train", "epoch": 1, "iter": 154, "lr": 0.00016, "memory": 49164, "data_time": 0.07981, "loss_cls_0": 0.82948, "loss_box_0": 1.7019, "loss_cns_0": 0.61924, "loss_yns_0": 0.14687, "loss_cls_1": 0.90055, "loss_box_1": 1.69974, "loss_cns_1": 0.64516, "loss_yns_1": 0.14812, "loss_cls_2": 0.94669, "loss_box_2": 1.66731, "loss_cns_2": 0.65207, "loss_yns_2": 0.14771, "loss_cls_3": 0.97238, "loss_box_3": 1.65271, "loss_cns_3": 0.65216, "loss_yns_3": 0.15296, "loss_cls_4": 0.95657, "loss_box_4": 1.65902, "loss_cns_4": 0.65416, "loss_yns_4": 0.15126, "loss_cls_5": 0.96366, "loss_box_5": 1.66231, "loss_cns_5": 0.65237, "loss_yns_5": 0.14903, "loss_cls_dn_0": 0.21973, "loss_box_dn_0": 0.76965, "loss_cls_dn_1": 0.14718, "loss_box_dn_1": 0.75573, "loss_cls_dn_2": 0.15107, "loss_box_dn_2": 0.74339, "loss_cls_dn_3": 0.15612, "loss_box_dn_3": 0.74843, "loss_cls_dn_4": 0.15757, "loss_box_dn_4": 0.74929, "loss_cls_dn_5": 0.16033, "loss_box_dn_5": 0.75587, "loss_dense_depth": 0.73006, "loss": 26.62787, "grad_norm": 40.523, "time": 1.65726} -{"mode": "train", "epoch": 1, "iter": 155, "lr": 0.00016, "memory": 49164, "data_time": 0.07676, "loss_cls_0": 0.80443, "loss_box_0": 1.6887, "loss_cns_0": 0.62345, "loss_yns_0": 0.14872, "loss_cls_1": 0.88735, "loss_box_1": 1.7003, "loss_cns_1": 0.64621, "loss_yns_1": 0.14777, "loss_cls_2": 0.92686, "loss_box_2": 1.66029, "loss_cns_2": 0.65484, "loss_yns_2": 0.14967, "loss_cls_3": 0.94224, "loss_box_3": 1.64935, "loss_cns_3": 0.65276, "loss_yns_3": 0.15, "loss_cls_4": 0.93295, "loss_box_4": 1.6821, "loss_cns_4": 0.65136, "loss_yns_4": 0.14831, "loss_cls_5": 0.94366, "loss_box_5": 1.69888, "loss_cns_5": 0.64962, "loss_yns_5": 0.15304, "loss_cls_dn_0": 0.21347, "loss_box_dn_0": 0.77505, "loss_cls_dn_1": 0.1456, "loss_box_dn_1": 0.74768, "loss_cls_dn_2": 0.14691, "loss_box_dn_2": 0.73507, "loss_cls_dn_3": 0.15436, "loss_box_dn_3": 0.74254, "loss_cls_dn_4": 0.15121, "loss_box_dn_4": 0.76052, "loss_cls_dn_5": 0.16355, "loss_box_dn_5": 0.77204, "loss_dense_depth": 0.79456, "loss": 26.59541, "grad_norm": 52.73148, "time": 1.56947} -{"mode": "train", "epoch": 1, "iter": 156, "lr": 0.00016, "memory": 49164, "data_time": 0.11034, "loss_cls_0": 0.82533, "loss_box_0": 1.71323, "loss_cns_0": 0.62192, "loss_yns_0": 0.15013, "loss_cls_1": 0.88557, "loss_box_1": 1.69221, "loss_cns_1": 0.64639, "loss_yns_1": 0.14921, "loss_cls_2": 0.91987, "loss_box_2": 1.65933, "loss_cns_2": 0.65966, "loss_yns_2": 0.15284, "loss_cls_3": 0.93094, "loss_box_3": 1.63945, "loss_cns_3": 0.65659, "loss_yns_3": 0.15024, "loss_cls_4": 0.9383, "loss_box_4": 1.64135, "loss_cns_4": 0.6554, "loss_yns_4": 0.15019, "loss_cls_5": 0.94153, "loss_box_5": 1.65692, "loss_cns_5": 0.65874, "loss_yns_5": 0.15226, "loss_cls_dn_0": 0.21811, "loss_box_dn_0": 0.77961, "loss_cls_dn_1": 0.1472, "loss_box_dn_1": 0.77014, "loss_cls_dn_2": 0.14835, "loss_box_dn_2": 0.7637, "loss_cls_dn_3": 0.15316, "loss_box_dn_3": 0.76925, "loss_cls_dn_4": 0.15239, "loss_box_dn_4": 0.78496, "loss_cls_dn_5": 0.16454, "loss_box_dn_5": 0.79998, "loss_dense_depth": 0.75781, "loss": 26.65678, "grad_norm": 44.46866, "time": 1.60626} -{"mode": "train", "epoch": 1, "iter": 157, "lr": 0.00016, "memory": 49164, "data_time": 0.07606, "loss_cls_0": 0.82135, "loss_box_0": 1.72232, "loss_cns_0": 0.62143, "loss_yns_0": 0.15217, "loss_cls_1": 0.88388, "loss_box_1": 1.72739, "loss_cns_1": 0.64376, "loss_yns_1": 0.14913, "loss_cls_2": 0.92889, "loss_box_2": 1.69202, "loss_cns_2": 0.66305, "loss_yns_2": 0.15247, "loss_cls_3": 0.93981, "loss_box_3": 1.69299, "loss_cns_3": 0.65521, "loss_yns_3": 0.15222, "loss_cls_4": 0.95311, "loss_box_4": 1.68951, "loss_cns_4": 0.65257, "loss_yns_4": 0.1515, "loss_cls_5": 0.93807, "loss_box_5": 1.72943, "loss_cns_5": 0.6534, "loss_yns_5": 0.15344, "loss_cls_dn_0": 0.21338, "loss_box_dn_0": 0.77624, "loss_cls_dn_1": 0.14447, "loss_box_dn_1": 0.75883, "loss_cls_dn_2": 0.14756, "loss_box_dn_2": 0.76392, "loss_cls_dn_3": 0.14929, "loss_box_dn_3": 0.78235, "loss_cls_dn_4": 0.15269, "loss_box_dn_4": 0.79869, "loss_cls_dn_5": 0.15736, "loss_box_dn_5": 0.82207, "loss_dense_depth": 0.75953, "loss": 26.94548, "grad_norm": 53.0739, "time": 1.57346} -{"mode": "train", "epoch": 1, "iter": 158, "lr": 0.00016, "memory": 49164, "data_time": 0.07787, "loss_cls_0": 0.79261, "loss_box_0": 1.71169, "loss_cns_0": 0.6191, "loss_yns_0": 0.15228, "loss_cls_1": 0.87092, "loss_box_1": 1.71395, "loss_cns_1": 0.64814, "loss_yns_1": 0.15122, "loss_cls_2": 0.92125, "loss_box_2": 1.66342, "loss_cns_2": 0.66306, "loss_yns_2": 0.15189, "loss_cls_3": 0.94541, "loss_box_3": 1.66658, "loss_cns_3": 0.65923, "loss_yns_3": 0.15282, "loss_cls_4": 0.9215, "loss_box_4": 1.67134, "loss_cns_4": 0.6581, "loss_yns_4": 0.15201, "loss_cls_5": 0.92165, "loss_box_5": 1.68834, "loss_cns_5": 0.65844, "loss_yns_5": 0.15875, "loss_cls_dn_0": 0.20745, "loss_box_dn_0": 0.77386, "loss_cls_dn_1": 0.13904, "loss_box_dn_1": 0.78106, "loss_cls_dn_2": 0.14663, "loss_box_dn_2": 0.78009, "loss_cls_dn_3": 0.14818, "loss_box_dn_3": 0.80124, "loss_cls_dn_4": 0.15249, "loss_box_dn_4": 0.81278, "loss_cls_dn_5": 0.16045, "loss_box_dn_5": 0.83012, "loss_dense_depth": 0.74175, "loss": 26.78885, "grad_norm": 49.84777, "time": 1.62818} -{"mode": "train", "epoch": 1, "iter": 159, "lr": 0.00016, "memory": 49164, "data_time": 0.0853, "loss_cls_0": 0.79253, "loss_box_0": 1.73024, "loss_cns_0": 0.61736, "loss_yns_0": 0.1529, "loss_cls_1": 0.87963, "loss_box_1": 1.71528, "loss_cns_1": 0.65346, "loss_yns_1": 0.15259, "loss_cls_2": 0.91338, "loss_box_2": 1.69467, "loss_cns_2": 0.65677, "loss_yns_2": 0.15219, "loss_cls_3": 0.94077, "loss_box_3": 1.69259, "loss_cns_3": 0.65748, "loss_yns_3": 0.15255, "loss_cls_4": 0.92812, "loss_box_4": 1.68069, "loss_cns_4": 0.65734, "loss_yns_4": 0.15459, "loss_cls_5": 0.94164, "loss_box_5": 1.65986, "loss_cns_5": 0.6595, "loss_yns_5": 0.1547, "loss_cls_dn_0": 0.20864, "loss_box_dn_0": 0.77803, "loss_cls_dn_1": 0.1407, "loss_box_dn_1": 0.78834, "loss_cls_dn_2": 0.14529, "loss_box_dn_2": 0.79415, "loss_cls_dn_3": 0.14871, "loss_box_dn_3": 0.81179, "loss_cls_dn_4": 0.15728, "loss_box_dn_4": 0.81795, "loss_cls_dn_5": 0.16482, "loss_box_dn_5": 0.82237, "loss_dense_depth": 0.79196, "loss": 26.96085, "grad_norm": 59.10673, "time": 1.5835} -{"mode": "train", "epoch": 1, "iter": 160, "lr": 0.00016, "memory": 49164, "data_time": 0.08983, "loss_cls_0": 0.79289, "loss_box_0": 1.71376, "loss_cns_0": 0.61539, "loss_yns_0": 0.15354, "loss_cls_1": 0.88043, "loss_box_1": 1.69907, "loss_cns_1": 0.65262, "loss_yns_1": 0.15253, "loss_cls_2": 0.91245, "loss_box_2": 1.68821, "loss_cns_2": 0.65384, "loss_yns_2": 0.15249, "loss_cls_3": 0.92451, "loss_box_3": 1.68101, "loss_cns_3": 0.65577, "loss_yns_3": 0.15364, "loss_cls_4": 0.91688, "loss_box_4": 1.66829, "loss_cns_4": 0.65528, "loss_yns_4": 0.15362, "loss_cls_5": 0.93437, "loss_box_5": 1.65081, "loss_cns_5": 0.65711, "loss_yns_5": 0.15385, "loss_cls_dn_0": 0.20962, "loss_box_dn_0": 0.7726, "loss_cls_dn_1": 0.14083, "loss_box_dn_1": 0.77122, "loss_cls_dn_2": 0.14131, "loss_box_dn_2": 0.78332, "loss_cls_dn_3": 0.14484, "loss_box_dn_3": 0.79075, "loss_cls_dn_4": 0.15144, "loss_box_dn_4": 0.79824, "loss_cls_dn_5": 0.15745, "loss_box_dn_5": 0.8032, "loss_dense_depth": 0.75244, "loss": 26.68962, "grad_norm": 45.64686, "time": 1.59074} -{"mode": "train", "epoch": 1, "iter": 161, "lr": 0.00016, "memory": 49164, "data_time": 0.11392, "loss_cls_0": 0.7841, "loss_box_0": 1.72253, "loss_cns_0": 0.61829, "loss_yns_0": 0.15341, "loss_cls_1": 0.87738, "loss_box_1": 1.73177, "loss_cns_1": 0.65066, "loss_yns_1": 0.15541, "loss_cls_2": 0.90441, "loss_box_2": 1.65998, "loss_cns_2": 0.65808, "loss_yns_2": 0.15482, "loss_cls_3": 0.93251, "loss_box_3": 1.622, "loss_cns_3": 0.65803, "loss_yns_3": 0.1568, "loss_cls_4": 0.92248, "loss_box_4": 1.63014, "loss_cns_4": 0.65855, "loss_yns_4": 0.15546, "loss_cls_5": 0.93291, "loss_box_5": 1.62828, "loss_cns_5": 0.66123, "loss_yns_5": 0.15528, "loss_cls_dn_0": 0.20564, "loss_box_dn_0": 0.77745, "loss_cls_dn_1": 0.13997, "loss_box_dn_1": 0.73766, "loss_cls_dn_2": 0.14068, "loss_box_dn_2": 0.72762, "loss_cls_dn_3": 0.14327, "loss_box_dn_3": 0.72413, "loss_cls_dn_4": 0.14761, "loss_box_dn_4": 0.7374, "loss_cls_dn_5": 0.1521, "loss_box_dn_5": 0.74995, "loss_dense_depth": 0.79715, "loss": 26.36512, "grad_norm": 39.22342, "time": 1.68706} -{"mode": "train", "epoch": 1, "iter": 162, "lr": 0.00016, "memory": 49164, "data_time": 0.16769, "loss_cls_0": 0.81239, "loss_box_0": 1.75051, "loss_cns_0": 0.61623, "loss_yns_0": 0.15492, "loss_cls_1": 0.8781, "loss_box_1": 1.75283, "loss_cns_1": 0.64549, "loss_yns_1": 0.15603, "loss_cls_2": 0.91704, "loss_box_2": 1.67802, "loss_cns_2": 0.65347, "loss_yns_2": 0.15738, "loss_cls_3": 0.93519, "loss_box_3": 1.66019, "loss_cns_3": 0.65639, "loss_yns_3": 0.15791, "loss_cls_4": 0.9243, "loss_box_4": 1.66311, "loss_cns_4": 0.65602, "loss_yns_4": 0.15736, "loss_cls_5": 0.92987, "loss_box_5": 1.65685, "loss_cns_5": 0.65904, "loss_yns_5": 0.15836, "loss_cls_dn_0": 0.20924, "loss_box_dn_0": 0.78916, "loss_cls_dn_1": 0.13541, "loss_box_dn_1": 0.7504, "loss_cls_dn_2": 0.14079, "loss_box_dn_2": 0.72875, "loss_cls_dn_3": 0.14485, "loss_box_dn_3": 0.73098, "loss_cls_dn_4": 0.14605, "loss_box_dn_4": 0.74068, "loss_cls_dn_5": 0.15159, "loss_box_dn_5": 0.75078, "loss_dense_depth": 0.84614, "loss": 26.65184, "grad_norm": 49.99268, "time": 1.67617} -{"mode": "train", "epoch": 1, "iter": 163, "lr": 0.00016, "memory": 49164, "data_time": 0.07427, "loss_cls_0": 0.79814, "loss_box_0": 1.74171, "loss_cns_0": 0.62086, "loss_yns_0": 0.15635, "loss_cls_1": 0.89037, "loss_box_1": 1.69904, "loss_cns_1": 0.65041, "loss_yns_1": 0.15597, "loss_cls_2": 0.91708, "loss_box_2": 1.6601, "loss_cns_2": 0.65494, "loss_yns_2": 0.15707, "loss_cls_3": 0.91331, "loss_box_3": 1.63602, "loss_cns_3": 0.658, "loss_yns_3": 0.15729, "loss_cls_4": 0.91729, "loss_box_4": 1.63323, "loss_cns_4": 0.65753, "loss_yns_4": 0.15771, "loss_cls_5": 0.92459, "loss_box_5": 1.63662, "loss_cns_5": 0.65618, "loss_yns_5": 0.15702, "loss_cls_dn_0": 0.2076, "loss_box_dn_0": 0.76765, "loss_cls_dn_1": 0.14109, "loss_box_dn_1": 0.73513, "loss_cls_dn_2": 0.1431, "loss_box_dn_2": 0.7212, "loss_cls_dn_3": 0.14512, "loss_box_dn_3": 0.71985, "loss_cls_dn_4": 0.15304, "loss_box_dn_4": 0.72866, "loss_cls_dn_5": 0.15414, "loss_box_dn_5": 0.73903, "loss_dense_depth": 0.73168, "loss": 26.29409, "grad_norm": 31.2308, "time": 1.56194} -{"mode": "train", "epoch": 1, "iter": 164, "lr": 0.00017, "memory": 49164, "data_time": 0.07258, "loss_cls_0": 0.80993, "loss_box_0": 1.74193, "loss_cns_0": 0.62083, "loss_yns_0": 0.15703, "loss_cls_1": 0.89006, "loss_box_1": 1.67778, "loss_cns_1": 0.65684, "loss_yns_1": 0.15636, "loss_cls_2": 0.91033, "loss_box_2": 1.67319, "loss_cns_2": 0.65836, "loss_yns_2": 0.1555, "loss_cls_3": 0.91618, "loss_box_3": 1.63292, "loss_cns_3": 0.66071, "loss_yns_3": 0.15822, "loss_cls_4": 0.9143, "loss_box_4": 1.63684, "loss_cns_4": 0.65967, "loss_yns_4": 0.15577, "loss_cls_5": 0.92636, "loss_box_5": 1.63596, "loss_cns_5": 0.66003, "loss_yns_5": 0.15772, "loss_cls_dn_0": 0.20808, "loss_box_dn_0": 0.77064, "loss_cls_dn_1": 0.13591, "loss_box_dn_1": 0.73458, "loss_cls_dn_2": 0.13873, "loss_box_dn_2": 0.72949, "loss_cls_dn_3": 0.14143, "loss_box_dn_3": 0.72048, "loss_cls_dn_4": 0.14922, "loss_box_dn_4": 0.73127, "loss_cls_dn_5": 0.15179, "loss_box_dn_5": 0.73962, "loss_dense_depth": 0.84828, "loss": 26.42235, "grad_norm": 46.86898, "time": 1.57433} -{"mode": "train", "epoch": 1, "iter": 165, "lr": 0.00017, "memory": 49164, "data_time": 0.09917, "loss_cls_0": 0.82832, "loss_box_0": 1.75563, "loss_cns_0": 0.61323, "loss_yns_0": 0.15587, "loss_cls_1": 0.88836, "loss_box_1": 1.68213, "loss_cns_1": 0.6544, "loss_yns_1": 0.15817, "loss_cls_2": 0.91306, "loss_box_2": 1.66529, "loss_cns_2": 0.65898, "loss_yns_2": 0.159, "loss_cls_3": 0.93471, "loss_box_3": 1.63202, "loss_cns_3": 0.6613, "loss_yns_3": 0.15978, "loss_cls_4": 0.91995, "loss_box_4": 1.64502, "loss_cns_4": 0.66068, "loss_yns_4": 0.15775, "loss_cls_5": 0.93029, "loss_box_5": 1.6432, "loss_cns_5": 0.66475, "loss_yns_5": 0.15991, "loss_cls_dn_0": 0.21078, "loss_box_dn_0": 0.77053, "loss_cls_dn_1": 0.13517, "loss_box_dn_1": 0.73369, "loss_cls_dn_2": 0.13769, "loss_box_dn_2": 0.72678, "loss_cls_dn_3": 0.14074, "loss_box_dn_3": 0.72724, "loss_cls_dn_4": 0.14161, "loss_box_dn_4": 0.74445, "loss_cls_dn_5": 0.15283, "loss_box_dn_5": 0.75323, "loss_dense_depth": 0.75144, "loss": 26.42798, "grad_norm": 45.00686, "time": 1.65046} -{"mode": "train", "epoch": 1, "iter": 166, "lr": 0.00017, "memory": 49164, "data_time": 0.07396, "loss_cls_0": 0.79694, "loss_box_0": 1.76045, "loss_cns_0": 0.61173, "loss_yns_0": 0.15587, "loss_cls_1": 0.89688, "loss_box_1": 1.66356, "loss_cns_1": 0.6565, "loss_yns_1": 0.1581, "loss_cls_2": 0.91023, "loss_box_2": 1.64702, "loss_cns_2": 0.6596, "loss_yns_2": 0.15943, "loss_cls_3": 0.92396, "loss_box_3": 1.63567, "loss_cns_3": 0.66152, "loss_yns_3": 0.1577, "loss_cls_4": 0.9228, "loss_box_4": 1.63242, "loss_cns_4": 0.66185, "loss_yns_4": 0.15949, "loss_cls_5": 0.92741, "loss_box_5": 1.62393, "loss_cns_5": 0.66459, "loss_yns_5": 0.16088, "loss_cls_dn_0": 0.20318, "loss_box_dn_0": 0.76306, "loss_cls_dn_1": 0.13544, "loss_box_dn_1": 0.74236, "loss_cls_dn_2": 0.13694, "loss_box_dn_2": 0.73879, "loss_cls_dn_3": 0.13953, "loss_box_dn_3": 0.74952, "loss_cls_dn_4": 0.14163, "loss_box_dn_4": 0.76369, "loss_cls_dn_5": 0.14942, "loss_box_dn_5": 0.76587, "loss_dense_depth": 0.81293, "loss": 26.45088, "grad_norm": 46.7545, "time": 1.58942} -{"mode": "train", "epoch": 1, "iter": 167, "lr": 0.00017, "memory": 49164, "data_time": 0.10182, "loss_cls_0": 0.84702, "loss_box_0": 1.778, "loss_cns_0": 0.61474, "loss_yns_0": 0.1618, "loss_cls_1": 0.91538, "loss_box_1": 1.68276, "loss_cns_1": 0.65084, "loss_yns_1": 0.15938, "loss_cls_2": 0.93671, "loss_box_2": 1.65165, "loss_cns_2": 0.65322, "loss_yns_2": 0.15805, "loss_cls_3": 0.95038, "loss_box_3": 1.64301, "loss_cns_3": 0.65755, "loss_yns_3": 0.16201, "loss_cls_4": 0.94733, "loss_box_4": 1.63666, "loss_cns_4": 0.65857, "loss_yns_4": 0.15915, "loss_cls_5": 0.95103, "loss_box_5": 1.63374, "loss_cns_5": 0.65845, "loss_yns_5": 0.16034, "loss_cls_dn_0": 0.21248, "loss_box_dn_0": 0.77193, "loss_cls_dn_1": 0.14039, "loss_box_dn_1": 0.76756, "loss_cls_dn_2": 0.14217, "loss_box_dn_2": 0.76822, "loss_cls_dn_3": 0.1442, "loss_box_dn_3": 0.77431, "loss_cls_dn_4": 0.14608, "loss_box_dn_4": 0.78009, "loss_cls_dn_5": 0.14955, "loss_box_dn_5": 0.78523, "loss_dense_depth": 0.80302, "loss": 26.813, "grad_norm": 38.86308, "time": 6.9954} -{"mode": "train", "epoch": 1, "iter": 168, "lr": 0.00017, "memory": 49164, "data_time": 0.06957, "loss_cls_0": 0.82819, "loss_box_0": 1.74063, "loss_cns_0": 0.62162, "loss_yns_0": 0.16207, "loss_cls_1": 0.89839, "loss_box_1": 1.67678, "loss_cns_1": 0.65323, "loss_yns_1": 0.16039, "loss_cls_2": 0.92535, "loss_box_2": 1.63403, "loss_cns_2": 0.65918, "loss_yns_2": 0.16182, "loss_cls_3": 0.93713, "loss_box_3": 1.63508, "loss_cns_3": 0.65968, "loss_yns_3": 0.16625, "loss_cls_4": 0.94666, "loss_box_4": 1.62921, "loss_cns_4": 0.66135, "loss_yns_4": 0.16732, "loss_cls_5": 0.94883, "loss_box_5": 1.6255, "loss_cns_5": 0.65961, "loss_yns_5": 0.16093, "loss_cls_dn_0": 0.20737, "loss_box_dn_0": 0.76163, "loss_cls_dn_1": 0.13977, "loss_box_dn_1": 0.75933, "loss_cls_dn_2": 0.14336, "loss_box_dn_2": 0.75998, "loss_cls_dn_3": 0.14402, "loss_box_dn_3": 0.77191, "loss_cls_dn_4": 0.14994, "loss_box_dn_4": 0.77833, "loss_cls_dn_5": 0.15636, "loss_box_dn_5": 0.79154, "loss_dense_depth": 0.82188, "loss": 26.70464, "grad_norm": 47.24394, "time": 1.52637} -{"mode": "train", "epoch": 1, "iter": 169, "lr": 0.00017, "memory": 49164, "data_time": 0.07028, "loss_cls_0": 0.86887, "loss_box_0": 1.75357, "loss_cns_0": 0.61443, "loss_yns_0": 0.15915, "loss_cls_1": 0.92109, "loss_box_1": 1.70469, "loss_cns_1": 0.64859, "loss_yns_1": 0.16297, "loss_cls_2": 0.94262, "loss_box_2": 1.66262, "loss_cns_2": 0.65576, "loss_yns_2": 0.16298, "loss_cls_3": 0.95558, "loss_box_3": 1.67763, "loss_cns_3": 0.65735, "loss_yns_3": 0.16454, "loss_cls_4": 0.95372, "loss_box_4": 1.6658, "loss_cns_4": 0.65773, "loss_yns_4": 0.16521, "loss_cls_5": 0.95921, "loss_box_5": 1.65582, "loss_cns_5": 0.65767, "loss_yns_5": 0.16037, "loss_cls_dn_0": 0.21752, "loss_box_dn_0": 0.76703, "loss_cls_dn_1": 0.13851, "loss_box_dn_1": 0.75658, "loss_cls_dn_2": 0.14109, "loss_box_dn_2": 0.75457, "loss_cls_dn_3": 0.14135, "loss_box_dn_3": 0.76784, "loss_cls_dn_4": 0.14658, "loss_box_dn_4": 0.77176, "loss_cls_dn_5": 0.1548, "loss_box_dn_5": 0.78372, "loss_dense_depth": 0.81904, "loss": 26.94838, "grad_norm": 42.16739, "time": 1.54169} -{"mode": "train", "epoch": 1, "iter": 170, "lr": 0.00017, "memory": 49164, "data_time": 0.07405, "loss_cls_0": 0.84729, "loss_box_0": 1.74165, "loss_cns_0": 0.62038, "loss_yns_0": 0.16041, "loss_cls_1": 0.92301, "loss_box_1": 1.65988, "loss_cns_1": 0.65481, "loss_yns_1": 0.16048, "loss_cls_2": 0.94033, "loss_box_2": 1.61831, "loss_cns_2": 0.65988, "loss_yns_2": 0.15901, "loss_cls_3": 0.94919, "loss_box_3": 1.58935, "loss_cns_3": 0.65999, "loss_yns_3": 0.15937, "loss_cls_4": 0.95761, "loss_box_4": 1.58743, "loss_cns_4": 0.6604, "loss_yns_4": 0.158, "loss_cls_5": 0.95436, "loss_box_5": 1.58491, "loss_cns_5": 0.66279, "loss_yns_5": 0.15849, "loss_cls_dn_0": 0.21398, "loss_box_dn_0": 0.76308, "loss_cls_dn_1": 0.13819, "loss_box_dn_1": 0.72061, "loss_cls_dn_2": 0.14128, "loss_box_dn_2": 0.71954, "loss_cls_dn_3": 0.1438, "loss_box_dn_3": 0.71271, "loss_cls_dn_4": 0.1482, "loss_box_dn_4": 0.71794, "loss_cls_dn_5": 0.1536, "loss_box_dn_5": 0.72546, "loss_dense_depth": 0.77586, "loss": 26.30159, "grad_norm": 32.59871, "time": 1.54845} -{"mode": "train", "epoch": 1, "iter": 171, "lr": 0.00017, "memory": 49164, "data_time": 0.07594, "loss_cls_0": 0.8373, "loss_box_0": 1.71681, "loss_cns_0": 0.62083, "loss_yns_0": 0.1594, "loss_cls_1": 0.90729, "loss_box_1": 1.67535, "loss_cns_1": 0.65083, "loss_yns_1": 0.15831, "loss_cls_2": 0.95112, "loss_box_2": 1.64311, "loss_cns_2": 0.65733, "loss_yns_2": 0.15782, "loss_cls_3": 0.93866, "loss_box_3": 1.61504, "loss_cns_3": 0.65901, "loss_yns_3": 0.16134, "loss_cls_4": 0.94401, "loss_box_4": 1.61622, "loss_cns_4": 0.65798, "loss_yns_4": 0.16252, "loss_cls_5": 0.94678, "loss_box_5": 1.60882, "loss_cns_5": 0.66016, "loss_yns_5": 0.15618, "loss_cls_dn_0": 0.20919, "loss_box_dn_0": 0.75949, "loss_cls_dn_1": 0.13917, "loss_box_dn_1": 0.70808, "loss_cls_dn_2": 0.14269, "loss_box_dn_2": 0.70877, "loss_cls_dn_3": 0.14314, "loss_box_dn_3": 0.70694, "loss_cls_dn_4": 0.14747, "loss_box_dn_4": 0.71583, "loss_cls_dn_5": 0.15202, "loss_box_dn_5": 0.71439, "loss_dense_depth": 0.78787, "loss": 26.2973, "grad_norm": 43.79013, "time": 1.55818} -{"mode": "train", "epoch": 1, "iter": 172, "lr": 0.00017, "memory": 49164, "data_time": 0.10171, "loss_cls_0": 0.83519, "loss_box_0": 1.74546, "loss_cns_0": 0.61737, "loss_yns_0": 0.15921, "loss_cls_1": 0.9231, "loss_box_1": 1.67452, "loss_cns_1": 0.64754, "loss_yns_1": 0.1561, "loss_cls_2": 0.94547, "loss_box_2": 1.65486, "loss_cns_2": 0.65612, "loss_yns_2": 0.16026, "loss_cls_3": 0.95615, "loss_box_3": 1.65051, "loss_cns_3": 0.65794, "loss_yns_3": 0.16226, "loss_cls_4": 0.96217, "loss_box_4": 1.64894, "loss_cns_4": 0.6571, "loss_yns_4": 0.16389, "loss_cls_5": 0.95681, "loss_box_5": 1.65234, "loss_cns_5": 0.6573, "loss_yns_5": 0.15994, "loss_cls_dn_0": 0.21196, "loss_box_dn_0": 0.75925, "loss_cls_dn_1": 0.13538, "loss_box_dn_1": 0.72659, "loss_cls_dn_2": 0.13685, "loss_box_dn_2": 0.72456, "loss_cls_dn_3": 0.15035, "loss_box_dn_3": 0.73584, "loss_cls_dn_4": 0.14701, "loss_box_dn_4": 0.75253, "loss_cls_dn_5": 0.15525, "loss_box_dn_5": 0.76172, "loss_dense_depth": 0.7481, "loss": 26.60594, "grad_norm": 44.55742, "time": 1.63344} -{"mode": "train", "epoch": 1, "iter": 173, "lr": 0.00017, "memory": 49164, "data_time": 0.07745, "loss_cls_0": 0.82211, "loss_box_0": 1.71637, "loss_cns_0": 0.62319, "loss_yns_0": 0.15674, "loss_cls_1": 0.90879, "loss_box_1": 1.6791, "loss_cns_1": 0.65198, "loss_yns_1": 0.15568, "loss_cls_2": 0.93504, "loss_box_2": 1.63939, "loss_cns_2": 0.65791, "loss_yns_2": 0.15597, "loss_cls_3": 0.94783, "loss_box_3": 1.64976, "loss_cns_3": 0.65725, "loss_yns_3": 0.15702, "loss_cls_4": 0.95929, "loss_box_4": 1.65343, "loss_cns_4": 0.65797, "loss_yns_4": 0.15728, "loss_cls_5": 0.95759, "loss_box_5": 1.66513, "loss_cns_5": 0.65661, "loss_yns_5": 0.15611, "loss_cls_dn_0": 0.21162, "loss_box_dn_0": 0.75839, "loss_cls_dn_1": 0.13668, "loss_box_dn_1": 0.74091, "loss_cls_dn_2": 0.13778, "loss_box_dn_2": 0.73465, "loss_cls_dn_3": 0.14338, "loss_box_dn_3": 0.74958, "loss_cls_dn_4": 0.14313, "loss_box_dn_4": 0.76063, "loss_cls_dn_5": 0.14684, "loss_box_dn_5": 0.7801, "loss_dense_depth": 0.79207, "loss": 26.61328, "grad_norm": 53.33952, "time": 1.5814} -{"mode": "train", "epoch": 1, "iter": 174, "lr": 0.00017, "memory": 49164, "data_time": 0.07586, "loss_cls_0": 0.83206, "loss_box_0": 1.74186, "loss_cns_0": 0.61361, "loss_yns_0": 0.15713, "loss_cls_1": 0.91905, "loss_box_1": 1.66726, "loss_cns_1": 0.64838, "loss_yns_1": 0.15754, "loss_cls_2": 0.94836, "loss_box_2": 1.6681, "loss_cns_2": 0.65427, "loss_yns_2": 0.15563, "loss_cls_3": 0.93266, "loss_box_3": 1.65769, "loss_cns_3": 0.65429, "loss_yns_3": 0.1549, "loss_cls_4": 0.9383, "loss_box_4": 1.6484, "loss_cns_4": 0.6573, "loss_yns_4": 0.15533, "loss_cls_5": 0.96266, "loss_box_5": 1.65319, "loss_cns_5": 0.6598, "loss_yns_5": 0.15678, "loss_cls_dn_0": 0.21114, "loss_box_dn_0": 0.7673, "loss_cls_dn_1": 0.13704, "loss_box_dn_1": 0.75062, "loss_cls_dn_2": 0.14021, "loss_box_dn_2": 0.75239, "loss_cls_dn_3": 0.14618, "loss_box_dn_3": 0.75986, "loss_cls_dn_4": 0.14527, "loss_box_dn_4": 0.77096, "loss_cls_dn_5": 0.16195, "loss_box_dn_5": 0.78904, "loss_dense_depth": 0.77742, "loss": 26.7039, "grad_norm": 44.22031, "time": 1.56522} -{"mode": "train", "epoch": 1, "iter": 175, "lr": 0.00017, "memory": 49164, "data_time": 0.07246, "loss_cls_0": 0.82009, "loss_box_0": 1.72113, "loss_cns_0": 0.6185, "loss_yns_0": 0.15923, "loss_cls_1": 0.90053, "loss_box_1": 1.66609, "loss_cns_1": 0.65224, "loss_yns_1": 0.15996, "loss_cls_2": 0.92743, "loss_box_2": 1.67195, "loss_cns_2": 0.65356, "loss_yns_2": 0.15887, "loss_cls_3": 0.93845, "loss_box_3": 1.6548, "loss_cns_3": 0.65727, "loss_yns_3": 0.16353, "loss_cls_4": 0.9416, "loss_box_4": 1.65664, "loss_cns_4": 0.65694, "loss_yns_4": 0.16111, "loss_cls_5": 0.94235, "loss_box_5": 1.66235, "loss_cns_5": 0.65801, "loss_yns_5": 0.15704, "loss_cls_dn_0": 0.21275, "loss_box_dn_0": 0.75342, "loss_cls_dn_1": 0.13362, "loss_box_dn_1": 0.74604, "loss_cls_dn_2": 0.1347, "loss_box_dn_2": 0.75652, "loss_cls_dn_3": 0.13525, "loss_box_dn_3": 0.76137, "loss_cls_dn_4": 0.13743, "loss_box_dn_4": 0.78041, "loss_cls_dn_5": 0.14494, "loss_box_dn_5": 0.79713, "loss_dense_depth": 0.7862, "loss": 26.63946, "grad_norm": 50.58326, "time": 1.56774} -{"mode": "train", "epoch": 1, "iter": 176, "lr": 0.00017, "memory": 49164, "data_time": 0.07486, "loss_cls_0": 0.81477, "loss_box_0": 1.70395, "loss_cns_0": 0.62064, "loss_yns_0": 0.1545, "loss_cls_1": 0.89691, "loss_box_1": 1.65729, "loss_cns_1": 0.65042, "loss_yns_1": 0.15357, "loss_cls_2": 0.92518, "loss_box_2": 1.63631, "loss_cns_2": 0.65664, "loss_yns_2": 0.15549, "loss_cls_3": 0.92877, "loss_box_3": 1.62513, "loss_cns_3": 0.66178, "loss_yns_3": 0.16303, "loss_cls_4": 0.94379, "loss_box_4": 1.62427, "loss_cns_4": 0.66056, "loss_yns_4": 0.16003, "loss_cls_5": 0.94264, "loss_box_5": 1.63363, "loss_cns_5": 0.65818, "loss_yns_5": 0.15308, "loss_cls_dn_0": 0.21317, "loss_box_dn_0": 0.76541, "loss_cls_dn_1": 0.13731, "loss_box_dn_1": 0.75845, "loss_cls_dn_2": 0.13675, "loss_box_dn_2": 0.75383, "loss_cls_dn_3": 0.1384, "loss_box_dn_3": 0.75746, "loss_cls_dn_4": 0.13986, "loss_box_dn_4": 0.76878, "loss_cls_dn_5": 0.1438, "loss_box_dn_5": 0.78183, "loss_dense_depth": 0.80015, "loss": 26.47575, "grad_norm": 36.53755, "time": 1.61216} -{"mode": "train", "epoch": 1, "iter": 177, "lr": 0.00017, "memory": 49164, "data_time": 0.07114, "loss_cls_0": 0.80499, "loss_box_0": 1.69903, "loss_cns_0": 0.61624, "loss_yns_0": 0.14987, "loss_cls_1": 0.88416, "loss_box_1": 1.6485, "loss_cns_1": 0.64899, "loss_yns_1": 0.14919, "loss_cls_2": 0.92288, "loss_box_2": 1.61645, "loss_cns_2": 0.65588, "loss_yns_2": 0.14771, "loss_cls_3": 0.94441, "loss_box_3": 1.61027, "loss_cns_3": 0.65578, "loss_yns_3": 0.15154, "loss_cls_4": 0.94491, "loss_box_4": 1.61309, "loss_cns_4": 0.65663, "loss_yns_4": 0.14964, "loss_cls_5": 0.93029, "loss_box_5": 1.60132, "loss_cns_5": 0.65809, "loss_yns_5": 0.15081, "loss_cls_dn_0": 0.21232, "loss_box_dn_0": 0.77372, "loss_cls_dn_1": 0.13606, "loss_box_dn_1": 0.76621, "loss_cls_dn_2": 0.13587, "loss_box_dn_2": 0.74902, "loss_cls_dn_3": 0.13972, "loss_box_dn_3": 0.74729, "loss_cls_dn_4": 0.14455, "loss_box_dn_4": 0.7533, "loss_cls_dn_5": 0.14622, "loss_box_dn_5": 0.74974, "loss_dense_depth": 0.78655, "loss": 26.25124, "grad_norm": 44.07105, "time": 1.56879} -{"mode": "train", "epoch": 1, "iter": 178, "lr": 0.00017, "memory": 49164, "data_time": 0.07484, "loss_cls_0": 0.79807, "loss_box_0": 1.70116, "loss_cns_0": 0.61723, "loss_yns_0": 0.15062, "loss_cls_1": 0.88161, "loss_box_1": 1.68121, "loss_cns_1": 0.64648, "loss_yns_1": 0.14932, "loss_cls_2": 0.91861, "loss_box_2": 1.63322, "loss_cns_2": 0.65406, "loss_yns_2": 0.14872, "loss_cls_3": 0.91999, "loss_box_3": 1.60384, "loss_cns_3": 0.65442, "loss_yns_3": 0.14926, "loss_cls_4": 0.92572, "loss_box_4": 1.59359, "loss_cns_4": 0.65329, "loss_yns_4": 0.14998, "loss_cls_5": 0.92863, "loss_box_5": 1.62114, "loss_cns_5": 0.65792, "loss_yns_5": 0.15, "loss_cls_dn_0": 0.20944, "loss_box_dn_0": 0.76791, "loss_cls_dn_1": 0.13737, "loss_box_dn_1": 0.72456, "loss_cls_dn_2": 0.14197, "loss_box_dn_2": 0.71274, "loss_cls_dn_3": 0.14845, "loss_box_dn_3": 0.70557, "loss_cls_dn_4": 0.14444, "loss_box_dn_4": 0.70952, "loss_cls_dn_5": 0.15747, "loss_box_dn_5": 0.71643, "loss_dense_depth": 0.78135, "loss": 26.04531, "grad_norm": 33.81805, "time": 1.56637} -{"mode": "train", "epoch": 1, "iter": 179, "lr": 0.00017, "memory": 49164, "data_time": 0.08115, "loss_cls_0": 0.79827, "loss_box_0": 1.71575, "loss_cns_0": 0.61405, "loss_yns_0": 0.15068, "loss_cls_1": 0.89392, "loss_box_1": 1.65528, "loss_cns_1": 0.64402, "loss_yns_1": 0.1526, "loss_cls_2": 0.91209, "loss_box_2": 1.6097, "loss_cns_2": 0.65348, "loss_yns_2": 0.15261, "loss_cls_3": 0.94374, "loss_box_3": 1.59292, "loss_cns_3": 0.65402, "loss_yns_3": 0.15258, "loss_cls_4": 0.96676, "loss_box_4": 1.58193, "loss_cns_4": 0.64988, "loss_yns_4": 0.15344, "loss_cls_5": 0.94712, "loss_box_5": 1.61137, "loss_cns_5": 0.65403, "loss_yns_5": 0.15212, "loss_cls_dn_0": 0.21099, "loss_box_dn_0": 0.77613, "loss_cls_dn_1": 0.1374, "loss_box_dn_1": 0.72222, "loss_cls_dn_2": 0.13822, "loss_box_dn_2": 0.71368, "loss_cls_dn_3": 0.1385, "loss_box_dn_3": 0.71725, "loss_cls_dn_4": 0.14468, "loss_box_dn_4": 0.72393, "loss_cls_dn_5": 0.14604, "loss_box_dn_5": 0.74338, "loss_dense_depth": 0.80712, "loss": 26.1319, "grad_norm": 40.28945, "time": 1.59837} -{"mode": "train", "epoch": 1, "iter": 180, "lr": 0.00017, "memory": 49164, "data_time": 0.0785, "loss_cls_0": 0.79343, "loss_box_0": 1.68239, "loss_cns_0": 0.62534, "loss_yns_0": 0.14953, "loss_cls_1": 0.87908, "loss_box_1": 1.63181, "loss_cns_1": 0.6499, "loss_yns_1": 0.14815, "loss_cls_2": 0.90753, "loss_box_2": 1.61255, "loss_cns_2": 0.65824, "loss_yns_2": 0.15124, "loss_cls_3": 0.92959, "loss_box_3": 1.58586, "loss_cns_3": 0.66627, "loss_yns_3": 0.15566, "loss_cls_4": 0.92841, "loss_box_4": 1.58284, "loss_cns_4": 0.66157, "loss_yns_4": 0.15434, "loss_cls_5": 0.92518, "loss_box_5": 1.58598, "loss_cns_5": 0.65964, "loss_yns_5": 0.15154, "loss_cls_dn_0": 0.20992, "loss_box_dn_0": 0.76554, "loss_cls_dn_1": 0.13519, "loss_box_dn_1": 0.72862, "loss_cls_dn_2": 0.13532, "loss_box_dn_2": 0.72554, "loss_cls_dn_3": 0.13602, "loss_box_dn_3": 0.7338, "loss_cls_dn_4": 0.13919, "loss_box_dn_4": 0.74494, "loss_cls_dn_5": 0.1447, "loss_box_dn_5": 0.76703, "loss_dense_depth": 0.77564, "loss": 26.01751, "grad_norm": 36.07664, "time": 1.57399} -{"mode": "train", "epoch": 1, "iter": 181, "lr": 0.00017, "memory": 49164, "data_time": 0.11045, "loss_cls_0": 0.79201, "loss_box_0": 1.71385, "loss_cns_0": 0.61884, "loss_yns_0": 0.15022, "loss_cls_1": 0.88345, "loss_box_1": 1.64227, "loss_cns_1": 0.65226, "loss_yns_1": 0.14998, "loss_cls_2": 0.91129, "loss_box_2": 1.61268, "loss_cns_2": 0.65466, "loss_yns_2": 0.14995, "loss_cls_3": 0.96804, "loss_box_3": 1.5942, "loss_cns_3": 0.65641, "loss_yns_3": 0.15189, "loss_cls_4": 1.01521, "loss_box_4": 1.59001, "loss_cns_4": 0.65586, "loss_yns_4": 0.15068, "loss_cls_5": 0.93139, "loss_box_5": 1.60615, "loss_cns_5": 0.65447, "loss_yns_5": 0.15029, "loss_cls_dn_0": 0.2094, "loss_box_dn_0": 0.76681, "loss_cls_dn_1": 0.13396, "loss_box_dn_1": 0.74493, "loss_cls_dn_2": 0.13414, "loss_box_dn_2": 0.74553, "loss_cls_dn_3": 0.13475, "loss_box_dn_3": 0.75679, "loss_cls_dn_4": 0.13794, "loss_box_dn_4": 0.76734, "loss_cls_dn_5": 0.14047, "loss_box_dn_5": 0.79454, "loss_dense_depth": 0.83143, "loss": 26.35409, "grad_norm": 50.29567, "time": 1.68501} -{"mode": "train", "epoch": 1, "iter": 182, "lr": 0.00017, "memory": 49164, "data_time": 0.16458, "loss_cls_0": 0.78184, "loss_box_0": 1.67415, "loss_cns_0": 0.62302, "loss_yns_0": 0.14993, "loss_cls_1": 0.88244, "loss_box_1": 1.64848, "loss_cns_1": 0.65455, "loss_yns_1": 0.15315, "loss_cls_2": 0.90173, "loss_box_2": 1.62206, "loss_cns_2": 0.6554, "loss_yns_2": 0.14792, "loss_cls_3": 0.90576, "loss_box_3": 1.61466, "loss_cns_3": 0.65746, "loss_yns_3": 0.14909, "loss_cls_4": 0.91206, "loss_box_4": 1.61341, "loss_cns_4": 0.6556, "loss_yns_4": 0.14891, "loss_cls_5": 0.91039, "loss_box_5": 1.62566, "loss_cns_5": 0.6556, "loss_yns_5": 0.14755, "loss_cls_dn_0": 0.20681, "loss_box_dn_0": 0.76657, "loss_cls_dn_1": 0.13417, "loss_box_dn_1": 0.77739, "loss_cls_dn_2": 0.13796, "loss_box_dn_2": 0.77803, "loss_cls_dn_3": 0.14709, "loss_box_dn_3": 0.78448, "loss_cls_dn_4": 0.14635, "loss_box_dn_4": 0.79512, "loss_cls_dn_5": 0.1557, "loss_box_dn_5": 0.81621, "loss_dense_depth": 0.76122, "loss": 26.29791, "grad_norm": 36.00284, "time": 1.64083} -{"mode": "train", "epoch": 1, "iter": 183, "lr": 0.00017, "memory": 49164, "data_time": 0.07505, "loss_cls_0": 0.77597, "loss_box_0": 1.65763, "loss_cns_0": 0.62302, "loss_yns_0": 0.14643, "loss_cls_1": 0.8781, "loss_box_1": 1.5801, "loss_cns_1": 0.65677, "loss_yns_1": 0.14693, "loss_cls_2": 0.91018, "loss_box_2": 1.53783, "loss_cns_2": 0.65757, "loss_yns_2": 0.14526, "loss_cls_3": 0.9135, "loss_box_3": 1.51484, "loss_cns_3": 0.66419, "loss_yns_3": 0.14627, "loss_cls_4": 0.918, "loss_box_4": 1.51768, "loss_cns_4": 0.66311, "loss_yns_4": 0.14637, "loss_cls_5": 0.90384, "loss_box_5": 1.51558, "loss_cns_5": 0.66224, "loss_yns_5": 0.14616, "loss_cls_dn_0": 0.2051, "loss_box_dn_0": 0.75456, "loss_cls_dn_1": 0.13088, "loss_box_dn_1": 0.73897, "loss_cls_dn_2": 0.13874, "loss_box_dn_2": 0.73504, "loss_cls_dn_3": 0.14401, "loss_box_dn_3": 0.7297, "loss_cls_dn_4": 0.14346, "loss_box_dn_4": 0.73523, "loss_cls_dn_5": 0.14774, "loss_box_dn_5": 0.73937, "loss_dense_depth": 0.80112, "loss": 25.57148, "grad_norm": 37.68125, "time": 1.62224} -{"mode": "train", "epoch": 1, "iter": 184, "lr": 0.00017, "memory": 49164, "data_time": 0.11888, "loss_cls_0": 0.80445, "loss_box_0": 1.67746, "loss_cns_0": 0.62594, "loss_yns_0": 0.14392, "loss_cls_1": 0.88393, "loss_box_1": 1.6155, "loss_cns_1": 0.65812, "loss_yns_1": 0.14344, "loss_cls_2": 0.90057, "loss_box_2": 1.58929, "loss_cns_2": 0.6608, "loss_yns_2": 0.14485, "loss_cls_3": 0.91163, "loss_box_3": 1.56637, "loss_cns_3": 0.66542, "loss_yns_3": 0.14516, "loss_cls_4": 0.92069, "loss_box_4": 1.55594, "loss_cns_4": 0.66353, "loss_yns_4": 0.14531, "loss_cls_5": 0.92795, "loss_box_5": 1.55452, "loss_cns_5": 0.6631, "loss_yns_5": 0.14647, "loss_cls_dn_0": 0.21, "loss_box_dn_0": 0.76353, "loss_cls_dn_1": 0.13099, "loss_box_dn_1": 0.71072, "loss_cls_dn_2": 0.1352, "loss_box_dn_2": 0.70725, "loss_cls_dn_3": 0.13546, "loss_box_dn_3": 0.70751, "loss_cls_dn_4": 0.14067, "loss_box_dn_4": 0.70871, "loss_cls_dn_5": 0.14123, "loss_box_dn_5": 0.71247, "loss_dense_depth": 0.80627, "loss": 25.72437, "grad_norm": 35.07913, "time": 1.61902} -{"mode": "train", "epoch": 1, "iter": 185, "lr": 0.00017, "memory": 49164, "data_time": 0.06886, "loss_cls_0": 0.79353, "loss_box_0": 1.67015, "loss_cns_0": 0.62142, "loss_yns_0": 0.14752, "loss_cls_1": 0.89132, "loss_box_1": 1.59609, "loss_cns_1": 0.65811, "loss_yns_1": 0.1451, "loss_cls_2": 0.90413, "loss_box_2": 1.55939, "loss_cns_2": 0.66086, "loss_yns_2": 0.1446, "loss_cls_3": 0.90523, "loss_box_3": 1.55051, "loss_cns_3": 0.66035, "loss_yns_3": 0.14524, "loss_cls_4": 0.91401, "loss_box_4": 1.55075, "loss_cns_4": 0.66119, "loss_yns_4": 0.1445, "loss_cls_5": 0.91404, "loss_box_5": 1.54009, "loss_cns_5": 0.66075, "loss_yns_5": 0.14528, "loss_cls_dn_0": 0.21151, "loss_box_dn_0": 0.76334, "loss_cls_dn_1": 0.13372, "loss_box_dn_1": 0.71233, "loss_cls_dn_2": 0.13515, "loss_box_dn_2": 0.70403, "loss_cls_dn_3": 0.13631, "loss_box_dn_3": 0.71068, "loss_cls_dn_4": 0.14266, "loss_box_dn_4": 0.71775, "loss_cls_dn_5": 0.14497, "loss_box_dn_5": 0.7203, "loss_dense_depth": 0.82333, "loss": 25.64022, "grad_norm": 30.5972, "time": 1.61989} -{"mode": "train", "epoch": 1, "iter": 186, "lr": 0.00017, "memory": 49164, "data_time": 0.07379, "loss_cls_0": 0.80703, "loss_box_0": 1.64681, "loss_cns_0": 0.61567, "loss_yns_0": 0.14639, "loss_cls_1": 0.89693, "loss_box_1": 1.58895, "loss_cns_1": 0.66018, "loss_yns_1": 0.14773, "loss_cls_2": 0.90975, "loss_box_2": 1.54677, "loss_cns_2": 0.66299, "loss_yns_2": 0.14624, "loss_cls_3": 0.92098, "loss_box_3": 1.54079, "loss_cns_3": 0.66312, "loss_yns_3": 0.14654, "loss_cls_4": 0.92943, "loss_box_4": 1.57315, "loss_cns_4": 0.66228, "loss_yns_4": 0.14688, "loss_cls_5": 0.92508, "loss_box_5": 1.5668, "loss_cns_5": 0.66098, "loss_yns_5": 0.14664, "loss_cls_dn_0": 0.20643, "loss_box_dn_0": 0.76738, "loss_cls_dn_1": 0.13325, "loss_box_dn_1": 0.71937, "loss_cls_dn_2": 0.13424, "loss_box_dn_2": 0.71424, "loss_cls_dn_3": 0.13901, "loss_box_dn_3": 0.72837, "loss_cls_dn_4": 0.14678, "loss_box_dn_4": 0.75205, "loss_cls_dn_5": 0.14852, "loss_box_dn_5": 0.76173, "loss_dense_depth": 0.8222, "loss": 25.83167, "grad_norm": 41.34472, "time": 1.58745} -{"mode": "train", "epoch": 1, "iter": 187, "lr": 0.00017, "memory": 49164, "data_time": 0.07496, "loss_cls_0": 0.81031, "loss_box_0": 1.66262, "loss_cns_0": 0.62026, "loss_yns_0": 0.14545, "loss_cls_1": 0.91171, "loss_box_1": 1.60534, "loss_cns_1": 0.65655, "loss_yns_1": 0.14826, "loss_cls_2": 0.91927, "loss_box_2": 1.59066, "loss_cns_2": 0.65967, "loss_yns_2": 0.14889, "loss_cls_3": 0.93539, "loss_box_3": 1.57416, "loss_cns_3": 0.65983, "loss_yns_3": 0.15007, "loss_cls_4": 0.94067, "loss_box_4": 1.57888, "loss_cns_4": 0.6593, "loss_yns_4": 0.14994, "loss_cls_5": 0.93647, "loss_box_5": 1.58607, "loss_cns_5": 0.65825, "loss_yns_5": 0.14868, "loss_cls_dn_0": 0.20878, "loss_box_dn_0": 0.76266, "loss_cls_dn_1": 0.13061, "loss_box_dn_1": 0.74274, "loss_cls_dn_2": 0.13018, "loss_box_dn_2": 0.74279, "loss_cls_dn_3": 0.13429, "loss_box_dn_3": 0.75248, "loss_cls_dn_4": 0.14022, "loss_box_dn_4": 0.77255, "loss_cls_dn_5": 0.14546, "loss_box_dn_5": 0.78803, "loss_dense_depth": 0.82455, "loss": 26.13208, "grad_norm": 40.18722, "time": 1.57778} -{"mode": "train", "epoch": 1, "iter": 188, "lr": 0.00017, "memory": 49164, "data_time": 0.07335, "loss_cls_0": 0.80997, "loss_box_0": 1.71303, "loss_cns_0": 0.62084, "loss_yns_0": 0.14562, "loss_cls_1": 0.89709, "loss_box_1": 1.62622, "loss_cns_1": 0.65571, "loss_yns_1": 0.14461, "loss_cls_2": 0.91693, "loss_box_2": 1.61169, "loss_cns_2": 0.65932, "loss_yns_2": 0.14499, "loss_cls_3": 0.91657, "loss_box_3": 1.58116, "loss_cns_3": 0.65751, "loss_yns_3": 0.14618, "loss_cls_4": 0.92575, "loss_box_4": 1.58371, "loss_cns_4": 0.6572, "loss_yns_4": 0.14514, "loss_cls_5": 0.91948, "loss_box_5": 1.58954, "loss_cns_5": 0.65734, "loss_yns_5": 0.14455, "loss_cls_dn_0": 0.20488, "loss_box_dn_0": 0.76109, "loss_cls_dn_1": 0.13315, "loss_box_dn_1": 0.76381, "loss_cls_dn_2": 0.13406, "loss_box_dn_2": 0.76491, "loss_cls_dn_3": 0.13574, "loss_box_dn_3": 0.76134, "loss_cls_dn_4": 0.14151, "loss_box_dn_4": 0.77432, "loss_cls_dn_5": 0.1452, "loss_box_dn_5": 0.78323, "loss_dense_depth": 0.81423, "loss": 26.18761, "grad_norm": 41.56697, "time": 1.612} -{"mode": "train", "epoch": 1, "iter": 189, "lr": 0.00018, "memory": 49164, "data_time": 0.07299, "loss_cls_0": 0.81806, "loss_box_0": 1.66784, "loss_cns_0": 0.63111, "loss_yns_0": 0.15245, "loss_cls_1": 0.89231, "loss_box_1": 1.619, "loss_cns_1": 0.6597, "loss_yns_1": 0.14925, "loss_cls_2": 0.92082, "loss_box_2": 1.61285, "loss_cns_2": 0.65808, "loss_yns_2": 0.14846, "loss_cls_3": 0.94062, "loss_box_3": 1.5898, "loss_cns_3": 0.65886, "loss_yns_3": 0.15081, "loss_cls_4": 0.93618, "loss_box_4": 1.56542, "loss_cns_4": 0.66076, "loss_yns_4": 0.15172, "loss_cls_5": 0.93543, "loss_box_5": 1.56675, "loss_cns_5": 0.66012, "loss_yns_5": 0.15023, "loss_cls_dn_0": 0.20729, "loss_box_dn_0": 0.75711, "loss_cls_dn_1": 0.13633, "loss_box_dn_1": 0.75046, "loss_cls_dn_2": 0.13715, "loss_box_dn_2": 0.74511, "loss_cls_dn_3": 0.13897, "loss_box_dn_3": 0.73848, "loss_cls_dn_4": 0.14981, "loss_box_dn_4": 0.73359, "loss_cls_dn_5": 0.15512, "loss_box_dn_5": 0.73281, "loss_dense_depth": 0.82045, "loss": 26.0993, "grad_norm": 51.62281, "time": 1.64565} -{"mode": "train", "epoch": 1, "iter": 190, "lr": 0.00018, "memory": 49164, "data_time": 0.11819, "loss_cls_0": 0.81815, "loss_box_0": 1.70457, "loss_cns_0": 0.62384, "loss_yns_0": 0.15328, "loss_cls_1": 0.89644, "loss_box_1": 1.66275, "loss_cns_1": 0.6529, "loss_yns_1": 0.15002, "loss_cls_2": 0.93333, "loss_box_2": 1.6556, "loss_cns_2": 0.65565, "loss_yns_2": 0.15108, "loss_cls_3": 0.95183, "loss_box_3": 1.63253, "loss_cns_3": 0.656, "loss_yns_3": 0.1542, "loss_cls_4": 0.94403, "loss_box_4": 1.61403, "loss_cns_4": 0.65734, "loss_yns_4": 0.15496, "loss_cls_5": 0.93627, "loss_box_5": 1.62283, "loss_cns_5": 0.65744, "loss_yns_5": 0.15313, "loss_cls_dn_0": 0.21404, "loss_box_dn_0": 0.76347, "loss_cls_dn_1": 0.13732, "loss_box_dn_1": 0.71796, "loss_cls_dn_2": 0.14182, "loss_box_dn_2": 0.71154, "loss_cls_dn_3": 0.1421, "loss_box_dn_3": 0.71096, "loss_cls_dn_4": 0.14797, "loss_box_dn_4": 0.70923, "loss_cls_dn_5": 0.15324, "loss_box_dn_5": 0.71759, "loss_dense_depth": 0.82623, "loss": 26.28569, "grad_norm": 46.32416, "time": 1.59778} -{"mode": "train", "epoch": 1, "iter": 191, "lr": 0.00018, "memory": 49164, "data_time": 0.07268, "loss_cls_0": 0.81658, "loss_box_0": 1.70048, "loss_cns_0": 0.61675, "loss_yns_0": 0.15036, "loss_cls_1": 0.8969, "loss_box_1": 1.64735, "loss_cns_1": 0.65254, "loss_yns_1": 0.15226, "loss_cls_2": 0.93703, "loss_box_2": 1.62572, "loss_cns_2": 0.65487, "loss_yns_2": 0.15173, "loss_cls_3": 0.93146, "loss_box_3": 1.62099, "loss_cns_3": 0.65687, "loss_yns_3": 0.15439, "loss_cls_4": 0.9397, "loss_box_4": 1.64336, "loss_cns_4": 0.65283, "loss_yns_4": 0.15517, "loss_cls_5": 0.93608, "loss_box_5": 1.6501, "loss_cns_5": 0.65222, "loss_yns_5": 0.15254, "loss_cls_dn_0": 0.21477, "loss_box_dn_0": 0.75232, "loss_cls_dn_1": 0.13422, "loss_box_dn_1": 0.71294, "loss_cls_dn_2": 0.13789, "loss_box_dn_2": 0.71548, "loss_cls_dn_3": 0.13792, "loss_box_dn_3": 0.73437, "loss_cls_dn_4": 0.13944, "loss_box_dn_4": 0.75402, "loss_cls_dn_5": 0.14869, "loss_box_dn_5": 0.77568, "loss_dense_depth": 0.78831, "loss": 26.2943, "grad_norm": 52.97769, "time": 1.55483} -{"mode": "train", "epoch": 1, "iter": 192, "lr": 0.00018, "memory": 49164, "data_time": 0.07929, "loss_cls_0": 0.81609, "loss_box_0": 1.69344, "loss_cns_0": 0.62601, "loss_yns_0": 0.15108, "loss_cls_1": 0.89706, "loss_box_1": 1.68691, "loss_cns_1": 0.65297, "loss_yns_1": 0.15035, "loss_cls_2": 0.92651, "loss_box_2": 1.63186, "loss_cns_2": 0.6558, "loss_yns_2": 0.15019, "loss_cls_3": 0.93704, "loss_box_3": 1.64299, "loss_cns_3": 0.65694, "loss_yns_3": 0.15153, "loss_cls_4": 0.93236, "loss_box_4": 1.64469, "loss_cns_4": 0.65544, "loss_yns_4": 0.15266, "loss_cls_5": 0.93812, "loss_box_5": 1.63268, "loss_cns_5": 0.65486, "loss_yns_5": 0.15356, "loss_cls_dn_0": 0.21335, "loss_box_dn_0": 0.75124, "loss_cls_dn_1": 0.13453, "loss_box_dn_1": 0.75299, "loss_cls_dn_2": 0.1373, "loss_box_dn_2": 0.74954, "loss_cls_dn_3": 0.14179, "loss_box_dn_3": 0.77956, "loss_cls_dn_4": 0.14536, "loss_box_dn_4": 0.79199, "loss_cls_dn_5": 0.14943, "loss_box_dn_5": 0.81004, "loss_dense_depth": 0.77019, "loss": 26.51841, "grad_norm": 46.75566, "time": 1.56809} -{"mode": "train", "epoch": 1, "iter": 193, "lr": 0.00018, "memory": 49164, "data_time": 0.07581, "loss_cls_0": 0.79272, "loss_box_0": 1.71582, "loss_cns_0": 0.62667, "loss_yns_0": 0.14901, "loss_cls_1": 0.88463, "loss_box_1": 1.67567, "loss_cns_1": 0.65649, "loss_yns_1": 0.14833, "loss_cls_2": 0.90717, "loss_box_2": 1.61814, "loss_cns_2": 0.65649, "loss_yns_2": 0.14763, "loss_cls_3": 0.93571, "loss_box_3": 1.61698, "loss_cns_3": 0.65426, "loss_yns_3": 0.14848, "loss_cls_4": 0.92605, "loss_box_4": 1.61981, "loss_cns_4": 0.65832, "loss_yns_4": 0.14912, "loss_cls_5": 0.93164, "loss_box_5": 1.63372, "loss_cns_5": 0.65599, "loss_yns_5": 0.15084, "loss_cls_dn_0": 0.2081, "loss_box_dn_0": 0.77209, "loss_cls_dn_1": 0.13437, "loss_box_dn_1": 0.80059, "loss_cls_dn_2": 0.13527, "loss_box_dn_2": 0.79665, "loss_cls_dn_3": 0.14002, "loss_box_dn_3": 0.82514, "loss_cls_dn_4": 0.14327, "loss_box_dn_4": 0.83908, "loss_cls_dn_5": 0.14594, "loss_box_dn_5": 0.86059, "loss_dense_depth": 0.78581, "loss": 26.64659, "grad_norm": 54.76633, "time": 1.64618} -{"mode": "train", "epoch": 1, "iter": 194, "lr": 0.00018, "memory": 49164, "data_time": 0.07467, "loss_cls_0": 0.80485, "loss_box_0": 1.73437, "loss_cns_0": 0.62541, "loss_yns_0": 0.15237, "loss_cls_1": 0.91535, "loss_box_1": 1.69918, "loss_cns_1": 0.65299, "loss_yns_1": 0.15109, "loss_cls_2": 0.95564, "loss_box_2": 1.66703, "loss_cns_2": 0.65382, "loss_yns_2": 0.1515, "loss_cls_3": 0.93076, "loss_box_3": 1.6437, "loss_cns_3": 0.65362, "loss_yns_3": 0.15085, "loss_cls_4": 0.91664, "loss_box_4": 1.6665, "loss_cns_4": 0.65939, "loss_yns_4": 0.15383, "loss_cls_5": 0.92425, "loss_box_5": 1.65532, "loss_cns_5": 0.65269, "loss_yns_5": 0.15116, "loss_cls_dn_0": 0.20818, "loss_box_dn_0": 0.76209, "loss_cls_dn_1": 0.13083, "loss_box_dn_1": 0.81283, "loss_cls_dn_2": 0.13431, "loss_box_dn_2": 0.81114, "loss_cls_dn_3": 0.13382, "loss_box_dn_3": 0.82342, "loss_cls_dn_4": 0.13588, "loss_box_dn_4": 0.84456, "loss_cls_dn_5": 0.14203, "loss_box_dn_5": 0.85235, "loss_dense_depth": 0.77309, "loss": 26.88684, "grad_norm": 46.24131, "time": 1.561} -{"mode": "train", "epoch": 1, "iter": 195, "lr": 0.00018, "memory": 49164, "data_time": 0.09115, "loss_cls_0": 0.76824, "loss_box_0": 1.69631, "loss_cns_0": 0.63133, "loss_yns_0": 0.15129, "loss_cls_1": 0.8713, "loss_box_1": 1.62429, "loss_cns_1": 0.65436, "loss_yns_1": 0.14655, "loss_cls_2": 0.88411, "loss_box_2": 1.61636, "loss_cns_2": 0.65761, "loss_yns_2": 0.14613, "loss_cls_3": 0.89393, "loss_box_3": 1.61426, "loss_cns_3": 0.65596, "loss_yns_3": 0.148, "loss_cls_4": 0.89767, "loss_box_4": 1.6182, "loss_cns_4": 0.65859, "loss_yns_4": 0.14545, "loss_cls_5": 0.91169, "loss_box_5": 1.58288, "loss_cns_5": 0.65736, "loss_yns_5": 0.14652, "loss_cls_dn_0": 0.19753, "loss_box_dn_0": 0.75561, "loss_cls_dn_1": 0.12619, "loss_box_dn_1": 0.7779, "loss_cls_dn_2": 0.1291, "loss_box_dn_2": 0.78071, "loss_cls_dn_3": 0.13278, "loss_box_dn_3": 0.78356, "loss_cls_dn_4": 0.13456, "loss_box_dn_4": 0.79825, "loss_cls_dn_5": 0.13894, "loss_box_dn_5": 0.78628, "loss_dense_depth": 0.78911, "loss": 26.1089, "grad_norm": 61.27783, "time": 1.62817} -{"mode": "train", "epoch": 1, "iter": 196, "lr": 0.00018, "memory": 49164, "data_time": 0.07262, "loss_cls_0": 0.75737, "loss_box_0": 1.67586, "loss_cns_0": 0.62702, "loss_yns_0": 0.14898, "loss_cls_1": 0.91595, "loss_box_1": 1.58873, "loss_cns_1": 0.65283, "loss_yns_1": 0.14937, "loss_cls_2": 0.89488, "loss_box_2": 1.56628, "loss_cns_2": 0.65968, "loss_yns_2": 0.14902, "loss_cls_3": 0.87635, "loss_box_3": 1.55462, "loss_cns_3": 0.66336, "loss_yns_3": 0.14934, "loss_cls_4": 0.88553, "loss_box_4": 1.5502, "loss_cns_4": 0.66031, "loss_yns_4": 0.14769, "loss_cls_5": 0.90967, "loss_box_5": 1.53653, "loss_cns_5": 0.66106, "loss_yns_5": 0.14748, "loss_cls_dn_0": 0.19689, "loss_box_dn_0": 0.75775, "loss_cls_dn_1": 0.13036, "loss_box_dn_1": 0.77905, "loss_cls_dn_2": 0.13152, "loss_box_dn_2": 0.76847, "loss_cls_dn_3": 0.1393, "loss_box_dn_3": 0.76064, "loss_cls_dn_4": 0.14114, "loss_box_dn_4": 0.76119, "loss_cls_dn_5": 0.14258, "loss_box_dn_5": 0.75393, "loss_dense_depth": 0.75689, "loss": 25.7478, "grad_norm": 42.36605, "time": 1.5661} -{"mode": "train", "epoch": 1, "iter": 197, "lr": 0.00018, "memory": 49164, "data_time": 0.08046, "loss_cls_0": 0.79679, "loss_box_0": 1.73544, "loss_cns_0": 0.62037, "loss_yns_0": 0.14922, "loss_cls_1": 0.90463, "loss_box_1": 1.63806, "loss_cns_1": 0.65085, "loss_yns_1": 0.15084, "loss_cls_2": 0.90262, "loss_box_2": 1.65771, "loss_cns_2": 0.65521, "loss_yns_2": 0.15049, "loss_cls_3": 0.90239, "loss_box_3": 1.63209, "loss_cns_3": 0.65958, "loss_yns_3": 0.15056, "loss_cls_4": 0.89638, "loss_box_4": 1.62433, "loss_cns_4": 0.65764, "loss_yns_4": 0.151, "loss_cls_5": 0.89789, "loss_box_5": 1.64906, "loss_cns_5": 0.65408, "loss_yns_5": 0.15017, "loss_cls_dn_0": 0.20266, "loss_box_dn_0": 0.74948, "loss_cls_dn_1": 0.13206, "loss_box_dn_1": 0.70898, "loss_cls_dn_2": 0.13082, "loss_box_dn_2": 0.71052, "loss_cls_dn_3": 0.13603, "loss_box_dn_3": 0.70652, "loss_cls_dn_4": 0.14012, "loss_box_dn_4": 0.70325, "loss_cls_dn_5": 0.14505, "loss_box_dn_5": 0.7172, "loss_dense_depth": 0.80127, "loss": 26.02135, "grad_norm": 49.61551, "time": 1.64446} -{"mode": "train", "epoch": 1, "iter": 198, "lr": 0.00018, "memory": 49164, "data_time": 0.07495, "loss_cls_0": 0.80984, "loss_box_0": 1.75077, "loss_cns_0": 0.61906, "loss_yns_0": 0.15524, "loss_cls_1": 0.88175, "loss_box_1": 1.6438, "loss_cns_1": 0.65306, "loss_yns_1": 0.15387, "loss_cls_2": 0.90228, "loss_box_2": 1.6382, "loss_cns_2": 0.65282, "loss_yns_2": 0.15267, "loss_cls_3": 0.90762, "loss_box_3": 1.63381, "loss_cns_3": 0.65452, "loss_yns_3": 0.15238, "loss_cls_4": 0.90812, "loss_box_4": 1.62712, "loss_cns_4": 0.65574, "loss_yns_4": 0.15261, "loss_cls_5": 0.90876, "loss_box_5": 1.62472, "loss_cns_5": 0.65101, "loss_yns_5": 0.15489, "loss_cls_dn_0": 0.20163, "loss_box_dn_0": 0.76308, "loss_cls_dn_1": 0.13317, "loss_box_dn_1": 0.712, "loss_cls_dn_2": 0.13409, "loss_box_dn_2": 0.71872, "loss_cls_dn_3": 0.13577, "loss_box_dn_3": 0.73247, "loss_cls_dn_4": 0.1417, "loss_box_dn_4": 0.73784, "loss_cls_dn_5": 0.14535, "loss_box_dn_5": 0.74914, "loss_dense_depth": 0.79912, "loss": 26.14873, "grad_norm": 47.48387, "time": 1.57592} -{"mode": "train", "epoch": 1, "iter": 199, "lr": 0.00018, "memory": 49164, "data_time": 0.08362, "loss_cls_0": 0.79216, "loss_box_0": 1.74383, "loss_cns_0": 0.61129, "loss_yns_0": 0.15476, "loss_cls_1": 0.89454, "loss_box_1": 1.67316, "loss_cns_1": 0.65443, "loss_yns_1": 0.15489, "loss_cls_2": 0.90243, "loss_box_2": 1.65425, "loss_cns_2": 0.65606, "loss_yns_2": 0.15529, "loss_cls_3": 0.90143, "loss_box_3": 1.67033, "loss_cns_3": 0.65235, "loss_yns_3": 0.15467, "loss_cls_4": 0.91036, "loss_box_4": 1.64847, "loss_cns_4": 0.64949, "loss_yns_4": 0.15331, "loss_cls_5": 0.89981, "loss_box_5": 1.66333, "loss_cns_5": 0.64989, "loss_yns_5": 0.15538, "loss_cls_dn_0": 0.19813, "loss_box_dn_0": 0.76232, "loss_cls_dn_1": 0.12932, "loss_box_dn_1": 0.74404, "loss_cls_dn_2": 0.133, "loss_box_dn_2": 0.74649, "loss_cls_dn_3": 0.14227, "loss_box_dn_3": 0.77117, "loss_cls_dn_4": 0.14405, "loss_box_dn_4": 0.78268, "loss_cls_dn_5": 0.14386, "loss_box_dn_5": 0.80504, "loss_dense_depth": 0.78109, "loss": 26.43936, "grad_norm": 50.17291, "time": 1.56543} -{"mode": "train", "epoch": 1, "iter": 200, "lr": 0.00018, "memory": 49164, "data_time": 0.08266, "loss_cls_0": 0.79867, "loss_box_0": 1.72913, "loss_cns_0": 0.60865, "loss_yns_0": 0.15032, "loss_cls_1": 0.9202, "loss_box_1": 1.69885, "loss_cns_1": 0.65651, "loss_yns_1": 0.15196, "loss_cls_2": 0.9052, "loss_box_2": 1.65831, "loss_cns_2": 0.65458, "loss_yns_2": 0.15136, "loss_cls_3": 0.89072, "loss_box_3": 1.65624, "loss_cns_3": 0.65425, "loss_yns_3": 0.15009, "loss_cls_4": 0.89483, "loss_box_4": 1.64522, "loss_cns_4": 0.65116, "loss_yns_4": 0.15029, "loss_cls_5": 0.90958, "loss_box_5": 1.65504, "loss_cns_5": 0.65292, "loss_yns_5": 0.1497, "loss_cls_dn_0": 0.20179, "loss_box_dn_0": 0.7549, "loss_cls_dn_1": 0.12903, "loss_box_dn_1": 0.77131, "loss_cls_dn_2": 0.12923, "loss_box_dn_2": 0.77307, "loss_cls_dn_3": 0.13796, "loss_box_dn_3": 0.78104, "loss_cls_dn_4": 0.13541, "loss_box_dn_4": 0.78854, "loss_cls_dn_5": 0.13723, "loss_box_dn_5": 0.80909, "loss_dense_depth": 0.78233, "loss": 26.47471, "grad_norm": 40.38892, "time": 1.59605} -{"mode": "train", "epoch": 1, "iter": 201, "lr": 0.00018, "memory": 49164, "data_time": 0.1132, "loss_cls_0": 0.80296, "loss_box_0": 1.70407, "loss_cns_0": 0.61433, "loss_yns_0": 0.14851, "loss_cls_1": 0.88827, "loss_box_1": 1.72475, "loss_cns_1": 0.65323, "loss_yns_1": 0.15036, "loss_cls_2": 0.90041, "loss_box_2": 1.68889, "loss_cns_2": 0.6502, "loss_yns_2": 0.14866, "loss_cls_3": 0.90442, "loss_box_3": 1.66612, "loss_cns_3": 0.6506, "loss_yns_3": 0.14878, "loss_cls_4": 0.90644, "loss_box_4": 1.68166, "loss_cns_4": 0.64991, "loss_yns_4": 0.14825, "loss_cls_5": 0.93012, "loss_box_5": 1.65872, "loss_cns_5": 0.65298, "loss_yns_5": 0.15059, "loss_cls_dn_0": 0.19953, "loss_box_dn_0": 0.7579, "loss_cls_dn_1": 0.12391, "loss_box_dn_1": 0.79466, "loss_cls_dn_2": 0.12535, "loss_box_dn_2": 0.79563, "loss_cls_dn_3": 0.13398, "loss_box_dn_3": 0.78889, "loss_cls_dn_4": 0.13765, "loss_box_dn_4": 0.80183, "loss_cls_dn_5": 0.14672, "loss_box_dn_5": 0.80275, "loss_dense_depth": 0.74904, "loss": 26.58105, "grad_norm": 51.67408, "time": 1.67644} -{"mode": "train", "epoch": 1, "iter": 202, "lr": 0.00018, "memory": 49164, "data_time": 0.17115, "loss_cls_0": 0.7737, "loss_box_0": 1.68389, "loss_cns_0": 0.61809, "loss_yns_0": 0.15117, "loss_cls_1": 0.87942, "loss_box_1": 1.65205, "loss_cns_1": 0.64803, "loss_yns_1": 0.15064, "loss_cls_2": 0.88443, "loss_box_2": 1.62027, "loss_cns_2": 0.65596, "loss_yns_2": 0.15105, "loss_cls_3": 0.89263, "loss_box_3": 1.60924, "loss_cns_3": 0.6528, "loss_yns_3": 0.15054, "loss_cls_4": 0.89844, "loss_box_4": 1.60321, "loss_cns_4": 0.65363, "loss_yns_4": 0.15098, "loss_cls_5": 0.91098, "loss_box_5": 1.588, "loss_cns_5": 0.65584, "loss_yns_5": 0.15144, "loss_cls_dn_0": 0.19251, "loss_box_dn_0": 0.76351, "loss_cls_dn_1": 0.1259, "loss_box_dn_1": 0.77634, "loss_cls_dn_2": 0.12716, "loss_box_dn_2": 0.76356, "loss_cls_dn_3": 0.12843, "loss_box_dn_3": 0.76132, "loss_cls_dn_4": 0.13678, "loss_box_dn_4": 0.76694, "loss_cls_dn_5": 0.1484, "loss_box_dn_5": 0.76615, "loss_dense_depth": 0.785, "loss": 26.02844, "grad_norm": 36.06106, "time": 1.70192} -{"mode": "train", "epoch": 1, "iter": 203, "lr": 0.00018, "memory": 49164, "data_time": 0.07477, "loss_cls_0": 0.80374, "loss_box_0": 1.71331, "loss_cns_0": 0.61755, "loss_yns_0": 0.1502, "loss_cls_1": 0.86612, "loss_box_1": 1.71542, "loss_cns_1": 0.64182, "loss_yns_1": 0.14829, "loss_cls_2": 0.91162, "loss_box_2": 1.69898, "loss_cns_2": 0.64373, "loss_yns_2": 0.15013, "loss_cls_3": 0.92645, "loss_box_3": 1.69033, "loss_cns_3": 0.64718, "loss_yns_3": 0.15326, "loss_cls_4": 0.9091, "loss_box_4": 1.68543, "loss_cns_4": 0.64833, "loss_yns_4": 0.15326, "loss_cls_5": 0.92035, "loss_box_5": 1.69673, "loss_cns_5": 0.64767, "loss_yns_5": 0.15227, "loss_cls_dn_0": 0.19288, "loss_box_dn_0": 0.75694, "loss_cls_dn_1": 0.12388, "loss_box_dn_1": 0.77403, "loss_cls_dn_2": 0.12772, "loss_box_dn_2": 0.75769, "loss_cls_dn_3": 0.12881, "loss_box_dn_3": 0.75363, "loss_cls_dn_4": 0.12982, "loss_box_dn_4": 0.75239, "loss_cls_dn_5": 0.13544, "loss_box_dn_5": 0.76059, "loss_dense_depth": 0.80105, "loss": 26.48617, "grad_norm": 54.73073, "time": 1.57209} -{"mode": "train", "epoch": 1, "iter": 204, "lr": 0.00018, "memory": 49164, "data_time": 0.09116, "loss_cls_0": 0.78934, "loss_box_0": 1.70924, "loss_cns_0": 0.62478, "loss_yns_0": 0.14881, "loss_cls_1": 0.8837, "loss_box_1": 1.68632, "loss_cns_1": 0.65138, "loss_yns_1": 0.14618, "loss_cls_2": 0.9017, "loss_box_2": 1.65183, "loss_cns_2": 0.65195, "loss_yns_2": 0.14875, "loss_cls_3": 0.91079, "loss_box_3": 1.62321, "loss_cns_3": 0.65008, "loss_yns_3": 0.15042, "loss_cls_4": 0.89734, "loss_box_4": 1.64517, "loss_cns_4": 0.65593, "loss_yns_4": 0.14986, "loss_cls_5": 0.91994, "loss_box_5": 1.64667, "loss_cns_5": 0.65674, "loss_yns_5": 0.15117, "loss_cls_dn_0": 0.19708, "loss_box_dn_0": 0.74958, "loss_cls_dn_1": 0.12109, "loss_box_dn_1": 0.69668, "loss_cls_dn_2": 0.122, "loss_box_dn_2": 0.68163, "loss_cls_dn_3": 0.12449, "loss_box_dn_3": 0.68417, "loss_cls_dn_4": 0.1256, "loss_box_dn_4": 0.69649, "loss_cls_dn_5": 0.13545, "loss_box_dn_5": 0.71007, "loss_dense_depth": 0.80636, "loss": 25.90202, "grad_norm": 44.18729, "time": 1.59617} -{"mode": "train", "epoch": 1, "iter": 205, "lr": 0.00018, "memory": 49164, "data_time": 0.07804, "loss_cls_0": 0.78621, "loss_box_0": 1.75217, "loss_cns_0": 0.61794, "loss_yns_0": 0.15142, "loss_cls_1": 0.89471, "loss_box_1": 1.64907, "loss_cns_1": 0.64298, "loss_yns_1": 0.14753, "loss_cls_2": 0.92168, "loss_box_2": 1.611, "loss_cns_2": 0.64802, "loss_yns_2": 0.14827, "loss_cls_3": 0.93102, "loss_box_3": 1.61783, "loss_cns_3": 0.64509, "loss_yns_3": 0.14825, "loss_cls_4": 0.91625, "loss_box_4": 1.66016, "loss_cns_4": 0.65191, "loss_yns_4": 0.14996, "loss_cls_5": 0.92474, "loss_box_5": 1.66294, "loss_cns_5": 0.65022, "loss_yns_5": 0.15085, "loss_cls_dn_0": 0.20365, "loss_box_dn_0": 0.77351, "loss_cls_dn_1": 0.1324, "loss_box_dn_1": 0.71741, "loss_cls_dn_2": 0.13177, "loss_box_dn_2": 0.71468, "loss_cls_dn_3": 0.13209, "loss_box_dn_3": 0.73844, "loss_cls_dn_4": 0.13226, "loss_box_dn_4": 0.76321, "loss_cls_dn_5": 0.1407, "loss_box_dn_5": 0.79047, "loss_dense_depth": 0.79165, "loss": 26.24245, "grad_norm": 60.03418, "time": 1.69261} -{"mode": "train", "epoch": 1, "iter": 206, "lr": 0.00018, "memory": 49164, "data_time": 0.07453, "loss_cls_0": 0.79778, "loss_box_0": 1.72187, "loss_cns_0": 0.62557, "loss_yns_0": 0.15215, "loss_cls_1": 0.88045, "loss_box_1": 1.69816, "loss_cns_1": 0.6505, "loss_yns_1": 0.14993, "loss_cls_2": 0.90868, "loss_box_2": 1.64566, "loss_cns_2": 0.6568, "loss_yns_2": 0.15182, "loss_cls_3": 0.91296, "loss_box_3": 1.65234, "loss_cns_3": 0.65362, "loss_yns_3": 0.15105, "loss_cls_4": 0.914, "loss_box_4": 1.66284, "loss_cns_4": 0.65514, "loss_yns_4": 0.15199, "loss_cls_5": 0.92642, "loss_box_5": 1.68521, "loss_cns_5": 0.6618, "loss_yns_5": 0.15203, "loss_cls_dn_0": 0.20377, "loss_box_dn_0": 0.76355, "loss_cls_dn_1": 0.12703, "loss_box_dn_1": 0.75113, "loss_cls_dn_2": 0.12969, "loss_box_dn_2": 0.74817, "loss_cls_dn_3": 0.1305, "loss_box_dn_3": 0.76951, "loss_cls_dn_4": 0.13214, "loss_box_dn_4": 0.79248, "loss_cls_dn_5": 0.14406, "loss_box_dn_5": 0.82852, "loss_dense_depth": 0.80839, "loss": 26.54771, "grad_norm": 48.64428, "time": 1.57698} -{"mode": "train", "epoch": 1, "iter": 207, "lr": 0.00018, "memory": 49164, "data_time": 0.09145, "loss_cls_0": 0.78734, "loss_box_0": 1.72775, "loss_cns_0": 0.62396, "loss_yns_0": 0.15372, "loss_cls_1": 0.90606, "loss_box_1": 1.69979, "loss_cns_1": 0.65483, "loss_yns_1": 0.15074, "loss_cls_2": 0.91926, "loss_box_2": 1.66994, "loss_cns_2": 0.65885, "loss_yns_2": 0.15265, "loss_cls_3": 0.92068, "loss_box_3": 1.65595, "loss_cns_3": 0.65492, "loss_yns_3": 0.15279, "loss_cls_4": 0.91083, "loss_box_4": 1.66791, "loss_cns_4": 0.65534, "loss_yns_4": 0.15415, "loss_cls_5": 0.93206, "loss_box_5": 1.66519, "loss_cns_5": 0.66209, "loss_yns_5": 0.15201, "loss_cls_dn_0": 0.20414, "loss_box_dn_0": 0.75339, "loss_cls_dn_1": 0.1337, "loss_box_dn_1": 0.77275, "loss_cls_dn_2": 0.14297, "loss_box_dn_2": 0.77056, "loss_cls_dn_3": 0.1367, "loss_box_dn_3": 0.78351, "loss_cls_dn_4": 0.13597, "loss_box_dn_4": 0.80692, "loss_cls_dn_5": 0.14563, "loss_box_dn_5": 0.82883, "loss_dense_depth": 0.7957, "loss": 26.69957, "grad_norm": 57.28479, "time": 1.58173} -{"mode": "train", "epoch": 1, "iter": 208, "lr": 0.00018, "memory": 49164, "data_time": 0.07411, "loss_cls_0": 0.79141, "loss_box_0": 1.73117, "loss_cns_0": 0.62263, "loss_yns_0": 0.15474, "loss_cls_1": 0.90139, "loss_box_1": 1.7325, "loss_cns_1": 0.65316, "loss_yns_1": 0.15351, "loss_cls_2": 0.90739, "loss_box_2": 1.67214, "loss_cns_2": 0.65268, "loss_yns_2": 0.15471, "loss_cls_3": 0.91254, "loss_box_3": 1.65752, "loss_cns_3": 0.6539, "loss_yns_3": 0.15538, "loss_cls_4": 0.90685, "loss_box_4": 1.67589, "loss_cns_4": 0.65289, "loss_yns_4": 0.15568, "loss_cls_5": 0.92635, "loss_box_5": 1.66087, "loss_cns_5": 0.65311, "loss_yns_5": 0.15378, "loss_cls_dn_0": 0.20209, "loss_box_dn_0": 0.76331, "loss_cls_dn_1": 0.12969, "loss_box_dn_1": 0.76789, "loss_cls_dn_2": 0.13606, "loss_box_dn_2": 0.75326, "loss_cls_dn_3": 0.1325, "loss_box_dn_3": 0.75667, "loss_cls_dn_4": 0.13387, "loss_box_dn_4": 0.77556, "loss_cls_dn_5": 0.14059, "loss_box_dn_5": 0.78342, "loss_dense_depth": 0.76729, "loss": 26.53437, "grad_norm": 56.33257, "time": 1.59601} -{"mode": "train", "epoch": 1, "iter": 209, "lr": 0.00018, "memory": 49164, "data_time": 0.10775, "loss_cls_0": 0.82447, "loss_box_0": 1.75137, "loss_cns_0": 0.61603, "loss_yns_0": 0.15197, "loss_cls_1": 0.89764, "loss_box_1": 1.69838, "loss_cns_1": 0.65168, "loss_yns_1": 0.1505, "loss_cls_2": 0.90826, "loss_box_2": 1.65915, "loss_cns_2": 0.65267, "loss_yns_2": 0.15001, "loss_cls_3": 0.92181, "loss_box_3": 1.65085, "loss_cns_3": 0.65484, "loss_yns_3": 0.15109, "loss_cls_4": 0.92378, "loss_box_4": 1.63338, "loss_cns_4": 0.65773, "loss_yns_4": 0.1499, "loss_cls_5": 0.92873, "loss_box_5": 1.64875, "loss_cns_5": 0.65649, "loss_yns_5": 0.1509, "loss_cls_dn_0": 0.20698, "loss_box_dn_0": 0.75993, "loss_cls_dn_1": 0.12775, "loss_box_dn_1": 0.7539, "loss_cls_dn_2": 0.12841, "loss_box_dn_2": 0.7469, "loss_cls_dn_3": 0.1323, "loss_box_dn_3": 0.74671, "loss_cls_dn_4": 0.13959, "loss_box_dn_4": 0.74856, "loss_cls_dn_5": 0.14634, "loss_box_dn_5": 0.75904, "loss_dense_depth": 0.82204, "loss": 26.45886, "grad_norm": 48.31931, "time": 1.60549} -{"mode": "train", "epoch": 1, "iter": 210, "lr": 0.00018, "memory": 49164, "data_time": 0.07331, "loss_cls_0": 0.78217, "loss_box_0": 1.71801, "loss_cns_0": 0.6193, "loss_yns_0": 0.1513, "loss_cls_1": 0.87843, "loss_box_1": 1.65013, "loss_cns_1": 0.65212, "loss_yns_1": 0.14804, "loss_cls_2": 0.90129, "loss_box_2": 1.62788, "loss_cns_2": 0.65363, "loss_yns_2": 0.14997, "loss_cls_3": 0.91514, "loss_box_3": 1.61328, "loss_cns_3": 0.65435, "loss_yns_3": 0.14962, "loss_cls_4": 0.91309, "loss_box_4": 1.58952, "loss_cns_4": 0.65876, "loss_yns_4": 0.15013, "loss_cls_5": 0.90103, "loss_box_5": 1.61126, "loss_cns_5": 0.65933, "loss_yns_5": 0.14942, "loss_cls_dn_0": 0.191, "loss_box_dn_0": 0.75381, "loss_cls_dn_1": 0.1269, "loss_box_dn_1": 0.73273, "loss_cls_dn_2": 0.12574, "loss_box_dn_2": 0.72529, "loss_cls_dn_3": 0.12818, "loss_box_dn_3": 0.71719, "loss_cls_dn_4": 0.13517, "loss_box_dn_4": 0.71056, "loss_cls_dn_5": 0.13577, "loss_box_dn_5": 0.71867, "loss_dense_depth": 0.75899, "loss": 25.85719, "grad_norm": 46.29446, "time": 1.59811} -{"mode": "train", "epoch": 1, "iter": 211, "lr": 0.00018, "memory": 49164, "data_time": 0.07305, "loss_cls_0": 0.78344, "loss_box_0": 1.70313, "loss_cns_0": 0.62364, "loss_yns_0": 0.14967, "loss_cls_1": 0.88592, "loss_box_1": 1.60985, "loss_cns_1": 0.65511, "loss_yns_1": 0.14892, "loss_cls_2": 0.90781, "loss_box_2": 1.57201, "loss_cns_2": 0.65708, "loss_yns_2": 0.14944, "loss_cls_3": 0.92016, "loss_box_3": 1.5767, "loss_cns_3": 0.65888, "loss_yns_3": 0.14901, "loss_cls_4": 0.91289, "loss_box_4": 1.57995, "loss_cns_4": 0.65867, "loss_yns_4": 0.14964, "loss_cls_5": 0.91175, "loss_box_5": 1.58232, "loss_cns_5": 0.65692, "loss_yns_5": 0.14867, "loss_cls_dn_0": 0.1893, "loss_box_dn_0": 0.75152, "loss_cls_dn_1": 0.12421, "loss_box_dn_1": 0.69902, "loss_cls_dn_2": 0.12167, "loss_box_dn_2": 0.68849, "loss_cls_dn_3": 0.12293, "loss_box_dn_3": 0.69843, "loss_cls_dn_4": 0.12983, "loss_box_dn_4": 0.71076, "loss_cls_dn_5": 0.13195, "loss_box_dn_5": 0.71865, "loss_dense_depth": 0.76985, "loss": 25.60821, "grad_norm": 49.26228, "time": 1.56177} -{"mode": "train", "epoch": 1, "iter": 212, "lr": 0.00018, "memory": 49164, "data_time": 0.12696, "loss_cls_0": 0.78395, "loss_box_0": 1.69923, "loss_cns_0": 0.62776, "loss_yns_0": 0.14747, "loss_cls_1": 0.88151, "loss_box_1": 1.65132, "loss_cns_1": 0.65034, "loss_yns_1": 0.14608, "loss_cls_2": 0.89001, "loss_box_2": 1.63139, "loss_cns_2": 0.65322, "loss_yns_2": 0.14601, "loss_cls_3": 0.89283, "loss_box_3": 1.64666, "loss_cns_3": 0.65739, "loss_yns_3": 0.14777, "loss_cls_4": 0.89724, "loss_box_4": 1.64986, "loss_cns_4": 0.66008, "loss_yns_4": 0.14755, "loss_cls_5": 0.90611, "loss_box_5": 1.6572, "loss_cns_5": 0.65293, "loss_yns_5": 0.14706, "loss_cls_dn_0": 0.19669, "loss_box_dn_0": 0.75359, "loss_cls_dn_1": 0.12891, "loss_box_dn_1": 0.71523, "loss_cls_dn_2": 0.12734, "loss_box_dn_2": 0.7216, "loss_cls_dn_3": 0.12665, "loss_box_dn_3": 0.74393, "loss_cls_dn_4": 0.12977, "loss_box_dn_4": 0.7629, "loss_cls_dn_5": 0.13299, "loss_box_dn_5": 0.77962, "loss_dense_depth": 0.76314, "loss": 26.05335, "grad_norm": 59.34396, "time": 1.67967} -{"mode": "train", "epoch": 1, "iter": 213, "lr": 0.00018, "memory": 49164, "data_time": 0.07929, "loss_cls_0": 0.79037, "loss_box_0": 1.68838, "loss_cns_0": 0.6289, "loss_yns_0": 0.15309, "loss_cls_1": 0.85908, "loss_box_1": 1.63914, "loss_cns_1": 0.65003, "loss_yns_1": 0.14941, "loss_cls_2": 0.87531, "loss_box_2": 1.61304, "loss_cns_2": 0.65638, "loss_yns_2": 0.15047, "loss_cls_3": 0.88621, "loss_box_3": 1.58986, "loss_cns_3": 0.65607, "loss_yns_3": 0.14921, "loss_cls_4": 0.88127, "loss_box_4": 1.60105, "loss_cns_4": 0.65731, "loss_yns_4": 0.14833, "loss_cls_5": 0.8795, "loss_box_5": 1.61054, "loss_cns_5": 0.65617, "loss_yns_5": 0.14861, "loss_cls_dn_0": 0.1982, "loss_box_dn_0": 0.74794, "loss_cls_dn_1": 0.12623, "loss_box_dn_1": 0.75894, "loss_cls_dn_2": 0.12755, "loss_box_dn_2": 0.77192, "loss_cls_dn_3": 0.126, "loss_box_dn_3": 0.79405, "loss_cls_dn_4": 0.13233, "loss_box_dn_4": 0.81729, "loss_cls_dn_5": 0.14091, "loss_box_dn_5": 0.83939, "loss_dense_depth": 0.79142, "loss": 26.08993, "grad_norm": 36.28946, "time": 1.64058} -{"mode": "train", "epoch": 1, "iter": 214, "lr": 0.00019, "memory": 49164, "data_time": 0.10597, "loss_cls_0": 0.78263, "loss_box_0": 1.7086, "loss_cns_0": 0.62555, "loss_yns_0": 0.153, "loss_cls_1": 0.864, "loss_box_1": 1.65067, "loss_cns_1": 0.64967, "loss_yns_1": 0.15086, "loss_cls_2": 0.89392, "loss_box_2": 1.63602, "loss_cns_2": 0.65332, "loss_yns_2": 0.15104, "loss_cls_3": 0.9151, "loss_box_3": 1.6102, "loss_cns_3": 0.65498, "loss_yns_3": 0.15092, "loss_cls_4": 0.89641, "loss_box_4": 1.64085, "loss_cns_4": 0.65254, "loss_yns_4": 0.15034, "loss_cls_5": 0.90442, "loss_box_5": 1.64356, "loss_cns_5": 0.65191, "loss_yns_5": 0.15141, "loss_cls_dn_0": 0.19618, "loss_box_dn_0": 0.75902, "loss_cls_dn_1": 0.12517, "loss_box_dn_1": 0.78784, "loss_cls_dn_2": 0.12624, "loss_box_dn_2": 0.79875, "loss_cls_dn_3": 0.12909, "loss_box_dn_3": 0.80977, "loss_cls_dn_4": 0.13959, "loss_box_dn_4": 0.82876, "loss_cls_dn_5": 0.14741, "loss_box_dn_5": 0.8439, "loss_dense_depth": 0.8079, "loss": 26.44155, "grad_norm": 54.90149, "time": 1.66028} -{"mode": "train", "epoch": 1, "iter": 215, "lr": 0.00019, "memory": 49164, "data_time": 0.09028, "loss_cls_0": 0.81259, "loss_box_0": 1.70579, "loss_cns_0": 0.6213, "loss_yns_0": 0.15077, "loss_cls_1": 0.88959, "loss_box_1": 1.65667, "loss_cns_1": 0.65136, "loss_yns_1": 0.15121, "loss_cls_2": 0.89785, "loss_box_2": 1.6399, "loss_cns_2": 0.65382, "loss_yns_2": 0.15074, "loss_cls_3": 0.92092, "loss_box_3": 1.59751, "loss_cns_3": 0.65776, "loss_yns_3": 0.1516, "loss_cls_4": 0.9082, "loss_box_4": 1.60552, "loss_cns_4": 0.65643, "loss_yns_4": 0.15041, "loss_cls_5": 0.92046, "loss_box_5": 1.60469, "loss_cns_5": 0.65618, "loss_yns_5": 0.1506, "loss_cls_dn_0": 0.20505, "loss_box_dn_0": 0.74962, "loss_cls_dn_1": 0.13299, "loss_box_dn_1": 0.76744, "loss_cls_dn_2": 0.13065, "loss_box_dn_2": 0.76824, "loss_cls_dn_3": 0.1341, "loss_box_dn_3": 0.7613, "loss_cls_dn_4": 0.13861, "loss_box_dn_4": 0.76619, "loss_cls_dn_5": 0.1416, "loss_box_dn_5": 0.7722, "loss_dense_depth": 0.80802, "loss": 26.23785, "grad_norm": 42.94695, "time": 1.58353} -{"mode": "train", "epoch": 1, "iter": 216, "lr": 0.00019, "memory": 49164, "data_time": 0.0772, "loss_cls_0": 0.81894, "loss_box_0": 1.7395, "loss_cns_0": 0.6193, "loss_yns_0": 0.14941, "loss_cls_1": 0.92152, "loss_box_1": 1.6689, "loss_cns_1": 0.65452, "loss_yns_1": 0.15479, "loss_cls_2": 0.90885, "loss_box_2": 1.61671, "loss_cns_2": 0.65577, "loss_yns_2": 0.15149, "loss_cls_3": 0.9138, "loss_box_3": 1.59236, "loss_cns_3": 0.65531, "loss_yns_3": 0.1523, "loss_cls_4": 0.90988, "loss_box_4": 1.59179, "loss_cns_4": 0.657, "loss_yns_4": 0.15263, "loss_cls_5": 0.9219, "loss_box_5": 1.59646, "loss_cns_5": 0.65727, "loss_yns_5": 0.15262, "loss_cls_dn_0": 0.20777, "loss_box_dn_0": 0.76753, "loss_cls_dn_1": 0.13925, "loss_box_dn_1": 0.77329, "loss_cls_dn_2": 0.13459, "loss_box_dn_2": 0.75803, "loss_cls_dn_3": 0.13948, "loss_box_dn_3": 0.74468, "loss_cls_dn_4": 0.14466, "loss_box_dn_4": 0.74306, "loss_cls_dn_5": 0.14648, "loss_box_dn_5": 0.74301, "loss_dense_depth": 0.7935, "loss": 26.24834, "grad_norm": 41.35535, "time": 1.56626} -{"mode": "train", "epoch": 1, "iter": 217, "lr": 0.00019, "memory": 49164, "data_time": 0.0749, "loss_cls_0": 0.79175, "loss_box_0": 1.70743, "loss_cns_0": 0.62219, "loss_yns_0": 0.15769, "loss_cls_1": 0.85989, "loss_box_1": 1.65155, "loss_cns_1": 0.65093, "loss_yns_1": 0.15605, "loss_cls_2": 0.8796, "loss_box_2": 1.59974, "loss_cns_2": 0.64983, "loss_yns_2": 0.15462, "loss_cls_3": 0.88525, "loss_box_3": 1.58499, "loss_cns_3": 0.65996, "loss_yns_3": 0.15458, "loss_cls_4": 0.88969, "loss_box_4": 1.57761, "loss_cns_4": 0.65266, "loss_yns_4": 0.15412, "loss_cls_5": 0.89309, "loss_box_5": 1.58773, "loss_cns_5": 0.65555, "loss_yns_5": 0.15499, "loss_cls_dn_0": 0.19233, "loss_box_dn_0": 0.75594, "loss_cls_dn_1": 0.12823, "loss_box_dn_1": 0.73104, "loss_cls_dn_2": 0.13284, "loss_box_dn_2": 0.71356, "loss_cls_dn_3": 0.13576, "loss_box_dn_3": 0.71116, "loss_cls_dn_4": 0.14362, "loss_box_dn_4": 0.71366, "loss_cls_dn_5": 0.1501, "loss_box_dn_5": 0.71992, "loss_dense_depth": 0.73792, "loss": 25.69758, "grad_norm": 53.99391, "time": 1.56106} -{"mode": "train", "epoch": 1, "iter": 218, "lr": 0.00019, "memory": 49164, "data_time": 0.09213, "loss_cls_0": 0.78519, "loss_box_0": 1.68004, "loss_cns_0": 0.62561, "loss_yns_0": 0.15371, "loss_cls_1": 0.87888, "loss_box_1": 1.64359, "loss_cns_1": 0.65152, "loss_yns_1": 0.15073, "loss_cls_2": 0.88796, "loss_box_2": 1.60653, "loss_cns_2": 0.64828, "loss_yns_2": 0.15088, "loss_cls_3": 0.88623, "loss_box_3": 1.60358, "loss_cns_3": 0.66127, "loss_yns_3": 0.15312, "loss_cls_4": 0.87848, "loss_box_4": 1.60423, "loss_cns_4": 0.65648, "loss_yns_4": 0.15368, "loss_cls_5": 0.88771, "loss_box_5": 1.606, "loss_cns_5": 0.65748, "loss_yns_5": 0.15476, "loss_cls_dn_0": 0.19229, "loss_box_dn_0": 0.76122, "loss_cls_dn_1": 0.13654, "loss_box_dn_1": 0.69745, "loss_cls_dn_2": 0.14313, "loss_box_dn_2": 0.69243, "loss_cls_dn_3": 0.1429, "loss_box_dn_3": 0.70341, "loss_cls_dn_4": 0.14431, "loss_box_dn_4": 0.71801, "loss_cls_dn_5": 0.15029, "loss_box_dn_5": 0.73513, "loss_dense_depth": 0.75236, "loss": 25.7354, "grad_norm": 46.58086, "time": 1.58572} -{"mode": "train", "epoch": 1, "iter": 219, "lr": 0.00019, "memory": 49164, "data_time": 0.08427, "loss_cls_0": 0.79516, "loss_box_0": 1.71854, "loss_cns_0": 0.62096, "loss_yns_0": 0.15417, "loss_cls_1": 0.88909, "loss_box_1": 1.6796, "loss_cns_1": 0.64709, "loss_yns_1": 0.15148, "loss_cls_2": 0.8921, "loss_box_2": 1.63922, "loss_cns_2": 0.65151, "loss_yns_2": 0.15085, "loss_cls_3": 0.90563, "loss_box_3": 1.61989, "loss_cns_3": 0.65476, "loss_yns_3": 0.15293, "loss_cls_4": 0.91027, "loss_box_4": 1.61706, "loss_cns_4": 0.65392, "loss_yns_4": 0.15341, "loss_cls_5": 0.9153, "loss_box_5": 1.63216, "loss_cns_5": 0.65538, "loss_yns_5": 0.15418, "loss_cls_dn_0": 0.20247, "loss_box_dn_0": 0.76083, "loss_cls_dn_1": 0.13758, "loss_box_dn_1": 0.73514, "loss_cls_dn_2": 0.13799, "loss_box_dn_2": 0.74145, "loss_cls_dn_3": 0.13629, "loss_box_dn_3": 0.7623, "loss_cls_dn_4": 0.1388, "loss_box_dn_4": 0.78647, "loss_cls_dn_5": 0.14607, "loss_box_dn_5": 0.81826, "loss_dense_depth": 0.73171, "loss": 26.25002, "grad_norm": 50.19263, "time": 1.58636} -{"mode": "train", "epoch": 1, "iter": 220, "lr": 0.00019, "memory": 49164, "data_time": 0.09132, "loss_cls_0": 0.78048, "loss_box_0": 1.70379, "loss_cns_0": 0.62804, "loss_yns_0": 0.15218, "loss_cls_1": 0.85832, "loss_box_1": 1.66067, "loss_cns_1": 0.65298, "loss_yns_1": 0.15093, "loss_cls_2": 0.8685, "loss_box_2": 1.64857, "loss_cns_2": 0.65471, "loss_yns_2": 0.15046, "loss_cls_3": 0.90348, "loss_box_3": 1.61991, "loss_cns_3": 0.66046, "loss_yns_3": 0.15303, "loss_cls_4": 0.91996, "loss_box_4": 1.624, "loss_cns_4": 0.65658, "loss_yns_4": 0.15331, "loss_cls_5": 0.90481, "loss_box_5": 1.63209, "loss_cns_5": 0.66172, "loss_yns_5": 0.15281, "loss_cls_dn_0": 0.19817, "loss_box_dn_0": 0.74876, "loss_cls_dn_1": 0.13369, "loss_box_dn_1": 0.74448, "loss_cls_dn_2": 0.13218, "loss_box_dn_2": 0.74987, "loss_cls_dn_3": 0.13333, "loss_box_dn_3": 0.76911, "loss_cls_dn_4": 0.14144, "loss_box_dn_4": 0.79231, "loss_cls_dn_5": 0.14382, "loss_box_dn_5": 0.82578, "loss_dense_depth": 0.74998, "loss": 26.21473, "grad_norm": 57.41725, "time": 1.61471} -{"mode": "train", "epoch": 1, "iter": 221, "lr": 0.00019, "memory": 49164, "data_time": 0.11768, "loss_cls_0": 0.78234, "loss_box_0": 1.71467, "loss_cns_0": 0.62436, "loss_yns_0": 0.15162, "loss_cls_1": 0.86017, "loss_box_1": 1.6697, "loss_cns_1": 0.65436, "loss_yns_1": 0.15161, "loss_cls_2": 0.88272, "loss_box_2": 1.68376, "loss_cns_2": 0.65361, "loss_yns_2": 0.15212, "loss_cls_3": 0.895, "loss_box_3": 1.65345, "loss_cns_3": 0.66207, "loss_yns_3": 0.15492, "loss_cls_4": 0.88757, "loss_box_4": 1.65664, "loss_cns_4": 0.65334, "loss_yns_4": 0.15338, "loss_cls_5": 0.88326, "loss_box_5": 1.64644, "loss_cns_5": 0.66247, "loss_yns_5": 0.15216, "loss_cls_dn_0": 0.19626, "loss_box_dn_0": 0.76043, "loss_cls_dn_1": 0.1372, "loss_box_dn_1": 0.76803, "loss_cls_dn_2": 0.13515, "loss_box_dn_2": 0.77095, "loss_cls_dn_3": 0.1362, "loss_box_dn_3": 0.78056, "loss_cls_dn_4": 0.13887, "loss_box_dn_4": 0.79726, "loss_cls_dn_5": 0.1398, "loss_box_dn_5": 0.81973, "loss_dense_depth": 0.76267, "loss": 26.38486, "grad_norm": 54.06062, "time": 1.67473} -{"mode": "train", "epoch": 1, "iter": 222, "lr": 0.00019, "memory": 49164, "data_time": 0.1689, "loss_cls_0": 0.78048, "loss_box_0": 1.69457, "loss_cns_0": 0.62381, "loss_yns_0": 0.15088, "loss_cls_1": 0.86389, "loss_box_1": 1.61348, "loss_cns_1": 0.65743, "loss_yns_1": 0.14878, "loss_cls_2": 0.88703, "loss_box_2": 1.60216, "loss_cns_2": 0.65919, "loss_yns_2": 0.14991, "loss_cls_3": 0.87945, "loss_box_3": 1.57986, "loss_cns_3": 0.66554, "loss_yns_3": 0.15156, "loss_cls_4": 0.8885, "loss_box_4": 1.56653, "loss_cns_4": 0.66184, "loss_yns_4": 0.1497, "loss_cls_5": 0.88325, "loss_box_5": 1.56606, "loss_cns_5": 0.66372, "loss_yns_5": 0.14941, "loss_cls_dn_0": 0.19195, "loss_box_dn_0": 0.75349, "loss_cls_dn_1": 0.12554, "loss_box_dn_1": 0.74647, "loss_cls_dn_2": 0.13341, "loss_box_dn_2": 0.74199, "loss_cls_dn_3": 0.13065, "loss_box_dn_3": 0.73939, "loss_cls_dn_4": 0.132, "loss_box_dn_4": 0.74291, "loss_cls_dn_5": 0.13792, "loss_box_dn_5": 0.7578, "loss_dense_depth": 0.74281, "loss": 25.71337, "grad_norm": 40.10612, "time": 1.64799} -{"mode": "train", "epoch": 1, "iter": 223, "lr": 0.00019, "memory": 49164, "data_time": 0.07533, "loss_cls_0": 0.81415, "loss_box_0": 1.70001, "loss_cns_0": 0.62177, "loss_yns_0": 0.15272, "loss_cls_1": 0.88596, "loss_box_1": 1.63724, "loss_cns_1": 0.65645, "loss_yns_1": 0.14942, "loss_cls_2": 0.88741, "loss_box_2": 1.60294, "loss_cns_2": 0.65702, "loss_yns_2": 0.15107, "loss_cls_3": 0.926, "loss_box_3": 1.57005, "loss_cns_3": 0.65604, "loss_yns_3": 0.15045, "loss_cls_4": 0.92906, "loss_box_4": 1.55985, "loss_cns_4": 0.65872, "loss_yns_4": 0.15301, "loss_cls_5": 0.91573, "loss_box_5": 1.57943, "loss_cns_5": 0.65718, "loss_yns_5": 0.15171, "loss_cls_dn_0": 0.20483, "loss_box_dn_0": 0.75059, "loss_cls_dn_1": 0.13294, "loss_box_dn_1": 0.75047, "loss_cls_dn_2": 0.13784, "loss_box_dn_2": 0.73558, "loss_cls_dn_3": 0.13745, "loss_box_dn_3": 0.72341, "loss_cls_dn_4": 0.14214, "loss_box_dn_4": 0.7181, "loss_cls_dn_5": 0.14497, "loss_box_dn_5": 0.72943, "loss_dense_depth": 0.76108, "loss": 25.89221, "grad_norm": 46.07729, "time": 1.56319} -{"mode": "train", "epoch": 1, "iter": 224, "lr": 0.00019, "memory": 49164, "data_time": 0.07209, "loss_cls_0": 0.78388, "loss_box_0": 1.67379, "loss_cns_0": 0.62293, "loss_yns_0": 0.14972, "loss_cls_1": 0.86154, "loss_box_1": 1.66713, "loss_cns_1": 0.6576, "loss_yns_1": 0.14944, "loss_cls_2": 0.87096, "loss_box_2": 1.60608, "loss_cns_2": 0.6561, "loss_yns_2": 0.15048, "loss_cls_3": 0.88972, "loss_box_3": 1.58322, "loss_cns_3": 0.656, "loss_yns_3": 0.15306, "loss_cls_4": 0.89041, "loss_box_4": 1.58302, "loss_cns_4": 0.65993, "loss_yns_4": 0.1529, "loss_cls_5": 0.8726, "loss_box_5": 1.5952, "loss_cns_5": 0.66277, "loss_yns_5": 0.14978, "loss_cls_dn_0": 0.19806, "loss_box_dn_0": 0.75155, "loss_cls_dn_1": 0.1293, "loss_box_dn_1": 0.71749, "loss_cls_dn_2": 0.12871, "loss_box_dn_2": 0.69868, "loss_cls_dn_3": 0.12904, "loss_box_dn_3": 0.69661, "loss_cls_dn_4": 0.13333, "loss_box_dn_4": 0.69956, "loss_cls_dn_5": 0.13607, "loss_box_dn_5": 0.7093, "loss_dense_depth": 0.73192, "loss": 25.55787, "grad_norm": 37.25158, "time": 1.56255} -{"mode": "train", "epoch": 1, "iter": 225, "lr": 0.00019, "memory": 49164, "data_time": 0.0707, "loss_cls_0": 0.77155, "loss_box_0": 1.71026, "loss_cns_0": 0.62308, "loss_yns_0": 0.15218, "loss_cls_1": 0.87491, "loss_box_1": 1.69286, "loss_cns_1": 0.65548, "loss_yns_1": 0.15118, "loss_cls_2": 0.8761, "loss_box_2": 1.65749, "loss_cns_2": 0.65817, "loss_yns_2": 0.15479, "loss_cls_3": 0.90801, "loss_box_3": 1.65999, "loss_cns_3": 0.65924, "loss_yns_3": 0.1531, "loss_cls_4": 0.90785, "loss_box_4": 1.67752, "loss_cns_4": 0.66139, "loss_yns_4": 0.15305, "loss_cls_5": 0.9506, "loss_box_5": 1.69125, "loss_cns_5": 0.66537, "loss_yns_5": 0.15525, "loss_cls_dn_0": 0.19743, "loss_box_dn_0": 0.74998, "loss_cls_dn_1": 0.12758, "loss_box_dn_1": 0.72781, "loss_cls_dn_2": 0.1272, "loss_box_dn_2": 0.72996, "loss_cls_dn_3": 0.12902, "loss_box_dn_3": 0.7516, "loss_cls_dn_4": 0.12965, "loss_box_dn_4": 0.77579, "loss_cls_dn_5": 0.13563, "loss_box_dn_5": 0.79383, "loss_dense_depth": 0.72572, "loss": 26.32187, "grad_norm": 54.14425, "time": 1.63551} -{"mode": "train", "epoch": 1, "iter": 226, "lr": 0.00019, "memory": 49164, "data_time": 0.07987, "loss_cls_0": 0.77948, "loss_box_0": 1.67586, "loss_cns_0": 0.62837, "loss_yns_0": 0.15168, "loss_cls_1": 0.90088, "loss_box_1": 1.64355, "loss_cns_1": 0.65915, "loss_yns_1": 0.15244, "loss_cls_2": 0.90624, "loss_box_2": 1.64048, "loss_cns_2": 0.65936, "loss_yns_2": 0.15231, "loss_cls_3": 0.93468, "loss_box_3": 1.64677, "loss_cns_3": 0.66131, "loss_yns_3": 0.15073, "loss_cls_4": 0.91982, "loss_box_4": 1.6539, "loss_cns_4": 0.6593, "loss_yns_4": 0.15105, "loss_cls_5": 0.91026, "loss_box_5": 1.6532, "loss_cns_5": 0.65941, "loss_yns_5": 0.15366, "loss_cls_dn_0": 0.195, "loss_box_dn_0": 0.75153, "loss_cls_dn_1": 0.13331, "loss_box_dn_1": 0.77886, "loss_cls_dn_2": 0.13023, "loss_box_dn_2": 0.79541, "loss_cls_dn_3": 0.13267, "loss_box_dn_3": 0.8161, "loss_cls_dn_4": 0.13248, "loss_box_dn_4": 0.83926, "loss_cls_dn_5": 0.13603, "loss_box_dn_5": 0.86017, "loss_dense_depth": 0.7233, "loss": 26.52825, "grad_norm": 49.20581, "time": 1.59498} -{"mode": "train", "epoch": 1, "iter": 227, "lr": 0.00019, "memory": 49164, "data_time": 0.07369, "loss_cls_0": 0.77051, "loss_box_0": 1.68586, "loss_cns_0": 0.61884, "loss_yns_0": 0.14691, "loss_cls_1": 0.86954, "loss_box_1": 1.66061, "loss_cns_1": 0.65183, "loss_yns_1": 0.14663, "loss_cls_2": 0.90883, "loss_box_2": 1.67212, "loss_cns_2": 0.65285, "loss_yns_2": 0.14748, "loss_cls_3": 0.89297, "loss_box_3": 1.64707, "loss_cns_3": 0.65739, "loss_yns_3": 0.14764, "loss_cls_4": 0.89448, "loss_box_4": 1.65329, "loss_cns_4": 0.65472, "loss_yns_4": 0.14888, "loss_cls_5": 0.91807, "loss_box_5": 1.65524, "loss_cns_5": 0.65202, "loss_yns_5": 0.147, "loss_cls_dn_0": 0.1951, "loss_box_dn_0": 0.75857, "loss_cls_dn_1": 0.1254, "loss_box_dn_1": 0.78893, "loss_cls_dn_2": 0.13012, "loss_box_dn_2": 0.8135, "loss_cls_dn_3": 0.13488, "loss_box_dn_3": 0.81311, "loss_cls_dn_4": 0.13015, "loss_box_dn_4": 0.83083, "loss_cls_dn_5": 0.13709, "loss_box_dn_5": 0.8539, "loss_dense_depth": 0.73162, "loss": 26.44399, "grad_norm": 51.46517, "time": 1.60086} -{"mode": "train", "epoch": 1, "iter": 228, "lr": 0.00019, "memory": 49164, "data_time": 0.07343, "loss_cls_0": 0.7752, "loss_box_0": 1.69765, "loss_cns_0": 0.62634, "loss_yns_0": 0.14874, "loss_cls_1": 0.903, "loss_box_1": 1.63279, "loss_cns_1": 0.64951, "loss_yns_1": 0.14722, "loss_cls_2": 0.91498, "loss_box_2": 1.63832, "loss_cns_2": 0.6535, "loss_yns_2": 0.14865, "loss_cls_3": 0.89133, "loss_box_3": 1.61241, "loss_cns_3": 0.66017, "loss_yns_3": 0.14916, "loss_cls_4": 0.89171, "loss_box_4": 1.61834, "loss_cns_4": 0.65725, "loss_yns_4": 0.15051, "loss_cls_5": 0.93378, "loss_box_5": 1.61934, "loss_cns_5": 0.65153, "loss_yns_5": 0.14853, "loss_cls_dn_0": 0.19665, "loss_box_dn_0": 0.74843, "loss_cls_dn_1": 0.12687, "loss_box_dn_1": 0.81383, "loss_cls_dn_2": 0.12928, "loss_box_dn_2": 0.82736, "loss_cls_dn_3": 0.13323, "loss_box_dn_3": 0.82202, "loss_cls_dn_4": 0.1307, "loss_box_dn_4": 0.83653, "loss_cls_dn_5": 0.13797, "loss_box_dn_5": 0.86048, "loss_dense_depth": 0.73577, "loss": 26.41907, "grad_norm": 44.94754, "time": 1.55392} -{"mode": "train", "epoch": 1, "iter": 229, "lr": 0.00019, "memory": 49164, "data_time": 0.07354, "loss_cls_0": 0.80958, "loss_box_0": 1.7398, "loss_cns_0": 0.6244, "loss_yns_0": 0.1485, "loss_cls_1": 0.86681, "loss_box_1": 1.67332, "loss_cns_1": 0.65017, "loss_yns_1": 0.14607, "loss_cls_2": 0.87032, "loss_box_2": 1.64042, "loss_cns_2": 0.65897, "loss_yns_2": 0.14762, "loss_cls_3": 0.89625, "loss_box_3": 1.62192, "loss_cns_3": 0.66238, "loss_yns_3": 0.14769, "loss_cls_4": 0.88419, "loss_box_4": 1.62871, "loss_cns_4": 0.66095, "loss_yns_4": 0.15002, "loss_cls_5": 0.88229, "loss_box_5": 1.62966, "loss_cns_5": 0.66041, "loss_yns_5": 0.14901, "loss_cls_dn_0": 0.18829, "loss_box_dn_0": 0.7475, "loss_cls_dn_1": 0.12844, "loss_box_dn_1": 0.78906, "loss_cls_dn_2": 0.1272, "loss_box_dn_2": 0.77959, "loss_cls_dn_3": 0.13045, "loss_box_dn_3": 0.77277, "loss_cls_dn_4": 0.13523, "loss_box_dn_4": 0.78135, "loss_cls_dn_5": 0.13742, "loss_box_dn_5": 0.78823, "loss_dense_depth": 0.73738, "loss": 26.19235, "grad_norm": 44.67694, "time": 1.55729} -{"mode": "train", "epoch": 1, "iter": 230, "lr": 0.00019, "memory": 49164, "data_time": 0.0741, "loss_cls_0": 0.78538, "loss_box_0": 1.73517, "loss_cns_0": 0.62589, "loss_yns_0": 0.14846, "loss_cls_1": 0.86686, "loss_box_1": 1.6769, "loss_cns_1": 0.65107, "loss_yns_1": 0.14809, "loss_cls_2": 0.9114, "loss_box_2": 1.62591, "loss_cns_2": 0.65881, "loss_yns_2": 0.14835, "loss_cls_3": 0.91079, "loss_box_3": 1.60362, "loss_cns_3": 0.66143, "loss_yns_3": 0.14739, "loss_cls_4": 0.89504, "loss_box_4": 1.61031, "loss_cns_4": 0.6637, "loss_yns_4": 0.14725, "loss_cls_5": 0.90394, "loss_box_5": 1.6041, "loss_cns_5": 0.66525, "loss_yns_5": 0.14994, "loss_cls_dn_0": 0.19064, "loss_box_dn_0": 0.75651, "loss_cls_dn_1": 0.12283, "loss_box_dn_1": 0.74377, "loss_cls_dn_2": 0.1206, "loss_box_dn_2": 0.72284, "loss_cls_dn_3": 0.12353, "loss_box_dn_3": 0.71698, "loss_cls_dn_4": 0.1252, "loss_box_dn_4": 0.72078, "loss_cls_dn_5": 0.12909, "loss_box_dn_5": 0.72351, "loss_dense_depth": 0.73146, "loss": 25.87279, "grad_norm": 33.59467, "time": 1.56555} -{"mode": "train", "epoch": 1, "iter": 231, "lr": 0.00019, "memory": 49164, "data_time": 0.07627, "loss_cls_0": 0.78551, "loss_box_0": 1.65103, "loss_cns_0": 0.61938, "loss_yns_0": 0.1441, "loss_cls_1": 0.83828, "loss_box_1": 1.55866, "loss_cns_1": 0.65388, "loss_yns_1": 0.14539, "loss_cls_2": 0.86989, "loss_box_2": 1.57763, "loss_cns_2": 0.66035, "loss_yns_2": 0.147, "loss_cls_3": 0.92379, "loss_box_3": 1.50877, "loss_cns_3": 0.66471, "loss_yns_3": 0.14541, "loss_cls_4": 0.91576, "loss_box_4": 1.52563, "loss_cns_4": 0.66502, "loss_yns_4": 0.14749, "loss_cls_5": 0.91649, "loss_box_5": 1.55323, "loss_cns_5": 0.67158, "loss_yns_5": 0.15045, "loss_cls_dn_0": 0.18698, "loss_box_dn_0": 0.7544, "loss_cls_dn_1": 0.11958, "loss_box_dn_1": 0.6944, "loss_cls_dn_2": 0.1221, "loss_box_dn_2": 0.6979, "loss_cls_dn_3": 0.13376, "loss_box_dn_3": 0.69814, "loss_cls_dn_4": 0.13428, "loss_box_dn_4": 0.71418, "loss_cls_dn_5": 0.14161, "loss_box_dn_5": 0.73281, "loss_dense_depth": 0.71993, "loss": 25.2895, "grad_norm": 51.74443, "time": 1.57748} -{"mode": "train", "epoch": 1, "iter": 232, "lr": 0.00019, "memory": 49164, "data_time": 0.11072, "loss_cls_0": 0.82974, "loss_box_0": 1.62669, "loss_cns_0": 0.60697, "loss_yns_0": 0.13933, "loss_cls_1": 0.85046, "loss_box_1": 1.61369, "loss_cns_1": 0.65111, "loss_yns_1": 0.1452, "loss_cls_2": 0.86765, "loss_box_2": 1.60905, "loss_cns_2": 0.65372, "loss_yns_2": 0.14572, "loss_cls_3": 0.93862, "loss_box_3": 1.49481, "loss_cns_3": 0.64011, "loss_yns_3": 0.14082, "loss_cls_4": 0.9425, "loss_box_4": 1.56519, "loss_cns_4": 0.65141, "loss_yns_4": 0.14606, "loss_cls_5": 0.89376, "loss_box_5": 1.63301, "loss_cns_5": 0.65742, "loss_yns_5": 0.14832, "loss_cls_dn_0": 0.18677, "loss_box_dn_0": 0.76826, "loss_cls_dn_1": 0.12618, "loss_box_dn_1": 0.73097, "loss_cls_dn_2": 0.12395, "loss_box_dn_2": 0.74406, "loss_cls_dn_3": 0.13111, "loss_box_dn_3": 0.75293, "loss_cls_dn_4": 0.14191, "loss_box_dn_4": 0.77826, "loss_cls_dn_5": 0.14243, "loss_box_dn_5": 0.81036, "loss_dense_depth": 0.72489, "loss": 25.75342, "grad_norm": 53.17388, "time": 1.61538} -{"mode": "train", "epoch": 1, "iter": 233, "lr": 0.00019, "memory": 49164, "data_time": 0.08734, "loss_cls_0": 0.81653, "loss_box_0": 1.63954, "loss_cns_0": 0.61683, "loss_yns_0": 0.14565, "loss_cls_1": 0.88254, "loss_box_1": 1.60843, "loss_cns_1": 0.65132, "loss_yns_1": 0.14818, "loss_cls_2": 0.87946, "loss_box_2": 1.60588, "loss_cns_2": 0.65526, "loss_yns_2": 0.14713, "loss_cls_3": 0.90194, "loss_box_3": 1.5461, "loss_cns_3": 0.65397, "loss_yns_3": 0.14757, "loss_cls_4": 0.91371, "loss_box_4": 1.56367, "loss_cns_4": 0.65756, "loss_yns_4": 0.14742, "loss_cls_5": 0.9362, "loss_box_5": 1.60172, "loss_cns_5": 0.65804, "loss_yns_5": 0.1481, "loss_cls_dn_0": 0.18468, "loss_box_dn_0": 0.75414, "loss_cls_dn_1": 0.1301, "loss_box_dn_1": 0.76309, "loss_cls_dn_2": 0.13131, "loss_box_dn_2": 0.77507, "loss_cls_dn_3": 0.13151, "loss_box_dn_3": 0.78864, "loss_cls_dn_4": 0.13509, "loss_box_dn_4": 0.80814, "loss_cls_dn_5": 0.14327, "loss_box_dn_5": 0.83795, "loss_dense_depth": 0.72673, "loss": 25.98248, "grad_norm": 48.63376, "time": 1.56985} -{"mode": "train", "epoch": 1, "iter": 234, "lr": 0.00019, "memory": 49164, "data_time": 0.07308, "loss_cls_0": 0.76892, "loss_box_0": 1.6613, "loss_cns_0": 0.62817, "loss_yns_0": 0.14825, "loss_cls_1": 0.91576, "loss_box_1": 1.62464, "loss_cns_1": 0.65397, "loss_yns_1": 0.15157, "loss_cls_2": 0.90275, "loss_box_2": 1.57847, "loss_cns_2": 0.65797, "loss_yns_2": 0.15188, "loss_cls_3": 0.92678, "loss_box_3": 1.54408, "loss_cns_3": 0.66525, "loss_yns_3": 0.1532, "loss_cls_4": 0.9303, "loss_box_4": 1.53927, "loss_cns_4": 0.66232, "loss_yns_4": 0.1511, "loss_cls_5": 0.97782, "loss_box_5": 1.55526, "loss_cns_5": 0.66576, "loss_yns_5": 0.15261, "loss_cls_dn_0": 0.17373, "loss_box_dn_0": 0.74887, "loss_cls_dn_1": 0.12132, "loss_box_dn_1": 0.752, "loss_cls_dn_2": 0.12334, "loss_box_dn_2": 0.74685, "loss_cls_dn_3": 0.1256, "loss_box_dn_3": 0.75058, "loss_cls_dn_4": 0.12718, "loss_box_dn_4": 0.76173, "loss_cls_dn_5": 0.14065, "loss_box_dn_5": 0.78246, "loss_dense_depth": 0.72743, "loss": 25.84915, "grad_norm": 39.51593, "time": 1.58478} -{"mode": "train", "epoch": 1, "iter": 235, "lr": 0.00019, "memory": 49164, "data_time": 0.07254, "loss_cls_0": 0.78365, "loss_box_0": 1.67406, "loss_cns_0": 0.63303, "loss_yns_0": 0.14987, "loss_cls_1": 0.86646, "loss_box_1": 1.59244, "loss_cns_1": 0.65735, "loss_yns_1": 0.14825, "loss_cls_2": 0.89766, "loss_box_2": 1.57577, "loss_cns_2": 0.6594, "loss_yns_2": 0.14933, "loss_cls_3": 0.90245, "loss_box_3": 1.545, "loss_cns_3": 0.65962, "loss_yns_3": 0.14959, "loss_cls_4": 0.90531, "loss_box_4": 1.549, "loss_cns_4": 0.66234, "loss_yns_4": 0.15119, "loss_cls_5": 0.92314, "loss_box_5": 1.53422, "loss_cns_5": 0.66139, "loss_yns_5": 0.14974, "loss_cls_dn_0": 0.177, "loss_box_dn_0": 0.73934, "loss_cls_dn_1": 0.12033, "loss_box_dn_1": 0.71607, "loss_cls_dn_2": 0.12376, "loss_box_dn_2": 0.71194, "loss_cls_dn_3": 0.12583, "loss_box_dn_3": 0.70663, "loss_cls_dn_4": 0.1283, "loss_box_dn_4": 0.71191, "loss_cls_dn_5": 0.13571, "loss_box_dn_5": 0.71961, "loss_dense_depth": 0.76359, "loss": 25.46028, "grad_norm": 44.61768, "time": 1.5594} -{"mode": "train", "epoch": 1, "iter": 236, "lr": 0.00019, "memory": 49164, "data_time": 0.09059, "loss_cls_0": 0.76276, "loss_box_0": 1.68835, "loss_cns_0": 0.62592, "loss_yns_0": 0.15151, "loss_cls_1": 0.90295, "loss_box_1": 1.57348, "loss_cns_1": 0.64671, "loss_yns_1": 0.14851, "loss_cls_2": 0.91538, "loss_box_2": 1.54456, "loss_cns_2": 0.64382, "loss_yns_2": 0.14754, "loss_cls_3": 0.8881, "loss_box_3": 1.53897, "loss_cns_3": 0.65448, "loss_yns_3": 0.15134, "loss_cls_4": 0.89534, "loss_box_4": 1.54145, "loss_cns_4": 0.65942, "loss_yns_4": 0.15253, "loss_cls_5": 0.93592, "loss_box_5": 1.50116, "loss_cns_5": 0.64925, "loss_yns_5": 0.1498, "loss_cls_dn_0": 0.17185, "loss_box_dn_0": 0.74064, "loss_cls_dn_1": 0.12351, "loss_box_dn_1": 0.6909, "loss_cls_dn_2": 0.12571, "loss_box_dn_2": 0.68184, "loss_cls_dn_3": 0.12667, "loss_box_dn_3": 0.67132, "loss_cls_dn_4": 0.1281, "loss_box_dn_4": 0.67472, "loss_cls_dn_5": 0.14373, "loss_box_dn_5": 0.68373, "loss_dense_depth": 0.7398, "loss": 25.17172, "grad_norm": 42.4786, "time": 1.64052} -{"mode": "train", "epoch": 1, "iter": 237, "lr": 0.00019, "memory": 49164, "data_time": 0.11768, "loss_cls_0": 0.77263, "loss_box_0": 1.68036, "loss_cns_0": 0.62892, "loss_yns_0": 0.15056, "loss_cls_1": 0.87678, "loss_box_1": 1.55042, "loss_cns_1": 0.65416, "loss_yns_1": 0.15053, "loss_cls_2": 0.87558, "loss_box_2": 1.52363, "loss_cns_2": 0.65476, "loss_yns_2": 0.15043, "loss_cls_3": 0.87302, "loss_box_3": 1.52442, "loss_cns_3": 0.66421, "loss_yns_3": 0.15131, "loss_cls_4": 0.87805, "loss_box_4": 1.49698, "loss_cns_4": 0.66513, "loss_yns_4": 0.15199, "loss_cls_5": 0.91658, "loss_box_5": 1.49414, "loss_cns_5": 0.65899, "loss_yns_5": 0.15054, "loss_cls_dn_0": 0.17823, "loss_box_dn_0": 0.75841, "loss_cls_dn_1": 0.12162, "loss_box_dn_1": 0.68734, "loss_cls_dn_2": 0.12312, "loss_box_dn_2": 0.67898, "loss_cls_dn_3": 0.12272, "loss_box_dn_3": 0.68037, "loss_cls_dn_4": 0.12201, "loss_box_dn_4": 0.67866, "loss_cls_dn_5": 0.12851, "loss_box_dn_5": 0.69421, "loss_dense_depth": 0.7002, "loss": 24.96848, "grad_norm": 36.48956, "time": 1.61079} -{"mode": "train", "epoch": 1, "iter": 238, "lr": 0.00019, "memory": 49164, "data_time": 0.07558, "loss_cls_0": 0.7704, "loss_box_0": 1.64344, "loss_cns_0": 0.63399, "loss_yns_0": 0.15177, "loss_cls_1": 0.87607, "loss_box_1": 1.55698, "loss_cns_1": 0.66095, "loss_yns_1": 0.15229, "loss_cls_2": 0.88736, "loss_box_2": 1.52376, "loss_cns_2": 0.66272, "loss_yns_2": 0.15188, "loss_cls_3": 0.92675, "loss_box_3": 1.52403, "loss_cns_3": 0.67042, "loss_yns_3": 0.15263, "loss_cls_4": 0.90959, "loss_box_4": 1.51157, "loss_cns_4": 0.66555, "loss_yns_4": 0.15289, "loss_cls_5": 0.89422, "loss_box_5": 1.51595, "loss_cns_5": 0.66357, "loss_yns_5": 0.15275, "loss_cls_dn_0": 0.17403, "loss_box_dn_0": 0.74761, "loss_cls_dn_1": 0.11977, "loss_box_dn_1": 0.70185, "loss_cls_dn_2": 0.12332, "loss_box_dn_2": 0.70098, "loss_cls_dn_3": 0.1285, "loss_box_dn_3": 0.71199, "loss_cls_dn_4": 0.12757, "loss_box_dn_4": 0.71621, "loss_cls_dn_5": 0.12971, "loss_box_dn_5": 0.73055, "loss_dense_depth": 0.66669, "loss": 25.1903, "grad_norm": 39.14441, "time": 1.56383} -{"mode": "train", "epoch": 1, "iter": 239, "lr": 0.0002, "memory": 49164, "data_time": 0.0833, "loss_cls_0": 0.76357, "loss_box_0": 1.59745, "loss_cns_0": 0.62796, "loss_yns_0": 0.15042, "loss_cls_1": 0.87796, "loss_box_1": 1.51057, "loss_cns_1": 0.66369, "loss_yns_1": 0.15205, "loss_cls_2": 0.90939, "loss_box_2": 1.48378, "loss_cns_2": 0.66413, "loss_yns_2": 0.15199, "loss_cls_3": 0.89307, "loss_box_3": 1.48603, "loss_cns_3": 0.66802, "loss_yns_3": 0.15243, "loss_cls_4": 0.88389, "loss_box_4": 1.51303, "loss_cns_4": 0.66206, "loss_yns_4": 0.15103, "loss_cls_5": 0.8899, "loss_box_5": 1.52079, "loss_cns_5": 0.66331, "loss_yns_5": 0.15188, "loss_cls_dn_0": 0.17348, "loss_box_dn_0": 0.74875, "loss_cls_dn_1": 0.12317, "loss_box_dn_1": 0.71045, "loss_cls_dn_2": 0.12444, "loss_box_dn_2": 0.71171, "loss_cls_dn_3": 0.12608, "loss_box_dn_3": 0.73036, "loss_cls_dn_4": 0.12879, "loss_box_dn_4": 0.748, "loss_cls_dn_5": 0.13433, "loss_box_dn_5": 0.76639, "loss_dense_depth": 0.70129, "loss": 25.11563, "grad_norm": 37.95193, "time": 1.57256} -{"mode": "train", "epoch": 1, "iter": 240, "lr": 0.0002, "memory": 49164, "data_time": 0.07885, "loss_cls_0": 0.77505, "loss_box_0": 1.6013, "loss_cns_0": 0.61968, "loss_yns_0": 0.14815, "loss_cls_1": 0.88494, "loss_box_1": 1.50187, "loss_cns_1": 0.66627, "loss_yns_1": 0.15096, "loss_cls_2": 0.89351, "loss_box_2": 1.48268, "loss_cns_2": 0.66785, "loss_yns_2": 0.15108, "loss_cls_3": 0.90465, "loss_box_3": 1.47989, "loss_cns_3": 0.66653, "loss_yns_3": 0.15184, "loss_cls_4": 0.90112, "loss_box_4": 1.49395, "loss_cns_4": 0.66562, "loss_yns_4": 0.14935, "loss_cls_5": 0.89588, "loss_box_5": 1.50488, "loss_cns_5": 0.66451, "loss_yns_5": 0.15059, "loss_cls_dn_0": 0.17816, "loss_box_dn_0": 0.75637, "loss_cls_dn_1": 0.12414, "loss_box_dn_1": 0.73723, "loss_cls_dn_2": 0.12294, "loss_box_dn_2": 0.74321, "loss_cls_dn_3": 0.12859, "loss_box_dn_3": 0.76284, "loss_cls_dn_4": 0.13263, "loss_box_dn_4": 0.78346, "loss_cls_dn_5": 0.13431, "loss_box_dn_5": 0.80358, "loss_dense_depth": 0.71977, "loss": 25.29936, "grad_norm": 43.18041, "time": 1.56655} -{"mode": "train", "epoch": 1, "iter": 241, "lr": 0.0002, "memory": 49164, "data_time": 0.10876, "loss_cls_0": 0.79709, "loss_box_0": 1.58786, "loss_cns_0": 0.61353, "loss_yns_0": 0.15026, "loss_cls_1": 0.90208, "loss_box_1": 1.55312, "loss_cns_1": 0.66312, "loss_yns_1": 0.15452, "loss_cls_2": 0.91001, "loss_box_2": 1.5269, "loss_cns_2": 0.66358, "loss_yns_2": 0.15301, "loss_cls_3": 0.94282, "loss_box_3": 1.52224, "loss_cns_3": 0.66489, "loss_yns_3": 0.15305, "loss_cls_4": 0.92379, "loss_box_4": 1.50952, "loss_cns_4": 0.66315, "loss_yns_4": 0.15176, "loss_cls_5": 0.91505, "loss_box_5": 1.52572, "loss_cns_5": 0.66074, "loss_yns_5": 0.1519, "loss_cls_dn_0": 0.18318, "loss_box_dn_0": 0.74709, "loss_cls_dn_1": 0.12562, "loss_box_dn_1": 0.7557, "loss_cls_dn_2": 0.12563, "loss_box_dn_2": 0.75701, "loss_cls_dn_3": 0.13461, "loss_box_dn_3": 0.76509, "loss_cls_dn_4": 0.13191, "loss_box_dn_4": 0.77404, "loss_cls_dn_5": 0.13373, "loss_box_dn_5": 0.7899, "loss_dense_depth": 0.74704, "loss": 25.63028, "grad_norm": 38.35988, "time": 1.66709} -{"mode": "train", "epoch": 1, "iter": 242, "lr": 0.0002, "memory": 49164, "data_time": 0.17087, "loss_cls_0": 0.77366, "loss_box_0": 1.60141, "loss_cns_0": 0.62359, "loss_yns_0": 0.15187, "loss_cls_1": 0.89039, "loss_box_1": 1.57176, "loss_cns_1": 0.66248, "loss_yns_1": 0.15442, "loss_cls_2": 0.91133, "loss_box_2": 1.54859, "loss_cns_2": 0.66345, "loss_yns_2": 0.15326, "loss_cls_3": 0.93697, "loss_box_3": 1.51637, "loss_cns_3": 0.66923, "loss_yns_3": 0.15373, "loss_cls_4": 0.90517, "loss_box_4": 1.50568, "loss_cns_4": 0.66502, "loss_yns_4": 0.15289, "loss_cls_5": 0.92071, "loss_box_5": 1.52101, "loss_cns_5": 0.66324, "loss_yns_5": 0.15294, "loss_cls_dn_0": 0.17333, "loss_box_dn_0": 0.73936, "loss_cls_dn_1": 0.12582, "loss_box_dn_1": 0.74105, "loss_cls_dn_2": 0.12571, "loss_box_dn_2": 0.74079, "loss_cls_dn_3": 0.13219, "loss_box_dn_3": 0.72741, "loss_cls_dn_4": 0.12832, "loss_box_dn_4": 0.73185, "loss_cls_dn_5": 0.13423, "loss_box_dn_5": 0.73519, "loss_dense_depth": 0.69062, "loss": 25.39506, "grad_norm": 39.87461, "time": 1.64634} -{"mode": "train", "epoch": 1, "iter": 243, "lr": 0.0002, "memory": 49164, "data_time": 0.07064, "loss_cls_0": 0.76774, "loss_box_0": 1.62871, "loss_cns_0": 0.62874, "loss_yns_0": 0.15395, "loss_cls_1": 0.87105, "loss_box_1": 1.58409, "loss_cns_1": 0.65877, "loss_yns_1": 0.15254, "loss_cls_2": 0.9029, "loss_box_2": 1.53863, "loss_cns_2": 0.66482, "loss_yns_2": 0.15167, "loss_cls_3": 0.90494, "loss_box_3": 1.51978, "loss_cns_3": 0.66675, "loss_yns_3": 0.15299, "loss_cls_4": 0.89119, "loss_box_4": 1.54039, "loss_cns_4": 0.66282, "loss_yns_4": 0.15238, "loss_cls_5": 0.90567, "loss_box_5": 1.53274, "loss_cns_5": 0.66252, "loss_yns_5": 0.15342, "loss_cls_dn_0": 0.17494, "loss_box_dn_0": 0.741, "loss_cls_dn_1": 0.12497, "loss_box_dn_1": 0.69086, "loss_cls_dn_2": 0.12496, "loss_box_dn_2": 0.6868, "loss_cls_dn_3": 0.13065, "loss_box_dn_3": 0.67796, "loss_cls_dn_4": 0.13102, "loss_box_dn_4": 0.69119, "loss_cls_dn_5": 0.1385, "loss_box_dn_5": 0.69154, "loss_dense_depth": 0.69846, "loss": 25.15202, "grad_norm": 41.51031, "time": 1.59283} -{"mode": "train", "epoch": 1, "iter": 244, "lr": 0.0002, "memory": 49164, "data_time": 0.12235, "loss_cls_0": 0.77854, "loss_box_0": 1.65838, "loss_cns_0": 0.62853, "loss_yns_0": 0.15472, "loss_cls_1": 0.86272, "loss_box_1": 1.58446, "loss_cns_1": 0.6575, "loss_yns_1": 0.15444, "loss_cls_2": 0.89254, "loss_box_2": 1.55401, "loss_cns_2": 0.66156, "loss_yns_2": 0.15461, "loss_cls_3": 0.89112, "loss_box_3": 1.5622, "loss_cns_3": 0.66489, "loss_yns_3": 0.15477, "loss_cls_4": 0.8889, "loss_box_4": 1.55626, "loss_cns_4": 0.66147, "loss_yns_4": 0.15483, "loss_cls_5": 0.89456, "loss_box_5": 1.56273, "loss_cns_5": 0.66253, "loss_yns_5": 0.15671, "loss_cls_dn_0": 0.17936, "loss_box_dn_0": 0.74581, "loss_cls_dn_1": 0.12485, "loss_box_dn_1": 0.68958, "loss_cls_dn_2": 0.12393, "loss_box_dn_2": 0.68121, "loss_cls_dn_3": 0.12774, "loss_box_dn_3": 0.69081, "loss_cls_dn_4": 0.13237, "loss_box_dn_4": 0.6996, "loss_cls_dn_5": 0.13518, "loss_box_dn_5": 0.71606, "loss_dense_depth": 0.77216, "loss": 25.3716, "grad_norm": 34.4308, "time": 1.61502} -{"mode": "train", "epoch": 1, "iter": 245, "lr": 0.0002, "memory": 49164, "data_time": 0.09533, "loss_cls_0": 0.74801, "loss_box_0": 1.64296, "loss_cns_0": 0.62559, "loss_yns_0": 0.15237, "loss_cls_1": 0.84849, "loss_box_1": 1.5755, "loss_cns_1": 0.64734, "loss_yns_1": 0.1523, "loss_cls_2": 0.85709, "loss_box_2": 1.53375, "loss_cns_2": 0.65732, "loss_yns_2": 0.15198, "loss_cls_3": 0.87595, "loss_box_3": 1.53538, "loss_cns_3": 0.65621, "loss_yns_3": 0.15305, "loss_cls_4": 0.86696, "loss_box_4": 1.51842, "loss_cns_4": 0.65729, "loss_yns_4": 0.15272, "loss_cls_5": 0.88663, "loss_box_5": 1.52568, "loss_cns_5": 0.65685, "loss_yns_5": 0.15234, "loss_cls_dn_0": 0.17427, "loss_box_dn_0": 0.73491, "loss_cls_dn_1": 0.12799, "loss_box_dn_1": 0.69453, "loss_cls_dn_2": 0.12901, "loss_box_dn_2": 0.68787, "loss_cls_dn_3": 0.13537, "loss_box_dn_3": 0.70405, "loss_cls_dn_4": 0.13429, "loss_box_dn_4": 0.71367, "loss_cls_dn_5": 0.13948, "loss_box_dn_5": 0.7391, "loss_dense_depth": 0.70928, "loss": 25.05395, "grad_norm": 41.0922, "time": 1.61584} -{"mode": "train", "epoch": 1, "iter": 246, "lr": 0.0002, "memory": 49164, "data_time": 0.06877, "loss_cls_0": 0.74523, "loss_box_0": 1.65843, "loss_cns_0": 0.62571, "loss_yns_0": 0.15221, "loss_cls_1": 0.81631, "loss_box_1": 1.60695, "loss_cns_1": 0.64617, "loss_yns_1": 0.15109, "loss_cls_2": 0.83745, "loss_box_2": 1.55072, "loss_cns_2": 0.65662, "loss_yns_2": 0.15167, "loss_cls_3": 0.85538, "loss_box_3": 1.53876, "loss_cns_3": 0.65965, "loss_yns_3": 0.15227, "loss_cls_4": 0.85065, "loss_box_4": 1.51874, "loss_cns_4": 0.65903, "loss_yns_4": 0.15128, "loss_cls_5": 0.89252, "loss_box_5": 1.49954, "loss_cns_5": 0.65284, "loss_yns_5": 0.15124, "loss_cls_dn_0": 0.16898, "loss_box_dn_0": 0.74451, "loss_cls_dn_1": 0.12559, "loss_box_dn_1": 0.72078, "loss_cls_dn_2": 0.1276, "loss_box_dn_2": 0.71297, "loss_cls_dn_3": 0.13224, "loss_box_dn_3": 0.72631, "loss_cls_dn_4": 0.13446, "loss_box_dn_4": 0.73388, "loss_cls_dn_5": 0.14395, "loss_box_dn_5": 0.75197, "loss_dense_depth": 0.71766, "loss": 25.12134, "grad_norm": 39.37664, "time": 1.69254} -{"mode": "train", "epoch": 1, "iter": 247, "lr": 0.0002, "memory": 49164, "data_time": 0.07231, "loss_cls_0": 0.78133, "loss_box_0": 1.67424, "loss_cns_0": 0.62434, "loss_yns_0": 0.15575, "loss_cls_1": 0.82558, "loss_box_1": 1.57137, "loss_cns_1": 0.64441, "loss_yns_1": 0.15342, "loss_cls_2": 0.85905, "loss_box_2": 1.53608, "loss_cns_2": 0.65681, "loss_yns_2": 0.15436, "loss_cls_3": 0.86611, "loss_box_3": 1.51181, "loss_cns_3": 0.66599, "loss_yns_3": 0.15516, "loss_cls_4": 0.86948, "loss_box_4": 1.50921, "loss_cns_4": 0.66461, "loss_yns_4": 0.15375, "loss_cls_5": 0.88868, "loss_box_5": 1.50704, "loss_cns_5": 0.65997, "loss_yns_5": 0.15423, "loss_cls_dn_0": 0.16968, "loss_box_dn_0": 0.75287, "loss_cls_dn_1": 0.12625, "loss_box_dn_1": 0.72846, "loss_cls_dn_2": 0.12688, "loss_box_dn_2": 0.72218, "loss_cls_dn_3": 0.12893, "loss_box_dn_3": 0.72835, "loss_cls_dn_4": 0.13411, "loss_box_dn_4": 0.74, "loss_cls_dn_5": 0.13974, "loss_box_dn_5": 0.75363, "loss_dense_depth": 0.73361, "loss": 25.22747, "grad_norm": 31.10679, "time": 1.56808} -{"mode": "train", "epoch": 1, "iter": 248, "lr": 0.0002, "memory": 49164, "data_time": 0.07615, "loss_cls_0": 0.76684, "loss_box_0": 1.65676, "loss_cns_0": 0.62624, "loss_yns_0": 0.1517, "loss_cls_1": 0.83069, "loss_box_1": 1.58887, "loss_cns_1": 0.65041, "loss_yns_1": 0.15116, "loss_cls_2": 0.85332, "loss_box_2": 1.54155, "loss_cns_2": 0.65815, "loss_yns_2": 0.15221, "loss_cls_3": 0.85665, "loss_box_3": 1.53543, "loss_cns_3": 0.66525, "loss_yns_3": 0.15186, "loss_cls_4": 0.8683, "loss_box_4": 1.52423, "loss_cns_4": 0.66749, "loss_yns_4": 0.15197, "loss_cls_5": 0.88224, "loss_box_5": 1.53884, "loss_cns_5": 0.66834, "loss_yns_5": 0.15433, "loss_cls_dn_0": 0.17105, "loss_box_dn_0": 0.73979, "loss_cls_dn_1": 0.12579, "loss_box_dn_1": 0.71792, "loss_cls_dn_2": 0.12562, "loss_box_dn_2": 0.71308, "loss_cls_dn_3": 0.12742, "loss_box_dn_3": 0.72173, "loss_cls_dn_4": 0.13004, "loss_box_dn_4": 0.73145, "loss_cls_dn_5": 0.13453, "loss_box_dn_5": 0.74375, "loss_dense_depth": 0.72476, "loss": 25.19974, "grad_norm": 41.81688, "time": 1.56101} -{"mode": "train", "epoch": 1, "iter": 249, "lr": 0.0002, "memory": 49164, "data_time": 0.09674, "loss_cls_0": 0.77095, "loss_box_0": 1.66613, "loss_cns_0": 0.63128, "loss_yns_0": 0.15482, "loss_cls_1": 0.8451, "loss_box_1": 1.55128, "loss_cns_1": 0.66405, "loss_yns_1": 0.15395, "loss_cls_2": 0.85481, "loss_box_2": 1.53488, "loss_cns_2": 0.66595, "loss_yns_2": 0.15186, "loss_cls_3": 0.86761, "loss_box_3": 1.54045, "loss_cns_3": 0.66579, "loss_yns_3": 0.15288, "loss_cls_4": 0.87381, "loss_box_4": 1.52959, "loss_cns_4": 0.66781, "loss_yns_4": 0.15233, "loss_cls_5": 0.87221, "loss_box_5": 1.53096, "loss_cns_5": 0.66746, "loss_yns_5": 0.1556, "loss_cls_dn_0": 0.17324, "loss_box_dn_0": 0.74995, "loss_cls_dn_1": 0.12304, "loss_box_dn_1": 0.70299, "loss_cls_dn_2": 0.12169, "loss_box_dn_2": 0.69664, "loss_cls_dn_3": 0.1218, "loss_box_dn_3": 0.70482, "loss_cls_dn_4": 0.12445, "loss_box_dn_4": 0.70855, "loss_cls_dn_5": 0.12962, "loss_box_dn_5": 0.71336, "loss_dense_depth": 0.69173, "loss": 25.08344, "grad_norm": 33.86974, "time": 1.59331} -{"mode": "train", "epoch": 1, "iter": 250, "lr": 0.0002, "memory": 49164, "data_time": 0.07331, "loss_cls_0": 0.73474, "loss_box_0": 1.66511, "loss_cns_0": 0.63036, "loss_yns_0": 0.15221, "loss_cls_1": 0.83169, "loss_box_1": 1.54279, "loss_cns_1": 0.6629, "loss_yns_1": 0.15231, "loss_cls_2": 0.84147, "loss_box_2": 1.53304, "loss_cns_2": 0.66174, "loss_yns_2": 0.15212, "loss_cls_3": 0.83695, "loss_box_3": 1.51015, "loss_cns_3": 0.66196, "loss_yns_3": 0.15233, "loss_cls_4": 0.83801, "loss_box_4": 1.53106, "loss_cns_4": 0.66654, "loss_yns_4": 0.14976, "loss_cls_5": 0.84134, "loss_box_5": 1.53623, "loss_cns_5": 0.66528, "loss_yns_5": 0.15223, "loss_cls_dn_0": 0.16853, "loss_box_dn_0": 0.73975, "loss_cls_dn_1": 0.12234, "loss_box_dn_1": 0.68719, "loss_cls_dn_2": 0.12158, "loss_box_dn_2": 0.68998, "loss_cls_dn_3": 0.12097, "loss_box_dn_3": 0.69051, "loss_cls_dn_4": 0.12451, "loss_box_dn_4": 0.70431, "loss_cls_dn_5": 0.12801, "loss_box_dn_5": 0.7174, "loss_dense_depth": 0.68761, "loss": 24.805, "grad_norm": 40.83947, "time": 1.55882} -{"mode": "train", "epoch": 1, "iter": 251, "lr": 0.0002, "memory": 49164, "data_time": 0.07207, "loss_cls_0": 0.75489, "loss_box_0": 1.63978, "loss_cns_0": 0.63042, "loss_yns_0": 0.14827, "loss_cls_1": 0.84551, "loss_box_1": 1.54899, "loss_cns_1": 0.66018, "loss_yns_1": 0.14976, "loss_cls_2": 0.8615, "loss_box_2": 1.51674, "loss_cns_2": 0.66157, "loss_yns_2": 0.15028, "loss_cls_3": 0.86144, "loss_box_3": 1.51349, "loss_cns_3": 0.66212, "loss_yns_3": 0.14923, "loss_cls_4": 0.86782, "loss_box_4": 1.5249, "loss_cns_4": 0.66834, "loss_yns_4": 0.14894, "loss_cls_5": 0.86198, "loss_box_5": 1.53244, "loss_cns_5": 0.66473, "loss_yns_5": 0.14998, "loss_cls_dn_0": 0.16981, "loss_box_dn_0": 0.74502, "loss_cls_dn_1": 0.11851, "loss_box_dn_1": 0.70799, "loss_cls_dn_2": 0.12032, "loss_box_dn_2": 0.7035, "loss_cls_dn_3": 0.12202, "loss_box_dn_3": 0.71377, "loss_cls_dn_4": 0.13282, "loss_box_dn_4": 0.72454, "loss_cls_dn_5": 0.13174, "loss_box_dn_5": 0.7421, "loss_dense_depth": 0.72034, "loss": 25.02583, "grad_norm": 43.2766, "time": 1.6123} -{"mode": "train", "epoch": 1, "iter": 252, "lr": 0.0002, "memory": 49164, "data_time": 0.07115, "loss_cls_0": 0.73366, "loss_box_0": 1.6749, "loss_cns_0": 0.62685, "loss_yns_0": 0.15023, "loss_cls_1": 0.83273, "loss_box_1": 1.54148, "loss_cns_1": 0.66132, "loss_yns_1": 0.15011, "loss_cls_2": 0.83964, "loss_box_2": 1.52324, "loss_cns_2": 0.66358, "loss_yns_2": 0.15015, "loss_cls_3": 0.85335, "loss_box_3": 1.52515, "loss_cns_3": 0.66687, "loss_yns_3": 0.15015, "loss_cls_4": 0.8471, "loss_box_4": 1.51478, "loss_cns_4": 0.67116, "loss_yns_4": 0.14848, "loss_cls_5": 0.84711, "loss_box_5": 1.52914, "loss_cns_5": 0.6662, "loss_yns_5": 0.1498, "loss_cls_dn_0": 0.16527, "loss_box_dn_0": 0.74162, "loss_cls_dn_1": 0.11536, "loss_box_dn_1": 0.71416, "loss_cls_dn_2": 0.11733, "loss_box_dn_2": 0.70817, "loss_cls_dn_3": 0.11643, "loss_box_dn_3": 0.71584, "loss_cls_dn_4": 0.12351, "loss_box_dn_4": 0.72203, "loss_cls_dn_5": 0.12451, "loss_box_dn_5": 0.74386, "loss_dense_depth": 0.69114, "loss": 24.91642, "grad_norm": 34.36688, "time": 1.5925} -{"mode": "train", "epoch": 1, "iter": 253, "lr": 0.0002, "memory": 49164, "data_time": 0.07106, "loss_cls_0": 0.75013, "loss_box_0": 1.67313, "loss_cns_0": 0.62848, "loss_yns_0": 0.15244, "loss_cls_1": 0.84671, "loss_box_1": 1.52229, "loss_cns_1": 0.66373, "loss_yns_1": 0.1475, "loss_cls_2": 0.84179, "loss_box_2": 1.50627, "loss_cns_2": 0.66344, "loss_yns_2": 0.14714, "loss_cls_3": 0.8444, "loss_box_3": 1.48616, "loss_cns_3": 0.66608, "loss_yns_3": 0.14716, "loss_cls_4": 0.84492, "loss_box_4": 1.49619, "loss_cns_4": 0.66618, "loss_yns_4": 0.14565, "loss_cls_5": 0.85945, "loss_box_5": 1.50414, "loss_cns_5": 0.66794, "loss_yns_5": 0.14722, "loss_cls_dn_0": 0.16793, "loss_box_dn_0": 0.747, "loss_cls_dn_1": 0.1206, "loss_box_dn_1": 0.72413, "loss_cls_dn_2": 0.11918, "loss_box_dn_2": 0.72839, "loss_cls_dn_3": 0.12073, "loss_box_dn_3": 0.72761, "loss_cls_dn_4": 0.12321, "loss_box_dn_4": 0.74152, "loss_cls_dn_5": 0.13217, "loss_box_dn_5": 0.75967, "loss_dense_depth": 0.7671, "loss": 24.99777, "grad_norm": 45.07278, "time": 1.57763} -{"mode": "train", "epoch": 1, "iter": 254, "lr": 0.0002, "memory": 49164, "data_time": 0.07095, "loss_cls_0": 0.74455, "loss_box_0": 1.6799, "loss_cns_0": 0.62684, "loss_yns_0": 0.15065, "loss_cls_1": 0.84339, "loss_box_1": 1.57757, "loss_cns_1": 0.66416, "loss_yns_1": 0.14925, "loss_cls_2": 0.84422, "loss_box_2": 1.54043, "loss_cns_2": 0.66206, "loss_yns_2": 0.15057, "loss_cls_3": 0.84485, "loss_box_3": 1.51756, "loss_cns_3": 0.66318, "loss_yns_3": 0.14832, "loss_cls_4": 0.84796, "loss_box_4": 1.53155, "loss_cns_4": 0.66411, "loss_yns_4": 0.14913, "loss_cls_5": 0.85207, "loss_box_5": 1.54595, "loss_cns_5": 0.66477, "loss_yns_5": 0.15106, "loss_cls_dn_0": 0.16582, "loss_box_dn_0": 0.74521, "loss_cls_dn_1": 0.12501, "loss_box_dn_1": 0.73201, "loss_cls_dn_2": 0.12273, "loss_box_dn_2": 0.73016, "loss_cls_dn_3": 0.12361, "loss_box_dn_3": 0.7215, "loss_cls_dn_4": 0.12691, "loss_box_dn_4": 0.72612, "loss_cls_dn_5": 0.13441, "loss_box_dn_5": 0.74247, "loss_dense_depth": 0.72354, "loss": 25.1336, "grad_norm": 43.33118, "time": 1.58468} -{"mode": "train", "epoch": 1, "iter": 255, "lr": 0.0002, "memory": 49164, "data_time": 0.08653, "loss_cls_0": 0.73578, "loss_box_0": 1.65862, "loss_cns_0": 0.63042, "loss_yns_0": 0.15055, "loss_cls_1": 0.82876, "loss_box_1": 1.54875, "loss_cns_1": 0.66655, "loss_yns_1": 0.15097, "loss_cls_2": 0.8343, "loss_box_2": 1.52785, "loss_cns_2": 0.66411, "loss_yns_2": 0.15111, "loss_cls_3": 0.83852, "loss_box_3": 1.51024, "loss_cns_3": 0.66741, "loss_yns_3": 0.1503, "loss_cls_4": 0.83998, "loss_box_4": 1.50641, "loss_cns_4": 0.66973, "loss_yns_4": 0.14954, "loss_cls_5": 0.84217, "loss_box_5": 1.52428, "loss_cns_5": 0.66571, "loss_yns_5": 0.15016, "loss_cls_dn_0": 0.16407, "loss_box_dn_0": 0.74044, "loss_cls_dn_1": 0.12327, "loss_box_dn_1": 0.7171, "loss_cls_dn_2": 0.12238, "loss_box_dn_2": 0.71407, "loss_cls_dn_3": 0.1225, "loss_box_dn_3": 0.70525, "loss_cls_dn_4": 0.12891, "loss_box_dn_4": 0.70055, "loss_cls_dn_5": 0.12651, "loss_box_dn_5": 0.71322, "loss_dense_depth": 0.72608, "loss": 24.86655, "grad_norm": 38.69933, "time": 1.59725} -{"mode": "train", "epoch": 1, "iter": 256, "lr": 0.0002, "memory": 49164, "data_time": 0.07387, "loss_cls_0": 0.74541, "loss_box_0": 1.66021, "loss_cns_0": 0.62153, "loss_yns_0": 0.14936, "loss_cls_1": 0.83003, "loss_box_1": 1.52578, "loss_cns_1": 0.66572, "loss_yns_1": 0.14799, "loss_cls_2": 0.841, "loss_box_2": 1.49647, "loss_cns_2": 0.66804, "loss_yns_2": 0.15013, "loss_cls_3": 0.84364, "loss_box_3": 1.51168, "loss_cns_3": 0.66722, "loss_yns_3": 0.15022, "loss_cls_4": 0.85393, "loss_box_4": 1.47364, "loss_cns_4": 0.66837, "loss_yns_4": 0.14889, "loss_cls_5": 0.85303, "loss_box_5": 1.48325, "loss_cns_5": 0.66543, "loss_yns_5": 0.15029, "loss_cls_dn_0": 0.16069, "loss_box_dn_0": 0.73611, "loss_cls_dn_1": 0.11403, "loss_box_dn_1": 0.6801, "loss_cls_dn_2": 0.11474, "loss_box_dn_2": 0.67338, "loss_cls_dn_3": 0.11915, "loss_box_dn_3": 0.68312, "loss_cls_dn_4": 0.12516, "loss_box_dn_4": 0.68091, "loss_cls_dn_5": 0.11934, "loss_box_dn_5": 0.68788, "loss_dense_depth": 0.72672, "loss": 24.59261, "grad_norm": 44.36017, "time": 1.56743} -{"mode": "train", "epoch": 1, "iter": 257, "lr": 0.0002, "memory": 49164, "data_time": 0.07694, "loss_cls_0": 0.76374, "loss_box_0": 1.68065, "loss_cns_0": 0.62426, "loss_yns_0": 0.15125, "loss_cls_1": 0.83776, "loss_box_1": 1.56671, "loss_cns_1": 0.66073, "loss_yns_1": 0.15028, "loss_cls_2": 0.85615, "loss_box_2": 1.5468, "loss_cns_2": 0.66258, "loss_yns_2": 0.1523, "loss_cls_3": 0.85034, "loss_box_3": 1.54205, "loss_cns_3": 0.66597, "loss_yns_3": 0.14913, "loss_cls_4": 0.85266, "loss_box_4": 1.51966, "loss_cns_4": 0.66529, "loss_yns_4": 0.14891, "loss_cls_5": 0.8571, "loss_box_5": 1.52906, "loss_cns_5": 0.66369, "loss_yns_5": 0.14794, "loss_cls_dn_0": 0.16089, "loss_box_dn_0": 0.73883, "loss_cls_dn_1": 0.11397, "loss_box_dn_1": 0.67363, "loss_cls_dn_2": 0.1165, "loss_box_dn_2": 0.67268, "loss_cls_dn_3": 0.12123, "loss_box_dn_3": 0.68214, "loss_cls_dn_4": 0.12179, "loss_box_dn_4": 0.68135, "loss_cls_dn_5": 0.12311, "loss_box_dn_5": 0.69926, "loss_dense_depth": 0.75607, "loss": 24.90647, "grad_norm": 34.59807, "time": 1.58082} -{"mode": "train", "epoch": 1, "iter": 258, "lr": 0.0002, "memory": 49164, "data_time": 0.07336, "loss_cls_0": 0.75219, "loss_box_0": 1.70197, "loss_cns_0": 0.61633, "loss_yns_0": 0.14836, "loss_cls_1": 0.84581, "loss_box_1": 1.58827, "loss_cns_1": 0.65614, "loss_yns_1": 0.14828, "loss_cls_2": 0.85369, "loss_box_2": 1.5539, "loss_cns_2": 0.66433, "loss_yns_2": 0.14877, "loss_cls_3": 0.85698, "loss_box_3": 1.53221, "loss_cns_3": 0.66619, "loss_yns_3": 0.14664, "loss_cls_4": 0.85288, "loss_box_4": 1.52083, "loss_cns_4": 0.66383, "loss_yns_4": 0.14663, "loss_cls_5": 0.85611, "loss_box_5": 1.51322, "loss_cns_5": 0.66399, "loss_yns_5": 0.14828, "loss_cls_dn_0": 0.16018, "loss_box_dn_0": 0.73571, "loss_cls_dn_1": 0.11759, "loss_box_dn_1": 0.6691, "loss_cls_dn_2": 0.11756, "loss_box_dn_2": 0.6661, "loss_cls_dn_3": 0.11699, "loss_box_dn_3": 0.66988, "loss_cls_dn_4": 0.11822, "loss_box_dn_4": 0.67307, "loss_cls_dn_5": 0.12167, "loss_box_dn_5": 0.68655, "loss_dense_depth": 0.76326, "loss": 24.86171, "grad_norm": 36.2046, "time": 1.59062} -{"mode": "train", "epoch": 1, "iter": 259, "lr": 0.0002, "memory": 49164, "data_time": 0.08526, "loss_cls_0": 0.79426, "loss_box_0": 1.70417, "loss_cns_0": 0.62199, "loss_yns_0": 0.14723, "loss_cls_1": 0.85952, "loss_box_1": 1.64795, "loss_cns_1": 0.6532, "loss_yns_1": 0.14498, "loss_cls_2": 0.88291, "loss_box_2": 1.60252, "loss_cns_2": 0.65505, "loss_yns_2": 0.14467, "loss_cls_3": 0.88468, "loss_box_3": 1.61596, "loss_cns_3": 0.66275, "loss_yns_3": 0.14522, "loss_cls_4": 0.88917, "loss_box_4": 1.61467, "loss_cns_4": 0.66753, "loss_yns_4": 0.14589, "loss_cls_5": 0.89019, "loss_box_5": 1.58633, "loss_cns_5": 0.65497, "loss_yns_5": 0.14606, "loss_cls_dn_0": 0.16579, "loss_box_dn_0": 0.74163, "loss_cls_dn_1": 0.11788, "loss_box_dn_1": 0.67595, "loss_cls_dn_2": 0.11767, "loss_box_dn_2": 0.66936, "loss_cls_dn_3": 0.11957, "loss_box_dn_3": 0.6793, "loss_cls_dn_4": 0.12259, "loss_box_dn_4": 0.68645, "loss_cls_dn_5": 0.12488, "loss_box_dn_5": 0.69446, "loss_dense_depth": 0.82343, "loss": 25.50083, "grad_norm": 41.04888, "time": 1.57564} -{"mode": "train", "epoch": 1, "iter": 260, "lr": 0.0002, "memory": 49164, "data_time": 0.0815, "loss_cls_0": 0.78294, "loss_box_0": 1.67907, "loss_cns_0": 0.62961, "loss_yns_0": 0.15064, "loss_cls_1": 0.83667, "loss_box_1": 1.61904, "loss_cns_1": 0.65456, "loss_yns_1": 0.14546, "loss_cls_2": 0.86332, "loss_box_2": 1.57769, "loss_cns_2": 0.65686, "loss_yns_2": 0.14657, "loss_cls_3": 0.87507, "loss_box_3": 1.56705, "loss_cns_3": 0.66294, "loss_yns_3": 0.14487, "loss_cls_4": 0.87391, "loss_box_4": 1.56644, "loss_cns_4": 0.66679, "loss_yns_4": 0.14564, "loss_cls_5": 0.87275, "loss_box_5": 1.56337, "loss_cns_5": 0.66011, "loss_yns_5": 0.14587, "loss_cls_dn_0": 0.1591, "loss_box_dn_0": 0.74111, "loss_cls_dn_1": 0.11984, "loss_box_dn_1": 0.68666, "loss_cls_dn_2": 0.12044, "loss_box_dn_2": 0.6786, "loss_cls_dn_3": 0.12156, "loss_box_dn_3": 0.67774, "loss_cls_dn_4": 0.12305, "loss_box_dn_4": 0.68101, "loss_cls_dn_5": 0.12689, "loss_box_dn_5": 0.69053, "loss_dense_depth": 0.75258, "loss": 25.16634, "grad_norm": 26.70441, "time": 8.54749} -{"mode": "train", "epoch": 1, "iter": 261, "lr": 0.0002, "memory": 49164, "data_time": 0.10027, "loss_cls_0": 0.78094, "loss_box_0": 1.68803, "loss_cns_0": 0.62188, "loss_yns_0": 0.14841, "loss_cls_1": 0.8396, "loss_box_1": 1.58801, "loss_cns_1": 0.65303, "loss_yns_1": 0.14554, "loss_cls_2": 0.86569, "loss_box_2": 1.5377, "loss_cns_2": 0.6567, "loss_yns_2": 0.14417, "loss_cls_3": 0.8635, "loss_box_3": 1.5364, "loss_cns_3": 0.65679, "loss_yns_3": 0.14353, "loss_cls_4": 0.88562, "loss_box_4": 1.53359, "loss_cns_4": 0.65855, "loss_yns_4": 0.14397, "loss_cls_5": 0.8875, "loss_box_5": 1.53023, "loss_cns_5": 0.658, "loss_yns_5": 0.14231, "loss_cls_dn_0": 0.16119, "loss_box_dn_0": 0.74195, "loss_cls_dn_1": 0.12207, "loss_box_dn_1": 0.68668, "loss_cls_dn_2": 0.12261, "loss_box_dn_2": 0.67699, "loss_cls_dn_3": 0.1234, "loss_box_dn_3": 0.67631, "loss_cls_dn_4": 0.13103, "loss_box_dn_4": 0.67864, "loss_cls_dn_5": 0.13447, "loss_box_dn_5": 0.68112, "loss_dense_depth": 0.93396, "loss": 25.18013, "grad_norm": 38.76659, "time": 1.62373} -{"mode": "train", "epoch": 1, "iter": 262, "lr": 0.0002, "memory": 49164, "data_time": 0.15997, "loss_cls_0": 0.78777, "loss_box_0": 1.71077, "loss_cns_0": 0.61876, "loss_yns_0": 0.14573, "loss_cls_1": 0.84475, "loss_box_1": 1.56131, "loss_cns_1": 0.65787, "loss_yns_1": 0.14404, "loss_cls_2": 0.86411, "loss_box_2": 1.53145, "loss_cns_2": 0.65721, "loss_yns_2": 0.14325, "loss_cls_3": 0.8702, "loss_box_3": 1.51655, "loss_cns_3": 0.6648, "loss_yns_3": 0.14459, "loss_cls_4": 0.88271, "loss_box_4": 1.5138, "loss_cns_4": 0.66544, "loss_yns_4": 0.14392, "loss_cls_5": 0.89626, "loss_box_5": 1.49548, "loss_cns_5": 0.66168, "loss_yns_5": 0.144, "loss_cls_dn_0": 0.16244, "loss_box_dn_0": 0.74507, "loss_cls_dn_1": 0.11833, "loss_box_dn_1": 0.67466, "loss_cls_dn_2": 0.11878, "loss_box_dn_2": 0.66437, "loss_cls_dn_3": 0.11988, "loss_box_dn_3": 0.66084, "loss_cls_dn_4": 0.12691, "loss_box_dn_4": 0.66014, "loss_cls_dn_5": 0.12882, "loss_box_dn_5": 0.65806, "loss_dense_depth": 0.90964, "loss": 25.01438, "grad_norm": 28.3236, "time": 1.62383} -{"mode": "train", "epoch": 1, "iter": 263, "lr": 0.0002, "memory": 49164, "data_time": 0.07036, "loss_cls_0": 0.77418, "loss_box_0": 1.68467, "loss_cns_0": 0.62588, "loss_yns_0": 0.14315, "loss_cls_1": 0.83367, "loss_box_1": 1.58479, "loss_cns_1": 0.6553, "loss_yns_1": 0.14058, "loss_cls_2": 0.85374, "loss_box_2": 1.54206, "loss_cns_2": 0.65788, "loss_yns_2": 0.14184, "loss_cls_3": 0.86134, "loss_box_3": 1.53633, "loss_cns_3": 0.66037, "loss_yns_3": 0.13947, "loss_cls_4": 0.85918, "loss_box_4": 1.52627, "loss_cns_4": 0.6626, "loss_yns_4": 0.13941, "loss_cls_5": 0.88653, "loss_box_5": 1.52795, "loss_cns_5": 0.65839, "loss_yns_5": 0.14129, "loss_cls_dn_0": 0.16233, "loss_box_dn_0": 0.74536, "loss_cls_dn_1": 0.11309, "loss_box_dn_1": 0.66872, "loss_cls_dn_2": 0.11466, "loss_box_dn_2": 0.65571, "loss_cls_dn_3": 0.11651, "loss_box_dn_3": 0.65662, "loss_cls_dn_4": 0.11796, "loss_box_dn_4": 0.65622, "loss_cls_dn_5": 0.1221, "loss_box_dn_5": 0.6633, "loss_dense_depth": 0.82938, "loss": 24.85883, "grad_norm": 20.29599, "time": 1.56332} -{"mode": "train", "epoch": 1, "iter": 264, "lr": 0.00021, "memory": 49164, "data_time": 0.07825, "loss_cls_0": 0.76326, "loss_box_0": 1.71108, "loss_cns_0": 0.6263, "loss_yns_0": 0.1445, "loss_cls_1": 0.85429, "loss_box_1": 1.59966, "loss_cns_1": 0.66047, "loss_yns_1": 0.14171, "loss_cls_2": 0.8714, "loss_box_2": 1.55036, "loss_cns_2": 0.66361, "loss_yns_2": 0.14193, "loss_cls_3": 0.86197, "loss_box_3": 1.54868, "loss_cns_3": 0.66222, "loss_yns_3": 0.14013, "loss_cls_4": 0.87558, "loss_box_4": 1.5362, "loss_cns_4": 0.66439, "loss_yns_4": 0.14082, "loss_cls_5": 0.87434, "loss_box_5": 1.53421, "loss_cns_5": 0.66307, "loss_yns_5": 0.14132, "loss_cls_dn_0": 0.15691, "loss_box_dn_0": 0.74219, "loss_cls_dn_1": 0.11567, "loss_box_dn_1": 0.68603, "loss_cls_dn_2": 0.11877, "loss_box_dn_2": 0.67042, "loss_cls_dn_3": 0.11636, "loss_box_dn_3": 0.67936, "loss_cls_dn_4": 0.12172, "loss_box_dn_4": 0.68492, "loss_cls_dn_5": 0.12376, "loss_box_dn_5": 0.69452, "loss_dense_depth": 0.99774, "loss": 25.2799, "grad_norm": 38.27979, "time": 1.60303} -{"mode": "train", "epoch": 1, "iter": 265, "lr": 0.00021, "memory": 49164, "data_time": 0.07479, "loss_cls_0": 0.78694, "loss_box_0": 1.71338, "loss_cns_0": 0.63045, "loss_yns_0": 0.14564, "loss_cls_1": 0.86163, "loss_box_1": 1.58499, "loss_cns_1": 0.66018, "loss_yns_1": 0.14367, "loss_cls_2": 0.86687, "loss_box_2": 1.57362, "loss_cns_2": 0.66331, "loss_yns_2": 0.14109, "loss_cls_3": 0.87136, "loss_box_3": 1.55629, "loss_cns_3": 0.66078, "loss_yns_3": 0.14143, "loss_cls_4": 0.87642, "loss_box_4": 1.55954, "loss_cns_4": 0.66183, "loss_yns_4": 0.14256, "loss_cls_5": 0.88212, "loss_box_5": 1.55482, "loss_cns_5": 0.66081, "loss_yns_5": 0.14244, "loss_cls_dn_0": 0.16775, "loss_box_dn_0": 0.73909, "loss_cls_dn_1": 0.11579, "loss_box_dn_1": 0.69695, "loss_cls_dn_2": 0.11602, "loss_box_dn_2": 0.69272, "loss_cls_dn_3": 0.11712, "loss_box_dn_3": 0.70038, "loss_cls_dn_4": 0.12279, "loss_box_dn_4": 0.71362, "loss_cls_dn_5": 0.12222, "loss_box_dn_5": 0.72549, "loss_dense_depth": 0.81904, "loss": 25.33115, "grad_norm": 33.59963, "time": 1.6107} -{"mode": "train", "epoch": 1, "iter": 266, "lr": 0.00021, "memory": 49164, "data_time": 0.07348, "loss_cls_0": 0.76846, "loss_box_0": 1.66573, "loss_cns_0": 0.63086, "loss_yns_0": 0.14409, "loss_cls_1": 0.83485, "loss_box_1": 1.55265, "loss_cns_1": 0.66059, "loss_yns_1": 0.14049, "loss_cls_2": 0.83961, "loss_box_2": 1.53462, "loss_cns_2": 0.66251, "loss_yns_2": 0.14083, "loss_cls_3": 0.83762, "loss_box_3": 1.52383, "loss_cns_3": 0.66512, "loss_yns_3": 0.14208, "loss_cls_4": 0.8416, "loss_box_4": 1.52265, "loss_cns_4": 0.6646, "loss_yns_4": 0.14303, "loss_cls_5": 0.85593, "loss_box_5": 1.51292, "loss_cns_5": 0.6668, "loss_yns_5": 0.14281, "loss_cls_dn_0": 0.16977, "loss_box_dn_0": 0.74266, "loss_cls_dn_1": 0.11687, "loss_box_dn_1": 0.71846, "loss_cls_dn_2": 0.11656, "loss_box_dn_2": 0.71717, "loss_cls_dn_3": 0.11913, "loss_box_dn_3": 0.71872, "loss_cls_dn_4": 0.12204, "loss_box_dn_4": 0.72981, "loss_cls_dn_5": 0.12239, "loss_box_dn_5": 0.73705, "loss_dense_depth": 0.91473, "loss": 25.13962, "grad_norm": 33.56141, "time": 1.59359} -{"mode": "train", "epoch": 1, "iter": 267, "lr": 0.00021, "memory": 49164, "data_time": 0.10512, "loss_cls_0": 0.74619, "loss_box_0": 1.65441, "loss_cns_0": 0.62693, "loss_yns_0": 0.1416, "loss_cls_1": 0.82671, "loss_box_1": 1.54321, "loss_cns_1": 0.65957, "loss_yns_1": 0.1386, "loss_cls_2": 0.83869, "loss_box_2": 1.50652, "loss_cns_2": 0.65959, "loss_yns_2": 0.13957, "loss_cls_3": 0.83563, "loss_box_3": 1.5087, "loss_cns_3": 0.66415, "loss_yns_3": 0.14336, "loss_cls_4": 0.84369, "loss_box_4": 1.50808, "loss_cns_4": 0.66418, "loss_yns_4": 0.1445, "loss_cls_5": 0.8592, "loss_box_5": 1.50315, "loss_cns_5": 0.66553, "loss_yns_5": 0.14402, "loss_cls_dn_0": 0.16115, "loss_box_dn_0": 0.73925, "loss_cls_dn_1": 0.11403, "loss_box_dn_1": 0.71618, "loss_cls_dn_2": 0.11486, "loss_box_dn_2": 0.70681, "loss_cls_dn_3": 0.11723, "loss_box_dn_3": 0.70294, "loss_cls_dn_4": 0.11687, "loss_box_dn_4": 0.70509, "loss_cls_dn_5": 0.11841, "loss_box_dn_5": 0.7101, "loss_dense_depth": 0.84418, "loss": 24.83288, "grad_norm": 40.43327, "time": 1.61978} -{"mode": "train", "epoch": 1, "iter": 268, "lr": 0.00021, "memory": 49164, "data_time": 0.07699, "loss_cls_0": 0.74979, "loss_box_0": 1.63592, "loss_cns_0": 0.63507, "loss_yns_0": 0.14525, "loss_cls_1": 0.82231, "loss_box_1": 1.5046, "loss_cns_1": 0.66741, "loss_yns_1": 0.14502, "loss_cls_2": 0.83499, "loss_box_2": 1.46297, "loss_cns_2": 0.67379, "loss_yns_2": 0.14432, "loss_cls_3": 0.84112, "loss_box_3": 1.46487, "loss_cns_3": 0.66816, "loss_yns_3": 0.14533, "loss_cls_4": 0.83934, "loss_box_4": 1.4582, "loss_cns_4": 0.66785, "loss_yns_4": 0.14536, "loss_cls_5": 0.84835, "loss_box_5": 1.44957, "loss_cns_5": 0.66919, "loss_yns_5": 0.14581, "loss_cls_dn_0": 0.16009, "loss_box_dn_0": 0.7356, "loss_cls_dn_1": 0.11903, "loss_box_dn_1": 0.69723, "loss_cls_dn_2": 0.11958, "loss_box_dn_2": 0.68524, "loss_cls_dn_3": 0.12082, "loss_box_dn_3": 0.68176, "loss_cls_dn_4": 0.1203, "loss_box_dn_4": 0.67925, "loss_cls_dn_5": 0.12098, "loss_box_dn_5": 0.67828, "loss_dense_depth": 0.83341, "loss": 24.51618, "grad_norm": 25.82797, "time": 1.58569} -{"mode": "train", "epoch": 1, "iter": 269, "lr": 0.00021, "memory": 49164, "data_time": 0.07426, "loss_cls_0": 0.75777, "loss_box_0": 1.63647, "loss_cns_0": 0.63157, "loss_yns_0": 0.14514, "loss_cls_1": 0.82606, "loss_box_1": 1.518, "loss_cns_1": 0.66097, "loss_yns_1": 0.14372, "loss_cls_2": 0.8398, "loss_box_2": 1.48947, "loss_cns_2": 0.66265, "loss_yns_2": 0.14472, "loss_cls_3": 0.84414, "loss_box_3": 1.50494, "loss_cns_3": 0.65962, "loss_yns_3": 0.14586, "loss_cls_4": 0.86107, "loss_box_4": 1.4878, "loss_cns_4": 0.6586, "loss_yns_4": 0.14548, "loss_cls_5": 0.84882, "loss_box_5": 1.49744, "loss_cns_5": 0.66075, "loss_yns_5": 0.14493, "loss_cls_dn_0": 0.15887, "loss_box_dn_0": 0.7406, "loss_cls_dn_1": 0.11411, "loss_box_dn_1": 0.68678, "loss_cls_dn_2": 0.11286, "loss_box_dn_2": 0.67775, "loss_cls_dn_3": 0.11335, "loss_box_dn_3": 0.67903, "loss_cls_dn_4": 0.12136, "loss_box_dn_4": 0.67715, "loss_cls_dn_5": 0.11961, "loss_box_dn_5": 0.67835, "loss_dense_depth": 0.86588, "loss": 24.66149, "grad_norm": 52.50675, "time": 1.62295} -{"mode": "train", "epoch": 1, "iter": 270, "lr": 0.00021, "memory": 49164, "data_time": 0.07978, "loss_cls_0": 0.78871, "loss_box_0": 1.69731, "loss_cns_0": 0.62122, "loss_yns_0": 0.14757, "loss_cls_1": 0.84656, "loss_box_1": 1.55401, "loss_cns_1": 0.65888, "loss_yns_1": 0.14584, "loss_cls_2": 0.87084, "loss_box_2": 1.51643, "loss_cns_2": 0.65975, "loss_yns_2": 0.14634, "loss_cls_3": 0.8807, "loss_box_3": 1.50325, "loss_cns_3": 0.65714, "loss_yns_3": 0.14695, "loss_cls_4": 0.88417, "loss_box_4": 1.48862, "loss_cns_4": 0.65355, "loss_yns_4": 0.14337, "loss_cls_5": 0.88023, "loss_box_5": 1.52129, "loss_cns_5": 0.66003, "loss_yns_5": 0.1447, "loss_cls_dn_0": 0.16678, "loss_box_dn_0": 0.75311, "loss_cls_dn_1": 0.11199, "loss_box_dn_1": 0.68221, "loss_cls_dn_2": 0.11443, "loss_box_dn_2": 0.67176, "loss_cls_dn_3": 0.11996, "loss_box_dn_3": 0.67618, "loss_cls_dn_4": 0.12361, "loss_box_dn_4": 0.68704, "loss_cls_dn_5": 0.11985, "loss_box_dn_5": 0.69996, "loss_dense_depth": 0.99604, "loss": 25.1404, "grad_norm": 36.19583, "time": 1.61685} -{"mode": "train", "epoch": 1, "iter": 271, "lr": 0.00021, "memory": 49164, "data_time": 0.07668, "loss_cls_0": 0.72784, "loss_box_0": 1.67067, "loss_cns_0": 0.62932, "loss_yns_0": 0.14762, "loss_cls_1": 0.82373, "loss_box_1": 1.54654, "loss_cns_1": 0.65914, "loss_yns_1": 0.14501, "loss_cls_2": 0.85431, "loss_box_2": 1.53595, "loss_cns_2": 0.66354, "loss_yns_2": 0.14521, "loss_cls_3": 0.86399, "loss_box_3": 1.50506, "loss_cns_3": 0.66135, "loss_yns_3": 0.14519, "loss_cls_4": 0.86995, "loss_box_4": 1.50058, "loss_cns_4": 0.66209, "loss_yns_4": 0.14341, "loss_cls_5": 0.87938, "loss_box_5": 1.50021, "loss_cns_5": 0.66249, "loss_yns_5": 0.14477, "loss_cls_dn_0": 0.15683, "loss_box_dn_0": 0.75099, "loss_cls_dn_1": 0.11078, "loss_box_dn_1": 0.69429, "loss_cls_dn_2": 0.11207, "loss_box_dn_2": 0.70208, "loss_cls_dn_3": 0.11736, "loss_box_dn_3": 0.70018, "loss_cls_dn_4": 0.11619, "loss_box_dn_4": 0.71162, "loss_cls_dn_5": 0.11627, "loss_box_dn_5": 0.72872, "loss_dense_depth": 0.81767, "loss": 24.9224, "grad_norm": 59.63836, "time": 1.62517} -{"mode": "train", "epoch": 1, "iter": 272, "lr": 0.00021, "memory": 49164, "data_time": 0.10906, "loss_cls_0": 0.75318, "loss_box_0": 1.62484, "loss_cns_0": 0.62789, "loss_yns_0": 0.14605, "loss_cls_1": 0.84102, "loss_box_1": 1.54307, "loss_cns_1": 0.66178, "loss_yns_1": 0.14582, "loss_cls_2": 0.855, "loss_box_2": 1.52469, "loss_cns_2": 0.66518, "loss_yns_2": 0.14543, "loss_cls_3": 0.86421, "loss_box_3": 1.49281, "loss_cns_3": 0.66332, "loss_yns_3": 0.14516, "loss_cls_4": 0.86562, "loss_box_4": 1.48675, "loss_cns_4": 0.66531, "loss_yns_4": 0.14601, "loss_cls_5": 0.85868, "loss_box_5": 1.48476, "loss_cns_5": 0.66401, "loss_yns_5": 0.14376, "loss_cls_dn_0": 0.15803, "loss_box_dn_0": 0.73314, "loss_cls_dn_1": 0.11013, "loss_box_dn_1": 0.71322, "loss_cls_dn_2": 0.11088, "loss_box_dn_2": 0.72026, "loss_cls_dn_3": 0.11491, "loss_box_dn_3": 0.71769, "loss_cls_dn_4": 0.11608, "loss_box_dn_4": 0.72949, "loss_cls_dn_5": 0.11637, "loss_box_dn_5": 0.74658, "loss_dense_depth": 0.85162, "loss": 24.95276, "grad_norm": 46.12945, "time": 1.62713} -{"mode": "train", "epoch": 1, "iter": 273, "lr": 0.00021, "memory": 49164, "data_time": 0.07483, "loss_cls_0": 0.77037, "loss_box_0": 1.65213, "loss_cns_0": 0.61124, "loss_yns_0": 0.1422, "loss_cls_1": 0.82753, "loss_box_1": 1.53937, "loss_cns_1": 0.65838, "loss_yns_1": 0.14279, "loss_cls_2": 0.83557, "loss_box_2": 1.52627, "loss_cns_2": 0.66278, "loss_yns_2": 0.14393, "loss_cls_3": 0.83877, "loss_box_3": 1.51169, "loss_cns_3": 0.65774, "loss_yns_3": 0.1424, "loss_cls_4": 0.84666, "loss_box_4": 1.52254, "loss_cns_4": 0.65579, "loss_yns_4": 0.14267, "loss_cls_5": 0.88057, "loss_box_5": 1.49761, "loss_cns_5": 0.65185, "loss_yns_5": 0.14053, "loss_cls_dn_0": 0.16231, "loss_box_dn_0": 0.778, "loss_cls_dn_1": 0.11358, "loss_box_dn_1": 0.7188, "loss_cls_dn_2": 0.1117, "loss_box_dn_2": 0.71411, "loss_cls_dn_3": 0.11301, "loss_box_dn_3": 0.71514, "loss_cls_dn_4": 0.11518, "loss_box_dn_4": 0.72949, "loss_cls_dn_5": 0.11668, "loss_box_dn_5": 0.73419, "loss_dense_depth": 0.79015, "loss": 24.91373, "grad_norm": 56.30774, "time": 1.56027} -{"mode": "train", "epoch": 1, "iter": 274, "lr": 0.00021, "memory": 49164, "data_time": 1.55243, "loss_cls_0": 0.76196, "loss_box_0": 1.63804, "loss_cns_0": 0.62258, "loss_yns_0": 0.14339, "loss_cls_1": 0.81244, "loss_box_1": 1.50051, "loss_cns_1": 0.65772, "loss_yns_1": 0.14127, "loss_cls_2": 0.81726, "loss_box_2": 1.46067, "loss_cns_2": 0.65957, "loss_yns_2": 0.14132, "loss_cls_3": 0.82368, "loss_box_3": 1.43982, "loss_cns_3": 0.6583, "loss_yns_3": 0.14124, "loss_cls_4": 0.83924, "loss_box_4": 1.44024, "loss_cns_4": 0.65885, "loss_yns_4": 0.14073, "loss_cls_5": 0.844, "loss_box_5": 1.42314, "loss_cns_5": 0.66083, "loss_yns_5": 0.1405, "loss_cls_dn_0": 0.16168, "loss_box_dn_0": 0.77441, "loss_cls_dn_1": 0.10777, "loss_box_dn_1": 0.68041, "loss_cls_dn_2": 0.10573, "loss_box_dn_2": 0.6612, "loss_cls_dn_3": 0.10627, "loss_box_dn_3": 0.65224, "loss_cls_dn_4": 0.11428, "loss_box_dn_4": 0.6564, "loss_cls_dn_5": 0.11432, "loss_box_dn_5": 0.65652, "loss_dense_depth": 0.8716, "loss": 24.23014, "grad_norm": 37.75471, "time": 3.07652} -{"mode": "train", "epoch": 1, "iter": 275, "lr": 0.00021, "memory": 49164, "data_time": 0.07149, "loss_cls_0": 0.73854, "loss_box_0": 1.6527, "loss_cns_0": 0.62812, "loss_yns_0": 0.14354, "loss_cls_1": 0.79112, "loss_box_1": 1.52209, "loss_cns_1": 0.65552, "loss_yns_1": 0.14009, "loss_cls_2": 0.80753, "loss_box_2": 1.51896, "loss_cns_2": 0.65447, "loss_yns_2": 0.14115, "loss_cls_3": 0.82993, "loss_box_3": 1.51093, "loss_cns_3": 0.65642, "loss_yns_3": 0.14047, "loss_cls_4": 0.85673, "loss_box_4": 1.50135, "loss_cns_4": 0.65591, "loss_yns_4": 0.14252, "loss_cls_5": 0.84743, "loss_box_5": 1.51409, "loss_cns_5": 0.65445, "loss_yns_5": 0.14012, "loss_cls_dn_0": 0.15839, "loss_box_dn_0": 0.74616, "loss_cls_dn_1": 0.1049, "loss_box_dn_1": 0.65477, "loss_cls_dn_2": 0.10623, "loss_box_dn_2": 0.65245, "loss_cls_dn_3": 0.11089, "loss_box_dn_3": 0.64936, "loss_cls_dn_4": 0.11822, "loss_box_dn_4": 0.64701, "loss_cls_dn_5": 0.11726, "loss_box_dn_5": 0.65936, "loss_dense_depth": 0.84671, "loss": 24.41589, "grad_norm": 61.40615, "time": 1.58426} -{"mode": "train", "epoch": 1, "iter": 276, "lr": 0.00021, "memory": 49164, "data_time": 0.07108, "loss_cls_0": 0.75079, "loss_box_0": 1.66832, "loss_cns_0": 0.6267, "loss_yns_0": 0.14292, "loss_cls_1": 0.81272, "loss_box_1": 1.53798, "loss_cns_1": 0.65382, "loss_yns_1": 0.14165, "loss_cls_2": 0.82239, "loss_box_2": 1.50733, "loss_cns_2": 0.65709, "loss_yns_2": 0.14162, "loss_cls_3": 0.84528, "loss_box_3": 1.51066, "loss_cns_3": 0.65606, "loss_yns_3": 0.13978, "loss_cls_4": 0.85999, "loss_box_4": 1.49217, "loss_cns_4": 0.65901, "loss_yns_4": 0.14147, "loss_cls_5": 0.85778, "loss_box_5": 1.50311, "loss_cns_5": 0.66053, "loss_yns_5": 0.14169, "loss_cls_dn_0": 0.16004, "loss_box_dn_0": 0.74883, "loss_cls_dn_1": 0.10863, "loss_box_dn_1": 0.65588, "loss_cls_dn_2": 0.10726, "loss_box_dn_2": 0.64577, "loss_cls_dn_3": 0.11133, "loss_box_dn_3": 0.65167, "loss_cls_dn_4": 0.11609, "loss_box_dn_4": 0.6495, "loss_cls_dn_5": 0.11537, "loss_box_dn_5": 0.66195, "loss_dense_depth": 0.82643, "loss": 24.4896, "grad_norm": 54.2076, "time": 1.54954} -{"mode": "train", "epoch": 1, "iter": 277, "lr": 0.00021, "memory": 49164, "data_time": 0.07349, "loss_cls_0": 0.74752, "loss_box_0": 1.66508, "loss_cns_0": 0.62952, "loss_yns_0": 0.14469, "loss_cls_1": 0.81578, "loss_box_1": 1.53759, "loss_cns_1": 0.65736, "loss_yns_1": 0.14347, "loss_cls_2": 0.81797, "loss_box_2": 1.52531, "loss_cns_2": 0.66278, "loss_yns_2": 0.14203, "loss_cls_3": 0.82456, "loss_box_3": 1.53279, "loss_cns_3": 0.65702, "loss_yns_3": 0.14086, "loss_cls_4": 0.83521, "loss_box_4": 1.52911, "loss_cns_4": 0.66221, "loss_yns_4": 0.14098, "loss_cls_5": 0.83887, "loss_box_5": 1.51694, "loss_cns_5": 0.66021, "loss_yns_5": 0.1418, "loss_cls_dn_0": 0.15578, "loss_box_dn_0": 0.73865, "loss_cls_dn_1": 0.1053, "loss_box_dn_1": 0.66137, "loss_cls_dn_2": 0.10515, "loss_box_dn_2": 0.65677, "loss_cls_dn_3": 0.10792, "loss_box_dn_3": 0.67034, "loss_cls_dn_4": 0.1171, "loss_box_dn_4": 0.6802, "loss_cls_dn_5": 0.11831, "loss_box_dn_5": 0.68694, "loss_dense_depth": 0.83599, "loss": 24.60949, "grad_norm": 57.47681, "time": 1.58726} -{"mode": "train", "epoch": 1, "iter": 278, "lr": 0.00021, "memory": 49164, "data_time": 0.07111, "loss_cls_0": 0.75226, "loss_box_0": 1.63033, "loss_cns_0": 0.62792, "loss_yns_0": 0.1434, "loss_cls_1": 0.83042, "loss_box_1": 1.5411, "loss_cns_1": 0.65651, "loss_yns_1": 0.14215, "loss_cls_2": 0.83872, "loss_box_2": 1.56395, "loss_cns_2": 0.66011, "loss_yns_2": 0.14101, "loss_cls_3": 0.8556, "loss_box_3": 1.55064, "loss_cns_3": 0.65337, "loss_yns_3": 0.14018, "loss_cls_4": 0.85821, "loss_box_4": 1.54902, "loss_cns_4": 0.65675, "loss_yns_4": 0.13925, "loss_cls_5": 0.8677, "loss_box_5": 1.53575, "loss_cns_5": 0.65518, "loss_yns_5": 0.14013, "loss_cls_dn_0": 0.15239, "loss_box_dn_0": 0.73006, "loss_cls_dn_1": 0.1055, "loss_box_dn_1": 0.6942, "loss_cls_dn_2": 0.10857, "loss_box_dn_2": 0.69838, "loss_cls_dn_3": 0.11222, "loss_box_dn_3": 0.704, "loss_cls_dn_4": 0.11628, "loss_box_dn_4": 0.71222, "loss_cls_dn_5": 0.11759, "loss_box_dn_5": 0.72153, "loss_dense_depth": 0.75524, "loss": 24.85781, "grad_norm": 60.73838, "time": 1.56607} -{"mode": "train", "epoch": 1, "iter": 279, "lr": 0.00021, "memory": 49164, "data_time": 0.08144, "loss_cls_0": 0.75284, "loss_box_0": 1.61667, "loss_cns_0": 0.62911, "loss_yns_0": 0.14462, "loss_cls_1": 0.83912, "loss_box_1": 1.51319, "loss_cns_1": 0.65523, "loss_yns_1": 0.14393, "loss_cls_2": 0.8454, "loss_box_2": 1.50014, "loss_cns_2": 0.66428, "loss_yns_2": 0.14294, "loss_cls_3": 0.8442, "loss_box_3": 1.47958, "loss_cns_3": 0.66489, "loss_yns_3": 0.14368, "loss_cls_4": 0.84958, "loss_box_4": 1.48152, "loss_cns_4": 0.66468, "loss_yns_4": 0.141, "loss_cls_5": 0.8632, "loss_box_5": 1.48345, "loss_cns_5": 0.66525, "loss_yns_5": 0.14162, "loss_cls_dn_0": 0.15594, "loss_box_dn_0": 0.73489, "loss_cls_dn_1": 0.11022, "loss_box_dn_1": 0.70008, "loss_cls_dn_2": 0.11403, "loss_box_dn_2": 0.69448, "loss_cls_dn_3": 0.11339, "loss_box_dn_3": 0.69611, "loss_cls_dn_4": 0.11653, "loss_box_dn_4": 0.69868, "loss_cls_dn_5": 0.12014, "loss_box_dn_5": 0.70845, "loss_dense_depth": 0.831, "loss": 24.66407, "grad_norm": 40.25448, "time": 1.5769} -{"mode": "train", "epoch": 1, "iter": 280, "lr": 0.00021, "memory": 49164, "data_time": 0.09574, "loss_cls_0": 0.74623, "loss_box_0": 1.63261, "loss_cns_0": 0.62991, "loss_yns_0": 0.14487, "loss_cls_1": 0.82247, "loss_box_1": 1.51425, "loss_cns_1": 0.65893, "loss_yns_1": 0.14291, "loss_cls_2": 0.83564, "loss_box_2": 1.53704, "loss_cns_2": 0.65845, "loss_yns_2": 0.14279, "loss_cls_3": 0.83268, "loss_box_3": 1.51852, "loss_cns_3": 0.66817, "loss_yns_3": 0.14251, "loss_cls_4": 0.84751, "loss_box_4": 1.51994, "loss_cns_4": 0.66524, "loss_yns_4": 0.14188, "loss_cls_5": 0.84461, "loss_box_5": 1.51763, "loss_cns_5": 0.66952, "loss_yns_5": 0.14229, "loss_cls_dn_0": 0.15026, "loss_box_dn_0": 0.74004, "loss_cls_dn_1": 0.10971, "loss_box_dn_1": 0.70696, "loss_cls_dn_2": 0.11011, "loss_box_dn_2": 0.71073, "loss_cls_dn_3": 0.11128, "loss_box_dn_3": 0.7094, "loss_cls_dn_4": 0.11374, "loss_box_dn_4": 0.71435, "loss_cls_dn_5": 0.11763, "loss_box_dn_5": 0.72124, "loss_dense_depth": 0.79187, "loss": 24.78392, "grad_norm": 50.55715, "time": 1.59954} -{"mode": "train", "epoch": 1, "iter": 281, "lr": 0.00021, "memory": 49164, "data_time": 0.10911, "loss_cls_0": 0.74101, "loss_box_0": 1.6133, "loss_cns_0": 0.63575, "loss_yns_0": 0.14369, "loss_cls_1": 0.83311, "loss_box_1": 1.47749, "loss_cns_1": 0.66409, "loss_yns_1": 0.13949, "loss_cls_2": 0.85272, "loss_box_2": 1.48693, "loss_cns_2": 0.66205, "loss_yns_2": 0.13927, "loss_cls_3": 0.84787, "loss_box_3": 1.46809, "loss_cns_3": 0.67126, "loss_yns_3": 0.14025, "loss_cls_4": 0.85782, "loss_box_4": 1.4507, "loss_cns_4": 0.67015, "loss_yns_4": 0.14124, "loss_cls_5": 0.85015, "loss_box_5": 1.44877, "loss_cns_5": 0.67039, "loss_yns_5": 0.14053, "loss_cls_dn_0": 0.1533, "loss_box_dn_0": 0.73283, "loss_cls_dn_1": 0.10858, "loss_box_dn_1": 0.68205, "loss_cls_dn_2": 0.10983, "loss_box_dn_2": 0.67829, "loss_cls_dn_3": 0.11533, "loss_box_dn_3": 0.67025, "loss_cls_dn_4": 0.11273, "loss_box_dn_4": 0.66699, "loss_cls_dn_5": 0.11614, "loss_box_dn_5": 0.67059, "loss_dense_depth": 0.82007, "loss": 24.38309, "grad_norm": 35.47084, "time": 1.6798} -{"mode": "train", "epoch": 1, "iter": 282, "lr": 0.00021, "memory": 49164, "data_time": 0.20244, "loss_cls_0": 0.74583, "loss_box_0": 1.6477, "loss_cns_0": 0.6273, "loss_yns_0": 0.14706, "loss_cls_1": 0.83754, "loss_box_1": 1.49955, "loss_cns_1": 0.66014, "loss_yns_1": 0.14508, "loss_cls_2": 0.84964, "loss_box_2": 1.4637, "loss_cns_2": 0.6642, "loss_yns_2": 0.14532, "loss_cls_3": 0.84078, "loss_box_3": 1.47867, "loss_cns_3": 0.66028, "loss_yns_3": 0.14471, "loss_cls_4": 0.85403, "loss_box_4": 1.46431, "loss_cns_4": 0.65965, "loss_yns_4": 0.14599, "loss_cls_5": 0.85504, "loss_box_5": 1.46967, "loss_cns_5": 0.65812, "loss_yns_5": 0.1461, "loss_cls_dn_0": 0.16471, "loss_box_dn_0": 0.75423, "loss_cls_dn_1": 0.11303, "loss_box_dn_1": 0.6816, "loss_cls_dn_2": 0.11515, "loss_box_dn_2": 0.66942, "loss_cls_dn_3": 0.11919, "loss_box_dn_3": 0.67138, "loss_cls_dn_4": 0.117, "loss_box_dn_4": 0.67536, "loss_cls_dn_5": 0.11772, "loss_box_dn_5": 0.68537, "loss_dense_depth": 0.73867, "loss": 24.43324, "grad_norm": 45.34087, "time": 1.70178} -{"mode": "train", "epoch": 1, "iter": 283, "lr": 0.00021, "memory": 49164, "data_time": 0.06229, "loss_cls_0": 0.75373, "loss_box_0": 1.63598, "loss_cns_0": 0.62728, "loss_yns_0": 0.14686, "loss_cls_1": 0.82361, "loss_box_1": 1.48868, "loss_cns_1": 0.65712, "loss_yns_1": 0.14685, "loss_cls_2": 0.8397, "loss_box_2": 1.45938, "loss_cns_2": 0.6588, "loss_yns_2": 0.14447, "loss_cls_3": 0.84215, "loss_box_3": 1.47357, "loss_cns_3": 0.66512, "loss_yns_3": 0.145, "loss_cls_4": 0.85595, "loss_box_4": 1.45026, "loss_cns_4": 0.66026, "loss_yns_4": 0.14505, "loss_cls_5": 0.86886, "loss_box_5": 1.44743, "loss_cns_5": 0.66413, "loss_yns_5": 0.14503, "loss_cls_dn_0": 0.16003, "loss_box_dn_0": 0.74014, "loss_cls_dn_1": 0.11301, "loss_box_dn_1": 0.67469, "loss_cls_dn_2": 0.11079, "loss_box_dn_2": 0.66239, "loss_cls_dn_3": 0.1119, "loss_box_dn_3": 0.66827, "loss_cls_dn_4": 0.11383, "loss_box_dn_4": 0.67516, "loss_cls_dn_5": 0.1164, "loss_box_dn_5": 0.67953, "loss_dense_depth": 0.77742, "loss": 24.34881, "grad_norm": 41.21923, "time": 1.55316} -{"mode": "train", "epoch": 1, "iter": 284, "lr": 0.00021, "memory": 49164, "data_time": 0.07478, "loss_cls_0": 0.74821, "loss_box_0": 1.61248, "loss_cns_0": 0.62997, "loss_yns_0": 0.14847, "loss_cls_1": 0.81666, "loss_box_1": 1.54304, "loss_cns_1": 0.65255, "loss_yns_1": 0.14832, "loss_cls_2": 0.83988, "loss_box_2": 1.47917, "loss_cns_2": 0.65429, "loss_yns_2": 0.14704, "loss_cls_3": 0.83367, "loss_box_3": 1.48262, "loss_cns_3": 0.66499, "loss_yns_3": 0.14754, "loss_cls_4": 0.83723, "loss_box_4": 1.48966, "loss_cns_4": 0.65756, "loss_yns_4": 0.1471, "loss_cls_5": 0.86258, "loss_box_5": 1.46686, "loss_cns_5": 0.6605, "loss_yns_5": 0.14653, "loss_cls_dn_0": 0.15529, "loss_box_dn_0": 0.73691, "loss_cls_dn_1": 0.11066, "loss_box_dn_1": 0.68898, "loss_cls_dn_2": 0.11222, "loss_box_dn_2": 0.67174, "loss_cls_dn_3": 0.11367, "loss_box_dn_3": 0.6813, "loss_cls_dn_4": 0.11451, "loss_box_dn_4": 0.6871, "loss_cls_dn_5": 0.11865, "loss_box_dn_5": 0.69056, "loss_dense_depth": 0.76247, "loss": 24.461, "grad_norm": 44.4423, "time": 1.58811} -{"mode": "train", "epoch": 1, "iter": 285, "lr": 0.00021, "memory": 49164, "data_time": 0.10395, "loss_cls_0": 0.77618, "loss_box_0": 1.64558, "loss_cns_0": 0.63145, "loss_yns_0": 0.15096, "loss_cls_1": 0.82909, "loss_box_1": 1.54619, "loss_cns_1": 0.65227, "loss_yns_1": 0.14861, "loss_cls_2": 0.84551, "loss_box_2": 1.50936, "loss_cns_2": 0.65795, "loss_yns_2": 0.14929, "loss_cls_3": 0.84888, "loss_box_3": 1.5038, "loss_cns_3": 0.65945, "loss_yns_3": 0.14866, "loss_cls_4": 0.85912, "loss_box_4": 1.48263, "loss_cns_4": 0.65722, "loss_yns_4": 0.148, "loss_cls_5": 0.86348, "loss_box_5": 1.47783, "loss_cns_5": 0.65551, "loss_yns_5": 0.14852, "loss_cls_dn_0": 0.16803, "loss_box_dn_0": 0.74503, "loss_cls_dn_1": 0.1107, "loss_box_dn_1": 0.68679, "loss_cls_dn_2": 0.11168, "loss_box_dn_2": 0.67663, "loss_cls_dn_3": 0.11565, "loss_box_dn_3": 0.68708, "loss_cls_dn_4": 0.11553, "loss_box_dn_4": 0.68652, "loss_cls_dn_5": 0.11774, "loss_box_dn_5": 0.69775, "loss_dense_depth": 0.81522, "loss": 24.72987, "grad_norm": 43.21061, "time": 1.70849} -{"mode": "train", "epoch": 1, "iter": 286, "lr": 0.00021, "memory": 49164, "data_time": 0.1157, "loss_cls_0": 0.75209, "loss_box_0": 1.59814, "loss_cns_0": 0.62673, "loss_yns_0": 0.14936, "loss_cls_1": 0.83894, "loss_box_1": 1.50051, "loss_cns_1": 0.65836, "loss_yns_1": 0.14968, "loss_cls_2": 0.8513, "loss_box_2": 1.48524, "loss_cns_2": 0.66159, "loss_yns_2": 0.14861, "loss_cls_3": 0.85434, "loss_box_3": 1.48575, "loss_cns_3": 0.65622, "loss_yns_3": 0.14803, "loss_cls_4": 0.87372, "loss_box_4": 1.48204, "loss_cns_4": 0.65577, "loss_yns_4": 0.1478, "loss_cls_5": 0.87746, "loss_box_5": 1.48504, "loss_cns_5": 0.65581, "loss_yns_5": 0.14778, "loss_cls_dn_0": 0.16263, "loss_box_dn_0": 0.7297, "loss_cls_dn_1": 0.10974, "loss_box_dn_1": 0.67747, "loss_cls_dn_2": 0.10709, "loss_box_dn_2": 0.67633, "loss_cls_dn_3": 0.10839, "loss_box_dn_3": 0.68192, "loss_cls_dn_4": 0.11011, "loss_box_dn_4": 0.6861, "loss_cls_dn_5": 0.11285, "loss_box_dn_5": 0.69495, "loss_dense_depth": 0.75379, "loss": 24.50135, "grad_norm": 44.16928, "time": 1.70222} -{"mode": "train", "epoch": 1, "iter": 287, "lr": 0.00021, "memory": 49164, "data_time": 0.077, "loss_cls_0": 0.75974, "loss_box_0": 1.58516, "loss_cns_0": 0.62378, "loss_yns_0": 0.14914, "loss_cls_1": 0.85039, "loss_box_1": 1.53086, "loss_cns_1": 0.65986, "loss_yns_1": 0.15052, "loss_cls_2": 0.86352, "loss_box_2": 1.50661, "loss_cns_2": 0.66372, "loss_yns_2": 0.1496, "loss_cls_3": 0.86677, "loss_box_3": 1.4936, "loss_cns_3": 0.66245, "loss_yns_3": 0.14984, "loss_cls_4": 0.86955, "loss_box_4": 1.49942, "loss_cns_4": 0.65666, "loss_yns_4": 0.14936, "loss_cls_5": 0.88391, "loss_box_5": 1.49634, "loss_cns_5": 0.66429, "loss_yns_5": 0.15079, "loss_cls_dn_0": 0.15471, "loss_box_dn_0": 0.74237, "loss_cls_dn_1": 0.11291, "loss_box_dn_1": 0.68329, "loss_cls_dn_2": 0.11175, "loss_box_dn_2": 0.68005, "loss_cls_dn_3": 0.11429, "loss_box_dn_3": 0.67554, "loss_cls_dn_4": 0.11742, "loss_box_dn_4": 0.68341, "loss_cls_dn_5": 0.11849, "loss_box_dn_5": 0.68349, "loss_dense_depth": 0.7501, "loss": 24.6637, "grad_norm": 37.77774, "time": 1.55336} -{"mode": "train", "epoch": 1, "iter": 288, "lr": 0.00021, "memory": 49164, "data_time": 0.07544, "loss_cls_0": 0.78326, "loss_box_0": 1.57132, "loss_cns_0": 0.6166, "loss_yns_0": 0.14612, "loss_cls_1": 0.8408, "loss_box_1": 1.53366, "loss_cns_1": 0.65737, "loss_yns_1": 0.14974, "loss_cls_2": 0.84038, "loss_box_2": 1.52586, "loss_cns_2": 0.65684, "loss_yns_2": 0.14953, "loss_cls_3": 0.86162, "loss_box_3": 1.50623, "loss_cns_3": 0.66093, "loss_yns_3": 0.14876, "loss_cls_4": 0.87868, "loss_box_4": 1.47652, "loss_cns_4": 0.65722, "loss_yns_4": 0.14867, "loss_cls_5": 0.8706, "loss_box_5": 1.49884, "loss_cns_5": 0.66681, "loss_yns_5": 0.15133, "loss_cls_dn_0": 0.15697, "loss_box_dn_0": 0.74236, "loss_cls_dn_1": 0.11033, "loss_box_dn_1": 0.67888, "loss_cls_dn_2": 0.10985, "loss_box_dn_2": 0.67603, "loss_cls_dn_3": 0.11503, "loss_box_dn_3": 0.66509, "loss_cls_dn_4": 0.11426, "loss_box_dn_4": 0.65847, "loss_cls_dn_5": 0.11308, "loss_box_dn_5": 0.6618, "loss_dense_depth": 0.75587, "loss": 24.55571, "grad_norm": 45.25956, "time": 1.564} -{"mode": "train", "epoch": 1, "iter": 289, "lr": 0.00022, "memory": 49164, "data_time": 0.07503, "loss_cls_0": 0.73637, "loss_box_0": 1.57837, "loss_cns_0": 0.62911, "loss_yns_0": 0.14966, "loss_cls_1": 0.80337, "loss_box_1": 1.50271, "loss_cns_1": 0.65879, "loss_yns_1": 0.1514, "loss_cls_2": 0.81834, "loss_box_2": 1.49855, "loss_cns_2": 0.65988, "loss_yns_2": 0.15065, "loss_cls_3": 0.82924, "loss_box_3": 1.47657, "loss_cns_3": 0.65868, "loss_yns_3": 0.14835, "loss_cls_4": 0.8495, "loss_box_4": 1.43732, "loss_cns_4": 0.6574, "loss_yns_4": 0.1481, "loss_cls_5": 0.83831, "loss_box_5": 1.47213, "loss_cns_5": 0.66061, "loss_yns_5": 0.14788, "loss_cls_dn_0": 0.15729, "loss_box_dn_0": 0.72507, "loss_cls_dn_1": 0.10839, "loss_box_dn_1": 0.66062, "loss_cls_dn_2": 0.10835, "loss_box_dn_2": 0.65489, "loss_cls_dn_3": 0.10884, "loss_box_dn_3": 0.65208, "loss_cls_dn_4": 0.10832, "loss_box_dn_4": 0.64733, "loss_cls_dn_5": 0.10974, "loss_box_dn_5": 0.65336, "loss_dense_depth": 0.72675, "loss": 24.08229, "grad_norm": 38.80302, "time": 1.58759} -{"mode": "train", "epoch": 1, "iter": 290, "lr": 0.00022, "memory": 49164, "data_time": 0.07318, "loss_cls_0": 0.76484, "loss_box_0": 1.64372, "loss_cns_0": 0.6268, "loss_yns_0": 0.15029, "loss_cls_1": 0.83047, "loss_box_1": 1.51948, "loss_cns_1": 0.65716, "loss_yns_1": 0.15004, "loss_cls_2": 0.84899, "loss_box_2": 1.48537, "loss_cns_2": 0.66163, "loss_yns_2": 0.15004, "loss_cls_3": 0.85354, "loss_box_3": 1.46481, "loss_cns_3": 0.65566, "loss_yns_3": 0.14945, "loss_cls_4": 0.85952, "loss_box_4": 1.49125, "loss_cns_4": 0.65683, "loss_yns_4": 0.14885, "loss_cls_5": 0.86177, "loss_box_5": 1.45598, "loss_cns_5": 0.65602, "loss_yns_5": 0.14777, "loss_cls_dn_0": 0.15477, "loss_box_dn_0": 0.73876, "loss_cls_dn_1": 0.11099, "loss_box_dn_1": 0.67327, "loss_cls_dn_2": 0.11061, "loss_box_dn_2": 0.6582, "loss_cls_dn_3": 0.11298, "loss_box_dn_3": 0.65624, "loss_cls_dn_4": 0.11332, "loss_box_dn_4": 0.67659, "loss_cls_dn_5": 0.11795, "loss_box_dn_5": 0.66874, "loss_dense_depth": 0.75793, "loss": 24.44065, "grad_norm": 43.65224, "time": 1.62089} -{"mode": "train", "epoch": 1, "iter": 291, "lr": 0.00022, "memory": 49164, "data_time": 0.07674, "loss_cls_0": 0.7896, "loss_box_0": 1.65418, "loss_cns_0": 0.63417, "loss_yns_0": 0.15461, "loss_cls_1": 0.8365, "loss_box_1": 1.53198, "loss_cns_1": 0.65877, "loss_yns_1": 0.1521, "loss_cls_2": 0.84862, "loss_box_2": 1.50124, "loss_cns_2": 0.66811, "loss_yns_2": 0.1522, "loss_cls_3": 0.86107, "loss_box_3": 1.4815, "loss_cns_3": 0.66414, "loss_yns_3": 0.15283, "loss_cls_4": 0.87073, "loss_box_4": 1.50318, "loss_cns_4": 0.6641, "loss_yns_4": 0.15183, "loss_cls_5": 0.87743, "loss_box_5": 1.4608, "loss_cns_5": 0.66307, "loss_yns_5": 0.15173, "loss_cls_dn_0": 0.15179, "loss_box_dn_0": 0.73705, "loss_cls_dn_1": 0.10497, "loss_box_dn_1": 0.67976, "loss_cls_dn_2": 0.1076, "loss_box_dn_2": 0.6682, "loss_cls_dn_3": 0.10969, "loss_box_dn_3": 0.6675, "loss_cls_dn_4": 0.11101, "loss_box_dn_4": 0.68398, "loss_cls_dn_5": 0.11882, "loss_box_dn_5": 0.67889, "loss_dense_depth": 0.76783, "loss": 24.67159, "grad_norm": 41.92731, "time": 1.58241} -{"mode": "train", "epoch": 1, "iter": 292, "lr": 0.00022, "memory": 49164, "data_time": 0.07768, "loss_cls_0": 0.7408, "loss_box_0": 1.63941, "loss_cns_0": 0.63231, "loss_yns_0": 0.15002, "loss_cls_1": 0.80806, "loss_box_1": 1.53942, "loss_cns_1": 0.65577, "loss_yns_1": 0.14884, "loss_cls_2": 0.82406, "loss_box_2": 1.50993, "loss_cns_2": 0.66382, "loss_yns_2": 0.15039, "loss_cls_3": 0.83307, "loss_box_3": 1.49708, "loss_cns_3": 0.66574, "loss_yns_3": 0.15105, "loss_cls_4": 0.84323, "loss_box_4": 1.48675, "loss_cns_4": 0.66699, "loss_yns_4": 0.15123, "loss_cls_5": 0.84996, "loss_box_5": 1.47922, "loss_cns_5": 0.66779, "loss_yns_5": 0.14927, "loss_cls_dn_0": 0.15119, "loss_box_dn_0": 0.73548, "loss_cls_dn_1": 0.10864, "loss_box_dn_1": 0.66885, "loss_cls_dn_2": 0.10899, "loss_box_dn_2": 0.6631, "loss_cls_dn_3": 0.10963, "loss_box_dn_3": 0.66405, "loss_cls_dn_4": 0.11323, "loss_box_dn_4": 0.6664, "loss_cls_dn_5": 0.1148, "loss_box_dn_5": 0.67001, "loss_dense_depth": 0.72348, "loss": 24.40204, "grad_norm": 38.75624, "time": 1.57784} -{"mode": "train", "epoch": 1, "iter": 293, "lr": 0.00022, "memory": 49164, "data_time": 0.13958, "loss_cls_0": 0.75402, "loss_box_0": 1.67392, "loss_cns_0": 0.62974, "loss_yns_0": 0.14766, "loss_cls_1": 0.82984, "loss_box_1": 1.51013, "loss_cns_1": 0.66121, "loss_yns_1": 0.14384, "loss_cls_2": 0.83137, "loss_box_2": 1.48335, "loss_cns_2": 0.66137, "loss_yns_2": 0.14479, "loss_cls_3": 0.84263, "loss_box_3": 1.4631, "loss_cns_3": 0.66421, "loss_yns_3": 0.14523, "loss_cls_4": 0.85637, "loss_box_4": 1.47413, "loss_cns_4": 0.6628, "loss_yns_4": 0.14433, "loss_cls_5": 0.87611, "loss_box_5": 1.45779, "loss_cns_5": 0.66427, "loss_yns_5": 0.14418, "loss_cls_dn_0": 0.15161, "loss_box_dn_0": 0.732, "loss_cls_dn_1": 0.1071, "loss_box_dn_1": 0.67181, "loss_cls_dn_2": 0.10527, "loss_box_dn_2": 0.66276, "loss_cls_dn_3": 0.10613, "loss_box_dn_3": 0.6591, "loss_cls_dn_4": 0.11124, "loss_box_dn_4": 0.66771, "loss_cls_dn_5": 0.11159, "loss_box_dn_5": 0.66748, "loss_dense_depth": 0.76122, "loss": 24.38141, "grad_norm": 29.80162, "time": 1.64028} -{"mode": "train", "epoch": 1, "iter": 294, "lr": 0.00022, "memory": 49164, "data_time": 0.06913, "loss_cls_0": 0.76413, "loss_box_0": 1.65596, "loss_cns_0": 0.62679, "loss_yns_0": 0.14785, "loss_cls_1": 0.82064, "loss_box_1": 1.50065, "loss_cns_1": 0.66402, "loss_yns_1": 0.14431, "loss_cls_2": 0.83737, "loss_box_2": 1.47848, "loss_cns_2": 0.66906, "loss_yns_2": 0.1435, "loss_cls_3": 0.85096, "loss_box_3": 1.45926, "loss_cns_3": 0.66222, "loss_yns_3": 0.1438, "loss_cls_4": 0.86003, "loss_box_4": 1.46253, "loss_cns_4": 0.6636, "loss_yns_4": 0.14447, "loss_cls_5": 0.87505, "loss_box_5": 1.45357, "loss_cns_5": 0.66377, "loss_yns_5": 0.14455, "loss_cls_dn_0": 0.14684, "loss_box_dn_0": 0.73285, "loss_cls_dn_1": 0.10762, "loss_box_dn_1": 0.66655, "loss_cls_dn_2": 0.10749, "loss_box_dn_2": 0.65649, "loss_cls_dn_3": 0.10935, "loss_box_dn_3": 0.65244, "loss_cls_dn_4": 0.10862, "loss_box_dn_4": 0.65542, "loss_cls_dn_5": 0.11384, "loss_box_dn_5": 0.65479, "loss_dense_depth": 0.74455, "loss": 24.2934, "grad_norm": 32.14201, "time": 1.56625} -{"mode": "train", "epoch": 1, "iter": 295, "lr": 0.00022, "memory": 49164, "data_time": 0.07188, "loss_cls_0": 0.77101, "loss_box_0": 1.62877, "loss_cns_0": 0.62717, "loss_yns_0": 0.14913, "loss_cls_1": 0.84775, "loss_box_1": 1.49142, "loss_cns_1": 0.66203, "loss_yns_1": 0.1468, "loss_cls_2": 0.85753, "loss_box_2": 1.46366, "loss_cns_2": 0.67848, "loss_yns_2": 0.14576, "loss_cls_3": 0.85849, "loss_box_3": 1.44895, "loss_cns_3": 0.6653, "loss_yns_3": 0.14711, "loss_cls_4": 0.87457, "loss_box_4": 1.44564, "loss_cns_4": 0.6628, "loss_yns_4": 0.14647, "loss_cls_5": 0.87599, "loss_box_5": 1.43817, "loss_cns_5": 0.66678, "loss_yns_5": 0.14772, "loss_cls_dn_0": 0.14601, "loss_box_dn_0": 0.73046, "loss_cls_dn_1": 0.10769, "loss_box_dn_1": 0.65698, "loss_cls_dn_2": 0.10899, "loss_box_dn_2": 0.64011, "loss_cls_dn_3": 0.11147, "loss_box_dn_3": 0.63677, "loss_cls_dn_4": 0.11701, "loss_box_dn_4": 0.63903, "loss_cls_dn_5": 0.11724, "loss_box_dn_5": 0.63526, "loss_dense_depth": 0.75134, "loss": 24.24585, "grad_norm": 22.12576, "time": 1.58059} -{"mode": "train", "epoch": 1, "iter": 296, "lr": 0.00022, "memory": 49164, "data_time": 0.07496, "loss_cls_0": 0.75731, "loss_box_0": 1.64653, "loss_cns_0": 0.62805, "loss_yns_0": 0.14791, "loss_cls_1": 0.84146, "loss_box_1": 1.51087, "loss_cns_1": 0.65746, "loss_yns_1": 0.14765, "loss_cls_2": 0.8484, "loss_box_2": 1.48554, "loss_cns_2": 0.6599, "loss_yns_2": 0.14747, "loss_cls_3": 0.85305, "loss_box_3": 1.4763, "loss_cns_3": 0.66356, "loss_yns_3": 0.14787, "loss_cls_4": 0.87506, "loss_box_4": 1.46288, "loss_cns_4": 0.66081, "loss_yns_4": 0.14746, "loss_cls_5": 0.86189, "loss_box_5": 1.46183, "loss_cns_5": 0.6615, "loss_yns_5": 0.14714, "loss_cls_dn_0": 0.15115, "loss_box_dn_0": 0.73051, "loss_cls_dn_1": 0.10481, "loss_box_dn_1": 0.65366, "loss_cls_dn_2": 0.10765, "loss_box_dn_2": 0.64008, "loss_cls_dn_3": 0.11019, "loss_box_dn_3": 0.64172, "loss_cls_dn_4": 0.11853, "loss_box_dn_4": 0.64475, "loss_cls_dn_5": 0.11108, "loss_box_dn_5": 0.64659, "loss_dense_depth": 0.75025, "loss": 24.30891, "grad_norm": 29.74428, "time": 1.61556} -{"mode": "train", "epoch": 1, "iter": 297, "lr": 0.00022, "memory": 49164, "data_time": 0.07319, "loss_cls_0": 0.79783, "loss_box_0": 1.69528, "loss_cns_0": 0.62842, "loss_yns_0": 0.14962, "loss_cls_1": 0.84398, "loss_box_1": 1.57481, "loss_cns_1": 0.65632, "loss_yns_1": 0.15011, "loss_cls_2": 0.85822, "loss_box_2": 1.53903, "loss_cns_2": 0.66122, "loss_yns_2": 0.14882, "loss_cls_3": 0.86859, "loss_box_3": 1.54127, "loss_cns_3": 0.65691, "loss_yns_3": 0.14842, "loss_cls_4": 0.8777, "loss_box_4": 1.53809, "loss_cns_4": 0.65786, "loss_yns_4": 0.149, "loss_cls_5": 0.87463, "loss_box_5": 1.53192, "loss_cns_5": 0.6564, "loss_yns_5": 0.14839, "loss_cls_dn_0": 0.1569, "loss_box_dn_0": 0.73668, "loss_cls_dn_1": 0.10703, "loss_box_dn_1": 0.65528, "loss_cls_dn_2": 0.1099, "loss_box_dn_2": 0.64788, "loss_cls_dn_3": 0.11065, "loss_box_dn_3": 0.65212, "loss_cls_dn_4": 0.11293, "loss_box_dn_4": 0.65928, "loss_cls_dn_5": 0.11475, "loss_box_dn_5": 0.6716, "loss_dense_depth": 0.74279, "loss": 24.83065, "grad_norm": 41.46824, "time": 1.55945} -{"mode": "train", "epoch": 1, "iter": 298, "lr": 0.00022, "memory": 49164, "data_time": 0.07585, "loss_cls_0": 0.79334, "loss_box_0": 1.64502, "loss_cns_0": 0.63391, "loss_yns_0": 0.15143, "loss_cls_1": 0.85134, "loss_box_1": 1.56942, "loss_cns_1": 0.65508, "loss_yns_1": 0.14775, "loss_cls_2": 0.86671, "loss_box_2": 1.53628, "loss_cns_2": 0.66769, "loss_yns_2": 0.14844, "loss_cls_3": 0.89019, "loss_box_3": 1.53894, "loss_cns_3": 0.65944, "loss_yns_3": 0.14806, "loss_cls_4": 0.88538, "loss_box_4": 1.524, "loss_cns_4": 0.65909, "loss_yns_4": 0.14845, "loss_cls_5": 0.87671, "loss_box_5": 1.52302, "loss_cns_5": 0.65952, "loss_yns_5": 0.14866, "loss_cls_dn_0": 0.14866, "loss_box_dn_0": 0.72759, "loss_cls_dn_1": 0.10974, "loss_box_dn_1": 0.66322, "loss_cls_dn_2": 0.11222, "loss_box_dn_2": 0.65736, "loss_cls_dn_3": 0.11122, "loss_box_dn_3": 0.66304, "loss_cls_dn_4": 0.11571, "loss_box_dn_4": 0.66941, "loss_cls_dn_5": 0.11756, "loss_box_dn_5": 0.68431, "loss_dense_depth": 0.74907, "loss": 24.85697, "grad_norm": 27.49586, "time": 1.60405} -{"mode": "train", "epoch": 1, "iter": 299, "lr": 0.00022, "memory": 49164, "data_time": 0.0902, "loss_cls_0": 0.78852, "loss_box_0": 1.65594, "loss_cns_0": 0.62489, "loss_yns_0": 0.15297, "loss_cls_1": 0.87914, "loss_box_1": 1.5536, "loss_cns_1": 0.64651, "loss_yns_1": 0.14677, "loss_cls_2": 0.89644, "loss_box_2": 1.50562, "loss_cns_2": 0.65627, "loss_yns_2": 0.14789, "loss_cls_3": 0.87959, "loss_box_3": 1.50511, "loss_cns_3": 0.65656, "loss_yns_3": 0.1483, "loss_cls_4": 0.86889, "loss_box_4": 1.51315, "loss_cns_4": 0.65992, "loss_yns_4": 0.15106, "loss_cls_5": 0.89646, "loss_box_5": 1.5173, "loss_cns_5": 0.66026, "loss_yns_5": 0.1505, "loss_cls_dn_0": 0.15643, "loss_box_dn_0": 0.73677, "loss_cls_dn_1": 0.10741, "loss_box_dn_1": 0.67018, "loss_cls_dn_2": 0.10755, "loss_box_dn_2": 0.66426, "loss_cls_dn_3": 0.10769, "loss_box_dn_3": 0.66547, "loss_cls_dn_4": 0.11251, "loss_box_dn_4": 0.67424, "loss_cls_dn_5": 0.11315, "loss_box_dn_5": 0.6869, "loss_dense_depth": 0.76727, "loss": 24.83149, "grad_norm": 48.1803, "time": 1.58283} -{"mode": "train", "epoch": 1, "iter": 300, "lr": 0.00022, "memory": 49164, "data_time": 0.0801, "loss_cls_0": 0.80911, "loss_box_0": 1.63732, "loss_cns_0": 0.62547, "loss_yns_0": 0.15134, "loss_cls_1": 0.87301, "loss_box_1": 1.53359, "loss_cns_1": 0.64495, "loss_yns_1": 0.14489, "loss_cls_2": 0.88425, "loss_box_2": 1.49908, "loss_cns_2": 0.65026, "loss_yns_2": 0.14674, "loss_cls_3": 0.88119, "loss_box_3": 1.47945, "loss_cns_3": 0.65361, "loss_yns_3": 0.14641, "loss_cls_4": 0.86263, "loss_box_4": 1.49711, "loss_cns_4": 0.65836, "loss_yns_4": 0.14692, "loss_cls_5": 0.86917, "loss_box_5": 1.49891, "loss_cns_5": 0.6604, "loss_yns_5": 0.14848, "loss_cls_dn_0": 0.17572, "loss_box_dn_0": 0.73695, "loss_cls_dn_1": 0.10988, "loss_box_dn_1": 0.67656, "loss_cls_dn_2": 0.11039, "loss_box_dn_2": 0.66404, "loss_cls_dn_3": 0.11234, "loss_box_dn_3": 0.65719, "loss_cls_dn_4": 0.11553, "loss_box_dn_4": 0.65729, "loss_cls_dn_5": 0.11493, "loss_box_dn_5": 0.65997, "loss_dense_depth": 0.78164, "loss": 24.67508, "grad_norm": 32.51467, "time": 1.59873} -{"mode": "train", "epoch": 1, "iter": 301, "lr": 0.00022, "memory": 49164, "data_time": 0.10712, "loss_cls_0": 0.79909, "loss_box_0": 1.63753, "loss_cns_0": 0.62888, "loss_yns_0": 0.14698, "loss_cls_1": 0.86759, "loss_box_1": 1.58404, "loss_cns_1": 0.65457, "loss_yns_1": 0.14596, "loss_cls_2": 0.88158, "loss_box_2": 1.55368, "loss_cns_2": 0.65529, "loss_yns_2": 0.14799, "loss_cls_3": 0.8846, "loss_box_3": 1.54952, "loss_cns_3": 0.65704, "loss_yns_3": 0.14625, "loss_cls_4": 0.87776, "loss_box_4": 1.56279, "loss_cns_4": 0.65547, "loss_yns_4": 0.14774, "loss_cls_5": 0.88362, "loss_box_5": 1.54293, "loss_cns_5": 0.65438, "loss_yns_5": 0.14762, "loss_cls_dn_0": 0.15215, "loss_box_dn_0": 0.73259, "loss_cls_dn_1": 0.10927, "loss_box_dn_1": 0.65925, "loss_cls_dn_2": 0.11058, "loss_box_dn_2": 0.65214, "loss_cls_dn_3": 0.1091, "loss_box_dn_3": 0.65006, "loss_cls_dn_4": 0.11123, "loss_box_dn_4": 0.65227, "loss_cls_dn_5": 0.11299, "loss_box_dn_5": 0.64637, "loss_dense_depth": 0.78278, "loss": 24.89369, "grad_norm": 45.09535, "time": 1.65555} -{"mode": "train", "epoch": 1, "iter": 302, "lr": 0.00022, "memory": 49164, "data_time": 0.1646, "loss_cls_0": 0.78208, "loss_box_0": 1.6452, "loss_cns_0": 0.6302, "loss_yns_0": 0.15115, "loss_cls_1": 0.86258, "loss_box_1": 1.56579, "loss_cns_1": 0.65568, "loss_yns_1": 0.1508, "loss_cls_2": 0.8848, "loss_box_2": 1.52801, "loss_cns_2": 0.65807, "loss_yns_2": 0.15107, "loss_cls_3": 0.87322, "loss_box_3": 1.5193, "loss_cns_3": 0.66014, "loss_yns_3": 0.15023, "loss_cls_4": 0.87207, "loss_box_4": 1.52956, "loss_cns_4": 0.66066, "loss_yns_4": 0.15083, "loss_cls_5": 0.89479, "loss_box_5": 1.52098, "loss_cns_5": 0.66043, "loss_yns_5": 0.15154, "loss_cls_dn_0": 0.15322, "loss_box_dn_0": 0.72303, "loss_cls_dn_1": 0.1105, "loss_box_dn_1": 0.65192, "loss_cls_dn_2": 0.11295, "loss_box_dn_2": 0.64282, "loss_cls_dn_3": 0.11069, "loss_box_dn_3": 0.64183, "loss_cls_dn_4": 0.11253, "loss_box_dn_4": 0.64842, "loss_cls_dn_5": 0.1159, "loss_box_dn_5": 0.64458, "loss_dense_depth": 0.81921, "loss": 24.79677, "grad_norm": 46.9765, "time": 1.6817} -{"mode": "train", "epoch": 1, "iter": 303, "lr": 0.00022, "memory": 49164, "data_time": 0.07105, "loss_cls_0": 0.81825, "loss_box_0": 1.70162, "loss_cns_0": 0.62575, "loss_yns_0": 0.15631, "loss_cls_1": 0.90909, "loss_box_1": 1.57908, "loss_cns_1": 0.66119, "loss_yns_1": 0.15472, "loss_cls_2": 0.90558, "loss_box_2": 1.56495, "loss_cns_2": 0.6662, "loss_yns_2": 0.15421, "loss_cls_3": 0.89974, "loss_box_3": 1.56857, "loss_cns_3": 0.66218, "loss_yns_3": 0.15552, "loss_cls_4": 0.91409, "loss_box_4": 1.56174, "loss_cns_4": 0.68253, "loss_yns_4": 0.15618, "loss_cls_5": 0.90926, "loss_box_5": 1.5638, "loss_cns_5": 0.66606, "loss_yns_5": 0.15548, "loss_cls_dn_0": 0.18674, "loss_box_dn_0": 0.74586, "loss_cls_dn_1": 0.111, "loss_box_dn_1": 0.66206, "loss_cls_dn_2": 0.11137, "loss_box_dn_2": 0.65569, "loss_cls_dn_3": 0.11406, "loss_box_dn_3": 0.658, "loss_cls_dn_4": 0.1122, "loss_box_dn_4": 0.66191, "loss_cls_dn_5": 0.11214, "loss_box_dn_5": 0.665, "loss_dense_depth": 0.7749, "loss": 25.363, "grad_norm": 46.21587, "time": 1.55228} -{"mode": "train", "epoch": 1, "iter": 304, "lr": 0.00022, "memory": 49164, "data_time": 0.07439, "loss_cls_0": 0.77322, "loss_box_0": 1.66226, "loss_cns_0": 0.62881, "loss_yns_0": 0.15327, "loss_cls_1": 0.86727, "loss_box_1": 1.51651, "loss_cns_1": 0.66217, "loss_yns_1": 0.15202, "loss_cls_2": 0.86308, "loss_box_2": 1.51759, "loss_cns_2": 0.66967, "loss_yns_2": 0.15193, "loss_cls_3": 0.86103, "loss_box_3": 1.5308, "loss_cns_3": 0.66135, "loss_yns_3": 0.15029, "loss_cls_4": 0.86868, "loss_box_4": 1.52756, "loss_cns_4": 0.66708, "loss_yns_4": 0.15236, "loss_cls_5": 0.85538, "loss_box_5": 1.52782, "loss_cns_5": 0.66108, "loss_yns_5": 0.15299, "loss_cls_dn_0": 0.16072, "loss_box_dn_0": 0.73358, "loss_cls_dn_1": 0.10959, "loss_box_dn_1": 0.65875, "loss_cls_dn_2": 0.10968, "loss_box_dn_2": 0.659, "loss_cls_dn_3": 0.11433, "loss_box_dn_3": 0.66908, "loss_cls_dn_4": 0.11318, "loss_box_dn_4": 0.67859, "loss_cls_dn_5": 0.11399, "loss_box_dn_5": 0.68782, "loss_dense_depth": 0.78537, "loss": 24.82789, "grad_norm": 43.13141, "time": 1.61445} -{"mode": "train", "epoch": 1, "iter": 305, "lr": 0.00022, "memory": 49164, "data_time": 0.07328, "loss_cls_0": 0.74368, "loss_box_0": 1.67038, "loss_cns_0": 0.62694, "loss_yns_0": 0.15155, "loss_cls_1": 0.82587, "loss_box_1": 1.53157, "loss_cns_1": 0.66068, "loss_yns_1": 0.15262, "loss_cls_2": 0.83451, "loss_box_2": 1.51854, "loss_cns_2": 0.66885, "loss_yns_2": 0.15037, "loss_cls_3": 0.84719, "loss_box_3": 1.47592, "loss_cns_3": 0.65905, "loss_yns_3": 0.14894, "loss_cls_4": 0.8531, "loss_box_4": 1.45447, "loss_cns_4": 0.65627, "loss_yns_4": 0.14917, "loss_cls_5": 0.84505, "loss_box_5": 1.48814, "loss_cns_5": 0.65774, "loss_yns_5": 0.15199, "loss_cls_dn_0": 0.14471, "loss_box_dn_0": 0.73585, "loss_cls_dn_1": 0.1085, "loss_box_dn_1": 0.68192, "loss_cls_dn_2": 0.10671, "loss_box_dn_2": 0.68409, "loss_cls_dn_3": 0.10856, "loss_box_dn_3": 0.69092, "loss_cls_dn_4": 0.11327, "loss_box_dn_4": 0.69772, "loss_cls_dn_5": 0.11332, "loss_box_dn_5": 0.71175, "loss_dense_depth": 0.72562, "loss": 24.54553, "grad_norm": 50.84257, "time": 1.62863} -{"mode": "train", "epoch": 1, "iter": 306, "lr": 0.00022, "memory": 49164, "data_time": 0.07851, "loss_cls_0": 0.77008, "loss_box_0": 1.65089, "loss_cns_0": 0.62247, "loss_yns_0": 0.14872, "loss_cls_1": 0.84613, "loss_box_1": 1.51381, "loss_cns_1": 0.65655, "loss_yns_1": 0.15177, "loss_cls_2": 0.85193, "loss_box_2": 1.48911, "loss_cns_2": 0.66025, "loss_yns_2": 0.14853, "loss_cls_3": 0.85793, "loss_box_3": 1.46028, "loss_cns_3": 0.65558, "loss_yns_3": 0.14783, "loss_cls_4": 0.85717, "loss_box_4": 1.439, "loss_cns_4": 0.65877, "loss_yns_4": 0.15044, "loss_cls_5": 0.86439, "loss_box_5": 1.43983, "loss_cns_5": 0.65153, "loss_yns_5": 0.14861, "loss_cls_dn_0": 0.16015, "loss_box_dn_0": 0.73344, "loss_cls_dn_1": 0.11469, "loss_box_dn_1": 0.68986, "loss_cls_dn_2": 0.11267, "loss_box_dn_2": 0.68284, "loss_cls_dn_3": 0.11252, "loss_box_dn_3": 0.68568, "loss_cls_dn_4": 0.11835, "loss_box_dn_4": 0.68604, "loss_cls_dn_5": 0.11862, "loss_box_dn_5": 0.69664, "loss_dense_depth": 0.79227, "loss": 24.54539, "grad_norm": 38.13459, "time": 1.65999} -{"mode": "train", "epoch": 1, "iter": 307, "lr": 0.00022, "memory": 49164, "data_time": 0.07824, "loss_cls_0": 0.75416, "loss_box_0": 1.6392, "loss_cns_0": 0.62888, "loss_yns_0": 0.14873, "loss_cls_1": 0.8379, "loss_box_1": 1.5099, "loss_cns_1": 0.65643, "loss_yns_1": 0.15104, "loss_cls_2": 0.83497, "loss_box_2": 1.50751, "loss_cns_2": 0.65893, "loss_yns_2": 0.15157, "loss_cls_3": 0.8416, "loss_box_3": 1.4963, "loss_cns_3": 0.66281, "loss_yns_3": 0.15058, "loss_cls_4": 0.84679, "loss_box_4": 1.49344, "loss_cns_4": 0.66177, "loss_yns_4": 0.1524, "loss_cls_5": 0.84383, "loss_box_5": 1.47794, "loss_cns_5": 0.65835, "loss_yns_5": 0.15041, "loss_cls_dn_0": 0.16497, "loss_box_dn_0": 0.72926, "loss_cls_dn_1": 0.11179, "loss_box_dn_1": 0.67624, "loss_cls_dn_2": 0.10907, "loss_box_dn_2": 0.67009, "loss_cls_dn_3": 0.10919, "loss_box_dn_3": 0.66564, "loss_cls_dn_4": 0.11204, "loss_box_dn_4": 0.67031, "loss_cls_dn_5": 0.11295, "loss_box_dn_5": 0.67669, "loss_dense_depth": 0.72739, "loss": 24.45107, "grad_norm": 42.03093, "time": 1.58711} -{"mode": "train", "epoch": 1, "iter": 308, "lr": 0.00022, "memory": 49164, "data_time": 0.11152, "loss_cls_0": 0.76962, "loss_box_0": 1.62056, "loss_cns_0": 0.63098, "loss_yns_0": 0.14808, "loss_cls_1": 0.84031, "loss_box_1": 1.53369, "loss_cns_1": 0.65263, "loss_yns_1": 0.14684, "loss_cls_2": 0.83625, "loss_box_2": 1.50946, "loss_cns_2": 0.6621, "loss_yns_2": 0.14785, "loss_cls_3": 0.85651, "loss_box_3": 1.5309, "loss_cns_3": 0.66609, "loss_yns_3": 0.14902, "loss_cls_4": 0.86027, "loss_box_4": 1.51044, "loss_cns_4": 0.66302, "loss_yns_4": 0.14867, "loss_cls_5": 0.85053, "loss_box_5": 1.49997, "loss_cns_5": 0.66116, "loss_yns_5": 0.14885, "loss_cls_dn_0": 0.16043, "loss_box_dn_0": 0.73551, "loss_cls_dn_1": 0.11339, "loss_box_dn_1": 0.67622, "loss_cls_dn_2": 0.11157, "loss_box_dn_2": 0.6653, "loss_cls_dn_3": 0.11462, "loss_box_dn_3": 0.66907, "loss_cls_dn_4": 0.11851, "loss_box_dn_4": 0.6653, "loss_cls_dn_5": 0.1188, "loss_box_dn_5": 0.66857, "loss_dense_depth": 0.79043, "loss": 24.6515, "grad_norm": 40.15565, "time": 1.59372} -{"mode": "train", "epoch": 1, "iter": 309, "lr": 0.00022, "memory": 49164, "data_time": 0.07313, "loss_cls_0": 0.7682, "loss_box_0": 1.62714, "loss_cns_0": 0.62758, "loss_yns_0": 0.14992, "loss_cls_1": 0.82787, "loss_box_1": 1.54463, "loss_cns_1": 0.65925, "loss_yns_1": 0.14737, "loss_cls_2": 0.83655, "loss_box_2": 1.53364, "loss_cns_2": 0.66884, "loss_yns_2": 0.14851, "loss_cls_3": 0.85592, "loss_box_3": 1.55247, "loss_cns_3": 0.66117, "loss_yns_3": 0.14777, "loss_cls_4": 0.86013, "loss_box_4": 1.51077, "loss_cns_4": 0.66189, "loss_yns_4": 0.14799, "loss_cls_5": 0.86158, "loss_box_5": 1.49442, "loss_cns_5": 0.65871, "loss_yns_5": 0.14876, "loss_cls_dn_0": 0.15219, "loss_box_dn_0": 0.72803, "loss_cls_dn_1": 0.11491, "loss_box_dn_1": 0.65869, "loss_cls_dn_2": 0.11595, "loss_box_dn_2": 0.65318, "loss_cls_dn_3": 0.12003, "loss_box_dn_3": 0.66747, "loss_cls_dn_4": 0.12449, "loss_box_dn_4": 0.65655, "loss_cls_dn_5": 0.12164, "loss_box_dn_5": 0.65658, "loss_dense_depth": 0.7221, "loss": 24.59287, "grad_norm": 43.01013, "time": 1.57679} -{"mode": "train", "epoch": 1, "iter": 310, "lr": 0.00022, "memory": 49164, "data_time": 0.10632, "loss_cls_0": 0.7485, "loss_box_0": 1.62443, "loss_cns_0": 0.63412, "loss_yns_0": 0.15133, "loss_cls_1": 0.81401, "loss_box_1": 1.52095, "loss_cns_1": 0.66266, "loss_yns_1": 0.14772, "loss_cls_2": 0.83647, "loss_box_2": 1.49143, "loss_cns_2": 0.66666, "loss_yns_2": 0.14698, "loss_cls_3": 0.85054, "loss_box_3": 1.5011, "loss_cns_3": 0.66319, "loss_yns_3": 0.1468, "loss_cls_4": 0.85533, "loss_box_4": 1.492, "loss_cns_4": 0.66521, "loss_yns_4": 0.14684, "loss_cls_5": 0.86433, "loss_box_5": 1.47217, "loss_cns_5": 0.66646, "loss_yns_5": 0.1467, "loss_cls_dn_0": 0.14879, "loss_box_dn_0": 0.7313, "loss_cls_dn_1": 0.11296, "loss_box_dn_1": 0.65355, "loss_cls_dn_2": 0.1131, "loss_box_dn_2": 0.64352, "loss_cls_dn_3": 0.11389, "loss_box_dn_3": 0.65658, "loss_cls_dn_4": 0.11744, "loss_box_dn_4": 0.66317, "loss_cls_dn_5": 0.11592, "loss_box_dn_5": 0.666, "loss_dense_depth": 0.77542, "loss": 24.42757, "grad_norm": 36.65202, "time": 1.63803} -{"mode": "train", "epoch": 1, "iter": 311, "lr": 0.00022, "memory": 49164, "data_time": 0.09947, "loss_cls_0": 0.75953, "loss_box_0": 1.6295, "loss_cns_0": 0.62966, "loss_yns_0": 0.15094, "loss_cls_1": 0.82241, "loss_box_1": 1.53697, "loss_cns_1": 0.6596, "loss_yns_1": 0.14563, "loss_cls_2": 0.84911, "loss_box_2": 1.50491, "loss_cns_2": 0.65962, "loss_yns_2": 0.1459, "loss_cls_3": 0.87477, "loss_box_3": 1.49511, "loss_cns_3": 0.66171, "loss_yns_3": 0.14747, "loss_cls_4": 0.86954, "loss_box_4": 1.51317, "loss_cns_4": 0.66155, "loss_yns_4": 0.14611, "loss_cls_5": 0.86921, "loss_box_5": 1.53719, "loss_cns_5": 0.66105, "loss_yns_5": 0.14739, "loss_cls_dn_0": 0.15547, "loss_box_dn_0": 0.73596, "loss_cls_dn_1": 0.11646, "loss_box_dn_1": 0.66281, "loss_cls_dn_2": 0.11352, "loss_box_dn_2": 0.65673, "loss_cls_dn_3": 0.11497, "loss_box_dn_3": 0.65939, "loss_cls_dn_4": 0.11699, "loss_box_dn_4": 0.67795, "loss_cls_dn_5": 0.11963, "loss_box_dn_5": 0.69631, "loss_dense_depth": 0.74612, "loss": 24.65037, "grad_norm": 45.33052, "time": 1.58956} -{"mode": "train", "epoch": 1, "iter": 312, "lr": 0.00022, "memory": 49164, "data_time": 0.10339, "loss_cls_0": 0.77923, "loss_box_0": 1.63689, "loss_cns_0": 0.62762, "loss_yns_0": 0.14877, "loss_cls_1": 0.83284, "loss_box_1": 1.5585, "loss_cns_1": 0.65648, "loss_yns_1": 0.14687, "loss_cls_2": 0.84934, "loss_box_2": 1.52176, "loss_cns_2": 0.65877, "loss_yns_2": 0.14696, "loss_cls_3": 0.85964, "loss_box_3": 1.50926, "loss_cns_3": 0.66017, "loss_yns_3": 0.147, "loss_cls_4": 0.86894, "loss_box_4": 1.50721, "loss_cns_4": 0.66412, "loss_yns_4": 0.14658, "loss_cls_5": 0.86662, "loss_box_5": 1.51331, "loss_cns_5": 0.65964, "loss_yns_5": 0.1473, "loss_cls_dn_0": 0.15834, "loss_box_dn_0": 0.72737, "loss_cls_dn_1": 0.11646, "loss_box_dn_1": 0.66513, "loss_cls_dn_2": 0.11772, "loss_box_dn_2": 0.65662, "loss_cls_dn_3": 0.12209, "loss_box_dn_3": 0.65864, "loss_cls_dn_4": 0.12256, "loss_box_dn_4": 0.66443, "loss_cls_dn_5": 0.12271, "loss_box_dn_5": 0.678, "loss_dense_depth": 0.77621, "loss": 24.7001, "grad_norm": 42.84145, "time": 1.59515} -{"mode": "train", "epoch": 1, "iter": 313, "lr": 0.00022, "memory": 49164, "data_time": 0.0736, "loss_cls_0": 0.76741, "loss_box_0": 1.58089, "loss_cns_0": 0.63086, "loss_yns_0": 0.1505, "loss_cls_1": 0.82861, "loss_box_1": 1.50128, "loss_cns_1": 0.65914, "loss_yns_1": 0.15076, "loss_cls_2": 0.83936, "loss_box_2": 1.45357, "loss_cns_2": 0.66153, "loss_yns_2": 0.14749, "loss_cls_3": 0.84645, "loss_box_3": 1.4559, "loss_cns_3": 0.66006, "loss_yns_3": 0.14792, "loss_cls_4": 0.86038, "loss_box_4": 1.43798, "loss_cns_4": 0.66289, "loss_yns_4": 0.14735, "loss_cls_5": 0.84901, "loss_box_5": 1.42943, "loss_cns_5": 0.6594, "loss_yns_5": 0.14778, "loss_cls_dn_0": 0.15277, "loss_box_dn_0": 0.72502, "loss_cls_dn_1": 0.11249, "loss_box_dn_1": 0.67493, "loss_cls_dn_2": 0.11158, "loss_box_dn_2": 0.65795, "loss_cls_dn_3": 0.11638, "loss_box_dn_3": 0.66168, "loss_cls_dn_4": 0.11679, "loss_box_dn_4": 0.65782, "loss_cls_dn_5": 0.11437, "loss_box_dn_5": 0.65995, "loss_dense_depth": 0.70789, "loss": 24.14557, "grad_norm": 29.73556, "time": 1.57705} -{"mode": "train", "epoch": 1, "iter": 314, "lr": 0.00023, "memory": 49164, "data_time": 0.09856, "loss_cls_0": 0.77603, "loss_box_0": 1.61119, "loss_cns_0": 0.62772, "loss_yns_0": 0.14974, "loss_cls_1": 0.83278, "loss_box_1": 1.48596, "loss_cns_1": 0.66057, "loss_yns_1": 0.14792, "loss_cls_2": 0.84803, "loss_box_2": 1.4566, "loss_cns_2": 0.66136, "loss_yns_2": 0.14605, "loss_cls_3": 0.85285, "loss_box_3": 1.46859, "loss_cns_3": 0.65932, "loss_yns_3": 0.14658, "loss_cls_4": 0.85972, "loss_box_4": 1.43884, "loss_cns_4": 0.66115, "loss_yns_4": 0.14653, "loss_cls_5": 0.86209, "loss_box_5": 1.44666, "loss_cns_5": 0.66049, "loss_yns_5": 0.14771, "loss_cls_dn_0": 0.15244, "loss_box_dn_0": 0.72508, "loss_cls_dn_1": 0.11657, "loss_box_dn_1": 0.6645, "loss_cls_dn_2": 0.12174, "loss_box_dn_2": 0.66024, "loss_cls_dn_3": 0.12527, "loss_box_dn_3": 0.66338, "loss_cls_dn_4": 0.12611, "loss_box_dn_4": 0.6491, "loss_cls_dn_5": 0.12411, "loss_box_dn_5": 0.6529, "loss_dense_depth": 0.78025, "loss": 24.31618, "grad_norm": 40.063, "time": 1.61062} -{"mode": "train", "epoch": 1, "iter": 315, "lr": 0.00023, "memory": 49164, "data_time": 0.10209, "loss_cls_0": 0.76431, "loss_box_0": 1.62994, "loss_cns_0": 0.62782, "loss_yns_0": 0.14982, "loss_cls_1": 0.83304, "loss_box_1": 1.48288, "loss_cns_1": 0.65522, "loss_yns_1": 0.14847, "loss_cls_2": 0.84246, "loss_box_2": 1.46015, "loss_cns_2": 0.6578, "loss_yns_2": 0.14725, "loss_cls_3": 0.84582, "loss_box_3": 1.46273, "loss_cns_3": 0.65632, "loss_yns_3": 0.14803, "loss_cls_4": 0.8513, "loss_box_4": 1.44997, "loss_cns_4": 0.65897, "loss_yns_4": 0.14969, "loss_cls_5": 0.85528, "loss_box_5": 1.45072, "loss_cns_5": 0.65712, "loss_yns_5": 0.14964, "loss_cls_dn_0": 0.14708, "loss_box_dn_0": 0.73537, "loss_cls_dn_1": 0.11176, "loss_box_dn_1": 0.66612, "loss_cls_dn_2": 0.11064, "loss_box_dn_2": 0.65843, "loss_cls_dn_3": 0.11402, "loss_box_dn_3": 0.65927, "loss_cls_dn_4": 0.11335, "loss_box_dn_4": 0.65055, "loss_cls_dn_5": 0.1142, "loss_box_dn_5": 0.65282, "loss_dense_depth": 0.74234, "loss": 24.21069, "grad_norm": 28.05566, "time": 1.58963} -{"mode": "train", "epoch": 1, "iter": 316, "lr": 0.00023, "memory": 49164, "data_time": 0.10931, "loss_cls_0": 0.75982, "loss_box_0": 1.57118, "loss_cns_0": 0.63099, "loss_yns_0": 0.14657, "loss_cls_1": 0.83281, "loss_box_1": 1.45789, "loss_cns_1": 0.65658, "loss_yns_1": 0.14549, "loss_cls_2": 0.84364, "loss_box_2": 1.43194, "loss_cns_2": 0.65965, "loss_yns_2": 0.1464, "loss_cls_3": 0.84191, "loss_box_3": 1.41533, "loss_cns_3": 0.65577, "loss_yns_3": 0.14511, "loss_cls_4": 0.84343, "loss_box_4": 1.42867, "loss_cns_4": 0.65721, "loss_yns_4": 0.14598, "loss_cls_5": 0.84837, "loss_box_5": 1.43195, "loss_cns_5": 0.65682, "loss_yns_5": 0.14603, "loss_cls_dn_0": 0.14738, "loss_box_dn_0": 0.72596, "loss_cls_dn_1": 0.11423, "loss_box_dn_1": 0.64687, "loss_cls_dn_2": 0.11831, "loss_box_dn_2": 0.6349, "loss_cls_dn_3": 0.12117, "loss_box_dn_3": 0.63445, "loss_cls_dn_4": 0.11944, "loss_box_dn_4": 0.63793, "loss_cls_dn_5": 0.11872, "loss_box_dn_5": 0.64383, "loss_dense_depth": 0.75127, "loss": 23.91401, "grad_norm": 37.87156, "time": 1.5999} -{"mode": "train", "epoch": 1, "iter": 317, "lr": 0.00023, "memory": 49164, "data_time": 0.07305, "loss_cls_0": 0.73347, "loss_box_0": 1.58187, "loss_cns_0": 0.63292, "loss_yns_0": 0.14522, "loss_cls_1": 0.81903, "loss_box_1": 1.48569, "loss_cns_1": 0.65875, "loss_yns_1": 0.14647, "loss_cls_2": 0.82217, "loss_box_2": 1.45445, "loss_cns_2": 0.66028, "loss_yns_2": 0.14759, "loss_cls_3": 0.81931, "loss_box_3": 1.44737, "loss_cns_3": 0.66025, "loss_yns_3": 0.14508, "loss_cls_4": 0.82546, "loss_box_4": 1.45245, "loss_cns_4": 0.66006, "loss_yns_4": 0.14577, "loss_cls_5": 0.82875, "loss_box_5": 1.45007, "loss_cns_5": 0.65965, "loss_yns_5": 0.14584, "loss_cls_dn_0": 0.14337, "loss_box_dn_0": 0.73592, "loss_cls_dn_1": 0.11408, "loss_box_dn_1": 0.65707, "loss_cls_dn_2": 0.11764, "loss_box_dn_2": 0.64592, "loss_cls_dn_3": 0.11766, "loss_box_dn_3": 0.65209, "loss_cls_dn_4": 0.11487, "loss_box_dn_4": 0.65747, "loss_cls_dn_5": 0.11659, "loss_box_dn_5": 0.66438, "loss_dense_depth": 0.70738, "loss": 23.9724, "grad_norm": 30.41676, "time": 1.589} -{"mode": "train", "epoch": 1, "iter": 318, "lr": 0.00023, "memory": 49164, "data_time": 0.07364, "loss_cls_0": 0.77487, "loss_box_0": 1.58862, "loss_cns_0": 0.63205, "loss_yns_0": 0.14928, "loss_cls_1": 0.83693, "loss_box_1": 1.48831, "loss_cns_1": 0.6626, "loss_yns_1": 0.14693, "loss_cls_2": 0.84214, "loss_box_2": 1.4727, "loss_cns_2": 0.66252, "loss_yns_2": 0.1479, "loss_cls_3": 0.85776, "loss_box_3": 1.47122, "loss_cns_3": 0.66155, "loss_yns_3": 0.14703, "loss_cls_4": 0.85234, "loss_box_4": 1.48487, "loss_cns_4": 0.6598, "loss_yns_4": 0.14588, "loss_cls_5": 0.85665, "loss_box_5": 1.48261, "loss_cns_5": 0.65984, "loss_yns_5": 0.14601, "loss_cls_dn_0": 0.14474, "loss_box_dn_0": 0.72474, "loss_cls_dn_1": 0.10422, "loss_box_dn_1": 0.66773, "loss_cls_dn_2": 0.10313, "loss_box_dn_2": 0.66658, "loss_cls_dn_3": 0.1122, "loss_box_dn_3": 0.6743, "loss_cls_dn_4": 0.10812, "loss_box_dn_4": 0.6856, "loss_cls_dn_5": 0.11138, "loss_box_dn_5": 0.69112, "loss_dense_depth": 0.7621, "loss": 24.38635, "grad_norm": 41.02284, "time": 1.61544} -{"mode": "train", "epoch": 1, "iter": 319, "lr": 0.00023, "memory": 49164, "data_time": 0.10603, "loss_cls_0": 0.75118, "loss_box_0": 1.59907, "loss_cns_0": 0.63526, "loss_yns_0": 0.14837, "loss_cls_1": 0.82276, "loss_box_1": 1.48104, "loss_cns_1": 0.66469, "loss_yns_1": 0.14624, "loss_cls_2": 0.83147, "loss_box_2": 1.44979, "loss_cns_2": 0.66326, "loss_yns_2": 0.14557, "loss_cls_3": 0.83762, "loss_box_3": 1.44578, "loss_cns_3": 0.66309, "loss_yns_3": 0.14591, "loss_cls_4": 0.83156, "loss_box_4": 1.44039, "loss_cns_4": 0.66165, "loss_yns_4": 0.14456, "loss_cls_5": 0.83759, "loss_box_5": 1.43474, "loss_cns_5": 0.66182, "loss_yns_5": 0.14463, "loss_cls_dn_0": 0.13884, "loss_box_dn_0": 0.73105, "loss_cls_dn_1": 0.10362, "loss_box_dn_1": 0.67207, "loss_cls_dn_2": 0.10472, "loss_box_dn_2": 0.67208, "loss_cls_dn_3": 0.11485, "loss_box_dn_3": 0.67032, "loss_cls_dn_4": 0.10572, "loss_box_dn_4": 0.66984, "loss_cls_dn_5": 0.10738, "loss_box_dn_5": 0.67394, "loss_dense_depth": 0.72898, "loss": 24.08148, "grad_norm": 33.67077, "time": 1.59521} -{"mode": "train", "epoch": 1, "iter": 320, "lr": 0.00023, "memory": 49164, "data_time": 0.08257, "loss_cls_0": 0.73687, "loss_box_0": 1.60804, "loss_cns_0": 0.63584, "loss_yns_0": 0.1475, "loss_cls_1": 0.81947, "loss_box_1": 1.47487, "loss_cns_1": 0.66427, "loss_yns_1": 0.1452, "loss_cls_2": 0.82321, "loss_box_2": 1.44556, "loss_cns_2": 0.66195, "loss_yns_2": 0.14616, "loss_cls_3": 0.8148, "loss_box_3": 1.42577, "loss_cns_3": 0.6601, "loss_yns_3": 0.14377, "loss_cls_4": 0.82063, "loss_box_4": 1.43409, "loss_cns_4": 0.66029, "loss_yns_4": 0.14397, "loss_cls_5": 0.83034, "loss_box_5": 1.43257, "loss_cns_5": 0.66205, "loss_yns_5": 0.14361, "loss_cls_dn_0": 0.136, "loss_box_dn_0": 0.71969, "loss_cls_dn_1": 0.10407, "loss_box_dn_1": 0.66396, "loss_cls_dn_2": 0.10929, "loss_box_dn_2": 0.6642, "loss_cls_dn_3": 0.1144, "loss_box_dn_3": 0.65221, "loss_cls_dn_4": 0.10999, "loss_box_dn_4": 0.65285, "loss_cls_dn_5": 0.11024, "loss_box_dn_5": 0.65446, "loss_dense_depth": 0.70225, "loss": 23.8745, "grad_norm": 34.14489, "time": 1.60455} -{"mode": "train", "epoch": 1, "iter": 321, "lr": 0.00023, "memory": 49164, "data_time": 0.10955, "loss_cls_0": 0.70722, "loss_box_0": 1.58037, "loss_cns_0": 0.63517, "loss_yns_0": 0.14546, "loss_cls_1": 0.7944, "loss_box_1": 1.46185, "loss_cns_1": 0.66221, "loss_yns_1": 0.14514, "loss_cls_2": 0.79739, "loss_box_2": 1.43845, "loss_cns_2": 0.6546, "loss_yns_2": 0.14483, "loss_cls_3": 0.79561, "loss_box_3": 1.42899, "loss_cns_3": 0.65757, "loss_yns_3": 0.14295, "loss_cls_4": 0.79699, "loss_box_4": 1.4325, "loss_cns_4": 0.66279, "loss_yns_4": 0.14318, "loss_cls_5": 0.80304, "loss_box_5": 1.4291, "loss_cns_5": 0.66212, "loss_yns_5": 0.14453, "loss_cls_dn_0": 0.13548, "loss_box_dn_0": 0.72476, "loss_cls_dn_1": 0.10582, "loss_box_dn_1": 0.66277, "loss_cls_dn_2": 0.11098, "loss_box_dn_2": 0.65075, "loss_cls_dn_3": 0.11168, "loss_box_dn_3": 0.64169, "loss_cls_dn_4": 0.11035, "loss_box_dn_4": 0.63712, "loss_cls_dn_5": 0.1104, "loss_box_dn_5": 0.63953, "loss_dense_depth": 0.68368, "loss": 23.59144, "grad_norm": 38.25582, "time": 1.65599} -{"mode": "train", "epoch": 1, "iter": 322, "lr": 0.00023, "memory": 49164, "data_time": 0.18875, "loss_cls_0": 0.73309, "loss_box_0": 1.57657, "loss_cns_0": 0.63319, "loss_yns_0": 0.14745, "loss_cls_1": 0.80556, "loss_box_1": 1.42271, "loss_cns_1": 0.66371, "loss_yns_1": 0.14507, "loss_cls_2": 0.8182, "loss_box_2": 1.38701, "loss_cns_2": 0.65699, "loss_yns_2": 0.14469, "loss_cls_3": 0.83027, "loss_box_3": 1.38627, "loss_cns_3": 0.66254, "loss_yns_3": 0.14441, "loss_cls_4": 0.82386, "loss_box_4": 1.39539, "loss_cns_4": 0.66273, "loss_yns_4": 0.14398, "loss_cls_5": 0.84958, "loss_box_5": 1.40788, "loss_cns_5": 0.66385, "loss_yns_5": 0.14653, "loss_cls_dn_0": 0.13901, "loss_box_dn_0": 0.72289, "loss_cls_dn_1": 0.10298, "loss_box_dn_1": 0.65953, "loss_cls_dn_2": 0.10407, "loss_box_dn_2": 0.6488, "loss_cls_dn_3": 0.1055, "loss_box_dn_3": 0.64558, "loss_cls_dn_4": 0.10513, "loss_box_dn_4": 0.6531, "loss_cls_dn_5": 0.10884, "loss_box_dn_5": 0.6596, "loss_dense_depth": 0.68157, "loss": 23.58814, "grad_norm": 29.3308, "time": 1.66758} -{"mode": "train", "epoch": 1, "iter": 323, "lr": 0.00023, "memory": 49164, "data_time": 0.07583, "loss_cls_0": 0.72647, "loss_box_0": 1.56941, "loss_cns_0": 0.63412, "loss_yns_0": 0.14184, "loss_cls_1": 0.79047, "loss_box_1": 1.47233, "loss_cns_1": 0.66373, "loss_yns_1": 0.1425, "loss_cls_2": 0.80405, "loss_box_2": 1.44488, "loss_cns_2": 0.6636, "loss_yns_2": 0.14374, "loss_cls_3": 0.80509, "loss_box_3": 1.4363, "loss_cns_3": 0.66249, "loss_yns_3": 0.14393, "loss_cls_4": 0.8159, "loss_box_4": 1.44031, "loss_cns_4": 0.66253, "loss_yns_4": 0.14529, "loss_cls_5": 0.81834, "loss_box_5": 1.45713, "loss_cns_5": 0.66585, "loss_yns_5": 0.14597, "loss_cls_dn_0": 0.13996, "loss_box_dn_0": 0.72305, "loss_cls_dn_1": 0.09862, "loss_box_dn_1": 0.6633, "loss_cls_dn_2": 0.09968, "loss_box_dn_2": 0.65587, "loss_cls_dn_3": 0.10228, "loss_box_dn_3": 0.65766, "loss_cls_dn_4": 0.10516, "loss_box_dn_4": 0.67158, "loss_cls_dn_5": 0.10527, "loss_box_dn_5": 0.68762, "loss_dense_depth": 0.67687, "loss": 23.78316, "grad_norm": 36.32337, "time": 1.57653} -{"mode": "train", "epoch": 1, "iter": 324, "lr": 0.00023, "memory": 49164, "data_time": 0.07701, "loss_cls_0": 0.71923, "loss_box_0": 1.61453, "loss_cns_0": 0.6335, "loss_yns_0": 0.14328, "loss_cls_1": 0.79525, "loss_box_1": 1.48832, "loss_cns_1": 0.66259, "loss_yns_1": 0.14387, "loss_cls_2": 0.80324, "loss_box_2": 1.47392, "loss_cns_2": 0.66199, "loss_yns_2": 0.14426, "loss_cls_3": 0.80755, "loss_box_3": 1.45092, "loss_cns_3": 0.66391, "loss_yns_3": 0.14308, "loss_cls_4": 0.81619, "loss_box_4": 1.46062, "loss_cns_4": 0.6647, "loss_yns_4": 0.1456, "loss_cls_5": 0.81401, "loss_box_5": 1.4646, "loss_cns_5": 0.6704, "loss_yns_5": 0.14689, "loss_cls_dn_0": 0.13461, "loss_box_dn_0": 0.71857, "loss_cls_dn_1": 0.10033, "loss_box_dn_1": 0.66209, "loss_cls_dn_2": 0.10112, "loss_box_dn_2": 0.65173, "loss_cls_dn_3": 0.10207, "loss_box_dn_3": 0.64391, "loss_cls_dn_4": 0.10373, "loss_box_dn_4": 0.65591, "loss_cls_dn_5": 0.10436, "loss_box_dn_5": 0.66475, "loss_dense_depth": 0.69019, "loss": 23.86583, "grad_norm": 30.14779, "time": 1.61917} -{"mode": "train", "epoch": 1, "iter": 325, "lr": 0.00023, "memory": 49164, "data_time": 0.07364, "loss_cls_0": 0.71004, "loss_box_0": 1.62119, "loss_cns_0": 0.64066, "loss_yns_0": 0.14543, "loss_cls_1": 0.78856, "loss_box_1": 1.47246, "loss_cns_1": 0.66671, "loss_yns_1": 0.14368, "loss_cls_2": 0.80029, "loss_box_2": 1.4629, "loss_cns_2": 0.66696, "loss_yns_2": 0.14524, "loss_cls_3": 0.79968, "loss_box_3": 1.43967, "loss_cns_3": 0.66529, "loss_yns_3": 0.14384, "loss_cls_4": 0.81199, "loss_box_4": 1.43592, "loss_cns_4": 0.66695, "loss_yns_4": 0.14392, "loss_cls_5": 0.81245, "loss_box_5": 1.43451, "loss_cns_5": 0.66735, "loss_yns_5": 0.14436, "loss_cls_dn_0": 0.13409, "loss_box_dn_0": 0.72094, "loss_cls_dn_1": 0.10209, "loss_box_dn_1": 0.65089, "loss_cls_dn_2": 0.10353, "loss_box_dn_2": 0.64326, "loss_cls_dn_3": 0.10369, "loss_box_dn_3": 0.63534, "loss_cls_dn_4": 0.10588, "loss_box_dn_4": 0.63579, "loss_cls_dn_5": 0.11024, "loss_box_dn_5": 0.64111, "loss_dense_depth": 0.69418, "loss": 23.71106, "grad_norm": 36.94784, "time": 1.64111} -{"mode": "train", "epoch": 1, "iter": 326, "lr": 0.00023, "memory": 49164, "data_time": 0.07267, "loss_cls_0": 0.70908, "loss_box_0": 1.59764, "loss_cns_0": 0.64038, "loss_yns_0": 0.14584, "loss_cls_1": 0.77469, "loss_box_1": 1.4883, "loss_cns_1": 0.66277, "loss_yns_1": 0.14147, "loss_cls_2": 0.78611, "loss_box_2": 1.45325, "loss_cns_2": 0.66411, "loss_yns_2": 0.14432, "loss_cls_3": 0.79318, "loss_box_3": 1.43495, "loss_cns_3": 0.66389, "loss_yns_3": 0.14322, "loss_cls_4": 0.79901, "loss_box_4": 1.42799, "loss_cns_4": 0.66526, "loss_yns_4": 0.14305, "loss_cls_5": 0.80014, "loss_box_5": 1.43729, "loss_cns_5": 0.66435, "loss_yns_5": 0.14188, "loss_cls_dn_0": 0.12805, "loss_box_dn_0": 0.71288, "loss_cls_dn_1": 0.10178, "loss_box_dn_1": 0.63323, "loss_cls_dn_2": 0.10059, "loss_box_dn_2": 0.62034, "loss_cls_dn_3": 0.10041, "loss_box_dn_3": 0.615, "loss_cls_dn_4": 0.10372, "loss_box_dn_4": 0.61354, "loss_cls_dn_5": 0.10511, "loss_box_dn_5": 0.62307, "loss_dense_depth": 0.6801, "loss": 23.45997, "grad_norm": 33.69952, "time": 1.71173} -{"mode": "train", "epoch": 1, "iter": 327, "lr": 0.00023, "memory": 49164, "data_time": 0.10445, "loss_cls_0": 0.71789, "loss_box_0": 1.60113, "loss_cns_0": 0.63511, "loss_yns_0": 0.14124, "loss_cls_1": 0.78822, "loss_box_1": 1.4674, "loss_cns_1": 0.66149, "loss_yns_1": 0.13947, "loss_cls_2": 0.80745, "loss_box_2": 1.44185, "loss_cns_2": 0.65818, "loss_yns_2": 0.14058, "loss_cls_3": 0.80991, "loss_box_3": 1.4492, "loss_cns_3": 0.65958, "loss_yns_3": 0.14016, "loss_cls_4": 0.80399, "loss_box_4": 1.45086, "loss_cns_4": 0.66523, "loss_yns_4": 0.14248, "loss_cls_5": 0.80248, "loss_box_5": 1.43738, "loss_cns_5": 0.66254, "loss_yns_5": 0.1407, "loss_cls_dn_0": 0.13522, "loss_box_dn_0": 0.71761, "loss_cls_dn_1": 0.10266, "loss_box_dn_1": 0.63134, "loss_cls_dn_2": 0.10224, "loss_box_dn_2": 0.61926, "loss_cls_dn_3": 0.1026, "loss_box_dn_3": 0.62399, "loss_cls_dn_4": 0.10472, "loss_box_dn_4": 0.62133, "loss_cls_dn_5": 0.10453, "loss_box_dn_5": 0.62396, "loss_dense_depth": 0.67517, "loss": 23.52914, "grad_norm": 31.48841, "time": 1.6021} -{"mode": "train", "epoch": 1, "iter": 328, "lr": 0.00023, "memory": 49164, "data_time": 0.10409, "loss_cls_0": 0.71644, "loss_box_0": 1.60827, "loss_cns_0": 0.6356, "loss_yns_0": 0.14332, "loss_cls_1": 0.79321, "loss_box_1": 1.49318, "loss_cns_1": 0.66298, "loss_yns_1": 0.14087, "loss_cls_2": 0.79517, "loss_box_2": 1.44195, "loss_cns_2": 0.66225, "loss_yns_2": 0.14006, "loss_cls_3": 0.80258, "loss_box_3": 1.45681, "loss_cns_3": 0.66493, "loss_yns_3": 0.14138, "loss_cls_4": 0.82695, "loss_box_4": 1.45089, "loss_cns_4": 0.66611, "loss_yns_4": 0.14263, "loss_cls_5": 0.80955, "loss_box_5": 1.43022, "loss_cns_5": 0.66468, "loss_yns_5": 0.14146, "loss_cls_dn_0": 0.13376, "loss_box_dn_0": 0.7254, "loss_cls_dn_1": 0.10261, "loss_box_dn_1": 0.64796, "loss_cls_dn_2": 0.10275, "loss_box_dn_2": 0.63706, "loss_cls_dn_3": 0.10325, "loss_box_dn_3": 0.64639, "loss_cls_dn_4": 0.10538, "loss_box_dn_4": 0.64444, "loss_cls_dn_5": 0.10575, "loss_box_dn_5": 0.64757, "loss_dense_depth": 0.70548, "loss": 23.73926, "grad_norm": 42.93483, "time": 1.58638} -{"mode": "train", "epoch": 1, "iter": 329, "lr": 0.00023, "memory": 49164, "data_time": 0.07534, "loss_cls_0": 0.71697, "loss_box_0": 1.6187, "loss_cns_0": 0.63797, "loss_yns_0": 0.14258, "loss_cls_1": 0.79519, "loss_box_1": 1.50042, "loss_cns_1": 0.66624, "loss_yns_1": 0.14122, "loss_cls_2": 0.80035, "loss_box_2": 1.47309, "loss_cns_2": 0.66564, "loss_yns_2": 0.14036, "loss_cls_3": 0.81039, "loss_box_3": 1.46217, "loss_cns_3": 0.66602, "loss_yns_3": 0.13978, "loss_cls_4": 0.81905, "loss_box_4": 1.45465, "loss_cns_4": 0.66873, "loss_yns_4": 0.14088, "loss_cls_5": 0.81849, "loss_box_5": 1.44916, "loss_cns_5": 0.66654, "loss_yns_5": 0.13949, "loss_cls_dn_0": 0.13624, "loss_box_dn_0": 0.71868, "loss_cls_dn_1": 0.10452, "loss_box_dn_1": 0.65658, "loss_cls_dn_2": 0.1029, "loss_box_dn_2": 0.65264, "loss_cls_dn_3": 0.10222, "loss_box_dn_3": 0.65527, "loss_cls_dn_4": 0.10619, "loss_box_dn_4": 0.65052, "loss_cls_dn_5": 0.10707, "loss_box_dn_5": 0.65685, "loss_dense_depth": 0.70499, "loss": 23.88872, "grad_norm": 31.86511, "time": 1.59825} -{"mode": "train", "epoch": 1, "iter": 330, "lr": 0.00023, "memory": 49164, "data_time": 0.0838, "loss_cls_0": 0.70737, "loss_box_0": 1.61242, "loss_cns_0": 0.63977, "loss_yns_0": 0.14236, "loss_cls_1": 0.78931, "loss_box_1": 1.48652, "loss_cns_1": 0.66611, "loss_yns_1": 0.1406, "loss_cls_2": 0.78615, "loss_box_2": 1.4659, "loss_cns_2": 0.66736, "loss_yns_2": 0.14083, "loss_cls_3": 0.80853, "loss_box_3": 1.44712, "loss_cns_3": 0.66625, "loss_yns_3": 0.1381, "loss_cls_4": 0.81643, "loss_box_4": 1.45474, "loss_cns_4": 0.66847, "loss_yns_4": 0.13958, "loss_cls_5": 0.80044, "loss_box_5": 1.44021, "loss_cns_5": 0.66777, "loss_yns_5": 0.13892, "loss_cls_dn_0": 0.13917, "loss_box_dn_0": 0.72719, "loss_cls_dn_1": 0.10264, "loss_box_dn_1": 0.66149, "loss_cls_dn_2": 0.10144, "loss_box_dn_2": 0.65468, "loss_cls_dn_3": 0.10265, "loss_box_dn_3": 0.64897, "loss_cls_dn_4": 0.10498, "loss_box_dn_4": 0.64935, "loss_cls_dn_5": 0.10577, "loss_box_dn_5": 0.64948, "loss_dense_depth": 0.68817, "loss": 23.76727, "grad_norm": 47.59902, "time": 1.56802} -{"mode": "train", "epoch": 1, "iter": 331, "lr": 0.00023, "memory": 49164, "data_time": 0.09328, "loss_cls_0": 0.71997, "loss_box_0": 1.61613, "loss_cns_0": 0.63394, "loss_yns_0": 0.14404, "loss_cls_1": 0.79557, "loss_box_1": 1.49817, "loss_cns_1": 0.66065, "loss_yns_1": 0.14139, "loss_cls_2": 0.79967, "loss_box_2": 1.46748, "loss_cns_2": 0.66428, "loss_yns_2": 0.14305, "loss_cls_3": 0.80495, "loss_box_3": 1.45661, "loss_cns_3": 0.66405, "loss_yns_3": 0.14102, "loss_cls_4": 0.82358, "loss_box_4": 1.46188, "loss_cns_4": 0.66231, "loss_yns_4": 0.14034, "loss_cls_5": 0.81941, "loss_box_5": 1.44688, "loss_cns_5": 0.66218, "loss_yns_5": 0.14092, "loss_cls_dn_0": 0.1415, "loss_box_dn_0": 0.72195, "loss_cls_dn_1": 0.10468, "loss_box_dn_1": 0.64912, "loss_cls_dn_2": 0.1053, "loss_box_dn_2": 0.63054, "loss_cls_dn_3": 0.10699, "loss_box_dn_3": 0.62319, "loss_cls_dn_4": 0.1086, "loss_box_dn_4": 0.6287, "loss_cls_dn_5": 0.11084, "loss_box_dn_5": 0.62605, "loss_dense_depth": 0.69421, "loss": 23.76014, "grad_norm": 43.05459, "time": 1.66211} -{"mode": "train", "epoch": 1, "iter": 332, "lr": 0.00023, "memory": 49164, "data_time": 0.08006, "loss_cls_0": 0.72366, "loss_box_0": 1.6109, "loss_cns_0": 0.6378, "loss_yns_0": 0.1452, "loss_cls_1": 0.78319, "loss_box_1": 1.46199, "loss_cns_1": 0.66559, "loss_yns_1": 0.14231, "loss_cls_2": 0.8073, "loss_box_2": 1.44338, "loss_cns_2": 0.6661, "loss_yns_2": 0.14199, "loss_cls_3": 0.82019, "loss_box_3": 1.45671, "loss_cns_3": 0.66801, "loss_yns_3": 0.1428, "loss_cls_4": 0.81299, "loss_box_4": 1.44336, "loss_cns_4": 0.6703, "loss_yns_4": 0.14336, "loss_cls_5": 0.80593, "loss_box_5": 1.41911, "loss_cns_5": 0.66778, "loss_yns_5": 0.14186, "loss_cls_dn_0": 0.13764, "loss_box_dn_0": 0.73399, "loss_cls_dn_1": 0.10427, "loss_box_dn_1": 0.64504, "loss_cls_dn_2": 0.10259, "loss_box_dn_2": 0.63143, "loss_cls_dn_3": 0.10446, "loss_box_dn_3": 0.63781, "loss_cls_dn_4": 0.10669, "loss_box_dn_4": 0.63418, "loss_cls_dn_5": 0.10717, "loss_box_dn_5": 0.63154, "loss_dense_depth": 0.68263, "loss": 23.68123, "grad_norm": 48.99204, "time": 1.57534} -{"mode": "train", "epoch": 1, "iter": 333, "lr": 0.00023, "memory": 49164, "data_time": 0.07774, "loss_cls_0": 0.73666, "loss_box_0": 1.60544, "loss_cns_0": 0.63711, "loss_yns_0": 0.14383, "loss_cls_1": 0.80608, "loss_box_1": 1.47976, "loss_cns_1": 0.66453, "loss_yns_1": 0.14161, "loss_cls_2": 0.82402, "loss_box_2": 1.46046, "loss_cns_2": 0.66546, "loss_yns_2": 0.14181, "loss_cls_3": 0.83935, "loss_box_3": 1.4726, "loss_cns_3": 0.66624, "loss_yns_3": 0.14116, "loss_cls_4": 0.84138, "loss_box_4": 1.47061, "loss_cns_4": 0.66526, "loss_yns_4": 0.14325, "loss_cls_5": 0.8273, "loss_box_5": 1.44812, "loss_cns_5": 0.66244, "loss_yns_5": 0.14086, "loss_cls_dn_0": 0.14469, "loss_box_dn_0": 0.72463, "loss_cls_dn_1": 0.10405, "loss_box_dn_1": 0.65154, "loss_cls_dn_2": 0.10453, "loss_box_dn_2": 0.64649, "loss_cls_dn_3": 0.11049, "loss_box_dn_3": 0.65747, "loss_cls_dn_4": 0.10876, "loss_box_dn_4": 0.65901, "loss_cls_dn_5": 0.10725, "loss_box_dn_5": 0.66502, "loss_dense_depth": 0.73006, "loss": 24.03931, "grad_norm": 55.26593, "time": 1.57054} -{"mode": "train", "epoch": 1, "iter": 334, "lr": 0.00023, "memory": 49164, "data_time": 0.07728, "loss_cls_0": 0.74389, "loss_box_0": 1.62558, "loss_cns_0": 0.62616, "loss_yns_0": 0.14438, "loss_cls_1": 0.82978, "loss_box_1": 1.48766, "loss_cns_1": 0.65963, "loss_yns_1": 0.14154, "loss_cls_2": 0.82892, "loss_box_2": 1.49876, "loss_cns_2": 0.66373, "loss_yns_2": 0.14465, "loss_cls_3": 0.83624, "loss_box_3": 1.4917, "loss_cns_3": 0.66382, "loss_yns_3": 0.14331, "loss_cls_4": 0.84208, "loss_box_4": 1.50997, "loss_cns_4": 0.66768, "loss_yns_4": 0.14504, "loss_cls_5": 0.85935, "loss_box_5": 1.4985, "loss_cns_5": 0.6625, "loss_yns_5": 0.14367, "loss_cls_dn_0": 0.14043, "loss_box_dn_0": 0.72993, "loss_cls_dn_1": 0.11201, "loss_box_dn_1": 0.67654, "loss_cls_dn_2": 0.1109, "loss_box_dn_2": 0.67398, "loss_cls_dn_3": 0.11529, "loss_box_dn_3": 0.67724, "loss_cls_dn_4": 0.11149, "loss_box_dn_4": 0.69113, "loss_cls_dn_5": 0.1144, "loss_box_dn_5": 0.69824, "loss_dense_depth": 0.72408, "loss": 24.43419, "grad_norm": 61.3368, "time": 1.57199} -{"mode": "train", "epoch": 1, "iter": 335, "lr": 0.00023, "memory": 49164, "data_time": 0.07405, "loss_cls_0": 0.73359, "loss_box_0": 1.62139, "loss_cns_0": 0.633, "loss_yns_0": 0.14471, "loss_cls_1": 0.80908, "loss_box_1": 1.48725, "loss_cns_1": 0.66009, "loss_yns_1": 0.14248, "loss_cls_2": 0.81471, "loss_box_2": 1.48916, "loss_cns_2": 0.6642, "loss_yns_2": 0.14313, "loss_cls_3": 0.81242, "loss_box_3": 1.47783, "loss_cns_3": 0.66474, "loss_yns_3": 0.14255, "loss_cls_4": 0.81362, "loss_box_4": 1.49947, "loss_cns_4": 0.67305, "loss_yns_4": 0.14198, "loss_cls_5": 0.82344, "loss_box_5": 1.48047, "loss_cns_5": 0.66813, "loss_yns_5": 0.1418, "loss_cls_dn_0": 0.13937, "loss_box_dn_0": 0.73172, "loss_cls_dn_1": 0.10871, "loss_box_dn_1": 0.68244, "loss_cls_dn_2": 0.10461, "loss_box_dn_2": 0.67545, "loss_cls_dn_3": 0.10558, "loss_box_dn_3": 0.6748, "loss_cls_dn_4": 0.10481, "loss_box_dn_4": 0.6909, "loss_cls_dn_5": 0.10996, "loss_box_dn_5": 0.69006, "loss_dense_depth": 0.69288, "loss": 24.19358, "grad_norm": 46.71754, "time": 1.56254} -{"mode": "train", "epoch": 1, "iter": 336, "lr": 0.00023, "memory": 49164, "data_time": 0.07263, "loss_cls_0": 0.72381, "loss_box_0": 1.6256, "loss_cns_0": 0.63138, "loss_yns_0": 0.1425, "loss_cls_1": 0.8009, "loss_box_1": 1.55464, "loss_cns_1": 0.65856, "loss_yns_1": 0.1404, "loss_cls_2": 0.82696, "loss_box_2": 1.50378, "loss_cns_2": 0.66359, "loss_yns_2": 0.14134, "loss_cls_3": 0.83425, "loss_box_3": 1.52116, "loss_cns_3": 0.66441, "loss_yns_3": 0.14138, "loss_cls_4": 0.83157, "loss_box_4": 1.53089, "loss_cns_4": 0.66655, "loss_yns_4": 0.14228, "loss_cls_5": 0.82351, "loss_box_5": 1.52727, "loss_cns_5": 0.66329, "loss_yns_5": 0.14077, "loss_cls_dn_0": 0.13703, "loss_box_dn_0": 0.7268, "loss_cls_dn_1": 0.10472, "loss_box_dn_1": 0.67958, "loss_cls_dn_2": 0.10412, "loss_box_dn_2": 0.67095, "loss_cls_dn_3": 0.10539, "loss_box_dn_3": 0.67846, "loss_cls_dn_4": 0.1081, "loss_box_dn_4": 0.67996, "loss_cls_dn_5": 0.11091, "loss_box_dn_5": 0.68799, "loss_dense_depth": 0.701, "loss": 24.39576, "grad_norm": 57.21959, "time": 1.5602} -{"mode": "train", "epoch": 1, "iter": 337, "lr": 0.00023, "memory": 49164, "data_time": 0.10563, "loss_cls_0": 0.72732, "loss_box_0": 1.66042, "loss_cns_0": 0.6286, "loss_yns_0": 0.14197, "loss_cls_1": 0.80735, "loss_box_1": 1.6186, "loss_cns_1": 0.65774, "loss_yns_1": 0.14089, "loss_cls_2": 0.82853, "loss_box_2": 1.53348, "loss_cns_2": 0.66336, "loss_yns_2": 0.13964, "loss_cls_3": 0.8349, "loss_box_3": 1.53625, "loss_cns_3": 0.66184, "loss_yns_3": 0.1404, "loss_cls_4": 0.82917, "loss_box_4": 1.53432, "loss_cns_4": 0.65915, "loss_yns_4": 0.14365, "loss_cls_5": 0.83547, "loss_box_5": 1.54434, "loss_cns_5": 0.66049, "loss_yns_5": 0.1419, "loss_cls_dn_0": 0.14237, "loss_box_dn_0": 0.72527, "loss_cls_dn_1": 0.10887, "loss_box_dn_1": 0.68007, "loss_cls_dn_2": 0.11308, "loss_box_dn_2": 0.66225, "loss_cls_dn_3": 0.11582, "loss_box_dn_3": 0.66617, "loss_cls_dn_4": 0.12015, "loss_box_dn_4": 0.66194, "loss_cls_dn_5": 0.11988, "loss_box_dn_5": 0.66984, "loss_dense_depth": 0.70095, "loss": 24.55644, "grad_norm": 67.20949, "time": 1.62326} -{"mode": "train", "epoch": 1, "iter": 338, "lr": 0.00023, "memory": 49164, "data_time": 0.07365, "loss_cls_0": 0.72811, "loss_box_0": 1.59968, "loss_cns_0": 0.63736, "loss_yns_0": 0.14148, "loss_cls_1": 0.80902, "loss_box_1": 1.53865, "loss_cns_1": 0.6595, "loss_yns_1": 0.13961, "loss_cls_2": 0.81399, "loss_box_2": 1.51048, "loss_cns_2": 0.66082, "loss_yns_2": 0.1397, "loss_cls_3": 0.80969, "loss_box_3": 1.50101, "loss_cns_3": 0.65854, "loss_yns_3": 0.14093, "loss_cls_4": 0.82548, "loss_box_4": 1.50418, "loss_cns_4": 0.6584, "loss_yns_4": 0.14355, "loss_cls_5": 0.82927, "loss_box_5": 1.49467, "loss_cns_5": 0.66025, "loss_yns_5": 0.14207, "loss_cls_dn_0": 0.14041, "loss_box_dn_0": 0.72427, "loss_cls_dn_1": 0.11239, "loss_box_dn_1": 0.65962, "loss_cls_dn_2": 0.11466, "loss_box_dn_2": 0.65264, "loss_cls_dn_3": 0.11804, "loss_box_dn_3": 0.65616, "loss_cls_dn_4": 0.12285, "loss_box_dn_4": 0.65683, "loss_cls_dn_5": 0.11929, "loss_box_dn_5": 0.65195, "loss_dense_depth": 0.71094, "loss": 24.18647, "grad_norm": 56.86732, "time": 1.56202} -{"mode": "train", "epoch": 1, "iter": 339, "lr": 0.00023, "memory": 49164, "data_time": 0.08606, "loss_cls_0": 0.72136, "loss_box_0": 1.60746, "loss_cns_0": 0.63186, "loss_yns_0": 0.14006, "loss_cls_1": 0.80664, "loss_box_1": 1.49779, "loss_cns_1": 0.66297, "loss_yns_1": 0.1387, "loss_cls_2": 0.81234, "loss_box_2": 1.49094, "loss_cns_2": 0.66295, "loss_yns_2": 0.13863, "loss_cls_3": 0.81439, "loss_box_3": 1.48216, "loss_cns_3": 0.66096, "loss_yns_3": 0.14028, "loss_cls_4": 0.81972, "loss_box_4": 1.50239, "loss_cns_4": 0.6625, "loss_yns_4": 0.14299, "loss_cls_5": 0.83037, "loss_box_5": 1.47269, "loss_cns_5": 0.66406, "loss_yns_5": 0.14167, "loss_cls_dn_0": 0.13998, "loss_box_dn_0": 0.72678, "loss_cls_dn_1": 0.10899, "loss_box_dn_1": 0.65711, "loss_cls_dn_2": 0.11014, "loss_box_dn_2": 0.65218, "loss_cls_dn_3": 0.11019, "loss_box_dn_3": 0.65476, "loss_cls_dn_4": 0.11136, "loss_box_dn_4": 0.66597, "loss_cls_dn_5": 0.11075, "loss_box_dn_5": 0.66042, "loss_dense_depth": 0.70041, "loss": 24.05491, "grad_norm": 44.17866, "time": 1.58126} -{"mode": "train", "epoch": 1, "iter": 340, "lr": 0.00024, "memory": 49164, "data_time": 0.08148, "loss_cls_0": 0.7521, "loss_box_0": 1.60201, "loss_cns_0": 0.63115, "loss_yns_0": 0.14151, "loss_cls_1": 0.81154, "loss_box_1": 1.50564, "loss_cns_1": 0.66259, "loss_yns_1": 0.13909, "loss_cls_2": 0.82869, "loss_box_2": 1.47716, "loss_cns_2": 0.66751, "loss_yns_2": 0.14038, "loss_cls_3": 0.83709, "loss_box_3": 1.46865, "loss_cns_3": 0.66717, "loss_yns_3": 0.14058, "loss_cls_4": 0.83447, "loss_box_4": 1.4886, "loss_cns_4": 0.67413, "loss_yns_4": 0.14244, "loss_cls_5": 0.85304, "loss_box_5": 1.48798, "loss_cns_5": 0.67711, "loss_yns_5": 0.14321, "loss_cls_dn_0": 0.13681, "loss_box_dn_0": 0.72847, "loss_cls_dn_1": 0.11086, "loss_box_dn_1": 0.66872, "loss_cls_dn_2": 0.11473, "loss_box_dn_2": 0.65796, "loss_cls_dn_3": 0.11452, "loss_box_dn_3": 0.66194, "loss_cls_dn_4": 0.11444, "loss_box_dn_4": 0.66679, "loss_cls_dn_5": 0.12227, "loss_box_dn_5": 0.66978, "loss_dense_depth": 0.70795, "loss": 24.24909, "grad_norm": 51.0514, "time": 1.57217} -{"mode": "train", "epoch": 1, "iter": 341, "lr": 0.00024, "memory": 49164, "data_time": 0.11621, "loss_cls_0": 0.74537, "loss_box_0": 1.5934, "loss_cns_0": 0.63363, "loss_yns_0": 0.14091, "loss_cls_1": 0.80189, "loss_box_1": 1.53029, "loss_cns_1": 0.65855, "loss_yns_1": 0.14013, "loss_cls_2": 0.8131, "loss_box_2": 1.4979, "loss_cns_2": 0.65566, "loss_yns_2": 0.13856, "loss_cls_3": 0.81778, "loss_box_3": 1.49346, "loss_cns_3": 0.65913, "loss_yns_3": 0.13818, "loss_cls_4": 0.82272, "loss_box_4": 1.49099, "loss_cns_4": 0.6635, "loss_yns_4": 0.13923, "loss_cls_5": 0.83672, "loss_box_5": 1.50211, "loss_cns_5": 0.66221, "loss_yns_5": 0.14222, "loss_cls_dn_0": 0.13884, "loss_box_dn_0": 0.7238, "loss_cls_dn_1": 0.10807, "loss_box_dn_1": 0.66629, "loss_cls_dn_2": 0.10961, "loss_box_dn_2": 0.65749, "loss_cls_dn_3": 0.11001, "loss_box_dn_3": 0.65499, "loss_cls_dn_4": 0.11126, "loss_box_dn_4": 0.65363, "loss_cls_dn_5": 0.11493, "loss_box_dn_5": 0.65832, "loss_dense_depth": 0.72745, "loss": 24.15235, "grad_norm": 40.71144, "time": 1.70011} -{"mode": "train", "epoch": 1, "iter": 342, "lr": 0.00024, "memory": 49164, "data_time": 0.17001, "loss_cls_0": 0.73585, "loss_box_0": 1.59776, "loss_cns_0": 0.63348, "loss_yns_0": 0.14352, "loss_cls_1": 0.81194, "loss_box_1": 1.55392, "loss_cns_1": 0.66084, "loss_yns_1": 0.14511, "loss_cls_2": 0.82397, "loss_box_2": 1.51807, "loss_cns_2": 0.65821, "loss_yns_2": 0.14274, "loss_cls_3": 0.82894, "loss_box_3": 1.51947, "loss_cns_3": 0.65912, "loss_yns_3": 0.14292, "loss_cls_4": 0.84387, "loss_box_4": 1.51795, "loss_cns_4": 0.65853, "loss_yns_4": 0.14936, "loss_cls_5": 0.83876, "loss_box_5": 1.50924, "loss_cns_5": 0.66307, "loss_yns_5": 0.14566, "loss_cls_dn_0": 0.1333, "loss_box_dn_0": 0.72124, "loss_cls_dn_1": 0.10816, "loss_box_dn_1": 0.65526, "loss_cls_dn_2": 0.10568, "loss_box_dn_2": 0.6391, "loss_cls_dn_3": 0.10903, "loss_box_dn_3": 0.64062, "loss_cls_dn_4": 0.1068, "loss_box_dn_4": 0.64575, "loss_cls_dn_5": 0.10645, "loss_box_dn_5": 0.64311, "loss_dense_depth": 0.71531, "loss": 24.23212, "grad_norm": 43.87159, "time": 1.66197} -{"mode": "train", "epoch": 1, "iter": 343, "lr": 0.00024, "memory": 49164, "data_time": 0.07055, "loss_cls_0": 0.7454, "loss_box_0": 1.6349, "loss_cns_0": 0.63385, "loss_yns_0": 0.14243, "loss_cls_1": 0.81085, "loss_box_1": 1.58591, "loss_cns_1": 0.65811, "loss_yns_1": 0.14014, "loss_cls_2": 0.81926, "loss_box_2": 1.52299, "loss_cns_2": 0.66137, "loss_yns_2": 0.13881, "loss_cls_3": 0.81828, "loss_box_3": 1.52497, "loss_cns_3": 0.66139, "loss_yns_3": 0.13928, "loss_cls_4": 0.83451, "loss_box_4": 1.53679, "loss_cns_4": 0.66177, "loss_yns_4": 0.14512, "loss_cls_5": 0.83161, "loss_box_5": 1.52883, "loss_cns_5": 0.66642, "loss_yns_5": 0.14137, "loss_cls_dn_0": 0.133, "loss_box_dn_0": 0.71886, "loss_cls_dn_1": 0.10604, "loss_box_dn_1": 0.66818, "loss_cls_dn_2": 0.10445, "loss_box_dn_2": 0.64176, "loss_cls_dn_3": 0.10569, "loss_box_dn_3": 0.64614, "loss_cls_dn_4": 0.10587, "loss_box_dn_4": 0.65706, "loss_cls_dn_5": 0.10606, "loss_box_dn_5": 0.65376, "loss_dense_depth": 0.70781, "loss": 24.33904, "grad_norm": 38.16846, "time": 1.56428} -{"mode": "train", "epoch": 1, "iter": 344, "lr": 0.00024, "memory": 49164, "data_time": 0.10876, "loss_cls_0": 0.76807, "loss_box_0": 1.65742, "loss_cns_0": 0.62936, "loss_yns_0": 0.14229, "loss_cls_1": 0.83261, "loss_box_1": 1.54262, "loss_cns_1": 0.65643, "loss_yns_1": 0.13703, "loss_cls_2": 0.82496, "loss_box_2": 1.53017, "loss_cns_2": 0.65878, "loss_yns_2": 0.13714, "loss_cls_3": 0.83304, "loss_box_3": 1.50501, "loss_cns_3": 0.65836, "loss_yns_3": 0.13628, "loss_cls_4": 0.84457, "loss_box_4": 1.52607, "loss_cns_4": 0.6619, "loss_yns_4": 0.13844, "loss_cls_5": 0.84726, "loss_box_5": 1.49715, "loss_cns_5": 0.66043, "loss_yns_5": 0.1382, "loss_cls_dn_0": 0.13768, "loss_box_dn_0": 0.72923, "loss_cls_dn_1": 0.10283, "loss_box_dn_1": 0.65574, "loss_cls_dn_2": 0.10393, "loss_box_dn_2": 0.65254, "loss_cls_dn_3": 0.10483, "loss_box_dn_3": 0.64402, "loss_cls_dn_4": 0.10619, "loss_box_dn_4": 0.64992, "loss_cls_dn_5": 0.10754, "loss_box_dn_5": 0.63982, "loss_dense_depth": 0.72729, "loss": 24.32516, "grad_norm": 42.36044, "time": 1.60926} -{"mode": "train", "epoch": 1, "iter": 345, "lr": 0.00024, "memory": 49164, "data_time": 0.07384, "loss_cls_0": 0.72497, "loss_box_0": 1.63016, "loss_cns_0": 0.63515, "loss_yns_0": 0.13879, "loss_cls_1": 0.79674, "loss_box_1": 1.51859, "loss_cns_1": 0.65816, "loss_yns_1": 0.1375, "loss_cls_2": 0.79921, "loss_box_2": 1.52365, "loss_cns_2": 0.65844, "loss_yns_2": 0.13795, "loss_cls_3": 0.80462, "loss_box_3": 1.49645, "loss_cns_3": 0.66323, "loss_yns_3": 0.13724, "loss_cls_4": 0.82289, "loss_box_4": 1.4916, "loss_cns_4": 0.66171, "loss_yns_4": 0.13715, "loss_cls_5": 0.82162, "loss_box_5": 1.4796, "loss_cns_5": 0.66117, "loss_yns_5": 0.13739, "loss_cls_dn_0": 0.13622, "loss_box_dn_0": 0.71337, "loss_cls_dn_1": 0.10247, "loss_box_dn_1": 0.63924, "loss_cls_dn_2": 0.10433, "loss_box_dn_2": 0.64257, "loss_cls_dn_3": 0.10576, "loss_box_dn_3": 0.63542, "loss_cls_dn_4": 0.10628, "loss_box_dn_4": 0.6342, "loss_cls_dn_5": 0.10764, "loss_box_dn_5": 0.63526, "loss_dense_depth": 0.70086, "loss": 23.93758, "grad_norm": 40.43537, "time": 1.63948} -{"mode": "train", "epoch": 1, "iter": 346, "lr": 0.00024, "memory": 49164, "data_time": 0.07708, "loss_cls_0": 0.74083, "loss_box_0": 1.62936, "loss_cns_0": 0.62905, "loss_yns_0": 0.13877, "loss_cls_1": 0.80933, "loss_box_1": 1.48521, "loss_cns_1": 0.66722, "loss_yns_1": 0.13588, "loss_cls_2": 0.82453, "loss_box_2": 1.47007, "loss_cns_2": 0.66818, "loss_yns_2": 0.13685, "loss_cls_3": 0.83329, "loss_box_3": 1.46443, "loss_cns_3": 0.66842, "loss_yns_3": 0.13561, "loss_cls_4": 0.84522, "loss_box_4": 1.4676, "loss_cns_4": 0.66307, "loss_yns_4": 0.13585, "loss_cls_5": 0.84297, "loss_box_5": 1.49182, "loss_cns_5": 0.66004, "loss_yns_5": 0.13516, "loss_cls_dn_0": 0.13591, "loss_box_dn_0": 0.72266, "loss_cls_dn_1": 0.10521, "loss_box_dn_1": 0.64907, "loss_cls_dn_2": 0.10614, "loss_box_dn_2": 0.64092, "loss_cls_dn_3": 0.10858, "loss_box_dn_3": 0.6408, "loss_cls_dn_4": 0.10855, "loss_box_dn_4": 0.6485, "loss_cls_dn_5": 0.11208, "loss_box_dn_5": 0.66184, "loss_dense_depth": 0.70521, "loss": 24.02425, "grad_norm": 40.31718, "time": 1.69639} -{"mode": "train", "epoch": 1, "iter": 347, "lr": 0.00024, "memory": 49164, "data_time": 0.07548, "loss_cls_0": 0.71297, "loss_box_0": 1.5889, "loss_cns_0": 0.63385, "loss_yns_0": 0.14029, "loss_cls_1": 0.78896, "loss_box_1": 1.45045, "loss_cns_1": 0.66599, "loss_yns_1": 0.13791, "loss_cls_2": 0.79952, "loss_box_2": 1.43258, "loss_cns_2": 0.6731, "loss_yns_2": 0.13733, "loss_cls_3": 0.81405, "loss_box_3": 1.42212, "loss_cns_3": 0.66765, "loss_yns_3": 0.1379, "loss_cls_4": 0.82489, "loss_box_4": 1.41449, "loss_cns_4": 0.66502, "loss_yns_4": 0.13734, "loss_cls_5": 0.82339, "loss_box_5": 1.41244, "loss_cns_5": 0.66405, "loss_yns_5": 0.13685, "loss_cls_dn_0": 0.12936, "loss_box_dn_0": 0.70825, "loss_cls_dn_1": 0.09958, "loss_box_dn_1": 0.65353, "loss_cls_dn_2": 0.09815, "loss_box_dn_2": 0.64818, "loss_cls_dn_3": 0.10268, "loss_box_dn_3": 0.64529, "loss_cls_dn_4": 0.10386, "loss_box_dn_4": 0.64954, "loss_cls_dn_5": 0.10446, "loss_box_dn_5": 0.65605, "loss_dense_depth": 0.70302, "loss": 23.58402, "grad_norm": 31.36348, "time": 1.56217} -{"mode": "train", "epoch": 1, "iter": 348, "lr": 0.00024, "memory": 49164, "data_time": 0.07762, "loss_cls_0": 0.73203, "loss_box_0": 1.59513, "loss_cns_0": 0.63216, "loss_yns_0": 0.14177, "loss_cls_1": 0.78781, "loss_box_1": 1.47512, "loss_cns_1": 0.65959, "loss_yns_1": 0.1399, "loss_cls_2": 0.81156, "loss_box_2": 1.4293, "loss_cns_2": 0.66774, "loss_yns_2": 0.13737, "loss_cls_3": 0.82651, "loss_box_3": 1.41723, "loss_cns_3": 0.66338, "loss_yns_3": 0.13747, "loss_cls_4": 0.83952, "loss_box_4": 1.39951, "loss_cns_4": 0.66095, "loss_yns_4": 0.13935, "loss_cls_5": 0.8289, "loss_box_5": 1.40304, "loss_cns_5": 0.66057, "loss_yns_5": 0.13805, "loss_cls_dn_0": 0.13349, "loss_box_dn_0": 0.71662, "loss_cls_dn_1": 0.10071, "loss_box_dn_1": 0.66046, "loss_cls_dn_2": 0.10073, "loss_box_dn_2": 0.65525, "loss_cls_dn_3": 0.10578, "loss_box_dn_3": 0.65374, "loss_cls_dn_4": 0.10604, "loss_box_dn_4": 0.65238, "loss_cls_dn_5": 0.10305, "loss_box_dn_5": 0.65798, "loss_dense_depth": 0.73109, "loss": 23.70128, "grad_norm": 42.05435, "time": 1.64213} -{"mode": "train", "epoch": 1, "iter": 349, "lr": 0.00024, "memory": 49164, "data_time": 0.09513, "loss_cls_0": 0.71899, "loss_box_0": 1.59643, "loss_cns_0": 0.63156, "loss_yns_0": 0.14094, "loss_cls_1": 0.77104, "loss_box_1": 1.49711, "loss_cns_1": 0.66164, "loss_yns_1": 0.13835, "loss_cls_2": 0.79694, "loss_box_2": 1.42419, "loss_cns_2": 0.66516, "loss_yns_2": 0.13758, "loss_cls_3": 0.80656, "loss_box_3": 1.40565, "loss_cns_3": 0.66868, "loss_yns_3": 0.13817, "loss_cls_4": 0.81754, "loss_box_4": 1.40183, "loss_cns_4": 0.66685, "loss_yns_4": 0.13796, "loss_cls_5": 0.82589, "loss_box_5": 1.39766, "loss_cns_5": 0.66696, "loss_yns_5": 0.13813, "loss_cls_dn_0": 0.13736, "loss_box_dn_0": 0.7124, "loss_cls_dn_1": 0.09966, "loss_box_dn_1": 0.65635, "loss_cls_dn_2": 0.09968, "loss_box_dn_2": 0.64021, "loss_cls_dn_3": 0.10014, "loss_box_dn_3": 0.63725, "loss_cls_dn_4": 0.1039, "loss_box_dn_4": 0.63555, "loss_cls_dn_5": 0.10634, "loss_box_dn_5": 0.63577, "loss_dense_depth": 0.71205, "loss": 23.52848, "grad_norm": 31.08712, "time": 1.60499} -{"mode": "train", "epoch": 1, "iter": 350, "lr": 0.00024, "memory": 49164, "data_time": 0.07948, "loss_cls_0": 0.70929, "loss_box_0": 1.57559, "loss_cns_0": 0.63628, "loss_yns_0": 0.14036, "loss_cls_1": 0.76588, "loss_box_1": 1.4315, "loss_cns_1": 0.66609, "loss_yns_1": 0.13809, "loss_cls_2": 0.79835, "loss_box_2": 1.42658, "loss_cns_2": 0.67229, "loss_yns_2": 0.13782, "loss_cls_3": 0.81273, "loss_box_3": 1.42331, "loss_cns_3": 0.67781, "loss_yns_3": 0.13817, "loss_cls_4": 0.82028, "loss_box_4": 1.42447, "loss_cns_4": 0.67193, "loss_yns_4": 0.13764, "loss_cls_5": 0.81431, "loss_box_5": 1.39571, "loss_cns_5": 0.66812, "loss_yns_5": 0.13778, "loss_cls_dn_0": 0.12821, "loss_box_dn_0": 0.7174, "loss_cls_dn_1": 0.09911, "loss_box_dn_1": 0.64239, "loss_cls_dn_2": 0.10184, "loss_box_dn_2": 0.6376, "loss_cls_dn_3": 0.1005, "loss_box_dn_3": 0.6357, "loss_cls_dn_4": 0.1046, "loss_box_dn_4": 0.64353, "loss_cls_dn_5": 0.10502, "loss_box_dn_5": 0.63612, "loss_dense_depth": 0.70463, "loss": 23.47705, "grad_norm": 46.1965, "time": 1.59563} -{"mode": "train", "epoch": 1, "iter": 351, "lr": 0.00024, "memory": 49164, "data_time": 0.08095, "loss_cls_0": 0.74611, "loss_box_0": 1.59166, "loss_cns_0": 0.6324, "loss_yns_0": 0.14331, "loss_cls_1": 0.8019, "loss_box_1": 1.43543, "loss_cns_1": 0.66679, "loss_yns_1": 0.14249, "loss_cls_2": 0.83434, "loss_box_2": 1.43412, "loss_cns_2": 0.67193, "loss_yns_2": 0.14244, "loss_cls_3": 0.8556, "loss_box_3": 1.41467, "loss_cns_3": 0.67141, "loss_yns_3": 0.1427, "loss_cls_4": 0.85499, "loss_box_4": 1.41433, "loss_cns_4": 0.66699, "loss_yns_4": 0.14318, "loss_cls_5": 0.84461, "loss_box_5": 1.3947, "loss_cns_5": 0.66588, "loss_yns_5": 0.14334, "loss_cls_dn_0": 0.12918, "loss_box_dn_0": 0.70891, "loss_cls_dn_1": 0.10136, "loss_box_dn_1": 0.63308, "loss_cls_dn_2": 0.10288, "loss_box_dn_2": 0.63157, "loss_cls_dn_3": 0.10481, "loss_box_dn_3": 0.62431, "loss_cls_dn_4": 0.10813, "loss_box_dn_4": 0.63153, "loss_cls_dn_5": 0.10767, "loss_box_dn_5": 0.63054, "loss_dense_depth": 0.72294, "loss": 23.69222, "grad_norm": 41.64109, "time": 1.55919} -{"mode": "train", "epoch": 1, "iter": 352, "lr": 0.00024, "memory": 49164, "data_time": 0.07783, "loss_cls_0": 0.73781, "loss_box_0": 1.59921, "loss_cns_0": 0.63194, "loss_yns_0": 0.14606, "loss_cls_1": 0.81421, "loss_box_1": 1.44477, "loss_cns_1": 0.66069, "loss_yns_1": 0.14501, "loss_cls_2": 0.82892, "loss_box_2": 1.42888, "loss_cns_2": 0.66318, "loss_yns_2": 0.14399, "loss_cls_3": 0.83685, "loss_box_3": 1.39743, "loss_cns_3": 0.66199, "loss_yns_3": 0.14371, "loss_cls_4": 0.84175, "loss_box_4": 1.39918, "loss_cns_4": 0.66118, "loss_yns_4": 0.1442, "loss_cls_5": 0.84289, "loss_box_5": 1.41258, "loss_cns_5": 0.66151, "loss_yns_5": 0.14413, "loss_cls_dn_0": 0.13214, "loss_box_dn_0": 0.72334, "loss_cls_dn_1": 0.10204, "loss_box_dn_1": 0.64508, "loss_cls_dn_2": 0.10328, "loss_box_dn_2": 0.64169, "loss_cls_dn_3": 0.10437, "loss_box_dn_3": 0.63107, "loss_cls_dn_4": 0.10536, "loss_box_dn_4": 0.63692, "loss_cls_dn_5": 0.10754, "loss_box_dn_5": 0.64534, "loss_dense_depth": 0.73173, "loss": 23.70198, "grad_norm": 35.34069, "time": 1.56313} -{"mode": "train", "epoch": 1, "iter": 353, "lr": 0.00024, "memory": 49164, "data_time": 0.07389, "loss_cls_0": 0.72698, "loss_box_0": 1.57786, "loss_cns_0": 0.63673, "loss_yns_0": 0.14544, "loss_cls_1": 0.79481, "loss_box_1": 1.39841, "loss_cns_1": 0.66348, "loss_yns_1": 0.14324, "loss_cls_2": 0.81602, "loss_box_2": 1.38391, "loss_cns_2": 0.66425, "loss_yns_2": 0.14294, "loss_cls_3": 0.82099, "loss_box_3": 1.36765, "loss_cns_3": 0.66309, "loss_yns_3": 0.14356, "loss_cls_4": 0.8222, "loss_box_4": 1.37731, "loss_cns_4": 0.66493, "loss_yns_4": 0.14274, "loss_cls_5": 0.82735, "loss_box_5": 1.35884, "loss_cns_5": 0.66383, "loss_yns_5": 0.14285, "loss_cls_dn_0": 0.13559, "loss_box_dn_0": 0.71818, "loss_cls_dn_1": 0.10169, "loss_box_dn_1": 0.63168, "loss_cls_dn_2": 0.10512, "loss_box_dn_2": 0.62266, "loss_cls_dn_3": 0.10186, "loss_box_dn_3": 0.62081, "loss_cls_dn_4": 0.10485, "loss_box_dn_4": 0.6274, "loss_cls_dn_5": 0.11035, "loss_box_dn_5": 0.62241, "loss_dense_depth": 0.68796, "loss": 23.27997, "grad_norm": 32.38664, "time": 1.57196} -{"mode": "train", "epoch": 1, "iter": 354, "lr": 0.00024, "memory": 49164, "data_time": 0.11406, "loss_cls_0": 0.75712, "loss_box_0": 1.60247, "loss_cns_0": 0.63344, "loss_yns_0": 0.14374, "loss_cls_1": 0.81153, "loss_box_1": 1.44225, "loss_cns_1": 0.65781, "loss_yns_1": 0.14155, "loss_cls_2": 0.81707, "loss_box_2": 1.38712, "loss_cns_2": 0.65871, "loss_yns_2": 0.13916, "loss_cls_3": 0.82358, "loss_box_3": 1.37907, "loss_cns_3": 0.65775, "loss_yns_3": 0.13918, "loss_cls_4": 0.82986, "loss_box_4": 1.38115, "loss_cns_4": 0.6618, "loss_yns_4": 0.14026, "loss_cls_5": 0.82918, "loss_box_5": 1.37448, "loss_cns_5": 0.65943, "loss_yns_5": 0.14076, "loss_cls_dn_0": 0.12997, "loss_box_dn_0": 0.72354, "loss_cls_dn_1": 0.09903, "loss_box_dn_1": 0.63467, "loss_cls_dn_2": 0.09987, "loss_box_dn_2": 0.61395, "loss_cls_dn_3": 0.09768, "loss_box_dn_3": 0.61355, "loss_cls_dn_4": 0.09881, "loss_box_dn_4": 0.61604, "loss_cls_dn_5": 0.10172, "loss_box_dn_5": 0.61727, "loss_dense_depth": 0.70804, "loss": 23.36263, "grad_norm": 36.34373, "time": 1.60763} -{"mode": "train", "epoch": 1, "iter": 355, "lr": 0.00024, "memory": 49164, "data_time": 0.07307, "loss_cls_0": 0.73185, "loss_box_0": 1.57659, "loss_cns_0": 0.63244, "loss_yns_0": 0.14155, "loss_cls_1": 0.8218, "loss_box_1": 1.43847, "loss_cns_1": 0.65895, "loss_yns_1": 0.14139, "loss_cls_2": 0.82154, "loss_box_2": 1.39714, "loss_cns_2": 0.65882, "loss_yns_2": 0.14083, "loss_cls_3": 0.82443, "loss_box_3": 1.40312, "loss_cns_3": 0.66129, "loss_yns_3": 0.14097, "loss_cls_4": 0.82946, "loss_box_4": 1.39655, "loss_cns_4": 0.66146, "loss_yns_4": 0.14106, "loss_cls_5": 0.82342, "loss_box_5": 1.38912, "loss_cns_5": 0.65793, "loss_yns_5": 0.14051, "loss_cls_dn_0": 0.13384, "loss_box_dn_0": 0.71663, "loss_cls_dn_1": 0.09912, "loss_box_dn_1": 0.63307, "loss_cls_dn_2": 0.09721, "loss_box_dn_2": 0.61629, "loss_cls_dn_3": 0.09755, "loss_box_dn_3": 0.61774, "loss_cls_dn_4": 0.10001, "loss_box_dn_4": 0.61894, "loss_cls_dn_5": 0.10209, "loss_box_dn_5": 0.61746, "loss_dense_depth": 0.69183, "loss": 23.37249, "grad_norm": 30.74401, "time": 1.59157} -{"mode": "train", "epoch": 1, "iter": 356, "lr": 0.00024, "memory": 49164, "data_time": 0.09231, "loss_cls_0": 0.75149, "loss_box_0": 1.56368, "loss_cns_0": 0.63665, "loss_yns_0": 0.14557, "loss_cls_1": 0.82482, "loss_box_1": 1.4449, "loss_cns_1": 0.66463, "loss_yns_1": 0.14463, "loss_cls_2": 0.82784, "loss_box_2": 1.41708, "loss_cns_2": 0.66252, "loss_yns_2": 0.14512, "loss_cls_3": 0.83761, "loss_box_3": 1.40414, "loss_cns_3": 0.66541, "loss_yns_3": 0.1446, "loss_cls_4": 0.84376, "loss_box_4": 1.41144, "loss_cns_4": 0.66377, "loss_yns_4": 0.145, "loss_cls_5": 0.85668, "loss_box_5": 1.38782, "loss_cns_5": 0.66091, "loss_yns_5": 0.14582, "loss_cls_dn_0": 0.1362, "loss_box_dn_0": 0.72363, "loss_cls_dn_1": 0.10073, "loss_box_dn_1": 0.61684, "loss_cls_dn_2": 0.09886, "loss_box_dn_2": 0.60932, "loss_cls_dn_3": 0.10144, "loss_box_dn_3": 0.60804, "loss_cls_dn_4": 0.10112, "loss_box_dn_4": 0.61005, "loss_cls_dn_5": 0.10704, "loss_box_dn_5": 0.60491, "loss_dense_depth": 0.72334, "loss": 23.53744, "grad_norm": 44.30978, "time": 1.57881} -{"mode": "train", "epoch": 1, "iter": 357, "lr": 0.00024, "memory": 49164, "data_time": 0.07881, "loss_cls_0": 0.76339, "loss_box_0": 1.57243, "loss_cns_0": 0.63003, "loss_yns_0": 0.14222, "loss_cls_1": 0.83306, "loss_box_1": 1.50709, "loss_cns_1": 0.65781, "loss_yns_1": 0.14295, "loss_cls_2": 0.80798, "loss_box_2": 1.46521, "loss_cns_2": 0.6599, "loss_yns_2": 0.14209, "loss_cls_3": 0.82678, "loss_box_3": 1.43089, "loss_cns_3": 0.65832, "loss_yns_3": 0.14208, "loss_cls_4": 0.83113, "loss_box_4": 1.42816, "loss_cns_4": 0.6589, "loss_yns_4": 0.14171, "loss_cls_5": 0.84043, "loss_box_5": 1.42755, "loss_cns_5": 0.66094, "loss_yns_5": 0.14231, "loss_cls_dn_0": 0.13611, "loss_box_dn_0": 0.72147, "loss_cls_dn_1": 0.09982, "loss_box_dn_1": 0.63953, "loss_cls_dn_2": 0.09741, "loss_box_dn_2": 0.62612, "loss_cls_dn_3": 0.10176, "loss_box_dn_3": 0.61589, "loss_cls_dn_4": 0.10133, "loss_box_dn_4": 0.61141, "loss_cls_dn_5": 0.10367, "loss_box_dn_5": 0.61322, "loss_dense_depth": 0.68708, "loss": 23.66818, "grad_norm": 35.24477, "time": 1.57065} -{"mode": "train", "epoch": 1, "iter": 358, "lr": 0.00024, "memory": 49164, "data_time": 0.07574, "loss_cls_0": 0.73731, "loss_box_0": 1.55656, "loss_cns_0": 0.63393, "loss_yns_0": 0.14439, "loss_cls_1": 0.79816, "loss_box_1": 1.48069, "loss_cns_1": 0.65512, "loss_yns_1": 0.14133, "loss_cls_2": 0.79403, "loss_box_2": 1.47972, "loss_cns_2": 0.65708, "loss_yns_2": 0.14108, "loss_cls_3": 0.8005, "loss_box_3": 1.46428, "loss_cns_3": 0.65573, "loss_yns_3": 0.14156, "loss_cls_4": 0.80759, "loss_box_4": 1.47661, "loss_cns_4": 0.66137, "loss_yns_4": 0.14077, "loss_cls_5": 0.81114, "loss_box_5": 1.49673, "loss_cns_5": 0.6582, "loss_yns_5": 0.1406, "loss_cls_dn_0": 0.13553, "loss_box_dn_0": 0.71731, "loss_cls_dn_1": 0.10225, "loss_box_dn_1": 0.62755, "loss_cls_dn_2": 0.09932, "loss_box_dn_2": 0.62633, "loss_cls_dn_3": 0.10087, "loss_box_dn_3": 0.62254, "loss_cls_dn_4": 0.10164, "loss_box_dn_4": 0.62774, "loss_cls_dn_5": 0.10461, "loss_box_dn_5": 0.63916, "loss_dense_depth": 0.68933, "loss": 23.66868, "grad_norm": 45.15971, "time": 1.57686} -{"mode": "train", "epoch": 1, "iter": 359, "lr": 0.00024, "memory": 49164, "data_time": 0.22122, "loss_cls_0": 0.70754, "loss_box_0": 1.5397, "loss_cns_0": 0.64164, "loss_yns_0": 0.13789, "loss_cls_1": 0.76924, "loss_box_1": 1.42534, "loss_cns_1": 0.66222, "loss_yns_1": 0.1387, "loss_cls_2": 0.79219, "loss_box_2": 1.41774, "loss_cns_2": 0.66344, "loss_yns_2": 0.138, "loss_cls_3": 0.78748, "loss_box_3": 1.39805, "loss_cns_3": 0.66791, "loss_yns_3": 0.13796, "loss_cls_4": 0.7882, "loss_box_4": 1.40297, "loss_cns_4": 0.66685, "loss_yns_4": 0.13746, "loss_cls_5": 0.79273, "loss_box_5": 1.41264, "loss_cns_5": 0.66396, "loss_yns_5": 0.13764, "loss_cls_dn_0": 0.13405, "loss_box_dn_0": 0.71906, "loss_cls_dn_1": 0.10324, "loss_box_dn_1": 0.63105, "loss_cls_dn_2": 0.1034, "loss_box_dn_2": 0.62702, "loss_cls_dn_3": 0.10142, "loss_box_dn_3": 0.62278, "loss_cls_dn_4": 0.10391, "loss_box_dn_4": 0.62827, "loss_cls_dn_5": 0.10946, "loss_box_dn_5": 0.63504, "loss_dense_depth": 0.67186, "loss": 23.21806, "grad_norm": 41.16728, "time": 1.69159} -{"mode": "train", "epoch": 1, "iter": 360, "lr": 0.00024, "memory": 49164, "data_time": 0.09766, "loss_cls_0": 0.74024, "loss_box_0": 1.56468, "loss_cns_0": 0.63957, "loss_yns_0": 0.14084, "loss_cls_1": 0.79181, "loss_box_1": 1.46283, "loss_cns_1": 0.66356, "loss_yns_1": 0.13998, "loss_cls_2": 0.79842, "loss_box_2": 1.43522, "loss_cns_2": 0.66467, "loss_yns_2": 0.13989, "loss_cls_3": 0.79676, "loss_box_3": 1.42714, "loss_cns_3": 0.66889, "loss_yns_3": 0.13931, "loss_cls_4": 0.80156, "loss_box_4": 1.44189, "loss_cns_4": 0.66707, "loss_yns_4": 0.13852, "loss_cls_5": 0.81151, "loss_box_5": 1.45501, "loss_cns_5": 0.66688, "loss_yns_5": 0.13885, "loss_cls_dn_0": 0.13264, "loss_box_dn_0": 0.71791, "loss_cls_dn_1": 0.10289, "loss_box_dn_1": 0.63562, "loss_cls_dn_2": 0.10406, "loss_box_dn_2": 0.61569, "loss_cls_dn_3": 0.1021, "loss_box_dn_3": 0.6164, "loss_cls_dn_4": 0.1049, "loss_box_dn_4": 0.62231, "loss_cls_dn_5": 0.10688, "loss_box_dn_5": 0.63057, "loss_dense_depth": 0.68859, "loss": 23.51565, "grad_norm": 38.03099, "time": 1.59187} -{"mode": "train", "epoch": 1, "iter": 361, "lr": 0.00024, "memory": 49164, "data_time": 0.08195, "loss_cls_0": 0.73862, "loss_box_0": 1.55366, "loss_cns_0": 0.64064, "loss_yns_0": 0.14459, "loss_cls_1": 0.80711, "loss_box_1": 1.44505, "loss_cns_1": 0.66348, "loss_yns_1": 0.14251, "loss_cls_2": 0.79931, "loss_box_2": 1.42022, "loss_cns_2": 0.66494, "loss_yns_2": 0.14224, "loss_cls_3": 0.80992, "loss_box_3": 1.41031, "loss_cns_3": 0.66531, "loss_yns_3": 0.14148, "loss_cls_4": 0.81152, "loss_box_4": 1.443, "loss_cns_4": 0.66611, "loss_yns_4": 0.14059, "loss_cls_5": 0.82292, "loss_box_5": 1.46572, "loss_cns_5": 0.66765, "loss_yns_5": 0.14128, "loss_cls_dn_0": 0.13056, "loss_box_dn_0": 0.71298, "loss_cls_dn_1": 0.10266, "loss_box_dn_1": 0.62764, "loss_cls_dn_2": 0.10261, "loss_box_dn_2": 0.61028, "loss_cls_dn_3": 0.10097, "loss_box_dn_3": 0.61082, "loss_cls_dn_4": 0.10246, "loss_box_dn_4": 0.62052, "loss_cls_dn_5": 0.10468, "loss_box_dn_5": 0.63328, "loss_dense_depth": 0.68959, "loss": 23.49723, "grad_norm": 33.81722, "time": 1.649} -{"mode": "train", "epoch": 1, "iter": 362, "lr": 0.00024, "memory": 49164, "data_time": 0.10053, "loss_cls_0": 0.74624, "loss_box_0": 1.55384, "loss_cns_0": 0.63358, "loss_yns_0": 0.14058, "loss_cls_1": 0.81127, "loss_box_1": 1.41822, "loss_cns_1": 0.65736, "loss_yns_1": 0.141, "loss_cls_2": 0.8206, "loss_box_2": 1.389, "loss_cns_2": 0.66082, "loss_yns_2": 0.14037, "loss_cls_3": 0.82422, "loss_box_3": 1.38751, "loss_cns_3": 0.66018, "loss_yns_3": 0.14073, "loss_cls_4": 0.82444, "loss_box_4": 1.39024, "loss_cns_4": 0.65936, "loss_yns_4": 0.14079, "loss_cls_5": 0.83025, "loss_box_5": 1.4021, "loss_cns_5": 0.66025, "loss_yns_5": 0.14065, "loss_cls_dn_0": 0.13602, "loss_box_dn_0": 0.71728, "loss_cls_dn_1": 0.10147, "loss_box_dn_1": 0.62048, "loss_cls_dn_2": 0.10142, "loss_box_dn_2": 0.61196, "loss_cls_dn_3": 0.09845, "loss_box_dn_3": 0.61134, "loss_cls_dn_4": 0.09845, "loss_box_dn_4": 0.61347, "loss_cls_dn_5": 0.10184, "loss_box_dn_5": 0.62109, "loss_dense_depth": 0.70203, "loss": 23.3089, "grad_norm": 34.21217, "time": 1.65429} -{"mode": "train", "epoch": 1, "iter": 363, "lr": 0.00024, "memory": 49164, "data_time": 0.08608, "loss_cls_0": 0.7388, "loss_box_0": 1.56047, "loss_cns_0": 0.63855, "loss_yns_0": 0.13912, "loss_cls_1": 0.79917, "loss_box_1": 1.4443, "loss_cns_1": 0.65949, "loss_yns_1": 0.14042, "loss_cls_2": 0.81319, "loss_box_2": 1.40119, "loss_cns_2": 0.66235, "loss_yns_2": 0.14027, "loss_cls_3": 0.82421, "loss_box_3": 1.38498, "loss_cns_3": 0.6646, "loss_yns_3": 0.14064, "loss_cls_4": 0.81355, "loss_box_4": 1.401, "loss_cns_4": 0.66406, "loss_yns_4": 0.13916, "loss_cls_5": 0.82035, "loss_box_5": 1.39667, "loss_cns_5": 0.66196, "loss_yns_5": 0.14027, "loss_cls_dn_0": 0.12846, "loss_box_dn_0": 0.72532, "loss_cls_dn_1": 0.09605, "loss_box_dn_1": 0.63826, "loss_cls_dn_2": 0.09615, "loss_box_dn_2": 0.62294, "loss_cls_dn_3": 0.09502, "loss_box_dn_3": 0.62359, "loss_cls_dn_4": 0.09334, "loss_box_dn_4": 0.62965, "loss_cls_dn_5": 0.09572, "loss_box_dn_5": 0.6338, "loss_dense_depth": 0.69496, "loss": 23.362, "grad_norm": 31.93637, "time": 1.60457} -{"mode": "train", "epoch": 1, "iter": 364, "lr": 0.00024, "memory": 49164, "data_time": 0.07852, "loss_cls_0": 0.7374, "loss_box_0": 1.57447, "loss_cns_0": 0.64074, "loss_yns_0": 0.13791, "loss_cls_1": 0.79691, "loss_box_1": 1.46654, "loss_cns_1": 0.65968, "loss_yns_1": 0.14031, "loss_cls_2": 0.79983, "loss_box_2": 1.45739, "loss_cns_2": 0.66258, "loss_yns_2": 0.13913, "loss_cls_3": 0.82483, "loss_box_3": 1.41401, "loss_cns_3": 0.66111, "loss_yns_3": 0.13728, "loss_cls_4": 0.8122, "loss_box_4": 1.41279, "loss_cns_4": 0.66135, "loss_yns_4": 0.13723, "loss_cls_5": 0.81776, "loss_box_5": 1.41652, "loss_cns_5": 0.66115, "loss_yns_5": 0.13844, "loss_cls_dn_0": 0.12994, "loss_box_dn_0": 0.72613, "loss_cls_dn_1": 0.0963, "loss_box_dn_1": 0.64469, "loss_cls_dn_2": 0.09705, "loss_box_dn_2": 0.63571, "loss_cls_dn_3": 0.09853, "loss_box_dn_3": 0.62602, "loss_cls_dn_4": 0.09622, "loss_box_dn_4": 0.62559, "loss_cls_dn_5": 0.09822, "loss_box_dn_5": 0.62989, "loss_dense_depth": 0.68042, "loss": 23.49229, "grad_norm": 36.30132, "time": 1.62034} -{"mode": "train", "epoch": 1, "iter": 365, "lr": 0.00025, "memory": 49164, "data_time": 0.07924, "loss_cls_0": 0.73276, "loss_box_0": 1.58629, "loss_cns_0": 0.64106, "loss_yns_0": 0.13754, "loss_cls_1": 0.80647, "loss_box_1": 1.46848, "loss_cns_1": 0.66174, "loss_yns_1": 0.13735, "loss_cls_2": 0.8134, "loss_box_2": 1.44246, "loss_cns_2": 0.66597, "loss_yns_2": 0.13746, "loss_cls_3": 0.82091, "loss_box_3": 1.42216, "loss_cns_3": 0.66167, "loss_yns_3": 0.13696, "loss_cls_4": 0.80862, "loss_box_4": 1.41914, "loss_cns_4": 0.66228, "loss_yns_4": 0.13814, "loss_cls_5": 0.81657, "loss_box_5": 1.43482, "loss_cns_5": 0.66378, "loss_yns_5": 0.13828, "loss_cls_dn_0": 0.12703, "loss_box_dn_0": 0.715, "loss_cls_dn_1": 0.09917, "loss_box_dn_1": 0.63415, "loss_cls_dn_2": 0.09786, "loss_box_dn_2": 0.61922, "loss_cls_dn_3": 0.09916, "loss_box_dn_3": 0.61039, "loss_cls_dn_4": 0.09834, "loss_box_dn_4": 0.60821, "loss_cls_dn_5": 0.10034, "loss_box_dn_5": 0.61758, "loss_dense_depth": 0.67631, "loss": 23.45705, "grad_norm": 24.14205, "time": 1.63803} -{"mode": "train", "epoch": 1, "iter": 366, "lr": 0.00025, "memory": 49164, "data_time": 0.07876, "loss_cls_0": 0.69456, "loss_box_0": 1.55916, "loss_cns_0": 0.63926, "loss_yns_0": 0.13613, "loss_cls_1": 0.76272, "loss_box_1": 1.43293, "loss_cns_1": 0.6624, "loss_yns_1": 0.13413, "loss_cls_2": 0.7683, "loss_box_2": 1.39711, "loss_cns_2": 0.66059, "loss_yns_2": 0.13487, "loss_cls_3": 0.76531, "loss_box_3": 1.38716, "loss_cns_3": 0.65884, "loss_yns_3": 0.13537, "loss_cls_4": 0.76559, "loss_box_4": 1.39626, "loss_cns_4": 0.65955, "loss_yns_4": 0.13655, "loss_cls_5": 0.77667, "loss_box_5": 1.3945, "loss_cns_5": 0.66229, "loss_yns_5": 0.138, "loss_cls_dn_0": 0.12404, "loss_box_dn_0": 0.71477, "loss_cls_dn_1": 0.09455, "loss_box_dn_1": 0.62977, "loss_cls_dn_2": 0.09365, "loss_box_dn_2": 0.61217, "loss_cls_dn_3": 0.09292, "loss_box_dn_3": 0.60584, "loss_cls_dn_4": 0.09344, "loss_box_dn_4": 0.60869, "loss_cls_dn_5": 0.09502, "loss_box_dn_5": 0.61043, "loss_dense_depth": 0.67977, "loss": 22.91331, "grad_norm": 29.16234, "time": 1.64771} -{"mode": "train", "epoch": 1, "iter": 367, "lr": 0.00025, "memory": 49164, "data_time": 0.10078, "loss_cls_0": 0.71707, "loss_box_0": 1.56717, "loss_cns_0": 0.63783, "loss_yns_0": 0.13771, "loss_cls_1": 0.77827, "loss_box_1": 1.44081, "loss_cns_1": 0.66512, "loss_yns_1": 0.13818, "loss_cls_2": 0.78872, "loss_box_2": 1.39642, "loss_cns_2": 0.66167, "loss_yns_2": 0.13629, "loss_cls_3": 0.79141, "loss_box_3": 1.40695, "loss_cns_3": 0.66375, "loss_yns_3": 0.13742, "loss_cls_4": 0.80124, "loss_box_4": 1.40906, "loss_cns_4": 0.66276, "loss_yns_4": 0.13612, "loss_cls_5": 0.80709, "loss_box_5": 1.42279, "loss_cns_5": 0.66606, "loss_yns_5": 0.13802, "loss_cls_dn_0": 0.13112, "loss_box_dn_0": 0.71429, "loss_cls_dn_1": 0.09619, "loss_box_dn_1": 0.62796, "loss_cls_dn_2": 0.09627, "loss_box_dn_2": 0.61327, "loss_cls_dn_3": 0.0961, "loss_box_dn_3": 0.61283, "loss_cls_dn_4": 0.10011, "loss_box_dn_4": 0.6153, "loss_cls_dn_5": 0.09866, "loss_box_dn_5": 0.62346, "loss_dense_depth": 0.71436, "loss": 23.24788, "grad_norm": 34.17976, "time": 1.58928} -{"mode": "train", "epoch": 1, "iter": 368, "lr": 0.00025, "memory": 49164, "data_time": 0.07377, "loss_cls_0": 0.70582, "loss_box_0": 1.58558, "loss_cns_0": 0.63532, "loss_yns_0": 0.13604, "loss_cls_1": 0.77076, "loss_box_1": 1.44396, "loss_cns_1": 0.66348, "loss_yns_1": 0.13704, "loss_cls_2": 0.78669, "loss_box_2": 1.39636, "loss_cns_2": 0.66028, "loss_yns_2": 0.13255, "loss_cls_3": 0.78524, "loss_box_3": 1.38407, "loss_cns_3": 0.6617, "loss_yns_3": 0.13273, "loss_cls_4": 0.78808, "loss_box_4": 1.39, "loss_cns_4": 0.6642, "loss_yns_4": 0.13368, "loss_cls_5": 0.79285, "loss_box_5": 1.39951, "loss_cns_5": 0.66637, "loss_yns_5": 0.13384, "loss_cls_dn_0": 0.12687, "loss_box_dn_0": 0.71504, "loss_cls_dn_1": 0.09834, "loss_box_dn_1": 0.62746, "loss_cls_dn_2": 0.09688, "loss_box_dn_2": 0.61627, "loss_cls_dn_3": 0.09762, "loss_box_dn_3": 0.61152, "loss_cls_dn_4": 0.09895, "loss_box_dn_4": 0.61405, "loss_cls_dn_5": 0.10024, "loss_box_dn_5": 0.62105, "loss_dense_depth": 0.74899, "loss": 23.15946, "grad_norm": 27.89022, "time": 1.5855} -{"mode": "train", "epoch": 1, "iter": 369, "lr": 0.00025, "memory": 49164, "data_time": 0.07576, "loss_cls_0": 0.69613, "loss_box_0": 1.55671, "loss_cns_0": 0.64225, "loss_yns_0": 0.13425, "loss_cls_1": 0.76437, "loss_box_1": 1.42496, "loss_cns_1": 0.66744, "loss_yns_1": 0.12993, "loss_cls_2": 0.78371, "loss_box_2": 1.39086, "loss_cns_2": 0.66978, "loss_yns_2": 0.13006, "loss_cls_3": 0.77358, "loss_box_3": 1.36729, "loss_cns_3": 0.66749, "loss_yns_3": 0.13155, "loss_cls_4": 0.78836, "loss_box_4": 1.37475, "loss_cns_4": 0.66696, "loss_yns_4": 0.13351, "loss_cls_5": 0.79332, "loss_box_5": 1.37501, "loss_cns_5": 0.66775, "loss_yns_5": 0.13287, "loss_cls_dn_0": 0.11986, "loss_box_dn_0": 0.70493, "loss_cls_dn_1": 0.09678, "loss_box_dn_1": 0.62814, "loss_cls_dn_2": 0.09758, "loss_box_dn_2": 0.61533, "loss_cls_dn_3": 0.09983, "loss_box_dn_3": 0.60709, "loss_cls_dn_4": 0.09786, "loss_box_dn_4": 0.61217, "loss_cls_dn_5": 0.09966, "loss_box_dn_5": 0.61783, "loss_dense_depth": 0.84814, "loss": 23.10808, "grad_norm": 33.07095, "time": 1.56834} -{"mode": "train", "epoch": 1, "iter": 370, "lr": 0.00025, "memory": 49164, "data_time": 0.0751, "loss_cls_0": 0.70383, "loss_box_0": 1.5951, "loss_cns_0": 0.63921, "loss_yns_0": 0.13491, "loss_cls_1": 0.77966, "loss_box_1": 1.41392, "loss_cns_1": 0.66839, "loss_yns_1": 0.13506, "loss_cls_2": 0.79929, "loss_box_2": 1.37012, "loss_cns_2": 0.66746, "loss_yns_2": 0.13333, "loss_cls_3": 0.79411, "loss_box_3": 1.36106, "loss_cns_3": 0.66601, "loss_yns_3": 0.1342, "loss_cls_4": 0.79719, "loss_box_4": 1.37714, "loss_cns_4": 0.66734, "loss_yns_4": 0.13385, "loss_cls_5": 0.80477, "loss_box_5": 1.36486, "loss_cns_5": 0.66663, "loss_yns_5": 0.1337, "loss_cls_dn_0": 0.13172, "loss_box_dn_0": 0.72749, "loss_cls_dn_1": 0.09943, "loss_box_dn_1": 0.63357, "loss_cls_dn_2": 0.09868, "loss_box_dn_2": 0.61562, "loss_cls_dn_3": 0.09868, "loss_box_dn_3": 0.61146, "loss_cls_dn_4": 0.0985, "loss_box_dn_4": 0.61419, "loss_cls_dn_5": 0.09935, "loss_box_dn_5": 0.61072, "loss_dense_depth": 0.78938, "loss": 23.16993, "grad_norm": 28.13429, "time": 1.56712} -{"mode": "train", "epoch": 1, "iter": 371, "lr": 0.00025, "memory": 49164, "data_time": 0.07493, "loss_cls_0": 0.68685, "loss_box_0": 1.54021, "loss_cns_0": 0.64132, "loss_yns_0": 0.1348, "loss_cls_1": 0.7631, "loss_box_1": 1.41015, "loss_cns_1": 0.66495, "loss_yns_1": 0.13682, "loss_cls_2": 0.77983, "loss_box_2": 1.36943, "loss_cns_2": 0.66366, "loss_yns_2": 0.13691, "loss_cls_3": 0.77988, "loss_box_3": 1.35039, "loss_cns_3": 0.66109, "loss_yns_3": 0.13659, "loss_cls_4": 0.77818, "loss_box_4": 1.36268, "loss_cns_4": 0.66286, "loss_yns_4": 0.13605, "loss_cls_5": 0.78493, "loss_box_5": 1.3518, "loss_cns_5": 0.66, "loss_yns_5": 0.13535, "loss_cls_dn_0": 0.11877, "loss_box_dn_0": 0.7149, "loss_cls_dn_1": 0.09588, "loss_box_dn_1": 0.62226, "loss_cls_dn_2": 0.09419, "loss_box_dn_2": 0.60629, "loss_cls_dn_3": 0.09466, "loss_box_dn_3": 0.59862, "loss_cls_dn_4": 0.0944, "loss_box_dn_4": 0.60004, "loss_cls_dn_5": 0.09598, "loss_box_dn_5": 0.59653, "loss_dense_depth": 0.71675, "loss": 22.77709, "grad_norm": 26.16096, "time": 1.5604} -{"mode": "train", "epoch": 1, "iter": 372, "lr": 0.00025, "memory": 49164, "data_time": 0.08957, "loss_cls_0": 0.71528, "loss_box_0": 1.57139, "loss_cns_0": 0.63044, "loss_yns_0": 0.1377, "loss_cls_1": 0.77771, "loss_box_1": 1.45374, "loss_cns_1": 0.65943, "loss_yns_1": 0.13708, "loss_cls_2": 0.79614, "loss_box_2": 1.41829, "loss_cns_2": 0.6585, "loss_yns_2": 0.13683, "loss_cls_3": 0.79011, "loss_box_3": 1.4144, "loss_cns_3": 0.66148, "loss_yns_3": 0.1365, "loss_cls_4": 0.79137, "loss_box_4": 1.42852, "loss_cns_4": 0.65825, "loss_yns_4": 0.1382, "loss_cls_5": 0.80137, "loss_box_5": 1.42047, "loss_cns_5": 0.65631, "loss_yns_5": 0.1372, "loss_cls_dn_0": 0.12189, "loss_box_dn_0": 0.73456, "loss_cls_dn_1": 0.09576, "loss_box_dn_1": 0.62121, "loss_cls_dn_2": 0.0969, "loss_box_dn_2": 0.605, "loss_cls_dn_3": 0.09712, "loss_box_dn_3": 0.59873, "loss_cls_dn_4": 0.09517, "loss_box_dn_4": 0.60295, "loss_cls_dn_5": 0.09678, "loss_box_dn_5": 0.60536, "loss_dense_depth": 0.74273, "loss": 23.24088, "grad_norm": 36.66177, "time": 1.59487} -{"mode": "train", "epoch": 1, "iter": 373, "lr": 0.00025, "memory": 49164, "data_time": 0.08131, "loss_cls_0": 0.75031, "loss_box_0": 1.59965, "loss_cns_0": 0.63472, "loss_yns_0": 0.14213, "loss_cls_1": 0.80091, "loss_box_1": 1.45752, "loss_cns_1": 0.65629, "loss_yns_1": 0.1376, "loss_cls_2": 0.82114, "loss_box_2": 1.43657, "loss_cns_2": 0.66306, "loss_yns_2": 0.13709, "loss_cls_3": 0.82561, "loss_box_3": 1.40811, "loss_cns_3": 0.66008, "loss_yns_3": 0.13903, "loss_cls_4": 0.80952, "loss_box_4": 1.41625, "loss_cns_4": 0.66005, "loss_yns_4": 0.13918, "loss_cls_5": 0.81728, "loss_box_5": 1.422, "loss_cns_5": 0.65975, "loss_yns_5": 0.14032, "loss_cls_dn_0": 0.12611, "loss_box_dn_0": 0.72233, "loss_cls_dn_1": 0.09768, "loss_box_dn_1": 0.63088, "loss_cls_dn_2": 0.09861, "loss_box_dn_2": 0.61894, "loss_cls_dn_3": 0.10086, "loss_box_dn_3": 0.60827, "loss_cls_dn_4": 0.09565, "loss_box_dn_4": 0.60852, "loss_cls_dn_5": 0.09443, "loss_box_dn_5": 0.61144, "loss_dense_depth": 0.74913, "loss": 23.49702, "grad_norm": 30.56927, "time": 1.58972} -{"mode": "train", "epoch": 1, "iter": 374, "lr": 0.00025, "memory": 49164, "data_time": 0.07742, "loss_cls_0": 0.72695, "loss_box_0": 1.58412, "loss_cns_0": 0.62836, "loss_yns_0": 0.14141, "loss_cls_1": 0.79302, "loss_box_1": 1.42785, "loss_cns_1": 0.65829, "loss_yns_1": 0.13767, "loss_cls_2": 0.80201, "loss_box_2": 1.40593, "loss_cns_2": 0.66442, "loss_yns_2": 0.13827, "loss_cls_3": 0.80189, "loss_box_3": 1.38979, "loss_cns_3": 0.66015, "loss_yns_3": 0.13965, "loss_cls_4": 0.80238, "loss_box_4": 1.39811, "loss_cns_4": 0.65833, "loss_yns_4": 0.13934, "loss_cls_5": 0.80974, "loss_box_5": 1.40387, "loss_cns_5": 0.6591, "loss_yns_5": 0.13988, "loss_cls_dn_0": 0.12219, "loss_box_dn_0": 0.71931, "loss_cls_dn_1": 0.09468, "loss_box_dn_1": 0.62131, "loss_cls_dn_2": 0.09464, "loss_box_dn_2": 0.60811, "loss_cls_dn_3": 0.09651, "loss_box_dn_3": 0.60318, "loss_cls_dn_4": 0.09348, "loss_box_dn_4": 0.60948, "loss_cls_dn_5": 0.09287, "loss_box_dn_5": 0.61293, "loss_dense_depth": 0.72589, "loss": 23.20511, "grad_norm": 24.93296, "time": 1.57833} -{"mode": "train", "epoch": 1, "iter": 375, "lr": 0.00025, "memory": 49164, "data_time": 0.09598, "loss_cls_0": 0.73089, "loss_box_0": 1.55553, "loss_cns_0": 0.638, "loss_yns_0": 0.14101, "loss_cls_1": 0.79913, "loss_box_1": 1.41624, "loss_cns_1": 0.66344, "loss_yns_1": 0.13836, "loss_cls_2": 0.80856, "loss_box_2": 1.40484, "loss_cns_2": 0.66512, "loss_yns_2": 0.13937, "loss_cls_3": 0.80875, "loss_box_3": 1.4016, "loss_cns_3": 0.66227, "loss_yns_3": 0.13787, "loss_cls_4": 0.80989, "loss_box_4": 1.38704, "loss_cns_4": 0.662, "loss_yns_4": 0.13732, "loss_cls_5": 0.81589, "loss_box_5": 1.38938, "loss_cns_5": 0.66422, "loss_yns_5": 0.13763, "loss_cls_dn_0": 0.11936, "loss_box_dn_0": 0.70875, "loss_cls_dn_1": 0.09232, "loss_box_dn_1": 0.61325, "loss_cls_dn_2": 0.09012, "loss_box_dn_2": 0.60279, "loss_cls_dn_3": 0.09322, "loss_box_dn_3": 0.60251, "loss_cls_dn_4": 0.09076, "loss_box_dn_4": 0.5997, "loss_cls_dn_5": 0.09184, "loss_box_dn_5": 0.60161, "loss_dense_depth": 0.78683, "loss": 23.20741, "grad_norm": 32.10767, "time": 1.63044} -{"mode": "train", "epoch": 1, "iter": 376, "lr": 0.00025, "memory": 49164, "data_time": 0.09502, "loss_cls_0": 0.75774, "loss_box_0": 1.58408, "loss_cns_0": 0.63518, "loss_yns_0": 0.14224, "loss_cls_1": 0.81482, "loss_box_1": 1.42656, "loss_cns_1": 0.66304, "loss_yns_1": 0.13908, "loss_cls_2": 0.81769, "loss_box_2": 1.40545, "loss_cns_2": 0.66214, "loss_yns_2": 0.13943, "loss_cls_3": 0.83013, "loss_box_3": 1.39259, "loss_cns_3": 0.66287, "loss_yns_3": 0.13838, "loss_cls_4": 0.82106, "loss_box_4": 1.39147, "loss_cns_4": 0.66159, "loss_yns_4": 0.13829, "loss_cls_5": 0.82685, "loss_box_5": 1.40167, "loss_cns_5": 0.66179, "loss_yns_5": 0.13891, "loss_cls_dn_0": 0.12673, "loss_box_dn_0": 0.72162, "loss_cls_dn_1": 0.09475, "loss_box_dn_1": 0.63837, "loss_cls_dn_2": 0.09247, "loss_box_dn_2": 0.6239, "loss_cls_dn_3": 0.09394, "loss_box_dn_3": 0.62032, "loss_cls_dn_4": 0.09175, "loss_box_dn_4": 0.61937, "loss_cls_dn_5": 0.0928, "loss_box_dn_5": 0.62597, "loss_dense_depth": 0.79328, "loss": 23.48832, "grad_norm": 29.51187, "time": 1.65111} -{"mode": "train", "epoch": 1, "iter": 377, "lr": 0.00025, "memory": 49164, "data_time": 0.07832, "loss_cls_0": 0.74263, "loss_box_0": 1.53833, "loss_cns_0": 0.63832, "loss_yns_0": 0.14261, "loss_cls_1": 0.79239, "loss_box_1": 1.39887, "loss_cns_1": 0.66286, "loss_yns_1": 0.13878, "loss_cls_2": 0.79438, "loss_box_2": 1.38028, "loss_cns_2": 0.6592, "loss_yns_2": 0.13734, "loss_cls_3": 0.80218, "loss_box_3": 1.36845, "loss_cns_3": 0.66462, "loss_yns_3": 0.13811, "loss_cls_4": 0.79865, "loss_box_4": 1.37443, "loss_cns_4": 0.66328, "loss_yns_4": 0.13885, "loss_cls_5": 0.80663, "loss_box_5": 1.36695, "loss_cns_5": 0.66432, "loss_yns_5": 0.1384, "loss_cls_dn_0": 0.11967, "loss_box_dn_0": 0.72827, "loss_cls_dn_1": 0.0941, "loss_box_dn_1": 0.6453, "loss_cls_dn_2": 0.09334, "loss_box_dn_2": 0.6263, "loss_cls_dn_3": 0.0962, "loss_box_dn_3": 0.6183, "loss_cls_dn_4": 0.09408, "loss_box_dn_4": 0.62013, "loss_cls_dn_5": 0.09325, "loss_box_dn_5": 0.62111, "loss_dense_depth": 0.76849, "loss": 23.1694, "grad_norm": 23.79237, "time": 1.58685} -{"mode": "train", "epoch": 1, "iter": 378, "lr": 0.00025, "memory": 49164, "data_time": 0.11255, "loss_cls_0": 0.73194, "loss_box_0": 1.54724, "loss_cns_0": 0.63905, "loss_yns_0": 0.14371, "loss_cls_1": 0.78324, "loss_box_1": 1.40259, "loss_cns_1": 0.66225, "loss_yns_1": 0.14038, "loss_cls_2": 0.78204, "loss_box_2": 1.37928, "loss_cns_2": 0.66176, "loss_yns_2": 0.13955, "loss_cls_3": 0.79518, "loss_box_3": 1.3667, "loss_cns_3": 0.66189, "loss_yns_3": 0.13971, "loss_cls_4": 0.79759, "loss_box_4": 1.3721, "loss_cns_4": 0.66324, "loss_yns_4": 0.14046, "loss_cls_5": 0.80593, "loss_box_5": 1.3708, "loss_cns_5": 0.66375, "loss_yns_5": 0.14057, "loss_cls_dn_0": 0.12111, "loss_box_dn_0": 0.72044, "loss_cls_dn_1": 0.09439, "loss_box_dn_1": 0.63723, "loss_cls_dn_2": 0.09393, "loss_box_dn_2": 0.62216, "loss_cls_dn_3": 0.09979, "loss_box_dn_3": 0.62027, "loss_cls_dn_4": 0.0925, "loss_box_dn_4": 0.62252, "loss_cls_dn_5": 0.09227, "loss_box_dn_5": 0.62568, "loss_dense_depth": 0.78025, "loss": 23.15347, "grad_norm": 31.92193, "time": 1.59931} -{"mode": "train", "epoch": 1, "iter": 379, "lr": 0.00025, "memory": 49164, "data_time": 0.18982, "loss_cls_0": 0.72814, "loss_box_0": 1.56349, "loss_cns_0": 0.63746, "loss_yns_0": 0.14442, "loss_cls_1": 0.80318, "loss_box_1": 1.42525, "loss_cns_1": 0.66023, "loss_yns_1": 0.13902, "loss_cls_2": 0.80852, "loss_box_2": 1.38658, "loss_cns_2": 0.66422, "loss_yns_2": 0.13899, "loss_cls_3": 0.80999, "loss_box_3": 1.38981, "loss_cns_3": 0.65941, "loss_yns_3": 0.13945, "loss_cls_4": 0.83076, "loss_box_4": 1.38004, "loss_cns_4": 0.66127, "loss_yns_4": 0.13844, "loss_cls_5": 0.83024, "loss_box_5": 1.38413, "loss_cns_5": 0.65942, "loss_yns_5": 0.13869, "loss_cls_dn_0": 0.11678, "loss_box_dn_0": 0.71059, "loss_cls_dn_1": 0.09023, "loss_box_dn_1": 0.63277, "loss_cls_dn_2": 0.08943, "loss_box_dn_2": 0.61926, "loss_cls_dn_3": 0.0898, "loss_box_dn_3": 0.62073, "loss_cls_dn_4": 0.09334, "loss_box_dn_4": 0.62328, "loss_cls_dn_5": 0.09378, "loss_box_dn_5": 0.63221, "loss_dense_depth": 0.77598, "loss": 23.30931, "grad_norm": 39.54122, "time": 1.65554} -{"mode": "train", "epoch": 1, "iter": 380, "lr": 0.00025, "memory": 49164, "data_time": 0.08386, "loss_cls_0": 0.74613, "loss_box_0": 1.55206, "loss_cns_0": 0.6404, "loss_yns_0": 0.1401, "loss_cls_1": 0.80303, "loss_box_1": 1.4072, "loss_cns_1": 0.65907, "loss_yns_1": 0.13666, "loss_cls_2": 0.81322, "loss_box_2": 1.37198, "loss_cns_2": 0.65766, "loss_yns_2": 0.13593, "loss_cls_3": 0.82128, "loss_box_3": 1.33782, "loss_cns_3": 0.65647, "loss_yns_3": 0.13561, "loss_cls_4": 0.82814, "loss_box_4": 1.32631, "loss_cns_4": 0.65541, "loss_yns_4": 0.136, "loss_cls_5": 0.82868, "loss_box_5": 1.3325, "loss_cns_5": 0.65196, "loss_yns_5": 0.13529, "loss_cls_dn_0": 0.12047, "loss_box_dn_0": 0.71472, "loss_cls_dn_1": 0.09269, "loss_box_dn_1": 0.63535, "loss_cls_dn_2": 0.0926, "loss_box_dn_2": 0.6279, "loss_cls_dn_3": 0.0932, "loss_box_dn_3": 0.61417, "loss_cls_dn_4": 0.10075, "loss_box_dn_4": 0.61492, "loss_cls_dn_5": 0.10057, "loss_box_dn_5": 0.6225, "loss_dense_depth": 0.76636, "loss": 23.10509, "grad_norm": 30.46765, "time": 1.58619} -{"mode": "train", "epoch": 1, "iter": 381, "lr": 0.00025, "memory": 49164, "data_time": 0.08128, "loss_cls_0": 0.72705, "loss_box_0": 1.51018, "loss_cns_0": 0.63991, "loss_yns_0": 0.13929, "loss_cls_1": 0.77202, "loss_box_1": 1.40068, "loss_cns_1": 0.65931, "loss_yns_1": 0.13474, "loss_cls_2": 0.77866, "loss_box_2": 1.38942, "loss_cns_2": 0.65848, "loss_yns_2": 0.13423, "loss_cls_3": 0.79189, "loss_box_3": 1.35898, "loss_cns_3": 0.66008, "loss_yns_3": 0.134, "loss_cls_4": 0.79498, "loss_box_4": 1.37181, "loss_cns_4": 0.65899, "loss_yns_4": 0.1352, "loss_cls_5": 0.80804, "loss_box_5": 1.36758, "loss_cns_5": 0.65671, "loss_yns_5": 0.1354, "loss_cls_dn_0": 0.12132, "loss_box_dn_0": 0.71282, "loss_cls_dn_1": 0.09168, "loss_box_dn_1": 0.62099, "loss_cls_dn_2": 0.09091, "loss_box_dn_2": 0.61739, "loss_cls_dn_3": 0.09078, "loss_box_dn_3": 0.60396, "loss_cls_dn_4": 0.09126, "loss_box_dn_4": 0.60883, "loss_cls_dn_5": 0.09547, "loss_box_dn_5": 0.61402, "loss_dense_depth": 0.71359, "loss": 22.89064, "grad_norm": 35.28202, "time": 1.70229} -{"mode": "train", "epoch": 1, "iter": 382, "lr": 0.00025, "memory": 49164, "data_time": 0.08545, "loss_cls_0": 0.70568, "loss_box_0": 1.53561, "loss_cns_0": 0.63777, "loss_yns_0": 0.13863, "loss_cls_1": 0.77803, "loss_box_1": 1.40874, "loss_cns_1": 0.66815, "loss_yns_1": 0.13623, "loss_cls_2": 0.79475, "loss_box_2": 1.3983, "loss_cns_2": 0.66555, "loss_yns_2": 0.13432, "loss_cls_3": 0.80474, "loss_box_3": 1.37647, "loss_cns_3": 0.66702, "loss_yns_3": 0.13607, "loss_cls_4": 0.81578, "loss_box_4": 1.37557, "loss_cns_4": 0.66709, "loss_yns_4": 0.13633, "loss_cls_5": 0.82012, "loss_box_5": 1.39157, "loss_cns_5": 0.66623, "loss_yns_5": 0.13624, "loss_cls_dn_0": 0.11897, "loss_box_dn_0": 0.71757, "loss_cls_dn_1": 0.09211, "loss_box_dn_1": 0.62052, "loss_cls_dn_2": 0.09074, "loss_box_dn_2": 0.60701, "loss_cls_dn_3": 0.09084, "loss_box_dn_3": 0.60044, "loss_cls_dn_4": 0.09082, "loss_box_dn_4": 0.60473, "loss_cls_dn_5": 0.09524, "loss_box_dn_5": 0.61216, "loss_dense_depth": 0.72069, "loss": 23.05685, "grad_norm": 32.36615, "time": 1.69645} -{"mode": "train", "epoch": 1, "iter": 383, "lr": 0.00025, "memory": 49164, "data_time": 0.0761, "loss_cls_0": 0.74493, "loss_box_0": 1.50957, "loss_cns_0": 0.62933, "loss_yns_0": 0.1361, "loss_cls_1": 0.80325, "loss_box_1": 1.40207, "loss_cns_1": 0.67008, "loss_yns_1": 0.13526, "loss_cls_2": 0.80367, "loss_box_2": 1.37419, "loss_cns_2": 0.66766, "loss_yns_2": 0.13506, "loss_cls_3": 0.81762, "loss_box_3": 1.37224, "loss_cns_3": 0.66705, "loss_yns_3": 0.1349, "loss_cls_4": 0.82351, "loss_box_4": 1.35772, "loss_cns_4": 0.66582, "loss_yns_4": 0.13521, "loss_cls_5": 0.82919, "loss_box_5": 1.37336, "loss_cns_5": 0.66642, "loss_yns_5": 0.13608, "loss_cls_dn_0": 0.11664, "loss_box_dn_0": 0.71218, "loss_cls_dn_1": 0.09504, "loss_box_dn_1": 0.62143, "loss_cls_dn_2": 0.09258, "loss_box_dn_2": 0.60352, "loss_cls_dn_3": 0.09701, "loss_box_dn_3": 0.60488, "loss_cls_dn_4": 0.09502, "loss_box_dn_4": 0.60293, "loss_cls_dn_5": 0.09555, "loss_box_dn_5": 0.60959, "loss_dense_depth": 0.69787, "loss": 23.03452, "grad_norm": 29.4554, "time": 3.29334} -{"mode": "train", "epoch": 1, "iter": 384, "lr": 0.00025, "memory": 49164, "data_time": 0.08041, "loss_cls_0": 0.74845, "loss_box_0": 1.52204, "loss_cns_0": 0.6303, "loss_yns_0": 0.13637, "loss_cls_1": 0.8029, "loss_box_1": 1.43853, "loss_cns_1": 0.66392, "loss_yns_1": 0.13627, "loss_cls_2": 0.81166, "loss_box_2": 1.41725, "loss_cns_2": 0.66332, "loss_yns_2": 0.13567, "loss_cls_3": 0.82646, "loss_box_3": 1.40313, "loss_cns_3": 0.66102, "loss_yns_3": 0.13553, "loss_cls_4": 0.83782, "loss_box_4": 1.40871, "loss_cns_4": 0.66368, "loss_yns_4": 0.13495, "loss_cls_5": 0.82948, "loss_box_5": 1.4122, "loss_cns_5": 0.66195, "loss_yns_5": 0.13379, "loss_cls_dn_0": 0.11736, "loss_box_dn_0": 0.70691, "loss_cls_dn_1": 0.09037, "loss_box_dn_1": 0.62156, "loss_cls_dn_2": 0.09049, "loss_box_dn_2": 0.61267, "loss_cls_dn_3": 0.0917, "loss_box_dn_3": 0.60895, "loss_cls_dn_4": 0.09078, "loss_box_dn_4": 0.61213, "loss_cls_dn_5": 0.09036, "loss_box_dn_5": 0.61277, "loss_dense_depth": 0.71913, "loss": 23.2806, "grad_norm": 28.26247, "time": 1.63985} -{"mode": "train", "epoch": 1, "iter": 385, "lr": 0.00025, "memory": 49164, "data_time": 0.07542, "loss_cls_0": 0.7327, "loss_box_0": 1.55137, "loss_cns_0": 0.64908, "loss_yns_0": 0.14023, "loss_cls_1": 0.78668, "loss_box_1": 1.41954, "loss_cns_1": 0.66758, "loss_yns_1": 0.13943, "loss_cls_2": 0.79176, "loss_box_2": 1.41181, "loss_cns_2": 0.66933, "loss_yns_2": 0.13517, "loss_cls_3": 0.8041, "loss_box_3": 1.39405, "loss_cns_3": 0.66523, "loss_yns_3": 0.13467, "loss_cls_4": 0.8112, "loss_box_4": 1.40177, "loss_cns_4": 0.66595, "loss_yns_4": 0.13464, "loss_cls_5": 0.80827, "loss_box_5": 1.40808, "loss_cns_5": 0.66718, "loss_yns_5": 0.13512, "loss_cls_dn_0": 0.11645, "loss_box_dn_0": 0.7055, "loss_cls_dn_1": 0.08706, "loss_box_dn_1": 0.61951, "loss_cls_dn_2": 0.0851, "loss_box_dn_2": 0.61341, "loss_cls_dn_3": 0.08514, "loss_box_dn_3": 0.60931, "loss_cls_dn_4": 0.08844, "loss_box_dn_4": 0.61535, "loss_cls_dn_5": 0.09114, "loss_box_dn_5": 0.61812, "loss_dense_depth": 0.6866, "loss": 23.14606, "grad_norm": 32.45523, "time": 1.63449} -{"mode": "train", "epoch": 1, "iter": 386, "lr": 0.00025, "memory": 49164, "data_time": 0.07513, "loss_cls_0": 0.74592, "loss_box_0": 1.5578, "loss_cns_0": 0.64452, "loss_yns_0": 0.13581, "loss_cls_1": 0.79185, "loss_box_1": 1.45724, "loss_cns_1": 0.66324, "loss_yns_1": 0.13569, "loss_cls_2": 0.79264, "loss_box_2": 1.46418, "loss_cns_2": 0.66373, "loss_yns_2": 0.133, "loss_cls_3": 0.8074, "loss_box_3": 1.45544, "loss_cns_3": 0.66287, "loss_yns_3": 0.13173, "loss_cls_4": 0.80493, "loss_box_4": 1.45294, "loss_cns_4": 0.66476, "loss_yns_4": 0.13231, "loss_cls_5": 0.816, "loss_box_5": 1.46808, "loss_cns_5": 0.66595, "loss_yns_5": 0.13305, "loss_cls_dn_0": 0.11855, "loss_box_dn_0": 0.71154, "loss_cls_dn_1": 0.08896, "loss_box_dn_1": 0.61493, "loss_cls_dn_2": 0.08682, "loss_box_dn_2": 0.61172, "loss_cls_dn_3": 0.08811, "loss_box_dn_3": 0.61177, "loss_cls_dn_4": 0.09081, "loss_box_dn_4": 0.61063, "loss_cls_dn_5": 0.09351, "loss_box_dn_5": 0.61977, "loss_dense_depth": 0.7253, "loss": 23.4535, "grad_norm": 34.63688, "time": 1.65581} -{"mode": "train", "epoch": 1, "iter": 387, "lr": 0.00025, "memory": 49164, "data_time": 1.55184, "loss_cls_0": 0.69993, "loss_box_0": 1.56242, "loss_cns_0": 0.64121, "loss_yns_0": 0.13564, "loss_cls_1": 0.76384, "loss_box_1": 1.42949, "loss_cns_1": 0.66451, "loss_yns_1": 0.13444, "loss_cls_2": 0.76017, "loss_box_2": 1.39859, "loss_cns_2": 0.6598, "loss_yns_2": 0.13098, "loss_cls_3": 0.76718, "loss_box_3": 1.40736, "loss_cns_3": 0.66159, "loss_yns_3": 0.13054, "loss_cls_4": 0.77946, "loss_box_4": 1.38038, "loss_cns_4": 0.66167, "loss_yns_4": 0.12977, "loss_cls_5": 0.78258, "loss_box_5": 1.39761, "loss_cns_5": 0.66459, "loss_yns_5": 0.13172, "loss_cls_dn_0": 0.11527, "loss_box_dn_0": 0.70439, "loss_cls_dn_1": 0.08963, "loss_box_dn_1": 0.63012, "loss_cls_dn_2": 0.08641, "loss_box_dn_2": 0.61909, "loss_cls_dn_3": 0.08706, "loss_box_dn_3": 0.6206, "loss_cls_dn_4": 0.08859, "loss_box_dn_4": 0.61399, "loss_cls_dn_5": 0.09054, "loss_box_dn_5": 0.61919, "loss_dense_depth": 0.71411, "loss": 22.9545, "grad_norm": 30.63864, "time": 3.35441} -{"mode": "train", "epoch": 1, "iter": 388, "lr": 0.00025, "memory": 49164, "data_time": 0.07542, "loss_cls_0": 0.70155, "loss_box_0": 1.5404, "loss_cns_0": 0.64041, "loss_yns_0": 0.13644, "loss_cls_1": 0.75725, "loss_box_1": 1.4057, "loss_cns_1": 0.66234, "loss_yns_1": 0.13394, "loss_cls_2": 0.7565, "loss_box_2": 1.39106, "loss_cns_2": 0.66033, "loss_yns_2": 0.13161, "loss_cls_3": 0.76184, "loss_box_3": 1.39827, "loss_cns_3": 0.66331, "loss_yns_3": 0.1318, "loss_cls_4": 0.77957, "loss_box_4": 1.37637, "loss_cns_4": 0.6623, "loss_yns_4": 0.13097, "loss_cls_5": 0.78008, "loss_box_5": 1.38276, "loss_cns_5": 0.66539, "loss_yns_5": 0.13209, "loss_cls_dn_0": 0.11024, "loss_box_dn_0": 0.71353, "loss_cls_dn_1": 0.08997, "loss_box_dn_1": 0.63002, "loss_cls_dn_2": 0.08687, "loss_box_dn_2": 0.61934, "loss_cls_dn_3": 0.0886, "loss_box_dn_3": 0.62133, "loss_cls_dn_4": 0.08898, "loss_box_dn_4": 0.61582, "loss_cls_dn_5": 0.08941, "loss_box_dn_5": 0.62096, "loss_dense_depth": 0.70324, "loss": 22.86057, "grad_norm": 31.67681, "time": 1.55344} -{"mode": "train", "epoch": 1, "iter": 389, "lr": 0.00025, "memory": 49164, "data_time": 0.07057, "loss_cls_0": 0.7031, "loss_box_0": 1.51779, "loss_cns_0": 0.64576, "loss_yns_0": 0.13838, "loss_cls_1": 0.75646, "loss_box_1": 1.41578, "loss_cns_1": 0.66427, "loss_yns_1": 0.13744, "loss_cls_2": 0.76233, "loss_box_2": 1.38673, "loss_cns_2": 0.66514, "loss_yns_2": 0.13464, "loss_cls_3": 0.76986, "loss_box_3": 1.3701, "loss_cns_3": 0.66195, "loss_yns_3": 0.13428, "loss_cls_4": 0.77541, "loss_box_4": 1.37852, "loss_cns_4": 0.66256, "loss_yns_4": 0.13427, "loss_cls_5": 0.77665, "loss_box_5": 1.37234, "loss_cns_5": 0.66351, "loss_yns_5": 0.1358, "loss_cls_dn_0": 0.11116, "loss_box_dn_0": 0.70768, "loss_cls_dn_1": 0.08854, "loss_box_dn_1": 0.6389, "loss_cls_dn_2": 0.08794, "loss_box_dn_2": 0.62344, "loss_cls_dn_3": 0.08676, "loss_box_dn_3": 0.61492, "loss_cls_dn_4": 0.08616, "loss_box_dn_4": 0.61985, "loss_cls_dn_5": 0.08885, "loss_box_dn_5": 0.62378, "loss_dense_depth": 0.71538, "loss": 22.85647, "grad_norm": 27.76526, "time": 8.50284} -{"mode": "train", "epoch": 1, "iter": 390, "lr": 0.00026, "memory": 49164, "data_time": 0.06813, "loss_cls_0": 0.70962, "loss_box_0": 1.55047, "loss_cns_0": 0.63955, "loss_yns_0": 0.1419, "loss_cls_1": 0.7685, "loss_box_1": 1.4129, "loss_cns_1": 0.66648, "loss_yns_1": 0.14107, "loss_cls_2": 0.7726, "loss_box_2": 1.38173, "loss_cns_2": 0.66751, "loss_yns_2": 0.13841, "loss_cls_3": 0.77153, "loss_box_3": 1.37203, "loss_cns_3": 0.66376, "loss_yns_3": 0.13815, "loss_cls_4": 0.78179, "loss_box_4": 1.38252, "loss_cns_4": 0.66375, "loss_yns_4": 0.13964, "loss_cls_5": 0.78703, "loss_box_5": 1.36994, "loss_cns_5": 0.6645, "loss_yns_5": 0.13898, "loss_cls_dn_0": 0.11316, "loss_box_dn_0": 0.71264, "loss_cls_dn_1": 0.09077, "loss_box_dn_1": 0.62863, "loss_cls_dn_2": 0.09112, "loss_box_dn_2": 0.60923, "loss_cls_dn_3": 0.08882, "loss_box_dn_3": 0.60571, "loss_cls_dn_4": 0.091, "loss_box_dn_4": 0.60827, "loss_cls_dn_5": 0.09427, "loss_box_dn_5": 0.60825, "loss_dense_depth": 0.68883, "loss": 22.89506, "grad_norm": 30.06877, "time": 1.53986} -{"mode": "train", "epoch": 1, "iter": 391, "lr": 0.00026, "memory": 49164, "data_time": 0.07129, "loss_cls_0": 0.69058, "loss_box_0": 1.52799, "loss_cns_0": 0.63863, "loss_yns_0": 0.13803, "loss_cls_1": 0.7603, "loss_box_1": 1.38303, "loss_cns_1": 0.66664, "loss_yns_1": 0.13636, "loss_cls_2": 0.75446, "loss_box_2": 1.35797, "loss_cns_2": 0.66713, "loss_yns_2": 0.13391, "loss_cls_3": 0.76246, "loss_box_3": 1.36767, "loss_cns_3": 0.66399, "loss_yns_3": 0.13339, "loss_cls_4": 0.7651, "loss_box_4": 1.3473, "loss_cns_4": 0.66603, "loss_yns_4": 0.13509, "loss_cls_5": 0.7771, "loss_box_5": 1.34104, "loss_cns_5": 0.66551, "loss_yns_5": 0.13493, "loss_cls_dn_0": 0.1145, "loss_box_dn_0": 0.70724, "loss_cls_dn_1": 0.09172, "loss_box_dn_1": 0.63222, "loss_cls_dn_2": 0.09003, "loss_box_dn_2": 0.61388, "loss_cls_dn_3": 0.09046, "loss_box_dn_3": 0.62145, "loss_cls_dn_4": 0.09221, "loss_box_dn_4": 0.61417, "loss_cls_dn_5": 0.09368, "loss_box_dn_5": 0.61506, "loss_dense_depth": 0.69898, "loss": 22.69024, "grad_norm": 25.38313, "time": 1.55241} -{"mode": "train", "epoch": 1, "iter": 392, "lr": 0.00026, "memory": 49164, "data_time": 0.07398, "loss_cls_0": 0.71306, "loss_box_0": 1.54701, "loss_cns_0": 0.63759, "loss_yns_0": 0.14089, "loss_cls_1": 0.7786, "loss_box_1": 1.40936, "loss_cns_1": 0.66567, "loss_yns_1": 0.13924, "loss_cls_2": 0.7839, "loss_box_2": 1.39399, "loss_cns_2": 0.66342, "loss_yns_2": 0.14031, "loss_cls_3": 0.79307, "loss_box_3": 1.38511, "loss_cns_3": 0.66309, "loss_yns_3": 0.139, "loss_cls_4": 0.79017, "loss_box_4": 1.38945, "loss_cns_4": 0.66409, "loss_yns_4": 0.13992, "loss_cls_5": 0.80768, "loss_box_5": 1.39892, "loss_cns_5": 0.6629, "loss_yns_5": 0.14142, "loss_cls_dn_0": 0.11937, "loss_box_dn_0": 0.7078, "loss_cls_dn_1": 0.09341, "loss_box_dn_1": 0.62579, "loss_cls_dn_2": 0.09029, "loss_box_dn_2": 0.61458, "loss_cls_dn_3": 0.0928, "loss_box_dn_3": 0.61607, "loss_cls_dn_4": 0.09242, "loss_box_dn_4": 0.61932, "loss_cls_dn_5": 0.09351, "loss_box_dn_5": 0.62566, "loss_dense_depth": 0.7069, "loss": 23.0858, "grad_norm": 29.96707, "time": 1.59082} -{"mode": "train", "epoch": 1, "iter": 393, "lr": 0.00026, "memory": 49164, "data_time": 0.11289, "loss_cls_0": 0.72212, "loss_box_0": 1.54214, "loss_cns_0": 0.63667, "loss_yns_0": 0.14124, "loss_cls_1": 0.78562, "loss_box_1": 1.39033, "loss_cns_1": 0.66433, "loss_yns_1": 0.14082, "loss_cls_2": 0.78635, "loss_box_2": 1.3736, "loss_cns_2": 0.66445, "loss_yns_2": 0.14084, "loss_cls_3": 0.79547, "loss_box_3": 1.35363, "loss_cns_3": 0.66222, "loss_yns_3": 0.1381, "loss_cls_4": 0.79805, "loss_box_4": 1.36488, "loss_cns_4": 0.66332, "loss_yns_4": 0.13697, "loss_cls_5": 0.80033, "loss_box_5": 1.36067, "loss_cns_5": 0.66177, "loss_yns_5": 0.1372, "loss_cls_dn_0": 0.11465, "loss_box_dn_0": 0.71775, "loss_cls_dn_1": 0.09698, "loss_box_dn_1": 0.63034, "loss_cls_dn_2": 0.09378, "loss_box_dn_2": 0.61966, "loss_cls_dn_3": 0.09293, "loss_box_dn_3": 0.61323, "loss_cls_dn_4": 0.0912, "loss_box_dn_4": 0.61795, "loss_cls_dn_5": 0.09117, "loss_box_dn_5": 0.61445, "loss_dense_depth": 0.71428, "loss": 22.96949, "grad_norm": 23.85957, "time": 1.60886} -{"mode": "train", "epoch": 1, "iter": 394, "lr": 0.00026, "memory": 49164, "data_time": 0.0738, "loss_cls_0": 0.70342, "loss_box_0": 1.49782, "loss_cns_0": 0.64126, "loss_yns_0": 0.14206, "loss_cls_1": 0.77825, "loss_box_1": 1.40275, "loss_cns_1": 0.66299, "loss_yns_1": 0.13978, "loss_cls_2": 0.78132, "loss_box_2": 1.36573, "loss_cns_2": 0.66264, "loss_yns_2": 0.14116, "loss_cls_3": 0.78953, "loss_box_3": 1.34749, "loss_cns_3": 0.65973, "loss_yns_3": 0.13916, "loss_cls_4": 0.78618, "loss_box_4": 1.34002, "loss_cns_4": 0.66188, "loss_yns_4": 0.13972, "loss_cls_5": 0.78678, "loss_box_5": 1.35322, "loss_cns_5": 0.6589, "loss_yns_5": 0.13894, "loss_cls_dn_0": 0.11718, "loss_box_dn_0": 0.70598, "loss_cls_dn_1": 0.09236, "loss_box_dn_1": 0.62221, "loss_cls_dn_2": 0.08985, "loss_box_dn_2": 0.60594, "loss_cls_dn_3": 0.09109, "loss_box_dn_3": 0.60239, "loss_cls_dn_4": 0.09273, "loss_box_dn_4": 0.59794, "loss_cls_dn_5": 0.09574, "loss_box_dn_5": 0.60615, "loss_dense_depth": 0.71852, "loss": 22.75876, "grad_norm": 34.41349, "time": 1.56242} -{"mode": "train", "epoch": 1, "iter": 395, "lr": 0.00026, "memory": 49164, "data_time": 0.0788, "loss_cls_0": 0.70314, "loss_box_0": 1.51111, "loss_cns_0": 0.63729, "loss_yns_0": 0.14183, "loss_cls_1": 0.77937, "loss_box_1": 1.37554, "loss_cns_1": 0.66332, "loss_yns_1": 0.14051, "loss_cls_2": 0.78464, "loss_box_2": 1.32431, "loss_cns_2": 0.66056, "loss_yns_2": 0.14109, "loss_cls_3": 0.79409, "loss_box_3": 1.32277, "loss_cns_3": 0.65796, "loss_yns_3": 0.14058, "loss_cls_4": 0.78976, "loss_box_4": 1.33235, "loss_cns_4": 0.66039, "loss_yns_4": 0.13939, "loss_cls_5": 0.7911, "loss_box_5": 1.33407, "loss_cns_5": 0.65462, "loss_yns_5": 0.1403, "loss_cls_dn_0": 0.11797, "loss_box_dn_0": 0.70454, "loss_cls_dn_1": 0.09043, "loss_box_dn_1": 0.63184, "loss_cls_dn_2": 0.09014, "loss_box_dn_2": 0.60563, "loss_cls_dn_3": 0.09345, "loss_box_dn_3": 0.60799, "loss_cls_dn_4": 0.093, "loss_box_dn_4": 0.60924, "loss_cls_dn_5": 0.09208, "loss_box_dn_5": 0.61726, "loss_dense_depth": 0.73082, "loss": 22.7045, "grad_norm": 36.37959, "time": 1.56825} -{"mode": "train", "epoch": 1, "iter": 396, "lr": 0.00026, "memory": 49164, "data_time": 0.07937, "loss_cls_0": 0.69891, "loss_box_0": 1.49795, "loss_cns_0": 0.63762, "loss_yns_0": 0.14232, "loss_cls_1": 0.78266, "loss_box_1": 1.35832, "loss_cns_1": 0.66646, "loss_yns_1": 0.14094, "loss_cls_2": 0.79798, "loss_box_2": 1.32307, "loss_cns_2": 0.6631, "loss_yns_2": 0.14241, "loss_cls_3": 0.80277, "loss_box_3": 1.31338, "loss_cns_3": 0.66161, "loss_yns_3": 0.14012, "loss_cls_4": 0.79855, "loss_box_4": 1.31923, "loss_cns_4": 0.66439, "loss_yns_4": 0.13954, "loss_cls_5": 0.80304, "loss_box_5": 1.32294, "loss_cns_5": 0.66257, "loss_yns_5": 0.14197, "loss_cls_dn_0": 0.11249, "loss_box_dn_0": 0.71157, "loss_cls_dn_1": 0.09063, "loss_box_dn_1": 0.6256, "loss_cls_dn_2": 0.0908, "loss_box_dn_2": 0.61179, "loss_cls_dn_3": 0.09296, "loss_box_dn_3": 0.61058, "loss_cls_dn_4": 0.09119, "loss_box_dn_4": 0.61314, "loss_cls_dn_5": 0.09104, "loss_box_dn_5": 0.62216, "loss_dense_depth": 0.70311, "loss": 22.68891, "grad_norm": 31.90498, "time": 1.6221} -{"mode": "train", "epoch": 1, "iter": 397, "lr": 0.00026, "memory": 49164, "data_time": 0.08292, "loss_cls_0": 0.72184, "loss_box_0": 1.52467, "loss_cns_0": 0.63886, "loss_yns_0": 0.14198, "loss_cls_1": 0.78691, "loss_box_1": 1.34967, "loss_cns_1": 0.66455, "loss_yns_1": 0.14187, "loss_cls_2": 0.79562, "loss_box_2": 1.33817, "loss_cns_2": 0.66446, "loss_yns_2": 0.14467, "loss_cls_3": 0.80551, "loss_box_3": 1.32504, "loss_cns_3": 0.66355, "loss_yns_3": 0.14099, "loss_cls_4": 0.81051, "loss_box_4": 1.32576, "loss_cns_4": 0.66547, "loss_yns_4": 0.14098, "loss_cls_5": 0.80501, "loss_box_5": 1.32877, "loss_cns_5": 0.66485, "loss_yns_5": 0.14191, "loss_cls_dn_0": 0.11446, "loss_box_dn_0": 0.69857, "loss_cls_dn_1": 0.09252, "loss_box_dn_1": 0.62159, "loss_cls_dn_2": 0.09107, "loss_box_dn_2": 0.61204, "loss_cls_dn_3": 0.09225, "loss_box_dn_3": 0.61085, "loss_cls_dn_4": 0.09274, "loss_box_dn_4": 0.61027, "loss_cls_dn_5": 0.09529, "loss_box_dn_5": 0.61672, "loss_dense_depth": 0.69311, "loss": 22.77308, "grad_norm": 27.52711, "time": 1.57602} -{"mode": "train", "epoch": 1, "iter": 398, "lr": 0.00026, "memory": 49164, "data_time": 0.11311, "loss_cls_0": 0.72563, "loss_box_0": 1.54168, "loss_cns_0": 0.63414, "loss_yns_0": 0.14175, "loss_cls_1": 0.79222, "loss_box_1": 1.39386, "loss_cns_1": 0.66266, "loss_yns_1": 0.14217, "loss_cls_2": 0.79564, "loss_box_2": 1.35802, "loss_cns_2": 0.6621, "loss_yns_2": 0.14049, "loss_cls_3": 0.80459, "loss_box_3": 1.34637, "loss_cns_3": 0.65961, "loss_yns_3": 0.13936, "loss_cls_4": 0.80706, "loss_box_4": 1.34409, "loss_cns_4": 0.66044, "loss_yns_4": 0.13969, "loss_cls_5": 0.81828, "loss_box_5": 1.34302, "loss_cns_5": 0.66192, "loss_yns_5": 0.1396, "loss_cls_dn_0": 0.11431, "loss_box_dn_0": 0.704, "loss_cls_dn_1": 0.08903, "loss_box_dn_1": 0.61728, "loss_cls_dn_2": 0.08703, "loss_box_dn_2": 0.60127, "loss_cls_dn_3": 0.08745, "loss_box_dn_3": 0.5992, "loss_cls_dn_4": 0.08856, "loss_box_dn_4": 0.59737, "loss_cls_dn_5": 0.09157, "loss_box_dn_5": 0.59714, "loss_dense_depth": 0.76615, "loss": 22.89474, "grad_norm": 28.36757, "time": 1.61082} -{"mode": "train", "epoch": 1, "iter": 399, "lr": 0.00026, "memory": 49164, "data_time": 0.18654, "loss_cls_0": 0.71452, "loss_box_0": 1.51991, "loss_cns_0": 0.63665, "loss_yns_0": 0.14279, "loss_cls_1": 0.77375, "loss_box_1": 1.41169, "loss_cns_1": 0.66215, "loss_yns_1": 0.14039, "loss_cls_2": 0.78663, "loss_box_2": 1.36464, "loss_cns_2": 0.66292, "loss_yns_2": 0.14334, "loss_cls_3": 0.7989, "loss_box_3": 1.35739, "loss_cns_3": 0.6632, "loss_yns_3": 0.14113, "loss_cls_4": 0.80314, "loss_box_4": 1.37088, "loss_cns_4": 0.66517, "loss_yns_4": 0.14344, "loss_cls_5": 0.80623, "loss_box_5": 1.35908, "loss_cns_5": 0.66507, "loss_yns_5": 0.14515, "loss_cls_dn_0": 0.11534, "loss_box_dn_0": 0.71018, "loss_cls_dn_1": 0.09036, "loss_box_dn_1": 0.62521, "loss_cls_dn_2": 0.08969, "loss_box_dn_2": 0.59755, "loss_cls_dn_3": 0.08856, "loss_box_dn_3": 0.59585, "loss_cls_dn_4": 0.08872, "loss_box_dn_4": 0.60303, "loss_cls_dn_5": 0.08913, "loss_box_dn_5": 0.60238, "loss_dense_depth": 0.78863, "loss": 22.9628, "grad_norm": 37.11226, "time": 1.66253} -{"mode": "train", "epoch": 1, "iter": 400, "lr": 0.00026, "memory": 49164, "data_time": 0.07621, "loss_cls_0": 0.73077, "loss_box_0": 1.51959, "loss_cns_0": 0.64067, "loss_yns_0": 0.144, "loss_cls_1": 0.78796, "loss_box_1": 1.38046, "loss_cns_1": 0.66914, "loss_yns_1": 0.14171, "loss_cls_2": 0.79121, "loss_box_2": 1.3411, "loss_cns_2": 0.66594, "loss_yns_2": 0.14642, "loss_cls_3": 0.80081, "loss_box_3": 1.33645, "loss_cns_3": 0.66508, "loss_yns_3": 0.14404, "loss_cls_4": 0.8253, "loss_box_4": 1.33647, "loss_cns_4": 0.66254, "loss_yns_4": 0.14371, "loss_cls_5": 0.81725, "loss_box_5": 1.34891, "loss_cns_5": 0.66811, "loss_yns_5": 0.14689, "loss_cls_dn_0": 0.11454, "loss_box_dn_0": 0.71442, "loss_cls_dn_1": 0.09292, "loss_box_dn_1": 0.64223, "loss_cls_dn_2": 0.09207, "loss_box_dn_2": 0.6331, "loss_cls_dn_3": 0.09085, "loss_box_dn_3": 0.63711, "loss_cls_dn_4": 0.09257, "loss_box_dn_4": 0.6498, "loss_cls_dn_5": 0.09398, "loss_box_dn_5": 0.66163, "loss_dense_depth": 0.71818, "loss": 23.08793, "grad_norm": 37.1855, "time": 3.31814} -{"mode": "train", "epoch": 1, "iter": 401, "lr": 0.00026, "memory": 49164, "data_time": 0.07167, "loss_cls_0": 0.73827, "loss_box_0": 1.54486, "loss_cns_0": 0.6379, "loss_yns_0": 0.14066, "loss_cls_1": 0.79861, "loss_box_1": 1.40916, "loss_cns_1": 0.66475, "loss_yns_1": 0.14066, "loss_cls_2": 0.79575, "loss_box_2": 1.38851, "loss_cns_2": 0.66509, "loss_yns_2": 0.14301, "loss_cls_3": 0.79624, "loss_box_3": 1.37026, "loss_cns_3": 0.66423, "loss_yns_3": 0.1396, "loss_cls_4": 0.8211, "loss_box_4": 1.36548, "loss_cns_4": 0.6617, "loss_yns_4": 0.13777, "loss_cls_5": 0.81383, "loss_box_5": 1.38636, "loss_cns_5": 0.66497, "loss_yns_5": 0.13949, "loss_cls_dn_0": 0.11545, "loss_box_dn_0": 0.70964, "loss_cls_dn_1": 0.09629, "loss_box_dn_1": 0.65439, "loss_cls_dn_2": 0.09429, "loss_box_dn_2": 0.65554, "loss_cls_dn_3": 0.0934, "loss_box_dn_3": 0.65813, "loss_cls_dn_4": 0.09666, "loss_box_dn_4": 0.66949, "loss_cls_dn_5": 0.0953, "loss_box_dn_5": 0.68422, "loss_dense_depth": 0.8067, "loss": 23.45776, "grad_norm": 36.40279, "time": 1.65688} -{"mode": "train", "epoch": 1, "iter": 402, "lr": 0.00026, "memory": 49164, "data_time": 0.08432, "loss_cls_0": 0.70804, "loss_box_0": 1.54276, "loss_cns_0": 0.6419, "loss_yns_0": 0.14106, "loss_cls_1": 0.76843, "loss_box_1": 1.41734, "loss_cns_1": 0.66514, "loss_yns_1": 0.13823, "loss_cls_2": 0.78182, "loss_box_2": 1.39794, "loss_cns_2": 0.66854, "loss_yns_2": 0.13901, "loss_cls_3": 0.79696, "loss_box_3": 1.39114, "loss_cns_3": 0.66593, "loss_yns_3": 0.13774, "loss_cls_4": 0.79203, "loss_box_4": 1.39802, "loss_cns_4": 0.66602, "loss_yns_4": 0.13975, "loss_cls_5": 0.78844, "loss_box_5": 1.40743, "loss_cns_5": 0.66556, "loss_yns_5": 0.13845, "loss_cls_dn_0": 0.1149, "loss_box_dn_0": 0.70908, "loss_cls_dn_1": 0.09072, "loss_box_dn_1": 0.67147, "loss_cls_dn_2": 0.08984, "loss_box_dn_2": 0.66762, "loss_cls_dn_3": 0.09083, "loss_box_dn_3": 0.67057, "loss_cls_dn_4": 0.09234, "loss_box_dn_4": 0.67811, "loss_cls_dn_5": 0.09236, "loss_box_dn_5": 0.68518, "loss_dense_depth": 0.7326, "loss": 23.38331, "grad_norm": 40.95105, "time": 1.61951} -{"mode": "train", "epoch": 1, "iter": 403, "lr": 0.00026, "memory": 49164, "data_time": 0.09235, "loss_cls_0": 0.72269, "loss_box_0": 1.57278, "loss_cns_0": 0.6401, "loss_yns_0": 0.14018, "loss_cls_1": 0.77668, "loss_box_1": 1.42587, "loss_cns_1": 0.66721, "loss_yns_1": 0.13668, "loss_cls_2": 0.79782, "loss_box_2": 1.40051, "loss_cns_2": 0.67103, "loss_yns_2": 0.137, "loss_cls_3": 0.8039, "loss_box_3": 1.38387, "loss_cns_3": 0.66659, "loss_yns_3": 0.13607, "loss_cls_4": 0.79383, "loss_box_4": 1.38161, "loss_cns_4": 0.66792, "loss_yns_4": 0.13852, "loss_cls_5": 0.79784, "loss_box_5": 1.38565, "loss_cns_5": 0.66764, "loss_yns_5": 0.13711, "loss_cls_dn_0": 0.11785, "loss_box_dn_0": 0.72337, "loss_cls_dn_1": 0.08994, "loss_box_dn_1": 0.65151, "loss_cls_dn_2": 0.09123, "loss_box_dn_2": 0.64345, "loss_cls_dn_3": 0.09259, "loss_box_dn_3": 0.63979, "loss_cls_dn_4": 0.09067, "loss_box_dn_4": 0.64174, "loss_cls_dn_5": 0.09217, "loss_box_dn_5": 0.64567, "loss_dense_depth": 0.75603, "loss": 23.32511, "grad_norm": 32.2789, "time": 1.58231} -{"mode": "train", "epoch": 1, "iter": 404, "lr": 0.00026, "memory": 49164, "data_time": 0.07419, "loss_cls_0": 0.72033, "loss_box_0": 1.55714, "loss_cns_0": 0.6386, "loss_yns_0": 0.13887, "loss_cls_1": 0.7683, "loss_box_1": 1.43519, "loss_cns_1": 0.66427, "loss_yns_1": 0.13504, "loss_cls_2": 0.78409, "loss_box_2": 1.40014, "loss_cns_2": 0.66332, "loss_yns_2": 0.13469, "loss_cls_3": 0.78514, "loss_box_3": 1.38762, "loss_cns_3": 0.66243, "loss_yns_3": 0.13478, "loss_cls_4": 0.79096, "loss_box_4": 1.39133, "loss_cns_4": 0.66161, "loss_yns_4": 0.13616, "loss_cls_5": 0.79749, "loss_box_5": 1.38095, "loss_cns_5": 0.66045, "loss_yns_5": 0.13732, "loss_cls_dn_0": 0.11158, "loss_box_dn_0": 0.71158, "loss_cls_dn_1": 0.08874, "loss_box_dn_1": 0.61718, "loss_cls_dn_2": 0.0883, "loss_box_dn_2": 0.60336, "loss_cls_dn_3": 0.08864, "loss_box_dn_3": 0.59914, "loss_cls_dn_4": 0.0889, "loss_box_dn_4": 0.6015, "loss_cls_dn_5": 0.0879, "loss_box_dn_5": 0.60203, "loss_dense_depth": 0.78334, "loss": 23.0384, "grad_norm": 44.46696, "time": 1.59952} -{"mode": "train", "epoch": 1, "iter": 405, "lr": 0.00026, "memory": 49164, "data_time": 0.07559, "loss_cls_0": 0.73355, "loss_box_0": 1.57296, "loss_cns_0": 0.63941, "loss_yns_0": 0.13824, "loss_cls_1": 0.77138, "loss_box_1": 1.43142, "loss_cns_1": 0.65986, "loss_yns_1": 0.13533, "loss_cls_2": 0.79138, "loss_box_2": 1.41728, "loss_cns_2": 0.66172, "loss_yns_2": 0.1343, "loss_cls_3": 0.80946, "loss_box_3": 1.40036, "loss_cns_3": 0.65905, "loss_yns_3": 0.13452, "loss_cls_4": 0.80115, "loss_box_4": 1.39828, "loss_cns_4": 0.66087, "loss_yns_4": 0.13531, "loss_cls_5": 0.81379, "loss_box_5": 1.38333, "loss_cns_5": 0.65524, "loss_yns_5": 0.13515, "loss_cls_dn_0": 0.12127, "loss_box_dn_0": 0.71809, "loss_cls_dn_1": 0.08747, "loss_box_dn_1": 0.60837, "loss_cls_dn_2": 0.08786, "loss_box_dn_2": 0.60151, "loss_cls_dn_3": 0.08931, "loss_box_dn_3": 0.60319, "loss_cls_dn_4": 0.08939, "loss_box_dn_4": 0.60045, "loss_cls_dn_5": 0.08957, "loss_box_dn_5": 0.60437, "loss_dense_depth": 0.71642, "loss": 23.0906, "grad_norm": 32.69555, "time": 1.62455} -{"mode": "train", "epoch": 1, "iter": 406, "lr": 0.00026, "memory": 49164, "data_time": 0.0775, "loss_cls_0": 0.74675, "loss_box_0": 1.55565, "loss_cns_0": 0.64141, "loss_yns_0": 0.13724, "loss_cls_1": 0.77862, "loss_box_1": 1.41505, "loss_cns_1": 0.66287, "loss_yns_1": 0.1339, "loss_cls_2": 0.79513, "loss_box_2": 1.41239, "loss_cns_2": 0.66907, "loss_yns_2": 0.13561, "loss_cls_3": 0.80777, "loss_box_3": 1.39213, "loss_cns_3": 0.66501, "loss_yns_3": 0.13508, "loss_cls_4": 0.81172, "loss_box_4": 1.3885, "loss_cns_4": 0.66661, "loss_yns_4": 0.13405, "loss_cls_5": 0.79305, "loss_box_5": 1.40573, "loss_cns_5": 0.66273, "loss_yns_5": 0.13374, "loss_cls_dn_0": 0.12056, "loss_box_dn_0": 0.71329, "loss_cls_dn_1": 0.09079, "loss_box_dn_1": 0.61211, "loss_cls_dn_2": 0.09158, "loss_box_dn_2": 0.60645, "loss_cls_dn_3": 0.09179, "loss_box_dn_3": 0.60674, "loss_cls_dn_4": 0.09219, "loss_box_dn_4": 0.61233, "loss_cls_dn_5": 0.09094, "loss_box_dn_5": 0.62589, "loss_dense_depth": 0.75183, "loss": 23.18635, "grad_norm": 41.36072, "time": 1.65071} -{"mode": "train", "epoch": 1, "iter": 407, "lr": 0.00026, "memory": 49164, "data_time": 0.10123, "loss_cls_0": 0.73966, "loss_box_0": 1.56978, "loss_cns_0": 0.63881, "loss_yns_0": 0.1336, "loss_cls_1": 0.79001, "loss_box_1": 1.40065, "loss_cns_1": 0.66582, "loss_yns_1": 0.13152, "loss_cls_2": 0.79664, "loss_box_2": 1.37777, "loss_cns_2": 0.66939, "loss_yns_2": 0.13195, "loss_cls_3": 0.8106, "loss_box_3": 1.36399, "loss_cns_3": 0.66647, "loss_yns_3": 0.13207, "loss_cls_4": 0.818, "loss_box_4": 1.37719, "loss_cns_4": 0.66865, "loss_yns_4": 0.13247, "loss_cls_5": 0.8139, "loss_box_5": 1.36854, "loss_cns_5": 0.66737, "loss_yns_5": 0.13206, "loss_cls_dn_0": 0.11886, "loss_box_dn_0": 0.71902, "loss_cls_dn_1": 0.0921, "loss_box_dn_1": 0.6359, "loss_cls_dn_2": 0.09122, "loss_box_dn_2": 0.62323, "loss_cls_dn_3": 0.09139, "loss_box_dn_3": 0.62585, "loss_cls_dn_4": 0.09198, "loss_box_dn_4": 0.64043, "loss_cls_dn_5": 0.09099, "loss_box_dn_5": 0.64827, "loss_dense_depth": 0.7321, "loss": 23.19826, "grad_norm": 31.47994, "time": 1.59928} -{"mode": "train", "epoch": 1, "iter": 408, "lr": 0.00026, "memory": 49164, "data_time": 0.07614, "loss_cls_0": 0.72564, "loss_box_0": 1.54205, "loss_cns_0": 0.63561, "loss_yns_0": 0.13799, "loss_cls_1": 0.7821, "loss_box_1": 1.37767, "loss_cns_1": 0.66687, "loss_yns_1": 0.13498, "loss_cls_2": 0.79471, "loss_box_2": 1.35286, "loss_cns_2": 0.66702, "loss_yns_2": 0.13455, "loss_cls_3": 0.80581, "loss_box_3": 1.3367, "loss_cns_3": 0.67084, "loss_yns_3": 0.13475, "loss_cls_4": 0.80409, "loss_box_4": 1.34241, "loss_cns_4": 0.67044, "loss_yns_4": 0.13622, "loss_cls_5": 0.79668, "loss_box_5": 1.3384, "loss_cns_5": 0.66965, "loss_yns_5": 0.13762, "loss_cls_dn_0": 0.11671, "loss_box_dn_0": 0.71828, "loss_cls_dn_1": 0.08941, "loss_box_dn_1": 0.64352, "loss_cls_dn_2": 0.08863, "loss_box_dn_2": 0.63085, "loss_cls_dn_3": 0.08856, "loss_box_dn_3": 0.62769, "loss_cls_dn_4": 0.09147, "loss_box_dn_4": 0.63335, "loss_cls_dn_5": 0.08952, "loss_box_dn_5": 0.63988, "loss_dense_depth": 0.7298, "loss": 22.98331, "grad_norm": 27.61463, "time": 1.56435} -{"mode": "train", "epoch": 1, "iter": 409, "lr": 0.00026, "memory": 49164, "data_time": 0.07593, "loss_cls_0": 0.71341, "loss_box_0": 1.54994, "loss_cns_0": 0.63933, "loss_yns_0": 0.13885, "loss_cls_1": 0.786, "loss_box_1": 1.41804, "loss_cns_1": 0.66572, "loss_yns_1": 0.13671, "loss_cls_2": 0.78878, "loss_box_2": 1.38614, "loss_cns_2": 0.66387, "loss_yns_2": 0.13645, "loss_cls_3": 0.80096, "loss_box_3": 1.37037, "loss_cns_3": 0.66332, "loss_yns_3": 0.13521, "loss_cls_4": 0.80144, "loss_box_4": 1.38331, "loss_cns_4": 0.66397, "loss_yns_4": 0.13613, "loss_cls_5": 0.79452, "loss_box_5": 1.39746, "loss_cns_5": 0.66545, "loss_yns_5": 0.13685, "loss_cls_dn_0": 0.11497, "loss_box_dn_0": 0.70775, "loss_cls_dn_1": 0.09096, "loss_box_dn_1": 0.63043, "loss_cls_dn_2": 0.09079, "loss_box_dn_2": 0.61429, "loss_cls_dn_3": 0.09102, "loss_box_dn_3": 0.61122, "loss_cls_dn_4": 0.09265, "loss_box_dn_4": 0.6157, "loss_cls_dn_5": 0.09307, "loss_box_dn_5": 0.6197, "loss_dense_depth": 0.71969, "loss": 23.06449, "grad_norm": 35.87351, "time": 1.56955} -{"mode": "train", "epoch": 1, "iter": 410, "lr": 0.00026, "memory": 49164, "data_time": 0.10137, "loss_cls_0": 0.7164, "loss_box_0": 1.53694, "loss_cns_0": 0.64524, "loss_yns_0": 0.14104, "loss_cls_1": 0.78027, "loss_box_1": 1.37937, "loss_cns_1": 0.6663, "loss_yns_1": 0.13786, "loss_cls_2": 0.77709, "loss_box_2": 1.36886, "loss_cns_2": 0.6674, "loss_yns_2": 0.13671, "loss_cls_3": 0.79288, "loss_box_3": 1.35046, "loss_cns_3": 0.66663, "loss_yns_3": 0.13723, "loss_cls_4": 0.78649, "loss_box_4": 1.35508, "loss_cns_4": 0.66627, "loss_yns_4": 0.13768, "loss_cls_5": 0.80047, "loss_box_5": 1.36348, "loss_cns_5": 0.66881, "loss_yns_5": 0.13834, "loss_cls_dn_0": 0.1113, "loss_box_dn_0": 0.70729, "loss_cls_dn_1": 0.0909, "loss_box_dn_1": 0.61012, "loss_cls_dn_2": 0.09009, "loss_box_dn_2": 0.5996, "loss_cls_dn_3": 0.09048, "loss_box_dn_3": 0.59168, "loss_cls_dn_4": 0.0904, "loss_box_dn_4": 0.59211, "loss_cls_dn_5": 0.09295, "loss_box_dn_5": 0.59255, "loss_dense_depth": 0.73385, "loss": 22.81066, "grad_norm": 27.47747, "time": 1.60595} -{"mode": "train", "epoch": 1, "iter": 411, "lr": 0.00026, "memory": 49164, "data_time": 0.0989, "loss_cls_0": 0.72397, "loss_box_0": 1.54462, "loss_cns_0": 0.63985, "loss_yns_0": 0.13785, "loss_cls_1": 0.78229, "loss_box_1": 1.37958, "loss_cns_1": 0.66651, "loss_yns_1": 0.13592, "loss_cls_2": 0.79389, "loss_box_2": 1.3835, "loss_cns_2": 0.66931, "loss_yns_2": 0.13511, "loss_cls_3": 0.80525, "loss_box_3": 1.35525, "loss_cns_3": 0.66606, "loss_yns_3": 0.13517, "loss_cls_4": 0.80325, "loss_box_4": 1.35911, "loss_cns_4": 0.66652, "loss_yns_4": 0.13592, "loss_cls_5": 0.81685, "loss_box_5": 1.37294, "loss_cns_5": 0.66731, "loss_yns_5": 0.13591, "loss_cls_dn_0": 0.11444, "loss_box_dn_0": 0.71509, "loss_cls_dn_1": 0.0893, "loss_box_dn_1": 0.62401, "loss_cls_dn_2": 0.08779, "loss_box_dn_2": 0.61946, "loss_cls_dn_3": 0.08921, "loss_box_dn_3": 0.61015, "loss_cls_dn_4": 0.08914, "loss_box_dn_4": 0.61246, "loss_cls_dn_5": 0.09039, "loss_box_dn_5": 0.62281, "loss_dense_depth": 0.7193, "loss": 22.99548, "grad_norm": 31.4195, "time": 1.59919} -{"mode": "train", "epoch": 1, "iter": 412, "lr": 0.00026, "memory": 49164, "data_time": 0.09825, "loss_cls_0": 0.70911, "loss_box_0": 1.52059, "loss_cns_0": 0.64272, "loss_yns_0": 0.13624, "loss_cls_1": 0.78094, "loss_box_1": 1.38011, "loss_cns_1": 0.66746, "loss_yns_1": 0.13434, "loss_cls_2": 0.78048, "loss_box_2": 1.37207, "loss_cns_2": 0.66965, "loss_yns_2": 0.13573, "loss_cls_3": 0.78808, "loss_box_3": 1.34793, "loss_cns_3": 0.66582, "loss_yns_3": 0.13522, "loss_cls_4": 0.79223, "loss_box_4": 1.35057, "loss_cns_4": 0.66582, "loss_yns_4": 0.13605, "loss_cls_5": 0.79612, "loss_box_5": 1.36463, "loss_cns_5": 0.6674, "loss_yns_5": 0.13572, "loss_cls_dn_0": 0.1139, "loss_box_dn_0": 0.72118, "loss_cls_dn_1": 0.09045, "loss_box_dn_1": 0.63232, "loss_cls_dn_2": 0.08857, "loss_box_dn_2": 0.6219, "loss_cls_dn_3": 0.09018, "loss_box_dn_3": 0.61567, "loss_cls_dn_4": 0.08981, "loss_box_dn_4": 0.62131, "loss_cls_dn_5": 0.09053, "loss_box_dn_5": 0.62893, "loss_dense_depth": 0.69059, "loss": 22.87038, "grad_norm": 32.17812, "time": 1.60477} -{"mode": "train", "epoch": 1, "iter": 413, "lr": 0.00026, "memory": 49164, "data_time": 0.10375, "loss_cls_0": 0.72578, "loss_box_0": 1.52556, "loss_cns_0": 0.63741, "loss_yns_0": 0.1381, "loss_cls_1": 0.79285, "loss_box_1": 1.37612, "loss_cns_1": 0.65982, "loss_yns_1": 0.13681, "loss_cls_2": 0.77875, "loss_box_2": 1.36421, "loss_cns_2": 0.66159, "loss_yns_2": 0.13646, "loss_cls_3": 0.78695, "loss_box_3": 1.3525, "loss_cns_3": 0.66263, "loss_yns_3": 0.13686, "loss_cls_4": 0.79113, "loss_box_4": 1.34111, "loss_cns_4": 0.65858, "loss_yns_4": 0.13629, "loss_cls_5": 0.81872, "loss_box_5": 1.32989, "loss_cns_5": 0.65832, "loss_yns_5": 0.13518, "loss_cls_dn_0": 0.10915, "loss_box_dn_0": 0.70887, "loss_cls_dn_1": 0.09026, "loss_box_dn_1": 0.63563, "loss_cls_dn_2": 0.08759, "loss_box_dn_2": 0.62134, "loss_cls_dn_3": 0.08693, "loss_box_dn_3": 0.61872, "loss_cls_dn_4": 0.08793, "loss_box_dn_4": 0.61952, "loss_cls_dn_5": 0.08966, "loss_box_dn_5": 0.61896, "loss_dense_depth": 0.72655, "loss": 22.8427, "grad_norm": 31.16446, "time": 1.60052} -{"mode": "train", "epoch": 1, "iter": 414, "lr": 0.00026, "memory": 49164, "data_time": 0.07136, "loss_cls_0": 0.71339, "loss_box_0": 1.56486, "loss_cns_0": 0.64628, "loss_yns_0": 0.14222, "loss_cls_1": 0.76822, "loss_box_1": 1.39743, "loss_cns_1": 0.66542, "loss_yns_1": 0.13844, "loss_cls_2": 0.76671, "loss_box_2": 1.37514, "loss_cns_2": 0.66394, "loss_yns_2": 0.13806, "loss_cls_3": 0.77538, "loss_box_3": 1.36477, "loss_cns_3": 0.66479, "loss_yns_3": 0.13753, "loss_cls_4": 0.78101, "loss_box_4": 1.357, "loss_cns_4": 0.6631, "loss_yns_4": 0.1392, "loss_cls_5": 0.78993, "loss_box_5": 1.36751, "loss_cns_5": 0.6634, "loss_yns_5": 0.13763, "loss_cls_dn_0": 0.11389, "loss_box_dn_0": 0.70891, "loss_cls_dn_1": 0.09123, "loss_box_dn_1": 0.62845, "loss_cls_dn_2": 0.08935, "loss_box_dn_2": 0.61228, "loss_cls_dn_3": 0.08964, "loss_box_dn_3": 0.60708, "loss_cls_dn_4": 0.08992, "loss_box_dn_4": 0.60615, "loss_cls_dn_5": 0.09086, "loss_box_dn_5": 0.61412, "loss_dense_depth": 0.72541, "loss": 22.88867, "grad_norm": 29.45526, "time": 1.57042} -{"mode": "train", "epoch": 1, "iter": 415, "lr": 0.00027, "memory": 49164, "data_time": 0.08852, "loss_cls_0": 0.72973, "loss_box_0": 1.53677, "loss_cns_0": 0.64884, "loss_yns_0": 0.14204, "loss_cls_1": 0.77972, "loss_box_1": 1.3831, "loss_cns_1": 0.66568, "loss_yns_1": 0.13693, "loss_cls_2": 0.77957, "loss_box_2": 1.36429, "loss_cns_2": 0.66415, "loss_yns_2": 0.13595, "loss_cls_3": 0.7909, "loss_box_3": 1.36787, "loss_cns_3": 0.66359, "loss_yns_3": 0.13692, "loss_cls_4": 0.79913, "loss_box_4": 1.34854, "loss_cns_4": 0.66269, "loss_yns_4": 0.13571, "loss_cls_5": 0.8105, "loss_box_5": 1.34429, "loss_cns_5": 0.66302, "loss_yns_5": 0.13557, "loss_cls_dn_0": 0.11524, "loss_box_dn_0": 0.70787, "loss_cls_dn_1": 0.09118, "loss_box_dn_1": 0.62003, "loss_cls_dn_2": 0.08949, "loss_box_dn_2": 0.60103, "loss_cls_dn_3": 0.09092, "loss_box_dn_3": 0.60075, "loss_cls_dn_4": 0.0898, "loss_box_dn_4": 0.59595, "loss_cls_dn_5": 0.09222, "loss_box_dn_5": 0.5986, "loss_dense_depth": 0.69754, "loss": 22.8161, "grad_norm": 40.11823, "time": 1.61617} -{"mode": "train", "epoch": 1, "iter": 416, "lr": 0.00027, "memory": 49164, "data_time": 0.11181, "loss_cls_0": 0.71416, "loss_box_0": 1.55735, "loss_cns_0": 0.64568, "loss_yns_0": 0.14316, "loss_cls_1": 0.78511, "loss_box_1": 1.38209, "loss_cns_1": 0.66556, "loss_yns_1": 0.13767, "loss_cls_2": 0.7774, "loss_box_2": 1.37111, "loss_cns_2": 0.66427, "loss_yns_2": 0.1379, "loss_cls_3": 0.78495, "loss_box_3": 1.36215, "loss_cns_3": 0.66305, "loss_yns_3": 0.13897, "loss_cls_4": 0.79825, "loss_box_4": 1.35689, "loss_cns_4": 0.66382, "loss_yns_4": 0.13949, "loss_cls_5": 0.79926, "loss_box_5": 1.3588, "loss_cns_5": 0.66428, "loss_yns_5": 0.13911, "loss_cls_dn_0": 0.11309, "loss_box_dn_0": 0.70683, "loss_cls_dn_1": 0.09394, "loss_box_dn_1": 0.61073, "loss_cls_dn_2": 0.09156, "loss_box_dn_2": 0.59776, "loss_cls_dn_3": 0.09315, "loss_box_dn_3": 0.5984, "loss_cls_dn_4": 0.09481, "loss_box_dn_4": 0.60093, "loss_cls_dn_5": 0.09475, "loss_box_dn_5": 0.6031, "loss_dense_depth": 0.71289, "loss": 22.86243, "grad_norm": 32.50106, "time": 1.60549} -{"mode": "train", "epoch": 1, "iter": 417, "lr": 0.00027, "memory": 49164, "data_time": 0.09849, "loss_cls_0": 0.72312, "loss_box_0": 1.56454, "loss_cns_0": 0.64106, "loss_yns_0": 0.1423, "loss_cls_1": 0.79152, "loss_box_1": 1.38045, "loss_cns_1": 0.66595, "loss_yns_1": 0.13917, "loss_cls_2": 0.79758, "loss_box_2": 1.37308, "loss_cns_2": 0.66456, "loss_yns_2": 0.13809, "loss_cls_3": 0.81339, "loss_box_3": 1.37082, "loss_cns_3": 0.6666, "loss_yns_3": 0.13829, "loss_cls_4": 0.82073, "loss_box_4": 1.36965, "loss_cns_4": 0.66622, "loss_yns_4": 0.13873, "loss_cls_5": 0.80064, "loss_box_5": 1.37096, "loss_cns_5": 0.66647, "loss_yns_5": 0.13913, "loss_cls_dn_0": 0.11479, "loss_box_dn_0": 0.70744, "loss_cls_dn_1": 0.09653, "loss_box_dn_1": 0.61616, "loss_cls_dn_2": 0.09432, "loss_box_dn_2": 0.60703, "loss_cls_dn_3": 0.09648, "loss_box_dn_3": 0.60985, "loss_cls_dn_4": 0.09779, "loss_box_dn_4": 0.61229, "loss_cls_dn_5": 0.09854, "loss_box_dn_5": 0.61536, "loss_dense_depth": 0.71135, "loss": 23.06099, "grad_norm": 37.27745, "time": 1.5921} -{"mode": "train", "epoch": 1, "iter": 418, "lr": 0.00027, "memory": 49164, "data_time": 0.07981, "loss_cls_0": 0.70054, "loss_box_0": 1.56457, "loss_cns_0": 0.6403, "loss_yns_0": 0.14382, "loss_cls_1": 0.7699, "loss_box_1": 1.4017, "loss_cns_1": 0.66202, "loss_yns_1": 0.14216, "loss_cls_2": 0.76694, "loss_box_2": 1.39016, "loss_cns_2": 0.66566, "loss_yns_2": 0.14097, "loss_cls_3": 0.77702, "loss_box_3": 1.36991, "loss_cns_3": 0.66237, "loss_yns_3": 0.14109, "loss_cls_4": 0.7809, "loss_box_4": 1.37212, "loss_cns_4": 0.66473, "loss_yns_4": 0.14358, "loss_cls_5": 0.78188, "loss_box_5": 1.37746, "loss_cns_5": 0.66359, "loss_yns_5": 0.14294, "loss_cls_dn_0": 0.10818, "loss_box_dn_0": 0.70159, "loss_cls_dn_1": 0.09175, "loss_box_dn_1": 0.62443, "loss_cls_dn_2": 0.0904, "loss_box_dn_2": 0.61178, "loss_cls_dn_3": 0.08976, "loss_box_dn_3": 0.61151, "loss_cls_dn_4": 0.09039, "loss_box_dn_4": 0.60941, "loss_cls_dn_5": 0.09416, "loss_box_dn_5": 0.61505, "loss_dense_depth": 0.70225, "loss": 22.90698, "grad_norm": 28.83083, "time": 1.56724} -{"mode": "train", "epoch": 1, "iter": 419, "lr": 0.00027, "memory": 49164, "data_time": 0.18069, "loss_cls_0": 0.75381, "loss_box_0": 1.61114, "loss_cns_0": 0.6361, "loss_yns_0": 0.14157, "loss_cls_1": 0.79238, "loss_box_1": 1.45531, "loss_cns_1": 0.66492, "loss_yns_1": 0.14165, "loss_cls_2": 0.80852, "loss_box_2": 1.42875, "loss_cns_2": 0.66612, "loss_yns_2": 0.13963, "loss_cls_3": 0.82542, "loss_box_3": 1.40806, "loss_cns_3": 0.66576, "loss_yns_3": 0.13934, "loss_cls_4": 0.83319, "loss_box_4": 1.40376, "loss_cns_4": 0.666, "loss_yns_4": 0.13967, "loss_cls_5": 0.83601, "loss_box_5": 1.40222, "loss_cns_5": 0.66536, "loss_yns_5": 0.14007, "loss_cls_dn_0": 0.11523, "loss_box_dn_0": 0.70341, "loss_cls_dn_1": 0.09131, "loss_box_dn_1": 0.63118, "loss_cls_dn_2": 0.09067, "loss_box_dn_2": 0.61442, "loss_cls_dn_3": 0.08928, "loss_box_dn_3": 0.61235, "loss_cls_dn_4": 0.09082, "loss_box_dn_4": 0.61093, "loss_cls_dn_5": 0.09133, "loss_box_dn_5": 0.60805, "loss_dense_depth": 0.74174, "loss": 23.45544, "grad_norm": 28.05375, "time": 1.65098} -{"mode": "train", "epoch": 1, "iter": 420, "lr": 0.00027, "memory": 49164, "data_time": 0.12255, "loss_cls_0": 0.71436, "loss_box_0": 1.57856, "loss_cns_0": 0.63776, "loss_yns_0": 0.14046, "loss_cls_1": 0.79479, "loss_box_1": 1.40443, "loss_cns_1": 0.66509, "loss_yns_1": 0.13805, "loss_cls_2": 0.80395, "loss_box_2": 1.37527, "loss_cns_2": 0.66544, "loss_yns_2": 0.13748, "loss_cls_3": 0.81104, "loss_box_3": 1.36825, "loss_cns_3": 0.66528, "loss_yns_3": 0.13667, "loss_cls_4": 0.81669, "loss_box_4": 1.37874, "loss_cns_4": 0.66581, "loss_yns_4": 0.13624, "loss_cls_5": 0.80394, "loss_box_5": 1.37119, "loss_cns_5": 0.66429, "loss_yns_5": 0.13787, "loss_cls_dn_0": 0.11047, "loss_box_dn_0": 0.70474, "loss_cls_dn_1": 0.09312, "loss_box_dn_1": 0.62472, "loss_cls_dn_2": 0.09456, "loss_box_dn_2": 0.60773, "loss_cls_dn_3": 0.09537, "loss_box_dn_3": 0.60629, "loss_cls_dn_4": 0.09652, "loss_box_dn_4": 0.61082, "loss_cls_dn_5": 0.09432, "loss_box_dn_5": 0.60941, "loss_dense_depth": 0.73213, "loss": 23.09184, "grad_norm": 25.86705, "time": 1.61343} -{"mode": "train", "epoch": 1, "iter": 421, "lr": 0.00027, "memory": 49164, "data_time": 0.07565, "loss_cls_0": 0.74311, "loss_box_0": 1.59265, "loss_cns_0": 0.62859, "loss_yns_0": 0.13496, "loss_cls_1": 0.78675, "loss_box_1": 1.41674, "loss_cns_1": 0.65882, "loss_yns_1": 0.13618, "loss_cls_2": 0.80005, "loss_box_2": 1.38014, "loss_cns_2": 0.6598, "loss_yns_2": 0.13606, "loss_cls_3": 0.81205, "loss_box_3": 1.38375, "loss_cns_3": 0.6594, "loss_yns_3": 0.1357, "loss_cls_4": 0.81083, "loss_box_4": 1.37972, "loss_cns_4": 0.66231, "loss_yns_4": 0.13549, "loss_cls_5": 0.82722, "loss_box_5": 1.36646, "loss_cns_5": 0.65933, "loss_yns_5": 0.13603, "loss_cls_dn_0": 0.11142, "loss_box_dn_0": 0.70218, "loss_cls_dn_1": 0.09029, "loss_box_dn_1": 0.63922, "loss_cls_dn_2": 0.09125, "loss_box_dn_2": 0.61701, "loss_cls_dn_3": 0.09419, "loss_box_dn_3": 0.61849, "loss_cls_dn_4": 0.09053, "loss_box_dn_4": 0.62109, "loss_cls_dn_5": 0.08891, "loss_box_dn_5": 0.62633, "loss_dense_depth": 0.71796, "loss": 23.15103, "grad_norm": 29.70999, "time": 1.66624} -{"mode": "train", "epoch": 1, "iter": 422, "lr": 0.00027, "memory": 49164, "data_time": 0.09813, "loss_cls_0": 0.70651, "loss_box_0": 1.58837, "loss_cns_0": 0.64128, "loss_yns_0": 0.13786, "loss_cls_1": 0.77587, "loss_box_1": 1.40178, "loss_cns_1": 0.66301, "loss_yns_1": 0.13277, "loss_cls_2": 0.78917, "loss_box_2": 1.36966, "loss_cns_2": 0.66301, "loss_yns_2": 0.13388, "loss_cls_3": 0.79834, "loss_box_3": 1.37675, "loss_cns_3": 0.6625, "loss_yns_3": 0.13312, "loss_cls_4": 0.80303, "loss_box_4": 1.38147, "loss_cns_4": 0.66178, "loss_yns_4": 0.13535, "loss_cls_5": 0.79684, "loss_box_5": 1.39021, "loss_cns_5": 0.66162, "loss_yns_5": 0.13424, "loss_cls_dn_0": 0.11389, "loss_box_dn_0": 0.71244, "loss_cls_dn_1": 0.08988, "loss_box_dn_1": 0.63134, "loss_cls_dn_2": 0.08917, "loss_box_dn_2": 0.60947, "loss_cls_dn_3": 0.09092, "loss_box_dn_3": 0.61089, "loss_cls_dn_4": 0.08919, "loss_box_dn_4": 0.61992, "loss_cls_dn_5": 0.08718, "loss_box_dn_5": 0.62931, "loss_dense_depth": 0.7049, "loss": 23.0169, "grad_norm": 39.67963, "time": 1.65307} -{"mode": "train", "epoch": 1, "iter": 423, "lr": 0.00027, "memory": 49164, "data_time": 0.07351, "loss_cls_0": 0.71151, "loss_box_0": 1.57481, "loss_cns_0": 0.64397, "loss_yns_0": 0.13768, "loss_cls_1": 0.77445, "loss_box_1": 1.38851, "loss_cns_1": 0.66826, "loss_yns_1": 0.13629, "loss_cls_2": 0.77768, "loss_box_2": 1.35808, "loss_cns_2": 0.66624, "loss_yns_2": 0.13486, "loss_cls_3": 0.78386, "loss_box_3": 1.34393, "loss_cns_3": 0.66625, "loss_yns_3": 0.13543, "loss_cls_4": 0.78283, "loss_box_4": 1.35636, "loss_cns_4": 0.66549, "loss_yns_4": 0.13945, "loss_cls_5": 0.78753, "loss_box_5": 1.36225, "loss_cns_5": 0.66441, "loss_yns_5": 0.13899, "loss_cls_dn_0": 0.11641, "loss_box_dn_0": 0.70769, "loss_cls_dn_1": 0.0917, "loss_box_dn_1": 0.62724, "loss_cls_dn_2": 0.09237, "loss_box_dn_2": 0.6107, "loss_cls_dn_3": 0.0927, "loss_box_dn_3": 0.60719, "loss_cls_dn_4": 0.09509, "loss_box_dn_4": 0.61584, "loss_cls_dn_5": 0.09279, "loss_box_dn_5": 0.6172, "loss_dense_depth": 0.70024, "loss": 22.86629, "grad_norm": 31.40513, "time": 1.57148} -{"mode": "train", "epoch": 1, "iter": 424, "lr": 0.00027, "memory": 49164, "data_time": 0.0733, "loss_cls_0": 0.7414, "loss_box_0": 1.62027, "loss_cns_0": 0.64219, "loss_yns_0": 0.13836, "loss_cls_1": 0.78071, "loss_box_1": 1.44976, "loss_cns_1": 0.66553, "loss_yns_1": 0.13415, "loss_cls_2": 0.78413, "loss_box_2": 1.42978, "loss_cns_2": 0.66572, "loss_yns_2": 0.13438, "loss_cls_3": 0.78768, "loss_box_3": 1.40512, "loss_cns_3": 0.66475, "loss_yns_3": 0.13307, "loss_cls_4": 0.80426, "loss_box_4": 1.39795, "loss_cns_4": 0.66181, "loss_yns_4": 0.13195, "loss_cls_5": 0.80217, "loss_box_5": 1.39739, "loss_cns_5": 0.66154, "loss_yns_5": 0.13318, "loss_cls_dn_0": 0.11778, "loss_box_dn_0": 0.70748, "loss_cls_dn_1": 0.09562, "loss_box_dn_1": 0.63911, "loss_cls_dn_2": 0.09461, "loss_box_dn_2": 0.62783, "loss_cls_dn_3": 0.09571, "loss_box_dn_3": 0.61842, "loss_cls_dn_4": 0.10212, "loss_box_dn_4": 0.61715, "loss_cls_dn_5": 0.09758, "loss_box_dn_5": 0.61657, "loss_dense_depth": 0.72836, "loss": 23.32558, "grad_norm": 43.01676, "time": 1.669} -{"mode": "train", "epoch": 1, "iter": 425, "lr": 0.00027, "memory": 49164, "data_time": 0.09421, "loss_cls_0": 0.70941, "loss_box_0": 1.59422, "loss_cns_0": 0.64525, "loss_yns_0": 0.13628, "loss_cls_1": 0.76911, "loss_box_1": 1.38199, "loss_cns_1": 0.66519, "loss_yns_1": 0.13025, "loss_cls_2": 0.78306, "loss_box_2": 1.38669, "loss_cns_2": 0.6682, "loss_yns_2": 0.1292, "loss_cls_3": 0.78911, "loss_box_3": 1.37362, "loss_cns_3": 0.66672, "loss_yns_3": 0.1295, "loss_cls_4": 0.80233, "loss_box_4": 1.36919, "loss_cns_4": 0.66855, "loss_yns_4": 0.12909, "loss_cls_5": 0.79734, "loss_box_5": 1.38282, "loss_cns_5": 0.67026, "loss_yns_5": 0.13022, "loss_cls_dn_0": 0.11547, "loss_box_dn_0": 0.71191, "loss_cls_dn_1": 0.09545, "loss_box_dn_1": 0.63157, "loss_cls_dn_2": 0.09319, "loss_box_dn_2": 0.62341, "loss_cls_dn_3": 0.09368, "loss_box_dn_3": 0.62356, "loss_cls_dn_4": 0.09616, "loss_box_dn_4": 0.62542, "loss_cls_dn_5": 0.09391, "loss_box_dn_5": 0.63325, "loss_dense_depth": 0.69874, "loss": 23.04334, "grad_norm": 39.48336, "time": 1.66603} -{"mode": "train", "epoch": 1, "iter": 426, "lr": 0.00027, "memory": 49164, "data_time": 1.54056, "loss_cls_0": 0.71257, "loss_box_0": 1.55489, "loss_cns_0": 0.64091, "loss_yns_0": 0.13938, "loss_cls_1": 0.77089, "loss_box_1": 1.41669, "loss_cns_1": 0.66641, "loss_yns_1": 0.13933, "loss_cls_2": 0.78969, "loss_box_2": 1.39572, "loss_cns_2": 0.66743, "loss_yns_2": 0.13825, "loss_cls_3": 0.79934, "loss_box_3": 1.39442, "loss_cns_3": 0.6672, "loss_yns_3": 0.13618, "loss_cls_4": 0.79741, "loss_box_4": 1.37683, "loss_cns_4": 0.66761, "loss_yns_4": 0.13568, "loss_cls_5": 0.80083, "loss_box_5": 1.37694, "loss_cns_5": 0.66764, "loss_yns_5": 0.13461, "loss_cls_dn_0": 0.11697, "loss_box_dn_0": 0.71048, "loss_cls_dn_1": 0.09087, "loss_box_dn_1": 0.63307, "loss_cls_dn_2": 0.08978, "loss_box_dn_2": 0.62217, "loss_cls_dn_3": 0.08976, "loss_box_dn_3": 0.63287, "loss_cls_dn_4": 0.09318, "loss_box_dn_4": 0.6332, "loss_cls_dn_5": 0.09374, "loss_box_dn_5": 0.63881, "loss_dense_depth": 0.69482, "loss": 23.12659, "grad_norm": 41.07704, "time": 3.16493} -{"mode": "train", "epoch": 1, "iter": 427, "lr": 0.00027, "memory": 49164, "data_time": 0.07064, "loss_cls_0": 0.71091, "loss_box_0": 1.56286, "loss_cns_0": 0.63808, "loss_yns_0": 0.138, "loss_cls_1": 0.78237, "loss_box_1": 1.39722, "loss_cns_1": 0.66492, "loss_yns_1": 0.13606, "loss_cls_2": 0.78716, "loss_box_2": 1.38811, "loss_cns_2": 0.66569, "loss_yns_2": 0.13559, "loss_cls_3": 0.80672, "loss_box_3": 1.38071, "loss_cns_3": 0.66554, "loss_yns_3": 0.13379, "loss_cls_4": 0.81186, "loss_box_4": 1.36783, "loss_cns_4": 0.66306, "loss_yns_4": 0.13425, "loss_cls_5": 0.80808, "loss_box_5": 1.37804, "loss_cns_5": 0.66666, "loss_yns_5": 0.13371, "loss_cls_dn_0": 0.11975, "loss_box_dn_0": 0.72019, "loss_cls_dn_1": 0.093, "loss_box_dn_1": 0.63919, "loss_cls_dn_2": 0.09203, "loss_box_dn_2": 0.63662, "loss_cls_dn_3": 0.0954, "loss_box_dn_3": 0.63677, "loss_cls_dn_4": 0.09521, "loss_box_dn_4": 0.64148, "loss_cls_dn_5": 0.09364, "loss_box_dn_5": 0.64716, "loss_dense_depth": 0.69743, "loss": 23.16512, "grad_norm": 40.71473, "time": 1.54422} -{"mode": "train", "epoch": 1, "iter": 428, "lr": 0.00027, "memory": 49164, "data_time": 0.07053, "loss_cls_0": 0.70923, "loss_box_0": 1.57035, "loss_cns_0": 0.64108, "loss_yns_0": 0.1432, "loss_cls_1": 0.76199, "loss_box_1": 1.41918, "loss_cns_1": 0.66383, "loss_yns_1": 0.13824, "loss_cls_2": 0.77448, "loss_box_2": 1.39744, "loss_cns_2": 0.66478, "loss_yns_2": 0.13807, "loss_cls_3": 0.78602, "loss_box_3": 1.36767, "loss_cns_3": 0.6646, "loss_yns_3": 0.13803, "loss_cls_4": 0.8019, "loss_box_4": 1.35415, "loss_cns_4": 0.66362, "loss_yns_4": 0.13827, "loss_cls_5": 0.79875, "loss_box_5": 1.36842, "loss_cns_5": 0.66654, "loss_yns_5": 0.1382, "loss_cls_dn_0": 0.1182, "loss_box_dn_0": 0.71458, "loss_cls_dn_1": 0.09464, "loss_box_dn_1": 0.64864, "loss_cls_dn_2": 0.09393, "loss_box_dn_2": 0.64469, "loss_cls_dn_3": 0.0981, "loss_box_dn_3": 0.63414, "loss_cls_dn_4": 0.09661, "loss_box_dn_4": 0.63386, "loss_cls_dn_5": 0.09661, "loss_box_dn_5": 0.64076, "loss_dense_depth": 0.7043, "loss": 23.1271, "grad_norm": 30.5194, "time": 1.55252} -{"mode": "train", "epoch": 1, "iter": 429, "lr": 0.00027, "memory": 49164, "data_time": 0.07698, "loss_cls_0": 0.69944, "loss_box_0": 1.57442, "loss_cns_0": 0.64035, "loss_yns_0": 0.14053, "loss_cls_1": 0.74736, "loss_box_1": 1.41, "loss_cns_1": 0.66064, "loss_yns_1": 0.13689, "loss_cls_2": 0.76839, "loss_box_2": 1.39202, "loss_cns_2": 0.66069, "loss_yns_2": 0.13443, "loss_cls_3": 0.77882, "loss_box_3": 1.3629, "loss_cns_3": 0.66027, "loss_yns_3": 0.13435, "loss_cls_4": 0.79162, "loss_box_4": 1.37318, "loss_cns_4": 0.66227, "loss_yns_4": 0.13626, "loss_cls_5": 0.79337, "loss_box_5": 1.36409, "loss_cns_5": 0.66271, "loss_yns_5": 0.13889, "loss_cls_dn_0": 0.1091, "loss_box_dn_0": 0.71203, "loss_cls_dn_1": 0.09143, "loss_box_dn_1": 0.62775, "loss_cls_dn_2": 0.09073, "loss_box_dn_2": 0.62645, "loss_cls_dn_3": 0.09297, "loss_box_dn_3": 0.61328, "loss_cls_dn_4": 0.09296, "loss_box_dn_4": 0.61462, "loss_cls_dn_5": 0.09412, "loss_box_dn_5": 0.60871, "loss_dense_depth": 0.72991, "loss": 22.92798, "grad_norm": 39.7986, "time": 1.55663} -{"mode": "train", "epoch": 1, "iter": 430, "lr": 0.00027, "memory": 49164, "data_time": 0.10231, "loss_cls_0": 0.70105, "loss_box_0": 1.55424, "loss_cns_0": 0.64406, "loss_yns_0": 0.13722, "loss_cls_1": 0.76306, "loss_box_1": 1.39659, "loss_cns_1": 0.66335, "loss_yns_1": 0.13517, "loss_cls_2": 0.77003, "loss_box_2": 1.36809, "loss_cns_2": 0.66068, "loss_yns_2": 0.13447, "loss_cls_3": 0.77541, "loss_box_3": 1.34591, "loss_cns_3": 0.65878, "loss_yns_3": 0.13201, "loss_cls_4": 0.7848, "loss_box_4": 1.36818, "loss_cns_4": 0.66256, "loss_yns_4": 0.13491, "loss_cls_5": 0.79132, "loss_box_5": 1.36713, "loss_cns_5": 0.66418, "loss_yns_5": 0.13615, "loss_cls_dn_0": 0.11333, "loss_box_dn_0": 0.70848, "loss_cls_dn_1": 0.09548, "loss_box_dn_1": 0.60936, "loss_cls_dn_2": 0.09631, "loss_box_dn_2": 0.59152, "loss_cls_dn_3": 0.0968, "loss_box_dn_3": 0.58335, "loss_cls_dn_4": 0.09883, "loss_box_dn_4": 0.58682, "loss_cls_dn_5": 0.09887, "loss_box_dn_5": 0.58623, "loss_dense_depth": 0.69185, "loss": 22.70658, "grad_norm": 28.9265, "time": 1.60809} -{"mode": "train", "epoch": 1, "iter": 431, "lr": 0.00027, "memory": 49164, "data_time": 0.07564, "loss_cls_0": 0.70875, "loss_box_0": 1.55849, "loss_cns_0": 0.64093, "loss_yns_0": 0.13865, "loss_cls_1": 0.7766, "loss_box_1": 1.3975, "loss_cns_1": 0.66377, "loss_yns_1": 0.13494, "loss_cls_2": 0.78206, "loss_box_2": 1.35956, "loss_cns_2": 0.66039, "loss_yns_2": 0.13493, "loss_cls_3": 0.80063, "loss_box_3": 1.32834, "loss_cns_3": 0.65553, "loss_yns_3": 0.13158, "loss_cls_4": 0.80013, "loss_box_4": 1.35362, "loss_cns_4": 0.66395, "loss_yns_4": 0.13433, "loss_cls_5": 0.80706, "loss_box_5": 1.37168, "loss_cns_5": 0.66704, "loss_yns_5": 0.13455, "loss_cls_dn_0": 0.11813, "loss_box_dn_0": 0.7089, "loss_cls_dn_1": 0.09691, "loss_box_dn_1": 0.62818, "loss_cls_dn_2": 0.09898, "loss_box_dn_2": 0.60829, "loss_cls_dn_3": 0.0994, "loss_box_dn_3": 0.60667, "loss_cls_dn_4": 0.10191, "loss_box_dn_4": 0.60667, "loss_cls_dn_5": 0.102, "loss_box_dn_5": 0.61786, "loss_dense_depth": 0.7216, "loss": 22.92052, "grad_norm": 31.24308, "time": 1.5737} -{"mode": "train", "epoch": 1, "iter": 432, "lr": 0.00027, "memory": 49164, "data_time": 0.07488, "loss_cls_0": 0.71868, "loss_box_0": 1.56902, "loss_cns_0": 0.6396, "loss_yns_0": 0.14005, "loss_cls_1": 0.77831, "loss_box_1": 1.40837, "loss_cns_1": 0.6635, "loss_yns_1": 0.13669, "loss_cls_2": 0.78442, "loss_box_2": 1.39315, "loss_cns_2": 0.66228, "loss_yns_2": 0.13754, "loss_cls_3": 0.80715, "loss_box_3": 1.36294, "loss_cns_3": 0.65949, "loss_yns_3": 0.13607, "loss_cls_4": 0.80145, "loss_box_4": 1.38414, "loss_cns_4": 0.66456, "loss_yns_4": 0.13638, "loss_cls_5": 0.80268, "loss_box_5": 1.39243, "loss_cns_5": 0.66488, "loss_yns_5": 0.13806, "loss_cls_dn_0": 0.11908, "loss_box_dn_0": 0.7162, "loss_cls_dn_1": 0.10045, "loss_box_dn_1": 0.6404, "loss_cls_dn_2": 0.10235, "loss_box_dn_2": 0.63352, "loss_cls_dn_3": 0.10379, "loss_box_dn_3": 0.63226, "loss_cls_dn_4": 0.10518, "loss_box_dn_4": 0.6376, "loss_cls_dn_5": 0.10345, "loss_box_dn_5": 0.64609, "loss_dense_depth": 0.71365, "loss": 23.23587, "grad_norm": 32.81714, "time": 1.57427} -{"mode": "train", "epoch": 1, "iter": 433, "lr": 0.00027, "memory": 49164, "data_time": 0.0889, "loss_cls_0": 0.7001, "loss_box_0": 1.58753, "loss_cns_0": 0.64077, "loss_yns_0": 0.1398, "loss_cls_1": 0.77106, "loss_box_1": 1.42712, "loss_cns_1": 0.66417, "loss_yns_1": 0.13558, "loss_cls_2": 0.77938, "loss_box_2": 1.41011, "loss_cns_2": 0.66368, "loss_yns_2": 0.13376, "loss_cls_3": 0.7843, "loss_box_3": 1.38633, "loss_cns_3": 0.66384, "loss_yns_3": 0.13389, "loss_cls_4": 0.7846, "loss_box_4": 1.38822, "loss_cns_4": 0.66508, "loss_yns_4": 0.13408, "loss_cls_5": 0.78464, "loss_box_5": 1.38326, "loss_cns_5": 0.66377, "loss_yns_5": 0.1344, "loss_cls_dn_0": 0.11295, "loss_box_dn_0": 0.71139, "loss_cls_dn_1": 0.10378, "loss_box_dn_1": 0.65146, "loss_cls_dn_2": 0.10216, "loss_box_dn_2": 0.64388, "loss_cls_dn_3": 0.10106, "loss_box_dn_3": 0.64047, "loss_cls_dn_4": 0.1022, "loss_box_dn_4": 0.6429, "loss_cls_dn_5": 0.10218, "loss_box_dn_5": 0.6461, "loss_dense_depth": 0.69124, "loss": 23.2112, "grad_norm": 33.36891, "time": 1.62478} -{"mode": "train", "epoch": 1, "iter": 434, "lr": 0.00027, "memory": 49164, "data_time": 0.08644, "loss_cls_0": 0.72099, "loss_box_0": 1.59013, "loss_cns_0": 0.63967, "loss_yns_0": 0.14052, "loss_cls_1": 0.77138, "loss_box_1": 1.44863, "loss_cns_1": 0.66688, "loss_yns_1": 0.13766, "loss_cls_2": 0.7784, "loss_box_2": 1.42213, "loss_cns_2": 0.66133, "loss_yns_2": 0.13716, "loss_cls_3": 0.78519, "loss_box_3": 1.40094, "loss_cns_3": 0.66414, "loss_yns_3": 0.138, "loss_cls_4": 0.78, "loss_box_4": 1.39609, "loss_cns_4": 0.66211, "loss_yns_4": 0.13766, "loss_cls_5": 0.79001, "loss_box_5": 1.39183, "loss_cns_5": 0.66222, "loss_yns_5": 0.13711, "loss_cls_dn_0": 0.11284, "loss_box_dn_0": 0.71069, "loss_cls_dn_1": 0.0963, "loss_box_dn_1": 0.6542, "loss_cls_dn_2": 0.09571, "loss_box_dn_2": 0.63701, "loss_cls_dn_3": 0.0959, "loss_box_dn_3": 0.62791, "loss_cls_dn_4": 0.10004, "loss_box_dn_4": 0.62483, "loss_cls_dn_5": 0.09902, "loss_box_dn_5": 0.62327, "loss_dense_depth": 0.70098, "loss": 23.23886, "grad_norm": 25.91738, "time": 1.5747} -{"mode": "train", "epoch": 1, "iter": 435, "lr": 0.00027, "memory": 49164, "data_time": 0.09798, "loss_cls_0": 0.73004, "loss_box_0": 1.62075, "loss_cns_0": 0.63618, "loss_yns_0": 0.14053, "loss_cls_1": 0.78444, "loss_box_1": 1.44857, "loss_cns_1": 0.66362, "loss_yns_1": 0.13712, "loss_cls_2": 0.78917, "loss_box_2": 1.42856, "loss_cns_2": 0.66106, "loss_yns_2": 0.13693, "loss_cls_3": 0.80269, "loss_box_3": 1.41664, "loss_cns_3": 0.66088, "loss_yns_3": 0.13727, "loss_cls_4": 0.79855, "loss_box_4": 1.41031, "loss_cns_4": 0.66307, "loss_yns_4": 0.13656, "loss_cls_5": 0.80686, "loss_box_5": 1.40371, "loss_cns_5": 0.66048, "loss_yns_5": 0.13635, "loss_cls_dn_0": 0.11474, "loss_box_dn_0": 0.70876, "loss_cls_dn_1": 0.09457, "loss_box_dn_1": 0.63442, "loss_cls_dn_2": 0.09342, "loss_box_dn_2": 0.61754, "loss_cls_dn_3": 0.09686, "loss_box_dn_3": 0.61234, "loss_cls_dn_4": 0.0979, "loss_box_dn_4": 0.61096, "loss_cls_dn_5": 0.09557, "loss_box_dn_5": 0.61503, "loss_dense_depth": 0.70075, "loss": 23.30322, "grad_norm": 35.00614, "time": 1.61727} -{"mode": "train", "epoch": 1, "iter": 436, "lr": 0.00027, "memory": 49164, "data_time": 0.10418, "loss_cls_0": 0.71572, "loss_box_0": 1.59536, "loss_cns_0": 0.6361, "loss_yns_0": 0.14077, "loss_cls_1": 0.7989, "loss_box_1": 1.45687, "loss_cns_1": 0.66493, "loss_yns_1": 0.13726, "loss_cls_2": 0.80015, "loss_box_2": 1.42223, "loss_cns_2": 0.66623, "loss_yns_2": 0.138, "loss_cls_3": 0.79938, "loss_box_3": 1.4283, "loss_cns_3": 0.66273, "loss_yns_3": 0.13725, "loss_cls_4": 0.80005, "loss_box_4": 1.42206, "loss_cns_4": 0.66931, "loss_yns_4": 0.13621, "loss_cls_5": 0.80416, "loss_box_5": 1.42956, "loss_cns_5": 0.66287, "loss_yns_5": 0.13636, "loss_cls_dn_0": 0.11128, "loss_box_dn_0": 0.71421, "loss_cls_dn_1": 0.09138, "loss_box_dn_1": 0.62884, "loss_cls_dn_2": 0.08883, "loss_box_dn_2": 0.61373, "loss_cls_dn_3": 0.09129, "loss_box_dn_3": 0.61837, "loss_cls_dn_4": 0.09442, "loss_box_dn_4": 0.61631, "loss_cls_dn_5": 0.09144, "loss_box_dn_5": 0.62606, "loss_dense_depth": 0.69213, "loss": 23.33906, "grad_norm": 30.89602, "time": 1.59698} -{"mode": "train", "epoch": 1, "iter": 437, "lr": 0.00027, "memory": 49164, "data_time": 0.08438, "loss_cls_0": 0.69375, "loss_box_0": 1.58093, "loss_cns_0": 0.63926, "loss_yns_0": 0.14129, "loss_cls_1": 0.78942, "loss_box_1": 1.4492, "loss_cns_1": 0.66519, "loss_yns_1": 0.13681, "loss_cls_2": 0.78982, "loss_box_2": 1.42433, "loss_cns_2": 0.66616, "loss_yns_2": 0.13678, "loss_cls_3": 0.78642, "loss_box_3": 1.42117, "loss_cns_3": 0.66193, "loss_yns_3": 0.13687, "loss_cls_4": 0.78625, "loss_box_4": 1.41922, "loss_cns_4": 0.66434, "loss_yns_4": 0.1366, "loss_cls_5": 0.79409, "loss_box_5": 1.42122, "loss_cns_5": 0.66354, "loss_yns_5": 0.13729, "loss_cls_dn_0": 0.10809, "loss_box_dn_0": 0.70903, "loss_cls_dn_1": 0.09366, "loss_box_dn_1": 0.62942, "loss_cls_dn_2": 0.09176, "loss_box_dn_2": 0.61426, "loss_cls_dn_3": 0.09169, "loss_box_dn_3": 0.61615, "loss_cls_dn_4": 0.09433, "loss_box_dn_4": 0.61718, "loss_cls_dn_5": 0.09451, "loss_box_dn_5": 0.62366, "loss_dense_depth": 0.67621, "loss": 23.20182, "grad_norm": 27.66455, "time": 1.57609} -{"mode": "train", "epoch": 1, "iter": 438, "lr": 0.00027, "memory": 49164, "data_time": 0.0836, "loss_cls_0": 0.7317, "loss_box_0": 1.62033, "loss_cns_0": 0.63292, "loss_yns_0": 0.13977, "loss_cls_1": 0.79427, "loss_box_1": 1.42056, "loss_cns_1": 0.66324, "loss_yns_1": 0.13862, "loss_cls_2": 0.79988, "loss_box_2": 1.40478, "loss_cns_2": 0.66073, "loss_yns_2": 0.13801, "loss_cls_3": 0.80026, "loss_box_3": 1.42605, "loss_cns_3": 0.66147, "loss_yns_3": 0.13747, "loss_cls_4": 0.80995, "loss_box_4": 1.41534, "loss_cns_4": 0.66244, "loss_yns_4": 0.13804, "loss_cls_5": 0.8184, "loss_box_5": 1.40723, "loss_cns_5": 0.66163, "loss_yns_5": 0.13834, "loss_cls_dn_0": 0.11577, "loss_box_dn_0": 0.72035, "loss_cls_dn_1": 0.09716, "loss_box_dn_1": 0.62286, "loss_cls_dn_2": 0.09606, "loss_box_dn_2": 0.60913, "loss_cls_dn_3": 0.09481, "loss_box_dn_3": 0.61581, "loss_cls_dn_4": 0.09505, "loss_box_dn_4": 0.61627, "loss_cls_dn_5": 0.09702, "loss_box_dn_5": 0.61851, "loss_dense_depth": 0.69184, "loss": 23.31207, "grad_norm": 32.53771, "time": 1.59329} -{"mode": "train", "epoch": 1, "iter": 439, "lr": 0.00027, "memory": 49164, "data_time": 0.18277, "loss_cls_0": 0.71777, "loss_box_0": 1.58434, "loss_cns_0": 0.63816, "loss_yns_0": 0.1384, "loss_cls_1": 0.77266, "loss_box_1": 1.40654, "loss_cns_1": 0.65907, "loss_yns_1": 0.13642, "loss_cls_2": 0.77439, "loss_box_2": 1.40803, "loss_cns_2": 0.66252, "loss_yns_2": 0.13795, "loss_cls_3": 0.78864, "loss_box_3": 1.39097, "loss_cns_3": 0.65893, "loss_yns_3": 0.13579, "loss_cls_4": 0.78688, "loss_box_4": 1.38471, "loss_cns_4": 0.66402, "loss_yns_4": 0.13567, "loss_cls_5": 0.78534, "loss_box_5": 1.38416, "loss_cns_5": 0.66185, "loss_yns_5": 0.13593, "loss_cls_dn_0": 0.10845, "loss_box_dn_0": 0.70683, "loss_cls_dn_1": 0.09474, "loss_box_dn_1": 0.61864, "loss_cls_dn_2": 0.09229, "loss_box_dn_2": 0.60551, "loss_cls_dn_3": 0.09113, "loss_box_dn_3": 0.60251, "loss_cls_dn_4": 0.09216, "loss_box_dn_4": 0.60477, "loss_cls_dn_5": 0.09195, "loss_box_dn_5": 0.60487, "loss_dense_depth": 0.70374, "loss": 22.9667, "grad_norm": 33.12938, "time": 1.65292} -{"mode": "train", "epoch": 1, "iter": 440, "lr": 0.00028, "memory": 49164, "data_time": 0.07353, "loss_cls_0": 0.73522, "loss_box_0": 1.59235, "loss_cns_0": 0.63956, "loss_yns_0": 0.13991, "loss_cls_1": 0.77751, "loss_box_1": 1.41227, "loss_cns_1": 0.66289, "loss_yns_1": 0.13537, "loss_cls_2": 0.77887, "loss_box_2": 1.41627, "loss_cns_2": 0.66527, "loss_yns_2": 0.13576, "loss_cls_3": 0.78313, "loss_box_3": 1.39146, "loss_cns_3": 0.66203, "loss_yns_3": 0.13674, "loss_cls_4": 0.78464, "loss_box_4": 1.39437, "loss_cns_4": 0.66716, "loss_yns_4": 0.13718, "loss_cls_5": 0.79462, "loss_box_5": 1.40017, "loss_cns_5": 0.66408, "loss_yns_5": 0.13781, "loss_cls_dn_0": 0.1108, "loss_box_dn_0": 0.70892, "loss_cls_dn_1": 0.09282, "loss_box_dn_1": 0.61187, "loss_cls_dn_2": 0.09015, "loss_box_dn_2": 0.60096, "loss_cls_dn_3": 0.09306, "loss_box_dn_3": 0.59348, "loss_cls_dn_4": 0.09416, "loss_box_dn_4": 0.59843, "loss_cls_dn_5": 0.09158, "loss_box_dn_5": 0.59768, "loss_dense_depth": 0.70435, "loss": 23.03293, "grad_norm": 26.22157, "time": 1.616} -{"mode": "train", "epoch": 1, "iter": 441, "lr": 0.00028, "memory": 49164, "data_time": 0.0773, "loss_cls_0": 0.71784, "loss_box_0": 1.57997, "loss_cns_0": 0.63569, "loss_yns_0": 0.13955, "loss_cls_1": 0.77357, "loss_box_1": 1.44351, "loss_cns_1": 0.66291, "loss_yns_1": 0.13712, "loss_cls_2": 0.78637, "loss_box_2": 1.4202, "loss_cns_2": 0.66231, "loss_yns_2": 0.13666, "loss_cls_3": 0.79039, "loss_box_3": 1.40305, "loss_cns_3": 0.66108, "loss_yns_3": 0.13601, "loss_cls_4": 0.7888, "loss_box_4": 1.40297, "loss_cns_4": 0.66356, "loss_yns_4": 0.13679, "loss_cls_5": 0.79361, "loss_box_5": 1.39947, "loss_cns_5": 0.66247, "loss_yns_5": 0.13683, "loss_cls_dn_0": 0.10714, "loss_box_dn_0": 0.70109, "loss_cls_dn_1": 0.09166, "loss_box_dn_1": 0.61502, "loss_cls_dn_2": 0.09028, "loss_box_dn_2": 0.59892, "loss_cls_dn_3": 0.09401, "loss_box_dn_3": 0.59786, "loss_cls_dn_4": 0.09523, "loss_box_dn_4": 0.59981, "loss_cls_dn_5": 0.09498, "loss_box_dn_5": 0.60056, "loss_dense_depth": 0.73252, "loss": 23.0898, "grad_norm": 30.68122, "time": 1.67471} -{"mode": "train", "epoch": 1, "iter": 442, "lr": 0.00028, "memory": 49164, "data_time": 0.08949, "loss_cls_0": 0.7267, "loss_box_0": 1.55682, "loss_cns_0": 0.63929, "loss_yns_0": 0.13864, "loss_cls_1": 0.78113, "loss_box_1": 1.44239, "loss_cns_1": 0.66474, "loss_yns_1": 0.13712, "loss_cls_2": 0.7883, "loss_box_2": 1.41485, "loss_cns_2": 0.66131, "loss_yns_2": 0.13625, "loss_cls_3": 0.78802, "loss_box_3": 1.42591, "loss_cns_3": 0.66034, "loss_yns_3": 0.13589, "loss_cls_4": 0.79406, "loss_box_4": 1.44195, "loss_cns_4": 0.66226, "loss_yns_4": 0.13712, "loss_cls_5": 0.79782, "loss_box_5": 1.42993, "loss_cns_5": 0.66179, "loss_yns_5": 0.13727, "loss_cls_dn_0": 0.10718, "loss_box_dn_0": 0.7083, "loss_cls_dn_1": 0.08944, "loss_box_dn_1": 0.62786, "loss_cls_dn_2": 0.08902, "loss_box_dn_2": 0.61277, "loss_cls_dn_3": 0.09096, "loss_box_dn_3": 0.62022, "loss_cls_dn_4": 0.09087, "loss_box_dn_4": 0.62851, "loss_cls_dn_5": 0.09434, "loss_box_dn_5": 0.62852, "loss_dense_depth": 0.72743, "loss": 23.27532, "grad_norm": 42.10378, "time": 1.65155} -{"mode": "train", "epoch": 1, "iter": 443, "lr": 0.00028, "memory": 49164, "data_time": 0.07429, "loss_cls_0": 0.7174, "loss_box_0": 1.53686, "loss_cns_0": 0.64159, "loss_yns_0": 0.1402, "loss_cls_1": 0.77804, "loss_box_1": 1.43865, "loss_cns_1": 0.66785, "loss_yns_1": 0.13734, "loss_cls_2": 0.78262, "loss_box_2": 1.41535, "loss_cns_2": 0.66602, "loss_yns_2": 0.13803, "loss_cls_3": 0.78553, "loss_box_3": 1.40693, "loss_cns_3": 0.66378, "loss_yns_3": 0.13649, "loss_cls_4": 0.79774, "loss_box_4": 1.40503, "loss_cns_4": 0.66553, "loss_yns_4": 0.13737, "loss_cls_5": 0.79099, "loss_box_5": 1.41006, "loss_cns_5": 0.66529, "loss_yns_5": 0.13746, "loss_cls_dn_0": 0.10907, "loss_box_dn_0": 0.70823, "loss_cls_dn_1": 0.09029, "loss_box_dn_1": 0.62202, "loss_cls_dn_2": 0.09099, "loss_box_dn_2": 0.60562, "loss_cls_dn_3": 0.09034, "loss_box_dn_3": 0.60256, "loss_cls_dn_4": 0.09023, "loss_box_dn_4": 0.60411, "loss_cls_dn_5": 0.09256, "loss_box_dn_5": 0.60761, "loss_dense_depth": 0.70626, "loss": 23.08201, "grad_norm": 26.8994, "time": 1.59219} -{"mode": "train", "epoch": 1, "iter": 444, "lr": 0.00028, "memory": 49164, "data_time": 0.1195, "loss_cls_0": 0.71963, "loss_box_0": 1.57216, "loss_cns_0": 0.64266, "loss_yns_0": 0.14007, "loss_cls_1": 0.77637, "loss_box_1": 1.43543, "loss_cns_1": 0.66512, "loss_yns_1": 0.13646, "loss_cls_2": 0.77831, "loss_box_2": 1.43541, "loss_cns_2": 0.66415, "loss_yns_2": 0.13713, "loss_cls_3": 0.78315, "loss_box_3": 1.3974, "loss_cns_3": 0.66081, "loss_yns_3": 0.13674, "loss_cls_4": 0.78995, "loss_box_4": 1.41037, "loss_cns_4": 0.66393, "loss_yns_4": 0.13639, "loss_cls_5": 0.79172, "loss_box_5": 1.41971, "loss_cns_5": 0.66413, "loss_yns_5": 0.13644, "loss_cls_dn_0": 0.10523, "loss_box_dn_0": 0.71428, "loss_cls_dn_1": 0.08834, "loss_box_dn_1": 0.62233, "loss_cls_dn_2": 0.0886, "loss_box_dn_2": 0.61427, "loss_cls_dn_3": 0.08859, "loss_box_dn_3": 0.60066, "loss_cls_dn_4": 0.08949, "loss_box_dn_4": 0.60409, "loss_cls_dn_5": 0.08946, "loss_box_dn_5": 0.60749, "loss_dense_depth": 0.80711, "loss": 23.2136, "grad_norm": 38.07526, "time": 1.615} -{"mode": "train", "epoch": 1, "iter": 445, "lr": 0.00028, "memory": 49164, "data_time": 0.07134, "loss_cls_0": 0.74846, "loss_box_0": 1.5944, "loss_cns_0": 0.64206, "loss_yns_0": 0.14225, "loss_cls_1": 0.78669, "loss_box_1": 1.43493, "loss_cns_1": 0.66017, "loss_yns_1": 0.13926, "loss_cls_2": 0.79713, "loss_box_2": 1.43285, "loss_cns_2": 0.66201, "loss_yns_2": 0.13968, "loss_cls_3": 0.8029, "loss_box_3": 1.40338, "loss_cns_3": 0.66052, "loss_yns_3": 0.13843, "loss_cls_4": 0.80778, "loss_box_4": 1.40292, "loss_cns_4": 0.66129, "loss_yns_4": 0.13896, "loss_cls_5": 0.81385, "loss_box_5": 1.39183, "loss_cns_5": 0.6581, "loss_yns_5": 0.13787, "loss_cls_dn_0": 0.10827, "loss_box_dn_0": 0.71443, "loss_cls_dn_1": 0.0895, "loss_box_dn_1": 0.61556, "loss_cls_dn_2": 0.08988, "loss_box_dn_2": 0.6095, "loss_cls_dn_3": 0.09357, "loss_box_dn_3": 0.60216, "loss_cls_dn_4": 0.0912, "loss_box_dn_4": 0.60143, "loss_cls_dn_5": 0.09347, "loss_box_dn_5": 0.59876, "loss_dense_depth": 0.7033, "loss": 23.20873, "grad_norm": 33.29106, "time": 1.63823} -{"mode": "train", "epoch": 1, "iter": 446, "lr": 0.00028, "memory": 49164, "data_time": 0.10683, "loss_cls_0": 0.7378, "loss_box_0": 1.55374, "loss_cns_0": 0.64121, "loss_yns_0": 0.14161, "loss_cls_1": 0.78432, "loss_box_1": 1.4009, "loss_cns_1": 0.66108, "loss_yns_1": 0.13595, "loss_cls_2": 0.79845, "loss_box_2": 1.38067, "loss_cns_2": 0.66276, "loss_yns_2": 0.13638, "loss_cls_3": 0.79856, "loss_box_3": 1.36463, "loss_cns_3": 0.66186, "loss_yns_3": 0.13556, "loss_cls_4": 0.79735, "loss_box_4": 1.36379, "loss_cns_4": 0.66189, "loss_yns_4": 0.1363, "loss_cls_5": 0.79541, "loss_box_5": 1.3594, "loss_cns_5": 0.6609, "loss_yns_5": 0.13633, "loss_cls_dn_0": 0.10636, "loss_box_dn_0": 0.71074, "loss_cls_dn_1": 0.08878, "loss_box_dn_1": 0.61489, "loss_cls_dn_2": 0.08942, "loss_box_dn_2": 0.59856, "loss_cls_dn_3": 0.09233, "loss_box_dn_3": 0.59316, "loss_cls_dn_4": 0.09115, "loss_box_dn_4": 0.59722, "loss_cls_dn_5": 0.09362, "loss_box_dn_5": 0.59641, "loss_dense_depth": 0.71612, "loss": 22.89561, "grad_norm": 27.94825, "time": 1.70766} -{"mode": "train", "epoch": 1, "iter": 447, "lr": 0.00028, "memory": 49164, "data_time": 0.10915, "loss_cls_0": 0.7273, "loss_box_0": 1.54419, "loss_cns_0": 0.64329, "loss_yns_0": 0.142, "loss_cls_1": 0.78795, "loss_box_1": 1.40805, "loss_cns_1": 0.66544, "loss_yns_1": 0.13953, "loss_cls_2": 0.79396, "loss_box_2": 1.38168, "loss_cns_2": 0.66577, "loss_yns_2": 0.13828, "loss_cls_3": 0.79953, "loss_box_3": 1.38831, "loss_cns_3": 0.66327, "loss_yns_3": 0.13893, "loss_cls_4": 0.80388, "loss_box_4": 1.38418, "loss_cns_4": 0.66248, "loss_yns_4": 0.1397, "loss_cls_5": 0.80363, "loss_box_5": 1.38889, "loss_cns_5": 0.6609, "loss_yns_5": 0.13875, "loss_cls_dn_0": 0.10586, "loss_box_dn_0": 0.70741, "loss_cls_dn_1": 0.08978, "loss_box_dn_1": 0.62469, "loss_cls_dn_2": 0.08883, "loss_box_dn_2": 0.60609, "loss_cls_dn_3": 0.08971, "loss_box_dn_3": 0.60798, "loss_cls_dn_4": 0.08881, "loss_box_dn_4": 0.61109, "loss_cls_dn_5": 0.08863, "loss_box_dn_5": 0.61745, "loss_dense_depth": 0.74702, "loss": 23.08325, "grad_norm": 30.65774, "time": 1.58659} -{"mode": "train", "epoch": 1, "iter": 448, "lr": 0.00028, "memory": 49164, "data_time": 0.07173, "loss_cls_0": 0.71382, "loss_box_0": 1.55635, "loss_cns_0": 0.64487, "loss_yns_0": 0.14039, "loss_cls_1": 0.78477, "loss_box_1": 1.41278, "loss_cns_1": 0.66863, "loss_yns_1": 0.13993, "loss_cls_2": 0.78942, "loss_box_2": 1.39997, "loss_cns_2": 0.67238, "loss_yns_2": 0.13905, "loss_cls_3": 0.79931, "loss_box_3": 1.39414, "loss_cns_3": 0.66728, "loss_yns_3": 0.13726, "loss_cls_4": 0.80424, "loss_box_4": 1.40219, "loss_cns_4": 0.6654, "loss_yns_4": 0.13773, "loss_cls_5": 0.80626, "loss_box_5": 1.4051, "loss_cns_5": 0.66585, "loss_yns_5": 0.13806, "loss_cls_dn_0": 0.10953, "loss_box_dn_0": 0.70148, "loss_cls_dn_1": 0.09086, "loss_box_dn_1": 0.61836, "loss_cls_dn_2": 0.08859, "loss_box_dn_2": 0.61019, "loss_cls_dn_3": 0.08855, "loss_box_dn_3": 0.60934, "loss_cls_dn_4": 0.08712, "loss_box_dn_4": 0.61503, "loss_cls_dn_5": 0.08767, "loss_box_dn_5": 0.61944, "loss_dense_depth": 0.67746, "loss": 23.0888, "grad_norm": 33.14681, "time": 1.55208} -{"mode": "train", "epoch": 1, "iter": 449, "lr": 0.00028, "memory": 49164, "data_time": 0.0706, "loss_cls_0": 0.70997, "loss_box_0": 1.5317, "loss_cns_0": 0.63942, "loss_yns_0": 0.13726, "loss_cls_1": 0.78499, "loss_box_1": 1.42472, "loss_cns_1": 0.66536, "loss_yns_1": 0.13758, "loss_cls_2": 0.79428, "loss_box_2": 1.40231, "loss_cns_2": 0.66895, "loss_yns_2": 0.13994, "loss_cls_3": 0.79887, "loss_box_3": 1.38995, "loss_cns_3": 0.66741, "loss_yns_3": 0.13803, "loss_cls_4": 0.81005, "loss_box_4": 1.38943, "loss_cns_4": 0.66722, "loss_yns_4": 0.14003, "loss_cls_5": 0.80489, "loss_box_5": 1.38217, "loss_cns_5": 0.66695, "loss_yns_5": 0.13742, "loss_cls_dn_0": 0.1095, "loss_box_dn_0": 0.70056, "loss_cls_dn_1": 0.09158, "loss_box_dn_1": 0.61826, "loss_cls_dn_2": 0.09, "loss_box_dn_2": 0.60829, "loss_cls_dn_3": 0.08997, "loss_box_dn_3": 0.60223, "loss_cls_dn_4": 0.0901, "loss_box_dn_4": 0.6016, "loss_cls_dn_5": 0.08985, "loss_box_dn_5": 0.59951, "loss_dense_depth": 0.70075, "loss": 23.02113, "grad_norm": 31.9611, "time": 1.61329} -{"mode": "train", "epoch": 1, "iter": 450, "lr": 0.00028, "memory": 49164, "data_time": 0.07245, "loss_cls_0": 0.73775, "loss_box_0": 1.56611, "loss_cns_0": 0.62943, "loss_yns_0": 0.14088, "loss_cls_1": 0.78524, "loss_box_1": 1.44031, "loss_cns_1": 0.66518, "loss_yns_1": 0.13824, "loss_cls_2": 0.7993, "loss_box_2": 1.4101, "loss_cns_2": 0.6646, "loss_yns_2": 0.14027, "loss_cls_3": 0.81288, "loss_box_3": 1.39546, "loss_cns_3": 0.66211, "loss_yns_3": 0.13799, "loss_cls_4": 0.81071, "loss_box_4": 1.3858, "loss_cns_4": 0.66356, "loss_yns_4": 0.13829, "loss_cls_5": 0.81791, "loss_box_5": 1.39333, "loss_cns_5": 0.66446, "loss_yns_5": 0.13708, "loss_cls_dn_0": 0.10806, "loss_box_dn_0": 0.70417, "loss_cls_dn_1": 0.0899, "loss_box_dn_1": 0.61666, "loss_cls_dn_2": 0.09141, "loss_box_dn_2": 0.59615, "loss_cls_dn_3": 0.09193, "loss_box_dn_3": 0.59087, "loss_cls_dn_4": 0.09063, "loss_box_dn_4": 0.58385, "loss_cls_dn_5": 0.09288, "loss_box_dn_5": 0.5906, "loss_dense_depth": 0.71135, "loss": 23.09546, "grad_norm": 25.86131, "time": 1.55631} -{"mode": "train", "epoch": 1, "iter": 451, "lr": 0.00028, "memory": 49164, "data_time": 0.07362, "loss_cls_0": 0.73426, "loss_box_0": 1.57472, "loss_cns_0": 0.63972, "loss_yns_0": 0.13923, "loss_cls_1": 0.77125, "loss_box_1": 1.44945, "loss_cns_1": 0.66296, "loss_yns_1": 0.13619, "loss_cls_2": 0.7778, "loss_box_2": 1.41894, "loss_cns_2": 0.66498, "loss_yns_2": 0.13617, "loss_cls_3": 0.78092, "loss_box_3": 1.40666, "loss_cns_3": 0.66346, "loss_yns_3": 0.13633, "loss_cls_4": 0.79105, "loss_box_4": 1.41159, "loss_cns_4": 0.66302, "loss_yns_4": 0.13648, "loss_cls_5": 0.79821, "loss_box_5": 1.42586, "loss_cns_5": 0.665, "loss_yns_5": 0.13675, "loss_cls_dn_0": 0.10778, "loss_box_dn_0": 0.70904, "loss_cls_dn_1": 0.08974, "loss_box_dn_1": 0.61473, "loss_cls_dn_2": 0.08905, "loss_box_dn_2": 0.60054, "loss_cls_dn_3": 0.08763, "loss_box_dn_3": 0.60044, "loss_cls_dn_4": 0.0878, "loss_box_dn_4": 0.60691, "loss_cls_dn_5": 0.08947, "loss_box_dn_5": 0.61732, "loss_dense_depth": 0.70198, "loss": 23.12342, "grad_norm": 29.74716, "time": 1.59766} -{"mode": "train", "epoch": 1, "iter": 452, "lr": 0.00028, "memory": 49164, "data_time": 0.07391, "loss_cls_0": 0.71489, "loss_box_0": 1.57539, "loss_cns_0": 0.63646, "loss_yns_0": 0.13801, "loss_cls_1": 0.76356, "loss_box_1": 1.45562, "loss_cns_1": 0.65981, "loss_yns_1": 0.13534, "loss_cls_2": 0.77118, "loss_box_2": 1.42183, "loss_cns_2": 0.66033, "loss_yns_2": 0.13383, "loss_cls_3": 0.78439, "loss_box_3": 1.41418, "loss_cns_3": 0.65985, "loss_yns_3": 0.13436, "loss_cls_4": 0.78844, "loss_box_4": 1.40196, "loss_cns_4": 0.65385, "loss_yns_4": 0.13315, "loss_cls_5": 0.78626, "loss_box_5": 1.42268, "loss_cns_5": 0.66186, "loss_yns_5": 0.13511, "loss_cls_dn_0": 0.10374, "loss_box_dn_0": 0.70237, "loss_cls_dn_1": 0.08869, "loss_box_dn_1": 0.61917, "loss_cls_dn_2": 0.08707, "loss_box_dn_2": 0.60741, "loss_cls_dn_3": 0.08819, "loss_box_dn_3": 0.61006, "loss_cls_dn_4": 0.0892, "loss_box_dn_4": 0.61966, "loss_cls_dn_5": 0.09015, "loss_box_dn_5": 0.6238, "loss_dense_depth": 0.68691, "loss": 23.05876, "grad_norm": 34.67639, "time": 1.56736} -{"mode": "train", "epoch": 1, "iter": 453, "lr": 0.00028, "memory": 49164, "data_time": 0.07518, "loss_cls_0": 0.69993, "loss_box_0": 1.54281, "loss_cns_0": 0.63761, "loss_yns_0": 0.13851, "loss_cls_1": 0.7673, "loss_box_1": 1.42656, "loss_cns_1": 0.65323, "loss_yns_1": 0.13602, "loss_cls_2": 0.77394, "loss_box_2": 1.38053, "loss_cns_2": 0.6557, "loss_yns_2": 0.13595, "loss_cls_3": 0.78741, "loss_box_3": 1.35683, "loss_cns_3": 0.65191, "loss_yns_3": 0.13552, "loss_cls_4": 0.78884, "loss_box_4": 1.3149, "loss_cns_4": 0.64495, "loss_yns_4": 0.13296, "loss_cls_5": 0.77818, "loss_box_5": 1.37334, "loss_cns_5": 0.65511, "loss_yns_5": 0.13731, "loss_cls_dn_0": 0.10233, "loss_box_dn_0": 0.70651, "loss_cls_dn_1": 0.08764, "loss_box_dn_1": 0.64404, "loss_cls_dn_2": 0.08797, "loss_box_dn_2": 0.62166, "loss_cls_dn_3": 0.0879, "loss_box_dn_3": 0.61905, "loss_cls_dn_4": 0.08917, "loss_box_dn_4": 0.61876, "loss_cls_dn_5": 0.08923, "loss_box_dn_5": 0.62681, "loss_dense_depth": 0.67624, "loss": 22.76264, "grad_norm": 32.85595, "time": 1.57958} -{"mode": "train", "epoch": 1, "iter": 454, "lr": 0.00028, "memory": 49164, "data_time": 0.09459, "loss_cls_0": 0.72346, "loss_box_0": 1.58459, "loss_cns_0": 0.63675, "loss_yns_0": 0.13938, "loss_cls_1": 0.76502, "loss_box_1": 1.42933, "loss_cns_1": 0.65525, "loss_yns_1": 0.1365, "loss_cls_2": 0.78023, "loss_box_2": 1.39186, "loss_cns_2": 0.65783, "loss_yns_2": 0.13578, "loss_cls_3": 0.78994, "loss_box_3": 1.3866, "loss_cns_3": 0.65709, "loss_yns_3": 0.13463, "loss_cls_4": 0.79336, "loss_box_4": 1.36979, "loss_cns_4": 0.6522, "loss_yns_4": 0.1332, "loss_cls_5": 0.79584, "loss_box_5": 1.40098, "loss_cns_5": 0.65861, "loss_yns_5": 0.13597, "loss_cls_dn_0": 0.10933, "loss_box_dn_0": 0.71044, "loss_cls_dn_1": 0.09307, "loss_box_dn_1": 0.62803, "loss_cls_dn_2": 0.09149, "loss_box_dn_2": 0.60614, "loss_cls_dn_3": 0.09091, "loss_box_dn_3": 0.60486, "loss_cls_dn_4": 0.09323, "loss_box_dn_4": 0.60101, "loss_cls_dn_5": 0.09483, "loss_box_dn_5": 0.6062, "loss_dense_depth": 0.69295, "loss": 22.96666, "grad_norm": 26.50565, "time": 1.65109} -{"mode": "train", "epoch": 1, "iter": 455, "lr": 0.00028, "memory": 49164, "data_time": 0.08683, "loss_cls_0": 0.71719, "loss_box_0": 1.5673, "loss_cns_0": 0.64491, "loss_yns_0": 0.13947, "loss_cls_1": 0.76436, "loss_box_1": 1.42211, "loss_cns_1": 0.66658, "loss_yns_1": 0.13771, "loss_cls_2": 0.77612, "loss_box_2": 1.39049, "loss_cns_2": 0.67084, "loss_yns_2": 0.13714, "loss_cls_3": 0.79242, "loss_box_3": 1.39236, "loss_cns_3": 0.66899, "loss_yns_3": 0.13716, "loss_cls_4": 0.79247, "loss_box_4": 1.3909, "loss_cns_4": 0.6684, "loss_yns_4": 0.13682, "loss_cls_5": 0.804, "loss_box_5": 1.39832, "loss_cns_5": 0.66748, "loss_yns_5": 0.13784, "loss_cls_dn_0": 0.10576, "loss_box_dn_0": 0.69893, "loss_cls_dn_1": 0.09195, "loss_box_dn_1": 0.61032, "loss_cls_dn_2": 0.08954, "loss_box_dn_2": 0.59266, "loss_cls_dn_3": 0.09209, "loss_box_dn_3": 0.59616, "loss_cls_dn_4": 0.09513, "loss_box_dn_4": 0.5979, "loss_cls_dn_5": 0.09638, "loss_box_dn_5": 0.60107, "loss_dense_depth": 0.68233, "loss": 22.97161, "grad_norm": 32.61551, "time": 1.57724} -{"mode": "train", "epoch": 1, "iter": 456, "lr": 0.00028, "memory": 49164, "data_time": 0.08153, "loss_cls_0": 0.72854, "loss_box_0": 1.545, "loss_cns_0": 0.64315, "loss_yns_0": 0.14168, "loss_cls_1": 0.76099, "loss_box_1": 1.43452, "loss_cns_1": 0.66345, "loss_yns_1": 0.13874, "loss_cls_2": 0.77487, "loss_box_2": 1.40561, "loss_cns_2": 0.663, "loss_yns_2": 0.13857, "loss_cls_3": 0.78906, "loss_box_3": 1.40158, "loss_cns_3": 0.66097, "loss_yns_3": 0.14113, "loss_cls_4": 0.79723, "loss_box_4": 1.40389, "loss_cns_4": 0.6623, "loss_yns_4": 0.1419, "loss_cls_5": 0.79282, "loss_box_5": 1.40022, "loss_cns_5": 0.66335, "loss_yns_5": 0.13912, "loss_cls_dn_0": 0.10771, "loss_box_dn_0": 0.70207, "loss_cls_dn_1": 0.08712, "loss_box_dn_1": 0.62131, "loss_cls_dn_2": 0.08678, "loss_box_dn_2": 0.60555, "loss_cls_dn_3": 0.09178, "loss_box_dn_3": 0.60878, "loss_cls_dn_4": 0.09751, "loss_box_dn_4": 0.61356, "loss_cls_dn_5": 0.09361, "loss_box_dn_5": 0.61825, "loss_dense_depth": 0.67483, "loss": 23.04053, "grad_norm": 32.86898, "time": 1.56637} -{"mode": "train", "epoch": 1, "iter": 457, "lr": 0.00028, "memory": 49164, "data_time": 0.07918, "loss_cls_0": 0.74496, "loss_box_0": 1.54488, "loss_cns_0": 0.63553, "loss_yns_0": 0.13993, "loss_cls_1": 0.782, "loss_box_1": 1.46181, "loss_cns_1": 0.66723, "loss_yns_1": 0.1398, "loss_cls_2": 0.79215, "loss_box_2": 1.42239, "loss_cns_2": 0.66226, "loss_yns_2": 0.13953, "loss_cls_3": 0.79253, "loss_box_3": 1.42107, "loss_cns_3": 0.66461, "loss_yns_3": 0.1412, "loss_cls_4": 0.80717, "loss_box_4": 1.41947, "loss_cns_4": 0.66516, "loss_yns_4": 0.14228, "loss_cls_5": 0.81224, "loss_box_5": 1.41533, "loss_cns_5": 0.66541, "loss_yns_5": 0.14026, "loss_cls_dn_0": 0.11597, "loss_box_dn_0": 0.71372, "loss_cls_dn_1": 0.09161, "loss_box_dn_1": 0.62419, "loss_cls_dn_2": 0.09065, "loss_box_dn_2": 0.6036, "loss_cls_dn_3": 0.09209, "loss_box_dn_3": 0.60327, "loss_cls_dn_4": 0.09353, "loss_box_dn_4": 0.60259, "loss_cls_dn_5": 0.09274, "loss_box_dn_5": 0.60614, "loss_dense_depth": 0.70977, "loss": 23.25907, "grad_norm": 28.96177, "time": 1.61874} -{"mode": "train", "epoch": 1, "iter": 458, "lr": 0.00028, "memory": 49164, "data_time": 0.08008, "loss_cls_0": 0.71485, "loss_box_0": 1.57069, "loss_cns_0": 0.63369, "loss_yns_0": 0.1428, "loss_cls_1": 0.77727, "loss_box_1": 1.46187, "loss_cns_1": 0.66604, "loss_yns_1": 0.14523, "loss_cls_2": 0.78177, "loss_box_2": 1.41878, "loss_cns_2": 0.66222, "loss_yns_2": 0.14364, "loss_cls_3": 0.79569, "loss_box_3": 1.40336, "loss_cns_3": 0.66322, "loss_yns_3": 0.14298, "loss_cls_4": 0.81484, "loss_box_4": 1.42492, "loss_cns_4": 0.66639, "loss_yns_4": 0.14301, "loss_cls_5": 0.81085, "loss_box_5": 1.41791, "loss_cns_5": 0.66642, "loss_yns_5": 0.14402, "loss_cls_dn_0": 0.10994, "loss_box_dn_0": 0.7081, "loss_cls_dn_1": 0.0967, "loss_box_dn_1": 0.63344, "loss_cls_dn_2": 0.09291, "loss_box_dn_2": 0.61219, "loss_cls_dn_3": 0.09343, "loss_box_dn_3": 0.60972, "loss_cls_dn_4": 0.0952, "loss_box_dn_4": 0.61336, "loss_cls_dn_5": 0.0969, "loss_box_dn_5": 0.61794, "loss_dense_depth": 0.69934, "loss": 23.29161, "grad_norm": 30.57393, "time": 1.5759} -{"mode": "train", "epoch": 1, "iter": 459, "lr": 0.00028, "memory": 49164, "data_time": 0.20938, "loss_cls_0": 0.71815, "loss_box_0": 1.56251, "loss_cns_0": 0.63984, "loss_yns_0": 0.14071, "loss_cls_1": 0.76854, "loss_box_1": 1.43038, "loss_cns_1": 0.66197, "loss_yns_1": 0.13942, "loss_cls_2": 0.77508, "loss_box_2": 1.39671, "loss_cns_2": 0.66244, "loss_yns_2": 0.13779, "loss_cls_3": 0.79084, "loss_box_3": 1.37476, "loss_cns_3": 0.6576, "loss_yns_3": 0.13832, "loss_cls_4": 0.78964, "loss_box_4": 1.39106, "loss_cns_4": 0.66099, "loss_yns_4": 0.13907, "loss_cls_5": 0.80326, "loss_box_5": 1.37848, "loss_cns_5": 0.66218, "loss_yns_5": 0.13895, "loss_cls_dn_0": 0.10621, "loss_box_dn_0": 0.70641, "loss_cls_dn_1": 0.09722, "loss_box_dn_1": 0.62976, "loss_cls_dn_2": 0.09484, "loss_box_dn_2": 0.60637, "loss_cls_dn_3": 0.09678, "loss_box_dn_3": 0.60711, "loss_cls_dn_4": 0.09714, "loss_box_dn_4": 0.61112, "loss_cls_dn_5": 0.09858, "loss_box_dn_5": 0.60673, "loss_dense_depth": 0.71322, "loss": 23.03018, "grad_norm": 31.87708, "time": 1.6951} -{"mode": "train", "epoch": 1, "iter": 460, "lr": 0.00028, "memory": 49164, "data_time": 0.07627, "loss_cls_0": 0.7654, "loss_box_0": 1.566, "loss_cns_0": 0.63935, "loss_yns_0": 0.14433, "loss_cls_1": 0.7834, "loss_box_1": 1.39006, "loss_cns_1": 0.66049, "loss_yns_1": 0.139, "loss_cls_2": 0.78736, "loss_box_2": 1.35303, "loss_cns_2": 0.66438, "loss_yns_2": 0.13706, "loss_cls_3": 0.79729, "loss_box_3": 1.3346, "loss_cns_3": 0.65993, "loss_yns_3": 0.14237, "loss_cls_4": 0.8077, "loss_box_4": 1.33945, "loss_cns_4": 0.66278, "loss_yns_4": 0.14202, "loss_cls_5": 0.82038, "loss_box_5": 1.34265, "loss_cns_5": 0.66311, "loss_yns_5": 0.1391, "loss_cls_dn_0": 0.11426, "loss_box_dn_0": 0.70972, "loss_cls_dn_1": 0.09747, "loss_box_dn_1": 0.63367, "loss_cls_dn_2": 0.09499, "loss_box_dn_2": 0.60518, "loss_cls_dn_3": 0.09574, "loss_box_dn_3": 0.59825, "loss_cls_dn_4": 0.09568, "loss_box_dn_4": 0.59749, "loss_cls_dn_5": 0.09671, "loss_box_dn_5": 0.60115, "loss_dense_depth": 0.69571, "loss": 22.91727, "grad_norm": 23.79872, "time": 1.63083} -{"mode": "train", "epoch": 1, "iter": 461, "lr": 0.00028, "memory": 49164, "data_time": 0.07686, "loss_cls_0": 0.73803, "loss_box_0": 1.57457, "loss_cns_0": 0.6347, "loss_yns_0": 0.14161, "loss_cls_1": 0.7884, "loss_box_1": 1.40527, "loss_cns_1": 0.66092, "loss_yns_1": 0.13863, "loss_cls_2": 0.79251, "loss_box_2": 1.38469, "loss_cns_2": 0.66277, "loss_yns_2": 0.13821, "loss_cls_3": 0.80691, "loss_box_3": 1.37354, "loss_cns_3": 0.66302, "loss_yns_3": 0.13936, "loss_cls_4": 0.80699, "loss_box_4": 1.37042, "loss_cns_4": 0.66221, "loss_yns_4": 0.13799, "loss_cls_5": 0.81194, "loss_box_5": 1.37898, "loss_cns_5": 0.66314, "loss_yns_5": 0.13827, "loss_cls_dn_0": 0.1088, "loss_box_dn_0": 0.70607, "loss_cls_dn_1": 0.09369, "loss_box_dn_1": 0.61006, "loss_cls_dn_2": 0.09261, "loss_box_dn_2": 0.59332, "loss_cls_dn_3": 0.0922, "loss_box_dn_3": 0.5916, "loss_cls_dn_4": 0.09445, "loss_box_dn_4": 0.59543, "loss_cls_dn_5": 0.09402, "loss_box_dn_5": 0.6066, "loss_dense_depth": 0.68938, "loss": 22.98129, "grad_norm": 35.48255, "time": 1.66667} -{"mode": "train", "epoch": 1, "iter": 462, "lr": 0.00028, "memory": 49164, "data_time": 0.08749, "loss_cls_0": 0.7459, "loss_box_0": 1.56023, "loss_cns_0": 0.63765, "loss_yns_0": 0.14216, "loss_cls_1": 0.79077, "loss_box_1": 1.4053, "loss_cns_1": 0.66315, "loss_yns_1": 0.13913, "loss_cls_2": 0.79436, "loss_box_2": 1.38603, "loss_cns_2": 0.66135, "loss_yns_2": 0.1403, "loss_cls_3": 0.80731, "loss_box_3": 1.37838, "loss_cns_3": 0.66411, "loss_yns_3": 0.13945, "loss_cls_4": 0.83029, "loss_box_4": 1.37435, "loss_cns_4": 0.66744, "loss_yns_4": 0.13997, "loss_cls_5": 0.81793, "loss_box_5": 1.39214, "loss_cns_5": 0.6673, "loss_yns_5": 0.13872, "loss_cls_dn_0": 0.11064, "loss_box_dn_0": 0.71287, "loss_cls_dn_1": 0.09439, "loss_box_dn_1": 0.63233, "loss_cls_dn_2": 0.09476, "loss_box_dn_2": 0.61212, "loss_cls_dn_3": 0.0948, "loss_box_dn_3": 0.61515, "loss_cls_dn_4": 0.09749, "loss_box_dn_4": 0.61535, "loss_cls_dn_5": 0.09528, "loss_box_dn_5": 0.62499, "loss_dense_depth": 0.68923, "loss": 23.17313, "grad_norm": 41.89635, "time": 1.62424} -{"mode": "train", "epoch": 1, "iter": 463, "lr": 0.00028, "memory": 49164, "data_time": 0.07184, "loss_cls_0": 0.73485, "loss_box_0": 1.52654, "loss_cns_0": 0.6485, "loss_yns_0": 0.14087, "loss_cls_1": 0.77426, "loss_box_1": 1.40185, "loss_cns_1": 0.66856, "loss_yns_1": 0.13867, "loss_cls_2": 0.77814, "loss_box_2": 1.37912, "loss_cns_2": 0.66706, "loss_yns_2": 0.14148, "loss_cls_3": 0.7961, "loss_box_3": 1.35378, "loss_cns_3": 0.66907, "loss_yns_3": 0.14028, "loss_cls_4": 0.79916, "loss_box_4": 1.35138, "loss_cns_4": 0.67072, "loss_yns_4": 0.14144, "loss_cls_5": 0.81524, "loss_box_5": 1.37353, "loss_cns_5": 0.67113, "loss_yns_5": 0.13995, "loss_cls_dn_0": 0.10684, "loss_box_dn_0": 0.71112, "loss_cls_dn_1": 0.09285, "loss_box_dn_1": 0.6248, "loss_cls_dn_2": 0.09295, "loss_box_dn_2": 0.5986, "loss_cls_dn_3": 0.0924, "loss_box_dn_3": 0.59069, "loss_cls_dn_4": 0.09292, "loss_box_dn_4": 0.58577, "loss_cls_dn_5": 0.09204, "loss_box_dn_5": 0.59638, "loss_dense_depth": 0.66384, "loss": 22.8629, "grad_norm": 32.85303, "time": 1.57111} -{"mode": "train", "epoch": 1, "iter": 464, "lr": 0.00028, "memory": 49164, "data_time": 0.07324, "loss_cls_0": 0.72147, "loss_box_0": 1.54031, "loss_cns_0": 0.64752, "loss_yns_0": 0.1416, "loss_cls_1": 0.77111, "loss_box_1": 1.38689, "loss_cns_1": 0.66702, "loss_yns_1": 0.13761, "loss_cls_2": 0.7729, "loss_box_2": 1.36165, "loss_cns_2": 0.66559, "loss_yns_2": 0.14131, "loss_cls_3": 0.78029, "loss_box_3": 1.35289, "loss_cns_3": 0.66615, "loss_yns_3": 0.14154, "loss_cls_4": 0.7873, "loss_box_4": 1.35955, "loss_cns_4": 0.66798, "loss_yns_4": 0.14551, "loss_cls_5": 0.79691, "loss_box_5": 1.35057, "loss_cns_5": 0.66779, "loss_yns_5": 0.14081, "loss_cls_dn_0": 0.10624, "loss_box_dn_0": 0.71586, "loss_cls_dn_1": 0.09089, "loss_box_dn_1": 0.62319, "loss_cls_dn_2": 0.08983, "loss_box_dn_2": 0.59376, "loss_cls_dn_3": 0.08813, "loss_box_dn_3": 0.59032, "loss_cls_dn_4": 0.08841, "loss_box_dn_4": 0.59385, "loss_cls_dn_5": 0.08839, "loss_box_dn_5": 0.59328, "loss_dense_depth": 0.66894, "loss": 22.74336, "grad_norm": 32.16875, "time": 1.61922} -{"mode": "train", "epoch": 1, "iter": 465, "lr": 0.00029, "memory": 49164, "data_time": 0.07243, "loss_cls_0": 0.72169, "loss_box_0": 1.56748, "loss_cns_0": 0.64945, "loss_yns_0": 0.14072, "loss_cls_1": 0.77432, "loss_box_1": 1.38041, "loss_cns_1": 0.6685, "loss_yns_1": 0.13481, "loss_cls_2": 0.78181, "loss_box_2": 1.36179, "loss_cns_2": 0.66478, "loss_yns_2": 0.13727, "loss_cls_3": 0.78493, "loss_box_3": 1.36217, "loss_cns_3": 0.66827, "loss_yns_3": 0.13886, "loss_cls_4": 0.79363, "loss_box_4": 1.37154, "loss_cns_4": 0.66835, "loss_yns_4": 0.13717, "loss_cls_5": 0.80798, "loss_box_5": 1.35674, "loss_cns_5": 0.66869, "loss_yns_5": 0.1362, "loss_cls_dn_0": 0.10782, "loss_box_dn_0": 0.70561, "loss_cls_dn_1": 0.09179, "loss_box_dn_1": 0.60542, "loss_cls_dn_2": 0.09128, "loss_box_dn_2": 0.5875, "loss_cls_dn_3": 0.09102, "loss_box_dn_3": 0.58721, "loss_cls_dn_4": 0.09028, "loss_box_dn_4": 0.59502, "loss_cls_dn_5": 0.09294, "loss_box_dn_5": 0.59426, "loss_dense_depth": 0.6782, "loss": 22.79593, "grad_norm": 33.27055, "time": 1.63848} -{"mode": "train", "epoch": 1, "iter": 466, "lr": 0.00029, "memory": 49164, "data_time": 0.07276, "loss_cls_0": 0.72785, "loss_box_0": 1.54805, "loss_cns_0": 0.64349, "loss_yns_0": 0.14043, "loss_cls_1": 0.78487, "loss_box_1": 1.39565, "loss_cns_1": 0.66787, "loss_yns_1": 0.1378, "loss_cls_2": 0.79123, "loss_box_2": 1.38315, "loss_cns_2": 0.66529, "loss_yns_2": 0.13833, "loss_cls_3": 0.79442, "loss_box_3": 1.3665, "loss_cns_3": 0.66659, "loss_yns_3": 0.1376, "loss_cls_4": 0.80828, "loss_box_4": 1.38489, "loss_cns_4": 0.66793, "loss_yns_4": 0.13776, "loss_cls_5": 0.82163, "loss_box_5": 1.36807, "loss_cns_5": 0.66622, "loss_yns_5": 0.13655, "loss_cls_dn_0": 0.11313, "loss_box_dn_0": 0.71789, "loss_cls_dn_1": 0.09162, "loss_box_dn_1": 0.60364, "loss_cls_dn_2": 0.09265, "loss_box_dn_2": 0.59403, "loss_cls_dn_3": 0.09249, "loss_box_dn_3": 0.58849, "loss_cls_dn_4": 0.09083, "loss_box_dn_4": 0.59652, "loss_cls_dn_5": 0.09473, "loss_box_dn_5": 0.59628, "loss_dense_depth": 0.68823, "loss": 22.94099, "grad_norm": 40.49921, "time": 1.64774} -{"mode": "train", "epoch": 1, "iter": 467, "lr": 0.00029, "memory": 49164, "data_time": 0.07322, "loss_cls_0": 0.7224, "loss_box_0": 1.54272, "loss_cns_0": 0.64429, "loss_yns_0": 0.13872, "loss_cls_1": 0.78036, "loss_box_1": 1.41619, "loss_cns_1": 0.66787, "loss_yns_1": 0.1357, "loss_cls_2": 0.79116, "loss_box_2": 1.39285, "loss_cns_2": 0.66684, "loss_yns_2": 0.13892, "loss_cls_3": 0.79955, "loss_box_3": 1.38802, "loss_cns_3": 0.66818, "loss_yns_3": 0.13687, "loss_cls_4": 0.81695, "loss_box_4": 1.389, "loss_cns_4": 0.6712, "loss_yns_4": 0.13891, "loss_cls_5": 0.8148, "loss_box_5": 1.3746, "loss_cns_5": 0.66669, "loss_yns_5": 0.13611, "loss_cls_dn_0": 0.10949, "loss_box_dn_0": 0.70929, "loss_cls_dn_1": 0.09265, "loss_box_dn_1": 0.61824, "loss_cls_dn_2": 0.09184, "loss_box_dn_2": 0.60315, "loss_cls_dn_3": 0.09038, "loss_box_dn_3": 0.60141, "loss_cls_dn_4": 0.09205, "loss_box_dn_4": 0.60265, "loss_cls_dn_5": 0.09196, "loss_box_dn_5": 0.60343, "loss_dense_depth": 0.67866, "loss": 23.02411, "grad_norm": 35.25317, "time": 1.55498} -{"mode": "train", "epoch": 1, "iter": 468, "lr": 0.00029, "memory": 49164, "data_time": 0.07441, "loss_cls_0": 0.73191, "loss_box_0": 1.54997, "loss_cns_0": 0.63812, "loss_yns_0": 0.13898, "loss_cls_1": 0.80081, "loss_box_1": 1.44519, "loss_cns_1": 0.66544, "loss_yns_1": 0.13554, "loss_cls_2": 0.80294, "loss_box_2": 1.42937, "loss_cns_2": 0.66486, "loss_yns_2": 0.13999, "loss_cls_3": 0.8182, "loss_box_3": 1.4322, "loss_cns_3": 0.66683, "loss_yns_3": 0.13678, "loss_cls_4": 0.81785, "loss_box_4": 1.42037, "loss_cns_4": 0.67059, "loss_yns_4": 0.1395, "loss_cls_5": 0.82588, "loss_box_5": 1.43, "loss_cns_5": 0.66858, "loss_yns_5": 0.13532, "loss_cls_dn_0": 0.11005, "loss_box_dn_0": 0.70781, "loss_cls_dn_1": 0.09442, "loss_box_dn_1": 0.61899, "loss_cls_dn_2": 0.0935, "loss_box_dn_2": 0.60383, "loss_cls_dn_3": 0.09323, "loss_box_dn_3": 0.60415, "loss_cls_dn_4": 0.09384, "loss_box_dn_4": 0.60286, "loss_cls_dn_5": 0.09303, "loss_box_dn_5": 0.61343, "loss_dense_depth": 0.67804, "loss": 23.31237, "grad_norm": 34.75251, "time": 6.93729} -{"mode": "train", "epoch": 1, "iter": 469, "lr": 0.00029, "memory": 49164, "data_time": 0.06951, "loss_cls_0": 0.74236, "loss_box_0": 1.55944, "loss_cns_0": 0.63539, "loss_yns_0": 0.1427, "loss_cls_1": 0.81523, "loss_box_1": 1.42157, "loss_cns_1": 0.66622, "loss_yns_1": 0.13918, "loss_cls_2": 0.82039, "loss_box_2": 1.38859, "loss_cns_2": 0.66529, "loss_yns_2": 0.14145, "loss_cls_3": 0.82334, "loss_box_3": 1.38468, "loss_cns_3": 0.6672, "loss_yns_3": 0.14059, "loss_cls_4": 0.83672, "loss_box_4": 1.37915, "loss_cns_4": 0.66692, "loss_yns_4": 0.14153, "loss_cls_5": 0.84951, "loss_box_5": 1.39248, "loss_cns_5": 0.66478, "loss_yns_5": 0.14024, "loss_cls_dn_0": 0.11214, "loss_box_dn_0": 0.71456, "loss_cls_dn_1": 0.0951, "loss_box_dn_1": 0.62348, "loss_cls_dn_2": 0.09447, "loss_box_dn_2": 0.6058, "loss_cls_dn_3": 0.09309, "loss_box_dn_3": 0.60616, "loss_cls_dn_4": 0.09316, "loss_box_dn_4": 0.61058, "loss_cls_dn_5": 0.09453, "loss_box_dn_5": 0.62016, "loss_dense_depth": 0.68848, "loss": 23.27665, "grad_norm": 42.89842, "time": 1.53322} -{"mode": "train", "epoch": 1, "iter": 470, "lr": 0.00029, "memory": 49164, "data_time": 0.0706, "loss_cls_0": 0.73983, "loss_box_0": 1.56636, "loss_cns_0": 0.63926, "loss_yns_0": 0.14364, "loss_cls_1": 0.81641, "loss_box_1": 1.40859, "loss_cns_1": 0.66318, "loss_yns_1": 0.14108, "loss_cls_2": 0.82835, "loss_box_2": 1.39007, "loss_cns_2": 0.66074, "loss_yns_2": 0.14055, "loss_cls_3": 0.83396, "loss_box_3": 1.38622, "loss_cns_3": 0.66175, "loss_yns_3": 0.14106, "loss_cls_4": 0.84138, "loss_box_4": 1.37147, "loss_cns_4": 0.65977, "loss_yns_4": 0.13988, "loss_cls_5": 0.84612, "loss_box_5": 1.37516, "loss_cns_5": 0.6623, "loss_yns_5": 0.1402, "loss_cls_dn_0": 0.11384, "loss_box_dn_0": 0.70829, "loss_cls_dn_1": 0.09366, "loss_box_dn_1": 0.60986, "loss_cls_dn_2": 0.09179, "loss_box_dn_2": 0.59202, "loss_cls_dn_3": 0.09176, "loss_box_dn_3": 0.59015, "loss_cls_dn_4": 0.09198, "loss_box_dn_4": 0.58885, "loss_cls_dn_5": 0.09209, "loss_box_dn_5": 0.58942, "loss_dense_depth": 0.67353, "loss": 23.12456, "grad_norm": 30.02024, "time": 1.53396} -{"mode": "train", "epoch": 1, "iter": 471, "lr": 0.00029, "memory": 49164, "data_time": 0.07255, "loss_cls_0": 0.7443, "loss_box_0": 1.58341, "loss_cns_0": 0.63558, "loss_yns_0": 0.14039, "loss_cls_1": 0.80092, "loss_box_1": 1.42416, "loss_cns_1": 0.66418, "loss_yns_1": 0.13955, "loss_cls_2": 0.81515, "loss_box_2": 1.41863, "loss_cns_2": 0.66035, "loss_yns_2": 0.14049, "loss_cls_3": 0.81721, "loss_box_3": 1.39911, "loss_cns_3": 0.65992, "loss_yns_3": 0.13997, "loss_cls_4": 0.81397, "loss_box_4": 1.38726, "loss_cns_4": 0.66199, "loss_yns_4": 0.13945, "loss_cls_5": 0.81817, "loss_box_5": 1.39331, "loss_cns_5": 0.66553, "loss_yns_5": 0.13874, "loss_cls_dn_0": 0.11355, "loss_box_dn_0": 0.71676, "loss_cls_dn_1": 0.09199, "loss_box_dn_1": 0.60759, "loss_cls_dn_2": 0.09022, "loss_box_dn_2": 0.5956, "loss_cls_dn_3": 0.08988, "loss_box_dn_3": 0.58915, "loss_cls_dn_4": 0.09126, "loss_box_dn_4": 0.58423, "loss_cls_dn_5": 0.0915, "loss_box_dn_5": 0.59072, "loss_dense_depth": 0.73664, "loss": 23.1908, "grad_norm": 43.10093, "time": 1.5436} -{"mode": "train", "epoch": 1, "iter": 472, "lr": 0.00029, "memory": 49164, "data_time": 0.07593, "loss_cls_0": 0.7138, "loss_box_0": 1.51667, "loss_cns_0": 0.64268, "loss_yns_0": 0.14202, "loss_cls_1": 0.79196, "loss_box_1": 1.36992, "loss_cns_1": 0.67082, "loss_yns_1": 0.14155, "loss_cls_2": 0.80006, "loss_box_2": 1.34102, "loss_cns_2": 0.66414, "loss_yns_2": 0.14261, "loss_cls_3": 0.80178, "loss_box_3": 1.35434, "loss_cns_3": 0.66786, "loss_yns_3": 0.1414, "loss_cls_4": 0.80447, "loss_box_4": 1.34194, "loss_cns_4": 0.67234, "loss_yns_4": 0.14237, "loss_cls_5": 0.81594, "loss_box_5": 1.35299, "loss_cns_5": 0.66948, "loss_yns_5": 0.14479, "loss_cls_dn_0": 0.1088, "loss_box_dn_0": 0.69574, "loss_cls_dn_1": 0.09045, "loss_box_dn_1": 0.60947, "loss_cls_dn_2": 0.08886, "loss_box_dn_2": 0.59987, "loss_cls_dn_3": 0.08892, "loss_box_dn_3": 0.60678, "loss_cls_dn_4": 0.09177, "loss_box_dn_4": 0.60519, "loss_cls_dn_5": 0.09081, "loss_box_dn_5": 0.61625, "loss_dense_depth": 0.7036, "loss": 22.84347, "grad_norm": 40.68065, "time": 1.68065} -{"mode": "train", "epoch": 1, "iter": 473, "lr": 0.00029, "memory": 49164, "data_time": 0.06931, "loss_cls_0": 0.71345, "loss_box_0": 1.52125, "loss_cns_0": 0.63932, "loss_yns_0": 0.14254, "loss_cls_1": 0.78617, "loss_box_1": 1.38669, "loss_cns_1": 0.66953, "loss_yns_1": 0.1396, "loss_cls_2": 0.7928, "loss_box_2": 1.35105, "loss_cns_2": 0.66523, "loss_yns_2": 0.14058, "loss_cls_3": 0.78251, "loss_box_3": 1.36589, "loss_cns_3": 0.66747, "loss_yns_3": 0.13932, "loss_cls_4": 0.78967, "loss_box_4": 1.35096, "loss_cns_4": 0.66918, "loss_yns_4": 0.13843, "loss_cls_5": 0.78798, "loss_box_5": 1.34722, "loss_cns_5": 0.66721, "loss_yns_5": 0.13923, "loss_cls_dn_0": 0.10878, "loss_box_dn_0": 0.70191, "loss_cls_dn_1": 0.0931, "loss_box_dn_1": 0.62627, "loss_cls_dn_2": 0.09037, "loss_box_dn_2": 0.61446, "loss_cls_dn_3": 0.09028, "loss_box_dn_3": 0.62022, "loss_cls_dn_4": 0.09081, "loss_box_dn_4": 0.61923, "loss_cls_dn_5": 0.09, "loss_box_dn_5": 0.61943, "loss_dense_depth": 0.75951, "loss": 22.91766, "grad_norm": 42.19437, "time": 1.5534} -{"mode": "train", "epoch": 1, "iter": 474, "lr": 0.00029, "memory": 49164, "data_time": 0.0746, "loss_cls_0": 0.71718, "loss_box_0": 1.54056, "loss_cns_0": 0.63511, "loss_yns_0": 0.14094, "loss_cls_1": 0.78095, "loss_box_1": 1.38753, "loss_cns_1": 0.6573, "loss_yns_1": 0.13796, "loss_cls_2": 0.78252, "loss_box_2": 1.3586, "loss_cns_2": 0.65956, "loss_yns_2": 0.14015, "loss_cls_3": 0.77948, "loss_box_3": 1.35813, "loss_cns_3": 0.65855, "loss_yns_3": 0.13944, "loss_cls_4": 0.7866, "loss_box_4": 1.36306, "loss_cns_4": 0.66071, "loss_yns_4": 0.14354, "loss_cls_5": 0.77832, "loss_box_5": 1.33818, "loss_cns_5": 0.66006, "loss_yns_5": 0.13985, "loss_cls_dn_0": 0.10825, "loss_box_dn_0": 0.69707, "loss_cls_dn_1": 0.09391, "loss_box_dn_1": 0.62783, "loss_cls_dn_2": 0.09227, "loss_box_dn_2": 0.61126, "loss_cls_dn_3": 0.09367, "loss_box_dn_3": 0.60899, "loss_cls_dn_4": 0.09518, "loss_box_dn_4": 0.60882, "loss_cls_dn_5": 0.09617, "loss_box_dn_5": 0.59849, "loss_dense_depth": 0.69694, "loss": 22.77316, "grad_norm": 42.67072, "time": 1.60391} -{"mode": "train", "epoch": 1, "iter": 475, "lr": 0.00029, "memory": 49164, "data_time": 0.10349, "loss_cls_0": 0.70113, "loss_box_0": 1.52088, "loss_cns_0": 0.63515, "loss_yns_0": 0.14118, "loss_cls_1": 0.75426, "loss_box_1": 1.3604, "loss_cns_1": 0.65649, "loss_yns_1": 0.13712, "loss_cls_2": 0.76008, "loss_box_2": 1.33312, "loss_cns_2": 0.65766, "loss_yns_2": 0.13669, "loss_cls_3": 0.75891, "loss_box_3": 1.35463, "loss_cns_3": 0.65766, "loss_yns_3": 0.13863, "loss_cls_4": 0.75898, "loss_box_4": 1.33817, "loss_cns_4": 0.65829, "loss_yns_4": 0.14174, "loss_cls_5": 0.7613, "loss_box_5": 1.32512, "loss_cns_5": 0.66168, "loss_yns_5": 0.13734, "loss_cls_dn_0": 0.10499, "loss_box_dn_0": 0.70458, "loss_cls_dn_1": 0.09276, "loss_box_dn_1": 0.60634, "loss_cls_dn_2": 0.09112, "loss_box_dn_2": 0.58619, "loss_cls_dn_3": 0.0902, "loss_box_dn_3": 0.58693, "loss_cls_dn_4": 0.09113, "loss_box_dn_4": 0.57888, "loss_cls_dn_5": 0.09403, "loss_box_dn_5": 0.57674, "loss_dense_depth": 0.69915, "loss": 22.38964, "grad_norm": 40.4125, "time": 1.63601} -{"mode": "train", "epoch": 1, "iter": 476, "lr": 0.00029, "memory": 49164, "data_time": 0.09135, "loss_cls_0": 0.70957, "loss_box_0": 1.52612, "loss_cns_0": 0.63941, "loss_yns_0": 0.14165, "loss_cls_1": 0.74714, "loss_box_1": 1.37942, "loss_cns_1": 0.66472, "loss_yns_1": 0.13782, "loss_cls_2": 0.75939, "loss_box_2": 1.33702, "loss_cns_2": 0.66534, "loss_yns_2": 0.13748, "loss_cls_3": 0.77214, "loss_box_3": 1.35822, "loss_cns_3": 0.66479, "loss_yns_3": 0.13778, "loss_cls_4": 0.77401, "loss_box_4": 1.36366, "loss_cns_4": 0.66695, "loss_yns_4": 0.13785, "loss_cls_5": 0.75249, "loss_box_5": 1.38758, "loss_cns_5": 0.66899, "loss_yns_5": 0.13804, "loss_cls_dn_0": 0.10302, "loss_box_dn_0": 0.71262, "loss_cls_dn_1": 0.08838, "loss_box_dn_1": 0.59835, "loss_cls_dn_2": 0.08596, "loss_box_dn_2": 0.57611, "loss_cls_dn_3": 0.08493, "loss_box_dn_3": 0.58397, "loss_cls_dn_4": 0.08465, "loss_box_dn_4": 0.58943, "loss_cls_dn_5": 0.08926, "loss_box_dn_5": 0.60005, "loss_dense_depth": 0.68521, "loss": 22.54954, "grad_norm": 39.26561, "time": 1.58211} -{"mode": "train", "epoch": 1, "iter": 477, "lr": 0.00029, "memory": 49164, "data_time": 0.08256, "loss_cls_0": 0.7212, "loss_box_0": 1.53759, "loss_cns_0": 0.63917, "loss_yns_0": 0.14104, "loss_cls_1": 0.77768, "loss_box_1": 1.38282, "loss_cns_1": 0.67159, "loss_yns_1": 0.13768, "loss_cls_2": 0.77342, "loss_box_2": 1.38159, "loss_cns_2": 0.66539, "loss_yns_2": 0.13702, "loss_cls_3": 0.78107, "loss_box_3": 1.38552, "loss_cns_3": 0.66403, "loss_yns_3": 0.13683, "loss_cls_4": 0.78243, "loss_box_4": 1.37956, "loss_cns_4": 0.66624, "loss_yns_4": 0.14366, "loss_cls_5": 0.78018, "loss_box_5": 1.38567, "loss_cns_5": 0.66614, "loss_yns_5": 0.13763, "loss_cls_dn_0": 0.10533, "loss_box_dn_0": 0.70724, "loss_cls_dn_1": 0.09138, "loss_box_dn_1": 0.59397, "loss_cls_dn_2": 0.09127, "loss_box_dn_2": 0.58631, "loss_cls_dn_3": 0.09408, "loss_box_dn_3": 0.59057, "loss_cls_dn_4": 0.09392, "loss_box_dn_4": 0.59164, "loss_cls_dn_5": 0.09648, "loss_box_dn_5": 0.59475, "loss_dense_depth": 0.69803, "loss": 22.81014, "grad_norm": 39.12618, "time": 1.57599} -{"mode": "train", "epoch": 1, "iter": 478, "lr": 0.00029, "memory": 49164, "data_time": 0.08176, "loss_cls_0": 0.70264, "loss_box_0": 1.53744, "loss_cns_0": 0.63638, "loss_yns_0": 0.14015, "loss_cls_1": 0.76949, "loss_box_1": 1.39382, "loss_cns_1": 0.66446, "loss_yns_1": 0.13556, "loss_cls_2": 0.76207, "loss_box_2": 1.37734, "loss_cns_2": 0.66415, "loss_yns_2": 0.13767, "loss_cls_3": 0.76332, "loss_box_3": 1.3515, "loss_cns_3": 0.66145, "loss_yns_3": 0.13778, "loss_cls_4": 0.77015, "loss_box_4": 1.35844, "loss_cns_4": 0.66511, "loss_yns_4": 0.14643, "loss_cls_5": 0.77221, "loss_box_5": 1.36065, "loss_cns_5": 0.66396, "loss_yns_5": 0.13807, "loss_cls_dn_0": 0.10488, "loss_box_dn_0": 0.70881, "loss_cls_dn_1": 0.09033, "loss_box_dn_1": 0.60235, "loss_cls_dn_2": 0.08959, "loss_box_dn_2": 0.58459, "loss_cls_dn_3": 0.0902, "loss_box_dn_3": 0.57499, "loss_cls_dn_4": 0.09111, "loss_box_dn_4": 0.57513, "loss_cls_dn_5": 0.0901, "loss_box_dn_5": 0.57708, "loss_dense_depth": 0.67977, "loss": 22.56918, "grad_norm": 28.94848, "time": 1.57204} -{"mode": "train", "epoch": 1, "iter": 479, "lr": 0.00029, "memory": 49164, "data_time": 0.1911, "loss_cls_0": 0.69156, "loss_box_0": 1.5014, "loss_cns_0": 0.63539, "loss_yns_0": 0.13889, "loss_cls_1": 0.76358, "loss_box_1": 1.356, "loss_cns_1": 0.66635, "loss_yns_1": 0.13707, "loss_cls_2": 0.75749, "loss_box_2": 1.35089, "loss_cns_2": 0.66673, "loss_yns_2": 0.13623, "loss_cls_3": 0.76383, "loss_box_3": 1.34561, "loss_cns_3": 0.66397, "loss_yns_3": 0.13611, "loss_cls_4": 0.76903, "loss_box_4": 1.35556, "loss_cns_4": 0.66736, "loss_yns_4": 0.13916, "loss_cls_5": 0.7627, "loss_box_5": 1.37431, "loss_cns_5": 0.66775, "loss_yns_5": 0.13664, "loss_cls_dn_0": 0.09843, "loss_box_dn_0": 0.70641, "loss_cls_dn_1": 0.09013, "loss_box_dn_1": 0.5908, "loss_cls_dn_2": 0.08879, "loss_box_dn_2": 0.57822, "loss_cls_dn_3": 0.08638, "loss_box_dn_3": 0.57534, "loss_cls_dn_4": 0.08885, "loss_box_dn_4": 0.5815, "loss_cls_dn_5": 0.0921, "loss_box_dn_5": 0.59084, "loss_dense_depth": 0.68099, "loss": 22.43238, "grad_norm": 44.82185, "time": 1.67159} -{"mode": "train", "epoch": 1, "iter": 480, "lr": 0.00029, "memory": 49164, "data_time": 0.07503, "loss_cls_0": 0.7329, "loss_box_0": 1.50567, "loss_cns_0": 0.63604, "loss_yns_0": 0.13848, "loss_cls_1": 0.77459, "loss_box_1": 1.37339, "loss_cns_1": 0.66506, "loss_yns_1": 0.13466, "loss_cls_2": 0.76975, "loss_box_2": 1.34144, "loss_cns_2": 0.6628, "loss_yns_2": 0.13267, "loss_cls_3": 0.77162, "loss_box_3": 1.34697, "loss_cns_3": 0.66202, "loss_yns_3": 0.1351, "loss_cls_4": 0.77292, "loss_box_4": 1.36364, "loss_cns_4": 0.66366, "loss_yns_4": 0.13392, "loss_cls_5": 0.77355, "loss_box_5": 1.39151, "loss_cns_5": 0.66581, "loss_yns_5": 0.13483, "loss_cls_dn_0": 0.10367, "loss_box_dn_0": 0.70827, "loss_cls_dn_1": 0.09429, "loss_box_dn_1": 0.59388, "loss_cls_dn_2": 0.09131, "loss_box_dn_2": 0.57567, "loss_cls_dn_3": 0.08813, "loss_box_dn_3": 0.57587, "loss_cls_dn_4": 0.09163, "loss_box_dn_4": 0.58375, "loss_cls_dn_5": 0.09449, "loss_box_dn_5": 0.59459, "loss_dense_depth": 0.6923, "loss": 22.57086, "grad_norm": 36.14632, "time": 1.57623} -{"mode": "train", "epoch": 1, "iter": 481, "lr": 0.00029, "memory": 49164, "data_time": 0.07314, "loss_cls_0": 0.70431, "loss_box_0": 1.46684, "loss_cns_0": 0.64247, "loss_yns_0": 0.13511, "loss_cls_1": 0.74456, "loss_box_1": 1.36315, "loss_cns_1": 0.66637, "loss_yns_1": 0.13138, "loss_cls_2": 0.74645, "loss_box_2": 1.31725, "loss_cns_2": 0.66313, "loss_yns_2": 0.13012, "loss_cls_3": 0.75586, "loss_box_3": 1.32223, "loss_cns_3": 0.66307, "loss_yns_3": 0.13153, "loss_cls_4": 0.7532, "loss_box_4": 1.33206, "loss_cns_4": 0.66382, "loss_yns_4": 0.13608, "loss_cls_5": 0.75635, "loss_box_5": 1.3617, "loss_cns_5": 0.66573, "loss_yns_5": 0.13177, "loss_cls_dn_0": 0.10156, "loss_box_dn_0": 0.70427, "loss_cls_dn_1": 0.09139, "loss_box_dn_1": 0.59404, "loss_cls_dn_2": 0.08775, "loss_box_dn_2": 0.56981, "loss_cls_dn_3": 0.08496, "loss_box_dn_3": 0.57063, "loss_cls_dn_4": 0.08585, "loss_box_dn_4": 0.57661, "loss_cls_dn_5": 0.08814, "loss_box_dn_5": 0.58791, "loss_dense_depth": 0.67127, "loss": 22.19873, "grad_norm": 44.74041, "time": 1.70665} -{"mode": "train", "epoch": 1, "iter": 482, "lr": 0.00029, "memory": 49164, "data_time": 0.08881, "loss_cls_0": 0.7147, "loss_box_0": 1.47864, "loss_cns_0": 0.64534, "loss_yns_0": 0.1369, "loss_cls_1": 0.75124, "loss_box_1": 1.34725, "loss_cns_1": 0.66426, "loss_yns_1": 0.13209, "loss_cls_2": 0.75567, "loss_box_2": 1.32019, "loss_cns_2": 0.66222, "loss_yns_2": 0.12968, "loss_cls_3": 0.76357, "loss_box_3": 1.31897, "loss_cns_3": 0.66169, "loss_yns_3": 0.13017, "loss_cls_4": 0.76725, "loss_box_4": 1.32698, "loss_cns_4": 0.66383, "loss_yns_4": 0.13557, "loss_cls_5": 0.76581, "loss_box_5": 1.32904, "loss_cns_5": 0.66396, "loss_yns_5": 0.13118, "loss_cls_dn_0": 0.10074, "loss_box_dn_0": 0.70011, "loss_cls_dn_1": 0.08739, "loss_box_dn_1": 0.57442, "loss_cls_dn_2": 0.08632, "loss_box_dn_2": 0.55785, "loss_cls_dn_3": 0.08591, "loss_box_dn_3": 0.55833, "loss_cls_dn_4": 0.08606, "loss_box_dn_4": 0.56021, "loss_cls_dn_5": 0.08648, "loss_box_dn_5": 0.56096, "loss_dense_depth": 0.69009, "loss": 22.13107, "grad_norm": 32.86944, "time": 1.64211} -{"mode": "train", "epoch": 1, "iter": 483, "lr": 0.00029, "memory": 49164, "data_time": 0.07364, "loss_cls_0": 0.68889, "loss_box_0": 1.49689, "loss_cns_0": 0.64578, "loss_yns_0": 0.13284, "loss_cls_1": 0.73432, "loss_box_1": 1.40768, "loss_cns_1": 0.66257, "loss_yns_1": 0.12958, "loss_cls_2": 0.74145, "loss_box_2": 1.39611, "loss_cns_2": 0.66259, "loss_yns_2": 0.12988, "loss_cls_3": 0.75053, "loss_box_3": 1.37497, "loss_cns_3": 0.66304, "loss_yns_3": 0.13037, "loss_cls_4": 0.75592, "loss_box_4": 1.37067, "loss_cns_4": 0.66344, "loss_yns_4": 0.13452, "loss_cls_5": 0.75441, "loss_box_5": 1.37411, "loss_cns_5": 0.66402, "loss_yns_5": 0.1304, "loss_cls_dn_0": 0.09954, "loss_box_dn_0": 0.69944, "loss_cls_dn_1": 0.08381, "loss_box_dn_1": 0.57675, "loss_cls_dn_2": 0.08338, "loss_box_dn_2": 0.56095, "loss_cls_dn_3": 0.08307, "loss_box_dn_3": 0.55613, "loss_cls_dn_4": 0.08492, "loss_box_dn_4": 0.5547, "loss_cls_dn_5": 0.08359, "loss_box_dn_5": 0.55567, "loss_dense_depth": 0.6609, "loss": 22.27781, "grad_norm": 45.99876, "time": 1.57} -{"mode": "train", "epoch": 1, "iter": 484, "lr": 0.00029, "memory": 49164, "data_time": 0.11362, "loss_cls_0": 0.69531, "loss_box_0": 1.51006, "loss_cns_0": 0.64316, "loss_yns_0": 0.1356, "loss_cls_1": 0.74719, "loss_box_1": 1.42262, "loss_cns_1": 0.66357, "loss_yns_1": 0.13209, "loss_cls_2": 0.75647, "loss_box_2": 1.38321, "loss_cns_2": 0.6647, "loss_yns_2": 0.1314, "loss_cls_3": 0.7605, "loss_box_3": 1.37054, "loss_cns_3": 0.66314, "loss_yns_3": 0.13085, "loss_cls_4": 0.75533, "loss_box_4": 1.37163, "loss_cns_4": 0.66338, "loss_yns_4": 0.13425, "loss_cls_5": 0.75787, "loss_box_5": 1.37351, "loss_cns_5": 0.66318, "loss_yns_5": 0.13211, "loss_cls_dn_0": 0.09972, "loss_box_dn_0": 0.69722, "loss_cls_dn_1": 0.08974, "loss_box_dn_1": 0.61002, "loss_cls_dn_2": 0.09007, "loss_box_dn_2": 0.58508, "loss_cls_dn_3": 0.0893, "loss_box_dn_3": 0.58243, "loss_cls_dn_4": 0.08933, "loss_box_dn_4": 0.58588, "loss_cls_dn_5": 0.08869, "loss_box_dn_5": 0.58876, "loss_dense_depth": 0.68006, "loss": 22.53795, "grad_norm": 41.88458, "time": 1.66187} -{"mode": "train", "epoch": 1, "iter": 485, "lr": 0.00029, "memory": 49164, "data_time": 0.07493, "loss_cls_0": 0.67201, "loss_box_0": 1.48301, "loss_cns_0": 0.64367, "loss_yns_0": 0.13023, "loss_cls_1": 0.73529, "loss_box_1": 1.37268, "loss_cns_1": 0.66301, "loss_yns_1": 0.12831, "loss_cls_2": 0.73614, "loss_box_2": 1.34485, "loss_cns_2": 0.66291, "loss_yns_2": 0.12759, "loss_cls_3": 0.74697, "loss_box_3": 1.33635, "loss_cns_3": 0.66397, "loss_yns_3": 0.12802, "loss_cls_4": 0.73753, "loss_box_4": 1.33175, "loss_cns_4": 0.6649, "loss_yns_4": 0.12867, "loss_cls_5": 0.73661, "loss_box_5": 1.31961, "loss_cns_5": 0.66164, "loss_yns_5": 0.12884, "loss_cls_dn_0": 0.09916, "loss_box_dn_0": 0.70378, "loss_cls_dn_1": 0.09069, "loss_box_dn_1": 0.61185, "loss_cls_dn_2": 0.09024, "loss_box_dn_2": 0.59268, "loss_cls_dn_3": 0.08934, "loss_box_dn_3": 0.58575, "loss_cls_dn_4": 0.0889, "loss_box_dn_4": 0.58641, "loss_cls_dn_5": 0.08865, "loss_box_dn_5": 0.58403, "loss_dense_depth": 0.63542, "loss": 22.13145, "grad_norm": 47.11884, "time": 1.65562} -{"mode": "train", "epoch": 1, "iter": 486, "lr": 0.00029, "memory": 49164, "data_time": 0.07474, "loss_cls_0": 0.67446, "loss_box_0": 1.46969, "loss_cns_0": 0.64423, "loss_yns_0": 0.12881, "loss_cls_1": 0.72222, "loss_box_1": 1.38168, "loss_cns_1": 0.65322, "loss_yns_1": 0.1258, "loss_cls_2": 0.72312, "loss_box_2": 1.36876, "loss_cns_2": 0.65184, "loss_yns_2": 0.12485, "loss_cls_3": 0.72826, "loss_box_3": 1.34873, "loss_cns_3": 0.6568, "loss_yns_3": 0.12543, "loss_cls_4": 0.73644, "loss_box_4": 1.31328, "loss_cns_4": 0.65299, "loss_yns_4": 0.12356, "loss_cls_5": 0.74568, "loss_box_5": 1.32644, "loss_cns_5": 0.65772, "loss_yns_5": 0.12524, "loss_cls_dn_0": 0.09611, "loss_box_dn_0": 0.70087, "loss_cls_dn_1": 0.08885, "loss_box_dn_1": 0.60275, "loss_cls_dn_2": 0.08746, "loss_box_dn_2": 0.58433, "loss_cls_dn_3": 0.08767, "loss_box_dn_3": 0.57322, "loss_cls_dn_4": 0.08694, "loss_box_dn_4": 0.56484, "loss_cls_dn_5": 0.08664, "loss_box_dn_5": 0.56607, "loss_dense_depth": 0.66965, "loss": 22.00464, "grad_norm": 33.90984, "time": 1.66197} -{"mode": "train", "epoch": 1, "iter": 487, "lr": 0.00029, "memory": 49164, "data_time": 0.07581, "loss_cls_0": 0.67084, "loss_box_0": 1.49837, "loss_cns_0": 0.64614, "loss_yns_0": 0.13211, "loss_cls_1": 0.71794, "loss_box_1": 1.40501, "loss_cns_1": 0.65868, "loss_yns_1": 0.12907, "loss_cls_2": 0.71691, "loss_box_2": 1.38306, "loss_cns_2": 0.65685, "loss_yns_2": 0.12702, "loss_cls_3": 0.73158, "loss_box_3": 1.36196, "loss_cns_3": 0.65972, "loss_yns_3": 0.12725, "loss_cls_4": 0.7375, "loss_box_4": 1.37082, "loss_cns_4": 0.659, "loss_yns_4": 0.12989, "loss_cls_5": 0.74459, "loss_box_5": 1.3907, "loss_cns_5": 0.66113, "loss_yns_5": 0.12791, "loss_cls_dn_0": 0.09904, "loss_box_dn_0": 0.7272, "loss_cls_dn_1": 0.08696, "loss_box_dn_1": 0.61141, "loss_cls_dn_2": 0.08642, "loss_box_dn_2": 0.59052, "loss_cls_dn_3": 0.08496, "loss_box_dn_3": 0.58341, "loss_cls_dn_4": 0.08579, "loss_box_dn_4": 0.58642, "loss_cls_dn_5": 0.08613, "loss_box_dn_5": 0.59691, "loss_dense_depth": 0.65507, "loss": 22.3243, "grad_norm": 48.34028, "time": 1.56186} -{"mode": "train", "epoch": 1, "iter": 488, "lr": 0.00029, "memory": 49164, "data_time": 0.10126, "loss_cls_0": 0.68385, "loss_box_0": 1.48911, "loss_cns_0": 0.6437, "loss_yns_0": 0.13417, "loss_cls_1": 0.7404, "loss_box_1": 1.38982, "loss_cns_1": 0.66345, "loss_yns_1": 0.1297, "loss_cls_2": 0.73832, "loss_box_2": 1.3454, "loss_cns_2": 0.6627, "loss_yns_2": 0.12781, "loss_cls_3": 0.7397, "loss_box_3": 1.33138, "loss_cns_3": 0.6637, "loss_yns_3": 0.12865, "loss_cls_4": 0.74187, "loss_box_4": 1.3337, "loss_cns_4": 0.67066, "loss_yns_4": 0.13397, "loss_cls_5": 0.74481, "loss_box_5": 1.35258, "loss_cns_5": 0.66481, "loss_yns_5": 0.12994, "loss_cls_dn_0": 0.09845, "loss_box_dn_0": 0.70428, "loss_cls_dn_1": 0.08585, "loss_box_dn_1": 0.60163, "loss_cls_dn_2": 0.08796, "loss_box_dn_2": 0.57572, "loss_cls_dn_3": 0.08572, "loss_box_dn_3": 0.57319, "loss_cls_dn_4": 0.08632, "loss_box_dn_4": 0.57402, "loss_cls_dn_5": 0.08672, "loss_box_dn_5": 0.58402, "loss_dense_depth": 0.68662, "loss": 22.2147, "grad_norm": 37.13873, "time": 1.66236} -{"mode": "train", "epoch": 1, "iter": 489, "lr": 0.00029, "memory": 49164, "data_time": 0.07496, "loss_cls_0": 0.6981, "loss_box_0": 1.53119, "loss_cns_0": 0.63594, "loss_yns_0": 0.13672, "loss_cls_1": 0.75279, "loss_box_1": 1.41247, "loss_cns_1": 0.66358, "loss_yns_1": 0.13421, "loss_cls_2": 0.76381, "loss_box_2": 1.40313, "loss_cns_2": 0.66087, "loss_yns_2": 0.13294, "loss_cls_3": 0.77398, "loss_box_3": 1.40717, "loss_cns_3": 0.65942, "loss_yns_3": 0.13424, "loss_cls_4": 0.76543, "loss_box_4": 1.39618, "loss_cns_4": 0.66218, "loss_yns_4": 0.13786, "loss_cls_5": 0.77209, "loss_box_5": 1.39657, "loss_cns_5": 0.65843, "loss_yns_5": 0.13523, "loss_cls_dn_0": 0.09662, "loss_box_dn_0": 0.7142, "loss_cls_dn_1": 0.08693, "loss_box_dn_1": 0.60547, "loss_cls_dn_2": 0.08726, "loss_box_dn_2": 0.59141, "loss_cls_dn_3": 0.08614, "loss_box_dn_3": 0.59171, "loss_cls_dn_4": 0.0861, "loss_box_dn_4": 0.58522, "loss_cls_dn_5": 0.08679, "loss_box_dn_5": 0.58815, "loss_dense_depth": 0.66743, "loss": 22.69794, "grad_norm": 47.47519, "time": 1.58704} -{"mode": "train", "epoch": 1, "iter": 490, "lr": 0.0003, "memory": 49164, "data_time": 0.06964, "loss_cls_0": 0.71158, "loss_box_0": 1.55482, "loss_cns_0": 0.6334, "loss_yns_0": 0.13931, "loss_cls_1": 0.76649, "loss_box_1": 1.42309, "loss_cns_1": 0.66235, "loss_yns_1": 0.13708, "loss_cls_2": 0.77004, "loss_box_2": 1.39959, "loss_cns_2": 0.65978, "loss_yns_2": 0.13598, "loss_cls_3": 0.78395, "loss_box_3": 1.40148, "loss_cns_3": 0.66108, "loss_yns_3": 0.13733, "loss_cls_4": 0.78232, "loss_box_4": 1.3991, "loss_cns_4": 0.6624, "loss_yns_4": 0.13878, "loss_cls_5": 0.79224, "loss_box_5": 1.39633, "loss_cns_5": 0.66113, "loss_yns_5": 0.13729, "loss_cls_dn_0": 0.10121, "loss_box_dn_0": 0.73072, "loss_cls_dn_1": 0.09164, "loss_box_dn_1": 0.61855, "loss_cls_dn_2": 0.09057, "loss_box_dn_2": 0.59961, "loss_cls_dn_3": 0.08992, "loss_box_dn_3": 0.59962, "loss_cls_dn_4": 0.09118, "loss_box_dn_4": 0.59585, "loss_cls_dn_5": 0.09173, "loss_box_dn_5": 0.59848, "loss_dense_depth": 0.71088, "loss": 22.95691, "grad_norm": 41.05606, "time": 1.55521} -{"mode": "train", "epoch": 1, "iter": 491, "lr": 0.0003, "memory": 49164, "data_time": 0.07532, "loss_cls_0": 0.71743, "loss_box_0": 1.53608, "loss_cns_0": 0.63894, "loss_yns_0": 0.13944, "loss_cls_1": 0.77748, "loss_box_1": 1.41907, "loss_cns_1": 0.66581, "loss_yns_1": 0.13631, "loss_cls_2": 0.78047, "loss_box_2": 1.38536, "loss_cns_2": 0.662, "loss_yns_2": 0.1363, "loss_cls_3": 0.78916, "loss_box_3": 1.38613, "loss_cns_3": 0.66327, "loss_yns_3": 0.13742, "loss_cls_4": 0.79186, "loss_box_4": 1.38467, "loss_cns_4": 0.66434, "loss_yns_4": 0.13721, "loss_cls_5": 0.79114, "loss_box_5": 1.38591, "loss_cns_5": 0.66367, "loss_yns_5": 0.13688, "loss_cls_dn_0": 0.10205, "loss_box_dn_0": 0.70594, "loss_cls_dn_1": 0.09026, "loss_box_dn_1": 0.62255, "loss_cls_dn_2": 0.09041, "loss_box_dn_2": 0.60597, "loss_cls_dn_3": 0.09013, "loss_box_dn_3": 0.60607, "loss_cls_dn_4": 0.09271, "loss_box_dn_4": 0.60466, "loss_cls_dn_5": 0.09155, "loss_box_dn_5": 0.6077, "loss_dense_depth": 0.67325, "loss": 22.90963, "grad_norm": 40.05081, "time": 1.57074} -{"mode": "train", "epoch": 1, "iter": 492, "lr": 0.0003, "memory": 49164, "data_time": 0.073, "loss_cls_0": 0.6782, "loss_box_0": 1.54611, "loss_cns_0": 0.64075, "loss_yns_0": 0.13927, "loss_cls_1": 0.76963, "loss_box_1": 1.37971, "loss_cns_1": 0.66617, "loss_yns_1": 0.13692, "loss_cls_2": 0.76764, "loss_box_2": 1.3643, "loss_cns_2": 0.66374, "loss_yns_2": 0.1361, "loss_cls_3": 0.78977, "loss_box_3": 1.35907, "loss_cns_3": 0.66352, "loss_yns_3": 0.13544, "loss_cls_4": 0.78205, "loss_box_4": 1.35638, "loss_cns_4": 0.66339, "loss_yns_4": 0.13535, "loss_cls_5": 0.77248, "loss_box_5": 1.35515, "loss_cns_5": 0.66251, "loss_yns_5": 0.13633, "loss_cls_dn_0": 0.10076, "loss_box_dn_0": 0.70352, "loss_cls_dn_1": 0.08617, "loss_box_dn_1": 0.6134, "loss_cls_dn_2": 0.08537, "loss_box_dn_2": 0.6017, "loss_cls_dn_3": 0.08653, "loss_box_dn_3": 0.59746, "loss_cls_dn_4": 0.08817, "loss_box_dn_4": 0.59449, "loss_cls_dn_5": 0.08506, "loss_box_dn_5": 0.59561, "loss_dense_depth": 0.67563, "loss": 22.61385, "grad_norm": 36.50474, "time": 1.58487} -{"mode": "train", "epoch": 1, "iter": 493, "lr": 0.0003, "memory": 49164, "data_time": 0.077, "loss_cls_0": 0.70669, "loss_box_0": 1.55199, "loss_cns_0": 0.63579, "loss_yns_0": 0.13978, "loss_cls_1": 0.77221, "loss_box_1": 1.41473, "loss_cns_1": 0.66379, "loss_yns_1": 0.13983, "loss_cls_2": 0.77842, "loss_box_2": 1.36084, "loss_cns_2": 0.6605, "loss_yns_2": 0.13806, "loss_cls_3": 0.78432, "loss_box_3": 1.35851, "loss_cns_3": 0.65942, "loss_yns_3": 0.13737, "loss_cls_4": 0.79083, "loss_box_4": 1.36237, "loss_cns_4": 0.65887, "loss_yns_4": 0.13945, "loss_cls_5": 0.79701, "loss_box_5": 1.36793, "loss_cns_5": 0.65762, "loss_yns_5": 0.13845, "loss_cls_dn_0": 0.10512, "loss_box_dn_0": 0.70349, "loss_cls_dn_1": 0.09118, "loss_box_dn_1": 0.61401, "loss_cls_dn_2": 0.09126, "loss_box_dn_2": 0.58367, "loss_cls_dn_3": 0.09216, "loss_box_dn_3": 0.5807, "loss_cls_dn_4": 0.09489, "loss_box_dn_4": 0.58204, "loss_cls_dn_5": 0.09051, "loss_box_dn_5": 0.58629, "loss_dense_depth": 0.67407, "loss": 22.70414, "grad_norm": 33.09213, "time": 1.57495} -{"mode": "train", "epoch": 1, "iter": 494, "lr": 0.0003, "memory": 49164, "data_time": 0.07424, "loss_cls_0": 0.69743, "loss_box_0": 1.51156, "loss_cns_0": 0.64166, "loss_yns_0": 0.13713, "loss_cls_1": 0.76774, "loss_box_1": 1.44382, "loss_cns_1": 0.6586, "loss_yns_1": 0.13619, "loss_cls_2": 0.77442, "loss_box_2": 1.38256, "loss_cns_2": 0.65611, "loss_yns_2": 0.13698, "loss_cls_3": 0.77501, "loss_box_3": 1.37467, "loss_cns_3": 0.65609, "loss_yns_3": 0.13606, "loss_cls_4": 0.78214, "loss_box_4": 1.38215, "loss_cns_4": 0.66083, "loss_yns_4": 0.1403, "loss_cls_5": 0.79303, "loss_box_5": 1.38879, "loss_cns_5": 0.65623, "loss_yns_5": 0.13557, "loss_cls_dn_0": 0.10173, "loss_box_dn_0": 0.6953, "loss_cls_dn_1": 0.08843, "loss_box_dn_1": 0.6195, "loss_cls_dn_2": 0.08989, "loss_box_dn_2": 0.58671, "loss_cls_dn_3": 0.08884, "loss_box_dn_3": 0.58519, "loss_cls_dn_4": 0.09045, "loss_box_dn_4": 0.58813, "loss_cls_dn_5": 0.0892, "loss_box_dn_5": 0.59594, "loss_dense_depth": 0.68492, "loss": 22.7293, "grad_norm": 37.09588, "time": 1.57315} -{"mode": "train", "epoch": 1, "iter": 495, "lr": 0.0003, "memory": 49164, "data_time": 0.07489, "loss_cls_0": 0.71535, "loss_box_0": 1.53517, "loss_cns_0": 0.64657, "loss_yns_0": 0.13948, "loss_cls_1": 0.74427, "loss_box_1": 1.40592, "loss_cns_1": 0.65896, "loss_yns_1": 0.13416, "loss_cls_2": 0.75874, "loss_box_2": 1.37117, "loss_cns_2": 0.65674, "loss_yns_2": 0.13266, "loss_cls_3": 0.76125, "loss_box_3": 1.3828, "loss_cns_3": 0.65856, "loss_yns_3": 0.13368, "loss_cls_4": 0.76187, "loss_box_4": 1.39542, "loss_cns_4": 0.6631, "loss_yns_4": 0.13402, "loss_cls_5": 0.76871, "loss_box_5": 1.39092, "loss_cns_5": 0.65605, "loss_yns_5": 0.13373, "loss_cls_dn_0": 0.10619, "loss_box_dn_0": 0.70964, "loss_cls_dn_1": 0.08747, "loss_box_dn_1": 0.60978, "loss_cls_dn_2": 0.08731, "loss_box_dn_2": 0.58859, "loss_cls_dn_3": 0.08567, "loss_box_dn_3": 0.5873, "loss_cls_dn_4": 0.08615, "loss_box_dn_4": 0.59105, "loss_cls_dn_5": 0.08751, "loss_box_dn_5": 0.59323, "loss_dense_depth": 0.68088, "loss": 22.64005, "grad_norm": 32.99357, "time": 1.6433} -{"mode": "train", "epoch": 1, "iter": 496, "lr": 0.0003, "memory": 49164, "data_time": 0.08214, "loss_cls_0": 0.71848, "loss_box_0": 1.52933, "loss_cns_0": 0.64641, "loss_yns_0": 0.13837, "loss_cls_1": 0.75185, "loss_box_1": 1.42392, "loss_cns_1": 0.659, "loss_yns_1": 0.13461, "loss_cls_2": 0.76001, "loss_box_2": 1.39187, "loss_cns_2": 0.6586, "loss_yns_2": 0.1341, "loss_cls_3": 0.76072, "loss_box_3": 1.37824, "loss_cns_3": 0.65796, "loss_yns_3": 0.13433, "loss_cls_4": 0.76498, "loss_box_4": 1.38739, "loss_cns_4": 0.65852, "loss_yns_4": 0.13398, "loss_cls_5": 0.77447, "loss_box_5": 1.39455, "loss_cns_5": 0.65907, "loss_yns_5": 0.1345, "loss_cls_dn_0": 0.10435, "loss_box_dn_0": 0.69863, "loss_cls_dn_1": 0.08571, "loss_box_dn_1": 0.61177, "loss_cls_dn_2": 0.08485, "loss_box_dn_2": 0.59238, "loss_cls_dn_3": 0.08378, "loss_box_dn_3": 0.58467, "loss_cls_dn_4": 0.08748, "loss_box_dn_4": 0.58548, "loss_cls_dn_5": 0.08587, "loss_box_dn_5": 0.58945, "loss_dense_depth": 0.67119, "loss": 22.65087, "grad_norm": 29.6501, "time": 1.57728} -{"mode": "train", "epoch": 1, "iter": 497, "lr": 0.0003, "memory": 49164, "data_time": 0.08448, "loss_cls_0": 0.69276, "loss_box_0": 1.54184, "loss_cns_0": 0.64081, "loss_yns_0": 0.13658, "loss_cls_1": 0.74522, "loss_box_1": 1.3681, "loss_cns_1": 0.66267, "loss_yns_1": 0.13551, "loss_cls_2": 0.75283, "loss_box_2": 1.37667, "loss_cns_2": 0.66343, "loss_yns_2": 0.13485, "loss_cls_3": 0.7648, "loss_box_3": 1.35387, "loss_cns_3": 0.66316, "loss_yns_3": 0.1351, "loss_cls_4": 0.76169, "loss_box_4": 1.36473, "loss_cns_4": 0.66447, "loss_yns_4": 0.1352, "loss_cls_5": 0.76211, "loss_box_5": 1.3612, "loss_cns_5": 0.66471, "loss_yns_5": 0.13563, "loss_cls_dn_0": 0.0938, "loss_box_dn_0": 0.70702, "loss_cls_dn_1": 0.08342, "loss_box_dn_1": 0.59099, "loss_cls_dn_2": 0.08279, "loss_box_dn_2": 0.59325, "loss_cls_dn_3": 0.08255, "loss_box_dn_3": 0.58841, "loss_cls_dn_4": 0.08691, "loss_box_dn_4": 0.58963, "loss_cls_dn_5": 0.0858, "loss_box_dn_5": 0.58911, "loss_dense_depth": 0.67783, "loss": 22.46942, "grad_norm": 42.39557, "time": 1.58224} -{"mode": "train", "epoch": 1, "iter": 498, "lr": 0.0003, "memory": 49164, "data_time": 0.08067, "loss_cls_0": 0.71832, "loss_box_0": 1.53259, "loss_cns_0": 0.6355, "loss_yns_0": 0.13521, "loss_cls_1": 0.74235, "loss_box_1": 1.3585, "loss_cns_1": 0.66726, "loss_yns_1": 0.13428, "loss_cls_2": 0.75738, "loss_box_2": 1.34808, "loss_cns_2": 0.66672, "loss_yns_2": 0.13338, "loss_cls_3": 0.75874, "loss_box_3": 1.34227, "loss_cns_3": 0.66718, "loss_yns_3": 0.13399, "loss_cls_4": 0.76188, "loss_box_4": 1.35546, "loss_cns_4": 0.66738, "loss_yns_4": 0.13504, "loss_cls_5": 0.7673, "loss_box_5": 1.35849, "loss_cns_5": 0.66912, "loss_yns_5": 0.13474, "loss_cls_dn_0": 0.09906, "loss_box_dn_0": 0.69876, "loss_cls_dn_1": 0.08461, "loss_box_dn_1": 0.61807, "loss_cls_dn_2": 0.0838, "loss_box_dn_2": 0.61517, "loss_cls_dn_3": 0.08373, "loss_box_dn_3": 0.61652, "loss_cls_dn_4": 0.08766, "loss_box_dn_4": 0.62468, "loss_cls_dn_5": 0.08854, "loss_box_dn_5": 0.62736, "loss_dense_depth": 0.69598, "loss": 22.6051, "grad_norm": 39.67081, "time": 1.57759} -{"mode": "train", "epoch": 1, "iter": 499, "lr": 0.0003, "memory": 49164, "data_time": 0.21583, "loss_cls_0": 0.75062, "loss_box_0": 1.51228, "loss_cns_0": 0.64083, "loss_yns_0": 0.13868, "loss_cls_1": 0.77063, "loss_box_1": 1.40249, "loss_cns_1": 0.66774, "loss_yns_1": 0.13776, "loss_cls_2": 0.78158, "loss_box_2": 1.37469, "loss_cns_2": 0.66871, "loss_yns_2": 0.13666, "loss_cls_3": 0.78182, "loss_box_3": 1.36943, "loss_cns_3": 0.66888, "loss_yns_3": 0.13759, "loss_cls_4": 0.78236, "loss_box_4": 1.36304, "loss_cns_4": 0.67271, "loss_yns_4": 0.13883, "loss_cls_5": 0.79227, "loss_box_5": 1.35436, "loss_cns_5": 0.6681, "loss_yns_5": 0.13718, "loss_cls_dn_0": 0.11306, "loss_box_dn_0": 0.69271, "loss_cls_dn_1": 0.08721, "loss_box_dn_1": 0.63041, "loss_cls_dn_2": 0.08708, "loss_box_dn_2": 0.6104, "loss_cls_dn_3": 0.08757, "loss_box_dn_3": 0.60718, "loss_cls_dn_4": 0.08789, "loss_box_dn_4": 0.60498, "loss_cls_dn_5": 0.09002, "loss_box_dn_5": 0.60266, "loss_dense_depth": 0.69657, "loss": 22.84696, "grad_norm": 42.02497, "time": 1.69587} -{"mode": "train", "epoch": 1, "iter": 500, "lr": 0.0003, "memory": 49164, "data_time": 0.07472, "loss_cls_0": 0.70285, "loss_box_0": 1.51073, "loss_cns_0": 0.64823, "loss_yns_0": 0.13826, "loss_cls_1": 0.75672, "loss_box_1": 1.36483, "loss_cns_1": 0.67103, "loss_yns_1": 0.13571, "loss_cls_2": 0.769, "loss_box_2": 1.3487, "loss_cns_2": 0.66949, "loss_yns_2": 0.13363, "loss_cls_3": 0.76368, "loss_box_3": 1.32822, "loss_cns_3": 0.66785, "loss_yns_3": 0.13412, "loss_cls_4": 0.77137, "loss_box_4": 1.31911, "loss_cns_4": 0.67237, "loss_yns_4": 0.1353, "loss_cls_5": 0.77648, "loss_box_5": 1.31298, "loss_cns_5": 0.66835, "loss_yns_5": 0.13425, "loss_cls_dn_0": 0.10417, "loss_box_dn_0": 0.7009, "loss_cls_dn_1": 0.08504, "loss_box_dn_1": 0.62628, "loss_cls_dn_2": 0.08471, "loss_box_dn_2": 0.60869, "loss_cls_dn_3": 0.08384, "loss_box_dn_3": 0.59843, "loss_cls_dn_4": 0.08326, "loss_box_dn_4": 0.59253, "loss_cls_dn_5": 0.08464, "loss_box_dn_5": 0.5902, "loss_dense_depth": 0.68741, "loss": 22.46336, "grad_norm": 41.88338, "time": 1.56068} -{"mode": "train", "epoch": 1, "iter": 501, "lr": 0.0003, "memory": 49164, "data_time": 0.07403, "loss_cls_0": 0.7001, "loss_box_0": 1.49544, "loss_cns_0": 0.64159, "loss_yns_0": 0.13844, "loss_cls_1": 0.76744, "loss_box_1": 1.34076, "loss_cns_1": 0.66748, "loss_yns_1": 0.13468, "loss_cls_2": 0.77347, "loss_box_2": 1.3357, "loss_cns_2": 0.66588, "loss_yns_2": 0.13363, "loss_cls_3": 0.76392, "loss_box_3": 1.32097, "loss_cns_3": 0.66518, "loss_yns_3": 0.13281, "loss_cls_4": 0.77998, "loss_box_4": 1.32658, "loss_cns_4": 0.66579, "loss_yns_4": 0.13463, "loss_cls_5": 0.78067, "loss_box_5": 1.32587, "loss_cns_5": 0.66628, "loss_yns_5": 0.13408, "loss_cls_dn_0": 0.09155, "loss_box_dn_0": 0.69804, "loss_cls_dn_1": 0.08186, "loss_box_dn_1": 0.5977, "loss_cls_dn_2": 0.08173, "loss_box_dn_2": 0.58471, "loss_cls_dn_3": 0.08053, "loss_box_dn_3": 0.57623, "loss_cls_dn_4": 0.08122, "loss_box_dn_4": 0.58179, "loss_cls_dn_5": 0.08157, "loss_box_dn_5": 0.58616, "loss_dense_depth": 0.67884, "loss": 22.29326, "grad_norm": 35.02705, "time": 1.70589} -{"mode": "train", "epoch": 1, "iter": 502, "lr": 0.0003, "memory": 49164, "data_time": 0.08925, "loss_cls_0": 0.72662, "loss_box_0": 1.51396, "loss_cns_0": 0.63703, "loss_yns_0": 0.13694, "loss_cls_1": 0.77424, "loss_box_1": 1.40081, "loss_cns_1": 0.66442, "loss_yns_1": 0.13378, "loss_cls_2": 0.78162, "loss_box_2": 1.35945, "loss_cns_2": 0.66394, "loss_yns_2": 0.13121, "loss_cls_3": 0.77856, "loss_box_3": 1.33133, "loss_cns_3": 0.66263, "loss_yns_3": 0.13109, "loss_cls_4": 0.78527, "loss_box_4": 1.35046, "loss_cns_4": 0.66403, "loss_yns_4": 0.13266, "loss_cls_5": 0.7943, "loss_box_5": 1.33787, "loss_cns_5": 0.66662, "loss_yns_5": 0.13316, "loss_cls_dn_0": 0.10368, "loss_box_dn_0": 0.6951, "loss_cls_dn_1": 0.08529, "loss_box_dn_1": 0.6, "loss_cls_dn_2": 0.08464, "loss_box_dn_2": 0.57882, "loss_cls_dn_3": 0.08369, "loss_box_dn_3": 0.57178, "loss_cls_dn_4": 0.08351, "loss_box_dn_4": 0.57813, "loss_cls_dn_5": 0.08526, "loss_box_dn_5": 0.58103, "loss_dense_depth": 0.67403, "loss": 22.49698, "grad_norm": 35.29139, "time": 1.67545} -{"mode": "train", "epoch": 1, "iter": 503, "lr": 0.0003, "memory": 49164, "data_time": 0.07244, "loss_cls_0": 0.69155, "loss_box_0": 1.50636, "loss_cns_0": 0.63966, "loss_yns_0": 0.13808, "loss_cls_1": 0.74767, "loss_box_1": 1.35988, "loss_cns_1": 0.66006, "loss_yns_1": 0.13508, "loss_cls_2": 0.77369, "loss_box_2": 1.31074, "loss_cns_2": 0.65558, "loss_yns_2": 0.13357, "loss_cls_3": 0.76691, "loss_box_3": 1.31528, "loss_cns_3": 0.65681, "loss_yns_3": 0.13409, "loss_cls_4": 0.7712, "loss_box_4": 1.30862, "loss_cns_4": 0.65662, "loss_yns_4": 0.1339, "loss_cls_5": 0.77026, "loss_box_5": 1.31219, "loss_cns_5": 0.65955, "loss_yns_5": 0.13436, "loss_cls_dn_0": 0.09678, "loss_box_dn_0": 0.70506, "loss_cls_dn_1": 0.07903, "loss_box_dn_1": 0.58075, "loss_cls_dn_2": 0.07901, "loss_box_dn_2": 0.56148, "loss_cls_dn_3": 0.07774, "loss_box_dn_3": 0.55634, "loss_cls_dn_4": 0.07785, "loss_box_dn_4": 0.55613, "loss_cls_dn_5": 0.07814, "loss_box_dn_5": 0.5603, "loss_dense_depth": 0.63833, "loss": 22.01865, "grad_norm": 30.70079, "time": 1.57967} -{"mode": "train", "epoch": 1, "iter": 504, "lr": 0.0003, "memory": 49164, "data_time": 0.07154, "loss_cls_0": 0.72901, "loss_box_0": 1.49288, "loss_cns_0": 0.63279, "loss_yns_0": 0.13674, "loss_cls_1": 0.7602, "loss_box_1": 1.3371, "loss_cns_1": 0.65802, "loss_yns_1": 0.13475, "loss_cls_2": 0.77646, "loss_box_2": 1.30408, "loss_cns_2": 0.65503, "loss_yns_2": 0.13265, "loss_cls_3": 0.78059, "loss_box_3": 1.31015, "loss_cns_3": 0.65817, "loss_yns_3": 0.13335, "loss_cls_4": 0.78249, "loss_box_4": 1.30523, "loss_cns_4": 0.6581, "loss_yns_4": 0.13451, "loss_cls_5": 0.78751, "loss_box_5": 1.31641, "loss_cns_5": 0.65961, "loss_yns_5": 0.13428, "loss_cls_dn_0": 0.0952, "loss_box_dn_0": 0.6948, "loss_cls_dn_1": 0.08102, "loss_box_dn_1": 0.56822, "loss_cls_dn_2": 0.08064, "loss_box_dn_2": 0.55235, "loss_cls_dn_3": 0.08083, "loss_box_dn_3": 0.54653, "loss_cls_dn_4": 0.08054, "loss_box_dn_4": 0.54542, "loss_cls_dn_5": 0.08108, "loss_box_dn_5": 0.55097, "loss_dense_depth": 0.70949, "loss": 22.07719, "grad_norm": 23.95937, "time": 1.60649} -{"mode": "train", "epoch": 1, "iter": 505, "lr": 0.0003, "memory": 49164, "data_time": 0.08972, "loss_cls_0": 0.71053, "loss_box_0": 1.50133, "loss_cns_0": 0.64018, "loss_yns_0": 0.13758, "loss_cls_1": 0.77793, "loss_box_1": 1.33934, "loss_cns_1": 0.66025, "loss_yns_1": 0.13462, "loss_cls_2": 0.7874, "loss_box_2": 1.31932, "loss_cns_2": 0.66079, "loss_yns_2": 0.13338, "loss_cls_3": 0.78801, "loss_box_3": 1.32544, "loss_cns_3": 0.66113, "loss_yns_3": 0.13323, "loss_cls_4": 0.78861, "loss_box_4": 1.33236, "loss_cns_4": 0.65983, "loss_yns_4": 0.13639, "loss_cls_5": 0.78909, "loss_box_5": 1.32774, "loss_cns_5": 0.66091, "loss_yns_5": 0.13577, "loss_cls_dn_0": 0.0985, "loss_box_dn_0": 0.71267, "loss_cls_dn_1": 0.08326, "loss_box_dn_1": 0.58266, "loss_cls_dn_2": 0.08241, "loss_box_dn_2": 0.56378, "loss_cls_dn_3": 0.08128, "loss_box_dn_3": 0.56549, "loss_cls_dn_4": 0.0818, "loss_box_dn_4": 0.57, "loss_cls_dn_5": 0.08348, "loss_box_dn_5": 0.57006, "loss_dense_depth": 0.69618, "loss": 22.31271, "grad_norm": 32.76394, "time": 1.68684} -{"mode": "train", "epoch": 1, "iter": 506, "lr": 0.0003, "memory": 49164, "data_time": 0.07723, "loss_cls_0": 0.69474, "loss_box_0": 1.49711, "loss_cns_0": 0.63646, "loss_yns_0": 0.13665, "loss_cls_1": 0.76728, "loss_box_1": 1.34312, "loss_cns_1": 0.66166, "loss_yns_1": 0.13193, "loss_cls_2": 0.79015, "loss_box_2": 1.31228, "loss_cns_2": 0.6612, "loss_yns_2": 0.13142, "loss_cls_3": 0.78438, "loss_box_3": 1.31879, "loss_cns_3": 0.66148, "loss_yns_3": 0.13212, "loss_cls_4": 0.78122, "loss_box_4": 1.31592, "loss_cns_4": 0.66194, "loss_yns_4": 0.1326, "loss_cls_5": 0.78879, "loss_box_5": 1.32699, "loss_cns_5": 0.66305, "loss_yns_5": 0.13213, "loss_cls_dn_0": 0.10041, "loss_box_dn_0": 0.69953, "loss_cls_dn_1": 0.08434, "loss_box_dn_1": 0.59456, "loss_cls_dn_2": 0.08404, "loss_box_dn_2": 0.5745, "loss_cls_dn_3": 0.08301, "loss_box_dn_3": 0.57421, "loss_cls_dn_4": 0.08381, "loss_box_dn_4": 0.57269, "loss_cls_dn_5": 0.08679, "loss_box_dn_5": 0.58062, "loss_dense_depth": 0.66713, "loss": 22.24906, "grad_norm": 26.87975, "time": 1.65607} -{"mode": "train", "epoch": 1, "iter": 507, "lr": 0.0003, "memory": 49164, "data_time": 0.07465, "loss_cls_0": 0.72223, "loss_box_0": 1.48653, "loss_cns_0": 0.63094, "loss_yns_0": 0.13843, "loss_cls_1": 0.77115, "loss_box_1": 1.32426, "loss_cns_1": 0.66451, "loss_yns_1": 0.1352, "loss_cls_2": 0.79167, "loss_box_2": 1.2882, "loss_cns_2": 0.66378, "loss_yns_2": 0.13437, "loss_cls_3": 0.79326, "loss_box_3": 1.28478, "loss_cns_3": 0.6644, "loss_yns_3": 0.13418, "loss_cls_4": 0.78984, "loss_box_4": 1.29976, "loss_cns_4": 0.66552, "loss_yns_4": 0.13534, "loss_cls_5": 0.79247, "loss_box_5": 1.29762, "loss_cns_5": 0.66669, "loss_yns_5": 0.13529, "loss_cls_dn_0": 0.09682, "loss_box_dn_0": 0.70335, "loss_cls_dn_1": 0.08565, "loss_box_dn_1": 0.59043, "loss_cls_dn_2": 0.08455, "loss_box_dn_2": 0.56881, "loss_cls_dn_3": 0.08644, "loss_box_dn_3": 0.56261, "loss_cls_dn_4": 0.08429, "loss_box_dn_4": 0.56929, "loss_cls_dn_5": 0.08516, "loss_box_dn_5": 0.56879, "loss_dense_depth": 0.70387, "loss": 22.20045, "grad_norm": 35.77492, "time": 1.55319} -{"mode": "train", "epoch": 1, "iter": 508, "lr": 0.0003, "memory": 49164, "data_time": 0.07352, "loss_cls_0": 0.68101, "loss_box_0": 1.4561, "loss_cns_0": 0.64413, "loss_yns_0": 0.1368, "loss_cls_1": 0.7558, "loss_box_1": 1.28294, "loss_cns_1": 0.66782, "loss_yns_1": 0.13368, "loss_cls_2": 0.77352, "loss_box_2": 1.25524, "loss_cns_2": 0.66655, "loss_yns_2": 0.13344, "loss_cls_3": 0.76406, "loss_box_3": 1.25466, "loss_cns_3": 0.66538, "loss_yns_3": 0.13382, "loss_cls_4": 0.75909, "loss_box_4": 1.25465, "loss_cns_4": 0.66638, "loss_yns_4": 0.13421, "loss_cls_5": 0.75718, "loss_box_5": 1.25647, "loss_cns_5": 0.66432, "loss_yns_5": 0.13425, "loss_cls_dn_0": 0.0951, "loss_box_dn_0": 0.7001, "loss_cls_dn_1": 0.08577, "loss_box_dn_1": 0.57269, "loss_cls_dn_2": 0.08695, "loss_box_dn_2": 0.55928, "loss_cls_dn_3": 0.08393, "loss_box_dn_3": 0.55595, "loss_cls_dn_4": 0.08301, "loss_box_dn_4": 0.55661, "loss_cls_dn_5": 0.08447, "loss_box_dn_5": 0.55888, "loss_dense_depth": 0.65143, "loss": 21.70563, "grad_norm": 27.20355, "time": 1.55247} -{"mode": "train", "epoch": 1, "iter": 509, "lr": 0.0003, "memory": 49164, "data_time": 0.07139, "loss_cls_0": 0.72985, "loss_box_0": 1.50056, "loss_cns_0": 0.63522, "loss_yns_0": 0.14201, "loss_cls_1": 0.75853, "loss_box_1": 1.37157, "loss_cns_1": 0.66155, "loss_yns_1": 0.13937, "loss_cls_2": 0.78291, "loss_box_2": 1.35038, "loss_cns_2": 0.66601, "loss_yns_2": 0.14138, "loss_cls_3": 0.77235, "loss_box_3": 1.32174, "loss_cns_3": 0.66444, "loss_yns_3": 0.13813, "loss_cls_4": 0.77128, "loss_box_4": 1.31827, "loss_cns_4": 0.66325, "loss_yns_4": 0.13819, "loss_cls_5": 0.77718, "loss_box_5": 1.31308, "loss_cns_5": 0.66442, "loss_yns_5": 0.13841, "loss_cls_dn_0": 0.09821, "loss_box_dn_0": 0.69627, "loss_cls_dn_1": 0.08511, "loss_box_dn_1": 0.59469, "loss_cls_dn_2": 0.09065, "loss_box_dn_2": 0.57869, "loss_cls_dn_3": 0.08401, "loss_box_dn_3": 0.57027, "loss_cls_dn_4": 0.08631, "loss_box_dn_4": 0.56641, "loss_cls_dn_5": 0.08746, "loss_box_dn_5": 0.56769, "loss_dense_depth": 0.71036, "loss": 22.37619, "grad_norm": 43.70464, "time": 1.59062} diff --git a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20260210_153254.log b/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20260210_153254.log deleted file mode 100644 index 1111ac0cd9490da100dd0784a354c305ebf5eb6e..0000000000000000000000000000000000000000 --- a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20260210_153254.log +++ /dev/null @@ -1,3221 +0,0 @@ -2026-02-10 15:32:54,327 - mmdet - INFO - Environment info: ------------------------------------------------------------- -sys.platform: linux -Python: 3.10.12 (main, Aug 15 2025, 14:32:43) [GCC 11.4.0] -CUDA available: True -GPU 0,1,2,3,4,5,6,7: BW200_LC, UBB BW1000_LC -CUDA_HOME: /opt/dtk -NVCC: Not Available -GCC: x86_64-linux-gnu-gcc (Ubuntu 11.4.0-1ubuntu1~22.04) 11.4.0 -PyTorch: 2.5.1 -PyTorch compiling details: PyTorch built with: - - GCC 10.3 - - C++ Version: 201703 - - Intel(R) Math Kernel Library Version 2020.0.4 Product Build 20200917 for Intel(R) 64 architecture applications - - OpenMP 201511 (a.k.a. OpenMP 4.5) - - LAPACK is enabled (usually provided by MKL) - - NNPACK is enabled - - CPU capability usage: AVX2 - - HIP Runtime 6.3.25521 - - MIOpen 2.18.0 - - Magma 2.8.0 - - Build settings: BLAS_INFO=mkl, BUILD_TYPE=Release, CXX_COMPILER=/opt/rh/gcc-toolset-10/root/usr/bin/c++, CXX_FLAGS= -D_GLIBCXX_USE_CXX11_ABI=1 -fvisibility-inlines-hidden -DUSE_PTHREADPOOL -DNDEBUG -DUSE_KINETO -DLIBKINETO_NOCUPTI -DLIBKINETO_NOXPUPTI=ON -DUSE_FBGEMM -DUSE_PYTORCH_QNNPACK -DUSE_XNNPACK -DSYMBOLICATE_MOBILE_DEBUG_HANDLE -O2 -fPIC -Wall -Wextra -Werror=return-type -Werror=non-virtual-dtor -Werror=bool-operation -Wnarrowing -Wno-missing-field-initializers -Wno-type-limits -Wno-array-bounds -Wno-unknown-pragmas -Wno-unused-parameter -Wno-strict-overflow -Wno-strict-aliasing -Wno-stringop-overflow -Wsuggest-override -Wno-psabi -Wno-error=old-style-cast -Wno-missing-braces -fdiagnostics-color=always -faligned-new -Wno-unused-but-set-variable -Wno-maybe-uninitialized -fno-math-errno -fno-trapping-math -Werror=format -Wno-stringop-overflow, FORCE_FALLBACK_CUDA_MPI=1, LAPACK_INFO=mkl, PERF_WITH_AVX=1, PERF_WITH_AVX2=1, TORCH_VERSION=2.5.1, USE_CUDA=0, USE_CUDNN=OFF, USE_CUSPARSELT=OFF, USE_EXCEPTION_PTR=1, USE_GFLAGS=1, USE_GLOG=1, USE_GLOO=1, USE_MKL=ON, USE_MKLDNN=0, USE_MPI=1, USE_NCCL=1, USE_NNPACK=ON, USE_OPENMP=1, USE_ROCM=ON, USE_ROCM_KERNEL_ASSERT=OFF, - -TorchVision: 0.20.1 -OpenCV: 4.12.0 -MMCV: 1.6.1 -MMCV Compiler: GCC 10.3 -MMCV CUDA Compiler: rocm not available -MMDetection: 2.25.1+ ------------------------------------------------------------- - -2026-02-10 15:32:55,331 - mmdet - INFO - Distributed training: False -2026-02-10 15:32:56,499 - mmdet - INFO - Config: -plugin = True -plugin_dir = 'projects/mmdet3d_plugin/' -dist_params = dict(backend='nccl') -log_level = 'INFO' -work_dir = './work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704' -total_batch_size = 160 -num_gpus = 8 -batch_size = 20 -num_iters_per_epoch = 175 -num_epochs = 100 -checkpoint_epoch_interval = 20 -checkpoint_config = dict(interval=3500) -log_config = dict( - interval=1, - hooks=[ - dict(type='TextLoggerHook', by_epoch=False), - dict(type='TensorboardLoggerHook') - ]) -load_from = None -resume_from = None -workflow = [('train', 1)] -fp16 = dict(loss_scale=32.0) -input_shape = (704, 256) -tracking_test = True -tracking_threshold = 0.2 -class_names = [ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier', - 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' -] -num_classes = 10 -embed_dims = 256 -num_groups = 8 -num_decoder = 6 -num_single_frame_decoder = 1 -use_deformable_func = True -strides = [4, 8, 16, 32] -num_levels = 4 -num_depth_layers = 3 -drop_out = 0.1 -temporal = True -decouple_attn = True -with_quality_estimation = True -model = dict( - type='Sparse4D', - use_grid_mask=True, - use_deformable_func=True, - img_backbone=dict( - type='ResNet', - depth=50, - num_stages=4, - frozen_stages=-1, - norm_eval=False, - style='pytorch', - with_cp=True, - out_indices=(0, 1, 2, 3), - norm_cfg=dict(type='BN', requires_grad=True), - pretrained='ckpt/resnet50-19c8e357.pth'), - img_neck=dict( - type='FPN', - num_outs=4, - start_level=0, - out_channels=256, - add_extra_convs='on_output', - relu_before_extra_convs=True, - in_channels=[256, 512, 1024, 2048]), - depth_branch=dict( - type='DenseDepthNet', - embed_dims=256, - num_depth_layers=3, - loss_weight=0.2), - head=dict( - type='Sparse4DHead', - cls_threshold_to_reg=0.05, - decouple_attn=True, - instance_bank=dict( - type='InstanceBank', - num_anchor=900, - embed_dims=256, - anchor='nuscenes_kmeans900.npy', - anchor_handler=dict(type='SparseBox3DKeyPointsGenerator'), - num_temp_instances=600, - confidence_decay=0.6, - feat_grad=False), - anchor_encoder=dict( - type='SparseBox3DEncoder', - vel_dims=3, - embed_dims=[128, 32, 32, 64], - mode='cat', - output_fc=False, - in_loops=1, - out_loops=4), - num_single_frame_decoder=1, - operation_order=[ - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine' - ], - temp_graph_model=dict( - type='MultiheadAttention', - embed_dims=512, - num_heads=8, - batch_first=True, - dropout=0.1), - graph_model=dict( - type='MultiheadAttention', - embed_dims=512, - num_heads=8, - batch_first=True, - dropout=0.1), - norm_layer=dict(type='LN', normalized_shape=256), - ffn=dict( - type='AsymmetricFFN', - in_channels=512, - pre_norm=dict(type='LN'), - embed_dims=256, - feedforward_channels=1024, - num_fcs=2, - ffn_drop=0.1, - act_cfg=dict(type='ReLU', inplace=True)), - deformable_model=dict( - type='DeformableFeatureAggregation', - embed_dims=256, - num_groups=8, - num_levels=4, - num_cams=6, - attn_drop=0.15, - use_deformable_func=True, - use_camera_embed=True, - residual_mode='cat', - kps_generator=dict( - type='SparseBox3DKeyPointsGenerator', - num_learnable_pts=6, - fix_scale=[[0, 0, 0], [0.45, 0, 0], [-0.45, 0, 0], - [0, 0.45, 0], [0, -0.45, 0], [0, 0, 0.45], - [0, 0, -0.45]])), - refine_layer=dict( - type='SparseBox3DRefinementModule', - embed_dims=256, - num_cls=10, - refine_yaw=True, - with_quality_estimation=True), - sampler=dict( - type='SparseBox3DTarget', - num_dn_groups=5, - num_temp_dn_groups=3, - dn_noise_scale=[2.0, 2.0, 2.0, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5], - max_dn_gt=32, - add_neg_dn=True, - cls_weight=2.0, - box_weight=0.25, - reg_weights=[2.0, 2.0, 2.0, 0.5, 0.5, 0.5, 0.0, 0.0, 0.0, 0.0], - cls_wise_reg_weights=dict( - {9: [2.0, 2.0, 2.0, 1.0, 1.0, 1.0, 0.0, 0.0, 1.0, 1.0]})), - loss_cls=dict( - type='FocalLoss', - use_sigmoid=True, - gamma=2.0, - alpha=0.25, - loss_weight=2.0), - loss_reg=dict( - type='SparseBox3DLoss', - loss_box=dict(type='L1Loss', loss_weight=0.25), - loss_centerness=dict(type='CrossEntropyLoss', use_sigmoid=True), - loss_yawness=dict(type='GaussianFocalLoss'), - cls_allow_reverse=[5]), - decoder=dict(type='SparseBox3DDecoder'), - reg_weights=[2.0, 2.0, 2.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0])) -dataset_type = 'NuScenes3DDetTrackDataset' -data_root = 'data/nuscenes/' -anno_root = 'data/nuscenes_anno_pkls/' -file_client_args = dict(backend='disk') -img_norm_cfg = dict( - mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375], to_rgb=True) -train_pipeline = [ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict( - type='LoadPointsFromFile', - coord_type='LIDAR', - load_dim=5, - use_dim=5, - file_client_args=dict(backend='disk')), - dict(type='ResizeCropFlipImage'), - dict(type='MultiScaleDepthMapGenerator', downsample=[4, 8, 16]), - dict(type='BBoxRotation'), - dict(type='PhotoMetricDistortionMultiViewImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict( - type='CircleObjectRangeFilter', - class_dist_thred=[55, 55, 55, 55, 55, 55, 55, 55, 55, 55]), - dict( - type='InstanceNameFilter', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ]), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=[ - 'img', 'timestamp', 'projection_mat', 'image_wh', 'gt_depth', - 'focal', 'gt_bboxes_3d', 'gt_labels_3d' - ], - meta_keys=['T_global', 'T_global_inv', 'timestamp', 'instance_id']) -] -test_pipeline = [ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='ResizeCropFlipImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=['img', 'timestamp', 'projection_mat', 'image_wh'], - meta_keys=['T_global', 'T_global_inv', 'timestamp']) -] -input_modality = dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False) -data_basic_config = dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier', - 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval') -data_aug_conf = dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]) -data = dict( - samples_per_gpu=20, - workers_per_gpu=20, - train=dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval', - ann_file='data/nuscenes_anno_pkls/nuscenes_infos_train.pkl', - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict( - type='LoadPointsFromFile', - coord_type='LIDAR', - load_dim=5, - use_dim=5, - file_client_args=dict(backend='disk')), - dict(type='ResizeCropFlipImage'), - dict(type='MultiScaleDepthMapGenerator', downsample=[4, 8, 16]), - dict(type='BBoxRotation'), - dict(type='PhotoMetricDistortionMultiViewImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict( - type='CircleObjectRangeFilter', - class_dist_thred=[55, 55, 55, 55, 55, 55, 55, 55, 55, 55]), - dict( - type='InstanceNameFilter', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', - 'traffic_cone' - ]), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=[ - 'img', 'timestamp', 'projection_mat', 'image_wh', - 'gt_depth', 'focal', 'gt_bboxes_3d', 'gt_labels_3d' - ], - meta_keys=[ - 'T_global', 'T_global_inv', 'timestamp', 'instance_id' - ]) - ], - test_mode=False, - data_aug_conf=dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]), - with_seq_flag=True, - sequences_split_num=2, - keep_consistent_seq_aug=True), - val=dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval', - ann_file='data/nuscenes_anno_pkls/nuscenes_infos_val.pkl', - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='ResizeCropFlipImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=['img', 'timestamp', 'projection_mat', 'image_wh'], - meta_keys=['T_global', 'T_global_inv', 'timestamp']) - ], - data_aug_conf=dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]), - test_mode=True, - tracking=True, - tracking_threshold=0.2), - test=dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval', - ann_file='data/nuscenes_anno_pkls/nuscenes_infos_val.pkl', - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='ResizeCropFlipImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=['img', 'timestamp', 'projection_mat', 'image_wh'], - meta_keys=['T_global', 'T_global_inv', 'timestamp']) - ], - data_aug_conf=dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]), - test_mode=True, - tracking=True, - tracking_threshold=0.2)) -optimizer = dict( - type='AdamW', - lr=0.0006, - weight_decay=0.001, - paramwise_cfg=dict(custom_keys=dict(img_backbone=dict(lr_mult=0.5)))) -optimizer_config = dict(grad_clip=dict(max_norm=25, norm_type=2)) -lr_config = dict( - policy='CosineAnnealing', - warmup='linear', - warmup_iters=500, - warmup_ratio=0.3333333333333333, - min_lr_ratio=0.001) -runner = dict(type='IterBasedRunner', max_iters=17500) -vis_pipeline = [ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='Collect', keys=['img'], meta_keys=['timestamp', 'lidar2img']) -] -evaluation = dict( - interval=3500, - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict( - type='Collect', keys=['img'], meta_keys=['timestamp', 'lidar2img']) - ]) -gpu_ids = range(0, 1) - -2026-02-10 15:32:56,499 - mmdet - INFO - Set random seed to 0, deterministic: False -2026-02-10 15:32:57,060 - mmdet - INFO - initialize ResNet with init_cfg {'type': 'Pretrained', 'checkpoint': 'ckpt/resnet50-19c8e357.pth'} -2026-02-10 15:32:57,345 - mmdet - INFO - initialize FPN with init_cfg {'type': 'Xavier', 'layer': 'Conv2d', 'distribution': 'uniform'} -Name of parameter - Initialization information - -img_backbone.conv1.weight - torch.Size([64, 3, 7, 7]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.bn1.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.bn1.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.conv1.weight - torch.Size([64, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn1.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn1.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.conv2.weight - torch.Size([64, 64, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn2.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn2.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.conv3.weight - torch.Size([256, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn3.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn3.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.downsample.0.weight - torch.Size([256, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.downsample.1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.downsample.1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.conv1.weight - torch.Size([64, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn1.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn1.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.conv2.weight - torch.Size([64, 64, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn2.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn2.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.conv3.weight - torch.Size([256, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn3.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn3.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.conv1.weight - torch.Size([64, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn1.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn1.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.conv2.weight - torch.Size([64, 64, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn2.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn2.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.conv3.weight - torch.Size([256, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn3.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn3.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.conv1.weight - torch.Size([128, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn1.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn1.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.conv2.weight - torch.Size([128, 128, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn2.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn2.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.conv3.weight - torch.Size([512, 128, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn3.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn3.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.downsample.0.weight - torch.Size([512, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.downsample.1.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.downsample.1.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.conv1.weight - torch.Size([128, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn1.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn1.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.conv2.weight - torch.Size([128, 128, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn2.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn2.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.conv3.weight - torch.Size([512, 128, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn3.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn3.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.conv1.weight - torch.Size([128, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn1.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn1.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.conv2.weight - torch.Size([128, 128, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn2.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn2.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.conv3.weight - torch.Size([512, 128, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn3.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn3.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.conv1.weight - torch.Size([128, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn1.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn1.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.conv2.weight - torch.Size([128, 128, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn2.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn2.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.conv3.weight - torch.Size([512, 128, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn3.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn3.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.conv1.weight - torch.Size([256, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.downsample.0.weight - torch.Size([1024, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.downsample.1.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.downsample.1.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.conv1.weight - torch.Size([512, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn1.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn1.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.conv2.weight - torch.Size([512, 512, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn2.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn2.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.conv3.weight - torch.Size([2048, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn3.weight - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn3.bias - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.downsample.0.weight - torch.Size([2048, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.downsample.1.weight - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.downsample.1.bias - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.conv1.weight - torch.Size([512, 2048, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn1.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn1.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.conv2.weight - torch.Size([512, 512, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn2.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn2.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.conv3.weight - torch.Size([2048, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn3.weight - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn3.bias - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.conv1.weight - torch.Size([512, 2048, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn1.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn1.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.conv2.weight - torch.Size([512, 512, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn2.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn2.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.conv3.weight - torch.Size([2048, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn3.weight - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn3.bias - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_neck.lateral_convs.0.conv.weight - torch.Size([256, 256, 1, 1]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.lateral_convs.0.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.lateral_convs.1.conv.weight - torch.Size([256, 512, 1, 1]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.lateral_convs.1.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.lateral_convs.2.conv.weight - torch.Size([256, 1024, 1, 1]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.lateral_convs.2.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.lateral_convs.3.conv.weight - torch.Size([256, 2048, 1, 1]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.lateral_convs.3.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.fpn_convs.0.conv.weight - torch.Size([256, 256, 3, 3]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.fpn_convs.0.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.fpn_convs.1.conv.weight - torch.Size([256, 256, 3, 3]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.fpn_convs.1.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.fpn_convs.2.conv.weight - torch.Size([256, 256, 3, 3]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.fpn_convs.2.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.fpn_convs.3.conv.weight - torch.Size([256, 256, 3, 3]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.fpn_convs.3.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.instance_bank.anchor - torch.Size([900, 11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.instance_bank.instance_feature - torch.Size([900, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.instance_bank.anchor_handler.fix_scale - torch.Size([1, 3]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.0.weight - torch.Size([128, 3]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.0.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.2.weight - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.2.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.3.weight - torch.Size([128, 128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.3.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.5.weight - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.5.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.6.weight - torch.Size([128, 128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.6.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.8.weight - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.8.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.9.weight - torch.Size([128, 128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.9.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.11.weight - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.11.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.0.weight - torch.Size([32, 3]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.0.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.2.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.2.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.3.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.3.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.5.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.5.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.6.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.6.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.8.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.8.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.9.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.9.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.11.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.11.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.0.weight - torch.Size([32, 2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.0.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.2.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.2.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.3.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.3.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.5.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.5.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.6.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.6.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.8.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.8.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.9.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.9.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.11.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.11.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.0.weight - torch.Size([64, 3]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.0.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.2.weight - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.2.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.3.weight - torch.Size([64, 64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.3.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.5.weight - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.5.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.6.weight - torch.Size([64, 64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.6.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.8.weight - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.8.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.9.weight - torch.Size([64, 64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.9.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.11.weight - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.11.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.1.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.1.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.1.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.1.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.1.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.1.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.1.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.1.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.3.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.4.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.4.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.4.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.4.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.5.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.5.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.5.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.5.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.6.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.6.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.8.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.8.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.8.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.8.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.8.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.8.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.8.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.8.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.10.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.11.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.11.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.11.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.11.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.12.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.12.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.12.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.12.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.13.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.13.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.15.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.15.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.15.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.15.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.15.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.15.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.15.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.15.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.16.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.16.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.17.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.18.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.18.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.18.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.18.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.19.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.19.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.19.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.19.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.20.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.20.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.22.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.22.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.22.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.22.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.22.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.22.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.22.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.22.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.23.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.23.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.24.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.25.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.25.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.25.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.25.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.26.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.26.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.26.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.26.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.27.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.27.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.29.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.29.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.29.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.29.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.29.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.29.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.29.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.29.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.30.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.30.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.31.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.32.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.32.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.32.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.32.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.33.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.33.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.33.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.33.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.34.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.34.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.36.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.36.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.36.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.36.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.36.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.36.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.36.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.36.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.37.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.37.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.38.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.fc_before.weight - torch.Size([512, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.fc_after.weight - torch.Size([256, 512]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.0.weight - torch.Size([1, 256, 1, 1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.0.bias - torch.Size([1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.1.weight - torch.Size([1, 256, 1, 1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.1.bias - torch.Size([1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.2.weight - torch.Size([1, 256, 1, 1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.2.bias - torch.Size([1]): -The value is the same before and after calling `init_weights` of Sparse4D -2026-02-10 15:32:57,692 - mmdet - INFO - Model: -Sparse4D( - (img_backbone): ResNet( - (conv1): Conv2d(3, 64, kernel_size=(7, 7), stride=(2, 2), padding=(3, 3), bias=False) - (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (maxpool): MaxPool2d(kernel_size=3, stride=2, padding=1, dilation=1, ceil_mode=False) - (layer1): ResLayer( - (0): Bottleneck( - (conv1): Conv2d(64, 64, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (downsample): Sequential( - (0): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - ) - ) - (1): Bottleneck( - (conv1): Conv2d(256, 64, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (2): Bottleneck( - (conv1): Conv2d(256, 64, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - ) - (layer2): ResLayer( - (0): Bottleneck( - (conv1): Conv2d(256, 128, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (downsample): Sequential( - (0): Conv2d(256, 512, kernel_size=(1, 1), stride=(2, 2), bias=False) - (1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - ) - ) - (1): Bottleneck( - (conv1): Conv2d(512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (2): Bottleneck( - (conv1): Conv2d(512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (3): Bottleneck( - (conv1): Conv2d(512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - ) - (layer3): ResLayer( - (0): Bottleneck( - (conv1): Conv2d(512, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (downsample): Sequential( - (0): Conv2d(512, 1024, kernel_size=(1, 1), stride=(2, 2), bias=False) - (1): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - ) - ) - (1): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (2): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (3): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (4): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (5): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - ) - (layer4): ResLayer( - (0): Bottleneck( - (conv1): Conv2d(1024, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(512, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (downsample): Sequential( - (0): Conv2d(1024, 2048, kernel_size=(1, 1), stride=(2, 2), bias=False) - (1): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - ) - ) - (1): Bottleneck( - (conv1): Conv2d(2048, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(512, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (2): Bottleneck( - (conv1): Conv2d(2048, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(512, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - ) - ) - init_cfg={'type': 'Pretrained', 'checkpoint': 'ckpt/resnet50-19c8e357.pth'} - (img_neck): FPN( - (lateral_convs): ModuleList( - (0): ConvModule( - (conv): Conv2d(256, 256, kernel_size=(1, 1), stride=(1, 1)) - ) - (1): ConvModule( - (conv): Conv2d(512, 256, kernel_size=(1, 1), stride=(1, 1)) - ) - (2): ConvModule( - (conv): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1)) - ) - (3): ConvModule( - (conv): Conv2d(2048, 256, kernel_size=(1, 1), stride=(1, 1)) - ) - ) - (fpn_convs): ModuleList( - (0-3): 4 x ConvModule( - (conv): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1)) - ) - ) - ) - init_cfg={'type': 'Xavier', 'layer': 'Conv2d', 'distribution': 'uniform'} - (head): Sparse4DHead( - (instance_bank): InstanceBank( - (anchor_handler): SparseBox3DKeyPointsGenerator() - ) - (anchor_encoder): SparseBox3DEncoder( - (pos_fc): Sequential( - (0): Linear(in_features=3, out_features=128, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((128,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=128, out_features=128, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((128,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=128, out_features=128, bias=True) - (7): ReLU(inplace=True) - (8): LayerNorm((128,), eps=1e-05, elementwise_affine=True) - (9): Linear(in_features=128, out_features=128, bias=True) - (10): ReLU(inplace=True) - (11): LayerNorm((128,), eps=1e-05, elementwise_affine=True) - ) - (size_fc): Sequential( - (0): Linear(in_features=3, out_features=32, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=32, out_features=32, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=32, out_features=32, bias=True) - (7): ReLU(inplace=True) - (8): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (9): Linear(in_features=32, out_features=32, bias=True) - (10): ReLU(inplace=True) - (11): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - ) - (yaw_fc): Sequential( - (0): Linear(in_features=2, out_features=32, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=32, out_features=32, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=32, out_features=32, bias=True) - (7): ReLU(inplace=True) - (8): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (9): Linear(in_features=32, out_features=32, bias=True) - (10): ReLU(inplace=True) - (11): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - ) - (vel_fc): Sequential( - (0): Linear(in_features=3, out_features=64, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((64,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=64, out_features=64, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((64,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=64, out_features=64, bias=True) - (7): ReLU(inplace=True) - (8): LayerNorm((64,), eps=1e-05, elementwise_affine=True) - (9): Linear(in_features=64, out_features=64, bias=True) - (10): ReLU(inplace=True) - (11): LayerNorm((64,), eps=1e-05, elementwise_affine=True) - ) - ) - (loss_cls): FocalLoss() - (loss_reg): SparseBox3DLoss( - (loss_box): L1Loss() - (loss_cns): CrossEntropyLoss(avg_non_ignore=False) - (loss_yns): GaussianFocalLoss() - ) - (layers): ModuleList( - (0): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (1): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (4-5): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (6): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (7): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (8): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (11-12): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (13): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (14): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (15): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (16): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (17): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (18-19): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (20): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (21): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (22): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (23): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (24): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (25-26): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (27): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (28): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (29): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (30): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (31): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (32-33): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (34): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (35): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (36): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (37): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (38): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - ) - (fc_before): Linear(in_features=256, out_features=512, bias=False) - (fc_after): Linear(in_features=512, out_features=256, bias=False) - ) - (depth_branch): DenseDepthNet( - (depth_layers): ModuleList( - (0-2): 3 x Conv2d(256, 1, kernel_size=(1, 1), stride=(1, 1)) - ) - ) - (grid_mask): GridMask() -) -2026-02-10 15:33:11,751 - mmdet - INFO - Start running, host: root@bw28, work_dir: /workspace/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704 -2026-02-10 15:33:11,751 - mmdet - INFO - Hooks will be executed in the following order: -before_run: -(VERY_HIGH ) CosineAnnealingLrUpdaterHook -(ABOVE_NORMAL) Fp16OptimizerHook -(NORMAL ) CheckpointHook -(NORMAL ) EvalHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -before_train_epoch: -(VERY_HIGH ) CosineAnnealingLrUpdaterHook -(NORMAL ) EvalHook -(LOW ) IterTimerHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -before_train_iter: -(VERY_HIGH ) CosineAnnealingLrUpdaterHook -(NORMAL ) EvalHook -(LOW ) IterTimerHook - -------------------- -after_train_iter: -(ABOVE_NORMAL) Fp16OptimizerHook -(NORMAL ) CheckpointHook -(NORMAL ) EvalHook -(NORMAL ) ProfilerHook -(LOW ) IterTimerHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -after_train_epoch: -(NORMAL ) CheckpointHook -(NORMAL ) EvalHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -before_val_epoch: -(LOW ) IterTimerHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -before_val_iter: -(LOW ) IterTimerHook - -------------------- -after_val_iter: -(LOW ) IterTimerHook - -------------------- -after_val_epoch: -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -after_run: -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -2026-02-10 15:33:11,751 - mmdet - INFO - workflow: [('train', 1)], max: 17500 iters -2026-02-10 15:33:11,754 - mmdet - INFO - Checkpoints will be saved to /workspace/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704 by HardDiskBackend. diff --git a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20260210_153254.log.json b/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20260210_153254.log.json deleted file mode 100644 index f691fdae6b2eb0b287f959b7a72d73034a4da3e6..0000000000000000000000000000000000000000 --- a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20260210_153254.log.json +++ /dev/null @@ -1 +0,0 @@ -{"env_info": "sys.platform: linux\nPython: 3.10.12 (main, Aug 15 2025, 14:32:43) [GCC 11.4.0]\nCUDA available: True\nGPU 0,1,2,3,4,5,6,7: BW200_LC, UBB BW1000_LC\nCUDA_HOME: /opt/dtk\nNVCC: Not Available\nGCC: x86_64-linux-gnu-gcc (Ubuntu 11.4.0-1ubuntu1~22.04) 11.4.0\nPyTorch: 2.5.1\nPyTorch compiling details: PyTorch built with:\n - GCC 10.3\n - C++ Version: 201703\n - Intel(R) Math Kernel Library Version 2020.0.4 Product Build 20200917 for Intel(R) 64 architecture applications\n - OpenMP 201511 (a.k.a. OpenMP 4.5)\n - LAPACK is enabled (usually provided by MKL)\n - NNPACK is enabled\n - CPU capability usage: AVX2\n - HIP Runtime 6.3.25521\n - MIOpen 2.18.0\n - Magma 2.8.0\n - Build settings: BLAS_INFO=mkl, BUILD_TYPE=Release, CXX_COMPILER=/opt/rh/gcc-toolset-10/root/usr/bin/c++, CXX_FLAGS= -D_GLIBCXX_USE_CXX11_ABI=1 -fvisibility-inlines-hidden -DUSE_PTHREADPOOL -DNDEBUG -DUSE_KINETO -DLIBKINETO_NOCUPTI -DLIBKINETO_NOXPUPTI=ON -DUSE_FBGEMM -DUSE_PYTORCH_QNNPACK -DUSE_XNNPACK -DSYMBOLICATE_MOBILE_DEBUG_HANDLE -O2 -fPIC -Wall -Wextra -Werror=return-type -Werror=non-virtual-dtor -Werror=bool-operation -Wnarrowing -Wno-missing-field-initializers -Wno-type-limits -Wno-array-bounds -Wno-unknown-pragmas -Wno-unused-parameter -Wno-strict-overflow -Wno-strict-aliasing -Wno-stringop-overflow -Wsuggest-override -Wno-psabi -Wno-error=old-style-cast -Wno-missing-braces -fdiagnostics-color=always -faligned-new -Wno-unused-but-set-variable -Wno-maybe-uninitialized -fno-math-errno -fno-trapping-math -Werror=format -Wno-stringop-overflow, FORCE_FALLBACK_CUDA_MPI=1, LAPACK_INFO=mkl, PERF_WITH_AVX=1, PERF_WITH_AVX2=1, TORCH_VERSION=2.5.1, USE_CUDA=0, USE_CUDNN=OFF, USE_CUSPARSELT=OFF, USE_EXCEPTION_PTR=1, USE_GFLAGS=1, USE_GLOG=1, USE_GLOO=1, USE_MKL=ON, USE_MKLDNN=0, USE_MPI=1, USE_NCCL=1, USE_NNPACK=ON, USE_OPENMP=1, USE_ROCM=ON, USE_ROCM_KERNEL_ASSERT=OFF, \n\nTorchVision: 0.20.1\nOpenCV: 4.12.0\nMMCV: 1.6.1\nMMCV Compiler: GCC 10.3\nMMCV CUDA Compiler: rocm not available\nMMDetection: 2.25.1+", "config": "plugin = True\nplugin_dir = 'projects/mmdet3d_plugin/'\ndist_params = dict(backend='nccl')\nlog_level = 'INFO'\nwork_dir = './work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704'\ntotal_batch_size = 160\nnum_gpus = 8\nbatch_size = 20\nnum_iters_per_epoch = 175\nnum_epochs = 100\ncheckpoint_epoch_interval = 20\ncheckpoint_config = dict(interval=3500)\nlog_config = dict(\n interval=1,\n hooks=[\n dict(type='TextLoggerHook', by_epoch=False),\n dict(type='TensorboardLoggerHook')\n ])\nload_from = None\nresume_from = None\nworkflow = [('train', 1)]\nfp16 = dict(loss_scale=32.0)\ninput_shape = (704, 256)\ntracking_test = True\ntracking_threshold = 0.2\nclass_names = [\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier',\n 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n]\nnum_classes = 10\nembed_dims = 256\nnum_groups = 8\nnum_decoder = 6\nnum_single_frame_decoder = 1\nuse_deformable_func = True\nstrides = [4, 8, 16, 32]\nnum_levels = 4\nnum_depth_layers = 3\ndrop_out = 0.1\ntemporal = True\ndecouple_attn = True\nwith_quality_estimation = True\nmodel = dict(\n type='Sparse4D',\n use_grid_mask=True,\n use_deformable_func=True,\n img_backbone=dict(\n type='ResNet',\n depth=50,\n num_stages=4,\n frozen_stages=-1,\n norm_eval=False,\n style='pytorch',\n with_cp=True,\n out_indices=(0, 1, 2, 3),\n norm_cfg=dict(type='BN', requires_grad=True),\n pretrained='ckpt/resnet50-19c8e357.pth'),\n img_neck=dict(\n type='FPN',\n num_outs=4,\n start_level=0,\n out_channels=256,\n add_extra_convs='on_output',\n relu_before_extra_convs=True,\n in_channels=[256, 512, 1024, 2048]),\n depth_branch=dict(\n type='DenseDepthNet',\n embed_dims=256,\n num_depth_layers=3,\n loss_weight=0.2),\n head=dict(\n type='Sparse4DHead',\n cls_threshold_to_reg=0.05,\n decouple_attn=True,\n instance_bank=dict(\n type='InstanceBank',\n num_anchor=900,\n embed_dims=256,\n anchor='nuscenes_kmeans900.npy',\n anchor_handler=dict(type='SparseBox3DKeyPointsGenerator'),\n num_temp_instances=600,\n confidence_decay=0.6,\n feat_grad=False),\n anchor_encoder=dict(\n type='SparseBox3DEncoder',\n vel_dims=3,\n embed_dims=[128, 32, 32, 64],\n mode='cat',\n output_fc=False,\n in_loops=1,\n out_loops=4),\n num_single_frame_decoder=1,\n operation_order=[\n 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm',\n 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm',\n 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm',\n 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm',\n 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm',\n 'deformable', 'ffn', 'norm', 'refine'\n ],\n temp_graph_model=dict(\n type='MultiheadAttention',\n embed_dims=512,\n num_heads=8,\n batch_first=True,\n dropout=0.1),\n graph_model=dict(\n type='MultiheadAttention',\n embed_dims=512,\n num_heads=8,\n batch_first=True,\n dropout=0.1),\n norm_layer=dict(type='LN', normalized_shape=256),\n ffn=dict(\n type='AsymmetricFFN',\n in_channels=512,\n pre_norm=dict(type='LN'),\n embed_dims=256,\n feedforward_channels=1024,\n num_fcs=2,\n ffn_drop=0.1,\n act_cfg=dict(type='ReLU', inplace=True)),\n deformable_model=dict(\n type='DeformableFeatureAggregation',\n embed_dims=256,\n num_groups=8,\n num_levels=4,\n num_cams=6,\n attn_drop=0.15,\n use_deformable_func=True,\n use_camera_embed=True,\n residual_mode='cat',\n kps_generator=dict(\n type='SparseBox3DKeyPointsGenerator',\n num_learnable_pts=6,\n fix_scale=[[0, 0, 0], [0.45, 0, 0], [-0.45, 0, 0],\n [0, 0.45, 0], [0, -0.45, 0], [0, 0, 0.45],\n [0, 0, -0.45]])),\n refine_layer=dict(\n type='SparseBox3DRefinementModule',\n embed_dims=256,\n num_cls=10,\n refine_yaw=True,\n with_quality_estimation=True),\n sampler=dict(\n type='SparseBox3DTarget',\n num_dn_groups=5,\n num_temp_dn_groups=3,\n dn_noise_scale=[2.0, 2.0, 2.0, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5],\n max_dn_gt=32,\n add_neg_dn=True,\n cls_weight=2.0,\n box_weight=0.25,\n reg_weights=[2.0, 2.0, 2.0, 0.5, 0.5, 0.5, 0.0, 0.0, 0.0, 0.0],\n cls_wise_reg_weights=dict(\n {9: [2.0, 2.0, 2.0, 1.0, 1.0, 1.0, 0.0, 0.0, 1.0, 1.0]})),\n loss_cls=dict(\n type='FocalLoss',\n use_sigmoid=True,\n gamma=2.0,\n alpha=0.25,\n loss_weight=2.0),\n loss_reg=dict(\n type='SparseBox3DLoss',\n loss_box=dict(type='L1Loss', loss_weight=0.25),\n loss_centerness=dict(type='CrossEntropyLoss', use_sigmoid=True),\n loss_yawness=dict(type='GaussianFocalLoss'),\n cls_allow_reverse=[5]),\n decoder=dict(type='SparseBox3DDecoder'),\n reg_weights=[2.0, 2.0, 2.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0]))\ndataset_type = 'NuScenes3DDetTrackDataset'\ndata_root = 'data/nuscenes/'\nanno_root = 'data/nuscenes_anno_pkls/'\nfile_client_args = dict(backend='disk')\nimg_norm_cfg = dict(\n mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375], to_rgb=True)\ntrain_pipeline = [\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(\n type='LoadPointsFromFile',\n coord_type='LIDAR',\n load_dim=5,\n use_dim=5,\n file_client_args=dict(backend='disk')),\n dict(type='ResizeCropFlipImage'),\n dict(type='MultiScaleDepthMapGenerator', downsample=[4, 8, 16]),\n dict(type='BBoxRotation'),\n dict(type='PhotoMetricDistortionMultiViewImage'),\n dict(\n type='NormalizeMultiviewImage',\n mean=[123.675, 116.28, 103.53],\n std=[58.395, 57.12, 57.375],\n to_rgb=True),\n dict(\n type='CircleObjectRangeFilter',\n class_dist_thred=[55, 55, 55, 55, 55, 55, 55, 55, 55, 55]),\n dict(\n type='InstanceNameFilter',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ]),\n dict(type='NuScenesSparse4DAdaptor'),\n dict(\n type='Collect',\n keys=[\n 'img', 'timestamp', 'projection_mat', 'image_wh', 'gt_depth',\n 'focal', 'gt_bboxes_3d', 'gt_labels_3d'\n ],\n meta_keys=['T_global', 'T_global_inv', 'timestamp', 'instance_id'])\n]\ntest_pipeline = [\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(type='ResizeCropFlipImage'),\n dict(\n type='NormalizeMultiviewImage',\n mean=[123.675, 116.28, 103.53],\n std=[58.395, 57.12, 57.375],\n to_rgb=True),\n dict(type='NuScenesSparse4DAdaptor'),\n dict(\n type='Collect',\n keys=['img', 'timestamp', 'projection_mat', 'image_wh'],\n meta_keys=['T_global', 'T_global_inv', 'timestamp'])\n]\ninput_modality = dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False)\ndata_basic_config = dict(\n type='NuScenes3DDetTrackDataset',\n data_root='data/nuscenes/',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier',\n 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ],\n modality=dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False),\n version='v1.0-trainval')\ndata_aug_conf = dict(\n resize_lim=(0.4, 0.47),\n final_dim=(256, 704),\n bot_pct_lim=(0.0, 0.0),\n rot_lim=(-5.4, 5.4),\n H=900,\n W=1600,\n rand_flip=True,\n rot3d_range=[-0.3925, 0.3925])\ndata = dict(\n samples_per_gpu=20,\n workers_per_gpu=20,\n train=dict(\n type='NuScenes3DDetTrackDataset',\n data_root='data/nuscenes/',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ],\n modality=dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False),\n version='v1.0-trainval',\n ann_file='data/nuscenes_anno_pkls/nuscenes_infos_train.pkl',\n pipeline=[\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(\n type='LoadPointsFromFile',\n coord_type='LIDAR',\n load_dim=5,\n use_dim=5,\n file_client_args=dict(backend='disk')),\n dict(type='ResizeCropFlipImage'),\n dict(type='MultiScaleDepthMapGenerator', downsample=[4, 8, 16]),\n dict(type='BBoxRotation'),\n dict(type='PhotoMetricDistortionMultiViewImage'),\n dict(\n type='NormalizeMultiviewImage',\n mean=[123.675, 116.28, 103.53],\n std=[58.395, 57.12, 57.375],\n to_rgb=True),\n dict(\n type='CircleObjectRangeFilter',\n class_dist_thred=[55, 55, 55, 55, 55, 55, 55, 55, 55, 55]),\n dict(\n type='InstanceNameFilter',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian',\n 'traffic_cone'\n ]),\n dict(type='NuScenesSparse4DAdaptor'),\n dict(\n type='Collect',\n keys=[\n 'img', 'timestamp', 'projection_mat', 'image_wh',\n 'gt_depth', 'focal', 'gt_bboxes_3d', 'gt_labels_3d'\n ],\n meta_keys=[\n 'T_global', 'T_global_inv', 'timestamp', 'instance_id'\n ])\n ],\n test_mode=False,\n data_aug_conf=dict(\n resize_lim=(0.4, 0.47),\n final_dim=(256, 704),\n bot_pct_lim=(0.0, 0.0),\n rot_lim=(-5.4, 5.4),\n H=900,\n W=1600,\n rand_flip=True,\n rot3d_range=[-0.3925, 0.3925]),\n with_seq_flag=True,\n sequences_split_num=2,\n keep_consistent_seq_aug=True),\n val=dict(\n type='NuScenes3DDetTrackDataset',\n data_root='data/nuscenes/',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ],\n modality=dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False),\n version='v1.0-trainval',\n ann_file='data/nuscenes_anno_pkls/nuscenes_infos_val.pkl',\n pipeline=[\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(type='ResizeCropFlipImage'),\n dict(\n type='NormalizeMultiviewImage',\n mean=[123.675, 116.28, 103.53],\n std=[58.395, 57.12, 57.375],\n to_rgb=True),\n dict(type='NuScenesSparse4DAdaptor'),\n dict(\n type='Collect',\n keys=['img', 'timestamp', 'projection_mat', 'image_wh'],\n meta_keys=['T_global', 'T_global_inv', 'timestamp'])\n ],\n data_aug_conf=dict(\n resize_lim=(0.4, 0.47),\n final_dim=(256, 704),\n bot_pct_lim=(0.0, 0.0),\n rot_lim=(-5.4, 5.4),\n H=900,\n W=1600,\n rand_flip=True,\n rot3d_range=[-0.3925, 0.3925]),\n test_mode=True,\n tracking=True,\n tracking_threshold=0.2),\n test=dict(\n type='NuScenes3DDetTrackDataset',\n data_root='data/nuscenes/',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ],\n modality=dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False),\n version='v1.0-trainval',\n ann_file='data/nuscenes_anno_pkls/nuscenes_infos_val.pkl',\n pipeline=[\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(type='ResizeCropFlipImage'),\n dict(\n type='NormalizeMultiviewImage',\n mean=[123.675, 116.28, 103.53],\n std=[58.395, 57.12, 57.375],\n to_rgb=True),\n dict(type='NuScenesSparse4DAdaptor'),\n dict(\n type='Collect',\n keys=['img', 'timestamp', 'projection_mat', 'image_wh'],\n meta_keys=['T_global', 'T_global_inv', 'timestamp'])\n ],\n data_aug_conf=dict(\n resize_lim=(0.4, 0.47),\n final_dim=(256, 704),\n bot_pct_lim=(0.0, 0.0),\n rot_lim=(-5.4, 5.4),\n H=900,\n W=1600,\n rand_flip=True,\n rot3d_range=[-0.3925, 0.3925]),\n test_mode=True,\n tracking=True,\n tracking_threshold=0.2))\noptimizer = dict(\n type='AdamW',\n lr=0.0006,\n weight_decay=0.001,\n paramwise_cfg=dict(custom_keys=dict(img_backbone=dict(lr_mult=0.5))))\noptimizer_config = dict(grad_clip=dict(max_norm=25, norm_type=2))\nlr_config = dict(\n policy='CosineAnnealing',\n warmup='linear',\n warmup_iters=500,\n warmup_ratio=0.3333333333333333,\n min_lr_ratio=0.001)\nrunner = dict(type='IterBasedRunner', max_iters=17500)\nvis_pipeline = [\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(type='Collect', keys=['img'], meta_keys=['timestamp', 'lidar2img'])\n]\nevaluation = dict(\n interval=3500,\n pipeline=[\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(\n type='Collect', keys=['img'], meta_keys=['timestamp', 'lidar2img'])\n ])\ngpu_ids = range(0, 1)\n", "seed": 0, "exp_name": "sparse4dv3_temporal_r50_1x8_bs6_256x704.py"} diff --git a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20260210_153448.log b/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20260210_153448.log deleted file mode 100644 index 39cc811b5fac2737d977515890078db1a71391ad..0000000000000000000000000000000000000000 --- a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20260210_153448.log +++ /dev/null @@ -1,3221 +0,0 @@ -2026-02-10 15:34:48,802 - mmdet - INFO - Environment info: ------------------------------------------------------------- -sys.platform: linux -Python: 3.10.12 (main, Aug 15 2025, 14:32:43) [GCC 11.4.0] -CUDA available: True -GPU 0,1,2,3,4,5,6,7: BW200_LC, UBB BW1000_LC -CUDA_HOME: /opt/dtk -NVCC: Not Available -GCC: x86_64-linux-gnu-gcc (Ubuntu 11.4.0-1ubuntu1~22.04) 11.4.0 -PyTorch: 2.5.1 -PyTorch compiling details: PyTorch built with: - - GCC 10.3 - - C++ Version: 201703 - - Intel(R) Math Kernel Library Version 2020.0.4 Product Build 20200917 for Intel(R) 64 architecture applications - - OpenMP 201511 (a.k.a. OpenMP 4.5) - - LAPACK is enabled (usually provided by MKL) - - NNPACK is enabled - - CPU capability usage: AVX2 - - HIP Runtime 6.3.25521 - - MIOpen 2.18.0 - - Magma 2.8.0 - - Build settings: BLAS_INFO=mkl, BUILD_TYPE=Release, CXX_COMPILER=/opt/rh/gcc-toolset-10/root/usr/bin/c++, CXX_FLAGS= -D_GLIBCXX_USE_CXX11_ABI=1 -fvisibility-inlines-hidden -DUSE_PTHREADPOOL -DNDEBUG -DUSE_KINETO -DLIBKINETO_NOCUPTI -DLIBKINETO_NOXPUPTI=ON -DUSE_FBGEMM -DUSE_PYTORCH_QNNPACK -DUSE_XNNPACK -DSYMBOLICATE_MOBILE_DEBUG_HANDLE -O2 -fPIC -Wall -Wextra -Werror=return-type -Werror=non-virtual-dtor -Werror=bool-operation -Wnarrowing -Wno-missing-field-initializers -Wno-type-limits -Wno-array-bounds -Wno-unknown-pragmas -Wno-unused-parameter -Wno-strict-overflow -Wno-strict-aliasing -Wno-stringop-overflow -Wsuggest-override -Wno-psabi -Wno-error=old-style-cast -Wno-missing-braces -fdiagnostics-color=always -faligned-new -Wno-unused-but-set-variable -Wno-maybe-uninitialized -fno-math-errno -fno-trapping-math -Werror=format -Wno-stringop-overflow, FORCE_FALLBACK_CUDA_MPI=1, LAPACK_INFO=mkl, PERF_WITH_AVX=1, PERF_WITH_AVX2=1, TORCH_VERSION=2.5.1, USE_CUDA=0, USE_CUDNN=OFF, USE_CUSPARSELT=OFF, USE_EXCEPTION_PTR=1, USE_GFLAGS=1, USE_GLOG=1, USE_GLOO=1, USE_MKL=ON, USE_MKLDNN=0, USE_MPI=1, USE_NCCL=1, USE_NNPACK=ON, USE_OPENMP=1, USE_ROCM=ON, USE_ROCM_KERNEL_ASSERT=OFF, - -TorchVision: 0.20.1 -OpenCV: 4.12.0 -MMCV: 1.6.1 -MMCV Compiler: GCC 10.3 -MMCV CUDA Compiler: rocm not available -MMDetection: 2.25.1+ ------------------------------------------------------------- - -2026-02-10 15:34:49,801 - mmdet - INFO - Distributed training: True -2026-02-10 15:34:50,763 - mmdet - INFO - Config: -plugin = True -plugin_dir = 'projects/mmdet3d_plugin/' -dist_params = dict(backend='nccl') -log_level = 'INFO' -work_dir = './work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704' -total_batch_size = 160 -num_gpus = 8 -batch_size = 20 -num_iters_per_epoch = 175 -num_epochs = 100 -checkpoint_epoch_interval = 20 -checkpoint_config = dict(interval=3500) -log_config = dict( - interval=1, - hooks=[ - dict(type='TextLoggerHook', by_epoch=False), - dict(type='TensorboardLoggerHook') - ]) -load_from = None -resume_from = None -workflow = [('train', 1)] -fp16 = dict(loss_scale=32.0) -input_shape = (704, 256) -tracking_test = True -tracking_threshold = 0.2 -class_names = [ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier', - 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' -] -num_classes = 10 -embed_dims = 256 -num_groups = 8 -num_decoder = 6 -num_single_frame_decoder = 1 -use_deformable_func = True -strides = [4, 8, 16, 32] -num_levels = 4 -num_depth_layers = 3 -drop_out = 0.1 -temporal = True -decouple_attn = True -with_quality_estimation = True -model = dict( - type='Sparse4D', - use_grid_mask=True, - use_deformable_func=True, - img_backbone=dict( - type='ResNet', - depth=50, - num_stages=4, - frozen_stages=-1, - norm_eval=False, - style='pytorch', - with_cp=True, - out_indices=(0, 1, 2, 3), - norm_cfg=dict(type='BN', requires_grad=True), - pretrained='ckpt/resnet50-19c8e357.pth'), - img_neck=dict( - type='FPN', - num_outs=4, - start_level=0, - out_channels=256, - add_extra_convs='on_output', - relu_before_extra_convs=True, - in_channels=[256, 512, 1024, 2048]), - depth_branch=dict( - type='DenseDepthNet', - embed_dims=256, - num_depth_layers=3, - loss_weight=0.2), - head=dict( - type='Sparse4DHead', - cls_threshold_to_reg=0.05, - decouple_attn=True, - instance_bank=dict( - type='InstanceBank', - num_anchor=900, - embed_dims=256, - anchor='nuscenes_kmeans900.npy', - anchor_handler=dict(type='SparseBox3DKeyPointsGenerator'), - num_temp_instances=600, - confidence_decay=0.6, - feat_grad=False), - anchor_encoder=dict( - type='SparseBox3DEncoder', - vel_dims=3, - embed_dims=[128, 32, 32, 64], - mode='cat', - output_fc=False, - in_loops=1, - out_loops=4), - num_single_frame_decoder=1, - operation_order=[ - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine' - ], - temp_graph_model=dict( - type='MultiheadAttention', - embed_dims=512, - num_heads=8, - batch_first=True, - dropout=0.1), - graph_model=dict( - type='MultiheadAttention', - embed_dims=512, - num_heads=8, - batch_first=True, - dropout=0.1), - norm_layer=dict(type='LN', normalized_shape=256), - ffn=dict( - type='AsymmetricFFN', - in_channels=512, - pre_norm=dict(type='LN'), - embed_dims=256, - feedforward_channels=1024, - num_fcs=2, - ffn_drop=0.1, - act_cfg=dict(type='ReLU', inplace=True)), - deformable_model=dict( - type='DeformableFeatureAggregation', - embed_dims=256, - num_groups=8, - num_levels=4, - num_cams=6, - attn_drop=0.15, - use_deformable_func=True, - use_camera_embed=True, - residual_mode='cat', - kps_generator=dict( - type='SparseBox3DKeyPointsGenerator', - num_learnable_pts=6, - fix_scale=[[0, 0, 0], [0.45, 0, 0], [-0.45, 0, 0], - [0, 0.45, 0], [0, -0.45, 0], [0, 0, 0.45], - [0, 0, -0.45]])), - refine_layer=dict( - type='SparseBox3DRefinementModule', - embed_dims=256, - num_cls=10, - refine_yaw=True, - with_quality_estimation=True), - sampler=dict( - type='SparseBox3DTarget', - num_dn_groups=5, - num_temp_dn_groups=3, - dn_noise_scale=[2.0, 2.0, 2.0, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5], - max_dn_gt=32, - add_neg_dn=True, - cls_weight=2.0, - box_weight=0.25, - reg_weights=[2.0, 2.0, 2.0, 0.5, 0.5, 0.5, 0.0, 0.0, 0.0, 0.0], - cls_wise_reg_weights=dict( - {9: [2.0, 2.0, 2.0, 1.0, 1.0, 1.0, 0.0, 0.0, 1.0, 1.0]})), - loss_cls=dict( - type='FocalLoss', - use_sigmoid=True, - gamma=2.0, - alpha=0.25, - loss_weight=2.0), - loss_reg=dict( - type='SparseBox3DLoss', - loss_box=dict(type='L1Loss', loss_weight=0.25), - loss_centerness=dict(type='CrossEntropyLoss', use_sigmoid=True), - loss_yawness=dict(type='GaussianFocalLoss'), - cls_allow_reverse=[5]), - decoder=dict(type='SparseBox3DDecoder'), - reg_weights=[2.0, 2.0, 2.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0])) -dataset_type = 'NuScenes3DDetTrackDataset' -data_root = 'data/nuscenes/' -anno_root = 'data/nuscenes_anno_pkls/' -file_client_args = dict(backend='disk') -img_norm_cfg = dict( - mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375], to_rgb=True) -train_pipeline = [ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict( - type='LoadPointsFromFile', - coord_type='LIDAR', - load_dim=5, - use_dim=5, - file_client_args=dict(backend='disk')), - dict(type='ResizeCropFlipImage'), - dict(type='MultiScaleDepthMapGenerator', downsample=[4, 8, 16]), - dict(type='BBoxRotation'), - dict(type='PhotoMetricDistortionMultiViewImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict( - type='CircleObjectRangeFilter', - class_dist_thred=[55, 55, 55, 55, 55, 55, 55, 55, 55, 55]), - dict( - type='InstanceNameFilter', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ]), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=[ - 'img', 'timestamp', 'projection_mat', 'image_wh', 'gt_depth', - 'focal', 'gt_bboxes_3d', 'gt_labels_3d' - ], - meta_keys=['T_global', 'T_global_inv', 'timestamp', 'instance_id']) -] -test_pipeline = [ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='ResizeCropFlipImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=['img', 'timestamp', 'projection_mat', 'image_wh'], - meta_keys=['T_global', 'T_global_inv', 'timestamp']) -] -input_modality = dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False) -data_basic_config = dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier', - 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval') -data_aug_conf = dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]) -data = dict( - samples_per_gpu=20, - workers_per_gpu=20, - train=dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval', - ann_file='data/nuscenes_anno_pkls/nuscenes_infos_train.pkl', - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict( - type='LoadPointsFromFile', - coord_type='LIDAR', - load_dim=5, - use_dim=5, - file_client_args=dict(backend='disk')), - dict(type='ResizeCropFlipImage'), - dict(type='MultiScaleDepthMapGenerator', downsample=[4, 8, 16]), - dict(type='BBoxRotation'), - dict(type='PhotoMetricDistortionMultiViewImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict( - type='CircleObjectRangeFilter', - class_dist_thred=[55, 55, 55, 55, 55, 55, 55, 55, 55, 55]), - dict( - type='InstanceNameFilter', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', - 'traffic_cone' - ]), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=[ - 'img', 'timestamp', 'projection_mat', 'image_wh', - 'gt_depth', 'focal', 'gt_bboxes_3d', 'gt_labels_3d' - ], - meta_keys=[ - 'T_global', 'T_global_inv', 'timestamp', 'instance_id' - ]) - ], - test_mode=False, - data_aug_conf=dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]), - with_seq_flag=True, - sequences_split_num=2, - keep_consistent_seq_aug=True), - val=dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval', - ann_file='data/nuscenes_anno_pkls/nuscenes_infos_val.pkl', - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='ResizeCropFlipImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=['img', 'timestamp', 'projection_mat', 'image_wh'], - meta_keys=['T_global', 'T_global_inv', 'timestamp']) - ], - data_aug_conf=dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]), - test_mode=True, - tracking=True, - tracking_threshold=0.2), - test=dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval', - ann_file='data/nuscenes_anno_pkls/nuscenes_infos_val.pkl', - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='ResizeCropFlipImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=['img', 'timestamp', 'projection_mat', 'image_wh'], - meta_keys=['T_global', 'T_global_inv', 'timestamp']) - ], - data_aug_conf=dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]), - test_mode=True, - tracking=True, - tracking_threshold=0.2)) -optimizer = dict( - type='AdamW', - lr=0.0006, - weight_decay=0.001, - paramwise_cfg=dict(custom_keys=dict(img_backbone=dict(lr_mult=0.5)))) -optimizer_config = dict(grad_clip=dict(max_norm=25, norm_type=2)) -lr_config = dict( - policy='CosineAnnealing', - warmup='linear', - warmup_iters=500, - warmup_ratio=0.3333333333333333, - min_lr_ratio=0.001) -runner = dict(type='IterBasedRunner', max_iters=17500) -vis_pipeline = [ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='Collect', keys=['img'], meta_keys=['timestamp', 'lidar2img']) -] -evaluation = dict( - interval=3500, - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict( - type='Collect', keys=['img'], meta_keys=['timestamp', 'lidar2img']) - ]) -gpu_ids = range(0, 8) - -2026-02-10 15:34:50,763 - mmdet - INFO - Set random seed to 0, deterministic: False -2026-02-10 15:34:51,165 - mmdet - INFO - initialize ResNet with init_cfg {'type': 'Pretrained', 'checkpoint': 'ckpt/resnet50-19c8e357.pth'} -2026-02-10 15:34:51,701 - mmdet - INFO - initialize FPN with init_cfg {'type': 'Xavier', 'layer': 'Conv2d', 'distribution': 'uniform'} -Name of parameter - Initialization information - -img_backbone.conv1.weight - torch.Size([64, 3, 7, 7]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.bn1.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.bn1.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.conv1.weight - torch.Size([64, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn1.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn1.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.conv2.weight - torch.Size([64, 64, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn2.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn2.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.conv3.weight - torch.Size([256, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn3.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn3.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.downsample.0.weight - torch.Size([256, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.downsample.1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.downsample.1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.conv1.weight - torch.Size([64, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn1.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn1.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.conv2.weight - torch.Size([64, 64, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn2.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn2.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.conv3.weight - torch.Size([256, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn3.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn3.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.conv1.weight - torch.Size([64, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn1.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn1.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.conv2.weight - torch.Size([64, 64, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn2.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn2.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.conv3.weight - torch.Size([256, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn3.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn3.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.conv1.weight - torch.Size([128, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn1.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn1.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.conv2.weight - torch.Size([128, 128, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn2.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn2.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.conv3.weight - torch.Size([512, 128, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn3.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn3.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.downsample.0.weight - torch.Size([512, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.downsample.1.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.downsample.1.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.conv1.weight - torch.Size([128, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn1.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn1.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.conv2.weight - torch.Size([128, 128, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn2.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn2.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.conv3.weight - torch.Size([512, 128, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn3.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn3.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.conv1.weight - torch.Size([128, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn1.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn1.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.conv2.weight - torch.Size([128, 128, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn2.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn2.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.conv3.weight - torch.Size([512, 128, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn3.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn3.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.conv1.weight - torch.Size([128, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn1.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn1.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.conv2.weight - torch.Size([128, 128, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn2.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn2.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.conv3.weight - torch.Size([512, 128, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn3.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn3.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.conv1.weight - torch.Size([256, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.downsample.0.weight - torch.Size([1024, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.downsample.1.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.downsample.1.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.conv1.weight - torch.Size([512, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn1.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn1.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.conv2.weight - torch.Size([512, 512, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn2.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn2.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.conv3.weight - torch.Size([2048, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn3.weight - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn3.bias - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.downsample.0.weight - torch.Size([2048, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.downsample.1.weight - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.downsample.1.bias - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.conv1.weight - torch.Size([512, 2048, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn1.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn1.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.conv2.weight - torch.Size([512, 512, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn2.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn2.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.conv3.weight - torch.Size([2048, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn3.weight - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn3.bias - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.conv1.weight - torch.Size([512, 2048, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn1.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn1.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.conv2.weight - torch.Size([512, 512, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn2.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn2.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.conv3.weight - torch.Size([2048, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn3.weight - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn3.bias - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_neck.lateral_convs.0.conv.weight - torch.Size([256, 256, 1, 1]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.lateral_convs.0.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.lateral_convs.1.conv.weight - torch.Size([256, 512, 1, 1]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.lateral_convs.1.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.lateral_convs.2.conv.weight - torch.Size([256, 1024, 1, 1]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.lateral_convs.2.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.lateral_convs.3.conv.weight - torch.Size([256, 2048, 1, 1]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.lateral_convs.3.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.fpn_convs.0.conv.weight - torch.Size([256, 256, 3, 3]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.fpn_convs.0.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.fpn_convs.1.conv.weight - torch.Size([256, 256, 3, 3]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.fpn_convs.1.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.fpn_convs.2.conv.weight - torch.Size([256, 256, 3, 3]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.fpn_convs.2.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.fpn_convs.3.conv.weight - torch.Size([256, 256, 3, 3]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.fpn_convs.3.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.instance_bank.anchor - torch.Size([900, 11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.instance_bank.instance_feature - torch.Size([900, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.instance_bank.anchor_handler.fix_scale - torch.Size([1, 3]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.0.weight - torch.Size([128, 3]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.0.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.2.weight - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.2.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.3.weight - torch.Size([128, 128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.3.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.5.weight - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.5.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.6.weight - torch.Size([128, 128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.6.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.8.weight - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.8.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.9.weight - torch.Size([128, 128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.9.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.11.weight - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.11.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.0.weight - torch.Size([32, 3]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.0.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.2.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.2.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.3.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.3.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.5.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.5.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.6.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.6.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.8.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.8.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.9.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.9.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.11.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.11.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.0.weight - torch.Size([32, 2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.0.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.2.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.2.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.3.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.3.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.5.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.5.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.6.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.6.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.8.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.8.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.9.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.9.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.11.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.11.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.0.weight - torch.Size([64, 3]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.0.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.2.weight - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.2.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.3.weight - torch.Size([64, 64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.3.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.5.weight - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.5.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.6.weight - torch.Size([64, 64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.6.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.8.weight - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.8.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.9.weight - torch.Size([64, 64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.9.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.11.weight - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.11.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.1.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.1.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.1.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.1.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.1.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.1.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.1.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.1.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.3.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.4.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.4.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.4.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.4.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.5.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.5.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.5.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.5.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.6.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.6.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.8.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.8.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.8.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.8.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.8.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.8.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.8.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.8.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.10.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.11.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.11.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.11.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.11.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.12.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.12.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.12.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.12.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.13.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.13.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.15.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.15.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.15.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.15.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.15.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.15.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.15.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.15.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.16.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.16.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.17.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.18.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.18.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.18.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.18.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.19.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.19.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.19.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.19.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.20.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.20.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.22.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.22.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.22.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.22.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.22.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.22.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.22.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.22.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.23.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.23.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.24.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.25.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.25.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.25.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.25.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.26.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.26.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.26.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.26.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.27.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.27.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.29.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.29.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.29.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.29.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.29.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.29.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.29.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.29.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.30.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.30.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.31.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.32.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.32.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.32.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.32.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.33.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.33.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.33.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.33.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.34.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.34.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.36.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.36.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.36.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.36.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.36.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.36.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.36.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.36.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.37.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.37.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.38.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.fc_before.weight - torch.Size([512, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.fc_after.weight - torch.Size([256, 512]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.0.weight - torch.Size([1, 256, 1, 1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.0.bias - torch.Size([1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.1.weight - torch.Size([1, 256, 1, 1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.1.bias - torch.Size([1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.2.weight - torch.Size([1, 256, 1, 1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.2.bias - torch.Size([1]): -The value is the same before and after calling `init_weights` of Sparse4D -2026-02-10 15:34:51,832 - mmdet - INFO - Model: -Sparse4D( - (img_backbone): ResNet( - (conv1): Conv2d(3, 64, kernel_size=(7, 7), stride=(2, 2), padding=(3, 3), bias=False) - (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (maxpool): MaxPool2d(kernel_size=3, stride=2, padding=1, dilation=1, ceil_mode=False) - (layer1): ResLayer( - (0): Bottleneck( - (conv1): Conv2d(64, 64, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (downsample): Sequential( - (0): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - ) - ) - (1): Bottleneck( - (conv1): Conv2d(256, 64, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (2): Bottleneck( - (conv1): Conv2d(256, 64, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - ) - (layer2): ResLayer( - (0): Bottleneck( - (conv1): Conv2d(256, 128, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (downsample): Sequential( - (0): Conv2d(256, 512, kernel_size=(1, 1), stride=(2, 2), bias=False) - (1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - ) - ) - (1): Bottleneck( - (conv1): Conv2d(512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (2): Bottleneck( - (conv1): Conv2d(512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (3): Bottleneck( - (conv1): Conv2d(512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - ) - (layer3): ResLayer( - (0): Bottleneck( - (conv1): Conv2d(512, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (downsample): Sequential( - (0): Conv2d(512, 1024, kernel_size=(1, 1), stride=(2, 2), bias=False) - (1): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - ) - ) - (1): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (2): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (3): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (4): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (5): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - ) - (layer4): ResLayer( - (0): Bottleneck( - (conv1): Conv2d(1024, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(512, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (downsample): Sequential( - (0): Conv2d(1024, 2048, kernel_size=(1, 1), stride=(2, 2), bias=False) - (1): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - ) - ) - (1): Bottleneck( - (conv1): Conv2d(2048, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(512, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (2): Bottleneck( - (conv1): Conv2d(2048, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(512, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - ) - ) - init_cfg={'type': 'Pretrained', 'checkpoint': 'ckpt/resnet50-19c8e357.pth'} - (img_neck): FPN( - (lateral_convs): ModuleList( - (0): ConvModule( - (conv): Conv2d(256, 256, kernel_size=(1, 1), stride=(1, 1)) - ) - (1): ConvModule( - (conv): Conv2d(512, 256, kernel_size=(1, 1), stride=(1, 1)) - ) - (2): ConvModule( - (conv): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1)) - ) - (3): ConvModule( - (conv): Conv2d(2048, 256, kernel_size=(1, 1), stride=(1, 1)) - ) - ) - (fpn_convs): ModuleList( - (0-3): 4 x ConvModule( - (conv): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1)) - ) - ) - ) - init_cfg={'type': 'Xavier', 'layer': 'Conv2d', 'distribution': 'uniform'} - (head): Sparse4DHead( - (instance_bank): InstanceBank( - (anchor_handler): SparseBox3DKeyPointsGenerator() - ) - (anchor_encoder): SparseBox3DEncoder( - (pos_fc): Sequential( - (0): Linear(in_features=3, out_features=128, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((128,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=128, out_features=128, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((128,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=128, out_features=128, bias=True) - (7): ReLU(inplace=True) - (8): LayerNorm((128,), eps=1e-05, elementwise_affine=True) - (9): Linear(in_features=128, out_features=128, bias=True) - (10): ReLU(inplace=True) - (11): LayerNorm((128,), eps=1e-05, elementwise_affine=True) - ) - (size_fc): Sequential( - (0): Linear(in_features=3, out_features=32, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=32, out_features=32, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=32, out_features=32, bias=True) - (7): ReLU(inplace=True) - (8): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (9): Linear(in_features=32, out_features=32, bias=True) - (10): ReLU(inplace=True) - (11): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - ) - (yaw_fc): Sequential( - (0): Linear(in_features=2, out_features=32, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=32, out_features=32, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=32, out_features=32, bias=True) - (7): ReLU(inplace=True) - (8): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (9): Linear(in_features=32, out_features=32, bias=True) - (10): ReLU(inplace=True) - (11): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - ) - (vel_fc): Sequential( - (0): Linear(in_features=3, out_features=64, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((64,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=64, out_features=64, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((64,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=64, out_features=64, bias=True) - (7): ReLU(inplace=True) - (8): LayerNorm((64,), eps=1e-05, elementwise_affine=True) - (9): Linear(in_features=64, out_features=64, bias=True) - (10): ReLU(inplace=True) - (11): LayerNorm((64,), eps=1e-05, elementwise_affine=True) - ) - ) - (loss_cls): FocalLoss() - (loss_reg): SparseBox3DLoss( - (loss_box): L1Loss() - (loss_cns): CrossEntropyLoss(avg_non_ignore=False) - (loss_yns): GaussianFocalLoss() - ) - (layers): ModuleList( - (0): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (1): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (4-5): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (6): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (7): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (8): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (11-12): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (13): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (14): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (15): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (16): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (17): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (18-19): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (20): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (21): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (22): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (23): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (24): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (25-26): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (27): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (28): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (29): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (30): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (31): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (32-33): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (34): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (35): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (36): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (37): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (38): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - ) - (fc_before): Linear(in_features=256, out_features=512, bias=False) - (fc_after): Linear(in_features=512, out_features=256, bias=False) - ) - (depth_branch): DenseDepthNet( - (depth_layers): ModuleList( - (0-2): 3 x Conv2d(256, 1, kernel_size=(1, 1), stride=(1, 1)) - ) - ) - (grid_mask): GridMask() -) -2026-02-10 15:35:07,329 - mmdet - INFO - Start running, host: root@bw28, work_dir: /workspace/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704 -2026-02-10 15:35:07,329 - mmdet - INFO - Hooks will be executed in the following order: -before_run: -(VERY_HIGH ) CosineAnnealingLrUpdaterHook -(ABOVE_NORMAL) Fp16OptimizerHook -(NORMAL ) CheckpointHook -(NORMAL ) CustomDistEvalHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -before_train_epoch: -(VERY_HIGH ) CosineAnnealingLrUpdaterHook -(NORMAL ) CustomDistEvalHook -(LOW ) IterTimerHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -before_train_iter: -(VERY_HIGH ) CosineAnnealingLrUpdaterHook -(NORMAL ) CustomDistEvalHook -(LOW ) IterTimerHook - -------------------- -after_train_iter: -(ABOVE_NORMAL) Fp16OptimizerHook -(NORMAL ) CheckpointHook -(NORMAL ) CustomDistEvalHook -(NORMAL ) ProfilerHook -(LOW ) IterTimerHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -after_train_epoch: -(NORMAL ) CheckpointHook -(NORMAL ) CustomDistEvalHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -before_val_epoch: -(LOW ) IterTimerHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -before_val_iter: -(LOW ) IterTimerHook - -------------------- -after_val_iter: -(LOW ) IterTimerHook - -------------------- -after_val_epoch: -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -after_run: -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -2026-02-10 15:35:07,330 - mmdet - INFO - workflow: [('train', 1)], max: 17500 iters -2026-02-10 15:35:07,332 - mmdet - INFO - Checkpoints will be saved to /workspace/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704 by HardDiskBackend. diff --git a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20260210_153448.log.json b/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20260210_153448.log.json deleted file mode 100644 index 9b7a36a4c4faf65b6fbf808cfd0607ffa5984a6f..0000000000000000000000000000000000000000 --- a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20260210_153448.log.json +++ /dev/null @@ -1 +0,0 @@ -{"env_info": "sys.platform: linux\nPython: 3.10.12 (main, Aug 15 2025, 14:32:43) [GCC 11.4.0]\nCUDA available: True\nGPU 0,1,2,3,4,5,6,7: BW200_LC, UBB BW1000_LC\nCUDA_HOME: /opt/dtk\nNVCC: Not Available\nGCC: x86_64-linux-gnu-gcc (Ubuntu 11.4.0-1ubuntu1~22.04) 11.4.0\nPyTorch: 2.5.1\nPyTorch compiling details: PyTorch built with:\n - GCC 10.3\n - C++ Version: 201703\n - Intel(R) Math Kernel Library Version 2020.0.4 Product Build 20200917 for Intel(R) 64 architecture applications\n - OpenMP 201511 (a.k.a. OpenMP 4.5)\n - LAPACK is enabled (usually provided by MKL)\n - NNPACK is enabled\n - CPU capability usage: AVX2\n - HIP Runtime 6.3.25521\n - MIOpen 2.18.0\n - Magma 2.8.0\n - Build settings: BLAS_INFO=mkl, BUILD_TYPE=Release, CXX_COMPILER=/opt/rh/gcc-toolset-10/root/usr/bin/c++, CXX_FLAGS= -D_GLIBCXX_USE_CXX11_ABI=1 -fvisibility-inlines-hidden -DUSE_PTHREADPOOL -DNDEBUG -DUSE_KINETO -DLIBKINETO_NOCUPTI -DLIBKINETO_NOXPUPTI=ON -DUSE_FBGEMM -DUSE_PYTORCH_QNNPACK -DUSE_XNNPACK -DSYMBOLICATE_MOBILE_DEBUG_HANDLE -O2 -fPIC -Wall -Wextra -Werror=return-type -Werror=non-virtual-dtor -Werror=bool-operation -Wnarrowing -Wno-missing-field-initializers -Wno-type-limits -Wno-array-bounds -Wno-unknown-pragmas -Wno-unused-parameter -Wno-strict-overflow -Wno-strict-aliasing -Wno-stringop-overflow -Wsuggest-override -Wno-psabi -Wno-error=old-style-cast -Wno-missing-braces -fdiagnostics-color=always -faligned-new -Wno-unused-but-set-variable -Wno-maybe-uninitialized -fno-math-errno -fno-trapping-math -Werror=format -Wno-stringop-overflow, FORCE_FALLBACK_CUDA_MPI=1, LAPACK_INFO=mkl, PERF_WITH_AVX=1, PERF_WITH_AVX2=1, TORCH_VERSION=2.5.1, USE_CUDA=0, USE_CUDNN=OFF, USE_CUSPARSELT=OFF, USE_EXCEPTION_PTR=1, USE_GFLAGS=1, USE_GLOG=1, USE_GLOO=1, USE_MKL=ON, USE_MKLDNN=0, USE_MPI=1, USE_NCCL=1, USE_NNPACK=ON, USE_OPENMP=1, USE_ROCM=ON, USE_ROCM_KERNEL_ASSERT=OFF, \n\nTorchVision: 0.20.1\nOpenCV: 4.12.0\nMMCV: 1.6.1\nMMCV Compiler: GCC 10.3\nMMCV CUDA Compiler: rocm not available\nMMDetection: 2.25.1+", "config": "plugin = True\nplugin_dir = 'projects/mmdet3d_plugin/'\ndist_params = dict(backend='nccl')\nlog_level = 'INFO'\nwork_dir = './work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704'\ntotal_batch_size = 160\nnum_gpus = 8\nbatch_size = 20\nnum_iters_per_epoch = 175\nnum_epochs = 100\ncheckpoint_epoch_interval = 20\ncheckpoint_config = dict(interval=3500)\nlog_config = dict(\n interval=1,\n hooks=[\n dict(type='TextLoggerHook', by_epoch=False),\n dict(type='TensorboardLoggerHook')\n ])\nload_from = None\nresume_from = None\nworkflow = [('train', 1)]\nfp16 = dict(loss_scale=32.0)\ninput_shape = (704, 256)\ntracking_test = True\ntracking_threshold = 0.2\nclass_names = [\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier',\n 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n]\nnum_classes = 10\nembed_dims = 256\nnum_groups = 8\nnum_decoder = 6\nnum_single_frame_decoder = 1\nuse_deformable_func = True\nstrides = [4, 8, 16, 32]\nnum_levels = 4\nnum_depth_layers = 3\ndrop_out = 0.1\ntemporal = True\ndecouple_attn = True\nwith_quality_estimation = True\nmodel = dict(\n type='Sparse4D',\n use_grid_mask=True,\n use_deformable_func=True,\n img_backbone=dict(\n type='ResNet',\n depth=50,\n num_stages=4,\n frozen_stages=-1,\n norm_eval=False,\n style='pytorch',\n with_cp=True,\n out_indices=(0, 1, 2, 3),\n norm_cfg=dict(type='BN', requires_grad=True),\n pretrained='ckpt/resnet50-19c8e357.pth'),\n img_neck=dict(\n type='FPN',\n num_outs=4,\n start_level=0,\n out_channels=256,\n add_extra_convs='on_output',\n relu_before_extra_convs=True,\n in_channels=[256, 512, 1024, 2048]),\n depth_branch=dict(\n type='DenseDepthNet',\n embed_dims=256,\n num_depth_layers=3,\n loss_weight=0.2),\n head=dict(\n type='Sparse4DHead',\n cls_threshold_to_reg=0.05,\n decouple_attn=True,\n instance_bank=dict(\n type='InstanceBank',\n num_anchor=900,\n embed_dims=256,\n anchor='nuscenes_kmeans900.npy',\n anchor_handler=dict(type='SparseBox3DKeyPointsGenerator'),\n num_temp_instances=600,\n confidence_decay=0.6,\n feat_grad=False),\n anchor_encoder=dict(\n type='SparseBox3DEncoder',\n vel_dims=3,\n embed_dims=[128, 32, 32, 64],\n mode='cat',\n output_fc=False,\n in_loops=1,\n out_loops=4),\n num_single_frame_decoder=1,\n operation_order=[\n 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm',\n 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm',\n 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm',\n 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm',\n 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm',\n 'deformable', 'ffn', 'norm', 'refine'\n ],\n temp_graph_model=dict(\n type='MultiheadAttention',\n embed_dims=512,\n num_heads=8,\n batch_first=True,\n dropout=0.1),\n graph_model=dict(\n type='MultiheadAttention',\n embed_dims=512,\n num_heads=8,\n batch_first=True,\n dropout=0.1),\n norm_layer=dict(type='LN', normalized_shape=256),\n ffn=dict(\n type='AsymmetricFFN',\n in_channels=512,\n pre_norm=dict(type='LN'),\n embed_dims=256,\n feedforward_channels=1024,\n num_fcs=2,\n ffn_drop=0.1,\n act_cfg=dict(type='ReLU', inplace=True)),\n deformable_model=dict(\n type='DeformableFeatureAggregation',\n embed_dims=256,\n num_groups=8,\n num_levels=4,\n num_cams=6,\n attn_drop=0.15,\n use_deformable_func=True,\n use_camera_embed=True,\n residual_mode='cat',\n kps_generator=dict(\n type='SparseBox3DKeyPointsGenerator',\n num_learnable_pts=6,\n fix_scale=[[0, 0, 0], [0.45, 0, 0], [-0.45, 0, 0],\n [0, 0.45, 0], [0, -0.45, 0], [0, 0, 0.45],\n [0, 0, -0.45]])),\n refine_layer=dict(\n type='SparseBox3DRefinementModule',\n embed_dims=256,\n num_cls=10,\n refine_yaw=True,\n with_quality_estimation=True),\n sampler=dict(\n type='SparseBox3DTarget',\n num_dn_groups=5,\n num_temp_dn_groups=3,\n dn_noise_scale=[2.0, 2.0, 2.0, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5],\n max_dn_gt=32,\n add_neg_dn=True,\n cls_weight=2.0,\n box_weight=0.25,\n reg_weights=[2.0, 2.0, 2.0, 0.5, 0.5, 0.5, 0.0, 0.0, 0.0, 0.0],\n cls_wise_reg_weights=dict(\n {9: [2.0, 2.0, 2.0, 1.0, 1.0, 1.0, 0.0, 0.0, 1.0, 1.0]})),\n loss_cls=dict(\n type='FocalLoss',\n use_sigmoid=True,\n gamma=2.0,\n alpha=0.25,\n loss_weight=2.0),\n loss_reg=dict(\n type='SparseBox3DLoss',\n loss_box=dict(type='L1Loss', loss_weight=0.25),\n loss_centerness=dict(type='CrossEntropyLoss', use_sigmoid=True),\n loss_yawness=dict(type='GaussianFocalLoss'),\n cls_allow_reverse=[5]),\n decoder=dict(type='SparseBox3DDecoder'),\n reg_weights=[2.0, 2.0, 2.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0]))\ndataset_type = 'NuScenes3DDetTrackDataset'\ndata_root = 'data/nuscenes/'\nanno_root = 'data/nuscenes_anno_pkls/'\nfile_client_args = dict(backend='disk')\nimg_norm_cfg = dict(\n mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375], to_rgb=True)\ntrain_pipeline = [\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(\n type='LoadPointsFromFile',\n coord_type='LIDAR',\n load_dim=5,\n use_dim=5,\n file_client_args=dict(backend='disk')),\n dict(type='ResizeCropFlipImage'),\n dict(type='MultiScaleDepthMapGenerator', downsample=[4, 8, 16]),\n dict(type='BBoxRotation'),\n dict(type='PhotoMetricDistortionMultiViewImage'),\n dict(\n type='NormalizeMultiviewImage',\n mean=[123.675, 116.28, 103.53],\n std=[58.395, 57.12, 57.375],\n to_rgb=True),\n dict(\n type='CircleObjectRangeFilter',\n class_dist_thred=[55, 55, 55, 55, 55, 55, 55, 55, 55, 55]),\n dict(\n type='InstanceNameFilter',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ]),\n dict(type='NuScenesSparse4DAdaptor'),\n dict(\n type='Collect',\n keys=[\n 'img', 'timestamp', 'projection_mat', 'image_wh', 'gt_depth',\n 'focal', 'gt_bboxes_3d', 'gt_labels_3d'\n ],\n meta_keys=['T_global', 'T_global_inv', 'timestamp', 'instance_id'])\n]\ntest_pipeline = [\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(type='ResizeCropFlipImage'),\n dict(\n type='NormalizeMultiviewImage',\n mean=[123.675, 116.28, 103.53],\n std=[58.395, 57.12, 57.375],\n to_rgb=True),\n dict(type='NuScenesSparse4DAdaptor'),\n dict(\n type='Collect',\n keys=['img', 'timestamp', 'projection_mat', 'image_wh'],\n meta_keys=['T_global', 'T_global_inv', 'timestamp'])\n]\ninput_modality = dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False)\ndata_basic_config = dict(\n type='NuScenes3DDetTrackDataset',\n data_root='data/nuscenes/',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier',\n 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ],\n modality=dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False),\n version='v1.0-trainval')\ndata_aug_conf = dict(\n resize_lim=(0.4, 0.47),\n final_dim=(256, 704),\n bot_pct_lim=(0.0, 0.0),\n rot_lim=(-5.4, 5.4),\n H=900,\n W=1600,\n rand_flip=True,\n rot3d_range=[-0.3925, 0.3925])\ndata = dict(\n samples_per_gpu=20,\n workers_per_gpu=20,\n train=dict(\n type='NuScenes3DDetTrackDataset',\n data_root='data/nuscenes/',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ],\n modality=dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False),\n version='v1.0-trainval',\n ann_file='data/nuscenes_anno_pkls/nuscenes_infos_train.pkl',\n pipeline=[\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(\n type='LoadPointsFromFile',\n coord_type='LIDAR',\n load_dim=5,\n use_dim=5,\n file_client_args=dict(backend='disk')),\n dict(type='ResizeCropFlipImage'),\n dict(type='MultiScaleDepthMapGenerator', downsample=[4, 8, 16]),\n dict(type='BBoxRotation'),\n dict(type='PhotoMetricDistortionMultiViewImage'),\n dict(\n type='NormalizeMultiviewImage',\n mean=[123.675, 116.28, 103.53],\n std=[58.395, 57.12, 57.375],\n to_rgb=True),\n dict(\n type='CircleObjectRangeFilter',\n class_dist_thred=[55, 55, 55, 55, 55, 55, 55, 55, 55, 55]),\n dict(\n type='InstanceNameFilter',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian',\n 'traffic_cone'\n ]),\n dict(type='NuScenesSparse4DAdaptor'),\n dict(\n type='Collect',\n keys=[\n 'img', 'timestamp', 'projection_mat', 'image_wh',\n 'gt_depth', 'focal', 'gt_bboxes_3d', 'gt_labels_3d'\n ],\n meta_keys=[\n 'T_global', 'T_global_inv', 'timestamp', 'instance_id'\n ])\n ],\n test_mode=False,\n data_aug_conf=dict(\n resize_lim=(0.4, 0.47),\n final_dim=(256, 704),\n bot_pct_lim=(0.0, 0.0),\n rot_lim=(-5.4, 5.4),\n H=900,\n W=1600,\n rand_flip=True,\n rot3d_range=[-0.3925, 0.3925]),\n with_seq_flag=True,\n sequences_split_num=2,\n keep_consistent_seq_aug=True),\n val=dict(\n type='NuScenes3DDetTrackDataset',\n data_root='data/nuscenes/',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ],\n modality=dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False),\n version='v1.0-trainval',\n ann_file='data/nuscenes_anno_pkls/nuscenes_infos_val.pkl',\n pipeline=[\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(type='ResizeCropFlipImage'),\n dict(\n type='NormalizeMultiviewImage',\n mean=[123.675, 116.28, 103.53],\n std=[58.395, 57.12, 57.375],\n to_rgb=True),\n dict(type='NuScenesSparse4DAdaptor'),\n dict(\n type='Collect',\n keys=['img', 'timestamp', 'projection_mat', 'image_wh'],\n meta_keys=['T_global', 'T_global_inv', 'timestamp'])\n ],\n data_aug_conf=dict(\n resize_lim=(0.4, 0.47),\n final_dim=(256, 704),\n bot_pct_lim=(0.0, 0.0),\n rot_lim=(-5.4, 5.4),\n H=900,\n W=1600,\n rand_flip=True,\n rot3d_range=[-0.3925, 0.3925]),\n test_mode=True,\n tracking=True,\n tracking_threshold=0.2),\n test=dict(\n type='NuScenes3DDetTrackDataset',\n data_root='data/nuscenes/',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ],\n modality=dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False),\n version='v1.0-trainval',\n ann_file='data/nuscenes_anno_pkls/nuscenes_infos_val.pkl',\n pipeline=[\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(type='ResizeCropFlipImage'),\n dict(\n type='NormalizeMultiviewImage',\n mean=[123.675, 116.28, 103.53],\n std=[58.395, 57.12, 57.375],\n to_rgb=True),\n dict(type='NuScenesSparse4DAdaptor'),\n dict(\n type='Collect',\n keys=['img', 'timestamp', 'projection_mat', 'image_wh'],\n meta_keys=['T_global', 'T_global_inv', 'timestamp'])\n ],\n data_aug_conf=dict(\n resize_lim=(0.4, 0.47),\n final_dim=(256, 704),\n bot_pct_lim=(0.0, 0.0),\n rot_lim=(-5.4, 5.4),\n H=900,\n W=1600,\n rand_flip=True,\n rot3d_range=[-0.3925, 0.3925]),\n test_mode=True,\n tracking=True,\n tracking_threshold=0.2))\noptimizer = dict(\n type='AdamW',\n lr=0.0006,\n weight_decay=0.001,\n paramwise_cfg=dict(custom_keys=dict(img_backbone=dict(lr_mult=0.5))))\noptimizer_config = dict(grad_clip=dict(max_norm=25, norm_type=2))\nlr_config = dict(\n policy='CosineAnnealing',\n warmup='linear',\n warmup_iters=500,\n warmup_ratio=0.3333333333333333,\n min_lr_ratio=0.001)\nrunner = dict(type='IterBasedRunner', max_iters=17500)\nvis_pipeline = [\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(type='Collect', keys=['img'], meta_keys=['timestamp', 'lidar2img'])\n]\nevaluation = dict(\n interval=3500,\n pipeline=[\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(\n type='Collect', keys=['img'], meta_keys=['timestamp', 'lidar2img'])\n ])\ngpu_ids = range(0, 8)\n", "seed": 0, "exp_name": "sparse4dv3_temporal_r50_1x8_bs6_256x704.py"} diff --git a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20260210_153802.log b/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20260210_153802.log deleted file mode 100644 index c53c65ad6fb87c434887e2704ea86f829b886be6..0000000000000000000000000000000000000000 --- a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20260210_153802.log +++ /dev/null @@ -1,3221 +0,0 @@ -2026-02-10 15:38:02,653 - mmdet - INFO - Environment info: ------------------------------------------------------------- -sys.platform: linux -Python: 3.10.12 (main, Aug 15 2025, 14:32:43) [GCC 11.4.0] -CUDA available: True -GPU 0,1,2,3,4,5,6,7: BW200_LC, UBB BW1000_LC -CUDA_HOME: /opt/dtk -NVCC: Not Available -GCC: x86_64-linux-gnu-gcc (Ubuntu 11.4.0-1ubuntu1~22.04) 11.4.0 -PyTorch: 2.5.1 -PyTorch compiling details: PyTorch built with: - - GCC 10.3 - - C++ Version: 201703 - - Intel(R) Math Kernel Library Version 2020.0.4 Product Build 20200917 for Intel(R) 64 architecture applications - - OpenMP 201511 (a.k.a. OpenMP 4.5) - - LAPACK is enabled (usually provided by MKL) - - NNPACK is enabled - - CPU capability usage: AVX2 - - HIP Runtime 6.3.25521 - - MIOpen 2.18.0 - - Magma 2.8.0 - - Build settings: BLAS_INFO=mkl, BUILD_TYPE=Release, CXX_COMPILER=/opt/rh/gcc-toolset-10/root/usr/bin/c++, CXX_FLAGS= -D_GLIBCXX_USE_CXX11_ABI=1 -fvisibility-inlines-hidden -DUSE_PTHREADPOOL -DNDEBUG -DUSE_KINETO -DLIBKINETO_NOCUPTI -DLIBKINETO_NOXPUPTI=ON -DUSE_FBGEMM -DUSE_PYTORCH_QNNPACK -DUSE_XNNPACK -DSYMBOLICATE_MOBILE_DEBUG_HANDLE -O2 -fPIC -Wall -Wextra -Werror=return-type -Werror=non-virtual-dtor -Werror=bool-operation -Wnarrowing -Wno-missing-field-initializers -Wno-type-limits -Wno-array-bounds -Wno-unknown-pragmas -Wno-unused-parameter -Wno-strict-overflow -Wno-strict-aliasing -Wno-stringop-overflow -Wsuggest-override -Wno-psabi -Wno-error=old-style-cast -Wno-missing-braces -fdiagnostics-color=always -faligned-new -Wno-unused-but-set-variable -Wno-maybe-uninitialized -fno-math-errno -fno-trapping-math -Werror=format -Wno-stringop-overflow, FORCE_FALLBACK_CUDA_MPI=1, LAPACK_INFO=mkl, PERF_WITH_AVX=1, PERF_WITH_AVX2=1, TORCH_VERSION=2.5.1, USE_CUDA=0, USE_CUDNN=OFF, USE_CUSPARSELT=OFF, USE_EXCEPTION_PTR=1, USE_GFLAGS=1, USE_GLOG=1, USE_GLOO=1, USE_MKL=ON, USE_MKLDNN=0, USE_MPI=1, USE_NCCL=1, USE_NNPACK=ON, USE_OPENMP=1, USE_ROCM=ON, USE_ROCM_KERNEL_ASSERT=OFF, - -TorchVision: 0.20.1 -OpenCV: 4.12.0 -MMCV: 1.6.1 -MMCV Compiler: GCC 10.3 -MMCV CUDA Compiler: rocm not available -MMDetection: 2.25.1+ ------------------------------------------------------------- - -2026-02-10 15:38:03,659 - mmdet - INFO - Distributed training: False -2026-02-10 15:38:04,625 - mmdet - INFO - Config: -plugin = True -plugin_dir = 'projects/mmdet3d_plugin/' -dist_params = dict(backend='nccl') -log_level = 'INFO' -work_dir = './work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704' -total_batch_size = 160 -num_gpus = 8 -batch_size = 20 -num_iters_per_epoch = 175 -num_epochs = 100 -checkpoint_epoch_interval = 20 -checkpoint_config = dict(interval=3500) -log_config = dict( - interval=1, - hooks=[ - dict(type='TextLoggerHook', by_epoch=False), - dict(type='TensorboardLoggerHook') - ]) -load_from = None -resume_from = None -workflow = [('train', 1)] -fp16 = dict(loss_scale=32.0) -input_shape = (704, 256) -tracking_test = True -tracking_threshold = 0.2 -class_names = [ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier', - 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' -] -num_classes = 10 -embed_dims = 256 -num_groups = 8 -num_decoder = 6 -num_single_frame_decoder = 1 -use_deformable_func = True -strides = [4, 8, 16, 32] -num_levels = 4 -num_depth_layers = 3 -drop_out = 0.1 -temporal = True -decouple_attn = True -with_quality_estimation = True -model = dict( - type='Sparse4D', - use_grid_mask=True, - use_deformable_func=True, - img_backbone=dict( - type='ResNet', - depth=50, - num_stages=4, - frozen_stages=-1, - norm_eval=False, - style='pytorch', - with_cp=True, - out_indices=(0, 1, 2, 3), - norm_cfg=dict(type='BN', requires_grad=True), - pretrained='ckpt/resnet50-19c8e357.pth'), - img_neck=dict( - type='FPN', - num_outs=4, - start_level=0, - out_channels=256, - add_extra_convs='on_output', - relu_before_extra_convs=True, - in_channels=[256, 512, 1024, 2048]), - depth_branch=dict( - type='DenseDepthNet', - embed_dims=256, - num_depth_layers=3, - loss_weight=0.2), - head=dict( - type='Sparse4DHead', - cls_threshold_to_reg=0.05, - decouple_attn=True, - instance_bank=dict( - type='InstanceBank', - num_anchor=900, - embed_dims=256, - anchor='nuscenes_kmeans900.npy', - anchor_handler=dict(type='SparseBox3DKeyPointsGenerator'), - num_temp_instances=600, - confidence_decay=0.6, - feat_grad=False), - anchor_encoder=dict( - type='SparseBox3DEncoder', - vel_dims=3, - embed_dims=[128, 32, 32, 64], - mode='cat', - output_fc=False, - in_loops=1, - out_loops=4), - num_single_frame_decoder=1, - operation_order=[ - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine' - ], - temp_graph_model=dict( - type='MultiheadAttention', - embed_dims=512, - num_heads=8, - batch_first=True, - dropout=0.1), - graph_model=dict( - type='MultiheadAttention', - embed_dims=512, - num_heads=8, - batch_first=True, - dropout=0.1), - norm_layer=dict(type='LN', normalized_shape=256), - ffn=dict( - type='AsymmetricFFN', - in_channels=512, - pre_norm=dict(type='LN'), - embed_dims=256, - feedforward_channels=1024, - num_fcs=2, - ffn_drop=0.1, - act_cfg=dict(type='ReLU', inplace=True)), - deformable_model=dict( - type='DeformableFeatureAggregation', - embed_dims=256, - num_groups=8, - num_levels=4, - num_cams=6, - attn_drop=0.15, - use_deformable_func=True, - use_camera_embed=True, - residual_mode='cat', - kps_generator=dict( - type='SparseBox3DKeyPointsGenerator', - num_learnable_pts=6, - fix_scale=[[0, 0, 0], [0.45, 0, 0], [-0.45, 0, 0], - [0, 0.45, 0], [0, -0.45, 0], [0, 0, 0.45], - [0, 0, -0.45]])), - refine_layer=dict( - type='SparseBox3DRefinementModule', - embed_dims=256, - num_cls=10, - refine_yaw=True, - with_quality_estimation=True), - sampler=dict( - type='SparseBox3DTarget', - num_dn_groups=5, - num_temp_dn_groups=3, - dn_noise_scale=[2.0, 2.0, 2.0, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5], - max_dn_gt=32, - add_neg_dn=True, - cls_weight=2.0, - box_weight=0.25, - reg_weights=[2.0, 2.0, 2.0, 0.5, 0.5, 0.5, 0.0, 0.0, 0.0, 0.0], - cls_wise_reg_weights=dict( - {9: [2.0, 2.0, 2.0, 1.0, 1.0, 1.0, 0.0, 0.0, 1.0, 1.0]})), - loss_cls=dict( - type='FocalLoss', - use_sigmoid=True, - gamma=2.0, - alpha=0.25, - loss_weight=2.0), - loss_reg=dict( - type='SparseBox3DLoss', - loss_box=dict(type='L1Loss', loss_weight=0.25), - loss_centerness=dict(type='CrossEntropyLoss', use_sigmoid=True), - loss_yawness=dict(type='GaussianFocalLoss'), - cls_allow_reverse=[5]), - decoder=dict(type='SparseBox3DDecoder'), - reg_weights=[2.0, 2.0, 2.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0])) -dataset_type = 'NuScenes3DDetTrackDataset' -data_root = 'data/nuscenes/' -anno_root = 'data/nuscenes_anno_pkls/' -file_client_args = dict(backend='disk') -img_norm_cfg = dict( - mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375], to_rgb=True) -train_pipeline = [ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict( - type='LoadPointsFromFile', - coord_type='LIDAR', - load_dim=5, - use_dim=5, - file_client_args=dict(backend='disk')), - dict(type='ResizeCropFlipImage'), - dict(type='MultiScaleDepthMapGenerator', downsample=[4, 8, 16]), - dict(type='BBoxRotation'), - dict(type='PhotoMetricDistortionMultiViewImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict( - type='CircleObjectRangeFilter', - class_dist_thred=[55, 55, 55, 55, 55, 55, 55, 55, 55, 55]), - dict( - type='InstanceNameFilter', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ]), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=[ - 'img', 'timestamp', 'projection_mat', 'image_wh', 'gt_depth', - 'focal', 'gt_bboxes_3d', 'gt_labels_3d' - ], - meta_keys=['T_global', 'T_global_inv', 'timestamp', 'instance_id']) -] -test_pipeline = [ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='ResizeCropFlipImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=['img', 'timestamp', 'projection_mat', 'image_wh'], - meta_keys=['T_global', 'T_global_inv', 'timestamp']) -] -input_modality = dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False) -data_basic_config = dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier', - 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval') -data_aug_conf = dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]) -data = dict( - samples_per_gpu=20, - workers_per_gpu=20, - train=dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval', - ann_file='data/nuscenes_anno_pkls/nuscenes_infos_train.pkl', - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict( - type='LoadPointsFromFile', - coord_type='LIDAR', - load_dim=5, - use_dim=5, - file_client_args=dict(backend='disk')), - dict(type='ResizeCropFlipImage'), - dict(type='MultiScaleDepthMapGenerator', downsample=[4, 8, 16]), - dict(type='BBoxRotation'), - dict(type='PhotoMetricDistortionMultiViewImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict( - type='CircleObjectRangeFilter', - class_dist_thred=[55, 55, 55, 55, 55, 55, 55, 55, 55, 55]), - dict( - type='InstanceNameFilter', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', - 'traffic_cone' - ]), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=[ - 'img', 'timestamp', 'projection_mat', 'image_wh', - 'gt_depth', 'focal', 'gt_bboxes_3d', 'gt_labels_3d' - ], - meta_keys=[ - 'T_global', 'T_global_inv', 'timestamp', 'instance_id' - ]) - ], - test_mode=False, - data_aug_conf=dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]), - with_seq_flag=True, - sequences_split_num=2, - keep_consistent_seq_aug=True), - val=dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval', - ann_file='data/nuscenes_anno_pkls/nuscenes_infos_val.pkl', - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='ResizeCropFlipImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=['img', 'timestamp', 'projection_mat', 'image_wh'], - meta_keys=['T_global', 'T_global_inv', 'timestamp']) - ], - data_aug_conf=dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]), - test_mode=True, - tracking=True, - tracking_threshold=0.2), - test=dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval', - ann_file='data/nuscenes_anno_pkls/nuscenes_infos_val.pkl', - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='ResizeCropFlipImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=['img', 'timestamp', 'projection_mat', 'image_wh'], - meta_keys=['T_global', 'T_global_inv', 'timestamp']) - ], - data_aug_conf=dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]), - test_mode=True, - tracking=True, - tracking_threshold=0.2)) -optimizer = dict( - type='AdamW', - lr=0.0006, - weight_decay=0.001, - paramwise_cfg=dict(custom_keys=dict(img_backbone=dict(lr_mult=0.5)))) -optimizer_config = dict(grad_clip=dict(max_norm=25, norm_type=2)) -lr_config = dict( - policy='CosineAnnealing', - warmup='linear', - warmup_iters=500, - warmup_ratio=0.3333333333333333, - min_lr_ratio=0.001) -runner = dict(type='IterBasedRunner', max_iters=17500) -vis_pipeline = [ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='Collect', keys=['img'], meta_keys=['timestamp', 'lidar2img']) -] -evaluation = dict( - interval=3500, - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict( - type='Collect', keys=['img'], meta_keys=['timestamp', 'lidar2img']) - ]) -gpu_ids = range(0, 1) - -2026-02-10 15:38:04,625 - mmdet - INFO - Set random seed to 0, deterministic: False -2026-02-10 15:38:05,197 - mmdet - INFO - initialize ResNet with init_cfg {'type': 'Pretrained', 'checkpoint': 'ckpt/resnet50-19c8e357.pth'} -2026-02-10 15:38:05,747 - mmdet - INFO - initialize FPN with init_cfg {'type': 'Xavier', 'layer': 'Conv2d', 'distribution': 'uniform'} -Name of parameter - Initialization information - -img_backbone.conv1.weight - torch.Size([64, 3, 7, 7]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.bn1.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.bn1.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.conv1.weight - torch.Size([64, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn1.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn1.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.conv2.weight - torch.Size([64, 64, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn2.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn2.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.conv3.weight - torch.Size([256, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn3.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn3.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.downsample.0.weight - torch.Size([256, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.downsample.1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.downsample.1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.conv1.weight - torch.Size([64, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn1.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn1.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.conv2.weight - torch.Size([64, 64, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn2.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn2.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.conv3.weight - torch.Size([256, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn3.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn3.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.conv1.weight - torch.Size([64, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn1.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn1.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.conv2.weight - torch.Size([64, 64, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn2.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn2.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.conv3.weight - torch.Size([256, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn3.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn3.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.conv1.weight - torch.Size([128, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn1.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn1.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.conv2.weight - torch.Size([128, 128, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn2.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn2.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.conv3.weight - torch.Size([512, 128, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn3.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn3.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.downsample.0.weight - torch.Size([512, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.downsample.1.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.downsample.1.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.conv1.weight - torch.Size([128, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn1.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn1.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.conv2.weight - torch.Size([128, 128, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn2.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn2.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.conv3.weight - torch.Size([512, 128, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn3.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn3.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.conv1.weight - torch.Size([128, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn1.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn1.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.conv2.weight - torch.Size([128, 128, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn2.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn2.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.conv3.weight - torch.Size([512, 128, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn3.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn3.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.conv1.weight - torch.Size([128, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn1.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn1.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.conv2.weight - torch.Size([128, 128, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn2.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn2.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.conv3.weight - torch.Size([512, 128, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn3.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn3.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.conv1.weight - torch.Size([256, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.downsample.0.weight - torch.Size([1024, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.downsample.1.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.downsample.1.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.conv1.weight - torch.Size([512, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn1.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn1.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.conv2.weight - torch.Size([512, 512, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn2.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn2.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.conv3.weight - torch.Size([2048, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn3.weight - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn3.bias - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.downsample.0.weight - torch.Size([2048, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.downsample.1.weight - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.downsample.1.bias - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.conv1.weight - torch.Size([512, 2048, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn1.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn1.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.conv2.weight - torch.Size([512, 512, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn2.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn2.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.conv3.weight - torch.Size([2048, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn3.weight - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn3.bias - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.conv1.weight - torch.Size([512, 2048, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn1.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn1.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.conv2.weight - torch.Size([512, 512, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn2.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn2.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.conv3.weight - torch.Size([2048, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn3.weight - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn3.bias - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_neck.lateral_convs.0.conv.weight - torch.Size([256, 256, 1, 1]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.lateral_convs.0.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.lateral_convs.1.conv.weight - torch.Size([256, 512, 1, 1]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.lateral_convs.1.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.lateral_convs.2.conv.weight - torch.Size([256, 1024, 1, 1]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.lateral_convs.2.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.lateral_convs.3.conv.weight - torch.Size([256, 2048, 1, 1]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.lateral_convs.3.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.fpn_convs.0.conv.weight - torch.Size([256, 256, 3, 3]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.fpn_convs.0.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.fpn_convs.1.conv.weight - torch.Size([256, 256, 3, 3]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.fpn_convs.1.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.fpn_convs.2.conv.weight - torch.Size([256, 256, 3, 3]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.fpn_convs.2.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.fpn_convs.3.conv.weight - torch.Size([256, 256, 3, 3]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.fpn_convs.3.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.instance_bank.anchor - torch.Size([900, 11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.instance_bank.instance_feature - torch.Size([900, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.instance_bank.anchor_handler.fix_scale - torch.Size([1, 3]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.0.weight - torch.Size([128, 3]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.0.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.2.weight - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.2.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.3.weight - torch.Size([128, 128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.3.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.5.weight - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.5.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.6.weight - torch.Size([128, 128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.6.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.8.weight - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.8.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.9.weight - torch.Size([128, 128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.9.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.11.weight - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.11.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.0.weight - torch.Size([32, 3]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.0.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.2.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.2.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.3.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.3.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.5.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.5.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.6.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.6.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.8.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.8.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.9.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.9.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.11.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.11.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.0.weight - torch.Size([32, 2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.0.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.2.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.2.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.3.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.3.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.5.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.5.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.6.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.6.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.8.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.8.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.9.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.9.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.11.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.11.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.0.weight - torch.Size([64, 3]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.0.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.2.weight - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.2.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.3.weight - torch.Size([64, 64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.3.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.5.weight - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.5.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.6.weight - torch.Size([64, 64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.6.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.8.weight - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.8.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.9.weight - torch.Size([64, 64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.9.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.11.weight - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.11.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.1.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.1.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.1.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.1.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.1.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.1.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.1.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.1.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.3.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.4.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.4.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.4.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.4.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.5.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.5.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.5.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.5.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.6.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.6.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.8.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.8.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.8.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.8.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.8.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.8.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.8.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.8.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.10.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.11.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.11.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.11.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.11.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.12.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.12.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.12.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.12.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.13.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.13.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.15.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.15.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.15.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.15.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.15.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.15.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.15.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.15.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.16.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.16.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.17.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.18.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.18.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.18.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.18.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.19.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.19.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.19.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.19.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.20.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.20.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.22.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.22.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.22.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.22.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.22.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.22.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.22.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.22.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.23.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.23.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.24.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.25.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.25.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.25.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.25.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.26.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.26.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.26.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.26.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.27.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.27.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.29.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.29.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.29.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.29.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.29.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.29.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.29.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.29.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.30.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.30.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.31.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.32.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.32.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.32.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.32.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.33.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.33.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.33.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.33.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.34.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.34.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.36.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.36.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.36.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.36.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.36.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.36.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.36.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.36.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.37.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.37.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.38.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.fc_before.weight - torch.Size([512, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.fc_after.weight - torch.Size([256, 512]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.0.weight - torch.Size([1, 256, 1, 1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.0.bias - torch.Size([1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.1.weight - torch.Size([1, 256, 1, 1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.1.bias - torch.Size([1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.2.weight - torch.Size([1, 256, 1, 1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.2.bias - torch.Size([1]): -The value is the same before and after calling `init_weights` of Sparse4D -2026-02-10 15:38:06,059 - mmdet - INFO - Model: -Sparse4D( - (img_backbone): ResNet( - (conv1): Conv2d(3, 64, kernel_size=(7, 7), stride=(2, 2), padding=(3, 3), bias=False) - (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (maxpool): MaxPool2d(kernel_size=3, stride=2, padding=1, dilation=1, ceil_mode=False) - (layer1): ResLayer( - (0): Bottleneck( - (conv1): Conv2d(64, 64, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (downsample): Sequential( - (0): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - ) - ) - (1): Bottleneck( - (conv1): Conv2d(256, 64, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (2): Bottleneck( - (conv1): Conv2d(256, 64, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - ) - (layer2): ResLayer( - (0): Bottleneck( - (conv1): Conv2d(256, 128, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (downsample): Sequential( - (0): Conv2d(256, 512, kernel_size=(1, 1), stride=(2, 2), bias=False) - (1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - ) - ) - (1): Bottleneck( - (conv1): Conv2d(512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (2): Bottleneck( - (conv1): Conv2d(512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (3): Bottleneck( - (conv1): Conv2d(512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - ) - (layer3): ResLayer( - (0): Bottleneck( - (conv1): Conv2d(512, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (downsample): Sequential( - (0): Conv2d(512, 1024, kernel_size=(1, 1), stride=(2, 2), bias=False) - (1): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - ) - ) - (1): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (2): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (3): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (4): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (5): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - ) - (layer4): ResLayer( - (0): Bottleneck( - (conv1): Conv2d(1024, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(512, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (downsample): Sequential( - (0): Conv2d(1024, 2048, kernel_size=(1, 1), stride=(2, 2), bias=False) - (1): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - ) - ) - (1): Bottleneck( - (conv1): Conv2d(2048, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(512, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (2): Bottleneck( - (conv1): Conv2d(2048, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(512, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - ) - ) - init_cfg={'type': 'Pretrained', 'checkpoint': 'ckpt/resnet50-19c8e357.pth'} - (img_neck): FPN( - (lateral_convs): ModuleList( - (0): ConvModule( - (conv): Conv2d(256, 256, kernel_size=(1, 1), stride=(1, 1)) - ) - (1): ConvModule( - (conv): Conv2d(512, 256, kernel_size=(1, 1), stride=(1, 1)) - ) - (2): ConvModule( - (conv): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1)) - ) - (3): ConvModule( - (conv): Conv2d(2048, 256, kernel_size=(1, 1), stride=(1, 1)) - ) - ) - (fpn_convs): ModuleList( - (0-3): 4 x ConvModule( - (conv): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1)) - ) - ) - ) - init_cfg={'type': 'Xavier', 'layer': 'Conv2d', 'distribution': 'uniform'} - (head): Sparse4DHead( - (instance_bank): InstanceBank( - (anchor_handler): SparseBox3DKeyPointsGenerator() - ) - (anchor_encoder): SparseBox3DEncoder( - (pos_fc): Sequential( - (0): Linear(in_features=3, out_features=128, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((128,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=128, out_features=128, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((128,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=128, out_features=128, bias=True) - (7): ReLU(inplace=True) - (8): LayerNorm((128,), eps=1e-05, elementwise_affine=True) - (9): Linear(in_features=128, out_features=128, bias=True) - (10): ReLU(inplace=True) - (11): LayerNorm((128,), eps=1e-05, elementwise_affine=True) - ) - (size_fc): Sequential( - (0): Linear(in_features=3, out_features=32, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=32, out_features=32, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=32, out_features=32, bias=True) - (7): ReLU(inplace=True) - (8): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (9): Linear(in_features=32, out_features=32, bias=True) - (10): ReLU(inplace=True) - (11): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - ) - (yaw_fc): Sequential( - (0): Linear(in_features=2, out_features=32, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=32, out_features=32, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=32, out_features=32, bias=True) - (7): ReLU(inplace=True) - (8): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (9): Linear(in_features=32, out_features=32, bias=True) - (10): ReLU(inplace=True) - (11): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - ) - (vel_fc): Sequential( - (0): Linear(in_features=3, out_features=64, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((64,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=64, out_features=64, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((64,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=64, out_features=64, bias=True) - (7): ReLU(inplace=True) - (8): LayerNorm((64,), eps=1e-05, elementwise_affine=True) - (9): Linear(in_features=64, out_features=64, bias=True) - (10): ReLU(inplace=True) - (11): LayerNorm((64,), eps=1e-05, elementwise_affine=True) - ) - ) - (loss_cls): FocalLoss() - (loss_reg): SparseBox3DLoss( - (loss_box): L1Loss() - (loss_cns): CrossEntropyLoss(avg_non_ignore=False) - (loss_yns): GaussianFocalLoss() - ) - (layers): ModuleList( - (0): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (1): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (4-5): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (6): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (7): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (8): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (11-12): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (13): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (14): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (15): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (16): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (17): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (18-19): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (20): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (21): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (22): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (23): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (24): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (25-26): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (27): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (28): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (29): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (30): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (31): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (32-33): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (34): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (35): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (36): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (37): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (38): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - ) - (fc_before): Linear(in_features=256, out_features=512, bias=False) - (fc_after): Linear(in_features=512, out_features=256, bias=False) - ) - (depth_branch): DenseDepthNet( - (depth_layers): ModuleList( - (0-2): 3 x Conv2d(256, 1, kernel_size=(1, 1), stride=(1, 1)) - ) - ) - (grid_mask): GridMask() -) -2026-02-10 15:38:20,192 - mmdet - INFO - Start running, host: root@bw28, work_dir: /workspace/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704 -2026-02-10 15:38:20,192 - mmdet - INFO - Hooks will be executed in the following order: -before_run: -(VERY_HIGH ) CosineAnnealingLrUpdaterHook -(ABOVE_NORMAL) Fp16OptimizerHook -(NORMAL ) CheckpointHook -(NORMAL ) EvalHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -before_train_epoch: -(VERY_HIGH ) CosineAnnealingLrUpdaterHook -(NORMAL ) EvalHook -(LOW ) IterTimerHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -before_train_iter: -(VERY_HIGH ) CosineAnnealingLrUpdaterHook -(NORMAL ) EvalHook -(LOW ) IterTimerHook - -------------------- -after_train_iter: -(ABOVE_NORMAL) Fp16OptimizerHook -(NORMAL ) CheckpointHook -(NORMAL ) EvalHook -(NORMAL ) ProfilerHook -(LOW ) IterTimerHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -after_train_epoch: -(NORMAL ) CheckpointHook -(NORMAL ) EvalHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -before_val_epoch: -(LOW ) IterTimerHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -before_val_iter: -(LOW ) IterTimerHook - -------------------- -after_val_iter: -(LOW ) IterTimerHook - -------------------- -after_val_epoch: -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -after_run: -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -2026-02-10 15:38:20,193 - mmdet - INFO - workflow: [('train', 1)], max: 17500 iters -2026-02-10 15:38:20,195 - mmdet - INFO - Checkpoints will be saved to /workspace/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704 by HardDiskBackend. diff --git a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20260210_153802.log.json b/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20260210_153802.log.json deleted file mode 100644 index f691fdae6b2eb0b287f959b7a72d73034a4da3e6..0000000000000000000000000000000000000000 --- a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20260210_153802.log.json +++ /dev/null @@ -1 +0,0 @@ -{"env_info": "sys.platform: linux\nPython: 3.10.12 (main, Aug 15 2025, 14:32:43) [GCC 11.4.0]\nCUDA available: True\nGPU 0,1,2,3,4,5,6,7: BW200_LC, UBB BW1000_LC\nCUDA_HOME: /opt/dtk\nNVCC: Not Available\nGCC: x86_64-linux-gnu-gcc (Ubuntu 11.4.0-1ubuntu1~22.04) 11.4.0\nPyTorch: 2.5.1\nPyTorch compiling details: PyTorch built with:\n - GCC 10.3\n - C++ Version: 201703\n - Intel(R) Math Kernel Library Version 2020.0.4 Product Build 20200917 for Intel(R) 64 architecture applications\n - OpenMP 201511 (a.k.a. OpenMP 4.5)\n - LAPACK is enabled (usually provided by MKL)\n - NNPACK is enabled\n - CPU capability usage: AVX2\n - HIP Runtime 6.3.25521\n - MIOpen 2.18.0\n - Magma 2.8.0\n - Build settings: BLAS_INFO=mkl, BUILD_TYPE=Release, CXX_COMPILER=/opt/rh/gcc-toolset-10/root/usr/bin/c++, CXX_FLAGS= -D_GLIBCXX_USE_CXX11_ABI=1 -fvisibility-inlines-hidden -DUSE_PTHREADPOOL -DNDEBUG -DUSE_KINETO -DLIBKINETO_NOCUPTI -DLIBKINETO_NOXPUPTI=ON -DUSE_FBGEMM -DUSE_PYTORCH_QNNPACK -DUSE_XNNPACK -DSYMBOLICATE_MOBILE_DEBUG_HANDLE -O2 -fPIC -Wall -Wextra -Werror=return-type -Werror=non-virtual-dtor -Werror=bool-operation -Wnarrowing -Wno-missing-field-initializers -Wno-type-limits -Wno-array-bounds -Wno-unknown-pragmas -Wno-unused-parameter -Wno-strict-overflow -Wno-strict-aliasing -Wno-stringop-overflow -Wsuggest-override -Wno-psabi -Wno-error=old-style-cast -Wno-missing-braces -fdiagnostics-color=always -faligned-new -Wno-unused-but-set-variable -Wno-maybe-uninitialized -fno-math-errno -fno-trapping-math -Werror=format -Wno-stringop-overflow, FORCE_FALLBACK_CUDA_MPI=1, LAPACK_INFO=mkl, PERF_WITH_AVX=1, PERF_WITH_AVX2=1, TORCH_VERSION=2.5.1, USE_CUDA=0, USE_CUDNN=OFF, USE_CUSPARSELT=OFF, USE_EXCEPTION_PTR=1, USE_GFLAGS=1, USE_GLOG=1, USE_GLOO=1, USE_MKL=ON, USE_MKLDNN=0, USE_MPI=1, USE_NCCL=1, USE_NNPACK=ON, USE_OPENMP=1, USE_ROCM=ON, USE_ROCM_KERNEL_ASSERT=OFF, \n\nTorchVision: 0.20.1\nOpenCV: 4.12.0\nMMCV: 1.6.1\nMMCV Compiler: GCC 10.3\nMMCV CUDA Compiler: rocm not available\nMMDetection: 2.25.1+", "config": "plugin = True\nplugin_dir = 'projects/mmdet3d_plugin/'\ndist_params = dict(backend='nccl')\nlog_level = 'INFO'\nwork_dir = './work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704'\ntotal_batch_size = 160\nnum_gpus = 8\nbatch_size = 20\nnum_iters_per_epoch = 175\nnum_epochs = 100\ncheckpoint_epoch_interval = 20\ncheckpoint_config = dict(interval=3500)\nlog_config = dict(\n interval=1,\n hooks=[\n dict(type='TextLoggerHook', by_epoch=False),\n dict(type='TensorboardLoggerHook')\n ])\nload_from = None\nresume_from = None\nworkflow = [('train', 1)]\nfp16 = dict(loss_scale=32.0)\ninput_shape = (704, 256)\ntracking_test = True\ntracking_threshold = 0.2\nclass_names = [\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier',\n 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n]\nnum_classes = 10\nembed_dims = 256\nnum_groups = 8\nnum_decoder = 6\nnum_single_frame_decoder = 1\nuse_deformable_func = True\nstrides = [4, 8, 16, 32]\nnum_levels = 4\nnum_depth_layers = 3\ndrop_out = 0.1\ntemporal = True\ndecouple_attn = True\nwith_quality_estimation = True\nmodel = dict(\n type='Sparse4D',\n use_grid_mask=True,\n use_deformable_func=True,\n img_backbone=dict(\n type='ResNet',\n depth=50,\n num_stages=4,\n frozen_stages=-1,\n norm_eval=False,\n style='pytorch',\n with_cp=True,\n out_indices=(0, 1, 2, 3),\n norm_cfg=dict(type='BN', requires_grad=True),\n pretrained='ckpt/resnet50-19c8e357.pth'),\n img_neck=dict(\n type='FPN',\n num_outs=4,\n start_level=0,\n out_channels=256,\n add_extra_convs='on_output',\n relu_before_extra_convs=True,\n in_channels=[256, 512, 1024, 2048]),\n depth_branch=dict(\n type='DenseDepthNet',\n embed_dims=256,\n num_depth_layers=3,\n loss_weight=0.2),\n head=dict(\n type='Sparse4DHead',\n cls_threshold_to_reg=0.05,\n decouple_attn=True,\n instance_bank=dict(\n type='InstanceBank',\n num_anchor=900,\n embed_dims=256,\n anchor='nuscenes_kmeans900.npy',\n anchor_handler=dict(type='SparseBox3DKeyPointsGenerator'),\n num_temp_instances=600,\n confidence_decay=0.6,\n feat_grad=False),\n anchor_encoder=dict(\n type='SparseBox3DEncoder',\n vel_dims=3,\n embed_dims=[128, 32, 32, 64],\n mode='cat',\n output_fc=False,\n in_loops=1,\n out_loops=4),\n num_single_frame_decoder=1,\n operation_order=[\n 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm',\n 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm',\n 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm',\n 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm',\n 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm',\n 'deformable', 'ffn', 'norm', 'refine'\n ],\n temp_graph_model=dict(\n type='MultiheadAttention',\n embed_dims=512,\n num_heads=8,\n batch_first=True,\n dropout=0.1),\n graph_model=dict(\n type='MultiheadAttention',\n embed_dims=512,\n num_heads=8,\n batch_first=True,\n dropout=0.1),\n norm_layer=dict(type='LN', normalized_shape=256),\n ffn=dict(\n type='AsymmetricFFN',\n in_channels=512,\n pre_norm=dict(type='LN'),\n embed_dims=256,\n feedforward_channels=1024,\n num_fcs=2,\n ffn_drop=0.1,\n act_cfg=dict(type='ReLU', inplace=True)),\n deformable_model=dict(\n type='DeformableFeatureAggregation',\n embed_dims=256,\n num_groups=8,\n num_levels=4,\n num_cams=6,\n attn_drop=0.15,\n use_deformable_func=True,\n use_camera_embed=True,\n residual_mode='cat',\n kps_generator=dict(\n type='SparseBox3DKeyPointsGenerator',\n num_learnable_pts=6,\n fix_scale=[[0, 0, 0], [0.45, 0, 0], [-0.45, 0, 0],\n [0, 0.45, 0], [0, -0.45, 0], [0, 0, 0.45],\n [0, 0, -0.45]])),\n refine_layer=dict(\n type='SparseBox3DRefinementModule',\n embed_dims=256,\n num_cls=10,\n refine_yaw=True,\n with_quality_estimation=True),\n sampler=dict(\n type='SparseBox3DTarget',\n num_dn_groups=5,\n num_temp_dn_groups=3,\n dn_noise_scale=[2.0, 2.0, 2.0, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5],\n max_dn_gt=32,\n add_neg_dn=True,\n cls_weight=2.0,\n box_weight=0.25,\n reg_weights=[2.0, 2.0, 2.0, 0.5, 0.5, 0.5, 0.0, 0.0, 0.0, 0.0],\n cls_wise_reg_weights=dict(\n {9: [2.0, 2.0, 2.0, 1.0, 1.0, 1.0, 0.0, 0.0, 1.0, 1.0]})),\n loss_cls=dict(\n type='FocalLoss',\n use_sigmoid=True,\n gamma=2.0,\n alpha=0.25,\n loss_weight=2.0),\n loss_reg=dict(\n type='SparseBox3DLoss',\n loss_box=dict(type='L1Loss', loss_weight=0.25),\n loss_centerness=dict(type='CrossEntropyLoss', use_sigmoid=True),\n loss_yawness=dict(type='GaussianFocalLoss'),\n cls_allow_reverse=[5]),\n decoder=dict(type='SparseBox3DDecoder'),\n reg_weights=[2.0, 2.0, 2.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0]))\ndataset_type = 'NuScenes3DDetTrackDataset'\ndata_root = 'data/nuscenes/'\nanno_root = 'data/nuscenes_anno_pkls/'\nfile_client_args = dict(backend='disk')\nimg_norm_cfg = dict(\n mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375], to_rgb=True)\ntrain_pipeline = [\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(\n type='LoadPointsFromFile',\n coord_type='LIDAR',\n load_dim=5,\n use_dim=5,\n file_client_args=dict(backend='disk')),\n dict(type='ResizeCropFlipImage'),\n dict(type='MultiScaleDepthMapGenerator', downsample=[4, 8, 16]),\n dict(type='BBoxRotation'),\n dict(type='PhotoMetricDistortionMultiViewImage'),\n dict(\n type='NormalizeMultiviewImage',\n mean=[123.675, 116.28, 103.53],\n std=[58.395, 57.12, 57.375],\n to_rgb=True),\n dict(\n type='CircleObjectRangeFilter',\n class_dist_thred=[55, 55, 55, 55, 55, 55, 55, 55, 55, 55]),\n dict(\n type='InstanceNameFilter',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ]),\n dict(type='NuScenesSparse4DAdaptor'),\n dict(\n type='Collect',\n keys=[\n 'img', 'timestamp', 'projection_mat', 'image_wh', 'gt_depth',\n 'focal', 'gt_bboxes_3d', 'gt_labels_3d'\n ],\n meta_keys=['T_global', 'T_global_inv', 'timestamp', 'instance_id'])\n]\ntest_pipeline = [\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(type='ResizeCropFlipImage'),\n dict(\n type='NormalizeMultiviewImage',\n mean=[123.675, 116.28, 103.53],\n std=[58.395, 57.12, 57.375],\n to_rgb=True),\n dict(type='NuScenesSparse4DAdaptor'),\n dict(\n type='Collect',\n keys=['img', 'timestamp', 'projection_mat', 'image_wh'],\n meta_keys=['T_global', 'T_global_inv', 'timestamp'])\n]\ninput_modality = dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False)\ndata_basic_config = dict(\n type='NuScenes3DDetTrackDataset',\n data_root='data/nuscenes/',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier',\n 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ],\n modality=dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False),\n version='v1.0-trainval')\ndata_aug_conf = dict(\n resize_lim=(0.4, 0.47),\n final_dim=(256, 704),\n bot_pct_lim=(0.0, 0.0),\n rot_lim=(-5.4, 5.4),\n H=900,\n W=1600,\n rand_flip=True,\n rot3d_range=[-0.3925, 0.3925])\ndata = dict(\n samples_per_gpu=20,\n workers_per_gpu=20,\n train=dict(\n type='NuScenes3DDetTrackDataset',\n data_root='data/nuscenes/',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ],\n modality=dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False),\n version='v1.0-trainval',\n ann_file='data/nuscenes_anno_pkls/nuscenes_infos_train.pkl',\n pipeline=[\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(\n type='LoadPointsFromFile',\n coord_type='LIDAR',\n load_dim=5,\n use_dim=5,\n file_client_args=dict(backend='disk')),\n dict(type='ResizeCropFlipImage'),\n dict(type='MultiScaleDepthMapGenerator', downsample=[4, 8, 16]),\n dict(type='BBoxRotation'),\n dict(type='PhotoMetricDistortionMultiViewImage'),\n dict(\n type='NormalizeMultiviewImage',\n mean=[123.675, 116.28, 103.53],\n std=[58.395, 57.12, 57.375],\n to_rgb=True),\n dict(\n type='CircleObjectRangeFilter',\n class_dist_thred=[55, 55, 55, 55, 55, 55, 55, 55, 55, 55]),\n dict(\n type='InstanceNameFilter',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian',\n 'traffic_cone'\n ]),\n dict(type='NuScenesSparse4DAdaptor'),\n dict(\n type='Collect',\n keys=[\n 'img', 'timestamp', 'projection_mat', 'image_wh',\n 'gt_depth', 'focal', 'gt_bboxes_3d', 'gt_labels_3d'\n ],\n meta_keys=[\n 'T_global', 'T_global_inv', 'timestamp', 'instance_id'\n ])\n ],\n test_mode=False,\n data_aug_conf=dict(\n resize_lim=(0.4, 0.47),\n final_dim=(256, 704),\n bot_pct_lim=(0.0, 0.0),\n rot_lim=(-5.4, 5.4),\n H=900,\n W=1600,\n rand_flip=True,\n rot3d_range=[-0.3925, 0.3925]),\n with_seq_flag=True,\n sequences_split_num=2,\n keep_consistent_seq_aug=True),\n val=dict(\n type='NuScenes3DDetTrackDataset',\n data_root='data/nuscenes/',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ],\n modality=dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False),\n version='v1.0-trainval',\n ann_file='data/nuscenes_anno_pkls/nuscenes_infos_val.pkl',\n pipeline=[\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(type='ResizeCropFlipImage'),\n dict(\n type='NormalizeMultiviewImage',\n mean=[123.675, 116.28, 103.53],\n std=[58.395, 57.12, 57.375],\n to_rgb=True),\n dict(type='NuScenesSparse4DAdaptor'),\n dict(\n type='Collect',\n keys=['img', 'timestamp', 'projection_mat', 'image_wh'],\n meta_keys=['T_global', 'T_global_inv', 'timestamp'])\n ],\n data_aug_conf=dict(\n resize_lim=(0.4, 0.47),\n final_dim=(256, 704),\n bot_pct_lim=(0.0, 0.0),\n rot_lim=(-5.4, 5.4),\n H=900,\n W=1600,\n rand_flip=True,\n rot3d_range=[-0.3925, 0.3925]),\n test_mode=True,\n tracking=True,\n tracking_threshold=0.2),\n test=dict(\n type='NuScenes3DDetTrackDataset',\n data_root='data/nuscenes/',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ],\n modality=dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False),\n version='v1.0-trainval',\n ann_file='data/nuscenes_anno_pkls/nuscenes_infos_val.pkl',\n pipeline=[\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(type='ResizeCropFlipImage'),\n dict(\n type='NormalizeMultiviewImage',\n mean=[123.675, 116.28, 103.53],\n std=[58.395, 57.12, 57.375],\n to_rgb=True),\n dict(type='NuScenesSparse4DAdaptor'),\n dict(\n type='Collect',\n keys=['img', 'timestamp', 'projection_mat', 'image_wh'],\n meta_keys=['T_global', 'T_global_inv', 'timestamp'])\n ],\n data_aug_conf=dict(\n resize_lim=(0.4, 0.47),\n final_dim=(256, 704),\n bot_pct_lim=(0.0, 0.0),\n rot_lim=(-5.4, 5.4),\n H=900,\n W=1600,\n rand_flip=True,\n rot3d_range=[-0.3925, 0.3925]),\n test_mode=True,\n tracking=True,\n tracking_threshold=0.2))\noptimizer = dict(\n type='AdamW',\n lr=0.0006,\n weight_decay=0.001,\n paramwise_cfg=dict(custom_keys=dict(img_backbone=dict(lr_mult=0.5))))\noptimizer_config = dict(grad_clip=dict(max_norm=25, norm_type=2))\nlr_config = dict(\n policy='CosineAnnealing',\n warmup='linear',\n warmup_iters=500,\n warmup_ratio=0.3333333333333333,\n min_lr_ratio=0.001)\nrunner = dict(type='IterBasedRunner', max_iters=17500)\nvis_pipeline = [\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(type='Collect', keys=['img'], meta_keys=['timestamp', 'lidar2img'])\n]\nevaluation = dict(\n interval=3500,\n pipeline=[\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(\n type='Collect', keys=['img'], meta_keys=['timestamp', 'lidar2img'])\n ])\ngpu_ids = range(0, 1)\n", "seed": 0, "exp_name": "sparse4dv3_temporal_r50_1x8_bs6_256x704.py"} diff --git a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20260210_160324.log b/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20260210_160324.log deleted file mode 100644 index bcadcf0d03f861e9446562b2c0334352d7710e6e..0000000000000000000000000000000000000000 --- a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20260210_160324.log +++ /dev/null @@ -1,3221 +0,0 @@ -2026-02-10 16:03:24,744 - mmdet - INFO - Environment info: ------------------------------------------------------------- -sys.platform: linux -Python: 3.10.12 (main, Aug 15 2025, 14:32:43) [GCC 11.4.0] -CUDA available: True -GPU 0,1,2,3,4,5,6,7: BW200_LC, UBB BW1000_LC -CUDA_HOME: /opt/dtk -NVCC: Not Available -GCC: x86_64-linux-gnu-gcc (Ubuntu 11.4.0-1ubuntu1~22.04) 11.4.0 -PyTorch: 2.5.1 -PyTorch compiling details: PyTorch built with: - - GCC 10.3 - - C++ Version: 201703 - - Intel(R) Math Kernel Library Version 2020.0.4 Product Build 20200917 for Intel(R) 64 architecture applications - - OpenMP 201511 (a.k.a. OpenMP 4.5) - - LAPACK is enabled (usually provided by MKL) - - NNPACK is enabled - - CPU capability usage: AVX2 - - HIP Runtime 6.3.25521 - - MIOpen 2.18.0 - - Magma 2.8.0 - - Build settings: BLAS_INFO=mkl, BUILD_TYPE=Release, CXX_COMPILER=/opt/rh/gcc-toolset-10/root/usr/bin/c++, CXX_FLAGS= -D_GLIBCXX_USE_CXX11_ABI=1 -fvisibility-inlines-hidden -DUSE_PTHREADPOOL -DNDEBUG -DUSE_KINETO -DLIBKINETO_NOCUPTI -DLIBKINETO_NOXPUPTI=ON -DUSE_FBGEMM -DUSE_PYTORCH_QNNPACK -DUSE_XNNPACK -DSYMBOLICATE_MOBILE_DEBUG_HANDLE -O2 -fPIC -Wall -Wextra -Werror=return-type -Werror=non-virtual-dtor -Werror=bool-operation -Wnarrowing -Wno-missing-field-initializers -Wno-type-limits -Wno-array-bounds -Wno-unknown-pragmas -Wno-unused-parameter -Wno-strict-overflow -Wno-strict-aliasing -Wno-stringop-overflow -Wsuggest-override -Wno-psabi -Wno-error=old-style-cast -Wno-missing-braces -fdiagnostics-color=always -faligned-new -Wno-unused-but-set-variable -Wno-maybe-uninitialized -fno-math-errno -fno-trapping-math -Werror=format -Wno-stringop-overflow, FORCE_FALLBACK_CUDA_MPI=1, LAPACK_INFO=mkl, PERF_WITH_AVX=1, PERF_WITH_AVX2=1, TORCH_VERSION=2.5.1, USE_CUDA=0, USE_CUDNN=OFF, USE_CUSPARSELT=OFF, USE_EXCEPTION_PTR=1, USE_GFLAGS=1, USE_GLOG=1, USE_GLOO=1, USE_MKL=ON, USE_MKLDNN=0, USE_MPI=1, USE_NCCL=1, USE_NNPACK=ON, USE_OPENMP=1, USE_ROCM=ON, USE_ROCM_KERNEL_ASSERT=OFF, - -TorchVision: 0.20.1 -OpenCV: 4.12.0 -MMCV: 1.6.1 -MMCV Compiler: GCC 10.3 -MMCV CUDA Compiler: rocm not available -MMDetection: 2.25.1+ ------------------------------------------------------------- - -2026-02-10 16:03:25,714 - mmdet - INFO - Distributed training: False -2026-02-10 16:03:26,685 - mmdet - INFO - Config: -plugin = True -plugin_dir = 'projects/mmdet3d_plugin/' -dist_params = dict(backend='nccl') -log_level = 'INFO' -work_dir = './work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704' -total_batch_size = 160 -num_gpus = 8 -batch_size = 20 -num_iters_per_epoch = 175 -num_epochs = 100 -checkpoint_epoch_interval = 20 -checkpoint_config = dict(interval=3500) -log_config = dict( - interval=1, - hooks=[ - dict(type='TextLoggerHook', by_epoch=False), - dict(type='TensorboardLoggerHook') - ]) -load_from = None -resume_from = None -workflow = [('train', 1)] -fp16 = dict(loss_scale=32.0) -input_shape = (704, 256) -tracking_test = True -tracking_threshold = 0.2 -class_names = [ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier', - 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' -] -num_classes = 10 -embed_dims = 256 -num_groups = 8 -num_decoder = 6 -num_single_frame_decoder = 1 -use_deformable_func = True -strides = [4, 8, 16, 32] -num_levels = 4 -num_depth_layers = 3 -drop_out = 0.1 -temporal = True -decouple_attn = True -with_quality_estimation = True -model = dict( - type='Sparse4D', - use_grid_mask=True, - use_deformable_func=True, - img_backbone=dict( - type='ResNet', - depth=50, - num_stages=4, - frozen_stages=-1, - norm_eval=False, - style='pytorch', - with_cp=True, - out_indices=(0, 1, 2, 3), - norm_cfg=dict(type='BN', requires_grad=True), - pretrained='ckpt/resnet50-19c8e357.pth'), - img_neck=dict( - type='FPN', - num_outs=4, - start_level=0, - out_channels=256, - add_extra_convs='on_output', - relu_before_extra_convs=True, - in_channels=[256, 512, 1024, 2048]), - depth_branch=dict( - type='DenseDepthNet', - embed_dims=256, - num_depth_layers=3, - loss_weight=0.2), - head=dict( - type='Sparse4DHead', - cls_threshold_to_reg=0.05, - decouple_attn=True, - instance_bank=dict( - type='InstanceBank', - num_anchor=900, - embed_dims=256, - anchor='nuscenes_kmeans900.npy', - anchor_handler=dict(type='SparseBox3DKeyPointsGenerator'), - num_temp_instances=600, - confidence_decay=0.6, - feat_grad=False), - anchor_encoder=dict( - type='SparseBox3DEncoder', - vel_dims=3, - embed_dims=[128, 32, 32, 64], - mode='cat', - output_fc=False, - in_loops=1, - out_loops=4), - num_single_frame_decoder=1, - operation_order=[ - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine' - ], - temp_graph_model=dict( - type='MultiheadAttention', - embed_dims=512, - num_heads=8, - batch_first=True, - dropout=0.1), - graph_model=dict( - type='MultiheadAttention', - embed_dims=512, - num_heads=8, - batch_first=True, - dropout=0.1), - norm_layer=dict(type='LN', normalized_shape=256), - ffn=dict( - type='AsymmetricFFN', - in_channels=512, - pre_norm=dict(type='LN'), - embed_dims=256, - feedforward_channels=1024, - num_fcs=2, - ffn_drop=0.1, - act_cfg=dict(type='ReLU', inplace=True)), - deformable_model=dict( - type='DeformableFeatureAggregation', - embed_dims=256, - num_groups=8, - num_levels=4, - num_cams=6, - attn_drop=0.15, - use_deformable_func=True, - use_camera_embed=True, - residual_mode='cat', - kps_generator=dict( - type='SparseBox3DKeyPointsGenerator', - num_learnable_pts=6, - fix_scale=[[0, 0, 0], [0.45, 0, 0], [-0.45, 0, 0], - [0, 0.45, 0], [0, -0.45, 0], [0, 0, 0.45], - [0, 0, -0.45]])), - refine_layer=dict( - type='SparseBox3DRefinementModule', - embed_dims=256, - num_cls=10, - refine_yaw=True, - with_quality_estimation=True), - sampler=dict( - type='SparseBox3DTarget', - num_dn_groups=5, - num_temp_dn_groups=3, - dn_noise_scale=[2.0, 2.0, 2.0, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5], - max_dn_gt=32, - add_neg_dn=True, - cls_weight=2.0, - box_weight=0.25, - reg_weights=[2.0, 2.0, 2.0, 0.5, 0.5, 0.5, 0.0, 0.0, 0.0, 0.0], - cls_wise_reg_weights=dict( - {9: [2.0, 2.0, 2.0, 1.0, 1.0, 1.0, 0.0, 0.0, 1.0, 1.0]})), - loss_cls=dict( - type='FocalLoss', - use_sigmoid=True, - gamma=2.0, - alpha=0.25, - loss_weight=2.0), - loss_reg=dict( - type='SparseBox3DLoss', - loss_box=dict(type='L1Loss', loss_weight=0.25), - loss_centerness=dict(type='CrossEntropyLoss', use_sigmoid=True), - loss_yawness=dict(type='GaussianFocalLoss'), - cls_allow_reverse=[5]), - decoder=dict(type='SparseBox3DDecoder'), - reg_weights=[2.0, 2.0, 2.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0])) -dataset_type = 'NuScenes3DDetTrackDataset' -data_root = 'data/nuscenes/' -anno_root = 'data/nuscenes_anno_pkls/' -file_client_args = dict(backend='disk') -img_norm_cfg = dict( - mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375], to_rgb=True) -train_pipeline = [ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict( - type='LoadPointsFromFile', - coord_type='LIDAR', - load_dim=5, - use_dim=5, - file_client_args=dict(backend='disk')), - dict(type='ResizeCropFlipImage'), - dict(type='MultiScaleDepthMapGenerator', downsample=[4, 8, 16]), - dict(type='BBoxRotation'), - dict(type='PhotoMetricDistortionMultiViewImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict( - type='CircleObjectRangeFilter', - class_dist_thred=[55, 55, 55, 55, 55, 55, 55, 55, 55, 55]), - dict( - type='InstanceNameFilter', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ]), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=[ - 'img', 'timestamp', 'projection_mat', 'image_wh', 'gt_depth', - 'focal', 'gt_bboxes_3d', 'gt_labels_3d' - ], - meta_keys=['T_global', 'T_global_inv', 'timestamp', 'instance_id']) -] -test_pipeline = [ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='ResizeCropFlipImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=['img', 'timestamp', 'projection_mat', 'image_wh'], - meta_keys=['T_global', 'T_global_inv', 'timestamp']) -] -input_modality = dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False) -data_basic_config = dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier', - 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval') -data_aug_conf = dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]) -data = dict( - samples_per_gpu=20, - workers_per_gpu=20, - train=dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval', - ann_file='data/nuscenes_anno_pkls/nuscenes_infos_train.pkl', - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict( - type='LoadPointsFromFile', - coord_type='LIDAR', - load_dim=5, - use_dim=5, - file_client_args=dict(backend='disk')), - dict(type='ResizeCropFlipImage'), - dict(type='MultiScaleDepthMapGenerator', downsample=[4, 8, 16]), - dict(type='BBoxRotation'), - dict(type='PhotoMetricDistortionMultiViewImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict( - type='CircleObjectRangeFilter', - class_dist_thred=[55, 55, 55, 55, 55, 55, 55, 55, 55, 55]), - dict( - type='InstanceNameFilter', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', - 'traffic_cone' - ]), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=[ - 'img', 'timestamp', 'projection_mat', 'image_wh', - 'gt_depth', 'focal', 'gt_bboxes_3d', 'gt_labels_3d' - ], - meta_keys=[ - 'T_global', 'T_global_inv', 'timestamp', 'instance_id' - ]) - ], - test_mode=False, - data_aug_conf=dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]), - with_seq_flag=True, - sequences_split_num=2, - keep_consistent_seq_aug=True), - val=dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval', - ann_file='data/nuscenes_anno_pkls/nuscenes_infos_val.pkl', - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='ResizeCropFlipImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=['img', 'timestamp', 'projection_mat', 'image_wh'], - meta_keys=['T_global', 'T_global_inv', 'timestamp']) - ], - data_aug_conf=dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]), - test_mode=True, - tracking=True, - tracking_threshold=0.2), - test=dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval', - ann_file='data/nuscenes_anno_pkls/nuscenes_infos_val.pkl', - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='ResizeCropFlipImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=['img', 'timestamp', 'projection_mat', 'image_wh'], - meta_keys=['T_global', 'T_global_inv', 'timestamp']) - ], - data_aug_conf=dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]), - test_mode=True, - tracking=True, - tracking_threshold=0.2)) -optimizer = dict( - type='AdamW', - lr=0.0006, - weight_decay=0.001, - paramwise_cfg=dict(custom_keys=dict(img_backbone=dict(lr_mult=0.5)))) -optimizer_config = dict(grad_clip=dict(max_norm=25, norm_type=2)) -lr_config = dict( - policy='CosineAnnealing', - warmup='linear', - warmup_iters=500, - warmup_ratio=0.3333333333333333, - min_lr_ratio=0.001) -runner = dict(type='IterBasedRunner', max_iters=17500) -vis_pipeline = [ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='Collect', keys=['img'], meta_keys=['timestamp', 'lidar2img']) -] -evaluation = dict( - interval=3500, - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict( - type='Collect', keys=['img'], meta_keys=['timestamp', 'lidar2img']) - ]) -gpu_ids = range(0, 1) - -2026-02-10 16:03:26,686 - mmdet - INFO - Set random seed to 0, deterministic: False -2026-02-10 16:03:27,245 - mmdet - INFO - initialize ResNet with init_cfg {'type': 'Pretrained', 'checkpoint': 'ckpt/resnet50-19c8e357.pth'} -2026-02-10 16:03:27,765 - mmdet - INFO - initialize FPN with init_cfg {'type': 'Xavier', 'layer': 'Conv2d', 'distribution': 'uniform'} -Name of parameter - Initialization information - -img_backbone.conv1.weight - torch.Size([64, 3, 7, 7]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.bn1.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.bn1.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.conv1.weight - torch.Size([64, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn1.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn1.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.conv2.weight - torch.Size([64, 64, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn2.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn2.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.conv3.weight - torch.Size([256, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn3.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn3.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.downsample.0.weight - torch.Size([256, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.downsample.1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.downsample.1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.conv1.weight - torch.Size([64, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn1.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn1.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.conv2.weight - torch.Size([64, 64, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn2.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn2.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.conv3.weight - torch.Size([256, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn3.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn3.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.conv1.weight - torch.Size([64, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn1.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn1.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.conv2.weight - torch.Size([64, 64, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn2.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn2.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.conv3.weight - torch.Size([256, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn3.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn3.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.conv1.weight - torch.Size([128, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn1.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn1.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.conv2.weight - torch.Size([128, 128, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn2.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn2.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.conv3.weight - torch.Size([512, 128, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn3.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn3.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.downsample.0.weight - torch.Size([512, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.downsample.1.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.downsample.1.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.conv1.weight - torch.Size([128, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn1.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn1.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.conv2.weight - torch.Size([128, 128, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn2.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn2.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.conv3.weight - torch.Size([512, 128, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn3.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn3.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.conv1.weight - torch.Size([128, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn1.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn1.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.conv2.weight - torch.Size([128, 128, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn2.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn2.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.conv3.weight - torch.Size([512, 128, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn3.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn3.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.conv1.weight - torch.Size([128, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn1.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn1.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.conv2.weight - torch.Size([128, 128, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn2.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn2.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.conv3.weight - torch.Size([512, 128, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn3.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn3.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.conv1.weight - torch.Size([256, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.downsample.0.weight - torch.Size([1024, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.downsample.1.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.downsample.1.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.conv1.weight - torch.Size([512, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn1.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn1.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.conv2.weight - torch.Size([512, 512, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn2.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn2.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.conv3.weight - torch.Size([2048, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn3.weight - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn3.bias - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.downsample.0.weight - torch.Size([2048, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.downsample.1.weight - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.downsample.1.bias - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.conv1.weight - torch.Size([512, 2048, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn1.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn1.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.conv2.weight - torch.Size([512, 512, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn2.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn2.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.conv3.weight - torch.Size([2048, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn3.weight - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn3.bias - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.conv1.weight - torch.Size([512, 2048, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn1.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn1.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.conv2.weight - torch.Size([512, 512, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn2.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn2.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.conv3.weight - torch.Size([2048, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn3.weight - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn3.bias - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_neck.lateral_convs.0.conv.weight - torch.Size([256, 256, 1, 1]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.lateral_convs.0.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.lateral_convs.1.conv.weight - torch.Size([256, 512, 1, 1]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.lateral_convs.1.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.lateral_convs.2.conv.weight - torch.Size([256, 1024, 1, 1]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.lateral_convs.2.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.lateral_convs.3.conv.weight - torch.Size([256, 2048, 1, 1]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.lateral_convs.3.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.fpn_convs.0.conv.weight - torch.Size([256, 256, 3, 3]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.fpn_convs.0.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.fpn_convs.1.conv.weight - torch.Size([256, 256, 3, 3]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.fpn_convs.1.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.fpn_convs.2.conv.weight - torch.Size([256, 256, 3, 3]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.fpn_convs.2.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.fpn_convs.3.conv.weight - torch.Size([256, 256, 3, 3]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.fpn_convs.3.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.instance_bank.anchor - torch.Size([900, 11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.instance_bank.instance_feature - torch.Size([900, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.instance_bank.anchor_handler.fix_scale - torch.Size([1, 3]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.0.weight - torch.Size([128, 3]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.0.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.2.weight - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.2.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.3.weight - torch.Size([128, 128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.3.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.5.weight - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.5.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.6.weight - torch.Size([128, 128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.6.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.8.weight - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.8.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.9.weight - torch.Size([128, 128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.9.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.11.weight - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.11.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.0.weight - torch.Size([32, 3]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.0.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.2.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.2.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.3.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.3.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.5.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.5.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.6.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.6.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.8.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.8.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.9.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.9.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.11.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.11.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.0.weight - torch.Size([32, 2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.0.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.2.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.2.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.3.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.3.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.5.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.5.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.6.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.6.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.8.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.8.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.9.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.9.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.11.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.11.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.0.weight - torch.Size([64, 3]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.0.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.2.weight - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.2.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.3.weight - torch.Size([64, 64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.3.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.5.weight - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.5.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.6.weight - torch.Size([64, 64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.6.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.8.weight - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.8.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.9.weight - torch.Size([64, 64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.9.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.11.weight - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.11.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.1.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.1.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.1.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.1.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.1.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.1.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.1.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.1.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.3.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.4.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.4.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.4.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.4.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.5.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.5.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.5.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.5.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.6.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.6.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.8.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.8.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.8.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.8.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.8.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.8.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.8.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.8.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.10.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.11.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.11.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.11.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.11.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.12.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.12.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.12.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.12.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.13.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.13.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.15.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.15.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.15.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.15.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.15.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.15.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.15.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.15.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.16.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.16.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.17.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.18.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.18.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.18.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.18.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.19.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.19.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.19.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.19.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.20.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.20.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.22.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.22.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.22.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.22.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.22.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.22.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.22.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.22.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.23.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.23.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.24.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.25.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.25.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.25.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.25.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.26.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.26.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.26.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.26.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.27.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.27.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.29.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.29.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.29.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.29.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.29.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.29.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.29.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.29.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.30.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.30.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.31.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.32.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.32.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.32.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.32.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.33.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.33.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.33.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.33.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.34.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.34.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.36.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.36.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.36.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.36.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.36.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.36.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.36.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.36.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.37.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.37.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.38.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.fc_before.weight - torch.Size([512, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.fc_after.weight - torch.Size([256, 512]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.0.weight - torch.Size([1, 256, 1, 1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.0.bias - torch.Size([1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.1.weight - torch.Size([1, 256, 1, 1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.1.bias - torch.Size([1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.2.weight - torch.Size([1, 256, 1, 1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.2.bias - torch.Size([1]): -The value is the same before and after calling `init_weights` of Sparse4D -2026-02-10 16:03:28,129 - mmdet - INFO - Model: -Sparse4D( - (img_backbone): ResNet( - (conv1): Conv2d(3, 64, kernel_size=(7, 7), stride=(2, 2), padding=(3, 3), bias=False) - (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (maxpool): MaxPool2d(kernel_size=3, stride=2, padding=1, dilation=1, ceil_mode=False) - (layer1): ResLayer( - (0): Bottleneck( - (conv1): Conv2d(64, 64, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (downsample): Sequential( - (0): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - ) - ) - (1): Bottleneck( - (conv1): Conv2d(256, 64, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (2): Bottleneck( - (conv1): Conv2d(256, 64, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - ) - (layer2): ResLayer( - (0): Bottleneck( - (conv1): Conv2d(256, 128, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (downsample): Sequential( - (0): Conv2d(256, 512, kernel_size=(1, 1), stride=(2, 2), bias=False) - (1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - ) - ) - (1): Bottleneck( - (conv1): Conv2d(512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (2): Bottleneck( - (conv1): Conv2d(512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (3): Bottleneck( - (conv1): Conv2d(512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - ) - (layer3): ResLayer( - (0): Bottleneck( - (conv1): Conv2d(512, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (downsample): Sequential( - (0): Conv2d(512, 1024, kernel_size=(1, 1), stride=(2, 2), bias=False) - (1): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - ) - ) - (1): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (2): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (3): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (4): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (5): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - ) - (layer4): ResLayer( - (0): Bottleneck( - (conv1): Conv2d(1024, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(512, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (downsample): Sequential( - (0): Conv2d(1024, 2048, kernel_size=(1, 1), stride=(2, 2), bias=False) - (1): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - ) - ) - (1): Bottleneck( - (conv1): Conv2d(2048, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(512, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (2): Bottleneck( - (conv1): Conv2d(2048, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(512, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - ) - ) - init_cfg={'type': 'Pretrained', 'checkpoint': 'ckpt/resnet50-19c8e357.pth'} - (img_neck): FPN( - (lateral_convs): ModuleList( - (0): ConvModule( - (conv): Conv2d(256, 256, kernel_size=(1, 1), stride=(1, 1)) - ) - (1): ConvModule( - (conv): Conv2d(512, 256, kernel_size=(1, 1), stride=(1, 1)) - ) - (2): ConvModule( - (conv): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1)) - ) - (3): ConvModule( - (conv): Conv2d(2048, 256, kernel_size=(1, 1), stride=(1, 1)) - ) - ) - (fpn_convs): ModuleList( - (0-3): 4 x ConvModule( - (conv): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1)) - ) - ) - ) - init_cfg={'type': 'Xavier', 'layer': 'Conv2d', 'distribution': 'uniform'} - (head): Sparse4DHead( - (instance_bank): InstanceBank( - (anchor_handler): SparseBox3DKeyPointsGenerator() - ) - (anchor_encoder): SparseBox3DEncoder( - (pos_fc): Sequential( - (0): Linear(in_features=3, out_features=128, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((128,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=128, out_features=128, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((128,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=128, out_features=128, bias=True) - (7): ReLU(inplace=True) - (8): LayerNorm((128,), eps=1e-05, elementwise_affine=True) - (9): Linear(in_features=128, out_features=128, bias=True) - (10): ReLU(inplace=True) - (11): LayerNorm((128,), eps=1e-05, elementwise_affine=True) - ) - (size_fc): Sequential( - (0): Linear(in_features=3, out_features=32, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=32, out_features=32, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=32, out_features=32, bias=True) - (7): ReLU(inplace=True) - (8): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (9): Linear(in_features=32, out_features=32, bias=True) - (10): ReLU(inplace=True) - (11): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - ) - (yaw_fc): Sequential( - (0): Linear(in_features=2, out_features=32, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=32, out_features=32, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=32, out_features=32, bias=True) - (7): ReLU(inplace=True) - (8): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (9): Linear(in_features=32, out_features=32, bias=True) - (10): ReLU(inplace=True) - (11): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - ) - (vel_fc): Sequential( - (0): Linear(in_features=3, out_features=64, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((64,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=64, out_features=64, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((64,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=64, out_features=64, bias=True) - (7): ReLU(inplace=True) - (8): LayerNorm((64,), eps=1e-05, elementwise_affine=True) - (9): Linear(in_features=64, out_features=64, bias=True) - (10): ReLU(inplace=True) - (11): LayerNorm((64,), eps=1e-05, elementwise_affine=True) - ) - ) - (loss_cls): FocalLoss() - (loss_reg): SparseBox3DLoss( - (loss_box): L1Loss() - (loss_cns): CrossEntropyLoss(avg_non_ignore=False) - (loss_yns): GaussianFocalLoss() - ) - (layers): ModuleList( - (0): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (1): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (4-5): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (6): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (7): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (8): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (11-12): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (13): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (14): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (15): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (16): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (17): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (18-19): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (20): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (21): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (22): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (23): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (24): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (25-26): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (27): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (28): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (29): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (30): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (31): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (32-33): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (34): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (35): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (36): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (37): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (38): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - ) - (fc_before): Linear(in_features=256, out_features=512, bias=False) - (fc_after): Linear(in_features=512, out_features=256, bias=False) - ) - (depth_branch): DenseDepthNet( - (depth_layers): ModuleList( - (0-2): 3 x Conv2d(256, 1, kernel_size=(1, 1), stride=(1, 1)) - ) - ) - (grid_mask): GridMask() -) -2026-02-10 16:03:41,847 - mmdet - INFO - Start running, host: root@bw28, work_dir: /workspace/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704 -2026-02-10 16:03:41,847 - mmdet - INFO - Hooks will be executed in the following order: -before_run: -(VERY_HIGH ) CosineAnnealingLrUpdaterHook -(ABOVE_NORMAL) Fp16OptimizerHook -(NORMAL ) CheckpointHook -(NORMAL ) EvalHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -before_train_epoch: -(VERY_HIGH ) CosineAnnealingLrUpdaterHook -(NORMAL ) EvalHook -(LOW ) IterTimerHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -before_train_iter: -(VERY_HIGH ) CosineAnnealingLrUpdaterHook -(NORMAL ) EvalHook -(LOW ) IterTimerHook - -------------------- -after_train_iter: -(ABOVE_NORMAL) Fp16OptimizerHook -(NORMAL ) CheckpointHook -(NORMAL ) EvalHook -(NORMAL ) ProfilerHook -(LOW ) IterTimerHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -after_train_epoch: -(NORMAL ) CheckpointHook -(NORMAL ) EvalHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -before_val_epoch: -(LOW ) IterTimerHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -before_val_iter: -(LOW ) IterTimerHook - -------------------- -after_val_iter: -(LOW ) IterTimerHook - -------------------- -after_val_epoch: -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -after_run: -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -2026-02-10 16:03:41,847 - mmdet - INFO - workflow: [('train', 1)], max: 17500 iters -2026-02-10 16:03:41,850 - mmdet - INFO - Checkpoints will be saved to /workspace/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704 by HardDiskBackend. diff --git a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20260210_160324.log.json b/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20260210_160324.log.json deleted file mode 100644 index f691fdae6b2eb0b287f959b7a72d73034a4da3e6..0000000000000000000000000000000000000000 --- a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20260210_160324.log.json +++ /dev/null @@ -1 +0,0 @@ -{"env_info": "sys.platform: linux\nPython: 3.10.12 (main, Aug 15 2025, 14:32:43) [GCC 11.4.0]\nCUDA available: True\nGPU 0,1,2,3,4,5,6,7: BW200_LC, UBB BW1000_LC\nCUDA_HOME: /opt/dtk\nNVCC: Not Available\nGCC: x86_64-linux-gnu-gcc (Ubuntu 11.4.0-1ubuntu1~22.04) 11.4.0\nPyTorch: 2.5.1\nPyTorch compiling details: PyTorch built with:\n - GCC 10.3\n - C++ Version: 201703\n - Intel(R) Math Kernel Library Version 2020.0.4 Product Build 20200917 for Intel(R) 64 architecture applications\n - OpenMP 201511 (a.k.a. OpenMP 4.5)\n - LAPACK is enabled (usually provided by MKL)\n - NNPACK is enabled\n - CPU capability usage: AVX2\n - HIP Runtime 6.3.25521\n - MIOpen 2.18.0\n - Magma 2.8.0\n - Build settings: BLAS_INFO=mkl, BUILD_TYPE=Release, CXX_COMPILER=/opt/rh/gcc-toolset-10/root/usr/bin/c++, CXX_FLAGS= -D_GLIBCXX_USE_CXX11_ABI=1 -fvisibility-inlines-hidden -DUSE_PTHREADPOOL -DNDEBUG -DUSE_KINETO -DLIBKINETO_NOCUPTI -DLIBKINETO_NOXPUPTI=ON -DUSE_FBGEMM -DUSE_PYTORCH_QNNPACK -DUSE_XNNPACK -DSYMBOLICATE_MOBILE_DEBUG_HANDLE -O2 -fPIC -Wall -Wextra -Werror=return-type -Werror=non-virtual-dtor -Werror=bool-operation -Wnarrowing -Wno-missing-field-initializers -Wno-type-limits -Wno-array-bounds -Wno-unknown-pragmas -Wno-unused-parameter -Wno-strict-overflow -Wno-strict-aliasing -Wno-stringop-overflow -Wsuggest-override -Wno-psabi -Wno-error=old-style-cast -Wno-missing-braces -fdiagnostics-color=always -faligned-new -Wno-unused-but-set-variable -Wno-maybe-uninitialized -fno-math-errno -fno-trapping-math -Werror=format -Wno-stringop-overflow, FORCE_FALLBACK_CUDA_MPI=1, LAPACK_INFO=mkl, PERF_WITH_AVX=1, PERF_WITH_AVX2=1, TORCH_VERSION=2.5.1, USE_CUDA=0, USE_CUDNN=OFF, USE_CUSPARSELT=OFF, USE_EXCEPTION_PTR=1, USE_GFLAGS=1, USE_GLOG=1, USE_GLOO=1, USE_MKL=ON, USE_MKLDNN=0, USE_MPI=1, USE_NCCL=1, USE_NNPACK=ON, USE_OPENMP=1, USE_ROCM=ON, USE_ROCM_KERNEL_ASSERT=OFF, \n\nTorchVision: 0.20.1\nOpenCV: 4.12.0\nMMCV: 1.6.1\nMMCV Compiler: GCC 10.3\nMMCV CUDA Compiler: rocm not available\nMMDetection: 2.25.1+", "config": "plugin = True\nplugin_dir = 'projects/mmdet3d_plugin/'\ndist_params = dict(backend='nccl')\nlog_level = 'INFO'\nwork_dir = './work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704'\ntotal_batch_size = 160\nnum_gpus = 8\nbatch_size = 20\nnum_iters_per_epoch = 175\nnum_epochs = 100\ncheckpoint_epoch_interval = 20\ncheckpoint_config = dict(interval=3500)\nlog_config = dict(\n interval=1,\n hooks=[\n dict(type='TextLoggerHook', by_epoch=False),\n dict(type='TensorboardLoggerHook')\n ])\nload_from = None\nresume_from = None\nworkflow = [('train', 1)]\nfp16 = dict(loss_scale=32.0)\ninput_shape = (704, 256)\ntracking_test = True\ntracking_threshold = 0.2\nclass_names = [\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier',\n 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n]\nnum_classes = 10\nembed_dims = 256\nnum_groups = 8\nnum_decoder = 6\nnum_single_frame_decoder = 1\nuse_deformable_func = True\nstrides = [4, 8, 16, 32]\nnum_levels = 4\nnum_depth_layers = 3\ndrop_out = 0.1\ntemporal = True\ndecouple_attn = True\nwith_quality_estimation = True\nmodel = dict(\n type='Sparse4D',\n use_grid_mask=True,\n use_deformable_func=True,\n img_backbone=dict(\n type='ResNet',\n depth=50,\n num_stages=4,\n frozen_stages=-1,\n norm_eval=False,\n style='pytorch',\n with_cp=True,\n out_indices=(0, 1, 2, 3),\n norm_cfg=dict(type='BN', requires_grad=True),\n pretrained='ckpt/resnet50-19c8e357.pth'),\n img_neck=dict(\n type='FPN',\n num_outs=4,\n start_level=0,\n out_channels=256,\n add_extra_convs='on_output',\n relu_before_extra_convs=True,\n in_channels=[256, 512, 1024, 2048]),\n depth_branch=dict(\n type='DenseDepthNet',\n embed_dims=256,\n num_depth_layers=3,\n loss_weight=0.2),\n head=dict(\n type='Sparse4DHead',\n cls_threshold_to_reg=0.05,\n decouple_attn=True,\n instance_bank=dict(\n type='InstanceBank',\n num_anchor=900,\n embed_dims=256,\n anchor='nuscenes_kmeans900.npy',\n anchor_handler=dict(type='SparseBox3DKeyPointsGenerator'),\n num_temp_instances=600,\n confidence_decay=0.6,\n feat_grad=False),\n anchor_encoder=dict(\n type='SparseBox3DEncoder',\n vel_dims=3,\n embed_dims=[128, 32, 32, 64],\n mode='cat',\n output_fc=False,\n in_loops=1,\n out_loops=4),\n num_single_frame_decoder=1,\n operation_order=[\n 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm',\n 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm',\n 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm',\n 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm',\n 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm',\n 'deformable', 'ffn', 'norm', 'refine'\n ],\n temp_graph_model=dict(\n type='MultiheadAttention',\n embed_dims=512,\n num_heads=8,\n batch_first=True,\n dropout=0.1),\n graph_model=dict(\n type='MultiheadAttention',\n embed_dims=512,\n num_heads=8,\n batch_first=True,\n dropout=0.1),\n norm_layer=dict(type='LN', normalized_shape=256),\n ffn=dict(\n type='AsymmetricFFN',\n in_channels=512,\n pre_norm=dict(type='LN'),\n embed_dims=256,\n feedforward_channels=1024,\n num_fcs=2,\n ffn_drop=0.1,\n act_cfg=dict(type='ReLU', inplace=True)),\n deformable_model=dict(\n type='DeformableFeatureAggregation',\n embed_dims=256,\n num_groups=8,\n num_levels=4,\n num_cams=6,\n attn_drop=0.15,\n use_deformable_func=True,\n use_camera_embed=True,\n residual_mode='cat',\n kps_generator=dict(\n type='SparseBox3DKeyPointsGenerator',\n num_learnable_pts=6,\n fix_scale=[[0, 0, 0], [0.45, 0, 0], [-0.45, 0, 0],\n [0, 0.45, 0], [0, -0.45, 0], [0, 0, 0.45],\n [0, 0, -0.45]])),\n refine_layer=dict(\n type='SparseBox3DRefinementModule',\n embed_dims=256,\n num_cls=10,\n refine_yaw=True,\n with_quality_estimation=True),\n sampler=dict(\n type='SparseBox3DTarget',\n num_dn_groups=5,\n num_temp_dn_groups=3,\n dn_noise_scale=[2.0, 2.0, 2.0, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5],\n max_dn_gt=32,\n add_neg_dn=True,\n cls_weight=2.0,\n box_weight=0.25,\n reg_weights=[2.0, 2.0, 2.0, 0.5, 0.5, 0.5, 0.0, 0.0, 0.0, 0.0],\n cls_wise_reg_weights=dict(\n {9: [2.0, 2.0, 2.0, 1.0, 1.0, 1.0, 0.0, 0.0, 1.0, 1.0]})),\n loss_cls=dict(\n type='FocalLoss',\n use_sigmoid=True,\n gamma=2.0,\n alpha=0.25,\n loss_weight=2.0),\n loss_reg=dict(\n type='SparseBox3DLoss',\n loss_box=dict(type='L1Loss', loss_weight=0.25),\n loss_centerness=dict(type='CrossEntropyLoss', use_sigmoid=True),\n loss_yawness=dict(type='GaussianFocalLoss'),\n cls_allow_reverse=[5]),\n decoder=dict(type='SparseBox3DDecoder'),\n reg_weights=[2.0, 2.0, 2.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0]))\ndataset_type = 'NuScenes3DDetTrackDataset'\ndata_root = 'data/nuscenes/'\nanno_root = 'data/nuscenes_anno_pkls/'\nfile_client_args = dict(backend='disk')\nimg_norm_cfg = dict(\n mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375], to_rgb=True)\ntrain_pipeline = [\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(\n type='LoadPointsFromFile',\n coord_type='LIDAR',\n load_dim=5,\n use_dim=5,\n file_client_args=dict(backend='disk')),\n dict(type='ResizeCropFlipImage'),\n dict(type='MultiScaleDepthMapGenerator', downsample=[4, 8, 16]),\n dict(type='BBoxRotation'),\n dict(type='PhotoMetricDistortionMultiViewImage'),\n dict(\n type='NormalizeMultiviewImage',\n mean=[123.675, 116.28, 103.53],\n std=[58.395, 57.12, 57.375],\n to_rgb=True),\n dict(\n type='CircleObjectRangeFilter',\n class_dist_thred=[55, 55, 55, 55, 55, 55, 55, 55, 55, 55]),\n dict(\n type='InstanceNameFilter',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ]),\n dict(type='NuScenesSparse4DAdaptor'),\n dict(\n type='Collect',\n keys=[\n 'img', 'timestamp', 'projection_mat', 'image_wh', 'gt_depth',\n 'focal', 'gt_bboxes_3d', 'gt_labels_3d'\n ],\n meta_keys=['T_global', 'T_global_inv', 'timestamp', 'instance_id'])\n]\ntest_pipeline = [\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(type='ResizeCropFlipImage'),\n dict(\n type='NormalizeMultiviewImage',\n mean=[123.675, 116.28, 103.53],\n std=[58.395, 57.12, 57.375],\n to_rgb=True),\n dict(type='NuScenesSparse4DAdaptor'),\n dict(\n type='Collect',\n keys=['img', 'timestamp', 'projection_mat', 'image_wh'],\n meta_keys=['T_global', 'T_global_inv', 'timestamp'])\n]\ninput_modality = dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False)\ndata_basic_config = dict(\n type='NuScenes3DDetTrackDataset',\n data_root='data/nuscenes/',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier',\n 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ],\n modality=dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False),\n version='v1.0-trainval')\ndata_aug_conf = dict(\n resize_lim=(0.4, 0.47),\n final_dim=(256, 704),\n bot_pct_lim=(0.0, 0.0),\n rot_lim=(-5.4, 5.4),\n H=900,\n W=1600,\n rand_flip=True,\n rot3d_range=[-0.3925, 0.3925])\ndata = dict(\n samples_per_gpu=20,\n workers_per_gpu=20,\n train=dict(\n type='NuScenes3DDetTrackDataset',\n data_root='data/nuscenes/',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ],\n modality=dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False),\n version='v1.0-trainval',\n ann_file='data/nuscenes_anno_pkls/nuscenes_infos_train.pkl',\n pipeline=[\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(\n type='LoadPointsFromFile',\n coord_type='LIDAR',\n load_dim=5,\n use_dim=5,\n file_client_args=dict(backend='disk')),\n dict(type='ResizeCropFlipImage'),\n dict(type='MultiScaleDepthMapGenerator', downsample=[4, 8, 16]),\n dict(type='BBoxRotation'),\n dict(type='PhotoMetricDistortionMultiViewImage'),\n dict(\n type='NormalizeMultiviewImage',\n mean=[123.675, 116.28, 103.53],\n std=[58.395, 57.12, 57.375],\n to_rgb=True),\n dict(\n type='CircleObjectRangeFilter',\n class_dist_thred=[55, 55, 55, 55, 55, 55, 55, 55, 55, 55]),\n dict(\n type='InstanceNameFilter',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian',\n 'traffic_cone'\n ]),\n dict(type='NuScenesSparse4DAdaptor'),\n dict(\n type='Collect',\n keys=[\n 'img', 'timestamp', 'projection_mat', 'image_wh',\n 'gt_depth', 'focal', 'gt_bboxes_3d', 'gt_labels_3d'\n ],\n meta_keys=[\n 'T_global', 'T_global_inv', 'timestamp', 'instance_id'\n ])\n ],\n test_mode=False,\n data_aug_conf=dict(\n resize_lim=(0.4, 0.47),\n final_dim=(256, 704),\n bot_pct_lim=(0.0, 0.0),\n rot_lim=(-5.4, 5.4),\n H=900,\n W=1600,\n rand_flip=True,\n rot3d_range=[-0.3925, 0.3925]),\n with_seq_flag=True,\n sequences_split_num=2,\n keep_consistent_seq_aug=True),\n val=dict(\n type='NuScenes3DDetTrackDataset',\n data_root='data/nuscenes/',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ],\n modality=dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False),\n version='v1.0-trainval',\n ann_file='data/nuscenes_anno_pkls/nuscenes_infos_val.pkl',\n pipeline=[\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(type='ResizeCropFlipImage'),\n dict(\n type='NormalizeMultiviewImage',\n mean=[123.675, 116.28, 103.53],\n std=[58.395, 57.12, 57.375],\n to_rgb=True),\n dict(type='NuScenesSparse4DAdaptor'),\n dict(\n type='Collect',\n keys=['img', 'timestamp', 'projection_mat', 'image_wh'],\n meta_keys=['T_global', 'T_global_inv', 'timestamp'])\n ],\n data_aug_conf=dict(\n resize_lim=(0.4, 0.47),\n final_dim=(256, 704),\n bot_pct_lim=(0.0, 0.0),\n rot_lim=(-5.4, 5.4),\n H=900,\n W=1600,\n rand_flip=True,\n rot3d_range=[-0.3925, 0.3925]),\n test_mode=True,\n tracking=True,\n tracking_threshold=0.2),\n test=dict(\n type='NuScenes3DDetTrackDataset',\n data_root='data/nuscenes/',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ],\n modality=dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False),\n version='v1.0-trainval',\n ann_file='data/nuscenes_anno_pkls/nuscenes_infos_val.pkl',\n pipeline=[\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(type='ResizeCropFlipImage'),\n dict(\n type='NormalizeMultiviewImage',\n mean=[123.675, 116.28, 103.53],\n std=[58.395, 57.12, 57.375],\n to_rgb=True),\n dict(type='NuScenesSparse4DAdaptor'),\n dict(\n type='Collect',\n keys=['img', 'timestamp', 'projection_mat', 'image_wh'],\n meta_keys=['T_global', 'T_global_inv', 'timestamp'])\n ],\n data_aug_conf=dict(\n resize_lim=(0.4, 0.47),\n final_dim=(256, 704),\n bot_pct_lim=(0.0, 0.0),\n rot_lim=(-5.4, 5.4),\n H=900,\n W=1600,\n rand_flip=True,\n rot3d_range=[-0.3925, 0.3925]),\n test_mode=True,\n tracking=True,\n tracking_threshold=0.2))\noptimizer = dict(\n type='AdamW',\n lr=0.0006,\n weight_decay=0.001,\n paramwise_cfg=dict(custom_keys=dict(img_backbone=dict(lr_mult=0.5))))\noptimizer_config = dict(grad_clip=dict(max_norm=25, norm_type=2))\nlr_config = dict(\n policy='CosineAnnealing',\n warmup='linear',\n warmup_iters=500,\n warmup_ratio=0.3333333333333333,\n min_lr_ratio=0.001)\nrunner = dict(type='IterBasedRunner', max_iters=17500)\nvis_pipeline = [\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(type='Collect', keys=['img'], meta_keys=['timestamp', 'lidar2img'])\n]\nevaluation = dict(\n interval=3500,\n pipeline=[\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(\n type='Collect', keys=['img'], meta_keys=['timestamp', 'lidar2img'])\n ])\ngpu_ids = range(0, 1)\n", "seed": 0, "exp_name": "sparse4dv3_temporal_r50_1x8_bs6_256x704.py"} diff --git a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20260210_160806.log b/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20260210_160806.log deleted file mode 100644 index 5611c03440dca0cff3746be3ebb97bba308e4b00..0000000000000000000000000000000000000000 --- a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20260210_160806.log +++ /dev/null @@ -1,3225 +0,0 @@ -2026-02-10 16:08:06,732 - mmdet - INFO - Environment info: ------------------------------------------------------------- -sys.platform: linux -Python: 3.10.12 (main, Aug 15 2025, 14:32:43) [GCC 11.4.0] -CUDA available: True -GPU 0,1,2,3,4,5,6,7: BW200_LC, UBB BW1000_LC -CUDA_HOME: /opt/dtk/cuda/cuda -NVCC: Cuda compilation tools, release 12.6, V12.6.77 -clang version 17.0.0 -Target: x86_64-unknown-linux-gnu -Thread model: posix -InstalledDir: /opt/dtk-25.04.4/llvm/bi -GCC: x86_64-linux-gnu-gcc (Ubuntu 11.4.0-1ubuntu1~22.04) 11.4.0 -PyTorch: 2.5.1 -PyTorch compiling details: PyTorch built with: - - GCC 10.3 - - C++ Version: 201703 - - Intel(R) Math Kernel Library Version 2020.0.4 Product Build 20200917 for Intel(R) 64 architecture applications - - OpenMP 201511 (a.k.a. OpenMP 4.5) - - LAPACK is enabled (usually provided by MKL) - - NNPACK is enabled - - CPU capability usage: AVX2 - - HIP Runtime 6.3.25521 - - MIOpen 2.18.0 - - Magma 2.8.0 - - Build settings: BLAS_INFO=mkl, BUILD_TYPE=Release, CXX_COMPILER=/opt/rh/gcc-toolset-10/root/usr/bin/c++, CXX_FLAGS= -D_GLIBCXX_USE_CXX11_ABI=1 -fvisibility-inlines-hidden -DUSE_PTHREADPOOL -DNDEBUG -DUSE_KINETO -DLIBKINETO_NOCUPTI -DLIBKINETO_NOXPUPTI=ON -DUSE_FBGEMM -DUSE_PYTORCH_QNNPACK -DUSE_XNNPACK -DSYMBOLICATE_MOBILE_DEBUG_HANDLE -O2 -fPIC -Wall -Wextra -Werror=return-type -Werror=non-virtual-dtor -Werror=bool-operation -Wnarrowing -Wno-missing-field-initializers -Wno-type-limits -Wno-array-bounds -Wno-unknown-pragmas -Wno-unused-parameter -Wno-strict-overflow -Wno-strict-aliasing -Wno-stringop-overflow -Wsuggest-override -Wno-psabi -Wno-error=old-style-cast -Wno-missing-braces -fdiagnostics-color=always -faligned-new -Wno-unused-but-set-variable -Wno-maybe-uninitialized -fno-math-errno -fno-trapping-math -Werror=format -Wno-stringop-overflow, FORCE_FALLBACK_CUDA_MPI=1, LAPACK_INFO=mkl, PERF_WITH_AVX=1, PERF_WITH_AVX2=1, TORCH_VERSION=2.5.1, USE_CUDA=0, USE_CUDNN=OFF, USE_CUSPARSELT=OFF, USE_EXCEPTION_PTR=1, USE_GFLAGS=1, USE_GLOG=1, USE_GLOO=1, USE_MKL=ON, USE_MKLDNN=0, USE_MPI=1, USE_NCCL=1, USE_NNPACK=ON, USE_OPENMP=1, USE_ROCM=ON, USE_ROCM_KERNEL_ASSERT=OFF, - -TorchVision: 0.20.1 -OpenCV: 4.12.0 -MMCV: 1.6.1 -MMCV Compiler: GCC 10.3 -MMCV CUDA Compiler: rocm not available -MMDetection: 2.25.1+ ------------------------------------------------------------- - -2026-02-10 16:08:07,744 - mmdet - INFO - Distributed training: False -2026-02-10 16:08:08,962 - mmdet - INFO - Config: -plugin = True -plugin_dir = 'projects/mmdet3d_plugin/' -dist_params = dict(backend='nccl') -log_level = 'INFO' -work_dir = './work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704' -total_batch_size = 160 -num_gpus = 8 -batch_size = 20 -num_iters_per_epoch = 175 -num_epochs = 100 -checkpoint_epoch_interval = 20 -checkpoint_config = dict(interval=3500) -log_config = dict( - interval=1, - hooks=[ - dict(type='TextLoggerHook', by_epoch=False), - dict(type='TensorboardLoggerHook') - ]) -load_from = None -resume_from = None -workflow = [('train', 1)] -fp16 = dict(loss_scale=32.0) -input_shape = (704, 256) -tracking_test = True -tracking_threshold = 0.2 -class_names = [ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier', - 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' -] -num_classes = 10 -embed_dims = 256 -num_groups = 8 -num_decoder = 6 -num_single_frame_decoder = 1 -use_deformable_func = True -strides = [4, 8, 16, 32] -num_levels = 4 -num_depth_layers = 3 -drop_out = 0.1 -temporal = True -decouple_attn = True -with_quality_estimation = True -model = dict( - type='Sparse4D', - use_grid_mask=True, - use_deformable_func=True, - img_backbone=dict( - type='ResNet', - depth=50, - num_stages=4, - frozen_stages=-1, - norm_eval=False, - style='pytorch', - with_cp=True, - out_indices=(0, 1, 2, 3), - norm_cfg=dict(type='BN', requires_grad=True), - pretrained='ckpt/resnet50-19c8e357.pth'), - img_neck=dict( - type='FPN', - num_outs=4, - start_level=0, - out_channels=256, - add_extra_convs='on_output', - relu_before_extra_convs=True, - in_channels=[256, 512, 1024, 2048]), - depth_branch=dict( - type='DenseDepthNet', - embed_dims=256, - num_depth_layers=3, - loss_weight=0.2), - head=dict( - type='Sparse4DHead', - cls_threshold_to_reg=0.05, - decouple_attn=True, - instance_bank=dict( - type='InstanceBank', - num_anchor=900, - embed_dims=256, - anchor='nuscenes_kmeans900.npy', - anchor_handler=dict(type='SparseBox3DKeyPointsGenerator'), - num_temp_instances=600, - confidence_decay=0.6, - feat_grad=False), - anchor_encoder=dict( - type='SparseBox3DEncoder', - vel_dims=3, - embed_dims=[128, 32, 32, 64], - mode='cat', - output_fc=False, - in_loops=1, - out_loops=4), - num_single_frame_decoder=1, - operation_order=[ - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine' - ], - temp_graph_model=dict( - type='MultiheadAttention', - embed_dims=512, - num_heads=8, - batch_first=True, - dropout=0.1), - graph_model=dict( - type='MultiheadAttention', - embed_dims=512, - num_heads=8, - batch_first=True, - dropout=0.1), - norm_layer=dict(type='LN', normalized_shape=256), - ffn=dict( - type='AsymmetricFFN', - in_channels=512, - pre_norm=dict(type='LN'), - embed_dims=256, - feedforward_channels=1024, - num_fcs=2, - ffn_drop=0.1, - act_cfg=dict(type='ReLU', inplace=True)), - deformable_model=dict( - type='DeformableFeatureAggregation', - embed_dims=256, - num_groups=8, - num_levels=4, - num_cams=6, - attn_drop=0.15, - use_deformable_func=True, - use_camera_embed=True, - residual_mode='cat', - kps_generator=dict( - type='SparseBox3DKeyPointsGenerator', - num_learnable_pts=6, - fix_scale=[[0, 0, 0], [0.45, 0, 0], [-0.45, 0, 0], - [0, 0.45, 0], [0, -0.45, 0], [0, 0, 0.45], - [0, 0, -0.45]])), - refine_layer=dict( - type='SparseBox3DRefinementModule', - embed_dims=256, - num_cls=10, - refine_yaw=True, - with_quality_estimation=True), - sampler=dict( - type='SparseBox3DTarget', - num_dn_groups=5, - num_temp_dn_groups=3, - dn_noise_scale=[2.0, 2.0, 2.0, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5], - max_dn_gt=32, - add_neg_dn=True, - cls_weight=2.0, - box_weight=0.25, - reg_weights=[2.0, 2.0, 2.0, 0.5, 0.5, 0.5, 0.0, 0.0, 0.0, 0.0], - cls_wise_reg_weights=dict( - {9: [2.0, 2.0, 2.0, 1.0, 1.0, 1.0, 0.0, 0.0, 1.0, 1.0]})), - loss_cls=dict( - type='FocalLoss', - use_sigmoid=True, - gamma=2.0, - alpha=0.25, - loss_weight=2.0), - loss_reg=dict( - type='SparseBox3DLoss', - loss_box=dict(type='L1Loss', loss_weight=0.25), - loss_centerness=dict(type='CrossEntropyLoss', use_sigmoid=True), - loss_yawness=dict(type='GaussianFocalLoss'), - cls_allow_reverse=[5]), - decoder=dict(type='SparseBox3DDecoder'), - reg_weights=[2.0, 2.0, 2.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0])) -dataset_type = 'NuScenes3DDetTrackDataset' -data_root = 'data/nuscenes/' -anno_root = 'data/nuscenes_anno_pkls/' -file_client_args = dict(backend='disk') -img_norm_cfg = dict( - mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375], to_rgb=True) -train_pipeline = [ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict( - type='LoadPointsFromFile', - coord_type='LIDAR', - load_dim=5, - use_dim=5, - file_client_args=dict(backend='disk')), - dict(type='ResizeCropFlipImage'), - dict(type='MultiScaleDepthMapGenerator', downsample=[4, 8, 16]), - dict(type='BBoxRotation'), - dict(type='PhotoMetricDistortionMultiViewImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict( - type='CircleObjectRangeFilter', - class_dist_thred=[55, 55, 55, 55, 55, 55, 55, 55, 55, 55]), - dict( - type='InstanceNameFilter', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ]), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=[ - 'img', 'timestamp', 'projection_mat', 'image_wh', 'gt_depth', - 'focal', 'gt_bboxes_3d', 'gt_labels_3d' - ], - meta_keys=['T_global', 'T_global_inv', 'timestamp', 'instance_id']) -] -test_pipeline = [ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='ResizeCropFlipImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=['img', 'timestamp', 'projection_mat', 'image_wh'], - meta_keys=['T_global', 'T_global_inv', 'timestamp']) -] -input_modality = dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False) -data_basic_config = dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier', - 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval') -data_aug_conf = dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]) -data = dict( - samples_per_gpu=20, - workers_per_gpu=20, - train=dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval', - ann_file='data/nuscenes_anno_pkls/nuscenes_infos_train.pkl', - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict( - type='LoadPointsFromFile', - coord_type='LIDAR', - load_dim=5, - use_dim=5, - file_client_args=dict(backend='disk')), - dict(type='ResizeCropFlipImage'), - dict(type='MultiScaleDepthMapGenerator', downsample=[4, 8, 16]), - dict(type='BBoxRotation'), - dict(type='PhotoMetricDistortionMultiViewImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict( - type='CircleObjectRangeFilter', - class_dist_thred=[55, 55, 55, 55, 55, 55, 55, 55, 55, 55]), - dict( - type='InstanceNameFilter', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', - 'traffic_cone' - ]), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=[ - 'img', 'timestamp', 'projection_mat', 'image_wh', - 'gt_depth', 'focal', 'gt_bboxes_3d', 'gt_labels_3d' - ], - meta_keys=[ - 'T_global', 'T_global_inv', 'timestamp', 'instance_id' - ]) - ], - test_mode=False, - data_aug_conf=dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]), - with_seq_flag=True, - sequences_split_num=2, - keep_consistent_seq_aug=True), - val=dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval', - ann_file='data/nuscenes_anno_pkls/nuscenes_infos_val.pkl', - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='ResizeCropFlipImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=['img', 'timestamp', 'projection_mat', 'image_wh'], - meta_keys=['T_global', 'T_global_inv', 'timestamp']) - ], - data_aug_conf=dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]), - test_mode=True, - tracking=True, - tracking_threshold=0.2), - test=dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval', - ann_file='data/nuscenes_anno_pkls/nuscenes_infos_val.pkl', - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='ResizeCropFlipImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=['img', 'timestamp', 'projection_mat', 'image_wh'], - meta_keys=['T_global', 'T_global_inv', 'timestamp']) - ], - data_aug_conf=dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]), - test_mode=True, - tracking=True, - tracking_threshold=0.2)) -optimizer = dict( - type='AdamW', - lr=0.0006, - weight_decay=0.001, - paramwise_cfg=dict(custom_keys=dict(img_backbone=dict(lr_mult=0.5)))) -optimizer_config = dict(grad_clip=dict(max_norm=25, norm_type=2)) -lr_config = dict( - policy='CosineAnnealing', - warmup='linear', - warmup_iters=500, - warmup_ratio=0.3333333333333333, - min_lr_ratio=0.001) -runner = dict(type='IterBasedRunner', max_iters=17500) -vis_pipeline = [ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='Collect', keys=['img'], meta_keys=['timestamp', 'lidar2img']) -] -evaluation = dict( - interval=3500, - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict( - type='Collect', keys=['img'], meta_keys=['timestamp', 'lidar2img']) - ]) -gpu_ids = range(0, 1) - -2026-02-10 16:08:08,963 - mmdet - INFO - Set random seed to 0, deterministic: False -2026-02-10 16:08:09,531 - mmdet - INFO - initialize ResNet with init_cfg {'type': 'Pretrained', 'checkpoint': 'ckpt/resnet50-19c8e357.pth'} -2026-02-10 16:08:09,760 - mmdet - INFO - initialize FPN with init_cfg {'type': 'Xavier', 'layer': 'Conv2d', 'distribution': 'uniform'} -Name of parameter - Initialization information - -img_backbone.conv1.weight - torch.Size([64, 3, 7, 7]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.bn1.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.bn1.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.conv1.weight - torch.Size([64, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn1.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn1.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.conv2.weight - torch.Size([64, 64, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn2.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn2.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.conv3.weight - torch.Size([256, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn3.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn3.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.downsample.0.weight - torch.Size([256, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.downsample.1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.downsample.1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.conv1.weight - torch.Size([64, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn1.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn1.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.conv2.weight - torch.Size([64, 64, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn2.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn2.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.conv3.weight - torch.Size([256, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn3.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn3.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.conv1.weight - torch.Size([64, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn1.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn1.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.conv2.weight - torch.Size([64, 64, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn2.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn2.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.conv3.weight - torch.Size([256, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn3.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn3.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.conv1.weight - torch.Size([128, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn1.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn1.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.conv2.weight - torch.Size([128, 128, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn2.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn2.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.conv3.weight - torch.Size([512, 128, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn3.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn3.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.downsample.0.weight - torch.Size([512, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.downsample.1.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.downsample.1.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.conv1.weight - torch.Size([128, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn1.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn1.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.conv2.weight - torch.Size([128, 128, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn2.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn2.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.conv3.weight - torch.Size([512, 128, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn3.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn3.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.conv1.weight - torch.Size([128, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn1.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn1.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.conv2.weight - torch.Size([128, 128, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn2.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn2.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.conv3.weight - torch.Size([512, 128, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn3.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn3.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.conv1.weight - torch.Size([128, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn1.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn1.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.conv2.weight - torch.Size([128, 128, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn2.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn2.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.conv3.weight - torch.Size([512, 128, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn3.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn3.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.conv1.weight - torch.Size([256, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.downsample.0.weight - torch.Size([1024, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.downsample.1.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.downsample.1.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.conv1.weight - torch.Size([512, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn1.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn1.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.conv2.weight - torch.Size([512, 512, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn2.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn2.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.conv3.weight - torch.Size([2048, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn3.weight - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn3.bias - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.downsample.0.weight - torch.Size([2048, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.downsample.1.weight - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.downsample.1.bias - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.conv1.weight - torch.Size([512, 2048, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn1.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn1.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.conv2.weight - torch.Size([512, 512, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn2.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn2.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.conv3.weight - torch.Size([2048, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn3.weight - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn3.bias - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.conv1.weight - torch.Size([512, 2048, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn1.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn1.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.conv2.weight - torch.Size([512, 512, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn2.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn2.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.conv3.weight - torch.Size([2048, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn3.weight - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn3.bias - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_neck.lateral_convs.0.conv.weight - torch.Size([256, 256, 1, 1]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.lateral_convs.0.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.lateral_convs.1.conv.weight - torch.Size([256, 512, 1, 1]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.lateral_convs.1.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.lateral_convs.2.conv.weight - torch.Size([256, 1024, 1, 1]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.lateral_convs.2.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.lateral_convs.3.conv.weight - torch.Size([256, 2048, 1, 1]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.lateral_convs.3.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.fpn_convs.0.conv.weight - torch.Size([256, 256, 3, 3]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.fpn_convs.0.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.fpn_convs.1.conv.weight - torch.Size([256, 256, 3, 3]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.fpn_convs.1.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.fpn_convs.2.conv.weight - torch.Size([256, 256, 3, 3]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.fpn_convs.2.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.fpn_convs.3.conv.weight - torch.Size([256, 256, 3, 3]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.fpn_convs.3.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.instance_bank.anchor - torch.Size([900, 11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.instance_bank.instance_feature - torch.Size([900, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.instance_bank.anchor_handler.fix_scale - torch.Size([1, 3]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.0.weight - torch.Size([128, 3]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.0.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.2.weight - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.2.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.3.weight - torch.Size([128, 128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.3.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.5.weight - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.5.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.6.weight - torch.Size([128, 128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.6.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.8.weight - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.8.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.9.weight - torch.Size([128, 128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.9.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.11.weight - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.11.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.0.weight - torch.Size([32, 3]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.0.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.2.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.2.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.3.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.3.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.5.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.5.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.6.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.6.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.8.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.8.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.9.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.9.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.11.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.11.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.0.weight - torch.Size([32, 2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.0.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.2.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.2.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.3.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.3.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.5.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.5.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.6.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.6.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.8.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.8.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.9.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.9.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.11.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.11.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.0.weight - torch.Size([64, 3]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.0.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.2.weight - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.2.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.3.weight - torch.Size([64, 64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.3.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.5.weight - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.5.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.6.weight - torch.Size([64, 64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.6.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.8.weight - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.8.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.9.weight - torch.Size([64, 64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.9.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.11.weight - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.11.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.1.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.1.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.1.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.1.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.1.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.1.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.1.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.1.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.3.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.4.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.4.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.4.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.4.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.5.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.5.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.5.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.5.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.6.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.6.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.8.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.8.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.8.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.8.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.8.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.8.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.8.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.8.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.10.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.11.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.11.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.11.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.11.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.12.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.12.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.12.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.12.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.13.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.13.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.15.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.15.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.15.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.15.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.15.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.15.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.15.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.15.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.16.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.16.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.17.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.18.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.18.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.18.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.18.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.19.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.19.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.19.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.19.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.20.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.20.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.22.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.22.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.22.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.22.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.22.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.22.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.22.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.22.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.23.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.23.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.24.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.25.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.25.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.25.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.25.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.26.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.26.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.26.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.26.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.27.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.27.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.29.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.29.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.29.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.29.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.29.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.29.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.29.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.29.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.30.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.30.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.31.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.32.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.32.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.32.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.32.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.33.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.33.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.33.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.33.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.34.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.34.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.36.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.36.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.36.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.36.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.36.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.36.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.36.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.36.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.37.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.37.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.38.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.fc_before.weight - torch.Size([512, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.fc_after.weight - torch.Size([256, 512]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.0.weight - torch.Size([1, 256, 1, 1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.0.bias - torch.Size([1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.1.weight - torch.Size([1, 256, 1, 1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.1.bias - torch.Size([1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.2.weight - torch.Size([1, 256, 1, 1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.2.bias - torch.Size([1]): -The value is the same before and after calling `init_weights` of Sparse4D -2026-02-10 16:08:10,175 - mmdet - INFO - Model: -Sparse4D( - (img_backbone): ResNet( - (conv1): Conv2d(3, 64, kernel_size=(7, 7), stride=(2, 2), padding=(3, 3), bias=False) - (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (maxpool): MaxPool2d(kernel_size=3, stride=2, padding=1, dilation=1, ceil_mode=False) - (layer1): ResLayer( - (0): Bottleneck( - (conv1): Conv2d(64, 64, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (downsample): Sequential( - (0): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - ) - ) - (1): Bottleneck( - (conv1): Conv2d(256, 64, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (2): Bottleneck( - (conv1): Conv2d(256, 64, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - ) - (layer2): ResLayer( - (0): Bottleneck( - (conv1): Conv2d(256, 128, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (downsample): Sequential( - (0): Conv2d(256, 512, kernel_size=(1, 1), stride=(2, 2), bias=False) - (1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - ) - ) - (1): Bottleneck( - (conv1): Conv2d(512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (2): Bottleneck( - (conv1): Conv2d(512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (3): Bottleneck( - (conv1): Conv2d(512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - ) - (layer3): ResLayer( - (0): Bottleneck( - (conv1): Conv2d(512, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (downsample): Sequential( - (0): Conv2d(512, 1024, kernel_size=(1, 1), stride=(2, 2), bias=False) - (1): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - ) - ) - (1): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (2): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (3): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (4): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (5): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - ) - (layer4): ResLayer( - (0): Bottleneck( - (conv1): Conv2d(1024, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(512, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (downsample): Sequential( - (0): Conv2d(1024, 2048, kernel_size=(1, 1), stride=(2, 2), bias=False) - (1): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - ) - ) - (1): Bottleneck( - (conv1): Conv2d(2048, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(512, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (2): Bottleneck( - (conv1): Conv2d(2048, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(512, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - ) - ) - init_cfg={'type': 'Pretrained', 'checkpoint': 'ckpt/resnet50-19c8e357.pth'} - (img_neck): FPN( - (lateral_convs): ModuleList( - (0): ConvModule( - (conv): Conv2d(256, 256, kernel_size=(1, 1), stride=(1, 1)) - ) - (1): ConvModule( - (conv): Conv2d(512, 256, kernel_size=(1, 1), stride=(1, 1)) - ) - (2): ConvModule( - (conv): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1)) - ) - (3): ConvModule( - (conv): Conv2d(2048, 256, kernel_size=(1, 1), stride=(1, 1)) - ) - ) - (fpn_convs): ModuleList( - (0-3): 4 x ConvModule( - (conv): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1)) - ) - ) - ) - init_cfg={'type': 'Xavier', 'layer': 'Conv2d', 'distribution': 'uniform'} - (head): Sparse4DHead( - (instance_bank): InstanceBank( - (anchor_handler): SparseBox3DKeyPointsGenerator() - ) - (anchor_encoder): SparseBox3DEncoder( - (pos_fc): Sequential( - (0): Linear(in_features=3, out_features=128, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((128,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=128, out_features=128, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((128,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=128, out_features=128, bias=True) - (7): ReLU(inplace=True) - (8): LayerNorm((128,), eps=1e-05, elementwise_affine=True) - (9): Linear(in_features=128, out_features=128, bias=True) - (10): ReLU(inplace=True) - (11): LayerNorm((128,), eps=1e-05, elementwise_affine=True) - ) - (size_fc): Sequential( - (0): Linear(in_features=3, out_features=32, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=32, out_features=32, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=32, out_features=32, bias=True) - (7): ReLU(inplace=True) - (8): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (9): Linear(in_features=32, out_features=32, bias=True) - (10): ReLU(inplace=True) - (11): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - ) - (yaw_fc): Sequential( - (0): Linear(in_features=2, out_features=32, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=32, out_features=32, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=32, out_features=32, bias=True) - (7): ReLU(inplace=True) - (8): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (9): Linear(in_features=32, out_features=32, bias=True) - (10): ReLU(inplace=True) - (11): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - ) - (vel_fc): Sequential( - (0): Linear(in_features=3, out_features=64, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((64,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=64, out_features=64, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((64,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=64, out_features=64, bias=True) - (7): ReLU(inplace=True) - (8): LayerNorm((64,), eps=1e-05, elementwise_affine=True) - (9): Linear(in_features=64, out_features=64, bias=True) - (10): ReLU(inplace=True) - (11): LayerNorm((64,), eps=1e-05, elementwise_affine=True) - ) - ) - (loss_cls): FocalLoss() - (loss_reg): SparseBox3DLoss( - (loss_box): L1Loss() - (loss_cns): CrossEntropyLoss(avg_non_ignore=False) - (loss_yns): GaussianFocalLoss() - ) - (layers): ModuleList( - (0): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (1): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (4-5): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (6): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (7): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (8): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (11-12): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (13): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (14): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (15): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (16): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (17): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (18-19): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (20): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (21): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (22): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (23): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (24): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (25-26): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (27): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (28): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (29): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (30): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (31): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (32-33): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (34): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (35): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (36): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (37): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (38): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - ) - (fc_before): Linear(in_features=256, out_features=512, bias=False) - (fc_after): Linear(in_features=512, out_features=256, bias=False) - ) - (depth_branch): DenseDepthNet( - (depth_layers): ModuleList( - (0-2): 3 x Conv2d(256, 1, kernel_size=(1, 1), stride=(1, 1)) - ) - ) - (grid_mask): GridMask() -) -2026-02-10 16:08:24,596 - mmdet - INFO - Start running, host: root@bw28, work_dir: /workspace/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704 -2026-02-10 16:08:24,596 - mmdet - INFO - Hooks will be executed in the following order: -before_run: -(VERY_HIGH ) CosineAnnealingLrUpdaterHook -(ABOVE_NORMAL) Fp16OptimizerHook -(NORMAL ) CheckpointHook -(NORMAL ) EvalHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -before_train_epoch: -(VERY_HIGH ) CosineAnnealingLrUpdaterHook -(NORMAL ) EvalHook -(LOW ) IterTimerHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -before_train_iter: -(VERY_HIGH ) CosineAnnealingLrUpdaterHook -(NORMAL ) EvalHook -(LOW ) IterTimerHook - -------------------- -after_train_iter: -(ABOVE_NORMAL) Fp16OptimizerHook -(NORMAL ) CheckpointHook -(NORMAL ) EvalHook -(NORMAL ) ProfilerHook -(LOW ) IterTimerHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -after_train_epoch: -(NORMAL ) CheckpointHook -(NORMAL ) EvalHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -before_val_epoch: -(LOW ) IterTimerHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -before_val_iter: -(LOW ) IterTimerHook - -------------------- -after_val_iter: -(LOW ) IterTimerHook - -------------------- -after_val_epoch: -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -after_run: -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -2026-02-10 16:08:24,597 - mmdet - INFO - workflow: [('train', 1)], max: 17500 iters -2026-02-10 16:08:24,599 - mmdet - INFO - Checkpoints will be saved to /workspace/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704 by HardDiskBackend. diff --git a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20260210_160806.log.json b/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20260210_160806.log.json deleted file mode 100644 index e824b9dc86300b4dd4e73ef6967ad89df0a91b4d..0000000000000000000000000000000000000000 --- a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20260210_160806.log.json +++ /dev/null @@ -1 +0,0 @@ -{"env_info": "sys.platform: linux\nPython: 3.10.12 (main, Aug 15 2025, 14:32:43) [GCC 11.4.0]\nCUDA available: True\nGPU 0,1,2,3,4,5,6,7: BW200_LC, UBB BW1000_LC\nCUDA_HOME: /opt/dtk/cuda/cuda\nNVCC: Cuda compilation tools, release 12.6, V12.6.77\nclang version 17.0.0\nTarget: x86_64-unknown-linux-gnu\nThread model: posix\nInstalledDir: /opt/dtk-25.04.4/llvm/bi\nGCC: x86_64-linux-gnu-gcc (Ubuntu 11.4.0-1ubuntu1~22.04) 11.4.0\nPyTorch: 2.5.1\nPyTorch compiling details: PyTorch built with:\n - GCC 10.3\n - C++ Version: 201703\n - Intel(R) Math Kernel Library Version 2020.0.4 Product Build 20200917 for Intel(R) 64 architecture applications\n - OpenMP 201511 (a.k.a. OpenMP 4.5)\n - LAPACK is enabled (usually provided by MKL)\n - NNPACK is enabled\n - CPU capability usage: AVX2\n - HIP Runtime 6.3.25521\n - MIOpen 2.18.0\n - Magma 2.8.0\n - Build settings: BLAS_INFO=mkl, BUILD_TYPE=Release, CXX_COMPILER=/opt/rh/gcc-toolset-10/root/usr/bin/c++, CXX_FLAGS= -D_GLIBCXX_USE_CXX11_ABI=1 -fvisibility-inlines-hidden -DUSE_PTHREADPOOL -DNDEBUG -DUSE_KINETO -DLIBKINETO_NOCUPTI -DLIBKINETO_NOXPUPTI=ON -DUSE_FBGEMM -DUSE_PYTORCH_QNNPACK -DUSE_XNNPACK -DSYMBOLICATE_MOBILE_DEBUG_HANDLE -O2 -fPIC -Wall -Wextra -Werror=return-type -Werror=non-virtual-dtor -Werror=bool-operation -Wnarrowing -Wno-missing-field-initializers -Wno-type-limits -Wno-array-bounds -Wno-unknown-pragmas -Wno-unused-parameter -Wno-strict-overflow -Wno-strict-aliasing -Wno-stringop-overflow -Wsuggest-override -Wno-psabi -Wno-error=old-style-cast -Wno-missing-braces -fdiagnostics-color=always -faligned-new -Wno-unused-but-set-variable -Wno-maybe-uninitialized -fno-math-errno -fno-trapping-math -Werror=format -Wno-stringop-overflow, FORCE_FALLBACK_CUDA_MPI=1, LAPACK_INFO=mkl, PERF_WITH_AVX=1, PERF_WITH_AVX2=1, TORCH_VERSION=2.5.1, USE_CUDA=0, USE_CUDNN=OFF, USE_CUSPARSELT=OFF, USE_EXCEPTION_PTR=1, USE_GFLAGS=1, USE_GLOG=1, USE_GLOO=1, USE_MKL=ON, USE_MKLDNN=0, USE_MPI=1, USE_NCCL=1, USE_NNPACK=ON, USE_OPENMP=1, USE_ROCM=ON, USE_ROCM_KERNEL_ASSERT=OFF, \n\nTorchVision: 0.20.1\nOpenCV: 4.12.0\nMMCV: 1.6.1\nMMCV Compiler: GCC 10.3\nMMCV CUDA Compiler: rocm not available\nMMDetection: 2.25.1+", "config": "plugin = True\nplugin_dir = 'projects/mmdet3d_plugin/'\ndist_params = dict(backend='nccl')\nlog_level = 'INFO'\nwork_dir = './work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704'\ntotal_batch_size = 160\nnum_gpus = 8\nbatch_size = 20\nnum_iters_per_epoch = 175\nnum_epochs = 100\ncheckpoint_epoch_interval = 20\ncheckpoint_config = dict(interval=3500)\nlog_config = dict(\n interval=1,\n hooks=[\n dict(type='TextLoggerHook', by_epoch=False),\n dict(type='TensorboardLoggerHook')\n ])\nload_from = None\nresume_from = None\nworkflow = [('train', 1)]\nfp16 = dict(loss_scale=32.0)\ninput_shape = (704, 256)\ntracking_test = True\ntracking_threshold = 0.2\nclass_names = [\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier',\n 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n]\nnum_classes = 10\nembed_dims = 256\nnum_groups = 8\nnum_decoder = 6\nnum_single_frame_decoder = 1\nuse_deformable_func = True\nstrides = [4, 8, 16, 32]\nnum_levels = 4\nnum_depth_layers = 3\ndrop_out = 0.1\ntemporal = True\ndecouple_attn = True\nwith_quality_estimation = True\nmodel = dict(\n type='Sparse4D',\n use_grid_mask=True,\n use_deformable_func=True,\n img_backbone=dict(\n type='ResNet',\n depth=50,\n num_stages=4,\n frozen_stages=-1,\n norm_eval=False,\n style='pytorch',\n with_cp=True,\n out_indices=(0, 1, 2, 3),\n norm_cfg=dict(type='BN', requires_grad=True),\n pretrained='ckpt/resnet50-19c8e357.pth'),\n img_neck=dict(\n type='FPN',\n num_outs=4,\n start_level=0,\n out_channels=256,\n add_extra_convs='on_output',\n relu_before_extra_convs=True,\n in_channels=[256, 512, 1024, 2048]),\n depth_branch=dict(\n type='DenseDepthNet',\n embed_dims=256,\n num_depth_layers=3,\n loss_weight=0.2),\n head=dict(\n type='Sparse4DHead',\n cls_threshold_to_reg=0.05,\n decouple_attn=True,\n instance_bank=dict(\n type='InstanceBank',\n num_anchor=900,\n embed_dims=256,\n anchor='nuscenes_kmeans900.npy',\n anchor_handler=dict(type='SparseBox3DKeyPointsGenerator'),\n num_temp_instances=600,\n confidence_decay=0.6,\n feat_grad=False),\n anchor_encoder=dict(\n type='SparseBox3DEncoder',\n vel_dims=3,\n embed_dims=[128, 32, 32, 64],\n mode='cat',\n output_fc=False,\n in_loops=1,\n out_loops=4),\n num_single_frame_decoder=1,\n operation_order=[\n 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm',\n 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm',\n 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm',\n 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm',\n 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm',\n 'deformable', 'ffn', 'norm', 'refine'\n ],\n temp_graph_model=dict(\n type='MultiheadAttention',\n embed_dims=512,\n num_heads=8,\n batch_first=True,\n dropout=0.1),\n graph_model=dict(\n type='MultiheadAttention',\n embed_dims=512,\n num_heads=8,\n batch_first=True,\n dropout=0.1),\n norm_layer=dict(type='LN', normalized_shape=256),\n ffn=dict(\n type='AsymmetricFFN',\n in_channels=512,\n pre_norm=dict(type='LN'),\n embed_dims=256,\n feedforward_channels=1024,\n num_fcs=2,\n ffn_drop=0.1,\n act_cfg=dict(type='ReLU', inplace=True)),\n deformable_model=dict(\n type='DeformableFeatureAggregation',\n embed_dims=256,\n num_groups=8,\n num_levels=4,\n num_cams=6,\n attn_drop=0.15,\n use_deformable_func=True,\n use_camera_embed=True,\n residual_mode='cat',\n kps_generator=dict(\n type='SparseBox3DKeyPointsGenerator',\n num_learnable_pts=6,\n fix_scale=[[0, 0, 0], [0.45, 0, 0], [-0.45, 0, 0],\n [0, 0.45, 0], [0, -0.45, 0], [0, 0, 0.45],\n [0, 0, -0.45]])),\n refine_layer=dict(\n type='SparseBox3DRefinementModule',\n embed_dims=256,\n num_cls=10,\n refine_yaw=True,\n with_quality_estimation=True),\n sampler=dict(\n type='SparseBox3DTarget',\n num_dn_groups=5,\n num_temp_dn_groups=3,\n dn_noise_scale=[2.0, 2.0, 2.0, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5],\n max_dn_gt=32,\n add_neg_dn=True,\n cls_weight=2.0,\n box_weight=0.25,\n reg_weights=[2.0, 2.0, 2.0, 0.5, 0.5, 0.5, 0.0, 0.0, 0.0, 0.0],\n cls_wise_reg_weights=dict(\n {9: [2.0, 2.0, 2.0, 1.0, 1.0, 1.0, 0.0, 0.0, 1.0, 1.0]})),\n loss_cls=dict(\n type='FocalLoss',\n use_sigmoid=True,\n gamma=2.0,\n alpha=0.25,\n loss_weight=2.0),\n loss_reg=dict(\n type='SparseBox3DLoss',\n loss_box=dict(type='L1Loss', loss_weight=0.25),\n loss_centerness=dict(type='CrossEntropyLoss', use_sigmoid=True),\n loss_yawness=dict(type='GaussianFocalLoss'),\n cls_allow_reverse=[5]),\n decoder=dict(type='SparseBox3DDecoder'),\n reg_weights=[2.0, 2.0, 2.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0]))\ndataset_type = 'NuScenes3DDetTrackDataset'\ndata_root = 'data/nuscenes/'\nanno_root = 'data/nuscenes_anno_pkls/'\nfile_client_args = dict(backend='disk')\nimg_norm_cfg = dict(\n mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375], to_rgb=True)\ntrain_pipeline = [\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(\n type='LoadPointsFromFile',\n coord_type='LIDAR',\n load_dim=5,\n use_dim=5,\n file_client_args=dict(backend='disk')),\n dict(type='ResizeCropFlipImage'),\n dict(type='MultiScaleDepthMapGenerator', downsample=[4, 8, 16]),\n dict(type='BBoxRotation'),\n dict(type='PhotoMetricDistortionMultiViewImage'),\n dict(\n type='NormalizeMultiviewImage',\n mean=[123.675, 116.28, 103.53],\n std=[58.395, 57.12, 57.375],\n to_rgb=True),\n dict(\n type='CircleObjectRangeFilter',\n class_dist_thred=[55, 55, 55, 55, 55, 55, 55, 55, 55, 55]),\n dict(\n type='InstanceNameFilter',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ]),\n dict(type='NuScenesSparse4DAdaptor'),\n dict(\n type='Collect',\n keys=[\n 'img', 'timestamp', 'projection_mat', 'image_wh', 'gt_depth',\n 'focal', 'gt_bboxes_3d', 'gt_labels_3d'\n ],\n meta_keys=['T_global', 'T_global_inv', 'timestamp', 'instance_id'])\n]\ntest_pipeline = [\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(type='ResizeCropFlipImage'),\n dict(\n type='NormalizeMultiviewImage',\n mean=[123.675, 116.28, 103.53],\n std=[58.395, 57.12, 57.375],\n to_rgb=True),\n dict(type='NuScenesSparse4DAdaptor'),\n dict(\n type='Collect',\n keys=['img', 'timestamp', 'projection_mat', 'image_wh'],\n meta_keys=['T_global', 'T_global_inv', 'timestamp'])\n]\ninput_modality = dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False)\ndata_basic_config = dict(\n type='NuScenes3DDetTrackDataset',\n data_root='data/nuscenes/',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier',\n 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ],\n modality=dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False),\n version='v1.0-trainval')\ndata_aug_conf = dict(\n resize_lim=(0.4, 0.47),\n final_dim=(256, 704),\n bot_pct_lim=(0.0, 0.0),\n rot_lim=(-5.4, 5.4),\n H=900,\n W=1600,\n rand_flip=True,\n rot3d_range=[-0.3925, 0.3925])\ndata = dict(\n samples_per_gpu=20,\n workers_per_gpu=20,\n train=dict(\n type='NuScenes3DDetTrackDataset',\n data_root='data/nuscenes/',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ],\n modality=dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False),\n version='v1.0-trainval',\n ann_file='data/nuscenes_anno_pkls/nuscenes_infos_train.pkl',\n pipeline=[\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(\n type='LoadPointsFromFile',\n coord_type='LIDAR',\n load_dim=5,\n use_dim=5,\n file_client_args=dict(backend='disk')),\n dict(type='ResizeCropFlipImage'),\n dict(type='MultiScaleDepthMapGenerator', downsample=[4, 8, 16]),\n dict(type='BBoxRotation'),\n dict(type='PhotoMetricDistortionMultiViewImage'),\n dict(\n type='NormalizeMultiviewImage',\n mean=[123.675, 116.28, 103.53],\n std=[58.395, 57.12, 57.375],\n to_rgb=True),\n dict(\n type='CircleObjectRangeFilter',\n class_dist_thred=[55, 55, 55, 55, 55, 55, 55, 55, 55, 55]),\n dict(\n type='InstanceNameFilter',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian',\n 'traffic_cone'\n ]),\n dict(type='NuScenesSparse4DAdaptor'),\n dict(\n type='Collect',\n keys=[\n 'img', 'timestamp', 'projection_mat', 'image_wh',\n 'gt_depth', 'focal', 'gt_bboxes_3d', 'gt_labels_3d'\n ],\n meta_keys=[\n 'T_global', 'T_global_inv', 'timestamp', 'instance_id'\n ])\n ],\n test_mode=False,\n data_aug_conf=dict(\n resize_lim=(0.4, 0.47),\n final_dim=(256, 704),\n bot_pct_lim=(0.0, 0.0),\n rot_lim=(-5.4, 5.4),\n H=900,\n W=1600,\n rand_flip=True,\n rot3d_range=[-0.3925, 0.3925]),\n with_seq_flag=True,\n sequences_split_num=2,\n keep_consistent_seq_aug=True),\n val=dict(\n type='NuScenes3DDetTrackDataset',\n data_root='data/nuscenes/',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ],\n modality=dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False),\n version='v1.0-trainval',\n ann_file='data/nuscenes_anno_pkls/nuscenes_infos_val.pkl',\n pipeline=[\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(type='ResizeCropFlipImage'),\n dict(\n type='NormalizeMultiviewImage',\n mean=[123.675, 116.28, 103.53],\n std=[58.395, 57.12, 57.375],\n to_rgb=True),\n dict(type='NuScenesSparse4DAdaptor'),\n dict(\n type='Collect',\n keys=['img', 'timestamp', 'projection_mat', 'image_wh'],\n meta_keys=['T_global', 'T_global_inv', 'timestamp'])\n ],\n data_aug_conf=dict(\n resize_lim=(0.4, 0.47),\n final_dim=(256, 704),\n bot_pct_lim=(0.0, 0.0),\n rot_lim=(-5.4, 5.4),\n H=900,\n W=1600,\n rand_flip=True,\n rot3d_range=[-0.3925, 0.3925]),\n test_mode=True,\n tracking=True,\n tracking_threshold=0.2),\n test=dict(\n type='NuScenes3DDetTrackDataset',\n data_root='data/nuscenes/',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ],\n modality=dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False),\n version='v1.0-trainval',\n ann_file='data/nuscenes_anno_pkls/nuscenes_infos_val.pkl',\n pipeline=[\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(type='ResizeCropFlipImage'),\n dict(\n type='NormalizeMultiviewImage',\n mean=[123.675, 116.28, 103.53],\n std=[58.395, 57.12, 57.375],\n to_rgb=True),\n dict(type='NuScenesSparse4DAdaptor'),\n dict(\n type='Collect',\n keys=['img', 'timestamp', 'projection_mat', 'image_wh'],\n meta_keys=['T_global', 'T_global_inv', 'timestamp'])\n ],\n data_aug_conf=dict(\n resize_lim=(0.4, 0.47),\n final_dim=(256, 704),\n bot_pct_lim=(0.0, 0.0),\n rot_lim=(-5.4, 5.4),\n H=900,\n W=1600,\n rand_flip=True,\n rot3d_range=[-0.3925, 0.3925]),\n test_mode=True,\n tracking=True,\n tracking_threshold=0.2))\noptimizer = dict(\n type='AdamW',\n lr=0.0006,\n weight_decay=0.001,\n paramwise_cfg=dict(custom_keys=dict(img_backbone=dict(lr_mult=0.5))))\noptimizer_config = dict(grad_clip=dict(max_norm=25, norm_type=2))\nlr_config = dict(\n policy='CosineAnnealing',\n warmup='linear',\n warmup_iters=500,\n warmup_ratio=0.3333333333333333,\n min_lr_ratio=0.001)\nrunner = dict(type='IterBasedRunner', max_iters=17500)\nvis_pipeline = [\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(type='Collect', keys=['img'], meta_keys=['timestamp', 'lidar2img'])\n]\nevaluation = dict(\n interval=3500,\n pipeline=[\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(\n type='Collect', keys=['img'], meta_keys=['timestamp', 'lidar2img'])\n ])\ngpu_ids = range(0, 1)\n", "seed": 0, "exp_name": "sparse4dv3_temporal_r50_1x8_bs6_256x704.py"} diff --git a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20260210_160910.log b/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20260210_160910.log deleted file mode 100644 index 18ab36f1b99ffd335a415c9c4f2cb7d02ebb4007..0000000000000000000000000000000000000000 --- a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20260210_160910.log +++ /dev/null @@ -1,3221 +0,0 @@ -2026-02-10 16:09:10,792 - mmdet - INFO - Environment info: ------------------------------------------------------------- -sys.platform: linux -Python: 3.10.12 (main, Aug 15 2025, 14:32:43) [GCC 11.4.0] -CUDA available: True -GPU 0,1,2,3,4,5,6,7: BW200_LC, UBB BW1000_LC -CUDA_HOME: /opt/dtk -NVCC: Not Available -GCC: x86_64-linux-gnu-gcc (Ubuntu 11.4.0-1ubuntu1~22.04) 11.4.0 -PyTorch: 2.5.1 -PyTorch compiling details: PyTorch built with: - - GCC 10.3 - - C++ Version: 201703 - - Intel(R) Math Kernel Library Version 2020.0.4 Product Build 20200917 for Intel(R) 64 architecture applications - - OpenMP 201511 (a.k.a. OpenMP 4.5) - - LAPACK is enabled (usually provided by MKL) - - NNPACK is enabled - - CPU capability usage: AVX2 - - HIP Runtime 6.3.25521 - - MIOpen 2.18.0 - - Magma 2.8.0 - - Build settings: BLAS_INFO=mkl, BUILD_TYPE=Release, CXX_COMPILER=/opt/rh/gcc-toolset-10/root/usr/bin/c++, CXX_FLAGS= -D_GLIBCXX_USE_CXX11_ABI=1 -fvisibility-inlines-hidden -DUSE_PTHREADPOOL -DNDEBUG -DUSE_KINETO -DLIBKINETO_NOCUPTI -DLIBKINETO_NOXPUPTI=ON -DUSE_FBGEMM -DUSE_PYTORCH_QNNPACK -DUSE_XNNPACK -DSYMBOLICATE_MOBILE_DEBUG_HANDLE -O2 -fPIC -Wall -Wextra -Werror=return-type -Werror=non-virtual-dtor -Werror=bool-operation -Wnarrowing -Wno-missing-field-initializers -Wno-type-limits -Wno-array-bounds -Wno-unknown-pragmas -Wno-unused-parameter -Wno-strict-overflow -Wno-strict-aliasing -Wno-stringop-overflow -Wsuggest-override -Wno-psabi -Wno-error=old-style-cast -Wno-missing-braces -fdiagnostics-color=always -faligned-new -Wno-unused-but-set-variable -Wno-maybe-uninitialized -fno-math-errno -fno-trapping-math -Werror=format -Wno-stringop-overflow, FORCE_FALLBACK_CUDA_MPI=1, LAPACK_INFO=mkl, PERF_WITH_AVX=1, PERF_WITH_AVX2=1, TORCH_VERSION=2.5.1, USE_CUDA=0, USE_CUDNN=OFF, USE_CUSPARSELT=OFF, USE_EXCEPTION_PTR=1, USE_GFLAGS=1, USE_GLOG=1, USE_GLOO=1, USE_MKL=ON, USE_MKLDNN=0, USE_MPI=1, USE_NCCL=1, USE_NNPACK=ON, USE_OPENMP=1, USE_ROCM=ON, USE_ROCM_KERNEL_ASSERT=OFF, - -TorchVision: 0.20.1 -OpenCV: 4.12.0 -MMCV: 1.6.1 -MMCV Compiler: GCC 10.3 -MMCV CUDA Compiler: rocm not available -MMDetection: 2.25.1+ ------------------------------------------------------------- - -2026-02-10 16:09:11,816 - mmdet - INFO - Distributed training: False -2026-02-10 16:09:12,787 - mmdet - INFO - Config: -plugin = True -plugin_dir = 'projects/mmdet3d_plugin/' -dist_params = dict(backend='nccl') -log_level = 'INFO' -work_dir = './work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704' -total_batch_size = 160 -num_gpus = 8 -batch_size = 20 -num_iters_per_epoch = 175 -num_epochs = 100 -checkpoint_epoch_interval = 20 -checkpoint_config = dict(interval=3500) -log_config = dict( - interval=1, - hooks=[ - dict(type='TextLoggerHook', by_epoch=False), - dict(type='TensorboardLoggerHook') - ]) -load_from = None -resume_from = None -workflow = [('train', 1)] -fp16 = dict(loss_scale=32.0) -input_shape = (704, 256) -tracking_test = True -tracking_threshold = 0.2 -class_names = [ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier', - 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' -] -num_classes = 10 -embed_dims = 256 -num_groups = 8 -num_decoder = 6 -num_single_frame_decoder = 1 -use_deformable_func = True -strides = [4, 8, 16, 32] -num_levels = 4 -num_depth_layers = 3 -drop_out = 0.1 -temporal = True -decouple_attn = True -with_quality_estimation = True -model = dict( - type='Sparse4D', - use_grid_mask=True, - use_deformable_func=True, - img_backbone=dict( - type='ResNet', - depth=50, - num_stages=4, - frozen_stages=-1, - norm_eval=False, - style='pytorch', - with_cp=True, - out_indices=(0, 1, 2, 3), - norm_cfg=dict(type='BN', requires_grad=True), - pretrained='ckpt/resnet50-19c8e357.pth'), - img_neck=dict( - type='FPN', - num_outs=4, - start_level=0, - out_channels=256, - add_extra_convs='on_output', - relu_before_extra_convs=True, - in_channels=[256, 512, 1024, 2048]), - depth_branch=dict( - type='DenseDepthNet', - embed_dims=256, - num_depth_layers=3, - loss_weight=0.2), - head=dict( - type='Sparse4DHead', - cls_threshold_to_reg=0.05, - decouple_attn=True, - instance_bank=dict( - type='InstanceBank', - num_anchor=900, - embed_dims=256, - anchor='nuscenes_kmeans900.npy', - anchor_handler=dict(type='SparseBox3DKeyPointsGenerator'), - num_temp_instances=600, - confidence_decay=0.6, - feat_grad=False), - anchor_encoder=dict( - type='SparseBox3DEncoder', - vel_dims=3, - embed_dims=[128, 32, 32, 64], - mode='cat', - output_fc=False, - in_loops=1, - out_loops=4), - num_single_frame_decoder=1, - operation_order=[ - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine' - ], - temp_graph_model=dict( - type='MultiheadAttention', - embed_dims=512, - num_heads=8, - batch_first=True, - dropout=0.1), - graph_model=dict( - type='MultiheadAttention', - embed_dims=512, - num_heads=8, - batch_first=True, - dropout=0.1), - norm_layer=dict(type='LN', normalized_shape=256), - ffn=dict( - type='AsymmetricFFN', - in_channels=512, - pre_norm=dict(type='LN'), - embed_dims=256, - feedforward_channels=1024, - num_fcs=2, - ffn_drop=0.1, - act_cfg=dict(type='ReLU', inplace=True)), - deformable_model=dict( - type='DeformableFeatureAggregation', - embed_dims=256, - num_groups=8, - num_levels=4, - num_cams=6, - attn_drop=0.15, - use_deformable_func=True, - use_camera_embed=True, - residual_mode='cat', - kps_generator=dict( - type='SparseBox3DKeyPointsGenerator', - num_learnable_pts=6, - fix_scale=[[0, 0, 0], [0.45, 0, 0], [-0.45, 0, 0], - [0, 0.45, 0], [0, -0.45, 0], [0, 0, 0.45], - [0, 0, -0.45]])), - refine_layer=dict( - type='SparseBox3DRefinementModule', - embed_dims=256, - num_cls=10, - refine_yaw=True, - with_quality_estimation=True), - sampler=dict( - type='SparseBox3DTarget', - num_dn_groups=5, - num_temp_dn_groups=3, - dn_noise_scale=[2.0, 2.0, 2.0, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5], - max_dn_gt=32, - add_neg_dn=True, - cls_weight=2.0, - box_weight=0.25, - reg_weights=[2.0, 2.0, 2.0, 0.5, 0.5, 0.5, 0.0, 0.0, 0.0, 0.0], - cls_wise_reg_weights=dict( - {9: [2.0, 2.0, 2.0, 1.0, 1.0, 1.0, 0.0, 0.0, 1.0, 1.0]})), - loss_cls=dict( - type='FocalLoss', - use_sigmoid=True, - gamma=2.0, - alpha=0.25, - loss_weight=2.0), - loss_reg=dict( - type='SparseBox3DLoss', - loss_box=dict(type='L1Loss', loss_weight=0.25), - loss_centerness=dict(type='CrossEntropyLoss', use_sigmoid=True), - loss_yawness=dict(type='GaussianFocalLoss'), - cls_allow_reverse=[5]), - decoder=dict(type='SparseBox3DDecoder'), - reg_weights=[2.0, 2.0, 2.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0])) -dataset_type = 'NuScenes3DDetTrackDataset' -data_root = 'data/nuscenes/' -anno_root = 'data/nuscenes_anno_pkls/' -file_client_args = dict(backend='disk') -img_norm_cfg = dict( - mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375], to_rgb=True) -train_pipeline = [ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict( - type='LoadPointsFromFile', - coord_type='LIDAR', - load_dim=5, - use_dim=5, - file_client_args=dict(backend='disk')), - dict(type='ResizeCropFlipImage'), - dict(type='MultiScaleDepthMapGenerator', downsample=[4, 8, 16]), - dict(type='BBoxRotation'), - dict(type='PhotoMetricDistortionMultiViewImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict( - type='CircleObjectRangeFilter', - class_dist_thred=[55, 55, 55, 55, 55, 55, 55, 55, 55, 55]), - dict( - type='InstanceNameFilter', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ]), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=[ - 'img', 'timestamp', 'projection_mat', 'image_wh', 'gt_depth', - 'focal', 'gt_bboxes_3d', 'gt_labels_3d' - ], - meta_keys=['T_global', 'T_global_inv', 'timestamp', 'instance_id']) -] -test_pipeline = [ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='ResizeCropFlipImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=['img', 'timestamp', 'projection_mat', 'image_wh'], - meta_keys=['T_global', 'T_global_inv', 'timestamp']) -] -input_modality = dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False) -data_basic_config = dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier', - 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval') -data_aug_conf = dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]) -data = dict( - samples_per_gpu=20, - workers_per_gpu=20, - train=dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval', - ann_file='data/nuscenes_anno_pkls/nuscenes_infos_train.pkl', - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict( - type='LoadPointsFromFile', - coord_type='LIDAR', - load_dim=5, - use_dim=5, - file_client_args=dict(backend='disk')), - dict(type='ResizeCropFlipImage'), - dict(type='MultiScaleDepthMapGenerator', downsample=[4, 8, 16]), - dict(type='BBoxRotation'), - dict(type='PhotoMetricDistortionMultiViewImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict( - type='CircleObjectRangeFilter', - class_dist_thred=[55, 55, 55, 55, 55, 55, 55, 55, 55, 55]), - dict( - type='InstanceNameFilter', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', - 'traffic_cone' - ]), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=[ - 'img', 'timestamp', 'projection_mat', 'image_wh', - 'gt_depth', 'focal', 'gt_bboxes_3d', 'gt_labels_3d' - ], - meta_keys=[ - 'T_global', 'T_global_inv', 'timestamp', 'instance_id' - ]) - ], - test_mode=False, - data_aug_conf=dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]), - with_seq_flag=True, - sequences_split_num=2, - keep_consistent_seq_aug=True), - val=dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval', - ann_file='data/nuscenes_anno_pkls/nuscenes_infos_val.pkl', - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='ResizeCropFlipImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=['img', 'timestamp', 'projection_mat', 'image_wh'], - meta_keys=['T_global', 'T_global_inv', 'timestamp']) - ], - data_aug_conf=dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]), - test_mode=True, - tracking=True, - tracking_threshold=0.2), - test=dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval', - ann_file='data/nuscenes_anno_pkls/nuscenes_infos_val.pkl', - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='ResizeCropFlipImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=['img', 'timestamp', 'projection_mat', 'image_wh'], - meta_keys=['T_global', 'T_global_inv', 'timestamp']) - ], - data_aug_conf=dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]), - test_mode=True, - tracking=True, - tracking_threshold=0.2)) -optimizer = dict( - type='AdamW', - lr=0.0006, - weight_decay=0.001, - paramwise_cfg=dict(custom_keys=dict(img_backbone=dict(lr_mult=0.5)))) -optimizer_config = dict(grad_clip=dict(max_norm=25, norm_type=2)) -lr_config = dict( - policy='CosineAnnealing', - warmup='linear', - warmup_iters=500, - warmup_ratio=0.3333333333333333, - min_lr_ratio=0.001) -runner = dict(type='IterBasedRunner', max_iters=17500) -vis_pipeline = [ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='Collect', keys=['img'], meta_keys=['timestamp', 'lidar2img']) -] -evaluation = dict( - interval=3500, - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict( - type='Collect', keys=['img'], meta_keys=['timestamp', 'lidar2img']) - ]) -gpu_ids = range(0, 1) - -2026-02-10 16:09:12,788 - mmdet - INFO - Set random seed to 0, deterministic: False -2026-02-10 16:09:13,342 - mmdet - INFO - initialize ResNet with init_cfg {'type': 'Pretrained', 'checkpoint': 'ckpt/resnet50-19c8e357.pth'} -2026-02-10 16:09:13,741 - mmdet - INFO - initialize FPN with init_cfg {'type': 'Xavier', 'layer': 'Conv2d', 'distribution': 'uniform'} -Name of parameter - Initialization information - -img_backbone.conv1.weight - torch.Size([64, 3, 7, 7]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.bn1.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.bn1.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.conv1.weight - torch.Size([64, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn1.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn1.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.conv2.weight - torch.Size([64, 64, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn2.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn2.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.conv3.weight - torch.Size([256, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn3.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn3.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.downsample.0.weight - torch.Size([256, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.downsample.1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.downsample.1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.conv1.weight - torch.Size([64, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn1.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn1.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.conv2.weight - torch.Size([64, 64, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn2.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn2.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.conv3.weight - torch.Size([256, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn3.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn3.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.conv1.weight - torch.Size([64, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn1.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn1.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.conv2.weight - torch.Size([64, 64, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn2.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn2.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.conv3.weight - torch.Size([256, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn3.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn3.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.conv1.weight - torch.Size([128, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn1.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn1.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.conv2.weight - torch.Size([128, 128, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn2.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn2.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.conv3.weight - torch.Size([512, 128, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn3.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn3.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.downsample.0.weight - torch.Size([512, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.downsample.1.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.downsample.1.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.conv1.weight - torch.Size([128, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn1.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn1.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.conv2.weight - torch.Size([128, 128, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn2.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn2.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.conv3.weight - torch.Size([512, 128, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn3.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn3.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.conv1.weight - torch.Size([128, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn1.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn1.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.conv2.weight - torch.Size([128, 128, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn2.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn2.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.conv3.weight - torch.Size([512, 128, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn3.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn3.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.conv1.weight - torch.Size([128, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn1.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn1.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.conv2.weight - torch.Size([128, 128, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn2.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn2.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.conv3.weight - torch.Size([512, 128, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn3.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn3.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.conv1.weight - torch.Size([256, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.downsample.0.weight - torch.Size([1024, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.downsample.1.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.downsample.1.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.conv1.weight - torch.Size([512, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn1.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn1.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.conv2.weight - torch.Size([512, 512, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn2.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn2.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.conv3.weight - torch.Size([2048, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn3.weight - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn3.bias - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.downsample.0.weight - torch.Size([2048, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.downsample.1.weight - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.downsample.1.bias - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.conv1.weight - torch.Size([512, 2048, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn1.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn1.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.conv2.weight - torch.Size([512, 512, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn2.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn2.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.conv3.weight - torch.Size([2048, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn3.weight - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn3.bias - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.conv1.weight - torch.Size([512, 2048, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn1.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn1.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.conv2.weight - torch.Size([512, 512, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn2.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn2.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.conv3.weight - torch.Size([2048, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn3.weight - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn3.bias - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_neck.lateral_convs.0.conv.weight - torch.Size([256, 256, 1, 1]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.lateral_convs.0.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.lateral_convs.1.conv.weight - torch.Size([256, 512, 1, 1]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.lateral_convs.1.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.lateral_convs.2.conv.weight - torch.Size([256, 1024, 1, 1]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.lateral_convs.2.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.lateral_convs.3.conv.weight - torch.Size([256, 2048, 1, 1]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.lateral_convs.3.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.fpn_convs.0.conv.weight - torch.Size([256, 256, 3, 3]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.fpn_convs.0.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.fpn_convs.1.conv.weight - torch.Size([256, 256, 3, 3]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.fpn_convs.1.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.fpn_convs.2.conv.weight - torch.Size([256, 256, 3, 3]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.fpn_convs.2.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.fpn_convs.3.conv.weight - torch.Size([256, 256, 3, 3]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.fpn_convs.3.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.instance_bank.anchor - torch.Size([900, 11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.instance_bank.instance_feature - torch.Size([900, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.instance_bank.anchor_handler.fix_scale - torch.Size([1, 3]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.0.weight - torch.Size([128, 3]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.0.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.2.weight - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.2.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.3.weight - torch.Size([128, 128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.3.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.5.weight - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.5.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.6.weight - torch.Size([128, 128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.6.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.8.weight - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.8.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.9.weight - torch.Size([128, 128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.9.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.11.weight - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.11.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.0.weight - torch.Size([32, 3]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.0.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.2.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.2.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.3.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.3.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.5.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.5.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.6.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.6.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.8.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.8.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.9.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.9.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.11.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.11.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.0.weight - torch.Size([32, 2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.0.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.2.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.2.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.3.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.3.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.5.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.5.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.6.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.6.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.8.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.8.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.9.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.9.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.11.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.11.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.0.weight - torch.Size([64, 3]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.0.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.2.weight - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.2.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.3.weight - torch.Size([64, 64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.3.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.5.weight - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.5.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.6.weight - torch.Size([64, 64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.6.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.8.weight - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.8.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.9.weight - torch.Size([64, 64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.9.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.11.weight - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.11.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.1.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.1.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.1.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.1.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.1.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.1.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.1.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.1.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.3.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.4.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.4.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.4.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.4.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.5.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.5.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.5.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.5.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.6.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.6.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.8.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.8.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.8.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.8.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.8.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.8.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.8.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.8.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.10.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.11.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.11.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.11.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.11.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.12.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.12.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.12.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.12.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.13.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.13.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.15.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.15.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.15.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.15.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.15.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.15.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.15.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.15.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.16.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.16.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.17.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.18.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.18.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.18.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.18.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.19.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.19.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.19.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.19.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.20.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.20.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.22.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.22.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.22.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.22.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.22.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.22.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.22.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.22.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.23.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.23.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.24.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.25.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.25.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.25.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.25.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.26.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.26.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.26.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.26.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.27.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.27.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.29.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.29.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.29.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.29.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.29.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.29.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.29.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.29.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.30.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.30.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.31.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.32.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.32.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.32.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.32.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.33.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.33.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.33.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.33.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.34.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.34.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.36.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.36.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.36.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.36.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.36.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.36.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.36.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.36.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.37.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.37.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.38.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.fc_before.weight - torch.Size([512, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.fc_after.weight - torch.Size([256, 512]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.0.weight - torch.Size([1, 256, 1, 1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.0.bias - torch.Size([1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.1.weight - torch.Size([1, 256, 1, 1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.1.bias - torch.Size([1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.2.weight - torch.Size([1, 256, 1, 1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.2.bias - torch.Size([1]): -The value is the same before and after calling `init_weights` of Sparse4D -2026-02-10 16:09:14,133 - mmdet - INFO - Model: -Sparse4D( - (img_backbone): ResNet( - (conv1): Conv2d(3, 64, kernel_size=(7, 7), stride=(2, 2), padding=(3, 3), bias=False) - (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (maxpool): MaxPool2d(kernel_size=3, stride=2, padding=1, dilation=1, ceil_mode=False) - (layer1): ResLayer( - (0): Bottleneck( - (conv1): Conv2d(64, 64, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (downsample): Sequential( - (0): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - ) - ) - (1): Bottleneck( - (conv1): Conv2d(256, 64, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (2): Bottleneck( - (conv1): Conv2d(256, 64, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - ) - (layer2): ResLayer( - (0): Bottleneck( - (conv1): Conv2d(256, 128, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (downsample): Sequential( - (0): Conv2d(256, 512, kernel_size=(1, 1), stride=(2, 2), bias=False) - (1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - ) - ) - (1): Bottleneck( - (conv1): Conv2d(512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (2): Bottleneck( - (conv1): Conv2d(512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (3): Bottleneck( - (conv1): Conv2d(512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - ) - (layer3): ResLayer( - (0): Bottleneck( - (conv1): Conv2d(512, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (downsample): Sequential( - (0): Conv2d(512, 1024, kernel_size=(1, 1), stride=(2, 2), bias=False) - (1): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - ) - ) - (1): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (2): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (3): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (4): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (5): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - ) - (layer4): ResLayer( - (0): Bottleneck( - (conv1): Conv2d(1024, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(512, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (downsample): Sequential( - (0): Conv2d(1024, 2048, kernel_size=(1, 1), stride=(2, 2), bias=False) - (1): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - ) - ) - (1): Bottleneck( - (conv1): Conv2d(2048, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(512, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (2): Bottleneck( - (conv1): Conv2d(2048, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(512, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - ) - ) - init_cfg={'type': 'Pretrained', 'checkpoint': 'ckpt/resnet50-19c8e357.pth'} - (img_neck): FPN( - (lateral_convs): ModuleList( - (0): ConvModule( - (conv): Conv2d(256, 256, kernel_size=(1, 1), stride=(1, 1)) - ) - (1): ConvModule( - (conv): Conv2d(512, 256, kernel_size=(1, 1), stride=(1, 1)) - ) - (2): ConvModule( - (conv): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1)) - ) - (3): ConvModule( - (conv): Conv2d(2048, 256, kernel_size=(1, 1), stride=(1, 1)) - ) - ) - (fpn_convs): ModuleList( - (0-3): 4 x ConvModule( - (conv): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1)) - ) - ) - ) - init_cfg={'type': 'Xavier', 'layer': 'Conv2d', 'distribution': 'uniform'} - (head): Sparse4DHead( - (instance_bank): InstanceBank( - (anchor_handler): SparseBox3DKeyPointsGenerator() - ) - (anchor_encoder): SparseBox3DEncoder( - (pos_fc): Sequential( - (0): Linear(in_features=3, out_features=128, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((128,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=128, out_features=128, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((128,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=128, out_features=128, bias=True) - (7): ReLU(inplace=True) - (8): LayerNorm((128,), eps=1e-05, elementwise_affine=True) - (9): Linear(in_features=128, out_features=128, bias=True) - (10): ReLU(inplace=True) - (11): LayerNorm((128,), eps=1e-05, elementwise_affine=True) - ) - (size_fc): Sequential( - (0): Linear(in_features=3, out_features=32, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=32, out_features=32, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=32, out_features=32, bias=True) - (7): ReLU(inplace=True) - (8): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (9): Linear(in_features=32, out_features=32, bias=True) - (10): ReLU(inplace=True) - (11): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - ) - (yaw_fc): Sequential( - (0): Linear(in_features=2, out_features=32, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=32, out_features=32, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=32, out_features=32, bias=True) - (7): ReLU(inplace=True) - (8): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (9): Linear(in_features=32, out_features=32, bias=True) - (10): ReLU(inplace=True) - (11): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - ) - (vel_fc): Sequential( - (0): Linear(in_features=3, out_features=64, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((64,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=64, out_features=64, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((64,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=64, out_features=64, bias=True) - (7): ReLU(inplace=True) - (8): LayerNorm((64,), eps=1e-05, elementwise_affine=True) - (9): Linear(in_features=64, out_features=64, bias=True) - (10): ReLU(inplace=True) - (11): LayerNorm((64,), eps=1e-05, elementwise_affine=True) - ) - ) - (loss_cls): FocalLoss() - (loss_reg): SparseBox3DLoss( - (loss_box): L1Loss() - (loss_cns): CrossEntropyLoss(avg_non_ignore=False) - (loss_yns): GaussianFocalLoss() - ) - (layers): ModuleList( - (0): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (1): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (4-5): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (6): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (7): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (8): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (11-12): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (13): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (14): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (15): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (16): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (17): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (18-19): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (20): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (21): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (22): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (23): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (24): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (25-26): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (27): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (28): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (29): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (30): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (31): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (32-33): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (34): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (35): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (36): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (37): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (38): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - ) - (fc_before): Linear(in_features=256, out_features=512, bias=False) - (fc_after): Linear(in_features=512, out_features=256, bias=False) - ) - (depth_branch): DenseDepthNet( - (depth_layers): ModuleList( - (0-2): 3 x Conv2d(256, 1, kernel_size=(1, 1), stride=(1, 1)) - ) - ) - (grid_mask): GridMask() -) -2026-02-10 16:09:27,367 - mmdet - INFO - Start running, host: root@bw28, work_dir: /workspace/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704 -2026-02-10 16:09:27,368 - mmdet - INFO - Hooks will be executed in the following order: -before_run: -(VERY_HIGH ) CosineAnnealingLrUpdaterHook -(ABOVE_NORMAL) Fp16OptimizerHook -(NORMAL ) CheckpointHook -(NORMAL ) EvalHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -before_train_epoch: -(VERY_HIGH ) CosineAnnealingLrUpdaterHook -(NORMAL ) EvalHook -(LOW ) IterTimerHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -before_train_iter: -(VERY_HIGH ) CosineAnnealingLrUpdaterHook -(NORMAL ) EvalHook -(LOW ) IterTimerHook - -------------------- -after_train_iter: -(ABOVE_NORMAL) Fp16OptimizerHook -(NORMAL ) CheckpointHook -(NORMAL ) EvalHook -(NORMAL ) ProfilerHook -(LOW ) IterTimerHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -after_train_epoch: -(NORMAL ) CheckpointHook -(NORMAL ) EvalHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -before_val_epoch: -(LOW ) IterTimerHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -before_val_iter: -(LOW ) IterTimerHook - -------------------- -after_val_iter: -(LOW ) IterTimerHook - -------------------- -after_val_epoch: -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -after_run: -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -2026-02-10 16:09:27,368 - mmdet - INFO - workflow: [('train', 1)], max: 17500 iters -2026-02-10 16:09:27,370 - mmdet - INFO - Checkpoints will be saved to /workspace/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704 by HardDiskBackend. diff --git a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20260210_160910.log.json b/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20260210_160910.log.json deleted file mode 100644 index f691fdae6b2eb0b287f959b7a72d73034a4da3e6..0000000000000000000000000000000000000000 --- a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20260210_160910.log.json +++ /dev/null @@ -1 +0,0 @@ -{"env_info": "sys.platform: linux\nPython: 3.10.12 (main, Aug 15 2025, 14:32:43) [GCC 11.4.0]\nCUDA available: True\nGPU 0,1,2,3,4,5,6,7: BW200_LC, UBB BW1000_LC\nCUDA_HOME: /opt/dtk\nNVCC: Not Available\nGCC: x86_64-linux-gnu-gcc (Ubuntu 11.4.0-1ubuntu1~22.04) 11.4.0\nPyTorch: 2.5.1\nPyTorch compiling details: PyTorch built with:\n - GCC 10.3\n - C++ Version: 201703\n - Intel(R) Math Kernel Library Version 2020.0.4 Product Build 20200917 for Intel(R) 64 architecture applications\n - OpenMP 201511 (a.k.a. OpenMP 4.5)\n - LAPACK is enabled (usually provided by MKL)\n - NNPACK is enabled\n - CPU capability usage: AVX2\n - HIP Runtime 6.3.25521\n - MIOpen 2.18.0\n - Magma 2.8.0\n - Build settings: BLAS_INFO=mkl, BUILD_TYPE=Release, CXX_COMPILER=/opt/rh/gcc-toolset-10/root/usr/bin/c++, CXX_FLAGS= -D_GLIBCXX_USE_CXX11_ABI=1 -fvisibility-inlines-hidden -DUSE_PTHREADPOOL -DNDEBUG -DUSE_KINETO -DLIBKINETO_NOCUPTI -DLIBKINETO_NOXPUPTI=ON -DUSE_FBGEMM -DUSE_PYTORCH_QNNPACK -DUSE_XNNPACK -DSYMBOLICATE_MOBILE_DEBUG_HANDLE -O2 -fPIC -Wall -Wextra -Werror=return-type -Werror=non-virtual-dtor -Werror=bool-operation -Wnarrowing -Wno-missing-field-initializers -Wno-type-limits -Wno-array-bounds -Wno-unknown-pragmas -Wno-unused-parameter -Wno-strict-overflow -Wno-strict-aliasing -Wno-stringop-overflow -Wsuggest-override -Wno-psabi -Wno-error=old-style-cast -Wno-missing-braces -fdiagnostics-color=always -faligned-new -Wno-unused-but-set-variable -Wno-maybe-uninitialized -fno-math-errno -fno-trapping-math -Werror=format -Wno-stringop-overflow, FORCE_FALLBACK_CUDA_MPI=1, LAPACK_INFO=mkl, PERF_WITH_AVX=1, PERF_WITH_AVX2=1, TORCH_VERSION=2.5.1, USE_CUDA=0, USE_CUDNN=OFF, USE_CUSPARSELT=OFF, USE_EXCEPTION_PTR=1, USE_GFLAGS=1, USE_GLOG=1, USE_GLOO=1, USE_MKL=ON, USE_MKLDNN=0, USE_MPI=1, USE_NCCL=1, USE_NNPACK=ON, USE_OPENMP=1, USE_ROCM=ON, USE_ROCM_KERNEL_ASSERT=OFF, \n\nTorchVision: 0.20.1\nOpenCV: 4.12.0\nMMCV: 1.6.1\nMMCV Compiler: GCC 10.3\nMMCV CUDA Compiler: rocm not available\nMMDetection: 2.25.1+", "config": "plugin = True\nplugin_dir = 'projects/mmdet3d_plugin/'\ndist_params = dict(backend='nccl')\nlog_level = 'INFO'\nwork_dir = './work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704'\ntotal_batch_size = 160\nnum_gpus = 8\nbatch_size = 20\nnum_iters_per_epoch = 175\nnum_epochs = 100\ncheckpoint_epoch_interval = 20\ncheckpoint_config = dict(interval=3500)\nlog_config = dict(\n interval=1,\n hooks=[\n dict(type='TextLoggerHook', by_epoch=False),\n dict(type='TensorboardLoggerHook')\n ])\nload_from = None\nresume_from = None\nworkflow = [('train', 1)]\nfp16 = dict(loss_scale=32.0)\ninput_shape = (704, 256)\ntracking_test = True\ntracking_threshold = 0.2\nclass_names = [\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier',\n 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n]\nnum_classes = 10\nembed_dims = 256\nnum_groups = 8\nnum_decoder = 6\nnum_single_frame_decoder = 1\nuse_deformable_func = True\nstrides = [4, 8, 16, 32]\nnum_levels = 4\nnum_depth_layers = 3\ndrop_out = 0.1\ntemporal = True\ndecouple_attn = True\nwith_quality_estimation = True\nmodel = dict(\n type='Sparse4D',\n use_grid_mask=True,\n use_deformable_func=True,\n img_backbone=dict(\n type='ResNet',\n depth=50,\n num_stages=4,\n frozen_stages=-1,\n norm_eval=False,\n style='pytorch',\n with_cp=True,\n out_indices=(0, 1, 2, 3),\n norm_cfg=dict(type='BN', requires_grad=True),\n pretrained='ckpt/resnet50-19c8e357.pth'),\n img_neck=dict(\n type='FPN',\n num_outs=4,\n start_level=0,\n out_channels=256,\n add_extra_convs='on_output',\n relu_before_extra_convs=True,\n in_channels=[256, 512, 1024, 2048]),\n depth_branch=dict(\n type='DenseDepthNet',\n embed_dims=256,\n num_depth_layers=3,\n loss_weight=0.2),\n head=dict(\n type='Sparse4DHead',\n cls_threshold_to_reg=0.05,\n decouple_attn=True,\n instance_bank=dict(\n type='InstanceBank',\n num_anchor=900,\n embed_dims=256,\n anchor='nuscenes_kmeans900.npy',\n anchor_handler=dict(type='SparseBox3DKeyPointsGenerator'),\n num_temp_instances=600,\n confidence_decay=0.6,\n feat_grad=False),\n anchor_encoder=dict(\n type='SparseBox3DEncoder',\n vel_dims=3,\n embed_dims=[128, 32, 32, 64],\n mode='cat',\n output_fc=False,\n in_loops=1,\n out_loops=4),\n num_single_frame_decoder=1,\n operation_order=[\n 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm',\n 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm',\n 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm',\n 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm',\n 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm',\n 'deformable', 'ffn', 'norm', 'refine'\n ],\n temp_graph_model=dict(\n type='MultiheadAttention',\n embed_dims=512,\n num_heads=8,\n batch_first=True,\n dropout=0.1),\n graph_model=dict(\n type='MultiheadAttention',\n embed_dims=512,\n num_heads=8,\n batch_first=True,\n dropout=0.1),\n norm_layer=dict(type='LN', normalized_shape=256),\n ffn=dict(\n type='AsymmetricFFN',\n in_channels=512,\n pre_norm=dict(type='LN'),\n embed_dims=256,\n feedforward_channels=1024,\n num_fcs=2,\n ffn_drop=0.1,\n act_cfg=dict(type='ReLU', inplace=True)),\n deformable_model=dict(\n type='DeformableFeatureAggregation',\n embed_dims=256,\n num_groups=8,\n num_levels=4,\n num_cams=6,\n attn_drop=0.15,\n use_deformable_func=True,\n use_camera_embed=True,\n residual_mode='cat',\n kps_generator=dict(\n type='SparseBox3DKeyPointsGenerator',\n num_learnable_pts=6,\n fix_scale=[[0, 0, 0], [0.45, 0, 0], [-0.45, 0, 0],\n [0, 0.45, 0], [0, -0.45, 0], [0, 0, 0.45],\n [0, 0, -0.45]])),\n refine_layer=dict(\n type='SparseBox3DRefinementModule',\n embed_dims=256,\n num_cls=10,\n refine_yaw=True,\n with_quality_estimation=True),\n sampler=dict(\n type='SparseBox3DTarget',\n num_dn_groups=5,\n num_temp_dn_groups=3,\n dn_noise_scale=[2.0, 2.0, 2.0, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5],\n max_dn_gt=32,\n add_neg_dn=True,\n cls_weight=2.0,\n box_weight=0.25,\n reg_weights=[2.0, 2.0, 2.0, 0.5, 0.5, 0.5, 0.0, 0.0, 0.0, 0.0],\n cls_wise_reg_weights=dict(\n {9: [2.0, 2.0, 2.0, 1.0, 1.0, 1.0, 0.0, 0.0, 1.0, 1.0]})),\n loss_cls=dict(\n type='FocalLoss',\n use_sigmoid=True,\n gamma=2.0,\n alpha=0.25,\n loss_weight=2.0),\n loss_reg=dict(\n type='SparseBox3DLoss',\n loss_box=dict(type='L1Loss', loss_weight=0.25),\n loss_centerness=dict(type='CrossEntropyLoss', use_sigmoid=True),\n loss_yawness=dict(type='GaussianFocalLoss'),\n cls_allow_reverse=[5]),\n decoder=dict(type='SparseBox3DDecoder'),\n reg_weights=[2.0, 2.0, 2.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0]))\ndataset_type = 'NuScenes3DDetTrackDataset'\ndata_root = 'data/nuscenes/'\nanno_root = 'data/nuscenes_anno_pkls/'\nfile_client_args = dict(backend='disk')\nimg_norm_cfg = dict(\n mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375], to_rgb=True)\ntrain_pipeline = [\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(\n type='LoadPointsFromFile',\n coord_type='LIDAR',\n load_dim=5,\n use_dim=5,\n file_client_args=dict(backend='disk')),\n dict(type='ResizeCropFlipImage'),\n dict(type='MultiScaleDepthMapGenerator', downsample=[4, 8, 16]),\n dict(type='BBoxRotation'),\n dict(type='PhotoMetricDistortionMultiViewImage'),\n dict(\n type='NormalizeMultiviewImage',\n mean=[123.675, 116.28, 103.53],\n std=[58.395, 57.12, 57.375],\n to_rgb=True),\n dict(\n type='CircleObjectRangeFilter',\n class_dist_thred=[55, 55, 55, 55, 55, 55, 55, 55, 55, 55]),\n dict(\n type='InstanceNameFilter',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ]),\n dict(type='NuScenesSparse4DAdaptor'),\n dict(\n type='Collect',\n keys=[\n 'img', 'timestamp', 'projection_mat', 'image_wh', 'gt_depth',\n 'focal', 'gt_bboxes_3d', 'gt_labels_3d'\n ],\n meta_keys=['T_global', 'T_global_inv', 'timestamp', 'instance_id'])\n]\ntest_pipeline = [\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(type='ResizeCropFlipImage'),\n dict(\n type='NormalizeMultiviewImage',\n mean=[123.675, 116.28, 103.53],\n std=[58.395, 57.12, 57.375],\n to_rgb=True),\n dict(type='NuScenesSparse4DAdaptor'),\n dict(\n type='Collect',\n keys=['img', 'timestamp', 'projection_mat', 'image_wh'],\n meta_keys=['T_global', 'T_global_inv', 'timestamp'])\n]\ninput_modality = dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False)\ndata_basic_config = dict(\n type='NuScenes3DDetTrackDataset',\n data_root='data/nuscenes/',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier',\n 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ],\n modality=dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False),\n version='v1.0-trainval')\ndata_aug_conf = dict(\n resize_lim=(0.4, 0.47),\n final_dim=(256, 704),\n bot_pct_lim=(0.0, 0.0),\n rot_lim=(-5.4, 5.4),\n H=900,\n W=1600,\n rand_flip=True,\n rot3d_range=[-0.3925, 0.3925])\ndata = dict(\n samples_per_gpu=20,\n workers_per_gpu=20,\n train=dict(\n type='NuScenes3DDetTrackDataset',\n data_root='data/nuscenes/',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ],\n modality=dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False),\n version='v1.0-trainval',\n ann_file='data/nuscenes_anno_pkls/nuscenes_infos_train.pkl',\n pipeline=[\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(\n type='LoadPointsFromFile',\n coord_type='LIDAR',\n load_dim=5,\n use_dim=5,\n file_client_args=dict(backend='disk')),\n dict(type='ResizeCropFlipImage'),\n dict(type='MultiScaleDepthMapGenerator', downsample=[4, 8, 16]),\n dict(type='BBoxRotation'),\n dict(type='PhotoMetricDistortionMultiViewImage'),\n dict(\n type='NormalizeMultiviewImage',\n mean=[123.675, 116.28, 103.53],\n std=[58.395, 57.12, 57.375],\n to_rgb=True),\n dict(\n type='CircleObjectRangeFilter',\n class_dist_thred=[55, 55, 55, 55, 55, 55, 55, 55, 55, 55]),\n dict(\n type='InstanceNameFilter',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian',\n 'traffic_cone'\n ]),\n dict(type='NuScenesSparse4DAdaptor'),\n dict(\n type='Collect',\n keys=[\n 'img', 'timestamp', 'projection_mat', 'image_wh',\n 'gt_depth', 'focal', 'gt_bboxes_3d', 'gt_labels_3d'\n ],\n meta_keys=[\n 'T_global', 'T_global_inv', 'timestamp', 'instance_id'\n ])\n ],\n test_mode=False,\n data_aug_conf=dict(\n resize_lim=(0.4, 0.47),\n final_dim=(256, 704),\n bot_pct_lim=(0.0, 0.0),\n rot_lim=(-5.4, 5.4),\n H=900,\n W=1600,\n rand_flip=True,\n rot3d_range=[-0.3925, 0.3925]),\n with_seq_flag=True,\n sequences_split_num=2,\n keep_consistent_seq_aug=True),\n val=dict(\n type='NuScenes3DDetTrackDataset',\n data_root='data/nuscenes/',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ],\n modality=dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False),\n version='v1.0-trainval',\n ann_file='data/nuscenes_anno_pkls/nuscenes_infos_val.pkl',\n pipeline=[\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(type='ResizeCropFlipImage'),\n dict(\n type='NormalizeMultiviewImage',\n mean=[123.675, 116.28, 103.53],\n std=[58.395, 57.12, 57.375],\n to_rgb=True),\n dict(type='NuScenesSparse4DAdaptor'),\n dict(\n type='Collect',\n keys=['img', 'timestamp', 'projection_mat', 'image_wh'],\n meta_keys=['T_global', 'T_global_inv', 'timestamp'])\n ],\n data_aug_conf=dict(\n resize_lim=(0.4, 0.47),\n final_dim=(256, 704),\n bot_pct_lim=(0.0, 0.0),\n rot_lim=(-5.4, 5.4),\n H=900,\n W=1600,\n rand_flip=True,\n rot3d_range=[-0.3925, 0.3925]),\n test_mode=True,\n tracking=True,\n tracking_threshold=0.2),\n test=dict(\n type='NuScenes3DDetTrackDataset',\n data_root='data/nuscenes/',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ],\n modality=dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False),\n version='v1.0-trainval',\n ann_file='data/nuscenes_anno_pkls/nuscenes_infos_val.pkl',\n pipeline=[\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(type='ResizeCropFlipImage'),\n dict(\n type='NormalizeMultiviewImage',\n mean=[123.675, 116.28, 103.53],\n std=[58.395, 57.12, 57.375],\n to_rgb=True),\n dict(type='NuScenesSparse4DAdaptor'),\n dict(\n type='Collect',\n keys=['img', 'timestamp', 'projection_mat', 'image_wh'],\n meta_keys=['T_global', 'T_global_inv', 'timestamp'])\n ],\n data_aug_conf=dict(\n resize_lim=(0.4, 0.47),\n final_dim=(256, 704),\n bot_pct_lim=(0.0, 0.0),\n rot_lim=(-5.4, 5.4),\n H=900,\n W=1600,\n rand_flip=True,\n rot3d_range=[-0.3925, 0.3925]),\n test_mode=True,\n tracking=True,\n tracking_threshold=0.2))\noptimizer = dict(\n type='AdamW',\n lr=0.0006,\n weight_decay=0.001,\n paramwise_cfg=dict(custom_keys=dict(img_backbone=dict(lr_mult=0.5))))\noptimizer_config = dict(grad_clip=dict(max_norm=25, norm_type=2))\nlr_config = dict(\n policy='CosineAnnealing',\n warmup='linear',\n warmup_iters=500,\n warmup_ratio=0.3333333333333333,\n min_lr_ratio=0.001)\nrunner = dict(type='IterBasedRunner', max_iters=17500)\nvis_pipeline = [\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(type='Collect', keys=['img'], meta_keys=['timestamp', 'lidar2img'])\n]\nevaluation = dict(\n interval=3500,\n pipeline=[\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(\n type='Collect', keys=['img'], meta_keys=['timestamp', 'lidar2img'])\n ])\ngpu_ids = range(0, 1)\n", "seed": 0, "exp_name": "sparse4dv3_temporal_r50_1x8_bs6_256x704.py"} diff --git a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20260210_161912.log b/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20260210_161912.log deleted file mode 100644 index c250abf638fb12ca0a5421fc148cc37528715417..0000000000000000000000000000000000000000 --- a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20260210_161912.log +++ /dev/null @@ -1,3221 +0,0 @@ -2026-02-10 16:19:12,076 - mmdet - INFO - Environment info: ------------------------------------------------------------- -sys.platform: linux -Python: 3.10.12 (main, Aug 15 2025, 14:32:43) [GCC 11.4.0] -CUDA available: True -GPU 0,1,2,3,4,5,6,7: BW200_LC, UBB BW1000_LC -CUDA_HOME: /opt/dtk -NVCC: Not Available -GCC: x86_64-linux-gnu-gcc (Ubuntu 11.4.0-1ubuntu1~22.04) 11.4.0 -PyTorch: 2.5.1 -PyTorch compiling details: PyTorch built with: - - GCC 10.3 - - C++ Version: 201703 - - Intel(R) Math Kernel Library Version 2020.0.4 Product Build 20200917 for Intel(R) 64 architecture applications - - OpenMP 201511 (a.k.a. OpenMP 4.5) - - LAPACK is enabled (usually provided by MKL) - - NNPACK is enabled - - CPU capability usage: AVX2 - - HIP Runtime 6.3.25521 - - MIOpen 2.18.0 - - Magma 2.8.0 - - Build settings: BLAS_INFO=mkl, BUILD_TYPE=Release, CXX_COMPILER=/opt/rh/gcc-toolset-10/root/usr/bin/c++, CXX_FLAGS= -D_GLIBCXX_USE_CXX11_ABI=1 -fvisibility-inlines-hidden -DUSE_PTHREADPOOL -DNDEBUG -DUSE_KINETO -DLIBKINETO_NOCUPTI -DLIBKINETO_NOXPUPTI=ON -DUSE_FBGEMM -DUSE_PYTORCH_QNNPACK -DUSE_XNNPACK -DSYMBOLICATE_MOBILE_DEBUG_HANDLE -O2 -fPIC -Wall -Wextra -Werror=return-type -Werror=non-virtual-dtor -Werror=bool-operation -Wnarrowing -Wno-missing-field-initializers -Wno-type-limits -Wno-array-bounds -Wno-unknown-pragmas -Wno-unused-parameter -Wno-strict-overflow -Wno-strict-aliasing -Wno-stringop-overflow -Wsuggest-override -Wno-psabi -Wno-error=old-style-cast -Wno-missing-braces -fdiagnostics-color=always -faligned-new -Wno-unused-but-set-variable -Wno-maybe-uninitialized -fno-math-errno -fno-trapping-math -Werror=format -Wno-stringop-overflow, FORCE_FALLBACK_CUDA_MPI=1, LAPACK_INFO=mkl, PERF_WITH_AVX=1, PERF_WITH_AVX2=1, TORCH_VERSION=2.5.1, USE_CUDA=0, USE_CUDNN=OFF, USE_CUSPARSELT=OFF, USE_EXCEPTION_PTR=1, USE_GFLAGS=1, USE_GLOG=1, USE_GLOO=1, USE_MKL=ON, USE_MKLDNN=0, USE_MPI=1, USE_NCCL=1, USE_NNPACK=ON, USE_OPENMP=1, USE_ROCM=ON, USE_ROCM_KERNEL_ASSERT=OFF, - -TorchVision: 0.20.1 -OpenCV: 4.12.0 -MMCV: 1.6.1 -MMCV Compiler: GCC 10.3 -MMCV CUDA Compiler: rocm not available -MMDetection: 2.25.1+ ------------------------------------------------------------- - -2026-02-10 16:19:13,062 - mmdet - INFO - Distributed training: True -2026-02-10 16:19:14,021 - mmdet - INFO - Config: -plugin = True -plugin_dir = 'projects/mmdet3d_plugin/' -dist_params = dict(backend='nccl') -log_level = 'INFO' -work_dir = './work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704' -total_batch_size = 160 -num_gpus = 8 -batch_size = 20 -num_iters_per_epoch = 175 -num_epochs = 100 -checkpoint_epoch_interval = 20 -checkpoint_config = dict(interval=3500) -log_config = dict( - interval=1, - hooks=[ - dict(type='TextLoggerHook', by_epoch=False), - dict(type='TensorboardLoggerHook') - ]) -load_from = None -resume_from = None -workflow = [('train', 1)] -fp16 = dict(loss_scale=32.0) -input_shape = (704, 256) -tracking_test = True -tracking_threshold = 0.2 -class_names = [ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier', - 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' -] -num_classes = 10 -embed_dims = 256 -num_groups = 8 -num_decoder = 6 -num_single_frame_decoder = 1 -use_deformable_func = True -strides = [4, 8, 16, 32] -num_levels = 4 -num_depth_layers = 3 -drop_out = 0.1 -temporal = True -decouple_attn = True -with_quality_estimation = True -model = dict( - type='Sparse4D', - use_grid_mask=True, - use_deformable_func=True, - img_backbone=dict( - type='ResNet', - depth=50, - num_stages=4, - frozen_stages=-1, - norm_eval=False, - style='pytorch', - with_cp=True, - out_indices=(0, 1, 2, 3), - norm_cfg=dict(type='BN', requires_grad=True), - pretrained='ckpt/resnet50-19c8e357.pth'), - img_neck=dict( - type='FPN', - num_outs=4, - start_level=0, - out_channels=256, - add_extra_convs='on_output', - relu_before_extra_convs=True, - in_channels=[256, 512, 1024, 2048]), - depth_branch=dict( - type='DenseDepthNet', - embed_dims=256, - num_depth_layers=3, - loss_weight=0.2), - head=dict( - type='Sparse4DHead', - cls_threshold_to_reg=0.05, - decouple_attn=True, - instance_bank=dict( - type='InstanceBank', - num_anchor=900, - embed_dims=256, - anchor='nuscenes_kmeans900.npy', - anchor_handler=dict(type='SparseBox3DKeyPointsGenerator'), - num_temp_instances=600, - confidence_decay=0.6, - feat_grad=False), - anchor_encoder=dict( - type='SparseBox3DEncoder', - vel_dims=3, - embed_dims=[128, 32, 32, 64], - mode='cat', - output_fc=False, - in_loops=1, - out_loops=4), - num_single_frame_decoder=1, - operation_order=[ - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine' - ], - temp_graph_model=dict( - type='MultiheadAttention', - embed_dims=512, - num_heads=8, - batch_first=True, - dropout=0.1), - graph_model=dict( - type='MultiheadAttention', - embed_dims=512, - num_heads=8, - batch_first=True, - dropout=0.1), - norm_layer=dict(type='LN', normalized_shape=256), - ffn=dict( - type='AsymmetricFFN', - in_channels=512, - pre_norm=dict(type='LN'), - embed_dims=256, - feedforward_channels=1024, - num_fcs=2, - ffn_drop=0.1, - act_cfg=dict(type='ReLU', inplace=True)), - deformable_model=dict( - type='DeformableFeatureAggregation', - embed_dims=256, - num_groups=8, - num_levels=4, - num_cams=6, - attn_drop=0.15, - use_deformable_func=True, - use_camera_embed=True, - residual_mode='cat', - kps_generator=dict( - type='SparseBox3DKeyPointsGenerator', - num_learnable_pts=6, - fix_scale=[[0, 0, 0], [0.45, 0, 0], [-0.45, 0, 0], - [0, 0.45, 0], [0, -0.45, 0], [0, 0, 0.45], - [0, 0, -0.45]])), - refine_layer=dict( - type='SparseBox3DRefinementModule', - embed_dims=256, - num_cls=10, - refine_yaw=True, - with_quality_estimation=True), - sampler=dict( - type='SparseBox3DTarget', - num_dn_groups=5, - num_temp_dn_groups=3, - dn_noise_scale=[2.0, 2.0, 2.0, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5], - max_dn_gt=32, - add_neg_dn=True, - cls_weight=2.0, - box_weight=0.25, - reg_weights=[2.0, 2.0, 2.0, 0.5, 0.5, 0.5, 0.0, 0.0, 0.0, 0.0], - cls_wise_reg_weights=dict( - {9: [2.0, 2.0, 2.0, 1.0, 1.0, 1.0, 0.0, 0.0, 1.0, 1.0]})), - loss_cls=dict( - type='FocalLoss', - use_sigmoid=True, - gamma=2.0, - alpha=0.25, - loss_weight=2.0), - loss_reg=dict( - type='SparseBox3DLoss', - loss_box=dict(type='L1Loss', loss_weight=0.25), - loss_centerness=dict(type='CrossEntropyLoss', use_sigmoid=True), - loss_yawness=dict(type='GaussianFocalLoss'), - cls_allow_reverse=[5]), - decoder=dict(type='SparseBox3DDecoder'), - reg_weights=[2.0, 2.0, 2.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0])) -dataset_type = 'NuScenes3DDetTrackDataset' -data_root = 'data/nuscenes/' -anno_root = 'data/nuscenes_anno_pkls/' -file_client_args = dict(backend='disk') -img_norm_cfg = dict( - mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375], to_rgb=True) -train_pipeline = [ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict( - type='LoadPointsFromFile', - coord_type='LIDAR', - load_dim=5, - use_dim=5, - file_client_args=dict(backend='disk')), - dict(type='ResizeCropFlipImage'), - dict(type='MultiScaleDepthMapGenerator', downsample=[4, 8, 16]), - dict(type='BBoxRotation'), - dict(type='PhotoMetricDistortionMultiViewImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict( - type='CircleObjectRangeFilter', - class_dist_thred=[55, 55, 55, 55, 55, 55, 55, 55, 55, 55]), - dict( - type='InstanceNameFilter', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ]), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=[ - 'img', 'timestamp', 'projection_mat', 'image_wh', 'gt_depth', - 'focal', 'gt_bboxes_3d', 'gt_labels_3d' - ], - meta_keys=['T_global', 'T_global_inv', 'timestamp', 'instance_id']) -] -test_pipeline = [ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='ResizeCropFlipImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=['img', 'timestamp', 'projection_mat', 'image_wh'], - meta_keys=['T_global', 'T_global_inv', 'timestamp']) -] -input_modality = dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False) -data_basic_config = dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier', - 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval') -data_aug_conf = dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]) -data = dict( - samples_per_gpu=20, - workers_per_gpu=20, - train=dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval', - ann_file='data/nuscenes_anno_pkls/nuscenes_infos_train.pkl', - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict( - type='LoadPointsFromFile', - coord_type='LIDAR', - load_dim=5, - use_dim=5, - file_client_args=dict(backend='disk')), - dict(type='ResizeCropFlipImage'), - dict(type='MultiScaleDepthMapGenerator', downsample=[4, 8, 16]), - dict(type='BBoxRotation'), - dict(type='PhotoMetricDistortionMultiViewImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict( - type='CircleObjectRangeFilter', - class_dist_thred=[55, 55, 55, 55, 55, 55, 55, 55, 55, 55]), - dict( - type='InstanceNameFilter', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', - 'traffic_cone' - ]), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=[ - 'img', 'timestamp', 'projection_mat', 'image_wh', - 'gt_depth', 'focal', 'gt_bboxes_3d', 'gt_labels_3d' - ], - meta_keys=[ - 'T_global', 'T_global_inv', 'timestamp', 'instance_id' - ]) - ], - test_mode=False, - data_aug_conf=dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]), - with_seq_flag=True, - sequences_split_num=2, - keep_consistent_seq_aug=True), - val=dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval', - ann_file='data/nuscenes_anno_pkls/nuscenes_infos_val.pkl', - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='ResizeCropFlipImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=['img', 'timestamp', 'projection_mat', 'image_wh'], - meta_keys=['T_global', 'T_global_inv', 'timestamp']) - ], - data_aug_conf=dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]), - test_mode=True, - tracking=True, - tracking_threshold=0.2), - test=dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval', - ann_file='data/nuscenes_anno_pkls/nuscenes_infos_val.pkl', - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='ResizeCropFlipImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=['img', 'timestamp', 'projection_mat', 'image_wh'], - meta_keys=['T_global', 'T_global_inv', 'timestamp']) - ], - data_aug_conf=dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]), - test_mode=True, - tracking=True, - tracking_threshold=0.2)) -optimizer = dict( - type='AdamW', - lr=0.0006, - weight_decay=0.001, - paramwise_cfg=dict(custom_keys=dict(img_backbone=dict(lr_mult=0.5)))) -optimizer_config = dict(grad_clip=dict(max_norm=25, norm_type=2)) -lr_config = dict( - policy='CosineAnnealing', - warmup='linear', - warmup_iters=500, - warmup_ratio=0.3333333333333333, - min_lr_ratio=0.001) -runner = dict(type='IterBasedRunner', max_iters=17500) -vis_pipeline = [ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='Collect', keys=['img'], meta_keys=['timestamp', 'lidar2img']) -] -evaluation = dict( - interval=3500, - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict( - type='Collect', keys=['img'], meta_keys=['timestamp', 'lidar2img']) - ]) -gpu_ids = range(0, 8) - -2026-02-10 16:19:14,022 - mmdet - INFO - Set random seed to 0, deterministic: False -2026-02-10 16:19:14,427 - mmdet - INFO - initialize ResNet with init_cfg {'type': 'Pretrained', 'checkpoint': 'ckpt/resnet50-19c8e357.pth'} -2026-02-10 16:19:14,939 - mmdet - INFO - initialize FPN with init_cfg {'type': 'Xavier', 'layer': 'Conv2d', 'distribution': 'uniform'} -Name of parameter - Initialization information - -img_backbone.conv1.weight - torch.Size([64, 3, 7, 7]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.bn1.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.bn1.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.conv1.weight - torch.Size([64, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn1.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn1.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.conv2.weight - torch.Size([64, 64, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn2.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn2.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.conv3.weight - torch.Size([256, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn3.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn3.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.downsample.0.weight - torch.Size([256, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.downsample.1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.downsample.1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.conv1.weight - torch.Size([64, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn1.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn1.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.conv2.weight - torch.Size([64, 64, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn2.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn2.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.conv3.weight - torch.Size([256, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn3.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn3.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.conv1.weight - torch.Size([64, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn1.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn1.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.conv2.weight - torch.Size([64, 64, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn2.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn2.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.conv3.weight - torch.Size([256, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn3.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn3.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.conv1.weight - torch.Size([128, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn1.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn1.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.conv2.weight - torch.Size([128, 128, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn2.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn2.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.conv3.weight - torch.Size([512, 128, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn3.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn3.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.downsample.0.weight - torch.Size([512, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.downsample.1.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.downsample.1.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.conv1.weight - torch.Size([128, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn1.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn1.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.conv2.weight - torch.Size([128, 128, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn2.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn2.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.conv3.weight - torch.Size([512, 128, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn3.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn3.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.conv1.weight - torch.Size([128, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn1.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn1.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.conv2.weight - torch.Size([128, 128, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn2.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn2.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.conv3.weight - torch.Size([512, 128, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn3.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn3.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.conv1.weight - torch.Size([128, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn1.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn1.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.conv2.weight - torch.Size([128, 128, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn2.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn2.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.conv3.weight - torch.Size([512, 128, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn3.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn3.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.conv1.weight - torch.Size([256, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.downsample.0.weight - torch.Size([1024, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.downsample.1.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.downsample.1.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.conv1.weight - torch.Size([512, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn1.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn1.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.conv2.weight - torch.Size([512, 512, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn2.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn2.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.conv3.weight - torch.Size([2048, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn3.weight - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn3.bias - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.downsample.0.weight - torch.Size([2048, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.downsample.1.weight - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.downsample.1.bias - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.conv1.weight - torch.Size([512, 2048, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn1.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn1.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.conv2.weight - torch.Size([512, 512, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn2.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn2.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.conv3.weight - torch.Size([2048, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn3.weight - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn3.bias - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.conv1.weight - torch.Size([512, 2048, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn1.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn1.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.conv2.weight - torch.Size([512, 512, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn2.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn2.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.conv3.weight - torch.Size([2048, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn3.weight - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn3.bias - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_neck.lateral_convs.0.conv.weight - torch.Size([256, 256, 1, 1]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.lateral_convs.0.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.lateral_convs.1.conv.weight - torch.Size([256, 512, 1, 1]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.lateral_convs.1.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.lateral_convs.2.conv.weight - torch.Size([256, 1024, 1, 1]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.lateral_convs.2.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.lateral_convs.3.conv.weight - torch.Size([256, 2048, 1, 1]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.lateral_convs.3.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.fpn_convs.0.conv.weight - torch.Size([256, 256, 3, 3]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.fpn_convs.0.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.fpn_convs.1.conv.weight - torch.Size([256, 256, 3, 3]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.fpn_convs.1.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.fpn_convs.2.conv.weight - torch.Size([256, 256, 3, 3]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.fpn_convs.2.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.fpn_convs.3.conv.weight - torch.Size([256, 256, 3, 3]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.fpn_convs.3.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.instance_bank.anchor - torch.Size([900, 11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.instance_bank.instance_feature - torch.Size([900, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.instance_bank.anchor_handler.fix_scale - torch.Size([1, 3]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.0.weight - torch.Size([128, 3]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.0.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.2.weight - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.2.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.3.weight - torch.Size([128, 128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.3.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.5.weight - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.5.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.6.weight - torch.Size([128, 128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.6.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.8.weight - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.8.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.9.weight - torch.Size([128, 128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.9.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.11.weight - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.11.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.0.weight - torch.Size([32, 3]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.0.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.2.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.2.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.3.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.3.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.5.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.5.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.6.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.6.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.8.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.8.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.9.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.9.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.11.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.11.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.0.weight - torch.Size([32, 2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.0.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.2.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.2.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.3.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.3.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.5.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.5.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.6.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.6.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.8.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.8.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.9.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.9.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.11.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.11.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.0.weight - torch.Size([64, 3]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.0.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.2.weight - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.2.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.3.weight - torch.Size([64, 64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.3.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.5.weight - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.5.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.6.weight - torch.Size([64, 64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.6.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.8.weight - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.8.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.9.weight - torch.Size([64, 64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.9.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.11.weight - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.11.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.1.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.1.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.1.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.1.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.1.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.1.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.1.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.1.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.3.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.4.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.4.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.4.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.4.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.5.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.5.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.5.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.5.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.6.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.6.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.8.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.8.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.8.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.8.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.8.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.8.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.8.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.8.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.10.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.11.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.11.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.11.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.11.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.12.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.12.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.12.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.12.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.13.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.13.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.15.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.15.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.15.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.15.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.15.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.15.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.15.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.15.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.16.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.16.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.17.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.18.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.18.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.18.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.18.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.19.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.19.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.19.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.19.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.20.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.20.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.22.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.22.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.22.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.22.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.22.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.22.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.22.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.22.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.23.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.23.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.24.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.25.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.25.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.25.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.25.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.26.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.26.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.26.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.26.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.27.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.27.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.29.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.29.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.29.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.29.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.29.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.29.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.29.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.29.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.30.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.30.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.31.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.32.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.32.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.32.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.32.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.33.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.33.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.33.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.33.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.34.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.34.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.36.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.36.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.36.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.36.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.36.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.36.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.36.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.36.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.37.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.37.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.38.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.fc_before.weight - torch.Size([512, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.fc_after.weight - torch.Size([256, 512]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.0.weight - torch.Size([1, 256, 1, 1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.0.bias - torch.Size([1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.1.weight - torch.Size([1, 256, 1, 1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.1.bias - torch.Size([1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.2.weight - torch.Size([1, 256, 1, 1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.2.bias - torch.Size([1]): -The value is the same before and after calling `init_weights` of Sparse4D -2026-02-10 16:19:15,075 - mmdet - INFO - Model: -Sparse4D( - (img_backbone): ResNet( - (conv1): Conv2d(3, 64, kernel_size=(7, 7), stride=(2, 2), padding=(3, 3), bias=False) - (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (maxpool): MaxPool2d(kernel_size=3, stride=2, padding=1, dilation=1, ceil_mode=False) - (layer1): ResLayer( - (0): Bottleneck( - (conv1): Conv2d(64, 64, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (downsample): Sequential( - (0): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - ) - ) - (1): Bottleneck( - (conv1): Conv2d(256, 64, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (2): Bottleneck( - (conv1): Conv2d(256, 64, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - ) - (layer2): ResLayer( - (0): Bottleneck( - (conv1): Conv2d(256, 128, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (downsample): Sequential( - (0): Conv2d(256, 512, kernel_size=(1, 1), stride=(2, 2), bias=False) - (1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - ) - ) - (1): Bottleneck( - (conv1): Conv2d(512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (2): Bottleneck( - (conv1): Conv2d(512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (3): Bottleneck( - (conv1): Conv2d(512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - ) - (layer3): ResLayer( - (0): Bottleneck( - (conv1): Conv2d(512, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (downsample): Sequential( - (0): Conv2d(512, 1024, kernel_size=(1, 1), stride=(2, 2), bias=False) - (1): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - ) - ) - (1): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (2): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (3): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (4): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (5): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - ) - (layer4): ResLayer( - (0): Bottleneck( - (conv1): Conv2d(1024, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(512, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (downsample): Sequential( - (0): Conv2d(1024, 2048, kernel_size=(1, 1), stride=(2, 2), bias=False) - (1): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - ) - ) - (1): Bottleneck( - (conv1): Conv2d(2048, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(512, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (2): Bottleneck( - (conv1): Conv2d(2048, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(512, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - ) - ) - init_cfg={'type': 'Pretrained', 'checkpoint': 'ckpt/resnet50-19c8e357.pth'} - (img_neck): FPN( - (lateral_convs): ModuleList( - (0): ConvModule( - (conv): Conv2d(256, 256, kernel_size=(1, 1), stride=(1, 1)) - ) - (1): ConvModule( - (conv): Conv2d(512, 256, kernel_size=(1, 1), stride=(1, 1)) - ) - (2): ConvModule( - (conv): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1)) - ) - (3): ConvModule( - (conv): Conv2d(2048, 256, kernel_size=(1, 1), stride=(1, 1)) - ) - ) - (fpn_convs): ModuleList( - (0-3): 4 x ConvModule( - (conv): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1)) - ) - ) - ) - init_cfg={'type': 'Xavier', 'layer': 'Conv2d', 'distribution': 'uniform'} - (head): Sparse4DHead( - (instance_bank): InstanceBank( - (anchor_handler): SparseBox3DKeyPointsGenerator() - ) - (anchor_encoder): SparseBox3DEncoder( - (pos_fc): Sequential( - (0): Linear(in_features=3, out_features=128, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((128,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=128, out_features=128, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((128,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=128, out_features=128, bias=True) - (7): ReLU(inplace=True) - (8): LayerNorm((128,), eps=1e-05, elementwise_affine=True) - (9): Linear(in_features=128, out_features=128, bias=True) - (10): ReLU(inplace=True) - (11): LayerNorm((128,), eps=1e-05, elementwise_affine=True) - ) - (size_fc): Sequential( - (0): Linear(in_features=3, out_features=32, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=32, out_features=32, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=32, out_features=32, bias=True) - (7): ReLU(inplace=True) - (8): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (9): Linear(in_features=32, out_features=32, bias=True) - (10): ReLU(inplace=True) - (11): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - ) - (yaw_fc): Sequential( - (0): Linear(in_features=2, out_features=32, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=32, out_features=32, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=32, out_features=32, bias=True) - (7): ReLU(inplace=True) - (8): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (9): Linear(in_features=32, out_features=32, bias=True) - (10): ReLU(inplace=True) - (11): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - ) - (vel_fc): Sequential( - (0): Linear(in_features=3, out_features=64, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((64,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=64, out_features=64, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((64,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=64, out_features=64, bias=True) - (7): ReLU(inplace=True) - (8): LayerNorm((64,), eps=1e-05, elementwise_affine=True) - (9): Linear(in_features=64, out_features=64, bias=True) - (10): ReLU(inplace=True) - (11): LayerNorm((64,), eps=1e-05, elementwise_affine=True) - ) - ) - (loss_cls): FocalLoss() - (loss_reg): SparseBox3DLoss( - (loss_box): L1Loss() - (loss_cns): CrossEntropyLoss(avg_non_ignore=False) - (loss_yns): GaussianFocalLoss() - ) - (layers): ModuleList( - (0): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (1): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (4-5): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (6): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (7): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (8): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (11-12): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (13): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (14): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (15): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (16): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (17): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (18-19): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (20): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (21): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (22): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (23): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (24): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (25-26): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (27): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (28): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (29): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (30): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (31): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (32-33): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (34): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (35): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (36): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (37): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (38): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - ) - (fc_before): Linear(in_features=256, out_features=512, bias=False) - (fc_after): Linear(in_features=512, out_features=256, bias=False) - ) - (depth_branch): DenseDepthNet( - (depth_layers): ModuleList( - (0-2): 3 x Conv2d(256, 1, kernel_size=(1, 1), stride=(1, 1)) - ) - ) - (grid_mask): GridMask() -) -2026-02-10 16:19:30,471 - mmdet - INFO - Start running, host: root@bw28, work_dir: /workspace/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704 -2026-02-10 16:19:30,472 - mmdet - INFO - Hooks will be executed in the following order: -before_run: -(VERY_HIGH ) CosineAnnealingLrUpdaterHook -(ABOVE_NORMAL) Fp16OptimizerHook -(NORMAL ) CheckpointHook -(NORMAL ) CustomDistEvalHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -before_train_epoch: -(VERY_HIGH ) CosineAnnealingLrUpdaterHook -(NORMAL ) CustomDistEvalHook -(LOW ) IterTimerHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -before_train_iter: -(VERY_HIGH ) CosineAnnealingLrUpdaterHook -(NORMAL ) CustomDistEvalHook -(LOW ) IterTimerHook - -------------------- -after_train_iter: -(ABOVE_NORMAL) Fp16OptimizerHook -(NORMAL ) CheckpointHook -(NORMAL ) CustomDistEvalHook -(NORMAL ) ProfilerHook -(LOW ) IterTimerHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -after_train_epoch: -(NORMAL ) CheckpointHook -(NORMAL ) CustomDistEvalHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -before_val_epoch: -(LOW ) IterTimerHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -before_val_iter: -(LOW ) IterTimerHook - -------------------- -after_val_iter: -(LOW ) IterTimerHook - -------------------- -after_val_epoch: -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -after_run: -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -2026-02-10 16:19:30,472 - mmdet - INFO - workflow: [('train', 1)], max: 17500 iters -2026-02-10 16:19:30,474 - mmdet - INFO - Checkpoints will be saved to /workspace/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704 by HardDiskBackend. diff --git a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20260210_161912.log.json b/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20260210_161912.log.json deleted file mode 100644 index 9b7a36a4c4faf65b6fbf808cfd0607ffa5984a6f..0000000000000000000000000000000000000000 --- a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20260210_161912.log.json +++ /dev/null @@ -1 +0,0 @@ -{"env_info": "sys.platform: linux\nPython: 3.10.12 (main, Aug 15 2025, 14:32:43) [GCC 11.4.0]\nCUDA available: True\nGPU 0,1,2,3,4,5,6,7: BW200_LC, UBB BW1000_LC\nCUDA_HOME: /opt/dtk\nNVCC: Not Available\nGCC: x86_64-linux-gnu-gcc (Ubuntu 11.4.0-1ubuntu1~22.04) 11.4.0\nPyTorch: 2.5.1\nPyTorch compiling details: PyTorch built with:\n - GCC 10.3\n - C++ Version: 201703\n - Intel(R) Math Kernel Library Version 2020.0.4 Product Build 20200917 for Intel(R) 64 architecture applications\n - OpenMP 201511 (a.k.a. OpenMP 4.5)\n - LAPACK is enabled (usually provided by MKL)\n - NNPACK is enabled\n - CPU capability usage: AVX2\n - HIP Runtime 6.3.25521\n - MIOpen 2.18.0\n - Magma 2.8.0\n - Build settings: BLAS_INFO=mkl, BUILD_TYPE=Release, CXX_COMPILER=/opt/rh/gcc-toolset-10/root/usr/bin/c++, CXX_FLAGS= -D_GLIBCXX_USE_CXX11_ABI=1 -fvisibility-inlines-hidden -DUSE_PTHREADPOOL -DNDEBUG -DUSE_KINETO -DLIBKINETO_NOCUPTI -DLIBKINETO_NOXPUPTI=ON -DUSE_FBGEMM -DUSE_PYTORCH_QNNPACK -DUSE_XNNPACK -DSYMBOLICATE_MOBILE_DEBUG_HANDLE -O2 -fPIC -Wall -Wextra -Werror=return-type -Werror=non-virtual-dtor -Werror=bool-operation -Wnarrowing -Wno-missing-field-initializers -Wno-type-limits -Wno-array-bounds -Wno-unknown-pragmas -Wno-unused-parameter -Wno-strict-overflow -Wno-strict-aliasing -Wno-stringop-overflow -Wsuggest-override -Wno-psabi -Wno-error=old-style-cast -Wno-missing-braces -fdiagnostics-color=always -faligned-new -Wno-unused-but-set-variable -Wno-maybe-uninitialized -fno-math-errno -fno-trapping-math -Werror=format -Wno-stringop-overflow, FORCE_FALLBACK_CUDA_MPI=1, LAPACK_INFO=mkl, PERF_WITH_AVX=1, PERF_WITH_AVX2=1, TORCH_VERSION=2.5.1, USE_CUDA=0, USE_CUDNN=OFF, USE_CUSPARSELT=OFF, USE_EXCEPTION_PTR=1, USE_GFLAGS=1, USE_GLOG=1, USE_GLOO=1, USE_MKL=ON, USE_MKLDNN=0, USE_MPI=1, USE_NCCL=1, USE_NNPACK=ON, USE_OPENMP=1, USE_ROCM=ON, USE_ROCM_KERNEL_ASSERT=OFF, \n\nTorchVision: 0.20.1\nOpenCV: 4.12.0\nMMCV: 1.6.1\nMMCV Compiler: GCC 10.3\nMMCV CUDA Compiler: rocm not available\nMMDetection: 2.25.1+", "config": "plugin = True\nplugin_dir = 'projects/mmdet3d_plugin/'\ndist_params = dict(backend='nccl')\nlog_level = 'INFO'\nwork_dir = './work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704'\ntotal_batch_size = 160\nnum_gpus = 8\nbatch_size = 20\nnum_iters_per_epoch = 175\nnum_epochs = 100\ncheckpoint_epoch_interval = 20\ncheckpoint_config = dict(interval=3500)\nlog_config = dict(\n interval=1,\n hooks=[\n dict(type='TextLoggerHook', by_epoch=False),\n dict(type='TensorboardLoggerHook')\n ])\nload_from = None\nresume_from = None\nworkflow = [('train', 1)]\nfp16 = dict(loss_scale=32.0)\ninput_shape = (704, 256)\ntracking_test = True\ntracking_threshold = 0.2\nclass_names = [\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier',\n 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n]\nnum_classes = 10\nembed_dims = 256\nnum_groups = 8\nnum_decoder = 6\nnum_single_frame_decoder = 1\nuse_deformable_func = True\nstrides = [4, 8, 16, 32]\nnum_levels = 4\nnum_depth_layers = 3\ndrop_out = 0.1\ntemporal = True\ndecouple_attn = True\nwith_quality_estimation = True\nmodel = dict(\n type='Sparse4D',\n use_grid_mask=True,\n use_deformable_func=True,\n img_backbone=dict(\n type='ResNet',\n depth=50,\n num_stages=4,\n frozen_stages=-1,\n norm_eval=False,\n style='pytorch',\n with_cp=True,\n out_indices=(0, 1, 2, 3),\n norm_cfg=dict(type='BN', requires_grad=True),\n pretrained='ckpt/resnet50-19c8e357.pth'),\n img_neck=dict(\n type='FPN',\n num_outs=4,\n start_level=0,\n out_channels=256,\n add_extra_convs='on_output',\n relu_before_extra_convs=True,\n in_channels=[256, 512, 1024, 2048]),\n depth_branch=dict(\n type='DenseDepthNet',\n embed_dims=256,\n num_depth_layers=3,\n loss_weight=0.2),\n head=dict(\n type='Sparse4DHead',\n cls_threshold_to_reg=0.05,\n decouple_attn=True,\n instance_bank=dict(\n type='InstanceBank',\n num_anchor=900,\n embed_dims=256,\n anchor='nuscenes_kmeans900.npy',\n anchor_handler=dict(type='SparseBox3DKeyPointsGenerator'),\n num_temp_instances=600,\n confidence_decay=0.6,\n feat_grad=False),\n anchor_encoder=dict(\n type='SparseBox3DEncoder',\n vel_dims=3,\n embed_dims=[128, 32, 32, 64],\n mode='cat',\n output_fc=False,\n in_loops=1,\n out_loops=4),\n num_single_frame_decoder=1,\n operation_order=[\n 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm',\n 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm',\n 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm',\n 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm',\n 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm',\n 'deformable', 'ffn', 'norm', 'refine'\n ],\n temp_graph_model=dict(\n type='MultiheadAttention',\n embed_dims=512,\n num_heads=8,\n batch_first=True,\n dropout=0.1),\n graph_model=dict(\n type='MultiheadAttention',\n embed_dims=512,\n num_heads=8,\n batch_first=True,\n dropout=0.1),\n norm_layer=dict(type='LN', normalized_shape=256),\n ffn=dict(\n type='AsymmetricFFN',\n in_channels=512,\n pre_norm=dict(type='LN'),\n embed_dims=256,\n feedforward_channels=1024,\n num_fcs=2,\n ffn_drop=0.1,\n act_cfg=dict(type='ReLU', inplace=True)),\n deformable_model=dict(\n type='DeformableFeatureAggregation',\n embed_dims=256,\n num_groups=8,\n num_levels=4,\n num_cams=6,\n attn_drop=0.15,\n use_deformable_func=True,\n use_camera_embed=True,\n residual_mode='cat',\n kps_generator=dict(\n type='SparseBox3DKeyPointsGenerator',\n num_learnable_pts=6,\n fix_scale=[[0, 0, 0], [0.45, 0, 0], [-0.45, 0, 0],\n [0, 0.45, 0], [0, -0.45, 0], [0, 0, 0.45],\n [0, 0, -0.45]])),\n refine_layer=dict(\n type='SparseBox3DRefinementModule',\n embed_dims=256,\n num_cls=10,\n refine_yaw=True,\n with_quality_estimation=True),\n sampler=dict(\n type='SparseBox3DTarget',\n num_dn_groups=5,\n num_temp_dn_groups=3,\n dn_noise_scale=[2.0, 2.0, 2.0, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5],\n max_dn_gt=32,\n add_neg_dn=True,\n cls_weight=2.0,\n box_weight=0.25,\n reg_weights=[2.0, 2.0, 2.0, 0.5, 0.5, 0.5, 0.0, 0.0, 0.0, 0.0],\n cls_wise_reg_weights=dict(\n {9: [2.0, 2.0, 2.0, 1.0, 1.0, 1.0, 0.0, 0.0, 1.0, 1.0]})),\n loss_cls=dict(\n type='FocalLoss',\n use_sigmoid=True,\n gamma=2.0,\n alpha=0.25,\n loss_weight=2.0),\n loss_reg=dict(\n type='SparseBox3DLoss',\n loss_box=dict(type='L1Loss', loss_weight=0.25),\n loss_centerness=dict(type='CrossEntropyLoss', use_sigmoid=True),\n loss_yawness=dict(type='GaussianFocalLoss'),\n cls_allow_reverse=[5]),\n decoder=dict(type='SparseBox3DDecoder'),\n reg_weights=[2.0, 2.0, 2.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0]))\ndataset_type = 'NuScenes3DDetTrackDataset'\ndata_root = 'data/nuscenes/'\nanno_root = 'data/nuscenes_anno_pkls/'\nfile_client_args = dict(backend='disk')\nimg_norm_cfg = dict(\n mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375], to_rgb=True)\ntrain_pipeline = [\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(\n type='LoadPointsFromFile',\n coord_type='LIDAR',\n load_dim=5,\n use_dim=5,\n file_client_args=dict(backend='disk')),\n dict(type='ResizeCropFlipImage'),\n dict(type='MultiScaleDepthMapGenerator', downsample=[4, 8, 16]),\n dict(type='BBoxRotation'),\n dict(type='PhotoMetricDistortionMultiViewImage'),\n dict(\n type='NormalizeMultiviewImage',\n mean=[123.675, 116.28, 103.53],\n std=[58.395, 57.12, 57.375],\n to_rgb=True),\n dict(\n type='CircleObjectRangeFilter',\n class_dist_thred=[55, 55, 55, 55, 55, 55, 55, 55, 55, 55]),\n dict(\n type='InstanceNameFilter',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ]),\n dict(type='NuScenesSparse4DAdaptor'),\n dict(\n type='Collect',\n keys=[\n 'img', 'timestamp', 'projection_mat', 'image_wh', 'gt_depth',\n 'focal', 'gt_bboxes_3d', 'gt_labels_3d'\n ],\n meta_keys=['T_global', 'T_global_inv', 'timestamp', 'instance_id'])\n]\ntest_pipeline = [\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(type='ResizeCropFlipImage'),\n dict(\n type='NormalizeMultiviewImage',\n mean=[123.675, 116.28, 103.53],\n std=[58.395, 57.12, 57.375],\n to_rgb=True),\n dict(type='NuScenesSparse4DAdaptor'),\n dict(\n type='Collect',\n keys=['img', 'timestamp', 'projection_mat', 'image_wh'],\n meta_keys=['T_global', 'T_global_inv', 'timestamp'])\n]\ninput_modality = dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False)\ndata_basic_config = dict(\n type='NuScenes3DDetTrackDataset',\n data_root='data/nuscenes/',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier',\n 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ],\n modality=dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False),\n version='v1.0-trainval')\ndata_aug_conf = dict(\n resize_lim=(0.4, 0.47),\n final_dim=(256, 704),\n bot_pct_lim=(0.0, 0.0),\n rot_lim=(-5.4, 5.4),\n H=900,\n W=1600,\n rand_flip=True,\n rot3d_range=[-0.3925, 0.3925])\ndata = dict(\n samples_per_gpu=20,\n workers_per_gpu=20,\n train=dict(\n type='NuScenes3DDetTrackDataset',\n data_root='data/nuscenes/',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ],\n modality=dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False),\n version='v1.0-trainval',\n ann_file='data/nuscenes_anno_pkls/nuscenes_infos_train.pkl',\n pipeline=[\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(\n type='LoadPointsFromFile',\n coord_type='LIDAR',\n load_dim=5,\n use_dim=5,\n file_client_args=dict(backend='disk')),\n dict(type='ResizeCropFlipImage'),\n dict(type='MultiScaleDepthMapGenerator', downsample=[4, 8, 16]),\n dict(type='BBoxRotation'),\n dict(type='PhotoMetricDistortionMultiViewImage'),\n dict(\n type='NormalizeMultiviewImage',\n mean=[123.675, 116.28, 103.53],\n std=[58.395, 57.12, 57.375],\n to_rgb=True),\n dict(\n type='CircleObjectRangeFilter',\n class_dist_thred=[55, 55, 55, 55, 55, 55, 55, 55, 55, 55]),\n dict(\n type='InstanceNameFilter',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian',\n 'traffic_cone'\n ]),\n dict(type='NuScenesSparse4DAdaptor'),\n dict(\n type='Collect',\n keys=[\n 'img', 'timestamp', 'projection_mat', 'image_wh',\n 'gt_depth', 'focal', 'gt_bboxes_3d', 'gt_labels_3d'\n ],\n meta_keys=[\n 'T_global', 'T_global_inv', 'timestamp', 'instance_id'\n ])\n ],\n test_mode=False,\n data_aug_conf=dict(\n resize_lim=(0.4, 0.47),\n final_dim=(256, 704),\n bot_pct_lim=(0.0, 0.0),\n rot_lim=(-5.4, 5.4),\n H=900,\n W=1600,\n rand_flip=True,\n rot3d_range=[-0.3925, 0.3925]),\n with_seq_flag=True,\n sequences_split_num=2,\n keep_consistent_seq_aug=True),\n val=dict(\n type='NuScenes3DDetTrackDataset',\n data_root='data/nuscenes/',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ],\n modality=dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False),\n version='v1.0-trainval',\n ann_file='data/nuscenes_anno_pkls/nuscenes_infos_val.pkl',\n pipeline=[\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(type='ResizeCropFlipImage'),\n dict(\n type='NormalizeMultiviewImage',\n mean=[123.675, 116.28, 103.53],\n std=[58.395, 57.12, 57.375],\n to_rgb=True),\n dict(type='NuScenesSparse4DAdaptor'),\n dict(\n type='Collect',\n keys=['img', 'timestamp', 'projection_mat', 'image_wh'],\n meta_keys=['T_global', 'T_global_inv', 'timestamp'])\n ],\n data_aug_conf=dict(\n resize_lim=(0.4, 0.47),\n final_dim=(256, 704),\n bot_pct_lim=(0.0, 0.0),\n rot_lim=(-5.4, 5.4),\n H=900,\n W=1600,\n rand_flip=True,\n rot3d_range=[-0.3925, 0.3925]),\n test_mode=True,\n tracking=True,\n tracking_threshold=0.2),\n test=dict(\n type='NuScenes3DDetTrackDataset',\n data_root='data/nuscenes/',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ],\n modality=dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False),\n version='v1.0-trainval',\n ann_file='data/nuscenes_anno_pkls/nuscenes_infos_val.pkl',\n pipeline=[\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(type='ResizeCropFlipImage'),\n dict(\n type='NormalizeMultiviewImage',\n mean=[123.675, 116.28, 103.53],\n std=[58.395, 57.12, 57.375],\n to_rgb=True),\n dict(type='NuScenesSparse4DAdaptor'),\n dict(\n type='Collect',\n keys=['img', 'timestamp', 'projection_mat', 'image_wh'],\n meta_keys=['T_global', 'T_global_inv', 'timestamp'])\n ],\n data_aug_conf=dict(\n resize_lim=(0.4, 0.47),\n final_dim=(256, 704),\n bot_pct_lim=(0.0, 0.0),\n rot_lim=(-5.4, 5.4),\n H=900,\n W=1600,\n rand_flip=True,\n rot3d_range=[-0.3925, 0.3925]),\n test_mode=True,\n tracking=True,\n tracking_threshold=0.2))\noptimizer = dict(\n type='AdamW',\n lr=0.0006,\n weight_decay=0.001,\n paramwise_cfg=dict(custom_keys=dict(img_backbone=dict(lr_mult=0.5))))\noptimizer_config = dict(grad_clip=dict(max_norm=25, norm_type=2))\nlr_config = dict(\n policy='CosineAnnealing',\n warmup='linear',\n warmup_iters=500,\n warmup_ratio=0.3333333333333333,\n min_lr_ratio=0.001)\nrunner = dict(type='IterBasedRunner', max_iters=17500)\nvis_pipeline = [\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(type='Collect', keys=['img'], meta_keys=['timestamp', 'lidar2img'])\n]\nevaluation = dict(\n interval=3500,\n pipeline=[\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(\n type='Collect', keys=['img'], meta_keys=['timestamp', 'lidar2img'])\n ])\ngpu_ids = range(0, 8)\n", "seed": 0, "exp_name": "sparse4dv3_temporal_r50_1x8_bs6_256x704.py"} diff --git a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20260210_162320.log b/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20260210_162320.log deleted file mode 100644 index 27ccf442d94d90d45c4b250685e6d4e40af4e0f6..0000000000000000000000000000000000000000 --- a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20260210_162320.log +++ /dev/null @@ -1,3243 +0,0 @@ -2026-02-10 16:23:20,889 - mmdet - INFO - Environment info: ------------------------------------------------------------- -sys.platform: linux -Python: 3.10.12 (main, Aug 15 2025, 14:32:43) [GCC 11.4.0] -CUDA available: True -GPU 0,1,2,3,4,5,6,7: BW200_LC, UBB BW1000_LC -CUDA_HOME: /opt/dtk -NVCC: Not Available -GCC: x86_64-linux-gnu-gcc (Ubuntu 11.4.0-1ubuntu1~22.04) 11.4.0 -PyTorch: 2.5.1 -PyTorch compiling details: PyTorch built with: - - GCC 10.3 - - C++ Version: 201703 - - Intel(R) Math Kernel Library Version 2020.0.4 Product Build 20200917 for Intel(R) 64 architecture applications - - OpenMP 201511 (a.k.a. OpenMP 4.5) - - LAPACK is enabled (usually provided by MKL) - - NNPACK is enabled - - CPU capability usage: AVX2 - - HIP Runtime 6.3.25521 - - MIOpen 2.18.0 - - Magma 2.8.0 - - Build settings: BLAS_INFO=mkl, BUILD_TYPE=Release, CXX_COMPILER=/opt/rh/gcc-toolset-10/root/usr/bin/c++, CXX_FLAGS= -D_GLIBCXX_USE_CXX11_ABI=1 -fvisibility-inlines-hidden -DUSE_PTHREADPOOL -DNDEBUG -DUSE_KINETO -DLIBKINETO_NOCUPTI -DLIBKINETO_NOXPUPTI=ON -DUSE_FBGEMM -DUSE_PYTORCH_QNNPACK -DUSE_XNNPACK -DSYMBOLICATE_MOBILE_DEBUG_HANDLE -O2 -fPIC -Wall -Wextra -Werror=return-type -Werror=non-virtual-dtor -Werror=bool-operation -Wnarrowing -Wno-missing-field-initializers -Wno-type-limits -Wno-array-bounds -Wno-unknown-pragmas -Wno-unused-parameter -Wno-strict-overflow -Wno-strict-aliasing -Wno-stringop-overflow -Wsuggest-override -Wno-psabi -Wno-error=old-style-cast -Wno-missing-braces -fdiagnostics-color=always -faligned-new -Wno-unused-but-set-variable -Wno-maybe-uninitialized -fno-math-errno -fno-trapping-math -Werror=format -Wno-stringop-overflow, FORCE_FALLBACK_CUDA_MPI=1, LAPACK_INFO=mkl, PERF_WITH_AVX=1, PERF_WITH_AVX2=1, TORCH_VERSION=2.5.1, USE_CUDA=0, USE_CUDNN=OFF, USE_CUSPARSELT=OFF, USE_EXCEPTION_PTR=1, USE_GFLAGS=1, USE_GLOG=1, USE_GLOO=1, USE_MKL=ON, USE_MKLDNN=0, USE_MPI=1, USE_NCCL=1, USE_NNPACK=ON, USE_OPENMP=1, USE_ROCM=ON, USE_ROCM_KERNEL_ASSERT=OFF, - -TorchVision: 0.20.1 -OpenCV: 4.12.0 -MMCV: 1.6.1 -MMCV Compiler: GCC 10.3 -MMCV CUDA Compiler: rocm not available -MMDetection: 2.25.1+ ------------------------------------------------------------- - -2026-02-10 16:23:21,889 - mmdet - INFO - Distributed training: False -2026-02-10 16:23:22,865 - mmdet - INFO - Config: -plugin = True -plugin_dir = 'projects/mmdet3d_plugin/' -dist_params = dict(backend='nccl') -log_level = 'INFO' -work_dir = './work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704' -total_batch_size = 160 -num_gpus = 8 -batch_size = 20 -num_iters_per_epoch = 175 -num_epochs = 100 -checkpoint_epoch_interval = 20 -checkpoint_config = dict(interval=3500) -log_config = dict( - interval=1, - hooks=[ - dict(type='TextLoggerHook', by_epoch=False), - dict(type='TensorboardLoggerHook') - ]) -load_from = None -resume_from = None -workflow = [('train', 1)] -fp16 = dict(loss_scale=32.0) -input_shape = (704, 256) -tracking_test = True -tracking_threshold = 0.2 -class_names = [ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier', - 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' -] -num_classes = 10 -embed_dims = 256 -num_groups = 8 -num_decoder = 6 -num_single_frame_decoder = 1 -use_deformable_func = True -strides = [4, 8, 16, 32] -num_levels = 4 -num_depth_layers = 3 -drop_out = 0.1 -temporal = True -decouple_attn = True -with_quality_estimation = True -model = dict( - type='Sparse4D', - use_grid_mask=True, - use_deformable_func=True, - img_backbone=dict( - type='ResNet', - depth=50, - num_stages=4, - frozen_stages=-1, - norm_eval=False, - style='pytorch', - with_cp=True, - out_indices=(0, 1, 2, 3), - norm_cfg=dict(type='BN', requires_grad=True), - pretrained='ckpt/resnet50-19c8e357.pth'), - img_neck=dict( - type='FPN', - num_outs=4, - start_level=0, - out_channels=256, - add_extra_convs='on_output', - relu_before_extra_convs=True, - in_channels=[256, 512, 1024, 2048]), - depth_branch=dict( - type='DenseDepthNet', - embed_dims=256, - num_depth_layers=3, - loss_weight=0.2), - head=dict( - type='Sparse4DHead', - cls_threshold_to_reg=0.05, - decouple_attn=True, - instance_bank=dict( - type='InstanceBank', - num_anchor=900, - embed_dims=256, - anchor='nuscenes_kmeans900.npy', - anchor_handler=dict(type='SparseBox3DKeyPointsGenerator'), - num_temp_instances=600, - confidence_decay=0.6, - feat_grad=False), - anchor_encoder=dict( - type='SparseBox3DEncoder', - vel_dims=3, - embed_dims=[128, 32, 32, 64], - mode='cat', - output_fc=False, - in_loops=1, - out_loops=4), - num_single_frame_decoder=1, - operation_order=[ - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine' - ], - temp_graph_model=dict( - type='MultiheadAttention', - embed_dims=512, - num_heads=8, - batch_first=True, - dropout=0.1), - graph_model=dict( - type='MultiheadAttention', - embed_dims=512, - num_heads=8, - batch_first=True, - dropout=0.1), - norm_layer=dict(type='LN', normalized_shape=256), - ffn=dict( - type='AsymmetricFFN', - in_channels=512, - pre_norm=dict(type='LN'), - embed_dims=256, - feedforward_channels=1024, - num_fcs=2, - ffn_drop=0.1, - act_cfg=dict(type='ReLU', inplace=True)), - deformable_model=dict( - type='DeformableFeatureAggregation', - embed_dims=256, - num_groups=8, - num_levels=4, - num_cams=6, - attn_drop=0.15, - use_deformable_func=True, - use_camera_embed=True, - residual_mode='cat', - kps_generator=dict( - type='SparseBox3DKeyPointsGenerator', - num_learnable_pts=6, - fix_scale=[[0, 0, 0], [0.45, 0, 0], [-0.45, 0, 0], - [0, 0.45, 0], [0, -0.45, 0], [0, 0, 0.45], - [0, 0, -0.45]])), - refine_layer=dict( - type='SparseBox3DRefinementModule', - embed_dims=256, - num_cls=10, - refine_yaw=True, - with_quality_estimation=True), - sampler=dict( - type='SparseBox3DTarget', - num_dn_groups=5, - num_temp_dn_groups=3, - dn_noise_scale=[2.0, 2.0, 2.0, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5], - max_dn_gt=32, - add_neg_dn=True, - cls_weight=2.0, - box_weight=0.25, - reg_weights=[2.0, 2.0, 2.0, 0.5, 0.5, 0.5, 0.0, 0.0, 0.0, 0.0], - cls_wise_reg_weights=dict( - {9: [2.0, 2.0, 2.0, 1.0, 1.0, 1.0, 0.0, 0.0, 1.0, 1.0]})), - loss_cls=dict( - type='FocalLoss', - use_sigmoid=True, - gamma=2.0, - alpha=0.25, - loss_weight=2.0), - loss_reg=dict( - type='SparseBox3DLoss', - loss_box=dict(type='L1Loss', loss_weight=0.25), - loss_centerness=dict(type='CrossEntropyLoss', use_sigmoid=True), - loss_yawness=dict(type='GaussianFocalLoss'), - cls_allow_reverse=[5]), - decoder=dict(type='SparseBox3DDecoder'), - reg_weights=[2.0, 2.0, 2.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0])) -dataset_type = 'NuScenes3DDetTrackDataset' -data_root = 'data/nuscenes/' -anno_root = 'data/nuscenes_anno_pkls/' -file_client_args = dict(backend='disk') -img_norm_cfg = dict( - mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375], to_rgb=True) -train_pipeline = [ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict( - type='LoadPointsFromFile', - coord_type='LIDAR', - load_dim=5, - use_dim=5, - file_client_args=dict(backend='disk')), - dict(type='ResizeCropFlipImage'), - dict(type='MultiScaleDepthMapGenerator', downsample=[4, 8, 16]), - dict(type='BBoxRotation'), - dict(type='PhotoMetricDistortionMultiViewImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict( - type='CircleObjectRangeFilter', - class_dist_thred=[55, 55, 55, 55, 55, 55, 55, 55, 55, 55]), - dict( - type='InstanceNameFilter', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ]), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=[ - 'img', 'timestamp', 'projection_mat', 'image_wh', 'gt_depth', - 'focal', 'gt_bboxes_3d', 'gt_labels_3d' - ], - meta_keys=['T_global', 'T_global_inv', 'timestamp', 'instance_id']) -] -test_pipeline = [ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='ResizeCropFlipImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=['img', 'timestamp', 'projection_mat', 'image_wh'], - meta_keys=['T_global', 'T_global_inv', 'timestamp']) -] -input_modality = dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False) -data_basic_config = dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier', - 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval') -data_aug_conf = dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]) -data = dict( - samples_per_gpu=20, - workers_per_gpu=20, - train=dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval', - ann_file='data/nuscenes_anno_pkls/nuscenes_infos_train.pkl', - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict( - type='LoadPointsFromFile', - coord_type='LIDAR', - load_dim=5, - use_dim=5, - file_client_args=dict(backend='disk')), - dict(type='ResizeCropFlipImage'), - dict(type='MultiScaleDepthMapGenerator', downsample=[4, 8, 16]), - dict(type='BBoxRotation'), - dict(type='PhotoMetricDistortionMultiViewImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict( - type='CircleObjectRangeFilter', - class_dist_thred=[55, 55, 55, 55, 55, 55, 55, 55, 55, 55]), - dict( - type='InstanceNameFilter', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', - 'traffic_cone' - ]), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=[ - 'img', 'timestamp', 'projection_mat', 'image_wh', - 'gt_depth', 'focal', 'gt_bboxes_3d', 'gt_labels_3d' - ], - meta_keys=[ - 'T_global', 'T_global_inv', 'timestamp', 'instance_id' - ]) - ], - test_mode=False, - data_aug_conf=dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]), - with_seq_flag=True, - sequences_split_num=2, - keep_consistent_seq_aug=True), - val=dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval', - ann_file='data/nuscenes_anno_pkls/nuscenes_infos_val.pkl', - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='ResizeCropFlipImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=['img', 'timestamp', 'projection_mat', 'image_wh'], - meta_keys=['T_global', 'T_global_inv', 'timestamp']) - ], - data_aug_conf=dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]), - test_mode=True, - tracking=True, - tracking_threshold=0.2), - test=dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval', - ann_file='data/nuscenes_anno_pkls/nuscenes_infos_val.pkl', - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='ResizeCropFlipImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=['img', 'timestamp', 'projection_mat', 'image_wh'], - meta_keys=['T_global', 'T_global_inv', 'timestamp']) - ], - data_aug_conf=dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]), - test_mode=True, - tracking=True, - tracking_threshold=0.2)) -optimizer = dict( - type='AdamW', - lr=0.0006, - weight_decay=0.001, - paramwise_cfg=dict(custom_keys=dict(img_backbone=dict(lr_mult=0.5)))) -optimizer_config = dict(grad_clip=dict(max_norm=25, norm_type=2)) -lr_config = dict( - policy='CosineAnnealing', - warmup='linear', - warmup_iters=500, - warmup_ratio=0.3333333333333333, - min_lr_ratio=0.001) -runner = dict(type='IterBasedRunner', max_iters=17500) -vis_pipeline = [ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='Collect', keys=['img'], meta_keys=['timestamp', 'lidar2img']) -] -evaluation = dict( - interval=3500, - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict( - type='Collect', keys=['img'], meta_keys=['timestamp', 'lidar2img']) - ]) -gpu_ids = range(0, 1) - -2026-02-10 16:23:22,866 - mmdet - INFO - Set random seed to 0, deterministic: False -2026-02-10 16:23:23,424 - mmdet - INFO - initialize ResNet with init_cfg {'type': 'Pretrained', 'checkpoint': 'ckpt/resnet50-19c8e357.pth'} -2026-02-10 16:23:23,648 - mmdet - INFO - initialize FPN with init_cfg {'type': 'Xavier', 'layer': 'Conv2d', 'distribution': 'uniform'} -Name of parameter - Initialization information - -img_backbone.conv1.weight - torch.Size([64, 3, 7, 7]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.bn1.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.bn1.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.conv1.weight - torch.Size([64, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn1.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn1.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.conv2.weight - torch.Size([64, 64, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn2.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn2.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.conv3.weight - torch.Size([256, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn3.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn3.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.downsample.0.weight - torch.Size([256, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.downsample.1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.downsample.1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.conv1.weight - torch.Size([64, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn1.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn1.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.conv2.weight - torch.Size([64, 64, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn2.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn2.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.conv3.weight - torch.Size([256, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn3.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn3.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.conv1.weight - torch.Size([64, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn1.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn1.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.conv2.weight - torch.Size([64, 64, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn2.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn2.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.conv3.weight - torch.Size([256, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn3.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn3.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.conv1.weight - torch.Size([128, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn1.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn1.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.conv2.weight - torch.Size([128, 128, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn2.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn2.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.conv3.weight - torch.Size([512, 128, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn3.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn3.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.downsample.0.weight - torch.Size([512, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.downsample.1.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.downsample.1.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.conv1.weight - torch.Size([128, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn1.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn1.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.conv2.weight - torch.Size([128, 128, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn2.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn2.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.conv3.weight - torch.Size([512, 128, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn3.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn3.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.conv1.weight - torch.Size([128, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn1.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn1.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.conv2.weight - torch.Size([128, 128, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn2.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn2.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.conv3.weight - torch.Size([512, 128, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn3.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn3.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.conv1.weight - torch.Size([128, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn1.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn1.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.conv2.weight - torch.Size([128, 128, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn2.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn2.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.conv3.weight - torch.Size([512, 128, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn3.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn3.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.conv1.weight - torch.Size([256, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.downsample.0.weight - torch.Size([1024, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.downsample.1.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.downsample.1.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.conv1.weight - torch.Size([512, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn1.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn1.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.conv2.weight - torch.Size([512, 512, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn2.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn2.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.conv3.weight - torch.Size([2048, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn3.weight - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn3.bias - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.downsample.0.weight - torch.Size([2048, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.downsample.1.weight - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.downsample.1.bias - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.conv1.weight - torch.Size([512, 2048, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn1.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn1.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.conv2.weight - torch.Size([512, 512, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn2.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn2.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.conv3.weight - torch.Size([2048, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn3.weight - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn3.bias - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.conv1.weight - torch.Size([512, 2048, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn1.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn1.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.conv2.weight - torch.Size([512, 512, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn2.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn2.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.conv3.weight - torch.Size([2048, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn3.weight - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn3.bias - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_neck.lateral_convs.0.conv.weight - torch.Size([256, 256, 1, 1]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.lateral_convs.0.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.lateral_convs.1.conv.weight - torch.Size([256, 512, 1, 1]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.lateral_convs.1.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.lateral_convs.2.conv.weight - torch.Size([256, 1024, 1, 1]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.lateral_convs.2.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.lateral_convs.3.conv.weight - torch.Size([256, 2048, 1, 1]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.lateral_convs.3.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.fpn_convs.0.conv.weight - torch.Size([256, 256, 3, 3]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.fpn_convs.0.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.fpn_convs.1.conv.weight - torch.Size([256, 256, 3, 3]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.fpn_convs.1.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.fpn_convs.2.conv.weight - torch.Size([256, 256, 3, 3]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.fpn_convs.2.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.fpn_convs.3.conv.weight - torch.Size([256, 256, 3, 3]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.fpn_convs.3.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.instance_bank.anchor - torch.Size([900, 11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.instance_bank.instance_feature - torch.Size([900, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.instance_bank.anchor_handler.fix_scale - torch.Size([1, 3]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.0.weight - torch.Size([128, 3]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.0.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.2.weight - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.2.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.3.weight - torch.Size([128, 128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.3.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.5.weight - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.5.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.6.weight - torch.Size([128, 128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.6.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.8.weight - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.8.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.9.weight - torch.Size([128, 128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.9.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.11.weight - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.11.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.0.weight - torch.Size([32, 3]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.0.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.2.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.2.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.3.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.3.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.5.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.5.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.6.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.6.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.8.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.8.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.9.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.9.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.11.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.11.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.0.weight - torch.Size([32, 2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.0.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.2.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.2.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.3.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.3.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.5.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.5.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.6.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.6.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.8.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.8.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.9.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.9.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.11.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.11.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.0.weight - torch.Size([64, 3]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.0.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.2.weight - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.2.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.3.weight - torch.Size([64, 64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.3.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.5.weight - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.5.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.6.weight - torch.Size([64, 64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.6.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.8.weight - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.8.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.9.weight - torch.Size([64, 64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.9.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.11.weight - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.11.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.1.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.1.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.1.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.1.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.1.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.1.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.1.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.1.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.3.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.4.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.4.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.4.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.4.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.5.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.5.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.5.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.5.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.6.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.6.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.8.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.8.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.8.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.8.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.8.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.8.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.8.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.8.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.10.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.11.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.11.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.11.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.11.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.12.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.12.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.12.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.12.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.13.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.13.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.15.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.15.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.15.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.15.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.15.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.15.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.15.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.15.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.16.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.16.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.17.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.18.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.18.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.18.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.18.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.19.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.19.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.19.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.19.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.20.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.20.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.22.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.22.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.22.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.22.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.22.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.22.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.22.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.22.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.23.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.23.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.24.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.25.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.25.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.25.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.25.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.26.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.26.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.26.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.26.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.27.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.27.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.29.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.29.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.29.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.29.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.29.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.29.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.29.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.29.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.30.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.30.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.31.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.32.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.32.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.32.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.32.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.33.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.33.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.33.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.33.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.34.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.34.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.36.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.36.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.36.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.36.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.36.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.36.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.36.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.36.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.37.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.37.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.38.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.fc_before.weight - torch.Size([512, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.fc_after.weight - torch.Size([256, 512]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.0.weight - torch.Size([1, 256, 1, 1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.0.bias - torch.Size([1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.1.weight - torch.Size([1, 256, 1, 1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.1.bias - torch.Size([1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.2.weight - torch.Size([1, 256, 1, 1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.2.bias - torch.Size([1]): -The value is the same before and after calling `init_weights` of Sparse4D -2026-02-10 16:23:24,023 - mmdet - INFO - Model: -Sparse4D( - (img_backbone): ResNet( - (conv1): Conv2d(3, 64, kernel_size=(7, 7), stride=(2, 2), padding=(3, 3), bias=False) - (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (maxpool): MaxPool2d(kernel_size=3, stride=2, padding=1, dilation=1, ceil_mode=False) - (layer1): ResLayer( - (0): Bottleneck( - (conv1): Conv2d(64, 64, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (downsample): Sequential( - (0): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - ) - ) - (1): Bottleneck( - (conv1): Conv2d(256, 64, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (2): Bottleneck( - (conv1): Conv2d(256, 64, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - ) - (layer2): ResLayer( - (0): Bottleneck( - (conv1): Conv2d(256, 128, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (downsample): Sequential( - (0): Conv2d(256, 512, kernel_size=(1, 1), stride=(2, 2), bias=False) - (1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - ) - ) - (1): Bottleneck( - (conv1): Conv2d(512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (2): Bottleneck( - (conv1): Conv2d(512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (3): Bottleneck( - (conv1): Conv2d(512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - ) - (layer3): ResLayer( - (0): Bottleneck( - (conv1): Conv2d(512, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (downsample): Sequential( - (0): Conv2d(512, 1024, kernel_size=(1, 1), stride=(2, 2), bias=False) - (1): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - ) - ) - (1): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (2): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (3): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (4): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (5): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - ) - (layer4): ResLayer( - (0): Bottleneck( - (conv1): Conv2d(1024, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(512, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (downsample): Sequential( - (0): Conv2d(1024, 2048, kernel_size=(1, 1), stride=(2, 2), bias=False) - (1): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - ) - ) - (1): Bottleneck( - (conv1): Conv2d(2048, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(512, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (2): Bottleneck( - (conv1): Conv2d(2048, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(512, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - ) - ) - init_cfg={'type': 'Pretrained', 'checkpoint': 'ckpt/resnet50-19c8e357.pth'} - (img_neck): FPN( - (lateral_convs): ModuleList( - (0): ConvModule( - (conv): Conv2d(256, 256, kernel_size=(1, 1), stride=(1, 1)) - ) - (1): ConvModule( - (conv): Conv2d(512, 256, kernel_size=(1, 1), stride=(1, 1)) - ) - (2): ConvModule( - (conv): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1)) - ) - (3): ConvModule( - (conv): Conv2d(2048, 256, kernel_size=(1, 1), stride=(1, 1)) - ) - ) - (fpn_convs): ModuleList( - (0-3): 4 x ConvModule( - (conv): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1)) - ) - ) - ) - init_cfg={'type': 'Xavier', 'layer': 'Conv2d', 'distribution': 'uniform'} - (head): Sparse4DHead( - (instance_bank): InstanceBank( - (anchor_handler): SparseBox3DKeyPointsGenerator() - ) - (anchor_encoder): SparseBox3DEncoder( - (pos_fc): Sequential( - (0): Linear(in_features=3, out_features=128, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((128,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=128, out_features=128, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((128,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=128, out_features=128, bias=True) - (7): ReLU(inplace=True) - (8): LayerNorm((128,), eps=1e-05, elementwise_affine=True) - (9): Linear(in_features=128, out_features=128, bias=True) - (10): ReLU(inplace=True) - (11): LayerNorm((128,), eps=1e-05, elementwise_affine=True) - ) - (size_fc): Sequential( - (0): Linear(in_features=3, out_features=32, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=32, out_features=32, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=32, out_features=32, bias=True) - (7): ReLU(inplace=True) - (8): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (9): Linear(in_features=32, out_features=32, bias=True) - (10): ReLU(inplace=True) - (11): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - ) - (yaw_fc): Sequential( - (0): Linear(in_features=2, out_features=32, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=32, out_features=32, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=32, out_features=32, bias=True) - (7): ReLU(inplace=True) - (8): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (9): Linear(in_features=32, out_features=32, bias=True) - (10): ReLU(inplace=True) - (11): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - ) - (vel_fc): Sequential( - (0): Linear(in_features=3, out_features=64, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((64,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=64, out_features=64, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((64,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=64, out_features=64, bias=True) - (7): ReLU(inplace=True) - (8): LayerNorm((64,), eps=1e-05, elementwise_affine=True) - (9): Linear(in_features=64, out_features=64, bias=True) - (10): ReLU(inplace=True) - (11): LayerNorm((64,), eps=1e-05, elementwise_affine=True) - ) - ) - (loss_cls): FocalLoss() - (loss_reg): SparseBox3DLoss( - (loss_box): L1Loss() - (loss_cns): CrossEntropyLoss(avg_non_ignore=False) - (loss_yns): GaussianFocalLoss() - ) - (layers): ModuleList( - (0): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (1): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (4-5): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (6): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (7): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (8): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (11-12): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (13): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (14): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (15): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (16): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (17): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (18-19): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (20): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (21): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (22): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (23): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (24): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (25-26): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (27): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (28): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (29): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (30): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (31): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (32-33): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (34): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (35): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (36): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (37): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (38): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - ) - (fc_before): Linear(in_features=256, out_features=512, bias=False) - (fc_after): Linear(in_features=512, out_features=256, bias=False) - ) - (depth_branch): DenseDepthNet( - (depth_layers): ModuleList( - (0-2): 3 x Conv2d(256, 1, kernel_size=(1, 1), stride=(1, 1)) - ) - ) - (grid_mask): GridMask() -) -2026-02-10 16:23:38,729 - mmdet - INFO - Start running, host: root@bw28, work_dir: /workspace/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704 -2026-02-10 16:23:38,729 - mmdet - INFO - Hooks will be executed in the following order: -before_run: -(VERY_HIGH ) CosineAnnealingLrUpdaterHook -(ABOVE_NORMAL) Fp16OptimizerHook -(NORMAL ) CheckpointHook -(NORMAL ) EvalHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -before_train_epoch: -(VERY_HIGH ) CosineAnnealingLrUpdaterHook -(NORMAL ) EvalHook -(LOW ) IterTimerHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -before_train_iter: -(VERY_HIGH ) CosineAnnealingLrUpdaterHook -(NORMAL ) EvalHook -(LOW ) IterTimerHook - -------------------- -after_train_iter: -(ABOVE_NORMAL) Fp16OptimizerHook -(NORMAL ) CheckpointHook -(NORMAL ) EvalHook -(NORMAL ) ProfilerHook -(LOW ) IterTimerHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -after_train_epoch: -(NORMAL ) CheckpointHook -(NORMAL ) EvalHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -before_val_epoch: -(LOW ) IterTimerHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -before_val_iter: -(LOW ) IterTimerHook - -------------------- -after_val_iter: -(LOW ) IterTimerHook - -------------------- -after_val_epoch: -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -after_run: -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -2026-02-10 16:23:38,729 - mmdet - INFO - workflow: [('train', 1)], max: 17500 iters -2026-02-10 16:23:38,732 - mmdet - INFO - Checkpoints will be saved to /workspace/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704 by HardDiskBackend. -2026-02-10 16:24:08,957 - mmdet - INFO - Iter [1/17500] lr: 1.000e-04, eta: 5 days, 21:57:22, time: 29.204, data_time: 8.286, memory: 48982, loss_cls_0: 2.3562, loss_box_0: 0.0000, loss_cns_0: 0.0000, loss_yns_0: 0.0000, loss_cls_1: 2.1552, loss_box_1: 0.1646, loss_cns_1: 0.0317, loss_yns_1: 0.0078, loss_cls_2: 2.4044, loss_box_2: 0.0000, loss_cns_2: 0.0000, loss_yns_2: 0.0000, loss_cls_3: 2.4750, loss_box_3: 0.0277, loss_cns_3: 0.0050, loss_yns_3: 0.0026, loss_cls_4: 1.9946, loss_box_4: 0.4235, loss_cns_4: 0.0563, loss_yns_4: 0.0272, loss_cls_5: 2.3639, loss_box_5: 0.0000, loss_cns_5: 0.0000, loss_yns_5: 0.0000, loss_cls_dn_0: 1.1947, loss_box_dn_0: 1.4557, loss_cls_dn_1: 1.1237, loss_box_dn_1: 1.7309, loss_cls_dn_2: 1.1991, loss_box_dn_2: 1.9799, loss_cls_dn_3: 1.1733, loss_box_dn_3: 2.2453, loss_cls_dn_4: 1.0500, loss_box_dn_4: 2.4168, loss_cls_dn_5: 1.2066, loss_box_dn_5: 2.6647, loss_dense_depth: 1.8661, loss: 35.8026, grad_norm: 371.3917 -2026-02-10 16:24:11,936 - mmdet - INFO - Iter [2/17500] lr: 1.004e-04, eta: 3 days, 6:12:54, time: 2.979, data_time: 0.059, memory: 48982, loss_cls_0: 2.1086, loss_box_0: 0.0000, loss_cns_0: 0.0000, loss_yns_0: 0.0000, loss_cls_1: 2.0818, loss_box_1: 0.1080, loss_cns_1: 0.0201, loss_yns_1: 0.0067, loss_cls_2: 2.2031, loss_box_2: 0.0502, loss_cns_2: 0.0055, loss_yns_2: 0.0017, loss_cls_3: 2.0120, loss_box_3: 0.1965, loss_cns_3: 0.0272, loss_yns_3: 0.0071, loss_cls_4: 1.8525, loss_box_4: 0.8447, loss_cns_4: 0.0930, loss_yns_4: 0.0338, loss_cls_5: 1.9991, loss_box_5: 0.9981, loss_cns_5: 0.1234, loss_yns_5: 0.0537, loss_cls_dn_0: 1.0740, loss_box_dn_0: 1.3617, loss_cls_dn_1: 1.0214, loss_box_dn_1: 2.4690, loss_cls_dn_2: 1.0327, loss_box_dn_2: 2.5740, loss_cls_dn_3: 0.9651, loss_box_dn_3: 2.6374, loss_cls_dn_4: 0.8941, loss_box_dn_4: 2.8635, loss_cls_dn_5: 0.9958, loss_box_dn_5: 3.1175, loss_dense_depth: 1.7784, loss: 37.6114, grad_norm: 80.3595 -2026-02-10 16:24:13,827 - mmdet - INFO - Iter [3/17500] lr: 1.008e-04, eta: 2 days, 7:12:11, time: 1.890, data_time: 0.061, memory: 48982, loss_cls_0: 1.4741, loss_box_0: 2.3963, loss_cns_0: 0.8099, loss_yns_0: 0.2458, loss_cls_1: 1.8236, loss_box_1: 0.3521, loss_cns_1: 0.0525, loss_yns_1: 0.0194, loss_cls_2: 1.8759, loss_box_2: 1.4691, loss_cns_2: 0.1836, loss_yns_2: 0.0796, loss_cls_3: 1.6312, loss_box_3: 3.9890, loss_cns_3: 0.4602, loss_yns_3: 0.2045, loss_cls_4: 1.5899, loss_box_4: 3.5747, loss_cns_4: 0.3419, loss_yns_4: 0.1526, loss_cls_5: 1.6728, loss_box_5: 3.6146, loss_cns_5: 0.3143, loss_yns_5: 0.1263, loss_cls_dn_0: 0.7514, loss_box_dn_0: 1.1602, loss_cls_dn_1: 0.8873, loss_box_dn_1: 2.4191, loss_cls_dn_2: 0.8678, loss_box_dn_2: 2.6708, loss_cls_dn_3: 0.7502, loss_box_dn_3: 2.8696, loss_cls_dn_4: 0.7562, loss_box_dn_4: 3.1068, loss_cls_dn_5: 0.8192, loss_box_dn_5: 3.3473, loss_dense_depth: 1.7425, loss: 50.6022, grad_norm: 120.9407 -2026-02-10 16:24:15,753 - mmdet - INFO - Iter [4/17500] lr: 1.012e-04, eta: 1 day, 19:44:25, time: 1.926, data_time: 0.057, memory: 48982, loss_cls_0: 1.3256, loss_box_0: 2.7074, loss_cns_0: 0.6568, loss_yns_0: 0.2231, loss_cls_1: 1.6315, loss_box_1: 1.8735, loss_cns_1: 0.2549, loss_yns_1: 0.1515, loss_cls_2: 1.6787, loss_box_2: 3.4923, loss_cns_2: 0.3997, loss_yns_2: 0.1754, loss_cls_3: 1.4509, loss_box_3: 4.7330, loss_cns_3: 0.4729, loss_yns_3: 0.2492, loss_cls_4: 1.4127, loss_box_4: 5.3446, loss_cns_4: 0.3340, loss_yns_4: 0.1943, loss_cls_5: 1.5105, loss_box_5: 5.3208, loss_cns_5: 0.4439, loss_yns_5: 0.1939, loss_cls_dn_0: 0.6367, loss_box_dn_0: 1.1799, loss_cls_dn_1: 0.7961, loss_box_dn_1: 2.9262, loss_cls_dn_2: 0.7573, loss_box_dn_2: 3.0538, loss_cls_dn_3: 0.6448, loss_box_dn_3: 3.3261, loss_cls_dn_4: 0.6516, loss_box_dn_4: 3.5712, loss_cls_dn_5: 0.7010, loss_box_dn_5: 3.7995, loss_dense_depth: 1.7113, loss: 59.9865, grad_norm: 147.5616 -2026-02-10 16:24:17,659 - mmdet - INFO - Iter [5/17500] lr: 1.016e-04, eta: 1 day, 12:50:33, time: 1.906, data_time: 0.062, memory: 48982, loss_cls_0: 1.2920, loss_box_0: 2.7951, loss_cns_0: 0.5304, loss_yns_0: 0.1840, loss_cls_1: 1.5237, loss_box_1: 3.3567, loss_cns_1: 0.3599, loss_yns_1: 0.1749, loss_cls_2: 1.5681, loss_box_2: 3.9760, loss_cns_2: 0.4093, loss_yns_2: 0.1915, loss_cls_3: 1.4263, loss_box_3: 4.3021, loss_cns_3: 0.4318, loss_yns_3: 0.2109, loss_cls_4: 1.3872, loss_box_4: 4.4946, loss_cns_4: 0.3666, loss_yns_4: 0.1979, loss_cls_5: 1.3847, loss_box_5: 4.8085, loss_cns_5: 0.4424, loss_yns_5: 0.2308, loss_cls_dn_0: 0.5577, loss_box_dn_0: 1.2501, loss_cls_dn_1: 0.7311, loss_box_dn_1: 2.6224, loss_cls_dn_2: 0.6998, loss_box_dn_2: 2.7394, loss_cls_dn_3: 0.5797, loss_box_dn_3: 2.9493, loss_cls_dn_4: 0.5880, loss_box_dn_4: 3.1282, loss_cls_dn_5: 0.6167, loss_box_dn_5: 3.2923, loss_dense_depth: 1.7004, loss: 57.5002, grad_norm: 143.9374 -2026-02-10 16:24:19,567 - mmdet - INFO - Iter [6/17500] lr: 1.020e-04, eta: 1 day, 8:14:45, time: 1.908, data_time: 0.060, memory: 48982, loss_cls_0: 1.3293, loss_box_0: 2.6043, loss_cns_0: 0.5307, loss_yns_0: 0.2162, loss_cls_1: 1.4462, loss_box_1: 3.3424, loss_cns_1: 0.3852, loss_yns_1: 0.1771, loss_cls_2: 1.4554, loss_box_2: 3.8996, loss_cns_2: 0.3995, loss_yns_2: 0.1895, loss_cls_3: 1.3933, loss_box_3: 4.0199, loss_cns_3: 0.3904, loss_yns_3: 0.2049, loss_cls_4: 1.3167, loss_box_4: 4.3894, loss_cns_4: 0.3197, loss_yns_4: 0.2112, loss_cls_5: 1.3418, loss_box_5: 4.6433, loss_cns_5: 0.3519, loss_yns_5: 0.2253, loss_cls_dn_0: 0.5042, loss_box_dn_0: 1.2122, loss_cls_dn_1: 0.6809, loss_box_dn_1: 2.8371, loss_cls_dn_2: 0.6182, loss_box_dn_2: 2.8610, loss_cls_dn_3: 0.5264, loss_box_dn_3: 2.9067, loss_cls_dn_4: 0.5211, loss_box_dn_4: 3.0714, loss_cls_dn_5: 0.5310, loss_box_dn_5: 3.1659, loss_dense_depth: 1.6649, loss: 55.8841, grad_norm: 124.1443 -2026-02-10 16:24:21,467 - mmdet - INFO - Iter [7/17500] lr: 1.024e-04, eta: 1 day, 4:57:23, time: 1.900, data_time: 0.053, memory: 48982, loss_cls_0: 1.2949, loss_box_0: 2.4497, loss_cns_0: 0.6728, loss_yns_0: 0.1980, loss_cls_1: 1.3142, loss_box_1: 3.5442, loss_cns_1: 0.4433, loss_yns_1: 0.2234, loss_cls_2: 1.3858, loss_box_2: 3.6590, loss_cns_2: 0.4360, loss_yns_2: 0.1933, loss_cls_3: 1.3071, loss_box_3: 3.5782, loss_cns_3: 0.4434, loss_yns_3: 0.1878, loss_cls_4: 1.3652, loss_box_4: 3.9445, loss_cns_4: 0.3718, loss_yns_4: 0.1966, loss_cls_5: 1.2681, loss_box_5: 4.1201, loss_cns_5: 0.3624, loss_yns_5: 0.2169, loss_cls_dn_0: 0.4955, loss_box_dn_0: 1.1637, loss_cls_dn_1: 0.5905, loss_box_dn_1: 2.8034, loss_cls_dn_2: 0.5515, loss_box_dn_2: 2.7992, loss_cls_dn_3: 0.4977, loss_box_dn_3: 2.7494, loss_cls_dn_4: 0.4674, loss_box_dn_4: 2.8962, loss_cls_dn_5: 0.4654, loss_box_dn_5: 2.9643, loss_dense_depth: 1.5775, loss: 53.1981, grad_norm: 115.3753 -2026-02-10 16:24:23,394 - mmdet - INFO - Iter [8/17500] lr: 1.028e-04, eta: 1 day, 2:30:20, time: 1.927, data_time: 0.056, memory: 48982, loss_cls_0: 1.2176, loss_box_0: 2.3947, loss_cns_0: 0.7324, loss_yns_0: 0.1826, loss_cls_1: 1.3070, loss_box_1: 3.6043, loss_cns_1: 0.4362, loss_yns_1: 0.2152, loss_cls_2: 1.4082, loss_box_2: 3.6203, loss_cns_2: 0.4344, loss_yns_2: 0.1877, loss_cls_3: 1.2654, loss_box_3: 3.4266, loss_cns_3: 0.4683, loss_yns_3: 0.1943, loss_cls_4: 1.3156, loss_box_4: 3.5008, loss_cns_4: 0.4865, loss_yns_4: 0.1949, loss_cls_5: 1.3093, loss_box_5: 3.5460, loss_cns_5: 0.5383, loss_yns_5: 0.1982, loss_cls_dn_0: 0.4998, loss_box_dn_0: 1.1170, loss_cls_dn_1: 0.5678, loss_box_dn_1: 1.7700, loss_cls_dn_2: 0.5390, loss_box_dn_2: 1.7942, loss_cls_dn_3: 0.5003, loss_box_dn_3: 1.7411, loss_cls_dn_4: 0.4635, loss_box_dn_4: 1.8677, loss_cls_dn_5: 0.4494, loss_box_dn_5: 1.9369, loss_dense_depth: 1.5408, loss: 46.9722, grad_norm: 95.8328 -2026-02-10 16:24:25,358 - mmdet - INFO - Iter [9/17500] lr: 1.032e-04, eta: 1 day, 0:37:10, time: 1.964, data_time: 0.068, memory: 48982, loss_cls_0: 1.1863, loss_box_0: 2.3541, loss_cns_0: 0.6830, loss_yns_0: 0.1872, loss_cls_1: 1.3307, loss_box_1: 3.5613, loss_cns_1: 0.4241, loss_yns_1: 0.1880, loss_cls_2: 1.3859, loss_box_2: 3.5440, loss_cns_2: 0.4010, loss_yns_2: 0.1947, loss_cls_3: 1.2383, loss_box_3: 3.5700, loss_cns_3: 0.3969, loss_yns_3: 0.2036, loss_cls_4: 1.2413, loss_box_4: 3.5115, loss_cns_4: 0.4161, loss_yns_4: 0.1778, loss_cls_5: 1.2995, loss_box_5: 3.5503, loss_cns_5: 0.4632, loss_yns_5: 0.2352, loss_cls_dn_0: 0.5093, loss_box_dn_0: 1.0798, loss_cls_dn_1: 0.4990, loss_box_dn_1: 1.6927, loss_cls_dn_2: 0.5110, loss_box_dn_2: 1.7309, loss_cls_dn_3: 0.5010, loss_box_dn_3: 1.8045, loss_cls_dn_4: 0.4660, loss_box_dn_4: 1.8245, loss_cls_dn_5: 0.4407, loss_box_dn_5: 1.9141, loss_dense_depth: 1.4885, loss: 46.2060, grad_norm: 80.6772 -2026-02-10 16:24:27,298 - mmdet - INFO - Iter [10/17500] lr: 1.036e-04, eta: 23:05:56, time: 1.940, data_time: 0.068, memory: 48982, loss_cls_0: 1.1589, loss_box_0: 2.3497, loss_cns_0: 0.6366, loss_yns_0: 0.1812, loss_cls_1: 1.2783, loss_box_1: 3.3871, loss_cns_1: 0.4275, loss_yns_1: 0.1888, loss_cls_2: 1.2589, loss_box_2: 3.3593, loss_cns_2: 0.4434, loss_yns_2: 0.1812, loss_cls_3: 1.2397, loss_box_3: 3.6077, loss_cns_3: 0.4500, loss_yns_3: 0.1961, loss_cls_4: 1.2181, loss_box_4: 3.4081, loss_cns_4: 0.4616, loss_yns_4: 0.1809, loss_cls_5: 1.2257, loss_box_5: 3.4697, loss_cns_5: 0.4980, loss_yns_5: 0.1894, loss_cls_dn_0: 0.4999, loss_box_dn_0: 1.0562, loss_cls_dn_1: 0.4625, loss_box_dn_1: 1.8615, loss_cls_dn_2: 0.5001, loss_box_dn_2: 1.9231, loss_cls_dn_3: 0.4872, loss_box_dn_3: 2.0545, loss_cls_dn_4: 0.4692, loss_box_dn_4: 2.0045, loss_cls_dn_5: 0.4317, loss_box_dn_5: 2.0821, loss_dense_depth: 1.4397, loss: 46.2681, grad_norm: 90.0191 -2026-02-10 16:24:29,222 - mmdet - INFO - Iter [11/17500] lr: 1.040e-04, eta: 21:50:50, time: 1.923, data_time: 0.082, memory: 48982, loss_cls_0: 1.1828, loss_box_0: 2.3536, loss_cns_0: 0.6029, loss_yns_0: 0.1728, loss_cls_1: 1.2364, loss_box_1: 3.2042, loss_cns_1: 0.4754, loss_yns_1: 0.1860, loss_cls_2: 1.2466, loss_box_2: 3.1340, loss_cns_2: 0.4892, loss_yns_2: 0.1833, loss_cls_3: 1.2128, loss_box_3: 3.3905, loss_cns_3: 0.5031, loss_yns_3: 0.1971, loss_cls_4: 1.2163, loss_box_4: 3.2169, loss_cns_4: 0.4679, loss_yns_4: 0.1937, loss_cls_5: 1.2390, loss_box_5: 3.2329, loss_cns_5: 0.5077, loss_yns_5: 0.1913, loss_cls_dn_0: 0.4843, loss_box_dn_0: 1.0739, loss_cls_dn_1: 0.4393, loss_box_dn_1: 2.1116, loss_cls_dn_2: 0.4953, loss_box_dn_2: 2.1327, loss_cls_dn_3: 0.4536, loss_box_dn_3: 2.2288, loss_cls_dn_4: 0.4518, loss_box_dn_4: 2.1616, loss_cls_dn_5: 0.4211, loss_box_dn_5: 2.1859, loss_dense_depth: 1.4611, loss: 46.1372, grad_norm: 65.7246 -2026-02-10 16:24:31,111 - mmdet - INFO - Iter [12/17500] lr: 1.044e-04, eta: 20:47:26, time: 1.890, data_time: 0.058, memory: 48982, loss_cls_0: 1.1595, loss_box_0: 2.3865, loss_cns_0: 0.5904, loss_yns_0: 0.1724, loss_cls_1: 1.1859, loss_box_1: 2.9759, loss_cns_1: 0.5365, loss_yns_1: 0.1770, loss_cls_2: 1.2071, loss_box_2: 2.9559, loss_cns_2: 0.5170, loss_yns_2: 0.1780, loss_cls_3: 1.1772, loss_box_3: 3.0361, loss_cns_3: 0.5274, loss_yns_3: 0.1760, loss_cls_4: 1.1883, loss_box_4: 3.0687, loss_cns_4: 0.4991, loss_yns_4: 0.1804, loss_cls_5: 1.2358, loss_box_5: 3.2420, loss_cns_5: 0.5387, loss_yns_5: 0.1885, loss_cls_dn_0: 0.4524, loss_box_dn_0: 1.1149, loss_cls_dn_1: 0.4241, loss_box_dn_1: 2.4298, loss_cls_dn_2: 0.4763, loss_box_dn_2: 2.4283, loss_cls_dn_3: 0.4076, loss_box_dn_3: 2.4293, loss_cls_dn_4: 0.4253, loss_box_dn_4: 2.4341, loss_cls_dn_5: 0.4129, loss_box_dn_5: 2.4911, loss_dense_depth: 1.3725, loss: 46.3990, grad_norm: 65.0204 -2026-02-10 16:24:33,020 - mmdet - INFO - Iter [13/17500] lr: 1.048e-04, eta: 19:54:12, time: 1.909, data_time: 0.055, memory: 48982, loss_cls_0: 1.1696, loss_box_0: 2.4044, loss_cns_0: 0.5914, loss_yns_0: 0.1743, loss_cls_1: 1.1673, loss_box_1: 3.0107, loss_cns_1: 0.5150, loss_yns_1: 0.1723, loss_cls_2: 1.1922, loss_box_2: 3.0976, loss_cns_2: 0.4965, loss_yns_2: 0.1780, loss_cls_3: 1.1746, loss_box_3: 3.0752, loss_cns_3: 0.5191, loss_yns_3: 0.1809, loss_cls_4: 1.1500, loss_box_4: 3.2085, loss_cns_4: 0.5097, loss_yns_4: 0.1753, loss_cls_5: 1.1948, loss_box_5: 3.2592, loss_cns_5: 0.5135, loss_yns_5: 0.1751, loss_cls_dn_0: 0.4506, loss_box_dn_0: 1.1238, loss_cls_dn_1: 0.4514, loss_box_dn_1: 2.0346, loss_cls_dn_2: 0.4855, loss_box_dn_2: 2.0712, loss_cls_dn_3: 0.4240, loss_box_dn_3: 2.1504, loss_cls_dn_4: 0.4371, loss_box_dn_4: 2.3167, loss_cls_dn_5: 0.4488, loss_box_dn_5: 2.3761, loss_dense_depth: 1.3461, loss: 45.4215, grad_norm: 83.7972 -2026-02-10 16:24:34,933 - mmdet - INFO - Iter [14/17500] lr: 1.052e-04, eta: 19:08:40, time: 1.913, data_time: 0.058, memory: 48982, loss_cls_0: 1.1837, loss_box_0: 2.4529, loss_cns_0: 0.6016, loss_yns_0: 0.1625, loss_cls_1: 1.1984, loss_box_1: 2.9060, loss_cns_1: 0.5199, loss_yns_1: 0.1686, loss_cls_2: 1.2594, loss_box_2: 2.9963, loss_cns_2: 0.5217, loss_yns_2: 0.1851, loss_cls_3: 1.2190, loss_box_3: 2.9051, loss_cns_3: 0.5356, loss_yns_3: 0.1778, loss_cls_4: 1.2019, loss_box_4: 2.9608, loss_cns_4: 0.5355, loss_yns_4: 0.1754, loss_cls_5: 1.2235, loss_box_5: 2.9376, loss_cns_5: 0.5421, loss_yns_5: 0.1641, loss_cls_dn_0: 0.4501, loss_box_dn_0: 1.1239, loss_cls_dn_1: 0.4702, loss_box_dn_1: 1.6939, loss_cls_dn_2: 0.4808, loss_box_dn_2: 1.6950, loss_cls_dn_3: 0.4333, loss_box_dn_3: 1.7892, loss_cls_dn_4: 0.4362, loss_box_dn_4: 1.9611, loss_cls_dn_5: 0.4796, loss_box_dn_5: 2.0076, loss_dense_depth: 1.3309, loss: 43.0863, grad_norm: 77.5273 -2026-02-10 16:24:36,836 - mmdet - INFO - Iter [15/17500] lr: 1.056e-04, eta: 18:28:59, time: 1.902, data_time: 0.056, memory: 48982, loss_cls_0: 1.1355, loss_box_0: 2.4066, loss_cns_0: 0.6015, loss_yns_0: 0.1633, loss_cls_1: 1.1868, loss_box_1: 2.8821, loss_cns_1: 0.5276, loss_yns_1: 0.1668, loss_cls_2: 1.2999, loss_box_2: 2.8234, loss_cns_2: 0.5504, loss_yns_2: 0.1878, loss_cls_3: 1.1803, loss_box_3: 2.7350, loss_cns_3: 0.5375, loss_yns_3: 0.1827, loss_cls_4: 1.2177, loss_box_4: 2.8013, loss_cns_4: 0.5454, loss_yns_4: 0.1691, loss_cls_5: 1.1811, loss_box_5: 2.8732, loss_cns_5: 0.5503, loss_yns_5: 0.1683, loss_cls_dn_0: 0.4481, loss_box_dn_0: 1.1214, loss_cls_dn_1: 0.4503, loss_box_dn_1: 1.6465, loss_cls_dn_2: 0.4425, loss_box_dn_2: 1.5842, loss_cls_dn_3: 0.4187, loss_box_dn_3: 1.6320, loss_cls_dn_4: 0.4099, loss_box_dn_4: 1.7362, loss_cls_dn_5: 0.4590, loss_box_dn_5: 1.7995, loss_dense_depth: 1.1967, loss: 41.4186, grad_norm: 74.1926 -2026-02-10 16:24:38,746 - mmdet - INFO - Iter [16/17500] lr: 1.060e-04, eta: 17:54:25, time: 1.911, data_time: 0.056, memory: 48982, loss_cls_0: 1.1742, loss_box_0: 2.3201, loss_cns_0: 0.5949, loss_yns_0: 0.1643, loss_cls_1: 1.2316, loss_box_1: 2.8616, loss_cns_1: 0.5372, loss_yns_1: 0.1755, loss_cls_2: 1.2980, loss_box_2: 2.7661, loss_cns_2: 0.5399, loss_yns_2: 0.1735, loss_cls_3: 1.2242, loss_box_3: 2.7405, loss_cns_3: 0.5433, loss_yns_3: 0.1787, loss_cls_4: 1.2656, loss_box_4: 2.7815, loss_cns_4: 0.5536, loss_yns_4: 0.1718, loss_cls_5: 1.2244, loss_box_5: 2.8997, loss_cns_5: 0.5526, loss_yns_5: 0.1749, loss_cls_dn_0: 0.4677, loss_box_dn_0: 1.0833, loss_cls_dn_1: 0.4333, loss_box_dn_1: 1.6555, loss_cls_dn_2: 0.4317, loss_box_dn_2: 1.5642, loss_cls_dn_3: 0.4282, loss_box_dn_3: 1.6091, loss_cls_dn_4: 0.3986, loss_box_dn_4: 1.6605, loss_cls_dn_5: 0.4385, loss_box_dn_5: 1.7428, loss_dense_depth: 1.3008, loss: 41.3621, grad_norm: 60.9556 -2026-02-10 16:24:40,659 - mmdet - INFO - Iter [17/17500] lr: 1.064e-04, eta: 17:23:56, time: 1.913, data_time: 0.059, memory: 48982, loss_cls_0: 1.1491, loss_box_0: 2.3350, loss_cns_0: 0.5817, loss_yns_0: 0.1713, loss_cls_1: 1.2305, loss_box_1: 2.8996, loss_cns_1: 0.5387, loss_yns_1: 0.1749, loss_cls_2: 1.2430, loss_box_2: 2.8150, loss_cns_2: 0.5429, loss_yns_2: 0.1837, loss_cls_3: 1.2141, loss_box_3: 2.8037, loss_cns_3: 0.5310, loss_yns_3: 0.1836, loss_cls_4: 1.2487, loss_box_4: 2.8550, loss_cns_4: 0.5403, loss_yns_4: 0.1675, loss_cls_5: 1.2533, loss_box_5: 2.9523, loss_cns_5: 0.5473, loss_yns_5: 0.1800, loss_cls_dn_0: 0.4595, loss_box_dn_0: 1.0762, loss_cls_dn_1: 0.4040, loss_box_dn_1: 1.6948, loss_cls_dn_2: 0.4251, loss_box_dn_2: 1.6045, loss_cls_dn_3: 0.4099, loss_box_dn_3: 1.6715, loss_cls_dn_4: 0.3899, loss_box_dn_4: 1.6879, loss_cls_dn_5: 0.4092, loss_box_dn_5: 1.7960, loss_dense_depth: 1.2201, loss: 41.5911, grad_norm: 55.4273 -2026-02-10 16:24:42,568 - mmdet - INFO - Iter [18/17500] lr: 1.068e-04, eta: 16:56:47, time: 1.909, data_time: 0.060, memory: 48982, loss_cls_0: 1.1515, loss_box_0: 2.2969, loss_cns_0: 0.5778, loss_yns_0: 0.1744, loss_cls_1: 1.2128, loss_box_1: 2.8443, loss_cns_1: 0.5359, loss_yns_1: 0.1706, loss_cls_2: 1.2086, loss_box_2: 2.8043, loss_cns_2: 0.5352, loss_yns_2: 0.1750, loss_cls_3: 1.1994, loss_box_3: 2.7540, loss_cns_3: 0.5423, loss_yns_3: 0.1823, loss_cls_4: 1.2137, loss_box_4: 2.8058, loss_cns_4: 0.5442, loss_yns_4: 0.1714, loss_cls_5: 1.2410, loss_box_5: 2.8629, loss_cns_5: 0.5596, loss_yns_5: 0.1834, loss_cls_dn_0: 0.4269, loss_box_dn_0: 1.0631, loss_cls_dn_1: 0.3961, loss_box_dn_1: 1.6488, loss_cls_dn_2: 0.4320, loss_box_dn_2: 1.6176, loss_cls_dn_3: 0.3861, loss_box_dn_3: 1.6968, loss_cls_dn_4: 0.3959, loss_box_dn_4: 1.7235, loss_cls_dn_5: 0.3972, loss_box_dn_5: 1.8718, loss_dense_depth: 1.1561, loss: 41.1591, grad_norm: 60.7094 -2026-02-10 16:24:44,475 - mmdet - INFO - Iter [19/17500] lr: 1.072e-04, eta: 16:32:28, time: 1.907, data_time: 0.060, memory: 48982, loss_cls_0: 1.1398, loss_box_0: 2.3630, loss_cns_0: 0.5497, loss_yns_0: 0.1770, loss_cls_1: 1.2019, loss_box_1: 3.0391, loss_cns_1: 0.5083, loss_yns_1: 0.1812, loss_cls_2: 1.1990, loss_box_2: 3.0084, loss_cns_2: 0.5155, loss_yns_2: 0.1730, loss_cls_3: 1.2186, loss_box_3: 3.0036, loss_cns_3: 0.5252, loss_yns_3: 0.1715, loss_cls_4: 1.1861, loss_box_4: 3.0344, loss_cns_4: 0.5122, loss_yns_4: 0.1647, loss_cls_5: 1.2087, loss_box_5: 3.0914, loss_cns_5: 0.5167, loss_yns_5: 0.1758, loss_cls_dn_0: 0.4275, loss_box_dn_0: 1.0627, loss_cls_dn_1: 0.4180, loss_box_dn_1: 1.4123, loss_cls_dn_2: 0.4555, loss_box_dn_2: 1.5224, loss_cls_dn_3: 0.3882, loss_box_dn_3: 1.6533, loss_cls_dn_4: 0.4203, loss_box_dn_4: 1.7119, loss_cls_dn_5: 0.4075, loss_box_dn_5: 1.8616, loss_dense_depth: 1.1188, loss: 41.7245, grad_norm: 68.2989 -2026-02-10 16:24:46,376 - mmdet - INFO - Iter [20/17500] lr: 1.076e-04, eta: 16:10:28, time: 1.900, data_time: 0.054, memory: 48982, loss_cls_0: 1.1395, loss_box_0: 2.3982, loss_cns_0: 0.5489, loss_yns_0: 0.1797, loss_cls_1: 1.2039, loss_box_1: 2.9053, loss_cns_1: 0.5224, loss_yns_1: 0.1918, loss_cls_2: 1.2099, loss_box_2: 2.8834, loss_cns_2: 0.5465, loss_yns_2: 0.1754, loss_cls_3: 1.2266, loss_box_3: 2.8572, loss_cns_3: 0.5476, loss_yns_3: 0.1790, loss_cls_4: 1.1748, loss_box_4: 2.8716, loss_cns_4: 0.5521, loss_yns_4: 0.1698, loss_cls_5: 1.2289, loss_box_5: 2.8589, loss_cns_5: 0.5350, loss_yns_5: 0.1675, loss_cls_dn_0: 0.4589, loss_box_dn_0: 1.0327, loss_cls_dn_1: 0.4118, loss_box_dn_1: 1.6526, loss_cls_dn_2: 0.4345, loss_box_dn_2: 1.7580, loss_cls_dn_3: 0.3783, loss_box_dn_3: 1.8332, loss_cls_dn_4: 0.4048, loss_box_dn_4: 1.8530, loss_cls_dn_5: 0.3932, loss_box_dn_5: 1.9424, loss_dense_depth: 1.1680, loss: 41.9956, grad_norm: 58.1297 -2026-02-10 16:24:48,398 - mmdet - INFO - Iter [21/17500] lr: 1.080e-04, eta: 15:52:14, time: 2.022, data_time: 0.165, memory: 48982, loss_cls_0: 1.1452, loss_box_0: 2.3601, loss_cns_0: 0.5638, loss_yns_0: 0.1757, loss_cls_1: 1.2161, loss_box_1: 2.6281, loss_cns_1: 0.5721, loss_yns_1: 0.1886, loss_cls_2: 1.2254, loss_box_2: 2.6668, loss_cns_2: 0.5720, loss_yns_2: 0.1744, loss_cls_3: 1.2164, loss_box_3: 2.6104, loss_cns_3: 0.5625, loss_yns_3: 0.1733, loss_cls_4: 1.1881, loss_box_4: 2.6280, loss_cns_4: 0.5870, loss_yns_4: 0.1709, loss_cls_5: 1.2454, loss_box_5: 2.6534, loss_cns_5: 0.5686, loss_yns_5: 0.1675, loss_cls_dn_0: 0.4599, loss_box_dn_0: 1.0506, loss_cls_dn_1: 0.4122, loss_box_dn_1: 1.3581, loss_cls_dn_2: 0.4181, loss_box_dn_2: 1.4265, loss_cls_dn_3: 0.3901, loss_box_dn_3: 1.4431, loss_cls_dn_4: 0.4033, loss_box_dn_4: 1.4455, loss_cls_dn_5: 0.4036, loss_box_dn_5: 1.4881, loss_dense_depth: 1.0933, loss: 39.0521, grad_norm: 50.8031 -2026-02-10 16:24:54,398 - mmdet - INFO - Iter [22/17500] lr: 1.084e-04, eta: 16:28:19, time: 5.997, data_time: 0.153, memory: 48982, loss_cls_0: 1.1313, loss_box_0: 2.2738, loss_cns_0: 0.5938, loss_yns_0: 0.1783, loss_cls_1: 1.2127, loss_box_1: 2.4730, loss_cns_1: 0.6170, loss_yns_1: 0.1790, loss_cls_2: 1.2076, loss_box_2: 2.4970, loss_cns_2: 0.5917, loss_yns_2: 0.1786, loss_cls_3: 1.1866, loss_box_3: 2.4966, loss_cns_3: 0.5860, loss_yns_3: 0.1707, loss_cls_4: 1.1897, loss_box_4: 2.4767, loss_cns_4: 0.6131, loss_yns_4: 0.1697, loss_cls_5: 1.2090, loss_box_5: 2.5179, loss_cns_5: 0.5970, loss_yns_5: 0.1729, loss_cls_dn_0: 0.4160, loss_box_dn_0: 1.0353, loss_cls_dn_1: 0.3967, loss_box_dn_1: 1.1479, loss_cls_dn_2: 0.3939, loss_box_dn_2: 1.1652, loss_cls_dn_3: 0.4000, loss_box_dn_3: 1.1808, loss_cls_dn_4: 0.3943, loss_box_dn_4: 1.1992, loss_cls_dn_5: 0.4154, loss_box_dn_5: 1.2086, loss_dense_depth: 1.1150, loss: 36.9879, grad_norm: 48.8411 diff --git a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20260210_162320.log.json b/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20260210_162320.log.json deleted file mode 100644 index c3b0effeb11b845f44b98b3cc6d0ee823a924898..0000000000000000000000000000000000000000 --- a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20260210_162320.log.json +++ /dev/null @@ -1,23 +0,0 @@ -{"env_info": "sys.platform: linux\nPython: 3.10.12 (main, Aug 15 2025, 14:32:43) [GCC 11.4.0]\nCUDA available: True\nGPU 0,1,2,3,4,5,6,7: BW200_LC, UBB BW1000_LC\nCUDA_HOME: /opt/dtk\nNVCC: Not Available\nGCC: x86_64-linux-gnu-gcc (Ubuntu 11.4.0-1ubuntu1~22.04) 11.4.0\nPyTorch: 2.5.1\nPyTorch compiling details: PyTorch built with:\n - GCC 10.3\n - C++ Version: 201703\n - Intel(R) Math Kernel Library Version 2020.0.4 Product Build 20200917 for Intel(R) 64 architecture applications\n - OpenMP 201511 (a.k.a. OpenMP 4.5)\n - LAPACK is enabled (usually provided by MKL)\n - NNPACK is enabled\n - CPU capability usage: AVX2\n - HIP Runtime 6.3.25521\n - MIOpen 2.18.0\n - Magma 2.8.0\n - Build settings: BLAS_INFO=mkl, BUILD_TYPE=Release, CXX_COMPILER=/opt/rh/gcc-toolset-10/root/usr/bin/c++, CXX_FLAGS= -D_GLIBCXX_USE_CXX11_ABI=1 -fvisibility-inlines-hidden -DUSE_PTHREADPOOL -DNDEBUG -DUSE_KINETO -DLIBKINETO_NOCUPTI -DLIBKINETO_NOXPUPTI=ON -DUSE_FBGEMM -DUSE_PYTORCH_QNNPACK -DUSE_XNNPACK -DSYMBOLICATE_MOBILE_DEBUG_HANDLE -O2 -fPIC -Wall -Wextra -Werror=return-type -Werror=non-virtual-dtor -Werror=bool-operation -Wnarrowing -Wno-missing-field-initializers -Wno-type-limits -Wno-array-bounds -Wno-unknown-pragmas -Wno-unused-parameter -Wno-strict-overflow -Wno-strict-aliasing -Wno-stringop-overflow -Wsuggest-override -Wno-psabi -Wno-error=old-style-cast -Wno-missing-braces -fdiagnostics-color=always -faligned-new -Wno-unused-but-set-variable -Wno-maybe-uninitialized -fno-math-errno -fno-trapping-math -Werror=format -Wno-stringop-overflow, FORCE_FALLBACK_CUDA_MPI=1, LAPACK_INFO=mkl, PERF_WITH_AVX=1, PERF_WITH_AVX2=1, TORCH_VERSION=2.5.1, USE_CUDA=0, USE_CUDNN=OFF, USE_CUSPARSELT=OFF, USE_EXCEPTION_PTR=1, USE_GFLAGS=1, USE_GLOG=1, USE_GLOO=1, USE_MKL=ON, USE_MKLDNN=0, USE_MPI=1, USE_NCCL=1, USE_NNPACK=ON, USE_OPENMP=1, USE_ROCM=ON, USE_ROCM_KERNEL_ASSERT=OFF, \n\nTorchVision: 0.20.1\nOpenCV: 4.12.0\nMMCV: 1.6.1\nMMCV Compiler: GCC 10.3\nMMCV CUDA Compiler: rocm not available\nMMDetection: 2.25.1+", "config": "plugin = True\nplugin_dir = 'projects/mmdet3d_plugin/'\ndist_params = dict(backend='nccl')\nlog_level = 'INFO'\nwork_dir = './work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704'\ntotal_batch_size = 160\nnum_gpus = 8\nbatch_size = 20\nnum_iters_per_epoch = 175\nnum_epochs = 100\ncheckpoint_epoch_interval = 20\ncheckpoint_config = dict(interval=3500)\nlog_config = dict(\n interval=1,\n hooks=[\n dict(type='TextLoggerHook', by_epoch=False),\n dict(type='TensorboardLoggerHook')\n ])\nload_from = None\nresume_from = None\nworkflow = [('train', 1)]\nfp16 = dict(loss_scale=32.0)\ninput_shape = (704, 256)\ntracking_test = True\ntracking_threshold = 0.2\nclass_names = [\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier',\n 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n]\nnum_classes = 10\nembed_dims = 256\nnum_groups = 8\nnum_decoder = 6\nnum_single_frame_decoder = 1\nuse_deformable_func = True\nstrides = [4, 8, 16, 32]\nnum_levels = 4\nnum_depth_layers = 3\ndrop_out = 0.1\ntemporal = True\ndecouple_attn = True\nwith_quality_estimation = True\nmodel = dict(\n type='Sparse4D',\n use_grid_mask=True,\n use_deformable_func=True,\n img_backbone=dict(\n type='ResNet',\n depth=50,\n num_stages=4,\n frozen_stages=-1,\n norm_eval=False,\n style='pytorch',\n with_cp=True,\n out_indices=(0, 1, 2, 3),\n norm_cfg=dict(type='BN', requires_grad=True),\n pretrained='ckpt/resnet50-19c8e357.pth'),\n img_neck=dict(\n type='FPN',\n num_outs=4,\n start_level=0,\n out_channels=256,\n add_extra_convs='on_output',\n relu_before_extra_convs=True,\n in_channels=[256, 512, 1024, 2048]),\n depth_branch=dict(\n type='DenseDepthNet',\n embed_dims=256,\n num_depth_layers=3,\n loss_weight=0.2),\n head=dict(\n type='Sparse4DHead',\n cls_threshold_to_reg=0.05,\n decouple_attn=True,\n instance_bank=dict(\n type='InstanceBank',\n num_anchor=900,\n embed_dims=256,\n anchor='nuscenes_kmeans900.npy',\n anchor_handler=dict(type='SparseBox3DKeyPointsGenerator'),\n num_temp_instances=600,\n confidence_decay=0.6,\n feat_grad=False),\n anchor_encoder=dict(\n type='SparseBox3DEncoder',\n vel_dims=3,\n embed_dims=[128, 32, 32, 64],\n mode='cat',\n output_fc=False,\n in_loops=1,\n out_loops=4),\n num_single_frame_decoder=1,\n operation_order=[\n 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm',\n 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm',\n 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm',\n 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm',\n 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm',\n 'deformable', 'ffn', 'norm', 'refine'\n ],\n temp_graph_model=dict(\n type='MultiheadAttention',\n embed_dims=512,\n num_heads=8,\n batch_first=True,\n dropout=0.1),\n graph_model=dict(\n type='MultiheadAttention',\n embed_dims=512,\n num_heads=8,\n batch_first=True,\n dropout=0.1),\n norm_layer=dict(type='LN', normalized_shape=256),\n ffn=dict(\n type='AsymmetricFFN',\n in_channels=512,\n pre_norm=dict(type='LN'),\n embed_dims=256,\n feedforward_channels=1024,\n num_fcs=2,\n ffn_drop=0.1,\n act_cfg=dict(type='ReLU', inplace=True)),\n deformable_model=dict(\n type='DeformableFeatureAggregation',\n embed_dims=256,\n num_groups=8,\n num_levels=4,\n num_cams=6,\n attn_drop=0.15,\n use_deformable_func=True,\n use_camera_embed=True,\n residual_mode='cat',\n kps_generator=dict(\n type='SparseBox3DKeyPointsGenerator',\n num_learnable_pts=6,\n fix_scale=[[0, 0, 0], [0.45, 0, 0], [-0.45, 0, 0],\n [0, 0.45, 0], [0, -0.45, 0], [0, 0, 0.45],\n [0, 0, -0.45]])),\n refine_layer=dict(\n type='SparseBox3DRefinementModule',\n embed_dims=256,\n num_cls=10,\n refine_yaw=True,\n with_quality_estimation=True),\n sampler=dict(\n type='SparseBox3DTarget',\n num_dn_groups=5,\n num_temp_dn_groups=3,\n dn_noise_scale=[2.0, 2.0, 2.0, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5],\n max_dn_gt=32,\n add_neg_dn=True,\n cls_weight=2.0,\n box_weight=0.25,\n reg_weights=[2.0, 2.0, 2.0, 0.5, 0.5, 0.5, 0.0, 0.0, 0.0, 0.0],\n cls_wise_reg_weights=dict(\n {9: [2.0, 2.0, 2.0, 1.0, 1.0, 1.0, 0.0, 0.0, 1.0, 1.0]})),\n loss_cls=dict(\n type='FocalLoss',\n use_sigmoid=True,\n gamma=2.0,\n alpha=0.25,\n loss_weight=2.0),\n loss_reg=dict(\n type='SparseBox3DLoss',\n loss_box=dict(type='L1Loss', loss_weight=0.25),\n loss_centerness=dict(type='CrossEntropyLoss', use_sigmoid=True),\n loss_yawness=dict(type='GaussianFocalLoss'),\n cls_allow_reverse=[5]),\n decoder=dict(type='SparseBox3DDecoder'),\n reg_weights=[2.0, 2.0, 2.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0]))\ndataset_type = 'NuScenes3DDetTrackDataset'\ndata_root = 'data/nuscenes/'\nanno_root = 'data/nuscenes_anno_pkls/'\nfile_client_args = dict(backend='disk')\nimg_norm_cfg = dict(\n mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375], to_rgb=True)\ntrain_pipeline = [\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(\n type='LoadPointsFromFile',\n coord_type='LIDAR',\n load_dim=5,\n use_dim=5,\n file_client_args=dict(backend='disk')),\n dict(type='ResizeCropFlipImage'),\n dict(type='MultiScaleDepthMapGenerator', downsample=[4, 8, 16]),\n dict(type='BBoxRotation'),\n dict(type='PhotoMetricDistortionMultiViewImage'),\n dict(\n type='NormalizeMultiviewImage',\n mean=[123.675, 116.28, 103.53],\n std=[58.395, 57.12, 57.375],\n to_rgb=True),\n dict(\n type='CircleObjectRangeFilter',\n class_dist_thred=[55, 55, 55, 55, 55, 55, 55, 55, 55, 55]),\n dict(\n type='InstanceNameFilter',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ]),\n dict(type='NuScenesSparse4DAdaptor'),\n dict(\n type='Collect',\n keys=[\n 'img', 'timestamp', 'projection_mat', 'image_wh', 'gt_depth',\n 'focal', 'gt_bboxes_3d', 'gt_labels_3d'\n ],\n meta_keys=['T_global', 'T_global_inv', 'timestamp', 'instance_id'])\n]\ntest_pipeline = [\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(type='ResizeCropFlipImage'),\n dict(\n type='NormalizeMultiviewImage',\n mean=[123.675, 116.28, 103.53],\n std=[58.395, 57.12, 57.375],\n to_rgb=True),\n dict(type='NuScenesSparse4DAdaptor'),\n dict(\n type='Collect',\n keys=['img', 'timestamp', 'projection_mat', 'image_wh'],\n meta_keys=['T_global', 'T_global_inv', 'timestamp'])\n]\ninput_modality = dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False)\ndata_basic_config = dict(\n type='NuScenes3DDetTrackDataset',\n data_root='data/nuscenes/',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier',\n 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ],\n modality=dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False),\n version='v1.0-trainval')\ndata_aug_conf = dict(\n resize_lim=(0.4, 0.47),\n final_dim=(256, 704),\n bot_pct_lim=(0.0, 0.0),\n rot_lim=(-5.4, 5.4),\n H=900,\n W=1600,\n rand_flip=True,\n rot3d_range=[-0.3925, 0.3925])\ndata = dict(\n samples_per_gpu=20,\n workers_per_gpu=20,\n train=dict(\n type='NuScenes3DDetTrackDataset',\n data_root='data/nuscenes/',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ],\n modality=dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False),\n version='v1.0-trainval',\n ann_file='data/nuscenes_anno_pkls/nuscenes_infos_train.pkl',\n pipeline=[\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(\n type='LoadPointsFromFile',\n coord_type='LIDAR',\n load_dim=5,\n use_dim=5,\n file_client_args=dict(backend='disk')),\n dict(type='ResizeCropFlipImage'),\n dict(type='MultiScaleDepthMapGenerator', downsample=[4, 8, 16]),\n dict(type='BBoxRotation'),\n dict(type='PhotoMetricDistortionMultiViewImage'),\n dict(\n type='NormalizeMultiviewImage',\n mean=[123.675, 116.28, 103.53],\n std=[58.395, 57.12, 57.375],\n to_rgb=True),\n dict(\n type='CircleObjectRangeFilter',\n class_dist_thred=[55, 55, 55, 55, 55, 55, 55, 55, 55, 55]),\n dict(\n type='InstanceNameFilter',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian',\n 'traffic_cone'\n ]),\n dict(type='NuScenesSparse4DAdaptor'),\n dict(\n type='Collect',\n keys=[\n 'img', 'timestamp', 'projection_mat', 'image_wh',\n 'gt_depth', 'focal', 'gt_bboxes_3d', 'gt_labels_3d'\n ],\n meta_keys=[\n 'T_global', 'T_global_inv', 'timestamp', 'instance_id'\n ])\n ],\n test_mode=False,\n data_aug_conf=dict(\n resize_lim=(0.4, 0.47),\n final_dim=(256, 704),\n bot_pct_lim=(0.0, 0.0),\n rot_lim=(-5.4, 5.4),\n H=900,\n W=1600,\n rand_flip=True,\n rot3d_range=[-0.3925, 0.3925]),\n with_seq_flag=True,\n sequences_split_num=2,\n keep_consistent_seq_aug=True),\n val=dict(\n type='NuScenes3DDetTrackDataset',\n data_root='data/nuscenes/',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ],\n modality=dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False),\n version='v1.0-trainval',\n ann_file='data/nuscenes_anno_pkls/nuscenes_infos_val.pkl',\n pipeline=[\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(type='ResizeCropFlipImage'),\n dict(\n type='NormalizeMultiviewImage',\n mean=[123.675, 116.28, 103.53],\n std=[58.395, 57.12, 57.375],\n to_rgb=True),\n dict(type='NuScenesSparse4DAdaptor'),\n dict(\n type='Collect',\n keys=['img', 'timestamp', 'projection_mat', 'image_wh'],\n meta_keys=['T_global', 'T_global_inv', 'timestamp'])\n ],\n data_aug_conf=dict(\n resize_lim=(0.4, 0.47),\n final_dim=(256, 704),\n bot_pct_lim=(0.0, 0.0),\n rot_lim=(-5.4, 5.4),\n H=900,\n W=1600,\n rand_flip=True,\n rot3d_range=[-0.3925, 0.3925]),\n test_mode=True,\n tracking=True,\n tracking_threshold=0.2),\n test=dict(\n type='NuScenes3DDetTrackDataset',\n data_root='data/nuscenes/',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ],\n modality=dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False),\n version='v1.0-trainval',\n ann_file='data/nuscenes_anno_pkls/nuscenes_infos_val.pkl',\n pipeline=[\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(type='ResizeCropFlipImage'),\n dict(\n type='NormalizeMultiviewImage',\n mean=[123.675, 116.28, 103.53],\n std=[58.395, 57.12, 57.375],\n to_rgb=True),\n dict(type='NuScenesSparse4DAdaptor'),\n dict(\n type='Collect',\n keys=['img', 'timestamp', 'projection_mat', 'image_wh'],\n meta_keys=['T_global', 'T_global_inv', 'timestamp'])\n ],\n data_aug_conf=dict(\n resize_lim=(0.4, 0.47),\n final_dim=(256, 704),\n bot_pct_lim=(0.0, 0.0),\n rot_lim=(-5.4, 5.4),\n H=900,\n W=1600,\n rand_flip=True,\n rot3d_range=[-0.3925, 0.3925]),\n test_mode=True,\n tracking=True,\n tracking_threshold=0.2))\noptimizer = dict(\n type='AdamW',\n lr=0.0006,\n weight_decay=0.001,\n paramwise_cfg=dict(custom_keys=dict(img_backbone=dict(lr_mult=0.5))))\noptimizer_config = dict(grad_clip=dict(max_norm=25, norm_type=2))\nlr_config = dict(\n policy='CosineAnnealing',\n warmup='linear',\n warmup_iters=500,\n warmup_ratio=0.3333333333333333,\n min_lr_ratio=0.001)\nrunner = dict(type='IterBasedRunner', max_iters=17500)\nvis_pipeline = [\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(type='Collect', keys=['img'], meta_keys=['timestamp', 'lidar2img'])\n]\nevaluation = dict(\n interval=3500,\n pipeline=[\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(\n type='Collect', keys=['img'], meta_keys=['timestamp', 'lidar2img'])\n ])\ngpu_ids = range(0, 1)\n", "seed": 0, "exp_name": "sparse4dv3_temporal_r50_1x8_bs6_256x704.py"} -{"mode": "train", "epoch": 1, "iter": 1, "lr": 0.0001, "memory": 48982, "data_time": 8.28572, "loss_cls_0": 2.35619, "loss_box_0": 0.0, "loss_cns_0": 0.0, "loss_yns_0": 0.0, "loss_cls_1": 2.15524, "loss_box_1": 0.1646, "loss_cns_1": 0.0317, "loss_yns_1": 0.00776, "loss_cls_2": 2.40444, "loss_box_2": 0.0, "loss_cns_2": 0.0, "loss_yns_2": 0.0, "loss_cls_3": 2.47504, "loss_box_3": 0.02766, "loss_cns_3": 0.00504, "loss_yns_3": 0.00258, "loss_cls_4": 1.99462, "loss_box_4": 0.42352, "loss_cns_4": 0.05633, "loss_yns_4": 0.02724, "loss_cls_5": 2.36391, "loss_box_5": 0.0, "loss_cns_5": 0.0, "loss_yns_5": 0.0, "loss_cls_dn_0": 1.19465, "loss_box_dn_0": 1.4557, "loss_cls_dn_1": 1.12369, "loss_box_dn_1": 1.73094, "loss_cls_dn_2": 1.19905, "loss_box_dn_2": 1.97989, "loss_cls_dn_3": 1.17326, "loss_box_dn_3": 2.24526, "loss_cls_dn_4": 1.04999, "loss_box_dn_4": 2.41683, "loss_cls_dn_5": 1.20662, "loss_box_dn_5": 2.66472, "loss_dense_depth": 1.86612, "loss": 35.8026, "grad_norm": 371.39169, "time": 29.2041} -{"mode": "train", "epoch": 1, "iter": 2, "lr": 0.0001, "memory": 48982, "data_time": 0.05889, "loss_cls_0": 2.10864, "loss_box_0": 0.0, "loss_cns_0": 0.0, "loss_yns_0": 0.0, "loss_cls_1": 2.08176, "loss_box_1": 0.10803, "loss_cns_1": 0.02006, "loss_yns_1": 0.00669, "loss_cls_2": 2.20313, "loss_box_2": 0.05018, "loss_cns_2": 0.00552, "loss_yns_2": 0.00168, "loss_cls_3": 2.01201, "loss_box_3": 0.19652, "loss_cns_3": 0.02724, "loss_yns_3": 0.00706, "loss_cls_4": 1.85246, "loss_box_4": 0.84468, "loss_cns_4": 0.09304, "loss_yns_4": 0.03379, "loss_cls_5": 1.99912, "loss_box_5": 0.99813, "loss_cns_5": 0.12343, "loss_yns_5": 0.05365, "loss_cls_dn_0": 1.07397, "loss_box_dn_0": 1.36174, "loss_cls_dn_1": 1.02139, "loss_box_dn_1": 2.46903, "loss_cls_dn_2": 1.03273, "loss_box_dn_2": 2.57402, "loss_cls_dn_3": 0.96509, "loss_box_dn_3": 2.63736, "loss_cls_dn_4": 0.89406, "loss_box_dn_4": 2.86349, "loss_cls_dn_5": 0.99582, "loss_box_dn_5": 3.11749, "loss_dense_depth": 1.77839, "loss": 37.61139, "grad_norm": 80.35947, "time": 2.97948} -{"mode": "train", "epoch": 1, "iter": 3, "lr": 0.0001, "memory": 48982, "data_time": 0.06068, "loss_cls_0": 1.47415, "loss_box_0": 2.39625, "loss_cns_0": 0.80986, "loss_yns_0": 0.24582, "loss_cls_1": 1.82361, "loss_box_1": 0.3521, "loss_cns_1": 0.0525, "loss_yns_1": 0.01943, "loss_cls_2": 1.87586, "loss_box_2": 1.46907, "loss_cns_2": 0.18357, "loss_yns_2": 0.07959, "loss_cls_3": 1.63124, "loss_box_3": 3.98897, "loss_cns_3": 0.46023, "loss_yns_3": 0.20449, "loss_cls_4": 1.58986, "loss_box_4": 3.57471, "loss_cns_4": 0.34192, "loss_yns_4": 0.15264, "loss_cls_5": 1.67282, "loss_box_5": 3.61456, "loss_cns_5": 0.31428, "loss_yns_5": 0.12634, "loss_cls_dn_0": 0.75138, "loss_box_dn_0": 1.1602, "loss_cls_dn_1": 0.88731, "loss_box_dn_1": 2.41907, "loss_cls_dn_2": 0.86783, "loss_box_dn_2": 2.67075, "loss_cls_dn_3": 0.75016, "loss_box_dn_3": 2.86963, "loss_cls_dn_4": 0.75623, "loss_box_dn_4": 3.10682, "loss_cls_dn_5": 0.8192, "loss_box_dn_5": 3.34725, "loss_dense_depth": 1.74252, "loss": 50.60225, "grad_norm": 120.94074, "time": 1.89048} -{"mode": "train", "epoch": 1, "iter": 4, "lr": 0.0001, "memory": 48982, "data_time": 0.05663, "loss_cls_0": 1.32556, "loss_box_0": 2.70742, "loss_cns_0": 0.65678, "loss_yns_0": 0.22306, "loss_cls_1": 1.63152, "loss_box_1": 1.87345, "loss_cns_1": 0.25486, "loss_yns_1": 0.1515, "loss_cls_2": 1.67871, "loss_box_2": 3.49226, "loss_cns_2": 0.39974, "loss_yns_2": 0.17542, "loss_cls_3": 1.45093, "loss_box_3": 4.73298, "loss_cns_3": 0.47291, "loss_yns_3": 0.24918, "loss_cls_4": 1.4127, "loss_box_4": 5.34461, "loss_cns_4": 0.33402, "loss_yns_4": 0.19426, "loss_cls_5": 1.51049, "loss_box_5": 5.32079, "loss_cns_5": 0.44385, "loss_yns_5": 0.19392, "loss_cls_dn_0": 0.6367, "loss_box_dn_0": 1.17985, "loss_cls_dn_1": 0.79607, "loss_box_dn_1": 2.92625, "loss_cls_dn_2": 0.7573, "loss_box_dn_2": 3.05384, "loss_cls_dn_3": 0.64483, "loss_box_dn_3": 3.32612, "loss_cls_dn_4": 0.65156, "loss_box_dn_4": 3.57118, "loss_cls_dn_5": 0.70102, "loss_box_dn_5": 3.7995, "loss_dense_depth": 1.71135, "loss": 59.98648, "grad_norm": 147.5616, "time": 1.92624} -{"mode": "train", "epoch": 1, "iter": 5, "lr": 0.0001, "memory": 48982, "data_time": 0.06183, "loss_cls_0": 1.29196, "loss_box_0": 2.79507, "loss_cns_0": 0.5304, "loss_yns_0": 0.18397, "loss_cls_1": 1.52373, "loss_box_1": 3.35673, "loss_cns_1": 0.35991, "loss_yns_1": 0.17485, "loss_cls_2": 1.5681, "loss_box_2": 3.976, "loss_cns_2": 0.40929, "loss_yns_2": 0.19151, "loss_cls_3": 1.4263, "loss_box_3": 4.30207, "loss_cns_3": 0.43175, "loss_yns_3": 0.2109, "loss_cls_4": 1.38721, "loss_box_4": 4.4946, "loss_cns_4": 0.36663, "loss_yns_4": 0.19788, "loss_cls_5": 1.38468, "loss_box_5": 4.80849, "loss_cns_5": 0.44243, "loss_yns_5": 0.23076, "loss_cls_dn_0": 0.55773, "loss_box_dn_0": 1.25005, "loss_cls_dn_1": 0.73113, "loss_box_dn_1": 2.62241, "loss_cls_dn_2": 0.69977, "loss_box_dn_2": 2.73939, "loss_cls_dn_3": 0.57969, "loss_box_dn_3": 2.9493, "loss_cls_dn_4": 0.58796, "loss_box_dn_4": 3.12819, "loss_cls_dn_5": 0.61666, "loss_box_dn_5": 3.29229, "loss_dense_depth": 1.70038, "loss": 57.50016, "grad_norm": 143.93739, "time": 1.9059} -{"mode": "train", "epoch": 1, "iter": 6, "lr": 0.0001, "memory": 48982, "data_time": 0.06019, "loss_cls_0": 1.32925, "loss_box_0": 2.60431, "loss_cns_0": 0.53066, "loss_yns_0": 0.21624, "loss_cls_1": 1.4462, "loss_box_1": 3.34244, "loss_cns_1": 0.38516, "loss_yns_1": 0.17712, "loss_cls_2": 1.45544, "loss_box_2": 3.89955, "loss_cns_2": 0.39953, "loss_yns_2": 0.18955, "loss_cls_3": 1.39334, "loss_box_3": 4.01987, "loss_cns_3": 0.3904, "loss_yns_3": 0.20487, "loss_cls_4": 1.31665, "loss_box_4": 4.38939, "loss_cns_4": 0.31966, "loss_yns_4": 0.21122, "loss_cls_5": 1.34178, "loss_box_5": 4.64332, "loss_cns_5": 0.3519, "loss_yns_5": 0.22528, "loss_cls_dn_0": 0.50421, "loss_box_dn_0": 1.21217, "loss_cls_dn_1": 0.68089, "loss_box_dn_1": 2.83711, "loss_cls_dn_2": 0.61823, "loss_box_dn_2": 2.86104, "loss_cls_dn_3": 0.52637, "loss_box_dn_3": 2.90666, "loss_cls_dn_4": 0.52106, "loss_box_dn_4": 3.07139, "loss_cls_dn_5": 0.53101, "loss_box_dn_5": 3.1659, "loss_dense_depth": 1.66491, "loss": 55.88407, "grad_norm": 124.14434, "time": 1.90817} -{"mode": "train", "epoch": 1, "iter": 7, "lr": 0.0001, "memory": 48982, "data_time": 0.05333, "loss_cls_0": 1.29488, "loss_box_0": 2.44969, "loss_cns_0": 0.67278, "loss_yns_0": 0.19797, "loss_cls_1": 1.31416, "loss_box_1": 3.54422, "loss_cns_1": 0.44326, "loss_yns_1": 0.22336, "loss_cls_2": 1.38582, "loss_box_2": 3.65897, "loss_cns_2": 0.43599, "loss_yns_2": 0.19326, "loss_cls_3": 1.30714, "loss_box_3": 3.57818, "loss_cns_3": 0.44339, "loss_yns_3": 0.18777, "loss_cls_4": 1.36522, "loss_box_4": 3.94452, "loss_cns_4": 0.37179, "loss_yns_4": 0.19656, "loss_cls_5": 1.26811, "loss_box_5": 4.12012, "loss_cns_5": 0.36237, "loss_yns_5": 0.21686, "loss_cls_dn_0": 0.49547, "loss_box_dn_0": 1.16367, "loss_cls_dn_1": 0.59053, "loss_box_dn_1": 2.80335, "loss_cls_dn_2": 0.55154, "loss_box_dn_2": 2.79924, "loss_cls_dn_3": 0.49773, "loss_box_dn_3": 2.74945, "loss_cls_dn_4": 0.46737, "loss_box_dn_4": 2.8962, "loss_cls_dn_5": 0.46539, "loss_box_dn_5": 2.96428, "loss_dense_depth": 1.57747, "loss": 53.1981, "grad_norm": 115.37534, "time": 1.89979} -{"mode": "train", "epoch": 1, "iter": 8, "lr": 0.0001, "memory": 48982, "data_time": 0.05574, "loss_cls_0": 1.21757, "loss_box_0": 2.39466, "loss_cns_0": 0.7324, "loss_yns_0": 0.18257, "loss_cls_1": 1.30702, "loss_box_1": 3.60428, "loss_cns_1": 0.43616, "loss_yns_1": 0.2152, "loss_cls_2": 1.40819, "loss_box_2": 3.62034, "loss_cns_2": 0.43443, "loss_yns_2": 0.18768, "loss_cls_3": 1.26541, "loss_box_3": 3.42662, "loss_cns_3": 0.46835, "loss_yns_3": 0.19425, "loss_cls_4": 1.31561, "loss_box_4": 3.50075, "loss_cns_4": 0.48647, "loss_yns_4": 0.19485, "loss_cls_5": 1.3093, "loss_box_5": 3.54605, "loss_cns_5": 0.53828, "loss_yns_5": 0.19821, "loss_cls_dn_0": 0.49985, "loss_box_dn_0": 1.11696, "loss_cls_dn_1": 0.56778, "loss_box_dn_1": 1.76998, "loss_cls_dn_2": 0.53905, "loss_box_dn_2": 1.79421, "loss_cls_dn_3": 0.50032, "loss_box_dn_3": 1.74105, "loss_cls_dn_4": 0.4635, "loss_box_dn_4": 1.86773, "loss_cls_dn_5": 0.44936, "loss_box_dn_5": 1.93693, "loss_dense_depth": 1.54079, "loss": 46.97217, "grad_norm": 95.83277, "time": 1.92672} -{"mode": "train", "epoch": 1, "iter": 9, "lr": 0.0001, "memory": 48982, "data_time": 0.06797, "loss_cls_0": 1.18628, "loss_box_0": 2.35407, "loss_cns_0": 0.68303, "loss_yns_0": 0.18722, "loss_cls_1": 1.33071, "loss_box_1": 3.5613, "loss_cns_1": 0.42415, "loss_yns_1": 0.18796, "loss_cls_2": 1.38587, "loss_box_2": 3.54403, "loss_cns_2": 0.40102, "loss_yns_2": 0.19474, "loss_cls_3": 1.2383, "loss_box_3": 3.57001, "loss_cns_3": 0.39695, "loss_yns_3": 0.20359, "loss_cls_4": 1.24127, "loss_box_4": 3.51146, "loss_cns_4": 0.41608, "loss_yns_4": 0.17782, "loss_cls_5": 1.29947, "loss_box_5": 3.55032, "loss_cns_5": 0.46323, "loss_yns_5": 0.23522, "loss_cls_dn_0": 0.50927, "loss_box_dn_0": 1.07979, "loss_cls_dn_1": 0.499, "loss_box_dn_1": 1.69271, "loss_cls_dn_2": 0.51104, "loss_box_dn_2": 1.73089, "loss_cls_dn_3": 0.50104, "loss_box_dn_3": 1.80446, "loss_cls_dn_4": 0.46598, "loss_box_dn_4": 1.82453, "loss_cls_dn_5": 0.44071, "loss_box_dn_5": 1.91405, "loss_dense_depth": 1.48849, "loss": 46.20604, "grad_norm": 80.67724, "time": 1.96413} -{"mode": "train", "epoch": 1, "iter": 10, "lr": 0.0001, "memory": 48982, "data_time": 0.06834, "loss_cls_0": 1.15887, "loss_box_0": 2.34966, "loss_cns_0": 0.63664, "loss_yns_0": 0.18119, "loss_cls_1": 1.27834, "loss_box_1": 3.38713, "loss_cns_1": 0.4275, "loss_yns_1": 0.18882, "loss_cls_2": 1.25891, "loss_box_2": 3.35935, "loss_cns_2": 0.44342, "loss_yns_2": 0.18118, "loss_cls_3": 1.23968, "loss_box_3": 3.60767, "loss_cns_3": 0.45004, "loss_yns_3": 0.19612, "loss_cls_4": 1.21806, "loss_box_4": 3.40808, "loss_cns_4": 0.46158, "loss_yns_4": 0.18086, "loss_cls_5": 1.22574, "loss_box_5": 3.46972, "loss_cns_5": 0.49797, "loss_yns_5": 0.18937, "loss_cls_dn_0": 0.49985, "loss_box_dn_0": 1.05617, "loss_cls_dn_1": 0.46253, "loss_box_dn_1": 1.8615, "loss_cls_dn_2": 0.50011, "loss_box_dn_2": 1.92305, "loss_cls_dn_3": 0.4872, "loss_box_dn_3": 2.05454, "loss_cls_dn_4": 0.46917, "loss_box_dn_4": 2.00453, "loss_cls_dn_5": 0.43165, "loss_box_dn_5": 2.08215, "loss_dense_depth": 1.43974, "loss": 46.26806, "grad_norm": 90.01911, "time": 1.94044} -{"mode": "train", "epoch": 1, "iter": 11, "lr": 0.0001, "memory": 48982, "data_time": 0.08227, "loss_cls_0": 1.18283, "loss_box_0": 2.35362, "loss_cns_0": 0.60293, "loss_yns_0": 0.17276, "loss_cls_1": 1.23645, "loss_box_1": 3.20423, "loss_cns_1": 0.47535, "loss_yns_1": 0.18598, "loss_cls_2": 1.2466, "loss_box_2": 3.13401, "loss_cns_2": 0.48916, "loss_yns_2": 0.18329, "loss_cls_3": 1.21283, "loss_box_3": 3.39055, "loss_cns_3": 0.50305, "loss_yns_3": 0.19713, "loss_cls_4": 1.21629, "loss_box_4": 3.21685, "loss_cns_4": 0.46786, "loss_yns_4": 0.19369, "loss_cls_5": 1.23899, "loss_box_5": 3.23288, "loss_cns_5": 0.50767, "loss_yns_5": 0.19127, "loss_cls_dn_0": 0.48432, "loss_box_dn_0": 1.07387, "loss_cls_dn_1": 0.43931, "loss_box_dn_1": 2.11155, "loss_cls_dn_2": 0.49528, "loss_box_dn_2": 2.13273, "loss_cls_dn_3": 0.45358, "loss_box_dn_3": 2.22883, "loss_cls_dn_4": 0.4518, "loss_box_dn_4": 2.1616, "loss_cls_dn_5": 0.42111, "loss_box_dn_5": 2.18592, "loss_dense_depth": 1.46107, "loss": 46.13725, "grad_norm": 65.72465, "time": 1.92333} -{"mode": "train", "epoch": 1, "iter": 12, "lr": 0.0001, "memory": 48982, "data_time": 0.05806, "loss_cls_0": 1.15953, "loss_box_0": 2.38653, "loss_cns_0": 0.59036, "loss_yns_0": 0.17241, "loss_cls_1": 1.18587, "loss_box_1": 2.9759, "loss_cns_1": 0.53653, "loss_yns_1": 0.17698, "loss_cls_2": 1.20715, "loss_box_2": 2.95591, "loss_cns_2": 0.51695, "loss_yns_2": 0.17803, "loss_cls_3": 1.17721, "loss_box_3": 3.03609, "loss_cns_3": 0.52738, "loss_yns_3": 0.17603, "loss_cls_4": 1.18832, "loss_box_4": 3.06874, "loss_cns_4": 0.49906, "loss_yns_4": 0.18044, "loss_cls_5": 1.23582, "loss_box_5": 3.242, "loss_cns_5": 0.53875, "loss_yns_5": 0.18846, "loss_cls_dn_0": 0.45239, "loss_box_dn_0": 1.11489, "loss_cls_dn_1": 0.42413, "loss_box_dn_1": 2.42976, "loss_cls_dn_2": 0.47627, "loss_box_dn_2": 2.42834, "loss_cls_dn_3": 0.40764, "loss_box_dn_3": 2.42925, "loss_cls_dn_4": 0.4253, "loss_box_dn_4": 2.43414, "loss_cls_dn_5": 0.41287, "loss_box_dn_5": 2.4911, "loss_dense_depth": 1.37249, "loss": 46.399, "grad_norm": 65.02036, "time": 1.88972} -{"mode": "train", "epoch": 1, "iter": 13, "lr": 0.0001, "memory": 48982, "data_time": 0.05508, "loss_cls_0": 1.16964, "loss_box_0": 2.40444, "loss_cns_0": 0.5914, "loss_yns_0": 0.17429, "loss_cls_1": 1.16731, "loss_box_1": 3.01072, "loss_cns_1": 0.51505, "loss_yns_1": 0.17226, "loss_cls_2": 1.19225, "loss_box_2": 3.09755, "loss_cns_2": 0.49647, "loss_yns_2": 0.17797, "loss_cls_3": 1.17456, "loss_box_3": 3.07524, "loss_cns_3": 0.51909, "loss_yns_3": 0.18088, "loss_cls_4": 1.15, "loss_box_4": 3.20847, "loss_cns_4": 0.50973, "loss_yns_4": 0.17527, "loss_cls_5": 1.19479, "loss_box_5": 3.25918, "loss_cns_5": 0.51351, "loss_yns_5": 0.17507, "loss_cls_dn_0": 0.45064, "loss_box_dn_0": 1.1238, "loss_cls_dn_1": 0.4514, "loss_box_dn_1": 2.03456, "loss_cls_dn_2": 0.48551, "loss_box_dn_2": 2.07122, "loss_cls_dn_3": 0.42405, "loss_box_dn_3": 2.15044, "loss_cls_dn_4": 0.43707, "loss_box_dn_4": 2.31673, "loss_cls_dn_5": 0.44881, "loss_box_dn_5": 2.37606, "loss_dense_depth": 1.34606, "loss": 45.42147, "grad_norm": 83.79718, "time": 1.90853} -{"mode": "train", "epoch": 1, "iter": 14, "lr": 0.00011, "memory": 48982, "data_time": 0.05832, "loss_cls_0": 1.18373, "loss_box_0": 2.45292, "loss_cns_0": 0.60157, "loss_yns_0": 0.16247, "loss_cls_1": 1.19838, "loss_box_1": 2.90597, "loss_cns_1": 0.51994, "loss_yns_1": 0.16862, "loss_cls_2": 1.25939, "loss_box_2": 2.99634, "loss_cns_2": 0.52175, "loss_yns_2": 0.18511, "loss_cls_3": 1.21896, "loss_box_3": 2.90511, "loss_cns_3": 0.53559, "loss_yns_3": 0.17785, "loss_cls_4": 1.20189, "loss_box_4": 2.96082, "loss_cns_4": 0.53552, "loss_yns_4": 0.17538, "loss_cls_5": 1.22348, "loss_box_5": 2.93763, "loss_cns_5": 0.54208, "loss_yns_5": 0.16408, "loss_cls_dn_0": 0.4501, "loss_box_dn_0": 1.12391, "loss_cls_dn_1": 0.47021, "loss_box_dn_1": 1.69385, "loss_cls_dn_2": 0.48078, "loss_box_dn_2": 1.69498, "loss_cls_dn_3": 0.43332, "loss_box_dn_3": 1.78918, "loss_cls_dn_4": 0.43619, "loss_box_dn_4": 1.96111, "loss_cls_dn_5": 0.47965, "loss_box_dn_5": 2.00763, "loss_dense_depth": 1.33086, "loss": 43.08632, "grad_norm": 77.52726, "time": 1.91335} -{"mode": "train", "epoch": 1, "iter": 15, "lr": 0.00011, "memory": 48982, "data_time": 0.0558, "loss_cls_0": 1.13549, "loss_box_0": 2.40663, "loss_cns_0": 0.60151, "loss_yns_0": 0.16326, "loss_cls_1": 1.18679, "loss_box_1": 2.8821, "loss_cns_1": 0.52755, "loss_yns_1": 0.16683, "loss_cls_2": 1.29994, "loss_box_2": 2.82341, "loss_cns_2": 0.55037, "loss_yns_2": 0.18783, "loss_cls_3": 1.18032, "loss_box_3": 2.73504, "loss_cns_3": 0.53751, "loss_yns_3": 0.18267, "loss_cls_4": 1.21769, "loss_box_4": 2.80129, "loss_cns_4": 0.54538, "loss_yns_4": 0.16915, "loss_cls_5": 1.18109, "loss_box_5": 2.87321, "loss_cns_5": 0.55032, "loss_yns_5": 0.16829, "loss_cls_dn_0": 0.44811, "loss_box_dn_0": 1.12136, "loss_cls_dn_1": 0.4503, "loss_box_dn_1": 1.64649, "loss_cls_dn_2": 0.4425, "loss_box_dn_2": 1.58416, "loss_cls_dn_3": 0.41869, "loss_box_dn_3": 1.63203, "loss_cls_dn_4": 0.40991, "loss_box_dn_4": 1.73616, "loss_cls_dn_5": 0.45899, "loss_box_dn_5": 1.79951, "loss_dense_depth": 1.19672, "loss": 41.41863, "grad_norm": 74.19264, "time": 1.90238} -{"mode": "train", "epoch": 1, "iter": 16, "lr": 0.00011, "memory": 48982, "data_time": 0.05577, "loss_cls_0": 1.17423, "loss_box_0": 2.32014, "loss_cns_0": 0.59494, "loss_yns_0": 0.16432, "loss_cls_1": 1.2316, "loss_box_1": 2.86164, "loss_cns_1": 0.53719, "loss_yns_1": 0.17554, "loss_cls_2": 1.29804, "loss_box_2": 2.76607, "loss_cns_2": 0.53994, "loss_yns_2": 0.17347, "loss_cls_3": 1.22425, "loss_box_3": 2.74052, "loss_cns_3": 0.54333, "loss_yns_3": 0.17871, "loss_cls_4": 1.26559, "loss_box_4": 2.78149, "loss_cns_4": 0.55355, "loss_yns_4": 0.17182, "loss_cls_5": 1.2244, "loss_box_5": 2.89965, "loss_cns_5": 0.55263, "loss_yns_5": 0.17486, "loss_cls_dn_0": 0.46769, "loss_box_dn_0": 1.08327, "loss_cls_dn_1": 0.43334, "loss_box_dn_1": 1.65551, "loss_cls_dn_2": 0.43174, "loss_box_dn_2": 1.5642, "loss_cls_dn_3": 0.42818, "loss_box_dn_3": 1.60907, "loss_cls_dn_4": 0.3986, "loss_box_dn_4": 1.66047, "loss_cls_dn_5": 0.43848, "loss_box_dn_5": 1.74279, "loss_dense_depth": 1.30083, "loss": 41.36208, "grad_norm": 60.95555, "time": 1.91071} -{"mode": "train", "epoch": 1, "iter": 17, "lr": 0.00011, "memory": 48982, "data_time": 0.05935, "loss_cls_0": 1.14913, "loss_box_0": 2.335, "loss_cns_0": 0.58173, "loss_yns_0": 0.17132, "loss_cls_1": 1.23054, "loss_box_1": 2.89963, "loss_cns_1": 0.53872, "loss_yns_1": 0.17487, "loss_cls_2": 1.243, "loss_box_2": 2.81503, "loss_cns_2": 0.54294, "loss_yns_2": 0.18369, "loss_cls_3": 1.21405, "loss_box_3": 2.80375, "loss_cns_3": 0.53096, "loss_yns_3": 0.18359, "loss_cls_4": 1.24868, "loss_box_4": 2.85497, "loss_cns_4": 0.54034, "loss_yns_4": 0.1675, "loss_cls_5": 1.25329, "loss_box_5": 2.95232, "loss_cns_5": 0.54727, "loss_yns_5": 0.18, "loss_cls_dn_0": 0.45953, "loss_box_dn_0": 1.07624, "loss_cls_dn_1": 0.40404, "loss_box_dn_1": 1.69483, "loss_cls_dn_2": 0.42509, "loss_box_dn_2": 1.60449, "loss_cls_dn_3": 0.40991, "loss_box_dn_3": 1.67152, "loss_cls_dn_4": 0.3899, "loss_box_dn_4": 1.68795, "loss_cls_dn_5": 0.40919, "loss_box_dn_5": 1.79598, "loss_dense_depth": 1.22011, "loss": 41.5911, "grad_norm": 55.4273, "time": 1.91257} -{"mode": "train", "epoch": 1, "iter": 18, "lr": 0.00011, "memory": 48982, "data_time": 0.06021, "loss_cls_0": 1.15152, "loss_box_0": 2.29688, "loss_cns_0": 0.57777, "loss_yns_0": 0.17435, "loss_cls_1": 1.21276, "loss_box_1": 2.84427, "loss_cns_1": 0.53594, "loss_yns_1": 0.17065, "loss_cls_2": 1.20859, "loss_box_2": 2.80429, "loss_cns_2": 0.53519, "loss_yns_2": 0.17496, "loss_cls_3": 1.19937, "loss_box_3": 2.75398, "loss_cns_3": 0.5423, "loss_yns_3": 0.18226, "loss_cls_4": 1.21374, "loss_box_4": 2.80581, "loss_cns_4": 0.54422, "loss_yns_4": 0.17144, "loss_cls_5": 1.24097, "loss_box_5": 2.8629, "loss_cns_5": 0.55957, "loss_yns_5": 0.18344, "loss_cls_dn_0": 0.42692, "loss_box_dn_0": 1.06308, "loss_cls_dn_1": 0.39611, "loss_box_dn_1": 1.64878, "loss_cls_dn_2": 0.43203, "loss_box_dn_2": 1.61759, "loss_cls_dn_3": 0.38612, "loss_box_dn_3": 1.69684, "loss_cls_dn_4": 0.39587, "loss_box_dn_4": 1.7235, "loss_cls_dn_5": 0.3972, "loss_box_dn_5": 1.8718, "loss_dense_depth": 1.15607, "loss": 41.15906, "grad_norm": 60.70938, "time": 1.90912} -{"mode": "train", "epoch": 1, "iter": 19, "lr": 0.00011, "memory": 48982, "data_time": 0.0598, "loss_cls_0": 1.13979, "loss_box_0": 2.36298, "loss_cns_0": 0.54971, "loss_yns_0": 0.17695, "loss_cls_1": 1.20187, "loss_box_1": 3.03914, "loss_cns_1": 0.50832, "loss_yns_1": 0.18116, "loss_cls_2": 1.19901, "loss_box_2": 3.00836, "loss_cns_2": 0.5155, "loss_yns_2": 0.17297, "loss_cls_3": 1.21858, "loss_box_3": 3.0036, "loss_cns_3": 0.52521, "loss_yns_3": 0.17148, "loss_cls_4": 1.18606, "loss_box_4": 3.03441, "loss_cns_4": 0.51222, "loss_yns_4": 0.16475, "loss_cls_5": 1.20875, "loss_box_5": 3.09138, "loss_cns_5": 0.51667, "loss_yns_5": 0.17576, "loss_cls_dn_0": 0.42745, "loss_box_dn_0": 1.06273, "loss_cls_dn_1": 0.41798, "loss_box_dn_1": 1.41232, "loss_cls_dn_2": 0.45552, "loss_box_dn_2": 1.52236, "loss_cls_dn_3": 0.38819, "loss_box_dn_3": 1.6533, "loss_cls_dn_4": 0.4203, "loss_box_dn_4": 1.71187, "loss_cls_dn_5": 0.40749, "loss_box_dn_5": 1.86157, "loss_dense_depth": 1.11876, "loss": 41.72446, "grad_norm": 68.29885, "time": 1.90729} -{"mode": "train", "epoch": 1, "iter": 20, "lr": 0.00011, "memory": 48982, "data_time": 0.05366, "loss_cls_0": 1.1395, "loss_box_0": 2.39819, "loss_cns_0": 0.54889, "loss_yns_0": 0.17974, "loss_cls_1": 1.20386, "loss_box_1": 2.90531, "loss_cns_1": 0.5224, "loss_yns_1": 0.19182, "loss_cls_2": 1.20994, "loss_box_2": 2.88343, "loss_cns_2": 0.54654, "loss_yns_2": 0.17541, "loss_cls_3": 1.22661, "loss_box_3": 2.85722, "loss_cns_3": 0.54758, "loss_yns_3": 0.17904, "loss_cls_4": 1.17476, "loss_box_4": 2.87158, "loss_cns_4": 0.55206, "loss_yns_4": 0.16978, "loss_cls_5": 1.22893, "loss_box_5": 2.85895, "loss_cns_5": 0.53498, "loss_yns_5": 0.16754, "loss_cls_dn_0": 0.45894, "loss_box_dn_0": 1.0327, "loss_cls_dn_1": 0.41181, "loss_box_dn_1": 1.65262, "loss_cls_dn_2": 0.43453, "loss_box_dn_2": 1.75798, "loss_cls_dn_3": 0.37827, "loss_box_dn_3": 1.83323, "loss_cls_dn_4": 0.40476, "loss_box_dn_4": 1.853, "loss_cls_dn_5": 0.39325, "loss_box_dn_5": 1.94239, "loss_dense_depth": 1.16801, "loss": 41.99556, "grad_norm": 58.12973, "time": 1.9002} -{"mode": "train", "epoch": 1, "iter": 21, "lr": 0.00011, "memory": 48982, "data_time": 0.16549, "loss_cls_0": 1.14517, "loss_box_0": 2.36011, "loss_cns_0": 0.56381, "loss_yns_0": 0.17572, "loss_cls_1": 1.21614, "loss_box_1": 2.62805, "loss_cns_1": 0.57213, "loss_yns_1": 0.18862, "loss_cls_2": 1.22543, "loss_box_2": 2.66675, "loss_cns_2": 0.57197, "loss_yns_2": 0.17445, "loss_cls_3": 1.21641, "loss_box_3": 2.61041, "loss_cns_3": 0.5625, "loss_yns_3": 0.17328, "loss_cls_4": 1.1881, "loss_box_4": 2.62797, "loss_cns_4": 0.58703, "loss_yns_4": 0.17088, "loss_cls_5": 1.24543, "loss_box_5": 2.6534, "loss_cns_5": 0.56863, "loss_yns_5": 0.16752, "loss_cls_dn_0": 0.45988, "loss_box_dn_0": 1.05061, "loss_cls_dn_1": 0.41216, "loss_box_dn_1": 1.35809, "loss_cls_dn_2": 0.41811, "loss_box_dn_2": 1.42652, "loss_cls_dn_3": 0.39006, "loss_box_dn_3": 1.44306, "loss_cls_dn_4": 0.40328, "loss_box_dn_4": 1.44545, "loss_cls_dn_5": 0.40357, "loss_box_dn_5": 1.48812, "loss_dense_depth": 1.09329, "loss": 39.05211, "grad_norm": 50.80311, "time": 2.02169} -{"mode": "train", "epoch": 1, "iter": 22, "lr": 0.00011, "memory": 48982, "data_time": 0.15293, "loss_cls_0": 1.13126, "loss_box_0": 2.27377, "loss_cns_0": 0.59382, "loss_yns_0": 0.17832, "loss_cls_1": 1.21269, "loss_box_1": 2.47296, "loss_cns_1": 0.61697, "loss_yns_1": 0.17898, "loss_cls_2": 1.20762, "loss_box_2": 2.49702, "loss_cns_2": 0.59172, "loss_yns_2": 0.17858, "loss_cls_3": 1.18665, "loss_box_3": 2.49655, "loss_cns_3": 0.58602, "loss_yns_3": 0.17066, "loss_cls_4": 1.18973, "loss_box_4": 2.4767, "loss_cns_4": 0.6131, "loss_yns_4": 0.16974, "loss_cls_5": 1.20902, "loss_box_5": 2.51787, "loss_cns_5": 0.59695, "loss_yns_5": 0.1729, "loss_cls_dn_0": 0.41599, "loss_box_dn_0": 1.0353, "loss_cls_dn_1": 0.39669, "loss_box_dn_1": 1.14788, "loss_cls_dn_2": 0.39392, "loss_box_dn_2": 1.16517, "loss_cls_dn_3": 0.39999, "loss_box_dn_3": 1.18085, "loss_cls_dn_4": 0.39427, "loss_box_dn_4": 1.1992, "loss_cls_dn_5": 0.4154, "loss_box_dn_5": 1.20865, "loss_dense_depth": 1.11498, "loss": 36.98787, "grad_norm": 48.84113, "time": 5.99743} diff --git a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20260210_162616.log b/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20260210_162616.log deleted file mode 100644 index f8e2eab774c319c81f8f96bd29ddddc208946f29..0000000000000000000000000000000000000000 --- a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20260210_162616.log +++ /dev/null @@ -1,3231 +0,0 @@ -2026-02-10 16:26:16,270 - mmdet - INFO - Environment info: ------------------------------------------------------------- -sys.platform: linux -Python: 3.10.12 (main, Aug 15 2025, 14:32:43) [GCC 11.4.0] -CUDA available: True -GPU 0,1,2,3,4,5,6,7: BW200_LC, UBB BW1000_LC -CUDA_HOME: /opt/dtk -NVCC: Not Available -GCC: x86_64-linux-gnu-gcc (Ubuntu 11.4.0-1ubuntu1~22.04) 11.4.0 -PyTorch: 2.5.1 -PyTorch compiling details: PyTorch built with: - - GCC 10.3 - - C++ Version: 201703 - - Intel(R) Math Kernel Library Version 2020.0.4 Product Build 20200917 for Intel(R) 64 architecture applications - - OpenMP 201511 (a.k.a. OpenMP 4.5) - - LAPACK is enabled (usually provided by MKL) - - NNPACK is enabled - - CPU capability usage: AVX2 - - HIP Runtime 6.3.25521 - - MIOpen 2.18.0 - - Magma 2.8.0 - - Build settings: BLAS_INFO=mkl, BUILD_TYPE=Release, CXX_COMPILER=/opt/rh/gcc-toolset-10/root/usr/bin/c++, CXX_FLAGS= -D_GLIBCXX_USE_CXX11_ABI=1 -fvisibility-inlines-hidden -DUSE_PTHREADPOOL -DNDEBUG -DUSE_KINETO -DLIBKINETO_NOCUPTI -DLIBKINETO_NOXPUPTI=ON -DUSE_FBGEMM -DUSE_PYTORCH_QNNPACK -DUSE_XNNPACK -DSYMBOLICATE_MOBILE_DEBUG_HANDLE -O2 -fPIC -Wall -Wextra -Werror=return-type -Werror=non-virtual-dtor -Werror=bool-operation -Wnarrowing -Wno-missing-field-initializers -Wno-type-limits -Wno-array-bounds -Wno-unknown-pragmas -Wno-unused-parameter -Wno-strict-overflow -Wno-strict-aliasing -Wno-stringop-overflow -Wsuggest-override -Wno-psabi -Wno-error=old-style-cast -Wno-missing-braces -fdiagnostics-color=always -faligned-new -Wno-unused-but-set-variable -Wno-maybe-uninitialized -fno-math-errno -fno-trapping-math -Werror=format -Wno-stringop-overflow, FORCE_FALLBACK_CUDA_MPI=1, LAPACK_INFO=mkl, PERF_WITH_AVX=1, PERF_WITH_AVX2=1, TORCH_VERSION=2.5.1, USE_CUDA=0, USE_CUDNN=OFF, USE_CUSPARSELT=OFF, USE_EXCEPTION_PTR=1, USE_GFLAGS=1, USE_GLOG=1, USE_GLOO=1, USE_MKL=ON, USE_MKLDNN=0, USE_MPI=1, USE_NCCL=1, USE_NNPACK=ON, USE_OPENMP=1, USE_ROCM=ON, USE_ROCM_KERNEL_ASSERT=OFF, - -TorchVision: 0.20.1 -OpenCV: 4.12.0 -MMCV: 1.6.1 -MMCV Compiler: GCC 10.3 -MMCV CUDA Compiler: rocm not available -MMDetection: 2.25.1+ ------------------------------------------------------------- - -2026-02-10 16:26:17,235 - mmdet - INFO - Distributed training: False -2026-02-10 16:26:18,198 - mmdet - INFO - Config: -plugin = True -plugin_dir = 'projects/mmdet3d_plugin/' -dist_params = dict(backend='nccl') -log_level = 'INFO' -work_dir = './work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704' -total_batch_size = 80 -num_gpus = 8 -batch_size = 10 -num_iters_per_epoch = 351 -num_epochs = 100 -checkpoint_epoch_interval = 20 -checkpoint_config = dict(interval=7020) -log_config = dict( - interval=1, - hooks=[ - dict(type='TextLoggerHook', by_epoch=False), - dict(type='TensorboardLoggerHook') - ]) -load_from = None -resume_from = None -workflow = [('train', 1)] -fp16 = dict(loss_scale=32.0) -input_shape = (704, 256) -tracking_test = True -tracking_threshold = 0.2 -class_names = [ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier', - 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' -] -num_classes = 10 -embed_dims = 256 -num_groups = 8 -num_decoder = 6 -num_single_frame_decoder = 1 -use_deformable_func = True -strides = [4, 8, 16, 32] -num_levels = 4 -num_depth_layers = 3 -drop_out = 0.1 -temporal = True -decouple_attn = True -with_quality_estimation = True -model = dict( - type='Sparse4D', - use_grid_mask=True, - use_deformable_func=True, - img_backbone=dict( - type='ResNet', - depth=50, - num_stages=4, - frozen_stages=-1, - norm_eval=False, - style='pytorch', - with_cp=True, - out_indices=(0, 1, 2, 3), - norm_cfg=dict(type='BN', requires_grad=True), - pretrained='ckpt/resnet50-19c8e357.pth'), - img_neck=dict( - type='FPN', - num_outs=4, - start_level=0, - out_channels=256, - add_extra_convs='on_output', - relu_before_extra_convs=True, - in_channels=[256, 512, 1024, 2048]), - depth_branch=dict( - type='DenseDepthNet', - embed_dims=256, - num_depth_layers=3, - loss_weight=0.2), - head=dict( - type='Sparse4DHead', - cls_threshold_to_reg=0.05, - decouple_attn=True, - instance_bank=dict( - type='InstanceBank', - num_anchor=900, - embed_dims=256, - anchor='nuscenes_kmeans900.npy', - anchor_handler=dict(type='SparseBox3DKeyPointsGenerator'), - num_temp_instances=600, - confidence_decay=0.6, - feat_grad=False), - anchor_encoder=dict( - type='SparseBox3DEncoder', - vel_dims=3, - embed_dims=[128, 32, 32, 64], - mode='cat', - output_fc=False, - in_loops=1, - out_loops=4), - num_single_frame_decoder=1, - operation_order=[ - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine' - ], - temp_graph_model=dict( - type='MultiheadAttention', - embed_dims=512, - num_heads=8, - batch_first=True, - dropout=0.1), - graph_model=dict( - type='MultiheadAttention', - embed_dims=512, - num_heads=8, - batch_first=True, - dropout=0.1), - norm_layer=dict(type='LN', normalized_shape=256), - ffn=dict( - type='AsymmetricFFN', - in_channels=512, - pre_norm=dict(type='LN'), - embed_dims=256, - feedforward_channels=1024, - num_fcs=2, - ffn_drop=0.1, - act_cfg=dict(type='ReLU', inplace=True)), - deformable_model=dict( - type='DeformableFeatureAggregation', - embed_dims=256, - num_groups=8, - num_levels=4, - num_cams=6, - attn_drop=0.15, - use_deformable_func=True, - use_camera_embed=True, - residual_mode='cat', - kps_generator=dict( - type='SparseBox3DKeyPointsGenerator', - num_learnable_pts=6, - fix_scale=[[0, 0, 0], [0.45, 0, 0], [-0.45, 0, 0], - [0, 0.45, 0], [0, -0.45, 0], [0, 0, 0.45], - [0, 0, -0.45]])), - refine_layer=dict( - type='SparseBox3DRefinementModule', - embed_dims=256, - num_cls=10, - refine_yaw=True, - with_quality_estimation=True), - sampler=dict( - type='SparseBox3DTarget', - num_dn_groups=5, - num_temp_dn_groups=3, - dn_noise_scale=[2.0, 2.0, 2.0, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5], - max_dn_gt=32, - add_neg_dn=True, - cls_weight=2.0, - box_weight=0.25, - reg_weights=[2.0, 2.0, 2.0, 0.5, 0.5, 0.5, 0.0, 0.0, 0.0, 0.0], - cls_wise_reg_weights=dict( - {9: [2.0, 2.0, 2.0, 1.0, 1.0, 1.0, 0.0, 0.0, 1.0, 1.0]})), - loss_cls=dict( - type='FocalLoss', - use_sigmoid=True, - gamma=2.0, - alpha=0.25, - loss_weight=2.0), - loss_reg=dict( - type='SparseBox3DLoss', - loss_box=dict(type='L1Loss', loss_weight=0.25), - loss_centerness=dict(type='CrossEntropyLoss', use_sigmoid=True), - loss_yawness=dict(type='GaussianFocalLoss'), - cls_allow_reverse=[5]), - decoder=dict(type='SparseBox3DDecoder'), - reg_weights=[2.0, 2.0, 2.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0])) -dataset_type = 'NuScenes3DDetTrackDataset' -data_root = 'data/nuscenes/' -anno_root = 'data/nuscenes_anno_pkls/' -file_client_args = dict(backend='disk') -img_norm_cfg = dict( - mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375], to_rgb=True) -train_pipeline = [ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict( - type='LoadPointsFromFile', - coord_type='LIDAR', - load_dim=5, - use_dim=5, - file_client_args=dict(backend='disk')), - dict(type='ResizeCropFlipImage'), - dict(type='MultiScaleDepthMapGenerator', downsample=[4, 8, 16]), - dict(type='BBoxRotation'), - dict(type='PhotoMetricDistortionMultiViewImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict( - type='CircleObjectRangeFilter', - class_dist_thred=[55, 55, 55, 55, 55, 55, 55, 55, 55, 55]), - dict( - type='InstanceNameFilter', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ]), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=[ - 'img', 'timestamp', 'projection_mat', 'image_wh', 'gt_depth', - 'focal', 'gt_bboxes_3d', 'gt_labels_3d' - ], - meta_keys=['T_global', 'T_global_inv', 'timestamp', 'instance_id']) -] -test_pipeline = [ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='ResizeCropFlipImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=['img', 'timestamp', 'projection_mat', 'image_wh'], - meta_keys=['T_global', 'T_global_inv', 'timestamp']) -] -input_modality = dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False) -data_basic_config = dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier', - 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval') -data_aug_conf = dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]) -data = dict( - samples_per_gpu=10, - workers_per_gpu=10, - train=dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval', - ann_file='data/nuscenes_anno_pkls/nuscenes_infos_train.pkl', - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict( - type='LoadPointsFromFile', - coord_type='LIDAR', - load_dim=5, - use_dim=5, - file_client_args=dict(backend='disk')), - dict(type='ResizeCropFlipImage'), - dict(type='MultiScaleDepthMapGenerator', downsample=[4, 8, 16]), - dict(type='BBoxRotation'), - dict(type='PhotoMetricDistortionMultiViewImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict( - type='CircleObjectRangeFilter', - class_dist_thred=[55, 55, 55, 55, 55, 55, 55, 55, 55, 55]), - dict( - type='InstanceNameFilter', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', - 'traffic_cone' - ]), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=[ - 'img', 'timestamp', 'projection_mat', 'image_wh', - 'gt_depth', 'focal', 'gt_bboxes_3d', 'gt_labels_3d' - ], - meta_keys=[ - 'T_global', 'T_global_inv', 'timestamp', 'instance_id' - ]) - ], - test_mode=False, - data_aug_conf=dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]), - with_seq_flag=True, - sequences_split_num=2, - keep_consistent_seq_aug=True), - val=dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval', - ann_file='data/nuscenes_anno_pkls/nuscenes_infos_val.pkl', - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='ResizeCropFlipImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=['img', 'timestamp', 'projection_mat', 'image_wh'], - meta_keys=['T_global', 'T_global_inv', 'timestamp']) - ], - data_aug_conf=dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]), - test_mode=True, - tracking=True, - tracking_threshold=0.2), - test=dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval', - ann_file='data/nuscenes_anno_pkls/nuscenes_infos_val.pkl', - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='ResizeCropFlipImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=['img', 'timestamp', 'projection_mat', 'image_wh'], - meta_keys=['T_global', 'T_global_inv', 'timestamp']) - ], - data_aug_conf=dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]), - test_mode=True, - tracking=True, - tracking_threshold=0.2)) -optimizer = dict( - type='AdamW', - lr=0.0006, - weight_decay=0.001, - paramwise_cfg=dict(custom_keys=dict(img_backbone=dict(lr_mult=0.5)))) -optimizer_config = dict(grad_clip=dict(max_norm=25, norm_type=2)) -lr_config = dict( - policy='CosineAnnealing', - warmup='linear', - warmup_iters=500, - warmup_ratio=0.3333333333333333, - min_lr_ratio=0.001) -runner = dict(type='IterBasedRunner', max_iters=35100) -vis_pipeline = [ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='Collect', keys=['img'], meta_keys=['timestamp', 'lidar2img']) -] -evaluation = dict( - interval=7020, - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict( - type='Collect', keys=['img'], meta_keys=['timestamp', 'lidar2img']) - ]) -gpu_ids = range(0, 1) - -2026-02-10 16:26:18,198 - mmdet - INFO - Set random seed to 0, deterministic: False -2026-02-10 16:26:18,797 - mmdet - INFO - initialize ResNet with init_cfg {'type': 'Pretrained', 'checkpoint': 'ckpt/resnet50-19c8e357.pth'} -2026-02-10 16:26:19,024 - mmdet - INFO - initialize FPN with init_cfg {'type': 'Xavier', 'layer': 'Conv2d', 'distribution': 'uniform'} -Name of parameter - Initialization information - -img_backbone.conv1.weight - torch.Size([64, 3, 7, 7]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.bn1.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.bn1.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.conv1.weight - torch.Size([64, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn1.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn1.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.conv2.weight - torch.Size([64, 64, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn2.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn2.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.conv3.weight - torch.Size([256, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn3.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn3.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.downsample.0.weight - torch.Size([256, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.downsample.1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.downsample.1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.conv1.weight - torch.Size([64, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn1.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn1.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.conv2.weight - torch.Size([64, 64, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn2.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn2.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.conv3.weight - torch.Size([256, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn3.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn3.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.conv1.weight - torch.Size([64, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn1.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn1.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.conv2.weight - torch.Size([64, 64, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn2.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn2.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.conv3.weight - torch.Size([256, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn3.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn3.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.conv1.weight - torch.Size([128, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn1.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn1.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.conv2.weight - torch.Size([128, 128, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn2.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn2.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.conv3.weight - torch.Size([512, 128, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn3.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn3.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.downsample.0.weight - torch.Size([512, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.downsample.1.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.downsample.1.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.conv1.weight - torch.Size([128, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn1.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn1.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.conv2.weight - torch.Size([128, 128, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn2.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn2.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.conv3.weight - torch.Size([512, 128, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn3.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn3.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.conv1.weight - torch.Size([128, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn1.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn1.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.conv2.weight - torch.Size([128, 128, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn2.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn2.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.conv3.weight - torch.Size([512, 128, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn3.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn3.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.conv1.weight - torch.Size([128, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn1.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn1.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.conv2.weight - torch.Size([128, 128, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn2.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn2.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.conv3.weight - torch.Size([512, 128, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn3.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn3.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.conv1.weight - torch.Size([256, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.downsample.0.weight - torch.Size([1024, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.downsample.1.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.downsample.1.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.conv1.weight - torch.Size([512, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn1.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn1.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.conv2.weight - torch.Size([512, 512, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn2.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn2.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.conv3.weight - torch.Size([2048, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn3.weight - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn3.bias - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.downsample.0.weight - torch.Size([2048, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.downsample.1.weight - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.downsample.1.bias - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.conv1.weight - torch.Size([512, 2048, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn1.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn1.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.conv2.weight - torch.Size([512, 512, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn2.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn2.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.conv3.weight - torch.Size([2048, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn3.weight - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn3.bias - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.conv1.weight - torch.Size([512, 2048, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn1.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn1.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.conv2.weight - torch.Size([512, 512, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn2.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn2.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.conv3.weight - torch.Size([2048, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn3.weight - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn3.bias - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_neck.lateral_convs.0.conv.weight - torch.Size([256, 256, 1, 1]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.lateral_convs.0.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.lateral_convs.1.conv.weight - torch.Size([256, 512, 1, 1]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.lateral_convs.1.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.lateral_convs.2.conv.weight - torch.Size([256, 1024, 1, 1]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.lateral_convs.2.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.lateral_convs.3.conv.weight - torch.Size([256, 2048, 1, 1]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.lateral_convs.3.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.fpn_convs.0.conv.weight - torch.Size([256, 256, 3, 3]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.fpn_convs.0.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.fpn_convs.1.conv.weight - torch.Size([256, 256, 3, 3]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.fpn_convs.1.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.fpn_convs.2.conv.weight - torch.Size([256, 256, 3, 3]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.fpn_convs.2.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.fpn_convs.3.conv.weight - torch.Size([256, 256, 3, 3]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.fpn_convs.3.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.instance_bank.anchor - torch.Size([900, 11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.instance_bank.instance_feature - torch.Size([900, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.instance_bank.anchor_handler.fix_scale - torch.Size([1, 3]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.0.weight - torch.Size([128, 3]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.0.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.2.weight - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.2.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.3.weight - torch.Size([128, 128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.3.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.5.weight - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.5.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.6.weight - torch.Size([128, 128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.6.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.8.weight - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.8.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.9.weight - torch.Size([128, 128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.9.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.11.weight - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.11.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.0.weight - torch.Size([32, 3]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.0.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.2.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.2.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.3.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.3.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.5.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.5.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.6.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.6.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.8.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.8.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.9.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.9.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.11.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.11.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.0.weight - torch.Size([32, 2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.0.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.2.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.2.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.3.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.3.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.5.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.5.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.6.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.6.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.8.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.8.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.9.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.9.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.11.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.11.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.0.weight - torch.Size([64, 3]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.0.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.2.weight - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.2.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.3.weight - torch.Size([64, 64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.3.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.5.weight - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.5.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.6.weight - torch.Size([64, 64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.6.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.8.weight - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.8.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.9.weight - torch.Size([64, 64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.9.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.11.weight - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.11.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.1.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.1.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.1.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.1.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.1.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.1.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.1.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.1.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.3.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.4.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.4.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.4.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.4.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.5.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.5.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.5.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.5.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.6.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.6.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.8.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.8.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.8.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.8.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.8.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.8.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.8.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.8.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.10.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.11.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.11.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.11.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.11.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.12.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.12.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.12.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.12.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.13.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.13.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.15.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.15.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.15.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.15.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.15.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.15.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.15.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.15.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.16.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.16.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.17.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.18.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.18.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.18.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.18.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.19.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.19.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.19.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.19.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.20.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.20.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.22.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.22.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.22.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.22.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.22.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.22.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.22.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.22.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.23.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.23.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.24.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.25.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.25.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.25.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.25.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.26.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.26.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.26.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.26.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.27.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.27.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.29.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.29.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.29.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.29.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.29.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.29.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.29.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.29.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.30.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.30.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.31.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.32.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.32.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.32.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.32.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.33.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.33.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.33.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.33.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.34.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.34.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.36.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.36.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.36.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.36.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.36.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.36.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.36.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.36.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.37.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.37.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.38.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.fc_before.weight - torch.Size([512, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.fc_after.weight - torch.Size([256, 512]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.0.weight - torch.Size([1, 256, 1, 1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.0.bias - torch.Size([1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.1.weight - torch.Size([1, 256, 1, 1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.1.bias - torch.Size([1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.2.weight - torch.Size([1, 256, 1, 1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.2.bias - torch.Size([1]): -The value is the same before and after calling `init_weights` of Sparse4D -2026-02-10 16:26:19,401 - mmdet - INFO - Model: -Sparse4D( - (img_backbone): ResNet( - (conv1): Conv2d(3, 64, kernel_size=(7, 7), stride=(2, 2), padding=(3, 3), bias=False) - (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (maxpool): MaxPool2d(kernel_size=3, stride=2, padding=1, dilation=1, ceil_mode=False) - (layer1): ResLayer( - (0): Bottleneck( - (conv1): Conv2d(64, 64, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (downsample): Sequential( - (0): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - ) - ) - (1): Bottleneck( - (conv1): Conv2d(256, 64, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (2): Bottleneck( - (conv1): Conv2d(256, 64, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - ) - (layer2): ResLayer( - (0): Bottleneck( - (conv1): Conv2d(256, 128, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (downsample): Sequential( - (0): Conv2d(256, 512, kernel_size=(1, 1), stride=(2, 2), bias=False) - (1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - ) - ) - (1): Bottleneck( - (conv1): Conv2d(512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (2): Bottleneck( - (conv1): Conv2d(512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (3): Bottleneck( - (conv1): Conv2d(512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - ) - (layer3): ResLayer( - (0): Bottleneck( - (conv1): Conv2d(512, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (downsample): Sequential( - (0): Conv2d(512, 1024, kernel_size=(1, 1), stride=(2, 2), bias=False) - (1): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - ) - ) - (1): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (2): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (3): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (4): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (5): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - ) - (layer4): ResLayer( - (0): Bottleneck( - (conv1): Conv2d(1024, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(512, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (downsample): Sequential( - (0): Conv2d(1024, 2048, kernel_size=(1, 1), stride=(2, 2), bias=False) - (1): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - ) - ) - (1): Bottleneck( - (conv1): Conv2d(2048, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(512, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (2): Bottleneck( - (conv1): Conv2d(2048, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(512, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - ) - ) - init_cfg={'type': 'Pretrained', 'checkpoint': 'ckpt/resnet50-19c8e357.pth'} - (img_neck): FPN( - (lateral_convs): ModuleList( - (0): ConvModule( - (conv): Conv2d(256, 256, kernel_size=(1, 1), stride=(1, 1)) - ) - (1): ConvModule( - (conv): Conv2d(512, 256, kernel_size=(1, 1), stride=(1, 1)) - ) - (2): ConvModule( - (conv): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1)) - ) - (3): ConvModule( - (conv): Conv2d(2048, 256, kernel_size=(1, 1), stride=(1, 1)) - ) - ) - (fpn_convs): ModuleList( - (0-3): 4 x ConvModule( - (conv): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1)) - ) - ) - ) - init_cfg={'type': 'Xavier', 'layer': 'Conv2d', 'distribution': 'uniform'} - (head): Sparse4DHead( - (instance_bank): InstanceBank( - (anchor_handler): SparseBox3DKeyPointsGenerator() - ) - (anchor_encoder): SparseBox3DEncoder( - (pos_fc): Sequential( - (0): Linear(in_features=3, out_features=128, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((128,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=128, out_features=128, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((128,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=128, out_features=128, bias=True) - (7): ReLU(inplace=True) - (8): LayerNorm((128,), eps=1e-05, elementwise_affine=True) - (9): Linear(in_features=128, out_features=128, bias=True) - (10): ReLU(inplace=True) - (11): LayerNorm((128,), eps=1e-05, elementwise_affine=True) - ) - (size_fc): Sequential( - (0): Linear(in_features=3, out_features=32, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=32, out_features=32, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=32, out_features=32, bias=True) - (7): ReLU(inplace=True) - (8): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (9): Linear(in_features=32, out_features=32, bias=True) - (10): ReLU(inplace=True) - (11): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - ) - (yaw_fc): Sequential( - (0): Linear(in_features=2, out_features=32, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=32, out_features=32, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=32, out_features=32, bias=True) - (7): ReLU(inplace=True) - (8): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (9): Linear(in_features=32, out_features=32, bias=True) - (10): ReLU(inplace=True) - (11): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - ) - (vel_fc): Sequential( - (0): Linear(in_features=3, out_features=64, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((64,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=64, out_features=64, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((64,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=64, out_features=64, bias=True) - (7): ReLU(inplace=True) - (8): LayerNorm((64,), eps=1e-05, elementwise_affine=True) - (9): Linear(in_features=64, out_features=64, bias=True) - (10): ReLU(inplace=True) - (11): LayerNorm((64,), eps=1e-05, elementwise_affine=True) - ) - ) - (loss_cls): FocalLoss() - (loss_reg): SparseBox3DLoss( - (loss_box): L1Loss() - (loss_cns): CrossEntropyLoss(avg_non_ignore=False) - (loss_yns): GaussianFocalLoss() - ) - (layers): ModuleList( - (0): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (1): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (4-5): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (6): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (7): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (8): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (11-12): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (13): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (14): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (15): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (16): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (17): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (18-19): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (20): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (21): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (22): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (23): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (24): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (25-26): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (27): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (28): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (29): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (30): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (31): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (32-33): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (34): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (35): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (36): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (37): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (38): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - ) - (fc_before): Linear(in_features=256, out_features=512, bias=False) - (fc_after): Linear(in_features=512, out_features=256, bias=False) - ) - (depth_branch): DenseDepthNet( - (depth_layers): ModuleList( - (0-2): 3 x Conv2d(256, 1, kernel_size=(1, 1), stride=(1, 1)) - ) - ) - (grid_mask): GridMask() -) -2026-02-10 16:26:33,227 - mmdet - INFO - Start running, host: root@bw28, work_dir: /workspace/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704 -2026-02-10 16:26:33,227 - mmdet - INFO - Hooks will be executed in the following order: -before_run: -(VERY_HIGH ) CosineAnnealingLrUpdaterHook -(ABOVE_NORMAL) Fp16OptimizerHook -(NORMAL ) CheckpointHook -(NORMAL ) EvalHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -before_train_epoch: -(VERY_HIGH ) CosineAnnealingLrUpdaterHook -(NORMAL ) EvalHook -(LOW ) IterTimerHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -before_train_iter: -(VERY_HIGH ) CosineAnnealingLrUpdaterHook -(NORMAL ) EvalHook -(LOW ) IterTimerHook - -------------------- -after_train_iter: -(ABOVE_NORMAL) Fp16OptimizerHook -(NORMAL ) CheckpointHook -(NORMAL ) EvalHook -(NORMAL ) ProfilerHook -(LOW ) IterTimerHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -after_train_epoch: -(NORMAL ) CheckpointHook -(NORMAL ) EvalHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -before_val_epoch: -(LOW ) IterTimerHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -before_val_iter: -(LOW ) IterTimerHook - -------------------- -after_val_iter: -(LOW ) IterTimerHook - -------------------- -after_val_epoch: -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -after_run: -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -2026-02-10 16:26:33,228 - mmdet - INFO - workflow: [('train', 1)], max: 35100 iters -2026-02-10 16:26:33,230 - mmdet - INFO - Checkpoints will be saved to /workspace/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704 by HardDiskBackend. -2026-02-10 16:26:55,799 - mmdet - INFO - Iter [1/35100] lr: 1.000e-04, eta: 8 days, 22:19:22, time: 21.982, data_time: 3.919, memory: 24611, loss_cls_0: 2.3066, loss_box_0: 0.0000, loss_cns_0: 0.0000, loss_yns_0: 0.0000, loss_cls_1: 2.1660, loss_box_1: 0.1165, loss_cns_1: 0.0237, loss_yns_1: 0.0049, loss_cls_2: 2.2464, loss_box_2: 0.0000, loss_cns_2: 0.0000, loss_yns_2: 0.0000, loss_cls_3: 2.3115, loss_box_3: 0.0524, loss_cns_3: 0.0085, loss_yns_3: 0.0008, loss_cls_4: 1.9817, loss_box_4: 0.3494, loss_cns_4: 0.0475, loss_yns_4: 0.0321, loss_cls_5: 2.3490, loss_box_5: 0.0000, loss_cns_5: 0.0000, loss_yns_5: 0.0000, loss_cls_dn_0: 1.1647, loss_box_dn_0: 1.4583, loss_cls_dn_1: 1.1569, loss_box_dn_1: 1.7270, loss_cls_dn_2: 1.1336, loss_box_dn_2: 1.9434, loss_cls_dn_3: 1.1271, loss_box_dn_3: 2.2150, loss_cls_dn_4: 1.0498, loss_box_dn_4: 2.4035, loss_cls_dn_5: 1.1831, loss_box_dn_5: 2.6805, loss_dense_depth: 1.8309, loss: 35.0708, grad_norm: 552.7686 -2026-02-10 16:26:57,953 - mmdet - INFO - Iter [2/35100] lr: 1.004e-04, eta: 4 days, 21:39:31, time: 2.154, data_time: 0.038, memory: 24611, loss_cls_0: 2.1843, loss_box_0: 0.0000, loss_cns_0: 0.0000, loss_yns_0: 0.0000, loss_cls_1: 1.9559, loss_box_1: 0.5730, loss_cns_1: 0.0725, loss_yns_1: 0.0185, loss_cls_2: 2.0600, loss_box_2: 0.0000, loss_cns_2: 0.0000, loss_yns_2: 0.0000, loss_cls_3: 1.9910, loss_box_3: 0.1840, loss_cns_3: 0.0266, loss_yns_3: 0.0103, loss_cls_4: 1.7822, loss_box_4: 1.4303, loss_cns_4: 0.1507, loss_yns_4: 0.0449, loss_cls_5: 1.9488, loss_box_5: 1.3291, loss_cns_5: 0.1652, loss_yns_5: 0.0664, loss_cls_dn_0: 1.1055, loss_box_dn_0: 1.2615, loss_cls_dn_1: 0.9639, loss_box_dn_1: 2.5890, loss_cls_dn_2: 0.9461, loss_box_dn_2: 2.6759, loss_cls_dn_3: 0.9394, loss_box_dn_3: 2.7226, loss_cls_dn_4: 0.8565, loss_box_dn_4: 2.8931, loss_cls_dn_5: 0.9667, loss_box_dn_5: 3.1169, loss_dense_depth: 1.7817, loss: 38.8125, grad_norm: 82.9143 -2026-02-10 16:26:59,059 - mmdet - INFO - Iter [3/35100] lr: 1.008e-04, eta: 3 days, 10:01:50, time: 1.106, data_time: 0.038, memory: 24611, loss_cls_0: 1.6642, loss_box_0: 2.2488, loss_cns_0: 0.6734, loss_yns_0: 0.5446, loss_cls_1: 1.7714, loss_box_1: 1.7659, loss_cns_1: 0.2835, loss_yns_1: 0.0799, loss_cls_2: 1.7267, loss_box_2: 2.6875, loss_cns_2: 0.3651, loss_yns_2: 0.1088, loss_cls_3: 1.6372, loss_box_3: 3.3641, loss_cns_3: 0.3806, loss_yns_3: 0.2207, loss_cls_4: 1.5800, loss_box_4: 4.0215, loss_cns_4: 0.4482, loss_yns_4: 0.1288, loss_cls_5: 1.6133, loss_box_5: 3.8723, loss_cns_5: 0.3336, loss_yns_5: 0.1544, loss_cls_dn_0: 0.8230, loss_box_dn_0: 1.1693, loss_cls_dn_1: 0.8186, loss_box_dn_1: 2.3477, loss_cls_dn_2: 0.7843, loss_box_dn_2: 2.5177, loss_cls_dn_3: 0.7234, loss_box_dn_3: 2.7278, loss_cls_dn_4: 0.6981, loss_box_dn_4: 2.8969, loss_cls_dn_5: 0.7729, loss_box_dn_5: 3.1261, loss_dense_depth: 1.7790, loss: 52.8592, grad_norm: 132.7128 -2026-02-10 16:27:00,172 - mmdet - INFO - Iter [4/35100] lr: 1.012e-04, eta: 2 days, 16:14:01, time: 1.113, data_time: 0.039, memory: 24611, loss_cls_0: 1.4907, loss_box_0: 2.9416, loss_cns_0: 0.6722, loss_yns_0: 0.2949, loss_cls_1: 1.6427, loss_box_1: 2.6664, loss_cns_1: 0.3880, loss_yns_1: 0.1396, loss_cls_2: 1.6195, loss_box_2: 3.7900, loss_cns_2: 0.4975, loss_yns_2: 0.1808, loss_cls_3: 1.4945, loss_box_3: 4.0943, loss_cns_3: 0.4754, loss_yns_3: 0.2364, loss_cls_4: 1.4486, loss_box_4: 4.6173, loss_cns_4: 0.3862, loss_yns_4: 0.2188, loss_cls_5: 1.5827, loss_box_5: 5.0142, loss_cns_5: 0.4289, loss_yns_5: 0.1947, loss_cls_dn_0: 0.7026, loss_box_dn_0: 1.3018, loss_cls_dn_1: 0.7477, loss_box_dn_1: 2.5121, loss_cls_dn_2: 0.7106, loss_box_dn_2: 2.6008, loss_cls_dn_3: 0.6586, loss_box_dn_3: 2.8447, loss_cls_dn_4: 0.6312, loss_box_dn_4: 3.0401, loss_cls_dn_5: 0.6740, loss_box_dn_5: 3.2560, loss_dense_depth: 1.6563, loss: 57.8526, grad_norm: 144.7561 -2026-02-10 16:27:01,311 - mmdet - INFO - Iter [5/35100] lr: 1.016e-04, eta: 2 days, 5:36:21, time: 1.139, data_time: 0.038, memory: 24611, loss_cls_0: 1.3736, loss_box_0: 2.9184, loss_cns_0: 0.5734, loss_yns_0: 0.1719, loss_cls_1: 1.5909, loss_box_1: 3.9228, loss_cns_1: 0.3929, loss_yns_1: 0.1704, loss_cls_2: 1.5918, loss_box_2: 3.8586, loss_cns_2: 0.4057, loss_yns_2: 0.1622, loss_cls_3: 1.4265, loss_box_3: 4.1254, loss_cns_3: 0.4671, loss_yns_3: 0.1871, loss_cls_4: 1.4647, loss_box_4: 4.3441, loss_cns_4: 0.4293, loss_yns_4: 0.1842, loss_cls_5: 1.4795, loss_box_5: 4.6468, loss_cns_5: 0.4322, loss_yns_5: 0.1698, loss_cls_dn_0: 0.5892, loss_box_dn_0: 1.2602, loss_cls_dn_1: 0.6954, loss_box_dn_1: 2.4955, loss_cls_dn_2: 0.6787, loss_box_dn_2: 2.6214, loss_cls_dn_3: 0.5866, loss_box_dn_3: 2.8200, loss_cls_dn_4: 0.5705, loss_box_dn_4: 2.9960, loss_cls_dn_5: 0.6032, loss_box_dn_5: 3.1172, loss_dense_depth: 1.6806, loss: 57.2039, grad_norm: 210.7267 -2026-02-10 16:27:02,435 - mmdet - INFO - Iter [6/35100] lr: 1.020e-04, eta: 1 day, 22:29:48, time: 1.124, data_time: 0.039, memory: 24611, loss_cls_0: 1.4006, loss_box_0: 2.6324, loss_cns_0: 0.5579, loss_yns_0: 0.2341, loss_cls_1: 1.5327, loss_box_1: 4.1901, loss_cns_1: 0.4165, loss_yns_1: 0.1676, loss_cls_2: 1.5457, loss_box_2: 4.3059, loss_cns_2: 0.3914, loss_yns_2: 0.1636, loss_cls_3: 1.4699, loss_box_3: 4.3766, loss_cns_3: 0.4124, loss_yns_3: 0.1677, loss_cls_4: 1.4552, loss_box_4: 4.7259, loss_cns_4: 0.3590, loss_yns_4: 0.2268, loss_cls_5: 1.4104, loss_box_5: 4.8407, loss_cns_5: 0.3687, loss_yns_5: 0.2103, loss_cls_dn_0: 0.5300, loss_box_dn_0: 1.1384, loss_cls_dn_1: 0.6629, loss_box_dn_1: 2.9004, loss_cls_dn_2: 0.6204, loss_box_dn_2: 2.9267, loss_cls_dn_3: 0.5417, loss_box_dn_3: 3.0538, loss_cls_dn_4: 0.5498, loss_box_dn_4: 3.2128, loss_cls_dn_5: 0.5487, loss_box_dn_5: 3.2511, loss_dense_depth: 1.6535, loss: 59.1525, grad_norm: 147.1835 -2026-02-10 16:27:03,557 - mmdet - INFO - Iter [7/35100] lr: 1.024e-04, eta: 1 day, 17:24:58, time: 1.122, data_time: 0.040, memory: 24611, loss_cls_0: 1.3283, loss_box_0: 2.4248, loss_cns_0: 0.6700, loss_yns_0: 0.2512, loss_cls_1: 1.4654, loss_box_1: 3.8779, loss_cns_1: 0.4420, loss_yns_1: 0.1800, loss_cls_2: 1.5532, loss_box_2: 3.7679, loss_cns_2: 0.4088, loss_yns_2: 0.1906, loss_cls_3: 1.4667, loss_box_3: 3.5995, loss_cns_3: 0.4257, loss_yns_3: 0.2046, loss_cls_4: 1.3903, loss_box_4: 3.9225, loss_cns_4: 0.3850, loss_yns_4: 0.2229, loss_cls_5: 1.3345, loss_box_5: 4.1766, loss_cns_5: 0.3954, loss_yns_5: 0.1985, loss_cls_dn_0: 0.5413, loss_box_dn_0: 1.1187, loss_cls_dn_1: 0.6126, loss_box_dn_1: 3.3573, loss_cls_dn_2: 0.5773, loss_box_dn_2: 3.2641, loss_cls_dn_3: 0.5120, loss_box_dn_3: 3.2404, loss_cls_dn_4: 0.5189, loss_box_dn_4: 3.3049, loss_cls_dn_5: 0.5052, loss_box_dn_5: 3.4169, loss_dense_depth: 1.5924, loss: 56.8443, grad_norm: 114.3462 -2026-02-10 16:27:04,679 - mmdet - INFO - Iter [8/35100] lr: 1.028e-04, eta: 1 day, 13:36:16, time: 1.121, data_time: 0.037, memory: 24611, loss_cls_0: 1.2762, loss_box_0: 2.3461, loss_cns_0: 0.6742, loss_yns_0: 0.1997, loss_cls_1: 1.3779, loss_box_1: 3.6697, loss_cns_1: 0.4537, loss_yns_1: 0.1663, loss_cls_2: 1.5221, loss_box_2: 3.5544, loss_cns_2: 0.4122, loss_yns_2: 0.1870, loss_cls_3: 1.3911, loss_box_3: 3.5476, loss_cns_3: 0.3964, loss_yns_3: 0.2064, loss_cls_4: 1.4155, loss_box_4: 3.7930, loss_cns_4: 0.3931, loss_yns_4: 0.1682, loss_cls_5: 1.3942, loss_box_5: 3.9147, loss_cns_5: 0.4223, loss_yns_5: 0.1990, loss_cls_dn_0: 0.5352, loss_box_dn_0: 1.0530, loss_cls_dn_1: 0.6172, loss_box_dn_1: 2.0363, loss_cls_dn_2: 0.5917, loss_box_dn_2: 1.8779, loss_cls_dn_3: 0.5247, loss_box_dn_3: 1.8641, loss_cls_dn_4: 0.5094, loss_box_dn_4: 2.0075, loss_cls_dn_5: 0.4971, loss_box_dn_5: 2.2047, loss_dense_depth: 1.5219, loss: 48.9219, grad_norm: 101.2006 -2026-02-10 16:27:05,803 - mmdet - INFO - Iter [9/35100] lr: 1.032e-04, eta: 1 day, 10:38:36, time: 1.125, data_time: 0.038, memory: 24611, loss_cls_0: 1.2642, loss_box_0: 2.3870, loss_cns_0: 0.6785, loss_yns_0: 0.1741, loss_cls_1: 1.4247, loss_box_1: 3.7892, loss_cns_1: 0.4286, loss_yns_1: 0.1771, loss_cls_2: 1.4165, loss_box_2: 3.6992, loss_cns_2: 0.3974, loss_yns_2: 0.1683, loss_cls_3: 1.3397, loss_box_3: 3.8785, loss_cns_3: 0.3675, loss_yns_3: 0.1977, loss_cls_4: 1.3890, loss_box_4: 3.9961, loss_cns_4: 0.3430, loss_yns_4: 0.1758, loss_cls_5: 1.4147, loss_box_5: 4.0497, loss_cns_5: 0.3738, loss_yns_5: 0.1802, loss_cls_dn_0: 0.5386, loss_box_dn_0: 1.0622, loss_cls_dn_1: 0.5673, loss_box_dn_1: 1.8283, loss_cls_dn_2: 0.5687, loss_box_dn_2: 1.8429, loss_cls_dn_3: 0.5119, loss_box_dn_3: 1.9575, loss_cls_dn_4: 0.5125, loss_box_dn_4: 2.0325, loss_cls_dn_5: 0.4815, loss_box_dn_5: 2.1449, loss_dense_depth: 1.4660, loss: 49.2251, grad_norm: 106.0484 -2026-02-10 16:27:06,923 - mmdet - INFO - Iter [10/35100] lr: 1.036e-04, eta: 1 day, 8:16:09, time: 1.119, data_time: 0.036, memory: 24611, loss_cls_0: 1.2539, loss_box_0: 2.3782, loss_cns_0: 0.6266, loss_yns_0: 0.1675, loss_cls_1: 1.3682, loss_box_1: 3.3882, loss_cns_1: 0.4556, loss_yns_1: 0.1765, loss_cls_2: 1.3511, loss_box_2: 3.5524, loss_cns_2: 0.4368, loss_yns_2: 0.1727, loss_cls_3: 1.3237, loss_box_3: 3.7702, loss_cns_3: 0.4285, loss_yns_3: 0.2075, loss_cls_4: 1.3475, loss_box_4: 3.6337, loss_cns_4: 0.4325, loss_yns_4: 0.1676, loss_cls_5: 1.3364, loss_box_5: 3.6077, loss_cns_5: 0.4950, loss_yns_5: 0.1827, loss_cls_dn_0: 0.5269, loss_box_dn_0: 1.0756, loss_cls_dn_1: 0.5091, loss_box_dn_1: 1.8940, loss_cls_dn_2: 0.5489, loss_box_dn_2: 1.9816, loss_cls_dn_3: 0.4966, loss_box_dn_3: 2.0842, loss_cls_dn_4: 0.4939, loss_box_dn_4: 2.1016, loss_cls_dn_5: 0.4622, loss_box_dn_5: 2.1453, loss_dense_depth: 1.4707, loss: 48.0513, grad_norm: 108.4478 diff --git a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20260210_162616.log.json b/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20260210_162616.log.json deleted file mode 100644 index bcaf762d86fb7650c80bdbd8422518370266eff1..0000000000000000000000000000000000000000 --- a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20260210_162616.log.json +++ /dev/null @@ -1,11 +0,0 @@ -{"env_info": "sys.platform: linux\nPython: 3.10.12 (main, Aug 15 2025, 14:32:43) [GCC 11.4.0]\nCUDA available: True\nGPU 0,1,2,3,4,5,6,7: BW200_LC, UBB BW1000_LC\nCUDA_HOME: /opt/dtk\nNVCC: Not Available\nGCC: x86_64-linux-gnu-gcc (Ubuntu 11.4.0-1ubuntu1~22.04) 11.4.0\nPyTorch: 2.5.1\nPyTorch compiling details: PyTorch built with:\n - GCC 10.3\n - C++ Version: 201703\n - Intel(R) Math Kernel Library Version 2020.0.4 Product Build 20200917 for Intel(R) 64 architecture applications\n - OpenMP 201511 (a.k.a. OpenMP 4.5)\n - LAPACK is enabled (usually provided by MKL)\n - NNPACK is enabled\n - CPU capability usage: AVX2\n - HIP Runtime 6.3.25521\n - MIOpen 2.18.0\n - Magma 2.8.0\n - Build settings: BLAS_INFO=mkl, BUILD_TYPE=Release, CXX_COMPILER=/opt/rh/gcc-toolset-10/root/usr/bin/c++, CXX_FLAGS= -D_GLIBCXX_USE_CXX11_ABI=1 -fvisibility-inlines-hidden -DUSE_PTHREADPOOL -DNDEBUG -DUSE_KINETO -DLIBKINETO_NOCUPTI -DLIBKINETO_NOXPUPTI=ON -DUSE_FBGEMM -DUSE_PYTORCH_QNNPACK -DUSE_XNNPACK -DSYMBOLICATE_MOBILE_DEBUG_HANDLE -O2 -fPIC -Wall -Wextra -Werror=return-type -Werror=non-virtual-dtor -Werror=bool-operation -Wnarrowing -Wno-missing-field-initializers -Wno-type-limits -Wno-array-bounds -Wno-unknown-pragmas -Wno-unused-parameter -Wno-strict-overflow -Wno-strict-aliasing -Wno-stringop-overflow -Wsuggest-override -Wno-psabi -Wno-error=old-style-cast -Wno-missing-braces -fdiagnostics-color=always -faligned-new -Wno-unused-but-set-variable -Wno-maybe-uninitialized -fno-math-errno -fno-trapping-math -Werror=format -Wno-stringop-overflow, FORCE_FALLBACK_CUDA_MPI=1, LAPACK_INFO=mkl, PERF_WITH_AVX=1, PERF_WITH_AVX2=1, TORCH_VERSION=2.5.1, USE_CUDA=0, USE_CUDNN=OFF, USE_CUSPARSELT=OFF, USE_EXCEPTION_PTR=1, USE_GFLAGS=1, USE_GLOG=1, USE_GLOO=1, USE_MKL=ON, USE_MKLDNN=0, USE_MPI=1, USE_NCCL=1, USE_NNPACK=ON, USE_OPENMP=1, USE_ROCM=ON, USE_ROCM_KERNEL_ASSERT=OFF, \n\nTorchVision: 0.20.1\nOpenCV: 4.12.0\nMMCV: 1.6.1\nMMCV Compiler: GCC 10.3\nMMCV CUDA Compiler: rocm not available\nMMDetection: 2.25.1+", "config": "plugin = True\nplugin_dir = 'projects/mmdet3d_plugin/'\ndist_params = dict(backend='nccl')\nlog_level = 'INFO'\nwork_dir = './work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704'\ntotal_batch_size = 80\nnum_gpus = 8\nbatch_size = 10\nnum_iters_per_epoch = 351\nnum_epochs = 100\ncheckpoint_epoch_interval = 20\ncheckpoint_config = dict(interval=7020)\nlog_config = dict(\n interval=1,\n hooks=[\n dict(type='TextLoggerHook', by_epoch=False),\n dict(type='TensorboardLoggerHook')\n ])\nload_from = None\nresume_from = None\nworkflow = [('train', 1)]\nfp16 = dict(loss_scale=32.0)\ninput_shape = (704, 256)\ntracking_test = True\ntracking_threshold = 0.2\nclass_names = [\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier',\n 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n]\nnum_classes = 10\nembed_dims = 256\nnum_groups = 8\nnum_decoder = 6\nnum_single_frame_decoder = 1\nuse_deformable_func = True\nstrides = [4, 8, 16, 32]\nnum_levels = 4\nnum_depth_layers = 3\ndrop_out = 0.1\ntemporal = True\ndecouple_attn = True\nwith_quality_estimation = True\nmodel = dict(\n type='Sparse4D',\n use_grid_mask=True,\n use_deformable_func=True,\n img_backbone=dict(\n type='ResNet',\n depth=50,\n num_stages=4,\n frozen_stages=-1,\n norm_eval=False,\n style='pytorch',\n with_cp=True,\n out_indices=(0, 1, 2, 3),\n norm_cfg=dict(type='BN', requires_grad=True),\n pretrained='ckpt/resnet50-19c8e357.pth'),\n img_neck=dict(\n type='FPN',\n num_outs=4,\n start_level=0,\n out_channels=256,\n add_extra_convs='on_output',\n relu_before_extra_convs=True,\n in_channels=[256, 512, 1024, 2048]),\n depth_branch=dict(\n type='DenseDepthNet',\n embed_dims=256,\n num_depth_layers=3,\n loss_weight=0.2),\n head=dict(\n type='Sparse4DHead',\n cls_threshold_to_reg=0.05,\n decouple_attn=True,\n instance_bank=dict(\n type='InstanceBank',\n num_anchor=900,\n embed_dims=256,\n anchor='nuscenes_kmeans900.npy',\n anchor_handler=dict(type='SparseBox3DKeyPointsGenerator'),\n num_temp_instances=600,\n confidence_decay=0.6,\n feat_grad=False),\n anchor_encoder=dict(\n type='SparseBox3DEncoder',\n vel_dims=3,\n embed_dims=[128, 32, 32, 64],\n mode='cat',\n output_fc=False,\n in_loops=1,\n out_loops=4),\n num_single_frame_decoder=1,\n operation_order=[\n 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm',\n 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm',\n 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm',\n 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm',\n 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm',\n 'deformable', 'ffn', 'norm', 'refine'\n ],\n temp_graph_model=dict(\n type='MultiheadAttention',\n embed_dims=512,\n num_heads=8,\n batch_first=True,\n dropout=0.1),\n graph_model=dict(\n type='MultiheadAttention',\n embed_dims=512,\n num_heads=8,\n batch_first=True,\n dropout=0.1),\n norm_layer=dict(type='LN', normalized_shape=256),\n ffn=dict(\n type='AsymmetricFFN',\n in_channels=512,\n pre_norm=dict(type='LN'),\n embed_dims=256,\n feedforward_channels=1024,\n num_fcs=2,\n ffn_drop=0.1,\n act_cfg=dict(type='ReLU', inplace=True)),\n deformable_model=dict(\n type='DeformableFeatureAggregation',\n embed_dims=256,\n num_groups=8,\n num_levels=4,\n num_cams=6,\n attn_drop=0.15,\n use_deformable_func=True,\n use_camera_embed=True,\n residual_mode='cat',\n kps_generator=dict(\n type='SparseBox3DKeyPointsGenerator',\n num_learnable_pts=6,\n fix_scale=[[0, 0, 0], [0.45, 0, 0], [-0.45, 0, 0],\n [0, 0.45, 0], [0, -0.45, 0], [0, 0, 0.45],\n [0, 0, -0.45]])),\n refine_layer=dict(\n type='SparseBox3DRefinementModule',\n embed_dims=256,\n num_cls=10,\n refine_yaw=True,\n with_quality_estimation=True),\n sampler=dict(\n type='SparseBox3DTarget',\n num_dn_groups=5,\n num_temp_dn_groups=3,\n dn_noise_scale=[2.0, 2.0, 2.0, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5],\n max_dn_gt=32,\n add_neg_dn=True,\n cls_weight=2.0,\n box_weight=0.25,\n reg_weights=[2.0, 2.0, 2.0, 0.5, 0.5, 0.5, 0.0, 0.0, 0.0, 0.0],\n cls_wise_reg_weights=dict(\n {9: [2.0, 2.0, 2.0, 1.0, 1.0, 1.0, 0.0, 0.0, 1.0, 1.0]})),\n loss_cls=dict(\n type='FocalLoss',\n use_sigmoid=True,\n gamma=2.0,\n alpha=0.25,\n loss_weight=2.0),\n loss_reg=dict(\n type='SparseBox3DLoss',\n loss_box=dict(type='L1Loss', loss_weight=0.25),\n loss_centerness=dict(type='CrossEntropyLoss', use_sigmoid=True),\n loss_yawness=dict(type='GaussianFocalLoss'),\n cls_allow_reverse=[5]),\n decoder=dict(type='SparseBox3DDecoder'),\n reg_weights=[2.0, 2.0, 2.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0]))\ndataset_type = 'NuScenes3DDetTrackDataset'\ndata_root = 'data/nuscenes/'\nanno_root = 'data/nuscenes_anno_pkls/'\nfile_client_args = dict(backend='disk')\nimg_norm_cfg = dict(\n mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375], to_rgb=True)\ntrain_pipeline = [\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(\n type='LoadPointsFromFile',\n coord_type='LIDAR',\n load_dim=5,\n use_dim=5,\n file_client_args=dict(backend='disk')),\n dict(type='ResizeCropFlipImage'),\n dict(type='MultiScaleDepthMapGenerator', downsample=[4, 8, 16]),\n dict(type='BBoxRotation'),\n dict(type='PhotoMetricDistortionMultiViewImage'),\n dict(\n type='NormalizeMultiviewImage',\n mean=[123.675, 116.28, 103.53],\n std=[58.395, 57.12, 57.375],\n to_rgb=True),\n dict(\n type='CircleObjectRangeFilter',\n class_dist_thred=[55, 55, 55, 55, 55, 55, 55, 55, 55, 55]),\n dict(\n type='InstanceNameFilter',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ]),\n dict(type='NuScenesSparse4DAdaptor'),\n dict(\n type='Collect',\n keys=[\n 'img', 'timestamp', 'projection_mat', 'image_wh', 'gt_depth',\n 'focal', 'gt_bboxes_3d', 'gt_labels_3d'\n ],\n meta_keys=['T_global', 'T_global_inv', 'timestamp', 'instance_id'])\n]\ntest_pipeline = [\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(type='ResizeCropFlipImage'),\n dict(\n type='NormalizeMultiviewImage',\n mean=[123.675, 116.28, 103.53],\n std=[58.395, 57.12, 57.375],\n to_rgb=True),\n dict(type='NuScenesSparse4DAdaptor'),\n dict(\n type='Collect',\n keys=['img', 'timestamp', 'projection_mat', 'image_wh'],\n meta_keys=['T_global', 'T_global_inv', 'timestamp'])\n]\ninput_modality = dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False)\ndata_basic_config = dict(\n type='NuScenes3DDetTrackDataset',\n data_root='data/nuscenes/',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier',\n 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ],\n modality=dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False),\n version='v1.0-trainval')\ndata_aug_conf = dict(\n resize_lim=(0.4, 0.47),\n final_dim=(256, 704),\n bot_pct_lim=(0.0, 0.0),\n rot_lim=(-5.4, 5.4),\n H=900,\n W=1600,\n rand_flip=True,\n rot3d_range=[-0.3925, 0.3925])\ndata = dict(\n samples_per_gpu=10,\n workers_per_gpu=10,\n train=dict(\n type='NuScenes3DDetTrackDataset',\n data_root='data/nuscenes/',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ],\n modality=dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False),\n version='v1.0-trainval',\n ann_file='data/nuscenes_anno_pkls/nuscenes_infos_train.pkl',\n pipeline=[\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(\n type='LoadPointsFromFile',\n coord_type='LIDAR',\n load_dim=5,\n use_dim=5,\n file_client_args=dict(backend='disk')),\n dict(type='ResizeCropFlipImage'),\n dict(type='MultiScaleDepthMapGenerator', downsample=[4, 8, 16]),\n dict(type='BBoxRotation'),\n dict(type='PhotoMetricDistortionMultiViewImage'),\n dict(\n type='NormalizeMultiviewImage',\n mean=[123.675, 116.28, 103.53],\n std=[58.395, 57.12, 57.375],\n to_rgb=True),\n dict(\n type='CircleObjectRangeFilter',\n class_dist_thred=[55, 55, 55, 55, 55, 55, 55, 55, 55, 55]),\n dict(\n type='InstanceNameFilter',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian',\n 'traffic_cone'\n ]),\n dict(type='NuScenesSparse4DAdaptor'),\n dict(\n type='Collect',\n keys=[\n 'img', 'timestamp', 'projection_mat', 'image_wh',\n 'gt_depth', 'focal', 'gt_bboxes_3d', 'gt_labels_3d'\n ],\n meta_keys=[\n 'T_global', 'T_global_inv', 'timestamp', 'instance_id'\n ])\n ],\n test_mode=False,\n data_aug_conf=dict(\n resize_lim=(0.4, 0.47),\n final_dim=(256, 704),\n bot_pct_lim=(0.0, 0.0),\n rot_lim=(-5.4, 5.4),\n H=900,\n W=1600,\n rand_flip=True,\n rot3d_range=[-0.3925, 0.3925]),\n with_seq_flag=True,\n sequences_split_num=2,\n keep_consistent_seq_aug=True),\n val=dict(\n type='NuScenes3DDetTrackDataset',\n data_root='data/nuscenes/',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ],\n modality=dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False),\n version='v1.0-trainval',\n ann_file='data/nuscenes_anno_pkls/nuscenes_infos_val.pkl',\n pipeline=[\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(type='ResizeCropFlipImage'),\n dict(\n type='NormalizeMultiviewImage',\n mean=[123.675, 116.28, 103.53],\n std=[58.395, 57.12, 57.375],\n to_rgb=True),\n dict(type='NuScenesSparse4DAdaptor'),\n dict(\n type='Collect',\n keys=['img', 'timestamp', 'projection_mat', 'image_wh'],\n meta_keys=['T_global', 'T_global_inv', 'timestamp'])\n ],\n data_aug_conf=dict(\n resize_lim=(0.4, 0.47),\n final_dim=(256, 704),\n bot_pct_lim=(0.0, 0.0),\n rot_lim=(-5.4, 5.4),\n H=900,\n W=1600,\n rand_flip=True,\n rot3d_range=[-0.3925, 0.3925]),\n test_mode=True,\n tracking=True,\n tracking_threshold=0.2),\n test=dict(\n type='NuScenes3DDetTrackDataset',\n data_root='data/nuscenes/',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ],\n modality=dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False),\n version='v1.0-trainval',\n ann_file='data/nuscenes_anno_pkls/nuscenes_infos_val.pkl',\n pipeline=[\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(type='ResizeCropFlipImage'),\n dict(\n type='NormalizeMultiviewImage',\n mean=[123.675, 116.28, 103.53],\n std=[58.395, 57.12, 57.375],\n to_rgb=True),\n dict(type='NuScenesSparse4DAdaptor'),\n dict(\n type='Collect',\n keys=['img', 'timestamp', 'projection_mat', 'image_wh'],\n meta_keys=['T_global', 'T_global_inv', 'timestamp'])\n ],\n data_aug_conf=dict(\n resize_lim=(0.4, 0.47),\n final_dim=(256, 704),\n bot_pct_lim=(0.0, 0.0),\n rot_lim=(-5.4, 5.4),\n H=900,\n W=1600,\n rand_flip=True,\n rot3d_range=[-0.3925, 0.3925]),\n test_mode=True,\n tracking=True,\n tracking_threshold=0.2))\noptimizer = dict(\n type='AdamW',\n lr=0.0006,\n weight_decay=0.001,\n paramwise_cfg=dict(custom_keys=dict(img_backbone=dict(lr_mult=0.5))))\noptimizer_config = dict(grad_clip=dict(max_norm=25, norm_type=2))\nlr_config = dict(\n policy='CosineAnnealing',\n warmup='linear',\n warmup_iters=500,\n warmup_ratio=0.3333333333333333,\n min_lr_ratio=0.001)\nrunner = dict(type='IterBasedRunner', max_iters=35100)\nvis_pipeline = [\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(type='Collect', keys=['img'], meta_keys=['timestamp', 'lidar2img'])\n]\nevaluation = dict(\n interval=7020,\n pipeline=[\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(\n type='Collect', keys=['img'], meta_keys=['timestamp', 'lidar2img'])\n ])\ngpu_ids = range(0, 1)\n", "seed": 0, "exp_name": "sparse4dv3_temporal_r50_1x8_bs6_256x704.py"} -{"mode": "train", "epoch": 1, "iter": 1, "lr": 0.0001, "memory": 24611, "data_time": 3.91893, "loss_cls_0": 2.30663, "loss_box_0": 0.0, "loss_cns_0": 0.0, "loss_yns_0": 0.0, "loss_cls_1": 2.16604, "loss_box_1": 0.11646, "loss_cns_1": 0.02369, "loss_yns_1": 0.00487, "loss_cls_2": 2.24642, "loss_box_2": 0.0, "loss_cns_2": 0.0, "loss_yns_2": 0.0, "loss_cls_3": 2.31147, "loss_box_3": 0.05239, "loss_cns_3": 0.00847, "loss_yns_3": 0.00082, "loss_cls_4": 1.9817, "loss_box_4": 0.34944, "loss_cns_4": 0.04751, "loss_yns_4": 0.03208, "loss_cls_5": 2.34902, "loss_box_5": 0.0, "loss_cns_5": 0.0, "loss_yns_5": 0.0, "loss_cls_dn_0": 1.1647, "loss_box_dn_0": 1.4583, "loss_cls_dn_1": 1.15688, "loss_box_dn_1": 1.72702, "loss_cls_dn_2": 1.13358, "loss_box_dn_2": 1.94337, "loss_cls_dn_3": 1.12714, "loss_box_dn_3": 2.21498, "loss_cls_dn_4": 1.04984, "loss_box_dn_4": 2.40351, "loss_cls_dn_5": 1.18307, "loss_box_dn_5": 2.68052, "loss_dense_depth": 1.83092, "loss": 35.07083, "grad_norm": 552.76855, "time": 21.98247} -{"mode": "train", "epoch": 1, "iter": 2, "lr": 0.0001, "memory": 24611, "data_time": 0.03771, "loss_cls_0": 2.18432, "loss_box_0": 0.0, "loss_cns_0": 0.0, "loss_yns_0": 0.0, "loss_cls_1": 1.95592, "loss_box_1": 0.57295, "loss_cns_1": 0.07251, "loss_yns_1": 0.01848, "loss_cls_2": 2.05999, "loss_box_2": 0.0, "loss_cns_2": 0.0, "loss_yns_2": 0.0, "loss_cls_3": 1.991, "loss_box_3": 0.18398, "loss_cns_3": 0.02664, "loss_yns_3": 0.0103, "loss_cls_4": 1.78215, "loss_box_4": 1.43034, "loss_cns_4": 0.15074, "loss_yns_4": 0.04494, "loss_cls_5": 1.94877, "loss_box_5": 1.32906, "loss_cns_5": 0.16515, "loss_yns_5": 0.06641, "loss_cls_dn_0": 1.10547, "loss_box_dn_0": 1.26149, "loss_cls_dn_1": 0.96388, "loss_box_dn_1": 2.589, "loss_cls_dn_2": 0.94613, "loss_box_dn_2": 2.67592, "loss_cls_dn_3": 0.9394, "loss_box_dn_3": 2.7226, "loss_cls_dn_4": 0.8565, "loss_box_dn_4": 2.89309, "loss_cls_dn_5": 0.9667, "loss_box_dn_5": 3.11688, "loss_dense_depth": 1.78172, "loss": 38.81245, "grad_norm": 82.91432, "time": 2.15405} -{"mode": "train", "epoch": 1, "iter": 3, "lr": 0.0001, "memory": 24611, "data_time": 0.03768, "loss_cls_0": 1.66421, "loss_box_0": 2.24881, "loss_cns_0": 0.6734, "loss_yns_0": 0.54463, "loss_cls_1": 1.7714, "loss_box_1": 1.76585, "loss_cns_1": 0.2835, "loss_yns_1": 0.07991, "loss_cls_2": 1.72673, "loss_box_2": 2.68747, "loss_cns_2": 0.36512, "loss_yns_2": 0.10875, "loss_cls_3": 1.63724, "loss_box_3": 3.36415, "loss_cns_3": 0.38064, "loss_yns_3": 0.22071, "loss_cls_4": 1.57999, "loss_box_4": 4.02145, "loss_cns_4": 0.44822, "loss_yns_4": 0.12883, "loss_cls_5": 1.61327, "loss_box_5": 3.87227, "loss_cns_5": 0.33362, "loss_yns_5": 0.15439, "loss_cls_dn_0": 0.82297, "loss_box_dn_0": 1.16926, "loss_cls_dn_1": 0.81862, "loss_box_dn_1": 2.34765, "loss_cls_dn_2": 0.7843, "loss_box_dn_2": 2.51772, "loss_cls_dn_3": 0.72339, "loss_box_dn_3": 2.72782, "loss_cls_dn_4": 0.69812, "loss_box_dn_4": 2.89686, "loss_cls_dn_5": 0.77291, "loss_box_dn_5": 3.12608, "loss_dense_depth": 1.77897, "loss": 52.85923, "grad_norm": 132.71275, "time": 1.10585} -{"mode": "train", "epoch": 1, "iter": 4, "lr": 0.0001, "memory": 24611, "data_time": 0.03857, "loss_cls_0": 1.4907, "loss_box_0": 2.94162, "loss_cns_0": 0.67218, "loss_yns_0": 0.29489, "loss_cls_1": 1.64272, "loss_box_1": 2.66643, "loss_cns_1": 0.388, "loss_yns_1": 0.13956, "loss_cls_2": 1.61947, "loss_box_2": 3.79002, "loss_cns_2": 0.49747, "loss_yns_2": 0.18079, "loss_cls_3": 1.49455, "loss_box_3": 4.09433, "loss_cns_3": 0.4754, "loss_yns_3": 0.23638, "loss_cls_4": 1.44855, "loss_box_4": 4.61731, "loss_cns_4": 0.38622, "loss_yns_4": 0.21875, "loss_cls_5": 1.58266, "loss_box_5": 5.01423, "loss_cns_5": 0.42892, "loss_yns_5": 0.19474, "loss_cls_dn_0": 0.70264, "loss_box_dn_0": 1.30184, "loss_cls_dn_1": 0.74773, "loss_box_dn_1": 2.51209, "loss_cls_dn_2": 0.71059, "loss_box_dn_2": 2.60082, "loss_cls_dn_3": 0.65856, "loss_box_dn_3": 2.84472, "loss_cls_dn_4": 0.63123, "loss_box_dn_4": 3.04012, "loss_cls_dn_5": 0.67402, "loss_box_dn_5": 3.25604, "loss_dense_depth": 1.65628, "loss": 57.85255, "grad_norm": 144.75609, "time": 1.11293} -{"mode": "train", "epoch": 1, "iter": 5, "lr": 0.0001, "memory": 24611, "data_time": 0.03831, "loss_cls_0": 1.37363, "loss_box_0": 2.91841, "loss_cns_0": 0.57343, "loss_yns_0": 0.17187, "loss_cls_1": 1.59089, "loss_box_1": 3.92281, "loss_cns_1": 0.39293, "loss_yns_1": 0.17037, "loss_cls_2": 1.5918, "loss_box_2": 3.85864, "loss_cns_2": 0.40566, "loss_yns_2": 0.16218, "loss_cls_3": 1.42649, "loss_box_3": 4.1254, "loss_cns_3": 0.46714, "loss_yns_3": 0.18707, "loss_cls_4": 1.46471, "loss_box_4": 4.34412, "loss_cns_4": 0.42928, "loss_yns_4": 0.18421, "loss_cls_5": 1.47945, "loss_box_5": 4.64682, "loss_cns_5": 0.43224, "loss_yns_5": 0.16982, "loss_cls_dn_0": 0.58922, "loss_box_dn_0": 1.26018, "loss_cls_dn_1": 0.69539, "loss_box_dn_1": 2.49545, "loss_cls_dn_2": 0.67866, "loss_box_dn_2": 2.62137, "loss_cls_dn_3": 0.58665, "loss_box_dn_3": 2.82004, "loss_cls_dn_4": 0.5705, "loss_box_dn_4": 2.99598, "loss_cls_dn_5": 0.60323, "loss_box_dn_5": 3.1172, "loss_dense_depth": 1.68064, "loss": 57.2039, "grad_norm": 210.72673, "time": 1.13888} -{"mode": "train", "epoch": 1, "iter": 6, "lr": 0.0001, "memory": 24611, "data_time": 0.03853, "loss_cls_0": 1.40063, "loss_box_0": 2.63245, "loss_cns_0": 0.55789, "loss_yns_0": 0.23414, "loss_cls_1": 1.53268, "loss_box_1": 4.19013, "loss_cns_1": 0.41648, "loss_yns_1": 0.1676, "loss_cls_2": 1.54565, "loss_box_2": 4.30595, "loss_cns_2": 0.39136, "loss_yns_2": 0.16363, "loss_cls_3": 1.46988, "loss_box_3": 4.37662, "loss_cns_3": 0.41237, "loss_yns_3": 0.16768, "loss_cls_4": 1.45524, "loss_box_4": 4.72589, "loss_cns_4": 0.35899, "loss_yns_4": 0.22685, "loss_cls_5": 1.41042, "loss_box_5": 4.84067, "loss_cns_5": 0.36866, "loss_yns_5": 0.21032, "loss_cls_dn_0": 0.53004, "loss_box_dn_0": 1.13839, "loss_cls_dn_1": 0.66293, "loss_box_dn_1": 2.90043, "loss_cls_dn_2": 0.62038, "loss_box_dn_2": 2.92671, "loss_cls_dn_3": 0.54166, "loss_box_dn_3": 3.0538, "loss_cls_dn_4": 0.54985, "loss_box_dn_4": 3.21278, "loss_cls_dn_5": 0.54875, "loss_box_dn_5": 3.25108, "loss_dense_depth": 1.65348, "loss": 59.15246, "grad_norm": 147.18355, "time": 1.12413} -{"mode": "train", "epoch": 1, "iter": 7, "lr": 0.0001, "memory": 24611, "data_time": 0.03969, "loss_cls_0": 1.32831, "loss_box_0": 2.42479, "loss_cns_0": 0.66999, "loss_yns_0": 0.25123, "loss_cls_1": 1.46543, "loss_box_1": 3.87786, "loss_cns_1": 0.44204, "loss_yns_1": 0.18003, "loss_cls_2": 1.55323, "loss_box_2": 3.76785, "loss_cns_2": 0.40879, "loss_yns_2": 0.19059, "loss_cls_3": 1.46666, "loss_box_3": 3.59951, "loss_cns_3": 0.42573, "loss_yns_3": 0.20456, "loss_cls_4": 1.39027, "loss_box_4": 3.92253, "loss_cns_4": 0.38499, "loss_yns_4": 0.22287, "loss_cls_5": 1.33454, "loss_box_5": 4.17657, "loss_cns_5": 0.39541, "loss_yns_5": 0.19855, "loss_cls_dn_0": 0.54133, "loss_box_dn_0": 1.11868, "loss_cls_dn_1": 0.6126, "loss_box_dn_1": 3.35731, "loss_cls_dn_2": 0.5773, "loss_box_dn_2": 3.26407, "loss_cls_dn_3": 0.51205, "loss_box_dn_3": 3.2404, "loss_cls_dn_4": 0.51887, "loss_box_dn_4": 3.3049, "loss_cls_dn_5": 0.50523, "loss_box_dn_5": 3.41687, "loss_dense_depth": 1.5924, "loss": 56.84434, "grad_norm": 114.3462, "time": 1.12237} -{"mode": "train", "epoch": 1, "iter": 8, "lr": 0.0001, "memory": 24611, "data_time": 0.03745, "loss_cls_0": 1.27622, "loss_box_0": 2.34612, "loss_cns_0": 0.6742, "loss_yns_0": 0.19967, "loss_cls_1": 1.37794, "loss_box_1": 3.66971, "loss_cns_1": 0.4537, "loss_yns_1": 0.16629, "loss_cls_2": 1.52208, "loss_box_2": 3.55444, "loss_cns_2": 0.41221, "loss_yns_2": 0.18705, "loss_cls_3": 1.39114, "loss_box_3": 3.54757, "loss_cns_3": 0.3964, "loss_yns_3": 0.20636, "loss_cls_4": 1.41546, "loss_box_4": 3.79295, "loss_cns_4": 0.3931, "loss_yns_4": 0.16817, "loss_cls_5": 1.39416, "loss_box_5": 3.91474, "loss_cns_5": 0.42232, "loss_yns_5": 0.19899, "loss_cls_dn_0": 0.53521, "loss_box_dn_0": 1.05299, "loss_cls_dn_1": 0.6172, "loss_box_dn_1": 2.03628, "loss_cls_dn_2": 0.5917, "loss_box_dn_2": 1.87792, "loss_cls_dn_3": 0.52474, "loss_box_dn_3": 1.86412, "loss_cls_dn_4": 0.50944, "loss_box_dn_4": 2.00754, "loss_cls_dn_5": 0.4971, "loss_box_dn_5": 2.2047, "loss_dense_depth": 1.52191, "loss": 48.92186, "grad_norm": 101.20061, "time": 1.12131} -{"mode": "train", "epoch": 1, "iter": 9, "lr": 0.0001, "memory": 24611, "data_time": 0.03819, "loss_cls_0": 1.26418, "loss_box_0": 2.38698, "loss_cns_0": 0.67848, "loss_yns_0": 0.17411, "loss_cls_1": 1.4247, "loss_box_1": 3.78921, "loss_cns_1": 0.42865, "loss_yns_1": 0.17709, "loss_cls_2": 1.41646, "loss_box_2": 3.69922, "loss_cns_2": 0.39736, "loss_yns_2": 0.16827, "loss_cls_3": 1.33973, "loss_box_3": 3.87854, "loss_cns_3": 0.36751, "loss_yns_3": 0.19768, "loss_cls_4": 1.38902, "loss_box_4": 3.99612, "loss_cns_4": 0.34295, "loss_yns_4": 0.17576, "loss_cls_5": 1.41468, "loss_box_5": 4.04968, "loss_cns_5": 0.37383, "loss_yns_5": 0.18017, "loss_cls_dn_0": 0.53858, "loss_box_dn_0": 1.06219, "loss_cls_dn_1": 0.5673, "loss_box_dn_1": 1.82828, "loss_cls_dn_2": 0.56867, "loss_box_dn_2": 1.84288, "loss_cls_dn_3": 0.51195, "loss_box_dn_3": 1.95749, "loss_cls_dn_4": 0.51249, "loss_box_dn_4": 2.0325, "loss_cls_dn_5": 0.48149, "loss_box_dn_5": 2.1449, "loss_dense_depth": 1.46599, "loss": 49.2251, "grad_norm": 106.04836, "time": 1.12493} -{"mode": "train", "epoch": 1, "iter": 10, "lr": 0.0001, "memory": 24611, "data_time": 0.036, "loss_cls_0": 1.25394, "loss_box_0": 2.37821, "loss_cns_0": 0.62657, "loss_yns_0": 0.16749, "loss_cls_1": 1.3682, "loss_box_1": 3.38818, "loss_cns_1": 0.45564, "loss_yns_1": 0.17654, "loss_cls_2": 1.35111, "loss_box_2": 3.55236, "loss_cns_2": 0.43676, "loss_yns_2": 0.17269, "loss_cls_3": 1.32365, "loss_box_3": 3.77024, "loss_cns_3": 0.42846, "loss_yns_3": 0.20747, "loss_cls_4": 1.34754, "loss_box_4": 3.63374, "loss_cns_4": 0.43249, "loss_yns_4": 0.16757, "loss_cls_5": 1.33636, "loss_box_5": 3.60772, "loss_cns_5": 0.49503, "loss_yns_5": 0.18271, "loss_cls_dn_0": 0.52693, "loss_box_dn_0": 1.07563, "loss_cls_dn_1": 0.5091, "loss_box_dn_1": 1.89404, "loss_cls_dn_2": 0.54889, "loss_box_dn_2": 1.98159, "loss_cls_dn_3": 0.49662, "loss_box_dn_3": 2.08418, "loss_cls_dn_4": 0.49389, "loss_box_dn_4": 2.1016, "loss_cls_dn_5": 0.46216, "loss_box_dn_5": 2.14529, "loss_dense_depth": 1.47073, "loss": 48.05132, "grad_norm": 108.44775, "time": 1.1191} diff --git a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20260210_162728.log b/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20260210_162728.log deleted file mode 100644 index cee66985d12ecc8b11fac905c8e0a321480292d1..0000000000000000000000000000000000000000 --- a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20260210_162728.log +++ /dev/null @@ -1,3243 +0,0 @@ -2026-02-10 16:27:28,135 - mmdet - INFO - Environment info: ------------------------------------------------------------- -sys.platform: linux -Python: 3.10.12 (main, Aug 15 2025, 14:32:43) [GCC 11.4.0] -CUDA available: True -GPU 0,1,2,3,4,5,6,7: BW200_LC, UBB BW1000_LC -CUDA_HOME: /opt/dtk -NVCC: Not Available -GCC: x86_64-linux-gnu-gcc (Ubuntu 11.4.0-1ubuntu1~22.04) 11.4.0 -PyTorch: 2.5.1 -PyTorch compiling details: PyTorch built with: - - GCC 10.3 - - C++ Version: 201703 - - Intel(R) Math Kernel Library Version 2020.0.4 Product Build 20200917 for Intel(R) 64 architecture applications - - OpenMP 201511 (a.k.a. OpenMP 4.5) - - LAPACK is enabled (usually provided by MKL) - - NNPACK is enabled - - CPU capability usage: AVX2 - - HIP Runtime 6.3.25521 - - MIOpen 2.18.0 - - Magma 2.8.0 - - Build settings: BLAS_INFO=mkl, BUILD_TYPE=Release, CXX_COMPILER=/opt/rh/gcc-toolset-10/root/usr/bin/c++, CXX_FLAGS= -D_GLIBCXX_USE_CXX11_ABI=1 -fvisibility-inlines-hidden -DUSE_PTHREADPOOL -DNDEBUG -DUSE_KINETO -DLIBKINETO_NOCUPTI -DLIBKINETO_NOXPUPTI=ON -DUSE_FBGEMM -DUSE_PYTORCH_QNNPACK -DUSE_XNNPACK -DSYMBOLICATE_MOBILE_DEBUG_HANDLE -O2 -fPIC -Wall -Wextra -Werror=return-type -Werror=non-virtual-dtor -Werror=bool-operation -Wnarrowing -Wno-missing-field-initializers -Wno-type-limits -Wno-array-bounds -Wno-unknown-pragmas -Wno-unused-parameter -Wno-strict-overflow -Wno-strict-aliasing -Wno-stringop-overflow -Wsuggest-override -Wno-psabi -Wno-error=old-style-cast -Wno-missing-braces -fdiagnostics-color=always -faligned-new -Wno-unused-but-set-variable -Wno-maybe-uninitialized -fno-math-errno -fno-trapping-math -Werror=format -Wno-stringop-overflow, FORCE_FALLBACK_CUDA_MPI=1, LAPACK_INFO=mkl, PERF_WITH_AVX=1, PERF_WITH_AVX2=1, TORCH_VERSION=2.5.1, USE_CUDA=0, USE_CUDNN=OFF, USE_CUSPARSELT=OFF, USE_EXCEPTION_PTR=1, USE_GFLAGS=1, USE_GLOG=1, USE_GLOO=1, USE_MKL=ON, USE_MKLDNN=0, USE_MPI=1, USE_NCCL=1, USE_NNPACK=ON, USE_OPENMP=1, USE_ROCM=ON, USE_ROCM_KERNEL_ASSERT=OFF, - -TorchVision: 0.20.1 -OpenCV: 4.12.0 -MMCV: 1.6.1 -MMCV Compiler: GCC 10.3 -MMCV CUDA Compiler: rocm not available -MMDetection: 2.25.1+ ------------------------------------------------------------- - -2026-02-10 16:27:29,123 - mmdet - INFO - Distributed training: False -2026-02-10 16:27:30,098 - mmdet - INFO - Config: -plugin = True -plugin_dir = 'projects/mmdet3d_plugin/' -dist_params = dict(backend='nccl') -log_level = 'INFO' -work_dir = './work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704' -total_batch_size = 80 -num_gpus = 8 -batch_size = 10 -num_iters_per_epoch = 351 -num_epochs = 100 -checkpoint_epoch_interval = 20 -checkpoint_config = dict(interval=7020) -log_config = dict( - interval=1, - hooks=[ - dict(type='TextLoggerHook', by_epoch=False), - dict(type='TensorboardLoggerHook') - ]) -load_from = None -resume_from = None -workflow = [('train', 1)] -fp16 = dict(loss_scale=32.0) -input_shape = (704, 256) -tracking_test = True -tracking_threshold = 0.2 -class_names = [ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier', - 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' -] -num_classes = 10 -embed_dims = 256 -num_groups = 8 -num_decoder = 6 -num_single_frame_decoder = 1 -use_deformable_func = True -strides = [4, 8, 16, 32] -num_levels = 4 -num_depth_layers = 3 -drop_out = 0.1 -temporal = True -decouple_attn = True -with_quality_estimation = True -model = dict( - type='Sparse4D', - use_grid_mask=True, - use_deformable_func=True, - img_backbone=dict( - type='ResNet', - depth=50, - num_stages=4, - frozen_stages=-1, - norm_eval=False, - style='pytorch', - with_cp=True, - out_indices=(0, 1, 2, 3), - norm_cfg=dict(type='BN', requires_grad=True), - pretrained='ckpt/resnet50-19c8e357.pth'), - img_neck=dict( - type='FPN', - num_outs=4, - start_level=0, - out_channels=256, - add_extra_convs='on_output', - relu_before_extra_convs=True, - in_channels=[256, 512, 1024, 2048]), - depth_branch=dict( - type='DenseDepthNet', - embed_dims=256, - num_depth_layers=3, - loss_weight=0.2), - head=dict( - type='Sparse4DHead', - cls_threshold_to_reg=0.05, - decouple_attn=True, - instance_bank=dict( - type='InstanceBank', - num_anchor=900, - embed_dims=256, - anchor='nuscenes_kmeans900.npy', - anchor_handler=dict(type='SparseBox3DKeyPointsGenerator'), - num_temp_instances=600, - confidence_decay=0.6, - feat_grad=False), - anchor_encoder=dict( - type='SparseBox3DEncoder', - vel_dims=3, - embed_dims=[128, 32, 32, 64], - mode='cat', - output_fc=False, - in_loops=1, - out_loops=4), - num_single_frame_decoder=1, - operation_order=[ - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine' - ], - temp_graph_model=dict( - type='MultiheadAttention', - embed_dims=512, - num_heads=8, - batch_first=True, - dropout=0.1), - graph_model=dict( - type='MultiheadAttention', - embed_dims=512, - num_heads=8, - batch_first=True, - dropout=0.1), - norm_layer=dict(type='LN', normalized_shape=256), - ffn=dict( - type='AsymmetricFFN', - in_channels=512, - pre_norm=dict(type='LN'), - embed_dims=256, - feedforward_channels=1024, - num_fcs=2, - ffn_drop=0.1, - act_cfg=dict(type='ReLU', inplace=True)), - deformable_model=dict( - type='DeformableFeatureAggregation', - embed_dims=256, - num_groups=8, - num_levels=4, - num_cams=6, - attn_drop=0.15, - use_deformable_func=True, - use_camera_embed=True, - residual_mode='cat', - kps_generator=dict( - type='SparseBox3DKeyPointsGenerator', - num_learnable_pts=6, - fix_scale=[[0, 0, 0], [0.45, 0, 0], [-0.45, 0, 0], - [0, 0.45, 0], [0, -0.45, 0], [0, 0, 0.45], - [0, 0, -0.45]])), - refine_layer=dict( - type='SparseBox3DRefinementModule', - embed_dims=256, - num_cls=10, - refine_yaw=True, - with_quality_estimation=True), - sampler=dict( - type='SparseBox3DTarget', - num_dn_groups=5, - num_temp_dn_groups=3, - dn_noise_scale=[2.0, 2.0, 2.0, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5], - max_dn_gt=32, - add_neg_dn=True, - cls_weight=2.0, - box_weight=0.25, - reg_weights=[2.0, 2.0, 2.0, 0.5, 0.5, 0.5, 0.0, 0.0, 0.0, 0.0], - cls_wise_reg_weights=dict( - {9: [2.0, 2.0, 2.0, 1.0, 1.0, 1.0, 0.0, 0.0, 1.0, 1.0]})), - loss_cls=dict( - type='FocalLoss', - use_sigmoid=True, - gamma=2.0, - alpha=0.25, - loss_weight=2.0), - loss_reg=dict( - type='SparseBox3DLoss', - loss_box=dict(type='L1Loss', loss_weight=0.25), - loss_centerness=dict(type='CrossEntropyLoss', use_sigmoid=True), - loss_yawness=dict(type='GaussianFocalLoss'), - cls_allow_reverse=[5]), - decoder=dict(type='SparseBox3DDecoder'), - reg_weights=[2.0, 2.0, 2.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0])) -dataset_type = 'NuScenes3DDetTrackDataset' -data_root = 'data/nuscenes/' -anno_root = 'data/nuscenes_anno_pkls/' -file_client_args = dict(backend='disk') -img_norm_cfg = dict( - mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375], to_rgb=True) -train_pipeline = [ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict( - type='LoadPointsFromFile', - coord_type='LIDAR', - load_dim=5, - use_dim=5, - file_client_args=dict(backend='disk')), - dict(type='ResizeCropFlipImage'), - dict(type='MultiScaleDepthMapGenerator', downsample=[4, 8, 16]), - dict(type='BBoxRotation'), - dict(type='PhotoMetricDistortionMultiViewImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict( - type='CircleObjectRangeFilter', - class_dist_thred=[55, 55, 55, 55, 55, 55, 55, 55, 55, 55]), - dict( - type='InstanceNameFilter', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ]), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=[ - 'img', 'timestamp', 'projection_mat', 'image_wh', 'gt_depth', - 'focal', 'gt_bboxes_3d', 'gt_labels_3d' - ], - meta_keys=['T_global', 'T_global_inv', 'timestamp', 'instance_id']) -] -test_pipeline = [ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='ResizeCropFlipImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=['img', 'timestamp', 'projection_mat', 'image_wh'], - meta_keys=['T_global', 'T_global_inv', 'timestamp']) -] -input_modality = dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False) -data_basic_config = dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier', - 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval') -data_aug_conf = dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]) -data = dict( - samples_per_gpu=10, - workers_per_gpu=10, - train=dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval', - ann_file='data/nuscenes_anno_pkls/nuscenes_infos_train.pkl', - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict( - type='LoadPointsFromFile', - coord_type='LIDAR', - load_dim=5, - use_dim=5, - file_client_args=dict(backend='disk')), - dict(type='ResizeCropFlipImage'), - dict(type='MultiScaleDepthMapGenerator', downsample=[4, 8, 16]), - dict(type='BBoxRotation'), - dict(type='PhotoMetricDistortionMultiViewImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict( - type='CircleObjectRangeFilter', - class_dist_thred=[55, 55, 55, 55, 55, 55, 55, 55, 55, 55]), - dict( - type='InstanceNameFilter', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', - 'traffic_cone' - ]), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=[ - 'img', 'timestamp', 'projection_mat', 'image_wh', - 'gt_depth', 'focal', 'gt_bboxes_3d', 'gt_labels_3d' - ], - meta_keys=[ - 'T_global', 'T_global_inv', 'timestamp', 'instance_id' - ]) - ], - test_mode=False, - data_aug_conf=dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]), - with_seq_flag=True, - sequences_split_num=2, - keep_consistent_seq_aug=True), - val=dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval', - ann_file='data/nuscenes_anno_pkls/nuscenes_infos_val.pkl', - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='ResizeCropFlipImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=['img', 'timestamp', 'projection_mat', 'image_wh'], - meta_keys=['T_global', 'T_global_inv', 'timestamp']) - ], - data_aug_conf=dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]), - test_mode=True, - tracking=True, - tracking_threshold=0.2), - test=dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval', - ann_file='data/nuscenes_anno_pkls/nuscenes_infos_val.pkl', - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='ResizeCropFlipImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=['img', 'timestamp', 'projection_mat', 'image_wh'], - meta_keys=['T_global', 'T_global_inv', 'timestamp']) - ], - data_aug_conf=dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]), - test_mode=True, - tracking=True, - tracking_threshold=0.2)) -optimizer = dict( - type='AdamW', - lr=0.0006, - weight_decay=0.001, - paramwise_cfg=dict(custom_keys=dict(img_backbone=dict(lr_mult=0.5)))) -optimizer_config = dict(grad_clip=dict(max_norm=25, norm_type=2)) -lr_config = dict( - policy='CosineAnnealing', - warmup='linear', - warmup_iters=500, - warmup_ratio=0.3333333333333333, - min_lr_ratio=0.001) -runner = dict(type='IterBasedRunner', max_iters=35100) -vis_pipeline = [ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='Collect', keys=['img'], meta_keys=['timestamp', 'lidar2img']) -] -evaluation = dict( - interval=7020, - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict( - type='Collect', keys=['img'], meta_keys=['timestamp', 'lidar2img']) - ]) -gpu_ids = range(0, 1) - -2026-02-10 16:27:30,099 - mmdet - INFO - Set random seed to 0, deterministic: False -2026-02-10 16:27:30,650 - mmdet - INFO - initialize ResNet with init_cfg {'type': 'Pretrained', 'checkpoint': 'ckpt/resnet50-19c8e357.pth'} -2026-02-10 16:27:30,871 - mmdet - INFO - initialize FPN with init_cfg {'type': 'Xavier', 'layer': 'Conv2d', 'distribution': 'uniform'} -Name of parameter - Initialization information - -img_backbone.conv1.weight - torch.Size([64, 3, 7, 7]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.bn1.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.bn1.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.conv1.weight - torch.Size([64, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn1.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn1.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.conv2.weight - torch.Size([64, 64, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn2.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn2.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.conv3.weight - torch.Size([256, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn3.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn3.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.downsample.0.weight - torch.Size([256, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.downsample.1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.downsample.1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.conv1.weight - torch.Size([64, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn1.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn1.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.conv2.weight - torch.Size([64, 64, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn2.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn2.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.conv3.weight - torch.Size([256, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn3.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn3.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.conv1.weight - torch.Size([64, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn1.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn1.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.conv2.weight - torch.Size([64, 64, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn2.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn2.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.conv3.weight - torch.Size([256, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn3.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn3.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.conv1.weight - torch.Size([128, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn1.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn1.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.conv2.weight - torch.Size([128, 128, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn2.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn2.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.conv3.weight - torch.Size([512, 128, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn3.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn3.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.downsample.0.weight - torch.Size([512, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.downsample.1.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.downsample.1.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.conv1.weight - torch.Size([128, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn1.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn1.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.conv2.weight - torch.Size([128, 128, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn2.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn2.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.conv3.weight - torch.Size([512, 128, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn3.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn3.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.conv1.weight - torch.Size([128, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn1.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn1.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.conv2.weight - torch.Size([128, 128, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn2.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn2.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.conv3.weight - torch.Size([512, 128, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn3.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn3.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.conv1.weight - torch.Size([128, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn1.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn1.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.conv2.weight - torch.Size([128, 128, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn2.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn2.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.conv3.weight - torch.Size([512, 128, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn3.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn3.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.conv1.weight - torch.Size([256, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.downsample.0.weight - torch.Size([1024, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.downsample.1.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.downsample.1.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.conv1.weight - torch.Size([512, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn1.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn1.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.conv2.weight - torch.Size([512, 512, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn2.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn2.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.conv3.weight - torch.Size([2048, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn3.weight - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn3.bias - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.downsample.0.weight - torch.Size([2048, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.downsample.1.weight - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.downsample.1.bias - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.conv1.weight - torch.Size([512, 2048, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn1.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn1.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.conv2.weight - torch.Size([512, 512, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn2.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn2.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.conv3.weight - torch.Size([2048, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn3.weight - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn3.bias - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.conv1.weight - torch.Size([512, 2048, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn1.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn1.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.conv2.weight - torch.Size([512, 512, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn2.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn2.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.conv3.weight - torch.Size([2048, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn3.weight - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn3.bias - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_neck.lateral_convs.0.conv.weight - torch.Size([256, 256, 1, 1]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.lateral_convs.0.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.lateral_convs.1.conv.weight - torch.Size([256, 512, 1, 1]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.lateral_convs.1.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.lateral_convs.2.conv.weight - torch.Size([256, 1024, 1, 1]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.lateral_convs.2.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.lateral_convs.3.conv.weight - torch.Size([256, 2048, 1, 1]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.lateral_convs.3.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.fpn_convs.0.conv.weight - torch.Size([256, 256, 3, 3]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.fpn_convs.0.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.fpn_convs.1.conv.weight - torch.Size([256, 256, 3, 3]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.fpn_convs.1.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.fpn_convs.2.conv.weight - torch.Size([256, 256, 3, 3]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.fpn_convs.2.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.fpn_convs.3.conv.weight - torch.Size([256, 256, 3, 3]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.fpn_convs.3.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.instance_bank.anchor - torch.Size([900, 11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.instance_bank.instance_feature - torch.Size([900, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.instance_bank.anchor_handler.fix_scale - torch.Size([1, 3]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.0.weight - torch.Size([128, 3]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.0.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.2.weight - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.2.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.3.weight - torch.Size([128, 128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.3.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.5.weight - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.5.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.6.weight - torch.Size([128, 128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.6.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.8.weight - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.8.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.9.weight - torch.Size([128, 128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.9.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.11.weight - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.11.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.0.weight - torch.Size([32, 3]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.0.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.2.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.2.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.3.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.3.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.5.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.5.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.6.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.6.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.8.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.8.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.9.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.9.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.11.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.11.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.0.weight - torch.Size([32, 2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.0.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.2.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.2.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.3.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.3.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.5.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.5.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.6.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.6.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.8.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.8.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.9.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.9.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.11.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.11.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.0.weight - torch.Size([64, 3]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.0.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.2.weight - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.2.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.3.weight - torch.Size([64, 64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.3.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.5.weight - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.5.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.6.weight - torch.Size([64, 64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.6.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.8.weight - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.8.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.9.weight - torch.Size([64, 64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.9.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.11.weight - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.11.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.1.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.1.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.1.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.1.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.1.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.1.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.1.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.1.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.3.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.4.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.4.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.4.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.4.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.5.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.5.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.5.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.5.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.6.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.6.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.8.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.8.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.8.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.8.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.8.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.8.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.8.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.8.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.10.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.11.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.11.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.11.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.11.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.12.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.12.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.12.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.12.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.13.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.13.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.15.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.15.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.15.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.15.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.15.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.15.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.15.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.15.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.16.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.16.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.17.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.18.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.18.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.18.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.18.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.19.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.19.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.19.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.19.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.20.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.20.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.22.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.22.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.22.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.22.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.22.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.22.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.22.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.22.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.23.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.23.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.24.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.25.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.25.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.25.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.25.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.26.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.26.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.26.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.26.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.27.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.27.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.29.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.29.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.29.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.29.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.29.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.29.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.29.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.29.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.30.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.30.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.31.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.32.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.32.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.32.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.32.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.33.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.33.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.33.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.33.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.34.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.34.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.36.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.36.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.36.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.36.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.36.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.36.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.36.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.36.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.37.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.37.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.38.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.fc_before.weight - torch.Size([512, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.fc_after.weight - torch.Size([256, 512]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.0.weight - torch.Size([1, 256, 1, 1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.0.bias - torch.Size([1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.1.weight - torch.Size([1, 256, 1, 1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.1.bias - torch.Size([1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.2.weight - torch.Size([1, 256, 1, 1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.2.bias - torch.Size([1]): -The value is the same before and after calling `init_weights` of Sparse4D -2026-02-10 16:27:31,234 - mmdet - INFO - Model: -Sparse4D( - (img_backbone): ResNet( - (conv1): Conv2d(3, 64, kernel_size=(7, 7), stride=(2, 2), padding=(3, 3), bias=False) - (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (maxpool): MaxPool2d(kernel_size=3, stride=2, padding=1, dilation=1, ceil_mode=False) - (layer1): ResLayer( - (0): Bottleneck( - (conv1): Conv2d(64, 64, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (downsample): Sequential( - (0): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - ) - ) - (1): Bottleneck( - (conv1): Conv2d(256, 64, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (2): Bottleneck( - (conv1): Conv2d(256, 64, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - ) - (layer2): ResLayer( - (0): Bottleneck( - (conv1): Conv2d(256, 128, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (downsample): Sequential( - (0): Conv2d(256, 512, kernel_size=(1, 1), stride=(2, 2), bias=False) - (1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - ) - ) - (1): Bottleneck( - (conv1): Conv2d(512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (2): Bottleneck( - (conv1): Conv2d(512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (3): Bottleneck( - (conv1): Conv2d(512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - ) - (layer3): ResLayer( - (0): Bottleneck( - (conv1): Conv2d(512, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (downsample): Sequential( - (0): Conv2d(512, 1024, kernel_size=(1, 1), stride=(2, 2), bias=False) - (1): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - ) - ) - (1): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (2): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (3): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (4): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (5): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - ) - (layer4): ResLayer( - (0): Bottleneck( - (conv1): Conv2d(1024, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(512, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (downsample): Sequential( - (0): Conv2d(1024, 2048, kernel_size=(1, 1), stride=(2, 2), bias=False) - (1): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - ) - ) - (1): Bottleneck( - (conv1): Conv2d(2048, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(512, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (2): Bottleneck( - (conv1): Conv2d(2048, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(512, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - ) - ) - init_cfg={'type': 'Pretrained', 'checkpoint': 'ckpt/resnet50-19c8e357.pth'} - (img_neck): FPN( - (lateral_convs): ModuleList( - (0): ConvModule( - (conv): Conv2d(256, 256, kernel_size=(1, 1), stride=(1, 1)) - ) - (1): ConvModule( - (conv): Conv2d(512, 256, kernel_size=(1, 1), stride=(1, 1)) - ) - (2): ConvModule( - (conv): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1)) - ) - (3): ConvModule( - (conv): Conv2d(2048, 256, kernel_size=(1, 1), stride=(1, 1)) - ) - ) - (fpn_convs): ModuleList( - (0-3): 4 x ConvModule( - (conv): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1)) - ) - ) - ) - init_cfg={'type': 'Xavier', 'layer': 'Conv2d', 'distribution': 'uniform'} - (head): Sparse4DHead( - (instance_bank): InstanceBank( - (anchor_handler): SparseBox3DKeyPointsGenerator() - ) - (anchor_encoder): SparseBox3DEncoder( - (pos_fc): Sequential( - (0): Linear(in_features=3, out_features=128, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((128,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=128, out_features=128, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((128,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=128, out_features=128, bias=True) - (7): ReLU(inplace=True) - (8): LayerNorm((128,), eps=1e-05, elementwise_affine=True) - (9): Linear(in_features=128, out_features=128, bias=True) - (10): ReLU(inplace=True) - (11): LayerNorm((128,), eps=1e-05, elementwise_affine=True) - ) - (size_fc): Sequential( - (0): Linear(in_features=3, out_features=32, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=32, out_features=32, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=32, out_features=32, bias=True) - (7): ReLU(inplace=True) - (8): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (9): Linear(in_features=32, out_features=32, bias=True) - (10): ReLU(inplace=True) - (11): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - ) - (yaw_fc): Sequential( - (0): Linear(in_features=2, out_features=32, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=32, out_features=32, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=32, out_features=32, bias=True) - (7): ReLU(inplace=True) - (8): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (9): Linear(in_features=32, out_features=32, bias=True) - (10): ReLU(inplace=True) - (11): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - ) - (vel_fc): Sequential( - (0): Linear(in_features=3, out_features=64, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((64,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=64, out_features=64, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((64,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=64, out_features=64, bias=True) - (7): ReLU(inplace=True) - (8): LayerNorm((64,), eps=1e-05, elementwise_affine=True) - (9): Linear(in_features=64, out_features=64, bias=True) - (10): ReLU(inplace=True) - (11): LayerNorm((64,), eps=1e-05, elementwise_affine=True) - ) - ) - (loss_cls): FocalLoss() - (loss_reg): SparseBox3DLoss( - (loss_box): L1Loss() - (loss_cns): CrossEntropyLoss(avg_non_ignore=False) - (loss_yns): GaussianFocalLoss() - ) - (layers): ModuleList( - (0): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (1): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (4-5): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (6): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (7): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (8): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (11-12): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (13): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (14): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (15): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (16): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (17): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (18-19): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (20): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (21): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (22): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (23): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (24): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (25-26): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (27): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (28): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (29): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (30): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (31): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (32-33): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (34): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (35): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (36): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (37): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (38): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - ) - (fc_before): Linear(in_features=256, out_features=512, bias=False) - (fc_after): Linear(in_features=512, out_features=256, bias=False) - ) - (depth_branch): DenseDepthNet( - (depth_layers): ModuleList( - (0-2): 3 x Conv2d(256, 1, kernel_size=(1, 1), stride=(1, 1)) - ) - ) - (grid_mask): GridMask() -) -2026-02-10 16:27:45,281 - mmdet - INFO - Start running, host: root@bw28, work_dir: /workspace/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704 -2026-02-10 16:27:45,281 - mmdet - INFO - Hooks will be executed in the following order: -before_run: -(VERY_HIGH ) CosineAnnealingLrUpdaterHook -(ABOVE_NORMAL) Fp16OptimizerHook -(NORMAL ) CheckpointHook -(NORMAL ) EvalHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -before_train_epoch: -(VERY_HIGH ) CosineAnnealingLrUpdaterHook -(NORMAL ) EvalHook -(LOW ) IterTimerHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -before_train_iter: -(VERY_HIGH ) CosineAnnealingLrUpdaterHook -(NORMAL ) EvalHook -(LOW ) IterTimerHook - -------------------- -after_train_iter: -(ABOVE_NORMAL) Fp16OptimizerHook -(NORMAL ) CheckpointHook -(NORMAL ) EvalHook -(NORMAL ) ProfilerHook -(LOW ) IterTimerHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -after_train_epoch: -(NORMAL ) CheckpointHook -(NORMAL ) EvalHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -before_val_epoch: -(LOW ) IterTimerHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -before_val_iter: -(LOW ) IterTimerHook - -------------------- -after_val_iter: -(LOW ) IterTimerHook - -------------------- -after_val_epoch: -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -after_run: -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -2026-02-10 16:27:45,282 - mmdet - INFO - workflow: [('train', 1)], max: 35100 iters -2026-02-10 16:27:45,284 - mmdet - INFO - Checkpoints will be saved to /workspace/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704 by HardDiskBackend. -2026-02-10 16:28:04,348 - mmdet - INFO - Iter [1/35100] lr: 1.000e-04, eta: 7 days, 12:52:04, time: 18.551, data_time: 3.981, memory: 24610, loss_cls_0: 2.3066, loss_box_0: 0.0000, loss_cns_0: 0.0000, loss_yns_0: 0.0000, loss_cls_1: 2.1660, loss_box_1: 0.1165, loss_cns_1: 0.0237, loss_yns_1: 0.0049, loss_cls_2: 2.2464, loss_box_2: 0.0000, loss_cns_2: 0.0000, loss_yns_2: 0.0000, loss_cls_3: 2.3115, loss_box_3: 0.0524, loss_cns_3: 0.0085, loss_yns_3: 0.0008, loss_cls_4: 1.9817, loss_box_4: 0.3494, loss_cns_4: 0.0475, loss_yns_4: 0.0321, loss_cls_5: 2.3490, loss_box_5: 0.0000, loss_cns_5: 0.0000, loss_yns_5: 0.0000, loss_cls_dn_0: 1.1647, loss_box_dn_0: 1.4583, loss_cls_dn_1: 1.1569, loss_box_dn_1: 1.7270, loss_cls_dn_2: 1.1336, loss_box_dn_2: 1.9434, loss_cls_dn_3: 1.1271, loss_box_dn_3: 2.2150, loss_cls_dn_4: 1.0498, loss_box_dn_4: 2.4035, loss_cls_dn_5: 1.1831, loss_box_dn_5: 2.6805, loss_dense_depth: 1.8309, loss: 35.0708, grad_norm: 552.7689 -2026-02-10 16:28:05,953 - mmdet - INFO - Iter [2/35100] lr: 1.004e-04, eta: 4 days, 2:15:17, time: 1.605, data_time: 0.046, memory: 24610, loss_cls_0: 2.1843, loss_box_0: 0.0000, loss_cns_0: 0.0000, loss_yns_0: 0.0000, loss_cls_1: 1.9628, loss_box_1: 0.5669, loss_cns_1: 0.0800, loss_yns_1: 0.0206, loss_cls_2: 2.0682, loss_box_2: 0.0313, loss_cns_2: 0.0017, loss_yns_2: 0.0008, loss_cls_3: 1.9859, loss_box_3: 0.1833, loss_cns_3: 0.0297, loss_yns_3: 0.0126, loss_cls_4: 1.7897, loss_box_4: 1.3336, loss_cns_4: 0.1416, loss_yns_4: 0.0456, loss_cls_5: 1.9402, loss_box_5: 1.2836, loss_cns_5: 0.1587, loss_yns_5: 0.0651, loss_cls_dn_0: 1.1054, loss_box_dn_0: 1.2616, loss_cls_dn_1: 0.9639, loss_box_dn_1: 2.5890, loss_cls_dn_2: 0.9461, loss_box_dn_2: 2.6759, loss_cls_dn_3: 0.9394, loss_box_dn_3: 2.7226, loss_cls_dn_4: 0.8565, loss_box_dn_4: 2.8931, loss_cls_dn_5: 0.9667, loss_box_dn_5: 3.1169, loss_dense_depth: 1.7817, loss: 38.7050, grad_norm: 82.5949 -2026-02-10 16:28:07,056 - mmdet - INFO - Iter [3/35100] lr: 1.008e-04, eta: 2 days, 21:05:17, time: 1.104, data_time: 0.040, memory: 24610, loss_cls_0: 1.6516, loss_box_0: 1.9371, loss_cns_0: 0.5917, loss_yns_0: 0.5043, loss_cls_1: 1.7872, loss_box_1: 1.8665, loss_cns_1: 0.3044, loss_yns_1: 0.0811, loss_cls_2: 1.7317, loss_box_2: 2.5637, loss_cns_2: 0.3186, loss_yns_2: 0.1233, loss_cls_3: 1.6194, loss_box_3: 3.4816, loss_cns_3: 0.4248, loss_yns_3: 0.2382, loss_cls_4: 1.5643, loss_box_4: 4.1664, loss_cns_4: 0.4432, loss_yns_4: 0.1240, loss_cls_5: 1.6238, loss_box_5: 4.0111, loss_cns_5: 0.3518, loss_yns_5: 0.1662, loss_cls_dn_0: 0.8184, loss_box_dn_0: 1.1690, loss_cls_dn_1: 0.8201, loss_box_dn_1: 2.3224, loss_cls_dn_2: 0.7816, loss_box_dn_2: 2.4949, loss_cls_dn_3: 0.7246, loss_box_dn_3: 2.7203, loss_cls_dn_4: 0.7022, loss_box_dn_4: 2.9195, loss_cls_dn_5: 0.7728, loss_box_dn_5: 3.1393, loss_dense_depth: 1.7989, loss: 52.8600, grad_norm: 120.7034 -2026-02-10 16:28:08,161 - mmdet - INFO - Iter [4/35100] lr: 1.012e-04, eta: 2 days, 6:30:27, time: 1.105, data_time: 0.040, memory: 24610, loss_cls_0: 1.4669, loss_box_0: 2.8534, loss_cns_0: 0.6930, loss_yns_0: 0.2783, loss_cls_1: 1.6797, loss_box_1: 2.6238, loss_cns_1: 0.3208, loss_yns_1: 0.1076, loss_cls_2: 1.6314, loss_box_2: 3.5497, loss_cns_2: 0.4407, loss_yns_2: 0.1609, loss_cls_3: 1.4964, loss_box_3: 3.6841, loss_cns_3: 0.4887, loss_yns_3: 0.2342, loss_cls_4: 1.4394, loss_box_4: 4.2500, loss_cns_4: 0.4115, loss_yns_4: 0.2266, loss_cls_5: 1.5408, loss_box_5: 4.6250, loss_cns_5: 0.4529, loss_yns_5: 0.1983, loss_cls_dn_0: 0.6844, loss_box_dn_0: 1.2427, loss_cls_dn_1: 0.7474, loss_box_dn_1: 2.4669, loss_cls_dn_2: 0.7041, loss_box_dn_2: 2.5019, loss_cls_dn_3: 0.6555, loss_box_dn_3: 2.7140, loss_cls_dn_4: 0.6263, loss_box_dn_4: 2.8782, loss_cls_dn_5: 0.6684, loss_box_dn_5: 3.0910, loss_dense_depth: 1.6706, loss: 55.5056, grad_norm: 119.5921 -2026-02-10 16:28:09,270 - mmdet - INFO - Iter [5/35100] lr: 1.016e-04, eta: 1 day, 21:45:56, time: 1.108, data_time: 0.041, memory: 24610, loss_cls_0: 1.4241, loss_box_0: 2.8527, loss_cns_0: 0.5541, loss_yns_0: 0.1646, loss_cls_1: 1.5953, loss_box_1: 3.8224, loss_cns_1: 0.3895, loss_yns_1: 0.1738, loss_cls_2: 1.5701, loss_box_2: 4.0394, loss_cns_2: 0.4190, loss_yns_2: 0.1675, loss_cls_3: 1.4107, loss_box_3: 4.0328, loss_cns_3: 0.4511, loss_yns_3: 0.1951, loss_cls_4: 1.4841, loss_box_4: 4.0425, loss_cns_4: 0.4325, loss_yns_4: 0.1944, loss_cls_5: 1.4769, loss_box_5: 4.1814, loss_cns_5: 0.4715, loss_yns_5: 0.1773, loss_cls_dn_0: 0.5825, loss_box_dn_0: 1.2238, loss_cls_dn_1: 0.6907, loss_box_dn_1: 2.4086, loss_cls_dn_2: 0.6706, loss_box_dn_2: 2.4778, loss_cls_dn_3: 0.5847, loss_box_dn_3: 2.5950, loss_cls_dn_4: 0.5572, loss_box_dn_4: 2.7483, loss_cls_dn_5: 0.5976, loss_box_dn_5: 2.8383, loss_dense_depth: 1.7168, loss: 55.4151, grad_norm: 127.8555 -2026-02-10 16:28:10,380 - mmdet - INFO - Iter [6/35100] lr: 1.020e-04, eta: 1 day, 15:56:24, time: 1.110, data_time: 0.037, memory: 24610, loss_cls_0: 1.3911, loss_box_0: 2.6336, loss_cns_0: 0.5805, loss_yns_0: 0.1911, loss_cls_1: 1.5284, loss_box_1: 3.8461, loss_cns_1: 0.4590, loss_yns_1: 0.1794, loss_cls_2: 1.5435, loss_box_2: 3.8214, loss_cns_2: 0.4694, loss_yns_2: 0.1675, loss_cls_3: 1.4735, loss_box_3: 3.6338, loss_cns_3: 0.4970, loss_yns_3: 0.2105, loss_cls_4: 1.4232, loss_box_4: 3.7274, loss_cns_4: 0.4831, loss_yns_4: 0.1999, loss_cls_5: 1.3667, loss_box_5: 3.8768, loss_cns_5: 0.4934, loss_yns_5: 0.1956, loss_cls_dn_0: 0.5319, loss_box_dn_0: 1.1525, loss_cls_dn_1: 0.6488, loss_box_dn_1: 2.6211, loss_cls_dn_2: 0.6033, loss_box_dn_2: 2.5429, loss_cls_dn_3: 0.5313, loss_box_dn_3: 2.5955, loss_cls_dn_4: 0.5285, loss_box_dn_4: 2.7064, loss_cls_dn_5: 0.5459, loss_box_dn_5: 2.8212, loss_dense_depth: 1.6406, loss: 53.8619, grad_norm: 96.4519 -2026-02-10 16:28:11,492 - mmdet - INFO - Iter [7/35100] lr: 1.024e-04, eta: 1 day, 11:46:58, time: 1.113, data_time: 0.041, memory: 24610, loss_cls_0: 1.3129, loss_box_0: 2.3846, loss_cns_0: 0.6822, loss_yns_0: 0.2316, loss_cls_1: 1.4656, loss_box_1: 3.6979, loss_cns_1: 0.4907, loss_yns_1: 0.1879, loss_cls_2: 1.5858, loss_box_2: 3.6480, loss_cns_2: 0.4843, loss_yns_2: 0.1569, loss_cls_3: 1.4454, loss_box_3: 3.6136, loss_cns_3: 0.4681, loss_yns_3: 0.2082, loss_cls_4: 1.4447, loss_box_4: 3.6879, loss_cns_4: 0.5112, loss_yns_4: 0.1848, loss_cls_5: 1.3505, loss_box_5: 3.8406, loss_cns_5: 0.5406, loss_yns_5: 0.2162, loss_cls_dn_0: 0.5396, loss_box_dn_0: 1.0971, loss_cls_dn_1: 0.5922, loss_box_dn_1: 2.7835, loss_cls_dn_2: 0.5513, loss_box_dn_2: 2.7011, loss_cls_dn_3: 0.4883, loss_box_dn_3: 2.6916, loss_cls_dn_4: 0.4963, loss_box_dn_4: 2.7747, loss_cls_dn_5: 0.4965, loss_box_dn_5: 2.9268, loss_dense_depth: 1.5759, loss: 53.5551, grad_norm: 87.8074 -2026-02-10 16:28:12,599 - mmdet - INFO - Iter [8/35100] lr: 1.028e-04, eta: 1 day, 8:39:26, time: 1.106, data_time: 0.041, memory: 24610, loss_cls_0: 1.2824, loss_box_0: 2.2970, loss_cns_0: 0.6683, loss_yns_0: 0.1905, loss_cls_1: 1.3998, loss_box_1: 3.3490, loss_cns_1: 0.4807, loss_yns_1: 0.1801, loss_cls_2: 1.4719, loss_box_2: 3.5176, loss_cns_2: 0.4757, loss_yns_2: 0.1767, loss_cls_3: 1.3627, loss_box_3: 3.4969, loss_cns_3: 0.4584, loss_yns_3: 0.1827, loss_cls_4: 1.4338, loss_box_4: 3.4934, loss_cns_4: 0.4915, loss_yns_4: 0.1700, loss_cls_5: 1.3996, loss_box_5: 3.5806, loss_cns_5: 0.5466, loss_yns_5: 0.1794, loss_cls_dn_0: 0.5301, loss_box_dn_0: 1.0490, loss_cls_dn_1: 0.5879, loss_box_dn_1: 1.6883, loss_cls_dn_2: 0.5600, loss_box_dn_2: 1.7368, loss_cls_dn_3: 0.5020, loss_box_dn_3: 1.7807, loss_cls_dn_4: 0.5086, loss_box_dn_4: 1.8833, loss_cls_dn_5: 0.4914, loss_box_dn_5: 1.9710, loss_dense_depth: 1.5308, loss: 47.1056, grad_norm: 90.8255 -2026-02-10 16:28:13,705 - mmdet - INFO - Iter [9/35100] lr: 1.032e-04, eta: 1 day, 6:13:32, time: 1.106, data_time: 0.036, memory: 24610, loss_cls_0: 1.2589, loss_box_0: 2.2269, loss_cns_0: 0.6765, loss_yns_0: 0.1684, loss_cls_1: 1.4357, loss_box_1: 3.5077, loss_cns_1: 0.4700, loss_yns_1: 0.1706, loss_cls_2: 1.3769, loss_box_2: 3.6324, loss_cns_2: 0.4773, loss_yns_2: 0.1742, loss_cls_3: 1.3150, loss_box_3: 3.6767, loss_cns_3: 0.4930, loss_yns_3: 0.1941, loss_cls_4: 1.3738, loss_box_4: 3.6514, loss_cns_4: 0.5132, loss_yns_4: 0.1831, loss_cls_5: 1.3503, loss_box_5: 3.7080, loss_cns_5: 0.4964, loss_yns_5: 0.1611, loss_cls_dn_0: 0.5300, loss_box_dn_0: 1.0492, loss_cls_dn_1: 0.5293, loss_box_dn_1: 1.5668, loss_cls_dn_2: 0.5585, loss_box_dn_2: 1.6630, loss_cls_dn_3: 0.4946, loss_box_dn_3: 1.7994, loss_cls_dn_4: 0.5146, loss_box_dn_4: 1.8399, loss_cls_dn_5: 0.4801, loss_box_dn_5: 1.9119, loss_dense_depth: 1.4425, loss: 47.0716, grad_norm: 94.8319 -2026-02-10 16:28:14,811 - mmdet - INFO - Iter [10/35100] lr: 1.036e-04, eta: 1 day, 4:16:50, time: 1.106, data_time: 0.035, memory: 24610, loss_cls_0: 1.2727, loss_box_0: 2.2852, loss_cns_0: 0.6205, loss_yns_0: 0.1578, loss_cls_1: 1.3521, loss_box_1: 3.2777, loss_cns_1: 0.4673, loss_yns_1: 0.1825, loss_cls_2: 1.3386, loss_box_2: 3.2648, loss_cns_2: 0.4842, loss_yns_2: 0.1558, loss_cls_3: 1.2767, loss_box_3: 3.4457, loss_cns_3: 0.5209, loss_yns_3: 0.1693, loss_cls_4: 1.2861, loss_box_4: 3.4148, loss_cns_4: 0.4939, loss_yns_4: 0.1909, loss_cls_5: 1.2983, loss_box_5: 3.4927, loss_cns_5: 0.4649, loss_yns_5: 0.1701, loss_cls_dn_0: 0.5182, loss_box_dn_0: 1.0704, loss_cls_dn_1: 0.4808, loss_box_dn_1: 1.7781, loss_cls_dn_2: 0.5368, loss_box_dn_2: 1.8065, loss_cls_dn_3: 0.4801, loss_box_dn_3: 1.9056, loss_cls_dn_4: 0.4812, loss_box_dn_4: 1.9307, loss_cls_dn_5: 0.4545, loss_box_dn_5: 1.9834, loss_dense_depth: 1.4933, loss: 46.0032, grad_norm: 79.2678 -2026-02-10 16:28:15,999 - mmdet - INFO - Iter [11/35100] lr: 1.040e-04, eta: 1 day, 2:45:42, time: 1.188, data_time: 0.097, memory: 24610, loss_cls_0: 1.2265, loss_box_0: 2.2948, loss_cns_0: 0.6140, loss_yns_0: 0.1594, loss_cls_1: 1.2282, loss_box_1: 3.1768, loss_cns_1: 0.5345, loss_yns_1: 0.1626, loss_cls_2: 1.3719, loss_box_2: 3.0956, loss_cns_2: 0.5294, loss_yns_2: 0.1693, loss_cls_3: 1.2710, loss_box_3: 3.2119, loss_cns_3: 0.5452, loss_yns_3: 0.1804, loss_cls_4: 1.2694, loss_box_4: 3.2011, loss_cns_4: 0.5214, loss_yns_4: 0.1759, loss_cls_5: 1.2900, loss_box_5: 3.2229, loss_cns_5: 0.5054, loss_yns_5: 0.1876, loss_cls_dn_0: 0.4637, loss_box_dn_0: 1.0932, loss_cls_dn_1: 0.4391, loss_box_dn_1: 1.9481, loss_cls_dn_2: 0.4979, loss_box_dn_2: 1.9292, loss_cls_dn_3: 0.4584, loss_box_dn_3: 1.9919, loss_cls_dn_4: 0.4455, loss_box_dn_4: 1.9882, loss_cls_dn_5: 0.4347, loss_box_dn_5: 2.0279, loss_dense_depth: 1.4198, loss: 45.2828, grad_norm: 61.1572 -2026-02-10 16:28:17,125 - mmdet - INFO - Iter [12/35100] lr: 1.044e-04, eta: 1 day, 1:26:44, time: 1.126, data_time: 0.041, memory: 24610, loss_cls_0: 1.2203, loss_box_0: 2.2783, loss_cns_0: 0.6060, loss_yns_0: 0.1686, loss_cls_1: 1.2412, loss_box_1: 3.1833, loss_cns_1: 0.4919, loss_yns_1: 0.1737, loss_cls_2: 1.3211, loss_box_2: 3.1436, loss_cns_2: 0.5047, loss_yns_2: 0.2045, loss_cls_3: 1.3035, loss_box_3: 3.1612, loss_cns_3: 0.5060, loss_yns_3: 0.1699, loss_cls_4: 1.2742, loss_box_4: 3.1987, loss_cns_4: 0.4977, loss_yns_4: 0.1921, loss_cls_5: 1.2835, loss_box_5: 3.2411, loss_cns_5: 0.5093, loss_yns_5: 0.1748, loss_cls_dn_0: 0.4449, loss_box_dn_0: 1.0400, loss_cls_dn_1: 0.4312, loss_box_dn_1: 2.2757, loss_cls_dn_2: 0.4716, loss_box_dn_2: 2.2550, loss_cls_dn_3: 0.4482, loss_box_dn_3: 2.2576, loss_cls_dn_4: 0.4346, loss_box_dn_4: 2.2688, loss_cls_dn_5: 0.4285, loss_box_dn_5: 2.3025, loss_dense_depth: 1.3996, loss: 46.5075, grad_norm: 65.2148 -2026-02-10 16:28:18,241 - mmdet - INFO - Iter [13/35100] lr: 1.048e-04, eta: 1 day, 0:19:26, time: 1.116, data_time: 0.043, memory: 24610, loss_cls_0: 1.2053, loss_box_0: 2.2993, loss_cns_0: 0.6206, loss_yns_0: 0.1592, loss_cls_1: 1.2885, loss_box_1: 3.1507, loss_cns_1: 0.5209, loss_yns_1: 0.1575, loss_cls_2: 1.2577, loss_box_2: 3.2292, loss_cns_2: 0.5113, loss_yns_2: 0.1773, loss_cls_3: 1.2607, loss_box_3: 3.2341, loss_cns_3: 0.4950, loss_yns_3: 0.1634, loss_cls_4: 1.2623, loss_box_4: 3.3537, loss_cns_4: 0.5141, loss_yns_4: 0.1716, loss_cls_5: 1.2797, loss_box_5: 3.4863, loss_cns_5: 0.5044, loss_yns_5: 0.1829, loss_cls_dn_0: 0.4323, loss_box_dn_0: 1.0276, loss_cls_dn_1: 0.4378, loss_box_dn_1: 1.7616, loss_cls_dn_2: 0.4630, loss_box_dn_2: 1.8028, loss_cls_dn_3: 0.4483, loss_box_dn_3: 1.8596, loss_cls_dn_4: 0.4382, loss_box_dn_4: 1.9779, loss_cls_dn_5: 0.4410, loss_box_dn_5: 2.0556, loss_dense_depth: 1.3251, loss: 44.9564, grad_norm: 83.3312 -2026-02-10 16:28:19,353 - mmdet - INFO - Iter [14/35100] lr: 1.052e-04, eta: 23:21:37, time: 1.112, data_time: 0.036, memory: 24610, loss_cls_0: 1.1911, loss_box_0: 2.3300, loss_cns_0: 0.6232, loss_yns_0: 0.1765, loss_cls_1: 1.2635, loss_box_1: 2.7008, loss_cns_1: 0.5619, loss_yns_1: 0.1780, loss_cls_2: 1.2408, loss_box_2: 2.8155, loss_cns_2: 0.5566, loss_yns_2: 0.1763, loss_cls_3: 1.2242, loss_box_3: 2.8422, loss_cns_3: 0.5433, loss_yns_3: 0.1756, loss_cls_4: 1.2140, loss_box_4: 2.8720, loss_cns_4: 0.5444, loss_yns_4: 0.1737, loss_cls_5: 1.2449, loss_box_5: 2.9296, loss_cns_5: 0.5567, loss_yns_5: 0.1726, loss_cls_dn_0: 0.4522, loss_box_dn_0: 1.0189, loss_cls_dn_1: 0.4686, loss_box_dn_1: 1.5461, loss_cls_dn_2: 0.4636, loss_box_dn_2: 1.5785, loss_cls_dn_3: 0.4489, loss_box_dn_3: 1.6056, loss_cls_dn_4: 0.4364, loss_box_dn_4: 1.6546, loss_cls_dn_5: 0.4571, loss_box_dn_5: 1.7257, loss_dense_depth: 1.4500, loss: 41.6133, grad_norm: 76.3792 -2026-02-10 16:28:20,480 - mmdet - INFO - Iter [15/35100] lr: 1.056e-04, eta: 22:32:04, time: 1.127, data_time: 0.040, memory: 24610, loss_cls_0: 1.0828, loss_box_0: 2.3990, loss_cns_0: 0.6060, loss_yns_0: 0.2007, loss_cls_1: 1.1359, loss_box_1: 2.5313, loss_cns_1: 0.5609, loss_yns_1: 0.1792, loss_cls_2: 1.2859, loss_box_2: 2.5029, loss_cns_2: 0.5592, loss_yns_2: 0.2055, loss_cls_3: 1.2459, loss_box_3: 2.5938, loss_cns_3: 0.5521, loss_yns_3: 0.1943, loss_cls_4: 1.1467, loss_box_4: 2.6475, loss_cns_4: 0.5460, loss_yns_4: 0.1745, loss_cls_5: 1.2028, loss_box_5: 2.6986, loss_cns_5: 0.5558, loss_yns_5: 0.2140, loss_cls_dn_0: 0.4609, loss_box_dn_0: 1.0376, loss_cls_dn_1: 0.4774, loss_box_dn_1: 1.4297, loss_cls_dn_2: 0.5261, loss_box_dn_2: 1.3838, loss_cls_dn_3: 0.4898, loss_box_dn_3: 1.4458, loss_cls_dn_4: 0.4735, loss_box_dn_4: 1.4632, loss_cls_dn_5: 0.5188, loss_box_dn_5: 1.5223, loss_dense_depth: 1.3272, loss: 39.5779, grad_norm: 84.2397 -2026-02-10 16:28:21,598 - mmdet - INFO - Iter [16/35100] lr: 1.060e-04, eta: 21:48:23, time: 1.118, data_time: 0.035, memory: 24610, loss_cls_0: 1.1034, loss_box_0: 2.3098, loss_cns_0: 0.5918, loss_yns_0: 0.2083, loss_cls_1: 1.2108, loss_box_1: 2.7324, loss_cns_1: 0.5375, loss_yns_1: 0.1837, loss_cls_2: 1.4188, loss_box_2: 2.6890, loss_cns_2: 0.5249, loss_yns_2: 0.1940, loss_cls_3: 1.3647, loss_box_3: 2.7971, loss_cns_3: 0.5232, loss_yns_3: 0.1919, loss_cls_4: 1.1757, loss_box_4: 2.8400, loss_cns_4: 0.5144, loss_yns_4: 0.1838, loss_cls_5: 1.2448, loss_box_5: 2.9347, loss_cns_5: 0.5232, loss_yns_5: 0.2205, loss_cls_dn_0: 0.4706, loss_box_dn_0: 1.0331, loss_cls_dn_1: 0.4821, loss_box_dn_1: 1.4334, loss_cls_dn_2: 0.5325, loss_box_dn_2: 1.3502, loss_cls_dn_3: 0.4900, loss_box_dn_3: 1.4481, loss_cls_dn_4: 0.4506, loss_box_dn_4: 1.5033, loss_cls_dn_5: 0.5165, loss_box_dn_5: 1.5951, loss_dense_depth: 1.2708, loss: 40.7947, grad_norm: 77.7138 -2026-02-10 16:28:22,721 - mmdet - INFO - Iter [17/35100] lr: 1.064e-04, eta: 21:10:00, time: 1.123, data_time: 0.036, memory: 24610, loss_cls_0: 1.1043, loss_box_0: 2.3164, loss_cns_0: 0.5783, loss_yns_0: 0.1934, loss_cls_1: 1.1852, loss_box_1: 2.8828, loss_cns_1: 0.5209, loss_yns_1: 0.1853, loss_cls_2: 1.2838, loss_box_2: 2.8597, loss_cns_2: 0.5075, loss_yns_2: 0.1886, loss_cls_3: 1.2468, loss_box_3: 2.8998, loss_cns_3: 0.4992, loss_yns_3: 0.1853, loss_cls_4: 1.1720, loss_box_4: 2.9977, loss_cns_4: 0.4945, loss_yns_4: 0.1936, loss_cls_5: 1.2101, loss_box_5: 3.0621, loss_cns_5: 0.4817, loss_yns_5: 0.1898, loss_cls_dn_0: 0.4550, loss_box_dn_0: 1.0241, loss_cls_dn_1: 0.4582, loss_box_dn_1: 1.6158, loss_cls_dn_2: 0.5018, loss_box_dn_2: 1.5630, loss_cls_dn_3: 0.4645, loss_box_dn_3: 1.6366, loss_cls_dn_4: 0.4244, loss_box_dn_4: 1.7343, loss_cls_dn_5: 0.4826, loss_box_dn_5: 1.8274, loss_dense_depth: 1.2707, loss: 41.8970, grad_norm: 67.2793 -2026-02-10 16:28:23,843 - mmdet - INFO - Iter [18/35100] lr: 1.068e-04, eta: 20:35:51, time: 1.122, data_time: 0.037, memory: 24610, loss_cls_0: 1.0876, loss_box_0: 2.2832, loss_cns_0: 0.5634, loss_yns_0: 0.1836, loss_cls_1: 1.1754, loss_box_1: 2.7271, loss_cns_1: 0.5107, loss_yns_1: 0.1781, loss_cls_2: 1.1964, loss_box_2: 2.7907, loss_cns_2: 0.4848, loss_yns_2: 0.1820, loss_cls_3: 1.1768, loss_box_3: 2.8110, loss_cns_3: 0.4839, loss_yns_3: 0.1997, loss_cls_4: 1.2307, loss_box_4: 2.8862, loss_cns_4: 0.4785, loss_yns_4: 0.1851, loss_cls_5: 1.1828, loss_box_5: 2.9169, loss_cns_5: 0.4742, loss_yns_5: 0.1826, loss_cls_dn_0: 0.4321, loss_box_dn_0: 1.0261, loss_cls_dn_1: 0.4421, loss_box_dn_1: 1.6792, loss_cls_dn_2: 0.4822, loss_box_dn_2: 1.6705, loss_cls_dn_3: 0.4513, loss_box_dn_3: 1.7151, loss_cls_dn_4: 0.4054, loss_box_dn_4: 1.7874, loss_cls_dn_5: 0.4510, loss_box_dn_5: 1.8567, loss_dense_depth: 1.2686, loss: 41.2388, grad_norm: 64.4959 -2026-02-10 16:28:24,983 - mmdet - INFO - Iter [19/35100] lr: 1.072e-04, eta: 20:05:51, time: 1.140, data_time: 0.044, memory: 24610, loss_cls_0: 1.0779, loss_box_0: 2.2639, loss_cns_0: 0.5621, loss_yns_0: 0.1812, loss_cls_1: 1.1600, loss_box_1: 2.6478, loss_cns_1: 0.5205, loss_yns_1: 0.1791, loss_cls_2: 1.2247, loss_box_2: 2.6741, loss_cns_2: 0.5231, loss_yns_2: 0.1820, loss_cls_3: 1.2330, loss_box_3: 2.6417, loss_cns_3: 0.5330, loss_yns_3: 0.1932, loss_cls_4: 1.2158, loss_box_4: 2.6783, loss_cns_4: 0.5175, loss_yns_4: 0.1788, loss_cls_5: 1.2959, loss_box_5: 2.7224, loss_cns_5: 0.5280, loss_yns_5: 0.1927, loss_cls_dn_0: 0.4185, loss_box_dn_0: 1.0249, loss_cls_dn_1: 0.4357, loss_box_dn_1: 1.2703, loss_cls_dn_2: 0.4680, loss_box_dn_2: 1.3778, loss_cls_dn_3: 0.4501, loss_box_dn_3: 1.4014, loss_cls_dn_4: 0.4050, loss_box_dn_4: 1.4670, loss_cls_dn_5: 0.4333, loss_box_dn_5: 1.5359, loss_dense_depth: 1.2540, loss: 39.0685, grad_norm: 64.5376 -2026-02-10 16:28:26,099 - mmdet - INFO - Iter [20/35100] lr: 1.076e-04, eta: 19:38:09, time: 1.116, data_time: 0.040, memory: 24610, loss_cls_0: 1.0710, loss_box_0: 2.2106, loss_cns_0: 0.5705, loss_yns_0: 0.1792, loss_cls_1: 1.1213, loss_box_1: 2.8212, loss_cns_1: 0.5432, loss_yns_1: 0.1844, loss_cls_2: 1.2260, loss_box_2: 2.8577, loss_cns_2: 0.5459, loss_yns_2: 0.1946, loss_cls_3: 1.1766, loss_box_3: 2.8723, loss_cns_3: 0.5473, loss_yns_3: 0.1772, loss_cls_4: 1.1270, loss_box_4: 2.9436, loss_cns_4: 0.5350, loss_yns_4: 0.1834, loss_cls_5: 1.2566, loss_box_5: 3.0706, loss_cns_5: 0.5134, loss_yns_5: 0.2013, loss_cls_dn_0: 0.4075, loss_box_dn_0: 1.0400, loss_cls_dn_1: 0.3999, loss_box_dn_1: 1.4761, loss_cls_dn_2: 0.4209, loss_box_dn_2: 1.5305, loss_cls_dn_3: 0.4175, loss_box_dn_3: 1.5210, loss_cls_dn_4: 0.3817, loss_box_dn_4: 1.5394, loss_cls_dn_5: 0.3958, loss_box_dn_5: 1.6167, loss_dense_depth: 1.1804, loss: 40.4567, grad_norm: 80.6266 -2026-02-10 16:28:27,317 - mmdet - INFO - Iter [21/35100] lr: 1.080e-04, eta: 19:15:55, time: 1.217, data_time: 0.088, memory: 24610, loss_cls_0: 1.0859, loss_box_0: 2.2495, loss_cns_0: 0.5802, loss_yns_0: 0.1762, loss_cls_1: 1.1047, loss_box_1: 2.8579, loss_cns_1: 0.5124, loss_yns_1: 0.1829, loss_cls_2: 1.1825, loss_box_2: 2.8607, loss_cns_2: 0.5165, loss_yns_2: 0.1964, loss_cls_3: 1.1592, loss_box_3: 2.8799, loss_cns_3: 0.5418, loss_yns_3: 0.1757, loss_cls_4: 1.1152, loss_box_4: 2.9177, loss_cns_4: 0.5501, loss_yns_4: 0.1851, loss_cls_5: 1.1671, loss_box_5: 3.0284, loss_cns_5: 0.5326, loss_yns_5: 0.1850, loss_cls_dn_0: 0.4061, loss_box_dn_0: 1.0428, loss_cls_dn_1: 0.4052, loss_box_dn_1: 1.2745, loss_cls_dn_2: 0.4126, loss_box_dn_2: 1.3306, loss_cls_dn_3: 0.4244, loss_box_dn_3: 1.3621, loss_cls_dn_4: 0.3947, loss_box_dn_4: 1.4019, loss_cls_dn_5: 0.4015, loss_box_dn_5: 1.4620, loss_dense_depth: 1.1845, loss: 39.4464, grad_norm: 75.1468 -2026-02-10 16:28:32,017 - mmdet - INFO - Iter [22/35100] lr: 1.084e-04, eta: 20:28:12, time: 4.699, data_time: 0.047, memory: 24610, loss_cls_0: 1.0320, loss_box_0: 2.2304, loss_cns_0: 0.5760, loss_yns_0: 0.1714, loss_cls_1: 1.0577, loss_box_1: 3.0822, loss_cns_1: 0.4876, loss_yns_1: 0.1841, loss_cls_2: 1.1087, loss_box_2: 3.1011, loss_cns_2: 0.4905, loss_yns_2: 0.2002, loss_cls_3: 1.1153, loss_box_3: 3.0700, loss_cns_3: 0.5009, loss_yns_3: 0.1928, loss_cls_4: 1.1054, loss_box_4: 3.0852, loss_cns_4: 0.4992, loss_yns_4: 0.1716, loss_cls_5: 1.1099, loss_box_5: 3.1077, loss_cns_5: 0.5031, loss_yns_5: 0.1812, loss_cls_dn_0: 0.4173, loss_box_dn_0: 1.0361, loss_cls_dn_1: 0.4183, loss_box_dn_1: 1.2250, loss_cls_dn_2: 0.4188, loss_box_dn_2: 1.2762, loss_cls_dn_3: 0.4383, loss_box_dn_3: 1.3381, loss_cls_dn_4: 0.4108, loss_box_dn_4: 1.4603, loss_cls_dn_5: 0.4226, loss_box_dn_5: 1.4730, loss_dense_depth: 1.1969, loss: 39.8961, grad_norm: 70.0092 diff --git a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20260210_162728.log.json b/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20260210_162728.log.json deleted file mode 100644 index feaa33d0778a404ea9f4dc622fc84d5fc8419f1d..0000000000000000000000000000000000000000 --- a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20260210_162728.log.json +++ /dev/null @@ -1,23 +0,0 @@ -{"env_info": "sys.platform: linux\nPython: 3.10.12 (main, Aug 15 2025, 14:32:43) [GCC 11.4.0]\nCUDA available: True\nGPU 0,1,2,3,4,5,6,7: BW200_LC, UBB BW1000_LC\nCUDA_HOME: /opt/dtk\nNVCC: Not Available\nGCC: x86_64-linux-gnu-gcc (Ubuntu 11.4.0-1ubuntu1~22.04) 11.4.0\nPyTorch: 2.5.1\nPyTorch compiling details: PyTorch built with:\n - GCC 10.3\n - C++ Version: 201703\n - Intel(R) Math Kernel Library Version 2020.0.4 Product Build 20200917 for Intel(R) 64 architecture applications\n - OpenMP 201511 (a.k.a. OpenMP 4.5)\n - LAPACK is enabled (usually provided by MKL)\n - NNPACK is enabled\n - CPU capability usage: AVX2\n - HIP Runtime 6.3.25521\n - MIOpen 2.18.0\n - Magma 2.8.0\n - Build settings: BLAS_INFO=mkl, BUILD_TYPE=Release, CXX_COMPILER=/opt/rh/gcc-toolset-10/root/usr/bin/c++, CXX_FLAGS= -D_GLIBCXX_USE_CXX11_ABI=1 -fvisibility-inlines-hidden -DUSE_PTHREADPOOL -DNDEBUG -DUSE_KINETO -DLIBKINETO_NOCUPTI -DLIBKINETO_NOXPUPTI=ON -DUSE_FBGEMM -DUSE_PYTORCH_QNNPACK -DUSE_XNNPACK -DSYMBOLICATE_MOBILE_DEBUG_HANDLE -O2 -fPIC -Wall -Wextra -Werror=return-type -Werror=non-virtual-dtor -Werror=bool-operation -Wnarrowing -Wno-missing-field-initializers -Wno-type-limits -Wno-array-bounds -Wno-unknown-pragmas -Wno-unused-parameter -Wno-strict-overflow -Wno-strict-aliasing -Wno-stringop-overflow -Wsuggest-override -Wno-psabi -Wno-error=old-style-cast -Wno-missing-braces -fdiagnostics-color=always -faligned-new -Wno-unused-but-set-variable -Wno-maybe-uninitialized -fno-math-errno -fno-trapping-math -Werror=format -Wno-stringop-overflow, FORCE_FALLBACK_CUDA_MPI=1, LAPACK_INFO=mkl, PERF_WITH_AVX=1, PERF_WITH_AVX2=1, TORCH_VERSION=2.5.1, USE_CUDA=0, USE_CUDNN=OFF, USE_CUSPARSELT=OFF, USE_EXCEPTION_PTR=1, USE_GFLAGS=1, USE_GLOG=1, USE_GLOO=1, USE_MKL=ON, USE_MKLDNN=0, USE_MPI=1, USE_NCCL=1, USE_NNPACK=ON, USE_OPENMP=1, USE_ROCM=ON, USE_ROCM_KERNEL_ASSERT=OFF, \n\nTorchVision: 0.20.1\nOpenCV: 4.12.0\nMMCV: 1.6.1\nMMCV Compiler: GCC 10.3\nMMCV CUDA Compiler: rocm not available\nMMDetection: 2.25.1+", "config": "plugin = True\nplugin_dir = 'projects/mmdet3d_plugin/'\ndist_params = dict(backend='nccl')\nlog_level = 'INFO'\nwork_dir = './work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704'\ntotal_batch_size = 80\nnum_gpus = 8\nbatch_size = 10\nnum_iters_per_epoch = 351\nnum_epochs = 100\ncheckpoint_epoch_interval = 20\ncheckpoint_config = dict(interval=7020)\nlog_config = dict(\n interval=1,\n hooks=[\n dict(type='TextLoggerHook', by_epoch=False),\n dict(type='TensorboardLoggerHook')\n ])\nload_from = None\nresume_from = None\nworkflow = [('train', 1)]\nfp16 = dict(loss_scale=32.0)\ninput_shape = (704, 256)\ntracking_test = True\ntracking_threshold = 0.2\nclass_names = [\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier',\n 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n]\nnum_classes = 10\nembed_dims = 256\nnum_groups = 8\nnum_decoder = 6\nnum_single_frame_decoder = 1\nuse_deformable_func = True\nstrides = [4, 8, 16, 32]\nnum_levels = 4\nnum_depth_layers = 3\ndrop_out = 0.1\ntemporal = True\ndecouple_attn = True\nwith_quality_estimation = True\nmodel = dict(\n type='Sparse4D',\n use_grid_mask=True,\n use_deformable_func=True,\n img_backbone=dict(\n type='ResNet',\n depth=50,\n num_stages=4,\n frozen_stages=-1,\n norm_eval=False,\n style='pytorch',\n with_cp=True,\n out_indices=(0, 1, 2, 3),\n norm_cfg=dict(type='BN', requires_grad=True),\n pretrained='ckpt/resnet50-19c8e357.pth'),\n img_neck=dict(\n type='FPN',\n num_outs=4,\n start_level=0,\n out_channels=256,\n add_extra_convs='on_output',\n relu_before_extra_convs=True,\n in_channels=[256, 512, 1024, 2048]),\n depth_branch=dict(\n type='DenseDepthNet',\n embed_dims=256,\n num_depth_layers=3,\n loss_weight=0.2),\n head=dict(\n type='Sparse4DHead',\n cls_threshold_to_reg=0.05,\n decouple_attn=True,\n instance_bank=dict(\n type='InstanceBank',\n num_anchor=900,\n embed_dims=256,\n anchor='nuscenes_kmeans900.npy',\n anchor_handler=dict(type='SparseBox3DKeyPointsGenerator'),\n num_temp_instances=600,\n confidence_decay=0.6,\n feat_grad=False),\n anchor_encoder=dict(\n type='SparseBox3DEncoder',\n vel_dims=3,\n embed_dims=[128, 32, 32, 64],\n mode='cat',\n output_fc=False,\n in_loops=1,\n out_loops=4),\n num_single_frame_decoder=1,\n operation_order=[\n 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm',\n 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm',\n 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm',\n 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm',\n 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm',\n 'deformable', 'ffn', 'norm', 'refine'\n ],\n temp_graph_model=dict(\n type='MultiheadAttention',\n embed_dims=512,\n num_heads=8,\n batch_first=True,\n dropout=0.1),\n graph_model=dict(\n type='MultiheadAttention',\n embed_dims=512,\n num_heads=8,\n batch_first=True,\n dropout=0.1),\n norm_layer=dict(type='LN', normalized_shape=256),\n ffn=dict(\n type='AsymmetricFFN',\n in_channels=512,\n pre_norm=dict(type='LN'),\n embed_dims=256,\n feedforward_channels=1024,\n num_fcs=2,\n ffn_drop=0.1,\n act_cfg=dict(type='ReLU', inplace=True)),\n deformable_model=dict(\n type='DeformableFeatureAggregation',\n embed_dims=256,\n num_groups=8,\n num_levels=4,\n num_cams=6,\n attn_drop=0.15,\n use_deformable_func=True,\n use_camera_embed=True,\n residual_mode='cat',\n kps_generator=dict(\n type='SparseBox3DKeyPointsGenerator',\n num_learnable_pts=6,\n fix_scale=[[0, 0, 0], [0.45, 0, 0], [-0.45, 0, 0],\n [0, 0.45, 0], [0, -0.45, 0], [0, 0, 0.45],\n [0, 0, -0.45]])),\n refine_layer=dict(\n type='SparseBox3DRefinementModule',\n embed_dims=256,\n num_cls=10,\n refine_yaw=True,\n with_quality_estimation=True),\n sampler=dict(\n type='SparseBox3DTarget',\n num_dn_groups=5,\n num_temp_dn_groups=3,\n dn_noise_scale=[2.0, 2.0, 2.0, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5],\n max_dn_gt=32,\n add_neg_dn=True,\n cls_weight=2.0,\n box_weight=0.25,\n reg_weights=[2.0, 2.0, 2.0, 0.5, 0.5, 0.5, 0.0, 0.0, 0.0, 0.0],\n cls_wise_reg_weights=dict(\n {9: [2.0, 2.0, 2.0, 1.0, 1.0, 1.0, 0.0, 0.0, 1.0, 1.0]})),\n loss_cls=dict(\n type='FocalLoss',\n use_sigmoid=True,\n gamma=2.0,\n alpha=0.25,\n loss_weight=2.0),\n loss_reg=dict(\n type='SparseBox3DLoss',\n loss_box=dict(type='L1Loss', loss_weight=0.25),\n loss_centerness=dict(type='CrossEntropyLoss', use_sigmoid=True),\n loss_yawness=dict(type='GaussianFocalLoss'),\n cls_allow_reverse=[5]),\n decoder=dict(type='SparseBox3DDecoder'),\n reg_weights=[2.0, 2.0, 2.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0]))\ndataset_type = 'NuScenes3DDetTrackDataset'\ndata_root = 'data/nuscenes/'\nanno_root = 'data/nuscenes_anno_pkls/'\nfile_client_args = dict(backend='disk')\nimg_norm_cfg = dict(\n mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375], to_rgb=True)\ntrain_pipeline = [\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(\n type='LoadPointsFromFile',\n coord_type='LIDAR',\n load_dim=5,\n use_dim=5,\n file_client_args=dict(backend='disk')),\n dict(type='ResizeCropFlipImage'),\n dict(type='MultiScaleDepthMapGenerator', downsample=[4, 8, 16]),\n dict(type='BBoxRotation'),\n dict(type='PhotoMetricDistortionMultiViewImage'),\n dict(\n type='NormalizeMultiviewImage',\n mean=[123.675, 116.28, 103.53],\n std=[58.395, 57.12, 57.375],\n to_rgb=True),\n dict(\n type='CircleObjectRangeFilter',\n class_dist_thred=[55, 55, 55, 55, 55, 55, 55, 55, 55, 55]),\n dict(\n type='InstanceNameFilter',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ]),\n dict(type='NuScenesSparse4DAdaptor'),\n dict(\n type='Collect',\n keys=[\n 'img', 'timestamp', 'projection_mat', 'image_wh', 'gt_depth',\n 'focal', 'gt_bboxes_3d', 'gt_labels_3d'\n ],\n meta_keys=['T_global', 'T_global_inv', 'timestamp', 'instance_id'])\n]\ntest_pipeline = [\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(type='ResizeCropFlipImage'),\n dict(\n type='NormalizeMultiviewImage',\n mean=[123.675, 116.28, 103.53],\n std=[58.395, 57.12, 57.375],\n to_rgb=True),\n dict(type='NuScenesSparse4DAdaptor'),\n dict(\n type='Collect',\n keys=['img', 'timestamp', 'projection_mat', 'image_wh'],\n meta_keys=['T_global', 'T_global_inv', 'timestamp'])\n]\ninput_modality = dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False)\ndata_basic_config = dict(\n type='NuScenes3DDetTrackDataset',\n data_root='data/nuscenes/',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier',\n 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ],\n modality=dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False),\n version='v1.0-trainval')\ndata_aug_conf = dict(\n resize_lim=(0.4, 0.47),\n final_dim=(256, 704),\n bot_pct_lim=(0.0, 0.0),\n rot_lim=(-5.4, 5.4),\n H=900,\n W=1600,\n rand_flip=True,\n rot3d_range=[-0.3925, 0.3925])\ndata = dict(\n samples_per_gpu=10,\n workers_per_gpu=10,\n train=dict(\n type='NuScenes3DDetTrackDataset',\n data_root='data/nuscenes/',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ],\n modality=dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False),\n version='v1.0-trainval',\n ann_file='data/nuscenes_anno_pkls/nuscenes_infos_train.pkl',\n pipeline=[\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(\n type='LoadPointsFromFile',\n coord_type='LIDAR',\n load_dim=5,\n use_dim=5,\n file_client_args=dict(backend='disk')),\n dict(type='ResizeCropFlipImage'),\n dict(type='MultiScaleDepthMapGenerator', downsample=[4, 8, 16]),\n dict(type='BBoxRotation'),\n dict(type='PhotoMetricDistortionMultiViewImage'),\n dict(\n type='NormalizeMultiviewImage',\n mean=[123.675, 116.28, 103.53],\n std=[58.395, 57.12, 57.375],\n to_rgb=True),\n dict(\n type='CircleObjectRangeFilter',\n class_dist_thred=[55, 55, 55, 55, 55, 55, 55, 55, 55, 55]),\n dict(\n type='InstanceNameFilter',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian',\n 'traffic_cone'\n ]),\n dict(type='NuScenesSparse4DAdaptor'),\n dict(\n type='Collect',\n keys=[\n 'img', 'timestamp', 'projection_mat', 'image_wh',\n 'gt_depth', 'focal', 'gt_bboxes_3d', 'gt_labels_3d'\n ],\n meta_keys=[\n 'T_global', 'T_global_inv', 'timestamp', 'instance_id'\n ])\n ],\n test_mode=False,\n data_aug_conf=dict(\n resize_lim=(0.4, 0.47),\n final_dim=(256, 704),\n bot_pct_lim=(0.0, 0.0),\n rot_lim=(-5.4, 5.4),\n H=900,\n W=1600,\n rand_flip=True,\n rot3d_range=[-0.3925, 0.3925]),\n with_seq_flag=True,\n sequences_split_num=2,\n keep_consistent_seq_aug=True),\n val=dict(\n type='NuScenes3DDetTrackDataset',\n data_root='data/nuscenes/',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ],\n modality=dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False),\n version='v1.0-trainval',\n ann_file='data/nuscenes_anno_pkls/nuscenes_infos_val.pkl',\n pipeline=[\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(type='ResizeCropFlipImage'),\n dict(\n type='NormalizeMultiviewImage',\n mean=[123.675, 116.28, 103.53],\n std=[58.395, 57.12, 57.375],\n to_rgb=True),\n dict(type='NuScenesSparse4DAdaptor'),\n dict(\n type='Collect',\n keys=['img', 'timestamp', 'projection_mat', 'image_wh'],\n meta_keys=['T_global', 'T_global_inv', 'timestamp'])\n ],\n data_aug_conf=dict(\n resize_lim=(0.4, 0.47),\n final_dim=(256, 704),\n bot_pct_lim=(0.0, 0.0),\n rot_lim=(-5.4, 5.4),\n H=900,\n W=1600,\n rand_flip=True,\n rot3d_range=[-0.3925, 0.3925]),\n test_mode=True,\n tracking=True,\n tracking_threshold=0.2),\n test=dict(\n type='NuScenes3DDetTrackDataset',\n data_root='data/nuscenes/',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ],\n modality=dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False),\n version='v1.0-trainval',\n ann_file='data/nuscenes_anno_pkls/nuscenes_infos_val.pkl',\n pipeline=[\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(type='ResizeCropFlipImage'),\n dict(\n type='NormalizeMultiviewImage',\n mean=[123.675, 116.28, 103.53],\n std=[58.395, 57.12, 57.375],\n to_rgb=True),\n dict(type='NuScenesSparse4DAdaptor'),\n dict(\n type='Collect',\n keys=['img', 'timestamp', 'projection_mat', 'image_wh'],\n meta_keys=['T_global', 'T_global_inv', 'timestamp'])\n ],\n data_aug_conf=dict(\n resize_lim=(0.4, 0.47),\n final_dim=(256, 704),\n bot_pct_lim=(0.0, 0.0),\n rot_lim=(-5.4, 5.4),\n H=900,\n W=1600,\n rand_flip=True,\n rot3d_range=[-0.3925, 0.3925]),\n test_mode=True,\n tracking=True,\n tracking_threshold=0.2))\noptimizer = dict(\n type='AdamW',\n lr=0.0006,\n weight_decay=0.001,\n paramwise_cfg=dict(custom_keys=dict(img_backbone=dict(lr_mult=0.5))))\noptimizer_config = dict(grad_clip=dict(max_norm=25, norm_type=2))\nlr_config = dict(\n policy='CosineAnnealing',\n warmup='linear',\n warmup_iters=500,\n warmup_ratio=0.3333333333333333,\n min_lr_ratio=0.001)\nrunner = dict(type='IterBasedRunner', max_iters=35100)\nvis_pipeline = [\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(type='Collect', keys=['img'], meta_keys=['timestamp', 'lidar2img'])\n]\nevaluation = dict(\n interval=7020,\n pipeline=[\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(\n type='Collect', keys=['img'], meta_keys=['timestamp', 'lidar2img'])\n ])\ngpu_ids = range(0, 1)\n", "seed": 0, "exp_name": "sparse4dv3_temporal_r50_1x8_bs6_256x704.py"} -{"mode": "train", "epoch": 1, "iter": 1, "lr": 0.0001, "memory": 24610, "data_time": 3.98057, "loss_cls_0": 2.30663, "loss_box_0": 0.0, "loss_cns_0": 0.0, "loss_yns_0": 0.0, "loss_cls_1": 2.16604, "loss_box_1": 0.11646, "loss_cns_1": 0.02369, "loss_yns_1": 0.00487, "loss_cls_2": 2.24642, "loss_box_2": 0.0, "loss_cns_2": 0.0, "loss_yns_2": 0.0, "loss_cls_3": 2.31147, "loss_box_3": 0.05239, "loss_cns_3": 0.00847, "loss_yns_3": 0.00082, "loss_cls_4": 1.9817, "loss_box_4": 0.34944, "loss_cns_4": 0.04751, "loss_yns_4": 0.03208, "loss_cls_5": 2.34902, "loss_box_5": 0.0, "loss_cns_5": 0.0, "loss_yns_5": 0.0, "loss_cls_dn_0": 1.1647, "loss_box_dn_0": 1.4583, "loss_cls_dn_1": 1.15688, "loss_box_dn_1": 1.72702, "loss_cls_dn_2": 1.13358, "loss_box_dn_2": 1.94337, "loss_cls_dn_3": 1.12714, "loss_box_dn_3": 2.21498, "loss_cls_dn_4": 1.04984, "loss_box_dn_4": 2.40351, "loss_cls_dn_5": 1.18307, "loss_box_dn_5": 2.68052, "loss_dense_depth": 1.83092, "loss": 35.07083, "grad_norm": 552.76892, "time": 18.55108} -{"mode": "train", "epoch": 1, "iter": 2, "lr": 0.0001, "memory": 24610, "data_time": 0.04637, "loss_cls_0": 2.18432, "loss_box_0": 0.0, "loss_cns_0": 0.0, "loss_yns_0": 0.0, "loss_cls_1": 1.96277, "loss_box_1": 0.56686, "loss_cns_1": 0.07997, "loss_yns_1": 0.02058, "loss_cls_2": 2.06824, "loss_box_2": 0.03134, "loss_cns_2": 0.00173, "loss_yns_2": 0.00085, "loss_cls_3": 1.98587, "loss_box_3": 0.18329, "loss_cns_3": 0.02968, "loss_yns_3": 0.01256, "loss_cls_4": 1.78967, "loss_box_4": 1.33363, "loss_cns_4": 0.1416, "loss_yns_4": 0.04559, "loss_cls_5": 1.94021, "loss_box_5": 1.28361, "loss_cns_5": 0.1587, "loss_yns_5": 0.06509, "loss_cls_dn_0": 1.10544, "loss_box_dn_0": 1.26155, "loss_cls_dn_1": 0.96388, "loss_box_dn_1": 2.58898, "loss_cls_dn_2": 0.94614, "loss_box_dn_2": 2.67589, "loss_cls_dn_3": 0.9394, "loss_box_dn_3": 2.72258, "loss_cls_dn_4": 0.85647, "loss_box_dn_4": 2.89311, "loss_cls_dn_5": 0.96672, "loss_box_dn_5": 3.11693, "loss_dense_depth": 1.78171, "loss": 38.70496, "grad_norm": 82.59489, "time": 1.60494} -{"mode": "train", "epoch": 1, "iter": 3, "lr": 0.0001, "memory": 24610, "data_time": 0.04025, "loss_cls_0": 1.65162, "loss_box_0": 1.93713, "loss_cns_0": 0.59172, "loss_yns_0": 0.50428, "loss_cls_1": 1.78719, "loss_box_1": 1.86653, "loss_cns_1": 0.3044, "loss_yns_1": 0.08113, "loss_cls_2": 1.73174, "loss_box_2": 2.56371, "loss_cns_2": 0.31855, "loss_yns_2": 0.12334, "loss_cls_3": 1.61935, "loss_box_3": 3.48156, "loss_cns_3": 0.42481, "loss_yns_3": 0.23822, "loss_cls_4": 1.56431, "loss_box_4": 4.16641, "loss_cns_4": 0.44319, "loss_yns_4": 0.12397, "loss_cls_5": 1.62375, "loss_box_5": 4.01113, "loss_cns_5": 0.35185, "loss_yns_5": 0.16616, "loss_cls_dn_0": 0.81844, "loss_box_dn_0": 1.16895, "loss_cls_dn_1": 0.82006, "loss_box_dn_1": 2.32237, "loss_cls_dn_2": 0.78162, "loss_box_dn_2": 2.49491, "loss_cls_dn_3": 0.7246, "loss_box_dn_3": 2.72033, "loss_cls_dn_4": 0.70216, "loss_box_dn_4": 2.91948, "loss_cls_dn_5": 0.77281, "loss_box_dn_5": 3.1393, "loss_dense_depth": 1.79892, "loss": 52.86003, "grad_norm": 120.70339, "time": 1.10367} -{"mode": "train", "epoch": 1, "iter": 4, "lr": 0.0001, "memory": 24610, "data_time": 0.04015, "loss_cls_0": 1.4669, "loss_box_0": 2.85343, "loss_cns_0": 0.69301, "loss_yns_0": 0.27831, "loss_cls_1": 1.67974, "loss_box_1": 2.62379, "loss_cns_1": 0.32076, "loss_yns_1": 0.10762, "loss_cls_2": 1.63142, "loss_box_2": 3.54968, "loss_cns_2": 0.4407, "loss_yns_2": 0.16092, "loss_cls_3": 1.49641, "loss_box_3": 3.68407, "loss_cns_3": 0.48868, "loss_yns_3": 0.23419, "loss_cls_4": 1.43944, "loss_box_4": 4.24996, "loss_cns_4": 0.41154, "loss_yns_4": 0.22664, "loss_cls_5": 1.54081, "loss_box_5": 4.62502, "loss_cns_5": 0.45289, "loss_yns_5": 0.19832, "loss_cls_dn_0": 0.68436, "loss_box_dn_0": 1.2427, "loss_cls_dn_1": 0.74741, "loss_box_dn_1": 2.46691, "loss_cls_dn_2": 0.70412, "loss_box_dn_2": 2.50193, "loss_cls_dn_3": 0.6555, "loss_box_dn_3": 2.71395, "loss_cls_dn_4": 0.62627, "loss_box_dn_4": 2.87817, "loss_cls_dn_5": 0.66842, "loss_box_dn_5": 3.09102, "loss_dense_depth": 1.6706, "loss": 55.50559, "grad_norm": 119.59214, "time": 1.10498} -{"mode": "train", "epoch": 1, "iter": 5, "lr": 0.0001, "memory": 24610, "data_time": 0.04052, "loss_cls_0": 1.42409, "loss_box_0": 2.85271, "loss_cns_0": 0.55408, "loss_yns_0": 0.16463, "loss_cls_1": 1.59531, "loss_box_1": 3.82238, "loss_cns_1": 0.38955, "loss_yns_1": 0.17379, "loss_cls_2": 1.57009, "loss_box_2": 4.03945, "loss_cns_2": 0.41903, "loss_yns_2": 0.16747, "loss_cls_3": 1.41072, "loss_box_3": 4.0328, "loss_cns_3": 0.45115, "loss_yns_3": 0.19508, "loss_cls_4": 1.48411, "loss_box_4": 4.04254, "loss_cns_4": 0.43249, "loss_yns_4": 0.19445, "loss_cls_5": 1.4769, "loss_box_5": 4.18145, "loss_cns_5": 0.47147, "loss_yns_5": 0.17734, "loss_cls_dn_0": 0.58254, "loss_box_dn_0": 1.22379, "loss_cls_dn_1": 0.69075, "loss_box_dn_1": 2.40863, "loss_cls_dn_2": 0.67061, "loss_box_dn_2": 2.47777, "loss_cls_dn_3": 0.58467, "loss_box_dn_3": 2.59501, "loss_cls_dn_4": 0.55723, "loss_box_dn_4": 2.7483, "loss_cls_dn_5": 0.59763, "loss_box_dn_5": 2.83826, "loss_dense_depth": 1.71681, "loss": 55.41506, "grad_norm": 127.85551, "time": 1.10823} -{"mode": "train", "epoch": 1, "iter": 6, "lr": 0.0001, "memory": 24610, "data_time": 0.03669, "loss_cls_0": 1.39112, "loss_box_0": 2.63356, "loss_cns_0": 0.58055, "loss_yns_0": 0.19106, "loss_cls_1": 1.52841, "loss_box_1": 3.84606, "loss_cns_1": 0.45899, "loss_yns_1": 0.17944, "loss_cls_2": 1.54351, "loss_box_2": 3.82138, "loss_cns_2": 0.46939, "loss_yns_2": 0.16751, "loss_cls_3": 1.47349, "loss_box_3": 3.63384, "loss_cns_3": 0.49699, "loss_yns_3": 0.21049, "loss_cls_4": 1.42324, "loss_box_4": 3.72741, "loss_cns_4": 0.48311, "loss_yns_4": 0.1999, "loss_cls_5": 1.36674, "loss_box_5": 3.8768, "loss_cns_5": 0.49344, "loss_yns_5": 0.1956, "loss_cls_dn_0": 0.53187, "loss_box_dn_0": 1.15254, "loss_cls_dn_1": 0.64877, "loss_box_dn_1": 2.62108, "loss_cls_dn_2": 0.60331, "loss_box_dn_2": 2.54291, "loss_cls_dn_3": 0.5313, "loss_box_dn_3": 2.59549, "loss_cls_dn_4": 0.52847, "loss_box_dn_4": 2.70635, "loss_cls_dn_5": 0.54592, "loss_box_dn_5": 2.82125, "loss_dense_depth": 1.64064, "loss": 53.86193, "grad_norm": 96.45193, "time": 1.10982} -{"mode": "train", "epoch": 1, "iter": 7, "lr": 0.0001, "memory": 24610, "data_time": 0.04098, "loss_cls_0": 1.31288, "loss_box_0": 2.38463, "loss_cns_0": 0.68216, "loss_yns_0": 0.23162, "loss_cls_1": 1.46562, "loss_box_1": 3.69785, "loss_cns_1": 0.49073, "loss_yns_1": 0.18794, "loss_cls_2": 1.58583, "loss_box_2": 3.64796, "loss_cns_2": 0.48434, "loss_yns_2": 0.15689, "loss_cls_3": 1.44536, "loss_box_3": 3.61358, "loss_cns_3": 0.46807, "loss_yns_3": 0.2082, "loss_cls_4": 1.44465, "loss_box_4": 3.68789, "loss_cns_4": 0.51119, "loss_yns_4": 0.18475, "loss_cls_5": 1.35051, "loss_box_5": 3.84059, "loss_cns_5": 0.54063, "loss_yns_5": 0.21624, "loss_cls_dn_0": 0.5396, "loss_box_dn_0": 1.09708, "loss_cls_dn_1": 0.59218, "loss_box_dn_1": 2.78354, "loss_cls_dn_2": 0.55129, "loss_box_dn_2": 2.70108, "loss_cls_dn_3": 0.48825, "loss_box_dn_3": 2.69163, "loss_cls_dn_4": 0.49633, "loss_box_dn_4": 2.77469, "loss_cls_dn_5": 0.49654, "loss_box_dn_5": 2.92684, "loss_dense_depth": 1.57594, "loss": 53.55511, "grad_norm": 87.80739, "time": 1.11272} -{"mode": "train", "epoch": 1, "iter": 8, "lr": 0.0001, "memory": 24610, "data_time": 0.04087, "loss_cls_0": 1.28242, "loss_box_0": 2.29705, "loss_cns_0": 0.66831, "loss_yns_0": 0.19046, "loss_cls_1": 1.39981, "loss_box_1": 3.34903, "loss_cns_1": 0.48067, "loss_yns_1": 0.18009, "loss_cls_2": 1.47193, "loss_box_2": 3.51757, "loss_cns_2": 0.47575, "loss_yns_2": 0.17673, "loss_cls_3": 1.36272, "loss_box_3": 3.49688, "loss_cns_3": 0.45838, "loss_yns_3": 0.18273, "loss_cls_4": 1.43378, "loss_box_4": 3.49343, "loss_cns_4": 0.49155, "loss_yns_4": 0.16998, "loss_cls_5": 1.39964, "loss_box_5": 3.58064, "loss_cns_5": 0.54665, "loss_yns_5": 0.1794, "loss_cls_dn_0": 0.53009, "loss_box_dn_0": 1.04898, "loss_cls_dn_1": 0.58786, "loss_box_dn_1": 1.68834, "loss_cls_dn_2": 0.55995, "loss_box_dn_2": 1.73685, "loss_cls_dn_3": 0.50199, "loss_box_dn_3": 1.78072, "loss_cls_dn_4": 0.50865, "loss_box_dn_4": 1.88332, "loss_cls_dn_5": 0.49142, "loss_box_dn_5": 1.971, "loss_dense_depth": 1.53081, "loss": 47.10556, "grad_norm": 90.82548, "time": 1.10643} -{"mode": "train", "epoch": 1, "iter": 9, "lr": 0.0001, "memory": 24610, "data_time": 0.03575, "loss_cls_0": 1.25892, "loss_box_0": 2.22689, "loss_cns_0": 0.67653, "loss_yns_0": 0.16845, "loss_cls_1": 1.43574, "loss_box_1": 3.50766, "loss_cns_1": 0.46999, "loss_yns_1": 0.17061, "loss_cls_2": 1.37694, "loss_box_2": 3.63239, "loss_cns_2": 0.47732, "loss_yns_2": 0.17415, "loss_cls_3": 1.31495, "loss_box_3": 3.67668, "loss_cns_3": 0.49298, "loss_yns_3": 0.19409, "loss_cls_4": 1.37384, "loss_box_4": 3.65143, "loss_cns_4": 0.51317, "loss_yns_4": 0.18314, "loss_cls_5": 1.35035, "loss_box_5": 3.70795, "loss_cns_5": 0.49643, "loss_yns_5": 0.16105, "loss_cls_dn_0": 0.52999, "loss_box_dn_0": 1.0492, "loss_cls_dn_1": 0.52929, "loss_box_dn_1": 1.56682, "loss_cls_dn_2": 0.55852, "loss_box_dn_2": 1.66302, "loss_cls_dn_3": 0.49461, "loss_box_dn_3": 1.79943, "loss_cls_dn_4": 0.51461, "loss_box_dn_4": 1.8399, "loss_cls_dn_5": 0.48009, "loss_box_dn_5": 1.91193, "loss_dense_depth": 1.44251, "loss": 47.07157, "grad_norm": 94.83189, "time": 1.10594} -{"mode": "train", "epoch": 1, "iter": 10, "lr": 0.0001, "memory": 24610, "data_time": 0.03526, "loss_cls_0": 1.27274, "loss_box_0": 2.28519, "loss_cns_0": 0.62052, "loss_yns_0": 0.15785, "loss_cls_1": 1.35207, "loss_box_1": 3.27768, "loss_cns_1": 0.46726, "loss_yns_1": 0.18254, "loss_cls_2": 1.33858, "loss_box_2": 3.26481, "loss_cns_2": 0.48423, "loss_yns_2": 0.15581, "loss_cls_3": 1.2767, "loss_box_3": 3.44572, "loss_cns_3": 0.52089, "loss_yns_3": 0.16929, "loss_cls_4": 1.28606, "loss_box_4": 3.41479, "loss_cns_4": 0.49386, "loss_yns_4": 0.19092, "loss_cls_5": 1.29827, "loss_box_5": 3.49275, "loss_cns_5": 0.46492, "loss_yns_5": 0.17009, "loss_cls_dn_0": 0.51819, "loss_box_dn_0": 1.07039, "loss_cls_dn_1": 0.48081, "loss_box_dn_1": 1.77812, "loss_cls_dn_2": 0.53678, "loss_box_dn_2": 1.80652, "loss_cls_dn_3": 0.48014, "loss_box_dn_3": 1.90555, "loss_cls_dn_4": 0.4812, "loss_box_dn_4": 1.93074, "loss_cls_dn_5": 0.4545, "loss_box_dn_5": 1.98345, "loss_dense_depth": 1.49331, "loss": 46.00323, "grad_norm": 79.26781, "time": 1.10639} -{"mode": "train", "epoch": 1, "iter": 11, "lr": 0.0001, "memory": 24610, "data_time": 0.0975, "loss_cls_0": 1.22653, "loss_box_0": 2.29476, "loss_cns_0": 0.61396, "loss_yns_0": 0.15944, "loss_cls_1": 1.22818, "loss_box_1": 3.17683, "loss_cns_1": 0.53454, "loss_yns_1": 0.16258, "loss_cls_2": 1.37189, "loss_box_2": 3.0956, "loss_cns_2": 0.52941, "loss_yns_2": 0.1693, "loss_cls_3": 1.27103, "loss_box_3": 3.21188, "loss_cns_3": 0.54519, "loss_yns_3": 0.18043, "loss_cls_4": 1.26942, "loss_box_4": 3.20106, "loss_cns_4": 0.5214, "loss_yns_4": 0.1759, "loss_cls_5": 1.29003, "loss_box_5": 3.22291, "loss_cns_5": 0.50545, "loss_yns_5": 0.18756, "loss_cls_dn_0": 0.46368, "loss_box_dn_0": 1.09315, "loss_cls_dn_1": 0.43905, "loss_box_dn_1": 1.94808, "loss_cls_dn_2": 0.49789, "loss_box_dn_2": 1.9292, "loss_cls_dn_3": 0.45842, "loss_box_dn_3": 1.99189, "loss_cls_dn_4": 0.44545, "loss_box_dn_4": 1.98816, "loss_cls_dn_5": 0.43474, "loss_box_dn_5": 2.02793, "loss_dense_depth": 1.41983, "loss": 45.28276, "grad_norm": 61.15717, "time": 1.18817} -{"mode": "train", "epoch": 1, "iter": 12, "lr": 0.0001, "memory": 24610, "data_time": 0.0407, "loss_cls_0": 1.2203, "loss_box_0": 2.27827, "loss_cns_0": 0.60599, "loss_yns_0": 0.16864, "loss_cls_1": 1.24125, "loss_box_1": 3.1833, "loss_cns_1": 0.49186, "loss_yns_1": 0.17369, "loss_cls_2": 1.32107, "loss_box_2": 3.14362, "loss_cns_2": 0.50465, "loss_yns_2": 0.20451, "loss_cls_3": 1.30347, "loss_box_3": 3.16121, "loss_cns_3": 0.50603, "loss_yns_3": 0.16989, "loss_cls_4": 1.27424, "loss_box_4": 3.19871, "loss_cns_4": 0.49766, "loss_yns_4": 0.19213, "loss_cls_5": 1.28354, "loss_box_5": 3.24107, "loss_cns_5": 0.50929, "loss_yns_5": 0.1748, "loss_cls_dn_0": 0.44487, "loss_box_dn_0": 1.04, "loss_cls_dn_1": 0.43118, "loss_box_dn_1": 2.27567, "loss_cls_dn_2": 0.47159, "loss_box_dn_2": 2.25502, "loss_cls_dn_3": 0.44825, "loss_box_dn_3": 2.25761, "loss_cls_dn_4": 0.43463, "loss_box_dn_4": 2.26885, "loss_cls_dn_5": 0.42854, "loss_box_dn_5": 2.30253, "loss_dense_depth": 1.39957, "loss": 46.50754, "grad_norm": 65.2148, "time": 1.12616} -{"mode": "train", "epoch": 1, "iter": 13, "lr": 0.0001, "memory": 24610, "data_time": 0.04256, "loss_cls_0": 1.20531, "loss_box_0": 2.29932, "loss_cns_0": 0.62059, "loss_yns_0": 0.1592, "loss_cls_1": 1.28855, "loss_box_1": 3.15073, "loss_cns_1": 0.52085, "loss_yns_1": 0.15746, "loss_cls_2": 1.25767, "loss_box_2": 3.22917, "loss_cns_2": 0.51131, "loss_yns_2": 0.17732, "loss_cls_3": 1.26067, "loss_box_3": 3.23408, "loss_cns_3": 0.49497, "loss_yns_3": 0.16339, "loss_cls_4": 1.2623, "loss_box_4": 3.35373, "loss_cns_4": 0.51406, "loss_yns_4": 0.17162, "loss_cls_5": 1.27971, "loss_box_5": 3.48625, "loss_cns_5": 0.50438, "loss_yns_5": 0.18286, "loss_cls_dn_0": 0.43233, "loss_box_dn_0": 1.0276, "loss_cls_dn_1": 0.43784, "loss_box_dn_1": 1.76155, "loss_cls_dn_2": 0.46302, "loss_box_dn_2": 1.8028, "loss_cls_dn_3": 0.44828, "loss_box_dn_3": 1.85957, "loss_cls_dn_4": 0.43824, "loss_box_dn_4": 1.97794, "loss_cls_dn_5": 0.44095, "loss_box_dn_5": 2.0556, "loss_dense_depth": 1.32515, "loss": 44.95638, "grad_norm": 83.33117, "time": 1.11553} -{"mode": "train", "epoch": 1, "iter": 14, "lr": 0.00011, "memory": 24610, "data_time": 0.03554, "loss_cls_0": 1.19108, "loss_box_0": 2.33004, "loss_cns_0": 0.62319, "loss_yns_0": 0.1765, "loss_cls_1": 1.26354, "loss_box_1": 2.70077, "loss_cns_1": 0.56187, "loss_yns_1": 0.17797, "loss_cls_2": 1.24081, "loss_box_2": 2.81552, "loss_cns_2": 0.55662, "loss_yns_2": 0.17631, "loss_cls_3": 1.22418, "loss_box_3": 2.8422, "loss_cns_3": 0.54325, "loss_yns_3": 0.17564, "loss_cls_4": 1.21396, "loss_box_4": 2.87199, "loss_cns_4": 0.54443, "loss_yns_4": 0.17369, "loss_cls_5": 1.24488, "loss_box_5": 2.92961, "loss_cns_5": 0.55672, "loss_yns_5": 0.17255, "loss_cls_dn_0": 0.45217, "loss_box_dn_0": 1.01887, "loss_cls_dn_1": 0.46855, "loss_box_dn_1": 1.54606, "loss_cls_dn_2": 0.46364, "loss_box_dn_2": 1.5785, "loss_cls_dn_3": 0.44891, "loss_box_dn_3": 1.60562, "loss_cls_dn_4": 0.43636, "loss_box_dn_4": 1.6546, "loss_cls_dn_5": 0.45707, "loss_box_dn_5": 1.72566, "loss_dense_depth": 1.45001, "loss": 41.61334, "grad_norm": 76.3792, "time": 1.11234} -{"mode": "train", "epoch": 1, "iter": 15, "lr": 0.00011, "memory": 24610, "data_time": 0.03958, "loss_cls_0": 1.0828, "loss_box_0": 2.39904, "loss_cns_0": 0.60597, "loss_yns_0": 0.20074, "loss_cls_1": 1.13595, "loss_box_1": 2.53133, "loss_cns_1": 0.56094, "loss_yns_1": 0.17922, "loss_cls_2": 1.28591, "loss_box_2": 2.50287, "loss_cns_2": 0.55917, "loss_yns_2": 0.2055, "loss_cls_3": 1.24592, "loss_box_3": 2.59383, "loss_cns_3": 0.55215, "loss_yns_3": 0.19434, "loss_cls_4": 1.14672, "loss_box_4": 2.64749, "loss_cns_4": 0.54596, "loss_yns_4": 0.17448, "loss_cls_5": 1.20278, "loss_box_5": 2.69861, "loss_cns_5": 0.55584, "loss_yns_5": 0.21402, "loss_cls_dn_0": 0.46094, "loss_box_dn_0": 1.03755, "loss_cls_dn_1": 0.47738, "loss_box_dn_1": 1.42973, "loss_cls_dn_2": 0.52607, "loss_box_dn_2": 1.38384, "loss_cls_dn_3": 0.48981, "loss_box_dn_3": 1.44583, "loss_cls_dn_4": 0.47352, "loss_box_dn_4": 1.46324, "loss_cls_dn_5": 0.51881, "loss_box_dn_5": 1.52233, "loss_dense_depth": 1.32725, "loss": 39.57785, "grad_norm": 84.23974, "time": 1.12705} -{"mode": "train", "epoch": 1, "iter": 16, "lr": 0.00011, "memory": 24610, "data_time": 0.03507, "loss_cls_0": 1.10337, "loss_box_0": 2.30977, "loss_cns_0": 0.59182, "loss_yns_0": 0.20833, "loss_cls_1": 1.21081, "loss_box_1": 2.73237, "loss_cns_1": 0.53748, "loss_yns_1": 0.18371, "loss_cls_2": 1.41882, "loss_box_2": 2.68901, "loss_cns_2": 0.52485, "loss_yns_2": 0.19402, "loss_cls_3": 1.36465, "loss_box_3": 2.7971, "loss_cns_3": 0.52317, "loss_yns_3": 0.19195, "loss_cls_4": 1.17566, "loss_box_4": 2.83997, "loss_cns_4": 0.51443, "loss_yns_4": 0.18379, "loss_cls_5": 1.24485, "loss_box_5": 2.93471, "loss_cns_5": 0.52318, "loss_yns_5": 0.22054, "loss_cls_dn_0": 0.47063, "loss_box_dn_0": 1.03307, "loss_cls_dn_1": 0.48206, "loss_box_dn_1": 1.43345, "loss_cls_dn_2": 0.53247, "loss_box_dn_2": 1.35025, "loss_cls_dn_3": 0.49004, "loss_box_dn_3": 1.44808, "loss_cls_dn_4": 0.45059, "loss_box_dn_4": 1.50329, "loss_cls_dn_5": 0.51652, "loss_box_dn_5": 1.59507, "loss_dense_depth": 1.27082, "loss": 40.7947, "grad_norm": 77.71381, "time": 1.11779} -{"mode": "train", "epoch": 1, "iter": 17, "lr": 0.00011, "memory": 24610, "data_time": 0.03632, "loss_cls_0": 1.10428, "loss_box_0": 2.3164, "loss_cns_0": 0.57826, "loss_yns_0": 0.19337, "loss_cls_1": 1.18522, "loss_box_1": 2.8828, "loss_cns_1": 0.52093, "loss_yns_1": 0.18528, "loss_cls_2": 1.28379, "loss_box_2": 2.85971, "loss_cns_2": 0.50753, "loss_yns_2": 0.18855, "loss_cls_3": 1.24685, "loss_box_3": 2.89979, "loss_cns_3": 0.49917, "loss_yns_3": 0.18534, "loss_cls_4": 1.17202, "loss_box_4": 2.99772, "loss_cns_4": 0.49447, "loss_yns_4": 0.19362, "loss_cls_5": 1.21008, "loss_box_5": 3.06205, "loss_cns_5": 0.48166, "loss_yns_5": 0.18981, "loss_cls_dn_0": 0.45496, "loss_box_dn_0": 1.02412, "loss_cls_dn_1": 0.45821, "loss_box_dn_1": 1.61577, "loss_cls_dn_2": 0.50182, "loss_box_dn_2": 1.563, "loss_cls_dn_3": 0.46446, "loss_box_dn_3": 1.63663, "loss_cls_dn_4": 0.42437, "loss_box_dn_4": 1.73429, "loss_cls_dn_5": 0.48258, "loss_box_dn_5": 1.82744, "loss_dense_depth": 1.27066, "loss": 41.897, "grad_norm": 67.27931, "time": 1.1227} -{"mode": "train", "epoch": 1, "iter": 18, "lr": 0.00011, "memory": 24610, "data_time": 0.03665, "loss_cls_0": 1.08763, "loss_box_0": 2.28317, "loss_cns_0": 0.56341, "loss_yns_0": 0.18355, "loss_cls_1": 1.17543, "loss_box_1": 2.72715, "loss_cns_1": 0.51069, "loss_yns_1": 0.17807, "loss_cls_2": 1.19638, "loss_box_2": 2.79071, "loss_cns_2": 0.48483, "loss_yns_2": 0.18203, "loss_cls_3": 1.17676, "loss_box_3": 2.81104, "loss_cns_3": 0.48385, "loss_yns_3": 0.19974, "loss_cls_4": 1.23068, "loss_box_4": 2.88617, "loss_cns_4": 0.47846, "loss_yns_4": 0.18506, "loss_cls_5": 1.1828, "loss_box_5": 2.91687, "loss_cns_5": 0.47418, "loss_yns_5": 0.18256, "loss_cls_dn_0": 0.43213, "loss_box_dn_0": 1.02608, "loss_cls_dn_1": 0.44208, "loss_box_dn_1": 1.6792, "loss_cls_dn_2": 0.48219, "loss_box_dn_2": 1.67047, "loss_cls_dn_3": 0.45125, "loss_box_dn_3": 1.71508, "loss_cls_dn_4": 0.4054, "loss_box_dn_4": 1.78737, "loss_cls_dn_5": 0.45101, "loss_box_dn_5": 1.85675, "loss_dense_depth": 1.2686, "loss": 41.2388, "grad_norm": 64.49593, "time": 1.12176} -{"mode": "train", "epoch": 1, "iter": 19, "lr": 0.00011, "memory": 24610, "data_time": 0.04444, "loss_cls_0": 1.07787, "loss_box_0": 2.26391, "loss_cns_0": 0.56208, "loss_yns_0": 0.18121, "loss_cls_1": 1.16, "loss_box_1": 2.64779, "loss_cns_1": 0.52052, "loss_yns_1": 0.17913, "loss_cls_2": 1.2247, "loss_box_2": 2.67409, "loss_cns_2": 0.52311, "loss_yns_2": 0.18195, "loss_cls_3": 1.23301, "loss_box_3": 2.64169, "loss_cns_3": 0.53296, "loss_yns_3": 0.19317, "loss_cls_4": 1.21583, "loss_box_4": 2.67834, "loss_cns_4": 0.51748, "loss_yns_4": 0.17881, "loss_cls_5": 1.29587, "loss_box_5": 2.72237, "loss_cns_5": 0.528, "loss_yns_5": 0.19266, "loss_cls_dn_0": 0.41848, "loss_box_dn_0": 1.0249, "loss_cls_dn_1": 0.4357, "loss_box_dn_1": 1.27034, "loss_cls_dn_2": 0.46802, "loss_box_dn_2": 1.37784, "loss_cls_dn_3": 0.4501, "loss_box_dn_3": 1.40138, "loss_cls_dn_4": 0.40497, "loss_box_dn_4": 1.46702, "loss_cls_dn_5": 0.43326, "loss_box_dn_5": 1.53593, "loss_dense_depth": 1.254, "loss": 39.06849, "grad_norm": 64.53758, "time": 1.1403} -{"mode": "train", "epoch": 1, "iter": 20, "lr": 0.00011, "memory": 24610, "data_time": 0.03967, "loss_cls_0": 1.07102, "loss_box_0": 2.21063, "loss_cns_0": 0.57045, "loss_yns_0": 0.17918, "loss_cls_1": 1.12127, "loss_box_1": 2.82119, "loss_cns_1": 0.54321, "loss_yns_1": 0.18436, "loss_cls_2": 1.22599, "loss_box_2": 2.85766, "loss_cns_2": 0.54587, "loss_yns_2": 0.19463, "loss_cls_3": 1.17655, "loss_box_3": 2.87226, "loss_cns_3": 0.54726, "loss_yns_3": 0.17715, "loss_cls_4": 1.12704, "loss_box_4": 2.94356, "loss_cns_4": 0.53498, "loss_yns_4": 0.18344, "loss_cls_5": 1.25657, "loss_box_5": 3.07056, "loss_cns_5": 0.51343, "loss_yns_5": 0.20129, "loss_cls_dn_0": 0.40748, "loss_box_dn_0": 1.03998, "loss_cls_dn_1": 0.39986, "loss_box_dn_1": 1.47612, "loss_cls_dn_2": 0.42093, "loss_box_dn_2": 1.53048, "loss_cls_dn_3": 0.41746, "loss_box_dn_3": 1.52101, "loss_cls_dn_4": 0.3817, "loss_box_dn_4": 1.53939, "loss_cls_dn_5": 0.39578, "loss_box_dn_5": 1.61666, "loss_dense_depth": 1.18036, "loss": 40.45675, "grad_norm": 80.6266, "time": 1.11603} -{"mode": "train", "epoch": 1, "iter": 21, "lr": 0.00011, "memory": 24610, "data_time": 0.08796, "loss_cls_0": 1.08592, "loss_box_0": 2.24947, "loss_cns_0": 0.58024, "loss_yns_0": 0.17617, "loss_cls_1": 1.10468, "loss_box_1": 2.85794, "loss_cns_1": 0.51244, "loss_yns_1": 0.18291, "loss_cls_2": 1.18248, "loss_box_2": 2.86065, "loss_cns_2": 0.51648, "loss_yns_2": 0.19639, "loss_cls_3": 1.15917, "loss_box_3": 2.87986, "loss_cns_3": 0.54178, "loss_yns_3": 0.17573, "loss_cls_4": 1.11524, "loss_box_4": 2.91768, "loss_cns_4": 0.55015, "loss_yns_4": 0.18508, "loss_cls_5": 1.16708, "loss_box_5": 3.02838, "loss_cns_5": 0.53261, "loss_yns_5": 0.18502, "loss_cls_dn_0": 0.40609, "loss_box_dn_0": 1.04277, "loss_cls_dn_1": 0.40517, "loss_box_dn_1": 1.27454, "loss_cls_dn_2": 0.41257, "loss_box_dn_2": 1.33064, "loss_cls_dn_3": 0.42439, "loss_box_dn_3": 1.36208, "loss_cls_dn_4": 0.39474, "loss_box_dn_4": 1.40194, "loss_cls_dn_5": 0.40151, "loss_box_dn_5": 1.46198, "loss_dense_depth": 1.18445, "loss": 39.44642, "grad_norm": 75.14684, "time": 1.21738} -{"mode": "train", "epoch": 1, "iter": 22, "lr": 0.00011, "memory": 24610, "data_time": 0.04665, "loss_cls_0": 1.032, "loss_box_0": 2.23037, "loss_cns_0": 0.57604, "loss_yns_0": 0.17138, "loss_cls_1": 1.05766, "loss_box_1": 3.08225, "loss_cns_1": 0.48763, "loss_yns_1": 0.18415, "loss_cls_2": 1.10866, "loss_box_2": 3.10112, "loss_cns_2": 0.49052, "loss_yns_2": 0.20018, "loss_cls_3": 1.11534, "loss_box_3": 3.07003, "loss_cns_3": 0.50088, "loss_yns_3": 0.19284, "loss_cls_4": 1.10544, "loss_box_4": 3.08521, "loss_cns_4": 0.49925, "loss_yns_4": 0.1716, "loss_cls_5": 1.10993, "loss_box_5": 3.10767, "loss_cns_5": 0.50314, "loss_yns_5": 0.18116, "loss_cls_dn_0": 0.41727, "loss_box_dn_0": 1.03615, "loss_cls_dn_1": 0.41833, "loss_box_dn_1": 1.22496, "loss_cls_dn_2": 0.41878, "loss_box_dn_2": 1.27621, "loss_cls_dn_3": 0.43825, "loss_box_dn_3": 1.33808, "loss_cls_dn_4": 0.41083, "loss_box_dn_4": 1.46028, "loss_cls_dn_5": 0.42263, "loss_box_dn_5": 1.47299, "loss_dense_depth": 1.19693, "loss": 39.89611, "grad_norm": 70.00922, "time": 4.69862} diff --git a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20260210_162905.log b/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20260210_162905.log deleted file mode 100644 index c96e58f9788a0a16b227f69836095ac6cd6715a2..0000000000000000000000000000000000000000 --- a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20260210_162905.log +++ /dev/null @@ -1,3242 +0,0 @@ -2026-02-10 16:29:05,786 - mmdet - INFO - Environment info: ------------------------------------------------------------- -sys.platform: linux -Python: 3.10.12 (main, Aug 15 2025, 14:32:43) [GCC 11.4.0] -CUDA available: True -GPU 0,1,2,3,4,5,6,7: BW200_LC, UBB BW1000_LC -CUDA_HOME: /opt/dtk -NVCC: Not Available -GCC: x86_64-linux-gnu-gcc (Ubuntu 11.4.0-1ubuntu1~22.04) 11.4.0 -PyTorch: 2.5.1 -PyTorch compiling details: PyTorch built with: - - GCC 10.3 - - C++ Version: 201703 - - Intel(R) Math Kernel Library Version 2020.0.4 Product Build 20200917 for Intel(R) 64 architecture applications - - OpenMP 201511 (a.k.a. OpenMP 4.5) - - LAPACK is enabled (usually provided by MKL) - - NNPACK is enabled - - CPU capability usage: AVX2 - - HIP Runtime 6.3.25521 - - MIOpen 2.18.0 - - Magma 2.8.0 - - Build settings: BLAS_INFO=mkl, BUILD_TYPE=Release, CXX_COMPILER=/opt/rh/gcc-toolset-10/root/usr/bin/c++, CXX_FLAGS= -D_GLIBCXX_USE_CXX11_ABI=1 -fvisibility-inlines-hidden -DUSE_PTHREADPOOL -DNDEBUG -DUSE_KINETO -DLIBKINETO_NOCUPTI -DLIBKINETO_NOXPUPTI=ON -DUSE_FBGEMM -DUSE_PYTORCH_QNNPACK -DUSE_XNNPACK -DSYMBOLICATE_MOBILE_DEBUG_HANDLE -O2 -fPIC -Wall -Wextra -Werror=return-type -Werror=non-virtual-dtor -Werror=bool-operation -Wnarrowing -Wno-missing-field-initializers -Wno-type-limits -Wno-array-bounds -Wno-unknown-pragmas -Wno-unused-parameter -Wno-strict-overflow -Wno-strict-aliasing -Wno-stringop-overflow -Wsuggest-override -Wno-psabi -Wno-error=old-style-cast -Wno-missing-braces -fdiagnostics-color=always -faligned-new -Wno-unused-but-set-variable -Wno-maybe-uninitialized -fno-math-errno -fno-trapping-math -Werror=format -Wno-stringop-overflow, FORCE_FALLBACK_CUDA_MPI=1, LAPACK_INFO=mkl, PERF_WITH_AVX=1, PERF_WITH_AVX2=1, TORCH_VERSION=2.5.1, USE_CUDA=0, USE_CUDNN=OFF, USE_CUSPARSELT=OFF, USE_EXCEPTION_PTR=1, USE_GFLAGS=1, USE_GLOG=1, USE_GLOO=1, USE_MKL=ON, USE_MKLDNN=0, USE_MPI=1, USE_NCCL=1, USE_NNPACK=ON, USE_OPENMP=1, USE_ROCM=ON, USE_ROCM_KERNEL_ASSERT=OFF, - -TorchVision: 0.20.1 -OpenCV: 4.12.0 -MMCV: 1.6.1 -MMCV Compiler: GCC 10.3 -MMCV CUDA Compiler: rocm not available -MMDetection: 2.25.1+ ------------------------------------------------------------- - -2026-02-10 16:29:06,799 - mmdet - INFO - Distributed training: False -2026-02-10 16:29:07,774 - mmdet - INFO - Config: -plugin = True -plugin_dir = 'projects/mmdet3d_plugin/' -dist_params = dict(backend='nccl') -log_level = 'INFO' -work_dir = './work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704' -total_batch_size = 120 -num_gpus = 8 -batch_size = 15 -num_iters_per_epoch = 234 -num_epochs = 100 -checkpoint_epoch_interval = 20 -checkpoint_config = dict(interval=4680) -log_config = dict( - interval=1, - hooks=[ - dict(type='TextLoggerHook', by_epoch=False), - dict(type='TensorboardLoggerHook') - ]) -load_from = None -resume_from = None -workflow = [('train', 1)] -fp16 = dict(loss_scale=32.0) -input_shape = (704, 256) -tracking_test = True -tracking_threshold = 0.2 -class_names = [ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier', - 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' -] -num_classes = 10 -embed_dims = 256 -num_groups = 8 -num_decoder = 6 -num_single_frame_decoder = 1 -use_deformable_func = True -strides = [4, 8, 16, 32] -num_levels = 4 -num_depth_layers = 3 -drop_out = 0.1 -temporal = True -decouple_attn = True -with_quality_estimation = True -model = dict( - type='Sparse4D', - use_grid_mask=True, - use_deformable_func=True, - img_backbone=dict( - type='ResNet', - depth=50, - num_stages=4, - frozen_stages=-1, - norm_eval=False, - style='pytorch', - with_cp=True, - out_indices=(0, 1, 2, 3), - norm_cfg=dict(type='BN', requires_grad=True), - pretrained='ckpt/resnet50-19c8e357.pth'), - img_neck=dict( - type='FPN', - num_outs=4, - start_level=0, - out_channels=256, - add_extra_convs='on_output', - relu_before_extra_convs=True, - in_channels=[256, 512, 1024, 2048]), - depth_branch=dict( - type='DenseDepthNet', - embed_dims=256, - num_depth_layers=3, - loss_weight=0.2), - head=dict( - type='Sparse4DHead', - cls_threshold_to_reg=0.05, - decouple_attn=True, - instance_bank=dict( - type='InstanceBank', - num_anchor=900, - embed_dims=256, - anchor='nuscenes_kmeans900.npy', - anchor_handler=dict(type='SparseBox3DKeyPointsGenerator'), - num_temp_instances=600, - confidence_decay=0.6, - feat_grad=False), - anchor_encoder=dict( - type='SparseBox3DEncoder', - vel_dims=3, - embed_dims=[128, 32, 32, 64], - mode='cat', - output_fc=False, - in_loops=1, - out_loops=4), - num_single_frame_decoder=1, - operation_order=[ - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine' - ], - temp_graph_model=dict( - type='MultiheadAttention', - embed_dims=512, - num_heads=8, - batch_first=True, - dropout=0.1), - graph_model=dict( - type='MultiheadAttention', - embed_dims=512, - num_heads=8, - batch_first=True, - dropout=0.1), - norm_layer=dict(type='LN', normalized_shape=256), - ffn=dict( - type='AsymmetricFFN', - in_channels=512, - pre_norm=dict(type='LN'), - embed_dims=256, - feedforward_channels=1024, - num_fcs=2, - ffn_drop=0.1, - act_cfg=dict(type='ReLU', inplace=True)), - deformable_model=dict( - type='DeformableFeatureAggregation', - embed_dims=256, - num_groups=8, - num_levels=4, - num_cams=6, - attn_drop=0.15, - use_deformable_func=True, - use_camera_embed=True, - residual_mode='cat', - kps_generator=dict( - type='SparseBox3DKeyPointsGenerator', - num_learnable_pts=6, - fix_scale=[[0, 0, 0], [0.45, 0, 0], [-0.45, 0, 0], - [0, 0.45, 0], [0, -0.45, 0], [0, 0, 0.45], - [0, 0, -0.45]])), - refine_layer=dict( - type='SparseBox3DRefinementModule', - embed_dims=256, - num_cls=10, - refine_yaw=True, - with_quality_estimation=True), - sampler=dict( - type='SparseBox3DTarget', - num_dn_groups=5, - num_temp_dn_groups=3, - dn_noise_scale=[2.0, 2.0, 2.0, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5], - max_dn_gt=32, - add_neg_dn=True, - cls_weight=2.0, - box_weight=0.25, - reg_weights=[2.0, 2.0, 2.0, 0.5, 0.5, 0.5, 0.0, 0.0, 0.0, 0.0], - cls_wise_reg_weights=dict( - {9: [2.0, 2.0, 2.0, 1.0, 1.0, 1.0, 0.0, 0.0, 1.0, 1.0]})), - loss_cls=dict( - type='FocalLoss', - use_sigmoid=True, - gamma=2.0, - alpha=0.25, - loss_weight=2.0), - loss_reg=dict( - type='SparseBox3DLoss', - loss_box=dict(type='L1Loss', loss_weight=0.25), - loss_centerness=dict(type='CrossEntropyLoss', use_sigmoid=True), - loss_yawness=dict(type='GaussianFocalLoss'), - cls_allow_reverse=[5]), - decoder=dict(type='SparseBox3DDecoder'), - reg_weights=[2.0, 2.0, 2.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0])) -dataset_type = 'NuScenes3DDetTrackDataset' -data_root = 'data/nuscenes/' -anno_root = 'data/nuscenes_anno_pkls/' -file_client_args = dict(backend='disk') -img_norm_cfg = dict( - mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375], to_rgb=True) -train_pipeline = [ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict( - type='LoadPointsFromFile', - coord_type='LIDAR', - load_dim=5, - use_dim=5, - file_client_args=dict(backend='disk')), - dict(type='ResizeCropFlipImage'), - dict(type='MultiScaleDepthMapGenerator', downsample=[4, 8, 16]), - dict(type='BBoxRotation'), - dict(type='PhotoMetricDistortionMultiViewImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict( - type='CircleObjectRangeFilter', - class_dist_thred=[55, 55, 55, 55, 55, 55, 55, 55, 55, 55]), - dict( - type='InstanceNameFilter', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ]), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=[ - 'img', 'timestamp', 'projection_mat', 'image_wh', 'gt_depth', - 'focal', 'gt_bboxes_3d', 'gt_labels_3d' - ], - meta_keys=['T_global', 'T_global_inv', 'timestamp', 'instance_id']) -] -test_pipeline = [ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='ResizeCropFlipImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=['img', 'timestamp', 'projection_mat', 'image_wh'], - meta_keys=['T_global', 'T_global_inv', 'timestamp']) -] -input_modality = dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False) -data_basic_config = dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier', - 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval') -data_aug_conf = dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]) -data = dict( - samples_per_gpu=15, - workers_per_gpu=15, - train=dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval', - ann_file='data/nuscenes_anno_pkls/nuscenes_infos_train.pkl', - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict( - type='LoadPointsFromFile', - coord_type='LIDAR', - load_dim=5, - use_dim=5, - file_client_args=dict(backend='disk')), - dict(type='ResizeCropFlipImage'), - dict(type='MultiScaleDepthMapGenerator', downsample=[4, 8, 16]), - dict(type='BBoxRotation'), - dict(type='PhotoMetricDistortionMultiViewImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict( - type='CircleObjectRangeFilter', - class_dist_thred=[55, 55, 55, 55, 55, 55, 55, 55, 55, 55]), - dict( - type='InstanceNameFilter', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', - 'traffic_cone' - ]), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=[ - 'img', 'timestamp', 'projection_mat', 'image_wh', - 'gt_depth', 'focal', 'gt_bboxes_3d', 'gt_labels_3d' - ], - meta_keys=[ - 'T_global', 'T_global_inv', 'timestamp', 'instance_id' - ]) - ], - test_mode=False, - data_aug_conf=dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]), - with_seq_flag=True, - sequences_split_num=2, - keep_consistent_seq_aug=True), - val=dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval', - ann_file='data/nuscenes_anno_pkls/nuscenes_infos_val.pkl', - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='ResizeCropFlipImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=['img', 'timestamp', 'projection_mat', 'image_wh'], - meta_keys=['T_global', 'T_global_inv', 'timestamp']) - ], - data_aug_conf=dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]), - test_mode=True, - tracking=True, - tracking_threshold=0.2), - test=dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval', - ann_file='data/nuscenes_anno_pkls/nuscenes_infos_val.pkl', - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='ResizeCropFlipImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=['img', 'timestamp', 'projection_mat', 'image_wh'], - meta_keys=['T_global', 'T_global_inv', 'timestamp']) - ], - data_aug_conf=dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]), - test_mode=True, - tracking=True, - tracking_threshold=0.2)) -optimizer = dict( - type='AdamW', - lr=0.0006, - weight_decay=0.001, - paramwise_cfg=dict(custom_keys=dict(img_backbone=dict(lr_mult=0.5)))) -optimizer_config = dict(grad_clip=dict(max_norm=25, norm_type=2)) -lr_config = dict( - policy='CosineAnnealing', - warmup='linear', - warmup_iters=500, - warmup_ratio=0.3333333333333333, - min_lr_ratio=0.001) -runner = dict(type='IterBasedRunner', max_iters=23400) -vis_pipeline = [ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='Collect', keys=['img'], meta_keys=['timestamp', 'lidar2img']) -] -evaluation = dict( - interval=4680, - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict( - type='Collect', keys=['img'], meta_keys=['timestamp', 'lidar2img']) - ]) -gpu_ids = range(0, 1) - -2026-02-10 16:29:07,774 - mmdet - INFO - Set random seed to 0, deterministic: False -2026-02-10 16:29:08,323 - mmdet - INFO - initialize ResNet with init_cfg {'type': 'Pretrained', 'checkpoint': 'ckpt/resnet50-19c8e357.pth'} -2026-02-10 16:29:08,537 - mmdet - INFO - initialize FPN with init_cfg {'type': 'Xavier', 'layer': 'Conv2d', 'distribution': 'uniform'} -Name of parameter - Initialization information - -img_backbone.conv1.weight - torch.Size([64, 3, 7, 7]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.bn1.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.bn1.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.conv1.weight - torch.Size([64, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn1.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn1.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.conv2.weight - torch.Size([64, 64, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn2.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn2.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.conv3.weight - torch.Size([256, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn3.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn3.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.downsample.0.weight - torch.Size([256, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.downsample.1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.downsample.1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.conv1.weight - torch.Size([64, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn1.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn1.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.conv2.weight - torch.Size([64, 64, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn2.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn2.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.conv3.weight - torch.Size([256, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn3.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn3.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.conv1.weight - torch.Size([64, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn1.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn1.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.conv2.weight - torch.Size([64, 64, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn2.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn2.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.conv3.weight - torch.Size([256, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn3.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn3.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.conv1.weight - torch.Size([128, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn1.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn1.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.conv2.weight - torch.Size([128, 128, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn2.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn2.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.conv3.weight - torch.Size([512, 128, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn3.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn3.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.downsample.0.weight - torch.Size([512, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.downsample.1.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.downsample.1.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.conv1.weight - torch.Size([128, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn1.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn1.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.conv2.weight - torch.Size([128, 128, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn2.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn2.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.conv3.weight - torch.Size([512, 128, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn3.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn3.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.conv1.weight - torch.Size([128, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn1.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn1.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.conv2.weight - torch.Size([128, 128, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn2.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn2.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.conv3.weight - torch.Size([512, 128, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn3.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn3.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.conv1.weight - torch.Size([128, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn1.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn1.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.conv2.weight - torch.Size([128, 128, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn2.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn2.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.conv3.weight - torch.Size([512, 128, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn3.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn3.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.conv1.weight - torch.Size([256, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.downsample.0.weight - torch.Size([1024, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.downsample.1.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.downsample.1.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.conv1.weight - torch.Size([512, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn1.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn1.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.conv2.weight - torch.Size([512, 512, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn2.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn2.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.conv3.weight - torch.Size([2048, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn3.weight - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn3.bias - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.downsample.0.weight - torch.Size([2048, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.downsample.1.weight - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.downsample.1.bias - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.conv1.weight - torch.Size([512, 2048, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn1.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn1.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.conv2.weight - torch.Size([512, 512, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn2.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn2.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.conv3.weight - torch.Size([2048, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn3.weight - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn3.bias - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.conv1.weight - torch.Size([512, 2048, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn1.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn1.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.conv2.weight - torch.Size([512, 512, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn2.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn2.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.conv3.weight - torch.Size([2048, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn3.weight - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn3.bias - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_neck.lateral_convs.0.conv.weight - torch.Size([256, 256, 1, 1]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.lateral_convs.0.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.lateral_convs.1.conv.weight - torch.Size([256, 512, 1, 1]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.lateral_convs.1.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.lateral_convs.2.conv.weight - torch.Size([256, 1024, 1, 1]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.lateral_convs.2.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.lateral_convs.3.conv.weight - torch.Size([256, 2048, 1, 1]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.lateral_convs.3.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.fpn_convs.0.conv.weight - torch.Size([256, 256, 3, 3]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.fpn_convs.0.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.fpn_convs.1.conv.weight - torch.Size([256, 256, 3, 3]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.fpn_convs.1.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.fpn_convs.2.conv.weight - torch.Size([256, 256, 3, 3]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.fpn_convs.2.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.fpn_convs.3.conv.weight - torch.Size([256, 256, 3, 3]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.fpn_convs.3.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.instance_bank.anchor - torch.Size([900, 11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.instance_bank.instance_feature - torch.Size([900, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.instance_bank.anchor_handler.fix_scale - torch.Size([1, 3]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.0.weight - torch.Size([128, 3]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.0.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.2.weight - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.2.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.3.weight - torch.Size([128, 128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.3.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.5.weight - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.5.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.6.weight - torch.Size([128, 128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.6.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.8.weight - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.8.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.9.weight - torch.Size([128, 128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.9.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.11.weight - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.11.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.0.weight - torch.Size([32, 3]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.0.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.2.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.2.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.3.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.3.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.5.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.5.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.6.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.6.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.8.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.8.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.9.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.9.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.11.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.11.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.0.weight - torch.Size([32, 2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.0.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.2.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.2.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.3.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.3.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.5.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.5.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.6.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.6.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.8.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.8.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.9.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.9.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.11.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.11.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.0.weight - torch.Size([64, 3]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.0.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.2.weight - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.2.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.3.weight - torch.Size([64, 64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.3.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.5.weight - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.5.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.6.weight - torch.Size([64, 64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.6.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.8.weight - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.8.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.9.weight - torch.Size([64, 64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.9.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.11.weight - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.11.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.1.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.1.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.1.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.1.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.1.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.1.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.1.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.1.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.3.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.4.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.4.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.4.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.4.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.5.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.5.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.5.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.5.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.6.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.6.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.8.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.8.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.8.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.8.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.8.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.8.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.8.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.8.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.10.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.11.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.11.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.11.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.11.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.12.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.12.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.12.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.12.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.13.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.13.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.15.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.15.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.15.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.15.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.15.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.15.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.15.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.15.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.16.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.16.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.17.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.18.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.18.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.18.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.18.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.19.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.19.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.19.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.19.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.20.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.20.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.22.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.22.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.22.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.22.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.22.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.22.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.22.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.22.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.23.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.23.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.24.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.25.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.25.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.25.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.25.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.26.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.26.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.26.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.26.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.27.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.27.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.29.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.29.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.29.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.29.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.29.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.29.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.29.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.29.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.30.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.30.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.31.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.32.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.32.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.32.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.32.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.33.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.33.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.33.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.33.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.34.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.34.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.36.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.36.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.36.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.36.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.36.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.36.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.36.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.36.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.37.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.37.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.38.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.fc_before.weight - torch.Size([512, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.fc_after.weight - torch.Size([256, 512]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.0.weight - torch.Size([1, 256, 1, 1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.0.bias - torch.Size([1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.1.weight - torch.Size([1, 256, 1, 1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.1.bias - torch.Size([1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.2.weight - torch.Size([1, 256, 1, 1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.2.bias - torch.Size([1]): -The value is the same before and after calling `init_weights` of Sparse4D -2026-02-10 16:29:08,905 - mmdet - INFO - Model: -Sparse4D( - (img_backbone): ResNet( - (conv1): Conv2d(3, 64, kernel_size=(7, 7), stride=(2, 2), padding=(3, 3), bias=False) - (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (maxpool): MaxPool2d(kernel_size=3, stride=2, padding=1, dilation=1, ceil_mode=False) - (layer1): ResLayer( - (0): Bottleneck( - (conv1): Conv2d(64, 64, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (downsample): Sequential( - (0): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - ) - ) - (1): Bottleneck( - (conv1): Conv2d(256, 64, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (2): Bottleneck( - (conv1): Conv2d(256, 64, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - ) - (layer2): ResLayer( - (0): Bottleneck( - (conv1): Conv2d(256, 128, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (downsample): Sequential( - (0): Conv2d(256, 512, kernel_size=(1, 1), stride=(2, 2), bias=False) - (1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - ) - ) - (1): Bottleneck( - (conv1): Conv2d(512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (2): Bottleneck( - (conv1): Conv2d(512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (3): Bottleneck( - (conv1): Conv2d(512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - ) - (layer3): ResLayer( - (0): Bottleneck( - (conv1): Conv2d(512, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (downsample): Sequential( - (0): Conv2d(512, 1024, kernel_size=(1, 1), stride=(2, 2), bias=False) - (1): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - ) - ) - (1): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (2): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (3): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (4): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (5): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - ) - (layer4): ResLayer( - (0): Bottleneck( - (conv1): Conv2d(1024, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(512, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (downsample): Sequential( - (0): Conv2d(1024, 2048, kernel_size=(1, 1), stride=(2, 2), bias=False) - (1): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - ) - ) - (1): Bottleneck( - (conv1): Conv2d(2048, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(512, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (2): Bottleneck( - (conv1): Conv2d(2048, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(512, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - ) - ) - init_cfg={'type': 'Pretrained', 'checkpoint': 'ckpt/resnet50-19c8e357.pth'} - (img_neck): FPN( - (lateral_convs): ModuleList( - (0): ConvModule( - (conv): Conv2d(256, 256, kernel_size=(1, 1), stride=(1, 1)) - ) - (1): ConvModule( - (conv): Conv2d(512, 256, kernel_size=(1, 1), stride=(1, 1)) - ) - (2): ConvModule( - (conv): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1)) - ) - (3): ConvModule( - (conv): Conv2d(2048, 256, kernel_size=(1, 1), stride=(1, 1)) - ) - ) - (fpn_convs): ModuleList( - (0-3): 4 x ConvModule( - (conv): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1)) - ) - ) - ) - init_cfg={'type': 'Xavier', 'layer': 'Conv2d', 'distribution': 'uniform'} - (head): Sparse4DHead( - (instance_bank): InstanceBank( - (anchor_handler): SparseBox3DKeyPointsGenerator() - ) - (anchor_encoder): SparseBox3DEncoder( - (pos_fc): Sequential( - (0): Linear(in_features=3, out_features=128, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((128,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=128, out_features=128, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((128,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=128, out_features=128, bias=True) - (7): ReLU(inplace=True) - (8): LayerNorm((128,), eps=1e-05, elementwise_affine=True) - (9): Linear(in_features=128, out_features=128, bias=True) - (10): ReLU(inplace=True) - (11): LayerNorm((128,), eps=1e-05, elementwise_affine=True) - ) - (size_fc): Sequential( - (0): Linear(in_features=3, out_features=32, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=32, out_features=32, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=32, out_features=32, bias=True) - (7): ReLU(inplace=True) - (8): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (9): Linear(in_features=32, out_features=32, bias=True) - (10): ReLU(inplace=True) - (11): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - ) - (yaw_fc): Sequential( - (0): Linear(in_features=2, out_features=32, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=32, out_features=32, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=32, out_features=32, bias=True) - (7): ReLU(inplace=True) - (8): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (9): Linear(in_features=32, out_features=32, bias=True) - (10): ReLU(inplace=True) - (11): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - ) - (vel_fc): Sequential( - (0): Linear(in_features=3, out_features=64, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((64,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=64, out_features=64, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((64,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=64, out_features=64, bias=True) - (7): ReLU(inplace=True) - (8): LayerNorm((64,), eps=1e-05, elementwise_affine=True) - (9): Linear(in_features=64, out_features=64, bias=True) - (10): ReLU(inplace=True) - (11): LayerNorm((64,), eps=1e-05, elementwise_affine=True) - ) - ) - (loss_cls): FocalLoss() - (loss_reg): SparseBox3DLoss( - (loss_box): L1Loss() - (loss_cns): CrossEntropyLoss(avg_non_ignore=False) - (loss_yns): GaussianFocalLoss() - ) - (layers): ModuleList( - (0): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (1): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (4-5): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (6): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (7): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (8): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (11-12): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (13): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (14): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (15): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (16): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (17): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (18-19): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (20): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (21): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (22): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (23): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (24): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (25-26): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (27): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (28): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (29): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (30): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (31): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (32-33): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (34): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (35): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (36): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (37): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (38): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - ) - (fc_before): Linear(in_features=256, out_features=512, bias=False) - (fc_after): Linear(in_features=512, out_features=256, bias=False) - ) - (depth_branch): DenseDepthNet( - (depth_layers): ModuleList( - (0-2): 3 x Conv2d(256, 1, kernel_size=(1, 1), stride=(1, 1)) - ) - ) - (grid_mask): GridMask() -) -2026-02-10 16:29:22,891 - mmdet - INFO - Start running, host: root@bw28, work_dir: /workspace/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704 -2026-02-10 16:29:22,892 - mmdet - INFO - Hooks will be executed in the following order: -before_run: -(VERY_HIGH ) CosineAnnealingLrUpdaterHook -(ABOVE_NORMAL) Fp16OptimizerHook -(NORMAL ) CheckpointHook -(NORMAL ) EvalHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -before_train_epoch: -(VERY_HIGH ) CosineAnnealingLrUpdaterHook -(NORMAL ) EvalHook -(LOW ) IterTimerHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -before_train_iter: -(VERY_HIGH ) CosineAnnealingLrUpdaterHook -(NORMAL ) EvalHook -(LOW ) IterTimerHook - -------------------- -after_train_iter: -(ABOVE_NORMAL) Fp16OptimizerHook -(NORMAL ) CheckpointHook -(NORMAL ) EvalHook -(NORMAL ) ProfilerHook -(LOW ) IterTimerHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -after_train_epoch: -(NORMAL ) CheckpointHook -(NORMAL ) EvalHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -before_val_epoch: -(LOW ) IterTimerHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -before_val_iter: -(LOW ) IterTimerHook - -------------------- -after_val_iter: -(LOW ) IterTimerHook - -------------------- -after_val_epoch: -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -after_run: -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -2026-02-10 16:29:22,892 - mmdet - INFO - workflow: [('train', 1)], max: 23400 iters -2026-02-10 16:29:22,894 - mmdet - INFO - Checkpoints will be saved to /workspace/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704 by HardDiskBackend. -2026-02-10 16:29:48,152 - mmdet - INFO - Iter [1/23400] lr: 1.000e-04, eta: 6 days, 15:39:52, time: 24.565, data_time: 5.767, memory: 36808, loss_cls_0: 2.3084, loss_box_0: 0.0000, loss_cns_0: 0.0000, loss_yns_0: 0.0000, loss_cls_1: 2.1821, loss_box_1: 0.0723, loss_cns_1: 0.0158, loss_yns_1: 0.0054, loss_cls_2: 2.3956, loss_box_2: 0.0000, loss_cns_2: 0.0000, loss_yns_2: 0.0000, loss_cls_3: 2.4343, loss_box_3: 0.0271, loss_cns_3: 0.0051, loss_yns_3: 0.0018, loss_cls_4: 1.9844, loss_box_4: 0.4677, loss_cns_4: 0.0537, loss_yns_4: 0.0268, loss_cls_5: 2.3104, loss_box_5: 0.0000, loss_cns_5: 0.0000, loss_yns_5: 0.0000, loss_cls_dn_0: 1.1695, loss_box_dn_0: 1.4651, loss_cls_dn_1: 1.1522, loss_box_dn_1: 1.7464, loss_cls_dn_2: 1.1890, loss_box_dn_2: 1.9870, loss_cls_dn_3: 1.1522, loss_box_dn_3: 2.2653, loss_cls_dn_4: 1.0405, loss_box_dn_4: 2.4319, loss_cls_dn_5: 1.1751, loss_box_dn_5: 2.6881, loss_dense_depth: 1.8449, loss: 35.5981, grad_norm: 428.5826 -2026-02-10 16:29:50,352 - mmdet - INFO - Iter [2/23400] lr: 1.004e-04, eta: 3 days, 14:58:35, time: 2.200, data_time: 0.052, memory: 36808, loss_cls_0: 2.1470, loss_box_0: 0.0388, loss_cns_0: 0.0125, loss_yns_0: 0.0014, loss_cls_1: 2.0484, loss_box_1: 0.2609, loss_cns_1: 0.0520, loss_yns_1: 0.0183, loss_cls_2: 2.1636, loss_box_2: 0.1117, loss_cns_2: 0.0107, loss_yns_2: 0.0037, loss_cls_3: 2.0164, loss_box_3: 0.1596, loss_cns_3: 0.0204, loss_yns_3: 0.0063, loss_cls_4: 1.8526, loss_box_4: 0.7409, loss_cns_4: 0.0903, loss_yns_4: 0.0282, loss_cls_5: 1.9376, loss_box_5: 1.2267, loss_cns_5: 0.1453, loss_yns_5: 0.0619, loss_cls_dn_0: 1.0958, loss_box_dn_0: 1.2221, loss_cls_dn_1: 1.0006, loss_box_dn_1: 2.5245, loss_cls_dn_2: 1.0118, loss_box_dn_2: 2.5888, loss_cls_dn_3: 0.9480, loss_box_dn_3: 2.6751, loss_cls_dn_4: 0.8793, loss_box_dn_4: 2.9051, loss_cls_dn_5: 0.9603, loss_box_dn_5: 3.1259, loss_dense_depth: 1.7774, loss: 37.8701, grad_norm: 74.8507 -2026-02-10 16:29:51,855 - mmdet - INFO - Iter [3/23400] lr: 1.008e-04, eta: 2 days, 13:14:16, time: 1.503, data_time: 0.045, memory: 36808, loss_cls_0: 1.6083, loss_box_0: 2.6876, loss_cns_0: 0.6574, loss_yns_0: 0.3505, loss_cls_1: 1.8442, loss_box_1: 0.9271, loss_cns_1: 0.1628, loss_yns_1: 0.0424, loss_cls_2: 1.8614, loss_box_2: 1.6759, loss_cns_2: 0.1748, loss_yns_2: 0.0835, loss_cls_3: 1.6606, loss_box_3: 2.6097, loss_cns_3: 0.2607, loss_yns_3: 0.1617, loss_cls_4: 1.5519, loss_box_4: 4.2399, loss_cns_4: 0.4228, loss_yns_4: 0.1638, loss_cls_5: 1.6686, loss_box_5: 2.9198, loss_cns_5: 0.2274, loss_yns_5: 0.1035, loss_cls_dn_0: 0.8190, loss_box_dn_0: 1.2459, loss_cls_dn_1: 0.8447, loss_box_dn_1: 2.4317, loss_cls_dn_2: 0.8349, loss_box_dn_2: 2.5661, loss_cls_dn_3: 0.7479, loss_box_dn_3: 2.7701, loss_cls_dn_4: 0.7163, loss_box_dn_4: 2.9617, loss_cls_dn_5: 0.7775, loss_box_dn_5: 3.1463, loss_dense_depth: 1.7575, loss: 49.6862, grad_norm: 100.8311 -2026-02-10 16:29:53,362 - mmdet - INFO - Iter [4/23400] lr: 1.012e-04, eta: 2 days, 0:22:34, time: 1.508, data_time: 0.044, memory: 36808, loss_cls_0: 1.3984, loss_box_0: 2.7386, loss_cns_0: 0.6217, loss_yns_0: 0.2357, loss_cls_1: 1.6464, loss_box_1: 2.3824, loss_cns_1: 0.3478, loss_yns_1: 0.1235, loss_cls_2: 1.7054, loss_box_2: 3.1103, loss_cns_2: 0.3773, loss_yns_2: 0.1455, loss_cls_3: 1.4657, loss_box_3: 4.3098, loss_cns_3: 0.4589, loss_yns_3: 0.2636, loss_cls_4: 1.4014, loss_box_4: 4.7024, loss_cns_4: 0.3821, loss_yns_4: 0.1729, loss_cls_5: 1.5221, loss_box_5: 4.8082, loss_cns_5: 0.4500, loss_yns_5: 0.1959, loss_cls_dn_0: 0.6983, loss_box_dn_0: 1.1479, loss_cls_dn_1: 0.7646, loss_box_dn_1: 2.5196, loss_cls_dn_2: 0.7560, loss_box_dn_2: 2.5956, loss_cls_dn_3: 0.6581, loss_box_dn_3: 2.7837, loss_cls_dn_4: 0.6245, loss_box_dn_4: 2.9134, loss_cls_dn_5: 0.6745, loss_box_dn_5: 3.0462, loss_dense_depth: 1.6257, loss: 55.7740, grad_norm: 127.1702 -2026-02-10 16:29:54,879 - mmdet - INFO - Iter [5/23400] lr: 1.016e-04, eta: 1 day, 16:40:11, time: 1.516, data_time: 0.049, memory: 36808, loss_cls_0: 1.3004, loss_box_0: 2.7016, loss_cns_0: 0.5562, loss_yns_0: 0.1835, loss_cls_1: 1.4874, loss_box_1: 3.3336, loss_cns_1: 0.3740, loss_yns_1: 0.1593, loss_cls_2: 1.5411, loss_box_2: 3.7620, loss_cns_2: 0.3637, loss_yns_2: 0.1867, loss_cls_3: 1.3870, loss_box_3: 4.1031, loss_cns_3: 0.4255, loss_yns_3: 0.2264, loss_cls_4: 1.3980, loss_box_4: 4.2318, loss_cns_4: 0.3929, loss_yns_4: 0.1842, loss_cls_5: 1.4482, loss_box_5: 4.5088, loss_cns_5: 0.4390, loss_yns_5: 0.1871, loss_cls_dn_0: 0.6183, loss_box_dn_0: 1.1658, loss_cls_dn_1: 0.6891, loss_box_dn_1: 2.3754, loss_cls_dn_2: 0.6935, loss_box_dn_2: 2.5426, loss_cls_dn_3: 0.5784, loss_box_dn_3: 2.6172, loss_cls_dn_4: 0.5515, loss_box_dn_4: 2.7378, loss_cls_dn_5: 0.5954, loss_box_dn_5: 2.8802, loss_dense_depth: 1.6349, loss: 54.5619, grad_norm: 116.1075 -2026-02-10 16:29:56,396 - mmdet - INFO - Iter [6/23400] lr: 1.020e-04, eta: 1 day, 11:32:02, time: 1.518, data_time: 0.046, memory: 36808, loss_cls_0: 1.2990, loss_box_0: 2.6393, loss_cns_0: 0.5968, loss_yns_0: 0.2197, loss_cls_1: 1.3902, loss_box_1: 3.4420, loss_cns_1: 0.3853, loss_yns_1: 0.1826, loss_cls_2: 1.4174, loss_box_2: 3.6814, loss_cns_2: 0.3992, loss_yns_2: 0.1884, loss_cls_3: 1.3294, loss_box_3: 3.6081, loss_cns_3: 0.4352, loss_yns_3: 0.2193, loss_cls_4: 1.3407, loss_box_4: 3.8303, loss_cns_4: 0.3976, loss_yns_4: 0.2072, loss_cls_5: 1.3896, loss_box_5: 4.2016, loss_cns_5: 0.3672, loss_yns_5: 0.1908, loss_cls_dn_0: 0.5501, loss_box_dn_0: 1.1572, loss_cls_dn_1: 0.6397, loss_box_dn_1: 2.7929, loss_cls_dn_2: 0.6391, loss_box_dn_2: 2.8185, loss_cls_dn_3: 0.5430, loss_box_dn_3: 2.8286, loss_cls_dn_4: 0.5083, loss_box_dn_4: 2.9636, loss_cls_dn_5: 0.5298, loss_box_dn_5: 3.1027, loss_dense_depth: 1.6617, loss: 54.0938, grad_norm: 133.9767 -2026-02-10 16:29:57,898 - mmdet - INFO - Iter [7/23400] lr: 1.024e-04, eta: 1 day, 7:51:02, time: 1.502, data_time: 0.042, memory: 36808, loss_cls_0: 1.4108, loss_box_0: 2.4541, loss_cns_0: 0.6773, loss_yns_0: 0.2056, loss_cls_1: 1.2889, loss_box_1: 3.7671, loss_cns_1: 0.4065, loss_yns_1: 0.1798, loss_cls_2: 1.3664, loss_box_2: 3.8412, loss_cns_2: 0.3971, loss_yns_2: 0.1756, loss_cls_3: 1.3261, loss_box_3: 3.7651, loss_cns_3: 0.4019, loss_yns_3: 0.2029, loss_cls_4: 1.3275, loss_box_4: 3.9706, loss_cns_4: 0.4169, loss_yns_4: 0.1932, loss_cls_5: 1.3130, loss_box_5: 4.0409, loss_cns_5: 0.4369, loss_yns_5: 0.1824, loss_cls_dn_0: 0.5069, loss_box_dn_0: 1.0952, loss_cls_dn_1: 0.5705, loss_box_dn_1: 3.0692, loss_cls_dn_2: 0.5769, loss_box_dn_2: 3.0175, loss_cls_dn_3: 0.4980, loss_box_dn_3: 2.9908, loss_cls_dn_4: 0.4684, loss_box_dn_4: 3.0733, loss_cls_dn_5: 0.4677, loss_box_dn_5: 3.1429, loss_dense_depth: 1.5134, loss: 54.7385, grad_norm: 95.2195 -2026-02-10 16:29:59,417 - mmdet - INFO - Iter [8/23400] lr: 1.028e-04, eta: 1 day, 5:06:04, time: 1.518, data_time: 0.047, memory: 36808, loss_cls_0: 1.2607, loss_box_0: 2.3687, loss_cns_0: 0.6602, loss_yns_0: 0.1686, loss_cls_1: 1.2905, loss_box_1: 3.6318, loss_cns_1: 0.4501, loss_yns_1: 0.1727, loss_cls_2: 1.3852, loss_box_2: 3.6448, loss_cns_2: 0.3863, loss_yns_2: 0.1810, loss_cls_3: 1.3301, loss_box_3: 3.7482, loss_cns_3: 0.3687, loss_yns_3: 0.1894, loss_cls_4: 1.2985, loss_box_4: 3.8233, loss_cns_4: 0.3890, loss_yns_4: 0.1761, loss_cls_5: 1.3248, loss_box_5: 3.8647, loss_cns_5: 0.4540, loss_yns_5: 0.1832, loss_cls_dn_0: 0.5126, loss_box_dn_0: 1.0715, loss_cls_dn_1: 0.5545, loss_box_dn_1: 1.9139, loss_cls_dn_2: 0.5661, loss_box_dn_2: 1.8732, loss_cls_dn_3: 0.4976, loss_box_dn_3: 1.9534, loss_cls_dn_4: 0.4870, loss_box_dn_4: 2.0394, loss_cls_dn_5: 0.4900, loss_box_dn_5: 2.1666, loss_dense_depth: 1.5623, loss: 48.4386, grad_norm: 82.1008 -2026-02-10 16:30:00,950 - mmdet - INFO - Iter [9/23400] lr: 1.032e-04, eta: 1 day, 2:58:24, time: 1.533, data_time: 0.051, memory: 36808, loss_cls_0: 1.2017, loss_box_0: 2.4188, loss_cns_0: 0.6062, loss_yns_0: 0.1672, loss_cls_1: 1.3346, loss_box_1: 3.2931, loss_cns_1: 0.5292, loss_yns_1: 0.1782, loss_cls_2: 1.4053, loss_box_2: 3.2669, loss_cns_2: 0.4862, loss_yns_2: 0.1793, loss_cls_3: 1.3163, loss_box_3: 3.4551, loss_cns_3: 0.4773, loss_yns_3: 0.1739, loss_cls_4: 1.2656, loss_box_4: 3.4091, loss_cns_4: 0.4815, loss_yns_4: 0.1806, loss_cls_5: 1.3306, loss_box_5: 3.4383, loss_cns_5: 0.5458, loss_yns_5: 0.1799, loss_cls_dn_0: 0.5213, loss_box_dn_0: 1.0458, loss_cls_dn_1: 0.4948, loss_box_dn_1: 1.5746, loss_cls_dn_2: 0.5193, loss_box_dn_2: 1.5881, loss_cls_dn_3: 0.4723, loss_box_dn_3: 1.7387, loss_cls_dn_4: 0.4832, loss_box_dn_4: 1.7548, loss_cls_dn_5: 0.4782, loss_box_dn_5: 1.8507, loss_dense_depth: 1.4189, loss: 45.2616, grad_norm: 71.8648 -2026-02-10 16:30:02,498 - mmdet - INFO - Iter [10/23400] lr: 1.036e-04, eta: 1 day, 1:16:50, time: 1.548, data_time: 0.046, memory: 36808, loss_cls_0: 1.1964, loss_box_0: 2.4670, loss_cns_0: 0.5694, loss_yns_0: 0.1655, loss_cls_1: 1.3155, loss_box_1: 3.3003, loss_cns_1: 0.4488, loss_yns_1: 0.1760, loss_cls_2: 1.3099, loss_box_2: 3.1998, loss_cns_2: 0.4716, loss_yns_2: 0.1742, loss_cls_3: 1.2503, loss_box_3: 3.4266, loss_cns_3: 0.5010, loss_yns_3: 0.1816, loss_cls_4: 1.2535, loss_box_4: 3.3637, loss_cns_4: 0.4958, loss_yns_4: 0.1679, loss_cls_5: 1.2925, loss_box_5: 3.4883, loss_cns_5: 0.4651, loss_yns_5: 0.1752, loss_cls_dn_0: 0.5190, loss_box_dn_0: 1.0752, loss_cls_dn_1: 0.4464, loss_box_dn_1: 1.7312, loss_cls_dn_2: 0.4882, loss_box_dn_2: 1.7761, loss_cls_dn_3: 0.4569, loss_box_dn_3: 1.9156, loss_cls_dn_4: 0.4641, loss_box_dn_4: 1.9157, loss_cls_dn_5: 0.4554, loss_box_dn_5: 2.0384, loss_dense_depth: 1.3976, loss: 45.5356, grad_norm: 87.4993 -2026-02-10 16:30:04,057 - mmdet - INFO - Iter [11/23400] lr: 1.040e-04, eta: 23:54:09, time: 1.559, data_time: 0.058, memory: 36808, loss_cls_0: 1.2027, loss_box_0: 2.4147, loss_cns_0: 0.5844, loss_yns_0: 0.1684, loss_cls_1: 1.2503, loss_box_1: 3.4091, loss_cns_1: 0.4129, loss_yns_1: 0.1643, loss_cls_2: 1.2897, loss_box_2: 3.3828, loss_cns_2: 0.4375, loss_yns_2: 0.1745, loss_cls_3: 1.2394, loss_box_3: 3.5024, loss_cns_3: 0.4495, loss_yns_3: 0.1780, loss_cls_4: 1.2333, loss_box_4: 3.4243, loss_cns_4: 0.4345, loss_yns_4: 0.1743, loss_cls_5: 1.3084, loss_box_5: 3.4568, loss_cns_5: 0.4294, loss_yns_5: 0.1912, loss_cls_dn_0: 0.5093, loss_box_dn_0: 1.0536, loss_cls_dn_1: 0.4210, loss_box_dn_1: 1.9490, loss_cls_dn_2: 0.4814, loss_box_dn_2: 2.0095, loss_cls_dn_3: 0.4581, loss_box_dn_3: 2.1086, loss_cls_dn_4: 0.4424, loss_box_dn_4: 2.0731, loss_cls_dn_5: 0.4410, loss_box_dn_5: 2.1520, loss_dense_depth: 1.4064, loss: 46.4183, grad_norm: 83.9635 -2026-02-10 16:30:05,609 - mmdet - INFO - Iter [12/23400] lr: 1.044e-04, eta: 22:44:59, time: 1.552, data_time: 0.052, memory: 36808, loss_cls_0: 1.1602, loss_box_0: 2.4033, loss_cns_0: 0.5922, loss_yns_0: 0.1672, loss_cls_1: 1.2448, loss_box_1: 3.0581, loss_cns_1: 0.5139, loss_yns_1: 0.1709, loss_cls_2: 1.2897, loss_box_2: 3.0714, loss_cns_2: 0.5047, loss_yns_2: 0.1883, loss_cls_3: 1.2206, loss_box_3: 3.0898, loss_cns_3: 0.5043, loss_yns_3: 0.1668, loss_cls_4: 1.2188, loss_box_4: 3.1071, loss_cns_4: 0.5058, loss_yns_4: 0.1880, loss_cls_5: 1.2946, loss_box_5: 3.2667, loss_cns_5: 0.5015, loss_yns_5: 0.1946, loss_cls_dn_0: 0.4682, loss_box_dn_0: 1.0730, loss_cls_dn_1: 0.4032, loss_box_dn_1: 2.1145, loss_cls_dn_2: 0.4656, loss_box_dn_2: 2.1527, loss_cls_dn_3: 0.4417, loss_box_dn_3: 2.1829, loss_cls_dn_4: 0.4216, loss_box_dn_4: 2.1519, loss_cls_dn_5: 0.4242, loss_box_dn_5: 2.1899, loss_dense_depth: 1.2968, loss: 45.4093, grad_norm: 79.3916 -2026-02-10 16:30:07,157 - mmdet - INFO - Iter [13/23400] lr: 1.048e-04, eta: 21:46:20, time: 1.548, data_time: 0.050, memory: 36808, loss_cls_0: 1.1722, loss_box_0: 2.4181, loss_cns_0: 0.5912, loss_yns_0: 0.1757, loss_cls_1: 1.2268, loss_box_1: 3.0468, loss_cns_1: 0.5141, loss_yns_1: 0.1725, loss_cls_2: 1.2775, loss_box_2: 3.0364, loss_cns_2: 0.5032, loss_yns_2: 0.1790, loss_cls_3: 1.2198, loss_box_3: 3.1171, loss_cns_3: 0.4901, loss_yns_3: 0.1739, loss_cls_4: 1.2118, loss_box_4: 3.1737, loss_cns_4: 0.5000, loss_yns_4: 0.1855, loss_cls_5: 1.2532, loss_box_5: 3.2958, loss_cns_5: 0.4870, loss_yns_5: 0.2128, loss_cls_dn_0: 0.4644, loss_box_dn_0: 1.0588, loss_cls_dn_1: 0.4460, loss_box_dn_1: 1.5930, loss_cls_dn_2: 0.5111, loss_box_dn_2: 1.6448, loss_cls_dn_3: 0.4671, loss_box_dn_3: 1.6546, loss_cls_dn_4: 0.4585, loss_box_dn_4: 1.7323, loss_cls_dn_5: 0.4700, loss_box_dn_5: 1.8026, loss_dense_depth: 1.3367, loss: 43.2743, grad_norm: 87.1659 -2026-02-10 16:30:08,705 - mmdet - INFO - Iter [14/23400] lr: 1.052e-04, eta: 20:56:06, time: 1.549, data_time: 0.052, memory: 36808, loss_cls_0: 1.1654, loss_box_0: 2.2685, loss_cns_0: 0.6066, loss_yns_0: 0.1816, loss_cls_1: 1.1888, loss_box_1: 2.5956, loss_cns_1: 0.5904, loss_yns_1: 0.1729, loss_cls_2: 1.2125, loss_box_2: 2.6351, loss_cns_2: 0.5762, loss_yns_2: 0.1929, loss_cls_3: 1.1877, loss_box_3: 2.6463, loss_cns_3: 0.5799, loss_yns_3: 0.1914, loss_cls_4: 1.2000, loss_box_4: 2.7850, loss_cns_4: 0.5859, loss_yns_4: 0.1894, loss_cls_5: 1.2198, loss_box_5: 2.8225, loss_cns_5: 0.5826, loss_yns_5: 0.1981, loss_cls_dn_0: 0.4006, loss_box_dn_0: 1.0328, loss_cls_dn_1: 0.4645, loss_box_dn_1: 1.3337, loss_cls_dn_2: 0.4878, loss_box_dn_2: 1.3719, loss_cls_dn_3: 0.4424, loss_box_dn_3: 1.3736, loss_cls_dn_4: 0.4369, loss_box_dn_4: 1.5163, loss_cls_dn_5: 0.4787, loss_box_dn_5: 1.5951, loss_dense_depth: 1.1604, loss: 39.6698, grad_norm: 65.3186 -2026-02-10 16:30:10,243 - mmdet - INFO - Iter [15/23400] lr: 1.056e-04, eta: 20:12:14, time: 1.537, data_time: 0.045, memory: 36808, loss_cls_0: 1.2096, loss_box_0: 2.2532, loss_cns_0: 0.6075, loss_yns_0: 0.1768, loss_cls_1: 1.1867, loss_box_1: 2.5472, loss_cns_1: 0.5779, loss_yns_1: 0.1705, loss_cls_2: 1.2388, loss_box_2: 2.5897, loss_cns_2: 0.5932, loss_yns_2: 0.1804, loss_cls_3: 1.2126, loss_box_3: 2.6993, loss_cns_3: 0.5826, loss_yns_3: 0.1763, loss_cls_4: 1.2150, loss_box_4: 2.8584, loss_cns_4: 0.5496, loss_yns_4: 0.1856, loss_cls_5: 1.1845, loss_box_5: 2.8908, loss_cns_5: 0.5676, loss_yns_5: 0.1809, loss_cls_dn_0: 0.4305, loss_box_dn_0: 1.0373, loss_cls_dn_1: 0.4715, loss_box_dn_1: 1.3595, loss_cls_dn_2: 0.4923, loss_box_dn_2: 1.3613, loss_cls_dn_3: 0.4351, loss_box_dn_3: 1.4000, loss_cls_dn_4: 0.4280, loss_box_dn_4: 1.5242, loss_cls_dn_5: 0.4774, loss_box_dn_5: 1.5819, loss_dense_depth: 1.2517, loss: 39.8856, grad_norm: 98.3994 -2026-02-10 16:30:11,929 - mmdet - INFO - Iter [16/23400] lr: 1.060e-04, eta: 19:37:31, time: 1.687, data_time: 0.154, memory: 36808, loss_cls_0: 1.1493, loss_box_0: 2.1349, loss_cns_0: 0.6148, loss_yns_0: 0.1795, loss_cls_1: 1.1842, loss_box_1: 2.7490, loss_cns_1: 0.5387, loss_yns_1: 0.1664, loss_cls_2: 1.2014, loss_box_2: 2.7377, loss_cns_2: 0.5423, loss_yns_2: 0.1764, loss_cls_3: 1.1874, loss_box_3: 2.8370, loss_cns_3: 0.5457, loss_yns_3: 0.1726, loss_cls_4: 1.2169, loss_box_4: 2.9541, loss_cns_4: 0.5258, loss_yns_4: 0.1786, loss_cls_5: 1.1750, loss_box_5: 3.0764, loss_cns_5: 0.5237, loss_yns_5: 0.1708, loss_cls_dn_0: 0.4188, loss_box_dn_0: 1.0350, loss_cls_dn_1: 0.4495, loss_box_dn_1: 1.4046, loss_cls_dn_2: 0.4417, loss_box_dn_2: 1.3849, loss_cls_dn_3: 0.4011, loss_box_dn_3: 1.4501, loss_cls_dn_4: 0.3892, loss_box_dn_4: 1.5476, loss_cls_dn_5: 0.4410, loss_box_dn_5: 1.6280, loss_dense_depth: 1.2770, loss: 40.2071, grad_norm: 103.9818 -2026-02-10 16:30:13,488 - mmdet - INFO - Iter [17/23400] lr: 1.064e-04, eta: 19:03:55, time: 1.558, data_time: 0.052, memory: 36808, loss_cls_0: 1.1111, loss_box_0: 2.0698, loss_cns_0: 0.6296, loss_yns_0: 0.1813, loss_cls_1: 1.1920, loss_box_1: 3.1293, loss_cns_1: 0.5393, loss_yns_1: 0.1722, loss_cls_2: 1.2058, loss_box_2: 3.0389, loss_cns_2: 0.5411, loss_yns_2: 0.1754, loss_cls_3: 1.1735, loss_box_3: 3.1034, loss_cns_3: 0.5351, loss_yns_3: 0.1828, loss_cls_4: 1.2246, loss_box_4: 3.1187, loss_cns_4: 0.5467, loss_yns_4: 0.1754, loss_cls_5: 1.1837, loss_box_5: 3.2663, loss_cns_5: 0.5299, loss_yns_5: 0.1806, loss_cls_dn_0: 0.4195, loss_box_dn_0: 1.0400, loss_cls_dn_1: 0.4230, loss_box_dn_1: 1.5869, loss_cls_dn_2: 0.4064, loss_box_dn_2: 1.5440, loss_cls_dn_3: 0.3809, loss_box_dn_3: 1.6224, loss_cls_dn_4: 0.3671, loss_box_dn_4: 1.6827, loss_cls_dn_5: 0.4073, loss_box_dn_5: 1.7651, loss_dense_depth: 1.1885, loss: 42.0401, grad_norm: 63.7068 -2026-02-10 16:30:15,042 - mmdet - INFO - Iter [18/23400] lr: 1.068e-04, eta: 18:33:58, time: 1.554, data_time: 0.063, memory: 36808, loss_cls_0: 1.1050, loss_box_0: 2.0899, loss_cns_0: 0.6217, loss_yns_0: 0.1818, loss_cls_1: 1.1723, loss_box_1: 3.2380, loss_cns_1: 0.5355, loss_yns_1: 0.1786, loss_cls_2: 1.2036, loss_box_2: 3.2545, loss_cns_2: 0.5328, loss_yns_2: 0.1826, loss_cls_3: 1.1495, loss_box_3: 3.3669, loss_cns_3: 0.5136, loss_yns_3: 0.1831, loss_cls_4: 1.1626, loss_box_4: 3.2854, loss_cns_4: 0.5223, loss_yns_4: 0.1761, loss_cls_5: 1.1726, loss_box_5: 3.4332, loss_cns_5: 0.4933, loss_yns_5: 0.1784, loss_cls_dn_0: 0.4471, loss_box_dn_0: 1.0276, loss_cls_dn_1: 0.4061, loss_box_dn_1: 1.6232, loss_cls_dn_2: 0.3977, loss_box_dn_2: 1.6257, loss_cls_dn_3: 0.3848, loss_box_dn_3: 1.7225, loss_cls_dn_4: 0.3762, loss_box_dn_4: 1.7213, loss_cls_dn_5: 0.3986, loss_box_dn_5: 1.8231, loss_dense_depth: 1.2695, loss: 43.1567, grad_norm: 84.9090 -2026-02-10 16:30:16,587 - mmdet - INFO - Iter [19/23400] lr: 1.072e-04, eta: 18:06:59, time: 1.545, data_time: 0.042, memory: 36808, loss_cls_0: 1.0686, loss_box_0: 2.1131, loss_cns_0: 0.6089, loss_yns_0: 0.1777, loss_cls_1: 1.1444, loss_box_1: 3.2087, loss_cns_1: 0.5274, loss_yns_1: 0.1790, loss_cls_2: 1.1876, loss_box_2: 3.2886, loss_cns_2: 0.5203, loss_yns_2: 0.1848, loss_cls_3: 1.1388, loss_box_3: 3.3588, loss_cns_3: 0.5029, loss_yns_3: 0.1790, loss_cls_4: 1.1286, loss_box_4: 3.3412, loss_cns_4: 0.5070, loss_yns_4: 0.1736, loss_cls_5: 1.1590, loss_box_5: 3.4620, loss_cns_5: 0.4672, loss_yns_5: 0.1880, loss_cls_dn_0: 0.4353, loss_box_dn_0: 1.0155, loss_cls_dn_1: 0.4115, loss_box_dn_1: 1.3343, loss_cls_dn_2: 0.4187, loss_box_dn_2: 1.4283, loss_cls_dn_3: 0.4103, loss_box_dn_3: 1.4984, loss_cls_dn_4: 0.4212, loss_box_dn_4: 1.5011, loss_cls_dn_5: 0.4157, loss_box_dn_5: 1.5946, loss_dense_depth: 1.0988, loss: 41.7987, grad_norm: 103.0945 -2026-02-10 16:30:18,143 - mmdet - INFO - Iter [20/23400] lr: 1.076e-04, eta: 17:42:55, time: 1.556, data_time: 0.060, memory: 36808, loss_cls_0: 1.0802, loss_box_0: 2.0697, loss_cns_0: 0.6212, loss_yns_0: 0.1819, loss_cls_1: 1.1759, loss_box_1: 2.9677, loss_cns_1: 0.5305, loss_yns_1: 0.1813, loss_cls_2: 1.1593, loss_box_2: 3.0226, loss_cns_2: 0.5254, loss_yns_2: 0.1761, loss_cls_3: 1.1312, loss_box_3: 3.0116, loss_cns_3: 0.5336, loss_yns_3: 0.1831, loss_cls_4: 1.1310, loss_box_4: 3.1277, loss_cns_4: 0.5267, loss_yns_4: 0.1659, loss_cls_5: 1.1604, loss_box_5: 3.1942, loss_cns_5: 0.5155, loss_yns_5: 0.1867, loss_cls_dn_0: 0.4418, loss_box_dn_0: 1.0062, loss_cls_dn_1: 0.3787, loss_box_dn_1: 1.4394, loss_cls_dn_2: 0.3929, loss_box_dn_2: 1.4986, loss_cls_dn_3: 0.3890, loss_box_dn_3: 1.4974, loss_cls_dn_4: 0.4146, loss_box_dn_4: 1.5299, loss_cls_dn_5: 0.3829, loss_box_dn_5: 1.5659, loss_dense_depth: 1.2052, loss: 40.7020, grad_norm: 74.7449 -2026-02-10 16:30:19,678 - mmdet - INFO - Iter [21/23400] lr: 1.080e-04, eta: 17:20:44, time: 1.535, data_time: 0.043, memory: 36808, loss_cls_0: 1.0677, loss_box_0: 2.1353, loss_cns_0: 0.6037, loss_yns_0: 0.1755, loss_cls_1: 1.1903, loss_box_1: 2.8507, loss_cns_1: 0.5392, loss_yns_1: 0.1861, loss_cls_2: 1.1459, loss_box_2: 2.9680, loss_cns_2: 0.5538, loss_yns_2: 0.1815, loss_cls_3: 1.1289, loss_box_3: 2.9235, loss_cns_3: 0.5753, loss_yns_3: 0.1749, loss_cls_4: 1.1255, loss_box_4: 3.0294, loss_cns_4: 0.5745, loss_yns_4: 0.1790, loss_cls_5: 1.1608, loss_box_5: 3.0699, loss_cns_5: 0.5818, loss_yns_5: 0.1750, loss_cls_dn_0: 0.4348, loss_box_dn_0: 1.0041, loss_cls_dn_1: 0.3773, loss_box_dn_1: 1.2404, loss_cls_dn_2: 0.4092, loss_box_dn_2: 1.3252, loss_cls_dn_3: 0.3951, loss_box_dn_3: 1.3632, loss_cls_dn_4: 0.4177, loss_box_dn_4: 1.5070, loss_cls_dn_5: 0.3902, loss_box_dn_5: 1.5611, loss_dense_depth: 1.1564, loss: 39.8775, grad_norm: 77.1765 diff --git a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20260210_162905.log.json b/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20260210_162905.log.json deleted file mode 100644 index 5fc2d5701f4283376d5c13ecda3bf07b85a407ec..0000000000000000000000000000000000000000 --- a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20260210_162905.log.json +++ /dev/null @@ -1,22 +0,0 @@ -{"env_info": "sys.platform: linux\nPython: 3.10.12 (main, Aug 15 2025, 14:32:43) [GCC 11.4.0]\nCUDA available: True\nGPU 0,1,2,3,4,5,6,7: BW200_LC, UBB BW1000_LC\nCUDA_HOME: /opt/dtk\nNVCC: Not Available\nGCC: x86_64-linux-gnu-gcc (Ubuntu 11.4.0-1ubuntu1~22.04) 11.4.0\nPyTorch: 2.5.1\nPyTorch compiling details: PyTorch built with:\n - GCC 10.3\n - C++ Version: 201703\n - Intel(R) Math Kernel Library Version 2020.0.4 Product Build 20200917 for Intel(R) 64 architecture applications\n - OpenMP 201511 (a.k.a. OpenMP 4.5)\n - LAPACK is enabled (usually provided by MKL)\n - NNPACK is enabled\n - CPU capability usage: AVX2\n - HIP Runtime 6.3.25521\n - MIOpen 2.18.0\n - Magma 2.8.0\n - Build settings: BLAS_INFO=mkl, BUILD_TYPE=Release, CXX_COMPILER=/opt/rh/gcc-toolset-10/root/usr/bin/c++, CXX_FLAGS= -D_GLIBCXX_USE_CXX11_ABI=1 -fvisibility-inlines-hidden -DUSE_PTHREADPOOL -DNDEBUG -DUSE_KINETO -DLIBKINETO_NOCUPTI -DLIBKINETO_NOXPUPTI=ON -DUSE_FBGEMM -DUSE_PYTORCH_QNNPACK -DUSE_XNNPACK -DSYMBOLICATE_MOBILE_DEBUG_HANDLE -O2 -fPIC -Wall -Wextra -Werror=return-type -Werror=non-virtual-dtor -Werror=bool-operation -Wnarrowing -Wno-missing-field-initializers -Wno-type-limits -Wno-array-bounds -Wno-unknown-pragmas -Wno-unused-parameter -Wno-strict-overflow -Wno-strict-aliasing -Wno-stringop-overflow -Wsuggest-override -Wno-psabi -Wno-error=old-style-cast -Wno-missing-braces -fdiagnostics-color=always -faligned-new -Wno-unused-but-set-variable -Wno-maybe-uninitialized -fno-math-errno -fno-trapping-math -Werror=format -Wno-stringop-overflow, FORCE_FALLBACK_CUDA_MPI=1, LAPACK_INFO=mkl, PERF_WITH_AVX=1, PERF_WITH_AVX2=1, TORCH_VERSION=2.5.1, USE_CUDA=0, USE_CUDNN=OFF, USE_CUSPARSELT=OFF, USE_EXCEPTION_PTR=1, USE_GFLAGS=1, USE_GLOG=1, USE_GLOO=1, USE_MKL=ON, USE_MKLDNN=0, USE_MPI=1, USE_NCCL=1, USE_NNPACK=ON, USE_OPENMP=1, USE_ROCM=ON, USE_ROCM_KERNEL_ASSERT=OFF, \n\nTorchVision: 0.20.1\nOpenCV: 4.12.0\nMMCV: 1.6.1\nMMCV Compiler: GCC 10.3\nMMCV CUDA Compiler: rocm not available\nMMDetection: 2.25.1+", "config": "plugin = True\nplugin_dir = 'projects/mmdet3d_plugin/'\ndist_params = dict(backend='nccl')\nlog_level = 'INFO'\nwork_dir = './work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704'\ntotal_batch_size = 120\nnum_gpus = 8\nbatch_size = 15\nnum_iters_per_epoch = 234\nnum_epochs = 100\ncheckpoint_epoch_interval = 20\ncheckpoint_config = dict(interval=4680)\nlog_config = dict(\n interval=1,\n hooks=[\n dict(type='TextLoggerHook', by_epoch=False),\n dict(type='TensorboardLoggerHook')\n ])\nload_from = None\nresume_from = None\nworkflow = [('train', 1)]\nfp16 = dict(loss_scale=32.0)\ninput_shape = (704, 256)\ntracking_test = True\ntracking_threshold = 0.2\nclass_names = [\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier',\n 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n]\nnum_classes = 10\nembed_dims = 256\nnum_groups = 8\nnum_decoder = 6\nnum_single_frame_decoder = 1\nuse_deformable_func = True\nstrides = [4, 8, 16, 32]\nnum_levels = 4\nnum_depth_layers = 3\ndrop_out = 0.1\ntemporal = True\ndecouple_attn = True\nwith_quality_estimation = True\nmodel = dict(\n type='Sparse4D',\n use_grid_mask=True,\n use_deformable_func=True,\n img_backbone=dict(\n type='ResNet',\n depth=50,\n num_stages=4,\n frozen_stages=-1,\n norm_eval=False,\n style='pytorch',\n with_cp=True,\n out_indices=(0, 1, 2, 3),\n norm_cfg=dict(type='BN', requires_grad=True),\n pretrained='ckpt/resnet50-19c8e357.pth'),\n img_neck=dict(\n type='FPN',\n num_outs=4,\n start_level=0,\n out_channels=256,\n add_extra_convs='on_output',\n relu_before_extra_convs=True,\n in_channels=[256, 512, 1024, 2048]),\n depth_branch=dict(\n type='DenseDepthNet',\n embed_dims=256,\n num_depth_layers=3,\n loss_weight=0.2),\n head=dict(\n type='Sparse4DHead',\n cls_threshold_to_reg=0.05,\n decouple_attn=True,\n instance_bank=dict(\n type='InstanceBank',\n num_anchor=900,\n embed_dims=256,\n anchor='nuscenes_kmeans900.npy',\n anchor_handler=dict(type='SparseBox3DKeyPointsGenerator'),\n num_temp_instances=600,\n confidence_decay=0.6,\n feat_grad=False),\n anchor_encoder=dict(\n type='SparseBox3DEncoder',\n vel_dims=3,\n embed_dims=[128, 32, 32, 64],\n mode='cat',\n output_fc=False,\n in_loops=1,\n out_loops=4),\n num_single_frame_decoder=1,\n operation_order=[\n 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm',\n 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm',\n 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm',\n 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm',\n 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm',\n 'deformable', 'ffn', 'norm', 'refine'\n ],\n temp_graph_model=dict(\n type='MultiheadAttention',\n embed_dims=512,\n num_heads=8,\n batch_first=True,\n dropout=0.1),\n graph_model=dict(\n type='MultiheadAttention',\n embed_dims=512,\n num_heads=8,\n batch_first=True,\n dropout=0.1),\n norm_layer=dict(type='LN', normalized_shape=256),\n ffn=dict(\n type='AsymmetricFFN',\n in_channels=512,\n pre_norm=dict(type='LN'),\n embed_dims=256,\n feedforward_channels=1024,\n num_fcs=2,\n ffn_drop=0.1,\n act_cfg=dict(type='ReLU', inplace=True)),\n deformable_model=dict(\n type='DeformableFeatureAggregation',\n embed_dims=256,\n num_groups=8,\n num_levels=4,\n num_cams=6,\n attn_drop=0.15,\n use_deformable_func=True,\n use_camera_embed=True,\n residual_mode='cat',\n kps_generator=dict(\n type='SparseBox3DKeyPointsGenerator',\n num_learnable_pts=6,\n fix_scale=[[0, 0, 0], [0.45, 0, 0], [-0.45, 0, 0],\n [0, 0.45, 0], [0, -0.45, 0], [0, 0, 0.45],\n [0, 0, -0.45]])),\n refine_layer=dict(\n type='SparseBox3DRefinementModule',\n embed_dims=256,\n num_cls=10,\n refine_yaw=True,\n with_quality_estimation=True),\n sampler=dict(\n type='SparseBox3DTarget',\n num_dn_groups=5,\n num_temp_dn_groups=3,\n dn_noise_scale=[2.0, 2.0, 2.0, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5],\n max_dn_gt=32,\n add_neg_dn=True,\n cls_weight=2.0,\n box_weight=0.25,\n reg_weights=[2.0, 2.0, 2.0, 0.5, 0.5, 0.5, 0.0, 0.0, 0.0, 0.0],\n cls_wise_reg_weights=dict(\n {9: [2.0, 2.0, 2.0, 1.0, 1.0, 1.0, 0.0, 0.0, 1.0, 1.0]})),\n loss_cls=dict(\n type='FocalLoss',\n use_sigmoid=True,\n gamma=2.0,\n alpha=0.25,\n loss_weight=2.0),\n loss_reg=dict(\n type='SparseBox3DLoss',\n loss_box=dict(type='L1Loss', loss_weight=0.25),\n loss_centerness=dict(type='CrossEntropyLoss', use_sigmoid=True),\n loss_yawness=dict(type='GaussianFocalLoss'),\n cls_allow_reverse=[5]),\n decoder=dict(type='SparseBox3DDecoder'),\n reg_weights=[2.0, 2.0, 2.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0]))\ndataset_type = 'NuScenes3DDetTrackDataset'\ndata_root = 'data/nuscenes/'\nanno_root = 'data/nuscenes_anno_pkls/'\nfile_client_args = dict(backend='disk')\nimg_norm_cfg = dict(\n mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375], to_rgb=True)\ntrain_pipeline = [\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(\n type='LoadPointsFromFile',\n coord_type='LIDAR',\n load_dim=5,\n use_dim=5,\n file_client_args=dict(backend='disk')),\n dict(type='ResizeCropFlipImage'),\n dict(type='MultiScaleDepthMapGenerator', downsample=[4, 8, 16]),\n dict(type='BBoxRotation'),\n dict(type='PhotoMetricDistortionMultiViewImage'),\n dict(\n type='NormalizeMultiviewImage',\n mean=[123.675, 116.28, 103.53],\n std=[58.395, 57.12, 57.375],\n to_rgb=True),\n dict(\n type='CircleObjectRangeFilter',\n class_dist_thred=[55, 55, 55, 55, 55, 55, 55, 55, 55, 55]),\n dict(\n type='InstanceNameFilter',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ]),\n dict(type='NuScenesSparse4DAdaptor'),\n dict(\n type='Collect',\n keys=[\n 'img', 'timestamp', 'projection_mat', 'image_wh', 'gt_depth',\n 'focal', 'gt_bboxes_3d', 'gt_labels_3d'\n ],\n meta_keys=['T_global', 'T_global_inv', 'timestamp', 'instance_id'])\n]\ntest_pipeline = [\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(type='ResizeCropFlipImage'),\n dict(\n type='NormalizeMultiviewImage',\n mean=[123.675, 116.28, 103.53],\n std=[58.395, 57.12, 57.375],\n to_rgb=True),\n dict(type='NuScenesSparse4DAdaptor'),\n dict(\n type='Collect',\n keys=['img', 'timestamp', 'projection_mat', 'image_wh'],\n meta_keys=['T_global', 'T_global_inv', 'timestamp'])\n]\ninput_modality = dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False)\ndata_basic_config = dict(\n type='NuScenes3DDetTrackDataset',\n data_root='data/nuscenes/',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier',\n 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ],\n modality=dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False),\n version='v1.0-trainval')\ndata_aug_conf = dict(\n resize_lim=(0.4, 0.47),\n final_dim=(256, 704),\n bot_pct_lim=(0.0, 0.0),\n rot_lim=(-5.4, 5.4),\n H=900,\n W=1600,\n rand_flip=True,\n rot3d_range=[-0.3925, 0.3925])\ndata = dict(\n samples_per_gpu=15,\n workers_per_gpu=15,\n train=dict(\n type='NuScenes3DDetTrackDataset',\n data_root='data/nuscenes/',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ],\n modality=dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False),\n version='v1.0-trainval',\n ann_file='data/nuscenes_anno_pkls/nuscenes_infos_train.pkl',\n pipeline=[\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(\n type='LoadPointsFromFile',\n coord_type='LIDAR',\n load_dim=5,\n use_dim=5,\n file_client_args=dict(backend='disk')),\n dict(type='ResizeCropFlipImage'),\n dict(type='MultiScaleDepthMapGenerator', downsample=[4, 8, 16]),\n dict(type='BBoxRotation'),\n dict(type='PhotoMetricDistortionMultiViewImage'),\n dict(\n type='NormalizeMultiviewImage',\n mean=[123.675, 116.28, 103.53],\n std=[58.395, 57.12, 57.375],\n to_rgb=True),\n dict(\n type='CircleObjectRangeFilter',\n class_dist_thred=[55, 55, 55, 55, 55, 55, 55, 55, 55, 55]),\n dict(\n type='InstanceNameFilter',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian',\n 'traffic_cone'\n ]),\n dict(type='NuScenesSparse4DAdaptor'),\n dict(\n type='Collect',\n keys=[\n 'img', 'timestamp', 'projection_mat', 'image_wh',\n 'gt_depth', 'focal', 'gt_bboxes_3d', 'gt_labels_3d'\n ],\n meta_keys=[\n 'T_global', 'T_global_inv', 'timestamp', 'instance_id'\n ])\n ],\n test_mode=False,\n data_aug_conf=dict(\n resize_lim=(0.4, 0.47),\n final_dim=(256, 704),\n bot_pct_lim=(0.0, 0.0),\n rot_lim=(-5.4, 5.4),\n H=900,\n W=1600,\n rand_flip=True,\n rot3d_range=[-0.3925, 0.3925]),\n with_seq_flag=True,\n sequences_split_num=2,\n keep_consistent_seq_aug=True),\n val=dict(\n type='NuScenes3DDetTrackDataset',\n data_root='data/nuscenes/',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ],\n modality=dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False),\n version='v1.0-trainval',\n ann_file='data/nuscenes_anno_pkls/nuscenes_infos_val.pkl',\n pipeline=[\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(type='ResizeCropFlipImage'),\n dict(\n type='NormalizeMultiviewImage',\n mean=[123.675, 116.28, 103.53],\n std=[58.395, 57.12, 57.375],\n to_rgb=True),\n dict(type='NuScenesSparse4DAdaptor'),\n dict(\n type='Collect',\n keys=['img', 'timestamp', 'projection_mat', 'image_wh'],\n meta_keys=['T_global', 'T_global_inv', 'timestamp'])\n ],\n data_aug_conf=dict(\n resize_lim=(0.4, 0.47),\n final_dim=(256, 704),\n bot_pct_lim=(0.0, 0.0),\n rot_lim=(-5.4, 5.4),\n H=900,\n W=1600,\n rand_flip=True,\n rot3d_range=[-0.3925, 0.3925]),\n test_mode=True,\n tracking=True,\n tracking_threshold=0.2),\n test=dict(\n type='NuScenes3DDetTrackDataset',\n data_root='data/nuscenes/',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ],\n modality=dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False),\n version='v1.0-trainval',\n ann_file='data/nuscenes_anno_pkls/nuscenes_infos_val.pkl',\n pipeline=[\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(type='ResizeCropFlipImage'),\n dict(\n type='NormalizeMultiviewImage',\n mean=[123.675, 116.28, 103.53],\n std=[58.395, 57.12, 57.375],\n to_rgb=True),\n dict(type='NuScenesSparse4DAdaptor'),\n dict(\n type='Collect',\n keys=['img', 'timestamp', 'projection_mat', 'image_wh'],\n meta_keys=['T_global', 'T_global_inv', 'timestamp'])\n ],\n data_aug_conf=dict(\n resize_lim=(0.4, 0.47),\n final_dim=(256, 704),\n bot_pct_lim=(0.0, 0.0),\n rot_lim=(-5.4, 5.4),\n H=900,\n W=1600,\n rand_flip=True,\n rot3d_range=[-0.3925, 0.3925]),\n test_mode=True,\n tracking=True,\n tracking_threshold=0.2))\noptimizer = dict(\n type='AdamW',\n lr=0.0006,\n weight_decay=0.001,\n paramwise_cfg=dict(custom_keys=dict(img_backbone=dict(lr_mult=0.5))))\noptimizer_config = dict(grad_clip=dict(max_norm=25, norm_type=2))\nlr_config = dict(\n policy='CosineAnnealing',\n warmup='linear',\n warmup_iters=500,\n warmup_ratio=0.3333333333333333,\n min_lr_ratio=0.001)\nrunner = dict(type='IterBasedRunner', max_iters=23400)\nvis_pipeline = [\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(type='Collect', keys=['img'], meta_keys=['timestamp', 'lidar2img'])\n]\nevaluation = dict(\n interval=4680,\n pipeline=[\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(\n type='Collect', keys=['img'], meta_keys=['timestamp', 'lidar2img'])\n ])\ngpu_ids = range(0, 1)\n", "seed": 0, "exp_name": "sparse4dv3_temporal_r50_1x8_bs6_256x704.py"} -{"mode": "train", "epoch": 1, "iter": 1, "lr": 0.0001, "memory": 36808, "data_time": 5.76691, "loss_cls_0": 2.30839, "loss_box_0": 0.0, "loss_cns_0": 0.0, "loss_yns_0": 0.0, "loss_cls_1": 2.18209, "loss_box_1": 0.07232, "loss_cns_1": 0.01579, "loss_yns_1": 0.00541, "loss_cls_2": 2.39557, "loss_box_2": 0.0, "loss_cns_2": 0.0, "loss_yns_2": 0.0, "loss_cls_3": 2.43429, "loss_box_3": 0.02711, "loss_cns_3": 0.00505, "loss_yns_3": 0.00183, "loss_cls_4": 1.9844, "loss_box_4": 0.46771, "loss_cns_4": 0.05375, "loss_yns_4": 0.02685, "loss_cls_5": 2.31038, "loss_box_5": 0.0, "loss_cns_5": 0.0, "loss_yns_5": 0.0, "loss_cls_dn_0": 1.1695, "loss_box_dn_0": 1.46509, "loss_cls_dn_1": 1.15224, "loss_box_dn_1": 1.74642, "loss_cls_dn_2": 1.18898, "loss_box_dn_2": 1.98701, "loss_cls_dn_3": 1.15224, "loss_box_dn_3": 2.2653, "loss_cls_dn_4": 1.04055, "loss_box_dn_4": 2.43186, "loss_cls_dn_5": 1.17508, "loss_box_dn_5": 2.68808, "loss_dense_depth": 1.84485, "loss": 35.59811, "grad_norm": 428.58261, "time": 24.56481} -{"mode": "train", "epoch": 1, "iter": 2, "lr": 0.0001, "memory": 36808, "data_time": 0.05178, "loss_cls_0": 2.14696, "loss_box_0": 0.03881, "loss_cns_0": 0.01248, "loss_yns_0": 0.00138, "loss_cls_1": 2.04836, "loss_box_1": 0.26089, "loss_cns_1": 0.05203, "loss_yns_1": 0.01831, "loss_cls_2": 2.16365, "loss_box_2": 0.11168, "loss_cns_2": 0.01067, "loss_yns_2": 0.00374, "loss_cls_3": 2.0164, "loss_box_3": 0.15962, "loss_cns_3": 0.02036, "loss_yns_3": 0.00628, "loss_cls_4": 1.85262, "loss_box_4": 0.74086, "loss_cns_4": 0.09029, "loss_yns_4": 0.02821, "loss_cls_5": 1.9376, "loss_box_5": 1.22675, "loss_cns_5": 0.1453, "loss_yns_5": 0.06192, "loss_cls_dn_0": 1.09583, "loss_box_dn_0": 1.22212, "loss_cls_dn_1": 1.00057, "loss_box_dn_1": 2.52451, "loss_cls_dn_2": 1.01182, "loss_box_dn_2": 2.58878, "loss_cls_dn_3": 0.94804, "loss_box_dn_3": 2.67513, "loss_cls_dn_4": 0.87931, "loss_box_dn_4": 2.90513, "loss_cls_dn_5": 0.96029, "loss_box_dn_5": 3.12594, "loss_dense_depth": 1.77742, "loss": 37.87007, "grad_norm": 74.8507, "time": 2.19952} -{"mode": "train", "epoch": 1, "iter": 3, "lr": 0.0001, "memory": 36808, "data_time": 0.04474, "loss_cls_0": 1.60829, "loss_box_0": 2.68762, "loss_cns_0": 0.65737, "loss_yns_0": 0.35052, "loss_cls_1": 1.84416, "loss_box_1": 0.92713, "loss_cns_1": 0.16281, "loss_yns_1": 0.04245, "loss_cls_2": 1.86137, "loss_box_2": 1.67592, "loss_cns_2": 0.17476, "loss_yns_2": 0.08346, "loss_cls_3": 1.66056, "loss_box_3": 2.60972, "loss_cns_3": 0.26071, "loss_yns_3": 0.16171, "loss_cls_4": 1.5519, "loss_box_4": 4.23991, "loss_cns_4": 0.42276, "loss_yns_4": 0.16381, "loss_cls_5": 1.66865, "loss_box_5": 2.91982, "loss_cns_5": 0.22741, "loss_yns_5": 0.10353, "loss_cls_dn_0": 0.81905, "loss_box_dn_0": 1.24593, "loss_cls_dn_1": 0.84471, "loss_box_dn_1": 2.43175, "loss_cls_dn_2": 0.83491, "loss_box_dn_2": 2.5661, "loss_cls_dn_3": 0.74794, "loss_box_dn_3": 2.7701, "loss_cls_dn_4": 0.71634, "loss_box_dn_4": 2.96169, "loss_cls_dn_5": 0.77752, "loss_box_dn_5": 3.14629, "loss_dense_depth": 1.75753, "loss": 49.6862, "grad_norm": 100.83112, "time": 1.50294} -{"mode": "train", "epoch": 1, "iter": 4, "lr": 0.0001, "memory": 36808, "data_time": 0.04418, "loss_cls_0": 1.39845, "loss_box_0": 2.73856, "loss_cns_0": 0.62168, "loss_yns_0": 0.2357, "loss_cls_1": 1.64642, "loss_box_1": 2.38242, "loss_cns_1": 0.3478, "loss_yns_1": 0.12348, "loss_cls_2": 1.70544, "loss_box_2": 3.11034, "loss_cns_2": 0.37728, "loss_yns_2": 0.14547, "loss_cls_3": 1.46571, "loss_box_3": 4.3098, "loss_cns_3": 0.4589, "loss_yns_3": 0.26361, "loss_cls_4": 1.4014, "loss_box_4": 4.7024, "loss_cns_4": 0.38206, "loss_yns_4": 0.17288, "loss_cls_5": 1.52214, "loss_box_5": 4.80815, "loss_cns_5": 0.44998, "loss_yns_5": 0.1959, "loss_cls_dn_0": 0.69831, "loss_box_dn_0": 1.14786, "loss_cls_dn_1": 0.76455, "loss_box_dn_1": 2.51959, "loss_cls_dn_2": 0.756, "loss_box_dn_2": 2.59555, "loss_cls_dn_3": 0.65815, "loss_box_dn_3": 2.78373, "loss_cls_dn_4": 0.62446, "loss_box_dn_4": 2.91344, "loss_cls_dn_5": 0.67449, "loss_box_dn_5": 3.0462, "loss_dense_depth": 1.6257, "loss": 55.774, "grad_norm": 127.17017, "time": 1.50774} -{"mode": "train", "epoch": 1, "iter": 5, "lr": 0.0001, "memory": 36808, "data_time": 0.04946, "loss_cls_0": 1.30042, "loss_box_0": 2.70159, "loss_cns_0": 0.5562, "loss_yns_0": 0.18355, "loss_cls_1": 1.48741, "loss_box_1": 3.33355, "loss_cns_1": 0.37401, "loss_yns_1": 0.15929, "loss_cls_2": 1.54112, "loss_box_2": 3.762, "loss_cns_2": 0.36369, "loss_yns_2": 0.18672, "loss_cls_3": 1.38705, "loss_box_3": 4.10307, "loss_cns_3": 0.42555, "loss_yns_3": 0.22641, "loss_cls_4": 1.398, "loss_box_4": 4.23185, "loss_cns_4": 0.39292, "loss_yns_4": 0.18416, "loss_cls_5": 1.44825, "loss_box_5": 4.50882, "loss_cns_5": 0.43903, "loss_yns_5": 0.18708, "loss_cls_dn_0": 0.61832, "loss_box_dn_0": 1.16578, "loss_cls_dn_1": 0.68914, "loss_box_dn_1": 2.3754, "loss_cls_dn_2": 0.69348, "loss_box_dn_2": 2.54262, "loss_cls_dn_3": 0.57842, "loss_box_dn_3": 2.61716, "loss_cls_dn_4": 0.55149, "loss_box_dn_4": 2.73783, "loss_cls_dn_5": 0.59537, "loss_box_dn_5": 2.88024, "loss_dense_depth": 1.63494, "loss": 54.56193, "grad_norm": 116.10751, "time": 1.51609} -{"mode": "train", "epoch": 1, "iter": 6, "lr": 0.0001, "memory": 36808, "data_time": 0.04625, "loss_cls_0": 1.29895, "loss_box_0": 2.63932, "loss_cns_0": 0.5968, "loss_yns_0": 0.21969, "loss_cls_1": 1.3902, "loss_box_1": 3.44198, "loss_cns_1": 0.3853, "loss_yns_1": 0.18265, "loss_cls_2": 1.41744, "loss_box_2": 3.68139, "loss_cns_2": 0.39921, "loss_yns_2": 0.18844, "loss_cls_3": 1.32942, "loss_box_3": 3.60814, "loss_cns_3": 0.43518, "loss_yns_3": 0.21925, "loss_cls_4": 1.34075, "loss_box_4": 3.83031, "loss_cns_4": 0.39761, "loss_yns_4": 0.20719, "loss_cls_5": 1.38962, "loss_box_5": 4.20156, "loss_cns_5": 0.36724, "loss_yns_5": 0.19084, "loss_cls_dn_0": 0.55012, "loss_box_dn_0": 1.15723, "loss_cls_dn_1": 0.63966, "loss_box_dn_1": 2.79295, "loss_cls_dn_2": 0.63907, "loss_box_dn_2": 2.8185, "loss_cls_dn_3": 0.54303, "loss_box_dn_3": 2.82858, "loss_cls_dn_4": 0.50826, "loss_box_dn_4": 2.96365, "loss_cls_dn_5": 0.52984, "loss_box_dn_5": 3.10266, "loss_dense_depth": 1.66172, "loss": 54.09377, "grad_norm": 133.97675, "time": 1.51789} -{"mode": "train", "epoch": 1, "iter": 7, "lr": 0.0001, "memory": 36808, "data_time": 0.04249, "loss_cls_0": 1.41081, "loss_box_0": 2.45411, "loss_cns_0": 0.67727, "loss_yns_0": 0.20556, "loss_cls_1": 1.28885, "loss_box_1": 3.76708, "loss_cns_1": 0.40654, "loss_yns_1": 0.17979, "loss_cls_2": 1.36641, "loss_box_2": 3.8412, "loss_cns_2": 0.39712, "loss_yns_2": 0.17558, "loss_cls_3": 1.3261, "loss_box_3": 3.76515, "loss_cns_3": 0.4019, "loss_yns_3": 0.20289, "loss_cls_4": 1.32753, "loss_box_4": 3.97059, "loss_cns_4": 0.41694, "loss_yns_4": 0.19325, "loss_cls_5": 1.31302, "loss_box_5": 4.04089, "loss_cns_5": 0.43687, "loss_yns_5": 0.18244, "loss_cls_dn_0": 0.50691, "loss_box_dn_0": 1.09521, "loss_cls_dn_1": 0.57054, "loss_box_dn_1": 3.06915, "loss_cls_dn_2": 0.57687, "loss_box_dn_2": 3.01753, "loss_cls_dn_3": 0.49799, "loss_box_dn_3": 2.99081, "loss_cls_dn_4": 0.46841, "loss_box_dn_4": 3.07328, "loss_cls_dn_5": 0.46768, "loss_box_dn_5": 3.14285, "loss_dense_depth": 1.51339, "loss": 54.7385, "grad_norm": 95.21945, "time": 1.5019} -{"mode": "train", "epoch": 1, "iter": 8, "lr": 0.0001, "memory": 36808, "data_time": 0.04658, "loss_cls_0": 1.26075, "loss_box_0": 2.36868, "loss_cns_0": 0.66021, "loss_yns_0": 0.16862, "loss_cls_1": 1.29046, "loss_box_1": 3.63176, "loss_cns_1": 0.45015, "loss_yns_1": 0.17272, "loss_cls_2": 1.38519, "loss_box_2": 3.64481, "loss_cns_2": 0.38629, "loss_yns_2": 0.18101, "loss_cls_3": 1.33011, "loss_box_3": 3.74817, "loss_cns_3": 0.36868, "loss_yns_3": 0.18939, "loss_cls_4": 1.29846, "loss_box_4": 3.82328, "loss_cns_4": 0.38896, "loss_yns_4": 0.17606, "loss_cls_5": 1.32479, "loss_box_5": 3.86475, "loss_cns_5": 0.45401, "loss_yns_5": 0.18323, "loss_cls_dn_0": 0.51258, "loss_box_dn_0": 1.07149, "loss_cls_dn_1": 0.55449, "loss_box_dn_1": 1.91393, "loss_cls_dn_2": 0.56606, "loss_box_dn_2": 1.87324, "loss_cls_dn_3": 0.49756, "loss_box_dn_3": 1.95336, "loss_cls_dn_4": 0.48701, "loss_box_dn_4": 2.0394, "loss_cls_dn_5": 0.48996, "loss_box_dn_5": 2.16664, "loss_dense_depth": 1.56232, "loss": 48.43859, "grad_norm": 82.10076, "time": 1.5182} -{"mode": "train", "epoch": 1, "iter": 9, "lr": 0.0001, "memory": 36808, "data_time": 0.05147, "loss_cls_0": 1.20165, "loss_box_0": 2.41878, "loss_cns_0": 0.60623, "loss_yns_0": 0.16723, "loss_cls_1": 1.33455, "loss_box_1": 3.29309, "loss_cns_1": 0.52921, "loss_yns_1": 0.17821, "loss_cls_2": 1.40531, "loss_box_2": 3.26687, "loss_cns_2": 0.48616, "loss_yns_2": 0.17928, "loss_cls_3": 1.31634, "loss_box_3": 3.45514, "loss_cns_3": 0.47733, "loss_yns_3": 0.17395, "loss_cls_4": 1.26557, "loss_box_4": 3.40914, "loss_cns_4": 0.48149, "loss_yns_4": 0.18062, "loss_cls_5": 1.33056, "loss_box_5": 3.4383, "loss_cns_5": 0.54579, "loss_yns_5": 0.17993, "loss_cls_dn_0": 0.52128, "loss_box_dn_0": 1.04582, "loss_cls_dn_1": 0.49483, "loss_box_dn_1": 1.5746, "loss_cls_dn_2": 0.51932, "loss_box_dn_2": 1.58811, "loss_cls_dn_3": 0.47226, "loss_box_dn_3": 1.73872, "loss_cls_dn_4": 0.4832, "loss_box_dn_4": 1.75484, "loss_cls_dn_5": 0.47825, "loss_box_dn_5": 1.8507, "loss_dense_depth": 1.41894, "loss": 45.26162, "grad_norm": 71.86478, "time": 1.53314} -{"mode": "train", "epoch": 1, "iter": 10, "lr": 0.0001, "memory": 36808, "data_time": 0.04557, "loss_cls_0": 1.1964, "loss_box_0": 2.46699, "loss_cns_0": 0.56939, "loss_yns_0": 0.16548, "loss_cls_1": 1.31547, "loss_box_1": 3.30025, "loss_cns_1": 0.44881, "loss_yns_1": 0.17597, "loss_cls_2": 1.30989, "loss_box_2": 3.19977, "loss_cns_2": 0.47156, "loss_yns_2": 0.17422, "loss_cls_3": 1.25031, "loss_box_3": 3.4266, "loss_cns_3": 0.50105, "loss_yns_3": 0.18159, "loss_cls_4": 1.25349, "loss_box_4": 3.36372, "loss_cns_4": 0.49576, "loss_yns_4": 0.16787, "loss_cls_5": 1.29254, "loss_box_5": 3.48831, "loss_cns_5": 0.46507, "loss_yns_5": 0.17522, "loss_cls_dn_0": 0.51897, "loss_box_dn_0": 1.07522, "loss_cls_dn_1": 0.44644, "loss_box_dn_1": 1.73115, "loss_cls_dn_2": 0.48825, "loss_box_dn_2": 1.77608, "loss_cls_dn_3": 0.45691, "loss_box_dn_3": 1.91564, "loss_cls_dn_4": 0.46411, "loss_box_dn_4": 1.91569, "loss_cls_dn_5": 0.45538, "loss_box_dn_5": 2.03844, "loss_dense_depth": 1.39759, "loss": 45.53561, "grad_norm": 87.49928, "time": 1.5479} -{"mode": "train", "epoch": 1, "iter": 11, "lr": 0.0001, "memory": 36808, "data_time": 0.05762, "loss_cls_0": 1.20269, "loss_box_0": 2.41467, "loss_cns_0": 0.58439, "loss_yns_0": 0.16845, "loss_cls_1": 1.25025, "loss_box_1": 3.40911, "loss_cns_1": 0.41289, "loss_yns_1": 0.16427, "loss_cls_2": 1.28974, "loss_box_2": 3.38279, "loss_cns_2": 0.43751, "loss_yns_2": 0.17448, "loss_cls_3": 1.23939, "loss_box_3": 3.50243, "loss_cns_3": 0.44953, "loss_yns_3": 0.17801, "loss_cls_4": 1.23333, "loss_box_4": 3.42427, "loss_cns_4": 0.43449, "loss_yns_4": 0.17425, "loss_cls_5": 1.3084, "loss_box_5": 3.45679, "loss_cns_5": 0.42944, "loss_yns_5": 0.19125, "loss_cls_dn_0": 0.50935, "loss_box_dn_0": 1.05357, "loss_cls_dn_1": 0.42104, "loss_box_dn_1": 1.94903, "loss_cls_dn_2": 0.48144, "loss_box_dn_2": 2.00953, "loss_cls_dn_3": 0.45811, "loss_box_dn_3": 2.1086, "loss_cls_dn_4": 0.4424, "loss_box_dn_4": 2.0731, "loss_cls_dn_5": 0.44097, "loss_box_dn_5": 2.15197, "loss_dense_depth": 1.40644, "loss": 46.41834, "grad_norm": 83.96349, "time": 1.55933} -{"mode": "train", "epoch": 1, "iter": 12, "lr": 0.0001, "memory": 36808, "data_time": 0.0522, "loss_cls_0": 1.16015, "loss_box_0": 2.40329, "loss_cns_0": 0.59222, "loss_yns_0": 0.1672, "loss_cls_1": 1.24476, "loss_box_1": 3.05811, "loss_cns_1": 0.51386, "loss_yns_1": 0.17087, "loss_cls_2": 1.2897, "loss_box_2": 3.07141, "loss_cns_2": 0.50474, "loss_yns_2": 0.1883, "loss_cls_3": 1.22055, "loss_box_3": 3.08976, "loss_cns_3": 0.50435, "loss_yns_3": 0.16681, "loss_cls_4": 1.21884, "loss_box_4": 3.10709, "loss_cns_4": 0.50579, "loss_yns_4": 0.18796, "loss_cls_5": 1.29464, "loss_box_5": 3.26673, "loss_cns_5": 0.50153, "loss_yns_5": 0.19461, "loss_cls_dn_0": 0.46816, "loss_box_dn_0": 1.073, "loss_cls_dn_1": 0.40317, "loss_box_dn_1": 2.11449, "loss_cls_dn_2": 0.46564, "loss_box_dn_2": 2.15266, "loss_cls_dn_3": 0.44175, "loss_box_dn_3": 2.18287, "loss_cls_dn_4": 0.42157, "loss_box_dn_4": 2.15186, "loss_cls_dn_5": 0.42415, "loss_box_dn_5": 2.18995, "loss_dense_depth": 1.29678, "loss": 45.4093, "grad_norm": 79.39165, "time": 1.55179} -{"mode": "train", "epoch": 1, "iter": 13, "lr": 0.0001, "memory": 36808, "data_time": 0.04972, "loss_cls_0": 1.1722, "loss_box_0": 2.41806, "loss_cns_0": 0.59119, "loss_yns_0": 0.17569, "loss_cls_1": 1.22681, "loss_box_1": 3.04684, "loss_cns_1": 0.51412, "loss_yns_1": 0.17254, "loss_cls_2": 1.27752, "loss_box_2": 3.03638, "loss_cns_2": 0.50322, "loss_yns_2": 0.17905, "loss_cls_3": 1.21983, "loss_box_3": 3.1171, "loss_cns_3": 0.49009, "loss_yns_3": 0.17388, "loss_cls_4": 1.21182, "loss_box_4": 3.17369, "loss_cns_4": 0.50003, "loss_yns_4": 0.18549, "loss_cls_5": 1.2532, "loss_box_5": 3.29582, "loss_cns_5": 0.48702, "loss_yns_5": 0.21279, "loss_cls_dn_0": 0.46443, "loss_box_dn_0": 1.05881, "loss_cls_dn_1": 0.446, "loss_box_dn_1": 1.59304, "loss_cls_dn_2": 0.51105, "loss_box_dn_2": 1.64481, "loss_cls_dn_3": 0.46705, "loss_box_dn_3": 1.65465, "loss_cls_dn_4": 0.45847, "loss_box_dn_4": 1.73227, "loss_cls_dn_5": 0.47002, "loss_box_dn_5": 1.80259, "loss_dense_depth": 1.33671, "loss": 43.2743, "grad_norm": 87.16595, "time": 1.5477} -{"mode": "train", "epoch": 1, "iter": 14, "lr": 0.00011, "memory": 36808, "data_time": 0.05221, "loss_cls_0": 1.16538, "loss_box_0": 2.26851, "loss_cns_0": 0.60659, "loss_yns_0": 0.18163, "loss_cls_1": 1.18879, "loss_box_1": 2.59563, "loss_cns_1": 0.59043, "loss_yns_1": 0.17295, "loss_cls_2": 1.21247, "loss_box_2": 2.63511, "loss_cns_2": 0.57623, "loss_yns_2": 0.1929, "loss_cls_3": 1.18773, "loss_box_3": 2.64627, "loss_cns_3": 0.57987, "loss_yns_3": 0.1914, "loss_cls_4": 1.19996, "loss_box_4": 2.78502, "loss_cns_4": 0.58593, "loss_yns_4": 0.18937, "loss_cls_5": 1.21978, "loss_box_5": 2.82246, "loss_cns_5": 0.58257, "loss_yns_5": 0.1981, "loss_cls_dn_0": 0.40061, "loss_box_dn_0": 1.03284, "loss_cls_dn_1": 0.46453, "loss_box_dn_1": 1.33373, "loss_cls_dn_2": 0.48779, "loss_box_dn_2": 1.37192, "loss_cls_dn_3": 0.44238, "loss_box_dn_3": 1.37364, "loss_cls_dn_4": 0.4369, "loss_box_dn_4": 1.51626, "loss_cls_dn_5": 0.47869, "loss_box_dn_5": 1.59505, "loss_dense_depth": 1.16038, "loss": 39.66982, "grad_norm": 65.31857, "time": 1.54899} -{"mode": "train", "epoch": 1, "iter": 15, "lr": 0.00011, "memory": 36808, "data_time": 0.0452, "loss_cls_0": 1.20958, "loss_box_0": 2.25322, "loss_cns_0": 0.60753, "loss_yns_0": 0.17678, "loss_cls_1": 1.18671, "loss_box_1": 2.54722, "loss_cns_1": 0.57788, "loss_yns_1": 0.17051, "loss_cls_2": 1.23883, "loss_box_2": 2.58974, "loss_cns_2": 0.59317, "loss_yns_2": 0.18038, "loss_cls_3": 1.2126, "loss_box_3": 2.69933, "loss_cns_3": 0.58262, "loss_yns_3": 0.17628, "loss_cls_4": 1.21498, "loss_box_4": 2.8584, "loss_cns_4": 0.54965, "loss_yns_4": 0.18555, "loss_cls_5": 1.18447, "loss_box_5": 2.89077, "loss_cns_5": 0.56757, "loss_yns_5": 0.18088, "loss_cls_dn_0": 0.43053, "loss_box_dn_0": 1.03733, "loss_cls_dn_1": 0.47146, "loss_box_dn_1": 1.35951, "loss_cls_dn_2": 0.49232, "loss_box_dn_2": 1.36134, "loss_cls_dn_3": 0.43513, "loss_box_dn_3": 1.4, "loss_cls_dn_4": 0.42802, "loss_box_dn_4": 1.52422, "loss_cls_dn_5": 0.47741, "loss_box_dn_5": 1.58192, "loss_dense_depth": 1.25173, "loss": 39.88558, "grad_norm": 98.39938, "time": 1.53697} -{"mode": "train", "epoch": 1, "iter": 16, "lr": 0.00011, "memory": 36808, "data_time": 0.15419, "loss_cls_0": 1.14931, "loss_box_0": 2.13491, "loss_cns_0": 0.61475, "loss_yns_0": 0.17954, "loss_cls_1": 1.18417, "loss_box_1": 2.74897, "loss_cns_1": 0.53868, "loss_yns_1": 0.16638, "loss_cls_2": 1.2014, "loss_box_2": 2.73767, "loss_cns_2": 0.54229, "loss_yns_2": 0.17641, "loss_cls_3": 1.18735, "loss_box_3": 2.83704, "loss_cns_3": 0.54573, "loss_yns_3": 0.17265, "loss_cls_4": 1.21686, "loss_box_4": 2.95414, "loss_cns_4": 0.52579, "loss_yns_4": 0.17865, "loss_cls_5": 1.17502, "loss_box_5": 3.07642, "loss_cns_5": 0.52375, "loss_yns_5": 0.17076, "loss_cls_dn_0": 0.41883, "loss_box_dn_0": 1.03504, "loss_cls_dn_1": 0.44954, "loss_box_dn_1": 1.40461, "loss_cls_dn_2": 0.44165, "loss_box_dn_2": 1.38486, "loss_cls_dn_3": 0.4011, "loss_box_dn_3": 1.45008, "loss_cls_dn_4": 0.3892, "loss_box_dn_4": 1.54757, "loss_cls_dn_5": 0.441, "loss_box_dn_5": 1.62797, "loss_dense_depth": 1.27704, "loss": 40.20712, "grad_norm": 103.98182, "time": 1.68695} -{"mode": "train", "epoch": 1, "iter": 17, "lr": 0.00011, "memory": 36808, "data_time": 0.05194, "loss_cls_0": 1.11107, "loss_box_0": 2.06984, "loss_cns_0": 0.62959, "loss_yns_0": 0.18129, "loss_cls_1": 1.192, "loss_box_1": 3.12929, "loss_cns_1": 0.53932, "loss_yns_1": 0.17217, "loss_cls_2": 1.20576, "loss_box_2": 3.0389, "loss_cns_2": 0.54106, "loss_yns_2": 0.17537, "loss_cls_3": 1.17346, "loss_box_3": 3.10338, "loss_cns_3": 0.53513, "loss_yns_3": 0.18281, "loss_cls_4": 1.22457, "loss_box_4": 3.11871, "loss_cns_4": 0.5467, "loss_yns_4": 0.17539, "loss_cls_5": 1.18371, "loss_box_5": 3.26633, "loss_cns_5": 0.52987, "loss_yns_5": 0.18063, "loss_cls_dn_0": 0.41946, "loss_box_dn_0": 1.04004, "loss_cls_dn_1": 0.42302, "loss_box_dn_1": 1.58691, "loss_cls_dn_2": 0.40638, "loss_box_dn_2": 1.54399, "loss_cls_dn_3": 0.38088, "loss_box_dn_3": 1.62244, "loss_cls_dn_4": 0.3671, "loss_box_dn_4": 1.6827, "loss_cls_dn_5": 0.40729, "loss_box_dn_5": 1.76506, "loss_dense_depth": 1.18848, "loss": 42.04012, "grad_norm": 63.70678, "time": 1.55805} -{"mode": "train", "epoch": 1, "iter": 18, "lr": 0.00011, "memory": 36808, "data_time": 0.06292, "loss_cls_0": 1.105, "loss_box_0": 2.08987, "loss_cns_0": 0.62166, "loss_yns_0": 0.18183, "loss_cls_1": 1.17233, "loss_box_1": 3.23796, "loss_cns_1": 0.53554, "loss_yns_1": 0.17864, "loss_cls_2": 1.20363, "loss_box_2": 3.25447, "loss_cns_2": 0.53282, "loss_yns_2": 0.18258, "loss_cls_3": 1.14954, "loss_box_3": 3.36688, "loss_cns_3": 0.51357, "loss_yns_3": 0.18311, "loss_cls_4": 1.1626, "loss_box_4": 3.28539, "loss_cns_4": 0.52233, "loss_yns_4": 0.17608, "loss_cls_5": 1.17259, "loss_box_5": 3.4332, "loss_cns_5": 0.4933, "loss_yns_5": 0.17839, "loss_cls_dn_0": 0.44714, "loss_box_dn_0": 1.02759, "loss_cls_dn_1": 0.40609, "loss_box_dn_1": 1.62316, "loss_cls_dn_2": 0.39772, "loss_box_dn_2": 1.62569, "loss_cls_dn_3": 0.3848, "loss_box_dn_3": 1.7225, "loss_cls_dn_4": 0.37618, "loss_box_dn_4": 1.72133, "loss_cls_dn_5": 0.39858, "loss_box_dn_5": 1.82309, "loss_dense_depth": 1.26948, "loss": 43.15667, "grad_norm": 84.90897, "time": 1.55421} -{"mode": "train", "epoch": 1, "iter": 19, "lr": 0.00011, "memory": 36808, "data_time": 0.04182, "loss_cls_0": 1.06863, "loss_box_0": 2.1131, "loss_cns_0": 0.6089, "loss_yns_0": 0.17771, "loss_cls_1": 1.14435, "loss_box_1": 3.2087, "loss_cns_1": 0.52739, "loss_yns_1": 0.17898, "loss_cls_2": 1.18758, "loss_box_2": 3.2886, "loss_cns_2": 0.52025, "loss_yns_2": 0.18477, "loss_cls_3": 1.13878, "loss_box_3": 3.35881, "loss_cns_3": 0.50291, "loss_yns_3": 0.17901, "loss_cls_4": 1.12859, "loss_box_4": 3.34124, "loss_cns_4": 0.50702, "loss_yns_4": 0.17356, "loss_cls_5": 1.15897, "loss_box_5": 3.46202, "loss_cns_5": 0.46718, "loss_yns_5": 0.188, "loss_cls_dn_0": 0.43526, "loss_box_dn_0": 1.01549, "loss_cls_dn_1": 0.41151, "loss_box_dn_1": 1.33428, "loss_cls_dn_2": 0.41866, "loss_box_dn_2": 1.42833, "loss_cls_dn_3": 0.41035, "loss_box_dn_3": 1.49838, "loss_cls_dn_4": 0.42122, "loss_box_dn_4": 1.50115, "loss_cls_dn_5": 0.4157, "loss_box_dn_5": 1.59456, "loss_dense_depth": 1.0988, "loss": 41.79873, "grad_norm": 103.09447, "time": 1.54518} -{"mode": "train", "epoch": 1, "iter": 20, "lr": 0.00011, "memory": 36808, "data_time": 0.06018, "loss_cls_0": 1.08015, "loss_box_0": 2.06973, "loss_cns_0": 0.6212, "loss_yns_0": 0.18186, "loss_cls_1": 1.17587, "loss_box_1": 2.96767, "loss_cns_1": 0.53049, "loss_yns_1": 0.18128, "loss_cls_2": 1.15926, "loss_box_2": 3.02263, "loss_cns_2": 0.52541, "loss_yns_2": 0.17611, "loss_cls_3": 1.13124, "loss_box_3": 3.01158, "loss_cns_3": 0.53364, "loss_yns_3": 0.18309, "loss_cls_4": 1.131, "loss_box_4": 3.12765, "loss_cns_4": 0.52673, "loss_yns_4": 0.16592, "loss_cls_5": 1.16042, "loss_box_5": 3.1942, "loss_cns_5": 0.51551, "loss_yns_5": 0.18672, "loss_cls_dn_0": 0.4418, "loss_box_dn_0": 1.00616, "loss_cls_dn_1": 0.37868, "loss_box_dn_1": 1.43942, "loss_cls_dn_2": 0.39295, "loss_box_dn_2": 1.49863, "loss_cls_dn_3": 0.38901, "loss_box_dn_3": 1.49737, "loss_cls_dn_4": 0.41457, "loss_box_dn_4": 1.52989, "loss_cls_dn_5": 0.38293, "loss_box_dn_5": 1.56593, "loss_dense_depth": 1.20524, "loss": 40.70197, "grad_norm": 74.74489, "time": 1.55596} -{"mode": "train", "epoch": 1, "iter": 21, "lr": 0.00011, "memory": 36808, "data_time": 0.0433, "loss_cls_0": 1.06766, "loss_box_0": 2.13533, "loss_cns_0": 0.60371, "loss_yns_0": 0.17554, "loss_cls_1": 1.19029, "loss_box_1": 2.85066, "loss_cns_1": 0.53915, "loss_yns_1": 0.18612, "loss_cls_2": 1.14591, "loss_box_2": 2.96801, "loss_cns_2": 0.55383, "loss_yns_2": 0.18153, "loss_cls_3": 1.12892, "loss_box_3": 2.92346, "loss_cns_3": 0.57531, "loss_yns_3": 0.17487, "loss_cls_4": 1.12547, "loss_box_4": 3.02941, "loss_cns_4": 0.57446, "loss_yns_4": 0.17897, "loss_cls_5": 1.16079, "loss_box_5": 3.06988, "loss_cns_5": 0.58175, "loss_yns_5": 0.17496, "loss_cls_dn_0": 0.4348, "loss_box_dn_0": 1.00407, "loss_cls_dn_1": 0.37729, "loss_box_dn_1": 1.24038, "loss_cls_dn_2": 0.40922, "loss_box_dn_2": 1.32518, "loss_cls_dn_3": 0.39508, "loss_box_dn_3": 1.36319, "loss_cls_dn_4": 0.41765, "loss_box_dn_4": 1.50701, "loss_cls_dn_5": 0.39022, "loss_box_dn_5": 1.56106, "loss_dense_depth": 1.15639, "loss": 39.87751, "grad_norm": 77.17654, "time": 1.53489} diff --git a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20260306_175131.log b/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20260306_175131.log deleted file mode 100644 index d59e7b0b428705c2fe85addb4ca1d924e755164f..0000000000000000000000000000000000000000 --- a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20260306_175131.log +++ /dev/null @@ -1,3250 +0,0 @@ -2026-03-06 17:51:31,532 - mmdet - INFO - Environment info: ------------------------------------------------------------- -sys.platform: linux -Python: 3.10.12 (main, Aug 15 2025, 14:32:43) [GCC 11.4.0] -CUDA available: True -GPU 0,1,2,3,4,5,6,7: BW200_LC, UBB BW1000_LC -CUDA_HOME: /opt/dtk -NVCC: Not Available -GCC: x86_64-linux-gnu-gcc (Ubuntu 11.4.0-1ubuntu1~22.04) 11.4.0 -PyTorch: 2.5.1 -PyTorch compiling details: PyTorch built with: - - GCC 10.3 - - C++ Version: 201703 - - Intel(R) Math Kernel Library Version 2020.0.4 Product Build 20200917 for Intel(R) 64 architecture applications - - OpenMP 201511 (a.k.a. OpenMP 4.5) - - LAPACK is enabled (usually provided by MKL) - - NNPACK is enabled - - CPU capability usage: AVX2 - - HIP Runtime 6.3.25521 - - MIOpen 2.18.0 - - Magma 2.8.0 - - Build settings: BLAS_INFO=mkl, BUILD_TYPE=Release, CXX_COMPILER=/opt/rh/gcc-toolset-10/root/usr/bin/c++, CXX_FLAGS= -D_GLIBCXX_USE_CXX11_ABI=1 -fvisibility-inlines-hidden -DUSE_PTHREADPOOL -DNDEBUG -DUSE_KINETO -DLIBKINETO_NOCUPTI -DLIBKINETO_NOXPUPTI=ON -DUSE_FBGEMM -DUSE_PYTORCH_QNNPACK -DUSE_XNNPACK -DSYMBOLICATE_MOBILE_DEBUG_HANDLE -O2 -fPIC -Wall -Wextra -Werror=return-type -Werror=non-virtual-dtor -Werror=bool-operation -Wnarrowing -Wno-missing-field-initializers -Wno-type-limits -Wno-array-bounds -Wno-unknown-pragmas -Wno-unused-parameter -Wno-strict-overflow -Wno-strict-aliasing -Wno-stringop-overflow -Wsuggest-override -Wno-psabi -Wno-error=old-style-cast -Wno-missing-braces -fdiagnostics-color=always -faligned-new -Wno-unused-but-set-variable -Wno-maybe-uninitialized -fno-math-errno -fno-trapping-math -Werror=format -Wno-stringop-overflow, FORCE_FALLBACK_CUDA_MPI=1, LAPACK_INFO=mkl, PERF_WITH_AVX=1, PERF_WITH_AVX2=1, TORCH_VERSION=2.5.1, USE_CUDA=0, USE_CUDNN=OFF, USE_CUSPARSELT=OFF, USE_EXCEPTION_PTR=1, USE_GFLAGS=1, USE_GLOG=1, USE_GLOO=1, USE_MKL=ON, USE_MKLDNN=0, USE_MPI=1, USE_NCCL=1, USE_NNPACK=ON, USE_OPENMP=1, USE_ROCM=ON, USE_ROCM_KERNEL_ASSERT=OFF, - -TorchVision: 0.20.1 -OpenCV: 4.12.0 -MMCV: 1.6.1 -MMCV Compiler: GCC 10.3 -MMCV CUDA Compiler: rocm not available -MMDetection: 2.25.1+ ------------------------------------------------------------- - -2026-03-06 17:51:32,492 - mmdet - INFO - Distributed training: True -2026-03-06 17:51:33,432 - mmdet - INFO - Config: -plugin = True -plugin_dir = 'projects/mmdet3d_plugin/' -dist_params = dict(backend='nccl') -log_level = 'INFO' -work_dir = './work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704' -total_batch_size = 120 -num_gpus = 8 -batch_size = 15 -num_iters_per_epoch = 234 -num_epochs = 100 -checkpoint_epoch_interval = 20 -checkpoint_config = dict(interval=4680) -log_config = dict( - interval=1, - hooks=[ - dict(type='TextLoggerHook', by_epoch=False), - dict(type='TensorboardLoggerHook') - ]) -load_from = None -resume_from = None -workflow = [('train', 1)] -fp16 = dict(loss_scale=32.0) -input_shape = (704, 256) -tracking_test = True -tracking_threshold = 0.2 -class_names = [ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier', - 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' -] -num_classes = 10 -embed_dims = 256 -num_groups = 8 -num_decoder = 6 -num_single_frame_decoder = 1 -use_deformable_func = True -strides = [4, 8, 16, 32] -num_levels = 4 -num_depth_layers = 3 -drop_out = 0.1 -temporal = True -decouple_attn = True -with_quality_estimation = True -model = dict( - type='Sparse4D', - use_grid_mask=True, - use_deformable_func=True, - img_backbone=dict( - type='ResNet', - depth=50, - num_stages=4, - frozen_stages=-1, - norm_eval=False, - style='pytorch', - with_cp=True, - out_indices=(0, 1, 2, 3), - norm_cfg=dict(type='BN', requires_grad=True), - pretrained='ckpt/resnet50-19c8e357.pth'), - img_neck=dict( - type='FPN', - num_outs=4, - start_level=0, - out_channels=256, - add_extra_convs='on_output', - relu_before_extra_convs=True, - in_channels=[256, 512, 1024, 2048]), - depth_branch=dict( - type='DenseDepthNet', - embed_dims=256, - num_depth_layers=3, - loss_weight=0.2), - head=dict( - type='Sparse4DHead', - cls_threshold_to_reg=0.05, - decouple_attn=True, - instance_bank=dict( - type='InstanceBank', - num_anchor=900, - embed_dims=256, - anchor='nuscenes_kmeans900.npy', - anchor_handler=dict(type='SparseBox3DKeyPointsGenerator'), - num_temp_instances=600, - confidence_decay=0.6, - feat_grad=False), - anchor_encoder=dict( - type='SparseBox3DEncoder', - vel_dims=3, - embed_dims=[128, 32, 32, 64], - mode='cat', - output_fc=False, - in_loops=1, - out_loops=4), - num_single_frame_decoder=1, - operation_order=[ - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine' - ], - temp_graph_model=dict( - type='MultiheadAttention', - embed_dims=512, - num_heads=8, - batch_first=True, - dropout=0.1), - graph_model=dict( - type='MultiheadAttention', - embed_dims=512, - num_heads=8, - batch_first=True, - dropout=0.1), - norm_layer=dict(type='LN', normalized_shape=256), - ffn=dict( - type='AsymmetricFFN', - in_channels=512, - pre_norm=dict(type='LN'), - embed_dims=256, - feedforward_channels=1024, - num_fcs=2, - ffn_drop=0.1, - act_cfg=dict(type='ReLU', inplace=True)), - deformable_model=dict( - type='DeformableFeatureAggregation', - embed_dims=256, - num_groups=8, - num_levels=4, - num_cams=6, - attn_drop=0.15, - use_deformable_func=True, - use_camera_embed=True, - residual_mode='cat', - kps_generator=dict( - type='SparseBox3DKeyPointsGenerator', - num_learnable_pts=6, - fix_scale=[[0, 0, 0], [0.45, 0, 0], [-0.45, 0, 0], - [0, 0.45, 0], [0, -0.45, 0], [0, 0, 0.45], - [0, 0, -0.45]])), - refine_layer=dict( - type='SparseBox3DRefinementModule', - embed_dims=256, - num_cls=10, - refine_yaw=True, - with_quality_estimation=True), - sampler=dict( - type='SparseBox3DTarget', - num_dn_groups=5, - num_temp_dn_groups=3, - dn_noise_scale=[2.0, 2.0, 2.0, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5], - max_dn_gt=32, - add_neg_dn=True, - cls_weight=2.0, - box_weight=0.25, - reg_weights=[2.0, 2.0, 2.0, 0.5, 0.5, 0.5, 0.0, 0.0, 0.0, 0.0], - cls_wise_reg_weights=dict( - {9: [2.0, 2.0, 2.0, 1.0, 1.0, 1.0, 0.0, 0.0, 1.0, 1.0]})), - loss_cls=dict( - type='FocalLoss', - use_sigmoid=True, - gamma=2.0, - alpha=0.25, - loss_weight=2.0), - loss_reg=dict( - type='SparseBox3DLoss', - loss_box=dict(type='L1Loss', loss_weight=0.25), - loss_centerness=dict(type='CrossEntropyLoss', use_sigmoid=True), - loss_yawness=dict(type='GaussianFocalLoss'), - cls_allow_reverse=[5]), - decoder=dict(type='SparseBox3DDecoder'), - reg_weights=[2.0, 2.0, 2.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0])) -dataset_type = 'NuScenes3DDetTrackDataset' -data_root = 'data/nuscenes/' -anno_root = 'data/nuscenes_anno_pkls/' -file_client_args = dict(backend='disk') -img_norm_cfg = dict( - mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375], to_rgb=True) -train_pipeline = [ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict( - type='LoadPointsFromFile', - coord_type='LIDAR', - load_dim=5, - use_dim=5, - file_client_args=dict(backend='disk')), - dict(type='ResizeCropFlipImage'), - dict(type='MultiScaleDepthMapGenerator', downsample=[4, 8, 16]), - dict(type='BBoxRotation'), - dict(type='PhotoMetricDistortionMultiViewImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict( - type='CircleObjectRangeFilter', - class_dist_thred=[55, 55, 55, 55, 55, 55, 55, 55, 55, 55]), - dict( - type='InstanceNameFilter', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ]), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=[ - 'img', 'timestamp', 'projection_mat', 'image_wh', 'gt_depth', - 'focal', 'gt_bboxes_3d', 'gt_labels_3d' - ], - meta_keys=['T_global', 'T_global_inv', 'timestamp', 'instance_id']) -] -test_pipeline = [ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='ResizeCropFlipImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=['img', 'timestamp', 'projection_mat', 'image_wh'], - meta_keys=['T_global', 'T_global_inv', 'timestamp']) -] -input_modality = dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False) -data_basic_config = dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier', - 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval') -data_aug_conf = dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]) -data = dict( - samples_per_gpu=15, - workers_per_gpu=15, - train=dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval', - ann_file='data/nuscenes_anno_pkls/nuscenes_infos_train.pkl', - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict( - type='LoadPointsFromFile', - coord_type='LIDAR', - load_dim=5, - use_dim=5, - file_client_args=dict(backend='disk')), - dict(type='ResizeCropFlipImage'), - dict(type='MultiScaleDepthMapGenerator', downsample=[4, 8, 16]), - dict(type='BBoxRotation'), - dict(type='PhotoMetricDistortionMultiViewImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict( - type='CircleObjectRangeFilter', - class_dist_thred=[55, 55, 55, 55, 55, 55, 55, 55, 55, 55]), - dict( - type='InstanceNameFilter', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', - 'traffic_cone' - ]), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=[ - 'img', 'timestamp', 'projection_mat', 'image_wh', - 'gt_depth', 'focal', 'gt_bboxes_3d', 'gt_labels_3d' - ], - meta_keys=[ - 'T_global', 'T_global_inv', 'timestamp', 'instance_id' - ]) - ], - test_mode=False, - data_aug_conf=dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]), - with_seq_flag=True, - sequences_split_num=2, - keep_consistent_seq_aug=True), - val=dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval', - ann_file='data/nuscenes_anno_pkls/nuscenes_infos_val.pkl', - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='ResizeCropFlipImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=['img', 'timestamp', 'projection_mat', 'image_wh'], - meta_keys=['T_global', 'T_global_inv', 'timestamp']) - ], - data_aug_conf=dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]), - test_mode=True, - tracking=True, - tracking_threshold=0.2), - test=dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval', - ann_file='data/nuscenes_anno_pkls/nuscenes_infos_val.pkl', - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='ResizeCropFlipImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=['img', 'timestamp', 'projection_mat', 'image_wh'], - meta_keys=['T_global', 'T_global_inv', 'timestamp']) - ], - data_aug_conf=dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]), - test_mode=True, - tracking=True, - tracking_threshold=0.2)) -optimizer = dict( - type='AdamW', - lr=0.0006, - weight_decay=0.001, - paramwise_cfg=dict(custom_keys=dict(img_backbone=dict(lr_mult=0.5)))) -optimizer_config = dict(grad_clip=dict(max_norm=25, norm_type=2)) -lr_config = dict( - policy='CosineAnnealing', - warmup='linear', - warmup_iters=500, - warmup_ratio=0.3333333333333333, - min_lr_ratio=0.001) -runner = dict(type='IterBasedRunner', max_iters=23400) -vis_pipeline = [ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='Collect', keys=['img'], meta_keys=['timestamp', 'lidar2img']) -] -evaluation = dict( - interval=4680, - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict( - type='Collect', keys=['img'], meta_keys=['timestamp', 'lidar2img']) - ]) -gpu_ids = range(0, 8) - -2026-03-06 17:51:33,432 - mmdet - INFO - Set random seed to 0, deterministic: False -2026-03-06 17:51:33,825 - mmdet - INFO - initialize ResNet with init_cfg {'type': 'Pretrained', 'checkpoint': 'ckpt/resnet50-19c8e357.pth'} -2026-03-06 17:51:34,349 - mmdet - INFO - initialize FPN with init_cfg {'type': 'Xavier', 'layer': 'Conv2d', 'distribution': 'uniform'} -Name of parameter - Initialization information - -img_backbone.conv1.weight - torch.Size([64, 3, 7, 7]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.bn1.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.bn1.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.conv1.weight - torch.Size([64, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn1.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn1.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.conv2.weight - torch.Size([64, 64, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn2.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn2.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.conv3.weight - torch.Size([256, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn3.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn3.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.downsample.0.weight - torch.Size([256, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.downsample.1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.downsample.1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.conv1.weight - torch.Size([64, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn1.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn1.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.conv2.weight - torch.Size([64, 64, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn2.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn2.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.conv3.weight - torch.Size([256, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn3.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn3.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.conv1.weight - torch.Size([64, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn1.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn1.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.conv2.weight - torch.Size([64, 64, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn2.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn2.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.conv3.weight - torch.Size([256, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn3.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn3.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.conv1.weight - torch.Size([128, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn1.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn1.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.conv2.weight - torch.Size([128, 128, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn2.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn2.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.conv3.weight - torch.Size([512, 128, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn3.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn3.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.downsample.0.weight - torch.Size([512, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.downsample.1.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.downsample.1.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.conv1.weight - torch.Size([128, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn1.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn1.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.conv2.weight - torch.Size([128, 128, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn2.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn2.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.conv3.weight - torch.Size([512, 128, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn3.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn3.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.conv1.weight - torch.Size([128, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn1.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn1.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.conv2.weight - torch.Size([128, 128, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn2.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn2.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.conv3.weight - torch.Size([512, 128, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn3.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn3.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.conv1.weight - torch.Size([128, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn1.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn1.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.conv2.weight - torch.Size([128, 128, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn2.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn2.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.conv3.weight - torch.Size([512, 128, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn3.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn3.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.conv1.weight - torch.Size([256, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.downsample.0.weight - torch.Size([1024, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.downsample.1.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.downsample.1.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.conv1.weight - torch.Size([512, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn1.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn1.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.conv2.weight - torch.Size([512, 512, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn2.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn2.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.conv3.weight - torch.Size([2048, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn3.weight - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn3.bias - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.downsample.0.weight - torch.Size([2048, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.downsample.1.weight - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.downsample.1.bias - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.conv1.weight - torch.Size([512, 2048, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn1.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn1.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.conv2.weight - torch.Size([512, 512, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn2.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn2.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.conv3.weight - torch.Size([2048, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn3.weight - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn3.bias - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.conv1.weight - torch.Size([512, 2048, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn1.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn1.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.conv2.weight - torch.Size([512, 512, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn2.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn2.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.conv3.weight - torch.Size([2048, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn3.weight - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn3.bias - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_neck.lateral_convs.0.conv.weight - torch.Size([256, 256, 1, 1]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.lateral_convs.0.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.lateral_convs.1.conv.weight - torch.Size([256, 512, 1, 1]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.lateral_convs.1.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.lateral_convs.2.conv.weight - torch.Size([256, 1024, 1, 1]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.lateral_convs.2.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.lateral_convs.3.conv.weight - torch.Size([256, 2048, 1, 1]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.lateral_convs.3.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.fpn_convs.0.conv.weight - torch.Size([256, 256, 3, 3]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.fpn_convs.0.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.fpn_convs.1.conv.weight - torch.Size([256, 256, 3, 3]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.fpn_convs.1.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.fpn_convs.2.conv.weight - torch.Size([256, 256, 3, 3]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.fpn_convs.2.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.fpn_convs.3.conv.weight - torch.Size([256, 256, 3, 3]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.fpn_convs.3.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.instance_bank.anchor - torch.Size([900, 11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.instance_bank.instance_feature - torch.Size([900, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.instance_bank.anchor_handler.fix_scale - torch.Size([1, 3]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.0.weight - torch.Size([128, 3]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.0.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.2.weight - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.2.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.3.weight - torch.Size([128, 128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.3.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.5.weight - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.5.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.6.weight - torch.Size([128, 128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.6.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.8.weight - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.8.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.9.weight - torch.Size([128, 128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.9.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.11.weight - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.11.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.0.weight - torch.Size([32, 3]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.0.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.2.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.2.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.3.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.3.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.5.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.5.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.6.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.6.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.8.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.8.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.9.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.9.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.11.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.11.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.0.weight - torch.Size([32, 2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.0.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.2.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.2.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.3.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.3.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.5.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.5.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.6.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.6.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.8.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.8.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.9.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.9.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.11.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.11.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.0.weight - torch.Size([64, 3]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.0.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.2.weight - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.2.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.3.weight - torch.Size([64, 64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.3.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.5.weight - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.5.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.6.weight - torch.Size([64, 64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.6.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.8.weight - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.8.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.9.weight - torch.Size([64, 64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.9.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.11.weight - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.11.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.1.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.1.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.1.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.1.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.1.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.1.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.1.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.1.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.3.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.4.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.4.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.4.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.4.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.5.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.5.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.5.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.5.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.6.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.6.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.8.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.8.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.8.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.8.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.8.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.8.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.8.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.8.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.10.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.11.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.11.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.11.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.11.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.12.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.12.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.12.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.12.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.13.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.13.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.15.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.15.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.15.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.15.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.15.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.15.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.15.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.15.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.16.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.16.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.17.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.18.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.18.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.18.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.18.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.19.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.19.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.19.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.19.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.20.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.20.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.22.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.22.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.22.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.22.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.22.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.22.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.22.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.22.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.23.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.23.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.24.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.25.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.25.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.25.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.25.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.26.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.26.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.26.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.26.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.27.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.27.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.29.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.29.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.29.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.29.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.29.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.29.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.29.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.29.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.30.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.30.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.31.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.32.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.32.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.32.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.32.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.33.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.33.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.33.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.33.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.34.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.34.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.36.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.36.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.36.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.36.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.36.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.36.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.36.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.36.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.37.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.37.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.38.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.fc_before.weight - torch.Size([512, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.fc_after.weight - torch.Size([256, 512]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.0.weight - torch.Size([1, 256, 1, 1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.0.bias - torch.Size([1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.1.weight - torch.Size([1, 256, 1, 1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.1.bias - torch.Size([1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.2.weight - torch.Size([1, 256, 1, 1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.2.bias - torch.Size([1]): -The value is the same before and after calling `init_weights` of Sparse4D -2026-03-06 17:51:34,479 - mmdet - INFO - Model: -Sparse4D( - (img_backbone): ResNet( - (conv1): Conv2d(3, 64, kernel_size=(7, 7), stride=(2, 2), padding=(3, 3), bias=False) - (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (maxpool): MaxPool2d(kernel_size=3, stride=2, padding=1, dilation=1, ceil_mode=False) - (layer1): ResLayer( - (0): Bottleneck( - (conv1): Conv2d(64, 64, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (downsample): Sequential( - (0): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - ) - ) - (1): Bottleneck( - (conv1): Conv2d(256, 64, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (2): Bottleneck( - (conv1): Conv2d(256, 64, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - ) - (layer2): ResLayer( - (0): Bottleneck( - (conv1): Conv2d(256, 128, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (downsample): Sequential( - (0): Conv2d(256, 512, kernel_size=(1, 1), stride=(2, 2), bias=False) - (1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - ) - ) - (1): Bottleneck( - (conv1): Conv2d(512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (2): Bottleneck( - (conv1): Conv2d(512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (3): Bottleneck( - (conv1): Conv2d(512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - ) - (layer3): ResLayer( - (0): Bottleneck( - (conv1): Conv2d(512, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (downsample): Sequential( - (0): Conv2d(512, 1024, kernel_size=(1, 1), stride=(2, 2), bias=False) - (1): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - ) - ) - (1): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (2): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (3): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (4): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (5): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - ) - (layer4): ResLayer( - (0): Bottleneck( - (conv1): Conv2d(1024, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(512, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (downsample): Sequential( - (0): Conv2d(1024, 2048, kernel_size=(1, 1), stride=(2, 2), bias=False) - (1): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - ) - ) - (1): Bottleneck( - (conv1): Conv2d(2048, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(512, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (2): Bottleneck( - (conv1): Conv2d(2048, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(512, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - ) - ) - init_cfg={'type': 'Pretrained', 'checkpoint': 'ckpt/resnet50-19c8e357.pth'} - (img_neck): FPN( - (lateral_convs): ModuleList( - (0): ConvModule( - (conv): Conv2d(256, 256, kernel_size=(1, 1), stride=(1, 1)) - ) - (1): ConvModule( - (conv): Conv2d(512, 256, kernel_size=(1, 1), stride=(1, 1)) - ) - (2): ConvModule( - (conv): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1)) - ) - (3): ConvModule( - (conv): Conv2d(2048, 256, kernel_size=(1, 1), stride=(1, 1)) - ) - ) - (fpn_convs): ModuleList( - (0-3): 4 x ConvModule( - (conv): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1)) - ) - ) - ) - init_cfg={'type': 'Xavier', 'layer': 'Conv2d', 'distribution': 'uniform'} - (head): Sparse4DHead( - (instance_bank): InstanceBank( - (anchor_handler): SparseBox3DKeyPointsGenerator() - ) - (anchor_encoder): SparseBox3DEncoder( - (pos_fc): Sequential( - (0): Linear(in_features=3, out_features=128, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((128,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=128, out_features=128, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((128,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=128, out_features=128, bias=True) - (7): ReLU(inplace=True) - (8): LayerNorm((128,), eps=1e-05, elementwise_affine=True) - (9): Linear(in_features=128, out_features=128, bias=True) - (10): ReLU(inplace=True) - (11): LayerNorm((128,), eps=1e-05, elementwise_affine=True) - ) - (size_fc): Sequential( - (0): Linear(in_features=3, out_features=32, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=32, out_features=32, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=32, out_features=32, bias=True) - (7): ReLU(inplace=True) - (8): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (9): Linear(in_features=32, out_features=32, bias=True) - (10): ReLU(inplace=True) - (11): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - ) - (yaw_fc): Sequential( - (0): Linear(in_features=2, out_features=32, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=32, out_features=32, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=32, out_features=32, bias=True) - (7): ReLU(inplace=True) - (8): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (9): Linear(in_features=32, out_features=32, bias=True) - (10): ReLU(inplace=True) - (11): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - ) - (vel_fc): Sequential( - (0): Linear(in_features=3, out_features=64, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((64,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=64, out_features=64, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((64,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=64, out_features=64, bias=True) - (7): ReLU(inplace=True) - (8): LayerNorm((64,), eps=1e-05, elementwise_affine=True) - (9): Linear(in_features=64, out_features=64, bias=True) - (10): ReLU(inplace=True) - (11): LayerNorm((64,), eps=1e-05, elementwise_affine=True) - ) - ) - (loss_cls): FocalLoss() - (loss_reg): SparseBox3DLoss( - (loss_box): L1Loss() - (loss_cns): CrossEntropyLoss(avg_non_ignore=False) - (loss_yns): GaussianFocalLoss() - ) - (layers): ModuleList( - (0): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (1): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (4-5): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (6): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (7): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (8): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (11-12): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (13): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (14): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (15): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (16): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (17): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (18-19): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (20): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (21): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (22): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (23): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (24): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (25-26): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (27): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (28): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (29): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (30): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (31): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (32-33): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (34): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (35): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (36): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (37): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (38): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - ) - (fc_before): Linear(in_features=256, out_features=512, bias=False) - (fc_after): Linear(in_features=512, out_features=256, bias=False) - ) - (depth_branch): DenseDepthNet( - (depth_layers): ModuleList( - (0-2): 3 x Conv2d(256, 1, kernel_size=(1, 1), stride=(1, 1)) - ) - ) - (grid_mask): GridMask() -) -2026-03-06 17:51:49,773 - mmdet - INFO - Start running, host: root@bw28, work_dir: /workspace/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704 -2026-03-06 17:51:49,773 - mmdet - INFO - Hooks will be executed in the following order: -before_run: -(VERY_HIGH ) CosineAnnealingLrUpdaterHook -(ABOVE_NORMAL) Fp16OptimizerHook -(NORMAL ) CheckpointHook -(NORMAL ) CustomDistEvalHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -before_train_epoch: -(VERY_HIGH ) CosineAnnealingLrUpdaterHook -(NORMAL ) CustomDistEvalHook -(LOW ) IterTimerHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -before_train_iter: -(VERY_HIGH ) CosineAnnealingLrUpdaterHook -(NORMAL ) CustomDistEvalHook -(LOW ) IterTimerHook - -------------------- -after_train_iter: -(ABOVE_NORMAL) Fp16OptimizerHook -(NORMAL ) CheckpointHook -(NORMAL ) CustomDistEvalHook -(NORMAL ) ProfilerHook -(LOW ) IterTimerHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -after_train_epoch: -(NORMAL ) CheckpointHook -(NORMAL ) CustomDistEvalHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -before_val_epoch: -(LOW ) IterTimerHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -before_val_iter: -(LOW ) IterTimerHook - -------------------- -after_val_iter: -(LOW ) IterTimerHook - -------------------- -after_val_epoch: -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -after_run: -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -2026-03-06 17:51:49,773 - mmdet - INFO - workflow: [('train', 1)], max: 23400 iters -2026-03-06 17:51:49,776 - mmdet - INFO - Checkpoints will be saved to /workspace/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704 by HardDiskBackend. -2026-03-06 17:53:19,291 - mmdet - INFO - Iter [1/23400] lr: 1.000e-04, eta: 23 days, 23:40:39, time: 88.570, data_time: 7.300, memory: 36993, loss_cls_0: 2.3501, loss_box_0: 0.0070, loss_cns_0: 0.0014, loss_yns_0: 0.0006, loss_cls_1: 2.1458, loss_box_1: 0.0911, loss_cns_1: 0.0204, loss_yns_1: 0.0067, loss_cls_2: 2.3169, loss_box_2: 0.0049, loss_cns_2: 0.0006, loss_yns_2: 0.0004, loss_cls_3: 2.4079, loss_box_3: 0.0248, loss_cns_3: 0.0043, loss_yns_3: 0.0014, loss_cls_4: 2.0283, loss_box_4: 0.4383, loss_cns_4: 0.0573, loss_yns_4: 0.0275, loss_cls_5: 2.4272, loss_box_5: 0.0147, loss_cns_5: 0.0023, loss_yns_5: 0.0007, loss_cls_dn_0: 1.1907, loss_box_dn_0: 1.4621, loss_cls_dn_1: 1.1082, loss_box_dn_1: 1.7343, loss_cls_dn_2: 1.1703, loss_box_dn_2: 1.9761, loss_cls_dn_3: 1.1697, loss_box_dn_3: 2.2542, loss_cls_dn_4: 1.0559, loss_box_dn_4: 2.4402, loss_cls_dn_5: 1.2400, loss_box_dn_5: 2.6845, loss_dense_depth: 1.8372, loss: 35.7041, grad_norm: 276.2693 -2026-03-06 17:53:21,989 - mmdet - INFO - Iter [2/23400] lr: 1.004e-04, eta: 12 days, 8:29:33, time: 2.667, data_time: 0.096, memory: 36993, loss_cls_0: 2.0518, loss_box_0: 0.0141, loss_cns_0: 0.0039, loss_yns_0: 0.0014, loss_cls_1: 2.0030, loss_box_1: 0.2895, loss_cns_1: 0.0493, loss_yns_1: 0.0148, loss_cls_2: 2.0937, loss_box_2: 0.2782, loss_cns_2: 0.0279, loss_yns_2: 0.0109, loss_cls_3: 1.9677, loss_box_3: 0.3453, loss_cns_3: 0.0428, loss_yns_3: 0.0140, loss_cls_4: 1.8075, loss_box_4: 1.5462, loss_cns_4: 0.1502, loss_yns_4: 0.0539, loss_cls_5: 2.0562, loss_box_5: 0.5149, loss_cns_5: 0.0549, loss_yns_5: 0.0202, loss_cls_dn_0: 1.0345, loss_box_dn_0: 1.2420, loss_cls_dn_1: 0.9534, loss_box_dn_1: 2.4380, loss_cls_dn_2: 0.9654, loss_box_dn_2: 2.5596, loss_cls_dn_3: 0.9202, loss_box_dn_3: 2.6327, loss_cls_dn_4: 0.8406, loss_box_dn_4: 2.9066, loss_cls_dn_5: 0.9848, loss_box_dn_5: 3.1464, loss_dense_depth: 1.6948, loss: 37.7313, grad_norm: 66.7134 -2026-03-06 17:53:26,911 - mmdet - INFO - Iter [3/23400] lr: 1.008e-04, eta: 8 days, 16:24:47, time: 4.967, data_time: 0.090, memory: 36993, loss_cls_0: 1.4552, loss_box_0: 2.5756, loss_cns_0: 0.6477, loss_yns_0: 0.2164, loss_cls_1: 1.7910, loss_box_1: 1.7107, loss_cns_1: 0.2709, loss_yns_1: 0.0976, loss_cls_2: 1.7983, loss_box_2: 3.8143, loss_cns_2: 0.3408, loss_yns_2: 0.1900, loss_cls_3: 1.6378, loss_box_3: 4.7860, loss_cns_3: 0.4351, loss_yns_3: 0.2091, loss_cls_4: 1.5831, loss_box_4: 4.0586, loss_cns_4: 0.3678, loss_yns_4: 0.1577, loss_cls_5: 1.7110, loss_box_5: 2.5662, loss_cns_5: 0.1874, loss_yns_5: 0.0908, loss_cls_dn_0: 0.7040, loss_box_dn_0: 1.1772, loss_cls_dn_1: 0.8340, loss_box_dn_1: 2.4160, loss_cls_dn_2: 0.7988, loss_box_dn_2: 2.6092, loss_cls_dn_3: 0.7174, loss_box_dn_3: 2.8059, loss_cls_dn_4: 0.7205, loss_box_dn_4: 3.0356, loss_cls_dn_5: 0.8070, loss_box_dn_5: 3.2676, loss_dense_depth: 1.6433, loss: 54.2354, grad_norm: 103.5416 -2026-03-06 17:53:30,547 - mmdet - INFO - Iter [4/23400] lr: 1.012e-04, eta: 6 days, 18:09:40, time: 3.605, data_time: 0.071, memory: 36993, loss_cls_0: 1.3558, loss_box_0: 2.5524, loss_cns_0: 0.5636, loss_yns_0: 0.1828, loss_cls_1: 1.6188, loss_box_1: 3.0864, loss_cns_1: 0.4484, loss_yns_1: 0.1820, loss_cls_2: 1.7087, loss_box_2: 3.5576, loss_cns_2: 0.4572, loss_yns_2: 0.1870, loss_cls_3: 1.5213, loss_box_3: 4.0408, loss_cns_3: 0.4747, loss_yns_3: 0.2167, loss_cls_4: 1.4818, loss_box_4: 4.4395, loss_cns_4: 0.4026, loss_yns_4: 0.1951, loss_cls_5: 1.5067, loss_box_5: 4.6454, loss_cns_5: 0.4464, loss_yns_5: 0.2026, loss_cls_dn_0: 0.5831, loss_box_dn_0: 1.1581, loss_cls_dn_1: 0.7331, loss_box_dn_1: 2.5101, loss_cls_dn_2: 0.6907, loss_box_dn_2: 2.5535, loss_cls_dn_3: 0.6220, loss_box_dn_3: 2.7232, loss_cls_dn_4: 0.5987, loss_box_dn_4: 2.8833, loss_cls_dn_5: 0.6724, loss_box_dn_5: 3.0564, loss_dense_depth: 1.5817, loss: 55.8406, grad_norm: 110.8958 -2026-03-06 17:53:35,355 - mmdet - INFO - Iter [5/23400] lr: 1.016e-04, eta: 5 days, 16:00:47, time: 4.839, data_time: 0.103, memory: 36993, loss_cls_0: 1.3209, loss_box_0: 2.7044, loss_cns_0: 0.4910, loss_yns_0: 0.1971, loss_cls_1: 1.5553, loss_box_1: 3.7737, loss_cns_1: 0.3968, loss_yns_1: 0.2038, loss_cls_2: 1.5970, loss_box_2: 3.9253, loss_cns_2: 0.3958, loss_yns_2: 0.1972, loss_cls_3: 1.4460, loss_box_3: 4.0295, loss_cns_3: 0.3940, loss_yns_3: 0.2007, loss_cls_4: 1.3924, loss_box_4: 4.2760, loss_cns_4: 0.3695, loss_yns_4: 0.1954, loss_cls_5: 1.4006, loss_box_5: 4.4638, loss_cns_5: 0.4142, loss_yns_5: 0.2065, loss_cls_dn_0: 0.5485, loss_box_dn_0: 1.1991, loss_cls_dn_1: 0.6563, loss_box_dn_1: 2.1757, loss_cls_dn_2: 0.6484, loss_box_dn_2: 2.2879, loss_cls_dn_3: 0.5673, loss_box_dn_3: 2.3737, loss_cls_dn_4: 0.5570, loss_box_dn_4: 2.5422, loss_cls_dn_5: 0.5852, loss_box_dn_5: 2.6320, loss_dense_depth: 1.4978, loss: 53.8180, grad_norm: 112.2390 -2026-03-06 17:53:39,053 - mmdet - INFO - Iter [6/23400] lr: 1.020e-04, eta: 4 days, 21:20:33, time: 3.696, data_time: 0.046, memory: 36993, loss_cls_0: 1.3009, loss_box_0: 2.5771, loss_cns_0: 0.5798, loss_yns_0: 0.1842, loss_cls_1: 1.4824, loss_box_1: 3.8672, loss_cns_1: 0.3639, loss_yns_1: 0.1892, loss_cls_2: 1.4823, loss_box_2: 4.0520, loss_cns_2: 0.3625, loss_yns_2: 0.2030, loss_cls_3: 1.3731, loss_box_3: 4.0229, loss_cns_3: 0.3470, loss_yns_3: 0.2009, loss_cls_4: 1.3305, loss_box_4: 4.2834, loss_cns_4: 0.3021, loss_yns_4: 0.1964, loss_cls_5: 1.3357, loss_box_5: 4.4388, loss_cns_5: 0.2996, loss_yns_5: 0.2017, loss_cls_dn_0: 0.5157, loss_box_dn_0: 1.1711, loss_cls_dn_1: 0.5988, loss_box_dn_1: 2.3723, loss_cls_dn_2: 0.5955, loss_box_dn_2: 2.4209, loss_cls_dn_3: 0.5215, loss_box_dn_3: 2.4822, loss_cls_dn_4: 0.4932, loss_box_dn_4: 2.7006, loss_cls_dn_5: 0.5043, loss_box_dn_5: 2.7710, loss_dense_depth: 1.4442, loss: 53.5679, grad_norm: 123.6024 -2026-03-06 17:53:43,984 - mmdet - INFO - Iter [7/23400] lr: 1.024e-04, eta: 4 days, 9:09:12, time: 4.932, data_time: 0.063, memory: 36993, loss_cls_0: 1.2905, loss_box_0: 2.3124, loss_cns_0: 0.6932, loss_yns_0: 0.1770, loss_cls_1: 1.3724, loss_box_1: 3.4704, loss_cns_1: 0.5040, loss_yns_1: 0.1932, loss_cls_2: 1.3964, loss_box_2: 3.5526, loss_cns_2: 0.4785, loss_yns_2: 0.1961, loss_cls_3: 1.3069, loss_box_3: 3.5156, loss_cns_3: 0.5136, loss_yns_3: 0.1880, loss_cls_4: 1.3111, loss_box_4: 3.7983, loss_cns_4: 0.5016, loss_yns_4: 0.1891, loss_cls_5: 1.3382, loss_box_5: 3.9468, loss_cns_5: 0.4826, loss_yns_5: 0.2040, loss_cls_dn_0: 0.4953, loss_box_dn_0: 1.0884, loss_cls_dn_1: 0.5303, loss_box_dn_1: 2.4360, loss_cls_dn_2: 0.5351, loss_box_dn_2: 2.4404, loss_cls_dn_3: 0.4764, loss_box_dn_3: 2.5166, loss_cls_dn_4: 0.4418, loss_box_dn_4: 2.7286, loss_cls_dn_5: 0.4375, loss_box_dn_5: 2.7957, loss_dense_depth: 1.4576, loss: 51.3121, grad_norm: 108.6199 -2026-03-06 17:53:47,448 - mmdet - INFO - Iter [8/23400] lr: 1.028e-04, eta: 3 days, 22:49:10, time: 3.465, data_time: 0.047, memory: 36993, loss_cls_0: 1.2566, loss_box_0: 2.2200, loss_cns_0: 0.6819, loss_yns_0: 0.1788, loss_cls_1: 1.3185, loss_box_1: 3.3390, loss_cns_1: 0.5766, loss_yns_1: 0.1916, loss_cls_2: 1.3863, loss_box_2: 3.5019, loss_cns_2: 0.5113, loss_yns_2: 0.1930, loss_cls_3: 1.3063, loss_box_3: 3.6690, loss_cns_3: 0.4848, loss_yns_3: 0.1887, loss_cls_4: 1.3168, loss_box_4: 3.7307, loss_cns_4: 0.4777, loss_yns_4: 0.1906, loss_cls_5: 1.3327, loss_box_5: 3.7895, loss_cns_5: 0.4941, loss_yns_5: 0.1936, loss_cls_dn_0: 0.5079, loss_box_dn_0: 1.0366, loss_cls_dn_1: 0.5374, loss_box_dn_1: 1.8217, loss_cls_dn_2: 0.5486, loss_box_dn_2: 1.7938, loss_cls_dn_3: 0.4928, loss_box_dn_3: 1.9287, loss_cls_dn_4: 0.4690, loss_box_dn_4: 2.0441, loss_cls_dn_5: 0.4404, loss_box_dn_5: 2.0929, loss_dense_depth: 1.3430, loss: 47.5868, grad_norm: 95.4680 -2026-03-06 17:53:51,364 - mmdet - INFO - Iter [9/23400] lr: 1.032e-04, eta: 3 days, 15:06:25, time: 3.916, data_time: 0.046, memory: 36993, loss_cls_0: 1.2268, loss_box_0: 2.2247, loss_cns_0: 0.6352, loss_yns_0: 0.1757, loss_cls_1: 1.2579, loss_box_1: 3.1179, loss_cns_1: 0.5672, loss_yns_1: 0.1858, loss_cls_2: 1.3849, loss_box_2: 3.2863, loss_cns_2: 0.4994, loss_yns_2: 0.1957, loss_cls_3: 1.2748, loss_box_3: 3.4731, loss_cns_3: 0.4765, loss_yns_3: 0.1905, loss_cls_4: 1.2670, loss_box_4: 3.4369, loss_cns_4: 0.5018, loss_yns_4: 0.1891, loss_cls_5: 1.3111, loss_box_5: 3.5117, loss_cns_5: 0.5434, loss_yns_5: 0.1856, loss_cls_dn_0: 0.5163, loss_box_dn_0: 1.0183, loss_cls_dn_1: 0.5057, loss_box_dn_1: 1.4971, loss_cls_dn_2: 0.5297, loss_box_dn_2: 1.5366, loss_cls_dn_3: 0.4639, loss_box_dn_3: 1.6609, loss_cls_dn_4: 0.4628, loss_box_dn_4: 1.6551, loss_cls_dn_5: 0.4326, loss_box_dn_5: 1.7250, loss_dense_depth: 1.4418, loss: 44.5648, grad_norm: 89.8070 -2026-03-06 17:53:56,321 - mmdet - INFO - Iter [10/23400] lr: 1.036e-04, eta: 3 days, 9:36:47, time: 4.956, data_time: 0.048, memory: 36993, loss_cls_0: 1.2089, loss_box_0: 2.2573, loss_cns_0: 0.6094, loss_yns_0: 0.1726, loss_cls_1: 1.2502, loss_box_1: 2.9687, loss_cns_1: 0.5890, loss_yns_1: 0.1837, loss_cls_2: 1.3109, loss_box_2: 3.0309, loss_cns_2: 0.5715, loss_yns_2: 0.1896, loss_cls_3: 1.2717, loss_box_3: 3.1341, loss_cns_3: 0.6021, loss_yns_3: 0.1981, loss_cls_4: 1.2592, loss_box_4: 3.1735, loss_cns_4: 0.5919, loss_yns_4: 0.1850, loss_cls_5: 1.3187, loss_box_5: 3.2757, loss_cns_5: 0.5953, loss_yns_5: 0.1857, loss_cls_dn_0: 0.4891, loss_box_dn_0: 1.0499, loss_cls_dn_1: 0.4605, loss_box_dn_1: 1.5424, loss_cls_dn_2: 0.4873, loss_box_dn_2: 1.6071, loss_cls_dn_3: 0.4321, loss_box_dn_3: 1.7023, loss_cls_dn_4: 0.4386, loss_box_dn_4: 1.7476, loss_cls_dn_5: 0.4191, loss_box_dn_5: 1.8538, loss_dense_depth: 1.2643, loss: 43.6277, grad_norm: 66.6497 -2026-03-06 17:53:59,972 - mmdet - INFO - Iter [11/23400] lr: 1.040e-04, eta: 3 days, 4:19:40, time: 3.619, data_time: 0.047, memory: 36993, loss_cls_0: 1.2454, loss_box_0: 2.2428, loss_cns_0: 0.6200, loss_yns_0: 0.1728, loss_cls_1: 1.3111, loss_box_1: 2.9068, loss_cns_1: 0.5511, loss_yns_1: 0.1824, loss_cls_2: 1.2897, loss_box_2: 2.9376, loss_cns_2: 0.5472, loss_yns_2: 0.1914, loss_cls_3: 1.2658, loss_box_3: 2.9864, loss_cns_3: 0.5809, loss_yns_3: 0.1848, loss_cls_4: 1.2599, loss_box_4: 3.1661, loss_cns_4: 0.5231, loss_yns_4: 0.1852, loss_cls_5: 1.3160, loss_box_5: 3.3889, loss_cns_5: 0.4610, loss_yns_5: 0.1889, loss_cls_dn_0: 0.4612, loss_box_dn_0: 1.0332, loss_cls_dn_1: 0.4182, loss_box_dn_1: 1.7094, loss_cls_dn_2: 0.4454, loss_box_dn_2: 1.7884, loss_cls_dn_3: 0.4163, loss_box_dn_3: 1.8462, loss_cls_dn_4: 0.4060, loss_box_dn_4: 1.9590, loss_cls_dn_5: 0.4026, loss_box_dn_5: 2.1543, loss_dense_depth: 1.3472, loss: 44.0925, grad_norm: 83.8529 -2026-03-06 17:54:03,692 - mmdet - INFO - Iter [12/23400] lr: 1.044e-04, eta: 2 days, 23:59:47, time: 3.754, data_time: 0.090, memory: 36993, loss_cls_0: 1.2486, loss_box_0: 2.2226, loss_cns_0: 0.6390, loss_yns_0: 0.1716, loss_cls_1: 1.3011, loss_box_1: 3.0729, loss_cns_1: 0.4878, loss_yns_1: 0.1796, loss_cls_2: 1.3037, loss_box_2: 3.0265, loss_cns_2: 0.4931, loss_yns_2: 0.1875, loss_cls_3: 1.2465, loss_box_3: 3.0428, loss_cns_3: 0.5317, loss_yns_3: 0.1772, loss_cls_4: 1.2587, loss_box_4: 3.1651, loss_cns_4: 0.5051, loss_yns_4: 0.1812, loss_cls_5: 1.2904, loss_box_5: 3.6010, loss_cns_5: 0.4505, loss_yns_5: 0.1850, loss_cls_dn_0: 0.4649, loss_box_dn_0: 1.0345, loss_cls_dn_1: 0.3891, loss_box_dn_1: 2.0179, loss_cls_dn_2: 0.4213, loss_box_dn_2: 2.0477, loss_cls_dn_3: 0.3987, loss_box_dn_3: 2.0681, loss_cls_dn_4: 0.3903, loss_box_dn_4: 2.1731, loss_cls_dn_5: 0.3881, loss_box_dn_5: 2.3925, loss_dense_depth: 1.2304, loss: 45.3858, grad_norm: 90.0176 -2026-03-06 17:54:08,006 - mmdet - INFO - Iter [13/23400] lr: 1.048e-04, eta: 2 days, 20:35:46, time: 4.284, data_time: 0.053, memory: 36993, loss_cls_0: 1.2138, loss_box_0: 2.1968, loss_cns_0: 0.6368, loss_yns_0: 0.1708, loss_cls_1: 1.2571, loss_box_1: 3.0973, loss_cns_1: 0.4959, loss_yns_1: 0.1719, loss_cls_2: 1.2879, loss_box_2: 3.1174, loss_cns_2: 0.5025, loss_yns_2: 0.1828, loss_cls_3: 1.2444, loss_box_3: 3.0776, loss_cns_3: 0.5105, loss_yns_3: 0.1779, loss_cls_4: 1.2521, loss_box_4: 3.0777, loss_cns_4: 0.5049, loss_yns_4: 0.1779, loss_cls_5: 1.2682, loss_box_5: 3.2432, loss_cns_5: 0.5151, loss_yns_5: 0.1755, loss_cls_dn_0: 0.4774, loss_box_dn_0: 1.0098, loss_cls_dn_1: 0.4303, loss_box_dn_1: 1.7363, loss_cls_dn_2: 0.4777, loss_box_dn_2: 1.7302, loss_cls_dn_3: 0.4435, loss_box_dn_3: 1.7306, loss_cls_dn_4: 0.4446, loss_box_dn_4: 1.7899, loss_cls_dn_5: 0.4479, loss_box_dn_5: 1.9168, loss_dense_depth: 1.3024, loss: 43.4935, grad_norm: 64.3034 -2026-03-06 17:54:11,676 - mmdet - INFO - Iter [14/23400] lr: 1.052e-04, eta: 2 days, 17:24:37, time: 3.700, data_time: 0.075, memory: 36993, loss_cls_0: 1.1973, loss_box_0: 2.1861, loss_cns_0: 0.6269, loss_yns_0: 0.1715, loss_cls_1: 1.2750, loss_box_1: 2.9968, loss_cns_1: 0.5318, loss_yns_1: 0.1725, loss_cls_2: 1.2748, loss_box_2: 3.1696, loss_cns_2: 0.5172, loss_yns_2: 0.1775, loss_cls_3: 1.2745, loss_box_3: 3.1306, loss_cns_3: 0.5097, loss_yns_3: 0.1747, loss_cls_4: 1.2911, loss_box_4: 3.1579, loss_cns_4: 0.4943, loss_yns_4: 0.1773, loss_cls_5: 1.2851, loss_box_5: 3.1865, loss_cns_5: 0.4928, loss_yns_5: 0.1775, loss_cls_dn_0: 0.4812, loss_box_dn_0: 1.0091, loss_cls_dn_1: 0.4538, loss_box_dn_1: 1.5297, loss_cls_dn_2: 0.4971, loss_box_dn_2: 1.5410, loss_cls_dn_3: 0.4550, loss_box_dn_3: 1.5535, loss_cls_dn_4: 0.4684, loss_box_dn_4: 1.5761, loss_cls_dn_5: 0.4810, loss_box_dn_5: 1.5847, loss_dense_depth: 1.2152, loss: 42.4951, grad_norm: 69.1298 -2026-03-06 17:54:16,943 - mmdet - INFO - Iter [15/23400] lr: 1.056e-04, eta: 2 days, 15:19:36, time: 5.264, data_time: 0.047, memory: 36993, loss_cls_0: 1.2067, loss_box_0: 2.2590, loss_cns_0: 0.6107, loss_yns_0: 0.1725, loss_cls_1: 1.3232, loss_box_1: 2.7266, loss_cns_1: 0.5604, loss_yns_1: 0.1728, loss_cls_2: 1.2950, loss_box_2: 2.8880, loss_cns_2: 0.5613, loss_yns_2: 0.1780, loss_cls_3: 1.3165, loss_box_3: 2.9347, loss_cns_3: 0.5431, loss_yns_3: 0.1775, loss_cls_4: 1.3023, loss_box_4: 2.9944, loss_cns_4: 0.5328, loss_yns_4: 0.1752, loss_cls_5: 1.2974, loss_box_5: 3.0974, loss_cns_5: 0.5271, loss_yns_5: 0.1766, loss_cls_dn_0: 0.4902, loss_box_dn_0: 1.0016, loss_cls_dn_1: 0.4625, loss_box_dn_1: 1.5031, loss_cls_dn_2: 0.4882, loss_box_dn_2: 1.5330, loss_cls_dn_3: 0.4503, loss_box_dn_3: 1.6187, loss_cls_dn_4: 0.4639, loss_box_dn_4: 1.6568, loss_cls_dn_5: 0.4902, loss_box_dn_5: 1.7026, loss_dense_depth: 1.2156, loss: 42.1060, grad_norm: 96.4565 -2026-03-06 17:54:20,803 - mmdet - INFO - Iter [16/23400] lr: 1.060e-04, eta: 2 days, 12:55:23, time: 3.835, data_time: 0.096, memory: 36993, loss_cls_0: 1.2014, loss_box_0: 2.2444, loss_cns_0: 0.6015, loss_yns_0: 0.1744, loss_cls_1: 1.2913, loss_box_1: 2.8704, loss_cns_1: 0.5362, loss_yns_1: 0.1723, loss_cls_2: 1.2846, loss_box_2: 2.9132, loss_cns_2: 0.5357, loss_yns_2: 0.1784, loss_cls_3: 1.3031, loss_box_3: 2.9563, loss_cns_3: 0.5309, loss_yns_3: 0.1793, loss_cls_4: 1.2673, loss_box_4: 2.9715, loss_cns_4: 0.5338, loss_yns_4: 0.1722, loss_cls_5: 1.2743, loss_box_5: 3.0219, loss_cns_5: 0.5271, loss_yns_5: 0.1736, loss_cls_dn_0: 0.4948, loss_box_dn_0: 0.9988, loss_cls_dn_1: 0.4746, loss_box_dn_1: 1.5450, loss_cls_dn_2: 0.4762, loss_box_dn_2: 1.5322, loss_cls_dn_3: 0.4493, loss_box_dn_3: 1.6460, loss_cls_dn_4: 0.4546, loss_box_dn_4: 1.6956, loss_cls_dn_5: 0.4895, loss_box_dn_5: 1.7463, loss_dense_depth: 1.2673, loss: 42.1854, grad_norm: 74.0453 -2026-03-06 17:54:24,474 - mmdet - INFO - Iter [17/23400] lr: 1.064e-04, eta: 2 days, 10:44:25, time: 3.673, data_time: 0.074, memory: 36993, loss_cls_0: 1.1971, loss_box_0: 2.3120, loss_cns_0: 0.5875, loss_yns_0: 0.1730, loss_cls_1: 1.2669, loss_box_1: 3.1222, loss_cns_1: 0.5006, loss_yns_1: 0.1742, loss_cls_2: 1.2713, loss_box_2: 3.1475, loss_cns_2: 0.5091, loss_yns_2: 0.1768, loss_cls_3: 1.2864, loss_box_3: 3.2360, loss_cns_3: 0.5310, loss_yns_3: 0.1765, loss_cls_4: 1.2551, loss_box_4: 3.2315, loss_cns_4: 0.5183, loss_yns_4: 0.1722, loss_cls_5: 1.2678, loss_box_5: 3.2865, loss_cns_5: 0.5163, loss_yns_5: 0.1776, loss_cls_dn_0: 0.4890, loss_box_dn_0: 1.0220, loss_cls_dn_1: 0.4583, loss_box_dn_1: 1.8683, loss_cls_dn_2: 0.4556, loss_box_dn_2: 1.8447, loss_cls_dn_3: 0.4283, loss_box_dn_3: 1.9785, loss_cls_dn_4: 0.4366, loss_box_dn_4: 2.0202, loss_cls_dn_5: 0.4650, loss_box_dn_5: 2.0667, loss_dense_depth: 1.1878, loss: 44.8144, grad_norm: 81.0714 -2026-03-06 17:54:29,356 - mmdet - INFO - Iter [18/23400] lr: 1.068e-04, eta: 2 days, 9:14:13, time: 4.885, data_time: 0.071, memory: 36993, loss_cls_0: 1.2119, loss_box_0: 2.3322, loss_cns_0: 0.5872, loss_yns_0: 0.1732, loss_cls_1: 1.2471, loss_box_1: 3.3732, loss_cns_1: 0.4782, loss_yns_1: 0.1757, loss_cls_2: 1.2444, loss_box_2: 3.3887, loss_cns_2: 0.4871, loss_yns_2: 0.1781, loss_cls_3: 1.2566, loss_box_3: 3.4475, loss_cns_3: 0.5238, loss_yns_3: 0.1769, loss_cls_4: 1.2457, loss_box_4: 3.4031, loss_cns_4: 0.4984, loss_yns_4: 0.1719, loss_cls_5: 1.2550, loss_box_5: 3.4224, loss_cns_5: 0.4902, loss_yns_5: 0.1795, loss_cls_dn_0: 0.4783, loss_box_dn_0: 1.0259, loss_cls_dn_1: 0.4527, loss_box_dn_1: 2.0214, loss_cls_dn_2: 0.4489, loss_box_dn_2: 1.9881, loss_cls_dn_3: 0.4280, loss_box_dn_3: 2.0717, loss_cls_dn_4: 0.4367, loss_box_dn_4: 2.0793, loss_cls_dn_5: 0.4578, loss_box_dn_5: 2.1113, loss_dense_depth: 1.2748, loss: 46.2230, grad_norm: 78.8572 -2026-03-06 17:54:33,185 - mmdet - INFO - Iter [19/23400] lr: 1.072e-04, eta: 2 days, 7:32:11, time: 3.844, data_time: 0.083, memory: 36993, loss_cls_0: 1.2283, loss_box_0: 2.3027, loss_cns_0: 0.5951, loss_yns_0: 0.1730, loss_cls_1: 1.2329, loss_box_1: 3.2394, loss_cns_1: 0.5020, loss_yns_1: 0.1781, loss_cls_2: 1.2402, loss_box_2: 3.2189, loss_cns_2: 0.5135, loss_yns_2: 0.1774, loss_cls_3: 1.2455, loss_box_3: 3.2100, loss_cns_3: 0.5316, loss_yns_3: 0.1779, loss_cls_4: 1.2379, loss_box_4: 3.1912, loss_cns_4: 0.5246, loss_yns_4: 0.1712, loss_cls_5: 1.2494, loss_box_5: 3.2198, loss_cns_5: 0.5155, loss_yns_5: 0.1769, loss_cls_dn_0: 0.4600, loss_box_dn_0: 1.0170, loss_cls_dn_1: 0.4651, loss_box_dn_1: 1.4043, loss_cls_dn_2: 0.4535, loss_box_dn_2: 1.4042, loss_cls_dn_3: 0.4418, loss_box_dn_3: 1.4257, loss_cls_dn_4: 0.4513, loss_box_dn_4: 1.4481, loss_cls_dn_5: 0.4657, loss_box_dn_5: 1.5093, loss_dense_depth: 1.1960, loss: 42.1952, grad_norm: 49.0674 -2026-03-06 17:54:37,050 - mmdet - INFO - Iter [20/23400] lr: 1.076e-04, eta: 2 days, 6:00:23, time: 3.847, data_time: 0.076, memory: 36993, loss_cls_0: 1.1602, loss_box_0: 2.2734, loss_cns_0: 0.5917, loss_yns_0: 0.1691, loss_cls_1: 1.2028, loss_box_1: 3.0834, loss_cns_1: 0.5175, loss_yns_1: 0.1743, loss_cls_2: 1.2202, loss_box_2: 3.0889, loss_cns_2: 0.5299, loss_yns_2: 0.1750, loss_cls_3: 1.2292, loss_box_3: 3.0765, loss_cns_3: 0.5300, loss_yns_3: 0.1745, loss_cls_4: 1.2213, loss_box_4: 3.1489, loss_cns_4: 0.5288, loss_yns_4: 0.1701, loss_cls_5: 1.2399, loss_box_5: 3.1970, loss_cns_5: 0.5259, loss_yns_5: 0.1750, loss_cls_dn_0: 0.4693, loss_box_dn_0: 1.0074, loss_cls_dn_1: 0.4343, loss_box_dn_1: 1.7411, loss_cls_dn_2: 0.4268, loss_box_dn_2: 1.7120, loss_cls_dn_3: 0.4149, loss_box_dn_3: 1.6750, loss_cls_dn_4: 0.4234, loss_box_dn_4: 1.6842, loss_cls_dn_5: 0.4286, loss_box_dn_5: 1.7407, loss_dense_depth: 1.1571, loss: 42.7186, grad_norm: 64.6720 -2026-03-06 17:54:41,918 - mmdet - INFO - Iter [21/23400] lr: 1.080e-04, eta: 2 days, 4:56:16, time: 4.868, data_time: 0.071, memory: 36993, loss_cls_0: 1.1653, loss_box_0: 2.2209, loss_cns_0: 0.5940, loss_yns_0: 0.1707, loss_cls_1: 1.2206, loss_box_1: 3.0589, loss_cns_1: 0.5245, loss_yns_1: 0.1752, loss_cls_2: 1.2283, loss_box_2: 3.0520, loss_cns_2: 0.5245, loss_yns_2: 0.1773, loss_cls_3: 1.2401, loss_box_3: 3.0419, loss_cns_3: 0.5200, loss_yns_3: 0.1720, loss_cls_4: 1.2387, loss_box_4: 3.0714, loss_cns_4: 0.5187, loss_yns_4: 0.1723, loss_cls_5: 1.2522, loss_box_5: 3.1006, loss_cns_5: 0.5256, loss_yns_5: 0.1752, loss_cls_dn_0: 0.4892, loss_box_dn_0: 1.0091, loss_cls_dn_1: 0.4263, loss_box_dn_1: 1.3547, loss_cls_dn_2: 0.4216, loss_box_dn_2: 1.3490, loss_cls_dn_3: 0.4133, loss_box_dn_3: 1.3489, loss_cls_dn_4: 0.4153, loss_box_dn_4: 1.3771, loss_cls_dn_5: 0.4209, loss_box_dn_5: 1.4210, loss_dense_depth: 1.2325, loss: 40.8199, grad_norm: 61.7929 -2026-03-06 17:54:49,281 - mmdet - INFO - Iter [22/23400] lr: 1.084e-04, eta: 2 days, 4:42:39, time: 7.390, data_time: 0.094, memory: 36993, loss_cls_0: 1.1575, loss_box_0: 2.2419, loss_cns_0: 0.5878, loss_yns_0: 0.1711, loss_cls_1: 1.2435, loss_box_1: 2.9639, loss_cns_1: 0.5583, loss_yns_1: 0.1769, loss_cls_2: 1.2283, loss_box_2: 2.9883, loss_cns_2: 0.5413, loss_yns_2: 0.1775, loss_cls_3: 1.2406, loss_box_3: 2.9702, loss_cns_3: 0.5399, loss_yns_3: 0.1757, loss_cls_4: 1.2592, loss_box_4: 2.9574, loss_cns_4: 0.5475, loss_yns_4: 0.1753, loss_cls_5: 1.2528, loss_box_5: 2.9896, loss_cns_5: 0.5541, loss_yns_5: 0.1730, loss_cls_dn_0: 0.4895, loss_box_dn_0: 1.0009, loss_cls_dn_1: 0.4251, loss_box_dn_1: 1.1792, loss_cls_dn_2: 0.4283, loss_box_dn_2: 1.2311, loss_cls_dn_3: 0.4240, loss_box_dn_3: 1.2792, loss_cls_dn_4: 0.4164, loss_box_dn_4: 1.3347, loss_cls_dn_5: 0.4229, loss_box_dn_5: 1.3944, loss_dense_depth: 1.1298, loss: 40.0270, grad_norm: 69.7735 -2026-03-06 17:55:02,979 - mmdet - INFO - Iter [23/23400] lr: 1.088e-04, eta: 2 days, 6:17:03, time: 13.698, data_time: 0.043, memory: 36993, loss_cls_0: 1.1707, loss_box_0: 2.2535, loss_cns_0: 0.5888, loss_yns_0: 0.1696, loss_cls_1: 1.2312, loss_box_1: 3.0049, loss_cns_1: 0.5401, loss_yns_1: 0.1738, loss_cls_2: 1.2312, loss_box_2: 2.9794, loss_cns_2: 0.5411, loss_yns_2: 0.1785, loss_cls_3: 1.2229, loss_box_3: 2.9714, loss_cns_3: 0.5521, loss_yns_3: 0.1759, loss_cls_4: 1.2386, loss_box_4: 2.9827, loss_cns_4: 0.5526, loss_yns_4: 0.1741, loss_cls_5: 1.2362, loss_box_5: 3.0801, loss_cns_5: 0.5380, loss_yns_5: 0.1721, loss_cls_dn_0: 0.4771, loss_box_dn_0: 1.0134, loss_cls_dn_1: 0.4242, loss_box_dn_1: 1.2323, loss_cls_dn_2: 0.4313, loss_box_dn_2: 1.2787, loss_cls_dn_3: 0.4359, loss_box_dn_3: 1.3287, loss_cls_dn_4: 0.4261, loss_box_dn_4: 1.3844, loss_cls_dn_5: 0.4280, loss_box_dn_5: 1.4428, loss_dense_depth: 1.2506, loss: 40.5129, grad_norm: 75.0214 -2026-03-06 17:55:04,365 - mmdet - INFO - Iter [24/23400] lr: 1.092e-04, eta: 2 days, 4:23:41, time: 1.384, data_time: 0.042, memory: 36993, loss_cls_0: 1.1595, loss_box_0: 2.2497, loss_cns_0: 0.5918, loss_yns_0: 0.1675, loss_cls_1: 1.2070, loss_box_1: 3.1807, loss_cns_1: 0.5260, loss_yns_1: 0.1719, loss_cls_2: 1.2273, loss_box_2: 3.1499, loss_cns_2: 0.5364, loss_yns_2: 0.1746, loss_cls_3: 1.2036, loss_box_3: 3.1627, loss_cns_3: 0.5430, loss_yns_3: 0.1748, loss_cls_4: 1.2082, loss_box_4: 3.2141, loss_cns_4: 0.5219, loss_yns_4: 0.1726, loss_cls_5: 1.2198, loss_box_5: 3.3811, loss_cns_5: 0.4983, loss_yns_5: 0.1724, loss_cls_dn_0: 0.4407, loss_box_dn_0: 1.0087, loss_cls_dn_1: 0.4107, loss_box_dn_1: 1.3242, loss_cls_dn_2: 0.4100, loss_box_dn_2: 1.3499, loss_cls_dn_3: 0.4252, loss_box_dn_3: 1.3841, loss_cls_dn_4: 0.4240, loss_box_dn_4: 1.4373, loss_cls_dn_5: 0.4146, loss_box_dn_5: 1.4929, loss_dense_depth: 1.1123, loss: 41.4494, grad_norm: 73.1865 -2026-03-06 17:55:07,971 - mmdet - INFO - Iter [25/23400] lr: 1.096e-04, eta: 2 days, 3:14:00, time: 3.607, data_time: 0.047, memory: 36993, loss_cls_0: 1.1849, loss_box_0: 2.2333, loss_cns_0: 0.5948, loss_yns_0: 0.1684, loss_cls_1: 1.1907, loss_box_1: 3.1956, loss_cns_1: 0.5114, loss_yns_1: 0.1742, loss_cls_2: 1.2109, loss_box_2: 3.1565, loss_cns_2: 0.5298, loss_yns_2: 0.1747, loss_cls_3: 1.2024, loss_box_3: 3.1398, loss_cns_3: 0.5354, loss_yns_3: 0.1741, loss_cls_4: 1.2039, loss_box_4: 3.1925, loss_cns_4: 0.5167, loss_yns_4: 0.1723, loss_cls_5: 1.2170, loss_box_5: 3.2661, loss_cns_5: 0.4980, loss_yns_5: 0.1679, loss_cls_dn_0: 0.4291, loss_box_dn_0: 1.0010, loss_cls_dn_1: 0.4106, loss_box_dn_1: 1.3854, loss_cls_dn_2: 0.4083, loss_box_dn_2: 1.3812, loss_cls_dn_3: 0.4231, loss_box_dn_3: 1.4039, loss_cls_dn_4: 0.4325, loss_box_dn_4: 1.4501, loss_cls_dn_5: 0.4139, loss_box_dn_5: 1.4592, loss_dense_depth: 1.0752, loss: 41.2848, grad_norm: 69.6576 -2026-03-06 17:55:13,205 - mmdet - INFO - Iter [26/23400] lr: 1.100e-04, eta: 2 days, 2:34:03, time: 5.232, data_time: 0.047, memory: 36993, loss_cls_0: 1.1624, loss_box_0: 2.1705, loss_cns_0: 0.6032, loss_yns_0: 0.1682, loss_cls_1: 1.1788, loss_box_1: 3.0186, loss_cns_1: 0.5170, loss_yns_1: 0.1741, loss_cls_2: 1.1900, loss_box_2: 3.0048, loss_cns_2: 0.5373, loss_yns_2: 0.1763, loss_cls_3: 1.1962, loss_box_3: 2.9501, loss_cns_3: 0.5509, loss_yns_3: 0.1731, loss_cls_4: 1.2039, loss_box_4: 2.9582, loss_cns_4: 0.5420, loss_yns_4: 0.1758, loss_cls_5: 1.2093, loss_box_5: 2.9770, loss_cns_5: 0.5356, loss_yns_5: 0.1733, loss_cls_dn_0: 0.4204, loss_box_dn_0: 1.0002, loss_cls_dn_1: 0.4274, loss_box_dn_1: 1.2152, loss_cls_dn_2: 0.4207, loss_box_dn_2: 1.2231, loss_cls_dn_3: 0.4311, loss_box_dn_3: 1.2581, loss_cls_dn_4: 0.4437, loss_box_dn_4: 1.3205, loss_cls_dn_5: 0.4301, loss_box_dn_5: 1.3307, loss_dense_depth: 1.0880, loss: 39.5562, grad_norm: 59.3728 -2026-03-06 17:55:17,112 - mmdet - INFO - Iter [27/23400] lr: 1.104e-04, eta: 2 days, 1:37:56, time: 3.908, data_time: 0.056, memory: 36993, loss_cls_0: 1.1346, loss_box_0: 2.1563, loss_cns_0: 0.6059, loss_yns_0: 0.1678, loss_cls_1: 1.1856, loss_box_1: 2.8269, loss_cns_1: 0.5387, loss_yns_1: 0.1736, loss_cls_2: 1.1983, loss_box_2: 2.8448, loss_cns_2: 0.5485, loss_yns_2: 0.1790, loss_cls_3: 1.2160, loss_box_3: 2.7939, loss_cns_3: 0.5647, loss_yns_3: 0.1757, loss_cls_4: 1.2188, loss_box_4: 2.7541, loss_cns_4: 0.5648, loss_yns_4: 0.1764, loss_cls_5: 1.2269, loss_box_5: 2.8350, loss_cns_5: 0.5576, loss_yns_5: 0.1776, loss_cls_dn_0: 0.4401, loss_box_dn_0: 0.9895, loss_cls_dn_1: 0.4101, loss_box_dn_1: 1.2520, loss_cls_dn_2: 0.4145, loss_box_dn_2: 1.2694, loss_cls_dn_3: 0.4057, loss_box_dn_3: 1.2897, loss_cls_dn_4: 0.4156, loss_box_dn_4: 1.3315, loss_cls_dn_5: 0.4140, loss_box_dn_5: 1.3856, loss_dense_depth: 1.1269, loss: 38.9661, grad_norm: 80.3890 -2026-03-06 17:55:20,584 - mmdet - INFO - Iter [28/23400] lr: 1.108e-04, eta: 2 days, 0:39:45, time: 3.472, data_time: 0.056, memory: 36993, loss_cls_0: 1.1103, loss_box_0: 2.1259, loss_cns_0: 0.6132, loss_yns_0: 0.1679, loss_cls_1: 1.1814, loss_box_1: 2.7140, loss_cns_1: 0.5601, loss_yns_1: 0.1713, loss_cls_2: 1.1958, loss_box_2: 2.7326, loss_cns_2: 0.5601, loss_yns_2: 0.1770, loss_cls_3: 1.2152, loss_box_3: 2.7040, loss_cns_3: 0.5740, loss_yns_3: 0.1733, loss_cls_4: 1.2263, loss_box_4: 2.6428, loss_cns_4: 0.5712, loss_yns_4: 0.1728, loss_cls_5: 1.2248, loss_box_5: 2.7551, loss_cns_5: 0.5595, loss_yns_5: 0.1733, loss_cls_dn_0: 0.4594, loss_box_dn_0: 0.9908, loss_cls_dn_1: 0.3950, loss_box_dn_1: 1.2666, loss_cls_dn_2: 0.4078, loss_box_dn_2: 1.2736, loss_cls_dn_3: 0.3889, loss_box_dn_3: 1.2891, loss_cls_dn_4: 0.3923, loss_box_dn_4: 1.3127, loss_cls_dn_5: 0.4014, loss_box_dn_5: 1.4032, loss_dense_depth: 1.0684, loss: 38.3512, grad_norm: 84.9075 -2026-03-06 17:55:24,549 - mmdet - INFO - Iter [29/23400] lr: 1.112e-04, eta: 1 day, 23:52:11, time: 3.963, data_time: 0.047, memory: 36993, loss_cls_0: 1.0937, loss_box_0: 2.1210, loss_cns_0: 0.6105, loss_yns_0: 0.1731, loss_cls_1: 1.2013, loss_box_1: 2.3439, loss_cns_1: 0.5928, loss_yns_1: 0.1753, loss_cls_2: 1.1747, loss_box_2: 2.3297, loss_cns_2: 0.6082, loss_yns_2: 0.1771, loss_cls_3: 1.1899, loss_box_3: 2.3475, loss_cns_3: 0.6286, loss_yns_3: 0.1787, loss_cls_4: 1.2223, loss_box_4: 2.3136, loss_cns_4: 0.6131, loss_yns_4: 0.1759, loss_cls_5: 1.1889, loss_box_5: 2.4104, loss_cns_5: 0.6104, loss_yns_5: 0.1678, loss_cls_dn_0: 0.4596, loss_box_dn_0: 0.9903, loss_cls_dn_1: 0.3920, loss_box_dn_1: 1.1316, loss_cls_dn_2: 0.4049, loss_box_dn_2: 1.1330, loss_cls_dn_3: 0.3975, loss_box_dn_3: 1.1711, loss_cls_dn_4: 0.3911, loss_box_dn_4: 1.1946, loss_cls_dn_5: 0.4097, loss_box_dn_5: 1.3170, loss_dense_depth: 1.0182, loss: 36.0591, grad_norm: 73.4992 diff --git a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20260306_175131.log.json b/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20260306_175131.log.json deleted file mode 100644 index e25169bef187383c2ae2d6e416bb3fec0f5c11db..0000000000000000000000000000000000000000 --- a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20260306_175131.log.json +++ /dev/null @@ -1,30 +0,0 @@ -{"env_info": "sys.platform: linux\nPython: 3.10.12 (main, Aug 15 2025, 14:32:43) [GCC 11.4.0]\nCUDA available: True\nGPU 0,1,2,3,4,5,6,7: BW200_LC, UBB BW1000_LC\nCUDA_HOME: /opt/dtk\nNVCC: Not Available\nGCC: x86_64-linux-gnu-gcc (Ubuntu 11.4.0-1ubuntu1~22.04) 11.4.0\nPyTorch: 2.5.1\nPyTorch compiling details: PyTorch built with:\n - GCC 10.3\n - C++ Version: 201703\n - Intel(R) Math Kernel Library Version 2020.0.4 Product Build 20200917 for Intel(R) 64 architecture applications\n - OpenMP 201511 (a.k.a. OpenMP 4.5)\n - LAPACK is enabled (usually provided by MKL)\n - NNPACK is enabled\n - CPU capability usage: AVX2\n - HIP Runtime 6.3.25521\n - MIOpen 2.18.0\n - Magma 2.8.0\n - Build settings: BLAS_INFO=mkl, BUILD_TYPE=Release, CXX_COMPILER=/opt/rh/gcc-toolset-10/root/usr/bin/c++, CXX_FLAGS= -D_GLIBCXX_USE_CXX11_ABI=1 -fvisibility-inlines-hidden -DUSE_PTHREADPOOL -DNDEBUG -DUSE_KINETO -DLIBKINETO_NOCUPTI -DLIBKINETO_NOXPUPTI=ON -DUSE_FBGEMM -DUSE_PYTORCH_QNNPACK -DUSE_XNNPACK -DSYMBOLICATE_MOBILE_DEBUG_HANDLE -O2 -fPIC -Wall -Wextra -Werror=return-type -Werror=non-virtual-dtor -Werror=bool-operation -Wnarrowing -Wno-missing-field-initializers -Wno-type-limits -Wno-array-bounds -Wno-unknown-pragmas -Wno-unused-parameter -Wno-strict-overflow -Wno-strict-aliasing -Wno-stringop-overflow -Wsuggest-override -Wno-psabi -Wno-error=old-style-cast -Wno-missing-braces -fdiagnostics-color=always -faligned-new -Wno-unused-but-set-variable -Wno-maybe-uninitialized -fno-math-errno -fno-trapping-math -Werror=format -Wno-stringop-overflow, FORCE_FALLBACK_CUDA_MPI=1, LAPACK_INFO=mkl, PERF_WITH_AVX=1, PERF_WITH_AVX2=1, TORCH_VERSION=2.5.1, USE_CUDA=0, USE_CUDNN=OFF, USE_CUSPARSELT=OFF, USE_EXCEPTION_PTR=1, USE_GFLAGS=1, USE_GLOG=1, USE_GLOO=1, USE_MKL=ON, USE_MKLDNN=0, USE_MPI=1, USE_NCCL=1, USE_NNPACK=ON, USE_OPENMP=1, USE_ROCM=ON, USE_ROCM_KERNEL_ASSERT=OFF, \n\nTorchVision: 0.20.1\nOpenCV: 4.12.0\nMMCV: 1.6.1\nMMCV Compiler: GCC 10.3\nMMCV CUDA Compiler: rocm not available\nMMDetection: 2.25.1+", "config": "plugin = True\nplugin_dir = 'projects/mmdet3d_plugin/'\ndist_params = dict(backend='nccl')\nlog_level = 'INFO'\nwork_dir = './work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704'\ntotal_batch_size = 120\nnum_gpus = 8\nbatch_size = 15\nnum_iters_per_epoch = 234\nnum_epochs = 100\ncheckpoint_epoch_interval = 20\ncheckpoint_config = dict(interval=4680)\nlog_config = dict(\n interval=1,\n hooks=[\n dict(type='TextLoggerHook', by_epoch=False),\n dict(type='TensorboardLoggerHook')\n ])\nload_from = None\nresume_from = None\nworkflow = [('train', 1)]\nfp16 = dict(loss_scale=32.0)\ninput_shape = (704, 256)\ntracking_test = True\ntracking_threshold = 0.2\nclass_names = [\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier',\n 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n]\nnum_classes = 10\nembed_dims = 256\nnum_groups = 8\nnum_decoder = 6\nnum_single_frame_decoder = 1\nuse_deformable_func = True\nstrides = [4, 8, 16, 32]\nnum_levels = 4\nnum_depth_layers = 3\ndrop_out = 0.1\ntemporal = True\ndecouple_attn = True\nwith_quality_estimation = True\nmodel = dict(\n type='Sparse4D',\n use_grid_mask=True,\n use_deformable_func=True,\n img_backbone=dict(\n type='ResNet',\n depth=50,\n num_stages=4,\n frozen_stages=-1,\n norm_eval=False,\n style='pytorch',\n with_cp=True,\n out_indices=(0, 1, 2, 3),\n norm_cfg=dict(type='BN', requires_grad=True),\n pretrained='ckpt/resnet50-19c8e357.pth'),\n img_neck=dict(\n type='FPN',\n num_outs=4,\n start_level=0,\n out_channels=256,\n add_extra_convs='on_output',\n relu_before_extra_convs=True,\n in_channels=[256, 512, 1024, 2048]),\n depth_branch=dict(\n type='DenseDepthNet',\n embed_dims=256,\n num_depth_layers=3,\n loss_weight=0.2),\n head=dict(\n type='Sparse4DHead',\n cls_threshold_to_reg=0.05,\n decouple_attn=True,\n instance_bank=dict(\n type='InstanceBank',\n num_anchor=900,\n embed_dims=256,\n anchor='nuscenes_kmeans900.npy',\n anchor_handler=dict(type='SparseBox3DKeyPointsGenerator'),\n num_temp_instances=600,\n confidence_decay=0.6,\n feat_grad=False),\n anchor_encoder=dict(\n type='SparseBox3DEncoder',\n vel_dims=3,\n embed_dims=[128, 32, 32, 64],\n mode='cat',\n output_fc=False,\n in_loops=1,\n out_loops=4),\n num_single_frame_decoder=1,\n operation_order=[\n 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm',\n 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm',\n 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm',\n 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm',\n 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm',\n 'deformable', 'ffn', 'norm', 'refine'\n ],\n temp_graph_model=dict(\n type='MultiheadAttention',\n embed_dims=512,\n num_heads=8,\n batch_first=True,\n dropout=0.1),\n graph_model=dict(\n type='MultiheadAttention',\n embed_dims=512,\n num_heads=8,\n batch_first=True,\n dropout=0.1),\n norm_layer=dict(type='LN', normalized_shape=256),\n ffn=dict(\n type='AsymmetricFFN',\n in_channels=512,\n pre_norm=dict(type='LN'),\n embed_dims=256,\n feedforward_channels=1024,\n num_fcs=2,\n ffn_drop=0.1,\n act_cfg=dict(type='ReLU', inplace=True)),\n deformable_model=dict(\n type='DeformableFeatureAggregation',\n embed_dims=256,\n num_groups=8,\n num_levels=4,\n num_cams=6,\n attn_drop=0.15,\n use_deformable_func=True,\n use_camera_embed=True,\n residual_mode='cat',\n kps_generator=dict(\n type='SparseBox3DKeyPointsGenerator',\n num_learnable_pts=6,\n fix_scale=[[0, 0, 0], [0.45, 0, 0], [-0.45, 0, 0],\n [0, 0.45, 0], [0, -0.45, 0], [0, 0, 0.45],\n [0, 0, -0.45]])),\n refine_layer=dict(\n type='SparseBox3DRefinementModule',\n embed_dims=256,\n num_cls=10,\n refine_yaw=True,\n with_quality_estimation=True),\n sampler=dict(\n type='SparseBox3DTarget',\n num_dn_groups=5,\n num_temp_dn_groups=3,\n dn_noise_scale=[2.0, 2.0, 2.0, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5],\n max_dn_gt=32,\n add_neg_dn=True,\n cls_weight=2.0,\n box_weight=0.25,\n reg_weights=[2.0, 2.0, 2.0, 0.5, 0.5, 0.5, 0.0, 0.0, 0.0, 0.0],\n cls_wise_reg_weights=dict(\n {9: [2.0, 2.0, 2.0, 1.0, 1.0, 1.0, 0.0, 0.0, 1.0, 1.0]})),\n loss_cls=dict(\n type='FocalLoss',\n use_sigmoid=True,\n gamma=2.0,\n alpha=0.25,\n loss_weight=2.0),\n loss_reg=dict(\n type='SparseBox3DLoss',\n loss_box=dict(type='L1Loss', loss_weight=0.25),\n loss_centerness=dict(type='CrossEntropyLoss', use_sigmoid=True),\n loss_yawness=dict(type='GaussianFocalLoss'),\n cls_allow_reverse=[5]),\n decoder=dict(type='SparseBox3DDecoder'),\n reg_weights=[2.0, 2.0, 2.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0]))\ndataset_type = 'NuScenes3DDetTrackDataset'\ndata_root = 'data/nuscenes/'\nanno_root = 'data/nuscenes_anno_pkls/'\nfile_client_args = dict(backend='disk')\nimg_norm_cfg = dict(\n mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375], to_rgb=True)\ntrain_pipeline = [\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(\n type='LoadPointsFromFile',\n coord_type='LIDAR',\n load_dim=5,\n use_dim=5,\n file_client_args=dict(backend='disk')),\n dict(type='ResizeCropFlipImage'),\n dict(type='MultiScaleDepthMapGenerator', downsample=[4, 8, 16]),\n dict(type='BBoxRotation'),\n dict(type='PhotoMetricDistortionMultiViewImage'),\n dict(\n type='NormalizeMultiviewImage',\n mean=[123.675, 116.28, 103.53],\n std=[58.395, 57.12, 57.375],\n to_rgb=True),\n dict(\n type='CircleObjectRangeFilter',\n class_dist_thred=[55, 55, 55, 55, 55, 55, 55, 55, 55, 55]),\n dict(\n type='InstanceNameFilter',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ]),\n dict(type='NuScenesSparse4DAdaptor'),\n dict(\n type='Collect',\n keys=[\n 'img', 'timestamp', 'projection_mat', 'image_wh', 'gt_depth',\n 'focal', 'gt_bboxes_3d', 'gt_labels_3d'\n ],\n meta_keys=['T_global', 'T_global_inv', 'timestamp', 'instance_id'])\n]\ntest_pipeline = [\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(type='ResizeCropFlipImage'),\n dict(\n type='NormalizeMultiviewImage',\n mean=[123.675, 116.28, 103.53],\n std=[58.395, 57.12, 57.375],\n to_rgb=True),\n dict(type='NuScenesSparse4DAdaptor'),\n dict(\n type='Collect',\n keys=['img', 'timestamp', 'projection_mat', 'image_wh'],\n meta_keys=['T_global', 'T_global_inv', 'timestamp'])\n]\ninput_modality = dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False)\ndata_basic_config = dict(\n type='NuScenes3DDetTrackDataset',\n data_root='data/nuscenes/',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier',\n 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ],\n modality=dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False),\n version='v1.0-trainval')\ndata_aug_conf = dict(\n resize_lim=(0.4, 0.47),\n final_dim=(256, 704),\n bot_pct_lim=(0.0, 0.0),\n rot_lim=(-5.4, 5.4),\n H=900,\n W=1600,\n rand_flip=True,\n rot3d_range=[-0.3925, 0.3925])\ndata = dict(\n samples_per_gpu=15,\n workers_per_gpu=15,\n train=dict(\n type='NuScenes3DDetTrackDataset',\n data_root='data/nuscenes/',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ],\n modality=dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False),\n version='v1.0-trainval',\n ann_file='data/nuscenes_anno_pkls/nuscenes_infos_train.pkl',\n pipeline=[\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(\n type='LoadPointsFromFile',\n coord_type='LIDAR',\n load_dim=5,\n use_dim=5,\n file_client_args=dict(backend='disk')),\n dict(type='ResizeCropFlipImage'),\n dict(type='MultiScaleDepthMapGenerator', downsample=[4, 8, 16]),\n dict(type='BBoxRotation'),\n dict(type='PhotoMetricDistortionMultiViewImage'),\n dict(\n type='NormalizeMultiviewImage',\n mean=[123.675, 116.28, 103.53],\n std=[58.395, 57.12, 57.375],\n to_rgb=True),\n dict(\n type='CircleObjectRangeFilter',\n class_dist_thred=[55, 55, 55, 55, 55, 55, 55, 55, 55, 55]),\n dict(\n type='InstanceNameFilter',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian',\n 'traffic_cone'\n ]),\n dict(type='NuScenesSparse4DAdaptor'),\n dict(\n type='Collect',\n keys=[\n 'img', 'timestamp', 'projection_mat', 'image_wh',\n 'gt_depth', 'focal', 'gt_bboxes_3d', 'gt_labels_3d'\n ],\n meta_keys=[\n 'T_global', 'T_global_inv', 'timestamp', 'instance_id'\n ])\n ],\n test_mode=False,\n data_aug_conf=dict(\n resize_lim=(0.4, 0.47),\n final_dim=(256, 704),\n bot_pct_lim=(0.0, 0.0),\n rot_lim=(-5.4, 5.4),\n H=900,\n W=1600,\n rand_flip=True,\n rot3d_range=[-0.3925, 0.3925]),\n with_seq_flag=True,\n sequences_split_num=2,\n keep_consistent_seq_aug=True),\n val=dict(\n type='NuScenes3DDetTrackDataset',\n data_root='data/nuscenes/',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ],\n modality=dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False),\n version='v1.0-trainval',\n ann_file='data/nuscenes_anno_pkls/nuscenes_infos_val.pkl',\n pipeline=[\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(type='ResizeCropFlipImage'),\n dict(\n type='NormalizeMultiviewImage',\n mean=[123.675, 116.28, 103.53],\n std=[58.395, 57.12, 57.375],\n to_rgb=True),\n dict(type='NuScenesSparse4DAdaptor'),\n dict(\n type='Collect',\n keys=['img', 'timestamp', 'projection_mat', 'image_wh'],\n meta_keys=['T_global', 'T_global_inv', 'timestamp'])\n ],\n data_aug_conf=dict(\n resize_lim=(0.4, 0.47),\n final_dim=(256, 704),\n bot_pct_lim=(0.0, 0.0),\n rot_lim=(-5.4, 5.4),\n H=900,\n W=1600,\n rand_flip=True,\n rot3d_range=[-0.3925, 0.3925]),\n test_mode=True,\n tracking=True,\n tracking_threshold=0.2),\n test=dict(\n type='NuScenes3DDetTrackDataset',\n data_root='data/nuscenes/',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ],\n modality=dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False),\n version='v1.0-trainval',\n ann_file='data/nuscenes_anno_pkls/nuscenes_infos_val.pkl',\n pipeline=[\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(type='ResizeCropFlipImage'),\n dict(\n type='NormalizeMultiviewImage',\n mean=[123.675, 116.28, 103.53],\n std=[58.395, 57.12, 57.375],\n to_rgb=True),\n dict(type='NuScenesSparse4DAdaptor'),\n dict(\n type='Collect',\n keys=['img', 'timestamp', 'projection_mat', 'image_wh'],\n meta_keys=['T_global', 'T_global_inv', 'timestamp'])\n ],\n data_aug_conf=dict(\n resize_lim=(0.4, 0.47),\n final_dim=(256, 704),\n bot_pct_lim=(0.0, 0.0),\n rot_lim=(-5.4, 5.4),\n H=900,\n W=1600,\n rand_flip=True,\n rot3d_range=[-0.3925, 0.3925]),\n test_mode=True,\n tracking=True,\n tracking_threshold=0.2))\noptimizer = dict(\n type='AdamW',\n lr=0.0006,\n weight_decay=0.001,\n paramwise_cfg=dict(custom_keys=dict(img_backbone=dict(lr_mult=0.5))))\noptimizer_config = dict(grad_clip=dict(max_norm=25, norm_type=2))\nlr_config = dict(\n policy='CosineAnnealing',\n warmup='linear',\n warmup_iters=500,\n warmup_ratio=0.3333333333333333,\n min_lr_ratio=0.001)\nrunner = dict(type='IterBasedRunner', max_iters=23400)\nvis_pipeline = [\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(type='Collect', keys=['img'], meta_keys=['timestamp', 'lidar2img'])\n]\nevaluation = dict(\n interval=4680,\n pipeline=[\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(\n type='Collect', keys=['img'], meta_keys=['timestamp', 'lidar2img'])\n ])\ngpu_ids = range(0, 8)\n", "seed": 0, "exp_name": "sparse4dv3_temporal_r50_1x8_bs6_256x704.py"} -{"mode": "train", "epoch": 1, "iter": 1, "lr": 0.0001, "memory": 36993, "data_time": 7.29984, "loss_cls_0": 2.35007, "loss_box_0": 0.00697, "loss_cns_0": 0.00136, "loss_yns_0": 0.00057, "loss_cls_1": 2.14584, "loss_box_1": 0.09113, "loss_cns_1": 0.02044, "loss_yns_1": 0.00674, "loss_cls_2": 2.31689, "loss_box_2": 0.00494, "loss_cns_2": 0.00063, "loss_yns_2": 0.00043, "loss_cls_3": 2.40788, "loss_box_3": 0.02478, "loss_cns_3": 0.00431, "loss_yns_3": 0.00144, "loss_cls_4": 2.02826, "loss_box_4": 0.43832, "loss_cns_4": 0.0573, "loss_yns_4": 0.02754, "loss_cls_5": 2.42722, "loss_box_5": 0.01467, "loss_cns_5": 0.00227, "loss_yns_5": 0.00074, "loss_cls_dn_0": 1.19068, "loss_box_dn_0": 1.46215, "loss_cls_dn_1": 1.10816, "loss_box_dn_1": 1.73429, "loss_cls_dn_2": 1.17027, "loss_box_dn_2": 1.97606, "loss_cls_dn_3": 1.16972, "loss_box_dn_3": 2.25422, "loss_cls_dn_4": 1.05587, "loss_box_dn_4": 2.44015, "loss_cls_dn_5": 1.24004, "loss_box_dn_5": 2.68454, "loss_dense_depth": 1.83719, "loss": 35.7041, "grad_norm": 276.26926, "time": 88.56956} -{"mode": "train", "epoch": 1, "iter": 2, "lr": 0.0001, "memory": 36993, "data_time": 0.09631, "loss_cls_0": 2.05182, "loss_box_0": 0.01413, "loss_cns_0": 0.00386, "loss_yns_0": 0.00138, "loss_cls_1": 2.00304, "loss_box_1": 0.28946, "loss_cns_1": 0.04926, "loss_yns_1": 0.01478, "loss_cls_2": 2.09374, "loss_box_2": 0.27822, "loss_cns_2": 0.02792, "loss_yns_2": 0.01095, "loss_cls_3": 1.96766, "loss_box_3": 0.34526, "loss_cns_3": 0.04276, "loss_yns_3": 0.01397, "loss_cls_4": 1.80747, "loss_box_4": 1.54621, "loss_cns_4": 0.15023, "loss_yns_4": 0.05388, "loss_cls_5": 2.05617, "loss_box_5": 0.51493, "loss_cns_5": 0.05486, "loss_yns_5": 0.0202, "loss_cls_dn_0": 1.03454, "loss_box_dn_0": 1.24198, "loss_cls_dn_1": 0.95345, "loss_box_dn_1": 2.43805, "loss_cls_dn_2": 0.9654, "loss_box_dn_2": 2.55963, "loss_cls_dn_3": 0.92018, "loss_box_dn_3": 2.63266, "loss_cls_dn_4": 0.8406, "loss_box_dn_4": 2.90659, "loss_cls_dn_5": 0.98482, "loss_box_dn_5": 3.1464, "loss_dense_depth": 1.69484, "loss": 37.73128, "grad_norm": 66.71338, "time": 2.66678} -{"mode": "train", "epoch": 1, "iter": 3, "lr": 0.0001, "memory": 36993, "data_time": 0.09014, "loss_cls_0": 1.45518, "loss_box_0": 2.57564, "loss_cns_0": 0.64771, "loss_yns_0": 0.21636, "loss_cls_1": 1.79098, "loss_box_1": 1.71069, "loss_cns_1": 0.27093, "loss_yns_1": 0.09759, "loss_cls_2": 1.79831, "loss_box_2": 3.81427, "loss_cns_2": 0.34078, "loss_yns_2": 0.18998, "loss_cls_3": 1.63776, "loss_box_3": 4.78598, "loss_cns_3": 0.43509, "loss_yns_3": 0.20908, "loss_cls_4": 1.58312, "loss_box_4": 4.05863, "loss_cns_4": 0.36783, "loss_yns_4": 0.15765, "loss_cls_5": 1.71103, "loss_box_5": 2.56618, "loss_cns_5": 0.1874, "loss_yns_5": 0.0908, "loss_cls_dn_0": 0.70397, "loss_box_dn_0": 1.17723, "loss_cls_dn_1": 0.83402, "loss_box_dn_1": 2.41603, "loss_cls_dn_2": 0.79877, "loss_box_dn_2": 2.60924, "loss_cls_dn_3": 0.71741, "loss_box_dn_3": 2.80588, "loss_cls_dn_4": 0.72045, "loss_box_dn_4": 3.03558, "loss_cls_dn_5": 0.807, "loss_box_dn_5": 3.26759, "loss_dense_depth": 1.64329, "loss": 54.23544, "grad_norm": 103.54157, "time": 4.96674} -{"mode": "train", "epoch": 1, "iter": 4, "lr": 0.0001, "memory": 36993, "data_time": 0.07061, "loss_cls_0": 1.35582, "loss_box_0": 2.55239, "loss_cns_0": 0.56356, "loss_yns_0": 0.18279, "loss_cls_1": 1.61884, "loss_box_1": 3.08635, "loss_cns_1": 0.44843, "loss_yns_1": 0.18201, "loss_cls_2": 1.70868, "loss_box_2": 3.55763, "loss_cns_2": 0.45723, "loss_yns_2": 0.18696, "loss_cls_3": 1.52134, "loss_box_3": 4.04079, "loss_cns_3": 0.47467, "loss_yns_3": 0.2167, "loss_cls_4": 1.48181, "loss_box_4": 4.43954, "loss_cns_4": 0.40258, "loss_yns_4": 0.19512, "loss_cls_5": 1.50671, "loss_box_5": 4.64543, "loss_cns_5": 0.44637, "loss_yns_5": 0.20259, "loss_cls_dn_0": 0.58306, "loss_box_dn_0": 1.15811, "loss_cls_dn_1": 0.73311, "loss_box_dn_1": 2.51014, "loss_cls_dn_2": 0.69074, "loss_box_dn_2": 2.55346, "loss_cls_dn_3": 0.62199, "loss_box_dn_3": 2.72316, "loss_cls_dn_4": 0.59869, "loss_box_dn_4": 2.88329, "loss_cls_dn_5": 0.67238, "loss_box_dn_5": 3.0564, "loss_dense_depth": 1.58172, "loss": 55.8406, "grad_norm": 110.89578, "time": 3.60546} -{"mode": "train", "epoch": 1, "iter": 5, "lr": 0.0001, "memory": 36993, "data_time": 0.10292, "loss_cls_0": 1.32094, "loss_box_0": 2.70445, "loss_cns_0": 0.49098, "loss_yns_0": 0.19714, "loss_cls_1": 1.55527, "loss_box_1": 3.77372, "loss_cns_1": 0.39679, "loss_yns_1": 0.20377, "loss_cls_2": 1.59697, "loss_box_2": 3.92534, "loss_cns_2": 0.39584, "loss_yns_2": 0.19724, "loss_cls_3": 1.446, "loss_box_3": 4.02946, "loss_cns_3": 0.39397, "loss_yns_3": 0.20065, "loss_cls_4": 1.39244, "loss_box_4": 4.27597, "loss_cns_4": 0.36949, "loss_yns_4": 0.19545, "loss_cls_5": 1.40058, "loss_box_5": 4.46379, "loss_cns_5": 0.41416, "loss_yns_5": 0.20654, "loss_cls_dn_0": 0.54854, "loss_box_dn_0": 1.19911, "loss_cls_dn_1": 0.65628, "loss_box_dn_1": 2.17573, "loss_cls_dn_2": 0.64841, "loss_box_dn_2": 2.2879, "loss_cls_dn_3": 0.56726, "loss_box_dn_3": 2.37375, "loss_cls_dn_4": 0.55696, "loss_box_dn_4": 2.54221, "loss_cls_dn_5": 0.58518, "loss_box_dn_5": 2.63197, "loss_dense_depth": 1.49776, "loss": 53.81805, "grad_norm": 112.23904, "time": 4.83939} -{"mode": "train", "epoch": 1, "iter": 6, "lr": 0.0001, "memory": 36993, "data_time": 0.04551, "loss_cls_0": 1.30087, "loss_box_0": 2.5771, "loss_cns_0": 0.57977, "loss_yns_0": 0.18415, "loss_cls_1": 1.48242, "loss_box_1": 3.86718, "loss_cns_1": 0.3639, "loss_yns_1": 0.18924, "loss_cls_2": 1.48233, "loss_box_2": 4.05205, "loss_cns_2": 0.36247, "loss_yns_2": 0.20297, "loss_cls_3": 1.37313, "loss_box_3": 4.02294, "loss_cns_3": 0.34697, "loss_yns_3": 0.2009, "loss_cls_4": 1.33054, "loss_box_4": 4.28342, "loss_cns_4": 0.30211, "loss_yns_4": 0.19637, "loss_cls_5": 1.33569, "loss_box_5": 4.43882, "loss_cns_5": 0.29964, "loss_yns_5": 0.20172, "loss_cls_dn_0": 0.51568, "loss_box_dn_0": 1.1711, "loss_cls_dn_1": 0.59876, "loss_box_dn_1": 2.37228, "loss_cls_dn_2": 0.59547, "loss_box_dn_2": 2.42088, "loss_cls_dn_3": 0.52152, "loss_box_dn_3": 2.48216, "loss_cls_dn_4": 0.49324, "loss_box_dn_4": 2.70062, "loss_cls_dn_5": 0.50425, "loss_box_dn_5": 2.77101, "loss_dense_depth": 1.4442, "loss": 53.5679, "grad_norm": 123.60237, "time": 3.69619} -{"mode": "train", "epoch": 1, "iter": 7, "lr": 0.0001, "memory": 36993, "data_time": 0.06316, "loss_cls_0": 1.2905, "loss_box_0": 2.31244, "loss_cns_0": 0.69322, "loss_yns_0": 0.17702, "loss_cls_1": 1.37239, "loss_box_1": 3.47039, "loss_cns_1": 0.50397, "loss_yns_1": 0.19318, "loss_cls_2": 1.39639, "loss_box_2": 3.55257, "loss_cns_2": 0.47846, "loss_yns_2": 0.19606, "loss_cls_3": 1.30686, "loss_box_3": 3.51563, "loss_cns_3": 0.51362, "loss_yns_3": 0.18799, "loss_cls_4": 1.31111, "loss_box_4": 3.79829, "loss_cns_4": 0.5016, "loss_yns_4": 0.18912, "loss_cls_5": 1.33821, "loss_box_5": 3.94683, "loss_cns_5": 0.48258, "loss_yns_5": 0.20404, "loss_cls_dn_0": 0.49525, "loss_box_dn_0": 1.08841, "loss_cls_dn_1": 0.53028, "loss_box_dn_1": 2.436, "loss_cls_dn_2": 0.5351, "loss_box_dn_2": 2.44038, "loss_cls_dn_3": 0.47639, "loss_box_dn_3": 2.51658, "loss_cls_dn_4": 0.44179, "loss_box_dn_4": 2.72864, "loss_cls_dn_5": 0.43745, "loss_box_dn_5": 2.7957, "loss_dense_depth": 1.45764, "loss": 51.31209, "grad_norm": 108.61987, "time": 4.93209} -{"mode": "train", "epoch": 1, "iter": 8, "lr": 0.0001, "memory": 36993, "data_time": 0.04724, "loss_cls_0": 1.25658, "loss_box_0": 2.21996, "loss_cns_0": 0.68193, "loss_yns_0": 0.17879, "loss_cls_1": 1.31849, "loss_box_1": 3.33904, "loss_cns_1": 0.57656, "loss_yns_1": 0.19156, "loss_cls_2": 1.3863, "loss_box_2": 3.50186, "loss_cns_2": 0.51131, "loss_yns_2": 0.19302, "loss_cls_3": 1.30629, "loss_box_3": 3.66897, "loss_cns_3": 0.48477, "loss_yns_3": 0.18873, "loss_cls_4": 1.31683, "loss_box_4": 3.73065, "loss_cns_4": 0.47771, "loss_yns_4": 0.19059, "loss_cls_5": 1.33265, "loss_box_5": 3.78951, "loss_cns_5": 0.49413, "loss_yns_5": 0.19363, "loss_cls_dn_0": 0.50788, "loss_box_dn_0": 1.03658, "loss_cls_dn_1": 0.53743, "loss_box_dn_1": 1.82167, "loss_cls_dn_2": 0.54863, "loss_box_dn_2": 1.79383, "loss_cls_dn_3": 0.49276, "loss_box_dn_3": 1.92871, "loss_cls_dn_4": 0.46897, "loss_box_dn_4": 2.04409, "loss_cls_dn_5": 0.44043, "loss_box_dn_5": 2.09295, "loss_dense_depth": 1.343, "loss": 47.58678, "grad_norm": 95.46798, "time": 3.46466} -{"mode": "train", "epoch": 1, "iter": 9, "lr": 0.0001, "memory": 36993, "data_time": 0.04648, "loss_cls_0": 1.22675, "loss_box_0": 2.2247, "loss_cns_0": 0.63521, "loss_yns_0": 0.17575, "loss_cls_1": 1.25786, "loss_box_1": 3.11795, "loss_cns_1": 0.56715, "loss_yns_1": 0.18582, "loss_cls_2": 1.38488, "loss_box_2": 3.28625, "loss_cns_2": 0.49935, "loss_yns_2": 0.19568, "loss_cls_3": 1.27476, "loss_box_3": 3.47312, "loss_cns_3": 0.47651, "loss_yns_3": 0.19047, "loss_cls_4": 1.26699, "loss_box_4": 3.43691, "loss_cns_4": 0.5018, "loss_yns_4": 0.18915, "loss_cls_5": 1.31114, "loss_box_5": 3.51166, "loss_cns_5": 0.54344, "loss_yns_5": 0.18564, "loss_cls_dn_0": 0.51628, "loss_box_dn_0": 1.01828, "loss_cls_dn_1": 0.50574, "loss_box_dn_1": 1.49713, "loss_cls_dn_2": 0.52967, "loss_box_dn_2": 1.53658, "loss_cls_dn_3": 0.46391, "loss_box_dn_3": 1.66094, "loss_cls_dn_4": 0.46278, "loss_box_dn_4": 1.6551, "loss_cls_dn_5": 0.43259, "loss_box_dn_5": 1.72504, "loss_dense_depth": 1.44185, "loss": 44.56483, "grad_norm": 89.80695, "time": 3.91551} -{"mode": "train", "epoch": 1, "iter": 10, "lr": 0.0001, "memory": 36993, "data_time": 0.04835, "loss_cls_0": 1.20889, "loss_box_0": 2.25725, "loss_cns_0": 0.60943, "loss_yns_0": 0.17257, "loss_cls_1": 1.25023, "loss_box_1": 2.9687, "loss_cns_1": 0.58904, "loss_yns_1": 0.18367, "loss_cls_2": 1.31087, "loss_box_2": 3.03085, "loss_cns_2": 0.5715, "loss_yns_2": 0.18962, "loss_cls_3": 1.27174, "loss_box_3": 3.13408, "loss_cns_3": 0.60211, "loss_yns_3": 0.19814, "loss_cls_4": 1.25925, "loss_box_4": 3.17347, "loss_cns_4": 0.59185, "loss_yns_4": 0.18496, "loss_cls_5": 1.31866, "loss_box_5": 3.27571, "loss_cns_5": 0.59525, "loss_yns_5": 0.18574, "loss_cls_dn_0": 0.48913, "loss_box_dn_0": 1.04993, "loss_cls_dn_1": 0.46052, "loss_box_dn_1": 1.54239, "loss_cls_dn_2": 0.48726, "loss_box_dn_2": 1.60708, "loss_cls_dn_3": 0.43207, "loss_box_dn_3": 1.70228, "loss_cls_dn_4": 0.43862, "loss_box_dn_4": 1.74761, "loss_cls_dn_5": 0.41909, "loss_box_dn_5": 1.85381, "loss_dense_depth": 1.26434, "loss": 43.62771, "grad_norm": 66.64968, "time": 4.95625} -{"mode": "train", "epoch": 1, "iter": 11, "lr": 0.0001, "memory": 36993, "data_time": 0.04707, "loss_cls_0": 1.24535, "loss_box_0": 2.24276, "loss_cns_0": 0.62005, "loss_yns_0": 0.17284, "loss_cls_1": 1.31113, "loss_box_1": 2.9068, "loss_cns_1": 0.55105, "loss_yns_1": 0.18239, "loss_cls_2": 1.28968, "loss_box_2": 2.93762, "loss_cns_2": 0.54722, "loss_yns_2": 0.19142, "loss_cls_3": 1.26578, "loss_box_3": 2.98642, "loss_cns_3": 0.58087, "loss_yns_3": 0.18477, "loss_cls_4": 1.25987, "loss_box_4": 3.16609, "loss_cns_4": 0.52309, "loss_yns_4": 0.18519, "loss_cls_5": 1.31597, "loss_box_5": 3.38895, "loss_cns_5": 0.46104, "loss_yns_5": 0.18895, "loss_cls_dn_0": 0.46118, "loss_box_dn_0": 1.03322, "loss_cls_dn_1": 0.41815, "loss_box_dn_1": 1.70936, "loss_cls_dn_2": 0.44542, "loss_box_dn_2": 1.78838, "loss_cls_dn_3": 0.41634, "loss_box_dn_3": 1.84617, "loss_cls_dn_4": 0.40597, "loss_box_dn_4": 1.95897, "loss_cls_dn_5": 0.40261, "loss_box_dn_5": 2.15427, "loss_dense_depth": 1.34721, "loss": 44.09253, "grad_norm": 83.85289, "time": 3.61865} -{"mode": "train", "epoch": 1, "iter": 12, "lr": 0.0001, "memory": 36993, "data_time": 0.0896, "loss_cls_0": 1.24857, "loss_box_0": 2.22255, "loss_cns_0": 0.63898, "loss_yns_0": 0.1716, "loss_cls_1": 1.30115, "loss_box_1": 3.0729, "loss_cns_1": 0.48777, "loss_yns_1": 0.17959, "loss_cls_2": 1.30373, "loss_box_2": 3.02648, "loss_cns_2": 0.49311, "loss_yns_2": 0.18749, "loss_cls_3": 1.24653, "loss_box_3": 3.0428, "loss_cns_3": 0.53174, "loss_yns_3": 0.17724, "loss_cls_4": 1.25868, "loss_box_4": 3.16513, "loss_cns_4": 0.50505, "loss_yns_4": 0.18122, "loss_cls_5": 1.29039, "loss_box_5": 3.60101, "loss_cns_5": 0.45055, "loss_yns_5": 0.18498, "loss_cls_dn_0": 0.4649, "loss_box_dn_0": 1.03448, "loss_cls_dn_1": 0.38905, "loss_box_dn_1": 2.01787, "loss_cls_dn_2": 0.42135, "loss_box_dn_2": 2.04768, "loss_cls_dn_3": 0.39867, "loss_box_dn_3": 2.06811, "loss_cls_dn_4": 0.39031, "loss_box_dn_4": 2.17309, "loss_cls_dn_5": 0.3881, "loss_box_dn_5": 2.39253, "loss_dense_depth": 1.23037, "loss": 45.38576, "grad_norm": 90.01756, "time": 3.75381} -{"mode": "train", "epoch": 1, "iter": 13, "lr": 0.0001, "memory": 36993, "data_time": 0.05341, "loss_cls_0": 1.2138, "loss_box_0": 2.19681, "loss_cns_0": 0.63684, "loss_yns_0": 0.17079, "loss_cls_1": 1.2571, "loss_box_1": 3.09732, "loss_cns_1": 0.49594, "loss_yns_1": 0.17189, "loss_cls_2": 1.2879, "loss_box_2": 3.11738, "loss_cns_2": 0.50254, "loss_yns_2": 0.18276, "loss_cls_3": 1.24436, "loss_box_3": 3.07761, "loss_cns_3": 0.51052, "loss_yns_3": 0.17789, "loss_cls_4": 1.25209, "loss_box_4": 3.07772, "loss_cns_4": 0.50489, "loss_yns_4": 0.17795, "loss_cls_5": 1.26823, "loss_box_5": 3.24324, "loss_cns_5": 0.5151, "loss_yns_5": 0.17548, "loss_cls_dn_0": 0.47745, "loss_box_dn_0": 1.0098, "loss_cls_dn_1": 0.43029, "loss_box_dn_1": 1.73632, "loss_cls_dn_2": 0.47766, "loss_box_dn_2": 1.7302, "loss_cls_dn_3": 0.44349, "loss_box_dn_3": 1.73059, "loss_cls_dn_4": 0.44461, "loss_box_dn_4": 1.78991, "loss_cls_dn_5": 0.44787, "loss_box_dn_5": 1.91677, "loss_dense_depth": 1.30237, "loss": 43.49346, "grad_norm": 64.30343, "time": 4.28375} -{"mode": "train", "epoch": 1, "iter": 14, "lr": 0.00011, "memory": 36993, "data_time": 0.07489, "loss_cls_0": 1.19732, "loss_box_0": 2.18609, "loss_cns_0": 0.62691, "loss_yns_0": 0.17153, "loss_cls_1": 1.27498, "loss_box_1": 2.99684, "loss_cns_1": 0.53177, "loss_yns_1": 0.17254, "loss_cls_2": 1.27479, "loss_box_2": 3.16964, "loss_cns_2": 0.51719, "loss_yns_2": 0.17749, "loss_cls_3": 1.27448, "loss_box_3": 3.13064, "loss_cns_3": 0.50968, "loss_yns_3": 0.17466, "loss_cls_4": 1.29113, "loss_box_4": 3.15786, "loss_cns_4": 0.49427, "loss_yns_4": 0.17731, "loss_cls_5": 1.28512, "loss_box_5": 3.18646, "loss_cns_5": 0.49282, "loss_yns_5": 0.17753, "loss_cls_dn_0": 0.48122, "loss_box_dn_0": 1.00914, "loss_cls_dn_1": 0.45378, "loss_box_dn_1": 1.52973, "loss_cls_dn_2": 0.49713, "loss_box_dn_2": 1.54103, "loss_cls_dn_3": 0.45504, "loss_box_dn_3": 1.5535, "loss_cls_dn_4": 0.46838, "loss_box_dn_4": 1.57613, "loss_cls_dn_5": 0.48104, "loss_box_dn_5": 1.58471, "loss_dense_depth": 1.21519, "loss": 42.49506, "grad_norm": 69.12981, "time": 3.69976} -{"mode": "train", "epoch": 1, "iter": 15, "lr": 0.00011, "memory": 36993, "data_time": 0.04738, "loss_cls_0": 1.20665, "loss_box_0": 2.25896, "loss_cns_0": 0.61069, "loss_yns_0": 0.17255, "loss_cls_1": 1.32317, "loss_box_1": 2.72662, "loss_cns_1": 0.56042, "loss_yns_1": 0.17283, "loss_cls_2": 1.295, "loss_box_2": 2.88802, "loss_cns_2": 0.56127, "loss_yns_2": 0.17803, "loss_cls_3": 1.31654, "loss_box_3": 2.93469, "loss_cns_3": 0.54308, "loss_yns_3": 0.17754, "loss_cls_4": 1.30225, "loss_box_4": 2.99443, "loss_cns_4": 0.53279, "loss_yns_4": 0.17522, "loss_cls_5": 1.29735, "loss_box_5": 3.09737, "loss_cns_5": 0.52711, "loss_yns_5": 0.17661, "loss_cls_dn_0": 0.49021, "loss_box_dn_0": 1.00157, "loss_cls_dn_1": 0.46255, "loss_box_dn_1": 1.50313, "loss_cls_dn_2": 0.48824, "loss_box_dn_2": 1.53302, "loss_cls_dn_3": 0.45026, "loss_box_dn_3": 1.61871, "loss_cls_dn_4": 0.46392, "loss_box_dn_4": 1.65678, "loss_cls_dn_5": 0.49024, "loss_box_dn_5": 1.70262, "loss_dense_depth": 1.21556, "loss": 42.10601, "grad_norm": 96.45647, "time": 5.2639} -{"mode": "train", "epoch": 1, "iter": 16, "lr": 0.00011, "memory": 36993, "data_time": 0.0965, "loss_cls_0": 1.20139, "loss_box_0": 2.24438, "loss_cns_0": 0.60154, "loss_yns_0": 0.17439, "loss_cls_1": 1.29128, "loss_box_1": 2.87039, "loss_cns_1": 0.53624, "loss_yns_1": 0.17234, "loss_cls_2": 1.28459, "loss_box_2": 2.91322, "loss_cns_2": 0.5357, "loss_yns_2": 0.17843, "loss_cls_3": 1.30307, "loss_box_3": 2.95634, "loss_cns_3": 0.53089, "loss_yns_3": 0.17927, "loss_cls_4": 1.26734, "loss_box_4": 2.97145, "loss_cns_4": 0.53376, "loss_yns_4": 0.1722, "loss_cls_5": 1.27431, "loss_box_5": 3.02193, "loss_cns_5": 0.52713, "loss_yns_5": 0.17359, "loss_cls_dn_0": 0.49485, "loss_box_dn_0": 0.99879, "loss_cls_dn_1": 0.47464, "loss_box_dn_1": 1.54498, "loss_cls_dn_2": 0.47618, "loss_box_dn_2": 1.53221, "loss_cls_dn_3": 0.44928, "loss_box_dn_3": 1.64601, "loss_cls_dn_4": 0.45455, "loss_box_dn_4": 1.69561, "loss_cls_dn_5": 0.4895, "loss_box_dn_5": 1.7463, "loss_dense_depth": 1.26731, "loss": 42.18538, "grad_norm": 74.04528, "time": 3.83458} -{"mode": "train", "epoch": 1, "iter": 17, "lr": 0.00011, "memory": 36993, "data_time": 0.07408, "loss_cls_0": 1.19705, "loss_box_0": 2.31203, "loss_cns_0": 0.58748, "loss_yns_0": 0.17297, "loss_cls_1": 1.26693, "loss_box_1": 3.12221, "loss_cns_1": 0.50056, "loss_yns_1": 0.17421, "loss_cls_2": 1.27131, "loss_box_2": 3.14748, "loss_cns_2": 0.50911, "loss_yns_2": 0.17675, "loss_cls_3": 1.28644, "loss_box_3": 3.23603, "loss_cns_3": 0.53099, "loss_yns_3": 0.17655, "loss_cls_4": 1.25511, "loss_box_4": 3.23152, "loss_cns_4": 0.51828, "loss_yns_4": 0.17224, "loss_cls_5": 1.26776, "loss_box_5": 3.2865, "loss_cns_5": 0.5163, "loss_yns_5": 0.17761, "loss_cls_dn_0": 0.48903, "loss_box_dn_0": 1.02198, "loss_cls_dn_1": 0.4583, "loss_box_dn_1": 1.86828, "loss_cls_dn_2": 0.45558, "loss_box_dn_2": 1.84475, "loss_cls_dn_3": 0.42832, "loss_box_dn_3": 1.97849, "loss_cls_dn_4": 0.43657, "loss_box_dn_4": 2.02022, "loss_cls_dn_5": 0.46501, "loss_box_dn_5": 2.06667, "loss_dense_depth": 1.18778, "loss": 44.81444, "grad_norm": 81.07143, "time": 3.67304} -{"mode": "train", "epoch": 1, "iter": 18, "lr": 0.00011, "memory": 36993, "data_time": 0.07057, "loss_cls_0": 1.21185, "loss_box_0": 2.33223, "loss_cns_0": 0.58721, "loss_yns_0": 0.17322, "loss_cls_1": 1.24713, "loss_box_1": 3.37324, "loss_cns_1": 0.47822, "loss_yns_1": 0.17572, "loss_cls_2": 1.24443, "loss_box_2": 3.38866, "loss_cns_2": 0.48711, "loss_yns_2": 0.17809, "loss_cls_3": 1.25656, "loss_box_3": 3.44751, "loss_cns_3": 0.52384, "loss_yns_3": 0.17691, "loss_cls_4": 1.24575, "loss_box_4": 3.40311, "loss_cns_4": 0.49837, "loss_yns_4": 0.17194, "loss_cls_5": 1.25497, "loss_box_5": 3.42244, "loss_cns_5": 0.49016, "loss_yns_5": 0.17948, "loss_cls_dn_0": 0.47833, "loss_box_dn_0": 1.02592, "loss_cls_dn_1": 0.45266, "loss_box_dn_1": 2.02136, "loss_cls_dn_2": 0.44894, "loss_box_dn_2": 1.98808, "loss_cls_dn_3": 0.42801, "loss_box_dn_3": 2.07174, "loss_cls_dn_4": 0.43666, "loss_box_dn_4": 2.07927, "loss_cls_dn_5": 0.45779, "loss_box_dn_5": 2.11132, "loss_dense_depth": 1.2748, "loss": 46.22304, "grad_norm": 78.85723, "time": 4.88465} -{"mode": "train", "epoch": 1, "iter": 19, "lr": 0.00011, "memory": 36993, "data_time": 0.08273, "loss_cls_0": 1.22829, "loss_box_0": 2.30266, "loss_cns_0": 0.59511, "loss_yns_0": 0.17298, "loss_cls_1": 1.23286, "loss_box_1": 3.23938, "loss_cns_1": 0.50197, "loss_yns_1": 0.17813, "loss_cls_2": 1.24025, "loss_box_2": 3.21888, "loss_cns_2": 0.51347, "loss_yns_2": 0.17741, "loss_cls_3": 1.2455, "loss_box_3": 3.21, "loss_cns_3": 0.53159, "loss_yns_3": 0.17787, "loss_cls_4": 1.23788, "loss_box_4": 3.19124, "loss_cns_4": 0.52462, "loss_yns_4": 0.17125, "loss_cls_5": 1.24942, "loss_box_5": 3.21982, "loss_cns_5": 0.51554, "loss_yns_5": 0.17687, "loss_cls_dn_0": 0.46003, "loss_box_dn_0": 1.01699, "loss_cls_dn_1": 0.46514, "loss_box_dn_1": 1.40434, "loss_cls_dn_2": 0.45354, "loss_box_dn_2": 1.40417, "loss_cls_dn_3": 0.4418, "loss_box_dn_3": 1.42574, "loss_cls_dn_4": 0.4513, "loss_box_dn_4": 1.44807, "loss_cls_dn_5": 0.4657, "loss_box_dn_5": 1.50934, "loss_dense_depth": 1.19605, "loss": 42.1952, "grad_norm": 49.06742, "time": 3.84448} -{"mode": "train", "epoch": 1, "iter": 20, "lr": 0.00011, "memory": 36993, "data_time": 0.0764, "loss_cls_0": 1.1602, "loss_box_0": 2.27343, "loss_cns_0": 0.59174, "loss_yns_0": 0.16913, "loss_cls_1": 1.20278, "loss_box_1": 3.08336, "loss_cns_1": 0.51755, "loss_yns_1": 0.17426, "loss_cls_2": 1.22023, "loss_box_2": 3.08893, "loss_cns_2": 0.52994, "loss_yns_2": 0.17496, "loss_cls_3": 1.22922, "loss_box_3": 3.07653, "loss_cns_3": 0.53003, "loss_yns_3": 0.17452, "loss_cls_4": 1.22132, "loss_box_4": 3.14891, "loss_cns_4": 0.52885, "loss_yns_4": 0.17013, "loss_cls_5": 1.23988, "loss_box_5": 3.19704, "loss_cns_5": 0.52586, "loss_yns_5": 0.17505, "loss_cls_dn_0": 0.46933, "loss_box_dn_0": 1.00742, "loss_cls_dn_1": 0.43429, "loss_box_dn_1": 1.74109, "loss_cls_dn_2": 0.42675, "loss_box_dn_2": 1.71198, "loss_cls_dn_3": 0.41489, "loss_box_dn_3": 1.67505, "loss_cls_dn_4": 0.42336, "loss_box_dn_4": 1.6842, "loss_cls_dn_5": 0.42861, "loss_box_dn_5": 1.74073, "loss_dense_depth": 1.15708, "loss": 42.71861, "grad_norm": 64.67196, "time": 3.8469} -{"mode": "train", "epoch": 1, "iter": 21, "lr": 0.00011, "memory": 36993, "data_time": 0.07121, "loss_cls_0": 1.16531, "loss_box_0": 2.22094, "loss_cns_0": 0.59402, "loss_yns_0": 0.17068, "loss_cls_1": 1.22065, "loss_box_1": 3.05889, "loss_cns_1": 0.52451, "loss_yns_1": 0.17518, "loss_cls_2": 1.22832, "loss_box_2": 3.05196, "loss_cns_2": 0.52449, "loss_yns_2": 0.17733, "loss_cls_3": 1.24008, "loss_box_3": 3.04188, "loss_cns_3": 0.52002, "loss_yns_3": 0.17202, "loss_cls_4": 1.23868, "loss_box_4": 3.07144, "loss_cns_4": 0.51865, "loss_yns_4": 0.17226, "loss_cls_5": 1.25222, "loss_box_5": 3.10056, "loss_cns_5": 0.52561, "loss_yns_5": 0.17524, "loss_cls_dn_0": 0.48922, "loss_box_dn_0": 1.00909, "loss_cls_dn_1": 0.42631, "loss_box_dn_1": 1.35474, "loss_cls_dn_2": 0.42164, "loss_box_dn_2": 1.34899, "loss_cls_dn_3": 0.41328, "loss_box_dn_3": 1.34886, "loss_cls_dn_4": 0.41534, "loss_box_dn_4": 1.3771, "loss_cls_dn_5": 0.42088, "loss_box_dn_5": 1.42102, "loss_dense_depth": 1.23253, "loss": 40.81994, "grad_norm": 61.79292, "time": 4.86797} -{"mode": "train", "epoch": 1, "iter": 22, "lr": 0.00011, "memory": 36993, "data_time": 0.09391, "loss_cls_0": 1.15753, "loss_box_0": 2.24188, "loss_cns_0": 0.58782, "loss_yns_0": 0.17114, "loss_cls_1": 1.24353, "loss_box_1": 2.96387, "loss_cns_1": 0.55832, "loss_yns_1": 0.17688, "loss_cls_2": 1.22833, "loss_box_2": 2.98834, "loss_cns_2": 0.54126, "loss_yns_2": 0.17752, "loss_cls_3": 1.24062, "loss_box_3": 2.97017, "loss_cns_3": 0.53987, "loss_yns_3": 0.17574, "loss_cls_4": 1.25916, "loss_box_4": 2.95737, "loss_cns_4": 0.54751, "loss_yns_4": 0.17528, "loss_cls_5": 1.25282, "loss_box_5": 2.9896, "loss_cns_5": 0.55408, "loss_yns_5": 0.17298, "loss_cls_dn_0": 0.48947, "loss_box_dn_0": 1.00093, "loss_cls_dn_1": 0.4251, "loss_box_dn_1": 1.17918, "loss_cls_dn_2": 0.42829, "loss_box_dn_2": 1.23112, "loss_cls_dn_3": 0.42402, "loss_box_dn_3": 1.27919, "loss_cls_dn_4": 0.41637, "loss_box_dn_4": 1.33467, "loss_cls_dn_5": 0.4229, "loss_box_dn_5": 1.39438, "loss_dense_depth": 1.12978, "loss": 40.02703, "grad_norm": 69.77352, "time": 7.38977} -{"mode": "train", "epoch": 1, "iter": 23, "lr": 0.00011, "memory": 36993, "data_time": 0.04252, "loss_cls_0": 1.17074, "loss_box_0": 2.25349, "loss_cns_0": 0.58885, "loss_yns_0": 0.16957, "loss_cls_1": 1.23125, "loss_box_1": 3.00489, "loss_cns_1": 0.54011, "loss_yns_1": 0.17379, "loss_cls_2": 1.23122, "loss_box_2": 2.9794, "loss_cns_2": 0.54115, "loss_yns_2": 0.17852, "loss_cls_3": 1.22286, "loss_box_3": 2.97135, "loss_cns_3": 0.55205, "loss_yns_3": 0.17594, "loss_cls_4": 1.23857, "loss_box_4": 2.98273, "loss_cns_4": 0.55261, "loss_yns_4": 0.17408, "loss_cls_5": 1.23619, "loss_box_5": 3.08006, "loss_cns_5": 0.53804, "loss_yns_5": 0.17208, "loss_cls_dn_0": 0.47705, "loss_box_dn_0": 1.01338, "loss_cls_dn_1": 0.42419, "loss_box_dn_1": 1.23229, "loss_cls_dn_2": 0.43134, "loss_box_dn_2": 1.27868, "loss_cls_dn_3": 0.43587, "loss_box_dn_3": 1.32869, "loss_cls_dn_4": 0.42613, "loss_box_dn_4": 1.38442, "loss_cls_dn_5": 0.42802, "loss_box_dn_5": 1.44281, "loss_dense_depth": 1.25056, "loss": 40.51293, "grad_norm": 75.02144, "time": 13.69798} -{"mode": "train", "epoch": 1, "iter": 24, "lr": 0.00011, "memory": 36993, "data_time": 0.04197, "loss_cls_0": 1.15953, "loss_box_0": 2.2497, "loss_cns_0": 0.59176, "loss_yns_0": 0.1675, "loss_cls_1": 1.20703, "loss_box_1": 3.1807, "loss_cns_1": 0.52604, "loss_yns_1": 0.1719, "loss_cls_2": 1.22731, "loss_box_2": 3.14993, "loss_cns_2": 0.53637, "loss_yns_2": 0.17459, "loss_cls_3": 1.2036, "loss_box_3": 3.1627, "loss_cns_3": 0.543, "loss_yns_3": 0.17479, "loss_cls_4": 1.20823, "loss_box_4": 3.21415, "loss_cns_4": 0.52188, "loss_yns_4": 0.17264, "loss_cls_5": 1.21977, "loss_box_5": 3.38113, "loss_cns_5": 0.49826, "loss_yns_5": 0.17236, "loss_cls_dn_0": 0.4407, "loss_box_dn_0": 1.00865, "loss_cls_dn_1": 0.41071, "loss_box_dn_1": 1.32419, "loss_cls_dn_2": 0.41, "loss_box_dn_2": 1.34994, "loss_cls_dn_3": 0.42521, "loss_box_dn_3": 1.38409, "loss_cls_dn_4": 0.42396, "loss_box_dn_4": 1.43733, "loss_cls_dn_5": 0.41462, "loss_box_dn_5": 1.49291, "loss_dense_depth": 1.11228, "loss": 41.44944, "grad_norm": 73.18649, "time": 1.38423} -{"mode": "train", "epoch": 1, "iter": 25, "lr": 0.00011, "memory": 36993, "data_time": 0.04739, "loss_cls_0": 1.1849, "loss_box_0": 2.23331, "loss_cns_0": 0.59479, "loss_yns_0": 0.16842, "loss_cls_1": 1.19066, "loss_box_1": 3.19563, "loss_cns_1": 0.5114, "loss_yns_1": 0.1742, "loss_cls_2": 1.21086, "loss_box_2": 3.15649, "loss_cns_2": 0.52981, "loss_yns_2": 0.17469, "loss_cls_3": 1.20244, "loss_box_3": 3.13977, "loss_cns_3": 0.53541, "loss_yns_3": 0.17409, "loss_cls_4": 1.20394, "loss_box_4": 3.19248, "loss_cns_4": 0.5167, "loss_yns_4": 0.1723, "loss_cls_5": 1.21702, "loss_box_5": 3.26615, "loss_cns_5": 0.49798, "loss_yns_5": 0.16794, "loss_cls_dn_0": 0.42915, "loss_box_dn_0": 1.00101, "loss_cls_dn_1": 0.41059, "loss_box_dn_1": 1.38543, "loss_cls_dn_2": 0.40825, "loss_box_dn_2": 1.3812, "loss_cls_dn_3": 0.42309, "loss_box_dn_3": 1.40388, "loss_cls_dn_4": 0.43249, "loss_box_dn_4": 1.45013, "loss_cls_dn_5": 0.41388, "loss_box_dn_5": 1.45916, "loss_dense_depth": 1.07516, "loss": 41.28478, "grad_norm": 69.65756, "time": 3.6069} -{"mode": "train", "epoch": 1, "iter": 26, "lr": 0.00011, "memory": 36993, "data_time": 0.04735, "loss_cls_0": 1.16242, "loss_box_0": 2.17053, "loss_cns_0": 0.60323, "loss_yns_0": 0.16822, "loss_cls_1": 1.17877, "loss_box_1": 3.01862, "loss_cns_1": 0.51703, "loss_yns_1": 0.1741, "loss_cls_2": 1.18997, "loss_box_2": 3.00484, "loss_cns_2": 0.53731, "loss_yns_2": 0.17634, "loss_cls_3": 1.19625, "loss_box_3": 2.95014, "loss_cns_3": 0.55094, "loss_yns_3": 0.17313, "loss_cls_4": 1.20388, "loss_box_4": 2.95824, "loss_cns_4": 0.54198, "loss_yns_4": 0.17578, "loss_cls_5": 1.20931, "loss_box_5": 2.97698, "loss_cns_5": 0.5356, "loss_yns_5": 0.17326, "loss_cls_dn_0": 0.42039, "loss_box_dn_0": 1.00022, "loss_cls_dn_1": 0.42739, "loss_box_dn_1": 1.21523, "loss_cls_dn_2": 0.4207, "loss_box_dn_2": 1.22313, "loss_cls_dn_3": 0.43111, "loss_box_dn_3": 1.25811, "loss_cls_dn_4": 0.44373, "loss_box_dn_4": 1.32052, "loss_cls_dn_5": 0.43013, "loss_box_dn_5": 1.33068, "loss_dense_depth": 1.08799, "loss": 39.55621, "grad_norm": 59.37284, "time": 5.23248} -{"mode": "train", "epoch": 1, "iter": 27, "lr": 0.00011, "memory": 36993, "data_time": 0.05579, "loss_cls_0": 1.13456, "loss_box_0": 2.15626, "loss_cns_0": 0.60593, "loss_yns_0": 0.16782, "loss_cls_1": 1.18561, "loss_box_1": 2.82694, "loss_cns_1": 0.53871, "loss_yns_1": 0.17358, "loss_cls_2": 1.19827, "loss_box_2": 2.84477, "loss_cns_2": 0.54849, "loss_yns_2": 0.17898, "loss_cls_3": 1.21601, "loss_box_3": 2.79391, "loss_cns_3": 0.56466, "loss_yns_3": 0.17567, "loss_cls_4": 1.21883, "loss_box_4": 2.7541, "loss_cns_4": 0.56484, "loss_yns_4": 0.17638, "loss_cls_5": 1.22692, "loss_box_5": 2.83496, "loss_cns_5": 0.5576, "loss_yns_5": 0.17764, "loss_cls_dn_0": 0.44011, "loss_box_dn_0": 0.98953, "loss_cls_dn_1": 0.41011, "loss_box_dn_1": 1.25203, "loss_cls_dn_2": 0.41446, "loss_box_dn_2": 1.26941, "loss_cls_dn_3": 0.40567, "loss_box_dn_3": 1.28974, "loss_cls_dn_4": 0.41557, "loss_box_dn_4": 1.33145, "loss_cls_dn_5": 0.41404, "loss_box_dn_5": 1.38562, "loss_dense_depth": 1.12691, "loss": 38.96614, "grad_norm": 80.38898, "time": 3.90791} -{"mode": "train", "epoch": 1, "iter": 28, "lr": 0.00011, "memory": 36993, "data_time": 0.05561, "loss_cls_0": 1.11029, "loss_box_0": 2.12588, "loss_cns_0": 0.61324, "loss_yns_0": 0.1679, "loss_cls_1": 1.18137, "loss_box_1": 2.71398, "loss_cns_1": 0.56012, "loss_yns_1": 0.17134, "loss_cls_2": 1.19582, "loss_box_2": 2.73258, "loss_cns_2": 0.56006, "loss_yns_2": 0.17701, "loss_cls_3": 1.21517, "loss_box_3": 2.70402, "loss_cns_3": 0.57403, "loss_yns_3": 0.17327, "loss_cls_4": 1.22629, "loss_box_4": 2.64276, "loss_cns_4": 0.57116, "loss_yns_4": 0.17283, "loss_cls_5": 1.22484, "loss_box_5": 2.75511, "loss_cns_5": 0.55951, "loss_yns_5": 0.17332, "loss_cls_dn_0": 0.45939, "loss_box_dn_0": 0.99081, "loss_cls_dn_1": 0.39504, "loss_box_dn_1": 1.26664, "loss_cls_dn_2": 0.40784, "loss_box_dn_2": 1.27362, "loss_cls_dn_3": 0.38891, "loss_box_dn_3": 1.28909, "loss_cls_dn_4": 0.39228, "loss_box_dn_4": 1.31268, "loss_cls_dn_5": 0.40141, "loss_box_dn_5": 1.40315, "loss_dense_depth": 1.06842, "loss": 38.35118, "grad_norm": 84.90753, "time": 3.47177} -{"mode": "train", "epoch": 1, "iter": 29, "lr": 0.00011, "memory": 36993, "data_time": 0.04694, "loss_cls_0": 1.09372, "loss_box_0": 2.12104, "loss_cns_0": 0.61048, "loss_yns_0": 0.17311, "loss_cls_1": 1.2013, "loss_box_1": 2.34393, "loss_cns_1": 0.59281, "loss_yns_1": 0.17529, "loss_cls_2": 1.17466, "loss_box_2": 2.32971, "loss_cns_2": 0.60822, "loss_yns_2": 0.17706, "loss_cls_3": 1.1899, "loss_box_3": 2.34746, "loss_cns_3": 0.62861, "loss_yns_3": 0.1787, "loss_cls_4": 1.22235, "loss_box_4": 2.31362, "loss_cns_4": 0.61308, "loss_yns_4": 0.17592, "loss_cls_5": 1.18886, "loss_box_5": 2.41042, "loss_cns_5": 0.61044, "loss_yns_5": 0.1678, "loss_cls_dn_0": 0.4596, "loss_box_dn_0": 0.99035, "loss_cls_dn_1": 0.39197, "loss_box_dn_1": 1.13164, "loss_cls_dn_2": 0.40488, "loss_box_dn_2": 1.13298, "loss_cls_dn_3": 0.39752, "loss_box_dn_3": 1.1711, "loss_cls_dn_4": 0.39112, "loss_box_dn_4": 1.1946, "loss_cls_dn_5": 0.40967, "loss_box_dn_5": 1.31696, "loss_dense_depth": 1.01822, "loss": 36.05911, "grad_norm": 73.49917, "time": 3.96279} diff --git a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20260403_163837.log b/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20260403_163837.log deleted file mode 100644 index 3539599354db690134c3a9b7560f0fc4d7d371a0..0000000000000000000000000000000000000000 --- a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20260403_163837.log +++ /dev/null @@ -1,3221 +0,0 @@ -2026-04-03 16:38:37,164 - mmdet - INFO - Environment info: ------------------------------------------------------------- -sys.platform: linux -Python: 3.10.12 (main, Aug 15 2025, 14:32:43) [GCC 11.4.0] -CUDA available: True -GPU 0,1,2,3,4,5,6,7: BW1000_H -CUDA_HOME: /opt/dtk -NVCC: Not Available -GCC: x86_64-linux-gnu-gcc (Ubuntu 11.4.0-1ubuntu1~22.04) 11.4.0 -PyTorch: 2.5.1 -PyTorch compiling details: PyTorch built with: - - GCC 10.3 - - C++ Version: 201703 - - Intel(R) Math Kernel Library Version 2020.0.4 Product Build 20200917 for Intel(R) 64 architecture applications - - OpenMP 201511 (a.k.a. OpenMP 4.5) - - LAPACK is enabled (usually provided by MKL) - - NNPACK is enabled - - CPU capability usage: AVX512 - - HIP Runtime 6.3.25521 - - MIOpen 2.18.0 - - Magma 2.8.0 - - Build settings: BLAS_INFO=mkl, BUILD_TYPE=Release, CXX_COMPILER=/opt/rh/gcc-toolset-10/root/usr/bin/c++, CXX_FLAGS= -D_GLIBCXX_USE_CXX11_ABI=1 -fvisibility-inlines-hidden -DUSE_PTHREADPOOL -DNDEBUG -DUSE_KINETO -DLIBKINETO_NOCUPTI -DLIBKINETO_NOXPUPTI=ON -DUSE_FBGEMM -DUSE_PYTORCH_QNNPACK -DUSE_XNNPACK -DSYMBOLICATE_MOBILE_DEBUG_HANDLE -O2 -fPIC -Wall -Wextra -Werror=return-type -Werror=non-virtual-dtor -Werror=bool-operation -Wnarrowing -Wno-missing-field-initializers -Wno-type-limits -Wno-array-bounds -Wno-unknown-pragmas -Wno-unused-parameter -Wno-strict-overflow -Wno-strict-aliasing -Wno-stringop-overflow -Wsuggest-override -Wno-psabi -Wno-error=old-style-cast -Wno-missing-braces -fdiagnostics-color=always -faligned-new -Wno-unused-but-set-variable -Wno-maybe-uninitialized -fno-math-errno -fno-trapping-math -Werror=format -Wno-stringop-overflow, FORCE_FALLBACK_CUDA_MPI=1, LAPACK_INFO=mkl, PERF_WITH_AVX=1, PERF_WITH_AVX2=1, TORCH_VERSION=2.5.1, USE_CUDA=0, USE_CUDNN=OFF, USE_CUSPARSELT=OFF, USE_EXCEPTION_PTR=1, USE_GFLAGS=1, USE_GLOG=1, USE_GLOO=1, USE_MKL=ON, USE_MKLDNN=0, USE_MPI=1, USE_NCCL=1, USE_NNPACK=ON, USE_OPENMP=1, USE_ROCM=ON, USE_ROCM_KERNEL_ASSERT=OFF, - -TorchVision: 0.20.1 -OpenCV: 4.12.0 -MMCV: 1.6.1 -MMCV Compiler: GCC 10.3 -MMCV CUDA Compiler: rocm not available -MMDetection: 2.25.1+c41df4b ------------------------------------------------------------- - -2026-04-03 16:38:37,876 - mmdet - INFO - Distributed training: True -2026-04-03 16:38:38,686 - mmdet - INFO - Config: -plugin = True -plugin_dir = 'projects/mmdet3d_plugin/' -dist_params = dict(backend='nccl') -log_level = 'INFO' -work_dir = './work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704' -total_batch_size = 120 -num_gpus = 8 -batch_size = 15 -num_iters_per_epoch = 234 -num_epochs = 100 -checkpoint_epoch_interval = 20 -checkpoint_config = dict(interval=4680) -log_config = dict( - interval=1, - hooks=[ - dict(type='TextLoggerHook', by_epoch=False), - dict(type='TensorboardLoggerHook') - ]) -load_from = None -resume_from = None -workflow = [('train', 1)] -fp16 = dict(loss_scale=32.0) -input_shape = (704, 256) -tracking_test = True -tracking_threshold = 0.2 -class_names = [ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier', - 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' -] -num_classes = 10 -embed_dims = 256 -num_groups = 8 -num_decoder = 6 -num_single_frame_decoder = 1 -use_deformable_func = True -strides = [4, 8, 16, 32] -num_levels = 4 -num_depth_layers = 3 -drop_out = 0.1 -temporal = True -decouple_attn = True -with_quality_estimation = True -model = dict( - type='Sparse4D', - use_grid_mask=True, - use_deformable_func=True, - img_backbone=dict( - type='ResNet', - depth=50, - num_stages=4, - frozen_stages=-1, - norm_eval=False, - style='pytorch', - with_cp=True, - out_indices=(0, 1, 2, 3), - norm_cfg=dict(type='BN', requires_grad=True), - pretrained='ckpt/resnet50-19c8e357.pth'), - img_neck=dict( - type='FPN', - num_outs=4, - start_level=0, - out_channels=256, - add_extra_convs='on_output', - relu_before_extra_convs=True, - in_channels=[256, 512, 1024, 2048]), - depth_branch=dict( - type='DenseDepthNet', - embed_dims=256, - num_depth_layers=3, - loss_weight=0.2), - head=dict( - type='Sparse4DHead', - cls_threshold_to_reg=0.05, - decouple_attn=True, - instance_bank=dict( - type='InstanceBank', - num_anchor=900, - embed_dims=256, - anchor='nuscenes_kmeans900.npy', - anchor_handler=dict(type='SparseBox3DKeyPointsGenerator'), - num_temp_instances=600, - confidence_decay=0.6, - feat_grad=False), - anchor_encoder=dict( - type='SparseBox3DEncoder', - vel_dims=3, - embed_dims=[128, 32, 32, 64], - mode='cat', - output_fc=False, - in_loops=1, - out_loops=4), - num_single_frame_decoder=1, - operation_order=[ - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine' - ], - temp_graph_model=dict( - type='MultiheadAttention', - embed_dims=512, - num_heads=8, - batch_first=True, - dropout=0.1), - graph_model=dict( - type='MultiheadAttention', - embed_dims=512, - num_heads=8, - batch_first=True, - dropout=0.1), - norm_layer=dict(type='LN', normalized_shape=256), - ffn=dict( - type='AsymmetricFFN', - in_channels=512, - pre_norm=dict(type='LN'), - embed_dims=256, - feedforward_channels=1024, - num_fcs=2, - ffn_drop=0.1, - act_cfg=dict(type='ReLU', inplace=True)), - deformable_model=dict( - type='DeformableFeatureAggregation', - embed_dims=256, - num_groups=8, - num_levels=4, - num_cams=6, - attn_drop=0.15, - use_deformable_func=True, - use_camera_embed=True, - residual_mode='cat', - kps_generator=dict( - type='SparseBox3DKeyPointsGenerator', - num_learnable_pts=6, - fix_scale=[[0, 0, 0], [0.45, 0, 0], [-0.45, 0, 0], - [0, 0.45, 0], [0, -0.45, 0], [0, 0, 0.45], - [0, 0, -0.45]])), - refine_layer=dict( - type='SparseBox3DRefinementModule', - embed_dims=256, - num_cls=10, - refine_yaw=True, - with_quality_estimation=True), - sampler=dict( - type='SparseBox3DTarget', - num_dn_groups=5, - num_temp_dn_groups=3, - dn_noise_scale=[2.0, 2.0, 2.0, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5], - max_dn_gt=32, - add_neg_dn=True, - cls_weight=2.0, - box_weight=0.25, - reg_weights=[2.0, 2.0, 2.0, 0.5, 0.5, 0.5, 0.0, 0.0, 0.0, 0.0], - cls_wise_reg_weights=dict( - {9: [2.0, 2.0, 2.0, 1.0, 1.0, 1.0, 0.0, 0.0, 1.0, 1.0]})), - loss_cls=dict( - type='FocalLoss', - use_sigmoid=True, - gamma=2.0, - alpha=0.25, - loss_weight=2.0), - loss_reg=dict( - type='SparseBox3DLoss', - loss_box=dict(type='L1Loss', loss_weight=0.25), - loss_centerness=dict(type='CrossEntropyLoss', use_sigmoid=True), - loss_yawness=dict(type='GaussianFocalLoss'), - cls_allow_reverse=[5]), - decoder=dict(type='SparseBox3DDecoder'), - reg_weights=[2.0, 2.0, 2.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0])) -dataset_type = 'NuScenes3DDetTrackDataset' -data_root = 'data/nuscenes/' -anno_root = 'data/nuscenes_anno_pkls/' -file_client_args = dict(backend='disk') -img_norm_cfg = dict( - mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375], to_rgb=True) -train_pipeline = [ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict( - type='LoadPointsFromFile', - coord_type='LIDAR', - load_dim=5, - use_dim=5, - file_client_args=dict(backend='disk')), - dict(type='ResizeCropFlipImage'), - dict(type='MultiScaleDepthMapGenerator', downsample=[4, 8, 16]), - dict(type='BBoxRotation'), - dict(type='PhotoMetricDistortionMultiViewImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict( - type='CircleObjectRangeFilter', - class_dist_thred=[55, 55, 55, 55, 55, 55, 55, 55, 55, 55]), - dict( - type='InstanceNameFilter', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ]), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=[ - 'img', 'timestamp', 'projection_mat', 'image_wh', 'gt_depth', - 'focal', 'gt_bboxes_3d', 'gt_labels_3d' - ], - meta_keys=['T_global', 'T_global_inv', 'timestamp', 'instance_id']) -] -test_pipeline = [ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='ResizeCropFlipImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=['img', 'timestamp', 'projection_mat', 'image_wh'], - meta_keys=['T_global', 'T_global_inv', 'timestamp']) -] -input_modality = dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False) -data_basic_config = dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier', - 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval') -data_aug_conf = dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]) -data = dict( - samples_per_gpu=15, - workers_per_gpu=15, - train=dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval', - ann_file='data/nuscenes_anno_pkls/nuscenes_infos_train.pkl', - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict( - type='LoadPointsFromFile', - coord_type='LIDAR', - load_dim=5, - use_dim=5, - file_client_args=dict(backend='disk')), - dict(type='ResizeCropFlipImage'), - dict(type='MultiScaleDepthMapGenerator', downsample=[4, 8, 16]), - dict(type='BBoxRotation'), - dict(type='PhotoMetricDistortionMultiViewImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict( - type='CircleObjectRangeFilter', - class_dist_thred=[55, 55, 55, 55, 55, 55, 55, 55, 55, 55]), - dict( - type='InstanceNameFilter', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', - 'traffic_cone' - ]), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=[ - 'img', 'timestamp', 'projection_mat', 'image_wh', - 'gt_depth', 'focal', 'gt_bboxes_3d', 'gt_labels_3d' - ], - meta_keys=[ - 'T_global', 'T_global_inv', 'timestamp', 'instance_id' - ]) - ], - test_mode=False, - data_aug_conf=dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]), - with_seq_flag=True, - sequences_split_num=2, - keep_consistent_seq_aug=True), - val=dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval', - ann_file='data/nuscenes_anno_pkls/nuscenes_infos_val.pkl', - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='ResizeCropFlipImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=['img', 'timestamp', 'projection_mat', 'image_wh'], - meta_keys=['T_global', 'T_global_inv', 'timestamp']) - ], - data_aug_conf=dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]), - test_mode=True, - tracking=True, - tracking_threshold=0.2), - test=dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval', - ann_file='data/nuscenes_anno_pkls/nuscenes_infos_val.pkl', - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='ResizeCropFlipImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=['img', 'timestamp', 'projection_mat', 'image_wh'], - meta_keys=['T_global', 'T_global_inv', 'timestamp']) - ], - data_aug_conf=dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]), - test_mode=True, - tracking=True, - tracking_threshold=0.2)) -optimizer = dict( - type='AdamW', - lr=0.0006, - weight_decay=0.001, - paramwise_cfg=dict(custom_keys=dict(img_backbone=dict(lr_mult=0.5)))) -optimizer_config = dict(grad_clip=dict(max_norm=25, norm_type=2)) -lr_config = dict( - policy='CosineAnnealing', - warmup='linear', - warmup_iters=500, - warmup_ratio=0.3333333333333333, - min_lr_ratio=0.001) -runner = dict(type='IterBasedRunner', max_iters=23400) -vis_pipeline = [ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='Collect', keys=['img'], meta_keys=['timestamp', 'lidar2img']) -] -evaluation = dict( - interval=4680, - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict( - type='Collect', keys=['img'], meta_keys=['timestamp', 'lidar2img']) - ]) -gpu_ids = range(0, 8) - -2026-04-03 16:38:38,687 - mmdet - INFO - Set random seed to 0, deterministic: False -2026-04-03 16:38:38,990 - mmdet - INFO - initialize ResNet with init_cfg {'type': 'Pretrained', 'checkpoint': 'ckpt/resnet50-19c8e357.pth'} -2026-04-03 16:38:39,184 - mmdet - INFO - initialize FPN with init_cfg {'type': 'Xavier', 'layer': 'Conv2d', 'distribution': 'uniform'} -Name of parameter - Initialization information - -img_backbone.conv1.weight - torch.Size([64, 3, 7, 7]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.bn1.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.bn1.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.conv1.weight - torch.Size([64, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn1.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn1.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.conv2.weight - torch.Size([64, 64, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn2.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn2.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.conv3.weight - torch.Size([256, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn3.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn3.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.downsample.0.weight - torch.Size([256, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.downsample.1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.downsample.1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.conv1.weight - torch.Size([64, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn1.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn1.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.conv2.weight - torch.Size([64, 64, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn2.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn2.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.conv3.weight - torch.Size([256, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn3.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn3.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.conv1.weight - torch.Size([64, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn1.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn1.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.conv2.weight - torch.Size([64, 64, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn2.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn2.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.conv3.weight - torch.Size([256, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn3.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn3.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.conv1.weight - torch.Size([128, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn1.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn1.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.conv2.weight - torch.Size([128, 128, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn2.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn2.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.conv3.weight - torch.Size([512, 128, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn3.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn3.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.downsample.0.weight - torch.Size([512, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.downsample.1.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.downsample.1.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.conv1.weight - torch.Size([128, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn1.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn1.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.conv2.weight - torch.Size([128, 128, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn2.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn2.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.conv3.weight - torch.Size([512, 128, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn3.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn3.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.conv1.weight - torch.Size([128, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn1.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn1.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.conv2.weight - torch.Size([128, 128, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn2.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn2.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.conv3.weight - torch.Size([512, 128, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn3.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn3.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.conv1.weight - torch.Size([128, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn1.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn1.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.conv2.weight - torch.Size([128, 128, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn2.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn2.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.conv3.weight - torch.Size([512, 128, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn3.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn3.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.conv1.weight - torch.Size([256, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.downsample.0.weight - torch.Size([1024, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.downsample.1.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.downsample.1.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.conv1.weight - torch.Size([512, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn1.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn1.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.conv2.weight - torch.Size([512, 512, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn2.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn2.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.conv3.weight - torch.Size([2048, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn3.weight - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn3.bias - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.downsample.0.weight - torch.Size([2048, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.downsample.1.weight - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.downsample.1.bias - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.conv1.weight - torch.Size([512, 2048, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn1.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn1.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.conv2.weight - torch.Size([512, 512, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn2.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn2.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.conv3.weight - torch.Size([2048, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn3.weight - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn3.bias - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.conv1.weight - torch.Size([512, 2048, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn1.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn1.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.conv2.weight - torch.Size([512, 512, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn2.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn2.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.conv3.weight - torch.Size([2048, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn3.weight - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn3.bias - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_neck.lateral_convs.0.conv.weight - torch.Size([256, 256, 1, 1]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.lateral_convs.0.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.lateral_convs.1.conv.weight - torch.Size([256, 512, 1, 1]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.lateral_convs.1.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.lateral_convs.2.conv.weight - torch.Size([256, 1024, 1, 1]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.lateral_convs.2.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.lateral_convs.3.conv.weight - torch.Size([256, 2048, 1, 1]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.lateral_convs.3.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.fpn_convs.0.conv.weight - torch.Size([256, 256, 3, 3]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.fpn_convs.0.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.fpn_convs.1.conv.weight - torch.Size([256, 256, 3, 3]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.fpn_convs.1.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.fpn_convs.2.conv.weight - torch.Size([256, 256, 3, 3]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.fpn_convs.2.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.fpn_convs.3.conv.weight - torch.Size([256, 256, 3, 3]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.fpn_convs.3.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.instance_bank.anchor - torch.Size([900, 11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.instance_bank.instance_feature - torch.Size([900, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.instance_bank.anchor_handler.fix_scale - torch.Size([1, 3]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.0.weight - torch.Size([128, 3]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.0.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.2.weight - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.2.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.3.weight - torch.Size([128, 128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.3.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.5.weight - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.5.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.6.weight - torch.Size([128, 128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.6.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.8.weight - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.8.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.9.weight - torch.Size([128, 128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.9.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.11.weight - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.11.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.0.weight - torch.Size([32, 3]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.0.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.2.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.2.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.3.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.3.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.5.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.5.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.6.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.6.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.8.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.8.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.9.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.9.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.11.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.11.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.0.weight - torch.Size([32, 2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.0.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.2.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.2.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.3.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.3.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.5.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.5.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.6.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.6.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.8.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.8.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.9.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.9.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.11.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.11.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.0.weight - torch.Size([64, 3]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.0.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.2.weight - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.2.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.3.weight - torch.Size([64, 64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.3.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.5.weight - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.5.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.6.weight - torch.Size([64, 64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.6.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.8.weight - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.8.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.9.weight - torch.Size([64, 64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.9.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.11.weight - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.11.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.1.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.1.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.1.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.1.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.1.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.1.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.1.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.1.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.3.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.4.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.4.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.4.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.4.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.5.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.5.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.5.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.5.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.6.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.6.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.8.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.8.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.8.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.8.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.8.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.8.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.8.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.8.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.10.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.11.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.11.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.11.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.11.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.12.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.12.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.12.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.12.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.13.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.13.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.15.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.15.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.15.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.15.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.15.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.15.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.15.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.15.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.16.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.16.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.17.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.18.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.18.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.18.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.18.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.19.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.19.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.19.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.19.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.20.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.20.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.22.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.22.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.22.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.22.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.22.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.22.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.22.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.22.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.23.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.23.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.24.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.25.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.25.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.25.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.25.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.26.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.26.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.26.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.26.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.27.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.27.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.29.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.29.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.29.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.29.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.29.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.29.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.29.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.29.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.30.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.30.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.31.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.32.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.32.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.32.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.32.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.33.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.33.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.33.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.33.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.34.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.34.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.36.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.36.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.36.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.36.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.36.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.36.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.36.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.36.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.37.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.37.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.38.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.fc_before.weight - torch.Size([512, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.fc_after.weight - torch.Size([256, 512]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.0.weight - torch.Size([1, 256, 1, 1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.0.bias - torch.Size([1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.1.weight - torch.Size([1, 256, 1, 1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.1.bias - torch.Size([1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.2.weight - torch.Size([1, 256, 1, 1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.2.bias - torch.Size([1]): -The value is the same before and after calling `init_weights` of Sparse4D -2026-04-03 16:38:39,280 - mmdet - INFO - Model: -Sparse4D( - (img_backbone): ResNet( - (conv1): Conv2d(3, 64, kernel_size=(7, 7), stride=(2, 2), padding=(3, 3), bias=False) - (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (maxpool): MaxPool2d(kernel_size=3, stride=2, padding=1, dilation=1, ceil_mode=False) - (layer1): ResLayer( - (0): Bottleneck( - (conv1): Conv2d(64, 64, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (downsample): Sequential( - (0): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - ) - ) - (1): Bottleneck( - (conv1): Conv2d(256, 64, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (2): Bottleneck( - (conv1): Conv2d(256, 64, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - ) - (layer2): ResLayer( - (0): Bottleneck( - (conv1): Conv2d(256, 128, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (downsample): Sequential( - (0): Conv2d(256, 512, kernel_size=(1, 1), stride=(2, 2), bias=False) - (1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - ) - ) - (1): Bottleneck( - (conv1): Conv2d(512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (2): Bottleneck( - (conv1): Conv2d(512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (3): Bottleneck( - (conv1): Conv2d(512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - ) - (layer3): ResLayer( - (0): Bottleneck( - (conv1): Conv2d(512, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (downsample): Sequential( - (0): Conv2d(512, 1024, kernel_size=(1, 1), stride=(2, 2), bias=False) - (1): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - ) - ) - (1): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (2): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (3): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (4): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (5): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - ) - (layer4): ResLayer( - (0): Bottleneck( - (conv1): Conv2d(1024, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(512, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (downsample): Sequential( - (0): Conv2d(1024, 2048, kernel_size=(1, 1), stride=(2, 2), bias=False) - (1): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - ) - ) - (1): Bottleneck( - (conv1): Conv2d(2048, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(512, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (2): Bottleneck( - (conv1): Conv2d(2048, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(512, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - ) - ) - init_cfg={'type': 'Pretrained', 'checkpoint': 'ckpt/resnet50-19c8e357.pth'} - (img_neck): FPN( - (lateral_convs): ModuleList( - (0): ConvModule( - (conv): Conv2d(256, 256, kernel_size=(1, 1), stride=(1, 1)) - ) - (1): ConvModule( - (conv): Conv2d(512, 256, kernel_size=(1, 1), stride=(1, 1)) - ) - (2): ConvModule( - (conv): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1)) - ) - (3): ConvModule( - (conv): Conv2d(2048, 256, kernel_size=(1, 1), stride=(1, 1)) - ) - ) - (fpn_convs): ModuleList( - (0-3): 4 x ConvModule( - (conv): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1)) - ) - ) - ) - init_cfg={'type': 'Xavier', 'layer': 'Conv2d', 'distribution': 'uniform'} - (head): Sparse4DHead( - (instance_bank): InstanceBank( - (anchor_handler): SparseBox3DKeyPointsGenerator() - ) - (anchor_encoder): SparseBox3DEncoder( - (pos_fc): Sequential( - (0): Linear(in_features=3, out_features=128, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((128,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=128, out_features=128, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((128,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=128, out_features=128, bias=True) - (7): ReLU(inplace=True) - (8): LayerNorm((128,), eps=1e-05, elementwise_affine=True) - (9): Linear(in_features=128, out_features=128, bias=True) - (10): ReLU(inplace=True) - (11): LayerNorm((128,), eps=1e-05, elementwise_affine=True) - ) - (size_fc): Sequential( - (0): Linear(in_features=3, out_features=32, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=32, out_features=32, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=32, out_features=32, bias=True) - (7): ReLU(inplace=True) - (8): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (9): Linear(in_features=32, out_features=32, bias=True) - (10): ReLU(inplace=True) - (11): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - ) - (yaw_fc): Sequential( - (0): Linear(in_features=2, out_features=32, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=32, out_features=32, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=32, out_features=32, bias=True) - (7): ReLU(inplace=True) - (8): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (9): Linear(in_features=32, out_features=32, bias=True) - (10): ReLU(inplace=True) - (11): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - ) - (vel_fc): Sequential( - (0): Linear(in_features=3, out_features=64, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((64,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=64, out_features=64, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((64,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=64, out_features=64, bias=True) - (7): ReLU(inplace=True) - (8): LayerNorm((64,), eps=1e-05, elementwise_affine=True) - (9): Linear(in_features=64, out_features=64, bias=True) - (10): ReLU(inplace=True) - (11): LayerNorm((64,), eps=1e-05, elementwise_affine=True) - ) - ) - (loss_cls): FocalLoss() - (loss_reg): SparseBox3DLoss( - (loss_box): L1Loss() - (loss_cns): CrossEntropyLoss(avg_non_ignore=False) - (loss_yns): GaussianFocalLoss() - ) - (layers): ModuleList( - (0): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (1): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (4-5): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (6): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (7): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (8): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (11-12): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (13): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (14): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (15): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (16): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (17): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (18-19): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (20): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (21): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (22): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (23): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (24): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (25-26): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (27): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (28): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (29): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (30): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (31): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (32-33): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (34): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (35): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (36): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (37): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (38): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - ) - (fc_before): Linear(in_features=256, out_features=512, bias=False) - (fc_after): Linear(in_features=512, out_features=256, bias=False) - ) - (depth_branch): DenseDepthNet( - (depth_layers): ModuleList( - (0-2): 3 x Conv2d(256, 1, kernel_size=(1, 1), stride=(1, 1)) - ) - ) - (grid_mask): GridMask() -) -2026-04-03 16:38:50,090 - mmdet - INFO - Start running, host: root@bw61, work_dir: /workspace/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704 -2026-04-03 16:38:50,090 - mmdet - INFO - Hooks will be executed in the following order: -before_run: -(VERY_HIGH ) CosineAnnealingLrUpdaterHook -(ABOVE_NORMAL) Fp16OptimizerHook -(NORMAL ) CheckpointHook -(NORMAL ) CustomDistEvalHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -before_train_epoch: -(VERY_HIGH ) CosineAnnealingLrUpdaterHook -(NORMAL ) CustomDistEvalHook -(LOW ) IterTimerHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -before_train_iter: -(VERY_HIGH ) CosineAnnealingLrUpdaterHook -(NORMAL ) CustomDistEvalHook -(LOW ) IterTimerHook - -------------------- -after_train_iter: -(ABOVE_NORMAL) Fp16OptimizerHook -(NORMAL ) CheckpointHook -(NORMAL ) CustomDistEvalHook -(NORMAL ) ProfilerHook -(LOW ) IterTimerHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -after_train_epoch: -(NORMAL ) CheckpointHook -(NORMAL ) CustomDistEvalHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -before_val_epoch: -(LOW ) IterTimerHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -before_val_iter: -(LOW ) IterTimerHook - -------------------- -after_val_iter: -(LOW ) IterTimerHook - -------------------- -after_val_epoch: -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -after_run: -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -2026-04-03 16:38:50,091 - mmdet - INFO - workflow: [('train', 1)], max: 23400 iters -2026-04-03 16:38:50,092 - mmdet - INFO - Checkpoints will be saved to /workspace/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704 by HardDiskBackend. diff --git a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20260403_163837.log.json b/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20260403_163837.log.json deleted file mode 100644 index c0403e68d56949db5d99dc6457c849b6d349e806..0000000000000000000000000000000000000000 --- a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20260403_163837.log.json +++ /dev/null @@ -1 +0,0 @@ -{"env_info": "sys.platform: linux\nPython: 3.10.12 (main, Aug 15 2025, 14:32:43) [GCC 11.4.0]\nCUDA available: True\nGPU 0,1,2,3,4,5,6,7: BW1000_H\nCUDA_HOME: /opt/dtk\nNVCC: Not Available\nGCC: x86_64-linux-gnu-gcc (Ubuntu 11.4.0-1ubuntu1~22.04) 11.4.0\nPyTorch: 2.5.1\nPyTorch compiling details: PyTorch built with:\n - GCC 10.3\n - C++ Version: 201703\n - Intel(R) Math Kernel Library Version 2020.0.4 Product Build 20200917 for Intel(R) 64 architecture applications\n - OpenMP 201511 (a.k.a. OpenMP 4.5)\n - LAPACK is enabled (usually provided by MKL)\n - NNPACK is enabled\n - CPU capability usage: AVX512\n - HIP Runtime 6.3.25521\n - MIOpen 2.18.0\n - Magma 2.8.0\n - Build settings: BLAS_INFO=mkl, BUILD_TYPE=Release, CXX_COMPILER=/opt/rh/gcc-toolset-10/root/usr/bin/c++, CXX_FLAGS= -D_GLIBCXX_USE_CXX11_ABI=1 -fvisibility-inlines-hidden -DUSE_PTHREADPOOL -DNDEBUG -DUSE_KINETO -DLIBKINETO_NOCUPTI -DLIBKINETO_NOXPUPTI=ON -DUSE_FBGEMM -DUSE_PYTORCH_QNNPACK -DUSE_XNNPACK -DSYMBOLICATE_MOBILE_DEBUG_HANDLE -O2 -fPIC -Wall -Wextra -Werror=return-type -Werror=non-virtual-dtor -Werror=bool-operation -Wnarrowing -Wno-missing-field-initializers -Wno-type-limits -Wno-array-bounds -Wno-unknown-pragmas -Wno-unused-parameter -Wno-strict-overflow -Wno-strict-aliasing -Wno-stringop-overflow -Wsuggest-override -Wno-psabi -Wno-error=old-style-cast -Wno-missing-braces -fdiagnostics-color=always -faligned-new -Wno-unused-but-set-variable -Wno-maybe-uninitialized -fno-math-errno -fno-trapping-math -Werror=format -Wno-stringop-overflow, FORCE_FALLBACK_CUDA_MPI=1, LAPACK_INFO=mkl, PERF_WITH_AVX=1, PERF_WITH_AVX2=1, TORCH_VERSION=2.5.1, USE_CUDA=0, USE_CUDNN=OFF, USE_CUSPARSELT=OFF, USE_EXCEPTION_PTR=1, USE_GFLAGS=1, USE_GLOG=1, USE_GLOO=1, USE_MKL=ON, USE_MKLDNN=0, USE_MPI=1, USE_NCCL=1, USE_NNPACK=ON, USE_OPENMP=1, USE_ROCM=ON, USE_ROCM_KERNEL_ASSERT=OFF, \n\nTorchVision: 0.20.1\nOpenCV: 4.12.0\nMMCV: 1.6.1\nMMCV Compiler: GCC 10.3\nMMCV CUDA Compiler: rocm not available\nMMDetection: 2.25.1+c41df4b", "config": "plugin = True\nplugin_dir = 'projects/mmdet3d_plugin/'\ndist_params = dict(backend='nccl')\nlog_level = 'INFO'\nwork_dir = './work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704'\ntotal_batch_size = 120\nnum_gpus = 8\nbatch_size = 15\nnum_iters_per_epoch = 234\nnum_epochs = 100\ncheckpoint_epoch_interval = 20\ncheckpoint_config = dict(interval=4680)\nlog_config = dict(\n interval=1,\n hooks=[\n dict(type='TextLoggerHook', by_epoch=False),\n dict(type='TensorboardLoggerHook')\n ])\nload_from = None\nresume_from = None\nworkflow = [('train', 1)]\nfp16 = dict(loss_scale=32.0)\ninput_shape = (704, 256)\ntracking_test = True\ntracking_threshold = 0.2\nclass_names = [\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier',\n 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n]\nnum_classes = 10\nembed_dims = 256\nnum_groups = 8\nnum_decoder = 6\nnum_single_frame_decoder = 1\nuse_deformable_func = True\nstrides = [4, 8, 16, 32]\nnum_levels = 4\nnum_depth_layers = 3\ndrop_out = 0.1\ntemporal = True\ndecouple_attn = True\nwith_quality_estimation = True\nmodel = dict(\n type='Sparse4D',\n use_grid_mask=True,\n use_deformable_func=True,\n img_backbone=dict(\n type='ResNet',\n depth=50,\n num_stages=4,\n frozen_stages=-1,\n norm_eval=False,\n style='pytorch',\n with_cp=True,\n out_indices=(0, 1, 2, 3),\n norm_cfg=dict(type='BN', requires_grad=True),\n pretrained='ckpt/resnet50-19c8e357.pth'),\n img_neck=dict(\n type='FPN',\n num_outs=4,\n start_level=0,\n out_channels=256,\n add_extra_convs='on_output',\n relu_before_extra_convs=True,\n in_channels=[256, 512, 1024, 2048]),\n depth_branch=dict(\n type='DenseDepthNet',\n embed_dims=256,\n num_depth_layers=3,\n loss_weight=0.2),\n head=dict(\n type='Sparse4DHead',\n cls_threshold_to_reg=0.05,\n decouple_attn=True,\n instance_bank=dict(\n type='InstanceBank',\n num_anchor=900,\n embed_dims=256,\n anchor='nuscenes_kmeans900.npy',\n anchor_handler=dict(type='SparseBox3DKeyPointsGenerator'),\n num_temp_instances=600,\n confidence_decay=0.6,\n feat_grad=False),\n anchor_encoder=dict(\n type='SparseBox3DEncoder',\n vel_dims=3,\n embed_dims=[128, 32, 32, 64],\n mode='cat',\n output_fc=False,\n in_loops=1,\n out_loops=4),\n num_single_frame_decoder=1,\n operation_order=[\n 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm',\n 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm',\n 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm',\n 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm',\n 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm',\n 'deformable', 'ffn', 'norm', 'refine'\n ],\n temp_graph_model=dict(\n type='MultiheadAttention',\n embed_dims=512,\n num_heads=8,\n batch_first=True,\n dropout=0.1),\n graph_model=dict(\n type='MultiheadAttention',\n embed_dims=512,\n num_heads=8,\n batch_first=True,\n dropout=0.1),\n norm_layer=dict(type='LN', normalized_shape=256),\n ffn=dict(\n type='AsymmetricFFN',\n in_channels=512,\n pre_norm=dict(type='LN'),\n embed_dims=256,\n feedforward_channels=1024,\n num_fcs=2,\n ffn_drop=0.1,\n act_cfg=dict(type='ReLU', inplace=True)),\n deformable_model=dict(\n type='DeformableFeatureAggregation',\n embed_dims=256,\n num_groups=8,\n num_levels=4,\n num_cams=6,\n attn_drop=0.15,\n use_deformable_func=True,\n use_camera_embed=True,\n residual_mode='cat',\n kps_generator=dict(\n type='SparseBox3DKeyPointsGenerator',\n num_learnable_pts=6,\n fix_scale=[[0, 0, 0], [0.45, 0, 0], [-0.45, 0, 0],\n [0, 0.45, 0], [0, -0.45, 0], [0, 0, 0.45],\n [0, 0, -0.45]])),\n refine_layer=dict(\n type='SparseBox3DRefinementModule',\n embed_dims=256,\n num_cls=10,\n refine_yaw=True,\n with_quality_estimation=True),\n sampler=dict(\n type='SparseBox3DTarget',\n num_dn_groups=5,\n num_temp_dn_groups=3,\n dn_noise_scale=[2.0, 2.0, 2.0, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5],\n max_dn_gt=32,\n add_neg_dn=True,\n cls_weight=2.0,\n box_weight=0.25,\n reg_weights=[2.0, 2.0, 2.0, 0.5, 0.5, 0.5, 0.0, 0.0, 0.0, 0.0],\n cls_wise_reg_weights=dict(\n {9: [2.0, 2.0, 2.0, 1.0, 1.0, 1.0, 0.0, 0.0, 1.0, 1.0]})),\n loss_cls=dict(\n type='FocalLoss',\n use_sigmoid=True,\n gamma=2.0,\n alpha=0.25,\n loss_weight=2.0),\n loss_reg=dict(\n type='SparseBox3DLoss',\n loss_box=dict(type='L1Loss', loss_weight=0.25),\n loss_centerness=dict(type='CrossEntropyLoss', use_sigmoid=True),\n loss_yawness=dict(type='GaussianFocalLoss'),\n cls_allow_reverse=[5]),\n decoder=dict(type='SparseBox3DDecoder'),\n reg_weights=[2.0, 2.0, 2.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0]))\ndataset_type = 'NuScenes3DDetTrackDataset'\ndata_root = 'data/nuscenes/'\nanno_root = 'data/nuscenes_anno_pkls/'\nfile_client_args = dict(backend='disk')\nimg_norm_cfg = dict(\n mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375], to_rgb=True)\ntrain_pipeline = [\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(\n type='LoadPointsFromFile',\n coord_type='LIDAR',\n load_dim=5,\n use_dim=5,\n file_client_args=dict(backend='disk')),\n dict(type='ResizeCropFlipImage'),\n dict(type='MultiScaleDepthMapGenerator', downsample=[4, 8, 16]),\n dict(type='BBoxRotation'),\n dict(type='PhotoMetricDistortionMultiViewImage'),\n dict(\n type='NormalizeMultiviewImage',\n mean=[123.675, 116.28, 103.53],\n std=[58.395, 57.12, 57.375],\n to_rgb=True),\n dict(\n type='CircleObjectRangeFilter',\n class_dist_thred=[55, 55, 55, 55, 55, 55, 55, 55, 55, 55]),\n dict(\n type='InstanceNameFilter',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ]),\n dict(type='NuScenesSparse4DAdaptor'),\n dict(\n type='Collect',\n keys=[\n 'img', 'timestamp', 'projection_mat', 'image_wh', 'gt_depth',\n 'focal', 'gt_bboxes_3d', 'gt_labels_3d'\n ],\n meta_keys=['T_global', 'T_global_inv', 'timestamp', 'instance_id'])\n]\ntest_pipeline = [\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(type='ResizeCropFlipImage'),\n dict(\n type='NormalizeMultiviewImage',\n mean=[123.675, 116.28, 103.53],\n std=[58.395, 57.12, 57.375],\n to_rgb=True),\n dict(type='NuScenesSparse4DAdaptor'),\n dict(\n type='Collect',\n keys=['img', 'timestamp', 'projection_mat', 'image_wh'],\n meta_keys=['T_global', 'T_global_inv', 'timestamp'])\n]\ninput_modality = dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False)\ndata_basic_config = dict(\n type='NuScenes3DDetTrackDataset',\n data_root='data/nuscenes/',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier',\n 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ],\n modality=dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False),\n version='v1.0-trainval')\ndata_aug_conf = dict(\n resize_lim=(0.4, 0.47),\n final_dim=(256, 704),\n bot_pct_lim=(0.0, 0.0),\n rot_lim=(-5.4, 5.4),\n H=900,\n W=1600,\n rand_flip=True,\n rot3d_range=[-0.3925, 0.3925])\ndata = dict(\n samples_per_gpu=15,\n workers_per_gpu=15,\n train=dict(\n type='NuScenes3DDetTrackDataset',\n data_root='data/nuscenes/',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ],\n modality=dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False),\n version='v1.0-trainval',\n ann_file='data/nuscenes_anno_pkls/nuscenes_infos_train.pkl',\n pipeline=[\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(\n type='LoadPointsFromFile',\n coord_type='LIDAR',\n load_dim=5,\n use_dim=5,\n file_client_args=dict(backend='disk')),\n dict(type='ResizeCropFlipImage'),\n dict(type='MultiScaleDepthMapGenerator', downsample=[4, 8, 16]),\n dict(type='BBoxRotation'),\n dict(type='PhotoMetricDistortionMultiViewImage'),\n dict(\n type='NormalizeMultiviewImage',\n mean=[123.675, 116.28, 103.53],\n std=[58.395, 57.12, 57.375],\n to_rgb=True),\n dict(\n type='CircleObjectRangeFilter',\n class_dist_thred=[55, 55, 55, 55, 55, 55, 55, 55, 55, 55]),\n dict(\n type='InstanceNameFilter',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian',\n 'traffic_cone'\n ]),\n dict(type='NuScenesSparse4DAdaptor'),\n dict(\n type='Collect',\n keys=[\n 'img', 'timestamp', 'projection_mat', 'image_wh',\n 'gt_depth', 'focal', 'gt_bboxes_3d', 'gt_labels_3d'\n ],\n meta_keys=[\n 'T_global', 'T_global_inv', 'timestamp', 'instance_id'\n ])\n ],\n test_mode=False,\n data_aug_conf=dict(\n resize_lim=(0.4, 0.47),\n final_dim=(256, 704),\n bot_pct_lim=(0.0, 0.0),\n rot_lim=(-5.4, 5.4),\n H=900,\n W=1600,\n rand_flip=True,\n rot3d_range=[-0.3925, 0.3925]),\n with_seq_flag=True,\n sequences_split_num=2,\n keep_consistent_seq_aug=True),\n val=dict(\n type='NuScenes3DDetTrackDataset',\n data_root='data/nuscenes/',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ],\n modality=dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False),\n version='v1.0-trainval',\n ann_file='data/nuscenes_anno_pkls/nuscenes_infos_val.pkl',\n pipeline=[\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(type='ResizeCropFlipImage'),\n dict(\n type='NormalizeMultiviewImage',\n mean=[123.675, 116.28, 103.53],\n std=[58.395, 57.12, 57.375],\n to_rgb=True),\n dict(type='NuScenesSparse4DAdaptor'),\n dict(\n type='Collect',\n keys=['img', 'timestamp', 'projection_mat', 'image_wh'],\n meta_keys=['T_global', 'T_global_inv', 'timestamp'])\n ],\n data_aug_conf=dict(\n resize_lim=(0.4, 0.47),\n final_dim=(256, 704),\n bot_pct_lim=(0.0, 0.0),\n rot_lim=(-5.4, 5.4),\n H=900,\n W=1600,\n rand_flip=True,\n rot3d_range=[-0.3925, 0.3925]),\n test_mode=True,\n tracking=True,\n tracking_threshold=0.2),\n test=dict(\n type='NuScenes3DDetTrackDataset',\n data_root='data/nuscenes/',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ],\n modality=dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False),\n version='v1.0-trainval',\n ann_file='data/nuscenes_anno_pkls/nuscenes_infos_val.pkl',\n pipeline=[\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(type='ResizeCropFlipImage'),\n dict(\n type='NormalizeMultiviewImage',\n mean=[123.675, 116.28, 103.53],\n std=[58.395, 57.12, 57.375],\n to_rgb=True),\n dict(type='NuScenesSparse4DAdaptor'),\n dict(\n type='Collect',\n keys=['img', 'timestamp', 'projection_mat', 'image_wh'],\n meta_keys=['T_global', 'T_global_inv', 'timestamp'])\n ],\n data_aug_conf=dict(\n resize_lim=(0.4, 0.47),\n final_dim=(256, 704),\n bot_pct_lim=(0.0, 0.0),\n rot_lim=(-5.4, 5.4),\n H=900,\n W=1600,\n rand_flip=True,\n rot3d_range=[-0.3925, 0.3925]),\n test_mode=True,\n tracking=True,\n tracking_threshold=0.2))\noptimizer = dict(\n type='AdamW',\n lr=0.0006,\n weight_decay=0.001,\n paramwise_cfg=dict(custom_keys=dict(img_backbone=dict(lr_mult=0.5))))\noptimizer_config = dict(grad_clip=dict(max_norm=25, norm_type=2))\nlr_config = dict(\n policy='CosineAnnealing',\n warmup='linear',\n warmup_iters=500,\n warmup_ratio=0.3333333333333333,\n min_lr_ratio=0.001)\nrunner = dict(type='IterBasedRunner', max_iters=23400)\nvis_pipeline = [\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(type='Collect', keys=['img'], meta_keys=['timestamp', 'lidar2img'])\n]\nevaluation = dict(\n interval=4680,\n pipeline=[\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(\n type='Collect', keys=['img'], meta_keys=['timestamp', 'lidar2img'])\n ])\ngpu_ids = range(0, 8)\n", "seed": 0, "exp_name": "sparse4dv3_temporal_r50_1x8_bs6_256x704.py"} diff --git a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20260403_164418.log b/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20260403_164418.log deleted file mode 100644 index c3c46eb5b18d9aaca9e12b011acd23ffe513952b..0000000000000000000000000000000000000000 --- a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20260403_164418.log +++ /dev/null @@ -1,3221 +0,0 @@ -2026-04-03 16:44:18,166 - mmdet - INFO - Environment info: ------------------------------------------------------------- -sys.platform: linux -Python: 3.10.12 (main, Aug 15 2025, 14:32:43) [GCC 11.4.0] -CUDA available: True -GPU 0,1,2,3,4,5,6,7: BW1000_H -CUDA_HOME: /opt/dtk -NVCC: Not Available -GCC: x86_64-linux-gnu-gcc (Ubuntu 11.4.0-1ubuntu1~22.04) 11.4.0 -PyTorch: 2.5.1 -PyTorch compiling details: PyTorch built with: - - GCC 10.3 - - C++ Version: 201703 - - Intel(R) Math Kernel Library Version 2020.0.4 Product Build 20200917 for Intel(R) 64 architecture applications - - OpenMP 201511 (a.k.a. OpenMP 4.5) - - LAPACK is enabled (usually provided by MKL) - - NNPACK is enabled - - CPU capability usage: AVX512 - - HIP Runtime 6.3.25521 - - MIOpen 2.18.0 - - Magma 2.8.0 - - Build settings: BLAS_INFO=mkl, BUILD_TYPE=Release, CXX_COMPILER=/opt/rh/gcc-toolset-10/root/usr/bin/c++, CXX_FLAGS= -D_GLIBCXX_USE_CXX11_ABI=1 -fvisibility-inlines-hidden -DUSE_PTHREADPOOL -DNDEBUG -DUSE_KINETO -DLIBKINETO_NOCUPTI -DLIBKINETO_NOXPUPTI=ON -DUSE_FBGEMM -DUSE_PYTORCH_QNNPACK -DUSE_XNNPACK -DSYMBOLICATE_MOBILE_DEBUG_HANDLE -O2 -fPIC -Wall -Wextra -Werror=return-type -Werror=non-virtual-dtor -Werror=bool-operation -Wnarrowing -Wno-missing-field-initializers -Wno-type-limits -Wno-array-bounds -Wno-unknown-pragmas -Wno-unused-parameter -Wno-strict-overflow -Wno-strict-aliasing -Wno-stringop-overflow -Wsuggest-override -Wno-psabi -Wno-error=old-style-cast -Wno-missing-braces -fdiagnostics-color=always -faligned-new -Wno-unused-but-set-variable -Wno-maybe-uninitialized -fno-math-errno -fno-trapping-math -Werror=format -Wno-stringop-overflow, FORCE_FALLBACK_CUDA_MPI=1, LAPACK_INFO=mkl, PERF_WITH_AVX=1, PERF_WITH_AVX2=1, TORCH_VERSION=2.5.1, USE_CUDA=0, USE_CUDNN=OFF, USE_CUSPARSELT=OFF, USE_EXCEPTION_PTR=1, USE_GFLAGS=1, USE_GLOG=1, USE_GLOO=1, USE_MKL=ON, USE_MKLDNN=0, USE_MPI=1, USE_NCCL=1, USE_NNPACK=ON, USE_OPENMP=1, USE_ROCM=ON, USE_ROCM_KERNEL_ASSERT=OFF, - -TorchVision: 0.20.1 -OpenCV: 4.12.0 -MMCV: 1.6.1 -MMCV Compiler: GCC 10.3 -MMCV CUDA Compiler: rocm not available -MMDetection: 2.25.1+c41df4b ------------------------------------------------------------- - -2026-04-03 16:44:18,871 - mmdet - INFO - Distributed training: True -2026-04-03 16:44:19,545 - mmdet - INFO - Config: -plugin = True -plugin_dir = 'projects/mmdet3d_plugin/' -dist_params = dict(backend='nccl') -log_level = 'INFO' -work_dir = './work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704' -total_batch_size = 120 -num_gpus = 8 -batch_size = 15 -num_iters_per_epoch = 234 -num_epochs = 100 -checkpoint_epoch_interval = 20 -checkpoint_config = dict(interval=4680) -log_config = dict( - interval=1, - hooks=[ - dict(type='TextLoggerHook', by_epoch=False), - dict(type='TensorboardLoggerHook') - ]) -load_from = None -resume_from = None -workflow = [('train', 1)] -fp16 = dict(loss_scale=32.0) -input_shape = (704, 256) -tracking_test = True -tracking_threshold = 0.2 -class_names = [ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier', - 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' -] -num_classes = 10 -embed_dims = 256 -num_groups = 8 -num_decoder = 6 -num_single_frame_decoder = 1 -use_deformable_func = True -strides = [4, 8, 16, 32] -num_levels = 4 -num_depth_layers = 3 -drop_out = 0.1 -temporal = True -decouple_attn = True -with_quality_estimation = True -model = dict( - type='Sparse4D', - use_grid_mask=True, - use_deformable_func=True, - img_backbone=dict( - type='ResNet', - depth=50, - num_stages=4, - frozen_stages=-1, - norm_eval=False, - style='pytorch', - with_cp=True, - out_indices=(0, 1, 2, 3), - norm_cfg=dict(type='BN', requires_grad=True), - pretrained='ckpt/resnet50-19c8e357.pth'), - img_neck=dict( - type='FPN', - num_outs=4, - start_level=0, - out_channels=256, - add_extra_convs='on_output', - relu_before_extra_convs=True, - in_channels=[256, 512, 1024, 2048]), - depth_branch=dict( - type='DenseDepthNet', - embed_dims=256, - num_depth_layers=3, - loss_weight=0.2), - head=dict( - type='Sparse4DHead', - cls_threshold_to_reg=0.05, - decouple_attn=True, - instance_bank=dict( - type='InstanceBank', - num_anchor=900, - embed_dims=256, - anchor='nuscenes_kmeans900.npy', - anchor_handler=dict(type='SparseBox3DKeyPointsGenerator'), - num_temp_instances=600, - confidence_decay=0.6, - feat_grad=False), - anchor_encoder=dict( - type='SparseBox3DEncoder', - vel_dims=3, - embed_dims=[128, 32, 32, 64], - mode='cat', - output_fc=False, - in_loops=1, - out_loops=4), - num_single_frame_decoder=1, - operation_order=[ - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine' - ], - temp_graph_model=dict( - type='MultiheadAttention', - embed_dims=512, - num_heads=8, - batch_first=True, - dropout=0.1), - graph_model=dict( - type='MultiheadAttention', - embed_dims=512, - num_heads=8, - batch_first=True, - dropout=0.1), - norm_layer=dict(type='LN', normalized_shape=256), - ffn=dict( - type='AsymmetricFFN', - in_channels=512, - pre_norm=dict(type='LN'), - embed_dims=256, - feedforward_channels=1024, - num_fcs=2, - ffn_drop=0.1, - act_cfg=dict(type='ReLU', inplace=True)), - deformable_model=dict( - type='DeformableFeatureAggregation', - embed_dims=256, - num_groups=8, - num_levels=4, - num_cams=6, - attn_drop=0.15, - use_deformable_func=True, - use_camera_embed=True, - residual_mode='cat', - kps_generator=dict( - type='SparseBox3DKeyPointsGenerator', - num_learnable_pts=6, - fix_scale=[[0, 0, 0], [0.45, 0, 0], [-0.45, 0, 0], - [0, 0.45, 0], [0, -0.45, 0], [0, 0, 0.45], - [0, 0, -0.45]])), - refine_layer=dict( - type='SparseBox3DRefinementModule', - embed_dims=256, - num_cls=10, - refine_yaw=True, - with_quality_estimation=True), - sampler=dict( - type='SparseBox3DTarget', - num_dn_groups=5, - num_temp_dn_groups=3, - dn_noise_scale=[2.0, 2.0, 2.0, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5], - max_dn_gt=32, - add_neg_dn=True, - cls_weight=2.0, - box_weight=0.25, - reg_weights=[2.0, 2.0, 2.0, 0.5, 0.5, 0.5, 0.0, 0.0, 0.0, 0.0], - cls_wise_reg_weights=dict( - {9: [2.0, 2.0, 2.0, 1.0, 1.0, 1.0, 0.0, 0.0, 1.0, 1.0]})), - loss_cls=dict( - type='FocalLoss', - use_sigmoid=True, - gamma=2.0, - alpha=0.25, - loss_weight=2.0), - loss_reg=dict( - type='SparseBox3DLoss', - loss_box=dict(type='L1Loss', loss_weight=0.25), - loss_centerness=dict(type='CrossEntropyLoss', use_sigmoid=True), - loss_yawness=dict(type='GaussianFocalLoss'), - cls_allow_reverse=[5]), - decoder=dict(type='SparseBox3DDecoder'), - reg_weights=[2.0, 2.0, 2.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0])) -dataset_type = 'NuScenes3DDetTrackDataset' -data_root = 'data/nuscenes/' -anno_root = 'data/nuscenes_anno_pkls/' -file_client_args = dict(backend='disk') -img_norm_cfg = dict( - mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375], to_rgb=True) -train_pipeline = [ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict( - type='LoadPointsFromFile', - coord_type='LIDAR', - load_dim=5, - use_dim=5, - file_client_args=dict(backend='disk')), - dict(type='ResizeCropFlipImage'), - dict(type='MultiScaleDepthMapGenerator', downsample=[4, 8, 16]), - dict(type='BBoxRotation'), - dict(type='PhotoMetricDistortionMultiViewImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict( - type='CircleObjectRangeFilter', - class_dist_thred=[55, 55, 55, 55, 55, 55, 55, 55, 55, 55]), - dict( - type='InstanceNameFilter', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ]), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=[ - 'img', 'timestamp', 'projection_mat', 'image_wh', 'gt_depth', - 'focal', 'gt_bboxes_3d', 'gt_labels_3d' - ], - meta_keys=['T_global', 'T_global_inv', 'timestamp', 'instance_id']) -] -test_pipeline = [ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='ResizeCropFlipImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=['img', 'timestamp', 'projection_mat', 'image_wh'], - meta_keys=['T_global', 'T_global_inv', 'timestamp']) -] -input_modality = dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False) -data_basic_config = dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier', - 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval') -data_aug_conf = dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]) -data = dict( - samples_per_gpu=15, - workers_per_gpu=15, - train=dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval', - ann_file='data/nuscenes_anno_pkls/nuscenes_infos_train.pkl', - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict( - type='LoadPointsFromFile', - coord_type='LIDAR', - load_dim=5, - use_dim=5, - file_client_args=dict(backend='disk')), - dict(type='ResizeCropFlipImage'), - dict(type='MultiScaleDepthMapGenerator', downsample=[4, 8, 16]), - dict(type='BBoxRotation'), - dict(type='PhotoMetricDistortionMultiViewImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict( - type='CircleObjectRangeFilter', - class_dist_thred=[55, 55, 55, 55, 55, 55, 55, 55, 55, 55]), - dict( - type='InstanceNameFilter', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', - 'traffic_cone' - ]), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=[ - 'img', 'timestamp', 'projection_mat', 'image_wh', - 'gt_depth', 'focal', 'gt_bboxes_3d', 'gt_labels_3d' - ], - meta_keys=[ - 'T_global', 'T_global_inv', 'timestamp', 'instance_id' - ]) - ], - test_mode=False, - data_aug_conf=dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]), - with_seq_flag=True, - sequences_split_num=2, - keep_consistent_seq_aug=True), - val=dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval', - ann_file='data/nuscenes_anno_pkls/nuscenes_infos_val.pkl', - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='ResizeCropFlipImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=['img', 'timestamp', 'projection_mat', 'image_wh'], - meta_keys=['T_global', 'T_global_inv', 'timestamp']) - ], - data_aug_conf=dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]), - test_mode=True, - tracking=True, - tracking_threshold=0.2), - test=dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval', - ann_file='data/nuscenes_anno_pkls/nuscenes_infos_val.pkl', - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='ResizeCropFlipImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=['img', 'timestamp', 'projection_mat', 'image_wh'], - meta_keys=['T_global', 'T_global_inv', 'timestamp']) - ], - data_aug_conf=dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]), - test_mode=True, - tracking=True, - tracking_threshold=0.2)) -optimizer = dict( - type='AdamW', - lr=0.0006, - weight_decay=0.001, - paramwise_cfg=dict(custom_keys=dict(img_backbone=dict(lr_mult=0.5)))) -optimizer_config = dict(grad_clip=dict(max_norm=25, norm_type=2)) -lr_config = dict( - policy='CosineAnnealing', - warmup='linear', - warmup_iters=500, - warmup_ratio=0.3333333333333333, - min_lr_ratio=0.001) -runner = dict(type='IterBasedRunner', max_iters=23400) -vis_pipeline = [ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='Collect', keys=['img'], meta_keys=['timestamp', 'lidar2img']) -] -evaluation = dict( - interval=4680, - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict( - type='Collect', keys=['img'], meta_keys=['timestamp', 'lidar2img']) - ]) -gpu_ids = range(0, 8) - -2026-04-03 16:44:19,545 - mmdet - INFO - Set random seed to 0, deterministic: False -2026-04-03 16:44:19,852 - mmdet - INFO - initialize ResNet with init_cfg {'type': 'Pretrained', 'checkpoint': 'ckpt/resnet50-19c8e357.pth'} -2026-04-03 16:44:20,197 - mmdet - INFO - initialize FPN with init_cfg {'type': 'Xavier', 'layer': 'Conv2d', 'distribution': 'uniform'} -Name of parameter - Initialization information - -img_backbone.conv1.weight - torch.Size([64, 3, 7, 7]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.bn1.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.bn1.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.conv1.weight - torch.Size([64, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn1.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn1.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.conv2.weight - torch.Size([64, 64, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn2.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn2.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.conv3.weight - torch.Size([256, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn3.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn3.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.downsample.0.weight - torch.Size([256, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.downsample.1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.downsample.1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.conv1.weight - torch.Size([64, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn1.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn1.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.conv2.weight - torch.Size([64, 64, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn2.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn2.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.conv3.weight - torch.Size([256, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn3.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn3.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.conv1.weight - torch.Size([64, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn1.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn1.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.conv2.weight - torch.Size([64, 64, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn2.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn2.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.conv3.weight - torch.Size([256, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn3.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn3.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.conv1.weight - torch.Size([128, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn1.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn1.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.conv2.weight - torch.Size([128, 128, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn2.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn2.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.conv3.weight - torch.Size([512, 128, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn3.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn3.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.downsample.0.weight - torch.Size([512, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.downsample.1.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.downsample.1.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.conv1.weight - torch.Size([128, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn1.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn1.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.conv2.weight - torch.Size([128, 128, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn2.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn2.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.conv3.weight - torch.Size([512, 128, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn3.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn3.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.conv1.weight - torch.Size([128, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn1.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn1.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.conv2.weight - torch.Size([128, 128, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn2.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn2.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.conv3.weight - torch.Size([512, 128, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn3.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn3.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.conv1.weight - torch.Size([128, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn1.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn1.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.conv2.weight - torch.Size([128, 128, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn2.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn2.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.conv3.weight - torch.Size([512, 128, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn3.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn3.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.conv1.weight - torch.Size([256, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.downsample.0.weight - torch.Size([1024, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.downsample.1.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.downsample.1.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.conv1.weight - torch.Size([512, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn1.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn1.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.conv2.weight - torch.Size([512, 512, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn2.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn2.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.conv3.weight - torch.Size([2048, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn3.weight - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn3.bias - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.downsample.0.weight - torch.Size([2048, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.downsample.1.weight - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.downsample.1.bias - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.conv1.weight - torch.Size([512, 2048, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn1.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn1.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.conv2.weight - torch.Size([512, 512, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn2.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn2.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.conv3.weight - torch.Size([2048, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn3.weight - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn3.bias - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.conv1.weight - torch.Size([512, 2048, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn1.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn1.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.conv2.weight - torch.Size([512, 512, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn2.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn2.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.conv3.weight - torch.Size([2048, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn3.weight - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn3.bias - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_neck.lateral_convs.0.conv.weight - torch.Size([256, 256, 1, 1]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.lateral_convs.0.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.lateral_convs.1.conv.weight - torch.Size([256, 512, 1, 1]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.lateral_convs.1.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.lateral_convs.2.conv.weight - torch.Size([256, 1024, 1, 1]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.lateral_convs.2.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.lateral_convs.3.conv.weight - torch.Size([256, 2048, 1, 1]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.lateral_convs.3.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.fpn_convs.0.conv.weight - torch.Size([256, 256, 3, 3]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.fpn_convs.0.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.fpn_convs.1.conv.weight - torch.Size([256, 256, 3, 3]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.fpn_convs.1.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.fpn_convs.2.conv.weight - torch.Size([256, 256, 3, 3]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.fpn_convs.2.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.fpn_convs.3.conv.weight - torch.Size([256, 256, 3, 3]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.fpn_convs.3.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.instance_bank.anchor - torch.Size([900, 11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.instance_bank.instance_feature - torch.Size([900, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.instance_bank.anchor_handler.fix_scale - torch.Size([1, 3]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.0.weight - torch.Size([128, 3]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.0.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.2.weight - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.2.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.3.weight - torch.Size([128, 128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.3.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.5.weight - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.5.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.6.weight - torch.Size([128, 128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.6.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.8.weight - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.8.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.9.weight - torch.Size([128, 128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.9.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.11.weight - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.11.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.0.weight - torch.Size([32, 3]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.0.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.2.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.2.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.3.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.3.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.5.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.5.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.6.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.6.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.8.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.8.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.9.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.9.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.11.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.11.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.0.weight - torch.Size([32, 2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.0.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.2.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.2.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.3.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.3.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.5.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.5.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.6.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.6.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.8.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.8.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.9.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.9.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.11.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.11.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.0.weight - torch.Size([64, 3]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.0.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.2.weight - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.2.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.3.weight - torch.Size([64, 64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.3.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.5.weight - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.5.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.6.weight - torch.Size([64, 64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.6.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.8.weight - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.8.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.9.weight - torch.Size([64, 64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.9.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.11.weight - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.11.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.1.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.1.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.1.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.1.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.1.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.1.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.1.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.1.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.3.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.4.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.4.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.4.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.4.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.5.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.5.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.5.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.5.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.6.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.6.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.8.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.8.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.8.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.8.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.8.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.8.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.8.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.8.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.10.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.11.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.11.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.11.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.11.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.12.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.12.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.12.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.12.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.13.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.13.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.15.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.15.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.15.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.15.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.15.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.15.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.15.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.15.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.16.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.16.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.17.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.18.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.18.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.18.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.18.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.19.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.19.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.19.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.19.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.20.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.20.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.22.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.22.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.22.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.22.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.22.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.22.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.22.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.22.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.23.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.23.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.24.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.25.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.25.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.25.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.25.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.26.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.26.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.26.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.26.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.27.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.27.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.29.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.29.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.29.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.29.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.29.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.29.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.29.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.29.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.30.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.30.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.31.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.32.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.32.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.32.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.32.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.33.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.33.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.33.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.33.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.34.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.34.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.36.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.36.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.36.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.36.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.36.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.36.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.36.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.36.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.37.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.37.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.38.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.fc_before.weight - torch.Size([512, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.fc_after.weight - torch.Size([256, 512]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.0.weight - torch.Size([1, 256, 1, 1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.0.bias - torch.Size([1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.1.weight - torch.Size([1, 256, 1, 1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.1.bias - torch.Size([1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.2.weight - torch.Size([1, 256, 1, 1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.2.bias - torch.Size([1]): -The value is the same before and after calling `init_weights` of Sparse4D -2026-04-03 16:44:20,295 - mmdet - INFO - Model: -Sparse4D( - (img_backbone): ResNet( - (conv1): Conv2d(3, 64, kernel_size=(7, 7), stride=(2, 2), padding=(3, 3), bias=False) - (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (maxpool): MaxPool2d(kernel_size=3, stride=2, padding=1, dilation=1, ceil_mode=False) - (layer1): ResLayer( - (0): Bottleneck( - (conv1): Conv2d(64, 64, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (downsample): Sequential( - (0): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - ) - ) - (1): Bottleneck( - (conv1): Conv2d(256, 64, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (2): Bottleneck( - (conv1): Conv2d(256, 64, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - ) - (layer2): ResLayer( - (0): Bottleneck( - (conv1): Conv2d(256, 128, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (downsample): Sequential( - (0): Conv2d(256, 512, kernel_size=(1, 1), stride=(2, 2), bias=False) - (1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - ) - ) - (1): Bottleneck( - (conv1): Conv2d(512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (2): Bottleneck( - (conv1): Conv2d(512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (3): Bottleneck( - (conv1): Conv2d(512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - ) - (layer3): ResLayer( - (0): Bottleneck( - (conv1): Conv2d(512, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (downsample): Sequential( - (0): Conv2d(512, 1024, kernel_size=(1, 1), stride=(2, 2), bias=False) - (1): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - ) - ) - (1): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (2): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (3): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (4): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (5): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - ) - (layer4): ResLayer( - (0): Bottleneck( - (conv1): Conv2d(1024, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(512, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (downsample): Sequential( - (0): Conv2d(1024, 2048, kernel_size=(1, 1), stride=(2, 2), bias=False) - (1): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - ) - ) - (1): Bottleneck( - (conv1): Conv2d(2048, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(512, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (2): Bottleneck( - (conv1): Conv2d(2048, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(512, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - ) - ) - init_cfg={'type': 'Pretrained', 'checkpoint': 'ckpt/resnet50-19c8e357.pth'} - (img_neck): FPN( - (lateral_convs): ModuleList( - (0): ConvModule( - (conv): Conv2d(256, 256, kernel_size=(1, 1), stride=(1, 1)) - ) - (1): ConvModule( - (conv): Conv2d(512, 256, kernel_size=(1, 1), stride=(1, 1)) - ) - (2): ConvModule( - (conv): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1)) - ) - (3): ConvModule( - (conv): Conv2d(2048, 256, kernel_size=(1, 1), stride=(1, 1)) - ) - ) - (fpn_convs): ModuleList( - (0-3): 4 x ConvModule( - (conv): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1)) - ) - ) - ) - init_cfg={'type': 'Xavier', 'layer': 'Conv2d', 'distribution': 'uniform'} - (head): Sparse4DHead( - (instance_bank): InstanceBank( - (anchor_handler): SparseBox3DKeyPointsGenerator() - ) - (anchor_encoder): SparseBox3DEncoder( - (pos_fc): Sequential( - (0): Linear(in_features=3, out_features=128, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((128,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=128, out_features=128, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((128,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=128, out_features=128, bias=True) - (7): ReLU(inplace=True) - (8): LayerNorm((128,), eps=1e-05, elementwise_affine=True) - (9): Linear(in_features=128, out_features=128, bias=True) - (10): ReLU(inplace=True) - (11): LayerNorm((128,), eps=1e-05, elementwise_affine=True) - ) - (size_fc): Sequential( - (0): Linear(in_features=3, out_features=32, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=32, out_features=32, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=32, out_features=32, bias=True) - (7): ReLU(inplace=True) - (8): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (9): Linear(in_features=32, out_features=32, bias=True) - (10): ReLU(inplace=True) - (11): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - ) - (yaw_fc): Sequential( - (0): Linear(in_features=2, out_features=32, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=32, out_features=32, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=32, out_features=32, bias=True) - (7): ReLU(inplace=True) - (8): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (9): Linear(in_features=32, out_features=32, bias=True) - (10): ReLU(inplace=True) - (11): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - ) - (vel_fc): Sequential( - (0): Linear(in_features=3, out_features=64, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((64,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=64, out_features=64, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((64,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=64, out_features=64, bias=True) - (7): ReLU(inplace=True) - (8): LayerNorm((64,), eps=1e-05, elementwise_affine=True) - (9): Linear(in_features=64, out_features=64, bias=True) - (10): ReLU(inplace=True) - (11): LayerNorm((64,), eps=1e-05, elementwise_affine=True) - ) - ) - (loss_cls): FocalLoss() - (loss_reg): SparseBox3DLoss( - (loss_box): L1Loss() - (loss_cns): CrossEntropyLoss(avg_non_ignore=False) - (loss_yns): GaussianFocalLoss() - ) - (layers): ModuleList( - (0): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (1): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (4-5): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (6): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (7): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (8): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (11-12): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (13): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (14): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (15): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (16): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (17): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (18-19): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (20): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (21): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (22): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (23): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (24): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (25-26): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (27): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (28): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (29): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (30): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (31): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (32-33): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (34): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (35): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (36): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (37): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (38): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - ) - (fc_before): Linear(in_features=256, out_features=512, bias=False) - (fc_after): Linear(in_features=512, out_features=256, bias=False) - ) - (depth_branch): DenseDepthNet( - (depth_layers): ModuleList( - (0-2): 3 x Conv2d(256, 1, kernel_size=(1, 1), stride=(1, 1)) - ) - ) - (grid_mask): GridMask() -) -2026-04-03 16:44:31,118 - mmdet - INFO - Start running, host: root@bw61, work_dir: /workspace/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704 -2026-04-03 16:44:31,119 - mmdet - INFO - Hooks will be executed in the following order: -before_run: -(VERY_HIGH ) CosineAnnealingLrUpdaterHook -(ABOVE_NORMAL) Fp16OptimizerHook -(NORMAL ) CheckpointHook -(NORMAL ) CustomDistEvalHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -before_train_epoch: -(VERY_HIGH ) CosineAnnealingLrUpdaterHook -(NORMAL ) CustomDistEvalHook -(LOW ) IterTimerHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -before_train_iter: -(VERY_HIGH ) CosineAnnealingLrUpdaterHook -(NORMAL ) CustomDistEvalHook -(LOW ) IterTimerHook - -------------------- -after_train_iter: -(ABOVE_NORMAL) Fp16OptimizerHook -(NORMAL ) CheckpointHook -(NORMAL ) CustomDistEvalHook -(NORMAL ) ProfilerHook -(LOW ) IterTimerHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -after_train_epoch: -(NORMAL ) CheckpointHook -(NORMAL ) CustomDistEvalHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -before_val_epoch: -(LOW ) IterTimerHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -before_val_iter: -(LOW ) IterTimerHook - -------------------- -after_val_iter: -(LOW ) IterTimerHook - -------------------- -after_val_epoch: -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -after_run: -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -2026-04-03 16:44:31,119 - mmdet - INFO - workflow: [('train', 1)], max: 23400 iters -2026-04-03 16:44:31,121 - mmdet - INFO - Checkpoints will be saved to /workspace/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704 by HardDiskBackend. diff --git a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20260403_164418.log.json b/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20260403_164418.log.json deleted file mode 100644 index c0403e68d56949db5d99dc6457c849b6d349e806..0000000000000000000000000000000000000000 --- a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20260403_164418.log.json +++ /dev/null @@ -1 +0,0 @@ -{"env_info": "sys.platform: linux\nPython: 3.10.12 (main, Aug 15 2025, 14:32:43) [GCC 11.4.0]\nCUDA available: True\nGPU 0,1,2,3,4,5,6,7: BW1000_H\nCUDA_HOME: /opt/dtk\nNVCC: Not Available\nGCC: x86_64-linux-gnu-gcc (Ubuntu 11.4.0-1ubuntu1~22.04) 11.4.0\nPyTorch: 2.5.1\nPyTorch compiling details: PyTorch built with:\n - GCC 10.3\n - C++ Version: 201703\n - Intel(R) Math Kernel Library Version 2020.0.4 Product Build 20200917 for Intel(R) 64 architecture applications\n - OpenMP 201511 (a.k.a. OpenMP 4.5)\n - LAPACK is enabled (usually provided by MKL)\n - NNPACK is enabled\n - CPU capability usage: AVX512\n - HIP Runtime 6.3.25521\n - MIOpen 2.18.0\n - Magma 2.8.0\n - Build settings: BLAS_INFO=mkl, BUILD_TYPE=Release, CXX_COMPILER=/opt/rh/gcc-toolset-10/root/usr/bin/c++, CXX_FLAGS= -D_GLIBCXX_USE_CXX11_ABI=1 -fvisibility-inlines-hidden -DUSE_PTHREADPOOL -DNDEBUG -DUSE_KINETO -DLIBKINETO_NOCUPTI -DLIBKINETO_NOXPUPTI=ON -DUSE_FBGEMM -DUSE_PYTORCH_QNNPACK -DUSE_XNNPACK -DSYMBOLICATE_MOBILE_DEBUG_HANDLE -O2 -fPIC -Wall -Wextra -Werror=return-type -Werror=non-virtual-dtor -Werror=bool-operation -Wnarrowing -Wno-missing-field-initializers -Wno-type-limits -Wno-array-bounds -Wno-unknown-pragmas -Wno-unused-parameter -Wno-strict-overflow -Wno-strict-aliasing -Wno-stringop-overflow -Wsuggest-override -Wno-psabi -Wno-error=old-style-cast -Wno-missing-braces -fdiagnostics-color=always -faligned-new -Wno-unused-but-set-variable -Wno-maybe-uninitialized -fno-math-errno -fno-trapping-math -Werror=format -Wno-stringop-overflow, FORCE_FALLBACK_CUDA_MPI=1, LAPACK_INFO=mkl, PERF_WITH_AVX=1, PERF_WITH_AVX2=1, TORCH_VERSION=2.5.1, USE_CUDA=0, USE_CUDNN=OFF, USE_CUSPARSELT=OFF, USE_EXCEPTION_PTR=1, USE_GFLAGS=1, USE_GLOG=1, USE_GLOO=1, USE_MKL=ON, USE_MKLDNN=0, USE_MPI=1, USE_NCCL=1, USE_NNPACK=ON, USE_OPENMP=1, USE_ROCM=ON, USE_ROCM_KERNEL_ASSERT=OFF, \n\nTorchVision: 0.20.1\nOpenCV: 4.12.0\nMMCV: 1.6.1\nMMCV Compiler: GCC 10.3\nMMCV CUDA Compiler: rocm not available\nMMDetection: 2.25.1+c41df4b", "config": "plugin = True\nplugin_dir = 'projects/mmdet3d_plugin/'\ndist_params = dict(backend='nccl')\nlog_level = 'INFO'\nwork_dir = './work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704'\ntotal_batch_size = 120\nnum_gpus = 8\nbatch_size = 15\nnum_iters_per_epoch = 234\nnum_epochs = 100\ncheckpoint_epoch_interval = 20\ncheckpoint_config = dict(interval=4680)\nlog_config = dict(\n interval=1,\n hooks=[\n dict(type='TextLoggerHook', by_epoch=False),\n dict(type='TensorboardLoggerHook')\n ])\nload_from = None\nresume_from = None\nworkflow = [('train', 1)]\nfp16 = dict(loss_scale=32.0)\ninput_shape = (704, 256)\ntracking_test = True\ntracking_threshold = 0.2\nclass_names = [\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier',\n 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n]\nnum_classes = 10\nembed_dims = 256\nnum_groups = 8\nnum_decoder = 6\nnum_single_frame_decoder = 1\nuse_deformable_func = True\nstrides = [4, 8, 16, 32]\nnum_levels = 4\nnum_depth_layers = 3\ndrop_out = 0.1\ntemporal = True\ndecouple_attn = True\nwith_quality_estimation = True\nmodel = dict(\n type='Sparse4D',\n use_grid_mask=True,\n use_deformable_func=True,\n img_backbone=dict(\n type='ResNet',\n depth=50,\n num_stages=4,\n frozen_stages=-1,\n norm_eval=False,\n style='pytorch',\n with_cp=True,\n out_indices=(0, 1, 2, 3),\n norm_cfg=dict(type='BN', requires_grad=True),\n pretrained='ckpt/resnet50-19c8e357.pth'),\n img_neck=dict(\n type='FPN',\n num_outs=4,\n start_level=0,\n out_channels=256,\n add_extra_convs='on_output',\n relu_before_extra_convs=True,\n in_channels=[256, 512, 1024, 2048]),\n depth_branch=dict(\n type='DenseDepthNet',\n embed_dims=256,\n num_depth_layers=3,\n loss_weight=0.2),\n head=dict(\n type='Sparse4DHead',\n cls_threshold_to_reg=0.05,\n decouple_attn=True,\n instance_bank=dict(\n type='InstanceBank',\n num_anchor=900,\n embed_dims=256,\n anchor='nuscenes_kmeans900.npy',\n anchor_handler=dict(type='SparseBox3DKeyPointsGenerator'),\n num_temp_instances=600,\n confidence_decay=0.6,\n feat_grad=False),\n anchor_encoder=dict(\n type='SparseBox3DEncoder',\n vel_dims=3,\n embed_dims=[128, 32, 32, 64],\n mode='cat',\n output_fc=False,\n in_loops=1,\n out_loops=4),\n num_single_frame_decoder=1,\n operation_order=[\n 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm',\n 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm',\n 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm',\n 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm',\n 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm',\n 'deformable', 'ffn', 'norm', 'refine'\n ],\n temp_graph_model=dict(\n type='MultiheadAttention',\n embed_dims=512,\n num_heads=8,\n batch_first=True,\n dropout=0.1),\n graph_model=dict(\n type='MultiheadAttention',\n embed_dims=512,\n num_heads=8,\n batch_first=True,\n dropout=0.1),\n norm_layer=dict(type='LN', normalized_shape=256),\n ffn=dict(\n type='AsymmetricFFN',\n in_channels=512,\n pre_norm=dict(type='LN'),\n embed_dims=256,\n feedforward_channels=1024,\n num_fcs=2,\n ffn_drop=0.1,\n act_cfg=dict(type='ReLU', inplace=True)),\n deformable_model=dict(\n type='DeformableFeatureAggregation',\n embed_dims=256,\n num_groups=8,\n num_levels=4,\n num_cams=6,\n attn_drop=0.15,\n use_deformable_func=True,\n use_camera_embed=True,\n residual_mode='cat',\n kps_generator=dict(\n type='SparseBox3DKeyPointsGenerator',\n num_learnable_pts=6,\n fix_scale=[[0, 0, 0], [0.45, 0, 0], [-0.45, 0, 0],\n [0, 0.45, 0], [0, -0.45, 0], [0, 0, 0.45],\n [0, 0, -0.45]])),\n refine_layer=dict(\n type='SparseBox3DRefinementModule',\n embed_dims=256,\n num_cls=10,\n refine_yaw=True,\n with_quality_estimation=True),\n sampler=dict(\n type='SparseBox3DTarget',\n num_dn_groups=5,\n num_temp_dn_groups=3,\n dn_noise_scale=[2.0, 2.0, 2.0, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5],\n max_dn_gt=32,\n add_neg_dn=True,\n cls_weight=2.0,\n box_weight=0.25,\n reg_weights=[2.0, 2.0, 2.0, 0.5, 0.5, 0.5, 0.0, 0.0, 0.0, 0.0],\n cls_wise_reg_weights=dict(\n {9: [2.0, 2.0, 2.0, 1.0, 1.0, 1.0, 0.0, 0.0, 1.0, 1.0]})),\n loss_cls=dict(\n type='FocalLoss',\n use_sigmoid=True,\n gamma=2.0,\n alpha=0.25,\n loss_weight=2.0),\n loss_reg=dict(\n type='SparseBox3DLoss',\n loss_box=dict(type='L1Loss', loss_weight=0.25),\n loss_centerness=dict(type='CrossEntropyLoss', use_sigmoid=True),\n loss_yawness=dict(type='GaussianFocalLoss'),\n cls_allow_reverse=[5]),\n decoder=dict(type='SparseBox3DDecoder'),\n reg_weights=[2.0, 2.0, 2.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0]))\ndataset_type = 'NuScenes3DDetTrackDataset'\ndata_root = 'data/nuscenes/'\nanno_root = 'data/nuscenes_anno_pkls/'\nfile_client_args = dict(backend='disk')\nimg_norm_cfg = dict(\n mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375], to_rgb=True)\ntrain_pipeline = [\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(\n type='LoadPointsFromFile',\n coord_type='LIDAR',\n load_dim=5,\n use_dim=5,\n file_client_args=dict(backend='disk')),\n dict(type='ResizeCropFlipImage'),\n dict(type='MultiScaleDepthMapGenerator', downsample=[4, 8, 16]),\n dict(type='BBoxRotation'),\n dict(type='PhotoMetricDistortionMultiViewImage'),\n dict(\n type='NormalizeMultiviewImage',\n mean=[123.675, 116.28, 103.53],\n std=[58.395, 57.12, 57.375],\n to_rgb=True),\n dict(\n type='CircleObjectRangeFilter',\n class_dist_thred=[55, 55, 55, 55, 55, 55, 55, 55, 55, 55]),\n dict(\n type='InstanceNameFilter',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ]),\n dict(type='NuScenesSparse4DAdaptor'),\n dict(\n type='Collect',\n keys=[\n 'img', 'timestamp', 'projection_mat', 'image_wh', 'gt_depth',\n 'focal', 'gt_bboxes_3d', 'gt_labels_3d'\n ],\n meta_keys=['T_global', 'T_global_inv', 'timestamp', 'instance_id'])\n]\ntest_pipeline = [\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(type='ResizeCropFlipImage'),\n dict(\n type='NormalizeMultiviewImage',\n mean=[123.675, 116.28, 103.53],\n std=[58.395, 57.12, 57.375],\n to_rgb=True),\n dict(type='NuScenesSparse4DAdaptor'),\n dict(\n type='Collect',\n keys=['img', 'timestamp', 'projection_mat', 'image_wh'],\n meta_keys=['T_global', 'T_global_inv', 'timestamp'])\n]\ninput_modality = dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False)\ndata_basic_config = dict(\n type='NuScenes3DDetTrackDataset',\n data_root='data/nuscenes/',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier',\n 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ],\n modality=dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False),\n version='v1.0-trainval')\ndata_aug_conf = dict(\n resize_lim=(0.4, 0.47),\n final_dim=(256, 704),\n bot_pct_lim=(0.0, 0.0),\n rot_lim=(-5.4, 5.4),\n H=900,\n W=1600,\n rand_flip=True,\n rot3d_range=[-0.3925, 0.3925])\ndata = dict(\n samples_per_gpu=15,\n workers_per_gpu=15,\n train=dict(\n type='NuScenes3DDetTrackDataset',\n data_root='data/nuscenes/',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ],\n modality=dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False),\n version='v1.0-trainval',\n ann_file='data/nuscenes_anno_pkls/nuscenes_infos_train.pkl',\n pipeline=[\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(\n type='LoadPointsFromFile',\n coord_type='LIDAR',\n load_dim=5,\n use_dim=5,\n file_client_args=dict(backend='disk')),\n dict(type='ResizeCropFlipImage'),\n dict(type='MultiScaleDepthMapGenerator', downsample=[4, 8, 16]),\n dict(type='BBoxRotation'),\n dict(type='PhotoMetricDistortionMultiViewImage'),\n dict(\n type='NormalizeMultiviewImage',\n mean=[123.675, 116.28, 103.53],\n std=[58.395, 57.12, 57.375],\n to_rgb=True),\n dict(\n type='CircleObjectRangeFilter',\n class_dist_thred=[55, 55, 55, 55, 55, 55, 55, 55, 55, 55]),\n dict(\n type='InstanceNameFilter',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian',\n 'traffic_cone'\n ]),\n dict(type='NuScenesSparse4DAdaptor'),\n dict(\n type='Collect',\n keys=[\n 'img', 'timestamp', 'projection_mat', 'image_wh',\n 'gt_depth', 'focal', 'gt_bboxes_3d', 'gt_labels_3d'\n ],\n meta_keys=[\n 'T_global', 'T_global_inv', 'timestamp', 'instance_id'\n ])\n ],\n test_mode=False,\n data_aug_conf=dict(\n resize_lim=(0.4, 0.47),\n final_dim=(256, 704),\n bot_pct_lim=(0.0, 0.0),\n rot_lim=(-5.4, 5.4),\n H=900,\n W=1600,\n rand_flip=True,\n rot3d_range=[-0.3925, 0.3925]),\n with_seq_flag=True,\n sequences_split_num=2,\n keep_consistent_seq_aug=True),\n val=dict(\n type='NuScenes3DDetTrackDataset',\n data_root='data/nuscenes/',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ],\n modality=dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False),\n version='v1.0-trainval',\n ann_file='data/nuscenes_anno_pkls/nuscenes_infos_val.pkl',\n pipeline=[\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(type='ResizeCropFlipImage'),\n dict(\n type='NormalizeMultiviewImage',\n mean=[123.675, 116.28, 103.53],\n std=[58.395, 57.12, 57.375],\n to_rgb=True),\n dict(type='NuScenesSparse4DAdaptor'),\n dict(\n type='Collect',\n keys=['img', 'timestamp', 'projection_mat', 'image_wh'],\n meta_keys=['T_global', 'T_global_inv', 'timestamp'])\n ],\n data_aug_conf=dict(\n resize_lim=(0.4, 0.47),\n final_dim=(256, 704),\n bot_pct_lim=(0.0, 0.0),\n rot_lim=(-5.4, 5.4),\n H=900,\n W=1600,\n rand_flip=True,\n rot3d_range=[-0.3925, 0.3925]),\n test_mode=True,\n tracking=True,\n tracking_threshold=0.2),\n test=dict(\n type='NuScenes3DDetTrackDataset',\n data_root='data/nuscenes/',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ],\n modality=dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False),\n version='v1.0-trainval',\n ann_file='data/nuscenes_anno_pkls/nuscenes_infos_val.pkl',\n pipeline=[\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(type='ResizeCropFlipImage'),\n dict(\n type='NormalizeMultiviewImage',\n mean=[123.675, 116.28, 103.53],\n std=[58.395, 57.12, 57.375],\n to_rgb=True),\n dict(type='NuScenesSparse4DAdaptor'),\n dict(\n type='Collect',\n keys=['img', 'timestamp', 'projection_mat', 'image_wh'],\n meta_keys=['T_global', 'T_global_inv', 'timestamp'])\n ],\n data_aug_conf=dict(\n resize_lim=(0.4, 0.47),\n final_dim=(256, 704),\n bot_pct_lim=(0.0, 0.0),\n rot_lim=(-5.4, 5.4),\n H=900,\n W=1600,\n rand_flip=True,\n rot3d_range=[-0.3925, 0.3925]),\n test_mode=True,\n tracking=True,\n tracking_threshold=0.2))\noptimizer = dict(\n type='AdamW',\n lr=0.0006,\n weight_decay=0.001,\n paramwise_cfg=dict(custom_keys=dict(img_backbone=dict(lr_mult=0.5))))\noptimizer_config = dict(grad_clip=dict(max_norm=25, norm_type=2))\nlr_config = dict(\n policy='CosineAnnealing',\n warmup='linear',\n warmup_iters=500,\n warmup_ratio=0.3333333333333333,\n min_lr_ratio=0.001)\nrunner = dict(type='IterBasedRunner', max_iters=23400)\nvis_pipeline = [\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(type='Collect', keys=['img'], meta_keys=['timestamp', 'lidar2img'])\n]\nevaluation = dict(\n interval=4680,\n pipeline=[\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(\n type='Collect', keys=['img'], meta_keys=['timestamp', 'lidar2img'])\n ])\ngpu_ids = range(0, 8)\n", "seed": 0, "exp_name": "sparse4dv3_temporal_r50_1x8_bs6_256x704.py"} diff --git a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20260403_165004.log b/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20260403_165004.log deleted file mode 100644 index 067b66af935189b3b591b6f796bda16faeee942b..0000000000000000000000000000000000000000 --- a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20260403_165004.log +++ /dev/null @@ -1,3227 +0,0 @@ -2026-04-03 16:50:04,816 - mmdet - INFO - Environment info: ------------------------------------------------------------- -sys.platform: linux -Python: 3.10.12 (main, Aug 15 2025, 14:32:43) [GCC 11.4.0] -CUDA available: True -GPU 0,1,2,3,4,5,6,7: BW1000_H -CUDA_HOME: /opt/dtk -NVCC: Not Available -GCC: x86_64-linux-gnu-gcc (Ubuntu 11.4.0-1ubuntu1~22.04) 11.4.0 -PyTorch: 2.5.1 -PyTorch compiling details: PyTorch built with: - - GCC 10.3 - - C++ Version: 201703 - - Intel(R) Math Kernel Library Version 2020.0.4 Product Build 20200917 for Intel(R) 64 architecture applications - - OpenMP 201511 (a.k.a. OpenMP 4.5) - - LAPACK is enabled (usually provided by MKL) - - NNPACK is enabled - - CPU capability usage: AVX512 - - HIP Runtime 6.3.25521 - - MIOpen 2.18.0 - - Magma 2.8.0 - - Build settings: BLAS_INFO=mkl, BUILD_TYPE=Release, CXX_COMPILER=/opt/rh/gcc-toolset-10/root/usr/bin/c++, CXX_FLAGS= -D_GLIBCXX_USE_CXX11_ABI=1 -fvisibility-inlines-hidden -DUSE_PTHREADPOOL -DNDEBUG -DUSE_KINETO -DLIBKINETO_NOCUPTI -DLIBKINETO_NOXPUPTI=ON -DUSE_FBGEMM -DUSE_PYTORCH_QNNPACK -DUSE_XNNPACK -DSYMBOLICATE_MOBILE_DEBUG_HANDLE -O2 -fPIC -Wall -Wextra -Werror=return-type -Werror=non-virtual-dtor -Werror=bool-operation -Wnarrowing -Wno-missing-field-initializers -Wno-type-limits -Wno-array-bounds -Wno-unknown-pragmas -Wno-unused-parameter -Wno-strict-overflow -Wno-strict-aliasing -Wno-stringop-overflow -Wsuggest-override -Wno-psabi -Wno-error=old-style-cast -Wno-missing-braces -fdiagnostics-color=always -faligned-new -Wno-unused-but-set-variable -Wno-maybe-uninitialized -fno-math-errno -fno-trapping-math -Werror=format -Wno-stringop-overflow, FORCE_FALLBACK_CUDA_MPI=1, LAPACK_INFO=mkl, PERF_WITH_AVX=1, PERF_WITH_AVX2=1, TORCH_VERSION=2.5.1, USE_CUDA=0, USE_CUDNN=OFF, USE_CUSPARSELT=OFF, USE_EXCEPTION_PTR=1, USE_GFLAGS=1, USE_GLOG=1, USE_GLOO=1, USE_MKL=ON, USE_MKLDNN=0, USE_MPI=1, USE_NCCL=1, USE_NNPACK=ON, USE_OPENMP=1, USE_ROCM=ON, USE_ROCM_KERNEL_ASSERT=OFF, - -TorchVision: 0.20.1 -OpenCV: 4.12.0 -MMCV: 1.6.1 -MMCV Compiler: GCC 10.3 -MMCV CUDA Compiler: rocm not available -MMDetection: 2.25.1+c41df4b ------------------------------------------------------------- - -2026-04-03 16:50:05,507 - mmdet - INFO - Distributed training: True -2026-04-03 16:50:06,180 - mmdet - INFO - Config: -plugin = True -plugin_dir = 'projects/mmdet3d_plugin/' -dist_params = dict(backend='nccl') -log_level = 'INFO' -work_dir = './work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704' -total_batch_size = 120 -num_gpus = 8 -batch_size = 15 -num_iters_per_epoch = 234 -num_epochs = 100 -checkpoint_epoch_interval = 20 -checkpoint_config = dict(interval=4680) -log_config = dict( - interval=1, - hooks=[ - dict(type='TextLoggerHook', by_epoch=False), - dict(type='TensorboardLoggerHook') - ]) -load_from = None -resume_from = None -workflow = [('train', 1)] -fp16 = dict(loss_scale=32.0) -input_shape = (704, 256) -tracking_test = True -tracking_threshold = 0.2 -class_names = [ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier', - 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' -] -num_classes = 10 -embed_dims = 256 -num_groups = 8 -num_decoder = 6 -num_single_frame_decoder = 1 -use_deformable_func = True -strides = [4, 8, 16, 32] -num_levels = 4 -num_depth_layers = 3 -drop_out = 0.1 -temporal = True -decouple_attn = True -with_quality_estimation = True -model = dict( - type='Sparse4D', - use_grid_mask=True, - use_deformable_func=True, - img_backbone=dict( - type='ResNet', - depth=50, - num_stages=4, - frozen_stages=-1, - norm_eval=False, - style='pytorch', - with_cp=True, - out_indices=(0, 1, 2, 3), - norm_cfg=dict(type='BN', requires_grad=True), - pretrained='ckpt/resnet50-19c8e357.pth'), - img_neck=dict( - type='FPN', - num_outs=4, - start_level=0, - out_channels=256, - add_extra_convs='on_output', - relu_before_extra_convs=True, - in_channels=[256, 512, 1024, 2048]), - depth_branch=dict( - type='DenseDepthNet', - embed_dims=256, - num_depth_layers=3, - loss_weight=0.2), - head=dict( - type='Sparse4DHead', - cls_threshold_to_reg=0.05, - decouple_attn=True, - instance_bank=dict( - type='InstanceBank', - num_anchor=900, - embed_dims=256, - anchor='nuscenes_kmeans900.npy', - anchor_handler=dict(type='SparseBox3DKeyPointsGenerator'), - num_temp_instances=600, - confidence_decay=0.6, - feat_grad=False), - anchor_encoder=dict( - type='SparseBox3DEncoder', - vel_dims=3, - embed_dims=[128, 32, 32, 64], - mode='cat', - output_fc=False, - in_loops=1, - out_loops=4), - num_single_frame_decoder=1, - operation_order=[ - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine' - ], - temp_graph_model=dict( - type='MultiheadAttention', - embed_dims=512, - num_heads=8, - batch_first=True, - dropout=0.1), - graph_model=dict( - type='MultiheadAttention', - embed_dims=512, - num_heads=8, - batch_first=True, - dropout=0.1), - norm_layer=dict(type='LN', normalized_shape=256), - ffn=dict( - type='AsymmetricFFN', - in_channels=512, - pre_norm=dict(type='LN'), - embed_dims=256, - feedforward_channels=1024, - num_fcs=2, - ffn_drop=0.1, - act_cfg=dict(type='ReLU', inplace=True)), - deformable_model=dict( - type='DeformableFeatureAggregation', - embed_dims=256, - num_groups=8, - num_levels=4, - num_cams=6, - attn_drop=0.15, - use_deformable_func=True, - use_camera_embed=True, - residual_mode='cat', - kps_generator=dict( - type='SparseBox3DKeyPointsGenerator', - num_learnable_pts=6, - fix_scale=[[0, 0, 0], [0.45, 0, 0], [-0.45, 0, 0], - [0, 0.45, 0], [0, -0.45, 0], [0, 0, 0.45], - [0, 0, -0.45]])), - refine_layer=dict( - type='SparseBox3DRefinementModule', - embed_dims=256, - num_cls=10, - refine_yaw=True, - with_quality_estimation=True), - sampler=dict( - type='SparseBox3DTarget', - num_dn_groups=5, - num_temp_dn_groups=3, - dn_noise_scale=[2.0, 2.0, 2.0, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5], - max_dn_gt=32, - add_neg_dn=True, - cls_weight=2.0, - box_weight=0.25, - reg_weights=[2.0, 2.0, 2.0, 0.5, 0.5, 0.5, 0.0, 0.0, 0.0, 0.0], - cls_wise_reg_weights=dict( - {9: [2.0, 2.0, 2.0, 1.0, 1.0, 1.0, 0.0, 0.0, 1.0, 1.0]})), - loss_cls=dict( - type='FocalLoss', - use_sigmoid=True, - gamma=2.0, - alpha=0.25, - loss_weight=2.0), - loss_reg=dict( - type='SparseBox3DLoss', - loss_box=dict(type='L1Loss', loss_weight=0.25), - loss_centerness=dict(type='CrossEntropyLoss', use_sigmoid=True), - loss_yawness=dict(type='GaussianFocalLoss'), - cls_allow_reverse=[5]), - decoder=dict(type='SparseBox3DDecoder'), - reg_weights=[2.0, 2.0, 2.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0])) -dataset_type = 'NuScenes3DDetTrackDataset' -data_root = 'data/nuscenes/' -anno_root = 'data/nuscenes_anno_pkls/' -file_client_args = dict(backend='disk') -img_norm_cfg = dict( - mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375], to_rgb=True) -train_pipeline = [ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict( - type='LoadPointsFromFile', - coord_type='LIDAR', - load_dim=5, - use_dim=5, - file_client_args=dict(backend='disk')), - dict(type='ResizeCropFlipImage'), - dict(type='MultiScaleDepthMapGenerator', downsample=[4, 8, 16]), - dict(type='BBoxRotation'), - dict(type='PhotoMetricDistortionMultiViewImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict( - type='CircleObjectRangeFilter', - class_dist_thred=[55, 55, 55, 55, 55, 55, 55, 55, 55, 55]), - dict( - type='InstanceNameFilter', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ]), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=[ - 'img', 'timestamp', 'projection_mat', 'image_wh', 'gt_depth', - 'focal', 'gt_bboxes_3d', 'gt_labels_3d' - ], - meta_keys=['T_global', 'T_global_inv', 'timestamp', 'instance_id']) -] -test_pipeline = [ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='ResizeCropFlipImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=['img', 'timestamp', 'projection_mat', 'image_wh'], - meta_keys=['T_global', 'T_global_inv', 'timestamp']) -] -input_modality = dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False) -data_basic_config = dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier', - 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval') -data_aug_conf = dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]) -data = dict( - samples_per_gpu=15, - workers_per_gpu=15, - train=dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval', - ann_file='data/nuscenes_anno_pkls/nuscenes_infos_train.pkl', - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict( - type='LoadPointsFromFile', - coord_type='LIDAR', - load_dim=5, - use_dim=5, - file_client_args=dict(backend='disk')), - dict(type='ResizeCropFlipImage'), - dict(type='MultiScaleDepthMapGenerator', downsample=[4, 8, 16]), - dict(type='BBoxRotation'), - dict(type='PhotoMetricDistortionMultiViewImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict( - type='CircleObjectRangeFilter', - class_dist_thred=[55, 55, 55, 55, 55, 55, 55, 55, 55, 55]), - dict( - type='InstanceNameFilter', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', - 'traffic_cone' - ]), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=[ - 'img', 'timestamp', 'projection_mat', 'image_wh', - 'gt_depth', 'focal', 'gt_bboxes_3d', 'gt_labels_3d' - ], - meta_keys=[ - 'T_global', 'T_global_inv', 'timestamp', 'instance_id' - ]) - ], - test_mode=False, - data_aug_conf=dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]), - with_seq_flag=True, - sequences_split_num=2, - keep_consistent_seq_aug=True), - val=dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval', - ann_file='data/nuscenes_anno_pkls/nuscenes_infos_val.pkl', - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='ResizeCropFlipImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=['img', 'timestamp', 'projection_mat', 'image_wh'], - meta_keys=['T_global', 'T_global_inv', 'timestamp']) - ], - data_aug_conf=dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]), - test_mode=True, - tracking=True, - tracking_threshold=0.2), - test=dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval', - ann_file='data/nuscenes_anno_pkls/nuscenes_infos_val.pkl', - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='ResizeCropFlipImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=['img', 'timestamp', 'projection_mat', 'image_wh'], - meta_keys=['T_global', 'T_global_inv', 'timestamp']) - ], - data_aug_conf=dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]), - test_mode=True, - tracking=True, - tracking_threshold=0.2)) -optimizer = dict( - type='AdamW', - lr=0.0006, - weight_decay=0.001, - paramwise_cfg=dict(custom_keys=dict(img_backbone=dict(lr_mult=0.5)))) -optimizer_config = dict(grad_clip=dict(max_norm=25, norm_type=2)) -lr_config = dict( - policy='CosineAnnealing', - warmup='linear', - warmup_iters=500, - warmup_ratio=0.3333333333333333, - min_lr_ratio=0.001) -runner = dict(type='IterBasedRunner', max_iters=23400) -vis_pipeline = [ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='Collect', keys=['img'], meta_keys=['timestamp', 'lidar2img']) -] -evaluation = dict( - interval=4680, - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict( - type='Collect', keys=['img'], meta_keys=['timestamp', 'lidar2img']) - ]) -gpu_ids = range(0, 8) - -2026-04-03 16:50:06,180 - mmdet - INFO - Set random seed to 0, deterministic: False -2026-04-03 16:50:06,484 - mmdet - INFO - initialize ResNet with init_cfg {'type': 'Pretrained', 'checkpoint': 'ckpt/resnet50-19c8e357.pth'} -2026-04-03 16:50:06,817 - mmdet - INFO - initialize FPN with init_cfg {'type': 'Xavier', 'layer': 'Conv2d', 'distribution': 'uniform'} -Name of parameter - Initialization information - -img_backbone.conv1.weight - torch.Size([64, 3, 7, 7]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.bn1.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.bn1.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.conv1.weight - torch.Size([64, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn1.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn1.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.conv2.weight - torch.Size([64, 64, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn2.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn2.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.conv3.weight - torch.Size([256, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn3.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.bn3.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.downsample.0.weight - torch.Size([256, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.downsample.1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.0.downsample.1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.conv1.weight - torch.Size([64, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn1.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn1.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.conv2.weight - torch.Size([64, 64, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn2.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn2.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.conv3.weight - torch.Size([256, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn3.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.1.bn3.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.conv1.weight - torch.Size([64, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn1.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn1.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.conv2.weight - torch.Size([64, 64, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn2.weight - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn2.bias - torch.Size([64]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.conv3.weight - torch.Size([256, 64, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn3.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer1.2.bn3.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.conv1.weight - torch.Size([128, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn1.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn1.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.conv2.weight - torch.Size([128, 128, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn2.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn2.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.conv3.weight - torch.Size([512, 128, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn3.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.bn3.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.downsample.0.weight - torch.Size([512, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.downsample.1.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.0.downsample.1.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.conv1.weight - torch.Size([128, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn1.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn1.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.conv2.weight - torch.Size([128, 128, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn2.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn2.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.conv3.weight - torch.Size([512, 128, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn3.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.1.bn3.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.conv1.weight - torch.Size([128, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn1.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn1.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.conv2.weight - torch.Size([128, 128, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn2.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn2.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.conv3.weight - torch.Size([512, 128, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn3.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.2.bn3.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.conv1.weight - torch.Size([128, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn1.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn1.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.conv2.weight - torch.Size([128, 128, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn2.weight - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn2.bias - torch.Size([128]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.conv3.weight - torch.Size([512, 128, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn3.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer2.3.bn3.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.conv1.weight - torch.Size([256, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.downsample.0.weight - torch.Size([1024, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.downsample.1.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.0.downsample.1.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.1.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.2.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.3.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.4.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.conv1.weight - torch.Size([256, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn1.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn1.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.conv2.weight - torch.Size([256, 256, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn2.weight - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn2.bias - torch.Size([256]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.conv3.weight - torch.Size([1024, 256, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn3.weight - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer3.5.bn3.bias - torch.Size([1024]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.conv1.weight - torch.Size([512, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn1.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn1.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.conv2.weight - torch.Size([512, 512, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn2.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn2.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.conv3.weight - torch.Size([2048, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn3.weight - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.bn3.bias - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.downsample.0.weight - torch.Size([2048, 1024, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.downsample.1.weight - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.0.downsample.1.bias - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.conv1.weight - torch.Size([512, 2048, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn1.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn1.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.conv2.weight - torch.Size([512, 512, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn2.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn2.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.conv3.weight - torch.Size([2048, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn3.weight - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.1.bn3.bias - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.conv1.weight - torch.Size([512, 2048, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn1.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn1.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.conv2.weight - torch.Size([512, 512, 3, 3]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn2.weight - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn2.bias - torch.Size([512]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.conv3.weight - torch.Size([2048, 512, 1, 1]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn3.weight - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_backbone.layer4.2.bn3.bias - torch.Size([2048]): -PretrainedInit: load from ckpt/resnet50-19c8e357.pth - -img_neck.lateral_convs.0.conv.weight - torch.Size([256, 256, 1, 1]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.lateral_convs.0.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.lateral_convs.1.conv.weight - torch.Size([256, 512, 1, 1]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.lateral_convs.1.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.lateral_convs.2.conv.weight - torch.Size([256, 1024, 1, 1]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.lateral_convs.2.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.lateral_convs.3.conv.weight - torch.Size([256, 2048, 1, 1]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.lateral_convs.3.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.fpn_convs.0.conv.weight - torch.Size([256, 256, 3, 3]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.fpn_convs.0.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.fpn_convs.1.conv.weight - torch.Size([256, 256, 3, 3]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.fpn_convs.1.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.fpn_convs.2.conv.weight - torch.Size([256, 256, 3, 3]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.fpn_convs.2.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -img_neck.fpn_convs.3.conv.weight - torch.Size([256, 256, 3, 3]): -XavierInit: gain=1, distribution=uniform, bias=0 - -img_neck.fpn_convs.3.conv.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.instance_bank.anchor - torch.Size([900, 11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.instance_bank.instance_feature - torch.Size([900, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.instance_bank.anchor_handler.fix_scale - torch.Size([1, 3]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.0.weight - torch.Size([128, 3]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.0.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.2.weight - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.2.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.3.weight - torch.Size([128, 128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.3.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.5.weight - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.5.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.6.weight - torch.Size([128, 128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.6.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.8.weight - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.8.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.9.weight - torch.Size([128, 128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.9.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.11.weight - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.pos_fc.11.bias - torch.Size([128]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.0.weight - torch.Size([32, 3]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.0.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.2.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.2.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.3.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.3.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.5.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.5.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.6.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.6.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.8.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.8.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.9.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.9.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.11.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.size_fc.11.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.0.weight - torch.Size([32, 2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.0.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.2.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.2.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.3.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.3.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.5.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.5.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.6.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.6.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.8.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.8.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.9.weight - torch.Size([32, 32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.9.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.11.weight - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.yaw_fc.11.bias - torch.Size([32]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.0.weight - torch.Size([64, 3]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.0.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.2.weight - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.2.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.3.weight - torch.Size([64, 64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.3.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.5.weight - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.5.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.6.weight - torch.Size([64, 64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.6.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.8.weight - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.8.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.9.weight - torch.Size([64, 64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.9.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.11.weight - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.anchor_encoder.vel_fc.11.bias - torch.Size([64]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.0.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.0.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.1.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.1.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.1.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.1.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.1.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.1.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.1.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.1.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.3.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.3.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.4.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.4.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.4.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.4.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.5.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.5.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.5.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.5.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.6.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.6.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.7.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.7.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.8.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.8.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.8.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.8.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.8.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.8.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.8.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.8.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.10.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.10.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.11.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.11.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.11.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.11.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.12.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.12.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.12.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.12.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.13.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.13.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.14.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.14.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.15.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.15.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.15.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.15.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.15.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.15.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.15.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.15.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.16.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.16.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.17.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.17.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.18.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.18.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.18.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.18.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.19.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.19.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.19.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.19.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.20.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.20.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.21.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.21.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.22.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.22.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.22.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.22.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.22.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.22.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.22.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.22.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.23.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.23.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.24.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.24.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.25.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.25.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.25.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.25.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.26.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.26.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.26.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.26.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.27.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.27.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.28.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.28.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.29.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.29.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.29.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.29.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.29.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.29.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.29.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.29.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.30.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.30.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.31.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.31.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.32.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.32.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.32.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.32.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.33.attn.in_proj_weight - torch.Size([1536, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.33.attn.in_proj_bias - torch.Size([1536]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.33.attn.out_proj.weight - torch.Size([512, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.33.attn.out_proj.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.34.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.34.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.kps_generator.fix_scale - torch.Size([7, 3]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.kps_generator.learnable_fc.weight - torch.Size([18, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.kps_generator.learnable_fc.bias - torch.Size([18]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.output_proj.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.output_proj.bias - torch.Size([256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.camera_encoder.0.weight - torch.Size([256, 12]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.camera_encoder.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.3.weight - torch.Size([256, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.camera_encoder.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.camera_encoder.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.35.weights_fc.weight - torch.Size([416, 256]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.35.weights_fc.bias - torch.Size([416]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.36.pre_norm.weight - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.36.pre_norm.bias - torch.Size([512]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.36.layers.0.0.weight - torch.Size([1024, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.36.layers.0.0.bias - torch.Size([1024]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.36.layers.1.weight - torch.Size([256, 1024]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.36.layers.1.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.36.identity_fc.weight - torch.Size([256, 512]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.36.identity_fc.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.37.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.37.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.2.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.4.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.4.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.5.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.7.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.7.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.9.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.9.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.10.weight - torch.Size([11, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.10.bias - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.layers.11.scale - torch.Size([11]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.6.weight - torch.Size([10, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.cls_layers.6.bias - torch.Size([10]): -Initialized by user-defined `init_weights` in Sparse4DHead - -head.layers.38.quality_layers.0.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.0.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.2.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.2.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.3.weight - torch.Size([256, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.3.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.5.weight - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.5.bias - torch.Size([256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.6.weight - torch.Size([2, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.layers.38.quality_layers.6.bias - torch.Size([2]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.fc_before.weight - torch.Size([512, 256]): -The value is the same before and after calling `init_weights` of Sparse4D - -head.fc_after.weight - torch.Size([256, 512]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.0.weight - torch.Size([1, 256, 1, 1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.0.bias - torch.Size([1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.1.weight - torch.Size([1, 256, 1, 1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.1.bias - torch.Size([1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.2.weight - torch.Size([1, 256, 1, 1]): -The value is the same before and after calling `init_weights` of Sparse4D - -depth_branch.depth_layers.2.bias - torch.Size([1]): -The value is the same before and after calling `init_weights` of Sparse4D -2026-04-03 16:50:06,914 - mmdet - INFO - Model: -Sparse4D( - (img_backbone): ResNet( - (conv1): Conv2d(3, 64, kernel_size=(7, 7), stride=(2, 2), padding=(3, 3), bias=False) - (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (maxpool): MaxPool2d(kernel_size=3, stride=2, padding=1, dilation=1, ceil_mode=False) - (layer1): ResLayer( - (0): Bottleneck( - (conv1): Conv2d(64, 64, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (downsample): Sequential( - (0): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - ) - ) - (1): Bottleneck( - (conv1): Conv2d(256, 64, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (2): Bottleneck( - (conv1): Conv2d(256, 64, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - ) - (layer2): ResLayer( - (0): Bottleneck( - (conv1): Conv2d(256, 128, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (downsample): Sequential( - (0): Conv2d(256, 512, kernel_size=(1, 1), stride=(2, 2), bias=False) - (1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - ) - ) - (1): Bottleneck( - (conv1): Conv2d(512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (2): Bottleneck( - (conv1): Conv2d(512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (3): Bottleneck( - (conv1): Conv2d(512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - ) - (layer3): ResLayer( - (0): Bottleneck( - (conv1): Conv2d(512, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (downsample): Sequential( - (0): Conv2d(512, 1024, kernel_size=(1, 1), stride=(2, 2), bias=False) - (1): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - ) - ) - (1): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (2): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (3): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (4): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (5): Bottleneck( - (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - ) - (layer4): ResLayer( - (0): Bottleneck( - (conv1): Conv2d(1024, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(512, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - (downsample): Sequential( - (0): Conv2d(1024, 2048, kernel_size=(1, 1), stride=(2, 2), bias=False) - (1): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - ) - ) - (1): Bottleneck( - (conv1): Conv2d(2048, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(512, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - (2): Bottleneck( - (conv1): Conv2d(2048, 512, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) - (bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (conv3): Conv2d(512, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False) - (bn3): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) - (relu): ReLU(inplace=True) - ) - ) - ) - init_cfg={'type': 'Pretrained', 'checkpoint': 'ckpt/resnet50-19c8e357.pth'} - (img_neck): FPN( - (lateral_convs): ModuleList( - (0): ConvModule( - (conv): Conv2d(256, 256, kernel_size=(1, 1), stride=(1, 1)) - ) - (1): ConvModule( - (conv): Conv2d(512, 256, kernel_size=(1, 1), stride=(1, 1)) - ) - (2): ConvModule( - (conv): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1)) - ) - (3): ConvModule( - (conv): Conv2d(2048, 256, kernel_size=(1, 1), stride=(1, 1)) - ) - ) - (fpn_convs): ModuleList( - (0-3): 4 x ConvModule( - (conv): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1)) - ) - ) - ) - init_cfg={'type': 'Xavier', 'layer': 'Conv2d', 'distribution': 'uniform'} - (head): Sparse4DHead( - (instance_bank): InstanceBank( - (anchor_handler): SparseBox3DKeyPointsGenerator() - ) - (anchor_encoder): SparseBox3DEncoder( - (pos_fc): Sequential( - (0): Linear(in_features=3, out_features=128, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((128,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=128, out_features=128, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((128,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=128, out_features=128, bias=True) - (7): ReLU(inplace=True) - (8): LayerNorm((128,), eps=1e-05, elementwise_affine=True) - (9): Linear(in_features=128, out_features=128, bias=True) - (10): ReLU(inplace=True) - (11): LayerNorm((128,), eps=1e-05, elementwise_affine=True) - ) - (size_fc): Sequential( - (0): Linear(in_features=3, out_features=32, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=32, out_features=32, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=32, out_features=32, bias=True) - (7): ReLU(inplace=True) - (8): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (9): Linear(in_features=32, out_features=32, bias=True) - (10): ReLU(inplace=True) - (11): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - ) - (yaw_fc): Sequential( - (0): Linear(in_features=2, out_features=32, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=32, out_features=32, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=32, out_features=32, bias=True) - (7): ReLU(inplace=True) - (8): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - (9): Linear(in_features=32, out_features=32, bias=True) - (10): ReLU(inplace=True) - (11): LayerNorm((32,), eps=1e-05, elementwise_affine=True) - ) - (vel_fc): Sequential( - (0): Linear(in_features=3, out_features=64, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((64,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=64, out_features=64, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((64,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=64, out_features=64, bias=True) - (7): ReLU(inplace=True) - (8): LayerNorm((64,), eps=1e-05, elementwise_affine=True) - (9): Linear(in_features=64, out_features=64, bias=True) - (10): ReLU(inplace=True) - (11): LayerNorm((64,), eps=1e-05, elementwise_affine=True) - ) - ) - (loss_cls): FocalLoss() - (loss_reg): SparseBox3DLoss( - (loss_box): L1Loss() - (loss_cns): CrossEntropyLoss(avg_non_ignore=False) - (loss_yns): GaussianFocalLoss() - ) - (layers): ModuleList( - (0): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (1): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (4-5): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (6): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (7): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (8): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (11-12): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (13): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (14): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (15): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (16): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (17): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (18-19): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (20): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (21): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (22): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (23): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (24): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (25-26): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (27): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (28): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (29): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (30): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (31): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - (32-33): 2 x MultiheadAttention( - (attn): MultiheadAttention( - (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True) - ) - (proj_drop): Dropout(p=0.0, inplace=False) - (dropout_layer): Dropout(p=0.1, inplace=False) - ) - (34): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (35): DeformableFeatureAggregation( - (proj_drop): Dropout(p=0.0, inplace=False) - (kps_generator): SparseBox3DKeyPointsGenerator( - (learnable_fc): Linear(in_features=256, out_features=18, bias=True) - ) - (output_proj): Linear(in_features=256, out_features=256, bias=True) - (camera_encoder): Sequential( - (0): Linear(in_features=12, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - ) - (weights_fc): Linear(in_features=256, out_features=416, bias=True) - ) - (36): AsymmetricFFN( - (activate): ReLU(inplace=True) - (pre_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True) - (layers): Sequential( - (0): Sequential( - (0): Linear(in_features=512, out_features=1024, bias=True) - (1): ReLU(inplace=True) - (2): Dropout(p=0.1, inplace=False) - ) - (1): Linear(in_features=1024, out_features=256, bias=True) - (2): Dropout(p=0.1, inplace=False) - ) - (dropout_layer): Identity() - (identity_fc): Linear(in_features=512, out_features=256, bias=True) - ) - (37): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (38): SparseBox3DRefinementModule( - (layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): Linear(in_features=256, out_features=256, bias=True) - (3): ReLU(inplace=True) - (4): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (5): Linear(in_features=256, out_features=256, bias=True) - (6): ReLU(inplace=True) - (7): Linear(in_features=256, out_features=256, bias=True) - (8): ReLU(inplace=True) - (9): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (10): Linear(in_features=256, out_features=11, bias=True) - (11): Scale() - ) - (cls_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=10, bias=True) - ) - (quality_layers): Sequential( - (0): Linear(in_features=256, out_features=256, bias=True) - (1): ReLU(inplace=True) - (2): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (3): Linear(in_features=256, out_features=256, bias=True) - (4): ReLU(inplace=True) - (5): LayerNorm((256,), eps=1e-05, elementwise_affine=True) - (6): Linear(in_features=256, out_features=2, bias=True) - ) - ) - ) - (fc_before): Linear(in_features=256, out_features=512, bias=False) - (fc_after): Linear(in_features=512, out_features=256, bias=False) - ) - (depth_branch): DenseDepthNet( - (depth_layers): ModuleList( - (0-2): 3 x Conv2d(256, 1, kernel_size=(1, 1), stride=(1, 1)) - ) - ) - (grid_mask): GridMask() -) -2026-04-03 16:50:17,763 - mmdet - INFO - Start running, host: root@bw61, work_dir: /workspace/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704 -2026-04-03 16:50:17,764 - mmdet - INFO - Hooks will be executed in the following order: -before_run: -(VERY_HIGH ) CosineAnnealingLrUpdaterHook -(ABOVE_NORMAL) Fp16OptimizerHook -(NORMAL ) CheckpointHook -(NORMAL ) CustomDistEvalHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -before_train_epoch: -(VERY_HIGH ) CosineAnnealingLrUpdaterHook -(NORMAL ) CustomDistEvalHook -(LOW ) IterTimerHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -before_train_iter: -(VERY_HIGH ) CosineAnnealingLrUpdaterHook -(NORMAL ) CustomDistEvalHook -(LOW ) IterTimerHook - -------------------- -after_train_iter: -(ABOVE_NORMAL) Fp16OptimizerHook -(NORMAL ) CheckpointHook -(NORMAL ) CustomDistEvalHook -(NORMAL ) ProfilerHook -(LOW ) IterTimerHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -after_train_epoch: -(NORMAL ) CheckpointHook -(NORMAL ) CustomDistEvalHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -before_val_epoch: -(LOW ) IterTimerHook -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -before_val_iter: -(LOW ) IterTimerHook - -------------------- -after_val_iter: -(LOW ) IterTimerHook - -------------------- -after_val_epoch: -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -after_run: -(VERY_LOW ) TextLoggerHook -(VERY_LOW ) TensorboardLoggerHook - -------------------- -2026-04-03 16:50:17,764 - mmdet - INFO - workflow: [('train', 1)], max: 23400 iters -2026-04-03 16:50:17,765 - mmdet - INFO - Checkpoints will be saved to /workspace/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704 by HardDiskBackend. -2026-04-03 16:51:53,722 - mmdet - INFO - Iter [1/23400] lr: 1.000e-04, eta: 25 days, 19:20:27, time: 95.287, data_time: 9.999, memory: 36997, loss_cls_0: 2.3501, loss_box_0: 0.0070, loss_cns_0: 0.0014, loss_yns_0: 0.0006, loss_cls_1: 2.1456, loss_box_1: 0.0897, loss_cns_1: 0.0202, loss_yns_1: 0.0066, loss_cls_2: 2.3166, loss_box_2: 0.0049, loss_cns_2: 0.0006, loss_yns_2: 0.0004, loss_cls_3: 2.4076, loss_box_3: 0.0274, loss_cns_3: 0.0048, loss_yns_3: 0.0016, loss_cls_4: 2.0284, loss_box_4: 0.4352, loss_cns_4: 0.0570, loss_yns_4: 0.0273, loss_cls_5: 2.4276, loss_box_5: 0.0131, loss_cns_5: 0.0020, loss_yns_5: 0.0007, loss_cls_dn_0: 1.1907, loss_box_dn_0: 1.4621, loss_cls_dn_1: 1.1081, loss_box_dn_1: 1.7343, loss_cls_dn_2: 1.1703, loss_box_dn_2: 1.9760, loss_cls_dn_3: 1.1697, loss_box_dn_3: 2.2541, loss_cls_dn_4: 1.0559, loss_box_dn_4: 2.4399, loss_cls_dn_5: 1.2401, loss_box_dn_5: 2.6845, loss_dense_depth: 1.8372, loss: 35.6994, grad_norm: 273.3486 -2026-04-03 16:51:55,915 - mmdet - INFO - Iter [2/23400] lr: 1.004e-04, eta: 13 days, 4:44:34, time: 2.180, data_time: 0.044, memory: 36997, loss_cls_0: 2.0166, loss_box_0: 0.0236, loss_cns_0: 0.0061, loss_yns_0: 0.0022, loss_cls_1: 2.0335, loss_box_1: 0.1354, loss_cns_1: 0.0251, loss_yns_1: 0.0077, loss_cls_2: 2.1187, loss_box_2: 0.2061, loss_cns_2: 0.0206, loss_yns_2: 0.0089, loss_cls_3: 1.9766, loss_box_3: 0.3063, loss_cns_3: 0.0389, loss_yns_3: 0.0133, loss_cls_4: 1.8127, loss_box_4: 1.4683, loss_cns_4: 0.1428, loss_yns_4: 0.0471, loss_cls_5: 2.0603, loss_box_5: 0.5356, loss_cns_5: 0.0559, loss_yns_5: 0.0206, loss_cls_dn_0: 1.0146, loss_box_dn_0: 1.2785, loss_cls_dn_1: 0.9576, loss_box_dn_1: 2.4414, loss_cls_dn_2: 0.9726, loss_box_dn_2: 2.5615, loss_cls_dn_3: 0.9198, loss_box_dn_3: 2.6295, loss_cls_dn_4: 0.8419, loss_box_dn_4: 2.8946, loss_cls_dn_5: 0.9851, loss_box_dn_5: 3.1360, loss_dense_depth: 1.6984, loss: 37.4147, grad_norm: 64.3749 -2026-04-03 16:51:57,216 - mmdet - INFO - Iter [3/23400] lr: 1.008e-04, eta: 8 days, 21:58:02, time: 1.299, data_time: 0.057, memory: 36997, loss_cls_0: 1.4379, loss_box_0: 2.5016, loss_cns_0: 0.6289, loss_yns_0: 0.2332, loss_cls_1: 1.7769, loss_box_1: 1.7257, loss_cns_1: 0.2803, loss_yns_1: 0.1036, loss_cls_2: 1.8045, loss_box_2: 3.6869, loss_cns_2: 0.3380, loss_yns_2: 0.1847, loss_cls_3: 1.6327, loss_box_3: 4.7280, loss_cns_3: 0.4363, loss_yns_3: 0.1993, loss_cls_4: 1.5866, loss_box_4: 4.0411, loss_cns_4: 0.3571, loss_yns_4: 0.1627, loss_cls_5: 1.7100, loss_box_5: 2.6877, loss_cns_5: 0.1967, loss_yns_5: 0.0892, loss_cls_dn_0: 0.6913, loss_box_dn_0: 1.1465, loss_cls_dn_1: 0.8323, loss_box_dn_1: 2.4695, loss_cls_dn_2: 0.8086, loss_box_dn_2: 2.6850, loss_cls_dn_3: 0.7201, loss_box_dn_3: 2.8650, loss_cls_dn_4: 0.7243, loss_box_dn_4: 3.1038, loss_cls_dn_5: 0.8077, loss_box_dn_5: 3.3483, loss_dense_depth: 1.6403, loss: 54.3725, grad_norm: 98.3870 -2026-04-03 16:51:58,483 - mmdet - INFO - Iter [4/23400] lr: 1.012e-04, eta: 6 days, 18:31:45, time: 1.268, data_time: 0.055, memory: 36997, loss_cls_0: 1.4084, loss_box_0: 2.5848, loss_cns_0: 0.5437, loss_yns_0: 0.1792, loss_cls_1: 1.6035, loss_box_1: 3.0533, loss_cns_1: 0.4472, loss_yns_1: 0.1782, loss_cls_2: 1.7106, loss_box_2: 3.6142, loss_cns_2: 0.4547, loss_yns_2: 0.1921, loss_cls_3: 1.5283, loss_box_3: 4.1035, loss_cns_3: 0.4746, loss_yns_3: 0.2234, loss_cls_4: 1.4767, loss_box_4: 4.6554, loss_cns_4: 0.3786, loss_yns_4: 0.1952, loss_cls_5: 1.5134, loss_box_5: 4.7963, loss_cns_5: 0.4382, loss_yns_5: 0.2003, loss_cls_dn_0: 0.5469, loss_box_dn_0: 1.1705, loss_cls_dn_1: 0.7323, loss_box_dn_1: 2.6399, loss_cls_dn_2: 0.6963, loss_box_dn_2: 2.7055, loss_cls_dn_3: 0.6232, loss_box_dn_3: 2.8766, loss_cls_dn_4: 0.6080, loss_box_dn_4: 3.0555, loss_cls_dn_5: 0.6795, loss_box_dn_5: 3.2298, loss_dense_depth: 1.5973, loss: 57.1152, grad_norm: 127.2257 -2026-04-03 16:51:59,805 - mmdet - INFO - Iter [5/23400] lr: 1.016e-04, eta: 5 days, 11:44:28, time: 1.326, data_time: 0.058, memory: 36997, loss_cls_0: 1.3268, loss_box_0: 2.7566, loss_cns_0: 0.4860, loss_yns_0: 0.1968, loss_cls_1: 1.5403, loss_box_1: 3.8564, loss_cns_1: 0.4049, loss_yns_1: 0.2045, loss_cls_2: 1.6347, loss_box_2: 4.0082, loss_cns_2: 0.3957, loss_yns_2: 0.1917, loss_cls_3: 1.4663, loss_box_3: 4.1163, loss_cns_3: 0.3974, loss_yns_3: 0.2066, loss_cls_4: 1.3999, loss_box_4: 4.3196, loss_cns_4: 0.3751, loss_yns_4: 0.1995, loss_cls_5: 1.3978, loss_box_5: 4.5025, loss_cns_5: 0.4280, loss_yns_5: 0.2114, loss_cls_dn_0: 0.5380, loss_box_dn_0: 1.2446, loss_cls_dn_1: 0.6613, loss_box_dn_1: 2.2641, loss_cls_dn_2: 0.6522, loss_box_dn_2: 2.3789, loss_cls_dn_3: 0.5671, loss_box_dn_3: 2.4651, loss_cls_dn_4: 0.5675, loss_box_dn_4: 2.6250, loss_cls_dn_5: 0.5955, loss_box_dn_5: 2.7148, loss_dense_depth: 1.5489, loss: 54.8458, grad_norm: 111.1918 -2026-04-03 16:52:01,096 - mmdet - INFO - Iter [6/23400] lr: 1.020e-04, eta: 4 days, 15:10:29, time: 1.288, data_time: 0.055, memory: 36997, loss_cls_0: 1.3241, loss_box_0: 2.6535, loss_cns_0: 0.5906, loss_yns_0: 0.1921, loss_cls_1: 1.4745, loss_box_1: 3.9117, loss_cns_1: 0.3939, loss_yns_1: 0.2022, loss_cls_2: 1.5216, loss_box_2: 4.0790, loss_cns_2: 0.3705, loss_yns_2: 0.1985, loss_cls_3: 1.3767, loss_box_3: 4.0531, loss_cns_3: 0.3575, loss_yns_3: 0.1952, loss_cls_4: 1.3495, loss_box_4: 4.3165, loss_cns_4: 0.3198, loss_yns_4: 0.2027, loss_cls_5: 1.3385, loss_box_5: 4.4109, loss_cns_5: 0.3238, loss_yns_5: 0.2155, loss_cls_dn_0: 0.5339, loss_box_dn_0: 1.2204, loss_cls_dn_1: 0.6045, loss_box_dn_1: 2.4081, loss_cls_dn_2: 0.6006, loss_box_dn_2: 2.4905, loss_cls_dn_3: 0.5263, loss_box_dn_3: 2.5153, loss_cls_dn_4: 0.4997, loss_box_dn_4: 2.7344, loss_cls_dn_5: 0.5107, loss_box_dn_5: 2.7709, loss_dense_depth: 1.4746, loss: 54.2617, grad_norm: 125.3949 diff --git a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20260403_165004.log.json b/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20260403_165004.log.json deleted file mode 100644 index c099e72a9078b3c51e605359ced16ad9b7d60c93..0000000000000000000000000000000000000000 --- a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/20260403_165004.log.json +++ /dev/null @@ -1,7 +0,0 @@ -{"env_info": "sys.platform: linux\nPython: 3.10.12 (main, Aug 15 2025, 14:32:43) [GCC 11.4.0]\nCUDA available: True\nGPU 0,1,2,3,4,5,6,7: BW1000_H\nCUDA_HOME: /opt/dtk\nNVCC: Not Available\nGCC: x86_64-linux-gnu-gcc (Ubuntu 11.4.0-1ubuntu1~22.04) 11.4.0\nPyTorch: 2.5.1\nPyTorch compiling details: PyTorch built with:\n - GCC 10.3\n - C++ Version: 201703\n - Intel(R) Math Kernel Library Version 2020.0.4 Product Build 20200917 for Intel(R) 64 architecture applications\n - OpenMP 201511 (a.k.a. OpenMP 4.5)\n - LAPACK is enabled (usually provided by MKL)\n - NNPACK is enabled\n - CPU capability usage: AVX512\n - HIP Runtime 6.3.25521\n - MIOpen 2.18.0\n - Magma 2.8.0\n - Build settings: BLAS_INFO=mkl, BUILD_TYPE=Release, CXX_COMPILER=/opt/rh/gcc-toolset-10/root/usr/bin/c++, CXX_FLAGS= -D_GLIBCXX_USE_CXX11_ABI=1 -fvisibility-inlines-hidden -DUSE_PTHREADPOOL -DNDEBUG -DUSE_KINETO -DLIBKINETO_NOCUPTI -DLIBKINETO_NOXPUPTI=ON -DUSE_FBGEMM -DUSE_PYTORCH_QNNPACK -DUSE_XNNPACK -DSYMBOLICATE_MOBILE_DEBUG_HANDLE -O2 -fPIC -Wall -Wextra -Werror=return-type -Werror=non-virtual-dtor -Werror=bool-operation -Wnarrowing -Wno-missing-field-initializers -Wno-type-limits -Wno-array-bounds -Wno-unknown-pragmas -Wno-unused-parameter -Wno-strict-overflow -Wno-strict-aliasing -Wno-stringop-overflow -Wsuggest-override -Wno-psabi -Wno-error=old-style-cast -Wno-missing-braces -fdiagnostics-color=always -faligned-new -Wno-unused-but-set-variable -Wno-maybe-uninitialized -fno-math-errno -fno-trapping-math -Werror=format -Wno-stringop-overflow, FORCE_FALLBACK_CUDA_MPI=1, LAPACK_INFO=mkl, PERF_WITH_AVX=1, PERF_WITH_AVX2=1, TORCH_VERSION=2.5.1, USE_CUDA=0, USE_CUDNN=OFF, USE_CUSPARSELT=OFF, USE_EXCEPTION_PTR=1, USE_GFLAGS=1, USE_GLOG=1, USE_GLOO=1, USE_MKL=ON, USE_MKLDNN=0, USE_MPI=1, USE_NCCL=1, USE_NNPACK=ON, USE_OPENMP=1, USE_ROCM=ON, USE_ROCM_KERNEL_ASSERT=OFF, \n\nTorchVision: 0.20.1\nOpenCV: 4.12.0\nMMCV: 1.6.1\nMMCV Compiler: GCC 10.3\nMMCV CUDA Compiler: rocm not available\nMMDetection: 2.25.1+c41df4b", "config": "plugin = True\nplugin_dir = 'projects/mmdet3d_plugin/'\ndist_params = dict(backend='nccl')\nlog_level = 'INFO'\nwork_dir = './work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704'\ntotal_batch_size = 120\nnum_gpus = 8\nbatch_size = 15\nnum_iters_per_epoch = 234\nnum_epochs = 100\ncheckpoint_epoch_interval = 20\ncheckpoint_config = dict(interval=4680)\nlog_config = dict(\n interval=1,\n hooks=[\n dict(type='TextLoggerHook', by_epoch=False),\n dict(type='TensorboardLoggerHook')\n ])\nload_from = None\nresume_from = None\nworkflow = [('train', 1)]\nfp16 = dict(loss_scale=32.0)\ninput_shape = (704, 256)\ntracking_test = True\ntracking_threshold = 0.2\nclass_names = [\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier',\n 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n]\nnum_classes = 10\nembed_dims = 256\nnum_groups = 8\nnum_decoder = 6\nnum_single_frame_decoder = 1\nuse_deformable_func = True\nstrides = [4, 8, 16, 32]\nnum_levels = 4\nnum_depth_layers = 3\ndrop_out = 0.1\ntemporal = True\ndecouple_attn = True\nwith_quality_estimation = True\nmodel = dict(\n type='Sparse4D',\n use_grid_mask=True,\n use_deformable_func=True,\n img_backbone=dict(\n type='ResNet',\n depth=50,\n num_stages=4,\n frozen_stages=-1,\n norm_eval=False,\n style='pytorch',\n with_cp=True,\n out_indices=(0, 1, 2, 3),\n norm_cfg=dict(type='BN', requires_grad=True),\n pretrained='ckpt/resnet50-19c8e357.pth'),\n img_neck=dict(\n type='FPN',\n num_outs=4,\n start_level=0,\n out_channels=256,\n add_extra_convs='on_output',\n relu_before_extra_convs=True,\n in_channels=[256, 512, 1024, 2048]),\n depth_branch=dict(\n type='DenseDepthNet',\n embed_dims=256,\n num_depth_layers=3,\n loss_weight=0.2),\n head=dict(\n type='Sparse4DHead',\n cls_threshold_to_reg=0.05,\n decouple_attn=True,\n instance_bank=dict(\n type='InstanceBank',\n num_anchor=900,\n embed_dims=256,\n anchor='nuscenes_kmeans900.npy',\n anchor_handler=dict(type='SparseBox3DKeyPointsGenerator'),\n num_temp_instances=600,\n confidence_decay=0.6,\n feat_grad=False),\n anchor_encoder=dict(\n type='SparseBox3DEncoder',\n vel_dims=3,\n embed_dims=[128, 32, 32, 64],\n mode='cat',\n output_fc=False,\n in_loops=1,\n out_loops=4),\n num_single_frame_decoder=1,\n operation_order=[\n 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm',\n 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm',\n 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm',\n 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm',\n 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm',\n 'deformable', 'ffn', 'norm', 'refine'\n ],\n temp_graph_model=dict(\n type='MultiheadAttention',\n embed_dims=512,\n num_heads=8,\n batch_first=True,\n dropout=0.1),\n graph_model=dict(\n type='MultiheadAttention',\n embed_dims=512,\n num_heads=8,\n batch_first=True,\n dropout=0.1),\n norm_layer=dict(type='LN', normalized_shape=256),\n ffn=dict(\n type='AsymmetricFFN',\n in_channels=512,\n pre_norm=dict(type='LN'),\n embed_dims=256,\n feedforward_channels=1024,\n num_fcs=2,\n ffn_drop=0.1,\n act_cfg=dict(type='ReLU', inplace=True)),\n deformable_model=dict(\n type='DeformableFeatureAggregation',\n embed_dims=256,\n num_groups=8,\n num_levels=4,\n num_cams=6,\n attn_drop=0.15,\n use_deformable_func=True,\n use_camera_embed=True,\n residual_mode='cat',\n kps_generator=dict(\n type='SparseBox3DKeyPointsGenerator',\n num_learnable_pts=6,\n fix_scale=[[0, 0, 0], [0.45, 0, 0], [-0.45, 0, 0],\n [0, 0.45, 0], [0, -0.45, 0], [0, 0, 0.45],\n [0, 0, -0.45]])),\n refine_layer=dict(\n type='SparseBox3DRefinementModule',\n embed_dims=256,\n num_cls=10,\n refine_yaw=True,\n with_quality_estimation=True),\n sampler=dict(\n type='SparseBox3DTarget',\n num_dn_groups=5,\n num_temp_dn_groups=3,\n dn_noise_scale=[2.0, 2.0, 2.0, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5],\n max_dn_gt=32,\n add_neg_dn=True,\n cls_weight=2.0,\n box_weight=0.25,\n reg_weights=[2.0, 2.0, 2.0, 0.5, 0.5, 0.5, 0.0, 0.0, 0.0, 0.0],\n cls_wise_reg_weights=dict(\n {9: [2.0, 2.0, 2.0, 1.0, 1.0, 1.0, 0.0, 0.0, 1.0, 1.0]})),\n loss_cls=dict(\n type='FocalLoss',\n use_sigmoid=True,\n gamma=2.0,\n alpha=0.25,\n loss_weight=2.0),\n loss_reg=dict(\n type='SparseBox3DLoss',\n loss_box=dict(type='L1Loss', loss_weight=0.25),\n loss_centerness=dict(type='CrossEntropyLoss', use_sigmoid=True),\n loss_yawness=dict(type='GaussianFocalLoss'),\n cls_allow_reverse=[5]),\n decoder=dict(type='SparseBox3DDecoder'),\n reg_weights=[2.0, 2.0, 2.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0]))\ndataset_type = 'NuScenes3DDetTrackDataset'\ndata_root = 'data/nuscenes/'\nanno_root = 'data/nuscenes_anno_pkls/'\nfile_client_args = dict(backend='disk')\nimg_norm_cfg = dict(\n mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375], to_rgb=True)\ntrain_pipeline = [\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(\n type='LoadPointsFromFile',\n coord_type='LIDAR',\n load_dim=5,\n use_dim=5,\n file_client_args=dict(backend='disk')),\n dict(type='ResizeCropFlipImage'),\n dict(type='MultiScaleDepthMapGenerator', downsample=[4, 8, 16]),\n dict(type='BBoxRotation'),\n dict(type='PhotoMetricDistortionMultiViewImage'),\n dict(\n type='NormalizeMultiviewImage',\n mean=[123.675, 116.28, 103.53],\n std=[58.395, 57.12, 57.375],\n to_rgb=True),\n dict(\n type='CircleObjectRangeFilter',\n class_dist_thred=[55, 55, 55, 55, 55, 55, 55, 55, 55, 55]),\n dict(\n type='InstanceNameFilter',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ]),\n dict(type='NuScenesSparse4DAdaptor'),\n dict(\n type='Collect',\n keys=[\n 'img', 'timestamp', 'projection_mat', 'image_wh', 'gt_depth',\n 'focal', 'gt_bboxes_3d', 'gt_labels_3d'\n ],\n meta_keys=['T_global', 'T_global_inv', 'timestamp', 'instance_id'])\n]\ntest_pipeline = [\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(type='ResizeCropFlipImage'),\n dict(\n type='NormalizeMultiviewImage',\n mean=[123.675, 116.28, 103.53],\n std=[58.395, 57.12, 57.375],\n to_rgb=True),\n dict(type='NuScenesSparse4DAdaptor'),\n dict(\n type='Collect',\n keys=['img', 'timestamp', 'projection_mat', 'image_wh'],\n meta_keys=['T_global', 'T_global_inv', 'timestamp'])\n]\ninput_modality = dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False)\ndata_basic_config = dict(\n type='NuScenes3DDetTrackDataset',\n data_root='data/nuscenes/',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier',\n 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ],\n modality=dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False),\n version='v1.0-trainval')\ndata_aug_conf = dict(\n resize_lim=(0.4, 0.47),\n final_dim=(256, 704),\n bot_pct_lim=(0.0, 0.0),\n rot_lim=(-5.4, 5.4),\n H=900,\n W=1600,\n rand_flip=True,\n rot3d_range=[-0.3925, 0.3925])\ndata = dict(\n samples_per_gpu=15,\n workers_per_gpu=15,\n train=dict(\n type='NuScenes3DDetTrackDataset',\n data_root='data/nuscenes/',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ],\n modality=dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False),\n version='v1.0-trainval',\n ann_file='data/nuscenes_anno_pkls/nuscenes_infos_train.pkl',\n pipeline=[\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(\n type='LoadPointsFromFile',\n coord_type='LIDAR',\n load_dim=5,\n use_dim=5,\n file_client_args=dict(backend='disk')),\n dict(type='ResizeCropFlipImage'),\n dict(type='MultiScaleDepthMapGenerator', downsample=[4, 8, 16]),\n dict(type='BBoxRotation'),\n dict(type='PhotoMetricDistortionMultiViewImage'),\n dict(\n type='NormalizeMultiviewImage',\n mean=[123.675, 116.28, 103.53],\n std=[58.395, 57.12, 57.375],\n to_rgb=True),\n dict(\n type='CircleObjectRangeFilter',\n class_dist_thred=[55, 55, 55, 55, 55, 55, 55, 55, 55, 55]),\n dict(\n type='InstanceNameFilter',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian',\n 'traffic_cone'\n ]),\n dict(type='NuScenesSparse4DAdaptor'),\n dict(\n type='Collect',\n keys=[\n 'img', 'timestamp', 'projection_mat', 'image_wh',\n 'gt_depth', 'focal', 'gt_bboxes_3d', 'gt_labels_3d'\n ],\n meta_keys=[\n 'T_global', 'T_global_inv', 'timestamp', 'instance_id'\n ])\n ],\n test_mode=False,\n data_aug_conf=dict(\n resize_lim=(0.4, 0.47),\n final_dim=(256, 704),\n bot_pct_lim=(0.0, 0.0),\n rot_lim=(-5.4, 5.4),\n H=900,\n W=1600,\n rand_flip=True,\n rot3d_range=[-0.3925, 0.3925]),\n with_seq_flag=True,\n sequences_split_num=2,\n keep_consistent_seq_aug=True),\n val=dict(\n type='NuScenes3DDetTrackDataset',\n data_root='data/nuscenes/',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ],\n modality=dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False),\n version='v1.0-trainval',\n ann_file='data/nuscenes_anno_pkls/nuscenes_infos_val.pkl',\n pipeline=[\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(type='ResizeCropFlipImage'),\n dict(\n type='NormalizeMultiviewImage',\n mean=[123.675, 116.28, 103.53],\n std=[58.395, 57.12, 57.375],\n to_rgb=True),\n dict(type='NuScenesSparse4DAdaptor'),\n dict(\n type='Collect',\n keys=['img', 'timestamp', 'projection_mat', 'image_wh'],\n meta_keys=['T_global', 'T_global_inv', 'timestamp'])\n ],\n data_aug_conf=dict(\n resize_lim=(0.4, 0.47),\n final_dim=(256, 704),\n bot_pct_lim=(0.0, 0.0),\n rot_lim=(-5.4, 5.4),\n H=900,\n W=1600,\n rand_flip=True,\n rot3d_range=[-0.3925, 0.3925]),\n test_mode=True,\n tracking=True,\n tracking_threshold=0.2),\n test=dict(\n type='NuScenes3DDetTrackDataset',\n data_root='data/nuscenes/',\n classes=[\n 'car', 'truck', 'construction_vehicle', 'bus', 'trailer',\n 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'\n ],\n modality=dict(\n use_lidar=False,\n use_camera=True,\n use_radar=False,\n use_map=False,\n use_external=False),\n version='v1.0-trainval',\n ann_file='data/nuscenes_anno_pkls/nuscenes_infos_val.pkl',\n pipeline=[\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(type='ResizeCropFlipImage'),\n dict(\n type='NormalizeMultiviewImage',\n mean=[123.675, 116.28, 103.53],\n std=[58.395, 57.12, 57.375],\n to_rgb=True),\n dict(type='NuScenesSparse4DAdaptor'),\n dict(\n type='Collect',\n keys=['img', 'timestamp', 'projection_mat', 'image_wh'],\n meta_keys=['T_global', 'T_global_inv', 'timestamp'])\n ],\n data_aug_conf=dict(\n resize_lim=(0.4, 0.47),\n final_dim=(256, 704),\n bot_pct_lim=(0.0, 0.0),\n rot_lim=(-5.4, 5.4),\n H=900,\n W=1600,\n rand_flip=True,\n rot3d_range=[-0.3925, 0.3925]),\n test_mode=True,\n tracking=True,\n tracking_threshold=0.2))\noptimizer = dict(\n type='AdamW',\n lr=0.0006,\n weight_decay=0.001,\n paramwise_cfg=dict(custom_keys=dict(img_backbone=dict(lr_mult=0.5))))\noptimizer_config = dict(grad_clip=dict(max_norm=25, norm_type=2))\nlr_config = dict(\n policy='CosineAnnealing',\n warmup='linear',\n warmup_iters=500,\n warmup_ratio=0.3333333333333333,\n min_lr_ratio=0.001)\nrunner = dict(type='IterBasedRunner', max_iters=23400)\nvis_pipeline = [\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(type='Collect', keys=['img'], meta_keys=['timestamp', 'lidar2img'])\n]\nevaluation = dict(\n interval=4680,\n pipeline=[\n dict(type='LoadMultiViewImageFromFiles', to_float32=True),\n dict(\n type='Collect', keys=['img'], meta_keys=['timestamp', 'lidar2img'])\n ])\ngpu_ids = range(0, 8)\n", "seed": 0, "exp_name": "sparse4dv3_temporal_r50_1x8_bs6_256x704.py"} -{"mode": "train", "epoch": 1, "iter": 1, "lr": 0.0001, "memory": 36997, "data_time": 9.99901, "loss_cls_0": 2.35014, "loss_box_0": 0.00697, "loss_cns_0": 0.00136, "loss_yns_0": 0.00057, "loss_cls_1": 2.14563, "loss_box_1": 0.08973, "loss_cns_1": 0.02019, "loss_yns_1": 0.00655, "loss_cls_2": 2.31656, "loss_box_2": 0.00492, "loss_cns_2": 0.00063, "loss_yns_2": 0.00043, "loss_cls_3": 2.40764, "loss_box_3": 0.02743, "loss_cns_3": 0.00483, "loss_yns_3": 0.00157, "loss_cls_4": 2.02842, "loss_box_4": 0.4352, "loss_cns_4": 0.05697, "loss_yns_4": 0.02725, "loss_cls_5": 2.42757, "loss_box_5": 0.01306, "loss_cns_5": 0.00202, "loss_yns_5": 0.00073, "loss_cls_dn_0": 1.19068, "loss_box_dn_0": 1.46213, "loss_cls_dn_1": 1.10814, "loss_box_dn_1": 1.73429, "loss_cls_dn_2": 1.17028, "loss_box_dn_2": 1.97599, "loss_cls_dn_3": 1.1697, "loss_box_dn_3": 2.25415, "loss_cls_dn_4": 1.05594, "loss_box_dn_4": 2.43994, "loss_cls_dn_5": 1.24009, "loss_box_dn_5": 2.68449, "loss_dense_depth": 1.83719, "loss": 35.69938, "grad_norm": 273.34857, "time": 95.28729} -{"mode": "train", "epoch": 1, "iter": 2, "lr": 0.0001, "memory": 36997, "data_time": 0.04388, "loss_cls_0": 2.01664, "loss_box_0": 0.02364, "loss_cns_0": 0.00614, "loss_yns_0": 0.00218, "loss_cls_1": 2.03348, "loss_box_1": 0.13543, "loss_cns_1": 0.02513, "loss_yns_1": 0.0077, "loss_cls_2": 2.11873, "loss_box_2": 0.20605, "loss_cns_2": 0.02055, "loss_yns_2": 0.00891, "loss_cls_3": 1.97661, "loss_box_3": 0.30635, "loss_cns_3": 0.03889, "loss_yns_3": 0.01332, "loss_cls_4": 1.81272, "loss_box_4": 1.46831, "loss_cns_4": 0.14281, "loss_yns_4": 0.04714, "loss_cls_5": 2.06031, "loss_box_5": 0.5356, "loss_cns_5": 0.05587, "loss_yns_5": 0.02064, "loss_cls_dn_0": 1.0146, "loss_box_dn_0": 1.27848, "loss_cls_dn_1": 0.95763, "loss_box_dn_1": 2.44142, "loss_cls_dn_2": 0.9726, "loss_box_dn_2": 2.56149, "loss_cls_dn_3": 0.91976, "loss_box_dn_3": 2.62951, "loss_cls_dn_4": 0.8419, "loss_box_dn_4": 2.89462, "loss_cls_dn_5": 0.98512, "loss_box_dn_5": 3.13601, "loss_dense_depth": 1.6984, "loss": 37.41468, "grad_norm": 64.37494, "time": 2.1804} -{"mode": "train", "epoch": 1, "iter": 3, "lr": 0.0001, "memory": 36997, "data_time": 0.05704, "loss_cls_0": 1.43787, "loss_box_0": 2.50161, "loss_cns_0": 0.6289, "loss_yns_0": 0.23316, "loss_cls_1": 1.77689, "loss_box_1": 1.72573, "loss_cns_1": 0.28033, "loss_yns_1": 0.10363, "loss_cls_2": 1.80445, "loss_box_2": 3.68688, "loss_cns_2": 0.33801, "loss_yns_2": 0.1847, "loss_cls_3": 1.63267, "loss_box_3": 4.72795, "loss_cns_3": 0.43634, "loss_yns_3": 0.19932, "loss_cls_4": 1.58661, "loss_box_4": 4.04114, "loss_cns_4": 0.3571, "loss_yns_4": 0.16273, "loss_cls_5": 1.71004, "loss_box_5": 2.68769, "loss_cns_5": 0.19674, "loss_yns_5": 0.08917, "loss_cls_dn_0": 0.69129, "loss_box_dn_0": 1.14652, "loss_cls_dn_1": 0.83227, "loss_box_dn_1": 2.46955, "loss_cls_dn_2": 0.80856, "loss_box_dn_2": 2.68503, "loss_cls_dn_3": 0.72006, "loss_box_dn_3": 2.86504, "loss_cls_dn_4": 0.72435, "loss_box_dn_4": 3.10385, "loss_cls_dn_5": 0.80769, "loss_box_dn_5": 3.3483, "loss_dense_depth": 1.64027, "loss": 54.37245, "grad_norm": 98.38701, "time": 1.29916} -{"mode": "train", "epoch": 1, "iter": 4, "lr": 0.0001, "memory": 36997, "data_time": 0.05471, "loss_cls_0": 1.40837, "loss_box_0": 2.58483, "loss_cns_0": 0.54366, "loss_yns_0": 0.17917, "loss_cls_1": 1.60351, "loss_box_1": 3.05331, "loss_cns_1": 0.44718, "loss_yns_1": 0.17818, "loss_cls_2": 1.71063, "loss_box_2": 3.61415, "loss_cns_2": 0.45474, "loss_yns_2": 0.1921, "loss_cls_3": 1.52833, "loss_box_3": 4.10346, "loss_cns_3": 0.47463, "loss_yns_3": 0.2234, "loss_cls_4": 1.47671, "loss_box_4": 4.65535, "loss_cns_4": 0.37861, "loss_yns_4": 0.19522, "loss_cls_5": 1.51338, "loss_box_5": 4.79634, "loss_cns_5": 0.43818, "loss_yns_5": 0.20032, "loss_cls_dn_0": 0.54691, "loss_box_dn_0": 1.17053, "loss_cls_dn_1": 0.73232, "loss_box_dn_1": 2.6399, "loss_cls_dn_2": 0.69635, "loss_box_dn_2": 2.70546, "loss_cls_dn_3": 0.62322, "loss_box_dn_3": 2.8766, "loss_cls_dn_4": 0.60797, "loss_box_dn_4": 3.05554, "loss_cls_dn_5": 0.67954, "loss_box_dn_5": 3.22982, "loss_dense_depth": 1.59727, "loss": 57.11519, "grad_norm": 127.22569, "time": 1.26824} -{"mode": "train", "epoch": 1, "iter": 5, "lr": 0.0001, "memory": 36997, "data_time": 0.05817, "loss_cls_0": 1.32677, "loss_box_0": 2.75655, "loss_cns_0": 0.486, "loss_yns_0": 0.19684, "loss_cls_1": 1.54025, "loss_box_1": 3.85636, "loss_cns_1": 0.40488, "loss_yns_1": 0.20445, "loss_cls_2": 1.63473, "loss_box_2": 4.00819, "loss_cns_2": 0.39565, "loss_yns_2": 0.19168, "loss_cls_3": 1.4663, "loss_box_3": 4.11631, "loss_cns_3": 0.39743, "loss_yns_3": 0.20664, "loss_cls_4": 1.39985, "loss_box_4": 4.31962, "loss_cns_4": 0.3751, "loss_yns_4": 0.19948, "loss_cls_5": 1.39778, "loss_box_5": 4.50248, "loss_cns_5": 0.42799, "loss_yns_5": 0.21135, "loss_cls_dn_0": 0.53796, "loss_box_dn_0": 1.24461, "loss_cls_dn_1": 0.66127, "loss_box_dn_1": 2.26407, "loss_cls_dn_2": 0.65224, "loss_box_dn_2": 2.3789, "loss_cls_dn_3": 0.56714, "loss_box_dn_3": 2.46506, "loss_cls_dn_4": 0.56753, "loss_box_dn_4": 2.62499, "loss_cls_dn_5": 0.59553, "loss_box_dn_5": 2.71484, "loss_dense_depth": 1.54893, "loss": 54.84576, "grad_norm": 111.19184, "time": 1.32603} -{"mode": "train", "epoch": 1, "iter": 6, "lr": 0.0001, "memory": 36997, "data_time": 0.05501, "loss_cls_0": 1.32413, "loss_box_0": 2.65355, "loss_cns_0": 0.59056, "loss_yns_0": 0.19214, "loss_cls_1": 1.47446, "loss_box_1": 3.9117, "loss_cns_1": 0.39393, "loss_yns_1": 0.20225, "loss_cls_2": 1.52156, "loss_box_2": 4.07898, "loss_cns_2": 0.37053, "loss_yns_2": 0.19848, "loss_cls_3": 1.37674, "loss_box_3": 4.05308, "loss_cns_3": 0.35748, "loss_yns_3": 0.19518, "loss_cls_4": 1.34949, "loss_box_4": 4.31655, "loss_cns_4": 0.31982, "loss_yns_4": 0.20266, "loss_cls_5": 1.33846, "loss_box_5": 4.41086, "loss_cns_5": 0.32379, "loss_yns_5": 0.21549, "loss_cls_dn_0": 0.53385, "loss_box_dn_0": 1.22038, "loss_cls_dn_1": 0.6045, "loss_box_dn_1": 2.40813, "loss_cls_dn_2": 0.60061, "loss_box_dn_2": 2.49051, "loss_cls_dn_3": 0.52626, "loss_box_dn_3": 2.51531, "loss_cls_dn_4": 0.49966, "loss_box_dn_4": 2.7344, "loss_cls_dn_5": 0.51073, "loss_box_dn_5": 2.77088, "loss_dense_depth": 1.4746, "loss": 54.26169, "grad_norm": 125.39494, "time": 1.28811} diff --git a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/sparse4dv3_temporal_r50_1x8_bs6_256x704.py b/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/sparse4dv3_temporal_r50_1x8_bs6_256x704.py deleted file mode 100644 index e8feeb1341ef87e886158a72d99fb5a42a48bb9f..0000000000000000000000000000000000000000 --- a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/sparse4dv3_temporal_r50_1x8_bs6_256x704.py +++ /dev/null @@ -1,430 +0,0 @@ -plugin = True -plugin_dir = 'projects/mmdet3d_plugin/' -dist_params = dict(backend='nccl') -log_level = 'INFO' -work_dir = './work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704' -total_batch_size = 120 -num_gpus = 8 -batch_size = 15 -num_iters_per_epoch = 234 -num_epochs = 100 -checkpoint_epoch_interval = 20 -checkpoint_config = dict(interval=4680) -log_config = dict( - interval=1, - hooks=[ - dict(type='TextLoggerHook', by_epoch=False), - dict(type='TensorboardLoggerHook') - ]) -load_from = None -resume_from = None -workflow = [('train', 1)] -fp16 = dict(loss_scale=32.0) -input_shape = (704, 256) -tracking_test = True -tracking_threshold = 0.2 -class_names = [ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier', - 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' -] -num_classes = 10 -embed_dims = 256 -num_groups = 8 -num_decoder = 6 -num_single_frame_decoder = 1 -use_deformable_func = True -strides = [4, 8, 16, 32] -num_levels = 4 -num_depth_layers = 3 -drop_out = 0.1 -temporal = True -decouple_attn = True -with_quality_estimation = True -model = dict( - type='Sparse4D', - use_grid_mask=True, - use_deformable_func=True, - img_backbone=dict( - type='ResNet', - depth=50, - num_stages=4, - frozen_stages=-1, - norm_eval=False, - style='pytorch', - with_cp=True, - out_indices=(0, 1, 2, 3), - norm_cfg=dict(type='BN', requires_grad=True), - pretrained='ckpt/resnet50-19c8e357.pth'), - img_neck=dict( - type='FPN', - num_outs=4, - start_level=0, - out_channels=256, - add_extra_convs='on_output', - relu_before_extra_convs=True, - in_channels=[256, 512, 1024, 2048]), - depth_branch=dict( - type='DenseDepthNet', - embed_dims=256, - num_depth_layers=3, - loss_weight=0.2), - head=dict( - type='Sparse4DHead', - cls_threshold_to_reg=0.05, - decouple_attn=True, - instance_bank=dict( - type='InstanceBank', - num_anchor=900, - embed_dims=256, - anchor='nuscenes_kmeans900.npy', - anchor_handler=dict(type='SparseBox3DKeyPointsGenerator'), - num_temp_instances=600, - confidence_decay=0.6, - feat_grad=False), - anchor_encoder=dict( - type='SparseBox3DEncoder', - vel_dims=3, - embed_dims=[128, 32, 32, 64], - mode='cat', - output_fc=False, - in_loops=1, - out_loops=4), - num_single_frame_decoder=1, - operation_order=[ - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine', 'temp_gnn', 'gnn', 'norm', - 'deformable', 'ffn', 'norm', 'refine' - ], - temp_graph_model=dict( - type='MultiheadAttention', - embed_dims=512, - num_heads=8, - batch_first=True, - dropout=0.1), - graph_model=dict( - type='MultiheadAttention', - embed_dims=512, - num_heads=8, - batch_first=True, - dropout=0.1), - norm_layer=dict(type='LN', normalized_shape=256), - ffn=dict( - type='AsymmetricFFN', - in_channels=512, - pre_norm=dict(type='LN'), - embed_dims=256, - feedforward_channels=1024, - num_fcs=2, - ffn_drop=0.1, - act_cfg=dict(type='ReLU', inplace=True)), - deformable_model=dict( - type='DeformableFeatureAggregation', - embed_dims=256, - num_groups=8, - num_levels=4, - num_cams=6, - attn_drop=0.15, - use_deformable_func=True, - use_camera_embed=True, - residual_mode='cat', - kps_generator=dict( - type='SparseBox3DKeyPointsGenerator', - num_learnable_pts=6, - fix_scale=[[0, 0, 0], [0.45, 0, 0], [-0.45, 0, 0], - [0, 0.45, 0], [0, -0.45, 0], [0, 0, 0.45], - [0, 0, -0.45]])), - refine_layer=dict( - type='SparseBox3DRefinementModule', - embed_dims=256, - num_cls=10, - refine_yaw=True, - with_quality_estimation=True), - sampler=dict( - type='SparseBox3DTarget', - num_dn_groups=5, - num_temp_dn_groups=3, - dn_noise_scale=[2.0, 2.0, 2.0, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5], - max_dn_gt=32, - add_neg_dn=True, - cls_weight=2.0, - box_weight=0.25, - reg_weights=[2.0, 2.0, 2.0, 0.5, 0.5, 0.5, 0.0, 0.0, 0.0, 0.0], - cls_wise_reg_weights=dict( - {9: [2.0, 2.0, 2.0, 1.0, 1.0, 1.0, 0.0, 0.0, 1.0, 1.0]})), - loss_cls=dict( - type='FocalLoss', - use_sigmoid=True, - gamma=2.0, - alpha=0.25, - loss_weight=2.0), - loss_reg=dict( - type='SparseBox3DLoss', - loss_box=dict(type='L1Loss', loss_weight=0.25), - loss_centerness=dict(type='CrossEntropyLoss', use_sigmoid=True), - loss_yawness=dict(type='GaussianFocalLoss'), - cls_allow_reverse=[5]), - decoder=dict(type='SparseBox3DDecoder'), - reg_weights=[2.0, 2.0, 2.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0])) -dataset_type = 'NuScenes3DDetTrackDataset' -data_root = 'data/nuscenes/' -anno_root = 'data/nuscenes_anno_pkls/' -file_client_args = dict(backend='disk') -img_norm_cfg = dict( - mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375], to_rgb=True) -train_pipeline = [ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict( - type='LoadPointsFromFile', - coord_type='LIDAR', - load_dim=5, - use_dim=5, - file_client_args=dict(backend='disk')), - dict(type='ResizeCropFlipImage'), - dict(type='MultiScaleDepthMapGenerator', downsample=[4, 8, 16]), - dict(type='BBoxRotation'), - dict(type='PhotoMetricDistortionMultiViewImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict( - type='CircleObjectRangeFilter', - class_dist_thred=[55, 55, 55, 55, 55, 55, 55, 55, 55, 55]), - dict( - type='InstanceNameFilter', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ]), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=[ - 'img', 'timestamp', 'projection_mat', 'image_wh', 'gt_depth', - 'focal', 'gt_bboxes_3d', 'gt_labels_3d' - ], - meta_keys=['T_global', 'T_global_inv', 'timestamp', 'instance_id']) -] -test_pipeline = [ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='ResizeCropFlipImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=['img', 'timestamp', 'projection_mat', 'image_wh'], - meta_keys=['T_global', 'T_global_inv', 'timestamp']) -] -input_modality = dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False) -data_basic_config = dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier', - 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval') -data_aug_conf = dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]) -data = dict( - samples_per_gpu=15, - workers_per_gpu=15, - train=dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval', - ann_file='data/nuscenes_anno_pkls/nuscenes_infos_train.pkl', - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict( - type='LoadPointsFromFile', - coord_type='LIDAR', - load_dim=5, - use_dim=5, - file_client_args=dict(backend='disk')), - dict(type='ResizeCropFlipImage'), - dict(type='MultiScaleDepthMapGenerator', downsample=[4, 8, 16]), - dict(type='BBoxRotation'), - dict(type='PhotoMetricDistortionMultiViewImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict( - type='CircleObjectRangeFilter', - class_dist_thred=[55, 55, 55, 55, 55, 55, 55, 55, 55, 55]), - dict( - type='InstanceNameFilter', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', - 'traffic_cone' - ]), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=[ - 'img', 'timestamp', 'projection_mat', 'image_wh', - 'gt_depth', 'focal', 'gt_bboxes_3d', 'gt_labels_3d' - ], - meta_keys=[ - 'T_global', 'T_global_inv', 'timestamp', 'instance_id' - ]) - ], - test_mode=False, - data_aug_conf=dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]), - with_seq_flag=True, - sequences_split_num=2, - keep_consistent_seq_aug=True), - val=dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval', - ann_file='data/nuscenes_anno_pkls/nuscenes_infos_val.pkl', - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='ResizeCropFlipImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=['img', 'timestamp', 'projection_mat', 'image_wh'], - meta_keys=['T_global', 'T_global_inv', 'timestamp']) - ], - data_aug_conf=dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]), - test_mode=True, - tracking=True, - tracking_threshold=0.2), - test=dict( - type='NuScenes3DDetTrackDataset', - data_root='data/nuscenes/', - classes=[ - 'car', 'truck', 'construction_vehicle', 'bus', 'trailer', - 'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone' - ], - modality=dict( - use_lidar=False, - use_camera=True, - use_radar=False, - use_map=False, - use_external=False), - version='v1.0-trainval', - ann_file='data/nuscenes_anno_pkls/nuscenes_infos_val.pkl', - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='ResizeCropFlipImage'), - dict( - type='NormalizeMultiviewImage', - mean=[123.675, 116.28, 103.53], - std=[58.395, 57.12, 57.375], - to_rgb=True), - dict(type='NuScenesSparse4DAdaptor'), - dict( - type='Collect', - keys=['img', 'timestamp', 'projection_mat', 'image_wh'], - meta_keys=['T_global', 'T_global_inv', 'timestamp']) - ], - data_aug_conf=dict( - resize_lim=(0.4, 0.47), - final_dim=(256, 704), - bot_pct_lim=(0.0, 0.0), - rot_lim=(-5.4, 5.4), - H=900, - W=1600, - rand_flip=True, - rot3d_range=[-0.3925, 0.3925]), - test_mode=True, - tracking=True, - tracking_threshold=0.2)) -optimizer = dict( - type='AdamW', - lr=0.0006, - weight_decay=0.001, - paramwise_cfg=dict(custom_keys=dict(img_backbone=dict(lr_mult=0.5)))) -optimizer_config = dict(grad_clip=dict(max_norm=25, norm_type=2)) -lr_config = dict( - policy='CosineAnnealing', - warmup='linear', - warmup_iters=500, - warmup_ratio=0.3333333333333333, - min_lr_ratio=0.001) -runner = dict(type='IterBasedRunner', max_iters=23400) -vis_pipeline = [ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict(type='Collect', keys=['img'], meta_keys=['timestamp', 'lidar2img']) -] -evaluation = dict( - interval=4680, - pipeline=[ - dict(type='LoadMultiViewImageFromFiles', to_float32=True), - dict( - type='Collect', keys=['img'], meta_keys=['timestamp', 'lidar2img']) - ]) -gpu_ids = range(0, 8) diff --git a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/tf_logs/events.out.tfevents.1762928919.VM-120-96-tencentos.834.0 b/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/tf_logs/events.out.tfevents.1762928919.VM-120-96-tencentos.834.0 deleted file mode 100644 index f917b3ffa5e5ce35429a854d2d3e43e218d4ce74..0000000000000000000000000000000000000000 Binary files a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/tf_logs/events.out.tfevents.1762928919.VM-120-96-tencentos.834.0 and /dev/null differ diff --git a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/tf_logs/events.out.tfevents.1762931765.VM-120-96-tencentos.182643.0 b/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/tf_logs/events.out.tfevents.1762931765.VM-120-96-tencentos.182643.0 deleted file mode 100644 index 55a94ce10d3238c73969fc012e51912cd07e8172..0000000000000000000000000000000000000000 Binary files a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/tf_logs/events.out.tfevents.1762931765.VM-120-96-tencentos.182643.0 and /dev/null differ diff --git a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/tf_logs/events.out.tfevents.1762948631.VM-120-96-tencentos.360450.0 b/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/tf_logs/events.out.tfevents.1762948631.VM-120-96-tencentos.360450.0 deleted file mode 100644 index 71353557dd83449a1380522747f4a111c4d85917..0000000000000000000000000000000000000000 Binary files a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/tf_logs/events.out.tfevents.1762948631.VM-120-96-tencentos.360450.0 and /dev/null differ diff --git a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/tf_logs/events.out.tfevents.1762949116.VM-120-96-tencentos.538076.0 b/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/tf_logs/events.out.tfevents.1762949116.VM-120-96-tencentos.538076.0 deleted file mode 100644 index 1c58b49cf39b108c23be5719f4b72eaaf72c1785..0000000000000000000000000000000000000000 Binary files a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/tf_logs/events.out.tfevents.1762949116.VM-120-96-tencentos.538076.0 and /dev/null differ diff --git a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/tf_logs/events.out.tfevents.1762953661.VM-120-96-tencentos.716132.0 b/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/tf_logs/events.out.tfevents.1762953661.VM-120-96-tencentos.716132.0 deleted file mode 100644 index b6af30e374bef7fa10e7d455925a0a0b68f9079f..0000000000000000000000000000000000000000 Binary files a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/tf_logs/events.out.tfevents.1762953661.VM-120-96-tencentos.716132.0 and /dev/null differ diff --git a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/tf_logs/events.out.tfevents.1762954071.VM-120-96-tencentos.893865.0 b/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/tf_logs/events.out.tfevents.1762954071.VM-120-96-tencentos.893865.0 deleted file mode 100644 index 4d952966db1fb82a09cf88755ce52f1094aba4dd..0000000000000000000000000000000000000000 Binary files a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/tf_logs/events.out.tfevents.1762954071.VM-120-96-tencentos.893865.0 and /dev/null differ diff --git a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/tf_logs/events.out.tfevents.1763002135.VM-120-96-tencentos.1071669.0 b/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/tf_logs/events.out.tfevents.1763002135.VM-120-96-tencentos.1071669.0 deleted file mode 100644 index 58a5e60ae66597dd94e8f7c54cc89622983ab8ce..0000000000000000000000000000000000000000 Binary files a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/tf_logs/events.out.tfevents.1763002135.VM-120-96-tencentos.1071669.0 and /dev/null differ diff --git a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/tf_logs/events.out.tfevents.1763016089.VM-120-96-tencentos.1249487.0 b/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/tf_logs/events.out.tfevents.1763016089.VM-120-96-tencentos.1249487.0 deleted file mode 100644 index bb353a1a56cecb32e874b80d530cac4b0d24258b..0000000000000000000000000000000000000000 Binary files a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/tf_logs/events.out.tfevents.1763016089.VM-120-96-tencentos.1249487.0 and /dev/null differ diff --git a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/tf_logs/events.out.tfevents.1763018046.VM-120-96-tencentos.1307434.0 b/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/tf_logs/events.out.tfevents.1763018046.VM-120-96-tencentos.1307434.0 deleted file mode 100644 index 675b63fbbcd39ca578e8b68726a8966364f73e42..0000000000000000000000000000000000000000 Binary files a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/tf_logs/events.out.tfevents.1763018046.VM-120-96-tencentos.1307434.0 and /dev/null differ diff --git a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/tf_logs/events.out.tfevents.1763018246.VM-120-96-tencentos.1365094.0 b/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/tf_logs/events.out.tfevents.1763018246.VM-120-96-tencentos.1365094.0 deleted file mode 100644 index 0d18efc040eabdddda183ee4c26eebd4838bd84f..0000000000000000000000000000000000000000 Binary files a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/tf_logs/events.out.tfevents.1763018246.VM-120-96-tencentos.1365094.0 and /dev/null differ diff --git a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/tf_logs/events.out.tfevents.1763018609.VM-120-96-tencentos.1423102.0 b/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/tf_logs/events.out.tfevents.1763018609.VM-120-96-tencentos.1423102.0 deleted file mode 100644 index d8cff38221a524022021c729aed48594bf33dfc8..0000000000000000000000000000000000000000 Binary files a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/tf_logs/events.out.tfevents.1763018609.VM-120-96-tencentos.1423102.0 and /dev/null differ diff --git a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/tf_logs/events.out.tfevents.1763019118.VM-120-96-tencentos.1480783.0 b/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/tf_logs/events.out.tfevents.1763019118.VM-120-96-tencentos.1480783.0 deleted file mode 100644 index aab7e2b864f46312327f059d3b10bf87031b6302..0000000000000000000000000000000000000000 Binary files a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/tf_logs/events.out.tfevents.1763019118.VM-120-96-tencentos.1480783.0 and /dev/null differ diff --git a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/tf_logs/events.out.tfevents.1763360280.VM-120-96-tencentos.1658802.0 b/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/tf_logs/events.out.tfevents.1763360280.VM-120-96-tencentos.1658802.0 deleted file mode 100644 index 37971413f42256475675b0508c20a7de0f950809..0000000000000000000000000000000000000000 Binary files a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/tf_logs/events.out.tfevents.1763360280.VM-120-96-tencentos.1658802.0 and /dev/null differ diff --git a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/tf_logs/events.out.tfevents.1768396926.bw150-1.18275.0 b/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/tf_logs/events.out.tfevents.1768396926.bw150-1.18275.0 deleted file mode 100644 index 4245be588075a650c394b87b86f198ffabb04db6..0000000000000000000000000000000000000000 Binary files a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/tf_logs/events.out.tfevents.1768396926.bw150-1.18275.0 and /dev/null differ diff --git a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/tf_logs/events.out.tfevents.1768397036.bw150-1.29719.0 b/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/tf_logs/events.out.tfevents.1768397036.bw150-1.29719.0 deleted file mode 100644 index 6c2d3155a30d1db932ad927a1e7476a17a3e7d10..0000000000000000000000000000000000000000 Binary files a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/tf_logs/events.out.tfevents.1768397036.bw150-1.29719.0 and /dev/null differ diff --git a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/tf_logs/events.out.tfevents.1768397360.bw150-1.71671.0 b/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/tf_logs/events.out.tfevents.1768397360.bw150-1.71671.0 deleted file mode 100644 index 64b9c41e65afdee05c92dd7f01d484acf9208f15..0000000000000000000000000000000000000000 Binary files a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/tf_logs/events.out.tfevents.1768397360.bw150-1.71671.0 and /dev/null differ diff --git a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/tf_logs/events.out.tfevents.1770708791.bw28.1304.0 b/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/tf_logs/events.out.tfevents.1770708791.bw28.1304.0 deleted file mode 100644 index 161d53027f68fed19a3d67a10634b10ea691ac0e..0000000000000000000000000000000000000000 Binary files a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/tf_logs/events.out.tfevents.1770708791.bw28.1304.0 and /dev/null differ diff --git a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/tf_logs/events.out.tfevents.1770708907.bw28.5408.0 b/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/tf_logs/events.out.tfevents.1770708907.bw28.5408.0 deleted file mode 100644 index e045ea3b6faefdf8ea954fa1ad58aafc094c5e96..0000000000000000000000000000000000000000 Binary files a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/tf_logs/events.out.tfevents.1770708907.bw28.5408.0 and /dev/null differ diff --git a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/tf_logs/events.out.tfevents.1770709100.bw28.36160.0 b/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/tf_logs/events.out.tfevents.1770709100.bw28.36160.0 deleted file mode 100644 index c0b1ac44659ba93dabcb250c4aca6264b2325daf..0000000000000000000000000000000000000000 Binary files a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/tf_logs/events.out.tfevents.1770709100.bw28.36160.0 and /dev/null differ diff --git a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/tf_logs/events.out.tfevents.1770710621.bw28.558.0 b/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/tf_logs/events.out.tfevents.1770710621.bw28.558.0 deleted file mode 100644 index aa910bf9ce6199fe95b2d8c6648a7cb24b85397d..0000000000000000000000000000000000000000 Binary files a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/tf_logs/events.out.tfevents.1770710621.bw28.558.0 and /dev/null differ diff --git a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/tf_logs/events.out.tfevents.1770710904.bw28.4965.0 b/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/tf_logs/events.out.tfevents.1770710904.bw28.4965.0 deleted file mode 100644 index b92c27c1ab8179aa713686ac8a11ada664ad6912..0000000000000000000000000000000000000000 Binary files a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/tf_logs/events.out.tfevents.1770710904.bw28.4965.0 and /dev/null differ diff --git a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/tf_logs/events.out.tfevents.1770710967.bw28.9100.0 b/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/tf_logs/events.out.tfevents.1770710967.bw28.9100.0 deleted file mode 100644 index baaf5c77d4f62eb2706005731f378a91793d363e..0000000000000000000000000000000000000000 Binary files a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/tf_logs/events.out.tfevents.1770710967.bw28.9100.0 and /dev/null differ diff --git a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/tf_logs/events.out.tfevents.1770711570.bw28.13140.0 b/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/tf_logs/events.out.tfevents.1770711570.bw28.13140.0 deleted file mode 100644 index 5a09fbf546dd4c304aab99539b12897112329148..0000000000000000000000000000000000000000 Binary files a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/tf_logs/events.out.tfevents.1770711570.bw28.13140.0 and /dev/null differ diff --git a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/tf_logs/events.out.tfevents.1770711818.bw28.55.0 b/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/tf_logs/events.out.tfevents.1770711818.bw28.55.0 deleted file mode 100644 index c057c2796a2ecb3e6cc261d7a12f721915c83ac9..0000000000000000000000000000000000000000 Binary files a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/tf_logs/events.out.tfevents.1770711818.bw28.55.0 and /dev/null differ diff --git a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/tf_logs/events.out.tfevents.1770711993.bw28.4728.0 b/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/tf_logs/events.out.tfevents.1770711993.bw28.4728.0 deleted file mode 100644 index 3890ab7f2e9585071fad09dc8868ba2b9fe3877f..0000000000000000000000000000000000000000 Binary files a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/tf_logs/events.out.tfevents.1770711993.bw28.4728.0 and /dev/null differ diff --git a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/tf_logs/events.out.tfevents.1770712065.bw28.8008.0 b/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/tf_logs/events.out.tfevents.1770712065.bw28.8008.0 deleted file mode 100644 index 40221f4f104d720bad33baa581570bc520af5f49..0000000000000000000000000000000000000000 Binary files a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/tf_logs/events.out.tfevents.1770712065.bw28.8008.0 and /dev/null differ diff --git a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/tf_logs/events.out.tfevents.1770712163.bw28.10688.0 b/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/tf_logs/events.out.tfevents.1770712163.bw28.10688.0 deleted file mode 100644 index d765dda8f3e4cba6427764ddd6d6044d3be91ae3..0000000000000000000000000000000000000000 Binary files a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/tf_logs/events.out.tfevents.1770712163.bw28.10688.0 and /dev/null differ diff --git a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/tf_logs/events.out.tfevents.1772790709.bw28.14810.0 b/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/tf_logs/events.out.tfevents.1772790709.bw28.14810.0 deleted file mode 100644 index a8d21878292d64c0a32f6ba20ecdbc92eacdf8a6..0000000000000000000000000000000000000000 Binary files a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/tf_logs/events.out.tfevents.1772790709.bw28.14810.0 and /dev/null differ diff --git a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/tf_logs/events.out.tfevents.1775205530.bw61.1690.0 b/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/tf_logs/events.out.tfevents.1775205530.bw61.1690.0 deleted file mode 100644 index aa0f10d264ad80e64afd02f0d6947c0fb0283bfe..0000000000000000000000000000000000000000 Binary files a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/tf_logs/events.out.tfevents.1775205530.bw61.1690.0 and /dev/null differ diff --git a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/tf_logs/events.out.tfevents.1775205871.bw61.43457.0 b/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/tf_logs/events.out.tfevents.1775205871.bw61.43457.0 deleted file mode 100644 index aa286c70712a8dcff1320f21c604468fde4223cb..0000000000000000000000000000000000000000 Binary files a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/tf_logs/events.out.tfevents.1775205871.bw61.43457.0 and /dev/null differ diff --git a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/tf_logs/events.out.tfevents.1775206217.bw61.83594.0 b/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/tf_logs/events.out.tfevents.1775206217.bw61.83594.0 deleted file mode 100644 index 3e9c9ea30457c723ac955f3bc66bc29699228099..0000000000000000000000000000000000000000 Binary files a/docker-hub/Sparse4D/Sparse4D/work_dirs/sparse4dv3_temporal_r50_1x8_bs6_256x704/tf_logs/events.out.tfevents.1775206217.bw61.83594.0 and /dev/null differ